摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Dr. A. R. Butler Professor E. C. Constable Dr. T. C. Gallagher Professor D. M. P. Mingos FRS Professor J. F. Stoddart FRS Consulting Editors Dr. G. G. Baht-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor B. T. Golding Professor M. Green Professor A. Hamnett Dr. T. M. Herrington Professor R. Hillman Professor R. Keese Dr. T. H. Lilley Dr. H. Maskill Professor A. de Meijere Professor J. N. Miller Professor S. M. Roberts Professor B. H. Robinson Professor M. R. Smyth Dr. A. J. Stace Staff Editor Mr. K. J. Wilkinson University of Sussex University of Leicester University of St.Andrews University of Basel, Switzerland University of Bristol Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich, Switzerland University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Bath University of Newcastle upon Tyne University of Reading University of Leicester University of Bern, Switzerland University of Sheffield University of Newcastle upon Tyne University of Gottingen, Germany Loughborough University of Technology U n iversity of Exeter University of East Anglia Dublin City University, Republic of Ireland University of Sussex Royal Society of Chemistry, Cambridge It is intended that Chemical Society Reviews will have the broad appeal necessary for researchers to benefit from an awareness of advances in areas outside their own specialities.Deliberate efforts will be made to solicit authors and articles from Europe which present a truly international outlook on the major advances in a wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative ovbrviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca. 30 references), but should act as a springboard to further reading.In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context. Review articles must be short, around 8-1 0 journal pages in extent. In consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Information to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis (including a selection of the literature references that will be cited in the review and a brief academic CV of the author), rather than the completed article, should be submitted to the Senior Editor (Reviews), Books and Reviews Department, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 4WF. @ The Royal Society of Chemistry, 1994 All Rights Reserved No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic or mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Typeset by Servis Filmsetting Ltd. Printed in Great Britain by B lackbear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99423FX021

出版商:RSC    

年代:1994

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99423BX023

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

ISSN 0306-001 2 CSRVBR 23(6) 363-442 (1 994) Chemical Society Reviews Volume 23 Issue 6 Pages 363-442 December 1994 Benzotriazole-mediated Arylalkylation and Heteroarylalkylation By Alan R. Katritzky and Xiangfu Lan (pp. 363-373) An electron donor can activate a benzotriazole group through a conjugated system of type D-(CH =CH)-CH,Bt. If the conjugated system is a benzene or heterocycle this enables arylalkylation and heteroarylation reactions in which Bt is replaced by alkyl or aryl groups from Grignard reagents, by hydrogen from LiAIH,, and by other nucleophiles including electron-rich aromatic systems, alcohols, thiols, etc. Such reactions enable specific para-substitution of anilines, ortho-substitution of phenols, and elaboration of phenol ethers and heterocycles.Their scope is increased by the possibility of deprotonation of the original system at the CH, and subsequent reaction with electrophiles before the Bt displacement. The Dynamics of Photodissociation By Richard N. Dixon (pp. 375-385) The use of lasers, both to inititiate photochemical decomposition and to probe the primary products, has greatly advanced our knowledge of the dynamics of photodissociation. This article uses case histories for excitation of H,O, H,O,, HONO, HCO, NH,, HNO, and CH, to illustrate the diversity of mechanisms. These cover dissociation through motion on a single potential energy surface; dissociation mediated by surface crossings; and dissociation following internal conversion to the molecular ground state.Implications for larger molecules and future work are also discussed. Molecular Mechanics Force Fields for Cyclopentadienyl Complexes By Brice Bosnich (pp. 387-395) A self-consistent molecular mechanics force field is developed for the reproduction of structures and energy differences of bis-cyclopentadienyl (Cp) complexes. The method is used to identify the forces which control Cp ring rotations, the conformations of substituted Cp complexes, and provides an explanation for the existence of bent metallocenes of the alkaline earth and lanthanide metals and possibly for the bent structures of the divalent silicon group. The method provides an accurate method of reproducing structures and can incorporate the structural effects caused by crystal packing forces. HAWORTH MEMORIAL LECTURE.Experiments Directed Towards Glycoconjugate Synthesis By Tornoya Ogawa (pp. 397-407) Total synthesis of complex carbohydrates and glycoconjugates are described. Cycloglycosylation was carried out by use of either anomeric fluorides or thioglycosides to obtain cyclodextrins and their analogues. Other targets for synthetic studies have been selected mainly from bio-funtional glycoconjugates that include gangliosides, Nod-factors, N- and 0-linked glycans of glycoproteins, plant cell wall oligosaccharide acting as endogenous plant hormones, GPI anchor (glycosyl phosphatidylinositol anchor), and functional domains of glycosaminoglycan. Key synthetic transformations in the experiments were highlighted in terms of efficiency and stereoselectivity.Pericyclic Key Reactions in Biological Systems and Biomimetic Syntheses By Ulf Pindur and Gunter H. Schneider (pp. 409-41 5) A variety of pericyclic key reactions in biological systems are explored and lead stereospecifically to many natural products. Some examples of pericyclic reactions taking place in biological systems are discussed and classified on the basis of reaction mechanism type. Some biomimetic syntheses are also included that complete the biosynthetic steps. There is evidence that some of the biological pericyclic reactions leading to enantiomerically pure compounds are catalysed by enzymes. Surfactant Systems: Their Use in Drug Delivery By M. Jayne Lawrence (pp. 41 7-424) Amphiphilic molecules frequently aggregate in solution to form a rich variety of phase structures.This review explains why there is considerable pharmaceutical interest in these structures and discusses the work performed to date examining their use as vehicles for drug delivery. Biological Activity, Reactivity, and Use of Chromotropic Acid and its Derivatives By Jan Duda (pp. 425-428) Complex-forming properties of chromotropic acid and its halogen derivatives in relation to metal ions are reviewed. Various reactions of these compounds, including oxidation reactions, and their applications are considered. The biological activities of chromotropic acid and its derivatives are also discussed, including their activity in the inhibition of HIV. Mechanistic and Structural Investigations Based on the lsokinetic Relationship By Wolfgang Linert (pp.429-438) The rapid development of the Isokinetic Relationship IKR, especially concerning its applications to actual chemical reaction series, is reviewed. It is proved on the basis of several examples that the IKR is extremely sensitive to changes in reaction pathways, and that this can be used as a powerful tool in distinguishing changes within a given series of reactions, provided temperature-dependent data are available. It is hoped that the examples presented in this Review will encourage others to use the effect for detailed investigation of their own reaction series, especially as a simplified, reliable statistical analysis is now available. 1994 Indexes (pp. 439-442) Articles that will appear in forthcoming issues include Structure-Property Relationships in Superconducting Cuprates C.N.R.Rao and A.K. Ganguli Stereochemical, Mechanistic, and Structural Features of Enzyme-catalysed Phosphate Monoester Hydrolyses D. Gani and J. Wilkie CENTENARY LECTURE. Bridgehead Unsaturation in Compounds of Nature: A Proper Forum for Unleashing the Potential of Organic Synthesis L.A. Paquette Conservation of Waterlogged Wood B. Kaye Synthesis of Diarylketones through Carbonylative Coupling J.-J. Brunet and R. Chauvin Transannular Interactions in Difunctional Medium Rings -Modelling Bimolecular Reactions P. Rademacher Potential Surfaces and Dynamics of Weakly Bound Trimers: Perspectives from High Resolution IR Spectroscopy M.A.Suhm and D.J. Nesbitt Chemical Society Reviews (ISSN 03064012) is published bi-monthly by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge, CB4 4WF, England. All orders accompanied with payment should be sent directly to The Royal Society of Chemistry, Turpin Distribution Services Ltd., Blackhorse Road, Letchworth, Herts., SG6 1HN, U.K. NB Turpin Distribution Services Ltd., distributors, is wholly owned by The Royal Society of Chemistry. 1994 annual subscription rate E.C. 299.00, U.S.A. $186.00, Canada Ell l.OO+ GST, Rest of World E106.00.Customers should make payments by cheque in sterling payable on a U.K. clearing bank or in U.S. dollars payable on a U.S. clearing bank. Second class postage is paid at Jamaica, N.Y. 11431. Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. All other despatches outside the U.K. by Bulk Airmail within Europe and Accelerated Surface Post outside Europe. PRINTED IN THE U.K. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E30.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.

ISSN:0306-0012    

DOI:10.1039/CS99423FP039

出版商:RSC    

年代:1994

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99423BP041

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Benzotriazole-mediated Arylalkylation and Heteroarylalkylation Alan R. Katritzky Center for Heterocyclic Compounds Department of Chemistry University of Florida Gainesville FL 326 I 1 -7200 U.S.A. Xiangfu Lan Sandoz Chemicals Corporation Mallard Creek Research Center P.0. Box 669304 Charlotte NC 28266 U.S.A. 1 Introduction A recent review discussed reactions important for the prep- aration of useful benzotriazolyl intermediates and the displace- ments by nucleophiles of the benzotriazole groups in such derivatives of (1). * We rationalized that these displacements of benzotriazole were assisted by the lone electron pair on the heteroatom substituents Y. Thus initial ionization of (1) occurred to give the reactive intermediate iminiums (2) which then reacted with nucleophiles to give the final products (3) (Scheme 1). Scheme 1 In reactions of this type the benzotriazole group and the heteroatom are connected to the same carbon. In recent years it has been shown in our group that such assistance by an electron pair could also be effected through a conjugated system as shown in compounds of type (4). The displacement of benzotria-zole by nucleophiles is thus realized through conjugation as shown in Scheme 2. was covered in a previous review.2 We now describe displace- ments by nucleophiles such as Grignard reagents electron-rich aromatic and heteroaromatic compounds active CH acids alcohols thiols etc. of benzotriazole groups activated by conju- gation through benzene rings to give substituted aromatic compounds. Furthermore the electron-withdrawing ability of benzotriazole further activates the directly attached benzylic carbon and renders the alpha protons acidic. Such compounds can thus be lithiated and substituents introduced by subsequent reaction with electrophiles. The displacement of the benzotria- zole group in the resulting derivatives will afford trisubstituted methanes. The general scheme is represented in Scheme 3 where E = H corresponds to the parent benzotriazole derivatives E # H to derivatives obtained via lithiation. An electron pair on the heteroatom in substituent Y can render assistance from both the ortho and the para positions as shown in Scheme 3. The heteroatom substituent Y can be an amino or substituted amino group as in the case of anilines the hydroxy group of a phenol or the alkoxy group of a phenol ether. Scheme 2 In the present overview we discuss such conjugated systems specifically those involving benzene rings. Some work on the similar system with conjugation through a simple vinyl group Xiangfu Lan born in 1965 in Jiangxi China received his B.En. at the Beijing Institute of Technology and his Ph.D. in 1991 at the University of Flor-ida under the supervision of Professor A. R. Katritzky. He is now a R&D Chemist with Sandoz Chemicals Corpor-ation at Charlotte North Carolina U.S.A. 363 Y = R2N OH OR Scheme 3 Alan Katritzky (b. 1928 London) was educated at Oxford (D. Phil. Robinson) where he carried out independent research from 1953. He moved to Cambridge in I958 (Lecturer and Fellow of Churchill) then to East Anglia to found the School of Chemical Sciences and finally to Florida in 1980 where he is Kenan Professor and Director of the Center for Heterocyclic Compounds. A light-hearted account of his lije is published in J. Het. Chem. 1994,31,pp. 569-602 and an overview of his scientijic work in Heterocyles 1994 37 pp. 3-130. 364 Similar reactions with heterocycles will be described in which the assistance comes from a free electron pair of the heteroatom incorporated in a heterocycle such as furan thiophene and indole as shown in Scheme 4 -+2) E+ Bt (19)(20) R =H R = Me ~ (21) R E' E Yield (%) a H PhCH,Br PhCH 88 b H 4-MeC,H4CH0 4-MeC,H4CH(OH) 71 c Me PhCH,Br PhCH 81 d Me 4-MeC,H4 4-MeC,H4CH(OH) 53 (13) x=o s Nu LR2 Scheme 4 2 Elaboration of Benzotriazole Derivatives via Lithiation Heteroatom assisted lithiations have attracted considerable attention in organic synthesis 1-and 2-Benzylbenzotriazoles have been shown to undergo lithiation at the benzylic carbon atom4 and similar reactions were demonstrated in reaction sequences leading to the preparation of aromatic ketones 4-(Benzotriazol-1-ylmethyl)anilines likewise react with BunLi and quenching the anions with a variety of electrophiles affords the expected substituted products in good yields Various electro- philes including alkyl halides aldehydes and ketones etc ,were employed The lithiations of N,N-dialkylaniline derivatives were covered in a previous review We have recently successfully extended such lithiations to (19) and (20) where the amino group is an NH or an NHMe by using two equivalents of BunLi6 (Table 1) 1) 2 eq n-BuLi Table 1 Lithiation of 4-(benzotriazol-1-ylmethyl)aniline ( 19) and 4- (benzotriazol-1-ylmethyl)-N-methylaniline(20) Lithiation also occurred smoothly in 1 -[(methoxyphenyl)-methyl]benzotriazole (22) good to excellent yields were obtained with a variety of electrophiles This lithiation sequence also succeeds for the naphthalene system (24) (Tables 2 3) The phenolic hydroxy group required a modification to the lithiation procedure due to two factors the initially formed phenoxide possessed less acidic methylene protons and the system formed a less soluble dilithiated product 8-1 Treatment with BunLi or Bu'Li in THF of o-(benzotriazol- 1-ylmethy1)phe- nols of type (26) gave satisfactory results only when the phenolic hydroxy group was protected by trimethylsilylation A one-pot process was developed by treatment of the substrate (26) with one equivalent of BunLi followed by one equivalent of tri- methylsilyl chloride to give (27) Further addition of one equiva- lent of BunLi and subsequent treatment with one equivalent of the electrophile gave (28) The protecting group was easily removed by stirring (28) in an acidic ethanolic solution Good CHEMICAL SOCIETY REVIEWS 1994 1) n-BuLi THF -78°C 2 h (23) R E+ E Yield (%) a 4-Me0 Me1 Me 80 b 4 Me0 PhCH2Br PhCH 85 C 4-Me0 Ph,C=O Ph,C( OH) 80 d 4-Me0 CO CO,H 78 e 2-Me0-3-Me Me1 Me 90 f 2-Me0-3 Me PhCH,Br PhCH 70 9 2-Me0-3-Me 4-MeCGH,CH0 4-MeC,H4CH(OH) 87 h 2-Me0-3-Me (CH,)SC=O (CH2)5C(0H) 80 I 2-Me0 3-Me Ph,C=O P h,C( OH) 75 I 2-Me0 3-Me PhC0,Et PhC=O 76 k 2 4 6-(MeO) Me1 Me 85 I 2 4 6-(MeO) (CH,)SC=O (CH,),C(OH) 72 Table 2 Lithiation of 1-[(methoxyphenyl)methyl]benzotriazole(22) 1)n-BuLi THF-OMe OMe a Me1 Me 98 b 4-MeC,H4CH0 4-MeC6H,CH(OH) 60 Table 3 Lithiation of I-(benzotriazol-1-ylmethyl)-2-methoxynaphtha-lene (24) overall yields of the desired products (29) were obtained A variety of electrophiles including alkyl halides aldehydes ketones and carbon dioxide were employed (Table 4) 1-Methyl-3-(benzotriazol-1-ylmethyl)indole (30) underwent smooth lithiation at the methylene carbon and the anion reacted with methyl iodide benzophenone phenyl isocyanate and diphenyl disulfide to give the desired derivatives (31) in good yields (Table 5) l2 Similarly we found recently' that 2-(benzotriazol- 1-ylme-thy1)indole (32) reacted with two equivalents of n-butyllithium to give a dianion (33) The dianion was quenched with one equivalent of an alkyl halide to give alkylated indole (35) and with three equivalents of methyl iodide to afford the dialkylated compound (34) in both cases in excellent yields (Table 6) 3 Displacement of Benzotriazole by Grignard Reagents and Hydride We have found that treatment with RMgBr or LiAIH of the parent products Y-C,H,-CH,-Bt as well as the Y-C,H,-CHR- Bt derivatives obtained via lithiation gave the expected alkyl aromatic compounds Y-C,H4-CHR-R' 3.1 Preparation of 4-Alkylanilines 4-(Benzotriazol-1-ylmethyl)anilines and their derivatives (36) obtained via lithiation reacted with an excess of Grignard reagents in refluxing benzene or toluene to give the desired 4-alkylanilines (38)6 (Table 7) Presumably cations (37) are the reactive intermediates A hydroxy functional group was readily BENZOTRIAZOLE-MEDIATED ARYLALKYLATIONS-A R KATRITZKY AND XIANGFU LAN oTMSBt 1) n-BuLi ?TMS ~~ ~ (29) R E+ E Yield (Yo) a H PhCHO PhCH(0H) 68 b Me Me1 Me 71 c Me nBul n Bu 50 d Me CO CO,H 71 e Me Ph,CO Ph,C(OH) 62 Table 4 Lithiation of o-(benzotriazol- 1-ylmethy1)phenols (26) E I Me (31) (31) E+ E Yield (%) a Met Me 83 b Ph,C=O Ph,C(OH) 95 C PhN=C=O PhNHC=O 90 d PhSSPh PhS 58 Table 5 Lithiation of I-methyl-3-(benzotrtazol- I-ylmethy1)indole (30) 2 eq n BuLflHF0-i.3OTLBt -N or 78OC 5h mBt I I H LI LI+ Me Me (32) (34) 1 eq RX (35) Fix R Yield (%) A R I c PhCH,Br PhCH 81 (35) Table 6 Lithiation of 2-(benzotriazol-1-ylmethyl)mdole (32) introduced in the alkyl substituent as shown by the examples of (38 1,m) Noteworthy is the stability of the initially formed alkoxide under such vigorous conditions In such a manner an R3CHE group is introduced at the para position of the anilines where E evolves from the electrophile and R3 from the Grignard reagent Good to excellent yields of (38) were usually obtained Such 4-alkylanilines are generally not easily available by other methods For example 4-benzyl- and 4-pentyl-N,N-dimethyla- niline were previously prepared by acidic reduction14 from the appropriate ketone and alcohol but such methods obviously suffer from the unavailability of the starting materials Classical Friedel-Crafts reactions are not generally applicable to the preparation of 4-alkylanilines owing to the deactivation effect of nitrogen on the Lewis acid catalysts Thus our method offers considerable advantages of easily available starting materials high yields and generality Also a hydroxy functional group could easily be introduced Under similar reaction conditions the benzotriazole group in derivatives of type (36) was replaced by hydride by the action of LiAlH,6 or sodium in piperidine l6 3.2 Preparation of ortho-Alkylsubstituted Phenols Just as for the (4-benzotriazolylalkyl)anilinesdescribed above o-(benzotriazolylalky1)phenols (39) reacted with Grignard rea- gents or LiAlH to give ortho-alkylsubstituted phenols (41)' (Table 8) Similarly the naphthol derivatives (42) under such conditions afforded the corresponding 1-alkyl-2-naphthols (44) (Table 9) The net effect of these transformations is the replace- ment of an ortho-ring hydrogen by an RECH group where E evolves from the electrophile (in the case of phenol but alternati- vely from the aldehyde in the case of naphthol) and R from the Grignard reagent or from LiAIH In this way normal as well as branched chain alkyl groups are easily introduced in moderate to excellent yields into the position ortho to a phenolic OH-group In support of our proposal' that these reactions involve the o-quinone methides (40) and (43) as intermediates such hetero- dienes were successfully trapped by the dienophiles ethyl vinyl ether and 1-vinyl-2-pyrrolidinone to give chroman derivatives (47) and (49) in excellent yields1* (Tables 10 1 1) CHEMICAL SOCIETY REVIEWS 1994 (38) R’ R2 E Reagent R3 Solvent Time (h) Yield (“A) a H H H PhMgBr Ph benzene 48 48 b H H H n-BuMgBr n-Bu benzene 48 50 c H MeH PhMgBr Ph toluene 12 36 d H MeH n-BuMgBr n-Bu benzene 24 30 e H H PhCH PhMgBr Ph benzene 17 68 f Me Me H PhMgBr Ph benzene 24 97 g Et Et H PhMgBr Ph benzene 24 92 h Me Me H n-BuMgBr n-Bu benzene 25 52 i Et Et H n-BuMgBr n-Bu benzene 17 78 j Me Me Me PhMgBr Ph benzene 10 82 k Me Me PhCH PhMgBr Ph benzene 10 91 I Me Me (CH,),C(OH) PhMgBr Ph toluene 18 81 m Me Me PhCH(0H) PhMgBr Ph toluene 18 69 n Me Me PhCH LiAIH H benzene 4 78 o Et Et Me LiAIH H toluene 13 77 p Me Me Et Na H piperidine 24 45 Table 7 Displacement of benzotriazole in 4-(benzotriazol- 1-ylalkyl)anilines (36) by Grignard reagents or LiAIH or Na/piperidine 150 OC tube) &osealed (41) E R Reagent Solvent Time (h) Yield (yo) ~~ a H Ph PhMgBr toluene 72 45 (47) R’ R2 Time (h) Yield (“A) b H n-Bu n-BuMgBr toluene 24 29 ~~ a ~ H OEt 36 95 c H H LiAIH toluene 48 50 b Ph OEt 5 92 d Me Ph PhMgBr THF 17 50 c Ph PYrr 3 82 e Me PhCH PhCH,MgBr THF 12 80 d 4-Me2NC,H OEt 5 92 f Me H LIAIH THF 48 66 e 4-Me2NC,H pyrr 3 87 g n-Bu H LIAIH TH F 48 62 Table 10 Preparation of chroman derivatives (47) Table 8 Displacement of benzotriazole in o-(benzotriazol- 1-ylalkyl)-phenols (39) by Grignard reagents or LiAlH R’I R’ (44) E Reagent R Time (h) Yield (YO) (49) R’ R2 Yield (Yo) a Ph PhMgBr Ph 22 66 a Me H 92 b t-BU H 91 b Ph PhCH,MgBr PhCH 3 86 c t-Bu Me 93 c Ph LiAIH H 22 90 d 4-Me2NC,H PhMgBr Ph 12 68 Table 11 Preparation of chroman derivatives (49) 0 4-Me2NC,H PhCH,MgBr PhCH 2 90 1 4-Me2NC,H LiAIH H 3 94 Taw 9 Displacement of benzotriazole in 1-(a-benzotriazol-1-ylalkyl)-2-naphthols (42) by Grignard reagents or LiAIH BENZOTRIAZOLE-MEDIATED ARYLALKYLATIONS-A R KATRITZKY AND XIANGFU LAN 367 3.3 Preparation of Alkyl-substituted Aryl Ethers dyes in the production of recording materials as antioxidants The benzotriazole group in 1-[(methoxyphenyl)alkyl]benzotria-for lubricating oils as curing agents of epoxy resins and as zoles (50) was displaced by Grignard reagents or organozinc electrically insulating composite materials reagents to give the alkyl-substituted aryl ethers (51) Sodium in Methylenebisanilines have been prepared by the reaction of piperidine also caused reduction of the benzotriazole group' an arylamine with formaldehyde in the presence of concentrated (Table 12) 1-(1-Benzotriazol-1-ylethyl)-2-methoxynaphthalene hydrochloric 25 and by the reaction of N-(alkoxyme- (52) reacted with PhMgBr to give compound (53) in 69% yield thy1)arylamines under acidic conditions 26 These two methods (Scheme 5) work only for symmetrical analogues Organomercury(I1) com- pounds were reported as intermediates for the preparation of R2 R2 both symmetrical and unsymmetrical analogues * organo-1 I mercurials however are toxic and difficult to use industrially 4- (Hydroxymethy1)-N N-dialkylanilines can provide unsymmetri- cal methylenebi~anilines,~~ but are reportedly rather un- stable No other general methods are available However we have found that the stable and easily accessible 4-(benzotriazol- Condition 1-ylmethy1)anilines(58) react with anilines bearing an NH an (51) R' R2 Reagent Solvent (oc h) Yield (%) NHR or an NR group with or without other ring substituents ~~ a 2 Me0 3 Me Me PhZnBr toluene reflux (72) 75 to give both symmetrical and unsymmetrical methylenebis- b 246 (MeO) H PhMgBr toluene reflux (8) 56 anilines (59) in excellent yields3 (Table 13) c 34(Me0) Et Na piperidrne 100 (24) 54 R4 Table 12 Preparation of alkyl substituted aryl ethers (45) I u PhMgBr toluene . MeYPh &OM. reflux 24 h LBt reflux 72 h '2' -2) 1 M KOH 03"""69% \/ (58) (59) (52) (53) (59) R1 R2 R3 R4 Condition Yield ("lo) a H H H H A 80 Scheme 5 b H H H 2Me A 78 c H H H 2CI A 72 3.4 Preparation of Substituted Indoles d e H H H M e Me Me H H A A 66 70 The benzotriazole moiety in compound (54) was displaced with ethyl magnesium bromide to give compound (55) in 47% yield f g MeH Me Me H Me H H A B 83 97 (Scheme 6) l2 Similarly compound (56) reacted with methyl magnesium iodide in toluene to afford compound (57) in 90% yield (Scheme 7) l3 h I 1 Me Me Et Me Et Et Et Et Et H H H B B B 96 99 90 k Et Me Me H B 100 I Et Me Et H B 99 Table 13 Preparation of methylenebisanilines (59) 4 1 2 Diarjlmethanes and their Hetero Analogues Such displacements were also successful with other electron-rich aromatic compounds such as 1,3-dimethoxybenzene 1,3,5-tri- Scheme 6 methoxybenzene and 2-naphthol to give diary1 methanes (60) and (61) Similar reactions with electron-rich heterocycles such toluene as indole N-methylindole pyrrole and N-methylpyrrole,+ MeMgl ,-. oT-&0-5Bt reflux 3h afforded the hetero analogues (62) and (63) respe~tively~~ N 90% (Table 14) I II (57) 4 1 3 Substituted Diarylmethanes Scheme 7 Di- and triarylmethanes containing electron-donating groups in the ortho or para positions are of considerable importance as 4 Displacement of Benzot r iazo le by Electron-they are leuco dyes which on hydride abstraction by oxidizing rich Aromatic and Heteroaromatic agents give coloured cations of the type of Michler's hydro1 Compounds (64) Crystal Violet (65),and Malachite Green (66) 34 The derivatives (68) (obtained via lithiation) react with aniline 4.1 With 4-(Benzotriazol-l-ylalkyl)anilineDerivatives and indole to give the substituted diarylmethanes (67) and the heteroaryl analogues (69) in good yields 33 It is noteworthy that 4 I 1 Methylenebisanilines under the acidic conditions the hydroxyl group is stable as Methylenebisanilines are well known compounds A large shown in cases of (69c-e) (Table 15) For the derivative from number of papers and patents exist dealing with the preparation benzophenone a mixture of four products (72)-(75) was and extensive applications of such compounds l9 22 Thus they obtained This strengthens our belief that in such reactions are used as curing agents for epoxy resins and urethane elas- benzotriazole leaves initially forming a relatively stable benzylic tomers as intermediates in the preparation of polyurethanes in cation which can then be trapped by various nucleophiles Thus the synthesis of polyamides in the preparation of azo and other the initially formed cation (71) can be trapped by indole to give CHEMICAL SOCIETY REVIEWS 1994 OH ‘R2 Reaction conditions (1) 50% aq. MeOH conc. HCI reflux; (2) 1 M KOH. (a) 1,3-Dimethoxy- or 1,3,5-Irimethoxybenzene;(b) 2-Naphthol; (c) lndole or N-methylindole; (d) Pyrrole or N-methylpyrrole. Yield (Time h) R’ R2 (60) (62) (63) a H H 53% (72) 92% (48) 29% (55) b Me H 50% (72) 96% (7) 52% (21) c El H 68% (72) 95% (72) 41% (120) d H OMe 80%(57) o Me OMe 73%(27) f Et OMe 72% (72) g H Me 85% (72) h Me Me 98% (20) 45% (24) i Et Me 82% (44) Table 14 Preparation of diarylmethanes (60) and (61) and their hetero analogues (62) and (63) Me,N QH R (64) Michler’s Hydro1 (65)R = NMe Crystal Violet (66) R = H Malachite Green lndoleNMe 1) 50% aq. MeOH NMe,IE=Me conc. HCI reflux PhNEt2 f--2) 1 M KOH 1) HOAc Q.,reflux 4 h 2) 1 M KOH E NMe H67% (69) E Time (h) Yield (Yo) a Me 18 91 b PhCH 40 82 (-)(OH 17 86 d MeCHOH 72 38 0 N~CHOH 20 70 Table 15 Preparation of substituted diarylmethanes (67) and heteroaryl analogues (69) BENZOTRIAZOLE-MEDIATED ARYLALKYLATIONS-A R KATRITZKY AND XIANGFU LAN the regular product (72) or by MeOH to give (73) Dehydration of (72) afforded (75) and migration of a phenyl group gave (74) (Scheme 8) 4.2 With o-(Benzotriazol-1-ylalky1)phenolDerivatives 4 2 I o of-Methylenebisphenols Symmetrical u 0‘-methylenebisphenols are well known but unsymmetrical analogues are far less investigated Part of the reason can be attributed to the difficulty of their preparation The methylenebisphenols are important precursors to calix[n]ar- enes which act as molecular receptors or enzyme mimics 36 38 HO I .-\/+ HO 0 Ph (73) (74) NMe <’ I1 l-i l-i (75) Reaction condition (1) 50% aq MeOH conc HCI (2) 1 M KOH Yield (%) 26 9 5 10 Scheme 8 2eq NaOPt ’PrOHreflux 40h R2&/ R2 -@Bt +A’ (76) (78) R‘ R2 Yield (“1.) aHH 28 b H 4Me 57 c H 4tBu 39 d H 35diMe 55 e H 4Ph 28 f tBu 4Me 62 Table 16 Preparation of o o methylenebisphenols (78) 1 3 dimethoxy benzene OH OMe p TsOH (cat ) 369 Two papers35 39 have described a general procedure for the preparation of both symmetrical and unsymmetrical o 0‘-alkyli-denebisphenols by using the magnesium salt of benzylic alco- hols However the procedure started with a limited number of uncommon substituted o-hydroxybenzaldehydes as starting materials We found that displacement of the benzotriazole group In the o-(benzotriazol-1-ylmethyl)phenols (76) could be effected by phenols in the presence of sodium isopropoxide to afford symmetrical as well as unsymmetrical o o’-methylenebisphenols (78) in moderate to good yields40 (Table 16) Similar displace- ment with derivatives bearing a substituent at the methylene carbon (obtained vra lithiation) is still under investigation 4 2 2 Diarylmethanes The benzotriazole group was also displaced by 1,3-dimethoxy- benzene and indole to give the desired diarylmethanes (79) and (80) respe~tively~~(Scheme 9) 4.3 With 1-[(Methoxyaryl)alkyl]benzotriazoles 4 3 I Diarylmethanes Similar to the aniline and phenol derivatives 1-[(methoxyaryl)- alkyl]benzotriazoles react with electron-rich aromatic com- pounds such as N N-dimethylaniline 1,3,5-trirnethoxybenzene and 1,3-dimethoxybenzene to give the expected products such as (82) and (83) ’In the case of 1 ,3-dimethoxybenzene1 in addition to the simple product (84) disubstituted product (85) is also formed (Scheme 10) 1-( 1-Benzotriazol-1-ylalkyl)-2-methoxy-naphthalene (86) reacts similarly with 2-methoxynaphthalene NN-dimethylaniline and indole to give (87) (88) and (89) respectively (Scheme 11) 4.4 With 1-(Diarylalky1)benzotriazoles 4 4 I Asymmetric Triarylmethanes We recently found4’ that 1-(diarylalkyl)benzotriazoles can be obtained from the reactions of an aromatic compound with 1-(benzenesulfony1)benzotriazoleand an aromatic aldehyde The derivatives thus obtained can then react with an electron-rich OMe OMe PhNMe reflux 70 h OM-. NMe (82) (81) 1 3 5 trimethoxy benzene Ar,CH 40% (83) 1 3 dimethoxybenzene reflux 24 h Ar (84) (38%) (85) (48%) Reaction condition (1) 50% aq MeOH conc HCI reflux (2) 1 M KOH Scheme 10 lndole I I closed vial ?H ? closedvial 150% 30 h 15OoC 35h OMe 34% (79) Scheme 9 370 CHEMICAL SOCIETY REVIEWS 1994 aromatic or heteroaromatic compound to give asymmetrical triarylmethanes. Thus compound (90) was treated with N,N-dimethylaniline and indoles in CH,CI in the presence of zinc chloride to give compounds (91) and (92) in good yields (Table 17). (86) 4.5 With (a-Benzo triazol ylalk yl)-su bsti tu ted Heterocycles E=Me PhNMe 4.5.1 I,]-Bis(heteroaryl)alkanes reflux 73 h (a-Benzotriazolylalky1)-substitutedheterocycles (93) and (96)I52% OMe react with 2-methylfuran,2-methylthiophene,N-methylpyrrole,E=Me N-methylaniline and N,N-dimethylaniline to give I 1-bis (heteroary1)alkanes (94)-(95) (97-(98) and diarylmethanes (99)-( 100)in good to excellent yields12,42 (Schemes 12,13).The derivatives obtained via lithiation of 1-methyl-3-(benzotriazol-H 1-ylmethyl)indole also reacted with N-methylaniline and a (89) Grignard reagent to afford compounds (1 02)-( 104) (Scheme 14).12Reaction condition (1) 50% aq MeOH conc HCI reflux (2) 1 M KOH Scheme 11 NMe2 I ZnCl-JCH2C12 Me PhNMe,5536.Me Bt Me Me OMe aH H 57 c H Me 70 b Me H 85 Table 17 Preparation of asymmetrical triarylmethanes (91) and (92) (94) X = 0.R = Me 92% (95) X = NMe R = H 52% Scheme 12 (97) x= 0 84% (98) x= s 90% I (96) \ i,PhNMeR (99) R=H 53% (100) R = Me 74% R N' \ he Scheme 13 BENZOTRIAZOLE-MEDIATED ARYLALKYLATIONS-A R KATRITZKY AND XIANGFU LAN 37 1 Scheme 14 NMe NMe,I ~1 = H I CH,(COPh) toluene/ZnBr 725% kH,CH (COP h)2 ~ R’ El El Et Et R2 El H El H R3 Et Bun -(108) 43% 48% -(109) -44% 71% Table 18 Displacement of benzotriazole in 4-(benzotriazol- 1-ylaklyl)- anilines by active methylene compounds alcohols and thiols 5 Displacement of Benzotriazole by Other Nucleophiles 5.1 From 4-(Benzo triazol- 1-y lalky1)anilines 4-(Benzotriazol-1-ylalkyl)anilines (1 06) react with other nucleo- philes Active methylene alcohols and thiophe- nol’ each give the corresponding para-substituted products (105) (107)-(109) The active methylene compounds yield different products depending on the reaction medium used Thus in anhydrous/aprotic conditions with ZnBr as the catalyst compound (105) was obtained While in aqueous acid one of the acyl groups was removed by hydrolysis to give simple ketones (107) (Table 18) 5.2 From o-(Benzotriazol-1-ylmethy1)phenols As we have shown earlier in the present account reactions of o-(benzotriazol-1 -ylmethy1)phenols involve o-quinone methides as the reactive intermediates which are trapped by dienophiles via [2 + 41 cycloaddition We have further demonstrated that such o-quinone methides are Michael acceptors 43 44 thus o-(benzotriazol-I-ylmethy1)phenols(1 1 1) react with thiols alco- hols amines and active methylene compounds to give the substituted phenols (I lo) (1 12)-(114) 45 With diethyl methyl- malonate ester exchange was observed in isopropoxide to give product (1 14) in 53% yield (Table 19) 5.3 From 1-[(Methoxyphenyl)alkyl]benzotriazoles Similar to the aniline and phenol derivatives 1-[(methoxyphe-nyl)alkyl]benzotriazoles (1 15) also reacted with phenol and thiophenol to give substituted phenyl ethers and sulfides (1 16)’ (Table 20) 5.4 From (Benzotriazolylalky1)indoles The benzotriazole moiety in (benzotriazolylalky1)indoles(1 17) was also displaced by a thiophenol group to give substituted indole (1 18) (Scheme 15) 6 Summary N-(Arylmethyl)benzotriazoles containing in the aryl ring an ortho or para electron-donating group such as amino hydroxy or methoxy undergo lithiation at the methylene carbons to give anions which react with a variety ofelectrophiles The benzotria- zole groups in the parent derivatives as well as those obtained via lithiation have been displaced by Grignard reagents and reduc- tively removed with LiAlH or Na/piperidine to give alkyl- substituted aromatic compounds These N-(arylmethy1)- and N-(arylalky1)benzotriazoles are also efficient arylalkylating rea- gents for electron-rich aromatic and heteroaromatic com- pounds alcohols thiols amines and active methylene com- pounds This general methodology has also been extended to heteroaryl analogues in which the ring heteroatom acts as the electron-donating group In general the reactions proceed in good to excellent yields and isolation and purification of the products was simple In many instances this new methodology represents the method of choice for the preparation of whole classes of compounds CHEMICAL SOCIETY REVIEWS 1994 Acknowledgment We acknowledge the help of our many collea- OH SR2 R2SNa OH 61 ~1 =H OH OF? gues who have contributed to this review and in particular Dr IP~ON~P~OH reflux reflw Jamshed Lam who was associated with much of this work 7 PrONd'PrOH 'PrONd 'PIOH 1 reflu reflux 40 h 53% + OH 2 3 4 5 6 7 8 ~~~ ~ ~ ~ (110) R' R2 Time(h) Yield(%) 9 a H Ph 40 55 10 11b H n-C,H17 40 69 I2 C 2-b~ Il-C,H17 25 74 13 d 2-Me n-C,H17 30 74 14 15 16(112) R' R2 Time(h) Yield ("YO) ~~ a H Et 40 26 17 18 b H 'Pr 40 60 19 (113) R' NR2R3 Time (h) Yield ("A) 20 a H N(CH2CH2),0 40 66 21 b H NMePh 24 33 22 Table 19 Displacement of benzotriazole In o-(benzotriazol- 1-ylmethyl)-phenols (11 I) by thiols alcohols active methylene com-23 pounds and amines 24 25 26 27 28 29 30 (116) R' R2 X Solvent Temp ('C) Yield(%) 31 a H Et 0 PhOH 180 33 32 b H H 0 PhOH 180 35 33 c OMe H 0 PhOH 180 27 34 d H Et S toluene 80 45 e OMe H S toluene 80 48 35 36 f H H S toluene 40 56 37 Table 20 Displacement of benzotriazole in 1-(methoxyphenylalkyl)ben-38 zotriazoles (1 15) by phenol and thiophenol 39 Y WBt NaSPh 73% I Me (117) Scheme 15 References A R Katritzky X Lan and W -Q Fan Benzotriazole as a Synthetic Auxiliary Benzotriazolylalkylations and Benzotriazole Mediated Heteroalkylation Sinthesis 1994,445 A R Katritzky S Rachwal and G J Hitchings Tetrahedron 1991 47,2683 H W Gschwend and H R Rodriguez Org React 1979,26 1 A R Katritzky and J N Lam Heteroatom Chem 1990 I 21 A R Katritzky and W Kuzmierkiewicz J Chem Soc Perkin Trans I 1989,819 A R Katritzky H Lang and X Lan Tetrahedron 1993 49,7445 A R Katritzky X Lan and J N Lam Chem Ber ,1991,124 1819 H Gilman and J W Morton Jr Org React 1954,8,258 R R Fraser M Bresse and T S Mansour J Am Chem Sue 1983 105,7790 N S Narasimhan and R S Mali Sjnthesis 1983,957 G H Posnerand K A Canella J Am Chem Soc ,1985,107,2571 A R Katritzky L Xie and D Cundy Synrh Commun 1994 in press A R Katritzky J Li and C V Stevens unpublished results G W Gribble W J Kelly and S E Emery Sjnthesis 1978 763 C F Nutaitis and J E Bernardo Sjnth Commun 1990,20,487 A R Katritzky S C Jurczyk M Szajda I V Shcherbakova and J N Lam Svnthesis 1993,499 A R Katritzky X Lan and J N Lam Chem Ber ,199 1,124,1809 A R Katritzky and X Lan Sjnthesis 1992 761 Sanyo Chemical Industries Ltd ,Japanese Patent 59,208,546 (1984) Chem Abstr 1985 102,212749a Nippon Mining Co ,Ltd ,Japanese Patent 59,230,094(1 984) Chem Abstr 1985 103 73675d T K Bilozor S S Polevik G A Voloshin and L F Tikhonina kv Vyssh Uchebn Zaved Khim Khim Tekhnol 1987 30 72 Chem Abstr 1988 108,23307t S Ishii N Takahashi T Kamimura and Y Nakano Japanese Patent 63 45,242 (1988) Chem Abstr 1988 109,74590b J H Gorlin J Chem SOC 1955,83 I E Pollak and G F Gnllot J Org Chem 1967,32 3101 V Peesapati P L Pauson and R A Pethrick J Chem Res (S),1987 194 J Barluenga A M Bayon P Campos G Asenslo E Gonzales- Nufiez and Y Molina J Chem Soc Perkin Trans 1 1988 1631 J Barluenga P J Campos M A Roy G Asensio J Chem Soc Chem Commun 1979,339 J Barluenga P J Campos M A Roy G Asensio J Chem SOC Perkin Trans I 1980 1420 G Sunagawa T Ichii and N Yoshida Pharm Bull (Japan) 1955 3 109 Chem Abstr 1956,50 10054b G Sunagawa Pharm Bull (Japan) 1955 3 116 Chem Abstr 1956,50 10055a G Sunagawa Pharm Bull (Japan) 1955 3 124 Chem Abstr 1956,50 10055d A R Katritzky X Lan and J N Lam Svnrhesis 1990 341 A R Katritzky X Lan and J N Lam J Org Chem 1991 56 4397 H Zollinger 'Color Chemistry' VCH Publishers Weinheim 1987 P 59 A Pochini and R Ungaro Sjnthesis 1975,617 V Bohmer F Marschollek and L Zetta J Org Chem 1987 52 3200 A Wolff V Bohmer W Vogt F Ugozzoli and G D Andreetti J Org Chem 1990,55 5665 A McKervey and V Bohmer Chem Brit 1992 724 G Casiraghi G Casnati M Cornia A Pochini G Sartori dnd R Ungaro J Chem SOCPerkin Trans I 1978,322 Y BENZOTRIAZOLE-MEDIATED ARYLALKYLATIONS-A. R. KATRITZKY AND XIANGFU LAN 40 A. R. Katritzky H. Lang Z. Zhang. and X. Lan J. Org. Chem. 43 B. Loubinoux. J. Miazimbakana and P. Gerardin Tetrahedron 1994 in press. Lett. 1989 1939. 41 A. R. Katritzky V. Gupta C. Garot C. V. Stevens and M. F. 44 T. Inoue S. Inoue and K. Sato Bull. Chem. SOC.Jpn. 1990 63 Gordeev Heterocycles 1994.38 345. 1647. 42 A. R. Katritzky L. Xie and W.-Q. Fan J. Org. Chem. 1993 58 45 A. R. Katritzky Z. Zhang. X. Lan and H. Lang J. Org. Chem. 4376. 1994.59 1900.

ISSN:0306-0012    

DOI:10.1039/CS9942300363

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

LIVERSIDGE LECTURE. The Dynamics of Photodissociation Richard N. Dixon School of Chemistry, University of Bristol, Bristol BS8 ITS, U.K. 1 Introduction A major object of modern chemical physics has been to obtain a detailed understanding of the factors that control chemical reaction dynamics, with the eventual goal of designing reactions to have a predetermined outcome. This is a field in which there has been a particularly strong interplay between ongoing experi- mentation and parallel theoretical developments. Molecular beams and lasers have been instrumental in many recent advances, most particularly in the study of photodissociation. This is inherently one of the simpler types of reaction in that the products are already in close contact in the reactant, and its detailed study has greatly advanced our theoretical understand- ing of reaction dynamics.These advances have highlighted two extremes for the distribution of the available energy over product motion; namely, those cases where the energy becomes statistically distributed over all possible product states, and those exhibiting specific dynamical control. Photochemically initiated reactions are of course also of intrinsic importance to us in such diverse ways as the generation of the ozone layer in the upper atmosphere whichprotects us from ultraviolet irradiation; in the photosynthetic cycles which provide much of our food; and in the recording of information by photographic means. Early photochemical experiments were initiated using flash lamps, which usually give out a range of photon energies. While such experiments often led to the identification of the reaction pathways and the detection of unstable intermediates, they did Richard Dixon is the AIfred Capper Pass Professor of Chemistry at the University of Bristol. He was educated at the Judd School, Tonbridge, at King's College, London, and at St.Catharine's College, Cambridge. His interest in spectroscopy was kindled by W. C. Price at King's, where he graduated with first class honours in physics. A desire to move to chemistry led to a Ph.D. in infrared spectroscopy under N. Sheppard at Cambridge. He spent two veurs as an analyst with the Atomic Energy Authority. This ww followed by three postdoctoral years in Canada.Returning to England, he spent one year as an ICIfellow and nine years as a lecturer in chemistry at Shejield University. He was the Sorby research fellow of the Royal Society,for the last five ofthese years. He MUS appointed to the chair of theoretical chemistry at Bristol University in 1969, and the Capper Pass chair in 1990. Dixon 's early spectroscopic research was concernedprincipally ItTith molecular structure, initially of stable molecules but then increasingly of transient species. This broadened into an interest in photochemistry and time-dependent molecular phenomena. His experimental studies have been underpinned by the development oftheoreti- cal models of molecular struc- ture and quantum dynamics.Previous recognition includes the RSC Corday- Morgan medal (1966) and the spectro- scopy award (1984). He was elected to the fellowship ofthe Royal Society in 1986. not necessarily reveal the detailed motion during the primary bond-breaking step. The use of lasers, and particularly tuneable dye-lasers, both to initiate photochemical decomposition and to probe the identity of, and energy disposal within, the primary reaction products, has removed this limitation. These advances are a consequence of four properties of lasers: monochromaticity, which often permits parent molecule activa- tion and product detection at a quantum specific level; intensity, which enhances the sensitivity of product detection, permitting low enough pressures in the gas phase to avoid collisional relaxation, and also facilitates multiphoton excitation to high energy states; directionality and polarization, which can be used to probe the correlation between the vectorial properties of the transition dipoles and the rotational or translational motions of the products, thereby giving insight into the stereo-specificity of the primary reaction step; and pulse duration, which at its shortest can match the femtosecond time-scale of the motions of the atoms in a molecule.Diatomic molecules have the most simple dynamics of photo- dissociation because the (one) vibrational coordinate is also the reaction coordinate (Figure 1). When a diatomic molecule is excited by absorption of a photon to an electronic state having a potential energy curve repulsive between the atoms they will recoil from one another on the time-scale of a vibrational period (-30 fs).This is termed direct dissociation, and is characterized by a structureless absorption spectrum for the parent molecule (recall the Uncertainty Principle). In contrast, a highly struc- \ A*+B xE" Q, w lnternuclear distance Absorption Figure 1 Schematic representation of potential curves and the corres- ponding absorption spectra for a diatomic molecule AB. Electronic state I is bound, but its levels are predissociated through interaction with the continuum of state 2. The absorption from state 0 to state 1 therefore consists of broadened lines, with widths inversely pro- portional to the dissociation lifetime. Electronic state 3 is directly dissociative, and the absorption spectrum from state 0 to state 3 consists of a broad continuum.375 376 tured absorption spectrum indicates the presence of a potential well for the excited state, giving rise to long-lived states which may often be characterized by excited state quantum numbers. Any dissociation from these states occurs on a slower time-scale than for direct dissociation. This is termedpredissociation, and is brought about by quantum tunnelling through a barrier, or by a change of electronic state near a curve crossing. In either case, any excess energy above the electronic energy of the separated atoms (the available energy) must appear in their mutual recoil: Polyatomic molecules have a much richer diversity of disso- ciation mechanisms, since energy can flow between many inter- nal modes of excitation, and the distinction between direct and indirect dissociation is less clear-cut than in diatomic molecules. In his classic book Herzberg' classified polyatomic predissocia- tion mechanisms into case I, predissociation by electronic transition; case 11, predissociation by vibration; and case 111, predissociation by rotation.It is now recognized that it is possible for these processes to occur simultaneously with rates that may vary with the vibrational or rotational quantum numbers, giving a quantum state specificity to the mechanism. Energy flow between internal modes during dissociation is not limited to long-lived states. In direct dissociation of a poly- atomic molecule into fragments X and Y the potential will be repulsive along the coordinate Rxu between the centres of mass of the fragments X and Y, thereby promoting recoil.But the structures of minimum energy for X and Y in the parent molecule may differ from those in the isolated fragments after dissociation, or the atoms initially connected by the breaking bond may be dynamically coupled to the vibrational modes within X and Y. In either case the motion of the atoms during dissociation is not simply one of extension of RXY,and part of the available energy ends up as internal excitation of either product instead of mutual recoil: In an ideal experiment the distribution of motion between all these degrees of freedom would be determined in full, albeit with some redundancy through conservation laws.The internal energies of the fragments are usually probed using laser-induced fluorescence (LIF) or resonance-enhanced multi-photon ioniza- tion (REMPI) spectroscopy, with relative populations obtained via line intensities. The distribution of recoil velocities is mea- sured through the Doppler-broadening of lines in the LIF or REMPI excitation spectra, or with higher resolution by free time-of-flight (TOF) over a known pathlength using electron bombardment ionization or REMPI with detection via ions. In molecules with no more than about six atoms good quality potential energy surfaces may be available through quantum chemical calculations.Where known, these have provided a valuable framework with which to interpret photochemical experiments, particularly when the time-dependent nuclear Schrodinger equation is used to calculate the nuclear motion on a single excited surface. However, many dissociation products are open-shell species, and when two such entities are brought together several potential surfaces of the combined system correlate with each dissociation limit. The consequence is that there are usually many surface crossings in polyatomic mole- cules, and the different channels may even lead to different chemical products, such as H,O* -+ H + OH or H, + 0. This article summarizes recent advances in photodissociation, with particular emphasis on systems with which the author has been involved.The molecular systems are grouped according to their basic mechanisms. Section 2 covers dissociation on a single potential energy surface: Section 3 dissociation mediated by a crossing of potential energy surfaces; and Section 4 dissociation following internal conversion to the vibrational continuum of CHEMICAL SOCIETY REVIEWS, 1994 the molecular ground state. Section 5 summarizes ongoing developments. 2 Dissociation on a Single Potential Energy Surface Where dissociation takes place on a single potential energy surface the Born-Oppenheimer separation of electronic and nuclear motion is usually valid. It is then sufficient to consider the dynamics of the nuclear motion independently of that of the electrons, although there may be some recoupling of electron spin or orbital motion as the products are formed at long range.This is the case in each of the examples given below. A quantitatively accurate theoretical description of the dynamics requires a quantum treatment, but a swarm of classical trajector- ies with starting conditions that mimic the distribution of positions and momenta of a Franck-Condon wavepacket is equally informative. 2.1 H,O A State A simple example of a direct dissociation is proxided b,y the dissociation of water vapour excited in its first (AIBl-XIAl) electronic absorption band near 165 nm,2 in that most of the available energy appears as mutual recoil of H from OH (Table 1). Table 1 The distribution of the available energy for photolysis of H20 at 157 nm Fraction in translational recoil ft -90% Fraction in OH rotation fr -1.2% Fraction in OH vibration L -10% When the H,O is jet cooled to ca.10K only about 240 cm-appears as OH rotational energy. This modest rotation derives largely from the zero-point energy of the H,O bending vibration v2which is a disappearing mode. The two H atoms share most of the initial HOH bending kinetic energy of v2/4 -41 5 cm -* in the form of relative angular oscillations of the bonds. Upon photo- lysis this correlates with tangential motion of one H atom and rotation of its OH partner, thereby imparting an average of -210 cm-I of rotational energy to the OH fragment.Two points of note in this system are: (i) with excitation by linearly polarized li ht the axis of OH rotation is strongly aligned parallel to th f.electric field. Since the transition moment for 'B,-'A, is perpendicular to the H20 plane, the sense of this alignment confirms that the rotational motion takes place in the instantaneous plane of the parent molecule; (ii) one set of A-doublets of the OH (X217) radical product is much more strongly populated than the other, indicating that the unpaired electron left behind in OH is strongly constrained to lie in a 2p~ orbital perpendicular to this plane of rotation. This is a clear example of an orbital conservation rule: one electron in water is excited from a 2p~lone-pair orbital to an in-plane anti-bonding orbital, and goes away on the departing H atom, leaving its unpaired partner behind in the 2pn orbital of the rotating OH molecule: H,O (...,2paI,2pb1,3.~a*a,;A'B,)-, H(lsa';2S) + OH(.* .,2p7rd2, 2pna'' ;X I7") (3) Considerably more energy appears as OH vibration, with population in u = 0,1, and 2. Dynamical calculations on an ab initio potential surface have shown that to a considerable degree this excitation has its origin in the symmetry of H20. Excitation at the ground-state geometry (a vertical transition in accordance with the Franck-Condon principle) leads to a region of the upper-state potential where its slope results in an initial repul- THE DYNAMICS OF PHOTODISSOCIATION-R. N. DIXON sion along the symmetric bond-stretching coordinate.However, at longer range there is a high barrier inhibiting dissociation along this coordinate because this would require both bonds to break. A small probability of vibration back to the Franck- Condon region gives rise to the observed slight undulations on the absorption band profile, with a vibrational spacing of -1900 cm-I, but in general this barrier deflects the motion towards stretching of only one bond. Even so, some of the initial symmetric excitation remains as vibration of the unbroken OH bond as the H and OH move apart, the extent of which increases with increase in the available energy. 2.2 H,O, A State The direct association of hydrogen peroxid: is strikingly differ- ent. Photolysis within its first continuum (A'A-X'A)at 266 or 248 nm releases about 11% of the available energy into OH rotation but gives no vibrational e~citation.~ However, the novel feature revealed by a polarization study of OH Doppler line profiles4 is the opposite signs of the effective recoil aniso- tropy parameters fle% for related pairs of transitions when recorded with the detection geometry of Figure 2.This indicates that there is a strong correlation between the fragment rotatio- nal and translational vectors JOHand vOH,with a clear prefer- ence for rotation around the axis of recoil. This non-trivial result cannot be the result of initial zero-point energy of disappearing bending vibrations, but is the dynamical consequence of forces that act during the dissociation.10 1 II 0' I I I I I 1 309.85 309.87 309.89 h I nm Figure 2 Line profiles, Doppler-broadened by the recoil velocity u, in the LIF excitation spectrum of the OH fragments from photolysis of H,O, at 266 nm. The photolysis (polarization axis cp) and analysis (polarization axis ea) laser beams were counter-propagated, and the fluorescence was detected with a photomultiplier (P.M.) at right angles. The two transitions have the same lower level, and thus probe the same group of molecules, but with opposite angular anisotropy relative to the axes of their rotational angular momentum J. The contrasting profiles show that u and J have a parallel correlation. Repulsion between lone-pair electrons on the two 0 atoms bestows a skewed equilibrium structure to H20, in its ground state, with a dihedral angle of -11 1.5" between the two OH bonds.The lowest energy electronic transition involves the promotion of one of these lone-pair electrons to an anti-bonding orbital, as in H,O. This excitation generates repulsion between the 0 atoms, resulting in HO-OH recoil, but it also removes the twisting force so that the initial motion is for the H atoms to move towards planarity as the fragments move apart. The result of this torque is that the OH radicals recoil in a corkscrew fashion, with much greater excitation of rotational motion than in OH from H20 (Figure 3). These dynamics appear to be a Figure 3 The dynamics of the dissociation of H,O, following excitation in its first absorption band, for which the transition dipole p bisects the angle between the two OH bonds.The H atoms move towards a trans-planar geometry as the OH radicals start to separate. Thus both the recoil velocities u and the most probable OH rotational vectors J are perpendicular to the dipole axis and parallel to each other. general characteristic of the peroxide bond, and a very similar behaviour has been found for the OH fragment from heavier peroxides such as tertiary butyl peroxide. 2.3 HONO A State Nitrous acid, HONO, provides a good example of a molecule with an excited state which is sufficiently long-lived to permit transient vibration. The near ultraviolet absorption spectra of nitrous acid, and of all alkyl nitrites, show diffuse overlapping bands with a mean spacing of V; -1100 cm-' characteristic of the vibration of a weak terminal NO bond. Dissociation to give the two molecular products (OH and NO) is mediated by a partial energy-transfer between the vibration of this terminal bond and stretching of the central bond.The Doppler widths of the lines in the spectra of both products show that about two thirds of the available energy is liberated in their mutual recoil. Dissociation of HONO at 355 nm, which is the most studied wavelength, proceeds via the 2; band (excitation to the quasi- bound level with u; = 2). This yields OH exclusively in its ground vibrational level, with little rotational excitation -2 10 cm-I). In contrast the vibrational levels of NO are populated up to v = 3, each with a high and inverted rotational excitation (Table 2).The OH line profiles, together with momentum conservation, show thatf, -60% for the combined OH and NO recoil fraction of the energy, and that all the available energy has been accounted for.5,6 Table 2 The distribution of the available energy for photolysis of HONO at 355 nm Fragment molecule OH NO Fraction in translational recoil fi 38% 22% Fraction in rotation Fraction in vibration f,fv 2%-0% 13% 25% Electronic orbital alignment IZ(A"):IZ(A') 1 :3.4 2.8:1 Polarization studies confirm that the rotational motions of both the OH and NO fragments are largely constrained to the original HONO plane. The observed A-doublet propensities (Table 2) show that an in-plane n-orbital (na')is favoured for the unpaired electron of the OH, but an out-of-plane n-orbital (na") for the NO electron.This is consistent with an electronic rearrangement in which the central N-0 0-bond is broken following an out-of-plane n*-n excitation within the -NO group (Figure 4). CHEMICAL SOCIETY REVIEWS, 1994 Figure 4 Electronic rearrangements during the photolysis of nitrous acid. The n*-n excitation weakens both the HO-NO and HON-0 bonds. Stretching of the central bond is then associated with a back transfer between in-plane orbitals, restoring the strength of the terminal NO bond (see the potential contours of Figure 6), leaving one unpaired electron on each rotating fragment with the indicated stereospecificities.As in the case of H,O discussed above, the modest rotation of the OH fragment derives from disappearing bending modes, of which the in-plane L HON vibration has the highest frequency. In contrast, the NO rotation is far too energetic for such an origin (Figure 5). The contrasting patterns of OH and NO rotational excitation are readily understood if much of the available energy is released in repulsion directed between the central 0 and N atoms. A force along this bond passes close to the centre of mass of OH, but exerts a strong torque around the more distant centre of mass of NO. Thus a significant fraction of this repulsive energy release will be channelled into NO rotation, but very little into OH rotation.Model calculations show that the NO rotational excitation is slightly less than predicted by this simple impulse picture. The explanation of the vibrational branching has been guided Figure 5 The LIF excitation spectrum of the NO(A-X) 0-2 band following photolysis of HONO at 355 nm. The profile of this band reflects the population distribution within the u" = 2 lower state. This is approximately Gaussian in rotational quantum number, with a mean value of J,,, = 26 and a standard deviation of uJ= 6.5. by ab initio calculation of the potential energy surface, and is illustrated by classical trajectories in Figure 6 (note that the complete mechanism involves synchronous oscillation of the LONO angle as well as stretching motions).We can consider this dissociation to occur in two steps. In the first slow step (T 5 90 fs) V-V transfer partially redistributes the excitation energy between vibration along the (H0)N-0 and (H)O-N(0) coordinates until the motion is directed towards a well-defined saddle point. Most of the available energy is then released impulsively in a fast second-step during which there is little change in the vibrational distribution of the emerging NO, and the motion is largely confined to a plane thereby generating the translational recoil and NO rotation. The dissociation of alkyl nitrites follows a similar pattern, albeit with some vibrational excitation of the alkyl radicals. This family therefore typifies Herzberg's predissociation by vibration (case 11).The essential feature which gives rise to this behaviour is the rapid change in bonding as the central ON bond is stretched. This results in the shallow potential minimum which can support transient vib- ration, with dissociation driven by the anharmonicity of this potential. 3 Dissociation Mediated by Surface Crossings There are three common types of surface crossing for which symmetry plays an important role. In molecules such as H20, HCO, and SiF: there are states which are orbitally degenerate for a linear geometry, but this degeneracy is lifted by bending 1.6 4 .1.4:9 a 1.2 1.o 1.2 1.4 1.6 1.8 2.0 2.2 RHO-,, 1 A Figure 6 A representative trajectory showing the relative motions in the ON and NO coordinates of HONO during photodissociation. The initial motion starts in the Franck-Condon region of coordinates (hatched), and is principally a vibration of the terminal NO bond. However, part of the vibrational energy becomes redistributed into motion of the central ON bond, thereby facilitating escape from the shallow potential well.I I I I I 248 247 246 245 244 h/nm THE DYNAMICS OF PHOTODISSOCIATION-R N DIXON with a splitting which initially increases quadratically with the amplitude of bending The two component states have different symmetry in the lower point group, and can only be mixed through electronic Coriolis interactions in a rotating molecule (the Renner-Teller effect) In the second type, for molecules such as H2S or NH,, two independent surfaces of different character cross in configurations of high symmetry, but have the same character on distortion to a lower point group This results in a 'conical intersection' of the potential hypersurfaces in which the splitting between them is initially linear in two or more nuclear coordinates Finally, for molecules such as CH,, there is a conical intersection which derives from an orbital degeneracy in a non-linear configuration of high symmetry (the Jahn-Teller effect) In both these latter cases the interstate coupling is vibronic in origin and is independent of rotation The examples below illustrate each of these types of surface intersection We will see that there are cases where a surface is subject to more than one of these crossing mechanisms 3.1 The A2A"and X2A'States of HCO The emission spectrum known as the 'hydrocarbon flame bands', which has also recently been observed in LIF, provides experimental data for a large number of vibrational levels of HCO in its electronic ground-state 'Many of these bands show d rotational diffuseness which arises from the predissociation of those lower levels which lie above the dissociation limit for HCO (R2A')-+ H + CO (XIC+) This is another example of Herz- berg's predissociation by vibration (case 11) These levels will not be discussed further save to note that their resonance energies and widths are well accounted for by a three-dimensional time- dependent wavepacket calculation using an ab znztzo potential energy surface The widths show systematic trends with increases in the vibrational quantum numbers, but not simply with the level energies * The A2A"state also has a structure of predissociated vibronic levels The longest-lived excited levels are those with K, = 0, and the dissociation rate increases with increase in K, and decreases with increase in 21; (Figure 7) This is in quantitative accord with a model calculation9 for a Renner-Teller (Coriolis) induced coupling to the ground-state continuum (Figure 8), in keeping with the knowledge that this excited state has the wrong symmetry to correlate adiabatically with ground-state products which are the only ones energetically accessible A second more surprising consequence of the Coriolis nature of this surface crossing concerns the observed recoil anisotropy of both the H(D) and CO dEsociation products from HCO and from DCO O The A2A"-,X2A'transition moment is perpendi- cular to the HCO plane It would therefore be anticipated that nl I I I 1 "0 5 10 15 20 "2' Figure 7 The variation with the quantum numbers for bending (6)and a-axis rotation (K,) of the linewidths in the A2A" excited state of HCO 0 experimental values, __ calculation I \ /**' Linear Bent Figure8 The dissociation of a quasi-bound vibrational level of the A2A" state of HCO at energy E by coupling to the continuum of the %A' ground state (hatched) The surface-crossing is mediated by electron- nuclear Coriolis forces in near linear HCO The bold arrows indicate the variation of the electronic energy along the dissociation path recoil would take place preferentially in a plane perpendicular to the axis of the electric vector of a linearly-polarized photolysis laser, leading to a negative recoil anisotropy parameter /3 In fact /.3 is found to vary with the vibronic band chosen for excitation, but is either zero or positive, the largest values being between 0 7 and 1 0 This IS best explained in terms of a simplified represen- tation of the stationary state wavefunctions for the coupled states, in which we consider explicitly only the rotation about the molecular a-axis Let A [= 1 for HCO A/X2IIU],conjugate to the body-fixed electronic angle a, and K, conjugate to the a-axis rotational angle 4, be good quantum numbers let 8 be the LHCO angle, Y the CO bond-length, and R the H-CO dissociation coordinate The greatest recoil anisotropy is found to be for excitation via the K' = 1 tK" = 0 sub-bands (Table 3) For these, with 4 = 0 defined by the axis of the laser polarization, the two-component wavefunction for the excited state K doublet is In this equation the A and Born-Oppenheimer states are assumed to have a common electronic wavefunction &(re) in all coordinates except the angle a, and the u functions are vibra- tional wavefunctions associated with these two surfaces The excitation_ transition moment from the ground-state couples to the first (A2A") term, which has the major amplitude For K = 1 this peaks at 4 = 90" as expected for this perpendicular transi- tion Even so, only the second (X2A') term with the minor amplitude uA(0, Y, R)extends out to dissociation at R = co,and this component peaks at + = 0 leading to a positive value for /3 Table 3 presents the comparison between the observed values of Table 3 A comparison of the observed and calculated recoil anisotropy parameter /3 for HCO (DCO) A2A" .+H(D) + COX'C+ K' sub-band /3 observed p calculated 0-0 tl<Q> 0 25 0 13 0-Ot 1(R) 00 -0 06 I+ 1 to 07-1 0 1 04 I+, 1-It2 00 0 08 2+, 2-24-1 0 1 4 15 0 12 6 and those calculated with wavefunctions such as in equation 4, with the further assumptions that the H atom departs at 60"to the a-axis which is the minimum energy path to dissociation on the ground-state surface, and with allowance for some degrada- tion of ,8 through end-over-end rotation of the HCO.lo Thus this unusual recoil anisotropy comes about because of the need for the electrons and nuclei to transiently exchange angular momentum in order to facilitate the non-adiabatic surface crossing at linearity. For the K' = 1 +-K" = 0 sub-bands the most probable plane of dissociation is at right angles to that of excitation. The two-component wavefunction of equation 4 also leads to an understanding of the energy distribution among the vibra- tional and rotational levels of the CO photof!agment.ll In a time-dependent view of dissociation from the A state of HCO, the early motio? of,a wavepacket representing the vertical excitation from X to A (the first component in equation 4) will be towards linearity, as indicated in Figure 8.The Coriolis coupling will then generate the second component on the ground-state surface. In three dimensions this second component will initially be tightly constrained around a saddle-point as indicated in Figure 9, but will then follow a bending trajectory towards the energy minimum. This figure presents a two-dimensional cut of such a three-dimensional se5ond-component wavefunction 15.9fs after ccossing from the A-state surface, calculated using the ab initio X-state surface. By this time the wavepacket has bounced offa steep repulsive wall on the surface; part has been reflected back towards the linear saddle-point, and part has been deflected out towards dissociation over a region of the potential which is almost flat.180" e 90 0 0.9 2.5 4.1 R/A Figure 9 The time-evolution of a wavepacket component on the ab initio ground-state potential energy surface of HCO. R and 0 are Jacobi coordinates for H relative to CO. A; a component appropriate to crossing from the linear excited state: B; the real part of this wave- packet component 15.9fs later. The asymptotic analysis of the wave- packet to give CO level populations is carried out at R,. The energy interval between potential contours is 2000cm-I; three contours are labelled with the values of V/lo00 cm- l. Fourier analysis of this outgoing wave leads to predicated CO level populations as a function of the HCO energy.Figure 10 presents this analysis for excitation of HCO at 14908 cm-1.8 This predicts the greatest vibrational population of CO to be for vco = 0, but with extensive rotational excitation peaking between J(C0)= 35 and 40. This is in good agreement with experimental measurements following photolysis through a number of HCO* states in this general energy range. One aspect of this population distribution is that less than half of the available energy is released as internal energy of the CO product. This arises because the angular momentum composition of the outgoing wavepacket is established at quite short range, where CHEMICAL SOCIETY REVIEWS, 1994 100 75 50 t .-0 U .S 25 3 Q0 Q00.z 50 -rn 2 25 0 25 v(C0) = 2 & '-A' 0 10 20 30 40 JW) Figure 10 Calculated CO vibration-rotation product distribution following dissociation of HCO through the (O,O, 15) vibrational resonance in the A2A"state at 14908cm-'.much of the energy is locked up in centrifugal motion which evolves to become product recoil. 3.2 The AlA'; State of NH, This next example involves a vibronic rather than a Coriolis coupling mechanism. Whereas ammonia is pyramidal in its ground state, it is planar in its known excited states, all of which are sufficiently long-Lived to show at least vibrational structure. The first excited AIA; state correlates adiabatically over a small potential barrier with the ground-state products H + NH,(X2Bl) for motion along a planar dissociation path.It might therefore have been anticipated that excitation to this planar state, reached in the region of 200 nm, would result in most of the energy being released in translational recoil follow- ing dissociation on this single surface. However, early attempts to study the energy disposal following the photodissociation of ammonia were frustrated by an inability to assign the spectra of the transient product (presumed to be NH,), recorded either in emission or by LIF, because of congestion and the ubiquity of unknown spectral lines. The breakthrough came with the deve- lopment of a method of recording 33-atom time-of-flight spectra with high resolution and sensitivity.12 TOF spectra of the nascent H-atoms from monochromatic photolysis of NH, (or D atoms from ND,) show many sharp peaks, each of which correlates through energy conservation with an internal energy level of the partner NH, (or ND,), (see eqcation 2).Excitation to the zero-point level of the A-state near 216 nm gives the simplest TOF spectrum, most of the intensity being concentrated into a single series of peaks spanning the full range of the available energy of 1.08 eV (Figure 11). Analysis of this spectru-m, with the aid of a calculated energy level manifold for NH, (X2B1), reveals that this series corresponds to rotational excitation concentrated about an axis parallel to H--.H, the a- inertial axis, with N = &. Photolysis at shorter wavelengths proceeds via discrete vibrational levels of NH3 in which the out- of-plane vibration V; is excited.The trend is for strong popula- tion inversion of the NH, states as u; (NH,) is increased, accompanied by a lowering of the average recoil energy of the two partners. The NH, motion remains concentrated in a-axis rotation, but there is also some excitation of levels with N = KO+ 1 and an increasing excitation of the NH, bending THE DYNAMICS OF PHOTODISSOCIATION-R. N. DIXON 1.o F I I, e=oo o-8 I I II I 1.o 0.6 0.4 0.2 n-1000 3000 5000 7000 9000 TKER / cm-’ Figure 11 Spectra of the total kinetic energy release (TKER) recorded with H atoms from photolysis of NH, in its A-2 0; band at 46 197cm-’.The spectrum recorded with recoil parallel to eP (B = 0) is biased towards low recoil velocities and high NH, angular momentum: whereas that perpendicular to ep (0 = 90) is biased towards high velocities and low NH, angular momentum. vibration (Figure 12).In addition to this general trend there is an alternation of internal energy pattern as u; increases from 0 to The interpretation of these observations has been greatly aided by the results of model dynamical calculations on ab initio potential energy surfaces. These highlight four aspects of the dissociation mechanism. For u; = 0 or 1 dissociation can only proceed by quantum tunnelling through a barrier along any one of the NH stretching coordinates, the rates being an order of magnitude slower for ND, than for NH,.A symmetry con-straint for non-rotating ammonia requires two quanta of V; for conversion to one quantum of bond stretching, and for c; 22 dissociation is initiated by such a vibrational redistribution. Thus dissociation from c; = 2 is mediated via anharmonic l’-l,’ transfer to the stretching continuum associated with tl; = 0, and 2:; = 3 via 21; = 1, as clearly demonstrated by comparisons Figure 12 Assignment of the NH, internal energy spectrum from photolysis of NH, in its A-2 2;band at 47 11Ocm-l, with detection of H atoms perpendicular to ep. vs=2,N-Ka 8 5 10 between the product distributions for nu; excitation. There is also some evidence that vibration-rotation Coriolis coupling between bending and stretching can give an additional rota-tional enhancement of dissociation mediated via dc; = -1.I4 However, for all values of 21; of NH,, the final dissociation step to ground state H + NH, involves a dynarnical interaction between motion on the first excited state and on th: lowe$ surface.The dominant feature of this step is that the A and X surfaces exhibit a conical intersection in each H-NH, exit channel for planar ammonia (Figure 13). The A surface there-fore correlates with e-ucitedstateproducts for non-planar geome-tries, but this excited state dissociation channel is closed energe-tically from the lower 7i; levels of excited NH,. Internal conversion between these states can only proceed by funnelling of trajectories originating in non-planar configurations through these conical intersections, where strong forces amplify the initial inversion motion to generate the high a-axis rotation of the NH, product.The final energy disposal is therefore very dependent on the value of u; (NH,), and is strongly influenced by the precise geometry and forces acting in the vicinity of this transition structure. The observed polarization dependence of the TOF spectra (Figure 11) has revealed a striking aspect of the dissociation dynamics. Many of the departing H atoms follow trajectories lying close to the original plane of the excited molecule (i.e. perpendicular to cp). However, where the NH, molecule is left in certain very high angular momentum statesthe H atom follows a path more nearly perpendicular to this plane (i.e.parallel to cP). The exact path followed appears to result from a subtle interplay 65,000 V/cm’ Fig_ure13 The dependence of the potential energysurfaces for the ,?and Xstates of NH, on R(H-NH,) and the out-of-planeangle 0, showing their conical intersection. B is proportionalto the inversioncoordinate q2 for short R, whereas when R +cc it becomes the a-axis azimuthal angle of the NH, fragment. 11 15 IIIIII IIIIIIIIIII II I I Iv2=0, N-K, 10 15 20 Ka hhOOO0QQrlQ qw N -Ka+l 1 0 0.4 0.8 1.2 NH, internal energy / eV between angular momentum constraints, the heights of centrifu- gal barriers and the impact distance at which the H atom breaks free 3.3 The &A, State of H,O The second excited (B'A,) state of H20, r-ached near 120 nm, has many features in common with the A state of ammonia Again there is a conical intersection of the excited- and ground- state potential energy surfaces in each exit channel leading to H + OH But in addition thereJs also a se2m of intersection between the surfaces for the B'A, and AiBl states which comprise a single 'nu state for linear H20 These surface crossings are both associated with strong angular forces con- trolling the dissociation dynamics which are therefore subject to both vibronic and Coriolis coupling mechanisms in this case Earlier studies of the photodissociation of water vapour at room temperature, recording either the fluorescence of nascent excited OH(A2C+)or LIF detection of ground state OH(X217), suggested that the Renner-Teller (Coriolis) mechanism through the knear 'nuintersection provided the major exit route from the B1 A, state Unfortunately, an energy-dependent predisso- ciation in the A state of OH restricts the range of OH levels that can be monitored in these ways More recently, the Rydberg H- atom TOF technique has facilitated the observation of the full OH product population distribution (Figure 14) l6 The H- atom yields through each of the available channels following photolysis at 121 6 nm are given in Table 4 The OH A-state rotational population distribution agrees closely with that der- ived from spontaneous fluorescence It is strongly inverted peaking at N' = 20, which is close to the energetic limit The major ground-state population is even more inverted, peaking at N" = 45[N" = 63 for OD from D20],although this is well !elow $e energetic limit of N" = 60 [N" = 83 for OD] Since the A and X surfaces of H20 both correlate with H + OH(X), and the OH A-doublet splittings are not resolved, this result does not in the filst instacce allow differentiation between the penner-Teller (B'A, +ALBl) and conical intersection (B'A, +Xi A,) mechanisms Even so, a striking feature of this OH(X) population distribu- tion is that it exhibits a significant shoulder at N" = 40 (even more marked near N" = 50 for OD from D20, Figure 15) The interpretation of these distributions has been greatly aided by comparison with theoretical calculations using a time-depen- dent wavepacket study of the dissociation dynamics, using two- dimensional potential energy surfaces derived from ab znztzo calculations (I e with one fixzd bond-length) These show that for final dissociation on the A-state surface the product rotation Y 10000 20000 30000 40000 TKER / cm ' Figure 14 Total kinetic energy spectrum of the fragments recorded with H atoms from photolysis ofjet-cooled H,O at 121 6 nm The OH(X) and OH(A) internal energy scales are taken from the literature CHEMICAL SOCIETY REVIEWS, 1994 Table 4 H-atom yields through the available channels for dissociation of H20(D20)at 121 6 nm Channel H2O DzO HID + OH/OD(X) 64% 62Y0 H/D + OH/OD(A) 14% 30% H/D + HID + o(3~) 22% 8 Yo Qg 60*It a, -2 a, [r I\ 2o t f t 0 1 10 20 30 40 50 60 70 80 N(OD) Figure 15 Experimentally derived rotational state population distribu- tion in the OD (X, L = 0) fragments arising from 121 6 nm photodisso- ciation of D,O is entirely generated on the B surface while the H20molecule is openingo!t towards linearity and one bond is stretching (Figure 16a), the A surface has no angular anisotropy at lopg range (Figure 16c) In contrast, for final diss.ciation on the X surface the rotation gained vzu motion on the B surface is augmented by further angular accelera_tion arising from the anisotropy of the conical feature on the X surface (Figure 16b) In either case a centrifugal barrier prevents the population of the highest OH levels which are energetically accessible at infinite separation of H from OH The overall population distribution for OH(X) is therefore the sum of two components, one peaking at a higher value of N" than the other Since the Coriolis mechanism depends on rotation of the parent molecule, the relative contri- butions of these two components will be proporkonal to the expectation value of (K,') for the excited H20 (B'A,) mole- cules, and will therefore vary with the temperature of the H,O sample On this basis it has been concluded that at the beam tempera- ture of cu 100-1 50 K used for the TOF experiment (at ?hi+ the value of (K,') = 2 5 for H20* and 3 7 for D20*) the B +A electronic Coriolis coupling makes a significant contribution to the overall dissociation [most obviously manifest by the shoulder near N" = 50 in the O,D(X,> product rotational distri- bution, Figure 151, but that the B +X vibronic coupling mecha- nism dominates Lowering the temperature of a thermal beam would lower the range of K, values excited, (K," -1) being approximately proportional to the absolute temperature Thus at a low enough temperature the Coriolis coupling would almost cease to operate, and the dissociation-to give ground-state OH would proceed almost entirely via the B --f X route and lead only to the A-doublets of e,A' symmetry 4 Dissociation by Internal Conversion The examples of photodissociation discussed above have the common feature that the product population distribution is dynamically controlled, with only a sub-set of the molecular degrees of freedom implicated in the dynamics This will tend to be the case where the surfaces are such that classical trajectories issuing from the Franck-Condon region cluster on an excited- THE DYNAMICS OF PHOTODISSOCIATION-R.N. DIXON B 'A, n 0 0.55 1.55 2.55 3.55 4.55 RIA X 'A, n 0 0 45 0.55 1.55 2.55 3.55 4.55 RIA state surface as they move out to dissociation, even when this motion involves a surface crossing. However, the concept of nuclear motion on potential energy surfaces is based on the Born-Oppenheimer approximation that nuclei are infinitely heavier than electrons, and thus move independently. All excited states are embedded in the upper reaches of the molecular ground-state and of any other lower states, and even a weak electron-nuclear coupling may mix levels of similar energies in different electronic states.In conse- quence, in large molecules with high level densities the excited states usually exhibit a rapid decay by internal conversion to the ground state, or inter-system crossing to a manifold of states of different spin multiplicity (e.g. triplet states), even though there may be no surface crossings to provide a direct pathway between the states. These decay routes can be expected to play an important role in any photochemistry by opening an indirect path to dissociation on a lower electronic surface, resulting in an energy disposal in the photo-products which may tend to be statistically rather than dynamically controlled. 4.1 The AIA"State of HNO Many of the levels of the A state of HNO are long-lived, with a radiative lifetime of ca.25 ps.The LIF excitation spectrum is highly structured, but exhibits a breaking-off in the band structure for all excited levels which lie above -16450 cm-l, except for a few levels with J' = 0. Direct dissociation on the A 'B, n 0 0.55 1.55 2.55 3.55 4.55 RIA Figure 16 Adiabatic components of an H,O wavepacket for K, = 2 resonantly driven at the energy of Lyman-a. 8 is the Jacobi angle between r(0H) and R(OH), with 8 = 0 for linear H-OH. In this coordinate system passage through linearity corresponds to reflection at 8 = 0 or 8=n.The wavepacket first evolves from the Franck- Condon region (filled ellipse) on the BIAl surface (a), from which dissociation results in the fragments H + OH(A). It then branches onto the XIBl ground-state surface (b) through vibro_nic coupling at a conical intersection (all values of KJ, and onto the A'B, surface (c) through electronic Coriolis coupling at linearity (K,>0 only), both leading to H + OH(X). For each component the wavefunction con- tours are drawn at lYmaxl/8,ratio chosen purely for clarity of presentation. Potential energy contours are separated by 5000 cm-', with selected indications of V/1000cm -l. excited surface at this energy is hindered by a substantial potential barrier, but predissociation nevertheless occurs to H(*S) + NO(X211).Below this limit there are numerous small rotational perturbations. In this example the levels with J' = 0 have odcj parity to inversion, and cannot mix with J"= 0 levels of the XlA' ground state which have even parity. Internal conversion is therefore forbidden at any energy for such levels. However, for each value of J greater than 0 there are levels of both parities in each state, so that there is no such strict symmetry constraint in the rotating molecule. Dissociation via internal conversion is promoted through an electronic Coriolis mechanism, and Figure 17 shows how the fluorescence quantum yield falls with increasing J-value in the 101-400 vibronic band. The observed fluorescence breaking-off limit for J' > 0 is the dissociation energy on the ground state surface, with J as the only controlling quantum number. The perturbations in the structured part of the spectrum arise through this same interac- tion mechanism.The dissociation products of HNO have not yet been charac- terized, but this same chromophore is present in all nitroso compounds. Larger RNO molecules have torsional vibrations in which some of the atoms move perpendicular to the CNO local plane of symmetry. These vibrations play the same role as toes rotation in HNO in promoting internal conversion from the A to the X state, but for all values of S including zero. Consequently all levels above the dissociation threshold lead to very rapid fragmentation and complete loss of fluorescence.NCNO disso- 384 3 2 1 pp,(J) I I I 16945 16955 ' cm Figure 17 The laser-induced fluorescence excitation spectrum of ambient temperature HNO in a band lying above the dissociation threshold to H + NO The rapid drop in intensity with increasein J' is a consequence of rotationally induced internal conversion to the continuum of the ground state, followed by dissociation (pP,(1) is the only transition that terminates on J' = 0 ) ciation has been studied in detail, with particular attention to energies near the dissociation threshold This provides a good example of a statistical rotational energy disposal in the disso- ciation products, although when a vibrational channel becomes open at higher energy the product vibration is not statistically equilibrated with rotation 4.2 The A'T, State of CH, In contrast to HNO, the first excited state of methane is orbitally degenerate The photodissociation of methane at 121 6 nm has been studied using the technique of H-atom photofragment translational spectroscopy It was concluded from the analysis of the TOF spectrum that simple C-H bond fission is the dominant primary process at this energy The resulting CH, fragments are formed with very high levels of internal excitation, such that some 25% possess so much energy that they undergo subsequent unimolecular decay The experiments do not provide a unique determination of this secondary decay process, but arguments based on unimolecular rate theory suggest that predominantly it will yield CH and H, fragments The energy dependence of the TOF spectrum up to the onset of the CH: unimolecular decay can be simulated with a statistical model which uses the density of states for the six (anharmonic) vibrational modes of CH,, but ignoring rotation (Figure 18) Above the dissociation limit for CH: +CH + H, the secondary decay rate estimated using RRKM theory rises 10 -08 c 3 g 06 TKER rnax a H+CH+H, Y -2 04 0, TKER rnax v) 02 0 0 10000 20000 30000 40000 TKER / cm-I Figure 18 The spectrum of total kinetic energy release (***) recorded with H atoms from photolysis of CH, at 121 6 nm, and its statistical simulation (-) Area A is the calculated vibrational density of states for CH, from CH, --+ H + CH, up to the dissociation limit D,(CH-H,) Abovethis limit (areaB) rapid dissociation of CHilimits the available density of states Area Cis attributed to direct three body dissociation into H + CH + H, CHEMICAL SOCIETY REVIEWS 1994 very rapidly It has then been assumed that departure of the TOF profile from the predictions of the H + CH, statistical model comes about because simultaneous dissociation of CHq* into H + CH + H2 takes over from the sequential dissociation, leading to a modification of the relevant density of states It is interesting to enquire as to why this dissociation of methane is essentially statistical, whereas that discussed above for ammonia is dynamically controlled An important difference is that the 'T, excited electronic state of CH, probably suffers from substantial Jahn-Teller distortion Furthermore, both of the triply degenerate vibrational modes v3 and V, are of appro- priate symmetry not only to induce Jahn-Teller distortion of the excited state, but also to promote vibronic coupling between the excited state and the ground state We do not believe that the good agreement with the results of a statistical calculation necessarily implies that the time scale of the CH, dissociation is sufficiently long for complete randomization of all modes in the photoexcited molecule, but rather that the strong Jahn-Teller distortions and vibronic couplings that occur whilst the mole- cule evolves from the Franck-Condon region of the excited state into the H + CH, exit channel of the ground state lead to population of a very wide spread of product vibrational states 5 Discussion The molecular systems discussed above have been chosen to illustrate the variety of dissociation mechanisms and energy disposal in small polyatomic molecules An important question that follows is whether we are yet in a position to predict the likely outcome of any given photodissoclation process In favourable cases the relevant potential energy surfaces may be available from ab znztzo calculation or semi-empirical modelling Alternatively, prediction may be based on analogies with weil- studied systems Whether dissociation will take place on one, or more than one, surface will depend critically on both the energies and associated internal coordinates of any surface crossings, on the strengths of their associated inter-state couplings, and on the Frank-Condon region of coordinates from the initial state For example, in H2S there is a conical intersection of the first two excited surfaces close to the Frank-Condon region for excitation from the ground state With stretching of either bond one of these surfaces correlates smoothly with the known ground state products H + SH (or HS + H) If, however, one H atom is substituted by ki methyl group the symmetry is lowered, such that the C-S bond strength is substantially lower (3 15 eV) than that of the S-H bond (3 75 eV), and the crossing of the corresponding surfaces is no longer in the Franck-Condon region Dissociation in the wavelength range 275-220 nm via the first excited llA" state results in fission of the stronger S-H bond Shorter wavelengths favour population of the second excited 2lA" state, which is bound with respect to both S-H and C-S bond fission, but has an extended C-S bond length Dissociation via this second excited state is believed to involve C-S bond stretching, followed by a crossing to the first excited surface which is unbound with respect to both CH,S + H and CH, + SH at these energies Experiment shows that the latter fragmentation channel becomes increasingly important at exci- tation wavelengths shorter than 220 nm 2o These, and other related, observations can be rationalized in the light of recently reported ab znztzo surfaces, but would have been hard to predict by simple analogy In contrast, the potential energy surface for the diss_ociation of the first _excited state of the methyl radical, CH3(B2Ai) -,H + CH2(A1Ai), must be crcssed at long range 1," its exit channel by the surface for CH,(X2B,) -+ H + CH,(X3B,), with the possibility of coupling through vibronic or Coriolis forces as in ammonia 21 Even so, the adiabatic route is found to be domi- nant, so the coupling is presumably too weak to be effective 22 (An earlier contradictory observation2 was incorrectly interpreted ) A second aspect concerns the observation that a statistical population distribution in the products is a characteristic of THE DYNAMICS OF PHOTODISSOCIATION-R N DIXON dissociation via internal conversion, especially with increasing molecular size But consider now the dissociation of deuterated ammonia m_olecules With ND, the outcome of dissociation through the A state is very similar to that described in Section 3 2 for NH,, apart from a closer energy spacing because of the increased mass For NHD,, TOF spectra have been recorded for both H and D atom products, with a branching ratio of about 5 1 The H-atom TOF spectrum again shows the simple regular structure associated with a-axis rotation of ND,, but that of the D-atom is densely structured, almost continuous, and unassign- able The constrast between one dynamically controlled mechanism, and a second apparently statistically controlled mechanism in competition, indicates that this must be an exit channel effect The most likely explanation is that the torque on the NHD product generated by a breaking N-D bond is perpendicular to that bond, and thereby makes an angle of about 18" to the a-axis of NHD The resultant tumbling motion then leads to population of most of the energetically accessible rotational states for this channel only The size of molecules for which ab znctzo calculations of potential energy surfaces are feasible is increasing rapidly This has already proved its value in a molecule such as CH,SH, where the number of active coordinates is restricted to the C-S-H backbone However, in general it is at present premature to expect accurate predictions of the dissociation dynamics of most polyatomics in advance of experiment Photodissociation has often been referred to as a half-colli- sion A full bimolecular reaction can be considered as two such half-collisions back to back with a range of close encounters The experimental and theoretical developments that have been so fruitful in studies of gas-phase photodissociation are now being applied to bimolecular reaction dynamics, and are also finding applications in such fields as the photochemisty of molecules on surfaces Acknowdedgements I am indebted to many gifted graduate students and other collaborators who have contributed to this work, and in particular to my colleagues Professor M N R Ashfold, Drs G G Balint-Kurti and C M Western, and Mr K N Rosser Support from the S E R C for this research is also gratefully acknowledged 6 References 1 G Herzberg, 'Molecular Spectra and Molecular Structure I11 Electronic Spectra and Electronic Structure of Polyatomic Mole- cules', Van Nostrand, Princeton, 1966 2 P Andresen, G S Ondrey, B Titze, and E W Rothe, J Chem Phvs , 1984,80,2548 3 K -H Gericke, S Klee, F J Comes, and R N Dixon, J Chem Phvs , 1986,85,4463 4 R N Dixon, J Chem Phjs , 1986,85, 1866 5 R Vasudev, R N Zare, and R N Dixon, J Chem Pht F , 1984,80, 4863 6 R N Dixon and H Rieley, J Chem Phjs , 1989,91,2308 7 R N Dixon, Trans Faruday Soc , 1969,65,3141 8 R N Dixon, J Chem SOC Faradaj Trans, 1992,88, 2575 9 R N Dixon, MoI Phvs , 1985,54,333 10 S H Kable, J -C Loison,D W Neyer,P L Houston, I Burak,and R N Dixon, J Phys Chem , 1991,95,8013 11 S H Kable, J -C Loison, P L Houston, and I Burak, J Phvs Chem , 1990,92,6332 I2 J Biesner, L Schnieder, J Schmeer.G Ahlers, Xiaoxiang Xie, K H Welge, M N R Ashfold, and R N Dixon, J Chem Phjs , 1988,88, 3607 13 J Biesner, L Schnieder, G Ahlers, Xiaoxiang Xie, K H Welge, M N R Ashfold, and R N Dixon, J Chem Phys , 1989,91, 2901 14 M N R Ashfold, R N Dixon, S J Irving, H -M Koeppe,, W Meier, J R Nightingale, L Schnieder, and K H Welge, Philos Trans Roy Soc London A, 1990,332,315 15 H J Krautwald, L Schnieder, K H Welge, and M N R Ashfold, Farad Discuss Chem SOC, 1986,82,99 16 D H Mordaunt, M N R Ashfold, and R N Dixon, J Chem Phbs , 1994, 100,7360 17 R N Dixon, K B Jones, M Noble, and S Carter, MoI Phjs , 1981, 42,455 18 I Nadler, M Noble, H Reisler, and C Wittig, J Chem Phys ,1985, 82,2608 19 D H Mordaunt,I R Lambert,G P Morley,M N R Ashfold,R N Dixon, C M Western, L Schnieder, and K H Welge, J Chem Phrs , 1993,98,2054 20 S H S Wilson, M N R Ashfold, and R N Dixon, J Chem Phys , 1994, in press 21 S H S Wilson, M N R Ashfold, and R N Dixon, Chem Phjs Lett, 1994,222,457 22 S H S Wilson,J D Howe, K N Rosser, M N R Ashfold,and R N Dixon, Chem Phjs Lett, 1994,227,456 23 D H Mordaunt, Ph D thesis, University of Bristol, 1994, D H Mordaunt, R N Dixon, M N R Ashfold, and L Schneider, in preparation

ISSN:0306-0012    

DOI:10.1039/CS9942300375

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Molecular Mechanics Force Fields for Cyclopentadienyl Complexes B. Bosnich Department of Chemistry, The University of Chicago, 5735 South Ellis Avenue, Chicago, Illinois, 60637, U.S.A. 1 Introduction the early work of Allinger4 is notable for it is now one of the most The notion that molecules might be treated as ball and spring comprehensive programs for organic molecules. In contrast to organic force fields, those involving inorganic objects has pervaded chemical thinking from the time that molecular geometry was recognized. Although primitive at first, the idea that stretching bonds and bending angles from their ’natural’ lengths and angles resembled deformations of a spring was widely accepted. Moreover, it was noted that atoms of a molecule occupied a volume and may carry partial charges.Consequently, these atoms were not expected to penetrate each others volume and their positions could be governed to some extent by charge-charge interaction. It was therefore logical to conclude that bond length and angle deformations, steric and charge interactions would raise the energy of a molecule com- pared to an analogous molecule possessing fewer deformations and interactions. It could then be said that one molecule was more ‘strained’ than the other. If one were able to calculate the energy of each of the components contributing to the strain one would have a quantitative measure of the total strain energy. Implied in this quantitative model is the expectation that variations in structure could be determined by finding the least strained structure of the molecule. This promise of calculating the strain energy and structure of molecules had to await the implementation of a suitable mathematical framework for assessing strain, of methods for finding the least strained struc- ture, and the advent of fast computational machines.The formal basis for what is now called molecular mechanics derives from ideas of Andrews’ and Hill;2 but it was Wes- theimer,3 while at The University of Chicago, who first imple- mented a molecular mechanics force field for the racemization of certain hindered biphenyls. In his considerations, Westheimer assumed that hindered biphenyl racemization was basically governed by non-bonded interactions and by bond strain along the rotational trajectory of the rotating phenyl groups.This seminal work demonstrated the utility of the empirical molecu- lar mechanics method and the development of the field has continued so that, for organic molecules, the method is routinely used. Although a number of force fields have been developed, Brice Bosnich Mias born in Australia, received his undergraduate degree at The University ofsj’dney and his Ph.D. from the John Curtin School of Medical Research in Canberra. He then jour- neyed to University College, London Mihere he wws u postdoctoral fellow- and eventuallj’ a lecturer. He then moved to The Universitj. of Toronto and,Jinally, to The University of Chicago in 1987 \i,here he is noit. a Projessor of Chemistrj?.His research encom- passes a diverse runge of interests which include, the development of’ asjwmetric cataljxts and studies of their mechanisms, synthesis, and reactivity of organometallic and coordina- tion compounds and occasional ,jbrays into quasi-theoretical studies such as those described here. A constant thread bt’hich has characterized his wwrk is an interest in stereochemistry, particularly the chirality qf molecules. He is the 1995196 Nyholm Lecturer. This article is dedicated in memorjj of Ron NjTholm. complexes are less well developed -particularly those that attempt to model organometallic n-complexes. Starting with the work of Hawkins5 and Snow6 there now exists a reliable body of empirical information which can be used to reproduce structures of first row transition metal coordination compounds contain- ing nitrogen and oxygen donor atoms.These coordination compound force fields’ were derived on a less formal basis than their organic counterparts. Force constants were derived by trial and error and certain arbitrary assumptions about non-bonded interactions were incorporated. Even so, for practical purposes, the methods work well if only the reproduction of structure is required. Until very recently, very little attention has been given to transition metal organometallic complexes, particularly those incorporating n-ligands. These systems are conceptually more difficult to handle because of the peculiarity of the topologies of binding.These require special treatments. This review describes our attempts to develop self-consistent force fields for organometallic n-complexes. We deal with linear metallocenes of the type [M(Cp’),] and bent metallocenes of the type, [M(Cp’),X,] where Cp’ is a generic cyclopentadienyl ligand which may or may not be substituted. In addition we illustrate by molecular mechanics why certain [M(Cp*)J, Cp* is the pentamethylcyclopentadienyl ligand, complexes of alkaline earth and lanthanide elements are bent instead of being linear as intuition might suggest. 2 The Force Field The total molecular mechanics energy, ET,is given by the sum of the energies associated with bond stretching, Eb,angle bending, E,, torsional deformations, E,, and van der Waals interactions, -bVDW-In these equations kb, k,, k, are the respective force constants, r, and 8, are the ‘equilibrium’ bond lengths and angles, respecti- vely, n is the periodicity and 19 is the angle.In the Lennard-Jones VDW equation, is the well-depth of the ithand jth interactions, ro = +(rll+ rJ1)and r,, is the internuclear distance. It will be noted that the harmonic Hooke’s law is used for the bond and angle terms and the truncated Fourier expression is used for dihedral terms. More realistic and fancier equations can be substituted but experience suggests that these simple representations suffice. Most organic atoms have been assigned van der Waals para- meters which adequately generate the energies of these non- bonded interactions.Parameters for metals are not known but we have found that assigning values of -0.001 kcal/mole for E and r, = 1.O A for all metals leads to satisfactory results. Chang- ing ro to 2.0A and E to -0.01 kcal/mole did not affect the structures calculated. In many force fields electrostatic interac- tions are added but this requires the ability to calculate partial charges on each of the atoms and a knowledge of the micro- scopic dielectric constant. For organometallic complexes there is no reliable method of calculating partial charges with any accuracy. The electrostatic contribution is therefore omitted. At this stage it is important to note that the energy associated with the first three terms assumes that an equilibrium value can 387 388 CHEMICAL SOCIETY REVIEWS 1994 be determined In other words, there exist bond lengths and C1 D c2 angles which are ‘natural’ for the system in question, so that when Y = ro and 6 = 6, there are no contributions to the strain energy To illustrate the problem consider the angle bending term in 5-coordinate complexes For distorted 5-coordinate structures, do we assume that the final structure is derived from a trigonal bipyramid or square pyramid structure? Similarly, in the present context, we might debate the ‘natural’ angles in [M(Cp),X,] complexes This dilemma is neither commonly recognized nor is it a trivial problem in molecular mechanics calculations We require some formal method of determining ‘equilibrium’ parameters Clearly, the final minimized structure will depend critically on the equilibrium parameters as well as on the associated force constants So will the energy In this respect one should be circumspect in using some of the commercial programs which attempt to derive structures of organometallic compounds Their ‘success’ is based on selecting equilibrium bond lengths and angles which are essentially the same as the final structure The structure is driven to the desired result by application of impossibly large force constants In a sense these programs are almost self-fulfilling One might say that they are ball and stick rather than ball and spring representations 3 A Molecular Mechanics Topology for Organometallic n-complexes Organometallic complexes bearing 7-r-bonding ligands present special challenges in defining the topology For example, in the case of an olefin complex, do we define the topology as consist- ing of two bonds to the metal, one for each of the carbon atoms, or do we make one bond to the centroid of the double bond7 In the case of ferrocene, do we make ten bonds orjust two, one each to the two ring centroids? Although not insurmountable, the topology defined by ten bonds in ferrocene presents difficulties in defining the molecular mechanics force field because of the presence of contiguous three-membered rings One can simplify the topology by defining a dummy atom (D) at the ring centroids and applying appropriate force constants to the bonds and angles associated with it Such a topology, although a fiction, can be used to reproduce structures but it has difficulties The first is that unrealistically large force constants are needed to maintain the dummy atom at the centroid during minimization The second is that spurious vibrational modes are generated when the force field is used to reproduce the vibrational spec- trum Since we use vibrational spectra to derive self-consistent force constants, this scheme will not generate the proper force field The use of the dummy atom topology is simple and appealing, however, and we have devised a way of retaining its simplicity but at the same time constructing a physically realistic model of the forces on the molecule This can be illustrated by the simple example of ethylene 7-r-bonded to a metal (Figure 1) where C, and C, are the ethylene carbon atoms and D is the dummy atom placed at the centroid in this case, but, in what follows, it is not required that it be at the centroid, M is the metal The x-coordinate of the dummy atom X, is defined by the x-coordi- nates of the carbon atoms, Xc, and Xc, as follows where XMis the x-coordinate of the metal For the M-D stretch the energy, E, is E = kb(Y -and dE/& = 2kb(r -r,j also dE/dXD = (dE/drj(dr/dXDj= [2kb(r -v,)][l] = 2kb(r -Y,) The force on the dummy atom can be transferred to C, by application of the chain rule 0-0-I 0 -Y Ir Figure 1 Thus half of the force on the dummy atom has been passed on to C, Similarly, the other half can be passed on to C,, leaving no forces on the dummy atom The second derivative needed for the vibrational analysis can be transferred to the carbon atoms in a similar way It is clear that this method can be applied to more complex systems Thus the simple dummy atom topology can be used to provide a rigorous description of the forces We note that during minimization the position of the dummy atom is recalcu- lated after each step 4 Derivation of Force Constants In order to derive a force field one requires to know force constants and their corresponding equilibrium parameters and parameters for the VDW non-bonded interactions The VDW parameters have been developed for most of the atoms of concern here, except for the metals The procedure for finding the force constants and equilibrium values is an iterative process and relies on no assumptions about bonding, although bonding schemes can be used as a guide The first step is to select a parent molecule of known structure This molecule must contain all the elements we wish to define for the force field and which is considered to be the least strained Thus we would select a [M(Cp),] complex for the linear metallocenes and [M(Cp),Cl,] for those said to be bent The assumption is that substitution of the Cp rings will lead to more strain than is present in the parent Cp-containing molecule Unless we are to make assumptions based on bonding schemes we usually know only two things about the parent molecule, its structure and its vibrational spectrum We require that the (approximate) molecular mech- anics force field reproduce both the structure and the vibrational spectrum using a set of force constants, equilibrium bond lengths and angles, and VDW parameters Although there are a number of approaches to this problem, we have adopted the following Having chosen a parent molecule, each of the bonds and angles is assigned a force constant and an equilibrium value The choice of the force constant is basically an educated guess initially and the choice of the initial set of equilibrium para- meters can be based on the crystal structure values or based on some supposed bonding scheme A computer-driven routine can be implemented which, with the given set of force constants, searches for equilibrium values which will minimize to the known structure The molecular mechanics force field is then used to generate the vibrational spectrum This is compared with experiment, adjustments are made to the force constants and the process is repeated until the structure and the vibrational spectrum is reproduced This procedure gives us a set of self- consistent force constants and equilibrium values For many molecules this process provides a unique solution, but not always For example, with the [M(Cp),Cl,] complexes there are many solutions to the skeletal bond angle equilibrium values These solutions all reproduce the structure and vibrational spectrum with the same set of force constants The minimized energies, of course, are all different Requiring, by definition, that the parent molecule have the lowest strain energy, a decision on the preferred equilibrium parameters can be made by the requirement that they provide the lowest minimized energy MOLECULAR MECHANICS FORCE FIELDS FOR CYCLOPENTADIENYL COMPLEXES-B BOSNICH 5 Force Constants For the force constants of [M(Cp),] and [M(Cp),Cl,] com-plexes, it is convenient to divide them into those associated with the Cp rings and those related to the skeletal deformations involving metal-ligand bonds This separation is an approxima- tion, but a good one, because the internal ligand modes are well separated in energy from the skeletal modes We found that to a very good approximation the internal Cp force constants can be transferred from metal to metal but the skeletal force constants are strongly metal-dependent For linear metallocenes four skeletal force constants are required in the dummy atom formu- lation, (1) and (2) They involve Cp tilting, y, metal angle bending, ,8,metal-dummy stretching, a, and dihedral rotation of the Cp rings, a There is a similar but more extensive set of skeletal force constants for the [M(Cp),Cl,] systems The dihed- ral rotation force constant was found to have a small force constant for linear metallocenes of the transition metals but, for [M(Cp),CI,]M = Ti.Zr, Hf systems, we have demonstrated that no dihedral force constant for the Cp ring rotations need be used Although the process of arriving at the force constants can be tedious they are self-consistent because they are based on the vibrational spectra N For the calculations described here and for the derivation of the force constants the CHARMM9 suite of programs was used It was interfaced with CHEM-X1 O which was used as a graphics and input front end The CHARMM program was modified to accommodate our force field formulation of organometallic T-complexes Parameters for organic groups attached to the Cp rings were those included in the standard CHARMM parameter list 6 Linear Metallocenes of the Transition Metals The structure of [Fe(Cp),] has a history of controversy' related to relative rotation of the Cp rings Both X-ray and neutron diffraction analyses of the crystals indicate disorder even at low temperatures Gas phase electron diffraction indicates that the ferrocene molecule prefers to adopt an eclipsed conformation with a small internal rotational barrier of about 0 9 kcal/mole From vibrational data we calculated the barrier to be 0 7 kcal/ mole A prevalent supposition favouring the staggered confor- mation is that the non-bonded interactions are greater in the eclipsed form Molecular mechanics can resolve this issue Table 1 contains the energies associated with the various terms in the two forms The total energy difference between the two forms found by molecular mechanics (-0 8 kcal/mole) is consistent with experiment Whereas it is true that VDW interactions are slightly less for the staggered form, the major difference resides in the torsion term which is equal to that derived from vibratio- nal data for ring rotation Thus molecular mechanics provides a clear demonstration that the impediment to Cp rotation in Table 1 Energy terms (kcal/mole) of the eclipsed and staggered forms of ferrocene Bond Angle Torsion VDW Total Eclipsed 0079 0000 0004 -5225 -5 142 Staggered 0073 0000 0 724 -5251 -4454 ferrocene is electronic, rather than steric, in origin Both ruthe- nocene and osmocene are found to exist in the eclipsed form in the solid l2 From the vibrational data of ruthenocene we calculated a rotational barrier of 6 8 kcal/mole A thermal motion study of crystalline ruthenocene estimates a barrier of 8 I kcal/mole No data are available to estimate the barrier for osmocene but we suspect it is higher As for ferrocene, molecular mechanics indicates that the rotational barrier of ruthenocene is electronic in origin The barrier for [Co(Cp),] was calculated' + to be the same as ferrocene suggesting that the rotational barrier increases for the higher mass elements of the same electronic configuration The small barrier in ferrocene suggests that substitution could lead to a preference for the staggered conformation This is the case for the decamethyl derivative, [Fe((CH,),Cp),] which in the crystal is perfectly staggered (a = 36") l4 Using the crystallo- graphic coordinates, gas phase electron diffraction indicates that the molecule is staggered with a rotation barrier of about I kcal/ mole Upon minimization, we find that [Fe((CH,),Cp),] has a Structure which is partially staggered (a = 18") Two perspec- tives of the minimized structure are shown in (3)and (4) It will be noted that the orientations of the methyl group hydrogen atoms are such that, in both rings, one hydrogen atom of each methyl group lies approximately in the Cp plane, the other two hydrogen atoms lie above and below the Cp plane Further, if we refer to the in-plane hydrogen atoms as the 'head', one ring has a head-to-tail clockwise sequence whereas the other is oriented head-to-tail anticlockwise In the crystal the hydrogen atoms orientations are more complex with a number of the hydrogen atoms disposed perpendicular to the Cp plane away from the metal On the assumptions that, first, the hydrogen atom conformations can affect the stagger angle and, second, that crystal packing forces can affect the hydrogen atom orien- tations, molecular mechanics calculations were performed on [Fe((CH,),Cp),] in its crystal environment The resultant struc- ture was perfectly staggered (a = 36") but the individual methyl group hydrogen atom orientations were the same as shown in (3) and (4) except that the hydrogen atoms were head-to-tail in the same direction for both rings Although there is a cluster of these conformations of similar energy, the energy of the crystallogra- phically found conformations is about 5 kcal/mole less stable than that of the minimized structure We noted earlier that the rotational barrier of ruthenocene was much higher than that of ferrocene and it might be that, for example, [Ru((Cl),Cp),], exists in an eclipsed conformation This is found to be the case both by experiment and by the present calculations Unfortuna- tely, the [Ru((CH,),Cp),] molecule is disordered in the solid state and the conformation is uncertain (3) (4) So far the problem of the rotamers of the linear metallocenes has been compared with experiment These questions do not test the validity of the derived skeletal force constants because little skeletal strain is contained in these molecules The two strapped ferrocenes, (5)' and (6)' 6, do have skeletal strain, particularly the latter For (6) we developed a number of spectroscopically based silicon parameters We show the superimposed structures of the calculated and observed structures in (7) and (8) It will be seen that the fit is almost identical in both cases giving assurance PhpSi (5) Q-(7) (9) that the force field is a good one and that the methods described can be extended to other systems.7 Structures of Bent Metallocenes [M(Cp*),] We now turn to a structural feature associated with certain [M(Cp*),] complexes which has been the subject of extensive debate. It was discovered that [M(Cp*),] (M = Ca, Sr, Ba, Sm, Eu, Yb) complexes did not possess parallel Cp* ring dispositions as is found in, for example, [Fe(Cp*),]. Rather, the rings were tilted, (9).17 This is not a consequence of crystal packing, although packing effects can alter the tilt angle. A number of these molecules were found to be bent in the gas phase.l8 The [Mg(Cp*),] complex, however, contains parallel rings as does its parent, [Mg(Cp),].The other alkaline earth and lanthanide metals give polymeric structures in the solid state when Cp is incorporated. Before these so called bent metallocenes were discovered it was known that [M(Cp),] complexes of Ge, Sn and Pb were bent and the facile assumption was made that a 'stereochemically active' lone pair of electrons was responsible for the tilt of the Cp rings. The [Sn(Ph,C,),] complex was found to be linear, however, and it was assumed that steric repulsion emanating from the phenyl groups led to the linear structure.Similarly, the complex, [Si(Cp*),] exists as both a linear and bent structure in the solid. These two forms presumably reflect the exigencies of crystal packing. Given the inter- or intra-molecular CHEMICAL SOCIETY REVIEWS, 1994 steric effects can result in a linear structure one is left questioning whether the putative lone pair of electrons is really responsible for the bent structures observed with the Si, Ge, Sn and Pb complexes. It is possible that these complexes belong to the same category as the alkaline earth and lanthanide complexes. The existence of the bent metallocenes provoked numerous theoretical studies' 9~20seeking an electronic explanation for the bent structures. None was forthcoming, except that there appears to be an electronic component which may induce the Mg complexes to be linear.20 Given the flexible predictive power of theoretical calculations on molecules of such complexity it seemed prudent to search elsewhere for an explanation. It was noted2' that in the bent [M(Cp*),] metallocenes the shortest interligand methyl group VDW contact distances were constant (-4.1 A).Consequently, the tilt angle p, (9), decreases with increases in metal radius. The possible implication was that the tilt in these complexes was governed by VDW attractive forces of the ligands. It will be recalled that the VDW expression has both an attractive and repulsive part and that in the absence of other restraints an assembly of atoms will rest in an energy well.What is counterintuitive about this explanation is the expectation that a tilted unsymmetrical structure rather than a symmetrical topology will be produced. This hypothesis can be tested by the molecular mechanics force field derived here. In order to derive a generic force field for these complexes we used the vibrational data of the [Mg(Cp),] complex which gave force constants for the M-D stretch and for the y-bending term, (1). The D-M-D (6)bend force constant was set to zero in order to determine how the VDW interactions would affect the tilt of the Cp* rings. No force constant was applied to Cp* rotation. The results are collected in Table 2.22The symbols Y,, px, a, refer to bond lengths and angles found in the crystal, rg,psrefer to those in the gas phase.Where two entries appear, they refer to different molecules in the unit cell of the crystal. The energy difference, in kcal/mole, dET and dEvDw refer to the total and VDW energies, respectively, and LIE refers to the difference; E for the bent, minus E for the linear forms. Thus the bent form is more stable than the linear form in all cases. (For the Mg complex the difference is trivial compared to the thermal energy at 25 "C.) Inspection of Table 2 reveals a remarkable agreement with experiment when potential deviations due to crystal pack- ing are considered and when the assumptions used in the gas phase electron diffraction analysis are recognized. Further, the difference in energy between the bent and linear forms is almost totally accounted for by the differences in VDW energy (LIEvDw),confirming the original hypothesis.These small energy differences indicate that these are floppy molecules which have ready access to both linear and bent forms at 25 "C. Two other features should be noted. First, the larger the metal radius the greater the tilt, that is the ,%angle decreases with increase in radius. This is expected if VDW attractive forces are the important contributor to the structure. Second, the VDW energy differences, dEvDw, increase in magnitude with increase in metal radius. This is because the VDW energy decreases in the linear form with increase in metal radius. Table 2 Calculated and observed bond lengths and angles and the energy differences between the bent and linear forms of [M(Cp),*I complexes Bond Length Tilt Angle Stagger Angle Energy Difference (4 ("1 ("1 (kcal/mole) rCdk rx rg Bcalc Px Bg acdc a, A ET A EVDW Mg 2.00 2.02 172 180 20 -0.08 0.10 Yb 2.33 2.33 153 158 22 -1.21 -1.07 Ca 2.35 2.33,2.36 2.31 152 146, 148 154 19 19,25 -1.29 -1.15 Sr 2.48 2.47 147 149 18 -1.65 -1.58 Sm 2.52 2.53 145 140 18 19 -1.74 -1.68 Eu 2.52 2.53 145 140 18 19 -1.74 -1.68 Ba 2.73 2.70,2.78 2.63 138 131 148 26 32,28 -2.03 -2.10 MOLECULAR MECHANICS FORCE FIELDS FOR CYCLOPENTADIENYL COMPLEXES-B BOSNICH 391 If, as it appears, VDW attractive forces control the topology of these metallocenes it is probable that with appropriate Cp substitution there will exist linear forms of these metals We calculated that the bis-penta-iso-propylcyclopentadienylcom-plexes of all of the metals listed in Table 2 should be linear If these can be made, it will be interesting to see if this prediction is correct It should be noted, however, that full iso-propyl substi- tution is required to obtain the linear forms by our calculations The [Ba((Pr,),C,H,)2] complex is found to be bent both by experiment and by our calculations Returning to the bent metallocenes of Si, Ge, Sn, and Pb, for which stereochemically active lone pair electrons are invoked in order to explain the topology, our calculations predict that these molecules will be bent because of VDW attractive forces in both the Cp* and Cp complexes Although these calculations do not preclude the stereochemical effect of lone pair electrons, they do indicate that the structures can be explained without invoking them The results of the molecular mechanics calculations on these bent metallocenes demonstrate the usefulness of the method Without this technique it would not be possible to define the 'mechanical' forces which lead to the bent structures Moreover, it gives pause to explanations of these structural features on the basis of molecular orbital calculations V 8 Structures of [M(Cp),X,] Complexes Derivatives of complexes of the types, [M(Cp),Cl,](M = Ti, Zr, Hf) are finding increasing application in stereoselective transfor- mations 23 Currently the chiral complexes of the l~gand,~ ethy-lene-l ,l'-bis(tetrahydroindenyl), (lo), have proven to be the most effective precursors for these stereoselective transforma- tions Perhaps the most impressive discovery was that, after suitable activation, the Zr complex of (10) was an excellent catalyst for the isotactic polymerization of propylene 25 The mechanism of this stereoselectivity is believed to occur by successive head-to-tail insertions of the propylene molecules as shown in Scheme 1 where the mean molecular planes of the two tetrahydroindenyl groups are shown as bars and Pis the polymer chain The scheme, as drawn, shows that each successive inser- tion involves the same (prochiral) face of propylene so that the pseudo-chiral centres of the product polymer are all of the same configuration It is supposed that the face selection is to some extent governed by the chiral disposition of the tetrahydroinde- nyl groups In the scheme this is implied by the assumption that the methyl group of propylene experiences less steric interaction in the shown orientation than when the olefin face is reversed leading to methyl group interaction with the lower drawn tetrahydroindenyl group The selectivity almost certainly encompasses more steric and orienting effects than is implied in the simple scheme but ultimately the isotacticity of the product can be traced back to the chiral structure of the catalyst Thus the first prerequisite to delineating the origins of the stereoselec- Scheme I tivity provided by these catalysts and to designing new catalysts is a precise understanding of the steric interactions that can occur This is provided by molecular mechanics In this section we do not attempt to give an explanation for any stereoselective reaction Rather we develop an accurate force field for these complexes which is necessary before such explanations can be at tempted Table 3 contains structural data for a number of complexes for which the crystal structures have been reported, superimpo- sitions of calculated and crystal structures are shown In the jargon of the trade such presentations are called 'rigid fits' where the superimposition is made by selecting the coordinates of a number of key atoms of each of the molecules and then superimposing them by assigning different weights to the atom coordinates For the present purposes the two dummy atoms (Cp centroids), the two chloro ligands, and the metal were rigidly fitted with the metal being assigned ten times the weight over the other atoms Cyclopentadienyl ligands are designated as Cp for an unsubstituted Cp, Cp* for pentamethyl Cp, and Cp' means a generic substituted Cp ligand The angle designation DCp*-CCp*-CMerefers to the out-of-plane angle formed by the carbon atom of the methyl group substituent attached to the Cp* ring Positive values indicate that the methyl group is tilted away from the metal The Cp' ligands can rotate with respect to Table 3 Calculated and observed structures of [M(Cp'),CI,] Rigid Fit Geometry Calculated Observed 2 07A 2 07A 2 09A 2 l0A 2 37A 2 35A DCp-T1-DCp* Dc,-TI-Cl 133" 105" 132" 105" Dcp*-TI-Cl Cl-TI-Cl 107" 94" 107" 94" DCp*-CCp*-CMeCCp-DCp-DCp*-CCp* BCl-Ti-DCp*-C,-p* Bcl-T1-Dcp-Ccp 5"--10" 29" 13" 14" 4"-8" 39" 0" 36" 2 13A 2 358, 138" 104" 93" 3"- 16" 34" 0", 32" Hf-Dcp 2 19A 2 18A Hf-Dc,.2 20A 2 198, Hf-Cl 243A 241A Dcp-Hf-Dcp* 132" 131" Dcp-Hf-C1 105" 105" Dcp*-Hf-Cl 107" 107" C1-Hf-C1 94" 96" DCp*-CCp*-CMe 4"-7" 3"-7" Ccp-DCp-DCP*-CCp* 27" 38" Bcl-Hf-DC,-Cq 14" 35" Bcl-Hf-DcP*-Ccp* 11" 0" Zr-Dcp 2218, 221A Zr-Dcp. 2 22A 2 22A Zr-C1 244A 2 44A Dcp-Zr-Dcp. 131" 130" Dcp-Zr-CI 105" 105" Dcp*-Zr-C1 107" 107" C1-Zr-C1 97" 98" DCp*-CCp*-CMe 3"--7" 4"-6" CCp-DCp-DCp*-CCpi 28" 39" Bc, -Zr -DCp-CCp 14" 36" BCI-Zr-DCpl-CCp* 12" 0" each other and with respect to the two chloro groups The rotations are defined in terms of (four-atom) dihedral angles The relative orientation of the Cp' rings with respect to each other is defined by the dihedral angle Ccp-DCp -DCp -Ccp and the smallest angle is quoted The Cp' orientation with respect to the two chloro groups is defined by the sequence BCI-M-DCp- Ccp where Bcl is a point which is in the Cl-M-CI plane and bisects this angle It has two values, one for each Cp' ring, and can be positive (clockwise) or negative (anticlockwise) The smallest angle is quoted Inspection of Table 3 reveals that our force field reproduces the skeletal bond lengths and angles with a high degree of precision The tilting of the methyl groups out of the Cp* plane is also well reproduced Although some of the Cp' dihedral angles show a good correspondence between the observed and calcu- lated values, a number of these angles do not match well Aside from crystal packing forces which could affect the Cp' dihedral orientations there are two other possible sources for the disparity between the calculated and observed structures That crystal packing effects are real, and not an all-purpose incan- tation, will be demonstrated presently The first possible reason for the dihedral differences could be because we did not include dihedral force constants in our calculations That these force constants are zero or nearly zero can be demonstrated in a number of ways Perhaps the most persuasive was the minimiza- tion of [Ti(Cp*),CI,] where the coordinates of all of the atoms of the Cp* rings were fixed to their crystallographic positions but the rings were allowed to rotate without dihedral restraint and the derived force constants were applied to the skeletal modes The resultant minimized structure was identical to that observed in the crystal This observation indicates that the orientations of the Cp* ligands is essentially governed by intramolecular non- bonded interactions and, parenthetically, that crystal packing forces do not control the Cp* orientations in this case If this be so, why is it that the Cp' orientations are not always well reproduced? The major reason for the lack of correspondence for the torsion angles is related to variations in the C-C bond lengths of the Cp' ligands that are observed These bond lengths can vary from as much as 1 38 to 1 43 A in some cases Our force field has the same force constant for all C-C bonds of the Cp' ligand and the minimized structures almost always give a C-C bond distance of 1 40A for these metallocenes Because of the C-C bond length variations, small differences in non-bonded interac- tions lead to torsion angles different from those calculated for Cp' rings with identical C-C distances The C-C variations can be caused by steric strain which the force field can calculate, but if, as is probable, the Variations have an electronic origin, molecular mechanics is mute on this issue Although the conse- quences of these C-C bond variations are minor and are probably inconsequential in assessing steric interactions, the variations do point to the difficulty of developing general force fields for complexes where transferability of the ligand force field is desired A clear example of the potential problem is the case of olefin complexes which can be described as metallocyclo- propenes at one extreme and metallocyclopropanes at the other canonical bonding extreme Clearly, force constants and equili- brium bond distances derived for one extreme are not transfer- able to the other, and it may be necessary to narrowly specify the group of olefin compounds to which a particular force field applies For the present Cp' systems the ligand force field can be transferred without seriously affecting either the structures or the energy differences between isomers Table 4lists superimposed structures and selected parameter comparisons for a number of strapped metallocenes The strap joining the two Cp' rings can consist of one or more atoms Inspection of Table 4reveals that excellent fits for the calculated and observed structures are generally found The two structures with silicon straps, [Zr(Me,Si(C,H,),)Cl,] and the racemic CHEMICAL SOCIETY REVIEWS, 1994 cenes, [Ti(C,H,(C,H,),)CI,] and [Hf(C,H,(C,H,),)Cl,] are also well reproduced except that the strap in the former is rotated further from the CI-Ti-CI bisector in the calculated structure than in the crystal structure A similar greater rotation is observed in the calculated structure of the meso isomer of [Ti(Me,C2(3-Bu,C,H,)2)C12]Aside from this the fit is excellent It is possible that this greater rotation of the straps in the calculated structures could be caused by crystal packing effects An opportunity to test this supposition is provided by the racemic [Ti(C,H,(3-ButC,H,),)C12] complex which crystallizes In monoclinic and tetragonal modifications, in which the rotation of the strap is different for the two forms 26 The calculated and observed structures are shown in Table 4without hydrogen atoms It will be noted that the calculated strap is spanned by those of the two crystal structures The hydrogen atoms were located for the tetragonal form but not for the monoclinic Since the hydrogen atoms were located for the tetragonal form it is possible to test the assertion that torsional rotation of the strap is controlled to some extent by crystal packing forces Crystal packing minimization calculation on the tetragonal form was carried out inthe following way The selected molecule was minimized in the presence of 16 of its rigidly positioned crystal nearest neighbours Included were all surrounding mole- cules, the Ti atoms of which were within 12 5 8, of the Ti atom of the molecule to be minimized In this way the crystal forces exerted on the molecule to be minimized were replicated Although this method is approximate it allows for ready calcula- tion The two superimposed structures are shown in (1 1) It can be seen that the fit is almost perfect lending credence to the upposition that crystal packing forces can affect the rotational structures of these systems Table 5 shows the superimposed structures of the Ti and Zr complexes of the ethylene-1,l'-bis(tetrahydroindenyl), (C2H,(THIND),), ligand The fit for the Zr complex is excellent but the minimized structure of the Ti complex has its strap rotated from the Cl-Ti-CI bisector A crystal packing calcula- isomer of [Z~(M~,S~(~-BU'-~-M~(C~H,)~))C~,],are C, sym-tion was performed on the racemic crystal2' of [Ti(C,H,metric The chirality of the latter arises from the particular (THIND),Cl,] The superimposed structures are shown in (1 2)binding of the Cp' groups The three-carbon strapped metallo- where the fit is almost perfect MOLECULAR MECHANICS FORCE FIELDS FOR CYCLOPENTADIENYL COMPLEXES-B BOSNICH 393 Table 4 Calculated and observed structures of strapped metallocenes Rigid Fit Geometry Calculated Observed 2 06A 2 06A 2 38A 2 37A 132" 133" 106" 106" 94" 94" 7" 8" 0" 3" 135", -135" -143", 145" 2 178,, 2 17A 2 188, 2 1781 2438,,244A 2 41 8,.2 43 8, 131" 130" 107" 107" 95" 96" 6" 8" 0" 2" -136", 136" -143". 141" Zr-D( 221A 2 20A Zr-CI 243A 2 44A Dcp -Zr-Dcp 126" 125" D,,, -Zr-C1 107" 108" Cl-Zr-Cl 98" 98" Dc, -CCp -Si 18" 17" CCp-si-c[ ,, 96" 93" CMe-Si-CMe 112" 116" cvc-si-cc" 112" 112" 0" 0" 180", 180" 180".180" [Zr(Me,Si(C,H,),)Cl,] monoclinic tetra onal Ti-Dc,, 2 098,,2 088, 2 lOA, 2 098, 2 111.2 IOA Ti-C1 2 35A, 2 32A 2 35A, 2 32 "A 2 38k2 33A D,, -Ti-D,, 131" 127" 129" D,-, -Ti-Cl 105"-108" 106"--110" 105"-108" Cl-Ti-Cl 95" 96" 97" D( -Ti-CBu' 16", 11" 13", 10" 10" 8" D,, -Ti-C, 2", 1" 2" 1" ccp -Dcp -Dcp -ccp 18" 20" 22" C', -C,-C,-Cc, 41" 45" 46" Bcl-Ti-Dc, -Ccp -171". 155" -164". 147" 175", 166" I U( -[Ti(C2H4(3Bu'C,H3),)Cl2] C refers to the Cp cdrbon atom bonded to the strap An inspection of the structure (12) reveals that the molecule is Table 5 Calculated and observed structures of strapped a diastereomer with three sources of chirality, one due to the tetrahydroindenyl complexes conformation of the strap, one due to the conformation of the cyclohexenyl groups, and one due to the binding of the tetra- Rigid Fit Geometry Calculated Observed hydroindenyl groups to the metal The chirality of the confor- Tl-DC, 2 09A 2 l0A mations can be defined by the four-atom puckering shown Ti-Cl 2 348, 2 35A below If the dotted line, drawn behind the four-atom sequence, Dc -Ti-DCp 131" 128" is regarded as the major axis of a helix then the 2ndand 3rdatomsDCp -Ti-CI 106" 107" either form a right-handed helix, 8, or a left-handed helix, h TheCI-TI-C1 95" 96" Dcp -ccp -c, 2" 0" chirality of the tetrahydroindenyl binding to the Ti is specified CCP-DCp-DCp-CCp 19" 20" by defining the chirality of the Cp' carbon atom bonded to the Ccp -C,-C,-CC, 42" 46" strap carbon atom The strap carbon has the lowest priority and BcI-Ti-DCp -Ccp -165", -177" -171", -171" hence a clockwise sequence, Ti -+cyclohexenyl Cp carbon -+Cp I U( -[TI(C,H~(THIND),)CI,~ carbon, obtains for the Cp' binding in (12) and hence the Cp' binding is R,R The diastereomer (12) can be written as R,R-Zr- D,, 2 22'4 221A [Ti(8-C2H4(h-THIND)2C12]or simply as 6-R,R-h,h It should be Zr-C1 2 438, 2 44A noted for future discussion that, for example, the mirror image DCp -Zr-Dep 125" 125" isomers 8-R,R-h,h and h-S,S-8,8 are of the same energy D,, -Zr-C1 105", 11 1" 108", 107" We have investigated the minimized energies of all of the Cl-Zr-CI 97" 99" DCp -cc, -c, 1" 1" conformational diastereomers of R,R-[M(C2H4(THIND),)C12] Cc, -Dc, -Dcp -Ccp 20" 20" (M = Ti, Zr, Hf) complexes We find that the S-R,R-h,h (or A-C,, -C,-C,-Ccp 45" 49" S,S-8,8)diastereomer found in the crystal is not the most stable B,-,-Zr-De,-Cc, 171", 171" 171", 171" I uc-[Zr(C,H,(TH IN D),)Cl,] C refers to the Cp cdrbon dtom bonded to the strdp 6 h conformer The A-R R-6,6 diastereomer is 2 2 to 3 kcal/mole more stable than the 6-R R-A,A conformer found in the crystal Presumably the 6-R R-h,h form produces the less soluble crystal and under different solvent conditions other conformers might be isolated if conformer interconversion is facile Table 6 lists the relative strain energies found for the different conformers Table 6 Relative energies of the various conformers of R,R-[M(C,H,(THIND),)Cl,] complexes M = Ti Zr Hf Conformer A E (kcal/mole) A-R R-6,6 A-R R-A,6 6-R R-6,X 6-R R-6,6 6-R R-A,X* X R R-A,A 0 16 18 20 30 34 0 15 15 18 22 29 0 15 16 18 24 30 * Conformer found by ,k’ ray diffraction for dl1 metals Given the small energy differences calculated for the various conformers, it is probable that these molecules undergo rapid interconversion between conformers in solution The activation energy for conformer interconversion is likely to be small because in some crystal structures the cyclohexenyl carbon atoms are found to be disordered because of rapid conforma- tional interconversion in the crystal The flexible nature of parts of these molecules suggests that during enantioselective catalysis there is a considerable amount of low energy ligand steric accommodation available to meet the steric demands of sub- strate reaction Thus any attempts at defining the stem origins of enantioselective reactions promoted by these systems requires recognition of the conformational flexibility as well as the steric strain that may accompany the reaction intermediates and transition states It is clear that just simply accepting the crystal structure coordinates as a steric framework for assessing steric interactions is likely to be a poor approximation, although it is commonly used 9 Perspective It is hoped that this overview of our work in the development of molecular mechanics force fields of metallocene complexes has provided the reader with a comprehension of the potential power of the technique The method, if properly used, can predict structures accurately, it can define differences in strain energy accurately, and it is capable of defining the origins of structural features As we have seen, the old controversy con- cerning the relative orientations of the Cp rings in ferrocene has been defined in terms of the force field contribution to the orientation Similarly, the bent metallocenes of the alkaline earths, the lanthanides and, perhaps of the divalent silicon group of metals now have a cogent explanation for their structures The reproduction of the structures and conformational energy differences associated with the strapped and unstrapped metal- locenes of Ti, Zr, and Hf both provide the basis for assessing steric interactions of stereoselective reactions promoted by derivatives of these complexes The effects of crystal packing on structures can also be determined One of the important aspects demonstrated by this work is that certain structural features can be explained by molecular mechanics without resorting to molecular orbital calculations Organic chemists have routinely used molecular mechanics in order to understand strain and steric hinderance for some time Organometallic chemists, however, are presented by a consider- ably more formidable challenge because, unlike organic chemists, they are faced with innumerable bonding schemes and structures When confronted with the prospect of developing CHEMICAL SOCIETY REVIEWS, 1994 force fields for organometallic complexes one has three choices The first is to do nothing and hope that the issue will either go away or will be resolved some other way The second is to develop generic force fields for a wide variety of structures and ligand types This second response is the one adopted by a number of commercial software houses If the objective is to be able to reproduce a wide variety of structures to a tolerable degree of accuracy and to obtain an approximate estimate of non-bonded interactions, the generic force fields suffice at least until more sophisticated force fields are developed Provided these generic force fields are not pushed beyond what they are capable of doing they can be very useful because they amount to a sophisticated method of model building superior to the mechanical balls-and-sticks found in most laboratories The third approach is to begin by carefully selecting certain general classes of organometallic systems and then constructing a self- consistent force field for these types of complexes This is the approach we have adopted in the expectation that certain narrowly defined structural issues could be unambiguously resolved We expect that in future all three approaches will be adopted but for those concerned with structure and reactivity the generic and rigorous force fields will be applied according to inclinations If work continues in developing self-consistent force fields one might expect that organometallic molecular mechanics will be applied as routinely as is now the case for purely organic systems Our expectation is that work is likely to expand rapidly in this area because the notion of treating molecules as spring and ball entities remains as attractive now as it did over a century ago Acknowledgements I am grateful to my coworkers Dr T N Doman and Mr T K Hollis and to Professor C R Landis who provided us with access to the CHARMM program The work was supported by the National Institutes of Health My indeb- tedness to my co-workers goes beyond normal civilities This is because of my singular ignorance and apprehension of computers I am yet to be persuaded that they are not an obstacle to thought and do not amount to little more than another file in the register of human folly 10 References 1 D H Andrews, Phys Rev, 1930,36,544 2 T L Hill, J Chem Phys, 1946,14,465 3 (a)F H Westheimer and J E Mayer, J Chem Phys , 1946,14,733 (b)F H Westheimer, J Chem Phjs , 1947,15,252 4 U Burkert and N L Allinger, ‘Molecular Mechanics, ACS Mono- graph, No 177, 1982 5 J R Gollogly and C J Hawkins, Znorg Chem , 1969,8, 1168 6 M R Snow, J Am Chem Soc, 1970,92,3610 7 R D Hancock, Prog Inorg Chem , 1989,37, 187 8 T N Doman, C R Landis, and B Bosnich, J Am Chem Soc, 1992,114,7264 9 R Brooks, R E Bruccoleri, B D Olafson, D J States, S Swaminathan, and M Karplus, J Comput Chem , 1983,4, 187 10 CHEM-X, developed and distributed by Chemical Design Ltd Oxford, England 11 A Haaland, Acc Chem Res , 1979, 12, 415 and references cited therein 12 P Seller and J D Dunitz, Acta Crystallogr Sect B, 1980,36,2946 13 V F Jellinek, Naturforsch B, 1959, 14, 737 14 D P Freyberg, J L Robbins, K N Raymond, and J C Smart, J Am Chem Soc , 1979, 101,892 15 M Hillman and J D Austin, Organometallzcs, 1987, 6, 1737 16 H Stoekli Evans, A G Osborne, and R H Whiteley, Helv Chzm Acta, 1976, 59, 2402 17 J K Burdett, in ‘Accurate Molecular Structures Their Determi- nation and Importance’, ed A Domenicano and I Hargittai, Oxford University Press, New York, 1992, p 501 and references cited therein 18 R A Anderson, R Blom, C J Burns, and H V Volder, J Chem Soc Chem Commun , 1987, 768 and references cited therein 19 M Kaupp, P R Schleyer, M Dolg, and H Stoll, J Am Chem Soc , 1992, 114, 8202 and references cited therein 20 R Blom, K Faegri, and H V Volden, Organometallrcs, 1990,9,372 and references cited therein MOLECULAR MECHANICS FORCE FIELDS FOR CYCLOPENTADIENYL COMPLEXES-B BOSNICH 21 R A Anderson, R Blom, J M Boncella, C J Burns, and H V 25 J A Ewen, J Am Chem Soc , 1984,106,6355 (6)W Kaminsky, K Volder, Acta Chem Scand Ser A, 1987,41,24 Kulper,H H Brintzinger,andF R W P Wild, AngeM Chem Int 22 T K Hollis, J K Burdett, and B Bosnich, Organometallics, 1993, Ed Engl, 1985,24, 507 12, 3385 and references cited therein 26 S Collins, Y Hong, and N J Taylor, Organometallics, 1990, 9, 23 R L Halterman, Chem Rev, 1992, 92, 965 and references cited 2695 therein 27 S Collins,B A Kuntz,N J Taylor,andD G Ward,J Organomet 24 (a)F R W P Wild, L Zsolnai, G Huttner, and H H Brintzinger, Chem , 1988,342,21 J Organomet Chem, 1982, 232, 233 (b) F R W P Wild, M Wasiucionek, G Huttner, and H H Brintzinger, J Organomet Chem , 1985,288, 63

ISSN:0306-0012    

DOI:10.1039/CS9942300387

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

HAWORTH MEMORIAL LECTURE. Experiments Directed Towards Glycoconjugate Synthesis* T. Ogawa The Institute of Physical and Chemical Research (RIKEN), Wako-shi, Saitama, 351-01, Japan and Graduate School for Animal Resource Science and Veterinary Medical Science, The University of Tokyo, Yayoi, Bunkyo- ku, Tokyo, 113, Japan 1 Introduction Remarkable advances' in synthetic methods and strategies in the field of complex oligosaccharides have been made in the past decades. In this article, highlights of our synthetic studies on glycoconjugates will be described. Particular emphasis will be placed on strategies for the total syntheses of complex carbohyd- rates and glycoconjugates such as oligosaccharins, glycolipids, cycloglycans, glycoproteins, and proteoglycans.2 Exploitation of New Methods and Strategies In 1976 a new method for highly regioselective and efficient acylation as well as alkylation was discovered by employing tin(rv)-mediated activation2 of hydroxyl groups. This techno- logy allowed the possibility of establishing synthetic strategies for the regioselective extension of glycan chains at the branching points of complex glycans. A conceptually new approach to glycosylation was developed in 1987 by Y. Ito3 employing alkyl phenylsulfenate as a new type of glycosyl acceptor which was readily activated in the presence of hard Lewis acid such as TMSOTf to generate a soft electro- phile PhSOTf and a hard nucleophile ROTMS. Upon reaction with either glycals or thioglycosides, which may be regarded as soft nucleophiles, a high yield of 0-glycosides was obtained under extremely mild conditions.Use of soft electrophile PhSeOTf4 in order to activate S-glycosides was also developed and applied successfully in the synthesis of complex carbohyd- rate sequences. Activation of thioglycosides was also executed efficiently by S. Sato5 in the presence of Bu:NBr-CuBr,-AgOTf (or HgBr,) to give 0-glycosides in a highly stereocontrolled manner. In order to attain high stereocontrol for glycosylation using sialic acid derivatives as glycosyl donor, 8-equatorially oriented phenylthio or phenylseleno groups6 were introduced as a stereo- controlling auxiliary at C-3. This new strategy for the introduc- tion of an a-D-NeuAc residue into carbohydrate sequences was recently implemented in the first total synthesis of disialosyl gangliosides, GD3.Tomoya Ogawa received his Ph.D. (1967)from the University of Tokyo under Professor M. Matsui and began his career as an assistant at the same university in 1967. He moved to RIKEN in 1968, and was appointed head of the laboratory for syn-thetic cellular chemistry in 1979. For two years from I972 he was a post-doctoral research fellow with Professor S. Hanessian at the University of Montreal. He returned to RIKEN in 1974. Since I990 he has also been a professor of cellular biochemistry at the graduate school of the Univer- sity of Tokq'o. 397 In the case of alcohols with low reactivity such as ceramide derivatives, a major side-reaction led to the formation of ortho- esters instead of 1,2-trans glycosides. To avoid this annoying outcome, both pivaloyl and trimethybenzoyl groups were intro- duced as 0-2 auxiliaries in the glycosyl donor which eventually improved the coupling efficiency dramatically.' 3 Total Synthesis of Oligosaccharins Some plant cell wall fragments have been chemically and physiologically characterized as oligosaccharins.8 Oligogalact- uronic acid (l), a pectin fragment, has been claimed to act as an endogenous regulator molecule which enhances the resistance of plant cells against invasive pathogens, and a fragment of hemi- cellulose xyloglucan (5) has been shown to be a factor controll- ing plant cell growth and differentiation.In 1984, Y. Nakahara began synthetic studies on pectin fragment (1) with the aim of confirming its physiological func- tions in plants. The designed synthetic blocks (2),(3), and (4) proved to be highly efficient in executing a-stereoselective glyco- sylations under Mukaiyama conditions, first between (3) and (4) (63%) and then between the product of the latter coupling and (2) (64%). Eventually a-( 144) linked dodecagalactoside was obtainedg as a properly protected key intermediate, which was then oxidized, deprotected, and purified by FPLC to give (1) in 1989. Biotesting showed (1) to be physiologically equivalent to the natural sample in eliciting phytoalexin biosynthesis and accumulation in soybean. Another oligosaccharide (5), a fragment of xyloglucan, was reconstructed by K.Sakai'O by coupling tetrasaccharide block (6) with disaccharide block (7) in high stereoselectivity and then coupling the hexasaccharide block so obtained with trisacchar- ide block (8), though in low stereoselectivity (a:/3=3: 1). The product was deprotected to yield (3,which inhibited the auxin- stimulated growth of etiolated pea stem segments in the same manner as the natural product. In the course of these experi- ments directed toward a total synthesis of (3,an efficient way to achieve a-D-xylosylation at primary hydroxyl groups had to be developed. This was only possible using the novel copper- mediated activation procedure for thioglycosides mentioned above. During experimentation aimed at the elucidation of solution conformation of phytoalexin elicitor-active /3-D-glucohexa-oside, N.Hong developed a method for the stereocontrolled regiospecific introduction of deuterium atoms' * to obtain deu- terium labelled analogues such as (9). One of the reactions for the key stereoselective coupling was between deuterated trisac- charide donor (1 0) and trisaccharide accepter (1 1) which was carried out in 63% yield. 4 Total Syntheses of Glycolipids Among many biologically significant glycosphingolipids we have studied so far, I will first focus here on the ganglio- * Based on a Haworth Memorial Lecture delivered at the Spring Meeting of the Royal Society of Chemistry Carbohydrate Group on March 29th 1993 at the University of Dundee and at the Annual Chemical Congress of the Royal Society of Chemistry on April 7th 1993 at the University of Southampton.gangliosides GM, (12) and GD, (16). When M. Sugimoto started synthetic studies on these molecules around 1980, no clear experimental evidence had been reported for the introduc- tion of sialic acid at secondary hydroxyl groups. We therefore first studied carefully the coupling reactions between sialosyl donor (13) and disaccharide glycosyl acceptor (14). In 1984 to our delight, after intensive purification of the reaction mixture, the (2+3)-linked trisaccharides were isolated for the first time (in 18% yield) and we were able to characterize them well spectroscopically. * Though the ratio of the products with U-and unnatural ,&configuration was 1 :2, this observation was CHEMICAL SOCIETY REVIEWS, 1994 clear evidence for the first time that we could employ the conventional sialosyl donor (13), known since 1966, to obtain not only a major elimination reaction leading to the formation of an undesired 2,3-dehydro-compound, but also the desired substitution at the C-2 position of (13) upon reaction with a rather sterically demanding alcohol such as (14).From the results of these experiments we deduced that the product obtained', by Shapiro in 1973 was in fact not the a-but 8-sialylated compound. Encouraged by this observation, a total synthesis of the ganglioside GM, (12) was subsequently achieved for the first time in 1985. The ceramide part-structure EXPERIMENTS DIRECTED TOWARDS GLYCOCONJUGATE SYNTHESIS-T OGAWA (15) was synthesized by K Koike from D-glucose in an efficient manner in 1984 In order to control the course of glycosylation sterically with a sialosyl donor, the novel glycosyl donor (17) was designed by Y Ito This proved to be a highly successful synthetic manoeuvre, and based on this advance the first total synthesis14 of ganglio- side GD, (16) as achieved in 1989 Coupling of disialosyl donor (17) with a disaccharide block (18) proceeded with high stereo- control (a p = 60 1) in 48% yield Since the tetrasaccharide thus obtained from (17) and (18) carries a pivaloyl group as an auxiliary (vide ante) at the 0-2 of the reducing end glucose residue, the coupling between ceramide derivative (1 5) and tetrasaccharide glycosyl donor was executed smoothly and the product was successfully transformed into the target compound (16)A first total synthesis of a further chain-extended ganglio- ganglioside (19) was also achievedI5 in 1990 The coupling of (20) with (21) was carried out in the presence of BF, OEt, in (CH,Cl), to afford the desired p(1+3) linked hexasaccharide and the undesired p( 1 +4) linked isomer in 43 and 7% yield, respectively Functional group manipulations of the /3( 1-3) linked hexasaccharide, coupling with ceramide derivative (1 5), and final deprotection gave (19) In addition to the studies on ganglio-gangliosides, other series of glycosphingolipids also attracted our synthetic curiosity Thus globo-series glycolipids such as SSEA-3 (22) (Stage Speci- fic Embrionic Antigen-3) and Forssman antigen (23) were synthesized by S Nunomura for the first time in 198816 and 1989,” respectively Carbohydrate sequences (22) and (23) were synthesized bj use of a common glycotriosyl acceptor (26) and glycobiosyl donors (24) and (25), respectively The coupling between the imidate (24) and a glycosyl acceptor (26) gave only CHEMICAL SOCIETY REVIEWS, 1994 22% of the pentasaccharide based on (24).In fact, 44% of (24) was converted into an undesired rearranged anomeric trichloro- acetamide. However, we were pleased to observe that Cu2+- mediated glyco~ylation~ of (26) with the thioglycoside (25) afforded a 78% yield of a mixture of a-and /3-linked pentasac- charide in a ratio of 1:lO. Couplings of the carbohydrate sequences with a ceramide derivative (15) were achieved in the usual way and the products were then converted into (22) and (23), respectively.Another important group of glycosphingolipids as targets for our synthetic studies was the so-called neolacto-series. S. Sat0 first approached the synthesis' of dimeric LeX antigen (27). The crucial coupling between (28) and (29) was carried out in 62% yield in a highly regioselective manner. The octasaccharide intermediate was then converted into (27) in 9 steps in 15% overall yield. This bond disconnection strategy has also been applied19 recently in the first synthesis of sialyl dimeric Lex antigen (30).Another neolacto-series glycosphingolipid, L2/HNK-1 anti-gen (3l), isolated from the nervous system was synthesized20 via EXPERIMENTS DIRECTED TOWARDS GLYCOCONJUGATE SYNTHESIS-T. OGAWA 401 a key intermediate (32) by T. Nakano in 1991. Neolacto-series glycosphingolipid (33) with multiple branches was synthesized2 by Y. Matsuzaki in 1992. Simultaneous couplings between a glycotriosyl donor (34) and three hydroxyl groups of the glyco- hexaosyl acceptor (35) were achieved by the Suzuki procedurez2 to give stereoselectively 7 1YOof pentadecasaccharide, which was in turn converted into (33) in 8 steps in 5% overall yield. In continuation of the neolacto-series syntheses, the complex pentaantennary structure of I-type glycan (36) was synthesized in 1993 by employing essentially the same strategy as used in the case of (33).Two key intermediates (37) and (38) were used as glycosyl donor and a glycosyl acceptor, respectively. After glycosylation, the desired protected pentacosasaccharide was isolated in 37% yield and was then completely deprotected in four steps in 37% overall yield to give (36).23The structure was confirmed by 'H-NMR and FAB-MSz4 Apart from the glycosphingolipids so far discussed, glycosyl- phosphatidylinositol anchor (39) was a particularly challenging target for synthesis. In 1991 C. Murakata reported a first synthesisz5 of (39). In consequence of a personal communica- tion,26 however, we now have to revise our previous assignment for the synthetic structure of (39) to that of its diastereoisomer (40)which contains L-myoinositol instead of the D-enantiomer.Thus in fact a synthesis of unnaturalGP1 anchor (40)was carried out by us employing regioselectively protected heptasaccharide (42) and two hydrogen phosphonate synthons (41) and (43). Another example I would like to note here is the recent synthetic on Nod factor glycolipid (44) carried out by S. Ikeshita. In this case, the introduction of the activated fatty acid (45) onto the carbohydrate backbone was planned to be the last step of the synthetic sequence. A deblocked key intermediate (46) was employed so that we could easily introduce a variety of fatty acids and thus study the biological specificity of interac- tions between host plants and microbes in terms of fatty acid structures.5 Synthesis of Cycloglycans Cyclodextrins are a group of naturally occurring cycloglycans well known for their ability as host molecules to recognize hydrophobic guest molecules. In 1985, Y. Takahashi exa-mined2* for the first time non-enzymatic ‘cycloglycosylation’ of an acyclic glucohexaosyl fluoride (48) by utilizing Mukaiyama conditions and successfully isolated a cyclic product (in 21 YO yield) that was successfully converted into a-cyclodextrin (47). Thus the total synthesis of (47) was accomplished 70 years after Schardinger first described its properties in reliable detail in 1920. After completing our synthesis of y-cyclodextrin in 1987, we then turned our efforts to the synthesis of other (1 -+ 4)-linked cyclooligosaccharides.By employing a linear thioglycoside such as (49) a high conversion into a-(1 -,4) linked cyclomannohex- aose (74%) was achieved by M. Mori in the presence of a soft electrophile such as PhSeOTf.3 In addition, further transforma- tion of the cycloglycan into functionalized cycloglycans such as (50) could now be executed.29 By employing a-( 1 +4)-linked oligolactose (51) as the key intermediate, trigonally shaped cyclooligolactose (52) was prepared30 by H. Kuyama in 1993. Further experiments have established synthetic routes to the homologous cyclooligolactoses (53) and (54). Energy minimized structures (59, (56), and (57) derived by T. Nukada using a modified MM2 program (Daikin Co.) indicated unique struc- tures for synthetic cycloglycans (52), (53), and (54), respectively.CHEMICAL SOCIETY REVIEWS, 1994 Practical applications expected for these kinds of artificial cycloglycans remain to be shown. 6 Glycoprotein Glycans Glycans of glycoproteins are classified into two major groups: N-linked types in which glycan chains are linked to an Asn residue, and 0-linked types in which glycan chains are linked to either a Ser or a Thr residue. Synthetic approaches towards typical structures (58) and (6 1) are highlighted here. The bian- tennary structure (58) carries two identical tetrasaccaride branches. Because of this characteristic, a tetrasaccharide block (59) and a trisaccharide block (60) were designed as glycosyl donor and glycosyl acceptor, respectively.The trichloroacetimi- date (59) was coupled with diol(60) under Schmidt conditions to give the desired a-(1 -,6) and a-( 1 -,3) linked undecasaccharide product in (56% yield) which was deprotected to afford (58). This crucial experiment completed the first total synthesis31 of a very typical biantennary complex type N-glycan of glycoprotein in 1986. Glycophorin A is one of the abundant glycoproteins embed- ded in the plasma membrane of human red blood cells and is rich in carbohydrate content. As part of our project on the synthesis of clustered 0-linked glycans on a peptide core, a partial structure of glycophorine A which corresponds to the N-terminal heptapeptide (61) was chosen as a model target.Based on the novel strategy14 for the a-stereoselective introduction of sialic acid as well as the conventional Fmoc for the peptide chain elongation, tetrasaccharide-serine/threonine blocks (62) and (63) were designed and subsequently synthesized in highly stereocontrolled manner by Y. Nakahara. A chain elongation from (64) by repeated use of (62) and (63) was carried out successfully. The product was finally depr~tected~~ to give the N-terminal heptapeptide structure (6 1) that represents the human blood group M epitope. Based on these achievements in both N- and 0-linked glycopeptide synthesis, we are continuing EXPERIMENTS DIRECTED TOWARDS GLYCOCONJUGATE SYNTHESIS-T. OGAWA CHEMICAL SOCIETY REVIEWS, 1994 EXPERIMENTS DIRECTED TOWARDS GLYCOCONJUGATE SYNTHESIS-T.OGAWA our synthetic challenge of dissecting the structure and function of biologically active glycoproteins. 7 Glycosaminoglycans and Proteoglycans Glycan structures that belong to proteoglycan or glycosamino- glycan (GAG) can be classified into four major groups, (i) hyaluronan, (ii) chondroitin and dermatan sulfate, (iii) heparin and heparan sulfate, and (iv) keratan sulfate. From our synthe- tic studies on these glycans, three examples are highlighted as follows. A synthetic approach taken by T. Slaghek towards a hyaluronic acid fragment tetrasaccharide (65) was designed by use of a key intermediate (66) that was prepared from each monosaccharide units. Oxidative conversion of (66) into (65) was executed in seven steps in 44% overall yield.34 Synthetic studies have also been directed toward the carbo- CHEMICAL SOCIETY REVIEWS, 1994 hydrate sequence (67) linked to a peptide backbone via a serine residue in chondroitin sulfate.Towards this goal, a key interme- diate (68) was designed for the purpose of the regioselective introduction of the two sulfate groups of (67). Compound (68) was reconstructed from (69), (70), and (7 1). First two trisaccar- ide blocks (69) and (70) were coupled in the presence of (Bu,N),CuBr, and AgOTf in 75% yield. The hexasaccharide that was isolated was further transformed into the key interme- diate (68) in eight steps in 32% overall yield. Regioselective introduction of sulfate into (68) and complete deprotection to give (67) was executed3 by F.Goto in five steps in 36% overall. A dermatan sulfate hexasaccharide sequence (72) was shown to have high affinity towards heparin cofactor II.36 Synthetic experiments directed towards (72) were carried out by designing EXPERIMENTS DIRECTED TOWARDS GLYCOCONJUGATE SYNTHESIS-T OGAWA (73) as a key precursor molecule, which was in turn synthesized from two glycosyl donors (74) and (75),and a glycosyl acceptor (76) Transformation of (73) into (72) was performed in eight steps in 11% overall yield Unambiguous of (72) afforded supporting evidence for the proposed structure of the biologically active domain of dermatan sulfate In summary, my co-workers and myself have developed versatile and unambiguous synthetic approaches towards com- plex structures of carbohydrate sequences that occur in Nature Acknowledgments Financial supports from the Science and Technology Agency and the Ministry of Education, Science, and Culture of the Japanese Government are deeply appreciated I wish to express my sincere thanks to all of my co-workers for their immense contributions to the projects which I have des- cribed in this review, and to Ms A Takahashi for her skilful technical assistance 8 References 1 R U Lemieux, Chem Soc Rev , 1978,7,423, H Paulsen, ibid ,1984, 13,15, R R Schmidt, Angeu Chem Int Ed Engl ,1986,25,212, H Kunz, zbid, 1987, 26, 194, H Paulsen, rbid, 1990, 29, 823, K C Nicolaou, Aldrichimica Acra, 1993, 26, 63 2 T Ogawa and M Matsui, Carbohydr Res , 1977, 56, C1 3 Y Ito and T Ogawa, Tetrahedron Lett , 1987,28,2723,4701, 1988, 29, 1061 4 Y Ito and T Ogawa, Tetrahedron Lett , 1987,28, 6221 5 S Sato, M Mori, Y Ito, and T Ogawa, Carbohvdr Res ,1986,155 C6 6 Y Ito and T Ogawa, Tetrahedron Lett, 1988,29, 3987 7 S Sato, S Nunomura, T Nakano, Y Ito, and T Ogawa, Tetrahed-ron Lett , 1988, 29,4097 8 A G Darvill and P Albersheim, Ann Rev Plant Physiol ,1984,35, 243 9 Y Nakahara and T Ogawa, Carbohvdr Res .1990,205, 147 10 K Sakai, Y Nakahara, and T Ogawa, Tetrahedron Lett , 1990,31, 3035 11 N Hong, Y Nakahara, and T Ogawa, Proc Jap Acad, 1993,69, Ser B 55 12 M Sugimoto and T Ogawa, Gljcoconjugate J , 1985, 2, 5, Carbo-hbdr Rey , 1985, 135, C5 13 D Shapiro, Pure App[ Chem , 1974,2, 153 14 Y Ito, M Numata, M Sugimoto, and T Ogawa, J Am Chem Soc , 1989,111,8508 15 M Sugimoto, K Fujikura, S Nunomura, Y Ito, and T Ogawa, Tetrahedron Lett, 1990, 31, 1435 16 S Nunomura and T Ogawa, Tetrahedron Lett, 1988,29, 5681 17 S Nunomura, M Mori, Y Ito, and T Ogawa, Tetrahedron Lett, 1989,30,67 13 18 S Sato, Y Ito, and T Ogawa, Tetrahedron Lett, 1988,29, 5267 19 M Iida, S Nunomura, M Numata, M Sugimoto, K Tomita, and T Ogawa, Jap Soc Biosci Biotech Agrochem Ann Meeting Abstract, 1993,417 20 T Nakano, Y Ito, and T Ogawa, Tetrahedron Lett ,1991,32.1569, Carbohydr Res , 1993,243,43 21 Y Matsuzaki, Y Ito, and T Ogawa, Tetrahedron Lett, 1992, 33, 6343 22 K Suzuki, H Maeta, and T Matsumoto, Tetrahedron Lett, 1989, 30,4853 23 Y Matsuzaki. Y Ito, Y Nakahara, and T Ogawa, Tetrahedron Lett, 1993,34, 1061 24 T 11, Y Ohashi, Y Matsuzaki, T Ogawa, and Y Nagai, Org Mass Spectrometrj, 1993, 28, 1340 25 C Murakata and T Ogawa, Tetrahedron Lett, 1991, 32, 671, Carbohvdr Res , 1992,235,95 26 According to the personal communication with Dr R Baker at Merck Sharp & Dohme (Terlings Park), the assignments for the absolute configuration of D-and L-myo-inositol derivatives reported in the paper of D C Billington, R Baker, J J Kulagowski, and I M Mawer, J Chem Soc Chem Commun , 1987, 314, should be reversed, and the correction will appear in the same Journal 27 S Ikeshita and T Ogawa, Glycoconjugate J ,1994,11,257 28 T Ogawa and Y Takahashi, Carbohydr Res , 1985,138, C5, 1987, 164,277 29 M Mori, Y Ito, and T Ogawa, Tetrahedron Lett , 1990,31, 3029 30 H Kuyama, T Nukada, Y Nakahara, and T Ogawa, Tetrahedron Lett , 1993,34,2 17 1 31 T Ogawa, M Sugimoto, T Kitajima, K K Sadozai, and T Nukada, Tetrahedron Letl , 1986,27,5739, T Ogawa, Y Nakahara, H Yamamoto, T Nukada, T Kitajima, and M Sugimoto, Pure Appl Chem , 1984,56,779 32 L A Carpino and G Y Han, J Org Chem, 1972,37,3404 33 Y Nakahara and T Ogawa, Tetrahedron Lett , 1994,35,3321 34 T M Slaghek, T K Hypponen, and T Ogawa, Tetrahedron Lett , 1993,34, 7939 35 F Goto and T Ogawa, Pure Appl Chem , 1993,65,793 36 M M Maimone and D M Tollefsen, J Biol Chem , 1990, 256, 18263 37 F Goto and T Ogawa, Bioorg Med Chem Lett, 1994,4,619

ISSN:0306-0012    

DOI:10.1039/CS9942300397

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Pericyclic Key Reactions in Biological Systems and Biomimetic Syntheses Ulf Pindur” and Gunter H. Schneider Department of Chemistry and Pharmacy, Institute of Pharmacy, University of Mainz, 0-55099 Mainz, Federal Republic of Germany 1 Introduction The pericyclic reactions. of organic chemistry, in contrast to polar or radical structural transformations of molecules, are one-step processes proceeding through a cyclic transition-state structure.1,2 Thus, they constitute reactions of the concerted type. Pericyclic reactions are not only of particular interest because of their stereospecific course and their broad prepara- tive significance in the chemistry of drugs and natural products but also for theoretical reasons. As was realized by Woodward and Hoffmann,, Zimrne~mann,~ Fukui,6u and others, De~ar,~ the stereochemical progress is determined by the symmetry of the molecular orbitals participating in the reaction (the so-called HOMO and LUMO according to the frontier molecular orbital concept6).Thus, with the help of, for example, Woodward and Hoffmann’s concept on the ‘principle of the control of orbital symmetry’, the stereochemical route of a thermally or photoche- mically induced pericyclic reaction and thus the resultant pro- duct spectrum can be predicted.1.2.6 Figure 1 shows a typical, thermally allowed pericyclic 6-electron reaction in hydrocarbon chemistry together with the appropriate Woodward-Hoffmann nomenclat~re.~.~ The transition-state structures derived from quantum chemi- cal ab initio calculations’ exhibit a great variety from strongly bonded, closed shell to weakly bonded, flexible forms with diradical character. Pericyclic reactions possess an extremely wide potential for application in organic synthesis since most of them can be realized according to stereochemically preplanned structural concepts.Creative successes have, in particular, been reported in the fields of the total synthesis of natural products and the development of drugs.6h In the field of the specific molecular designing of drugs, pericyclic synthesis concepts open the way to unimagined and in some cases fascinating possibili- ties ranging through to the asymmetric synthesis of physiologi- cally active, enantiomerically pure substances.Pericyclic reac- tions also occur in natural biological systems; for example, in some biochemical processes and the biosynthesis of numerous secondary metabolites, especially in the plant kingdom. The stereospecific production of enantiomerically pure natural pro- ducts points in many cases to a highly ordered, enzymatic catalysis. These naturally occurring, thermally or photochemi- Gunter H. Schneider was born in Bad Hersfeld, Germany in 1965. From 1985 to 1987 he received training as a chemical laboratory assistant. After acquiring prac- tical experience in the chemical industry he studied chemistry at the University of Mainz from 1987. Since 1993 he has been a scientific member of the Institute for Pharmacy at Mainz; he obtained his Dip- loma degree in I994 and is now doing post-graduate research work in Professor Pindur ’s group.His research involves mechanistic studies of the colour reactions of tricyclic, ant idepressive drugs. 409 cally allowed rearrangements are of great interest for the in situ generation of pharmacologically or biologically relevant, reac- tive intermediates. On the other hand, the pericyclic reactions confirmed in biological systems provide valuable chemical infor- mation for the conception of new natural product synthesis. The elucidation of biosynthetic mechanisms and the frequently associated biomimetic synthetic investigations (simulation of synthetic steps of biosynthesis) can be planned rather precisely when pericyclic reactions are involved. In the present review, some examples of typical and outstand- ing pericyclic reactions taking place in biological systems are discussed and classified on the basis of the type of the reaction mechanism.In addition to natural processes, some biomimetic syntheses are also included. In many of the selected examples, the ‘formal’ pericyclic mode of formation of structurally clari- fied (and often enantiomerically pure) natural products can be predicted by means of the retrosynthetic analysis principle. 2 Reactions in Biological Systems and Biomimetic Syntheses 2.1 Electrocyclic Reactions The formation of vitamin D, (cholecalciferol) in the outer skin regions proceeds from 7-dehydrocholesterol (I), a ubiquitous companion of cholesterol, under the influence of sun light (UV light, A = 275-3 10nm).’ Photochemically induced conrotatory cycloreversion in the steroid ring B gives rise to precalciferol D, (2) which is ultimately transformed into vitamin D, by way of a both thermally and orbital-symmetry allowed [1,7]-sigmatropic H-shft from C-19 to C-9 [(I) represents the thermodynamically more stable conformer in the crystal] (Figure 2a).’ Experimental evidence for the entire reaction process has been obtained Uv Pindur was born in Bad Darkau, Germany in 1943.He obtained his Diplom degree in pharmacy and food chemistry at the University of Marburg in 197111972. He earned his Dr. rer. nat. degree in synthetic organic chemistry in 1974 and his Habilitation in pharmaceutical chemistry in 1980 at the University of Marburg and was appointed Privat Dozent.From 1980 to 1985 he was Professor of Pharmaceutical Chemistry at the University of Wiirzburg and at the end of 1985 he was appointed to his present position, Professor of Pharmaceutical Chemistry at the University of Mainz where he has been head of the Institute for Pharmacy since 1990. His re-search interests include pericyclic reactions of heterocyclic systems to give natural products and biologically active compounds (DNA attacking molecules), applications of theor-etical pharmaceutical chemistrv for drug design, and analytical aspects of the mecha- nisms of colour reac-tions of drugs. CHEMICAL SOCIETY REVIEWS, 1994 [ ls,5s] sigmatropic0rearrangement [3s,3s] sigmatropic rearrangement " Cope rearrangement 'I [,6] disrotatoric electrocyclic reaction L J +[x2s, 2s + 02s 1 pericyclic reaction " Alder -En reaction " [n4s &I+ cycloaddition " Diels -Alder reaction 'I [x4s + 02s + moa]chelotropic reaction + :CH, " Carben -1,4-cyclo-addition " [x2s + 2s Jsl+ (".II group-transfer reaction +l)IIH Figure 1 Examples of the simple basic structures in hydrocarbon chemistry for the usual, thermally allowed pericyclic reactions in which six electrons participate. The respective transition state struc- tures are shown on the right in square brackets. The carbon atoms in the reaction partners may individually be exchanged for heteroatoms (e.g.0,S, N) (hetero-Diels-Alder reaction, hetero-Cope rearrange- ment, 1,3-dipolar cycloaddition with heterodipoles, etc.). through isotopic labelling, kinetic measurements, and determi- nation of the quantum yield. In reality, however, more complex equilibrium systems are involved.8 The frontier orbitals (HOMO and LUMO) of the parent structure 1,3,5-hexatriene [7~ system of (2)] according to the Woodward-Hoffmann rules are displayed qualitatively in Figure 2b. For the course of the cyclization, referred to the 1,3,5-hexatriene/l,3-cyclohexadiene equilibrium, the HOMO of the triene is relevant in the case of thermal excitation and the corresponding LUMO in the case of photochemical excitation (disrotatory or conrotatory process, respectively).In the case of the formation of vitamin D, in the organism, ring B of 7-dehydrocholesterol(l) corresponds to the 1,3-cyclohexadiene structure in Figure 2b. The transformation of chorismine to prephenic acid [(4)+(5)] which occurs in plants and microorganisms in the synthesis of aromatic amino-acids (phenylalanine, tyrosine) is a stereospecific Claisen rearrangement of the hetero-Cope type, i.e. a [3,3]-sigmatropicprocess (see Figure l).9JO This process is catalysed by the genuine enzyme chorismate mutase -a key HO Figure 2 Photochemically initiated biosynthesis of vitamin D, (3) from 7-dehydrocholesterol (1). The technical production of vitamin D, from ergosterol is based upon an analogous (biomirnetic) concept.Figure 2b Qualitative frontier orbital model according to the Wood- ward-Hoffmann principle of retention of orbital symmetry for the 1,6-electrocyclization of 1,3,5-hexatriene to 1,3-~yclohexadiene. In the biosynthesis of vitamin D, the symmetry of the LUMO in ring B has a decisive role in the control of the reaction (orbital control). enzyme of the shikimate pathway -through degradation of the transition state structure (I) (Figure 3),lobThis reaction can also be accelerated by the action of so-called catalytic antibodies. OU Catalytic antibodies have increasingly gained importance in recent years as enzyme equivalents in organic and bioorganic chemistry.196 In order to obtain specific antibodies for the catalysis of the (4)/(5)-rearrangement, the compounds (6a, 6b) are used as haptenes for simulation (as mimeticum) of the Claisen transition state I.loO The enediyne cytostatic/antibiotic agents constitute a class of new natural compounds of bacterial origin now undergoing rapid development.These DNA-cleaving natural products, PERICYCLIC KEY REACTIONS IN BIOLOGICAL SYSTEMS-U. PINDUR AND G. H. SCHNEIDER 41 I 0-o&.,, 0 0 H 11 N-COMe chorismate mutase . QCo; 1. nucleophile attack Box,,,,OH OH 2. conjugate addition H -0ocL -0oc-L 0 Figure 3 Enzyme-catalysed [3,3]-sigmatropic rearrangement of choris- mine to prephenic acid (chorismate-prephenate rearrangement).9.’0u The hapten molecules (6) are used to generate catalytic antibodies as an enzyme equivalent.loU which combine an unusual molecular structure with practically unbelievable biological activities and fascinating modes of action, now occupy a central position as novel lead structures in medicinal chemistry.’ Members of this substance groups are, for example, calicheamicins (Table 1) isolated from Micromo-nospora echinospora ssp. calichensis. lU They are active against Gram-positive and Gram-negative bacteria. Most important, however, is their extraordinary activity against P338 mouse tumour, leukaemia L1210 cell lines, and those of neoplasms such as colon 26 and B- 16 melanoma at optimal doses of 0.15 to 5 pg/kg body weight. The biological activities are attributed to sequence-specific DNA damage; thus, for example, calicheami- cin y\ (70 binds specifically to TCCT, CTCT, and TTTT sequences in the minor groove of B-DNA.Here, the oligosac- charide part, essential for the molecular recognition, is oriented towards the 3’ end of the DNA fragment (Figures 4and 5). It is currently assumed that the oligosaccharide unit constitutes the Table 1 The calicheamicin family (7a-g) X (a) Calicheamicin fi$ Br (b) Calicheamicin yyr Br (c) Calicheamicin a: I (d) Calicheamicin a: I (e) Calicheamicin fi: I (f) Calicheamicin y\ I (g) Calicheamicin 6: I 0-sugar MeS/’? (7) cd 3.16A [MM2] (8) :Nu cd 3.35A [MM2] Bergman-cyclization 0 0 N-COMe [DNA-cleavageI0-sugar 0-sugar (9) 0 N-COMe 0-sugar 0oc\-OH-Calicheamicin8 * Figure 4 Mechanism for the cleavage of DNA by calicheamicin y: (70.According to MM2 force-field calculations,’ lUthe distance dc in (70is shortened by 0.19 8, on transformation to (8). Rha = EEd OMe OH hi = R’ R2 R3 Rha Ami CHMe, Rha Ami Et H Ami Et Rha H - Rha Arni CHMe, Rha Arni Et Rha Ami Me CHEMICAL SOCIETY REVIEWS, 1994 Figure 5 Computer-generated molecular model showing the minor groove binding of calicheamicin 0; (1 1) to double-stranded DNA along the TCCT site * Id (Reproduced with permission from Chemistry CG Biology Ild) most important part of the molecule for recognition and bind- ing. Then a nucleophile (e.g. glutathione), possibly activated intramolecularly by a basic nitrogen atom, attacks the central sulfur atom of the unusual trisulfide linkage with formation of a thiolate (bioreduction process, see Figure 4).On account of the geometry of the allylic double bond, the thiolate moiety is in an ideal position to attack intramolecularly the a,p-unsaturated ketone incorporated in the neighbouring six-membered ring to furnish compound (8). This reaction, in which the sp2 carbon atom at the site of attack changes its hydridization to sp3,is an indispensable prerequisite for the subsequent Bergman cycliza- tion (6-electron electrocyclization to a 1,4-sigma diradical) to furnish the benzoid-type, highly strained diradical (9) ( = cycloaromatization via a benzoid I ,4-diradical). The reactive diradical is then in the correct position to extract two hydrogen radicals from the DNA, one from the C5' position of desoxycyti- dine (C) and the other from a ribose of the opposite strand.The resulting DNA radicals then react with molecular oxygen via cleavage of the double strand.' la In the mean time, several synthetic studies on the molecular design of calicheamicin as a lead structure with the aim of developing highly selective anti-tumour agents have been repor- ted.' la-' lr One result of this work was the design of calicheami- cin Ol, (1 1) based on the known mechanism of the biological action of calicheamicin yi (70.'lb In place of the trisulfide unit of the original, this novel derivative possesses a thioacetyl group as the reactive, electrophilic centre.Calicheamicin Ol, (1 1) also reacts differently to its natural analogue: it is activated under neutral or mild basic conditions, e.g., by hydroxide ions. The 1,4-phenyIene diradicals derived from (1 1) also induce cleavage of the double-stranded B-DNA effectively and selectively at TCCT, CTCT, and TTTT sequences (concentration for biologi- cal activity: < 10 M). 2.2 DielsAlder Reactions In the past few years there has been an explosion of reports about numerous inter- and intramolecular Diels-Alder reac-tions (or formal Diels-Alder reactions) occurring in plants and microorganisms. Although definitive and experimentally con- firmed evidence on the actual mechanisms is not available in all cases, highly feasible results, especially in the field of phyto- toxins, have been presented on the basis of biosynthetic experi- ments (feeding experiments with cell cultures, use of labelled compounds, selective treatment with enzymes, and biomimetic syntheses).In some cases, in particular those involving enantio- merically pure target molecules, an enzymatic catalysis of the [4~J+ ,n,]-cycloaddition by an apparently genuinely existing 'Diels-Alderase' has been discussed. The biosyntheses of alkaloids of the Iboga and Aspidosperma types [e.g.catharanthine (14) or tabersonine (1 5)] are assumed to start from stemmadenine (12) (Figure 6). Heterolytic ring open- ing with elimination of water then leads to the postulated dehydrosecodine (1 3) which can undergo cyclization from two different orientations to furnish ultimately catharanthine (14) or tabersonine (15), respectively (Figure 6). 4-1 An isomeric dehydrosecodine of the type (1 3) has also been proposed as the biogenetic precursor of the Aspidosperma alkaloid pseudotaber- sonine. Several biomimetic syntheses of Iboga and Aspidos-perma alkaloids on the basis of this intramolecular enamine- acrylate reaction (Figure 6) have been described in detail.' 6-1 The more stable dihydropyridone (1 6), an oxosecodine deriva- PERICYCLIC KEY REACTIONS IN BIOLOGICAL SYSTEMS-U q,!-,!q (13) COOMe Figure6 Possible biosyntheses of Ihoga and Aspidospernia alkaloids and biomimetic synthetic strategies via 15-oxovincadifformine (1 7) and the 15-siloxycatharanthine ( 18) l6 tive, was used in some of the regio-controlled cycloadditions as an equivalent for dehydrosecodine By means of a related reaction, it should be possible to synthesize the dimeric indole alkaloids presecamines (19) and (20), isolated from the leaves or roots of Rhazja species, from Diels-Alder reactions of the secodines (21a) and (21b) which also possess the 2-vinylindole x-system (Figure 7) Parasitic lower fungi such as Alternuria solani (Fungi imper- fecti) growing on Solanaceae species produce the solanopyrones A (22), B (23),C (24), and D (25) originating from the polyketide metabolism (Figure 8) 2o * These constituent substances could be responsible for the hazardous potato disease known as potato blight in North America Biomimetic investigations20 21 with the appropriate diene-dienophile precursor (26) (Figure 8)strongly support the genuine production of these analogues of the naturally occurring hydroxymethylglutaryl-coenzyme A reductase inhibitors (e g lovastatin from Aspergzllus ter reus) Since optically pure compounds are formed, an enzymatic catalysis has been discussed Hypotensive chalcones and the dimerization product kuwa- none J have been isolated from the bark of the mulberry tree Morus alba L 22 Biogenetic experiments with Morus alba cell cultures involving feeding with the 0-methylated genuine chal- cone (29) revealed that the formation of the di-0-methylated kuwanone J (32) proceeds via an enzyme-catalysed, formal Diels-Alder reaction between the enone unit in (30) and the 2- methyl- 1,3-butadiene component in the dehydro derivative (31) (Figure 9) 22 In addition to a series of sesquiterpene lactones, several novel PINDUR AND G H SCHNEIDER 41 3 oR$TN oR$TD \ \ \ \ N N C0,Me C0,Me pJ-7-\ N H C0,Me (21) a R=X b R=Y Figure 7 The dimeric indole dlkaloids presecamines (19) dnd (20) dnd their putative biogenetic precursors (2ld) and (2 1b) CHO CHO O&OCHS ///,Oh //,,O& 63\ H H H (22) R= CHO (24) (23) R= CH,OH n nVy” n toluene170 -190 ‘C-OXOCH, (71 Yo) 4,?li/-’&+ &H Figure 8 Phytotoxin lactones, (22)-(25), produced by the seconddry metabolism of Alternuria solanr and a biomimetic synthesis concept for the total synthesis of (22) and (25) involving an intrdmoleculdr Diels Alder reaction of (26) which leads primarily to the opticdlly pure diastereomers (27) and (28) bis-sesquiterpene lactones (33)-(37), have been isolated from Helenrum autumnale L and their structures elucidated (Figure 10) 23 2s From this series, for example, the natural product (33) can be considered as a formal [4 + 21-cycloadductof isoalanto-lactone (38) and zingiberene (39) where the semicyclic double bond of the a-methylene-7-butyrolactone acts as the dienophile CHEMICAL SOCIETY REVIEWS.1994 (29) OCH, OH I1 Figure 9 Biogenesis of the dimerization product (32) by asymmetrical Diels-Alder reaction of (31) and (30) in cell cultures of Morus afba L 22 A Diels-Alder reaction Figure 10 Novel bis-sesquiterpene lactones from Helenium autumnale L and their formal biosynthesis via a Diels-Alder reaction 23 component and the cyclohexadiene ring of zingiberene (39) as the diene component For this transformation, a thermal Diels- Alder reaction has also been realized on a preparative scale 24 The biosynthetic key step to (35)-(37) can be constructed analogously, but in these cases a formal hetero-Diels-Alder reaction could lead to the target compounds (35)-(37) The phenylphenalenone derivative lachananthocarpone (4 1a) is the major pigment found in Lachnanthes tznctorza Ell (Hae- modoraceae) (Figure 11) 26 The biosynthetic pathway to (41a) and its analogues (41b) and (41c) was proposed from the results of feeding the plants with [2-14C]tyrosine and [l-I3C]phenylala- nine, respectively The formation of a diarylheptanoid (40) as diene4ienophile system and its subsequent Diels-Alder reac-tion should represent the basic principle of the biosynthetic key step (Figure 11) Furthermore, in addition to providing a new and presumably general synthetic approach to 2-hydroxyphenalenones, the syn- thesis of (41a) from (42) (Figure 12) reveals that the 9-phenyl- phenalenone system present in the pigments of plants of the Haemodoraceae family can indeed be constructed from suitably substituted 1,7-diarylheptanoid ortho-quinones In this respect, the synthesis may be considered as further substantiation of the hypothesis shown in Figure 11 26 Heliocide H, (48), an insecticidal sesterterpenoid, was iso- lated from cotton (Gossypzum hzrsutum L ) 27 On the basis of zn vztro Diels-Alder synthetic studies at room temperature, its biogenetic formation is presumed to be the result of a stereospe- cific [4 + 21 process between hemigossypolone (46) and myrcene (47) (Figure 13), probably vzcl a highly ordered endo-transition state Myrcene will encounter less steric interaction by approaching the quinone ring in such a way that the myrcene alkenyl side-chain and the quinone isopropyl group are as far apart as possible This would result in the construction of the side-chain at C-18 Many further pericyclic reactions certainly still remain to be detected in biological processes and a particular challenge will be the isolation and characterization of the responsible enzymes -leading, we predict, to a renewed boom in preparative, pericyclic synthesis of active principles according to the concept of ‘enzy-matic organic synthesis’ PERICYCLIC KEY REACTIONS IN BIOLOGICAL SYSTEMS-U PINDUR AND G H SCHNEIDER Me 00 (40) OH R OH (41) a R=H b R=OMe c R=OH Figure 11 Biogenetic pathway to naturally occurring phenylphenale-nones as plant pigments via an intramolecular Diels-Alder step 26 HO Me Figure 12 Biomimetic synthesis of (41a) from an appropriate diene-dienophile precursor (43) 26 Me Me Me Figure 13 Possible biogenetic process leading to heliocide H, (48) via Diels-Alder reactions 27 3 References 1 K N Houk,Y Li,and J D Evanseck, Angebi Chem ,Znt Ed Engl, 1992, 31, 682, and references cited therein 2 J March, ‘Advanced Organic Chemistry’, 4th Edn , Wiley, New York, 1992 3 R B Woodward and R Hoffmann, Angeu Chem , 1969,8,781 4 H E Zimmerman, Acc Chem Res , 1971,4,272 5 M J S Dewar, Angeu Chem ,Int Ed Engl , 1971, 10,761 6 (a) K Fukui, Acc Chem Res , 1971, 4, 57 (b)G Desimoni, G Tacconi, A Barco, and G P Pollini, ‘Natural Product Synthesis through Pericyclic Reactions’, ACS Monograph 180, American Chemical Society, Washington, DC, 1983 7 W G Dauben and R B Phillips, J Am Chem SOC,1982,104,355 8 (a)J E Havinga and J L M A Schlatmann, Tetrahedron, 1961,16, 146 (6)J D Enans, G Y Shen, and W H Okamura, J Am Chem Soc, 1991,113,3873 9 Y Asano, J J Lee, T L Shieh, F Spreafico, C Kowal, and H G Floss, J Am Chem SOC, 1985,107,4314 10 (a)A Bentley, CRC Crrt Rev Biochem Mol Biol , 1990,25,307 (b) W W Smith and P A Bartlett, J Org Chem ,1993,58,7308 (c) For several review articles on catalytic antibodies in synthetic chemistry, see Ace Chem Res , 1993,26, 391453 1I (a) K C Nicolaou and W -M Dai, AngeM Chem , Int Ed Engl, 1991, 30, 1387 (b) K C Nicolaou, T Li, M Nakada, C W Hummel, A Hiatt, and W Wrasilo, Angew Chem , Int Ed Engl, 1994,33, 183 (c) For several detailed articles, see the special issue of Tetrahedron, 1994,50, 1341 et seq (d)K C Nicolaou, Chemistry & Biology (introductory issue), 1994, XXVI 12 A Ichihara, Synth Org Chem (Jpn ), 1992, 50, 96 13 D Voet and J G Voet, ‘Biochemistry’,Wiley, New York, 1990 14 E Wenkert, J Am Chem Soc , 1962,84,98 A I Scott, Acc Chem Res , 1970,3, 151 15 W A Carroll dnd P A Grieco, J Am Chem Soc , 1993,115, 1164 16 W G Bornmann and M E Kuehne, J Org Chem , 1992,57, 1752 17 R J Sundberg and R J Cherney, J Org Chem , 1990,55,6028 18 R J Sundberg and J D Bloom, J Org Chem, 1981,46,4836 19 G A Cordel1,G F Smith,andG N Smith,J Chem SOC,Chem Commun , 1970, 191 20 H Oikawa, T Yokota, A Ichihara, and S Sakamura,J Chem Soc Chem Commun , 1989, 1284 21 A Ichihara, M Miki, H Tazaki, and S Sakamura, Tetrahedron Lett , 1987,28, 1175 22 A Ichihara, in ‘Studies in Natural Products Chemistry’, Vol 4, ‘Stereoselective Synthesis (Part C)’, ed Atta-ur-Rahman, Elsevier, Amsterdam, 1989, p 579 23 H Haberlein and R Matusch, Dtsch Apoth Ztg , 1987, 127, 2057 24 R Matusch and H Haberlein, Liebigs Ann Chem , 1987,455 25 R Matusch and H Haberlein, Helv Chim Acta , 1987, 70, 342 26 A C Bazan, J M Edwards, and U Weiss, Tetrahedron, 1978, 34, 3005 27 R D Stipanovic, A A Bell, D H O’Brien, and M J Lukefahr, Tetrahedron Lett, 1977, 567

ISSN:0306-0012    

DOI:10.1039/CS9942300409

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Surfactant Systems Their Use in Drug Delivery M.Jayne Lawrence Department of Pharmacy King’s College London University of London Manresa Road London SW3 6LX 1 Introduction Molecules or ions which are amphiphilic that is contain both a hydrophobic and hydrophilic part in aqueous solution fre- quently assemble at interfaces and self-associate in an attempt to sequester their apolar regions from contact with the aqueous phase. This self-association gives rise to a rich variety of phase structures (Figure 1). Aggregation is not however just limited to aqueous solution; it is sometimes observed in non-aqueous polar solvents such as ethylene glycol and non-polar solvents such as hexane (in the latter case giving rise to inverse structures). Over the years several of the phase structures produced by surfactants have been of interest to the pharmaceutical scientist either as drug vehicles/carriers or more recently as targetting systems. In the former application the surfactant system takes no part in the biodistribution of the drug it carries acting purely as the vehicle. In the second case the surfactant system in some way ‘conveys’ the drug to the desired (or target) site in the body and deposits it. Targetting can take one of two forms; namely ‘passive’ targetting which relies on the natural biodistribution of the carrier or ‘active’ targetting in which the carrier is in some way directed to the desired site frequently by the use of targetting ligands expressed on the surface of the carrier. Both types of targetting have the advantage of protecting the body from any unwanted side-effects of the drug while at the same time achieving the desired concentration of drug at the target site. By far the majority of work examining the potential of surfactant systems in drug delivery has explored their use as drug carriers; for example non-ionic micelles have been widely inves- tigated as a means of producing a clear stable solution of a poorly water-soluble drug suitable for intravenous or oral administration.’J However during the past twenty years or so as the importance of drug targetting has been realized a number of surfactant systems such as phospholipid or non-ionic surfac- tant vesicles have been extensively investigated as targetting systems. Despite all the research into the potential use of surfactant phase structures for drug delivery such phase structures have not made much of an impact on the formulation scene; there are M. Jayne Lauwnce graduated in Pharmacy .from Liverpool Polytechnic (B.Sc.) in 1981 and qualijied us a member of the Royal Pharmaceutical So-ciety of Great Britain in 1982. She received her Ph.D. degree in 1985 from the Universitjj of Manchester. Since 1984 she has been a Lecturer in the Pharmacy Department King’s College London. Her research interests cover the design syn- thesis and physio-chemical churucterization of surfactant systems and membrane transport. 417 only a few marketed preparations that could be considered to be drug-containing surfactant systems in either the United King- dom or the United States. Consequently the true potential of surfactant formulations particularly of non-ionic surfactants has perhaps not been fully realized. In order to appreciate the potential and also the limitations of such systems an understand- ing of the phase behaviour of surfactants is essential. The following account therefore describes the phase behaviour of surfactants with reference to their physico-chemical properties relevant to their use as drug delivery systems. It also details some of the work performed to date investigating the use of surfactant systems -in particular those produced from the less toxic non- ionic surfactants -for drug delivery.* 2 Phase Behaviour of Surfactants 2.1 Equilibrium Phase Structures Although surfactants self-associate in a wide variety of solvents their state of aggregation varies considerably between solvents (Table 1). As water or a buffered aqueous solution is the usual continuum for most drug delivery systems it is important to understand (and predict) the range of equilibrium phase struc- tures commonly encountered in such solutions. Mention will be made of the phase structures encountered in other solvents where appropriate. When a surfactant is dispersed in water just above the limit of its aqueous solubility (i.e. above its critical micelle concent- ration cmc) it generally aggregates depending upon its molecu- lar ge~metry,~ into one of four types of structure namely the isotropic micellar phase and the liquid crystalline hexagonal lamellar and cubic phases. The aforementioned phases. with the exception of the lamellar phase can either be in a normal or reverse orientation. Recently in addition to these commonly encountered phase structures there has been an increasing number of more unusual aggregates such as helical bilayers6 and fibre gels7 reported. Upon increasing the surfactant concentration well above the cmc there are generally changes in aggregate or phase structure. The order of phase structures formed upon increasing surfactant concentration generally follows a well-defined sequence (Figure 2) with a ‘mirror plane’ through the lamellar phase such that normal phase structures can be considered to be ‘oil-in-water’ while reverse structures can be thought of as ‘water-in-oil’.* Most surfactants however exhibit only a portion of this sequence depending upon the aggregate type initially formed at the cmc and the resulting interaggregate forces e~perienced.~ Although the same phase structures are observed in other non- aqueous polar solvents the sequence of phases is sometimes very different and appears to depend both upon the molecular geometry and the nature of the polar head-solvent interactions. 2.1.I Implications for Drug Delivery An understanding of the phase behaviour of surfactants is essential for the efficient use of surface active systems in drug delivery. For example after introduction into the body the surfactant system may depending upon its route of administ- ration undergo a large dilution. If the surfactant is diluted below its cmc precipitation of transported drug may occur. This precipitation may have very serious consequences especially if 418 CHEMICAL SOCIETY REVIEWS 1994 t 1 Surfactant Molecules Spherical Micelles Rod-shaped Micelles Hexagonal Phase Lamellar Phase Reverse Hexagonal Phase Reverse Micelles Figure 1 Table 1 Self-association in solvents Class of solvents Example of class Type of Aggregate Class A Class B Water glycerol ethylene glycol Hexane benzene cyclohexane Normal Reverse Class C Methanol ethanol No aggregate formation the drug is being administered intravenously. Ideally therefore the cmc should be a low as possible in order to avoid such problems. Surfactants that form lamellar phases at their cmc generally do so at much lower concentrations than those surfac- tants which initially form micelles. Since non-ionic surfactants generally exhibit lower cmc’s than ionic surfactants they are preferred for the purposes of drug delivery especially when a micellar solution is being investigated as the drug delivery vehicle. In a similar vein if a concentrated surfactant solution is administered it may experience a sufficient dilution to induce a phase change say for example from an hexagonal to a micellar phase. As the drug-carrying capacity of each aggregate type may differ such a phase change could have very serious implications such as dose dumping within the body. When considering using a surfactant system as a drug delivery vehicle it should also be borne in mind that phase transitions can also be induced by a change in temperature and that normal human body tempera- ture is typically 12 degrees above ambient. Other problems to be aware of are hysteresis effects. These are particularily common in cubic phases and may have important consequences for drug delivery. For example certain cubic phases have been shown to be pseudo-stable for very long periods at temperatures at which some other form of aggregate would normally be formed.6 A knowledge of the various biological surface-active agents which the surfactant aggregate may encounter in vivo is also essential as these may alter or even destroy the aggregate. For example the endogenous micelle-forming bile salts encountered in the small intestine have been shown to solubilize orally administered liposomes thereby releasing any water-soluble solute trapped inside the carrier. 2.3 Modified Phase Structures In addition to the equilibrium phase structures mentioned above there are a few non-equilibrium and modified surfactant phase structures that are currently finding application in drug delivery. Increasing surfactant concentration ‘oil-in-water’ ‘mirror plane’ ‘water-in-oil’ I H,O Micelle (L,) <Hexagonal (H,) < Lamellar (La)< Reversed Hexagonal (H,) < Reversed Micelle (LJ Solid I I I I I I I‘ II I I I II I Cubic (I,) Cubic (V,) { Cubid (V,) Cubic (I,) I Figure 2 Idealized phase sequence in surfactant-water systems. (Modified from reference 6; terminology as in reference 7.) SURFACTANT SYSTEMS THEIR USE IN DRUG DELIVERY-M 2 3 I Vesicles Vesicles are generally formed by dispersing lamellar phases in an excess of water' (or non-aqueous polar solvents such as ethy- lene glycol dimethylformamide) or in the case of reversed vesicles in an excess of oil l The resulting vesicles are approxi- mately spherical structures dispersed in a water or an oil continuum Vesicles produced from phospholipids have been widely investigated as drug delivery vehicles Unlike the phase structures mentioned earlier however these non-equilibrium structures are prepared using methods such as sonication and will eventually re-equilibrate back into the lamellar phases from which they originate l1 This inherent instability has caused considerable problems with the wide-spread commercial exploi- tation of vesicular delivery systems For a few surfactants however the vesicular phase is an equilibrium structure for example the ionic ganglioside GM3 a glucosidic amphiphile of biological origin forms vesicles spontaneously in water while some combinations of non-ionic surfactants have been shown to form reversed vesicles spontaneously l 2 3 2 Pofjmerized Aggregates Attempts have been made to use polymerization to stabilize various nascent phase structures for example micelles,' cubic phases,16 and vesicles With the exception of micelles (which as yet it has not proven possible to polymerize) polymerization of these structures gives aggregates exhibiting the significant advantage that they do not suffer break down upon dilution in vzvo However because of their size (ranging from tens to hundreds of nm) these aggregates can cause problems as they are not readily excreted from the body hence such systems will probably have limited clinical use although they may have a use in oral administration In an attempt to overcome the problem biodegradable polymerized aggregates are presently being inves- tigated When preparing drug-containing polymerized aggre- gates it is important to choose the appropriate stage for drug addition adding the drug before polymerization may cause the drug to be irreversibly bound in the aggregate while addition of the drug after polymerization may lead to low levels of entrapment 2.4 Drug Aggregates A number of drugs are themselves amphiphilic and may aggre- gate into various structures most frequently small micellar type structures In these cases the drug aggregate could act as its own vehicle if the drug loading were not too high It has been postulated that the formation of vesicles consisting of pure drug (pharmacosomes) should also be feasible l9 Unfortunately most drugs are not lipophilic enough to form vesicles easily without derivatization with materials like fatty acids l9 However with certain drugs it may be possible to produce vesicles over a narrow pH range using the appropriate ratio of amphiphilic salt to free drug The tight control over pH that would be necessary however means that such vesicles are unlikely to provide useful drug delivery systems An alternative approach to producing pharmacosomes has recently been reported in which a biode- gradable micelle-forming drug conjunct has been synthesized from the hydrophobic drug adriamycin and a polymer com- posed of polyoxyethylene glycol and polyaspartic acid 2o This approach has the benefit that while it may be possible to dilute out the micelle the drug will probably not precipitate because of the water solubility of the monomeric drug conjunct Since neither of these types of derivatized drug structures consist of drug alone they can therefore not be considered to be true drug aggregates 2.5 Influence of Oil When oil is added to a binary mixture of surfactant and water a whole variety of phase structures may be formed Several of these structures currently have a use in drug delivery for J LAWRENCE example microemulsions macroemulsions and multiple emul- sions * Others such as self-emulsifying systems2 and vesicles encapsulated in water-in-oil emulsions are at present under investigation 22 3 Choice of Phase Structure for Drug Delivery When choosing a phase structure for drug delivery a number of factors need to be considered in particular how the physico- chemical properties of the phase structure relate to the intended application If for example a surfactant system is required for topical use the phase structure chosen should be of sufficiently high viscosity to enable the preparation to be retained on the skin surface while at the same time allowing it to be spread readily over the surface of the skin In contrast if a system is intended for administration intravenously it should be of suffi- ciently low viscosity not to cause pain upon injection Another important factor to be considered is the capacity of the aggregate for the drug to be incorporated Micellar solutions by virtue of low surfactant concentrations generally exhibit the lowest capacity for drug while in contrast cubic and other liquid crystalline phases can frequently tolerate very high drug load- ings 23 24 Recently it has been realized that the toxicity of a particular surfactant may depend upon the nature of its aggre- gate For example the same surfactant has been shown to exhibit a significantly reduced toxicity when present in a vesicu- lar as opposed to a micellar solution Table 2 gives some of the physico-chemical characteristics important for formulation purposes together with the possible pharmaceutical applications of each phase structure It should be noted that while Table 2 gives the average properties of each phase the variations in each case maybe quite significant For example while solutions containing spherical micelles generally exhibit low viscosities those containing fong rod shaped micelles frequently exhibit very high viscosities Similarly cubic phases can display a wide range of stiffness some samples are as hard as plexiglass while in others the phases are sufficiently flexible that they almost flow ti It is important when considering the use of surfactant phase structures as delivery vehicles to remember that a surfactant aggregate cannot be considered an inert carrier and that the drug and indeed other additives such as preservatives and flavourings* may (depending upon the amount present) dra- matically alter the cmc and in some cases the type and range of aggregates formed Unfortunately very little work has been performed in this area and is difficult to predict the effect of a drug (or indeed any other additive) on a phase structure as it is expected to vary according to whether the additive (a) is water soluble (b) adsorbs at the aggregate surface (c) co-aggregates with the surfactant or (d) resides in the interior of the aggregate Evidence suggests however that the phase structure experiences the most disruption when the additive is itself surface active For example the presence of the drug lignocaine hydrochloride at concentrations greater than about 5 wt% converts the cubic structure formed from 10 wt% monoolein in water into a lamellar phase O The influence of the presence of drug is further complicated because most drugs are administered as salts hence the amount of amphiphilic salt to lipophilic free drug varies according to pH Consequently the effect of the drug on the phase structure may vary with the pH of the surrounding environment This effect is more likely to be significant if ionic surfactants are used Yet another complication is that if the drug promotes a phase transition this transition may conceivably be reversed as the release of a surface-active drug from the aggre- gate proceeds lo This phase reversal may lead to two different patterns of drug release * Flavourings are very important if surfactants are to be given orally surfac tants do not taste very pleasant Also because of their effect on membranes surfactants may numb the patient s mouth CHEMICAL SOCIETY REVIEWS 1994 Table 2 Some physico-chemical properties and potential pharmaceutical applications of surfactant phase structure Phdse Structure Micelles Cubic Phase Hexagonal Lamellar Vesicles Solid Appearance Clear non-birefringent Clear non-birefringent Clearicloudy birefringent Clear,lCloudy birefringent Clear/cloudy birefringent Waxy solid VlSCOSlty Low Least viscous phase Very high Most viscous phase Viscous Fairly viscous Low viscosity Stiff Solubilization Capdcity Low dmphiphilic and non- polar solutes only High amphiphilic and non- poldr solutes Low water-soluble solutes Probably high amphiphilic and non-polar solutes Low water-soluble solutes Probably high amphiphilic and non-polar solutes Low water-soluble solutes High amphiphilic and non- polar solutes* Low water-soluble solutes Not known Surfactant Concentration 0-25 yo Varies Generally gredter than 30% Wide range possible Wide range possible Fairly low Generally less than 10 wt% 100 wt% Possible Use Solution for most routes of delivery Protection of labile compounds Viscous preparation for sustained release intramuscular subcutaneous oral and topical Protection of labile compounds Sustained release particularly topical Sustained release particularly topicdl Most routes of administration except oral Protection of labile compounds Solid dispersion for oral use * The solubilization capacity recorded here refers to vesicles produced by non equilibrium methods those formed spontaneously are expected to exhibit very low capacity for amphiphilic and non polar drugs (see Section 5 4) 4 Choice of Surfactant Surfactants are well known to exert a wide range of biological pharmacological and toxicological effects on man Therefore the single most important factor in the choice of a surfactant or combination of surfactants is toxicity Unfortunately this property is hard to assess The reasons for this are many not the least being the difficulty in finding an appropriate measure of toxicity especially when screening new surfactants Generally dcute oral toxicological studies are routinely performed on all new surfactants regardless of their intended usage Although this information is valuable it cannot adequately predict chronic toxicity A further complication is the understandable reluc- tance of the Pharmaceutical Companies to enter into the full scale chronic toxicity studies needed for a proper assessment of a new surfactant for drug delivery purposes a toxicity study currently costs in the order of 10million GB pounds Only a very limited number of surfactants are generally considered for formulation purposes Usually only those surfactants are used that have been used in pharmaceutical formulations for many years and are therefore generally recognized as safe even though some of these surfactants may themselves not have been tested for chronic toxicity' From a toxicological point of view non-ionic surfactants are generally regarded as the most suitable for pharmaceutical formulation * Even so the range of non-ionic surfactants used is very limited Tween 80 [polyoxyethylene (20) sorbitan mono- oleyl ether] and Cremorphor EL [polyoxyethylene (40) castor oil] are probably the two most common There are however a large number of non-ionic surfactants commercially available Some of the more common examples are shown in Table 2 A surfactant is composed of three distinct portions a hydrophilic segment a hydrophobic portion and a semi-polar linker Consequently it is theoretically possible to join together any combination of segments to produce a surfactant with the required properties. for example biodegradable surfactants can be readily achieved by the use of an ester linkage while bilayer (vesicle) and micelle forming surfactants can be produced from dialkyl and monoalkyl chain surfactants respectively Despite the wide range of surfactants potentially available most workers tend to use surfactants that have been previously used in formulation thereby limiting themselves considerably There is however a real need to produce new surfactants in order to realize the full potential of surfactant systems in drug delivery Yet the number of surfactants that can be synthesized is enormous In an attempt to address the problem of design and synthesis of new biocompatible surfactants a program VESICA25 has been developed with a view to predicting which potential surfactants would preferentially form a particular aggregate type In this way the number of surfactants that need to be synthesized could be greatly reduced 5 Phase Structures in Drug Delivery 5.1 Normal Micelles The increased solubility in a micellar solution of an organic substance insoluble or sparingly soluble in water is a well estdblished phenomenon Indeed the solubilization of water- insoluble drugs by micelles has long been investigated as a means of improving solubility for drug delivery in particular for parenteral or oral administration but also for ophthalmic topical rectal and nasal delivery The protection of labile drugs from the environment through solubilization within micelles has also been examined Consequently an enormous number of papers examine the incorporation of a wide variety of drugs into micelles formed from a large variety of surfactants and in particular non-ionic surfactants of the type shown in Table 3 There are however only a few products on the market that can be considered to be micellar systems This is mainly because solubilization capacity is usually too low to be of practical use with only a few mg of drug solubilized per g of surfactant As the average dose of a drug is in the order of tens of mg and as the concentration of the micellar solution is never more than 20 wt% surfactant this means that solubilization is not feasibleexcept in a few instances where very potent lipophilic drugs e g testosterone are incorporated Attempts have been made to design non-ionic surfactants with an improved solubilization capacity An early approach involved the production of larger micelles Despite an increased micelle size solubilization decreased upon lengthening the hyd- rophobic chain this decrease was attributed to deleterious SURFACTANT SYSTEMS THEIR USE IN DRUG DELIVERY-M Table 3 Commonly encountered non-ionic surfactants Common Hydrophilic Group Hydrophobic Group Linker Moeity Name Cholesterol Ether Solulan Long chain alcohol Ether Brij Long chain acid Ester MYl-11 Long chain acid Sorbitan ring Tween Polyoxyethylene Alkvl phenol Ether Triton Alkil amide Amide ---Alkyl amine Amine ---Polyoxypropylene Ether Pluronic Long chain Ester ---trig1 ycerides Sugar Long chain alcohol Ether ___-Long chain acid Ester ---Sorbitan ring Long chain acid Ester Span Crown ether Long chain acid Ether ---Tertiary amine oxide Long alkyl chain - - - - ---changes in the polyoxyethylene chains nearest to the core the main locus of solubilization for most drugs 26 As the amount of drug solubilized in the core is usually less than a few percent of the total drug incorporated in the micelle the same group attempted to promote solubilization in this region by the introduction of a semi-polar group into the hydrophobic chain Incorporating a single ether linkage in the hydrophobe resulted in a marked reduction in the tendency to aggregate and as a consequence a significant reduction in solubilization This modification was obviously counter-productive and suggests that solubilization cannot be improved by altering the nature of the hydrophobic region and that it may be better to consider replacing the usual polyoxyethylene head group Data do sug- gest that it may be feasible to achieve significant increases in solubilization by using alternative head groups such as the amine oxides 28 Even if it is possible to increase solubilization to a sufficient degree (ideally to about a 100 mg per g of surfactant) there are still a number of problems with the use of micellar solutions for drug delivery One of the major problems is the large dilution the system experiences upon administration This dilution is par- ticularly large after oral and intravenous administration and can cause the unwanted precipitation of drug In the case of oral delivery this may lead to irritation of the gastrointestinal tract while in the case of intravenous administration pain may be experienced upon injection Other complicating factors experienced when using micellar solutions include the concomitant solubilization of other addi- tives such as preservatives and sweetening agents some surfac- tants taste foul especially if administered as a solution Depend- ing upon their relative sites of incorporation in the micelles this co-solubilization can either lead to a decrease or increase in drug solubilization This potential problem of concomitant solubili- zation of additives is not just limited to micellar systems and is encountered with all surfactant systems Owing to their labile nature micelles can only be used as drug carriers and not as targetting systems although there is a small amount of evidence that suggests it may be possible to alter the biodistribution of a drug by administering it in a micellar solution 29 This alteration has however been attributed (at least in part) to a direct effect of the surfactant (in this case the non-ionic surfactant Tween 80) on biomembrane permeability most micelle-forming surfactants are known to influence the permeability of biomembranes Furthermore as most of the surfactan ts used for drug delivery are not readily biodegradable their activity is retained for long periods in the body Although drug solubilization in micelles has been extensively investigated much less work has been performed examining the influence on drug transfer of solubilization in micelles Accord- J LAWRENCE 42 I ing to the limited evidence available micellar solubilization reduces the rate of mass transfer of most drugs across inert membranes In the body this effect appears to be counter- balanced by the fact that the surfactant can frequently increase membrane permeability 5.2 Cubic Phases Cubic phases have received a considerable amount of attention as putative drug delivery systems lo 23 30-35 One interesting cubic phase is that formed by the polyoxyethylene-polyoxypro-pylene co-block polymer pluronic F127 This particularly attractive system has a high solubilizing capacity and is generally considered to be relatively non-toxic In aqueous solution at concentrations greater than 20 wt% F127 is transformed upon heating from a low viscosity transparent (micellar) solution at room temperature to a solid clear gel (cubic phase) at body temperature Other members of the pluronic series also undergo a liquid to gel transformation at around body temperature but only at higher surfactant concentrations (namely 30 wt% and above) 33 This thermal gelation which is reversible upon cool- ing has a number of important applications in drug delivery For example a solution poured onto the skin or injected into the body will gel to form a solid sustained release depot Further- more since the gelation is reversible removal from the skin is facilitated by simply immersing or irrigating the skin with cool water Removal from a body cavity is more difficult however and would require a surgical procedure In order to circumvent this problem some workers are currently trying to synthesize biodegradable surfactants that will undergo a thermal reversible gelation at a similar temperature to that of F127 34 To date the cubic phase of F127 has been investigated for a wide range of applications including topical delivery covering of burn wounds ophthalmic delivery rectal delivery as a vehicle for injectables by both intramuscular and subcutaneous routes and as a bioadhesive 30 33 Drug release from the cubic phase is governed by the physico-chemical properties of the solute and the concentration of the surfactant 30 31 As a general rule increasing solute lipophilicity and/or increasing surfactant con- centration leads to a decrease in release rate As a consequence the cubic phase of F127 has considerable potentidl as d sustained release preparation Another cubic phase undergoing extensive studies for phar- maceutical purposes is that formed by monoolein and water lo 23 32 A great advantage to the use of monoolein is that it is subject to enzymatic lipolysis in a wide range of tissues and is therefore considered to be biodegradable With respect to drug delivery the most interesting property of the cubic phase formed by monoolein is its ability to co-exist with water at body temperature As a consequence it is possible to formulate a system so that when added to water it does not undergo a phase change None of the other long chain monoglycerides with the exception of monoerucin and sunflower oil monoglycerides have been reported to form a cubic phase over a temperature range suitable for exploitation in drug delivery The cubic phase of monoolein occurring at about 50-60 wt% monoolein has been shown to incorporate at levels up to 5-10 wt% a large range of drugs of very different size and polarity including a number of proteins and oligopeptides without experiencing a phase change 32 At higher levels of incorporated drug depending upon the nature of the drug phase changes may be observed The reason proposed for the ability of the cubic phase to solubilize such a wide range of drugs is its very large interfacial area -in the order of 400 m2/g cubic phase 32 As with F127 the cubic phase of monoolein has been shown to extend significantly the release of bioactive substances both m vztro and m vivo 32 Again in agreement with F127 the pattern and rate of release will be very dependent upon the nature of the drug The cubic phase also has the advantage of being able to reduce the enzymatic degradation of the incorporated proteins and peptides possibly because the enzyme has restricted access to the substrate 32 The cubic phase of monoolein has been proposed as a vehicle for drug uptake from the gastrointestinal tract or as a subcutaneous or intramuscular depot for sustained release although in the latter examples because of the viscosity of the phase discomfort would be experienced upon injection Discomfort can be overcome however by formulating the monoolein to undergo a phase transition to the cubic phase on injection This phase transition can be achieved in one of two ways (1) by exploiting the transformation from a relatively low- viscosity lamellar phase at room temperature to the stiffer cubic phase present at body temperature (11) by utilization of the transition from a lamellar phase to a cubic phase upon addition of water 23 Formulating monoolein in either manner creates a precursor that is easily handled and can be injected without causing distress 23 The use of monoolein in combination with other surfactants for example the non-ionic surfactant poly- oxyethylene (20) oleyl ether (Brij 96) has also been studied in order to promote favourable phase transitions 23 One property of the cubic phase of monoolein that does not seem to have been exploited yet is its bioadhesive properties (it appears that most cubic phases are bioadhesive) As a result of these properties the cubic phase could have some use in rectal and vaginal delivery 5 2 I Cubosomes The cubic phase has been dispersed by homogenization with the aid of F127 and lecithin to produce so-called cubosomes lo Cubosomes have a particle size distribution similar to that found in commercially available oil-in-water emulsions intended for parenteral nutrition The ‘cubic phase emulsion’ contains water and it is hoped that this will extend drug release zn vzvo To date the only cubic phases that have been investigated for their use as drug delivery systems are those formed by F 127 and monoolein Yet cubic phases are commonly found in a wide variety of surfactant systems Many of these cubic phases may have a place in drug delivery -and since they can be found in non-aqueous polar solvents such as ethylene glycol and since such solvents frequently exhibit a higher capacity than water for many hydrophilic drugs the possibility exists to increase the loading of some cubic phases 5.3 Liquid Crystalline Phases Liquid crystalline lamellar phase structures are currently recog- nized as important in pharmaceutical formulation Hydrophilic creams are oil-in-water mixtures stabilized by lamellar struc- tures and it has been suggested that the lamellar structures within hydrophilic creams are sometimes the factor controlling release of drug from the system To date very little work has examined the possibility of using lamellar phases or indeed hexagonal phases for drug delivery Yet a large number of surfactants particularly those formed from non-ionic surfac- tants form liquid crystalline phases over a wide range of surfactant concentrations In addition most of the work that has been performed examining the use of liquid crystalline phases in drug delivery has not bothered to characterize the nature of the phase structure that is whether the liquid crystalline phase is hexagonal or lamellar in nature yet it is known that the release pattern differs depending upon the phase structure present 5 3 1 Lamellar Phases Only a small amount of work reported in the literature specifi- cally examines the use of lamellar phases Yet lamellar phase structures exhibit interesting solubility properties in the lamel- lar structure lipophilic bilayers alternate with hydrophilic layers which contain interlamellar water hence it is possible to incor- porate water-soluble oil-soluble and amphiphilic drugs Furth- ermore evidence suggests that some drugs are more soluble in the liquid crystalline lamellar phase than in isotropic liquids of similar composition 24 Generally a drug permeating through a lamellar gel network CHEMICAL SOCIETY REVIEWS 1994 may follow an interlamellar or translamellar route depending on local rates of diffusion and partition Extremely lipophilic drugs will probably be trapped inside the lipophilic bilayer~,~~ while extremely hydrophilic drugs will permeate through the hydrophilic regions between the lamellae and amphiphilic drugs may move both between and across the lamellae 36 In the latter case interesting release patterns have been predicted theoretically 36 For extremely hydrophilic drugs the interlamel- lar aqueous channels behave as pores the tortuosity of which is determined by the amount of free water and the orientation of the lamellae 36 The diffusion coefficient of a drug within a lamellar phase is about one to two orders of magnitude smaller than that in solution 37 As a result of their control over drug release it has been suggested that liquid crystalline phases and in particular lamellar phases are potentially very useful systems for the topical delivery of drugs 24 In addition if the release rate from the surfactant system is less than the diffusion of the drug through the skin then the surfactant system can be used as a topical controlled-release preparation One potential problem with topical application of lamellar phases is that dehydration of the skin may occur resulting in irritation Reverse micelles containing drug and lecithin in oil have recently been investigated as a precursor to a sustained release lamellar phase formulation 38 By clever formulation it should be possible to produce a reverse micellar solution of drug which transforms into a liquid crystalline system on contact with biofluids The feasibility of such an approach has been demon- strated zn vztro using an oily solution containing reverse micelles consisting of phospholipid and drug On contact with aqueous media this solution was shown to change its microstructure from spherical or cylinderical micelles to lamellar liquid crystals As the diffusion was smaller by a factor of 100 in the lamellar phase compared to the oily solution the formulation has potential as a sustained-release preparation for intramuscular or subcuta-neous administration Further as the diffusion of the drug was also dependent upon the thickness of the liquid crystalline layer which was in turn influenced by whether free acid or base was solubilized in the system it may be possible to achieve a fine tuning of release properties 5.4 Helical Bilayers These more unusual phase structures have recently been investi- gated as a drug delivery vehicle 39 The core of these tubular-like structures has been filled with a polymer matrix containing drug in an attempt to produce a sustained-release formulation animal studies showed that slow release of drugs occurred for up to 5 days 5.5 Vesicles Since the realization in the seventies that phospholipid vesicles (or liposomes) had potential as drug delivery systems vesicles have probably been the most extensively investigated of all surfactant systems Vesicles (niosomes) produced from non- ionic surfactants have also been widely studied 40 Their higher chemical stability better chemical definition and reduced cost mean that niosomes have a number of advantages over liposomes As vesicles are generally non-equilibrium structures a large number of different types of vesicles can be produced The nature of the preparation of the vesicle can determine its physical ~tability,~ an important consideration when using vesicles for drug delivery In addition the choice of vesicle type also depends both on the nature of the drug to be encapsulated and on the desired route of administration The type of vesicle is critically important only for hydrophilic drugs as different vesicle types can encapsulate different amounts of aqueous phase and consequently different amounts of water-soluble drug Lipid-soluble drugs are readily entrapped in the hydro- phobic bilayer structure and as a consequence are less sensitive to vesicle type One advantage of using vesicles formed by non- equilibrium methods is that they are not normally broken down SURFACTANT SYSTEMS THEIR USE IN DRUG DELIVERY-M upon dilution They are however liable to destruction in the presence of biological surfactants such as the bile salts and lysolecithin The main route of administration of vesicles is by intravenous injection Unfortunately most vesicles are removed rapidly from the systemic circulation by the fixed macrophages of the liver However by clever manipulation of the formulation for exam- ple by coating phospholipid vesicles with a hydrophilic polymer such as polyoxyethylene glycol uptake by the liver can be reduced thereby retaining the vesicles in the circulation for longer periods and allowing them to act as sustained release vehicles By incorporating targetting ligands on the surface of such vesicles it then becomes feasible to direct the vesicles to certain organs and deposit them there This well-established approach using liposomes has not yet been investigated using niosomes Other routes of administration that have been exa- mined using niosomes include the topical route 41 In addition the nasal occular oral rectal and pulmonary routes have all been extensively examined using vesicles prepared from phospholipids To date no work has been reported investigating the potential of spontaneously formed vesicles for drug delivery One reason is that the only biocompatible surfactant producing these vesi- cles GM3 is prohibitively expensive Also with this type of vesicle entrapment of non-polar or amphiphilic drugs will probably be difficult Another problem with using sponta- neously formed vesicles is that they are very small and as a consequence would be expected to exhibit a very low capacity for water-soluble drugs One advantage of this very small size however is that the vesicles would probably avoid uptake by the fixed macrophages of the liver Another advantage would be their improved stability A patent has been published claiming a wide range of pharma- ceutical uses for the recently discovered reverse vesicle these applications include the topical nasal rectal and parenteral routes of administration 42 Reverse vesicles have the potential to protect sensitive compounds from the environment 5.6 Reverse Micelles A possible exploitation of the association of surfactant in non- polar media is the production of reverse micellar solutions containing drug for use in the production of therapeutic aerosols from pressurized metered dose inhalers These are currently the major devices used in the delivery of drugs to the respiratory tract This situation is unlikely to change in the near future although fluorinated gases will replace the currently used chlorofluorocarbon propellants To date only one study has been performed and although this study examined the use of phospholipids rather than non-ionic surfactants it did demon- strate the concept 43 The work showed that the level of drug delivered m vim from the reverse micellar solution was compar- able with that obtained from the commercially available suspen- sion formulation For hydrophilic drugs the loading achievable in reverse micelles is limited by the amount of water solubilized in the core of the aggregate it may be possible to improve this incorporation by replacing water with a non-aqueous polar solvent such as glycerol or polyoxyethylene glycol Other potential applications of reverse micelles include the protection of labile drugs via the oral subcutaneous and intramuscular routes However these possibilities remain as yet untried 5.7 Solid Surfactant One serious problem facing the pharmaceutical scientist is the formulation of poorly water-soluble drugs Frequently the oral bioavailability of such drugs is very poor as a result of their slow dissolution in the aqueous medium of the gastrointestinal tract It is difficult to find water-soluble excipients that completely dissolve the active ingredients after addition of water that maintain the drug in solution for long periods even upon dilution do not impair absorption and are non-toxic In an J LAWRENCE attempt to solve these problems one group has examined the production of a solid solution of drug (cyclosporin) and non- toxic surfactant (sugar esters) 44 The loading of the poorly- soluble drug used was about 14 wt% The solid solution readily dissolves in the contents of the small intestine to form a clear micellar solution of solubilizate Presenting a drug in this way should overcome some of the problems inherent when using micellar formulations Furthermore as the solution is solid it should be relatively easy to enclose the drug within a capsule although as the formulation is hygroscopic care will be needed when storing the capsules An added advantage is the possibility of directly compressing the solid solution to form tablets Unfortunately while the method has significant benefits it is limited to surfactants that are solid at room temperature although a solid solution can be produced from a liquid surfac- tant such as Tween SO through admixture of a polymer such as polyoxyethylene glycol 45 The technique is also limited to drugs that are readily soluble in the micellar solution formed from the surfactant in the small intestine 6 Conclusion A number of equilibrium surfactant structures and related systems have considerable potential as delivery systems for a wide range of drugs Some of the more unusual aggregates such as fibre gels have no obvious use in pharmacy at the moment but may prove to be exploitable in the near future Most surfactant systems with the exception of ligand modified vesicles have little potential as targetting devices Before attempting to formu- late a drug the limitations of each type of system need to be thoroughly understood For example it is no use trying to increase the aqueous solubility of a water-soluble hydrophilic drug in an aqueous-based surfactant system However it may be beneficial to formulate a hydrophilic drug in a surfactant system if a protective effect or a sustained release is required Similarly it would be of little advantage formulating a drug requiring a very high dose in a micellar solution The most serious problem with formulating drugs in surfac-tant systems is the paucity of suitable biodegradable surfactants commercially available Until this situation is rectified surfac- tants will not live up to their full potential as delivery vehicles and possibly as targetting systems With all the current interest in the area there is hope that this situation will at least in part be rectified in the near future 7 References 1 D Attwood and A TFlorence ‘Surfactant Systems Their Chemistry Pharmacy and Biology’ Chapman and Hall London 1983 2 A T Florence ‘Techniques of Solubilization of Drugs’ ed S H Yalkowsky Marcel Dekker Inc ,New York 1981 3 E Tomlinson in ‘Site-specific Drug Delivery’ ed E Tomlinson John Wiley and Sons Ltd ,Chichester 1986 4 A Y Ozer A A Hincal and J A Bouwstra Eur J Pharm Biopharm 1991,37 75 5 J Israelachvili ‘Intermolecular and Surface Forces’ 2nd Edition Academic Press London 199 1 6 K Fontell Colloid Polym Sci 1990,268 264 7 D J Mitchell G J T Tiddy L Waring T Bostock and M P McDonald J Chem Soc Faradav Trans I 1983,79,975 8 A S Rudolph J M Calvert M E Ayers and J M Schnur J Am Chem Soc 1989,111,8516 9 J -H Fuhrop S Svenson C Boettcher E Rossler and H -M Vieth J Am Chem Soc 1990 112,4307 10 S Engstrom and L Engstrom Int J Pharm 1992,79 113 11 M J Lawrence in ‘Controlled Particle Droplet and Bubble Forma- tion’ ed D J Wedlock Butterworth-Heinemann Oxford 1994 12 H Kunieda K Nakamura and D F Evans J Am Chem SOC 1991,113 1051 13 M Corti L Cantu and P Salina Adv Colloid Interface Scr 1991 36 153 14 H Kunieda K Nakamura U Olsson and B Lindman J Phys Chem 1993,97,9525 15 D Cochin F Candu and R Zana Macromolecules 1993,26,5755 16 P Strom J Colloid Interface Sci 1992 154 184 17 J H Fendler and P Tundo Acc Chem Res 1984,17,3 18 W J Bailey and L -L Zhou ACS Symp Ser 1991,469,285 19 M 0 Vaizoglu and P P Speiser Acta Pharm Suec ,1986,23 163 20 M Yokoyama M Miyauchi N Yamada TOkano Y Sakurai K Kataoka and S Inoue J Control Re1 1990 11,269 21 S A Charman W N Charman M C Rogge. T D Wilson F J Dutko and C W Pouton Pharm Res 1992,9 87 22 T Yoshioka and A T Florence Int J Pharm ,In/ J -Pharm 1994 108 117 23 S Engstrom L Lindahl R Wallin and J Engblom Int J Pharm 1992,86 137 24 S Wahlgren A L Lindstrom and S E Friberg J Pharm Sci 1984,73 1484 25 S M Lawrence M J Lawrence and D J Barlow J Mol Graphics 1991,9,218 26 P H Elworthy and M S Patel J Pharm Pharmacol ,1982,34,543 27 D Attwood P H Elworthy and M J Lawrence J Pharm Pharmacol 1989,41 585 28 M J Lawrence and F Devinsky J Pharm Pharmacol 1988,40 125P 29 M N Azmin A T Florence R M Handjani-Vila J F B Stuart G Vanlerberghe md J S Whittaker J Pharm Pharmaocol ,1985 37,237 30 J C Gilbert J Hadgraft A Bye and G Brookes Int J Pharm 1986,32,223 CHEMICAL SOCIETY REVIEWS 1994 31 P -C Chen-Chow and S G Frank Znt J Pharm 198 1,8,89 32 B Ericsson P 0 Erikson J E Lofroth and S Engstrom ACS Symp Ser 199 1,469,25 1 33 S Miyazaki C Yokouchi TNakamura N Hashiguchi W -M Hou and M Takada Chem Pharm Bull 1986,34 1801 34 L Martini D Attwood J H Collett C V Nicholas S Tanodek-aew N -J Deng F Heatley and C Booth J Chem SOC Farada-y Trans 1994,90 1961 35 H G Ibrahim J Pharm Sci 1989,78,683 36 H E Bodde T De Vringer and H E Junginger Progr Colloid Polymer SCI 1986,72 37 37 C C Mueller-Goymann and S G Frank Int J Pharm 1986,29 I47 38 C C Mueller-Goymann and H -J Hamann J Control Re1 1993 23 165 39 F R Service Science 1994 252 3 16 40 H E J Hofland J A Bouwstra G S Goons F Spies H Talsma and H E Junginger J Colloid Interface Sci 1993 161 366 41 H E J Hofland J A Bouwstra M Ponec H E Bodde F Spies J Coos Verhoef and H E Junginger J Control Re1 1991 16 155 42 Patent application A6 1 K9/ 127 43 R E Evans and S J Farr J Biopharm Sci 1992,3,33 44 L Hahn and H Sucker Pharm Res 1989,6,958 45 K R Morris G T Knipp and A T M Serajuddin J Pharm Sci 1992,81 1185

ISSN:0306-0012    

DOI:10.1039/CS9942300417

出版商:RSC    

年代:1994

数据来源: RSC