摘要:

ISSN 0306-001 2 CSRVBR 22(6) 361-442 (1 993) Chemical Society Reviews Volume 22 Issue 6 Pages 361-442 December 1993 TILDEN LECTURE. Organometallic Intermediates: Ultimate Reagents By Robin N. Perutz (pp. 361 -369) Short-lived intermediates play a central role, through their structure and reactivity, in controlling photochemical reactions of organo-transition metal complexes. Molecules such as Ru(dmpe), (dmpe = Me,PCH,CH,PMe,) can be studied directly by time-resolved spectroscopy and matrix isolation to reveal key features of reaction mechanisms. Photochemical reductive elimination of H, may be complete in picoseconds and re-addition of H, in tens of nanoseconds. Excited states of organometallics are more elusive, but sometimes emission and excitation spectra are sufficiently well resolved to estimate the change in geometry in the excited state.Determination of Molecular Conformation from Large Amplitude Vibrations in Electronic Spectra of Organic Molecules in a Supersonic Jet By J. Michael Hollas (pp. 371 -382) Molecules having low wavenumber, large amplitude vibrations are easily distorted in the direction of the corresponding normal mode. Therefore the conformation may be changed appreciably when a molecule is excited from its ground state to another electronic state and there is a tendency for low wavenumber vibrations to have large Franck-Condon factors in electronic transitions. The resulting vibrational progressions can be used to access many levels of what may be inversion, ring-puckering, or torsional vibrations.Fitting these levels to a model potential allows the conformation of the molecule to be obtained. Enantioselective and Diastereoselective Molecular Recognition of Neutral Molecules By Thomas H. Webb and Craig S. Wilcox (pp. 383-395) Biological processes rely upon the specificity of biomolecular interactions. Shape-selective transport and catalysis are facilitated by proteins that show high selectivity for specific carbohydrates, nucleotides, lipids, or peptides. Contemporary chemists are striving to create synthetic receptors that will show similar levels of selectivity. The purpose of these efforts is to illuminate the general mechanisms used by Nature to achieve I I molecular shape-selectivity, and to provide new synthetic receptors that can be used in separation technologies, drug delivery, or as the basis of new catalytic processes.Recent progress toward this objective is reviewed. The Hydrogen Bond and Crystal Engineering By Christer B. Aakeroy and Kenneth R. Seddon (pp. 397-407) The hydrogen bond, particularly between ions, is much stronger than commonly acknowledged, and the number of recognized hydrogen-bond donors (such as C-H) and acceptors (such as aromatic r-clouds) is rapidly expanding. We present here an overview of the hydrogen bond, and show how hydrogen-bond patterns in the solid state can be classified using a topological analysis. We conclude by illustrating the use of the hydrogen bond as a synthetic vector for crystal engineering, and consider the problems of polymorphism (a matter of immense current interest, given the multi-billion dollar law suit of Glaxo versus Novopharm,which has polymorphism as its cynosure).The Properties of Organic Liquids that are Relevant to their Use as Solvating Solvents By Y. Marcus (pp. 409-41 6) The solvation properties of solvents (solvent effects) depend mainly on their polarity/polarizability (accounting also for dispersion interactions), hydrogen bond donation and acceptance abilities, and cohesive energy density, that are orthogonal to each other. These are best measured by the Kamlet-Taft solvatochromic parameters T*, 01, and /3 and the square of Hildebrand’s solubility parameter, .8h, respectively.The first three are presented for a large set of solvents, together with several other solvation indices that are correlated with them.Water Purification by Semiconductor Photocatalysis By Andrew Mills, Richard H. Davies, and David Worsley (pp. 41 7-425) The basic principles of the photooxidative mineralization of organic pollutants by 02,sensitized by TiO,, are described. The kinetics of this process as a function of [TiO,], [organic pollutant], [O,], light intensity, temperature, pH, and the type of anion present are discussed, and a general kinetic model is presented. Standard test and demonstration systems for water purification by TiO, photocatalysis are described and other novel applications of semiconductor photocatalysis are outlined. Mechanisms of Solvolytic Alkene-forming Elimination Reactions By Alf Thibblin (pp.427-433) Different types of alkene-forming elimination reactions that accompany substitution by the solvent as well as heterolysis reactions in non-nucleophilic solvents are treated. The article discusses kinetic deuterium isotope effects for such reactions and summarizes the use of extreme kinetic deuterium isotope effects as a mechanistic probe of reaction-branching in solvolytic reactions. Mechanistic borderlines are discussed. Electrochemistry in Media of Low Dielectric Constant By Andrew Abbott (pp. 435-440) This article describes the use of low dielectric constant solvents in electrochemical investigations. It outlines how such solvents can be made conducting, how their solvent properties can be characterized, and how various electroactive solutes can be dissolved in them.Some of the practical applications of these solvents are discussed together with some of the theoretical implications that a low dielectric constant has upon solution chemistry. Articles that will appear in forthcoming issues include Magnetic Field Gradients in NMR Friend or Foe7 T. J. Norwood Helical Poly(isocyanides) R. J. M. Nolte INDUSTRIAL LECTURE Polyelectrolyte Materials-Reflections on a Highly Charged Topic J. W. Nicholson The Kirkwood-Buff Theory of Solutions and its Application K. E. Newman Tetrathiafulvalenes as Building-blocks in Supramolecular Chemistry T. Jergensen, T. H. Hansen, and J. Becker Solvent Structure and Perturbations in Solutions of Chemical and Biological Importance J.L. Finney and A. K. Soper Thin Film Diamond by Chemical Vapour Deposition Methods M. N. R. Ashfold, P. W. May, C. A. Rego, and N. M. Everitt Thermodynamics of Polar Additives in Surfactant Solutions R. De Lisi and S. Milioto Electrophoresis of Small Particles C. Boxall 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 &90 00, U S A $198 00, Canada El04 00+ GST, Rest of World &99 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 I1431 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 UK Members of the Royal Society of Chemistry may subscribe to Chemical Sociefy Revieus at 230 00 per annum, they should place their orders on the Annual Subscription renewal forms in the usual way

ISSN:0306-0012    

DOI:10.1039/CS99322FP019

出版商:RSC    

年代:1993

数据来源: RSC

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 Ir. J. M. Brown Ir. J. Burgess Ir. N. Cape f'rofessor A. Hamnett IIr. T. M. Herrington I'rofessor R. Hillman I'rofessor R. Keese IIt-. T. H. Lilley I3.H. Maskill I3ofessor Dr. A. de Meijere I>rofessor J. N. Miller I3ofessor S. M. Roberts I'rofessor B. H. Robinson IDr. A. J. Stace Staff Editors Mr. K. J. Wilkinson Dr. J. A. Rhodes Dr. M. Sugden University of Sussex University of Lei ceste r 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 Loug hborough University of Technology University of Exeter University of East Anglia University 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 trulyinternational 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 A4jAmerican 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 Black bear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99322FX022

出版商:RSC    

年代:1993

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99322BX024

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

TILDEN LECTURE. Organometallic Intermediates Ultimate Reagents Robin N. Perutz Department of Chemistry University of York York YO I 5DD U.K. 1 Reaction Intermediates Since this review revolves around the study of reaction inter- mediates it is worth starting by rehearsing the benefits of such research. The first purpose is to determine the molecular and electronic structure of the intermediates many of which are open-shell molecules. The molecular and electronic structure of open-shell molecules cannot be deduced reliably from the study of closed-shell molecules. You may satisfy yourself of this assertion by considering whether you can understand the methyl radical simply from knowledge of methane and ammonia or whether the structure of Cr(CO) can be deduced by studying Cr(CO) and Fe(CO) (see below).The second feature of reaction intermediates follows from the first because they often have unexpected structures they illustrate new principles of bonding. Moreover by their very nature they are more reactive than closed-shell molecules so they illustrate new principles of reactivity. Again Cr(CO) will serve as an illustration. Thirdly reaction intermediates have a crucial bearing on reaction mech- anisms. If you don't understand the intermediates you don't understand the mechanisms. Unfortunately too many authors still concentrate on kinetics at the expense of reaction inter- mediates; a knowledge of kinetics is not enough to deduce mechanism. Finally there are often incidental benefits to study- ing intermediates for instance in synthesis. It should already be apparent that such high expectations from the study of reaction intermediates cannot be fulfilled unless the maximum possible spectroscopic information is available one technique will rarely be sufficient. Before moving to recent work in York I will illustrate the success of these principles by comparing knowledge of Cr(CO) in 1978 to that today. Cr(CO) is an intermediate which is usually generated photochemically en route to substitution products as in equation 1. By 1978 most of the matrix isolation work on Cr(CO) was complete (for an introduction to matrix photochemistry see ref. 1). We had demonstrated by isotopic labelling that Cr(CO) has a square pyramidal structure [unlike Fe(CO),] and the structure Robin Perutz has been at the University of York since 1983 whese he is now Professor of Chemistry and Royal Society-British Gas Senior Research FelloM,. In his research he combines prepara- tive photochemistry of organo-metallics with matrix isolation and time-resolved spectro-scopy with a special interest in C-H bond activation. Prior to the move to York he held Demonstratorships at Oxford (1977-83) and Edinburgh (1975-77) and a European Exchange Fellowship in Miil-heim (1975-76). His Ph.D. work partly in Cambridge partly in Newcastle was super-vized by J. J.Turner. 36 1 had been rationalized by molecular orbital arguments.2 We had shown by an investigation of the visible absorption spectrum in a variety of matrices both pure and mixed that the sixth coordi- nation position of Cr(CO) was blocked by a 'token' ligand (a phrase coined later).3 Methane xenon and even argon could function as token ligands though we could not estimate the magnitude of their interaction energy. A small change in the axial-radial bond angle accompanied a change in the token ligand. At this stage it wasn't possible to establish that Cr(CO) was ever free of a token ligand even in a neon matrix. In a further group of early experiments the token ligand was inter- changed by irradiating selectively into the lowest energy absorp- tion of the molecule. Such reactions were shown by polarized photochemistry and spectroscopy to occur by expulsion of one token ligand followed by a Berry pseudorotation and uptake of another token ligand (Scheme 1).* We also found evidence that photolysis of Cr(CO) initially generated Cr(CO) in a vibra- tionally or possibly electronically excited state. X * X C4" D3h C4" Scheme 1 Interconversion of X.Cr(CO) and Y .Cr(CO) viu a photo-induced expulsion of X pseudorotation of Cr(CO) and uptake of Y in a matrix (X and Y represent different species in a mixed matrix e.g. Ar and Xe).The scheme is exactly as proposed in 1978.2Notice that the D3,,intermediate is vibrationally excited (4). Nowadays Cr(CO) is a standard test molecule for photo- chemical studies alongside stilbene and [Ru(bpy),12 +. Knowl-edge has been transformed by time-resolved spectroscopy both with infrared and visible detection by photoacoustic calori- metry and other techniques employed in,fluid phases. We know that the square pyramidal structure is maintained both in solution and in the gas We know that Cr(CO) is generated within 1 ps of light absorption and that it is formed in vibrationally excited states which relax over tens if not hundreds of picosecond^.^^^ Even at very early stages of react- ion it is coordinated by a token ligand e.g. an alkane.The alkane -metal bond strength has been estimated as c'a. 35 kJ mol-in solution by photoacoustic calorimetry.* A remarkably similar estimate has been made for Xe..-Cr(CO) in liquid xenon.9 Naked Cr(CO) can now be observed in the gas phase and proves essentially identical to Cr(CO) in neon matri~es.~ Otheralkane.-.M(CO) complexesandXe.*.M(CO) (M = Cr Mo W) complexes have been observed in the gas phase; the metal-alkane and metal-xenon bond energies agree with the solution values.' O,' Curiously though the methane-W(CO) bond is appreciably weaker than the bonds to higher alkanes. The influence of this token ligand is so important that it is hard to conceive of Cr(CO) without it in solution. Nevertheless Spears et al. succeeded in photodissociating cyclohexane from C6H1 Cr(CO) and generated naked Cr(CO) albeit vibra- tionally excited for a few picoseconds in cyclohexane solution.' It is worth stressing that these alkane-Cr(CO) and xenon- Cr(CO) bonds are twice as strong as conventional hydrogen bonds and have a decisive mechanistic impact. Chromium pentacarbonyl and related molecules are probably the most powerful uncharged Lewis acids knownThe first step in react- ion is usually dissociation of the token ligand and takes place dramatically faster when it coordinates most weakly (eg in perfluoroalkanes)The idea of a simple 18-1618 electron mechanism has been superseded by something appreciably more subtle (equation 2) Although naked Cr(CO) still plays a role it certainly does not live long enough to ‘wait’ for a collision in dilute solution Instead I think of it picking up a new ligand from within its solvent cage via a pseudorotation on a picose- cond timescaleThis summary of the Cr(CO) story has concen- trated on the coordination to the token ligand for more details the reader is referred toTurner’s recent assessment of this remarkable molecule 2 Strategy and Goals Carbon-hydrogen bond activation reactions (see Scheme 2) have been intimately linked to photochemistry l4 Usually such reactions are initiated by photodissociation of H CO or C,H from a transition metal complex Alternatively they may be promoted thermally by dissociation of an alkane from an alkyl hydride or arene from an aryl hydride complexThese pre- cursors are often made photochemically themselves The C-H activation process can result in insertion into C-H bonds of alkanes arenes or alkenes provided that the intermediate has the right characteristics l5 A principal goal of the studies discussed in this lecture has been to establish the nature of the intermediates in these reactions In the previous section I demonstrated that information can be obtained both from matrix isolation and from time-resolved spectroscopy in solu- tion In the examples which follow we have used matrix isolation with detection by IR and UV/visible absorption spec- troscopy and time-resolved spectroscopy An essential adjunct to these techniques is a knowledge of the final products which is usually obtained from NMR spectroscopy ,M,/== ethene R = alkyl L = CO ethene etc Scheme 2 Principal reactions involved in C-H activation processes Although C-H activation reactions are commonly initiated photochemically it is rarely if ever possible to observe excited states of the precursorsThe reason is beginning to become clear through studies on a picosecond timescale (Section 4) In contrast there are other organometallics with long-lived excited states which exhibit spectra of sufficient quality that the problem of distortion in the excited state may be tackled Rhenocene a highly reactive molecule itself has spectral characteristics appropriate to a study of excited state distortion and is the subject of Section 5 In the final section I return to preparative studies and show how new work on C-F bond activation has been stimulated by studies of transients in C-H activation CHEMICAL SOCIETY REVIEWS 1993 3 Photodissociation of H from M(dmpe),H (M = Fe Ru) Recombination or Subsequent Reaction The first intermolecular C-H activation reaction of an arene discovered by Chatt and Davidson was brought about by reduction of Ru(dmpe),Cl (dmpe = Me,PCH,CH,PMe,) in the presence of naphthalene to form Ru(dmpe),(naphthyl)H This naphthyl hydride complex could be used to react with other arenes eliminating naphthalene the iron analogue behaved similarly It was not until much later that such reactions were investigated by irradiation of M(dmpe),H (1) Dihydride complexes have the attraction as precursors that the irradiation can be carried out at low temperature enabling labile products to be observed In this way Field et a1 demonstrated insertion of iron into C-H bonds of alkanes even including methane All of these reactions were postulated to proceed via M(dmpe) (M = Fe Ru) intermediates yet such species had never been observed All other cases of insertion into alkanes to form alkyl hydride complexes involved second and third row transition metals l4The reversal of reactivity in this case making the iron complex more reactive than the ruthenium one demonstrates that there is something very unusual about this pair of intermediates The white dihydride complex Ru(dmpe),H is readily sub- limed co-condensation with methane followed by UV photo- lysis at 12 K causes growth of three product bands in the visible absorption spectrum (Figure la) Subsequent selective photolysis and trapping experiments provide circumstantial evidence for the formation of Ru(dmpe) (Scheme 3)This complex is generated both in argon and in methane matrices in two forms one with H trapped close-by is converted back into Ru(dmpe),H on selective photolysis In the other form H has diffused too far for recombination and another reaction ensues Both forms exhibit the three-band spectrum but with slightly different maxima l6 h>400nm Ru(dmpe)2 Ar or CHA ’50’0C0’Ar\ rP P’ I hv other Ru-CO -productsP‘ I LP Scheme 3 Matrix photochemistry of Ru(dmpe),H (Reproduced with permission from C Hall et a/ J Am Chem Soc 1992 114 7425) When the same precursor Ru(dmpe),H is studied by laser flash photolysis in cyclohexane solution at room temperature an intense transient signal is observedThe spectrum of this transient proves remarkably similar to the spectrum observed in matrices (Figure 1 b)The transient absorbance decays back to the original level by second order kinetics over ca 80 ps as the TILDEN LECTURE ORGANOMETALLIC INTERMEDIATES ULTIMATE REAGENTS-R N PERUTZ 005 -040 (a) Methane matrix 12 K I 400 500 600 700 800 wavelength Vnm 004 AA 0.0. (b) Cyclohexane solution 300 K 400 ns after flash 0 0003 o*... 0. 0.. 00 V 00. 00. 0 o.o. 0 0 000 0 0 .a 0 0 O.0 0 0 0 00 0 ...0 0 0. 0.001 0 000400 500 600 700 800 wavelength Unm Figure I (a) Matrix UV/vis spectrum recorded after UV photolysis of Ru(dmpe),H in a methane matrix at 12 K Note that all the features are due to the product Ru(dmpe) (b) UV/vis spectrum recorded 400 ns after pulsed laser photolysis (308 nm) of a solution of Ru(dmpe),H in cyclohexane solution at 300 K showing formation of Ru(dmpe)The solution spectrum closely resembles the matrix spectrum but is not identical because of the changes in temperature medium and the perturbation of the matrix spectrum by expelled hydrogen (Reproduced with permission from C Hall et af J Am Chem Soc 1992 114 7425 ) precursor is regenerated (Figure 2a) As a working hypothesis we postulate that photolysis causes expulsion of H yielding Ru(dmpe) and H in equal concentrations which must then recombine by second order kinetics (equation 3) hvRu(dmpe),H eRu(dmpe) + H (3) If correct the rate of reaction should be increased by addition of HThis is indeed the case even very low concentrations of added H increase the rate of decay of the transient and turn the reaction kinetics to first order (Figure 2b) Variation of the partial pressure of dihydrogen demonstrates that the reaction of Ru(dmpe) with H is essentially diffusion controlled I e there is minimal barrier to reaction (Figure 3) By this stage the evidence of reversibility of reaction in the presence of added hydrogen precludes phosphine decoordination in the initial photochemical step the extreme reactivity of the transient excludes any 18-electron products (y2-H2 isomer cyclometalla- tion etc ) the matrix spectrum establishes that we are looking at the ground state of a reaction intermediate and not the excited state of a precursor In short the identity of Ru(dmpe) is established Addition of other substrates enables us to study kinetics of reaction of Ru(dmpe) leading to Ru(dmpe),L and Ru(dmpe),- (X)H products (Scheme 4)Although Ru(dmpe) proves to react with ethene under these conditions it is disappointing to find no reaction with benzene l6 It appears that the back reaction with H is so fast that benzene competes only very inefficiently with dihydrogen In an attempt to overcome this problem we have used Ru(dmpe),(C,H,) as a precursor for Ru(dmpe) but once more we observe no reaction with benzene l8 In further studies 004 -8 003-C e % ," 002-001 -1 1 I I I I 1 I I 1 0 40 80 120 160 Tirnelp (b) Argon + 15 torr Hp '-.. 000-.. . 'a-,..._-*...*. /.-.&A .-#-ad I I I I I I I 6 T v) 0 0 005 010 015 020 025 030 035 Partial pressure/atm Figure 3 Plot of the pseudo first-order rate constant versus partial pressure of hydrogen for the decay of Ru(dmpe) generated by laser flash photolysis of Ru(dmpe),H Note that the solubility of hydro- gen in cyclohexane is ca 10 mol dm atm Belt et a1 have shown that the quantum yield for loss of H from Ru(dmpe),H is as high as 0 85They have also used photo- acoustic calorimetry to determine Ru-H Ru-CO and Ru-N bond energies l9 Solution studies of Fe(dmpe),H by Field et a1 had led us to expect that there would be substantial differences between Fe(dmpe) and Ru(dmpe) but the nature of those differences 3 64 CHEMICAL SOCIETY REVIEWS 1993 ,Ru-L or PI cp PP Lf L = CO C2H4 Me3CNC PMe3 67 H P+X cp X = SiEtg CH*=CH Scheme 4Transient photochemistry of Ru(dmpe),H in solution O O8 r pentane solution AA 300K 1 ps after flash 000 400 500 unm600 Figure4Transient spectrum of Fe(dmpe) recorded 1 ps following laser flash photolysis (308 nm) of Fe(dmpe),H in pentane at room temperature under a hydrogen atmosphereThere are no peaks detectable at wavelengths longer than the region shown This spec- trum should be contrasted with that shown in Figure 1b (Reproduced with permission from M K Whittlesey et al J Am Chem Soc 1993,115,8627 ) was quite unanticipated Figure 4 shows the spectrum of Fe(dmpe) measured 1 ps after a laser flash there is just one band and it is in the near ultraviolet not in the visible part of the spectrum 2oThe reaction kinetics are quite different too in the absence of added dihydrogen Fe(dmpe) reacts by first order kinetics with the alkane solvent over ca 10 milliseconds Figure 5 summarizes the comparison between Ru(dmpe) and Fe(dmpe) The reaction with dihydrogen has a rate constant 7000 times smaller than the ruthenium analogue Another oxidative addition reaction that with Et,SiH is also far slower for Fe(dmpe) than for Ru(dmpe) Comparison of activation parameters for reaction with Et,SiH shows that AHf is smaller for Ru and Asf is less negative (Ru AHf = 8 9 f1 1 kJ mol- l Asf = -53 f4 J K-' mol-' Fe AHt = 22 4 f1 8 kJ mol-' Asf = -87 f6 J K-mol-') With the iron intermediate we can study reactions with members of each group of hydrocar- bons alkenes arenes and alkanesThe reaction with benzene is complicated by saturation kinetics at very high concentration but the maximum second order rate constant for the reaction is lo6 dm3 mol -s -l more than four orders of magnitude faster than the reactions with alkanes 2o Unlike the oxidative addition reactions of (y5-CSR5)Rh(PMe3) (R = H Me) with benzene,,l any y2-arene intermediates play only a slight role in reaction Previous attempts to measure activation barriers for reaction with alkanes have been thwarted by the lack of a barrier (e g for CpRhCO) 22 In contrast the reaction of Fe(dmpe) with alkanes lends itself well to detailed kinetic studyThe activation barrier turns out to have a major entropic component (AH*= 25 Of5 9 kJ mol-' Ast = -125*22 J K-l mol-') There is one pair of reactions for which Ru(dmpe) and Fe(dmpe) resemble one another the reactions with CO yield rate constants close to the diffusion limit for both of them This observation probably indicates that both intermediates are spin- paired since any spin interconversion is expected to create an entropic barrier as has been observed for Fe(CO) 23 24 Another common feature is the absence of evidence for specific solvation effects as had been found for Cr(CO) P-Ru-P P'1 Spectra 467,555,745 nm 355 nm Specific Solvation insignificant Very weak Second Order Rate Constants lo co Et3SiH cyclopentene H2 and benzene max cyclopentene Et3SiH 2 pentane cyclohexane Figure 5 Schematic comparison of Ru(dmpe) and Fe(dmpe) contrast- ing their UV/vis spectra and reactivitiesThe latter are expressed as log k where k is the second order rate constant for reaction with the substrate at room temperature deduced by laser flash photolysis The structures shown account most satisfactorily for the spectra and kinetics (Reproduced with permission from M K Whittlesey et af J Am Chem Soc 1993,115,8627 ) With enormous differences in spectra and in kinetics we infer that Fe(dmpe) and Ru(dmpe) have different molecular struc- tures The most satisfactory postulate is that Ru(dmpe) is square planar and Fe(dmpe) has a butterfly structure (Figure 5) 2o Without isotopic data to check the structures we have to look to other means of verifying these postulates Preliminary evidence from analogues with tetradentate phosphines which constrain the intermediates to a non-planar structure does lend support to the proposed structures In conclusion the behaviour of Fe(dmpe) is almost paradoxicalThe reduction in reactivity towards H when compared to Ru(dmpe) is one of the features that makes it more reactive towards hydrocarbons especially arenes and alkanes This effect can be compared to solar energy conversion only when the rate of the back reaction is low (in that case electron-hole recombination) is it possible for a productive reaction to occur Of course moderating the rate of the back reaction is not the only precondition for successful alkane activation 4 How Long Does Reductive EliminationTake? Picosecond Photochemistry of Ru (dmpe),H We already know that photodissociation of CO from Cr(CO) is complete in less than a picosecond The succeeding 10-O s are the domain of vibrational relaxation The studies described in the previous section show that the reaction of H with Ru(dmpe) is diffusion controlled That means that the exten- sion of the H-H bond the formation of two Ru-H bonds and the change in the structure of the RuP skeleton all take place without any activation barrier in a smooth process Now we can shift the focus of attention to the reverse reaction reductive elimination of H from Ru(dmpe),H Conventional tests with isotopic crossover show that photo-induced reductive elimina- tion is concerted What does that mean in terms of time-resolved spectroscopy7 In short how long does reductive elimination of H take (equation 4)3 TILDEN LECTURE ORGANOMETALLIC INTERMEDIATES ULTIMATE REAGENTS-R N PERUTZ 365 PP (4) Break No Ru-H bonds Make H-H bond Flatten RuP4skeleton The reversibility of the photodissociation of H from Ru(dmpe),H and the characteristic transient spectrum of Ru(dmpe) are features which make Ru(dmpe),H suitable for study on a picosecond timescale We undertook a preliminary study at the Rutherford Appleton Laboratory in which we irradiated a sample of Ru(dmpe),H in cyclohexane under a hydrogen atmosphere with picosecond pulses of 300 nm radia- tion We probed the sample at wavelengths close to the absorp- tion maxima which we had determined for Ru(dmpe) pre-viously Each measurement was made at a preset delay relative to the initial laser pulse Figure 6 shows a rise in absorption at each of the three wavelengths corresponding to the three maxima The rise time of the signal ca 16 ps is probably instrument determined No rise was detected when the measurements were made at wavelengths corresponding to the troughs in the Ru(dmpe) spectrum Following the rise we find only slight decay over the following 4 nsThe decay has two components the slower rate constant matches that determined previously for recombination of Ru(dmpe) with HThe faster rate constant 2 5 x lo9 s l may correspond to vibrational relaxation or geminate recombinationThus these experiments are consistent with expulsion of H and formation of Ru(dmpe) from Ru(dmpe),H within ca 16 ps of the original laser pulse The expulsion of H from Ru(dmpe),H is calculated to release excess energy of 3 15 kJ mol- A large proportion of this excess is likely to be removed as vibrational excitation of the expelled H since the photochemical act involves compression of the H-H bond We are now building apparatus designed to deter- mine full UV/vis spectra on the picosecond timescale If our conclusions are confirmed by more detailed studies it will give new meaning to the idea of a concerted reaction 26 5 Structure of Excited States. Reactive Open- She1I Metallocenes. Photochemistry is intimately connected with the structure of excited state molecules yet knowledge of the geometry of 720 nm AAA~ 540 nm... 470nm \ 50 ?.0 I 50 I 100 I 150 I 200 250 AA Timelps Risetirne 16 ps probably instrument limited No decay up to 4 ns No transient absorption in troughs Figure 6 Kinetic traces showing the loss in transmission over the time range 0-250 ps following photolysis of Ru(dmpe),H in cyclohexane under an argon atmosphere At right is shown the spectrum of Ru(dmpe) from Figure lbThe three kinetic traces are monitored point-by-point at wavelengths close to the three maxima No signal is observed when monitoring in the troughs The signals are assigned provisiondlly Ru(dmpe) indicating that it is formed within 16 ps of the initial flash excited states of polyatomic molecules is sparse in the extreme Often we are able to determine whether the symmetry of the excited state differs from that of the ground state but it is very rare even to estimate the extension or deformation of bonds quantitativelyThe experiments described above show that the excited state of Ru(dmpe),H is probably dissociative If we are to probe organometalhc excited states we require molecules with relatively long-lived excited states which do not undergo photoreactions We also need a probe of the excited state and ground state geometries Rhenocene fits these requirements We showed about ten years ago that rhenocene Re(y5-C,H,) can be generated in a nitrogen matrix by photolysis of Re(?,- C5H,),H (equation 5) hv 12K l -Re + H (5)N2 or Ar I We studied its properties by IR and UV/vis absorption by magnetic circular dichroism and by laser-induced fluores- cenceThis 17-electron metallocene proves to have a parallel ring structure with an orbitally degenerate ground state and a low-lying ligand-to-metal charge-transfer (LMCT) excited state very much like the ferrocenium ion The vibrational fine structure in the absorption spectrum also resembles that found for [Fe(y5-C,H,),]+ but the excited state of rhenocene has the additional property that it fluoresces Recently we have investi- gated rhenocene in great detail our probe is a pulsed tunable laser which excites the emission very selectively The principles of these laser-induced fluorescence (LIF) experiments are summarized in Figure 7The molecule of interest shows an absorption band with vibrational fine struc- ture If the transition is fully allowed as for rhenocene the fine structure corresponds to totally symmetric vibrational modes v' of the electronic excited state For a metallocene there are only four such modes and only one of them involves the metal- ring bonds directly vzz v4 the ring-metal-ring symmetric stretching mode When the sample is irradiated into this band it emits from the same electronic excited state In the case of rhenocene the electronic excited state relaxes down to its lowest vibrational level before emitting so the emission spectrum is very close to the mirror image of the absorption spectrumThis time the vibrational separation corresponds to the electronic ground state V" If the intensity of a certain emission band is probed as a function of the excitation frequency we obtain an c-urptionV Emission spectrum emission MeasureI as f(ve,) at fixed Vex Excitation spectrum excitation MeasureI as f(vex) at fixed v Lifetime MeasureI as f(delay time) after laser pulse t t Figure 7 Schematic diagram illustrating the principles of laser-induced fluorescence The top spectrum represents an absorption exhibiting a vibrational progression V' (upper state frequency) The middle spec- trum shows the corresponding fluorescence emission spectrum (vibra tional frequency V" for the lower state)The bottom spectrum shows the corresponding fluorescence excitation spectrum which is recorded point-by-point in the experiments described and closely resembles the absorption spectrum in simple cases excitation spectrum In dilute solution the exytation spectrum should resemble the absorption spectrum but in the matrix it has great advantages in selectivity as explained below The emission and excitation spectra carry two sets of information From the changes in vibrational frequencies it should be possible to deduce the changes in the force constants concerned with the totally symmetric modes From the pattern of intensit- ies of the vibrational components the Franck-Condon profile it should be possible to deduce the changes in the corresponding geometric parameters When compared to solution spectra matrix spectra have the advantage that they are much sharper but the snag that there are often more bands than expected because of matrix effects There are two causes of such effects which concern us the presence of the guest molecule in different but specific trapping sites and the presence of several different conformers of the guest (Figure 8) As a result the absorption spectrum consists of the sum of the spectra of the guest in each conformer and each site However by judicious choice of laser wavelength the emission and excitation spectra can be restricted to one site or conformer (or a group of almost identical sites) Such selective excitation can simplify the spectrum dramatically TraDDlna sites Conformers @@@ Matrix sites and conformers are reproducible and specific Figure 8 Schematic diagram illustrating the origins of matrix splittings A metallocene is shown above surrounded by 5,6 or 8 matrix atoms Below it is trapped in three different conformations DSd,D5h,and D The matrix absorption spectra represent the sum of the spectra in all the different trapping sites and all the diflerent conformations The lowest energy absorption band of rhenocene isolated in a nitrogen matrix is shown in Figure 9a 29 30 It shows a vibration- al progression of frequency about 300 cm -corresponding to the ring-metal-ring symmetric stretching frequency v but the progression is poorly resolved and each component appears to show further ill-defined structure In contrast the emission spectrum (Figure 9b) obtained by laser excitation at the position marked by an arrow is much sharper Since it is resolved to the baseline the value of the ground-state frequency (vi = 325 cm ') can be measured preciselyThe spectrum illustrated also shows the first component of the progression in the symmetric C-H deformation mode (v = 826 cm ') When the excitation wavelength is changed systematic changes occur in the emission spectrum corresponding to the excitation of different conformer/sites of the matrix-isolated rhenocene If the intensity of the strongest emission maximurn (double arrow) of Figure 9b is measured as a function of excitation wavelength the excitation spectrum (Figure 9c) is obtainedThis excitation spectrum corresponds to a single site or conformer of rhenocene Thus the complex absorption spectrum has been deconvoluted experimentally As in the idealized spectrum of Figure 7 the excitation and emission spectra are close to mirror images of one another with a common (0,O) bandThe excitation spectrum shows that v4 is increased by 17cm in the excited state while u2 and v3 (the C-H deformation at ca 750 cm and the ring breathing mode at ca 1100 cm-l) decreaseThese spectra establish unequivocally that it is possible to obtain high resolu- tion electronic spectra of organometallics They may be com- CHEMICAL SOCIETY REVIEWS 1993 '0x1041 1 (b) Emission I Re in N2 Matrix Emissioncounts Laser energy Emissioncounts 3 470 490 Unm 500 Figure 9 Spectra of rhenocene Re(TS-C,Hs) isolated in a nitrogen matrix at 12K (a)partial absorption spectrum (b)emission spectrum excited at 473 5 nm (double arrow) (c) excitation spectrum showing the variation in intensity of the most intense emission maximum of Figure 9b with excitation wavelength Notice that the emission and excitation spectra are much sharper than the absorption spectrum because they are sitejconformer selective (Reproduced with permission from J N Hill et a1 Coord Chew Rev 1991,111 111 ) pared to the single crystal spectra of forbidden bands in rutheno- cene3132 and to the recent spectra of Z~(T~-C,H,)measured in supersonic jets 33 The electronic transition giving rise to the rhenocene spec- trum has been assigned as a LMCT transition from the el levels of the C,H rings (C-C bonding but Re-C non-bonding) to the e2glevels of the metal (principally d y2 -,l but Re-C bond-ing) *The increase of ca 5% in the ring-metal-ring stretching frequency and the decrease in the ring breathing frequency of ca 3% fit this assignment beautifully We now turn to the problem of extracting geometric infor- mation Zink has developed programs based on the Heller wavepacket method for simulating vibrational fine structure in electronic spectra 34 In this method the evolution of a wave- packet on an excited state is followed in the time domain via the time-dependent Schrodinger equationThe frequency domain corresponds to the Fourier-transform of the evolution in the time-domainThis method lends itself to the simulation of electronic spectra containing progressions in more than one frequency It can also take into account the changes in frequency between ground and excited states Figure 10 shows the observed and calculated excitation and emission spectra for rhenocene The spectra are calculated with one distortion para- meter each for v2,v3 and v4 the same parameters are used for the excitation as for the emission spectra The fit of the calculated emission spectrum is close to perfect that for the excitation spectra reproduces the first few features well but is less good at the high frequency extreme If we assume that the normal coordinate corresponding to u4 is identical to the symmetry coordinate we can deduce the change of the metal-ring-centroid distance in the excited state (Table 1) Since the simulation method involves a square root these calculations do not dis- tinguish between a bond compression and an extension How-ever in combination with the frequency data we deduce that the metal to ring-centroid distance is compressed by 0 045 8 in the LMCT excited state of rhenocene For comparisonTable I also gives values from measurements on tungstenocene which has similar emission properties to rhenocene In this molecule the same vibrational mode v4. acts as the principal carrier of the progression yet there is negligible change in its frequencyThus the frequency and the geometry do not change together The simulations of the spectra show that * In terms of spin orbit states this transition is described ds E 29 to E in the D ,double group TILDEN LECTURE ORGANOMETALLIC INTERMEDIATES ULTIMATE REAGENTS-R N PERUTZ 1 e Re in N2 Matrix 21 8 21 4 21 206 202 198 194 19 V/I o3cm-' Figure 10 Observed and calculated excitation and emission spectra for Re(775-CsH,)z in a nitrogen matrix The calculated spectra are derived by the Heller method Table 1 Calculated change in metal to ring-centroid distance and ring-metal-ring symmetric stretching frequency (v4) of metallocenes shown as (Excited state -Ground state) (7 '-cs H s z Re (77 '-c5 H 5 z (77 '-c5 H5 z Ru Transition LMCT" LMCT" Ligand Fieldh (v& -v;)/cm + 17 -4 -52 6r(Cp-M)/A f0 045 f0 049 20 12 J N Hill D PhilThesis University of York 1993 h References 31 32 the ring-to-metal bond is compressed by a comparable distance to that in rhenocene Similar calculations have been performed on the ligand-field (LF) excited state of ruthenocene observed by emission from crystals at low temperature As expected LF excitation populates metal-ring antibonding orbitals so there is a very large extension of the metal-ring distance and reduction in v4 In the next stage of this work we will determine the force field of rhenocene in order to obtain an accurate transformation of the change in normal coordinates to a change in geometry This section has been dominated by investigations of rheno- cene an unstable molecule known only in the matrix environ- ment or postulated as a reaction intermediateThe permethy- lated analogue decamethylrhenocene is a volatile crystalline compound which is made by solution photolysis of Re(q5- C,Me,),H (an early extension of the matrix studies by Cloke et al ) The electronic properties of decamethylrhenocene in solu- tion and in matrices correspond exactly to those for matrix- isolated rhenocene It also exhibits vibrationally resolved LIF spectra which are even sharper than those of rhenocene 29 30 35 6 Coordination and Activation of HexafIuorobenzene Studies of reaction intermediates should be intimately linked to studies of the associated reactions by conventional techniques of preparative chemistry particularly by NMR and X-ray crystal- lographyThe value of such a link for understanding the transient photochemistry has been amply illustrated above There is an equal benefit in the other direction studies of transient chemistry stimulate new preparative chemistry It has long been postulated that activation of arene C-H bonds is preceded by coordination of the arene though two carbon atoms $-coordination When investigating the photoreaction of Rh(~5-C,H,)(PMe3)(C2H4) with benzene we found evidence for just such an intermediate Rh(qs-Cs H s)(PMe3)(~2-C6H6) and showed that it had a lifetime of CQ 1 ms at room temperature (Scheme 5) 36 For comparison we required an inert solvent which was incapable of C-H activation of any sort Simon Belt hit on the idea of using hexafluorobenzene which had recently been described as an inert solventThe transient UV/vis trace now showed a rise but no decay hexafluorobenzene was reactive Further investigation by NMR and X-ray crystallogra- phy showed that this reaction provided a means of arresting the oxidative addition at the stage of T2-coordination the product was the air-stable stereochemically rigid complex Rh(v5- CsHs)(PMe3)(q2-C6F6) This discovery signalled the start of hexafluorobenzene chemistry for us and the start of controlling arene activation 36 We have now characterized several com- plexes of hexafluorobenzene involving coordination to rho- dium iridium or rhenium we have obtained crystal structures of complexes with $-and coordination and obtained NMR evidence for q6-coordination (Scheme 6a+) When q2-coordi- nated two of the C-F bonds are distorted out of the plane of the remaining C6F4 unit by ca 43" and the carbon skeleton is distorted to generate a coordinated double bond and an unco- ordinated diene 36-38 When ~~-coordinated the hexafluoro- benzene is folded like a book this time there is a coordinated diene unit and one uncoordinated short C=C double bond 37 I Oh M~,P/' ""/ k millisec hv Flash ' -Me3P/"\ e H Ph dh ~H'=480+18kJmol' Me3P/ \ H AS*=288f67 J K' mollH Scheme 5Transient photochemistry of R~(T~-C,H,)complexes in benzene showing v2-C6H6 intermediate detected prior to C-H insertion In the initial example of hexafluorobenzene coordination we did arrest the oxidative addition process but in other examples we have observed C-F activation sometimes accompanied by HF elimination (Scheme 6d-f) 38 In one case we even turned the tables entirely when Re(q5-C,MeS)(CO) is irradiated in benzene the reaction stops at q*-coordination but in hexafluoro- benzene we observe C-F activation accompanied by intramo- lecular C-H activation and elimination of HF (Scheme 6f)139 Returning to the initial reactions of the Rh(~5-C5H,)(PMe3) fragment with arenes the reaction with C6F6 provided the first lesson for us in how to control the reaction with arenes We can now steer this reaction to obtain entirely C-H oxidative addi- tion products (eg with benzene or with C6F,H) entirely 7'-coordination (eg with C6F6 or with naphthalene) or an I I Ir FF *H2 I/ I I dh ;&R[ co Fo X ray (d) C F insertion (e) C F insertion (f) C F and C kiinsertion HF elimination HF elimination Scheme 6 Coordination and C-F activation products from reactions with hexafluorobenzene equilibrium mixture between C-H insertion and $-coordina- tion (eg with 1,3-C6H4(CF3),) 7 Challenges In this lecture I have tried at several stages to indicate the challenges which lie ahead Rather than summarize the results I now wish to draw together some of the unsolved problems (I) How fast is photo-induced reductive elimination? We know that photodissociation of CO from a metal carbonyl can be complete in < 1ps Reductive elimination involves bond break- ing and forming combined with molecular rearrangement yet preliminary indications from our work on Ru(dmpe),H sug-gest that reductive elimination may be almost as rapid Such studies will enable us to reframe our notions of reaction mech- anisms especially of concerted processes in terms of a time sequence (11) Measurement of barriers to oxidative addition of the solvent Many C-H activation reactions involve reaction with the solvent Such processes do not require diffusion in the conven- tional sense so their second order rate constants can exceed lo9 or loio dm3 rno1-l s-’ We have been able to measure the barrier to oxidative addition of alkane solvent to Fe(dm~e),,~O but in other cases oxidative addition of the solvent occurs within the rise time of nanosecond apparatus [eg reaction of (77,-C,H,)RhCO with cyclohexane] 22The obvious alternative of using an inert solvent and adding alkane is complicated by the lack of an inert solvent For instance Xe coordinates so strongly to CpRhCO that liquid xenon is unsuitable for such study 40 41 Liquid krypton is better but even then coordination of krypton competes with coordination of alkane For these reasons it will be necessary to use picosecond spectroscopy to measure the barriers to oxidative addition of the solvent When the molecules contain carbonyl groups the methods of ultrafast infrared spectroscopy hold particular promise l2 42 (111) Structure and stabilization of alkane complexes In the first section I indicated the importance of specific coordination of alkanes to Cr(CO)There is evidence now from time-resolved spectro~copy,~ l2 from matrix isolation 43 from isotopic 6-substit~tion,~~-~~from kinetic isotope effects,44 and from theory4*-49 that such complexes are very important as reaction intermediates As yet it has not proved possible to determine any of their structures we have had to rely on the analogies of agostic comple~es,~~ ~2-silanecomplexes,Sor metal tetrahydro- boratesThe key to structure determination is stabilization the (OC),Cr-alkane bond has an energy of about 35-40 kJ mol-l If we are to obtain NMR or crystallographic infor- mation we probably need to learn how to increase the bond energy substantially or to hold the alkane in place within a larger structure (iv) Structure of the excitedstate In Section 5 I described our attempts to determine the differences between the structures of rhenocene in the ground and the LMCT excited state by analysis of emission and excitation spectra As yet the extent of infor- mation on excited state geometries of transition metal com- plexes is tiny and on organometallics almost zero In addition to our own approach two others are being employed Resonance Raman excitation profiles offer similar information to fluores- cence excitation spectra but are usually less well resolved 52 Time-resolved infrared spectroscopy is also providing infor- mation about metal carbonyls with long-lived charge transfer excited states In that case a change in force constant is readily accessible but it is not trivial to convert it to a change in bond lengthThe most complete description of the excited state would be obtained from an X-ray crystallographic difference map measured during excitation of a single crystal -such measure- ments should be accessible very soon 54 (v) Structure-sensitive methods for transients without carhonyl groups zn fluid phasesThe methods of time-resolved infrared spectroscopy have proved particularly effective for the determi- nation of the structure of transient intermediates in the reactions of metal carbonyls However as is evident from the work on CHEMICAL SOCIETY REVIEWS 1993 M(dmpe) complexes we still lack methods suitable for other organometallics and coordination complexes In that case we have to resort to devising analogous complexes which have constrained geometries and comparing their UV/vis spectra Sometimes information may be obtained by time-resolved resonance Raman spectroscopy as for the excited state of [Ru(bpy),12 + 56 Another alternative is suggested by the laser- induced fluorescence spectra of rhenocene It should be possible to vaporize the precursor molecules and entrain them in a supersonic jet Following photolysis the product metallocenes could be detected by LIF yielding spectra which should be similar to the matrix spectra Indeed jet cooling has been employed to obtain LIF spectra of Zn(q5-C,H,) and related molecules 33 (vi) Applications of C-H ActivationThe reactions of alkanes with transition metal complexes have disproved the usual ideas of lack of reactivity of alkanes even methane will undergo productive insertion reactions at extremely low temperature Despite much effort however progress in putting these remark- able reactions to use has been slowThe feasibility of dehydroge- nating or carbonylating alkanes has been demonstrated but these reactions require far more development before they can fulfil their industrial promise l4 The possibilities for the future are summed up by a letter from Professor Joseph Chatt commenting on the detection of Ru(dmpe)2 ‘We tried to get argon to behave as a ligand on N,-binding centres at room temperature but nothing came of it Of course the single atom cannot polarise as can the nitrogen molecule but in this field keep an open mind about it ’ AcknowledgementsThe work described in this article represents the work of members of the group in York who have contributed by their experimental work their ideas and their enthusiasm over several years Important contributions have been made by T W Bell ST Belt C Hall J N Hill A McCamley R Osman M G Partridge A D Rooney and M K Whittlesey W D Jones and A H Klahn spent their leave in York and had a major impact on the direction of the research In addition I have benefitted enormously from the collaboration of S E J Bell and R J Mawby in York F G N Cloke (Sussex) L D Field (Sydney) M Helliwell (Manchester) and J Zink (UCLA) as well as the staff of the Rutherford Appleton Laboratory A Langley A W Parker and STavenderThe work has been supported extremely generously by The Royal Society British Gas plc and SERC Additional funds have been provided by the European Commission NATO and BP Chemicals Ltd 8 References 1 P Grebenik R Grinter and R N Perutz Chem SOCRev ,1988,17 453 2 J K Burdett J M Gryzbowski R N Perutz M Poliakoff J J Turner and R F Turner Inorg Chem ,1978,17,147 and references therein 3 G R Dobson P M Hodges M A Healy M Poliakoff J J Turner S Firth and K J Asali J Am Chem Soc 1987,109,4218 4 S P Church F -W Grevels H Herrmann and K Schaffner,Znorg Chem ,1985,24,418 5 T A Seder S P Church A J Ouderkirk and E Weitz J Am Chem Soc 1985 107 1432 6 A G Joly and K A Nelson Chem Phys 1991 152,69 7 S C Yu X Xu R Lingle and J B Hopkins J Am Chem Soc 1990 112 3668 8 J M Morse G H Parker andT J Burkey Organometalhcs 1989 8,2471 9 B H Weiller J Am Chem SOC,1992,114 10910 10 C E Brown,Y Ishakawa,P A Hackett,andD M Rayner,J Am Chem Soc ,1990,112,2530 I1 J R Wells and E Weitz J Am Chem SOC,1992 114,2783 12 J R Sprague S M Arrivo andK G Spears J Phys Chem ,1991 95 10528 13 J J Turner in ‘Photoprocesses in Transition Metal Complexes TILDEN LECTURE ORGANOMETALLIC INTERMEDIATES ULTIMATE REAGENTS-R N PERUTZ Biosystems and Other Molecules Experiment and Theory’ ed E Kochanski Kluwer Academic Dordrecht 1992 p 125 14 R N Perutz Chem Ind 1991 462 ‘Activation and Functionalisa- tion of Alkanes’ ed C L Hill Wiley New York 1989 15 R N Perutz Pure Appl Chem 1990,62 1103 16 C Hall W D Jones R J Mawby R Osman R N Perutz and M K Whittlesey J Am Chem SOC,1992,114,7425 and references therein 17 L D Field A V George and B A Messerle J Chem SOC Chem Commun 1991 1339 18 R Osman and R N Perutz to be published 19 ST Belt J C Scaiano and M K Whittlesey J Am Chem Soc 1993 115 1921 20 M K Whittlesey R J Mawby R Osman R N Perutz L D Field M P Wilkinson and M W George J Am Chem Soc ,1993 115 21 S T Belt L Dong S B Duckett W D Jones M G Partridge and R N Perutz J Chem SOC Chem Commun 1991,266 22 S T Belt F -W Grevels W E Klotzbucher A McCamley and R N Perutz J Am Chem SOC,1989 111,8373 23 E Weitz J Phys Chem ,1987,91 3945 24 R Ryther and E Weitz J Phys Chem 1991,95,9841 25 R Osman R N Perutz and A D Rooney Ann Rep Central Laser Facrlrtb Rutherford Appleton Laboratory 1992 155 26 F Bernard] M Olivucci M A Robb and GTonachini J Am Chem Soc 1992,114 5805 27 P A Cox,P Grebenik,R N Perutz,R G Graham,andR Grinter Chem Phys Lett 1984 108,415 28 K D Warren Struct Bonding (Berlin) 1976,27,45 29 S E J Bell J N Hill A McCamley and R N Perutz J Phys Chem 1990,94,3876 30 J N Hill A McCamley and R N Perutz Coord Chem Rev ,1991 111,111 31 G J Hollingsworth K S Kim Shin and J I Zink Inorg Chem 1990,29,2501 32 H Riesen E Krausz W Luginbuhl M Biner H U Gudel and A Ludi J Chem Phys 1992,96,4131 33 E S J Robles A M Ellis andT A Miller J Phys Chem 1992 96,3247 34 J I Zink and K S Kim Shin Adv Photochem 1991 16 119 35 J A Bandy F G N Cloke G Cooper J P Day R B Girling R G Graham J C Green R Grinter and R N Perutz J Am Chem SOC.1988,110 5039 36 S T Belt S B Duckett M Helliwell and R N Perutz J Chern Soc Chem Commun 1989 1372 37 T W Bell M Helliwell M G Partridge and R N Perutz Organometallrcs 1992 11 19 1 1 38 S T Belt M Helliwell W D Jones M G Partridge and R N Perutz J Am Chem SOC 1993,115 1429 39 A H Klahn,M H Moore,andR N Perutz J Chem Soc Chem Commun 1992 1699 40 B H Weiller E P Wasserman R G Bergman C B Moore and G C Pimentel J Am Chem Soc 1989,111,8288 41 B H Weiller E P Wasserman C B Moore and R G Bergman J Am Chem Soc 1993,115,4326 42 J N Moore P A Hansen and R M Hochstrasser J Am Chem SOC,1989,111,4563 43 Z H Kafafi R H Hauge and J L Margrave J Am Chem Soc 1985,107,6134 44 R A Periana and R G Bergman J Am Chem Soc 1986 108 7332 45 R M Bullock C E L Headford K M Hennessey S E Kegley and J R Norton J Am Chem SOC,1989,111,3897 46 G Parkin and J E Bercaw Organometallics 1989,81 1172 47 G L Gould and D M Heinekey J Am Chem Soc 1989 111 5502 48 N Koga and K Morokuma J Phys Chem 1990,94,5454 J Am Chem Soc 1993,115,6883 49 N Re M Rosi A Sgamelloti C Floriani and M F Guest J Chem Soc DaltonTrans 1992 1821 T Ziegler E Folgd and A Berces J Am Chem Soc 1993 115 636 T R Cundari Organometallics 1993 12 1988 J Song and M B Hall Organo-metallrcs 1993 12 31 18 50 M Brookhart M L H Green and L L Wong Prog lnorg Chem 1988,36 1 51 U Schubert Adv Organomet Chem 1990,30 151 52 K S Kim Shin R J H Clark and J I Zink J Am Chem Soc 1990 112 3754 53 P Glyn,F P A Johnson,M W George,A J Lees,andJ J Turner Inorg Chem 1991,30 3453 54 K Moffat Y Chen K Ng D McRee and E D Getzoff Phzloy Trans R Soc London Ser A 1992,340 175 55 M Poliakoff and E Weitz Adv Organomet Chem 1986,25,227 56 D E Morris and W H Woodruff Adv Spectrosc 1987 14 285

ISSN:0306-0012    

DOI:10.1039/CS9932200361

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Determination of Molecular Conformation from Large Amplitude Vibrations in Electronic Spectra of Organic Molecules in a Supersonic Jet J. Michael Hollas Department of Chemistry University of Reading Whiteknights Reading RG6 2AD Berkshire U.K. 1 Introduction The more dramatic examples of the double helical conformation of DNA and the science of drug design demonstrate the great importance of molecular conformation in chemistry and bio- chemistry. At the other end of the scale of molecular size the discovery that the equilibrium conformation of ethane is eclipsed rather than staggered and a theoretical understanding of the nature of the energy barrier between equivalent staggered conformations stimulated comparable investigations of similar molecules such as methyl alcohol and acetaldehyde. There tends to be some confusion regarding the use of the words 'conformation' and 'structure' in the context of polyato- mic molecules. Structure usually refers to quantitative bond lengths and angles as might be obtained in a microwave spectroscopic or electron diffraction experiment. On the other hand conformation is usually concerned with more qualitative properties such as planarity or non-planarity eclipsed or stag- gered. However what we may regard as conformational studies are also concerned with energies for example how much energy is required to go from the staggered to the eclipsed form of ethane or from cis-to trans-l,2-difluoroethylene?They may be concerned also with structural parameters such as the distance out-of-plane of the hydrogen atoms of the NH,-group in aniline. So the confusion remains to some extent and even the Oxford English Dictionary gives one of the meanings of conformation as 'structure'! In this Review though I shall be concerned more with molecular conformation as we understand it generally than with details of bond lengths and angles. Carbon suboxide C,O is an example of a polyatomic molecule which might be linear rather than bent about the central carbon atom. One method of finding out is to identify as many as possible of the vibrational levels of the corresponding bending vibration.These can be fitted to an appropriate poten- tial function and if the potential shows a minimum at a bending angle of 180" it is linear. If it is a W-shaped potential with a symmetrical double minimum the molecule is bent. In the case of C,O it is linear but very floppy and easily bent. However this ~ Dr. Hollas studied for his BSc. degree at University College London M-here he went on to a Ph.D. under the supervision of Professor G. W. King and then Professor D. P. Craig. In 1959 he tt-as arzurded a Postdoctoral Fellowship at the National Research Council in OttaMva in the Division of Pure Physics headedby Dr. G. Herzberg. From 1961-64 he M'US a Senior Research Fell0 M in the Basic Physics Division of the National Physical Labora- tory (Teddington) whose Su-perintendant was Professor J. A. Pople. In 1964 he was ap-pointed as a Lecturer at the University of Reading u.here he is now a Reader. His interests are in applying the techniques of high resolution electronic spectroscopy to tz?hat are bj usual standards large organic molecules. 371 linear conformation cannot be referred to strictly as the equilibrium conformation. It is the equilibrium conformation so far as the bending vibration is concerned but the zero-point conformation so far as all other vibrations are concerned. We must also be careful to appreciate that this bending potential will be affected when other vibrations in the molecule are excited. This is what is generally referred to as anharmonicity but it is of a rather special kind. In non-linear polyatomic molecules two particularly import- ant aspects of conformation are (i) whether a molecule is planar or has a planar skeleton as in the examples of aniline (1) and 9,lO-dihydrophenanthrene (2) (ii) whether a substituent attached to an aromatic ring is coplanar eclipsed or staggered with respect to the ring as in the examples of styrene (3). toluene (4),and 3-aminobenzotrifluoride (5). 2 Experimental andTheoretical Methods We are all familiar with the general shape of the potential function for a bound diatomic molecule the Morse potential provides a reasonable approximation.To determine this poten- tial from experimental data requires as many vibrational energy levels as possible to be known -preferably including levels close to the dissociation limit. If the electronic state concerned is the ground state these levels are usually obtained by emission from an excited electronic state.This excited state should have an appreciably different equilibrium bond length so that the Franck-Condon principle results in appreciable intensity of transitions from vibrational levels of the excited state to high- lying levels of the ground state.To determine vibrational levels in an excited electronic state absorption from the ground statc may be used and the Frank-Condon principle again relied upon to allow access to high-lying vibrational levels. In polyatomic molecules the same general methods apply but particularly in larger molecules the determination of vibratio- nal potentials is usually confined to the ground singlet state So and the first excited singlet state S for closed shell molecules. For molecules in which spin-orbit coupling is appreciable investigation of vibrational potentials in the first excited triplet stateT may also be possible. Large polyatomic molecules have many vibrational degrees of freedom 3N -6 where N is the number of atoms and small rotational constants.This results in crowded low-lying rotatio- nal and vibrational energy levels which are relatively highly populated at room temperature and a gas phase absorption spectrum which is increasingly likely to be so congested as to appear almost featureless. For example although rotational fine structure can be observed in the S -Soabsorption spectrum of (l) (3),and (4) only broad vibrational structure is shown by (2) and (5). An extremely successful way of getting round the problem of spectral congestion is to seed the molecules into a supersonic jet. If a noble gas which is often helium at a pressure of the order of 1 atm is pumped through a small pinhole or slit with a diameter or width of the order of 100 pm into a low pressure chamber the atoms attain an extremely low translational tem- perature.This may be around 1 K or less if the backing pressure is higher. When the molecules of interest are seeded into this so-called supersonic jet of helium they are drastically cooled. Rotational cooling is more efficient than vibrational cooling because of the relative closeness of the energy levels. Typical 3 72 CHEMICAL SOCIETY REVIEWS 1993 (23) F% 0 (311 rotational and vibrational temperatures achieved are 10 K and 100 K respectively but these may be lower with a higher pressure of helium. If argon is used instead of helium there is a tendency because of the greater polarizability of argon for van der Waals complexes to be formed.These are of great interest in their own right but will not be considered in this Review. A supersonic jet may flow continuously or be pulsed. A pulsed (3) F F F@@FF F F F 'y".qH2FF (19) jet is particularly appropriate if a pulsed laser is being used to obtain spectra and it also conserves material. The outer parts of the jet may be skimmed away if particularly narrow ranges of rotational temperatures and molecule velocities are required. The rotational and vibrational cooling removes much of the congestion in the electronic spectra of molecules seeded into a supersonic jet. As a result S,-So absorption and fluorescence spectra are mostly very sharp but this is not always the case. Rotational and vibrational levels may be broadened due to efficient competing processes which occur on a time scale which is short compared to the fluoroscence lifetime of the S state. Examples of such processes are vibrational predissociation and intramolecular vibrational relaxation. The absorption and fluorescence processes are investigated by laser irradiation the laser beam intersecting the jet about 1 cm downstream from the pinhole or slit. Direct absorption has been MOLECULAR CONFORMATION OF ORGANIC MOLECULES IN A SUPERSONIC JET-J M HOLLAS used in supersonic jets but it is more usual to use an indirect method of monitoring absorptionThe two most common methods fluorescence excitation (FE) and resonant multipho- ton ionization (REMPI) are illustrated in Figure 1 V V FE REMPI SVLF (or DF) Figure 1 Fluorescence excitation (FE) 2 + 2 resonant multiphoton ionization (REMPI) and single vibronic level fluorescence (SVLF),or dispersed fluorescence (DF) processes To obtain a FE spectrum the laser wavelength is tuned across the S,-S absorption system and the intensity of total undis- persed fluorescence monitoredThe FE spectrum corresponds exactly to an absorption spectrum if the quantum yield of fluorescence is constant for all vibronic levels of S More often the quantum yield declines with increasing vibrational exci- tation so that fluorescence intensities decrease compared to absorption intensities The possibility of obtaining REMPI spectra depends on the high radiation density provided by a laser source Because of this high density a molecule may absorb more than one photon Such a multiphoton process may take several forms and all depend on an intermediate step prior to ionization in which the molecule is excited to the S state this is the reason why it is called a resonant process Figure 1 illustrates the case where the molecule absorbs two photons to take it to SThe first photon takes it to a virtual state V,just as in a Raman scattering process and the second takes it to STwo more photons are absorbed to ionize the molecule and the complete process is sometimes referred to as a 2 + 2 multiphoton ionization process Alternatively a 1 + 1 process may be used in which one photon takes the molecule to S and a second ionizes it When the S,-S system is in the near ultraviolet the single photon is a visible photon in the 2 + 2 process but an ultraviolet photon in the I + 1 process In either process the laser rddiation is tuned so that one or two photons scan the S,-S system while the total number of ions is collected and counted These ions may be passed into a time-of-flight mass spectrometer to ascertain that the REMPI spectrum corres- ponds to the correct molecular species Both FE and REMPI spectra provide information on vibra- tional levels in S To obtain vibrational levels in So the technique of single vibronic level fluorescence (SVLF) alternatively called dis- persed fluorescence (DF) is usedThis is illustrated in Figure 1 The laser wavelength is tuned to match a particular absorption band to take the molecule into a single vibronic level This may be the zero-point level or an excited vibrational level within S Apart from the very low vibrational and rotational tempera- tures produced in a supersonic jet there is a further great advantage in that the molecules are in collision-free conditions As a result molecules excited to a single vibronic level are not transferred to other vibronic levels due to collisions occurring during the fluorescence lifetime In a SVLF spectrum the fluorescence from the single vibronic level of S that has been populated is dispersed to give vibratio- nal levels in So The observation of progressions in SVLF FE or REMPI spectra allowing access to higher vibrational levels of Soor S is highly desirable and depends on there being an appreciable change of curvature of the potential energy curve or of the equilibrium conformation of the molecule between So and S Figure 2 illustrates a typical experimental arrangement for obtaining FE and SVLF spectraThe pulsed tunable dye laser is pumped by a Nd YAG or excimer laser but a high resolution CW ring dye laser may also be used In the figure the supersonic jet is directed vertically downwards into the vacuum pumping systemThe dye laser radiation may have to be frequency doubled when working in the near ultraviolet region but the beam of visible or ultraviolet radiation intersects the supersonic jet in the centre of the vacuum chamber Fluorescence is collected in a direction perpendicular to the laser and molecular beams In the arrangement in Figure 2 a set of four concave mirrors is used to collect the fluorescence more efficiently but other optical arrangements may be used The collected fluores- cence is directed into a spectrometer with photomultiplier detection For a FE spectrum the laser is tuned through the S,-So absorption system and with the angle of the grating in the spectrometer set to zero the total undispersed fluorescence is collected For REMPI spectra ion collection plates are used or the ions are directed into a time-of-flight mass spectrometer For a SVLF spectrum the laser wavelength is fixed on a particular absorption band and the grating of the spectrometer rotdted to scan through the region of the dispersed fluorescence Calib-ration of SVLF spectra may be achieved with an iron hollow cathode source Fe-Ne1lampVacuum + ,I\ Trigger delay Figure2Typical experimental arrangement for obtaining FE and SVLF spectra of molecules in a supersonic jet When the exciting laser radiation is pulsed the photomultip- lier detects pulsed fluorescence the length of the pulses being determined by the lifetime of the fluorescenceThese pulses are integrated by the boxcar and the signal output to d chdrt recorder or microcomputer Other techniques for observing vibrational levels by laser excitation such as fluorescence depletion and stimulated emis- sion pumping (SEP) are used but will not be discussed here There are two general types of vibration that we shall be concerned with One of these IS inversion as in the well-known inversion vibration of ammonia Examples are the out-of-plane bending vibration of the hydrogen atoms of the amino group of aniline (1) and the out-of-plane bending of the two benzene rings about the central carbon and oxygen atoms joining them in 374 xanthene (6) The second type is a torsional vibration examples of which are the torsional motion of the vinyl group relative to the benzene ring in (3) and of the CF group relative to the benzene ring in (5) For an inversion vibration also called a ring-puckering vibration in the case of a closed ring system as in (6) the form of the potential energy function which is adopted must be able to show a symmetrical double minimum as it does2 in the Sostate of (1)There have been several suggested forms for this potential V(Q),where Q is the inversion vibrational coordinate but V(Q)= 3aQ’ + b exp( -cQ’) (1) is one of the more successful The first term is that of a harmonic oscillator and the second term introduces an energy barrier of height b,at Q = 0 between the two equivalent minima of the W-shaped potential A useful alternative W-shaped potential to that in equation 1 1s If A is zero the potential is a steep-sided quartic but the addition of the AQ2term with negative A,introduces an energy barrier at Q = 0 of height b = A2/4B (3) For a molecule which may be non-planar or have in it a non- planar group the determination of the barrier height or show- ing that it is zero is vital in determining molecular conformation For torsional motion the most useful general form of the potential is (4) where 4is the torsional angle andTI is a non-zero integer Which terms are dominant in the summation depends on the molecule concerned For example in the planar (3) it is V and in (4) it is v6 In these cases the barrier height between equivalent confor- mations is V and V6,respectively 3 Examples 3.1 Inversion Vibrations When conformational investigations using spectroscopic tech- niques were confined mostly to small molecules the use of inversion vibrations in structure determination was applied to molecules such as ammonia (NH,) and formaldehyde (H,CO) Ammonia is pyramidal in its ground state but planar in several of the lower energy excited states while formaldehyde is planar in its ground state but pyramidal in its S excited state These examples illustrate a possible source of ambiguity regarding what we call an inversion vibration Strictly an inversion vibration is one which can take a non-planar molecule through an energy barrier to an equivalent configuration related to the initial one by a reflection through the plane of the planar high energy configuration If the energy barrier in the resulting W-shaped vibrational potential function is reduced to zero the molecule is planar and the inversion vibration becomes an out-of-plane bending vibration For example the inversion vibration of pyramidal ammonia in So and of non-planar formaldehyde in S become out-of-plane bending vibrations in planar excited states of ammonia and in the So state of formaldehyde Now that larger molecules can be investigated by the tech- niques described in Section 2 we encounter a wider variety of inversion vibrations although a similar example to those of ammonia and formaldehyde is the inversion or out-of-plane bending vibration of the amino hydrogens in aniline (1) which CHEMICAL SOCIETY REVIEWS. 1993 can be regarded as a phenyl derivative of ammoniaThe molecule is non-planar in So,with a barrier to planarity of 547 cm (6 54 kJ mol- ’) and planar in S1with an almost purely quartic potential Symmetry arguments and selection rules will not be discussed here in detail but it is important to realize that when splitting of vibrational levels occurs due to tunnelling through an inversion barrier the symmetry classification of the split levels must be according to the point group of the planar structure For example for pyramidal ammonia we must use the D3h,and not the C3,,,point group and for non-planar aniline (I) the C, and not the C,s point groupdThen in the absence of vibronic coupling the vibrational selection rule is Av = 0 2,4 6 except in molecules which have very low symmetry such as those belonging to the C,point group in these cases A v is unrestricted Some rather different types of inversion or out-of-plane bending vibrations which have been investigated by supersonic jet electronic spectroscopy are (a) the symmetrical (b3Jout-of-plane bending vibration of the four fluorine atoms in 1,2,4,5-tetrafluorobenzene (7) (b) the symmetrical (b3J out-of-plane bending vibration of all the fluorine atoms in perfluoronaphtha- lene (8) (c) the out-of-plane h bending variation of the CH group in position-2 in 1,3-benzodioxole (9) (d) the out-of-plane b bending vibration of the BH group in catecholborane (lo) and the out-of-plane b bending vibration about the central carbon and oxygen atoms in xanthene (6) All of these are important vibrations in respect of molecular conformation If any of them shows a potential energy minimurn corresponding to a non-planar structure of the molecular skeleton it will also show a second identical minimum corresponding to the mirror image non-planar structure just as for the ground state of ammonia If the energy barrier separating the two minima in the W-shaped potential is sufficiently low and the reduced mass for the vibration sufficiently small quantum mechanical tunnelling through the barrier results in splitting and gross anharmonicity of the energy levels in each well of the potential For out-of-plane ring vibrations such as those referred to above in (9) and (lo) an alternative description as ring-pucker- ing vibrations is often used A simple example of a ring-puckering vibration is the out-of-plane vibration of the CH group in position-3 in cyclopenteneThis vibration could equally well be called an inversion vibration or an out-of-plane bending vibration if the molecular skeleton is non-planar or planar respectively Supersonic jet electronic spectroscopy has opened up the possibility of investigating many vibrations of this type and of relating them to conformation in much larger molecules than previously and there have been many surprises among the results One example is illustrated by part of the FE spectrum of (7) in a supersonic jet4 shown in Figure 3There is a long progression with dv even in V ,,the symmetrical (h3L,)out-of-plane bending vibration of the four fluorine atoms Application of the Franck-Condon principle to this unexpected observation indicates an appreciable change of geometry in the direction of motion of V ,,from So to STherefore since the molecule is planar in Soit must be non-planar in SThe observed vibratio- nal levels in S have been fitted to the potential of equation 1 and are shown in Figure 4The molecule is non-planar in S Although the barrier to planarity is only 78 cm-’ (0 93 kJ mol l) corresponding to an out-of-plane angle of about 11’,the effect on the spectrum is seen in Figure 3 to be dramatic and is an excellent illustration of how sensitive this technique is in detect- ing changes of geometry A simple rationalization based on qualitative valence theory arguments of why such a geometry change occurs between the ground and an excited electronic state is not often possible but in this case there is a convincing argument It has been shown More rigorously the symbolism employing the permutation inversion sym metry introduced by Longuet-Higg~ns~ may be used MOLECULAR CONFORMATION OF ORGANIC MOLECULES IN A SUPERSONIC JET-J M HOLLAS 375 x= c2 c-I Figure 3 Part of the S,-So fluorescence excitation spectrum of J 1,2,4,5-tetrafluorobenzene 28 2 284 286 288 (Reproduced by permission from J Phys Chem 1986,90 3948 ) Waveleng th/nm Figure5 Part of the S,-So fluorescence excitation spectrum of xdnthene (Reproduced by permission from J Chem Phqs 1993 98,836 ) I I2351182 4% I 0 9 Figure 4The h fluorine inversion potential for 1,2,45-tetrafluoroben-zene in S (Reproduced by permission from J Phjs Chem 1986,90 3948 ) that as more hydrogen atoms of benzene are substituted by fluorine atoms the energy of a no* state is successively lowered in energy When four fluorines are present it is lowered so much that it is involved in strong vibronic coupling with the S1,TT* state conferring on it C-F antibonding character A similar geometry change from So to S has been found in perfluoronaphthalene (8) which is planar in So In the FE spectrum in a supersonic jets there is a progression with d veven in the symmetric (b3Jout-of-plane bending vibration of all the fluorine atomsThe energy levels of this vibration in S are anharmonic and were fitted to the potential in equation 1The barrier to planarity is 14 cm-’ (0 17 kJ mol l) Although this barrier is very smdl it can be determined accurately because of the extreme sensitivity of the lower vibrational levels to the perturbing effect of the barrier It is probable that the reason for the barrier is again the mixing by vibronic coupling of the ~n* with a TG* state lowered in energy by the fluorine atoms The inversion vibration of xanthene (6) is very strongly active in the FE spectrum,6 part of which is shown in Figure 5 This indicates dn appreciable change of geometry from So to S Figure 6 shows the inversion potentials calculated for both electronic states with the potential in equation 1 using the inversion energy levels obtained from the FE spectrum for S -9-Figure 6 Inversion potentials for xanthene in Soand S (Reproduced by permission from J Chem Phys 1993,98 836 ) and various SVLF spectra for SoThe molecule is non-planar in So,with a 49 cm-l (0 59 kJ mol I) barrier to planarity but planar in S Because of the barrier in So,and the large reduced mass for this inversion motion there is no observed doubling of the 11 = 0 level in So One important result of this is that in the progression from this level observed in the FE spectrum the molecule behaves as if it has a rigidly non-planar skeleton and belongs to the C,point group with only a plane of symmetryThe inversion vibration IS then an a’,totally symmetric vibration and dv can be even and odd On the other hand transitions in the SVLF spectra from the planar C2”,S state obey the d v even selection rule because the inversion vibration is now a non-totally sym- metric b,,vibration Transitions obeying this selection rule are shown In Figure 6 There is an interesting comparison between the geometry of xanthene and that of 9,lO-dihydroanthracene (1 1) which has a non-planar skeleton in both the Soand S electronic states The barrier decreases from 615cm -(7 36 kJ mol -in Soto 80 cm-(0 96 kJ mol-') in S corresponding to an increase in the angle between the two benzene rings from 144 6" in Soto 164 1" in S The structural differences in xanthene compared to (1 I) in So and S are in agreement with the expectation that the replace-ment of CH by oxygen favours a planar conformation 9,lO-Dihydrophenanthrene (2) has a strongly non-planar twisted conformation in So,with a barrier of 2460 cm (31 6 kJ mol I) and in S with a barrier of 1430 cm ') (17 1 kJ mol-')The reason for this is that the CH groups in the 9-and 10-positions strongly favour a staggered conformation This has been shown to be the case in benzodioxan (12) which has a twisted Cz,structure with a barrier to planarity in both electro-nic states which is too high to be determined but which is probably more than 4000 cm-' (48 kJ mol l) The molecules 1,3-benzodioxole (9) and catecholborane (10) are interesting examples of partly saturated five-membered rings attached to a benzene ring and the question is whether the skeletal atoms of the five-membered ring are coplanar with the benzene ring Figure 7 shows the SVLF spectrumloof (9) with excitation in the 0; band (the pure electronic transition from the zero-point level of So to that of S,)There is a strong band labelled 20 involving two quanta of LJ,~in So,which is 102 cm-' to low wavenumber of 0; but no band at twice this interval The vibration responsible vz0 must therefore be anharmomic indi-cating non-planarity in So Interpretation of the gas phase far infrared spectrum' apparently showed that the levels of the puckering vibration of the CH group in the 2-position are more' or less harmonic and that the molecule is planarThe 102 cm interval in the SVLF spectrum could not involve the b bending vibration about the central C-C bond and the only remaining possibility seemed to be that it involves two quanta of the a twisting vibration of the oxygen atoms Therefore the non-planarity in Sowas attributed to a C twisted conformation lo The barrier to planarity is 157 cm-' (1 88 cm kJ mol ') However this is not the end of the story Investigation of the microwave spectrum of (9) in the zero-point level and several low-lying vibrational levels' has shown that the anharmonic Figure 7 The 0; SVLF spectrum of 1,3-benzod1ox1le (Reproduced by permission from Chem Phjs Lett 1989,157 183 ) t Q Om-CHEMICAL SOCIETY REVIEWS. 1993 vibration is the puckering vibration of the CH group and that the non-planarity involves this group only It is the values of the rotational constants in the excited vibrational levelswhich allow this assignment of the vibration concernedThe far infrared spectrum has also been reinterpreted satisfactorily and a slightly modified barrier to planarity of 126 cm (1 51 kJ mol I) obtained Catecholborane (10) has proved rather less of a problem SVLF spectra show13 no low wavenumber anharmonic vib-rations and together with the gas-phase absorption spectrum these show that the molecule is planar in both So and S The rationalization of this is that there is some -rr-electronconjuga-tion around the five-membered ring involving the otherwise empty 2p orbital on the boron atom stabilizing the planar conformation 3.2Torsional Vibrations The energy barrier to torsional motion generally decreases as the number of times the equivalent equilibrium configuration is encountered during a 360" rotation increases In the case of toluene (4)this number is six and the six-fold energy barrier is indeed very low It has been found from the microwave spectrum,14 to be 4 875 cm (0 058 kJ mol-') in So The torsional potential and energy barrier of (4) in S have been determined from a REMPI spectrum With the toluene in a supersonicjet weak transitions from the zero-point level of So to various torsional levels in S were observed and fitted to a potential like that in equation 4 but with only V6 non-zero The molecular symmetry group3 to which toluene belongs is G but this is isomorphous with the D3,?point group and it is that which we shall use hereThe torsional vibrational levels are labelled in the free internal rotation limit with the quantum number m = 0 1 2 3 all levels with m>O being doubly degenerate When a sixfold barrier is introduced the degeneracy of the m = 3,6,9 levels is split For example the m = 3 level is split into a 34 and 34 pair It has been shown' that the order of these two levels depends on the sign of v6 The 3a;' level is the higher one if V6 is negative and this is the case only for a staggered as opposed to an eclipsed equilibrium configuration In the case of toluene V,is negative in both Soand S (its value is -25 cm-in S,)showing that the staggered configuration is the more stable in both electronic states but not by very much The situation is similar in 4-substituted toluenes For exam-ple in 4-fl~orotoluene'~(13) V6 is -4 8 cm-' (-0 057 kJ 00 0-m c L 1 I I I I 2500 2000 1500 1000 500 0 Akm-' MOLECULAR CONFORMATION OF ORGANIC MOLECULES IN A SUPERSONIC JET-J M HOLLAS mol ')in Soand -33 7 cm (-0 403 kJ mol l) in S It is not surprising that the equilibrium configuration is staggered in both states and that the barrier to be surmounted in going to the eclipsed configuration is very similar to that in toluene in both electronic states When the substituent is in the 2-or 3-position as in 2-fluorotoluene (14) and 3-fluorotoluene (1 5),the methyl torsional potential is dramatically different Figure 8 shows the methyl torsional bands associated with the 0; band in the FE spectrum of (14)The high intensity of these bands indicates from the Franck-Condon principle an appreciable change of torsional potential from So to SThis molecule belongs to the G molecular symmetry group isomorphous with the C3 point group In the supersonic jet nearly all the molecules are frozen down into the Ocr and le levels (0 and 1 are the rn quantum numbers) transitions from the upper le to the lower Oa level are nuclear spin forbidden and can never occur even on collision The selection rules are that only a,-a a2 a and e-e transitions are normally allowed except that a1-a2,a2-a1 transitions may occur weakly due to interaction between torsional and overall rotational motion In Figure 8 several transitions to a and e torsional levels in S and a weak one to the 3a2 level are clearly observedThese have been fitted to a torsional potential for S for which V = 169 cm and r/6 = -5 3 cm l and SVLF spectra allow the torsional potential to be obtained for So,for which V = 228 1 cm and V6 = -26 4 cm I Le 50 3756 2 0 300r-I I t 60' + 120'1 I I -120' -60. 0' Figure 9 Methyl torsional potentials and energy levels for 2 fluoro 0; 37561 Scrn' + II I II I I 0 LL 12 7 51 7 835 1293 45 5 Oal 4e 5e 4L 37600 I 1% 5 toluene in So and S (Reproduced by permission from J Phys Chem 1985 89 5617 )bCH3F are so largeThe methyl torsional potential in S is very similar to that of toluene but completely different in So The methyl torsional potential in 3-fluorotoluene (15)' is very different from that in (16) In So V = 16 9 cm and V = -5 3 cm so that the equilibrium configuration is stag- gered with a barrier of only 16 9 cm ,(0 202 kJ mol ,) to go to the eclipsed configuration In S however V = 123 7 cm and V6 = -26 4 cm the equilibrium configuration is also staggered but with a much higher barrier of 123 7 cm (1 479 kJ mol ') The methyl torsional potentials are rather similar when the substituent in the ring is an amino-group instead of a fluorine atom In 2-aminotoluene (16) the equilibrium configuration is eclipsed in Soand S with barrier heights of 703 cm (8 41 kJ mol ') and 40 cm (048 kJ mol l) respectively In 3- aminotoluene (17) there is a very low barrier in So for which V = 9 cm and V = -10 cm and a barrier of 317 cm (379kJmol 1)inS,,forwhichV,=317cm 'and?',= -19 cm and the equilibrium configuration is eclipsed The effect on the CH torsional potential of the position of substitution in the indole molecule (1 8) has been investigated for 1- 4- 5- 6- and 7-methylindole l9The changes in the potential are appreciable in both Soand S and are a useful probe of the electronic structure of the indole molecule in the region of substitution Until recently very little was known about the torsional potential for the CF group attached to a benzene ringTwo examples that have been investigated through their supersonic jet fluorescence spectra are 2-and 3-aminobenzotrifluoride (19) and (5) Figure 10 shows the FE spectrum of(5) close to the 0; band 2o There is almost no similarity between this and the corresponding spectrum of 3-aminotoluene even though the torsional poten- tials in Soand S are remarkably similar for both molecules The A-37800/cm-' Wavenurnbe r Figure 8 FE spectrum of 2 fluorotoluene showing methyl torsional bands in the region of the 0; band (Reproduced by permission from J Phys Chem 1985 89 5617 ) The potentials for Soand S are shown in Figure 9 In Sothe molecule is eclipsed with a surprisingly high barrier of 228 1 cm (2 728 kJ mol ') in going to the staggered configuration for which 4= 0" In S the fact that V3and V6 are similar in magnitude leads to a complex potential showing that the mole- cule is quasi-staggered there is a small energy barrier of 2 9 cm at 4= 0" but it is below the zero-point levelThe larger barrier at 4= 60" is only 24 7 cm (0 295 kJ mol ') above the minimum It is because the molecule goes from an eclipsed ground state to an essentially staggered excited state that the Franck-Condon factors for the torsional transitions in Figure 8 3 78 CHEMICAL SOCIETY REVIEWS 1993 160 1 1' I 0 50 100 AT/cm-' Y V+ Figure 10 FE spectrum of 3-aminobenzotrifluoride showing CF,-tor-sional bands in the region of the 0 band (Reproduced by permission from Chem Phys Lerf 1991 183 377 ) differences in the spectra are due mainly to the fact that the CF group is very much heavier than the CH groupThe internal rotation constant F which is inversely proportional to the moment of inertia for internal rotation is about 0 29 cm- for CF and 5 cm-' for CH An important consequence of this is that for the same torsional potential the energy levels for CF are very much more closely spaced than they would be for CH In other words the torsional motion of CF tends to behave much more like a torsional vibration with little tunnelling through energy barriers whereas CH torsional motion more often resembles free internal rotation with a much greater tendency for tunnelling The torsional potentials for (5) in So and S are shown in Figure 11 In So V = 9 cm-' and V6= -10 cm-l giving a very shallow potential with the equilibrium configuration about 25" from the eclipsed configuration In S V = 155 cm-I and V = -40 cm-' resulting in an eclipsed configurationThe change in equilibrium geometry from Soto S results in the long progressions in the torsional vibration v which dominate the spectrum in Figure 10 The S potential in Figure 11 shows that due to the large reduced mass of the CF group appreciable tunnelling through the torsional barrier occurs only at the V~ = 10 level This figure also shows why even under the extremely low temperature conditions of the Jet progressions are observed in the FE spectrum with V = 1,2 and 3 in SoThe reason again is that the massive CF group causes these levels to be so low-lying that they are appreciably populated The comparison between the torsional potentials for 3-amino- toluene' and 3-aminoben~otrifluoride~~ shows them to be remarkably similar Those for Soare identical both molecules show a very low barrier In S,,both are eclipsed with quite a high barrier although it is twice as high in 3-aminotoluene These observations parallel the effect of substituting fluorine in the 3- position in toluene l7 Long CF,-torsional progressions are also observed in the fluorescence spectra of (1 9) These have been assigned2 to give the torsional parameters V = 450 cm and V = 83 cm-' in So and V3=220 cm-I and V,= -65 cm-' in SThe equilibrium configuration is eclipsed in both states with barrier heights of 450 cm-(5 38 kJ rnol ') in Soand 220 cm (2 63 kJ mol l) in S The fairly high barrier in So is comparable with that of 703 cm in (16) and 228 cm in (14) In S,,however the barrier in ( 19)is much higher than in (1 6) and (1 4) which are 1 '0I I 9i120 8 7iI i 6 80 3 4 340 7 1 0 D 33 2 '0 -60" 0 60" cp Figure 11 CF,-torsionalpotentials and energy levels for 3-aminobenzo-trifluoride in S (upper) and So(lower) electronic states (Reproduced by permission from Chern Phjs Lett 1991 183 377 ) much more nearly freely rotating Weak hydrogen bonding between an amino hydrogen and a fluorine atom may be responsible A rather different type of molecule with an interesting CF torsional problem is trifluoronitrosomethane (20) Microwave spectroscopy and electron diffraction studies have shown that the equilibrium configuration in So is eclipsedThe S,-So electronic transition involves a n*-n promotion which results in a staggered equilibrium configuration 22 There is barrier of 238 4 cm (2 851 kJ mol l) in So,for which V = 238 4 cm and V = -5 8 cm l and a barrier of 601 5 cm (7 194 kJ rnol ')inS,,forwhich V = 601 5cm 'and V = -1 Ocm Unusual cases of toluene substituted in the ring are the 2- 3- and 4-methylbenzyl radicals generated by laser photolysis of the appropriate a-chloroxylene followed by cooling in a supersonic jet 23 FE and SVLF spectra were obtained and interpreted to give methyl torsional potentials in the Do and D,states (these are doublet states)The highest barriers were found in 2-methylbenzyl(21) for which V = 754 cm (9 02 kJ rnol *) in Do and 362 cm-I (4 33 kJ mol ') in D When there is monosubstitution in the methyl group of toluene the torsional energy barrier is considerably increased For example in benzyl alcohol (22) the parameters in the torsional potential are24 V2 = 140 cm-' in So and V2 = 330 cm l V4 = 3 cm in S the barrier height being 140 cm (1 67 kJ mol-I) in Soand 330 cm-' (3 95 kJ mol l) The question of the planarity of styrene (3) was long-standing and unsolved until the use of laser fluorescence techniques So far as the geometry is concerned there are two competing effects Planarity is favoured because it results in maximum 7~- MOLECULAR CONFORMATION OF ORGANIC MOLECULES IN A SUPERSONIC JET-J M HOLLAS electron conjugation between the ring and the substituent whereas this configuration results in maximum steric hindrance The first proof of the planarity of (3) came from SVLF spectra obtained in the gas phase under low-pressure collision-free conditions 25 As many energy levels as possible of the C( I)-C(a) torsional vibration ~42,are required to obtain the parameters in the torsional potential of equation 4 Obser-vations up to v42 = 4 together with some dv = 2 transitions observed in the gas phase Raman spectrum which accessed higher levels resulted in V = 1145 cm and V4 = -278 cmThis potential is plotted in Figure 12 which also shows in the upper curve the effect of putting V4 = 0The comparison illustrates the drastic effect of the large negative value of V4 It pulls down the potential making it very flat-bottomed The molecule is planar because the minimum is at 4= O" but it takes only 4 8 kJ rnol I of energy to twist the vinyl group 50" out-of-plane The effect of V4 on the lower energy levels is to close them up so that for example the v = 1-0 separation is only 38 cm l and to cause them to diverge up to about v = 9 This very unusual behaviour can be attributed to the effect of n-electron conjugation winning the fight against steric hindrance but only Just With slightly improved data26 the values of V2 = 1070 cm and V4 = -275 cm have been obtainedThe energy barrier in only I 50 90 $/deg -Figure 12 Torsiondl potential and energy levels for styrene in So The upper curve with the dashed energy levels is plotted for V4 = 0 to illustrdte the effect of the large negative value of V in the potential in the lower curve (Reproduced by permission from Chem Phjs Lett 1980 75 94 ) -&I going to the perpendicular configuration is 1070 cm (12 80 kJ mol l) In the S state of (3) the evidence points to planarity with a more purely V potential resulting in the lower torsional vibra- tional levels being almost harmonic However there is an unusually strong Duschinsky effect resulting in the normal coordinates which describe C( 1)-C(a) torsion and out-of-plane bend about C(1) in Sobeing heavily mixed in SThe fact that torsional motion is distributed between two vibrations in S means that it is not possible to determine the torsional barrier in S but it almost certainly considerably higher than in So The more rigid planarity in S is consistent with the simple valence picture in which the r-electron conjugation between C(1) and C(a)is increased compared to So When a fluorine atom is substituted in the 2-position as in 2- fluorostyrene (23) there is a possibility of two planar rotational isomers (rotamers) In fact only the trans rotamer (23) is known presumably because steric hindrance destabilizes the c IS rotamer Figure 13 shows the SVLF spectrum of (23) in a supersonic jet with excitation in the 0; band 27The high intensity of the progression in the C(1)-C(a) torsional vib- ration v4* compared to that observed for styrene is due to a larger change in the shape of the torsional potential from So to S In So,V = 895 cm and V4 = -333 cm resulting in a small barrier to planarity of 15 cm (0 18 kJ mol I) dt 4= 0" This barrier IS below the zero-point level and the molecule is said to be quasiplanar Although this level is above the barrier the v = 0 wavefunction is concentrated above the minima rather than at 4= O" so that the molecule spends more time out-of- plane when in the v = 0 state that it would if there were no barrier In fact it can be shown that there is a barrier at 4= 0"if When a methyl group is substituted in the 2-position in 2-methylstyrene (24) once again only the trans rotamer is found SVLF spectra in a supersonic jet show even longer progressions in the C( 1)-C(a) torsional vibration 28The Sotorsional poten- tial with V = 670 cm and V4 = -420 cm l is illustrated in Figure 14 and shows that the vinyl group is now rigidly about 33" out-of-plane The barrier to planarity is 151 cm (1 81 kJ mol ') The effect of increasing non-planarity when a methyl group is Figure 13 The 0; SVLF spectrum of rranr 2 fluorostyrene (Reproduced by permission from Chem Phjs Lert 1989 154 14 0 0. h OrJ-N qr(N 3000 2000 1000 0 -Aikrn-' v- CHEMICAL SOCIETY REVIEWS I993 42 12 10 8 6 4 2,3 091 I I I 1 I -60 -30 0 30 60 Angle/" Figure 14The C(1)-C(a) torsional potential for the So state of trans-2-methylstyrene (Reproduced by permission from J Chem SOC Faraday Trans 1991 87,3335) substituted in the 2-position in styrene to give the trans rotamer appears to be due to an increasing steric hindrance between the substituent and the hydrogen atom in the a-position A similar effect results when a methyl group is substituted on the a-carbon atom to give a-methylstyrene (25) A REMPI spectrum of this shows a very long progression in the C(1)-C(a) torsional vibration due to a W-shaped potential in Soin which the substituted vinyl group is twisted out-of-plane by 45" This group is coplanar with the benzene ring in S Substitution in the 3-position in styrene is more likely to result in similar amounts of two rotamersThe FE spectrum of 3- fluorostyrene shows two 0; bands,30 275 cm-' apart which are attributed to the cu-(26) and trans-(27) rotamers SVLF spectra obtained with excitation into these and other bands have allowed the observation of C(1)-C(a) torsional levels up to v = 8 in the CIS and v = 4 in the trans rotamers These levels were fitted to a single torsional potential for So,with V = 220 cm- V = 1040 cm-' and V4 = -247 cm-' which is shown in Figure 15 The values of V and V4 are similar to those for styrene and result in the characteristic flat-bottomed minima for each rotamer The value of 220 cm-I for V corresponds to the energy difference (2 63 kJ mol-l) between the two rotamers These results agree with those of a microwave investigation3 in that the cu-rotamer is the more stable but V was estimated3 to be 26 cm with a large uncertainty of 46 cm- Reasons for the discrepancy include the assumptions that in the electronic spectrum the transition probabilities for the two 0; transitions are equal and that in the microwave spectrum the two perma- nent dipole moments are equal The trans-stilbene molecule (28) presents an interesting torsio- nal and structural problem which is closely related to that in styrene (3) One difference however is that since there are two phenyl groups there are two torsional vibrationsThe super- cis trans 1200 900 1 T-I5;;600 4 v 1 300 0 -90 0 90 180 270 - cpldeg Figure 15The C(1)-C(a) torsional potential for the Sostate of LL\-dnd trans-3-fluorostyrene (Reproduced by permission from Chem Phvs Lett 1989.154 228 ) sonic jet SVLF spectra with excitation in the 0 and other bands progressions with dv even in the a torsional vibration v~~ Torsional levels up to v3 = 9 were observed and fitted to a potential for So with V2= 305 cm- and V4 = -85 cm-Because of the large negative value of V4 the potential is qualitatively similar to that for the torsional vibration of styr- ene 26 The potential is very flat-bottomed and may even show a small energy barrier to the planar configuration The molecule is planar or possibly quasiplanar but is very easily twisted even more so than styrene for example an energy equivalent to only 100 cm-' (1 2 kJ mol-') twists it through a torsional angle of about 50" and the barrier height in going to the 90" configuration is only 305 cm-I (3 65 kJ mol l) The configuration of the biphenyl molecule (29) is interesting because of the possible effect of steric hindrance between hydrogen atoms on either side of the bond joining the two benzene ringsThe effect may be sufficiently large to cause the equilibrium configuration to be twisted about the central bond Unlike the case of trans-stilbene (28) biphenyl has only one torsional vibration with a symmetry involving one ring twist- ing relative to the other The motion analogous to the other torsional vibration in stilbene becomes an overall rotational motion In the REMPI spectra33 of (29) in a supersonic jet there are long progressions in the torsional vibration indicating a large change of torsional angle from So to S Fitting levels up to v= 13inSi resultedinvaluesof V2= 1195~m-l~V4= -190 cm-' and V6= 18 cm-l corresponding to a single minimum and a planar configuration In So,the potential was obtained from only one torsional vibrational interval and the intensity distribution in the progressions in the REMPI spectraThe values of V = 50 cm-I and V4 = -148 cm-' for So are not therefore very reliable They correspond to a W-shaped poten- tial with an equilibrium torsional angle of about 44" and a barrier to planarity of about 125 cm-I (1 50 kJ mol-') What-ever the uncertainty regarding the So state potential it is clear that because the molecule is planar in S and the torsional progressions are very long there must be appreciable non- planarity in So Tolane (30) is a planar molecule which contains two benzene rings and like biphenyl has only one torsional vibration and this has a symmetry Several SVLF of 30 are shown in MOLECULAR CONFORMATION OF ORGANIC MOLECULES IN A SUPERSONIC JET-J M HOLLAS 38 1 Figure 16These spectra were obtained with excitation into the vT=0 1 2 3 and 4 levels of the torsional vibration vT of an excited electronic state This excited state may be S the highest of three predicted low-lying singlet states Because the torsional transitions must obey the dv even selection rule the odd- quantum levels could only be accessed by excitation into theTi and bands [The vT = 1 level in Sois sufficiently low-lying ( 17 cm l) to be appreciably populated in the supersonic jet 3 The spectra in Figure 16 show that torsional levels up to vT = 14 were observed and these were fitted to a pure V potential with V = 202 cm (2 42 kJ mol I) and shown in Figure 17 _L '51 T3 II Figure 16 SVLF spectra of tolane with excitation into various torsional vibrational levels (Reproduced by permission from J Phys Chem 1984 88 171 1 ) In the FE spectrum of tolane a short progression in the torsional vibration with d v even is observed 34 35 The vibration wavenumber is 48 cm and V = 1590 cm (19 0 kJ mol ') One of the newer techniques that of threshold photoelectron spectroscopy has been applied35 to tolane in order to obtain the torsional potential in the Do (doublet) ground state of the singly charged positive ionThis technique involves multiphoton ioni- zation of tolane in which one photon takes it to a vibronic level of S and a second photon of a different colour (wavelength) from the first and tunable ionizes the molecule to produce the ion in the Do state Tuning the second photon through the ionization band system takes the ion to various vibrational levels of Do Unlike ordinary photoelectron spectroscopy in threshold photoelectron spectroscopy the ions are produced under field- free conditions and extracted with a pulsed electric field in this case of I V cm A torsional vibration wavenumber of 54 cm and a value of V2= 1980 cm were obtained for the Do stateThe torsional progression in the threshold photoelectron spectrum is only short This is a consequence of the torsional potentials in the S state of the molecule and the Do state of the ion being similar leading to small Franck-Condon factors cm-' 300-04 I L'O 0" I I I I -90" 0" +goo <p Figure 17 The torsional potential dnd energy levels in the So state of tolane (Reproduced by permission from J Phbs Chem 1984 88 1711 ) 4 Conclusions Low wavenumber vibrations have large amplitudes and this property makes them of particular interest from the point of view of molecular structure and conformation Because they have large amplitudes they are very susceptible and therefore the corresponding molecular structural parameter is very sus- ceptible to changes in the immediate molecular environment One important effect of the environment concerns the phase of the molecular sample For example it is a well-known but not always well-remembered fact that low wavenumber vibrations below about 300 cm l have wavenumbers whlch are of the order of 10 cm higher in the liquid than in the gas phase An example of this is the out-of-plane boat-form vibration of the oxygen atoms in p-benzoquinone (3 1) which has a wavenumber of 108 cm in the liquid and 87 cm in the gas phaseThe reason for an increase in wavenumber in the liquid is that the large amplitude motions of the atoms are much more restricted by neighbouring molecules resulting in a reduction of the amplitude than are motions of smaller amplitude For similar reasons the structure of a molecule in a crystal or perhaps adsorbed onto a surface is likely to be affected in those parts of the molecule where vibration has a large amplitude Even if the equilibrium structure remains the same the shape of the potential for the large amplitude motion may be changed In small molecules it is extremely unusual for a vibration to have a wavenumber less than 100 cm In this review we have seen that this is not the case for larger moleculesThe smallest vibration wavenumbers in the molecules discussed here are 7 cm in the S state of 3-aminobenzotrifluoride (5)and 8 cm in the So state of trans-stilbene (28) Clearly some larger molecules can be very floppy indeed The importance of supersonic jet electronic spectroscopy in the determination of molecular conformation or structure can- not be overemphasizedTwo of the more traditional methods have been X-ray crystallography and electron diffraction The problem with X-ray crystallography is that it relates to molecu- lar structure in the pure crystal which may be different from that in the free molecule particularly when the molecule has large amplitude motions Electron diffraction gives a vibrationally averaged structure and in larger molecules of fairly low sym- metry there are likely to be too many overlapping peaks in the radial distribution function for small structural differences to be distinguishable Far infrared spectroscopy can be useful for observing low wavenumber vibrations but very long absorption pathlengths are often necessary and the resulting spectra can be extremely complex making assignments difficult Microwave spectroscopy is a very powerful tool in distinguishing possible gross moleculdr structures such as czs-and trans-isomers and in 382 determining whether a molecule is planarThis technique can also give information relating to potential functions for large amplitude vibrations from the determination of rotational con- stants in excited states of these vibrations The study of electronic spectra of molecules seeded into a supersonic jet started almost twenty years ago and it IS now one of the most powerful techniques for investigation of molecular structure In a supersonic jet the molecules are so rotationally and vibrationally cold that their electronic spectra are dramati- cally simplified to such an extent that vibrational progressions stand out with great clarity By contrast in the gas phase at ambient temperatures they are often overlaid by rotational and vibrational structure caused by appreciable population of rota- tional and vibrational levels which are drastically depopulated in the jet However we still have to rely on the Franck-Condon factors to inject intensity into these progressions But because large amplitude motions are particularly sensitive to changes in the immediate molecular environment the promotion of an electron to a higher orbital in going to an excited electronic state is very often sufficient to cause such a change and produce a progressionThen the fitting of the observed vibrational levels of the large amplitude vibration to a model potential provides accurate structural information and magnitudes of energy barriers to changes in that structureThis information is not confined to the ground electronic state but is extended to low- lying excited electronic states and more recently to states of the positive ion 5 References 1 R E Smalley L Wharton and D H Levy Ace Chem Res 1977 10 139 2 J M Hollas M R Howson T Ridley and L Halonen Chem Phjs Lett 1983,98 61 1 3 H C Longuet-Higgins Mol Phvs 1963,6,445 4 K Okuyama T Kakinuma M Fujii N Mikami and M Ito J Phjs Chem 1986,90,3948 5T Chakraborty D Nath and M Chowdhury J Chem Phjs ,1992 96,6456 6 T Chakraborty and E C Lim J Chern Phys 1993,98,836 7 Y -D Shin H Saigusa M Z Zgierski,F Zerbetto and E C Lim J Chem Phjr 1991,94,3511 CHEMICAL SOCIETY REVIEWS 1993 8 T Chakrahorty and M Chowdhury Chem Phvs Lett. 1991 177 223 9 R D Gordon and J M Hollas J Chem P~JF 1993,99 3380 10 K H Hassan and J M Hollas Chem Phjs Lett 1989,157 183 I1 J A Duckett T L Smithson and H Wieser Chem Phbs Lett 1979,64 261 12 W Caminati S Melandri G Corbelli L B Favero and R Meyer Mol Phvs ,in press 13 A C P Alves K H Hassan and J M Hollas J Chem Soc Furudui Trans 1990,86 3341 14 H D Rudolph H Dreizler A Jaescke and P Wendling 2 Nuturforsch 1967 22A 940 15 P J Breen J A Warren E R Bernstein and J I Seeman J Chem Phrs 1987,87 1917 16 R D Gordon and J M Hollas Chem Phvs Lett 1989 164 255 17 K Okuyama N Mikami and M Ito J Phjs Chem ,1985,89,5617 18 K Okuyama N Mikami and M Ito Law Chem 1987.7 197 19 G Bickel G W Ledch D R Demmer J W Hager dnd S C Wallace J Chem Phjs 1988 88 I 20 R D Gordon J M Hollas P J A Ribeiro-Claro and J J C Teixeird-Dias Chem Phys Lett 1991 183 377 21 R D Gordon J M Hollas P J A Robeiro-Claro dnd J J C Teixeira-Dias Chem Phyr Lett 1993 211 392 22 J A Dyet M R S McCoustra and J Pfab J Chem Soc Furudut Trans 2 1988,84,463 23T -Y D Lin and T A Miller J Phis Chem 1990,94 3554 24 H -S Im E R Bernstein H V Secor dnd J I Seemdn J Am Chem Soc 1991 113,4422 25 J M Hollas and T Ridley Chem Phjs Lett 1980.75,94 26 J M Hollas H Musa T Ridley P H Turner K H Weissen-berger and V Fawcett J Mol Spectrosc 1982 94,437 27 J M Hollas and M Z bin Hussein Chem Phis Lett 1989 154 14 28 J M Hollas and P F Taday J Chem Soc Furuduj Trans 1991 87 3335 29 V H Grassian E R Bernstein H V Secor and J I Seeman J Phi s Chem 1990 94,669 1 30 J M Hollas and M Z bin Hussein Chem Phis Lett 1989 154 228 31 R M Villmanan and J L Alonso Chem Pht F Lett 1989,159,97 32 T Suzuki N Mikami and M Ito J Phjs Chem 1986,90,6431 33 H -S Im and E R Bernstein J Chem Phi r 1988,88,7337 34 K Okuyama T Hasegawa M Ito and N Mikami J Phvs Chem 1984,88 171 1 35 K Okuyama M C R Cockett dnd K Kimura J Chem Phis 1992,97 1649

ISSN:0306-0012    

DOI:10.1039/CS9932200371

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Enantioselect ive and Diastereoselective Molecular Recognition of Neutral Molecules Thomas H. Webb and Craig S. Wilcox Department of Chemistry University of Pittsburgh Pittsburgh PA 15260 U.S.A. 1 Introduction Selective molecular interactions that control or initiate specific physical functions are the essence of biological chemistry. The field of molecular recognition is concerned with studies of such phenomena. Biology without molecular recognition is unima- ginable. Vital biochemical processes such as molecular trans- port genetic information processing and protein assembly involve molecular recognition and complexation as an essential action. An elucidation of the rules and restrictions which govern these intermolecular interactions is important for the under- standing and manipulation of these processes.Therefore the design and synthesis of synthetic receptors has become an important and rapidly growing field of chemistry. The exquisite control and efficiency seen in biochemical systems is a consequence of the great selectivity of molecular interactions that have evolved in biological organisms. A great deal remains to be learned about the ways in which selectivity can be achieved with synthetic receptors. Over the past 20 years studies in molecular recognition have given chemists the tools that are needed to design build and evaluate enantioselective and diastereoselective host molecules and considerable atten- tion is now being focused on this area. Enantioselective host molecules carry great potential for synthetic separative and analytical purposes.This review will discuss the enantioselective and diastereoselective complexation of neutral organic mole- cules by synthetic hosts in solution and assess progress in the field. Binding studies involving charged molecules and binding studies where no attempt was made to test for diastereoselecti- vity or enantioselectivity will not be discussed. The forces involved in neutral molecular recognition and complexation are non-covalent intermolecular interactions -principally dipole-dipole interactions dipole-induced dipole interactions hydrogen bonding London dispersion forces 7~-stacking interactions charge transfer interactions and hydro- phobic or solvophobic effects. It has been pointed out that in order for a receptor to exhibit enantioselectivity it must have at least three points of interaction with one of the guest enan- tiomers at least one of which must be steriochemically depen- Of course two adjacent molecules interact at all points simulta- neously. In analysing selectivity one is often faced with the question of whether or not there indeed exist three (and only three) clearly identifiable regions of the system that are essential for enantioselective binding. The two molecules involved in an association event are often called ‘host and guest’ or ‘receptor and substrate’. ‘Host’ and ‘guest’ (and receptor and substrate) are rather ambiguous descriptors and one person’s host can be another person’s guest. Kitaigorodski recognized that when crystals form the convex portions of one molecule fit into the concave portions of the next molecule.6 When the two molecules that constitute the host- guest system interact according to Kitaigorodskii’s model it is usual to identify the molecule presenting the convex surface as the ‘guest’ and the ‘host’ is the molecule presenting the more concave aspect. Unfortunately the nomenclature is complicated by the fact that both interacting molecules may be convex there may be no definable concave surface.The final definition of host and guest in such systems is arbitrary. Fortunately molecules behave the same no matter what label the chemist may assign! To help organize this review host compounds have been divided into two general classes.These classes are distinguished primarily by their size and by the shape of their binding sites.The first class consists of relatively small molecules with convex binding sites. Since by definition guests also bear convex binding sites these hosts usually interact with a limited portion (less than half) of the guest surface. They are typically used as chiral solvating agents or as chiral stationary phases for HPLC. The second class consists of relatively large molecules with concave binding sites. These hosts typically bind a guest by encapsulating it within a cleft or pocket. Since the binding site of the host is concave and the binding site of the guest is convex they usually interact with each other at many different points over the surface of the guest.These hosts normally exhibit substantially stronger binding than the first class and they have usually been more thoroughly studied and characterized. dent. These interactions may be either attractive or repul~ive.~.~ Thomas H. Webb was born in 1962 in Aberdeen Maryland. He received his B.S.from the University of Virginia in 1984 and is currently finishing his Ph.D. at the University of Pittsburgh with Craig Wilcox. His dissertation work is on the subject of hydrophobic molecu- lar recognition of neutral molecules. Craig Wilcox was raised in a small town in Illinois. His interests in chemistry date from early childhood. During his undergraduate studies at Illinois Institute ofTechnology he worked with Kenneth Kopple Sidney Miller and David Gutman before moving to Caltech to study synthetic organic chemistrjs Mith Robert E. Ireland. After completing his dissertation on the total synthesis of polyether anti-biotics he studied physicul or-ganic chemistry jor tu.0 jteurs with Ronald Breslow at Colum- bia University. He is currently Professor of Chemistry at University qf’ Pittsburgh where he is studying shape-selective aspects of molecular recogni- tion intermolecular forces and biomimet ic c*a talys is. 383 384 2 Chiral Solvating Agents In 1965 Mislow and Raban first proposed that a non-racemic chiral solvent could cause a signal separation in the NMR spectra of enantiomeric solutes ’This phenomenon was first demonstrated by Pirkle in 1966 when he reported different 19F-NMR signals for the two enantiomers of (trifluoromethy1)phe- nylcarbinol when dissolved in optically active 1-phenylethyla-mine This was soon extended to ‘H-NMR when he reported similar results with racemic phenylisopropylcarbinol dissolved in optically active I-( 1-naphthy1)ethylamine Many other examples of such chiral solvating agents have since followed The primary applications for chiral solvating agents are for the determination of enantiomeric purity and absolute configu- ration Chiral solvating agents (CSAs) have a number of advan- tages over other techniques for these purposes More recent techniques such as chiral HPLC which involve the physical separation of the enantiomers can be difficult and time consum- ing Spectroscopic methods such as NMR are much faster and less expensive Measurements of specific rotation while very easy are unreliable and are prone to error caused by contami- nants Also they can not be used to determine the enantiomeric purity of compounds not previously characterizedTo give separate NMR signals the target enantiomers must be con- verted into diastereomers Adding a chiral solvating agent or lanthanide chiral shift reagent to bind to the target enantiomers is easier than chemically derivatizing them with covalently attached chiral auxiliaries Chiral solvating agents that do not contain metals are often preferable because they do not cause the spectroscopic line broadening often experienced with lanthanide chiral shift reagents In NMR studies the signal separation of enantiomers in the presence of a CSA is caused by the formation of temporary diastereomeric complexes between the solute and the CSA as shown in equations 1 and 2 where H-( +) is the optically active host CSA and G-(+) and G-(-) are the enantiomeric solutes The NMR signal observed for each enantiomer is the time- averaged signal of both the complexed and the uncomplexed materialThis can give rise to signal non-equivalence in two ways First a difference in association constants Kf and K between the two enantiomers and the CSA can cause one enantiomer to be preferentially bound This gives rise to differ- ent time-averaged NMR signals Second the two enantiomers may have the same association constants with the CSAs and therefore be bound in equal proportions but the two diastereo- meric complexes thus formed may have intrinsically different spectra Most cases will involve a combination of these two mechanisms Usually large association constants are desirable in order to maximize the amount of shift separation and to allow low concentration of solute and of CSA to be used Among the most widely employed CSAs due both to their broad efficacy and their commercial availability are the aryltri- fluoromethylcarbinols (1) and 1-arylethylamines (2) developed by Pirkle and co-workersThese CSAs can bind to the solutes via two-point hydrogen bonding as shown in (3)and (4),and rely CHEMICAL SOCIETY REVIEWS 1993 on the proximity of the aryl group to induce magnetic aniso- tropy CSAs of type 1 require solutes containing hydrogen bond accepting groups (A) and CSAs of type 2 require solutes containing hydrogen bond donating groups (D) Both types of CSAs also require a solute to contain a second functional group that is capable of binding to either the acidic carbinyl hydrogen or the aryl group of the CSA CSAs of type 1 have been found to be effective for solutes such as carboxylic esters lactones ethers aryl amines amine oxides oxaziridines phosphine oxides sulfinates sultines sulfinate esters sulfoxides sulfites and sulfinamides CSAs of type 2 have been found to be effective for solutes such as alcohols carboxylic acids and amides (5) Cl (7) A number of different CSAs have also been developed by Toda and co-workersThese CSAs can induce signal separation in hydrogen bond-accepting solutes 2,2’-Dihydroxy-l 1-binaphthyl (5) has been found to cause signal separation for amines alcohols sulfoxides and selenoxides O 1,6-Di(o-chlor-opheny1)-1,6-diphenylhexa-2,4-diyne-1,6-d1ol (6) has been found to cause signal separation for amines phosphine oxides and arsine oxides O 4,4’,6,6’-Tetrachloro-2,2‘-bis( hydroxydi- phenylmethy1)-biphenyl (7) has been found to cause signal separation for amines lactams amine N-oxides alcohols sul- foxides sulfoximines selenoxides phosphinates phosphine oxides and arsine oxidesThese compounds have also been used to perform optical resolution on enantiomeric solutes by cla thrate forma tion It is not necessary for a CSA to be capable of forming hydrogen bonds in order to bind to the solute CSAs have also been designed which rely entirely on charge-transfer forces R’YcozR YOzMe N4* (9) R R a Me H \ NO I NO b c Me Me Me Et ENANTIO-IDIASTEREO-SELECTIVERECOGNITION OF NEUTRAL MOLECULES-T H WEBB AND C S WILCOX Mannschreck et a1 have used the fluorene derivative (8) to induce signal splitting in the racemic carbazole derivatives (9) Conversely Balan and Gottlieb have used the optically active helicene (1 0) to induce signal splitting in racemic (8) A large number of other CSAs have also been developedTwo good (though slightly outdated) reviews of the area are avail- able l4 l5 Most of the concave host molecules discussed in Section 4 can also function as CSAs 3 Chiral Stationary Phases One of the principal techniques for the optical resolution of enantiomers is chiral chromatographyThe earliest chiral sta- tionary phases (CSPs) used for this purpose were natural materials such as wool paper or cellulose Modern CSPs are usually synthetically designed for the resolution of certain classes of enantiomers New CSPs can be designed solely for the separation of a specific enantiomer or they can be discovered by an educated process of trial and error Many CSPs have been discovered using the concept of reciprocity by which a molecule that is found to be resolvable on an existing CSP is itself attached to a solid support and tested as a CSP Several generations of reciprocal CSPs have been developed in this manner l6 In designing CSPs it must be remembered that unlike CSAs CSPs require that the two enantiomers have different associa- tion constants in order to achieve separationTo avoid long retention times and peak broadening it is best if the CSP exhibits weak binding (low K,) to the substrate but at the same time in order to achieve separation it must exhibit a large difference in binding strengths (AK,) between the two enantiomers There are several different broad and overlapping classes of CSPs used today These include chiral polymer-based protein- based host-guest complex ligand exchange and donor-accep- tor CSPs Most neutral organic compounds are separated on donor-acceptor (DA) CSPsThese involve interactions between neutral functionality using hydrogen bonding x donor-accep-tor dipole stacking and steric interactions for binding and enantioselection The first DA CSP was demonstrated in 1976 by Mikes et al when several helicenes were resolved using (I I) which probably relies entirely on x donor-acceptor interactions ' The most common type of DA CSP available commercially is the Pirkle type developed by W H Pirkle at the University of Illinois There are two types of Pirkle DA CSPs the x-electron acceptors and the n-electron donors l9 2o The Pirkle n-electron acceptor DA CSPs are based on 3,5-dinitrobenzoyl derivatives of phe- nylglycine (12) and of leucine (13)They are used to separate enantiomers which are r-electron donors and contain aromatic functionality The Pirkle n-electron donor DA CSPs are based on N-(2-naphthyl)alanine (14) These are used to separate enantiomers such as amines amino acids alcohols and thiols which have been derivatized with a 7-r-electron acceptor Typical derivatives are those formed with 3,5-dinitrobenzoyl chloride (1 5) or 3,5-dinitrophenyl isocyanate (1 6) Both types of Pirkle DA CSPs rely on x donor-acceptor hydrogen bonding dipole stacking and steric interactions to achieve selectivity and binding Almost all DA CSPs rely on TIT interactions as a major source of their activity though there are a few which do not One notable example of the latter type is a derivative of N N'-2,6-pyridinediylbis[(,S')-2-phenyl butanamide] (17) developed by Feibush et a1 21 It has been used to separate enantiomers of barbiturates glutarimides and hydantoinsThis DA CSP uses three hydrogen-bond interactions for binding which mimic those responsible for DNA base pairing Since these three hydrogen-bond interactions are co-planar however they are not capable of chiral discrimination This is achieved by the steric interactions of the chiral substituent groups A wide variety of other CSPs have also been developed several of which are available commercially Two excellent reviews of this area are available 22 23 4 Concave Hosts Concave hosts are differentiated from most CSAs and CSPs primarily by the nature of their binding sites Most CSAs and CSPs have convex binding sites and only interact with a minor portion of the guest surface Concave hosts interact with the guest in a much more intimate fashionThey have large concave binding sites and they enclose the guest in a cleft or pocket and interact with it from several converging directions This not only increases the strength of binding but the selectivity as well Concave hosts can often function as special CSAs or CSPs 4.1 Hydrogen Bonding Hosts Host molecules which utilize hydrogen bonding forces often exhibit strong binding as well as a high degree of enantioselec- tion This is due to both the strength and the high degree of directionality of hydrogen bonds Minor changes in the geo- metry of the hydrogen bonds will greatly weaken themThis directionality enables host molecules to be designed which bind guests strongly only in a particular conformation Minor devi- ations in the structure of the guests can greatly weaken binding by changing the conformation of the complex and distorting the geometry of the hydrogen bonds A large number of enantioselective hydrogen bonding hosts have been designed by Rebek and co-workers 24-28 These hosts CHEMICAL SOCIETY REVIEWS. 1993 (20a) X=H (20b) X=Nq CO2CH36Y H3C (21a) Y = NO2 (21b) Y =OH take advantage of the U-shaped relationship between the car- interactions between the phenyl ring of the guest and the boxyl functionalities in Kemp's triacid (1 8) and its derivatives acridine unit of the host Enantioselection is provided by steric By using two of these units separated by a spacer the Rebek repulsion between the ester of the guest and the bulky amide of team has been able to construct clefts containing convergently the host Structure (22) has been proposed as the conformation directed hydrogen bonding groups of this complex The first enantioselective hosts of this type consisted of clefts More recently the Rebek group has synthesized chiral clefts containing chiral secondary amides Host (19) acts as a chiral containing convergent lactams and cylic imides [(23) (24) and solvating agent for racemic alcohols such as a-phenylethanol (25)]The convergent lactams [(23) and (24)] have been found by and menthol 24 Hosts (20a and b) act as chiral solvating agents 'H-NMR to bind the diketopiperazines (26) and (27),26 27 L-for racemic phenylalanine derivatives (2 1a and b) 25The phenyl- hydantoins [(28)-(3 l)] and L-hydroorotic acid methyl ester alanine derivatives are bound not only by a hydrogen bond (32)23 with a high degree of enantioselectivity (Table 1) Host between the guest amine and the host acid but also by nstacking ( + )-(23) was found to bind cyclo-(L-leucyl-L-leucine) (27) with R 'R (26) R' = H R2 = 1-Bu (30) R = CH(CH3)z (27) R' = R2 = 1-Bu (31) R = CH(CHS)CHzCH3 (33) N-CBz-Al-A2-NHBn (34) A1 = glyane; A2 = L-leuane (35) A1 = L-leuane; A2 = glyane (36) A1 = L-isoleuane; A2 = glyane (37) A1 = A2 = L-wleuane (38) A1 = A2 = L-almne N-CBz = Bn = benzyl ENANTIO-/DIASTEREO-SELECTIVERECOGNITION OF NEUTRAL MOLECULES-T H WEBB AND C S WILCOX 387 Table 1 Association constantsa and AAG valuesb of host (23) with guests (26) and (27) host (24) with guests (28)-(32) dnd host (25) with guests (34)-(38) Association constants K (M l) for host AAGh Guest (-)-(23) (+1423) (-1424) (+1424) (-)-(25) (+)-(25) (kcal mol l) 73000 2900 -19 82000 840 -27 4800 1720 -06 1050 390 -06 7080 705 -14 7070 650 -14 380 2100 10 1672 1003 -03 4736 2340 -04 4250 2320 -04 62 134 05 320 405 02 Association constants were determined by NMR titrations in CDCI carried out at 25°C * AAC = AC (-) endntiomer -AG (+) enantiomerThe lower K.,s for this guest are probably due to the fact that the CDCI solution contained 10% d8THF an association constant of approximately 82000 M *,while host (-)-(23) bound it with an association constant of approxi- mately 840 M -This corresponds to a Ad G of 2 7 kcal mol- l Table2 Free energies of associationa for hosts (39a and b) which is among the largest observed for neutral substances The and various amides proposed conformation of the complex is shown in structure AG(kca1 (33) For optimal binding the guest is held in a rigid conforma- mol-') of tion by the four hydrogen bonds with the lactams Steric binding for repulsions between the R groups of the guest and the naphthyl AAGh Solvent (39a) (39b) (kcal mol ')spacer of the host distort the disfavoured complex severely Guest reducing both the strength and number of the hydrogen bonds (5')-PhCHMeNHCOMe C6D6 -3 04 thus reducing the strength of binding (R)-PhCHMeNHCOMe C6D6 -2 62 0 42 Host (25) was found to exhibit enantioselectivity with dipep- (5')-PhCHMeNHCOH C6D6 -3 18 tides (34)-(38) (Table 1) **It is believed that the lower enantio- (R)-PhCHMeNHCOH C6D6 -2 85 0 33 C6D6 -180selectivity found with the dipeptides is due to their greater (5')-PhCHMeNHCOEt (R)-PhCHMeNHCOEt C6D6 -155 0 25flexibility and their ability to adopt multiple binding (5')-1-NpCHMeNHCOMe -2 56 C6D6conformations (R)-I-NpCHMeNHCOMe C6D6 -231 0 25 (5')-BnOAlaNHCOMe -2 29 C6D6 (R)-BnOAlaNHCOMe C6D6 -181 0 48 (5')-MeOPGlyNHCOMe -191C6D6 (R)-MeOPGlyNHCOMe -2 06 -0 15 HY LN Ac-L- Ala-NHBn CDCI -2 36 Ac-D-Ala-NHBn CDCl -136 -10 Ph Ac-L- Ala-NHMe CDCI -202 I PhAc-D-Ala-NHMe CDCl -I91 -01 HX Ac-L- Ala-OBn CDCl -I27 Ac-D-Ala-OBn CDCl -0 86 -0 4' I Ac-L- Ala-OBn C6D6 -3 46 I[ Ac-D- Ala-OBn C6D6 -2 93 -05 Ac-L-Ala-L-Ala OBn CDCl -2 57 Ac-D- Ala-L- Ala-OBn CDC1 -1 67 -09 Ac-L- Ala-D- Ala-OBn CDCl -2 24 Ac-D- Ala-D-Ala-OBn CDCl -148 -08 (39a) Z =CH Ac-L- Ala-NH-t-but yl CDCl -2 35 (39b) Z =N Ac-D- Ala-NH-t-but yl CDCl -104 -1 3c Ac-L- Ala-NH-t-butyl C6D6 -4 38 Ac-D-Ala-NH-t-butyl C6D6 -3 31 -1 1 Still and co-workers have also designed hosts (39a and b) Free energies of association were determined by NMR titration at which bind amide and amino-acid derivatives (Table 2) 29 30 In host (39a) the amides are bound by the hydrogen bonds H;*-O 25°C AAG = AGs -AGRfor (39a) or AC -AC for (39b)These figures dre uncertain due to a low extent of saturation achieved in the titration and 0,-..H Enantioselectivity is due to the steric interactions of the other substituents In host (39b) an additional hydrogen bond is allowed H-...Z which can lead to enhanced enantio- selectivityTwo additional reports by this group are included in the addendum to this review proposed that this chiral recognition is due to the three interac- Mendoza and co-workers were able to design a host (40) to tions shown in (41)The carboxylate undergoes hydrogen bind enantioselectively to zwitterionic amino-acids containing bonding with the guanidinium unit the aromatic side-chain aromatic side-chains 31 When an aqueous solution of racemic undergoes n-stacking with the naphthalene unit and the ammo- Trp or Phe was extracted with a CH2CI solution of (40) only nium group binds with the crown ether Unlike the L-enan- the L-enantiomers were extracted An HPLC analysis of the tiomers the D-enantiomers of the amino-acids cannot undergo diastereomeric dipeptides prepared from the extracts and suit- all three interactions simultaneously able L-Leu derivatives indicated the amount of D-enantiomer to Diederich and co-workers have discovered that binaphthyl be less than 0 5% forTrp and less than 2% for Phe It has been derivatives (42a-d) will enantioselectively bind quinine (43) CHEMICAL SOCIETY REVIEWS 1993 Table 3 Association constantsa and AdG values for hosts (42a-d) and guests (43) and (44) Guest (43) (44) (40) (43) (44) (43) (44) (43) (44) Association constants K (M l) for host (R)-(42a) (94424 95 60 20 75 (R)-(42b) (S)-(42b) -1 L- -1 -1 1270 650 625 850 (R)-(424 (S)-(42d) 775 140 105 550 Ad@ (kcal mol ') 0 27 -0 74 --0 39 -0 18 0 99 -0 95 Free energies of association were determined by NMR titration In CDCI at 20°C AC = ACs -ACR No measurdble complexation was observed (42) R R' R" a PhCHz PhCHz H b PhCHz PhCHz CH3 c H H CH3 (45) d H PhCHz CH3 (44)Meoa and quinidine (44) in the major grooveThe association con-stants are shown inTable 3 32 When all of the hydroxyl groups on the binaphthyls are alkylated (42b) no binding is observed This indicates that hydrogen bonding is the principal attractive interaction involved although n-stacking interactions between the aromatic rings may be involved as well This discovery led them to design host (43,based on the 9,9'-spirobifluorene unit which contains a more rigid organized cleft than that of binaphthyl 33 Structure (45) exhibited strong binding and good enantioselectivity for a number of dicarboxy-lic acids [(47)-(55)] as shown in Table4 A comparison with the binaphthyl host derivative (46) containing analogous function-ality demonstrated the importance of the conformational inflexibility of the spirobifluorene unit to its chiral recognition abilities Hamilton and co-workers have designed a cleft molecule (56) based on the binaphthyl unit which binds tartaric acid deriva-tives selectively 34 Fluorescence spectroscopy in CH,C12 gave association constants of 3 0 x lo5 M-l for D-(-)-dibenzoyl tartaric acid and 3 6 x lo5 M-l for L-(+)-dibenzoyl tartaric acidThis gives a A d G of 0 11 kcal mol -The proposed binding conformationsare shown in (57) and (58) The enantioselectivity is believed to be due to unfavourable steric interactions between the benzoyl groups and the binaphthyl spacer in the D-(-)-(47) R = C~HC,CHZOCO (49) R = GHSCHZOCO (48) R = n-BuOCO (50) R = n-BuOCO (54) ENANTIO-/DIASTEREO-SELECTIVERECOGNITION OF NEUTRAL MOLECULES-T H WEBB AND C S WILCOX (55) Association conStantSaand and (46) and guests (47)-(55) for hosts (45) Association constants K Guest (M l) for host AAGh(kcal mol l) (R)-(45) (W45) 8 20 4200 -09 1400 4800 -07 I4000 3900 08 23000 10000 04 680 3400 -09 800 2200 -06 420 680 -03 490 11300 -18 (49) 20800 19400 01 (55) 8500 7200 01 Free energies of association were determined by NMR titration in CDCl at 20°C 'AAG= AC.7-AGR NH m-derivative For dipivaloyl tartaric acid derivatives the associa- tion constants are 1 01 x lo6 M for the D-( -)-enantiomer and 3 2 x lo5 M -for the L-( +)-enantiomer to give a AdG of -0 67 kcalmolThe increased selectivity for the D-( -)-enantiomer is believed to be due to stabilizing x-Me interactions between the binaphthyl unit and the trimethyl acetate groups Hara and co-workers have designed the host (R,R)-(59) to bind the diol (60) 35The orientation of the hydrogen bonding sites forces the two diol enantiomers to bind with a different twist in their orientations The difference in the steric interactions for each orientation results in different strengths of binding The host (R,R)-(59) binds diol (S,S)-(60) with an association con- stant of 25 0 M-l and diol (R,R)-(60) with an association constant of 5 8 M-' This corresponds to a AdG of -0 87 kcal mol- between the two enantiomers 4.2 Lipophilic Binding Hosts Since hydrogen bond interactions are less effective in aqueous media n-stacking and hydrophobic interactions are often used to bind a substrate in waterThe relatively non-directed nature of these forces makes the design of enantioselective hosts more challenging since small changes in binding conformation will not necessarily result in large changes in binding strength These hosts usually tightly encapsulate their guests making intimate contact over a large surface rather than at discrete points The first example of enantioselective recognition in aqueous media by a synthetic host was achieved by Koga and co-workers in 1984 36 They used a cyclophane host (61a) consisting of two diphenylmethane skeletons bridged by two chiral C,-chains derived from L-tartaric acid This was found to bind the aroma- tic carboxylic acids [(62)-(65)] in acidic D,O(pD = 1 2) Although association constants were not determined binding was evinced by large upfield shifts (AS up to -I 3 ppm) in the proton NMR signals of the guests Enantioselectivity was demonstrated by the signals of the two enantiomers being shifted to a different degree Further evidence of the enantio- selectivity of these hosts was given when asymmetric reductions of achiral arylglyoxylic acids (66)-(68) were performed on their inclusion complexes with hosts (61a) and (61 b) using NaBH 37 When the NaBH reacts with the bound guests it is believed that steric interactions with the host cause a preference for the reagent to approach from one face of the carbonyl over the otherThis gives rise to the reaction enantioselectivities shown in Table 5 Diederich and co-workers have been among the most prolific workers in this area producing a number of chiral lipophilic CHEMICAL SOCIETY REVIEWS 1993 Table 5Asymmetric reduction of arylglyoxylic acids (69a) R=Me2N R'=Me (69b) R=EtflCHZ R'=Et hostsTheir first attempt was with hosts (69a and b) using the tetrasubstituted biphenyl unit as the source of chirality 38 These hosts failed to bind aromatic guests however due to insufficient preorganization of the host cavities prior to complexation It was believed that the bridging aliphatic chains were free to approach each other thus closing the cavity (71) a R=C@H g Me R b R=C@c R=CONH2 d R = C-Me e R=CON~ f R=com<N S O\ The biphenyl unit was then replaced with a chiral source structurally related to the natural alkaloids latifine and cheryl- line to give the host (70) 39 40 NMR studies in aqueous solu- tions with 40-50% MeOH demonstrated enantioselectivity by showing different chemical shifts between the enantiomers of both naproxen (71a) and its methyl ester (71d) A high degree of overlap between the signals of the host and the guests prevented an accurate determination of the association constants from being made However very crude estimates were made indicat- ing association constants of approximately 50 M-' for (71a) and 300 M-for (71d)The low association constants observed are believed to be due to the narrowness of the cavity provided by the chiral spacer To alleviate this difficulty the host was improved by replacing the phenyl ring of the chiral spacer with a naphthyl unit to provide a host (72) with a wider cavity 41This host provides stronger binding to naproxen derivatives and demonstrates a moderate degree of enantioselectivity as shown in Table 6 A further improvement was made by using binaphthyl deriva- tives as chiral sources to give hosts (73) and (74) 37 32 42 The Table 6 Association constants" and ddG values for host (72) and various naproxene derivatives Association constants K (M I) for host Guest iR14.72) (s)-(72) AAG" (kcal mol I) (71a)' 930 810 0 08 (7lc) 450 420 0 04 (7 14 1130 1070 0 03 (71e) 1210 900 0 17 (710 730 470 0 26 (7 1gId 230 200 0 08 Free energies of association were determined by NMR titration in D,O/ CD,OD (60 40) at 20°C * AAC = dCs -ACR In 0 01 M D,O CD,OD (50 50) In 0 01 M DCIICD OD (60 40) \' -I R = CHZCH2N+Et3 A (74) " 'major groove' of the binaphthyl spacers was found to provide wide enough cavities for the inclusion of aromatic guests while retaining chiral discriminationThese receptors were found to give good binding and enantioselection of the naproxen deriva- tives (71a-f) (Table 7) In the anticipation that hosts containing two chiral spacers would give greater enantioselection than those containing one chiral and one achiral spacer a number of hosts [(75)-(77)] were synthesized which contained two chiral binaphthyl spacers linked by two C,-chains 43 These hosts were found to give very poor enantioselection (Table 8)This is believed to be due to the high degree of conformational flexibility available to these hosts A highly rigid host (78) composed of a Troger's base and a diphenylmethane unit linked by two ethenoanthracene moieties. was reported by Wilcox and co-workers 44 46 This is one of the few hosts that provide enantio- and diastereo-selection of neutral aliphatic and alicyclic substrates Determined by NM R studies in aqueous media (pD = 6 8) enantioselection has been observed for menthol (79),46 3,3-dimethylcyclohexanol and citronellol (81) 48 Substantial diastereoselection has also been observed between ( -)-menthol (79) (+ )-isomentho1 (82),4648 and (+)-neomenthol (83),47 and between czs-and trans-4-t-butylcyclohexanol(84)49These conclusions are based on the observation of differing chemical shifts between the enantiomers and the calculation of differing apparent associa- ENANTIO-/DIASTEREO-SELECTIVERECOGNITION OF NEUTRAL MOLECULES-T H WEBB AND C S WILCOX 39 1 Table 7 Association constants" and AAG values for hosts (73) and (74) with various naproxene derivatives Association constants Guest K (M-l) for host AAGb (kcal mol-') (R)-(73) (9473) (947 1a) (S)-(71b)" 2105 1040 2540 1335 0 16 0 15 (947 1c) 775 1010 0 15 (947 1d) 2075 31 10 0 23 (947 1e) 1760 2840 0 28 (9471f) 1405 2490 0 33 0 20 Free energies of association were determined by NMR titration in Dzo/ CD,OD (60 40) at 20°C AAG = ACs -AGR In 0 1 M DCI/CD,OD (60 40) ' In 0 01 M K,CO,/CD,OD (60 40) tion constants for enantiomeric and diastereomeric solutes Determination of precise association constants has been con- founded by the observation of slightly different apparent asso- ciation constants for different protons on a single guestThis perturbation is likely to be due to the effect of a small amount of higher order (2 1 host guest) binding Most lipophilic host molecules rely on n-stacking rather than hydrophobic interactions to achieve binding Since host (78) binds aliphatic and alicyclic guests n-stacking interactions are not expected and binding is believed to be achieved primarily by hydrophobic interactions Intermolecular interactions involving attractive forces between molecules such as hydrogen bonding or n-stacking are enthalpically favourable (AH <0) and entro- pically disfavourable (AS <0) According to the equation AC = AH -TAS the binding strength of host molecules which rely on such interactions should decrease as the temperature is increased Hydrophobic solute interactions are enthalpically disfavourable (AH > 0) and entropically favourable (AS >0) Therefore the binding strength of host molecules which rely on these interactions should increase as temperature is increased For example for the hydrogen-bonded complex (86) prepared by Adrian and W~lcox,~~ the binding strengths dropped precipi- Table 8 Association constantsa and AAC values for hosts (73 (76) and (77) with naproxene derivative (74f) Association constants K (M-I) for host Hosts (R)-(74f) (9-(74f) AAGh(kcal mol ') -1 -1 -(75)(76) 320 375 -0 09 (77) 43 5 455 -0 Free energies of association were determined by NMR titration in D,O] CD,OD (60 40) at 20°C * AAG = dGs -AGR No measurable complexation was observed tously as the temperature was increased In dry CDCl the association constants were 24000 M -at 283 K 9400 M -'at 293 K 4800 M-at 303 K 2100 M-at 313 K and 1000 M at 323 K A van't Hoff plot of this data revealed that AH = -14 3 kcal mol- and AS = -30 cal mol- K For a complex based on n-stacking interactions such as that between host (920 and isoquinoline prepared by Dougherty and co-worker~,~ ' the association constants again dropped steeply as the temperature increasedThe van't Hoff plot revealed that AH = -11 kcal-mol-and AS = -17 calmol- 'K Binding studies with host (78) and (-)-menthol (79) were performed at various tempera- tures and the association constants for one of the protons were found to be 3400 (+ 700 -500) M-at 296 K 4400 (+600 -500) M-' at 308 K 4200 (+ 400 -300) M-' at 318 K dnd 4500 (+ 1300 -900) M-at 328 K 47 Different association constants were observed at these temperatures for other pro- tonsThe observed differences in association constants calcu- lated for various guest protons indicate that this system is perturbed by small amounts of higher order binding Because of this and because the change in association constants over the observed temperature range is small compared with the uncer- tainties in the association constants accurate determination of AH and AS via a van't Hoff plot is not possible It is significant however that these association constants do not drop as the temperature is increased Instead they seem to rise slightly Similar behaviour was shown by all of the guest protons observedThis indicates that AH >0 and AS >0 and supports the hypothesis that binding in this case is achieved primarily by hydrophobic interactions Another host (85) has also been synthesized which replaces the diphenyl methane unit with another ethenoanthracene unit 52 Preliminary results indicate that this host exhibits both stronger binding and greater enantioselectivity than host (78) Murakami et a1 have designed a cage-type cyclophane (87) using L-and D-valine residues as chiral sources 53This host has been shown to enantioselectively bind the steroid hormones a-estradiol (88) p-estradiol (89) and estratriol (90) as shown in Table 9The chemical shifts induced by complexation indicate that it is the aromatic moieties of the steroids that are bound within the cavity This is further supported by the observation that testosterone (9 l) which does not contain an aromatic moiety does not bind to host (87) Dougherty and co-workers have also created a series of chiral hosts (92a-g) base on two linked ethenoanthracene units 54 These have been used to bind neutral achiral aromatic guests and to enantioselectively bind chiral cationic guests containing trimethylammonium substituents No results have been reported however in regard to their enantioselectivity for neutral chiral guests 4.3 Constrictive Binding Hosts Another class of host molecules exists which rather than relying on attractive interactions with the guests utilizes what has been termed 'constrictive binding'These hosts consist of rigid hollow armatures containing portals which provide access to the CHEMICAL SOCIETY REVIEWS 1993 -0,cY c0,-H,C ENANTIO-/DIASTEREO-SELECTIVERECOGNITIONOF NEUTRAL MOLECULES-T H WEBB AND C S WILCOX Table 9 Association constantsa and ddG values for host (87) with steroids (88)-(91) Association constants K (M-l) for host Guests (+)-(87) (-1-037) AAGh(kcal mol-l) (88) 460 1300 -0 62 (89) 760 700 0 05 (90) 360 520 -0 22 (91) -[ -f - Free energies of association were determined by NMR titration in D,O/ CD,OD (75 25) at 27°C * AAG = AG (-)-enantiomer -AC (+) endntiomer No measurable complexation was observed C 0 5 g x=$1 c)-"csog interior of the hostsThe relative size and shape of these portals compared to those of the guests imposes steric constraints which the guests must thermally overcome in order to enter or leave the hosts Cram and co-workers have developed a constrictive host (93) with chiral portals by linking two cavitands with four binaphthyl spacers 56 Host-guest complexes were formed by heating the host in neat guest cooling and isolating the complex The complex was then dissolved in CDCI at 23 "C and the rate of guest release was measured by 'H-NMR From the complexes of enantiomerically pure (R)and (S)host (93) and (S)-BrCH,CH(CH,)CH,CH the first-order rate constants for guest release were determined to be 4 4 x 1OP2h-for the (R)-(S)diastereomer and 6 2 x 10-3h-1 for the (S)-(S) diastereomerThis gave kRs/kss= 7 and ddG = 1 1 kcal mol at 23 "C Enantiomerically pure (9-host (93) and racemic BrCH,CH,CHBrCH gave a mixture of diastereomeric com- plexes in ratios ranging from 1 5 1 to 2 1 indicating a ddG of association of approximately 0 3 kcal mol-I at 100°C The dissociation rate constants were kr, =3 0 x lo-' h-l and kslow= 5 8 x 10 h-' This gavek~ast/ksl,w 5 and ddG = 1 0= kcal mol at 23 "C The less thermodynamically stable diaster- eomer gave the faster rate In the same manner enantiomerically pure (a-host (93) and racemic BrCH,CHBrCH,CH gave a diasteriomeric ratio of 2 1The dissociation rate constants werekf, = 1 21 x 10 h and kslow= 1 3 x lo- h-' This gave k~ast/k,lo,= 9 and Ad G = 1 3 kcal mol -at 23 "C In this case the more thermody- namically stable diastereomer gave the faster rate Collet and co-workers have also created a host (94) which appears to utilize constrictive binding 57 This host was com- plexed with racemic bromochlorofluoromethane (95) in CDCI and the association constants of the diastereomeric complexes were determined by 'H-NMR at 59 "C These were found to be 0 30 M -'for the (+)-(95) *( +)-(94) diastereomer and 0 22 M -for the (-)-(95)-(+)-(94) diastereomer to give a ddG of 0 21 kcal mol- (94) 5 Concluding Comments The field of molecular recognition is moving forward at an exciting pace Progress is being made along several pathsThe acceleration of chemical reactions through a template effect based on non-covalent interactions has been demonstrated in a totally synthetic system * New self-replicating systems (mole- cules that are catalysts for their own synthesis) have recently been developed 59 Experiments in crystal design for new mater- ials production are being pursued 6o Chemists have always sought new knowledge to support the development refinement and improvement of chemical techno- logies The obvious applications of enantioselective and diaster- eoselective receptors in separation and analysis are being vigor- ously pursued in many laboratories Most of the technological applications of enantioselective and diastereoselective synthetic receptors are yet to be inventedThe practical importance of shape-selective binding will be magnified when shape selectivity can be combined with catalytic capability Examples from the biological world provide some idea of the fantastic level of self- organization and control that can be based on shape-selective binding and catalysis Experiments described in this chapter provide the foundation for further advances in this promising new field 6 Addendum After submission of this review Still and co-workers published a report61 on a unique peptide-binding host (96) which can be synthesized in one step It is highly enantioselective exhibiting differences of free energy of binding between enantiomeric pairs of up to 3 0 kcal mol- l as shown inTable 10 This is the largest enantioselectivity so far reported The host binds simple pep- 394 CHEMICAL SOCIETY REVIEWS 1993 tides through four intermolecular hydrogen bonds With larger peptides the host seems to be able to use outlying amides to form Table 11 Free energies of associationa for hosts (97) and (98) hydrogen bonds with up to three amino acid residues and various peptides AC (kcal mol l) AAGh (kcal mol-l) of binding for for Peptides (97) (98) (97) (98) N-Boc-D- Ala-NHMe -17 -21 N-Boc-L-Ala-NHMe -39 -38 -22 -17 N- Boc-L-Ala-NHBn -14 N-Boc-L-Ala-NHtBu NCi N-Boc-D-Val-NHMe -15 -15 N-Boc-L- Ala-NHMe -44 -40 -29 -25 N-Boc-D-Leu-NHMe -15 -16 N-Boc-L-Leu-NHMe -41 -38 -26 -22 N-Boc-D-Ser-NHMe -38 -44 N-Boc-L-Ser-NHMe <-61 <-62 <-23 <-18 N-Boc-L-Ser(0Bn)-NHMe -3 1 N-Boc-D-Thr-NHMe -32 -36 N-Boc-L-Thr-NHMe < -6 2 Igd < -3 0 N-Ac-D- Ala-NHMe -27 N-Ac-L-Ala-NHMe -39 -12 N-Ac-D- Ala-NHt Bu -20 N-Ac-L- Ala-NHt Bu -30 -10 Table 10 Free energies of associationa for host (96) and Free energies of dssocidtion were determined by NMR titration of 0 5 mMvarious peptides (97) or (98) in CDCI at 25°C * AAC = ACL-ACD NC no complexation dG (kcal mol-I) of detected rl Ig loo large to measure accurately binding for peptide AAGh Guest Peptide 0-1 (D) (kcal mol-’) 7 ReferencesN-Ac-Gly-NHMe -19 N-Ac-Ala-NHMe -35 -22 -13 1 J -M Lehn Angew Chem Int Ed Engl 1988,27,89 N- Ac-Val-NHMe -50 -24 -26 2 D J Cram Angew Chem Int Ed Engl 1988,27 1009 N-Ac-Ile-NHMe -43 -24 -19 3 J Rebek Jr ,Science 1987,235 1478 N-Ac-Leu-NHMe -34 -24 -10 4 W H Pirkle and D W House J Org Chem 1979,44 1957 N-Ac-PGlyc-NHMe -59 -29 -30 5 C E Dalgliesh J Chem SOC,1952 3940 N- Ac-Phe-NHMe NCd -20 >+20 6 A I Kitaigorodsky ‘Molecular Crystals and Molecules’ Academic N-0c‘-Tyr-NHMe NC Press New York and London 1973 N-Ac-Ser-NHMe -35 -34 -01 7 M Raban and K MislowTetrahedron Lett 1965,48,4249 N- Ac- H Sed-NH Me -51 -37 -14 8 W H Pirkle J Am Chem Soc 1966,88 1837 N-Ac-Thr-NHMe -35 -29 -06 9T G Burlingame and W H Pirkle J Am Chem SOC,1966 88 4249N-Boc-Val-NHMe -28 -17 -11 N-Boc-Val-NH -49 -37 -12 10 FToda K Mori J Okada M Node A Itoh K Oomine and K Fuji Chem Lett 1988 131 N-Boc-Gly-Val-NHMe -62 -32 -30 11 F Toda R Toyotaka and H Fukuda Tetrahedron Asymmetry,N-Boc-Gly-Val-Gly-NHBn < -7 2 -46 <-26 1990,1 303 0 Free energies of association were determined by NMR titration of 0 5 mM 12 A Mannschreck P Rosa H Brockmann Jr and T Kemmer (96) in CDCI at 25 “C * AAG = ACL-ACD PGly phenylglycine Angew Chem Int Ed Engl 1978,17,940 fd NC no complexation detected Oc octanoyl HSer homoserine 13 A Balan and H E Gottlieb J Chem SOC Perkin Truns 2 1981 350 14 W H Pirkle and D J Hoover in ‘Topics in Stereochemistry’ ed N L Allinger E L Eliel and S H Wilen John Wiley and Sons Still and co-workers have also used hosts (97) and (98) to bind 1982 Vol 13 simple peptides 62These basket-shaped hosts are C symmetric 15 G R Weisman in ‘Asymmetric Synthesis’ ed J D Morrison and exist largely in a single family of closely related conforma- Academic Press New York 1983 Vol I Chap 8 tions They also exhibit extremely high enantioselectivity as 16 W H Pirkle and T J Sowin J Chromatogr 1987,396 83 shown in Table 11 17 F Mikes G Boshart and E J GiI-Av J Chem SOC Chem Commun 1976,99 18 F Mikes and G J Boshart Chromutogr 1978 149,455 0 19 W H Pirkle D W House and J M Finn J Chromatogr 1980 192 143 20 W H Pirkle andT C Pochapsky J Am Chem SOC,1986,108,352 21 B Feibush A Figueroa R Charles K D Onan P Feibush and B L Karger J Am Chem Soc 1986 108 3310 22 W H Pirkle andT C Pochapsky Chem Rev 1989,89,347 23 W H Pirkle and T C Pochapsky in ‘Advances in Chromato- graphy’ ed J C Giddings E Grushka and P R Brown Marcel Dekker 1987 Vol27 Chap 3 24 J Rebek Jr B Askew N Islam M Killorah D Nemeth and R Wolak J Am Chem Soc 1985 107,6736 25 J Rebek Jr ,B Askew P Ballester and M Doa J Am Chem Sot 1987,109,4119 26 K -S Jeong A V Muehldorf and J Rebek Jr ,J Am Chem Soc 1990 112,6144 H 27 K -S Jeong T Tjivikua A Muehldorf G Deslongchamps M Famulok and J Rebek Jr ,J Am Chem Soc 1991 113,201 (97) x= s 28 M Famulok J -S Jeong G Deslongchamps and J Rebek Jr (98) X = 0 Angew Chem Int Ed Engl 1991,30,858 ENANTIO-/DIASTEREO-SELECTIVERECOGNITION OF NEUTRAL MOLECULES-T. H. WEBB AND C. S. WILCOX 29 P. E. J. Sanderson J. D. Kilburn and W. C. Still,J.Am. Chem. SOC. 1989,111,8314. 30 R. Liu P. E. J. Sanderson and W. C. Still J. Org. Chem. 1990 55 5184. 31 A. Galan D. Andreu A. M. Echavarren P. Prados and J. de Mendoza J. Am. Chem. SOC.,1992,114 1511. 32 F. Diederich M. Hester and M. A. Uyeki Angew. Chem. Int. Ea’. Engl. 1988 27 1705. 33 V. Alcazar and F. Diederich Angew. Chem.. Int. Ed. Engl. 1992,31 1521. 34 F. Garcia-Tellado J. Albert and A. D. Hamilton J. Chem. SOC. Chem. Commun. 1991 1761. 35 Y. Dobashi A. Dobahi H. Ochiai and S. Hara J.Am. Chem. SOC. 1990,112,6121. 36 I.Takahashi K. Odashima and K. Koga Tetrahedron Lett. 1984 25 973. 37 I. Takahashi K. Odashima and K. Koga Chem. Pharm. Bull. 1985 33 3571. 38 Y. Rubin K. Dick F. Diederich andT. M. Georgiadis J. Org. Chem. 1986,51,3270. 39 R. Dharanipragada and F. Diederich Tetrahedron Lett. 1987 28 2443. 40 R. Dharanipragada S. B. Ferguson and F. Diederich J.Am. Chem. SOC.,1988 110 1679. 41 T. M. Georgiadis M. M. Georgiadis and F. Diederich J. Org. Chem. 1991,56,3362. 42 P. P. Castro T. M. Georgiadis and F. Diederich J. Org. Chem. 1989,54,5835. 43 P. P. Castro and F. Diederich Tetrahedron Lett. 1991,32,6277. 44 C. S. Wilcox and M. D. Cowart Tetrahedron Lett. 1986 5563. 45 M. D. Cowart I. Sucholeiki R. R. Bukownik and C. S. Wilcox J. Am. Chem. SOC.,1988,110,6204. 46 T. H. Webb H. Suh and C. S. Wilcox J.Am. Chem. SOC.,1991,113 8554. 47 T. H. Webb and C. S. WilcoxThe University of Pittsburgh unpublished results. 48 C. S. Wilcox T. H. Webb F. J. Zawacki N. M. Glagovich and H. Suh Supramolecular Chem. 1993 in press. 49 C. S. Wilcox J. C. Adrian Jr. T. H. Webband F. J .Zawacki J.Am. Chem. SOC.,1993 in press. 50 J. C. Adrian Jr. and C. S. Wilcox J. Am. Chem. SOC.,1992 114 1398. 51 D. A. Stauffer R. E. Barrans Jr. and D. A. Dougherty J. Org. Chem. 1990,55,2762. 52 N. M. Glagovich and C. S. Wilcox The University of Pittsburgh unpublished results. 53 Y. Murakami 0.Hayashida T. Ito and Y. Hisaeda Chem. Lett. 1992,497. 54 M. A. Petti T. J. Shepodd R. E. Barrans Jr. and D. A. Dougherty J. Am. Chem. SOC.,1988,110,6825. 55 D. J. Cram M. E. Tanner and C. B. Knobler J. Am. Chem. Soc. 1991,113,7717. 56 J. K. Judice and D. J. Cram J. Am. Chem. SOC.,1991,113,2790. 57 J. Canceill L. Lacombe and A. Collet J.Am. Chem. SOC.,1985,107 6993. 58 T. R. Kelly C. Zhao and G. J. Bridger J.Am. Chem. SOC.,1989,111 3744. 59 V. Rotello J.-I. Hong and J. Rebek Jr. J. Am. Chem. SOC.,1991 113,9422. 60 M. C. Etter Acc. Chem. Res. 1990,23 120. 61 S. S. Yoon and W. C. Still J. Am. Chem. SOC.,1993 115,823. 62 J.-I. Hong S. K. Namgoong A. Bernardi and W. C. Still J. Am. Chem. SOC.,1991,113,5111.

ISSN:0306-0012    

DOI:10.1039/CS9932200383

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

The Hydrogen Bond and Crystal Engineeringt Christer B. Aakeroy and Kenneth R. Seddon Department of Chemistry, David Keir Building, The Queen’s University of Belfast, Belfast BT9 5AG, Northern Ireland 1 Introduction The purpose of this review is to present a philosophy of crystal engineering. The chemist is comfortable and familiar with intramolecular bonding; our advanced knowledge of synthetic chemistry (which could almost be considered as the raison d’gtre of the chemist) is constructed around our understanding of the essential principles of covalent bonding. Less well-known and acceptable are the concepts of intermolecular bonds between molecules and/or ions (even the field of supramolecular chemistry has only just established itself),’q2 and our under- standing of the factors which control crystal habit and morpho- logy is rudimentary.The chemist, in designing molecules, rarely turns his attention to the crystalline form which that molecule will adopt in the solid state. The crystal form is usually a matter of serendipity; the ubiquitous occurrence of polymorphism (see Section 4.4)is either ignored or treated as a problem beyond control. We present here our thoughts on the field of ‘crystal engineer- ing’, which has been advanced in recent years by the elegant synthetic work and shrewd topological analysis of Margaret Etter.3 This is a field in its infancy; it is at the interface between a number of demanding disciplines, and has all the challenge and excitement expected of interdisciplinary research.We present the hydrogen bond as a synthetic ‘vector’ for granting topologi- cal control over crystalline form, and hence control over such crucial physical phenomena as optical properties, thermal stabi- lity, solubility, colour, conductivity, crystal habit, and mechani- cal strength. The significance of this area to industry and academia cannot be overstated. 2 The Hydrogen Bond The object of this section is not to define in detail what a hydrogen bond is, nor to exhaustively record experimental techniques for studying it, but to raise healthy questions in the mind of the reader. The field of hydrogen bonding tends to be clouded by preconception and prejudice about the nature, Christer Bjorn Aakeroy, although of Norwegian extraction, was educated in Sweden where he obtained an MSc.at Uppsala University in 1986. His participation in the Sussex-Uppsala exchange programme resulted in further studies at the University of Sussex where he acquired a D.Phi1. in 1990. He remained as a Research Fellow and was then awarded a Scholarship from the Royal Swedish Acad- emy of Science. This enabled him to spend nine months as a Research Fellow at the Univer- sity of Minnesota where he worked in the laboratory of the late Professor Etter. He has recently accepted a Lecture-ship in Inorganic Chemistry at the Queen’s University of Bevast. 397 strength, occurrence, and importance of the hydrogen bond. It is hoped that, by the end of this review, the reader will not dismiss it as a weak bond of relatively marginal importance to material chemists.2.1 What is a Hydrogen Atom? This is not a rhetorical question, and nor are we the first to raise it. The following is quoted verbatim from a recent paper by Cotton and ‘There is a kind of conventional wisdom that neutron diffraction finds hydrogen atoms better than X-ray diffraction does. But is this even a meaningful statement, let alone a true one? It can be argued that it is not meaningful and thus incapable of being true. The simple facts are that neutrons and X-rays see two different parts of the hydrogen atom and that these parts do not coincide. It is then a Solomonic question whether either technique is justifiably considered to ‘see’ the hydrogen atom.The neutron experiment sees, with considerable accuracy (ca. f0.001 A), the location of the hydrogen atom’s nucleus, the proton. In a very favourable case [.. .], the X-ray experiment sees, with less accuracy (ca. f0.02 A), the hydrogen atom’s electron cloud. Which of these is ‘thehydrogen atom’? Both the nucleus and the electron density of an atom are essential parts, and it is therefore impossible to assert rationally that the position of either the one or the other is ‘the’ position of the atom.’ Only Cotton has the standing, insight, and gall to ask questions like this in a manuscript primarily concerned with the crystal and molecular structure of {diethylbis( 1-pyrazoly1)borato)allyldi-carbonylmolybdenum(II)! It is a pity that this manuscript, principally of interest to organometallic chemists, may not attract the universal readership that it deserves.The question raised is of fundamental importance, and should be considered carefully and seriously by all chemists, especially those with a + This review IS dedicated to the memory of Professor Margaret C. Etter, whose contributions to the field of crystal engineering have been of inestimable value; her vital enthusiasm and inspired insights will be sadly missed. ~~ Professor K. R. Seddon was appointed to the Chair of Inorganic Chemistry at the Queen’s University of Bevast earlier this year; he was previously Reader in Experimental Chemistry at the Univer- sity of Sussex. His principal research interests include chemistry in ionic liquids, novel materials for non-linear optics, crystal engineering, coordina- tion chemistry (especially of the metal halides), and the application of chemical tech- niques to the ident$cation of dyes and mordants in ancient papers and textiles.He is co-author of a book ‘The Chemistry of Phosgene and Related Carbonyl Halides ’ which will be published in 1994. central interest in hydrogen bonding. If there is not universal agreement about the absolute positon of a hydrogen atom in a molecule or lattice, how can serious qualitative and quantitative studies of its bonding be made? In other words, if we are not sure what we mean by a hydrogen atom, how can we enter detailed discussions and calculations on hydrogen bonding? 2.2 What is a Hydrogen Bond? Again, this is not a rhetorical question: currently, a definitive answer does not exist.Ideas of what constitutes a hydrogen bond are in a constant state of flux. The following quotes, arranged chronologically, may illustrate the nature of the problem of producing even a simple definition: Latimer and Rodebush (1920) ‘Water 1.. .] shows tendencies both to add and give up hydrogen, which are nearly balanced. Then [. . .] a free pair of electrons on another water molecule might be able to exert sufficient force on a hydrogen held by a pair of electrons on another water molecule to bind the two molecules together. [. . .I. Indeed the liquid may be made up of large aggregates of molecules, continually breaking up and reforming under the influence of thermal agitation.[. ..I. Such an explanation amounts to saying that the hydrogen nucleus held between two octets constitutes a weak ‘bond’.’ Pauling (1940)6 ‘It has been recognized in recent years that under certain conditions an atom of hydrogen is attracted by rather strong forces to two atoms, instead of only one, so that it may be considered to be acting as a bond between them. This is called the hydrogen bond. It is now recognized that [. ..]the hydrogen bond is largely ionic in character, and is formed only between the most electronegative atoms. [. . .]. Although the hydrogen bond is not a strong bond (its bond energy [. . .] being only about 5 kcal/molef.), it has great significance in determining the properties of substances.’ Pimentel and McClellan (1960)’ ‘A hydrogen bond exists between a functional group A-H and an atom or a group of atoms B in the same or a different molecule when: (a) there is evidence of bond formation (association or chelation), (b) there is evidence that this new bond linking A-H and B specifically involves the hydrogen atom already bonded to A.’ Zeegers-Huyskens and Huyskens (1991)8 ‘Specific interactions are short-range site-bounded cohesion forces that considerably weaken a given chemical bond of one of the partners.Hydrogen bonding constitutes a particular case of specific interactions where the weakened chemical bond involves a hydro- gen atom and a more electronegative one (in general 0, N.S, halogens).’ Attempts at simpler explanations seem doomed to failure. The following quotation could form the basis of a critical finals examination question: Atkins (1989)9 ‘A hydrogen bond is a link formed by a hydrogen atom lying between two strongly electronegative atoms.’ This has the seductive appeal of appearing correct at first sight, but being in error in almost every detail and at every level of understanding. The problems of defining the hydrogen bond are manifest. All too frequently, current descriptions of hydrogen bonds include phrases which refer to them as ‘involving hydrogen bonded to an electronegative atom’, ‘thermodynamically weak’, or ‘essen- tially ionic in nature’.As can be seen, these hark back to Pauling,6 a definition that was insightful and visionary when proposed, but now should be viewed with the hindsight of fifty years of chemical progress: although many hydrogen bonds do CHEMICAL SOCIETY REVIEWS, 1993 fall within Pauling’s definition, it is now too restrictive, and precludes many examples of intermolecular bonding now universally accepted as hydrogen bonding (e.g. C-H * *. 0). Many definitions are empirical, usually boiling down to ‘a hydrogen bond exists where there is evidence that it exists’. Theoretical descriptions of the hydrogen bond are rapidly improving, but are extraordinarily sensitive to details of the basis set, and to electron-correlation effects: there is not yet general agreement as to whether, in weak hydrogen bonds, D-H -A, there is any significant electron density in the H -* A bond (i.e.is a hydrogen bond essentially electrostatic in nature, or is there a significant covalent contribution). This review is not the correct place to extend this controversy, fascinating though it is. A detailed, multidisciplinary study of this area is greatly needed -an updated and expanded version of the seminal volumes edited by Schuster et al. O is long overdue. Perhaps the last words in this Section should go to Samuel Butler: ‘A definition is the enclosing a wilderness of idea within a wall of words.’ Notebooks (1912) We will thus not attempt a formal definition here, but draw from the descriptions above as appropriate, recognizing that the value of the hydrogen-bond concept lies in the wilderness of idea, and not within the wall of words.2.3 Occurrence of Hydrogen Bonds Hydrogen bonds occur between atoms, molecules, or ions (positive or negative) in the gas, liquid, solid, or supercritical phases. Hydrogen bonds may be simple (involving only one donor and one acceptor), bifurcated (three-centre), or trifur- cated (four-centre) (set: Figure 1). Some hydrogen bond donors and acceptors are given in Table 1. However, there are now examples of aliphatic methylene protons acting as hydrogen bond donor^,^ and transition rnetal~,~ alkenes,’ alkynes,’ and aromatic n-clouds’ acting as hydrogen bond acceptors. More- over, it could be convincingly argued that even the hydrogen atom is not essential to a hydrogen bond, and that lithium could be considered to enter into multi-centred bonds which could be described, in the wilderness of idea, as ‘hydrogen bonds’.I3 Figure 1 Common hydrogen bond arrangements: note that the simple bond is rarely linear. 2.4 Characterization and Effects of Hydrogen Bonding Let us consider the simple, and most common, arrangement for a hydrogen bond (see Figure 1).In this description, where r(D-H) is shorter than r(H A), the element D is referred to as a hydrogen bond donor, and the element A as a hydrogen bond acceptor. Most hydrogen-bond acceptors (see Table 1) have one feature in common: they formally possess a lone-pair of elec-trons in conventional formalisms. Some unusual acceptors, such as transition metals, alkenes, and aromatic r-clouds, all have centres of high electron density (an occupied dz2orbital, or the 7~-molecular orbitals).Similarly, the elements, D, of the hydrogen THE HYDROGEN BOND AND CRYSTAL ENGINEERING-C. B. AAKEROY AND K. R. SEDDON Table 1 Some hydrogen bond donors and acceptors Donors Acceptors C-H C=C, CK, arenes N-H N P-H P 0-H 0 S-H S F-H F C1-H c1 Br-H Br I-H I bond donors (or, to be more precise, the functional groups or moieties of which D is a part) have the effect of removing electron density from the hydrogen atom, leaving it with a significant partial positive charge. Note, however, that this has little to do with our outmoded ideas of electronegativity (a concept that is perhaps best left for heated tutorial debates about the validity of imprecise definitions, and about the use of terms such as 'atoms within molecules'), and a lot more to do with the overall electronic structure of the molecule of which D-H forms a part.Thus, methane does not readily form strong directional hydrogen bonds (although methane activation WILL occur via a mechanism involving initial hydrogen bond formation, when a viable system is discovered), whereas all the C-H ring protons of the imidazolium cation form three-dimensional, structure- determining hydrogen-bonded networks. ' Clearly, then, a hydrogen bond cannot be defined in terms of the elements which might partake in it.Nevertheless, certain elements and functional groups exhibit a higher propensity than others to form hydrogen bonds, and it is these (which were, of course, the earliest recognized and most easily detected) which formed the basis of the Pauling definition (see Section 2.2). So, how is a hydrogen bond detected? Its nett effect, in the system D-H-*.A is to weaken the D-H bond (compared with D-H in an isolated system), and this is the basis of the Zeegers- Huyskens and Huyskens definition (see Section 2.2).* Thus, a wide ranging collection of spectroscopic, structural, and ther- modynamic techniques can be used to study the nature of hydrogen bonding, the most common being IR and NMR spectroscopy, and single-crystal X-ray diffraction (see Section 3).The use of these techniques, and many others, has been extensively reviewed el~ewhere.~.*~'~ 2.5 Strength of the Hydrogen Bond The thermodynamic strength of a hydrogen bond is, as might be expected, extremely ~ariable.~~*~'~ For neutral molecules, it normally lies in the range of 10-65 kJ mol- l, being greater than that found for van der Waals interactions (<8 kJ molt l), but weaker than conventional covalent bonds. However, when one component of the hydrogen bond is ionic, the range of bond strengths rises to 40-190 kJ mol-'. In order to place the strength of a hydrogen bond in perspective, a summary of some typical 0-bond strengths is presented in Figure 2, along with some of the stronger characterized hydrogen bonds (between an ion and a neutral molecule).8.1 It is clear that a strong hydrogen bond is energetically on a par with a weak covalent bond.3 Hydrogen Bonding in Crystals 3.1 Philosophy In many areas of chemistry, an X-ray single-crystal structure determination of a novel compound represents the solution to a particular problem. or the end of a specific project: the cynosure is usually the identity of the molecule itself, or some particular feature within its molecular structure. An alternative view, however, would be to treat the structural information as the beginning of a new venture, leading to questions of far reaching Figure 2 Bond epergies for a range of common 0-bonds compared with a number of the stronger hydrogen bonds. and fundamental importance regarding the interrelationships between molecules and ions in the solid state.Since the crystal structure represents a situation where all the bonding and non-bonding forces are poised at an energetic minimum (not necessarily a global minimum!), it contains all the information regarding the importance of, and balance between, intermolecular forces. If this information could be extracted and deconvoluted, then prospects of designing materials with speci- fic properties would be vastly enhanced. Consequently, it is of great importance to improve our understanding of the forces that determine the structures of crystalline materials, and single- crystal data provide a natural starting point. 3.2 Hydrogen-bond Geometry in Crystals The existence of hydrogen bonds in solids is often detected and determined purely by applying geometric criteria.7,1 These criteria, when based solely upon estimated van der Waals radii, are somewhat controversial,' 'and there have been several suggestions as to which radii are 'c~rrect'.~~' Depending8,19 upon which set is chosen, certain interactions may, or may not, be regarded as 'legimate' hydrogen bonds. This is clearly not an ideal situation, but there are no better ways to discriminate between very weak hydrogen bonds and close contacts gener- ated by lattice forces, especially since this question is difficult to answer unambiguously with current experimental techniques. Nevertheless, several important publications containing tabulated data and statistical analyses of the known geometries of a wide range of hydrogen-bonded materials are avail-These studies, in combination with the information available in the Cambridge Structural Database,? constitute a vital base from which our understanding of solid-state hydrogen bonds can be developed.Many surveys have correlated geometries of hydrogen bonds in the solid state with the nature and environment of the donor and acceptor groups, and some important trends have been identified.7.10,1l.l It has been demonstrated that longer, weaker hydrogen bonds are more likely to deviate from a linear arrange- ment. Furthermore, it is more common for N-H-..O bonds to deviate from linear arrangements than it is for 0-H -0bonds, even when their bond distances are similar.In addition, the covalent D-H bond is also found to be influenced by the H * -* A t Details avadable from Dr Olga Kennard, Cambridge Structural Database, Cambridge Cry3tallographic Data Centre, CAMBRIDGE CB2 IEW, U K distance: a shorter H A bond will lead to a longer D-H bond, as shown in Figure 3. lo20t OO\ 0.90‘ ‘ 1.2 1.4 1.6 1.8 2.0 2.2 H-*O (A) Figure 3 0-H distance as a function of H.--Odistance in a O-H.*-O system.(Reproduced with permission from reference 10.j In recent years, the importance, and frequency, of relatively weak interactions have been widely recognized, e.g. it has become clear that C-H X hydrogen bonds (where X = F, 0, N, C1, Br, or 1)14316320 may be of significant importance to the organization of the solid-state. Situations where C-H groups are found to participate in C-H*..X hydrogen bonds, are particularly common when they are found adjacent to a nitrogen atom (aliphatic or aromatic) and, hence, such hydrogen bonds may be very important in amino acid and nucleoside chemistry.3.3 The Influence of Hydrogen Bonding on the Lattice Energy of Crystals 3.3.1 Introduction Before the hydrogen bond can be ‘employed’ as a regiospecific, structure-controlling agent (i.e. as a synthetic vector), the ener- getic contribution made by hydrogen bonding to the lattice energy of a crystalline ionic material must be evaluated. Unless the hydrogen bond is seen to make a significant energetic contribution, the underlying assumptions about its usefulness in crystal engineering will be invalidated.The lattice energy, U,of an ionic crystalline material, MX, is often defined as the energy change associated with the process described by equation 1. The lattice energy of a solid is generally assumed to receive contributions from four main components: electrostatic, Ec; repulsive, E,; dispersive, Ed;and zero-point energy, E,, (equa-tion 2). The main dilemma with lattice energy calculations (see Section 3.3.3) results from the problems associated with the evaluation of the repulsive and the dispersive contributions. Such calculations are normally based upon extensive empirical parameterizations and, consequently, accurate lattice energy calculations require substantial effort.3.3.2 Experimental Lattice Energies Direct measurements of lattice energies are not feasible, but it is possible to relate the lattice energy of an ionic compound to various measurable thermodynamic quantities using a Born- Haber cycle, such as that illustrated for a Group 1 halide in Figure 4,where A HVMand AHDxare the enthalpies of vaporiza- tion and dissociation for the respective elements M and X, AHIE CHEMICAL SOCIETY REVIEWS, 1993 Figure 4 A typical Born-Haber cycle for a Group 1 halide. is the ionization energy, -AHE*is the electron affinity, AHfis the enthalpy of formation of the crystalline salt, and U,,, is the lattice energy of MX at 25 “Cand 1 atm. The lattice energy at 0 K, U,, is obtained by combining U,,, with the appropriate heat capacity correction.Contributions arising from the heat capacities, C,, of the species involved are normally small enough to be ignored. It is worth noting that values for lattice energies obtained from a Born-Haber cycle are often referred to as ‘experimental’ data, Uexp,even though these data have only been deduced from the experimental values for the various steps involved in the cycle. The accuracy of ‘experimental’ lattice energies is therefore determined by the reliability of the available data for the enthalpies of ionization, vaporization, etc. Consequently, ‘experimentally’ determined lattice energies are not necessarily accurate and, in many cases, a situation exists where only simulations/calculations can give an indication of the lattice energy of a material.3.3.3 Lattice Energy Calculations Recent years have seen considerable interest in nonlinear optical materials, notably those capable of second harmonic generation (SHG).,l Such materials, and the basis of their properties, have attracted both academic and commercial investigators. The structural investigation of a series of SHG-active dihyd- rogenphosphate salts of organic cations,, prompted our own interest in the role played by hydrogen bonding in the crystal structures of such salts, specifically concerning the energetic contribution made by hydrogen bonding to their lattice energies. Further, from the discussion in Section 3.3.2, it should be clear that accurate lattice energy calculations require much time and effort, and there would appear to be little point in developing separate potential models for all pair-wise interactions within these materials, if appropriate simplifications would yield an approximate value.Furthermore, since it is also impossible to assign an accurate value to the energy associated with each hydrogen bond in the solid state, the quality of the comparison will always be limited by estimates of hydrogen bond strengths. Hence, to a first approximation, it was assumed that in these ionic materials, the electrostatic forces would dominate, and the dispersive and repulsive forces would be of equal magnitude and, hence, would cancel each other.This approximation sim- plifies the lattice energy calculation significantly, as it has now been reduced to an Ewald summation of point charges. In the Ewald method, each point charge is replaced by a Gaussian charge distribution at the appropriate lattice site, resulting in a smoothly varying charge distribution, which leads to a quickly converging series. As the charge of an ion is a periodic function throughout the lattice, the Coulombic potential can be eva- luated using a Fourier transformation of the charge. A justification for the approximations adopted in this approach was provided by calculating the lattice energy, Ucal,for two salts with experimentally determined lattice energies, [NH4],[SnC16] and [NH,],[ReCl,]: the results are presented in Table 2, and justify the approximation that the dispersive and repulsive forces cancel.This approach has also been validated by extensive work carried out by Lubkowski et al. on a range of halide salts of mononitrogen bases.23 THE HYDROGEN BOND AND CRYSTAL ENGINEERING-C. B. AAKEROY AND K. R. SEDDON 40 1 Table 2 Lattice energies (kJ mol- l) and atomic charges for [NH41Z[MC161a M ZClh ZH' UCd uexp A Ud Sn -0.66 0.35 1329 1334 -t0.4% Re -0.56 0.35 1397 1390 -0.5% Energy terms are defined in the main text. zcI = atomic charge on chloride (H. D. B. Jenkins and K. F. Pratt, Adv. Inorg. Chem. Radiochem., 1979,22, 1). zH = atomic charge on hydrogen (A. Pullman and A. M. Armbruster, Int. J. Quant. Chem.Symp., 1974, 8S, 69). AU = lOO(U,,, -Ucal)/Uexp. 3.3.4 Lattice Energies of Organic Dihydrogenphosphates As the organic salts of dihydrogenphosphate consist of structur- ally complex ions without either a symmetrical, or a centralized, charge distribution, it was necessary to perform ab initio calcula-tions on the participating ions in order to obtain the desired information; charges obtained with the STO-3G basis set were used in these lattice energy calculation^.^^ Following the method outlined in Section 3.3.3, the lattice energies of a series of organic dihydrogenphosphate salts were calculated and compared with the strength of the hydrogen bonds present in each structure (Table 3).24 Table 3 Minimum calculated hydrogen bond contributions (kJ mol-I) to the total lattice energy of four dihydrogenphosphate salts, [AH][H,PO,] A Piperidine 3-hydroxypyridine 3-hydroxy-6-methylpyridine 4-h ydrox ypyridine uc/ EHB~ Utot~ hd 500 130 630 21% 545 135 680 20% 410 135 545 25% 515 135 650 21% (1 [Calculated lattice energy.h Hydrogen bond energy. Total lattice energy, L'cA + EHB loo(EHB/~IoI). Based upon extensive experimental and theoretical data,2 each 0-H -0interaction was assigned an energy content of 35 kJ mol-I, and each N-H interaction was assigned a value -*a0 of 30 kJ mol-' (both values are significantly lower -by a factor of two -than those found experimentally and theoretically for ionic hydrogen bonds of this type). By adding these values for each structure to give a total hydrogen bond energy, EHB, depending upon the number of hydrogen bonds, and comparing this with the total lattice energy, an estimate of the minimum contribution made by hydrogen bonding is obtained (Table 3).It should be pointed out that some of the hydrogen bond energy (i.e. the electrostatic part) has, in reality, already been included in the calculated lattice energy, Ucal.Hence, an incor- rect value (i.e. too large) for the total lattice energy U,,, will be the result if Ucal and EHB are simply added together. As no attempt has been made to estimate the relative magnitude of the electrostatic part of the hydrogen bond energy, it is not possible to obtain an accurate value for the contribution made by hydrogen bonding to the total lattice energy.However, the minimum level of contribution can be estimated. The results displayed in Table 3 show that hydrogen bonding provides a notable energetic contribution, 20-25%, to the total lattice energy of dihydrogenphosphates of organic cations. It is also likely that the true values are significantly higher than that, given the significant underestimation of EHB,and the overesti- mation of U,,,. Consequently, hydrogen bonding can act as a regiospecific, controlling and directing, structural tool in the crystal engineering of ionic and molecular solids. 3.4 Effects of Hydrogen Bonding on Charge Density The fundamental nature of chemical bonding and molecular structure is determined by the electron density distribution in the system.Information regarding this feature can be made avail- able through quantum mechanical calculations, but it can also be observed experimentally using high-intensity X-ray diffrac- tion measurements. * By combining such experiments with neutron diffraction data, it is possible to gain added insight into the redistribution of electron density that occurs upon bond formation, including hydrogen b~nding.~ Such X-N maps, based upon structure factor amplitudes and thermal parameters from neutron data, can furnish information about the deviation of electron density around the nuclei from spherical symmetry. Electron-density distribution studies of hydrogen-bonded complexes can be divided into two groups, (a) those involving weak or moderately strong hydrogen bonds, and (b) those containing strong hydrogen bonds.This division is made because of the presence of some distinctly different features within the two groups. The observable changes in electron distribution that take place in materials with weak, or moderately strong, hydrogen bonds are close to the limit of current experimental techniques, but many studies have obtained a good agreement between X-N maps and theoretical deformation maps, both showing the general features of such hydrogen bonded systems. These features are, commonly, a build up of electron density in the D-H bond, and in a space close to the acceptor atom in the H..-A interaction, which is accompanied by a decrease in the electron density closer to the hydrogen atom in the D-H...A bond,zs see Figure 5.I I I I I I I I I H.1 C\/ H" I 'IH ----H Figure 5 The difference density map for glycine, between a molecule in the crystal field and a free molecule. A schematic (right) shows the hydrogen-bond pattern. (Reproduced with permission from reference 28.) However, the same features can be obtained, qualitatively, by superimposing deformation maps of the isolated, unperturbed molecules in the system. Consequently, a weak or moderately strong hydrogen bond can be represented by a simple electro- static model, since contributions from charge transfer and exchange repulsion effects are significantly smaller and tend to cancel each other.The accuracy of this description is also supported by the fact that the strength of these hydrogen bonds (the total hydrogen bond energy) correlates well with the electrostatic energy of the interaction. Detailed theoretical cal- culations have indicated, though, that some electron migration does take place as a second-order effect.266 The situation regarding strong hydrogen bonds [most studies have been performed on systems with short O-H.--O bonds, with r(O***O)in the range of 0.24-0.25 nm], however, is more complex. Normally, the charge distribution is more symmetri- cally arranged around the centre of the hydrogen bond and there is less charge build up both in the 0-H bond and in the H...O region, see Figure 6.29It also becomes much more difficult to separate contributions from various components, e.g.polariza-tion vs. charge-transfer effects, as well as trying to partition the electron distribution arising from the different molecules/ions in the complex. Another question regarding electron distribution and hydro- gen bonding, which has attracted much attention, concerns the role played by the lone-pair as a hydrogen-bond acceptor. Both CHEMICAL SOCIETY REVIEWS, 1993 Figure 6 The experimental deformation density for u-C,O,H, 2H,O in the plane of the oxalic acid molecule (schematic top) (Reproduced with permission from reference 29 ) theoretical and experimental studies have been carried out in order to clarify the stereoelectronic effect caused by lone-pair regions around acceptor atoms in hydrogen-bond interactions, and the main issue has been whether, or not, it is possible always to view a lone-pair as the distinct acceptor of a hydrogen bond Theoretical studies of the electron density around C=O groups normal!y reveal a double maximum of electron density corresponding to the lone-pair regions, although it should be pointed out that because of the very nature of these deforma- tion-density calculations, only the difference between the mole- cular electron density and the superpositioning of spherical densities is evaluated, not an absolute location of excess electron density due to a specific orbital Attempted correlations of solid-state bond angles and lone- pair directionality hake not been completely conclusive lo l6 An extensive study of neutron-diffraction data of C=O * * -H angles found an accumulation of data around 120",albeit with several large variations Other studies have also highlighted the import- ance of correlating such data with steric factors, which may otherwise bias the existing data On the whole, however, double bonded oxygen atoms in certain groups, -C(O)OH and -C(O)O-, appear to favour a hydrogen bond approach at an angle of 120" In contrast, theoretical investigations of the water molecule and hydroxy groups normally only reveal one broad electron- density maximum, and a study of a large number of crystalline ROH found a tendency for the acceptor atom to occupy a position in the plane of the lone-pairs, although there was no accumulation of acceptor atoms in the tetrahedral direction within the plane Overall, there is no absolute correlation between lone-pair regions and acceptor angles and, hence, it would seem somewhat simplistic to view the 'classical' lone-pairs as specific receivers of hydrogen bonds Instead, it is probably more fruitful, albeit more complex, to investigate both steric factors and the whole electrostatic potential energy surface, in order to elucidate the preferred approach for a hydrogen-bond donor 4 Crystal Engineering 4.1 Rationale Within the fields of supramolecular chemistry, molecular recog- nition, and crystal engineering, it has been recognized that hydrogen bonding is an indispensable tool for designing molecu-lar aggregates 31 It is our firm belief that an improved understanding of hydrogen bonding in general, and hydrogen bonds in ionic solids in particular, can strengthen our awareness of the tools with which we may be able to 'persuade' molecules and ions to form specific aggregate structural units Since the properties of a solid are critically dependent upon its structure, it is feasible to design materials with desired characteristics by incorporating specific properties into the subunits There are clearly numerous diverse areas where such an approach could be very fruitful, e g studies of biomolecular substrate binding and recognition, reactivity and catalysis, development of improved detergents, the petrochemical industry, and the synthesis of novel nonlinear optical and ferroelectric materials 4.2 Pattern Recognition Probably the most well known, and most easily identifiable, hydrogen-bonded aggregate is a monocarboxylic acid dimer (Figure 7) Several groups have carried out extensive studies of solid-state structures of a wide range of related molecular and ionic solids, and these studies have led to the realization that hydrogen-bond aggregation is not random patterns do exist Certain functional groups, and ions, display a clear pattern preference, and this insight will have important consequences for the development of crystal engineering If certain molecular building blocks tend to crystallize in specific, energetically favourable, arrangements, then molecules containing these blocks can be encouraged to form aggregates with specific structural features Figure 7 The common hydrogen-bonded dimeric form of a carboxylic acid Lehn and co-workers have studied, inter aha,macromolecular recognition involving cryptands and other macrocyclics, inter- actions between phosphates and carboxylates with polyammo- nium macrocyclics, the formation of helical metal complexes of bipyridine strands, and supramolecular liquid crystalline polymers Other groups have investigated molecular tweezers, organometallics, and small organic molecules as subunits of extended structures 32 Although all of these studies fall into the broad categories of molecular recognition or crystal engineer- ing, the discussion in the remainder of this review will be focused upon relatively small organic/inorganic systems in the absence of preorganized, custom-designed cavities Such systems are more amenable to an examination and evaluation of hydrogen- bond selectivity, preference, and topology, without an a priorr directional bias The most extensive studies of organic hydrogen-bonded solid- state structures have been carried out by Etter and co-workers 33 Their work has identified the pattern preference displayed by a range of functional groups and molecular types, eg amides, diary1 ureas, imides, nitroanilines, and 2-aminopyrimidines A THE HYDROGEN BOND AND CRYSTAL ENGINEERING-C. B.AAKEROY AND K. R. SEDDON detailed description of some of these patterns will be found in Section 4.3.2.As a result of this work, a method for describing hydrogen-bond patterns has been developed, in addition to several general rules (guide-lines) regarding expected hydrogen- bond organization in organic solid-state stru~tures.~ Ionic compounds, both inorganic-organic salts and pure organic salts, have also been studied recently. A series of organic salts with the dihydrogenphosphate anion, [H,PO,] -,displayed certain specific, reoccurring, arrangements of the anionic network; the anions were lined together by short, strong hydro- gen bonds into specific chains (Figure S), sheets, or three- dimensional infinite arrangements.22 It is likely that the char- acter of the cation, and the nature of its hydrogen-bond interac- tions with the anionic matrix, will discriminate between the three possible structures.Similarly, a structural study of organic salts of hydrogen-~-tartrates~, showed that the anionic network was remarkably consistent, regardless of the nature of the cation. The strongly hydrogen-bonded sheet of anions created a matrix, see Figure 9, which dominated the structure and only allowed limited flexibility regarding the positioning and packing of the cations.34 The presence of rigid, frequently occurring, hydrogen-bonded aggregate structures in both molecular and ionic compounds, creates a necessary platform for intermolecular synthesis. By identifying, classifying, and rationalizing such networks, it may be possible to utilize them as active design tools in the crystal engineering of novel materials with specific structural features.4.3 Encoding Hydrogen-bonded Networks 4.3.I Pattern Designation Once it has been established, via crystallographic data, that there are several different hydrogen bonds within a material, how can this information be translated into a form, or code, which allows for the classification and recognition of hydrogen-bond interac- Figure 8 Chains of dihydrogenphosphate anions within the crystal structure of the piperidinium dihydrogenphosphate.22 Covalent bonds, red; hydrogen bonds, yellow and green. Figure 9 A two-dimensional sheet of hydrogen-L-tartrate anions, as found in piperazinium(2-t) bis-hydrogen-~-tartrate.~~Covalent bonds, red; hydrogen bonds, yellow and green.tions as aggregate structures? Furthermore, how can this infor- mation be communicated in a precise fashion? Clearly, in order to achieve this objective, it is essential to find a way of describing even complex hydrogen-bond patterns in a simple, yet compre- hensive, language. The most successful method for encoding hydrogen-bond patterns of organic solids has been developed by Margaret Etter and co-worker~,~ who have demonstrated that, by analysing hydrogen-bond interactions in organic molecular solids in a systematic and consistent fashion, it is possible to establish the pattern-preference displayed by many functional groups. The methodology, loosely based upon graph-theory, adopts a topo- logical approach to analysing hydrogen-bond patterns, but instead of viewing molecules as points and hydrogen bonds as lines, chemical structure and functionality have been retained.A graph set is specified using the pattern designator (G), its degree (n)and the number of donors (d)and acceptors (a): G, the descriptor referring to the pattern of hydrogen bonding, can be either S (for intramolecular bonds), C (for infinite chains), R (for intermolecular rings), or D (for non-cyclic dimers and other finite structures), and the parameter n refers to the number of atoms in a ring, or the repeat unit of a chain. Graph sets are assigned initially to motifs (hydrogen-bond patterns constructed by only one type of hydrogen bond), and then to higher-level networks (combinations of the relevant motifs).Thus, if the compound contains four different hydrogen bond types, then the first-level graph set, N, ,is a sequential listing of these four motifs. Higher-level graph sets are assigned to networks generated by combinations of different hydrogen bond types (e.g. the second-level graph set is created by combi- nations of two hydrogen-bond types, the third-level set by, combinations of three etc.). 4.3.2 Encoding in Practice The method outlined in Section 4.3.1 will now be demonstrated with a series of examples, illustrated in Figure 10, which shows the four fundamental motifs that may be generated by a hydrogen bond; dimers (D),chains (C),intramolecular bonds (S),and rings (R).Figure 10 Some examples of assigning graph sets to simple hydrogen- bonded ~ysterns.~ Examples of higher-level graphs sets are presented in Figure 11, where several hydrogen bonds are combined to form specific patterns. The compounds depicted in Figure 11 each contain two unique hydrogen bonds and, therefore, their first-level graph sets, N,, contain two motifs. The second-level graph sets, N2,describe patterns that are created by combining the unique hydrogen bonds within each structure. By applying this procedure in a systematic fashion, it is possible to characterize and recognize patterns that appear within crystals containing fundamentally different, as well as closely related, molecules. By correlating pattern preference with functional groups or molecular types, certain specific aggregate structures may be induced to form, and may be incorporated into a material by the introduction of a specific functionality.This encoding technique has also been employed in the analysis of closely related hydrogen-bond networks in poly- morph~,~where subtle differences in various hydrogen-bond arrangements can be very difficult to detect and describe. Furthermore, the realization that specific hydrogen-bonded aggregates occur very frequently should also influence the way in which we investigate and approach topics like protein recogni- tion and nucleation processes, as the ‘active’ species involved in such interactions may be hydrogen-bonded aggregates, and not isolated molecules or ions.CHEMICAL SOCIETY REVIEWS, 1993 Figure 11 Some examples of assigning graph sets to more complex hydrogen-bonded sy~tems.~ 4.3.3 Hydrogen-bond Directed Co-crystallization Etter and co-workers have employed co-crystallization tech- niques extensively, in order to identify competition in hydrogen- bond accepting/donating capability between different molecular types and functional group^.^ The co-crystals were prepared not only by traditional meth- ods (involving evaporation of a solution containing two, or more, components), but also by grinding the starting materials together in the solid-state. A prerequisite for the formation of co-crystals is that the heteromeric species, the co-crystal, con- tains stronger hydrogen-bonds than either of the homomeric starting materials.In addition, formation of a co-crystal from solution is obviously hampered if there is a large solubility differences between the various starting materials. Figure 12 shows some common hydrogen-bond motifs that have been identified in co-crystals of diarylurea derivatives (top) with a wide range of hydrogen-bond acceptors (e.g. triphenylphos-phine ~xide).~ Each pattern is identified with the relevant graph- set notation, described in Section 4.3.1. 4.4 Polymorphism Polymorphism is a unique feature of the solid-state, but although this phenomenon has been known for more than a century, surprisingly little use has been made of it in solid-state studies. The true extent of polymorphism among crystalline materials is very difficult to gauge.However, as it has been more often encountered in areas of research where structure is of paramount importance, we are in strong agreement with M~Crone~~who stated (in 1965): ‘[. ..] every compound has different polymorphic forms [. ..] and the number of forms known for a given compound is proportional to the time spent in research on that compound.’ It is generally accepted that differences in lattice energies between different polymorphs are in the region of 5-20 kJ mol-, and, especially for ionic compounds, these differences are small compared to the total lattice energies of such materials. Consequently, it would be fair to assume that, for example, different methods of recrystallization would be sufficient to induce the formation of a new structural form. If this is true, then THE HYDROGEN BOND AND CRYSTAL ENGINEERING-C. B.AAKERdY AND K. R. SEDDON Figure 12 Some common hydrogen-bond patterns found in diary1 urea crystals and co-cry~tals.~~~ the whole process of identifying and classifying a material, based upon a single-crystal study, may, on its own, be less than satisfactory. When selecting individual crystals for structure determination, it is not necessarily true that the chosen crystals are representative of the structure of the bulk material. In fact, normally, the ‘best’ crystal is chosen -by definition, that must be atypical of the majority of the material. It would therefore seem critical always to compare the structure of the single-crystal with that of the bulk material, obtainable from powder diffraction.By simulating a diffraction pattern from the single-crystal data, and comparing that with the powder pattern recorded experi- mentally on the bulk sample, it is very easy to make an assessment as to the structural purity of, and agreement between, single-crystal and bulk material. We strongly believe that this procedure should be carried out routinely, in parallel with single-crystal structure determinations. The question of structural purity is, in many areas of chemistry, as important as that of chemical purity. Polymorphic systems provide excellent opportunities to study specific chemical entities in different crystalline environments. Hence, information about the interplay between ionic or mole- cular geometry, and intermolecular and crystalline forces, may be extracted from such studies, e.g.the lattice energies of the observed structures can be calculated and evaluated against DSC measurements, the energy differences between observed ionic conformations can be estimated, and an analysis of the results may be correlated to a partitioning of the lattice energies into its individual atomic contributions. A careful investigation of closely related polymorphic salts facilitates a comparison between the relative differences dis- played by measurable thermodynamic quantities (obtainable from DSC techniques), and trends displayed by theoretically determined lattice energies.In this case, experimental and theoretical chemistry are combined into a very powerful probe of intermolecular interactions. Such an approach will also highlight the relative effects and importance of different ionic, or molecular, subunits, on the overall crystalline arrangement of each material. 4.5 Design of Crystals with Specific Structural Features and Properties Although we are still a long way away from being able to predict the precise structure of an unknown material, several imagina- tive efforts have been made at utilizing hydrogen bonding as a means of creating specific structural features. Wuest and co-workers3 have utilized the well known cyclic dimeric motif of lactams, cf. 2-pyridones, in combination with rigid spacers, in order to create new extended structural motifs (Figure 13).Ureylendicarboxylic acids have been employed by Fowler and co-~orkers~~~ as building blocks for several infinite two-dimensional hydrogen-bonded layer structures. Whitesides and co-worker~~~ have synthesized co-crystals of melamin deri- vatives and barbituric acids which contain extended hydrogen- bonded ‘tapes’. Figure 13 Novel structural motifs generated by isomeric derivatives of 2-pyrid0ne.~ Our own work has focused on the preparation of salts with predictable structural features, and recent efforts include the use of infinite sheets of hydr~gentartrate~~ anions as a means of imposing structural consistency within a series of non-linear optical salts (Figure 14 and Figure 15).Furthermore, in an attempt to produce transparent, colour- less, non-centrosymmetric crystals (a non-centrosymmetric medium is an absolute condition for certain non-linear optical effects) with a needle-like morphology (a condition required for the successful incorporation into thermally aligned polymer films),39 3-and 4-hydroxybenzoic acids were treated with a chiral amine, (8-1-phenylethylamine, to produce crystals with the desired proper tie^.^^ It was assumed that the anions would create infinite chains by a head-to-tail hydrogen bond, and that this aggregate structure would manifest itself macroscopically as needle-like crystals. Single-crystal studies of these two materials, both of which had a needle-like habit, revealed the presence of infinite chains of anions (Figure 16 and Figure 17), and, in the case of 1 -phenylethylammonium 4-hydroxybenzoate, the direc- tion of the infinite chain coincided with the long-axis of the needle.40 Further studies are also needed in order to identify correla- tions between microscopic structure and macroscopic appear- ance.Will the regions of strong hydrogen-bonds or the hydro- CHEMICAL SOCIETY REVIEWS, 1993 Figure 16 Infinite chains of anions, parallel to the LI axis, in (9-1-Figure 17 Infinite chains of anions, parallel to the b axis, in (9-1- phenylethylammonium 3-hydroxybenzoate, viewed down b.40 phenylethylammonium 4-hydroxybenzoate, viewed down c.~O Hydrogen bonds indicated by dotted lines.Hydrogen bonds indicated by dotted lines. THE HYDROGEN BOND AND CRYSTAL ENGINEERING-C B AAKEROY AND K R SEDDON phobic regions crystallize more quickly upon precipitation from aqueous solvent’ Answers to such questions are also likely to cast more light on the path towards successful crystal engineering 5 The Future The last five years have witnessed a significantly increased awareness of the importance of being able to understand and rationalize the effects of hydrogen bonding on the solid state As a consequence, several groups, worldwide, have made very important and useful contributions to the field of crystal engi- neering by employing hydrogen bonds as active design elements in the synthesis of novel materials and extended aggregates In the past few years, Zyss and co-workers have developed an independent parallel approach to the design of novel materials for non-linear optics, of combining organic cations (to carry the high optical polarizability) with hydrogen-bonding inorganic (or organic) anions (to provide thermal and structural stability), similar to that which we proposed in 1989 22 The resulting crystalline salt, 2-amino-5-nitropyridinium dihydrogenphos-phate, shows significant SHG-activity 41 Clearly, because of the potential impact that crystal engineer- ing may have on a range of areas with significant commercial interests, we are likely to see substantial advances being made, both theoretical and practical, during the next few years Control of crystalline structure is an ambitious, but achievable, target, and the next decade promises to be extremely exciting Acknowledgments We would like to thank Dr J McDonald (Harvard University) for invaluable comments regarding the graph-sets, and the Royal Swedish Academy of Sciences for generous financial support (to CBA) We are indebted to Drs D Braben dnd D Ray of Venture Research International for nurturing (financially and intellectually) our interest in this field, and to Drs C J Adams, G Armstrong, and J Warr (Unilever Research Laboratory), and the SERC for actively supporting our current research 6 References 1 J M Lehn, Angew Chem Int Ed Engl, 27,89, 1988, J M Lehn, Angew Chem Int Ed Engl , 1990,29, 130 2 J P Mathias and J F Stoddart, Chem Soc Rev, 1992,21,215 3 M C Etter, Ace Chem Res, 1990, 23, 120, M C Etter, J C MacDonald, and J Bernstein, Acta Crjstallugr Sect B, 1990, 46, 256 4 F A Cotton and R L Luck, Inorg Chem , 1989,28, 3210 5 W M Latimer and W H Rodebush, J Am Chem Soc, 1920,42, 1419 6 L Pauling, ‘The Nature of the Chemical Bond and the Structure of Molecules and Crystals -An Introduction to Modern Structural Chemistry’, 2nd Edn ,Oxford University Press, London, 1940 7 G C Pimentel and A L McClellan, ‘The Hydrogen Bond’, Free- man, San Francisco, 1960 8 Th Zeegers-Huyskens and P Huyskens, in ‘Intermolecular Forces -An Introduction to Modern Methods and Results’, ed P L Huyskens.W A P Luck, and T Zeegers-Huyskens, Springer- Verlag, Berlin, 1991, p 1 9 P W Atkins ‘General Chemistry’, Scientific American Books, New York, 1989 10 ‘The Hydrogen Bond Recent Developments in Theory and Experi- ments ed P Schuster, G Zundel, and C Sandorfy, Vols 1-111, North Holland, Amsterdam, 1976 11 M A Viswamitra, R Radhakrishna, J Bandekar, and G R DeSlrdJU, J Am Chem Soc , 1993, 115,4868 12 H S Rzepa, M L Webb, A M Z Slawin, and D J Williams, J Chem Soc Chem Commun , 1991,765, S S Al-Juaid, A K A Al- Nasr C Eaborn, and P B Hitchcock, J Chem Soc Chem Commun , 1991 1482, S S Al-Juaid, A K A Al-Nasr, C Eaborn, and P B Hitchcock, J Organomet Chem , 1992,429, C9 13 M C Etter and G Ranawake, J Am Chem Soc , 1992,114,4430 14 A K Abdul-Sada, A M Greenway, P B Hitchcock, T J Mohammed, K R Seddon, and J A Zora, J Chem SOC Chem Commun , 1986, 1753, A K Abdul-Sada, S Al-Juaid, A M Greenway, P B Hitchcock, M J Howells, K R Seddon, dnd T Welton, Struct Chem, 1990, 1, 391, A G Avent, P A Chaloner, M P Day, K R Seddon, and T Welton, in ‘Proceedings of the Seventh International Symposium on Molten Salts’, ed C L Hussey, J S Wilkes, S N Flegas, and Y Ito, The Electrochemical Society Inc ,Pennington, New Jersey, PV 90-17, 1990, p 98 15 R Yamdagni and P Kebarle, J Am Chem Soc , 1971,93,7 139 16 (a)R Taylor and 0 Kennard, Ace Chem Res , 1984,17,320 (b)R Taylor and 0 Kennard, J Am Chem Soc , 1982, 104,5063 17 G A Jeffrey and W Saenger, ‘Hydrogen Bonding in Biologicdl Structures’, Springer-Verlag, Berlin, I99 1 18 A Bondi, J Phys Chem , 1964,68,441 19 S C Nyburg and C H Faerman, Actu Crjstallogr Sect B, 1985, 41,360 20 G R Desiraju, Ace Chem Res , 1991, 24, 290 21 ‘Nonlinear Optical Properties of Organic Molecules dnd Crystals’, ed D S Chemla and J Zyss, Vols 1 and 2, Academic Press, New York, 1987 22 C B Aakeroy,P B Hitchcock, B D Moyle,andK R Seddon,J Chem Soc Chem Commun , 1989, 1856 23 J Lubkowski and J Blazejowski, J Phjs Chem , 199 1,95,23 1 I.J Lubkowski, P Dokurno, and J Blazejowski, Thermothzm Ac fa, 1991,176, 183 24 C B Aakeroy, M Leslie, and K R Seddon, Struct Chem , 1992.3, 56 25 (a) M Meot-Ner (Mautner), in ‘Molecular Structure and Energe- tics’, Vol 4, ed J F Liebman and A Greenberg, VCH Verlag, New York, 1987, p 72 (6)C A Deakyne, in ‘Molecular Structure and Energetics’, Vol 4, ed J F Liebman and A Greenberg, VCH Verlag, New York, 1987, p 105 26 (a) P Coppens, J Chem Educ, 1984, 61, 761 (b) ‘Electron dnd Magnetization Densities in Molecules and Crystals’, ed P Becker, Plenum, New York, 1980 27 I Olovsson, Croat Chem Acta, 1982,55, 171 28 J Almlof, 8, Kvick, and J 0 Thomas, J Chem Phys , 1973, 59, 3901 29 E D Stevens and P Coppens, Acta Crjstallogr Sect B, 1980, 36, 1864 30 J Kroon, J A Kanters, J G C M VanDuijneveldt-VdndeRijdt, F B Duijneveldt, and J A Vliegenhart, J Mol Struct , 1975, 24, 1975 3 1 G R Desiraju, ‘Crystal Engineering The Design of Organic Solids’, Elsevier, Amsterdam, I989 32 (a) M Haramata and C L Barnes, J Am Chem Soc , 1990, 112, 5655, (6)P J Fagan, M D Ward,and J C Calabrese, J Am Chem Soc ,1989,111,1698,(c)X Zhao, Y -L Chang, F W Fowler, and J W Lauher, J Am Chem Soc, 1990, 112, 6627, (d) F GdrCid-Tellado, S J Geib, S Goswami, andA D Hamilton, J Am Chem Soc , 1991, 113,9265 33 (a)S M Reutzel and M C Etter, J Phys Org Chem, 1992,544, (b)M C Etter and D A Adsmond, J Chem Soc Chem Commun , 1990, 589, (c) M C Etter, Z Urbanczyk-Lipkowska, M Zia-Ebrahimi, and T W Panunto, J Am Chem Soc , 1990, 112,8415, (d)M C Etter, J Phis Chem, 1991,95,4601 34 C B Aakeroy, P B Hitchcock, and K R Seddon, J Chem Sot Chem Commun , 1992, 553 35 J Bernstein, Acta Crystallogr Sect B.1991,47, 1004 36 W C McCrone, in ‘Physics and Chemistry of the Organic Solid State’, ed D Fox, M M Labes, and A Weissberger, Vol 2, Interscience, New York, 1965, p 725 37 (a)Y Ducharme and J D Wuest, J Org Chem , 1988,53,5789, (b) M Simard, D Su, and J D Wuest, J Am Chem Sot , 1991, 113, 4696 38 (a)J H Zerkowski, G T Seto, D H Wierda, and G M Whitesides, J Am Chem Soc, 1990, 112, 9025, (b) G T Seto and G M Whitesides, J Am Chem Soc , 1990, 112, 6409 39 (a) C B Aakeroy, N E Azoz, P D Calvert, M Kddim, A J McCaffery, and K R Seddon, ACS Sjmp Ser ,1991,455.5 16, C B Aakeroy, N Azoz, M Kadim, K R Seddon, and L Trowbridge, Ad\ Muter, 1993,5, 364 40 C B Aakeroy, G S Bahra, P B Hitchcock, Y Patell and K R Seddon, J Chem Soc Chem Commun , 1993, 152 41 R Masse dnd J Zyss, Mol Eng , 1991, 1, 141, Z Kotler, R Hierle, D Josse, J Zyss, and R Masse, J Opt Soc Am B, 1992,9, 534

ISSN:0306-0012    

DOI:10.1039/CS9932200397

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

The Properties of Organic Liquids that are Relevant to their Use as Solvating Solvents Y. Marcus Department of Chemistry, University of Leicester, Leicester LEI 7RH, U.K. * 1 Introduction Organic liquids are characterized by several properties that make them suitable for dissolving and for providing reaction media for various types of solutes. These properties include physical quantities, such as the liquid range (freezing to normal boiling temperatures), vapour pressure, density, refractive index, relative permittivity, etc., that are not further discussed here per se. The more ‘chemical’ properties to be discussed include polarity, ability to form hydrogen bonds, and structur- edness, among others. Linear free energy relationships (LFER) or linear solvation energy relationships (LSER) have been proposed that relate such properties to divers processes in solution: solubility, distribution between two liquids, retention in chromatography, rates of reactions, free energy and enthalpy of equilibria, wavelengths of light absorption, NMR chemical shifts, etc.In most cases, the quantity that describes the intensity or extent of such a process (called XYZ in the following for the sake of generality) depends on more than one solvent property. Of the many expressions that have been proposed for the description of LSERs, one that was found to be very successful is the Kamlet-Taft expression: XYZ= XYZ, + a-a + b.P + S.R* +... (1) where XYZ,, a, b, and s are (solvent-independent) coefficients characteristic of the process and indicative of its sensitivity to the accompanying solvent properties, a is the hydrogen bond dona- tion (HBD) ability of the solvent, p is its hydrogen bond acceptance (HBA) or electron pair donation ability to form a coordinative bond, and X* is its polarity/polarizability para- meter.Further terms (involving products of coefficients and solvent properties) may be added as required for specific pro- cesses. For some processes any of the coefficients XYZ,, a, 6, and/or s may be negligibly small, so that the corresponding terms do not play a role in the characterization of the solvent effects for these processes. The quantities a and fl are solvatochromic properties of the solvents, i.e., they are determined primarily by the energies of the longest wavelength absorption peaks of certain carefully selected probe solutes in the solvents in question, after subtrac- tion of the effect that non-HBD and/or non-HBA solvents Yizhak Marcus received his M.Sc.in chemistry in 1952 and the Ph.D. in I956 from the Hebrew University of Jerusalem. Until 1965 he was with the Israeli AEC, doing radiochemical research. In 1965 he was ap- pointed as Professor of Inor-ganic and Analytical Chem-istry at the Hebrew University. He has since been a visiting professor at universities in the U.S.A., Germany, Japan, and the U.K. He has authored three books and edited several more andpublished 200 research and review papers and book chapters. would have on the probe, determined in separate experiments.They have been designed and given numerical values so that ideally they describe exclusively the HBD and HBA properties of the solvents, not being affected by their other properties, such as polarity, polarizability, tightness of cohesion, etc. The solvatochromatic parameter 7~*,on the other hand, describes a combination of properties, the polarity and the polarizability of the solvents. For certain processes a modification term, -s. d .6, has to be added to equation 1 in order to describe the solvent polarizability correctly, where 6 = 1.O for aromatic solvents, 0.5 for polychlorinated (polyhalogenated?) aliphatic solvents, and 0 for all other aliphatic solvents, and 0 5 d 5 0.4, depending on the process. This is a less desirable feature of the parameter T*.A host of other solvent parameters have been proposed over the years to express solvent properties in this context. Some of these were called ‘polarity indices’, others ‘donor-’ and ‘accep- tor-numbers’, etc. Survived and of widespread use are many of these, including Dimroth and Reichardt’s ET(3O),* Kosower’s Z,3Mayer and Gutmann’s AN,4 Gutmann’s DIV,~and Swain et al.’s Acity and Basity (their symbols A and B are not employed here, to avoid confusion with other uses of these letters),6 to mention but a few that describe various aspects of polarity and donor-acceptor behaviour. Also important with regard to the solvation abilities of the solvents are physical properties such as Hildebrand’s solubility parameter 6H,’ and the relative permitti- vity (dielectric constant) E, the dipole moment p, and the refractive index n, among others.These quantities have been determined for a large number of solvents, whereas most other quantities are known for a limited number only. There are several computational methods for relating experi- mentally observed quantities XYZ to solvent properties accord- ing to equation 1 or to equivalent expressions employing differ- ent solvent parameters. One is stepwise multiple linear regression (SMLR), where solvent parameters are offered one by one to the statistical computer program, being accepted, rejected, or exchanged until certain statistical criteria are met.These might be the explanation of a major fraction of the variance of the data (say, >98%) and a maximal Fisher-F,,,, statistic form independent parameters and m + n data (solvent) points. Another method is principal component or factor analy- sis, in particular its target factor analysis (TFA) variant.8 This, again on the basis of statistical criteria, determines first how many independent basic factors are required for the explanation of most of the variance of the data, and then selects that many among solvent property vectors that describe the data most adequately. The former method (SMLR) has now been applied to a very extensive set of solvent properties that has not been considered previously for so many solvents of different classes, see Table 1.There are over 170 solvents for which the five parameters a,p, 7~*,6~,and ET(30)have been established. (There are many more for which one, mainly ET(30)or 6~,or two, both ET(30)and aH, are known.) There are 110 solvents for which DN, 52 for which AN, 61 for which Z, and 52 for which Acity and Basity are known in addition to the former five indices. (Each of these parameters is known for a few additional solvents, for which, however, most or all of a, /3, x*, and ET(30) are unknown.) Correlations among these parameters and between them and * Permanent address: Department of Inorganic and Analytical Chemistry, The Hebrew University of Jerusalem, Jerusalem 91904, Israel. 409 CHEMICAL SOCIETY REVIEWS, 1993 Table 1 The property parameters of organic solvents HBD ability a, HBA ability 8, polarity/polarizability T*,polarity ET(30), donor number DN, acceptor number AN, Aczty, Baszty, polarity Z, polarity Z Solvent a B Tr* Ed301 DN AN Acity Basity Z Z' perF-n-nexane 00 -08 -41 per F-Me-c-hexane 00 -06 -40 per F-decalin 00 -05 -32 Me,-silane 00 02 -09 30 7 2- Me- butane 00 01 -08 30 9 n-pentane 00 00 -08 31 1 n-hexane 00 00 -04 31 0 0 0 01 -01 n-heptane 00 00 -08 31 1 0 0 00 n-octane 00 00 01 31 1 n-decane 00 00 03 31 0 n-dodecane 00 00 05 31 1 c-hexane 00 00 00 30 9 0 02 06 60 1 cis-decalin 00 08 11 31 2 benzene 00 10 59 34 3 1 82 15 59 54 0 54 0 toluene 00 11 54 33 9 1 13 54 m-xylene 00 12 47 33 3 50 04 50 p-xy lene 00 12 43 33 1 50 06 50 mesitylene 00 13 41 32 9 10 0 styrene 00 12 34 8 50 water 117 47 1 09 63 1 18 0 54 8 1 00 1 00 94 6 89 6 methanol 98 66 60 55 4 30 0 41 3 75 50 83 6 79 4 ethanol 86 75 54 51 9 32 0 37 1 66 45 79 6 75 8 n-propanol 84 90 52 50 7 37 3 63 44 78 3 73 7 1-propanol 76 84 48 49 2 36 0 33 5 59 44 76 3 72 4 n-butanol 84 84 47 50 2 29 0 36 8 61 43 77 7 73 7 I-butanol 79 84 40 48 6 77 7 s-butanol 69 80 40 47 1 75 4 t-butanol 42 93 41 43 7 38 0 27 1 45 50 71 3 68 1 n-pentanol 84 86 40 49 1 25 0 77 6 72 9 I-pentanol 84 86 40 49 0 32 0 77 6 73 3 t-pentanol 28 93 40 41 1 66 6 n-hexanol 80 84 40 48 8 76 5 73 3 c-hexanol 66 84 45 46 9 25 0 75 0 n-octanol 77 81 40 48 3 32 0 n-decanol 70 82 45 47 7 73 3 benzyl alcohol 60 52 98 50 4 23 0 36 8 78 4 2-phenylethanol 64 61 88 49 5 23 0 33 8 3-phen ylpropanol 53 55 95 48 5 ally1 alcohol 84 90 52 52 1 2-chloroethanol 128 53 46 55 5 trifluoroethanol 151 00 73 59 8 53 8 hexafluoroiPrOH 1 96 00 65 65 3 66 7 ethanediol 90 52 92 56 3 20 0 78 84 85 I glycerol 121 51 62 57 0 19 0 82 7 phenol 1 65 30 72 53 4 11 0 m-cresol 1 13 34 68 52 4 50 4 p-cresol 1 64 34 68 53 3 m-chlorophenol 157 23 77 60 8 diethyl ether 00 47 27 34 5 19 2 12 34 di-n-propyl ether 00 46 27 34 0 18 0 di-1-propyl ether 00 49 27 34 0 19 0 di-n-butyl ether 00 46 27 33 0 19 0 06 28 60 1 di-C1Et ether 00 40 82 41 6 16 0 anisole 00 32 73 37 1 90 21 74 phenethole 00 30 69 36 6 80 58 9 dibenzyl ether 00 41 80 36 3 19 0 diphenyl ether 00 13 66 35 5 furan 00 14 36 0 60 tetrahydro furan 00 55 58 37 4 20 0 80 17 67 58 8 56 0 2-Me-THF 00 45 36 5 12 0 55 3 tetrahydropyran 00 54 51 36 6 22 0 dioxane 00 37 55 36 0 14 3 10 3 19 67 64 5 61 1 dioxolane 00 45 69 43 I dimethoxyethane 00 41 53 38 2 20 0 10 2 21 50 59 1 bis-MeOEt ether 00 40 64 38 6 99 18-cineole 00 61 34 0 24 0 acetone 08 43 71 42 2 17 0 12 5 25 81 65 7 61 8 2-butanone 06 48 67 41 3 17 4 23 74 64 0 60 4 c-pentanone 00 52 76 39 4 18 0 2-pentanone 05 50 65 41 1 63 3 3-pentanone 00 45 72 39 3 15 0 c-hexanone 00 53 76 39 8 18 0 25 79 Me-1-Bu ketone 02 48 65 39 4 62 0 58 3 2-heptanone 05 48 61 41 1 65 2 THE PROPERTIES OF ORGANIC LIQUIDS RELEVANT TO THEIR USE AS SOLVATING SOLVENTS-Y MARCUS 41 1 Table 1 contd Solvent a P x* Ed301 DN AN Acity Basity Z 2' dcetophenone 04 49 90 40 6 15 0 23 90 formic acid 123 38 65 57 7 19 0 83 6 118 51 82 6 dCetlC acid 1 12 45 64 55 2 20 0 52 9 93 13 79 2 79 2 propanoic acid 1 12 45 58 55 0 79 0 butanoic acid 1 10 45 56 54 4 78 3 pentanoic acid 1 19 45 54 55 3 79 5 hexanoic acid 1 22 45 52 55 4 79 6 heptanoic acid 1 20 45 50 55 0 79 0 acetic anhydride 00 29 76 43 9 10 5 methyl formate 00 37 62 45 0 70 3 66 6 ethyl formate 00 36 61 40 9 methyl acetate 00 42 60 40 0 16 3 10 7 ethyl acetate 00 45 55 38 1 17 1 93 21 59 64 0 58 7 propyl acetate 00 40 37 5 16 0 butyl acetate methyl propanoate 00 00 45 27 46 38 5 38 0 15 0 11 0 dimethyl carbonate 00 43 38 8 17 2 647 diethyl carbonate 00 40 45 37 0 16 0 64 6 ethylene carbonate 00 41 48 6 16 4 propylene CO, 00 40 83 46 6 15 1 18 3 72 4 methyl benzoate 00 38 38 1 15 0 ethyl benzoate 00 41 74 38 1 15 0 diMe-phthalate 00 78 82 40 7 ethyl Clacetate 00 35 70 39 4 13 0 ethyl C1,acetate 00 25 61 38 7 4-but yrolactone 00 49 87 44 3 18 0 17 3 fluorobenzene 00 07 62 37 0 30 60 2 p-difluorobenzene 00 03 58 36 4 hexafluorobenzene 00 02 33 34 2 1-chlorobutane 00 00 39 36 9 chlorobenzene 00 07 71 36 8 33 20 65 58 0 dichloromethane 13 10 82 40 7 10 20 4 33 80 64 7 59 3 1,1-dichloroethane 10 10 48 39 4 16 2 62 1 58 3 1,2-dichloroethane 00 10 81 41 3 0 16 7 30 82 64 3 o-dichlorobenzene 00 03 80 38 0 30 60 0 rn-dichlorobenzene 00 03 75 36 7 20 tr-diClethylene 00 00 44 41 9 chloroform 20 10 58 39 1 40 23 I 42 73 63 2 57 8 1,1,1 -Cl,ethane 00 00 49 36 2 trichloroethylene 00 05 53 35 9 16 54 C1,methane 00 10 28 32 4 0 86 09 34 C1,ethylene 00 05 28 31 9 10 25 1,1 ,2,2-C14ethane 00 00 95 39 4 64 3 1-bromobutane 00 13 50 36 6 dibromomethane 00 00 92 39 4 62 8 1,2-dibromoethane 00 00 75 38 3 60 0 bromoform 05 05 62 37 7 bromobenzene 00 06 79 36 6 30 22 66 59 2 1-iodobutane 00 23 47 34 9 diiodomethane 00 00 65 36 5 iodobenzene 00 06 81 36 2 40 butylamine 00 72 31 37 6 42 0 15 117 diaminoethane 13 1 43 47 42 0 55 0 20 9 pyrrolidine 16 70 39 39 1 piperidine 00 1 04 30 35 5 40 0 morpholine 29 70 39 41 0 17 5 diethylamine 03 70 24 35 4 50 0 94 triethylamine 00 71 14 32 1 61 0 14 08 19 tribut ylamine 00 62 16 32 1 50 0 diMe benzylamine 00 64 45 21 0 diMe cHexylamine 00 84 23 37 3 aniline 26 50 73 44 3 35 0 36 119 o-chloroaniline 25 40 83 45 5 31 0 N-methylaniline 17 47 82 42 5 33 0 40 1 07 dimethylaniline 00 43 73 36 5 27 0 pyridine 00 64 87 40 5 33 1 14 2 24 96 64 0 60 3 4-methylpyridine 00 67 84 39 6 34 0 2-fluoropyridine 00 51 84 42 4 perfluorop yridine 2- bromopyridine 00 00 16 53 53 1 00 36 3 41 3 3-bromopyridine 00 60 89 39 7 3,4-lutidine 00 78 73 38 9 2,6-lutidine 00 76 80 36 9 18 81 2-cyanopyridine 00 29 I 20 44 2 quinoline 00 64 92 39 4 32 0 acetonitrile 19 40 75 45 6 14 I 18 9 37 86 71 3 66 9 Table 1 contd Solvent a B x* ET(30) propanitrile 00 39 71 43 6 butanitrile 00 40 71 42 5 C1-acetonitrile 00 34 101 46 4 benzyl cyanide 00 41 1 00 42 7 benzoni trile 00 37 90 41 5 nitromethane 22 06 85 46 3 ni trobenzene 00 30 101 41 5 formamide 71 48 97 56 6 N-Me-formamide 62 80 90 54 1 dimethylformamide 00 69 88 43 8 diethylformamide 00 79 41 8 N-Me-acetamide 47 80 101 52 0 dimethylacetamide 00 76 88 43 7 diethylacetamide 2-pyrrolidinone 00 36 78 77 84 85 42 4 48 3 N-Me-pyrrolidinone 00 77 92 42 2 N-Me-caprolactam tetraMe-urea 00 00 69 80 83 41 6 41 0 tetraMe-guanidine 00 86 76 39 3 diMe-cyanamide carbon disulfide 00 00 64 07 72 61 43 8 32 8 dimethyl sulfide 00 34 57 26 8 diethyl sulfide di-i-propyl sulfide 00 00 37 38 46 36 35 7 34 9 di-n-butyl sulfide tetraCH, sulfide 00 00 38 44 36 62 34 9 36 7 pentaCH, sulfide dimethyl sulfoxide 00 00 36 76 61 1 00 35 9 45 1 tetraCH, sulfoxide 00 81 1 06 43 6 su1fo1an e 00 39 98 44 0 dimethyl sulfate trimethyl phosphate triethyl phosphate tributyl phosphate 00 00 00 00 36 77 77 80 78 72 72 65 43 6 41 7 39 6 Me, phosphoramide 00 1 05 87 40 9 other relevant quantities have now been explored with the following results Certain recommendations can be made on the basis of these results for the use of such solvent parameters 2 Correlations Among Parameters Table 2 describes the results of the binary correlations, in which one parameter, XYZ ( = a, p, T*, ET(30),DN, AN, 2,Aczty and Baszty) is tested against another, X (from the same list) in terms of the linear regression XYZ = XYZ, + x-x (2) The quantity presented is r(n),the correlation coefficient of the regression (equation 2), for n -2 degrees of freedom, where y1 is the number of pairs (X,XYZ) available for a given pair of parameters It is seen that for most cases the correlation is very slight For our large set of data the parameters a, p, and T* are essentially orthogonal to each other So are the donor number DN and the acceptor number AN (for the 45 solvents for which both are known) and the Aczty and Baszty (for all of the solvents for which both are known,6 excluding o-xylene, zso-octane, and trifluoroacetic acid, for which the other parameters are not known) It should be noted that ET(30),2,and DN, expressed in kcal mol-', and AN, expressed on a scale from 0 to 100, are not commensurate with a, p, and 7r*, that range mainly from 0 to 1 The normalized E$!= (ET(30)-30 7)/32 3, where 30 7 is the ET(30) of tetramethylsilane and 32 3 = 63 1 -30 7, with 63 1 the ET(30)of water, has already been introduced by Reichardt, placing Ep also in the range of 0 to 1 Similarly, the normalized ONN= DN/38 8, where 38 8 is the DN of hexamethyl phos- phoric triamide, and ANN= AN/54 8, where 54 8 is the AN of water, also place the normalized quantities mostly in the range CHEMICAL SOCIETY REVIEWS.1993 DN AN Acity Basrty Z Z' 16 1 16 6 67 8 10 0 15 1 11 9 15 5 30 87 65 0 60 6 27 20 5 39 92 71 2 68 2 44 14 8 29 86 24 0 39 8 66 1 00 83 3 27 0 32 1 26 6 16 0 30 93 68 4 65 3 30 9 77 9 27 8 13 6 27 97 66 9 32 2 13 6 27 3 13 3 27 I 29 6 92 17 0 20 10 38 29 8 19 3 34 1 08 70 2 67 0 14 8 19 2 70 6 23 0 16 3 26 0 64 6 23 7 99 61 3 38 8 10 6 00 107 62 8 0 to 1 If these normalized quantities are employed in expressions such as equation 2 (or equation 3 below), the coefficient x (and y) has to be multiplied by 32 4, 38 8, and 54 8 for EY, DNN,and ANN, respectively, and for the former also XYZ, must be modified Since most authors quote the non-normalized quanti- ties ET(30),DN, and AN, however, these are reported in Table 1 and used in the correlations The readers should have no difficulties in using the normalized quantities instead For those few cases in Table 2 where Y > 0 8, Table 3 presents the values of XYZ, and x It is seen that a, ET(30),AN, 2,and Aczty are interrelated, and so are DN with /3 and Baszty with T* For these cases it is worthwhile to explore correlations involving more than one independent variable Also shown in Table 3 are the results of the application of equation 2 to XYZ = Z', the value of the longest wavelength transition energy of 4-cyano- 1-ethylpyridinium iodide9 (in kcal mol- l, 1 kcal = 4 184 kJ) as a function of X = 2,the similar quantity for 4-methoxycarbonyl- I-ethylpyridinium iodide In this case XYZ, z 0, so that 2' is practically proportional to Z For two further cases the correla- tion according to equation 2 is sufficiently good to claim that XYZ is linearly correlated with X for Aczty with AN and for Z with ET(30) In a few cases there are outliers that are obviously based on faulty data (all the known data have been included in the correlations reported in Table 2) If these are excluded improved correlations according to equation 2 may be achieved, but this procedure should not be driven too far A few correlations with one independent variable but with data excluded are also known in Table 3 In the case of AN formic acid is an outlier, probably because it is a sufficiently strong acid to protonate, rather than hydrogen-bond to, the probe base, triethylphosphine oxide In the case of DN the aliphatic amines diethyl-, triethyl-, and tributylamine are outliers, either because their DNvalues are too high or because their p values are too lowlo (see also the THE PROPERTIES OF ORGANIC LIQUIDS RELEVANT TO THEIR USE AS SOLVATING SOLVENTS-Y.MARCUS 413 Table 2 Binary correlations of solvent parameters: the numbers in the first row of each entry are r, those in parenthesis in the 2nd row are n XlXYZ a B ff* ET(30) DN AN Z Acity Basity a 1 0.178 0.073 0.849 0.178 0.931 0.887 0.939 -0.071 (185) (174) (1 80) (110) (52) (61) (52) (52) 1 0.246 0.342 0.871 -0.044 0.504 0.329 0.329P (174) (1 80) (1 10) (52) (61) (52) (52) 77* 1 0.436 0.092 0.254 0.108 0.404 0.819 (1 69) (99) (52) (59) (52) (52) ET(30) 1 0.234 0.915 0.966 0.942 0.344 (1 09) (52) (61) (52) (52) DN 1 0.084 0.450 0.248 0.254 (45) (47) (48) (48) AN 1 0.920 0.980 0.020 (31) (32) (32) Z 1 0.926 0.068 (32) (32) Acity 1 0.152 (52) Table 3 The coefficients of equation 2 XYZ X XYZ, X o(XYZ) r n ET(30) a 38.2 f0.3 14.6 f0.7 3.9 0.8490 180 AN a 12.1 f1.1 33.6 f1.9 6.4 0.93 12 52 Z a 63.2 f0.3 19.5 f1.3 4.0 0.8870 61 Acity a 0.171 f0.015 0.67 f0.03, 0.09 0.9389 52 DN P 0.25 f1.18 40.4 * 2.2 6.2 0.8710 110 Basity 77* 0.086 f0.061 0.91 f0.09 0.18 0.8188 52 AN ET(30) -65.9 f 5.6 2.00 f0.12 7.1 0.9149 52 Acity ET(30) -0.99 f0.07 0.0317 f0.0016 0.09 0.9423 52 Z ET(30) 14.7 f1.9 1.236 f0.043 2.2 0.9661 61 Acity AN 0.034 f0.017 0.0158 f0.0006 0.06 0.9795 32 Z AN 53.3 f5.6 0.727 f0.079 3.8 0.9199 31 Z Acity 53.6 f5.2 40.3 f7.4 3.7 0.9260 32 z’ Z -0.03 f0.08 0.944 f0.002 0.5 0.9999 26 Improved correlations with (presumably faulty) data excluded AN a 12.2 f0.8 31.1 f1.4 4.7 0.9537 51 AN ET(30) -59.9 f4.1 1.850 f0.092 5.1 0.9441 51 (formic acid was excluded) DN B 0.5 f0.8 38.2 f1.5 4.3 0.9247 107 (diethyl-, triethyl-, and tributylamine were excluded) Z ET(30) 13.0 f1.8 1.27 f0.04 2.0 0.9721 60 (cyclohexme was excluded) Acity ET(30) -0.91 f0.05 0.0297 f0.0011 0.06 0.9662 50 Acity AN 0.02 f0.01 0.0171 f0.0004 0.03 0.9943 30 (formic acid and hexamethyl phosphoramide were excluded) Discussion).In the case of 2 the reported value for cyclohexane variable space is measured by o(XYZ),that should be minimal, is obviously too high, causing it to be an outlier.’ Formic acid, but has to be compared with the range of the XYZ values. The again, is an outlier in the case of the Acity, but so also is higher F, the better is the correlation, but it should be remem- hexamethyl phosphoramide, to which the arbitrary value of bered that it depends on the availability of data points, as Acity = 0 was assigned, whereas a value of Acity z0.2 is counted by n.expected in view of the values for N,N-dimethyl-formamide and The adequacy of the correlations depends on their purpose. If -acetamide. that is to learn what physical or chemical interactions are Correlations with a constant and two independent variables responsible for and contribute to a composite solvent para- according to equation 3 are shown in Table 4. meter, then correlations that explain 98 or even 93% of the variance may be adequate. If the purpose is to use the correlation XYZ = XYZ, + x.x + y.Y (3) for the prediction of ‘missing’ values of a parameter, then the value of a(XYZ)ought to be comparable with the expected error The quantities listed beside the explicit equation 3 are o(XYZ) in the experimental values of XYZ.For the values obtained (the standard deviation of the dependent variable), the range of spectroscopically as a transition energy expressed in kcal mol- XYZ values, r2 (the adjusted multiple correlation coefficient [ET(30),z]this error would be f0.4. For the statistically derived squared), the number nof data points (excluded were those listed parameters Acity and Basity this error would be f0.04.For the in the lower part of Table 3 as well as further ones), and the thermochemical value DN it is f1.0 kcal mol-l, and for AN F3,n-3statistic, for 3 and n -3 degrees of freedom. Listed in derived from NMR chemical shifts but normalized on a scale of Table 4 are only those correlations for which r2 > 0.90.The 0 to 100 this would be f0.5. The correlations in Table 4 do not scatter around the correlation vector in the 3-dimensional quite meet this criterion, since their a(XYZ) are a few times CHEMICAL SOCIETY REVIEWS, 1993 Table 4 Correlations according to equation 3 with r2 > 0 90 XYZ=XYZ,+ x.X+ b*Y dXYZ) range of XYZ rz n F ET(30) = 31 2 + 15 2.a + 11 5.x* 21 31 1-63 1 0 9585 166 92 1 (phenol m cresol and dimethylsulfide excluded) AN= -300 + 15 3.a + 1 Ol.ET(30) 24 0 0-54 8 0 9724 AN= 2 9 + 29 7.a + 14 O.n* 27 0 0-54 8 0 9667 48 48 792 652 (formic acid acetic acid m cresol and chloroform excluded) Z= 55 9 + 20 6.a + 10 2.n* 29 54 0-91 4 0 9429 55 212 (methyl formclte propylene carbonate and benzene excluded) Acltl = 0 03 + 0 64.a + 0 25.x* 0 05 0 00-1 00 0 9594 51 41 1 (hexamethyl phosphoramide excluded) Basitv = 0 04 + 0 94-n* + 0 035.p 0 07 0 00-1 08 0 9260 47 275 (acetic acid 1 butylamine aniline and carbon disulfide excluded) larger than the expected experimental errors The addition of a further variable was found to increase r2 and decrease o(XYZ) somewhat ET(30) benefits most from this with the term 3 4.p in addition to those in Q and T* [reducing o(XYZ) to 16 and increasing r2to 0 97551 An alternative would be the discarding of even more outliers, where now the justification is not the exclusion of expected faulty data but the further improvement of the fit It should be realized that according to the definition of Q as the measure of the ability of a solvent to donate a hydrogen atom towards the formation of a hydrogen bond, only protic and protogenic solvents have non-zero Q values These constitute only ca 30% of the solvents listed in Table 1, whereas the correlations with a shown in Tables 2 and 3 pertain to the entire set If only the subset of solvents with non-zero Q values were employed, somewhat different results could be expected, but then the correlations would have been skewed by the non- inclusion of zero values in the cases where the other parameters themselves have zero or low values In addition to the solvent parameters that measure the uncorrelated solvent properties ‘polarity’ (eg , T*), electron-pair donicity (HBA ability, e g ,p)and acceptance (HBD ability, e g ,Q) that are compared above, there are some other properties with which these parameters may be correlated These include properties obtained from physical measurements, such as the ‘tightness’ of the solvent, expressed as the Hildebrand solubility parameter SH or its square, the cohesive energy density Sh = (dvapN*-RT)/V*,where dvdPH*is the enthalpy of vapor- ization and V* is the volume, both per mole of the pure liquid at 298 15 K Also included are the relative permittivity (the dielec- tric constant) E or some function of it, e g , the polarization P = (E -1)/(2~+ l), and the polarizability, expressed as a func- tion of the refractive index aD (for the sodium D-line), R = (nb-1)/(2nb + 1) Table 5 shows that the solubility parameter and its square, which measures the work required to produce a cavity of unit volume in the solvent, are poorly correlated with any of the previously discussed parameters, so that this work is an indepen- dent property of the solvent The only exceptions are ET(30), which is somewhat correlated with tiH,and a, that is also slightly related to it It is not surprising that HBD solvents may be associated in the neat liquid form, hence have large cohesive energy densities, but this does not suffice for an acceptable Table 5 Non-correlations of solvent polarity and hydrogen bonding parameters with their structuredness, measured by the solubility parameter or its square Parameter a B x* ET(30) DN AN a.p n 121 121 121 121 85 50 121 r(sH) 0 583 0225 0540 0779 0 147 0 539 0549 r@i) 0533 0225 0475 0712 0 122 0495 0493 correlation It has been claimedI2 that the product of the HBD and HBA parameters, Q-P, can substitute for S& in certain correlations of processes with solvent properties As Table 5 shows, this might be true only for a limited set of selected solvents but the product Q-Pis even less correlated with 8f, than Q IS for our large set of very divers solvents The solvatochromic parameter 7~*that describes in a compo- site manner the polarity and polarizability of the solvents has been said to be well correlated with the product of the polariza- tion P and the polarizability R by means of equdtion 2 (XYZ = T*, X = P’R), albeit for certain limited sets of sol- vents For an extensive set, 1 15,of divers solvents the value of r for equation 2 is only 0 780 and o(T*) is 0 16 Application of equation 3 with X= P and Y = R yields similar results [y2 = 0 632, o(T*)= 0 161 If water, dioxane, and the alkanols dre excluded, then equation 2 yields for 90 solvents Y = 0 879 and o(T*)= 0 13 The use of modified functions of E and nD, such as having (E + 2) instead of (26 + 1) or (nD2+ 2) instead of (2n6+ 1) or the bulk quantities Y** P or V* -R did not improve the correlations However, the use of kirk wood'^'^ modified polarization function, X = P‘ = 9-g-~/(~-1) (E + 2) (wheregis the Kirkwood angular dipole correlation parameter) did produce with Y = R in equation 3 for 84 solvents for which T* and g1 are known the value r2 = 0 780 with o(T*) = 0 10, which is more promising, in view of the large variety of solvents included in the set Excluded, of course, are the non-polar solvents for whichg cannot be obtained since the dipole moment is zero 3 Solvatochromic Parameters Solvatochromic parameters, such as Q, p,and T* [also ET(30),Z and 2’1 have certain advantages over other parameters in that they are readily measurable by equipment to be found in most laboratories Of these, T* is measured directly, and is the mean of results for several probe (indicator) solutes 4-nitro-N N-di- ethylaniline (l), 3-nitro-N N-diethylaniline (2), 4-nitroanisole (3), 4-nitro- I -ethylbenzene (4), and 4-(2-nitroethenyl)anisole (5) l7 These probes are supposed to be insensitive to HBD and HBA properties of the solvents and to respond only to their polarity/polarizability The conversion expressions from the wavenumber v (in 1000cm-’) of the longest wavelength absorp- tion peak of a dilute solution of the probe in the solvent to 7~* values have been given” l8 (see also Table 6) The solvatochromic HBA-ability parameter /3 can also be obtained from measurements with probe solutes, however, a knowledge of the 7~*of the solvents IS generally required Suitable probes are 4-nitro-aniline (6), 4-nitrophenol (7), and tetramethylethylenediaminoacetylacetonato-copper(I1) perch-lorate (8) (for the latter, knowledge of T* is not required) The conversion expressions from the measured wavenumber v to /3 values, given T* values, have been presented18 (see also Table 6) Failing this direct measurement, provided the DN val~es,~lo and the T* and ET(30)values are known or can be estimated, the expression given in Table 4 can be inverted to give pvalues as THE PROPERTIES OF ORGANIC LIQUIDS RELEVANT TO THEIR USE AS SOLVATING SOLVENTS-Y MARCUS shown in Table 6, the sensitivity to DN being much larger than to the other two parameters The determination of a has, until recently, depended on the knowledge of certain other quantities (T*and in some cases also p), in addition to the solvatochromic or NMR data 2o Spectro-metry in the UV-visible region has been applied with 2,6- diphenyl-4-(2,4,6-triphenyl-1-pyridino)phenoxide (9) [z e , the ET(30) probe], 4-carbomethoxy- 1-ethylpyridinium iodide (1 0) (I e , the Z probe), 4-cyano-l-ethylpyridiniumiodide (1 1) (I e , the Z’ probe), bis( 1,lo-phenanthro1ine)-dicyano-iron(I1)(12), among others NMR with 31P was applied to triethylphosphine oxide (13) (I e ,the ANprobe) and with 13Cto N,N-dimethyl- or N N-diethylbenzamide [(14) and (15)] With the latter, large a/s (compare equation 1) were achieved, up to 4 74, depending on the chemical shift of which ring carbon was compared with that of the carbonyl one 21 Even much larger sensitivities to u with respect to T* (a/s-+ m ) were recently achieved with 13CNMR of pyridine-N-oxide (1 6) as the probe 22 The conversion expres- sions are shown in Table 6 Table 6 Expressions for the calculation of T*,p, and a and their standard deviations (5 Where applicable, the number of n of data and the ratio of the coefficients a, h, and s of equation 1 are also given Probe expression n 20 x* = 0 314.(27 52 -v) x* = 0 452.(25 52 -v) 0 06 0 06 x* = 0 427*(34 12 -v) 0 06 x* = 0 444.(29 96 -v) x* = 0 443.(37 67 -v) j3 = 0 358*(31 10 -v) -1 125.n*, bjs = 0 89 8 = 0 346*(35 045 -v) -0 57.77* -0 12.6,‘ 0 06 0 06 0 09 bls = 1 76 46 009 /3 = 0 358+ -18 76), bls + co /3 = 0 26.DN -0 00037.ET(30) -0 019.~* a = 0 0649*E,(30) -2 03 -0 72 x*, ajs= 1 39 17 009 90 0 10 138 0 13 a = 0 0485.2 -2 75 -0 46.x*, U~S= 2 19 55 0 12 a = 0 0514.2‘ -2 75 -0 46.n* 0 13 a = 0 375.(v -15 636) -0 45.x* + 0 27.8, ajs = 2 20 14 006 5) 5) (14),(15) (16) (16) 5) Q = 0 0337.AN -0 10 -u = 0 356-(d; -42 42) -a = 0 541 .[dj-41 98) -a = 0 694-(& -41 07) -a = -0 1624, + 2 43.a= -O174.di4+04O, a = 0 346.(d‘; -32 40) -0 47.n*, ajs = 2 12 0 53.x*, ajs = 1 87 0 21 .T*, ajs = 4 74 0 3h*, a/s = 3 18 ajs = +cc ajs = + co 0 42-n*. U~.S= 2 39 48 009 34 0 13 34 009 34 0 18 34 0 13 27 0 15 27 0 14 tl are the differences in chemical shifts S(C)-6(C = 0)in p p m of the 13C NMR signals of the zth ring carbon and the carbonyl carbon atoms in N N dimethyl or N N diethylbenzamide d,, IS the difference in chemical shifts 6(C ) -6(C,) and d3, is the difference 6(C3) -S(C,) in p p m for 13CNMR in pyridine N oxide The pohzabihty correction 6 is 1 0 for aromatic 0 5 for polychlorinated ahphdtic and 0 for all other aliphatic solvents 4 Discussion The results presented above show that there are four more or less independent solvent parameters that describe solvent properties relevant to the present discussion One is the hydrogen bond donation (HBD) ability, that is accounted for best by U,but is also described (along with a measure of the solvent polarity) by ET(30),Z or Z, AN, and the Aczty The second is the hydrogen bond acceptance (HBA) or electron pair donation ability, that is dccounted for best by p, but with which DNis also correlated It is interesting to note that, although DN is a measure of enthalpy, it is well correlated with Gibbs free energy quantities, such as p, ds also with others (eg , IR frequency shifts), a fact that was already commented upon lo The third is the polarity and polarizability of the solvent, measured T*,with which the Baszty is correlated, and also, for a limited list of solvents, the modified polarization P‘and the polarizability R The fourth is the solvent stiffness,15l6 measured by its cohesive energy density, 8)$, the work required to produce in the solvent a cavity of unit volume This is one measure of its structuredness l6 Some or all of these four solvent property parameters should be adequate for LSERs or QSARs (quantitative structure/ activity relationships) similar to equation 1 in all the 77 pro- cesses listed by Swain et a1 and the 560 processes listed by Taft et al 23 (some of which appear also on the former list6) or the very numerous processes discussed by Rei~hardt’~ as far ds solvent effects are concerned The processes include reaction kinetics and equilibria (z e, differences in solvent effects on initial and transition states and reactants and products, respecti- vely) and spectroscopic processes (I e , differences in solvent effects on the ground and excited states) for light absorption in the ultraviolet, visible, and infrared regions, and NMR chemicdl shifts and similar quantities It should be noted that the terms ‘acidity’ and ‘basicity’ as applied to solvents (to be distinguished from Aczty and Basztk6) are not employed in the present context They pertain to the complete transfer of a proton from the solvent to the solute (dcidity) or from the solute to the solvent (basicity), forming new species that are charged Such a process goes beyond solvation by the solvent which should be confined to adduct formation and hydrogen bonding or dipole -(induced) dipole interactions The assignment, sometimes found in the literature, of (Lewis) acidity to aprotic and non-protogenic solvents such as dimethyl- sulfoxide, N N-dimethylformamide, or hexamethyl phosphoric acid triamide implies the formation of a coordinative bond between an electron pair of a donor atom of a solute and the positive end of the dipole of the solvent molecule The evidence IS against this, the positive end of the dipole being well shielded It is more expedient to assign the solute-solvent interactions to dipole attraction In the present context, this is the responsibility of the T*term rather than that of the a term, and aprotic solvents therefore rightly have zero u values The question of whether quantities based on a single indicator probe, on the average of results from several probes, or as a statistical parameter derived from d large number of results (including reaction kinetics and equilibria as well as spectro- scopic data) are the best descriptors of solvent properties has been argued in the literature 23 24 Strictly speaking, the solvent effects observed for a given probe should not be readily transfer- able to any other solute, in particular one that has different functional groups A case in point is the different HBA proper- ties of solvents measured with 4-nitroaniline and with 4-nitro- phenol Practically, however, the main consideration should still be that the purpose of obtaining numerical values for solvent properties is their use as descriptors or predictors of the solvent effects on the behaviour of divers kinds of solutes and transition states The probes should thus act as stand-ins or substitutes for the ‘general solute’ Hence, if several probes of rather different chemical constitution provide concorddnt results for a given property (within a few percent of the total range of the quantity for the entire set of solvents), this would mean that they do indeed measure the property in a useful manner The average of the numerical values obtained from such probes is, therefore, a more meaningful quantity than the value obtained from any single probe, such as, e g , antimony pentachloride for the donor number DN5 or 2,6-diphenyl-4- (2,4,6-triphenyl-1-pyridino)phenoxide for the polarity index ET(30) l9 This is one advantage of the Kamlet-Taft U,p, and T* solvent parameters, which are averages of results from several probes, although this has also been regarded as one of their weaknesses 24 The uncertainties of the parameters quoted in Table 6 (20) reflect the spread of the values for the individual probes around the mean (Some other parameters used in physical organic chemistry, such as the Hammett acidity func- tions No, are also based on averages of several probes, with comparable uncertainties ) On the other hand, driving this procedure to the extreme of the statistical analysis of Swain et a1 has the disadvantage that new solvents are not readily added to the list of 61 considered by these authors without the use of 41 6 their particular statistical program and the many kinds of results (including reaction kinetics and equilibria as well as spectro- scopic data) they have employed There are many additional probe molecules that have found limited use for the determination of polarity, HBA, and HBD properties of solvents by means of UV-visible spectrometry (solvatochromatic indicators) or, say, IR band position and NMR chemical shift measurements Only a few can be men- tioned here Nicolet and Laurence, for instance, provided data from which r* of solvents can be obtained with indicators such as 2,4-dinitro-N,N-diethylaniline, 4-cyano-N,N-dimethylaniline, 4-acetyl-N,N-dimethylaniline, and 4-carbomethoxy-N,N-di-methylaniline, in addition to (1) and (3) mentioned above 25 They also provided data from which ,8 can be obtained with the indicator 4-aminoacetophenone, in addition to the indicators (6) and (7) listed above The HBA ability can be determined by the shift between the lowest and the second-lowest energy absorp- tion peaks of diacetylacetonatooxovanadium(Iv),dI and the NMR chemical shift of 23Na in dilute solutions of sodium iodide lo 27 In particular, the shift of the O-D stretching frequency of CH,OD in various solvents, forming the basis of Koppel and Palm’s well-known B scale2* was shown to be linear with DN, hence also with ,8 lo The lowest energy absorption peak of Michler’s ketone, 4,4’-bis(dimethy1amino)benzophe-none, was shown29 to conform very well to equation 3 with X = u and Y = n* The combination of Drago’s earlier E-C enthalpy-based specific interaction approach with his more recent spectroscopically-based non-specific interaction approach led him to a four-parameter expression30 similar in form to equation 1, but stressing other aspects of the solute- solvent interactions than the u-,~T* treatment stressed here The averaging of results obtained with several probes of differ- ent natures, shapes, and sizes, however, was recommended These are just examples of such correlations, reference 19 gives a wealth of further information 5 Summary The more widely used solvent parameters that have been pro- posed for the description of the polarity and the hydrogen bond and electron-pair donation and acceptance properties of more than 180 solvents have been compared and correlated These properties contribute to the exoergzc solute-solvent interactions that dre required for the solute to be soluble in the solvent in the first place and to solvent effects on spectra and reactions in the second place Three mutually independent quantities play roles in this respect these are measured by the solvatochromic parameters a (for HBD), ,8 (for HBA), and T* (for polarity/ polarizability) of the solvent The main endoergic contribution to solute-solvent interac- tions is the formation of a cavity in the solvent to accommodate the solute This is a chemical property of the solvent, depending on the association of its molecules in the liquid state It is measured by the cohesive energy density, Sb, which is indepen- dent of the former three parameters when all the solvents are considered together This measure of the stiffness or tightness of the solvent is related indirectly to its ‘structuredness’, which is measured by the entropy deficiency of the liquid solvent relative to the solvent in the ideal gas state, corrected for its compres- sion s This quantity, however, has not been calculated so far for most solvents at room temperature When only a small set of related solvents is considered, mutual correlations of the solvent parameters may arise, and this must be guarded against when causes for solvent effects are sought, since not all the four parameters need be operative for a given process The mutual orthogonality of the parameters employed must be tested and confirmed It must also be stressed that certain physical properties of solvents, such as the relative permittivity (dielectric constant, E) may be very important where charged solute species are con- CHEMICAL SOCIETY REVIEWS.1993 cerned Low E values lead to solute-solute interactions (ion pairing of unlike charged species) even in dilute solutions, but this effect is outside the scope of this paper Also, certain chemical properties of solvents, such as hydrophobicity and miscibility or mutual solubility with water are not directly relevant to the solvation ability of solvents, although they play important roles in chromatography or liquid-liquid distri-bution The solvating ability of solvents is described by the HBA and HBD abilities, the polarity, and tightness, without having to invoke their behaviour towards water However, when interac- tion takes place with a very ‘soft’ solute, the softness of the solvent should also be taken into account as an additional solvent property 31 Acknowledgement This work was inspired by the correlations on a more limited set of data, communicated to the author by Dr S Spange of the Friedrich-Schiller University in Jena, Germany, from the diploma thesis of D Keutel prepared under his guidance 6 References 1 M J Kamlet, J -L M Abboud, M H Abraham, and R W Taft J Org Cliem , 1983,48,2877 2 K Dimroth, Ch Reichardt, T Siepmann, and F Bohlmann, Liebigs Ann Clreni .1966, 661, 1 3 E M Kosower, J An? Cliem SOL, 1958,80, 5253 4 U Mayer, V Gutmann, and W Gerger, Monatsh Chem , 1975 106, 1235 5 V Gutmann and E Wychera, Inorg Nucl Chem Lett , 1966,2,257 6 C G Swain, M S Swain, A L Powell, dnd S Alunni, J Am Chem Soc , 1983, 105, 502 7 J H Hildebrand, J Am Cheni SOL, 1916,38, 1442 8 E R Malinowski, ‘Factor Analysis in Chemistry’ 2nd ed , Wiley Interscience, New York, 199 1 9 J Hormadaly and Y Marcus, J Ph~s Chem , 1979. 83, 2843, K Medda, M Pal, and S Bagchi, J Chem SOC Faradab Trans I 1988 84, 1501 10 Y Marcus, J Solution Cheni , 1984, 13, 599 11 K Medda, P Chatterjee, A K Chandar, and S Bagchi, J Chem SOC Perkin Trans 2, 1992, 343 12 D C Legget, J Solution Cheni , 1993, 22, 289 13 V Bekarek, J Plijs Chem , 1981,85,722 14 R H Cole, J Chem Phjs , 1957,27, 33 15 Y Marcus, J Solution Chem , 1992,21, 1216 16 H P BenettoandE F Caldin, J Chem Soc A 1971,2191 17 M J Kamlet, M E Jones, and R W Taft, J Chem Soc Perkin Trany 2, 1979, 345 18 Y Migron and Y Marcus, J Cliem Soc Faradai Trans , 1991 87 1339 19 Ch Reichardt, Chem Rev, 1994, 94, in the press, ‘Solvents and Solvent Effects in Organic Chemistry’, 2nd Edn , VCH, Weinheim (FRG), 1988 20 Y Marcus, J Solution Chem , 199I, 20, 929 21 H Schneider, Y Migron, and Y Marcus, Z Phvs Chem , 1992,175, 145 22 H Schneider, Y Badrieh, Y Migron, and Y Marcus, Z Ph~r Chem , 1993, 177, 143 23 R W Taft, J -L M Abboud, M J Kamlet, and M H Abraham, J Solution Chem , 1985, 14, 153, cf Appendix, pp 176-186 24 C Laurence, P Nicolet, and M Helbert, J Chem Soc Perkin Trans 2, 1986, 1081 25 P Nicolet and C Laurence, J Chem Soc Perkin Tran? 2, 1986, 1071 26 J Selbin and T R Orlando, J Inorg Nucl Chem , 1964,26, 37 27 A I Popov, Pure Appl Chem , 1975,41,275 28 I A Koppel and V A Palm.in ‘Advances in Linear Free Energy Relationships’, ed N B Chapman and J Shorter, Plenum Press, New York, 1972, p 2038 29 D Keutel, ‘Diplom Thesis’, University of Jena, 1991, S Spange and D Keutel, Liebigs Ann Chem , 1992, 423 30 R S Drdgo, D C Ferris, and N Wong, J Am Chem Soc, 1990, 112, 8953, R S Drago, J Chem Soc Perkin Trans, 1992, 1827, R S Drago, J Org Chem , 1992,57,6547 31 Y Marcus, J Phvs Chem , 1987,91,4442

ISSN:0306-0012    

DOI:10.1039/CS9932200409

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Water Purification by Semiconductor Photocatalysis Andrew Mills,” Richard H. Davies and David Worsley Department of Chemistry University College of Swansea Singleton Park Swansea SA2 8PP Wales 1 Introduction It has long been recognized’ that there are many reactions which can be promoted by light-activated solids which are not con- sumed in the overall reaction; such solids are often referred to as photocatalysts or photosensitisers and are invariably semicon- ductors. Probably the most well-established example of semi- conductor photocatalysis is paint chalking,2 which involves the photodegradation of the organic polymer part of the paint sensitized by the semiconductor pigment usually TiO,. Nowa- days this undesirable feature is often largely controlled by coating the pigment with a hydrous layer of an inert oxide such as silica alumina or zirconia. In recent years there has been a growing interest in the use of semiconductors as photosensitizers for the complete oxidative mineralization of pollutants by ~xygen.~ In semiconductor photocatalysis for water purification the pollutants are usually organic and therefore the overall process can be summarized by the following reaction equation semiconductororganic + 0 ullro-bdndgdp ,,ght CO + H,O + mineral acids pollutant (1) In contrast to semiconductor photomineralization as a method of water purification the destructive technologies cur- rently in use in the water industry such as chlorination and ozonation all use strong oxidants of a seriously hazardous and therefore undesirable nature.The predominant non-destructive technologies currently in use i.e. air-stripping and carbon absorption are also not without their problems. For example the removal of volatile contaminants by air-stripping converts a liquid contamination problem into an air pollution problem Richard H. Davies Chemistry B.Sc. (Hons.) University College of Swansea 1990; Ph.D. University of Wales U.K. 1993. Fields of interest include heterogeneous photochemistry especially photocatulj’sed Mater purijication and computer modelling of reaction kinetics. David A. Worsley Chemistry B.Sc. (Hons.) University College of Swanseu 1989; Ph.D. University of Wales U.K. 1993. Fields of interest include photocatalysed reactions and interfacial redox chem is trj’. and carbon absorption produces a hazardous solid which in turn must be disposed of. The purification of water by semiconductor photocatalysis is attracting a great deal of interest not only from research workers but also from water purification companies.This interest arises because (i) the mineral ‘effluent’ it produces is harmless to the environment (ii) the process of photominerali- zation can be turned on or off at the flick of a switch and (iii) there is a real possibility that it could be readily incorporated into existing UV water purification systems. 2 Semiconductor PhotocataIysis In a solid the electrons occupy energy bands as a consequence of the extended bonding network. In a semiconductor the highest occupied and lowest unoccupied energy bands are separated by a bandgap Ebg,a region devoid of energy levels. Activation of a semiconductor photocatalyst is achieved through the absorp- tion of a photon of ultra-bandgap energy which results in the promotion of an electron e- from the valence band into the conduction band with the concomitant generation of a hole h + in the valence band. For a photocatalyst to be efficient the different interfacial electron transfer processes involving e -and h+ reacting with adsorbed species must compete effectively with the major deactivation route of electron-hole recombina-tion~.~-Semiconductor photocatalysis has its origins in the substan- tial research effort in the seventies and early eighties into photoelectrochemical systems and micro-photoelectrochemical systems for solar to chemical energy conversion.The first photoelectrochemical system for splitting water into hydrogen and oxygen was developed by Fujishima and Honda6 and comprised aTiO semiconductor photoanode coupled up to a Pt cathode and a schematic illustration of this cell is given in Figure 1. After a few years research it was realised’ that micro- Fujishima-Honda cells comprisingTiO particles with deposits of Pt on them would also work as photocatalysts for splitting water as illustrated in Figure 1. As part of the research effort into solar-to-chemical energy conversion using photoelectrochemical cells some attention was given to the use of platinizedTiO powder as a photosensitizer for the production of hydrogen from biomass and water,* i.e. Andrefii Mills Chemistry B.Sc. (Hans.) University oj London 1979; Ph.D. Royal Institute 1982; Lectureship University College of Swansea 1982-. Fields of interest include homogeneous and heterogeneous photochemistry solar to chemical energ]’ con- version optical and electrical gas sensors redox catalysis and corrosion science. 417 e-I 1f[ "'r> Ebg TiO Electrode H+ + O2 Fujishima-Honda Photoelectrochermcal cell 1/2 H CB TI02 particle H2O H++ 1/2 0 Micro-photoelectrochemical cell Figure 1 Schematic illustrations of the Fujishima Honda photoelectro- chemical cell6 (top) and a micro version of the cell for the dissocta- tion of water into H and 0 Electron Energy I/ 02 CB VB Semiconductor Semconductor/soIutionsurface reactions Figure 2 Schematic representation of the band energetic model of the overall process of semiconductor photocatalysis for water purification TiO,/Pt biomass + H20 ultra bandgap hght "02+ H2 where the biomass was natural products (such as glucose ethanol cellulose etc ) food (such as sweet potatoes and fatty oil) grass clover wood (including cherry wood) green algae seaweed dead animals (e g cockroaches) and excrement (including human urine) * Of course for reaction 2 to work all irradiations had to be carried out in the absence of 0 It IS interesting to note however that had the systems been aerated rather than N,-purged then the reaction under study would have been the semiconductor-sensitized photo-oxidative miner- alization of the biomass The first clearest recognition and implementation of semicon- CHEMICAL SOCIETY REVIEWS 1993 ductor photocatalysis as a method of water purification came in 1983 from David 0111s and his co-workers through the publica- tion of two papers9 on the photomineralization of haloge- nated hydrocarbon contaminants including trichloroethylene dichloromethane chloroform and carbon tetrachloride sensit- ized byTi0 (Interestingly during conventional water chlori- nation procedures chloroform is the major halocarbon conta- minant formed and is a suspected carcinogen ) Subsequently it appeared' that reaction 1 might be limited to non-aromatic compounds however these fears were quickly eliminated through the work of Matthews,' Barbeni et a/ ,13 and Okamoto et a/,14who demonstrated thatTi0 could sensitize the photomineralization of chlorobenzene chlorophe- nol and phenol respectivelyThe list of organic pollutants which have been shown to be photomineralized via reaction 1 using Ti0 as the sensitizer is extensive,' as indicated by Table 1,l and increases daily 2.1 Semiconductor Photocatalysis for Water Purification Choice of Semiconductor There are a lot of different semiconducting materials which are readily available but only a few are suitable for sensitizing the photomineralization of a wide range of organic pollutants I e sensitizing reaction I A sensitizer for reaction 1 must be (1) photoactive (11) able to utilise visible and/or near UV light (111) biologically and chemically inert (iv) photostable (z e not liable to photoanodic corrosion for example) and (v) cheap In order for a semiconductor to be photochemically active as a sensitizer for reaction 1 the redox potential of the photogener- ated valence band hole must be sufficiently positive to generate absorbed OH radicals which can subsequently oxidize the organic pollutant. and the redox potential of the photogener- ated conductance band electron must be sufficiently negative to be able to reduce absorbed 0 to superoxide a schematic representation of the energetics associated with the overall process is given In Figure 2 Figure 3 illustrates band positions for a variety of different semiconductors and the redox poten- tials for the H,O/OH and O,/HO couple at pH 0 (The relationship between Ebg,in units of eV and the approximate E(O,/HO;) E(H20/0 H') I 2 8eVn*j(p+SrTi03 1 7eVCdSe 1 4eV CdTe 1 3 2eV ZnO I I CdS 1 2 4eV ZnS I 36eV I -2 -1 0 +1 +2 +3 V vs NHE Figure 3 Band positions of common n-type semiconductors used in photocatalysis and the redox potentials-of the H20:'OH' and O2 HO,' redox couples at pH 0 WATER PURIFICATION BY SEMICONDUCTOR PHOTOCATALYSIS-A MILLS R H DAVIES AND D WORSLEY ~~~ ~ Table 1 Photomineralization of organic pollutants sensitized byT10 Examples of compounds studied' ChSS ALKANES HALOALKANES ALIPHATIC ALCOHOLS ALIPHATIC CARBOXYLIC ACIDS ALKENES HALOALKENES AROMATICS HALOAROMATICS PHENOLS AROMATIC CARBOXYLIC ACIDS POLYMERS SURFACTANTS HERBICIDES PESTICIDES DYES Examples methane [sobutane pentane uooctane heptane n-dodecane cyclohexane. methylcyclohexane 1,4- methylcyclohexane paraffin mono- di- tri- dnd tetra-chloromethane fluorotrichloromethane 1,l- and 1,2-dichloroethane. I I 1,2- and I 1,2,2-tetrachloroethane,pentachloroethane 1,l-and 1,2-dibromoethane tribromoethane 1,2- dichloropropane 1-bromododecane 1,l -difluoro- 1,2-dichloroethane 1 I -difluoro- 1,2,2-trichloroethane 1,1,1 -trifluoro-2,2,2-trichloroethane,tributyl tin compounds methanol ethanol isopropyl alcohol cyclobutanol. n-propyl alcohol propan-2-01 butanol penta- 1,4-diol 2-elloxyethanol. 2-butoxyethanol dodecanol benzyl alcohol glucose sucrose formic ethanoic dimethylethanoic mono- di- and tri-chloroethanoic propanoic butanoic dodecanoic oxalic propene cyclohexene perchlorethene 1,2-dichloroethene 1 I 1-and I 1,2-trichloroethene tetrachloroethene mono- di- and tetra-fluoroethene 3,3,3-trifluoropropene hexafluoropropene benzene naphthalene chlorobenzene bromobenzene 2- 3- and 4-chlorophenol 2,4- and 3,4-dichlorophenol 2,4,5- and 2,4,6- trichlorophenol pentachlorophenol 2- 3- and 4-fluorophenol 2,4- and 3,4-difluorophenol 2,4- dinitrophenol 1,2-dichlorobenzene 1,2,4-trichlorobenzene 2,3- and 3,4-dichloronitrobenzene1.2-dichloronitrobenzene phenol hydroquinone methylhydroquinone catechol 4-methyl catechol 4-nitrocatechol resorcinol 2-naphthol 0-,m-,and p-cresol benzoic 4-amino benzoic 3-chloro-4-hydroxybenzoic,phthalic salicyclic m-and p-hydroxybenzoic 3- chlorohydroxybenzoic polyethylene PVC SDS p-nonyl phenyl polyoxyethylene ether polyethylene glycol p-nonyl phenyl ether sodium dodecyl benzene sulfonate benzyl dodecyl dimethyl ammonium chloride p-nonyl phenyl poly(oxyethylene)esters sodium benzene sulfonate paraxon malathion 4-nitrophenyl ethyl phosphinate 4-nitrophenyl isopropyl phosphinate I-hydroxy ethane- 1 I-diphosphonate 4-nitrophenyl diethyl phosphate trimethyl phosphate trimetyl phosphite dimethyl ammonium phosphodithionate tetrabutyl ammonium phosphate methyl viologen atrazine simazine prometon propetryne bentazon DDT parathion lindane methylene blue rhodamine B methyl orange fluorescein umbelliferone threshold wavelength of light hth in nm below which the semiconductor will absorb strongly I e hlh is given by the expression Ebs= 1240/hth) In many cases the semiconductor is liable to oxidative decomposition by the photogenerated hole It is generally found that only n-type semiconductor oxides are stable towards photoanodic corrosion although such oxides usually have bandgaps which are sufficiently large that the semiconductors absorb only UV light CdS is an example of a highly active semiconductor photosensitizer which has the highly desirable feature that it can be activated using visible light (thus sunlight could be used) but as is typical for visible light absorbing semiconductors it is liable to photoanodic corrosion CdS + 2h+-+Cd2++ Sl (3) and this feature renders it unacceptable as a photocatalyst for water purification Of all the different semiconductor photocatalysts tested suc- cessfully for reaction 1Ti0 appears the most active as illustrated by the results in Figure 4 for reaction 1 where the organic pollutant is pentachlorophenol (PCP) In addition althoughT10 is only a UV absorber (Ebg= 3 2 eV) it is cheap insoluble under most conditions photostable and non-toxic Indeed the complete absence of any biological activity asso- ciated with this bright white pigment allows it to be used in a wide range of domestic goods including sunblocks vitamin tablets and chicken roll' Although there are many different sources ofTiO Degussa P25Ti0 has effectively become a research standard because it has (1) a reasonably well-defined nature (z e typically a 70 30 anatase rutile mixture non-porous BET surface area = 55 f 15 m2 ggl average particle size 30 nm) and (11) a 17c v vv L IJ time/hours Figure 4 Photodegradation of pentachlorophenol (PCP) sensitized by dispersions of the following semiconductors +Ti0,. x ZnO 0 CdS 0 WO and A SnO In each case the solution was air- saturated and the semiconductor and initial PCP concentrations were 2g cm and 45 pmol dm respectively Data from reference 16 substantially higher photocatalytic activity than most other readily available samples ofT10 2.2 Kinetics of Pollutant Photomineralization Sensitized by TiO In the photomineralization of organic pollutants sensitized by TiO numerous studies3 l4 have reported that the initial rate r of disappearance of the pollutant S fits a Langmuir-Hinshel- wood kinetic scheme I e CHEMICAL SOCIETY REVIEWS 1993 c2 2 I Crystal Phase' Even after choosingT10 as the semiconductor photocatalyst l2 the choice of which crystalline form is still importantThe crystalline forms of Ti0 are anatase rutile and brookite O8 t206 II 0 200 400 600 800 1000 1200 Annealing Temperature / ("C) Figure 5 Initial rate of CO generated in the photomineralization of 4-chlorophenol sensitized by Ti0 versustemperature used to anneal the Ti0 Data from reference 19 (4) where [S] = initial concentration of the pollutant S and tradi- tionally K(S) is taken to represent the Langmuir absorption constant of the species S on the surface ofTiO and k(S)is a proportionality constant which provides a measure of the intrinsic reactivity of the photoactivated surface with S It is found that k(S)is proportional to e,,where I is the rate of light absorption and 0 is a power term which is equal to + or 1 at high or low light intensities respectively It is also found that k(S)is proportional to the fraction of 0 adsorbed on the TiO I e .f(O,) which is defined as follows (5) where KO is the Langmuir adsorption coefficient for O which appears to be non-competitively absorbed onT10 owing to its exclusive absorption at Ti"' sites hydroxyl radicals and organic substrates are believed to be exclusively absorbed at Tilv-lattice oxygen sites Thus a more complete form of equation 4 is given by the following expression where y is a proportionality constant Cunningham and his co-workers" and others have measured dark Langmuir adsorption isotherms for T10 for a variety of different organic pollutants and found them to be significantly smaller than the values of K(S)obtained from plots I/r versus l/[S] It appears likely that the value of K(S) derived from a kinetic study is not directly equivalent to the Langmuir absorp- tion coefficient for S on T10 Turchi and Ollisi8 have proposed four possible different mechanistic schemes involving OH' radi- cal attack inTi0,-sensitized organic photomineralization reac- tions all of which yield a rate equation of the form of equation 6 In each scheme the fundamental interpretation of the constant K(S) is different and none are simply equivalent to the dark Langmuir absorption coefficient In the Appendix we describe the two schemes which appear most likely given the increasing evidence that the OH' radical is bound to the s~rface,~ z e Case A adsorbed hydroxyl radical attack on adsorbed S and Case B adsorbed hydroxyl radical attack on free S In both cases a final working equation is developed I e equation A 19 in the Appen- dix which has an identical form to that of the empirically derived equation 6 Detailed studies of the kinetics of water purification photo- sensitized byT10 have identified the following important factors conditions used to prepare the T10 The popular Degussa P25 tO4 T10 used in most photomineralization studies is primarily anatase (70%) but can be converted into rutile Ti0 by anneal- nn"" iing at 800°C for 5h in air Such treatment reduces the photocata- although brookite is not commonly available The result of an initial study on the photomineralization of phenol indicated that rutile T10 is inactive as a catalyst although it is not clear why this should be given that the redox potentials of the valence and conductive bands for anatase and rutile T10 are quite similar However a more rigorous study using the same pollutants showed thatT10 was active or inactive according to the initial lytic activity of the Ti0 dramatically as indicated by the plot of initial rate of CO evolution versus annealing temperature for the photomineralization of 4-chlorophenol illustrated in Figure 5 l9 However this behaviour appears to be associated more with the concomitant decrease in specific surface area (from 50 down to 5 m2 g-l) than the conversion of the T10 from anatase into rutile Thus the crystal phase of T10 used may not be that important a factor but the method of its preparation is especially with regard to its porosity and surface area 2 2 2 TIO Concentration T10 is often used as a dispersion to sensitize reaction 1 and Figure 6(a) provides an illustration of a typical 'batch' reactor for this work 2o In 'batch' reactors the rate of photomineraliza- tion is often found to increase with increasing [TiO,] and to tend towards a limiting value at high concentrations (typically ca 0 5 mg ~m-~) In any commercial system it is more likely that the T10 will be fixed and the contaminated water will be flowed over it Such a 'flow' system would eliminate the need for filtration or settling/resuspension of the photocatalyst In the development of d wide range of flow reactorsT10 has been (1) incorporated in Nafion film ceramic film silica gel and (11) attached to glass in the form of tubing beads and mesh The tremendous tenacity with which T10 readily adheres to glass makes the production of a glass 'flow' reactor fairly easy typical glass flow reactors are illustrated in Figure 6(b) There have been many attempts3 to improve the photo- catalytic activity ofTiO in particular by depositing small amounts of metal (Pt Ru or Cu) on its surface or extending its II Figure 6 Schematic illustration of the two different types of photoredc- tor used in the study of water purificdtion by semiconductor photo- catalysis I r (a) a 'batch' photoreactor,20 comprising two half-cylinders (A) each containing 6 x 8 W Blacklight UVA lamps which when pushed together form the photoreactor and a reaction vessel (B) with a top (C) through which 0 gas is continuously bubbled through the reaction solution by means of a peristaltic pump (E) The reaction solution is usually stirred continuously by means of a magnetic stirrer (DL WATER PURIFICATION BY SEMICONDUCTOR PHOTOCATALYSIS-A MILLS R H DAVIES AND D WORSLEY 42 1 20 w UV fluorescent tube 0 80 160 240 320 outlet(1 \&-Ti02 coated glassmeshin annulus 0 0 Ti02 coated glass tube silicone rubber inletsealant I I1 (b) two 'flow' (I) comprisingTi0,-coated glass mesh in the annulus between a UVA lamp and the outer glass wall and (11) comprising a glass tube spiralled around a UVA lamp with a thin inside coating of TiO over which the polluted water is flowed wavelength range of photoactivity by dye sensitization or doping with transition metal ions (e g Fe"' or Cr"') To date this work has largely been unsuccessful and Degussa P25Ti0 remains the best readily available form of Ti0 for reaction 1 2 2 3 Organic Pollutant Concentration As noted earlier for most of the organic pollutants tested the kinetics of reaction 1 are described by equation 4,a Langmuir- Hinshelwood-type expression and this feature is rather nicely illustrated in Figure 7,a plot of rate of oxidation of pollutant to CO versus [pollutant] for a variety of different pollutants 22 It is clear that the kinetics of photomineralization will depend upon the ease with which it can be oxidized by the photogener- ated hole and how well it absorbs on the surface of Ti0 It is also worth noting that the absorption spectrum of the pollutant can drastically affect the kinetics of photocatalysis In particular if the pollutant is a strong UV absorber. then as its concentration is increased it will eventually begin significantly to screen theTi0 from the ultra-bandgap light and the kinetics of photomineralization will begin to deviate from equation 4,with the rate decreasing with increasing [pollutant] Many of the pollutants listed in Table 1 do not absorb significantly in the 300-400nm wavelength region and the problem of screening (as well as the homogeneous photochemical decomposition of the pollutant) is rendered insignificant through judicious choice of irradiation source (usually black-light bulb emission A, = 365nm) and irradiation vessel (usually borosilicate glass which cuts off light below 300nm) The effect of screening on the kinetics of photocatalysis can however be readily observed using dyestuffs such as rhodamine 6G Screening of theTi0 by highly coloured pollutants or deposits of particulate matter on the surface of Ti0 for that matter appears to be a quietly forgotten general major drawback of Ti0 photocatalysis as a method of water purification I I I I rc e,Y2 "0 20 40 60 80 [Pollutant]/ (mg dm ') Figure 7 Rates of oxidation to CO versus initial pollutant concent- ration for the following different pollutants (top to bottom) chloro- form (0),methanol (A) 4-chlorophenol(O) phenol (0),dcetic dad ( x ) and propan-2-01 ( + ) Data from ref 22 determined using d flow photoreactor (20 W UVA lamp 40°C and circulation rate = 300 cm3 min-I) 2 2 4 0 Concentration As you might expect given the reaction stoichiometry of equa- tion l photomineralization will not proceed unless 0 is pres- ent In most kinetic studies the observed variation of the rate of photomineralization as a function of [O,] is described very well by equation 6Thus a double reciprocal plot of the data yields a straight line and from the ratio of the intercept to the gradient of the straight line a value of K(0,) can be obtained Figure 8 illustrates plots of RCO2 versus %O and l/Rco versus 1/%0 determined23 as part of a kinetic study of the photomineraliza- tion of 4-chlorophenol by Degussa P25 TiO from this work a valued of 0 044 f0 005 kPa-was determined which is a typical for such work Thus increasing the 0 concentration from air- saturated (20% 0,)to 0,-saturated (100% 0,)conditions will typically only increase the rate by a factor of I 7 Oxidants other than O such as H,O S208,and Ag+ ions have been used with some success but in terms of a practical approach to water purification only 0 and H,O appear a viable choice of oxidant Although H,O does enhance the rate of photomineralization of some organics byTiO with others it 1/%0 000 001 002 003 004 005 006 19 16 13 52 10 20 40 60 80 100 120 %O Figure 8 Plots of the initial relative rate of CO generated in the photomineralization of 4-chlorophenol sensitized by TiO I e Rco versus [O,] and l/Rcoz versus l/[O,] Irradiations were carried out using the batch reactor illustrated in Figure 6(a) with pollutant = 4-chlorophenol (100 cm3 mol dm-3) [Ti0,]=05 mg cm 3 pH = 2T = 30°C The results of a least-squares analysis of the ddtd points in the double-reciprocal plot are as follows grddi-ent = 18 5 f2 4 %O intercept = 0 81 f0 07 and correlation coef- ficient = 0 9841 Data from reference 23 has a negative or no effect it has been suggested that H,O or the hydroperoxy radical may in fact function as a hydroxyl radical scavenger 2 2 5 Light Intensit) In most kinetic studies of reaction 1 the light intensities used have been such (typically > 6 x 1014 ultra-band gap photons cm- s ') that the rate has usually been found to be pro- portional to [:However some workers have successfully tested for the theory-predicted transition in dependence of rate from first-order to half-order as the light intensity is increased from low to high levels A notable example of this is the work of Egerton and King,24 on the photo-oxidation of isopropyl alcohol to acetone (on its way to complete mineralization) by rutileT10 in air-saturated solution the results of this work are illustrated in Figure 9 13 14 15 16 17 18 19 Loglo (light intensity) Figure 9 Rate of acetone generation from isopropyl alcohol photo- sensitized by pure rutile (400 mg in 40 cm3) versus incident light intensity Data from reference 24 2 2 6 Temperature The overall process of semiconductor photocatalysis is not usually found to be very temperature sensitive thus reported activation energies generally lie in the range 5-16 kJ mol-' 2 2 7 pH and Anions The pH of the aqueous solution significantly affectsTiO including the charge on the particles the size of the aggregates it forms and the positions of the conductance and valence bands It is somewhat surprising therefore that the rate of photocata- lysis IS not usually found to be strongly dependent upon pH typically varying by less than an order of magnitude from pH 2 to pH 12 Higher reaction rates for various Ti0,-sensitized photomineralizations have been reported at both low and high PH Matthews and his co-workersZ5 have carried out one of the most rigorous studies on the effect of anions on the rate of photomineralization of organic pollutants (salicyclic acid ani- line and ethanol) sensitized byTi0 From the results of this work it appears that perchlorate and nitrate have very little effect whereas sulfate chloride and phosphate at concent- rations > lop3mol dm-3 can reduce the rate of photominerali- zation by 20-70% due to their absorption at the oxidation sites on the Ti0 We have recently found that nitrate at concent- rations of ca 0 4 mol dm-3 can indeed reduce the rate of Ti0,- sensitized photomineralization of 4-chlorophenol by 50% not by blocking oxidation sites but rather by UV screening the T10 particles 23 2 2 8 Wear Eficiencj and Costings Most workers have not noticed an appreciable loss in the photocatalytic activity of Ti0 with its repeated use as a CHEMICAL SOCIETY REVIEWS 1993 photosensitizer for reaction 1 Figure 10illustrates the results for a typical set of successive kinetic runs in which 4-chlorophenol (0 155 x mol dm 3 is the pollutant in reaction 1 andT10 the photosensitizer 26 If semiconductor photocatalysis is to be used extensively as a method of water purification then much more significant extended wear tests will need to be carried out than have been to date 0 200 -400 600 800 1000 1200 time/mins Figure 10 Ten successive photomineralization cycles involving 4-chlor- ophenol and the same portion ofTiO The dotted lines correspond to the readjustment of the 4-chlorophenol concentration to LU 0 155 x 10 mol dm and to a dark period of 20 min so thdt the adsorption equilibrium is dttained exciting light 3 340 nm Data from reference 26 If semiconductor photocatalysis as a method of water purifi- cation is to have a viable commercial future its operating costs must compare favourably with those of its competitors A major component of these costs will be that of the electricity used to generate the necessary UV lightThe manufacturers of the 8 W blacklight (UVA) bulbs used in our work on the Ti0,-sensitized photomineralization of 4-chlorophenol quote a UV output at 344 nm of 0 8 W I e 10% electricity to light conversion efficiency Ferrioxalate actinometry carried out using the batch photoredctor illustrated in Figure 6(a) showedz3 that one 8 W blacklight bulb generated 5 x 1OI6 photons s in the wave- length region 300400 nm if we assume for the sake of simplicity that all these photons are of wavelength 355 nm then the electricity to light conversion efficiency is only 0 35% in our system it is not at all clear why this figure should be so very different to that reported by the manufacturer Using the photoreactor illustrated in Figure 6(a) with 11 lamps on the initial rate of destruction of 4-chlorophenol was determined as 6 x 10' molecules s and therefore the formal (maximum) quantum yield for the photomineralization of this pollutant is 0 01 I In most photomineralization studies the reported quantum yield lies typically in the range (1-0 l)Yo In 1987 0111s carried out a preliminary comparison of the process economics associated with the removal of PCBs from waste water using activated carbon UV-ozone and a near-UV semiconductor photocatalysis system 27The calculated operat- ing costs updated from 1987 to 1993 for a range of system sizes are given inTable 2 Although it is just a preliminary study the calculations do indicate that heterogeneous semiconductor pho- tocatalysis could be economically comparable with activated carbon systems on intermediate to large size water purification systems 2 2 9 4-ChlorophenollTi0,lAir A Standard Test Ststem As can be seen from Table 1 research into the photomineraliza- tion of organic compounds sensitized by semiconductors has progressed to such an extent that there is a real need to define a standard test system since it would help facilitate a comparison of results between groups using different experimental con- WATER PURIFICATION BY SEMICONDUCTOR PHOTOCATALYSIS-A MILLS R H DAVIES AND D WORSLEY Table 2 Estimated process costs for different water purification systemsu System size MGDh 0029 0 115 023 092 244 Cdrbon $779 425 3 19 221 195 UV ozone $1300 632 492 383 3 10 uv phOtOcdtdlySiS’ $985 436 321 232 200 The 1993 costs have been estimated from Ollis s original 1987 costsZ7 by assuming a 10’ o per year rise giving a multiple of 1 77 0111sassumed that the labour maintenance requirements will only be 2 4h/day MGD = million of gallons per day Incinerative regenerator included in larger sizes only No Ldrbon dispo\al costs included in small units ’ UV ozone process minus ozone generation dissolution ditions such as alternative semiconductors pollutants reac- tant concentrations irradiation sources and temperatures At present so disparate are the conditions used by the different groups in their research on semiconductor photocatalysis that such comparisons are usually meaningless As noted before the commonly accepted standard form of Ti0 for such work is Degussa P25TiO and this is usually used under air-saturated conditions Of the many organics studied in the photomineralization of pollutants sensitized byTiO 4-chlorophenol has attracted particular attention and appears a good candidate for a standard test organic pollutant We have suggested that Degussa P25 Ti0,/4-chlorophenol/air-saturated aqueous solution at pH 2 should be adopted as a standard test system for water purification by semiconductor photocalysis and in order to characterize this system thoroughly we have carried out a detailed study of the kinetics of photomineraliza- tionZ3 and the intermediatesZo involved in this standard test system 2 2 10 Methylene BluelTi0,lAir A Simple Visual Demonstration of Semiconductor Photocalysis Methylene blue is a brightly coloured commonly available water-soluble dye which is stable in air-saturated solution when irradiated with light of wavelength > 300 nm It has been established28 thatTi0 is able to photocatalyse the complete oxidative mineralization of methylene blue by 02,i e Ti0 air Ci6H18N3SC1 + 25:02 16C0 + 6H,O + 3HN03 + methylene blue H,SO + HCI (7) Because of the bright colour of the dye ‘pollutant’ the methylene blue/TiO,/air system provides a simple visual demonstration of the mineralization of organic pollutants by 0,,photosensitized byT10 In a typical demonstration using the batch irradiation system illustrated in Figure 6(a) 10mg of Degussa P25 Ti0 are dispersed in 100 cm3 of air-saturated water containing 1 05 x 10 mol dm methylene blue which is sufficient to turn the Ti0 dispersion very blue This reaction solution is then placed in a 100 cm3 Dreschel bottle and then irradiated with the 12 x 8 W UVA lamps of the photoreactor The variation of the absorbance due to the dye versus irradiation time is illustrated in Figure 11 and shows that the dye has apparently all been photomineralized within 6 min In fact it can be shown that the dye is bleached within 6 min and colourless intermediates have been generated quantitative evolution of CO according to the reaction stoichiometry in equation 7 takes about three times as long as the bleaching 2 2 11 Other Novel Applications ofTiO Photocatalysis There is a growing interest in the application of T10 photoca- talysis in areas other than the photomineralization of organic -16 I 1 0 4 8 time/min Figure 11 Plot of dbsorbance at 660 nm due to methylene blue (1 05 x 10 mol dm 3 recorded ds d function of irradidtion time in the presence of Ti0 (10 mg in 100cm3 of dir sdturated water) where the normalized absorbance of 1 IS equivalent to d true dye dbsorbdnce of 0 78 In this work the methylene blue samples were tdken from the reaction solution at reguldr intervdls dnd both dbsorbdnce dnd dbsorption spectra were recorded (see inset) pollutants and some of these other uses are listed inTable 3 Thus Matthews and his co-workers have reported that a Ti0 photocatalyst for reaction 1 can be used to carry out total organic carbon analysis (0 1-30 pg cm 3 in < 10 min the CO generated being detected quantitatively using a conductivity cell In this work it appears to be assumed that all organics will be totally mineralized by this technique despite the fact that there are known exceptions An instrument for the rapid deter- mination based on Ti0 photocatalysis of organic carbon in water has just been made commercially available -the SGE ANATOCTM Another intriguing use ofTi0 as a photocatalyst is in mediating the destruction of cancer cells 3o Not surprisingly the workers involved believe that the photoinduced death of the cancer cells was due to attack from photogenerated hydroxyl Table 3 Other novel applications of Ti0 photocatalysis Application Ref Solar to chemical energy conversion a Photoelectrochemical detector (photocurrent measured) for h flow injection analysis and liquid chromatography Photocatalytic oxidation system for total organic carbon c dnalysis Photoinduced cytotoxic action towards cancer cells d Photodeodorizer for kitchens and bathrooms e Photo-oxidation of oil slicks using Ti0 coated hollow glass f micro beads Recovery of platinum group metals from industrial wastes or gdilute solution Photoinduced detoxification of cyanides h K KdlydndSUnddrdm in Energy Resources through Photochemistry dnd Catalysis ed M Gratzel Academic Press New York 1983 Chapter 7 G N Brown J W Birks andC A Koval Anal Chem 1992 64 427 434 R W Matthews M AbdullahandG K C Low Anal Chim Acia 1990 233 171 179 * R Cdi Y KubotaT Shuin H Sdkei K Hdshimoto and A Fujishima Cancer Research 1992 52 2346-2348 T ogawd T Saito T HdSegdWd H Shinozaki K Hashimoto and A Fujishima in The First International Conference on Ti0 Photocatalytic Purification and Treatment of Water and Air Book of Abstracts London Ontario Cdnddd November 1992 p 192 I Rosenberg J R Brock andA Heller J Phby Chem 1992 96 3423 3428 E Borgarello R Harris and N Serpone Now J Chm 1985 9 743 N Serpone E Borgdrello and E Pelizzetti in Photocatalysis and EnvironmentTrends and Applications ed M Schiavello KIuwer Academic Publishers Dordrecht 1988 p 499 radicals and hydrogen peroxide It is surprising that more work is not in progress in the general area of photocatalytic steriliza- tion by Ti0 as a means of destroying bacteria viruses and cancer cells Ti0 photocatalysis can be used to destroy volatile organics in air thus it has potential as a photodeodorant It has been suggested3 that Ti0 could be used to deodorize cars kitchens and bathrooms Some research workers have even gone as far to develop an interior tile with a ca 1 pm thick Ti0 film for this purpose It has also been suggested32 recently thatTi0 photocatalysis could be used to assist in the clean-up of oil spills To this end Ti0 has been coated in part onto low-cost hollow glass microbeads (50-200 pm diameter) which float on water and are readily separated These oleophilic Ti0 photocatalyst coated beads clean up small-scale experimental oil spills by (1) stripping and aggregating the oil followed by mechanical clean-up (11) photoassisting the oxidation of the oil and (111) controlling the combustion of the oil films on water 3 Conclusion The purification of water by semiconductor photocatalysis is a rapidly growing area of interest to both research workers and water purification companies In this process invariably the pollutant is organic and the semiconductor isTi0 The range of organic pollutants which can be completely photomineralized by oxygen using Ti0 as the sensitizer is extensive and includes many aliphatics aromatics detergents dyes pesticides and herbicides T10 photocatalysis can also be used in (1) the measurement of total organic carbon (11) the killing of cancer cells (111) the removal of unwanted odours and (iv) the cleanup of oil spills 4 Appendix 4.1 Photochemical Reaction Schemes E-witation iiv 2 3 2 eV Ti0 -+h+e Back Reaction e +h+- heat ('4-2) Trapping Ti'"(0H /H,O) +h+- Ti"-OH' (A3a) Ti"-OH'- Ti'"(0H /H20)+h+ (A3b) +e KA4a-TI"' (A44 KA4hTi"' +0 -Ti"-O; (A4b) In the absence of 0 no photomineralization occurs thus the efficiency of trapping e ,8= is a function of [O,] (which is thought to be noncompetitively adsorbed onto the surface of the Ti0 atTi"' sites) If KA4b [O,] >> 1 it can be shown that aT (= {[Ti'I'] +[TiIV-O' ])/([e ] +[Ti"'] +[TiIV-0; I)) is given by the following expression Absorption of organic pollutant (S) and photogenerated interme- diates (Q,) site +S -k rd site +Q -Sdds (A51 Qj ads (A6) CHEMICAL SOCIETY REVIEWS 1993 Hvdroxyl radical attack on the initial organic pollutant generut- mg intermediates TI"-OH' +S-TI'" +Q (A7b) Hydroxyl radical attack on intermediates generating other intermediates where the different intermediates are identified by the different integer values ofj The parameterj will have all values which lie in the range m (= total number of intermediates generated) 3j 3 1 Other reactions e +TI'~-O; +2H+ =TiiV(H202) (A91 TI"-O; +H+ =TI"(HO;) Ti"(H207) +Ti"-OH' =TI'"(HO:) +TlIV(H,O) (A1 1) Reactions (A9-Al1) indicate that the reduction of 0 by e can lead to the formation of radicals such asTiIv(H0;) which may be capable of oxidizing the organic pollutant S As a result under steady state conditions (of electron [e I and hole [h+] concentration) the overall rate of formation of oxidizing radi- cals which is usually taken to be =8TkA3 [TiIV(OH / H,O)][h+], may be best expressed as = u8TkA3,[T1'V(OH /H,O)][h+], where for any irradiation system the term u is a constant with a value which lies in the range 1 d u <2 4.2 Kinetic Equations High absorbed light intensity Inbs where 4 is the quantum yield for reaction A 1 A is the area of a Ti0 particle normal to the illumination and Vp = volume of a Ti0 particle Low absorbed light intensitj where A is the surface area of aTi0 particle Case A The destruction of S occurs via hydroxyl radical attack on adsorbed S i e equation 7a and in the other major reactions all intermediates are adsorbed on the surface of the Ti0 [TI'V-OH'],s = aGTkA3d[T1'V(0H/H20)l[h+lSS (A 14) m kAjb +kA7a[Sadsl + 1kQ ads[QI adsl JI -d[S]/dt =~A,~[S,~~][TI'~-OH']~~AS (A 15) Case B The destruction of S occurs via hydroxyl radical attack of the non-absorbed S I e equation A7b and in the other major reactions all intermediates and S are not absorbed on the surface of theTi0 WATER PURIFICATION BY SEMICONDUCTOR PHOTOCATALYSIS-A MILLS R H DAVIES AND D WORSLEY Summar\ The final rate equations for both cases A and B have the same form I e where the different parameters are defined as follows CASE A CASE B KS kA7dKA5 [sltelAS/kA3 b kA7 blkA3 h K~~ kQj adsKQj ads[SitelAS/kA3b kQj/kA3h 5 References 1 C F Goodeve and J A Kitchener Trans Faraday Soc 1938,34 570 2 S P Pappas and R M Fischer J Paint Tech 1974,46,65 3 R W Matthews in 'Photochemical Conversion and Storage of Solar Energy ed E Pelizzetti and M Schiavello Kluwer Academic Publishers Dordrecht 1991 p 427 4 D F Ollis E Pellizetti and N Serpone in 'Photocatalysis Funda- mentals and Applications ed N Serpone and E Pelizzetti Wiley- Interscience New York 1989 Chapter 18 5 M A Fox and MT Dulay Chem Rev 1993,93,341 6 A Fujishima and K Honda Nature (London) 1972,238 37 7 A V Bulatov and M L Khidekel Zzv Akad SSSR Ser Khim 1976 1902 8 T Sakata and T Kawai Nouv J Chim 1981,5,279 9 A L PrudenandD F Ollis J Catal 1983,82,404 10 C -Y Hsiao C -L Lee and D F Ollis J Catal 1983 82 418 11 D F Ollis C -Y Hsiao L Budiman and C -L Lee J Catal 1984 88,89 12 R W Matthews J Catal 1986,97 565 13 M Barbeni E Pramauro E Pelizzetti E Borgarello M Gratzel and N Serpone Nouv J Chim 1984,8 547 14 K Okamoto Y Yamamoto H Tanaka and A Itaya Bull Chem SOC Jpn 1985,58,2023 15 S Morris 'Photocatalysis for Water Purification' Ph D Thesis University of Wales 1992Table I 1 and references therein 16 M Barbeni E Pramauro E Pehzzetti E Borgarello and N Serpone Chemosphere 1985 14 195 17 J Cunningham and G Al-Sayyed J Chem Soc Faraday Trany 1990,86 3935 18 C S Turchi and D F Ollis J Catal 1990 122 178 19 A Mills S Morris and R Davies J Photochern Photobiol A Chem 1993 71,285 and reference therein 20 A Mills and S Morris J Photochem Photobiol A Chem ,1993,70 183 21 R W Matthews Solar Energy 1987,38 405 22 R W Matthews J Catal 1988 111 264 23 A Mills and S Morris J Photochern Photobiol A Chem ,1993,71 75 24 T A Egertonand C J King J OiICol Chem Assoc ,1979,62,386 25 M Abdullah G K C Low and R W Matthews J Phvs Chem 1990.94.6820 26 G Al-Sayyed J -C D'Oliveira and P Pichat J Photochem Photo- biol A Chem 1991 58 99 27 D F Ollis in 'Photocatalysis and Environment Trends dnd Appli- cations' ed M Schiavello Kluwer Academic Publishers Dor- drecht 1988 p 663 28 R W Matthews J Chem Soc Faraday Trans I 1989,85 1291 29 R W Matthews,M Abdullah,andG K C Low Anal Chim Acta 1990,233 17 1 30 R Cai Y Kubota T Shuin H Sakai K Hashimoto dnd A Fujishima Cancer Research 1992,52 2346 31 T Ogawa T Saito T Hasegawa H Shinozaki K Hashimoto and A Fujishima in 'The First International Conference on Ti0 Photocatalytic Purification and Treatment of Water and Air Book of Abstracts' London Ontario Canada November 1992 p 192 32 A Heller and J R Brock in ref 3 1 p I7 and references therein

ISSN:0306-0012    

DOI:10.1039/CS9932200417

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Mechanisms of Solvolytic Alkene-Forming Elimination Reactions Alf Thibblin Institute of Chemistry University of Uppsala P. 0. Box 531 Uppsala S-751 21 Uppsala Sweden 1 Introduction Solvolytic elimination is often defined as elimination that is promoted (induced) by the solvent (SOH) i.e. the solvent acts as the hydron-abstracting base.' This review will discuss elimi- nation in a broader sense including all types of alkene-forming elimination reactions that accompany substitution by the sol- vent. Thus elimination through a carbocation intermediate as well as concerted pericyclic elimination (thermal or pyrolytic) and solvent-promoted elimination (E2) will be treated. It will also include heterolysis reactions in non-nucleophilic solvents in which an alkene is formed by a stepwise reaction through a carbocationic intermediate.The various types of mechanisms are summarized in Table 1. The recently published IUPAC recommendations for representation of reaction mechanisms are given in the last column of this table.2 Most of the review will deal with 1,2-elimination but exam- ples of 1,4-elimination will also be discussed. A few representa- tive examples of reactions that follow each type of mechanism will be presented. For a more complete account of solvolytic elimination reactions see for example a monograph of Saunders and Cockeril13 and a review by Baci~cchi.~ 2 Extreme Kinetic Deuterium Isotope Effects as a Probe of Reaction Branching Solvolysis reactions of secondary and tertiary substrates having at least one hydron in the P-position generally provide both a substitution product by reaction with the solvent or added nucleophile and an elimination product. Good nucleophilic solvents favour substitution and elimination is often minor under such conditions. When the experimental results indicate stepwise solvolysis the substitution and the elimination have frequently been postulated for the sake of mechanistic simpli- city to occur via a common carbocation intermediate. As indicated in equation 1 the common intermediate may be an ion pair or a free diffusionally-equilibrated carbocation that is formed by reversible or irreversible ionization. It may also be a carbocation-molecule pair (iondipole intermediate) formed from a substrate RX+ with a neutral leaving group e.g. an ammonium salt or a protonated ether or alcohol. As will be shown below the magnitude of the kinetic deuterium isotope effects for the separate reactions using substrates deuteriated in AlfThibhlin was born in Eskilstuna Sweden in 1949. He received the Ph. D.degree in chemistry in 1977at the University of Uppsalcr (with P. Ahlberg). After post- doctoral work at Brandeis with W. P. Jencks .for one year (77-78) he joined the Facultjq of Chemistry at the University of Uppsala He has now a research position financed by the Swedish Natural Research Council. His principal research interest lies in kinetic and me- chanistic investigation of reac-tive intermediates of carho-cation and carbanion type in so1volysis and hydr on -transfer reactions. Table 1 Mechanistic classification of solvolytic elimination reactions Dehydrona-Kinetic IUPAC tion of Base order Description Symbol symbol2 RX B- SOH 2nd" RX 1st R+ X-SOH B 2nd"-X-1st R+X-SOH B- X-1st R+ SOH B-1st concerted E2 concerted El pericyclic preequilibrium E2 ion pair irreversible El ion pair carbocation El AxhDHDN cyclo-DHDNA" DEA,,Dk Dd*AXhDH Dd+AxhDII " Pseudo-first order if solvent IS the base. the P-position may provide experimental evidence for branching through a common intermediate. Enlarged and attenuated isotope effects have been employed extensively in this laboratory as a mechanistic probe of coupled reactions through a common intermediate.5The probe was first used in studies of competing elimination and proton-transfer reactions that were shown to proceed through common hydro- gen-bonded carbanion intermediates.The method has also been found to be very useful in probing common intermediates in carbocation O Several of the reactions that will be discussed in this review have been studied using this probe and therefore it will be discussed in general terms in this section as an introduction. The mechanistic model of equation 1 corresponds to the relations between the phenomenological rate constants ksandkE for formation of substitution and elimination product respecti- vely and microscopic rate constants as described by equations 2-4. ROS k2 //RX 2 R+X-or R+ k \ k3 alkene ks = k,k,/(k_ + k + k3) kE = k,k,/(k-1 + k + k3) ks + kE= k,(k + k,)/(k-+ k + k3) The expressions for the isotope effects are Reaction branching may cause enlarged and attenuated iso- tope effects as will be discussed below.The isotope effect kp/kpis a secondary isotope effect and is expected to have a value of 1.10 f0.05/P-D." Also ky/kyis a secondary isotope effect; the 427 value should be close to unityThe primary isotope effect ky/ky on the other hand should have a substantial value l2 Let us assume for simplicity that internal return is negligible (k1<< k k3) It can be inferred from equation 6 that the isotope effect on the elimination reaction attains a maximum value of kE/kp= (kF/ky)(ky/k$’)when elimination is much slower than substitution (ky/ky)The isotope effect on the substitution reaction (equation 5) attains under these conditions a maximum value of kp << kg = ky/ky On the other hand fast elimination i e ky >> k? yields a minimum elimination isotope effect of kE/kF = ky/kf,and a minimum substitution isotope effect of kp/kp= (kF/k~)(k?/k~)(k$’/ky),which can be approximated to kp/kp= (ky/kY)(ky/ky) Fast internal return (k->> k k3)yields kp/kg = (ky/kf) (kY/k$’)(kPl/kk! 1) i e kE/kF = (ky/ky)(ky/k$’)This has the same effect on the elimination isotope effect as fast substitution Thus the observed elimination isotope effect is enlarged owing to multiplication by a factor that is larger than unity The above analysis shows that competition between ebmina- tion and substitution that occurs through a common intermediate may give rise to an enlargedkinetic deuterium isotope eflect for the elimination reaction and an attenuated isotope eflect for the substitution reactionThe relative amounts of the elimination and substitution products and the extent of internal return govern how large these eflects will be 3 Concerted Solvent-Promoted Elimination Concerted elimination with the solvent acting as the base may be considered to be the mechanistically most simple of the mechan- istic types of Table I Despite this simplicity the mechanism is very difficult to establish If the solvent is able to abstract a hydron the conjugate base (SO-) of the solvent or another added base of higher basicity than the solvent should be an even more efficient hydron abstractor as shown graphically in Figure 1The experimentally obtained rate constant (kg) which is measured without added base B- could be the pseudo-first- order reaction rate constant for a concerted solvent-promoted reaction or the first-order rate constant of a concerted pericyclic reaction It could also be the macroscopic rate constant of a multistep reaction through a carbocation or ion-pair inter- mediate When the lifetime of the carbocationic species is too short to be considered as an intermediate (t+< s) the reaction is enforced to be concerted l3 This situation will be discussed below in connection with ion-pair intermediates 0 [B-I Figure 1The expected increase in elimination rate constant as a function of added strong base (see Section 3) The solvolysis of a sulfonate ester in formic acid has been suggested to involve a concerted solvent-promoted elimination reaction (equation 7) l4 However there seems to be no conclu- sive evidence that the solvent is the active base formate anion within an ion pair (generated in the first step) or the leaving group might be the hydron acceptor 4 Stepwise Elimination via a ‘Free‘ Solvent-Equilibrated Carbocation (El ) The other extreme mechanism is the classical El mechanism The elimination occurs through ionization of the substrate to CHEMICAL SOCIETY REVIEWS 1993 NY” give the carbocation ion pair or the ion-molecule pair if the leaving group is a neutral speciesThe dissociation of this complex is much faster than its reaction to elimination product A parallel route to substitution product by reaction with solvent or added nucleophile has usually been assumed to occur by nucleophilic attack on the very same carbocation (equation 8) ROS k “/RX AR+ k\ alkene Pure El reactions as defined above are probably not common for solvolytic reactions of substrates with leaving groups that are negatively charged or are neutral but efficient bases There seems to be no evidence that solvolytic elimination from such sub- strates occurs mainly or exclusively from the ‘free’ diffusionally equilibrated carbocation The leaving group of an ion pair is often very efficient in promoting elimination by abstraction of a p-hydron even in highly aqueous media It has been reported that 1,l-diphenyl- ethyl derivatives Ph,CMeX react to elimination product through the ion pair l5The leaving groups AcO- and p-nitrobenzoate give rise to three times as much elimination as the leaving groups MeOH and HOAc in 20 vol% Me,SO in water However dissociation of the ion pair has been concluded to be faster than elimination since the measured nucleophilic selectivi- ties are very similar with different leaving groups Despite the relatively high stability of the free carbocation and the highly aqueous solvent most of the elimination arises from the ion pairs The free carbocation obviously yields less elimination than the ion pairs The solvolysis of (1) in ethanol at 57 3“Chas been found to provide 28% of alkene and 72% of ether (equation 9) l6The (9)\ MECHANISMS OF SOLVOLYTIC ALKENE-FORMING ELIMINATION REACTIONS-A THIBBLIN corresponding compound with only one ortho substituent reacts slower but gives a greater yield of alkene product The formation of the free carbocation was concluded to be rate limiting since a low concentration of triethylamine (3 5 mM) or sodium ethox- ide (3 I mM) does not significantly change product composition or overall rate The El mechanism (equation 8) was proposed A Hammett p of -2 5 was calculated from rate data of (I) combined with rate data for the substrate having a bromo- substituent on the 9-fluorenyl ring The kinetic deuterium isotope effect for the disappearance of the substrate (1) deuteriated in the ,&position was measured as k%,/k,Db”,= 1 62 at 57 3°C and was concluded to be in accord with a secondary isotope effect However the maximum second- ary P-deuterium isotope effect for this type of process has been estimated to be kH/kD2= 1 32 at 25 “C I e 1 l5//3-DThis discrepancy is probably due to significant internal return from an ion-molecule pair R+NMe It is very likely that the basic leaving group NMe promotes elimination by abstracting one of the benzylic hydrons and that substitution by the solvent occurs via the solvent-separated ion-molecule pair or the free carboca- tion Such a mechanism has been proposed for the solvolysis of a substrate with pyridine as leaving group (see Section 5)The result of an analysis of the kinetic deuterium isotope effects for the separate steps [which can be derived from the reported product ratios and the overall rate constant of (l)] is consistent with a branched mechanism (equation 10) through the ion- molecular pair 5 Irreversible Ion Pair (or Ion-Molecule Pair) (€7 IP) Rate-limiting formation of the ion pair followed by fast dehyd- ronation characterizes the El mechanism (equation 10 k-<< k + k,) The substitution product does not necessarily come directly from the ion pair It is presumably more common with a multistep process involving nucleophilic attack on the solvent-separated ion pair or the free carbocation Consistently an increase in the ionizing power of the solvent generally decreases the fraction of elimination at the expense of substitu- tionThe solvent-separated ion pair has been excluded from the reaction scheme of equation 10 for simplicity It is not needed for the following discussion ROS k ’/RX $ R+X-k\ alkene where k =k; + k;”B ] The solvolysis of the very reactive 1,l-diphenyl-1-chloroeth-ane [(2) equation 113 in mixtures of acetonitrile with methanol has been concluded to go through an almost irreversibly formed ion-pair intermediate O Analysis of the kinetic deuterium iso- tope effects for the elimination and substitution processes strongly indicates that these competing reactions have a common intermediate Very significant base catalysis from the leaving group indicates that this intermediate is the contact ion pair Ph,CCl 5 Ph,COMeICH,(D3) <AD3) I (1 1) (2) Ph,C=CH,(D,) Let us look at some of the experimental dataThe changes in the measured kinetic deuterium isotope effects are in accordance with the analysis of the isotope effects for competing elimination and substitution given above (see Section 2) The elimination isotope effect kF/kE3increases from 1 73 to 3 20 at 25°C when the fraction of substitution increases from 0 to 44% The substitution isotope effect kF/kp3 increases with increasing methanol content from 0 84 to 0 96 corresponding to 13 and 44% substitution respectively It does not seem possible to accommodate these isotope-effect values and trends in isotope- effect values in a reaction scheme with two competing parallel reactions that do not have an intermediate in common How-ever branching through a common intermediate as shown in equation 10 may account for the resultsThe intermediate undergoes some internal return at low methanol fraction of the solvent as indicated by the isotope effect value kH/kD3= 1 73 at 0% methanol Addition of a strong base increases the fraction of olefin product the fraction is larger for solvolysis in 2M sodium methoxide than in methanol Catalysis from added base has been found in some other solvolysis reactions via carbocationic intermediates 5-10 The ion pair was estimated to eliminate > 3000 times faster than the free carbocation in 0 4 vol% water in acetonitrileThe acid-catalysed El elimination of water from the corresponding alcohol in 25 vol% acetonitrile in water which involves rate- limiting hydron-transfer from the free carbocation (or possibly from the ion-molecule pair) is very sensitive to isotopic substitu- tion kE/kp3= 6 5 lo The substitution reaction with water which gives back starting material is a much faster reaction than elimination I e k >> k (equation lo) which in accord with equation 6 results in an enlarged elimination isotope effect Other examples of reactions following the El mechanism are the hydrolyses of (3) in water-acetonitrile mixtures (equation 12) ’The chloride was found to yield about 64% of olefin (4) and 36% of alcohol and only a trace of the thermodynamically more stable olefin (5) was formed in 25 vol% acetonitrile in water at 25 “CThe addition of the common ion C1- does not depress the disappearance of the substrate (kobs) substantially but catalyses the formation of alkene (4) This indicates rate-limiting ioniza- tion Also weak bases as well as the leaving group catalyse elimination from the ion pair The catalysis from substituted acetate anions was found to be small p = 0 05 As shown L=’H or ‘H X = CI,Br or OAc ” AL3C CL (5) in Figure 2 the catalysis from halide anions is described fairly well by the same Brsnsted line The presence of strong base on the other hand opens up a parallel bimolecular concerted elimination route (E2) l8This route provides exclusively the more stable olefin (5)and exhibits a large kinetic isotope effect of kH/kD= 8 1 (substrate deuteriated at the 9-position of the fluorene moiety) The intermediate showed very small discrimi- nation between the nucleophiles azide anion methanol and water Thus an azide anion is about five times more reactive CHEMICAL SOCIETY REVIEWS 1993 1 1 I I Br--IC H20 n IUI II 1 I -10 -8 -6 -4 -2 0 2 4 6 8 PK Figure 2 Br~rnstedplot for the dehydronation of the ion-pair inter- mediate formed from (3) (X = C1) with substituted acetate anions (e) in 25 vol% acetonitrile in water ionic strength 0 75 M maintained with sodium perchlorate The pK values refer to water than a solvent molecule towards the carbocation intermediate e ,kN,/kH -5The selectivity is so small that it may represent reaction within a pool of solvent molecules that are present at the time of ionization to the ion pair The rate constant for the reaction of the intermediate with water to form the alcohol was estimated at -4 x 10los based upon a diffusion-controlled reaction with azide anion with kd = 5 x lo9 M 's l7 l9 Accordingly the rate constant for deprotonation of the interme- diate by solvent water is -7 x 10'O s These rate constants are larger than or at least comparable to the estimated rate of diffusional separation of the ion pair Thus it was concluded that the dehydronation of the intermediate and the nucleophilic substitution are processes that occur mainly at the ion-pair stage before the ion pair undergoes diffusional separationThe elimi- nation reaction promoted by addition of acetate anion should occur by a stepwise preassociation mechanism in which the base comes into reaction position for hydron abstraction before the ionization to the ion pair The measured kinetic isotope effects support the equation 10 mechanism with kk << k + k The isotope effect on the disappearance of the substrate having the methyl groups fully deuteriated was measured as kH/kD6= 2 2 at 25°C This large secondary kinetic /3-deuterium isotope effect which corresponds to a value of I 14 per deuterium shows that the bonds to the hydrons are weakened considerably in the ionization stepThe kinetic isotope effect on substitution and elimination for the solvolysis of the chloride were measured as kp/kp6= 1 4 and kp/kE6= 3 7 These isotope effects are in accord with a mechan- ism in which a rate-limiting ionization step is followed by branching The competing paths show differences in sensitivity to isotopic substitution Owing to this competition the isotope effects on ks and kE are attenuated and enlarged respectively compared with the isotope effects on the rate-limiting ionization of the substrate The experimental data for the chloride in 25 vol% acetonitrile in water and the mechanistic model are consistent with ky/ky6= I 0 and ky/ky6= 2 8 (equation 10) The consistency of the measured isotope effects with equa- tions 4-6 both at low and high water concentration indicates that internal return from the ion pairs is not significantThe large ionization isotope effect kH/kD6= 2 2 suggests that the ionization is accompanied by considerable reorganization of the carbocation structure and the solvent These processes slow down the collapse of the ion pair back to covalent material 2o There are rather many examples in the literature on elimina- tion promoted by the leaving group from carbocation inter- mediates 21 Usually the alkene is formed in minor amount the substitution product being dominant in nucleophilic solvents An example is cumyl derivatives I e 2-X-2-phenylpropane which solvolyse in 25 volo/~ acetonitrile in water mainly to alcohol accompanied by 2-phenylpropene (equation 13)The kinetic results with hexadeuteriated substrates suggest a branched mechanism in which the elimination and the substitu- tion go through the same ion-pair intermediate Internal return is probably slow but there is no conclusive evidence for this The nucleophilic substitution occurs via the solvent-separated or the free carbocation but the elimination was concluded to be promoted by hydron-abstraction by the leaving group The elimination is also catalysed by added general bases The Brsnsted parameter value of /3 = 0 13 for cumyl chloride indi- cates an early transition state It is consistent with the relatively small kinetic isotope effect ky/ky6= 3 5 f0 2 that was mea- sured with acetate anion at 25 "C L = 'H or 2H Cleavage of carbon-carbon bonds in heterolysis of cumyl derivatives has been reportedThus the reaction of PhCMe,C(CN) in Me,SO yields 2-phenylpropene in almost quantitative yield,22 presumably through the ion pair The reaction is five times slower than the total rate of solvolysis in methanol The heterolysis of t-BuC(CN),NO to 2-propene in Me,SO was found to be eleven times faster than the heterolysis of t-BuC1 23 It is not clear if internal return is fast in these reactions It has also been suggested that the elimination from the cumyl derivative with pyridine as leaving group is promoted by the departing pyridine methanol as leaving group gives eight times less of olefin Thus the results indicate that the elimination occurs through the ion-molecule pair Both 1,2- and 1,4-elimination have been reported for the solvolysis of the allylic isomers (6)-OAc and (7)-OAc (equation 14)The elimination reactions are catalysed by the leaving group as well as by adding general bases Brransted parameters for the deprotonatron of the carbocation intermediates by acetate anions were measured as /3 = 0 16 and /3 = 0 14 for formation of alkene (8) and (9) respectively The kinetic and product data are consistent with the mechanism shown in equation I5 Two discrete ion-pair intermediates must be involved since product compositions are quite different for the two isomeric acetates The ionization of these acetates is not completely irreversible and some internal return accompanies the elimination and dissociationThe measured kinetic deuterium isotope effects for the acetates (deuteriated in the benzylic position as shown in equation 14) supports this mechanistic interpretation The 1.4- elimination of HOAc from (6)-OAc and (7)-OAc are not 'true' 1,4-elirninations but occur through intramolecular allylic re- arrangement of the ion pairs followed by 1,2-elimination pro- moted by the leaving acetate anion Both acetates undergo 1,2- elimination faster than 1,4-elimination 6 Pre-Equilibrium Ion Pair (or Ion-Molecular Pair) (El,) Elimination reactions that follow this type of mechanism involve rate-limiting hydron transfer directly from the ion pair The mechanism is otherwise quite similar to the El mechanism and is shown in equation 10 (k >> k + k,)The reaction to alkene with added base B is kinetically of second order and the mechanism is accordingly difficult to distinguish from E2 24 A large fraction of elimination product for this type of reaction seems to require a preassociation mechanism in which the dilute reactant B gets into reaction position before the bond to the leaving group is ruptured 25 A m uch more dominant MECHANISMS OF SOLVOLYTIC ALKENE-FORMING ELIMINATION REACTIONS-A THIBBLIN 43 1 alkene ia;:OAc a;;OH + OH If (9) -(6)+ -0Ac (7)' -0Ac -(8) allylic carbocation 1fast allyiic alcohols route to alkene is presumably with the solvent or the leaving group acting as the base The following scheme (equation 16) shows various ways of alkene formation directly from the substrate or through the ion pair k represents elimination promoted by the leaving group and/or by the solventThe reaction route via the preassociation complex (B- RX) followed by dehydronation (k,) is the ordin- ary base-promoted E2 mechanism The stepwise preassociation mechanism involves formation of the complex followed by ionization (k;)and base-promoted elimination from the ion pair (kh) This mechanism is preferred for the stepwise base-pro- moted elimination when collapse of (B-R+ X-) to (B- RX) is faster than B-can diffuse away z e ,k' > k- This may be the case when X is a potent nucleophile and when only minor reorgunization of carbocatzon structure and solvent is needed for the collapse to occur It is conceivable that some elimination reactions that have been classified as E2 with carbocationic transition states consist in fact of a mixture of concerted elimination and elimination through a reversibly formed ion-pair intermediateThis hypoth- esis is in accord with Bordwell's suggestion that in systems that undergo ionization,26 lyate ions promote elimination from ion pairs rather from the substrate itself The reason for this is that ion pairs are far superior to covalent substrate as hydron donors hl B RX B R+X alkene Accordingly elimination occurs via hydron abstraction from the B-R+X- complex (kk equation 16) rather than by con- certed elimination (k equation 16) from the preassociation complex When the base-ion-pair complex is too unstable to exist as an intermediate i e ,ti < 10 l3 s the reaction is forced to be concerted 27 Recently it was concluded that PhCH,CMe,Cl [(lo) equation 171reacts by an E2 mechanism with methoxide anion in metha- nol to give alkene (1 1)This elimination product is also formed by a stepwise route via a reversibly formed ion pair by dehydro- nation with solvent and added bases and probably also by the leaving group The other alkene (12) is only formed by the carbocationic route through reaction of the ion pair (equation 17) Solvolysis without any base present provides all three products Let us look briefly at the experimental results on which these mechanistic assignments are based 9-3 PhCY,COS &L3 (a) Y = 'H and L = 'H (b) Y = 2H and L = 'H (c) Y = 'H and L = ,H There are several indications for reversible ionization in methanol as well as in highly aqueous solventThus the solvolysis in 25 vol% acetonitrile in water is somewhat faster in the presence of azide anion or bromide anion than perchlorate anion which suggest nucleophilic attack on a reversibly formed ion-pair intermediate giving rise to a bimolecular contribution to the observed rate The isotope effect on the total reaction rate also suggests reversible ionization since k%,/kyis = 1 41 and 1 42 for reaction at 25°C in the aqueous medium and methanol respectively corresponding to an isotope effect of 1 19 per deuterium that is too large for a secondary ,&deuterium isotope effect Values of 1 10 f0 05//3-Dhave been reported for second- ary isotope effects The following expressions for the isotope effects can be derived from equations 17 and 18 reaction through the ion pair (cf Section 2) k!?/kP = (cq/eq)Wk?) kP/kP = (cq/eq)(k7/k?) kP/kF = (JqVq)(k?/k?) kk./k%s = (eq/=q)[(k? + k? + k!)/(k? -tk!? + k?)l where k&s = ks + kE + kEand (Qq/eq) ,/kD')= (ky/ky)(kH The isotope effects for the separate steps were calculated for solvolysis in methanol and without base addition with the help oftheseequations = 1 15ky/ky2= 1 16,ky/ky2 = 2 2 eq/@= 1 50 ky/k?" = 1 13 ky/k,D6= 2 3 The addition of methoxide anion to methanol increases the overall rate of disappearance of the substrateThis increase in total rate is caused by a large increase of elimination to give (1 1) but also by an increase in the rate of formation of (12) (equation 17) However the rate of formation of the ether decreasesThere is also a large increase in k%,/k:& and kp/kp2 but kp/kE2 and kF/kF6are not changed (Table 2) These results strongly indicate a parallel competing methoxide-promoted concerted E2 reac- tion (equation 18) Further indications are the observed decreases in k%,/k:& and kp/kE6 Table 2 Isotope effects for the reactions of (10) in methanol" at 25 00 f0 03°C Base i = D2 none 1 42 133 2 53 115 NaOAch 1 46 136 2 56 118 NaOMe' 2 39 1 46 3 89 115 i = D6 none 181 170 150 34 NaOAch 181 172 1 47 34 NaOMe' 135 1 47 116 34 3 74 vol% water 0 98 M 2 00 M The results do not indicate a parallel E2 reaction for formation of the other alkene (12) but are completely in accord with base- promoted and solvent-promoted elimination via the reversibly formed ion pair The experimentally measured isotope effect kF/kE2= 3 89 (Table 2) is the isotope effect for formation of alkene (1 1) both through the E2 route and the carbocationic pathThe assump- tion that the rate constant ratio k,/k (equation 18 pseudo-first order rate constants for reaction with solvent and base) is approximately the same with and without added base makes it possible to calculate the isotope effect for the E2 reaction with MeO- kH/kD2= 4 9The isotope effect for the carbocationic route to (1 1) is 2 5 and 1 5 for the di-deuteriated and the hexa- deuteriated substrates respectively The values are similar to those obtained without base 6.1 Mechanistic Borderline How are the borders between the different stepwise mechanisms of Table 1defined? The relative rates of the different microscopic processes determine which reaction path is the dominant one Jencks' definition which has been generally accepted12' 28 of a reaction intermediate as a molecular entity with a lifetime appreciably longer than a molecular vibration which is about lo-' s implies a rather sharp border between one-step and multi-step reactions A presumably rather common type of change of mechanism for a reaction involves two concurrent mechanisms having different transition-state structures ' la A change in experi- mental conditions or structure of the reactants lowers the energy of one of the transition states relative to the energy of the other which may induce a shift in the major reaction path Accord- ingly the reaction product may in principle be formed simulta- CHEMICAL SOCIETY REVIEWS 1993 neously by two parallel reactions 'reaction channels' 29 At the borderline both transition states are of equal energy Owing to a large difference in energy between the transition states one of the mechanisms frequently dominates and is the only mechan- ism observed The mechanistic change for reaction of (10) to (1 1) is con- cluded to be of this type Accordingly in pure methanol the E2 transition state (with methanol as hydron acceptor) is much higher in energy than the transition state of the stepwise reaction through the ion pairThus the elimination reaction exclusively employs the carbocationic path With methoxide the E2 transi- tion state is much lower in energy and can compete successfully with the dehydronation of the ion pair with methanol as well as with methoxide anionThe E2 reaction should be about four times faster than the methoxide-promoted reaction via the ion pair (Table 2 and equation 18) Apparently the methoxide- promoted reactions are very close to the borderline where both mechanisms have the same activation energy This borderline does not correspond to merging of transition-state structures 30 Is it also possible that the uncatalysed elimination to give (1 I) is a one-step solvent-promoted concerted E2 reaction? No since the isotope effect kp/kF6 is 1 50 in methanol but decreases to 1 16 in the presence of 2M sodium methoxideThe expected value for this secondary P-deuterium isotope effect on a one-step reaction should be very close to unity for reaction both with and without added base The values strongly indicate a stepwise mechanism for the reaction with pure solvent and elimination mainly through an E2 mechanism in the presence of a substan- tial amount of lyate anion Moreover a parallel stepwise preassociation mechanism (equation 16) or an enforced con- certed mechanism with a carbocation-like transition state for the methoxide-promoted reaction is not a reasonable alternative to the E2 mechanism for the same reason 6.2 Other Systems A study of a related system eliminated from t-BuC1 in basic methanol or methanol-Me,SO mixtures has revealed that EtS is a more efficient base than methoxide in pure methanol but not in solvent mixtures having a high proportion of Me,SO 31The results may be interpreted by a stepwise mechanism with an ion- pair intermediate formed in a pre-equilibrium step or with mixed stepwise and concerted elimination If the reactions are con- certed there is no need to invoke the E2C mechanism since the results of the base-promoted reactions are compatible with the theory of the variable E2 transition state theory Shiner and co-workers have concluded that the solvolysis of cyclopentyl p-bromobenzenesulfonate in aqueous hexafluoro- isopropanol involves reversible formation of the contact ion pair 32The stereochemistry of the elimination was studied by use of specifically deuteriated substrate The solvolysis of (13) in acetic acid which only provides elimination to alkene has been inferred to involve rate-limiting elimination through an ion pair 33 The main evidence was the value of the kinetic isotope effect kH/kD6= 2 87 which is too large for a purely secondary isotope effect The elimination may also occur through a reversibly formed ion-molecule pair Bordwel126 has suggested that the hydroxide- promoted elimination from neomenthyl trimethylammonium ion (14) in water occurs via a 'tightly solvated cation' Even strongly alkaline conditions could not completely suppress the competing first-order reactionThe possible role of the basic leaving group trimethylamine as the hydron abstractor has not been investigated However Bunton and co-workers have found MeI (EtO)*P-C-OMSII I0 Me H (14) H MECHANISMS OF SOLVOLYTIC ALKENE-FORMING ELIMINATION REACTIONS-A THIBBLIN that added amines are efficient in promoting elimination from ferrocenylalkyl carbocations in 50% acetonitrile in water 34 7 Concerted Pericyclic Elimination (El) This mechanism has a cyclic transition state e g (1 5) and (16) in which intramolecular hydron transfer to the leaving group is concerted with C-X bond cleavage It was suggested recently that destabilization of the carboca- tion intermediate of the stepwise solvolysis of cumyl derivatives RArCMe,X by electron-withdrawing ring substituent R leads to a change in mechanism to concerted pericyclic elimination Thus it was found that destabilization of the carbocation intermediate increases the amount of alkene product in 50 vol% trifluoroethanol in waterThis was interpreted as elimination through the ion pair which competes with nucleophilic substitu- tion by the solvent (at the ion-pair stage or through the free carbocation) Large destabilization OR+ b 0 34 yields a sub- stantial amount of alkene Addition of 0 50 M sodium azide provides 20-30% of the azide substitution product but has no acceleration effect on the overall rate Moreover it was found that the fraction of alkene is independent of azide anion concentration only the amounts of alcohol and ether decrease The substitution reaction with azide was concluded to proceed by a concerted preassociation mechanism (equation 19) RX R’X -ROS alkene -N RX-RN This mechanistic interpretation requires that a large part of the substrate is associated with azide anion as a preassociation complex and that this species reacts to give olefin with a rate similar to that of the un-preassociated substrate An alternative interpretation might be a mechanism in which the contact ion pair reacts on/y to elimination product and the substitution products originate from solvent-separated and free carbocation Consistently the measured Hammett and Win- stein-Grunwald parameters of p+ = -4 6 and mellm= 0 7 res-pectively for para-substituted chlorides suggest a polar transi- tion state Another speculative mechanistic interpretation is concerted solvent-promoted eliminationThe absence of detec- table catalysis with strong base may be due to a very small Brarnsted 6-parameter z e ,the catalysis from a solvent molecule is similar to that of an added strong base that is present at a much lower concentration AcknowledgementThe Swedish Natural Science Research Council supported this work 8 References 1 The hydron is a proton deuteron or triton Commission on Physical Organic Chemistry IUPAC Pure Appl Chem 1988,60 11 15 2 Commission on Physical Organic Chemistry IUPAC Pure Appl Chem 1989 61 23 R D Guthrie and W P Jencks Acc Chem Res 1989,22,343 3 W H Saunders Jr and A F Cockerill ‘Mechanisms of Elimination Reactions’ Wiley-Interscience New York N Y 1973 4 E Baciocchi Alkene-forming eliminations involving the carhon- halogen bond chapter 23 in ‘The Chemistry of Functional Groups’ Supplement D ed S Patai and Z Rappaport Wiley 1983 5 AThibblin and P Ahlberg Chem SOC Rev 1989 18 209 and references therein 6 A Thibblin J Chem SOC Perkins Trans 2 1986 321 7 A Thibblin J Am Chem SOC 1987 109,2071 8 A Thibblin J Am Chem SOC,1989 111 5412 9 A Thibblin J Phys Org Chem 1989 2 15 10 A Thibblin and H Sidhu J Am Chem SOC 1992,114,7403 11 K C Westaway ‘Isotopes in Organic Chemistry’ ed E Buncel and C C Lee Elsevier Amsterdam 1987 chapter 5 12The isotope effect kv/kQis not always a purely primary isotope effect For example if there are several /3-deuteriums ky/ky includes a small secondary isotope effect with an expected value of > 1 13 W P Jencks Chem SOC Rev 1982,10 345 14 H L Nyquist D A Davenport P Y Han J G Shih and T G Speechly J Org Chem 1992,57 1449 15 A Thibblin J Phys Org Chem 1992,5 367 16 P J Smith and J Pradhan Can J Chem 1986,64 1060 17 J P Richard and W P Jencks J Am Chem SOC,1984,106 1373 18 A Thibblin J Am Chem Soc 1988,110,4582 19 R A McClelland V M Kanagasabapathy N S Banait and S Steenken J Am Chem SOC 1991,113 1009 20 C Paradisi and J F Bunnett J Am Chem SOC 1985,107,8223 21 (a)A Thibblin and H Sidhu J Phvs Org Chem 1993 6 374 (h) references therein 22 H Hirota and T Mitsuhashi Chem Lett 1990,803 23 T Mitsuhashi and H Hirota J Chem Soc Chem Commun ,1990 3 24 24 R A Sneen Ace Chem Res 1973,6,46 25 A Thibblin and W P Jencks J Am Chem Soc ,1979.101,4963 26 F G Bordwell Ace Chem Res 1972,5,374 27 Commission on Physical Organic Chemistry IUPAC Pure Appl Chem 1983,55 1281 28 The definition has been critically discussed see ref 19 29 T W Bentley and I S Koo J Chem Soc Chem Commun 1988 41 T W Bentley and I S Koo J Chem SOC Perkin Trans 2,1989 1385 30 R A More O’Ferrall P J Warren and P M Ward Acta Univ Ups Symp Univ Ups 1978,12,209 W P Jencks Chem Soc Rev 1985 85 511 31 J F Bunnett and C A Migdal J Org Chem ,1989,54,3037,3041 32 R C Seib V J Shiner Jr ,V Sendijarevic and K Humski J Am Chem SOC 1978,100,8133 33 X Creary C C Geiger and K Hilton J Am Chem SOC ,1983,105 285 1 34 C A Bunton N Carrasco F Davoudzadeh and W E Watts J Chem SOC Perkin Trans 2 198 1,924 35 T L Aymes and J P Richard J Am Chem Soc ,1991,113,8960

ISSN:0306-0012    

DOI:10.1039/CS9932200427

出版商:RSC    

年代:1993

数据来源: RSC