摘要:

Functional cascade molecules Andreas Archut and Fritz Vögtle Cascade molecules/dendrimers are highly branched regularly built molecules that have now been known for two decades. While efforts in the early stage of their investigation were directed towards the development of higher generation structures and new dendritic architectures the design of functional dendrimers is more and more emphasized today. The latest literature in the field reflects the search for cascade molecules that accumulate certain functional groups resulting in novel properties—the first practical applications are in sight. Dendrimers that interact with and can be manipulated with light appear particularly attractive and hold promises for future developments. 1 Introduction Cascade molecules/dendrimers i.e.highly branched yet structurally perfect molecules have now been known for two decades. The term ‘cascade molecule’ indicates that such compounds usually originate from a stepwise synthetic procedure whereas the word ‘dendrimer’ is a combination of the Greek ‘dendron’ (tree branch) and ‘meros’ (part). Evolving around a core atom or molecule they possess repeating ‘generations’ of branches that branch again and again until an almost globular shape with a dense surface is reached. A manifold of different styles and designs of cascade molecules has so far been reported1 and the yearly number of publications on this topic is ever increasing since the first cascade molecules were reported twenty years ago.2 More than 2000 papers have now been published on the subject.In 1979 Denkewalter et al. released a patent on branched structures based on amino acids.3 Later the US industrial chemist D. A. Tomalia revived the concept of highly branched monodisperse molecules when he developed the family of so called polyamidoamine (PAMAM) dendrimers (Fig. 1).4 Only in the early 1990s when research on dendritic molecules began to develop explosively did a great number of chemists all over Andreas Archut was born in Bonn in 1970. He obtained his education in chemistry in Bonn and Los Angeles. Working with Nobel laureate George A. Olah he received the Master of Science degree from the University of Southern California in 1994. In 1997 he completed his PhD in the research group of Fritz V�ogtle at the University of Bonn with a thesis on functional photoswitchable dendrimers.Andreas Archut Fritz V�ogtle was born in Ehingen/ Danube in 1939. He studied chemistry in Freiburg and chemistry and medicine in Heidelberg. In 1965 he Kekul�e-Institut f�ur Organische Chemie und Biochemie der Universit�at Bonn Gerhard-Domagk-Str. 1 D-53121 Bonn Germany Fax Intl. +228-73-5662 E-mail voegtle@uni-bonn.de the world eventually begin to explore the scope of this new concept. 2 Evolution of dendritic structures When the dendrimer is build up synthetically from the core to the periphery as in the first examples,2–4 this is called a divergent synthetic strategy. When the dendritic branches are first built up and are then in a final step linked to the core molecule or atom like a wedge then this is known as a convergent synthesis.The larger the dendrimer becomes however the larger the number of bond formations per molecule and thus the greater the probability of defects in the structure. In addition as the dendrimers grow the terminal units move closer and steric hindrance becomes a problem in the ongoing synthesis. Calculations and the nature of exponentially increasing the number of terminal groups indicate that there is a limit to which the divergent strategy can be taken without significant deviations from the intended structure of the molecule. This is referred to as the ‘starburst effect’ a term coined by Tomalia for the upper generation limit above which only incomplete further conversion can be achieved due to steric crowding in the periphery.5 Apart from PAMAM dendrimers polyamine,6–8 carbosilane,9,10 and polyaryl ether11 dendrimers are amongst the most investigated and commonly used dendritic structures.Polyamine (Fig. 2) and PAMAM dendrimers are now even commercially available. Among others it is through the work of Newkome Fr�echet Meijer Moore and Majoral that more and more dendritic systems became known and the scope of their preparation was investigated. Newkome et al. have contributed a great number of new dendritic structures in the field e.g. globular dendritic poly-alcohols that exhibit micellular behaviour.12 Fr�echet et al. developed a family of polyaryl ether dendrimers.11 Meijer and coworkers have made available poly(propylene imine) den- Fritz V�ogtle received his PhD at the University of Heidelberg in the group of Heinz A.Staab. After his habilitation in 1969 in the same group he became a professor of chemistry at the University of Würzburg. Since 1975 he has been professor and director at the Kekul�e-Institut für Organische Chemie und Biochemie at the University of Bonn. Professor V�ogtle’s research focusses on supramolecular chemistry particularly on mechanically interlocked molecules like catenanes and rotaxanes on dendrimers and on host–guest chemistry. 233 Chemical Society Reviews 1998 volume 27 drimers to the scientific community that reach up to generation 5 with 64 amino groups in the periphery and they have used them for dendritic systems with promising abilities.8 Moore succeeded in preparing pure hydrocarbon dendrimers of a previously unseen size and molecular weight (up to 30 000 emu) and yet still remaining soluble.13 Phosphorus-containing dendrimers developed by Majoral et al.even surpass this weight record by far the heaviest species reaching a molecular weight of more than 3 million emu today.14 3 Functional dendrimers Fig. 1 PAMAM starburst dendrimers developed by Tomalia Fig. 2 The synthesis of polyamine dendrimers according to V�ogtle’s initial cascade molecules paper2 follows a stepwise (repetitive) procedure of consecutive Michael addition–reduction steps With a number of high generation ‘skeletons’ with multiple functional groups in the periphery the general interest of dendrimer research has experienced a shift from generations to functions with the intention to approach applications in both the life and material sciences.15,16 If one wants to introduce functionality to a dendrimer i.e.attach a shaped functional group like a chromophore ionophore receptor catalyst or molecular switch to a dendrimer there are two general ways to do so with readily available dendritic precursors (Fig. 3). Either such units can be connected to a dendritic skeleton with a number of reactive groups like amines alcohols or halides in the periphery or one can use a functional core unit which is then functionalised by reaction with a ‘dendreagent’ with dendritic ‘wedges’ (also referred to as ‘dendryl residues/substituents’).17 Chemical Society Reviews 1998 volume 27 234 Numerous periphery-functionalized dendrimers have been reported so far e.g. those with ferrocene units fatty acid moieties or sugar units. The main feature of such compounds is Fig. 3 Functionalization of dendrimers can be accomplished by three general approaches functionalization of a dendrimer at the periphery (b) attachment of ‘dendryl residues’ to a functional core (a); when dendrimers with functional groups in all regions (core branches periphery c) are desired a de novo synthesis is usually required Fig. 4 Balzani et al. have contributed various metal-polypyridine dendrimers with interesting electrochemical properties the accumulation of a defined number (multiplication) of functional groups around a central dendritic core.Among functionalized dendrimers those species containing chiral 235 Chemical Society Reviews 1998 volume 27 groups have become particularly important. Peerlings and Meijer have recently published a systematic approach to dendrimers possessing chire metal-containing dendrimers The metal–ligand–metal connectivity of ruthenium–bipyridine complexes and related coordination compounds has been used to assemble metal-containing dendrimers.19 The resulting cascade molecules are interesting for both their electrochemical and their luminescence properties. The first dendrimer containing a ruthenium–bipyridine complex as the core unit was described by Balzani et al.in 1994 (Fig. 4).20 Core-functionalized dendrimers in turn take advantage of the spatial demand of their dendritic wedges that can serve as shields for the central unit. One example is the dendritic ruthenium–bipyridine complex depicted in Fig. 5 it was found that the lifetime of the excited state of the ruthenium ion is significantly prolonged compared to the non-dendritic complex since the dendritic periphery prevents quenching processes with the solvent or with dissolved oxygen.21 Applications of this type Fig. 5 ‘Dendrylation’ of a ruthenium–bipyridine complex results in new properties. The shielding effect of the dendryl substituents renders the metal center less prone to quenching with solvent molecules and dissolved oxygen.Chemical Society Reviews 1998 volume 27 236 of ‘supramolecular dendrimer’ in medical diagnostics are envisioned.22 Dendrimers with a metal porphyrin unit as the core have become interesting model compounds for heme-containing proteins and as sterically hindered oxidation catalysts. Inoue et al. were the first to describe a dendrimer with a metal porphyrin as its core (Fig. 6).23 The convergent synthetic method of Fr�echet was used to prepare the dendrimer in which the photoactive metal porphyrin center is sterically shielded resulting in a size-dependent accessibility for quencher molecules. Diederich et al. have prepared dendrimers with a zinc– porphyrin core and Newkome-type polyether amide branches with the aim to fine tune and control the redox potential of the chromophore.24 Cyclovoltammograms of these dendrimers show that the first reduction potential of the zinc porphyrins decreases with increasing dendrimer generation and the authors rationalize this observation as being a result of the increasingly electron rich microenvironment created by the dendritic branches.The dendryl substituents ‘shield’ the porphyrin center and hence hinder the addition of electrons to it. Fig. 6 Porphyrin dendrimers are model compounds for heme-containing proteins 5 Photoswitchable azobenzene dendrimers Light is a useful way of manipulating molecular systems because its effect is fast mild and often reversible. Azobenzene derivatives have been used to construct photoswitchable devices for many years.25 Azobenzene-type compounds when they are not strongly sterically hindered do not show any appreciable fluorescence or phosphorescence but they can be easily and reversibly photoisomerized.The thermodynamically stable E-isomer can be photochemically converted to the Z-isomer which is converted back to the E-isomer by light excitation and thermally in the dark (Fig. 7).26,27 Azobenzene moieties have therefore been applied in the construction of photoresponsive molecular and supramolecular systems. hn1 N N N N Chemical Society Reviews 1998 volume 27 hn2 D Fig. 7 The E/Z isomerization of azobenzene compounds can be brought about by means of light A dendrimer bearing six peripheral azobenzene groups showing reversible switching behaviour was reported in 1993 (Fig.8).28 Recently Junge and McGrath have reported the synthesis of a two-directional dendrimer with an azobenzene group in the center and have investigated its E/Z isomerization induced by ultraviolet light (Fig. 9).29 Although the steric effect Fig. 8 The first photoswitchable dendrimer was reported by V�ogtle et al. 237 of the isomerization of the chromophore is probably a minor one the work has prompted speculation as to whether such molecules could be used to ‘grab’ molecules upon irradiation. It has been stated that in order to obtain such an effect there would have to be more switchable moieties present in the molecule.30 Such molecules are now available (vide infra). The Z/E photoisomerization of dendrimers with an azobenzene unit as the core has recently been claimed to occur by excitation with infrared radiation—with the dendrimer being able to harvest low-energy photons channel the absorbed energy to the core and bring about the chemical E/Z transformation.31 The latest report in the field of azobenzene dendrimers is a study by Vögtle Balzani et al.poly(propylene imine) dendrimers of the generations one through four equipped with up to 32 azobenzene units in the periphery were investigated as to their photochemical switching behaviour (Fig. 10).32 Interestingly measurements of the quantum yields of the photoisomerization and of the changes in the absorption spectra of these dendrimers show no significant difference to the results obtained with the corresponding monomers.In other words the azobenzene moieties behave independently even under increasingly close spatial proximity. In addition these periphery-functionalized dendrimers have been tested as materials for holographic data storage for the first time. Polymers with azobenzene moieties in the side chains are well known in holography. The structural rearrangement of the azo groups upon exposure to laser light is thought to be responsible for a change in the optical property of a holographic film. Although the results of these first experiments show that the dendrimers investigated are still less effective than the polymers commonly used a further fine-tuning of the dendrimers’ properties may lead to interesting optical applications for these dendrimers.32 Fig.9 Junge and McGrath have reported a dendrimer with an azobenzene core capable of undergoing photochemical isomerization Chemical Society Reviews 1998 volume 27 238 6 Molecular dendritic antennae Besides transition-metal sites other photo- and electro-chemically active organic functional groups have been introduced to dendritic structures. Xu and Moore have designed so-called ‘molecular antennae’ consisting of an electroluminescent luminophore equipped with dendryl residues capable of ‘funneling’ electrons to the focal point of the molecule.33 In another study the photoinduced electron transfer of a number of ‘dendryl-substituted’17 fluorophores was investigated.34 It has been found that the fluorescence maxima in organic solvents of these dendritic compounds is strongly generation dependent.Whereas the fluorophore (p-dimethoxybenzene) behaves like the corresponding non-dendritic molecule up to generation {4} of the dendryl residue an increasingly strong blue-shift was observed for generations {5} and {6}. It has been suggested that such compounds could be used for the design of electroluminescent diodes.35 7 Conclusion and outlook Today after several thousands of publications dealing with dendritic molecules the chemistry of dendrimers is reaching maturity at a great pace. Branched tree-like molecules are inspiring chemists from all fields and areas of chemistry since they open up routes to new materials unique substance properties and potential applications.Cascade molecules are the linking element between small (organic) molecules and high molecular weight macromolecules and in times of great interest in nanometer-size molecular arrays they are the ideal building blocks for nanoarchitecture. The first commercial applications have been realised e.g. in medical diagnostics and many more are presently envisaged. Photoactive dendrimers may play a prominent role in the future. Fig. 10 V�ogtle et al. have prepared dendrimers with up to 32 peripheral azobenzene groups that have been used as holography materials Fig. 11 Molecular antennae prepared by Moore et al. can funnel electrons to the focal point of the dendrimer 239 Chemical Society Reviews 1998 volume 27 8 References 1 G. R.me C.Moorefield and F. V�ogtle Dendritic Molecules Concepts Syntheses Perspectives 2nd edn. Wiley-VCH Weinheim (1998). 2 E. Buhleier W. Wehner and F. V�ogtle Synthesis 1978 155. 3 R. G. Denkewalter J. F. Kolc and W. J. Lukasavage U.S. Patent 4,410,688 (1983). 4 D. A. Tomalia A. M. Naylor and W. A. Goddard Angew. Chem. 1990 102 119; Angew. Chem. Int. Ed. Engl. 1990 29 138. 5 D. A. Tomalia Adv. Mater. 1994 6 529. 6 C. Wörner and R. M�ulhaupt Angew. Chem. 1993 105 1367; Angew. Chem. Int. Ed. Engl. 1993 32 1306. 7 R. Moors and F. V�ogtle Chem. Ber. 1993 126 2133. 8 E. M. M. de Brabander-van den Berg and E. W. Meijer Angew. Chem. 1993 105 1370; Angew. Chem. Int. Ed. Engl. 1993 32 1308. 9 D. Seyferth D. Y. Son A. L. Rheingold and R. L. Ostrander Organometallics 1994 13 2682.10 A. W. van der Made P. W. N. M. van Leeuwen J. C. de Wilde and R. A. C. Brandes Adv. Mater. 1993 5 466. 11 K. L. Wooley C. J. Hawker and J. M. J. Fr�echet J. Am. Chem. Soc. 1991 113 4252. 12 G. R. Newkome C. N. Moorefield and G. R. Baker Aldrichimica Acta 1992 25 (2) 31. 13 Z. Xu and J. S. Moore Angew. Chem. 1993 105 1394; Angew. Chem. Int. Ed. Engl. 1993 32 1354. 14 M. Slany M. Bardaji A.-M. Caminade B. Chaudret and J. P. Majoral Inorg. Chem. 1997 36 1939. 15 J. Issberner R. Moors and F. V�ogtle Angew. Chem. 1994 106 2507; Angew. Chem. Int. Ed. Engl. 1994 33 2413. 16 R. Moors and F. V�ogtle Adv. Dendritic Macromol. 1995 2 41. 17 Nomenclature proposed by M. Plevoets G. Nachtsheim and F. V�ogtle J. Prakt.Chem. Chem. Ztg. 1998 340 112. 18 H. W. I. Peerlings and E. W. Meijer Chem. Eur. J. 1997 3 1563. Chemical Society Reviews 1998 volume 27 240 19 J.-P. Sauvage J.-P. Collin J.-C. Chambron S. Guillerez C. Coudret V. Balzani F. Barigelletti L. De Cola and L. Flamigni Chem. Rev. 1994 94 993. 20 S. Serroni S. Campagna A. Juris M. Venturi V. Balzani and G. Denti Gazz. Chim. Ital. 1994 124 423. 21 J. Issberner F. V�ogtle L. De Cola and V. Balzani Chem. Eur. J. 1997 3 706. 22 R. Herrmann F. V�ogtle H.-P. Josel B. Frommberger G. Pappert and J. Issberner Patent DE 4439346A1 (1994). 23 R. H. Jin T. Aida and S. Inoue J. Chem. Soc. Chem Commun. 1993 1260. 24 P. J. Dandliker F. Diederich M. Gross C. B. Knobler A. Louati and E. M. Sanford Angew. Chem. 1994 106 1821; Angew. Chem. Int. Ed. Engl. 1994 33 1739. 25 V. Balzani and F. Scandola Supramolecular Photochemistry Horwood Chichester (1991). 26 H. Rau in Photochromism Molecules and Systems eds. H. D�urr and H. Bouas-Laurent Elsevier Amsterdam (1990). 27 G. S. Kumar and D. C. Neckers Chem. Rev. 1989 89 1915. 28 H.-B. Mekelburger K. Rissanen and F. V�ogtle Chem. Ber. 1993 126 1161. 29 D. M. Junge and D. V. McGrath J. Chem. Soc. Chem. Commun. 1997 857. 30 M. Freemantle Chem. Eng. News 1997 May 26 30. 31 D.-L. Jiang and T. Aida Nature 1997 388 454. 32 A. Archut F. V�ogtle L. De Cola G. A. Azzellini V. Balzani R. H. Berg and P. S. Ramanujam Chem. Eur. J. 1998 4 699. 33 Z. Xu and J. S. Moore Acta Polymer. 1994 45 83. 34 C. Devadoss P. Bharathi and J. S. Moore Angew. Chem. 1997 109 1706; Angew. Chem. Int. Ed. Engl. 1997 36 1709. 35 P.-W. Wang Y.-J. Liu C. Devadoss P. Bharathi and J. S. Moore Adv. Mater. 1996 8 237. Received 3rd September 1997

ISSN:0306-0012    

DOI:10.1039/a827233z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Surface-enhanced Raman scattering Alan Campion and Patanjali Kambhampati Department of Chemistry and Biochemistry The University of Texas at Austin Austin TX 78712 USA We present an introduction to surface-enhanced Raman scattering (SERS) which reviews the basic experimental facts and the essential features of the mechanisms which have been proposed to account for the observations. We then review very recent fundamental developments which include SERS from single particles and single molecules; SERS from fractal clusters and surfaces; and new insights into the chemical enhancement mechanism of SERS. 1 Introduction Surface-enhanced Raman scattering (SERS) was discovered twenty years ago. The field developed aggressively even explosively for the first decade or so and then settled down a bit as it entered its teenage years.Interest shifted from fundamentals to applications and a steady stream of papers was published in diverse fields which included electrochemistry analytical chemistry chemical physics solid state physics biophysics and even medicine for example. As SERS enters its third decade there has been a renewed interest in fundamentals especially in the short-range or chemical mechanism and single molecule detection has been achieved. Few scientists 25 years ago would have bet that Raman spectroscopy notorious for its difficulty and insensitivity at that time would have joined the ranks of single molecule spectroscopies before the end of the century. We present here a contemporary review of SERS with two objectives in mind.First we wish to provide an introduction to the field for scientists (and for students in particular) who may wish to conduct research in the area or use SERS techniques in their own research in other fields. Second we wish to highlight new areas of SERS research which we feel are particularly exciting and show promise for further development. Our goal is to provide the reader with sufficient background orientation and perspective to permit him or her to read the primary reviews and the original literature more easily. In keeping with the spirit of the Journal only the briefest historical development will be presented and the reader is referred to other review articles rather than the original literature for elaboration and further discussion.We acknowledge here the important contributions Alan Campion Alan Campion received his BA in chemistry from New College Sarasota Florida in 1972 and his PhD in chemical physics from UCLA in 1977 where he studied energy transfer in condensed matter with Mostafa El- Sayed. He was a National Science Foundation Postdoctoral Fellow in the laboratory of Charles Harris at UC Berkeley prior to joining the faculty of The University of Texas at Austin in 1980 where he is now Dow Chemical Company Professor of Chemistry. made by an enormous number of scientists from around the world over the past twenty years and regret that the format of this review does not permit us to recognize many individual contributions.The review is organized as follows. (We assume that the reader is familiar with the elementary aspects of Raman scattering; if not the monograph by Long1 is recommended.) A short history of the discovery of SERS is followed by a summary of the key experimental facts. Two classes of SERS mechanisms electromagnetic and chemical are introduced and it is shown how these mechanisms account for the experimental observations. Most of the features of the electromagnetic mechanism can be understood by examining the electrostatics of a polarizable metal sphere in a uniform external electric field; that model therefore will be discussed in some detail. Evidence that the electromagnetic mechanism is not the whole story will be presented to motivate a discussion of the chemical mechanism.Three areas of contemporary interest and activity are then reviewed in some detail. Recent advances in microscopy have made it possible to use single particles as SERS substrates and to obtain the Raman spectra of single molecules adsorbed on them; an introduction to this field and selected examples are presented. The importance of interparticle interactions is illustrated nowhere better than by fractal clusters. Recent advances in theory and computational techniques have provided a quantitative understanding of the localization of electromagnetic energy and its effects on both linear and nonlinear spectroscopies. Finally very recent studies on systems which show chemical enhancement without electromagnetic enhancement have provided a new and more detailed level of understanding about this mechanism.We conclude with mention of two areas of application which we find particularly promising and our assessment of the future of the field. 2 History and fundamentals SERS was discovered though not recognized as such by Fleischmann et al.2 in 1974 who observed intense Raman scattering from pyridine adsorbed onto a roughened silver Patanjali Kambhampati received his BA in chemistry from Carleton College in 1992 and is finishing his PhD in chemical physics at The University of Texas under the supervision of Professor Campion. Patanjali Kambhampati 241 Chemical Society Reviews 1998 volume 27 electrode surface from aqueous solution.The motivation for this work was to develop a chemically specific spectroscopic probe which could be used to study electrochemical processes in situ; Fleischmann’s approach was to roughen the electrode to increase its surface area and hence the number of adsorbed molecules available for study. Jeanmaire and Van Duyne3 and Albrecht and Creighton4 recognized independently that the large intensities observed could not be accounted for simply by the increase in the number of scatterers present and proposed that an enhancement of the scattered intensity occurred in the adsorbed state. Interestingly enough these papers presaged a debate about the SERS mechanism which ran furiously for nearly a decade and about which research is still being conducted.Jeanmaire and Van Duyne tentatively proposed an electric field enhancement mechanism whereas Albrecht and Creighton speculated that resonance Raman scattering from molecular electronic states broadened by their interaction with the metal surface might be responsible. As we shall see they were both right in concept though not in detail. SERS research accelerated dramatically in the early 1980s with contributions from chemists physicists and engineers from around the world. It is not hard to see the motivation for such interest. The effect was large completely unexpected hard to understand and of enormous practical utility if it could be understood and exploited. By 1985 certainly the experimental facts were generally agreed upon as were the essential features of the mechanisms.2.1 Experimental observations SERS has been observed for a very large number of molecules adsorbed on the surfaces of relatively few metals in a variety of morphologies and physical environments. Silver copper and gold have been far and away the dominant SERS substrates but work has been reported on the alkali metals and a few others. The largest enhancements occur for surfaces which are rough on the nanoscale (10–100 nm). These include electrode surfaces roughened by one or more oxidation–reduction cycles island films deposited on glass surfaces at elevated temperatures films deposited by evaporation or sputtering in vacuum onto cold (100 K) substrates colloids (especially aggregated colloids) single ellipsoidal nanoparticles and arrays of such particles prepared by lithographic techniques.It had been thought that surface roughness either atomic scale or nanoscale was required for SERS. Recent results discussed below show that roughness is not a requirement however. SERS differs in a number of ways from ordinary Raman spectroscopy of molecules and solids and even from unenhanced surface Raman spectroscopy.5 The intensities of the bands observed generally fall off with increasing vibrational frequency; C–H stretches for example tend to be relatively weak in SERS. Overtones and combination bands are not common. Selection rules are relaxed resulting in the appearance of normally forbidden Raman modes in the surface spectra. The spectra tend to be completely depolarized in contrast to solution spectra and those taken from molecules adsorbed on atomically smooth flat surfaces.Excitation profiles differ from the w4 dependence of nonresonant scattering; the broad resonances observed may be characteristic of the substrate the adsorbate or the combined system. Excitation profiles depend upon electrode potential in electrochemical experiments and may be different for different vibrational modes. The enhancement may be remarkably long ranged extending tens of nanometers from the surface depending upon the substrate morphology. Many mechanisms were proposed in the early days of SERS to account for the experimental facts mentioned above. A number of them turned out simply to be wrong and those that survived were quickly sorted into two classes which were called electromagnetic and chemical.As their names imply the former focus on the enhanced electromagnetic fields which can be supported on metal surfaces with appropriate morphologies and the latter on changes in the electronic structure of molecules Chemical Society Reviews 1998 volume 27 242 which occur upon adsorption and which can lead to resonance Raman scattering. We review briefly the salient features of these mechanisms below. 2.2 Electromagnetic enhancement The collective excitation of the electron gas of a conductor is called a plasmon; if the excitation is confined to the near surface region it is called a surface plasmon. Surface plasmons can either be propagating on the surface of a grating for example or localized on the surface of a spherical particle for example.Surface roughness or curvature is required for the excitation of surface plasmons by light. Perhaps the most familiar example of this phenomenon is Wood’s anomaly in which the reflectivity of a grating dips sharply at the frequency which excites the surface plasmon. The electromagnetic field of the light at the surface can be greatly enhanced under conditions of surface plasmon excitation; the amplification of both the incident laser field and the scattered Raman field through their interaction with the surface constitutes the electromagnetic SERS mechanism. There have been many versions of the electromagnetic theory developed over the years which treat physical situations of varying complexity at different levels of completeness.Model systems which have been treated include isolated spheres isolated ellipsoids interacting spheres interacting ellipsoids randomly rough surfaces treated as collections of hemispherical bumps or gratings and fractal surfaces for example. These systems have been analyzed with different degrees of sophistication. The simplest treatments invoke the electrostatic approximation using sharp boundaries and local bulk dielectric functions for the substrate. Full electrodynamic calculations have been carried out for the simpler systems and the effects of a nonlocal dielectric response have been discussed. These issues have been critically reviewed in the comprehensive article by Moskovits.6 The essential physics which underlies the electromagnetic mechanism is well-illustrated by the textbook example of a metal sphere in an external electric field.For a spherical particle whose radius is much smaller than the wavelength of light the electric field is uniform across the particle and the electrostatic (Rayleigh) approximation is a good one. The field induced at the surface of the sphere is related to the applied external (laser) field by eqn. (1) where e1(w) is the complex frequency- (1) Einduced = {[e1(w) 2 e2]/[e1(w) + 2e2]} Elaser dependent dielectric function of the metal and e2 is the relative permittivity of the ambient phase. This function is resonant at the frequency for which Re (e1) = 22e2. Excitation of the surface plasmon greatly increases the local field experienced by a molecule adsorbed on the surface of the particle.A very physical way to visualize this phenomenon is to consider the particle as having localized the plane wave of the light as a dipole field centered in the sphere which then decays with the dipole decay law away from the surface in all directions. The particle not only enhances the incident laser field but also the Raman scattered field. It acts as an antenna which amplifies the scattered light intensity. It is easy to see from the above discussion why small increases in the local field produce such large enhancements in the Raman scattering; the overall enhancement scales roughly as E4! This simple model rationalizes at least qualitatively most of the experimental observations.While we have focused on the sphere for simplicity the arguments which follow apply generally to the wide variety of surface morphologies which have been used in SERS research; the numerical factor of 2 in the resonance equation will simply be different for different structures. The dominance of the coinage metals and the alkali metals as SERS substrates arises simply because the resonance condition is satisfied at the visible frequencies commonly used for Raman spectroscopy. Other metals have their surface plasmon resonances in different regions of the electromagnetic spectrum and can in principle support SERS at those frequencies. In addition the imaginary part of the dielectric function (which measures losses in the solid) for the coinage and alkali metals is very small at the resonance frequency.Low loss materials sustain sharper and more intense resonances than those where scattering and other dissipative mechanisms are important. The materials requirement is fulfilled simply by selecting an excitation frequency for which Re e1 satisfies a resonance condition and Im e1 is as close to zero as possible. The model explains many of the other observations mentioned above. Since the Raman scattered light is of a different frequency than the incident laser the enhancement really goes as E2 laser E2 Raman. This means that both fields can be nearly resonant with the surface plasmon only for small frequency shifts. This fact explains the fall off in intensity observed for high frequency vibrational bands; the surface plasmon is excited by either the laser field or the Raman field but not both.The dipole decay law explains the range dependence of the phenomenon and the early controversy over that issue. The enhancement falls off as G = [r/(r + d)]12 for a single molecule located a distance d from the surface of a sphere of radius r or G = [r/(r + d)]10 for a monolayer of molecules. For large radii of curvature the effect looks long-ranged whereas for small radii it can appear to be a surface effect. That the enhancement can be long ranged provided strong evidence for the electromagnetic mechanism. The depolarization is easily explained by considering a SERS-active surface to be a heterogeneous collection of roughness features of different sizes and shapes onto which the molecules adsorb in a variety of orientations.The lack of motional averaging and the opportunity for multiple scattering both contribute to depolarization. 2.3 Chemical enhancement Several lines of evidence suggest that there is a second enhancement mechanism which operates independently of the electromagnetic mechanism; for systems in which both mechanisms are simultaneously operative the effects are multiplicative. As discussed below it has been very difficult to separate these effects on systems which support electromagnetic enhancement. The early evidence for the existence of chemical enhancement was therefore mostly inferential. Electromagnetic enhancement should be a nonselective amplifier for Raman scattering by all molecules adsorbed on a particular surface yet the molecules CO and N2 differ by a factor of 200 in their SERS intensities under the same experimental conditions.This result is very hard to explain invoking only electromagnetic enhancement. The polarizabilities of the molecules are nearly identical and even the most radical differences in orientation upon adsorption could not produce such a large difference. A second line of evidence in support of a chemical mechanism comes from potential-dependent electrochemical experiments. If the potential is scanned at a fixed laser frequency or the laser frequency is scanned at fixed potential broad resonances are observed.These observations can be explained by a resonance Raman mechanism in which either (a) the electronic states of the adsorbate are shifted and broadened by their interaction with the surface or (b) new electronic states which arise from chemisorption serve as resonant intermediate states in Raman scattering. The evidence to date supports the latter hypothesis. It is not uncommon that the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of the adsorbate are symmetrically disposed in energy with respect to the Fermi level of the metal (Fig. 1). In this case chargetransfer excitations (either from the metal to the molecule or vice versa) can occur at about half the energy of the intrinsic intramolecular excitations of the adsorbate.Molecules commonly studied by SERS typically have their lowest-lying electronic excitations in the near ultraviolet which would put the charge transfer excitations of this simple model in the visible region of the spectrum. Lombardi et al.7 have developed a theory which accounts for the potential-dependent excitation profiles mentioned above; it is very physical (based upon the standard second-order perturbation theory of Raman scattering) easy to understand and useful for extracting molecular information from SERS spectra. Their article is also a good source of references for both the experiments which motivated the theory as well as earlier theoretical treatments of the problem. Fig. 1 Typical energy level diagram for a molecule adsorbed on a metal surface.The occupied and unoccupied molecular orbitals are broadened into resonances by their interaction with the metal states; orbital occupancy is determined by the Fermi energy. Possible charge transfer excitations are shown. In this section we have provided a brief introduction to SERS. For readers interested in a comprehensive review of surfaceenhanced processes (including luminescence and photochemistry) emphasizing the physical aspects of the phenomena the excellent review by Moskovits6 is warmly recommended. There is a wonderful collection of reprints of original articles in the field carefully chosen and compiled by Kerker,8 which provides a sense of the excitement controversy and dynamics of the subject that is hard to convey in a review article.Two keynote issues of the Journal of Raman Spectroscopy,9,10 one published in 1991 and the second currently in preparation are also good collections of articles which include several reviews. Other useful references to which the reader is referred for more detail on various aspects of the subject include Birke et al.11 (general) Pettinger12 (electrochemistry) Cotton et al.13 (biomolecules) Nabiev et al.14 (medical applications) and Vo Dinh15 (sensors). We hope that these articles will provide the reader with easy access to the relevant literature of interest. 3 Single particles and single molecules There has been tremendous interest recently in the properties of nanoparticles in optical microscopies which provide greatly enhanced depth (confocal microscopy) and lateral (near field microscopy) resolution and in the ability to detect and characterize single molecules.Nanoparticles (with dimensions of the order of nanometers) can have properties which are intermediate in character between molecular and bulk; mesoscopic physics and the accompanying materials science are hot areas of current research. Similarly the new optical microscopies bridge the gap between the few micron length scale of conventional light microscopes and the atomic resolution of scanning tunneling and atomic force (AFM) microscopes. A 243 Chemical Society Reviews 1998 volume 27 very exciting new area of SERS research has drawn upon these fields recently in an attempt to better understand the electrodynamics of small particles and the SERS which results and to exploit this understanding to obtain the vibrational spectra of single molecules.Single molecule spectroscopy must certainly be the Holy Grail of analytical chemistry and the addition of a vibrational spectroscopy with its superior structural characterization to the electronic spectroscopies already developed is an important achievement. As mentioned in the introduction the possibility of using Raman spectroscopy of all things for single molecule spectroscopy seems almost ludicrous. Ordinary Raman cross sections are of the order 10230 cm2 molecule21 sr21 so even a watt of laser power focused into a square micron would produce only one scattered photon every few hours from a single molecule.Enormous enhancements must be created to make single molecule Raman spectroscopy a reality. Two groups have achieved this goal recently using very different approaches. Kneipp et al.16 reported the detection of single molecules of cresyl violet adsorbed on aggregated clusters of colloidal silver. They used near-infrared excitation which is not resonant with any intramolecular optical transitions of the dye but efficiently excites the plasmons of the fractal aggregate (vide infra ). Nie and Emory,17,18 on the other hand combined surface and resonance enhancement (SERRS) to produce the required sensitivity to detect a dye molecule adsorbed on the surface of a single silver particle. Nie and Emory used standard citrate reduction techniques to produce silver sols which comprised a heterogeneous collection of mostly unaggregated particles of various sizes and shapes.Surprisingly a small number of these particles showed extraordinarily high enhancements; they have been labelled ‘hot’ particles. These particles can be imaged by bandpass filtering the Raman scattering of adsorbed Rhodamine 6G. Combined optical and AFM measurements showed that most of these particles were isolated of dimension ca. 100 nm and have shapes which ranged from spherical to rod-like (Fig. 2). In agreement with the general features of the electromagnetic mechanism more intense scattering was observed when the laser polarization was aligned along the long axis of an ellipsoidal particle which is the most polarizable direction.Size selective fractionation through nanoporous membranes allows the enrichment of particles in a certain size range; in general agreement with the electromagnetic theory is the result that the resonance frequency is a function of particle size. The large enhancement factors observed cannot be accounted for solely by electromagnetic theory however and it is very hard to see why so few particles are ‘hot’ and why the distribution is so peaked about certain particle sizes for a given laser excitation wavelength. This behavior is in contrast to the situation for fractal surfaces in which the localization of electromagnetic energy at ‘hot spots’ is understood. Single molecule Raman spectra excited both in the near field and in the confocal mode are shown in Fig.3. Samples were prepared which had fewer than one adsorbed molecule per Fig. 2 Tapping mode atomic force microscopy images of selected silver nanoparticles. (A) Particles 1 and 2 showed significant Raman enhancement the rest did not. (B) A four-particle hot aggregate. (C) A rod-like hot particle. (D) A faceted hot particle. (Reproduced from Ref. 17 by permission). Chemical Society Reviews 1998 volume 27 244 particle on average by using appropriately dilute solutions based upon the known adsorption isotherm for the system. Confirmation that near field excitation does not dramatically alter the selection rules for SERS is shown by the similarity of the spectra. The increased signal-to-noise ratio observed for the confocal spectrum arises because the laser power is more efficiently transferred to the sample in this configuration.The total enhancement estimated for the system is an astonishing 1014–1015 which can be converted to a ‘Raman quantum yield’ of essentially unity! The authors suggest that this large enhancement factor can be reconciled with the more commonly measured values of 108–1010 for SERRS by considering the averaging that occurs in conventional measurements. They estimate that only one out of a hundred or a thousand particles is ‘hot’ and in addition that only about one in ten thousand surface sites leads to efficient enhancement. Support for the latter conclusion comes from the rapid saturation of the effect with surface coverage.The SERS signal is fully developed with only 3–4 molecules per particle. Recent microscopic examination of these particles shows well-developed facets which provide a variety of sites of atomic scale roughness and thus Fig. 3 Near-field (A) and confocal (B) SERS spectra of Rhodamine 6G adsorbed on single silver nanoparticles. (Reproduced from Ref. 18 by permission). various possibilities for adsorption. Although the extraordinarily large enhancement factors have been reported by two research groups for two very different physical systems a detailed understanding of the various contributions made by the several mechanisms operating is not yet in hand and promises to be a very interesting area for study. This field is moving very quickly and we can expect rapid advances on both the fundamental and applied fronts.Single silver nanoparticles have already been used to modify the tips of near field fiber probes in an effort to do near field SERS microscopy. Numerous schemes are being evaluated employing various combinations of near- and far-field excitation and detection to try and develop this technique. When perfected this single molecule vibrational spectroscopy will be a very powerful tool indeed. 4 Fractals It was recognized early in SERS research that systems of interacting particles produced the largest enhancements and a great deal of effort has been directed towards the study of the electrodynamics of coupled systems. It was also recognized early on that aggregated silver colloids produced fractal clusters.19 Concepts from these fields of inquiry have been brought together recently to analyze both the linear and nonlinear optical properties of fractals including of course SERS.Powerful new theoretical approaches have been developed and combined with near-field optical microscopy to investigate these phenomena in great detail. SERS has once again revitalized interest in a venerable subject the optical properties of small particles and stimulated research in areas outside its specific domain. A fractal object looks very much the same when examined at different magnifications. This property is called dilation symmetry. Fractals are characterized by the concepts of scale invariance and self-similarity.Unlike geometrical objects physical fractal structures (mountains clouds clusters and surfaces for example) follow scaling laws only over a limited range of length scales the limits of which are determined by real physical constraints. These could be the size of the sample at the macroscopic limit or the size of the molecule on the microscopic end for example. Fractals which obey different scaling laws in different directions or achieve self-similarity by anisotropic dilation are called self-affine. There are a number of good monographs in the general area of the physics of fractals; we have found the book by Gouyet20 to be very helpful. The averaged optical properties of composite systems have been well understood for a number of years using Maxwell– Garnett and more sophisticated effective medium theories.The measured long-wavelength optical absorption of these systems often exceeds the calculated values by many orders of magnitude however and a number of suggestions have been offered to explain the differences. Among these is that electromagnetic fields can become localized in the interparticle regions of clusters resulting in greatly enhanced absorption and other optical properties. This is the essential mechanism by which fractal aggregates produce a rich variety of interesting optical phenomena both linear and nonlinear. Fig. 4 shows a typical fractal cluster of colloidal silver particles prepared by reducing silver nitrate with sodium borohydride. The number of particles in a cluster of gyration radius RC is given by N = (RC/R0)D where R0 is a characteristic separation between colloidal particles and D is the fractal dimension.For this particular cluster the particles have typical diameters of 20 nm and D = 1.78. The contour diagrams in Fig. 4 show the computed intensity of the electric field which results when this cluster is illuminated with 500 nm light. Note the remarkable localization and greatly enhanced intensity which results from the fractal character of the aggregate. The local intensity enhancement can be as high as 103 and it is important to recognize that the fluctuations in the intensity can be as large as the intensity itself. This fact has important consequences for the spectroscopy and photochemistry of molecules adsorbed on the surfaces of these aggregates.Fig. 4 Local field intensities calculated for a simulated cluster of fractal dimension D = 1.78. (Reproduced from Ref. 22 by permission). There has been a tremendous body of theoretical work developed recently which is directed towards achieving a quantitative understanding of these phenomena. The excellent review by Shalaev21 summarizes this work in great detail. We highlight a few key developments to orient the reader. The field intensities shown in Fig. 4 are the results of a simulation based upon a scaling theory applied to ‘diluted’ aggregates which contain a few hundred particles. The theory clearly accounts at least qualitatively for the frequency and polarization dependent localization of electromagnetic energy in fractal clusters as demonstrated recently by its good agreement with the results of direct imaging of the fields using photon scanning tunneling microscopy.22 A more general theory has been presented recently21 which provides solutions to the coupled dipole equations using the exact (electrodynamic) operator for the dipole interaction rather than that of the static limit.Furthermore the theory is capable of examining much larger systems (up to 104 particles) which more realistically model the aggregates studied by experiments. A particularly nice result of these calculations which illustrates the essential physics is shown in Fig 5. Fig. 5 shows the wavelength-dependent local field enhancement factor calculated by simulation for three different kinds of 500-particle aggregates a fractal cluster (CCA) a system of close-packed spheres (CPSP) and a gas of spherical particles (RGP).The excitation of the surface plasmon of isolated spheres is clearly seen as the sharp feature below 400 nm for the gas and there is very little interaction among the particles as shown by the weak featureless response above that wavelength. For the crystalline system (CPSP) the single particle excitation is attenuated and intensity appears in the long-wavelength region of the spectrum as a result of the interparticle interactions. The largest enhancements in the local field intensity clearly occur for the fractal system (CCA) however. The symmetry breaking results in localization of the energy in regions of the aggregate which depend sensitively on wavelength and polarization and the additional enhancement can be huge factors of 103 or so.The fluctuations are large which means that the intensity of the electric field averaged over the aggregate ( < E2 > ) is greater 245 Chemical Society Reviews 1998 volume 27 Fig. 5 Wavelength dependence of the averaged local field intensity for a gas of spherical particles (RGP) a close-packed crystal of spherical particles (CPSP) and a fractal aggregate (CCA). (Reproduced from Ref. 21 by permission). than the square of the field averaged over the aggregate ( < E > 2). Recall that Fig. 5 illustrates this phenomenon for the linear optical response; the effect is even more pronounced for nonlinear responses which include four-wave mixing third harmonic generation and of course SERS.Excellent agreement has been reached between theory and experiment for SERS excitation profiles for molecules adsorbed on colloidal silver clusters. Although we have not discussed them in this review there is a similarly impressive body of work on self-affine surfaces important examples of which are the cold deposited silver films often used in SERS. The studies which have been carried out on both fractal clusters and self-affine surfaces have provided beautiful illustrations of the physics of these interesting systems as well as stimulated new theoretical approaches to study the electromagnetic properties of systems which are inhomogeneous on the nanometer length scale.The consequences of these discoveries for analytical applications of SERS are potentially significant. If signals arise largely from regions of the surface or cluster where the electromagnetic field is intense due to its spatial localization the possibility that the environment being sampled is not typical certainly exists and photochemical considerations may also become important. 5 Chemical enhancement on smooth surfaces 5.1 Introduction It has proved to be very difficult to study the chemical enhancement mechanism selectively for two reasons. First it is generally thought to contribute only a factor of 10–102 compared to factors of 104–107 for electromagnetic enhancement. Second almost any experimental parameter which can be varied to probe a system will have an influence via both mechanisms making the separation of effects difficult if not impossible.It is extremely important however to understand the chemical mechanism for both fundamental reasons and for its relevance to analytical applications. Since the effects are multiplicative unexpected chemical enhancement could lead to analytical conclusions which are not only quantitatively wrong but even qualitatively wrong. The earliest experimental evidence which suggested a connection between SERS and charge transfer excitations has been reviewed by Avouris and Demuth.23 They describe ultrahigh vacuum (UHV) experiments in which both SERS and a charge transfer excitation were observed for a Ag(111) surface onto which a grating was etched to allow for surface plasmon excitation.We reported24 shortly thereafter however that molecules physically adsorbed onto atomically smooth surfaces showed no enhancement apart from the small electromagnetic Chemical Society Reviews 1998 volume 27 246 effect expected from the surface simply acting as a mirror. This suggested to us (and others) that atomic scale roughness was a requirement for SERS though it was not clear whether the roughness merely provided chemisorption sites or also chemical enhancement. A systematic series of experiments was conducted to test this idea—atomic scale roughness was created on otherwise flat surfaces which provide only minor electromagnetic enhancement—but proved ultimately inconclusive.We rationalized our failure to observe the charge transfer mechanism for SERS directly by pointing out that the very large homogeneous linewidths of the transitions observed would result in only very small enhancements. We had in fact abandoned this line of inquiry when the adventitious discovery25 of SERS from pyromellitic dianhydride (PMDA) adsorbed on Cu(111) provided us with a new opportunity to investigate this problem of longstanding interest in an especially incisive way. 5.2 PMDA/Cu(111) experimental results Fig. 6 shows PMDA as both the free molecule and as the adsorbed carboxylate. As shown in Fig. 6 the molecule chemisorbs as a carboxylate by elimination of CO from one anhydride ring; the plane defined by the aromatic ring and the second anhydride group is oriented perpendicular to the surface.Fig. 7 shows the Raman spectrum of PMDA adsorbed on Cu(111) in UHV at 100 K excited at 647 and 725 nm. The 647.2 nm spectrum is clearly enhanced; the count rates of the most Fig. 6 Pyromellitic dianhydride as the free molecule and as the adsorbed carboxylate intense peaks are ca. 2000 cts s21W21compared with the typical unenhanced count rates we observe of ca. 20 cts s21W21. Fig. 7 Raman spectra of pyromellitic dianhydride adsorbed on Cu(111) excited at 725 nm (a) and 647 nm (b) To see if any new electronic excitations resulted from chemisorption we measured the electronic absorption spectrum of PMDA/Cu(111) using EELS (Fig. 8). Note the intense narrow transition at 1.9 eV which appears only in the monolayer spectrum.The intrinsic intramolecular excitations of PMDA a colorless compound occur in the ultraviolet and are observed in both the monolayer spectrum and with increased intensity in the spectra of condensed multilayers (not shown here). These results clearly demonstrate that a new low-energy electronically excited state is created when PMDA adsorbs on copper. To connect the results of our Raman scattering experiments with those of the EELS experiments it is necessary to show that the Raman excitation profile (dependence of intensities upon incident laser frequency) is related to the charge transfer absorption. Fig. 10 shows that the low-energy side of the excitation profile tracks the EELS spectrum well; experiments are underway to extend these measurements to higher energies.We discuss below the simulated spectra shown in Fig. 10. The PMDA/Cu(111) system provided a case in which the enhancement factor was large enough and the resonant behavior obvious enough that it was easy to conclude that chemical enhancement occurs. How do we establish the existence of chemical enhancement and determine its magnitude for weaker interactions? The answer is to apply a more sophisticated analysis. In the absence of coverage-dependent reorientations the relative intensities of all bands should show the same coverage and frequency dependence for unenhanced spectra; deviations from this behavior can be taken as evidence for chemical enhancement.Flat single crystal surfaces allow us to investigate the role of surface structure and to conduct polarization experiments to learn more about the nature of the excited state. We have observed significant differences in the spectra obtained from the Cu(111) and Cu(100) surfaces . It is reasonable to expect some differences even in the absence of SERS. The bonding configurations could be different on each surface and the Fig. 8 Electron energy loss spectrum of pyromellitic dianhydride (PMDA) adsorbed on Cu(111); the spectrum of clean Cu(111) is shown for comparison combined effects of the anisotropic molecular polarizability and surface screening would produce different spectra. For PMDA on copper however we could not rationalize the differences observed using such simple arguments and the results led us to consider the idea that the charge transfer states were different on the two surfaces.The Raman spectra obtained from both surfaces looked superficially similar but closer inspection revealed significant differences. The spectrum recorded from the (111) surface was ca. 2–3 times more intense than that from the (100) surface when excited with p-polarized (parallel to the plane of incidence) light and even the relative intensities were quite different for each crystal face. In addition both the total intensities and the relative intensities change markedly with laser polarization in a manner which is inconsistent with the predictions of classical electrodynamics. In particular the intensities observed with s-polarized (perpendicular to the plane of incidence) excitation were too high to be accounted for by consideration of the optical properties of metal surfaces alone.Clearly the surface electronic structure plays an important role in determining the character of the charge transfer state responsible for the chemical enhancement in SERS. 5.3 Theory and simulations To gain additional insight into the scattering mechanism we have carried out molecular spectroscopy simulations26 using the dynamical approach of Heller and coworkers.27,28 Equivalent results could be obtained using second order perturbation theory but we prefer the dynamical approach for its physical insight and computational convenience. The chemical enhancement mechanism of SERS can be considered a variation of ordinary resonance Raman scattering the differences being that the excited states are not purely intramolecular and that the screening orienting and damping properties of the metal surface must be considered.Within this framework the Raman intensity is determined by a few simple parameters which include the energy and linewidth of the excited state the displacement and curvature of the excited state potential energy surface (PES) along each normal coordinate and the magnitude of the transition moment component normal to the surface. Vertical excitation of the molecule from the ground vibrational level of the ground electronic state to the excited electronic state results in a vibrational wavepacket which is not 247 Chemical Society Reviews 1998 volume 27 an eigenfunction of the excited state Hamiltonian.It therefore evolves in time and later returns to the ground electronic state. As for ordinary resonance Raman scattering the excited state PES must be shifted significantly along some normal coordinate for that mode to be enhanced via Franck–Condon activity. Larger displacements produce greater overlaps between the vibrational wavefunctions of the ground and excited states. The intensity of Raman scattering follows a profile similar to that of the absorption spectrum of the charge transfer state. The largest enhancement occurs of course on resonance; the intensity and spectral width of the excitation profile is determined in part by the oscillator strength and lifetime of the excited state.The functional form of this dependence was discussed in the article by Lombardi et al.7 referred to earlier. Fig. 9 shows the results of a simulation of the Raman spectrum of adsorbed PMDA compared with the actual spectrum and Fig. 10 shows the results of the simulation of the EEL spectrum and the Raman excitation profile compared with the experimental results. A consistent set of parameters was used for all of the simulations; these included the vibrational frequencies and charge transfer absorption energy obtained from experiment and the width of the excited state resonance and a reference shift in the PES chosen to fit the spectra. The quality of the fits is excellent and allows us to reach certain conclusions about the excited states on the two surfaces.Fig. 9 Experimental (a) and simulated (b) Raman spectra of pyromellitic dianhydride adsorbed on Cu(100) for 647 nm excitation Franck–Condon activity is clearly shown by the vibronic shoulder observed in the EEL spectra and in the simulations. The differences in frequencies and relative intensities observed for the two surfaces (not shown here) tell us that the PES on Cu(111) is displaced more and of greater curvature than that on Cu(100). That the s-polarized spectra are more intense than predicted by the classical model suggests that the relevant transition dipole moments are oriented nearly parallel to the surface. Since the intensity of the p-polarized spectra relative to the s-polarized spectra is smaller for Cu(100) we conclude that the transition dipole moment is oriented more closely toward the surface for the PMDA/Cu(100) system than for PMDA/ Chemical Society Reviews 1998 volume 27 248 Fig.10 Experimental and simulated Raman excitation profiles (top) and EEL spectra (bottom) for pyromellitic dianhydride adsorbed on Cu(100) Cu(111). The latter conclusion also explains why the absolute intensity of the scattering is greater on the (111) surface. Metal surfaces screen the tangential components of electric fields and enhance normal components; all other things being equal systems in which the transition dipole moment is oriented along the surface normal will have the largest enhancements. The results of these experiments and simulations show clearly that the microscopic details of surface bonding and their effects on the electronic structure of the adsorbate are critical in determining chemical enhancement.5.4 The nature of the charge transfer state Although we have referred to the resonant intermediate state as a charge transfer state for consistency with the literature and because it is hard to see what else it could be we have only recently obtained definitive evidence of its character. We also discovered quite unexpectedly evidence for the spatial localization of these states. It is clear from Fig. 1 that changing the Fermi energy (or work function) without affecting the adsorbate orbital energies would shift the wavelength of the charge transfer transition.This idea was the basis for the interpretation of the potential-dependent electrochemical SERS experiments with the Fermi energy controlled by the applied potential. It has been known for some time that doping a surface with very small amounts of either electropositive or electronegative elements can also shift its Fermi energy significantly; this doping is easily accomplished in UHV experiments by dosing the appropriate elements onto the surface. Perturbations which shift the energy of the low energy state without affecting the adsorbate’s intramolecular excited states would provide strong evidence of a dynamic or photon driven charge transfer. In addition the direction of charge transfer is determined by the relation between the applied change in Fermi energy and the resulting shift in the transition energy.A red shift which results from raising the Fermi energy would mean that the charge transfer occurs from a filled metal orbital to an empty adsorbate orbital; a blue shift means the opposite. Our recent experiments show that doping the copper surface with a few percent of a monolayer of Cs results in the attenuation of the 1.9 eV peak in the EEL spectrum of adsorbed PMDA and the appearance of a new peak at 2.5 eV. The blue shift observed immediately establishes that the charge transfer is molecule ? metalin nature. The surprising result is that the energy of the new peak is independent of Cs coverage in contrast to what would be expected if doping produced a uniform increase in the Fermi energy of the entire surface.The intensity of the new peak does however increase with increasing Cs coverage. These observations have been taken as evidence that the alkali metal produces a local change in the metal work function which affects only nearby PMDA molecules. We estimate that a single Cs atom can influence the electronic structure of about four neighboring molecules. These experimental results are consistent with recent work on electron localization near ground state adsorbed atoms but provide the first evidence of which we are aware for the effect upon excited molecular states. These results are reported in detail elsewhere. 28 6 Applications There have been of the order of a thousand SERS papers published in the last five or six years and so it is clearly impossible to try and review them even in broad classes here.SERS has been used to investigate a wide variety of problems in science and the only unifying theme has been the use of a common technique. We would like to mention however a couple of lines of inquiry which we have found particularly interesting and which we feel may interest a general chemical audience. We also feel that they may make SERS more widely applicable as a tool in surface physics and chemistry. Natan’s group has published an interesting series of papers recently on the development of novel SERS substrates based upon the self assembly of gold colloids.30 Important characteristics of these systems include ease of preparation reproducibility stability compatibility with biomolecules and the important ability to tune the electromagnetic characteristics of the surface by controlling particle size and spacing.These workers have also prepared gold colloids overcoated with silver as useful SERS substrates. The versatility of this approach is a promising development for analytical applications. Weaver’s group has found a way to extend the SERS technique to transition metal surfaces by electrodepositing the metal of interest on a suitably prepared enhancing gold substrate. Although this idea had been tried before the inability to make pinhole-free overlayers prevented the technique from being generally useful. High quality films have been prepared by depositing slowly at constant cathode current rather than the more rapid constant potential deposition method formerly used.These films must be thick enough to behave chemically as the bulk metal of interest yet thin enough to support the electromagnetic enhancement of the underlying substrate. Promising applications of this in situ method to study gas phase heterogeneous catalytic reactions and electrocatalytic processes have recently been reported31 and the future of this area of application looks promising. 7 Conclusions and outlook SERS research continues at a brisk pace; hundreds of papers are published each year. Fundamental aspects of the phenomenon are still under investigation. The ability to isolate and characterize single particles and small clusters has provided us with a powerful new approach to the study of electromagnetic enhancement.Advances in computational techniques have allowed us to simulate accurately the electrodynamic response of very complicated structures. And the discovery of SERS from molecules adsorbed on flat single crystal surfaces has allowed us to study the chemical mechanism selectively. These advances together with revolutionary developments in instrumentation promise a bright future for SERS. Instrumental advances include compact efficient spectrometers based upon holographic filter and grating technology; charge-coupled device detectors with nearly unit quantum efficiency and Raman microscopes operating in both near-field and confocal modes. We believe that a comprehensive quantitative understanding of the mechanisms will be achieved during the next few years and that applications of this powerful molecular spectroscopy will be found in areas not yet imagined.8 Acknowledgments P. K. acknowledges the Welch Foundation and The Graduate School of The University of Texas for fellowship support. A. C. gratefully acknowledges the longstanding support of our research by the Welch Foundation and the National Science Foundation. 9 References 1 D. A. Long Raman Spectroscopy McGraw-Hill New York 1977. 2 M. Fleischmann P. J. Hendra and A. J. McQuillan Chem. Phys. Lett. 1974 26 163. 3 D. L. Jeanmaire and R. P. Van Duyne J. Electroanal. Chem. 1977 84 1. 4 M. G. Albrecht and J. A. Creighton J. Am.Chem. Soc. 1977 99 5215. 5 A. Campion Ann. Rev. Phys. Chem. ed. B. S. Rabinovitch J. M. Schurr and H. L. Strauss Annual Reviews Inc. Palo Alto CA 1985 p. 549. 6 M. Moskovits Rev. Mod. Phys. 1985 57 783. 7 J. R. Lombardi R. L. Birke T. Lu and J. Xu J. Chem. Phys. 1986 84 4174 and references therein. 8 M. Kerker SPIE Milestone Series ed. B. J. Thompson SPIE Optical Engineering Press Bellingham WA 1990 MS10. 9 J. Raman Spectrosc. 1991 22 No. (12). 10 J. Raman Spectrosc. Keynote Number Surface Enhanced Raman Spectroscopy New Trends and Applications 1998 29 in the press. 11 R. L. Birke T. Lu and J. R. Lombardi Tech. Charact. Electrodes Electrochem. Processes ed. R. Varma and J. R. Selman Wiley New York 1991 211. 12 B. Pettinger Adsorption of molecules at metal electrodes ed. J. Lipkowski and P. N. Ross VCH Publishers New York 1992 285. 13 T. M. Cotton J. H. Kim and R. E. Holt Adv. Biophys. Chem. 1992 2 115. 14 I. R. Nabiev K. V. Sokolov and M. Manfait Adv. Spectrosc. 1993 20 267. 15 T. Vo-Dinh Sens. Actuators B 1995 29 183. 16 K. Kneipp. Y. Wang H. Kneipp L. T. Perelman I. Itzkan R. R. Dasari and M. Feld Phys. Rev. Lett. 1997 78 1667. 17 S. Nie and S. R. Emory Science 1997 275 1102. 18 S. R. Emory and S. Nie Anal. Chem. 1997 69 2631. 19 D. A. Weitz and M. Olivera Phys. Rev. Lett. 1984 52 1433. 20 J. F. Gouyot Physics and Fractal Structures Masson Paris 1996. 21 V. M. Shalaev Phys. Rep. 1996 272 61. 22 D. P. Tsai J. Kovacs Z. Wang M. Moskovits V. M. Shalaev J. S. Suh and R. Botet Phys. Rev. Lett. 1994 72 4149. 23 Ph. Avouris and J. E. Demuth J. Chem. Phys. 1981 75 4783 and references therein. 24 A. Campion and D. R. Mullins Chem. Phys. Lett. 1983 94 576. 25 A. Campion J. E. Ivanecky III C. M. Child and M. Foster J. Am. Chem. Soc. 1995 117 11807. 26 P. Kambhampati C. M. Child M. C. Foster and A. Campion J. Chem. Phys. 1998 108 5013. 27 E. J. Heller Acc. Chem. Res. 1981 14 368. 249 Chemical Society Reviews 1998 volume 27 28 E. J. Heller R. L. Sundberg and D. Tannor J. Phys. Chem. 1982 86 1822. 29 P. Kambhampati M. C. Foster and A. Campion J. Chem. Phys. 1998 submitted. 30 K. C. Grabar R. G. Freeman M. B. Hommer and M. J. Natan Anal. Chem. 1995 67 735. Chemical Society Reviews 1998 volume 27 250 31 H. Y. H. Chan C. G. Takoudis and M. J. Weaver J. Catal. 1997 172 336. Paper 7/05148D Received 3rd February 1998 Accepted 24th February 1998

ISSN:0306-0012    

DOI:10.1039/a827241z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Cancer therapy a move to the molecular level F. Thomas Boyle and Gerard F. Costello* Zeneca Pharmaceuticals Mereside Alderley Park Macclesfield UK SK10 4TG E-mail Gerard.Costello@Alderley.Zeneca.Com The development of cancer therapeutics has traditionally been based on an empirical approach. As modern biological techniques have begun to open the way to understanding of key cellular processes at the individual protein level it has become possible to take a more mechanistic approach to the discovery of antitumour agents. This review describes some of the areas in which target mechanisms have been identified and the drugs which have been developed or are currently being investigated. Some of the interventions are not aimed primarily at the tumour but at the host systems.1 Introduction Cancer is not a single disease but a broad group characterised by uncontrolled proliferative growth and the spread of aberrant cells from their site of origin. At the simplest level cancer cells may be regarded as having lost touch with their environment so that they are no longer responsive to the controlling signals and interactions which occur continuously in normal healthy tissues. In general cancer incidence increases with age and most of the major cancers occur in localised tissues.1 This has led to them being described as solid tumours e.g. lung colon prostate to distinguish them from those such as the leukaemias (blood) and lymphomas. To some extent this also helps to explain why surgery and radiotherapy are predominant in cancer treatment.Chemotherapy though widely used is still a relatively minor weapon in the fight against solid tumour disease. Clinically cancers have been categorised by the organ or structure in which they originated e.g. breast colon. This has tended to reinforce the treatment by clinical speciality whilst Tom Boyle is a Zeneca Research Associate who was born in Cheshire in 1944. He obtained his BSc at the University of Salford before moving to ICI in 1964. After completing an industrially based MSc on ring opening reactions of pyridines he moved to the University of East Anglia to complete his PhD working with Richard A. Y. Jones and Alan Katritzky on the synthesis and tautomerism of N-oxides. He rejoined ICI at their Central Research Laboratories before moving to the Pharmaceutical Division in 1973 working on aspects of infection and animal health.The latter area took him to Australia in 1979 to work on the small molecule modulation of ruminant nutrition. On returning to the UK he focused his research interests in cancer and has been involved in the small molecule approaches to aromatase inhibition cytotoxic therapies based on the inhibition of thymidylate synthase antibody directed prodrug therapies (ADEPT) and inhibition of cell signalling and apoptopsis. Tom Boyle Activator sometimes obscuring any commonality of disease mechanism across tissue tumour types. This review will attempt to illustrate the opportunity provided by the rapidly increasing understanding of such underlying molecular mechanisms in cancer which has been made possible by the revolution in molecular and cell biology in recent years.To harness this insight most effectively clinical testing and practice may have to accommodate corresponding changes in tumour classification and treatment. 2 Cytotoxics Current chemotherapy consists of cytotoxic (cell-killing) agents and anti-hormonal drugs which reduce the proliferative drive to the tumour.2 Many compounds with good tumour cell-killing activity have been discovered but few have found clinical utility. This reflects a lack of discrimination between effects on tumour and normal tissue the cell-cycle dependency of many cytotoxic drugs and their frequent susceptibility to induced drug resistance.Significant side-effects such as nausea vomiting diarrhoea hair loss and serious infection are often encountered during chemotherapy since healthy tissue in the gastrointestinal tract hair follicle and bone marrow proliferates at least as rapidly as most tumours. Furthermore quiescent (nonproliferating) tumour cells can remain largely unaffected by treatment and may subsequently begin to divide and grow. Clinical strategies have been developed to address such issues. These involve cycles of therapy to allow recovery of normal tissue in between and attack on those tumour cells which have grown out since the previous treatment. Unfortunately this process may generate increased selection pressure for changes Gerard Costello was born in Scunthorpe England in 1952 and obtained his BA (Chemistry) from the University of Oxford in 1975.He moved to Leeds to work for his PhD with Edwin Saxton on total synthesis of Aspidosperma alkaloids. After gaining his PhD in 1978 he spent two years working in the laboratory of Professor Albert Eschenmoser at ETH Zurich on a NATO postdoctoral fellowship before joining ICI Pharmaceuticals at Alderley Park as a medicinal chemist in 1981. He has also worked in chemical process development and as an International Product Development Manager with responsibility for cancer projects from 1987–1991. Since 1991 he has been Section/Project Manager in the Cancer Research Department of Zeneca (formerly ICI) Pharmaceuticals.Gerard Costello 251 Chemical Society Reviews 1998 volume 27 which induce drug resistance in what is by definition a genetically-labile cell population. In the clinic early responses to therapy are often followed by disease progression or recurrence with reduced tumour susceptibility to the original or other drug treatment. To a greater or lesser extent this general profile applies to cytotoxic agents from a wide range of mechanistic classes e.g. alkylating agents DNA intercalators antifolates tubulin binders topoisomerase inhibitors. This includes many of the best known and most widely-used anticancer drugs such as cisplatin 1 doxorubicin 2 methotrexate 3 paclitaxel 4 and etoposide 5. For the most part cytotoxic drugs have been developed empirically and their major locus of action identified in parallel or afterwards.Information gained from clinical study has been used to derive new approaches based on mechanistic considerations in addition to the more traditional compound screening methods. As an example inhibition of transcription by targeting compounds to specific sequences in the minor groove of DNA is an area of much research activity which has been greatly assisted by advances in molecular structural techniques.3 In another area the powerful pairing of biosynthetic pathway elucidation (Fig. 1) and molecular modelling (Fig. 2) has led to a new class of antifolate agents which selectively inhibit thymidylate synthase (TS). This enzyme is critical for the de novo biosynthesis of thymidine the only nucleotide required exclusively for the synthesis of DNA rather than RNA.Inhibition of this enzyme is one of the actions of the widely-used agent 5-fluorouracil 6. The recent introduction of the new TS-specific drug raltitrexed 7 shown bound in the enzyme complex in Fig. 2 will allow a Chemical Society Reviews 1998 volume 27 252 methylene tetrahydrofolate tetrahydrofolate thymidylate synthase DHFR dihydrofolate reductase dihydrofolate Fig. 1 Thymidine biosynthesis pathway Fig. 2 Raltitrexed bound in the ternary complex of thymidylate synthase realistic assessment of the clinical advantages arising from such mechanistic selectivity.4 Despite an improved basis for designing ‘conventional’ DNA-targeted cytotoxic agents there must be a high risk that the inherent problems described above preclude any major therapeutic breakthrough with this category of drugs.At the same time it should be recognised that incremental improvements in the treatment of solid tumour disease in particular remain highly desirable. 3 Antibody-targeting In order to overcome the problem of normal tissue toxicity efforts have been made to achieve direct targeting of tumour cells usually by means of antibodies to tumour-specific antigens. This is the embodiment of the ‘magic bullet’ long sought after in cancer therapy. Despite much early promise there have not been any real successes with antibody treatment of major solid tumours.5 There are a number of problems which have been found in using antibody therapy: (i) It is remarkably difficult to achieve tumour-specific antibodies and also to have high affinity.(ii) Unlike the laboratory situation clinical tumours do not have consistent expression of target antigen throughout their mass. (iii) Antibodies are large molecules which do not penetrate solid tumours well. As a result only a very small amount of dosed antibody (much less than 1%) reaches the tumour and much of that localises to the vasculature. This means that systems using antibodies linked to radio-isotopes have the problem that the high doses given to achieve the desired effect at the tumour result in extended circulation times and most of the radiation being received by other tissues. Antibody–toxin and antibody–drug conjugates may suffer from their only being active against tumour cells bearing the relevant antigen and any instability in the chemical link to the antibody could result in undesirable systemic toxicity.To date there has been little success with antibodies alone. This could be a reflection of the majority of clinical studies being conducted with murine antibodies which suffer from both immunogenicity and poor recruitment of effector mechanisms. Whilst human antibodies might be more effective it is worth noting that T-cells are recruited in large numbers to some solid tumours apparently without the necessary activation to achieve cell killing. In fact new immunological stimulation approaches are being investigated based on T-cell signalling targets and there is also renewed interest in cancer therapeutic vaccines.3.1 ADEPT One of the approaches most likely to overcome the shortcomings described above is antibody-directed enzyme prodrug therapy (ADEPT).6,7 This is a two-phase therapy (Fig. 3) which uses an antibody–enzyme conjugate to achieve localisation to the tumour and follows up with a prodrug of low toxicity which is converted only by that enzyme to a very potent short-acting drug. In this way an amplification mechanism for targeted drug delivery is provided which can also achieve a ‘bystander effect’ and kill cells not bearing the antigen. Fig. 3 ADEPT system A system of this type is currently under clinical investigation. It uses a very selective antibody against carcinoembryonic antigen (CEA) and binds to most colon gastric and non-small cell lung cancers as well as many breast and ovarian tumours.The antibody is murine and it is linked to a bacterial enzyme carboxypeptidase G2 (CPG2). This means that the enzyme does not occur naturally in man and by designing the prodrug appropriately liberation of free drug away from the tumour is avoided. Careful consideration of the properties required for the prodrug–drug combination determined the design process. The drug had to (i) Be small enough to be readily-diffusible through the tumour mass. (ii) Show high cytotoxic potency against both dividing and quiescent cells. (iii) Have a rather short chemical half-life to avoid toxicity caused by escape of the drug from the tumour into the circulation.In contrast the prodrug had to (i) Show markedly less cytotoxicity than the drug since it would be administered systemically. (ii) Exhibit good enzyme kinetics as a substrate in order to allow rapid production of sufficient drug to have the desired effect. In this system the preferred drug 8 is an aromatic ‘mustard’ compound. The advantages of this class of alkylating agent are that they are not cell-cycle-specific so are also effective against quiescent cells and they tend to be less susceptible to induced resistance than most anti-cancer agents. From a medicinal chemistry viewpoint there is also a reasonable basis of understanding of how to modulate the cytotoxic potency of such agents.A thorough investigation of the interactions between the various component parts of the prodrug 9 was necessary before an optimal system was achieved. Obviously this antibody–enzyme conjugate will almost certainly be immunogenic in man. This may well restrict the number of doses which can be given to cancer patients. Research is continuing to examine the feasibility of producing humanised systems with much reduced potential for immunogenicity which could allow more treatment cycles to be undertaken. 4 Anti-hormonal agents Perhaps the first example of an area of medical treatment for cancer to benefit from a detailed understanding of biochemical mechanism is that of anti-hormonal therapy. It is interesting to note that the approach derives from surgical discoveries over the last century.Removal of the ovaries or testes was shown to give clinical responses in a significant proportion of breast and prostate cancer patients respectively. Responsive tumours were found to be dependent on the relevant sex hormone oestrogen or testosterone for their growth.8 Extensive research over many years into the biosynthesis and action of the sex hormones then provided the basis for targeting interventions and drug design (Fig. 4). 4.1 Anti-oestrogens Even with extensive background knowledge the first and still the most successful anti-hormonal drug is tamoxifen 10 which was discovered in a programme originally aimed at anti-fertility treatment. To some extent this reflected the concern that medical treatment would not be able to match the efficacy of surgery in cancer.It also resulted from the apparent paradox that this oestrogen receptor antagonist does not cause direct killing of breast cancer cells yet can achieve good clinical anti-tumour effects.9,10 Additional complexity was provided by the fact that tamoxifen acted as a full antagonist in some tissues and as a partial agonist in others even within the same species. A pure oestrogen receptor antagonist ICI182780 11 was subsequently discovered and is now in late-stage clinical trial. 4.2 Aromatase inhibition As indicated in Fig. 4 there are other points in the oestrogen biosynthetic pathway which offer potential for breast cancer 253 Chemical Society Reviews 1998 volume 27 Hypothalamus Testis Adrenal Androgens Testosterone Peripheral tissue (fat) Fig.4 Hypothalamic/pituitary axis and sex hormone action Scheme 1 Action of steroid aromatase Prostate Seminal vesicles treatment interventions. Inhibition of the enzyme steroid aromatase which effects the conversion of androgens to oestrogens (Scheme 1) has been a target of much recent research.11 Once again the interest was stimulated by clinical observation this time with the drug aminoglutethimide 12. Originally developed as an anti-convulsant this compound had been found to be a non-selective inhibitor of steroid biosynthesis. In particular it was shown to be an effective inhibitor of cytochrome P-450 enzymes many of which (including aromatase) are involved in the steroid pathways.Whilst interesting from a mechanistic viewpoint the more important Chemical Society Reviews 1998 volume 27 254 Pituitary Ovary Serum oestrogens Uterus Breast finding clinically was that the drug lowered circulating oestrogen levels by around 50% in post-menopausal women and the responses seen in breast cancer patients broadly reflected the reduction in hormone levels. Efforts were then focused not only on evaluating structures with known cytochrome P-450 inhibitory potential (especially from the anti-fungal area) but also on using molecular modelling based on homology with X-ray structures of bacterial enzymes to build in the selectivity required to avoid the severe side-effects seen with aminoglutethimide.12,13 This has proved a very successful approach and a number of compounds generally classified as ‘azoles’ have been evaluated clinically.Two recently introduced drugs from this class anastrozole 13 and letrazole 14 appear to deliver the improvements in sideeffects and clinical efficacy being sought. The increased efficacy results from a more profound lowering of circulating oestrogen levels. 4.3 LHRH agonists Luteinizing hormone releasing hormone (LHRH) agonists form a third class of hormonal therapy for breast cancer albeit only in pre-menopausal women. This limitation arises from their inhibitory effect on luteinizing hormone (LH) release from the pituitary (see Fig. 4) and consequent suppression of ovarian stimulation for oestrogen production.Whilst there are LHRH antagonists in clinical study the ability of the agonists to mimic successfully the natural inhibitory feedback effect of oestrogen at the pituitary by inducing LHRH receptor downregulation provides a rare superiority over direct blockade. The agonists are all close analogues of LHRH 15 but their potency has had to be even greater than that of the natural decapeptide hormone itself to achieve the downregulating effect.15 The structures of the two major drugs goserelin 16 and leuprorelin 17 are aligned by amino acid sequence for comparison. To realise the full clinical benefit of their biological action required the development of sustained-release formulations of one month duration. Novel technology was needed to produce the bio-degradable carriers which allowed the usual problems of rapid metabolic cleavage of peptides to be overcome.The basis of these formulations was a lactide–glycolide co-polymer. Very small quantities of these peptide agents (3.6 and 10.8 mg in the case of goserelin for formulations of one month and three month duration respectively) as injectable depot preparations are sufficient to suppress serum oestrogen levels into the menopausal range and maintain that suppression throughout these extended periods. Since the LHRH agonists act at the pituitary (Fig. 4) they also have an inhibitory effect in men on hormonal drive to the testes resulting in a reduction of serum testosterone to levels comparable with those achieved by surgical castration.By matching the effects of this standard treatment LHRH agonists have also become the first acceptable medical therapy for prostate cancer the second largest cause of cancer deaths in men. 4.4 Anti-androgens Whilst either surgical castration or treatment with LHRH agonists ablates testicular androgen production a secondary source is provided by the adrenals (Fig. 4). To achieve what is referred to as ‘total androgen blockade’ androgen receptor antagonists 14 have been introduced in combination with either surgery or LHRH agonist treatment. The prototypic antiandrogen is the non-steroidal compound flutamide 18. Its biological activity however derives mainly from a metabolite hydroxyflutamide 19 which is a much more potent androgen receptor antagonist.Subsequently bicalutamide 20 was developed from consideration of the hydroxyflutamide structure and this drug is active in its own right and appears very well tolerated. As yet no anti-androgen has gained use as a single therapy in prostate cancer though trials are taking place both in advanced and in early-stage disease. This forms a marked contrast with tamoxifen in breast cancer where oestrogen receptor blockade was the first successful approach. Another potential target for intervention in the androgen biosynthetic pathway is the enzyme 5a-reductase which converts testosterone to the much more potent androgen dihydrotestosterone (Scheme 2). Although 5a-reductase inhibitors have been developed for the treatment of benign prostatic hypertrophy (BPH),16 initial studies against the more demanding target of prostate cancer have not been as encouraging.Scheme 2 Action of 5a-reductase 4.5 Anti-hormonal profile Common features across the whole of anti-hormonal therapy are the need for continuous dosing of the agent over extended time periods (sometimes indefinitely) and the consequent requirement for a much better tolerability profile than for example cytotoxic therapy. These factors relate to the lack of direct cellkilling ability associated with this general therapeutic class. At the same time the improved tolerability is often associated with better quality of life for the patient and increased compliance with therapy. Although differing mechanistically from cytotoxics anti-hormonal agents are also subject to the development of resistance by the tumour.The timescale tends to be much longer than for cytotoxic therapy but resistance may emerge and in certain cases it has been known for the drug to become stimulatory for tumour growth. 5 Signal transduction inhibition Despite the great importance and value of anti-hormonal therapy perhaps its most important deficiency is that it is essentially limited to use in breast and prostate cancer. However the clinical profile shown by these agents opened up the prospect of being able to develop similar drugs for other tumours which were not hormonally-responsive.17 Without good clinical precedent to direct research towards a particular approach it was unclear for a very long time how to make progress towards this new goal.In this case the insight came from the laboratory and the rapid advances being made in the understanding of the genetic processes underlying cancer pathogenesis. Over the last two decades there has been a fundamental change in the approach to cancer research with recognition of the primacy of molecular mechanism. The first wave of drug candidates derived from this change is just coming through into the clinic. Amongst the first systems to have been investigated successfully are the growth factor signalling pathways (Fig. 5). Even as their detail and complexity have been emerging the potential for therapeutic intervention at several different levels between the cell membrane and the nucleus has become evident and a variety of molecular biological and ‘small molecule’ tools applied to validating genes as targets.5.1 Growth factor receptor antagonism Antagonism of growth factors at their cell surface receptors was an early approach which helped to open up the area for further investigation.18 Growth factor receptor blockade can be achieved in a number of cases but its impact has generally not been sufficient to affect the proliferation of representative 255 Chemical Society Reviews 1998 volume 27 Fig. 5 Growth factor signalling pathways tumour cells. This has been interpreted as a consequence of the inherent ‘redundancy’ within cell signalling pathways. This limits the extent to which proliferation differentiation or survival signals depend on the binding of an individual ligand to a single receptor type.Best effects have been claimed for lessselective agents which tends to support the view that selective growth factor antagonists will not be useful ‘stand-alone’ anticancer drugs. 5.2 Receptor tyrosine kinase inhibition As shown in Fig. 5 the next stage in the signalling cascade involves the phosphorylation of protein tyrosine moieties by the cytoplasmic domains of growth factor transmembrane receptors. The enzymic addition of phosphate groups is carried out by kinases. Protein tyrosine kinase activity (Scheme 3) is associated with many of the growth factor receptors and also with oncogenic non-receptor proteins such as src. A number of tyrosine kinases are overexpressed and/or show increased activity in human tumours.Whilst this is not proof of a causal relationship it provides some evidence for the contribution of tyrosine kinase activity to cancer growth. Scheme 3 Protein tyrosine kinase activity For these reasons inhibition of tyrosine kinases was seen as an attractive and chemically feasible opportunity. The main issue around these targets was one of selectivity. In this case selectivity refers to tumour versus normal cells and also across the different classes of tyrosine kinase. A combination of highthroughput screening and structure-based design approaches was used to derive the first compounds which served as Chemical Society Reviews 1998 volume 27 256 pharmacological tools to demonstrate that the principle of intervening at this level in signal transduction pathways was valid.Flavonoid natural products like quercetin 21 and genistein 22 were not generally selective and had other actions such as topoisomerase inhibition but they showed that compounds competitive with ATP could at least discriminate between tyrosine and serine/threonine kinases. Another natural product erbstatin 23 engendered great interest because of its simple structure. Many analogues of the tyrphostin type,19 represented by 24 and 25 were made and provided yet more support for the approach because they also showed a degree of discrimination between individual tyrosine kinases and some had anti-tumour activity in animal models.20 The epidermal growth factor (EGF) receptor was one of the first targets for drug discovery because of its known overexpression in human tumours such as non-small cell lung cancer (NSCLC) and head and neck cancer.Using the known sequence of the EGF receptor tyrosine kinase catalytic domain alongside X-ray crystal structures of cyclic adenosine monophosphate (cAMP)-dependent protein kinase a homology model could be built. Structural types identified by screening against the human enzyme derived from the membranes of A431 tumour cells were assessed against this model. A number of different compound classes showing reasonably selective inhibition of EGF receptor tyrosine kinase activity has been identified. Perhaps the most significant of these to date is the anilinoquinazoline class.These agents have also been shown to be competitive with ATP and have exhibited a high degree of selectivity. The prototypic structure represented by 26 was identified by targeted screening and was enhanced to produce compounds like 27 which is an extremely potent inhibitor and very selective for the EGF receptor. However the physical properties of these agents were not optimal for in vivo activity in animal tumour models. The breakthrough in this area came with the discovery of in vivo activity with the 6-aminoquinazoline compound 28. Although very much less potent against the enzyme than many of the compounds referred to above it had a better pharmacokinetic profile in animals. This was particularly important because these inhibitors had the desired profile incorporating a separation of anti-proliferative and cell-killing actions.In the absence of direct cell-killing it appears essential to have significant levels of compound in the blood at all times in order to see reproducible effects in the tumour models. Scheme 4 Action of ras farnesyl transferase Optimisation based around pharmacokinetic properties resulted in the compound ZD1839 29 which is currently undergoing clinical trials.21 Another compound 30 with very close structural similarity to 27 has also shown good in vivo activity. These agents and a small number of compounds directed against other tyrosine kinase targets should provide the first clinical test of whether intervention at this point in the signalling cascade can produce useful anti-tumour efficacy.Whilst aberrant signalling is a feature of many tumour cells the same pathways are essential for maintenance of some normal tissues. It is hoped that a balance similar to that seen in the animal models will be found in the clinic between activity against the poorly structured tumour and toxicity to the patient. 5.3 Ras inhibition Another key signalling system is the ras pathway (Fig. 5). Most interest in modifying the action of the ras oncogene has been focused on inhibition of the farnesyl transferase enzyme. In order to exert its functional effects ras has to be docked into the cell membrane. The cytosolic protein has to be modified at the C-terminus by addition of a lipophilic ‘tail’ which then anchors it into the cell membrane.A key step in this process is the addition of a farnesyl group to a cysteine thiol which is catalysed by farnesyl transferase (Scheme 4).22 Each of the different ras forms mutated or normal has a C–A–A–X terminal peptide sequence which confers farnesyl transferase substrate activity. A range of structures has been identified usually based around the tetrapeptide motif 31,23 amongst which there are very potent enzyme inhibitors 32–35. As yet none of these compounds is known to have progressed to the clinic. Structural studies on various SH2-containing proteins have provided some insights into how they perform their adapter 5.4 Other signalling pathway interventions SH2 (src-homology) domain containing adapter proteins determine which of the associated proteins will interact with tyrosine phosphorylation sites on receptors to propagate the signal (Fig.5). 257 Chemical Society Reviews 1998 volume 27 role.24 The N- and C-terminal SH2 domains of the p85 sub-unit of phosphatidyl inositol (PI) 3 kinase have provided excellent examples of protein structure determination by NMR methods. However the fact that the key interactions in this case are between two proteins at least one of which is phosphorylated has made it more difficult to find good chemical starting points from normal compound library screening. Even peptide-based medicinal chemistry approaches do not appear to have made much progress against this category of target and the first real breakthrough is still awaited.There has been a great deal of work done around the protein kinase C (PKC) family which has produced some interesting compounds though the problem of selectivity between isoforms of this important serine/threonine kinase has not been overcome. Another family of signalling proteins exciting great interest is the mitogen-activated protein (MAP) kinases which are involved in the linkage of tyrosine phosphorylation signals to serine/threonine phosphorylation signals. They are important enzymes in growth modulation signalling and have become leading drug discovery targets. When assessing intervention options closer to the cell nucleus like these there may be increased concern that the balance being sought between efficacy and toxicity will be shifted towards a profile more closely resembling that of the cytotoxic agents.6 Cell cycle modulation The concern about increased toxicity is even greater when considering intervention at the level of the cell cycle. The nuclear process of replication and division involves a number of phases (Fig. 6). Fig. 6 The cell cycle Rapid growth in understanding of the basic machinery has been accompanied by insight into how mitogenic and inhibitory pathways couple to the cell cycle and how it is deregulated in cancer.25 Entry into the cell cycle is controlled by a balance of activating factors such as mitogen and oncogene signals and inhibitory elements such as transforming growth factor (TGF)b and tumour suppression genes.6.1 Cyclin-dependent kinase inhibition One of the prime targets is cyclin-dependent kinase (CDK)4 which acts at the G1/S interface. The response to DNA damage in normal cells is to arrest the cell at this starting point of its cycle. CDK4 activity is known to be increased in a wide variety of solid tumours and this may be associated with overexpression of cyclin D1 TGF-b signalling defects and reduction or loss of the tumour suppressors p16 p21 and p53 (Fig. 7). Inhibition of CDK4 should block entry into the cell cycle but the degree of selectivity for tumour over normal cells has not yet been established. A broad spectrum CDK inhibitor flavopiridol 36 is currently under clinical investigation26 and appears to be showing a cytotoxic profile in line with expectation.X-Ray structural studies27 with CDK2 have been successful which provides extra information for design not only of Chemical Society Reviews 1998 volume 27 258 Inhibitor Inhibitor p53 Fig. 7 Pathway to cell cycle entry Activator 7 Apoptosis Entry into Cell Cycle inhibitors of that enzyme but also of other CDK enzymes by homology modelling. All of the approaches described in Sections 4–6 have as their main aim an anti-proliferative effect without direct cell-killing. There are also options for cell death approaches28 which can exploit the differences between tumour and normal cells and so avoid the drawbacks of the ‘conventional cytotoxics’. The most important area for consideration involves the process of apoptosis or programmed cell death.Apoptosis is an important and widespread biological process which seems to be complementary to mitosis (cell-division) in the regulation of cell populations. It plays a critical role in development and is often a result of tissue damage. Disruption or inhibition of apoptosis is frequently seen as a major component of malignancy. Induction of apoptosis either by blockade of survival signals or activation of programmed cell death signals is attractive because it is a process which does not occur randomly in all cells of a tissue. Some precedent for efficacy is provided by the fact that many of the known cytotoxic agents induce apoptosis albeit in a non-specific manner. Progress in understanding the range of mechanisms involved in apoptosis has occurred at a remarkable rate which reflects it being currently one of the most intensively studied biological areas.Once a cell is stimulated to enter the cell cycle signals at certain stages direct it either to complete the cycle or to undergo apoptosis (Fig. 8). Overexpression of the proto-oncogene bcl-2 seems to limit the effects of chemotherapy and radiation treatment.29 It appears to function as a negative regulator of apoptosis and much effort is being made to discover agents capable of blocking its action. Similarly loss of p53 activity by mutation of that tumour suppressor gene causes resistance to apoptosis induction. Both of these drug targets involve protein–protein interactions and initial approaches have consequently been built mainly around peptides to provide validation tools.An alternative approach is to reduce the enhanced survival signalling which may occur in tumour cells. There are targets of this type such as insulin-like growth factor (IGF)-1 receptor and focal adhesion kinase (FAK) activity which seem more Fig. 8 Cell death and survival signalling pathways akin to the anti-proliferative signalling targets. Intervention using more precedented medicinal chemistry approaches and low molecular-weight compounds seems more feasible in such systems. 8 Angiogenesis inhibition In addition to the majority of the treatments aimed solely at the tumour cell there are also therapeutic approaches targeted at the host or host–tumour interaction.The two major areas here are angiogenesis (generation of new blood levels from existing vasculature)30 and invasion31 (Fig. 9). Tumours require a blood supply in order to grow. Since angiogenesis in adults is normally a transient local process controlled by a balance of angiogenic and angiostatic factors tumours have to subvert this to achieve sustained blood vessel formation. In many cases this leads to tumour blood vessels being structurally abnormal and potentially usefully different from the rest of the vascular system. An anti-angiogenic agent should in principle be useful in all solid tumour disease to produce at least growth stasis. There are a number of biopharmaceutical approaches directed against angiogenesis including antibodies and angiostatic factors which have yet to be fully tested in the clinic.Some low molecular-weight compounds are also known to be anti-angiogenic including the natural product-derived TNP470 3732,33 and thalidomide 38. It is likely that the anti-angiogenic actions of the latter compound contribute to its well-known production of birth defects. Angiogenesis is a multiple-step process involving activation of endothelial cells synthesis and release of degradative enzymes migration and proliferation of the cells and then organisation and differentiation to form the new structure. Consequently it is not always possible to Fig. 9 Angiogenesis and invasion determine which steps are affected by a given compound. As the process has been studied in greater depth specific opportunities for therapeutic intervention have emerged.8.1 VEGF receptor tyrosine kinase inhibition Amongst the mechanisms involved the vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) receptor tyrosine kinases represent targets which look amenable to drug discovery efforts. Of the two the VEGF receptor would seem to offer the best chance for selectivity since it is found predominantly on the vascular endothelium and most research has focused on this target.34 VEGF is also known as vascular permeability factor and it is suggested that it facilitates tumour progression by stimulating angiogenesis and increasing vascular permeability. Tumour cells are known to make and secrete VEGF which then acts locally on endothelial cells.This should not only provide some tumour selectivity but also may help to avoid the sort of drug resistance mechanisms utilised by tumours because endothelial cells are not similarly genetically unstable. There are two forms of the VEGF receptor KDR and Flt against which compounds have been tested for their ability to inhibit tyrosine kinase activity. No strong evidence exists for one form being significantly more important than the other for the angiogenic process and most inhibitors seem to have at least some activity against both enzymes. Clinical assessment of drug candidates with different profiles against the receptor forms should be available soon as it is known that compounds have entered pre-clinical development.9 Anti-invasion approaches Invasion is another complex process which occurs in both normal and disease situations. With regard to tumours the definition of malignancy has always been made pathologically in terms of whether growth has been accompanied by invasion into other tissue. The three major components of tumour 259 Chemical Society Reviews 1998 volume 27 invasion are tissue degradation adhesion and migration. Contributory mechanism targets have been identified in all three areas and there is overlap in some cases with angiogenesis approaches. 9.1 MMP inhibition Most research has been carried out in the area of tissue degradation which is also of great relevance in diseases such as arthritis.Inhibition of matrix metalloproteinases (MMPs) has been by far the most investigated of the approaches.35 These enzymes constitute a family of zinc and calcium-dependent endoproteinases which is capable physiologically of breaking down all of the protein components in the extracellular matrix. Normal tissue remodelling involving MMPs occurs in processes such as wound healing and connective tissue maintenance but the same processes are important in tumour invasion and the enzymes have been found in a range of solid tumour types. The three major MMP classes are collagenases stromelysins and gelatinases. On the basis of tumour association and their ability to degrade basement membrane gelatinases are claimed to be the preferred target in cancer.It seems likely that as with some of the areas described above clinical testing of compounds with differing profiles against the MMP classes will determine which are most relevant. Initial medicinal chemistry interest has been centred around broad-spectrum peptidic structures bearing a zinc-binding ligand often a hydroxamic acid. Whilst this has resulted in extremely potent compounds being discovered and batimistat 39 and marimistat 40 being taken into the clinic there are a number of problems with inhibitors of this type. In particular they often have very poor aqueous solubility which can lead to difficult formulation and contribute to poor bioavailability and pharmacokinetics. Adverse effects such as joint pain have also been seen in the clinic and these findings cannot be attributed with any certainty to the general approach given the lack of enzyme selectivity with these compounds.The more recent availability of X-ray and NMR structures of collagenase and stromelysin combined with high-throughput screening should help to provide additional start points to those derived so far from rational design based on substrate cleavage sites. Modification of physical properties particularly by introduction of non-peptide structures and replacement of the widely used zinc ligands to improve pharmacokinetics and metabolism remains the goal of second generation MMP inhibitors in cancer. It is still much too early to say whether an anti-invasive agent will be sufficiently effective as a single agent in cancer.Although it is generally true that cancer treatments will involve multiple drug therapy there is a greater expectation that antiinvasives and anti-angiogenic agents will be used in combination with drugs targeting the tumour cell exclusively. 10 Conclusion The breadth of cancer therapeutic research means that only a limited illustrative coverage of a few key areas has been attempted. For example antisense oligonucleotide35,37 and gene therapy38 approaches to cancer have not been considered. In both cases the technologies and therapeutics differ sufficiently Chemical Society Reviews 1998 volume 27 260 from previous pharmaceutical systems to require fuller explanation. Furthermore whilst clinical studies are being conducted with examples of both types their prospects in solid tumour disease are probably confined to proof of principle in this phase.Similarly understanding of differentiation mechanisms is still at an early stage despite the interesting activities of retinoid compounds,39 and approaches to restoration of normal morphology and function to tumour cells are not sufficiently advanced for inclusion. Nevertheless the general message for cancer therapy is that a new era has begun. It started with the development of the techniques of molecular biology which allowed identification and investigation of individual components in key cell systems. This not only provided the basis for elucidating molecular mechanisms but also allowed the production of individual proteins or their relevant domains (often as the human version) for structural study and use in compound screening.Now that targets of particular relevance to tumours can be more readily identified drug discovery research has started to operate at the molecular level. The final phase requires that the clinical approach builds on this process and ensures that the developing speciality of molecular medicine becomes established in cancer. 11 Acknowledgements We thank Stephen Green Phillip Hedge and Donald Ogilvie for providing diagrams to us and Andrea Torkington for preparation of this manuscript. 12 References 1 Cancer Facts and Figures American Cancer Society Atlanta 1997 pp. 1–17. 2 B. Chabner Cancer Principles and Practice in Oncology ed.V. T. DeVita S. Hellman S. A. Rosenberg and J. B. Lippincott Philadelphia 1993 pp. 325–417. 3 J. O. Trent G. R. Clarke A. Kamur W. Wilson D. W. Boykin J. E. Hall R. R. Tidwell B. L. Blackburn and S. Neidle J. Med. Chem. 1996 39 4554. 4 A. L. Jackman and A. H. Calvert Ann. Oncol. 1995 6 871. 5 D. C. Blakey Acta Oncologica 1992 31 91. 6 K. D. Bagshawe Mol. Med. Today 1995 1 424. 7 R. Melton R. Knox and T. A. Connors Drugs of the Future 1996 21 167. 8 A. Howell R. B. Clarke and E. Anderson Endocrine-Related Cancer 1997 4 371. 9 T. J. Powles Semin. Oncol. 1997 24 Suppl. 1 S1-48-S1-54. 10 W. J. Gradishar and V. C. Jordan J. Clin. Oncol. 1997 15 840. 11 P. E. Goss and K. M. E. H. Gwyn J. Clin. Oncol. 1994 12 2460. 12 Y.-H. Kao L. L. Cam C.Laughton D. Zhou and S. Chen Cancer Res. 1996 56 3451. 13 S. Graham-Lorence B. Amarneh R. H. White J. A. Peterson and E. R. Simpson Protein Sci. 1995 4 1065. 14 W. R. Fair M. S. Cookson N. Stroumbakis D. Cohen A. G. Aprikian Y. Wan P. Russo S. M. Soloway and J. Sogani Urology 1997 43 (3A) Suppl 46. 15 B. J. A. Furr G. R. P. Blackledge and I. D. Cockshott Hormone Dependent Cancer ed. J. R. Pasqualini and B. S. Katzenellenbogen Dekker New York 1996 pp. 397–424. 16 J. L. Tenover G. A. Pagano A. S. Morton C. L. Liss and C. A. Bymes Clin. Ther. 1997 19 243. 17 G. Powis Curr. Opin. Oncol. 1995 554. 18 J. B. Trepel J. D. Moyser and F. Cuttita Biochem. Biophys. Res. Commun. 1988 156 1383. 19 A. Gazit N. Osherov C. Giton and A. Lavitzki J. Med.Chem. 1996 39 4905. 20 N. M. Law and N. B. Lydon Emerging Drugs 1996 241. 21 J. Woodburn A. J. Barker K. H. Gibson S. E. Ashton A. E. Wakeling B. J. Curry L. Scarlett and L. R. Henthorn J. Immunotherapy 1997 20 408. 22 M. Yongqi C. Omer and R. A. Gibbs J. Am. Chem. Soc. 1996 118 1817. 23 S. Graham and T. M. Williams Exp. Opin. Ther. Patents 1996 6 1295. 24 T. Pawson and J. D. Scott Science 1997 278 2075. 25 L. Meijer S. Guidet and H. Y. L. Tung Progress in Cell Cycle Research Plenum Press New York 1996 vol. 1 373. 26 H. H. Sedlacek J. Czech R. Nqaik G. Kaur W. Worland M. Losiewicz B. Parker B. Carlson A. Smith A. Senderowicz and E. Sausville Int. J. Oncol. 1996 1143. 27 W. F. Jr. De Azevedo H.-J. Mueller-Dieckmann U. S. Gahmen P. J. Worland E. Sausville and S.-H. Kim Proc. Natl. Acad. Sci. USA 1996 93 2735. 28 E. White Genes and Development 1995 1. 29 J. C. Reed Nature 1997 387 773. 30 W. Risau Nature 1997 386 671. 31 J. Folkman New Eng. J. Med. 1995 333 1757. 32 D. Ingber K. Takeshi S. Shoji T. Kanamura H. Brem and J. Folkman Nature 1990 340 555. 33 M. Skobe P. Rockwell N. Goldstein S. Vosseler and N. Fusenig Nature Med. 1997 3 1222. 34 L. Liotta Cancer Res. 1986 46 1. 35 H. S. Rammusen and G. M. Hockel Pharm News 1997 4 11. 36 C. A. Stein Cancer Principles and Practice in Oncology ed. V. T. DeVita S. Hellman S. A. Rosenberg and J. B. Lippincott Philadelphia 1993 pp. 2646–2648. 37 C. Helene New Approaches in Cancer Pharmacology Drug Design and Development ed. P. Workman Springer-Verlag Heidelberg 1992 vol. 1 pp. 13–24. 38 M. Singh V. Parikh and A. Sharma Drugs of the Future 1997 22 995. 39 L. M. De Luca FASEB J. 1993 2924. Received 23rd January 1998 Accepted 23rd February 1998 261 Chemical Society Reviews 1998 volume 27

ISSN:0306-0012    

DOI:10.1039/a827251z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Bent metallocenes revisited Jennifer C. Green Inorganic Chemistry Laboratory South Parks Road Oxford UK OX1 3QR The orbital structure of bent metallocenes and how their geometry depends on the number of d electrons are described. Bonding by a metallocene unit is exemplified by reference to the known hydrides. The reactivity of metallocene derivatives is illustrated with particular emphasis on the differences between ansa-bridged and unbridged compounds; the reactions include ring opening polymerisation of ferrocenophanes elimination from and addition to Group 6 metallocene derivatives and Ziegler–Natta polymerisation of olefins by Group 4 metallocenes. 1 Introduction Shortly after the discovery of ferrocene and the establishment of its sandwich structure with parallel rings I came the discovery that bis-cyclopentadienyl metal complexes could be synthesised in which the two C5-rings are inclined at an angle to one another and there are also additional ligands attached to the metal II.1 M L n M I II y x z The high symmetry of ferrocene and its first row transition metal analogues enabled qualitative molecular orbital methods to be applied successfully to describe its bonding with a high degree of confidence.Ligand field theory was used to treat the Jennifer Green gained her BA MA and DPhil at the University of Oxford the latter under supervision of Jack Linnett and Peter Atkins. She then held a Turner and Newall Research Fellowship after which she was appointed a Fellow and Tutor in Chemistry at St Hugh’s College.Her current position is that of University Reader in the Inorganic Chemistry Laboratory. Her research has focused on the electronic structure of transition metal compounds which she has investigated both theoretically and experimentally using photoelectron spectroscopy. Cl g M Cl M = Ti Zr Hf 131.0 129.3 127.1 g d electrons and rationalise spectroscopic and magnetic properties. The lower symmetry bent metallocenes were less tractable, 2 but physical studies3–5 and semi-empirical theoretical treatments6,7 produced a coherent model describing the orbital structure of these species which has been invaluable in interpreting the wide variety of basic chemical processes in which these bent metallocenes participate.The study of bent metallocenes has generated a wealth of chemistry which has demonstrated and led to understanding of many fundamental transformations; for example for the tungstenocene system these include photochemical reductive elimination of dihydrogen insertion of Cp2W [Cp = (h-C5H5)] into saturated sp3 C–H bonds first evidence for reversible a-H elimination and development of rules for predicting the regioselectivity of nucleophilic addition to organometallic cations.8 In recent years metallocenes have become of considerable commercial importance providing in combination with methyl alumoxane a new generation of Ziegler–Natta type polymerisation catalysts for production of polyethylene and polypropylene.9–11 The aim of this article is to review the bonding model for bent metallocenes and to show how their orbital structure relates to some of the unusual reactivity shown by these compounds. The selection of material is eclectic rather than comprehensive. In Fig. 1 the various structural parameters used in this review are defined for a bent metallocene unit (a) and for an ansabridged metallocene unit (b). d d f a e X b g M a b g M d d (b) (a) Fig. 1 Geometric parameters in a bent metallocene a = angle between the ring planes; b = angle between the normals from the metal to the ring planes; g = ring centroid-metal-ring centroid angle; d = angle between the ring plane and the metal-ring centroid vector; e = angle between the vectors from a bridging atom X to the ipso-carbons; f = angle between the ipsocarbon vector and the ring plane; d = the ring slippage the displacement of the ring centroid from the normal to the ring plane.There is generally found to be little departure from planarity of the cyclopentadienyl rings. The angle between the rings is normally defined in one of three ways either as a the angle between the ring planes or b the angle between the metal-ring normals (a + b = 180°) or g the angle between the vectors from the metal to the ring centroids. When the rings are linked by a bridge X e gives the angle at the bridging atom and f the angle between the ring plane and the vector from the ipso-carbon to the bridging atom. Chemical Society Reviews 1998 volume 27 263 2 MO model for a bent metallocene unit The orbital structure of ferrocene is well established; a schematic energy level diagram is given in Fig.2. The top four occupied energy levels are also illustrated in Fig. 3. In D5h symmetry† the top three occupied orbitals are to a first approximation non-bonding and principally metal 3d in character. The dz2 orbital has minimal overlap with the ring pp orbitals as its nodal cone intersects the region of maximum electron density of these orbitals. It is the major contributor to the a1A orbital where it is mixed with the 4s orbital and avoids a ligand x22y2 and dxy orbitals have the potential to contribution. The d back-bond into the empty ring orbitals of e2A symmetry but as these are rather high in energy the back bonding contribution is small.The principal bonding interaction is between the metal dxz and metal dyz orbitals and the ring e1B p orbitals. The e1A combinations of ring p orbitals are less effective at bonding as they can only combine with the metal p orbitals and these are high in energy. Similarly the mixing between the ring a1A p orbitals and the metal s orbital is much more effective than that Fig. 2 Schematic energy level diagram for a D5h metallocene; for ferrocene the levels are occupied up to the e2A orbitals and the antibonding e1B* orbital is empty † Ferrocene is known to have rapidly rotating rings even in the solid state. The energy barrier to ring rotation is estimated to be 5–10 kJ mol21. At low temperatures the relative orientation of the rings is closer to eclipsed than staggered.It is convenient for the purposes of this article to assume the D5h symmetry consistent with an eclipsed structure. The symmetry labels for the MO of ferrocene will thus differ from those found in most text books where D5d symmetry is assumed however the essential details of the bonding are the same. Chemical Society Reviews 1998 volume 27 264 1 between the ring a2B p combination and the metal pz orbital. Thus as a consequence of the high symmetry of ferrocene only three of the molecular orbitals have significant metal ligand bonding characteristics. The changes in the wave functions which occur when the molecule is bent with a = 35° are illustrated in Fig.3. Fig. 4 gives the variation of one electron energies as a function of angle a. Lowering the symmetry from D5h to C2v‡ causes two of the three occupied d orbitals to become the same symmetry a1; as the molecule bends they mix and move apart in energy the upper one 4a1 becoming less stable and the lower one 3a1 eventually more stable. The third d orbital becomes b1 in symmetry and increases in energy though not as much as the 4a orbital. Both the 4a1 and 2b1 orbitals are directed towards the open side of the metallocene wedge. These two orbitals have greater amplitude towards the open side of the wedge. As the rings are bent destructive interference between the metal d-orbitals and the ring pp orbitals where the rings approach each other leads to this asymmetry in electron distribution and the outward pointing direction of these orbitals.The 3a1 orbital resembles a dx2 orbital pointing along the x axis. All three orbitals have their maximum probability density in the xz plane. Removal of the symmetry constraint also affects the top occupied cyclopentadienyl orbitals. In the lower symmetry all four symmetry adapted linear combinations can interact with the metal d orbitals. The e1A orbitals become the 2a1 and 1b1 orbitals and drop in energy on bending becoming bonding rather than non-bonding. The drop in energy of these two orbitals is more or less mirrored by the rise in energy of the 4a1 and 2b1 orbitals. Of the e1B orbitals one becomes b2 in symmetry loses overlap with the metal orbitals and is raised in energy on bending; the other the 1a2 orbital remains almost constant in energy.The increase in the number of bonding interactions on bending suggests that bent structures might well be favourable however it is well established that all the isolated transition metal metallocenes have a parallel sandwich structure. Bent structures are favoured by the d0 pre-transition metals Ca Sr and Ba,12 though as the bonding forces involved with these elements are more electrostatic than covalent calculations show the structural preference to be slight. However there is a good case for believing that it is the d orbital occupancy that controls the angle between the rings of a metallocene. Fig. 5 shows how the calculated total energy of both ferrocene and the hypothetical triplet zirconocene vary with a.When all three d orbitals are fully occupied as in ferrocene the total energy of the molecule is raised significantly in energy on bending. However when the 4a1 orbital is unoccupied as in zirconocene with the configuration 3a1 1 2b1 1 the overall energy drops slightly. Thus ferrocene shows a strong preference for a parallel ring arrangement whereas the hypothetical moiety ZrCp2 would show little resistance to angle variation. Further theoretical investigation of the optimised structures of 4d metallocenes shows a definite pattern (Table 1).§ If the 4a1 orbital is occupied parallel or nearly parallel ring structures result but if it is unoccupied a bent structure with a > 30° is favoured.The small deviation from planarity for triplet MoCp2 and TcCp2 is expected on the basis of the Jahn–Teller theorem ‡ The highest possible symmetry for a bent metallocene unit is C2v. The axis system we assume throughout this article is shown in Fig. 1. Although there is universal agreement that the C2v axis lies in the z direction different choices of x and y axes lead to different labels for orbitals in a C2v molecule so care must always be taken when comparing results from different studies. C2v is conveniently a sub group of D5h so we can correlate the orbitals of a parallel and a bent metallocene. Our choice of axes means that the sh of the D5h molecule becomes the xz plane in C2v symmetry. § The results presented here are from density functional calculations using the Amsterdam Density Functional code ADF 2.0.1 Baerends and te Velde Vreije Universiteit Amsterdam 1996.Fig. 3 Isosurfaces for the orbitals of a bent metallocene unit. (These were generated using the Cerius2 package of Molecular Simulations Inc.) indeed matrix isolated molybdenocene has been shown to be bent.13 Tungstenocene has a parallel sandwich structure as spin orbit coupling suppresses the Jahn–Teller distortion. TiCp*2 [Cp* = (h-C5Me5)],14 a d2 metallocene has not been structurally characterised but may well be bent. Table 1 Optimised bending angles and average metal–carbon distances for the second row metallocenes. Metal Configuration 1 12b1 04a1 0 1 22b1 04a1 0 1 22b1 14a1 1 1 22b1 24a1 1 Y Zr Zr Nb Nb Mo Mo Tc Ru 3a 3a1 12b1 14a1 0 3a 3a1 12b1 14a1 1 3a1 22b1 14a1 0 3a 3a1 22b1 24a1 0 3a 3a1 22b1 24a1 2 Where two states with differing spins are possible the lower energy state is that of maximum multiplicity but the average Spin state M–C av/Å a° 0 40.9 2.62 36.3 2.49 47.8 2.46 2.43 40.5 2.37 18.4 2.33 35.3 2.29 11.2 2.26 2.21 doublet triplet singlet quartet doublet triplet singlet doublet singlet 0 Energy (eV) 2124.59 2125.96 2125.74 2127.63 2126.79 2127.28 2126.48 2128.36 2126.74 metal–carbon distance tends to be shorter in the low spin bent structures.The preference for the maximum number of unpaired electrons will be greater in the first transition series where exchange interactions are stronger and ligand field splittings less.In conclusion it is the d electron configuration that controls the geometry of the transition metal metallocenes. The commonly found parallel ring structures are a consequence of the d electrons avoiding anti-bonding interactions and minimising electron–electron repulsion. In the absence of these forces there is no inherent weakening of the metal ring bonding on bending. 3 Further bonding by a metallocene unit The 3a1 2b1 and 4a1 orbitals of a bent metallocene constitute the three frontier orbitals which can be used to bind further ligands. The simplest example of this is the formation of the hydrides MCp2H III (M = Re) MCp2H2 IV (M = Mo W) and MCp2H3 V (M = Nb Ta).In all cases the hydride ligands lie in the xz plane of the metallocene unit. 265 Chemical Society Reviews 1998 volume 27 H M M H M H H H H V III IV Fig. 4 Variation of one electron energies of ferrocene as a function of bending angle a Fig. 5 Variation of total energy of ferrocene and triplet zirconocene as a function of bending angle a In each case the metallocene unit provides a symmetry match to the symmetry adapted linear combinations (SALCs) of the hydrogens’ 1s orbitals (Fig. 6). III with one M–H bond has a d4 configuration with two lone pairs of a1 and b1 symmetry. IV has one lone pair of a1 symmetry and V none. Photoelectron studies of these compounds clearly show the high lying lone pairs which have low ionisation energies (IEs).The M–H bonding electrons have IEs of similar magnitude to the top occupied cyclopentadienyl orbitals.4 Chemical Society Reviews 1998 volume 27 266 Fig. 6 MO schemes for binding H1-3 to MCp2 to give III IV and V. The orbitals are drawn as projections on the xz plane. The HOMOs of III IV and V are denoted by a double arrow. There is good physical evidence for the spatial location of the a1 HOMO in a compound with the stoichiometry MCp2X2. In the series where X = Cl and M = Zr Nb and Mo the compounds have the configuration d0 d1 and d2 respectively. The Cl–M–Cl angle decreases along the series 97.1° (Zr) 85.6° (Nb) and 82.0° (Mo).3 The presence of electrons in a dx2 like orbital causes the Cl–M–Cl angle to close up.In MoCp H–Mo–H angle is 75.5°.15 Thus VI provides a better picture of the lone pair in a metallocene than does VII. An ESR study of a number of d1 compounds shows the half-occupied a1 orbital to consist mainly of dx2 with a small admixture of dy22z2.5 2H2 the In the trihydride TaCp2H3 the H–Ta–H angles are 63° and the adjacent H–H distance is 1.85 Å.16 Such close approach is unfavourable for larger ligands and there are no structurally characterised examples of metallocene units bound to three monodentate ligands other than hydrogen. Both NbCp2H3 and [MoCp2H3]+ show an 1H NMR spectrum in a certain temperature regime consistent with an AB2 spin system but with unusually large JAB coupling constants; that of the Nb compound is 20.4 Hz above 243 K while that of the Mo compound varies from ca.1000 Hz at 203 K to 450 Hz at 153 K.17 Such a massive value is due to quantum mechanical X X M M X X VII VI exchange coupling. It has been argued that the the main factor governing the magnitude of the exchange coupling is the stability of an [MCp2(h-H2)H]n+ structure VIII relative to the minimum energy trihydride.18 However Heinekey19 maintains that the formation of an H2 intermediate is not required but that a large quantum mechanical exchange coupling is favoured by a soft vibrational potential for the WHAHB wag. A low frequency for this motion allows close approach of the hydrogens and substantial overlap of the hydride wave functions leading to the observed tunnelling.H H M H VIII Deuteration of the d2 hydride WCp*2H2 to form the cation [WCp*2H2D]+ occurs preferentially between the two hydrogens (1) giving 90% of the central isotopomer (1a) rather than the lateral one (1b),20 which would be expected by direct protonation of the occupied 3a1 orbital. Attack is at the hydride ligands and is subject to charge rather than orbital control. Subsequent exchange between the central and lateral positions is found to occur at a faster rate than deprotonation and further deuteration demonstrating the presence of an intramolecular exchange mechanism. Kinetic studies on protonation of WCp2H2 suggests the presence of the intermediate [WCp2(h2-H2)H]+ in the protonation reaction.21 + + H D+ (1) Cp* Cp*2W Cp*2W 2W D H H H H D H 1a 1b Several other bonding situations for the versatile bent metallocene unit are discussed by Lauher and Hoffmann.7 A case not explicitly considered there is the bonding of a metallocene unit to an imido group to form MoCp2NR complexes.The NR22 ligand is a strong p donor and therefore has the capacity to form a triple bond to a transition metal. A metallocene fragment readily provides only one acceptor p orbital the 2b1 orbital in the xz plane. The dyz orbital is involved in bonding to the rings in the 2b2 orbital. The anti-bonding counterpart 3b2 descends in energy on bending but lies some 1.5 eV above the 4a1 orbital. It may be considered as a fourth frontier orbital of the MCp2 unit (Fig.6). In the compound MoCp2NBut the Mo–N–C angle is 177.7° indicating that both N pp orbitals are interacting with the metal and the Mo–N distance is consistent with a bond order between 2 and 3; both factors suggest a b2 MoUN p bonding interaction.22 Further evidence for donation from the imido group into the 3b2 orbital is the lengthening of the Mo–C distances opposite the NR group. In the 3b2 orbital the interaction between the metal and these carbons is anti-bonding (Fig. 7). Thus strong p donor ligands can compete effectively with the cyclopentadienyl groups for the metal d orbitals and consequently weaken metal– cyclopentadienyl bonding. It is found that the rings in such imido compounds are more readily displaced than in other metallocenes.Fig. 7 Iso-surface for the 3b2 orbital of a bent metallocene unit the fourth frontier orbital A similar bonding picture has been established for biscyclopentadienyl metal-oxo compounds.23 The lack of cylindrical symmetry in the metallocene unit which provides only one effective p acceptor orbital for the oxo group leads to a build up of charge on the oxygen making it susceptible to electrophilic attack. 4 Some structural features of substituted metallocenes Substitution of the hydrogens on a cyclopentadienyl group increases its size and can change its electron releasing properties.12 It can also convey chirality on the complex to which it is bound.24 The bulk of the substituents is important in determining the angle between the rings of a bent metallocene.Thus for a fixed metal–ring distance the larger the substituent the smaller the value of a needed to avoid steric interference between the groups. This is illustrated by the angles found for the tin metallocenes; SnCp2 has a = 46° SnCp*2 has a = 35.4° and Sn(h-C5Ph5)2 has a = 0°. As with group 2 metallocenes the electronic preference for a bent structure appears to be marginal and the angle is controlled by the size of the substituent. ansa-Metallocenes are those which have the two rings connected by a bridging group. Introduction of a bridge may constrain the geometry of the metallocene. A single atom bridge can increase a in comparison with the analogous compound with unbridged rings.Fig. 8 shows the angles g found for MCp2Cl2 where M = Ti Zr and Hf and for the analogues where the rings are bridged by SiMe2 or CMe2 groups.25 The M–Cp centroid and M–Cl distances do not vary much from those of the parent metallocenes but the Si bridge reduces g by 0–4° and the C bridge by up to 10–13°. A single Si bridge has much the same effect as a two atom C2Me4 bridge. The metal tends to lie closer to the ipso-carbons and the two carbons adjacent to it than to the two opposite it and the C–C bond opposite the ipso-carbons is shortened. When the rings are bent back by an ansa-bridge the coordination site at the metal is opened up and also the metal becomes more electrophilic.25,26 267 Chemical Society Reviews 1998 volume 27 Cl Cl g M g M Me2Si Cl Cl g 131.0 129.3 127.1 Cl g M g 128.7 125.4 126.8 R' R Si n M = Ti Zr Hf Si Me2C D (2) Fe Cl Fe R R' g 121.5 116.6 117.1 Fig.8 Cp(centroid)-metal-Cp(centroid) angles g° found for a series of Group 4 metallocene dichlorides The angles a b and f (Fig. 1b) are not given for these molecules but in a series of ansa-bridged niobocene metallocenes with CMe2 as the bridging group and similar g values a values of 63–68° are found with f values of 16–18°.27 A comparison with ansa-bridged ferrocenes also known as ferrocenophanes which have a single Si bridge shows that e values are similar but in contrast the values for a lie around 20° and f has a value of 37°.Because the d6 ferrocene unit is much more reluctant to bend there is considerable strain energy in forming the ansa-ring and this is evidenced by the greater departure from planarity at the ipso-carbons. 5 Comments on selected reactions of metallocenes and their derivatives The reactions described below are chosen to illustrate the role of the three co-planar orbitals and the interplay between orbital occupancy and inter-ring angle. In particular they focus on the differences between reactions of ansa-bridged metallocenes and the unbridged analogues. 5.1 Ring opening polymerisation of ferrocenophanes The ansa-ring strain present in d6 metallocenes has been utilised in forming ferrocene polymers.28 For example heating [1]silaferrocenophane induces quantitative exothermic ring opening polymerisation (ROP) [see (2)].This ROP reaction may also be initiated at room temperature by using ionic initiators such as ferrocenyllithium followed by hydrolytic work-up (3). Early transition metal ansa-metallocenes are thermally stable and show no tendency to rupture the ansa-bridge. These differences in reactivity are consistent with the energy change of bending the various dn configurations discussed above. Chemical Society Reviews 1998 volume 27 268 Me Fe Si Me Li Fe Me Si Me Fe Fe Fe Me Si n Me H2O or SiMe3Cl Me Si Me Fe Fe Fe Me Si n Me 5.2 Elimination from d2 metallocenes and addition to d4 metallocenes Brintzinger and co-workers6,29 showed that whereas MoCp2(CO) and WCp2(CO) are thermally stable compounds formation of CrCp2(CO) is reversible.All three compounds had near identical CO stretching frequencies suggesting similar binding in the three CO complexes. The difference in reactivity was analysed by a combination of semi-empirical MO theory and ligand field theory and attributed to the repulsion between the d electrons that needs to be overcome on forming the CO complex. The metallocene has a triplet ground state with a configuration of 3a1 22b1 14a1 1. The CO complex is a singlet with a d4 configuration 3a1 22b1 2. Thus a promotion term is involved in the bond formation. The energy difference between the triplet ground state and the singlet excited state is greater for a first row metallocene thus there is a bigger energy input in formation of CrCp2(CO) and conversely a greater energy gain on decomposition.Significant reactivity differences are found between ansabridged and non ansa-bridged group 6 metallocene derivatives. 30 The dihydride WCp2H2 photochemically eliminates dihydrogen forming the triplet tungstenocene intermediate13 which can for example insert into the C–H bonds of benzene [see (4)]. In contrast the ansa-bridged analogue W[(h5- C5H4)2CMe2]H2 is photochemically inert. C H hn W W H Similarly WCp2MeH thermally eliminates methane at 60 °C [see (5)] whereas W[(h5-C5H4)2CMe2]MeH is thermally stable at 110 °C. The contrast between the two latter compounds is of particular interest as they both show intramolecular hydrogen exchange between the methyl group and the metal hydride.For WCp2MeH this hydrogen exchange is only marginally faster Li E (3) E = H or SiMe3 n = 0–32 H 6H6 (4) W H W Me d2 3a1 2 D H W C Me D W CH3 D W C CH3 d2 3a1 2 W D than the elimination reaction.31 The exchange reactions pass through a mid-point in which a CH4 moiety is bound to the H H IX metal through two of its hydrogens IX and X.32 At this midpoint the d configurations are 3a1 22b1 2. The transition states for exchange are estimated to be of very similar energy. Elimination of methane from this mid-point is however calculated to be very different for the unbridged and ansa-bridged systems.The energy profiles for elimination are plotted in Fig. 9 as a function of W–C distance for a C2v system. The energy is referenced with respect to the methyl hydrides at zero. The midpoints for exchange IX and X are at the minima in the singlet energy curves. In both cases the tungstenocene product is expected to be in a triplet state. The unbridged compound has a close cross-over point to the triplet state and subsequent energy gain on forming triplet tungstenocene with parallel rings. In the ansa-bridged compound where the bridge prevents the rings achieving a parallel conformation the singlet–triplet cross-over is at a greater metal–methane distance and is of higher energy. Also the eventual triplet product is of higher energy than the transition state for the exchange.The driving force for reductive elimination in these Group 6 metallocene derivatives is seen to be the formation of the favoured parallel ring structure for the d4 configuration metallocene product. When the rings are constrained by the + CH (5) 4 W d4 e2"3a1'1 H W (6) CH2D H (7) W C CH2D H H H C C W H H H X Fig. 9 Energies of triplet and singlet C2v methane complexes (a) [W(h- C5H5)2(CH4)] and (b) [W((h-C5H4)2CH2)(CH4)] at varying W–C distance with respect to the energy of the corresponding methyl hydride ansa-bridge the four d electrons to avoid being spin paired have partially to occupy the high energy 4a1 orbital. In the reactions of d0 titanium metallocene derivatives Brintzinger has found that titanocene derivatives with an interannular ethylene bridge while resembling the unbridged analogue in many transformations involving Ti(iv) or Ti(iii) oxidation states do not undergo reactions which are thought to involve a free titanocene Ti(ii) intermediate.33 Theoretical estimates of the energy difference between triplet titanocene with an a value of 50° which is typical of that found in a Ti(iv) metallocene derivative and the optimised a value of 26° is 0.63 eV or 60.78 kJ mol21 .Such an energy gain available to the unbridged titanocene is not accessible if the inter-ring angle is constrained by an ansa-bridge; it may well account for the reactivity difference. For zirconocene the estimate is less 0.21 eV or 20.26 kJ mol21.This coupled with the greater difficulty in reducing Zr(iv) to Zr(ii) means less striking differences are expected in the reactions of ansa-bridged and non-bridged zirconocene derivatives. The principal differences found appear to be associated with the increased electrophilicity of the metal associated with the presence of the ansa-bridge.26 5.3 Zeigler–Natta polymerisation catalysed by metallocene derivatives Zirconocene dichloride in the presence of excess methylalumoxane (MAO) was found to catalyse the polymerisation of ethene to high density polypropylene.9 Variation of the cyclopentadienyl groups has led to effective processes for both isotactic and syndiotactic polymerisation of propylene. 10,11,34 There is abundant evidence that the active species in these metallocene polymerisation catalysts are monomeric cationic zirconocene alkyls e.g.[ZrCp2R]+. The role of the MAO is to generate and stabilise the d0 14 electron cationic species. The mechanism for Zeigler–Natta alkene polymerisation proposed by Coss�ee and Arlman35 (8) involves a four centre transition state (8c) in which the new C–C bond is forming. If a methyl group and a coordinated ethene are brought close to one another steric repulsion between the hydrogens occurs before the carbon atoms are close enough to form an incipient bond. Brookhart and Green proposed a modification of the Coss�ee– Arlman mechanism in which the alkyl group tilted and formed 269 Chemical Society Reviews 1998 volume 27 0.0 Erel / kJmol–1 2 H C Zr CH2 H X X (9a) P H b-agostic resting state R R X + X M M X X X (8a) vacant site P C H C Ti C + C C C Ti (8b) p-complex b-agostic p-complex R X M X X X (8c) transition state P C H C Ti C C C C (9d) Fig.10 Key steps on the energy profile for insertion of ethene into the Zr–C bond of [ZrCp2Et]+ as calculated by Lohrenz et al.38 g-agostic product The necessity for three available orbitals is underlined by the stability of the d2 cation [WCp2(h2-C2H4)Me]+. This cation fails to insert ethene into the W–C bond. Though the two d electrons will be involved in back-donation to the ethene resulting in stronger olefin binding than in a d0 analogue it is likely that the occupancy of the third orbital prevents the agostic bond formation needed to lower the activation energy barrier for the insertion step.The history of metallocene chemistry is a classic example of the symbiosis of experiment and theory in its broadest sense. Synthetic ingenuity detailed studies of mechanism and careful application of physical techniques have all played their part. In the future the relative availability and improved accuracy of modern computational methods should mean that they will also be useful tools to the synthetic chemist in planning experiments. 1 M. L. H. Green Organometallic Compounds 3rd edn. Methuen London 1968 vol. 2. 2 C. Balhausen and J. P. Dahl Acta Chem. Scand.1961 15 1333. 3 J. C. Green M. L. H. Green and C. K. Prout J. Chem. Soc. Chem. Commun. 1972 421. 4 J. C. Green S. E. Jackson and B. Higginson J. Chem. Soc. Dalton Trans. 1975 403. 5 J. L. Petersen and L. F. Dahl J. Am. Chem. Soc. 1975 97 6416 6422 and references therein. 6 H. H. Brintzinger L. L. Lohr and K. L. Tang Wong J. Am. Chem. Soc. 1975 97 5146. 7 J. W. Lauher and R. Hoffmann J. Am. Chem. Soc. 1976 98 1729 and references therein. 8 M. L. H. Green Pure Appl. Chem. 1978 50 27 and references therein. 9 A. Andersen H.-G. Cordes J. Herwig W. Kaminsky A. Merck R. Mottweiler J. Pein H. Sinn and H.-J. Vollmer Angew. Chem. Int. Ed. Engl. 1976 15 630. 10 J. A. Ewen J. Am. Chem. Soc. 1984 106 6355. 11 W. Kaminsky K. Külper H. H. Brintzinger and F.R. W. P. Wild Angew. Chem. Int. Ed. Engl. 1985 25 507. 12 M. L. Hays and T. P. Hanusa Adv. Organomet. Chem. 1996 40 117. 13 P. Grebnik R. Grinter and R. N. Perutz Chem. Soc. Rev. 1988 17 453. 14 J. E. Bercaw J. Am. Chem. Soc. 1974 96 5087. 15 A. J. Schultz K. L. Stearley J. M. Williams and R. Mink Inorg. Chem. 1977 16 3303. 16 R. D. Wilson T. F. Koettzle D. W. Hart Å. Kvick D. L. Tipton and R. Bau J. Am. Chem. Soc. 1977 99 1775. 17 D. M. Heinekey J. Am. Chem. Soc. 1991 113 6074. 18 S. Camanyes F. Maseras M. Moreno A. Led�os J. M. Lluch and J. Bertr�an J. Am. Chem. Soc. 1996 118 4617 and references therein. an a-agostic hydrogen bond to the metal (9a).36 Tilting of the methyl group relieves the steric hindrance betwen the alkyl and olefinic substituents in the transition state (9b).Formation of the agostic bond also provides an explanation for control of the stereochemistry of the polymerisation of propene since the substituents could lie either cis- or trans- with respect to the planar transition state. Extensive experimental and theoretical investigations37 have lent support to this picture of the reaction pathway. The results from one such theoretical study on the insertion of ethene into the metalUcarbon bond of [ZrCp2C2H5]+ are summarised in Fig. 10.38 The resting state of the process is calculated to be a b-agost. The olefin binds to this exothermically. The alkyl chain then rotates to form an a-agostic p-complex which undergoes the insertion process resulting in a g-agostic product which then returns to a b-agostic resting state.Throughout the insertion reaction the key atoms the alkyl a-carbon and a-agostic hydrogen the two olefinic carbons and the metal all lie in the same plane. To enable such a reaction pathway the metal must provide three co-planar orbitals one to bind the olefin one to form the metal–alkyl bond and one to form the agostic bond. Also the orbitals need to be sufficiently proximate for the C–C bond to form readily. The three frontier orbitals of a bent metallocene fulfill this role admirably. Chemical Society Reviews 1998 volume 27 (9c) 270 –20.0 –37.1 –34.1 2 C H H C 2 CH CH 2 2 Zr H Zr 3 CH2 CH H H2C X CH (8) a-agostic p-complex R X M X X X (8d) vacant site (9) a-agostic transition state 6 References (9b) P H Ti –32.1 –71.3 –98.1 2 H2 C H2 C H C CH2 CH2 Zr Zr Zr CH CH CH 3 3 CH CH C2H5 H H H b-agostic resting state 19 D.M. Heinekey A. S. Hinkle and J. D. Close J. Am. Chem. Soc. 1996 118 5353 and references therein. 20 G. Parkin and J. E. Bercaw J. Chem. Soc. Chem. Commun. 1989 255. 21 R. A. Henderson and K. E. Ogilvie J. Chem. Soc. Dalton Trans. 1993 3431. 22 J. C. Green M. L. H. Green J. T. James P. C. Konidaris G. H. Maunder and P. Mountford J. Chem. Soc. Chem. Commun. 1992 1361. 23 A. J. Bridgeman L. Davis S. J. Dixon J. C. Green and I. N. Wright J. Chem. Soc. Dalton Trans. 1995 1023. 24 R. L. Halterman Chem. Rev. 1992 92 965. 25 R. M. Shaltout J. Y. Corey and N. P. Rath J. Organomet. Chem. 1995 503 205. 26 T. K. Woo L. Fan and T. Ziegler Organometallics 1994 13 2252. 27 N. J. Bailey unpublished work. 28 I. Manners Adv. Organomet. Chem. 1995 37 131. 29 K. L. Tang Wong and H. H. Brintzinger J. Am. Chem. Soc. 1975 97 5143. 30 A. Chernega J. Cook M. L. H. Green L. Labella S. J. Simpson J. Souter and A. H. H. Stephens J. Chem. Soc. Dalton Trans. 1997 3225 and references therein. 31 R. M. Bullock C. E. L. Headford K. M. Hennessy S. E. Kegley and J. R. Norton J. Am. Chem. Soc. 1989 111 3897 and references therein. 32 J. C. Green and C. N. Jardine J. Chem. Soc. Dalton Trans. 1998 1057. 33 J. A. Smith and H. H. Brintzinger J. Organomet. Chem. 1981 218 159. 34 J. A. Ewen R. L. Jones A. Razavi and J. D. Ferrara J. Am. Chem. Soc. 1988 110 6255. 35 E. J. Arlman and P. Coss�ee J. Catal. 1964 3 99 and references therein. 36 M. Brookhart and M. L. H. Green J. Organomet. Chem. 1983 250 395. 37 R. H. Grubbs and G. W. Coates Acc. Chem. Res. 1996 29 85 and references therein. 38 J. C. W. Lohrenz T. K. Woo and T. Ziegler J. Am. Chem. Soc. 1995 117 12793 and references therein. Received 12th January 1998 Accepted 6th April 1998 271 Chemical Society Reviews 1998 volume

ISSN:0306-0012    

DOI:10.1039/a827263z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Is alkylation the main mechanism of action of the antimalarial drug artemisinin? Anne Robert and Bernard Meunier* Laboratoire de Chimie de Coordination du CNRS 205 route de Narbonne 31077 Toulouse cedex 4 France Artemisinin is a sesquiterpene lactone with an endoperoxide function essential for its antimalarial activity against chloroquine-resistant strains of Plasmodium falciparum. The mechanism of action of this paradigm molecule for endoperoxide-containing antimalarial drugs is still open to debate. Are the artemisinin derivatives only responsible for oxidative stress or are they able to alkylate heme and parasite proteins? The characterization of a covalent artemisininheme model adduct supports the role of C-centered radicals generated by the reductive activation of the peroxidic bond of this class of drugs.Artemether (an artemisinin analogue) and BO7 (a synthetic antimalarial trioxane) are also able to alkylate a porphyrin cycle. 1 Introduction Nearly two billion people are at risk of malaria and the incidence of this disease is dramatically increasing since many Plasmodium falciparum strains the parasite responsible for the majority of fatal malaria infection have now become resistant to chloroquine. Some strains have also developed resistance to mefloquine and even to the naturally occurring and highly efficient antimalarial quinine. A major event in the history of malaria was the discovery in the early 17th century of the antimalarial activity of the bark of Cinchona the ‘Peruvian fever tree’.In 1820 Pelletier and Caventou two French pharmacists isolated the two main alkaloids responsible for this activity quinine and cinchonine. Selective breeding of a new variety of Cinchona trees was then developed and the new trees were producing up to 13% quinine ten times more than the older varieties in cultivation. Quinine was the main treatment for malaria until the 1930s up to the development of synthetic antimalarials based on a quinoline moiety of which chloroquine mefloquine and amodiaquine are the most commonly used. Quinine is now considered as being too toxic for prophylaxis or routine treatment of malaria but cases of resistance are rare and this molecule is still used via intravenous infusion to treat severe malaria. Resistance to chloroquine first appeared in Thailand and South America in the early 1960s,1 and many strains of P.falciparum resistant to nearly all quinoline drugs are now present in large parts of the world. Resistance is now so widespread that chloroquine is virtually useless in some parts of the world.2 The alarming spread of drug resistance has led the WHO to predict that in the absence of new antimalarial strategies the number of people suffering from malaria will double by the year 2010. Thus to circumvent this phenomenon of drug resistance it is imperative that novel drugs should be developed to treat the desease. In 1967 the government of China launched a programme to discover new antimalarial drugs and indigeneous plants used in traditional medicine have been systematically examined.3,4 The antipyretic activity of a decoction of leaves of Artemisia annua was described as long ago as 340 and its antimalarial activity in 1596 (the year of publication of the Chinese Compendium of Materia Medica).In 1972 Chinese researchers recovered by extraction at low temperature from this plant a crystalline compound that they named qinghaosu (the name artemisinin is preferred by Chemical Abstracts RN 63968-64-9). Artemisinin H HO Quinine H N N HO N N Cl 3 Structure of some antimalarial drugs based on a quinoline moiety 1 is an enantiomerically pure sesquiterpene lactone bearing an endoperoxide function which has been proved to be essential for antimalarial activity the reduced peroxide deoxyartemisinin 5 (Scheme 1) being completely inactive.It is highly potent and is currently used in China South-East Asia and in some parts of Africa to treat more than a million people. However although they are highly active against polyresistant P. falciparum artemisinin and its effective analogues such as b-artemether 2b and sodium artesunate 2d have to be obtained by semi- Mefloquine 6 7 8 Chloroquine 5 6 2 7 1 8 H 5a 12a O O 12 H 8a 9 O 11O 10 O 1 O O O O O 1 Artemisinin a 6 C6H4F C6H4F O O CH3O 4 3 8 Cl 10 9 N NCF 3 5 4 2 3 1 3 8 H H 5 4a 7 7a 1 4 O 3 2 3 BO7 Structure of antimalarial drugs artemisinin b-artemether BO7 and arteflene.a Two different possible drawings are given for artemisinin. Chemical Society Reviews 1998 volume 27 4 Arteflene 5 H 6 4 7 O2 5a 3 1 H 8 O 12a 8a O 12 11O 10 9 O 1 H OH N N H N Amodiaquine NH CF O H 5a 12a H 8a 9 O 12 11O 10 H OR 2a R = H Dihydroartemisinin 2b R = Me b-Artemether 2c R = Et Arteether 2d R = C(O)CH2CH2COONa Artesunate 2 1 O 7 O 3 4 5 6 CF3 O CF 273 H Zn AcOH 2.5 h 3 1 2 O O H 4 2 O O 3 1 O O O H O O 1 H FeCl2•4H2O CH3CN 5–15 min H H 5 3 H 1 HO 2 H H O O O H3C C O O O O O 6 + (+5)aO For this purpose radioactive [14C]-artemisinin 1,20 [3H]-dihydroartemisinin 2a [3H]-arteether 2c or [14C]-arteflene 4 were incubated with P.falciparum infected erythrocytes.21 After treatment a proportion of the total parasite-associated radioactivity was bound to hemozoin as a low molecular mass artemisinin–heme adduct.20 Furthermore with radioactive dihydroartemisinin21 used at a pharmacologically relevant concentration (34 nm) this drug was able to alkylate some specific and not particularly abundant parasite proteins one of which has a similar size to that of ‘HRP’ (histidine-rich protein 42 kDa) a protein involved in the polymerization of heme in infected erythrocytes.22 The two other endoperoxides arteether 2c and arteflene 4 reacted with the same proteins.21 In contrast no proteins were alkylated under the same conditions in normal erythrocytes.None of the parasitic proteins were alkylated by the inactive reduced analogue deoxyartemisinin 5 supporting a specific mode of action of these peroxide-containing drugs. But the precise identities of the artemisinin–heme adduct(s) and target proteins are not yet known. Recent in vitro studies reported two possible modes of reactivity of artemisinin (i) is artemisinin an oxygen atom donor with respect to heme and (ii) is artemisinin an alkylating agent? 3 Is artemisinin an oxygen atom donor with respect to heme? We will consider two different cases the reactivity of artemisinin with simple transition metal ions and then with synthetic metalloporphyrins as heme models.7 Scheme 1 Reactivity of artemisinin according to the reaction conditions zinc powder in acetic acid or iron(ii) chloride in acetonitrile (after reference 23). a Compound 5 was also obtained when the reaction was carried in the presence of FeIIBr2 in tetrahydrofuran (reference 25). synthesis from extracts of the original plant Artemisia annua with all the drawbacks and cost that this entails [the growing of A. annua is possible only in limited geographical areas namely in the South Chinese and Vietnamese uplands and the yield of extraction is low (0.4%).5 Possible biosynthetic pathways have recently been described6]. Under these circumstances it will be difficult to extend the use of these artemisinin derivatives to the scale of billions of people.Fully synthetic compounds that share the benefits of artemisinin without its disadvantages (neurotoxicity has been reported at high doses of arteether 2c in monkeys7) and which can be made at low cost are clearly highly desirable. This has drawn attention to a new class of products having an endoperoxide function and are therefore capable of having a similar activity.8,9 Some of them have been prepared in particular RO 42-1611 (arteflene 4) and BO7 (‘Fenozan-50F’ 3) the latter being based on a cis-fused cyclopenteno- 1,2,4-trioxane exhibiting good activity on chloroquine-resistant Plasmodium strains and with a remarkably high safety margin.10,11 The development of this strategy obviously requires a precise knowledge about the mechanism of action of artemisinin at a molecular level.Like chloroquine and quinine these peroxidic antimalarial drugs act as blood schizonticides. However from the chemical structure of artemisinin and synthetic trioxanes it is apparent that these molecules share a common mode of action which will be different from that of the traditional quinoline-based antimalarials. In this review article we will focus on the molecular aspects of the mechanism of action of artemisinin (1) and the related semi-synthetic derivative b-artemether (2b) or a synthetic related peroxide having a significant biological activity BO7 (3). 2 What is known about the reactivity of artemisinin derivatives in infected erythrocytes? After infection of a person by the bite of an infected female Anopheles Plasmodium parasites first accumulate in hepatocytes then invade the erythrocytes for the next stage of their maturation.After a few days the infected red cells burst open and the merozoites are released causing the periodic fevers of malaria. These merozoites infect new erythrocytes and the intraerythrocyte cycle starts again. Within erythrocytes the parasite degrades the hemoglobin of the host and digests 30% or Chemical Society Reviews 1998 volume 27 274 more of the protein moiety using it as a source of amino-acids for the synthesis of its own proteins. The resulting free and potentially toxic heme residues are polymerized as a microcrystalline redox inactive iron(iii)–heme pigment called hemozoin which is insoluble in biological conditions and accumulates in the food vacuole.12,13 Only a small amount of heme is degraded by the parasite to be used as an iron source for its own metalloenzymes.Artemisinin and related peroxide-containing drugs are active on intraerythrocytic parasites at nanomolar concentrations the toxic concentration toward normal erythrocytes being in the range of micromolar. It has been proposed that free intraparasitic heme liberated during hemoglobin digestion might play an important role in the selective toxicity of artemisinin toward the parasite.14 Hemin was found to catalyze the reductive decomposition of artemisinin and dihydroartemisinin in vitro.15 Since the activity of artemisinin is inhibited by antioxidants (including catalase) it was initially proposed that the mechanism of action of this drug involved an oxidative stress leading to the destruction of the parasite.16 When incubated with normal erythrocytes artemisinin was shown to increase the methemoglobin concentration and to reduce slightly the intracellular glutathione and membrane fatty acid concentrations resulting in a dose-dependent increase of cell lysis.17 A combination of hemin and artemisinin oxidize erythrocyte membrane thiols in vitro whereas artemisinin alone or hemin alone has little effect.This oxidation is reduced in the presence of a free radical scavenger (a-tocopherol) or in the presence of deferoxamine which binds to iron ions resulting from the parasitic degradation of heme. Thus the artemisinin– hemin mediated oxidation reactions dependent on iron and mediated by free radicals were considered as due to an ironcatalyzed cleavage of the endoperoxide function of the drug.18 However it should be noted that all these experiments were made at concentrations ranging from 50 to 1000 mm i.e.at concentrations 103 to 105 times higher than effective in vitro antimalarial concentrations. It is therefore reasonable to consider that the parasite death in the presence of artemisinin is not due to non-specific or random cell damage caused by free radicals but might involve specific radicals and targets which have yet to be identified at the molecular level.19 3.1 In the presence of transition metal salts The formation of deoxyartemisinin 5 an artemisinin metabolite has been considered as an argument in favor of artemisinin acting as an oxygen atom donor to a metal complex (see reference 3 for the identification of artemisinin metabolites).Reaction of artemisinin in the presence of a Pd/CaCO3 catalyst caused the loss of a peroxidic oxygen atom to give the bis-ether derivative deoxyartemisinin 5.3 More recently it has been reported that the addition of zinc powder to a solution of artemisinin in acetic acid resulted after a few hours in a nearly quantitative conversion to deoxyartemisinin (Scheme 1). The same experiment carried out with 2b produced deoxyb-artemether.23 The same result was previously obtained with the synthetic trioxane 3.24 This deoxygenation is obviously a two-electron process which might be enzymatically induced in vivo leading to the inactive deoxy metabolite.However there is absolutely no indication that such a reaction is implicated in the parasiticidal event. In the presence of iron(ii) chloride tetrahydrate in acetonitrile for no more than fifteen minutes artemisinin was completely FeIII H • H O O 8 O 4 H C3—C4 cleavage 6 H2C• O 5 1 H3C C 3 O 2 O O H 9 O 9 FeIIIO O FeII H H O O H3C C O O O 6 Scheme 2 Mechanism of iron(ii) mediated degradation of artemisinin (the right part of the scheme—route 2—is reproduced from reference 8) Ph Chemical Society Reviews 1998 volume 27 H 4 2 O 3 1 H route 2 route 1 O O 1 O H 3 H O 1 HO 2 O H O O O O 7 epoxide opening converted to a ring-contracted furano acetate derivative 6 and 3a-hydroxydeoxyartemisinin 7 (resulting from a hydroxylation at position C4a of artemisinin) with yields of 78 and 17% respectively.24 In the presence of iron(ii) bromide in tetrahydrofuran, 25 the degradation of artemisinin produced the same products 6 7 and also 5 in 29 10 and 59% yields respectively.Posner et al. proposed that compounds 6 and 7 arose from an iron(ii)-induced homolytic cleavage of the peroxidic bond a RO· radical being formed either on O2 or on O1 of artemisinin (routes 1 and 2 respectively Scheme 2).8 The RO2· radical 8 (Scheme 2 route 1) rearranged by homolytic cleavage of the C3–C4 bond to produce the non-sterically hindered C4-centered primary radical 9.In the absence of an easy substrate to alkylate this radical might react with the Fe–O1 bond to give rise to the ring-contracted derivative 6 with release of a FeII salt. This reactivity which supposes the rapid isomerization of the O-centered radical to a C-centered radical by cleavage of the adjacent C–C bond has also been evidenced by Jefford using BO7 as substrate.24 Ha H FeIIIO • H O O O O H 1,5-H shift • HO O O 12 FeIIIO (a) O FeII H FeII H HO O O O 13 FeIV=O H Ph (b) rebound epoxidation S Me O S Me H • H 1 O H O FeIII—OH O 5 b-scission H H HO O O cyclization O O 14 + FeIII—O• OH 275 That the ester formation is a driving force of the rearrangement of the intermediate radical 9 was confirmed by calculations on models using 1-methoxycyclopentan-1-yloxyl radical 10 the heats of formation of 10 and 11 indicated that the conversion of the O-radical 10 to the C-radical 11 was a strongly exothermic process (Scheme 3).11 The analogue of compound 6 (but with an ethyl ether at position 12 instead of CNO) was one of the microbial metabolites of arteether 2c but was not found in mammalian metabolism.26 These data indicate that the reaction pathway with a C4-centered radical might be involved in vivo.H H H H O HH H • H O H H H 10 1-methoxycyclopentan-1-yloxyl Heats of formation; –59 kcal mol–1 276 Scheme 3 Isomerization of an O-centered radical to a C-centered radical (after reference 11) In route 2 of Scheme 2 the RO1· radical initially formed rearranged via a stereospecific 1,5-H shift from H the C4-centered secondary radical 12.It should be noted that a reaction pathway proceeding via a carbon-centered radical at position C4a has already been suspected to be important for the antimalarial activity of artemisinin.27 From radical 12 it has been proposed that deoxyartemisinin 5 was formed via a capture of H· and release of FeIII-OH the presence of cyclohexa- 1,4-diene as hydrogen atom donor increasing the yield of 5. In fact the initial alkoxyl radical RO1· is a better H-atom abstracting agent from an external source than the alkyl radical 12 (this has been evidenced in autoxidation reactions).The formation of 5 might therefore occur directly from RO1H via the release of FeIII-OH. From the C4-centered radical 12 the C3ahydroxy deoxyartemisinin 7 might be formed by two different pathways a direct intramolecular epoxidation with release of iron(ii) followed by a OH2 mediated intramolecular hydrolysis of the intermediate epoxide 13 [Scheme 2 route 2(a)] or by an indirect route involving the elimination of FeIII-O· from the C4a-radical 12 followed by epoxidation of the C3–C4 double bond of the intermediate 14 and then intramolecular opening of the a-oriented epoxide [Scheme 2 route 2(b)]. It should be noted that this latter proposed alternative pathway (and only that one) requires a high-valent epoxidizing iron-oxo species such as FeIII-O· (or FeIVNO).In fact the formation of all the characterized products from route 2 can be explained by route 2(a) alone without the second hypothetical route 2(b). Furthermore it should be noted that the aromatization of hexamethyl Dewar benzene sulfoxidation of thioanisole and hydroxylation of tetrahydronaphthalene all of which have been invoked as proof for an iron-oxo intermediate are reactions which can be achieved by efficient one-electron oxidants but which do not absolutely require an oxygen atom transfer. Hexamethyl Dewar benzene can be isomerized to hexamethylbenzene via a radical cation chain rearrangement even in the absence of high valent metal-oxo species [up to 48% isomerization was obtained in the presence of tris(p-bromophenyl) aluminium hexachloroantimonate alone].28 Methyl phenyl sulfoxide and hydroxytetrahydronaphthalene can be obtained by a one-electron oxidation followed by nucleophilic attack of a water molecule.One of the most efficient and selective reactions of a putative iron(iv or v)-oxo species should be the epoxidation of an electron-rich olefin such as cyclohexene. Exposure of artemisinin to one equivalent of FeCl2·4H2O in acetonitrile for 15 min at room temperature provided cleanly the same two products as above the furano acetate as the major product and 3-hydroxydeoxyartemisinin (6 and 7 respectively Scheme 1). In the presence of cyclohexene the same products were obtained in Chemical Society Reviews 1998 volume 27 d-radical of methyl pentanoate –77 kcal mol–1 H O 11 HHHH • H 3.2 In the presence of a metalloporphyrin We therefore decided to investigate the possibility of generating manganese(iv)-oxo or manganese(v)-oxo species from artemisinin and a synthetic manganese tetraarylporphyrin.31 Several attempts to epoxidize cyclohexene with artemisinin (2 equiv.compared to the substrate) in the presence of catalytic amounts of MnIII(TMP)Cl or MnII/III(Cl12TMP)Cl32 (5 mol% with respect to the olefin) were performed. These experiments were unsuccessful there was no olefin conversion and no traces of epoxide after 30 min at room temperature suggesting that no 4a leading to metal-oxo species was generated by artemisinin in the presence of a metalloporphyrin.H O HH HH In fact hydroperoxides are known to be poor oxygen atom donors with respect to metalloporphyrins or non-heme metal complexes,29,33 and no evidence has been reported up to now that dialkylperoxide compounds are better oxygen atom donors than hydroperoxides. The main reaction pathway when a peroxide is activated by a transition metal complex corresponds to the homolytic cleavage of the weak peroxidic O–O bond. 4 Is artemisinin an alkylating agent via C-centered radicals? The best way to answer this question will be to isolate and to characterize fully a covalent adduct of artemisinin with a low weight molecule like heme (or a heme model) or with a parasitic protein.In fact early studies on the activation of artemisinin derivatives have been described using simple metal salts and then developed with metalloporphyrins. 4.1 In the presence of metal salts Artemisinin was reported to form adduct(s) with heme after generating an unidentified oxy radical which after rearrangement can also alkylate malarial proteins.4,21 The reaction products of BO7 with iron(ii) chloride tetrahydrate have been carefully identified thus giving information about the possible intermediates.24 Under different reaction conditions 14 was a constant product of the reaction (Scheme 4). The origin of this product is rationalized by the rupture of the 84% and 8% respectively and no traces of cyclohexene oxide were detected.23 The same lack of epoxidation was observed with b-artemether and with a synthetic trioxane close to BO7 (bearing two phenyl groups instead of p-fluorophenyl at positions 6 and 7a) in the presence of FeIICl2 in acetonitrile acetic acid or tetrahydrofuran.24 However it must be noted that even with good oxygen atom donors such as PhIO NaOCl or KHSO5 no oxygen atom transfer reaction hydroxylation or epoxidation is efficiently catalyzed by simple iron salts.Only porphyrin or Schiff base ligands provide a suitable coordination sphere around manganese or iron to catalyze oxygenation reactions.29,30 H H H H C C6H4F 6H4F Fe2+ Fe2+ O O BO7 O– C C 6H4F O O• –O1 • 12 13 H+ H H H H C C6H4F 6H4F O Bu O 6H4F +H-atom C O O C 6H4F 6H4F HO HO • 14 13¢ Scheme 4 Mechanism of iron(ii) mediated degradation of BO7 (from reference 24) peroxidic bond of BO7 after a single electron transfer from an iron(ii) ion giving the radical anion 12 which quickly isomerizes to 13 or 13A after protonation.The alkyl radical 13A is trapped in the presence of a thiol to afford the pentanoate 14. The intermediate radical 13 is completely analogous to radical 9 (Scheme 2) produced by activation of artemisinin under similar conditions. This common pathway for artemisinin and BO7 is an indication that these compounds probably share the same reaction mode. The radicals R–CH2· 9 and 13 are potentially alkylating species. We therefore decided to trap these intermediates by characterization of covalent adducts between these radicals and a synthetic heme model.4.2 In the presence of a metalloporphyrin As a heme model we used a hydrophobic complex MnIITPP generated in situ by borohydride reduction of MnIII(TPP)Cl (TPP stands for the dianion of tetraphenylporphyrin). This synthetic analogue of heme was expected to provide a limited number of products because of its four-order symmetry. By reacting MnIII(TPP)Cl with three equivalents of artemisinin b-artemether or BO7 in the presence of borohydride in dichloromethane manganese chlorin-type adducts were formed between the macrocycle and one of these antimalarial drugs. The demetallation of these manganese(ii) adducts was achieved by adding a solution of acetic and hydrochloric acids (95 5 v/v) directly to the reaction mixture.In the cases of artemisinin or BO7 this treatment allowed the removal of manganese without any other modification of the corresponding adducts. However in the case of b-artemether this drastic demetallation procedure could not be applied owing to the lability of the B cycle of b-artemether itself under such strongly acidic conditions. A gentler demetallation was then carried out by transmetallation of the Mn(ii)-chlorin-b-artemether adduct to its analogous cadmium(ii) derivative followed by demetallation of the cadmium(ii) adduct under very mild conditions (Scheme 5). The resulting demetallated adducts between the Scheme 6 Mechanism of alkylation of the heme model MnIITPP by artemisinin drastic demetallation route HCl/AcOH 5/95 MnIITPC-artemisinin adduct H2-TPC-artemisinin adduct N2 30 min r.t.conditions as above adduct modifications MnIITPC-artemether adduct CH2Cl2 solution lmax = 440 nm H2-TPC-artemether adduct lmax = 420 nm soft demetallation route transmetallation Cd(NO3)2 in DMF 5 min r.t. air N2 30 min r.t. CdIITPC-artemether adduct demetallation AcOH/H2O 1/9 277 lmax = 434 nm Scheme 5 Demetallation of the manganese(ii)–tetraphenylchlorin–artemisinin and tetraphenylchlorin-b-artemether adducts tetraphenylchlorin and the trioxane-derived moiety were completely characterized by the usual analytical methods including 2D-NMR (adducts 16 and 17 with artemisinin and artemether respectively Scheme 6; adduct 21 with BO7 Scheme 7).In all three cases the covalent chlorin-drug adducts were not minor compounds 20–30% of pure adducts were obtained. These adducts resulted from addition at the b-pyrrolic position C2A of the macrocycle of a non-sterically hindered alkyl radical derived from the antimalarial drug namely radical 9 in the case of artemisinin (or its analogue in the case of artemether) and 13 in the case of BO7. A secondary alkyl radical was therefore formed at the adjacent position C3A of the macrocycle leading after an intramolecular electron abstraction by Mn(iii) to a carbocation at C3A. The addition of borohydride at this position produced the dihydropyrrole ring. Borohydride also mediated the reductive elimination of the ester at positions C12 of artemisinin and C4a of BO7.In the case of artemisinin the introduction of two hydrogen atoms from borohydride at positions C3A of the macrocycle and C12 of the artemisinin Chemical Society Reviews 1998 volume 27 Scheme 7 Mechanism of alkylation of the heme model MnIITPP by BO7 fragment was confirmed by using borodeuteride instead of borohydride as reducing agent. The reduction at position 12 did not occur in the case of b-artemether. The ‘entire’ drug was conserved in the addition product of tetraphenylchlorin and the b-artemether-derived radical. Adduct 17 was therefore isolated without loss of any fragment from the drug (for a preliminary communication about alkylating properties of artemisinin see ref. 34; complete characterization of these products including artemether and BO7 derived adducts and analysis of the mechanism of their formation have been reported in reference 35).The antimalarial activity of BO7 was not influenced by the absolute configuration of the molecule the pure enantiomers being no more active than the racemate.11 However its mode of action on the intraerythrocytic parasite was rationalized in terms of close docking by a twist-boat conformer of the 1,2,4-trioxane cycle with a heme molecule (Scheme 8). the close interaction between the iron(ii) of heme [or manganese(ii)] and the peroxide bond results in a one-electron transfer to the O–O antibonding orbital causing the fast scission of the O–O bond. Scheme 8 Nestling of the twisted-boat conformer of BO7 on the surface of heme and activation of the peroxide bond.(Reproduced with permission from reference 11 SET stands for single electron transfer). Chemical Society Reviews 1998 volume 27 278 The generated acetal radical then irreversibly isomerizes to a non bulky C-centered radical the ultimate alkylating and probably lethal agent. In addition it should be noted that BO7 (with a cyclopentane and two p-fluorophenyl substituents) is a highly hydrophobic molecule like artemisinin indicating that this non-aqueous solubility is commonly shared by promising antimalarial drugs.4 Among the semi-synthetic derivatives of artemisinin many hydrosoluble molecules have been synthesized bearing for instance an amine a glycosyl an ether or more recently a butyric acid side-chain.36 Only few of them are pharmacologically active and usually on a narrow spectrum of parasite strains considerably reducing interest in them.Furthermore conversion of esters derived from artemisinin to the corresponding acids drastically reduces their antimalarial activity suggesting again that the lipophilicity probably related to a fast diffusion through membranes plays an important role in determining the antimalarial activity of this series.4 Since heme itself is a rather hydrophobic molecule the lipophilicity and the boat conformation of 1,2,4-trioxanes are two necessary factors for their docking with heme and then for the initiation of the alkylation reaction. 5 Concluding remarks All these results based on a heme model clearly indicate that one of the major modes of reactivity of artemisinin and related synthetic trioxanes is their reductive activation leading to the homolytic cleavage of their peroxide bond.The resulting alkoxyl radicals are rearranged in non-sterically hindered C-centered radicals acting as powerful alkylating agents. These data on the alkylating properties of artemisinin derivatives also indicate that a heme molecule in the presence of artemisinin is not going to perform cytochrome P-450-type chemistry involving high-valent metal-oxo species. These alkylation reactions involve the generation of drug radicals in the vicinity of heme in a solvent-cage controlled reaction instead of as freely diffusible radicals. The characterization of these artemisinin-type adducts confirms that the alkylating properties of artemisinin is not limited to this natural compound but is a common feature probably required for the antimalarial activity of endoperoxide-containing drugs.These compounds should be able to alkylate either heme itself or parasitic proteins. This alkylation process which it is proposed occurs at a pharmacologically relevant concentration of the drug,21 would inhibit the proteases responsible for the hemoglobin digestion within infected erythrocytes thus starving the parasite of essential amino acids. The alkylation and inactivation of proteins involved in the heme polymerisation namely the ‘histidine-rich protein’ would poison the parasite with redox active heme molecules. High heme concentrations are supposed to be responsible for oxidative stress within the cell inducing a disruption of membranes and they have also been shown to inhibit a parasitic hemoglobinase.2 Alkylation of heme by a drug-derived radical may also be directly involved in the parasite death through the accumulation of non-polymerizable redox-active heme adducts which could also behave as inhibitors for heme-polymerisation enzymes. Finally the characterisation of these covalent adducts resulting from C-alkylation of a heme model by radicals derived either from artemisinin artemether or a synthetic peroxide will be useful for the interpretation of the mass spectra of parasitic proteins alkylated by artemisinin and related derivatives. It will also contribute to giving a better molecular basis for the rational design of new synthetic antimalarial drugs having an endoperoxide function.6 Acknowledgements This work was supported by CNRS and by a grant from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases (Director’s Initiative Fund). Professor Charles W. Jefford (University of Geneva Switzerland) is gratefully acknowledged for a gift of synthetic trioxane BO7 and Rh�one-Poulenc-Rorer Doma (Antony France) for a gift of b-artemether (Paluther®). 7 References 1 S. J. Foote and A. F. Cowman Acta Tropica 1994 56 157. 2 M. Foley and L. Tilley Int. J. Parasitol. 1997 27 213. 3 D. L. Klayman Science 1985 228 1049. 4 S. R. Meshnick T. E. Taylor and S. Kamchonwongpaisan Microbiol.Rev. 1996 60 301. 5 K. L. Chan C. K. H. Teo S. Jinadasa and K. H. Yuen Planta Med. 1995 61 285. 6 R. K. Haynes and S. C. Vonwiller Acc. Chem. Res. 1997 30 73. 7 J. M. Petras D. E. Kyle M. Gettayacamin G. D. Young R. A. Bauman H. K. Webster K. D. Corcoran J. O. Peggins M. A. Vane and T. G. Brewer Am. J. Trop. Med. Hyg. 1997 56 390. 8 J. N. Cumming P. Ploypradith and G. H. Posner Adv. Pharmacol. 1997 37 253. 9 C. W. Jefford Adv. Drug Res. 1997 29 271. 10 W. Peters B. L. Robinson G. Tovey J.-C. Rossier and C. W. Jefford Ann. Trop. Med. Parasitol 1993 87 111. 11 C. W. Jefford S. Kohmoto D. Jaggi G. Timari J.-C. Rossier M. Rudaz O. Barbuzzi D. G�erard U. Burger P. Kamalaprija J. Mareda G. Bernardinelli I. anares C. J. Canfield S.L. Fleck B. L. Robinson and W. Peters Helv. Chim. Acta 1995 78 647. 12 A. F. G. Slater W. J. Swiggard B. R. Orton W. D. Flitter D. E. Goldberg A. Cerami and G. B. Henderson Proc. Natl. Acad. Sci. USA 1991 88 325. 13 D. S. Bohle R. E. Dinnebier S. K. Madsen and P. W. Stephens J. Biol. Chem. 1997 272 713. 14 S. R. Meshnick A. Thomas A. Ranz C.-M. Xu and H.-Z. Pan Mol. Biochem. Parasitol. 1991 49 181. 15 F. Zhang D. K. Gosser Jr. and S. R. Meshnick Biochem. Pharmacol. 1992 43 1805. 16 S. R. Krungkrai and Y. Yuthavong Trans. R. Soc. Trop. Med. Hyg. 1987 81 710. 17 M. D. Scott S. R. Meshnick R. A. Williams and D. T.-Y. Chiu J. Lab. Clin. Med. 1989 114 401. 18 S. R. Meshnick Y.-Z. Yang V. Lima F. Kuypers S. Kamchonwongpaisan and Y. Yuthavong Antimicrob.Agents Chemother. 1993 37 1108. 19 S. R. Meshnick Lancet 1994 344 1441. 20 Y.-L. Hong Y.-Z. Yang and S. R. Meshnick Mol. Biochem. Parasitol. 1994 63 121. 21 W. Asawamahasakda I. Ittarat Y.-M. Pu H. Ziffer and S. R. Meshnick Antimicrob. Agents Chemother. 1994 38 1854. 22 D. J. Sullivan Jr. I. Y. Gluzman and D. E. Goldberg Science 1996 271 219. 23 C. W. Jefford M. G. H. Vicente Y. Jacquier F. Favarger J. Mareda P. Millasson-Schmidt G. Brunner and U. Burger Helv. Chim. Acta 1996 79 1475. 24 C. W. Jefford F. Favarger M. G. H. Vicente and Y. Jacquier Helv. Chim. Acta 1995 78 452. 25 G. H. Posner J. N. Cumming P. Ploypradith and C. H. Oh J. Am. Chem. Soc. 1995 117 5885. 26 I.-S. Lee and C. D. Hufford Pharmac. Ther. 1990 48 345. 27 G. H. Posner C. H. Oh D. Wang L. Gerena W. K. Milhous S. R. Meshnick and W. Asawamahasakda J. Med. Chem. 1994 37 1256. 28 T. G. Traylor and A. R. Miksztal J. Am. Chem. Soc. 1987 109 2770. 29 B. Meunier Chem. Rev. 1992 92 1411. 30 W. Nam and J. S. Valentine J. Am. Chem. Soc. 1993 115 1772. 31 A. Robert M. Boularan and B. Meunier C. R. Acad. Sci. Paris 1997 324II 59. 32 P. Hoffmann A. Robert and B. Meunier Bull. Soc. Chim. Fr. 1992 129 85. 33 P. A. MacFaul I. W. C. E. Arends K. U. Ingold and D. D. M. Wayner J. Chem. Soc. Perkin Trans. 2 1997 135. 34 A. Robert and B. Meunier J. Am. Chem. Soc. 1997 119 5968. 35 A. Robert and B. Meunier Chem. Eur. J. 1998 in the press. 36 A. J. Lin A. B. Zikry and D. E. Kyle J. Med. Chem. 1997 40 1396. Received 9th March 1998 Accepted 9th April 1998 279 Chemical Society Reviews 1998 volume 27

ISSN:0306-0012    

DOI:10.1039/a827273z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Dicatechol ligands novel building-blocks for metallo-supramolecular chemistry Markus Albrecht The self-assembly of helicate-type coordination compounds from oligodentate ligands and two or more metal ions is an important part of metallo-supramolecular chemistry. The design of appropriate ligands the fascinating structures of the self-assembled coordination compounds and the understanding of the mechanisms of metal-directed self-assembly processes are topics which have recently come into the focus of attention. In this paper the formation characterization and some properties of helicate-type compounds which are formed from oligo-catechol ligands are described. 1 Introduction Institut für Organische Chemie Universität Karlsruhe Richard-Willstätter-Allee 76131 Karlsruhe Germany Fig.1 Schematic representation of a double- and a triple-stranded helicate Helicity is a motif which is present in many macroscopic natural or artificial objects. Several people artists or architects as well as scientists are fascinated and inspired by the beauty of helical structures.1 O O O O Fe N N N O OH N O O O H NH O O O HN NH HN HN O NH HN NH O O O H On a microscopic (molecular) level helicity can be found for example in proteins or in double stranded DNA. A class of artificial helical supramolecular aggregates which has recently come into the focus of attention are the helicates. They are formed by spontaneous and cooperative self-assembly of two or three linear oligo-donor ligand strands and two or more metal ions.Due to the preferred coordination chemistry of the metal centers double- triple- or quadruple-stranded helicates are obtained (Fig. 1). Investigations into the formation structure and synthetic use of helicates have been mainly performed with nitrogen-donor ligands and only to a minor extend with oxygendonors. 1,2 O N HO N N O N O Fe O O O O [(1)3Fe2] 1-H2 (rhodoturulic acid) In 1968 Neilands reported the isolation and characterisation of the siderophore rhodoturulic acid (1-H2) the diketopiperazine of d-N-acetyl-l-(S)-d-N-hydroxyornithine.3 1-H2 bears two hydroxamic acid units as binding sites for metal ions and forms with iron(iii) ions enantiomerically pure binuclear D-configured helical complexes [(1)3Fe2].4 3Fe2] was the first isolated triple-stranded Although [(1) helicate only few examples of analogous oxygen donor complexes are described in the literature.As an example ligand NH HN O N N O OH OH OH OH O O 3-H4 O O O O 2 Markus Albrecht was born in 1964. He studied chemistry in Würzburg and Münster and obtained his Dr. rer. nat. in 1992. After one year as a postdoctoral fellow with Professor Kenneth N. Raymond in Berkeley he moved to the Institute of Organic Chemistry at the University of Karlsruhe and received his habilitation in 1997. His work on metal-directed self-assembly processes was honoured by the ‘ADUCJahrespreis 1996 für Habilitanden’. O O OH HO 2-H2 281 4-H Ligands 1- 2- 4-H2 and 3-H4 and the first triple-stranded helicate [(1)3Fe2] Chemical Society Reviews 1998 volume 27 4 form in the presence of titanium(iv) ions the 3Ti3].6 Just re- 2-H2 can be deprotonated twice and is then able to form a neutral triple-stranded helicate with iron(iii) ions.5 Three ligands of 3-H enantiomerically pure trinuclear helicate [(3) cently the ligand 4-H2 with two 1,3-dicarbonyl moieties was introduced and its helicates [(4)3M2] (M = Ti V Mn Fe) were described.7 However most of the oxygen donor ligands which were used for the self-assembly of helicate-type complexes contain catechol units as binding sites for metals.Herein we describe the metal directed self-assembly structure and some properties of helicate-type coordination compounds which are formed from oligo(catecholato) ligands.2 Dicatechol ligands A number of dicatechol ligands which bear different spacers were synthesised to enable the systematic study of the selfassembly of helicates with oxygen donor ligands. One class of such compounds possesses amide linkers in the spacer to connect two catechol units (5a–f-H4).8–12 A second class of ligands (6-H4 7-H6) contains pure alkyl chains as spacers.13215 The chiral ligands 8-H4–10a-H4 were prepared to investigate their use in the self-assembly of enantiomerically pure helicates. 9,10,16,17 3 Self-assembly of helicate-type coordination compounds 4[(5e)2- 3a Formation of triple-stranded dinuclear metal complexes from dicatechol ligands The ligands 5–10 were used in metal directed self-assembly processes.In one case a double-stranded helicate Na (MoO2)2] was obtained and could be structurally characterised. 12 All other experiments led to dinuclear (5 6 8–10)8–11,15–22 (Scheme 1) or trinuclear (7)23 triple-stranded complexes of gallium(iii) iron(iii) or titanium(iv). The dinuclear complexes are obtained in quantitative or close to quantitative yield. Only with the long chain ligands 6g,h-H4 OH HO OH = HO O NH (a) NH O (b) OH HO = (CH2) n [ n = 2 (c) 3 (d) 5 (e)] OH HO 5-H4 OH HO OH HO HO OH HO O NH 7-H6 HO OH O O HN NH R R = ( S)-CHMePh 10a-H4 R = Pri 10b-H4 Linear oligocatechol ligands OH Chemical Society Reviews 1998 volume 27 282 O M O M¢n 3 O M O 3 O HO HN O 5f-H HN 4 OH OH OH OH OH HO O NH Ph K6[(5a,b,f 10)3Ga2] [Et4N]6[(5d 8)3Ga2] [Et4N]6[(5c)3Fe2] M'4[(6a–f 9)3Ti2] (M¢ = Li Na K) Scheme 1 Formation of dinuclear helicate-type complexes The enantiomerically pure chiral helicates M is a mixture of isomers or oligomers detected by NMR spectroscopy.However the species which are formed in these reactions are still soluble. No precipitation of polymeric material is observed. For entropic reasons the dinuclear complexes are favoured with respect to oligomer or polymer formation. 4[(9)3Ti2] (M = Li Na K),17 [(3)3Ti3],6 K6[(10a)3Ga2],16 or [Et4N]6[(8)3Ga2]10 could be obtained from oxygen donor ligands. A high specific rotation of [a]D = +970 (±50) (c = 1) is observed for K4[(9)3Ti2] in methanol (free ligand [a]D = +53; c = 1 methanol).17 4 4 The cooperativity of the self-assembly process of the helicate-type complexes was shown for several dinuclear coordination compounds.Reaction of appropriate metal ions with an excess of ligands 5a,b,d,f-H 8,10 or 6b,c-H 18,19 in the presence of base selectively led to the binuclear complexes. No mononuclear species could be observed by NMR spectroscopy or mass spectrometry. A mixture of the ligands 5a,b,f-H4 leads in the presence of gallium tris(acetylacetonate) and potassium carbonate to a OH HO 2) n 6-H4 [ n = 1 (a) 2 (b) 3 (c) 4 (d) 5 (e) 6 (f) 8 (g) 10 (h)] OH OH HO O HN Ph 9-H4 8-H4 OH HO O 2 Mm+ base OH (CH HO O NH O HN R NH mixture of the three helicates [(5a)3Ga2]62 [(5b)3Ga2]62 and [(5f)3Ga2]62 in which every dinuclear complex contains only one kind of ligand.In this case the molecular self-recognition process is controlled by the different spacer length of the three ligands.8 3b The structure of triple-stranded dinuclear metal complexes The dinuclear coordination compounds can adopt different structures. Due to the spacer length of the ligands 6a,b,c,9-H4 binding of both binding sites of one ligand to the same titanium ion should not be possible (structure A). O O O Ti O O O O O Ti O O O O A O O O O O O Ti Ti O O O O O O O O O O O Ti O Ti O O O O O O helicate (L,L-form) tion.15,18,20,22 3Ti2]42.meso-helicate (D,L-form) Possible stuctures of dinuclear coordination compounds However there are two possible diastereomeric structures. If both complex units possess the same configuration the triplestranded helicate is formed. If the two units are differently configured the corresponding achiral meso-compound is obtained the meso-helicate. In the D3-symmetric helicate the ligands show an ‘S’-type conformation while the ligands of the C3h-symmetric meso-helicate adopt a ‘C’-type conforma- The helicate- as well as the meso-helicate-type structures can be observed for the alkyl-bridged derivatives [(6) X-Ray structural analyses could be obtained for the dinuclear titanium complexes Li4[(6a)3Ti2]·6DMF,20 Li4[(6b)3Ti2]·6 DMF·2H2O,19 4[(6c)3Ti2]·6DMF·5H2O,18 Na and K4[(6f)3Ti2]·5 DMF·3 H2O.15 3Ti2]42 with an odd number of methylene The tetraanions [(6) 3Ti2]42 [(6c)3Ti2]42} possess the meso-helicate units {[(6a) structure in the solid state.The same structure is observed in solution. On the other hand the compounds with an even spacer length {[(6b)3Ti2]42 [(6f)3Ti2]42} adopt the helicate structure (Fig. 2). In [(6f)3Ti2]42 the stereochemical information is transferred through seven s-bonds to influence the relative configuration of the complex moieties and form the chiral helicate. The results discussed above show that it is possible to obtain selectively either the chiral helicates or the achiral mesohelicates by choosing an appropriate spacer.The self-assembly process itself is highly diastereoselective (only one stereoisomer is observed) and can be controlled in the case of the alkylbridged dicatechol ligands 6. A ligand with an odd number of methylene units yields the meso-helicate while an even number leads to the corresponding helicate.The tetraanions [(6)3Ti2]42 represent structures of cage compounds in which internal oxygen atoms are present. The high negative charge and the Fig. 2 Molecular structures of the tetraanionic helicates [(6b)3Ti2]42 and [(6f)3Ti2]42 and the corresponding meso-helicates [(6a)3Ti2]42 and [(6c)3Ti2]42 cryptand type structure should be ideal for the encapsulation of cations. Thus in the solid state one of the counterions can be observed in the interior of [(6b)3Ti2]42 (Li)19 or [(6c)3Ti2]42 (Na).18 As a representative example the structure of the monomeric species of Li4[(6b)3Ti2]·6 DMF·2 H2O is shown in Fig.3.19 Three of the lithium ions are located outside of the cryptand type helicate (one is bound to four DMF molecules; two are bound to terminal oxygen atoms of the helicate and to DMF molecules which act as bridges to obtain a linear coordination polymer in the solid state). The fourth lithium cation is found in the interior of the tetraanion binding to two internal oxygen atoms of one of the ethylene linked dicatechol ligands 6b. Additionally this ion is coordinated to two water molecules which by hydrogen bonding to the remaining internal Chemical Society Reviews 1998 volume 27 Fig.3 Molecular structure of Li4[(6b)3Ti2]·6 DMF·2 H2O in the crystal 283 In [(6f) catecholate oxygen atoms are additionally fixed in the interior of this cavity. The arrangement of ligands around this lithium atom allows the stabilisation of an unusual distorted square planar coordination geometry.19 3Ti2]42 the cavity is much too large to encapsulate only one cation. Consequently two potassium cations (which are electronically saturated by coordination to DMF or water) are located in the interior of this tetraanionic cryptand in the solid state. K4[(6f)3Ti2]·5DMF·3H2O crystallises in the orthorhombic space group P212121 with spontaneous separation of the enantiomeric helices.15 The X-ray structures of the helicates K6[(5a)3Ga2],9,16 [Et4N]6[(5c)3Fe2],11 and [Et4N]6[(5d)3Ga2]10 are also described in the literature.They all possess a triple-stranded helicate structure. However no inclusion of guests could be observed for those compounds with amide linkages in the spacer. Only for [(2)3Fe2] could the binding of one molecule of water be shown in the solid state.5 3Ti2]42} The inclusion of sodium cations in the metallacryptands [(6b,c)3Ti2]42 in solution can be shown by 23Na NMR spectroscopy of the corresponding sodium salts. At room temperature a broad signal is observed for the sodium ions. The line width of the signal indicates that an exchange process between different cations takes place. Cooling of the NMR samples ([2H4]methanol) results in a splitting of the signals and at low temperature an intense signal can be observed for ‘free’ solvated sodium ions {d = 20.9 for [(6b)3Ti2]42}.Additionally a small signal is detected {d = 215.5 for [(6b) which is assigned to encapsulated sodium. The exchange barrier can be estimated to be approximately DG‡ = 32 kJ mol21.1,15 The observation of alkali metal cations which are bound to the helicates in the solid state and in solution shows that the counterions are part of the supramolecular structures and not just innocent spectators. Therefore those ions should have an influence on the properties as well as on the self-assembly of the cryptand-type helicates and meso-helicates. Fig. 4 Schematic representation of the inversion of dinuclear helicates and meso-helicates and the observation of the racemisation of K4[(6b)3Ti2] by dynamic 1H NMR spectroscopy (in D2O) 3- N O H 3c Dynamic behaviour of the dinuclear complexes A property of the triple-stranded helicates or meso-helicates which are formed from oligocatechol ligands is the racemisation or symmetrization of the metal complexes.This inversion process can be monitored by dynamic NMR spectroscopy using either diastereotopic protons of the spacer in the ligands 6 and 7 or of the substituents bound to the ligand 10b. 4[(6b)3Ti2] in D2O at O Ga O H N O 3 [(11)3Ga]3- Fig. 4 shows the 1H NMR spectra of K variable temperature.19 At room temperature signals of the aromatic units and two multiplets (d = 2.76 and 2.44) of the diastereotopic spacer protons can be observed.Heating the sample leads to coalescence of the signals of the alkyl protons at 328 K and at higher temperature only one sharp singlet is detected. From those results a DG‡ value of 64.4 kJ mol21 is estimated. Similar experiments with the alkyl-bridged complexes [(6)3Ti2]42 show that the symmetrization of the helicates or meso-helicates is influenced by several different factors:21 (1) The inversion barriers of the meso-helicates are higher than those observed for the helicates. (2) Within the helicate or meso-helicate series the free 3Ga2]62 This indicates that the inversion of the helicate [(10b) proceeds by a stepwise mechanism with a meso-helicate structure as intermediate. The analogous complex [(12)3Ga2]62 does not show interconversion of the C1- and C3-symmetric isomers during racemisation.This again shows that the inversion follows an intramolecular mechanism.9,16 The dinu- The inversion of [(6) energy barriers are lower for compounds with longer chain lengths. (3) Due to interaction of the cations with the tetraanions the cations possess an enormous influence on the inversion behaviour of the dinuclear complexes. 3Ti2]42 21 as well as of clear and trinuclear helicates M4[(6b)3Ti2] and M6[(7)3Ti3] (M = Na or K) possess approximately the same inversion barrier and therefore again a consecutive inversion of the metal complex units takes place. This means for the trinuclear complex M6[(7)3Ti3] that the inversion starts at one end of the helicate followed by inversion of the central unit and finally of the second terminus.23 [(10b)3Ga2]62 9,16 proceeds by a non-dissociative mechanism.Thus no ligand exchange can be observed in the presence of excess ligand. For [(10b)3Ga2]62 it could be shown that the inversion barrier DG‡ = 80 kJ mol21 is only 1.2 times higher than the one observed for the analogous mononuclear complex [(11)3Ga]32.24 Chemical Society Reviews 1998 volume 27 284 3d Template-assisted self-assembly In the alkyl-bridged helicate- and meso-helicate-type complexes M4[(6)3Ti2] the counterions are part of the molecular structure of the complexes and are not only innocent spectators. Therefore they should play an active part in the self-assembly process which leads to the dinuclear coordination compounds.For [(6a)3Ti2]42 this could be shown by a series of experiments. If a mixture of ligand 6a-H4 [TiO(acac)2] (acac = acetyl acetonate) and alkali metal carbonate is stirred in methanol overnight one would expect to obtain the dinuclear meso-helicate. However with potassium carbonate only a mixture of undefined products can be obtained which probably contains a variety of different oligomers. With sodium or lithium carbonate only one defined species can be observed in the NMR spectra. The same spectra are obtained by addition of respectively LiClO4 or NaClO4 to the mixture of oligomers which was originally produced in the presence of potassium carbonate.20,22 OH K2CO3 OH 3 undefined material OH OH 6a-H2 Li2CO3 or Na2CO3 O O O Ti O O O O O O Ti O + 2 [TiO(acac)2] 3Ti2]42 O O Scheme 2 Template-directed self-assembly of [(6a) The small cations (Li+ Na+) are able to stabilise the mesohelicate structure by binding to the oxygen atoms of the catecholate ligands.Potassium is too large and no stabilisation of the dinuclear complex can occur. Therefore lithium and sodium can act as a template while potassium cannot. However this templating does not proceed by inclusion of one cation in the interior of the cryptand-type meso-helicate. The X-ray structural analysis of Li4[(6a)3Ti2] reveals that no lithium ion is bound in the interior but three cations bind to the periphery of the tetraanion and form a ‘molecular box’ (Fig.5).20 In solution a different situation is found (Fig. 5). As was shown by 6Li and 1H NMR spectroscopy at low temperature only two of the lithium cations bind to the tetraanion. In the 6Li NMR spectrum of 6Li4[(6a)3Ti2] at 193 K two sharp signals can be observed at d = 1.36 and 0.99 for the two diastereotopic ions which are bound to the meso-helicate [(6a)3Ti2]42. Additionally a broad signal can be detected at d = 0.90 for the two solvated lithium ions. In the 1H NMR spectrum the signals split at low temperature to form three sets of signals due to the loss of symmetry.22 In the case of the self-assembly of [(6a)3Ti2]42 a templating effect by the counterions is observed (Scheme 2). The lithium or sodium cations can act as template while potassium is not able to induce a self-assembly process to form defined complexes.Similar template-assisted self-assembly processes also seem to be important in the formation of various other supramolecular species.25,26 Fig. 5 The molecular structure of Li3[(6a)3Ti2]2 in the solid state and of Li2[(6a)3Ti2]22 in solution 4 Self-assembly of dinuclear coordination compounds from directional and sequential ligands Supramolecular systems with a high content of information are obtained if directional or sequential ligands are used for the selfassembly of helicate type complexes. The ligands 12–149,16,22,27 possess two different binding sites for metal ions and thus two different orientations of the ligands are possible. Two isomeric coordination compounds can be formed.Different binding sites at the ligands (ambident chelating ligands) should enable the selective binding of different metals which in a triple-stranded system would lead to a situation as depicted in Fig. 6 Type 1 structure. Selective binding of the metal ions to the different binding sites of the ligand occurs. On the other hand if only one kind of metal ion is used a Type 2 structure should be obtained in a self-assembly process. In such a homobinuclear complex the two metal ions have a very similar electronic situation and charge separation is minimised.27 4a Directional systems The directional ligand 12-H4 with Ga(acac)3 in the presence of KOH yields the dinuclear helicate K6[(12)3Ga2] as a mixture of 3- (cis) and C1-symmetric (trans) isomers in a ratio of the C 1.00 2.86 which is close to the statistical mixture.Variable temperature NMR studies in D2O show that the two isomers racemise (LLÔDD) independently without reorientation of the ligands (cis–trans isomerization).9,16 3- and C1-symmetric meso-helicate-type A mixture of the C complexes M3[(13)3Ti2] (M = Li Na) was obtained from ligand 13-H4 and [TiO(acac)2] in the presence of M2CO3. Again 285 Chemical Society Reviews 1998 volume 27 HN O OH OH OH HN O OH OH HN O OH OH 13-H4 OH HN Fig. 6 Schematic representation of the relative orientation of sequential ligands in triple-stranded dinuclear coordination compounds O O O O Ga not observed K6 K6 O O O O O The directional ligands 12-H4 13-H4 and 14-H3 NH NH NH NH 12-H4 a close to statistical distribution (found 1 4; expected 1 3) of the two isomers could be observed.22 No significant selectivity can be found in the self-assembly of triple-stranded dinuclear helicate-type complexes from the directional ligands 12-H4 and 13-H4.The unsymmetric C1 and the C3-symmetric isomers are formed close to the statistical O O O O Ga 3 C3-symmetric 6[(12)3Ga2] C1-symmetric K6[(12)3Ga2] Possible isomerisation of K Chemical Society Reviews 1998 volume 27 286 OH 2 OH NH O Ga O NH NH O NH O O O O OH HN HN HN HN 14-H3 O O O O O NH Ga 2 ratio (3 1).The reason for this is that the two binding sites of the directional ligands geometrically and electronically are too similar which does not lead to a discrimination between the two binding sites during self-assembly. 4b Sequential systems A sequential ligand in which the two binding sites possess different denticity was first realised by Piguet and co-workers.28 The orientation of this sequential ligand upon complexation is controlled by the preferred coordination geometry of the metals which were used.28 A ligand which possesses two geometrically very similar binding sites for metals which are different in their electronic features is 14-H3.27 2 OH base NH base HO 3)3 3 2 Ti(OMe)4 or 2 Ga(NO 14-H3 n– 2 – O O O O M Ga OH Ti(OMe)4 and Ga(NO3)3 2 N O H2 O 2 O 2 NH N N H H2 NH O O H N 2 O O M O Ti O O O O O [(14) [(14) O 3M2] n– 3GaTi]2 – 4 13 M = Ti ( n = 1) M = Ga ( n = 3) Scheme 3 Formation of dinuclear coordination compounds from the sequential ligand 14-H3 Self-assembly of pseudo-meso-helicate type homodinuclear complexes [(14)3M2]n2 (M = Ga Ti; n = 3 1) proceeds by mixing of three equivalents of 14-H3 with two equivalents of Ti(OMe)4 or Ga(acac)3 in the presence of alkali metal carbonate as base (Scheme 3).NMR spectroscopy reveals that the dinuclear complexes possess no symmetry (C1) and a full set of signals is observed for the triple-stranded complexes (39 carbon atoms in the 13C NMR spectra).This is in accordance with a Type 2 structure (Fig. 6) which was expected for a homodinuclear complex. Two of the ligands 14 are oriented in one the third in the opposite direction. No signals of a homodinuclear complex which adopts a Type 1 structure can be detected by NMR spectroscopy. If a 1 1 mixture of Ti(OMe) and Ga(acac)3 is reacted with 14-H3 and alkali metal carbonate the heterodinuclear complex [(14)3TiGa]22 (Fig. 7) is exclusively formed in a cooperative self-assembly process. By FAB MS spectrometry only the heterodinuclear complex and no homodinuclear complexes can be observed. The 1H as well as C NMR spectra of [(14)3TiGa]22 are very simple which indicates that a C3-symmetric Type 1 complex is formed.Only one set of ligand signals can be detected by NMR spectroscopy. The formation of [(14)3TiGa]22 can also be observed in an NMR experiment by mixing of the heterodinuclear compounds [(14)3Ga2]32 and [(14)3Ti2]2 in a ration of 1 1.27 The X-ray structural analysis of K2[(14)3TiGa]·6 DMF·diethyl ether shows the same Type 1 structure in the solid state as was observed for the dianion [(14)3GaTi]22 in solution. The titanium is bound to three catecholate units while three aminophenolato ligands are coordinated to gallium. The complex experiences an additional stabilisation through internal hydrogen bonding of the amines to the internal oxygen atoms of Fig. 7 Molecular structure of [(14)3TiGa]22 in the solid state the catechols. Therefore no templating by alkali metal cations needs to take place.27 5 Conclusions Linear oligocatechol ligands which bear either amide or alkyl spacers are excellent organic building blocks for the selfassembly of metallo-supramolecular architectures.In this review we discussed a number of aspects of this chemistry (i) Triple-stranded helicate-type oligonuclear complexes are obtained in cooperative self-assembly processes. (ii) Self-recognition can lead to defined coordination compounds from a mixture of differently sized ligands. (iii) The relative stereochemistry of the complex units can be controlled in the case of alkyl-bridged systems and racemisation or inversion of the compounds can be observed by NMR spectroscopy. (iv) Helicate-type complexes with a cryptand-type structure are able to bind alkali or alkaline earth metal cations in their interior.(v) The self-assembly process can be assisted by certain templates (e.g. appropriate counterions). (vi) With directional or sequential ligands two different structures of metal complexes can be obtained. If the two binding sites of the ligand are different selective formation of Type 1 heterodinuclear and Type 2 homodinuclear complexes can be achieved. The results which were discussed show the high potential of oxygen donor ligands in metal-directed self-assembly processes and it is expected that even more complex and fascinating structures will be observed in the future.26,29 6 Acknowledgments I thank my co-workers Cyrill Riether Oliver Blau Matthias Schneider and Karen Witt for their ongoing contributions to the work which has been discussed here.X-Ray structure analyses were performed by Dr Roland Fröhlich Dr Sirpa Kotila and Dr Peter Burger and NMR spectroscopy by Dr Herbert Röttele. This work was financially supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. 7 References 1 J.-M. Lehn Supramolecular Chemistry VCH Weinheim 1995 and references cited therein. 2 C. Piguet G. Bernardinelli and G. Hopfgartner Chem. Rev. 1997 97 2005 and references cited therein. 3 C. L. Atkin and J. B. Neilands Biochemistry 1968 7 3734. 4 C. J. Carrano and K. N. Raymond J. Am. Chem. Soc. 1978 100 5 R. C. Scarrow D. L. White and K. N. Raymond J.Am. Chem. Soc. 6 E. J. Corey C. L. Cywin and M. C. Noe Tetrahedron Lett. 1994 35 7 V. A. Grillo E. J. Seddon C. M. Grant G. Aromí J. C. Bollinger K. 8 D. L. Caulder and K. N. Raymond Angew. Chem. Int. Ed. Engl. 1997 9 M. Meyer B. Kersting R. E. Powers and K. N. Raymond Inorg. Chem. 5371. 1985 107 6540. 69. Folting and G. Christou Chem. Commun. 1997 1561. 36 1440. 1997 36 5179. 10 E. J. Enemark and T. D. P. Stack Angew. Chem. Int. Ed. Engl. 1995 34 996. 11 E. J. Enemark and T. D. P. Stack Inorg. Chem. 1996 35 2719. 12 A.-K. Duhme Z. Dauter R. C. Hider and S. Pohl Inorg. Chem. 1996 35 3059. 13 M. Albrecht Tetrahedron 1996 52 2385. 14 M. Albrecht Synthesis 1996 230. 15 M. Albrecht H. Röttele and P. Burger Chem. Eur. J. 1996 2 1264. 16 B. Kersting M. Meyer R. E. Powers and K. N. Raymond J. Am. Chem. Soc. 1996 118 7221. 17 M. Albrecht Synlett 1996 565. 18 M. Albrecht and S. Kotila Angew. Chem. Int. Ed. Engl. 1995 34 2134. 19 M. Albrecht and S. Kotila Angew. Chem. Int. Ed. Engl. 1996 35 1208. 20 M. Albrecht and S. Kotila Chem. Commun. 1996 2309. 21 M. Albrecht H. Röttele and M. Schneider Chem. Ber. Recueil 1997 130 615. 22 M. Albrecht Chem. Eur. J. 1997 3 1466. 23 M. Albrecht and M. Schneider Chem. Commun. 1998 137. 24 B. Kersting J. R. Telford M. Meyer and K. N. Raymond J. Am. Chem. Soc. 1996 118 5712. 25 R. G. Chapman and J. C. Sherman Tetrahedron 1997 53 15911. 26 B. Hasenknopf J.-M. Lehn N. Boumediene A. Dupont-Gervais A. Van Dorsselaer B. Kneisel and D. Fenske J. Am. Chem. Soc. 1997 119 10956. 27 M. Albrecht and R. Fröhlich J. Am. Chem. Soc. 1997 119 1656. 28 C. Piguet J.-C. G. Bünzli G. Bernardinelli G. Hopfgartner S. Petoud and O. Schaad J. Am. Chem. Soc. 1996 118 6681 and references cited therein. 29 R. W. Saalfrank I. Bernt E. Uller and F. Hampel Angew. Chem. Int. Ed. Engl. 1997 36 2482. Received 10th February 1998 Accepted 8th April 1998 287 Chemical Society Reviews 1998 volume 27

ISSN:0306-0012    

DOI:10.1039/a827281z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Transition metals as structural components in the construction of molecular containers Christopher J. Jones School of Chemistry The University of Birmingham Birmingham UK B15 2TT This article will briefly review the development of metallomacrocycle syntheses and consider the distinction between equilibrium controlled self-assembly processes and kinetically controlled but metal directed macrocycle formation reactions. The use of octahedral rather than square planar or tetrahedral metal coordination geometries can introduce configurational isomerism into the structure and raises the issue of stereoselectivity in kinetically controlled metallomacrocycle formation reactions. The incorporaton of transition metal centres into cyclophane like structures introduces new functionality.This may contribute to the binding of guest molecules by the macrocycle and can give rise to novel electron transfer luminescence or magnetic properties. 1 Introduction Supramolecular chemistry that is the formation of new chemical structures by the assembly of molecular sub-units bound together by non-covalent interactions,1 is having a major impact on current chemical research and will continue to do so into the foreseeable future. Early examples of supramolecular assemblies have been provided by ‘host–guest’ compounds in which a larger ‘host’ molecule includes a smaller ‘guest’ molecule within a cavity in its structure.2 One well studied example is provided by tetra-p-tert-butylcalix[4]arene (bcalH4) [(Fig. 1(a)].3 The inclusion of toluene into the hydrophobic cavity of bcalH4 provides an example of an ‘inclusion compound’ [Fig.1(a)]. Because they contain molecular cavities and have the capacity to include organic molecules as guests cyclophanes have played a central role in the development of such ‘host–guest’ chemistry. Cyclophanes have been defined as ‘receptor molecules in which at least one aromatic ring is bridged by at least one n-membered aliphatic bridge’.4 The incorporation of transition metal centres into such structures can confer new properties on these potential host molecules. In particular transition metal centres might introduce Lewis acidity,5 magnetism,6 redox activity7 or luminescence8 properties into the macrocyclic structure. This may have important implications for the chemical reactivity or physical properties of Chris Jones graduated in 1967 (University of Sheffield) and after a PhD in 1970 (Sheffield with Professor J.A. McCleverty) followed by postdoctoral work (UCLA NATO Fellow with Professor M. F. Hawthorne and Imperial College ICI Fellow with Professor G. Wilkinson) joined the staff at the Harwell Laboratory in 1974. In 1981 he moved to the University of Birmingham where he is now a Senior Lecturer. 4} (M = Cr Mo W) units. O O PhP PPh Cl Cu Cu O O R R Cl I Cl R R O O Cu Cu Cl PPh PhP O O [{Cu(m2-Cl)}4(m4-I)(res(PPh)4)] a cyclophane host; for example optically driven charge transfer processes between the metal centres and ligands in the macrocycle structure may lead to novel electro-optical properties.Further far reaching aspects of the use of metal ions in supramolecular chemistry arise from the particular structural features associated with transition metal centres and the range a) But But C6H5Me OHHO OHHO Me But But L L b) LM N N M" L N N N N LM" N N M L L L M = M" = Pd2+; L-L = H2NCH2CH2NH2 M =M" = Pd2+ Pt2+; L-L = Ph2PCH2CH2CH2PPh2 M = Pd2+; L-L = H2NCH2CH2NH2; M"(L-L)=Re(CO)3Cl CO c) CO CO O M O CO P P OC CO M* OC O OC P O P O O O O O P O CO CO CO OC P M* P O OC OC O P M CO CO M = M* = Cr Mo W; M = Cr M* = Mo; M = W M* = Mo Fig. 1 (a) Tetra-p-tert-butylcalix[4]arene as a host for toluene.(b) Some cyclic tetramers constructed from metal centres and 4,4A-bipyridyl units. (c) A cyclic tetramer constructed from a ditopic phosphine ligand and {M(CO) Chemical Society Reviews 1998 volume 27 289 a) X Co X N N N X Co X N N N N X Co X Co X X N N N N N X Co X N N N N X N Co X N N = N N Co X X N N N N Co X N N X N N N X Co X X = NCS b) i) Mn N N Mn N Mn N N N N N N N N N Mn Mn N Mn N N N N N N N N N N Mn Mn N N N Mn N N N N N N Mn Mn N N N N N N N Mn N ii) N N = N N N N N N N Cu+ N N O O N O O N N Cu+ N N N Cu+ N of thermodynamic and kinetic stabilities their complexes display. The incorporation of transition metal centres as vertices in a macrocyclic structure offers ready access to square planar or octahedral geometries [Fig.1(b) 1(c)] in addition to the linear trigonal or tetrahedral geometries available at carbon or other light p-block element centres. A consideration of crystal field theory tells us that in the presence of strong field ligands the use of d8 metal centres such as Pd2+ or Pt2+ should favour a square planar geometry at the metal centre whilst d6 electron configurations as in Ru2+ or Co3+ should favour an octahedral geometry. The use of d10 ions such as Cu+ or Zn2+ might be expected to give rise to a tetrahedral coordination geometry. Transition metal complexes containing suitably oriented ligands which can be substituted by a polyfunctional ligand capable of linking metal centres provide a means of forming macrocycles in a single reaction step.In such reactions particular combinations of ligand structure and metal reactivity may be exploited to favour a particular outcome. Thus rigid linear bridging ligands may be combined with square planar metal centres to produce cyclic tetramers [Fig. 1(b)] if two ciscoordination sites are blocked by a non-labile ligand.9 In the absence of a blocking ligand polymeric structures may arise (Fig. 2). More flexible ligands or ligands which offer a dihedral angle between the donor atom sites may afford cyclic dimers (Fig. 3).9,10 Apart from these structural features the variety of thermodynamic and kinetic properties exhibited by transition metal centres allows access to a range of reactivities which may be exploited during macrocycle synthesis.The use of kinetically more inert metal centres provides access to those macrocyclic products which form most rapidly whilst with kinetically labile metal centres the thermodynamically most stable macrocyclic products will be obtained. This article aims to review the use of transition metal centres as structural components in the construction of macrocyclic compounds containing aromatic rings. For the purposes of this review the term ‘metallocyclophane’ will be used for molecules in which one or more transition metal centres are involved in bridging one or more aromatic rings so as to form a macrocyclic structure. The syntheses and functionality of metallocyclophanes will be considered including their inclusion properties electron transfer reactions luminescence and magnetic properties.c) N N 2 The construction of metallocyclophanes O O N N + N N 2 and 4,4A-bpy or pyrazine (pyz)12 2(L–L)2] (L–L = 4,4A-bpy pyz) Cu+ Fig. 2 (a) ‘Square’ coordination polymers derived from Co(NCS)2 and 4,4Abipyridine or pyrazine. (b) ‘Rhombohedral’ coordination polymers derived from Mn(BF4)2 and bbi [1,4-bis(imidazol-1-yl)butane]. (i) shows the chemical composition and (ii) the way in which two rhombohedral lattices interlock. (c) An example of a helicate complex in which two tritopic ligands form intertwined helices. 2]PF6}n [bpe = trans-1,2-bis-(4-pyridyl)ethene] {[Cu(bpe) five.13 In aqueous ethanol the reaction between 4,4A-bpy and [Pd(en)(NO3)2] affords [Pd(en)(4,4A-NC5H4C5H4N)]4 8+ [Fig.1(b)] in essentially 100% yield. Stang and co-workers have demonstrated that similar reactions can be performed in nonaqueous media by using [M(dppp)(OTf)2] [M = Pd Pt; dppp = Ph2P(CH2)3PPh2; OTf2 = CF3SO32] [Fig. 1(b)].14 The high yields obtained in these reactions result from the 90° angle between the 4,4A-bpy ligands bound to the Pd2+ centre and the linear disposition of the donor atoms in the 4,4A-bpy ligands. These factors create a situation in which the cyclic tetramer 2.1 Self assembly reactions involving nitrogen heterocycle ligands The utility of metal centres in the construction of metallocyclophanes first became widely recognised in 1990 after Fujita’s report9 of the formation of a cyclic tetramer [Pd(en)(4,4A-NC5H4C5H4N)]4 8+ in the reaction between the ditopic non-chelating ligand 4,4A-bipyridyl (4,4A-bpy) and the kinetically labile PdII complex [Pd(en)(NO3)2] (en = H2NCH2CH2NH2) [Fig.1(b)]. This complex was not in fact the first example of a transition metal based cyclic tetramer to be reported. The reactions of [M(CO)6] (M = Cr Mo W) with the ditopic phosphorus donor ligand P(OCH2)3P had earlier been found to afford tetranuclear metallomacrocycles [Fig. 1(c)].11 Both of these reactions involve metal centres in which all but two cis-coordination sites are blocked by relatively inert ligands. This is important for the formation of molecular species since the reactions between ditopic ligands such as 4,4A-bpy and an unencumbered metal centre are likely to lead to coordination polymers (Fig.2). One example of this is provided by the reactions between Co(NCS) which produce [Co(NCS) containing sheets of trans-{Co(NCS)2(L–L)}4 units [Fig. 2(a)]. More complex three-dimensional structures are possible with larger bridging ligands and the crystal structure of {[Cu- (dap)2]PF6}n (dap = 2,7-diazapyrene) contains three interpenetrating lattices of {Cu(dap)}4 ‘squares’ whilst {[Cu(4,4Abpy) 2]BF4}n contains four interpenetrating lattices and Chemical Society Reviews 1998 volume 27 290 a) Z H N 2 N N H N 2 Pd2+ Pd2+ N N N H2 Z N H2 Z = -CH2- -C(OH)2- >C=O >C=CH2 -CH2CH2- 1,4-CH2C6X4CH2- (X = H F) 4,4'-CH2C6F4C6F4CH2- 2,6-CH2C10F6CH2- b) N N N N Co Cl Cl Co Cl Cl N N N N c) O Zr Zr O O O d) N HB N BH N N N Ar N E E N N N Mo N N N Mo E E N Ar N O O E =NH; Ar = E = O; Ar = 1,3-E2C6H4 1,3-(ECH2)2C6H4 1,4-(ECH2)2C6H4 4,4'-(EC6H4)2CH2 4,4'-(EC6H4)2C=O E = S; Ar = 1,3-E2C6H4 1,3-(ECH2)2C6H4 1,4-(ECH2)2C6H4 4,4'-(EC6H4)2CH2 4,4'-(EC6H4)2C2H4 4,4'-(EC6H4)2C=O Fig.3 (a) Cyclic dimers constructed with flexible ditopic pyridyl ligands. (b) The proposed structure of [CoCl2(1,3-bimb)]2 [1,3-bimb = 1,3-bis(benzimidazol-1-ylmethyl)benzene]. (c) The structure of [Zr(h5- C5H5)2{1,3-(OCH2)2C6H4}]2. (d) Binuclear complexes containing octahedral metal centres derived from [Mo(NO)(Tp*)I2].offers the most energetically favourable structure. The equilibrium constant for formation of a metal complex depends upon the free energy change in the reaction which in turn contains contributions from both enthalpy and entropy changes. Normally the assembly of a number of free components into a single oligomeric molecule will involve a decrease in the disorder within a system and give rise to an unfavourable i.e. negative entropy change. However this may be compensated for by the enthalpy change associated with metal–ligand bond formation. In a molecule involving a square planar or octahedral metal centre the strongest metal–ligand bonds will form when the angle at the metal centre is 90°. Thus although assembling eight components is less entropically favourable than assembling four or six the structural constraints on the {Pd(en)}2+/4,4A-bpy system rule out the formation of cyclic dimers and favour cyclic tetramers in which the ‘square’ geometry gives stronger Pd–N bonds.Since the PdII centre is kinetically labile the system can rapidly explore the various combinations of metal and ditopic ligand available leading to the formation of the enthalpically favoured cyclic tetramer. This reaction constitutes an example of self-assembly in that it proceeds to equilibrium in high yield. 2O)4]2+.16 In contrast the reaction between 4,4A-bpy and the more kinetically inert PtII complex [Pt(en)(NO3)2] proceeds under kinetic control to give a mixture of oligomeric products.Only after heating at 100 °C for some weeks do these more rapidly formed kinetic products convert to the more thermodynamically favoured cyclic tetramer.9 Hunter et al. have developed a thermodynamic model of the self-assembly process.15 This relates the concentration range over which a self-assembled macrocyclic structure of a given size is stable to the concentration of the reactants and the strength of the interactions between them. Changing the structure of the ditopic ligand used in these selfassembly reactions can affect the degree of oligomerisation in the macrocycles formed. Thus the use of more flexible ligands which can accommodate the 90° angles at metal ions within a binuclear or trinuclear macrocycle may give rise to the formation of cyclic dimers or trimers in addition to cyclic tetramers.In the cases of the ditopic pyridyl ligands (4-NC5H4)2Z cyclic dimers were obtained with [Pt(en)(NO3)2] when Z = CH2CH2 C(OH)2 or 1,4-CH2C6F4CH2 [Fig. 3(a)] whereas when Z = C·C CHNCH or 1,4-C6H4 equilibrium mixtures of cyclic trimers and tetramers were obtained.9 However where Z = 4,4A-CH2C6F4C6F4-CH2 or 2,6-(CH2)2C10F6 a mixture of cyclic dimers and linear oligomers was formed. The outcome of such reactions may also be affected by the nature of the metal ion. Cyclic dimers are obtained from the reactions between 1,3- or 1,4-bis(benzimidazol-1-ylmethyl)benzene (1,3- or 1,4-bimb) and CoCl2·6H2O [Fig. 3(b)] [M(en)(NO3)2] (M = Pd or Pt) or However with NiCl [Cu(en)(H 2·6H2O and CuCl2·2H2O 1,3-bimb reacts to form insoluble complexes of metal:ligand stoichiometry 1 1 or 2 3.One way to favour cyclic dimer formation with square planar metal centres is to construct a ditopic ligand which incorporates a rigid 90° angle at its centre and contains two 4-pyridyl binding groups. Two such ligands could then complete a rectangle with two square planar metal centres without imposing significant ligand reorganisation energy requirements. Furthermore this ligand structure is likely to give optimum metal–ligand bonding in a cyclic dimer structure since this will not involve significant distortions from 90° for the angle subtended at the metal. A particular example is provided by the use of a rigid ditopic ligand in which two 4-pyridyl groups are arranged at 90° through being linked to an iodonium or Pt2+ ‘corner’ [Fig.4(a,b)]. Thus reaction of (4,4A-NC5H4C6H4)2I+ with [M(dppp)(OTf)2] affords a cyclic dimer which is structurally related to [Pd(dppp)(4-NC5H4)2]4 8+ but contains diagonally opposed iodonium centres in place of two metal ions.14 Extending this strategy to incorporate 120° angles in the ditopic ligand and 180° angles at the metal centres allows cyclic hexamers to be constructed. An example is provided by the reaction of (4-NC5H4)2CNO with [trans-{Pt(PPh3)2(C6H4- 4)2}(OTf)2] to give [trans-{Pt(PPh3)2(C6H4-4)2}2- {(4-NC5H4)2C = O}]6 12+ [Fig. 4(c)].17 Self-assembly reactions of this type have been exploited in the production of a wide range of other new complexes including bicyclic metallomacrocycles, 9 catenanes,9 nanometre sized metallomacrocycles [Fig.4(b,c)],14 chiral metallocyclophanes and metallomacrocycles peripherally functionalised by metallocene calix[4]arene or crown ether groups.18 Heteronuclear metallomacrocycles may also be prepared by self-assembly reactions involving the use of ditopic metallaligands such as cis- [Pt(dppp)(C6H4CN-4)2] and cis-[Pt(dppp)(C·CC5H4N-4)2] which produce tetranuclear cyclic dimers with {Pd(dppp)} centres [Fig. 4(a)]14 following a similar strategy to that of using (4,4A-NC5H4C6H4)2I+. Ditopic ligands containing the 4-pyridyl donor group have been particularly effective in producing cyclic dimers and tetramers but other nitrogen heterocycle based ligands can also be used.In one example a cyclopalladation reaction between 1,3-bis(1-methylbenzimidazol-2-yl)benzene (mbzimpH) and 291 Chemical Society Reviews 1998 volume 27 a) N M N M(OTf)2 N N N M N PPh2 PPh2 Pt Pt C N N P Ph P Ph2 2 or N N C N and M = M(dppp) cis-M(PEt3)2 M2+ = Pd2+ Pt2+ dppp = Ph2PCH2CH2CH2PPh2 NH OTf- = O3SCF3 - 2 Pt N N H2N N and M = Re(CO)3Cl N M M N N b) N N N M M M M N N N N N PPh2 M M(OTf)2 I+ N N N M M 2 P Ph Pt or N N N PPh3 PPh3 M M Pt+ Pt+ PPh3 PPh3 Fig. 4 (a) Cyclic tetramers constructed using ditopic ligands containing 90° angles at Pt2+ centres. (b) Nanometre sized cyclic tetramers constructed using ditopic ligands containing 90° angles at iodonium or Pt2+ centres.(c) Nanometre sized cyclic hexamers constructed using ditopic ligands containing 120° angles. 2 to form Pd(OAc)2 affords the cyclic trimer [Pd(mbzimp)(O2CMe)]3· 9MeCN [Fig. 5(a)]19 A cyclic trimer is also formed by the more flexible homoditopic ligand 1,2-bis(3A,5A-dimethylpyrazol-1- yl)ethane (dmpze) which reacts with PdCl [PdCl2(dmpze)]3 containing trans-Pd centres [Fig. 5(b)].20 In another example the short rigid bridging ligand imidazole (imdH) gives the cyclic trimer [Cu(tacn)(imd)]3 3+ (tacn = 1,4,7-triazacyclononane) which contains paramagnetic CuII centres [Fig. 5(c)].6 In a quite different way the self reaction of a peripherally hydroxy substituted porphyrin complex results in the formation of [Fe(ttpO)]3 (ttpOH = 2-hydroxytetra- 4A-tolylporphyrin) [Fig.5(d)].21 This heteroditopic binding motif results in the formation of the cyclic trimer which contrasts with the formation of a cyclic tetramer from 5,15-bis(4'-pyridyl)-2,8,12,18-tetrahexyl-3,7,13,17-tetramethylporphyrin which contains a linear array of 4-pyridyl groups.22 In these cases it is not obvious that the combinations Chemical Society Reviews 1998 volume 27 292 c) E TfO EOTf E *E TfO E*OTf E E* E* E = N E* = M E = M E* = N E E E* E* E* E E N E E E* E* E* E = N or M E* = M or N E* E E* E E E E* E* E* E* E E Key N N N N PPh3 PPh3 M M Pt+ Pt+ PPh3 PPh3 O N N N N O PPh3 PPh3 M M Pt+ Pt+ PPh3 PPh3 of metal centre and ditopic ligand should necessarily produce cyclic trimers rather than dimers or tetramers and the outcome of reactions involving flexible or heterotopic ligands having different binding sites is difficult to predict.The flexibility of ligands such as dmpze which contain two monodentate binding sites linked by saturated hydrocarbon chains offers much less control over the outcome of reactions with metal centres. Control is further reduced in reactions involving PdCl2 by the absence of a chelating co-ligand to direct the ditopic ligand to cis-coordination sites. However in the case of dmpze and PdCl2 binding of the ditopic ligand to the trans-sites on Pd2+ is advantageous in producing the cyclic trimer because of the conformation adopted by the dmpze ligand [Fig.5(b)]. In contrast the reaction of a similar ligand 1,4-bis(imidazol- 1-yl)butane (bbi) with Mn(BF4)2 affords a polymeric product [Mn(bbi)3](BF4)2. The structure of this compound contains rhombohedral arrays of eight octahedral Mn2+ ions at the b) a) Pd X Pd MeCN X Pd X Pd N N N = Pd Pd Pd mbzimpPd N 2 X = O2CMe d) c) 3+ N Fe N N Cu N N N N N N Cu Cu O N N N N N R = 4-MeC6H4 = 1,2-bis(3A,5A-dimethylpyrazol-1-yl)ethane]. (c) The structure of the cyclic trimer [Cu(tacn)(imd)] (d) The structural arrangement in [Fe(ttpO)]3 (ttpOH = 2-hydroxytetra-4A-tolylporphyrin). N corners of the rhombohedron linked by 12 bbi ligands which form the edges [Fig.2(b)].23 These rhombohedra are arranged in an interlocked structure in which two equivalent mutually interpenetrating three-dimensional networks are present. Ditopic ligands containing 2,2A-bpy binding groups provide a means of favouring macrocycle formation over polymerisation in that they reduce the number of ligands which can be bound to a particular metal centre and so may avoid the need for coligands to block some coordination sites.1 When suitable spacer groups are present in these ditopic ligands binuclear complexes may form which contain cavities capable of binding guest molecules [Fig. 6(a)].24 In some cases polytopic ligands containing several 2,2A-bipyridyl binding sites can form a helix in which metal atoms are embedded.One example is provided by a tritopic ligand containing three 2,2A-bipyridyl binding sites which reacts with Cu+ to form a trinuclear helicate complex [Fig. 2(c)].1,25 However flexible polytopic ligands with chelating termini can also form cyclic structures as found in pentanuclear [Fe5(tbp)5]10+ (tbp = 4,4A-bis[2-{4-(4A-methyl)- 2,2A-bipyridyl}ethyl]-2,2A-bipyridyl)26 [Fig. 7(a)] and octanuclear [Co8(bppz)12(ClO4)]3+ {bppz2 = bis[3-(2-pyridyl)- pyrazol-1-yl]dihydroborate}27 [Fig.7(b)]. The use of tritopic ligands in which three metal binding sites radiate from a central aryl ring offers another synthetic strategy for the construction of container molecules.1 Fujita et al. have shown9 that in the presence of p-methoxyphenyl acetate or xylene [Pd(en)2(NO3)2] and 1,3,5-tris(4A-pyridylmethyl)benzene (tpmb) react to give a bicyclic structure [{Pd(en)}3(tpmb)2]6+ [Fig.8(a)]. A more complex structure arises from the reaction of [Pd(en)2(NO3)2] with the rigid tritopic ligands 1,3,5-(4A-NC5H4Ar)3C3H3N3 (Ar = nothing 1,4-C6H4 4,4A-C6H4C6H4). In this case an octahedral array of six {Pd(en)}2+ groups is bound together by four tritopic ligands. The ligands are arranged in a tetrahedral array each binding to three Pd centres and occupying four triangular faces of the octahedron [Fig. 8(b)]. This structural arrangement is also found in the products of the reaction between 1,3,5-tris(4-pyridyl) ethynylbenzene and [M{(R)-(+)-binap}(OTf)2] [M = Pd Fig. 5 (a) The structure of [Pd(mbzimp)(O2CMe)]3.9MeCN [mbzimpH = (methylbenzimidazol-2-yl)benzene].(b) The structure of [PdCl2(dmpze)]3 [dmpze 3 3+ (tacn = 1,4,7-triazacyclononane imdH = imidazole). Pt; binap = 2,2A-bis(diphenylphosphino)-1,1A-binaphthyl].28 These latter structures contain chiral (R)-(+)-binap ligands and are members of the rare T symmetry group. The more flexible tritopic ligand 1,3,5-tris(pyrazol-1-ylmethyl)-2,4,6-triethylbenzene (tpteb) produces a complex of similar general structure and a ligand:metal ratio of 2:3 in its reaction with PdCl2.29 However in this case the six Pd centres have trans-geometries rather than the cis-geometry of the {Pd(en)}2+ moiety. The reaction proceeds to give the compound [(trans-PdCl2)6(tpteb)4]·2Me2SO·4H2O in 87% isolated yield. Thus the more flexible ligand is able to accommodate a transcoordination geometry at the metal whilst preserving the gross structural arrangement found for more rigid tritopic ligands combined with cis-coordinated metal centres.These findings further illustrate the extent to which bridging ligand structure may be exploited in the design and facile synthesis of new nanometre scale polynuclear structures with different metal centres. 2.2 Metal directed metallomacrocycle formation reactions The labile nature of the Pd2+ centre widely used in metallomacrocycle syntheses involving nitrogen heterocycle ligands leads to thermodynamically driven self-assembly reactions which often proceed in high yield. However it is also possible to construct metallo-macrocycles in reactions where the metal ligand bonding is more kinetically inert.In such cases it is the most rapidly formed reaction products that are isolated. These kinetic products may or may not be the most thermodynamically stable products and it is possible that they might subsequently convert to more thermodynamically stable forms where these exist. It is not always clear from literature reports whether metallomacrocycle formation reactions are self-assembly in the strict sense of being equilibrium driven or whether they are metal directed macrocycle formation reactions in which a kinetic product is being isolated. It is of course possible that the kinetic product is also the thermodynamic product. An early example of the metal directed construction of a metallomacrocycle host is provided by the work of Maverick Chemical Society Reviews 1998 volume 27 R N N N Cl N Cl Pd Pd Cl N Cl N N N N Cl N N N Pd Cl O Fe O O Fe O R N N Fe Fe R = R N N 293 4+ O a) O N N O O OMe H2O NZn N N OH2 N N ZnNN N OMe O O N N O O b) O O Cu O O N N O O Cu O MeO O - O MeO2C = MeO2C O - = Fe3+ MeO O 2C)C(O)]2C6H4}.c) 294 O Fig. 6 (a) A molecular host derived from a ligand containing 2,2'-bypyridyl termini. (b) A molecular host derived from a ligand containing b-diketonate termini. (c) The structure of the tricyclic metallomacrocycle [Fe4(dikp)6] {dikp = 1,4-[OC(MeO)C(MeO et al.who in 1984 first reported the synthesis of a binuclear complex of Cu2+ with a ditopic b-diketonate ligand [Fig. 6(b)].5 Tricyclic structures may also be obtained using ligands of this type and Saalfrank and co-workers have reported30 tetranuclear complexes of formula [Fe4(dikp)6] {dikp = 1,4-[OC(MeO)- C(MeO2C)C(O)]2C6H4} which contain six bridging ligands [Fig. 6(c)]. Non-chelating alkyl or aryloxide ligands have also been found to form metallomacrocycles. Stephan has shown31 that the reactions between [Zr(h5-C5H5)2Me2] and dihydroxy compounds such as 1,3-(HOCH2)2C6H4 afford macrocyclic products. The complex [Zr(h5-C5H5)2{1,3-(OCH2)2C6H4)2}] [Fig. 3(c)] constitutes an example of a metallocyclophane constructed from a ditopic ligand containing two monodentate alkoxide binding sites.Normally such linkages might be expected to be rather reactive and poorly suited to the purpose of metallomacrocycle synthesis. However the presence of the d0 Zr4+ centre allows the Zr–O bond to be strengthened by p-donation from filled oxygen p-orbitals to empty zirconium d-orbitals.31 Similar effects have been exploited in our own work on the construction of metallocyclophanes using octahedral {M(A)(Tp*)}2+ [Tp*2 = hydrotris(3,5-dimethylpyrazol- 1-yl)borate; A = NO; M = Mo W; A = O M = Mo] centres in combination with ditopic proligands carrying hydroxy thiol or amino groups [Fig. 3d)].7,10,32 In complexes of the formula [M(NO)(Tp*)(ER)2] (M = Mo W; E = O S NH; R = hydrocarbyl) the nitrosyl ligand is essentially linear and may be considered to act as a two-electron s-donor NO+ ligand.This requires a formal oxidation state of (ii) for the metal centre in [M(NO)(Tp*)(OR)2] (M = Mo W) so that the complexes are coordinatively unsaturated having a formally 16-electron d4 Chemical Society Reviews 1998 volume 27 a) 9+ Fe Fe Represents Cl Fe Fe N N N N Fe N N b) 3+ Co Co Co Co ClO4 Co Co Co Co Represents N N H2B - N N N The MoV complex [Mo(O)(Tp*)Cl N Fig. 7 (a) The pentanuclear metallomacrocycle [Fe5(tbp)5(Cl)]9+ (tbp = 3,3A-bis-3,3B-[2-{3B-(3A-methyl-2,2A-bipyridyl)ethyl}2,2'-bipyridyl]) containing an included chloride ion. (b) The octanuclear metallomacrocycle [Co8(bppz)12(ClO4)]3+ {bppz2 = bis[3-(2-pyridyl)pyrazol-1-yl]dihydroborate} containing an included perchlorate ion.metal centre. The nitrosyl ligand is a strong p-acceptor and can p-back bond to the filled metal dxz and dyz orbitals (Fig. 9) leaving dxy empty and free to act as a p-acceptor orbital towards filled p-orbitals on the donor atoms E. Thus the electron deficiency at the metal is relieved by Epp to Modp charge donation. This strengthens the M–E bond giving rise to the unusual stability of complexes containing monodentate alkoxide thiolate or organoamide ligands. The structurally similar MoV complex [Mo(NO)(Tp*)Cl2] shows a similar derivative chemistry to [Mo(NO)(Tp*)Cl2]. However in this case the formally d1MoV centre is bonded to a p-donor oxo-ligand so that the metal dxz and dyz orbitals interact with filled oxygen p-orbitals to give filled p(dxz dyz) levels leaving one electron to occupy the empty dxy-orbital (Fig.9).33 2] differs in reactivity 2] and its use in the syntheses of from [Mo(NO)(Tp*)I metallocyclophanes from ditopic hydroxy ligands is more problematic.32 Attempts to develop similar metallomacrocycle syntheses using [Re(O)(Tp*)Cl2] and dithiol ligands have met with no success and in this respect the chemistry of the ReV systems is quite different from that of the {Mo(O)}3+ and {Mo(NO)}3+ systems. In view of the general structural and chemical similarities between the ReV and MoV systems it is a) 3 2 + H N 2 X ONO2 = X Pd 2 b) 4 6 + ONO2 2 X N H = X NH and = Pd 2 N Ar NH N Ar = nothing xy 1 and dxy 2.Fig. 8 (a) The structure of the bicyclic metallomacrocycle [{Pd(en)}3(tpmb)2]6+ [tpmb = 1,3,5-tri(4'-pyridylmethyl)benzene]. (b) The structures of some hexanuclear metallomacrocycles. possible that this difference in behaviour arises from the different electron configurations of the metal centres. The ReV centre in [Re(O)(Tp*)X2] (X = Cl SR; R = alkyl or aryl) is formally d2 so that the dxy orbital is fully occupied and unable to enter into p-bonding interactions with filled p-orbitals on the donor atoms of the ligands X (Fig. 9). This is likely to labilise the Re–X bond compared to the Mo–X bonds in [Mo(O)(Tp*)X2] or [Mo(NO)(Tp*)X2] changing the balance between kinetic and thermodynamic control over the formation of reaction products.Furthermore the absence of p-bonding interactions will allow more facile rotation about the Re–X bond so that the group R may explore a larger volume of space. This may reduce the prospect of forming cyclic structures where X is a bifunctional potentially bridging ligand. In qualitative terms the increasing reluctance of {Mo(NO)(Tp*)}2+ {Mo(O)(Tp*)}2+ and {Re(O)(Tp*)}2+ to form metallomacrocycles follows their respective dxy orbital occupancies of dxy 0 d The formation of metallomacrocycles from complexes of the formula [M(A)(Tp*)X2] (A = NO M = Mo W X = I; A = O M = Mo X = Cl OMe) introduces some complications when compared to reactions involving kinetically labile square planar metal centres.Firstly the link between the metal and the ditopic ligand is no longer a direct linear M–N(pyridyl) linkage. Instead the M–E–R unit is present which introduces an angle at the donor atom E. This means that the structure is inherently less rigid and less readily controlled through suitable choice of the ditopic ligand structure. Secondly the incorporation of octahedral {M(A)(Tp*)} metal centres into the macrocycles introduces configurational isomerism into the structures. A cyclic dimer may exist as syn- or anti-isomers depending upon the 6+ (X-)6 X X N N = N 12+ (X-)12 X X 2 ONO2 OTf P Ph P Pd Ph2 OTf N and N Ar or ONO2 N N Ar or N N N a) z Ap A Ap A dyz dxz Mo Mo x y {M=O}1 {M-NO}4 b) NOp* {dxz d yz Op}* dxy Op {dxz d yz NOp}* dxy dxy dxz d yz Op {dxz d yz Op} c) A {dxz d yz NOp} dxy Mo E E Fig.9 (a) Schematic view of p-bonding between the ligand A and Mo in a complex of formula [Mo(A)(Tp*)(ER)2]. (b) Qualitative fragment molecular orbital diagram to show the effect of M–A p-bonding on the metal p-orbitals (t2g in strictly octahedral symmetry) of [M(A)(Tp*)(ER)2] (M = Mo Re). (c) Schematic view of p-bonding between the ligand ER and Mo in a complex of formula [Mo(A)(Tp*)(ER)2]. relative orientations of the ligands A in the {M(A)(Tp*)} units. In the case of a cyclic trimer syn,syn- and anti,syn-isomers are possible. This situation is exacerbated by a third difference; the chemistry involved is different from that associated with Pd2+/ pyridyl ligand complexation and involves more kinetically inert metal centres.This can lead to the formation of a number of products not all of them macrocyclic which have to be separated. A further effect of having non-labile metal centres is that the isomer distributions obtained will reflect the kinetics of macrocycle formation and will probably not represent a thermodynamically determined equilibrium distribution. An unexpected macrocycle formation process occurs when thiol ligands are used in reactions with [Mo(NO)(Tp*)I2]. Ligand oligomers form through disulfide bond formation and these species are isolated in the form of chelating ligands in mononuclear metallocyclophanes (Fig.10). Since it is known that [Mo(NO)(Tp*)I2] can act as an oxidant it is possible that this reactant oxidises some of the thiols to disulfides which then react to form chelate complexes.34 Because of the presence of different isomers of the metallomacrocyclic products of these reactions it is possible to determine whether under a given set of conditions a reaction is stereoselective towards one isomer. It turns out that the isomer distribution obtained can vary with ligand structure donor atom type and the nature of the metal centre involved. The reaction between [Mo(NO)(Tp*)I2] and 2,7-(HO)2C10H6 is essentially non-selective and ultimately produces similar proportions of syn- and anti-isomers although the syn-isomer forms more rapidly than the anti-isomer.The reaction betwen [Mo(NO)- (Tp*)I2] and 1,3-(HO)2C6H4 also produces both syn- and antiisomers of [Mo(NO)(Tp*)(1,3-O2C6H4 )]2 but the reaction involving 1,3-(HS)2C6H4 produces only anti- [Mo(NO)(Tp*)(1,3-S2C6H4 )]2 and the reaction between Chemical Society Reviews 1998 volume 27 295 a) N HB N N N N N Mo S S N S S O b) N HB S N N N S N N Mo S N S S S O Fig. 10 The structures of (a) [Mo(NO)(Tp*){1,4-(SCH2)2C6H4}2] and (b) [Mo(NO)(Tp*){1,4-(SCH2)2C6H4}3] [Mo(O)(Tp*)Cl2] and 1,3-(HO)2C6H4 produces only syn- [Mo(O)(Tp*)(1,3-O2C6H4 )]2. 2.3 Macrocyclic ligand directed metallomacrocycle formation reactions An alternative approach to the synthesis of metallocyclophanes is to add metal centres to a preformed macrocyclic cyclophane template.Puddephatt and co-workers have used a phosphite substituted calix[4]resorcinarene in this way to assemble four copper centres into a macrocyclic array [Fig. 11(a)].35 The aryl rings of cyclophanes may themselves be used to form p-complexes with metal ions. In this way cationic hosts such as [{Ru(h6-p-MeC6H4CHMe2)}4(calH2)]6+ (calH4 = calix[4]arene); [{Ir(h5-C5Me5)}2(bcalH3]3+ (bcalH4 = tetra-p-tert-butylcalix[ 4]arene) and [{Ru(h6-p-MeC6H4CHMe2)}2(h6 h6- ctv)]4+ (ctv = cyclotriveratrylene) have been synthesised (Fig. a) O O PhP PPh Cl Cu Cu O O R R Cl I Cl R R O O Cu Cu Cl PhP PPh O O [{Cu(m2-Cl)}4(m4-I)(res(PPh)4)] O O 4+ O Ru Ru O O Fig.11 The structures of some metallomacrocycles derived from preformed cyclophanes (a) [{Cu(m-Cl)}4(m4-I){(PhP)4res}]; (b) [(ML2)4(res)] (resH8 = tetrahexylcalix[4]resorcinarene); (c) [{Ru(h6-p-MeC6H4CHMe2)}2(cvt)]4+ (cvt = cyclotriveratrylene); (d) [{Ru(h6-p-MeC6H4CHMe2)}4(calH2)]6+ (calH4 Ru c) = calix[4]arene) O Chemical Society Reviews 1998 volume 27 296 11).36 The direct metallation of the hydroxy face of tetrahexylcalix[ 4]resorcinarene (resH8) is also possible [Fig.11(b)] and the reaction with [Zr(h5-C5H5)2Me2] leads to the elimination of CH4 and the formation of [{Zr(h5-C5H5)2}4(res)].37 A similar reaction with [Fe(phen)(mes)2] (phen = ortho-phenanthroline and mesH = mesitylene) affords the paramagnetic complex [{Fe(phen)}4(res)].In the presence of NEt3 the octahedral complex [Mo(NO)(Tp*)I2] also reacts with resH8 to form metallated species but these are insufficiently stable to allow separation by column chromatography. However if the steric bulk of the complex is reduced by removing the pyrazolylmethyl substituents and [Mo(NO)(Tp)I2] [Tp2 = hydrotris(pyrazol-1-yl)borate] is used then the metallated derivatives [{Mo(NO)(Tp)}n(resH(822n)] (n = 2 3 4) may be isolated [Fig. 11(b)].38 3 Inclusion properties of metallocyclophanes Having considered the synthetic approaches which may be used to construct metallocyclophanes it is important to consider their inclusion properties. Probably the earliest report of such behaviour is due to Maverick et al. who exploited the Lewis acid properties of the Cu2+ centres in a b-diketonate coordinated metallomacrocycle [Fig.6(b)] to bind nitrogen heterocycles. The compound containing a 2,7-naphthyl spacer between the copper bis-diketonate moieties was found to bind diaza[2,2]bicyclooctane selectively (K = 220 dm3 mol21) compared to other bases such as pyridine (K = 5 dm3 mol21) or pyrazine (K = 7 dm3 mol21).5 Elegant examples of the use of zinc as a Lewis acid centre to promote the binding of guest molecules within macrocyclic structures are provided by the work of Sanders and co-workers.39 Thus a cyclic trimer constructed with zinc porphyrin containing edges can bind complementary guest ligands such as s-tris(4-pyridyl)triazine very strongly (K > 109 dm3 mol21) (Fig.12). Although the zinc centres in these systems are not involved in maintaining the macrocyclic structure intact the yield of the macrocycle formation reaction can be improved if an s-tris(4-pyridyl)triazine ‘template’ is O ML b) O 2 L2M O O R R R R O O O O ML2 ML2 [{ML2}4(res)] R = n-C6H13 ML2 ={ Mo(NO)(Tp)} {Zr(h5-C5H5)2} {Fe(phen)} d) 6+ Ru O H Ru Ru O O H O added to orient the zinc porphyrin units during cyclisation of the acyclic trinuclear precursor. Zn N N N N N N Zn Zn 2 R R Me Me N N Zn Zn = N N Me R = CH2CH2CO2Me Me R 10H7 3)2] R Fig. 12 A trinuclear metallomacrocycle containing porphyrin bound Zn2+ ions acting as Lewis acid centres binding to a tritopic guest The hydrophobic cavity in metallocyclophanes can in suitable cases act as a host for hydrophobic guests.The binuclear complex [Pd(en){1,4-(4A-NC5H4CH2)2C6F4}]2 4+ 4 8+ [Fig. 1(b)] show shape selective molecular recognition [Fig. 3(a)] and the tetranuclear complex [Pd(en)(4,4Abpy)] of electron rich aromatic compounds.9 In the case of the cyclic dimer 1,4-(MeO)2C6H4 is bound more strongly (K = 2680 dm3 mol21) than 1,3-(MeO)2C6H4 (K = 1560 dm3 mol21) which in turn is bound more strongly than 1,2-(MeO)2C6H4 (K = 1300 dm3 mol21). The larger guest 2-(MeCONH)C10H7 is only weakly bound (K = 15 dm3 mol21). In the case of the cyclic tetramer the largest of these guests is the most strongly bound and binding constants decrease in the order 2-(MeCONH)C 1,3-(MeO)2C6H4 1,4-(MeO)2C6H4 1,2-(MeO)2C6H4 (respective K values 1800 580 330 30 dm3 mol21).Structural evidence has also been obtained for the incluson of 1,2- or 1,4-(MeO)2C6H4 into the electron deficient cavities formed by two pyromellitimide groups linked by {Zn(2,2A-bpy)2} moieties [Fig. 6(a)].24 A particular kind of molecular self-recognition is found in the formation of a [2]catenane {[Pd(en){1,4- (4A-NC5H4CH2)2-C6H4}]2}2 8+ from the binuclear complex [Pd(en){1,4-(4A-NC5H4CH2)2C6H4}]2 4+.9 In some cases molecules which become guests in metallomacrocycle structures may act as templates for the formation of the metallocyclophane. A particular example is provided by the formation of a bicyclic trinuclear host from [Pd(en)(NO and 1,3,5-(4-NC5H4CH2)C6H3 [Fig.8(a)]. The bicyclic structure only forms in the presence of specific templates such as 4-MeOC6H4CH2CO2Na which contain hydrophobic aryl components of suitable size to fit the cavity of the bicyclic metallocyclophane produced.9 Larger hexanuclear hosts may also be stabilised by the presence of guest molecules and although [{Pd(en)}6{1,3,5-(4-NC5H4CH2)C6N3}4]12+ [Fig. 8(b)] may be crystallised in the absence of guest molecules only the inclusion derivative containing four 1-adamantane carboxylate guests was sufficiently stable for X-ray crystallographic study.9 Solvent inclusion provides another example of ‘host–guest’ chemistry and several types of metallomacrocycle structure have been shown to form ‘host–guest’ complexes with solvent 3H3N2)3C6Et3 2O included in the cyclophane cavity.36 More specific 4(res)] [Fig.11(b)] reveals a _- molecules. The trinuclear compound derived from cyclopalladation of 1,3-bis(1-methylbenzimidazol-2-yl)benzene includes a molecule of MeCN in the central cavity [Fig. 5(a)].19 A hexanuclear compound derived from 1,3,5-(1A-C and PdCl2 similarly includes a molecule of Me2SO within the octahedral array of metal centres.29 In another example the structure of the cationic metallocyclophane [{Ir(h5- Et C5Me5)}2(bcalH3)](BF4)3·Et2O·MeNO2 contains a molecule of examples of solvent interactions with metallocyclophane hosts result from the H-bond acceptor properties of the nitric oxide ligand in complexes containing the {Mo(NO)Tp*} moiety.10,38 In particular the crystal structure of the tetrametallated calix[ 4]resorcinarene [{Mo(NO)Tp*} molecule of dichloromethane included within the cavity formed by the Tp* ligands and hydrogen bonded to one nitric oxide ligand.38 Although the H···O distance of 2.47 Å is greater than is normally associated with C–H···O hydrogen bonds in organic compounds it is less than the van der Waals sum for H and O of 2.6 Å and significantly shorter than the distance of ca.2.6 Å typically found for C–H hydrogen bonding to the oxygen atoms of terminally coordinated carbon monoxide ligands. Anion inclusion has also been observed in metallomacrocycle structures and the centre of the pentanuclear ring in [Fe5(tbp)5Cl]9+ is occupied by a chloride ion [Fig.7(a)]26 and of the octanuclear ring in [Co8(bppz)12(ClO4)](ClO4)3 by a perchlorate ion [Fig. 7(b)].27 The complicated polymeric structure of [Mn(bbi)3](BF4)2 includes two BF42 ions within each of the rhombohedral arrays from which the lattice is constructed [Fig. 2(b)].23 The incorporation of metal cations into preformed cyclophanes offers a more structured approach to the design of anion binding agents and the size of the central cavity formed by the four Cu2+ ions in [{Cu(m Cl)}4(X){(PhP)4res}] (X = m3-Cl m4-Br m4-I) [Fig. 11(a)] is of the correct size for symmetric binding to I2 ions but smaller Cl2 ions can only bind closely to three of the four Cu2+ ions asymmetrically.35 The addition of metal ions to the outer surface of cyclophanes through metallocene formation provides an alternative means of creating a host for anions as exemplified by the crystal structure of [{Ru(h6-p-MeC6H4CHMe2)}4- (calH2)](BF4)6.[Fig. 11(d)] in which one BF42 ion is included within the calixarene cavity.36 The structure of the trinuclear complex [{Ru(h6-p-MeC6H4CHMe2)}2(cvt)](OTf)4 [Fig. 11(c)] similarly includes a triflate ion in the cyclophane cavity and this structure shows selectivity for large tetrahedral anions such as [ReO4]2.36 4 Electrochemical magnetic and luminescence properties The introduction of transition metal centres into cyclophane structures can introduce new functionality to the macrocycle. Examples have already been described in which the cationic nature or Lewis acid properties of transition metal centres have been exploited to induce the binding of anions or Lewis base guests within the cyclophane cavity.Further examples of extended functionality are provided by cyclophanes containing metal centres with photoluminescence or photoelectron transfer properties metal centred redox properties and magnetic properties. An example of a photoluminescsent metallocyclophane was reported by Hupp et al.8 who synthesised a tetranuclear metallomacrocycle [{Pd(en)}(4,4A-bpy){Re(CO)3Cl}]2 4+ [Fig. 1(b)] which contains redox active rhenium and cationic palladium centres. The photoluminescence of the compound arises from metal to ligand charge transfer (MLCT) within the fac-{Re(CO)3Cl(4,4A-bpy)2} chromophore. The presence of the Pd2+ centre lowers the energy of the MLCT excited state by stabilising the p*-acceptor orbital of the 4,4A-bpy ligand.The addition of (NEt4)ClO4 increases the emission intensity 297 Chemical Society Reviews 1998 volume 27 presumably through ClO42 inclusion in the molecular cavity. This demonstrates the possible exploitation of metallocyclophanes containing photoactive centres in developing molecular sensing devices. Complexes containing the the {M(A)(Tp*)} (A = NO M = Mo W; A = O M = Mo) centres are known to be redox active and metallocyclophanes containing these groups are no exception. These compounds exhibit a variety of electrochemical behaviour depending on the nature of the ditopic ligand present the nature of the metal centre and the nuclearity of the metallocyclophane.Each metal centre should exhibit a one-electron reduction and in cases where the ditopic ligand is sufficiently long to support little or no interaction between the metal centres these processes will occur at similar potentials giving rise to only one reduction wave in the cyclic voltammogram of the complex. However when the ditopic ligand can support substantial interactions between the metal centres more than one reduction process will be observed [Fig. 13(a)]. In a cyclic dimer addition of the first electron will result in the reduction of one metal centre. The reduction potential of the second metal centre will then be shifted to more negative potential by the presence of the nearby reduced metal centre giving rise to two distinct reduction processes.In the case of a cyclic trimer after the first reduction the remaining two unreduced centres will be reduced at more negative potentials because of the effect of the single reduced metal centre. After addition of the second electron the remaining unreduced metal centre will be affected by two reduced centres shifting its reduction potential to an even more negative value. Thus three separate one-electron reductions would be expected in this case. A cyclic tetramer would be expected to show different behaviour. Reduction of the first metal centre will shift the reduction potentials of the adjacent metal centres to more negative values. However the fourth diagonally opposed metal centre is further away and may be little influenced by reduction of the first metal centre.In this case two electrons might enter the molecule at similar potentials. Following the same logic the two remaining unreduced metal centres might be expected to reduce at similar potentials but at more negative values than for the first two metal centres. These effects are observed in the electrochemistry of cyclic oligomers containing {M(A)(Tp*)} centres and bridging ligands which support some interaction between the redox sites [Fig. 13(b)].7,10 Redox active metallocyclophanes demonstrate important functionality in that host molecules which can undergo multiple reduction processes offer a means of storing electrons. In principle a metallocyclophane host containing four reduced metal centres might be capable of delivering four electrons into a reducible bound guest molecule.Furthermore reduction of a metallocyclophane leads to changes in the magnetic and optical properties of the molecule offering opportunities to electrochemically alter the reactivity magnetism or optical properties of and inclusion compound involving such a host. The use of paramagnetic metal centres in metallocyclophane construction not only provides access to paramagnetic potential host molecules but also allows intramolecular magnetic interactions to be studied within large molecular arrays. In cyclic trimers the problem of spin frustration arises. This is because two out of the three magnetic centres may couple antiferromagnetically but if the third couples antiferromagnetically to one it must be ferromagnetically coupled to the other.Three interacting S = 12 centres in an equilateral triangular arrangement give rise to a degenerate doublet of doublets. The degeneracy may be removed by low symmetry components of the exchange field and by intermolecular exchange interactions. Such effects have been successfully modelled for [Cu(tacn)- (imd)]3 3+ and the EPR spectrum simulated over the temperature range 4.2 to 300 K based on spin frustration low symmetry and intercluster exchange interactions.6 In the corresponding cyclic tetramer [Cu(tacn)(imd)]4 4+ antiferromagnetic coupling between adjacent copper centres might be expected to result in a Chemical Society Reviews 1998 volume 27 298 Fig.13 Examples of the electrochemistry of metallomacrocycles containing {Mo(NO)(Tp*)} centres (a) patterns of electron transfer into (i) binuclear (ii) trinuclear; and (iii) tetranuclear complexes. (b) Current�]potential plots (cyclic voltammograms) for the cyclic oligomers (i) [Mo(NO)(Tp*)(2,7- O2C10H6)]2; (ii) syn syn-[Mo(NO)(Tp*)(1,4-O2C6H4)]3; (iii) anti,syn,syn- [Mo(NO)(Tp*)(1,3-O2C6H4)]4. singt ground state and magnetic susceptibility measurements show this to be the case.40 5 Conclusion The chemistry of metallocyclophanes is currently much less well developed than that of their purely organic counterparts. However the results described here demonstrate the rich and varied possibilities for creating new molecular architectures constructed in facile self-assembly processes or by stereoselective kinetically controlled reactions.The products of such syntheses offer access to potential host molecules with electron transfer and magnetic or optical properties. Examples of such features have already been provided by the work carried out on metallocyclophanes so far providing prototypes for the design and synthesis of more elaborate functional molecular structures with exploitable properties. 6 Acknowledgements Support from the EPSRC and the Leverhulme Trust for work in Birmingham is gratefully acknowledged. 7 References 1 J.-M. Lehn Supramolecular Chemistry VCH Verlagsgesellschaft Weinheim 1995. 2 D. J. Cram and J. M. Cram Container Molecules and Their Guests Royal Society of Chemistry Monographs in Supramolecular Chemistry ed.J. F. Stoddart Royal Society of Chemistry Cambridge 1994. 3 C. D. Gutsche Calixarenes Royal Society of Chemistry Monographs in Supramolecular Chemistry ed. J. F. Stoddart Royal Society of Chemistry Cambridge 1989. 4 F. N. Diederich Cyclophanes Royal Society of Chemistry Monographs in Supramolecular Chemistry ed. J. F. Stoddart Royal Society of Chemistry Cambridge 1991. 5 A. W. Maverick S. C. Buckingham Q. Yao J. R. Bradbury and G. G. Stanley J. Am. Chem. Soc. 1986 108 7430 and references therein. 6 J. Padilla D. Gatteschi and P. Chaudhuri Inorg. Chim. Acta 1997 260 217. 7 F. S. McQuillan H. Chen T. A. Hamor and C. J. Jones Polyhedron 1996 15 3909. 8 R. V. Slone D. I. Yoon R.M. Calhoun and J. T. Hupp J. Am. Chem. Soc. 1995 117 11813. 9 M. Fujita and K. Ogura Bull. Chem. Soc. Jpn. 1996 69 1471. 10 F. S. McQuillan H. Chen T. A. Hamor and C. J. Jones Inorg. Chem. 1997 36 4458 and references therein. 11 P. M. Stricklen E. J. Volcko and J. G. Verkade J. Am. Chem. Soc. 1983 105 2494. 12 J. Lu T. Paliwala S. C. Lim C. Yu T. Niu and A. J. Jacobson Inorg. Chem. 1997 36 923 and references therein. 13 A. J. Blake N. R. Champness A. K. Khlobystov D. A. Lemenovskii W.-S. Li and M. S. Schröder Chem. Commun. 1997 1339 and 14 P. J. Stang and B. Olenyuk Acc. Chem. Res. 1997 30 502 and references therein. references therein. 15 X. Chi A. J. Guerin R. A. Haycock C. A. Hunter and L. D. Sarson J. Chem. Soc. Chem. Commun. 1995 2563.16 S. K. Chawla and B. K. Gill Polyhedron 1997 16 1315. 17 P. J. Stang N. E. Persky and J. Manna J. Am. Chem. Soc. 1997 119 4777 and references therein. 18 P. J. Stang D. H. Cao K. Chen G. M. Gray D. C. Muddiman and R. D. Smith J. Am. Chem. Soc. 1997 119 5163 and references therein. 19 S. Rüttimann G. Bernadinelli and A. F. Williams Angew. Chem. Int. Ed. Engl. 1993 32 392. 20 A. T. Baker J. K. Crass M. Maniska and D. C. Craig Inorg. Chim. Acta 1995 230 225. 21 J. Wojacczynski L. Latos-Grazynski M. M. Olmstead and A. L. Balch Inorg. Chem. 1997 36 4548. 22 P. J. Stang J. Fan and B. Olenyuk Chem. Commun. 1997 1453. 23 P. C. M. Duncan D. M. L. Goodgame S. Menzer and D. J. Williams Chem. Commun. 1996 2127. 24 M. A. Houghton A. Bilyk M. M. Harding P. Turner and T. W. Hambley J. Chem. Soc. Dalton Trans. 1997 2725 and references therein. 25 C. Piguet G. Bernardinelli and G. Hopfgartner Chem. Rev. 1997 97 2005. 26 B. Hasenkopf J. M. Lehn B. O. Kneisel G. Baum and D. Fenske Angew. Chem. Int. Ed. Engl. 1996 35 1838. 27 P. L. Jones K. J. Byrom J. C. Jeffery J. A. McCleverty and M. D. Ward Chem. Commun. 1997 1361. 28 P. J. Stang B. Olenyuk D. C. Muddiman and R. D. Smith Organometallics 1997 16 3094. 29 C. M. Hartshorn and P. J. Steel Chem. Commun. 1997 541. 30 R. W. Saalfrank R. Burak A. Breit D. Stalke R. Herbst-Irmer J. Daub M. Porsch E. Bill M. Müther and A. Trautwein Angew. Chem. Int. Ed. Engl. 1994 33 1621. 31 D. W. Stephan Organometallics 1990 9 2718. 32 T. E. Berridge and C. J. Jones Polyhedron 1997 16 3695. 33 B. L. Westcott and J. H. Enemark Inorg. Chem. 1997 36 5404. 34 H. A. Hinton H. Chen T. A. Hamor C. J. Jones F. S. McQuillan and M. S. Tolley Inorg. Chem. 1998 in the press. 35 W. Xu J. J Vittal and R. J Puddephatt J. Am. Chem. Soc. 1995 117 8362. 36 J. L. Atwood K. T. Holman and J. W. Steed Chem. Commun. 1996 1401 and references therein. 37 E. Solari W. Lesueur A. Klose K. Schenk C. Floriani A. Chiesi-Villa and C. Rizzoli Chem. Commun. 1996 807. 38 T. E. Berridge and C. J. Jones Polyhedron 1997 16 2329. 39 A. Vidal-Ferran N. Bampos and J. K. M. Sanders Inorg. Chem. 1997 36 6117. 40 J. Padilla D. Gatteschi and P. Chaudhuri I. Karpenstein M. Winter M. Lengen C. Butzlaff E. Bill A. X. Trautwein U. Fl�orke and H.-J. Haupt Inorg. Chem. 1993 32 888. Received 14th January 1998 Accepted 5th March 1998 299 Chemical Society Reviews 1998 volume

ISSN:0306-0012    

DOI:10.1039/a827289z

出版商:RSC    

年代:1998

数据来源: RSC