摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman)Professor M. J. Blandamer Dr. A. R. Butler Dr. E. C. Constable 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 Ilr. J. M. Brown Ilr. J. BurgessIIr. N. Cape I'rofessor A. Hamnett Ilr. T. M. HerringtonI'rofessor R. Hillman I'rofessor R. Keese Ilr. T. H. Lilley I3.H. Maskill I3ofessor Dr. A. de Meijere I'rofessor J. N. Miller I'rofessor S. M. Roberts I'rofessor B. H. Robinson Ilr. A. J. Stace Staff Editors Mr. K. J. Wilkinson Dr. J. A. Rhodes Dr. M. Sugden University of Sussex University of Leicester University of St.Andrews University of Cambridge 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 Leicester University of Bern University of Sheffield University of Newcastle upon Tyne University of Gottingen Loughborough University of Technology University of Exeter University of East AngliaUniversity of Sussex Royal Society of Chemistry, Cambridge 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 which present a truly international outlook on the major advances in a. wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca. 30 references), but should act as a springboard to further reading. In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context. Review articles must be short, around 6-8 journal pages in extent.In consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Instruction to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis, 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, 1993 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/CS99322FX009

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ISSN 0306-001 2 CSRVBR 22(3) 143-21 2 (1 993) Chemical Society Reviews Volume 22 Issue 3 Pages 143-212 June I993 CENTENARY LECTURE. The Pursuit of Selectivity in Radical Reactions By Athelstan L. J. Beckwith (pp. 1 43-1 51 ) The recognition of the factors which influence the chemo-, regio-, and stereo-selectivity of organic free radical reactions has underpinned recent developments in the application of such reactions to the synthesis of complex molecules. In this review the role of thermochemical, stereoelectronic, steric, and polar effects in determining selectivity will be discussed and illustrated with examples drawn mainly from recent work carried out at the Australian National University. The Nature of Ammonium and Methylammonium Halides in the Vapour Phase: Hydrogen Bonding versus Proton Transfer By A.C.Legon (pp. 153-1 63) The familiar white smoke, so redolent of school chemistry, is the ultimate product when ammonia and hydrogen chloride gases mix but what is the nature of the heterodimer first formed? Is it the simple hydrogen-bonded species H,N-..HCl or is the proton transferred to give the ion pair H,NH+ 0.-C1-? How is the extent of proton transfer affected by progressive methylation of NH,, by replacement of C1 with Br and then I, and by substitution of N by P? Recent advances in the rotational spectroscopy of supersonically expanded jets have led to answers to these fundamental questions. Discovery and Development of Anthracycline Antitumour Antibiotics By J. William Lown (pp.165-1 76) The armamentarium of the oncologist includes 40-50 clinically useful chemical agents. The paradigm of cytotoxic anticancer agents is doxorubicin, an anthracycline, which is amongst the most widely prescribed and effective of anticancer agents. This review attempts to summarize the discovery and development of the anthracyclines, their classification on the basis of structure and mechanism, and evidence on their several modes of action. Recent identification of cellular targets including topoisomerases and helicases present new challenges to the synthetic chemist and pharmacologist. Computer Simulations on Aqueous Solutions of Some Non-Electrolytes By Koichiro Nakanishi (pp. 1 77-1 82) The structure of aqueous solutions of non-electrolyte amphiphiles is often of complex nature and methods to investigate the solution structure experimentally are limited.Recent development of molecular simulation (Monte Carlo and molecular dynamics calculations) has made it possible to examine the structure and dynamics at the molecular level. Three representative cases, 2-methyl-propan-2-01, urea, and acetonitrile, are described. Biosynthetic Incorporation of Non-natural Amino Acids into Proteins By Josef Brunner (pp. 1 83-1 89) This article summarizes recently developed methods of incorporating non-natural amino acids into proteins. These methods rely on the chemical aminoacylation of artificial [(semi)synthetic or in vitro transcribed]transfer RNAs that recognize a unique ‘blank’ codon in appropriately engineered messenger RNAs.By introducing ‘designer’ amino acids, it is possible to generate novel probes of protein structure and function. Structural Systematics in Molecular Inorganic Chemistry By A. Guy Orpen (pp. 191-1 97) Applications of crystal structural data to molecular inorganic chemistry are described. The use of empirical correlations between structural parameters to provide insights into transition metal chemistry is illustrated. Examples are given of applications of this approach to determining typical molecular dimensions, testing theories of bonding, exploring conformational behaviour of flexible species, and analysing reaction pathways. The relationship between the experimental observations and the theory used in their interpretation is discussed. Some Recent Synthetic Routes to Thioketones and Thioaldehydes By William M.McGregor and David C. Sherrington (pp. 1 99-204) Thiocarbonyl compounds are reactive intermediates used to introduce sulfur heteroatoms into organic syntheses. Such species can either be synthesized and then further modified, or generated and reacted in situ. The crucial synthetic step in such syntheses is the generation of the inherently unstable C-S r-bond, thus creating a reactive ‘handle’. This review explores the plethora of methods for the generation of both transient and stable thioaldehydes and thioketones. Electrolytes in Binary Solvents: An Experimental Approach By S. Taniewska-Osihska (pp. 205-21 2) The lack of an adequate theory to describe the properties of electrolytic solutions in binary mixed solvents makes research on such systems important.This article collates the standard enthalpies of salt solutions in water-organic solvents with the properties of solvent component mixtures, in order to discuss the effect of ions on solvent structure. Plots of AH: for electrolytes with and without organic ions are compared in several groups of mixed solvents. The AH: maxima for organic ions are described by means of hydrophobic effects parameters of organic co-solvents. The role of hydrophobic interactions is shown by the comparison of AH: plots in water-organic and organic-organic solvents. Articles that will appear in forthcoming issues include Cholaphanes et af.;Steroids as Structural Components in Molecular Engineering A.P. Davis The Physiological Role of Nitric Oxide A. R. Butler and D. L. H. Williams The Lower Oxidation States of Indium D. G. Tuck Interplay of Theory and Experiment in the Determination of Transition-state Structure I. H. Williams Catalytic Antibodies: Mechanistic and Practical Considerations J. D. Stewart and S. J. Benkovic Interactions of Metal Ions with Nucleotides and Nucleic Acids and their Constituents H. Sigel Catalysis by Metal Ions in Reactions of Crown Ether Substrates R. Cacciapaglia and L. Mandolini Thermodynamics of Solvation in Mixed Solvents W. E. Waghorne The Chemistry of Cyclopropylmethyl and Related Radicals D. C. Nonhebel Electrochemistry in Media of Low Dielectric Constant A.Abbott Mechanisms of Solvolytic Alkene-forming Elimination Reactions A. Thibblin Chemical Society Reviews (ISSN 0306-0012) is published bi-monthly by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge, CB4 4WF, England. All orders accompanied with payment should be sent directly to The Royal Society of Chemistry,Turpin Distribution Services Ltd., Blackhorse Road, Letchworth, Herts., SG6 IHN, U.K. NB Turpin Distribution Services Ltd., distributors, is wholly owned by The Royal Society of Chemistry. 1993 annual subscription rate E.C. E90.00, U.S.A. $198.00, Canada &104.00+ GST, Rest of World f99.00. Customers should make payments by cheque in sterling payable on a U.K. clearing bank or in U.S. dollars payable on a U.S. clearing bank. Second class postage is paid at Jamaica, N.Y. 11431. Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. All other despatches outside the U.K. by Bulk Airmail within Europe and Accelerated Surface Post outside Europe. PRINTED IN THE U.K. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E30.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.

ISSN:0306-0012    

DOI:10.1039/CS99322FP011

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

CENTENARY LECTURE. The Pursuit of Selectivity in Radical Reactions Athelstan L. J. Beckwith Research School of Chemistry, Australian National University, Canberra, Australia, 260I 1 Introduction Moses Gomberg’s paper, ‘An Instance of Trivalent Carbon: Triphenylmethyl” appeared in 1900 in the December 5th issue of the Journal of the American Chemical Society. It was written in the first person in a delightful idiosyncratic style typified by the concluding cautionary note: ‘This work will be continued and I wish to reserve the field to myself. Gomberg need not have been concerned. Although gas phase radical processes received con- siderable attention during the next three decades, the possibility that some synthetically useful organic reactions in solution might involve free radical intermediates was not seriously con- sidered until W.A. Waters in Durham, D. H. Hey in London, and M. S. Kharasch in Chicago commenced the mechanistic studies that contributed so profoundly to the development of organic free radical chemistry. The English groups concentrated on the homolytic substitution of aromatic systems. This was understandable in the light of Ingold’s success in developing mechanistic theory through his scrutiny of electrophilic aroma- tic substitution, but unfortunate with the wisdom of hindsight, since free-radical reactions are now seen as being particularly useful for the construction of aliphatic and alicyclic systems. Kharasch, however, worked mainly on the free radical chemistry of relatively simple aliphatic compounds.He was so successful that by the time of his death he had defined and explored virtually all of the elementary mechanistic pathways available to free radicals, namely (i) coupling and its reverse reaction, homolysis (equation 1); (ii) homolytic substitution or SH2(equation 2);(iii) addition and its reverse reaction, 8-fission (equation 3); and (iv) electron transfer (equation 4). Athel Beckwith is a professor of chemistry at the Research School of Chemistry, Australian National University (ANU) in Can- berra. An honoursgraduate of the University of Western Australia (1953),he carried out his graduate research with the late W. A. Waters at Oxford University (D. Phil., 1956). After a short period with the CSIRO in Melbourne, he joined the Universityd- Adelaide in 1958,first as lecturer and later as Professor and Head of the Organic Chemistry Department, a position he held until he moved to the ANU in 1981.A frequent visiting lecturer in Europe and North America, he has spent extended periods at Imperial College London (1962), the University of York (1968), and Oxjord University (1974 and 1979). He was elected as Fellow of the Australian Academy of Science in 1974 (Vice-president) and of the Royal Society in 1989. Awards received include the Rennie, H. G. Smith, and Organic Chemistry Medal of the Royal Austra- lian Chemical Institute. Professor Beckwith’s re-search lies in the general area @-physical organic chemistry- and has been mainly concerned with reactive intermediates.It covers structural, kinetic, and mechanistic investigations in- volving ESR spectroscopy and laserflashphotolysis, the deve- lopment of theoretical ap-proaches to radical reactivity, the role of radical interme- diates in biological processes, and the application of radical reactions to synthesis. A’+B’ +A-B (1) A’ + B-DSA-B + D’ (2) A’ + B=D*A-B-D’ (3) A‘ + e+A-; A‘ -e*A+ etc. (4) Most free radical processes can be rationalized in terms of these elementary steps, or simple variations of them. Since Kharasch’s time only one completely new type of radical process, namely pericyclic reactions of radicals2 and radical- ions,3 has been added to this list. However, it is one thing to be able to recognize the types of reaction available to a reactive intermediate.It is another matter to be able to predict which pathway will be followed or, for complex substrates, what will be the regiochemistry and stereo- chemistry. It was this lack of predictability and selectivity that so delayed the realization of the synthetic potential of free radical chemistry. Long after Kharasch had defined their basic mecha- nisms, free radical reactions continued to be regarded by most organic chemists as erratic, capricious, and prone to give intractable mixtures; in short, as unsuitable for the efficient preparation of pure compounds. During the past decade this view has been radically ~hanged.~ It is now widely recognized that radical reactions, even with highly complex and heavily substituted substrates, can be con- ducted in a highly selective and efficient manner, and often display advantages over alternative ionic processes.Conse- quently, free radical methodology has become a major weapon in the armoury of the synthetic organic chemist, of particular value in natural product synthesi~.~ One might well ask what has brought about this change in attitude. In my view it is the development of our understanding of the factors that influence the various forms of selectivity -chemoselectivity, regioselectivity, and diastereoselectivity. With a knowledge of these factors it is often possible to define reaction conditions and reagents which will ensure that a reaction proceeds at one functional group in preference to another, at one position in preference to another, and with one diastereisomer in preference to another, or to afford one diastereisomer in prefer- ence to another.In this article I shall outline some of the experiments which contribute to the recognition and definition of the factors affecting the selectivity of radical processes. The examples will be drawn mainly from work carried out by my group;6 further relevant results may be found in papers and reviews by such pioneers in the field as Walling, Julia, Tedder, Ruchardt,and Ingold,’ and more recently by Curran, Porter, Newcomb, Giese, and others. 53839 Much of our knowledge about the factors influencing the selectivity of radical reactions comes from kinetic studies.This is understandable since a highly selective reaction is one in which the formation of the desired product is very much more rapid under the experimental conditions used than all other possible reactions. The factors which affect selectivity are therefore those factors which affect the energies of transition complexes and hence the magnitudes of the rate constants. The importance of thermochemistry was recognized in the earliest studies. The thermochemical approach was expressed in such generalizations as ‘radical reactions follow the most exo- thermic available pathway’ or ‘radical reactions afford the most stable possible product’. It is based on the assumption that activation enthalpies reflect reaction enthalpy changes, and it leads to the conclusion that the relative rates of related reactions 143 can be estimated from bond dissociation energies.It underlies Benson's approach to the quantitative calculation of kinetic parameters. O The thermochemical approach is especially useful for assess- ing the relative rates and directions of simple addition and ,$-fission processes (equation 3). Thus, when A' represents a carbon-centred radical and B=D a carbonxarbon double bond, the addition is exothermic. Such reactions are usually relatively fast, and the relative rate constants often roughly reflect the exothermicities. However, when B=D represents the carbonyl group, the reverse reaction, i.e. the ,&fission of alkoxy radicals, is favoured on thermochemical grounds.The same is true for homolytic substitution reactions (equation 2). Thus, the order of reactivity of various substrates with Bu,Sn' radicals, namely RI > RBr > RSeAr > RC1> RSAr > RSMe is roughly in the same order as the exothermicities of the transfer reaction. However, thermochemistry is not the only factor, nor even the predominant factor, affecting the outcome of many free radical processes. The others are: Stereoelectronic eflects: these reflect the way in which the requirement for overlap of frontier orbitals affects the energy of the transition structure. Polar eflects: these reflect the way in which the electronegativ- ities of the constituent atoms affect the energy of the transition structure.Steric eflects: these reflect the contribution of non-bonded interactions to the energy of the transition structure. The outcome of any particular reaction will reflect the subtle interplay of all of these factors. They apply to all of the general elementary radical reactions (equations 1-4) but are best illustrated by reference to addition and 8-fission reactions (equation 3) and to homolytic substitution reactions (equation 2); these processes will be the subject of this lecture. Further- more, I shall give special attention to intramolecular reactions. By comparison with their intermolecular analogues they often reveal more clearly the constraints that arise from the necessity for the intimate transition structure to be accommodated within the overall molecular architecture of the reactant.Furthermore, they are of especial interest because of their utility for the synthesis of complex natural products. 2 Intramolecular Addition and P-Fission Reactions The relative rates and regiochemistry of intramolecular addition and 8-fission processes clearly reveal the limitations of the thermochemical approach to selectivity. The opening of the cyclopropane ring in suitably constituted radicals provides a case in point. ,$-Fission of the cyclopropyl radical to give the ally1 radical, although highly exothermic (ca. 96 kJ m~l-~),lO has a rate constant at least six orders of magnitude less than that for the mildly exothermic ring opening of cyclopropylcarbinyl radical (E,,, x 29 kJ mol -l, k z1.3 x 1 O8 s -at 25 "C).Many substituted cyclopropylcarbinyl radicals undergo ring-opening even more rapidly. An explanation 6314,1 for these observations rests on the assumption that the transition structure (1) for fl-fission comprises a triangular array of centres arising from interaction of the SOMO orbital with the U* orbital (or possibly the u orbital) of the bond undergoing fission as illustrated in (2). This array is readily attained in cycloalkylcarbinyl radicals (3) in which rotation about the exocyclic bond allows coplanarity between the SOMO and O* orbitals to be reached, but not in the cyclopropyl radical (4)where the orbitals involved are essen- tially orthogonal. For the same reason, cyclobutylcarbinyl radicals undergo moderately fast ring-opening whereas the cyclobutyl radical does not.16 In bicyclic systems such stereo- electronic effects underly the direction of ring opening.' 5,16 Thus ring opening of (5) exclusively affords (6), the less stable of the two possible products (6) and (7),in clear contravention of the thermochemical guideline.The regioselectivity of the 8-CHEMICAL SOCIETY REVIEWS, 1993 (3) U (5) (4) 0 (7) fission of (5) reflects the ability of the SOMO orbital to overlap with the U* orbital of the cyclobutane bond exocyclic to the cyclopentane ring, whereas there is no such effective interaction possible with the U* orbital of the bond forming the ring junction. Stereoelectronic factors also play a dominant role in deter- mining the regiochemistry of many intramolecular addition reactions.The well known regioselective cyclization of hex-5- enyl radical (8) to give the less stable cyclopentylmethyl radical (9) in preference to the more stable cyclohexyl radical (lo), provides another clear contravention of predictions based on thermochemical criteria. The suggestion, first made more than twenty years ago, that this reaction is under stereoelectronic control,14 is now widely accepted. d""0'benzene, 80'-+ (8) (9)(~a.98%) (lO)(ca.2%) The preference for formation of the smaller possible ring (exo cyclization) also applies to a large number of substituted hexenyl radicals and related systems (e.g. Scheme 1 where B=D repre- sents C=C, C=O, N=N, C=N, C-C, C=N, etc.; A' represents C', Si', So,O', N', etc.; and n represents a chain of 1 to 5 atoms, not all of which are necessarily carbon atoms).exo endo Scheme 1 This behaviour is a reflection of the stereoelectronic demands of the intimate transition structure for homolytic addition which incorporates the three atoms involved in bond breaking and bond making at the corners of an obtuse triangle orthogonal to the nodal plant of the 7~system. Molecular orbital calculations1 on the transition structure (1 1) for addition of an alkyl radical to an olefin show that the bond being formed is very long (ca.2.4A)and forms an angle of about 106"with the ethylenic C-C bond. Formation of the transition complex is thought to involve interaction of the SOMO orbital with the vacant 7~*orbital as shown in (12). Essentially, this requires the radical centre to behave as a nucleophile. The transition structure should there- fore be dipolar, and its energy should be sensitive to the polar nature of substituents.The qualitative rationale for the preferred exo-cyclization of hex-5-enyl radical and related species rests on the hypothesis that the strain engendered in accommodating the mandatory THE PURSUIT OF SELECTIVITY IN RADICAL REACTIONS-A. disposition of reactive centres within the transition structure for 1,&ring closure outweighs those steric and thermochemical factors expected to favour the formation of the more stable possible product.Theoretical considerations support this view. The geometries and strain energies of the transition structures for exo- and endo-cyclization of the hex-5-enyl radical can be estimated by a combination of molecular orbital and molecular mechanics calculations. In our approach' the dimensions of the intimate transition structure, as determined by the MNDO- UHF method, were incorporated in the transition structures for 13-and 1,6-ring closure and the minimum strain energy of each was calculated by the MM2 program. In accord with the hypothesis that the strain engendered in accommodating the required disposition of reactive centres is greater for the endo- transition structure than it is for the exo, the calculated energy for (14) was found to be ca.11 kJ mol-l greater than for (13). The assumption made in this method that the dimensions of the intimate transition structure are invariant for a variety of cyclization reactions is clearly invalid and must lead to inaccur- ate outcomes. An alternative method devised by Houklg resolves this difficulty. Nevertheless, our method correctly pre- dicts the major isomer formed in more than sixty reactions, and is a useful adjunct to the use of radical cyclizations in synthesis. The calculated transition structure (1 3) for exo-cyclization of hex-5-enyl radical resembles cyclohexane in its chair form. For a typical monosubstituted system (e.g. the 4-methylhex-5-enyl radical) there are therefore two possible diastereoisomeric tran- sition structures: one (15) in which the substituent is pseudo- axial, and the other (16) in which it is pseudo-equatorial. The latter is expected to be of lower energy, and this is confirmed by calculation. Hence, cyclization of 4-substituted hexenyl radi- cals affords preferentially the trans-product.20 The observed preference for trans-cyclization of 2-substituted hexenyl systems and for cis-cyclization of 1-or 3-substituted systemsz0 can be similarly rationalized.Although calculations based on pseudo chair transition struc- tures such as (1 5) correctly predict the stereochemistry of the major cyclization product, they tend to overestimate the degree of selectivity. This led ourselves2' and Houk l9 to consider the possibility that pseudo boat structures might lie on the pathways to minor isomers.The cyclization of 3-t-butyl-hex-5-enyl radical (17) allows a clear cut distinction to be made. The calculated L. J. BECKWITH difference in strain energy between the chair-like transition structure (19) leading to the cis-isomer and that (20) leading to the trans is 19.5 kJ mol-l. On this basis the diastereoisomeric ratio cis:trans should be about 200: 1. However, the calculated difference between the chair-like structure (19) and the boat-like structure (21) is only 5.1 kJ mol-l and the predicted cis:trans ratio is 3.5:l. The experimental cis:trans ratio is 4.1:1.21 We conclude, therefore, that the minor diastereomer (trans) formed from cyclization of 3-t-butylhex-5-enyl radical (17) arises from the boat-like transition structure (21). The same is believed to be true for the cyclization of many other monosubstituted hexenyl systems.z The involvement of boat-like transition structures sets an upper limit to the cis:trans ratio since their energies are relatively insensitive to the size of the substituent.However, since the differences in strain energy appear mainly in the activation energy term of the Arrhenius equation the diastereo- selectivity is markedly increased by decrease in temperature. Thus the transxis ratio for cyclization of 4-methylhex-5-enyl radicals is ca. 3.7 at 80 "C,but is ca. 8.3 at -40 "C. , Ip+CH2* BU' BU' (17) [ (18)(cis. : trans = 4.1) :* ]* [ButL]* 4 [But But The formation of bicyclic and polycyclic systems by intramo- lecular homolytic addition to a suitably constituted cycloalkene illustrates another important facet of stereoelectronic effects.Since the intimate transition structure requires the approach of the radical centre orthogonal to the nodal plane of the 7~ system, the reaction can only proceed through that conformation of the ring which places the side chain in a pseudo-axial position (Figure 1). Consequently, such reactions afford exclusively cis-fused products; e.g. the formation of (23) by ring closure of (22) (R = Me).22 Figure 1 Transition structure for formation of a bicyclic system [ring closure of 4-(cyclohex-2-enyl)butyl radical]. This type of reaction also reveals the importance of polar and steric effects.Thus, the radical (24) containing a carbonyl group undergoes ring-closure some 40 times more rapidly than does (22) (R = H), which does not.22 This is a manifestation of the polar effect. The electron-accepting carbonyl group facilitates the reaction by stabilizing the dipolar transition state. This is a general phenomenon which has been extensively explored by Gie~e.~~ Radicals such as (22) which contain a substituent at one of the possible points of attack undergo reaction mainly at the unsub- stituted double bond. This illustrates another important genera- lization, namely that substituents on a double bond exert a powerful steric retardation on the rate of homolytic attack at the position of s~bstitution.~~Thus, the rate constant for the ring- closure onto the substituted double bond of (22) (R = Me) to give (25) is about 30 times less than that for addition on the unsubstituted double bond to give (23).In the light of this, and I46 CHEMICAL SOCIETY REVIEWS, 1993 similar results, it is clear that the observed Markownikoff mode of addition for intermolecular radical attack on unsymmetrical olefins reflects the steric hindrance of approach to the more substituted position rather than, as is often supposed, the greater stability of the more substituted adduct radical. Thus the outcome of such addition reactions depends more on steric than on thermochemical effects. R &MeOCO CH2* MeOCOMeQ 0 All of the factors affecting selectivity are nicely illustrated by the formation of the bicyclic compound (29) from the precursor (26).25 The noteworthy features of the mechanism are: (i) The chemoselective transfer of a bromine atom from (26) to the tin-centred radical to afford (27): an example of the import- ance of thermochemical factors, since this is a highly exothermic process.(ii) Cyclization of the radical (27) to give exclusively the smaller possible ring and the cis-fused product: manifestations of stereo-elec tronic effects. (iii) Highly efficient ring closure, despite steric hindrance by the methyl group, reflecting the polar effect of the electron- withdrawing substituent. (iv) Diastereoselective hydrogen transfer from stannane to the less sterically encumbered convex face of the radical (28).C02CH3 Bu3SnH Bu3Snm ,tJ ..H H (29) (28) Ring closure to form bicyclic systems also occurs readily in suitably substituted dihydropyridines. Thus, treatment of the bromo compound (30) with tributylstannane in the usual way gives the quinolizidine (31) in high yield.26 It is noteworthy that in this case the final atom transfer step is opposite in diastereo- selectivity to that observed for (28). This is attributed to rapid inversion at the nitrogen centre of the initially formed cis-fused radical (33) to give a species (34) with a pseudo-trans ring junction. The hydrogen transfer then proceeds from the less sterically hindered pseudo-axial direction to give exclusively the product (3 1) with an equatorial carbomethoxy group.Reduc- tion of (31) gives ( f)-epilupinane (32) in good overall yield.26 Cyclizations to give pyrrolizidines or quinolizidines also occur readily by intramolecular homolytic substitution in Br __t n appropriately constituted pyridones and dihydropyridones. Substituted radicals of the general type (36) undergo highly diastereoselective cyclization. Thus the compounds (35a-d) when treated with tributylstannane each give high yields (290%) of the corresponding products (37) formed by intra- molecular addition on the face of the dihydropyridone trans to the s~bstituent.~~ (35a; n = 1, R= Ph) (35b; n = 2, R= Ph) (35c; n = 1, R = CH3) (35d; n = 2, R = CH3) This outcome can be understood in the light of models of the transition structures incorporating the reactive centres in the usual triangular array orthogonal to the nodal plane of the x system.Figure 2 depicts the structures for the cyclization of (36d) viewed from above the plane of the heterocyclic ring. When the approach of the radical centre is anti to the methyl substituent it assumes a pseudo-axial position in the transition structure (38). However syn attack leads to a transition structure (39) in which the methyl is pseudo-equatorial. The strain energy of structure (38) leading to the preferred product, (37d), was found to be almost 12 kJ mol-l less than that of the alternative structure(39). The calculations indicate that in the higher energy (39) Figure 2 Transition structures for cyclization of the radical (36d).THE PURSUIT OF SELECTIVITY IN RADICAL REACTIONS-A. structure (39) there is a severe non-bonded interaction between the pseudo-equatorial methyl group and the amide oxygen atom. This is not present in (38) where the substituent is pseudo-axial. 3 Homolytic Substitution The preceding examples show how stereoelectronic factors strongly influence the outcome of many intramolecular homo- lytic addition and 8-fission reactions. The same applies to intramolecular homolytic substitution (&2) reactions. In general terms, there are two modes available for such a reaction affording two different types of product (Scheme 2). When the atom undergoing substitution lies at the remote end of the group (B-D) under attack, the reaction (endo-SH2) affords a re-arranged radical; when it lies at the nearer end of B-D, the reaction (exo-SH2)proceeds with ring formation and the expul- sion of D'.The outcome in specific cases is expected to depend on the ability of the alternative gross transition structures to accommodate the optimum shape and dimensions of the inti- mate transition structure. / A-B L,J exo D* \ A-D B* L"J endo Scheme 2 Examples of both modes of reaction have been well docu- mented. Hydrogen atom transfer from carbon to oxygen involves an approximately linear transition structure and lengths of about 1.4 8, for C---H and about 1.28, for O---H.This can be accommodated in a six-membered cyclic structure but not in smaller rings.28 It is not surprising, therefore, that the well-known Barton method for remote functionalization usually proceeds mainly, and sometimes exclusively, by 1,5-hydrogen atom tran~fer.~~ The Barton reaction is an example of an intramolecular homolytic substitution reaction that proceeds in the endo mode, and is regioselective because of stereoelectro- nic factors.Intramolecular homolytic substitution at sulfur, however, appears always to be confined to the exo mode. Thus, treatment of sulfides of the general type (40) with tributylstan- nane in low concentration affords only the cyclized product (42) and the appropriate alkane, RH.30,3 Clearly, the reaction involves exclusive exo homolytic substitution at sulfur in (41) with expulsion of an alkyl radical R' which then undertakes hydrogen atom transfer from the stannane.Stereoelectronic factors must be of prime importance since the more exothermic possible process would involve endo substitution to give the rearranged radicals (43). 6' (40) X"-CH2* (43) (44) Cyclization of radicals of the general type (41) is relatively fast with rate constants of the order of los s-at 80 0C.30The values of the relative rate constants reflect the relative stabilities of the radicals expelled, i.e. the rates of cyclization of (41) for various L. J. BECKWITH groups Rare in the order benzyl > t-butyl > methyl. Intramolec- ular attack by an alkyl radical on sulfur in suitably constituted substrates is much slower.32 Thus, radicals of the general type (44) where X is an alkyl group undergo cyclization with rate constants in the range lo2-lo4 s-l.The reaction is also slow when the leaving group is acyl or aroyl, but is much faster when it is an alkylthio group. Thus, k, for 44 (X = PhCO) is about 2 x lo4s-at 80 "C,while for (44) (X = Buts) it is about 2 x lo7 s-l at 80°C.30 A plausible explanation for these observations is that radical substitution at sulfur proceeds through an approximately linear transition structure and involves concerted bond-formation and bond-fission. Such an arrangement of centres could not be accommodated in the 5-or 6-membered cyclic transition struc- tures involved in endo substitution. Molecular orbital calcula- tions recently completed by Schiesser support this hyp~thesis.~~ Intramolecular homolytic substitution at the sulfur centre in appropriately constituted sulfoxides also proceeds under stereo- electronic control.Thus, treatment of the optically active sulfox- ide (45) with tributylstannane affords only the one enantiomer (46) of the exo-cyclization product. Thus this reaction is highly regioselective and also stereospecific since it proceeds with strict inversion of stere~chemistry.~ Presumably it involves an inti- mate transition structure (47);it represents, therefore, a homoly- tic analogue of the Walden inversion. Reactions involving intramolecular homolytic substitution of sulfur are useful for the preparation of a variety of complex heterocyclic systems.Examples include compounds e.g.(48)and (49) related to the penems and ~ephems;~O>~~ some of them show interesting biological activity. The formation of such com- pounds raises the question of whether the biosynthesis of penicillin involves a somewhat similar mechanism. The notion that the thiazolidine ring might be formed by intramolecular alkyl radical attack on sulfur attached to the enzyme through iron, appears to be compatible with all the results obtained from biosynthetic ~tudies.~ Buts Br C -(48)(89%) R Br (49) (60-70%) In all of the preceding examples of intramolecular sH2 reactions, the high regioselectivity, and, in the case of substitu- tion at sulfoxide sulfur, the stereospecificity, is imposed by the necessity for the intimate transition structure to be accommo- dated within the overall molecular architecture of the radical undergoing reaction. For intermolecular sH2 processes, this constraint no longer applies and the factors affecting selectivity are more subtle.Sometimes electronic interactions with neigh- bouring groups or atoms are important. For example, maximum stabilization of the transition structure for an SH2process will be attained when the bond being broken or formed can assume coplanarity with a neighbouringp or T orbital containing one or two electrons 24 An early example of this type of stereoelectronic effect was detected when trans-3-chloro-5-t-butylcyclohexanewas found to react with tributyltin radicals 10 times more rapidly than the cis-isomer 36 Since the chlorine is pseudo-axial in the trans- isomer (50), the transition structure can readily adopt the energetically favoured disposition of centres whereby the bond undergoing fission lies in the same plane as the adjacent 7~ orbital This would only be possible for the cu-isomer (51) if it were to adopt the alternative highly energetic conformation with the bulky t-butyl group in a pseudo-axial position Similarly, chlorination of 4-t-butylcyclohexene with t-butyl hypochlorite affords mainly the trans-isomer (5) 36 The inter- mediate allylic radical (52)undergoes preferential pseudo-axial bond formation, even though the product, being less stable than its isomer, is disfavoured on thermochemical grounds An early example of the effect of adjacent lone pairs on the diastereoselectivity of radical reactions was provided by an examination of the relative rates of reaction of some conforma- tionally locked dioxanes with t-butoxyl radical It was observed that the compound (53) (R = Me) in which the hydrogen at C-2 is equatorial reacts about 12 times more slowly than its more stable isomer (54)(R = Me) containing an axial hydrogen atom 37 This contravention of the thermochemical rule is con- sistent with the view that substitution at the axial hydrogen atom proceeds vza a transition structure in which the bond undergoing fission can interact efficiently with the adjacent oxygen lone pairs 37 Further confirmation that the reaction is relatively unaffected by thermochemical factors was obtained when the reactivity of the methoxy compounds was examined Because of the anomeric effect, the isomer (53) (R = OMe) with an axial methoxy substituent is the more stable Nevertheless, the prefer- ence for cleavage of the axial C-H bond in (54) (R = OMe) over that in czs-isomer (53)(R = OMe) was almost the same as that observed for the methyl-substituted compounds R H (53) (54) (55) Stereoelectronic factors should also favour axial attack on radicals denved from other cyclic ethers Examples of reactions of radicals derived from carbohydrates which display the pre- dicted diastereoselectivity have recently been reported The regiochemistry of intermolecular radical substitution is often determined by thermochemical factors For example, halogenations with t-butyl hypochlorite or N-bromosuccin-imide usually proceed through highly regioselective hydrogen- atom abstraction at allylic or benzylic positions because of the resonance stabilization of the resulting radicals However, recent experiments by Roberts39 have shown that in some hydrogen atom transfer reactions polar effects can play a crucial role A good example39 40 is provided by an ESR examination of hydrogen atom abstraction from butyrolactone When t-butoxyl radicals act as the hydrogen acceptor the reaction is confined to the position adjacent to the ether oxygen to afford the radical (55) Since the t-butoxyl radicals are strongly electro- philic the transition structures for hydrogen atom abstraction are dipolar The transition structure, on the pathway to (55),is stabilized by the electron donating character of the oxygen atom of the substrate However, hydrogen atom abstraction by the CHEMICAL SOCIETY REVIEWS, 1993 aminoboryl radical Et,N-BH,’ proceeds exclusively at the position adjacent to carbonyl Since the boryl radical is nucleo- phili~,~~the dipolar transition structure for hydrogen atom transfer on the pathway to (56) is stabilized by the electron accepting carbonyl group 4 Captodative Radicals and Related Species It has been suggestFd from time to time4’ 42 that carbon-centred radicals, eg D-C-X=Y, flanked by an electron accepting substituent X=Y (eg C=O, C=N) and an electron donating atom D (e g 0,N, S) are more stable than would be expected from the sum of the stabilizing effects of the two substituents In such ‘captodative’ radicals41 the presence of both types of substituents allows contributing structures to be drawn which indicate that conjugation extends from one end of the system to the other In other words, the two substituents act synergistically to bestow enhanced stabilization on the radical The concept of captodative stabilization has been criticized, and recent experimental seems to indicate that synergis- tic stabilization is, at best, relatively unimportant, a conclusion also reached in a comprehensive review 42 Nevertheless, it is clear that radicals with captodative substituents exhibit the restriction of rotation expected from systems with extended delocalization 44 Since radicals derived from such biologically significant compounds as a-hydroxy acids and a-amino acids are formally captodative, it is important to ascertain whether spe- cies of this type show any special selectivity arising either from their thermodynamic stabilization or the extended delocaliza- tion of the unpaired electron In order to avoid possible ambiguities in the interpretation of results of experiments with acylic radicals arising from the effect of non-bonded interactions on their conformations, we chose to work with cyclic compounds Bromination of the dioxolane (57) with N-bromosuccinimide under bromine atom chain con-ditions gave only the product (59) (> 90”/0), whereas similar bromination of the dioxolanone (60) occurred exclusively at the position adjacent to the carbonyl group to give (62) (> 92%) 45 Clearly hydrogen atom transfer to Br’, an electrophilic radical, affords only (58) from (57),and (61) from (60) This regioselecti- vity is unexpected in view of the preference for attack on butyrolactone by the electrophilic t-butoxyl radical at the position adjacent to the ether oxygen to give (55) Although it is tempting to propose that the selectivity of the bromination reaction reflects captodative stabilization of the radical (61) it is more likely that it represents the outcome of the subtle interplay of polar and thermochemical effects What is significant, however, is that the reaction with the substituted dioxolanone (63) (R = H) affords exclusively and in excellent yield (96%) the trans-brominated product (64) (R = H) 45 Similarly, bromination of disubstituted substrates of the general type (63) (R = alkyl) always occurs antr to the t-butyl substituent This propensity of radicals of the general type (65) to undergo stereoselective bond formation anti to the t-butyl substituent applies also to other radical reactions Thus, treat- ment of the bromo compound (64) (R = Me) with tributyltin deuteride affords (66) as the major diastereoisomer (> 90%), while the reaction of (64) (R = Me) with allyltributyltin gives a mixture of isomers (7 1) in which (67) is the major component 45 Surprisingly, the diastereoselectivity of the latter reaction is THE PURSUIT OF SELECTIVITY IN RADICAL REACTIONS-A.L. J. BECKWITH 149 greater than that (trans:cis= 1.9) for the allylation of (64) (R = H) to give (67) (R = H). Oxazolidinones and imidazolidinones of the general type (68) (R = H or Me; X = 0 or NMe) also undergo bromination exclusively at the 'captodative' The determination of the stereochemistry of the bromination products is difficult but the bulk of the evidence suggests that they have the general structure (69) (R = H or Me; X = 0or NMe). The situation for other radical reactions of oxazolidinones and imidazolidinones is more clearcut as some of the products have been examined by X-ray ~rystallography.~~ It appears that reactions involving the intermediacy of radicals (70) (X = 0or NMe) exhibit diastereo- selectivity opposite to that for the dioxolanones to afford products (69) arising from bond formation syn to the t-butyl group.Although compounds of the general type (70) (X = 0or NMe, R = H; R' = Ph) show relatively little diastereoselectivity in their reactions with tributylstannane, the selectivity is enhanced markedly when the size of the substituent R or of R' in the acyl group is increa~ed.~ The corresponding imidazolidi- nones behave similarly but usually with higher diastereoselecti- vity. The tendency for radicals of the general type (70) (X = 0or NMe) to undergo diastereoselective bond formation syn to the t- butyl group is exemplified by the highly diastereoselective formation of (71) and (72) by treatment of the appropriate precursors with tributyltin de~teride.~~ As yet, a satisfactory explanation for the highly selective behaviour of radicals of the general types (65) and (70) (X = 0 or NMe) has eluded us.In the case of radicals [e.g. (65)] derived from dioxolanones the preference for bond formation anti to the bulky t-butyl group may reflect steric hindrance to the syn face of the molecule. The fact that allylation of (64) (R = H) and (64) (R = Me) with allyltributylstannane occurs anti to the t-butyl group in both cases but that the diastereoselectivity is greater when R is methyl suggests that the reaction cannot be under thermochemical control. For reactions of radicals of the general type (70) (X = 0or NMe) the factors underlying diastereoselec- tivity are even more obscure; attempts to probe possible stereo- electronic effects by application of the usual molecular mecha- nics calculations have failed because of the lack of appropriate parameters.Radical reactions of appropriately constituted dioxolanones, oxazolidinones, and imidazolidinones are very useful for the enantioselective synthesis of a-hydroxy acids and a-amino acids. From the readily available enantiomers of lactic acid or malic acid it is a relatively straightforward matter to prepare either the (2R)-methylene compound (73) or its (2S)-enanti0mer.~~ Treat-ment of (73) either with tributylstannane and an alkyl iodide or bromide, or with an alkylmercuric halide and a borohydride reducing agent, affords the enantiomerically pure (2R, 5R) product (75) with high diastereoselectivity.The optically active hydroxy acid (76) can then be readily obtained by hydrolysis of (75). The (3-enantiomer of (76) can be similarly prepared from the (2S)enantiomer of (73).45 P40 (75) (76) The stannane and alkylmercuric halide procedures involve essentially the same mechanism. An alkyl radical generated by reaction of RX with Bu,Sn', or by a-fission of RHg', adds regioselectively to (73) to give the radical adduct (74) which then undergoes highly diastereoselective hydrogen atom transfer either from the stannane, or from the alkylmercuric hydride (RHgH) generated by borohydride reduction of the alkylmer- curic halide.In similar vein, the optically active (23-oxazolidinone (77), which can be readily prepared from (R)-alanine, undergoes addition of an alkyl radical to afford the radical (78). In this case hydrogen atom transfer occurssyn to the t-butyl group with high diastereoselectivity to give the (2S,5S)-oxazolidinone (79) from which the (9-a-amino acid (80) can be obtained in good yield.4S The preparation by this method of naturally occurring a-amino acids allows the stereochemistry of intermediates in the reaction sequence to be securely assigned.45 :N%i, :N 4CH2R PhCO PhCO A$H2A RCH,-CO,H FlH2PhdO Although it has not yet been possible to define the role of stereoelectronic effects in the reactions of such formally 'capto- dative' radicals as (74) and (78), there is good evidence that they are important in defining the stereochemical behaviour of some related species.Carbon centred radicals [e.g.(83)] flanked by an oxygen atom and a C-0 single bond can be formally regarded as captodative, since the former acts as the electron donor and the a* orbital of the latter acts as the electron acceptor. Evidence for a stabilizing interaction between the SOMO orbital, the a* orbital, and the filled oxygen orbital as indicated in structure (8 1) (the 'homoanomeric effect'47) comes from ESR ~tudies~~,~~ which show that radicals containing this type of system have high barriers to rotation about the C-0 and C-CO bonds, and assume a relatively stable conformation in which the C-0 bond is co-planar with the SOMO orbital.44 This has a dramatic effect on the diastereoselectivity of the reactions of suitably constituted radicals.Thus, the ESR spec- trum of the 2-butanoyloxycyclohexyl radical shows that it preferentially assumes the conformation (82) in which the ester group is pseudo-equatorial 48 49 Atom transfer to (82) from tributyltin deuteride occurs selectively from the axial direction to afford mainly the cu-isomer of (84) (X = CH,) 49 However, when the radical contains an oxygen atom u to the radical centre, it takes up the conformation (83) in which the ester group is axial The reaction with tributyltin deuteride then occurs on the less encumbered face of (83) anti to the acyloxy group to afford mainly the trans-isomer of (84) (X = 0)49 (83) (84) The evidence that radicals of the general type (81) are stabi- lized by interaction of the SOMO with both the adjacent lone pair and the adjacent U* orbital raises the question of whether a C-0 U* orbital alone will stabilize an adjacent carbon radical centre Earlier observations that xanthates and similar species derived from /3-alkoxy alcohols give good yields of deoxygen- ated products on treatment with tributylstannane under Barton conditions, whereas substrates lacking the oxygen substituent do not, prompted the suggestion that /3-alkoxyalkyl radicals are stabilized by the so-called ‘/3-oxygen effect’ so A stabilizing interaction between the SOMO and (I* orbitals would require their co-plananty as indicated in (85) However, ESR measure- ments on /3-alkoxyalkyl radicals indicate that the preferred conformation has the C-0 bond nearly orthogonal to the SOMO orbital s1 Furthermore, when a mixture of a xanthate (86) (X = CH,) with its analogue containing a /3-oxygen atom (86) (X = 0)is treated with tributylstannane, the two deoxygen- ated products (87) (X = CH,) and (87) (X = 0)are formed in approximately equal yields These and similar experiments involving the competitive formation of radicals both inter- and intra-molecularly suggest that the presence of a /3-oxygen substi- tuent has little effect on the ease of formation and, by impli- cation, on the stability of a carbon-centred radical The results previously adduced in support of the ‘/3-oxygen effect’ must, therefore, have some other explanation R-X- !? CH2-CH2-C-SMe R-X-CH,-CH, R-fl U U (85) Finally, I turn to another example of a highly diastereoselec- tive homolytic reaction, the outcome of which appears to reflect the conformational preference of the intermediate radical Treatment of the iodosulfoxide (88) with tributylstannane gave mainly one diastereoisomer (90) of the product (d e > 94%) 52 The reaction with allyltributylstannane was even more selective and afforded only one detectable diastereisomer (d e > 98%) of the product (91) The reactions of (88) with hex-1-ene and with hexa- 1,Sdiene were equally diastereoselective The relative stereochemistry of (9 1) was determined by X-ray crystallogra- phy, while assignments of stereochemistry for other products were based on the comparison of NMR spectra These results are all consistent with the view that the intermediate radical (89) has a relatively stable conformation (92), the preference for which results from the dielectric repulsion between the oxygens CHEMICAL SOCIETY REVIEWS, 1993 of the carbonyl and sulfoxide groups Bond formation, either by hydrogen atom transfer or by addition to a carbon<arbon double bond, is confined to the less encumbered face of (92) 01 0 I .OH 1 v Ar-&-$-CO,Et Ar-S-$ -C02Et ,S-y *gCO,Et Me Me Ar Me I Me-C-C0,Et eH2CH =CH2 Unfortunately, the hypothesis of restricted rotation in (92) lacks experimental support since we have, as yet, been unable to conduct successful ESR experiments However, the correspond- ing radical containing an arylthio group in place of the arylsulfi- nyl clearly exhibits a large barrier to rotation in the ESR,44 presumably because of interaction of the sulfur lone pair with the adjacent SOMO and T* orbitals Recently Porter, Curran, and Giese have described a number of highly diastereoselective radical reactions, the selectivity of which appears to depend upon the relative stability of the preferred conformations of radical intermediates 5 Conclusion In this article I have tried to give an overview, albeit one based mainly on our own work, of experiments that have contributed to our understanding of the factors affecting the selectivity of radical reactions, an understanding that has underpinned the dramatic developments during the past decade in the application of radical methodology to organic synthesis The picture is far from complete Although it is clear that stereoelectronic effects play a dominant role in determining the outcome of many simple homolytic intramolecular and substitution reactions, the selecti- vity exhibited by intermolecular radical reactions with complex substrates is often less easily rationalized Undoubtedly, in some cases thermochemical, polar, stereoelectronic, and steric effects all play a part The further investigation of the factors affecting the structure, stability, and reactivity of radicals, and the selectivity of their reactions remains an important and fascinat- ing field of research deserving the attention of spectroscopists, kineticists, and synthetic organic chemists I have no doubt that their endeavours will be well rewarded 6 References 1 M Gomberg, J Am Chem Soc , 1900,22,757 2 A L J Beckwith and P J Duggan, J Chem SOC Perkin Trans 2, 1992, 1777, and references cited therein 3 N L Bauld, Adv Electron Transfer Chem , 1992,2, 1 4 For a recent comprehensive review of organic free radical chemistry see A Ghosez, B Giese, W Mehl, J 0 Metzger, and H Zipse, in ‘Methoden der organische Chemie’, (Houben-Well), ed M Regitz and B Giese, Georg Thieme, Stuttgart, 1989, Vol E19a, Parts I and 2 5 C P Jasperse, D P Curran, and T L Feviz, Chem Rev , 1991,91, 1237 6 For a review of earlier work see A L J Beckwith, Tetrahedron, 1981, 37,3073 7 M Julia, Pure AppI Chem , 1974,40,553,A L J Beckwith and K U Ingold in ‘Rearrangements in Ground and Excited States’, ed P de Mayo, Academic Press, New York, 1980, Vol 1, p 161,M Julia, Acc Chem Res, 1971, 4, 386, J M Tedder and J C Walton, Tetrahedron, 1982, 38, 313, J M Tedder, Adv Phys Org Chem , 1978, 16, 51, C Walling, ‘Free Radicals in Solution’, Wiley, New York, 1957, C Ruchardt, Top Curr Chem , 1980,88, 1 8 N A Porter, B Giese, and D P Curran, Acc Chem Res ,1991,24, 296 9 D P Curran, Synthesis, 1988, 417, 489, B Giese, ‘Radicals in THE PURSUIT OF SELECTIVITY IN RADICAL REACTIONS-A Organic Synthesis Formation of Carbon-Carbon Bonds’, Perga- mon, Oxford, 1986, T V RajanBabu, Acc Chem Res , 1991, 24, 139, M Ramadah, Tetrahedron, 1987, 43, 3541, M Newcomb, Tetrahedron, 1993,49, 1 15 1 10 S W Benson, Thermochemical Kinetics’, Wiley, New York, 1973 11 H Fischer in ‘Substituent Effects in Radical Chemistry’, ed H G Viehe Z Janousek, and R Merenyi, Reidel, Dordrecht, 1986, NATO AS1 Series C, Vol 189, p 123 12 A J L Beckwith and P E Pigou, Aust J Chem , 1986,39,77 13 V W Bowry, J Lusztyk, andK U Ingold, J Am Chem Soc , 1991, 113,5687, and references cited therein 14 A L J Beckwith in Essays on Free Radical Chemistry’, ed R 0 C Norman, Chemical Society, London, 1970, p 239 15 A L J Beckwith and G Philhpou, Aust J Chem , 1976,29, 123 16 A L J BeckwithandG Moad, J Chem Soc Perkin Trans 2,1980, 1473 17 K N Houk, M N Paddon-Row, D C Spellmeyer, N G Rondan, and S Nagase, J Org Chem , 1986,51,2874 18 A L J Beckwith and C H Schiesser, Tetrahedron, 1985,41, 3925 19 D C Spellmeyer and K N Houk, J Org Chem , 1987,52,959 20 A L J Beckwith,C J Easton,T Lawrence,andA K Serelis,Aust J Chem , 1983,36,545 21 A L J Beckwith and J Zimmermann, J Org Chem ,1991,56,5791 22 A L J Beckwith and D H Roberts, J Am Chem Soc , 1986,108, 5893 23 B Giese Angew Chem Int Ed Engl, 1983,22,753 24 A L J Beckwith, C J Easton, and A K Serelis, J Chem Soc Chem Commun , 1980,482 25 S W Westwood, unpublished work 26 A L J Beckwith and S W Westwood, Tetrahedron, 1980,45,5269 27 A L J Beckwith, S Gerba, S Joseph, and R T A Mayadunne in ‘Abstracts of the 13th National Conference of the Organic Chemistry Division’, Royal Australian Chemical Institute, Mel- bourne, 1992, p 55 28 A E Dorigo and K N Houk, J Org Chem ,1988,53,1650, K N Houk, J A Tucker, and A E Dorigo, Acc Chem Res , 1990,23, 107 29 D H R Barton, Pure Appl Chem , 1968, 16, 1, R H Hesse, Adv Free-Radical Chem , 1969,3, 83 30 S A Munaweera Ph D Thesis, Australian National University, 1989 31 A L J Beckwith and D R Boate, J Chem Soc Chem Commun , 1986, 189 L J BECKWITH 32 J A Franz, D H Roberts, andK F Ferris, J Org Chem ,1987,52, 2256 33 J E Lyons andC H Schemer, J Chem Soc Perkin Trans 2,1992, 1655 34 A L J BeckwithandD R Boate, J Org Chem, 1988,53,4339 35 J E Baldwin in ‘Recent Advances in the Chemistry of 8-Lactam Antibiotics’, ed P H Bentley and R Southgate, Royal Society of Chemistry, London, 1989, p I, R M Adlington in ‘Molecular Mechanisms in Bioorganic Chemistry’, ed C Bleasedale and B T Golding, Royal Society of Chemistry, Cambridge, 1990, p 1 36 A L J Beckwith and S W Westwood, Aust J Chem, 1983, 36, 2123 37 A L J Beckwith and C J Easton, J Am Chem SOC,1981, 103, 615 38 D Crichand L B L Lim, J Chem Soc Perkin Trans I, 1991,2209, and references cited therein 39 P Kausha1,P L H Mok,andB P Roberts, J Chem Soc Perkin Trans 2, 1990, 1663 40 S Brumby, unpublished work 41 H G Viehe, Z Janousek, R Merenyi, and L Stella, Acc Chem Res , 1985,18, 148 42 R Sustmann and H G Korth, Adv Phys Org Chem ,1990,26,13 1 43 F G Bordwell, T Gallagher, and X Zhang, J Am Chem SOC, 1991,111,3495 44 A L J Beckwith and S Brumby, J Chem Soc Perkin Trans 2, 1987, 1801 45 A L J BeckwithandC L L Chai,J Chem Soc Chem Commun , 1990, 1087, C L L Chai, Ph D Thesis, Australian National University, 1989 46 A C Willis, A L J Beckwith, and M J Tozer, Acta Cryst, 1991, C47,2276 47 H -G Korth, R Sustmann, J Dupuis, and B Giese, J Chem Soc Perkin Trans 2, 1986, 1453 48 P J Duggan and S Brumby, unpublished work 49 P J Duggan, Ph D Thesis, Australian National University, 1990 50 D H R Barton, W Hartwig, and W B Motherwell, J Chem Soc Chem Commun , 1982, 447, for a review of Barton deoxygenation reactions see D Crich and L Quintero, Chem Rev ,1989,89,1413 51 A L J Beckwith,S Brumby,I G E Davison,P J Duggan,andR W Longmore, ‘Abstracts of the Sixth International Symposium on Organic Free Radicals’, Noorwijkerhout, 1992, p 344 52 A L J Beckwith, R Hersperger, and J M White, J Chew Soc Chem Commun ,1991, 1151

ISSN:0306-0012    

DOI:10.1039/CS9932200143

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

The Nature of Ammonium and Methylammonium Halides in the Vapour Phase: Hydrogen Bonding versus Proton Transfer A. C. Legon Department of Chemistry, University of Exeter, Stocker Road, Exeter EX4 4QD 1 Introduction A striking experience, common to most young chemists, is the veil of white fog that hangs eerily over the benches of a school laboratory after a period of vacant calm. The image is both enduring and thought provoking. When encountered in later life the phenomenon is redolent of chemistry in childhood. What is the smoke? Why and how is it formed? The smoke consists, of course, of solid particles of ammonium chloride and has its origin in the slow interdiffusion of the vapours that leak from reagent bottles of ‘0.880’ammonia and ‘conc.’ hydrochloric acid.We learn somewhat later that each solid particle is composed of regular, interpenetrating arrays of ammonium (NH4 +)and chloride (C1-) ions in a body-centred cubic lattice. Presumably, the separate vapours issuing from the two reagent bottles contain NH, and HCl molecules, respecti- vely. This provokes further questions: How do NH, and HCI molecules interact to produce the ionic solid? What is the stable product of the interaction of a single NH, molecule and a single HCl molecule? Is it the simple hydrogen-bonded dimer H,N-**HCl or is a proton transferred from one to the other to give the ion pair H,NH+ -**Cl-? If the former, how does the ionic solid result? Are clusters (H,N -HCl), produced by further interactions of H,N**.HCl in the vapour until, even- tually, the Coulombic stabilization associated with the ionic lattice facilitates proton transfers to give (H,NH *.*Cl-),,,? At + what value of m does this process occur? Are there analogues R, -,H,N - - HX for which the ion pair R, -,H,NH * * + X -is the most stable form in the dimer? What groups R and X and what values of n favour the ion pair? The work described in this review was stimulated by such childhood memories and more mature reflection.Its aim was to answer some of the questions posed above. It was enabled by the development of a powerful tool: rotational spectroscopy of supersonically expanded jets. This technique allows the isolation and detailed characterization of dimers such as (NH,,HCI) before clustering and precipitation can occur.Section 2 describes in outline the technique, the dimer properties to which it leads, and the special problems associated with its application to ammonium and methylammonium halides. The results for the heterodimer (NH,,HCI) in the vapour of A. C. Legon is the Professor of Physical Chemistry in the University of Exeter. He was born at Rookery Farm, near Sudbury in Sufolk but was educated in London: at the Coopers’ Company School, Bow and afterwards at Univer- sity College London. His recent research interests include a systematic investi-gation of the nature of hydro-gen-bonded dimers and other types of complex through the spectroscopy of supersonic jets. He was Tilden Lecturer and Medallist of the Royal Society of Chemistry for 1989-90.the archetypal substance ammonium chloride and their interpre- tation in the light of limiting hydrogen-bond H,N*.*HCl and ion-pair H,NH + C1-models are discussed in some detail in Section 3. The ways in which ammonium chloride might be modified to enhance ion-pair character in the heterodimer are explored in Section 4,wherein the conclusions available from analogous experiments conducted on carefully selected series of ammonium and methylammonium halides are also presented. The consistency of the experimentally derived conclusions of Section 4with simple energetic arguments is examined in Section 5 while the consequences of replacing N by its second row analogue P are reviewed in Section 6.2 How Can the Nature of Heterodimers (R,-,H,N,HX) in theVapour Phase be Established Experimentally? Much of our detailed knowledge of the geometry and electric charge distributions of simple molecules has been derived from spectroscopic constants gained by analysis of rotational spectra. But the application of rotational spectroscopy to heterodimers (R, -,H,N,HX) in the vapour of even the simplest members of the series, namely the ammonium and methylammonium halides, presents several problems which are discussed in 2.1 below. A form of spectroscopy that allows the problems to be overcome is outlined in 2.2 together with a summary of the spectroscopic constants and the molecular properties thereby available.The interpretation of spectroscopic constants to yield molecular properties requires limiting models for a hydrogen- bonded and an ion-pair heterodimer to be chosen. These are discussed in 2.3. 2.1 What Problems are Encountered in the Spectroscopy of Heterodimers (R, -.H,,N,HX)? Rotational spectroscopy is conducted on gases. The ammonium and methylammonium halides, which are the only members of the series (R, -,H,N,HX) so far investigated by rotational spectroscopy, have suitable vapour pressures at temperatures in the approximate range 200-300 “C. Unfortunately, at such temperatures and pressures the vapour is almost completely dissociated into the amine and HX, with only a tiny fraction of the equilibrium mixture present as the heterodimer.Cooling the vapour will increase the mole fraction of heterodimer but decreases the vapour pressure at the same time. It is difficult to find a compromise temperature at which the number density of heterodimers is detectable. What is required is a method of cooling the mixture of amine and HX but without the concomi- tant precipitation of the solid ammonium halide. Such a method exists. It involves supersonic expansion of a dilute mixture of the component substances in, for example, argon through a pin hole into a vacuum to form a jet. The properties of supersonic expansions are well known.’ Of most significance here are the efficient formation of the heterodimer by three-body collisions early in the expansion and the subsequent rapid onset of the collisionless phase of the expansion.Heterodimers surviving until the collisionless phase will then persist until they encounter a wall of the vacuum chamber. No further clustering or precipi- tation is possible and hence the molecules can be interrogated by microwave radiation at this stage to give their rotational spectra (see 2.2 below). Because of the very low effective temperature in 153 the expanded gas, almost all heterodimer molecules are in their vibrational ground state The mixture of the active components in argon can be achieved simply by entraining in argon the vapour in equili- brium with the heated salt When X = Br or I, however, the hot vapour is very reactive and attacks the containing metal vessel In such cases, a more satisfactory approach is to use the so-called fast-mixing technique The components are then held separately at room temperature until the point at which they expand, simultaneously and coaxially, into the vacuum Precipitation of the solid is again avoided and the resultingjet is nch in the heterodimer species of interest 2.2 Rotational Spectroscopy of Supersonic Jets containing (R,-.H,N,HX) The properties of the (R, -.H,N,HX) heterodimers discussed here have been obtained from their rotational spectra observed by a technique called pulsed-nozzle, Fourier-transform micro- wave spectroscopy A short pulse of the gas mixture of interest is expanded supersonically from a relatively high pressure through the nozzle (pin hole) into a vacuum When the gas pulse is in collisionless expansion, it interacts with a pulse of micro- wave radiation If the latter contains frequencies that coincide with a rotational transition of the heterodimer of interest, a macroscopic electric polarization is induced in the ensemble of molecules Compared with the pulse of microwave radiation, the macroscopic polarization of the sample is long lived and there- fore the spontaneous coherent radiation emitted when the polarization subsequently decays can be detected as a free induction decay in the time domain but in the absence of any background radiation Fourier-transformation of these signals leads to the usual rotational spectrum in the frequency domain The reader will be familiar with the pulsed NMR experiment in which a magnetic polarization is induced in an ensemble of nuclear spin vectors and the NMR transition is detected as a free induction decay The microwave experiment outlined above is the exact electric analogue of the NMR experiment The physics of the two methods is identical except that the polarization by microwave radiation involves the alignment of electric rather than magnetic dipoles and that the characteristic time of the free induction decay in the microwave region is much shorter In the present context, the essential feature of the pulsed- nozzle FT microwave spectrometer is the nozzle that produces the gas pulse Two types of nozzle used here have been referred to above the heated nozzle and the fast-mixing nozzle Both are based on a solenoid valve which forms the gas pulse In the heated nozzle, solid ammonium chloride (for example) is con- tained in a channel concentric with a 0 7 mm circular hole in a chamber attached to the base plate of the solenoid valve The chamber is heated to a temperature at which the vapour pressure above the solid is sufficient When the solenoid valve is activated, the 0 7 mm hole is opened and the vapour above the salt, entrained in a suitable quantity of argon, expands through it into the vacuum chamber of the spectrometer In this way a super- sonically expanded jet is produced The timescale of the expan- sion is very short the period between the equilibrium gas mixture entering the nozzle and the collisionless expansion phase in the vacuum chamber being only about lops As a result, the extent of clustering is kept small and little solid is formed The concentration of heterodimers (NH,,HCl) in the expanded gas is, on the other hand, substantial The fast-mixing nozzle consists of the assembly illustrated in Figure 1 It is attached coaxially to the base plate of a solenoid valve A pulse of gas mixture of (e g ) 1Yo trimethylamine in argon from the solenoid valve passes through the outer of the two concentric tubes, as indicated by the arrow Meanwhile, a mixture of (eg ) 30% HBr in argon flows continuously through the inner tube (0 25 mm internal diameter) The two gas flows meet only as they expand into the vacuum chamber At the boundary between the inner and outer components of the concentric flow which constitutes the jet, the heterodimer of CHEMICAL SOCIETY REVIEWS.1993 i Continuous flow of HBr in Arll Pulse of (CH3)3N in Ar from solenoid valve Figure 1 Fast mixing nozzle used to observe the rotational spectra of species such as [(CH,),N,HBr] The (CH,),N/Ar mixture is pulsed from a solenoid valve, onto the bottom of which the fast-mixing nozzle shown is attached The continuous flow of HBriAr from the inner tube meets the pulse of gas from the outer tube in a concentric flow and [(CH,),N,HBr] is formed at the confluence without precipitation (e g ) trimethylamine and HBr is formed in detectable quantity4 while precipitation of the solid is again minimal 2.3 Spectroscopic Constants and their Interpretation: Limiting Models Once observed by the above-described technique, the ground- state rotational spectra of heterodimers in the series (R,-.H,N,HX) can be analysed to give a variety of spectro- scopic constants These constants can be interpreted in turn to give the dimer properties and the nature of the intermolecular interaction can then be diagnosed with the help of models suitable for the two limiting cases, ze the simple hydrogen- bonded dimer and the simple ion-pair Table 1 sets out the spectroscopic phenomena/constants of principal interest in connection with this discussion of the ammonium and methylammonium halides and indicates the heterodimer properties to which they lead The final column of Table 1 gives some comments about models used All but one of the heterodimers considered here exhibit symmetric-top type rotational spectra This observation gives an immediate insight into the heterodimer geometry -it must have at least C, symmetry The rotational constants B, determined from analyses of such spectra are the ground-state values and are related to the distribution of mass (moment of inertia) in the vibrationless state in a complicated, but well understood, man- ner It is customary to use B, values as though they were equilibrium values (the differences are relatively small) to obtain bond lengths and angles (the so-called r,, and r,-values) Follow- ing this convention, B, values and their changes on isotopic substitution can be interpreted in terms of the separation of the (CH,),_.H,N and HX subunits in the heterodimer, if unper-turbed monomer geometries are assumed All bond lengths considered here are of the r,-type The centrifugal distortion constant DJof a symmetric-top heterodimer, which allows for the slight dependence of the molecular geometry on rotational state ignored in the rigid rotor limit, can be interpreted in terms of one measure of the strength of intermolecular binding by using a simple model The monomers are assumed rigid and unchanged in geometry on dimer formation The constant DJdepends in the quadratic approximation on only the intermolecular stretching force constant k, according to8 THE NATURE OF AMMONIUM AND METHYLAMMONIUM HALIDES IN THE VAPOUR PHASE-A C LEGON Table 1 Spectroscopic constants, properties and limiting models of heterodimers [(CH,), -,H,,HX] from rotational spectroscopy Spectroscopic property Heterodimer property Limiting model (1) Form of spectrum Symmetry Symmetric top (A>B=C) Asymmetric top (A > B > C) Semi-rigid rotor in either hydrogen-bond (CH,), ,H,N..*HCI or ion-pair (CH,), ,,H,NH+ *-*X limit Rotational constants A0,BO co Radial and angular geometry r(N*.*X) and relative orientation of Unchanged monomer geometries assumed, orientation (where appropriate) of subunits Centrifugal distortion constant DJ components Intermolecular stretching force constant k, fitted (a) Simple hydrogen- bond model," e g HCN --* HCl, k,=912Nm (b) Ion pair modelh eg Na+.-*Cl , k,= 108 6Nm X or I4N Electric field gradient (a) Simple hydrogen nuclear along principal bond model," e g quadrupole coupling constants x(X) or X( 14N) inertial axis at X or 4N HCN...HCl, X(~~CI) = -53 720(2) MHz (b) Ion-pair model,' e gc1 2 X(,~C~)= -5 643(4) Naf ...MHz C Legon E J Campbell and W H Flygare J Chem Phjs 1982 76 2267 and Ref 11 Calculated using W, = (~xc)l(k/p): from P L Clouser and W Gordy Phis Rev A 1964 134 A863 F H de Leeuw R Van Wdchem dnd A Dymdnus Symposium on Molecular Structure and Spectroscopy Ohio 1969 Abstract R5 where = mBasemHX/(mBase + mHX) and Bo,BBase, BHX are zero- point rotational constants of the heterodimer, the base, and the acid HX, respectively k, is the restoring force per unit infinitesi- mal extension of the intermolecular bond and is a measure of the energy required for unit infinitesimal extension It is an import- ant quantity here, for it changes by an order of magnitude between a typical hydrogen-bonded heterodimer (e g HCN * * * HCl) and a typical ion pair (e g Na + * C1 ) (see Table 1) The variation of k, along selected series [(CH,), -,H,N, HX], to be discussed below, provides a criterion of hydrogen-bond/ ion-pair character The spectroscopic quanti ties/phenomena that are perhaps most revealing of the nature of [(CH,),-.H,N,HX] in the vapour are the halogen and 4N-nuclear quadrupole coupling constants and the Stark effect The latter is the splitting induced in rotational transitions by a uniform applied electric field and can be analysed to give the electric dipole moment of the molecule This effect will not be used here Nuclear quadrupole hyperfine structure in rotational transitions arises from the coupling of a nuclear spin vector I with the rotational angular momentum vector J through the interaction of the electric quadrupole moment Q of the nucleus in question with the electric field gradient at that nucleus The number of discrete relative orientations of Z and J is limited to that group for which the magnitude of the resultant I+ J= F is given by {F(F+ l)h2)*, where Each orientation corresponds to a slightly different potential energy of interaction of Q with the electric field gradient and hence each rotational energy level, labelled by J, is split into 21 + 1 components (if J > I), labelled by F Rotational transi- tions exhibit a corresponding hyperfine structure, the pattern being determined by the Fand Jselection rules For a symmetric top molecule carrying on its axis a quadrupolar nucleus X, the associated observable spectroscopic quantity is the nuclear quadrupole coupling constant x(X) This is linearly related to the electric field gradient Vzz= -d2 V/dz2 at X along the molecular symmetry axis z by x(X) = -eV,Q (3) where e is the magnitude of the protonic charge As x(X) increases, the hyperfine splitting in a given transition increases Since VV7at X arises from the particular distribution of electrons and nuclei in the molecule, the quadrupolar nucleus, through equation 3, provides a probe of the electric charge distribution in the molecule, once x(X) is known Limiting values of x(X) in the free monomers and in molecules chosen to model simple hydrogen-bonded and ion-pair hetero- dimers will be important in what follows As an example, x(, 5Cl) will be consideredg but similar arguments apply to 14N, Br, and I nuclei and will be used later An isolated 35Cl- ion with the electronic configuration [K L3s23p6] is spherically symmetric and therefore Vzz= 0 and ~(~~cl) = 0 An isolated atom Tl[K L3s23ps] is generated when an electron is removed from a 3p orbital and the resulting V__leads to ~(~~cl) = -109 74 MHz It is convenient to describe an isolated hydrogen chloride molecule through the valence-bond structures H-Cl and H + C1 The former has a 3pf electron deficiency, like the C1 atom, while the latter has X(,~C~)= 0 The observed value', x("5Cl) = -67 62 MHz for gaseous HCl can then be readily understood through weighting the contributions of the valence bond structures When an axially symmetric molecule, like HCN, is brought up to HCl along the z-axis to form a simple hydrogen- bonded dimer HCN HCl (no proton transfer), x(,Tl) changes slightly (see Table 1) because of the changed electric field gradient at C1 due to the HCN electric charge distribution and because of the additional zero-point motion available to HCI in the dimer These contributions to X(,~CI) have been modelled (see Section 3 2 below) If the heterodimer (CH,), -,H,NH C1- were an ion pair, + * it could be viewed as arising when the appropriate methylammo- nium ion is brought up to an isolated C1- ion The quantity x(,Tl) = 0 appropriate to C1- in isolation then increases some- what in magnitude as a result of the distortion of the spherically symmetric charge distribution of the anion by the cation A suitable model for such an ion pair is an alkali chloride diatomic molecule in the gas phase For example, ~(~~cl) is only -5 643(4) MHz in Na+ .-.C1- (See Table 1) 3 The Heterodimer (NH,, HCI): A Case Study The ground-state rotational spectra of the most abundant isotopomer (l4NH3, H35Cl) of the ammonia-hydrogen chloride dimer and those obtained by single isotopic substitution at each different atom have been measured in the vapour above solid ammonium chloride The heated nozzle discussed above was used in the pulsed-nozzle FT microwave spectrometer in this case The set of spectroscopic constants determined from the spectra are recorded in Table 2 They can be interpreted, first qualitatively and then quantitatively, to establish the nature of the heterodimer (NH,,HCl) in the gas phase 3.1 Qualitative Interpretation of Spectroscopic Constants The rotational spectrum of (NH,,HCl) is of the symmetric-top type The only way in which HCl can be bound to NH, to achieve this result is in a heterodimer of C3" symmetry More- over, the only geometry of this symmetry that is consistent with the observed changes in the rotational constant B, (which is proportional to the moment of inertia Ib through B, = h/8r2Ib) CHEMICAL SOCIETY REVIEWS, 1993 Table 2 Observed ground-state spectroscopic constants of isotopomers of (NH,,HCl) in ammonium chloride vap0ur.O Is0 topomer B,/MHz Dj/kHz DJK/kHZ X(Cl)/MHZ x( 14N)/MHz ( 4NH3,H3 T1) 4243.2593( 16) 12.8(2) 371S(8) -47.607(9) -3.248(14) (l 5NH3,H35C1) 4098.3113( 12) 11.6(2) 344.2(5) -47.614(5) -(14NH,D,H3SCl)b 4033.8388( 16) 11.4(2) --47.481(9) -3.312(16) (14NH3,D3 T1) 4228.932(I) 12.6‘ --48.630(16) -3.27(2) ( 4NH3 ,H ’C1) 4 168.8 107(9) 12.0(1) 357.7(6) -37.531(6) -3.264(10) From Ref.3. Asymmetric rotor species. Value in B, column is (B, + C,)/2.(1 is the one3 in which the nuclei lie in the order H,N.**HCl.But the position of the hydrogen-bond proton along the C, axis in the equilibrium geometry is uncertain and each of the three models of C,, symmetry shown in Figure 2 is consistent with the observed rotational constants. The reason for the uncertainty in the position of this proton lies in its proximity to the dimer centre of mass. Then D-substitution leads to only a small change A4 in the equilibrium moment of inertia while the change in the zero- point motion attending this substitution tends to make dlb smaller than A& In any case, A4 depends on zk and hence we cannot tell on which side of the centre of mass the proton lies. How then can we discriminate between the three models (ion- pair, intermediate-type, and hydrogen-bonded) shown in Figure 2? *.*.*o-@ 6-) $”.....0.....@ .....@ Figure 2 Three possible structures of C3” symmetry for the (NH,.HCI) dimer in which the nuclei lie in the order H3N*.-HCI; namely, the hydrogen-bonded form, a form with partial proton transfer, and a form with complete proton transfer. A qualitative distinction is immediately possible from the magnitudes of X(~~CI) and k, (as determined from DJby the method outlined in Section 2.3). Table 3 lists X(~~CI) and k, for H,N.--HCl and a selection of model systems, namely the isolated HCl molecule, a typical hydrogen-bonded dimer HCN*-.HCI, and a typical ion pair Na+ .**Cl-. For H,N * * HC1, each of these quantities is closer to the correspond- ing value for HCN**.HCl than that of Na+ .*.C1-.Qualitati- vely, at least, we conclude that the simple hydrogen-bonded model H,N*.*HCl is more appropriate. Are the numerical values of X(,~C~) and k, also quantitatively consistent with this model? 3.2 Quantitative Interpretation of~(~~cl)and k, It can be shown that both x(”C1) and k, are quantitatively as expected for the hydrogen-bonded model H,N*.* HCl. Table 3 Comparison of C1 nuclear quadrupole coupling constants x(C1) and intermolecular stretching force constants k, of (NH,,HCl) and various model Assumed value, see Ref. 3 for justification. A consideration of the k, for an extended series of hetero- dimers B *** HX has revealed that the experimental values can be reproduced by a simple empirical equation k,=cEN (4) where c = 0.25 Nm-and E and N are numerical electrophili- cities and nucleophilicities assigned to the donor region of HX and the acceptor region of B, respectively.” If E for HF is chosen as 10, equation 4gives the set of N values for the series of B indicated in Figure 3 by using the k, of the corresponding B.*.HF.The plot ofk,versus Nfor the series B***HFis then by definition a straight line. When k, for the analogous series B * HCl is plotted against the same N values (also in Figure 3), the result is also a straight line through the origin. In this case, we note that the point for H,N-.*HCl lies on the straight line. Thus, k, for H,N-*.HCI is well behaved and has the value expected by extrapolation from the more weakly bound members of the series.But it can be shown that in the series Be.* HF there is only tiny HF bond lengthening on dimer12>13 formation. Hence, it appears that the above arguments indicate only negligible extension of HCI and therefore no significant extent of proton transfer in H,N*.*HCl. On the other hand, while k, for (CH,),N-..HF lies on its straight line 13.14 in Figure 3, that of (CH,),N-.*HCI certainly does not* (see Section 4.2). To show that x(, Tl) for H,N HC1 is quantitatively pre- dicted by using the simple hydrogen-bonded model of the dimer, 40-35-30-7 25-€ z bb20-;t 15-10-2 4 G 8 10 12 14 16 Nucleophilicity N of B Figure 3 Systematic relationship between the intermolecular stretching force constant k, of dimers Bas-HF and B*-*HCl and the nucleophili- city N of the donor region of B (see text for discussion).The N values of B are chosen so that the k, for B-.*HF lie on a straight line. The k, for B**-HCI are then also a linear function of the N values. The extrapolated value of k, for the hydrogen-bonded model of (CH,),N-.*HCI is indicated by the arrow. sys tems Molecule HCl HC14N-.*HC1 (l4NH3,HC1) Na+ ..*Cl- c1-Reason for choice ,Y(~TI)/MHz k,/Nrn-l Isolated component -67.6189“ -Hydrogen-bond model -53.720(2)h 9.126 Test Ion-pair model Free anion Ref. 10. See footnote a of Table 1. 1. e See footnote b of Table 1. -47.607(9)‘ 17.6(3)‘ -5.643(4)d 108.6‘ -0.0 c Ref.3. d See footnote c of Table THE NATURE OF AMMONIUM AND METHYLAMMONIUM HALIDES IN THE VAPOUR PHASE-A. C. LEGON we must calculate the change d L$! in the electric field gradient at C1 when NH, is brought up from infinite distance to its equilibrium position along z. We consider first the equilibrium (vibrationless) C,, geometry. The expression for d l"$; is then l6 In equation 5, F,, F,,, etc. are the electric field, the gradient of the electric field, etc. due to NH, but at the position along the symmetry axis z occupied by the C1 nucleus. (The convention V,, for an intrinsic field gradient and Fz,for that external to a group of charges is in use.) The terms gz',,, gz,,,,, etc. are the axial components of the HCl response tensors.Thus gz,',=is the electric field gradient at C1 along z induced by a unit external electric field, gZ,,= measures the additional electric field gradient induced at C1 in response to a unit electric field gradient along z,and so on. These g-tensor components have been calculated ab initio for HCl by Baker et a1.16 The values of F,, F,,, etc. have been estimated at the appropriate distance (see Section 3.3 below) from NH, along z by Fowler17 using a distributed multipole analysis to represent the electric charge distribution of NH,. Application of equation 5then leads to a value of d Ejwhich corresponds to a correction d~~(~~Cl) = -ed F'$i Q z 13 MHz to the free HCl coupling constant x~(~~C~) = -67.62 MHz. Hence, the equilibrium C1-nuclear quadrupole coupling constant for the hydrogen-bonded model of H3N**.HC1 is predicted to be X~(,~CI) -55 MHz, which is close to the observed ground- z state value x("C1) = -47.607(9) MHz.Even closer agreement between the model value X,(~~CI) and the experimental value x(,Tl) is obtained when the effects of zero-point averaging are added to the model. The value of the C1-nuclear quadrupole coupling constant in the zero-point state is given according to the model by where EA and LIE;are now the instantaneous values of the electric field gradient at C1 along the a-axis of the dimer appropriate to unperturbed HCI and induced by NH,, respecti- vely. The average is over the zero-point motion of H,N * HCl.The first term in equation 6 is simply (E;)= +(3cos20 -1) Ed = (P2(cos0)) F'$i,where 0 is the angle between the HCI direction and the instantaneous a-axis.The second term is much more complicated, as can be appreciated by examining the model tor the zero-point motion ofthe dimer shown in Figure 4. The distance between the subunit mass centres is assumed fixed while the rigid subunits execute angular oscillations + and 0 defined with reference to the instantaneous a-axis. Clearly, the NH, oscillation 4 affects F,, FZz,etc., while each g-tensor term contributing to A EA has an appropriate coefficient (P,(cos0)). In the absence of the detailed term by term corrections, it is an acceptable approximation to assume that (A V,,) x (P2(cos0))LIE:,where is the equilibrium value.A detailed discus- sion3 indicates that cos-(cos20)f = 15" is reasonable for H,N.-.HCI and hence equation 6 leads to the prediction of Figure 4 Definition of the angles 4, 8, a, /3, and the distance Y,, used to discuss the geometry and the interpretation of the nuclear quadrupole coupling constants of (NH,,HCl). n x(,sC1) z -49 MHz for the simple hydrogen-bonded model of H,N*..HCI. This value is close enough to the observed ground- state quantity to give confidence that the hydrogen-bonded model is quantitatively capable of accounting for the experimen- tal value of x(,sC1). Similar arguments to those above could be made to show that the decrease in magnitude of xo(14N) = -4.090 MHz in free 14NH3 to x(14N) = -3.248(14) MHz in H,14N-**HC1 (see Table 2) is also consistent with the simple hydrogen-bonded modeL3 Unfortunately, however, the response tensor compo- nents ggy3, etc.,are not available and quantitative comparison is not yet possible.3.3 The Geometry of H,N -HCI The arguments in Sections 3.1 and 3.2 establish that the hetero- dimer in ammonium chloride vapour is a simple hydrogen- bonded molecule H,N * -* HCl of C3" symmetry. In particular, it is unnecessary in interpreting the various spectroscopic con- stants to invoke any significant extent of proton transfer from HC1 to NH,. These conclusions are consistent with matrix isolation studies1 * and recent ab initio calculations. Having established the angular geometry H,N HCI, the radial geometry is available from the B, values under the assumption that the geometries of NH, and HCl survive dimer formation.The model used (see Figure 4) attempts to account for the contribution of the intermolecular bending modes to the zero-point motion by allowing the NH, and HC1 subunits to execute the angular oscillations a and 6 about their respective mass centres, the distance between which is fixed. It is readily shown that where rcm,a, and are defined in Figure 4 and cH3and #'' refer to the free monomers. In the approximation that the observed quantity for the heterodimer Ib = h/8& B, can be used in place of (Ibb), equation 7 provides a route to (r:m)+. To a high degree of approximation (COS2a) and (cosz/3) can be taken as equal to (cos2+) and (cos28), respectively, and the last two quantities have been established from x(14N) and ~(~~cl), respectively.The values of (T:~)+ for each of the symmetric-top isotopomers are recorded in Table 4. Once (r:m)j is available, the known geometries of NH, and HC1 allow r(N-*.Cl) to be calculated, and these too are given in Table 4. The modelling of X(,~CI) discussed in Section 3.2 employed the radial geometry displayed in Table 4. 4 Does Proton Transfer Occur in Gas-phase Ammonium or Methylammonium Halides? The conclusion of Section 3 is unambiguous. The lowest energy form of the heterodimer in the vapour of the prototype ammo- nium halide has the simple hydrogen-bonded structure H3N.*.HC1. Furthermore, no evidence was found for an ion- pair form H,NH + * -* C1- existing at a minimum of comparable potential energy. This raises an important question: Is it possible to modify H,N*.*HCl chemically so that the ion-pair becomes the more stable, and possibly the only, form? A chemist would Table 4 (r,",)* and r(N -a*C1) for symmetric-top isotopomers of H,N*..HCla Isotopomer <r:ln>% r(N * -* Cl)/A H, 14N* 0 H3 H, 5N*--H3 H, I4N--.D3T1 H, I4N*** H37C1 3 1654(2) 3 1614(2) 3 1367(2) 3 1673(2) 3 1364(7) 3 1358(7) 3 1410(11) 3 1363(6) For a detailed discussion of the values of (cos2a)and (cos2/3) used with equation 7 to evaluate these distances, see Ref 3 The errors in (r&,)* and r(N -* * CI) are those arising from the assumed error of f3" in aav= cos '(cos'~): = 15" attempt to answer this question by reference to the group of ammonium and methylammonium halides in Figure 5 It is known that the energy required to dissociate HX into H + and X- in the gas phase decreases along the series X = F, C1, Br, I Hence, the proton is more likely to be transferred from HX to (CH,), -,H,N as we move down a vertical column of Figure 5 On the other hand, the gas-phase proton affinity of ammonia is increased progressively with the stepwise methylation of the base Proton transfer might therefore be encouraged progressi- vely along the horizontal series (CH,), -nH,N.-* HCl as n decreases from 3 to 0 The bottom right-hand corner of Figure 5 therefore favours ion-pair structures while the top left should lead to hydrogen-bonded dimers Figure5 Series ofdimers (CH,), ,,H,N-*.HX investigated by rotation- al spectroscopy In the left-hand vertical series, the proton affinity of the base NH, is fixed while the ease of dissociation of the hydrogen halide increases from F to I In the central horizontal series, the proton affinity of the base increases progressively from left to right, while In the right-hand vertical series the proton affinity of the base is constant but has its maximum value All of the heterodimers identified in Figure 5 have been investigated through their rotational spectra, all but one by the pulsed-nozzle FT technique By consideration of the spectro- CHEMICAL SOCIETY REVIEWS, 1993 scopic and molecular properties thereby available, especially x(X) and k,, which featured predominantly in the discussion of ammonium chloride, the question asked above can be answered It is convenient first to consider the left-hand vertical series in Figure 5 to test the effect of weakening the HX bond while leaving the proton affinity of the base unchanged The horizon- tal series (CH,), -,H,N * * * HCl (n= 3,2,0) then allows the effect of methylation of ammonia to be investigated for a fixed HX Finally, the right-hand vertical series illustrates the effect of weakening the HX bond when the proton affinity of the base is maximized In the discussion that follows, we must bear in mind that spectroscopic techniques employing supersonic jets tend to detect only the vibrational ground-state of the lowest energy conformer of a molecule Hence, the conclusions drawn below pertain to the lowest temperature form of (CH,), -,H,N * -HX We cannot rule out a higher energy minimum on the potential surface since this would probably be depopulated in the expan- sion As we shall see, however, there are reasons for believing that in general only one stable form (either hydrogen-bonded or ion pair) exists 4.1 The Series H,N HX (X = C1, Br, I) The ground-state rotational spectrum of each member of the series H,N...HX (X = C1, Br, I) has been observed by the pulsed-nozzle FT microwave method and spectroscopic con- stants determined ,2o 21 Again, we focus attention on the critical quantities x(X) and k,, listed in Table 5, as the means to establish the nature of the lowest energy forms of the hetero- dimers Also given in Table 5 are the corresponding quantities for free HX, for HCN**.HX (the limiting hydrogen-bond model) and for Na + --* X -(the limiting ion-pair model) It is immediately evident that x(X) and k, are much closer in magnitude to their respective values in HCN.-*HX than in Na+ -*-X- We note also that the ratio x(X)/xHB(X) is almost constant across the series X = C1, Br, and I, as is the ratio k,/ kUHB Although a quantitative prediction of x("Cl) for the hydrogen-bonded model of H,N * -HCl was possible and gave good agreement with the observed value (see Section 3 2) such an approach is not possible for X = Br or I because the required response tensor components g:z -* are not available for HBr and HI Nevertheless, in view of the result for H,N.--HCl and the constancy of the ratios x(X)/xHB(X) and k,/k!B it is very likely that both H,N***HBr and H,N**.HI are simple hydro- gen-bonded dimers H,N * -* HF has been investigated using the molecular beam electric resonance technique 22 The DJvalue leads, via equation 1, to k, = 32 8 Nm-l which compares with the limiting values8 23 of 18 2 and 176 1 Nm-' for HCN..*HF and Na + * F-, respectively Again the hydrogen-bond limit seems appropriate Another quantity that indicates a simple hydrogen bond in H,N HX is the 14N nuclear quadrupole coupling constant 0..which has the values3 20-22 x(14N) = -3 283 MHz for X = F, Table 5 Comparison of halogen nuclear quadrupole coupling constants x(X) and intermolecular stretching force constants k, for heterodimers H,N--*HX with those of model systems Molecule 35c1 81Br 1271 X(X)IMHZ k,/Nm-' x(X)/MHz kJNm x(X)IMHz kJNm HX -67 6189" -444 681d - -1823 4' - HCi4N.--HX -53 720d 9 12d 356 232(9)' 8 I' -1475 7(1)f 4 561(2y H,i4N-.*HX -47 607(9)g 17 6(3)g 301 777h Na+...X--5 643(4)' 108 6k 48 508l 13 4(3)h93 7" -1324 891' -262 14" 7 18(9)' 77 0" Ref 10 * 0 B Dabbousi W L Meerts F H Deleeuw and A Dymanus Chem Phqs 1973 2 473 F C DeLucia P Helminger and W Gordy Phis Rev0 fA 1971 3 1849 d See footnote a of Table 1 e E J Campbell A C Legon and W H Flygare J Chem Phqs 1983 78 3494 and Ref I1 P W Fowler A C Legon, and SPeebles, unpublished observations R Ref 3 Ref 20 Ref 21 See footnote L of Table 1 See footnote b of Table 1 J Cederberg D Nitz A Kolan T Rasmussen K Hoffman and S Tufte Symposium on Molecular Structure and Spectroscopy Ohio 1985 Abstract MF6 Calculated using (27~~)(k/~)ifrom J R Rusk and W Gordy Phis Rev A 1962 127 817 C E Miller and J C Zorn J Chem Phjs 1969 50 3748W, = THE NATURE OF AMMONIUM AND METHYLAMMONIUM HALIDES IN THE VAPOUR PHASE-A.C. LEGON -3.248(14) MHz for X = C1, -3.188(8) MHz for X = Br, and -3.182(8) MHz for X = I. The constancy of these values and their small reduction in magnitude from the free NH, value ,yo( 14N)= -4.090 MHzpoints to a similar type of interaction in all three heterodimers involving only a weak perturbation of the electric field gradient at 14N in NH,.Recent ab initio calculations predict a simple hydrogen- bonded rather than an ion-pair form for H,N * -* HCl and .~~H,N.**HBr. Latajka et ~1find a long shallow minimum linking the ion-pair and hydrogen-bond forms H,NH I-+ and H,N***HI, both of which have similar energy. It is of interest to note that so far only J + 1tJ, K = 0 transitions have been fitted2' in the rotational spectrum of H,N.*.HI, with the K = 1 set oddly behaved. The odd behaviour could arise from the effect of a second minimum. 4.2 The Series (CH3),-,, H,N.-.HCI (n = 3,2, and 0) The experimentally determined quantities x(, Tl) and k, for each member of the series (CH,), -,H,N...HCl (n = 3,2,0) are compared in Table 6.,9' s,2 Also included are the corresponding quantities for the limiting hydrogen-bonded model HCN HCl and the limiting ion-pair model Na + -* C1-.Table 6 Comparison of CI nuclear quadrupole coupling constants X( sCl) and intermolecular stretching force constants k, for [(CH,), -,,H,N,HCl] with those of model systems Molecule x(~'CI)/MHZ k,/Nm --HC1 -67.6189" HCN ..*HCl -53.720' 9.12h H, 14N * *- HCl -47.607(9)' 17.6(3)'CH,NH, --.HCl -37.89(1)d -(CH,),N * -* HCI -21.625(5)' 84(3)' Na + *.-Cl--5.643(4Y 108.68 0 Ref. 10. h See footnote a of Table 1. Ref. 3. Ref. 25. p Ref. 15. f See footnote L' of Table I. g See footnote b of Table 1. The X(,~CI) values in Table 6 exhibit clearly a stepwise decrease in magnitude as NH, is progressively methylated, with an approximate decrement of 9 MHz per methyl group.There is a corresponding increase in k,, although no value of this quantity is available from the appropriate centrifugal distortion constant of CH,NH,..*HCl because the dimer does not have axial symmetry.25 Both trends are consistent with an increase in the extent of proton transfer from HCl to (CH,),-,H,N as n decreases. For n = 0, (CH,),N*..HCl, both x("Cl) and k, approach the values expected in the ion-pair limit, i.e. the Na+ .-.C1- values. Indeed, k, is so large in this case that the question of the validity of its determination from DJ using equation 1 must be kept in mind. A detailed analysis, reproduced elsewhere,26 predicts values of = -47.7 MHz and k, w 20 Nm-' for the hydrogen-bonded model (CH,),N*..HCI .The latter is four times smaller than the observed value and was obtained by extrapolating the k, versus Nline for B * * HCI in Figure 3 to the Nvalue for trimethylamine. Clearly, the ion-pair description (CH,), -,H,NH *+ C1-is more appropriate. However, an attempt to calculate x,(~~C~) by +starting from the ion-pair model (CH,),NH * C1- and using the charge distribution of (CH,),NH and the response tensors + of C1- ran into convergence problems.26 Hence, no quantitative conclusion about the relative contributions of (CH,),N HC1*a* and (CH,),NH+ ...Cl- to a valence bond description of the molecule is yet available. Nevertheless the conclusion of this section is clear: progressive methylation of H,N..- HCl leads in the limit to a heterodimer for which the simple hydrogen-bond description alone is apparently inadequate and for which a substantial contribution from (CH,),NH + .*.Cl-must be invoked.4.3 The Series (CH,),N -HX (X = F, C1, Br, I) The conclusion of Section 4.2 encourages us to examine the series of trimethylammonium halides shown in the extreme right hand column of Figure 5. Of the generalized heterodimers (CH,),-,H,N-..HX, the series with n = 0 is one in which the proton affinity of the fully methylated base is maintained at the maximum value while the proton affinity of X- decreases in the order F > C1 > Br > I. Of the systems considered in this article, the propensity to form an ion pair (CH,),-,H,NH+...X- should therefore be greatest for n = 0 and X = I.The important spectroscopic constants for the symmetric-top species n = 0, X = C1, Br, and I are collected in Table 7.4713- 5,27 Included for comparison are those for the limiting hydrogen-bond and ion- pair models (HCN * * * HX and Na+ * -* X-,respectively) and the series H,N-*.HX. We consider first and separately the dimer (CH,),N *..HF as a limiting example of a hydrogen-bonded complex where the base is very strong but negligible proton transfer occurs. It has been possible in this case to determine the position of the hydrogen bond proton pre~ise1y.l~ This is available from the H,F spin-spin coupling constant 0:: of (CH,),' 5N*.*HF which has been measured.The separation r of the H and F nuclei in the heterodimer is related to D?: through the expression. cc (r-3)(P2(cos8)) where the constant of proportionality involves the H and F nuclear magnetic moments and various universal constants and the term (P2(cos0)) = +(3cos28 -1) accounts for the angular oscillation of the HF subunit as defined in Figure 4. In the term (r-,), the average is over the HF stretching motion but with respect to the changed equilibrium length in the complex. Equation 8 therefore allows the definition ro + 6r = (r-,)-f of an operational HF bond length in the dimer. Clearly, the free HF bond length ro = (r-,)-4 can be similarly defined and is available from the known spin-spin coupling constant ofF.The result thereby obtained', for 6r in (CH,),N-.-HF is 0.041(11)A.The value of 8," = cos- 1(~os28)* used to extract ro + 6rfrom equation 8 is 14( I)" and is discussed in detail elsewhere. Even for this very strong hydrogen-bonded system (k, = 38.6 Nm- I), we conclude that there is only a small lengthening (-5% of the equilibrium value) of the HF bond when incorporated in the dimer. This is perhaps not unexpected since the HF bond is the most difficult to extend (k, = 966 Nm- I) of all single bonds. The discussion of Section4.2 has already dealt with the second member of the series, namely (CH,),N-*-HCI. Unlike the HF analogue, there is evidence of an appreciable extent of proton transfer as a result of weakening the HX bond.An examination of both the x(X) and the k,values in Table 7 reveals that a further weakening of HX increases this effect from X = C1 through X = Br to X = I. In fact, it is of interest to estimate crudely the fractional extent of proton transfer brought about by the complete methylation of H,N.*.HX to give (CH,),N-.-HX. This can be measured by (9) where x(A) is the halogen nuclear quadrupole coupling constant of the ammonium halide, x(T) refers to the trimethylammonium +halide, and x(1P) to the model ion pair Na ***X-.This formula assumes no proton transfer for the ammonium halide. The results calculated from Table 7 aref = 0.62, 0.80, and 0.93 for (CH,),N-..HX, X = C1, Br, and I, respectively. The above result is confirmed when the 14N nuclear quadru- pole coupling constants of the two series H,N*..HX and (CH,),N-*.HX (X = F, C1, Br, I) recorded in Table 8 are considered.Qualitatively, X( 4N) is effectively constant along the H,N HX series while its magnitude decreases by one half along the (CH,),N.**HX series. Unfortunately, x(I4N) for the trimethylammonium ion is not experimentally available but clearly -2.45 MHz is an upper limit to this quantity if CHEMICAL SOCIETY REVIEWS, 1993 Table 7 Comparison of halogen nuclear quadrupole coupling constants x(X) and intermolecular stretching force constants k, for [(CH,),N,HX] with those of model systemsa Molecule 35c1 81Br 1271 X(X)MHZ k,/Nm X(X)/MHZ k,/Nm X(X)/MHZ k,/Nm HCN.-*HX -53 720 9 12 H,I4N.* -HX -47 607(9) 17 6(3) (CH,), 14N * + HX -21 625(5)b 84 (3)h Na+ ...X -5 643(4) 108 6 356 232(9) 81 301 777 13 4(3) 99 645(7)' 82(3p 48 508 93 7 -1475 7(1) -1324 891(8) -341 204(14)d -262 14 4 561(2) 7 18(9) 66 5(2)d 77 0 Values for HCN HX H,N HX and Na+ X taken from Table 5 h Ref 15 Ref 4 d Ref 27 Table 8 A comparison of x(14N)/MHz among the ammonium and trimethylammonium halides H,N*-.HX (CH,),N*-*HX Free base -4 090" -5 502(3)' X=F -3 283' -4 764(3)dc1 -3 248(14)' -3 504(5)f Br -3 188(8>g -2 883(7)h I -3 182(8)' -2 451(8)1 0 M D Marshall and J S Muenter J Mol Spectrosr , 1971 39, 94 C A Rego R C Batten, and A C Legon J Chem Phys 1988 89 696 Ref 22 d Ref 14 Ref 3 f Ref 15 g Ref 20 h Ref 4 Ref 21 Ref 27 J (CH,),N HI is assumed to be wholly an ion pair If so, and if (CH,),N**.HF can be taken as the hydrogen-bond limit, then the fractional ionic character can also be defined by The results are 0 54,O 82, and 1 00 for (CH,),N...HX, X = C1, Br, and I, respectively This approach is obviously crude but the agreement with the values determined from the halogen cou- pling constants is acceptable It is of interest to compare the distances r(N*.*X) in the two series H,N**.HX and (CH,),N**.HX There is, however, a difficulty in applying the model of Section 3 3 and equation 7 to (CH,),N-..HBr and (CH,),N**.HI in which there is a prepon- derant contribution of the ion-pair valence bond structure to the description of the heterodimer Briefly, this lies in the structure of the ion (CH,),NH +,which is assumed unchanged on dimer formation An experimental geometry for free (CH3),NH is+ not available and this has been modelled by takin that of (CH,),N but with a proton added at a distance of 1 03xfrom N along the C, axis In applying equation 7, the oscillation angle pay= cos-(cos2/3)$ (see Figure 4 for definition) is clearly zero for X-, as is Ib, in the ion-pair limit Arguments given elsewhere lead to the assumption of aaV= COS-~(COS~~)*z lO(2)" for the +oscillation of the (CH,),NH subunit The heterodimers (CH,),N-.*HX, X = F and C1, and all of the H,N...HX have been analysed using the hydrogen-bond model, the detailed assumptions about a,, and payin each case being given in the relevant reference In fact the results for r(N-.*X), which are listed in Table 9, are not sensitive to which model is used when the above set of assumptions is made, as may be seen in (CH,),N -* HBr and (CH,),N * HI for which the results for both limits are given It is clear from Table 9 that there is a general shortening of r(N...X) when NH, becomes fully methylated 5 Are the Conclusions for (CH,),-,H,N ...HX Consistent with Simple Energetic Conside ratio ns? Arguments above based on x(X), x(14N), and k, for the series (CH,), -,H,N HX indicate that progressive methylation of Table 9 Comparison of the distances r(N X) in the series (CH,),N.*.HX and H,N-.*HX x H,N .* * HX r(N -* -X)/A Modela (CH, ,N*-.HX r(N -* .X)/d Model" Fc1 Br 2 716 3 137d 3 255f HB HB HB 2 5863(6)c 2 8164(3)' 2 9607g HB HB HB 2 95949 IP I 3 584h HB 3 190(2)' 3 190(3)' HB IP (1 HB = hydrogen bond model IP = ion pair model See text for cdiscussion 6 Ref 22 Ref 14 d Ref 3 e Ref 15 f Ref 20 K Ref 4 Calculated from data in Ref 21 Ref 27 NH, coupled with the progressive weakening of the HX bond with respect to the process HX = H+ + X- eventually leads to an ion pair in the gas phase when n = 0 and X = Br or I In the series when n = 0, X = F remains a simple hydrogen-bonded dimer while X = C1 is of the intermediate type The series H,N-**HX appears to exhibit no appreciable extent of proton transfer for any X Are these conclusions consistent with simple energetics? To answer this question, we shall examine the energy of the general process (CH,), ,H,N*-.HX = (CH,), ,H,NH+ .-ex (1 1) in which a proton is transferred from HX to the base in the isolated hydrogen-bonded dimer to give the isolated ion pair For simplicity, we shall assume that no significant change in the positions of other nuclei accompanies reaction 11 Is LIE,, negative for n = 0, X = Br and I, and positive in other cases? To find LIE, ,,we note that reaction 11 can be written as the sum of the following (CH,), ,,H,N*..HX = (CH,), ,H,N + HX (12) (CH,), ,H,N + H+ = (CH,), ,H,,NH+ (13) HX=H++X (14) (CH,), ,H,NH+ + X = (CH,), ,H,NH+ --.X (15) The required energy change is then LIE, = C,!2 ,LIE, Values of the various LIE, are listed in Table 10 and have been obtained as follows LIE,, is the negative of the gas-phase proton affinity of the base (CH,), -,H,N and values are readily available 28 LIE,, is given by the sum of the ionization potential of the H atom and the zero-point dissociation energy of HX minus the electron affinity of X, all of which are well known dE, ,is the electro- static energy gained when the ions (CH,), -,H,NH + and X-are brought to the appropriate distance r(N -- - X) from infinite separation It is assumed in calculating LIE,, that the cationic charge is located on N, that the repulsive contribution is negligible, and that any additional hydrogen-bond interaction N+ -H.*.X is sufficiently independent of X to allow its neglect when comparing relative magnitudes of LIE,, The values of r(N-..X) used are those given in Table 9 The final THE NATURE OF AMMONIUM AND METHYLAMMONIUM HALIDES IN THE VAPOUR PHASE-A C LEGON Table 10 Estimates of LIE, , for the reaction (CH,), -.H,N -* HX = (CH,), -.H,NH *+ X -, when n = 3 and 0 n=3 F 78 -858 1554 -513 261 c1 42 -858 1391 -443 132 Br 32 -858 1350 -427 97 I 17 -858 1312 -388 83 n=O F 117 -946 1554 -537 188 c1 55 -946 1391 -493 7 Br 49 -946 1350 -469 -16 I 37 -946 1312 -436 -33 See text for method of estimating AE,, the hydrogen bond dissociation energy AE,, is the negative of the gas-phase proton affinity of the base and vdlues are taken from Ref 28 but are scaled to the value recommended for NH, byC R Moyldndnd J I Brauman Annu Rev Phbs Chem 1983 34 187 AE,, is the dissociation energy for HX = H+ + X It is the sum of the ionization potential of H and the zero point dissociation energy for HX = H + X minus the electron affinity of X Values from P W Atkins Physicdl Chemistry Fourth Edition Oxford University Press Oxford 1990 Coulombic energy of the ion pair (CH,), H NH+ X when the positive charge is dssumed to reside on the N atom 6E,, = CIS12 AE (see text) quantity LIE, ,is the hydrogen-bond dissociation energy and is not experimentally available Latajka et a1 l9 have calculated A El ab initio for the hydrogen-bondedmodels H3N HCl and (CH,),N.--HCl For the remainder of the dimers in Table 10, LIE,,has been estimated from these values by assuming that it is proportional to k,, which is another measure of the strength of the hydrogen bond There is some evidence to support this assumption 29 The k, for n = 0, X = Br and I (which have significant ion-pair character) were estimated in the hydrogen bondlimit using N = 15 4 for (CH,),N from reference 13 and the appropriate Evalues for HX (10,5 0,4 25, and 3 2 for X = F, C1, Br, and 1)11 in equation 4 The relative errors incurred in this approximate procedure are not large and in any case LIE,, makes the smallest contribution to dE,, Table 10 shows clearly that LIE,, for the proton transfer process is large and positive for all H,N...HX, in agreement with our conclusion that all are simple hydrogen-bonded dimers On the other hand, for the series (CH,),N..-HX (X = F, C1, Br, I), LIE,, becomes progressively smaller and changes in sign between X = Cl and X = Br This pattern is in good qualitative agreement with the experimental conclusion described above, namely that for X = F the heterodimer con- tains a simple hydrogen bond, X = C1 is of intermediate char- acter while for X = Br and I the ion-pair description is more appropriate 6 What Happens when P Replaces N in [(CH.),-,H,N, HX]? The phosphorus analogues of the ammonium and methylam- monium halides are well known The solid phases of the phosphonium halides consist of ions but in the vapour they are described as being completely dissociated into PH, and HX Presumably, the same is true of the trimethylphosphonium halides, for example, and in the context of the discussion here a question of interest is How do the heterodimers [H,P,HX] and [(CH,),P,HX] differ from their nitrogen analogues in the vapour phase? Several members of these two P-containing series have been investigateds 30 34 through their rotational spectra in the manner described in Section 2 2 The aim was to determine the halogen nuclear quadrupole coupling constants x(X) and the intermolecular stretching force constants k, and use them as criteria of the nature of the interaction The experimental quantities are summarized in Table 11 It is immediately evident from Table 11 that heterodimers in the vapour phase of each of the phosphonium halides are of the simple hydrogen-bonded type H3P-..HX, the order of the nuclei having been established through isotopic substitution Thus, we note that the x(X) are very similar in sign and magnitude to those of the corresponding ammonium halides and the simple model dimers HCN HX (see Tables 5 and 11) The k, are also of the magnitude (-3-10 Nm- I) expected for a simple hydrogen-bonded species but are much smaller than observed for the ion-pair limiting cases (compare Tables 7 and 11) Indeed by assigning a nucleophilicity N = 4 4 to PH,, it is possible to use the established electrophilicities E = 10 0, 5 0, 4 25, and 3 2 for HX (X = F, C1, Br, and I, respectively) in equation 4to predict 33 satisfactorily the observed k, for each H,P.**HX Table 11 also allows the conclusion, by similar arguments, that (CH,),P.-.HX (X = C1 and Br) are of the simple hydrogen-bonded type 34 The experimentally established conclusions for the P analo- gues of the ammonium and trimethylammonium halides are reinforced when the simple energetic arguments of Section 5 are applied Table 12 displays LIE,, for each of the P-containing series The value of dE, in each case has been taken from the ab Table 12 Estimates of LIE, for the reaction (CH,),_,H,P*-*HX = (CH,),-,H,PH+ **-X-, when n = 3 and 0 X AE,/kJ mol -a I= 12 I= 13 I= 14 I= 15 I= 11 n=3 F 26 -795 1554 -420 365 c1 14 -795 1391 -358 252 Br 12 -795 1350 -343 224 I 8 -795 1312 -317 208 n=O Fb -----c1 25 -948 1391 -385' 83 Br 19d -948 1350 -370' 51 I 12d -948 1312 -347' 29 0 See footnotes to Table 10 for the definition of the various AE, h No estimates of AE,, and r(P*-*F) in (CH,),P*-*HF are available Calculated from experimental r(P---X)in Refs 5 and 34 d Ref 35 Calculated from e the ab intlio r(P---X)in Ref 35 Table 11 Comparison of halogen nuclear quadrupole coupling constants x(X) and intermolecular stretching force constants k, for [PH,,HX] and [(CH,),P,HX] with those of model systemsa Molecule 3 5c1 *lBr 1271 X(X)/MHZ k,/Nm-X(X)MHZ k,/Nm-' X(W/MHZ k,/Nm-HCN..-HX -53 270 9 12 356 232(9) 81 -1475 7(1) 4 561(2) H,P---HX -53 861(3)' 5 9h 357 521(6)' 4 3' -1461 022(8)d 3 409(2)" (CH,),P.* * HX -50 486(7)' 10 48(7)' 322 99(4)/ 8 28(4)rNa+...X -5 643(4) 108 6 48 508 93 7 -262 14 77 0 Values for HCN HX and Na' X taken from Table 5 A C Legon and L C Willoughby unpublished observations Ref 32 and Ref 11 d Ref 33 Ref 5 Ref 34 znitio calculations of Muller and Reinhold 35 The gas-phase proton affinities (-dE13) of PH, and (CH,),P are experimen- tally available* while the required dEl4 values are transferred from Table 10 Each dE,, has been estimated as described in Section 5 using the r(P..*X) taken from references 5,30-35, as appropriate It is clear from Table 12 that dE,, is large and positive in each heterodimer, thereby establishing that the simple hydrogen-bonded structure (CH,), -,H,P -HX (n = 0 and 3) is favoured energetically relative to the corresponding ion pair (CH3),-,nHnPH+ --*X- Why isit that (CH,),P.*-HX are not ion-pair heterodimers in the gas phase while (CH,),N...HX (X = Br and I) are, given that the gas-phase proton affinities (-LIE,,) of (CH,),N and (CH,),P are identical2 within experimental error? The answer is clear and lies in the relative magnitude of dE,, in the two series Obviously, r(P -* X) is greater than r(N * X) for a given X and hence LIE,, is more negative in the nitrogen analogue Since for a given X the difference in dE, is small and the LIE,, are identical, the predominant term in stabilizing the ion-pair form in the N series is dE,, This is of sufficient magnitude to make LIE, , negative for the trimethylammonium halides (X = Br and I) but not in the case of any of the trimethylphos- phonium halides In short, the large P atom ensures that LIE,,is insufficiently negative in the latter group 7 Conclusions The spectroscopic constants of the series of dimers (CH,), -,H,N-.*HX and their phosphorus analogues, obtained from rotational spectroscopy conducted on supersoni- cally expanded jets of appropriate gas mixtures in argon, allow the conclusion that the species H,N*..HX and H,P.-*HX can all be described as the simple hydrogen-bonded type, without the need to invoke an appreciable extent of proton transfer However, along the series (CH,),N -* * HX, where X = F, C1, Br, and I, the progressive weakening of the HX bond with respect to the dissociation products H+ and X- favours the ion-pair Experiment shows that for X = Br and I, the ion-pair form is the preponderant contribution to a simple valence-bond description of the molecule Across the series (CH,), -,H,N HC1, where n = 3,2, and 0, the progressive methylation of NH, increases its gas-phase proton affinity so that the extent of proton transfer becomes appreciable for (CH,),N HCl The series (CH,),P * HX, where X = C1 and Br, appears to behave differ- ently, with the simple hydrogen-bond model appropriate in both cases This behaviour contrasts with the observations for (CH,),N*..HX, where X = C1 and Br, and the difference is attributed to the decrease of Coulombic attraction in the gas- phase ion pair of the phosphorus compounds because of the larger radius of P than N These conclusions are broadly in agreement with a number of the more recent ab inztzo calculations’ 24 35 which indicate ion- pair forms for (CH,),N**.HX, where X = C1, Br, and I, but simple hydrogen-bonded structures for all other species dis- cussed in this article These ab znitio calculations also tend to find only a single minimum in the potential energy surface, rather than one associated with the hydrogen-bonded form and another associated with the ion-pair form This is physically reasonable, for the distance that the hydrogen-bond proton needs to move to yield the ion-pair form is only z0 5 8, in most cases This would imply a very sharp, almost singular potential energy barrier between the two forms So it is that the natures of the heterodimers in the vapour above the various ammonium halides [(CH,), -,H,N,HX] and phosphonium halides [(CH,), -.H,P,HX] have been established by experiment and theory Ammonium chloride itself holds a special position in these series as one of the archetypal donor- acceptor systems and consequently there is a history associated with attempts to characterize the interaction The author’s speculations in the Introduction are by no means the first For example, M~lliken,~alluded to inner complexes H,NH + -* * C1- and outer complexes H,N - - HCl in his classic CHEMICAL SOCIETY REVIEWS.1993 paper on electron donors and acceptors in 1952 but at that time there had been no experimental characterization of the hetero- dimer He returned to the theme in his Nobel Prize lecture,’ in 1966, in which he devoted some time to a discussion of the then unpublished but later famous early ab znztzo calculations by Clement1 38-40 These calculations were the first to provide a detailed description of the charge redistribution that occurs in the process of proton transfer and, although it now appears that they found an equilibrium geometry with too much ion-pair character for the (NH,,HCl) system, Mulliken’s comments36 on them are worth quoting, for presumably they are appropriate to cases such as (CH,),N*.* HX (X = C1, Brand I) “ Clementi’s calculations show a gradual transfer of charge from the NH, to the C1 atom, accompanied by some stretching of the H-CI distance, until at equilibrium a structure approaching that of an NHiCl- ion pair, but with considerable polarization of the C1- (H-bonding of NH; to Cl-) is attained The NH, + HC1 system is thus apparently an example of ion-pair formation rather than ordinary loose hydrogen bonding, however, the changes in charge distribution during the early stages of approach of the HC1 and NH, should probably be similar to those in ordinary H- ”bonding and thus instructive for the latter Clementi’s calculations and Mulliken’s comments stimulated new experimental attempts to characterize the heterodimer in ammonium chloride vapour (through mass spectrometry4 and electron diffra~tion~~) These were followed by larger and more refined ab znztzo calculations Raffenetti and Phillips,43 Latajka et a1 ,l and Brciz et a1 44, for example, showed conclusively that the equilibrium form of the dimer is H,N*** HCI with no second minimum corresponding to H,NH+ ***C1 ,in agreement with the experimental result discussed above Finally, the infrared spectroscopy of the complexes (CH,), -,H,N * HX isolated in argon matrices at low tempera- ture18 45 46 has identified N.*.H..*X antisymmetric stretching modes and allowed the existence of ion pairs (CH,) ,-,H,NH+..-X-to be postulated in the cases X = C1 and Br NMR spectroscopy of (CH,),N..-Br in the gas phase also indicates that this species should be classified as an ion pair 47 Acknowledgements I am pleased to have this opportunity to acknowledge the contribution of Elizabeth Goodwin, Nigel Howard, Joanna Thorn, Andrew Wallwork, and, especially, Charles Willoughby and Christopher Rego to our work on the series (CH,),-.H,Y ***HX(Y = N and P, X = F, C1, Br, I reviewed here 8 References 1 R E Smalley, L Wharton, and D H Levy, Acc Chem Res , 1977 10, 139 2 E J Goodwin,N W Howard,andA C Legon,Chem Phjs Lett 1986, 131, 319 3 N W Howardand A C Legon, J Chem Phys, 1988,88,4694 4 A C Legon, A L Wallwork, and C A Rego, J Chem Phjs ,1990 92,6397 5 A C Legon and C A Rego, J Chem SOC Faradav Trans .1990 86, 1915 6 T J Balle and W H Flygare, Rev Scr Instrum ,1981, 52, 33 7 A C Legon, Annu Rev Phys Chem ,1983,34,275 8 D J Millen, Can J Chem, 1985,63, 1477 9 W Gordy and R L Cook, in ‘Microwave Molecular Spectra’, Technique of Organic Chemistry, Vol IX, ed A Weissberger, Interscience, New York, 1970, Chapter 14 10 E W Kaiser, J Chem Phys , 1970,53, 1686 I1 A C Legon and D J Millen, J Am Chem SOC , 1987, 109, 356 12 A C Legonand D J Millen Proc R SOC London Ser A 1986, 40489 13 A C Legon and C A Rego, Chem Phjs Lett, 1989,154,468 14 A C Legon and C A Rego, Chem Phjs Lett, 1989 157,243 15 A C Legon and C A Rego J Chem SOC Chem Commun 1988 1496 J Chem Phvs 1989,90,6867 16 J Baker A D Buckingham, P W Fowler,P Lazzeretti, E Steiner and R Zanasi, J Chem Soc Faraday Trans 2, 1989,85,901 THE NATURE OF AMMONIUM AND METHYLAMMONIUM HALIDES IN THE VAPOUR PHASE-A C LEGON 17 P W Fowler, personal communication 18 A J Barnes, T R Beech, and Z Mielke, J Chem Soc Faraday Tram 2, 1984,80,455 19 Z Ldtajka, S Sakai, K Morokuma, and H Ratajczak, Chem Phys Lett, 1984,110,464 20 N W Howard and A C Legon, J Chem Phys, 1987,86,6722 21 A C Legon and D Stephenson, J Chem Soc Faraday Trans, 1992,88, 761 22 B J Howard and P R R Langridge-Smith, personal communica- tion of D,and x(14N) for H,N..*HF 23 Calculated from W, = (27~)l(k/p):and the value of W, given in S E Veazey and W Gordy, Phys Rev A, 1965,138, 1303 24 Z Ldtdjka, S Schemer, and H Ratajczdk, Chem Phys , 1992, 166, 25 A C Legon and C A Rego, Chem Phvs Lett, 1989,162,369 26 P W Fowler, A C Legon, C A Rego, andP Tole, Chem Phys, 1989 134,297 27 A C Legon and C A Rego, J Chem Phys, in press 28 S Ikuta,P Kebarle,G M Bancroft,T Chan,andR J Puddephatt, J Am Chem Soc , 1982,104,5899 29 A C Legon, D J Millen, and H M North, J Chem Phys, 1987, 86,2530 30 A C Legon and L C Willoughby, Chem Phys ,1983,74, 127 31 A C Legon and L C Willoughby, J Chem Soc Chem Commun , 1982,997 and unpublished observations 32 L C Willoughby and A C Legon, J Phys Chem ,1983,87,2085 33 N W Howard, A C Legon, and G J Luscombe, J Chem SOC Faraday Trans , 199 1,87, 507 34 A C Legon and J C Thorn, unpublished observations 35 B Muller and J Reinhold, Chem Phys Lett, 1992, 196,363 36 R S Mulliken, J Phys Chem , 1952,56,801 37 R S Mulliken, Science, 1967, 157, 13 38 E Clementi, J Chem Phys , 1967,46,3851 39 E Clementi, J Chem Phys , 1967,47,2323 40 E Clementi and J N Gayles, J Chem Phys , 1967,47, 3837 41 P Goldfinger and G Verhaegen, J Chem Phys , 1969,50, 1467 42 S Shibata, Acta Chem Scand, 1970,24, 705 43 R C Raffenetti and D H Phillips, J Chem Phys , 1979,71,5434 44 A Brciz, A Karpfen, H Lischka, and P Schuster, Chem Phys, 1984,89,337 45 A J Barnes, J N S Kuzniarski, and Z Mielke, J Chem SOC Faraday Trans 2, 1984,80,465 46 A J Barnes and M P Wright, J Chem SOC Faraday Trans 2, 1986,82, 153 47 N S Golubev and G S Denisov, Sov J Chem Phys , 1982,1,965

ISSN:0306-0012    

DOI:10.1039/CS9932200153

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Discovery and Development of Anthracycline Antitumour Antibiotics J. William Lown Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2 1 Introduction The clinical treatment of neoplastic diseases relies on the com- plementary procedures of surgery, radiation treatment, immu- notherapy, and chemotherapy. The latter technique has matured from its earliest applications of mustard alkylating agents in the 1940s to a vigorous and increasingly rationally- based discipline which is contributing significantly to the man- agement of human ma1ignancies.l Initially, as a result of the introduction of activity-based screens, the main contribution of the organic chemist was in isolation and structure elucidation. As the field of chemotherapy matured and several promising natural anticancer agents were identified' the interests of organic chemists turned towards total synthesis of such natural products.However, a more urgent need soon arose from the common experience of clinically limiting toxicities of most anticancer drugs, i.e. the necessity to develop less toxic clinical drug candidates.' Thus the role of the synthetic organic chemist or medicinal chemist turned towards analogue development during this, still largely empirical, phase of chemotherapy. Hand-in- hand with this considerable synthetic effort, which uncovered several promising clinical leads, biochemical pharmacology, or study of the mechanisms of action of clinical anticancer agents,' afforded deeper insight into drug metabolism and mode of action.More recently, therefore, the field of synthetic organic chemistry, which has become increasingly sophisticated in the interim and has been complemented by the methods of microbial chemistry, has been faced with new synthetic challenges, occa- sioned by the identification of hitherto unrecognized cellular targets for anticancer drugs, such as topoisomerases, and helicases. The armamentarium of the oncologist currently includes about 40-50 clinically useful chemical agents. The paradigm of cytotoxic anticancer agents is doxorubicin, an anthracycline, J. William Lown was born in Blyth, Northumbria and received the Ph.D. and D.I.C. degrees in organic chemistry at Imperial College in 1959.Following postdoctoral work with Professor R. U. Lemieux at the University of Alberta, Canada, he joined the faculty of that institution where, since 1974, he has been Professor of -Organic Chemistry. Professor Lown was Alberta Heritage Foundation for Medical Research Professor-in-Residence from 1984-85 and was appointed McCalla Research Professor in 1988. Dr. Lown has published over 280 refereed research articles mainly in theJield of cancer chemotherapy. He was awarded the Paul Ehrlich Prize for 1992 by the Association de Recherches Scientijique Paul Neumann for his work in chemotherapy and was elected Fellow of the International Union Against Cancer in 1992. Professor Lown serves as Associate Editor of Cancer Research and is on the editorial boards of Anti-Cancer Drug Research, Free Radical Biology and Medicine, J. Cellular Phar- macology, and Heterocyclic Communications.which is still amongst the most widely prescribed and effective of anticancer agents. The present review attempts to summarize the discovery and isolation of anthracyclines, the elucidation of their structure, stereochemistry, and absolute configuration, and synthetic efforts towards improvement of therapeutic efficacy. 2 Discovery and Isolation of the First Ant hracyclines 2.1 Structure Determination and Conventions The first anthracycline, whose structure was elucidated, p-rhodomycin I1 (l), was isolated from Streptomycespurpurascens by Brockmann and Bauer in 1950.2 Subsequently, in the late 1950s, a red compound isolated from Streptomyces sp.from a soil sample collected in India was shown by Arcamone and his colleagues at Farmitalia to be rhodomycin B (2), a component of the rhodomycin complex studied earlier by the German workers (Figure 1). Although the earliest discovered anthracyclines displayed potent antibacterial activity in culture, their high toxicity in mice precluded their further development as antitu- mour agents. Two principal classes of microbial products were investigated by Brockmann and co-workers.2 The first group which included the rhodomycinones, the isorhodomycinones, the rhodomycins, and the isorhodomycins, was isolated from Streptomyces purpurascens, while the second group, together with the pyrromycinones and the glycoside pyrromycin, were isolated from a related strain.The corresponding deoxyagly- cones (subsequently shown to be important metabolites of anthracyclines) were also isolated and characterized. These early studies served to identify some of the structural features and different substitution patterns characteristic of the aglycones of many anthracyclines subsequently identified. Another important structural feature established by the early workers was the stereochemistry of the sugar moieties rhodosa- mine (3a), 2-deoxy-~-fucose (3b), and rhodinose (3d) contained in the first group of anthracyclines investigated. A significant observation was that the isolation of 7-deoxy compounds such as c-rhodomycinone, (;-pyrromycinone, and r-isorhodomyci- none indicated that reductive elimination of the C-7 oxygen could take place during anthracycline biosynthesis.2 This subse- quently proved to be an important metabolic pathway of the anthracyclines.2.2 Biogenetic Considerations The biosynthesis of two different anthracyclinones, E-pyrromy- cinone and daunomycinone, were found to start with a propio- nate unit3 Acetate units are added to complete the polyketide intermediate, which is later transformed and developed into the anthra~yclines.~This hypothesis, subsequently confirmed by 14C labelling experiments, was used to account for the ethyl substituent in E-pyrromycinone. Such biogenetic schemes including the use of 13C labelling could then be used to rationa- lize the formation of a number of anthracyclines, including aklavinone, E-rhodomycinone, E-pyrromycinone, and daunoru- bicin, and to anticipate the formation of other structures.The biosynthesis of anthracycline antibiotics has been studied genetically using various blocked mutants of anthracycline- producing Streptomyces and by 3C NMR analy~is.~ The results of these studies show that there are two biosynthetic pathways in the formation of the polyketide: daunomycinone, pyrromyci- none, rhodomycinones, and aklavinone are derived from nine 165 CHEMICAL SOCIETY REVIEWS, 1993 0 OH OR@#p. cH3F0HHO OH 0 OH OR' (1) R = R = rhodosamine (2) R = H, W = rhodosamine OH 0 OH 6 Horn'%w (6) Pyrromycin Figure 1 Representative anthracycline structures and numbering system in the chromophore.acetate units and one unit of propionate; steffimycinone and nogalarol are built from ten acetate units. Several anthraquinones such as aklanonic acid have been isolated from Streptomyces sp. 21 MET43717 and blocked mutants of S.galibeus MA144-M1 as intermediates at an early stage in the biosynthesis of anthracyclines. Treatment of S. griseus with MNNG (methyl-N'-nitro-N-nitroguanidine) provided the blocked mutants OIP7 and IP3 which are incapable of producing daunorubicin. These mutants incorporate the total radioactivity of (U-"C) alkanonic acid into 6-rhodomycinone, daunomycinone, 7-deoxy-~-rhodomycinone,and IP/II (yhyd- roxy-bisanhydro-E-rhodomycinone).The closure of the decake- tide chain derived from acetate and/or propionate units subse- quently forms aklavinone via a series of polyketide intermediates; 1,8-dihydroxyanthraquinones,aklavinones, and aklanonic acid.Aklavinone is then converted into rhodomyci- nones and pyrromycinones by oxidation, and glycosidation leads to the corresponding biologically active glycosides (Scheme 1). In the biosynthetic pathway to daunorubicin and doxorubicin it is probable that daunomycinone (15) is synthesized from aklavinone (12) via E-rhodomycinone (1 3) because all daunoru- bicin products of Streptomyces can produce E-rhodomycinone concomitantly in the culture broth, and their blocked mutants capable of accumulating E-rhodomycinone and its glycosides or of accumulating aklavinone and 7-deoxyaklavinone have been i~olated.~In fact it has been confirmed that daunorubicin can be produced by a daunorubicin non-producing mutant of S.coeru-(3a) K = NMe2, (Rhodosamine) (3b) R" = OH, (2-Deoxy-L-fucose) (3c) Fi' = OH, (3' epirner of 3b, Boivinose (3d)R" = H (Rhodinose) (3e)K' = NH2, (Daunosamine) 0 leorubidus in the fermentation medium to which either E-rhodo- mycinone or aklavinone was added. In this manner the biosynthetic sequence of anthracyclines was deduced from microbial modification and glycosidation of various anthracyclinones using a variety of blocked mutants derived from S. galilaeus, S. coeruleorubidus, S. purpurascens, and S.peucetius. 3 Daunorubicin 3.1 Isolation, Degradative Studies, and Properties An important advance in anthracyclines was the discovery and isolation of daunorubicin in the early 1960s, since it was the first antibiotic of this class to show activity against acute leukaemia in man.The antibiotic was discovered independently in the laboratories of Farmitalia, where it was named daunomycin4" and, in those of Rh6ne-Poulenc where it was named rubid~mycin.~~ The first report on the antitumour activity of daunorubicin indicated exceptional pharmacological properties for this new antibiotic. The marked anticancer activity of daunorubicin led to clinical trials where the main therapeutic indications of the drug are leukaemias, especially acute lymphocytic leukaemia in children but also in Hodgkins's disease, lymphosarcoma, and reticular cell sarcoma. The most serious clinical limitation, which was recognized early, is that of dose-dependent cardiotox- icity (vide infra).lb 3.2 Stereochemistry and Absolute Configuration The positions of the substituents and the stereochemistry of the aglycone were then established by a series of chemical degrada- DISCOVERY AND DEVELOPMENT OF ANTHRACYCLINE ANTITUMOUR ANTIBIOTICS-J. W.LOWN COOR ' W H c H 3 0 0 0 0 COOR 0 COOR wHcH 3 -OH 0 OH OH Aklavinone OH(12) CWR0 1 OH 0 OH OH Me0 0 OH OH (13) &-Rhodomyanone (14) e-Pyrromycinone 0 OH 0 CH,O 0 OH OH (15) Daunomycinone Scheme 1 Hypothetical biosynthetic pathway of anthracyclines and distribution of 13C enrichment when [1-13C]acetate (a) or [2-I3C] acetate (*) were fed to cultures of S.peucetius. tions. The aminosugar component daunosamine (3e) proved to be 3-amino-2,3,6-trideoxy-~-lyxo-hexopyranose,a compound not previously found from natural sources. The absolute confi- guration was determined by comparison of the molar rotations of (3e) with those of known 2,6-dideoxyhexapyranosesand their methyl glycosides, and the result was confirmed by 'HNMR spectroscopy. Thus daunosamine has structure (3e). Thus the structure of daunorubicin is represented by (16)and was subse- quently confirmed by X-ray crystallographic analysis of the N-bromoace t y1 derivative of daunoru bicin. 0 OH 0 R cH3729OH (16) R = CH3, W = H (Daunorubicin) (17) R = CH3, K = OH (Doxorubicin) (18)R = K = H (Carminomycin) 0 OH WR* RO 0 OH OH (15) R = CH3, K = H (Daunomycinone) (19)R = CH3, W = OH (Adriamycinone) (20)R' = K = H (Carminomyanone) 4 Doxorubicin (Adriamycin) 4.1 Isolation and Anticancer Activity Doxorubicin (1 7) was isolated from the cultures of one of the varieties derived from Streptomyces peucetius strain, i.e., S.Peucetius var. caesius. This agent generated immediate excite- ment because of its outstanding antitumour properties which were established by Di Marco et aL6 Doxorubicin exhibited the same kind of marked inhibitory effect on tumour growth as daunorubicin but was generally more potent.6 Clinical trials confirmed the early promise of this valuable agent.The principal clinical limitation however remains, as in the case of daunorubicin, the risk of cardiotoxicity that ranges from a delayed and insidious cardiomyopathy to irreversible heart failure (videinfra).lb 5 Classification of Anthracyclines and Nomenclature 5.1 Classification on the Basis of Structure The P-rhodomycins I, 11,111, and IV and the related pyrromycins have been mentioned in connection with the pioneering struc- tural elucidation in Section 2.* The cinerubins A and B are related structurally and were isolated from strains of Strepto-myces. Workers at the Bristol-Myers Company reported the isolation of new anthracyclines from the bohemic acid complex including marcellomycin (22a), musettamycin (2 1a), rudolfo- mycin (21b), and the rhodirubins A and B (22b and c), which were shown by chemical transformations to be related to the pyrromycins.In 1975 investigations at the Sanraku-Ocean Co. and the Institute of Microbial Chemistry in Tokyo led to the isolation and identification of the anthracyclines aclacinomycins A and B.76These agents bear the same trisaccharides as cinerubins A and B respectively (5a, 5b), but the aglycone was novel and it is designated aklavinone (12). A related antibiotic was found to correspond to the aclacinomycins but lacking the two terminal CHEMICAL SOCIETY REVIEWS. 1993 Q$@ OH 0 OH 0 I 0 cH3+ cH3+RO 0 cH3+ (22a) R' = R2 = OH (Marcellomyan) (21a) R = H (Musettamyan) OH -(22b) R' =OH, R2 P H (Rhodirubin A)(21b) =0,-(22c) R' = R2 = H (Rhodrubin B) NH2 (Rudolfmycm) "%*%HO CH3OH cH37&-0H OH 0 OH OR2 CH,O OCH3 Nogalose (23) R' = C02CH3,R2 = Nogalose (Nogalamyan) (24) R' = C02CH3,R2 = CH3 (7-OMethylepinogalorI) (25) R' = H, R2 = Nogalose (Nogalamycin)) (26) R' = H, R2 = CH3 (7-OMethyfepinogalorol) sugar residues and therefore is equivalent to a 1 -deoxypyrromy- an structure Wiley and his co-workers at the Upjohn company isolated a group of anthracyclines [(23)-(26)] which differed substantially in the chromophore moiety The nogalamycins attracted the attention of synthetic chemists both because of their novel structural features and because certain examples exhibited superior antitumour activity The lead compound nogalamycin (23) was chemically modified to prepare the 'noga' and 'nogala' series of analogues Each of these compounds exists as two possible stereoisomers, namely the con and dzs configurations with respect to C-7, which are not interconvertible The struc- ture (7R)-O-methylnogarol (menogaril) (26) shows superior anticancer activity and entered Phase 1/11 clinical trials The structures of additional representative compounds belonging to other groups of anthracyclines are also given (Figure 2) and include baumycin A1 and A2 (27a and b), steffimycins A and B (28a and b), isoquinocycline A (29), elloramycin (30), and barminomycin I (31) 5.2 Classification on the Basis of Mechanism of Action It soon became evident that one of the significant modes of action of many anthracyclines was connected with their ability to bind intercalatively to double helical DNA (Figures 3 and 4) An obvious further classification followed into those anthracyc- lines (including daunorubicin, doxorubicin, and carminomycin) which bind to DNA and those agents which do not (principally semi-synthetic derivatives such as the AD series, including AD32 and AD41) The discovery of inhibition of topoisomerase I1 action by certain non-intercalative anthracy~lines,~ which has important implications in both mode of action and drug resis- tance, emphasizes the reality of this classification and presents new synthetic challenges (videinfra) More extensive and detailed examination of the biochemical pharmacology of anthracyclines revealed additional mechanis- tically related classes based on their ability to inhibit nucleolar RNA synthesis Two classes of anthracyclines are defined, based upon their selectivities for the inhibition of nucleolar precursor ribosomal RNA (NO-RNA) synthesis Class I anthracyclines are nucleolar non-selective, inhibiting both DNA and NO-RNA synthesis at approximately equivalent concentrations Class I1 anthracyclines are nucleolar selective, inhibiting NO-RNA syn- thesis at concentrations 200-1 300-fold lower than those required to inhibit DNA synthesis These two classes of agents may be further subdivided When comparing the affinity constants for binding to DNA, the Class I anthracyclines can be subdivided into carminomycin (low Kapp, apparent binding constants) versus pyrromycin and doxorubicin (intermediate and high Kapprespectively) In addition, the Class I1 anthracyclines marcellomycin, rudolfomycin, musettamycin, and aclacinomycin are distinct from 10-decarbomethoxy mar- cellomycin, l 0-descarbomethoxy rudolfomycin, collinemycin, and mimimycin, based on DNA binding characteristics and biological activities 6 Chemical Reaction and Transformations of Ant h racyc Iines 6.1 Reactions within the Aglycone Acylation of the alcoholic and phenolic hydroxyl groups occurs readily, except for the tertiary hydroxyl at position 9 The phenolic hydroxyl groups are methylated by dimethyl sulfate or methyl iodide and mild base such as potassium carbonate In contrast, methylation of the aliphatic hydroxyl group at position 7 requires more vigorous conditions Demethylation of anthracycline 0-methyl ethers may be effected with either a DISCOVERY AND DEVELOPMENT OF ANTHRACYCLINE ANTITUMOUR ANTIBIOTICS-J.W. LOWN 0 OH 00 mHa3 (33%cH30w@)@: CH-0 0 OH 0 OH 0 OH 0 OH 0 0 (28a)R = H (Steffimycin A) (29) lsoquinocyclineA (28a) R = CH3 (Steffimycin B) (27a) R = CH20H (Baumycin Al)(27b)R = COOH (BaurnycinA2) 0 OH 0 WHY y3 OH 0 1' p"". H3Cy&q" -3 0 0 OH CHI (30) Ellorarnycin Figure 2 Representative alternative anthracycline structures. w5 Figure 3 Daunorubicin-d(CpGpTpApCpG)complex showing intermo- lecular interactions. Lewis acid such as aluminium chloride in benzene, or under milder conditions with sodium thiocres~late.~~ Side-chain bromination at the 14 carbon of daunomycinone occurs readily and the bromine may be readily displaced with the sodium or potassium salts of carboxylic acidslO to give, for example, the valuable 14-octanoate series; or 14-ethers may be obtained by solvolysis in the presence of sodium triflate.A useful intermediate (Scheme 2) derived from the aglycone is the selectively protected triethoxycarbonyl derivative prepared with ethoxycarbonyl chloride and pyridine to give, for example, compound (32). This could be selectively dealkylated with (31) Barminomyan I Figure 4 The conformation of daunorubicin in the complex with d(CpGpTpApCpG) (filled bonds) compared with the conformation of daunorubicin crystallized by itself (open bonds).aluminium bromide to (33). Ethoxylation, followed by treat- ment with morpholine, permitted selective removal of the phe- nolic ethoxycarbonyl protection. Removal of the remaining ethoxycarbonyl group at the 7-hydroxyl position may be effected with sodium hydroxide, and glycosidation affords (37). Dehydration of the aliphatic ring of the chromophore occurs relatively readily under acidic conditions because the hydroxyl groups present are either tertiary or benzylic. Thus bis-hyd- ration of 7,9-diols results in aromatization of ring D. The 7- deoxy aglycones (40) cannot aromatize in ring D by dehydration, and instead one or other olefinic product is formed depending on the dehydrating agent used.Heating of (40)at 230 "C gives (43) together with the ring-opened compound (44), (Scheme 3). Side-chain oxidative degradation of the aglycone in 13-dihydro-N-trifluoroacetyl daunorubicin (46) may be effected by sodium metaperiodate, (Scheme 4)and affords the ketone (47) which, after reduction to the alcohol, may be used for the preparation of alternative derivatives. The 13-carbonyl group of both daunorubicin and doxorubicin is reduced to the alcohol by potassium borohydride. Conversion of the acetyl side chain of daunorubicin into the corresponding 9-ethyl group may be effected by reduction of the 13-thioketal with Raney nickel. CHEMICAL SOCIETY REVIEWS, 1993 c, 0 OH RO 0 OH OH (36) Scheme 2 Reaction conditions: (a), EtOCOCI, pyridine; (b), AIBr,; (c), RI, Ag,O; (d), morpholine; (e), OH-; (f), daunosaminyl chloride, CF3S03Ag, MeOH, OH-.OH 0 OH OH OH 0 OH OH (38) E-Pyrrom ycinone (39) q-Pyrromycinone OH 0 OH \\ (40) (Pyrromycinone \\ '1f (42) OH 0 OH OH 0 OH (44) (45) Scheme 3 Reaction conditions: (a), p-TSA, or HCl; (b), p-TSA; (c), P,O,; (d),HBr; (e) 230 "C; (f) pyridine, Et,N. However the 7-hydroxyl group is also subject to some hydrogen- olysis under these conditions. Alternatively sodium cyanoboro- hydride reduction of the 13-p-toluenesufonylhydrazoneof daunomycinone affords the 7,13-dideoxy derivative. 6.2 Reactions within the Aminosugar Moiety Methylation of the amino group in the daunosamine moiety of anthracyclines gives either the N,N-dimethyl derivative or the RO 0 OH OC02Et (35) 0 OH 0 RO 0 OH 0 (37' a3+HO 0 OH OH OH 0 OH Me6 0 OH 0 Me6 0 OH 0 I I sugar sugar (46) (47) OH 0 OH 0 I sugar Scheme 4 Reaction conditions: (a), NaIO,.quaternary ammonium salt, depending on the reaction con- ditions. Selective methylation of the 6-and 11-phenolic hydroxyl group of daunorubicin may be effected after protection of the amino group of the sugar moiety as its trifluoroacetyl derivative. Similarly N,N-dibenzyl derivatives of daunorubicin and doxor- ubicin proved to be valuable, both in terms of their antitumour activity and in metabolic studies. Acylation of the amino group in the daunosamine moiety is readily accomplished.The N-trifluoroacetyl derivates (AD ser- ies) are especially noteworthy because of their expression of antitumour activity in spite of lacking the ability to bind intercalatively to DNA. Representative chemical transforma- tions of daunosamine (3e) leading to an acosamine derivative (49) are shown in Scheme 5. Additional amino sugar derivatives are discussed in Section 8. 6.3 Reactions of the Glycosides Once the clinical potential of doxorubicin was recognized, a chemical transformation of intact glycosides that became of immediate interest was the interconversion of daunorubicin into doxorubicin. This could be effected (Scheme 6) by trifluoroace- tylation of daunorubicin, followed by 0-acyl exchange, to the N-trifluoroacetyl derivative, which on photohalogenation gave (51).13 Replacement of the iodo substitute with acetate gave (52),and the latter was converted into N-trifluoroacetyl doxoru- bicin (53) upon treatment with a weak base.Hydrolysis of (54) followed by acid treatment afforded doxorubicin (17). Subse- DISCOVERY AND DEVELOPMENT OF ANTHRACYCLINE ANTITUMOUR ANTIBIOTICS-J. W. LOWN HO HO HO OCH3 OCH3 4 c H 3 H d,( 3 4 3 e &CH3ToH NHCOCF, NHCOCF, HO NHCOCF, c1 Scheme 5 Representative chemical transformations of daunosamine 6.4 Metabolism leading to N,O-ditrifluoroacetyl-a-acosaminylchloride. Reaction The two main and characteristic routes of metabolic transfor- conditions: (a), MeOH/H +;(b), 1, (CF,CO),O; 2, MeOH; (c) RuO,/ mation of the anthracyclines in animals and in man are the KIO,; (d), NaBH,; (e) AcOH/H,O; (f), (CF,CO),O, HCl/Et,O.reduction of the side chain carbonyl group to a secondary alcohol and reductive deglycosidation with formation of 7-quently a more direct conversion of daunorubicin into doxoru- deoxyaglycones. l4 The enzyme catalysing the first reaction is a bicin was developed by electrophilic bromination of (50)at C-14 cytoplasmic aldoketo reductase, native to all tissues (although it followed by replacement of the bromine by hydroxyl with mild has been referred to as daunomycin reductase). It is also capable base treatment. of reducing daunomycinone, and other anthracycline aglycones, The reverse transformation, of doxorubicin into daunorubi- to the 13-dihydro compound (46).l4 These conversions are cin, was used for the synthesis of [14-14C]daunorubicin by stimulated by NADPH and do not require oxygen. Arcamone and co-workers.' [ 14-14C]Daunorubicin was con- The reductive scission of the glycosidic bond is catalysed by a verted into [ 14-14C] doxorubicin via the 14-bromo derivative as reductive hydrolase, the action of which appears to be rare since described previously. there are no other examples of enzyme catalysis of this otherwise chemically facile reaction. Rat liver microsomal preparations Scheme 6 Reaction conditions: (a) (CF,CO),O, MeOH; (b), I,, CaO, were found to convert both daunorubicin and daunorubicinol THF; (c), NaOAc; (d), Na,CO,; (e), OH-; (f) H+.into their aglycones.14 The scission of the glycosides by both 0 OH 0 HO 0 OH 0 0 OH 0 Me0 0 OH 0 Me0 0 OH 0 HO HO 0 OH ?, Me0 CH3?-;-I (54) NHCOCF, HO CHEMICAL SOCIETY REVIEWS, 1993 0 OH . .. Me0 0 OH 0 Me0 0 OH (46a) R = H (55a) R = H (46b) R = OH (55b) R =OH t OH \ 0 OH 0 0 OH 0fl. 0 OH OH __t Me0 0 OH Me0 0 OH Me0 0 OH (56a) R = H (57a) R = H (56b) R = OH (57b) R = OH ..*HO OH 00 OH I 0 OH . .. Me0 0 OH Me0 0 OH (59a)R = H (60a) R = H (59b) R = OH (60b) R = OH Figure 5 Principal metabolites of the anthracyclines routes is of immediate pharmacological and clinical concern because, while the aglycone and its 7-deoxy counterpart do not appear to contribute to the cytotoxic/anticancer activity, they appear to contribute to general side-effect toxicities Although the liver was the most effective organ in glycoside cleavage reactions, lung, kidney, brain, skeletal muscle, and heart also showed significant activities l4 The principal metabolites of the anthracyclines are also subject to conjugation The following aglycone-derived meta- bolites have been identified from the urine of patients following treatment with daunorubicin (Figure 5) 13-dihydrodauno-mycinine, 7-deoxydaunomycinone, 7-deoxy- 1 3-dihydrodauno- mycinine (55a), 4-methyl-7-deoxy- 13-dihydrodaunomycinone (58a) R = H (58b) R = OH OH 0 OH OH COOHHoHw0 OH OH (6la)R=H (6lb)R=OH vity in the construction of the chromophore, stereoselectivity at positions 7 and 9, stereocontrolled synthesis of L-daunosamine, and efficient glycosidic coupling procedures 7.2 Microbial Transformations Another valuable and complementary route to new and poten- tially useful anthracycline structures is by the action of microbial cultures, and particularly by the use of blocked mutants Since 1981 about 20 new analogues of daunorubicin, doxorubicin, and carminomycin have been reported The baumycin-producing Streptomyces D788,16 which is different from the well-known baumycin-producing strains of S coeruleorubidus and S peuce-tzus is transformed by treatment with MNNG to a mutant which lacks the ability to form 4'-0-substitution products Fermen-(58a), 4-demethyl-7-deoxy-13-dihydrodaunomyc~none-4-0-tation using this mutant, which cannot produce baumycins, sulfate (60a), 4-demethy1-7-deoxy-13-dihydrodaunomycinone-allows the accumulation of daunorubicin and concomitantly 4-0-fl-~-glucuronide(6 1a), and 7-deoxy- 13-dihydrodaunomy- cinone-l3-O-fl-~-glucuronide(59a) l4 7 Overview of Anthracycline Development 7.1 Synthetic Strategy The recognition of the anticancer activity of daunorubicin and particularly the clinical potential of doxorubicin began to attract the attention of synthetic organic chemists in the early 1970s The pioneering work in this field is due to Wong and co-workers who reported the synthesis of the racemic aglycone derivative 4- demethoxy-7-0-methyldaunomycinone The synthetic stra- tegy was essentially A + CD -,ABCD Subsequently Wong and co-workers reported the total synthesis of racemic daunomyci- none following essentially the same strategy Some of the synthetic approaches were adopted and adapted in the Farmita- ha laboratories where, for example, resolution of a key racemic substituted tetralin intermediate opened a route to a number of different daunomycinone analogues obtained in optically pure form Following this method both enantiomeric forms of 4-demethoxydaunomycinone were prepared Major synthetic questions that were subsequently addressed, in an effort which is international in scope, included regioselecti- produces a new water-soluble anthracycline D788-1, identified as 10-carboxyl- 13-deoxocarminomycin Further mutation of strain GI-1 with MNNG provided a doubly blocked mutant strain RPM-5 which produces an intensely potent minor compo- nent called oxaunomycin,' in addition to 13-deoxocarminomy- cin, 13-dihydrocarminomycin, and 10-decarboxy- 13-deoxycar- minomycin The structure of oxaunomycin is 7-0-(a-~-daunosamin yl)-fl- rhodom ycinone Two strains of Actznomadura roseoviolacea produce the new anthracyclines N-formyl- 13-dihydrocarminomycin' and akro- bomycin (9,lO-anhydro- 13-deoxocarminomycin) respecti-vely A blocked mutant MnW I of the carminomycin-producer Actinomadura roseoviolacea can convert anthracyclinones into the corresponding glycosides, e g conversion of E-rhodomyci- none into carminomycins and bio-inactive 4-0- (P-D-glucopyra- nosy1)-6-rhodomycinone, E-pyrromycinone to the novel 1-hyd-roxy- 1 1-deoxycarminomycin I1 In conjunction with biosynthetic studies of anthracycline oligosaccharides, the microbial glycosidation of biologically inactive anthracyclinones is aimed at preparing new anthracyc- lines with better therapeutic indices An aclacinomycin negative mutant strain KE303 derived from S gablaeus MA144-MI has the property of glycosidating the C-7 or C-10 position of various DISCOVERY AND DEVELOPMENT OF ANTHRACYCLINE ANTITUMOUR ANTIBIOTICS-J.W. LOWN (62a) R=H (62b) R =OH OH 0 OH 0 aglycones, e.g. 2-hydroxaklavinone, 8-rhodomycinone, p-isor- hodomycinone, ~-2-rhodomycinone, and a-citromycinone with mono, di, or trisaccharides.16 2-Hydroxyaclacinomycins A and B, betacalamycinsA, M,N, S,andT,16CG10,CG11,andCG12 especially valuable for uptake and cellular distribution studies.0 OH 0 6CH3 I YPOH 0 OH 0 CH30 0 OH 0 I Esters of doxorubicin were obtained by reaction of 13-bromodoxorubicin with the sodium salts of carboxylic acids. Variants of this type of derivative include double esters and thioester structures. Side chain fatty acid derivatives proved are produced by adding the corresponding aglycone to a culture of the blocked mutant KE303. The microbial reduction of daunorubicin to daunorubicinol first reported in 1975 can also be effected with strains of Streptomyces lavendulae, S. roseochromogenes, Corynebacter- ium simplex, and Bacterium cycloxydans. These strains are also capable of converting carminomycin into 13-dehydrocar- minomycin.Corynebacterium egui has been used for prepara- tive scale selective side-chain reduction of N-acetyldaunorubicin to N-acetyl- 13-dihydrodaunorubicin. 6h 8 Clinical Experience of Doxorubicin Leading to Analogue Development The initial observation of the outstanding antitumour activity of doxorubicinl directly stimulated the studies on the synthesis of anthracycline aglycones and total synthesis of the natural anti- biotics. Early clinical studies revealed toxicities however, especially the risk of cardiot~xicity.~ This led directly to analo- gue development, and therefore to a systematic exploration of anthracycline chemistry in the hope that such toxic effects could be minimized or eliminated.8.1 Side-chain Modification Obvious functional sites for derivatization in the anthracyclines were at C- 13 and C-14. This gave rise to products from reduction at C-13, derivatization of the C-13 carbonyl 14-esters and thioesters, 14-amino derivatives and compounds derived from oxidative degradation of the doxorubicin side chain and 14- ethers of doxorubicin. Reduction of the C- 13 carbonyl group may be effected microbially (vide supra) or chemically with potassium borohydride. The regioselectivity of the reaction is ensured by the rapid air oxidation of the hydroquinone. 13-Deoxo analogues of the anthracyclines (of interest because of the presence of the C-9 ethyl group in many promising natural anthracyclines) were produced by sodium cyanoborohydride reduction, daunomycinone or daunorubicin tosylhydrazone.Conventional functionalization of the carbonyl group occurs unremarkably to afford semicarbazones, thiosemicarbazones, oximes, and substituted hydrazones. A variation of the latter was to prepare linked anthracycline bishydrazones (62) by reaction with the corresponding dihydrazide in methanol. Of particular interest in this regard was the N-trifluoroacetyl- doxorubicin 14-valerate (63). The early recognition that intercalative binding of anthracyc- lines to doubled helical DNA was an important component of the cytotoxic action of anthracyclines led to the consideration of net charges on the molecule. Therefore amine functions were introduced at C- 14 by reaction of 14-bromodaunorubicin with amines. Periodate oxidation of 13-dihydroxyrubicin N-tri- fluoroacetate followed by reduction of the resulting C-9 formyl compound to a C-9 hydroxymethyl derivative and de-N-acyla- tion gave C-9-hydroxymethyldaunorubicin.Reaction of the hydroxyaldehyde with diazomethane affords both N-trifluoro- acetyldaunorubicin and an epoxide. Many of the derivatives obtained by functionalization at C-9 and C- 10 are obtained from the 8,lO-anhydro-N-trifluoroacetyl-daunorubicin which is obtained by treating daunorubicin hydrochloride with trifluoracetic anhydride in the presence of collidine at 0"C.l9 Catalytic hydrogenation of (64) in the presence of Pd/BaSO,, followed by base treatment gives a 9-deoxydaunorubicin.The presence of a methoxy substituent at C-8 in the steffimy- cins stimulated interest in developing routes to alternatively C-8 substituted structures. Protection of the C-9-OH and C-13 carbonyl in daunomycinone by treatment with dimethoxypro- pane and p-toluenesulfonic acid permits selective C-7-C-8 dehydration to give (65).20 Epoxidation of (65) proceeds nor- mally to give (66). Opening of the oxirane ring with methanol and mild acid gives the 7-methoxy-8-hydroxy derivative. Methy- lation with methyl iodide and sodium hydride in tetrahydro- furan and deprotection with trifluoroacetic acid gives the new 8- substituted aglycone (67). O 8.2 Chromophore Modification The recognition of the intercalative capacity of the anthracyc- lines and its bearing on their biological properties naturally focused attention on the chromophore.Development of effective routes, both total synthetic and semisynthetic, to 4-demethoxydaunorubicins produced especially potent analogues. * These developments focused attention on ring A of the chromophore and routes to 1,4- and 2,3-disubstituted anthracyclines were pursued. Additional stu- CHEMICAL SOCIETY REVIEWS, 1993 CH30 0 OR Q$@$&H3 CH30 0 0 OH 0 OH 0 dies along these lines led to 5-0-methyl and 11-0-methyl derivatives. In addition to the above chemical modifications of the chro- mophore, mechanistic considerations also influenced the direc- tion of synthetic studies. The recognition of the generation of free radical species following microsomal or chemical activation of the quinone moiety of anthracyclines raised the question of the contribution of this pathway to either the cytotoxicity or cardiotoxicity.2 Chromophore-modified glycosides were prepared by total synthesis in which the bioreducible quinone group was replaced by 7-pyrone (68) or 7-thiapyrone (69) moeities,z2 of which (68) proved to be cytotoxic.8.3 Aminosugar Modification The first synthesis of L-daunosamine was reported by Goodman et al. starting from ~-rhamnose.~~ An alternative procedure was developed by Horton and Weckerle starting from methyl-a-D- mannopyrano~ide.~~The synthesis of D-daunosamine and D,L- daunosamine have also been described.2 Systematic exploration of the L-arabino analogues of these glycosides led to the new clinical candidate drugs 4'-epidoxoru- bicin and 4'-epidaunor~bicin.~~ Additional sugar modifications are the 4-deoxy analogues, 4'-methyl derivatives, and 4'-C- methylated analogues.26 Configurational analogues of doxoru- bicin and daunorubicin that have been synthesized and eva- luated and include: (i) those belonging to the L-series but possessing the same configuration at C-1 '-(a-glycosides), i.e. the L-rib0 and L-xylo analogues; (ii) those belonging to the L-series but showing inverted configuration at C- 1'-@-glycosides); and (iii) those belonging to the D-series both in the a and /3 anomeric forms. ,z 8.4 Serendipitous Modification Results from the systematic modification of anthracyclines have been complemented by a number of fortuitous discoveries that have influenced the synthetic strategy of analogue development.The first involves the treatment of daunorubicin with methano- lic ammonia during certain manipulations that had the unexpec- ted result of producing a stable 5-imino derivative (70).27The imino group is presumably introduced selectively because of the additional stabilization afforded by the hydrogen bonds to the flanking oxygen functions. The 5-imino derivatives of daunorubicin and doxorubicin OCH3 OH OR 0 OH 0 OH OH .RZ 0 OH 0 N-N / CH30 NH OH 0 R3 R (70)R = daunosaminyl, R' = H or OH proved to be quite potent, and perhaps more significantly, less cardiot~xic.~This serendipitous finding led directly to the synthetic rationale for producing the promising anthrapyrazoles (71) and, retrospectively, in interpretations of the reactivity and toxicity properties of such agents as mitoxantrone.Another valuable observation emerged from Acton's group during their development of a general method for N-alkylation of daunorubicin and doxorubicin.2 * This method is based on the N,N-dimethylation of amines that uses formaldehyde as a source of the methyl groups, sodium cyanoborohydride as reducing agent, and aqueous acetonitrile as the solvent (Scheme 7). This selectivity under mild conditions is convenient for work with daunorubicin and doxorubicin because concomitant reduc- tion of the 13-ketone to the 13-dihydro derivative could be minimized as a side reaction.29 Similar use of sodium cyanobor- ohydride has recently afforded the intensely potent N-(5,5-diacetoxypent-1-y1)doxorubicin. O The morpholino anthracyclines are a special subclass of N-alkyl derivatives from doxorubicin and daunorubicin.Use of glutaraldehyde gave piperidino derivatives incorporating the amino-N in a new ring. As logical variants additionally incor- porating a ring-0, the morpholino analogues were synthesized by reductive alkylation with 2,2'-oxybisacetaldehyde. Neutral by-products of this synthesis unexpectedly included an excep- tionally potent derivative which proved to be the a-cyano- morpholine derivative (73).29 Formation of these a-cyano compounds can be explained by a mechanism of the reductive alkylation involving iminium inter- mediates in two stages.Iminium ions are excellent alkylating species and at either of the two intermediate stages can evidently capture CN-, which is present as an impurity in the NaBH3CN reagent or as a product of its consumption (Scheme 8). Not only are the cyanomorpholinoanthracyclines exceptio-nally potent as anticancer agents but they also appear to function quite differently from conventional anthracyclines in inducing, for example, DNA interstrand cro~s-links.~~ DISCOVERY AND DEVELOPMENT OF ANTHRACYCLINE ANTITUMOUR ANTIBIOTICS-J. W. LOWN 0 OH 0 Scheme 7 - motpholinea DNA or DOX I I Cqz&H* Ity0 cyanomorphokner Scheme 8 Intermediates in the formation of morpholinoanthracylines. 9 Synthetic Challenges in Response to Mechanistic Findings The anthracyclines are unique amongst clinical anticancer agents in terms of the breadth and depth of prolonged enquiry into all aspects of their properties.Among the most challenging problems has been the elucidation of their mode of action. The consensus view is that the anthracyclines exert several parallel cytotoxic mechanisms,22 which in turn pose new synthetic challenges for the organic chemist. Some of these challenges have been met, but with qualified success. Thus while an adequate biochemical basis has yet to be established for the cardiotoxicity of the anthracyclines, the pragmatic view was taken that it might be connected to redox cycling of the quinone moiety.22a This interpretation was offered post facto for the apparently lower cardiac toxicity of 5-iminodaun~rubicin~ 'and mit~xantrone,~and was used to justify the design and synthesis of the promising anthrapyra~oles~~ and related structures. A new challenge to synthetic chemists has arisen from the discovery of DNA topoisomerase I1 and helicases as an intracel- lular target of anthracy~lines.~ The DNA intercalating analo- gues of doxorubicin and daunorubicin as well as certain of the non-intercalating N-acylanthracyclines are apparently topoiso- merase 11-targeted drug^.^ Metabolic activation of 14-acyl ana- logues of N-acylanthracyclines by ubiquitous non-specific ester- ases is a prerequisite for the drug interaction with the enzyme- DNA complexes. The mapping analysis of DNA cleavage mediated by topoisomerase I1 has shown a similar pattern among various anthracy~lines.~ Molecular pharmacological evidence suggests that anthracyclines interfere with the break- age-reunion reaction of topo-I1 by (i) stabilizing the complex between the enzyme and DNA, (ii) changing its non-cleavable state into a cleavable one, thus (iii) triggering a sequence of events leading to cell death.9 Drugs with N-trifluoroacetyl or N-pentafluoropropionyl substitution of the sugar moiety are sub- stantially more effective than analogues with N,N-dibenzyl or N-acetyl substitution.Since some of the compounds known to inhibit topo-I1 DNA complexes are poor DNA binders, drug intercalation may not be a precondition of a drug-induced inhibition of topo-11.Doxorubicin and several analogues may bind to the enzyme-DNA adducts and form a ternary complex. Such a proposed complex implies that a topo-I1 molecule has a domain which represents a binding site for the drug. The domain may have become accessible following formation of the non- cleavable complex between topo-I1 and DNA. Subsequent binding of anthracyclines to topo-I1 inhibits the enzyme func- tion and stabilizes the cleavable c~mplex.~ The removal of bulky side chains of AD32 or AD143, which is a precondition of the drug interaction with topo-11, may reduce steric hindrance and facilitate the formation of drug-topo-I1 DNA comple~es.~ These studies suggest strategies for development of new anthracycline analogues.These would include identification of various substituents of anthracyclines responsible for topo-I1 inhibition. This, in turn, could lead to new analogues which, unlike doxorubicin, stabilize the topo-I1 DNA complex with high efficiency, and which may translate into higher anticancer potency . Additional synthetic challenges arise from the extremely serious problem of resistance to chemotherapy in the manage- ment of cancer patients. There are indications that resistance to topo-I1 mediated anthracyclines may be related to low levels of topo-I1 in quiescent human leukaemic and cancer cells.9 Syste- matic exploration of the structural requirements in anthracyc- lines for efficient topo-I1 inhibition may therefore also contri- bute to the resistance Such challenges arising directly from a more sophisticated view of anthracycline action will require the organic chemist to work more closely with enzymolo- gists and molecular pharmacologists in the future.10 Prospects Synthetic organic chemistry has contributed significantly in all phases of anthracycline development. These contributions include isolation and structure determination, establishment of absolute configuration, organization into different structural classes, exploration of the characteristic chemical properties and 176 reactions of anthracyclines, the development of procedures for total synthesis of steadily increasing sophistication and control, and finally, in response to clinical challenges, the generation of anthracycline analogues on a massive scale for biological testing During the long effort extending over 25 years or more the role of the organic chemist has changed The major challenge is no longer in the area of total synthesis of natural antibiotics Increasingly effective microbial transformations can generate new candidates for biological screens Analogue development in response to clinical observations of severe toxicity of existing anthracyclines encouraged exploration of the basic chemical properties of these agents Well over 2000 analogues have been synthesized to date in this largely pragmatic and semi-empirical development programme The immediate problem of cardiotox- icity has eased somewhat as a result of the introduction and co- administration of ICRF-187 Newer and greater challenges have emerged for the organic chemist from recent mechanistic stu- dies These include compounds synthesized especially to address questions related to topoisomerase I1 or helicase3 mediated cell effects and the possibly related questions of drug resistance 32 Additional synthetic challenges are posed by the need to develop improved cell delivery of anthracyclines by, for example, stealth liposomes 35 11 References (a)J A Montgomery, T P Johnston, and Y F Shealy, in ‘Burger’s Medicinal Chemistry’, 4th Edn ,Part TI, ed M E Wolff, Wiley- Interscience, New York, 1979, p 595 (b)F Arcamone, ‘Doxorubi- cin Anticancer Antibiotics Medicinal Chemistry’ A Series of Monographs, Vol 17, Academic Press, New York, 1981 H Brockmann and K Bauer, Naturewissenschaften, 1950,37,492 (a)T Oki, in ‘Anthracycline Antibiotics’, ed E Khadem, Academic Press, New York, 1982, p 75 (6) V P Marshall, ‘Biochemical Perception’, ed L N Ornston and S G Shgar, Academic Press, London, 1982, p 75 (a) A Grein, C Spalla, A Di Marco, and G Canevazzi, Giorn Microbiol, 1963, 11, 109 (b) M Dubost, P Gauter, R Maral, L Ninet, S Pinnert, J Preud’homme, and G H Werner, C R Acad Sci Paris, 1963, 257, 18 13 0 Kennard, R Angiuli, Fr Arcamone, E Foresti, N W Isaacs, D L Wampler, W D S Motherwell, and L Riva Di Sanseverino, Nature, New Biol, 1971,234,78 A Di Marco, M Gaetani, and B M Scarpinato, Cancer Chemother Rep, 1969,63,33 (a)D E Nettleton, Jr ,W T Bradner, J A Bush, A B Coon, J E Moseley, R W Myllymaki, F A O’Herron, R H Schreiber, and A L Vulcano, J Antibiot (Tokyo), 1977, 30, 525 (b) T Oki, Y Matsuzawa, A Yashimoto, K Numato, I Kitamura, S Hori, A Takematsu,, H Umezawa, M Ishizuka, H Noganawa, H Suda, M Hamada, and T Takeuchi, J Antibiot (Tokyo), 1975,28, 830 CHEMICAL SOCIETY REVIEWS 1993 8 P F Wiley, F A MacKellar, E L Caron, and R B Kelly, Tetrahedron Lett , 1968,4, 663 9 M Potmesil, in ‘Anthracycline and Anthracenedione-based Anti- cancer Agents’, Bioactive Molecules, ed J W Lown, Vol I11 Elsevier, Amsterdam, 1988, p 432 10 F Arcamone, G Franceschi, A Minghetti, S Penco, S Redaelli, A Di Marco, A M Casazza, T Dasdia, G DI Fronzo, F Guiliani, L Lenaz, A Necco, and C Soranzo, J Med Chem , 1974,17,335 11 E M Acton, in ‘Anthracyclines Current Status and New Develop- ments’, ed S T Crooke and S D Reich, Academic Press New York, 1980, p 15 12 M Israel, E J Modest, and E Frei, Cancer Res , 1975, 35, 1368 13 F Arcamone, W Barbieri, G Franceschi, and S Penco.Chim Znd (Milan), 1969, 51, 834 14 N R Bachur and M Gee, J Pharmacol Exp Ther , 1971,177,567 15 C M Wong, D Popien, R Schwenk, and J TeRoa, Can J Chem 1971,49,2712 16 S Fuji], K Kubo, 0 Johdo, A Yoshimoto, T Ishikura, H Naganawa, T Sawa, T Takeuchi, and H Umezawa, J Antibiotics (Tokyo), 1986,39,473 17 A Di Marco, M Gaetani, L Dorigotti, M Soldati, and 0 Belhni, Cancer Chemotherapy Rep , 1964,38, 3 1 18 F Arcamone, L Bernardi, B Patelh, and A Di Marco, Ger Patent 2557537 (July 8, 1976) Chem Abstr ,1976,85, 177866 19 S Penco, F Gozzi, A Vigevani, M Ballabio, and F Arcamone, Heterocycles, 1979, 13, 28 1 20 S Penco, F Angelucci, M Ballabio, A Vigevani, and F Arcamone, Tetrahedron Lett , 1980,21,2253 21 B Patelli, L Bernardi, F Arcamone, and A DiMarco, Ger Patent 2 525 633 (July 29, 1976) 22 (a)J W Lown, Advances Free Radical Biology andMedicine, 1985,1, 225 (b) J W Lown, S M Sondhi, and J A Plambeck, J Med Chem , 1986,29,2235 23 J P Marsh,C W Mosher, E M Acton, and L Goodman, Chem Commun , 1967,973 24 D Horton and W Weckerle, Carbohydrate Res , 1975,44,227 25 A C Richardson, Chem Commun , 1965,627 26 F Arcamone and S Penco, in ‘Anthracycline and Anthracenedione- based Anticancer Agents’, ed J W Lown, Elsevier, 1988, p 1 27 L Tong,D W Henry,andE M Acton, J Med Chem ,1979,22,36 28 E M Acton, in ‘Anthracyclines Current Status and New Develop- ments’, ed S T Crooke and S D Reich, Academic Press, 1980, p 15 29 E M Acton, K Wasserman, and R A Newman, in ‘Anthracycline and Anthracenedione-based Anticancer Agents’, ed J W Lown, Elsevier, 1988, p 54 30 A Chenf and D Farquar, J Med Chem , 1992,35,3208 31 W H M Peters and H M J Roelofs, Cancer Res , 1992,52, 1886 32 M A Graham, D R Newell, B J Foster, L A Gumbrell, K E Jenns, and A H Calvert, Cancer Res , 1992, 52, 603 33 P Genne, M T Dimanche-Boitrel, R Y Mauvernay, G Gutierrez, 0 Duchamp, J -M Petit, F Martin, and B Chauffert, Cancer Res , 1992,52,2797 34 N R Bachur, F Yu, R Johnson, R Hickey, Y Wu, and L Malkas, Mol Pharmacol, 1992,41,993 35 A A Gabizon, Cancer Res , I992,52, 89 1

ISSN:0306-0012    

DOI:10.1039/CS9932200165

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Computer Simulations on Aqueous Solutions of Some Non-electrolytes Koic hi ro Nakanishi Department of Industrial Chemistry, Kyoto University, Sak yo-ku Kyoto 606-01, Japan 1 Introduction Water is a very good solvent for many kinds of substance. If we confine the scope of solutes to non-electrolyte molecules, we can construct a series of solutes according to the magnitude of the cohesive energy density of the molecule in question. One exam- ple of such energy density series is shown in Figure 1, where one can see that the position of water is at the upper end (the highest energy density). By contrast, rare gases, fluorocarbons, and hydrocarbons are near the other (lower energy density) end of the series. It is well known that the latter substances show only a low solubility in water and that, although the composition ranges of their aqueous solutions are severely limited, there occurs a remarkable change in the structure of solvent water.The structure promotion in such dilute aqueous solutions is known as ‘hydrophobic hydration’. On the other hand, the positions of polar and associated molecules, such as urea, polyols, and hydrogen peroxide, are located near the position of water. Such molecules can be accommodated into water structure rather easily and their aqueous solutions can be treated as normal mixtures.2 In such aqueous solutions, structure enhancement of water might not exist, but rather a structural disruption has been suggested, as in the case of urea. Only hydrogen bonding interactions should be predominant.There are many other organic compounds between these two extremes. Detailed thermodynamic studies indicate that, except for some highly hydrophilic compounds mentioned above, their aqueous solutions can be characterized by the behaviour of the non-polar group in the molecule. Rare gases and alcohols behave similarly as far as the thermodynamic excess properties are concerned, and the role of polar groups in the latter molecules is simply to lower the free energy and to promote a high miscibility with water. Although important progress has been achieved experimen- tally in thermodynamic studies, knowledge at the molecular level is still insufficient. Recently, there has been much progress in molecular simulation because of rapid developments in supercomputers as well as in the techniques for calculating static and dynamic properties of molecular ensembles. The method of molecular sirnulati~n,~ which consists of molecular dynamics (MD) and Monte Carlo (MC) calculations, can serve as a complementary method to traditional macroscopic approaches.In this review we first give a brief description of how we can Koichiro Nakanishi was born in K3~oto (Japan) in 1931. He obtained his Ph.D. at Kyoto University in 1962. He was Assistant Professor at Shinshu University (Nagano, Japan) from 1961 to 1966. In 196314 he worked with Professor J. H. Hildebrand at the University of California, Berkeley. Since 1966 he has been at Kyoto University, where he has been Professor since 1987.In 19861 7 he was adjunct Associate Professor at the Institute for Molecular Science, Okazaki (Japan). large high Compound Water Hydrogen Peroxide Polyols(Ethy1ene Glycol) UreaI TCohesive Purine Polarity Energy Methanol Density Propan-2-01 Acetone Acetonit rile 2-Methylpropan-2-01 Fluoroalcohols Benzene Methane Carbon Tetrafluoride small Argon1 1low Figure 1 Cohesive energy density series for water and organic non-electrolytes. execute MD and/or MC calculations for aqueous solutions in which the hydrogen bonding interaction is predominant. We then discuss the results of molecular simulations of aqueous solutions of, mainly, 2-methylpropan-2-01 (TBA) and urea.There exist several reasons for the choice of solutes. First, these two molecules represent highly hydrophobic and hydrophilic compounds, respectively. Secondly, their molecular structures are such that conformational changes are relatively unimpor- tant. In other words, the internal rotations do not cause large conformational changes in the molecules. One may consider that if these two rather extreme cases can be fully understood, then the intermediate molecules might be more easily inter- preted. We shall show in a later part of this review, however, that one of the intermediate cases, acetonitrile-water mixtures, is actually highly complicated and interesting. 2 Method of Molecular Simulation The method of molecular simulation as applied to aqueous solutions of non-electrolytes consists of the following several processes.2.1 Selection of Potential Functions In almost all of the molecular simulations attempted so far, the pairwise additivity of molecular interactions is adopted. Thus, three kinds of pair potentials are necessary for binary mixtures. In the present case, one component is always water. For water- water interactions, some empirical and non-empirical models are available. We adopt the so-called TIP4P potential proposed by J~rgensen,~ and also the MCY potential by Clementi et We should point out that, although many water-water dimer potential functions including the above have been proposed,6 we are still looking for a better one. If we term the second component ‘solute’, though somewhat ambiguously, we must have water-solute and solute-solute potential functions. They are not available, except for some limited cases.In the case of non-polar molecules, solute-solute potentials can be expressed by the simple Lennard-Jones potential. Otherwise, we must use semi-empirical methods or derive potential functions for a dimer (super-molecule) on the basis of quantum chemical molecular orbital (MO) calculations. As a semi-empirical method, Jorgen- 177 sen and co-workers have proposed a useful set of potential parameters called TIPS or OPLS for various kinds of atoms and atomic groups, in order to avoid the unnecessary repetition of MO calculations This is essentially a potential energy version of the group contribution concept Thus, semi-empirical potential functions can be prepared easily by combining their parameters The TIP4P potential for the water dimer mentioned above is a typical example OPLS potentials are adopted for the aceto- nitrile dimer7 and water-acetonitrile interactions 2.2 Preparation of the Potential Function In the general case where the above-mentioned semi-empirical potentials cannot be utilized, or are shown to be inappropriate, one must proceed to the so-called 'super-molecule method', with the aid of extensive MO calculations Potential functions for TBA dimer,s urea d~mer,~TBA-water,'O and urea-water1 have been developed in this manner The method consists of the following procedures (1) The molecular conformation of each molecule concerned should be known Sometimes information is available from molecular spectral data Otherwise, the most stable one can be determined by the energy gradient method in MO calculations (11) Energies of the two molecules concerned, say, 1 and 2, and of the super-molecule (1 + 2) are determined by MO calcula- tions, using the appropriate basis set Then intermolecular interactions between 1 and 2, 412(Y), can be obtained by where molecules 1 and 2 are identical in the case of pure fluids (111) The determination of 412(Y) is repeated for each different mutual configuration of (1 + 2) and simultaneous optimiza- tion of parameters in an assumed potential function The functional form of potential implies the proposition of a model In this manner, an assumption is introduced into the potential function (iv) When the further addition of different configurations does not lead to an improvement in the potential parameter set, we can assume that a reasonable 'non-empirical' potential func- tion has been obtained A representative example of this procedure may be found fully described in literature for the case of TBA-water lo 2.3 MD and MC Simulations The methodology of molecular simulation for liquids and liquid mixtures is now well documented Only some important points will be commented upon Monte Carlo calculations have usually been done in NTV and NPT ensembles by the use of the Metropolis scheme The number of molecules N is 256 and the temperature T of the system is 298 15 K throughout the present calculations In addition, either the volume of the cell V,from the density data, or the pressure P (normally 1 atm ) is kept constant The Metropolis scheme is essentially the generation of the configurations of a molecular ensemble under certain restric- tions, and the total potential energy of the ensemble tends to decrease normally until a stationary state is established One needs at least lo6 N configurations for the equilibrium of the system and a further 2 x lo6Nconfigurations for the formation of a meaningful statistical mechanical ensemble Various struc- tural and thermodynamic properties of the system studied can be obtained from MC ensemble data Conventional molecular dynamics calculations have nor-mally been performed for systems with spherical symmetrical potentials in a microanonical (NEV) ensemble by the use of the Verlet scheme Since the temperature T is obtained as a result of the calculation, the constant temperature MD, proposed by Andersen, is sometimes used The molecular dynamics calcu- lation is essentially a numerical integration of the Newton-Euler equation of motion for the molecular ensemble The time step is CHEMICAL SOCIETY REVIEWS.1993 0 005 picoseconds and the number of steps is as large as lo5 The integration extends to a total of 20 -40 ps The result of calculations, namely the position and velocity of each molecule as a function of time, forms a statistical MD ensemble descrip- tion, from which various static and dynamic properties of the system studied can be derived In the present case, attention is focused mainly on the structure of the solution The fundamental property used to describe structural characteristics is the radial distribution func- tion (RDF) g(r) In addition, one can make use of various kinds of energy-structure distribution functions, among which the pair interaction distribution function, PIDF, might be the most useful PIDF is the distribution of the interaction energy of each pair of molecules in the system 3 Aqueous Solutions of TBA Among aliphatic alcohols miscible with water in all proportions, TBA has the largest and most compact hydrophobic group The minimum in the partial molar volume-composition relation in TBA aqueous solution is the most characteristic among organic non-electrolytic solutes This means that hydrophobic effects are largest in this solution Results of extensive MC and constant temperature MD calculations are available for 0 5,3 0, 8 0, and 17 mol YOaqueous solutions of TBA with the potential models described above lo l4 In the case of an infinitely dilute solution (1 TBA in 215 water = 0 5 rnol %), a hydration structure is clearly formed Figure 2 shows density diagrams for the distribution of the water molecules surrounding TBA For a comparison, similar density diagrams are included in the figure for water molecules around methanol or one specified water molecule This is essentially a graphic representation of the angle-dependent g(r) (or more exactly, pair correlation functions between the centres of mass of alcohol and water) It is seen that, while the hydration structure around methanol is spherical, around TBA it is rather spot-like, suggesting the formation of a stable hydration shell in the latter case Moreover, it is interesting to note that, as seen in Figure 3, the average potential energy of water molecules in the hydration shell of the hydrophobic groups of TBA is lower than that near the hydroxyl group of TBA Although qualitative observations, they are evidence for the existence of hydrophobic hydration Similar information can be obtained from the RDF itself Indeed, the first peak of the water-water RDF becomes signifi- cantly higher when a small amount (3 mol O/O) of TBA is introduced into pure water This is direct evidence for the structure enhancement of water in such dilute solutions Kauzemann proposed three different liquid structures for pure water (ice), depending on the time scale of observation l6 They are I(instantane0us)-structure, V(vibrational1y-averaged)-structure, and D(diffusiona1ly-averaged)-structureBoth V- and I-structures obtained from the results of MD calculation will give useful information on hydrophobic hydration and interac- tion(H1) The V-structure can be seen in the trajectory diagram obtained from the results of the MD calculation Here HI implies a kind of self-association of non-polar molecules in aqueous solution or of the tendency of non-polar groups to interact with each other Although not to be discussed in more detail here, this kind of hydrophobic interaction is proved to be of solvent-separated type(SSHI), suggested by Franks The presence of HI can be seen most definitely in the PIDF shown in Figure 4a, where a large peak for TBA-TBA interactions appears between -5 to + 10 kJ mol and the distribution completely disappears in the lower energy region, indicating the lack of hydrogen bonding interactions between TBA molecules Howevpr, as shown in Figure 4b, the TBA- TBA interaction has a small but definite peak between -10 and -20 kJ mol- in a more concentrated (17 mol YO)solution, indicating the presence of hydrogen bonding between TBA molecules in such a TBA/water molar ratio Summarizing the general impression obtained from MD and MC data, the introduction of TBA into water leads to an COMPUTER SIMULATIONS ON AQUEOUS SOLUTIONS OF SOME NON-ELECTROLYTES-K.NAKANISHI Figure 2 Density distribution diagrams of water molecules around water, methanol, and 2-methylpropan-2-01: (upper) oxygen atom, (lower) hydrogen atoms. -33 --34 -4 Aqueous Solutions of Urea Effects of urea and its alkyl derivatives on water structure have been studied extensively by experimental and model approaches. Conflicting interpretations are given for the results of these studies: one insists that urea may destroy water structure, while another suggests that urea may promote water structure. In addition, urea behaves as a protein denaturing agent in concentrated aqueous solutions.It is thus important to clarify at the molecular level the influence of urea on hydropho- bic effects. As previously noted, urea occupies a position on the hydro- 1-E"2 -35 -\ 1 -36 --37 --38 - I I I I I I 30 90 150 210 270 330 360 Wdeg Figure3 Angular dependence of the potential energy of water molecules hydrated around 2-methylpropan-2-01 (for details see ref. 10). enhancement of aqueous structure, an energetic stabilization of water molecules near hydrophobic groups of TBA, and solvent- separated hydrophobic interactions between TBA molecules. With an increase in TBA concentration, hydrophobic interac- tions, namely contacts between hydrophobic groups, are predo- minant below 10 mol YOsolution.The formation of hydrogen bonding between TBA molecules cannot be observed in this region. It is only above 10 mol YOthat the hydrogen bonds appear between TBA molecules. philic side of the series given in Figure 1. Urea seems to be a good partner of TBA, because of the conflicting effects on water structure. The procedure of determining a potential function, as explained in Section 2, is quite suitable for urea because it is a rather rigid molecule, internal rotations being unimportant, and there are strong hydrogen bonding interactions. There exist non-empirical potential functions for urea-water' and urea-urea9 interactions, and the results of MD simulations are a~ailable.~,~ According to quantum mecha- 1,1 nical MO calculations, water and urea form a ring-like dimer and the intermolecular energy is estimated to be as large as -40 kJ mol- l.However, an MD simulation of a 0.5 mol O/O (1 urea in 215 water) urea solution indicates that the structure of water is practically identical to that of pure water, except for a few water molecules attached to hydrogen bonding sites of urea. In Figure 5, the density distribution diagram for the distribution of water molecules around urea(D-structure) is superimposed on an isoenergy contour map for the urea-water potential. It is seen that the water distribution outside the hydration shell around urea has no special structural characteristics. There seems to be five strong hydrogen bonding sites in the hydration shell. Of these, one is at the potential minimum between two amino groups.The other four sites near the carbonyl group do not correspond to the minima in the urea-water potential. The resultant disadvantage in the energetic stabilization can be compensated by the harmonious connections between the hyd- ration shell and the surrounding water structures. 180 20 a 16 12 hQ Pure Water f 0 7 8 4 0 -20 0 20 V/kJ mol-' Figure 4 Pair interaction distribution functions in aqueous solutions of 2-methylpropan-2-01: (a) 8 mol%, (b) 17 mol%. Figure 5 Comparison of density distribution diagram of water mole- cules around urea with the isoenergy contour map of urea + water potential.In more concentrated solutions (8 and 17 mol YOsolutions), urea molecules tend to self-associate as seen by graphical presentations of I-and V-str~ctures.~~~ The pair interaction distribution functions plotted in Figure 6 can also afford inter- esting information on the self-association of urea. One can see two large peaks for urea-urea interactions near -20 --40 and -60 --80 kJ mol-', respectively. These are due to the self-association of urea. In contrast to the hydrophobic interac- tion in dilute TBA solution, it is through these strong hydrogen bonding interactions that urea can self-associate in water. 5 Aqueous Solutions of Acetonitrile As has been shown in the above two sections, TBA and urea represent two extreme cases of solutes in water and one expects that the behaviour of many other amphiphiles may easily be understood as intermediate between them without any special complication.This expectation has not been borne out in recent studies on acetonitrile (AN), however.20,2 There are many studies dealing with the thermodynamic CHEMICAL SOCIETY REVIEWS, 1993 VIkJ mol-' 20 :I j 16 ;I : 12 *Q 0 7 8 4 ,.l--I , , I I0-ioo -80 -60 -40 -20 0 20 40 V/kJ rnol-' Figure 6 Pair interaction distribution functions in aqueous solutions of urea. properties and solution structure for aqueous mixtures of this useful solvent. Computer simulation studies have also been carried out.22 It has been suggested that AN may form dimers in dilute aqueous solution.This should have a close connection with the fact that, although there is endothermic mixing for almost all compositions of AN-water mixtures, mixing is slightly exothermic in the highly dilute region of AN.23 The results of standard NPT Monte Carlo calculations are available for the following conditions and models.20 The tem- perature and pressure of the systems are 298.15 K and 1 atmosphere; the total number of molecules in the system is 21 6; the TIP4P model water-water potential and a three-site model for AN-AN were used. The water-AN potential used is of the conventional LJ x Coulomb type and the unlike interactions assumed are the geometric mean of each set of LJ parameters. This is somewhat different from the usual Lorent-Berthelot combining rule, but the difference is of a minor degree.Some details of molecular models are shown in Figure 7. Monte Carlo simulations cover the whole composition range of water + AN mixtures with 12 different molar concentrations. Particular emphasis is placed on the water-rich region. Much energetic and structural information has been obtained. The most direct results from the MC calculation are the excess molar enthalpies HE of water + AN solutions. In Figure 8, computer generated HEvalues are compared with experimental results.23 It is seen that the real mixtures are less endothermic but the tendency of exothermic mixing in the water-rich region can be well reproduced. The potential functions used are the best ones available for the two pure fluids and the results may COMPUTER SIMULATIONS ON AQUEOUS SOLUTIONS OF SOME NON-ELECTROLYTES-K NAKANISHI Acetonitrile (3 site model) do Water (TIP4P) (0 52e) Figure 7 Model for acetonitrile + water interactions 500 000 500 0 I....!I: ....1 00 05 10 Mole fraction of acetonitnle Figure 8 Molar excess enthalpies of acetonitrile + water mixtures Comparison between Monte Carlo(-) and calorimetric( ) results indicate that there is room for improving the agreement between simulation and experiment by using a more appropriate combin- ing rule for unlike interactions The solution structure can be quantitatively discussed by calculating various energetic and structural distribution func- tions However, a graphic display can furnish more impressive information Figures 9 and 10 show solution structures of some water-AN mixtures It is clearly seen that in water-rich regions, where the excess enthalpy is negative, two AN molecules tend to associate It IS also observed that in the equimolar region both water and AN molecules show clustering of the same kind, in other words, they are not in monodisperse states The presence of such ‘local heterogeneity’ might have a close relation with a large positive excess enthalpy The further analysis of pair formation of AN in dilute aqueous solutions would be impossible by conventional thermo- dynamic approaches Instead, one can use a statistical mechani- cal approach, namely, RISM(reference interaction site model) theory 24 Although this RISM theory has been applied to pure fluids, there is difficulty in its application to fluid mixtures However, dilute solutions containing only two AN molecules could be treated in the following manner 21 (1) Consider two AN solutes with a fixed mutual configuration as a super-molecule (AN),, the RISM calculation can be applied to the infinite dilution state of this molecule (11) The hydration energy 1s calculated for an infinite dilution state of (AN), Figure 9 Structure of acetonitrile + water mixture Two AN molecules in 214 water molecules Figure 10 Structure of acetonitrile + water mixture Clustering of AN(upper) and water(1ower) in equimolar solution 182 (iii) The above calculations are repeated by changing the confi- guration of the super-molecule.The most stable configuration of (AN), obtained from this calculation is such that the angle between two AN molecules is 60” and the distance between centres of mass is 0.4 nm (see Figure 11). Contrary to expectation, the CN group in one AN molecule comes into close contact with the CN group in the other AN molecule. This kind of ‘parallel’ orientation is the result of the potential of mean force from the surrounding water; the solute-solute interaction is repulsive (-11 kJ mol- l); contributions from solvent water are as large as -25 kJ mol- l, and resultant net energetic stabilization due to the pair forma- tion is thus about -14 kJ mol-l. Figure 11 Stable structure of acetonitrile dimer in water In order to confirm such unexpected pair formation, an MD simulation has also been done under the same conditions as those in the above RISM study.*’ The result clearly shows a long contact period between two CN groups.Figure 12is a plot of the site-site distance as a function of time steps in the MD simulation. I I I I i I I I 20000 40000 MD Steps Figure 12 Molecular dynamics results of two acetonitrile molecules in dilute aqueous solution. 6 Concluding Remarks Recent advances in computer simulation have made it possible to calculate various thermodynamic and transport properties of CHEMICAL SOCIETY REVIEWS, 1993 aqueous solutions of non-electrolytes. If calculated values of many physical properties agree consistently with experimental results, then various types of information on the molecular level obtained concurrently should generally give a valid picture.The MD and MC studies for TBA and urea solutions described in Sections 3 and 4furnish not only complementary information to various experimental studies but also give a more detailed insight at the molecular level. Similar studies on methanol’ 5.25 and some other solutes (eg. non-polar fluoroalco-h~ls,~’elc.) in water have already been performed. As an example, one can also see reasonable agreement of the calcu- lated excess enthalpy of AN-water mixtures with the experimen- tal data. However, further improvement and extensions of the method of molecular simulation are required because, for example, the calculation of the excess free energy cannot always be performed with confidence. Furthermore, new molecular approaches can- not be complete unless fluctuations and dynamics in liquid mixtures are discussed on a sound theoretical basis. Examples of this kind of approach, namely, one-particle dynamics, local equilibrium searches, and normal mode analysis, are already being applied to aqueous solutions of non-polar molecules.28 7 References 1 F.Franks, ‘Water: a Comprehensive Treatise’, Vol. 2, Plenum, New York, 1973. 2 J. S. Rowlinson, ‘Liquids and Liquid Mixtures’, 3rd Edn. 3 M. P. Allen and D. J. Tildesley, ‘Computer Simulation of Liquids’, Clarendon Press, Oxford, 1987. 4 W. L.Jorgensen, J. Chem. Phys., 1979,71, 5034. 5 0.Matsuoka, E. Clementi, and M. Yoshimine, J. Chem. Phys., 1976, 60, 1351. 6 J. L. Finney, J. E. Quin, and J. 0.Baum, Water Science Review, 1985, 1, 93. 7 M. L. Jorgensen and J. M. Briggs, Mol. Phys., 1988,63, 547. 8 H. Tanaka, H. Touhara, K. Nakanishi, and N. Watanabe, J. Chem. Phys., 1984, 80, 5170. 9 H. Tanaka, K. Nakanishi, and H. Touhara, J. Chem. Phys., 1985,82, 5 184. 10 K. Nakanishi, K. Ikari, S.Okazaki, and H. Touhara, J. Chem. Phys., 1984,80, 1656. I1 H. Tanaka, H. Touhara, and K. Nakanishi, J. Chem. Phys., 1984,81, 4065. 12 N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and E. Teller, J. Chem. Phys., 1953, 21, 1087. 13 L. Verlet, Phys. Rev., 1967, 159,98. 14 H. Tanaka and K. Nakanishi, Fluid Phase Equi., 1993, 83, 77. 15 S. Okazaki, H. Touhara, and K. Nakanishi, J. Chem Phys., 1983,78, 454. 16 D. Eisenberg and W. Kauzmann, ‘Structure and Properties of Water’, Oxford University Press, London, 1969. 17 F. Franks, Faraday Symp. Chem. Soc., 1982,17, 11. 18 (a)G. C. Kreshech and H. A. Scheraga, J. Phys. Chem., 1965,69, 1704. (b) R. H. Stokes, Aust. J. Chem., 1967, 20, 2987. (c) Y. Mizutani, Y. Kamogawa, and K. Nakanishi, J. Phys. Chem., 1989, 93, 5650. 19 R. A. Kuharski and R. J. Rossky, J. Am. Chem. SOC.,1984, 106, 7596. 20 Y. Sat0 and K. Nakanishi, to be published. 21 M. Matsumoto and K. Nakanishi, to be published. 22 H. Kovacs and A. Laaksonen, J. Am. Chem. Soc., 1991, 113,5596. 23 K. W. Morcom and R. N. Smith, J. Chem. Thermodyn., 1969,1,503. 24 D. Chandler and H. C. Andersen, J. Chem. Phys., 1972,57, 1930. 25 H. Tanaka and K. G. Gubbins, J. Chem. Phys., 1992,97,2626 26 S. Okazaki, K. Nakanishi, H. Touhara, and Y. Adachi, J. Chem. Phys., 1979, 71, 2421. 27 K. Kinugawa and K. Nakanishi, J. Chem. Phys., 1988,89,5834 28 H. Tanaka and K. Nakanishi, J Chrm. Phys., 1991,95,3717

ISSN:0306-0012    

DOI:10.1039/CS9932200177

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Biosynthetic Incorporation of Non-natural Amino Acids into Proteins Josef Brunner Department of Biochemistry ETH -Zentrum CH-8092 Zurich Switzerland 1 Introduction Since the elucidation of the structure of DNA in 1953 -the beginning of the modern era of molecular biology -many successful studies have been performed on the immensely com- plex and interrelated processes by which DNA is copied and then transcribed into RNA and how the four letter alphabet of the latter is translated into the entirely different alphabet of polypeptides. One of the most fascinating aspects that has emerged from this work is the extraordinary accuracy with which these processes occur. Indeed the maximal frequency with which wrong amino acids are inserted into a polypeptide chain is between only 1 in 1000and 1 in 10000.To achieve such fidelity the coordinated interplay of a large number of macro- molecules is necessary. Although a few synthetic nearly isosteric analogues of amino acids such as 3-fluorotyrosine or 6-fluorotryptophan are known to be accepted by the biosynthetic machinery it is quite clear that a general strategy to incorporate non-natural amino acids requires specific intervention at one or more steps of the translation. During recent years a general methodology has been developed which now permits a broad range of non-standard or non-natural amino acids to be incorporated at rationally selected sites and hence to generate polypeptides with novel structural and functional features. This offers a number of intriguing opportunities with far-reaching implications for both basic research and for the engineering of proteins in order to create novel types of catalysts or carriers of biologically active residues with pharmacologically interesting properties. This review is intended to give a brief summary of some of the more recent developments and achievements.The first part is concerned with a discussion of some of the basic elements of translation and of the manipulations that allow the bypassing of step(s) determining the high specificity of translation. In a subsequent section the experimental approaches which have been developed to build up and assemble the molecules essential for these manipulations are outlined. The third part then summarizes some of the results that demonstrate the successful incorporation of non-natural amino acids into proteins. A final section will be devoted to some applications of this new techno- logy. For reasons to be detailed there we shall examine some- what more closely the biosynthetic incorporation of photoacti- Josef Brunner u‘as born in 1943. After an apprenticeship as a laboratory technician study of chemistry at the HTL Burgdocs (I 964-471 and three years in industry he began studying biochemistry at the ETH Ziir-ich from ,tihere he received his Ph.D. (M-ith G. Semenza) in 1977. After postdoctoral Ltiork at Yale with Fred Richards (1977-79) he returned to the ETH ct.here lie has been an Oberassistant in the Depart- ment of Biochemistrjy. His principal research interest lies in the development of photo-crosslinking methods.for studjj- ing biologicul membranes. vatable amino acids into nascent polypeptides and the use of these polypeptides to identify biological targets. 2The Role ofTransfer RNAs in Protein Biosynthes i s As mentioned the ribosome-catalysed synthesis of protein is an immensely complex process and many important details still remain to be elucidated (an excellent overview is found in Stryer’s textbook ‘Biochemistry”). To understand the manipu- lations necessary to incorporate non-natural amino acids into polypeptides we must focus on the role of transfer RNAs (tRNAs) in the translation of the genetic information. The coded information is in the sequential triplets of messenger RNA (mRNA) each of which specifies a particular amino acid. In each cell there is a set of tRNAs which is responsible for recognizing each of the 61 amino acid-specifying triplets (codons) and for carrying the cognate activated amino acid to the ribosome where it participates in peptide synthesis.Thus t RNAs act as the crossroads between the information-contain- ing polynucleotide chains of the mRNA and those sites in the ribosome where this genetic information is expressed into the 20-letter alphabet of polypeptides. All cells are also endowed with a set of (at least) twenty enzymes called aminoacyl-tRNA synthe- tases. Each aminoacyl-tRNA synthetase recognizes in a highly specific manner one of the 20 proteinogenic amino acids and catalyses the formation of an ester bond between the amino acid and the cognate tRNA molecule(s). In general more than one species of tRNA can act as an acceptor for a particular amino acid and because of the ‘wobble’ in base-pairing some tRNAs can recognize more than one codon in the mRNA. Of the total 64 possible triplets 61 code for the 20 amino acids whereas the other 3 signal chain termination and normally are not read by tRNAs but rather recognized by proteins called release factors. Binding of a release factor to a termination codon somehow ‘activates’ peptidyl transferase so that it hydrolyses the peptidyl- tRNA ester bond allowing the polypeptide chain to leave the ribosome. As has become more and more clear during the past few years in addition to this normal decoding translation can involve non-standard decoding events such as ribosomal hop- ping frameshifting and reading through stop codons all at unexpectedly high levels and using surprising mechanisms. These are distinct from random translational errors and are largely determined by the structure of the mRNA being trans- lated. It is important to refer to this fact as it implies that the protein sequence cannot always be deduced from the nucleic acid sequence of the mRNA. Transfer RNA molecules are small polynucleotides contain- ing between 73 and 93 nucleotides. About half of their bases are paired and the molecules fold up into an L-shaped structure with two segments of double helices. As might be expected tRNAs contain two functionally important elements the amino acid attachment site which is at one end of the folded molecules and ~the codon recognition site the so-called anticodon -located at the other end of the structure. During protein synthesis the three anticodon bases interact with the codon bases of the mRNA positioning the amino acylmoiety such that it can participate in the peptidyl transferase reaction. At no time does the side chain of the amino acid directly interact with the mRNA template. One of the constant features of tRNAs is the sequence of four non-paired residues at the amino acid acceptor (3‘) end which all terminate in -CpCpA. Recent evidence suggests that this part of the tRNA interacts with a distinct highly conserved 183 region of 23s ribosomal RNA (rRNA) a finding which lends further support to the exciting hypothesis3 that peptide bond formation is catalysed solely by rRNAThis constant element of the amino acid acceptor end also provided a base for the development of a general methodology for the chemical misacyl- ation of tRNAs All (natural) tRNA molecules contain a number of unusual bases formed by enzymatic modification of the tRNA precursor moleculesThe role of these modifications is not fully clear as yet but as will be shown in this article they are not essential for effective participation in the peptidyl transferase reactionThey may be important however for proper recognition by the aminoacyl tRNA synthetases and also seem to stabilize the tRNA three-dimensional structure More than thirty years ago Chapeville et ul provided the first direct proof for the adapter function of the tRNA by showing that after the reductive desulfuration of cysteinyl- tRNACySto alanyl-tRNA cYs this alanyl residue is incorporated in response to a triplet coding for cysteine Using a related experi- mental approach Johnson et a1 later demonstrated that ribo- somes also accept structurally modified amino acids A lysyl- tRNA acetylated at the N‘-amino group was found to be incorporated into rabbit globin chains with nearly the same efficiency as the unmodified lysine 3 Experimental Approaches To site-specifically incorporate non-natural amino acids into polypeptides by ribosome-based translation a number of requirements must be fulfilled First an mRNA must be gener- ated in which the codon corresponding to the amino acid residue of interest is replaced by a codon not specifying one of the natural amino acids Secondly a tRNA capable of recognizing this ‘extra’ codon must be constructedThirdly a procedure must be found to charge this tRNA properly with the non- natural amino acid And fourthly a translation system compat- ible with the afore mentioned manipulations is required 3.1 In Vivo and in vitroTranslation The most common and most efficient translation systems are living cells However since all cells are surrounded by a mem- brane which acts as an effective diffusion barrier specific manipulation of an endogenous tRNA component in the man- ner outlined above is an extremely difficult task not attainable at present However it remains a great challenge for the future to overcome the various problems and to programme and manipu- late lzvzng cells in a way such that they take up a non-natural amino acid and efficiently incorporate it into a protein compo- nentThe reasons for this are that cells especially bacteria can grow on relatively cheap nutrients and produce high concent- rations of protein factors which are both important from a technological perspective of view Moreover eukaryotic cells can carry out a number of post-translational modifications such as proteolysis glycosylation phosphorylation sulfatation fatty acylation isoprenylation etc many of which cannot be obtained or with great difficulty only in m vztro systems (see below) For the proper functioning of many proteins post- translational modifications are essential An attractive zn vzvo system is offered by the oocytes of Xenopus luevuThese cells not only have an impressive diameter of about 1 mm but a significant proportion of their content is represented by components required for the post-fertilization protein synthesis during early embryonic developmentThese two facets have made oocytes a powerful system for studies of transcription replication assembly and translation of micro- injected macromolecules Although direct experimental evi- dence is still lacking it is quite certain that oocytes would also utilize micro-injected chemically misaminoacylated tRNA and produce very small amounts of mutant proteins containing a non-natural amino acid Although the amounts of protein generated in this way would hardly exceed those obtained from the more simple and economical zn vztro translation systems (see CHEMICAL SOCIETY REVIEWS 1993 below) the oocytes can correctly carry out post-translational modifications and direct proteins to specific intracellular com- partments and even export secretory proteins Thus far all successful approaches to incorporate non- natural amino acids into proteins have made use of suitably programmed zn vztro translation systems supplemented with an exogenous misacylated tRNA In vztro or cell-free translation is a well established and important technique to generate small amounts of protein Appropriate systems can be prepared easily from bacteria yeast as well as from cells of higher eukaryotes (plants animals) Most frequently used are extracts from Escherzchza colz and wheat germ and cell lysates from rabbit reticulocytes (erythrocyte precursors that make large amounts of globin) Before use these extracts are usually depleted of endogenous amino acids and energy sources and treated with ribonuclease to destroy their own mRNA Protein synthesis is then initiated by the addition of mRNA (obtained from natural sources or from zn vztro transcription of plasmid DNA) amino acids ATP GTP and an ATP-generating system consisting of creatine phosphate and creatine phosphokinase As mentioned previously one of the limitations of a cell-free translation is the relatively low yield of proteinTypically only a few polypeptide chains are produced per mRNA chain which corresponds to a few micrograms or a few tens of micrograms of protein per millilitre of cell lysate or extract It should be noted however that there may be ways to greatly increase translational efficien- ciesThus using continuous-flow systems zn vztro systems have been shown to be active for tens of hours and to produce up to several hundreds of polypeptide chains per copy of mRNA ’ 3.2 Choice of the tRNA Two conceptually different approaches have been described for the site-specific incorporation of non-natural amino acids into proteins In the first approach.* advantage is taken of the fact that either of the three stop codons that normally signal chain termination can also function to direct incorporation of an amino acid provided a charged tRNA is available that can effectively compete with the release factor(s) in reading the ‘stop’ codonThis principle called suppression is used by both bacteria and eukaryotic cells for example for the translational insertion of the non-standard amino acid selenocysteine into formate dehydrogenase in E colz and into glutathione peroxi- dases of mammalian cells In the second approach an extended genetic alphabet including a new base pair with a non-standard hydrogen bonding pattern is used to construct a 65th codon- anticodon pair l1 3 2 1 Generation of Suppressor tRNA For the purpose of incorporating a non-natural amino acid into a protein in response to a stop codon a suppressor tRNA IS needed that can effectively compete with the release factors Even though both bacteria and eukaryotic cells can under selective pressure make and use suppressor tRNAs suppression is usually weak Were it otherwise this would probably create a very serious problem as it would be generally difficult for the cell to terminate the translation of proteins correctly For this as well as for other reasons efficient suppressor tRNAs cannot be (easily) obtained from natural sources Efficient suppressor tRNAs can be constructed however by one of three procedures In the first the anticodon-replacement procedure of Bruce and Uhlenbeck * phenylalanine-specific tRNA from yeast (tRNAPhe) serves as the starting material (Figure 1)The presence of a hypermodified highly acid sensi- tive base (wyosine) within the anticodon loop of this tRNA provides an opportunity to open the loop chemically and remove the three adjacent anticodon bases by controlled diges- tion with ribonucleaseThis treatment also removes the two terminal nucleotides at the 3’-amino acid acceptor endThe anticodon loop is then restored by enzymatically inserting a synthetic tetranucleotide containing the nucleotide sequence BIOSYNTHETIC INCORPORATION OF NON-NATURAL AMINO ACIDS INTO PROTEINS-J BRUNNER AOH C C A CAP I I5 115' b) aniline yeast tRNAPhe CPUPApAT4 RNA-ligase 3 cn COH/ a)T4 polynucleotidkrnase ATP b) T4 RNA-ligase Figure 1 Steps involved in the construction of a 3'-abbreviated amber suppressor tRNA from yeast tRNAPhe by the anticodon replacement procedure complementary to the stop codon The missing nucleotides at the 3'-terminus are replaced during chemical aminoacylation (see below) The second procedure to prepare a suppressor tRNA is by run-off transcription of a DNA template which can be made by either of two related methods l3 The first utilizes plasmid amplification of a DNA template which contains both a T7 promoter and a suitable restriction site at the 3'-end of the tRNA gene Linearization of the overproduced plasmid followed by run-off transcription then affords the RNA with the desired sequence Alternatively functional suppressor tRNA can be obtained byT7 polymerase-catalysed transcription of a synthe- tic linear DNA template featuring a double stranded promoter region and a long 5'-overhang corresponding to the transcribed region By either of these methods milligram amounts of tRNA can be produced Unlike natural tRNA however these run-off transcripts lack hypermodified bases and there is a triphosphate rather than a monophosphate at their 5'-end The presence of this 5'-triphosphate was found to have no recognizable effect on the tRNA function during translation Third and finally it may be pointed out that functional tRNAs have also been obtained by solid phase chemical synthe- sis using the phosphoramidite method for nucleotide cou-pling l6 However despite the steady improvements of the corresponding methods the preparation of a ribonucleic acid of the size of a tRNA still represents a considerable task Because of the necessity to protect the 2'-hydroxyl group of the individual building blocks internucleotide phosphate bond formation is sterically hindered and slower than for the corresponding desoxyribonucleotides With the introduction and use of the N,N-diethylphosphoramidite ribonucleosides which are more reactive than the older diisopropylphosphoramidites and the use of optimized protocols for coupling and deprotection a 74- mer suppressor t RNA could be synthesized which after chemi- cal aminoacylation displayed the same suppression efficiency as the corresponding tRNA (same sequence) obtained by run-off transcription 3 2 2 Non-standard tRNA As discussed by Benner and co-workers O the geometry of the Watson-Crick base pairs can accommodate at least six mutually exclusive hydrogen bonding schemes of which for whatever reason only two are used by natural oligonucleotides One of the additional pairs is formed by iso-C and iso-G (Figure 2) In a joint effort the groups of Benner and Chamberlin' succeeded G Is0 0 OH \/ Figure 2 Structure of the iso-G-iso-C pair In forming a non-standard hydrogen bonding pattern in constructing by chemical synthesis a tRNA possessing the anticodon CU(iso-G) When chemically charged with an amino acid this tRNA correctly reads the complementary codon (iso- C)AG introduced semisynthetically into an appropriate mRNA (see below) 3.3 Aminoacylation of tRNA Although a variety of conditions (for example in heterologous systems and in the presence of organic solvents) have been reported to lead to increased levels of misacylation of tRNA natural aminoacyl-tRNA synthetases because of their specifi- city are unlikely to provide generally useful catalysts for charg- ing tRNA with non-natural amino acids More promising in this respect might be synthetases genetically engineered to display altered substrate specificities In this way it was possible to activate a number of phenylalanine analogues l7 In the longer term it may even be possible to replace synthetases by anti- bodies that catalyse the aminoacyl transfer reaction Initial work towards exploring this possibility was reported recently by Schultz and colleagues *They examined antibodies raised against a phosphonate diester which contains elements of the acyl donor the acyl acceptor and the leaving group in a tetrahedral geometry mimicking that of the probable transition state for the transesterification reaction One of the antibodies was indeed found to be a remarkably efficient catalyst that provided further important insight into the requirements for efficient aminoacylation catalysts Another approach to generate misaminoacylated tRNAs is by chemical modification of the amino acid side chain of natural aminoacyl-tRNAs However because of the high degree of selectivity required to modify the amino acid side chain of dn entire aminoacyl-tRNA molecule the scope of this approach is very limited Examples where this approach has been applied successfully are the above mentioned conversion of cysteinyl- tRNACyS into alanyl-tRNACYS and the acylation of the €-amino group of lysyl-tRNA (see also below) The only truly general method for the aminoacylation of tRNA is that developed by Hecht and his group l9 21 Key elements of this method are the chemical (amino)acylation of the 3'(2') hydroxyl group of the dinucleotide pCpA (or pdCpA) and the subsequentT4 RNA Iigase-mediated coupling of this amino- acylated pCpA (pdCpA) to an abbreviated tRNA missing the 3'- terminal nucleotides pCpA (Figure 3) Several procedures employing different protection/deprotec- tion schemes have been developed for the (amino)acylation of the 3'-terminal dinucleotide (discussed in reference 22)To simplify the procedure and to improve the stability of the intermediates the original procedures have been modified in two important respects First the pCpA was replaced by pdCpA a DNA/RNA hybridThe absence of a free hydroxyl group within the cytosine ribose leads to fewer side-products during acylation and equally important eliminates the risk of H +-catalysed isomerization of the 3'-5'-phosphodiester bond during deprotection and purification of the (amino)acylated pCpA 23 Second for protection of the a-amino group of the CHEMICAL SOCIETY REVIEWS 1993 0 R -amino aadslde cham P -H or protectinggroup Figure 3The key step in the chemical aminoacylation of tRNA amino acid protecting groups are employed that can be removed after the ligation reaction ze on the intact tRNA Since the ester bond of aminoacyl-p(d)CpA and aminoacyl-tRNA are far more sensitive to hydrolysis than that of the a-amino group-protected counterparts both the preparation of acyl-p(d)CpA and the ligation reaction are greatly facilitated Among the protecting groups found to satisfy most needs are pyroglutamyl-and the nitroveratryl group which can be removed enzymatically and photolytically respectively (Figure 4)24 25 Other amino protecting groups also reported to be compatible with this general scheme are the nitrophenylsulfenyl (reductive removal) and the 2-(4-biphenyl)-prop-2-yloxycarbo-nyl (Bpoc) groups (acidic removal) Much progress has been made also in defining conditions favouring mono U-acylation of unprotected p(d)CpA Presu-mably the most simple and efficient method currently available is that developed by Schultz and co-workers 22 In this procedure unprotected pdCpA is acylated with the cyanomethylester-activated Na-(nitroveratry1)-blockedamino acid to give high yields (up to 80%) of the 3'(2')-O-acyl-pdCpA which without further purification could be used for the ligation reaction It may be of interest that preferential or selective O-acylation of unprotected pCpA can also be accomplished with N,N'-carbo-nyldiimidazole-activated Boc-protected amino acids using an acetonitnle-tetrahydrofuran-water solvent system 26 Although the procedure of Schultz has allowed the prep-aration of a broad variety of aminoacyl-tRNAs it should be kept in mind that amino acids containing reactive side-chain functions require additional or alternative protection For example the nitroveratryl group cannot be used to protect the amino group of the light-sensitive 4-(trifluoromethyldiazirinyl)-phenylalanine a precursor of a highly reactive carbene used for photocrosslinking purposes 23The final sections of this article contain a brief presentation and discussion of work aimed at acylating tRNA with an amino acid carrying additional sensitive functions including a radioisotope of high specificradioactivity Figure 4 Schemes for the deprotection of W-nltroveratryl-and pyroglu-tamoyl-protected aminoacyl-tRNAs 0 4 Participation of Misacylated tRNA in Protein Synthesis Having outlined the experimental strategies to prepare misacy-lated (acy1)aminoacyl-tRNAs this section focuses on the behav-lour of these speciesduring ribosome-catalysed protein-synthe-sisThe key step in polypeptide chain elongation is peptide bond formation a reaction catalysed by peptidyltransferaseThe substrates for this process are peptzdyl-tRNA and amznoacyl-tRNA which are bound to distinct sites on the ribosome referred to as the P site (the peptidyl donor) and the A site (the peptidyl acceptor) respectivelyTo further refine our under-standing of this reaction and to define the structural and spatial parameters requisite for effective participation of aminoacyl-and peptidyl-tRNAs Hecht and colleaguesprepared a variety of structurally modified peptidyl-tRNAs and examined their abi-lity to bind to the P-site and to function as peptidyl donors for phenylalanyl-tRNA in the ribosomal A site 27 28 A surprising variety of the modified peptidyl-tRNAs were found to partici-pate in ribosome-medlated peptide formation or even in forma-tion of products with altered connectivity As an example N-(ch1oroacetyl)phenylalanyl-tRNAPhegave rise to two 'peptides' the expected N-chloroacetyl phe-phe and the product resulting from reaction of the Wamino group of the phenylalanyl-tRNA with the peptidyl-tRNA at an electrophilic position different from the peptidyl-tRNA carboxylate ester (Figure 5)Thus there seems to be considerable flexibilityin the peptidyl transfer-ase reaction which may have to do with the fact that the ribosomal RNA plays a more active role in the peptidyl transfer-ase reaction and that peptide bond formation may actually be catalysed by this component and not by ribosomal proteins With the accessibility of chemically misacylated amznoacyl-tRNAs the above studies could be extended to include also experiments with structurally altered substrates (peptidyl-tRNA acceptors) for the ribosomal A site Most of this work was done by Schultz and his group who demonstrated that a range of backbone and side-chain modified amino acids can be incor-porated into growing polypeptidesThe successful replace-ments include a,a-disubstituted amino acids N-alkyl amino acids and lactic acid an isoelectronic analogue of alanine D-Amino acids were not incorporated It is possible however that the discrimination against D-amino acids arises in part at the level of formation of the tRNA-EF-Tu-GTP ternary complex tRNAo~;~o~ocH3I tRNAoGNH2hv 0 + O$ocH3 OCH3OCH OCH R -aminoacidsldechain R -amino acid slde chain BIOSYNTHETIC INCORPORATION OF NON-NATURAL AMINO ACIDS INTO PROTEINS-J. BRUNNER 2) OH-Figure 5 Reactions catalysed by ribosomal peptidyl transferase using a chemically charged structurally modified peptidyl-tRNA as a P-site substrate. which carries the aminoacylated tRNA to the ribosome rather than in the peptidyltransferase reaction itself. Ribosomes seem also to tolerate amino acids with abnormally bulky side chains as is the case for a number of “-modified lysines including W-biotinyllysine and the fluoroescent probe W-fluoresceinthio- carbomoyllysine. A detailed analysis of read-through suppression and site- specificity of the incorporation of a non-natural amino acid into a short model peptide was reported by Chamberlain and collea- gues.Translations were done with reticulocyte lysate in the presence of tRNA suppressors that had been obtained by run- off transcription of a synthetic DNA template and chemically misacylated with [’ 51]iodotyrosine. Under optimal conditions the efficiency of suppression of the UAG (amber) was 6370 and the [ 51]iodotyrosine was incorporated exclusively at the position of the UGA stop codon. Consistent with observations also made by Schultz’s group suppression with aminoacylated t RNAs lacking hypennodified bases was strongly dependent upon the Mg2+ concentration in the translation system. Chamberlain and colleagues also compared amber (UAG) opal (UGA) and ochre (UAA) suppressors. Interestingly the opal suppressor gave the highest level of suppression for incorpora- tion of [ SI]iodotyrosine (8l “/O) while the ochre suppressor gave the lowest level (48%) compared with 63% for the amber suppressor. Thus far the highest level of in vitro incorporation of a non- natural amino acid into a polypeptide chain was achieved by employing the non-standard tRNA possessing the anticodon CU(iso-dG) (tRNACU(,so-dG)) and an mRNA containing the complementary codon (iso-C)AG. While as noted above the level of suppression of stop codons was between 48 and 81% read-through of the (iso-C)AG codon was 90%.The specificity of translation of this 65th codon was also high. Neither semi- synthetic suppressor tRNAc-A nor any natural tRNA allowed reading of the (iso-C)AG codon. When the ribosome encoun- tered the (iso-C)AG codon in the absence of charged tRNA- CU(lso-dG) the primary outcome was continued translation following a frameshift that skipped the iso-C base a finding strongly supporting the view that termination of translation depends upon binding of release factors to stop codons to prevent frameshifting. 5 Areas of Application With the accessibility of proteins containing ‘designer’ amino acids at selected positions new avenues for research in several areas have been opened.The remainder of this article summar- izes some recent biochemical and biophysical analyses of mutated proteins carried out recently by Schultz’s group as well as efforts in our laboratory to utilize the technology for the preparation of a new generation of macromolecular photoaffi- nity- and photocrosslinking probes. 5.1 Biochemical and Biophysical Analyses of Mutated Proteins In a first study8 reported by Schultz’s group large scale in vitro translations were used to prepare truncated versions of 8-lactamases wherein Phe-66 a residue near the active site and conserved in four class A 8-lactamases was replaced by several analogues of Phe. Yields of mutated p-lactamase obtained in this E. coli translation system were estimated to be 5.5 to 7.5 pg per millilitre which represents 15-20% suppression efficiency (note that this suppression efficiency is much lower than those referred to above obtained with the reticulocyte lysate). While several of these mutant proteins could be purified to near homogeneity and characterized by their K and k, values those mutants containing substitutions corresponding to the greatest steric perturbations were unstable and presumably were proteolysed during purification attempts.This latter obser- vation is consistent with earlier mutagenesis experiments which demonstrated that Phe-66 plays an important structural role. The site-specific incorporation of amino acids with side chains that meet specific steric electronic and other constraints can greatly extend the scope of mutagenesis in studying the folding packing and stabilities of proteins as was demonstrated by anotherThere specific replacements of an interior residue inT4 lysozyme were used to evaluate the effects on protein stability of the stepwise removal of methyl groups from the hydrophobic core of side-chain solvation of packing density and of side-chain conformational entropy. Perhaps not surprisingly all the mutations examined significantly influenced protein stability a point to consider in any mutational study. Proteins containing isotopically substituted amino acids at defined positions offer new opportunities for studying proteins by NMR spectros~opy.~~ To illustrate this a single Ala residue (Ala-82) inT4 lysozyme was replaced by [13C]alanine. Using 13C-filtered ‘HNMR spectroscopy the a-H and the CH of this single alanyl residue were clearly visible implying that NMR spectroscopy of mutant proteins may be successfully used to determine chemical shifts pK values and relaxation para- meters for individual amino acids in both native and denatured proteins. In vitro translation may also allow incorporation of amino acids containing nitroxide free radicals into proteins. These could serve then as electron spin resonance probes for a wide range of structural and dynamic ~tudies.~ Notably the fact that spectra can be collected from 1Opl-samples containing micromolar or sub-micromolar concentrations of protein ( < 10 pmoles) would make such applications particularly attractive. An important question still to be answered is whether conditions for translation can be found that preserve the sensitive nitroxide functionality . 5.2 In vitro Synthesis of Photocrosslinking Probes The terms photocrosslinking and photoaffinity labelling refer to a variety of techniques that very often serve to identify the receptor for a given polypeptide or In the conven- tional approaches this is usually pursued by covalently cross- linking the two interacting species through the action of homo- and heterobifunctional reagents. Efforts are now being made to extend this popular technique by directly incorporating a pho- toactivatable amino acid at a single predetermined site into the ligand and hence to generate photoaffinity probes of defined structure and reactivity The potential of this approach is already evident from a respectable number of studies in which photoactivatable amino acids were translationally incorporated into proteins destined to be targeted to and into membrane vesicles derived from the endoplasmic reticulum (reviewed in reference 32) Every protein destined for a specific cellular membrane or sub-cellular com- partment contains a corresponding 'address label' In the case of the endoplasmic reticulum this is the so-called signal sequence a predominantly hydrophobic stretch of 15-25 amino acids at or near the N-terminus of the newly synthesized protein When this signal emerges from the ribosome it is first recognized by a cytosolic factor the signal recognition particle (SRP) whose binding slows down or even stops further transla- tion until the ribosome-SRP-nascent chain complex docks to a receptor in the membrane where the signal is transferred to integral components to initiate membrane insertion/transloca- tion By means of proteins containing photoactivatable amino acids within or near the signal sequence three components could thus far be identified which transiently contact the signal and are functionally engaged in the processes of protein targeting to and translocation across membranesThere is hope that basically the same approach can be applied to identify the 'receptors' for other signals especially for those whose recognition is directed by sequence-independent colligative properties As discussed by R~thman,~~ this principle of recognition is relatively new in biology and may be important in other arenas such as in the distinction between folded and unfolded proteins by heat shock proteins or the targeting of transport vesicles for membrane fusion Since a photoactivatable residue may be positioned within or near such a signal without seriously affecting its interaction with the target photoactivatable proteins could emerge as powerful tools to characterize the interactions under- lying cellular recognition folding and sorting events However before the potential of this methodology can be fully exploited a difficult technical problem has to be solvedThis has to do with the very limited capacities of the zn vztro systems for studying these interactions with typical yields of crosslinking products being in the femtomole or sub-femtomole range Unfortunately no general and efficient method is thus far available to identify and characterize the crosslinked target components At least a part of this problem might be solved by adopting a strategy ('label-transfer crosslinking') successfully employed in conventional crosslinking 32 Accordingly the side chain of the photoactivatable amino acid would have to be made cleavable and equipped within the photoactivatable moiety with a reporter group (Figure 6) Following translational incorporation of such an amino acid and its crosslinking to a target molecule cleavage of the susceptible bond would result in the formal transfer of the reporter group onto the target thereby yielding simply labelled targets which should now be far more easy to characterize than the original crosslinking product While it is not difficult to select the individual functional elements their joining together into a compact structure representing the side chain of a tRNA- activated amino acid is not a trivial taskThe difficulty arises mainly from the fact that only a radioisotope reporter group of high specific radioactivity (e g 32P,35S,or 12sI) would give the necessary detection sensitivity and would also be small enough to be compatible with this approach While 32P and 3sSare not available in synthetically convenient forms manipulation of lZ5I for safety reasons is usually restricted to millicurie or chemically speaking to nanomole quantitiesThis poses further severe constraints on the design and synthesis of such a multi- functional amino acid Work in our laboratory is directed towards realizing this task 26The basic approach chosen is summarized in Figure 7 In this approach the side chain sulfhydryl function of either cysteinyl-pCpA (prior to ligation to abbreviated tRNASUp) or cysteinyl-tRNAsup (after ligation) is alkylated with a reagent CHEMICAL SOCIETY REVIEWS. 1993 CH-0 + wpT I I I Figure 6 Schematic representation of 'LabelTransfer Crosslinking' The side chain of the crosslinking amino acid contains a photoactiva- table group (filled circle) a radioisotope reporter group of high specific radioactivity (asterisk) and a cleavable element (a pair of triangles) Following covalent photocrosslinking to a target molecule the susceptible element is cleaved whereby the reporter group is formally transferred onto the target component combining the necessary elements for label-transfer crosslink- ingThe rate at which cysteine and other thiols react with N-alkyllmaleimides suggested that the reaction would be fast and specific enough to successfully join the two reactants even though both might be present initially in sub-millimolar con- centrations and the reaction time would have to be kept short (minutes) to minimize hydrolysis of the aminoacyl-tRNA (aminoacyl-pCpA) ester bond 2'(3')-0-cysteinyl-pCpA could be obtained in approximately 20% yield by acylation of pCpA with N-Boc-dimethylthiazoli- dine and subsequent N,S-deprotection of the mono 0-acyl product by successive treatments with trifluoroacetic acid and water Ligation of cysteinyl-pCpA to abbreviated tRNASuP then S-alkylating cysteinyl-pCpA reagent S-alkyl-cysteinyl-pCpA mwsuP(CA)YT4 RNA-ligase S-alkylatingreagentTT4 RNA-lgasetRtdUp (CA) + cysteinyl-pCpA + cysteinyl-tRNAsuP Figure 7 Scheme for the preparation of multifunctional tRNA-acti- vated amino acids by S-alkylation of a cysteinyl side chain afforded the desired cysteinyl-tRNASUp We could show also that the thiol group of cysteinyl-pCpA is fully available for S-alkylation by various N-alkylmaleimides including [ IITIDM-3 a reagent which at least formally meets the criteria for a label transfer photocrosslinker (Figure 8) In contrast S-modification of cysteinyl-tRNASUp proved more difficult and yields ranging from 20-90% have been achieved Presumably during work-up and isolation of the cysteinyl-tRNASuP a fraction of the thiol underwent oxidation Current efforts are aimed at preparing cysteinyl-tRNAsuP alkyl- ated with [1251]TIDM-3 of high specific radioactivity (> 100 Ci mmol -') and at assessing translational incorporation of the BIOSYNTHETIC INCORPORATION OF NON-NATURAL AMINO ACIDS INTO PROTEINS-J BRUNNER [’251]TIDM-3 Figure 8 Structure of [I2 51]TIDM-3 a thiol-specific (N-alkylmalei- mide) cleavable (carboxylate ester) label-transfer photocrossllnker that can be prepared with an extremely high specific radioactivity ( > 2000 Ci mmol -I) modified cysteine into polypeptides Although this modified cysteine is an abnormally bulky residue initial data indicate that it is accepted by the ribosomes Moreover when placed within the signal sequence of a nascent preprolactin chain this chain can be photocrosslinked to the signal sequence-binding subunit of SRP strongly supporting the view that the amino acid replacement did not abolish proper recognition In conclusion proteins containing non-natural or modified natural amino acids at rationally selected sites provide novel and unique tools that will provide further exciting new insights into the structure and function of individual proteins as well as into their interactions with other proteins or components of such complex systems as biological membranes organelles or even whole cells However before the full potential of this technology can be realized a number of challenging chemical and biochemi- cal problems remain to be solved AcknowledgementsThe author thanks Drs N Mantel and B Martoglio andT Weber and C Gahwiler for assistance in the preparation of the manuscript It is also a pleasure to acknow- ledge the financial support byThe Swiss National Science FoundationThe work on label-transfer photocrosslinking was carried out in the laboratory of Prof Semenza 6 References 1 L Stryer ‘Biochemistry’ 3rd Edition Feeman & Company New York 1988 2 J F Atkins R B Weiss and R F Gesterland Cell 1990 62,413 3 H F Noller V Hoffarth and L Zimniak Science 1992,256 1416 4 F Chapeville F Lippmann G von Ehrenstein B Weisblum W J Ray and S Benzer Proc Natl Acad Sci USA 1962,48 1086 A E Johnson W R Woodward E Herbert and J R Mennmger Biochemistry 1976 15 569 6 A Colman in ‘Transcription andTranslation’ ed B D Hames dnd S J Higgins IRL Press 1984 Oxford p 271 7 A S Spirin,V I Baranov,L A Ryabova,S Y Ovodov,andY B Alkahov Nature 1988 242 1162 8 C J Noren S J Anthony-Cahill M C Griffith and P G Schultz Science 1989,244 182 9 J D Bain C G GlabeT A Dlx and A R Chamberlin J Am Chem Soc 1989,111,8013 J A PicanlliT Krauch S E Moroney and S A Benner Nature 1990,343,33 11 J D Bain C Switzer A R Chamberlin and S A Benner Nature 1992,356,537 12 A G Bruce and 0 C Uhlenbeck Biochemistry 1982 21 855 13 C J Noren S J Anthony-Cahill D J Suich K A Noren M C Griffith and P G Schultz Nucleic Acid Res 1990 18 83 14 J D Bain D A Wacker E E Kuo M H Lyttle dnd A R Chamberlin J Org Chem 1991,56,4615 J D Bain E S Diala C G Glabe D A Wacker M H LyttleT A Dix and A R Chamberlin Biochemutry 1991,30 541 1 16 M H Lyttle P B Wright N D Sinha J D Bain and A R Chamberlain J Org Chem 1991,56,4608 17 P Kast Ph DThesis 1991 ETH Zurich 18 J R Jacobsen J R Prudent L Kochersperger S Yonkovich and P G Schultz Science 1992 256 365 19 J M Pezzuto and S M Hecht J Biol Chem 1980,255,865 T G Heckler Y ZanmaT Naka and S M Hecht J Biol Chem 1983,258,4492 21 T G Heckler L-H Chang,Y Zama,T Naka M S Chorghade and S M Hecht Biochemistry 1984,23 1468 22 S A Robertson J A Ellman and P G Schultz J Am Chem Soc 1991,113,2722 23 G Baldini B Martoglio A Schachenmann C Zugliani and J Brunner Biochemistry 1988 27 795 1 24 J R Roesser C Xu R C Payne C K Surrdtt and S M Hecht Biochemistry 1989 28 5 185 J Ellman D Mendel S Anthony-Cahill C J Noren and P G Schultz Methods Enzymol 199 1,202 301 26 B Martoglio Ph D Thesis 1992 ETH Zurich 27 J R Roesser M S Chorghade and S M Hecht Biochemistry 1986,25 6361 28 T G Heckler J R Roesser C Xu P -I Chang and S M Hecht Biochemistry 1988,27 7254 29 D Mendell J A Ellman Z Chang D L Veenstra P A Kollman and P G Schultz Science 1992,256 1798 J A Ellman B F Volkman D Mendel P G Schultz and D E Wemmer J Am Chem Soc 1992 114 7959 31 G L Millhauser Trends Biochem Sci 1992 17,448 32 J Brunner Annu Rev Biochem 1993,62,483 33 J E Rothman Nature 1989,340,433

ISSN:0306-0012    

DOI:10.1039/CS9932200183

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Structural Systematics in Molecular Inorganic Chemistry A. Guy Orpen School of Chemistry, University of Bristol, Bristol BS8 I TS I Introduction The ability crystallography gives us to visualize the world at the atomic level has revolutionized the way we look at chemistry. From the first structure determination -of NaCl- to the current time, our whole view of the gamut of chemical structure, from the simplest ionic arrays to the most complex biomolecules, has been shaped by the results of crystallography. The diversity of molecular structure is perhaps most notable in the field of transition metal chemistry where many of the traditional tenets of organic structural chemistry have had to be repeatedly re- evaluated. Because of the clarity and unequivocal answers it offers, in large areas of modern chemistry X-ray crystal structure analysis is often the analytical method of choice.In many laboratories, when a new organometallic or coordination com- plex is prepared its X-ray crystal structure is determined for the sake of unambiguous identification (and to expedite publica- tion!). Often this is an end in itself and the level of detail provided by the structure analysis sufficient to provide insight into the properties of the compound, be they spectroscopic, or reactivity, or otherwise. This article, however, addresses other applications of structural data to chemistry which profit from the maturity of the techniques of crystal structure determination. The advent of computer-controlled diffractometers and the development of increasingly powerful software and hardware packages for the analysis of diffraction data from single crystals has led to an explosive growth in available crystal structure data.Fortunately, crystallographers have made great efforts to provide these data in readily useable, computer accessible form. The largest collection of crystallographic data on molecular structure is that contained in the Cambridge Structural Data- base (CSD).l This database is devoted to compounds of ‘organic carbon’. In practice about 50% of the over 100000 crystal structures it currently (January 1993) contains fall within the domain of molecular inorganic chemistry. It therefore holds nearly all the crystal structures of complexes of the transition metals.This is a colossal store of structural information on the chemistry of these elements and offers opportunities for whole Guy Orpen was born in 1955 on the island of Trinidad. He was educated in the UK and subsequently obtained his B.Sc. at the University of Cape Town where he worked with Dr. L. R. Nassimbeni. He studied for his Ph.D. at Cambridge University under the guidance ofDr. G.M. Sheldrick and Professor J. Lewis, during which time he was awarded a Walter C. Humilton Scholar- ship to carry out neutron difiraction research at Brookhaven National Laboratory, USA.In 1979 he was appointed to a lectureship in Inorganic Chemistry at the University of Bristol, and in 1990 was pro- moted to Reader in Structural Chemistry. He has been awarded the Meldola and the Corday-Morgan Medals ojthe Royal Society ofchemistry.In 1992 he held a Ciba-Geigy Senior Research Fellowship. He is author and co-author of over 200 research papers in the field of structural chemistry . new areas of research in which collections of structures, rather than individuals, are the subject matter of the investigation. The data contained within the CSD or other crystallographic databases can be of value in a number of ways. At the simplest level, literature surveys may be carried out, or compounds related to one currently under study may be retrieved for purposes of comparison. An alternative approach is to examine the geometry of a molecular or sub-molecular fragment in the variety of environments in which it has been found in crystals.This systematic type of study, as applied to problems in coordi- nation and organometallic chemistry, is the subject of this article. At the heart of such systematic studies is enquiry into the relationship between the molecular connectivity and the shape(s) that result from it, and the properties that in turn result from the molecular shape. 2 Systematic Studies Of the many possible applications of structural data to inorganic chemistry this article focuses on four. Examples will be taken from our own work as well as those of others in the field but the reader should be aware that this is very far from a full overview of this area of structural chemistry and is referred to more complete recent surveys.3 Typical Molecular Dimensions One major application of the CSD and other databases has been to provide information on ‘typical’ dimensions for molecular parameters, such as bond lengths and angles. The function of such information is to provide a knowledge base to support efforts to build realistic structural models for molecular species in cases where direct observation of the geometry is either not practical or impossible. Thus reliable standard starting geome- tries are invaluable as starting points for quantum mechanical optimization of a molecular geometry, for use in modelling the interaction of a small molecule with a surface or enzyme, for use in a constrained or restrained crystallographic refinement, or for model-fitting of e.g.EXAFS data. The question of what is a ‘typical’ geometry arises immedi- ately. In principle, one can conceive of two approaches. First, the structure-analysis of simple archetypal molecules (methane, benzene, water, Ni(C0)4 ) provides precise dimensions for C-H, C-C bonds etc. in well defined environments. The second alternative is to examine the corresponding bond type or func- tional group in a wider range of environments, thus exploiting a database of all the relevant structural data. The advantage of the second type of study is that the bond length (for example) under study is observed under a variety of molecular and crystal environments. This allows for some estimate of the effects of changes in the environment on the molecular parameter under study, as well as those due to experimental uncertainties. Thus both the mean value of a parameter and its variability are important pieces of information and both need to be thought about in the discussion of suitably ‘typical’ dimensions for a structural unit.In two large studies of this type we have compiled tables of bond lengths from the CSD for organic3 and organometallic compounds, and coordination complexes of the transition ele- ment~.~These compilations are a summary of a large body of structural information and therefore also of a large body of chemistry. In the latter table the information is presented according to ligand and covers both metal-ligand and intra- 191 ligand bond distances The chemical content is clear Thus we see how metal-phosphorus bond lengths in phosphite complexes differ from those for related phosphines, or how alkylisocya- nides and the cyanide ligand differ in dimensions In addition, information on the effects on bond distances of metal oxidation state or coordination number and estimates of the variability of the bond lengths are provided, where possible One clear mess- age of these estimates is the much wider variation of bond lengths in the transition metal study than for the organic table In part this reflects the softness of metal-ligand bonds, I e their sensitivity to their chemical and crystal environment In addi- tion it reflects unresolved chemical information, in the sense of other factors not always explicitly taken into account in the tabulation (trans influence, coordination geometry, etc ) Clearly these tabulations have thrown numerous areas ripe for future study, as of the type described below 4 Testing Theories One particularly powerful application of structural data is in providing experimental tests of theories Such theories might range from qualitative, which for example might predict the extension of a bond length as a result of contraction of a bond angle, to quantitative, in which absolute values of structural parameters are obtained In either case, tests may be formulated and where possible evidence located in the CSD to allow for experimental evaluation of the theory or hypothesis While it is not possible to prove any scientific theory by testing it against experimental observations, it is of value to disprove those that fail to provide adequate explanations of observations Metal-ligand bonding is a particularly important field of theory for practical transition metal chemistry since ligands provide a powerful and subtle means of controlling reactivity, catalytic, and other properties of the metal Tertiary phosphine and related ligands [PA, (A = alkyl, aryl, OR, etc )] have been employed in this way and have attracted much attention from synthetic and theoretical chemists as a result Current theories of the metal-phosphine bond view it as comprising u and 7~ components, the latter being of debatable and certainly variable importance The nature of the 7~ acceptor function has also been a matter of dispute, with phosphorus 3d or P-A orbitals, or a (T* combination of the two, being suggested (see Figure 1) Using two different approaches we have investigated these hypotheses In the first we examined the molecular structures of pairs of transition metal phosphine complexes (L,M-PA,) related by redox couples We looked for changes in M-P and P-A bond lengths and A-P-A bond angles, as a consequence of metal- centred oxidation The changes observed are strikingly consis- tent M-P distances increased, P-A distances decreased, and A-P-A angles increased These effects were entirely consistent with significant M-P 7~ back-bonding in these complexes and with a notable P-A (T*component in the PA, 7~ acceptor function While this kind of study is a powerful test of theories of bonding and electronic structure7 it lacks generality, not least because of the difficulties associated with preparing crystals of pairs of complexes' More seriously, not all theories are testable Figure 1 Hybridized T acceptor orbitals on tertiary phosphine and related ligands PA, CHEMICAL SOCIETY REVIEWS 1993 by such a method For example, the redox-pair study said little about the effects of M-P u bonding in these complexes We therefore set out to conduct a more general study on triphenyl- phosphine species (Z-PPh,, Z = metal, C, 0,N, etc )8 in which we concentrated primarily on symmetrical distortions (I e pre-serving C,, local symmetry of the ZPC, unit) as a function of the element Z and the Z-P bond length This study provided further substantial support for (but as always, no proof ofl) a model of the bonding of triphenylphosphine in which u effects are largely dominant, but 7~ effects on geometry of the sort described above6 are notable The u effects are especially important when Z = p-block element (0,N, C, etc ), while for the transition elements, u and 7~ effects are broadly in balance leading to PPh, geometries close to that of PPh, itself In the study of chiral complexes [CpM(XO)Z(PPh,)] we obtained evidence for the asymmetry of 7~ back-bonding between metal and phosphine being sufficient to cause observable P-C bond length and C-P-C bond angle distortions 5 Conformational Analysis Flexible molecules present a particular challenge to the struc- tural chemist An individual structure determination tells us in some sense what the average (or possibly the equilibrium) geometry of a given species is in the state in which the experiment was made (e g the crystal in this article) However, it is more difficult to learn much about the motion of the molecule far from its equilibrium geometry (vibrational motion close to the equili- brium geometry is, however, amenable to study by crystallo- graphic methods -see, e g , reference 10) How then can we proceed when faced with molecules and ligands which we know to be flexible (or even fluxional) in solution7 One approach that offers promise is to select a molecular or sub-molecular fragment and examine its shape in a (large) number of crystal and molecular environments These environments can be expected to perturb the geometry of the fragment and allow inspection of the distortions which result In the case where no bonds are broken or formed as a result of the perturbations, these distor- tions are variations in the conformation of the fragment and may allow insight into (a) the types of conformation possible for the fragment, (b) how these conformers might interconvert, (c) which aspects of the molecular environment are responsible for the particular conformation adopted These problems are the concerns of conformational analysis, and their resolution is of central importance in a range of contemporary chemistry (e g drug design, homogeneous cata- lyst design) in which the shape (or range of possible shapes) of a flexible molecule may be critical to its properties Underlying this approach is the idea that the potential energy hypersurface (PEH) of the fragment may be read qualitatively (but nut quantitatively)i from the distribution of geometries observed We assume in this kind of study that the fragments adopt geometries that are relatively low in energy and therefore map out the low-lying regions of their common PEH While for molecular fragments it is literally the molecular PEH that is being probed, for sub-molecular fragments we are not inspecting a single PEH, but rather the common features of the various PEHs of the many different molecules in the set of structures studied It is clear that studies of this sort are bound to lead to a fuller understanding of the flexibility of a fragment than are individual structure analyses In a sense the range of crystal environments to which a chosen fragment is exposed in the structures studied mirrors the variation in local fields that a molecule might experience in solution Although this analogy is in general more applicable for a molecular species than for a sub- molecular fragment, it is clearly appropriate in the context of conformational analysis where the interactions between the fragment and its environment are, by definition not so strong as to cause bond cleavage or formation We and others have studied the conformations of both organic and inorganic systems in this way Thus when planning STRUCTURAL SYSTEMATICS IN MOLECULAR INORGANIC CHEMISTRY-A.G. ORPEN syntheses of extended saturated carbocyclic systems containing connected quaternary centres we wanted to know what confor- mations were likely to be adopted.12 The answer turned out to be surprisingly simple. We inspected the geometries of the many tetraalkyl ammonium salts whose crystal structures are known, since they of course contain quaternary centres attached to saturated carbon chains. All NEt; ions adopted one of only two conformers which have molecular symmetry close to DZd and S, respectively [see Figure 2 (a), (b)]. The same conformation about +N-C bonds was observed for all the NPr; and NBu? + ions.In addition all N-C-C-C torsion angles were close to 180",[see Figure 2 (c)], indicating that the quaternary centre controls the conformation of three carbon chains. In contrast both gauche and anticonformations were observed in the C-C-C-C chains of the NBu"; ions. These observations are in excellent accord with calculated (molecular mechanics) energies for the observed and other unobserved conformers. As a result we were able to predict two possible high symmetry polymeric carbocyclic struc- tures, made up of fused sixteen-membered rings of very similar energies. In the first, where the quarternary centres adopt the D2dconformation, a two-dimensional molecular mat is formed, and in the other, in which the quarternary centres have the S, conformation, a three-dimensional cage structure related to that of the ultramarines is obtained, (a) D2d Figure 2 Preferred conformations near the quarternary centres in NEt,+, NPr; ,and NBu; .Two distinct local symmetries are observed at the + + nitrogen: (a) and (b) S4.All N-C-C-C conformations are anti (c).The conformations of ring systems have been much studied in both organic and inorganic chemistries. One key problem is how to describe the ring conformation succinctly using the minimum number of parameters, and preferably ones that may be easily related to familiar intramolecular parameters such as torsion or dihedral angles. In our work in this area we have used the technique of principal component analysis (pca) to achieve these ends.Inspecting and understanding the torsion angle data which can be used to describe the conformations of, e.g., a five-membered ring is rendered awkward because there are five such torsion angles per ring. Thus a scatterplot of such a dataset would need, on the face of it, to be five-dimensional! In practice things are not as bad as they seem, since an n-membered ring is well known to have only n -3 degrees of conformational freedom (i.e. only two parameters or graphical axes are needed to display all the variations in conformations of a five-mem- bered ring). However, the choice of the 'best' n -3 parameters has been the subject of much discussion and research.Pca provides an objective means of reducing the dimensionality of a dataset (of any type) by extracting new axes on which to display the data which are linear combinations of the original para- meters. The new axes -the principal components -are orthogo- nal and are chosen such that the first such component (pcl) expresses the largest fraction of the total variance of the dataset, pc2 the next largest, and so on for as many pcs as are required to describe all the variance in the data set. Figure 3 illustrates such an analysis for a dataset that is nearly (but not quite) one dimensional. A second critical element of such studies is the correct treatment of the symmetry of the parameter space, for which a variety of treatments have been considered.For torsion angle data these essentially boil down to one of two approaches. Yt pcl Figure 3 Schematic of principal component analysis for a near one- dimensional dataset. A plot on axes corresponding to initial para- meters x and y (a) is converted to one on new axes on pcl and pc2, which are linear combinations of x and y (b). In these plots (and those of Figures 5, 6, 7, 9, and 10) each point represents an individual structure. In the first the initial set of torsion angle values is symmetry expanded to fill conformation space with the set of values corresponding to isometric conformations, while in the second all the conformations are symmetry transformed to lie in the asymmetric unit of conformation space. 331 The graphical identification of favoured conformations is greatly aided by pca of a torsion angle dataset since by definition it produces the most helpful, i.e.most spread out, scatterplots of the dataset.Thus in one study of this sort we examined struc- tures containing M,(p-dppm) [M = transition metal, p-dppm = bridging bis(diphenylphosphino)methane] fragments or M(dppe) [dppe = chelating bis(dipheny1phosphino)ethanel fragments (see Figure 4). We examined the conformational preferences of the five-membered MP,C2 (or M,P,C) rings and those of the attached phenyl groups. The M(dppe) fragments showed a preference for a twist (C,)conformation of the MP,C, five-membered ring (6 and A, see Figure 5), in which the P-C-C-P torsion angle was far from zero (typically ca.f 50"). In contrast, the M,P,C five-membered rings of the M,(p-dppm) predominantly had envelope (C,)conformations, with the meth- ylene carbon out of the plane of the near planar M,P, unit, i.e. having the PMMP torsion angle close to zero, and the MPCP torsion angles ca. f45". In these two systems the pcs generated by pca can be identified as measuring the 'twistiness' and 'envelopeness' of the conformation (i.e. the components of deviation from planarity along coordinates preserving C2or C, symmetry of the ring system, respectively). The distribution of structures indicated that in both cases the ring conformations interconvert by a pseudo-rotation pathway, similar to those / M~M M2(~-dppm) M(dppe) Figure 4 Torsion angles within the five-membered ring of M,(p-dppm) and M(dppe) fragments used in pca (see Figure 5).CHEMICAL SOCIETY REVIEWS, 1993 observed for other five-membered ring systems, in which planar intermediates (at the centre of the scatterplot in Figure 5) are avoided. The phenyl group orientations were indicative of a strong coupling of the rotations of the two phenyls on any one phosphorus in either M(dppe) or M,(p-dppm) fragments. For the M,(p-dppm) system transannular phenyl group interactions which lead to preferred conformations were seen. This was most pronounced for the envelope conformation of the five-mem- bered ring, and resulted in the two pseudo-axial phenyls being near parallel. In contrast transannular interactions appeared to be insignificant for the MM(dppe) system, with little or no sign of preferred conformations for the phenyl groups.Pca really comes into its own when dealing with more complex ring systems for which there is no analytical procedure analo- gous, for example, to the Cremer-Pople method.i46,16 Thus, for example, pca of the ten intra-ring torsion angles in diethylene- triamine complexes [L,M(dien)] gives a pca scatter-plot as shown in Figure 6." The two most important pcs display 89% of the total variance in the dataset. The plot clearly shows there to be four main conformer types (66 and AA, A6, SA). Of these only 6h is observed for mer-M(dien) species while 66 (and its mirror Figure 5 Scatterplot for two principal components obtained by pca of the M(dppe) five-membered ring torsion angle dataset.The principal components measure the degree of C, twist [pc(tw)] and C,envelope [pc(env)] character in the conformation of any given M(dppe) frag- ment. The pseudo-rotation pathway interconverting 6 and A con-formers via envelope intermediates is clearly visible. pc2 15.0 4.0 ---66(fac) 3.0 7 h6(fac) -2.0 -3.0 -hh(fac)-4.0 -J.v -7.v -c).v -L.v -1.v v.v 1.v L." PCl Figure 6 Pca scatterplot of the two most significant pcs from the pca of the ten intra-ring torsion angles of the dataset for M(dien). The four main conformer types and the corresponding coordination geome- tries are indicated. STRUCTURAL SYSTEMATICS IN MOLECULAR INORGANIC CHEMISTRY-A.G. ORPEN .X' 0' Figure 7 Scatterplot of torsion angles about the M-P (7)and P-C (wl, w2,and w,) bonds for the chiral metal complexes [CpM(XO)Z(PPh,)]. Note the vacant region of T near 20, 140, and 260", and the near continuous distribution of w values. image Ah) or A8 (which is not the mirror image of ah) is observed for fac-M(dien) species. In a different type of study,9 not involving pca, we analysed the conformations of PPh, complexed to the asymmetric metal fragment CpM(X0)Z (X = N, C; Z = any ligand). Clear corre- lation of phenyl group and PPh, orientations (see Figure 7) and preferred conformations were observed, consistent with res-tricted rotation about the M-P bond but a relatively low barrier to P-Ph rotation. This proposition has been confirmed by NMR spectroscopic evidence, and is in good agreement with molecular mechanics treatments of this system.6 Reaction Pathway Analysis Extension of PEH analysis has been taken to its logical conclu- sion in a series of outstanding studies by Burgi and Dunitz and their co-workers. They applied the structure correlation method to a variety of inter- and intramolecular reactions.19 The assumptions and hypotheses were essentially as described above except that in this context the reactions typically included bond formation and cleavage rather than conformational changes. In more recent extensions to this work, Burgi has demonstrated the possibility of structure-nergy correlations,20 as well as struc- ture-structure correlations (see discussion in reference 2 for example). We have looked at two systems in which bond formation and cleavage information is available from study of collections of structural data.In each case we sought information to allow us to distinguish between alternative mechanisms. In the first we analysed possible pathways by which heteronuclear metal clus- ters containing Ru,Au2 fragments might rearrange.2 In most known examples, these five atom fragments adopt a trigonal bipyramidal geometry with the two gold atoms in different sites, one axial and the other equatorial. From NMR studies it was known that the gold atoms in these clusters undergo rapid intramolecular site exchange with a modest energy barrier. The structural data available allowed a complete mapping of the trajectory followed in this exchange process and confirmed that a partial Berry pseudo-rotation path, in which a square pyrami- dal geometry is intermediate between two equivalent trigonal bipyramid cluster cores (see Figure 8), is followed. In contrast a number of other mechanisms could be ruled out as being inconsistent with the structural evidence.Formally the square pyramidal geometry has one less metal-metal bond (or short contact, since these species are surely not well described by Figure 8 Metal site exchange in Ru,Au, fragments. (a) Axialkquator- ial exchange of gold atoms observed by NMR spectroscopy. (b) A partial Berry pseudo-rotation mechanism for this process. (c) The trajectories of the gold atoms during first half of the Berry-like process in (b), the Ru, triangles are superimposed and example tbp and spy (dashed) frameworks are shown.localized two-electron metal-metal bonds) than does the trigo- nal bipyramidal species (8 compared with 9). In practice there is clear evidence of contraction of the metal-metal bonds around the open square face of the square pyramid apparently compen- sating for loss of bonding across the diagonal of this face. In the second of our studies in this area we have sought to understand the means by which a carbon monoxide ligand moves between metal atoms in dinuclear and trinuclear arrays.22 Such motion is ofcourse exceedingly well known in the chemistry of metal clusters from NMR spectroscopy, and relevant structural evidence abounds.In a revealing study, Crabtree and LavinZ3 were able to identify a full range of structures to map out the CO exchange process in diiron species. In our work we too observed clear pathways for bridge-terminal exchange (see Figure 9), and have extended this to include CO ligands attached to triangular metal arrays. In this work one of our aims was to gain insight into CO migration across the cluster and by implication into related process on metal surfaces, in which of course there are many triangular M, units (e.g. as on close-packed metal surfaces). One question that might be ans- werable in this way is how does CO move from a terminal site into a triply bridging position? The two extreme potential routes would be (i) directly from the terminal site through intermediate geometries of C, symmetry [see (a), Figure lo], and (ii) through p2-intermediates [see (b) and (c), Figure 101.Inspection of the distribution of CO geometries above the metal triangle (see Figure lo), with due allowance for the symmetry of the para- meter space, suggested that the path through (c) and (b) is the more likely, although there is a small number of structures in the region close to the intermediate geometries appropriate to path 196 0 3.501 Figure9 Distribution of Fe-C distance d, and d2in Fe,(CO) fragments in the CSD. The narrow band of values observed corresponds to the reaction pathway for terminal to p2-C0 site exchange. is.\,, M/ 0 1.o-0.5 -S2b , 0.0:. -0.5-----I---1 .o SZ \ Figure 10 Scatterplot (S2b vs. S2a) for M,(CO) fragment geometries (M, = Fe,, Ru,, Os,, CO,,Rh,, or Ir,). S2a = (2d, -d, -d3)/,/6; S2b = (d, -d3)/J2. Pathways for direct terminal to p3-C0 site exchange (a), p2-to p,-CO site exchange (b), and terminal to p2-C0 site exchange (c), are indicated. (a). Another aspect investigated in this study was the effect of carbonyl ligand position on the dimensions of the polymetal unit. For example, in Ru,(CO) species the range of Ru-Ru (single) bond distances was greatly reduced in those cases where the carbonyl is close to symmetrically bridging. However, when the CO was terminal Ru - - Ru distances spanned the range from 2.6 to 3.2& when the carbonyl was bridging the range was CHEMICAL SOCIETY REVIEWS, 1993 reduced to 2.6-2.85& and as a result the mean Ru-Ru distance for this class of structures was substantially lower than for the entire set of structures. Similar effects are observed for the M, system.The limits on the range of metal-metal distances presumably arise because of limit on the range of tenable M-C-M angles in bridging carbonyls. The M-C distances for p,-CO ligands were only slightly increased over the terminal values (by ca. 12%, see Figure 9). This is considerably smaller than for other bridge-terminal exchange processes studied in this way, e.g. O-H.-.O to 0* * * H-0,' for which the symmetrical intermediate has 0-H bonds ca. 25% longer than in the terminal cases.These latter variations in interatomic distances were quantitatively explained as arising from the conservation of bond order around the O-H*.*O system. The small M-C bond length extension is therefore consistent with this same principle, but only if some loss of bond order occurs elsewhere in the system (i.e. in the C-0 bond). Ofcourse this is precisely what is expected given the lower values of v(C0) and associated increases in C-0 bond lengths for bridging as compared with terminal carbonyl ligands. 7 Conclusions In many cases the determination of a crystal structure is an end in itself, and a crucial element in a research project. This review has indicated some ways in which a broader view of the structure, and in particular, its variability under perturbation, may be obtained.In part, this type of study seeks to offer some insight into the eternal conundrum of crystallography: 'so that is the structure in the crystal -but how different/similar is it in solution?' While crystallographic results alone cannot answer this question, we can learn something by seeing a fragment in a wide variety of crystal environments. The different fields perturb the fragment geometry and may be viewed as mimicking the variety of local fields that the same fragment would experience over time in a solution. Clearly this is a much simplified view and is not quantitative. However, we can safely say that a fragment that shows very little variation in geometry across a wide range of crystal (and/or molecular) environments will spend essen- tially all its time in such a geometry while in solution.Collections of structures may be of use in other ways too. There are clearly many physical observables against which one might choose to calibrate a molecular modelling method, includ- ing structural, thermodynamic, and spectroscopic in forma tion. Of these, perhaps the most reliable and abundant source of data is from three-dimensional molecular structures revealed by crystallography. Clearly one can model the molecular geometry found in an individual crystal structure and thereby assess the efficacy of a modelling procedure. However in most calculations the procedure predicts the properties of an isolated (gas-phase) molecule, usually at rest. In contrast the crystallography data refer to a molecule in a crystal field and undergoing vibrational motion.Two strategies might be considered to make a more appropriate comparison between the model and crystallo- graphic results: (1) to allow explicitly for the symmetry (includ- ing translational symmetry) of the solid state structure in the modelling procedure; (2) to place emphasis on the ability to reproduce the range of geometries adopted by a molecular or submolecular fragment in a large number of crystal structures. The aim of this latter approach is to find not just the calculated minimum energy molecular or fragment configuration, but also to examine its softness towards perturbation from whatever source (and by implication to learn something of the fragment's potential energy hypersurface).Clearly such studies are of particular utility when molecular flexibility or reactivity are of importance. This latter approach has been successfully used by KlebeZ4 and others. More direct relationships between molecu- lar structures and their energetics have been probed by Biirgi and his co-workers in structure-energy correlation studies. *JO Developments in the field of structural systematics may in a sense be largely limited by the imagination of the practitioners since there is such an enormous quantity of crystallographic STRUCTURAL SYSTEMATICS IN MOLECULAR INORGANIC CHEMISTRY-A. G. ORPEN data to hand. In the types of studies described above, the structure-structure correlation methodology can be extended to gain more insight into how structural parameters interact.For example, we might study how the conformation of ring systems affects, or is affected by, other aspects of geometry (e.g. large M-M distances are associated with rigorously envelope ring conformation in the M,(p-dppm) system).’ This approach is clearly linked to the desire to be able to engineer molecular structure so as to gain control of reactivity, catalytic, and other properties of the species. What is missing in much of this type of work is an understanding of the relationship beween the molecu- lar geometry and the reactivity or other property of interest. Such studies will require a substantial body of physico-chemical data for correlation with the structural data, but there is clearly an opportunity to aim for quantitative structure-activity rela-tionships of the sort that have been so successfully exploited in pharmaceutical chemistry.In the studies described above it has become clear that it is necessary to use techniques (frequently statistical) which are unfamiliar to many chemists in order to analyse and make the best use of the vast quantities of data in the CSD. These include the application of principal component analysis and the exploi- tation of the symmetry of parameter space, as described above. Substantial progress has been made in introducing statistical technique^'^^'^^^^ and adapting them to deal with the symmetry and periodicity properties of the parameter spaces of interest to structural chemists.Nevertheless it is clear that there remains much to be learnt about the most effective ways of extracting chemical understanding from structural data, and that it is through statistical methods that most of this progress is likely to be made. In addition to technical developments, the prospects for work in this area are perhaps brightest in the field of extramolecular chemistry -the interactions between molecular species in the crystalline state. Crystallography is, of course, by definition a unique source of information in this respect. The scope of the potential contribution of this approach to the study of solid- state reactivity, surface chemistry, and the field of molecular recognition in molecular inorganic chemistry is tremendous and has only begun to be recognized and exploited.26 The emphasis in this article has been on the use of empirical correlations between structural parameters.In fact, of course, any useful chemical interpretation of this information is obtained by adding a considerable patina of theory and hypoth- esis to the empirical observations. The reader should be aware of this. One particular consequence is that there may be several hypotheses that serve to rationalize a given set of observed correlations. This is no weakness of the approaches outlined, but merely another reminder of the way science To sum up, our aim is to understand first what shapes molecules have, then to think about why they have those shapes, and finally to explore the chemical consequences of their shape.The abundance of molecular structure data available, and current developments of methodology and technology, afford extraordinary opportunities for new structural science. Acknowledgements. I thank the Ciba-Geigy Foundation for the award of a Senior Research Fellowship, Professor D. Braga and his colleagues at the Dipartimento di Chimica ‘G. Ciamician’, Universita di Bologna for their hospitality during a sabbatical stay in Bologna during which time this article was drafted, my colleagues and co-workers at the University of Bristol, and the many chemists and crystallographers who synthesized and determined the structures of the compounds discussed.8 References 1 F. H. Allen, 0.Kennard, and R. Taylor, Ace. Chem. Res., 1983,16, 146;F. H. Allen, J. E. Davies, J. J. Galloy, 0.Johnson, 0.Kennard, C. F. Macrae, E. M. Mitchell, G. F. Mitchell, J. M. Smith, and D. G. Watson, J. Chem. In$ Comput. Sci., 1987,31, 187. 2 See H.-B. Burgi, in ‘Perspectives in Coordination Chemistry’, ed. A. F. Williams, C. Floriani, and A. E. Merbach, Verlag Helvetica Chimica Acta, Basel, 1992, p.1; T.P.E. Auf der Heyde, in ‘Structure Correlation’, ed. H.-B. Burgi and J. D. Dunitz, VCH, Weinheim, in the press. 3 F. H. Allen, 0.Kennard, D. G. Watson, L. Brammer, A. G. Orpen, and R. Taylor, J. Chem. Soc., Perkin Trans 2, 1987, S1. 4 A. G. Orpen, L. Brammer, F. H. Allen, 0.Kennard, D.G. Watson, and R. Taylor, J. Chem. Soc., Dalton Trans., 1989, S1. 5 K. R. Popper, ‘Conjectures and Refutations’, Routledge and Kegan Paul, London, 1963. 6 A. G. Orpen and N. G. Connelly, J. Chem. Soc., Chem. Commun., 1985, 1310; A. G. Orpen and N. G. Connelly, Organometallics,1990, 9, 1206. 7 See, e.g., R. P. Aggarwal, N. G. Connelly, M. C. Crespo, B. J. Dunne, P. M. Hopkins, and A. G. Orpen, J. Chem. Soc., Dalton Trans., 1992, 655. 8 B. J. Dunne, R. B. Morris, and A. G. Orpen, J. Chem. SOC.,Dalton Trans., 1991, 653. 9 S. E. Garner and A. G. Orpen, Acta Crystallogr., Sect. A, 1990,A46, C360; J. Chem. Soc., Dalton Trans., 1993, 533. 10 H. B. Burgi, A. Raselli, D. Braga, and F. Grepioni, Acta Crystallogr., Sect. B, 1992, B48,428. 11 J. D. Dunitz and H.-B. Burgi, Acta Crystallogr., Sect. B, 1988, B44, 445. 12 R. W. Alder, C. M. Maunder, and A. G. Orpen, Tetrahedron Lett., 1990,31,6717. 13 H. B. Burgi and K. Chandrasekhar, J. Am. Chem. Soc., 1983, 105, 7081; L. Nerrskov-Lauritsen and H. B. Biirgi, J. Comp. Chem., 1984, 6, 216. 14 (a)F. H. Allen, M. J. Doyle, and R. Taylor, Acta Crystallogr., Sect. B, 191, B47,29,41, and 50; (b)F. H. Allen, M. J. Doyle, and T. P. E. Auf der Heyde, Acta Crystallogr., Sect. B, 1991, B47,412. 15 D. A. V. Morton and A. G. Orpen, J. Chem. Soc., Dalton Trans., 1992,641. 16 D. Cremer and J. A. Pople, J. Am. Chem. Soc., 1975,97, 1358. 17 S. E. Garner and A. G. Orpen, unpublished results. 18 S. G. Davies, A. E. Derome, and J. P. McNally, J. Am. Chem. Soc., 1991,113,2854. 19 H. B. Burgi and J. D. Dunitz, Ace. Chem. Res., 1983,16, 153; H. B. Burgi, Inorg. Chem., 1973, 12, 2321. 20 H. B. Burgi and K. C. Dubler-Steudle, J.Am. Chem. Soc., 1988,110, 4953,729 1. 21 A. G. Orpen, and I. D. Salter, Organometallics,1991, 10, 111. 22 L. Brammer, N. S. Dhillon, D. A. V. Morton, K. A. MacPherson, and A. G. Orpen, to be submitted to Inorganica Chimica Acta. 23 R. H. Crabtree and M. Lavin, Znorg. Chem., 1986,25, 805. 24 G. Klebe, Struct. Chem., 1990, 1, 597. 25 R. Taylor, J. Mol. Graphics, 1986,4, 123. 26 See for example, D. Braga and F. Grepioni, Organometallics,1992, 11, 71 1, 1256.

ISSN:0306-0012    

DOI:10.1039/CS9932200191

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Some Recent Synthetic Routes to Thioketones and Thioaldehydes William M. McGregor” and David C. Sherrington Department of Pure and Applied Chemistry, Thomas Graham Building, University of Strathclyde, 295, Cathedral Street, Glasgow G 7 IXL I Introduction The chemistry of thioketones and thioaldehydes has been widely studied and their reactions re~iewedl-~ because they offer useful routes for the incorporation of sulfur heteroatoms into the synthesis of heterocycles or natural products. They have high reactivities associated with the poor orbital overlap of the (C-~P, S-3p)7r-bond. This has meant that only a few examples of stable, isolable thioketones have been made4-5 and the even greater reactivity of thioaldehydes delayed the isolation of the first examples6 * to as recently as 1982.The tendency with thiocar- bony1 compounds is for spontaneous oligomerization to occur unless there is electronic stabilization’ (thioamides, thionoes- ters, and thionoacids are usually fairly stable) or steric stabiliza- tion (thioketones and thioaldehydes need bulky substituents if they are to be isolable). Most of the studies on the reactivity of thiocarbonyl groups have been conducted on stable thioketones and these have shown general similarities between the chemistry of carbonyl and thiocarbonyl groups. A significant increase in reaction rate is observed, however, for thiocarbonyls because of the inherent instability of the C-S r-bond: for example, thiobenzophenone reacts about 2000 times faster1 than benzophenone when treated with phenylhydrazine to give the hydrazone. Similarly, the reduction of thioketones to thiols with sodium borohydride is much faster’ than the corresponding reduction of ketones to alcohols.The most significant differences between thiocarbonyl and carbonyl chemistry are observed in reactions involving nucleo- philes, olefins, and conjugated dienes. Thus, nucleophilic attack on carbonyl compounds is generally at the carbonyl carbon, but with thiocarbonyls it can occur either at carbon or sulfur. Examples of nucleophiles which attack thioketones at sulfur’ include phenyl lithium (to give thioethers) and sodium bisul- phite (to give thiosulfates). With mono-olefins (l), thiocarbonyl compounds (2) can undergo ‘ene’-type reactions (Scheme 1).For example thioben- zaldehyde reacts with (-)-p-pinene to give ‘ene’ reaction pro- duct~.~Both C-C 0-bond formation, to give a thiol (3), and C-S 0-bond formation, to give a sulfide (4),can be observed. Professor D. C. Sherrington was born in Liverpool andeducated at Waterloo Grammar School. He was awarded his Ph.D. at the University of Liverpool and was appointed Lecturer at the Univer- sity of Strathclyde in 1971. From 1983-87 he was a Senior Research Scientist with Unilever Research before re-turning to Strathclyde as Pro- fessor of Polymer Chemistry. In 1990 he became a Fellow oj the Royal Society of Edinburgh and he received his D.Sc. from the University of Liverpool in 1992. His research interest lies in the synthesis and chemical modijication of polymers for applications ranging from catalysis to high performance materials.II99 ‘S n (4) Scheme 1 ‘Ene’-reaction of thiocarbonyl compounds. Thiocarbonyl compounds undergo cycloaddition reactions much faster than carbonyl compounds. Thus the Diels-Alder [4 + 21 cycloaddition reaction (Scheme 2) between a conjugated diene (5) and a thioketone or thioaldehyde (2) gives stable cycloadducts (6) and, in particular, transient thiocarbonyls are routinely characterized as such cycloadducts. Scheme 2 Diels-Alder reaction of a thiocarbonyl compound. So despite their highly reactive nature, a wide range of literature is available on the reactions of both stable and transient thioketones and thioaldehydes and on their use in synthesis.O As the latter has become more widespread, so too has the variety of synthetic methods for the generation of thiocarbonyls become more extensive. This article concentrates on the plethora of synthetic approaches to thioketones and * Now at Ecole Europeenne des Hautes Etudes des Industries Chirnique de Strasbourg, Universite Louis Pasteur, 1, rue Blaise Pascal, 67008 Strasbourg, France. Dr. W. M. McGregor was born in Glasgow in 1965 andgained his Ph.D. on the study of thio-carbonyl compounds from the University of Glasgow in 1989. He has since been employed in a variety of postdoctoral positions and also as a research chemist with Scottish & New-castle Breweries.His research interests include organosulfur chemistry, host-guest chem-istry, and the design and syn- thesis of novel monomers. He is currently in Strasbourg studying complexes of calixarenes. CHEMICAL SOCIETY REVIEWS, 1993 thioaldehydes, rather than on their reactivity which is reviewed extensively elsewhere. 1-3 2 Methods for Converting Carbonyls into ThiocarbonyIs The availability of carbonyl compounds makes thionation of carbonyls (7) to give thiocarbonyls (2) a very desirable route (Scheme 3) and many methods of effecting this transformation are known. In the past thiocarbonyl transfer from thiophosgene and analogous compounds' 'has been used, but this method is often ineffective and is now rarely used. Many better methods and reagents are now known.reagent R R' Scheme 3 Thionation of carbonyl compound. 2.1 Hydrogen SulfidelAcid The use of hydrogen sulfide gas in the presence of an acid catalyst is a classical method for the generation of thiocarbonyl groups which has been used for over a century (it is much more commonly used than the related reagent H,S,/HCl). The acid catalyst is usually hydrogen chloride, which reversibly proto- nates the carbonyl group, facilitating substitution at carbon by H,S, and subsequent elimination to give the thiocarbonyl. This reagent (H,S/HCl) has also been used to convert ketals into thioketones. Despite the difficulties associated with the use of hydrogen sulfide gas this approach is still in use, partly because it gives good yields and clean products.This method', is straightforward when the starting material is a ketone (7) (with R and R' as alkyl or aryl) and gives good yields of pure thioketones (2), often isolated as cycloadducts. The synthesis of thioaldehydes with H,S/HCl is less straightforward, but a successful route to thioaldehyde adducts has been devised (Scheme 4).Although no free thioaldehyde is generated by this method, the cycloadducts are identical to those derived from the appropriate thioaldehyde by other routes. The silyated compound (8) is converted14 (Scheme 4)into the reactive thioketone (9) which can be trapped in situ as the Diels- Alder cycloadduct (10) and then can be desilylated with tetrabu- tyl ammonium fluoride (TBAF) to give the thioaldehyde adduct (1 1).Use of an analogue of (8), but with a chiral silicon centre, gives asymmetric induction and allows (10) and (1 1) to be made enantioselectively. 4b 2.2 Phosphorus-based Reagents Many variants on inorganic reagents containing phosphorus and sulfur have been reported, including PClJAl, S3/Na, SO,, P,S,/NEt,, RPS(OR'),, PSCl,(NMe,), -n, but the one most commonly used to convert ketones (7) into thioketones (2) is phosphorus pentasulfide (in fact P4S10) (Scheme 5). Scheme 5 Thionation with phosphorus pentasulfide. It was first used in the synthesis of Michler's thioketone in 1886 and since then has found extensive application in the synthesis of thioketones and related thiocarbonyl compounds.The main problem with this reagent is its insolubility in most organic solvents. The best results have, however, been achieved using phos- phorus pentasulfide as a homogeneous reagent (it is appreciably soluble in pyridine or diglyme) and many thionations have been successfully performed in pyridine,?' 6*1 ' to give a variety of stable or transient thioketones containing alkyl, alkenyl, aryl, or ester substituents. The main advantages of this method are good yields, no organic by-products, and facile purification of the thionated products. Another successful phosphorus-based reagent is 2,4-bis(4- methoxypheny1)-1,3-dithia-2,4-phosphetane-2,4-disulfide(13), known as Lawesson's reagent.I5 It is made (Scheme 6) by the action of phosphorus pentasulfide on anisole (12) and is con- sidered to be an organic analogue of P,S o.S --0Me OMe II P-s S Scheme 6 Synthesis of Lawesson's Reagent. This reagent is soluble in, for example, benzene, toluene, or xylene and has been used in these solvents to convert ketones (7) into thioketones (2) in good yields.15,18 The thioketones or thioketone derivatives, with alkyl or aryl substituents, made by this method usually require chromatographic purification to remove the anisole-related by-products. Rj$R R (11) I8 R Scheme 4 Synthesis of thioaldehyde adducts using H,S/HCl. SOME RECENT ROUTES TO THIOKETONES AND THIOALDEHYDES-W. M. McGREGOR AND D. C. SHERRINGTON 20 I 2.3 Silicon-or Tin-based Reagents Although, for example, SiS2 can be used l9 to convert carbonyls (7)1 into thiocarbonyls, more recently better silicon reagents have R R’become available. Ketones (7) can also be converted into transient alkyl- or aryl- PPh,H2Nm6substituted thioketones (2) with bis(trimethylsily1)sulfidein ace- II tonitrile (with catalytic amounts of CoCI, or CF,SO,SiMe,) at room temperature, making mild conditions possible in thiona- tion reactions.20 This reagent (in the presence of catalytic amounts of butyl lithium) also converts aldehydes into transient thioaldehydes206(R = alkyl, aryl).In both cases the transient thiocarbonyls are isolated as Diels-Alder cycloadducts in high yields. A similar reagent is bis(tricyclohexylstanny1)sulfide which, in the presence of catalytic amounts of boron trichloride, converts ketones into thioketones in high yield2OC (only stable thioke- tones have been reported).2.4 Via Dithiolanes Ketones are readily converted into 1,3-dithiolanes by reaction with ethane-l,2-dithiol using a suitable catalyst such as BF,.OEt,. Parent or S-substituted 1,3-dithiolanes are useful precursors of thioketones, giving either a stable or transient thioketone when treated with a base (Scheme 7). This type of chemistry cannot be applied to the synthesis of thioaldehydes because treatment with base of a dithiolane (or derivative) formed from an aldehyde preferentially gives an ‘umpolung’ reaction by removal of the hydrogen a to both sulfur atoms. Simple 1,3-dithiolanes (14) do dissociate (Scheme 7) on treatment with base to give thioketones (2) but it is possible to use a more reactive analogue, such as the S,S-dioxide (lsa), methyl sulfonium (1 5b) or even the silyl derivative (1 5c).Good yields of thioketones2 (or derivatives) are obtained from (14), (1 5a), (1 5b), and (1 5c) on treatment with a base, and from (1 5a) by thermolysis. 2.5 Via Hydrazones Some stable thioketones (with R and R’ = alkyl or aryl) have been made via hydrazones (1 7) by reaction with sulfur dichlor- ide22 (Scheme 8). Notably, the first stable thioaldehydeP (2) were also made from hydrazones (17) (R = hindered aryl, R’ = H) by reaction with sulfur dichloride and triethylamine. Similarly, conversion of a hydrazone (17) into the phosphora- zone (18) then heating with elemental sulfur also gave stable thioketones2, (2) (R and R’ = alkyl).0 s2c1\or S2CI2/base S Scheme 8 Conversion of hydrazones into thiocarbonyls. 3 Thiocarbonyl Compounds from Non- carbonyl Precursors 3.1 Elimination of HX A large variety of transient thi~aldehydes~~ (2) (R = H, ester, amide, aryl, phenacyl, cyano; R’ = H) have been made by base- induced elimination of HX (Scheme 9) from suitable sulfides (19) and reacted in situ, usually in Diels-Alder cycloadditions. Yields are often good with few by-products. Some thioketones2 have also been made by such base-induced eliminations (R = R’ = ester, aryl) giving either a stable thioketone or a Diels-Alder adduct of a transient thioketone.R’ R’R+ s/x base ~ H/ X = C1, S03-Na+,S02Ar, phthalimido Scheme 9 Elimination of HX from sulfides. Additionally, the thioaldehyde (21 a) and thioformaldehyde (21 b) have been recently made26 by gas-phase dehydrocyana- tion (Scheme 10) of (20a) and (20b). These very reactive species (21a) and (21b) have been detected by photoelectron spectroscopy. 1-1(15a) X = SO2 (15b) X = S+Me’XR,(1%) X = S(O)Si(r-Bu)Me,R Scheme 7 Base-mediated decomposition of 1,3-dithiolanes. (20a) R = Me (21a) R=Me (20b) R=H (21b) R =H Scheme 10 Dehydrocyanation of cyanothiols. 3.2 Using Xanthates Geminal dibromides (22a) and (22b) are converted by treatment with two equivalents of potassium 0-ethyl xanthate (Scheme 1 1) into transient thioketonesZ7 or thioaldehydes2 lC(2), trapped in situ as Diels-Alder cycloadducts.The mechanism is likely to involve displacement of bromide by xanthate. The second equivalent of xanthate then causes elimination to the thio- ketone. Yields vary from poor to good and since there is often a variety of by-products purification is usually necessary. (22a) R = R' = acyl, ester, phthalimido (22b) R = H ;R' = CY~XIO (2) Scheme 11 Conversion of geminal dibromides into thiocarbonyls. 3.3 Photolysis of Thioacetophenones A good photochemical approach to thioaldehydes and thio- ketones is by photolytic Norrish type-I1 cleavage of o-substi- tuted thioacetophenones (27) (Scheme 12), themselves easily made by reaction of a thiolate, (24) or (25), with a halide, (23) or (26).1 phTo+ Scheme 12 Photolytic cleavage of thioacetophenones. The first description of photolysis of a phenacyl sulfide (27) was in 1966 and was soon shown to result in thiocarbonyl compounds (2).z8 This approachz9 has subsequently been used to produce a stable thioketone (benzophenone), and transient thioaldehydes (R = H, R' = alkyl, aryl, phenacyl, ester, cyano), usually trapped in situ as Diels-Alder cycloadducts. Problems can occur when R and/or R' contain phenacyl groups. The substituents then contain the same chromophore as the photola- bile part of (27) and yields are reduced dramatically. 3.4 Pyrolytic Methods Pyrolysis is usually used only in the case of transient thioketones and thioaldehydes, as a thermally labile adduct is a convenient CHEMICAL SOCIETY REVIEWS, 1993 way of storing such transient species. The most common pyro- lytic method of generating thioketones and thioaldehydes is by the retro-Diels-Alder reaction (Scheme 13).Scheme 13 Generation of thiocarbonyls by the retro-Diels-Alder reaction. The anthracene adducts (29) and the cyclopentadiene adducts (30) of thioketones' and thi~aldehydes,~~~~~~~and other bridged cycloadducts,30 ,(R = H. R' = alkyl, aryl, cyano, ester, phenacyl; R = R' = ester) decompose thermally (Scheme 13) to give thiocarbonyl compounds (2). The ring strain associated with bridged cycloadducts makes retro-Diels-Alder reactions feasible and both (29) and (30) dissociate reversibly on heating to liberate the thiocarbonyl (2) which can be reacted in situ with another substrate, e.g.with a substituted butadiene. A more unusual reaction31a is the thermolysis of 3,3,5,5- tetraphenyl-l,2,4-trithiolane(3I) which decomposes above its melting point (124°C) to give a mixture of thiobenzophenone (32) and the ylid (33), which spontaneously decomposes to (32) and elemental sulfur (Scheme 14). The intermediacy of (33) has been proven by intercepting it by cycloaddition. Earlier reac- tion~~of this type gave thioformaldehyde (2 1b) by thermolysis at 840 K. (2lb)R=H(32) R = PhR h - RkS \ + s-s (31) Scheme 14 Thermolysis of 1,2,4-trithiolanes. Thiosulfinates (34) decompose thermally9 (Scheme 15) to give sulfinic acids (35) and thioaldehydes (36) (R = alkyl, aryl), trapped in situ as Diels-Alder cycloadducts.Flash vacuum pyrolysis (FVP) is a very clean and efficient way of generating (in the gas phase) transient species for spectro- scopic or chemical investigations at low temperatures. It gener- ally involves using temperatures in excess of 400°C and pres- sures of less than 10-4mbar for the vacuum pyrolysis, and isolation of the free thiocarbonyl compound in a liquid nitrogen trap or argon matrix. Many pyrolytic routes to thi~formaldehyde~ have been 2a described and it has been generated by FVP3 from (3 1) (Scheme 14) and from thioformaldehyde dimer. Interest in FVP gene- -II 0 OH H (34) R = Me, Ph (35) (36) Scheme 15 Thermolysis of thiosulfinates.SOME RECENT ROUTES TO THIOKETONES AND THIOALDEHYDES-W. M. McGREGOR AND D. C. SHERRINGTON ration of thioformaldehyde and related species, and their pho- toelectron spectra, is associated with the discovery of these species in interstellar space, and the photoelectron spectra of such thioaldehydes generated by FVP are routinely compared with microwave spectra from radio telescope^.^^^FVP has been used to generate thi~aldehydes~~~ from the adducts (37) and (40) and the subsequent reactions of the species, isolated at liquid nitrogen temperatures, have been investigated. For example, FVP of (37) also gives a thi~aldehyde~~ (38), which spontaneously tautomerizes to (39) (Scheme 16). Simi- larly, (40) with FVP gives the thioketones (41) isolated at low temperature (Scheme 17), and subsequently reacted with conju- gated dienes to give Diels-Alder cycloadducts.Scheme 16 Flash vacuum pyrolysis of (37). R R (40)R = H ,Me (41) Scheme 17 Flash vacuum pyrolysis of (40). 3.5 Miscellaneous Methods Fluorine-induced elimination of a-silyl disulfides (42) gives transient thioaldehydes (36), trapped in situ with conjugated dienes, under very mild conditions and with high yields24c (Scheme 18). SiMe2R CsF or S ____) R Bu4NF R (36) (43) Scheme 18 Elimination of a-silyl disulfides. An unusual synthesis (Scheme 19) of stable thioaldehydes (47) involves converting the disulfide (44) into the thiosulfinate (45) which can undergo elimination with perchloric acid to give the (46).Reaction with an aliphatic diamine and spontaneous rearrangement gives the stable thioaldehyde19 (47). Ph Ph (47) Scheme 19 Synthesis of a stable thioaldehyde. Investigations into glutathione peroxidase (a mammalian selenoenzyme) used the model compound (49) which when treated with benzyl thiol gives the thioselenoxide (50) (Scheme 20). Spontaneous syn-elimination of (50) then gives35 the un- stable species thiobenzaldehyde (51) which was trapped in situ with cyclopentadiene in high yield. i) Br2 0 H Ph ii))=(OSiMe3 M<NLe Se-Ph iii) O3Me iv) peracid (48) '0IIPhCH2SH (52) OH Me s,' .spontaneous 0 + Ph Scheme 20 Elimination of thioselinate. An involved route to the thioketone (57) from tris(trimethy1si- 1yl)methane (53) is shown in Scheme 21.Treatment of (53) with methyl lithium gives tris( trimethylsily1)methyl lithium which reacts with elemental sulfur and is subsequently quenched with aqueous acid to give a mixture of (54)and (55). The thiol(54) can be converted into the sulfenyl bromide (56). Pyrolysis of the tetrasulfide (55) or of the sulfenyl bromide (56) gives the stable thio ketone (57). i) MeLi HSC(SiMe3)3 ____) BrSC(SiM%)3i) MeLi ii) Br,-..ii) S8 HC(SiMe3)3 -+ (55) (57) Scheme 21 Synthesis of a silylated thioketone. A related synthesis (Scheme 22) gives the stable thioalde- hyde66tris(trimethylsilyl)ethanethial(60) in 16% yield by treat- ing tris(trimethylsily1)methane (53) with methyl lithium, then with 0-ethylthioformate to give (58) in low overall yield.Inter- estingly, even when heated to 80 "C,(58) cannot oligomerize (for steric reasons) but instead isomerizes to give (60) quantitatively. Treatment of the sulfide (6 1) with N-chlorosuccinimide (NCS) (Scheme 23) then with thioacetic acid and base gives the dithioacetal(62). Oxidation of (62) with rn-chloroperoxybenzoic i) MeLi HC(SiMq)3 -H iC(SiMe3)3YwSiMe3SiMe3 ii) (53) H OEt (58) (59) SiMe3 s' Scheme 22 Synthesis of a silylated thioaldehyde. R = alkyl, aryl, ester, carboxyalkyl, carboxyaryl Scheme 23 Thioaldehydes from dithioacetals acid (mCPBA) and treatment with triethylamine gives the thioaldehydeZg6 (36) The unstable thioaldehydes (36) were trapped zn sztu as their Diels-Alder cycloadducts An adaptation of the Vilsmeier-Haack aldehyde synthesis involves treatment of the enamines (63) with N,N-dimethyl formamide and phosphoryl chloride (Scheme 24), and gives the Vilsmeier salt (64), which can then be treated with aqueous NaSH to give the stable thioaldehyde~~ (65) + P02C1me2 - S H R’ q H k R “ R-R S H R R NaSH R R (63) (64) (65) R = ester, R’ = alkyl, aryl, R” = NH2 R = R’ = Ph, R” = rnorpholino Scheme 24 Vilsmeier-type synthesis of thioaldehydes An unusual to the preparation of the thioketone (68) involves treatment of diethyl 2-chloromalonate (66) with caesium carbonate and elemental sulfur (Scheme 25) to give the intermediate salt (67) This spontaneously decomposes to give the highly reactive thioketone (68), which was trapped zn sztu by Diels-Alder cycloaddition in very high yield (66) (67) (68) Scheme 25 Elimination of a caesium salt 4 Overview The synthesis of thioaldehydes and thioketones has been exten- sively investigated Both stable and transient thioaldehydes and thioketones can be made by a wide variety of routes, and good yields of the thiocarbonyl or thiocarbonyl reaction products can be obtained Transient examples can be trapped at low tempera- tures or generated and reacted m sztu A range of methodology is available for converting different functional groups Into thiocar- bony1 groups and this offers considerable scope for introduction of a sulfur heteroatom into the synthesis of natural products or other materials 5 References 1 F Duus, ‘Comprehensive Organic Chemistry’, Vol 3, ed D H R Barton and W D Ollis, 1979, Pergamon, Oxford, p 373 2 D Crich and L Quintero, Chem Rev , 1989,89, 1413 3 (a) P Metzner, Phosphorus Sulphur, 1991, 58, 295, (b) K Hartke, ibid , 1991, 58, 223 4 (a)J W Greidanus, Can J Chem ,1970,48,3530, (b)K F Wai and M P Sammes, J Chem Soc Perkzns Trans 1, 1991, 183 5 A Ricci, A Degl’Innocenti, M Fiorenza, P Dembach, N Rama-dan, G Secondi, and D R M Walton, Tetrahedron Lett, 1985,26, 1091 CHEMICAL SOCIETY REVIEWS, 1993 6 (a) R Okazaki, A Ishii, N Fukuda, H Oyama, and N Inamoto, Tetrahedron Lett ,1984,25,849, (b)R Okazaki, A Ishii, N Fukuda, and N Inamoto, J Am Chem Soc , 1987,109,279 7 B Adelaere, J -P Guemas, and H Quiniou, Bull Soc Chim Fr , 1987, 517 8 M Moraoka, T Yamamoto, K Enomoto, and T Takeshima, J Chem Soc Perkin Trans I, 1989, 1241 9 J E Baldwin and R C G Lopez, J Chem Soc Chem Commun 1982, 1029, Tetrahedron, 1983,39, 1487 10 (a)T A Shepherd and L N Jungheim, Tetrahedron Lett, 1988,29, 5061, (6)V Alcazar, I Tapia, and J R Moran, Tetrahedron, 1990, 46, 1057 11 C Larsen and D N Harpp, J Org Chem , 1980,45, 3713 12 R Mayer and H Berthold, Chem Ber , 1963, 96, 3096 13 (a)E J Corey, D Seebach, and R Freedmann, J Am Chem SOC , 1967,89,434, (b)C Fournier, D Paquer, and M Vazeux, Bull Soc Chim Fr , 1975,2753 14 (a) B F Bonini, G Mazzanti, P Zani, G Maccagnani, and E Foresti, J Chem Soc Perkin Trans I, 1988, 1499, (b)B F Bonini, G Mazzanti, P Zani, and G Maccagnani, J Chem SOC Chem Commun , 1988,365 15 (a)I Thomsen, K Clausen, S Scherbye, and S 0 Lawesson, Bull Soc Chim Belg , 1977, 86, 693, (c) B S Pederson, S Scheibye, N Nilsson, and S 0 Lawesson, zbzd, 1978,87, 223 16 P Beslin, D Lagain, and J Vialle, Tetrahedron, 1981,37, 3839 17 G W Kirby and W M McGregor, J Chem SOC Perkin Trans I, 1990,3175 18 (a) K F Wai and M P Sammei, J Chem Soc Perkins Trans 1, 1991, 183, (b)T Saito, Y Shundo, S Kitazawa, and S Motoki, ibzd, 1992, 600 19 F M Dean, J Goodchild, A W Hill, S Murray, and A Zahman, J Chem Soc Perkin Trans 1, 1975, 1335 20 (a) A Capperucci, A Degl’Innocenti, A Ricci, A Mordini, G Reginato, J Org Chem , 1991,56,7323, (b)M Segi, T Nakijima, S Suga, S Mural, I Ryu, 0 Ogawa, and N Sonoda, J Am Chem Soc , 1988, 110, 1976, (c) K Steliou and M Mrani, J Am Chem Soc , 1982, 104, 3104, (d)See also J M Kane, Synthesis, 1987, 10, 912 21 (a) E Schaumann, Bull SOC Chim Belg, 1986, 95, 995, (b) E Schaumann, S Winter-Extra, and G Ruhter, J Org Chem , 1990, 55, 4200, (c) B Schuler and W Sundermeyer, Tetrahedron Lett, 1989,30,4111 22 R Okazaki, K Inoue, and N Inamoto, Tetrahedron Lett , 1979,20, 3673 23 P de Mayo, G L D Petrasiunas, and A C Weedon, Tetrahedron Lett ,1978,19,462 I 24 (a)G W Kirby and A D Sclare, J Chem Soc Perkin Trans I, 1991,2329, (6)S S -M Choi, G W Kirby, and M P Mahajan, J Chem SOC Perkzns Trans I, 1992, 191, (c)G A Krafft and P T Meinke, Tetrahedron Lett , 1985, 26, 1947 25 (a)J L Kice and L Wedas, J Org Chem , 1985, 50, 32, (b) S D Larsen, J Am Chem Soc , 1988,100,5932 26 (a) L Waznek, J C Guillemin, P Guenot, Y Vallee, and J M Denis, Tetrahedron Lett, 1988, 29, 5899, (b) A C Gaumont, L Waznek, and J M Denis, Tetrahedron, 199 1,47,4927 27 K Friedrich and H -J Gallmeier, Tetrahedron Lett , 198 1,22,2971 28 (a)H Hogeveen and J J Smidt, Reel Trav Chim Pays-Bas, 1966, 87, 489, (b) B M Trost, J Am Chem Soc, 1967, 87, 138, (c) E Vedejs and D A Perry, J Am Chem SOC, 1983,105, 1683 29 (a)K Kimwa, H Niwa, and S Motoki, Bull Chem Soc Jpn , 1977, 50, 2751, (b)E Vedejs, J S Stults, and R G Wilde, J Am Chem Soc , 1988,110,5452 30 L F Lee, M G Dolson, R K Howe, and B R Stults, J Org Chem , 1985,50,3216 31 (a)R Huisgen and J Rapp, J Am Chem Soc ,1987,109,902, (6)V Eck, A Schweig, and H Vermeer, Tetrahedron Lett, 1979, 20, 59 32 (a)M E Jacox and D E Milligan, J Mol Spectrosc ,1975,58, 142, (b) H Bock, B Solouki, S Aygen, M Bankmann, 0 Breuer, R Dammel, J Dorr, M Haun, T Hirabayashi, D Jaculi, J Mintzer, S Mohmand, H Huller, P Rosmus, B Roth, J Wittmann, and H P Wolf, J Mol Struct ,1988, 173, 31 33 (a) F Bourdon, J -L Ripoll, Y Vallee, S Lacombe, and G Pfister- Guillouzo, J Org Chem , 1990, 55, 2596, (b) Y Vallee and J -L Ripoll, Phosphorus Sufjiur, 1991, 121 34 E Klinsberg, J Am Chem Soc , 1961,83,2934 35 (a) H J Reich and C P Jasperse, J Am Chem Soc, 1987, 109, 5549, (b) see also J L Kice and T W S Lee, J Am Chem Soc , 1978,100,5694 36 M M Abelman, Tetrahedron Lett, 1991,32,7389

ISSN:0306-0012    

DOI:10.1039/CS9932200199

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Electrolytes in Binary Solvents: An Experimental Approach S. Taniewska-Osinska Department of Physical Chemistry, University of Codi, Pomorska 18, 91 -4 16Lodi, Poland 1 Introduction abilities. Of the thermodynamic functions describing solvation Early investigations into electrolyte solutions focused on or solution processes, entropy is significantly connected with the solvent structure perturbations brought about by the dissolved aqueous systems, because of the universal availability and importance of water. Later, electrolytes in single organic sol- vents were investigated, and more recently similar studies on mixed solvent systems have been carried out. The latter find application in various technologies, as they offer a wide choice of solutions with appropriate properties.Although research on electrolyte solutions has received a great deal of attention, a general theory describing their proper- ties and structure has yet to be formulated and, to the best of the author’s knowledge, no theoretical attempt has even been made to describe ternary solutions containing electrolytes. Most investigations have focused on experimental methods affording the possibility of drawing conclusions about the properties of the systems under discussion. Information on electrolyte solutions in binary mixed solvents has so far come mainly from the use of thermodynamic, viscosimetric, electro- chemical, and other macroscopic methods. Few spectroscopic investigations of these systems have been carried out.Thermodynamic methods are important because changes in properties caused by variations of temperature, composition, and pressure can be studied without any reference to assump- tions, models, or hypotheses. This review will consider the properties of alkali metal halides and NaBPh,, as an example of salts with organic ions, in binary mixed solvents. Such a selection simplifies the task of interpre- tation, because the chosen electrolytes do not form complexes or enter into reactions with solvents. 2 Thermochemical Features Thermodynamic functions have shed some light on the energe- tics of solvation of ions in the chosen mixed solvents, the effects of ions upon the properties of mixed solvents (e.g. interactions and structure), and on the role of the solvents themselves, i.e.their bulk properties and their electron donor and acceptor Professor Stefania Taniewska-Osiriska is head of the Physical Chemistry Department at the University of L6di (Poland), where she obtained MSc. (1950) and Ph.D (1960) degrees and her Habil. D (I 965). After a postdoctoral fell0 wship (1962-1 964) at the Leningrad Technological Institute, working with Professor Mishchenko, she returned to Lcidi as an assistant professor in 1965. She was made associate professor in 1976, and fullpro- fessor in 1987. Her current in- terests include thermodyna-mics, and viscosity and other physico-chemical properties of electrolytic solutions. She has had 163 research papers and three monograph reviews pub- lished and has been the reci- pient of two research awards from the Ministry of Higher Education and 21 from the Rector of The University of L6di.She is the Editor of the journal Folia Chimica. ions. Unfortunately, only limited entropy data for electrolytes in mixed solvents are available. Enthalpy values are more widely available which is the reason why these are more frequently analysed. As is known, plots of entropy changes vs. salt concent- rations or mixed solvent compositions are analogous to corres- ponding enthalpy plots. In order to concentrate attention on the properties of electrolyte solutions in binary mixed solvents, enthalpies of various salt solutions are discussed in this article. It is worth remembering that the solvation of a salt in any solvent proceeds as the transfer of individual ions constituting an electrolyte from the ideal gas standard state into their standard state in solution.This process reflects the sum of all interactions occurring between dissolved ions and solvent. In such a solution the ion-ion interactions are absent, thereby simplifying the discussion of standard enthalpy data. 2.1 Solutions of Inorganic Electrolytes in Water-Organic Mixtures In order to study the character of plots of the enthalpies of electrolyte solvation in water-organic mixtures it is necessary to know the standard enthalpies of solution over the whole range of mixed-solvent compositions. To facilitate the analysis the entire range of AH,” =f(co-solvent content) is frequently divided into three areas, i.e.water-rich, intermediate, and co-solvent-rich regions. 2.1.1 Water-rich Region The properties of mixed solvents in this region show distinctive changes due to the effects of ions and, especially, of organic co- solvents on the three-dimensional structure of water. Figure 1 presents the standard enthalpies of NaI solution in various mixed solvents as a function of binary solvent compo- sitions. Sodium iodide has been predominantly employed from the inorganic salts because of its relatively high solubility in organic solvents. Moreover, extensive data on its heats of solution have been reported in the literature. It is known that curves of AH,”(NaI)vs. alcohol content pass through their maxima in the water-rich region (Figure la), as with the enthalpies of solution of other electrolytes and non- electrolytes.Plots of several properties (spectral, NMR, visco- sity, kinetics parameters of reactions, thermodynamic functions of complex formation processes, etc.) also show variations in the range of compositions corresponding to the AH,”maxima., It is relevant to mention here that there are differing views about the meaning of the extrema (positions, heights, depths) of curves that illustrate the properties of multi-component systems versus their compositions. Some authors5q6 consider extrema of AH: not to be significant, resulting only from the slopes of both ends of these curves (e.g. the McMillan-Mayer interaction coefficients).Other author~~*~-~ are of the opinion that the observed extrema reflect solution structure and intermolecular interactions. The presence of AH,” maxima in the water-rich region has been attributed to ‘iceberg’ formation, structure making or breaking effects, higher-order microphase transi- tions, the appearance of clathrate-like aggregates, and to hydro- phobic effects. Possibly these terms are simply different names for the same phenomena, but it has not been easy to demonstrate this supposition so far. In this article the assumption is made 205 CHEMICAL SOCIETY REVIEWS, 1993 ct 7 -15--201 -20 -30-251::1 I I I 10 20 30 40 50 10 20 30 40 10 20 30 40 mol % alcohol mol % organic solvent mol % organic solvent Figure 1 Standard enthalpies of solution of NaI in water-organic and anions with the solvent.mixtures as a function of organic solvent content at 25°C. (a): (1) The enthalpy maxima of NaI solutions are probably con- methanol; (2) ethanol; (3) propan-1-01; (4)propan-2-01; (5) t-butyl nected with particular difficulties in the formation of solvation alcohol. (b): (6) tetrahydrofuran; (7) hexamethylphosphoramide,ref. shells in the region of the hydrophobically ordered structure of 3. (c): (8) dimethylsulfoxide, ref. 2; (9) dimethylformamide, ref. 3; (10) the solution. The latter leads to the maxima of enthalpy, AH:. dioxane, ref. 2. On the other hand, the positions of AH: maxima for NaCl (Table l), KBr, etc. in water-alcohol mixtures are somewhat that the extrema originate from the hydrophobic hydration of different -as was observed for the enthalpy maxima of NaI and the organic components, and that hydrogen bonds play only a the NaCl in solutions containing propanols and t-butyl alco- minor role.h01.~ In the cases of methanol and ethanol, however, the shifts of In some aqueous non-electrolyte solutions the apparent molar the positions of maxima, if any, are within the limits of error. volume V, of the organic component passes through a mini- The differences in behaviour of NaCl and NaI probably result mum. According to Franks’O the position and depth of V, from a stronger relative affinity to water in the case of C1- due to minimum depend on the size and shape of the apolar (i.e.the smaller size of the chloride anion, its larger electric surface hydrophobic) groups of solute molecules. Therefore, positions density, and lower polarizability. The displacement of the of V, minima can be treated as measures of the hydrophobic AH:(NaCl) maxima positions in relation to the minima of effect. In Table 1 the positions of the minima of the apparent mixing enthalpy for water and alcohol suggests a change in the molar volume of alcohols in water (V,min)’ are juxtaposed structure of the binary solvent caused by sodium chloride. This with those for the maxima of standard enthalpies of solution of view is supported by Franks’ and De~noyers’~ opinion that NaI in these mixtures (dH:max) and with the minima of mixing NaCl causes ‘the shift in the equilibrium between different states enthalpies of water with co-solvents (HEmin).’ The positions of the alcohol-water mixtures’. Unlike the case for anions, the of all three extrema are close to each other. This observation positions of the maxima do not seem to depend on the character could be treated as evidence for the importance of hydrophobic of the cations.interactions in the enthalpy of mixing (except for solutions The pattern for NaI is particularly interesting. Feakins’ old containing methanol). rule ‘When in doubt, leave the iodide out’ can be replaced, at The proximity of the positions of dH,O(NaI) maxima and of least for the solvents in question, by ‘When in doubt, watch the HE minima suggests that NaI does not affect the solvent iodide’.It is possible that I -ion is hydrated in the region from structure. It can also be postulated that this absence of a NaI pure water to the HEminima (dH:maxima), at which point the effect is the result of equalization of the interactions of cations alcohol molecules invade the solvation shells. The Na + cation is Table 1 Minima positions of the apparent molar volume, V4, heat of mixing, HEfor water-organic mixtures and maxima positions of the standard enthalpies of NaI and NaCl solution in these mixtures* dH,D(NaI) HE dH,O(NaCl)b Organic solvent (min)ll ma^)^ (min)12.16,17 ma^)^,^ Methanol 17 30 17 Ethanol 12 15 12 Propan-1-01 6 5 7.5 Propan-2-01 9 10 12 t-Butyl alcohol 6 6 9 Tetrahydrofuran 2.5 15 25 Hexameth ylphosphoramide 1.5 20 10 * The accuracy with which the extrema positions were determined tends to differ, and is dependent on the data available for the above analysis. ELECTROLYTES IN BINARY SOLVENTS-S.TANIEWSKA-OSINSKA probably hydrated in a wide range of solvent compositions. In the case of chlorides, bromides, and other small anions the courses of these curves are different, thus showing a pertur- bation effect of such anions on the structure of mixed solvents. That the preferential solvation of ions plays some role here cannot be ruled out. Significant conclusions concerning the preferential solvation of both inorganic and organic ions in binary aqueous mixtures containing four alcohols (from methanol to t-butyl alcohol), acetone, and dimethylsulfoxide (DMSO) can be found in a series of important papers by Blandamer et al.l4 The authors analysed the transfer chemical potentials of simple and complex ions as the basis for a further analysis of kinetic data in systems involving ions in the chosen mixtures. Hawlicka,' applying the self-diffusion method to the consti- tuents of electrolyte solutions in water with methanol, propan- 1-01, and acetonitrile also expressed views on the preferential solvation of ions. Unfortunately, the conclusions drawn by various authors, using different experimental methods, are not always identical. Feakins et al.,s on the basis that the mixing enthalpy of water with methanol, HE, passes through a minimum at a mole fraction of methanol xMeOH = 0.3, while the enthalpies of transfer of alkali metal halides AH: have their minima in the region xMeOH = 0.1-0.2, have developed a general relationship HE/xalc=flA HPr/xalc).They applied the relationship to the alkali metal chlorides in water-methanol, and obtained a linear relation for a large part of the range of solvent composition.They concluded that neither the absence of a maximum in A H,O,, nor its exact position are of any significance for the structure of a mixed solvent. However, these authors did not take into account the fact that AH$ curves of MeI, MeClO,, etc., pass through the extrema at the same positions, like those for HEof water with alcohols (with the exception of methanol). Therefore, the above relationship seems to be more appropriate for solutions other than those used by Feakins et al.As well as alcohol-containing systems, there are also electro- lyte solutions in mixtures of water with aprotic solvents, viz. tetrahydrofuran and hexamethylphosphoramide (Figure 1b) in which the curves of AH,"(salt) =f(organic solvent content) also exhibit ma~ima.~ The maxima are shifted by anions (Table 1) in a manner similar to systems containing alcohols. However, the positions of the solution enthalpy maxima for alkali metal iodides are not close to those for the mixing enthalpy minima of these binaries,' 6,1 as is observed in the case of the methanol- water mixed solvent (Table 1). It can be postulated that the specific behaviour of methanol, tetrahydrofuran, and hexamethylphosphoramide as the compo- nents of electrolyte solutions involves equalizing, at least to some extent, the effects of the hydrophilic and hydrophobic groups in the solvent molecules.Solutions in binaries composed mainly of non-self-associating aprotic solvents do not display the AH,"maxima for NaI and other inorganic electrolytes (Figure lc). In these systems the interaction of the electron-donor groups of the solvent with water is a more important factor than the hydrocarbon-water interaction. Therefore, the ability to interact in a hydrophobic way, revealed by the V, minima measure, is not a sufficient criterion for a comprehensive description of these solutions. The solution enthalpy maxima of NaI do not occur in water mixtures with hydrophilic co-solvents (formamide, urea),3 whose interactions with water consist mainly in hydrogen-bond formation.Numerous association equilibria occur in these systems, varying with changes in composition of the mixed solvent. Electrolyte solutions in water-polyol mixed solvents have been investigated only to a limited extent. The curves of enthalpy of KI solution in water with ethane-l,2-diol or glycerol over the considered range of mixed solvent composition appear to be monotonic,2 pointing to similar properties for these co-solvents (Figure 2). These solvents exhibit a strong tendency to form intermolecular hydrogen bonds. The weak hydrophobicity of f 20L I 1 I 10 20 30 40 50 mol % organic solvent Figure 2 Standard enthalpies of solution of inorganic salts in water- polyol mixtures as a function of organic solvent content at 25 "C.(1) KI-water+thane-1,2-diol; (2) KI-water-glycerol, ref. 2; (3) NaCl- water-propane- 1,2-diol; (4) NaCl-water-butane- 1,4-diol, ref. 18. ethane- 1,2-diol and glycerol displayed, for example, by the shallow V, minima in mixtures with a high water content cannot be observed in electrolyte solutions involving these mixtures. Unlike the above mentioned solutions, the enthalpies of inorganic salt solutions in water with propane- 1,2-diol or butane-1,4-dio11 exhibit their maxima in the water-rich region. The stronger hydrophobic effect of butane- 1,4-diol, with its ethyl group unsubstituted by -OH groups, is demonstrated by the higher maximum.2.1.2 Enthalpic Pair-interaction Coeficients Standard enthalpies of electrolyte solutions provide some infor- mation about the interactions between the solute and the bulk solvent. In order to analyse the interactions between solute and organic co-solvent molecules in water, the pair interaction coefficients derived from the McMillan-Mayer theory were applied to the transfer enthalpy by Desnoyers et a1.I9 The enthalpic pair-interaction coefficients were evaluated for several electrolyte solutions in water-organic binary solvents (hES). The standard enthalpy of electrolyte solution in the mixture of water with a small amount of organic co-solvent can be pre- sented as follows: AH:(E in S + W) -AH:(E in W) = AHP,(E) = 2m,h~s+ 3mshEss+ .... (1) hence where AH,"is the standard enthalpy of electrolyte solution, AH?, is the transfer enthalpy of electrolyte from water to solvent S, and rn, is molality of solvent S. In Table 2 a number of hESvalues for NaI and NaCl with several co-solvents in water are presented.20 Different hESvalues are observed for chlorides and iodides in all binaries. Observed higher hEScoefficients, in their absolute values, for NaI in comparison with those for NaCl seem to indicate a greater affinity of NaI for organic solvent polar groups, probably as a result of the higher polarization I-ion. It has been also found that in aqueous systems the enthalpic pair interaction coefficients hESof NaCl or NaI with different non-electrolytes are linearly correlated with some properties of the non-electrolytes, such as molecular polarizability (ap),elec-tric permittivity (E), or Kosower acidity parameter (2).How-ever, these correlations are not general and hold only for groups 208 Table 2 Enthalpic pair-interaction coefficients, hES for NaI-non-electrolyte and NaC1-non-electrolyte in water2O hEs/J.kg.mol-’) Non-electrolyte NaI NaCl Methanol 3 14 300 Ethanol 596 580 Propan-1-01 780 740 Propan-2-01 1018 900 t-Butyl alcohol 1140 980 Tetrahydro furan 344 404 Hexamethylphosphoramide 564 842 Acetone -92 20 Dimethyl sulfoxide 629 -202 Formamide -696 -452 Dimethylformamide -350 -79 Acetonitrile -494 -286 of related compounds.For instance, a graph of hESas a function of the Kosower acidity parameter consists of two straight line segments: one for the pairs containing alcohols and the other for those with aprotic solvents.The best linear correlation, which encompasses almost all systems investigated so far, was obtained between the hEs(NaCl/NaI-non-electrolyte) values and the heat capacities of transfer of the non-electrolytes from the gas phase to high dilution in water (cp,O-Cpg),or, better still, of the heat capacities of interaction Cp(int) between a non-electrolyte and water (Figure 3). The possible reasons for the observed hESbehaviour were discussed in detail in reference 20. Cp(int)/J rno1-lK-l -120 -80 -40 0 40 80 111111~111111 1400 800 -EtOH2 6001 L I -0 or 5 200 -1 200 P -1000 -800 600 -600 THFj‘ZEt OH -400 -200 -0 --200 Nal -S --400 --600 FA --800 Ill 1 1 1 1 1111 1--120 -80 -40 0 40 80 Cp(int)/J mol-’ K-’ Figure 3 Enthalpic pair interaction coefficients,hESfor electrolyte-non-electrolyte as a function of the heat capacity of interaction, C,(int) of the non-electrolyte.MeEtOH = 2-methoxyethanol, EtEtOH = 2-ethoxyethanol, FA = formamide, DMF = N,N-dimethylformamide, AA = acetamide, DMA = N,N-dimethylacetamide, DME = 1,2-dimethoxythane, THF = tetrahydrofuran, DMSO = dimethyl sul-foxide, ACT = acetone, AN = acetonitrile, U = urea; data from ref. 20. CHEMICAL SOCIETY REVIEWS, 1993 2.1.3 Intermediate Region Having passed the region of composition at which solution enthalpy maxima occur, the AH: curves usually become smooth in the system containing alcohols etc.In the middle range of organic co-solvent contents, in all cases so far investigated, hydrophobic effects are already absent and changes in the equilibria of aggregates formed by hydrogen bonds occur conti-nuously. According to Naberukhin,2 water molecules form ‘globules’in this region, i.e. small ice-likeaggregates, and the co-solvent molecules probably form self-associates. Alternatively, entities containing molecules from both substances, linked by hydrogen bonds, may also form. By this hypothesis, the middle range of compositions would be microheterogeneous. This view is supported by the observation that for some mixtures, demix-ing occurs below critical solution temperature in this region.For example, a mixture of water with acetonitrile (approximately 3:2 molar ratio) separates into two phases at -1 “C. 2.1.4 Co-solvent-rich Region Enthalpy plots in the non-aqueous region are often featureless. However, in alcohol-rich regions the solution enthalpies of salts exhibit minima within the 70-90 mol% alcohol range. These minima were first observed for dH:(NaI) in water with buta-nol~,~and then for CaCl, in water-alcohol mixtures from methanol to propano1.22Sincethe minima of the electric permit-tivity and viscosity23of the water-butanol mixtures, without an electrolyte, correspond to the same alcohol content region as in the case of dH,O(salts) (Figure 4), it was concluded that such minima are due to solvent structure.This conclusion is in accordance with ideas about the formation of Brown-Ives ‘centrosymmetric’ associates, composed of one water and four alcohol molecule^.^ sfF1 4.6-4.2-20-3.8-16-40 60 80 100 mol Yot-butyl alcohol Figure 4 Viscosities, 7,and electric permittivities, E, of water-t-butyl alcohol mixtures and standard enthalpies of solution of Nal in these mixtures, AH!, at 26 “C (t-butyl alcohol-rich region), data from refs. 3.7, and 23. The analogous minimum of enthalpy of NaClO, solution was noticed in the 90 mol% tetrahydrofuran-rich region.24(In this range NaI is almost insoluble.) It should also be added that the mixing enthalpy of water with propanols, and butanols, as well as with tetrahydrofuran, exhibits a maximum in the same organic-rich area as the minimum in the case of the solution enthalpy of electrolyte. The evident minima of solution enthalpy are also observed for all solutes so far investigated in the acetonitrile-water system in the region of 80-90 mol% acetonitrile.’ Some authors attri-bute this minimum to the preferential hydration of ions, whereas ELECTROLYTES IN BINARY SOLVENTS-s.TANIEWSKA-OSINSKA others refer to the formation of complexes from molecules of acetonitrile and water, with a cation or even with both ions.It is noteworthy that the heat of mixing for water-acetonitrile mix-tures exhibits a minimum at 70 mol% organic solvent, and the azeotrope (t = 76 "C) contains about 69 mol% acetonitrile.These extrema in properties suggest that acetonitrile molecules form mixed associates with water which affect the solution enthalpies of electrolytes. 2.2 Solutions of Salts with Organic Ions That organic ions devoid of groups specifically interacting with solvent exhibit special properties when dissolved in water is well- known. In spite of their large size, in comparison with alkali metal and halide ions, they probably do not destroy the three- dimensional structure of water, but rather enhance it. That view can be supported by the large and positive heat capacities of electrolytes containing these ions in water, positive viscosity B coefficients from the Jones-Dole equation, and negative struc- tural entropies of hydration ASPstr, etc.These somewhat unex- pected values display some similarity to the behaviour of organic ions and non-polar solutes, in contrast to that of inorganic ions. The strange properties of the ions studied were attributed by many authors to the hydrophobic hydration and were also observed in water-organic binary solvents. In the water-rich range, curves illustrating the AH,"=florga-nic solvent content) of electrolytes containing ions with aryl groups exhibit maxima in all systems so far investigated. ,i2 5--27 Figure 5 shows plots of the standard solution enthalpies of the chosen electrolyte, NaBPh,.The appearance of high enthalpy extrema for electrolytes with organic ions can probably be attributed to competition between an organic co-solvent and organic ions for the water molecules needed to form hydration shells, quasi-clathrates, or other such structures.This phenome- non could possibly be due to hydrophobic effects. 4 401 2011 %"a ~ ~ ~l ing the degree of hydrophobicity of the organic co-solvent and comparing it to the characteristics of the AH:, maxima. A comprehensive theory of liquids or hydration that could be utilized to determine hydrophobic effects has yet to be postu- lated. Kessler,, * using the McMillan-Mayer theory, recently proposed the semi-empirical parameter (8B,,/dp)Tas a reason- able measure of the hydrophobic interactions of two dissolved molecules. B,, is the second virial coefficient of the power series of the osmotic pressure as a function of solution density.where ~7 denotes isothermic compressibility and a;2 the Lewis- Randall virial coefficient. The chosen parameter is positive for hydrophobic solutes in water and negative for hydrophilic ones. It was pointed out by DesnoyersZ9 that the partial molal heat capacities of solutes in water Cpg are associated with the hydrophobic hydration. As is known, the hydrophobic hyd- ration and hydrophobic interactions are of a common origin. The correlation between (8B,,/8p)T and C,02(Figure 6) shows that C,02can also quantitatively describe the hydrophobic effects. Knowing the parameters for hydrophobic effects, it is possible to compare them with the heights of the AHE(NaBPh,) maxima in plots of water-organic mixtures.The Na+ ion in the NaBPh, does not seem to have a significant influence on the heights of enthalpy maxima. t BuOH nBuOH 7 !i cI-\f 300-THF 7Y -200 l l.l l.l 20 40 60 80 100 mol % organic solvent Figure 5 Standard enthalpies of solution of NaBPh, in water-organic mixtures as a function of organic solvent content at 25°C. (1) methanol, ref, 25; (2) ethanol, ref. 26; (3) tetrahydrofuran, ref. 27; (4) acetonitrile; (5) dimethyl sulfoxide; (6)dioxane, ref. 2. In systems containing NaBPh,, or other salts with organic ions dissolved in binary organic solvents, the solution enthalpy maxima are absent. This observation supports the hypothesis regarding an hydrophobic origin for the maxima in water- containing systems.In solutions of one common salt, a signifi- cant effect for the hydrophobic hydration of co-solvents on the shape and position of maxima could be confirmed by determin- OF*\ 1 -1 0 1 -5 1 0 I 5 4 ab,iap. 1o2 (cm3 mot-' bar-') Figure 6 Correlation between the partial molal heat capacities of solutes in water, Co and the semi-empirical parameter, dB,,/ap; the correla- tion coeffic!& r = 0.943;data from refs. 11 and 28. (DEF = diethyl-formamide, D = dioxane, other symbols are the same as for Figure 3.) In Figure 7 dHpr(NaBPh,) maxima are presented as a func- tion of cj2of the organic co-solvents. The reference function Cj2 was chosen because it is known for a larger number of solvents than the Kesler coefficient.Two straight lines are produced: one corresponds to systems containing alcohols and the other to mostly aprotic solvents. The distance between these two lines is approximately equal to the hydrogen bond enthalpy. This pattern can be explained by the ability of alcohols to form more hydrogen bonds than is the case for the aprotic solvents. / / 0 / HMW /nPrOH /// Figure 7 Correlation between the transfer enthalpies of NaBPh, from water to water-organic mixtures, corresponding to H:max and the partial molal heat capacities of organic solvent in water, C& correla-tion coefficients, rl = 0.948 and rz = 0.995; data from refs. 2, 11, and 25-27. (HMPA = hexamethylphosphoramide, GE = ethane-I ,2-diol, other symbols are the same as for Figure 3.) 2.3 Solutions of Salts in Organic-organic Solvents Solutions of electrolytes in binary non-aqueous organic solvents are often used in Chemistry and Technology.Results from the studies of salt solutions in organic-organic binaries conducted by various experimental methods can be found in the literature. However, data obtained by thermochemical studies are quite scarce. Krest~v,~~ Taniewska-O~inska,~ and their co-workers have performed calorimetric measurements dealing with ther- mochemical properties of the discussed three-component system. A question considered in some papers concerned differ- ences in the behaviour of salt solutions in both aqueous and non- aqueous mixed solvents.Taking into consideration the similar molecular structure and dipole moments of water and methanol, the standard enthalpies of the chosen electrolyte (NaI) were measured in mixtures of methanol with three other alcohols, of diminishing electric permitti~ities.~ In the methanol-rich range, no maxima in AH,O(NaI) were observed (Figure 8), which shows the absence of an ordering structure effect or of a phenomenon analogous to the hydrophobic hydration (solvophobic solva- tion). This observation could be attributed to the difference in bulk structure between water and methanol. In some systems, such as NaI-methanol containing propan-l- 01 and propan-2-01, the curves of dH,O(NaI) vs. solvent compo- sition exhibit small maxima in the low methanol composition range. In these solutions, spatial associations presumably arise.Correcting for ionic association does not lead to the loss of these maxima.32 CHEMICAL SOCIETY REVIEWS, 1993 -201 -45 t -501 I -55 I,,,,, 20 40 60 80 1( mot % co-solvent Figure 8 Standard enthalpies of solution of NaI in methanol-organic solvent mixtures as a function of cosolvent content at 25 “C; data from ref. 31 (NM = nitromethane, other symbols are the same as for Figure 3). The enthalpy of solution for NaI in methanol with three aprotic solvents, of similar electric permittivities as that of methanol (Figure 8), was determined.31 It follows from the analysis of the plots of dH,O(NaI) =f(co-solvent content) in isodielectric mixtures that specific ion-solvent interactions are more important than the electric permittivity of the solvents.In the systems containing NaI, AgC1, or AgBr etc. in metha- nol with acetonitrile, enthalpy minima were observed in the acetonitrile-rich region (approximately 80 mol %).31,33 This observation is analogous to that made for the electrolyte-water- acetonitrile solutions. Consideration of this minimum, together with the results of spectroscopic investigations, sup-ports the view that acetonitrile forms complexes with water, and possibly with methanol. The cations are probably attached to the oxygen atoms of water, or methanol, thus increasing the solvation energy in the region under consideration. Glycerol is a solvent considered to be water-like because of its three-dimensional network of hydrogen bonds in the crystal and, probably, in the liquid state.Evidence supporting such a comparison is found in the monotonic plots of dH,”(NaI) vs. organic solvent content for the entire range of water-glycerol mixture compositions (Figure 2). As a comparison of these two liquids as components of ternary systems, Figure 9 shows the standard enthalpies of a solution of NaI in glycerol with three monohydroxyalcohols.34 In the glycerol-rich range the AH,” maxima are observed, as is the case for all salt solutions in water- alcohol mixtures. This could indicate the appearance of a structure ordering effect, analogous to the systems containing small amounts of alcohols in water. The minima of enthalpy in solutions with small amounts of glycerol may suggest the existence of some kind of associates.Unlike the character of AH,O(NaI) curves for glycerol with alcohols, the analogous functions for NaI solutions in glycerol with five diols (Figure 9) are monotonic, without distinct e~trema.~~Based on the observation mentioned above it can be ELECTROLYTES IN BINARY SOLVENTS-S TANIEWSKA-OSINSKA 21 1 radii are discrepant Conway prefers a method which uses the assumption A HsoI,(BPh,) = A HsoI,(Ph,P + or Ph,As +)based on the spherical shape of these organic ions, on their large size and thus small surface density of charge, and on the lack of specific interactions of these ions with all solvents These organic ions \ were the object of numerous investigations, and the results led some authors (Krishnan and Fr~edman~~ -281 and others3’) to conclude that there was a marked difference in behaviour between BPh, anion and Ph,P+ or Ph,AS+ cations The different properties of these ions are probably the result of the more hindered rotation of phenyl groups in the BPh; anion compared with the Ph4P+ cation, and from a difference in charge distribution on the surfaces of these ions This could be an explanation for why the application of known methods to the splitting of solvation or transfer enthalpies in simple solvents -’1-40 produces somewhat different ionic enthalpy values The situation is more complicated with respect to electrolyte T solutions in water-organic solvents The TATB (or TPTB) splitting method leads to different AH?, curves for cations and anions as a function of water-organic solvent compo-sitions 25 27 In the water-rich region the curves exhibit cationic maxima and anionic minima that correspond to the same organic component contents The reason for the strange shape -341 of plots of dHp,(ion) obtained by the TPTB method, according to Taniewska-Osinska and Nowi~ka,~~ seems to be the different interactions of organic ions with binary solvents, leading to a difference in hydrophobic hydration This observation supports the suggestion that the TPTB method is not generally applicable to solutions in water-organic mixtures On the other hand, adoption of the dHP,(Cs+) = AHP,(I-)-267-30 method, or others based on the equality of enthalpy of inorganic 20 40 60 80 100 mol % co-solvent Figure 9 Standard enthalpies of solution of NaI in glycerol-organic mixtures as a function of cosolvent content at 40 “C,data from ref 34 (1) methanol, (2) ethanol, (3) propan-1-01, (4) ethane-172-diol, (5) propane-1,3-dlol7 (6) propane- 1,2-diol, (7) butane-1,2-dio17 (8) butane- 1,4-diol supposed that glycerol is capable of solvophobic solvation, but to a lesser degree than water The enthalpic pair electrolyte -non-electrolyte interaction coefficients for all the systems men- tioned above vary linearly with the reciprocal electric permittivi- ties of the co-solvents The character of enthalpy plots may reflect not only the change in structure of one solvent affected by another or by a dissolved electrolyte, but also the change in composition of solvation shells, the variation of solvation numbers, or the selective solvation Unfortunately, the methods for the analysis of the latter phenomenon concern chiefly dG,” and not AH! Some of the attempts to determine selective solvation from enthalpy seem to be based on such simplifications, and they should not be applied to different systems without hesitation 3 Some Remarks on the Splitting of Enthalpy of Electrolytes into Ionic Contributions The enthalpies of solvation, or transfer of electrolytes, depend on the properties of the ions constituting the salts, as well as on the physicochemical properties of the solvents In order to establish the relationship between ion-solvent interactions and the charge sign, the character of the ion, and the properties of the solvent, it is necessary to divide enthalpies of solvation into their individual ionic contributions Ionic enthalpies cannot be obtained theoretically, but only by using extra-thermodynamic assumptions Methods of division have been described and critically analysed by C~nway,~ who considered methods based on splitting thermodynamic values for salts containing cations and anions of similar radii as being unsatisfactory for the case of small ions In his view, ionic size affects the properties of cations and anions in differing ways, and the existing values of ionic ions, does not cause extrema for enthalpies of individual ions in binary solvents where the AH: of whole salts do not exhibit them, eg in water with most aprotic solvents In the case of water-alcohol mixtures, the AH: curves of inorganic salts pass through maxima similar to those of the ionic enthalpies The suggestion regarding the preponderance of the split- ting method of AHP,(Cs+) = AHP,(I-) over AHP,(Ph,P+) = dHP,(BPh,) is in accordance with the view of Friedman et a1 36 and Somsen et al ,6 and is also supported by the author of this article 4 Conclusion The significant progress made in the study of electrolytes in binary solvents has been summarized in this review, the analysis having shown the increased understanding of the nature of the complex interactions between components of this kind of system It is reasonable to suppose that future studies will bring forth valuable ideas and results concerning, in particular, new systems, techniques, and methods of interpretation 5 References 1 K P Mishchenko and K P Poltoratskii, ‘Problems of Thermody- namics and Structure of Aqueous and Nonaqueous Electrolyte Solutions’, Plenum, New York, 1972 2 G M Poltoratskii, ‘Thermodynamic Characteristics of Non-aqueous Electrolyte Solutions’, Khimiya, Leningrad, 1984 (in Russian) 3 S Taniewska-Osinska and H Piekarski, Acta Univ Lodz Folia Chimica, 1986, 6, 1 4 S Taniewska-Osinska and H Piekarski, J Solution Chem , 1978,7, 891 5 E De Valera, D Feakins, and W E Waghorne, J Chem SOC, Faraday Trans I, 1983,79, 1061 6 M BOO~Jand G Somsen, Electrochim Acta, 1983,28, 1883 7 A C Brown and D I G Ives, J Chem Soc , 1962, 1908 8 R Bury, A Mayaffre, and M Chemla, J Chem Phys ,1977,74,745 9 F Franks and J E Desnoyers, in ‘Water Science Reviews 1 ,ed F Franks, Cambridge University Press, 1985, p 171 10 F Franks, J Chem SOC Faraday Trans 1, 1977,73,830 11 M Jozwiak, Thesis, University of Lodz, 1989 CHEMICAL SOCIETY REVIEWS, 1993 12 B.Marongiu, J. Ferino, R. Monaci, V. Solinas, and S. Torrazza, J. Mol. Liquids, 1984,28, 229. 13 Y.Pointud, J. Juillard, L. Avedikian, J. P. Morel, and M. Ducros, Thermochim. Acta, 1974,8,423. 14 M. J. Blandamer, B. Briggs, J. Burgess, P. Guardado, S. Radulovic, andC. D. Hubbard, J.Chem, SOC.,Faraday Trans. I, 1988,84,1243.15 E. Hawlicka, 2.Naturforsch., Teil A, 1986,41,939; 1988,43, 769. 16 0.Kiyohara and G. C. Benson, Can. J. Chem., 1977,55, 1354. 17 J. Jose, R. Philippe, andP. Clechet, Can.J. Chem.Eng., 1975,53,88. 18 G. A. Krestov, J. V. Egorova, and V. V. Korolev, Zh. Fiz. Khim., 1985,59,2327. 19 J. E. Desnoyers, G. Perron, L. Avedikian, and J. P. Morel, J. Solution Chem., 1976,5,631. 20 H. Piekarski and M. Tkaczyk, J. Chem. SOC.,Faraday Trans. I, 1991, 87, 3661. 21 Yu. I. Naberukhin and V. A. Rogov, Usp. Khim., 1971,40, 369. 22 S. Taniewska-Osinska and J. Barczynska, J. Chem. SOC.,Faraday Trans. I, 1984,80, 1409. 23 S. Taniewska-Osinska and A. Kacperska, Pol. J. Chem., 1979, 53, 1351, 1979,53, 1673. 24 S. Taniewska-Osinska, Z. Kozlowski, B.Nowicka, A. Bald, and A. Szejgis, J. Chem. Soc., Faraday Trans. I, 1989,85, 479. 25 M. H. Abraham, T. Hill, M. Ch. Ling, R. A. Schulz, and R. A. C. Watt, J. Chem. Soc., Faraday Trans. I, 1984,80,489. 26 E. M. Arnett, W. G. Bentrude, J. J. Burke, andP. McC. Duggleby, J. Am. Chem. SOC.,1965,87, 1541. 27 S. Taniewska-Osinska and B. Nowicka, Thermochim. Acta, 1987, 115, 129. 28 Yu. M. Kessler and A. L. Zaycev, ‘Solvophobic Effects’, Khimiya, Leningrad, 1989, p. 104, (in Russian). 29 0.Kiyohara, G. Perron, and J. E. Desnoyers, Can. J. Chem., 1975, 53, 3263. 30 G. A. Krestov, Zh. Fiz. Khim., 1966, 7, 608. 31 S. Taniewska-Osinska and A. Piekarska, J. Chem. SOC.,Faraday Trans. I, 1985, 81, 1913; A. Piekarska, H. Piekarski, and S. Taniewska-Osinska, ibid., 1986, 82, 5 13. 32 S.Taniewska-Osinska, A. Piekarska, A. Bald, and S. Szejgis, Phys.-Chem. Liq. 1990,21,217. 33 B. G. Cox and W. G. Waghorne, J. Chem. SOC.,Faraday Trans. 1, 1984,80, 1267; B. G. Cox, R. Natarajan, and W. E. Waghorne, ibid., 1979, 75, 86. 34 J. Woznicka and S. Taniewska-Osinska, J. Chem. SOC.,Faraday Trans. I, 1983,79,2879; 1986,82, 1299; J. Woinicka, STaniewska- Osinska, and A. Bald, ibid., 1989,85, 3335. 35 B. E. Conway, J. Solution Chem., 1978,7, 721. 36 C. V. Krishnan and H. L. Friedman, in ‘SoluteSolvent Interac- tions’, ed. J. F. Coetzee and C. D. Richtie, Marcel Dekker, New York, 1976, Chapter 1, p. 17. 37 J. E. Coetzeeand W. R. Sharpe, J. Phys. Chem., 1971,75,3141.

ISSN:0306-0012    

DOI:10.1039/CS9932200205

出版商:RSC    

年代:1993

数据来源: RSC