摘要:

Quarterly Reviews ~~ No2 Vol22 1968 Asymmetric Synthesis By D. R. Boyd and M. A. McKervey Page 95 The Interaction of Aromatic Nitro-compounds with Bases By E. Buncel A. R. Norris and K. E. Russell 123 The Reactions of Ions and Excited Atoms of the Inert Gases By B. Brocklehurst 147 Semiconduction and Photoconduction of Biological Pigments By R. J. Cherry 160 Liquid Crystals as Solvents in Nuclear Magnetic Resonance By G. R. Luckhurst 179 The Theory of Thermal Electron-transfer Reactions in Solution By I. Ruff 199 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents By J. I. G. Cadogan 222 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. 1. Such preliminary 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 Than& Press Margato.

ISSN:0009-2681    

DOI:10.1039/QR96822FP003

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Asymmetric Synthesis By D. R. Boyd and M. A. McKervey DEPARTMENT OF CHEMISTRY QUEEN’S UNIVERSITY BELFAST NORTHERN IRELAND 1 Introduction In a paper entitled ‘A Suggestion Looking to the Extension into space of the Structural formulas at Present Used in Chemistry and a Note upon the Relation between the Optical Activity and the Chemical Constitution of Organic Com- pounds’ van’t Hoff laid down a fundamental concept of stere0chemistry.l van’t Hoff noted that the great majority of optically active organic substances con- tained asymmetric carbon atoms and he recognised that if the carbon atoms were tetrahedral in character they would produce molecules related to each other as enantiomers. Pasteur mentions the tetrahedron in summarising his researches on molecular dissymmetry and as early as 1808 Wollaston predicted that one day chemists would have to think of atomic arrangements in three dimensions.Two years before van’t Hops paper appeared Hofmannz established that a sterically hindered amine such as dimethylmesitylamine did not react with methyl iodide and the concept that steric hindrance of a functional group by a neighbouring non-functional group can retard or prevent a chemical reaction was established. Victor Meye? illustrated this idea particularly well by showing that ortho-disubstituted benzoic acids are much more difficult to esterify and their esters are much more difficult to hydrolyse than if ortho substituents are absent whereas phenylacetic acids even if ortho-disubstituted are easily esterified and their esters are readily hydrolysed. Despite Wollaston’s remark- able understanding of three-dimensional atomic arrangements and the clear directive in the title of van’t Hoff’s paper most of the experiments performed in the ensuing years provided confirmation of the fundamental theory rather than establishing new lines of approach.Similarly the theory of steric effects remained undeveloped. A notable exception in this period was the discovery by KennelA*5 and Turners,7 and their co-workers that 6,6’-dinitrobiphenic acid could be resolved into optically active forms. This observation that rotation about the single bond joining the aromatic nuclei could be restricted because of the effective bulk of the substituents was of great significance because the ideas of van’t Hoff had J. H. van’t Hoff in ‘Classics in the Theory of Chemical Combination’ ed.0. T. Benfey A. W. Hofmann Ber. 1872,5,704; 1875 8 61. V. Meyer Ber. 1894 27 510. J. Kenner and W. V. Stubbings J. Chem. SOC. 1921 593. G. H. Christie and J. Kenner J . Chem. SOC. 1922 614. C. V. Ferriss and E. E. Turner J . Chem. SOC. 1920 1140. Dover Publications New York 1963 p. 151. ’ R. J. W. Le Fevre and E. E. Turner J. Chern. SOC. 1926 2476. 95 Asymmetric Synthesis been so successful that chemical thought tended to overlook the fact that these considerations were based on the principles of free rotation about single bonds and of restricted rotation about double bonds. Kenner and Turner and their co-workers showed that three-dimensional atomic arrangements and rotation about single bonds depend on the size of groups and that steric hind- rance as a kinetic phenomenon must depend on mechanism and it was not until later that mechanisms were sufficiently clearly defined and distinguished.The introduction of the concept of the transition state clearly showed that in order to be able to consider the incidence of steric and electronic effects in any chemical reaction it is essential to know the mechanism at least sufficiently well to be able to define the stereochemistry of the transition state because it is the difference between the magnitude of the effect in the ground state and in the transition state which is im?ortant. These ideas which are commonly grouped together in conformational analysis have been used recently to provide a qualitative under- standing of asymmetric synthesis. In 1874 le Be18 defined the concept of symmetric synthesis.Any synthesis of dissymmetric molecules starting from either symmetrical molecules or a racemic modification and using no optically active reagents or catalysts and no asym- metric physical influence always produces an equal number of the two enan- tiomericaliy related product molecules. The addition of ethylmagnesium iodide to acetaldehyde produces a dissym- metric molecule. There is no reason for the Grignard reagents to have anything but an equal probability of approaching from either side of the acetaldehyde molecule thus producing equal amounts of (+)- and (-)-butan-2-01. The transition states leading to (+)- and (-)-butan-2-01 are said to be enantiomeric and since acetaldehyde is a single non-dissymmetric molecule the free-energy difference between the transition states is zero.It follows that the two reactions occur at exactly the same rate and the products are formed in exactly equal amounts. le Be1 continued ' . . . this is not necessarily true of asymmetric bodies formed in the presence of other active bodies or transversed by circularly polarised light or in short when submitted to any cause whatever which favours the formation of one of the asymmetric isomers. Such conditions are excep- tional; and generally in the case of bodies prepared synthetically those which are active will escape the observation of the chemist unless he endeavours to separate the mixed isomer product the combined action of which upon polarised light is neutral.'8 When a dissymmetric grouping is already present in the molecule and a second grouping (e.g. an asymmetric carbon atom) is created an exactly 1 1 mixture * J.A. le Bel see ref. I p. 161. 96 Boyd and McKervey of the two possible stereoisomers (which are now diastereoisomers) is not expected. This follows from the fact that diastereoisomerically related transition states like diastereoisomerically related ground states differ in free energy and therefore in stability. This difference shows in the rates of formation and in the rates of reaction of stereoisomers and generally a reaction in which new dissymmetric groupings are produced in unequal amounts is called an asym- metric synthesis. The addition of methylmagnesium iodide to the optically active aldehyde (1) provides an e~ample.~ This reaction can yield two diastereoisomerically related alcohols (2) and (3); indeed both are formed but alcohol (3) predominates over (2) by about 2 to 1.The formation of products related as diastereoisomers is not the only route for asymmetric synthesis. Many examples are known in which enan- tiomerically related products are formed in unequal amounts. Such a case is the Ph ,C-Pri 9 . Ph\ PhCPr' + ,C-CH,MgCl - H0' A Et (4) reduction of isopropyl phenyl ketone with the Grignard (4) reagent from (+)-1- chloro-2-phenylbutane. Isopropylphenylmethanol (5) was obtained in 82 % optical yield.1° The magnitude of the asymmetric synthesis can be expressed as the percentage optical yield [a] (product) [a] (pure enantiomer) x 100 = % optical yield where [a] (product) is the optical rotation of the product of the asymmetric synthesis and [a] (pure enantiomer) is the maximum optical rotation obtained (usually) by resolution.The oxidation of sulphide (6) with (+)-peroxycamphoric acid (7) to give the sulphoxide (8) in 4.3% optical yield is an example in which the new dissymmetric grouping is not an asymmetric carbon atom.ll In these D. J. Cram and F. A. Abd Elhafez J. Amer. Chem. SOC. 1952 74 5828. lo J. S. Birtwistle K. Lee J. D. Morrison W. A. Sanderson and H. S. Mosher J. Org. Chem. 1964 29 37. A. Maccioni F. Montanari M. Secci and M. Tramontini Tetrahedron Letters 1961 607. 97 Asymmetric Synthesis two and in related examples the essential feature of the asymmetric synthesis is the diastereoisomeric nature of the transition states leading to enantiomeric products in unequal amounts. The alcoholysis of various anhydrides has been used by Mislow12 to illustrate the types of free-energy relationships found in the formation and reactions of stereoisomers.Three basic types of reaction which are under kinetic control are envisaged In type 1 AGO = 0 and AAG+ = 0; in type 2 AGO = 0 and AAG* # 0; and in type 3 AGO # 0 and AAG* # 0 where Go and G+ represent the free energies of the ground and transition states respectively. (This passage omits much of the detailed analysis and the reader is urged to consult Mislow’s book for amplification of the discussion merely summarised here. Mislow’s analysis is of further importance in that it identifies the concept of asymmetric synthesis within the framework of the general phenomenon of stereoselectivity.) Of the preceding examples the addition of methylmagnesium iodide to acetaldehyde belongs to type 1; the others belong to type 2.The earliest explanation for asymmetric synthesis postulated the dissymmetric polarisation of a symmetrical centre in a molecule under the influence of a nearby pre-existing dissymmetric grouping. Such polarisation was assumed to produce unequal amounts of two diastereoisomeric ‘activated species’ which react to give unequal amounts of stereoi~omers.l~,~* A more recent rationalisation of asymmetric synthesis is one in which the ideas described in the beginning of this Review are used to define topological differences between diastereoisomeric transition states. These differences are ascribed to non-bonded and dipole interactions between the reactants leading to a stereochemically favoured reaction path. Certain idealised rules and transition state models have been devised to rationalise topological difference between transition states and in the subsequent sections of this Review they will be applied to asymmetric synthesis in a few general reaction types.Morrison15 has emphasised the limitations of many of the models ‘It is important that one realise the empirical nature of all such models and appreciate their possible fallibility and theoretical nalvete. It is the opinion of the author that these transition-state models can be useful conceptual devices for correlating experimental results but they should not be given great credence as predictive tools unless a large number of examples have been successfully accommodated. The utility of the model for correlative and predictive purposes does not necessarily confirm it as an accurate representation of the product- controlling transition state.’ 2 Asymmetric Synthesis in Carbonyl-addition Reactions One of the first applications of the concept of steric control of asymmetric synthesis was to the reduction of carbonyl compounds by certain Grignard reagents containing /%hydrogen atoms.Whitmore suggested a transition state l2 K. Mislow ‘Introduction to Stereochemistry’ W. A. Benjamin New York 1965 p. 122. l3 T. M. Lowry and E. E. Walker Nature 1924 113 565. l4 An attempt to disprove this theory has been described by M. J. Kubitscheck and W. A. Bonner J. Org. Chem. 1961 26 2194; W. A. Bonner J. Amer. Chem. SOC. 1963 85,439. l5 J. D. Morrison in ‘Survey of Progress in Chemistry’ ed. A. F. Scott Academic Press New York 1966 vol. 3 p. 147. 98 Boyd and McKervey R Ph But C6H11 in which reduction proceeds via an essentially planar six-membered ring structure.On the basis of this model it was predicted that the alcohol resulting from the reduction of t-butyl methyl ketone (9) with (+)-2-methylbutylmagnesium chloride (10) would contain a preponderance of the (S)-( +)-enantiomer (1 l).LG Me Et Prn Bun Pri Bui C6H11 But Ph 3-9 5.7 5.9 7.2 24 9.9 25 16 13 11 11 11 5 6 2.5 16 3.6 8.8 8.9 11 2.1 16 2.5 25 Et\ H *"A 0 II (9) Me (lo) Me-C-But + C-CH2MgCl I 'Cl t -c1 t J. Me ,OH c' (11) But' 'H Bu OH Me' 'H CO' This was based on the idea that there would be less non-bonded interaction in the suggested transition state (A) in which the t-butyl group (larger than methyl) of the ketone was trans to the ethyl group (larger than methyl) attached to the asymmetric /3-carbon atom of the Grignard reagent.The ten- dency for the transition state to form with the t-butyl and ethyl groups cis should be less because of the apparently greater non-bonded interaction in transi- tion state (B). When the reaction was carried out at 20" the enantiomer of the product resulting from transition state (A) was found to predominate.17J8 Thus 0 0 0 II II C,H ,.CR II But C R PhCR v Et l6 The use of R and S to specify absolute configuration is described by R. S. Cahn C. K. Ingold and V. Prelog Experenfia 1956,12,81; R. S . Cahn J . Chem. Educ. 1964,41 1 1 6 508. l7 H. S. Mosher and E. La Combe J . Amer. Chem. SOC. 1950 72 3994 4991. l8 E. P. Burrows F. J. Welch and H. S. Mosher J . Amer. Chem. SOC. 1960 82,880; see also ref. 10. 99 Asymmetric Synthesis the direction of the asymmetric synthesis appears to be governed by differences in the free energies of the diastereoisomeric transition states which arise from differences in non-bonded interactions between the groups attached to the /3-carbon atom of the Grignard reagent and to the carbonyl group of the ketone.Amongst the possible difficulties which arise in deciding whether this interpretation is consistent with an experimental result is that of arranging the groups involved in order of effective size. The optical yields obtained for the reduction of three series of ketones with the Grignard reagent (10) are summarised in Table 1.17J8 If the assumption is made that the order of effective size for the three largest groups in the ketone series is phenyl > t-butyl > cyclohexyl and that the ethyl group is larger than the methyl group in the dissymmetric Grignard reagent then the predominant enantiomer of the product is that which arises from the transition state in which the larger group of the ketone is trans to the ethyl group of the Grignard reagent.On the surface this conclusion appears highly plausible in that it correlates the absolute con- figuration of the predominant enantiomer of the product in each case and that of (+)-l-chloro-2-methylbutane. However an examination of the percentages in Table 1 appear to diminish its credibility. When R is methyl ethyl and n-propyl the percentages are highest for the t-butyl ketones; when R is isopropyl the percentage is highest for the phenyl ketone and when R is isobutyl the per- centage is highest for the cyclohexyl ketone.The difference between the free energies of the two transition states would be expected to decrease as R increases in effective size yet the percentage for cyclohexyl phenyl ketone is much larger than that for methyl phenyl ketone and the percentage for cyclohexyl isobutyl ketone is much larger than that for cyclohexyl methyl ketone. Morrisonls suggests that since the energy difference between the two transition states is extremely sensitive to subtle changes in the nature of the groups being com- pressed in these transition states the phenyl and cyclohexyl groups may have effective sizes that are a function of their rotational conformations in the transition state. If by compression of the groups in the transition state is meant van der Waals compression it should be emphasised that since van der Waals interactions have the steepest potential-energy gradient with respect to distance of all the causes of strain all other parameters (torsional and angle deformations) will tend to change before van der Waals compression becomes important.Although restricted rotation of the phenyl and cyclohexyl groups may be impor- tant in any detailed analysis of the reaction the interpretative difficulties may be due to the limitations imposed by approximately planar transition-state models. For example one can draw non-planar transition-state models having topological characteristics in which the non-bonded interactions are of quite a different order. Recent results of Mosher and his co-workers emphasise that the difference in effective size of the two substituents attached to the carbonyl group is not the only factor responsible for determining the stereoselectivity of these asymmetric reductions.Reduction of [a-2H,]benzaldehyde with the optically active Grignard reagent from ( +)-l-chloro-2-methylbutane gave [a-2HJ(S)-( +)-benzyl alcohol in 19% optical yield; reduction of t-butyl phenyl ketone with the same reagent 100 Boyd and McKervey gave approximately the same optical yield.lS Several other types of asymmetric reduction have been rationalised in terms of non-bonded interactions in six- membered cyclic transition states; these have been reviewed by Morrison.16 In the examples of asymmetric reduction summarised in Table 1 the hydrogen transfer occurs from the asymmetric ,&arbon atom of the Grignard reagent.Morrison20 has established that asymmetric reduction is possible even when the carbon atom from which hydrogen transfer occurs is not formally asymmetric. )CH-CH,MqCl H '; (13) R I E t Ph ,OH c Predominant enant iomer H +-The environment of this hydrogen R 8" H is diastereoisomeric with the environment of ,C-C \this hydrogen by internal comparison. Thus reduction of isopropyl phenyl ketone (12) with the Grignard reagents prepared from (I?)-( -)-l-chloro-3-phenylpentane (1 3) and (I?)-( -)-1-chloro- 3-phenylbutane (14) gives (R)-( +)-isopropylphenylmethanol(l5) in 29 and 23 % optical yields respectively. The essential feature of this type of asymmetric reduction is the diastereoisomeric environments of the two hydrogen atoms available for transfer to the ketone. Other examples of this phenomenon may be found in the asymmetric reduction of a series of methyl ketones with the lithium aluminium hydride-3-O-benzyl-1,2-cyclohexylidene-a-~-glucofuranose com- + R\ ,c=o Me Me \ .C- R"' 1 H Predominant e na n t io me r (s) lev.E. Althouse D. M. Feigl W. A. Sanderson and H. S. Mosher J. Amer. Chem. SOC. 1966,88,3595; W. A. Sanderson and H. S. Mosher ibid. 1961,83,5033. 2o J. D. Morrison D. Black and R. Ridgeway Chem. Eng. News 1967 45 48. 101 Asymmetric Synthesis R Optical yield (%) Table 2 Ph But Prn CH,(CH& But CH = C(CH,)-CH 34.1 2.1 13.4 12.2 25.3 31.0 designated small (S) medium (M) and large (L) is so orientated that the diastereoisomeric faces of the carbonyl function are flanked by the groups S and M the reagent RIX in an addition reaction approaches the carbonyl group from the side of the group S provided that the reaction is non-catalytic and that the products are formed in a kinetically controlled process and not in a sub- sequent equilibration.Cram successfully correlated the configurations of a large number of compounds using six different reactions of this type the stereo- 21 s. R. Landor B. J. Miller and A. R. Tatchell J. Chem. Soc. (C) 1966 1822,2280; for the use of other sugar derivatives see 0. Cervinka and A. Fabryova Tetrahedron Lefters 1967 1179. 23 0. Cervinka Coll. Czech. Chem. Comm. 1965 30 1684. 1 02 S. R. Landor B. J. Miller and A. R. Tatchell J. Chem. SOC. (C) 1967 197. Boyd and McKervey chemical relationships between the asymmetric centres having been independ- ently determined in each case.The addition of methylmagnesium iodide to the aldehyde (1 8) provides an example. The model predicts that the erythro-isomer will predominate in the product if the assumption is made that the order of decreasing effective size on the asymmetric carbon atom is Ph > R > H. When R is r n e t h ~ l ~ ~ ~ and isopropy126 this is the experimental result. ery t h r o ihreo Table 3 R = Me R = Pri R = Et erythrolthreo ratio 2-4/1994 1.0-1.9/126 2 ~ 5 1 1 ~ ~ The decrease in the diastereoisomeric product ratio as R varies from methyl to isopropyl illustrates one of the interpretative difficulties namely that of arrang- ing the groups on the asymmetric carbon atom in order of effective size. The above results tend to indicate that the isopropyl group has a smaller effective size than that of methyl.While the actual size of a group is obviously constant its effect on conformational equilibria or on a nearby reaction site depends very much on its immediate environment and on the reaction mechanism. For example the isopropyl group when a substituent on a cyclohexane ring appears almost as ‘large’ as a t-butyl group in certain experiments and nearly the same size as a methyl group in other^.^',^^ It is important to the study of asymmetric synthesis in acyclic molecules that the order of effective size of groups established by n.m.r. methods shows significant differences from that of their preference for the equatorial position of a cyclohexane ring.29 Increasing the steric require- ment of the approaching reagent results in an increase in the diastereoisomeric ratio.Thus reaction of 3-phenylbutan-2-one with the methyl ethyl and phenyl Grignard reagent gives the expected evythro product predominating by a ratio of 2:1 3:1 and 5:1 respectively in the three cases.3o The proven utility of Cram’s model does not necessarily mean that it is an accurate representation of the diastereoisomeric transition states. It was em- s4 Y. Gault and H. Felkin Bull. SOC. chim. France 1960 1342. 86 D. J. Cram F. A. Abd Elhafez and H. Weingartner J. Amer. Chem. Soc. 1953,75,2293. 26 D. J. Cram F. A. Abd Elhafez and H. LeRoy Nyquist J. Amer. Chem. SOC. 1954,76,22. N . L. Allinger L. A. Freiberg and Shih-En Hu J. Amer. Chem. SOC. 1962 84,2836. 28 R. D. Stolow J. Amer. Chem. SOC. 1964 86,2170. 2o G. M. Whitesides J. P. Sevenair and R. W. Goetz J . Amer. Chem. SOC.1967 89 1135. Unpublished results of B. P. Thill quoted by E. L. Eliel N. L. Allinger S. J. Angyal and G. A. Morrison in ‘Conformational Analysis’ Interscience New York 1966 p. 34. Asymmetric Synthesis phasised earlier that in order to consider the incidence of steric effects in a chemical reaction it is essential to know its mechanism sufficiently well to be able to define the stereochemistry of the transition state. In the first place several investigations have established that the stable ground state conformation of a carbonyl compound is one in which the carbonyl function is eclipsed and the R group staggered,31,s2,33 and therefore is not thai implied in (17). However this is not a serious objection provided that the activation energy of the reaction is large compared with the energy barrier to rotation about the relevant sp3-sp2 carbon-carbon single bond.= If this condition obtains the diastereoisomeric product ratio should depend entirely on the free-energy differences between the two transition states.KarabatsosS has developed the Cram model to a point where semiquanti- tative predictions of product stereospecificity are possible. The following assump- tions are required. (a) The two transition states resemble the reactants i.e. little bond-breaking and -making has occurred at the transition states. Therefore the arrangement of the three groups on the asymmetric carbon atom with respect to the carbonyl group is similar to that in the ground state of the aIdehyde or ketone. (6) The two product-controlling transition-state models (19) and (20) (though these were not the only models considered%) have the incoming group R2 closest to the smallest group S.(c) The diastereoisomeric product R' (19) R ' (20) i (8) ratio A/B reflects the relative magnitude of the carbonyl-eclipsed group interac- tions M c3 0 in (19) and L t--t 0 in (20). Karabat~os~~ has calculated the free-energy differences (GA+ - G B ~ ) using only M * 0 and L +-+ 0 interac- tions for a large number of reactions for which the experimental free-energy differences are available. The experimental GA+ - GB+ values are obtained from the experimental product ratios A/B by use of the equation GA* - GB* = -RT In A/B The results are summarised in Table 4.55 Although the generally good agreement between the experimental and calculated free-energy differences must be fortui- a1 G.J. Karabatsos and N. Hsi J. Amer. Chem. SOC. 1965,87,2864. 3% G. J. Karabatsos and N. Hsi Tetrahedron 1967,23 1079. aa G. J. Karabatsos and K. L. Krumel Tetrahedron 1967 23 1097. a4 E. L. Eliel 'Stereochemistry of Carbon Compounds' McGraw-Hill New York 1962 p. 151. as G. J. Karabatsos J. Amer. Chem. SOC. 1967 89 1367. 104 Boyd and McKervey tous to some extent it does lend credence to the validity of the general approach. Karabatsos stresses the following factors which may impose serious limitations. (a) The model does not include solvation of the transition states. (b) All non- bonded interactions except M +-+ 0 and L +-+ 0 are neglec!ed; the experi- mental free-energy difference could depend on the effective size of R2. (c) The extent of bond-breaking and -making at the transition state could vary with each reaction.(d) Differences in the entropies of the two transition states could affect the experimental GA* - GB* values. Table 4 Experimental and calculated free-energy diferences between the two diastereoisomeric transition states GA+ - Ggf (cal./mole) No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Reaction Found Ref. PhMeHC-CHO + MeMgI PhMeHC-COMe + LiAlH PhMeHC-COEt + LiAlH PhMeHC-CHO + EtMgI PhMeHC-COPh + LiAlH PhMeHC-CHO + PhMgBr PhMeHC-CHO + MeMgBr PhMeHC-COMe + LiAlH PhMeHC-C(Me) = BMg+X + LiAlH PhMeHC-C(Me) = NMg+X + PhEtHC-CHO + MeMgI PhEtHC-COMe + LiAlH PhEtHC-CHO + EtMgBr PhEtHC-COEt + LiAlH PhEtMeC-COMe + EtLi PhEtMeC-COEt + MeLi PhEtMeC-CHO + PhMgBr PhEtMeC-COPh + LiAlH PhPrlHC-CHO + PriMgBr PhPrlHC-CHO + Pr'MgBr + 6MgBr2 PhPr'HC-CHO + PriLi C,H,,MeHC-CHO + MeMgI C,HllMeHC-CHO + MeLi C,H,,MeHC-CHO f MeLi NH$HCO,- (pentane) (pentane) (ether) -410 - 550 -410 - 650 - 830 - 830 - 460 - 660 (0 "1 - 400 - 420 - 550 - 650 - 650 - 650 - 530 - 830 -410 - 470 - 380 - 160 0 - 380 -220 -110 9 9 9 9 9 9 34 -700 34 (-70") a a 9 9 25 25 b b C C 26 26 26 d d d C ~ I C .~ ~ - 600 - 600 - 600 - 600 - 600 -600 -6600 - 600 - 700 - 7 0 - 500 -5500 - 5 0 -5500 -5300 -5500 -5500 -5500 -200 -2200 -2200 - 300 - 3 0 -3300 105 Asymmetric Synthesis Table kontinued No. Reaction GA# - G# (cal./mole) Found Ref. Calc.ss 25 C,H,,MeHC-COMe + NaBH - 320 d 26 C,H,,MeHC-COMe + LiAlH -2200 d 27 EtMeHC-CHO + MeMgBr - 220 24 28 EtMeHC-COMe + LiAlH -16 $32 24 +36 (-70") (0") (35") 29 PhPr'HC-COPr' + LiAlH -1380 26 30* C,H,,MeHC-COMe 4- Al(PriO) +380 d -3300 -3300 - 100 -100 -2200 -3300 * The model does not predict the stereochemical result.a D. J. Cram and J. E. McCarty J . Amer. Chem. SOC. 1954,76,5740; b D. J. Cram and J. D. Knight J . Amer. Chem. SOC. 1952 74 5835; D. J. Cram and J. Allinger J. Amer. Chem. SOC. 1954,76,4516; D. J. Cram and F. D. Greene J . Amer. Chem. SOC. 1953 75 6005. Cram and KopeckYs have considered a second model for systems in which one of the groups (e.g. OH) attached to the asymmetric carbon atom is capable of chelating with a metal atom in the reagent. This model involves a relatively rigid five-membered ring structure which fixes the conformation of the reacting species. For systems in which an amino- or hydroxyl group on the asymmetric carbon atom is the medium-sized group the open-chain and cyclic models predict the same stereochemical result.However when the chelating group is the small group only the rigid model (21) predicts the correct result. For example the open-chain model (17) predicts that the addition of methyl-lithium to 2-hydroxy-l,2-diphenylpropan-l-one (22) will give a meso/( &) product ratio of less than unity; the cyclic model (23) predicts a ratio greater than unity. When the reaction solvent was ether the experimental ratio was 8 11. The stereo- selectivity of reactions of this type is strongly dependent on the solvent3' and on Predominant isomer 36 D. J. Cram and K. R. Kopecky J. Amer. Chem. SOC. 1959 81,2748. D. J. Cram and D. R. Wilson J . Amer. Chem. SOC. 1963 85 1245. 106 Boyd and McKervey even slight variation of the reagent.In the addition of Grignard reagents to phenylacetoin HO-CMePhCOMe for example use of phenylmagnesium chloride or bromide gives a preponderance of the meso-glycol whereas with phenylmagnesium iodide the (&)-product predominates (ratios meso :( &) = 3.1 :1 2.1 :1 and 1 :2 A six-membered cyclic transition state has been suggested to account for the very high stereoselectivity observed in the reaction of phenylmagnesium bromide with the ketone (24).39 The dipolar model (25) has been used to account for the stereochemistry of additions of Grignard reagents and lithium alkyls to a-chloro-carbonyl c What has been said already about the diastereoisomeric faces of the carbonyl group in compounds of type (17) applies equally well to situations in which the carbonyl group is further removed from the asymmetric centre.Such a case is the asymmetric atrolactic acid synthesis described originally by McKenzie and Thompson>l and submitted to conformational analysis by Pre10g.~~ When (-)-menthol is esterified with phenylglyoxalyl chloride and the resulting (-)-menthy1 phenylglyoxylate (26) is treated with methylmagnesium iodide addition of the methyl group to the two diastereoisomeric faces of the ketonic carbonyl group will result in the esters (27) and (28); these are diastereoisomers which will usually be produced in unequal amounts. Quantitative hydrolysis of the ester mixture will result in unequal amounts of (R)-and (S)-atrolactic acid plus the original (-)-menthol. (Three cases in which McKenzie's work was in dis- agreement with Prelog's model were reinvestigated and the discrepancy proved to result from differential hydrolysis during isolation of the a-hydroxy-acid ; J.H. Stocker P. Sidisunthorn B. M. Benjamin and C. J. Collins J. Amer. Chem. SOC. 1960 82 3913. *O K. W. Bentley D. G. Hardy and B. Meek J . Amer. Chem. SOC. 1967,89 3273; see also S. Yamada and K. Koga Tetrahedron Letters 1967 1711. 40 J. W. Cornforth R. H. Cornforth and K. K. Mathew J. Chem. SOC. 1959 112. I1 A. McKenzie and H. B. Thompson J. Chern. SUC. 1905 1004; and many later papers. 4a V. Prelog Helv. Chim. Acta 1953 36 308. 107 Asymmetric Synthesis McMgI I Ph ,CO,H Ma 'OH ,c. Hydrolysis I Ph ,CO,H HO Me dCX*\ + @?)-(-I- Atrolactic acid (S)- (+)-Atrolactic acid Predominant ena n tiomer quantitative hydrolysis is essential since one diastereoisomer may hydrolyse faster than the other.) Prelog has suggested a model which satisfactorily explains this result and many others.The keto-ester grouping is assumed to lie essentially in a plane in which the two carbonyl groups have the transoid conformation and in which the smallest group (hydrogen) is eclipsed with the ketone carbonyl. Addition of methylmagnesium iodide will then be more rapid from the less hindered side of the carbonyl group [i.e. the side of the methylene at C(2)] leading to a preponderance of (R)-(-)-atrolactic acid. In cases where a reducing agent is used rather than a Grignard reagent unequal amounts of (+)- and (-)-mandelic acid or unequal amounts of (+)- and (-)-phenylethane-l,2-diol are obtained depending on the reducing agent used. In the above example of Prelog's rule (-)-menthol functions as an optical activating agent; and in general when the configuration of the activating alcohol is known that of the a-hydroxy-acid obtained in excess can be deduced and vice versa.The con- figuration of (-)-t-butyl ethanol and that of the hydroxyl-bearing carbon atom in a series of triterpenes and steroids have been determined by this m e f h ~ d . ~ ~ ~ ~ A judgment of the relative sizes S M and L of the groups attached to the asymmetric centre is necessary in each case. 43 V. Prelog E. Philbin E. Watanabe and M. Wilhelm Helv. Chim. Ada 1956 39 1086. 44 W. G. Dauben D. F. Dickel 0. Jeger and V. Prelog Helv. Chim. Acta 1953 36 325. 45 V. Prelog and H. Meier Helv Chim. Acta. 1953 36 320. 108 Boyd and McKervey 3 Asymmetric Synthesis in Additions to Carbon-Carbon and Carbon-Nitro- gen Double Bonds The most striking examples of asymmetric synthesis in olefin-addition reactions have been discovered by Brown and his co-workers using the hydroboronation reaction.46 Dialkylboranes prepared by hydroboronation of hindered olefins exhibit very high stereoselectivity towards olehs with different structural features.Brown has shown that (-)-sym-tetraisopinocampheyldiborane gives alcohols of very high optical purity when used as a reagent for the hydration of 1 ,Zdisubstituted cis ole fin^.^^ The hydroboronation of (+)-a-pinene (29) 4 & + (29) + @I-(-)- butan-2-01 Predominant enant iomer proceeds readily to give (-)-sym-tetraisopinopheyldiborane (30) by stereo specific cis addition of the boron-hydrogen bond to the double bond of (+)-a- pinene from the less hindered side of the molecule.Reaction of cis but-Zene with (30) followed by oxidation of the resulting diisopinocampheyl-2-butyl- borane (31) with alkaline hydrogen peroxide gives (R)-(-)-butan-2-01 of 87 % optical The hydroboronation of a number of 1 ,Zdisubstituted cis olefins proceeds similarly to give alcohols of high optical yield (Table 5 ) . 4 7 T h ~ exceptionally high values have been ascribed to an unusually good steric fit of the olefin and a particular conformation of the borane in a model in which the Table 5 a-Pinene used Olefin + cis-But-2-ene - cis-Pent-2-ene + cis-Hex-3-ene - cis-4-Methyl -pent-Zene + Norbornene 3. Bicyclohept adiene Alcohol Optical purity (%) ( -)-But=-2-01 87 ( +)-Pentan-2-01 82 ( -)-Hexan-3-01 91 (+)-4-Methyl -pentan-2-01 76 (- )-exo-Norborneol 67-70 ( + )-exo-Dehydro -norborneol 48-5 1 H.C. Brown ‘Hydroboration’ W. A. Benjamin New York. 1962. 47 H. C. Brown N. R. Ayyangar and G. Zweifel J. Amer. Chem. SOC. 1964,86,397. 109 Asymmetric Synthesis cis-addition of the boron-hydrogen bond on to the olefin proceeds via a four- centre transition state. Although (- >syrn-tetraisopinocampheyldiborane exists as the dimer the monomeric form (32) is used (for simplicity) in representations of the product-controlling transition states.48 The absolute configuration of (+)-a-pinene is known and (32) is considered to represent the most stable conformation of (-)-di-i~opinocampheylborane?~ The significant non-bonded H interactions in the transition-state models (33) and (34) are thought to be those between the methyl and hydrogen groups of the olefin and the hydrogen atom at C(3’) and the larger methylene group at C(4) of the borane.Model (33) has the methyl group and C(4) positioned away from each other whereas model (34) has these two groups in close proximity. On this basis (33) is preferred and it will lead to (R)-(-)-butan-2-01 after oxidation of the organoborane. Similar Me-- C- H H-- C- Me H H--$$@ M g c 4 ( ! .A M#,c” \dl 2 3 ’ ’& s 8<-eB*m’&’ (34) s’ ‘L (33) \ ;*** &.*M M s’ ‘L transition-state models may be used to predict the absolute configurations of alcohols resulting from the hydroboronation of 2-methylalk-l-enes with di- isopinocampheylborane. The significant non-bonded interactions are different and the optical yields are much lower than those obtained with 1 ,Zdisubstituted cis ole fin^.*^ Brown has also established that the models devised for cis and terminal olefins do not apply to trans and hindered olefins.For example applica- tion of the hydroboronation-oxidation sequence to trans-but-2-ene gives butan-2-01 of 13 % optical purity and of configuration opposite to that predicted by model (33).60 This discrepancy between the behaviour of cis and trans olefins was traced to a difference in mechanism for it was observed that the reaction of di-isopinocampheylborane with trans-but-Zene was much slower than that with cis-but-Zene and it was accompanied by the dissociation of the reagent into a-pinene and tri-isopinocampheyldiborane. Displacement of a-pinene from di- isopinocampheylborane was also observed in the hydroboronation of highly 48 For a recent publication describing the use of the dimer in representations of the transition state for these asymmetric hydroboration reactions see D.R. Brown S. F. A. Kettle J. McKenna and J. M. McKenna Chem. Comm. 1967,667. 49 G. Zweifel N. R. Ayyangar T. Munekata and H. C. Brown J. Amer. Chem. SOC. 1964 86 1076. H. C. Brown N. R. Ayyangar and G. Zweifel J . Amer. Chem. Suc. 1964 86 1071. 110 Boyd and McKervey hindered olefins. Although a model utilising tri-isopinocampheyldiborane has not been formulated Brown has suggested the following generalisation to account for the behaviour of trans and hindered olefins. ‘Whenever displacement of a-pinene occurs in stoicheiometric amounts in the hydroboronation of olefins with di-isopinocampheylborane the alcohol (or the olefin; this refers to the kinetic resolution of a racemic olefin with di-isopin~campheylborane~~) obtained will possess the configuration opposite from that predicted on the basis of the simple addition model.For such olefins the use of tri-isopinocampheyldiborane will yield the same result.’50 Streitwieser and his co-workers have suggested a model which accounts for the formation of [1-2Hl] (R)-(-)-butan-1-01 in the hydroboronationof [ l-2H,]cis-but-l-ene with (-)-di-isopinocan~pheyldiborane.~~ The composition of transition-state models for asymmetric synthesis in terms of reactants only should not be taken as evidence that solvation is unimportant. Solvent effects have been observed in the reactions of olefins with peroxy-acids. Some olefins give optically active epoxides when dissymmetric peroxy-acids are used as oxidants.For example epoxidation of styrene (35) with (+)- enant iomer peroxycamphoric acid (36) in chloroform gives (S)-( -)-styrene oxide (37) of 4.4% optical Additional results that have been obtained with (+)- peroxycamphoric acid are summarised in Table 6.52s53 The optical yields of these asymmetric epoxidations are extremely low; 4.4% being taken as the upper limit for the optical purity of the epoxides produced the free-energy difference between the diastereoisomeric transition states is about 45 cal./mole at the temperature of the epoxidation. Although a simple transition-state model Table 6 R in R-CH=CH Methyl n-Butyl n-Pentyl n-Hexyl Phenyl Cyclohexyl 2-Phenylet hyl t-Butyl Optical purity ( %) 2.0 2-5 1.8 2.5 1.7 4.4 3.7 2.1 Epxide Configuration S S S S S S S S ti1 A.Streitwieser jun. L. Verbit and R. Bittman J. Org. Chem. 1967,32 1530. ae R. C. Ewins H. B. Henbest and M. A. McKervey Chem. Comm. 1967 1085. p. 83. H. B. Henbest in ‘Organic Reaction Mechanisms’ Chem. SOC. Special Publ. No. 19 1965 111 Asymmetric Synthesis of the type (38) can be used to interpret the absolute configurations of the epoxides listed in Table 6 it does not take into account the effect of solvent on the reaction. The suggestion has been made that the non-bonded interac- tions at the transition state can be transmitted partly through interposing solvent molecules; if this is so then a transition state which is not well solvated could lead to a low optical yield whereas a transition state with which there is associated a definite solvent structure could lead to a higher optical yield.The results with the reaction of styrene with (+)-peroxycamphoric acid show that optical yields and reaction rates are highest in dichloromethane and chloroform. On the other hand ether and carbon tetrachloride produce a much slower reaction and smaller optical yields. For the range of solvents studied there is a correlation between (a) the optical yield in the epoxidation of styrene (b) the rate of this reaction and (c) the axial-equatorial product ratio in the epoxidation Me$ H~~~ Me&:bl of (39a) and (39b) (Table 7).52,65 For a more detailed discussion of solvent effects in the reaction of olefins with peroxy-acids see refs. 54 and 55. (390) Me Table 7 Axial attack (%) Solvent Optical yield (%) 104k(l.mole-l sec.-l) (39a) (39b) Et 2 0 2.0 cc14 2.0 CHzCl2 3-5 CHCl 4.4 c6H6 2.5 (0.3) 85 35 3.7 85 25 6.4 82 20 8.6 79 17 13-3 79 17 Asymmetric synthesis has been observed in the epoxidation of linalool (40). The epoxide (41) obtained after removal of the original asymmetric centre was found to be optically active.s6 Grundon and co-workers have used (+)-peroxy- tu P. Renolen and J. Ugelstad J. Chim. phys. 1960 57 634. 65 R. G. Carlson and N. S. Behn J. Org. Chem. 1967,32 1363. 66 G. V. Pigulevskii and G. V. Markina Doklady Akad. Nauk S.S.S.R. 1948 63 6277. (Chem. Abs. 1949,43,4628). 112 Boyd and McKervey Me /O\ Me /O\ F C -CH-(CH,),-?= CHMe - C-W(CH&Y-U-$- Me Me Me Ma/ Me / (4 I> camphoric acid (P.C.A.) in the asymmetric synthesis of the optically active quinoline alkaloids (42) and (43).57 Wie Me P.C.A .__c Although the optical yield in the asymmetric epoxidation of styrene is solvent- dependent the configuration of the predominant enantiomer of the product remains the same throughout the range of liquids studied. There is at least one example of an olefin addition in which both the stereochemical outcome and the optical yield are solvent-dependent. The base-catalysed condensation of (-)- menthylchloroacetate with ethyl acrylate gives optically active trans-cyclo- propane-1,Zdicarboxylic acid (44) plus a trace of the cis diacid (45) after removal of the optical activating menthyl group by hydrolysi~.~~,~~ The stereochemistry of this reaction depends markedly on the solvent; in toluene the trans diacid Me 57 R. M. Bowman and M.F. Grundon J. Chem. SOC. (C) 1967,2368; R. M. Bowman J. F' Collins and M. F. Grundon Chem. Comm. 1967 1131. Y . Inouye S. Inamasu M. Ohno T. Sugita and H. M . Walborsky J. Amer. Chem. Soc. 1961 83,2962. H. M. Walborsky and C. G. Pitt J . Amer. Chem. SOC. 1962 84 4831. 113 Asymmetric Synthesis obtained is laevorotatory (optical yield 1-8-3.1 %) whereas in dimethylforma- mide this acid has the opposite Configuration and the optical yield is higher (10.2-10-9 %).58959 It has been suggested60 that the actual asymmetric synthesis of the trans diacid is unaffected by a change of solvent but that the cis-trans isomer ratio is solvent-dependent; and that the change in configuration of the trans diacid is due to an asymmetric cis to trans isomerisation during the alkaline hydrolysis of the cis diester which is formed in the initial kinetically controlled process.Recent results of Inouye and his co-workers cast doubt on this inter- pretation These authors report that the cis-trans and the (+)-(-) stereo- n H Me Ma H Y ’.. 1 ol-naphthyl ccc C-CH2C02H + / \ Md’& selectivities are both solvent-dependent in the sodium hydride-catalysed con- densation of (-)-menthylchloropropionate and methyl methacrylate (a system in which isomerisation of the diesters is excluded).s1 Condensation of (-)- menthyl or (+)-bornyl 2-arylacrylates with dimethyloxosulphonium methoxide or dimethylsulphonium methoxide followed by alkaline hydrolysis gives optically active 2-arylcyclopropanecarboxylic acids.62 Partial asymmetric synthesis has been realised in the Diels-Alder reaction of butadiene with (-)-dimenthy1 fumarate both thermally and at lower tempera- tures with catalysis by Lewis acids.63 The predominant enantiomer of the product obtained in the presence of the Lewis acids was of opposite configuration from that produced in the thermal reacti0n.6~ A more recent study has revealed that the stereoselectivity of this reaction is subject to temperature and pressure effect^.^^,^^ Prelog’s interpretation of asymmetric synthesis in the catalytic hydrogenation of olefins is based on the generalisation that the stereochemical course can often be explained by assuming that the least hindered face of the molecule is pre- ferentially adsorbed on the catalyst surface and that addition of hydrogen occurs from the same side.6s For example the optically active /%methylcinnamate (46) on hydrogenation and hydrolysis yields preferentially (R)-( -)-#%phenyl- butyric acid (47).66 The ester is represented as reacting in the preferred con- formation (46) in which all the groups except those on the asymmetric carbon atom lie in a plane parallel to the catalyst surface.Addition of hydrogen then 6o L. L. McCoy J. Org. Chem. 1964,29,240. *s H. M. Walborsky L. Barash and T. C. Davis Tetrahedron 1963,19 2333. s4B. S. Elyanov E. I. Klabunovskii M. G. Gonikberg G. M. Parfenova and L. F. Godunova Izvest. Akad. Nauk. S.S.S.R. Ser. khim. 1966 9 1678. (Chem. Abs. 1967 66 6088). 66 J. Saver pnd J. Kredel Tetrahedron Letters 1966 6359. 66 V. Prelog and H. Scherrer Helv. Chim. Acta 1959 42 2227. Y. Inouye S. Inamasu and M. Horiike Chem. and Ind. 1967 1293. H. Nozaki H.Ito D. Tunernoto and K. Kondo Tetrahedron 1966 22 441. 114 Boyd and McKervey occurs predominantly from the side of the small group (in this case hydrogen). In this example the incipient asymmetric centre is four bond lengths removed from the optical activating group. Armsa7 has studied a number of examples in which these two centres are adjacent to each other. Thus by catalytic hydro- genation of the enantiomers of 3-ethylhept-3-en-2-01 to 3-ethylheptan-2-01 followed by oxidation to optically active 3-ethylheptan-2-one it was shown that asymmetric hydrogenation had occurred to the extent of 70%.s7 The partial asymmetric synthesis of optically active lupinine has been achieved by sodium borohydride reduction of the carbon-carbon double bond in the menthyl ester (48). Removal of the menthyl group by reductive cleavage with lithium alu- minium hydride gave (+)-lupinhe (49) in ca.10% optical yield.68 The asym- metric reduction of carbon-nitrogen double bonds has been examined as a possible route for the preparation of optically active arnines and amino-acids. A recent case involves the use of optically active organoborane derivatives in Table 8 (50)-R (1) CH (2) CH3 (4) CH3 (3) CH2CH2CH3 (5) CH2CHzCH3 Amine Reagent Configuration Optical purity ( %) DIPCB S 2.0- 3.3 TIPCB S 2.2- 3.2 TIPCB S 2.9-10.7 LHBIB R 19.5-24.0 LHBIB R 4.0- 4-3 the reduction of 2-methyl- and 2-propyl-Al-piperideine (5O).gs Reactions (1)-(3) (Table S) in which the A I-piperideines were treated with di-isopinocampheyl- borane (DIPCB) or tri-isopinocampheyldi borane (TIPCB) gave preferentially the (9-amine in low optical yield.On the other hand lithium hydro-(l-buty1)- bis(isopinocamphey1) borate (LHBIB) prepared from di-isopinocampheyl- 67 C . L. Arcus Proc. Chem. SOC. 1964 135. 6* S. I. Goldberg and 1. Ragade J. Org. Chem. 1967 32 1046. '' D. R. Boyd M. F. Grundon and W. R. Jackson Tetrahedron Lctrers 1967,2101. 115 Asymmetric Synthesis borane and n-butyl-lithium gave preferentially the (R)-amine [reactions (4) and (5)]. The important contributions of Hiskey and N~rthrop'~ to the asymmetric synthesis of amino-acids have been reviewed by Morrison.ls 4 Asymmetric Synthesis in Elimination Reactions Most examples of asymmetric synthesis in elimination reactions involve the formation of optically active olefins. Goldberg and Lam71 have investigated the pyrolysis of the four possible stereoisomers of optically active Lt-methylcyclo- Configuration uncertain 30% (53) (s) - (-1 Ph-y-0 the 250' 1 47% ZOOo\ 70% (s)-~-l - irons (R) - (+) " trans Scheme.Asymmetric synthesis in ester sulphoxide and amine oxide pyrolysis. The arrows indicate the reaction temperature the optical yield and the predominant enantiomer of the olefin. hexyl hydratropate (two of which are illustrated in the Scheme). In each case optically active 4-methylcyclohexene of very low optical purity (0-87 % maxi- mum) was obtained. Although transition-state models capable of accounting 70 R. G. Hiskey and R. C. Northrop J. Amer. Chem. SOC. 1961 83 4798; ibid. 1965 87 1753; see also K. Harada J. Org. Chem. 1967,32 1790 and K. Harada and K. Matsumoto ibid. p. 1794. 71 S. I.Goldberg and F-L. Lam J . Org. Chem. 1966 31 2336. 116 Boyd and McKervey for these results have been formulated the authors agree that the very small free-energy difference between the diastereoisomeric transition states makes interpretation difficult. In a conceptually similar study Goldberg and Sahli72 obtained optically active 4-methylcyclohexene in up to 70 % optical yield from the pyrolysis of (R)-( +)-trans- and (S)-(-)-trans-4-methylcyclohexyl p-tolyl sulphoxide (Scheme). The two low-energy transition states (50) and (51) which successfully correlate the configuration of the predominant enantiomer of the olefh with that of the sulphoxide are considered to be those in which the bulky p-tolyl group is positioned away from the hydrogen atoms of the cyclohexane ring. Pyrolysis of optically active N-methyl-N-(4-methylcyclohexyl)-N-pheny- lamine oxide also gives optically active 4-methyl~yclohexene.~~ Further aspects of the molecular dissymmetry of trans-cyclic olefins have been demonstrated by the preparation of optically active trans-cyclo-octene by an asymmetric Hofmann degradation of (-)-N-n-butyl-N-isobutyl-N-methylcyclo-octylammonium hydroxide; the normal Hofmann procedure i.e.pyrolysis of the hydroxide gave predominantly the (- )-enantiomer whereas treatment of the corresponding perchlorate with potassium amide in liquid ammonia gave the ( +)-enanti~mer.~~ The preparation of cis trans-l,5-cyclo-octadiene by an asymmetric Hofmann degradation has also been reported.75 5 Asmmyetric Synthesis of Sulphoxides The preceding discussion was illustrated primarily with examples in which the relative and absolute configurations of the molecules were well established by independent methods.Aysmmetric synthesis of sulphoxides differs notably in that much speculative work had been done before the absolute configurations of simple sulphoxides were known. There was ample scope therefore for pre- dictions based on empirical rules. The asymmetric oxidation of optically active methionine (54) was fist de- scribed by L a ~ i n e . ~ ~ The different methods used to introduce the second asym- metric centre (at sulphur) gave products (55) of different optical purities and Lavine concluded that this was due to the formation of different proportions of the two diastereoisomeric methionine sulphoxides. Cram and Pine7’ applied this technique to the determination of the configurations of the diastereoisomeric sulphoxides produced in the oxidation of (R)-( -)-2-octyl phenyl sulphide (56) 73 S.I. Goldberg and M. S. Sahli J . Org. Chem. 1967 32 2059. 73 G. Berti and G. Bellucci Tetrahedron Letters 1964 3853. 74 A. C. Cope W. R. Funke and F. N. Jones J. Amer. Chem. SOC. 1966 88,4693. 75 A. C. Cope C. F. Howell and A. Knowles J . Amer. Chem. SOC. 1962 84 3190. 76 T. F. Lavine J . Biol. Chem. 1947 169 477. ” D. J. Cram and S. H. Pine J . Amer. Chem. SOC. 1963 85 1096. 117 Asymmetric Synthesis with t-butyl hydroperoxide using as a basis the argument that oxidation occurs more rapidly on the diastereoisomericface of sulphur which has the electron pair situated between the hydrogen and n-hexyl groups of the adjacent asymmetric carbon atom (cf.Cram's rule for additions to carbonyl compounds). According to this reasoning the predominant sulphoxide (57) has the R-configuration at sulphur. Montanari and his c o - ~ o r k e r s ~ ~ oxidised o-(methy1thio)benzoic acid esters (59) of optically active alcohols and obtained the optically active sulphoxide (60) after removal of the optical activating group by hydrolysis. 0 Optical purity (%) Configuration R = (-)-menthy1 6.2 (S)-( - ) R = (+)-methylmesitylmethyl 20-5 (W-3 R = (-)-methyl-a-naphthylmethyl 8.9 (R)-( +) The absolute configuration of the predominant enantiomer of the product was predicted by assuming that the most probable conformation of the sulphide was PhCqH (61) in which the carbonyl oxygen is placed in the least hindered region with respect to the gIoups S M and L and that peroxy-acid attack occurs pre- ferentially from the side remote from the large group (cf.Prelog's rule for the atrolactic acid synthesis). Montanari and his c o - w ~ r k e r s ~ ~ ~ ~ also suggested a model which may be used to predict the stereochemical result of the oxidation of non-dissymmetric sulphides with dissymmetric peroxy-acids of known absolute configuration. Thus oxidation of n-alkyl phenyl sulphides with ( +)- 'a A. Maccioni F. Montanari M. Secci and M. Tramontini Tetrahedron Letters 1961 607. 'O A. Mayr F. Montanari and M. Tramontini Gazz. Chim. Ital. 1960 90 739; see also K. Balenovic N. Bregant and D. Francetic Tetrahedron Letters 1960 20. 118 Boyd and McKervey peroxycamphoric acid gave ( +)-n-alkyl phenyl sulphoxides (62) of low optical purity a similar reaction with t-butyl phenyl sulphide gave (-)-t-butyl phenyl sulphoxide (63).According to the authors the attack of peroxy-acid on sulphur occurs in a direction perpendicular to the C-S-C plane of the sulphide and the two most probable transition-state conformations are the staggered arrangements .. (64) and (65) in which the order of decreasing effective size of the groups attached to sulphur is t-butyl > phenyl > n-alkyl. This interpretation predicts the R- configuration for (62) and the S-configuration for (63). The question of the absolute configuration of sulphoxides was finally settled by Mislow et aLsousing a combination of X-ray analysis and optical rotatory dispersion. It was estab- lished that ( +)-alkyl aryl sulphoxides have the R-configuration.This conclusion was significant because Cram and Pine had assigned the R-configuration to the (-)-alkyl aryl sulphoxide (57) whereas Montanari had (correctly) assigned the R-configuration to (+)-alkyl aryl sulphoxides. Mislow points out that in using asymmetric synthesis to predict absolute configuration one always has a 50% chance of being correct and even though Montanari’s model gives the correct answer the arguments used in constructing the model are not necessarily correct. Mislow further contends that these models do not provide a qualitative estimate of the non-bonded interactions which indicate the diastereoisomeric transition state with the lower free energy because (a) there is no way of knowing the preferred arrangements of the peroxy-acid and the sulphide at the transition state (b) the r6le of the sulphoxide oxygen and of the lone pair on sulphur in conformational analysis is not well understood (c) a correlation between the stereochemical result cind the model is by no means unambiguous insofar as different models may lead to the same result and (d) ‘Conformational rules empirically derived from one type of system may not be legitimately extrapolated and transferred to another.’*O Mislow and his co-workerssl have shown that configurational assignments based on the transition-state models of Montanari fail in the case of n-alkyl benzyl sulphoxide; MontanariS2 claims that these results 8o K.Mislow M. M. Green P. Law J. T. Melillo T. Simmons and A. L. Ternay Jr. J . Amer. Chem. SOC. 1965 87 1958. 81 K. Mislow M. M. Green and M. Raban J . Amer.Chem. Soc. 1965 87 2761. 82 F. Montanari Tetrahedron Letters 1965 3367. 119 Asymmetric Synthesis are difficult to interpret and that solvent effects may change the sign of rotation and the optical rotatory dispersion curves of these sulphoxides. 6 Other Examples of Asymmetric Synthesis A. Intramolecular Rearrangements.-The transfer of asymmetry from one asym- metric centre to another via the intramolecular route has proved a valuable method for elucidating transition-state topology. This method developed mainly by Hill has been applied to the Cope rearrangement (66) (optical yield Me- Me\ I . - (s) - (+) - cis Me dWMe Me gMe 13% H (R) - (+) - trans P h G M e - Ph-C-H I 1 C-0 Me OH Me k Me Me (69) 120 Boyd and McKervey 94-96 %),83a the Stevens rearrangement (67) (optical yield probably high),83b the Claisen rearrangement (68) (optical yield not reported),83c and the intra- molecular 1,5-hydride transfer (69) (optical yield 15 %),83d the absolute con- figurations of the reactants and products having been independently determined in each case.The very high optical yield and the almost exclusive formation of the (S)-( +)-cis- and the (It)-( +)-trans-dienes from (66) not only confirm the concerted cyclic mechanism for the rearrangement but also indicate that the transition states have the chair conformations shown (the alternative boat con- formations predict the opposite stereochemical results). A transition state with cis phenyl and vinyl groups is required to explain the stereochemistry of rearrangement (67) and the conversion of the (R)-( +)-vinyl ether =into the (R)-( -)-aldehyde is consistent with the cyclic mechanism suggested for the Claisen rearrangement (68).The formation of the (S)-( +)-ketone in the acid-catalysed rearrangement (69) is in agreement with the suggestion that the transition state adopts a chair conformation in which the phenyl group at C(6) and the methyl group at C(2) occupy equatorial positions (cf. the reduction of ketones with optically active Grignard reagents). B. Asymmetric Catalysis.-Several attempts have been made to achieve asym- metric synthesis with use of non-enzymic homogeneous and heterogeneous asym- metric catalysts. The asymmetric hydrogenation of methyl acetoacetate with Raney nickel modified with tartaric acid has been reported.84 Examples from the field of homogeneous catalysis include the asymmetric oxidation of benzyl methyl sulphide by iodine in the presence of ( +)-2-methyl-2-phenylsuccinic acid.86 The low optical purity (6%) of the sulphoxide produced may be compared with the much higher optical purities (up to 100%) obtained in the oxidation of sulphides in the presence of growing serobic cultures of Aspergihs niger.88 Sheehan and Hunneman8' observed asymmetric synthesis using ( +)-N-[2- phenyl-2-(cyclohexanecarboxy)ethyl]-4-methylthiazolium bromide (70) as a catalyst in the benzoin condensation.Methylphenylketen reacts with methanol or ethanol in the presence of optically active bases such as brucine acetylquinine or acetylquinidine to give optically active a-phenylpropionate esters.88 The 83 (a) R. K. Hill and N. W. Gilman Chem. Comm. 1967,619; (b) R.K. Hill and T-H. Chan J . Amer. Chem. SOC. 1966 88 866; (c) R. K. Hill and A. G. Edwards Tetrahedron Letters 1964 3239; (d) R. K. Hill and R. M. Carlson J . Amer. Chem. SOC. 1965,87,2772. 84 Y. Izumi S. Tatsumi and M. Imaida Bull. Chem. SOC. Japan 1966 39 2223. 85 T. Higuchi I. H. Pitman and K. H. Gensch J . Amer. Chem. SOC. 1966 88 5676. 86 B. J. Auret D. R. Boyd and H. B. Henbest Chem. Comm. 1966 66. 87 J. C. Sheehan and D. H. Hunneman J . Amer. Chem. SOC. 1966 88 3666. 88 H. Pracejus Annalen 1960 634 9. 121 Asymmetric Synthesis stereoselectivity of this reaction was shown to be temperature-dependent. For example (R)-(-)-methyl a-phenylpropionate was obtained in up to 74 % optical yield from methylphenylketen and methanol in the presence of acetylquinine at - 110" at -60° the (S)-( +)-ester was the dominant product (optical yield 20%) and at 25" the optical yield approached zero. Pracejd8 suggests that complex formation between the keten and the amine is rate-determining at low temperatures whereas at higher temperatures the reaction is uncatalysed and therefore non-stereoselective. Grimshaw and co-w~rkers~~ have reported the asymmetric synthesis of 3,4-dihydro-4-methylcoumarin by electro-chemical reduction of 4-methylcoumarin in the presence of certain alkaloids. (D.R.B.) gratefully acknowledges an S.R.C. Postdoctoral Fellowship. 80 R. N. Gourley J. Grimshaw and P. G. Millar Chern. Cornrn. 1967 1278. 122

ISSN:0009-2681    

DOI:10.1039/QR9682200095

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Interaction of Aromatic Nitro-compounds with Bases By E. Buncel A. R. Norris and K. E. Russell QUEEN’S UNIVERSITY KINGSTON ONTARIO CANADA 1 Introduction Aromatic nitro-compounds interact with a variety of Lewis and Br~nsted bases to give typically brightly coloured solutions. The nature and the structure of the coloured species has been under investigation since before 190O1 and during the intervening years several specific types of interaction have been recognised. For the present purpose the historical approach will not be taken; rather the various interactions are treated according to the degree to which the base participates through its unshared electron pair with the nitro-compound. A partial transfer (e.g. through orbital overlap) of electronic charge from the base (Y or Y-) to the aromatic nucleus depleted ofv-electron density owing to the electronegative nitro-substituents gives rise to n-complexes (l) known as donor-acceptor or charge-transfer complexes after Mulliken.2 A further degree of interaction may lead to an electron from Y- becoming completely transferred to the nitro-compound in which case a radical-anion (2) is produced.In contrast to electron-transfer processes is the interaction by which the un- shared electron pair of Y- is used in formation of a covalent bond to an aro- matic carbon atom. The resulting species (3) is then a a-complex and will no longer have the benzenoid resonance intact. The best known of sucha-complexes is the red ‘Meisenheimer adduct’ (3; Y = OCH,) formed from 2,4,6-trinitro- anisole and methoxide ion in methan01.~ If the aromatic nitro-compound con- tains a displaceable group (e.g.halogen) then the base may act as a nucleophile and we enter the area of nucleophilic aromatic substitution. As a further possible interaction the basic reagent may take part in proton-abstraction processes. This possibility arises owing to the powerful electron-withdrawing capability of nitro-groups which may cause a hydrogen atom of the aromatic nitro- compound to become sufficiently acidic. Abstraction of a nuclear hydrogen would yield an aryl carbanion (e.g. 4) while abstraction of a hydrogen bound to an alpha carbon atom would give rise to a nitrobenzyl anion (e.g. 5). An enumeration of the various possible interactions is however only the first step in any consideration of these processes. From a practical viewpoint it would be important to know just which type of interaction would predominate in a given system and to have available definitive criteria for each interaction.One may ask some specific questions. (a) How is the nature of the species formed dependent on the number of nitro- (and other) substituents present? (b) For a given nitro-compound does the base have any influence on the outcome (e.g. C. A. Lobry de Bruyn Rec. Trav. chim. 1890 9 208; 1901 20 120; C. L. Jackson and W. F. Boos Amer. Chem. J. 1898 20,444. * R. S. Mulliken J . Amer. Chem. SOC. 1952 74 811. a J. Meisenheimer Annalen 1902 323 205. 123 The Interaction of Aromatic Nitro-compounds with Bases HO-/H,O as opposed to ButO-/ButOH)? (c) Does the nature of the medium have any effect (e.g. CH,O-/CH,OH as opposed to CH,O-/CH,.SOCH,) ? ( d ) Is it possible for more than one species (1-5) ti be present in a given system of a nitroaromatic compound and base? (e) What are the intimate mechanisms of the processes leading to the species (1)-(5)? (f’) Is there any relationship between rate processes (e.g.in nucleophilic aromatic substitution) and equili- brium processes [e.g. in formation of adduct (3)]? Another issue may be raised which has been a source of considerable con- fusion in the interpretation of these interactions. Paradoxically the cause of this confusion has been the most characteristic aspect of the interactions namely the colour of the resulting solutions. Colour formation has at various times been cited as evidence for the presence of all the species (1)-(5). A further point which needs critical examination is the relationship if any between the isolation and characterisation of a coloured adduct from the reaction of a nitro-compound with base and any claim that that adduct was directly responsible for the colour of the solution.This point applies particularly to the situation where more than one species is present in a given system. Evidence relating to various aspects of these interactions is discussed in this Review. The charge-transfer interaction is considered when there is evidence that it operates simultaneously with one of the other interactions but is not treated as a separate topic. This is in part due to the fact that charge-transfer complexes have been treated in detail re~ently.~ The first topic discussed is the formation of o-complexes and the related evidence from electronic absorption and nuclear magnetic resonance spectroscopy with some emphasis given to equilibrium constants and structural relationships.Proceeding to rate processes we consider nucleophilic aromatic substitution with emphasis on the evidence for a-complex intermediates and reactivity relationships. Proton-transfer pro- cesses are considered next and finally electron-transfer processes and the observation of radical anions by electron spin resonance spectroscopy. 2 Evidence from Ultraviolet-Visible Spectroscopy A characteristic feature of the complexes formed in the initial reversible inter- actions of aromatic nitro-compounds and a variety of bases is their absorption * (a) G. Briegleb ‘Elektronen-Donator-Acceptor Komplexe’ Springer-Verlag Berlin 1961 ; (b) Angew.Chem. Internat. Edn. 1964 3 617; (c) J. N. Murrell Quart. Rev. 1961 15 191; (d) L. J. Andrews and R. M. Keefer ‘Molecular Complexes in Organic Chemistry’ Holden- Day San Francisco 1964; (e) E. M. Kosower Progr. Phys. Org. Chem. 1965,3,81. 124 Buncel Norris and Russe 11 of light in the visible region. Among the systems investigated are (i) 3,3,5- trinitrobenzene and aliphatic amines in chloroform 1,4-dioxan and ethanol ; 5 9 6 (ii) 1,3,5-trinitrobenzene and a series of 1 -X-2,4,6-trinitrobenzenes in liquid ammonia;' (iii) a series of l-alkoxyl-2,4,6-trinitrobenzenes with alkoxide ions in a number of non-aqueous solvents;s (iv) 1,3-dinitrobenzene a series of l-X-2,4-dinitrobenzenes and 1,3,5-trinitrobenzene with acetone in basic acetone- water solution^;^-^^ and (v) 1,3,5-trinitrobenzene with a variety of anions in different solvents.In this section attention is focused upon the spectral proper- ties and equilibrium constants associated with the formation of the latter group of complexes. Representative data in Table 1 refer to the initial 1 :1 complexes formed at 25". Similar spectral properties are displayed by the complexes formed in systems (I) (ii) and (iii). Except for the spectra of the 1,3,5-trinitrobemene-hydroxide ion and the 1,3,5-trinitrobenzene-sulphite ion complexes in aqueous solution which possess a single sharpmaximum plus a shoulderon the long-wavelength side of this maxi- mum all the spectra display two distinct maxima in the 400-850 m,u region. The molar extinction coefficient ( E ) of the high-energy band typically a factor of from 1.3 to 2.5 times larger than that of the low-energy band is characterised by a value in the range 2-4 x lo4 1.mole-l cm.-l. These values of E are much higher than those associated with the charge-transfer complexes of aromatic nitro-compounds with aromatic amines and hydrocarbon^.^^ For a given 1,3,5-trinitrobenzene-anion complex the positions of maximum absorption the energy separation of the absorption maxima and the molar extinction coefficients at the positions of maximum absorption depend markedly on the solvent employed in studying the reaction. Unfortunately few systematic investigations have been carried out in connection with these aspects of the equilibrium reactions. Foster and Fyfea report that in the alkoxide ion-picryl ether interactions there is a bathochromic shift of the high-energy maximum and a hypsochromic shift of the low-energy band on altering the solvent from an ether to a more ionising solvent.The same effect has been noted in connection with the 1,3,5-trinitrobenzene-cyanide ion interacti~nl~~~ b although a simple relationship between the position of maximum absorption and a single solvent parameter such as its dielectric constant has not been observed. Much the same behaviour has been reported by Kimura Kawata and Nakadatelj in connection with the acetonate ion complex of 1,3-dinitrobenzene in a variety of solvents R. Foster J . Chem. SOC. 1959 3508. R. Foster and R. K. Mackie Tetrahedron 1961 16 119. ' R. Foster and R. K. Mackie Tetrahedron 1962 18 161. R. Foster and C. A. Fyfe Rev. Pure Appl. Chem. (Australia) 1966 16 61.'T. Canback Farm. Revy 1949 48 217. lo M. J. Newlands and F. Wild J . Chem. SOC. 1956 3686. l1 S. S. Gitis and A. Ya. Kaminskii J . Gen. Chem. (U.S.S.R.) 1960 30 3771. l2 R. Foster and R. K. Mackie Tetrahedron 1962 18 1131. l3 R. J. Pollitt and B. C. Saunders J . Chem. SOC. 1965 4615. l4 (a) A. R. Norris Canad. J. Chem. 1967 45 2703; (b) Ph.D. thesis Chicago 1962; (c) A. R. Norris and W. Proudlock unpublished results. l5 M. Kimura M. Kawata and M. Nakadate Chem. and Ind. 1965 2065. 125 2 CI h o\ Spectral properties and formation constants of some 1,3,5-trinitrobenzene-anion complexes Anion HO- CH30- CH3CH20- CH,CH,S- CH3COCH2- C,H5COCH2- SO3- CN- CH3NH- gH loN- Solvent Water Methanol Ethanol Dime t h yl formami de Acetone Ethanol Water Chloroform Dimethyl sulphoxide Acetonitrile A (mt-4 450 425 424 465 464 464 462 437 452 444 A2 (mI-4 475 (sh) 495 497 570 572 565 525 (sh) 555 528 521 €1 (1.mole-l cm.-l) 35,000 3 1,200 33,900 29,900 23,100 23,500 39,900 32,000 33,000 - (2.05)* €2 (1. mole-l cm.-l) 21,100 22,800 14,750 10,800 12,050 22,500 20,000 - - - K (1. mole-l) 15-4 2.7" 2O7Oc d - e - -f 2679 33,500h 2,000i 0.05j a T. Abe Bull. Chem. SOC. Japan 1960,33,41; b V. Gold and C. H. Rochester J . Chem. SOC. 1964 1692; C A. R. Norris Ph.D. Thesis Chicago 1962; d R. J. Pollitt and B. C. Saunders J . Chem. SOC. 1965,4615; e S. S. Gitis and A. Ya. Kaminskii J. Gen. Chem. U.S.S.R. 1963 33 3226; f M. Kimura M. Kawata and M. Nakadate Chem. and Ind. 1965 2065; g A. R. Norris Canad. J . Chem. 1967 45 175; h A. R. Norris Canad. J . Chem. 1967 45 2703; * M.R. Crampton and V. Gold J . Chem. SOC. (B) 1967,23;J G. Briegleb W. Liptay and M. Cantner 2. phys. Chem. (Frankfurt) 1960 26 55. * Ratio of to e2 at 25"c. Buncel Norris and Russell though these authors indicate that there exists no solvent effect on the position of maximum absorption in the case of the 1,3,5-trinitrobenzene-acetonate ion complex. Significantly perhaps the absorption maximum of the 2,4-dinitro- phenylacetonate ion shifts very little on changing the s01vent.l~ Studies have been carried out on the effect of solvent on the equilibrium constant ( K ) the heat of formation and the molar extinction coefficient ( E ) of the resulting complex in the reaction of 1,3,5-trinitrobenzene with cyanide ion.16 The solvent has a major effect upon the equilibrium constant; variations in K of lo4 are noted on changing the medium from methanol to acetone or dimethyl sulphoxide.For the same solvents the variation in E is less than two- fold with E being smaller in methanol. The equilibrium constant for formation of the 1,3,5-trinitrobenzene-methoxide ion complex is about 103 times greater in 1,4-dioxan and 1,2-dimethoxyethane than in methan01.l~~ Similar solvent effects have been noted in related studies on the 2,4- and 2,6-dinitroani~oles.~~ The narrow range over which the absorption maxima for the 1,3,5-trinitro- benzene-anion complexes occur and the similarity in their band profiles and molar extinction coefficients have been taken as evidence that the absorbing species have in all cases an essentially identical structure.14 One could argue perhaps that a similarity in the positions of absorption maxima might arise because all the anions had essentially the same ionisation potential.This seems unlikely in view of the widely varying contact-charge-transfer-to-solution bands exhibited by these anions.l8 The near-constancy of the absorption maxima and molar extinction coefficients of the 1,3,5-trinitrobenzene complexes with different anions is in marked contrast to the large variations in wavelengths of absorption maxima and molar extinction coefficients which occur in the 1,3,5-trinitro- benzene-aromatic hydrocarbon and 1,3,5-trinitrobenzene-aromatic amine complexes4a and suggests a different mode of complex formation in the two cases. Because of the close correspondence in the visible absorption spectra of the 1,3,5-trinitrobenzene-ethoxide ion and the 2,4,6-trinitroanisole-ethoxide ion complexes in solution the former was formulated as a a-complex in which the ethoxide ion is covalently bonded to a ring carbon atom and the net negative charge is shared among the three nitro-groups.lg This formulation has received strong support as a result of comparisons of the infrared spectra of the solid 1,3,5- trinitrobenzene-methoxide and the 2,4,6-trini troanisole-methoxide addi- tion products20a~20b and the nuclear magnetic resonance (n.m.r.) spectra of the same products dissolved in methano121,22 and in dimethyl s~lphoxide.~~ l6 E.Buncel A. R. Norris W. Proudlock and K. E. Russell unpublished results. l7 R. J. Pollitt and B. C. Saunders J. Chem. SOC. 1964 1132. l9 R. Foster Nature 1959 183 1042. 2o (a) R. Foster and R.K. Mackie J. Chem. SUC. 1963 3796; (b) R. Foster and D. L1. Ham- mick ibid. 1954 2153. M. R. Crampton and V. Gold J. Cliem. SOC. 1964 4293. 22 M. R. Crampton and V. Gold Chem. Comm. 1965,256. 23 K. L. Servis J. Amer. Chem. SOC. 1965 87 5495. R. J. Marcus Science 1956 123 399. 127 The Interaction of Aromatic Nitro-compounds with Bases As a result of the close similarity in the absorption properties of the 1,3,5- trini tro benzene-methoxide ion complex and other 1 3,5- trini tro benzene-anion complexes all the anions listed in the Table are thought to form a-complexes. Support for this formulation has been advanced for the lY3,5-trinitrobenzene- sulphi te the 1 3,5-trini trobenzene-cyanide ion and the 1,3,5-trini trobenzene- acetonate ion complexes on the basis of infrared s t ~ d i e s ~ ~ ~ y ~ ~ - ~ ~ and for the 1,3,5-trinitrobenzene-piperidine anion complex on the basis of n.m.r.and conductivity meas~rements.~~ The n.m.r. spectrum of the 1,3,5-trinitrobenzene- sulphite complex is also in accord with this form~1ation.l~~ Alternative structures of some of these 1,3,5-trinitrobenzene-anion complexes have however been advanced. Ainscough and Caldin,28 for instance formulated the 1,3,5-trinitrobenzene-ethoxide ion complex as a charge-transfer complex of unspecified geometry. Pollitt and Saunders13 suggest that the complex between 1,3,5-trinitrobenzene and acetonate ion contains that ion situated above one of the ring carbon atoms and donating a lone pair on the methylene group to the lowest availablen* orbital of the benzene ring to form a dative a bond.In this way the hydrogen atom at the ring position concerned is still in or near its original position and the basicn-structure is not grossly disturbed. Gitis Okseng- endler and K a m i n ~ k i i ~ ~ suggest that in the interaction of 1,3,5-trinitrobenzene with acetonate ion the absorption bands with maxima at 570 462 and 518 mp are associated with the mono- di- and tri-salts of acetonate ion with 1,3,5- trinitrobenzene. This formulation certainly seems incorrect ; Kimura Kawata and Nakadate15 separated the 1 :1 addition complex of 173,5-trinitrobenzene- acetonate ion and showed that it had in ethanol absorption maxima at 463 mp ( E = 27,600) and 562 mp (E = 13,700). The complex resulting from the inter- action of iodide ion and 1,3,5-trinitrobenzene has been formulated as a charge- transfer c ~ m p l e x ~ ~ ~ ~ a ‘Meisenheimer-like’ adduct ,140 and as an adduct involv- ing co-ordination of the iodide ion with a nitrogen of the nitro-gro~p.~~ In the case of iodide and possibly other anions the solvent in addition to influencing the values of K E and Amax may well determine whether 0- or wcomplex formaiion is the predominant reaction.For l-substituted 2,4,6-trinitrobenzenes complex formation may occur at both the 1- and the 3-position either preferentially or simultaneously and the absorption spectra of the two complexes may differ appreciably. In the inter- action of 2,4,6-trinitroanisole with methoxide ion attack occurs first at the 3-positionZ3 though the final product of the reaction contains methoxide ion 24 R. A. Henry J . Org. Chem. 1962 27 2637.25 R. C . Farmer J . Chem. SOC. 1959 3425 26 L. K. Dyall J . Chem. SOC. 1960 5160. 27 A. Ya. Kaminskii and S. S. Gitis J. Gen. Chem. (U.S.S.R.) 1964 34 3794. 28 J. B. Ainscough and E. F. Caldin J. Chem. SOC. 1956 2540. 29 S. S. Gitis G. M. Oksengendler and A. Ya. Kaminskii J . Gen. Chem. (U.S.S.R.) 1959 29 2948. 30 G. Briegleb W. Liptay and R. Fick Z . Elektrochem. 1962 66 851. 31 G. Briegleb W. Liptay and R. Fick 2. phys. Chew (Frankfurt) 1962 33 181. 32 A. B. Tronov J. Gen. Chern. (U.S.S.R.) 1965 35 1549. 128 Buncel Norris and Russell in the 1 - p o s i t i ~ n . ~ ~ y ~ ~ Unfortunately visible absorption measurements corre- sponding to these two a-complexes are not yet available. The observation that the cyanide complexes of 2,4,6-trinitrotoluene picryl chloride 2,4,6-trinitro- benzaldehyde and 2,4,6-trinitroanisole have absorption maxima at wave- lengths very little different from those of the 1,3,5-trinitrobenzene-cyanide ion complex may well indicate attack of cyanide ion at the 1-position in all cases.33 Consideration of the ultraviolet and visible absorption results so far reported for the interaction of aromatic nitro-compounds with both aliphatic amines and anions suggests that most of the data are associated with the formation of (T and Meisenheimer complexes.Few spectral data are available for aryl carbanion species e.g. (4) anion-nitroaromatic charge-transfer complexes (except 1,3,5- trinitrobenzene-iodide in certain solvent^),^^,^^ nitroaromatic radical anions and nitrobenzyl anions. The spectral properties of the latter two classes of compound are referred to later.3 Evidence from Nuclear Magnetic Resonance Spectroscopy Application of the n.m.r. method to the problem of the interaction of aromatic nitro-compounds with bases has provided the most direct evidence on the structures of the species that are present in solution. While some of the resulting evidence has confirmed previous thinking new facets to the problem have also been revealed. The n.m.r. spectra of the species resulting from the interaction of 1,3,5- trinitrobenzene and a variety of anions and aliphatic amines in a number of can be rationalised in terms of the formation of o-complexes of 1 :1 stoicheiometry (6) but are not explicable on the basis of a wcomplex formulation. The ring protons (Ha) absorb at low fields (6 = -8.9 to -8.7 p.p.m. doublet J = cn.1 c./sec.) while the hydrogen bonded to the carbon at which attack of the anion takes place (Hp) absorbs at higher fields (6 = -6.10 to -6.30 p.p.m. broad triplet J = ca. 1 c./sec.). The two resonances are of 33 W. Depew and A. R. Norris unpublished results. 34 M. R. Crampton and V. Gold Chem. Comm. 1965 549. 35 R. Foster and C. A. Fyfe Tetrahedron 1965 21 3363. 36 M. R. Crampton and V. Gold J . Chem. SOC. (B) 1966 893. 37 R. Foster and C . A. Fyfe J . Chem. SOC. (B) 1966 53. 129 The Interaction of Aromatic Nitro-compounds with Bases relative intensity 2:l. The positions of the resonance lines particularly the resonance line of H8 and the magnitudes of the spin-spin coupling constants of the ring protons seem to be solvent- and anion-dependent. For all the anion- 1,3,5-trinitrobenzene interactions studied so far n.m.r.spectra due to both species are observed in solutions containing both the complexed and uncom- plexed 1,3,5-trinitrobenzene implying that anion exchange between different 1,3,5-trinitrobenzene units is not a rapid process. The shifts of the resonance lines upfield on going from the parent aromatic molecule (6 = -9.2 p.p.m.) to the complex are said to be due to larger screening of the protons in the anionic s p e ~ i e s . ~ ~ ~ ~ Only in the case of the interactions of lY3,5-trinitrobenzene with aliphatic amines have simultaneous measurements of both n.m.r. and visible spectra been made as a function of the variation in the concentration of both component^.^^^^ Despite this the species in solution giving rise to the visible absorption is thought in all cases to be the same as that giving rise to the n.m.r.spectrum. At ratios of anion to 1,3,5-trinitrobenzene greater than unity the addition of a second anion may occur. This has been demonstrated only in the case of the interaction of methoxide with 1,3,5-trinitroben~ene.~~,~~ Proton loss may be occurring simultaneously at these higher methoxide ion-l,3,5-trinitrobenzene ratios as the 1,3,5-trinitrobenzene resonance becomes broader with increasing concentration of base without corresponding broadening of the lines due to complex.22 The reaction of 1-substituted 2,4,6-trinitrobenzenes with bases is slightly more involved as either a a-complex of the general type suggested by Meisen- heimer (7) or a o-complex formed by attack at C(3) (8) may be obtained.If in addition the substituent contains one or more acidic protons ionisation to the conjugate base e.g. (9) may occur in competition with a-complex formation and the conjugate base may itself undergo anion attack (10). Attack of the base at a carbon containing a nitro-group is not considered to be a likely process. In the interaction of 2,4,6-trinitroanisole with methoxide ion in dimethyl sulphoxide methoxide ion first adds at the 3-position of 2,4,6-trinitroanisole (8; R1=R2=OCH3) but this anion is unstable and transforms into a complex with the methoxide ion attached at the 1-position (8).353,40 Evidently addition of methoxide ion to an unsubstituted position to form the thermodynamically less stable product is more rapid than addition to a position carrying a methoxyl group to form the thermodynamically more stable product.This is contrary to what would be expected on the basis of Miller’s calculation^.^^ The stable 1 :1 adducts formed by addition of methoxide ion to the 1-position of 2,4,6-trinitroanisole have also been reported in acet~nitrile~~ and [2H6]- dimethyl sulphoxide.42 Addition of dimethylamine diethylamine or a i d e ion 88 M. R. Crampton and V. Gold J. Chem. SOC. (B) 1967 23. 39 K. L. Servis J. Amer. Chern. SOC. 1967 89 1508. 40 R. Foster C. A. Fyfe P. H. Emslie and M. I. Foreman Tetrahedron 1967 23 227. *l J. Miller J . Amer. Chem. SOC. 1963 85 1628. 4a P. Caveng P. B. Fischer E. Heilbronner A. L. Miller and H. Zollinger Helv. Chin?. Acta 1967 50 848. 130 Buncel Norris and Russell to 2,4,6-trinitroanisole in dimethyl sulphoxide yields the 1 -substituted Meisen- heimer complexes with no evidence for preliminary complex formation at the 3 -posit ion .39 sg3 In picramide and N-alkylpicramides ionisation to the conjugate base (9) occurs in competition with a-complex formation at C(3) (lo) the relative amounts depending on the structure of the the alkoxide ion em- ployed and the concentration of alcohol in the m e d i ~ m .~ ~ ~ ~ ~ In the series picra- mide N-methylpicramide and N-phenylpicramide the ratio of conjugate base to a-complex increases in the order -NH, -NH*CH, -NH.C,H when inter- action with methoxide ion is studied in dimethyl ~ u l p h o x i d e . ~ ~ ~ ~ ~ A second methoxide ion adds to the conjugate base of N-methylpicramide at high meth- oxide to N-methylpicramide ratios.39 In contrast n.m.r. studies suggest that neither the picramide anion nor the picrate ion undergo addition of either methoxide or hydroxide ion even at high anion concentrations.22 This observa- tion may be peculiar to the anions and the solvent systeni used in this st~dy.4~ Addition of methoxide ion to NN-dimethyl- or NN-diethylpicramide yields the 3-substituted a-complex while addition of the corresponding amines to 2,4,6-trinitroanisole yields the 1-substituted Meisenheimer c ~ m p l e x e s .~ ~ * ~ ~ A 2:l adduct with methoxide ion has been observed for NN-dimethylpicramide in dimethyl sulphoxide at high methoxide to dimethylpicramide ratios.36 Hydroxide ion and acetonate ion behave similarly to methoxide ion in their interactions with NN-dimethylpi~ramide.~~~~~ On the basis of n.m.r. studies the formation of benzyl-type anions from the interaction of methoxide ion and 2,4,6-trinitrotoluene in dimethyl sulphoxide- methanol mixtures is considered unlikely.39 However competing reactions at the high methoxide and 2,4,6-trinitrotoluene concentrations required in n.m.r.studies may prevent significant concentrations of 2,4,6-trinitrobenzylanion from being obtained. On the other hand 4-nitrobenzyl cyanide45 and anilines containing two nitro-groups either in the 2,4- or the 2,6-position ionise by proton loss when reacting with methoxide ion in dimethyl sulphoxide solution and the anilines show no evidence of addition reaction^.^^^^^ The interaction of methoxide ion with both 2,4-dinitroanisole or 2,6-&nitro- anisole in dimethyl sulphoxide gives rise to a complex containing the added methoxide ion at the l - p o ~ i t i o n .~ ~ ~ ~ ~ Addition of methoxide ion to 1,5-dimethoxy- 2,4-dinitrobenzene in dimethyl sulphoxide yields a product in which attack of methoxide ion has occurred at the l-po~ition.~~ Addition of acetone gives rise to only a slight shift attributed to the effect of solvent. The similarity in the visible absorption spectra of this adduct and that formed as a result of the interaction of 1,5-diethoxy-2,4-dinitrobenzene and acetonate ion in acetone raises some question about Gitis and KaminsWs formulation of the structure of the acetonate ion adductll [attack at C(3)]. 43 P. Caveng and H. Zollinger Helv. Chim. Acta 1967,50 861. I4 J. Murto Suomen Kern. 1961 B 34,92. O6 M. R. Crampton J . Chem. SOC. (B) 1967 85. 131 The Interaction of Aromatic Nitro-compounds with Bases 4 Janovsky Reaction The interaction of aromatic nitro-compounds with ketones in basic ketone- water solutions the J a n ~ v s k y ~ ~ ~ ~ reaction has attracted the attention of chemists for some time.C a n b a ~ k ~ ~ reviewed the early literature and suggested that for the case of interactions of 1,3-dinitrobenzene the coloured species was the cyclohexadienide anion (a-complex) formed by the attack of the conjugate base of the ketone on the 4-position of 1,3-dinitroben~ene.~~ On this interpretation this reaction is simply a special case of a-complex formation in which the base is an anion of a ketone usually the acetonate ion.50 Subsequent i n v e s t i g a t i ~ n s ~ ~ ~ ~ ~ ~ have offered considerable support for this formulation though questions have arisen concerning (a) the stoicheiometry of the coloured species (b) whether an oxygen atom or a carbon atom in the conjugate base of the ketone is attached to the ring carbon atom,52 (c) whether attachment occurs at both the 2- and the 4-position (in 1,3-dinitrobenzene) either simultaneously or consecutively,26 (d) what effects ring substituents other than hydrogen have on the point of attachment of the ketonate ion7952 and on the absorption spectrum and (e) whether a portion or all of the colour arises from a reduction product of the original ~ o m p l e x .~ ~ ? ~ ~ Despite a great deal of research in which the solid adducts of various ketonate ions and a number of di- and tri-nitro-compounds have been prepared and characterised according to comp~sition,~~*~~ decomposition temperat~re,~~ visible infrared and n.m.r.absorption ~ h a r a ~ t e r i ~ t i ~ ~ ~ and chemical behavio~r,~~3~~ some of these questions remain unanswered. (Demon- stration that the solid adduct which results from the 1,3-dinitrobenzene-acetonate ion interaction contains 1,3-dinitrobenzene and acetonate ion in a 1:l mole ratio and is deeply coloured strongly suggests but in no way proves that the coloured species in solution is the result of 1,3-dinitrobenzene and acetonate ion interacting in a 1 :1 mole ratio. Parallel infrared n.m.r. and visible absorp- tion studies of this complex in solution have not yet been carried out.) In addition the formulation of the Janovsky complex as a direct analogue of the Meisenheimer adducts has been q u e s t i ~ n e d . ~ ~ ~ ~ In the interaction of 1,3- dinitrobenzene with acetonate ion Pollitt and Saunder~~~ suggest that the colour in the system might arise owing to absorption by the conjugate base of 1,3- dinitrobenzene while in the case of the 1,3,5-trinitrobenzene-acetonate ion interaction the coloured species was formulated as a rather unusual a-complex.1° 46 J.V. Janovsky and L. Erb Ber. 1886 19 2155. 47 5. V. Janovsky Ber. 1891 24 971. 48 T. Canback Svensk Farm. Tidskr. 1949 53 151. 4g T. Canback Farm. Revy 1949 48 153. 5o R. Foster and R. K. Mackie Tetrahedron 1963 19 691. 51T. Abe Bull. Chem. SOC. Japan 1959 32 391. 52 S. S. Gitis and A. Ya. Kaminskii Doklady Akad. Nauk S.S.S.R. 1962 144 775. 63 M. Ishidate and T. Sakaguchi J. Pharm. SOC. Japan 1950 70,444. 64 T. J. King and C. E. Newall J. Chem. SOC. 1962 367.66 S. S. Gitis J. Gen. Chem. (U.S.S.R.) 1957 27 1956. 66 S. S. Gitis and A. Ya. Kaminskii J . Gen. Chem. (U.S.S.R.) 1963 33 3226. b7 T. Severin and R. Schrnitz Angew. Chem. Internat. Edn. 1963 2 266. 68 R. J. Pollitt and B. C. Saunders Proc. Chem. SOC. 1962 176. 132 Buncel Norris and Russell The second formulation based on qualitative observations concerning the magnitudes of the formation constants of acetonate and hydroxide ion adducts in dimethylformamide the effect of blocking groups in ring positions and differences in absorption spectra between what are considered to be two different types of complex may well be correct. Reliable values of equilibrium constants absorption maxima and molar extinction coefficients are required for both the ketonate ion complexes and other X- complexes in a variety of solvents in order to support or refute such arguments.It is possible that the solvent employed in such investigations may alter the mode of attachment of an ion to a nitro-compound or preferentially favour one reaction site over another. In addition a change in solvent may affect both the positions of maximum absorption and/or the molar extinction coefficients at these maxima or favour alternate reactions such as free-radical formation or aryl-carbanion formation. In the conditions of di- or tri-nitroaromatic compound in excess over the ketone in basic ethanolic solution (the Zimmerman59 reaction) solutions are produced whose spectral properties change rapidly with time and in which there is the eventual production of the corresponding dinitro- or trinitro-benzyl k e t ~ n e .~ ~ ~ ~ The reaction products arising from the interaction of 1,3-dinitro- benzene with acetonate ion in the early stages of the Janovsky reaction and in the late stages of the Zimmerman reaction are two distinct species as shown by differences in composition decomposition temperature infrared and visible absorption spectra and chemical properties .15p2’ 5 Nucleophilic Aromatic Substitution The present discussion is confined to benzenoid derivatives; for consideration of nucleophilic substitution in the heteroaromatic series the reader is referred to ref. 60. A. ‘Meisenheimer Structures’ as Intermediates in Displacement Reactions.- Evidence was presented in the early 1 9 5 0 ~ ~ ~ 9 ~ ~ that nucleophilic displacement in nitro- cyano- etc. activated aryl halides does not proceed by a synchronous s N 2 mechanism [transition state (1 l)] analogous to aliphatic substitution but that instead a two-stage mechanism obtains with a discrete intermediate [a ‘Meisenheimer structure’ (12)] intervening between reactants and products.The carbon undergoing displacement is tetrahedral in (12) and the negative charge originating from the unshared electron-pair of nucleophile Y is de- localised over the aromatic system. While the importance of charge delocalisa- tion has generally been accepted there has been discussion63 as to whether the cyclohexadienide structure represents a minimum or a maximum on the potential- 69 W. Zimmerman Z. physiol. Chem. 1937 245,47. 6o G. Illuminati Adv. Heterocyclic Chem. 1964 3 285. 61 J. F. Bunnett and R. E. Zahler Chem.Rev. 1951 49 273. 62 J. Miller Rev. Pure Appl. Chem. (Australia) 1951 1 171. 68 R .E. Parker and T 0. Read J . Chem. Soc. 1962 3149. 133 The Interaction of Aromatic Nitro-compounds with Bases energy profile for reaction; in the latter case a synchronous mechanism would obtain and the configuration of the single transition state would be given by (13). on the mechanism of the aromatic displacement process has been the r6le of ‘isolable intermediates’ the stable Meisenheimer adducts formed between alkyl picryl ethers and alkali- metal alkoxides (14; X = OR where R = alkyl). Can species such as (14; X = Hal) be demonstrated to be present in a reacting solution of a picryl halide and alkoxide ion? One possible approach might be to detect (14) by the observa- tion of colour.A rough estimate of the value of E for a complex between picryl halide and methoxide ion is given by the corresponding value for the 1,3,5- trinitrobenzene-methoxide ion complex in methanol.gab If it is assumed that an absorbance value of 0.1 at 495 mp would be detectable to the eye the minimum visual detectable concentration of (14; R=CH, X = Hal) would be 5 x mole 1.-l. In practice coloured solutions are not observed in the course of reac- tion of picryl halides with base?g The formation of coloured solutions in equili- brium processes continues to be used as evidence in support of the intermediate complex mechanism in rate processes. A recent study70 reports the following observations. 1 -Fluoro-2,4-dinitrobenzene and ethyl malonate interact in presence of triethylamine to give a coloured product which is stable in dimethyl- formamide (Amax = 397 and 510 mp) but on addition of water is transformed into the 2,4-dinitrophenylmalonate ethyl ester.The coloured product is taken to be the triethylammonium salt of cyclohexadienide anion corresponding to (12; R = NO2 X = F Y = ma10nate)~O (cf. also ref. 71). The above discussion must be critically examined however for there is no absolute connection between the presence of a species in detectable concentra- tion in a given reaction medium and the hypothesis that such a species is an intermediate along the reaction pathway of a particular rate process. That side equilibria are possible in suitable systems of aromatic nitro-compounds and bases is such n-complexes of the charge-transfer type2s4 are in One of the focal points of J.F. Bunnett Quart. Rev. 1958 12 1. J. F. Bunnett ‘Theoretical Organic Chemistry’ Butterworths Landon 1959 p. 144. J. Sauer and R. Huisgen Angew. Chem. 1960,72,294. e7 S. D. Ross Progr. Phys. Org. Chem. 1963,1 31. 6*V. Gold and C. H. Rochester J. Chem. Soc. 1964 (a) 1687 (b) 1692 (c) 1697 (4 1710 (e) 1727. 6n J. Murto Acfa Chem. Scad. 1966 20 303 310. 70 P. Baudet Helv. Chim. Acta 1966 49 545. 71 R. Bolton J. Miller and A. J. Parker Chern. and Ind. 1960 1026; 1963 492. 72 P. Caveng and H. Zollinger Helv. Chim. A m 1967 50 866. 134 Buncel Norris and Russell contrast to the a-bonded Meisenheimer structures. As an added consideration the wcomplexes themselves usually absorb in the visible region. For these reasons unequivocal evidence for a reaction intermediate such as (12) is sought through kinetic studies by choosing a system in which the reactive intermediate can be partitioned along two or more reaction paths.This can sometimes be accomplished by the addition of reagents which will divert the intermediate in a predictable manner and thereby cause a change in the overall rate expression. Evidence of this nature has been presented for electrophilic aromatic substitution.7s75 B. Kinetic Evidence for a a-Complex 1ntermediate.In recent years the reaction system chosen for the kinetic demonstration of an intermediate in nucleophilic aromatic substitution is that between Z-nitro- or 2,4-dinitro-aryl derivatives and secondary amines (eqn. 1). Owing to the presence of a labile proton in (15) which may be removed by a Bransted base (R2NH CH3C02- OH- etc.) this system lends itself to studies of base catalysis.The removal of the N-H proton which should itself be an easy process provides an alternative pathway of lower activation energy for the conversion of intermediate into products. Since R2N- is a poorer leaving group than R2NH the reversion of intermediate into reactants will occur less readily when proton loss from (15) has been effected and expulsion of X- will correspondingly be more favoured. X H R,NH + There are two essentia1,points to note about eqn. (1) (a) that conversion of intermediate (15) into products occurs by parallel competing paths and (6) that it implies some specific requirements with respect to the occurrence and the form of the base catalysis. The latter point is seen by examination of the kinetic expression (eqn.2) derived on basis of the steady-state treatment. in eqn. (2) the kSB[B] term refers to any base effective in catalysis (including R,NH; if more than one base is effective in a given system then additional such terms would be present). It is instructive to consider two limiting cases (i) when 7a H. Zollinger Experientia 1956 12 165. 7b B. T. Baliga and A. N. Bourns Canad. J. Chem. 1966,44,363,379. E. Grovenstein jun. and N. S. Aprahamian J . Amer. Chem. Soc. 1962,84,212. 135 The Interaction of Aromatic Nitro-compounds with Bases k- << k + k3B[B] then k = k, so that formation of the intermediate is rate- determining and base catalysis should not be observable; (ii) when k- >> k + k3B[B] we have a pre-equilibrium condition in which the product-forming processes are rate-determining and base catalysis is expected.These conditions can now be related with physical systems. Since case (i) will be relatively favoured when k is large it follows that in related compounds base catalysis should be favoured for the one with thepoorer leaving group (e.g. ArF in favour of ArCl). Also if k, k then base catalysis may be observed at low base concentration but will tend to diminish as [B] is increased; hence the sensitivity to base catalysis will decrease with increasing [B] and the overall rate constant will show a curvi- linear dependence on [B]. These predictions with respect to leaving-group and base-concentration dependence thus become a test of the intermediate complex hypo thesis. 7G Base catalysis is in fact observed in reaction of l-fluoro-2,4-dinitrobenzene with certain bases and it is much more effective than with l-chloro-2,4-dinitro- benzene in a particular system.Very marked catalysis by hydroxide and acetate ion is observed in the reaction of l-fluoro-2,4-dinitrobenzene with N-methy- laniline but catalysis is barely detectable with the chloro-compound.7G Similarly the reactions of 4-nitrophenyl phosphate with dimethylamine and with piperidine are catalysed by hydroxide ion.77 The reaction of 2,4-dinitrophenyl phenyl ether78 with piperidine is catalysed by piperidine and by hydroxide ion but when the ease of removal of the leaving group is improved by nitro-substitution (in the 2- and 4-positions) base catalysis is no longer observed.79 (The competing direct displacement of phenoxy- by hydroxide ion accounts for only 2% the total reaction in a medium of 40% dioxane-60% water.This is a result of the low nucleophilic reactivity of hydroxide ion in nucleophilic aromatic substitution.) Further the 4-nitrophenyl phosphate and 2,4-dinitrophenyl phenyl ether pro- cesses show pronounced curvilinear dependence of rate on the concentration of the catalysing base. The observation by Hart and Bournsso of a variable kinetic oxygen ( P / k l s ) isotope effect as a function of hydroxide-ion concentration in the reaction of 2,4-dinitrophenyl phenyl ether with piperidine can only be accommodated by a two-stage intermediate complex mechanism. C. Mechanism of Base Cata1ysis.-While the above observations are in full accord with the requirements of the two-step mechanism it is not clear why some bases are effective in catalysis while others are not.For instance the re- action of l-fluoro-2,4-dinitrobenzene with aniline or with t-butylamine is not catalysed by hydroxide ion.81 Also the reaction of 4-nitrophenyl phosphate with dimethylamine is not catalysed by ~iperidine,~~ even though piperidine acts as a catalyst in the reaction of 2,4-dinitrophenyl phenyl ether with pi~eridine.~~ 76 J. F. Bunnett and J. J. Randall J . Amer. Chem. SOC. 1958 80 6020. 17 A. J. Kirby and W. P. Jencks J . Amer. Chem. SOC. 1965 87 3217. 78 J. F. Bunnett and R. H. Garst J . Amer. Chem. Sac. 1965 87 3879. 79 J. F. Bunnett and C. Bernasconi J . Amer. Chem. SOC. 1965 87 5209. 80 C . R. Hart and A. N. Bourns Tetrahedron Letters 1966 2995. 81 J. F. Bunnett and J. H. Beale personal communication. 136 Buncel Norris and Russell It is difficult to accept the suggestion7' that proton abstraction from (15) should be subject to considerable steric effects.In certain systemss2 there appears to be a trend of a greater sensitivity for base catalysis with decreasing base strength of the reacting amine. The timing of the proton-transfer step between intermediate (15) and base needs consideration. If proton removal were part of the rate-determining step then a kinetic hydrogen-deuterium isotope effects3 should be observable. In practice the overall isotope effects are generally very close to unity (for a sum- mary of earlier work see refs. 67 and 84) and may also take inverse values. For example in the reactions of [arnino-2H&-anisidine with l-halogeno-2,4-dinitro- benzenes in benzene solution,85 k ~ / k ~ varies between 0.80 and 0.94 for the chloro-compound and between 0.95 and 1 a 0 5 for the fluoro-compound depend- ing on base concentration and temperature.Isotope effects of this low magnitude generally signify 'secondary effects's6 and their observation suggests that rupture of the N-H bond is not rate-determining. On the basis of theoretical considerationss7 the view would be taken that proton transfer from a nitrogen base should be a kinetically fast process for the cases considered here.However from a study of the reaction 2,4-dinitrophenyl phenyl ether with [N-,H2]- piperidine in dioxan-D,O catalysed by deuteroxide ion it was deducedso that the rate constant ratio k3H/k3D for conversion of intermediate into products is 1.80 which appears to be outside the range of secondary effects.B -+H Ij+$. . . The mechanism of base catalysis now favoured by most ~ ~ r k e r ~ is an initial equilibrium proton transfer followed by rate-determining general-acid- catalysed removal of the leaving group by BH+ as shown in transition state (16). This mechanism satisfies the requirements of the kinetic equation (2) but it raises certain questions as discussed below. An alternative possibility therefore is (cf. also ref. 80) that removal of the N-H proton occurs simultaneously with weakening of the C-X bond as in transition state (17) which bears resemblance to that in bimolecular (E2) eliminations. This suggestion may better account for 82 G. Becker C. F. Bernasconi and H. Zollinger Helv. Chim. Acta 1967 50 10. 83K. Wiberg Chem. Rev. 1955 55 713.84 H. Zollinger Adv. Phys. Org. Chem. 1964 2 163. 85 C. Bernasconi and H. Zollinger Helv. Chim. Acta 1967 50 3; 1966 49 2570. 86 E. A. Halevi Progr. Phys. Org. Chem. 1963 1 180. M. Eigen Angew. Chem. 1963 75 489. 137 The Interaction of Aromatic Nitro-compounds with Bases a hydrogen-deuterium isotope effect of 1.80 with respect to N-H,*O since hydrogen isotope effects in E2 processes are known to vary considerably in magnit~de.~*,~~ A further difficulty with the mechanism involving transition state (16) [and one which is perhaps met better by the E2 type transition state (17)] is that it would require that electrophilic catalysis (e.g. by Ag+) with respect to the leaving group be of fairly common occurrence in nucleophilic aromatic displacements. Whereas electrophilic catalysis in S N ~ displacement at saturated carbon is well establi~hed,~~ there appears to be only one report of metal-ion catalysis in the aromatic series the reaction of l-fluoro-2,4-dinitroben- zene with thiocyanate ion in methanol is accelerated by a factor of 2000 on the addition of thorium ions.S1 On the other hand a case of bifunctional catalysis in the aromatic series has recently been reported.a-Pyridone (which has previously been shown to be effective in bifunctional catalysiss2) strongly accelerates the reaction of l-fluoro-2,4-dinitrobenzene with piperidine. Transition state (18) is suggested to account for this obser~ation.~~ The possibility that piperidine itself may take part in bifunctional catalysis [transition state (19)] was suggested previ~usly.~~ The problem of the mechanism of nucleophilic aromatic substitution has also been approached through use of molecular orbital theory in the Huckel approxi- m a t i ~ n .~ ” ~ ~ A number of interesting correlations between molecular orbital quantities (rr-electron density localisation energy) and experimental quantities (rates activation energies) are reported but basically the calculations do not allow a clear differentiation between the one-step synchronous and the two- step intermediate complex mechanisms. In conclusion the general evidence supports very strongly the two-stage mechanism of nucleophilic aromatic substitution but there are a number of anomalies which are not yet fully understood. Thus the structure of the rate- determining transition state is not known with certainty. D. Quantitative Approach to Nucleophilic Aromatic Substitution.-A number of different approaches have been used in the quantitative correlation of re- activities in nucleophilic aromatic substitution processes.A thermochemical approach has been developed at length by Miller and his c o - w ~ r k e r s ; ~ ~ - ~ ~ E. Buncel and A. N. Bourns Canad. J. Chem. 1960 38,2457; A. N. Bourns and E. Buncel Ann. Rev. Phys. Chem. 1961,12 1. 8Q J. F. Bunnett Angew. Chem. Internat. Edn. 1962 1 225. Qo C. K. Ingold ‘Structure and Mechanism in Organic Chemistry’ Bell London 1953 p. 357. Q1 K. B. Lam and J. Miller Chem. Comm. 1966 642. 92 C. G. Swain and J. F. Brown J. Amer. Chem. SOC. 1952,74,2538. *a F. Pietra and D. Vitali Tetrahedron Letters 1966 5701. 94 B. Capon and C. W. Rees Ann. Reports 1964 61 278.Q6 C. Parkanyi and R. Zahradnik Coll. Czech. Chem. Comm. 1964 29,973. O6 J. Murto Suomen Kem. 1965 B 38,246. O7 S. Carra M. Raimondi and M. Simonetta Tetrahedron 1966 22 2683. Q8 J. Miller J. Amer. Chem. SOC. 1963 85 1628. OQ J. Miller and K. W. Wong J. Chem. SOC. 1965 5454. l o o J. Miller and K. W. Wong Austral. J . Chem. 1965 18 117. 101 D. L. Hill K. C. Ho and J. Miller J. Chem. SOC. (B) 1966 299. 104 K. C. Ho J. Miller and K. W. Wong J. Chem. SOC. (B) 1966 310. 138 Buncel Norris and Russell the method sets out to calculate activation energies on the basis of the inter- mediate complex mechanism. Transformation of the cyclohexadienide structure (20) into products (23) (or reactants) is considered in terms of stages (21) and (22) and the energies associated with these changes are then estimated.The energy levels of the initial and final states relative to that of the intermediate complex are thus obtained by taking into account terms due to changes in bonding in electron affinity in solvation and in delocalisation energy. The energy levels of the two transition states (24) and (25) for bond formation and bond rupture are related to the intermediate complex by means of an unusual application of the Hammond hypothesisla and are further modified by the inclusion of the ‘a-substituent effect’ term. The latter useful concept is introduced on the basis that a strongly electronegative atom X already attached to the aromatic carbon atom would lower the energy of the transition state for attack by a nucleophile. The greater mobility of fluorine than iodine amounting to 4 kcal./mole in activation-energy difference is thus ascribed to an electro- negativity effect alone.It is clear that this treatmentg* necessarily involves a number of assumptions and approximations since the required energy terms are not available for the actual processes under consideration and must be drawn from other systems. The solvation energy term being taken as an example the data used are heats of hydration and it is assumed that the change to methanol medium would affect all the values equally. The solvation energy of the intermediate complex for all cases of substitution is equated to that of the picrate ion. In view of these and other assumptions used it is remarkable that agreement between calculated and experimental activation energies (where known) is in fact very good (generally within 1-2 kcal./mole).A general cancellation of errors is indicated. As with some other semi-empirical methods of calculation one may take the view that the general applicability of the method to diverse processes is its own justification despite any assumptions that may be used. Some examples of Miller’s method of calculation are given in the Figure they are chosen to illustrate the changes in balancelMJo5 between the situations where formation of the intermediate is rate-determining and where its decom- position is the slow step and the influence upon this balance of factors such as lo3 G. S. Hammond J . Amer. Chem. SOC. 1955 77 334. lo4 G. S. Hammond and L. R. Parks J . Amer. Chem. SOC. 1955,77 340. lo6 J. F. Bunnett E. W. Garbisch,jun. and K. M. Pruitt J .Amer. Chem. SOC. 1957,79 385. 139 The Interaction of Aromatic Nitro-compounds with Bases + 14 - 31.5 + 11.5 +I35 (d Figure ( f Potential energy-reaction co-ordinate diagrams for activated nucleophilic aromatic substitution (energies are shown in kcal. mole-’) (a) 1-Iodo-4-nitrobenzene and CH,O-/CH,OH (b) 1 -Fluoro-4-nitrobenzene and CH,O-/CH,OH (c) l-Fluoro-4-nitrobenzene and CH,S-/CH,OH (d) l-Bromo-2,4-dinitrobenzene and I-/CH,OH (e) 1 -Fluoro-2,4-dinitrobenzene and I-/CH,OH (f) Methyl picryl ether and CH,O-/CH,OH nucleophile and leaving group. Comparison of (a) with (6) in the Figure shows the change from a ‘typical’ substitution in which bond formation is completely rate-deterrnininglo5 which results when the leaving group is changed. The approach to the second transition state becomes virtually rate-determining (Figure c) when additionally the nucleophile is changed to thiomethoxide.loO This tendency of sulphur nucleophileslo6-lo* to raise the energy barrier for bond rupture is predicted to be even more important with larger nucleophiles such as thiophenoxide.lo1,lo2 The iodide-bromide exchange (Figure d ) is predicted to be a slow reversible process; the very high activation energy calculated for iodide-fluoride exchange (Figure e) is in accord with the observed preference for solvolysis in this system.The important case of a stable intermediate complex is shown in (f); the low calculated activation energy is noteworthy (cf. refs. 68a 106 A. J. Parker in ‘Organic Sulfur Compounds’ ed. N. Kharasch Pergamon Press Oxford 1961 vol. 1 p. 103.lo7 J. F. Bunnett and J. D. Reinheimer J. Amer. Chern. SOC. 1962 84 3284. 108 R. F. Hudson and G. Klopman J . Chem. Suc. 1962 1062. 140 Buncel Norris and Russell 109-1 1 1). Miller's discussion of factors determining nucleophilic reactivity in these systems is related to the conclusions drawn from somewhat different points of view by Hudson,l12 Bunnett,l13 and Pearson.ll* 6 Proton-transfer Processes A. Nuclear Hydrogen Abstraction.-The possibility of the occurrence of nuclear hydrogen abstraction in the interaction of aromatic nitro-compounds with bases first arose in connection with the observations that solutions of some aromatic nitro-compounds in amines and in liquid ammonia are conducting. The small conductance of 1,3,5-trinitrobenzene in pyridine was explained by Lewis and Seaborg115 as being the result of partial proton transfer while com- plete ionisation (eqn.3) was postulated116 to take place in 2-aminoethanol on the basis of conductance and cryoscopic data. 2 C,H,,*NH + NO + c,H,,NH,'. (4) In contrast it was concluded117 from conductance and spectroscopic results that 1,3,5-trinitrobenzene interacts with piperidine in acetonitrile solution to give an adduct (eqn. 4). The intense colour and high conductivity of 1,3-dinitro- benzene in liquid ammonia was also explained on the basis of adduct forma- tion;l18 this formulation received support from spectral examination in liquid ammonia of a series of polynitroaromatic compounds including 1,3-dinitro- benzene and 1,3,5-trinitroben~ene.~ The use of isotopic exchange as a tool for the study of proton transfer in the nitroaromatic series was first reported on by Kharasch and his co-w~rkers.~~~ Treatment of 1,3,5-trinitrobenzene with O-O2~-sodium hydroxide in ethanol- D,O at 110" for 68 hr.resulted in dilution of the deuterium of the medium to an extent equivalent to exchange of over 2 atoms of hydrogen from the trinitro- log T. Abe T. Kumai and H. Arai Bull. Chem. SOC. Japan 1965 38 1526. ll1 J. H. Fendler J. Amer. Chem. SOC. 1966 88 1237. 112 R. F. Hudson Chimia (Aarau) 1962 16 173. 113 J. F. Bunnett Ann. Rev. Phys. Chem. 1963 14 271. I14R. G. Pearson J. Amer. Chem. SOC. 1963 85 3533; 1967 89 1827. 115 G. N. Lewis and G. T. Seaborg J. Amer. Chem. SOC. 1940 62,2122. 116 V. Baliah and V. Ramakrishnan Rec. Trav. chim. 1959,78 783; 1960 79 1150. 117 G. Briegleb W. Liptay and M.Cantner Z. phys. Chem. (Frankfurt) 1960 26 55. 11* J. D. Farr C. C. Bard and G. W. Wheland J. Amer. Chem. SOC. 1949 71 2013. llg M. S. Kharasch W. G. Brown and J. McNab J. Org. Chem. 1937 2 36. J. Murto and E. Kohvakka Suomen Kern. 1966 B 39 128. 141 The Interaction of Aromatic Nitro-compounds with Bases benzene. It was pointed out recently,120 however that nucleophilic displace- ment of a nitro-group in 1,3,5-trinitrobenzene by alkoxide ion occurs6** under even milder conditions than in the exchange study119 so that the observations of Kharasch cannot be taken at face value. No exchange of hydrogen occurred when 1,3,5trinitrobenzene was treated with 8~-sodium hydroxide in D,Ol2l or with pyridine-D ,O .122 Whereas the conflicting results discussed above place the nuclear proton transfer process in question other work is more conclusive.ShatenshteinlD observed that 1,3-dinitrobenzene in liquid ND undergoes significant exchange with the medium (at 50" and 24 hr. 0-3 atom hydrogenexchange). 1,3-Dinitro- benzene is also shown to undergo hydrogen exchange (in the 2-position as shown by n.m.r .589124) with NaOD in dimet hylformamide-D 20,58 methanol-D a0,126 dimet hyl sulp hoxide-D 20>24 and dimet hoxyethane-D ,O .lW Conclusive evidence was also given of hydrogen exchange in 1,3,5-trinitrobemene by the action of sodium methoxide in tritiated methanoll*O and by NaOD (0.01~) in dimethyl- f~rmarnide-D,O.~~~ The observation of catalysis of hydrogen exchange in 1,3- dinitrobenzene12* and in 1,3,5-trinitroben~enel~~ by a number of bases both anionic and neutral suggests that these aromatic hydrogen-abstraction processes are subject to Brarnsted base catalysis.While the feasibility of the proton-abstraction process in aromatic nitro- compounds can thus be considered as proven there has been controversy about the relationship between colour and the proton abstraction process in these systems. Some workers have s u g g e ~ t e d ~ ~ ~ ~ ~ that colour development in the inter- action of 1,3-dinitrobenzene with bases provides a quantitative measure of the formation of aryl carbanion. The problem was examined quantitatively by Crampton and Gold.20 In tritium-labelled methanol and methanol-dimethyl sulphoxide mixtures containing sodium methoxide both the rate of exchange and the colour intensity (Amax 430 520 mp) increase as the basicity of the medium (as measured by the H- acidity function) increases but beyond a certain basicity (If- e 21) both the rate and the absorbance level off.The development of colour appears to be instantaneous throughout the range of basicities. Even at the basicities corresponding to maximum absorbance the rate of tritium exchange still follows a first-order rate law so that the aryl carbanion cannot be present in significant concentration. The results are consistent both qualitatively and quantitatively with the reasoning that the coloured form is the predominant species present but is unreactive in the exchange process; the latter occurs by proton removal from 1,3-dinitrobenzene which has not been converted into the 120 M. R. Crampton and V. Gold J . Chem. SOC. (B) 1966,498. 122 R. E. Miller and W.F. K. Wynne-Jones J. Chem. SOC. 1959 2375. lz3A. I. Shatenshtein N. M. Dykno A. E. Izrailevich L. N. Vasil'eva and M. Faivush Doklady Akad. Nauk S.S.S.R. 1951 79 479. 126 R. Foster and R. K. Mackie Tetrahedron 1963 19 691. lZ7 E. Buncel and E. A. Symons Chem. Comm. 1967,771. lZ8 R. Schaal Compt. rend. 1954 238 2156. J. A. A. Ketelaar A. Bier and H. T. Vlaar Rec. Trav. chim. 1954 73 37. E. Buncel and W. A. Zabel J . Amer. Chem. SOC. 1967 89 3082. E. Buncel and E. A. Symons Canad. J . Chem. 1966,44,771. 142 Buncel Norris and Russell unreactive coloured form. From the characteristics of the visible absorption spectrum it is deduced that the coloured species is the Meisenheimer adduct [methoxyl at C(4)].12* The available evidence thus indicates that the dinitrophenyl anion and by analogy the trinitrophenyl anion is not formed in more than a very small con- centration in the interaction of aromatic nitro-compounds with bases.The suggested m e c h a n i ~ m ~ ~ ~ ~ ~ ~ of the exchange process for 173-dinitrobenzene and a base Y involves a rate-determining deprotonation to form the aryl carbanion (4) followed by rapid reaction of (4) with the solvent. The exchanging [C(2)] hydrogen is the most acidic one owing to the adjacent electronegative nitro- groups. Evidence for the importance of the inductive effect in aromatic hydrogen abstraction has been pre~ented.l~~-l~l Since however exchange in 173,5-trinitro- benzene is less easy than in 1,3-dinitroben~ene,l~~9~~~ additional effects must be operative. One such effect is presumably a greater tendency for conversion of substrate into the (unreactive) Meisenheimer adduct in the case of 173,5-trinitro- benzene.B. Formation of Nitrobenzyl Anions.-The introduction of nitro-groups into the aromatic ring of toluene and its derivatives increases their Brarnsted acid strength and in sufficiently basic solution it is possible to remove a proton to produce substituted benzyl anions. 4-Nitrobenzyl cyanide (104~) for example,132 reacts with sodium ethoxide (10-3~) in ethanol solution to give a red anion (Amax = 560 mp E = 2.6 x lo4 1. mole-l cm.-l) by proton and it is only at high hydroxyl-ion concentration in aqueous solution that another process possibly formation of a a-complex competes with the production of the anion.133 Caldin and Long1% have shown that under similar conditions 2,4,6-trinitrotoluene (TNT) gives a purple anion in the temperature range -80" to 20"; the anion has absorption maxima at 510 mp ( E M 12,000) and 370 mp ( ~ ~ 7 0 0 0 ) .They note that picrate ion in methanol has an absorption maximum close to 370 mp and on the basis of this and other evidence conclude that the purple product is the 2,4,6-trinitrobenzyl anion (TNT-). The reaction of ethoxide ion with TNT TNT + OEt- + TNT- + HOEt proceeds to an equilibrium position and the equilibrium constant increases from 60 l./mole at -78.5" to 2040 l./mole at 25". The large positive entropy change 27 cal. deg.-' molew1 in this reaction in ethanol solution is interpreted by Caldin as being due to the desolvation which occurs when the ethoxide ion with its charge localised mainly on the oxygen is replaced by the trinitrobenzyl anion in which the charge is delocalised.A significant part of this desolvation appears to occur in the formation of the transition state because the entropy of 12@ A. I. Shatenshtein Adv. Phys. Org. Chem. 1963 1 156. lao G. E. Hall E. M. Libby and E. L. James J . Org. Chem. 1963 28 31 1. 131 A. Streitweiser and J. H. Hammons Progr. Phys. Org. Chem. 1965 3,41. 132 E. F. Caldin and J. C. Harbron J. Chem. SOC. 1962 2314. lS4 E. F. Caldin and G. Long Proc. Roy. SOC. 1955 A 228 263. R. A. More O'Ferrall and J. H. Ridd J . Chem. Soc. 1963 5030. 143 The Interaction of Aromatic Nitro-compounds with Bases activation is positive (7 cal. deg.-l mole-l). The activation energy for the forward reaction 13.6 kcal./mole is not much larger than that observed for association reactions of ethoxide ion with TNT and 1,3,5-trinitrobenzene in ethanol solu- t i ~ n l ~ ~ and Caldin suggests that the process of desolvation makes a considerable contribution to the activation energy.The pKa of TNT in ethanol at 25" can be estimated from the ethoxide-TNT studies134 to be approximately 15.8 which is close to the value for phenol in ethan01.l~~ Deuterium exchange with the methyl hydrogens should occur under relatively mild basic conditions but only 23 % exchange was effected when TNT was allowed to stand for 2-3 weeks at 25" in pyridine-D,O solution.122 Ex- change occurs rapidly however when catalysed by deuteroxide ion in dimethyl- formamide-D20 mixtures.13' In view of the low pKa of TNT it should be possible to produce the purple TNT- in ethanol solution by use of weaker bases than ethoxide ion.The reaction of TNT- with 3-methylphenol in ethanol yields a measurable equilibrium con- centration of TNT-,138 inferring an attack by the 3-methylphenoxide ion on TNT but direct studies in an aprotic solvent have not been reported. In more strongly basic solutions a proton may be removed from 4-nitrotoluene or 2,4-dinitrotol~ene.l~~ With potassium t-butoxide in t-butyl alcohol (or dimethyl sulphoxide) in the absence of oxygen the nitro- and dinitro-toluene yield their radical anions as shown by means of electron spin resonance spectro- scopy. In the case of 2,4-dinitrotoluene significant amounts of radical anion are produced only when the solution is deficient in base. It is thought that the nitro- or dinitro-benzyl anion is first formed and that electron transfer occurs between this anion and the parent compound to produce the radical anion.139 According to Miller and Pobiner140 the concentration of 4-nitrobenzyl anion reached when potassium t-butoxide is added to a solution of 4-nitrobenzene in t-butyl alcohol is low and the anion has an absorption maximum at 362 mp.7 Formation of Radical Anions Nitroaromatic compounds can be converted into radical anions by accepting a single electron from a strong base. 4-Nitrotoluene for example139 has been observed to give a high concentration of radical anions when dissolved in t-butyl alcohol or t-butyl alcohol-dimethyl sulphoxide mixtures to which potassium t-butoxide has been added. Such experiments are performed in the absence of oxygen and the identification of the radical anion and the determination of its concentration are readily performed using electron spin resonance techniques.With t-butoxide ion in t-butyl alcohol 1,3-dinitrobenzene is a better acceptor than nitrobenzenelgl and a small concentration of the dinitrobenzene anion 135 E. F. Caldin J. Chem. SOC. 1959 3345. 136 B. D. England and D. A. House J. Chem. SOC. 1962,4421. 13' E. Buncel and J. Wood unpublished results. 13* J. A. Blake M. J. B. Evans and K. E. Russell Canad. J. Chem. 1966 44 1 19. 130 G. A. Russell and E. G. Janzen J. Amer. Chem. SOC. 1962 84,4153; 1967 89,300. 140 J. M. Miller and H. Pobiner Analyt. Chem. 1964 36 238. 141 G. A. Russell E. G. Janzen and E. T. Strom J. Amer. Chem. SOC. 1964 86 1807. 1 4 4 Buncel Norris and Russell radical is produced even when ethoxide ion is used as the donor in ethanol solution.However 173,5-trinitrobenzene and t-butoxide give a very low con- centration of radical anion and this is ascribed to the intervention of a-complex formation between the two reactants. Many other bases can be used to produce the radical anion of nitrobenzene in t-butyl alcohol-dimethyl sulphoxide mix- tures or of 1,3-dinitrobenzene in ethanol solution;141 these include the conjugate bases of fluorene indene propiophenone quinol and diethyl malonate. The ability of the basic anion to donate a single electron to the nitroaromatic com- pound is in part related to its base strength. The dithionite ion has also been successfully used to produce radical anions from a variety of nitroaromatic compounds.142 The rate of radical-anion formation is dependent on the solvent; the nitro- benzene radical anion is more rapidly formed from nitrobenzene and t-butoxide ion when the proportion of dimethyl sulphoxide in the solvent is increased.141 Electron spin resonance studies show that radical anions of nitroaromatic compounds may interact strongly with the solvent particularly when a hydroxylic solvent such as ethanol which can form hydrogen bonds to the nitro-groups is used.143 Interaction with the positive counter-ion is however usually weak except in solvents of low dielectric constant and solvating power.Many of the radical anions of nitroaromatic compounds are reported to be coloured under the conditions used for study of their electron spin resonance spectra. For example 1,3-dinitrobenzene gives a bright violet solution im- mediately after the start of electrolysis in dimethylformamide solution and 2,6- dinitrotoluene gives a bright red solution and it seems reasonable to associate the colours with the radical anions.144 Ward145 reports that the colours of solu- tions of 173-dinitrobenzene anion in 1 ,Zdimethoxyethane vary from a light yellow with the lithium counter-ion to a dark orange for the caesium counter-ion.On the other hand a purple colour observed in concentrated solutions of this anion has been ascribed to a decomposition p r 0 d u ~ t . l ~ ~ Miller and Wynne- Jones14' suggested that there would be an electronic level in the 1,3,5-trinitro- benzene radical anion so situated as to produce absorption of light in the visible region and it is interesting that an unstable green complex is reportedly formed by the reaction of sodium with 1,3,5-trinitrobenzene in dioxan.14* There have been surprisingly few visible absorption studies of radical anions of nitroaromatic compounds.Kemula and S i ~ d a l ~ ~ have reported that the radical anion of nitrobenzene generated electrochemically in dimethylformamide shows absorption maxima at 435 mp ( E = 1.5 x lo3 1. mole-l cm.-l) and 465 mp ( E = 1.25 x lo3 1. mole-l cm.-l). Chambers and Adams150 also report lo2 P. L. Kolker and W. A. Waters J . Chem. SOC. 1964 1136. 143 P. B. Ayscough F. P. Sargent and R. Wilson J . Chem. SOC. 1963 5418. 144 J. H. Freed and G. K. Fraenkel J. Chem. Phys. 1964 41 699. 145 R. L. Ward J . Chem. Phys. 1960 32 410. 146 M. J. Blandamer T. E. Gough J. M. Gross and M. C. R. Symons J . Chem. SOC. 1964 536.14' R. E. Miller and W. F. K. Wynne-Jones J . Chem. SOC. 1959 2375. 148 A. Mathias and E. Warhurst Trans. Faraday SOC. 1962 58 942. 14Q W. Kemula and R. Sioda Bull. Acad. polon. Sci. Ser. Sci. chim. 1963 11 395. 150 J. Q. Chambers and R. N. Adams Mol. Phys. 1965 9 413. 145 The Interaction of Aromatic Nitro-compounds with Bases an absorption maximum for this species at 465 mp in dimethylformamide. The position of maximum absorption is strongly solvent-dependent shifting to lower wavelengths with increasing percentage of water in dimethylformamide-water mixtures. In 1 ,Zdimethoxyethane the radical anion of nitrobenzene generated through reaction of sodium with nitrobenzene possesses an absorption maximum at 560 mp and an E estimated at ca. 1000.151 The extent to which radical anions are present under conditions which lead to significant concentrations of a-complexes in solution is uncertain.Gold and Rochester6* observed no electron spin resonance spectrum of solutions of 1,3,5-trinitrobenzene in methanolic sodium methoxide or ethanolic sodium hydroxide solution and Russell Janzen and Strom141 observed only a very low concentration of radical anions in t-butyl alcohol solutions of potassium t-butoxide. No electron spin resonance spectrum was given by a solution of 2,4,6 trinitroanisole in ethanolic sodium hydroxide.68 A spectrum was observed during the course of a slow reaction between 1,3,5-trinitrobenzene and diethy- lamine both in the presence and absence of but the importance of this radical or radical anion in complex formation is ~ncertain.~$~@ Similarly species having electron spin resonance spectra are generated in basic solutions of acetone and 1,3,5trinitroben~ene,6~ b9153 but whether these species are derived from the initially formed Meisenheimer complex and to what extent the radical is responsible for the colour of the resulting solutions is not known.A further complicating factor in these studies is the possibility of a light- induced production of radical anions. Thus irradiation of solutions containing nitrobenzene and methoxide ion in ethanol143 and 1,3,5trinitrobenzene and cyanide ion in chloroformlS gives rise to free-radical species. It is probable however that this is not a serious complication in the general studies of inter- actions of bases with aromatic nitro-compounds reported here. 161 A. Ishitani K. Kuwata H. Tsubomura and S. Nagakura BUN. Chern. SOC. Japan 1963 36 1357. 162 R. E. Miller and W. F. K. Wynne-Jones Nature 1960 186 149. A. R. Norris unpublished results. 146

ISSN:0009-2681    

DOI:10.1039/QR9682200123

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Reactions of Ions and Excited Atoms of the Inert Gases By B. Brocklehurst DEPARTMENT OF CHEMISTRY UNIVERSITY OF SHEFFIELD S3 7HF 1 Introduction This Review is not concerned with the stable compounds of the rare gases but with the transient species produced by light or fast particles (electrons in dis- charges and mass spectrometers high energy radiation). Ed (German ‘Edelgas’) is used to symbolise any of the gases but some generalisations apply only to the heavier gases-neon argon krypton and xenon; helium has a different elec- tronic structure and its excited states differ in some of their properties. Some of the ‘reactions’ involve only the transfer of charge or excitation to another atom or molecule though the charged or excited molecules produced often decompose immediately; subsequent ionisation of an excited molecule is called the Penning reaction1 Ed* + X+Ed + X+ + e- (1) The excitation energies of even the lowest excited states of the inert gases (see Table) are larger than the ionisation potentials of many molecules.However the ions of the inert gases are isoelectronic with halogen atoms so it is not surprising that they can abstract atoms such as hydrogen from molecules and so give rise to intermediates in which the inert-gas atom is bound to other atoms; e.g. Similar reactions involving excited atoms are few but there are many examples of associative ionisation Ed*+X-tEdX++e- (3) Intermediates like ArH+ can also be produced from ground-state atoms; e.g. Hs+ + He -+ HeH+ + H (4) Unlike the ground-state atoms which only form diatomic ‘molecules’ bound by van der Waals forces Ed+ and Ed* can be converted into Ed$ and Eda* in three-body collisions A.A. Kruithof and F. M. Penning Phjuica 1937 4 430. 147 The Reactions of Ions and Excited Atoms of the Inert Gases Ed2+ ions are also produced by the Hornbeck-Molnar process2 [reaction (3) X = Ed]. These species have binding energies of 1-2 ev. A comprehensive review of these reactions has been published recently3- some 285 were listed. Here only a few will be discussed with reference to their intrinsic interest and to their role in reactions sensitised by the rare gases. Reactions (2) and (4) are very simple reactions; also the kinetic energy of the ions can be varied over a wide range by means of electric fields. Detailed studies have therefore been made4 to test theories of reaction kinetics.The use of mercury photosensitisation is well known;5 traces of mercury absorb the 2537 8 resonance line strongly and can be used to excite molecules which do not themselves absorb conveniently in the near-ultraviolet region and as a source of reactive intermediates e.g. The inert gases can be used in the same way and the energies of Ed* are such as to produce ions and highly excited species which are important in radiation chemistry and in the chemistry of the upper atmosphere. The range of applica- tions is somewhat limited since many molecules themselves absorb strongly in the vacuum-u.v. region; also the experimental difficulties in this region (< 2000 A) are much greater and no window materials will transmit the He and Ne resonance lines. Unlike mercury however the inert gases can easily be handled at high pressures (used here to mean roughly 1-760 mm.).They can therefore form the bulk of mixtures with other substances; it is then not neces- sary to use specific excitation with light but the whole mixture can be excited by high-energy radiation or in a discharge. Most of the energy is absorbed by the inert gas and then transferred in one way or another to the reacting substance. In the case of discharges the pure gas can be excited in a flow system and the reactant added downstream. These methods are much easier to use than vacuum-u.v. sensitisation but interpretation of the results is more difficult since a wide range of ions and excited states of the inert gas are produced instead of a single excited state. It is convenient to describe first the experiments carried out at low pressures (< 1 mm.; photosensitisation mass spectrometry). Then a detailed description of the degradation of high-energy radiation in the pure inert gases is given as a preliminary to accounts of ionisation luminescence and chemical change in gas mixtures. Gas discharges are described briefly because of their complexity and smaller chemical interest ; more interesting are the recently developed discharge-flow systems. J. A. Hornbeck and J. P. Molnar Phys. Rev. 1951 84 621. G. von Biinau Fortschr. Chem. Forsch. 1965 5 347. Advances in Chemistry Series No. 58 ‘Ion-Molecule Reactions in the Gas Phase’ American H. E. Gunning and 0. P. Strausz Adv. Phofochem. 1963 1 209; R J. CvetanoviC Progr. Chemical Society Washington D.C. 1966. Reaction Kinetics 1964 2 39.148 Brocklehurst 2 Energy levels The energies6 and radiative lifetimes’ of the states derived from the first excited and the first ionised configurations of the inert gases are given in the Table; for comparison the resonance lines of mercury at 1849 and 2537 8 correspond to 6.70 and 4-89 ev. The ground-state atoms (Ne Ar Kr Xe) have the valency shell configuration s2p6; when an electron is excited to give s2p5s four states result. Russell-Saunders notation is commonly used as in the Table but because the outer s-electron occupies a much larger orbital than the others coupling of its orbital and spin moment to those of the core electrons is weak compared with coupling between orbital and spin. Therefore the terms ‘singlet’ and ‘triplet’ are not very meaningful especially in the heavier gases.This is illustrated by the radiative lifetimes shown in the Table even in neon the 3P state radiates very much faster than the triplets of e.g. carbon compounds. The 3P and 3P2 are metastable because of the dJ = 1 selection rule; however at high pressures radiation from the resonance levels IP and 3P, will be ‘imprisoned’ in the gas by repeated re-absorption by ground-state atoms so that all four levels can be described as ‘metastables’. Vacuum-u.v. photons from transitions between higher states and the ground state are also re-absorbed but these levels can emit photons in the visible and near4.r. region and so produce atoms in the metastable states. Helium has a different electron structure and the spin selection rule (0s = 0) does hold rigorously.The spectra of He are well known but there has been very little work on the other Ed2* molecules; they undergo transitions to the repulsive ground state giving rise to continua in the vacuum-u.v. region These continua are now widely used in absorption spectroscopy.* Ed2* - + E d + Ed + hv (8) Table Energy levels of the inert gases and their radiative lifetimes Met astables He Ne Ar Kr Xe I 2 3s 2 lS 19-8 1 20-61 - - 3p2 3p1 16-61 16.67 11.55 11-62 - 8.1 9-91 10-03 3.4 8.31 8.43 3.8 21 - - - 2 lP 21-21 0.56 16.7 1 11.72 10-56 9.44 3p0 - - - - I - lP1 16.84 1-5 11.82 1-8 10.64 3.3 9.57 3.2 Ions A 2S 24.58 - - - 2p3/2 ?P1/2 21-56 21-66 15.75 15-93 14.00 14-66 12.13 13.43 - - - - - - - - ev nsec. ev nsec. ev nsec. ev nsec. ev nsec. C. E. Moore Nat. Bur. Stand. Circular 467 vol.I TI I11 (1949 1952 1958). ’ P. G. Wilkinson Canad. J. Phys. 1967,45,1709; E. L. Lewis Proc. Phys. SOC. 1967,92,817. * P. G. Wilkinson J. Mol. Spectroscopy 1961 6 1. 149 The Reactions of Ions and Excited Atoms of the Inert Gases 3 Low-pressure Studies A. Photosensitised reactions.-In recent years there has been a considerable increase in photochemical work in the region 2000-1050 8 (below which lithium fluoride windows cease to tran~mit).~ A number of lamps have been constructed which emit the longer-wavelength resonance line (3P1 -+ ?Yo) of Ar Kr and Xe together with lower intensities of the shorter-wavelength one but surprisingly little work on sensitised reactions has been publi~hed.~,~ Of course many molecules absorb strongly in this region but there are exceptions such as nitrogen; the krypton-sensitised dissociation of nitrogen into atoms has been demonstrated by use of isotopes;1° in view of the difficulties associated with the use of 'active nitrogen' this is a potentially useful technique for study- ing reactions of the atoms.Also transfer of energy from Ed* does not neces- sarily produce the same states as direct excitation for example absorption of light by organic compounds mainly produces singlet states while transfer from a triplet excited atom is likely to produce triplet states. (It does not necessarily produce a triplet state since spin conservation is no more likely to hold than in the 3P1 - 'So radiative transition.) von Bunau and Schindlerll have compared xenon-sensitised photolysis of ethane with direct photolysis and find a larger proportion of products (C3H, C4H1J formed by a free-radical mechanism (primary splitting into 2CH3 or C,H + H); on the other hand krypton- sensitisadon favours products formed by direct molecular elimination (C,H4 + H2).Free radicals result from repulsive triplet states while ground-state mole- cular products must correlate with singlet states. The difference between krypton and xenon is probably due to the difference in the available energy and the configurations of the excited states. B. Mass Spectrometry.-At the low pressures normally used in the ion-source of a mass spectrometer primary ions do not collide with other molecules before extraction for analysis ; at higher pressures (10-3-10-2 mm.) secondary ions such as ArH+ from reaction (2) axe easily observed. Only in the last 15 years have these processes been studied in detail; because of their importance in radia- tion chemistry and elsewhere there is now great activity in this field.394J2 For certain identification of the primary ion a double mass spectrometer is used:13 the first analyser injects ions of known elm and kinetic energy into the ion-source of the second spectrometer where reaction occurs.Such apparatus is compli- cated and much work has been done in ordinary mass spectrometers. Changes in pressure and in the energy of the ionising electrons are used to identify the reactants producing a given secondary ion; e.g. secondary and primary ion should have the same appearance potential. The kinetic energy of the reactants in a mass spectrometer can be controlled J. R. McNesby and H. Okabe Adv.Photochern. 1964,3 157. lo W. Groth W. Pessara and H. J. Rommel Z. phys. Chenz. (Frankfurt) 1962 32 192. l1 G. von Biinau and R. N. Schindler J. Chem. Phys. 1966,44,420. l2 F. W. Lampe J. L. Franklin and F. H. Field Progr. Reaction Kinetics 1961 1 69. l3 E. Lindholm ref. 4 p. 1. 150 Brocklehurst by electric fields and reactions between ions and molecules can also be studied by molecular-beam techniques14J5 which give detailed information about the angular distribution of the recoiling products; in both cases data of considerable theoretical interest are obtained. C. Ion-Molecule Reactions Charge Transfer.-Processes in which a bond is formed between the reactants (or parts of them) are called ion-molecule reac- tions i.e. charge transfer is not included though it involves an ion and a mole- cule; since they are competing processes the two are considered together here.Two simple ion-molecule reactions (2) and (4) have already been mentioned and a number of other abstraction reactions are known,3 e.g. Xe+ + 0 + XeO+ + 0 Ed+ + CH --f EdH+ + CH Ed+ + CH +- EdCH,+ + H Ed+ + CH -f EdCH,+ + H Reaction (2) has been observed for Ar Kr and Xe. HeH+ produced in mix- tures of He and H was originally ascribed to reaction (2) but the ion is still present when the electron energy is less than the ionisation potential of helium showing that H2+ is the precursor. Reaction (4) is in fact endothermic and the appearance potential corresponds to vibrationally excited H2+ ions.ls Reaction (2) for helium would be exothermic by 8.3 ev so that if HeH+ were formed it would probably decompose into He and H+.It is simpler to regard such a reaction as charge transfer followed by decomposition of the newly formed ion [but see below-reaction (1 3)]. From these energy considerations it can be predicted that He+ and Ne+ will transfer charge in this way while the heavier gases will also undergo ion-molecule reactions. This is illustrated by reactions (lo) (ll) and (12) which have not been observed in the cases of He+ and Ne+. Further charge transfer from the lighter gases will cause more frag- mentation of the product ion because of the larger energy release. For example methane (reacting with 2 ev ions in a double mass spectrometer1') gives the following relative yields :-Xe+ -+ CH,+ @ CH,+ Kr+ -+ CH$ > CH3+ Ar+ - CH,+ > CH,+ 9 CH,+ Ne+ - CH,+ > CH+ He+ -f CH,+ > CH+ > CH,+ > C+.More detailed information about such dissociation processes has been obtained by the use of isotopes e.g. CH3CD,.CH3 has been used to distinguish between the formation of n-propyl and s-propyl ions after collision with inert-gas ions.ls Exothermic ion-molecule reactions are very fast at thermal energies,I2 faster than the normal collision rate of neutral species because of the long-range l4 A. R. Blythe M. A. D. Fluendy and K. P. Lawley Quart. Rev. 1966 20 465. l5 Z. Herman J. Kerstetter T. Rose and R. Wolfgang Discuss. Faraday Suc. 'Molecular Dynamics of the Chemical Reactions of Gases' 1967 to be published. l7 G. R. Hertel and W. S . Koski J. Amer. Chem. SOC 1965 87 1686. l8 J. H. Futrell and T. 0. Tiernan J. Chem. Phys. 1963 39 1539.L. Friedman ref. 4 p. 87. 151 The Reactions of Ions and Excited Atoms of the Inert Gases attraction due to ion-induced dipole interaction (energy cc l/P); the mechanism is described in terms of ‘orbiting colli~ions’~~ or ‘complex’ collisions. The life- time of the complex is comparable with its period of rotation. The calculated rate constants decrease with increasing kinetic energy and at high energies the reactions are described in terms of the ‘stripping’ model taken over from nuclear physics;20 in the extreme case an atom is ‘stripped’ from a molecule without any transfer of momentum to the rest of the molecule. An isotope effects should be apparent in the former case but not the latter and reactions (2) and (4) with HD have been extensively studied. However recent molecular-beam work16 has led to the suggestion that no complex is formed even at low energies.Another reaction of especial interest is (13) He+ + 0 -t He + Of + 0 (1 3) One long-standing problem in the chemistry of the upper atmosphere has been the mechanism by which helium atoms acquire enough energy to escape the earth’s gravitational field. He+ is produced in the atmosphere at the required rate by photoionisation; Friedman16 has measured the kinetic energy of O+ formed in reaction (8). This changes when 4He is replaced by 3He which shows that (8) is not a simple charge-transfer process (when all the kinetic energy would be released in the dissociation of O,+) but involves a transient excited He@ ion which dissociates giving He atoms having about 1 ev of kinetic energy. D. Ionising Reactions of Excited States.-The Penning reaction (1)’ is not easily studied in a mass spectrometer because the same ions are produced directly but associative reactions (3) are easily observed3 since new ions are produced.(Ed* here often refers to higher excited states rather than metastables.) X may be an alkali-metal atom Hg or a molecule (N, 0,’ CO C,H2 CH,); Ed is usually one of the heavier gases. Kr* reacts with I to give KrI+ + I but few reactions of this type are known. The Hornbeck-Molnar process [reaction (3); X = Ed] has been the subject of much detailed study because of its importance in producing the diatomic ions Ed,+. It was first postulated by Arnot and M’Ewen21 (who wrongly equated Ed* with the metastables; such a reaction would be endothermic by 2-3 ev) soon after the discovery of the Ed,+ ions in 1936.22 Kinetic studies using pulsed mass ~pectrornetry~~ show that the reactions are very fast (k N 1-2 x ~ m .~ sec.-l> and the participation of several states with lifetimes up to 5 x lo-‘ sec. has been detected. The appearance potentials of Ed,+ ions give lower limits for the binding energies of the ions the most accurate values [D(Ed-Ed+)-l ev] have been obtained using vacuum-u.v. excitati~n.~~ The various possible mixed processes giving ions such as HeAr+ have also been studied.25 l9 G. Gioumousis and D. P. Stevenson J. Chem. Phys. 1958 29 294. 2o A. Henglein ref. 4 p. 63. 21 F. L. Arnot and M. B. M’Ewen Proc. Roy. SOC. 1938 A 166 543. 32 0. Tuxen 2. Physik 1936 103 463. 23 W. Kaul P. Seyfried and R. Taubert 2. Narurforsch 1963 18a 432; 884.24 R. E. Huffman and D. H. Katayama J. Chem. Phys. 1966 45 138. 25 M. S. B. Munson F. IJ. Field and J. L. Franklin J. Phys. Chem. 1963 67 1542. 152 Brocklehurst 4 High-energy Irradiation That energy absorbed by an inert gas irradiated for instance with a-particles could cause the decomposition of an admixed gas was demonstrated by Lind many years ago and he has reviewed the early work in this field.26 Detailed interpretation of the results has had to await the arrival of modern techniques such as mass spectrometry. Absorption of high-energy radiation by any sub- stance produces a great range of excited neutral molecules ions and excited ions; yields of these species their decomposition and the subsequent reactions give very complicated reaction mechanisms so that any simplification of the initial processes is very desirable.It might be hoped that inert-gas sensitisation would reduce the range of initial states by transferring energy and charge at a few energies only e.g. those of the metastables and the ground-state ions. If this is to happen then decay of other excited states must be very rapid. The processes occurring in the pure gases are described first so that the use of this technique can be evaluated. Whatever the nature (a-particles electrons y-rays etc.) or energy of the primary radiation it produces many fast secondary electrons ( 1 ~ 1 0 0 0 ev) which dissipate most of the energy in a large number of collisions which may ionise or excite molecules. A useful parameter is the ‘ W-value’-the number of electron volts required to produce one ion pair.W depends on the nature of the substance and is typically 25-35 ev i.e. twice the ionisation potential-the remaining energy is dissipated mainly as electron excitation with a small fraction as thermal energy.27 The fast electrons eventually fall below the lowest excitation potential of the molecules they are then called sub-excitation electrons. In a pure substance their energy is eventually converted into heat but in mixtures they may specifically excite the substance with lower energy levels; no clear-cut evidence of this effect is available but it is likely to be important in the inert gases.27 A. Decay of Excitation in the Pure Gases.-Excited ions will probably decay rapidly by emission of radiation in the vacuum-u.v. region (where radiative life- times are very small sec.) to the two lowest states (see Table).Optical transitions between these two states are strictly forbidden; collisional conversion 2P,,2 -f 2P3,2 does not appear to have been studied despite its possible import- ance in charge-transfer processes. Neutral excited atoms will initially be in states which can radiate to the ground state but such radiation is imprisoned as already described ; unless the Hornbeck-Molnar process intervenes the higher excited states are converted into the four metastables by a series of transitions (radiative cascade) that will take about 10-7-10-6 sec. Studies of the decay of the metastables in discharge afterglows28 have shown 26 S. C. Lind ‘Radiation Chemistry of Gases’ Reinhold New York 1961. 27 R. L. Platznian Internat. J. Appl. Radiation Isotopes 1961 10 116; Radiation Res.1962 17 419. 28 L. B. Loeb ‘Basic Processes of Gaseous Electronics’ University of California Press Berkeley and Los Angeles 2nd edn. 1961; J. B. Hasted ‘Physics of Atomic Collisions’ Butterworths London 1964. 153 The Reactions of Ions and Excited Atom of the Inert Gases that at pressures high enough to prevent diffusion to the container walls (a few mm.) the 3P2 states are destroyed in three-body collisions [reaction (6)] rate constants have been measured.29 The He (2 3c) molecule is also metastable and is destroyed by diffusion to the cell walls or collision with imp~rities.~~ The other molecules radiate in the vacuum-u.v. region [reaction (S)] the radiative lifetime has been measured in one case-Ar2* 3.4 x sec30 The other metastable atoms are converted into (3P2) atoms in collisions measurements on neon31 show that this is a fairly slow process so that at higher pressures (a few 100 mm.) the other metastabjes are probably converted into molecules directly as well as via the 3P2 level.The monatomic ions are also removed in three-body collisions [reaction (5)]. This was first postulated by Bates32 to explain the observed rapid recombination of He2+ ions and electrons. Direct recombination Ed+ + e- should be very slow because the energy cannot be removed except by radiation (which is im- probable) or by simultaneous collision with a third body (compare the require- ments for atom-atom recombination). The diatomic ions in contrast can react rapidly (rate constants are ~ l O - ~ - - l 0 - ~ ~ m . ~ atom-l sec.-l) by dissociative recombination Ed,+ + e -+ Ed* + Ed (14) Some of the energy is released as kinetic energy but much is retained as electronic excitation of one of the atoms (which then passes through the sequence of events described above).The correctness of this mechanism has been demon- strated in an elegant experiment by Connor and BiondP who detected the Doppler broadening of the emission from recoiling Ne* atoms. B. Sensitised Processes.-It is necessary to consider the roles of Ed* (metastable and higher levels) Ed2* Ed+ Ed+* and Ed2+; possible effects of vacuum-u.v. photons and sub-excitation electrons are ignored here. Argon (300 mm.) with an additive (10 mm.) is taken as a typical mixture. The lifetime of Ed* etc. with respect to collisions with additive molecules is about sec. This is comparable with or perhaps greater than the radiative lifetime of Ed+* so that these species are not likely to be important.The Hornbeck-Molnar process is very fast (mean life of Ar* at this pressure 10-lo sec.) and this removes all Ed* with sufficient energy. Ed* with lower energy decay more slowly by radiation (10-s-10-7 sec.); these will therefore undergo collisions with the added molecules ; fortunately only a few levels new 14 ev are involved in Ar (near 14 ev)-the higher levels all give Ar2+. There is a similar situation in neon but a wider range of low-lying levels in Kr and Xe.6 *' A. V. Phelps Phys. Rev. 1955 99 1307. 30 L. Colli Phys. Rev. 1954 95 892. 31 A. V. Phelps Phys. Rev. 1959 114 1011. 32 D. R. Bates Phys. Rev. 1950 77 718. 33 T. R. Connor and M. A. Biondi Phys.Rev. 1965 140 A778. 154 Brocklehurst The three-body processes are somewhat slower; mean lives of Ed+ and Ed* with respect to conversion into Ed2+ and Ed2* are about lo-’ and sec. at this pressure. The lifetime of Ar2* is 3.4 x sec.?O while the rate of reaction (14) depends on the concentrations of Ed2+ and e-. The mean life of Ed2+ there- fore decreases with increasing dose rate and will vary between about sec. for y-irradiation and about sec. for bombardment in an electron accelerator. The conditions quoted appear to be useful; the sensitisers will be Ar+ Ar2+ (Hornbeck-Molnar only) Ar* (14 ev) and Ar* (metastable); Ed+* and Ed2* are not important. Further simplification would be difficult. After sec. only Ed,+ ions remain but additive pressures of only 0.01 mm. would have to be used; problems of chemical analysis would be very great though useful results have been obtained in the special case of luminescence (see below).A more detailed discussion of the decay processes has been published.= C. Ionisation Yields.-W values are obtained by measuring the number of ions produced by a fast particle as the ‘saturation current when an electric field is applied across an ionisation chamber. For many years values quoted for the light inert gases were in error because of the effect of impurities which produce more ions by the Penning reaction (1).27,35 For example a few parts in lo4 of impurities in helium reduce W from 41-3 to 30.35 Effects are less striking in the heavier gases because of the lower energies of their metastable states which lie below the ionisation potentials of many molecules.Jesse and Sada~skis~~ have made detailed studies of this effect and obtained relative rates of reaction between the metastables and various additives. Ionisation in argon mixtures due to the 14 ev levels mentioned above has also been detected.36 Some gases were found to af€ect W even though their ionisation potentials lay above the Ar metastables. A molecule excited above its ionisation potential does not necessarily ionise; the ‘super-excited molecule’ (e.g. one in which a strongly bound electron has been excited) may decompose into neutral fragments. This is demonstrated by the ~ b s e r v a t i o n ~ ~ ~ ~ that neon containing methane has a larger W value than CD,-Ne. The rates of ionisation of super-excited CH and CD are presumed to be the same so the competing decomposition must be faster in CH,.D. Luminescence of Irradiated Gases.-The inert gases give brighter lumines- cence than other gases and there have been many reports of their use in scintil- lation counters for detection of fast particles.38 Reproducible results are not easily obtained because of the effects of impurities; deliberate use of mixtures such as Ar + N, reduces the difficulties. These counters are useful because 34 B. Brocklehurst ‘Radiation Research Reviews’ ed. R. B. Cundall F. S. Dainton and G. 0. Phillips Elsevier Amsterdam to be published. 36 W. P. Jesse and J. Sadauskis Phys. Rev. 1952 88 417. 87 W. P. Jesse and R. L. Platzman Nature 1962 195 790 W. P. Jesse J. Chem. Phys. 1964 41 2060. 88 J. L. Teyssier D. Blanc and A. Godeau J. Phys. Radium 1963,24 55.G. S. Hurst T. E. Bortner and R. E. Glick J. Chem. Phys. 1965 42 713. 155 The Reactions of Ions and Excited Atoms of the Inert Gases their response varies linearly with energy even for very densely ionising particles (fission fragments). The impurity effects are most striking in helium where a few parts of nitrogen in lo9 have been detected by emission of the N2+ first negative bands (excited by collision between He2+ and N2).39 The effects are smaller in the other gases but a few p.p.m. of nitrogen in argon give rise to the N second positive bands (collision with Ar meta~tables).,~ In addition to these simple transfer processes dissociative reactions such as (13) and (14) have also been observed Ar*+H,O+Ar+H+OH* Xe* + O,+XeO* + 0 E. Radiation Chemistry.-The observed effects are often striking; e.g.in Xe-CH mixtures the yield of H is greater than that expected on the basis of energy absorbed by the methane but in Ar-CH mixtures energy absorbed by Ar is actually more efficient at producing H than that absorbed by CH4.41 Because of the complexity of the sensitisation process in only a few cases has it been possible to draw up a complete reaction scheme; inert gas-propane mix- tures have received the most detailed However it is often possible to pick out changes in relative yields of products due to known reactions e.g.,43 the different Ed+ + CH reactions1' decribed above. Xenon is of special interest in that it has a lower ionisation potential than many reactant gases. Excitation transfer alone would be expected and charge transfer to xenon should also reduce yields.But xenon often acts as efficiently as a sensitiser as the other gases perhaps via the formation of complex ions which decompose on recombination. For example XeC2H2+ has been observed in the mass spe~trometer.~ Highly specific effects of the individual gases are rare but one occurs in the radiolysis of acetylene. Formation of benzene is specifically sensitised by neon; this is ascribed to reaction (17).44 The other inert gas ions do not give C,H+. Yields of the main product-the polymer cuprene-are enhanced by all the gases; since Kr+ does not readily transfer charge to C2H2 in a mass spectrometer it has been argued that poly- merisation occurs by a free-radical mechanism, (though three-body ion-associa- tion processes might occur at the higher pressure).Reaction (18) has been used as a source of hydrogen atoms for reactions with other substances :46 Ne+ + C2H2 -+ Ne + C2H+ + H (17) Ed++ H,-+EdH++ H (1 8) 30 W. R. Bennett Ann. Phys. 1962 18 367. 40 B. Brocklehurst Trans. Faraday SOC. 1967 63 274. 41 R. W. Hummel Trans. Faraday SOC. 1966 62 59. 42 J. H. Futrell and T. 0. Tiernan J. Chem. Phys. 1962,37,1694; L. 1. Bone L. W. Sieck and J. H. Futrell ibid. 1966 44 3667. 43 G. G. Meisels W. H. Hamill and R. R. Williams J. Chem. Phys. 1956 25 790; V. Aquil- anti J. Phys. Chem. 1965 69 3434. 44 J. H. Futrell and L. W. Sieck J. Phys. Chem. 1965 69 892. 4c A. Maschke and F. W. Lampe J . Amer. Chem. SOC. 1964,86,569. 156 Broc klehurst With an excess of hydrogen H,+ may be formed EdH+ + H 3 Ed + H3+ This is important in the sensitised exchange reaction between H and D of which reaction (20) is the chain-propagating step.46 H3+ + D - H + HD,+ While He Ne and Ar sensitise this process Kr and especially Xe have a strong inhibiting action since reaction (19) proceeds in the reverse direction because of the high proton affinity of the heavier gases.5 Excitation in Discharges One important difference from high-energy excitation is that the power input is so much greater in a discharge (watt~lcm.~; cf. N lo-’ wattsl~m.~ for y-irradia- tion). As a result a much larger proportion of the atoms are excited or ionised and reactions between two active species become important ; for example metastable states destroy each other 2 Ed* -t Ed+ + Ed + e- (21) Reaction (21) is very fast;47 also the long-lived metastables may be excited to radiating states by fast electrons or deactivated in ‘superelastic collisions’ with slow ones (the reverse of excitation) so that they are less important in discharges.Another complication in gas mixtures is that the electrons are continually re- accelerated from low energies by the field and so can preferentially excite added molecules which have lower energy levels than the inert gases. The sub-excitation electron problem of radiation chemistry is much more important in discharges. Reaction (1)l was postulated many years ago to explain the ionisation effects in gas mixtures. Also well known is the ‘perturbation’ of molecular spectra by added inert gases;48 the observed bands are often reduced to those vibrational- electronic levels which lie close to the metastable levels of the inert gases.For example nitrogen with helium gives the first negative bands of N2+ with argon the second positive bands and with xenon the first positive bands. The ions of the inert gases have been shown to be involved in some cases but the phenomenon is probably more complex than previously suspected; e.g. recent workg9 on argon-hydrogen mixtures showed that the radiating states of argon and vacuum- U.V. photons were more important than the true metastable states. A. Lasers.-However these energy transfer processes occur they produce a limited range of excited states. This characteristic which means that high energy levels may be populated in preference to lower ones makes it possible to use them in gas-discharge lasers.50 The essential requirement for laser action is the 46 0.A. Schaeffer and S. 0. Thompson Radiation Res. 1959 10 671. 47 W. B. Hurt J Chem Phys 1966 45 2713. 48 E J. B. Willey ‘Collisions of the Second Kind‘ Arnold London 1937; R. Meyerott Phys. Rev. 1944 66 242. 4D S. Takezawa F. R. Innes and Y . Tanaka J. Chem. Phys. 1966 45 2000. W. R. Bennett Appl. Opt. Suppl. 1962 1 24; 1965 2 3. 157 3 The Reactions of Ions and Excited Atoms of the Inert Gases production of a population inversion i.e. there should be more molecules in the upper state than in the lower one to which stimulated emission takes place. One process which has been used is (22). NecP,, 3P,) + 0 - O(3 3P) + 0 + Ne; O(3 3P) -+ O(3 3S) 4- 8446 A B. Discharge Afterglows.-When a discharge is stopped the ions and excited atoms decay by the processes already described usually giving rise to light emission whence the name ‘afterglow’.Static afterglows following a pulsed discharge can be studied by absorption spectroscopy (metastables) or micro- wave techniques (electron concentrations).28 Many of the processes already discussed [e.g. reactions ( 5 ) and (6)] were studied in this way and their rate constants measured. This work has only become feasible in the last 10-15 years because of the need for advanced electronic techniques and bakeable vacuum systems to maintain the required gas purity. Static afterglows are most suitable for maintaining high purity but for reactions with other substances discharge- flow systems are more convenient. This method in which gas flows through a continuous discharge before reacting with a substance added downstream has long been familiar in work on ‘active nitrogen’,51 but has only recently been used with the inert gases where the decay of excitation is much faster.Added gases often produce luminescence and some attractive coloured photographs have been published.52 The relative concentrations of the active species are easily varied-a consider- able advantage over the use of high-energy radiation. For example in argonus metastable atoms predominate at low discharge currents ions at high currents. Pressure is also important in one set of measurements the predominant species in helium at 10 20 and 30 mm. were found to be He (23S) He+ and He$ re~pectively.~~ In other experiments argon has been used to remove helium metastables (by the Penning reaction)% and microwave fields have been used to heat ions and electrons and so enhance their loss by diffusion to the walls.s6 Using a mass spectrometer to detect ionic species the research group at the Environmental Science Services Administration’s laboratories at Boulder Colorado have recently studied a great variety of reactions and have made accurate measurements of the rates of several processes which are important in the upper atmo~phere.~~-~~ A number of charge-transfer processes have been B.Brocklehurst and K. R. Jennings Progr. Reaction Kinetics 1967 4 1. sz A. h%. Bass and H. P. Broida J . Res. Nut. Bur. Stand. Sect. A 1963 67 379. 68 J. F. Prince C. B. Collins and W. W. Robertson J . Chem. Phys. 1964 40 2619. 64 C. B. Collins and W. W. Robertson J. Chem. Phys. 1964 40 701. 66 F. C. Fehsenfeld A. L. Schmeltekopf and E.E. Ferguson Planet. Space Sci. 1965,13,219. 6aE. E. Ferguson F. C. Fehsenfeld D. B. Dunkin A. L. Schmeltekopf and H. I. Schiff Planet. Space Sci. 1964 12 1169. 15’ F. C. Fehsenfeld E. E. Ferguson and A. L. Schmeltekopf J . Chem. Phys. 1966 45 404. 68 E. E. Ferguson F. C. Fehsenfeld P. D. Goldan A. L. Schmeltekopf and H. I. Schiff Planet. Space Sci. 1965 13 823. 69 A. L. Schmeltekopf F. C. Fehsenfeld G. I. Gilman and E. E. Ferguson Planet. Space Sci. 1967 15,401. 158 Broc klehurst studied:' but the main interest has been in the subsequent reactions. For example one can produce N+ or N2+ by reaction (23) or (24) He+ + N -+ He + N+ + N He (2 3S) + N -+ He 4- Na+ + e and study their reactions with 0 added further downstream. The advantages of the flow technique are further illustrated by recent work on reactions with a second reactive species produced by a second discharge in the gas being added; addition of oxygen atoms (from a discharge in 02)58 and vibrationally excited nitrogen (N2t discharged N,)59 has permitted the study of reactions (25) and (26). Nz+ + 0 +NO+ + N O++N,t-+NO++N Reaction (26) which is very slow for N (v = 0) is very important in the iono- sphere since it converts an atomic ion into a molecular ion which can recombine rapidly with electrons; this controls the electron concentration in the F layer. Clearly the use of dischargeflow systems with the inert gases is a most valuable technique. 1 59

ISSN:0009-2681    

DOI:10.1039/QR9682200147

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Semiconduction and Photoconduction of Biological Pigments By R. J. Cherry UNILEVER RESEARCH LABORATORY THE FRYTHE WELWYN HERTS 1 Introduction There has in recent years been a growing interest in the properties of the organic solid state. This interest has no doubt been stimulated in part by the success achieved in understanding inorganic solids and the widespread technological applications of semiconductive phenomena which have followed. The study of the conductivity of organic solids received a further impetus however when Szent-Gyorgyi suggested that the transfer of electrons might play an important role in the functioning of biological systems.l Following this speculation studies have been made of the conductivity of macromolecules such as proteins2 and DNA,3 and naturally occurring pigments in particular chlorophyll and caro- tenoids.Some measurements have also been attempted with more complex biological materials such as dried chloroplasts and retinal rods. Apart from their biological importance the properties of some naturally occurring pigments are also of theoretical interest. In particular polyenes have attracted considerable attention from quantum chemist^.^ Various quantum- mechanical methods have been used in attempts to calculate the electronic energy levels of these molecules. The property of cis-trans-isomerism is a further aspect of polyenes which has been investigated in some detaL5 The purpose of this review is to bring together thevarious investigations of the conductivity of naturally occurring pigments and to give some consideration to the biological problems which have stimulated this research.The next section contains a brief account of some of the more important aspects of organic semi- conductors in general so that measurements with biological materials may be viewed in relation to our present understanding of this area. More detailed information may be found in some recent reviews6 Szent-Gyorgyi Science 1941 93 609. D. D. Eley ‘Horizons in Biochemistry’ ed. N. Kasha and B. Pullman Academic Press D. D. Eley and D. I. Spivey Trans. Faraday SOC. 1962 58 41 1. (a) C . A. Coulson Proc. Roy. SOC. 1939 A 169 413; (b) Y. Ooshika J. Phys. SOC. Japan 1957,12 1238; 1246; (c) H. C. Longuet-Higgins and L. Salem Proc. Roy. SOC. 1959 A 251 172; ( d ) H. Labhart J. Chem. Phys. 1957 27 957; (e) J. N. Murrell ‘The Theory of the Electronic Spectra of Organic Molecules’ Methuen London 1963.L. Zechmeister Fortschr. Chem. org. Naturstoffe 1960 18 223. (a) J. Kommandeur ‘Physics and Chemistry of the Organic Solid State’ ed. D. Fox M. M. Labes and A. Weissberger Interscience New York. 1965 vol. 11 p. 1 ; (b) Y. Okamoto and W. Brenner ‘Organic Semiconductors’ Reinbold New York 1964; (c) F. Gutman and L. E. Lyons ‘Organic Semiconductors’ Wiley New York 1967. New York 1962 341. 160 Cherry 2 Organic Semiconductors In inorganic semiconductors the electronic energy levels form energy bands which result from the strong overlap of the electronic orbitals of the individual atoms. These energy bands largely determine the electrical and optical properties of the semiconductor. In organic crystals however there is generally little over- lap between the orbitals of neighbouring molecules.As a result there is generally no gross difference between the absorption spectra of the crystal and vapour states. The optical absorption of the crystal may be successfully explained in terms of exciton theory.’ Both singlet and triplet exciton states occur in the crystal corresponding to the singlet and triplet states of the free molecule. An important consequence of exciton theory is that the excitation energy or exciton is able to migrate through the crystal. However no electrical charge can be transported by this process. The observation of semiconducting properties in crystals of conjugated molecules implies the presence of conduction states in addition to the non- conducting exciton states described above.In these molecules the n-electrons may be delocalised over the whole molecule. In the solid the overlap of the n-electron orbitals leads to the formation of conduction states in which electrons can move through the crystal. Because of the weak overlap the conduction bands are expected to extend over only narrow ranges of energy. As in inorganic semi- conductors the charge carriers may be electron vacancies or ‘holes’ as well as electrons. In a pure crystal the electrical conductivity varies with the absolute tempera- ture according to the equation u = uo exp - Ec/2kT (1) where o0 is a constant k is Boltzmann’s constant and Ec is the energy required to raise an electron to the conduction state. Much experimental effort par- ticularly with biological materials has gone into determining Ec from the temperature dependence of the conductivity.However considerable controversy has surrounded the interpretation of the value of Ec obtained from these experi- ments. Attempts have been made to link the value of Ec with the energies of the excited singlet or triplet states of the molecule.* The involvement of the triplet state in the conductivity appeared to be suggested by the close agreement which was sometimes found between Ec and the triplet-state energy. There have however always been strong theoretical objections to the validity of this sort of approach. A somewhat different approach is due to LyonslO who expressed the energy ’ A. S. Davydov ‘Theory of Molecular Excitons’ McGraw-Hill New York 1962. * (a) D. D. Eley G. D. Parfitt M. J. Perry and D. H. Taysum Trans.Faraday SOC. 1953 49 79; (6) D. J. Carswell J. Ferguson and L. E. Lyons Nature 1954 173 736; (c) D. C. Northrop and 0. Simpson Proc. Roy. SOC. 1956 A 234 124; (d) A. N. Terenin Radintekh. i Elektron. 1956 1 1127; (e) B. Rosenberg J. Chern. Phys. 1958 29 1108. G. G. B. Garrett ‘Semiconductors’ ed. N. B. Hannay Reinhold New York. 1959 p. 634. lo L. E. Lyons J. Chem. Soc. 1957 5001. 161 Semiconduction and Photoconduction of Biological Pigments of the conduction band in terms of the ionisation potential I and the electron affinity A of the molecule. EC = I - A - 2P (2) This expressed the energy required to remove an electron from one molecule and place it on another molecule at a distant point in the lattice. Pis the polarisa- tion energy produced by the interaction of both the electron and the positive ion with the surrounding molecules.It can be seen from equation (2) that Ec is not directly related to the energies of either the singlet or triplet excited states. In fact Ec could be above or below the first excited singlet state depending on the values of the parameters involved. Recent experiments using photoelectron emission and pulsed photoconductivity techniques have shown that the lowest conduction state in anthracene lies at about 3.7 ev which is greater than the lowest singlet-state energy of 3.15 ev.ll This result is particularly significant in view of the fact that previously measured values of Ec from conductivity experiments were found to be about 1.8-2.0 ev.6 Thus the conductivity measured in these experiments must have been extrinsic and the similarity with the triplet state energy (1.83 ev) was probablycoincidental.Further evidence to substantiate this conclusion is provided by the experi- ments of Becker Riehl and Baessler.12 They showed that the measured dark conductivity of anthracene was due to the injection of charge carriers by the electrode and concluded that the measured activation energy was the energy of the potential barrier existing at the electrode-crystal interface. Recent pulsed-light experiments have enabled the mobilities of the charge carriers in anthracene to be determined.13 Values in the range 0.1-1-0 cm.2/ sec./volt have been found for both electrons and holes. Mobility measurements are of considerable importance in that they provide evidence for the mechanism of conduction. The results obtained with anthracene are consistent with recent band-structure calculation^.^^ Until more mobility measurements have been carried out however it remains an open question how generally band theory may be applied to organic semiconductors.It seems likely that in some cases a mechanism in which charge carriers hop randomly from molecule to molecule may be more appropriate. It has long been known that photoconductivity excitation spectra usually resemble closely t.he crystal absorption spectra.6 As previously mentioned the absorption spectra of many organic crystals may be successfully explained in terms of exciton theory. Thus the first stage in photoconductivity is the produc- tion of the nonconducting exciton states. If as in anthracene the lowest con- ducting states lie above the lowest singlet state the difficulty arises of accounting l1 M.Pope and 3. Burgos MoZ. Crystals 1966 1 395. la G. Becker N. Riehl and H. Baessler Phys. Letters 1966,20,221. la (a) R. G. Kepler Phys. Rev. 1960 119 1226; (b) R. Raman L. Azarraga and S. P. McGlynn J. Chem. Phys. 1964 41 2516. l4 (a) J. I. Katz S. A. Rice S. Choi and J. Jortner J. Chem. Phys. 1963 39 1683; (b) R. Silbey J. Jortner S. A. Rice and M. T. Vala ibid. 1965 42 733. 162 Cherry for the extra energy needed to produce an electron-hole pair from singlet excitons. Experiments by Silver et aZ.15 suggested that dissociation may be achieved by the interaction of pairs of excitons. However more recent experi- ments utilizing the high light-intensities available from the ruby laser indicate that the dominant intrinsic mechanism for the production of charge carriers may be the photo-ionisation of singlet-exciton states.16 Experiments with light of somewhat higher energy than the conduction band have provided evidence that free charge carriers may also be produced by a direct transition to the conduction band.However an alternative explanation in terms of auto-ionisa- tion of a high-energy exciton state has also been suggested.17 The above intrinsic processes are of course of great theoretical interest. However in practice the photoconductivity of organic crystals is more often due to an extrinsic process. Dissociation of single excitons is achieved by inter- action with imperfections in the crystal particularly at the surface or the electrode-crystal interface.6 Apart from some charge-transfer complexes the dark and photo-currents observed in organic crystals are generally rather low.In anthracene for example about lo4 photons must be absorbed to produce one free charge carrier.13 How- ever donor and acceptor molecules adsorbed on the surface of organic semi- conductors may greatly increase the magnitude of the dark and photo-conduc- tivity. Recent experiments strongly indicate that these changes are due to induced carrier injection at the electrode-crystal interface and that the mobilities of the charge carriers are unaffected.18 3 Measurements with Biological Pigments A. Chlorophyll.-The chemical structures of chlorophyll a chlorophyll b and bacteriochlorophyll are well known (Figure 1) and the total synthesis of chloro- phyll a has been reported.lg All oxygen-producing organisms contain chlorophyll a.Higher plants green algae and euglenoids also contain the secondary pigment chlorophyll b. Bacteriochlorophyll occurs in the purple bacteria where the photosynthetic process is not linked with the production of molecular oxygen. The first measurements of photoconductivity in chlorophyll were made by Nelson.20 The sample was in the form of a film made by solvent evaporation of a chlorophyll solution. The photocurrents were of the order of 10-l2 amp. The excitation spectrum of methyl chlorophyllide was measured and found to resemble closely the optical absorption spectrum. Arnold and Maclay confirmed Nelson's studies and in addition observed M. Silver D. Olness M. Swicord and R. C. Jarnagin Phys. Rev. Letters 1%3,10,12. l6 (a) E. Courtens and A. Bergman Phys.Rev. 1967 156 948; (b) R. G. Kepler and J. Jortner Phys. Rev. Letters 1967 18 951. l7 (a) G. Castro and J. F. Hornig J. Chem. Phys. 1965,42,1459; (b) R. F. Chaiken and D. R. Kearns ibid. 1966 45 3966; (c) M. Pope and J. Burgos MoZ. Crystals 1967 3 215. (a) P. J. Reucroft 0. N. Rudyj and M. M. Labes J . Amer. Chem. SOC. 1963 85 2059; (b) P. J. Reucroft 0. N. Rudyj and M. M. Labes Mol. Crystals 1966,1,429. l9 (a) R. B. Woodward et al. J. Amer. Chern. SOC. 1960,82,3800; ( 6 ) M. Strell A. Kalojanoff and H. Koller Angew. Chem. 1960,72 169. 'O R. C. Nelson J. Chem. Phys. 1957 27 864. 163 Semiconduction and Photoconduction of Biological Pigments CH2 H II CH CH3 I CH2 -CH3 ' \ //' ' HC CH3 H H-C c=o I I CH2 I CO2CH3 02c 2 0 " 39 Bact er io chl or o phy 11 Fig. 1 Structural formulae of chlorophylls a and b and bacteriochlorophyll.164 Cherry photoconductivity in chlorophyll monolayers.21 They also showed that the photoconductivity of ,&carotene films could be sensitised in the region of the red chlorophyll absorption by the addition of small quantities of chlorophyll. Studies of a chlorophyll-carotene junction indicated that light induced a charge separation at the junction. Terenin and his associates have used the ‘condenser’ method to study photo- voltaic effects in chlorophyll.22 In this method the organic material is placed between semitransparent electrodes and is separated from the electrodes by thin transparent insulating sheets. Modulated light is passed through one electrode and produces electrons and holes close to the surface of the chlorophyll layer.A difference in the diffusion rates of electrons and holes across the sample produces a measurable photovoltage. If a constant voltage is applied to the condenser plates the signal will vary with direction of the applied field being larger when the illuminated electrode has the same sign as the more mobile charge carrier. In this way it was shown that the majority charge carriers in chlorophyll are positive holes. Terenin found that the spectral response of the photovoltage observed in microcrystalline chlorophyll had a peak at about 710 mp. The spectral response of the photovoltage in a layer of methyl chloro- phyllide was of interest in that the peak observed at 680 mp in a freshly prepared film shifted to 740 mp after several hours exposure to water vapour (Figure 2).During this time the magnitude of the response increased by a factor of three. This shift was almost certainly due to the crystallisation of the methyl chloro- phyllide in the presence of water.23 The dark conductivity of chlorophyll was studied by Rosenberg and Camiscoli using compressed microcrystalline powders.24 They found that the conductivity varied exponentially with temperature as in equation (1) and determined the value of Ec to be 1-12 ev for chlorophyll a and 1-44 ev for chlorophyll b. The activation energy for photoconduction was also measured in chlorophyll b and found to be 0-36 ev. Oxygen was found to increase both the dark and photo- conductivities of chlorophylls a and b over the values measured in vacuo. Such an effect is commonly found in organic semiconductors and in the light of recent experiments,ls appears more likely to be due to an increase in the concentration of free carriers rather than an increase in mobility as suggested by Rosenberg and Camiscoli.Eley and Snart25 determined Ec for a mixed film of chlorophyll a and b and found it to be 1.45 ev which is close to Rosenberg and Camiscoli’s value for crystalline chlorophyll b. They obtained however a much lower value (0.08 ev) for the photoconductivity activation energy. Measurements of complexes of chlorophyll and /%carotene with bovine plasma albumin indicated that these 21 W. Arnold and H. K. Maclay Brookhaven National Laboratory Symposia 1959 11 1. 22 (a) A. Terenin and E. Putzeiko J. Chim. phys. 1958 55 681 ; (b) A. Terenin E. Putzeiko and I. Akimov ibid. 1957 54 716; (c) A.Terenin E. Putzeiko and I. Akimov Discuss. Faraday SOC. 1959 27 83. 23 G. Sherman and S. F. Wang Nature 1966 212 588. 24 B. Rosenberg and J. F. Camiscoli J. Chem. Phys. 1961 35 982. z5 D. D. Eley and R. S. Snart Biochim. Biophys. Actu 1965,102 379. 165 Semiconduction and Photoconduction of Biological Pkments 400 500 600 700 800 900 171 p Fig. 2 Spectral response in vacuo of photo-e.m.f. in a methyl chlorophyllide (a & b) layer with microcrystals growing. Curve 1 layer deposited from a (C2H,),0+ petroleum solution; curve 2 after 10 hr. contact with H20 vapour at 20'; curve 3 after a repeated treatment by H20 vapour. The relative heights of the maxima are 1 :2 :3*5 in this sequence. (Reproduced by permission from A. Terenin E. Putzeiko and I. Akimov Discuss Faraday SOC.1959 27 83.) pigments reduced the semiconductivity activation energy of the protein. McCree26 investigated the photoconductivity of a stack of chlorophyll mono- layers laid down on a solid substrate. He concluded that an absorption of about lo9 photonslsec. was required to produce a current of one electron/sec. and hence that photoconductivity was a very inefficient process. Actually of more fundamental importance would be an estimate of the quatum efficiency for charge-carrier production but this involves the lifetime and mobility of the charge carriers which are unknown. If these parameters are rather low the quantum efficiency could be higher than is indicated by McCree's calculations. Nevertheless studies by the different investigators all suggest that chlorophyll is not a particularly efficient photoconductor.B. Carotenoids.-Polyene compounds play an important physiological role in both plants and animals. The yellow orange or red pigments occur naturally as the oxygen-free hydrocarbons the carotenes or as the oxygenated derivatives K. J. McCree Biochim. Biophys. Acta 1965 102 90. 166 Cherry the xanthophylls. The structure of one of the most commonly occurring caro- tenoids /%carotene is shown in Figure 3. This and several other carotenoid pigments have been synthesised in the lab~ratory.~’ 4’ 3’ 2’ 4 AI I - trans-# -carotene All-tmns- retinal Fig. 3 Structural formulae of p-carotene and retinal. Carotenoids are to be found in close association with chlorophyll in the chloroplast. Whether they have a mainly protective function or whether they participate more directly in the photosynthetic process is at present uncertain.Vitamin A is a lower isoprenologue of the monocyclic carotenoid pigments. The importance of vitamin A aldehyde (retinal) in the visual process is well estab- lished. The fkst suggestion that carotenoids should be semiconducting was made by Chynoweth and Schneider.28 Semi- and photo-conducting properties in p-carotene were subsequently observed by R~senberg.~~ The fkrotene was in the form of a glass consisting of a mixture of cis-trans-isomers which were produced by melting all-trans-p-carotene. Rosenberg found that the photo- conductivity excitation spectrum of the glass had its peak at 3500 A which is the region of absorption to the second excited singlet state (the ‘cis’ peak). This transition is forbidden in all-trans-/%carotene but allowed in the cis-isomers.Very little photoconductivity was observed in the region of the lowest excited singlet state. A similar result was obtained with a compressed powder of the 15,15’-cis-isomer. Rosenberg also found a large rectifying effect in the glass. When the illuminated electrode was positive the photocurrent was greater than when it was negative by a factor of about lo5 indicating that positive holes are much more mobile than electrons in this material. A number of determinations have now been made of the dark and photo- conductivity activation energies of various isomers of p-carotene and the values 0. Isler and P. Zeller Vitamins and Hormones 1957 15 31. 28 A. G. Chynoweth and W. G. Schneider J. Chem. Phys. 1954,22 1021. *O (a) B.Rosenberg J. Opt. SOC. Amer. 1958 48 581; (b) B. Rosenberg J. Chem. Phys. 1959 31 238. 167 Semiconduction and Photoconduction of Biological Pigments obtained are listed in the Table. The a.c. determination of Ec for all-trans-p- carotene appears to agree reasonably with the other results which are all from d.c. measurements. However since the specific resistance in the a.c. measurement was many orders of magnitude lower than in the d.c. it is likely that the a.c. losses are unconnected with the bulk conductivity. Table Semiconduction and photoconduction activation energies for various isomers of p-carotene Material Semiconduction Photoconduction activation energy activation energy (ev) (ev) All-trans-powder 1 *47a 0 ~ 3 7 4 ~ 15,15’-cis-Powder 1 *5ga 0 ~ 1 9 9 ~ p-Carotene glass 3 ~ 0 7 ~ 0.1 95 1 1,12 ; 1 1 ’ 12’-Di-cis-powder 2 ~ 7 6 ~ 0.OOc All-trans single crystal 1*45d - 13-Carotene film (a.c.measurement) 1.65” - a B. Rosenberg ‘Electrical Conductivity in Organic Solids’ ed. H. Kallman and M. Silver Interscience New York 1961 p. 291. B. Rosenberg,J. Chem. Phys. 1961,39,63. C B. Rosen- berg J. Opt. SOC. Amer. 1961 51 328. D. Chapman R. J. Cherry and A. Morrison Proc. Roy. SOC. 1967 A 301 173. e D. D. Eley and R. S. Smart Biochim. Biophys. Acta 1965,102 379. In order to account for the photoconductivity observed in the glass and’ 15,15’-cis-isomer Rosenberg suggested that charge carriers are formed from the triplet state and that in p-carotene the triplet state lies above the first but below the second excited singlet state i.e. at about 3.0 ev.This would explain why photoconductivity occurs mainly from the second singlet state since no net additional energy is required for intersystem crossing from this state to the triplet state. He also suggested that there is a potential barrier to intersystem crossing which accounts for the photoconductivity activation energy and claimed that the correlation of this activation energy with vibrational energies of the molecule gave support to his theory. Rosenberg also claimed that the energy assigned to the triplet state is corroborated by the semiconduction activa- tion energy of the glass (3-01 ev) which he identifies as the triplet-state energy and by Lewis and Kasha’s30 observation of a phosphorescent band at 3-1 ev in the related carotenoid lycopene. Several criticisms may be made of the above scheme.The correlation of the triplet-state energy with the semiconduction activation energy is doubtful (see previous section) and further Lewis and Kasha place the lowest triplet state at 2.25 ev in lycopene. The observed photoconductivity activation energies would be explained on any theory in which there existed a potential barrier to free carrier formation and thus these results do not specifically implicate the triplet state. 30 G. N. Lewis and M. Kasha J. Amer. Chem. SOC. 1944,66,2100. 168 Cherry The presence of oxygen increases both the dark and photo-conductivity of p - ~ a r o t e n e . ~ ~ ~ ~ ~ Oxygen decreases the dark conductivity activation energy of the all-trans-powder to 1.24 ev. The effect of the oxygen may be completely reversed by heating the /3-carotene above 80"c in vacuo.The effects of various other gases on the conductivity of p-carotene have also been examined.33 It was found that both electron-donor and electron-acceptor gases increased the conductivity. This result is consistent with the theoretical prediction that carotenoids should be both good electron donors and also good electron accept ors . /karotene is one of the few biological materials in which there has been an opportunity to make conductivity measurements with single crystals. Working with crystals rather than powders has the advantage of eliminating interparticle and surface effects as well as permitting studies of anisotropy to be carried out. All-trans-/%carotene may be crystallised from benzene-methanol solution in the form of hexagonal platelets.The crystal structure is monoclinic with two mole- cules per unit cell. The optical and electrical properties of these crystals have been studied by Chapman Cherry and Morrison.32 The value of Ec measured in a direction perpendicular to the ab plane of the crystal was 1-45 ev which agrees well with the previous measurements on powders and films. However in view of the very low conductivity (1O-l' ohm-l cm.-l at Z O O ) and the recent findings with anthracene discussed previously one would be hesitant about identifying this with an intrinsic value. The photoconductivity of the crystals showed little resemblance to that observed with powdered samples. The strongest photoconductivity occurred at energies well below the lowest absorption band while only relatively weak photoconductivity was observed in the region of strong absorption.With strongly absorbed light a rather weak rectifying effect was observed which indicated that the majority charge carriers were positive holes. Further studies with crystals of several related carotenoids as well as various isomeric forms of b-carotene revealed that a similar low-energy photoconduc- tivity band occurred in all these materials.35 The photoconductivity spectra obtained are shown in Figures 4 and 5. From these experiments it was tentatively concluded that the low-energy photoconductivity was most probably an intrinsic property of this type of molecule although the possibility of an impurity effect could not be completely eliminated. The optical transition responsible for the low-energy photoconductivity must be weak since it was not possible to observe any corresponding band in the absorption spectrum.Transitions which could conceivably be responsible for the observed photoconductivity include excitation of the triplet state a direct transition to a conducting state and excitation of charge-transfer states.11 31 B. Rosenberg J . Chem. Phys. 1961 34 812. 32 D. Chapman R. J. Cherry and A. Morrison Proc. Roy. SOC. 1967 A 301 173. 34 B. Pullman and A. Pullman 'Quantum Biochemistry' Interscience New York 1965 p. 440. 36 R. J. Cherry and D. Chapman Mol. Crystals 1967,3 251. R. J. Cherry and D. Chapman Nature 1967 215,956. 169 Semiconduction and Photoconduction of Biological Pigments Further studies are needed to determine whether in fact any of these transitions are involved in the low-energy photoconductivity of carotenoids.Two anomalous effects have been observed in /%carotene. RosenbergZ9 de- cn? Fig. 4 Photoconductivity excitation spectra of p-carotene. (A) Bulk Photoconductivity all-trans- @-carotene optical absorption (all trans) all- trans- 8-car0 tene surface electrodes sandwich cell -.-.-. 15,15’-cis-fl-carotene - - - - - - - - - (B) Surface Photoconduct ivity ----- 15,15’-cis-p-carotene (C) @-Carotene glass (Reproduced by permission from R. J. Cherry and D. Chapman Mol. CrystalsD 1967,3,25 1 .) ----- 170 Cherry A 20 - 10 - 0 20 - -I H 10 - 0 - 20.1 0 t u :j D l 9 30,000 20,000 10,000 I I cni’ Fig. 5 Photoconductivity excitation spectra of various carotenoids. (A) P-Apo- 8’-carotenal Bulk photoconductivity -.-. -. Surface photoconductivity (B) fl-Apo-8’-carotenoic acid ethyl ester.Surface photoconductivity (C) Canthaxanthin Surface photoconductivity (D) Lycopene Compressed powder (Reproduced by permission from R. J. Cherry and D. Chapman Mol. Crystals 1967,3,251.) scribed an e.m.f. which appeared across his p-carotene glass cells at high tem- peratures. He was unable to find any reasonable explanation of this effect. Chapman and Cherry observed current pulses which sometimes appeared when a d.c. voltage was applied to a p-carotene crystal.36 The variation of these pulses with light intensity temperature and voltage suggested that they were related to a similar effect which has been observed in inorganic semiconductor^.^^ 36 D. Chapman and R. J. Cherry Nature 1964,203,641. 37 D. C. Northrop P. R. Thornton and K. E. Trezise Solid State Electron.1964,7 17 171 Semiconduction and Photoconduction of Biologicol Pigments C. Other Pigments.-Cardew and measured the conductivity of ferric haem and found Ec to be 1-74 ev. Eley and S p i ~ e y ~ ~ measured activation energies for a series of porphyrins and dipyrromethanes. One of the principal conclusions of this work was that the central metal atom had little effect on the conductivity. The energy gaps were similar to those found in the structurally related phthalo- cyanines. Terenin Putzeiko and Akimov22c have measured photovoltages in a variety of pigments. They concluded that the excitation spectra were confined to the absorption spectra of the molecules and that the charge carriers were in most cases positive holes. Some preliminary studies of the conductivity of single crystals of all-trans- retinal and all-trans-retinoic acid have been attempted.40 However no measur- able dark or photo-currents could be detected in these crystals.Thus the conduc- tivity of these materials is probably even less than the value of 1O-l’ ohm-l cm.-l found with p-carotene. 4 Biological Aspects A. Photosynthesis.-In photosynthesis energy absorbed anywhere in an aggre- gate of chlorophyll molecules is transferred to a single reaction centre.41 The close packing of pigment molecules in the chloroplast lamellae and the high degree of order which is thought to exist at the molecular level suggested to some workers that the mechanism of energy transfer might be the diffusion of electrons and Studies with chlorophyll and other organic semiconductors however indicate that photoconductivity is too inefficient to be important in photosynthesis in this way.It is now generally considered that resonance transfer is the dominant mechanism by which energy is transferred to the reaction centre.*l Another possibility which has been considered is that electrons or holes may be transferred between different reactbn centres. The scheme suggested by Calvin43 is illustrated in Figure 6. In this scnerne the chlorophyll is visualised as being sandwiched between two layers one containing an electron acceptor possibly plastoquinone and the other containing an electron donor possibly a cytochrome. A quantum absorbed by the chlorophyll will migrate by resonance transfer to a suitable site near the quinone where electron transfer to the quinone will take place.Subsequently the electron vacancy or hole migrates along the array of chlorophyll molecules until it becomes adjacent to the cytochrome where an electron is transferred from the cytochrome to the chlorophyll. In this 38 M. H. Cardew and D. D. Eley Discuss. Furaduy Soc. 1959 27 115. 3g D. D. Eley and D. I. Spivey Trans. Furaduy SOC. 1962 58 405. 40 R. J. Cherry unpublished results. 41 (a) R. K. Clayton Ann. Rev. Plant Physiol. 1963 14 159; (6) L. N. M. Duysens Progr. Biophys. 1964 14 1. 42 (a) E. Katz ‘Photosynthesis in Plants’ ed. W. E. Loomis and J. Franck Iowa State College Press Ames Iowa 1949 p. 291; (b) D. F. Bradley and M. Calvin Proc. Nut. Acud. Sci. U.S.A. 1955 41 563. 43 (a) M. Calvin Rev. Mod. Phys. 1959 31 147; (6) M. Calvin J . Theor. Biol. 1961 2 258. 172 Cherry Acceptor Molecule (e.g Plastoquinone) Chlorophyll 1 ,a> el' Fig.6 Suggested scheme for electron-hole separation in photosynthesis. (In the above scheme an absorbed quantum produces an excitation in the chlorophyll which migrates to the electron- acceptor site. The excited electron is removed by the axeptor leaving a positive hole in the chlorophyll layer. The hole migrates to the electron donor where the transfer of an electron from the donor to the chlorophyll returns the chlorophyll to a neutral state.) way charge separation is achieved the oxidised donor becoming an oxidant and the electron in the quinone the reductant for subsequent reactions. An essential feature of this theory is that the interactions of donors and acceptors with the chlorophyll leads to a much higher quantum efficiency for photoconductivity in the chlorophyll layer than would be expected for chloro- phyll alone.A quantum efficiency of the order of unity has in fact been demon- strated in a model system composed of phthalocyanine as the semiconductor and chloranil as the electron acceptor.44 Calvin pointed out that the advantage of introducing photoconductivity into photosynthesis is that it enables the first oxidation and reduction reactions to occur on opposite sides of a lamellar structure. Thus the high-energy products are physically prevented from back-reacting with each other. Rabinowitch has remarked that separation could equally well be achieved if the reaction centres were located on opposite sides of the lambella but associated with the same chlorophyll molecule.In this case no conductivity would be required. Rabino- witch also raised a third possibility; namely that charge migration might occur I4 (a) D. R. Kearns and M. Calvin J . Chem. Plzys. 1958 29 950; (b) D. R. Kearns and M. Calvin J. Amer. Chem. SOC. 1961 83 2110; ( c ) D. R. Kearns G. Tollin and M. Calvin J. Chem. Phys. 1960 32 1020. 173 Semiconduction and Photoconduction of Biological Pigments along the chlorophyll layer as suggested by Calvin but that this process separated reaction centres which were on the same side of the chlorophyll layer. Until we possess a more detailed knowledge of the molecular structure of chloroplast lamellae there appears to be little reason for preferring one or other of these various possibilities. An important criticism of any charge-migration hypothesis is that the inter- action between chlorophyll molecules in the chloroplast may be so low that little or no conduction can occur.& Clayton4 has discussed the state of chlorophyll in vivo in some detail and concludes that there is probably some interaction between molecules but considerably less than in the crystalline state.It is not possible at present to decide whether the interactions are sufficiently low to rule out semiconductive mechanisms. Certainly crystallinity is not essential for the observation of semiconductive and photoconductive behaviour in organic s0lids.4~ The experimental evidence that photoconduction is important in photo- synthesis is largely inconclusive. Semiconductive properties have indeed been observed in dried chloroplast and chromatophore preparations ;21,48,49 it may always be argued however that the molecular interactions are changed during the drying process.The photocurrents observed in these preparations are always very low but this could be due to the insulating effect of lipid components and does not necessarily imply a low quantum efficiency for the production of free carriers. Various other phenomena including delayed light emission,60 light-induced spectral changes,49 and electron spin resonance s i g n a l ~ ~ s ~ ~ have been cited as evidence for the presence of free electrons and holes in the chloroplast. Dis- cussion of these rather complex effects may be found in several review article^.^^,^^ In geneiral other interpretations which do not require the presence of free charge carriers may be made of these phenomena.B. Vision.-In the visual process the absorption of light by the visual pigment leads to the appearance of a nervous impulse. Most visual pigments investigated have been found to consist of a complex of a protein opsin with retinal. Of the various possible isomers of retinal only the 11-cis-form has the ability to cam- bine with opsin to produce spectroscopically normal rh~dopsin.~~ When light is 45 E. Rabinowitch Discuss. Faraday SOC. 1959 27 161. 46 R. K. Clayton ‘Molecular Physics in Photosynthesis’ Blundell New York 1965. 47 J. Kommandeur G. J. C. Korinek and W. G. Schneider Canad. J. Chem. 1958,36 513. See also ref. 22 (c) and 29 (6). 48 (a) W. Arnold and H. K. Sherwood Proc. Nat. Acad. Sci. U.S.A. 1957 43 105; (b) S. Ichimura Biophys. J. 1960 1 99. 40 W.Arnold and R. K. Clayton Proc. Nat. Acad. Sci. U.S.A. 1960 46 769. 6o (a) W. Amold J. Phys. Chem. 1965 69 788; (6) G. Tollin E. Fujimuri and M. Calvin Proc. Nat. Acad. Sci. U.S.A. 1958 44 1035. 51 R. H. Ruby I. D. Kuntz and M. Calvin Bull. SOC. Chim. biol. 1964 46 1595. 54 B. Commoner ‘Light and Life’ ed. W. E. McElroy and B. Glass Johns Hopkins Balti- more Md. 1961 p. 356. 53 R. Hubbard and G. Wald J. gen. Physiol. 1952 36 269. 174 Cherry absorbed by rhodopsin the 1 l-cis-retinal is isomerised to the all-trans-form. Following this a number of thermal rearrangements of the rhodopsin molecule occur before the retinal becomes detached from the opsin. Subsequently the all-trans-retinal undergoes a regenerative cycle during which it is converted back into the 1 l-cis-form and re-attached to the protein.From the time constants of these various processes it is clear that the nervous impulse is generated very early on i.e. before the retinal becomes detached from the opshM The pigment molecules exist in what has been termed a ‘quasi-crystalline’ arrangement in the lamellae of the rods and c0nes.5~ This has suggested to some workers that a photoconductive mechanism might account for the production of an electrical signal following light a b s o r p t i ~ n . ~ ~ ~ ~ ~ ~ Rosenberg and his associate^^^(^-^) have used a cell consisting of a thin layer of fl-carotene glass between two conducting glass plates as a model of a visual receptor. With a suitable polarity of applied voltage it was possible to produce photoconductive and photovoltaic effects which occurred in opposite directions and with different time constants.These effects were seen to have something in common with the PI1 and PI11 components of the electr~retinogram.~~ Because the photocurrent depends on light which has passed through the bulk of the b-carotene while the photovoltage depends on light absorbed at the electrode- /3-carotene interface the spectral dependence of these two effects is different. Due to this and to the different time constants of the two effects the transient response of the cell varied with the wavelength of the incident light. Rosenberg suggested that a similar process might provide the basis of colour discrimination in the visual receptors. The above approach appears to take little account of the membranous structure of the visual receptors. It is difficult to see how some of the effects observed in the /3-carotene cell could occur in a membrane since this would be much too thin to show differential absorption between one side and the other.It has also been shown that the primary colour receptors are located in different cones.59 This is contrary to Rosenberg’s theory which predicts that individual cones should be capable of colour discrimination. Dingle and LucyGo have considered the role of the membrane in more detail and suggest that two retinal molecules back-to-back could provide a molecular pathway for the transmission of electrons through the membrane. They suggest 64 (a) G. A. J. Pitt and R. A. Morton Biochem. SOC. Symp. 1960,19 67; (b) M. F. Moody Biol. Rev. 1964 39,43; (c) H. H. Seliger and W. D. McElroy ‘Light Physical and Biological Action’ Academic Press New York 1965.s5 G. Wald Nature 1959 184 620. s0 H. Fernandez-Moran Rev. Mod. Phys. 1959 31 319. 67 (a) B. Rosenberg Photochem. and Photobiol. 1962 1 117; (6) B. Rosenberg R. J. Heck and K. Aziz J. Opt. SOC. Amer. 1964,54 1018; (c) B. Rosenberg R. J. Heck and K. Aziz Photochem. and Photobiol. 1965 4 351; (d) W. A. Hagins and W. H. Jennings Discuss. Faraday SOC. 1959 27 180. R. Granit ‘Receptors and Sensory Perception’ Yale University Press New Haven Conn. 1955. 69 W. B. Marks W. H. Dobelle and E. F. McNichol Science 1964 143 1181. 6o J. T. Dingle and J. A. Lucy Biol. Rev. 1965 40 422. 175 Semiconduction and Photoconduction of Biological Pigments that the all-trans-molecule may have a much higher electron mobility than the cis-isomer.Thus isomerisation following light absorption results in an increase in the electrical conductivity of the membrane. This may be compared with Rosenberg’s view that it is the cis-isomer which produces a photoconductive signal isomerisation to the insensitive all-trans-form being largely a protective device.29 In this respect it may be noted that single-crystal studies of p-carotene throw some doubt on the premise that the cis-isomer is necessarily more photo- conductive than the a l l - t r ~ n s . ~ ~ In any case it is by no means clear how the properties of a retinal-protein complex can be related to the properties of a carotenoid crystal. In fact the complex needs to be a considerably more efficient photoconductor than the crystal if photoconductivity is to have any importance in vision.Attempts to demonstrate photoconductivity in visual receptors include measurements with dried films of rod outer segments.s1 These experiments are subject to the same difficulties of interpretation as those with dried chloroplast preparations. Falk and Fatt have observed impedance changes in packed suspensions of frog rods following flash illumination.62 The response could be resolved into three components. One of these was shown to be due to the uptake of H+ by rhodopsin producing a change in buffer conductivity. A second component was thought to be due to an increase in ionic permeability of the rod surface mem- brane. It was suggested that the third component could be due to a change in conduction within the solid surface structure of the rod. The recent discovery of the so-called early receptor potential (E.R.P.) may be pertinent to the present disc~ssion~~.The E.R.P. which may be observed follow- ing illumination of the retina with intense light flashes has the following proper- ties; its rise time is of the order of microseconds it persists when the cells of eyes are depolarised and when the eye is frozen it resists anoxia and it has in the albino rat the action spectrum of rhodopsin. Thus the E.R.P. may represent an early and perhaps direct manifestation of the absorption of light by rhodopsin. Most of the studies of the E.R.P. have been made with extracellular recordings. More recently Smith and Brown64 have measured a similar early potential by means of intracellular recordings in the lateral eye of Limulus polyphemus. This potential they term a photoelectric potential (P.E.P.).The P.E.P. has many of the characteristics of the E.R.P. although it has not yet been shown to have the same action spectrum as the rhodopsin absorption. The magnitude of the P.E.P. depends on the membrane potential and it almost certainly arises in the cell membrane. a B. Rosenberg R. A. Orlando and J. M. Orlando Arch. Biochem. Biophys. 1961 93 395. 6a G. Falk and P. Fatt J. Physiol. 1963 167 36P; 1966 183 211; 1966 185 1OP; 1966 186 104P. e3 (a) K. T. Brown and M. Murakami Nafure 1964 201 626; (b) R. A. Cone ibid. 1964 204 736; (c) G. B. Arden and H. Ikeda ibid. 1965 208 1100. 64 T. G. Smith and J. E. Brown Nature 1966 212 1217. f 76 Cherry It is tempting to relate these early potentials to the production of free electrons and h0les.6~ The variation of P.E.P.amplitude with membrane potential strongly indicates a change in conductance in the cell membrane. Since the P.E.P. persists when the cell is depolarised by replacing sodium with potassium ions it is clear that changes in ionic permeability are ruled out and it appears rather unlikely that ions are involved at all. It seems most plausible that electrons or protons are produced following light absorption although there is at present little basis for choosing between these possibilities. It should be noted when considering these possibilities that although there are strong indications that the E.R.P. and P.E.P. are fundamentally concerned with the generation of the visual impulse there is not yet conclusive evidence that this is the case.64 Further light-induced early potentials have been observed in a variety of pigmented tissues.65 The properties of these potentials closely resemble the E.R.P.and there may well exist a common fundamental mechanism. C. Discussion.-Studies with crystals films and powders of biological pigments have largely been undertaken to provide a basis for considering the role of the pigment in the more complex biological situation. However the extrapolation of these measurements to biological systems presents a serious problem. Although much emphasis has been given to the ordered nature of photosynthetic and visual receptors it is clear that the differences between these structures and crystals are considerable. A further difficulty is that as yet only somewhat limited progress has been made towards an understanding of the solid state properties of biological pigments.These difficulties may be illustrated by considering the question of whether absorbed photons provide sufficient energy for the production of free charge carriers in vision and photosynthesis. The semiconductive activation energies obtained with the relevant pigments may be intrinsic but could also be deter- mined by defects in the crystal or by the properties of the electrode-crystal inter- face. Further the corresponding energy in the biological system may be affected by the environment of the pigment molecule. Because of these uncertainties it is not possible to give a reasonable answer to this question at present. One point which does clearly emerge is that there is no evidence for a high photoconductive efficiency in any of the pigments examined.In contrast both photosynthetic and visual receptors are very efficient in utilising the absorbed light energy. Thus photoconductive mechanisms can only be important in these systems if the photoconductive efficiency of the pigment is in some way enhanced. This could conceivably be achieved by donor-acceptor interactions with other molecules in the receptors. The concepts derived from studies of crystals have led to some interesting speculations about the functioning of highly ordered biological structures. One would expect these concepts to need modification eventually to take into account the detailed molecular structure of the biological system. In photosynthetic and G. B. Arden C. D. B. Bridges H. Ikeda and I. M. Siege] Nature 1966 212 1235. 177 Semiconduction and Photoconduction of Biological Pigments visual receptors the pigment molecules are organised in lipoprotein membranes.The transfer of an electron across the membrane may involve only one or two molecules. This process may contain features of both semiconduction and of donor-accep t or interact ion. There has of late been considerable interest in artificial phospholipid mem- braneP and recently preliminary attempts to incorporate pigments in these membranes have been de~cribed.~' It will be of interest to see whether the energy- transfer processes which have been suggested for biological membranes can be observed in these model systems. Studies of these artificial membranes may give insight into the processes which could occur in biological receptors. 66 A. H. Maddy C. Huang and T. E. Thompson Fed. Proc. 1966,25,935. ~3' R. B. Leslie and D. Chapman Chem. and Phys. Lipids 1967,1 143. 178

ISSN:0009-2681    

DOI:10.1039/QR9682200160

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Liquid Crystals as Solvents in Nuclear Magnetic Resonance By G. R. Luckhurst DEPARTMENT OF CHEMISTRY THE UNIVERSITY SOUTHAMPTON SO9 5NH 1 Introduction The molecules of solvents generally used in high resolution nuclear magnetic resonance (n.m.r.) do not adopt a preferred orientation when the sampIe is placed in a magnetic field and in this sense the solvents are isotropic. Analysis of the n.m.r. spectrum of a substance dissolved in such a solvent yields a set of chemical shifts and spin-spin coupling c0nstants.l These constants have proved to be invaluable to the organic chemist in the determination of the molecular structure of an unknown compound. However although the measured spin- spin coupling constants and chemical shifts are isotropic there are magnetic interactions within a molecule which are anisotropic that is their magnitude depends on the orientation of the molecule with respect to the magnetic field.Provided the molecule is undergoing rapid isotropic motion these interactions do not affect the positions of the lines within a spectrum; they can only influence the widths. If however the molecular motion can be restricted so that the molecule adopts a preferred orientation the positions and number of lines in the spectrum will depend on the anisotropic interactions and valuable information is gained. As we shall see it is possible to determine relative bond lengths bond angles and signs of spin-spin coupling constants. The nematic mesophases of several liquid crystals2 have proved to be ideal solvents to restrict the motion of solute molecules. This Review is intended for the chemist who wishes to understand the physical basis and significance of high-resolution n.m.r.experiments with use of liquid crystals as solvents. While the Review does not treat the subject with complete rigour it should enable the reader to tackle with some confidence the original papers as well as the excellent article by Buckingham and McLauchlan3 who deal with the theory in considerable detail. The reader is assumed to be con- versant with the article by Bovey;4 detailed knowledge of the nature of liquid crystals or the theory of nuclear dipolar coupling is not presupposed. 2 The Dipolar Coupling The magnitude of an isotropic spin-spin coupling constant (henceforth referred to as the spin-spin coupling constant) measured in n.m.r. depends on the nature J.W. Emsley J. Feeney and L. H. Sutcliffe ‘High Resolution Nuclear Magnetic Resonance Spectroscopy’ Pergamon Press London and New York vol. 1 1965. G. W. Gray ‘Molecular Structure and the Properties of Liquid Crystals’ Academic Press London and New York 1962. A. D. Buckingham and K. A. McLauchlan ‘Progress in n.m.r. Spectroscopy’ ed. J. W. Emsley J. Feeney and L. H. Sutcliffe Pergamon Press London and New York vol. 2,1967. F. A. Bovey Chem. Eng. News 1965,43,98. 179 Liquid Crystals as Solvents in Nuclear Magnetic Resonance of the bonds joining the two nuclei and this is one reason why n.m.r. is invalu- able to the chemist in molecular-structure determinations. However the two nuclei also interact through space rather like bar magnets and this interaction is known as the dipole-dipole or dipolar ~oupling.~ Suppose we measure the nuclear resonance spectrum of two protons A and B having the same chemical shift with the magnetic field H pointing along the interproton axis.In the absence of dipolar coupling the two protons would come into resonance at H, but because of the dipolar interactions the field felt by nucleus B is not equal to H but is instead H AH depending on the orientation of the nuclear spin of A . Thus the protons come into resonance at Ho + A H and at H - A H and the effect of the dipolar coupling is therefore to split the single line at Ho into a doublet centred on H, with a separation 2AH. In general the frequency separation between these lines for two protons iss Av= 3* (3 cos2+ - 1) 4nr3 where YH is the proton gyromagnetic ratio Y is the interproton distance and 4 is the angle between this axis and the magnetic field This formula contains a spin exchange contribution and so requires modification if the dipolar coupling is between two different nuclei e.g.carbon-13 and hydrogen; the doublet splitting is then given by y . . A A v = - zY3 (3 cosz+ - 1) 2.rrr3 The factor of $ has vanished because the resonance frequency of nucleus i is different from that of nucleusj (see ref. 6 p. 98). It is interesting to substitute some numbers into equation (l) for we find the separation A v for two protons 1 A apart to be 360.33 kHz when the field is along the interproton axis. One might therefore expect these large dipolar split- tings to dominate the n.m.r. spectrum for the largest proton spin-spin coupling rarely exceeds 20 Hz.The absence of any large splitting is however due to the rapid isotropic Brownian motion of the molecules. This movement means that is continually changing and to calculate the observed splitting (Av) from equation (1) we must take a time average over all molecular orientations that is ( A v ) = 3y2Hh - (3 cos24 - 1) 4i7r (3) where the angular brackets ( ) are intended to denote an average over all values of $. When we say that a molecule is moving isotropically we mean that all orientations are equally probable so (3 cos2# - 1) is zero and the dipolar splitting must vanish. On the other hand if all molecular orientations are not equally probable that is the motion is anisotropic then (3 c o s 2 ~ - 1) E. R. Andrew ‘Nuclear Magnetic Resonance’ Cambridge Univ. Press 1955 ch. 6.A. Abragam ‘The Principles of Nuclear Magnetism’ Oxford Univ. Press 1961 p. 217. 180 Luckhurst would not vanish and we would observe a splitting of the spectral line. Since the separation of the lines depends on r we would have a method of measuring both internuclear distanccs and indirectly bond angles. One obvious way of restricting the motion of a molecule is to freeze the com- pound and n.m.r. studies of powders but preferably single crystals have yielded internuclear di~tances.~ The technique is not generally applicable because the dipolar coupling to nuclei in neighbouring molecules often serves to broaden the lines and obscure the splitting. In some instances the molecules are well separ- ated by non-magnetic nuclei e.g. in inorganic hydrates and the doublet splitting can then be resolved.8 The intermolecular dipolar broadening can also be removed if the motion of the molecules is sufficiently rapid; however -to retain the splitting from intramolecular dipolar coupling the motion must be aniso- tropic.An apparently attractive method of restricting the motion of a molecule pos- sessing a dipole moment is to apply a large static electric field to the sample. Under these conditions the molecules would tend to be aligned with their dipole moments parallel to the electric field and the n.m.r. spectrum should reflect this preferred orientation. Buckingham and McLauchlang have shown that the application of an electric field to p-nitrotoluene does indeed result in a change in its n.m.r. spectrum. The increase in the separation between certain lines was attributed to an unaveraged dipolar coupling although this interpretation has been questioned by Sears and Hahnlo who have employed pulsed n.m.r.tech- niques. In view of this difficulty and the expected low degree of alignment it is fortunate that Saupe and Englertll have demonstrated the very high degree of alignment that can be obtained by using the nematic mesophase of a liquid crystal as a solvent. 3 Liquid Crystals A normal organic solid such as benzoic acid melts sharply at a given temperature to give a clear liquid and continued heating does not produce any further change until charring! However if the substance is capable of existing as a liquid crystal the sharp transition at the melting point produces a turbid liquid but on further heating there is a second phase transition when a clear isotropic liquid is produced.As we shall see it is the cloudy mesophase or liquid crystal- line state which interests us. Although the properties of the mesophase are well characterised experi- mentally no unique theory has been suggested to explain these properties. Indeed two opposing theories have been described; on the one hand there is the swarm theory suggested by Bose12 and on the other Zocher's distortion theory l3 ' R. E. Richards Quart. Rev. 1956 10 480. Z. M. El Saffar J. Chem. Phys. 1966 45,4643. A. D. Buckingham and K. A. McLauchlan Proc. Chem. SOC. 1963 144. lo R. E. J. Sears and E. L. Hahn J. Chem. Phys. 1966 45 2753; 1967 47 348. l1 A. Saupe and G. Englert Phys. Rev. Letters 1963 11 462. l2 E. Bose Z . Physik 1909 10 32. l3 H. Zocher Trans. Faraday SOC.1933 29 945. 181 Liquid Crystals as Solvents in Nuclear Magnetic Resonance Fortunately both theories give the same result when used to describe the posi- tions of the lines in a magnetic resonance spectrum.14 Since the distortion theory has been used as a basis for the description of the orientation of molecules in n.m.r. experiments we shall persist in its use. Three types of liquid crystal have been characterised and they differ in their local molecular order although the most important liquid crystal for magnetic resonance experiments is the nematic mesophase. We now describe the general properties of liquid crystals. The molecules of compounds which form a nematic mesophase are normally long and contain benzene rings for example p-azoxyanisole whose nematic range extends from 118" to 135".The intermolecular forces together with a co- operative effect tend to arrange the molecules with their long axes parallel. As the lateral distance between molecules increases the tendency for the long axes to be parallel decreases; however in the nematic mesophase the ordering is considerable extending over many thousands of molecules. The single degree of local ordering is shown in Figure 1. In the absence of any external restraint Figure 1 The local molecular arrangement for a nematic mesophase for example p-azoxyanisole the orientation of the local order in one region of the sample with respect to that in another region is completely random although the direction of order does change continuously (ref. 2 p. 76). When a magnetic field greater than lo00 gauss is applied to the sample however the long axes of the molecules tend to become aligned parallel to the magnetic field.15 It is useful to think of this effect in the following way.The magnetic field interacts with anisotropic components of the magnetic susceptibility in much the same manner as an electric field interacts with an electric dipole moment to produce partial alignment. The calculated degree of alignment for a single molecule of for example p-azoxy- anisole in a field of 20 ko is extremely small. The anisotropy in the magnetic susceptibility is however enhanced by the large degree of local order and this leads to the alignment of molecules in the mesophase. The viscosity of a nematic mesophase is comparable with that of a normal liquid such as benzene which implies that the molecular motion is sufficiently rapid to average out the inter- molecular dipolar coupling.Although the motion of the solute is also rapid the l4 G. R. Luckhurst Mol. Cryst. 1967 2 363. l5 W. Maier 2. Naturforsch. 1947 2a 458. 182 Luckhurst anisotropic environment is capable of restricting the rotational freedom of the solute and causing partial alignment. The molecules in a smectic mesophase are also aligned with their long axes parallel to one another but the local structure contains an extra degree of order which is illustrated in Figure 2. The additional order makes the structure of the Figure 2 The local structure of a smectic mesophase for example ethyl p-azoxybenzoate smectic mesophase similar to a solid and as we might expect the mesophase is rather viscous.By implication the molecular motion must be relatively slow and so the intermolecular dipolar coupling should broaden the n.m.r. lines; however no magnetic resonance experiment in which a smectic liquid crystal was used as a solvent has been reported. Again the molecules making up compounds which form smectic mesophases are rather long and narrow (e.g. salts of oleic acid) and frequently they contain aromatic rings carrying terminal substituents (e.g. ethyl p-azoxybenzoate). Cholesteryl esters form a third type of liquid crystal (the cholesteric meso- phase). Although cholesterol itself melts to an isotropic liquid most of its esters give liquid crystals whose structure is shown in Figure 3. Each molecule is represented by a single line and within a given layer the long axes of the mole- cules are parallel but the direction of alignment changes continuously on passing from one layer to the next.The resulting helical structure is responsible for the high optical activity of the cholesteric mesophase. The reason for the helix is hinted at by the following simple experiment. When right- or left-handed bolts are stacked with their long axes parallel the threads interact in such a way that the long axes change direction from one layer to the next and a helix results.ls l6 E. Sackmann S. Meiboom and L. C. Snyder J. Amer. Chem. SOC. 1967,89 5981 183 Liquid Crystals as Solvents in Nuclear Magnetic Resonance Figure 3 The helical structure of a cholesteryl liquid crystal for example cholesteryl propionate Because of the small anisotropy associated with the magnetic susceptibility of predominantly aliphatic compounds and the complex structure of the cholesteric mesophase it is not possible to say if solutes could be aligned with this meso- phase.In fact electron resonance experiments using cholesteryl propionate as a solvent for vanadyl acetylacetonate have failed to indicate any alignment,17 suggesting that any anisotropy is averaged to zero by the helical structure. Another experiment with bolts suggests a method of eliminating the twist for if one stacks bolts with alternate left- and right-handed threads their axes remain parallel. Indeed appropriate mixing of two cholesteric compounds with opposite rotations produces a solvent which does align benzene when placed in a magnetic field.16 Often aqueous solutions of certain soaps (salts of long-chain sulphonic acids) exhibit anisotropic behaviour.The local order is a function of composition and can resemble either that of a smectic or nematic mesophase,18 The viscosity of the mesophase is often extremely high and it is surprising that molecules dis- solved in them do give high-resolution n.m.r. spectra. It is even more surprising to find that methyl alcohol is highly aligned in such a solvent system.lg We shall not mention this class of liquid crystal solvent again although they could be particularly valuable for orienting lyophilic solutes. 4 Experimental Methods Although mixtures of cholesteric mesophases may form a useful class of solvents for alignment experiments present results suggest that the broad background l7 G. R. Luckhurst Thesis Cambridge England 1965 ch.7. l8 V. Luzzati H. Mustacchi and A. Skoulios Discuss. Faraday SOC. 1958 25 43. K. D. Lawson and T. J . Flautt J. Amer. Chem. Sor. 1967 89 5489. 184 Luckhurst n.m.r. signal may limit their usefulness in this field. We shall therefore concern ourselves entirely with nematic liquid crystal solvents. Addition of solutes to a liquid crystal lowers both the melting point and the nematic-isotropic transition temperature which may often become lower than the m.p.;20 in other words no mesophase exists or the mesophase becomes monotropic with respect to the solid. Clearly if the structure of the solute is similar to that of the solvent it will be possible to add a large amount of solute and under favourable conditions 25 moles % may be dissolved. For this and indeed other reasons it is necessary to use a very sensitive n.m.r.spectrometer to measure the spectra. When the spectrometer sensitivity is not sufficient one can use time-averaging techniques in which many spectra are added to the memory of a computer thus decreasing the random noise while increasing the signal intensity. Unfortunately although such techniques do produce improved results they are time-consuming and demand high instrument stability.21 The melting points of most nematogenic compounds especially those which are commercially available are above room temperature. The solution must therefore be heated in the n.m.r. probe and great care must be taken both to eliminate thermal gradients within the sample and to ensure a constant tem- perature (&0.1"). The necessity arises because the alignment is extremely temperature-dependent,22 and since the positions of the n.m.r.lines are related to the alignment they are temperature-dependent. Thus molecules in different regions of the sample would have different n.m.r. transition frequencies and because of molecular motion these frequencies would be modulated and so cause line-br~adening.~~ Recent work by Spiesecke and Belli~n-Jourdan~~ and by D e m ~ s ~ ~ has largely eliminated the need for an instrumental solution to the problem. They find that by mixing suitable nematic liquid crystals they can lower the melting point of the solid mixture to about 30" and still obtain a mesophase. In conventional n.m.r. spectrometers the sample is rotated about an axis perpendicular to the magnetic field in order to reduce field inhomogeneities.Since the molecules of a nematic mesophase are aligned parallel to the magnetic field such rotations should destroy the alignment and in the majority of experi- ments which have been reported the sample was not spun. In certain cases however sample rotation has been found to reduce the linewidth although it is not clear whether the improved resolution results from a reduction of magnetic field inhomogeneity or temperature gradients.26 For successful experiments the speed of spinning must remain constant and in the region of 5 to 10 c. per second. Because the sample is rotated about the axis parallel to the magnetic field in n.m.r. spectrometers having superconducting solenoids these instruments may prove to be ideally suited for liquid-crystal studies. 2o J. S. Dave and M.J. S. Dewar J . Chem. SOC. 1954,4616. 21 L. C. Snyder and S. Meiboom J. Chem. Phys. 1967 47 1480. 22 H. C. Longuet-Higgins and G. R. Luckhurst Mol. Phys. 1964 8 613. 23 A. Saupe 2. Naturforsch. 1965 20a 572. 24 H. Spiesecke and J. Bellion-Jourdan Angew. Chem. Internat. Edn. 1967 6 450. 25 D. Demus 2. Naturforsch. 1967 22a 285. 26 P. Diehl and C. L. Khetrapal Mol. Phys. 1968 14 283. 185 Liquid Crystals as Solvents in Nuclear Magnetic Resonance 5 Examples of the Theory By starting with the appropriate spin Hamiltonian and describing the average orientation of the molecule with an ordering matrix2' or with certain motional constants,28 it is possible to derive all the results one needs to analyse n.m.r. spectra of solutes in a nematic mesophase. This approach is valuable because it treats the problem with mathematical rigour but in this Review we shall con- centrate on the physical aspects of the problem.These are best dealt with by considering specific examples although we shall begin by introducing the concept of the ordering matrix developed by Sa~pe.~' A. The Ordering Matrix.-The analysis of the spectra of oriented molecules demands a mathematical description of the orientation and a convenient point to begin is the formula (3) for the average splitting which we write as In this equation we have introduced the direction cosine Iza which is the cosine of the angle between the two axes z the direction of the magnetic field and a the interproton axis. We now define a quantity Sa5 by the equation s a 5 = 8 <3lzalza - 1> ( 5 ) where S is subscribed with two a's because a occurs in both direction cosines.Since the value of (l~alza) is a measure of the alignment of axis a with respect to the magnetic field S, also gives a measure of the alignment but in a more con- venient form. When the motion is isotropic Saa is zero whereas if a is completely aligned along the z axis the maximum value of S, is unity since4 is zero. If a is perpendicular to z the minimum value is -&. The values of Sbb and S, are given by equations analogous to (5). It is possible to define quantities which involve the direction cosines of different axes for example S u b and in general Sij = < 3 I J j z - S i j ) (6) where 6 is the Kronecker delta taking the value 1 when i is the same as j and 0 when i = j . The S values can be arranged in a square array which is known as the ordering matrix.The properties of the matrix have been discussed elsewhereF8 and we shall mention just two of these. The sum of the diagonal elements (the trace of the matrix) is zero and this is proved simply by adding equations for s,, Sbb and sCc to give s a a + s b b + s c c = 3 (lzalza + I z b l z b + l z c l z c - 1 = (7) since the sum of 12za 12& and lzzc is unity.ag Combination of equations (4) and (5) gives 27 A. Saupe 2. Narurfursch. 1964 19a 161. 28 L. C. Snyder J. Chem. Phys. 1965 43,4041. 28 H. Margenau and G. M. Murphy 'The Mathematics of Physics and Chemistry' D. Van Nostrand Co. Inc. Princeton New Jersey 1955 2nd edn. vol. 1 p. 139. 186 Luckhurst (Av) = 3Y2Hh - s, 2.rrr3 which is a value for (dv) in terms of an ordering matrix for an axis system in which one axis is parallel to the interproton vector.Often such an axis system does not reflect the geometry of the molecule which really determines the value of the ordering matrix. It is useful therefore to be able to relate an ordering matrix in one axis system a b and c to that in another system a 8 and y. Because S transforms as a second-rank tensor the required relationship is C i j = a B. Equivalent Protons.-We begin by considering 1,2,3,5-tetrachlorobenzene which for our purposes can be regarded as a simple two spin system. Because the protons are equivalent the isotropic n.m.r. spectrum contains a single line. If however the spectrum is measured in the nematic mesophase then because the molecule is oriented with respect to the magnetic field we would expect to find the single line split into a doublet.The n.m.r. spectrum has been measured in 4,4'-di-n-heptyloxyazoxybenzene,30 a nematic liquid crystal and Figure 4 I28 Hz 0 128 Hz 1 I I 30 mG 0 30 mG + H Figure 4 The anisotropic n.m.r. spectrum of 1,2,3,5-tetrachlorobenzene in 4,4'-n-heptyloxy- azoxybenzene at 82" [Reproduced by permission from 2. Naturforsch 1964 19a 172) shows that we do indeed observe two lines which are separated by 256 Hz. The absence of the solvent spectrum is general and at first sight rather surprising because the n.m.r. spectrum of the isotropic phase is readily observed. The explanation lies in the fact that 4,4 '-di-n-heptyloxyazoxybenzene contains 38 protons; there are therefore 238 spin levels and if each nuclear transition had a different frequency the n.m.r.spectrum would be far too complex to measure. Fortunately most of the transitions are multiply degenerate and in fact it is fairly easy to observe the n.m.r. spectrum in the isotropic phase. The effect of so G. Englert and A. Saupe 2. Naturforsch. 1964 19a 172. 187 Liquid Crystah as Solvents in Nuclear Magnetic Resonance aligning the solvent molecules is to introduce a dipolar coupling between the protons which splits this degeneracy and so reduces the intensity of each peak to such an extent that they vanish in the noise. According to the theory we have developed the observed splitting is given by 360.33 r3 ( A v ) = - S b b where we have substituted for the constants in equation (9) and ( A v ) is given in kHz provided r is measured in A. In the equation b is the interproton axis and the complete molecular axis system is shown in Figure 5.Provided Cl Figure 5 The molecular axis system for 1,2,3,5-tetrachiorobenzene all in-plane orientations of the molecule are equally probable the ordering matrix will be axially symmetric in the axis system a b c. Further since the trace of the S matrix is zero we have Substitution of the interproton distance of 4.30 8 into equation (10) combined with equation (11) yields a value of 0.113 for See. Unfortunately the simple n.m.r. experiment does not yield the absolute magnitude of the splitting and so we cannot determine the sign of S, except when S is greater than 8. It must then be positive since the minimum value of S is - Q. Other experiments show that planar molecules tend to be aligned with their planes parallel to the magnetic field and so we can be sure S, is negative.Since S is both temperature- and concentration-dependent the measurement of the n.m.r. spectra of molecules in liquid crystals provides an excellent tech- nique for studying solvent-solute interactions. For example Spiesecke31 has measured the spectra of acetonitrile and its isonitrile isomer in various liquid- crystal solvents under identical conditions. In three different solvents the nitrile is always more aligned than the isonitrile. It is tempting and reasonable to think that larger molecules will be more aligned than smaller ones; Spiesecke’s result is then surprising since the isonitrile is longer than the nitrile. Clearly other interactions are involved and liquid-crystal measurements should prove useful in their investigation.When a molecule contains n equivalent protons the isotropic n.m.r. spectrum consists of a single line whereas the anisotropic spectrum contains n equally spaced lines with a binomial distribution of intensities. Thus the spectrum of 31 H. Spiesecke Euratom Italy personal communication. 188 Luckhursf axially symmetric the average splitting is also given by equation (10). By using this together with the same interproton distance of 4.30 A it is calculated that S, is 0-153 where c is perpendicular to the molecular plane. Because of the different experimental conditions used in the determinations the difference in the values of S, for the two solutes is not surprising. C. Non-equivalent Protons.-The first published high resolution n.m.r. spectrum of a compound oriented in a nematic mesophase was that of benzene dissolved in 4,4 ’-di-n-hexyloxyazoxybenzene.ll Since benzene possesses six ‘equivalent’ protons one might at first sight expect the anisotropic spectrum to contain six lines with binomial intensities 1 5 10 10 5 1.Figure 7 shows this supposition to be false for the spectrum contains a multitude of lines. The solution to this apparent dilemma is contained in the word ‘equivalent’. 1,3,5-trichlorobenzene is shown in Figure 8 and from the diagram we can see that proton 1 is separated from protons 2 and 3 by the same distance and because the ordering of all inter- proton axes is the same the 1-2 interaction is identical with the 1-3 and the 2-3. Proton 1 therefore interacts equally with the other protons in the molecule and we describe this by saying the protons are equivalent.In benzene however the 1-2 separation is different from the 1-3 which is also different from the 1-4. Thus proton 1 is not equally coupled to the other protons in benzene. It is easy to extend this argument to all the other protons and because each proton is coupled to at least one other proton to an extent which is not equal to its coupling to any other the protons are non-equivalent. Mushers2 has given a more rigorous 34 J. I. Musher J. Chem. Phys. 1967 46 1537. 189 4 1000 Hz 500 Hz 0 Hz 1 I L 1500 Hz I I I I 0 Hz I I 1000 Hz I H- 500 Hz I500 Hz I TMS 22J H-+ 9 Figure 7 The anisotropic spectrum of benzene in 4,4’-di-n-hexyloxyazoxybenzene at 85 ’. The theoretical reconstruction is given for Jmeta both positive (full lines) and negative (broken lines); clearly J is positive [Reproduced by per- mission from Z .Naturforsch. 1965 20a 5721 Luckhurst Figure 8 The equivalence of the protons in 1,3,5-rrichlorobenzene and benzene definition of equivalence and this has been shown to be identical33 for nuclei with spins I = 8 to that given earlier by Englert and Sa~pe.~O Application of the concept of equivalence is a useful aid to the interpretation of both isotropic and anisotropic n.m.r. spectra and has been employed in the partial analysis of the spectrum of ethyl iodide dissolved in 4,4 '-n-hexyloxyazoxybenzene.34 The anisotropic n.m.r. spectrum of benzene has been completely analysed by use of computer technique^,^^ and as one might expect the spectrum does contain a large amount of information.Indeed it is possible to determine the absolute magnitudes of the ortho and metu spin-spin coupling constants; S a ~ p e ~ ~ has arrived at identical conclusions by using group-theoretical methods. The next section indicates how the sign of the spin-spin coupling constant can be deter- mined from anisotropic n.m.r. spectra. D. The Sign of the Spin-Spin Coupling Constant.-The conceptual basis of the technique is best understood by considering a specific example and the deter- mination of the sign of JCH in acetonitrile is particularly illuminating. The n.m.r. spectrum of acetonitrile aligned in 4,4 '-di-n-hexyloxyazoxybenzene has been measured36 and the high degree of alignment of such a relatively small molecule may indicate a specific solute-solvent interaction. The molecule contains three equivalent protons and the anisotropic spectrum shown in Figure 9 consists of the expected 1 2:l triplet with a spacing of 3495 Hz.We are interested in the sign of JCH and since the natural abundance of a carbon isotope with a nuclear spin is small the methyl carbon was enriched with carbon-13 which has spin 3. The effect of the carbon-13-proton coupling is to split each line in the original three-line spectrum into a doublet. Because the enrichment was not complete the spectrum given in Figure 9 also contains a 1 2:l triplet coming from molecules without carbon-1 3. The magnitude of the proton dipolar splitting is given by 2 h < A v ) = -3yH - s, 4nr3 where r is the interproton distance and c is an axis parallel to the length of the 33 A. Saupe and J. Nehring J.Chem. Phys. 1967 47 5459; J. I. Musher ibid. p. 5460. 34 C. M. Woodman Mol. Phys. 1967 13 365. 35 L. C. Snyder and E. W. Anderson J. Amer. Chem. SOC. 1964 86 5023. 36 G. Englert and A. Saupe Mol. Cryst. 1966 1 503. 191 Liquid Crystals as Solvents in Nuclear Magnetic Resonance I I I I I I I I 1 I +--AvHn = 3495 cps- TMS I I I I I I I I I _ - I 1 I I I I I 1 I I I 1 I I I 1 I I I I I I I I Figure 9 The anisotropic n.m.r. spectra of acetonitrile and [2-1SC]acetonitrile in 4,4'-di-n- hexyloxyazoxybenzene [Reproduced by permission from Mol. Cryst. 1966 1 5031 molecule. It is important to note that since the three protons are equivalent the splitting does not depend on the magnitude of the proton spin-spin c o ~ p l i n g . ~ ~ ~ ~ This is not the case for the carbon-13-proton splitting for the nuclei cannot be equivalent and so the splitting is the sum of the spin-spin and the dipolar coupling < AV)CH = JCH + 'aA S, nR3 where R is the internuclear distance 01 is parallel to the internuclear axis and the dipolar splitting was cakulated from equation (2) because the nuclei are not identical.The element S, is readily related to the ordering matrix in the orthogonal axis system a b c through equation (9) and is (14) 192 Luckhurst where I, is the direction cosine between the carbon-hydrogen bond and the length of the molecule [N.B. the axial symmetry of the ordering matrix was used in deriving equation (14)]. It is straightforward to relate I, to the inter- nuclear distances R and r to give To proceed further we must substitute numbers into equation (15); ( d v ) ~ ~ is 1593 Hz and Jm measured by raising the temperature of the sample above the nematic-isotropic transition point is 136.5 Hz.The value of S, calculated from the proton-proton splitting of 3091 Hz (the value given in Figure 9 is incorrect3’) and an interproton distance of 1-805 8 is 0.1009. Although this calculation does not yield the sign of S,, we can be sure it is positive by analogy with the positive sign found for acetylenic compounds31 by use of a technique to be described in section 6. The value of S, together with R = 1.108 8 allows us to cakulate the last term in equation (15) to be 1459.9 Hz and putting the numbers together we find 1593 Hz must equal 1459-9 & 136.5 Hz an equation which can only be satisfied if J is positive as predicted by theory.38 An identical result has been found by studying both the proton and fluorine spectra of methyl fluoride dissolved in a liquid although the analysis is more complex.In the case of benzene itself the absolute signs of J came from a full analysis of the spectrum. The analysis reveals that the spectrum contains two types of line those whose positions depend on the partially averaged dipolar splitting and those whose positions depend also on the spin-spin coupling constants. By fitting the positions of the first class of line the relative bond lengths and the degree of alignment can be determined. With this knowledge the positions of the second class are fitted simply by varying the J’s thus determining their absolute magnitudes. E. Molecular Geometry.-The dipolar splitting which is observed in aniso- tropic n.m.r.spectra depends on two factors the degree of orientation S and the internuclear distance r. If liquid-crystal solvents are to be used to determine bond lengths we must have some way of measuring S. The most obvious and accurate way of making this measurement is to have an internal molecular standard. In other words we determine the dipolar splitting for two protons with a known internuclear distance and use this to calculate S. Armed with the value for the alignment we can calculate the r from the observed dipolar split- tings for the other magnetic nuclei. The basis for such a determination is well illustrated by our previous analysis of the splittings observed in [2-13C]acetonitrile. Equations (12) and (15) show that both the proton-proton splitting and the proton-carbon-13 splitting 37 A.Saupe University of Freiburg Germany personal communication. 38 D. M. Grant and M. Barfield ‘Advances in Magnetic Resonance’ ed. J. S. Waugh Academic Press New York 1965 vol. 1. 39 R. A. Bernheim and B. J. Lavery J . Amer. Chem. SOC. 1967 89 1279. 193 Liquid Crystals as Solvents in Nuclear Magnetic Resonance (adjusted for JCH) are proportional to Scc. By eliminating Scc we find equation and can determine the ratio r/R. The HCH bond angle is simply 2 sin-l r/R and values have been determined for this from the anisotropic n.m.r. spectra of several methyl compounds;40 the results of these measurements are given in the Table. Table Compound Solvent * from n.m.r. from microwave Ref. Bond angle Bond angle CH3CN I 109" 2' & 2' 109"16' a CH3NC.- 109'45' 7 109'7' b CH30H I1 110" 3' -l- 8' 109" 2' & 45' C CH31 1 111"42' f 2' 11 1'25' d * I is 4,4'-di-n-hexyloxyazoxybenzene and I1 is 4-n-octyloxybenzoic acid. t Ref. 31. I1 108'56' -f 2' a L. F. Thomas E. I. S. Sherrad and J. Sheridan Trans. Faraday SOC. 1955 51 619; b C. C. Costain J. Chem. Phys. 1958 29 864; C P. Venkateswarlu and W. Gordy J. Chem. Phys. 1955,23 1200; d S. L. Miller L. C. Aamodt G. Dousmanis C. H. Townes and J. Kraitch- man J. Chem. Phys. 1952 20 1112. The Table shows that the n.m.r. technique is capable of greater accuracy than microwave methods. In comparing the two sets of data one must realise that while the n.m.r. results were obtained in the liquid phase the microwave results are for molecules in the gas phase. Indeed one of the strengths of the liquid- crystal technique is its ability to determine molecular geometries in the liquid phase.The small differences in the bond angles may be attributed to the differ- ence in the nature of the phase. An alternative explanation is hinted at by the measurement of the anisotropic spectra of spherically symmetric molecules. Snyder and Meiboom*l have measured the n.m.r. spectra of tetramethylsilane and neopentane in 4,4 '-di-n-hexyloxyazoxybenzene. Since these molecules are not expected to be aligned by a nematic liquid crystal it is surprising to find that the spectra consist of a 1:2:1 triplet and not a single line. The splitting un- doubtedly comes from the dipolar coupling between the protons in a particular methyl group. It is thought that the methyl group is distorted by the neighbouring liquid-crystal molecules and if the splitting is to be observed the distortion must depend on the orientation of the methyl group with respect to the field.In the preceding sections we have been mainly concerned with first-order 40 A. Saupe G. Englert and A. Povh Adv. Chem. Ser. 1967 63 51. 41 L. C. Snyder and S . Meiboom J. Chem. Phys. 1966 44 4057. 194 Luckhurst I n.m.r. spectra; that is those in which the spacings between the lines are multiples of the coupling constants in the spin Hamiltonian. Let us now consider 33- dichlorobenzoic acid whose anisotropic n.m.r. spectrum has been measured in 6-n-hexyloxy-2-naphthoic acid.42 The solute is similar to 1,3,5-trichlorobenzene which was considered earlier but now the ordering matrix is not axially sym- metric because the carboxyl groups of both solute and solvent form hydrogen bonds,43 so that the 1-4 axis is more aligned than the 2-6.This added degree of alignment spoils the first-order analysis and instead of a 1 2 :1 triplet the spectrum contains eight lines. The analysis of the line separations is straightforward and it is fairly easy to obtain analytic expressions in terms of the dipolar coupling constants which are themselves given by equations similar to (12). Y Figure 10 A comparison of the experimental and theoretical n.m.r. spectrum of cyclopropane in a nematic mesophase. Since the spectrum is symmetrical only the low-field harf is given [Reproduced by permission from J. Chem. Phys. 1967 47 14801 When the molecule contains many protons interacting differently it is not possible to obtain analytic expressions for the line separations and computer techniques must be used to obtain the coupling constants.A particular tour de force of the liquid-crystal technique has been the determination of the geometry of cyclopropaneZ1 by measuring the n.m.r. spectra of molecules not containing carbon-13 and of those which do. The many magnetic interactions within cyclo- propane have been successfully analysed as one can see from Figure 10 where the theoretical spectrum is compared with the experimental one. Although the geometry determined by n.m.r. is in satisfactory agreement with that obtained from gas-phase electron diffraction the analysis of the dipolar coupling constants to obtain the bond lengths emphasised an interesting point.44 The dipolar couplings obtained from the spectrum are of course propor- tional to l/r3.Such a simple relationship however neglects the vibrations of the molecule and strictly speaking the dipolar coupling is proportional to an average over all vibrations i.e. to (l/r3). It is convenient to write r which is 42 G. Englert and A. Saupe 2. Nuturforsch. 1965 20a 1401. 43 J. C. Rowell W. D. Phillips L. R. Melby and M. Panar J. Chem. Phys. 1965 43 3442. 44 J. A. Ibers and D. P. Stevenson J. Chem. Phys. 1958 28 929. 195 Liquid Crystals as Solvents in Nuclear Magnetic Resonance the separation at any given time as the sum of the equilibrium internuclear distance re and an extension 6r. The function we require is ( l / ( r e + 6 ~ ) ~ ) and provided (&)/re is small we can expand this quantity as a power series in (&)/re to give Provided re is defined as the average value of r then (6r) vanishes and the dipolar coupling is proportional to (l/r?)[l + 6(a/re2)] where a is the root- mean-square vibrational displacement.Both X-ray and electron diffraction techniques measure (re + 6 r ) which is just the equilibrium separation and the values for n.m.r. will be related to these by The calculation of a for a given vibration is an extremely difficult task and vibrational effects may be a potential source of error in the use of n.m.r. to measure bond lengths. Fortunately in the case of cyclopropane the dis- crepancies are rather small in agreement with estimates4s of a/re for a methylene group of 0.1. Strictly the dipolar coupling is proportional to the average value of l / r 3 and S but when the molecule is rigid the effect of vibrations on S is negligible.If however the molecule is non-rigid such as n-pentane we should also have to allow for the dependence of S on the internal motion of the molecule. The examples of structure determination which have been discussed were useful in (a) establishing the validity of the liquid crystal technique and (6) show- ing the small effect on molecular dimensions on passing from the gas to the liquid phase. As one might hope the technique is being applied to unknown structures and a particularly fruitful field would seem to be organometalic complexes. For example one would like to know whether the protons in cyclobutadiene iron tricarbonyl shown in Figure 11 are arranged in a square or rectangular con- &!%-< H H v Figure 11 The structure of cyclobutadiene iron tricarbonyl ‘li 0.Bastiansen F. N. Fritsch and K. Hedberg Acta Cryst. 1964 17 538. 196 Luckhurst figuration. This is an ideal problem for liquid-crystal n.m.r. and the spectrum of the complex has been recorded in 4,4’-di-n-hexyloxyazoxybenzene.4e The spectrum is not first order and contains eight lines as one would expect for a square configuration although the intensities of the lines are not in com- plete agreement with theory. If the departure of the intensities from their theore- tical values is ascribed to a small distortion of the ring the ratio of unequal sides is found to be 0.9977 f 04045 that is the departure from a square configura- tion is very small. 6 The Chemical Shift Tensor The previous sections have shown how anisotropic n.m.r.spectra are dominated by the nuclear dipolar coupling. However the chemical shift can also be aniso- tropic and this anisotropy will be observed in the spectra as a shift in the posi- tions of the lines. Consider a single nucleus whose chemical shift tensor is crab where a b and c form a molecule fixed-axis system. The chemical shift (J measured from the isotropic spectrum is one third of the trace of crab but in the nematic mesophase the chemical shift (a) is given by This can be rewritten by use of the ordering matrix dehed in equation (6) as C i,j = a Under certain conditions equation (20) can be used to determine the signs of the elements of the ordering matrix. Consider the acetylenic proton in methyl- acetylene. The symmetry of the molecule demands that both s, and uab are axially symmetric about the long axis a.Equation (20) can therefore be written as where (do) is the difference in the chemical shift for the anisotropic and isotropic spectra. It is important to realise that the sign of (do) follows im- mediately from experiment and if the sign of (oaa - a b b ) is known the sign of S, can be determined. The difference (uaa - ebb) can be calculated with some certainty and is positive for acetylenic protons.47 Since (do) is positive the ordering matrix element Saa must also be positive,36 implying that methyl- acetylene is aligned with its long axis parallel to the magnetic field. The derivation of equation (19) contains a hidden assumption for it treats the 46 C. S. Yannoni G. P. Caesar and €3. P. Dailey J. Arner. Chem. SOC. 1967 89 2833.47 J. A. Pople Proc. Roy. SOC. 1957 A 239 541 197 Liquid Crystals as Solvents in Nuclear Magnetic Resonance molecule as if it were rotating in an isotropic potential. Clearly such an approxi- mation is not valid for the anisotropic fields created by the liquid-crystal solvent are responsible for the alignment. The anisotropy is expected to produce a change d '0 in the chemical shift even if the tensor were isotropic.48 There are two contributions to d Oa; the first results from the change in the volume suscep- tibility of the sample when the solvent passes from the isotropic to the nematic mesophase. (The volume susceptibility determines the value of the magnetic field inside the sample.) The other contribution depends on the anisotropy of the solvent effects often encountered in n.m.r. Since the total magnitude of the two effects is difficult to calculate the use of liquid crystals may not yield reliable values for the anisotropy in the chemical shift tensor. Although such comments cast doubt on the value of (o, - crbb) determined for the acetylenic proton in methylacetylene they do not affect the conclusion that S, is positive. I am grateful to Drs. Englert Meiboom Saupe and Snyder for permission to publish the spectra reproduced in this Review as well as to Dr. Saiko for permission to publish parts of the Review which had already appeared in the Oesterreichische Chemiker-Zeitung 1967 68 1 13. It is a pleasure to acknowledge useful discussions with Drs. Meiboom and Saupe. 1 48 A. D. Buckingham and E. E. Burnell J. Amer. Chem. SOC. 1967 89 3341. 198

ISSN:0009-2681    

DOI:10.1039/QR9682200179

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Theory of Thermal Electron-transfer Reactions in Solution By I. Ruff INSTITUTE OF INORGANIC AND ANALYTICAL CHEMISTEIY L . EOTVO UNIVERSITY BUDAPEST 1 Introduction Inorganic reactions in solution can be classified in general as belonging either to acid-base or to electron-transfer reactions. All reactions of mono- and poly- nuclear complex formation as well as those of precipitate formation an extreme case of the latter are included in the group of acid-base reactions if the defini- tion of acid and base is broad enough as e.g. Lewis’s one. Electron-transfer reactions take place between two or more chemical objects by electron donation and acceptance and these chemical objects are initially and finally moving independently of each other. Neutral molecules ions and compact materials (electrodes) may serve as chemical objects.The initial and final independence of the reactants excludes the electron-transfer processes within the same molecule such as the case of charge-transfer complexes. In most cases the electron-transfer reaction is accompanied by acid-base processes since the change in oxidation number which is the result of the reaction influences the acid-base properties of the reactant particles [see e.g. the iron(ii)-permanganate reaction]. It is generally assumed that in the elec- tronic excitation changes in the co-ordination sphere such as protonation or deprotonation i.e. acid-base phenomena precede the electronic jump. Electron transfer is however the principal process determining the total free-energy change predominantly through electronic energies.Sometimes the electronic jump is believed to take place simultaneously with the transfer of a ‘heavy’ particle such as proton (H atom transfer in organic oxidation) oxygen or any other ions atoms or groups of atoms. As these reactions can be regarded to be simultaneous electron-transfer and acid-base reactions it seems that there is no basic reason for distinguishing between electron and atom transfers in oxidation-reduction reactions as is done frequently. The terms ‘oxidation-reduction’ and ‘electron-transfer reactions’ are usually applied as identical ones. For simplicity the expression ‘oxidation-reduction’ will be used in this Review for conventional electron transfers with finite change in reaction free energy while ‘electron exchange’ will refer to those of zero net free energy change (isotopic exchanges between ions of the same element in different oxidation states).‘Electron transfer’ will cover both. Electron-transfer reactions play a very important r6le in almost every field of 199 The Theory of Thermul Electron-transfer Reactions in Solution chemistry and in certain important phenomena of biology such as photo- synthesis reactions in the respiratory chain of cells etc. This explains the enor- mous volume of experiments and theoretical work that have been published. In this Review only the theoretical treatments for homogeneous electron-transfer reactions will be considered with reference to heterogeneous electrode reactions only when the theory predicts relations for both. Experimental results will be quoted only in comparing them with theoretical predictions.‘Pure’ electro- chemical theories as well as those dealing with radiation- induced electron-transfer reactions will be omitted. A number of reviews and books have dealt with this field.1-9 Recent develop- ments however call for an up-to-date comparison of the applicability of the different theories. Hence instead of the concrete treatment the physical meaning of the models and the conclusions will be discussed first; theories starting from identical models but differing in mathematical apparatus will not be treated individually. Any theoretical approach for chemical reactions starts from assumptions on the microscopic processes i.e. on the model of the mechanism. Its correctness can then be confirmed by deducing relations for kinetic parameters that can be experimentally observed.Unfortunately the number of measurable quantities of these reactions is small; this is why the theories outlined by different authors are based on such different contradicting principles. At the present stage there is not any direct proof of the correctness of one of the activated complexes supposed so the applicability of a model can be judged from two viewpoints (i) which model is more plausible and (ii) which can predict results in satisfactory agreement with experiments over a broader range of parameters and reactions. 2 The Franck-Condon Principle and Its Application for Radiationless Transitions The Franck-Condon principle>0 so important in interpreting molecular spectra yields information on the probability of transitions among electronic levels due to different vibrational states.This schematic conception is that the transition of an electron from one level to the other cannot be accompanied by simul- taneous nuclear displacement since the mass of the electron is much smaller than that of the nuclei. If energy relations are schematically indicated in a diagram of energy as a function of internuclear distance as in Figure 1 electron transfer could occur only between terms having vibrational eigenfunction with a R. R. Edwards Ann. Rev. Nuclear Sci. 1952’1 301. C. B. Amphlett Quart. Rev. 1954 8 219. F. Basolo and R. G. Pearson ‘Oxidation-Reduction’ in ‘Mechanism of Inorganic Reac- H. Taube ‘Mechanism of Redox Reactions’ Adv. Znorg. Chem. Radiochem. 1959. J. Halpern Quart. Rev. 1961 15,228. H. Taube Proc. R. A. Welch Foundation ConJ Chem.Res. 1962 7. a R. A. Marcus Ann. Rev. Phys. Chem. 1964 15 155. E. S. Amis ‘Solvent Effects on Reaction Rates and Mechanisms’ Academic Presr New York and London 1966. l o E. U. Condon Phys. Rev. 1940,36 1121; E. U. Condon and G. H. Shorttoy ibid. 1931 37 1025. a00 * B. J. Zwolinsky R. J. Marcus and H. Eyring Chem. Rev. 1955 55 157. tions’ J. Wiley and Sons New York 1958. r t ---tr Figure 1 Schematic representation of the electronic transitions regulated by the Franck-Condon principle maximum at the same Y. (The density of the electron clouds is symbolised by hatching. For lower vibrational levels the maximum is somewhere in the middle of the potential well; for higher levels it is around the turning points that is at the wall of the potential well.) The ground state corresponding to the new bond relation created by electron transition can be occupied by the system only after the electron transition has taken place after a relaxation period determined by the nuclear masses.In Figure 1 electron transitions accompanied by bond relaxation are schema- tically represented. Electron transition can thus occur only vertically (r - constant) i.e. only the 1,l - 2,2 and 1,2 --f 2,l transitions are permitted. Libby’s fundamental idea was that these transition probabilities are valid for radiationless electron transitions and so for electron transfer reactions as well.l1J2 In this case the 1,l and 1,2 states belong to the reducing agent in the reaction the 2,l and 2,2 states belong to the oxidising particle. When the discussion is limited to electron-exchange reactions in which reactants and products are chemically identical the standard enthalpy change of the reaction need not be considered since the 1,l and 2,l levels are energetically identical.For instance let the 1,l level be the ground state of the aquated Fe2+ and consider reaction (1). Fe2+ + *Few + Fe* + *Fe2+ (1) (2) (1) (2) (1) It is evident that in the final state of the reaction the electronic energy is identical with that of the initial state. With the same example electron exchange can l1 W. F. Libby J . Phys. Chem. 1952,56,863. l* H. C. Hornig G. L. Zimmermann and W. F. Libby J. Amer. Chem. SOC. 1950,72,3808. 201 The Theory of Thermal Electron-transfer Reactions in Solution occur as follows (a) If the electron starts from the ground state of Fe2+ it can be ‘received‘ by Fe3+ only if the water molecules in the first co-ordination sphere are in an excited vibrational state corresponding to the 2,2 level.A part of the ativation energy thus has to be supported to excite the vibrational state of Fe3+ so that on receiving the electron the water molecules should be in an excited state corresponding to the Fe2+ ion. The reaction is then terminated by the 2,2 - 2,l transition i.e. by accepting the electron the formerly tervalent iron ion has become an excited bivalent iron ion then an Fe2+ ion in the ground state. (b) The reaction path via 1,l - 1,2 -+ 2,l states is completely analogous to what has been shown above; moreover it can easily be seen that its activation requirements are the same since in an electron exchange reaction the inter- mediate and ground levels are identical.Both in (a) and (b) a symmetrical arrangement of the neighbourhood with respect to the plane perpendicular to the Fe-Fe axis is produced so that the electron cannot ‘distinguish between its original and new owner’. There are intermediate cases of course when this symmetry requirement is fulfilled by simultaneous changes in both co-ordination spheres. These were essentially Libby’s assumptions. His most important conclusion was that for ion pairs for which there is no or only very small difference in the first co-ordination sphere of the oxidised and reduced form like Fe(CN)$- Fe(CN),& and MnOZ-MnO,- the Franck-Condon restriction does not lead to any appreciable energy barrier and these reactions are expected to proceed quickly. At the time of Libby’s report these electron-exchange reactions were known to be unmeasurably fast.13-15 Yet measurements on the Fe11-Fe=1,16 Ce*-C&v,l7 and EuU-EuItl systemsls yielded measurable and slower rates.This was in accordance with Libby’s ideas for the bond strength of water molecules is markedly different in these cases for ions of different charge.Numerical data cannot be derived by this theory since quantum mechanical knowledge of the system is needed and this is almost inaccessible. The reaction rate depends also on the distance between the two reacting particles. Therefore in addition to the activation energy due to the Franck- Condon restriction energy is required to bring the particles from an infinite distance to a distance necessary for reaction. This can be roughly approximated by the electrostatic repulsion if the solvent is considered to be a continuous dielectric medium.The value of the enthalpy of activation AH; is considerablyreduced by a negative ion placed between the ions. For this reason Libby attributed catalytic effects to small negative ions. This accelerating effect is the greatest when the negative ion is built in in the first co-ordination sphere of the reactant ions l3 N. A. Bonner and H. A. Potratz J. Amer. Chem. SOC. 1951 37 1845. l4 R. C. Thompson J. Amer. Chem. SOC. 1948 70 1045. 16 J. W. Cobble and A. W. Adamson J. Amer. Chem. SOC. 1950,72 2276. l6 J. Silverman and R. W. Dodson Brookhaven Quart. Report BNL-93 p. 65 1950. l7 J. W. Gryder and R. W. Dodson J. Amer. Chem. SOC. 1949,71 1894. D. J. Meier and C. S. Garner J. Amer.Chem. SOC. 1951 73 1894. 202 forming a bridge for electron transfer as for example in the activated complex (I) where X- may be F- C1- OH- etc. (H20)5Mnn+-X- Mn nS1 + (H,O) (1) Though Libby’s hypothesis is qualitative it has been the starting point for most of the theories developed so far. They can be divided into two groups (i) the so-called ‘outer sphere’ mechanism which deals with systems involving reactant ions of ‘hard‘ co-ordination spheres and (ii) the ‘inner sphere’ mech- anism. In this case the reactant ions are suitable for formation of the binuclear activated complex (I). The latter group of reactions seems to be the more difficult for theoretical treatment so we shall deal with it after the outer sphere transfers. 3 The Outer Sphere Mechanism A. Theories in which Direct Transfer is Assumed.-(i) Application of the tunnel efect.The first theory yielding quantitative results comparable with the measure- ments was published by Eyring et al.19 Libby’s assumption for the outer sphere mechanism was entirely retained but it was completed by an important aspect namely the transfer probability of the electron which can easily be calculated on the basis of quantum mechanical tunnelling. According to this model the main steps of the reaction are (1) The collision of the reactant species which needs an energy equal to the Coulomb repulsion. (2) The Franck-Condon restriction requires the rearrangement of the environment owing to the new charge distribution resulting from the transfer the water molecules being almost ‘frozen’ in the electrostatic field of the ions are to ‘melt’ to make the rearrangement possible.(3) The electron must appear where the oxidising ion is located and this does not occur with unit probability. The Franck-Condon principle as applied by Libby does not take into account that the electron must be transferred into a field of a nucleus chemically identical with the original but different in its space co-ordinates. The electron thus is not only subjected to energy changes but it must cover a definite route in space which is not without obstacles. The potential barrier to be passed by the electron is in this approach the sum of the Coulomb potential of the reactants (see Figure 2). Since to a first approximation this potential barrier can be replaced by a triangular potential the transfer probability X e is given by equation (2) where rn is the mass of the electron V is the height of the potential barrier and W is the kinetic energy of the electron.Vo can be calculated with respect to the ionic charges and the interionic distance rnlm while W can be obtained as the half of the potential energy of the electron according to the virial theorem. The total activation free energy is given by equation (3). The first and third R. J. Marcus B. J. Zwolinsky and H. Eyring J. Phys. Chem. 1954 58 432. 203 The Theory of Thermal Electron-transfer Reactions in Solution AFS = AF$rep. + AF$rearr. + AF$tun. (3) term of the right-hand side of equation (3) is determined by equation (2) and by a simple electrostatic interaction term respectively. Figure 2 Electrostatic potential barrier between two ions of positive charge (Reproduced from J.Phys. Chem. 1954 58,432 by permission) The a priori calculation of AFfrearr. is very difficult but this portion of the activation free energy was supposed to be approximately constant. Thus there should be parallelism between the total activation free energy calculated without dF$rearr. and the measured ones. The reaction of Fen to F f l was chosen for fitting and the 8.1 kcal./mole difference obtained was considered for each re- action as AFSrearr. Since the repulsion term is enhanced by decreasing rntm while the tunnelling term is reduced there is an optimal interionic distance r*n,m that can be calcu- lated by minimising the total free energy of activation. The minimum appearing at r *n,m represents the value which is due to the fastest path i.e.the observed one. The calculated and experimental values are compared in Table 1. Agreement between the values is not always satisfactory. The r*,, values show that apart from the reacting ions and the ligands in their first co-ordination sphere no other ,,lcle is implied in the activated complex as they generally approach one another up to the contact of these spheres. Thus the error in assumptions arising from treating the solvent as a dielectric medium is immediately revealed. Apart from its semi-empirical nature this theory has other essential deficiencies. On one hand it cannot interpret the electron exchange between negatively charged ions as e.g. Mn0,2-Mn04-; in this case there would be three potential wells for the electron; two in the vicinity of Mnm and MnvI and one between the two negative ions which is obviously meaningless.On the other hand it is 204 Table 1 Reaction r * n m (4 dF$ (kcal./mole) c o (en) 32+-Co(en) 3js ~ i + - ~ i ( OH)^+ V(OH)2+-V02+ Fe2+-Fe3+ Fe2+-Fe(OH)’+ Fe2+-FeC12+ Fe2+-FeC12+ Ce3-t-Ce4+ 5.9 3-3 4.4 5.0 4.9 4.9 3.4 9.3 Calc. 16-8 13-3 16.0 16.7‘ 15.1 15.1 13.1 19.1 Obs. 23-5 23.gC 1 7 ~ 2 ~ 1 6 ~ 7 ~ 12.2‘ 15.3’ 15.1 18.W a Fitted value. b W. B. Lewis C. D. Coryell and J. W. Irvine J. Chem. SOC. Suppl. No. 2 1949 s386. C G. Harbottle and R. W. Dodson J . Amer. Chem. SOC. 1951,13 2442. d S. C. Furman and C. S. Garner J. Amer. Chem. SOC. 1952,14,2333. See Ref. 32. f J. W. Gryder and R. W. Dodson J. Amer. Chem. SOC. 1951,73,2890. not clear why it is the kinetic energy of the electron of the ground state that is considered when the transfer state is vibrationally excited because of the free energy of rearrangement which it possesses.Marcus criticised* the electron-tunnelling hypothesis because of the omission of the probability factor p that expresses the number of times the electron ‘strikes’ the barrier wall (about 10lasec.-l). The right probability should be PXe instead of Xe. However in view of the method of deduction of equation (2),2O and the invariability of Xe with respect to the amplitude of the ‘striking’ electronic wave Marcus’s criticism can be rejected. For if an electron is in its initial state it will be on the other side of the barrier with probability X e independently of whether it strikes the wall p times forwards and PXe times backwards since the chance of finding an electron in its initial state is always Xe times greater than that of finding it in the final state.It must be mentioned that Randles21 published independently a theoretical treatment for electrode reactions on an almost identical basis to Eyring’s one. Weiss’s however which appeared in the same year does not take the rearrangement before the electron transfer into account. By refinement of the model Laidler23 eliminated some of the inadequacies but he could not solve the main contradictions. Two essential changes have been affected in the premises (i) In place of dielectric constant the refractive index which is more relevant to short distances has been considered. (ii) In place of the Franck-Condon restrictions the activation free energy (dFSdiff.) required for diffusion has been applied in place of APZrearr..The electrostatic energy barrier the electron has to overcome is lowered owing to (i) so the transition probability increases. By this the ‘defect’ observed by Eyring and his co-workers to be 8.1 kcal./mole is somewhat decreased. 2o Z. V. Shpolskii ‘Atomnaia fizika’ Moscow 1951 p. 431. a1 J. E. B. Randles Trans. Faraday SOC. 1952 48 828. 22 J. Weiss Proc. Roy. SOC. 1954 A 222 128. 23 K. J. Laidler. Canad. J. Chem. 1959 37 138. 205 The Theory of Thermal Electron-transfer Reactions in Solution Laidler carried out calculations for the experimentally best known F&F& reaction only. The results are exactly the same as the experimental values. This agreement is unique but the assumptions cannot be accepted. The discarding of the Franck-Condon restriction means in fact that the electron is in a steady ‘dispersed’ state with a decreasing amplitude defined by the tunnel effect in continuous ‘readiness’ for exchange around the reducing ion and the oxidising ion is permanently able to accept this electron.All the other portions of activa- tion energy serve only to bring the reactants closer to one another. From this it follows (as an extreme case) that a crystal composed of such aquated or complex ions should show metallic conductance. This however could not be expected since for example Prussian blue is a semiconductor i.e. the charge-transfer process is appreciably hindered. Later24 the model was corrected by the rearrangement of the environment and the diffusion term was omitted. In this form the model is similar to those in the following section with the only difference that tunneling has been retained.(ii) Calculation of energy relations of the activated complex by polarisation. The theory most frequently quoted is that of Marcus.25 His model is fundamentally an electrostatic concept too. It is to be considered as a great advance however that all parts of the activation free energy can be calculated a priori. Moreover this is the first theory suitable for deducing quantitative relations for redox reactions and for discussion of the rate of electro-chemical and chemical re- actions by identical principles. The model is as follows ions are considered to be metal spheres of radius Y,. Around them there is a solvent (or ligand) layer of thickness a - r assumed to be dielectrically saturated. The solvent outside this sphere of radius a is taken as a continuous medium with dielectric constant D,.In the activated complex the two spheres of ao,, and ao,m radii are in contact. Unlike the previous theories the tunnelling probability is estimated to be around unity owing to the high value of p so there is no need of free energy for the spatial transfer of the electron. Free-energy requirements show up in Coulomb repulsion and in the so-called ‘non-equilibrium’ polarisation. This renders it possible to calculate the energy required to fulfil the Franck-Condon principle in a classical way avoiding the difficult problems of quantum mechanical treat- ment. For in chemical reactions taking place by transition of an atom or a group of atoms the electronic orbitals in the activated complex overlap to a great extent while for electron-transfer reactions smaller interactions and orbital overlaps are expected.The interaction of the reacting ion pair is thus probably smaller than or about the same as that of the ions and the environment or of the activated complex and the environment. Thus while in the calculation of other reactions the interaction of partners and the solvent can be taken to be in equilibrium for electron transfers this assumption is not allowed. 24 K. J. Laidler and E. Sacher ‘Modern Aspects of Electrochemistry’ Butterworths London 1964 vol. 3 pp. 1-42. R. A. Marcus J. Chem. Phys. 1956 24 966; ibid. 1957 26 867; Canad. J. Chem. 1959 37 155; Discuss. Faraday Sac. 1960 29 21 ; J. Phys. Chem. 1963 67 853 ; J. Chem. Phys. 1965 43 679. 206 Ruf On the other hand the electron transfer is extremely fast.The new charge distribution affected in the activated complex can be followed only by the electronic polarisation among the different parts (atomic orientation and electronic) of the dielectric polarisation of the environment since the other two would involve nuclear displacement. That is the non-equilibrium state of the neighbouring solvent at the moment of the electronic jump arises from the inertness of the 'heavy' nuclei. Since detailed calculations are tedious only the final results can be reported here. Marcus estimated the slowest elementary step to be the collision in a suit- able polarised arrangement of the water molecules. Therefore the rate constant is given by equation (4) k=Zexp (-=) AF* free energy is AF* = m2h + z,Z Dsrn,m where where 2 is the number of collisions while the activation (4) 2h and In these equations D is the statical dielectric constant Do is the square of refraction index; zn,t and zm,t are the charges of the products and dz is the charge corresponding to the transferring electrons while d F" is the standard free energy of the reaction.(The thermodynamic quantities with asterisks are not identical with those of sign $ because of the appreciable difference between kT/h and Z.) Since the reaction is the fastest at the smallest rn,m we have equa- tion (8). Results so calculated are not in satisfactory agreement with the experi- mental values (Table 2). Calculating the size of the reacting spherical ions Marcus considered the first co-ordination spheres of the ions to be unaltered by the electron transfer only the polarisation free-energy change outside the first layer of the solvent being applied.This seems to cause merely the overestimation of rates (Table 2). By further refinement of the model Hush between 1957 and 1959,2s as well as a6 N. S . Hush 2. Electrochem. 1957 61 734; J. Chem. Phys. 1958 28 962; Trans. Faraday SOC. 1961 57 557. 207 The Theory of Thermal Electron-transfer Table 2 Reaction Mn0,2-Mn0,- Fe( CN),4-Fe( CN):- Mo(CN)~-MO(CN)Z- Fe2+-Fe3+ co2+-co3-t Reactions in Solution AFS (kcal./mole) (Calc.) (Obs.) 9-2 12.8" 10.1 12*7b 9.5 < 12.6' 9.8 16.3 9.9 16.4' a J. C. Sheppard and A. C. Wahl J . Amer. Chem. SOC. 1953 75 5133. A. C. Wahl and C. F. Deck J. Amer. Chem. SOC. 1954 76 4054. C R. L. Wolfgang J. Amer. Chem. SOC. 1952 74 6144.d See Ref. 32. CN. A. Bonner and J. P. Hunt J. Amer. Chem. SOC. 1952 74 1866. Marcus in 1960 introduced an additional term into the expression of free energy of activation which refers to changes in bond strength between the central ion and the water molecules just next to it. For transition-metal ions as Hush assumed this change in energy consists of two portions (i) the 'pure' electro- static interaction term due to the attraction between the central charge and the dipoles and (ii) the crystal-field stabilisation energy change corresponding to the ligand-electron and electron-electron interactions. The first portion has been estimated to be small ca. 2 kcal./mole while the second one is significant about 5 kcal./mole. For non-transition metals only the first one operates. The values calculated in this way fit far better the experimental ones than those of Marcus as shown in Table 3.Table 3 Reaction AH:. (kcal./mole) AS (e.u.) (Calc.) (Obs.) (Calc.) (Obs.) Fe3+-Fe2+ 10.8 9.9" - 32 - 25a v3+-v2+ 7.5 13*2b -31 -25b Ce4+-ce3+ 6-8 7.7 - 45 -40' NpO Z+-Np02+ 9.6 8.3' - 19 - 24d a See Ref. 35. b K. V. Krishnamurty and A. C. Wahl J. Amer. Chem. SOC. 1958 80 5921. C See Ref. f i n Table 1. d D. Cohen J. C. Sullivan and J. C. Hindman J. Amer. Chem. SUC. 1956 78 1540. Returning to Marcus's theory for the rate constant of redox reactions kl,t we can write equation (9) where kl, and kz,2 are the rate constants of the kl, = J(k1,l k2,2 Kof 1 (9) corresponding electron exchange reaction of the reactants KO is the equilibrium constant and (ln In f = kl 1 k2,2 z2 41n - 208 If In f is negligible and for similar systems for which kl,l and k,, do not alter much on going from one system to another one can expect the relation (1 1) to be valid where a is a constant.In k, = a + &ln KO (1 1) The validity of equation (11) was observed by Sutin and his co-workers investigating the oxidation of ferrous phenanthroline derivatives by ceriurn(~v)~’ and other reactions2* (Figure 3). Some other observation^^^ on exchange between 1 I I 1 I I I I I I I a -2 -4 -6 d -0 92 -# d G . h f d-‘ Figure 3 Experimental results confirming the relation in equation (1 1). The circles correspond to reactions between cerium(rv) and various substituted iron(r1) phenantrolines and between iron@) and iron(m) phenantroline derivatives (Reproduced from Inorg. Chem. 1963 2 9 17 by permission) ferrous and ferric complexes show similar relations but with respect to the absolute value of In KO the reason of which is not explained.The essential identity of the processes of electrochemical and chemical electron-transfer reactions prompts one to find a relationship between the rates of the two kinds of reaction. Marcus’s theory was the first suitable one for this. If equations (9-47) are rearranged corresponding to electrochemical reac- tions we get equation (12) where dF*c is the activation free-energy change AF*c 6 2AF*el (12) of the electronexchange reaction while dF*el is that of the reaction corre- sponding to the exchange current. The factor 2 in fact means that two particles (and their co-ordination sphere) participate in the chemical reaction while in 27 G.Dulz and N. Sutin Inorg. Chem. 1963 2 917. 28 R. J. Campion N. Purdie and N. Sutin Inorg. Chem. 1964 3 1091. 29 K. Backmann and K. H. Lieser Ber. Bunsenges. Phys. Chem. 1963,67,802; 1963,67,810; 2. phys. Chem. 1963,36,236; Symp. on Exchange Reactions Upton 1965 Paper SM-64/10. 209 The Theory of Thermal Electron-transfer Reactions in Solution electrode reactions only one do so. If polarisation processes in the adsorption layer of the electrode are not negligible the equality sign must be replaced by < . From equation (12) we have equation (13) where Zc is the collision number of the chemical reaction (ca. 10l1 mole-l sec.-l) while Zel is that of the electro- chemical one (ca. lo4 cm. sec.-l). Thus k ~ calculated on the basis of theexchange- current values can be compared with the rate constant of the electron-exchange reaction (Table 4).Values show good agreement thereby proving the identity of the fundamental processes. Table 4 System Fe( CN):-Fe( CN):- Mn0,2-Mn04- Fe2+-Fe3+ v2+.-v3+ Eu2+-Eu3-t TL+-TP+ CO(NH~)~+-CO(NH&,~+ &Wl0l1) 1 x 10-3b 4 x 10-7e 2 x l0-4C 9 x 10-Gd 6 x 10-sf 3 x 10-8g < 5 x 10-llh ~ ~ ~ / 1 0 4 a 1 x 10-5 7 x 10-7 7 x 10-7 4 x 10-7 3 x 2 x 10-8 5 x 10-12 a J. E. B. Randles and K. W. Somerton Trans. Faraday SOC. 1952 48 937. b A. C. Wahl 2. Electrochem. 1960 64 90. K. V. Krishnamurty and A. C. Wahl J. Amer. Chem. SOC. 1958 80 5921. f D. J. Meier and C. S. Garner J. Phys. Chem. 1952 56 833. g E. Roig and R. W. Dodson J. Phys. Chem. 1961 65 2175. h See R. A. Marcus’s explanation in J. Phys. Chem. 1963 67 853.See Ref. a in Table 3. See Ref. 32. An essentially similar model has been used but computed elegantly on quantum mechanical basis by Levich and his c o - w ~ r k e r s . ~ ~ ~ ~ ~ Their theory was outlined first for electrode reactions. The approximations differing in their mathematical nature from those of Marcus led to somewhat different results although in the classical limit the formulae fit Marcus’s ones. They used also a continuous dielectric medium but because of the complexity of the problem they had to neglect the effect of the inner co-ordination sphere. As has been pointed out all the theories based on the polarisation approach attribute a marked influence to the environment outside the first co-ordination sphere. This leads to a prediction that the rate of electron-transfer reactions should depend on the dielectric constant of the medium [see e.g.equations (5)-(7)]. Some investigations in mixed solvents however do not confirm this relation :32,33 the rate is essentially independent of the macroscopic dielectric 30 V. G. Levich and R. R. Dogonadze Doklady Akad. Nauk S.S.S.R. 1959 124 123; 1960 133 158. 31 R. R. Dogonadze Doklady Akad. Nauk S.S.S.R. 1960 133 1368; R. R. Dogonadze and Y. A. Chizmadzhev ibid. 1962 144 1077. 32 D. Cohen J. C. Sullivan E. S. Amis and J. C. Hindman J. Amer. Chem. SOC. 1956 78 1543. 33R. A. Horne Symp. on Exchange Reactions Upton 1965 Paper SM-64/18. 210 properties of the solvent in a broad range of D. The solvent component of lower dielectric constant does not affect the reaction as far as the selective solvation of the component of higher D is not disturbed.On the other hand these theories predict usually negative entropy of activa- tion. In several cases however a positive entropy change has been measured that cannot be interpreted on the basis of polarisation theory. B. Theories in which Indirect Electron Transfer is Assumed.-(i) Hydrogen- atom transfer. In the investigation of electron-transfer reactions the pH-depend- ence of the rate has been observed for many reactions. In the rate law however no integer was directly measured as the order of hydrogen ion concentration. By expansion used in such cases a variety of reaction orders have been obtained for the different reactions. For example the Fe2+ - Fe3+ exchange could be b k = a + - [H+l describedM by a rate constant (14) where a and b are constants.To interpret the pH-dependence it was assumed that the reaction path chang- ing with acidity corresponds to electron transfer between metal ions hydrolysed to varying extents. Thus the F$-FelI1 electron exchange can take place via two paths either by the reaction of Fe(H,0),2+ with Fe(H,O),w or of Fe(H,0)62f with Fe(H,0),0H2+. Dodson and his c o - w o ~ k e r s ~ ~ ~ conceived the activated complex of the second reaction as in (II). According to them electron transfer proceeds with the simul- (H,O),F$-O-H . . . . O-FG(H,O) I H 1 H 01) taneous vibration of the proton forming the bridge which really means the transition of a hydrogen atom. Amiss tried to treat quantitatively a similar model namely the acid-catalysed NpV-Npn exchange. Applying the procedure of Eyring and his co-workers he calculated the transfer probability Xe through a double potential barrier which is seen in Figure 4.However the value of 8.1 kcal./mole for dFfrearr. used by Eyring et al. seemed to be too large since it would result in an unbelievably large value for the Bohr radii of the electron in the corresponding ions. To obtain reasonable results the rearrangement free energy had to be lowered to about -4 kcal./mole but this is inconsistent with the Franck-Condon restric- tion. These contradictions may be caused by the rough approximations in the calculation whereas Dodson’s model is fundamentally correct. Reynolds and 34 J. Silverman and R. W. Dodson J. Phys. Chem. 1951 56 846. 35 R. W. Dodson and N. Davidson in the discussion of Ref. 11. 36 J. Hudis and R. W. Dodson J.Amer. Chem. SOC. 1956 78 911. 21 1 The Theory of T/iermal Electron-transfer Reactions in Solution ELECTRON COORDINATE Figure 4 Double potential barrier for the NpO,+-H+-NpO,a+ activated complex (Reproduced from E. S. Amis ‘Solvent Effects on Reaction Rates and Mechanisms’ Academic Press Inc. New York 1966 p. 110 by permission) Lumry3’ supposed the hydrogen ion to play a much moreextended r6le. They consider all electron transfers to proceed through an activated complex as in 011). H H (H,O),FeLO-H . ..O-H . . . 0-F&(H,O) I I H I ( n)l A (ID) This theory suggests that acid-catalysis should be observed for electron-transfer reactions since protonated Fe2+ ion is found at the left-hand end of the above chain. Such a phenomenon was observed for only a few reactions so the mech- anism did not seem generally valid.It is known that the mobility of H+ and OH- ions is greater by an order of magnitude in water than that of other ions. This can be explained by Grotthus’s concept38 that the ‘excess’ and ‘missing’ protons of the H,O+ ion and OH- ion respectively run through the chain of water molecules like a polarisation wave. Horne and Axelr0d,3~ renewing Reynolds and Lumry’s assumption attribute the mobility of ‘hydrogen atom’ to this mechanism. This has been said to be proved by the fact that the enthalpy and enthropy of activation for the Fe2+-Fe3+ electron exchange in ice was the same as in aqueous solution?0 37 W. L. Reynolds and R. Lumry J. Chem. Phys. 1955 23 2560. 38 C. J. T. Grotthus Ann. Chim. 1806 58 54. 39 R. A. Horne and E. H. Axelrod J. Chem.Phys. 1964,40 1518. 40 R. A. Horne J . Inorg. Nuclear Chem. 1963 25 1139. 21 2 Concentrations used were such (ca. 10-3~) that the ions were placed at an average distance of ca. 100 A. Under these conditions the electron had to pass through a layer of about 30 water molecules to reach the ion in the oxidised state. Home and Axelrod also assume that the total activation energy need of the reaction is consumed to form the chain of water molecules suitable to ‘transport’ the H atom. Thus energy is to be provided to produce the spatial arrangement of the water molecules of the chain. Obviously this energy is largely independent of the nature of the reactants (ca. 9.8-12.3 kcal./mole). The activation energies of the Fc-FeIIJ exchanges are apparently in agreement with this value (8-10 kcal./mole) but results obtained for the activation energies of other reactions do not agree with this suggesting that the agreement is probably accidental.The most striking discrepancy is however that the changes in the electronic energy throughout the transfer are not taken into account. It is evident that the electronic energy e.g. at an Fe2+ ion is not identical with that of the solvated hydrogen atom. If electronic excitation is required in addition to the 9-10 kcal./mole due to the suitable rotation how can such a low energy of activation be explained as the 4-6 kcal./mole value for the Fe(CN,&-Fe(CN):- exchange? Further contradiction exists concerning acid catalysis since this hypothesis also requires an H,O+ ion to transmit. Recently Sykes41 pointed out that a hydrated electron cannot exist as inter- mediate in electron-transfer reactions since the rate law for a reaction sequence such as (15) (16) is given by (17).This rate law is valid as far as the steady-state treatment is allowed to be used that is as far as the transferring charge can become independent in movement from the ion yielding it. Thus other things being equal it must be true for the Horne-Axelrod mechanism. Neither this rate law nor its limiting cases have been observed so far. (ii) The ‘band model’. Recently a new theory was published by based in part on the former concepts of indirect electron transfer but eliminating the contradiction within the assumptions and pointing to their extended applicability for both inner and outer sphere mechanisms. The fundamental supposition of this theory is connected with the question what is the intermediate ‘H atom’ like? Reynolds and Lumry as well as Horne A.G. Sykes ‘Kinetics of Inorganic Reactions’ Pergamon Press 1966 p. 222. 1. Ruff J. Phys. Chem. 1965 69 3183; Acta Chim. Acad. Sci. Hung. 1966 47,245; 1966 47,255; 1967 52 251. 213 The Theory of Thermal Electron-transfer Reactions in Solution and Axelrod speak of hydrogen atom as if it were a hydrogen atom as it would be in a gaseous phase but immersed in water without any alteration. This could not serve as a transporting particle since it could not move by a Grotthus mechanism because its proton would not take part in any hydrogen bonding. On the other hand if it is an H30 radical which is a reasonable assumption its life time (ca. s~c.*~) does not seem to permit its moving apart from its original position more or less close to the electron-donor ion.It can mediate the transfer however in such a fixed state if the electron cloud of H30 is widely spread. This can be expected to be so for the charge of the H30+ ion itself being the 'nucleus' of the quasi-hydrogen atom is spread over several water molecules and the wave function of the electron ordered around this charge must be highly extended in space. By the usual treatment for similar problems the Bohr radius of such' a 'hydrogen atom' has been estimated to be around 40 A. This value is of the same order of magnitude as the radius of impurity wave functions in semiconductors in which the dielectric constant is smaller but the effective mass for the electron is taken. It is noteworthy that Amis's treatment led to a similarly large value but whether this is due to the nature of the problem or to errors in approximations is not known.Thus H30+ can serve as a bridge for the transferring electron if its wave function covers both of the reacting ions i.e. even if they are so far apart as 80 A. Further the hydroxonium ion is not required to be either in the close vicinity of the reducing ion or between the reacting ions at the moment of the electronic excitation. The only restriction is that both ions must be inside the effectivity space of the H30 radical. The other feature of the H30 radical is that the energy of the electron bound to it is appreciably higher than that of any of the reactant ions. If not the water would decompose. It has been shown that these energies can be calculated by the normal oxidation-reduction potentials of the corresponding ions and so at least the difference between the potentials due to the process H30+ + e- $ H,O and Mn+ + e- + M (n - l) + must be supported as activation energy.The electron transfer takes place via two steps (18) and (19). If it is assumed M @ - + + H30+ --+ Mln+ + H,O (18) that H30 does not move apart from Mln+ i.e. it is not an independent particle Sykes's steady-state equation mentioned above should not be valid in this case. There is another possible way [(20) (21)] very similar to the former one where the reaction starts on the oxidising ion by electron acceptance and the 43 T. J. Sworski Adv. Chem. 1965 40 263. 214 intermediate particle is an OH radical. Every statement that has been written in connection with the H,O radical is true of this mechanism.Representing the possible states of the transferring electron as a function of the position co-ordinate and mark-in the extension of the corresponding electron clouds with the length of the horizontal lines due to the stationary levels (Figure 5) we can draw two approximately rectangular potential barriers 0 d Posiftron courdimte (broken lines). The upper one should be overcome in reaction (18) when in fact electron transfer takes place and the lower reversed barrier hinders the reaction (20) in which electron deficiency ix. a hole is transferred. The reactant levels in electron-exchange reactions are identical in energy (U = 0). The level of the reducing ion is occupied and that of the oxidising ion is unoccupied.For oxida- tion-reduction reactions the energy difference between the donor and acceptor level corresponds to that between the normal oxidation-reduction potentials of the reducing and oxidising ions respectively. This model is very similar to the band model of semiconductors the electron and hole transfer paths being analogous to the n and p type conduction respec- tively with the only exception that the H,O and OH states are not infinitely extended all over the system but only relatively with respect to the reactant distance. Two further possible reaction paths exist if one presumes tunnelling one through the upper barrier by electron tunnelling and another by hole tunnelling. Tunnelling occurs when the energy of activation is not enough for a ‘classical’ transfer above the barrier.Among the four possible paths that one will be Figure 5 Energy levels corresponding to the band model 215 The Theory of Thermal Electron-transfer Reactions in Solution favoured which requires the smallest free energy of activation. We shall turn now to the quantitative relationships based on this model. It has been shown that the entropyof activation can be calculated by equations ASi - R Clnpi i AStS = -2R In C* (22)-(25). Here Pe is the transition probability of the electron pi is the prob- ability of satisfaction of the ith configurational requirement of the activated complex (ASi$ represents that fraction of the collisions which take place at a favourable orientation) and c* is the so-called normalised concentration i.e. at this concentration the average distance between the reactants in solution is equal to the distance d between the reactants in the activated complex and is given by equation (26) where N is Avogadro’s number.AStS represents the entropy due to the change in the degrees of freedom of translation. In the case of transfer above the barrier the activation energy is larger or equal to U but then we have Pe = 1 if the reactants are situated inside the effectivity space of one H,O or OH radical. If we consider their above-mentioned Bohr radii this means that the reactants should approach one other at least to about 80 A Thus on the basis of equations (22)-(26) equation (27) follows since for the outer-sphere mechanism the reactants can be considered to be spheres i.e. there is no configurational restriction for the collision either.It has been an important result of this theory that it could explain the positive entropy of activation observed. The calculated value is compared with the experimental ones in Table 5 below the broken line. In view of the rough approxi- mations the agreement is quite satisfactory. All reactions listed in Table 5 proceed by hole transfer because the level of their oxidising reactivity is very near the lower energy limit (see Figure 5). Equation (20) shows that the rate of such hole transfers above the barrier should be a linear function of OH- con- centration i.e. the rate law should involve the reciprocal of the hydrogen-ion concentration. This pH-dependence should not be due to hydrolysis as was supposed in the original papers. The present theory predicts base catalysis unlike the former models for indirect transfer which would require acid catalysis in contradiction with experiments.Acid catalysis would appear in electron transfer above the barrier but this path would be favoured only in systems containing extremely strong reducing agents. In the case of the tunnel effect the fastest path corresponds to the minimum d. 21 6 Table 5 Reaction ASScalc. (e.u.) dSSobs. (e.u.) [H+]-order Hole transfer reactions NpvLNpv - 20 -24b +1 Fe(CN),3-Fe(CN),4- -21 -21d 0 F S F & - 23 -25' 0 MnM-MnVI - 16 - 16d 0 Vrv-VIII - 26 - 24" - 1 21 25f -2 21 219; 27h - 1 21 1 l i - 1 21 14j -1 21 22k -1 21 19&102 -1 Electron transfer reactions vn-v" - 28 - 25" 0 Co(NHdZ+-Co(NH.J:f - 32 -41n 0 Co(en) 32+-€0(en) 33+ - 32 -33p 0 CO(EDTA)~-CO(EDTA)- - 23 - 174 On Cem-CeN - 37 - 40f 0 a pH dependence due to equation (30).See Ref. b in Table 4. e See Ref. d in Table 1. f See Ref. f in Table 1. J. Shankar and B. C. de Souza J. Inorg. Nuclear Chem. 1963 24 693. h H. S. Habib and J. P. Hunt J. Amer. Chem. SOC.. 1966 88 1658. t L. E. Bennett and J. C. Sheppard J. Phys. Chem. 1962 66 1265. L. H. Sutcliffe and J. R. Weber Trans. Faraday SOC. 1956,52 1225. J. Halpern Canad. J. Chem. 1959,37 148. J. C. Sheppard J. Phys. Chem. 1964,68,1190. See Ref b in Table 3. N. S. Biradar D. W. Stransk M. S. Vaidhya G. J. Weston and D. J. Simpson Trans. Faraday SOC. 1959 55 1268. p See Ref. b in Table 1. Q A. W. Adamson and K. S. Vones J . Inorg. Nuclear Chem. 1956,3 203. For the outer sphere mechanism this was assumed to occur by the penetration of the second co-ordination spheres leading to a value of d about 9.4 A from which it follows that A&+ w 0.On the basis of the Gramow equation (28) the See Ref. d in Table 3. See Ref. 35. total entropy of activation is given by (29) since spherical symmetry is valid in R (29) this case too. Almost all the observed entropies of activation for electron exchange reactions are summarised in Table 5. The comparison seems to prove the applicability of this model. A slightly worse though satisfactory fitting has been observed for oxidation-reduction reactions. As tunnelling probability is independent of the 217 The Theory of Thermal Electron-transfer Reactions in So Iution height of the acceptor level the entropy of activation can be calculated by equa- tion (29) for both electron exchange and oxidation-reduction reactions.This contradicts the result in Marcus’s theory [equation (9)] but it arises from the approximation used for tunnelling; that is the changes in the barrier height caused by the overlap of the donor and acceptor levels cannot be allowed for by tunnelling but by the appropriate individual electronic orbitals which would be extremely difficult to calculate. For the tunnelling paths a slight pH-depend- ence can be expected; since the limit energies in Figure 5 vary linearly with pH In k = - aJ(U - Ea + 2-303 kTpH) + b (30) equation (30) can be obtained where a and b are given by equations (31) and (31) 47r h a = - dJ(2m) = 7.6 x lo6 g.-t cm.-l sec.-l (32). The sign before the term 2.303 kTpH is positive for electron transfer and negative for hole transfer.Equation (30) fits the pH-dependence of the reactions between Vrr and Urn Ferl and H,O, Puvr and Feu and ComEDTA and Con- EDTA with slopes 1.3 x los 6.8 x lo6 13 x lo6 and 9.7 x los respectively. An important feature of this theory is that the electrostatic interactions have been neglected because at the larger interionic distances supposed their effect is small. Thus the energy of activation is due to the rearrangement of the inner co-ordination spheres. This has been estimated by use of Van Vleck’s formulae by equation (22) where the terms of the right-hand side are the crystal field E a = (10Dq)ox - (10Dq)red (33) splittings of the (electron or hole) donor ion in its oxidised and reduced state respectively. This is somewhat similar to Hush’s treatment for estimating the crystal field stabilisation energy changes but theE a linear approximation has been used with respect to h.Because the acceptor ion is far from the donor it does not affect the excitation process i.e. the energy of activation is independent of the nature of the acceptor and of whether the actual path is electron or hole transfer. Some examples are shown in Table 6 which confirm this correlation. Equation (33) does not hold for the case when the electronic structure of the donor ion is much altered by the loss of the electron or hole e.g. when a high spin-low spin transformation takes place [Co(en)?+ - C~(en),~+]. Unlike Marcus’s theory the results of this treatment for electrode reactions show that there should be no difference in the energy of activation between electrochemical and chemical electron-transfer reactions (see e.g.the values marked by ‘electrode’ in Table 6). The difference exists only in the entropy terms since one ion may approach the electrode more easily than the other ion so that the tunnelling distance is decreased. Some calculated exchange current values are in good agreement with the measured ones. 21 8 CoIII -+ COII --f COII c o n 1 4 cox1 Fe(CN):- -f Fe(CN),4- Fe(CN):- -+ Fe(CN),*- CO(EDTA)~- -+ Co(EDTA)- Reaction Ea (talc.) partner (kcal./mole) VIII 11.2 VIII 13.2 UVI 13.2 Co(NH3),3+ 13.2 Ed1 13.2 FeII 9.5 H202 9.5 electrode 9.5 COII 24.6 FeII 24-6 C& 24-6 Tll 24.6 electrode - Co(EDTA)- 26.1 Fe( CN):- - Ea (obs.) (kcal. /mole) 10*7e 13.2" 7.3" 9.7 12.0 y 9.9d 9.48 9 E 21.69 28.5h 17.22 19.0' 26-4k 4*6d 4" 224 Co(NH3),2+ - Co(NH3):+ CO(NH,),~+ 64.0 13*5n C~(en),~+ - Co(en),3+ Co(en),3+ 64.0 1 4 ~ 3 ~ For the references denoted by roman letters see the same letter in Table 5 ; a T.W. Newton and F. B. Baker J. Phys. Chem. 1965 69 176. A. Zwinkel and H. Taube J. Amer. Chem. SOC. 1961 83 793. Y A. Adin and A. G. Sykes Nature 1966 209 804. J. Sobkowski Roczniki Chem. 1961 36 1503. See Ref. a in Table 4. 4 Inner Sphere Mechanism As has been noted the a pviori theoretical investigation of inner sphere electron- transfer reactions is more difficult owing to several complications which do not appear in outer sphere mechanism. These are (i) the lack of spherical symmetry of the reactants in their collision since they must collide in the direction of the bridging ligand; (ii) the close contact of the reactants that results in larger over- lap in the electronic orbitals; (iii) the lower symmetry makes the calculation of the rearrangement free energy more difficult; (iv) the nature of the bridging ligand cannot be considered exclusively to be a point charge or dipole but its electronic orbitals significantly influence the electron transfer.These complications cause our knowledge of the inner sphere mechanism to be based primarily on conclusions of experiments reviewed el~ewhere,~~' which fall outside our area of a priori theories. It is possible that some theoretical information may be obtained by a general quantum mechanical treatment but this would yield only qualitative results or by a simpler model like that of the outer sphere mechanism which could at least estimate the kinetic parameters.The quantum mechanical treatment as outlined by Halpern and Orge1,44 44 J. Halpern and L. E. Orgel Discuss. Faraduy SOC. 1960 29 32. 219 The Theory of Thermal Electron-transfer Reactions in Solution starts from the following assumptions the energy conditions of the binuclear activated complex are entirely identical with those obtained by the Marcus theory except that the transfer probability of the electron estimated to be about unity in the outer sphere case is now much less and markedly regulated by the nature of the bridging ligand. On this basis the detailed mechanism of the electron transfer can be direct double and superexchange. Slightly vulgarising the question one can say that direct exchange occurs when the time for which the transferring electron is located on the bridging ligand is much less than that for which the electron is in its initial or final positions in the course of its oscilla- tion between these.If not double exchange appears i.e. the electron jumps first to the bridge and after that to the acceptor ion [equation (34)] or first from the MlL-M2+ 4 [Ml+L-M2+] -+ M1+L-M2 (34) ligand to the acceptor then from the donor to the ligand [equation (25)]. M1L-M2+ - [MlLM2] -+ Ml+L-M (3 5 ) In superexchange the activated states in equations (34) and (35) are mixed MlL-M2+ -+ [Ml+L+M2] -+ Ml+L-M as in equation (36) that is the hole starting from M and the electron of MI are recombined at the ligand. Halpern and Orgel emphasise the r6le of conjugation of then electrons in organic bridging ligands and of the symmetry of the electronic orbitals in transition-metal ions as well.Conjugation is an accelerating effect owing to the high ‘conductivity’ of the bridge. The orbital symmetry of the transferring electron may affect the rate as follows in the case of eg and tSg electrons transfer via 0 and n type molecular orbitals is favoured respectively. Further on the oxidation-reduction potential of the ligand is predicted to be important in the mechanism of superexchange. All these qualitative theoretical results are in excellent agreement with the measured rates of different bridged systems. As can be seen there is a high similarity between the double exchange mech- anism represented by equations (34) and (35) and the electron and hole transfer paths in Ruff’s theory. In this way the band model was extended recently to the inner sphere me~hanism.4~ The simplified method of calculation of the transfer probability gives quantitative data for the entropy of activation.The approxi- mate rectangular potential barrier in Figure 5 is lowered if the oxidation- reduction potential of the bridging ligand differs favourably from the corre- sponding limit potential; at the same time the barrier width becomes smaller owing to the reduced tunnelling distance in the bridged complex. In addition the asymmetric behaviour of the reactants in collision should be taken into account. The entropy values calculated in this way for halides as bridging ligands are around -25 e.u. in good agreement with the measured ones. 46 I. Ruff J. Phys. Chem. in the press. 220 5 Conclusions Although the theories reviewed show essential differences in the model of electron-transfer reactions it is clear that the grounds of the detailed mechanism are revealed.The phenomena affecting the rate of these processes are in general (i) the collision of the reactants (ii) the rearrangement of the environment and (iii) the probability of the electronic jump. It is also obvious that an ab initio quantum mechanical treatment would be the most sophisticated theoretical approach but even if its difficulties are disregarded its results would be in- dividual which is never the purpose of any theory. Thus further development can be expected to be based on refinement of the approximations used so far perhaps with a unification of the advantageous properties of the various theories; it will be desirable to retain general validity since the purpose is to understand and explain the experimentally observed behaviour of oxidation-reduction reactions and not to replace measured values with calculated ones. 221 5

ISSN:0009-2681    

DOI:10.1039/QR9682200199

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents By J. I. G. Cadogan UNIVERSITY OF ST. ANDRBWS SCOTLAND Tervalent organophosphorus compounds (X,P) such as trialkyl- or triaryl- phosphines and trialkyl phosphites react with a wide variety of oxygen-contain- ing compounds to yield the corresponding quinquevalent derivatives (X,PO),l as in equation (1). The major driving force behind these reactions is the great X,P + zo + x,PO + z equation 1) strength of the P=O bond formed typical values for P=O bond dissociation energies in phosphates and phosphine oxides lying for example in the range of 12&150 kcal./mole,2 which can be compared with values in the range 50-70 kcal./mole for the +N-0- bond in amine oxides. Very many examples of the general reaction outlined in equation (1) have been known for some time but those involving reduction of nitro- and nitroso- compounds have received widespread attention during the last few years only.Such studies have led to the recognition of the reaction of reagents such as triethyl phosphite with aromatic nitro-compounds as one of considerable synthetical value and this Review is directed towards the mechanism and scope of this and related processes. Reduction of Aromatic Nitroso-compounds Most of the reported reactions of nitroso-compounds with tervalent phosphorus reagents have involved aromatic systems. The first record of such a reaction was by Hoffmann and Homer who stated in a reference to unpublished work that substituted nitrosobenzenes (ArNO; Ar = p-C1 p-Me p-NMea but not nitro- sobenzene itself reacted (equation 2) with triphenylphosphine to give the corre- sponding azoxybenzenes (ca.50 %). Bunyan and Cadogan5p6 later showed that ArNO + Ph,P -+ ArN=N(O)Ar + Ph,PO (Equation 2) reduction of nitrosobenzene by triphenylphosphine or triethyl phosphite did give azoxybenzene in low yield together with a polymer if the reaction which was exothermic was carried out in benzene. Reduction of o-ethylnitrosobenzene J. I. G. Cadogan Quart. Rev. 1962 16,208. a S. B. Hartley W. S. Holmes J. K. Jacques M. F. Mole and J. C. McCoubrey Quart Rev. 1963 17,204. * T. L. Cottrell ‘The Strengths of Chemical Bonds’ Butterworths London 1958. H. Hoffmann and L. Homer Angew. Chem. 1956,68,473 P. J. Bunyan and J. I. G. Cadogan Proc. Chem. SOC. 1962 78. P. J. Bunyan and J. I. G. Cadogan J. Chem. Soc.1963 42. 222 Cadogan and p-dimethylaminonitrosobenzene similarly gave the corresponding azoxy- compounds but in the latter case the reaction proceeded more slowly and triethyl N-p-dimethylaminophenylphosphorhidate (1) was isolated either directly or in the case of work-up involving activated alumina as its hydrolysis product diethyl N-p-dimethylaminophenylphosphoramidate (2) (Scheme 1). ArNO+(EtO)3P-+ArN =N (0)Ar + (E tO),PO B>Et-’ + EtO-PTNAr-+(EtO),P(O)NHAr EtO’ \H+ (1) Scheme These observations led to the suggestion ArNO + (EtO),P -t (EtO),PO + ArN ArN + ArNO + ArN=N(O)Ar ArN + (EtO),P 3 (EtO),P=NAr (2) (Ar = P-Me2N.C,H,-) 1 that the reactions could involve the (Equation 3) (Equation 4) (Equation 5 ) intermediacy of nitrenes [reactions (3)-(5)]. Thus in the case of p-dimethyl- aminonitrosobenzene the initial reaction would be expected to occur less readily and the derived nitrene would be more stable as a result of contributions of the form (3) hence increasing the chance of its capture by the strongly nucleophilic triethyl phosphite in a manner analogous to the capture of carbenes by triphenyl- ph~sphine.~ In accord with this equimolar amounts of the reactants gave the azoxy-compound (63.5 ‘4 and the phosphorhidate (1) (13 %) while a tenfold excess of triethyl phosphite over the nitroso-compound led to the predominance of the phosphorimidate (63-5 %) over the azoxy-compound (23 %).On the other hand if the reduction of o-ethylnitrosobenzene proceeds via a nitrene it might be expected that some insertion into the side chain would have been observed (equation 6).Bunyan and Cadogan detected no indoline however but pointed out that the absence of this product could be due to a greater ease of reaction of the nitrene if present with the nitroso-compound to give 2,2’-diethylazoxy- benzene (47 %). They also suggested that the reduction of o-dinitrosobenzene to benzofurazan (4) (Scheme 2) by triphenylphosphine and related reactions D. Seyferth S. 0. Grim and T. 0. Read J. Amer. Chem. SOC. 1960,82 1510. 223 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Scheme 2 reported contemporaneouslyY8 might proceed via a nitrene. It is of interest that o-nitronitrosobenzene is smoothly reduced by triethyl phosphite to the furazan oxide ( 5 ; 19%) which in turn gives the furazan (4) on further reduction under more forcing conditions (150°).9 It should be noted that the results of these experiments did not exclude routes (equation 7) to azoxybenzenes or a similar route (equation 8) to benzofurazan.ArNO dimer i- (EtO),P+ (EtO),PO +Ar N = N ( 0 ) Ar 4 (Equation 7) (Equation 8) Regardless of mechanistic detail however this reduction of nitroso-com- pounds under certain conditions provides a good route (equation 9) to azoxy- compounds and has been exploited as such.1° C6F5N0 f (EtO),P -+ C6F5N =N(0)C6F5 (80 %) (Equation 9) A more important synthetical development of this reduction involves the reductive cyclisation of 2-nitrosobiaryl.~~,~ to carbazoles. Thus 2-nitrosobiphenyl reacts (equation 10) within a few minutes with triethyl phosphite in benzene or ether at 0" to give carbazole (76%) and triethyl phosphate (84%).Triphenyl- phosphine is also effective but it is simpler in practice to use triethyl phosphite since both it and triethyl phosphate are easily removed by distillation. Phos- phorus trichloride does not react indicating that a strongly nucleophilic phos- phorus atom is required. With use of triethyl phosphite 3-o-nitrosophenyl- pyridine (6) gave a mixture (64%) of ~(81.5%) and 7-carboline (183%) (equa- tion l l) while 2-o-nitrosophenylpyridine gave an almost quantitative yield of pyrid[l ,Zb]indazole (7) (equation 12). The latter observation was considered6 to be in accord with the concept of an electron-deficient nitrogen intermediate which would react preferentially at the electron-rich ring-nitrogen atom. Further pyrid[lY2-b]indazole (57 %) has been obtained from the decomposition of 2-0-azidophenylpyridine~~~ there being a J.H. Boyer and S . E. Ellzey J. Org. Chem. 1961 26,4684. @ J. I. G. Cadogan M. Cameron-Wood R. K. Mackie and R. J. G. Searle J. Chem. Soc. 1965,4831. lo J. Burdon C. J. Morton and D. F. Thomas J. Chem. SOC. 1965 2621. l1 R. A. Abramovitch and K. A. H. Adams Cunud. J. Chem. 1961 39,2516. 224 Cadogan (Equation 11) (Equation 10) - + (Equation 12) considerable amount of evidence,12,1s particularly for photo-induced reactions,14 that this type of reaction proceeds via a nitrene. Convincing support for the intermediacy of a nitrene in the reduction of nitroso-compounds by triphenylphosp hine has recently been provided by Odum and Brenner15 following work on the photolysis16 and pyrolysis17 of phenyl azide in amines reactions which proceed via nitrenes which undergo ring expansion in the presence of amines with the ultimate formation of derivatives of 2-amino-3-H-azepines (9).Huisgen and his co-workers17 suggested that the reaction involved the 7-azabicyc10[4,1 ,O]heptatriene (8) e.g. equation (13). The reported isolation of azepines of this type from reactions of nitroso- benzene with triphenylphosphine in the presence of dialkylamines is therefore good evidence in favour of the participation of nitrenes15 in these cases also. Recent work by Sundberg18 also points to the intermediacy of nitrenes in the reactions of o-methyl- o-propyl- and o-butyl-nitrosobenzene carried out in excess of triethyl phosphite in the absence of a solvent. Under these conditions which are very different from those used by earlier worker^,^^^ which involved more nearly equimolar proportions of reactants in a solvent little of the corre- sponding (ca.1 %) azoxy-compounds were isolated and low yields (ca. 5-1 1 %) of triethyl N-arylphosphorimidates (1 ; Ar = O-Me.C,H,; o-Pr-C,H,; o-Bu-C,H,) were indicated by g.1.c. and n.m.r. studies on mixtures of products. l2 P. A. S. Smith and J. H. Boyer J. Amer. Chem. SOC. 1951 73 2435. l3 G. Smolinsky J. Amer. Chem. SOC. 1960 82 4717; 1961 83 2489; J. Org. Chem. 1961 26 4108. l4 G. Smolinsky E. Wasserman and W. A. Yager J Amer. Chent. SOC. 1964 86 3166; G. Bowes A. Reiser and H. Wagner Tetrahedron Letters 1966 2635. l5 R. A. Odum and M. Brenner J. Amer. Chem. SOC. 1966,88,2074. l6 W. von E. Doering and R. A. Odum Tetrahedron 1966 22 81. l7 R.Huisgen D. Vossins and M. Appl Chem. Ber. 1958 91 1; R. Huisgen and M. Appl ibid. p. 12. I6 R. J. Sundbtrg J. Amer. Chem. SOC. 1966 88 3781. 225 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents In addition small quantities (1-7%) of amines were detected by g.1.c. and in the case of o-nitrosotoluene at O" N-o-tolyl-a-methyl-a-(2-pyridyl) nitrone (10; R=Me; Ar=o-tolyl)) was isolated in varying amounts (ca. 2&30%) according to conditions. At 156" the corresponding anil (11) was also formed by deoxy- genation. The relatively lower yields of azoxy-compounds and higher yields of phosphorimidates (1) obtained in these cases compared with those studied by Bunyan and Cadogan6p6 are as expected as a result of the presence of a large excess of triethyl phosphite.The nature of the amines detected vary with reaction conditions; thus at 0" a-alkylanilines (ca. 2 %) are formed while slightly higher yields (ca. 5-7 %) are produced at 156" together with traces of what are believed to be indolines (12; R=Et or Me) formed by insertion reactions of a nitrene in the side chain. In view of the low yields and lack of evidence based on actual isolation of products little mechanistic significance can be attached to these observations as is readily pointed out by Sundberg. The mode of formation of the nitrone (10) is not yet established but Sundberg has drawn attention to a possible route involving first a nitrene and then a 7-azabicyclo[4,1 ,O]hepta-2,4,6-triene intermediate of the type discussed above which is supposed to react with another molecule of the nitroso-compound (Scheme 3) although at the high temperature and with a relatively low concentration of the nitroso-compound it would be surprising if this reaction between two highly reactive intermediates had a high probability of taking place.Scheme 3 It is clear therefore that several aspects of the mechanism of reduction of nitroso-compounds by tervalent. phosphorus reagents still remain to be clarified (See Appendix). Reduction of Aromatic Nitro-compounds In theory the triethyl phosphite-induced reductive cyclisation of 2-nitrosobiaryls to carbazoles is of some synthetical value. In practice the preparative significance of this reaction is reduced by the difficulties sometimes encountered in the 226 Cadogan preparation of the nitroso-compound from the corresponding nitrobiaryl.This difficulty was overcome and wide synthetical possibilities revealed by the demon~tration~,~~ that reaction of the parent 2-nitrobiphenyl with a slight excess over two equivalents of boiling triethyl phosphite gave carbazole in excellent yield (83%) (equation 14). Extensions of this reaction reported at the included the conversion of both cis- and trans-2-nitrostilbene and a-nitrostilbene into 2-phenylindole 2-nitrobenzylidene anilines into indazoles 2-nitroazoarenes into benzotriazoles and various dinitro-compounds into polycyclic derivatives containing five-membered rings. Since then many additional ramifications of the reductive cyclisation have been reported and for the purpose of this Review it is convenient to classify the results of the various investigations into the scope and mechanism of the reactions in terms of the type of ring system produced.Formation of Carbazoles and Related Systems.-The triethyl phosphite-induced reduction of 2-nitrobiphenyl has been extended to give various bromo- chloro- and methyl-carbazoles in moderate to good yields (35-83 2,2'-Dinitro- biphenyl on the other hand gave only a low (1.5%) yield of benzo[c]cinnoline (13) rather than the possible alternative (14),9 the former compound being readily produced (86 %) by reduction of benzo[c]cinnoline NN'-dioxide under similar conditions. It is of interest that 2,2'-dinitrobiphenyl is converted into benzo[c]cinnoline "'-dioxide by phosphine in allcali,2lU this reagent having previously been used to produce azoxy-compounds from nitroarenes21 Scheme 4 la J.I. G. Cadogan and M. Cameron-Wood Proc. Chem. SOC. 1962 361. *O J. I. G. Cadogan and R. J. G. Searle Chem. and Ind. 1963 1282; 1434. 81 (a) A. G. Bellaart Tetrahedron 1965 21 3285; (b) S. A. Buckler M. Epstein L. Doll and F. K. Lind J. Org. Chem. 1962 27 794. 227 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Apart from the above case 1,s-cyclisation is the preferred reaction in phosphite reductions thus 1-o-nitrophenylnaphthalene which presents two possible points of ring closure is reduced to 3,4-benzcarbazole (15) rather than the isomeric mesoben~acridine~ (1 6) (Scheme 4). Reductive cyclisations can be effected by other tervalent phosphorus reagents and a comparative study of their relative efficiencies has been made by Cadogan and Todd22a who followed the rate of removal of 2-nitrobiphenyl on reaction with an excess of the reducing agent (15-20 moles).Although the precision of the g.1.c. method of analysis which they used was not high (15%) the results (Table 1) are useful in that they establish the order of reactivity of the phosphorus reagents thus (EtO),PMe > (Et,N),P N (EtO)P(NEt,) > (EtO),P - (PriO),P > > PCl (inactive) and indicate that the reaction carried out in a polar solvent is not faster. Qualitative experiments established that tributyl- and triphenyl- phosphine were also suitable deoxygenating agents but that neither was as reactive as the phosphorous amides or diethyl methylphosphonite. The results suggest that the rate-determining step involves nucleophilic attack by phos- phorus but the extremely rapid reaction of diethyl methylphosphonite cannot be explained on this basis alone yet it would be surprising if such an enhance- ment in rate arose from simple steric effects alone.The method of analysis Table 1 Deoxygenation of 2-nitrobbhenyl (Pr 'O),P (EtO),P (EtO),P (EtO),P (EtO)P(NEt,) (Et,N),P x3p (EtO),P + Me2NCO-H (equimolar) (EtO),PMe Temp. 143.5" 155 145 135 144 121 111 61 ti (min.) 63 32 50 83 97 ca. 5 41 154 employed in these kinetic experiments unfortunately could be extended to include only two substituted nitrobiaryls (Table 2) but they show as would be expected for a mechanism involving deoxygenation rather than ring closure as the rate-determining step that the methyl group in 4-methyl-2'-nitrobiphenyl causes no change in rate compared with 2-nitrobiphenyl while the slight increase in rate observed with 4-bromo-2-nitrobiphenyl lends support to the probability of nucleophilic attack by phosphorus at the nitro-group.A key question of mechanism still to be answered in these reactions is whether the nitro-compound is reduced to the carbazole via the corresponding nitroso- compound. In practice this is difficult to establish of course because of the great ease with which 2-nitrosobiphenyl reacts with triethyl phosphite to give carbazole ra (a) J. I. G. Cadogan and M. J. Todd Chem. Cumm. 1967 178; (b) unpublished results; (c) J. I. G. Cadogan S. Kulik and M. J. Todd unpublished results. 228 Cadogan Table 2 Deoxygenation of nitrobiaryls by triethyl phosphite (145.5 ") X H H 4-Br Y th (min.) H 50 4'-Me 50 H 17 (ti = ca. 2 min. at 0").In the absence of evidence to the contrary however it remains a reasonable possibility (See Appendix). The main point of interest concerning the mechanism of the ring closure centres however on the possibility of either the participation of a nitrene formed from the intermediate nitroso-compound or by another more direct route (Scheme 5). The question of the point of initial attack by the phosphorus atom is also open to discussion. In the cases of both the nitro- and nitroso- groups it can be argued that initial attack occurs at the more positively polarised nitrogen atom followed by rearrangement possibly via a three-membered cyclic intermediate to a dipolar structure in which the phosphate leaving group is latent. Alternatively the latter may be formed by direct attack on the oxygen atom.Thus for reduction of the nitro group Scheme 6 might apply a similar dipolar intermediate (RO),P+-O-N-Ar being formulated in the case of the nitroso-compound. Deoxygenation of such acyclic and cyclic intermediates could also lead to the phosphorimidate (1) directly. Scheme 5 + ?- 0- /7 (RO),P-O-NAr-+ (RO),PO - ArNO IR0)jP +O=I$-Ar .T (ROi,P/p ' 0- Scheme 6 'Nil. Evidence for nitrenes is usually taken to be the occurrence of abstraction and insertion reactions with C-H bonds.23 Thus the formation of 8,lO-dimethyl- na R. A. Abramovitch and B. A. Davis Chem. Rev. 1964,64 149. 229 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents phenanthridine (16; 50%) 2,4,9-trimethylcarbazole (17; 5 %) and 2’-amino- 2,4,6-trimethylbiphenyl (18; 30%) has been attributed to the participation of a nitrene in the decomposition of 2’-azido-2,4,6-trimethylbiphenyl (19 ; X = N3) in hexadecane at 230”.% Following this useful results bearing on the mechanism of the phosphite-induced deoxygenation have been obtained from the reaction of the corresponding 2’-nitro-2,4,6-trimethylbiphenyl (19; X = NO2) with a slight excess of triethyl phosphite i.e.under the normal conditions of the reaction.a2a The formation of 2’-amino-2,4,6-trimethylbiphenyl (1 8; 13 %) in this case suggests abstraction by an electron-deficient nitrogen species such as a nitrene a suggestion supported by the formation of triethyl N-(2’,4’,6-trimethyl- biphenyl-Zyl) phosphorimidate (20; 15 ‘by presumably by reaction of the nitrene with excess of triethyl phosphite (Scheme 7).8,lO-Dimethylphenanthridine (1 6) and 2,4,9-trimethylcarbazole (17) were not formed under these conditions presumably because their formation would have involved the energetically less favourable insertion into strong bonds in competition with the easier pathway involving coupling with excess of phosphite present. In the corresponding reaction of the azide (19; X = N& the latter low- energy pathway is not available hence insertions occur. In accord with this deoxygenation of the nitrobiaryl in an excess (ca. 50 mol.) of (A) isopropyl- benzene and (B) t-butylbenzene under which conditions the coupling reaction to give the phosphorimidate (20) would be expected to be reduced and the inser- IX=N,; 230’ X=NO (EtO),P + PhCHMe (16)+(18)+(20) OMe (22) Me- ArN.+ PhCHMe? + ArNH + PhCMe,.-+ PhCMeiCMezPh (21) ArN. + (EtO),P + (EtO),P=NAr Scheme 7 tion reactions to be enhanced gave in addition to (18) and (20) the insertion product 8,lO-dimethylphenanthridine (16) [(A) 14 %; (B) 12x1. Further in (A) the amount of hydrogen abstraction increased [(18); 32%; cf. 19% in (B)] l4 G. Smolinsky J. Amer. Chem. Soc. 1960 82 4717. 230 Cadogan with the concomitant formation of bi-a-cumyl (21 ; 11 %). Thus when the possibility of coupling with triethyl phosphite is reduced the presumably less energetically favourable abstraction and insertion process occur to an appreci- able extent. Bi-a-cumyl can only arise by dimerisation of a-cumyl radicals produced by abstraction of a hydrogen atom from cumene by a triplet species presumably a nitrene (Scheme 7). The difference between the yields of 2’-amino- 2,4,6-trimethylbiphenyl(l8) formed in experiments (A) and (B) is almost exactly equivalent to the amount of bi-a-cumyl formed in the reaction in cumene t-butylbenzene having no readily abstractable hydrogen atom.In accord with these observations Smolinsky and FeueF have also isolated after hydrolysis of the products of the reaction of triethyl phosphite and 2’-nitro-2,4,6-trimethyl- biphenyl (19; X = NO& a trace of the amine (18) and a compound tentatively identified by n. m . r. as diet h yl N-(2’ 4’,6’-trimet h yl biphenyl-2-yl)p hosp horami- date (22). This appears to be a correct assignment because Cadogan and Todd established22a that the first formed phosphorimidate (20) is easily hydrolysed to the phosphoramidate (22) and in general the formation of phosphoramidate in such reactions usually points to the prior production (equation 15) of the phosphorimidate.6p26 (EtO)sP=NAr --t (EtO)2P(O)Nm (Equation 15) Related to this problem is the suggestion referred to above that phenyl- nitrene and 7-azabicycl0[4,1 ,O]hepta-2,4,6-triene (8) are successive intermediates in the ring expansion which occurs on decomposition of phenyl azide in amines,l6~l7 and as described earlier this ring expansion has been utilised16 to support the suggestion6 of nitrene participation in the deoxygenation of nitrobenzene by triphenylphosphine.Application of this test to the case of nitro- compounds was difficult at first in view of the very much more drastic conditions required but this difficulty disappeared with the demonstration of the high reactivity of diethyl methylphosphonite as a reducing agent.22a Thus reaction of the latter with 2-nitrobiphenyl in an excess of diethylamine gave 2-diethyl- amino-3H-3-phenylazepine (23; 13 in addition to carbazole (67 % compared with 86% in the absence of amine).At fist sight the formation of the azepine derivative suggests the participation of a nitrene in a manner analogous to that generally accepted15,16 following the suggestion by Huisgen and his co-workers17 (Scheme 8). Scheme 8 as G. Smolinsky and B. I. Feuer J. Org. Chem. 1966,31,3882. * J. I. G. Cadogan and H. N. Moulden J. Chem. SOC. 1961,3079. 231 Reduction of Nitro- and Nitroso-compounds by Tervafent Phosphorus Reagents Scheme 9 Scheme 10 The concomitant formation of carbazole in this reaction calls for a more detailed examination of the possibilities however.From the foregoing the most plausible explanation for the formation of carbazole involves attack by the nitrene at the 2’-position of the biaryl either by radical abstraction of a hydrogen atom followed by radical recombination or by direct insertion into the C-H bond (Scheme 9). The isolation of both carbazole and the azepine derivative suggests the following possibilities (1) That both products arise from the same intermediate which would be either the nitrene or the derived azabicyclo- heptatriene (Scheme S) or (2) that both intermediates are present in equilibrium (cf. Abramovitch and Davis23) carbazole arising from the nitrene and the azepine from the azabicycloheptatriene. If a common intermediate is involved (alternative 1) it appears to be unnecessary to invoke the sequence nitrene- azabicycloheptatriene-carbazole and an alternative explanation does not require the intermediacy of the highly strained azabicycloheptatriene system at all but rather requires a ring expansion concerted with attack by the nucleo- philic amine in competition with reaction of the nitrene with the aromatic ring (Scheme 10).Against this it would be expected that direct reaction of the nitrene with an amine would produce the diethylhydrazine ArNH.NEt,. Regardless of this detail however it is clear that the similarity of mechanism of the decomposi- tion of azides believed to involve nitrenes and of the deoxygenation of nitro- biaryls by triethyl phosphite is established. Formation of Indoles and Related Systems.-By analogy with the cyclisation of 2-nitrobiaryls to carbazole o-nitro-styrenes or -stilbenes would be expected to give rise to indoles and these possibilities have been realised experimentally (Scheme 1 1).Thus cis- and trans-Znitrostilbene and a-nitrostilbene give 2- phenylindole (85 58 and 16 % respecti~ely),~~~~ and a-(o-nitrophenyl)-2- chlorocinnamic acid gives 2-(o-chlorophenyl)-3-ethoxycarbonylindole (22b; 46%) on reaction with triethyl phosphite indicating the occurrence of esteri- 232 Cadogan fication presumably by triethyl phosphate during the reaction. The reductive cyclisation of 2'-nitro-a-stilbazole which presents two possible points of ring closure also gives rise to an indole (24) rather than proceeding via reaction at the electron-rich nitrogen to give a diazepine.22b o-Nitrostyrene and 2,2'- dinitrostilbene gave small yields (ca.1-2 %) only of indole and indolo[2,2-b]- indole (25) respectively while b-aitrostyrene gave no indole and o-nitrocin- namic acid was reduced and esterified to give a low yield of indole-2-carboxylic ester. J Scheme 11 H H Taylor and Garciaz7 have prepared in low yield two of the biologically interesting pyrrolo[3,2-d]pyrimidines (27) from the corresponding 5-nitro-6- styrylpyrimidine derivatives (26) thermally and by the irradiation in the presence of triethyl phosphite. It is not clear in the latter case whether the reaction is photochemically initiated because the experimental details provided indicate that 27 E. E. Taylor and E. C. Garcia J. Org. Chem. 1965 30 655. 233 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents some warming of the reaction mixture occurred.In this case therefore the products may be arising via a thermal process. It is tempting to extend the analogy between the reductive cyclisation of nitro- biaryls to carbazoles and of o-nitrostyrene derivatives to indoles to include a common mechanism but some interesting results reported by Sundberg in the course of an extension of this indole synthesis indicate that this may not be valid. Thus Sundberg2* confirmed the synthesis of 2-phenylindole from 2-nitro~tilbene~ and extended it to include 2-alkyl- (Me Et ; 50-60 % yields) and 2-acyl- (MeCO PhCO; 16%)-indoles. Of more interest are the by-products of these reactions. Thus deoxygenation of 2-nitro-trans-stilbe gave compounds formulated as 2,2’-diphenyl-3,3’-bi-indolyl(28 ; 7 %) and diethyl2-phenyl-3-indolylphosphonate (29; 1.6%) although the possibility that the latter is the phosphoramidate (30) is not rigorously excluded by the evidence presented.Postulating l-hydroxy-2- phenylindole (31) as an intermediate Sundberg proceeded to isolate this fFom a deoxygenation interrupted well before completion and further demonstrated that on being heated with triethyl phosphite this product gave a product dis- tribution similar to that obtained from the 2-nitro-trans-stilbene (Scheme 12). Bi-indolyls were also obtained from the deoxygenation of p-methyl- and p- propyl-o-nitrostyrene together with small amounts of the corresponding 1- ethoxy-2-alkylindoles but l-ethoxy-2-phenylindole was never detected among the products of the reaction of 2-nitro-trans-stilbene the most thoroughly investigated reaction.Scheme 12 The demonstration that 1-hydroxy-Zphenylindole is an intermediate indicated that in this case Scheme 13 may be followed at least in part thus raising an important point of difference between this and the corresponding reductive cyclisation of 2-nitrobiaryls. The mode of formation of the bi-indolyl in these reactions has not been estab- lished but it is significant that the yield of the diphenylbi-indolyl(28) is increased at the expense of 2-phenylindole if the proportion of triethyl phosphite is reduced by performing the reaction in triethyl phospkate or p-cymene. Thus an inter- s$ R. J. Sundberg J. Org. Chem. 1965 30 3604. 234 Cadogan $9 r:r; \ N-O /N-0 D h ( E t 016 m3P;% ' y-0 OH WPh 0 Scheme 13 mediate in the reaction of triethyl phosphite with 2-nitrostilbene can be diverted to the bi-indolyl (28) when the concentration of the phosphite is reduced but Sundberg notes that the yields of products are not high enough for it to be stated with certainty that this unknown intermediate is converted into 2-phenylindob in excess of the phosphite although this seems likely.If a triplet nitrene is an intermediate in this reaction it would have been expected that some 2-amino- trans-stilbene and ad-bi-p-cymyl would be formed. Sundberg did not comment on this possibility however and it is not clear from the experimental details provided whether these compounds were sought. In the corresponding reaction of #$%disubstituted o-nitrostyrenes,as indoles are again major products. Thus cyclohexylidene(cmitropheny1)methane (32) underwent ring closure with rearrangement to give 5,6,7,8,9,1O-hexahydrocyclo- hept[b]indole (33; 35 %) together with lower yields of the bi-indolinyl(34; 24 YJ and the spiro-indolinone (35; 8%) (Scheme 14) (See Appendix).Similarly @-dimethyl-o-nitrostyrene gave 2,3-dimethylindole (33 %) the indolinone (36; 11 'A and a very low yield of the bi-indolinyl(37) (Scheme 19 Scheme 14 Scheme I5 aB R. J. Sundberg and T. Yamazaki J. Org. Chem. 1967,32,290. 235 Reduction of Nitro and Nitroso-compounds by Tervalent Phosphorus Reagents while a-methyl-2’-nitrostilbene gave a high yield of the rearranged indole (38 ; 77%) together with the N-ethylated product (39; 21 %) presumably formed by alkylation of the first formed indole with triethyl phosphate (Scheme 16).Scheme 16 It is noteworthy that in this case in contrast to the reductive cyclisation of simpler o-nitrostyrenes discussed above none of the corresponding 1 -hydroxy- indoles were detected even in the case of interrupted reactions. Thus in these instances it cannot be said whether or not nitrenes or 1-hydroxyindoles are intermediates but it is clear from the rearrangement which occurs particularly in the case of a-methyl-2’-nitrostilbene that an electrophilic nitrogen species is involved and likely possibilities are outlined in Scheme 17. 0’ ‘&-P(OEt) ‘ U I W P’ Scheme 17 The genesis of the by-products is even less obvious; the formation of bi- indolinyls suggests a homolytic process possibly via an intermediate nitroso- compound (Scheme 18) or via a triplet nitrene (Scheme 19).Against Scheme 18 is the requirement of an electron donor for the first step to be possible and the unlikelihood of realising a sufficiently high concentration of the nitroso-com- pound under the experimental conditions. Scheme 19 allows a rationalisation of all the observed products in terms of a common intermediate. Further experi- mentation to obtain a better accountance of starting material is clearly necessary in these cases. Deoxygenation of p-nitrostyrene in the presence of triethyl phosphite has been reported to give a positive test for indole although it is clear that the yield is very Of more interest is the reported isolation of a low yield of phenyla~etonitriIe.~~ That this compound is also produced by photolysis of 30 (a) Ref. 23 refers to a persona1 communication from A.Weinstock; (6) J. H. Boyer W. E. Krueger and G. J. Mikol J. Amer. Chem. SOC. 1967 89 5504. 236 Cadogan l i J’ ti donor Scheme 19 /3-styrylazide has led Boyer and his co-workers tentatively to suggest that /3-styrylnitrene may be an intermediate in both reactions30b [equations (16) and (17).] PhCH-CHNO -+ PhCH=CH.N -+ PhCH,CN PhCH = CHN -t PhCH = CH+N -t PhCH,CN (Equation 16) (Equation 17) Formation of Indazoles Triazoles and Related Systems.-Following the sup- positionlg that an electrophilic nitrogen species such as a nitrene is an inter- mediate in the deoxygenation of 2-nitrobiaryls by triethyl phosphite it was shown early9*l9 that in the analogous reaction of 2-o-nitrophenylpyridine (40) ring closure occurred at the electron-rich ring-nitrogen atom rather than at carbon to give pyrid[l,2-b]indazole (41) in excellent yield.This opened the way to success- ful cyclisations of o-nitrobenzylideneanilines to 2-arylindazoles and of o-nitro- azoarenes to the corresponding benzot~-iazoles,~~~~ except in the case of 2-nitro- 4’-hydroxyazobenzene which undergoes ethylation31 as well as reductive cyclisa- tion to give 2-p-ethoxyphenyl-2H-benzotriazole (Scheme 20). Similarly o-nitro- benzaldehydeazine gives 2,2’-bi-2H-indazolyl (42) bis-[o-nitrobenzylidenel-p- phenylenediamine gives p-di-2H-indazol-2-ylbenzene (43) while the fused five- membered ring system dibenzo[b,fl-l,3a,4,6a,-tetra-azapentalene (44) is readily obtained from 2,2’-dini troazo benzene. 31 Cf. J. I. G. Cadogan J. Chem. SOC. 1957 1079. 237 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Ar=Ph,-P-C,H,*Me o-C,H,*Me p-C,H;OMe o-C,H,Br a-CIOH7; 35-83‘%1 Scheme 20 Extensions of these reactions which have been reported since then include (equation 18) the formation of pyrazolo[l,2-a]benzotriazole (46; 18 %) from o-nitrophenylpyrazole (49 32 and (equation 19) of the new benzotriazolo- naphthotriazine (48 ; 46 %) from the naphthotriazine (47).ss (45) (47) 4 N-N equation 19) \ ** Y.Y. Hung and B. M. Lynch J. Heterocyclic Chem. 1965 2,218. 8* H . Sieper Tetrahedron Letters 1967 1987. 238 Cadogan The reaction (equation 19) has been extended to include some C-methyl derivatives and to the formation of the isomeric triazolonaphthotriazine (49) from the corresponding triazine (equation 20),54 and of 13-oxobenzotriazolo [ 2,l-b]benzo[l,2-e]triazine (50) from 3,4-dihydro-4-oxo-l,2,3-benzotriazine (equation 21).= Similarly 2-(o-nitrophenyl)-2benzotriazole (51) 240-nitro- phenyl)-2H-triazole (52) and l-(o-nitropheny1)-lH-benzotriazole (53) are readily converted (61-88 %) into the mono- and dibenzo-tetra-azapentalenes (54) (55) and (56)88 (Scheme 21).Scheme 21 L=/ Scheme 22 Kauer and Carboni in their study of this reactionsg noted that it was strongly affected by the nature of the phosphorus compound; the reactions with rne more nucleophilic tributylphosphine were more rapid but the yields and product purity were higher with triethyl phosphite. The increased nucleophilicity of the former reagent led to a competing reaction at the carbonyl group in the case of the ester (57) to give a product tentatively formulated as (58) (Scheme 22).p4 H. Sieper and P. Tavs Annalen 1967,704 161. *s A. W. Murray and K. Vaughan Chern. Cornrn. 1967 1283. J. C. Rauer and R. A. Carboni J. Amer. Chem. Soc. 1967,89,2633. 239 Reduction of Nitro- and Nitroso-compounds by TervaIent Phosphorus Reagenrs Formation of Anthrani1s.-The reductive cyclisation of nitro-compounds involv- ing an electron-rich centre in the 5-position has also been extended3' to include the formation of anthranils from o-nitrophenyl ketones [equation (22)l. Thus 2-nitrobenzophenone gave 3-phenylanthranil (56 %) and 2-aminobenzophenone (19-5 %) the latter suggesting the intermediacy of a nitrene while 2'-nitro- chalcone and 5-chloro-2-nitroacetophenone similarly gave 3-styrylanthranil K (54 %) and 3-methyl-5-chloroanthranil (37 %).2-Nitroacetophenone on the other hand gave no cyclised product the only product isolated being diethyl N-(2-acetylphenyl)phosphoramidate (1 0 %),22 thus suggesting the intermediacy of the corresponding phosphorimidate and hence a nitrene (Scheme 23). Although Scheme 23 possible extension of this synthesis of anthranils has not received a great deal of attention there is some evidence that it is not of wide appli~ability.~~ Formation of Phenothiazines.-All the cyclisation reactions so far described have resulted in five-membered rings even in those cases where the possibility of six- membered ring formation also exist. In accord with this phenoxazine and di- hydrophenazine were not isolated from the reactions of 2-nitrodiphenyl ether and 2-nitrodiphenylamine with triethyl phosphite respecti~ely,~ although in these cases the tarry products should be further investigated.The corresponding reaction (23) of 2-nitrodiphenyl sulphide on the other hand gave a moderately good yield of phenothiazine (54%) together with a small amount of N-ethyl- phenothiazine formed by alkylation of the former and it appeared that satis- factory route to phenothiazines had been found.37 A more detailed examination of the reaction has recently revealed the opera- tion of a new molecular rearrangement in this reaction however. Thus while 4-methyl-2-nitrodiphenyl sulphide gives the expected 2-methylphenothiazine (36 %) and the corresponding N-ethyl derivative (25 %) the isomeric 4'-methyl-2- 37 J. I. G. Cadogan R. K. Mackie and M. J. Todd Chem. Comm. 1966,491. 38 Altaf-Ur-Rahman and A. J.Boulton Tetrahedron 1966 Suppl. 7 49. 240 Cadogan nit rodip hen yl sulp hide gives 3 -met h ylp heno t hiazine and its N-e t h y 1 derivative instead of the expected 2-methyl-derivatives (Scheme 24). Similarly 4’-t-butyl-2- Scheme 24 nitrodiphenyl sulphide gives 3- rather than 2-t-butylphenothiazine [equation (24)] while 4’-chloro-2-nitrodiphenyl sulphide gives 3- rather than 2-chloro- phenothiazine. These observations suggest that the six-membered ring is being formed after rearrangement of a five-membered intermediate formed by electro- philic attack at the electron-rich 1’-position (Scheme 25). It is probable although not yet established that the ring-closure involving cyclisation on to an unsubstituted ring also proceeds by such a route which is similar to that involved in the Hayashi rearrangement of carboxybenzophenones in strong acid39 (Scheme 26).It will be of interest to determine whether the reported thermally 0 ‘ 0 Scheme 26 39 Cf. R. B. Sandin R. Melby R. Crawford and D. McGreer J . Amcr. Chem. Soc. 1956 78 3817. 24 1 Reduction of Nitro- and Nitroso-compowtds by Tervalent Phosphorus Reagents induced cyclisation of 2-azidodiphenyl sulphide to phen~thiazine~~ also pro- ceeds via a similar rearrangement. Thus the formation of phenothiazines in the phosphite-induced deoxygenation conforms with other successful cyclisations in that five-membered ring forma- tion is preferred in the first instance. It is significant that the success of this cyclisation is reagent-dependent; thus reduction by the highly nucleophilic tributylphosphine and hexaethylphos- p hor ous t r iamide or with tripheny lp hosp hine or diet h ylmet h ylp hosphoni te gave tars only.In the last case dilution with hexadecane gave phenothiazine in 9% yield. Dilution thus appears to have a beneficial effect on the yield of cyclised products and in the cases of 3-t-butyl- 2-chloro- and 3-chloro-phenothiazines dilution increased the yields in each case (55 to 74; trace to 55; and 63 % respec- tively). The absence of bi-a-cumyl in these cases suggests the absence of triplet nitrene by comparison with the reaction of 2’-nitro-2,4,6-trimethylbiphenyl discussed above. It is possible that in the case of the phenothiazine a nitrene generated in a singlet state (see below) undergoes the cyclisation before d e activation to the triplet state occurs. Whereas in the case of 2’-nitro-2,4,6- trimethylbiphenyl where cyclisation is not the preferred reaction deactivation to the triplet state is competitive giving rise to some hydrogen abstraction from the solvent.Deoxygenatfon of 0-Nitroalkylbenzenes and of Nitrobenzene.-Sundberg has reported that treatment of 0-methyl- o-propyl- o-butyl- o-cyclohexyl-,18 and o-ethyl-nitrobenzene28 with excess of boiling triethyl phosphite [equation (25)] gives the corresponding triethyl N-alkylphosphorimidate (59) as the major identified product (ca. 35-50%). In addition minor amounts of products o-R.C,HdNO + (EtO),P -+ (EtO)sP-N.C,H,(o-R) (Equation 25) ascribed to abstraction and insertion reactions of intermediate nitrenes were detected. Thus o-propylnitrobenzene gave 2-methylindoline (7 %) o-propyl- aniline (6 %) and o-allylaniline (6 %); o-butylnitrobenzene gave 2-ethylindoline (9 “A 1,2,3,4-tetrahydr0-2-methylquinoline (2 %) and compounds believed to be isomeric butenylanilines (ca.5 %) ; while o-cyclohexylnitrobenzene gave a mixture of amines (16 %) containing cis- and trans-1,2,3,4,4a,9a-hexahydro- carbazole (Scheme 27). These results are considered to indicate the intermediacy of a nitrene in these deoxygenations a conclusion c o h e d by the recent work by Smolinsky and on the triethyl phosphite-induced reduction of optically active 2-nitro-(2’-methylbutyl)benzene (60; X = NO& in an extension of earlier work41 involving the pyrolysis of the corresponding 2-azido-derivative (60; X = N$. The latter reaction carried out in the vapour phase and in solu- B. B. Brown R. K. Putney R.F. Reinisch and P. A. S . Smith J. Amer. Chem. SOC. 1953 B. I. Feuer and G. Smolinsky J. Amer. Chem. Soc. 1964 86 3085. 75 6335. 242 Cadogan W o R > a K + mMe + rR (R=Me,Et) [ R = E t ) + unsaturated alkylanilines Scheme 21 tion gave active 2-ethyl-2-methylindoline (61) in each case thus suggesting direct insertion of a singlet nitrene into the G H bond at the 2-position of the side chain via a transition state such as (62). The alternative reaction involving triplet nitrene via radical abstraction to give the intermediate (63) followed by recombination would have led to extensive racemisation (Scheme 28). Scheme 28 It is noteworthy that a significantly greater degree of retention of configuration was observed in the vapour-phase reaction suggesting as would be expected that collisional deactivation of the singlet nitrene to the triplet occurred in the liquid phase followed by reaction via intermediate (63).However it appears that even if pyrolysis of an azide does give a singlet nitrene this will only react as such in the presence of a suitably constituted side chain since thermal de- composition of p-methoxyphenylazide in ~ u m e n e ~ ~ produces substantial yields of bi-a-cumyl a product which can only arise via radical intermediates. These observations are particularly relevant to the interpretation of some of the phosphite-induced deoxygenations of nitrobiaryls and nitrodiphenyl sulphides referred to above. In the case of the deoxygenation by triethyl phosphite of active 2-nitro-(2’- methylbuty1)benzene (60; X =NO& to which we return the product was partially active (ca.50 %) 2-ethyl-2-methylindoline (61) thus paralelling the results and hence the conclusions obtained from the pyrolysis of the corre- sponding azide in solution. Still further support for the concept of the inter- mediacy of a nitrene in these reactions comes from a recent reinvestigationa2b P. Walker and W. A. Waters J. Chem. Soc. 1962 1632. 243 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents of the deoxygenation by triethyl phosphite of o-methyl- and o-ethyl-nitrobenzene. It will be recalled that Sundberg1*s2a reported good yields of the triethyl N-alkyl- phosphorimidate in these reactions. Reinvestigation22 has confirmed that although these compounds are indeed formed but in a lower yield than pre- viously reported an additional product is present in each case.Thus 2-ethyl- nitrobenzene gives diethyl 2-ethyl-3-H-azepin-7-ylphosphonate (64; 21 %) and ethylene (ca. 50% based on the azepine). These most interesting products pre- sumably arise by a mechanism similar to that suggested above for the formation of diethylaminoazepines with the difference that triethyl phosphite acts as the nucleophile in this case to give a phosphonium intermediate (65) which then eliminates ethylene rather than undergo Arbusov-type isomerisation to give a C- or N-ethylated product (Scheme 29). The other products of the reaction were Scheme 29 triethyl N-(2-ethylphenyl)phosphorimidate (66) its hydrolysis product diethyl N-2-ethylpheny1)phosphoramidate (67) and diethyl N-ethyl-N-(2-ethylphenyl)- phosphoramidate (68).(EtO),P = NAr -+ (Et O),P(O)NH Ar (EtO),P(O)NEtAr (66) (67) (68) (Ar = o.EtC,H,) The reaction of o-nitrotoluene with triethyl phosphite followed a similar course as did that of nitrobenzene. In the latter case for example diethyl N-phenylphosphoramidate diethyl N-ethyl-N-phenylphosphoramidate diethyl 3H-azepin-7-ylphosphonate and ethylene were formed control experiments having established that triethyl N-phenylphosphorimidate is partially converted into the diethyl N-ethyl-N-phenylphosphoramidate but not into the corre- sponding azepin-7-ylphosphonate. This is in contrast to earlier failure to detect identifiable products in this r e a c t i ~ n . ~ ~ ~ ~ These observations coupled with the isolation of bi-a-cumyl (8%) from the reaction of o-nitroethylbenzene with triethyl phosphite in cumene,22b therefore provide a considerable body of circumstantial evidence in favour of the participa- tion of nitrenes in the reactions described in this section.Reactions of Bifunctional Aromatic Nitro-compounds.-Reactions of trialkyl phosphites with o-dinitrobenzene proceed mainly without deoxygenation to give 244 Cadogan dialkyl o-nitrophenylphosphonates (69 ; 75-85 %) and the alkyl nitrite?Sa Although a small quantity of triethyl phosphate is formed indicative of some deoxygenation as a competing side reaction the corresponding reduction products of the nitro-compound have not been isolated. The simplest route to the products of this reaction involves direct aromatic nucleophilic substitution by the phos- phite (Scheme 30) but a mechanism involving attack on one of the nitro-groups QTEt '6% Scheme 30 followed by rearrangement cannot be excluded at this stage.The reaction has been extended to include other tervalent phosphorus reagents such as diethyl methylphosphonite and ethyl diphenylphosphonite which give the correspond- ing o-nitrophenyl-phosphinate and -phosphine oxide.43 The reaction is of interest not only because it affords convenient routes to compounds containing o-nitrophenyl-phosphorus bonds but also because established examples of heterolytic aromatic substitution by phosphorus compounds are rare. It is also of interest in that it possibly sheds light on observations previously recorded but unexplained. Thus Horner and KliipfelM showed that o-dinitrobenzene and triethylphosphine gave a 1 :1-adduct of unknown structure while the reaction of triphenylphosphine with 4-nitropyridine 1 -oxide at 200" has been reported to lead to the evolution of nitrous fumes the fate of the remainder of the molecule being unkn0wn.4~ In view of the foregoing it is obviously possible these reactions proceed via displacement of a nitro-group from the aromatic ring.Triphenyl- phosphine and o-dinitrobenzene on the other hand are reported to give triphenylphosphine oxide after reaction in boiling benzene.8 The nitro-group displacement is most successful with o-dinitrobenzene. rn- and p-Dinitrobenzenes and o-chloronitrobenzene react with a variety of tervalent phosphorus compounds to give a complex mixture of products which although they have not been resolved yet do not include compounds such as diethyl 0- or p-nitrophenylphosphonate.It is also significant that Griffin and Obrycki have ob~erved"~ photolytic oxidation of triethyl phosphite by o-halogenonitrobenzenes but identification of the transformation products of the latter was not possible. In these cases the many possible competing side reactions e.g. formation of azepine amines. phosphorimidates and phosphoramidates are probably favoured. 4-Methyl-2'-nitrophenyl phenyl sulphoxine on the other hand gives a low yield (ca. 5 %) of diethyl o-nitrophenylphosphonate in addition to the pheno- 43 (a) J. I. G. Cadogan D. J. Sears and D. M. Smith Chem. Comm. 1966,491 ; (b) Unpub- lished results. 44 L. Horner and K. Kliipfel Annalen 1955 591 69. 45 E. Howard and W. F. Olszewski J. Amer. Chem. Soc. 1959 81 1483. 46 C. E. Griffin and K. Obrycki J. Org.Chem. 1968,33 632. 245 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents thiazine (5 %) on reaction with triethyl phosphite (Scheme 31).43b 1,2,4-Trinitro- benzene as might be expected gives a good yield of diethyl 2,4-dinitrophenyl- phosphonate on reaction with triethyl phosphite and the same compound is produced in lower yield from the corresponding reaction of 2,4dinitrochloro- benzene.& b. 0 Scheme 31 An interesting displacement closely related to those described above has been reported recently by Sieper.*' Treatment of 2H-2-(2-nitrophenyl)-naphtho [ 1,8-de]-1,2,3-triazine (70) with triethyl phosphite gave diethyl o-nitrophenyl- phosphonate and the ethylated triazines (71) and (72) in addition to the expected te t r a-azapent alene (7 3) (Scheme 3 2).Diet h y 1 o-ni tr op hen y Ip hos p honate was also produced as a side product in the reaction of triethyl phosphite with 3,4-dihydro-4-oxo- 1,2,3-ben~otriazine.~~ + + Scheme 32 Reduction of Aliphatic Nitro- and Nitroso-compounds Fewer reactions of aliphatic than of aromatic nitro- and nitroso-compounds with tervalent organophosphorus reagents have been reported. Simple nitro- alkanes** do not react at low temperatures and reactions at higher temperatures have not been investigated. Reactions involving the more reactive halogeno- nitroalkanes have been reported however although in some cases the low accountances of identified products lead to uncertainty over the participation of O7 H. Sieper Tetrahedron Letters 1967 1987. 48 S. Trippett B. J. Walker and H. Hoffmann J. Chem.SOC. 1965 7140. 246 Cadogan the nitro-gr0up.4~ Allen50 has shown that 2-chloro-2-nitropropane reacts with triethyl phosphite to give diethyl isopropylideneaminophosphate (74) triethyl phosphate and ethyl chloride. It is not known whether reduction to the nitroso- compound fist occurs followed by reaction of this with another molecule of phosphite it having been shown that 2-chloronitrosopropane and triethyl phosphite also give the same product (74) but much more easily or whether deoxygenation of an intermediate is involved (Scheme 33). Alternative routes involving nucleophilic attack on halogen to give an intermediate (RO),PCl+ Me,CNO,- cannot be discounted. A 0 79 ?- (EtO)3P*b0 =&-CMez -+ (EtO),P(O)Or;l= CMe + EtCl (EtOl,P 7) 1 (EtOi,i’\ O=&-CMc -+ (EtO),P(O)ON =CMez (74) Scheme 33 In an extension of these reactions it has been shown that gem-halogeno- nitr~so-~l and -nitr~-~~cycloalkanes give rise to intermediates which undergo the Beckrnann rearrangement.Thus at room temperature the nitroso-compound reacts to give a good yield of the corresponding lactam presumably as outlined in Scheme 34. The related nitro-compounds react similarly but at higher tem- peratures and again there is no evidence for or against the participation of the nitroso-compound (Scheme 35). / I = 1,2,3,4 8 + Ph,PO Scheme 34 Scheme 35 Certain 1 -bromo-1 -nitro-alkanes including 1 -bromo-1 -nitro-octane l-bromo- 1-nitropropane bromonitrophenylmethane and ethyl bromonitroacetate react $@ A E. Arbusov €3. A. Arbusov and B. P. Lugovkin Bull. Acad. Sci. U.R.S.S. CIasse xi.chim. 1947 538; G. Kamai Doklady Akad. Nauk S.S.S.R. 1951 79 795. so J. F. Allen J. Amer. Chem. SOC. 1957 79 3071. 61 M. Ohno and I. Sakai Tetrahedron Letters 1965,4541. 6a M. Ohno and N. Kawabe Tetrahedron Letters 1966 3935. 247 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents with triphenylphosphine to give the phosphine oxide and the corresponding nitrile possibly via attack on the aci-form of the nitro-compound (Scheme 36),53 but it is more likely that the reaction proceeds via initial nucleophilic attack on bromine as suggested by Speziale and Smith54 (Scheme 37). No evidence for the Scheme 36 P- P hj P-L Br -Gk- NO __I) P h,PB r+ R C H = ?- O- 4 0- + - + RCN RC 2 N- 0 t- Ph3P70-N= C - R ?-I H Scheme 37 intermediacy of the postulated nitrile oxide was found which is not surprising in view of the now known reactivity of triphenylphosphine towards nitrile oxides.55 Bromonitromethane and 1-bromo-1-nitroethane on the other hand give the corresponding or-hydroxyiminoalkylphosphonium bromides [equation (26)] and the change in the course of the reaction following increase in size of + 2Ph3P + RCHBraNO -+ Ph3P0 + Ph,PCR=NOH Br- (Equation 26) the alkyl group is clearly a point of interest.The possibility that both processes proceed via reaction (equation 26) as a first step followed in the case of R = Me and Et by addition (equation 27) of triphenylphosphine to the nitrile oxide - ..a + - + PhSP RCE N- 0 + Ph,PCR=N - 0 (Equation 27) probably can be discounted because of the relative ease with which triphenyl- phosphine reduces nitrile oxides although quantitative data would be necessary completely to rule out this alternati~e.~~ Smolinsky and F e ~ e r ~ have isolated phenylacetonitrile from the phosphite reduction of 1 -nitro-2-phenylethane thus recalling the corresponding reduction of P-nitrostyrene referred to above.That phenylacetaldoxime also gives the same product suggests in this case also that prior reduction of the mi-form of the nitro-compound occurs [reaction (28)]. PhCH,CH,NO + PhCH,*CH=N(O)OH -+ PhCH,*CH=NOH -+ PhCH,C_N (Equation 28) 53 S. Trippett and D. M. Walker J . Chem. SOC. 1960 2976. 54 A. J. Speziale and L. R. Smith J. Amer. Chem. Soc. 1962 84 1868. 55P. Griinanger Atti Accad. naz. Lincei 1964 36 387 (Chem. Abs. 1965 62 3973); C. Grundman and H. D. Frommeld J. Org. Chem. 1965 30 2077.248 Cadogan Appendix added in Proof It is known5s that reaction of 5-dimethylaminobenzofuroxan with nitrous acid gives 4-dimethylamino-7-nitrobenzofurazan (76). From this fact the formation of an intermediate nitroso-compound (75) was inferred this compound then being assumed to undergo rapid rearrangement to the furazan (76) (Scheme 38). I 0- Extension of this idea has led to evidence which tends to support the transient intermediacy of a nitroso-compound in the phosphite reduction of 3-methyl-7- nitroanthranil (77). 57 In this case the major product is 4-acetylbenzofuroxan (78) (Scheme 39). The analogy with Scheme 38 is obvious. In the absenceof Scheme 39 instances of actual isolation of an intermediate nitroso-compound in phosphite reductions of nitro-compounds this together with the observed formation of phenylacetonitrile in the reduction of 1 -nitr0-2-phenylethane~~ mentioned above is the best evidence so far produced for the participation of nitroso- intermediates in the phosphite reduction.Photochemically induced phosphite reduction of aromatic nitro-compounds (79) at room temperature has now been reported.58 In general yields of products are low and it is also stated that cyclisation of 2-nitrobiphenyl and 2-nitro- stilbene under these conditions occurs in lower but unstated yield compared with the thermal cyclisation. In all the cases (79) studied triethyl N-arylphos- phorimidates (80) are formed in moderate yields. o-Methyl nitro-compounds also give the corresponding N-aryl-2-acetimidylpyridines (8 1). The results of the reactions are summarised in Scheme 40 and Table 3.It will be recalled that products of this type (81) had been isolated previously from the reaction 68 A. J. Boulton P. B. Ghosh and A. R. Katritzky J. Chem. SOC. (B) 1966 1004. 68 R. J. Sundberg W. G . Adams R. H. Smith and D. E. Blackburn Tetrahedron Letters 1968 777. A. J. Boulton I. J. Fletcher and A. R. Katritzky Chem. Comm. 1968 62. 249 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents of nitroso-compounds with triethyl phosphit&* and that a not very satisfactory rationalisation of the mode of formation had been advanced (Scheme 3). The new and very significant experimental data presented by Sundberg and his co-workers68 now clearly indicates that the skelatal rearrangement to give the pyridine derivative is more complex than previoiisly suspected in that it involves the formation of a C-1 to C-3 link (Scheme 40).On the basis of these results Scheme 3 has been withdrawn and a satisfactory rationalisation is now awaited. Table 3 Products of photochemical phosphite reduction of (79) (Scheme 40) (79) Yield (%) H H H H 1 0 H H Me H <3 0 M e H H H 13 37 Me Me H H 12 18 Me H Me H 51 10 Me H H Me 14 4 H H Me0 H 34 0 R1 R2 R3 R4 (80) (81) Scheme 40 9 (EtO),P 3. ArNO,-+ArNO-+Ar~~O. (OEt) or ArN-P (OEt) azepines 5 -membered ' 0 heterocycles ArN P (OEt) ArN(O):NAr ArN -7 (OE t)3 or ArN .O - h (OEt) e tc. @ + (EtO),PO Scheme 41 250 Cadogan Such a rationalisation should embrace all of the various types of products which have been obtained from phosphite deoxygenations of nitro-compounds. A tentative interim suggestion for consideration while we await more experimental results is outlined in Scheme 41 which includes the new intermediate (83) which could in theory be stabilised when the nitro compound contains an o-methyl group and hence play a more significant part in the reaction. Scheme 42 An extension of the synthesis of indoles by deoxygenation of o-nitrostyrenes has been reported59 and is summarised in Scheme 42. These are the first instances of migration of a /3-substituent in a monosubstituted styrene earlier work having shown that one of the substituents in /Ip-disubstituted o-nitrostyrenes was prone to migration during deoxygenation. ~4 R. J. Sundberg J. Org. Chem. 1968,33 487. 251

ISSN:0009-2681    

DOI:10.1039/QR9682200222

出版商:RSC    

年代:1968

数据来源: RSC