摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Dr. A. R. Butler Professor B. T. Golding Professor M. Green Professor D. M. P. Mingos FRS Professor J. F. Stoddart Consulting Editors Dr. G. G. Balint-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor A. Hamnett Dr. T. M. Herrington Dr. R. Hillman Professor R. Keese Dr. T. H. Lilley Dr. H. Maskill Professor Dr. A. de Meijere Professor J. N. Miller Professor S. M.Roberts Professor B. H. Robinson Ds. A. J. Stace Staff Editors Mr. K. J. Wilkinson Dr. J. A. Rhodes University of Sussex University of Leicester University of St. Andrews University of Newcastle upon Tyne University of Bath Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Reading University of Bristol University of Bern University of Sheffield University of Newcastle upon Ty University of Gottingen Loughborough University of Tec University of Exeter University of East Anglia University of Sussex ie 1no logy Royal Society of Chemistry, Cambridge 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 and further afield, in order to present a truly international outlook on the major advances in a wide range of chemical areas. It is hoped that it will be of interest and help to students planning a career in research. In particular, it will be the place to find succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be comprehensive, too detailed, or heavily referenced; they should act as a springboard to further reading. Interdisciplinary awareness should thereby be heightened and the student should be stimulated to take a more professional interest in the topic of the review. Although the majority of articles are intended to be specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, 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, 1992 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 Blackbear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99221FX005

出版商:RSC    

年代:1992

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99221BX007

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Artemisinin (Qinghaosu): A New Type of Antimalarial Drug Anthony R. Butler Department of Chemistry, University of St. Andrews, St. Andrews, Fife KY16 9Sr Scotland Yu-Lin Wu Shanghai Institute of Organic Chemistry, Academia Sinica, Shanghai 200032, China 1 Introduction Malaria is probably older than mankind. Intermittent fevers with enlargement of the spleen have been described in medical writings from many civilizations since earliest times. In Britain the word ‘ague’ was used to describe the condition until the nineteenth century. The Hippocratic writings give one of the first descriptions, noting the different periods of onset -subter-tian, tertian, and quartan -and calling attention to the enlarged spleen. It is possible that malaria contributed to the decline of Greek civilization.Certainly malaria was a serious problem around the city of Rome, in the Roman Campagna, and near the Pontine marshes. Within the British Isles even Scotland was affected and the English fen country was badly infected. London was plagued until 1864 when the Thames embankment was built and flooding controlled. Information about the incidence of malaria outside Europe until the sixteenth century is scanty but it has been widespread during recent times. Coastal Africa, both east and west, was malarial but a natural immunity to the especially virulent form of the malarial parasite, Plasmodium falciparum, exists in some 10-30% of the indigenous popula- tions of central and western areas of Africa who carry the sickle cell trait in their red blood cells.However, European explorers, missionaries, and traders were, and still are, badly affected by this type of malaria. Malaria occurred extensively in India during colonial times. In ancient Chinese books it was recorded that malaria was widespread in areas south of the Yangtze River, especially amongst those newly immigrated from North China. It remained common in these areas, including the coastal cities of Shanghai and Kwangchow, until the beginning of the twentieth century. Now malaria is no longer a problem except on Hanan Island and in some districts along the southern border. For centuries malaria was a major problem in Central and South America, and in the rural south of the USA it was not eradicated until 1960.2 Malarial Parasite and the Mosquito as Vector Various writers had noted the connection between marshes with stagnant, smelly pools and malaria. Indeed, the word malaria comes from the Italian mal’aria or ‘bad air’ and the ancient Yu-Lin Wu was born in Zhejing, China in 1938. He graduated jrom Jianlin University in 1962 and later as a postgraduate from the Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, where he is currently research professor and the director of the State Keji Laboratory of Bio-organic and Natural Products Chemistrj?. His research interests are in the area of natural products and organic synthesis. Anthony R. Butler graduated from King’s College, University of London in 1958 and studied with Victor Gold for his Ph.D.He is currently Reader in Chemistry at the University of St. Andrews. His main research interest is the application of chemistry to medical problems. Chinese term zhangqi means much the same thing. The reason for the connection became clear only at the end of the nineteenth century. A French army surgeon working in Algeria, Alphonse Laveran, discovered that the condition is due to members of a family of protozoal parasites named Plasmodium which enter red blood corpuscles when in the human blood stream. Sir Patrick Manson, while working in China, found evidence that this parasite was carried by mosquitoes and that people became infected when bitten. These ideas were confirmed by Ronald Ross, a doctor in the Indian Medical Service, and announced in 1898.Further studies by a number of workers showed that humans are affected by four species of Plasmodium, three of which produce the mild forms of malaria by destroying red blood cells in peripheral capillaries and thus causing anaemia. The bouts of fever correspond to the reproductive cycle of the parasite. However, the most dangerous is the fourth species, Plasmodium ,faleiparum.In this case the infected red blood cells become sticky and form clumps in the capillaries of the deep organs of the body and cause microcirculatory arrest. If this happens in the brain delirium, coma, convulsions, and death may ensue. Cerebral malaria is by far the most serious form of the infection.Immunity to all four types of malaria is eventually acquired over a number of years as a result of frequent attacks of the condition and is maintained whilst the individual remains in the malarious area. Small children and those newly resident are therefore particularly vulnerable to attacks until their immunity is functional. 3 Eradication of Malaria Once the role of the mosquito had been established its eradica- tion became the key factor in the control of malaria. The draining of marshes to remove the mosquitoes’ habitat is highly effective but too expensive to implement in poor countries. The use of DDT to kill mosquitoes was, at one time, widely practised but the toxic properties of DDT, its increased cost following the oil crisis of 1974, and the appearance of DDT-resistant mosqui- toes have led to the re-emergence of malaria as a major world health problem.To make matters worse, strains of P. falciparum resistant to the principal antimalarial drug chloroquine and to some of the prophylactics have been detected in mosquitoes. Thus drugs can no longer treat all forms of malaria and it now seems unlikely that it will be possible, on a worldwide scale, to eliminate all the habitats of the mosquito by ecologically accep- table means. The size of the resulting problem can be seen from the statistics of malarial incidence in India: as a result of the eradication programme by 1961 there were fewer than 100000 cases, by 1971 the number had risen to 1 400 000 but by 1977 this had increased catastrophically to 50 million, many cases due to infection by P.falciparum. There were similar increases in other parts of the world and the overall total must be at least 200 million. In the absence of an effective vaccine the only answer seems to be improved drugs for the destruction of the malarial parasite, particularly P. falciparum, while in the human host. In the early 1980s the need for a new antimalarial drug became a matter of great urgency as malaria is, once again, the most important of all tropical diseases. 85 CHEMICAL SOCIETY REVIEWS, 1992 cH307p Hh-CHCH2CH2CH2NH2 I CH3 (7) 4 Quinine as an Antimalarial Drug The bark of the cinchona tree was used in South America by the indigenous people for the control of fevers.It was introduced into Europe by an Augustinian monk who had lived in Peru, Father Antonio de la Calancha, in the seventeenth century and proved very successful in the treatment of malaria. As it was distributed by Jesuits it became known as ‘Jesuits’ bark’. Cin- chona bark must be one of the most successful of all herbal remedies and illustrates the value of folk medicine. The active principle, quinine, was isolated by Pelletier and Caventou in 1820 and found to be more palatable than the nauseating powdered cinchona bark. The chemical structure of quinine (1) was elucidated in 1908 and key steps in its total synthesis were achieved by Woodward and Doering in 1944. 5 Synthetic Antimalarial Drugs During the 1920s a synthetic quinoline derivative, pamaquin (2), was found to be more effective than quinine in killing malarial parasites lodged in the liver.Also mepacrine (3) was developed as a synthetic alternative to quinine. During World War I1 there was a desperate need for alternatives to quinine as the Japanese cut off the supply of cinchona bark from Java and so the production of mepacrine was greatly increased. Further research led to the production of chloroquine (4) which has fewer side effects and does not turn the patient yellow. Prima- quine (5) is another quinoline derivative with antimalarial properties particularly effective against P. viva, the cause of benign tertian fever. A biguanidine compound proguanil(6) also has powerful antimalarial properties but is more generally used as a prophylactic. A pyrimidine derivative (7), pyrimethamine, by itself is used for suppression only.For treatment it is used in combination. Most drugs for the treatment of malaria are derivatives of quinoline and acridine and, until recently, there was no alterna- tive chemotherapy. Unfortunately none of the drugs mentioned above is particularly effective against P.falciparum, the form of malaria now making a comeback on the world scene. There is an additional problem facing the health authorities in the countries affected: malaria is an orphan disease. It occurs in countries too poor to pay the enormous development costs of a commercial drug. Fortunately a new lead compound has appeared and early prospects are very promising.(5) 6 Treatment of Malaria in Traditional Chinese Medicine An important part of traditional medicine in China is the use of herbs for the treatment of disease. In this, of course, it does not differ from the medicine of medieval Europe. Collections of Chinese herbal remedies or bencao have been published over the centuries. The first bencao, Shen Nong Bencao Jing, was pub- lished in the late Han (100-200 AD) and many others appeared at intervals up to modern times. The most comprehensive is probably the Bencao Gangmu or ‘Systematic Materia Medica’ by Li Shizhen published in 1596 AD. The first pharmacopoeia (i.e. a volume regulating the use of drugs or yaodian) in Chinese was published in 1930 AD and in it the authors dismissed the bencao as ‘nothing but waste paper’. Fortunately, since then traditional Chinese medicine has been restored to something like its former status -not instead of Western medicine but comple- mentary -and the bencao seen as valuable sources of infor- mation on traditional cures tested over several thousand years.In view of the uniqueness of Chinese flora and the early invention of printing in China (ca. 700 AD), which means that the bencao are less prone to copying errors than European herbals, it is not surprising that a number of interesting new lead compounds have been obtained. The classical remedy for malaria in traditional Chinese medi- cine is the root of Dichroafebrfuga Loureiro (changshan in Chinese). This material was investigated as part of the effort during World War 11 to find alternatives to quinine.An alkaloid, febrifugine (8), was extracted from the root and found to be 100 times more effective against P. cynomolgi than quinine. How-ever, even at subtoxic levels, febrifugine is a powerful emetic and this has effectively precluded its general clinical use in the treatment of malaria. Another traditional cure for fever, first mentioned in the bencao nearly 2000 years ago and many times subsequently, is the herb qinghao or caohao, Artemisia annua L. (sweet worm- wood or annual wormwood), a weedlike plant growing over large parts of China. In the Bencao Gangrnu it is stated rather concisely that qinghao ‘can cure malaria, fever, and cold’. The use of qinghao for the treatment of malaria is also described in ‘The Barefoot Doctor’s Manual’, a manual which did so much to improve primary health care in China.It was to this tra- ditional herbal remedy that Chinese scientists turned in an effort to fight the resurgence of malaria. 7 Isolation of Artemisinin Extraction of the dried leaves of qinghao with petroleum ether at low temperature and chromatography on silica gel with subse- quent recrystallization gave fine, colourless crystals, named ARTEMISININ (QINGHAOSU): A NEW TYPE OF ANTIMALARIAL DRUG-A. R. BUTLER AND YU-LIN WU qinghaosu (extract of qinghao) in Chinese but also given the Western name artemisinin. The yield was variable ranging from negligible quantities to almost 1O/O depending on the area from which the plant was collected.The formula C15H2205 sug- gested the compound to be a sesquiterpene and reaction with triphenylphosphine to give the phosphine oxide was consistent with the presence of a peroxide group. The structure of artemisi- nin, as well as its absolute configuration, was determined by X-ray diffraction as (9).l The lactone ring has a trans configu-ration. The most unusual feature of the chemical structure is the 1,2,4-trioxane ring which may also be viewed as a bridging peroxide group. Artemisinin is the only known 1,2,4-trioxane occurring in nature, although compounds with peroxide bridges are common, particularly in marine organisms. In extensive clinical trials in China2 artemisinin showed promise in the treatment of otherwise drug-resistent forms of malaria, notably P.jafciparum. This discovery has occasioned a considerable amount of research, in both China and the West, into the synthesis, biosynthesis, and biological action of artemisinin and related compounds. The rest of this review will attempt to give an account of some of that work. 0 (9) 8 Nomenclature Although the Chinese name qinghaosu, meaning extract of green plant, is attractive it is frequently misspelt. In a recent edition of a popular textbook on tropical medicine it is referred to as Quing Ha0 Hsu; in the modern pinyin romanization of Chinese ‘q’ is not followed by ‘u’and ‘hs’ is quite different from ‘s’. In qinghaosu the ‘q’ is pronounced like ‘ch’ in ‘cheat’.The systematic name is 3,6,9-trimethyl-9,1 Ob-epidioxyperhydropyr- ano[4,3,2-jk]benzoxepin-2-onebut this is hardly convenient for regular use. The name adopted by Chemical Abstracts is artemis- inin, derived from the plant which is its source, and artemisinin will be used in this review. The systematic numbering is shown in (9), but other systems have been used. 9 Total Synthesis of Artemisinin The rather unusual structure of artemisinin has meant that the molecule constitutes a stimulating synthetic challenge and a number of successful total syntheses have been reported. In most cases the trioxane ring has been formed by addition of singlet oxygen to an olefin in the presence of a photosensitizer followed by protonation and reaction with a carbonyl compound (Scheme 1).This approach to the synthesis of trioxanes has been fully explored by Jefford and his co-~orkers.~ Scheme 1 For the application of this procedure to the synthesis of artemisinin Schmid and Hofheinz4 started with ( -)-isopulegol (10) and the synthesis was completed in ten steps. The key intermediate was a benzyloxymenthone and the photosensitizer was Methylene Blue. Zhou and co-workers5 started with (+)-citronella1 (11) and the intermediate (12) was prepared in 19 steps. Photooxidation using Rose Bengal and acidification gave artemisinin in 28% yield from (12). Avery and co-workers6 used the cyclohexane (13) as the starting material and the peroxide bridge was introduced by the abnormal course of reaction of a vinylsilane with ozone at -78 “C.A formal stereoselective synthesis starting with (+ )-car-3-ene (14) has been devised by Ravindranathan and co-worker~.~ One of the key steps was an intramolecular Diels-Alder reaction for the conversion of (I 5) into an epimeric mixture of ethers (16). Treatment of the mixture with MCPBA gave a single epoxide, reduction of which with LAH produced a single tertiary alcohol. The synthesis pro- ceeded without stereochemical problems to give (17), the con- version of which into artemisinin had been accomplished in the synthetic route of Zhou et ~1.~mentioned above. H cH3 Although not a total synthesis there is an important pro- cedure, described by Wu and Ye,* for the conversion of artemisi- nic acid (18), which is a relatively abundant constituent of Artemisia annua, into artemisinin by photooxidation of a cyclic enol ether in the presence of Methylene Blue followed by treatment with trime t hylsil yltrifluorome thanesulfona te and regeneration of the carbonyl group at the 2-position by oxi- dation with ruthenium chloride-sodium periodate.10 Chemical Reactions of Artemisinin The most interesting aspect of artemisinin’s chemistry is the stability of the peroxide bridge. Indeed, artemisinin can be recovered unchanged from neutral solvents after several days at temperature up to 150 "C.At higher temperatures the peroxide bridge is destroyed and (19), (20), and (21) are formed, albeit in low yield^.^ A different range of products is obtained on dry heating.CH3 CH3 0+CH3 0 0wCH3 0+CH3 0 Reaction of artemisinin with acid in methanol produces mainly (22), which is converted by more concentrated acid into the diketone (23), a useful intermediate for the relay synthesis of artemisinin and its analogue. Compound (22) can be recyclized into artemisinin by TFA.l0 Artemisinin is quite sensitive to alkali and reaction with potassium carbonate results in substan- tial reorganization of the molecule to give, eventually, (24) as one product. In terms of the medicinal uses of artemisinin the most important reaction is its reduction. Lithium aluminium hydride exhaustive reduction gives a tetrahydroxy compound (25).Reaction with hydrogen over palladium results in loss of the peroxide bridge to give I I-deoxyartemisinin (26). However, reduction with borohydride leaves the peroxide bridge intact, giving the lactol2-hydroxy-2-deoxoartemisinin(27a). This is an important intermediate in the synthesis of artemisinin analogues. CH3 H3C0 0 H3COOC 11 Derivatives and Analogues of Artemisinin The naming of the derivatives of artemisinin is a matter of some confusion. In this review we have used the systematic numbering shown above but this may differ from that used in the original paper. The lactol (27a) has been used to prepareI3 nearly 50 derivatives [ethers (27b), esters (27c), and carbonates (27d)l of CHEMICAL SOCIETY REVIEWS, 1992 H3C 0-0+CH3 0+H3 0 OR (27)(a) R = H (b) R = alkyl (c) R = COR' (d) R = COOR' (e) R = OCOCHzCH2COOH (1) R = OCH~CGH~COOH artemisinin as part of a screening programme of antimalarial activity.Two other interesting derivatives of the lactol with water-solubilizing groups are artesunic acid (27e) and artelinic acid (27f).14 The antimalarial activity of all these compounds will be discussed later. 2-Deoxo-l l-deoxyartemisinin (28) has been prepared from artemisinin' and 1I-deoxyartemisinin (26) from artemisinic acid (18) by singlet oxygenation in the presence of Methylene Blue at room temperature. There has been a total synthesis of 3,6-didemethylartemisinin (29) from pyrrolidinocyclohexene and 1,4-dichlor0-2-butene in 8 steps.The peroxide bridge was introduced by ozonolysis at -78 "C with subsequent acid treatment using Amberlite 15. Similarly, 3,3a-secoartemisinin (30) has been prepared. A number of amino derivatives (3 1) have been prepared from the lactol(27a) by elimination of the elements of water, addition of bromine to give the 2,3-dibromo compound, and subsequent reaction with the appropriate amine, both benzenoid and hetero- aromatic. l9 An important derivative in attempts to understand the mechanism of artemisinin's antimalarial activity is desethano- artemisinin (32) in which the trioxane structure is retained but one of the rings has been lost. It was prepared from citronella1 (I 1) and the sensitizer for photooxidation was Rose Bengal.20 Three carba-analogues of artemisinin, [(33)-(35)], have been prepared.Artemisinic acid (1 8) was converted into a cyclic enol ether in four steps. Reaction of this with paraformaldehyde in the presence of boron trifluoride etherate afforded (33), and oxidation of (33) with ruthenium oxide gave (34) and (35).21 NHAr ARTEMISININ (QINGHAOSU): A NEW TYPE OF ANTIMALARIAL DRUG-A. R. BUTLER AND YU-LIN WU H 3 HiC.C 9 H 3ooc--C q 0 CH3 CH3 (33) (34) H3C ooc --o+ CH3 0 (35) 12 Biosynthesis of Artemisinin Artemisinin appears to be unique to Artemisia annua and is at a maximum in the upper leaves at the beginning of budding. Labelling experiments have identified two intermediates en route to artemisinin: mevalonic acid lactone (36) and artemisinic acid (1 8).22It has also been establishedz3 that isopentenyl pyrophos- phate is incorporated in artemisinin obtained from cell cultures of A.annua. Otherwise, little appears to be known about the biosynthetic pathway. The conversion of artemisinic acid (1 8) and dihydroartemisinic acid into artemisitene (37) may be significant in delineating the biosynthetic pathway.24 o+CH2 0 (37) 13 Antimalarial Activity of Artemisinin It is almost certain that the crucial structure in artemisinin which gives it its antimalarial activity is the peroxide bridge. Other parts of the molecule may be modified without loss of antimalar- ial activity. Removal of one or two of the methyl groups at positions 3 and 6, as in (29), leaves a molecule which is still active against P.falciparum.l7 The lactol(27a) obtained by reduction of artemisinin, although chemically unstable, is a better antima- larial agent than artemisinin itself, indicating clearly that the carbonyl group is not the essential structure.I5 On the other hand, 1 1 -deoxyartemisinin (26)16 and 2-deoxo-l l -deoxyar-temisinin (28)' are not active against the malarial parasite, as neither are the carba-analogues (33), (34), and (35).20 All these compounds lack the peroxide bridge.The converse appears also to be true. Many compounds not obviously related to artemisi- nin but which contain a peroxide group are antimalarial agents. Vennerstrom and co-workers2 have examined 23 peroxides of diverse chemical structures, including di-t-butyl peroxide, ascar- idole (38), and dihydroascaridole (39).Several are active in vitro against P. falciparum, although none is active in vivo. This suggests that the remainder of the artemisinin molecule is responsible for the delivery of the drug to the infected erythro- cyte in a still active form where it can exercise its toxicity towards the parasite. However, even in compounds closely related to artemisinin the possession of a peroxide bridge is not of itself sufficient condition for antimalarial activity.l0 In this regard information on the antimalarial activity of desethanoartemisi-nin (32), in which the peroxide bridge has been retained but one of the rings has been lost, is awaited with interest.20 CH3 y3I CH3ACH3 CH3&H3 (38) (39) Although the peroxide bridge may be the crucial structure in artemisinin, the rest of the molecule has a profound effect on the in vitro and in vivo antimalarial activity of artemisinin and related compounds.The ether derivative of 2-hydroxy-2-deoxo-artemisinin (27b, R = Et) is a better antimalarial agent than artemisinin itselP6 and is the form of artemisinin which is currently under commercial development; it is generally called arte-ether. Esters (27c, R' = alkyl) are generally as effective as artemisinin but the corresponding acids (27c, R' = H) are much less so. This may be a consequence of their lower solubility in lipids. Carbonates (27d, R' = alkyl) are the least effective of this group of comp~unds.'~ Sodium artelinate [the sodium salt of (27f)l is only slightly less effective than artemisinin in vitro and has the advantage of water solubility and can be administered orally.27A number of highly effective antimalarial agents have been obtained by replacing the carbonyl group of artemisinin by a substituted amino-group (31).18 The most potent is that with Ar = 3-fluorophenyl; the presence of bromine does not seem to have any deleterious effect.Quantum mechanical calculations give excellent correlation between calculation and experiment in cases where the structure has been determined by X-ray crystallography.28 Thus it may be possible to use quantum pharmacology to explore as yet unsyn- thesized compounds in the search for an even better antimalarial drug.14 Mode of Action Artemisinin and related compounds are normally administered as a solution or suspension in oil by intramuscular injection. From studies of artemisinin it was found that absorption is rapid with a peak serum level after two hours, an elimination half life of 1.6hours, and a mean retention time of 3.3 hours. None was detected in plasma after rectal or oral admini~tration.~~ The action of artemisinin on the malarial parasite appears to be completely different from that of chloroquine and this may well be why artemisinin is effective against parasites which have become chloroquine-resistant. Most research suggests that artemisinin acts by an oxidative mechanism and it effects changes in both red blood cells and in the limiting and other membranes of the malarial parasite.At concentrations much higher than those used clinically, artemisi- nin affects red blood cell deformability in a manner which suggests that it is acting as an efficient pro~xidant.~~ At even higher concentrations artemisinin brings about complete lysis of the red blood cells.31 The effect of artemisinin on parasite membranes after even a single dose to a mouse host is quite substantial within half an hour. There are alterations in riboso- mal organization and endoplasmic reticulum. Nuclear mem- brane blebbing develops after one hour and segregation of the nucleoplasm after three hours. Further degenerative changes, with disorganization and death, occur from 8 hours onwards.The morphological changes in ribosomes and endoplasmic reticulum correlate in time with the in vitro depression of protein synthesis observed in P. falciparum. Similarly, the onset of nucleoplasmic segregation correlates with the development of nucleic acid synthesis inhibition. Some evidence suggests that the drug may be localized in the membranes so that changes in membrane integrity might precede the early depression of protein synthesis.32 Changes in the phosphorus, potassium, and sodium contents of infected red blood cells occur on treatment with artemisinin consistent with alterations in the membrane.33 At a molecular level it is possible that the oxygen-oxygen bond of the peroxide bridge is broken due to an electron transfer with the generation of an oxygen-centred radical (Scheme 2).This radical species could be responsible for the destruction of the membranes of Plasmodium. Quantum mechanical calculations have shown that transfer of an electron to the peroxide moiety of artemisinin results in considerable lengthening of the oxygen- oxygen bond. * Scheme 2 15 Future Prospects It seems probable that artemisinin and its derivatives will play a substantial part in the fight against the resurgence of malaria, particularly falciparum malaria, throughout the world. In the absence of a vaccine the situation now is rather like that in Europe before the introduction of quinine. Clinical trials of arte-ether (27b, R = Et) are almost complete. Although artemether (27b, R = Me) is equally potent and both are soluble in oil, which is the normal mode of administration, arte-ether was selected for development for commercial reasons and also because metabolism in the body gives ethanol rather than the more toxic methanol.For commercial production artemisinin is obtained by extraction from cultivated A. annua and converted into arte-ether by reduction with sodium borohydride in metha- nol and subsequent reaction with ethanol and boron trifluoride ethe~ate.~~The more active form is the /3-epimer and this can be separated from the mixture of epimers either by column chroma- tography or fractional crystallization. The latter is more suitable for commercial production. The Kunming Pharmaceutical Company in Southern China is already in commercial produc- tion, while the Guilin Pharmaceutical Company produces sodium artesunate (27e), another effective antimalarial agent which has the advantage of water-solubility.But artemisinin and its derivatives have shortcomings as antimalarial agents. Although partially solved, low solubility in both oil and water is still a problem. Efficacy by oral administ- ration is very poor and administration by injection is a problem in countries where there are inadequate medical facilities. There is also a high rate of recrudesence in treated patients. Thus there is much work still to be done in providing the world with a really successful new antimalarial drug. In 1692 the Chinese Emperor Kangxi was cured of malaria by visiting Jesuits using cinchona bark.Today the tables have been reversed and China has provided the world with new hope in the fight against this most widespread of tropical diseases. Acknowledgements. The authors thank Dr. Kevin McCullough of Heriot Watt University for helpful discussions and Dr. Alan McNaught of the Royal Society of Chemistry for advice on nomenclature. 16 References The order of Chinese authors' names has been changed to that ofwestern custom. The names of most Chinese journals are given in romanized CHEMICAL SOCIETY REVIEWS, 1992 Chinese rather than in the latinized version; Huaxue Xuebao is Acta Chimica Sinica, Yaoxue Xuebao is Acta Pharm. Sinica, and Zhongguo Yaoli Xuebao is Acta Pharmacol.Sinica. Most references to Chinese journals are to the English language editions. 1 Qinghao Research Group, Kexue Tongbao, 1977,22, 142. 2 Qinghao Antimalarial Coordinating Research Group, Chinese Med. J., 1979,92, 81 1. 3 C. W. Jefford, F. Favarger, S. Ferro, D. Chambaz, A. Bringhen, G. Bernardinelli, and J. Boukouvalas, Helv. Chim. Acta, 1986,69, 1778. 4 G. Schmid and W. Hofheinz, J. Am. Chem. SOC., 1983,105,624. 5 Xing-xiang Xu, Jie Zhu, Da-zhong Huang, and Wei-shan Zhou, Tetrahedron, 1986,42, 819. 6 M. A. Avery, C. Jennings-White, and W. K. M. Chong, Tetrahedron Lett., 1987, 28,4629. 7 T. Ravindranathan, M. A. Kumar, R. B. Menon, and S. V. Hire- math, Tetrahedron Lett., 1990, 31, 755. 8 Bin Ye and Yu-lin Wu, J. Chem.Soc., Chem. Commun., 1990, 726. 9 X. D. Luo, H. J. C. Yeh, A. Brossi, J. L. Flippen-Anderson, and R. Gilardi, Heterocycles, 1985,23,881; A. J. Lin, D. L. Klayman, J. M. Hoch, J. V. Silverton, and C. F. George, J. Org. Chem., 1985, 50, 4504. 10 Ying Li, Pei-lin Yu, Yi-xin Chen, Jing-li Zhang, and Yu-lin Wu, Kexue Tongbao, 1986,31, 1038. 11 Mei-yi Zeng, Lan-na Li, Shu-feng Chen, Guang-yi Li, Xiao-tian Liang, M. Chen, and J. Clardy, Tetrahedron, 1983,39, 2941. 12 Jing-ming Liu, Mu-yun Ni, Ju-fen Fen, You-you Tu, Zhao-hua Wu, Yu-lin Wu, and Wei-shan Chou, Huaxue Xuebao, 1979,37, 129 13 Ying Li, Pei-lin Yu, Yi-xin Chen, Liang-quan Li, Yuan-zhu Gai, De- sheng Wang, and Ya-ping Zheng, Yaoxue Xuebao, 1981,16,429. 14 A. J. Lin, M. Lee, and D. L.Klayman, J. Med. Chem., 1989,32,1249. 15 M. Jung, X. Li, D. A. Bustos, H. N. ElSohly, J. D. McChesney, and W. K. Milhous, J. Med. Chem., 1990,33, 1516. 16 M. Jung, H. N. ElSohly, E. M. Croom, A. T. McPhail, and D. R. McPhail, J. Org. Chem., 1986, 51, 5417. 17 M. A. Avery, C. Jennings-White, and W. K. M. Chong, J. Org. Chem., 1989,54, 1792. 18 M. A. Avery, W. K. M. Chong, and G. Detre, Tetrahedron Lett., 1990,31, 1799. 19 A. J. Lin, Liang-quan Li, D. L. Klayman, C. F. George, and J. L. Flippen-Anderson, J. Med. Chem., 1990,33,2610. 20 Y. Imakura, T. Yokoi, T. Yamagishi, J. Koyama, H. Hu, D. R. McPhail, A. T. McPhail, and K. H. Lee, J. Chem. SOC., Chem. Commun., 1988,372. 21 Bin Ye and Yu-lin Wu, Tetrahedron, 1989, 45, 7287. 22 Jing-jian Huang, Feng-yi Zhou, Lian-fen Wu, and Gui-hui Zhen, Huaxue Xuebao, 1988, 383. 23 G. J. Kudakasseril, L. Lam, and E. J. Staba, Planta Med., 1987, 53, 280. 24 R. K. Haynes and S. C. Vonwiller, J. Chem. SOC., Chem. Commun. 1990,451. 25 J. L. Vennerstrom, N. Acton, A. J. Lin, and D. L. Klayman, Drug Design Delivery, 1989,4, 45. 26 Hao-ming Gu, Bao-feng Lii, and Zhi-xiang Qu, Yaoxue Xuebao, 1980, 1,48. 27 P. H. vanvianen, D. L. Klayman, A. J. Lin, C. B. Lugt, A. L. Engen, H. J. van der Laay, and B. Mons, Exp. Parasitol., 1990,70, 1 15. 28 A. R. Butler and C. Thomson, unpublished observations. 29 Kai-cun Zhao, Qi-ming Chen, and Zhen-yu Song, Yaoxue Xuebao, 1986, 21, 736. 30 M. D. Scott, S. R. Meshnick, R. A. Williams, D. T. Y. Chiu, H. C. Pan, B. H. Lubin and F. A, Kuypers, J. Lab. Clin. Med., 1989, 114, 401. 31 Gu Haoming, D. C. Warhurst, and W. Peters, Zhongguo Yaoli Xuebao, 1986,7,269. 32 D. S. Ellis, Z. Li, H. M. Gu, W. Peters, B. L. Robinson, G. Tovey, and D. C. Warhurst, Ann. Trop. Med. Parasitol., 1985,79, 367. 33 P. Lee, Z. Ye, K. van Dyke, and R. G. Kirk, Am. J. Trop. Med. Hyg., 1988,39, 157. 34 A. Brossi, B. Venugopalan, G. L. Dominguez, H. J. C. Yeh, J. L. Flippen-Anderson, P. Buchs, X. D. Luo, W. Milhous, and W. Peters J. Med. Chem., 1988, 31, 645.

ISSN:0306-0012    

DOI:10.1039/CS9922100085

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Calculating Molecular Spectra Jonathan Tennyson and Steven Miller Department of Physics and Astronomy, University College London, London WC1 E 6BT 1 Introduction Spectra, the absorption or emission of light at characteristic wavelengths, have long been known to provide accurate finger- prints of molecular species. Spectroscopic analysis is standard for characterizing species and for identifying and quantifying molecules in complex mixtures. It also provides information on such things as temperature and isotopic abundances. We are particularly interested in these properties as a diagnostic for ‘cool’ astronomical bodies. Astronomically, molecular spectra provide a unique handle on the physical conditions in environ- ments such as giant molecular clouds, planetary atmospheres, and cool stellar atmospheres.As a counter example infra red spectroscopy is widely used in the petroleum industry to monitor everything from the oxidation of oils in engines to the compo- sition of exhaust gases after combustion.’ The detailed information contained in the many transitions that characterize the spectrum of a molecule is a sensitive reflection of the underlying interactions within that molecule. Thus the rotational energy levels of a molecule are largely governed by the molecular geometry: microwave spectroscopy, the study of rotational transitions, has long provided the most accurate determination of molecular bond lengths. Conversely the vibrational energy levels of a molecule are determined by the ease with which the atoms can move relative to each other within the molecule: infra red spectroscopy, which covers the wave- lengths of the strongest vibrational transitions, provides, at least in principle, detailed and often very accurate information on how the atoms in a molecule interact.The interaction of atoms within a molecule is governed by the electronic potential energy surface of the system (see Section 2). Potential energy surfaces are important entities within chemistry Jonathan Tennyson studied Natural Sciences at King j. College Cambridge (B.A. 1977). In 1980 he completed a Ph.D. in Theoretical Chemistry at Sussex University under the supervision of Professor John Murrell. He then spent two years at the University of Nijmegen, The Netherlands, as a Royal Society Western Exchange Fellow followed by three years at the SERC Daresbury Laboratory where he worked on calculations of photo-ionization and electron molecule scattering.He moved in 1985 to the Department of Physics and Astronomy at University College London and was promoted to Reader in 1991. His current research interests include calculating the ro-vibra- tional spectra of molecules, the consequences of classical chaos on molecular spectra, electron (positron) molecule collision calcula- tions, and the calculation of molecular data for astrophysics (including observational studies). Jonathan Tennyson is the great, great grandson of AIfred Lord Tennyson. Steven Miller gained B.Sc. (1970) and Ph.D.(1975) degrees from Southampton University. Two years at UMIST as research fellow and two at Shefleld University were followed by seven years working as a political journalist before he joined the Molecular Physics Group at University College London to work on ro-vibrational transitions of triatomic molecules, including Hi. He has pioneered the use of Hi as an astronomical probe of planetary atmospheres and made the first identijication of Hi outside the solar system (in Supernova 1987A) in 1990. 91 as they determine not only spectra, but also reaction dynamics, transport properties, and interactions in the liquid and solid state. Although all these properties are sensitive to the potential, spectroscopy is undoubtedly the most accurate source of infor- mation on the potential.Unfortunately, except for diatomic molecules, there is no general method of extracting potentials from spectroscopic data. This leads to an alternative strategy. Potentials constructed by any means, from first principles application of quantum mechanics to guess work, are tested by using them to generate spectra which can be compared with observation. If the agreement is unsatisfactory the potential can be adjusted and the calculation repeated. Recorded spectra, particularly for polyatomic molecules, are often very complicated. They can contain many thousands of lines each corresponding to transitions between unknown levels. The process of assigning such spectra involves ascribing degrees of vibrational and rotational excitation to both the initial and final levels involved in the transition.Often this is quite straight- forward once a few assignments have been made but the first assignments may involve inspired guesswork and can be greatly aided by calculations. Similarly predictions of where the transi- tions can be found greatly aid the search for a particular species. Calculations of molecular spectra are thus important for testing and developing potential energy surfaces, interpreting laboratory data, and predicting spectra. However, even for a three-atom molecule, such as water, at room temperature the spectrum can contain a very large number of transitions. The total absorption of light as a function of wavelength, which is called the opacity of the system, is an important property of many bodies including the Earth’s atmosphere. A detailed knowledge of the opacity of the atmosphere is required to model the greenhouse effect.The peak of the Sun’s radiation is at optical wavelengths to which the Earth’s atmos- phere is largely transparent. The Earth is cooler which means that it re-emits radiation at longer wavelengths -mainly in the infrared. If this emission passes through the atmosphere then the earth will lose heat; if the atmosphere absorbs the emission then this heat is retained. This is a fine balance which is dependent on the components of gases in the atmosphere. Although CO, is well known as a greenhouse gas, many other atmospheric trace species such as methane, Chloro-Fluoro-Carbons (CFCs) and their derivatives, actually have a much greater potential to act as greenhouse gases.The generation of the many transitions involved in synthesizing opacities for these species is naturally handled computationally. We are currently embarking on such a project focusing on the behaviour of cool stellar atmospheres. 2 Potentials As mentioned above, potential energy surfaces form a common strand running through practically all areas of physical chemistry. Actually the very existence of these surfaces relies on an approximation, albeit one which usually is valid. Chemists generally consider molecules to be composed of electrons which are light and singly charged, and compound nuclei which are comparatively heavy (hydrogen is 1836 electron masses) and multiply charged.Given the large mass ratio, it is generally assumed that the electrons can relax instantaneously to any given nuclear geometry. It is within this framework, known as the Born-Oppenheimer approximation, that molecu- lar potential energy surfaces can be defined. The potential energy for a particular nuclear configuration, generally denoted I/, is a function only of the relative nuclear coordinates as the electronic motion is assumed to be fully adjusted to this geometry. Potentials can be calculated using first principles quantum mechanics or ab initio, by choosing a set of internuclear sepa- rations, solving the resulting Schrodinger equation for the electronic motion and repeating the procedure until a grid of points has been generated.These points must then be fitted to some continuous function to produce a full surface. Solution of the electronic Schrodinger equation is a formidable problem in its own right but one which has advanced rapidly with theoreti- cal developments and computer technology. It should be noted, however, that the generation of full ab initio potential energy surfaces for polyatomic molecules remains computationally expensive. The traditional method of extracting potentials from observed spectra has been via force constants. These constants underpin a theory of vibrational motion which assumes the nuclei undergo only small amplitude motion. The force con- stants represent the derivatives of the potential at the equili- brium geometry of the molecule and as such give a very high order representation of the surface at one point rather than a global surface.There have been a number of suggestions as to which coordinates give the best extrapolation of the potential away from equilibrium. But no completely satisfactory solution exists. An alternative method of obtaining potentials from experi- ment has been to make some initial guess at a potential function which contains parameters that can be optimized by comparison with experimental data. This procedure places great emphasis on the development of computationally efficient procedures for calculating spectra as this is the slowest step in the process.The method was first extensively used for atom-diatom van der Waals complexes such H,-X and HCl-X (where X = He, Ne, Ar, Kr).273 These weakly bound systems have the advantage that the relatively stiff diatomic vibrational mode can be neglected. The other vibrational modes undergo large amplitude motions which are poorly represented as an expansion about equili- brium. More recently this method has been applied to chemi- cally bound systems such as ~ater,~,~ H,S,5 and HCN.6 3 Diatomic Molecules Given a potential energy curve the calculation of rotation- vibration spectra for diatomic molecules is relatively straightfor- ward. For these systems the potential energy surface and hence also the vibrational wavefunction is one dimensional. This makes visualization of both these functions straightforward.Similarly the degree of vibrational excitation in a diatomic is given by the number of times the wavefunction passes through zero. These zero amplitude points are called nodes. A number of model potentials have been proposed for diato- mics. These include harmonic, Morse, and Lennard-Jones 6-1 2 potentials. The first two of these are illustrated in Figure 1. The harmonic potential is important not so much for its quantitative properties, but because most of the language used to interpret vibrational spectra is based upon it. Representing vibrational motion as harmonic oscillations is valid if the amplitude of these vibrational motions is small. In this approxi- mation vibrational energy levels are evenly spaced.However, the harmonic model neglects all possibility of molecular disso- ciation. This introduces so called anharmonic effects which usually cause the spacing between vibrational levels to decrease. Of course vibrational motion becomes increasingly large ampli- tude as dissociation is approached. The Morse potential, also shown in Figure 1, contains the basic properties of most diatomic molecules. It is strongly repulsive when the nuclei are close together, has a minimum at some intermediate distance and dissociates at long bond length. It is thus capable of representing basic anharmonic effects although for fully quantitative results more sophisticated func- tions are generally required. CHEMICAL SOCIETY REVIEWS.1992 Figure I Potential energy curves, V(R),for a diatomic with equilibrium bond length Reand dissociation energy D,.The blue curve is harmonic and the red curve is a Morse potential. Horizontal lines indicate vibrational energy levels. Both potentials have the same curvature (force constant) at R = Re. There are now many computational techniques for direct numerical integration of one-dimensional second-order differ- ential equations of type encountered for diatomic systems. This means that, given a potential, the rotation-vibration energy levels of a diatomic system can be routinely obtained. This is in contrast to polyatomic systems. 4 Coordinate Systems for Polyatomic MolecuIes If a molecule consists of N nuclei then 3N coordinates are required to describe the position of these nuclei in space.The translational motion of the molecule can be represented by the 3 coordinates of its centre-of-mass. Similarly for a non-linear molecule, rotational motion can be represented by a further 3 coordinates. This leaves 3N -6 (3N -5 for a linear molecule) coordinates to represent the vibrational motion of the molecule. These are usually described as internal coordinates. For diatomics the bond length of the molecule is also the natural internal coordinate of the system. Even for triatomic systems, however, there is no unique choice of internal coordi- nates which gives a good picture for all systems. Traditionally so called normal coordinates have been used to represent the motions of polyatomic molecules.These coordi- nates are obtained as the solution of the multi-dimensional harmonic problem assuming that only harmonic terms are retained in the molecular potential. But these coordinates are unsatisfactory in many cases. This is because, given enough internal energy, all molecules undergo large amplitude vibratio- nal motion which is not well represented within the harmonic approximation. In particular there is a technical problem that with these coordinates it is all too possible to leave the true domain of the problem. This is illustrated by Figure 1 where it can be seen that for high energies the harmonic potential crosses R = 0 and allows the vibrational wavefunction to sample nega- tive bond lengths.Normal coordinates have now been largely abandoned for accurate calculations on small molecules. Instead internal coor- dinates defined in terms of internuclear separations and asso- ciated angles are usually employed. Figure 2 illustrates 3 of many such triatomic coordinate systems. The internuclear coordinates (rl, r2, r3), Figure 2a, would appear a good coordinate choice for a number of triatomic systems. This is particularly so for the molecule Hi, discussed below, whose equilibrium geometry is an equilateral triangle and for which it is thus desirable to have coordinates reflecting the high symmetry of the system. Unfortunately these coordi- nates are inconvenient to work with as their allowed ranges are CALCULATING MOLECULAR SPECTRA-J.TENNYSON AND S. MILLER I 0 A3 r A2 Figure 2 Internal coordinates for triatomic systems: (a) bond length coordinates (rlrr2.r3);(b) bond length ~ bond angle coordinates (rl, r2,a);(c) scattering coordinates (r, R, 0). linked by a series of triangulation relationships of the form I r1 -. r3 I < y2 d r1 + r3. This makes numerical integration very difficult and as a consequence these coordinates are gener- ally not used. Bond length-bond angle coordinates (rl, r2, a), Figure 2b, give a good representation of many triatomics such as H20 and H2S. These coordinates have often been used for these mole- cules. Recently Radau coordinates, originally developed to study planetary motion about the Sun, have also been used for such systems.When the central atom is heavy, these coordinates are very similar to the bond length-bond angle coordinates but have the advantage that they yield a much simpler kinetic energy operator in the nuclear motion Hamiltonian. Scattering coordinates (Y, R, B), Figure 2c, like Radau coordi- nates, also give a simple (diagonal) kinetic energy operator. These coordinates got their name from their use to represent atom-diatom collisions: the intersection of Y and R being at the diatom centre-of-mass. They are appropriate not only for weakly bound atomcliatom van der Waals complexes but also for a molecule like HCN as the coordinates can link the equilibrium structure of the two linear isomers HCN and HNC. Even if there is no coordinate system in which the vibrational motion is separable, it is still usual to talk about a molecule having 3N -6 vibrational modes.Like the internal coordinates, how these modes are chosen is not unique. The process of partitioning the vibrational energy between the vibrational modes, known as assignment, will be discussed below. There are also many ways of defining the 3 coordinates that represent the overall rotational motion of the system. This is because these coordinate depend on how one fixes (‘embeds’) the x,y, and z axes to the molecule. Thus, for example, in the scattering coordinates of Figure 2c the z axis of the system is often embedded along either Y or R depending on what is most appropriate for the system.5 Hamiltonians The equations of motion for N interacting particles, or their quantum mechanical equivalent given by the Hamiltonian, can easily be written down. However, the removal of translational motion, combined with coordinate transformations to internal coordinates and a particular axis embedding means the Hamil- tonian itself must be transformed. Each time one defines a new set of internal coordinates or axis embedding it is necessary to construct a new Hamiltonian for the system. A general method for deriving these Hamiltonians has been developed by Sut- cliffe7 but its application is often algebraically messy. Unfortunate as this may seem, there is a more serious problem. The very process of defining internal coordinates and axis embeddings introduces geometries into the Hamiltonian where it is badly behaved.For example, these badly behaved geometries, or singularities, are often encountered when a molecule becomes linear. In this case the system only has 2 instead of 3 rotational degrees of freedom and 3N -5 instead of 3N -6 vibrational modes. Special care is thus required for any bent molecule which samples linear geometries. 6 The Variational Principle In its simplest form the Raleigh-Ritz Variational Principle states that for a given Hamiltonian operator the energy of any approximate wavefunction will always be greater or equal to the lowest (ground state) energy of the system. This theorem can also be extended to give upper bounds for the energy of excited states provided that the trial wavefunctions obey certain simple constraints.* The Variational Principle has been a rock upon which much of quantum chemistry has been founded.It allows trial wave- functions to be improved systematically with the best function being given unambiguously by the function with the lowest energy. Particularly powerful in variational calculations are basis functions. These allow the trial wavefunction to be expanded as linear combination of suitable functions: The coefficients, given by the vector c, can be varied until the best trial wavefunction is obtained. This procedure can be written in terms of matrices and is thus particularly well suited to computer implementations. These (‘secular’) matrices are diagonalized to yield the c’s as eigenvectors and energies as eigenvalues.In the secular matrix method, which is commonly used in all branches of quantum chemistry,* matrix elements are given by In this expression, His the secular matrix, fi the Hamiltonian for the system and the 4’s are the basis functions. The integral runs over all space with d7 being the appropriate volume element. Usually at least some of the integrals involved in this expression have to be evaluated using numerical quadrature. Provided the matrix elements are real, the secular matrix is symmetric (i.e.H.. = H..).‘J J:‘ For vibrational problems suitable basis functions include solutions of the harmonic oscillator and Morse problems dis- cussed in Section 3. The basis sets required to give a complete representation of the vibrational motion are always infinite.One’s aim is therefore to make a judicious choice of functions so that only a few are required to give a good representation of the wavefunction. Convergence is reached when the use of addi- tional functions only leads to an insignificant lowering of the energy. For rotational degrees of freedom the situation is simpler. For a given rotational quantum number, J, it is only necessary to have (2J + 1) functions to represent the rotational wavefunction fully. These functions, known as Wigner rotation rnatrice~,~ can thus be used to expand any rotational problem and their use is universal. For triatomics, variational calculations performed using basis sets give a computer time requirement which is almost always determined by the size of the secular matrix that needs diagona- lizing.A technique which has proved highly successful in molecular vibration-rotation calculations is the application of the Variational Principle in several steps. In such procedures a problem of reduced dimension is solved using a basis set expansion and the solutions of this problem used as the basis for CHEMICAL SOCIETY REVIEWS, 1992 a higher dimensional problem. If the reduced problem is well chosen then only a minority of its solutions are needed to obtain converged results for the full problem. This can lead to massive savings in computer time. Such methods have allowed the solution of many problems which would otherwise be beyond the scope of current computers.7 Molecular Properties The most obvious results obtained from solving the vibration- rotation problem are the (approximate) wavefunction of the system and its associated energy levels. However, quantum 3 mechanics tells us that with the wavefunction we can calculate all other knowable properties of the system. Of particular State 5'I[----interest in this context are the intensities of the transitions between the various states of the system. Most common transitions which involve absorbing or emit- ]!ting light are driven by dipoles. Thus the intensity of a pure rotational transition is proportional to the square of the perma- nent dipole of the molecule. For vibrational transitions the important property is the change in the dipole moment in the vibrational coordinate being excited.In the harmonic model of molecular vibrations this is simply given by the derivative of the dipole at the equilibrium geometry of the molecule. However, .-more accurate calculations require a knowledge of the dipole as 0 e 180 0 8 180 a function of the internal coordinates of the molecule, as well as the wavefunction of the initial and final state. In fact, because the dipole is a vector, 3 surfaces (2 for planar molecules such as (b)triatomics) are required. Transition dipoles not only give the strength of individual State 397 1 State 398 absorption or emission features, they can also be used to give I t fluorescence lifetimes of excited states.(Fluorescence lifetime is the average length of time an excited state will survive before decaying to a lower level by the spontaneous emission of a photon.) Dipole surfaces can also be used to give vibrationally resolved dipole moments. Similarly any other property of the molecule which is geometry dependent, such as bond lengths, rotational constants, or polarizabilities, can be obtained for different states of the system by using the wavefunction to do the appropriate average. The wavefunction of the system can always be labelled by the total angular momentum of the system, J, as this is a constant of motion. However it is usual to label molecules by the number of quanta of vibrational excitation in each vibrational mode of the system.Such labelling is often based on a harmonic model of the system and is always approximate. The process of attaching these labels is called assignment. Assigning levels is important as a wealth of detailed understanding of how systems behave has been developed in terms of these assignments. Furthermore, as experiments never record all transitions for a system, assign- 3 ments are crucial in any comparison between theory and 0experiment. 8 180 0 e 180 Usually low-lying levels can be assigned without difficulty either by studying energy patterns or by using the coefficients in Figure 3 Contour plots of the wavefunction of 8 vibrational states of the basis set expansion. For higher-lying levels making assign- LiCN: (a) low-lying states, (b) highly excited states. The plots are in ments can be very much trickier.The density of vibrational scattering coordinates with the CN bond length, r, frozen at its states increases with energy, making energy differences unreli- equilibrium value. 0 = 180" for linear LiNC and 0" for linear LiCN. able, and individual states may no longer be dominated by a Solid lines enclose regions where the wavefunction is positive and dashed lines regions of negative amplitude. Nodal planes occur where single basis function. the wavefunction has zero amplitude. The outer dotted curves repre- ~ Another technique for making assignments, used extensively sent the limits of the classically accessible potential for each state in by us, is visual inspection of the wavefunction.This is normally quantum mechanics the wavefunction can tunnel into this region. achieved by making contour plots of the wavefunction or cuts (After J. R. Henderson and J. Tennyson, Mol. Phys., 1990,69, 639.) through the wavefunction -see Figures 3 and 4. States that can easily be assigned are characterized by nodes in the wavefunc- tion which give a simple grid pattern; some examples are given in chaotic. How classical chaos manifests itself in quanta1 systems Figures 3a and 4a. For higher-lying states there may be no remains a controversial subject and is beyond the scope of this assignable nodal structure. This may be because inappropriate article. However, one property that is observed in both mecha- coordinates have been used for preparing the contour plots or nics is that even above the transition to classical chaos, some because the state may be inherently unassignable.assignable solutions are found. Figure 4b depicts wavefunctions Unassignable states can usually be associated with energy of the Hi molecular ion whose energies are well above the regions for which classical solutions of the same problem are classical transition to chaos. For three states the wavefunction CALCULATING MOLECULAR SPECTRA-J. TENNYSON AND S. MILLER 5.f3i 'I 't @(I State 3 I State 4 I 5 ! r3! ,-I. , , , I 1 3 5 1 3 5 R R I I I I I I I I State 148 State 149 I 5 r3 1 1 I I I I I I I I State 150 State 151 5 r3 1 1 3 5 1 3 5 R R Figure 4 Contour plots of the wavefunction 8 vibrational states of Hi: (a) low-lying states; (b) highly excited states.The plots are in scatter- ing coordinates with 8 frozen at 90".The contours are as for Figure 3. (After J. R. Henderson and J. Tennyson, Chem. Phys. Lett., 1990, 173, 133.) appears irregular. State number 150 has a clearly defined nodal structure spread along a half horseshoe shape -and is called a horseshoe state. O The energy levels of the system can also be used to give other properties of the system. The most obvious of these is the partition function, the factor which normalizes the Boltzmann distribution of molecules into particular states. The partition function is needed if one wants to synthesize spectra as a function of temperature.8 Sample Results Variational calculations have been performed on many tri- atomic and some tetratomic molecules. We choose a few of these to give a flavour of what can be achieved. 8.1 H:: from Jupiter to Chaos Perhaps the most exciting project that we have been involved in concerns the fundamental molecular ion Hi. This seemingly simple molecule consists of 2 electrons and 3 protons. Its relatively simple electronic structure means that potential energy surfaces for this molecule can be calculated more accu- rately by using first principles quantum mechanics', than by analysing experimental data. This situation is probably unique for a polyatomic. Hi is rapidly formed when H, is ionized by the exothermic reaction H, + H:+Hl+ H (3) As H, is the most abundant molecule in the universe, there are many astronomical situations where Hfis expected, although it remains largely unobserved.Similarly Hi is easily formed in the laboratory using a hydrogen discharge. Hfis an equilateral triangle in its equilibrium geometry. However, because of its light nuclei, Hi undergoes large ampli- tude vibrational motion. We have performed a series of variatio- nal calculations on this molecule in an attempt to aid both laboratory assignment of the complicated infra red spectra of this molecule and its astronomical detection. These calculations took a new significance following obser- vations of Jupiter by Drossart et aZ.13 These workers were studying a known hot region near Jupiter's south pole.In particular they were looking for very weak transitions due to molecular hydrogen. They observed these lines but at the same time saw 28 other, unexpected, transitions which they were unable to explain. It transpired that a spectrum like this had been observed at the Herzberg Institute of Astrophysics in Ottawa but only assigned as a result of these obser~ations.~~ Because it was thought that the experimental spectrum was probably due to Hf our calculations were enlisted. Figure 5 gives a comparison of our calculated spectrum and that observed by Drossart et aZ. Note that the observed line at 4721 cm-' is due to H,. The agreement is striking. Our calculations match the observed lines positions to within about 0.02%. There are a number of surprising aspects of this first extraterrestrial observation of Hi.The actual transitions involve jumps of two vibrational quanta. Conventional wisdom is that such transitions should be very weak, but the floppiness of Hi makes the two-quanta transitions nearly as strong as the funda- mental or one-quantum transition, which has since also been observed in Jupiter.l Another surprising outcome of this observation was that the observed spectrum could only be modelled by Hi with a temperature in the region of 1000 K. This is a high temperature for a planet whose body is at about 200 K. The observation of Hi in Jupiter has stimulated a very active area of astronomical research studying Hi in Jupiter and elsewhere.This will be the subject of another article by us in this journal. While the electronic properties of Hi may be fairly simple, its nuclear dynamics are extremely rich. Attention was focused on this by the observation of a very unusual spectrum by Carr- ington and co-workers. This experiment prepared Hi in a discharge and probed the resulting ions with an infra redlaser. A mass spectrometer was then used to monitor if any protons were produced by photodissociation. Photodissociation is a common technique for investigating molecular systems and the resulting spectra are often quasi- continuous with maybe rather lumpy features. The spectrum obtained by Carrington and co-workers was extraordinary for two reasons. The energy of the laser was only a small fraction of that required to dissociate the ground state of the Hf and the Figure 5 Observed (blue) and simulated (red) emission spectrum from Jupiter's southern polar region.The labels identify the rotational levels of Hi involved: R(J) is a J + 1 tJ transition. (Reproduced by permission from Chem. Br., 1990,26, 1069.) spectrum, taken over a small range of wavelengths, contained over 26 000 narrows lines. This spectrum, occurring as it does right at the dissociation limit of the molecule, presents a formidable challenge to theory. Further analysis of the experiments showed that the protons generally left the molecule with more energy than they got from the laser. This could only be rationalized by placing both the initial and final state of the system above the dissociation limit of the molecule.Such quasibound states, physicists call them shape resonances, are well known although such a profusion of them is unusual. They result from rotational excitation of the molecule which causes the potential to be distorted by a hump or barrier to dissociation. The molecule can get trapped behind this hump, but because of quantum mechanical tunnelling, the state only remains trapped for a finite time. The Uncertainty Principle means that states which are trapped only for a short time have large uncertainty in their energy and appear as broad lines in any spectrum. All the observed Hf lines are narrow suggesting that they are associated with long-lived states.The dissociation spectrum of Hf as observed has little struc- ture. However, if a spectrum is synthesized by broadening the observed lines, then the spectrum collapses to 4regularly spaced features. This intriguing piece of information has been a chal- lenge to theoreticians. Classical analysis of the Hi system suggests that the molecule is almost totally chaotic once the molecule has chance to become linear. Such geometries become accessible at energies about one third of the way to dissociation. In chaotic systems it is usual for some solutions ('trajectories') to behave in a regular or quasiperiodic fashion. Classical mechani- cians argue that the structure in the Hi spectrum is caused by some underlying regular motion with the many individual transitions being a reflection of the chaotic nature of the system.This explanation leaves two questions. What is the underlying regular motion and how does the fact that Hf obeys quantum and not classical mechanics alter this picture? One proposal for the regular motion is that the atoms undergo a 'horseshoe' motion. O In scattering coordinates, see Figure 2c, this motion can be described in terms of the coupled motions of r CHEMICAL SOCIETY REVIEWS. 1993 and R with 8fixed at 90". As R goes towards zero, r becomes large as the other two atoms move apart to allow the central atom through. As R moves away from zero, r decreases again. Plotting this motion as a function of r against R, where R has the range -co++ co,gives a horseshoe shape.Quantum mechanical calculations on these high energy regions are very difficult, not least because the density of states rises rapidly with energy. However modern supercomputers and the adaptation of the variational procedures described above to use methods based on finite elements rather than basis functions have allowed such calculations to be attempted. So, for example, a quantal calculation has recently estimated the position of every bound vibrational state of Hi. Figure 4shows wavefunctions produced by these calculations. The states of Hi are plotted in scattering coordinates with 8 fixed at 90".In these plots R has the range 0-+ co.State 150has a (half) horseshoe-like shape and is regular in the sense that nodes can easily be counted.The other states in Figure 4b, which are typical of many states in the high energy region, have irregular structures. Although there is clear evidence for horseshoe states in the quantum mechanical calculations, their exact role in the Hi dissociation spectrum remains controversial. So far the quantal calculations have not been sophisticated enough to generate actual spectra: in particular no-one has managed to study the effect of rotational excitation on the high-lying states. A decade after the first infra red photodissociation spectrum of Hi was recorded there is still clearly some way to go before it is fully understood. 8.2 Van der Waals Complexes:Ar-N, Van der Waals bonding is the weak attraction which results from charge clouds adjusting to each others instantaneous fluctua- tions.It is the weakest of the chemical bonds and thus van der Waals molecules are only very weakly bound. Van der Waals complexes are thus aggregates of stable, often closed shell, molecules. The most studied triatomic van der Waals systems are the complexes formed between the Noble gases (He, Ne, Ar . . . .) and diatomics such as H,, HF, HC1, and N,. These systems have been prototypical in the development of variational methods. This is because their flat potential energy surfaces mean that the concept of vibrational motion as a (small) displacement from an equilibrium geometry is unhelpful. Inter- CALCULATING MOLECULAR SPECTRA-J. TENNYSON AND S. MILLER nal coordinates, particularly scattering coordinates, have thus been used for some time for van der Waals complexes.As mentioned earlier, the wealth of experimental data on certain van der Waals systems has meant that empirical poten- tial energy surfaces have been derived for a number of complexes by performing cycles of calculation which involve guessing the surface, calculating the transition frequencies predicted by the surface, comparing with experiment, and repeating the pro- cedure until a satisfactory surface is pr~duced.~ Recently an attempt has been to use a similar procedure to generate a dipole surface for the Ar-N, complex.'* For Ar-N, the experimental spectrum was obtained in the infra redby using wavelengths in the region of the N, vibrational fundamental.' As excitation of the vibrational mode of isolated N, is forbidden, this experiment is sensitive to those N,'s which are part of a complex.Furthermore the transitions do not occur exactly at the frequency of the forbidden N, transitions because vibrational modes of the van der Waals complex can also be excited. The spectrum thus has several features, all of which are superimposed on a broad continuous background due to pres- sure broadening. Pressure broadening is the result of using relatively high pressures to produce the van der Waals com- plexes meaning that the molecules in the experimental cell can no longer be considered as isolated. The experiments on Ar-N, were performed at liquid nitrogen temperatures of about 77 K.For a van der Waals complex this is hot and means that for Ar-N, all levels are thermally occupied. The result is that instead of consisting of a series of discrete lines, the experimental spectrum comprises a number of features each containing many transitions. Indeed the calculations found that the strongest individual transitions were ten times weaker than the features in the experiment. It was thus necessary to consider some 15 000 transitions in synthesizing the spectrum. These were obtained by calculating all vibration-rotation states of the system which lie below the dissociation limit of the complex. Figure 6 compares the calculated and experimental spectra for the Ar-N, complex. The agreement is not spectacular but the comparison contains a wealth of information about the poten- tial used for the calculations and the interpretation of the experiment.For instance, the broad shoulder in the calculated spectrum 7 cm-from the N, fundamental frequency (taken at the origin on the figure) is almost certainly a genuine feature which was lost in the process of removing the background due to pressure broadening from the observed spectrum. Conversely, the misalignment of the peaks marked S(0)is due to the potential energy surface used in the calculation. It would appear the potential for the Ar to move about the N, was not flat enough in the low-energy region. It will also be noticed that the observed spectrum has a much greater extent than the calculated one.This Figure 6 Observed (blue) and simulated (red) infra red absorbtion spectrum of the Ar-N, van der Wads complex. Both spectra are for a temperature of 77 K and a density of 1.7 amagat. (After Garcia Allyon et al., Mol. Phys., 1990,71, 1043.) is because the calculations only considered truly bound states of the system, whereas the higher frequency features are due to quasibound states trapped behind rotational humps in the potential. 8.3 Potential Energy Barriers: Isomerization in LiCN Many molecules are found to have more than one stable structure. These different structures are called isomers and in general become more common as the number of atoms in a molecule increases since this also increases the number of candidate structures.Some triatomic systems exist as different isomers. For example hydrogen cyanide is found with two linear structures. Indeed, an unresolved astrophysical problem is why HNC is almost as abundant in interstellar clouds as its much more stable HCN isomer. Quantum chemical calculations have predicted that both linear forms of LiCN are also stable.20 However, in this case the LiNC isocyanide is predicted to be the more stable form. This has been confirmed by microwave experiments., ' Although HCN has been the subject of many variational calculations (eg. ref. 6), LiCN is actually easier to work on. This is because in HCN the H-CN stretching mode is at a similar frequency to the C-N stretch. The heavier Li atom means that the high fre- quency C-N mode is approximately decoupled from the other vibrational modes in the system.Furthermore, the theoretical barrier between the two linear isomers of LiCN is small, only 3500 cm-' (0.4 eV) or less than 5 quanta of the Li-CN stretch. As the LiCN is heavily ionic, the potential can be thought of as being Li + orbiting CN- with only a secondary sensitivity to the actual orientation of the CN-. LiCN has thus been the subject of a series of calculations all of which have frozen the CN motion. Within this model, the 6Yh state of the system is already in the region of the barrier. As recent calculations22 have obtained wavefunctions for the low- est 900 states of the system, the behaviour of the system below the barrier, in the region of the barrier, and well above the barrier can be studied.Plots of low-lying and high-lying states of LiCN are given in Figure 3. The lower states can all be easily assigned by their nodal patterns. However, as more energy is put into the Li-CN bending motion, the states become increasingly irregular. It is the bending motion which samples the isomerization barrier. The potential in this region is of course highly anharmonic and thus the assignments, based as they are on a harmonic picture of molecular vibrations, rapidly break down. Above the barrier to isomerization an increasing proportion of states (more than 90% above state 400) have a highly irregular appearance and cannot be assigned. However, in this region there are series of states with clear nodal patterns.Perhaps the most pronounced of these series is the overtones of the isocya- nide Li-CN stretch. These states go throughout the region studied. Thus state 400, depicted in Figure 3b, has 12 quanta of stretch. An interesting and topical question is how the many irregular states in a system such as LiCN would manifest themselves in observed spectra. Calculations of the conventional absorption spectrum of the molecule in its ground vibrational states show few unusual features. This is because the spectra are dominated by excitations of the Li-NC stretching mode. The ionic nature of the molecule means that this mode has a large dipole which leads to very intense transitions. Similar behaviour has been observed experimentally for HCN.-AT~rak-aiing-pi-cxure'dT-ihe'.bekvioakuT rne ekcfteti $rates of LiCN is obtained by considering their fluorescence life- time~.,~This is illustrated in Figure 7 which has an interesting structure: the assigned ('LiNC regular') states fall into a number of series. One series, for which the lifetime decreases with excitation, all have no stretching excitation and increasing quanta of bending energy. Conversely the other series are for 1, 2, 3, and 4quanta of stretching excitation respectively, with the lifetime getting shorter as the degree of stretching excitation 98 0 3000 0 1 0 0 0 A ., 0 0 LiNC regular a Unclear Chaotic 0 2000. 0 0 a A h 7 0 0 v6 0 0 0 Lu‘ 1000. 0 0 0 0 0 0 0 0 0 0 0- 1.o Figure 7 Fluorescence lifetime of excited vibrational states of LiNC with J = 0.Eiis the energy of each state relative to the LiNC ground states. Assignments as ‘regular’ or ‘chaotic’ were made by analysing contour plots of the wavefunctions such as those given in Figure 3. (Reproduced by permission from Chern. Phys, 1986, 104, 399.) increases. Within each series, increasing the bending excitation leads to slightly longer lifetimes. All this can be understood by remembering that because of its ionic nature the transitions in LiCN are very strong for stretch- ing states. Thus any excited state which can do so decays rapidly by losing a quantum of stretch. Conversely decaying by losing one or more quanta of bending excitation is slow.However, as the molecule is excited there is an increase in bend-stretch mixing. This means that the higher bending states take on a small but increasing stretch character which shortens their lifetime whereas increasing the bending character of the stretch excited states lengthens their lifetime. As the states become more excited the bend-stretch interac- tion becomes stronger. This leads to a host of states that have distorted wavefunctions which defy assignment. It also leads to a situation where all states have rather similar fluorescence life- times as represented by the clump of unassigned (‘chaotic’) states in Figure 7. Analysis of classical calculations on LiCN shows that bending excitation is the main cause of chaotic behaviour in the system.24 Indeed regular (‘quasiperiodic’) stretching states can be found over a wide energy range, possibly all the way to dissociation, provided that the bending excitation is kept low.Similar analysis of the quantal results24 show again that regular stretching states can be found over an extensive energy range, see state 400 of Figure 4b for example. A direct quantitative comparison of classical and quantal results is given by Figure 8. In this figure the energy is appor- tioned to either stretch or bending. The resulting state/trajectory is then assigned as either regular/quasiperiodic or irregular/ chaotic and the appropriate entry made. It is clear that there is a wide measure of agreement between the predictions in the two mechanics although the onset of irregular quantum states occurs at a slightly higher threshold energy.This behaviour has been called ‘quantum sluggishness’. 9 Conclusions In this article we have given a brief review of how one performs vibration-rotation calculations on small molecules. We have tried to explain why these calculations are important and to give the scope of problems that can be addressed with these calcula- tions by the discussion of some sample results. It will be noted that all the problems discussed are for triatomic molecules. In two of these cases the calculations involved finding all the bound states of the system -all the vibrational states of HS and all the vibrational and rotational states of Ar-N,.These and most other variational calculations on triatomics can now be per- formed so accurately that the major source of error is the CHEMICAL SOCIETY REVIEWS, 1992 Figure 8 Comparison of the quasiperiodic (blank) -chaotic classical domain (hatched in blue) with ‘regular’ -‘chaotic’ quantum states (red) for LiNC. Classical results were obtained by starting 50 trajec-tories with the energy indicated in the stretch and bend coordinates. Quanta1 results were obtained by partitioning energy between stretch- ing and bending modes. This cannot be done for most ‘chaotic’ states which therefore do not appear on the figure. (After J. Tennyson and S. C. Farantos, Chem. Phys., 1985,93,237.) potential energy surface used for the calculation.This is still true for the most accurate vibration-rotation calculations available, viz. our own on Hi, which reproduce a range of experimental data with an error of about 1 part in 5000. The advent of variational vibration-rotation calculations has really opened the way for serious and systematic theoretical analysis of the highly-excited states triatomic systems. Such states, as implied by the discussion of the Hi and LiCN problems, are often found at energies where classically one finds chaos. What the exact consequences of this are for quantum mechanical wavefunctions or indeed spectroscopy is still a matter for considerable speculation. At least theoreticians now have the necessary tools in their armoury to tackle such problems.Our discussion has concentrated almost exclusively on tri- atomic molecules. This does not imply that larger molecules are without interest. Quite the contrary, these systems are extremely challenging. A number of variational calculations on tetratomic systems have now been performed. The computational aspects of the tetratomic problem are rather different than for triatomics. Although it has not been considered here, in order to form the matrix elements necessary to construct the secular matrix, it is necessary to integrate over the coordinates of the problem. If any arbitrary potential function is to be used, this integration must be done numerically in 3N -6 dimensions. It turns out that for triatomic systems (N = 3) this integration is much less computationally demanding than the later step in the calcula- tion of diagonalizing the secular matrix.For tetratomic system (N = 4), 6D numerical integration must be performed. As the number of integration points increases roughly as M3N-6,even a modest number of points in each dimension such as M = 10 leads to a thousandfold increase in computer time requirements. It is clear that any serious advance in the area of calculating spectra for larger molecules using variational procedures must somehow first break this integration bottleneck. Acknowledgements. We thank our various collaborators, par- ticularly James Henderson and Brian Sutcliffe, from whose CALCULATING MOLECULAR SPECTRA-J. TENNYSON AND S.MILLER work we have quoted freely. We also gratefully acknowledge funding from the Science and Engineering Research Council, the EEC, the British Council, the Research Corporation Trust, and NATO. 10 References 1 J. Birnie, Spectrosc. World, 199 1, 3(3), 12. 2 R. J. Le Roy and J. S. Carley, Adv. Chem. Phys., 1980,42, 353. 3 J. M. Hutson, Annu. Rev. Phys. Chem., 1990,41, 123. 4 P. Jensen, J. Mol. Spectrosc., 1989, 133,438. 5 E. Kauppi and L. Halonen, J. Phys. Chem., 1990,94, 5779. 6 S. Carter, N. C. Handy, and I. M. Mills, Proc. R. Soc.London, Ser. A, 1990,332,309. 7 B. T. Sutcliffe, in ‘Methods in Computational Chemistry’, Vol. 5, ed. S. Wilson, Plenum, (in press). 8 R. McWeeny, ‘Methods of Molecular Quantum Mechanics’, 2nd edition, Academic, London, 1989, Chapter 2. 9 R. N. Zare, ‘Angular Momentum’, Wiley, 1990. 10 J. M. Gomez Llorente and E. Pollak, J.Chem. Phys., 1988,88,1195; 1989,90, 5406. 11 J. Tennyson, S. Miller, and J. R. Henderson, in ‘Methods in Computational Chemistry’, Vol. 5, ed. S. Wilson, Plenum, (in press). 12 W. Meyer, P. Botschwina, and P. G. Burton, J. Chem. Phys., 1986, 84, 891. 13 P. Drossart, J.-P. Maillard, J. Caldwell, S. J. Kim, J. K. G. Watson, W. A. Majewski, J. Tennyson, S. Miller, S. Atreya, J. Clarke, J. H. Waite Jr., and R. Wagener, Nature (London), 1989,340, 539. 14 W. A. Majewski, P. A. Feldman, J. K. G. Watson, S. Miller, and J. Tennyson, Astrophys. J., 1989,343, L51. 15 S. Miller, R. D. Joseph, and J. Tennyson, Astrophys. J., 1990, 360, L55. 16 A. Carrington, J. Buttenshaw, and R. A. Kennedy, Mol. Phys., 1982, 45,753; A. Carrington and R. A. Kennedy, J. Chem. Phys., 1984,81, 91. 17 J. R. Henderson and J. Tennyson, Chem. Phys. Lett., 1990,173,133. 18 A. Garcia Allyon, J. Santamaria, S. Miller, and J. Tennyson, Mol. Phys., 1990,71, 1043. 19 A. R. W. McKellar, J. Chem. Phys., 1988,88,4190. 20 R. Essers, J. Tennyson, and P. E. S. Wormer, Chem. Phys. Lelt., 1982,89, 223. 21 J. J. van Vaals, W. L. Meerts, and A. Dymanus, Chem. Phys., 1984, 77, 406 1. 22 J. R. Henderson and J. Tennyson, Mol. Phys., 1990,69,639. 23 J. Tennyson, G. Brocks, and S. C. Farantos, Chem. Phys., 1986,104, 399. 24 J. Tennyson and S. C. Farantos, Chem. Phys., 1985,93,237.

ISSN:0306-0012    

DOI:10.1039/CS9922100091

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Transmetallation and its Applications Geoffrey Davies" Department of Chemistry, Northeastern University, Boston, MA 021 15, U.S.A. Mohamed A. El-Sayed and Ahmed El-Toukhyt Department of Chemistry, Alexandria University, Alexandria, Egypt 1 Copper-Oxygen Reactions Copper(~)-O, reactions are crucial in Nature' and essential components of important industrial reactions that include the Wacker synthesis of acetaldehyde from ethylene and the produc- tion of phenylene oxide polymers from phenols, equation la., r + 02 py \ + H20 In this review we describe how a systematic study of aprotic Cul/O, chemistry led to the isolation of huge families of new heteropolymetallic molecules through the accidental discovery of tran~metallation.~ Also described are the use of heteropoly- metallic transmetallation products for the low-temperature syn- thesis of metals, alloys, and mixed metal oxides4 and how the transmetallation phenomenon might be extended to additional elements. Our story begins in 1976.After much effort we had managed to characterize py2CuC1, and a polymeric, soluble form of copper(I1) oxide, [pymCuO],, as the products of oxidation of slurried copper@) chloride by 0, in neat pyridine (p~).~ This Geoflrey Davies was born in Staflordshire, England. He has B.Sc. (1963), Ph.D. (1966), and D.Sc. (1987) degrees from Birm- ingham University. After postdoctoral work with Drs. Kenneth Kustin (Brandeis University), Norman Sutin (Brookhaven Natio- nal Laboratory), and Edward Caldin (University of Kent) he joined Northeastern University in 1971 and is a University Dis- tinguished Professor.His major interests are inorganic synthesis and catalytic mechanisms. Mohamed El-Sayed was born in Alexandria, Egypt. He has B.Sc. (1973) and M.Sc. (1977) degrees from Alexandria University anda Ph.D. degree (1982) from Northeastern. He is an Associate Professor of Chemistry at Alexandria and a Visiting Professor at the United Emirates University. His research concerns the pro- ducts, kinetics, and mechanisms of oxidation and transmetallation reactions. Ahmed El-Toukhy has B.Sc. (1970), M.Sc. (1974), and Ph.D. (1977) degrees from Alexandria University. He is Professor of Chemistry there and a Visiting Professor at Qatar University.He has worked especially on the synthesis and characterization of (N,S) ligand-metal complexes with biological activity and on metal-phosphazene materials. fundamental work was needed before we could investigate the mechanisms of catalytic reactions 1.6 Our next goal was to find ligands which give soluble halo(ami-ne)copper(I) complexes that react aprotically with 0, to give simple, isolatable oxocopper(I1) products {mixtures CuCl(s)/ py,CuCl,/[py,CuO],/py/O, are mechanistically intractable). s77 Hopefully, the oxocopper(rr) products would be characterizable catalysts for reaction 1. Although they are made from common materials, halo(oxo)- copper(r1) complexes are uncomfortable Neverthe-less, we found that simple monodentate ligands N = N,N-diethylnicotinamide (DENC) and ethylnicotinate (ENCA) and bidentate L = N,N,N',N'-tetraalkyldiaminessatisfied our solu- bility requirements and survive coordination to strongly oxidiz- ing oxocopper(I1) centres.'3 Subsequent workgJ O describes aprotic Cul/O, chemistry that encouraged substantial new work in this important research area.' We learned that oxocopper(I1) complexes can deproto- nate substrate~,~~'~ and that their ability to do so apparently depends on how the 0x0 groups are disposed in the overall molecular structure (Figure l).9-1 The 0x0 group disposition must depend on the ligand on copper because [NCuX],O, products from equation 2 do not initiate reaction 1 whereas products py3,,Cu4X402 are The different 0x0 group dispositions in Figure 1 are due to the oxidation mechanism.The slowest step is insertion of an 0, molecule through one of the six faces of a (N,py),Cu,X, cubane and not a redox reaction. Transfer of the third electron from Cd to 0, breaks and 0-0bond. If 0, insertion is completed before this third electron is transferred, the result is weakly basic 0.-O),N,Cu,X4 (Figure 1). On the other hand, if the leading 0 atom of 0, only reaches the centre of the (N,py),Cu,X, cubane when the third electron is transferred, then that leading atom becomes the central p4-0x0 group of ~4-O)py3,,Cu4X40. The other 0 species created when the 0-0 bond is broken does not enter the core of (N,py),Cu,X, and ends up coordinated as a very basic terminal 0x0 group (Figure l).9-12 But our work was far from complete: the primary oxocop- per@) products in Figure 1 disproportionate on attempted isolation as single crystals to (p4-O)(py,N),Cu4X6 molecules that do not initiate reactions 1.9-11 Excellent initiators [LCuX],O from equation 3 are even more reactive and have never been successfully cry~tallized.~-~ 2[LCuX], + 02 +2[LCUX],O (3) Then Fortune smiled. One of us had been studying M(NS), complexes.Here, NS are monoanionic hydrazinecarbodith- ioate Schiff base ligands (Figure 2).I39l4 Some M(NS), com- plexes have antitumour activity that can be tuned by choice of metal M and ligand NS. Their efficacy depends on M(NS),- DNA interactions. DNA is a potential ligand and is basic, so M(NS), complexes must be Lewis acids.This was confirmed by t Present address: Department of Chemistry, University of Qatar, P.O. Box 271 3, Doha, Qatar. N x-C"' \N 'N Forms stable dicarbonate No reaction with C02 ENCA PY DENC Figure I Proposed alternative dioxotetracopper(i1) products from the reactions of (N,py),Cu,X, complexes with 0, in aprotic solvents. o.l their tunable ability to form adducts with Lewis bases like pyridines. Aprotic Lewis acids M(NS), were being studied in Alexandria and Lewis bases like (p,-O)py3,,Cu,X,0 were being made in Boston. Perhaps these acids and bases would combine to form adducts? Adduct stability would vary with M and NS in M(NS), and with the different proposed 0x0 group dispositions in Figure 1.We carefully planned a basicity scale that would differentiate oxocopper(I1) core structures and ran the experiment. 2 Transmetallation The result was completely unexpected. Instead of just adduct formation, we found stoichiometric exchange of M from M(NS), for copper(I1) in (~~,N),CU,X,O,~ and [LCuX],O.' The reactions gave entirely new heteropolymetallic products. We called this new metal exchange phenomenon transmetalla- tion, which is the stoichiometric replacement of the metals in a polymetallic target with different metals from reagents called transmetall~tors.~Targets are Lewis bases and transmetallators are Lewis acids. 5-' Acid-base interactions in precursors organize metal exchange between the reactants. '6,1 Direct transmetallation gives heteropolymetallic products with the same number of metal centres as the target (for example, equations 4-6, 8, 9) but ,fragmentation reduces it (equation 7).Monotransmetallation replaces just one of several metals in a target. Targets can react selectively with transmetallator mix- tures.'* When the target contains more than one element, the replacement of a particular element is said to be specific.' 'Most importantly, transmetallation gives heteropolymetallic pro-ducts that otherwise cannot be obtained. (p4-O)N,Cu4X6 + xM(NS), -+ (p,-O)N,Cu, ,M,& + xCu(NS), (refs. 18, 19) (4) N,Cu,X, + Co(NS), 3 N,Cu,Co(NS),X, + N + Cu(NS)(s) (ref. 12) (5) (p,-O)N,Ni,X, + 2 Co(NS), 3 (p,-O)N,NiCo,X, + 2 Ni(NS), (ref.18) (6) (p,-O)N,Cu,X, + Hg(NS)2 3 (P~-O)N~CU~X,+ HgX, .NCu(NS), (ref. 18) (7) CHEMICAL SOCIETY REVIEWS, 1992 (p3-O)N3Ni,CoCl, + Zn(NS), -+(p,-O)N,NiCoZnCl,+ Ni(NS), (ref. 18) (9) Reactions 4 proceed in four distinct, stoichiometric steps.' For this reason they are sources of new heteropolymetallic molecules containing up to four different rnetals.l8 The first product of equation 5 is a mixed valence molecule that formally contains two copper(i), one copper(rI), and one cobalt(I1) centres. These and other partially transmetallated complexes have useful photoemissive properties.20 Reactions 6 and 9 demonstrate that transmetallation is not restricted to copper targets. 2~2'Scissor' transmetallators M(NS), (M = Cd, Hg, or Sn) snip one corner from the targets in reactions 7, which are the only known routes to (p,-O)N,Cu,X, molecules.The latter are transmetallated to a host of new heterometallic trimers. Equation 8 demonstrates that we can transmetallate cata-lytic2,6y1 targets to alter catalysis mechanisms. Reaction 9O shows that direct transmetallation is specific (nickel is replaced in preference to cobalt): the first product shown contains three different metals. * Subsequent work shows that the most fundamental require- ments for efficient transmetallation are labile targets and trans- metallators and a strong driving force for metal exchange. ''A list of other desirable target and transmetallator characteristics has been offered.16 3 Major Conclusions The metal nuclearity of neutral halo(amine)copper(r) complexes in aprotic solvents depends on the denticity of the amine ligand and the experimental condition^.^.^ O Bidentate ligands favour dimeric copper(1) complexes [LCuX], 7,1 whereas monodentate pyridines allow monomers, dimers, and tetramers to exist at different copper(r) concentrations.' O DENC spontaneously forms invariant, soluble complexes [NCuX], (X = C1 or Br) and has been the ligand in most of our work on tetranuclear target tran~metallation.~Alternative core structures in Figure 1 can be differentiated by a number of means, including their patterns and products of transmetallation with M(NS), reagents.Transmetallation can be used to synthesize large families of entirely new heteropolymetallic molecules under mild con-ditions.17.1sJ' The reactions are quantitative and the products are easily separated by gel permeation chromatography. Distinct transmetallation patterns depend on the target and transmetallator structures' '.14 and relative stabilities.22 Trans- metallation is selective' ,because of different transmetallator stabilities22 and transmetallation rate Step-wise, site- specific transmetallation has been demonstrated.' 2,13 Nucleo-philic centres in the reactants organize and facilitate specific transmetallation.6+1 The transmetallation phenomenon is not restricted to polynuclear copper and is very useful for heteropolymetallic product synthesis., 4 The Inherent Interest of Heteropolymetallic Molecules Heteropolymetallic chemistry is of obvious interest for several reason^.^ Molecular families containing different metals have inherent specific character and could catalyse specific, 'tunable' thermal or photochemical coordination and redox reactions.Communication between metals is affected by changing the metallic constitution of a molecule. * The availability of families with a common core structure but containing different and easily variable metals allows us to find out whether one metal knows that another is at some other molecular site. Lastly, a family of molecules, say (p,-O)N,Cu, -,Ni,X,, (from equation 4; x = W),can be reduced under very mild conditions to the .,alloy family Cu,_ ,Ni, These alloys might function as hetero- [LCuX],O + M(NS), --t [LCu(X,X)ML]O + Cu(NS), geneous catalysts that differ from materials with the same (ref.16) (8) atomic proportions prepared by other methods.24 TRANSMETALLATION AND ITS APPLICATIONS-G. DAVIES, M. A. EL-SAYED, AND A. EL-TOUKHY S S n Figure 2 Molecular structures of some M(NS), trans metal la tor^.'^ NS is S-methyl isopropylidenehydrazinecarbodithioatein molecules A, B, and D and S-methyl benzylidenehydrazinecarbodithioatein mole- cules C and E. 5 Further Development of Transmetallation Chemistry The limitations of current transmetallation and its application to alloy and mixed metal oxide synthesis are that (a) not all transmetallation reactions are direct:2 this limits the range of uniform compositions that can be obtained from trans-metallation product starting materials; (b) (p4-O)N4 (M ,M2,M3,M4)4X6 molecules cannot contain more than two zinc centres; (c) iron<opper transmetallation reactions are slow because transmetallators Fe(NS), .3 and co-products Cu(NS), have similar thermodynamic stability;21 (d) the approach cannot easily be extended to catalytic heavy metals like M = Pd or Pt because transmetallators M(NS), are substi- tution-inert and we have found no easily made, labile polymetal-lic targets containing such heavy metals.However, early indica- tions are that dithiophosphatometal complexes M(S,P(OR),), (R = alkyl or aryl) are good alternatives to M(NS), as trans- metal la tor^.^ Among their advantages are easy synthesis and availability for most of the metallic elements.There are many obvious uses of heteropolymetallic molecules, alloys, and mixed metals and their oxides that contain catalytic heavymetals.2 3.24.2 6 Acknowfrdgernents. We gratefully acknowledge the National Science Foundation (Grants CHE-8717556 and INT-8918985) and the Donors of the Petroleum Research Fund, administered by the American Chemical Society (Grants 20022-AC3 and 24132-AC3-C) for financial support of our work. The contribu- tions of our many co-workers are acknowledged by citation in the references. S S 6 References 1 Z. Tyeklar and K. D. Karlin, Acc. Chem. Res., 1989,22, 241. 2 H. L. Finkbeiner, A. S. Hay, and D. M. White, in 'Polymerization Processes', ed. C. E. Schildnecht and I.Skeist, Wiley-Intersclence, New York, 1977, p. 537 and references therein. 3 A. El-Toukhy, G.-Z.Cai, G. Davies, T. R. Gilbert, K. D. Onan, and M. Veidis, J. Am. Chem. Soc., 1984, 106, 4596. 4 G. Davies, B. C. Giessen, and H.-L. Shao, Mat. Lett., 1990, 9, 23 1. 5 I. Bodek and G. Davies, Inorg. Chem., 1978, 17, 1814. 6 M. A. El-Sayed, G. Davies, and T. S.Kasem, Inorg. Chem., 1990,29, 4730 and references therein. 7 G. Davies, M. F. El-Shazly, D. R. Kozlowski, C. E. Kramer, M. W. Rupich, and R. W. Slaven, Adv. Chem. Ser., 1979, 173, 178. 8 M. R. Churchill, G. Davies, M. A. El-Sayed, and J. P. Hutchinson, Inorg. Chem., 1982, 21, 1002. 9 G. Daviesand M. A. El-Sayed, Comments Inorg. Chem., 1985,4,151. 10 G. Davies and M. A. El-Sayed, Inorg.Chem., 1983,22, 1257. 11 G. Davies, M. A. El-Sayed, A. El-Toukhy, M. Henary, and C. A. Martin, Inorg. Chem., 1986,25,4479. 12 G. Davies, M. A. El-Sayed, A. El-Toukhy, M. Henary, and T. R. Gilbert, Inorg. Chem., 1986,25,2373and references therein. 13 M. F. Iskander, M. M. Mishrikey, L. El-Sayed, and A. El-Toukhy, Inorg. Chim. Acta, 1979, 41, 815. 14 K. D. Onan, G. Davies, M. A. El-Sayed, and A. El-Toukhy, Inorg. Chim. Acta, 1986, 119, 121 and references therein. 15 M. F. Iskander, L. El-Sayed, L. Labib, and A. El-Toukhy, Inorg. Chim. Acta, 1984, 86, 197. 16 G. Davies, N. El-Kady, M. A. El-Sayed, A. El-Toukhy, and M. R. Schure, Inorg. Chim. Acta, 1988, 149,45 and references therein. 17 S. Al-Shehri, G. Davies, M. A. El-Sayed, and A. El-Toukhy, Inorg. Chem., 1990,29, 1198, 1206. 18 A. Abu-Raqabah, G. Davies, M. A. El-Sayed, and A. El-Toukhy, Inorg. Chem., 1989,28, 1156. 19 G. Davies, M. A. El-Sayed, and A. El-Toukhy, Inorg. Chrm., 1986, 25, 2269. 20 M. Henary and J. I. Zink, J. Am. Chem. Soc., 1989, 111, 7409. 21 K. G. Caulton, G. Davies, and E. M. Holt, Polyhedron Report No. 33, Polyhedron, 1990,9, 2319. 22 G. Davies, M. A. El-Sayed, A. El-Toukhy, M. Henary, T. S. Kasem, and C. A. Martin, Inorg. Chem., 1986,25, 3904. 23 P. Braunstein, New J. Clzem., 1988, 12, 307. CHEMICAL SOCIETY REVIEWS, 1992 24 W. Romanowsky, 'Highly Dispersed Metals', Wiley, Chichester, Sayed, A. El-Toukhy, and X. Liu, unpublished results. 1987. 26 G. H. Via, K. F. Drake, Jr., G. Meitzner, F. H. Lytle, and J. H. 25 H. Abo-El-Dahabe, A. R. Barron, G. Davies, N. El-Kady, M. A. El- Sinfelt, Catal. Lett., 1990, 5, 25.

ISSN:0306-0012    

DOI:10.1039/CS9922100101

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Bridgehead Radicals John C. Walton Department of Chemistry, University of St, Andrews, St, Andrews, Fife KY16 9ST Scotland I Significance of Bridgehead Radicals Bridgehead species necessarily have virtually rigid molecular frameworks, with bond angles and dihedral angles which are fixed and known with reasonable certainty. Molecules with such well defined structures are invaluable probes for testing and designing theories about orbital interactions, bond interactions, a n aconformational effects on structure and reactivity, substituent effects, and a host of other geometry-dependent chemical propo- sitions. Many bridgehead species, particularly highly strained ones, have very unusual structures with abnormal bond lengths and angles. They have provided a certain stimulus, notably to those synthetic chemists with well developed pioneering instincts, to invent new and original processes whereby they can be made.Special zest is added to this type of work if it has been predicted by a particular hypothesis that the species in question will be unstable or forbidden. In this way organic structures have been pushed to the limits of stability and reactivity and the amazing flexibility of the carbon atom in adapting itself to virtually any geometry, no matter how distorted, has been highlighted.A remarkable illustration of this type of progress is provided by the chemistry of organic species with structures based on the regular polyhedra. There are only five regular polyhedra, known collectively as the Platonic solids.Three of these, the tetrahedron, the cube and the dodecahedron can serve as models for the structure of hydrocarbons C,H, (Scheme 1) in which every methine unit is a bridgehead. Removal of one hydrogen atom from these hydrocarbons yields the correspond- ing ‘Platonic Radicals’,* C,H,-which are neutral bridgehead species of exceptional symmetry. The Platonic hydrocarbons have stimulated a phenomenal amount of creative research3 and attention has turned recently to the generation and observation of reactive intermediates derived from their basic structures. The parent tetrahedrane is unstable under normal laboratory con- ditions so it is unlikely that the tetrahedryl radical will be detected, although sterically shielded derivatives are not imposs- ible.The cubyl radical is well characterized (vide infra) but rather little work has been done on the dodecahedryl radical and attempts to observe it spectroscopically have not ~ucceeded.~ Bridgehead radicals can be distinguished by the number of /3-hydrogens they contain. Three limiting types can be specified, firstly, the /3-methylene type (1) in which the radical bridgehead is flanked exclusively by methylene groups; examples include the 1-adamantyl and bicyclo[2.2.2]oct- 1 -yl radicals. Secondly, the /3-methine type (2) in which the radical centre is flanked by bridgehead methine groups; examples include the cubyl, dode- cahedryl, and prismyl radicals. Thirdly, the /3-quaternary type (3) in which the radical centre is flanked by quaternary carbon atoms; the triptycyl radical is an example of this type.Of course, a host of intermediate species exist with mixtures of adjacent methylene, methine, and quaternary groups, and a few bridge- head radicals with /3-heteroatoms have also been studied. Bridgehead radicals differ from other tertiary radicals in a number of important ways. It is known from spectroscopic and J. C. Walton received his BSc. and DSc. degrees from Shefield University and his Ph.D. degree from St. Andrews University where he is now Reader in Organic Chemistry. His research interests include synthetic and mechanistic studies of free radicals, oxidations with dioxygen, and organic superconductors. 1-H‘ 1-H‘ Scheme 1 theoretical work that the t-butyl radical is non-planar in its lowest energy configuration.However, the deviation from pla- narity is small and the barrier to inversion is very low so that for most chemical purposes t-butyl radicals behave as if they were planar. Bridgehead radicals are permanently pyramidal and far from planar, inversion is prevented, so their structures are ‘unnatural’, and they are expected to be destabilized, relative to corresponding acyclic tertiary radicals. The p-carbons of bridge- head radicals are ‘tied back’ by the cage structure so that the radical centre is sterically uncongested. Obviously, this factor will vary considerably, depending on the details of the structure, but in general bridgehead radicals will be more reactive as a consequence of this.The pyramidal configuration of the bridge- head radical centre ensures that the SOMO will be a 0-orbital with high s-character. The orientation ofthe SOMOwith respect to other bonds in the radical differs therefore from that of normal, nearly planar alkyl radicals, and this will particularly influence unimolecular reactions such as decomposition and rearrangement. The marked s-character is expected to modify their redox properties relative to acyclic tertiary radicals, making them more nucleophilic in behaviour. The t-butyl radical is ca. 37 kJ mol- more stable than the methyl radicalS and this is usually attributed to hyperconjugative and/or induc- tive effects, (4). In a bridgehead radical the orientation of the SOMO with respect to the orbitals of the /3-C-H bonds is usually less favourable for overlap and the rigid structure prevents rotation to improve this.In addition, hyperconjugative structures will contain very strained bridgehead alkene units. For example, in /3-methylene bridgehead radicals (1) hypercon-jugation leads to an alkene with a single bridgehead. In the Is-methine type (2) a double bridgehead structure (5) would be formed; /3-quaternary radicals (3) are inherently incapable of 105 H' (4) (5) hyperconjugation. Thus hyperconjugative structures will not make a significantly stabilizing contribution to the ground electronic state of any type of bridgehead radical. 2 Generation of Bridgehead Radicals The factors mentioned above seemed to indicate that bridgehead radicals would be a lot more difficult to generate than t-butyl radicals.Much of the early work concentrated on comparing the rates of formation of series of bridgehead radicals, derived from various precursors, with the rate of formation of the t-butyl radical. Many bridgehead carbo-cations were known to form with extreme difficulty and therefore comparison of their reacti- vity with that of the corresponding radicals gave additional motivation to this work. There have been three main approaches to the generation of bridgehead radicals. First, particular pre- cursors, designed to yield one specific radical, were carefully synthesized and the rates of radical formation were measured either absolutely, or relative to some standard such as t-butyl radicals.Much work was done with bridgehead perester thermo- lysis, bridgehead azoalkane thermolysis, and tin hydride reduc- tion of bridgehead halides. Appropriate precursors for ketone photolysis, decarbonylation of acyl radicals, and fragmentation of alkoxyl radicals were also examined. This work has been reviewed and evaluated by several authors.6 Most of these methods indicated that the rates of formation were in the following order: But > 1-Ad (6)> Bicyclo[2.2.2]oct-l-y1(7)> Bicyclo[2.2.l]hept-l-y1(8)> Cubyl (9) > 9-Triptycyl (10) but the relative rates varied greatly depending on the mode of generation, presumably because the extent of development of radical character in the transition state differed from one method to another.Thus, a tendency for more strained radicals to be formed with greater difficulty was perceived, although there were exceptions, such as the cubyl radical. The method which showed the greatest change across the series was usually azoalkane thermolysis, but even this technique showed a range of values far smaller, by many orders of magnitude, than had been observed for the generation of carbo-cations in solvolysis experiments. Thus, although bridgehead radicals are more difficult to generate than t-butyl radicals, and are therefore probably thermodynamically less stable, they have a greater toleration for the pyramidal geometry than analogous carbo- cations.This trend is strongly influenced by the amount of strain in the structure. Secondly, a few bridgehead radicals have been formed in ring closure reactions of alkenylbutyl and alkenylpropyl radicals. (7) CHEMICAL SOCIETY REVIEWS, 1992 For alkenyl radicals stabilized by electron-withdrawing substi- tuents at the radical centre, em-cyclization is reversible, which leads to thermodynamic control of the process, and conse- quently the products of endo-cyclization are favoured. Julia and Surzur' studied the ring closure of several cyclohexenyl- and cyclopentenyl-alkyl radicals of this type, e.g. (1 l), and showed that the main product was formed via the bridgehead radical, (12). Beckwith and co-workers showed that, as expected, cycli- zation of the 4-cyclopentenylbutyl radical (1 3) occurred most rapidly in the exo mode to give the spiro intermediate (14), but that formation of the bridgehead radical (1 5) was only about a factor of five slower.In the case of the 3-(2-methylenecyclohex- y1)propyl radical (16), where the pattern of substitution disfa- vours exo-cyclization, the bridgehead radical (1 7) was actually the major cyclized intermediate.8 In a particularly interesting study Yurchenko et ~1.~showed that free radical addition of CC1, or CBr, to diene (1 8) gave adamantane derivatives, formed via bridgehead radical (I 9), and nor-adamantane derivatives, derived from (20), in the ratio (19):(20) = 1:3. In all these examples the bridgehead species formed quite easily, consider- ing that endo cyclizations were involved, and any destabilization of the bridgehead radical seems to have played no part.The reasons for this are that none of the bridgehead radicals studied so far by this method contained much ring strain, but most importantly, that the transition state of intramolecular addition reactions is 'early' and unsymmetrical. That is, the development of bridgehead character in the product radical is not far advanced in the transition state so that its destabilization plays little part in controlling the cyclization.8 Obviously, this ring closure method has substantial untapped potential for the production of bridgehead radicals. Thirdly, a great variety of bridgehead radicals have been produced by hydrogen abstraction from precursors containing bridgehead hydrogens.Species such as halogen atoms, t-butoxyl radicals, aminyl radicals, aminium cation radicals, and poly- haloalkyl radicals have been employed. Usually bridgehead hydrogen abstraction takes place in competition with hydrogen abstraction from other sites in the molecule so that an immediate comparison of bridgehead reactivity with bridge reactivity is possible. In molecules like adamantane, bicyclo[2.2.2]octane, or bicyclo[3.2.2]nonane, with little or no ring strain, photochlori- nation1* and photobromination take place at about the same rate, on a per hydrogen basis, as in cyclohexane. The selectivity for bridgehead hydrogen, relative to secondary bridge hydro- gen, is also similar to that of tertiary hydrogen relative to secondary hydrogen in open chain hydrocarbons. For molecules with more strain, hydrogen abstraction from the bridgehead becomes much more difficult.Bicyclo[2.2. llheptane (norbor- nane) is by far the most studied. More than 95% of the reaction with chlorine atoms occurs at the C, bridges and only about 0.3% at the bridgehead. Larger radicals show a greater prefer- ence for abstraction from the bridgehead,l e.g. the proportion bridgehead products from norbornane increases as shown in Table 1. The factor controlling this selectivity is probably steric. The bridgehead hydrogens are more exposed and therefore larger radicals can approach them more easily than the C,-bridge hydrogens. Additional evidence in support of this 40% BRIDGEHEAD RADICALS-J.C. WALTON Table 1 Proportion of bridgehead hydrogen abstraction from norbornane by various radicals' OP1Oo+a Bridgehead Radical product (YO) C1' 0.3 Bu'O' 1.3 Et,NH+' 2.0 Pr', NH +' 4.0 (Me,Si),N' 4.0m+cb conclusion is displayed in Table 2. The extent of steric shielding of the hydrogens in the bridges increases as the size of the bicycloalkane decreases; at the same time, the bridgehead hydrogens become increasingly exposed. In line with this, the proportion of bridgehead hydrogen abstraction increases down the series and increases with the size of the attacking radical. The bridgehead radicals become more destabilized as the ring strain increases and this probably leads to a decrease in the rate of bridgehead H-abstraction, but this factor is evidently not so important as the steric effect.Bicyclo[ 1.1. llpentane is a particu- larly interesting example. The hydrogens from adjacent bridges lie extremely close to one another so that an attacking radical can only approach them with difficulty in 'co-linear' fashion. The majority of abstraction occurs therefore at the exposed bridgehead hydrogens. Competitive experiments showed that the bridgehead hydrogens themselves are deactivated, as would be expected for such a strained radical. 3 Spectroscopic Studies of Bridgehead Radicals A few bridgehead radicals have been studied by EPR spectros- copy and the data for most of these are displayed in Table 3.The EPR spectra showed that all these radicals had lifetimes in solution of the same order of magnitude as other transient alkyl radicals. Bicycle[ 1.1. llpent-l-yl (22), cubyl (9), and probably the others, decayed by bimolecular processes -almost certainly Table 3 EPR Parameters of bridgehead radicals Radical 1-Adamantyl (6)' Bicyclo[2.2.2]oct-1-yl (7) Quinuclidiene+' (5 1)29 Quinuclidin-4-yl (52)29 Bicyclo[2.2. llhept- l-yl (8) Bicyclo[2. I .I]hex-l-yl (21) Bicyclo[l.l.l]pent-l-yl (22)13 Cubyl (9)33 * lG=O.lmT g-Factor ---2.0029 2.0026 2.0026 2.0028 2.0028 Table 2 Proportion of bridgehead hydrogen abstraction ("/o) from bicyclo[n.m. llalkanes' O-I Bicycloalkane C1' Bu*O' (Me, Si), N' 0.3 1.3 4.0 <5 -0 29 62A exclusively combination rather than disproportionation.It is remarkable that radicals with as much strain as (22) and (9) could be directly observed, and this illustrates the difference in reactivity between bridgehead radicals and cations: the bridge- head cubyl and bicyclo[ 1.1. llpent- 1-yl carbo-cations cannot be observed by low-temperature NMR in super-ionizing media because they react or rearrange too rapidly. The g-factors are the same as those of other hydrocarbon radicals, but the hyperfine splittings (hfs) show noteworthy differences. In t-butyl radicals the /3-hydrogen hfs is 22.9 G and this large value is attributed to hyperconjugation, i.e. the SOMO and the C,-H bonds overlap effectively. All the bridgehead radicals have a(H& values much R n-radical Hfs/G* 4Hg) 6.6( 6H) 6.6(6H) 9.4( 6H) 7.0(6H) 9.8( H ,, 6exo) 2.4(2H7) 0.5(H ,6endo) 5.0(H5,6'"") 2.O(H ,6endo) 1.2(6H) 1 2.4( 3 H) o-radical a(other) 4.7(3Hy), 3.1(3H,5), 0.8(3HJ 0.9(6H,), 2.7( 1Hbr) 2.4(6Hy), 14.3(Hb,), 25.1(N) 1.8(N) 1 .23(H3,SeXo), 2.5(Hbr) 0.36(H,,sendo) 69.6(Hb,)8.2(3H), 6.3(Hbr) lower in magnitude.Two factors contribute to this. First, for the bridgehead radicals of Table 3, the dihedral angle 4 is non-zero (except for cubyl) and is fixed by the cage structure. In addition, the SOMO points away from the orbitals of the C,-H bond, i.e. the angle 0 between the axis of the SOMO and the orbitals of the C,-H bond, is greater than in planar n-radicals.Thus orbital overlap is diminished in comparison with t-butyl radicals. Secondly, hyperconjugative structures are of high energy, and will not contribute to the ground state, because they contain bridgehead alkene units. Hence, even in the cubyl radical (9), where the dihedral angle 4 is zero, and thus optimum for overlap, a(H& is still small. Most of the bridgehead radicals show large long-range hfs. Of particular interest are the splittings from bridgehead hydrogens elsewhere in the structures. In bicyclo[2.2.2]oct- I -yl(7), bicyclo- [l .l.llpent-l-yl (22), and cubyl (9) radicals the SOMO and the orbitals of the C-Hb, bond are exactly in line and at a angle of 180". In bicyclo[2.1. Ilhex- I -yl (21) and bicyclo[2.2.llhept- I -yl radicals (8) deviations from this ideal co-linearity are fairly small. Figure 1 shows a plot of the hfs from the bridgehead 807 1 6o 11 I 0 -20 1.5 2.0 2.5 3.0 r (C*-C~JA Figure 1 hydrogens, a(Hb,), against the distance in space between the radical centre and the carbon atom to which the bridgehead hydrogen is bonded, v[C'-Cb,], as calculated by the semi- empirical AM1 method. Radical (22) shows an enormous hfs from the 7-bridgehead hydrogen which is probably the result of reinforcement between through space (TS) and through bond (TB) effects. There is a steep decrease in a(Hg) as r[c'-cb,] increases up to ca. 2.1 A, followed by a gentle increase. It seems probable that the sharp decrease results from a rapid fall off in the TS effect with increasing distance; this effect being negligible much above 2.1 A.Other evidencel4>l shows that TS effects are not important beyond this distance in radicals. It is known, however, that TS effects between pairs of double bonds, and in cations, are transmitted much further than this. The reason for the difference observed for radicals remains puzzling. The TB effect should also die away as the number of intervening bonds increases, but this will be strongly modified by the dihedral angles between the bonds. It appears that for the cubyl radical (9) the orbital arrangement permits significant spin density to reach the bridgehead &hydrogen so that the graph shows an apparent increase; more data are needed to disclose the true trends.The hfs from bridgehead hydrogens which are not co- linear with the SOMO, e.g. the y-hydrogens in the adamantyl radical, are not aligned with those in Figure 1. Comparison of the experimental H, hfs with values calculated by the INDO method suggested that the radical centres in adamantyl (6) and bicyclo[2.2.2]oct-l -yl radicals (7) were pyr- amidal but slightly flattened out of the true tetrahedral geo- metry. The best experimental evidence about the geometry at bridgehead radical centres would come from the EPR I3C hfs. However, to date, U('~C~,*)has only been obtained for the adamantyl radical.' 'The observed value, of a( "CC,,.) = 136.7 CHEMICAL SOCIETY REVIEWS. 1992 G, showed that only slight relaxation towards planarity had occurred.The first band in the photoelectron spectrum of the adamantyl radical (6) was observed on flash vacuum photolysis of 1-adamantyl nitrite. The adiabatic ionization energy, IE,, of 6.21 eV was about as expected for a large tertiary radical. The sizeable difference between this and the vertical ionization energy, 6.36 eV, together with the band shape, indicated that a substantial geometrical reorganization took place upon ioniza- tion, i.e. that the I-adamantyl carbo-cation is much more flattened than the radical. 4 Homolytic Reactions at Bridgeheads Hydrogen and halogen abstraction from bridgehead sites have been referred to above. For highly strained bicycloalkanes containing small rings, radicals such as CCl,', (Me,Si),N', and ButO', do not transfer hydrogen from the bridgehead, but attack at the bridge(s) of the larger ring.19 In the bicyclo[n.1.O]alkanes (23) attack normally occurs in the larger ring, adjacent to the bridgehead, to give a bicyclo[n. 1.O]alk-2-yl radical (24) which rearranges rapidly by p-scission to either the cycloalk-3-enyl radical (25) or, for n > 3, to the cycloalkenylmethyl radical (26). 'QI--One consequence of this is that the smallest possible bridgehead radicals, (27) n = 1,2 etc., in which the bridgehead forms part of a 3-membered ring, have never been generated, or studied experimentally. Interestingly, with bicyclo[2.2.0]hexane (28), hydrogen abstraction does take place at the bridgehead as well as at the methylene adjacent to the bridgehead,20 and the bicyclo- [2.2.0]hex-I -yl radical (29) was observed by EPR spectroscopy.This surprising selectivity does not denote any increased reacti- vity of the bridgehead hydrogens, but is probably a consequence of steric shielding of the bridge hydrogens. For the larger members of the bicyclo[n.2.0]alkane series, e.g. (3l), abstraction did not take place at the bridgehead. For bicycloalkanes where the bridgehead forms part of a 3- membered ring, a second reaction can occur. This is homolytic substitution (SH2) (also known as displacement). Strangely, with only a few exceptions, the attacking radical must be a halogen atom for this reaction to supplant hydrogen abstrac- tion.Bromine atoms attack both bicyclo[ 1. I .O]butane and bicyc- 10[2.1 .O]pentane (32) exclusively at the bridgehead position, with cleavage of the inter-ring bond to give a 3-substituted cycloalkyl 03 BRIDGEHEAD RADICALS-J. C. WALTON Br*Br U radical such as (33). For the higher homologues, attack occurred at the bridgehead and at the C, bridge, e.g. for (23, n = 3) reactions (i) and (ii) were in competition. The ratio of outer- to inter-ring bond scission i.e. [(i) + (iia)]/(iib) rose from zero for (23) n = 1 and n = 2 to 0.6 and 6 for n = 3 and n = 4, respecti- vely. The ring strain released by inter-ring bond cleavage decreases sharply with increase in ring size and this accounts for the changeover. Chlorine atoms, and sometimes iodine atoms, behave similarly. Br' + (23)n=3 Recently, the scope of this type of process has increased substantially with the discovery that halogen atoms also take part in SH2 reactions at bridgeheads in polycycloalkanes con- taining condensed cyclobutane rings.Photobromination of bicyclo[2.2.0]hexane (28) leads to the formation of trans-and cis-dibromocyclohexane (35) via the SH2process. The homolytic substitution reactions which [ 1.1. llpropellane (36) takes part inz1 are even more remarkable. A variety of radicals such as MeCO', CCl,', PhS', PhSe', ButO', and halogen atoms cleave the unique inter-ring bond to give substituted bicyclo[ 1.1. llhex- anes (38). This is, of course, a new way of generating bridgehead radicals (37).It is evident that the bridgehead bicyclo[ 1.1. llhexyl radical (37) also substitutes at the quaternary carbon atom of (36) because dimers (39) and even oligomers (40) have been isolated in some instances. Bicyclo[ 1.1 .O]butane reacts in a similar way with several free radicals, but only halogen atoms cleave the inter-ring bond of bicyclo[2.1 .O]pentane. 5 Reactions of Bridgehead Radicals Bridgehead radicals will add to unsaturated molecules, abstract hydrogen or halogen, and take part in combination reactions in solution in the normal way via chain processes. In one of the most studied systems, bridgehead radicals were generated from the corresponding perester, and allowed to abstract halogen from a mixture of CC1, and CC1,Br. From the ratio of the bridgehead bromide to chloride produced in this way the relative rate constants for bromine and chlorine abstraction were deter- mined for a series of bridgehead radicals.22 The difference in activation enthalpies AH& -AH& was found to increase from homocubyl (and cubyl) to norborn- 1 -yl to bicyclo[2.2.2]oct- 1-yl to 1-adamantyl.This order agrees with expectation because the radical centres become progressively less exposed (i.e. front strain increases) along this series and ring strain decreases. The relative rates were also influenced by entropic factors but, in the appropriate temperature range, homocubyl (and cubyl) radicals were the least selective, and adamantyl radicals were the most selective of this series.The main type of rearrangement open to bridgehead radicals is ,!I-scission. For free radicals in general only species containing 3- or 4-membered rings readily undergo p-scission under normal solution phase conditions. Thus, facile rearrangements are not expected for adamantyl, bicyclo[2.2.2]oct- 1-yl, norborn- 1-yl etc. radicals, except in high temperature or high energy situations. It seems to be characteristic of bridgehead radicals that even highly strained ones rearrange with great reluctance. The cubyl radical, which contains ca. 14 kcal mol-' of strain per C-C bond (cf. cyclobutyl which contains ca. 6 kcal mol- of strain per C-C bond) takes part in reactions at 100°C and above without rearrangement. This is easily explained because 8-scission of (9) would produce the high energy bridgehead alkene (41).However, even bridgehead radicals with potentially strongly exothermic ring opening processes, such as bicyclo- [l.l.l]pent-l-y1(22), bicyclo[2.1 .l]hex-1-yl, or bicyclo[2.2.0]hex- l-y1(28), require forcing conditions for ,!I-scission to occur. This is exemplified by (22), for which /3-scission is of the well known cyclobutylmethyl to pent-1-enyl type, and should give the 3- methylenecyclobutyl radical (42). In fact rearranged products were not observed even at 150 "C; the activation energy for 8-scission being > 26 kcal mol- .l 3~23This can be accounted for in stereoelectronic terms. The SOMO and the orbitals of the bond due to break, CB-Cy, are very poorly aligned for overlap; in addition a great deal of structural reorganization must take place during rearrangement.Kinetically, therefore, rearrange- ment is disfavoured. CHEMICAL SOCIETY REVIEWS. 1992 Because bridgehead radicals maintain their structural inte- grity so tenaciously, homolytic methods have frequently been employed for manipulation of functional groups at bridgeheads, and for other synthetic purposes. Several syntheses of the cubane skeleton, including the original method of Eaton and produce cubane- 1,4-dicarboxylic acids which may be converted into the hydrocarbon via thermal decomposition of the perester. In recent years the Barton decarboxylation via the N-hydroxypyridine-2-thioneesters has replaced perester ther- molysis as the method of choice.This procedure was used to make a variety of highly strained bridgehead derivative^,^^,^ including bromo- and iodo-cubanes, -bicycle[1. I. llpentanes, and -bicyclo-[2.1. Ilhexanes. Cubane was directly converted into mono- and poly-iodides by photochemical reaction with t-butyl hypoiodite.26 The photochemical reaction of bicyclo[ 1. I. llpen-tane with oxalyl chloride gave the corresponding bridgehead acyl ~hloride.~' 6 Bridgehead Radicals Containing Heteroatoms According to an interesting stereoelectronic hypothesis, the SOMO on C, of a radical should interact (by conjugative electron delocalization) with the p-type lone pair of an adjacent oxygen.28 Consequently, any weakening of the C-H bond adjacent to oxygen in the parent ether, and concomitant acceler- ation of hydrogen abstraction, would be at a maximum when the dihedral angle + between the C-H bond and the p-type orbital on the oxygen(s) is 0", and would be at a minimum when this angle is 90".Hydrogen abstraction from a number of cyclic and bicyclic ethers was investigated in order to test this hypothesis.28 The hydrogens adjacent to oxygen in tetrahydrofuran, for which + = 30°, were found to be about 2900 times as reactive as those of cyclopentane, whereas hydrogen abstraction from (43), where += 90" for the two bridgehead hydrogens adjacent to the oxygen, was undetectably slow. For (44) only the radical formed by abstraction of Ha (+ = 30") was detected, with no trace of the radical which would be formed by loss of Hb (# = 70"), nor even that which would be formed by loss of H, though this has two neighbouring oxygens (+ = 50"and 70"). The bridgehead hydro- gens in (45) and (46) (+ = 30" and 70") were abstracted in preference to other hydrogens in these molecules.In (47) and (48) the bridgehead hydrogens adjacent to three oxygens (# = 90°, 90°, 90") were not abstracted. This, together with other evidence, showed that a pronounced stereoelectronic effect of the type outlined above is of crucial importance in hydrogen abstraction from ethers. H 0 H o (47) (48) A series of aza- and di-azabicyclo[2.2.2]octylspecies has been investigated. For diazabicyclo[2.2.2]octaneitself (49), and its radical cation (50), experimental and theoretical evidence showed that there is a very strong stabilizing interaction between the nitrogen orbitals.1s,29 The magnitudes of the EPR a(Hb,) values for the mono-aza quinuclidine cation radical (51) and the quinuclidin-4-yl radical (52) (see Table 3) indicate however that while there is some TB coupling to the 4-position it is not particularly strong.Chemical studies revealed that (52) is formed only three times faster than the hydrocarbon radical (7), free-radical chlorination of (52-H) showed normal bridgehead reactivity, and electrochemical reduction of the bromide pre- cursors of (52)and (7) showed that they have similar reduction potentials. It can be concluded that radicals (7), (51), and (52) are not stabilized by long-range electron delocalization.It is likely that (49) and (50) represent special cases in which electron delocalization, and hence stabilization, result from symmetry- induced degeneracy of the interacting (49) 4.4LG 7 Thermochemistry of Bridgehead Radicals Thermochemical and kinetic data provide the best means of quantifying the stability and reactivity of free radicals and are extremely useful aids to mechanistic analysis and in the design of syntheses. The enthalpies of formation of bridgehead radicals, dHf(Rbr'), and the C-Hb, bond dissociation energies of the corresponding hydrocarbons, DHo(R-Hb,), summarize and organize key facts. Leading information is thereby made avail- able in a readily usable form which enables one radical to be meaningfully compared with another, and with other reactive intermediates.A start has been made in the collection of this type of data for bridgehead radicals, but so far theoretical predictions outnumber reliable experimental facts.The heat of formation of the norbornyl radical (8) was derived from a study of the reaction of norbornyl iodide with hydrogen iodide.30 Recent work on this type of iodination system has indicated that radical heats of formation and the corresponding C-H bond dissociation energies can only be obtained accu- rately if the individual rates of the reactions of each radical with HI are determined.s For secondary and tertiary alkyl radicals this leads to an increase of ca. 3 kcal mol-' over previous A$'(R') and DHo(R-H) values. If we arbitrarily assume that a similar correction of the original results is needed for the norbornyl radical then the estimated A@ (norbornyl) and DHo (norbornyl-H) values become 35.6 and 99.7 kcal mol- respectively. In a different approach, the rates of abstraction of iodine by phenyl radicals from a series of alkyl iodides, including I-iodoadamantane, l-iodobicyclo[2.2.2]octane,and l-iodobicyc- 10[2.2.Ilheptane, were measured relative to the rate of bromine abstraction from bromotrichloromethane.3 The relative rates were corrected for polar effects by use of the Taft (T* parameters.A linear correlation of these modified rate constants [kl/ksr]cor with the DHo(R-H) values of the corresponding hydrocarbons was obtained.The bond dissociation energies of the bridgehead compounds were derived by comparing their corrected relative rates with this correlation. If the most recent valuess of DHo(R-H) are used for this correlation it becomes: BRIDGEHEAD RADICALS-J. C. WALTON Table 4 Thermochemical data for bridgehead and related radicals" DHo(R-H) DHo(R-H) AHdR') A Hf(R') Radical Exptl. AM 1 Exptl. AM 1 Bu' 95.9h 78.8 1 1 .tjh -2.7 Methyl 104.8h 34.9h C ycloprop yl 106.3 93.5 59.2 1-Adamantyl(6) 99.6' 88.0 15.7' -7.3 1-Adamantyl(6) 97.0d 13.3d Bicyclooct-1-yl (7) 97.7" 89.4 22.0d 1.3 Bicyclohept-1-yl (8) 97.7d 99.7 35.6d 33.1 Bicyclohept-I-yl (8) 101.8' 36.7' Cubyl(9) 106.8' 110.7 205.Y 209.8 Bicyclo[ 1.1 .O]but- 1-yl (23, n = 1) 116.7 142.7 Bicyclo[2.1.O]pent-l-y1(23,n = 2) 109.7 103.4 Bicyclo[3.1 .O]hex-l-yl (23, n = 3) 102.2 58.7 Bicyclo[2.2.0]hex- l-yl(28) 100.9 70.0 Bicyclo[3.2.0]hept-1-yl (3 1') 92.1 29.5 Bicyclo[1.1. llhex-1-yl (22) 108.7 136.6 Bicyclo[2.l.l]hept-l-yl(21) 105.9 32.9 Data in kcal mol-'.From ref. 5. From PES data, see ref. 18 and text. From correlation of relative rates of iodine abstraction by Ph' radicals with DHO fdata, see ref. 31 and text. From reaction of I-iodonorbornane with HI, see ref. 30 and text. From ah initio computed energy relative to t-butyl, see ref. 32 and text. DHo(R-H)/kcal mo1-l = 96.0 -8.021og[k~/k~,],,, with r2 = 0.975; the thermochemical data estimated from this equation are given in Table 4.From the difference in ionization energies of the t-butyl and 1-adamantyl radicals, and the known difference in hydride affini- ties of the corresponding cations, the bridgehead bond dissocia- tion energy of adamantane was found to be 3.7 kcal mol-l greater than the tertiary C-H bond dissociation energy of isobutane.' Using the most recent value for this latter quantity (95.9 kcal mo1-1)5 we find DHo(Ad-H) = 99.6 kcal mol-'. Experimental data are not available for any highly strained bridgehead radicals. The closest approach we can make to this comes from ah initio calculations comparing the computed energy of cubyl with that of t-butyl radicals.32 Incorporation of the experimental enthalpies of formation of cubane, t-butyl, etc.gives: DHO(cuby1-H) = 106.8 kcal mol- I. Thermochemical parameters for a series of strained bridgehead and related radicals, computed using the semi-empirical AM 1 approach, are also displayed in Table 4. The two experimental DHo(R-H) values for the norbornyl radical (8) are nearly in agreement; there is a greater discrepancy in the adamantyl data, but it is probably not beyond the combined experimental error. The DHo(R-H) values increase through the series from radical (6) to radical (9) (ignoring the higher experimental value for adamantyl); this is in line with the increase in internal strain through this series, the increasing s-character of the orbitals, and the decrease in front strain.For radicals (6)-(8) the DHo(R-H) values are all greater than that of isobutane, and less than that of methane, as would be expected. The 'experimental' C-H bond dissociation energy for cubane looks reasonable for such a strained molecule in that it is slightly greater than that of methane and comparable to that of cyclo- propane. When Bu'O' radicals abstracted hydrogen from a mixture of cubane and cyclopropane, both cyclopropyl and cubyl radicals were detected,33 which indicates that their bond dissociation energies are not very different. Unfortunately, the AM 1 calculated DHo(R-H) values show large differences from the experimental values (including the well established experi- mental values for methane, isobutane, etc.). The computed results are probably least reliable for radicals containing 3-and 4-membered rings; the limitations of the semi-empirical methods for these species are well known.However, the trend of increas- ing DHo(R-H) from radical (6) to (9) is reproduced by the calculations. The computed value for bicyclo[ 1.1.llpent- l-yl (22) is greater than that of cyclopropyl; this agrees with experi- ment in that only cyclopropyl radicals were detected on hydro- gen abstraction from a mixture of cyclopropane and bicyclo- [1.1.Ilpentane.' The computations also predict very high bridgehead C-H bond dissociation energies in bicyclo[ 1.1 .O]bu- tane and bicyclo[2.1 .O]pentane which is in line with expectation and suggests that these particular bridgehead radicals will be very difficult to generate and detect.8 Conclusions Bridgehead radicals are strongly pyramidal, much more so than acyclic tertiary radicals, yet they can be generated and observed with ease by the usual methods employed for free radicals. They do not disproportionate because this would produce highly strained bridgehead alkenes. Bridgehead radicals, even those with extremely high internal strain, either fail to rearrange, or do so reluctantly. Sometimes this is because 8-scission would produce an anti-Bredt alkene, in other cases unfavourable stereoelectronic effects inhibit rearrangement. In general, bridgehead radicals take part in conventional abstraction and addition reactions, but are more reactive and less selective than t-butyl radicals.Internal strain and the degree of steric 'expo- sure' of the radical centre play important parts in governing the reactivity. The well defined geometries possessed by bridgehead radicals have made them ideal tools for testing theories about hyperconjugation, through-space and through-bond effects, stereoelectronic effects, etc. Although a great variety of bridge- head radical reactions have been examined, and some of these are synthetically useful, quantitative studies of their thermo- chemistry and stability are in their infancy. The same is true of spectroscopic studies leading to structural information such as the extent of pyramidality at the radical centre. These are conspicuous areas for profitable future research.9 References and Notes 1 None of the Kepler-Poinsot polyhedra such as the stellu octungulu, or other combinations of the basic Platonic structures, are applicable as models of organic structures. 2 The name implies a self-contradictory character for these species! 3 See for example, G. W. Griffin and A. P. Marchand, Chem. Rev., 1989,89,997; L. A. Paquette, Chem. Rev., 1989,89, 1051, and other articles in the same issue. 4 J. Lusztyk, personal communication. 5 J. A. Seetula, J. J. Russell, and D. Gutman, J. Am. Chem. Soc., 1990, 112, 1347; D. Gutman, Ace. Chem. Res., 1990. 23,375. 6 R. C. Fort and P. von R. Schleyer, Adv. Alicyclic Chem., 1966,1,283: R. C. Bingham and P. von R. Schleyer, J. Am. Chem. Soc., 1971,93, 3189; C. Riichardt.Angew. Chem., Int. Ed. EngI., 1970, 9, 330; V. Golzke, F. Groeger, A. Oberlinner, and C. Riichardt, Nouv. J. Chim., 1978, 2, 169; P. S. Engel, Chem. Rev., 1980,80,99. 7 J.-M. Surzur, in ‘Reactive Intermediates’, ed. R. A. Abramovitch, Plenum, New York, 1982, Vol. 2, Chapter 3, p. 121. 8 A. L. J. Beckwith, I. A. Blair, and G. Phillipou, Tetrahedron Lett., 1976, 2251; A. L. J. Beckwith, G. Phillipou, and A. K. Serelis, Tetrahedron Lett., 1981, 281 1. 9 A. G. Yurchenko, L. A. Zosim, N. L. Dovgan, and N. S.Verpovsky, Tetrahedron Lett., 1976,4843. 10 M. L. Poutsma, in ‘Methods in Free-radical Chemistry’, ed. E. S. Huyser, Dekker, New York, 1969, Vol. 1, Chapter 3, p. 79. 11 C. V. Smith and W. E. Billups, J. Am. Chem. Soc., 1974,96,4307.12 J. C. Walton, J. Chem. Soc., Perkin Trans. 2, 1988, 1989. 13 B. Maillard and J. C. Walton, J. Chem. SOC.,Chem. Commun., 1983, 900. 14 J. C. Walton, J. Chem. Soc., Perkin Trans. 2, 1987, 231. 15 S. Bank, W. K. S. Cleveland, D. Griller, and K. U. Ingold, J. Am. Chem. SOC.,1979,101,3409. 16 P. J. Krusic, T. A. Rettig, and P. von R. Schleyer, J. Am. Chem. Soc., 1972, 94, 995. 17 S. P. Mishra and M. C. R. Symons, Tetrahedron Lett., 1973,2267. 18 G. H. Kruppa and J. L. Beauchamp, J. Am. Chem. SOC.,1986,108, 2 162. 19 K. U. Ingold and J. C. Walton, Ace. Chem. Rex, 1986, 19, 72. 20 J. C. Walton, J. Chem. Soc., Chem. Commun., 1987, 1252; J. C. Walton, J. Chem. Soc., Perkin Trans. 2, 1988, 1371. 21 K. B. Wiberg, S. T. Waddell, and K.Laidig, Tetrahedron Lett., 1986, 1553; A. C. Friedli, P. Kaszynski, and J. Michl, Tetrahedron Lett., CHEMICAL SOCIETY REVIEWS, 1992 1989,455; P. Kaszynski and J. Michl, J. Am. Chem. Soc., 1988,110, 5225; G. S. Murthy, K. Hassenruck, V. M. Lynch, and J. Michl, J. Am. Chem. SOC.,1989,111,7262. 22 C. Riichardt, K. Herwig, and S. Eichler, Tetrahedron Lett., 1969, 421; B. Giese, Tetrahedron Lett., 1979, 857; B. Giese and J. Stell-mach, Chem. Ber., 1980,113, 3294. 23 E. W. Della, P. E. Pigou, C. H. Schiesser, and D. K. Taylor, J. Org. Chem.,(in the press); E. W. Della and 3.Tsanaktsidis, Aust. J. Chem., 1989, 42, 61. 24 P. E. Eaton and T. W. Cole, J. Am. Chem. Soc., 1964,86,962,3157. 25 E. W. Della and D. K. Taylor, Aust. J. Chem., 1990,43, 945; P. E. Eaton and M. Maggini, J. Am. Chem. Soc., 1988, 110, 7230. 26 D. S. Reddy, M. Maggini, J. Tsanaktsidis, and P. E. Eaton, Tetrahedron Lett, 1990, 805. 27 K. B. Wiberg and V. Z. Williams, J. Org. Chem., 1970,35, 369. 28 V. Malatesta and K. U. Ingold, J. Am. Chem. SOC.,1981, 103, 609. 29 R. Hoffmann, A. Imamura, and W. J. Hehre, J. Am. Chem. SOC., 1968,90, 1499; T. M. McKinney and D. H. Geske, J. Am. Chem. SOC.,1965, 87, 3013; E. Heilbronner and K. A. Muszkat, J. Am. Chem. Soc., 1970,92,3818. 30 H. E. O’Neal, J. W. Bagg, and W. H. Richardson, Znt. J. Chem. Kinet., 1970, 2,493. 31 W. C. Danen, T. J. Tipton, and D. G. Saunders, J. Am. Chem. Soc., 1971,93, 5186. 32 D. A. Hrovat and W. T. Borden, J.Am. Chem. SOC.,1990,112,3227. 33 E. W. Della, G. M. Elsey, N. J. Head, and J. C. Walton, J. Chem. Soc., Chem. Commun., 1990, 1589.

ISSN:0306-0012    

DOI:10.1039/CS9922100105

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

RHON E-POULENC LECTURE* Search and Discovery of New Antitumour Compounds Pierre Potier lnstitut de Chimie des Substances Naturelles, Centre National de la Recherche Scientifique, Avenue de la Terrasse, 9I 198 Gif S/Yvette Cedex, France 1 Introduction Until the latter part of the nineteenth century, plants, minerals, and, more rarely, animals constituted the major sources of drugs. Consequently, natural products form the basis of thera- peutics. For more than a century the pharmaceutical industry maintained an important effort in trying to select new and interesting drugs from Nature but since 1970 this field of research has been practically abandoned. However, the tremen- dous progress realized in biology and pharmacology in recent times advocates another look at ‘old drugs’.Although some of these have been known for a very long time, knowledge of their pharmacological profile is often limited and would certainly be increased by looking at them from a new perspective. Further- more, most of the natural products so far structurally character- ized (and sometimes synthesized) would have never been created, even by the most ingenious chemist. The structure of morphine, for instance, is not that complex; however, we can hardly imagine this structure ever being synthesized de novo by Man. There are other facts that should encourage us to continue to study the natural world: Fewer than 10% of the plants on our planet have been examined for their biological properties. This ‘terra incognita’ is therefore immense.Among the major drugs which have been discovered in recent times, several have been isolated from natural sources: immuno- modulating substances (cyclosporin, FK 506), inhibitors of the biosynthesis of cholesterol, new antibiotics, new antitumour compounds, compounds interacting with the nervous system, etc. 2 Search and Discovery of New Antitumour Compounds Cancers and related diseases represent the second major cause of death for Man. The aetiology of cancers is multiple but great progress has been made in recent times towards the understand- ing of these diseases with the discovery of oncogenes. The links between these genetic ‘templates’, their expression products, and the various growth or differentiation factors are becoming more and more important.Such discoveries will undoubtedly lead to more rational approaches to the treatment of cancerous diseases. Pierre Potier was born near Paris in 1934 and obtained his Diploma in Pharmacy (1957) andPh.D. (1960)from the Univer- sity of Paris. He has been, since 1962, at the Institut de Chimie des Substances Naturelles at Gif-sur-Yvette with Professors M.-M. Janot, E. Lederer, D. H. R. Barton, and G. Ourisson. He became Director of this Institute in 1990 as well as, in the same year, Professor of Chemistry at the Muskum National d ’Histoire Naturelle in Paris. He has been a Visiting Professor at the University of Strathclyde (1984-1990). Professor Potier has published more than 300 papers in the fields of the chemistry of natural products, organic chemistry, and medicinal chemistry. He is a member of the French Academy of Sciences.For the time being, surgery, radiotherapy, and chemotherapy remain the three most important methods of treating cancers. Immunotherapy is still in its infancy but remains a serious hope. Cancer chemotherapy relies principally on the use of a number of natural products, intact or chemically modified, and secondly, on purely synthetic products (some of them, metho- trexate for instance, being the mimics of naturally occurring substances). These compounds have been classified according to their known (or supposed) mode of action: (i) the alkylating agents which are given to ‘alkylate’ several crucial biological targets (nucleic acids, for example) impair- ing their normal mode of functioning.These agents are: cyclophosphamide, nitrogen mustards, platinum derivatives, etc. (ii) the intercalating agents whose activity is based upon their ‘intercalation’ between the base pairs of nucleic acids (mostly, but certainly not exclusively, DNA). Examples of such com- pounds are adriamycin, platinum derivatives, and bleomycin. (iii)the antimetabolites which are compounds that play the role of lures vis-ci-vis metabolic processes: examples are metho- trexate, fluoro-uracil etc. (iv) the spindle poisons which either prevent the formation of the spindle (colchicine, maytansine, vinblastine-type com- pounds) or stabilize it (taxol and derivatives) during the cell division. We have chosen to work on the intercalating agents and, more intensively, on the spindle poisons.3 Antitumour Compounds of the Ellipticine Group Ellipticine (I) is an alkaloid isolated in 1959 by Horning, Goodwin, and co-workers’ from the leaves of Ochrosia elliptica Labill, (Apocynaceae) and found, since then, with its congener, olivacine (4)in many other plant sources belonging to the Apocynaceae and, less frequently, to the Loganiaceae families. Ochrosia oppositifolia (Lmk) K. Schum now known as Neisos-perma oppositifolia (Lmk) Forsburg, Sachet, Boiteau was des- cribed by Rumphius (von Rumpf) in 1742 who reported its use in folk medicine against cancers of the nose.2 A review of this subject has been p~blished.~ Ellipticine (1) and 9-methoxyellipticine (2) have been found to be highly active compounds on various experimental tumours and leukaemias; 9-methoxyellipticine is, at the moment, one of the substances having the broadest spectrum of antitumour activity. Our first chemical studies of this series of compounds were made by Poisson, Le Men, and Dat-Xuong in collabor- ation with biochemists, biologists, and physicians (Paoletti, Le Pecq, Mathe, and their co-workers).They were quickly able to demonstrate that DNAs were the biological target of this type of molecule. They focused their efforts on Celiptiump (3) which they found to be the most active compound of the series. This compound has been marketed but has had a rather limited career in cancer chemotherapy.This outcome does not necessar- ily mean that compounds of the ellipticine series [or of the * Delivered on 10th April 1991 at Imperial College London during the 150th Anniversary Annual Chemical Congress of the Royal Society of Chemistry. 113 I14 CHEMICAL SOCIETY REVIEWS, 1992 (1) R = H Ellipticine (3)ciliptium@ (4) Olivacine (2) R=OMe olivacine (4)series] cannot find their place in the armamentar- ium of cancer chemotherapy, and other compounds of this series should be evaluated. The mechanism of action of ellipticine in vivo has been a matter of considerable speculation more or less supported by experimentation, but which has been rarely con- firmed. Ellipticine derivatives generally bind to double-stranded DNA (1 0-to 10 -M), destroying kinoplastic DNA and base- pairing ability, thus denaturing DNA (for a review, see reference 3).The high affinity of ellipticine and its derivatives for DNA and their intercalation between DNA base pairs have long been considered the two main factors responsible for their anti- tumour activity. However, certain derivatives such as the 9- amino or 9-fluoroellipticine, although exhibiting quite a high affinity for DNA, have little or no in vivo activity. The strong activity of 9-hydroxyellipticine (2), an easily accessible metabo- lite of ellipticine, suggests that ellipticine (1) has to be metabo- lized to this compound before being active. This metabolization is enhanced in patients receiving barbiturates, which are known to be potent cytochrome P-450inducers.Taking into account that ellipticine has to be oxidized in vivo to be active, Dat-Xuong prepared Celiptiumm (3), a compound having a high affinity for DNA preparations. However, a hypothesis has been put forward that over-oxidation of 9-hydroxyellipticine or Celiptium@ occurs to give an imino-quinone system (see Figure 1). This reacts readily with various nucleophiles such as pyridine, amino acids, a-substi- tuted primary amines, sulfydryl derivatives, methanol, ethylene- glycol, ribonucleosides, and ribonucleotides to give the corres- ponding and rather stable adducts. It is important to emphasize that some of these reactions (or their equivalent) can take place in vivo and thus explain at least a part of the biological activity of ellipticine and structurally related compounds [in particular all those able to lead to an iminoquinone (or quinoid) structure].A given antitumour agent has to penetrate the cell, cross the cytoplasm, and enter the nucleus before being able to interfere with the DNA structures in the nucleus (unprotected by their usual chromatin). Upon crossing the cytoplasm they can well encounter other nucleic acid structures such as the RNAs, whose role is of crucial importance in the transfer of genetic infor- mation, protein synthesis, and a number of other basic biologi- cal processes (co-factors, ATP, GTP, etc.).The high reactivity of ribonucleosides or ribonucleotides with 9-hydroxyellipticine derivatives under oxidative conditions leading to the formation of regiospecific and stereoselective ketal linkages between 2’ and 3‘ hydroxyl groups of the ribose moiety of these compounds and the 10-position of ellipticine derivatives, suggests the occurrence of these reactions in vivo.Among the four naturally occurring ribonucleosides used in this reaction [A, G, T(or U), C], the order of reactivity for hydroxyellipticine is: A > G > > > T(U) > C. The decreasing reactivity in going from purines to pyrimidines is probably due to a better affinity (‘recognition’) of adenine and guanine for the pyrido-carbazole moiety of ellipticine, as has been shown by NMR measure- ment~.~The poly-A tail of the m-RNAs, or the ‘cap’ present at the 5’ end of m-RNAs, are obvious targets for such reactions.The formation of dioxolanes (with ribose or related glycolic structures) and oxazolines (with amino acids) are examples of the formation of reversible covalent bonds which play an import- ant role in the control of many biological processes. These bonds can easily be broken by slight changes of pH (or redox con- ditions), which can occur on passing from one cellular compart- ment to another. There are many biomolecules which can participate in such Figure 1 R’ I I Proteins or Disrnutation amino acids I O 0 I O=P-OH W’‘N I M R \/.- R H I Intercalates bet ween DNA base pairs (1) SEARCH AND DISCOVERY OF NEW ANTITUMOUR COMPOUNDS-P.POTIER reaction processes, in particular those able to generate iminoqui- noid systems (or their equivalent) by simple oxidation processes (serotonin, catecholamines, some steroid hormones, etc.).It is of interest to determine whether these highly bioactive substances can interact with proteins, nucleotide structures (or both in a ‘menaged. trois’),thereby temporarily blocking their function (or expression). 4 New Spindle Poisons One of the essential features of cell division is the formation (and disappearance) of the mitotic (or meiotic) spindle. The spindle plays an important role in the distribution of the chromosomes to the two daughter-cells resulting from normal cell division. Some natural products are known to impair cell division.Colchicine (5) was the first of these to be discovered; its pharmacological action is that of the plant which contains it, Colchicum autumnale L. (Liliaceae), used for the treatment of gout. Its biological target has been relatively recently identified as tubulin, a ubiquitous protein present in all eukaryotic orga- nisms. Tubulin is a heterodimeric protein (2 x 55000 daltons) which polymerizes (GTPase activity) leading to microtubules and microfilaments which, after further assembly, constitute the ~pindle.~Colchicine prevents the polymerization of tubulin into microtubules. Other natural products are also known to inter- fere with the formation of the spindle: podophyllotoxin (6), vinblastine (7) [or vincristine (S)], taxol(9), and their derivatives, etc.Microtubules are generally accompanied by the so called ‘MAPS’ (Microtubule Associated Proteins). These are proteins which play an important role in the various biological functions of microtubules or microfilaments (i.e. axonal transport, brain organization, hormone secretion, motility of various cells, etc.). One of them, T (tau) protein, has been recently shown to be involved in some degenerative nervous disorders such as Alz- heimer’s disease. These proteins may also have other important functions. Tubulin constitutes the major, if not unique, target of the spindle poisons, which inhibit polymerization of tubulin into microtubules or, less frequently, its depolymerization [like taxol(9)l.This property forms the basis of a simple and rapid biological test which can be used for selecting new spindle poisons. The test consists of measuring inhibition of polymeriza- tion (or depolymerization) of a preparation of tubulin (generally extracted from pig or sheep brain) by increasing quantities of a potential inhibitor. Me0Meom--NHCOMe OMe Me (5) Colchicine OH We have developed this test, based on Shelanski’s method,‘j and have found that it considerably shortens the time required to perform classical pharmacological testing on in vitro and in vivo preparations. It spares the lives of many animals as evaluation on animal tumours can then be restricted to those products being found active in the ‘tubulin test’.4.1 Navelbinem We first applied this screening method to the antitumour alkaloids of the vinblastine group.7 The biomimetic-type syn- thesis of this group of structurally complex natural products was achieved for the first time in our laboratory as early as 1974.8 This synthesis is based on the modified Polonovski reaction9 which uses the trifluoroacetic anhydride in place of the acetic anhydride originally used in the genuine Polonovski reaction. The imminium ion which is formed in the first stage of the reaction is stable in the presence of trifluoroacetate ion while, in the Polonovski reaction, this imminium suffers from the attack of the acetate ions. The ‘dimeric’ alkaloids of the vinblastine group (7) are formed by the coupling of two structural subunits, catharanthine (10) and vindoline (1 I), which are present in the same plant, the ‘Madagascan periwinkle’ [Catharanthus roseus G.Don (Apocynaceae)] and related specie^.^ When catharanthine N-oxide (12) is treated with trifluoroace- tic anhydride in the presence of vindoline (1 l), one gets, after reduction in situ, anhydrovinblastine (1 3) as a major product, possessing the 16’s configuration of the natural products. All previous synthetic efforts led invariably to the biologically inactive 16‘R configuration (Scheme 1). Anhydrovinblastine (13) constitutes the precursor of a number of other ‘dimeric’ indole alkaloids. Vinblastine (7) and vincristine (8) are obtained differentl~;~they are among the most often used compounds in the field of cancer chemotherapy.Our efforts were partly motivated by the prices of these drugs (3 M$/kg for vinblastine and 20 M$/kg for vincristine!). Also, synthesis has made avail- able compounds otherwise inaccessible by direct structural modifications of the natural products e.g. Navelbine‘Q or noranhydrovinblastine (14)’O (Scheme 2). Navelbinem is currently used in the treatment of non-small- cell lung cancer and of breast cancer. It is orally active and its use will almost certainly be extended to the treatment of other cancerous diseases. Vinblastine-type compounds interact, like other spindle poi- sons, with tubulin. The compounds have to be biosynthesized in such a way as to prevent them from reacting with the tubulin of OM, (6) Podophyllotoxine (7) (R = Me) Vinblastine (9) Taxol(8) (R = CHO) Vincristine CHEMICAL SOCIETY REVIEWS, 1992 0-Q-77 Q--+ C0,Me C02Me (10) Catharanthine (11) Vindoline (12) Catharanthine Nb-oxide - I (cF3c0)20 O-COCF,c-J qq CO2Me C0,Me (13) Anhydrovinblastine Scheme 1 Me- CO,Me Cb,Me- ’ ’ C02Me Anhydrovinblastine Nb-oxide (14) Nor-anhydrovinblastine= Navelbine@ Scheme 2 the plant which produces them.Otherwise, vinblastine would be toxic to Catharanthus spp., colchicine to Colchicum, etc. just as animal venoms should also be toxic to the animals which produce them! This problem is avoided by vinblastine and related compounds being biosynthesized and localized in vacuoles.There is no definitive proof that cytochrome P450-type enzymes are responsible for the coupling reaction between catharanthine (10) and vindoline (1 1) leading, after reduction (NADPH ?) to vinblastine-type compounds, although these enzymatic systems are known to be present in (or around) the vacuoles. NavelbineB (14) is formed from anhydrovinblastine (13), one of the biogenetic precursors of other dimeric alkaloids of Catharanthus spp.We think that NavelbineB (14) is probably not a naturally occurring compound: in order to obtain Navelbine,@ anhydrovinblastine has to be transformed by a modified Polo- novski reaction, implying the formation of the Nb’-oxide of anhydrovinblastine (or its biogenetic equivalent) followed by a fragmentation reaction (Scheme 2).However, the pH of the vacuoles is acidic,” as shown by 31PNMR measurements, which suggests that Nbfof anhydrovinblastine is protonated and not prone to further reaction. The biological mode of action of Navelbinea resembles that of other spindle poisons of the vinblastine group; however, it has structural peculiarities which can partly or entirely explain the distinct pharmacological profile of this new antitumour agent. Thus Navelbinem contains an eight-membered ring instead of the nine-membered ring found in vinblastine or related com- pounds. One knows’O that Nb’ of vinblastine-type compounds must be ‘free’ (protonated?) for these to exhibit antitumour activity. Whatever the exact mode of interaction of these types of compounds with tubulin, it is quite probable that after their ‘recognition’ by tubulin (hydrophobic interactions) further interactions come into play and protonation of Nbfof the upper part of vinblastine-type compounds by an amino acid of tubulin is possible.In the case of vinblastine-type compounds, there is no further consequence of this protonation. In the case of NavelbineB, however, protonation of Nb’ of the ‘gamine por- tion’ of that alkaloid (instead of the ‘tryptamine portion’ in the equivalent vinblastine-type compounds) can be followed by a fragmentation reaction which offers the possibility of nucleophi- lic addition of a suitable group of tubulin onto the newly formed conjugated imminium indolic ion (see Scheme 3).This newly formed bond is a reversible covalent bond whose role should be important in the control of the fate of tubulin, microtubules, spindle, etc. (see above concerning the formation of ellipticine derivatives with ribonucleosides or nucleotides). 4.2 TaxotGreB Taxus baccata L. (Taxaceae) is the most commonly encountered species of yew tree in Europe. There are other species of Taxus (or related genus) all over the world, but they do not appear to differ significantly in their content of secondary metabolites. The SEARCH AND DISCOVERY OF NEW ANTITUMOUR COMPOUNDS-P. POTIER Protein (tubulin) yrmyy Protein (tubulin) - I I 0 +/ H+ - C02Me fl Protein (tubulin) - I ,NH / (14) Navelbine@ Scheme 3 wood of the yew tree is dense, imputrescible, and mould- resistant, and has been largely used for making bows [those of the battle of Crecy (1 346) for instance] and furniture. However, it is slow growing and one or two centuries are necessary to get a good specimen of tree.Its toxicity has been known since Antiquity. Practically all parts of the tree are toxic with the exception of the red fleshy envelope of the fruit which is eaten by birds (and the seeds are therefore disseminated). The first reported work on Tuxus constituents is due to Lucas12 who, in 1856, isolated from the leaves a toxic alkaloid named taxin. In 1921 Wintersteinl3>l4 identified a degradation product of taxin -3-dimethylamino-3-phenylpropionic acid (1 5)-known as 'Winterstein's acid'. Lythgoe and co-workers1 showed that the skeleton of taxin called taxane, is of diterpenic nature.It was only in 1971 that Wall and co-workersl6 isolated taxol(l6) from the stem bark of an American yew, Tuxus brevifoliu Nutt., and disclosed its rather complex chemical structure. Taxol was later isolated from various other Tuxus species, although in relatively low yields (0.14.2 g/kg of stem bark). It possesses a cytotoxic activity due to its unique mode of action on the microtubule proteins responsible for the formation of the spindle during cell division. While all other known spindle poisons have been shown to interfere with the polymerization of tubulin (see above), taxol and its derivatives are known to stabilize the spindle or to promote the assembly of microtubules into microfi- laments and, finally, into the spindle. For a review on this topic, see reference 17.The cytotoxic properties of taxol were recognized by Wall and co-workers' on KB cells and, later, on leukaemias L1210, P388, and HeLa cells. This activity has been related to the interaction of taxol with the tubulin-microtubule system. Taxol and derivatives are the only products known to favour microtubule assembly into microfilaments and spindle; they stabilize the spindle which, after cell division, should normally vanish. Taxol has been submitted for clinical evaluation (phases I and 11) in both the United States and France and it appears to be an exceptionally promising drug. It has a very broad spectrum of PhOCNH 0 PhPOOH NMe, (15) Winterstein's acid (16) Taxol activity against leukaemias and solid tumours and has already been successfully used in the treatment of ovarian cancers where other therapies proved to be ineffective. Taxol is extracted from the stem bark of several species of yew (Tuxus spp). However, the isolation procedure is tedious, low- yielding, and an obvious ecological problem.There is a wide- spread concern that 'if taxol proves effective . . . the yew population could be so severely depleted that there would not be enough trees left to make treatment successful'. Obviously, other sources have to be found to meet with the expected increasing demand. The National Cancer Institute of the USA has contracted for 27 000 kg of yew stem-bark.l9 Attempts to achieve a total synthesis of taxol have not succeeded so far. However, we have found' an efficient solution to the problem of taxol supply. We analysed the various parts of Tuxus baccutu and isolated 10-deacetylbaccatin I11 (1 7), from the leaves of this species. These are quickly regenerated and can be harvested in rather large amounts without detrimental effects on the yew population. The production of taxol (16) (and derivatives) from 10-deacetylbaccatin I11 (1 7) is simple. Selective protection of (17) at the C-7 position (triethylsilylation) and acetylation at the C- 10 position is followed by forced acylation of the secondary alcohol at C-13 (largely unreactive under normal conditions) by the suitably protected N-benzoyl phenyl- isoserine side chain.Deprotection of both the C-7 and C-2' secondary alcohol functions leads to taxol(16)' (Scheme 4). In another approach20 10-deacetylbaccatin I11 (1 7), suitably protected at both the C-7 and C-10 positions is converted into the cinnamoyl ester at the C-13 position. The cinnamic double bond is oxyminated (Sharpless method), leading to the four different possible isomers (Scheme 5). The 2'R,3'Sisomer is then conveniently deprotected and N-benzoylated to give taxol(l6). The poor diastereoselectivity of this method can be improved by using chiral ligands during the oxy-amination reaction. This approach led to diastereomers of taxol necessary for Structure-Activity Relationship studies.In addition to taxol (16), we were able to obtain several derivatives, one of which was given the name Taxotkre@ (18) and has revealed interesting OCOPh (17) 10-Deacetylbaccatin111 CHEMICAL SOCIETY REVIEWS, 1992 OCOMe 0 OSiEt3 H O . - a T H O . - a o OH : dAc OH I dAc kOPh 6COPh (17) 10-Deacetylbaccatin III OH 6COPh 2’R,3’S and 2’S,3’R Zn/AcOH\\ ButoCONH 0 pharmacological properties.2 ,2 It has better bioavailability and pharmacological characteristics than taxol and is a promis- ing new anticancer agent (Scheme 6). The exact mode of action of taxol derivatives is not clearly understood. Although tubulin constitutes its major (if not unique) biological target, it remains to be seen whether some of the Microtubule Associated Proteins (MAP’S) play a role in the interaction between taxol derivatives and tubulin.Work is in progress which will allow a better understanding of the very promising therapeutic activity of this type of c~mpound.~ 5 Conclusion and Prospects Arthur Kornberg has written:23 ‘Much of life can be understood in rational terms if expressed in the language of chemistry. It is an international language, a language for all time, and a language that explains where we came from, what we are, and where the physical world will allow us to go. Chemical language has great esthetic beauty and links the physical sciences to the biological sciences. Unfortunately, the full use of this language to understand life processes is hindered by a gulf that separates chemistry and biology.This gulf is not nearly as wide as the one between the humanities and sciences. Yet, chemistry and biology are two distinc- tive cultures and the rift between them is serious, generally unappre- ciated and counterproductive.’ PhCOyHOCoMe 0 OSiEt, PhCoNHbR (16) TaxolPh do--@&^I OH : &c bCOPh Scheme 4 OH ! OAc OCOPh Oxyamination OsO4(l%) AgN03 BubpCNClNa 6COPh 2‘R,3’S and 2‘S,3‘R 6COPh (18) Tadre* Scheme 5 In this article we have tried to show that medicinal chemistry constitutes an unlimited field of research open to those being chemists and biologists at the same time: the biochemists of modern times. There is no logical boundary between chemistry and biology and careful chemical analysis of biological pro- cesses will continue to lead to fundamental discoveries. One consequence of those discoveries will inevitably be progress in therapeutics. 6 References 1 S.Goodwin, A. F. Smith, and E. C. Horning, J. Am. Chem. Soc., 1959,81, 1903. 2 Rumphius in ‘Herbarium Amboinense, 11’, 1742, p. 255. 3 V. K. Kansal and P. Potier, Tetrahedron, 1986,42, 2389. 4 D. Broek, R. Bartlett, K. Crawford, and P. Nurse, Nature, 1991,349, 388. 5 (a) P. Valenzuela, M. Quiroga, J. Zaldivar, W. J. Rutter, M. W. Kirchner, and D. W. Cleveland, Nature, 1981,289, 650. (b) L. E. Cameron, J. A. Hutsul, L. Thorlacius, and H. €3. Lejohn, J. Biol. Chem., 1990,265(25), 15 245.6 M. L. Shelanski, S. Gaskin, and C. R. Kantor, Proc. Natl. Acad. Sci. U.S.A., 1973,70, 765. 7 P. Potier, ‘Actualites de Chimie Therapeutique’, 14e Serie, Masson Edit., Paris, 1987, and references cited. 8 P. Potier, Pure Appl. Chem., 1986,58, 137. 9 D. Grierson, ‘Organic Reactions’, 1990, Vol. 39, ed. Lko Paquette, J. Wiley & Sons, 1990, p. 85. SEARCH AND DISCOVERY OF NEW ANTITUMOUR COMPOUNDS-P. POTIER Phdo.\HO. bCOPhOCOPh \ / I (17) 1O-DeacetylbaccatinIII (R = H) Baccatin III (R = COMe) HO--.-a. kOPh R’ = C02CH2CCl3 bH A; IR2= COMe or C02CH2CC13 OH OAc bCOPhVHCOPh VHCO$ut NHCOPh NHCO~BU‘ IbR4 OH OCOPh 6COPh (16) Taxol (R‘ = R4 = H, R2 = COMe (18) Taxot&e@ (R’ = R2 = H) Scheme 6 10 (a) Symposium on ‘Navelbine’, Pierre Fabre Oncologie, Biarritz, 19 (a)J.-N.Denis, A. E. Greene, D. Guenard, F. Gueritte-Voegelein, L. 2-3 November 1989. (b)UICC Symposium on ‘Navelbine: Recent Mangatal, and P. Potier, J. Am. Chem. Soc., 1988, 110, 5917 and results emphasize its potential’, Pierre Fabre Oncologie, Hamburg, references cited. (b)J.-N. Denis, A. E. Greene, A. A. Serra, and M.-J. Germany, August 21st, 1990. Luche, J. Org. Chem., 1986,51,46. 11 J. Guern, personal communication. 20 L. Mangatal, M.-T. Adeline, D. Guenard, F. Gueritte-Voegelein, 12 H. Lucas, Arch. Pharm., 1856,95, 145. and P. Potier, Tetrahedron, 1989,45,4177. 13 E. Winterstein and D. Iatrides, Z. Physiol. Chem., 1921, 240. 21 M. Colin, D. Guenard, F. Gueritte-Voegelein, and P. Potier, Eur. 14 E. Winterstein and A. Guyer, 2. Physiol. Chem., 1923, 175. Pat. Appl. EP 253 738 (Cl. C07D305/14) 20th Jan. 1988; Fr. Appl. 15 B. Lythgoe, K. Nakanishi, and S. Uyeo, Proc. Chem. Soc., 1964,301. 86/10400, 17th Jul. 1986, in Chem. Abstr., 1988,109,22762~. 16 H. C. Wani, H. L. Taylor, H. E. Wall, P. Coggon, and A. T. McPhail, 22 M. Colin, D. Guenard, F. Guiritte-Voegelein, and P. Potier (Rh6ne- J. Am. Chem. Soc., 1971,93,2325. Poulenc SantC) Eur. Pat. Appl. EP 253 739 (Cl; C07D305/14), 20th 17 F. Gueritte-Voegelein, D. Guenard, and P. Potier, J. Nat. Prod., Jan. 1988; Fr. Appl. 86/10401, 17th Jul. 1986, in Chem. Abstr., 1988, 1987,50, 9. 109,22763~. 18 P. B. Schiff, J. Fant, and S. B. Horwitz, Nature, 1979, 277, 665. 23 A. Kornberg, Biochemistry, 1987,26,6888.

ISSN:0306-0012    

DOI:10.1039/CS9922100113

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Structure, Dynamics, and Electronic Properties of Cobaltocene in SnS, -se,{~<X< 2) D. O'Hare Inorganic Chemistry Laboratory, South Parks Road, Oxford OX1 30R 1 Introduction Two-dimensional layered materials, such as the tin dichalcoge- nides SnS, -,Sex (0 < x < 2), have been much studied with regard to their pronounced electronic and structural aniso- tropy. The SnS, -$ex layered compounds crystallize in the Cd(OH),-type structure (space group P3ml) to form an iso- structural series of solid solutions. The three-dimensional structure is built from repeatedly stacked XMX lamellae bound together by van der Waals interactions between adjacent planes of hexagonally close-packed chalcogenide atoms (X). In SnX, (X = S, Se) the metal atoms are coordinated in nearly octahedral sites (Figure 1).4SnS2 @ =s =Sn5.89 8, l-Waals Gap Van der 4SnS2 Figure 1 Representation of the octahedral Cd(OH), type structure of SnS,. The layered MX, structure permits a variety of guest mole- cules to be inserted into the interlamellar gaps of the host material^.^ For example, the intercalation of metallic tantalum and niobium dichalcogenides by organic amines has led to the discovery of a new class of two-dimensional superconductors.4 This has raised questions as to the dimensionality of supercon-ductivity in layered structures and the relationship between charge density waves (CDWs) and the superconducting BCS mechanisms. There are also many examples of semiconducting layered structures being induced into a metallic state by intercalation of electron-donor guest species, for example, LiTiS,,6 K0,5WS,,' and K,.,MoS,.* The importance of electron-transfer from the guest to the host in these intercalation reactions has been recognized for some time.9 In view of this, the possibility of fine- tuning the electronic structure of the host material by intercala- tion has led to a vast effort towards the synthesis and characteri- zation of layered MX, materials (M = Ti, Mo, W, Ta, Nb; X = S, Se) intercalated by molecular guests such as hydrazine, organic amines, and metallocenes.O Almost no interest has been Dermot 0'Hare was educated at Balliol College, Oxford (B.A. 1982, D.Phil 1985). His D.Phil research was carried out with Professor M.L. H. Green F. R. S. on the activation of carbon-hydrogen bonds using metal atoms. In 1985 he was elected to an Exhibition of 1851 Research Fellowship and a Junior Research Fellowship at Wolfson College, Oxford. He was a temporary University Lecturer and Supernumerary Fellow of St John's College (1987-89), until his current appointment as University Lecturer in Inorganic Chemistry and Tutorial Fellow at Balliol College. He has co-authored over 60 publications and his research interests include synthetic organornetallic chemistry and its appli- cations in solid state chemistry. shown towards the intercalation of non-transition metal dichal- cogenides such as SnS, and SnSe,. ' This paper reviews our research over the past 18 months on the detailed characterization of the organometallic intercalates SnS,-,Se,{Co(~-Cp),),,,, (Cp = C,H,; 0 < x < 2).Full accounts of the crystal structure, photoelectron spectros- copy,I3 electronic conductivity,14 and solid state NMR experi-ments' of these materials have been published previously. 2 Synthesis of the Intercalates SnS2-~e,{Co(rl-Cp)2}0.3,*0.02 Single crystals of the SnS, -xSex hosts were all grown using the iodine vapour transport method. The high purity elements (> 99.99%) with a 1Yo molar quantity of phosphorus dopant and the transport agent I, (5 mg cm-3) were sealed in evacuated quartz ampoules (10 cm x 1.7 cm). A three-zone furnace provided a stable temperature gradient between the reaction zone (685 "C for SnS,) and the growth zone (645 "C for SnS,).The intercalates were prepared by adding the crystalline host (ca. 100 mg) to a solution of freshly sublimed Co(q-Cp), (ca. 150 mg) in acetonitrile (ca. 5 cm3) under a nitrogen atmosphere. In order to avoid damage to the brittle crystals the reaction was carried out at ca. 65 "C without stirring. After the reaction was complete, typically 5-21 days, the co(7-C~)~ solution was removed and the intercalate washed with acetonitrile (4 x 20 cm3) until the filtrate was colourless. The product crystals were deemed to be fully intercalated when all the host reflections had disappeared from the X-ray powder spectrum; we observed no evidence for staging at any point during the course of these reactions.Table 1 shows the lattice expansion, stoichiometries, appearance, and reaction conditions for the intercalates pro- duced in this study. The stoichiometry in each case was deter- mined by elemental microanalysis. 3 Characterization of the Intercalates SnS2-seK{c0(q-cP)2}0.33 f 0.02 3.1 X-Ray Crystal Structure Determination of sns2{c0(r)-cP)Z)0.33 The intercalation of an organometallic species into a layered host lattice has been known since Dines reported the intercala- tion of cobaltocene (CO(T-C~),> into ZrS, in 1975.l6 Since then a large number of related metallocenes have been intercalated Table 1 Stoichiometries, reaction conditions, lattice expansion, and appearance for some intercalated layered materials SnS2 -xsex{Co(T-CP),)O.33 *0.02{0-<21 Host Reaction conditions Guest stoichiometry T ("C); t (days) stoichiometry Ac (A) Colour SnS, 65;5 0.31 5.29 dark blue SnS,,,Se, 65;7 0.31 5.24 light blue SnS,,,Se,,, 65;9 0.3 1 5.20 light blue SnS,,,Se,,, 65;14 0.33 5.26 black SnS, IsSel.ss 65;17 0.33 5.50 black SnSe, 65;2 1 0.33 5.56 black 121 CHEMICAL SOCIETY REVIEWS, 1992 Perpendicular Parallel 6.96A 6.74 At Figure 2 Van der Waals dimensions of Co(?-Cp),.into layer lattices. While the physical properties of these early metallocene intercalates have been extensively studied, very little is known about their structure, and fundamental questions such as the orientation of the metallocene guests within the host remain uncertain or controversial.For example, although upon intercalation of cobaltocene into a lamellar metal dichalcogenide host the interlayer spacing typically increases from 5.89 to 11.18 8, (dc = 5.298,) as mea- sured by powder X-ray diffraction, due to the almost spherical shape of the cobaltocene molecule (Figure 2) little information can be determined from these lattice expansion measurements alone. However, we have recently been able to determine the three-dimensional structure of a single crystal of SnS, interca- lated with cobaltocene. Owing to the disorder and low crystalli- nity of these compounds it was necessary to use conventional X- ray film methods of recording the diffraction pattern, and to perform the data analysis sequentially.A one-dimensional analysis of only the 001 reflections gave the first indication that the cobaltocene guests molecules adopt an orientation with the molecular C, symmetry axis parallel to the planes of the host lattice layers. Least squares refinement of the observed structure factors derived from the zero-layer Weissenberg photograph (OM zone) confirmed that the cobaltocenes adopt a parallel orientation, and that the SnS, layers, while remaining intact on intercalation, undergo a shift relative to each other, leading to a doubling of the unit cell dimension along the c-axis. Finally, including all the observed reflections we derived a three-dimen- sional structure showing that the lattice layers shift in both the a and b directions. We believe this was the first report of a three- dimensional structure determination of an organometallic intercalate.3.2 Neutron Diffraction Experiments on S~S,(CO(&D,)~}~.~~ Comparison of the scattering factors for the constituent atoms using either X-rays or neutrons revealed that Co(~pCp), would contribute only 19% in an X-ray diffraction experiment whereas 44% of the scattering would be due to the CO(&~D~)~ in a analogous neutron diffraction experiment. Therefore neutron diffraction should be a more sensitive probe of the metallocene orientation. A major drawback of neutron diffraction studies on organic or organometallic compounds is the large incoherent scattering of hydrogen that leads to low quality diffraction patterns. This problem is overcome by using samples with a high level of deuterium enrichment, as incoherent scattering is much lower for deuterium.Thus time-of-flight neutron diffraction data were collected on an aligned ensemble of single crystals of SnS,{Co(T-C5D5)2}0,3 Least squares refinement of the occupancies of an initial model containing both the parallel and perpendicular extreme orientations refined to a parallel occupancy of 0.302(5) and a perpendicular occupancy of O.OOS(5). The overall agree- ment factor of 10.5% was achieved using a model containing a single parallel orientation for the Co(7-C,D5), which is in excellent agreement with the X-ray structure refinement. Figure 3 Van der Waals representation of the X-ray structure of SnS,{Co(?-CP),)o 3 1.3.3 Solid State ZHNMR Experiments on SnS2{Co(?7-C5D5)2>~.3~ In a solid state 2H NMR experiment the energetically most significant perturbation to the nuclear energy levels is the interaction of the nuclear quadrupole of the 2H (I = 1) nucleus with the external magnetic field. In the case of a deuteron in an axially symmetric environment a doublet is observed in the solid state spectrum whose separation depends on the orientation of the C-D bond with respect to the magnetic field. Thus solid state 2H NMR is ideally suited to the study of molecular orientation and dynamics in solids. The 2H NMR spectra of a single crystal of SnS,{Co(q-C5D5)2)0,33oriented with basal plane (crystallographic a-b plane) perpendicular and parallel to the static magnetic field are shown in Figure 4.The ,H NMRspectra are easily interpreted and immediately suggest the orientation of the cobaltocene molecules, without having to consider additional static disorder within the layers, or in-plane rotations of the cobaltocenes.In contrast to the powder lineshapes observed from a polycrystal- line sample, for an orientated single crystal we should observe only one sharp doublet per orientation. If the cobaltocene molecule adopts the perpendicular orientation, then only the x component of the electric field gradient (e.f.g.) tensor at the deuteron, eqSyx,has a non-zero projection along the field direc- tion; rotation about the CO-C~(,,,,,,,~~ axis will have no effect on eqxx. The spectrum obtained will therefore consist of a doublet with a splitting of ca.136 kHz. There are two possible low-energy motional processes postulated for the parallel orien- tation. Rotation about the CO-C~(,,,~~~~~) axis will cause an averaging of the z and y components of the e.f.g., eqLrand eq;,, and the resulting spectrum will reflect this averaging, giving a splitting of approximately 68kHz, as shown schematically in Figure 5. Additional in-plane (ab plane) rotation will cause no further averaging of the e.f.g. tensor. Only one doublet is observed with a splitting of 68 kHz when the crystal is orientated perpendicular to the magnetic field, this strongly suggests that the guest molecules adopt a single orientation in the host structure: the parallel orientation.3.4 Electrical Conductivity Experiments The temperature-dependence of the four-probe conductivity for single crystals of both the hosts and intercalates have been measured in the temperature range 2-300 K. Figure 6a presents a plot of log, Resistivity vs. Temperature for the entire series of Bol 1 1 _1 t ". 'I "' ' I +loo -100 hHZ Figure 4 The postulated motions for Co(+Z,D,), in the 2H-SnS, lattice in the perpendicular (a) and parallel (b) orientations and its predicted ,H NMR spectra when the static magnetic field is aligned perpendicular and parallel to the basal plane. host single crystals (phosphorus doped) SnS, -xSex, where x = 0.0,0.3,0.7, 1.3, 1.85, and 2.0. This plot dramatically shows the wide variation in conductivity behaviour in the host struc- tures as sulfur is replaced by selenium.A closer examination of the data presented in Figure 6 suggests that the host materials can be divided into two distinct classes on the basis of their resistivity variation with temperature. For class I (x = 0.0, 0.3, 0.7, and 1.3) the resistivity spans several orders of magnitude in the temperature range 10&300 K. For class 11 (x = 1.85 and 2.0) the resistivity decreases in the range 298 K to ca. 150 K then increases down to 2 K. The lowest resistivity corresponds to 140 and 160 K for x = 1.85 and x = 2.0 respectively. Materials in class I seem to be well described by an Arrhenius- type model for conductivity. Plots of log, Resistivity vs.T-for SnS,-,Se,, {x = 0.0, 0.3, 0.7, and 1.31) are linear; the observed activation energies (15,)decrease steadily from 0.45 eV (x = 0.0) to 0.09 eV (x = 1.3) across this series. The materials in class I1 can be modelled by considering them to be very low activation energy semiconductors (E, ca. lop2 eV), so that the carrier mobility term (p) becomes important compared to the carrier concentration (n) at elevated tempera- tures (T> 150 K). This would account for their metallic-like behaviour in the range 150-300 K. Previously, a similar temperature-dependence of the resistivity (p) for the organic I" " I "' ' I I"' 'I" " 1 +loo 0 -100 +loo 0 -100 kHz kHz Figure 5 ,H NMR room temperature spectra of a single crystal of SnS,{Co(7-C,D,),},,, ,oriented with the single crystal basal plane (a) perpendicular and (b) parallel to the static magnetic field.(The spike at 0 ppm is an artefact.) . --c Figure 6a Plot of log,, Resistivity vs. Temperature for the host single -2.00-0cn -SnSe2(Co(Cp)2)0.31 o2 {O < x d 2).crystals, SnS, -,S~,{CO(~-C~)~J~.~~ conductor (NMP),TCNQ, (NMP = N-methylphenazinium; TCNQ = 7,7',8,8'-tetracyano-p-quinodimethane} has been modelled using equation 1. * The temperature dependence of conductivity is considered to depend on the activation energy (E,) and the type of scattering mechanism in operation. Best fits for the resistivity versus temperature (100-300 K) data for x = 1.85 and x = 2.0 to the theoretical expression in equation 1 are given in Table 2.Table 2 summarizes the resistivity (p) at 298 K, the mobility parameter (a)and activation energy (E,) for the host crystals SnS, -uSe.r, (x = 0.0, 0.3, 0.7, 1.3, 1.85, and 2.0). The experimental data show a gradually decreasing activation energy in these host materials from SnS, (0.45 eV) to SnSe, (0.02 eV) via SnS,.,Se,., (0.09 eV). This is to be expected on the basis of Mott's impurity model for doped semicond~ctors,~~ which predicts that as the medium becomes more polarizable the energy required to ionize impurity electrons from the donor levels into the host conduction band will decrease. No work has been done to estimate the level of P doping throughout the series, 124 Table 2 Summary of important resistivity data for the hosts SnS,-,Se,;{O<x62} Hosts (x) 0.0 p (0cm, 298 K) 387.6 Ea (eV) 0.45 a - 0.3 7.67 0.37 - 0.7 2.87 0.28 - 1.3 1.21 0.09 - 1.85 0.1 14 0.025 1.70 2.0 0.054 0.020 1.72 but an estimate of 10, cm-3 for the carrier density in P-doped SnS, has been made.,, The conductivity behaviour of the so-called class I1 host materials can be interpreted in terms of the carrier mobility factor (p) in semiconductors with small activation energies (E, ca.lo-, eV). The model presented on the basis of equation 1 seems to work well above 100 K, but below this temperature there is some deviation -possibly arising from additional conduction via impurity sites. It has previously been demonstrated that in two-dimensional layered systems the carrier mobility (p) is highly temperature- dependent as in equation 2.,l This strong temperature-dependence above ca.100 K has been related to an optical phonon scattering mechanism. This is unique to two-dimensional layered materials, since the carriers are confined to individual XMX layers with mainly short-range interactions coupling the carriers to the optical modes of the lattice. These vibrational modes involve modulation of the XMX sandwich thickness in layered materials. In this study the exponent (a) in the mobility temperature- dependence expression in SnS,-,Sex crystals was found to be Q = 1.70, 1.72 for x = 1.85,2.0 respectively. This correlates well with theoretical predictions and other experimental data on layered systems,, suggesting that scattering of conduction electrons at T > 75 K may well be related to this mechanism in these particular SnS, -$ex hosts.In order to illustrate clearly the difference in electrical proper- ties of the intercalates at either end of the series, Figure 6b shows a plot of log,, resistivity vs. temperature variation for the entire intercalate single crystal series. As with the host materials it is convenient to divide the intercalate compounds into two distinct groups. Consider first the mainly sulfur-rich intercalate single crystals SnS,-,Se,{Co(q-Cp),),,,,, {x = 0.0, 0.3,0.7, and 1.3). For these samples a plot of log,, resistivity ~s.T-l/~is linear, and is in excellent agreement with the Mott variable-range hopping (VRH) law.Least squares fitting of the resistivity data to equation 322 gives o = 0.25 f0.02: (3) The values of p at 298 K, W, po, and Tofor each composition are given in Table 3. Notice that the room temperature resistivity increases for x = 0.3, 0.7, and 1.3 upon intercalation, whereas for x = 0.0 the opposite is true. Experiments have demonstrated that there is considerable anisotropy in the host single-crystal conductivity (pII/plcu. loo), whereas in the intercalate materials a much reduced anisotropy is observed (pII/phca. 10).However, a hopping mechanism in d-dimensions yields a T-(l/(l+d)expressi~n,~~so the conductivity data for the interca- lates strongly suggest an isotropic three-dimensional (T-1/4) hopping process rather than a two-dimensional (T-1/3) process.The experiments carried out to investigate the anisotropy of conduction in these intercalate systems suggest that the current CHEMICAL SOCIETY REVIEWS. 1992 Table 3 Summary of important resistivity data for the intercalates SnS, -,Se,{Co(q-Cp),},,33 ,,,, Intercalates (4 p(0cm,298K) po(xlO-lo) T,(x1O8K) 1/, T,(K) 0.0 3.90 1.54 3.67 4.0 -0.3 38.6 7.48 3.42 3.9 -0.7 69.2 9.77 3.60 3.9 -1.3 61.7 12.5 3.64 3.9 -1.85 1.1 x 10-2 ---5.7 2.0 1.1 x 10-2 ---6.1 carriers are not confined to a single layer to such an extent as in the host systems. The higher resistivity of some of these interca- lates relative to their host compounds (Table 3), despite substan- tial electron transfer to the Sn atoms, may depend on the limiting nature of the thermally activated hopping process rather than the presence of a large energy gap.The resistivity for the SnS, -,S~,{CO(~-C~)~}~,~~ ,,,, (x = 1.85 and 2.0) decreases with decreasing temperature as expected for metallic samples. Thus a clear semiconductor-to- metal transition appears to occur in the stoichiometry range 1.3 < x < 1.85. Remarkably, at 5.7 K (x = 1.85) the resistivity drops sharply (width 1.5 K). At 6.1 K(x = 2.0) a similar transi- tion (width 0.7 K) is observed with the resistivity falling to zero. This superconducting transition has been confirmed in the diselenide case by magnetic susceptibility measurements, which have demonstrated the Meissner effect below 6 K appropriate to a type I1 superconductor. At 4.2 K the critical fields H,, and H,, were 40 G and 700 G respectively. Below 3K extensive hysteresis of the magnetization is observed due to flux trapping within the samples.In the tin dichalcogenide intercalates the superconducting transition temperature (T,) increases as sulfur is replaced by selenium. This change can be understood in terms of the enhanced degree of electron charge transfer to the empty Sn(Ss, 5p) conduction band as suggested by XPES data (vide infra). This can be related to the increasing polarizability of the medium as selenium is added, since screening of the ionized electron from the [Co(q-Cp),] attractive potential becomes more effective. + This would probably lead to a greater value in N(EF)for the pure diselenide relative to the disulfide, which would be expected to lead to a decrease in the T, value as predicted by BCS theory.3.5 Photoelectron Spectroscopy Experiments Ultraviolet and X-ray photoelectron experiments were carried out on the host and intercalates in order to investigate the perturbations to the band structure and redox states of the host upon intercalation of cobaltocene. In order to carry out surface studies it is essential to have reproducibly clean surfaces. The preparation of clean, undisturbed crystal surfaces was achieved by cleavage of single crystals of both the host lattices and the intercalate samples in ultra-high vacuum (UHV) within the PES spectrometer. The resulting UV and X-ray photoelectron spec- tra of these materials were considered to be of high quality.In the X-ray photoelectron spectra of the hosts and interca- lates the main emission peaks due to Sn, S, and Se all remain essentially unchanged upon intercalation retaining similar bind- ing energies accompanied by a small increase in peak widths. However, for Sn(4d) emission a weak shoulder appears at lower binding energy to the main peak for all members of the series. By fitting Gaussian lineshapes to the Sn(4d) peak it was possible to calculate that the additional species has roughly 10% of the intensity of the main peak for the disulfide case, rising gradually through the series to roughly 12% of the main peak intensity for the diselenide case.The data are consistent with the formation of a reduced tin species as a result of electron transfer from the nineteen valence electron Co(q-Cp), guest molecule. The binding energy sepa- STRUCTURE, DYNAMICS, AND ELECTRONIC PROPERTIES OF COBALTOCENE IN SnS, -,Se,(O 6x 62)-D. O’HARE ration of the two species (LIE,= 1 eV) is constant as the selenium content changes. Similar binding energy shifts have been obtained for the intercalation of Cu into SnS,24 and Ag into SnSe,.25 It might be expected that a two-electron reduction of the tin site is occurring leading to the formation of a SnlI species. Consequently, two cobaltocene molecules are required to effect the complete reduction of a SnIV site.We have also investigated the cobalt emission peaks in the XPS spectra of the intercalate series. In all cases there appears to be at least two resolvable CO(~P,,~) emission peaks at ca. 780 and 782 eV. The ratio of these two peaks varies between 1:1.3 and 1:3.1 across the series. The binding energy separation (dEB= 1.8 eV) suggests that the two cobalt species are Coz + and Co3+. These two emission peaks have been assigned to Co(q-Cp), and [Co(q-Cp),] by comparison with the CO(~P,/,) + emission peaks from authentic samples of both Co(q-Cp), and [co(&p),] [PF,]. Such results were some of the first to suggest + that these classical redox intercalation reactions may not always proceed to complete electron-transfer from the guest to the host and that complex mixed valency species may be formed in the layers.Recent EPR studies on Cd,P,S,(Co(q-Cp),),.~ have also demonstrated the equilibrium of these neutral and ionized cobaltocenes between the layers of Cd2P,S,.26 The ultraviolet photoelectron (UPS) spectra of the host crystals show the valence band maximum shifting to lower binding energy as the selenium content increases, consistent with the decreasing band gap of the host. In each case there is a dramatic change in the UPS spectra upon intercalation. The spectra can be broadly interpreted as the summation of two sub- spectra due to the UPS spectra of host lattice and the cobalto- cene guests. The additional intensity in the band gap (lowest ionization band) stems from the transfer of electrons to the tin band.3.6 Impurity Band Model for the Electronic Structure A qualitative band model description would be useful in under- standing the process of electron transfer between the guest and the host. A Rigid Band Model approach would view the intercalate band structure as the sum of the guest and host valence bands (VB) together with the creation of a partially- filled conduction band (CB) by electron transfer. However, the observed changes near the Fermi level suggest that the process of electron transfer does not simply fill the empty Sn(Ss,5p) states in the conduction band to give a metallic system in all cases. The evidence indicates that there are strong electron localizing effects at play, especially in the sulfur-rich intercalates (x= 0.0, 0.3, 0.5),such that simple band theory is inapplicable.-Empty CB Sn 5sl5p +CoCp2 dn’ HOMO The impurity band model for heavily-doped semiconductors (e.g. P/Si) offers a useful approach to the understanding of the intercalate electronic structure. The overlap of the impurity orbitals is significant at high impurity concentrations resulting in the formation of an impurity band close to the conduction band of the semiconductor, as in Figure 7. However, metallic conductivity does not follow directly, since the localizing effects of the impurity potential may be significant. The Hubbard criterion states that the width of the impurity band ( W) must be greater than the electron repulsion (U), in order that a deloca- lized metallic system can form. Indeed, at a critical impurity concentration, the doped system may become metallic.The substitutional nature of the doping in P/Si is clearly distinct from the reaction that intercalates Co(q-Cp), into SnX, {X = S, Se} hosts, but the guest species can be considered to be an impurity sitting adjacent to the acceptor tin sites. The electron transfer is viewed as an overlap of empty Sn(5s,5p) states with the filled Co(q-Cp), dr* impurity states leading to the formation of an impurity band near the host conduction band. The transfer of an electron onto a SnIV site creates a strong electron-phonon interaction, such that a second trans- ferred electron gives a SnlI valency; SnlI1 is commonly observed to disproportionate.The Sn 5s2 states that constitute the impurity band form below the main empty conduction band states. The parent [Co(q-Cp),] + attractive potential adjacent to the reduced tin site tends to localize the electrons as well. Thus, these mixed valency materials may be pictured as having elec- trons hopping between the tin and cobaltocene sites. XPS is able to detect two tin oxidation states, the hopping being slow on a XPS timescale. For SnS,(Co(~-Cp),},,,, the effects of the lattice distortion at the tin site and the [Co(q-Cp),]+ impurity potential may be sufficient to localize the transferred electrons into an impurity band in the band gap. Conductivity studies have confirmed the semiconducting character of the disulfide intercalate.The impurity band width (W) depends on the impurity concentration (nd) and the width of the host conduction band. The electron repulsion energy (U) depends on the size of the impurity orbitals (Sn 5s2), which is directly related to the polarizability of the medium, i.e. the screening of the ionized electrons from [Co(q-Cp),] by the medium. Treating the + impurity orbitals as hydrogenic with radius aH, Mott deduced that the transition to the metallic state, as the electron repulsion effects within the impurity band are overcome, is achieved at nd’l3aH ca. 0.25.’’ As the sulfur is replaced by selenium in the intercalates the band width of the host conduction band increases and the polarizability of the medium increases, but the impurity con- I-, -Partially-filled CB B-E-1 Figure 7 Schematic band structure diagrams for SnS,{Co(r]-Cp),},,, and SnSe,{Co(r]-Cp)2}o,33. 126 centration remains constant (Sn:Co ca.3:1). Thus, the extent of Sn(545p) and Co(y-Cp), dn* overlap increases to give a shift of the impurity band, eventually giving an extended overlap with the host conduction band such that a transition to metallic behaviour at a critical selenium content (1.3 < x < 1.85) occurs. The electrons on the reduced tin sites may now be to some extent itinerant in the impurity band. As we have seen this is consistent with our conductivity measurements on SnSe,(Co(~-Cp),},~,, which confirm the metallic character of this material and reveal that it is a type I1 superconductor.4 Conclusions The intercalation of cobaltocene into single crystals of the layered tin dichalcogenides SnS, -xSex presented us with a unique opportunity to study in detail the structural and electro- nic changes taking place on intercalation of this highly electron- rich organometallic compound. The X-ray and neutron diffrac- tion analyses clearly show that the molecules adopt a highly ordered arrangement within the layers and surprisingly with the principal axis of the cobaltocene parallel to the basal planes of the lattice. The most remarkable observation is the dramatic change in the electrical conductivity across the series. The totally unexpected observation of superconductivity for SnS, -{Co(~-Cp),},,,, (1.85 6 x < 2) has given impetus to further work and demonstrated that there are still fascinating discover- ies to be made in this area of chemistry. Acknowledgments.The author is grateful to acknowledge sup- port from the Science and Engineering Research Council and the Nuffield Foundation. I also thank my co-workers (C. A. Formstone, E. T. FitzGerald, P. A. Cox, M. Kurmoo, K. Prout, H.-V. Wong, C. Grey, J. Hodby, J. S. 0.Evans, and S.J. Heyes) for their contributions to the work reported herein. 5 References 1 R. H. Friend and A. D. Yoffe, Adv. Phys., 1987,36, 1; ‘Physics and Chemistry of Materials with Layered Structure’, Vol. 4, ed. P. A. Lee, Reidel, Dordrecht, 1976. 2 H. P. B. Rimmington and A. A.Balchin, Phys. Status Solidi, 1971,6, K47. CHEMICAL SOCIETY REVIEWS, 1992 3 ‘Intercalation Chemistry’, ed. M. S. Whittingham and A. J. Jacob- son, Academic Press, New York, 1982. 4 F. R. Gamble, F. J. DiSalvo, R. A. Klemm, and T. H. Geballe, Science, 1970, 568. 5 D. C. Johnston, Solid State Commun., 1982,43, 533. 6 P. C. Klipstein and R. H. Friend, J. Phys. C., 1984, 17, 2713. 7 F. S. Ohuchi, W. Jaegermann, C. Pettenkofer, and B. A. Parkinson, Langmuir, 1989, 5, 439. 8 R. B. Somoano and A. Rembaum, Phys. Rev. Lett., 1971,27,402. 9 B. Bach and J. M. Thomas, J. Chem. Soc., Chem. Commun., 1972, 301. 10 M. S. Whittingham, Prog. Solid State Chem., 1978, 12, 41. 11 L. Benes, J. Votinsky, P. Lostak, J. Kalousova, J. Klikorka, Phys. Status Solidi, 1985, 89, K1; J.Votinsky, L. Benes, J. Kalousova, P. Lostak, and J. Klikorka, Chem. Papers, 1988, 42, 133. 12 D. O’Hare, J. S. 0. Evans, P. J. Wiseman, and K. Prout, Angeu. Chem., Int. Ed. Engl., 1991, 30, 1 156. 13 C. A. Formstone, E. T. FitzGerald, P. A. Cox, and D. O’Hare, Inorg Chem., 1990,29,3860. 14 C. A. Formstone, E. T. FitzGerald, P. A. Cox, D. O’Hare, and M. J. Kurmoo, J. Mater. Chem., 1991, 1, 51. 15 C. Grey, J. S. 0.Evans, D. O’Hare, and S. J. Heyes, J. Chem. Soc., Chem. Commun., 1991, 1380. 16 M. B. Dines, Science, 1975, 188, 1210. 17 R. P. Clement, W. B. Davies, K. A. Ford, M. L. H. Green, and A. J. Jacobson, Inorg. Chem., 1978, 17,2754. 18 J. S. Miller and A. J. Epstein, Angeu. Chem., Int. Ed. Engl., 1987,63, 287. 19 N. F. Mott and E. A. Davies, ‘Electronic Processes in Non- crystalline Materials’, Clarendon, Oxford, 2nd Edn. 1979; M. A. Kastner, R. J. Birgeneau, C. Y. Chen, Y.M. Chiang, D. R. Gabbe, H. P. Jenssen, T. Yunk, C. J. Peters, P. J. Picone, T. Thio, T. R. Thurston, and H. L. Tuller, Phys. Rev. B, 1988,37, 11 1. 20 B. Fotouhi, A. Katty, and R. Parsons, J. Electroanal. Chem., 1985, 183,303. 21 Y. Frongillo, M. Aubin, and S. Jandl, Can. J. Phys., 1985,63, 1405. 22 N. F. Mott and E. A. Davies, ‘Electronic Processes in Non- Crystalline Materials’, Clarendon, Oxford, 2nd Edn., 1979. 23 V. Ambegaokar, B. I. Halperin, and J. S. Langer, Phvs. Rev. B, 1971, 4, 2612. 24 F. S. Ohuchi, W. Jaegermann, and B. A. Parkinson, Surf. Sci., 1988, 194, L69. 25 C. Formstone and P. A. Cox, Unpublished work. 26 D. A. Cleary and A. H. Francis, J. Phys. Chem., 1985,89,97. 27 P. P. Edwards and M. J. Sienko, J. Am. Chem. Soc., 1981,103,2967.

ISSN:0306-0012    

DOI:10.1039/CS9922100121

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Surfactant Interactions with Biomembranes and Proteins M. N. Jones Department of Biochemistry and Molecular Biology, University of Manchester, Manchester M 139PT 1 Introduction Detergents are generally associated with cleaning processes from the mundane applications of dish and car washing to industrial laundering and the dispersion of oil slicks in the environment. Surface active agents (surfactants) are the main active constituents of detergent formulations. The large scale production of detergents has led to the release of surfactants into sewage treatment plants and ultimately given rise to their appearance in natural waters. Such pollution not only results in the build-up of unwanted foam at the aqueous-air interface but some surfactants present a potential toxic hazard to fish and other aquatic organisms.The realization that synthetic surfac- tants can be toxic, led to the development of straight-chain biodegradable surfactants and a general awareness of the need to control the contamination of natural waters by these mole- cules. Nowadays, the levels of surfactant contamination found in the environment are generally low and in the range 20-60 pg per litre’ although higher levels can occur in areas where raw sewage is released. Apart from the large scale use of detergents containing surfactants for cleaning purposes both natural surfactants such as the products of cholesterol metabolism [e.g. the bile acids and steroid glycosides (saponins) like digitonin and ouabain] and synthetic surfactants [e.g.sodium n-dodecyl sulfate (SDS) and n-octyl 8-D-ghcopyranoside (OBG)] have found widespread applications in biochemical research. In the biological sciences there are several major areas of research which make use of the interactions between surfactants and biological systems. These include the use of surfactants to solubiiize hydrophobic compo- nents of various tissues and cellular structures, particularly membrane components. The isolation of the anion and glucose transporters of the human erythrocyte and the reconstitution of receptors for insulin, acetylcholine,2 and the 8-adrenergic recep- tor are all carried out by processes requiring surfactants to breakdown the interactions between the protein and the mem- brane lipids. The resulting soluble protein-surfactant complexes can then be handled to separate and purify the transporter or receptor before reconstituting it into a model environment such as a vesicle or liposome where function can be studied in i~olation.~.~ One other important application of the use of a surfactant (SDS) is routine in most biochemical laboratories; this is polyac- rylamide gel electrophoresis (PAGE) in SDS, i.e.SDS-PAGE. The technique is used to determine the polypeptide composition of proteins and depends on the formation of SDS-polypeptide complexes, which, when subjected to an electrophoretic field in a Dr. M. N. Jones is a Reader in Physical Biochemistry in the Department of Biochemistry and Molecular Biologq, at the University of Manchester.He was trained as a chemist and his early research work was concerned with the physical properties of polymer solutions. During the late 1960she broadened his interests to the properties of association colloids and the forces acting in thin liquid films. These studies led to work on biological membranes and model membrane systems including liposomal dispersions. His current research is concerned with biochemical thermodynamics, the properties of liposomal dispersions and their use in drug delivery, and the characterization of biological macromolecules. polyacrylamide gel, separate according to their molecular weights. The resulting gel is stained for protein, usually with Coomassie blue dye. This reveals the number of bands and their position on the gel, thus giving the polypeptide composition of the sample, and the molecular weights of each band with reference to a calibration, using standard proteins of known molecular weight.The procedure is carried out with reduced SDS-saturated protein complexes, i.e. having all their disulfide bonds reduced to -SH groups, so that ideally the surfactant binds uniformly along the polypeptide chains and the charge per unit length is constant. Since the electrical force and the fric- tional force on the complexes both increase with polypeptide chain length, in the absence of the cross-linked gel matrix the complexes would migrate at the same speed, but the cross-linked nature of the gel ‘sieves’ them and the higher molecular weight complexes are retarded.The vast majority of molecular weights for protein subunits quoted in the biochemical literature have been obtained by this technique, although it is not without its pitfalls, particularly for proteins which are partially glycosylated. All surfactants consist of a hydrophobic residue(s) terminat- ing in a hydrophilic head group and can broadly be divided into anionic, cationic, or non-ionic depending on whether the head group is negatively, positively, or uncharged respectively. The essential feature of aqueous solutions of such amphipathic molecules is the formation of micelles above a critical concent- ration (the critical micelle concentration or cmc). Micelle forma- tion is, in general, a cooperative process and the cmc represents the limiting concentration of single (monomeric) molecules that can exist in solution.The magnitude of the cmc depends on the hydrophobic-hydrophilic balance of the monomer as well as on temperature and, for ionic surfactants, on the ionic strength of the solution. Increasing ionic strength usually depresses the cmc while for many surfactants the cmc often goes through a minimum as the temperature is raised from 0”C6It was first shown by Anson in 1939’ that many surfactants were potent denaturants ofhaemoglobin, but it was not until 1968 that it was established that surfactants such as SDS could form saturated complexes having approximately 1.4g of SDS per g protein (i.e. approximately one SDS molecule per two amino acid residues)* and that such binding involved interaction of the ‘monomeric’ surfactant with the pr~tein.~ This stimulated work on the precise nature of the interaction between proteins and amphipathic moleculeslo-lz which is the substance of this review.2 Interaction of Surfactants with Biomembranes The plasma membrane is the barrier between the cytoplasm of the cell and its environment and controls the entry of metabo- lites and other materials into and out of the cell. In mammalian cells this is the only barrier while bacteria and plant cells also have a cell wall to maintain structural integrity, although transport is still primarily controlled by the plasma membrane. The matrix of the plasma membrane of mammalian cells is a lipid bilayer in which are embedded receptors and transporter proteins required for cell function.These are the so-called ‘intrinsic proteins’ in contrast to ‘extrinsic proteins’ which are associated with the surfaces of the bilayers. Extrinsic proteins on the cytoplasmic side of the membrane bilayer form a rigid network or cytoskeleton which helps to maintain the integrity of the membrane and in some cases its shape, while the external 127 surface of the plasma membrane is coated with protruding oligosaccharide chains of intrinsic membrane glycoproteins forming a glycocalyx. Figure 1 shows the structure of the human erythrocyte membrane which is probably the most studied and most comprehensively understood of all cell membranes.The lipids forming membrane bilayers in cells are complex and varied. The major constituents of the bilayer are phospho- lipids, sphingolipids, glycolipids, and cholesterol; bacterial cell membranes contain no cholesterol. The lipids generally have two acyl chains although some phospholipids such as cardio- lipin have four. The acyl chain lengths are generally in the range C,, to CZ4and have varying degrees of unsaturation. The stability of the bilayer is critically dependent on the diacyl structure which enables the acyl chains of the lipids to pack in a lamellar form with limited head-group interaction. The bilayer is one form of smectic lyotropic mesophase. If one acyl chain is removed, e.g. by the action of a phospholipase (often present in venoms), head-group repulsion destabilizes the bilayer relative to the micellar structure.Like surfactants, membrane lipids have an amphipathic structure but the hydrophobic-hydrophilic balance is such that the molecules are much more hydrophobic and hence the concentration of monomeric species in equili- brium with the bilayer is vanishingly small (10-10-10-13M). The interactions between proteins and glycoproteins and the membrane lipids are largely hydrophobic. The transmembrane sequences of membrane receptors and transporters have a high proportion of hydrophobic amino acid residues and are fre- quently a-helical,13 so that the proteins are firmly anchored to the membrane but can still undergo lateral diffusion in the plane of the bilayer.It should be noted that the polypeptide chains of membrane transporters and receptors may cross the bilayer many times; the polypeptide chains of the anion transporter of the erythrocyte crosses the bilayer 14 times and the sugar transporter 12 times,14 while numerous receptors (e.g.PI and P2 adrenergic, MI and M2 muscarinic, and K-receptor) fit into a family having 7 transmembrane sections.' It is against this background that we must consider the interaction between surfactants and biomembranes. The sim- plest starting point is the isolated bilayer either in the form of a lamellar mesophase or a vesicle (liposome). Figure 2 shows the sequence of events that arise on exposing a bilayer to a typical surfactant such as SDS or the commonly used non-ionic Triton CHEMICAL SOCIETY REVIEWS, 1992 X-100 (polyethylene glycol,-, ,-p-t-octylphenol).The effective molar ratio of surfactant to bilayer lipid (&) in such a system is given by equation 1 .l6,l As & is increased the bilayer becomes saturated with surfactant (RY')after which it will be progressively disrupted, the phospho- lipid forming mixed micelles, until the system is completed solubilized(R%01)as mixed micelles with an increasing surfactant content, which will ultimately be in equilibrium with surfactant micelles. For surfactants with a low cmc Rewill be approxima- tely equal to the total surfactant to phospholipid molar ratio in the system when [S] >> [S]monomer. The process of bilayer disruption can be followed by using unilamellar vesicles encapsulating a radiolabelled or fluorescent solute and following its release as a function of surfactant concentration.Carboxyfluorescein is a useful tool in this respect in that at high concentration its .fluorescence is quenched. If carboxyfluorescein is encapsulated in vesicles at high concent- ration (i.e.in the quenched state) its release can be followed by the increase in fluorescence due to decrease in quenching. Figure 3 shows some data for Triton X-100 release of carboxyfluores- cien from egg phosphatidylcholine vesicles. The relationship between the cmc of a number of surfactants and the molar ratio of surfactant to phospholipid required to release 50% of encap-sulated carboxyfluorescein (R,,) and to produce 50% phospho-lipid solubilization (S,,) is shown by the data in Table 1.Both R,, and S,, increase with the cmc of the surfactant; there also is significant release of carboxyfluorescein below the cmc of the surfactant indicating that the surfactant monomer is incorpor- ated into the bilayer. The situation is considerably more complex when we come to consider surfactant interaction with a cell membrane. Figure 4 shows schematically the sequence of events on exposing a membrane to increasing amounts of surfactant. The mem- brane bilayer initially becomes saturated with surfactant, lyses and disrupts with concomitant solubilization as lipid-protein- surfactant complexes in equilibrium with mixed micelles. On further addition of surfactant the complexes lose lipid to mixed micelles, which at sufficient high surfactant concentration will be \I GlpphonnA Figure 1 Schematic model of the organization of proteins and glycoproteins in the human erythrocyte membrane.(Reproduced from reference 12, p. 3, by permission of Marcel Dekker Inc.) SURFACTANT INTERACTIONS WITH BIOMEMBRANES AND PROTEINS-M. N. JONES PL bilayer Surfactant/PL ratio 11+S (PL bilayer)S, saturated Resat 11+s (PL bilayer)S, + (Mixed micelle, PLS,,,) 11+s (Mixed micelle, PLS,) + (Surfactant micelles) R,so* Figure 2 Schematic representation of the sequence of events arising on exposure of a phospholipid (PL) to increasing amounts of surfactant(S).L4-k ' Ill 0 -5 -4 -3 -2 0 -5 -4 -3 -2 log [TRITON X -1001 (M) Figure 3 The release of encapsulated 6-carboxyfluorescein (6-CF) from phospholipid multilamellar liposomes by the non-ionic surfactant Triton X-100. (A) The decreases in fluorescence quenching on release of encapsulated 6-CF as a function of Triton X-100concentration. (B) % release of 6-CF and solubilization of phospholipid as a function of Triton X-100 concentration. The dotted line shows the procedure for defining Rso. (Reproduced from reference 18 by permission of Elsevier Science Publishers.) in equilibrium with surfactant micelles. The details of the membrane solubilization process will vary with the composition of the membrane but the final state, consisting of solubilized protein-surfactant complexes, demonstrates that the interac- tions between surfactant and proteins are in general stronger than between the proteins and membrane lipids.The detailed nature of protein-surfactant interactions can be readily studied with reference to proteins of known structure and for this it is appropriate to consider interactions between surfactants and soluble globular proteins. 3 Surfactant Interaction with Globular Proteins The interactions between surfactants and globular proteins have been studied using a wide range of physical methods, some of which are given in Table 2 together with the type of information a technique yields. Surfactants can be broadly divided into those which bind and initiate protein unfolding, i.e.denaturing surfac- tants, and those which only bind leaving the tertiary structure of the protein intact. Commonly used anionic surfactants, such as SDS and sodium n-dodecylsulfonate, generally denature pro-teins whereas non-ionic surfactants do not. For this reason the non-ionics are often preferred for membrane solubilization when enzyme, receptor, or transporter function is to be pre- served. There are, however, exceptions to these generalizations. Some proteins (glucose oxidase,20 bacterial catalase,2 papain Table 1 Surfactant release of carboxyfluorescein from phosphatidylcholine vesicles (From Riuz et al. 8, Surfactant cmc (mM)" Rsob &oc Triton X-100 0.24 0.35 1.7 SDS 1.33 1.2 2.5 Sodium cholate 3.oo 3.1 7.1 OBG 25.0 11.1 20.0 In 0.1M NaCI.b/c Molar ratio of surfactant to lipid required to release 50%0 of encapsulated carboxyfluorescein (R5,,)and to solubilize 50% (S5,,)of the lipid. The final phospholipid concentration was 1mM. Membrane ll+s Membrane S, Saturation 11+s Membrane S, Lysis 11+s Lipid-protein-S,,, + Lipid-S, Disruption and complex mixed micelle Solubilization 11+s Protein-S, + Lipid-S, + S, complex mixed micelle micelle Figure 4 Schematic representation of the sequence of events arising on exposure of a biomembrane to increasing amounts of surfactant (S). Table 2 Techniques used in the study of surfactant-globular protein interaction Technique Quantitative equilibrium dialysis Molecular sieve chromatography Titrimetry Calorimetry (microcalorimetry and titration calorimetry) Polyacrylamide gel electrophoresis Ultracentrifugation (sedimentation rate and equilibrium) Viscometry Static and dynamic light scattering UV difference spectroscopy Neutron scattering Enzyme kinetics Information obtained Binding isotherms, Gibbs energy of ligand binding Binding levels Proton binding in relation to surfactant binding Enthalpy of surfactant binding and protein unfolding Detection of specific complexes Sedimentation coefficients of protein-surfactant complexes, subunit dissociation and molecular weights Hydrodynamic volume and shape factors, protein unfolding Molecular weights, diffusion coefficients-complex dimensions Surfactant-induced conformational changes Structure of surfactant-protein complexesSurfactant-induced enzyme denaturation or activation I30 CHEMICAL SOCIETY REVIEWS, 1992 -+L +LP-PL, PLn Native Unfolded Figure 5 A schematic representation of the binding of surfactant ligands (L) to the native state of a protein P and subsequent unfolding process.(From reference 26.) and pepsin22 resist denaturation by SDS under some conditions and there are cases of enzyme activation by surfactants, e.g. Aspergillus niger catalase by SDS,23glucose-6-phosphatase by Triton X-100.24The bile salts sodium cholate and deoxycholate, although anionic, are non-denaturing and can activate some enzymes, e.g.phospholipase is activated by deoxycholate. The general pattern of interaction between surfactants and globular proteins is illustrated schematically in Figure 5. For anionic surfactants initial binding occurs to the cationic sites on the protein surface, specifically to the lysyl, histidyl, and arginyl amino acid side chains, whereas for non-ionic surfactants the binding sites will be hydrophobic patches on the protein surface and no further binding occurs after these are saturated. Anionics may, however, induce protein unfolding thus exposing many more hydrophobic binding sites previously buried in the core of the tertiary structure. The saturation of all potential binding sites is generally completed as the free surfactant concentration approaches the cmc.For reduced (no disulfide bonds) globular proteins at saturation, the protein binds between 1-2g of SDS per gram depending on the ionic strength of the solution. Figure 6 shows typical isotherms for SDS binding to lysozyme at low and high ionic strengths. The isotherms show the average number of surfactant molecules bound to the protein (t) as a function of the logarithm of the free surfactant concentration in equilibrium with the protein-surfactant complexes. At low free SDS concentrations the binding isotherms rise sharply as the cationic binding sites are saturated, after which binding increases more slowly before rising again as the free SDS concentration approaches the cmc. For low values of 8( < 18) the complexes precipitate but on further binding the solutions become only turbid and for t > 30 are optically clear.The formation of protein-surfactant complexes (PS,) can be represented as a series of equilibria: 0 6o r-II 50 40 -I> -30 20 -cmc cmc I I I I -5 -4 -3 log [SDS] Figure 6 Binding isotherms. determined by equilibrium dialysis, (V vs. log [SDS)) for the binding of sodium n-dodecylsulphate to lysozyme in solution at 25"C, pH 3.2; 0,ionic strength 0.0119 M; H, ionic strength 0.21 19 M. P+S SPS PS+S *PS, PS, + s +PS, Psi-1 + s +PS; PS,-1 + s= PS, The equilibrium constant for the i'th step is The average number (8) of surfactant molecules bound to a protein molecule is (4) and assuming that the intrinsic binding constants for each step in equation 2 are identical apart from a statistical factor gives equation 5 -nK[S]V=-I +K[S] This approach will not in general be valid since the binding of one surfactant molecule affects the binding of a second surfac- tant and so on.There are thus varying degrees of cooperativity between binding sites so that the K's are not the same. To account for this Hill2' proposed the following expression (equa- tion 6) -n,K[S]"HV=-1 + K[S]"" where nH is the cooperativity coefficient and K an intrinsic binding constant. If binding of ligands inhibits subsequent binding, nH < 1 and binding is negatively cooperative, if subse- quent binding is enhanced due to the ligands already bound nH > 1 and binding is positively cooperative.Figure 7 shows the way in which the shape of binding isotherms change with cooperativity for a hypothetical molecule with 50 binding sites " -a -7 -6 -5 -4 -3 -2 -1 log [ligand] Figure 7 Theoretical binding isotherms (C vs. log [ligand]) calculated from the Hill equation for a protein with 50 binding sites (intrinsic binding constant lo4)for a range of Hill coefficients (H.C.)from 0.5 to 7.5. (From reference 26.) SURFACTANT INTERACTIONS WITH BIOMEMBRANES AND PROTEINS-M. N. JONES 131 and an intrinsic binding constant of lo4. The isotherms become I--progressively steeper as nH increases and the binding process becomes more positively cooperative. An equation which is extensively used by biochemists is the Scatchard equation2 which follows from equation 6, when nH= 1, The Scatchard equation would apply if all the binding sites were identical and independent.Despite its shortcoming^^^ it is diagnostic of the type of cooperativity a system is di~playing.~~ Figure 8 shows the Scatchard plots for the binding isotherms given in Figure 7. For nH < 1 the Scatchard plots show negative curvature and for nH > 1 exhibit maxima. From equation 7 it follows that v/[S] vs. v extrapolates to give the total number of binding sites (n)when V/[S] -+ 0. When the Scatchard analysis is applied to protein-surfactant interactions examples of both negative and positive cooperativity are found for some systems. V V Figure 8 Theoretical Scatchard plots (B/[ligand] free vs.9) for the isotherms of Figure 7 for a protein with 50 binding sites (intrinsic binding constant lo4) for a range of Hill coefficients from 0.5 to 7.5. (From reference 26.) Figures 9 and 10 show Scatchard plots for the binding of SDS to bovinecatalase(R.M.M. = 245 000) in acid solutions. At pH 3.2 and 4.3 curves diagnostic of negative cooperativity are obtained which extrapolate to give values of n of 343 f6 and 333 f13 respectively, which are close to the number of cationic amino acid residues in the catalase molecule of 331 (1 12 lysyl, 86 histidyl, and 133 arginyl). At pH 6.4 a typical positively coopera- tive Scatchard plot is found corresponding to a Hill coefficient of 2.61 f 0.07.Table 3 presents data obtained by Scatchard analy- sis for other proteins where, like catalase, extrapolation gives values of the total number of specific binding sites very close to the number of cationic amino acid residues in the protein. The shape of the linear part of the Scatchard plots give values for the intrinsic binding constant and hence Gibbs energies of SDS binding (AG,). Also shown in Table 3 are the corresponding enthalpies of binding per mole of SDS (AH,) measured by microcalorimetry which combined with AGGgive TAS,. It can be seen that the enthalpies of binding are in general exothermic but small relative to TAS,. The large increases of entropy on binding are characteristic of a substantial hydrophobic contribution to Figure 9 Scatchard plots for sodium n-dodecylsulfate (SDS) on binding to bovine catalase at 25 "C: 0,pH 3.2;0,pH 4.3.(Reproduced from reference 24 by permission of the publishers, Butter- worth-Heinemann Ltd.) O 1 2 3 4 5 6 7 8 9 1011 12 1o-2 x v Figure 10 Scatchard plot for sodium n-dodecylsulfate (SDS) on binding to bovine catalase pH 6.4,25 "C. The solid line was fitted using the Hill equation for a total of 1190 binding sites (1.4g SDS per g catalase) giving an intrinsic binding constant 479 f 6 dm-3 mol-' and nH = 2.62f 0.07. (Reproduced from reference 35 by permission of the publishers, Butter- worth-Heinemann Ltd.) the binding process arising from the disordering of water molecules, concomitant with the partial removal of the alkyl chains of the surfactant from the aqueous environment. Thus initial binding of surfactants to proteins requires not only the ionic interaction of head groups with cationic sites but also binding of the alkyl chains to hydrophobic regions of the protein in the vicinity of the cationic sites.Confirmation of this comes from the observations that chemical modification of the cationic sites, such as acetylation of lysyl residues, shifts the Scatchard Table 3 Scatchard analysis of the binding of sodium n-dodecylsulfate to some globular proteins3' Protein (R.M.M.), pH No. of cationic residues n K(dm3mol-') AG, (kJmol-l) AH, (kJmol-l) TAS,,(kJmol-I) Ribonuclease A (13 682), 7 18 19 7.70~lo4 -27.9 -1.27 26.6 Ly sozyme (14 306), 3.2 18 18 3.93x 104 -26.2 -8.66 17.5 Ovalbumin (44000), 7.0 42 37 1.m105 -30.0 0 30.0 Glucose oxidase (147 000), 3.7 120 132 0.51 x lo4 -21.2 -3.29 17.9 Bovine catalase (245000), 3.2 331 343 9.53x 104 -28.4 -8.36 20.0 plots to give lower values of n, and reducing the alkyl chain length weakens binding.32 Despite these observations any Scat- chard analysis should be treated with a degree of caution, and it is not entirely clear why such a good correspondence between n and the number of cationic sites is obtained since the binding sites must only approximate to independence and are certainly not chemically identical.The pitfalls of this procedure can be seen from Figure 8a. The approximately linear portion of the curve for nH = 0.5 could be extrapolated to about 12 but such an extrapolation is meaningless for this model system.A more rigorous analysis of binding isotherms involves the CHEMICAL SOCIETY REVIEWS, 1992 Superficially these two proteins which are of very similar molecular mass and have the same number of cationic residues (18) and disulfide bonds (4)appear to bind SDS in a similar fashion with comparable Gibbs energies (dG,). However, the enthalpies of interaction are significantly different as shown in Figure 12. In particular lysozyme interacts exothermically throughout the range of V whereas for ribonuclease interaction at low 9 is endothermic and only becomes exothermic at high binding levels. They differ markedly in the ease of unfolding, the initial endothermic interaction seen for ribonuclease arises because the endothermic enthalpy associated with surfactant- inuse of the binding potential concept proposed by W~man~~ induced unfolding exceeds the exothermic enthalpy of surfac-which the binding potential n (P,T,p1,p2,...) at pressure P and temperature T is related to the binding (v) and chemical poten- tial of the ligand (p)as follows, v = (k) ap P.T The binding potential can be calculated by integration under the binding isotherm on the assumption that the chemical potential of the ligand is given by the ideal solution expression, thus B x = RT{Bdln[S] (9) 0 where R is the gas constant. Considering the formation of a specific complex (PS,), differentiation of equations 5 with respect to ln[S] followed by substitution into equation 9 and integration gives T = RTln(1 + K[S]”) (10) If it is assumed that for any given free surfactant concentration a complex PS, predominates then, TI = RTln(1 + KaPp[S]?) (1 1) from which it is possible to calculate an apparent binding constant (Kapp)at a given 1 and hence AG, from RTAC, = -In Kapp9 This procedure gives a profile of how AG, changes with 9.Figure 11 shows typical profiles for lysozyme and ribonuclease on interaction with SDS at pH 3.2 which demonstrates the initial ‘high energy’ binding at low V followed by progressively ‘lower energy’ binding as the proteins become saturated with surfactan t. 0-0 Ribonuclease pH 3.2 0-0 Lysozyme pH 3.2 -1 0 tant binding, the overall enthalpy only becoming exothermic at higher binding levels.The microcalorimetry thermograms for the SDS interaction with ribonuclease clearly show both exoth- ermic and endothermic component^.^^ In contrast lysozyme has a tertiary structure which is more resistant to unfolding by SDS and only at high binding levels is there evidence of a possible endothermic contribution to the overall exothermic enthalpy of binding. One of the most important lessons that can be learned from studies of protein-surfactant interactions is that although the overall nature of the interactions, specific-ionic interaction followed by non-specific hydrophobic interaction, is broadly similar for protein-anionic surfactant interactions the details of the process reflect the detailed tertiary structures of the proteins.It is important to note that surfactant denaturation of proteins occurs at surfactant concentrations which are far lower than those required for other commonly used denaturants such as urea (6-8 M) or guanidinium chloride (4-6 M) where the denaturation process depends primarily on the effect of the denaturant on the water structure and weakening of the hydro- phobic interactions in the tertiary structure of the proteins. In thermodynamic terms the Gibbs energies of surfactant binding (AG,) as saturation is approached are comparable to the Gibbs energies of micelle f~rmation.~ The protein-surfactant com-plexes saturate as the free surfactant in equilibrium with them approaches the cmc.Thus the complexes are more stable than micelles, the protein presenting a complementary amphipathic surface on which the surfactant can condense. 4 Models of Protein-Surfactant Complexes One of the difficult aspects of the study of protein-surfactant complexes is the determination of their structure. Yonath et studied the structure of lysozyme-SDS complexes by X-ray crystallography on cross-linked triclinic lysozyme crystals that had been soaked in 1.1 M SDS and then transferred to water or a lower concentration of SDS solution (0.35 M) to allow the protein to refold. It was necessary to use cross-linked crystals to prevent them dissolving on exposure to the high SDS concent- -3 nL”-401 1 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Number of ligands bound Number of ligands bound Figure 1I Gibbs energy per ligand bound (dG,) as a function of number Figure 12 Enthalpy of interaction of sodium n-dodecylsufate (SDS) of ligands bound (ij) for lysozyme (0)and ribonuclease (0)at pH 3.2, with lysozyme (0)and ribonuclease (0)as a function of the number ionic strength 0.0119 M, 25°C.of ligands bound at pH 3.2, ionic strength 0.01 19 M, 25°C. SURFACTANT INTERACTIONS WITH BIOMEMBRANES AND PROTEINS-M. N. JONES ration. Examination of the resulting ‘denatured-renatured’ crystals located three SDS molecules in the renatured structure. The agreement between the structure factors of the renatured lysozyme-SDS crystals and native cross-linked crystals was 17% (renatured in water) and 19% (renatured in 0.35 M SDS) and the minimum spacings in the X-ray pattern of renatured and native crystals were 2.9 and 1.1 8, respectively.Hence the conformation of the lysozyme in the renatured crystals was similar but not identical to that of native lysozyme. The need to cross-link the crystals detracts somewhat from the significance of the results although there were only a few cross-links and these were highly flexible. The essential common feature of the complexes was the location of the SDS molecules with the sulfate head group forming a salt bridge with positively charged amino residues and the hydrocarbon chain making hydrophobic contact with the tertiary structure, consistent with the pattern of binding discussed above.Of the three SDS molecules bound in the renatured crystal one formed a salt bridge with the terminal lysine with its alkyl chain penetrating deep into the hydrophobic core of the tertiary structure while the other two SDS molecules were bound to the protein surface, one being shared between two lysozyme mole- cules in the renatured crystal. The SDS molecule which pene- trated into the hydrophobic core could not be removed even after soaking in SDS-free water for an extended period; however its presence did not inhibit the formation of the tertiary struc- ture. Lysozyme has two domains separated by a cleft into which the natural substrate (cell wall polysaccharide) binds, and it is suggested that during renaturation the two domains fold separa- tely trapping the SDS molecule between them.During denatu- ration many more SDS molecules will be associated with the protein but most are lost when the renatured crystal is formed. The X-ray diffraction studies, while confirming the essential features of the binding process do not tell us about the structure of protein-surfactant complexes in solution when large numbers of surfactant molecules are bound. A variety of models have been proposed for the structure of SDS complexes with water- soluble proteins, usefully summarized by Ibel et ~21.~’as follows: (a) a model in which the protein organizes the SDS anions, into a ‘micelle complex’; (b) a model based on a ‘rod-like particle’ in which the protein forms the backbone of the complex with the SDS bound along the backbone, the particle having a length of 0.074 nm per amino acid residue; (c) a ‘pearl necklace’ model in which the flexible denatured polypeptide chain(s) of the protein has small spherical micelles clustered along it, the transmicellar regions of the polypeptide chain possibly forming a-helices; and (d) a ‘flexible helix’ model in which the SDS forms a flexible cylindrical micelle and the polypeptide chains are chemically wound around it.The stabilizing interaction proposed in this model is hydrogen-bonding between oxygens in the SDS head groups and nitrogens of the peptide bonds.38 The diversity of models reflect the conflicting ideas on the structure of the complexes and the difficulty of finding a technique which gives an unambiguous result.Some of the most Npis the number density of spheres of radius R, and g(r)is the pair-correlation function which is related to N(r) the number of individual scatterers within a sphere radius r. N(r) can be related to the fractal dimension (D)of the protein- surfactant complex by the relation N(r) = (Y/R)~.Since the distribution of micelles is dictated by the topology of the polypeptide backbone the fractal dimension is lower than 3, as it would be for freely diffusing micelles provided there were no intermicellar correlations. Thus the fractal dimension can be taken as an index of the topology of the surfactant-denatured protein. For increasing concentrations of dodecylsulfate (wt. “/o) the fractal dimension decreases from 2.3 YO), to 1.76 (3%) consistent with the a transition from a compact state to a more open random coil in which a string of constant-sized micelles are distributed along the hydrophobic patches of the denatured random coil.The ‘pearl necklace’ model for the BSA-dodecylsulfate com- plexes is different from the model which has been reported for the structure of the complexes formed between the deuterated bifunctional enzyme N-5’-phosphoribosylanthranilate/indole-3-glycerol-phosphate synthase (PRA-IGP) and SDS.3 This enzyme from Escherichia coli contains a single polypeptide chain of 452 amino acids (R.M.M. = 49 484) with no disulfide bonds. The deuterated molecule binds 1.26 g SDS per g protein (2 16 SDS molecules/452 amino acid residues).Neutron scattering was investigated from the whole molecule complex (W) and two SDS-complexed fragments produced by gentle hydrolysis with trypsin, a large fragment (L) containing 289 residues and a small fragment (S) containing 163 residues. Figure 13 shows the pair- distance distribution functions (PDDFs) of volume elements of the three SDS-complexed structures at vanishing contrast between the buffer-medium and the protein-surfactant phase. The small complex (S) gives a single peak corresponding to a single globular structure and a neutron scattering total dodecyl- chain volume (V,) of 26.0 f1 nm3 and 73 f3 C,,H,, chains. The large complex (L) gives two peaks which arise from two micelles associated with the C-and N-terminal ends of the polypeptide chain (Lc and LN)which are well separated.The first peak (the self peak) is due to interferences of pairs of volume elements within the same micelle and the second peak is due to 1000 750 mE-. h recent work has used the neutron-scattering techniq~e.~~,~~ While this method is capable of giving much information, the results have to be fitted to models and hence the conclusions drawn from the technique are always model-dependent. The complexes formed between lithium n-dodecylsulfate and bovine serum albumin (BSA) were studied by small angle neutron- ~cattering.~~BSA is a globular protein (R.M.M. = 67 000) with a single polypeptide chain and 17 disulfide linkages. The neutron-scattering data were interpreted in terms of a ‘pearl necklace’ model in which micelles of radius (R)equal to 1.8 nm (aggregation number 70 f20) were distributed along the polypeptide chain.The interparticle structure factor S(Q) which takes into account interparticle correlations is given by equation 13. S(Q)= I + Np[4~r2g(r)dr (13)0 Qr 500 a“ e w 250 0 RInm Figure 13 Pair-distance distribution functions (PDDFs) of volume elements situated in n-dodecylsulfate phases of SDS-(N-5‘-phosphori- bosylanthranilate isomerase/indole-3-glycerol-phosphatesynthase) complexes as observed by neutron scattering at vanishing contrast between the buffer and the protein-SDS phase. S (small fragment, R.M.M. = 17478; 163 amino acid residues.L (large fragment, R.M.M. = 32024; 289 amino acid residues. W (whole molecule, R.M.M. = 49484; 452 amino acid residues). (Reproduced by permission from reference 37.) interferences of pairs of volume elements in the micelles asso- ciated with Lc and LN; for these V, is 35.6 f1.5 nm3 (101 f4 SDS molecules) and 14.7 f0.8 nm3 (42 f2 SDS molecules) respectively. The whole molecule complex (W) gives two peaks and a shoulder. These are intepreted as the self peak (at low R), the central peak arising from interferences between pairs of volume elements situated in the core of a central micelle (W,) and the core of a micelle associated with either the C-terminal (W,) or N-terminal (W,) ends of the polypeptide chain, and finally the shoulder (at large R) arising from intereference between pairs of volume elements in Wc and WN.The number of SDS molecules in the three micelles WM, Wc, and WN are approximately 42, 101, and 73 respectively, giving the proposed structure shown in Figure 14 in which the polypeptide chain is wrapped around the three micelles to give what is described as a ‘protein-decorated micelle’ structure.In the structure it is assumed that the two interconnecting polypeptide segments, which may bind a small number of SDS molecules, are highly flexible as in the ‘pearl necklace’ model, and that the repulsive interaction between the micelles leads to an overall elongated conformation. 1 425 0 10 20 nm Figure 14 ‘Protein-decorated micelle’ model of the complexed formed between 2 16 sodium n’-dodecylsulfate molecules and the 452 amino acid residues of N-5‘-phosphoribosylanthranilateisomerase/indole-3-glycerol-phosphate synthase. The complex consists of three spherical micelles (grey areas) mostly of SDS alkyl chains with hydrophilic shells (black areas) occupied by polypeptide chains and sulfate head groups.(Reproduced by permission from reference 37.) There is clearly a considerable difference between the ‘pearl necklace’ and ‘decorated micelle’ models which may in part relate to the differences between the proteins, in particular the fact that BSA has a more restricted conformation because of disulfide linkages. The most important difference is that in the ‘pearl necklace’ model the polypeptide chain is believed to pass through micelles of constant size as opposed to around micelles of variable size in the ‘decorated micelle’ model.However, it is significant that for the decorated micelles the number of SDS molecules per amino acid residue are surprisingly uniform 0.45(S), 0.49(L), and 0.48(W). The structure of protein-non-ionic surfactant complexes is not in general complicated by protein unfolding and there seems little alternative but to envisage the binding of the surfactant molecules to hydrophobic patches on the protein surface. The binding of OBG to globular proteins has become a controversial issue in that no evidence for binding was found using molecular sieve chromatography40 whereas equilibrium dialysis gave bind- ing levels consistent with the formation of complexes in which the OBG adsorbed as a monolayer on the protein surface.For globular proteins covering a molecular weight range from 14000 to 350000,41 as the free OBG concentration approached the cmc, the binding levels increased with protein size and for many proteins the extent of binding calculated assuming the proteins are ellipsoidal in shape and coated with surfactant molecules agreed well with the experimentally measured binding levels at the cmc of the OBG. CHEMICAL SOCIETY REVIEWS. 1992 5 Molecular Dynamics of Protein-Surfactant Complexes A recent development in the study of protein-surfactant com-plexes is the application of the technique of molecular dyna- mic~.~~From a knowledge of the potential functions describing the molecular interactions in a protein, the force on each atom at some time t can be calculated.By use of Newton’s equations of motion it is then possible to calculate the acceleration of each atom and by iterative integration the velocity and position of each atom at time t + 6t, where 8t is of the order of 1 fs. By performing tens of thousands of such iterations the motion of the protein over a time period of 1&1000 ps can be followed which is sufficient to find the conformation of minimum energy. In principle such calculations should be carried out including a large number of water molecules but in practice the computatio- nal time required to include even a few tens of water molecules is often prohibitively long.The aqueous environment can be approximated to by using a radially dependent permittivity. Figure 15 shows a computer simulation of the proteins ribonuclease and lysozyme with ten bound SDS molecules at pH 3. The potential energies of binding are shown in Figure 16 and predict that SDS should have a greater affinity for lysozyme than for ribonuclease, particularly for the first two to three SDS molecules bound. This prediction is borne out by the experimen- tal values of dG, for very low values of 3 (Figure 1l), although at higher values of binding to ribonuclease is stronger (lower energy) than binding to lysozyme. Furthermore it is interesting that although these two proteins have structural similarities such as size and the number of disulfide linkages the structural disorganization caused by the binding of the surfactant ligands is considerably greater in the case of ribonuclease than it is for lysozyme. Figure 17 shows the native conformation of the two proteins on which has been superimposed the minimum energy conformation of the polypeptide chain after ten surfactant ligands have been bound.The change in conformation relative to the native state is clearly considerably larger for ribonuclease than for lysozyme as reflected in the root-mean-square (RMS) displacements of all the atomic positions as a function of binding (Figure 18). It is significant that the enthalpies of interaction of the two proteins with SDS (Figure 12) clearly reflect the differ- ences in the rigidity of their structures as suggested by the molecular dynamic calculations.The endothermicity of the initial interaction for ribonuclease, associated with a conforma- tional change, contrasts with the exothermicity of the interac- tion of SDS with the more rigid conformation of lysozyme. Although the problem of taking account of water molecules is not fully addressed in the molecular dynamic calculations the correspondence between the theoretical predictions and the experimental observations are sufficiently encouraging to sug- gest that this approach can be usefully applied to protein- surfactant interactions and should yield worthwhile results in the future. 6 References 1 M. A. Lewis and V.T. Wee, Environmental Toxicology and Chemistry, 1983,2, 105. 2 0. T. Jones, J. P. Earnest, and M. G. McNamee, in ‘Biological Membranes’, ed. J. B. C. Findlay and W. H. Evans, IRL Press, Oxford, 1987, Chapter 5, p. 139. 3 J. K. Nickson and M. N. Jones, Biochim. Biophys. Actu, 1982,690, 31. 4 M. N. Jones, J. E. More, and D. J. Riley, J. Receptor Res., 1986, 6, 361. 5 K. Weber and M. Osborn, J. Biol. Chem., 1969,244,4406. 6 R. J. Hunter, ‘Foundation of Colloid Science’, Volume 1, Clarendon Press, Oxford, 1987, Chapter 10, 564. 7 M. L. Anson, Science, 1939,90,256. 8 R. Pitt-Rivers and F. S. A. Impiombato, Biochem. J., 1968,109,825. 9 J. A. Reynolds and C.Tanford, Proc. Nutl. Acad. Sci., USA, 1970, 66, 1002. 10 J. Steinhardt and J.A. Reynolds, ‘Multiple Equilibrium in Proteins’, Academic Press, New York, 1969. SURFACTANT INTERACTIONS WITH BIOMEMBRANES AND PROTEINS-M. N. JONES .-@ Ribonuclease pH 3.2 -0 Figure 15 Molecular dynamic simulations of the structure of ribonuc- lease (a) and lysozyme (b) sodium n-dodecylsulfate (SDS) complexes with 10 surfactant ligands bound at pH 3. The hydrogens of the SDS alkyl chains are white; sulfur, yellow; oxygen, red; carbon, green; nitrogen blue. 11 M. N. Jones, 'Biological Interfaces', Elsevier, Amsterdam, 1975, Chapter 5, p. 101. 12 P. Agre and J. C. Parker, 'Red Blood Cell Membranes: Structure, Function, Chemical Implications', Marcel Dekker, New York, 1989, Chapter 1. 13 D. M. Engleman, T. A. Steitz, and A.Goldman, Ann. Rev. Biophys. Biophys. Chem., 1986,15,321. 14 D. J. Anstee, Vox Sang., 1990,58, 1. 15 Y. Masu, K. Nakayama, H. Tamaki, Y. Harada, M. Kuno, and S. Nakanishi, Nature, 1987,329, 836. 16 D. Lichtenberg, R. J. Robson, and E. A. Dennis, Biochim. Biophys. Acta, 1983,737, 285. 17 D. Lichtenberg, Biochim. Biophys. Acta, 1985,821,470. 18 J. Ruiz, F. M. Goni, and A. Alonso, Biochim. Biophys. Acta, 1988, 937, 127. 19 A. Helenius and K. Simons, Biochim. Biophys. Acta, 1975,425,29, F-l5'I o-oLysozymepH/w 3 -250L1 i i i 'I 'I" 0 12 3 45 6 7 8 91011' z Number of ligands bound Figure 16 Minimized potential energies of interaction of sodium n- dodecylsulfate (SDS) ligands (energy per ligand bound) with ribonuc- lease and lysozyme at pH 3.Figure 17 (a) Polypeptide chain conformation of native ribonuclease (pink) and the ribonuclease (SDS),, complex (green). (b) Polypeptide chain conformation of native lysozyme (blue) and lysozyme (SDS),, complex (green) at pH 3. 0, 1 I01 ’ ’ ’ I ’ I ’ I J 0 12 3 4 5 6 7 8 9101112 Number of ligands bound Figure 18 Root-mean-square displacements (RMS) of the polypeptide chains of ribonuclease and lysozyme sodium n-dodecylsulfate (SDS) complexes as a function of the number of SDS ligands bound at pH 3. 20 M. N. Jones, P. Manley, and A. E. Wilkinson, Biochem. J., 1982,203, 285. 21 M. N. Jones, P. Manley, P. J. W. Midgley, and A. E. Wilkinson, Biopolymers, 1982,21, 1435. 22 C. A. Nelson, J. Biol. Chem., 1971,246, 3895.23 M. N. Jones, A. Finn, A. Mosavi-Movahedi, and B. J. Waller, Biochim. Biophys. Acta, 1987,913, 395. 24 F. E. Beyhl, ZRCS Med. Sci., 1986, 14,417. CHEMICAL SOCIETY REVIEWS. 1992 25 M. Y.El-Sayert and M. F. Roberts, Biochim. Biophys. Acfa, 1985, 831, 133. 26 M. N. Jones and A. Brass, in ‘Food Polymers, Gels, and Colloids’, ed. E. Dickinson, Special Publication No. 82 The Royal Society of Chemistry, Cambridge, 1991, p. 65. 27 A. V. Hill, J. Physiol., 1910,40,40P. 28 G. Scatchard, Ann. N.Y.Acad. Sci., 1949,51, 660. 29 I. M. Klotz and D. L. Hunston, J. Biol. Chem., 1984,259, 10060. 30 G. Schwarz, Biophys. Struct. Mech., 1976, 2, 1. 31 M. N. Jones, in ‘Biochemical Thermodynamics’, ed. M. N. Jones, Elsevier, Amsterdam, 1988, Chapter 5, p. 228. 32 M. N. Jones and P. Manley, J. Chem. Soc., Faraday Trans. I, 1980, 76, 654. 33 J. Wyman, J. Mol. Biol., 1965, 11, 63 1. 34 M. I. Paz Andrade, E. Boitard, M. A. Saghal, P. Manley, M. N. Jones, and H. A. Skinner, J. Chem. SOC., Furaday Trans. I., 1981,77, 2939. 35 M. N. Jones and P. Manley, Int. J. Biol. Mucromol., 1982,4, 201. 36 A. Yonath, A. Podjarny, B. Honig, A. Sielecki, and W. Traub, Biochemistry, 1977, 16, 1418. 37 K. Ibel, R. P. May, K. Kirschner, H. Szadkowski, E. Mascher, and P. Lundahl, Eur. J. Biochem., 1990, 190, 31 1. 38 P. Lundahl, E. Greijer, M. Sandberg, S. Cardell, and K. D. Eriksson, Biochim. Biophys. Acta, 1986,873, 20. 39 S.-H. Chen and J. Teixeira, Phys. Rev. Lett., 1986, 57, 2583. 4o P. Lundahl, E. Mascher, K. Kameyama, and T. Takagi, J. Chrorna-togr., 1990, 518, 11 1. 41 J. Cordoba, M. D. Reboiras, and M. N. Jones, hit. J. Biol. Macromol., 1988, 10, 270. 42 J. A. McCammon and S. C. Harvey, ‘Dynamics of Proteins and Nucleic Acids’, Cambridge University Press, Cambridge, 1987.

ISSN:0306-0012    

DOI:10.1039/CS9922100127

出版商:RSC    

年代:1992

数据来源: RSC

摘要:

Modern Liquid Chromatography R. P. W. Scott Department of Chemistry, Birkbeck College, University of London, London WCl E 6BT, U.K. and Department of Chemistry, Georgetown University, Washington, DC 20057, U.S.A. 1 Introduction Liquid chromatography, (LC), was discovered nearly a century ago by a Russian botanist called Tswett who used a column packed with calcium carbonate to separate a number of plant pigments. The coloured bands produced on the adsorbent inspired the term chromatography to describe the separation process. Although colour has little to do with modern chromato- graphy, the name has persisted and, despite its irrelevance, is still used to describe all separation techniques that employ a mobile and stationary phase. Unfortunately, the work of Tswett was not immediately developed to any significant extent, partly due to the original paper being in Russian, and partly due to the condemnation of the method by Willstatter and Stoll.Willstat- ter and Stoll tried to repeat the work of Tswett but did not heed the advice of Tswett to avoid ‘aggressive’ adsorbents and so their experiments failed. Inaccurate and careless experimental work has always been a threat to scientific progress and recent years have shown that contemporary science is by no means immune to such threats. However, the mistake of Willstatter and Stoll was particularly unfortunate as, not only did it retard the development of a very useful separation technique, it also inhibited progress in many other fields of chemistry.It was not until thirty years had passed that Kuhn and his co- workers repeated Tswett’s original work successfully and separ- ated lutein and xanthine from a plant extract. Nevertheless, despite the success of Kuhn et a/. and the validation of Tswett’s experiments, progress continued to be slow and desultory. In 194I Martin and Synge introduced liquid-liquid chromatogra- phy by supporting the stationary phase, in this case water, on silica in the form of a packed bed and used it to separate some acetyl amino acids. In the same paper Martin and Synge suggested that it would be advantageous to replace the liquid mobile phase by a gas to improve the rate of transfer between the phases and thus improve the separation. The recommendation was not heeded and it was left to Martin and James to bring the concept to practical reality in the early fifties.Thus, Gas Chromatography (GC) was born and a new and important era of chromatographic development began. Gas chromatography grew from a laboratory novelty into a popular well established analytical technique in little more than Raymond P. W. Scott, D.Sc. (London), F.R.S.C., F.A.I.C., C. Chem. (U.K.), C. Chem. (U.S.A.) is a Consultant Chemist. He is Visiting Professor at the Department of Chemistry, George- town University, Washington D.C. and the Department of Chemistry, Birkheck College, University of London. He has been the recipient of the A.C.S. Chromatography Award (1977), the R.S.C. Award in Chemical Analysis and Instrumentation (19871, the U.S.S.R.Tswett Medal (1978), and the Martin Award (1982). Professor Scott has published more than 160 original papers in separation technology. He is the author of ‘Contempor-ary Liquid Chromatography’ (Wiley), ‘Liquid Chromatography Detectors’ (Elsevier), and ‘Liquid Chromatography Column Theory’ (Wiley), and edited ‘Gas Chromatography’ 1960 (But- terworths) and ‘Small Bore Columns’ (Wiley). His interests are primarily in separation science and analytical instrumentation. a decade. During that time, the foundations of chromatography column theory were laid down which were to guide the develop- ment of LC in the sixties and seventies. As a result of the glamourous success of GC, LC became the Cinderella of chro-matography and it was not until the major developments of GC were completed that scientists turned their attention to the development of LC.Today, LC is also a well established separation technique but its progress has been slow and arduous relative to that of GC. The difficulties encountered in the development of LC arose from two causes. Firstly, solutes have a very low diffusivity in liquids compared with a gas and thus, the kinetics of exchange between the phases are slow and special steps must be taken to achieve efficient separations. Secondly, very low concentrations of a solute in a liquid do not modify the properties of the liquid to the same extent that they do to a gas and this has made the development of LC detectors far more difficult.2 Classification of Chromatography The different forms of chromatography take their definitions from the physical nature of the mobile and stationary phases. Thus, in GC the mobile phase is a gas and in LC the mobile phase is a liquid. Furthermore, the stationary phase can be a liquid or a solid and thus there can be two forms of GC and LC; Gas Liquid Chromatography (GLC) and Gas Solid Chromatography (GSC) together with Liquid Liquid Chromatography (LLC) and Liquid Solid Chromatography (LSC). The classification of the different forms of chromatography are summarized in Table 1. There are some sub-classifications of chromatography based on the physical shape of the chromatographic system; for example column chromatography where the phase system is contained in tubular form and lamina chromatography where the phase system takes the form of a flat sheet or strip as in thin layer chromatography (TLC). The technique of Critical Fluid Chro- matography (CFC), where the mobile phase is a fluid operated above its critical temperature might also be considered a separ- ate class.However, if the mobile phase is in its critical state, it can, in fact, still be classed as a gas, albeit a dense gas, and consequently is another form of GLC or GSC. Table 1 The classification of chromatography ~ MOBILE PHASE STATIONARY PHASE I GAS LIQUID SOLID Gas chromatography Gas-liquid Gas-solid chromatography chromatography LIQUID Liquid Liquid-liquid Liquid-solidchromatography chromatography chromatography LC LLC LSC 137 3 The Separation Process Chromatography has been defined as: 'A separation process that is achieved by the distribution of substances between two phases, a stationary phase and a mobile phase.Those solutes distributed preferentially in the mobile phase will move more rapidly through the system than those distributed preferentially in the stationary phase. Thus, the solutes will elute in order of their increasing distribution coeffi- cients with respect to the stationary phase.' This definition is a little trite and, although it introduces the concept of a mobile and stationary phase which are essential characteristics of a chromatographic separation, it tends to obscure the basic process of retention in the term distribution. A solute is distributed between two phases as a result of the molecular forces that exist between the solute molecules and those of the two phases. The stronger the forces between the solute molecules and those of the stationary phase, the more the solute will be retained.Conversely, the stronger the interactions between the solute molecules and the mobile phase, the more rapidly will the solute pass through the column. Consequently, solute retention is controlled by molecular forces of which there are four basic types: ionic forces, polar forces, dispersive forces, and chemical forces. There could be considered a third type of molecular interaction, hydrogen bonding, but for the purpose of this discussion, forces due to hydrogen bonding will be classed as strong polar forces.3.1 Ionic Interactions Ionic interactions result from permanent electrical charges that exist on molecules when in the form of ions, for example organic acids, bases, and salts. Such interactions are exploited in ion exchange chromatography and it follows, that to retain anionic materials in a chromatographic system, the stationary phase should contain cations. Conversely, to retain cationic materials, the stationary phase should contain anions. The stationary phase can consist of an ion exchange resin or can take the form of an adsorbed ion exchanger on the surface of a reverse phase, such as an alkyl sulfonate. 3.2 Polar Interactions Polar interactions between molecules arise from permanent or induced dipoles and do not result from net charges as in the case of ionic interactions. Alcohols, ketones, and aldehydes are examples of polar substances having permanent dipoles, aroma- tic hydrocarbons such as benzene or toluene are examples of polarizable substances with no permanent dipoles.When a molecule carrying a permanent dipole approaches a polarizable molecule, the field from the permanent dipole induces a dipole in the polarizable molecule and thus, electrical interaction can occur. It follows, that selectively to retain a polar solute, the stationary phase must also be polar and contain, perhaps, hydroxyl groups. If the solutes to be separated are strongly polar, then a polarizable substance such as an aromatic hydro- carbon might be adequate as the stationary phase.However, to maintain strong polar interactions exclusively in the stationary phase (as opposed to the mobile phase) the mobile phase must be relatively non-polar or dispersive in nature. 3.3 Dispersive Interactions Dispersive interactions are more difficult to describe. Although electric in nature, they result from charge fluctuations rather than permanent electric charges on the molecule. Examples of purely dispersive interactions are the molecular forces that exist between hydrocarbon molecules. n-Heptane is not a gas due to the collective effect of all the dispersive interactions that hold the molecules together as a liquid. To retain solutes selectively, by dispersive interactions, the stationary phase must not contain polar or ionic substances but only hydrocarbon-type materials such as the reverse-bonded phases now so popular in LC.It CHEMICAL SOCIETY REVIEWS, 1992 follows, that to allow dispersive selectivity to dominate in the stationary phase, the mobile phase must be polar and signifi- cantly less dispersive. Hence the use of methanol-water and acetonitrile-water mixtures as mobile phases in reverse-phase chromatography systems. 3.4 Chemical Forces Chemical forces are normally irreversible in nature and thus the distribution coefficient of the solute with respect to the station- ary phase is infinite. Affinity chromatography is an example of the use of chemical forces in a separation process.The stationary phase is formed in such a manner that it will chemically react with one unique solute present in the sample and thus exclusively extract it from the other materials present. In any distribution system, it is rare that only one type of interaction is present and if this occurs, it will certainly be dispersive in nature. Polar interactions are always accompanied by dispersive interactions and ionic interactions will, in all probability, be accompanied by both polar and dispersive interactions. However, it is not merely the magnitude of the interacting forces between the solute and the stationary phase that will control the extent of retention, but also the amount of stationary phase present in the system and it's accessibility to the solutes.If the stationary phase is a porous solid, some solutes, for example those of small molecular size, can penetrate and interact with more stationary phase than larger molecules which are partially excluded. Under such circumstances the retention is at least partly controlled by size exclusion. This type of chroma- tography is called Size Exclusion Chromatography (SEC) but is still included within the classification of LSC or LLC. However, even in SEC, retention is not exclusively controlled by the size of the solute molecule, it is still partly controlled by molecular interactions between the solute and the two phases. 4 The Stationary Phase 4.1 Silica Gel The majority of stationary phases employed in LC are based on silica gel and, in fact, silica gel is probably the most important single substance involved in the technique.It is not only used as a stationary phase per se, but is also used as the matrix from which the so called 'bonded phases' are made. Silica gel has some unique properties that make it particularly useful in chromato- graphy and these properties arise from the way it is produced. It follows that the process used for the manufacture of silica gel is of sufficient importance that it merits some detailed attention. Silica gel is made in large batches by adding hydrochloric acid to a solution of sodium metasilicate. Initially, silicic acid is released but this quickly starts to condense with itself to form dimers, trimers, and eventually polymeric silicic acid.The polymer continues to grow until, at a particular size, the solution begins to gel. As a result of this process, primary particles of silica gel are formed which may have diameters ranging from a few Angstroms to many thousands of Angstroms. The size of the primary particles will depend, among other factors, on the temperature and pH of the mixture at the time of gelling. It is the formation of these primary particles that confers onto silica gel its high porosity and high surface area, which are so important in its use as a stationary phase or support in LC. Furthermore, it is the condensation of the surface silanol groups of the primary particles that causes their adhesion and the onset of gel forma- tion.After the gel has become solid, it is allowed to stand for a few days while condensation between the primary particles continues and the gel shrinks and exudes water. This process is called sinerisis and the firm gel that is finally produced is called the hydrogel. The hydrogel is then heated for a few hours at I20 "C and the resulting product is called the xerogelwhich, after grinding to an appropriate particle size, is the material employed as the stationary phase or support for LC and is called irregular MODERN LIQUID CHROMATOGRAPHY-R. P. W. SCOTT silica gel. Spherical silica gel is made from an emulsion of appropriate silane esters by the addition of hydrochloric acid followed by sinerisis and heat treatment.The particle size of the spherical silica gel is adjusted by controlling the dispersion of the ester in the emulsion. The physical properties of irregular silica gel are difficult to control during manufacture and thus, appro- priate material for LC is selected from the particular batch that provides the desired surface area and porosity. As the pore size is controlled by the size of the primary particles during the gelling process, the pores of the silica gel can also range from a few Angstroms to many thousands of Angstroms. The material can thus, act as an exclusion media, separating solutes on a basis of molecular size. If the mobile phase is chosen such that the interactions between the solutes and the two phases are closely similar, then the only effective mechanism for relative retention, will be size exclusion.The small molecules will enter all the pores and, therefore, will associate with all the stationary phase (or surface) and will be retained the most. The larger molecules will be excluded from some of the pores that are too small to enter and, therefore, associate with less stationary phase or surface and will be the least retained. However, even under conditions of normal solute retention, where there is a differential interaction of the solute between the two phases, the exclusion properties of the silica will still play a part. The restricted availability of the stationary phase to the larger molecules will reduce the extent of their interaction with it and their absolute retention will be proportionally reduced.For this reason, it is unwise to try to compare the chemical nature of two solutes from retention data obtained from a silica gel matrix. Such retention data would reflect true differences in molecular interaction only if both solutes were exposed to the same amount of stationary phase during elution. 4.2 Bonded Phases Bonded phases, and in particular the reversephases, are the most commonly used stationary phases in modern LC. The term reverse phase, though conventional, has no precise meaning. It originated in the work of Martin and Synge when they were investigating different phase systems. These authors replaced a system that employed water as the stationary phase and a hydrocarbon as the mobile phase by a hydrocarbon stationary phase and an aqueous mobile phase.This, was a form of phase reversal and, as the first stationary phase used was water, the alternative system employing a hydrocarbon stationary phase, was called the reverse phase. Today, it is a term given to stationary phases that are hydrocarbon in nature and, thus, retain solutes predominantly on the basis of dispersive interactions. Bonded phases are made from silica gel by reacting the surface hydroxyl groups with appropriate reagents to attach an organic group, or chain, to the silica. It is this organic chain that acts as the interacting moiety, retaining the solutes and producing the required separation. The reagents used are usually organic chlorsilanes or esters.There are basically three types of bonded phase; ‘brush’ type phases, ‘bulk’ phases, and oligomeric phases. Reacting silica, dried at about 150 “C,with dimethyloctylch- lorosilane contained in a solvent at elevated temperatures causes a dimethyloctylsilyl group to be attached to the surface by silicon-oxygen-silicon bonds. This is a fairly stable bonding sequence (far more stable than the carbon-oxygen-silicon bond which was the first chemical link used in bonded phase synthesis) but is not very stable at extremes of pH. The product is a surface covered with dimethyloctyl chains like bristles of a brush-hence the term ‘brush’ phase. If the silica surface is saturated with water and octyltrichlorsilane is used as the reagent, reaction occurs with both the hydroxyls of the silica surface and the adsorbed water, causing a crosslinking reaction and an octylsila- nyl polymer to be built up on the surface.Due to the polymeriza- tion process, the stationary phase has a multilayer character and, consequently, is termed a ‘bulk’ phase. Finally, ifmethyloc- tyldichlorsilane is used as the reagent in a sequence of synthetic steps, an oligomeric phase can be built up on the surface that is far more stable than either the ‘brush’ or ‘bulk’ phases. The silica is first reacted with the methyloctyldichlorsilane to link methyl- octylchlorosilyl groups to the surface. The bonded phase is then treated with water to generate methyloctylhydroxysilyl groups which, in turn, are then reacted with more methyloctyldichlor- silane attaching another methyloctylchlorosilyl group to the pre- vious group.This process can be repeated until eight or ten oligomers are linked to each other on the surface. The product is finally treated with trimethylchlorsilane to eliminate the last hydroxyl group. This final step is called capping. The oligomers are layered over the surface making the product extremely stable and exhibit almost no polar characteristics whatsoever. How- ever, due to the complexity of the synthesis, oligomeric phases are expensive to manufacture and, consequently, are not often used. The most popular reverse phase appears to be the brush type phase with paraffin chains having, four, eight, or eighteen carbon atom chains attached.These types of reverse phase have been termed, C-4, C-8, and C-18 respectively. The C-8 and C-18 phases are mainly used for solutes having relatively low molecu- lar weights whereas the C-4 phase is used for the separation of very large molecules. The C-4 reverse phase is also particularly useful in the separation of materials of biological origin that may be chemically labile or easily denatured. Using appropriate organic chlorsilanes, polar or polarizable groups such as nitriles or aromatic rings can be bonded to the silica to provide stationary phases covering a wide range of polarities. Bonded ion exchange materials have also been syn- thesized, although the most common types of ion exchange media are ion exchange resins in the form of tiny polymer beads.An interesting carbon stationary phase, introduced by Knox, is obtained by filling the pores of appropriately sized silica parti- cles with an organic polymer and carbonizing the product at elevated temperatures. The silica is removed from the product by treatment with strong alkali or hydrofluoric acid forming, perhaps, what might be termed, a true ‘reverse phase’. It would be a ‘reverse phase’ in the sense that the pores are where the primary particles of silica existed and the solid matrix now replaces the pores. The product is too active for use in chromato- graphy and so the carbon is graphitized by exposure to an argon plasma. This is a relatively new material and, due to the complexity of its manufacture, is expensive.Whether its perfor- mance relative to that of a conventional reverse phase merits the greater cost, remains to be established. 5 The Mobile Phase The stationary phase determines the character of the mobile phase that must to be used in order to achieve the required resolution. Unfortunately, the type of detector employed also imposes limitations on the choice of mobile phase. If, for example, a UV detector is used, then the mobile phase must also be transparent to the UV light of the operating wavelength. If silica gel, which is strongly polar, is employed as the stationary phase then a dispersive type solvent would be appro- priate for the mobile phase. Normal paraffins such as n-hexane or n-heptane would constitute the more dispersive types of mobile phase and would also be transparent to UV light. If the solutes were retained too strongly, the hydrocarbon could be mixed with methylene dichloride which is also transparent to UV light.Progressively more polar solvents could be mixed with the n-paraffin to increase the magnitude of the polar interactions in the mobile phase and elute the more strongly-absorbed solutes. Tetrahydrofuran, propanol, and methanol are examples of polar solvents that are transparent to the UV light and that can be mixed in small concentration with the n-paraffin to elute more strongly-retained solutes. Reverse phases, being dispersive in nature, require very polar mobile phases such as mixtures of water and methanol, acetoni- trile, or tetrahydrofuran.To increase the magnitude of the interactions in the mobile phase, and thus help elute strongly retained solutes, the proportion of the organic solvent in the mobile phase can be increased, relative to that of water. Very difficult separations may require subtle mixtures of all three solvents with water to optimize the separation. The need for such complex solvent mixtures, however, is rare in practice. Ion exchange resins can be, weak or strong, anion or cation exchangers and will require aqueous buffer solutions as mobile phases. The pH of the mobile phase must be adjusted to complement the type of stationary phase employed. In general, a suitable mobile phase can be chosen on a rational basis, provid- ing that the type of interactions that are occurring on the stationary phase are known and understood. Alternatively, the literature abounds with LC applications to help the novice choose an appropriate phase system for a specific separation problem. 6 The Liquid Chromatography Column The column is the heart of the liquid chromatograph and is where the separation takes place.It is usually a stainless steel tube a few centimetres long and a few millimetres wide, packed with particles of stationary phase a few microns in diameter. In the column the separation is completed and, despite the complex and perhaps glamorous appearance of the rest of the chromato- graph, the associated equipment is there merely to serve the column and help interpret the results.During the development of a chromatographic separation, two processes proceed pro- gressively and simultaneously in the column. Firstly, the bands of the individual solutes in the sample are moved apart as a result of their different interactions with the stationary phase. Secondly, as the bands are moved apart, they spread or disperse and tend to merge together, blurring the separation that has been obtained. The column, by appropriate design, must mini- mize this dispersion, so that, having been moved apart and separated, the individual solutes enter the detector as individual bands. Thus, to obtain maximum resolution, the column must move the bands as far apart as possible but, at the same time, keep each band as narrow as possible.The capacity of the column to restrain the dispersion of the solute bands is called the column efficiency. Column efficiency is measured in ‘theoretical plates’ and the greater the number of plates provided by the column the greater the separating capacity of the column. The major difference between modern LC columns and those of Tswett are that the former have much greater efficiencies. In order to design an efficient column the factors that control solute dispersion must be known and understood. A number of theories of band dispersion have been introduced and the one that is best supported by experimental results is that of Van Deemter et al. Van Deemter examined the different dispersion processes that could contribute to the total band variance per unit length of a column (H).The variance contribution of each process was then determined and the sum of all the variances provided a value for (H). The resulting equation would then show how (H) could be reduced and, consequently, the column made more efficient. Van Deemter et al. postulated that there were three basic dispersion processes that took place in a column the multipath eflect, longitudinal dzflusion, and resistance to mass transfer. 6.1 The Multipath Effect As the solute molecules pass through the interstices of the packing some molecules will, on a random basis, follow longer paths than others. As a consequence, some molecules pass ahead of the bulk of the molecules while others will lag behind.This results in band spreading and was given the term the ‘Multipath Effect’. Van Deemter et al. derived the following function for the variance contribution (a&,), from the multipath effect where (A) is a constant and (d,) the particle diameter of the packing. CHEMICAL SOCIETY REVIEWS, 1992 6.2 Longitudinal Diffusion Dispersion due to longitudinal diffusion is a band spreading process that results from the normal diffusion processes that occur in a liquid (mobile phase). Obviously, the longer the solute remains in the column, the longer this diffusion process continues and thus, the variance resulting from diffusion will be inversely proportional to the linear velocity of the mobile phase. Van Deemter et al.derived the following function for the variance contribution from longitudinal diffusion (06) where, (y) is a constant, (D,)is the diffusivity of the solute in the mobile phase and (u) is the linear mobile phase velocity. 6.3 The Resistance to Mass Transfer Band dispersion from resistance to mass transfer is the major factor contributing to band variance at high mobile phase velocities. The dispersion results from the finite time required for the solute to diffuse through the stationary phase in order to enter the mobile phase and, similarly, to diffuse through the mobile phase to return to the stationary phase. Solute molecules that have diffused further into the stationary phase, will take longer to diffuse back to the surface and enter the mobile phase, than those that were closer to the surface.While this transfer is taking place, those molecules that entered the mobile phase earlier will be have been swept further down the column by the moving phase, thus, ‘smearing’ the band along the column. Exactly the same type of band broadening occurs when the solute molecules pass from the mobile phase and enter the stationary phase. Consequently, Van Deemter et al. derived two functions to describe this type of dispersion, one for the variance contribution from the resistance to mass transfer in the mobile phase and another for the stationary phase. Obviously, the higher the mobile phase velocity the greater will be this type of dispersion. The functions derived by Van Deemter et al. are as follows: where (ah)is the variance contribution from the resistance to mass transfer in the mobile phase and f,(k‘) is a function of the distribution coefficient of the solute between the two phases and the stationary phase/mobile phase ratio.where (09 is the variance contribution from the resistance to mass transfer in the stationary phase and f,(k’) is another function of the distribution coefficient of the solute between the two phases and the stationary phase/mobile phase ratio. Summing the individual variances to obtain the total column variance per unit length (H) H= UkP + u&+ u& + u: 2AD,H=2Adp+-+-U fl(k‘)+ Dm + f,(k’)dfu-Ds or, BH= A +-+ (C, + C2)U U Now, in LC, D, -Ds,and dp>> df,thus, the resistance to mass transfer in the stationary phase is very small, i.e.C, >> C,, and H= A + -B + c,u U MODERN LIQUID CHROMATOGRAPHY-R. P. W. SCOTT rl I VARIANCE PER UNIT LENGTH Minimum Value of (H) MULTIPATH EFFECT LONG1TUDl NAL DIFFUSION Optimum -) LINEAR VELOCITY (u) Velocity Figure 1 Curve relating variance per unit length to linear velocity. Equation 2 is a hyperbolic function made up of a constant, a reciprocal function and a linear function. The relationship between (H)and the mobile phase linear velocity (u)is shown in Figure 1. It is seen that there is an optimum value of (u) that provides a minimum value for (H)which would correspond to a maximum column efficiency (n)where, n = I/H and (0is the column length.The optimum velocity, at which (H)will be a minimum can be determined by differentiating equation 2 and equating to zero when it is seen that, Furthermore, the minimum value of (H) can be obtained by inserting the expression for the optimum velocity in equation 2 Thus, knowing the functions for A, B, and C from the Van Deemter equation, the particle diameter of the packing and the diffusivity of the solute in the mobile phase, the theoretical column efficiency can be calculated. Conversely, the particle diameter can be calculated that will achieve a desired efficiency to achieve a given separation. Chromatography column theory has progressed to a level where the dimensions and operating conditions necessary to achieve a given analysis in the minimum time can be calculated.A discussion on the design of optimum LC columns, however, is outside the scope of this review. Examination of equation 1 indicates that the dominant factor that controls the minimum value of (H)is the particle diameter of the packing (dJ. It would appear to follow, that the maximum efficiency would be obtained from using the smallest possible particles. Unfortunately, it is not so simple, as reduction in the particle diameter also increases the column impedance to flow. Consequently, as the available column pressure is limited so, also, is the minimum particle diameter that can be used with a column of given length. Theory shows, that as a result of the dependence of both efficiency and flow rate on particle diameter, simple separations are best carried out on short wide columns packed with very small particles.Conversely, difficult sepa- rations require long narrow columns, packed with relatively large particles. The most common particle diameters that are commercially available for silica and bonded stationary phases are 3, 5, and 10 micron. A set of columns that would use these particle sizes to an advantage is given in Table 2. Liquid chromatography column theory is now well developed and columns that can provide very fast separations and very high resolution can now be designed and fabricated. An example Table 2 A practical set of columns employing readily available particle sizes Column Particle Column Column efficiency diameter length diameter 50 000 10micron 100 cm 1 mm 15000 5 micron 15 cm 2 mm 5000 3 micron 3 cm 3 mm of a very fast separation obtained from a relatively short column is shown in Figure 2.The separation was obtained employing a column 2.5 cm long, 3 mm in diameter packed with silica gel particles 3 pm in diameter. It is seen that the 5 components are separated in about 3.5 s. Separations as rapid as this are rarely required in general analysis, although they might be of use in following reaction kinetics. Nevertheless, the chromatogram shown in Figure 2 represents one of the fastest separations obtained with LC and also demonstrates the contribution that column theory has made to column design.0 1 2 3 4 Time(seconds) Figure 2 High speed isocratic separation of a five-component synthetic mixture. Packing, Hypersil 3 pm; column i.d., 0.26 cm; column length, 2.50 cm; mobile phase, 2.2% methyl acetate in n-pentane; linear velocity, 3.3 cmls. 1, p-xylene; 2, anisole; 3, nitrobenzene; 4,acetophenone; 5, dipropyl phthalate. Figure 3 shows the separation of a group of aromatic com- pounds extracted from coal and was obtained from a column 14 metres long, 1 mm in diameter packed with reverse phase particles 10pm in diameter. The column had an efficiency of over 500 000 theoretical plates and the elution time was over two and a half days. Unfortunately, very high efficiency columns must either be operated at exceedingly high pressures or a very long elution time must be tolerated.The inlet pressure used was 6000 p.s.i. and it was found that higher pressures were not possible due to leaks developing in the sample valve and excessive heat being generated in the column. The only alternative was to tolerate very long retention times. Such long elution times are rarely seen in normal LC analyses and would only be acceptable for very special samples. Nevertheless, the chromatogram shown in Figure 3 represents one of the highest column efficien- cies obtained with LC and again demonstrates the value of column theory in helping column design. The insert in Figure 3 is the first part of the chromatogram presented by the computer in an expanded form to illustrate the excellent resolution achieved.7 The Basic Liquid Chromatograph The basic liquid chromatograph is shown as a block diagram in Figure 4. The chromatograph consists of seven major components, a solvent supply module, an optional solvent programmer, a pump, the sample valve, a column oven, the detector, and the data acquisition, processing, and display system. 7.1 The Solvent Supply System The solvent supply system consists of a series of solvent reser- voirs, usually four in number, from which one or more solvents can be selected. They are constructed of stainless steel or glass and normally have a capacity of about one litre and some degassing facility such as a supply of helium gas to each reservoir. The degassing system is essential as, on mixing, some solvents evolve dissolved air which adversely effects the column performance and causes serious detector noise.7.2 The Solvent Programmer The solvent programmer has two functions which are usually programmed from a keyboard associated with the programmer. One function is to select a particular mixture of solvents to be used to develop a separation isocratically. The second is to arrange the composition of the mobile phase to change regu- larly, in a pre-defined manner, during the development of the separation. Many mixtures contain solutes covering a wide range of polarities. Consequently, if the separation is carried out isocratically, some solute will elute very rapidly and others will be eluted very late in the chromatogram. The separation time can be shortened by changing the mobile phase composition during development, accelerating the strongly retained solutes through the column with solvent mixtures of greater strength.The majority of samples can be satisfactorily separated by employing two solvents only, although a few may require ternary mixtures. It is very rare indeed that four solvents are necessary. Thus, most programmers are designed to program two or three solvents and, although a number of four solvent programmers are available, they would only be useful for very special samples. 7.3 The Solvent Pump Solvent pumps are usually piston operated and most contain two cylinders, operating alternately in parallel to reduce pres- sure pulses that can cause detector noise.All parts in contact with the solvent are made of stainless steel except for non-return valves, seats, and gaskets that may be made of sapphire or PTFE. Most pumps have a maximum pressures of 6000 or CHEMICAL SOCIETY REVIEWS, 1992 10000 p.s.i. Pumps can be made to operate at pressures above 10000 p.s.i. but it is not the pump that limits the maximum pressure that can be employed in the chromatographic system. One of the problems associated with operating columns at extremely high pressures is the heat that is generated, which can seriously effect column performance. Flow rate ranges for analytical purposes should extend from a few p1 per minute to about 10 ml per minute. 7.4 Sample Valves Sample valves can have internal or external loops and a range of sample volumes should be available from about 0.5 pl to 10 p1.For most purposes the valve can also be made of stainless steel but for biochemical separations the material may need to be bio-compatible. Many labile materials of biological origin easily degrade or denature in contact with heavy metals. For this reason sample valves are often made of titanium, or have titanium liners, as this material is bio-compatible. Sample valves can be made to withstand pressures up to 10000 p.s.i. but their lifetime at this pressure can be very limited. This is due to particulate contaminates in samples (which are almost imposs- ible to eliminate in practice) causing continual wear on the sealing surfaces of the valve.A sample valve is best operated at a maximum pressure of about 3000 p.s.i. to ensure a reasonably long life. 7.5 Column and Column Oven The column, as already stated can be of stainless steel, but for biochemical separations it, also, may need to be constructed from titanium. The column oven thermostats the column to ensure stability and helps dissipate the heat generated in the column. The retention time is linearly related to the distribution coefficient which, in turn, will depend on the operating tempera- ture. Consequently, for consistent retention data the column must be thermostated. Provision should also be made to bring the temperature of the mobile phase to that of the column before it enters the sample valve and column.As the mobile phase has a fairly high specific heat it is of little use to employ an air bath as a thermostat. A liquid thermostating medium is strongly recommended. 7.6 Detectors Books have been written solely on the subject of LC detectors so the treatment here must, of necessity, be somewhat cursory. An LC detector must respond to solute concentrations over the range of about to g/ml and have a linear response to concentration over at least three orders of magnitude. Prefera- MODERN LIQUID CHROMATOGRAPHY-R. P. W. SCOTT CHROMATOGRAM DISPLAY DATA ACQUISITION PROCESSING Figure 4 Block diagram of a liquid chromatograph. bly, the detector should respond to solutes present in the mobile phase but be insensitive to the mobile phase solvents themselves or changes in solvent composition.The detector should also be insensitive to changes in pressure, temperature, and flow rate. Unfortunately, an LC detector with all these attributes does not exist but some approach this performance. If a group of different detectors is available, then one can be chosen to have the attributes necessary for the particular sample in hand. There are four LC detectors in common use. The fixed and variable wavelength UV detector, the refractive index detector, the electrical conductivity detector, and the fluorescence detector. The fixed wavelength UV detector normally operates at 254 nm (the light emitted from a low pressure mercury lamp) although other wavelengths are available.It has a linear res- ponse of over three orders of magnitude extending from about 3 x to 1 x 10-g/ml, but this will vary somewhat with the extinction coefficient of the solute. If designed correctly, it is relatively insensitive to changes in temperature, pressure, and flow rate and thus, can be used satisfactorily with gradient elution. It is the most useful general detector available but is not suitable for detecting paraffins, aliphatic alcohols, sugars, etc. nor any substances that do not adsorb significantly at 254 nm. The variable wavelength detector normally employs a deuterium lamp that emits light from about 180 nm to about 400 nm and can take two forms: the dispersive detector and the diode array detector. The dispersive detector contains a monochrometer prior to the detector cell by which the wavelength required is selected.The flow can be stopped at any time and the contents of the cell scanned and a UV spectrum obtained if so desired. The diode array detector has a diffraction grating subsequent to the cell and disperses the transmitted light across a diode array, the outputs from which are stored in a memory bank. Thus, the contents of the detector cell is monitored simultaneously over a range of wave lengths and, at any time, a spectrum can be printed out. This detector also has a linear response of about three orders of magnitude extending from about 1 x lo-’ to 3 x lop4 g/ml but is slightly less sensitive than the fixed wave- length detector. In practice the variable wavelength detector is largely used to provide a choice of fixed wavelength detection using the specific wavelength that the solutes adsorb most strongly.With the exception of some aromatic compounds, UV spectra are not very informative relative to the IR or mass spectrum for structural elucidation. Consequently, the use of UV spectra obtained from variable wavelength detectors for solute identification is not very common. The refractive index detector is far less sensitive than the UV detector and has a linear dynamic range of less than three orders of magnitude from about 5 x lo-’ to about 2 x g/ml. It is very sensitive to temperature changes and thus should be thermostated. It is also sensitive to both changes in pressure and flow-rate and consequently is not suitable for use with gradient elution.Despite its disadvantages it is frequently used for detecting those solutes that do not adsorb in the UV and do not fluoresce. The electrical conductivity detector, as its name suggests, measures the conductivity of the mobile phase and is frequently used in ion exchange chromatography. It is relatively insensitive to flow rate and thus can be employed for gradient elution, providing the gradient does result in a continuous change in the conductivity of the mobile phase. It has a linear dynamic range of about three orders of magnitude from about 1 x low8to 1 x g/ml but this range varies a little from detector to detector. The fluorescence detector is the most sensitive detector in common use.It has a response ranging from about 1 x lop9to 1 x lop5 g/ml but, unfortunately, a linear dynamic range of only just over two orders of magnitude. Nevertheless, its high sensitivity and other features make it a very useful detector. The fluorescence detector is insensitive to temperature changes, pressure changes, and changes in flow rate and, consequently, can be employed with gradient elution. It also has the advantage of being a selective detector and thus can pick out a specific component of a mixture that fluoresces, from a host of other unresolved undetected peaks which do not fluoresce. 7.7 Data Acquisition, Processing, and Display The results obtained from a chromatographic separation, as depicted by the output from the detector, can be displayed at many levels of sophistication. The simplest way of displaying a chromatogram is to use a potentiometric recorder and many chromatographers still use this method of recording data.The heights of the peaks can be measured and used for quantitative analysis and the retention distance for peak identification. Most modern chromatographs, however, have A/D converters and the chromatogram is digitized and stored on disc by a computer. A simple program can then present a table of results that prints out the retention times and peak heights or peak areas and, if previously calibrated, a complete quantitative analysis. Very subtle algorithms can be included in the program to obtain accurate quantitative results from partially resolved peaks, or from small peaks eluted on the tails of large peaks.Unfortuna- tely, however clever the algorithm, it will always be a poor substitute for good chromatography. 8 Quantitative Analysis Quantitative analysis is carried out using peak heights or peak areas. Most manual analyses are carried out using peak heights, or the product of the peak height and peak width at half height. Peak heights, although simpler to use, are very dependent on a constant mobile phase flow rate and thus computer processed data usually employ peak area measurements. Under some circumstances, a normalization procedure can be employed and the percentage of any given component is expressed as the percentage area of the peak of the total area of all the peaks.This procedure is only valid if the detector has the same response to all solutes. This condition is rare, but can be used, for example, in the analysis of high molecular weight polymers when a refractive index detector is employed. Quantitative data are normally obtained using internaZ or external standards. If an internal standard is employed, the standard is added to the mixture and the ratio of the areas of the peaks of interest to that of the standard, corrected for their relative detector response, will give directly the percentage of each component. The response factors are obtained from calib- ration runs using known mixtures of the standard and the solutes of interest. If an external standard is used, it is run as a separate chromatogram and the areas of each peak in the sample chromatogram compared with that of the standard in the reference chromatogram. The external standard procedure is not as accurate as the internal standard, but eliminates the need to search for a substance to act as the internal standard that must be eluted in a position in the sample chromatogram where no other peaks occur.In fact, the external standard can be the same as the solute of interest thus eliminating the need for relative response factors. Experienced chromatographers, operating in a single laboratory, taking great care, can achieve a precision of about f 3%. However, between-laboratory trials have shown that, in many cases, a precision of f10% might be considered very satisfactory.High accuracy and precision is not easily obtained in LC analyses and the presentation of the results on a computer screen in glorious colour does nothing to improve the situation. The extra cost might be more usefully spent elsewhere in the apparatus. 9 Tandem Techniques Tandem techniques is the term originally given to combined instruments that included both a chromatograph and a spectro- meter. These twin systems were provoked by the development of high efficiency columns in GC which were applied to the separation of complex mixtures such as essential oils. Com- pounds of hitherto unknown structure were separated and it was immediately apparent that a method was needed to identify them.As a consequence, the mass spectrometer was directly associated with the gas chromatograph providing mass spectra of each component as it was eluted. The same motivation resulted from the development of high efficiency LC columns, to provoke the association of the liquid chromatograph with different types of spectrometer. Books have been written on this subject so the treatment of tandem techniques in this review must also be somewhat cursory in nature. The association of the liquid chromatograph with a spectrometer proved to be far more difficult than for a gas chromatograph. The first LC tandem technique was introduced by McLafferty et al. who passed a CHEMICAL SOCIETY REVIEWS, 1992 fraction of the eluent directly from the column into the mass spectrometer using solvent vapour as the chemical ionization agent.The result was crude, in the sense that the solute bands were badly dispersed in the interface between the two instru- ments, and much of the resolution was lost, but the feasibility of the system was demonstrated. A transport system was then introduced similar in principle to the transport LC detector. This consisted of a moving wire, or band, that passed through the LC column eluent, leaving a coating of solvent containing the solute on the wire. The solvent was evaporated and the residual solute, coated on the wire, passed through two vacuum locks into the mass spectrometer. Inside the mass spectrometer, and close to the ion source, the solute was thermally volatilized from the wire and the spectrum produced.This method had the advantage of producing electron impact spectra which are more useful in structural identification than chemical ionization spec- tra. The most common and sensitive method of sample introduc- tion for LC-MS today is the Thermospray interface. The eluent from the column is passed directly into the mass spectrometer through a tube, the end of which is heated, and the eluent is violently evaporated in spray form into the ion source. The solvent vapour is also used as the chemical ionizing agent. The Thermospray interface has proved very successful but expensive and a little cumbersome. The tandem technique LC-IR has also been developed, but has been found less useful than LC-MS.The system is less sensitive and IR spectra provide less information for structural identification than MS spectra. There is a problem with the column eluent as most solvents used in LC do not have a ‘transmission window’ over the wavelength range where the most useful data for structural information is obtained. There have been transport methods developed where the eluent is allowed to drip onto a KBr disc, the solvent evaporated and the disc then automatically placed in the light path of a IR spectro-meter. However, this procedure appears to have been developed more as an act of desperation than a serious practical technique. The most useful spectroscopic technique for structure elucida- tion is, without doubt, NMR.However, NMR is a relatively insensitive technique and it is difficult to obtain enough sample from a liquid chromatograph to provide useful spectra. On line LC-NMR has been attempted by employing a stainless steel, small bore column that projects into the field of the NMR magnet. Surprisingly, this does not seriously disturb the homo- geneity of the field and on-line spectra of good quality have been obtained. The liquid chromatograph has also been associated very successfully with the atomic absorption spectrometer and the combination has been used to follow heavy metal speciation in biological samples. The successful association of the Flame AA with the liquid chromatograph is a direct result of the high sensitivity of the spectroscopic technique and the simplicity of the interface.It must be said, however, that tandem techniques are bulky, usually difficult to operate, and very expensive. The best approach in LC to the structural identification of eluted solutes is still by an off-line procedure. The peaks of interest should be collected and submitted for spectroscopic examination as a separate experiment. An off-line, high resolution mass or NMR spectrum of an unknown solute will provide far greater struc- tural information than any tandem technique presently available. 10 The Present Status of LC LC development by the extrapolation of present knowledge is now proceeding very slowly and, indeed, the technique has nearly reached the stable condition of GC. Column theory has progressed to the point where the limits of resolution and speed of analysis, attainable from packed or capillary columns, is known and the possibilities and impossibilities of the technique recognized.Instrumentation continues to become more sophis- ticated, impressive, costly, and unfortunately sometimes irrele- vant, but the limits of column performance have not changed much in nearly a decade. This means that the separating MODERN LIQUID CHROMATOGRAPHY-R. P. W. SCOTT potential of LC has remained the same. There are, however, real needs in the biotechnology field to render the technique more amenable to samples of biological origin and in this area future developments may be exciting and rewarding. At any time, of course, a completely new concept could alter this relatively static condition dramatically, however, without such innovation, LC performance is likely to remain the same for some time to come. 11 Books Recommended for Further Reading 11.1 General LC ‘High Performance Liquid Chromatography’, ed. P. R. Brown and R. A. Hartwick, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, 1989. ‘Practical High Performance Liquid Chromatography’, ed. C. F. Simp-son, Heyden and Son Ltd., 1976. ‘Modern Practice of Liquid Chromatography’, ed. J. Kirkland, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, 1989. 11.2 Chromatography Theory R. P. W. Scott, ‘Liquid Chromatography Column Theory’, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, in press. A. S. Said, ‘Theory and Mathematics of Chromatography’, Huthig, Heidelberg, Basel, New York, 1981. J. C. Berridge, ‘Techniques for the Automated Optimization of HPLC Separations, 1985, John Wiley and Sons, New York, Chichester, Bris- bane, Toronto, Singapore, 1985. 11.3 Quantitative Analysis ‘Quantitative Analysis in Chromatographic Techniques’, ed. E. Katz, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, 1987. 11.4 Liquid Chromatography Detectors R. P. W. Scott, ‘Liquid Chromatography Detectors’, Elsevier, Amster- dam, Oxford, New York, Tokyo, 1986. T. M. Vickrey, ‘Liquid Chromatography Detectors’, Marcel Dekker, Inc., New York and Basel, 1983. 11.5 Applications J. F. Lawrence, ‘Liquid Chromatography in Environmental Analysis’, Himana Press, Clifton, New Jersey, 1984. K. Gooding and F. Regnier, ‘HPLC of Biological Molecules’, Marcel Dekker, Inc., New York and Basel, 1990.

ISSN:0306-0012    

DOI:10.1039/CS9922100137

出版商:RSC    

年代:1992

数据来源: RSC