摘要:

Volume 26 CHEMICAL SOCIETY Pages 239-326 August 1997 ISSN 0306-0012REVIEWS Issue 4 CSRVBR 26(4) 239-3 16 Surface scientific aspects in semiconductor electrochemistry H. J. Lewerenz 239-246 Conjugated polymers incorporating pendant functional groups-synthesis and characterisation Simon Higgins 247-258 1 positive eledrodr negative eleclrude 1 Rechargeable lithium batteries John R. Owen 259-268 Oxaziridine rearrangements in asymmetric synthesis Jeffrey Aube 269-278 MELDOLA LECTURE: understanding the properties of urea and thiourea compounds Kenneth D. M. Harris 279-290 e2Snz+ bacterial methylalion snl redistribution Speciation of trace metals in the environment Steve J. Hi1 29 1-298 Biota lment Developing the physical organic chemistry of Fischer carbene complexes Claude F.Bernasconi 299-308 The synthesis of molecular sieves from non-aqueous 309-3 18 solvents Russell E. Morris and Scott J. Weigel CCD cw DFWM DIAL FT-ICR FTIR ICP-AES ICP-MS charge c continuo degener different Fourier resonan Fourier inductiv atomic e 7induct i v Laser techniques for chemical analysis Richard Snook 319-326 Articles that will appear in forthcoming issues Pentafluorophenylboranes: from obscurity to applications Warren E. Piers and Tristram Chivers Selection approaches to catalytic systems Paul A. Brady and Jeremy K. M. Sanders Molecular modelling of electron transfer systems by noncovalently linked porphyrin-acceptor pairing Takashi Hayashi and Hisanobu Ogoshi Hydrogen isotope exchange reactions involving C-H (D,T) bonds Thomas Junk Asymmetric synthesis of building-blocks for peptides and peptidomimetics by means of the @-lactam synthon method Iwao Ojima and Francette Delaloge Approaches to the synthesis of ingenol Sanghee Kim and Jeffrey Winkler Photo-induced electron and energy transfer in non-covalently bonded supramolecular assemblies Michael D.Ward The mechanistic and evolutionary basis of stereospecificity for hydrogen transfers in enzyme-catalyzed processes Kevin A. Reynolds and Koren Holland Sandwich-type heteroleptic phthalocyaninato and porphyrinato metal complexes Dennis K. P. Ng and Jianzhuang Jiang Glycosylation employing bio-systems Vladimir Kren and Joachim Thiem Preparation of seven and larger membered heterocycles by electrophilic heteroatom cyclization Gerard Rousseau and Fadsi Homsi Molecular and chemical basis of prion-related diseases Sheila B.L. Ng and Andrew Doig Ultrasound in synthetic organic chemistry Timothy J. Mason The biomedical chemistry of technetium and rhenium Jonathan R. Dilworth and Suzanne J. Parrott Enzymes in organic synthesis: recent developments in aldol reactions and glycosylations Shuichi Takayama, Glenn J. McGarvey and Chi-Huey Wong Polymer-supported organic reactions: what takes place in the beads? Philip Hodge Equilibrium, frozen, excess and volumetric properties of dilute solutions Michael J. Blandamer New synthetic methods via radical cation fragmentation Mariella Mella, Maurizio Fagnoni, Mauro Freccero, Elisa Fasani, and Angelo Albini Asymmetric synthesis of amino acids using sulfinimines (thiooxime S-oxides) Franklin A. Davis, Ping Zhou and Bang-Chi Chen Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS99726FP011

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Chemical Society Reviews Editorial Board Jean-Pierre Sauvage (CNRS, Strasbourg) [Chair] Vicenzo Balzani (Bologna) Ed C. Constable (Basel) Chris Elschenbroich (Marburg) Tim C. Gallagher (Bristol) Editorial Office Martin Sugden (Managing Editor) David Bradley; Peter Whittington (Production) Debbie Halls (Editorial Secretary) http://chemistry .rsc.org/rsc tel: +44 (0)1223 420066 Chemical Society Reviews publishes concise, succinct and lightly referenced articles that provide an introductory overview to topics of current interest in chemistry. The articles appeal to the general research chemist as well as to the expert in the field and provide an essential starting point for further reading. Advanced undergraduates, postgraduates and experienced re- searchers should all benefit from reading Chemical Society Reviews. Chemical Society Reviews (ISSN 0306-0012) is published bimonthly by the Royal Society of Chemistry, Thomas Graham House, Science Park, Cambridge, UK CB4 4WF.1997 subscription rate: & 130 (USA $234). Customers in Canada will be charged the sterling price plus a surcharge to cover GST. Individuals can subscribe for &45 (USA $80) providing their institutional library takes a full price subscription. All orders accompanied by payment should be sent directly to The Royal Society of Chemistry, Turpin Distribution Services Ltd, Blackhorse Road, Letchworth, UK SG6 IHN. (NB Turpin Distribution Services Ltd., distributors, is wholly owned by the Royal Society of Chemistry.) Payment should be by cheque in pounds sterling payable on a UK clearing bank or in US dollars payable on a US clearing bank.Second class postage is paid at Zdenek Herman (Prague) Horst Kunz (Mainz) John P. Maier (Basel) D. Mike P. Mingos (Imperial) Jeremy K. M. Sanders (Cambridge) Royal Society of Chemistry Thomas Graham House Science Park Cambridge UK CB4 4WF csr@rsc.org fax: +44(0)1223 420247 The Editorial Board commissions articles that encourage international, interdisciplinary dialogues in chemical research. The Board welcomes any suggestions for new articles. A guide for authors and synopsis form can be found in the first issue of this year’s volume or on the RSC’s World-Wide Web home page (URL above).Alternatively, they can be requested from the Managing Editor, in paper or electronic form (postal and e- mail address above). Jamaica NY 1141-9998. Airfreight and mailing in the USA by Publications Expediting Services Inc., 200 Meacham Avenue, Elmont, NY I1003 and at additional mailing offices. US Postmaster: send address changes to Chemical Society Review, c/o Publication Expediting Services Inc., 200 Meacham Ave- nue, Elmont NY 11003. All dispatches outside UK by bulk airmail within Europe and Accelerated Surface Post outside Europe. 0 The Royal Society of Chemistry, 1997 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic, mechanical, recording, or otherwise, without the prior permission of the publishers. Typeset and printed in Great Britain by Black Bear Press Limited.

ISSN:0306-0012    

DOI:10.1039/CS99726FX013

出版商:RSC    

年代:1997

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99726BX015

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Surface scientific aspects in semiconductor electrochemistry H. J. Lewerenzt Brandenburgische Technische Universitat Cottbus, Faculty of Physics, Erich- Weinert-Str. 1, 03046 Cottbus, Germany Some of the fundamental properties of the reactive semicon- ductor-electrolyte interface are outlined and possibilities for electrochemical modification of semiconductor surfaces are discussed. The present status of investigating the physicochemical and morphological changes after (photo- electrochemical) processing is reviewed for selected exam- ples. The accessibility of near surface changes by a selection of surface sensitive techniques is presented and the informa- tion obtained by ultrahigh vacuum ex situ methods such as high resolution electron energy loss spectroscopy (HREELS), and ultraviolet photoelectron spectroscopy (UPS) is compared with in situ techniques such as Fourier transform infrared spectroscopy (FTIR).The suitability of electrochemistry-atomic force microscopy (EC-AFM) to follow in situ the surface microtopographic changes during electrochemical processes is emphasised. Examples are presented on the electrochemically hydrogenated Si (111) surface and on monitoring in situ the changes at the silicon- silicon oxide-electrolyte interfacial region during current oscillations. 1 Introduction The continuous improvement of investigative methods over the last 15 years has resulted in a wealth of new possibilities to Hans-Joachim Lewerenz studied physics at the Technical University Berlin, and obtained MS and PhD thesis at the Fritz- Haber Institut der Max Planck-Gesellschaft on photoemission from metals into electrolytes.He was a member of the technical staff for two postdoctoral years at Bell Laboratories, Murray Hill, working on semiconductor and photoelectrochemical interfaces, and photoelectrochemical solar cells. As a member of staffat the Brown Boveri Research Center in Switzerland, his research area was catalysis. He has worked at the Hahn- Meitner-Institute since 1982, conducting research on photo- voltaics, semiconductor materi- als science and photoelectro- chemistry. He has been an hon- orary Professor of Physics at the Technical University of Ber-lin since 1993, and is a visiting professor at the Technical Uni- versity Cottbus (1996197).He is currently head of the depart- ment of interfaces at the Hahn- Meitner-Institute. He has pub- lished over 140 scientific ar-I ticles and a book on 1 photovoltaics, and holds 13 I patents. t Permanent address: Hahn-Meitner-Institut Berlin GmbH, Glienicker Str. 100, 14109 Berlin, Germany. extract information on interfacial changes at semiconductors subjected to electrochemical or photoelectrochemical treat- ment.1-5 As a consequence, the field is characterized by the technical development of the analysis methods and simul- taneously by the advances in reproducible electrochemical surface conditioning.6-8 Many of the advances stem from purely fundamental research but frequently applications could be found.Among such examples are the electrochemical hydro- genation of silicon which occurred in the course of a still rather obscure current transient,3,5,6 smoothing of Si due to current oscillations,’ photoelectrochemical surface transformation of InP and CuInSe2 in electrochemical solar cells109l1 and photoelectrochemical etching of layered materials for solar cell application,12 for instance. There are, of course, many other results and applications but in this article I will restrict myself to examples which represent the technical advancement and diversity of the analysis methods employed and to specific recent selected results on electro- chemical semiconductor surface alteration. As an overview, Fig. 1 shows some of the areas of possible applications of semiconductor electrochemistry.semiconductor Fig. 1 Schematic on applications of semiconductor electrochemistry In sensors, for instance, the potential and current control which is typical for electrochemical processing can improve the selectivity of the species to be detected; solid state or photoelectrochemical solar cells can be improved by surface modifications which result in improved electronic properties and/or interfacial layers which inhibit or suppress leakage currents or chemical reactions with ambient or contact phases.l3,I4 In micromechanics, the extended application of electrochemical processing, particularly of Si, would allow high reproducibility of very well-ordered structures.Other important areas are electrodeposition of semiconducting films, 15 light-induced electrocatalysis16 and the characterization of the basic semiconductor properties following film preparation in a materials development laboratory. Chemical Society Reviews, 1997, volume 26 239 2 The reactive surface In comparison to surfaces exposed and investigated in an ultrahigh vacuum (UHV) environment, where the preparation of the solid surface itself is often problematic, the surface of a semiconductor ‘forms’ upon immersion into the respective electrolyte. Electrochemical parameters can influence the surface conditions via potential and current flow and subse- quently lead to alterations due to the interfacial reactivity. A view of the surface region is given in Fig. 2.Fig. 2(a-b) show the surface and interface properties on a more microscopic scale, whereas Fig. 2(c) displays some of the mesoscopic features. In concentrated electrolytes and not too highly doped semiconductors, the Galvani potential drop appears predom- inantly on the semiconductor side, since the respective capacitances CHH (outer Helmholtz layer) and Csc (space charge layer) can differ by a factor of ca. 100, depending on relative permittivities in the semiconductor and the Helmholtz layer and on the doping level as well as on ionic concentration semiconductor \i 0 0 0 0 0 0 0 0 \ \ inner outer n-semiconductor electrolyte ‘I II c-I‘ II ’I \ ’I -7-+ film formation in solution.Thus, the relative voltage drop is determined from eqn. (1). QF=Q(&+&) It is almost negligible on the electrolyte side (C, total capacitance). Applied potentials drop predominantly in the semiconductor surface and space charge region, allowing the investigation of the semiconductor using potential and current as parameters. The inner Helmholtz layer, characterized by the centre of specifically adsorbed ions is already an example for the reactivity of the surface. Defining the point of zero charge (pzc) as the potential at which ideally no excess charge exists on both sides of the junction (characterized experimentally by a minimum in the differential capacitance17 one observes for instance adsorption of anions negative from the pzc.This behaviour is attributed to the chemical reactivity of the semiconductor with a specific ion in solution which is strong E t vs < 0 EC EF E, n-semiconductor hv \ 4-dissolution Fig. 2 Schematic illustration of processes at the semlconductor electrolyte interface; (a) charge distribution, (b)energetic scheme for charge transfer processes, (c) photocorroslon with insoluble film formation or semiconductor dissolution (see also text) 240 Chemical Society Reviews, 1997, volume 26 enough to result in a partial removal of the solvation shell as indicated in Fig. 2(a)by the semicircles. As a consequence the position of the band edges can be altered compared to the so-called flat band situation where no band bending (i.e.electrical field) exists in the semiconductor.In the ideal case (absence of surface states, specific adsorption and corrosion reactions) the flatband potential and the pzc should coincide. Fig. 2(6) shows the charge transfer properties at the semiconductor-electrolyte junction. Depicted are redox reac- tions between electrons from the semiconductor conduction band and rechargeable species in the electrolyte [eqn. (2)]. In this case, the electron which is transferred in a tunnel process according to18 eqn. (3) (v frequency factor, K k,,, = IU(E)K(E,= Ef)Do(E,)D,(Ef)dEwith E, = E, --oo (3) transition probability, Do, D, density of occupied and un-occupied states) stems from the semiconductor conduction band.The expressions for cathodic current flow are given by i = qk,,fn, (0) (i, = i: -i;) with: (4) Since the charge transfer occurs in a narrow energy interval above and below the band edges, the integration with energy [eqn. (3)] is not necessary to a good approximation; Wred,ox describes the thermal distribution function for electronic states in the redox electrolyte with gaussian character;l8 n,(O) is the electron concentration at the surface in the conduction band without applied voltage and Cox the concentration of the oxidized species in solution. The voltage dependence at the semiconductor-electrolyte junction is given by the dependence of n, (x = 0, V,) (Va, applied voltage). With the usual approximations, l9 this change is given by a Boltzmann term and the current-voltage curve for an n-type semiconductor depicted here is given by eqn.(5). This is the diode equation known from solid state physics, except for the term i,," which is different from the reverse saturation current in semiconductor diodes, eqn. (6). ic.0 = qk,~'(Ec>Co,W,x(Ec>n,(x = 0, Va = 0) = 4k:K'(Ec>cRedWR,d(Ec)~c (6) Here N, is the effective density of states at the bottom of the conduction band. Finally, Fig. 2(c)shows the surface changes resulting from the interaction of electrons and holes with the semiconductor surface and the solution. For the case of photocorrosion20 for instance, the formation of a passivation layer occurs if the product of the corrosion process is insoluble in the respective electrolyte.An example is shown in eqn. (7). n-CdS + 2hf, (hv)4Cdz + Sfolm (7) In this case the light induced excess charges are the holes from the valence band of n-type cadmium sulfide. The process results in the formation of soluble Cd2+ ions and an insoluble sulfur film which finally passivates the surface, inhibiting further charge transfer (current flow). It should be noted that the energetics of reaction (7)are also influenced by the intermediate step, i.e. the formation of Cd+ and S-on the surface. In the case that the corrosion product is soluble, as for example for GaAs in aqueous solution, the semiconductor dissolves according to eqn. (8). GaAs + 6h+ (hv)+ aq +Gaz + AsO;, + 64: (8) The difference between photooxidation rate and dissolution rate determines whether a residual partly oxidized layer remains if the sample is removed from the solution or the light switched off since partly oxidized surfaces might not dissolve as well as fully oxidized ones (see above).An intermediate case in which current still flows in the presence of a thin interfacial film is found in the electropolishing of semiconductors. Here, a dynamic equilibrium exists between film formation rate and etch rate at a fixed film thickness and the electrical field at the electrolyte contact tends to smooth the surface since the carriers are following the trajectories of the field. 3 Selected surface analysis methods In this section, a selection of experimental methods for analysing the semiconductor surface during (in situ) and after electrochemical treatment (ex situ) will be presented.The selection is made on the basis of the corresponding results, which will be presented and discussed subsequently. 3.1 Combined electrochemistry-UHV surface analysis system The system shown in Fig. 3 is designed to allow the standard UHV surface analysis characterization directly after interrupt- ing or terminating the electrochemical process without contact with ambient air. Therefore, a specifically constructed (photo)electrochemistry chamber is attached to a commercial surface analysis system.6 This electrochemistry chamber is shown in detail in Fig. 3(a).Its main features are: (i) permanent flow of high purity dry N2 gas, the use of 02 free N2-purged ultrapure water (Millipore), the use of highest available purity chemicals (ultrapure or reagent grade) for preparing the solutions; (ii) the semiconductor sample is mounted on a stub for direct transfer into the UHV preparation chamber and the respective electrolyte is supplied via a glass rod (L) which can also be used as light-guide for photoelectrochemical experi- ments; (iii) a cylindrical Pt electrode and a Luggin capillary are used as counter and reference electrode, respectively; (iv) for interruption of the experiment, the electrolyte solution can be removed by jet-blowing with N2.This has the advantage of fast interruption (under potential, for instance) and since hydro- carbons tend to reside on the surface of water droplets, relatively low contamination of the semiconductor surface is expected; (v) after removal of the electrolyte, the semiconductor surface is thoroughly rinsed and dried in an N2 stream.Nevertheless, it should be noted that this rinsing and drying procedure can influence the surface at the microscopic level investigated here. Species which are bound strongly enough to withstand rinsing and which are not easily dissolved in water will remain on the surface. Also, the drying procedure as well as the following outgassing in the preparation chamber of the UHV-system will result in a reduction of water and weakly bound hydroxy groups on the surface; (vi) finally the sample is transferred from the preparation chamber to the analysis chamber if the pressure in the former reaches values below 10-9 hp.It turns out that this system and the procedure result in hydrocarbon contamination values after current flow in the electrochemical cell of ca. 0.15 monolayers. Consequently, all the established surface analysis methods can be applied, such as ultraviolet photoelectron spectroscopy (UPS), X-ray pho- toelectron spectroscopy (XPS), high resolution electron energy loss spectroscopy (HREELS) and low energy electron diffrac- tion (LEED], for instance. The results of a UPS and an HREELS experiment on electrochemically hydrogenated silicon will be shown in Section 4. 3.2 In situ Fourier transform infrared spectroscopy This method allows the measurement of interfacial changes during current flow by analysis of the vibrational losses from species formed in the electrochemical process.In the case of Si, which is presented here, a (1 1 1)-orientated single crystal is used in the attenuated total reflection-multiple internal reflection (ATR-MIR) mode (Fig. 4):the IR probe light enters the crystal Chemical Society Reviews, 1997, volume 26 241 Sol. Supply (4 RE L n LOC -1 Waste AC U Fig. 3 Combined electrochemistry-UHV system for ex situ surface analysis; (a)details of the electrochemical vessel, L light guide and solution supply, CE counter electrode, WE working electrode, RE reference electrode, LC Luggin capillary (see text); (b)overview of the integrated EC-UHV system; PC preparation chamber, AC analysis chamber, T1, T2, T3 transfer rod, EC electrochemistry chamber such that multiple total reflections occur at the silicon-electrolyte interface, the multiplicity of the reflections depend- ing on the angle of incidence and the thickness of the crystal.Since the IR light does not have to penetrate the electrolyte solution, reflection losses due to Si-H and Si-0 vibrations can be measured. The vibrations are excited by the evanescent wave directed parallel to the surface normal of the ATR crystal. Because of multiphonon absorption in the wavenumber region where the signal from the asymmetric Si-0 stretching mode is expected, a particularly designed electrochemical cell with a Si crystal of drastically reduced size and thickness was used to minimize the optical path length to 11 mm in the material.Hence the Si-H stretching mode, analysed with a large (7 cm2) crystal (as shown in Fig. 4) and the Si-0 stretching mode become experimentally accessible. The Si-F vibration, how- ever, lies outside our experimental range. The measuring cells are completed by a Pt counter electrode and a Ag/AgCl reference electrode as shown in Fig. 4. 3.3 In situ Electrochemistry-atomic force microscopy Since the changes at semiconductor surfaces often result in the formation of films which either are passivating or less conductive than the substrate material, the microtopographic analysis is difficult to perform using standard scanning tunnelling microscopy.21 In addition, the presence of the electrolyte and the problematic distinction between Faradaic and tunnelling currents makes this method less suitable.Instead, the scanning probe microscopy based on attractive and repulsive mechanical forces in the atomic region appears very well suited: insulating layers can be imaged and in addition the imaging process is not complicated by electronic processes. It allows a comparably accurate mapping of the surface micro- topography, provided that the height changes compared to the width of flat regions are not too large. Otherwise, one would also have to take into account the finite size and shape of the imaging needle. The AFM method uses the fact that the probing needle, once it is in attractive contact with the sample, is moving with the sample structure when measuring in the so-called constant force mode (Fig.5). The needle deviation is translated via a laser beam deflection system on a photodiode array allowing accurate height measurements (after calibration with well known steps on mica, for instance). Because of the laser light, the investigation of n-type Si appears problematic. In this case the light produces holes as excess carriers which can undesirably oxidize the surface. Consequently, anodic reactions of Si investigated by EC-AFM were performed with p-Si where the light-induced excess carriers are electrons and anodic polari- zation results in a deflection of electrons away from the interface. 4 Selected results and discussion 4.1 Fundamental electrochemical observations The (photo)electrochemistry of Si in aqueous fluoride-con- taining solutions is characterized by a series of surprising features.Among them is the observation of quantum yield multiplication,22 the occurrence of a dark current transient on formerly oxidized Si surfaces,23 the formation of ultrathin nanoporous Si in dilute ammonium fluoride solutions at potentials positive from flatband,24 an electropolishing regime and current oscillations. Fig. 6 shows a typical photocurrent-voltage curve (a)and a dark current transient (6) obtained on n-Si(ll1). The iph-V characteristic [Fig. 6(a)]exhibits two maxima and for V > 4 V, photocurrent oscillations set in. In the region between flatband and the first maximum, porous Si is formed, as evidenced by a series of in situ FTIR and high resolution scanning electron microscopy experiments.25 At larger anodic potentials, oxide formation and electropolishing of the surface followed by the formation of thicker oxides and current oscillations also take place as indicated in Fig.6. The dark current displayed in Fig. 6(6) is always observed when an oxidized Si sample is exposed to F- containing acidic solution. Following electroless etching of Si02 according to eqn. (9), the current rises when Si02+6HF + aq -+SiFz-+2H20 + 2HCl Si02 + 3HF-zaq +aq +SiFz-+H20 +OH-,, (9) the etch front reaches the silicon oxide-silicon interface thus making contact between the Si and the electrolyte.Ex situ XPS and UPS analyses using the system described in Section 3.1 (see also Fig. 3) have shown that the surface undergoes a transformation from oxidic coverage, enrichment of F- at the surface around the dark current (iD) maximum and increasing hydrogenation during the decay and levelling out of iD.26 Several electrochemical methods to prepare H-terminated Si(11l), Si( 1 13)27 and Si( 100) surfaces have been developed. Best results with respect to smoothness of the surface and high electronic quality were obtained by a two-step procedure in which the oxide etching is made at pH 4.0 and residual oxidic 242 Chemical Society Reviews, 1997, volume 26 Air Electrolyte Fig. 4 Experimental arrangement for FTIR measurements in the ATR-MIR configuration; W window, CS spacer, I/O in and outlet, D detector, CE, WE, RE, as in caption of Fig.3 Fig. 5 Experimental arrangement for in situ electrochemistry-AFM measurements; see indications in the figure traces are then removed by subsequent etching in a solution with pH 4.9, where the oxide etch rate28 [eqn. (lo)], is very small. k, = a[HF] +b[HF,] +C (10) Another procedure uses sustained photocurrent oscillations at large anodic potentials [+6 V (SCE)] followed by a change in potential to +0.5 V (SCE; standard calomel electrode) in the dark where the oxide present during oscillations is removed during the dark current transient. The latter method allows the preparation of surfaces with very high electronic quality (see below Section 4).4.2 The hydrogenated Si(111)-H (1 X 1) surface Using the combined electrochemistry-UHV surface analysis system (Fig. 3), the surface of Si( 1 1 1) has been analysed by high resolution energy loss spectroscopy (HREELS). 15 This tech- nique allows the detection of vibrational losses of primary low energy electrons impinging on the surface, measured in specula reflection geometry. The result of such an experiment in which the surface has been prepared by the aforementioned two-step electrochemical procedure is given in Fig. 7. The Si-H bending mode and the Si-H stretching mode are very pronounced (78 and 257 meV, respectively) although the excitation cross section is comparably low, revealing the high quality of the surface.Additional vibrations occur due to phonons or due to oxidic and hydroxidic remnants as indicated in Fig. 7. Small (a) oxide formation roughening.-I0 0 2 4 6 8 1ov ENvs.NHE 120-n-Si(l11)90 -60 -30 -I 1 I I 50 100 150 200 tls Fig. 6 Current characteristics of Si( 11 1) in aqueous ammonium fluoride solution; (a)photocurrent-voltage curve pH 4.0,O.1 M NH4F, light intensity 20 mW cm-2 (tungsten-iodine lamp); (b) dark current versus time; pH 0.1 M NH4F (NHE = normal hydrogen electrode) parts of the surface are covered by OH groups and some oxygen is found in the Si backbonds. The electronic band structure of ideally H-terminated Si(111) surfaces has been calculated theoretically29 and the data are shown in Fig. 8 together with a result from UPS measurements with variation of the polar angle of the emitted electrons.In Fig. 8(a) the electronic bands along symmetry lines of the surface Brillouin zone (BZ) are plotted. It can be seen that gap? in tJeproje_cted bu_lk basd structure exist, for instance along r -K, r -M and K -M, near the outer part of the surface BZ at energies between 4 and 7 eV below the valence band edge [zero energy on the scale of Fig. 8(a)]. The hydrogenation of the Chemical Society Reviews, 1997, volume 26 243 78 SI-HBend I Si-H + Phonon Si-H Stretch N-Hx Si-OH 2x78 419 I 454 I x 500 x 5000 I 1 I I 1 0 100 200 300 400 500 Energy loss / meV Fig. 7 High resolution electron energy loss spectrum of Si( 1 11) obtained by removal after the decay of the dark current at pH 4.0,subsequent immersion into a solution of 0.I M NhF with pH 4.9 and final removal after levelling out of the residual dark current (two step procedure); primary electron energy 6 eV, energy resolution 4 meV, background pressure 1 x 10-10 hP surface leads to additional electronic states which partly overlap with the bulk band structure (surface resonances) and partly lie within the forbidden regions.The latter surface states due to the hydrogenation would be expected at tilt angles of ca. 20" to 30" of the surface normal of the Si sample towards the electron analyser at energies below the valence band edge of 3.8 eV [small gap in Fig. 8(a)]and 4.5 eV (large gap). Fig.8(b) shows two photoelectron spectra of a Si surface treated by the two-step procedure for the dark current: at 8 = O", the surface normal points towards the electron analyser and the well known band structure features of Si due to emission from 3p levels (valence band top) and lower lying 3s levels are observed. At a 25" tilt angle, however, the spectrum is drastically changed. The emission from the Si 3p level at -1.3 eV is quenched according to the dispersion of the initial state towards K (rcorresponds to normal emission) which is lowered to ca. -2 eV at 0 = 25". Furthermore, a dispersion of the maximum at -7 eV (Si 3s, 3p) towards lower binding energy is noted, in accordancs with the calculated band structure (lower band at -7.8 eV at r).The most prominent feature, however, is the occurrence of a maximum at EB = -4.5 eV which can clearly be attributed to the Si-H surface band in excellent agreement with the theory.The expected small signal at -3.8 eV cannot be observed at this tilt angle. The process of surface hydrogenation can also be monitored in situ by FTIR. In this case, two surface treatments are compared: the first electropolishing treatment, keeping the sample for 15 min at +1.6 V (SCE) at a photocurrent density of iph = 400 pA cm-2; this treatment is known to produce smooth surfaces with very low density of interface states. The second procedure consists of the aforementioned combined photo- current oscillation and dark current decay. Fig. 9 shows the FTIR signals for electropolishing (curve b)and for the two step procedure involving oscillations (curve a) with reference to the Si-H free surface (oxidized Si) in the wavenumber region of the Si-H and Si=H2 stretching modes.A pronounced signal at ca. 2080 cm-1 (Si-H) and 2120 cm-1 (Si=H2) is found for both treatments indicating hydrogenation of the surface. From the difference spectrum (c) obtained by subtracting spectra (b)-(a)one sees a remaining signal which can be attributed to the Si=H2 stretching mode. Since two H atoms are exposed at steps in the direction of a (1 11) surface, the difference spectrum (c) shows that the combined oscillation procedure results in a reduced density of such steps compared to electropolishing, hence the surface after treatment (a) is smoother than after treatment (b).244 Chemical Society Reviews, 1997, volume 26 25" 0" I I 1 If' 1 I I -12 -10 -8 -6 -4 -2 0 Binding energy, €,/eV Fig. 8 Photoelectron spectroscopy on Si( 111) treated by the electrochemical two step procedure; (a) surface band structure calculation for hydrogen terminated Si( 111); (b)photoelectron spectra obtained for normal emission (0 = 0")and for a tilt angle 0 of 25" (see text); detector acceptance angle f15" 4.3 In situ analysis of dynamic processes during current oscillations The dependence of the oscillation period on solution composi- tion allows the adjustment of oscillations which are slow enough to be followed by in situ techniques such as FTIR, EC-AFM and microwave reflectivity, for instance.In the following the changes of the Si-0 vibrational absorption during photo- current oscillations and of the outer oxide topography will be investigated. With a modified ATR-MIR configuration using a miniaturized cell, the asymmetric Si-0 stretching vibration in the wavenumber region between 980 cm-l and 1260 cm-l could be monitored simultaneously with the current oscilla- tion. In a different experiment, the topography at the oxide- electrolyte interface was monitored by in situ AFM simul- taneous with the current oscillation at psi. The roughness was determined from line scans across the sample according to eqn. (11), where the surface topography is described by the L, L, R, = JjIf(X,Y)bdY (1 1) 00 function f(x, y) and L,,, L, are the lengths of the line scans.The two sets of experiments (photocurrent oscillation, Si-0 signal and current oscillation, roughness) were made under slightly different experimental conditions: the former experiment was performed in 0.1 M NaF on n-Si( 11 1) at pH 4.0 with a period of 95 s. The AFM experiment was made in 0.1 M NH4F at pH 4.2 (to increase the period since the recording period is 5 s with AFM) on p-Si( 1 11). The very characteristic behaviour of each experiment is condensed into Fig. 10 by adjusting the timescale accordingly and plotting the photocurrent, the roughness and the Si-0 absorption vs. time. It is found that the roughness is largest in the increasing region of the current and drops already when the photocurrent reaches its maximum value.The Si-0 signal varies by more than +20% (ca.k 5 monolayers under these experimental conditions) and is minimal at the time when the roughness is largest. These data cannot easily be interpreted and I will merely describe the findings: the maxima in the photocurrent are confined to a smaller time region than the more extended minima. This fits with the observation that the Si-0 signal exhibits two slopes. The formation (increasing Si-0 signal) appears to be faster than the chemical dissolution P 9 I I II I SiH 3 f"'"' 2000 2100 2200 Wavenumber / cm-l Fig. 9 FTIR spectra on Si( 111) for different surface conditioning; (a)after photocurrent oscillations at +6 V (SCE); (b)after electropolishing (see text); (c) difference spectrum (6)-(a);solution IM NaF pH 4.0 (smaller slope).Since the maximum oxide thickness is reached at the decreasing branch of the photocurrent, passivation is good and the oxide etches chemically obviously resulting in a smoothing of the outer oxide parts. Around the minimum oxide thickness the roughness reaches its maximum indicating that a rough surface of the inner part of the oxide has been reached by the etch front. With increasing current the then growing oxide at the silicon-silicon oxide interface results in a smoothing of the silicon oxide at the electrolyte interface due to simultaneous etching. Further investigations on the corresponding microwave reflectivity which allows us to distinguish between charge transfer and surface recombination losses are presently being performed.In addition a spatial coupling of the oscillations via the electrolyte has been found but it is too early yet to arrive at conclusive model descriptions. 5 Synopsis The modern surface analysis methods can be very well applied for the investigation of surface changes which occur during electrochemical processing of semiconductors. Several diffi- culties, however, have to be kept in mind: the analysis using ex-situ methods (UPS, XPS, HREELS and LEED) involves three steps which are not uncritical. First, the sample has to be removed from the electrochemical cell. Whether this is done under applied external potential or at the rest potential can already alter the surface condition.The second step involves the transfer from the electrochemical environment to, for instance, a UHV preparation chamber. It is very important that con- tamination from the respective laboratory ambient is suppressed as much as possible. The rinsing and drying procedure after removal from the cell also belongs to this transfer step and might lead to surface alterations due to the removal of less strongly bound adsorbates or to dissolution of species in the ultrapure oxygen free rinsing water. On the other hand, precipitates formed upon removal from the cell can be dissolved. The third step is the outgassing of the sample in the preparation chamber of a UHV system. Here, weakly bound water and hydroxy groups can evaporate leaving a surface with an eventually more open structure and a formerly hydrophilic surface can change to hydrophobic.The application of such techniques to hitherto unknown surface modifications is important and worthwhile even if not all the effects introduced during the three transfer steps are well known, but it should be t 0111111111111111 0 50 100 tls Fig. 10 Comparison of the temporal evolution of current, integrated Si-0 signal obtained from FTIR and roughness parameter during current oscillations on Si( 11 1) (see text) Chemical Society Reviews, 1997, volume 26 245 stressed that the uncertainties of the procedure are kept in mind when interpreting the obtained data.For the hydrogen termi- nated hydrophobic Si( 111) surface, these difficulties are marginal and we can make use of the HREELS experiment, for instance. This technique allows us to identify almost all the species present on and even below the topmost atomic layer with high resolution in the submonolayer range. The in situ methods give a more realistic description of the surface changes but since they include the analysis of water and hydroxy groups into surface films, it is difficult to judge whether the in situ monitored surface condition is stable for a subsequent process in a technology line. In addition, if one considers the example of the FTIR method, only Si-H and Si-0 species can be monitored since Si-F is outside the experimen- tally accessible range in the ATR-MIR configuration.There- fore, the information gained here is more direct but less general compared to an HREELS experiment. The in situ AFM method can give very important information on topographical changes not otherwise obtainable. It should be noted, though, that the images contain a temporal development during the scanning procedure and that the influence of illumination by the laser deflection system has to be excluded. It is this author’s opinion that due to the complexity of the sample handling procedures reliable results can only be obtained by a combination of ex situ and in situ techniques, particularly if the introduced surface modification is made for specific applications in semiconductor devices such as solar cells and transistors.6 Acknowledgements I am indebted for their important contributions and many enlightening and rewarding discussions to the following persons: Professor Karl Jacobi, Dr Markus Gruyter (Fritz- Haber-Institut der Max-Planck-Gesellschaft) and Dr Thomas Bitzer (International Surface Science Centre, Liverpool) for the HREELS investigation; Dr Jorg Rappich, Mr Stefan Rauscher (Hahn-Meitner-Institut Berlin) for the FTIR experiments; Dr Helmut Jungblut, Mr Oliver Nast (Hahn-Meitner-Institut) for the EC-AFM research, Mr Stefan Rauscher for the UPS investigation and Ms Wiebke Frandsen for drawing the figures. References 1 L. M. Peter, D. J. Blackwood and S. Pons, Phys. Rev. Lett., 1988, 62, 308. 2 R.A. Venkateswara, F. Ozanam and J. N. Chazalviel, J. Electrochem. Soc., 1991, 138, 153. 3 H. J. Lewerenz and T. Bitzer, J. Electrochem. Soc., 1992, 139, L21. 4 J. Stumper, R. Greef and L. M. Peter, J. Electroanal. Chem., 1991,310, 445. 5 H. J. Lewerenz, T. Bitzer, M. Gruyters and K. Jacobi, J. Electrochem. SOC., 1993, 140, L44. 6 T. Bitzer and H. J. Lewerenz, Suq. Sci., 1992, 2691270, 886. 7 D. J. Blackwood, A. Borazio, R. Greef, L. M. Peter and J. Stumper, Electrochim. Acta, 1992, 37, 889. 8 H. J. Lewerenz and H. Jungblut, in Semiconductor Micromachining, vol. I (Fundamentals),ed. S. A. Campbell and H. J. Lewerenz, Wiley, London, 1997, ch. 3. 9 J. Rappich and H. J. Lewerenz, Electrochim. Acta, 1996, 41, 675. 10 H. J. Lewerenz, D.E. Aspnes, B. Miller, D. L. Malm and A. Heller, J. Am. Chem. Soc., 1982,104, 3325, 11 S. Menezes, H. J. Lewerenz and K. J. Bachmann, Nature, 1983, 305, 615. 12 D. Mahalu, L. Margulis, A. Wold and R. Tenne, Phys. Rev. B, 1992,45, 1943. 13 A. Heller, Acc. Chem. Res., 1981, 14, 154 and references therein. 14 H. J. Lewerenz, M. Liibke, K. J. Bachmann and S. Menezes,Appl. P hys. Lett., 1981, 39, 798. 15 G. L. Schnabel and P. F. Schmidt, J. Electrochem. SOC., 1976, 123, 31OC. 16 H.-M. Kiihne and H. Tributsch, J. Electroanal. Chem., 1986, 201, 263. 17 C. H. Hamann and W. Vielstich, Electrochemie I, Verlag Chemie, Weinheim, 1975. 18 H. Gerischer, J. Electrochem. SOC., 1984, 131, 2452. 19 H. J. Lewerenz and H. Jungblut, Photovoltaik, Grundlagen und Anwendungen, Springer Verlag, Heidelberg, Berlin, New York, 1995. 20 H. Gerischer, in Semiconductor Liquid Junction Solar Cells, The Electrochemical Society, Proc. vol. 77-3, ed. A. Heller 1977, ch. 1. 21 S. Rauscher, T. Dittrich, M. Aggour, J. Rappich, H. Flietner and H. J. Lewerenz, Appl. Phys. Lett., 1995, 66, 3018. 22 H. J. Lewerenz, J. Stumper and L. M. Peter, Phys. Rev. Lett., 1988, 61, 1989. 23 M. Matsumara and S. R. Morrison, J. Electroanal. Chem., 1983, 147, 157. 24 J. Rappich, H. Jungblut, M. Aggour and H. J. Lewerenz,J. Electrochem. Soc., 1994, 141, L99. 25 T. Dittrich, S. Rauscher, V. Yu. Timoshenko, J. Rappich, I. Sieber, H. Flietner and H. J. Lewerenz, Appl. Phys. Lett., 1995, 67, 1134. 26 J. Rappich and H. J. Lewerenz, J. Electrochem. Soc., 1995, 142, 1233. 27 K. Jacobi, M. Gruyters, P. Geng, T.Bitzer, M. Aggour, S. Rauscher and H. J. Lewerenz, Phys. Rev. B, 1995, 51, 5437. 28 J. S. Judge, J. Electrochem. Soc., 1971, 118, 1772. 29 K. C. Pandey, Phys. Rev. B, 1976,14, 1557. Received, 24th March 1997 Accepted, 10th April 1997 246 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600239

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Conjugated polymers incorporating pendant functional groups-synthesis and characterisation Simon J. Higgins Department of Chemistry, University of Liverpool, Chemistry Building, Crown Street, Liverpool, UK L69 720. E-mail: shiggins@liv.ac:uk The subject of this review is the synthesis and electropolymerisation of pyrroles or thiophenes with pen- dant functional groups (metal complexes, ligands, bio- molecules for protein binding etc.) to afford electrodes coated with a polymeric matrix containing the desired functionality. Such electrodes are being examined in electro- catalysis and in sensor fabrication. More recently, well- defined, soluble polythiophenes with receptor groups for alkali metal ions or for non-covalent binding of electron- deficient aromatics have been made chemically.These polymers can be spin-coated onto electrode surfaces. Their optical and redox properties are changed as a consequence of reversible binding in readily detectable ways. Chemical synthesis of polymers avoids the quite positive potentials usually required for electropolymerisation, which can be a problem for delicate or expensive functional groups. An- other approach is to attach the functionality to a suitably reactive polymer after electropolymerisation. Electropoly- merisation can also result in poorly adherent or inho-mogeneous polymer coatings, and the review concludes with an approach to tackling this problem with another class of functionalised pyrroles-those designed for the fabrication of monolayers, self-assembled on gold electrodes.1 Introduction In 1979 a communication was published detailing how pyrrole, on electrooxidation in 0.1 M Et4NBF4 in 'wet' CH3CN, gave a conducting film of polypyrrole (later shown to have the idealised structure 1) on the working electrode surface.' Later, polythiophene (2) was prepared similarly. These materials, together with polyacetylene (synthesised chemically in the mid- 1970s), have some remarkable properties which have stimulated a very large and increasingly interdisciplinary research effort since their discovery; early work is described in the two-volume book by Skotheim.1 For example, they may be reversibly redox- cycled between the positively-charged, conducting (p-doped) state in which they are formed (shown in 1 and 2) and a neutral, essentially insulating state; during this redox switch, ions must move into or out of the film to balance the charge.Their Simon Higgins is a Lecturer in Inorganic Chemistry at the Uni- versity of Liverpool. His back- ground was in coordination and organometallic chemistry, be- fore be became interested in modified electrodes and conju- gated materials. His research interests include the synthesis and electrochemistry of new conjugated oligomers and poly- mers, and the development of new phosphine ligands and complexes for catalysis and electrocatalysis. H H nBF,,-1 + nBF4-2 conductivity therefore varies with applied potential, in a manner reminiscent of inorganic semiconductors (the term 'doping' to describe oxidation to the conducting form was purloined, inaccurately, from that field).Moreover, as the electronic structures of the redox states are different, they have different colours (i.e. the polymers are electrochromic). Polythiophene, for instance, is blue in the conducting form and red in the insulating form. Discovery of these materials came as electrochemists were increasingly becoming interested in modified electrodes, that is, in chemically anchoring electrocatalytic or other redox-active species to electrode surfaces in order to exert finer control over the reactivity of the electrode than could be obtained simply by varying the applied potential. The first account of deliberate electrode modification dates from 1973; the known strong adsorption of alkenes at platinum surfaces was used to coat a platinum electrode with a monolayer of a platinum complex of a pyridine ligand with a pendant alkene.2 Clearly, by using thiophenes or pyrroles bearing some functionality (for example, a metal complex, a metal ion binding site, an anionic or cationic group) as monomers for electropolymerisation, electrodes could be modified with a potentially conductive matrix containing these functional groups.It was perceived that such systems could be useful in, for example, electrosynthesis, novel types of sensor, or electrochromic devices. The aim of this review is to highlight some important factors in the synthesis, characterisation and applications of function- alised polythiophenes and polypyrroles.Coverage is not restricted to polymers generated electrochemically, since chem- ical routes, to functionalised polythiophenes in particular, have now become useful. Clearly, work in this field must begin with synthesis, of the monomer for electropolymerisation or of the polymer itself. The review too, therefore, begins with some general remarks about synthetic routes to monomers and polymers. Approaches to characterisation of conjugated polymer-modified electrodes are next discussed. The covalent incorporation of some specific classes of functional groups into polythiophenes and poly- pyrroles is then discussed in turn. Conjugated polymers bearing every conceivable type of functionality have been made, including C60 derivatives, optically active side-chains, per- fluoroalkyl groups, electron donors such as phenothiazine, Chemical Society Reviews, 1997, volume 26 247 acceptors such as viologens (or combinations of the two), liquid crystalline groups etc.An apology is offered in advance to workers who feel that their contribution has been omitted unreasonably. However, it is hoped that coverage of a few types of functional group in appropriate detail should better illustrate the area than a mere catalogue of every possibility. 2 Synthetic routes to functionalised thiophenes, pyrroles and their polymers Much effort has gone into the synthesis of new heterocyclic monomers based upon thiophene or pyrrole, in order to tune the electronic structures and hence the redox and optical properties of the resulting polymers; this fascinating area has been well reviewed very recently,3 so this discussion will focus on the preparation of monomers bearing pendant functional groups.Polypyrroles can be redox-cycled between insulating and conducting forms at potentials compatible with aqueous electrolytes, unlike polythiophenes, a clear advantage in possible sensing applications. Moreover, the positive potentials needed to oxidise pyrroles to polypyrroles are not so extreme as those needed to oxidise thiophenes to polythiophenes. Clearly, this is a key point since, to incorporate a given functional group into a conducting polymer using electrooxidation, that func- tional group must be stable at the positive potential necessary to generate the polymer.Another consideration is the nature of the connection between the functional group and the heterocyclic polymer backbone. In work with alkyl-substituted polythiophenes, it was shown that the incorporation of bulky groups reduced the conductivity of the doped form, and decreased the degree of conjugation, unless the bulky group was separated by at least a four-carbon chain from the polymer backbone.4 Therefore, to incorporate bulky functional groups without affecting adversely the redox and optical properties of the polymer, a flexible chain must be employed as a linker between the heterocycle and the functional group; the bulkier the functional group, the longer this chain must be.Additionally, electropolymerisation pro- ceeds via generation of heterocycle radical cations, which are appreciably acidic; neither thiophenes nor pyrroles will electro- polymerise in the presence of basic moieties5 and this needs to be taken into account. Treatment of pyrrole with potassium metal gives pyrrol- 1 -ylpotassium, which, although potentially an ambident nucleo- phile, usually undergoes electrophilic attack at nitrogen (Scheme 1, A).6 Since this is the most straightforward way of functionalising pyrroles, the great majority of studies of functionalised polypyrroles involve N-functionalised pyrrole monomers. Unfortunately, in work with simple alkyl or aryl substituents, it was shown that the film-forming properties of pyrroles were diminished, and that the conductivity of the resulting polypyrroles was curtailed, by N-substitution; 3-sub- stitution had a much less adverse effe~t.~ Potentially, 3-func- tionalised pyrroles are therefore very attractive. Unfortunately, the syntheses of these are not trivial; the chemistry of pyrrole is heavily dominated by electrophilic substitution at the 2-(or 2,5-) position(s).6 To obtain 3-functionalised pyrroles se-lectively, it is necessary to protect the 2- and 5-positions by attaching a bulky group to the nitrogen. Scheme 1, B, shows one approach that has been applied here.* An alternative is to use Freidel-Crafts acylation of pyrrole-N-toluene-p-sulfonamide, in which case deprotection must be performed by base hydrolysis.The syntheses of 3-functionalised thiophenes are more straightforward as suitable starting materials [e.g. 3-bromo-thiophene, 2-( 3-thienyl)ethanol] are stable and commercially available, and one does not need protecting groups to manip- ulate them. Soluble polythiophenes can readily be made by FeC13 or [Fe(acetyla~etonato)~] oxidation of 3-alkylthiophenes, but, like electropolymerisation, this can give regiorandom polymers (i.e. a mixture of head-to-head, head-to-tail and tail- 248 Chemical Society Reviews, 1997, volume 26 A I H B Br I I H \ TiPS iii c-f-VI iv I TiPS TiPS Scheme 1 (A) Typical synthesis of a ligand with a pendant N-functionalised pyrrole: '2 (i) potassium metal, tetrahydrofuran (thf); (ii) excess Br(CH&Br; (iii) thf (treatment of 4,4'-dimethyl-2,2'-bipyridine with BunLi affords the reagent shown).(B)Synthesis of a 3-functionalised pyrrole, illustrating a common protection strategy:8 (i) BuLi, then triisopropylsilyl (TiPS) chloride; (ii) N-bromosuccinimide; (iii) BuLi, then excess Br(CH2)5Br; (iv) Mg, Et20, then ferrocene carbaldehyde (FcCHO); (v) Bu~NF,thf-HzO. to-tail coupled monomer units; Scheme 2).4 Recently, the synthesis of polyalkylthiophenes with a near-exclusive head-to- tail arrangement of monomers has been acheived.9 These polymers also have relatively high average degrees of polymer- isation and narrow molecular weight ranges. Interestingly, this chemistry has since been extended to the preparation of poly(o- bromoalkyl)thiophenes,9 which will doubtless find application in the preparation of structurally ordered, functionalised materials.This is an important development because it is well-known that the key properties of polythiophenes (the redox potential for the p-doping/undoping process; the electronic absorption and luminescence emission maxima for the nto n* transition of the neutral polymer) depend on the degree of conjugation. In chemically-generated polythiophenes, this is controlled mainly by the inter-ring twist angle; the more twisted adjacent monomer units are with respect to each other, the lower is the degree of conjugation, the higher is the redox potential, and the higher is the energy of the x to n* tran~ition.~The inter-ring twist angle is controlled in the first instance by the size of neighbouring substituents in 3-(or 3,4-) substituted polymers.This effect is well illustrated by recent work on polythiophene- based light-emitting diodes; the energy of electroluminescence could be controlled across the visible region simply by varying the size of the substituents. lo Blue electroluminescence could be obtained using chemically-generated poly-3-methyl-4-cyclo- hexylthiophene, in which the very bulky cyclohexyl groups reduce the degree of conjugation enough to raise the n to n* R R IJ- iii R R R &Br BrMgA$ r IV P R>95% HT Scheme 2 (A) Chemical oxidation of 3-R-thiophene (R = n-hexyl or longer) gives a soluble polythiophene, but this may have head-to-tail, head- to-head (zig-zag line) and tail-to-tail (bar) linkages." (B) Synthesis of regioregular, soluble polythiophenes? (iii) N-bromosuccinimide, dmf; (iv) BunLi, then MgBr2-2Et20, Et20, -78 "C; (v) 5 mol% [NiC12(Ph2P( CH2I3PPh2)] catalyst.absorption energy into the UV, and hence the emission energy into the blue. 3 Conjugated polymer-modified electrode preparation and characterisation Electrodeposition of conjugated polymers is usually best performed using repetitive scan cyclic voltammetry (RSCV), with the anodic limit just beyond the potential at which the monomer becomes irreversibly oxidised (to polymer) and the cathodic limit at a potential just negative of the polymer redox chemistry.Chronoamperometry (i.e.stepping the potential to a value at which monomer oxidation takes place for a fixed length of time, while monitoring current flow) or chronopotentiometry (i.e.passing a known current for a fixed length of time while monitoring electrode potential) can also be employed. 1 1 Control over film thickness can be exerted by controlling the amount of charge passed during electrodeposition. However, control over other factors, such as the morphology of the films, is less easily exercised; some recent advances in this area are discussed in the final section. Spin-casting a film onto the electrode is an alternative to electrodeposition, if the polymer is soluble and can be chemically prepared.Cyclic voltammetry is easy to perform, with equipment which is widely available.*I It can be used to assess, in a preliminary way, whether the redox chemistry of the functional group has been affected by incorporation into the polymer matrix. For instance, one question that immediately arises is: if the functional group has a redox process at a potential where the polymer matrix is in its neutral (insulating) state, can electron transfer still occur between the electrode surface and the functional group? Sometimes, if for example the functional group is a charged metal complex, there is sufficient ionic conductivity within the film to allow electron transfer by a hopping mechanism even when the polymer backbone is electronically insulating.l2,I3 In favourable cases, differences between the electrochemistry of the polymer-trapped functional group, and that of the same functional group in solution, are due only to the fact that the former has the characteristics of a surface-localised process and the latter has the characteristics of a solution process, coupled with the fact that, since a redox group within the polymer undergoes electron transfer, charge- balancing ions will need to enter or leave the polymer film. The latter process results in kinetic limitations, evident on studying the voltammetry as a function of scan rate. Electrochemists have devised ways to apply many spectro- scopic and physical techniques to electrode surfaces in situ under potential control, and conducting polymer-modified electrodes in particular have been much studied.11v15 When coupled with the electrochemical data, these techniques can give invaluable information. Electronic spectroscopy is parti- cularly useful for metal complex-containing polymers. The development of Fourier transform infrared (FTIR) spec-trometers, coupled with advances in thin-layer reflectance and attenuated total internal reflection techniques, has made IR spectroscopy a valuable tool for examining the nature of the charge carriers in conducting polymersls The power of in situ spectroscopic techniques for studying functionalised conduct- ing polymers has recently been applied to electrodeposited copolymers of 3 (perchlorate salt) with 3-methylthi0phene.1~7~~ In situ electronic spectroscopy showed that the energy of the n-n* transition for the neutral form of the copolymer was higher than that for poly-3-methylthiophene (P3MT) itself, indicating a shorter mean conjugation length.In situ FTIR spectra further showed that P3MT has very uniform conjugation lengths; the lowest-energy electronic transition of the oxidised form of conducting polymers can be seen in the IR, and the energy of this transition did not appreciably change with potential for P3MT.15 The corresponding band in the copolymer was at lower energy, and its maximum did vary with potential. This is probably due to a combination of the shorter mean conjugation length, and a higher static dielectric constant for this copolymer incorporating ionic moieties.Thus, incorpora-tion of the metal complex into P3MT could be achieved while maintaining the redox activity of the polymer, but at a significant cost in terms of important conjugated polymer properties such as the degree of conjugation. 4 Electropolymerisation of polypyridine metal complexes bearing pendant pyrroles Metal complexes of polypyridine and porphyrin ligands often display both metal- and ligand-centred redox processes, and have been intensively studied as electrocatalysts. Much effort has therefore gone into incorporating these into polymer- modified electrodes. The majority of studies here involve pyrroles; work until 1990has been already reviewed.12 Work on the immobilisation of metalloporphyrins in electropolymerised films has more recently been reviewed by one of the main groups in this field,*3 so these will not be dealt with here.Early work on incorporating metal-polypyridine complexes into polypyrrole showed that with coordinatively saturated complexes such as [Ru(bpy)2(L>2]2+ (bpy = 2,2'-bipyridine; L = 4 or 5) or [Ru(bpy),(L-L)3 -,]2+ (L-L = 6 or 7),the film- forming ability of the complexes, and the stability to repetitive 49 QIY cp2 4 5 6 7 R = 2,3,4 Chemical Society Reviews, 1997, volume 26 249 voltammetry of the films once prepared, improved on increas- ing the number of pyrrole groups per metal centre.13 In fact, with only one pendant pyrrole per metal centre, stable films could rarely be generated at all.Clearly, complexes bearing more than one pendant pyrrole can cross-link through the metal centres, giving more insoluble and therefore more robust polymer films; similar results are obtained when forming ionically conducting polymer films by electroreduction of 4-vinyl-4'-methylbipyridine complexes.l6 This lesson resulted in the synthesis of complexes with ligands bearing additional pyrrole groups, for instance 8 and 9. The fate of the 'polypyrrole' formed on electrooxidation of RuII complexes of ligands bearing pendant pyrroles is an important question. l2 Usually, workers in this area grew films by repetitive cyclic voltammetry, with a very negative potential limit to allow the ligand-based reductions to be observed and the positive potential limit just beyond the RuII/RuIII redox wave, i.e.about +1.3 V (vs. Ag/lO mM Ag+). Although films could successfully be grown by this protocol, the poly-(N-functiona1ised)pyrrole redox chemistry expected at +0.6 V was never seen. Only when films were grown by RSCV with the anodic limit just positive of the onset of pyrrole oxidation, i.e. +0.75 V, was a redox-active polypyrrole obtained. l7 If the RuV Ru"* process was subsequently scanned, again the polypyrrole electrochemistry vanished; a broad wave superimposed on the RUT' oxidation was assigned to the irreversible oxidation of the polypyrrole backbone. Thus, electropolymerisation of pyrroles can often be used to modify an electrode with a polymer- encapsulated metal complex even if the 'polypyrrole' itself is irreversibly destroyed by oxidation as it forms; presumably there is enough residual ionic conductivity within such films to allow continued pyrrole electrooxidation on top of the irreversi- bly oxidised material on successive scans.The problem with complexes of the type [M(bpy)3].+ (M = 4d or 5d transition metal) is that, for electrocatalysis, one is limited to outer-sphere electron transfer processes because these complexes are inert to ligand substitution. More recent work has focussed on metal centres having potential free coordination sites which could participate in inner-sphere processes, and on more reactive organometallic complexes. The coordination chemistry of the metal sites within films can often be manipulated electrochemically.For example, [RuC12(7)2 J (n = 4) readily electropolymerised in acetonitrile to afford poly-[RuC12(7)2].1* This had a reversible RuI1/RulI1 redox wave at +0.3 V (vs. standard calomel electrode, SCE). When repeatedly redox cycled over the RuI1/RulI1 wave in aqueous CF3S03H, conversion, via poly-[RuCl(H20)(7)2]+, to poly-cis- [R~(H~0)~(7)2]~+occurred. Irradiation of the latter with visible light caused isomerisation to poly-tr~ns-[Ru(H20>2(7)2]~+.It is known that, in solution, cis-[Ru(bpy)2(H2O>2l2+ undergoes successive pH-dependent stepwise two-electron oxidations to unstable cis-[RuVI(O)2(bpy)2J2+, which rapidly loses bpy to form trans-[RuV1(0)2(OH)2(bpy)].The latter reaction was inhibited in polymer films.'* This is of some importance because cis-[RuVI(0)2(bpy)2]2+, though unstable, is a good oxidation catalyst. A poly-~is-[Ru(H20)2(7)2]~+film was used to oxidise benzyl alcohol to benzaldehyde at +1.15 V with 89% current efficiency and 2200 turnovers. Related Ru" complexes have been used in homogeneous conditions in organic synthesis for the catalysis of olefin epoxidation by 0x0-transfer reagents like PhIO (including asymmetric examples); it would be interesting to develop electrocatalytic methods for these important reactions. The complex [(Cp*)RhCl(bpy)]+ (Cp* = q-C5Me5) is a homogeneous catalyst for the reduction of the important redox enzyme cofactor NAD+ to 1,4-NADH, which is kinetically slow.Electrodes modified with [(Cp*)RhCl(bpy)]+ might therefore be valuable in enzyme-based sensors. In acetonitrile solution the 18-electron [(Cp*)RhCl(L-L)]+ (L-L = various substituted bpy and 1,lO-phenanthroline ligands) are reduced in a slow two-electron process, at ca. -1.0 V (Ag/lO mM Ag+) to unstable, formally 20e species, which are in equilibrium with the 18e Rhr [(Cp*)Rh(bpy)] formed by chloride ion loss; this is confirmed by the behaviour of this redox wave in the presence of added Cl-.19 At < -2.0 V, successive one-electron reductions of L-L occur, with loss of any remaining C1-; on sweeping positive, the two-electron oxidation of [(Cp*)RhI(L- L)] is seen, at more positive potentials than for oxidation of [(Cp*)RhCl(L-L)]-.Using the ligands L-L = 8, 9 and 10, polypyrrole films incorporating [(Cp*)RhCl(L-L)]+ were grown. In this work, the polypyrrole backbone retained its redox activity because modest positive potentials (+0.7 to +0.9 V) were used to prepare the films. Interestingly, the charge under the RhI1I/Rh1 redox wave for a film of [(Cp*)RhC1(8)]+ was smaller than that for the polypyrrole redox process. Presumably, electron transfer is inhibited by the polymer backbone being electronically insulating in the Rh1I1/Rh1 potential region, coupled with low redox conductivity. By either repetitive cyclic voltammetry over the RhIII/RhI redox wave, or by poising the potential of the film just negative of this process, the reactive, coordinatively unsaturated species [(Cp*)Rh(L-L)I2+ could be accumulated in the film on reoxidation.When the ligand-based reduction process was accessed, a sharp current spike was seen at the foot of the cathodic wave; presumably, the remaining RhlI1 complex became accessible to the electrode as the ligand- based reduction made the film more ionically conducting. In aqueous solution, a catalytic wave due to hydrogen evolution occurred (at low pH), at the onset of the RhlI1/Rhl wave. Presumably, the mechanism of H2 evolution involves the reaction of the electron-rich [(Cp*)RhI(bpy)] with H+ to generate the hydride [(Cp*)RhIJ1H(bpy)]+. Platinum metal-phosphine complexes have been used in many homogeneously-catalysed organic reactions. Recently, efforts have been made to couple this chemistry with electro- chemical processes via electrode modification.This has also stimulated progress in the study of the electrochemistry of platinum metal-phosphine complexes, which had previously been rather neglected. For instance, the complexes cis,truns- [Rh(H),(C1)2-.(PR3)2(L-L)] (n = 0, 1, 2; PR3 = PPh3, PEtPh,; L-L = bpy, 2,2'-bipyridyl-4,4'-isopropylcarboxylate,8, 11) have been the subject of a detailed voltammetric and spectroscopic study.20 Like the Cp* complexes described above, irreversible two-electron reduction at ca. -1.0 V (vs. Ag/lO mM Ag+, in acetonitrile) is coupled with loss of chloride ions, to give [RhI(PR&(L-L>]+; broad waves on sweeping the potential positive again were attributed to two-electron oxida- tions to [Rh111(PR3)2(L-L)(CH3CN)*]3+and [RhC12(PR&(L- L)]+.If the experiments were conducted in the presence of a proton source (formic acid or water), [Rh(H)2(PR&(L-L)]+ complexes were formed. It is interesting that hydride ligands so stabilise RhIII that no metal-centred reduction was seen for these species; only the reversible ligand-based reductions at < -2.0 V were evident. However, the hydrides were irreversibly oxidised, at ca. +1.0 V, to [Rh(PR3)2(L-L)(CH3CN)2]3+with evolution of protons (detected by pH measurement after exhaustive electrolyses). Polypyrrole films incorporating these metal centres were grown using the ligands 8 and 11. The metal- based redox processes within films were generally similar to 250 Chemical Society Reviews, 1997, volume 26 11 12 R = n-hexyl; phenyl groups on phosphorus omitted for clarity those in solution, except for slower kinetics for chloride ion uptake and loss. However, one remarkable finding was that in films of poly-[RhC12(PEtPh2)2(8)]+,in which the electroactivity of the polypyrrole backbone is again maintained, the film imposes a form of chemical rectification on the complex (Fig. 1).On sweeping the potential negative, the expected RhlI1/ RhI wave was not present, but on accessing the first bpy-based reduction process (onset ca. -Z .5 V) a sharp cathodic current spike was seen, due to the combined ligand and RhlII/Rhl reductions. If the scan range was subsequently maintained between -1.0 and -2.0 V, only the ligand-based, reversible redox process was seen; the complex was 'locked' in its Rhl form.If the positive potential limit was then extended to +0.6 V, a sharp anodic current spike at ca. 0.0 V, at the onset of polypyrrole oxidation, was seen, due to combined polyp yrrole and RhI/RhIII oxidations. The explanation for this phenomenon is presumably the same as for the current spike seen for the Cp* complex described above; in the potential region where the polypyrrole is in its electronically insulating state, and prior to the onset of the bpy-based reduction where the ionic con- ductivity of the film will rise, the film is too insulating to mediate electron transfer between the metal centres and the electrode surface.The dichloride complexes could be transformed into the dihydrides within the polymer films by redox-cycling in the presence of a proton donor. Polypyrrole-modified carbon electrodes containing the dihydride complexes were tried in the electrocatalytic reduction of activated alkenes, and ketones.20 Product yields and Coulombic efficiencies were low, but the principle was demonstrated; using poly-[RhC12(PEtPh2)2( 1l)]+, cyclohexanone was produced from cyclohex-2-enone at -1.4 V (vs.SCE) in 14.5% yield (47% current yield). A limitation with this chemistry is the reliance on pyrrole-functionalised bpy ligands. Routes to phosphine ligands bearing pendant pyrroles or thiophenes would enable the incorporation of metal-phosphine complexes into polymer-modified elec- trodes without bpy co-ligands.It has been shown that nucleophilic addition to the double bond of coordinated (Ph2P)2C=CH2 can furnish platinum metal complexes of functionalised diphosphine ligands in one step.21 Recently, we have used this chemistry to prepare complexes such as 12 and 13, which can be used as monomers for electropolymerisa- tions. If any metal complex is to be electropolymerised, clearly the metal centre must be stable at the positive potential necessary to grow the polymer film, and this can pose problems. Recently, it has been found that ligands such as 7 (n = 4) and 11 can themselves be electropolymerised in acetonitrile media, provid- ing that deprotonation of the intermediate pyrrole radical cation by the bpy is suppressed by the presence of strong acid (HC104).22 Care needs to be taken that acid-catalysed pyrrole Fig.1 (A) Cyclic voltammograms, in CH3CN + 0.1 M tetrabutylammonium perchlorate (TBAP), of a glassy carbon electrode (5 mm diameter, surface coverage 7 x 10-9 mol cm-2 Rh, 100 mV s-1) prepared by the oxidative electropolymerisation of 1 mM [RhC12(PEtPh2)2(S)]+ in CH3CN + 0.1 M TBAP. Curve (a),first negative-going scan, -0.5 to -2.0 to +0.6 to -0.5 V, showing the current 'spikes' due to combined bpy and Rh"' reductions at ca. -1.8 V, and to the combined polypyrrole and RhI oxidations at +0.2 V; curves (b)and (c), scans restricted to the negative and positive potential regions respectively (initial scans not shown).Here, the redox processes seen are due only to the bpy ligands and the polypyrrole backbone respectively; the complex is trapped as Rh1 {probably poly- [Rh(PEtPh2)2(8)]+} or RhlI1 (probably p0ly-[Rh(PEtPh2)2(8)(CH3CN)2]~+} respectively. (B) Cyclic voltammogram (5 mm diameter glassy carbon electrode, CH3CN + 0.1M TBAP) of [RhC12(PEtPh2)2(L-L)]+(L-L = 4,4'-bis(isopropoxycarbonyl)-2,2'-bipyridine)in solution, illustrating the RhV Rhl redox processes (cathodic wave at -1.1 V; anodic waves labelled Pa" and Pa1) expected for the metal centre in (A), but suppressed by the insulating polymer film. The two anodic waves associated with the RhVRhI process are assigned20 to the formation of [RhCI2(PEtPh&(L-L)]+ (pa1) and [Rh(PEtPh2)2(L-L)(CH,CN)2]3+(pa1')respectively. The two reversible redox processes at more negative potentials are due to the successive bpy ligand-based reductions.(Reproduced from ref. 20. Copyright Elsevier Science SA, 1995.) Potentials vs. Ad10-2~ Ag+. oligomerisation6 (to non-conjugated oligomers) is avoided. Voltammograms of the resulting films (on Pt electrodes) show an irreversible reduction at ca. -1.O V (YS.Ag/lO mM Ag+) due to reduction of bpy-bound protons to H2, and the usual pair of one-electron reductions of the bpy- to bpy2- at very negative potentials. Whereas MnI1 complexes of ligands like 8 and 11 could not be obtained by electropolymerisation of [Mn(L- L)3I2+ (the MnI1 centre underwent irreversible oxidation at the potentials required to oxidise the pendant pyrroles), modified electrodes incorporating [Mn(L-L)#+ could be made simply by soaking the poly-(H28)2+ or poly-(H211)2+ electrodes in a dimethyl sulfoxide (dmso) solution of MnC12.Poly-[M- (diene)(L-L)]+ (M = Rh, Ir; diene = cycloocta-1,5-diene or norbomadiene) films could similarly be prepared, by soaking previously-deprotonated films of poly-8 or poly-11 in solutions of [{ M(diene)C1}2] in dmso-Bun4NC104. Complexes such as these are precursors to homogeneous catalysts, and it will be Chemical Society Reviews, 1997, volume 26 251 interesting to see whether electrocatalytic reactions can be developed using these modified electrodes. Attempts to develop sensor applications for polypyrrole- bipyridine complex films have been described.The complex [Ru(7)2(14)](BF& (n = 4) has two pendant pyrroles available for electropolymerisation, and a diamide-functionalised bpy ligand.'7 The amide groups in RUT' complexes of 14 have been MeO 14 15 16 demonstrated to act as a receptor for halide ions in solution; the 1H chemical shift of the amide protons changes as a function of halide ion concentration, and the first ligand-based reduction potential moves by up to 50 mV to negative potential in the presence of C1- ions. There are complications with the growth of films containing [Ru(7)2(14)](BF& because pyrrole oxida- tion almost coincides with an irreversible peak due to the oxidation of the dimethoxybenzene units. It is possible that the polymerisation involves the latter as well as the pyrroles, and the 'polypyrrole' generated was not, in any case, electroactive. However, what is beyond doubt is that the electrochemistry of the film so formed does depend upon the electrolyte anion; a shift in the potential of the first ligand-based reduction (localised on the amide-functionalised bpy) is seen in the presence of C1-, but not in the presence of Br- or I-.Using solution electronic spectroscopy, with [R~(bpy)2(14)]~+ as a model complex, the association constant for chloride ion binding was 1.7 X 106 mol-l 1. However, the films did not slavishly mimic the solution behaviour of [Ru(bpy)2(14)I2+; low levels of F- had no measurable effect upon the solution redox chemistry of [R~(bpy)~(14)]~+ but did greatly perturb the electrochemistry of films of poly- [Ru(7)2(14)I2+.5 Host-guest chemistry in conjugated polymers The archetypal example of host-guest chemistry was the development of crown ethers (hosts) for selective complexation of alkali metal ions (guests). More recently, much effort has been expended in designing crown ethers with additional functionality which allows straightforward detection of metal ion binding, for example a ferrocenyl group whose redox potential changes as a function of metal ion binding. By building crown ether functions into conjugated materials, the possibility exists that the properties of the polymer might change in a detectable way as a function of metal ion binding by the crown ethers, allowing the design of potentially simple sensors. One obvious property to use is the redox switching potential of the polymer; it might be expected that this would be affected by the presence of ions within the neutral film as a consequence of cation binding by the pendant crown ethers.Early attempts did not seem to give much ground for optimism. For instance, an N-functionalised pyrrole bearing a 15-crown-5 substituent gave a polymer, the electrochemistry of which did not vary appreciably as a function of the presence of alkali metal cations in the bathing electrolyte.23 However, this is an instance where the synthesis of 3-functionalised pyrroles, though more difficult, was worthwhile. Monomers 15 and 16 could be electropolymerised on Pt discs (CH3CN, LiC104 electrolyte) to give polypyn-ole films with high conductivities (1 S cm-1) which could be redox-cycled in both aqueous and non- aqueous (acetonitrile) media with much lower redox potentials 252 Chemical Society Reviews, 1997, volume 26 than the N-functionalised materials.24 The redox wave for the doping/undoping process for poly-15 did not change appre- ciably as a function of electrolyte cation (Li+, Na+ or K+), perhaps because this amide derivative of 12-crown-4 is not a particularly good ligand for any alkali metal ion.However, if poly-16 was grown in LiC104 electrolyte, then cycled in NaC104 or KC104, the anodic peak for polypyn-ole oxidation shifted positive, irreversibly, by up to 0.4 V in CH3CN (Fig. 2).Fig. 2 (a) Cyclic voltammogram (E vs. SCE) of a poly-16 film, polymerisation charge 0.1 C cm-2, platinum electrode, sweep rate 50 mV s-l, CH3CN, electrolyte 0.1 M LiC104. (b)Successive voltammograms of the same film after transfer to CH3CN-O.l M NaC104. From ref. 24. Presumably the binding of Na+ or K+ by the larger crown in poly-16 makes it thermodynamically harder to generate the oxidised form of the polymer than when Li+ (weakly bound by larger crown ether cavities) is the electrolyte cation. This may be due to changes to the polymer secondary structure caused by cation binding, which make doping less favourable, as will be discussed further below. The advantage often enjoyed by polypyrroles of water compatibility was of no use in this instance as a similar cation response was not observed in aqueous electrolytes. Polythiophenes incorporating pendant 12-crown-4 sites have also been prepared, by electrooxidising 17 and 18 in CH3CN- Bun4NPF6.In contrast to poly-16, these respond most strongly to the presence of Li+ions, as expected given the binding preference of this smaller crown.25 Even very low concentra- tions (25 X M) of Li+ in CH3CN were enough to cause obvious changes in the voltammetry of poly-17 (Fig. 3). With increasing [Li+], the polymer redox wave shifted positive, broadened, and decreased in area. Significantly, model poly- 3-alkylthiophenes showed no change in their redox properties in the presence of Li+ and 12-crown-4; the binding site must be covalently anchored to the polymer for the effect to be apparent.Among the most interesting developments in this area have been reports of some soluble polythiophenes, generated chem- ically.26 Scheme 3 shows the syntheses of polythiophenes C.8 0.6 0.4 4:20.2 ‘-0 -0.2 \\-I I I I I I-0.4 I 0 0.2 0.4 0.6 0.8 1 1.2 EIV Fig. 3 Cyclic voltammograms (E 1’3.Ag/AgCl) of a poly-17 film (sweep rate 100 mV s-I) in CH3CN, 0.1 M TBAP in the absence of LiC104, then successively increasing concentrations; 10-4, 2 X 10-4, 3 X 10-4, 4 X 10-4, 5 X 10-4 M, of LiC104. The polymer redox process moves to more positive potentials and diminishes as LiC104 is added. (Reproduced from ref. 25. Copyright 1993, VCH Publishing Group). J A n = 1 only: 19 (x=4, 5) Me,Sn SnMe, n= 1: 20 n= 2:21 Scheme 3 Synthesis of ionoresponsive polythiophenes:26 (i) CuC12 oxidative coupling (ca.20% yield); (ii) BunLi (A not isolated); (iii) [Fe(acetylace ton at^)^] oxidation; (iv) Me3SnCl; (v) 5,5’-dibromo-2,2’-bithiophene, PdO catalyst. designed such that the binding of alkali metal ions by the crown-like oligooxyethylene moieties would alter the degree of inter-ring twisting in the polymer chain (see Fig. 4),and hence alter the colour of the polymer, as well as its redox properties. The potential advantage of the polythiophene acting as a ‘reporter’ of cation binding is that even if only a single binding site is occupied on a given polythiophene chain, the ‘twisting’ effect could be transmitted significantly to neighbouring rings, thus amplifying the response at low concentration of cation.One would predict a blue shift in the n-n* transition on cation binding, as the degree of conjugation is lowered with twisting, and this was indeed observed for these polymers. The magnitude of the shift also showed size selectivity. By analogy P O\ ++ M+M+ --M+M+ n 04 Fig. 4 The principle behind the design of the ionoresponsive polymers of ref. 26. Binding of an alkali metal cation causes twisting of the polythiophene chain away from optimum conjugation. with known binding constants for crown ethers, Na+ is expected to fit best the pentaoxy receptor, and K+ the hexaoxy receptor. The data supported this picture.Also notable was that the shifts were very large even though binding constants, measured for the monomers, were smaller than those found for crown ethers themselves, lending support to the idea that although binding is weak, it affects more of the chain than the immediate bithiophene unit concerned. One disadvantage of these other-wise elegant systems is the length (and overall yields) of the syntheses (Scheme 3), and it is interesting that similar findings have also been reported for regioregular poly[3-oligo(oxy-ethylene)-4-methylthiophene]~.*~ Calixarenes give very high binding constants for alkali metal ions, and as a result of their greater rigidity are also more size-selective than crown ethers. The calixarene-functionalised bithiophene 22 was therefore synthesised, characterised crys-BUt Bu‘ TE7h0 0 30-,-2,Q0 wowo3 22 23 tallographically, and its copolymer with 3,3’-bis(2-methoxy-ethoxy)-2,2’-bithiopheneprepared.26 In this instance, the bithio-phene exhibited considerable twisting in the absence of cation binding (inter-thiophene ring angle 68”), and a red shift was seen for the polymer in the presence of Na+; in this instance, binding of Na+ by the calixarene improved the conjugation of the polymer.Interestingly, the redox potential of the polymer, cast as a film onto interdigitated Pt microelectrodes, increased in the presence of Na+ in spite of this. Moreover, using a bipotentiostat and maintaining a small potential difference between two adjacent Pt microelectrodes overlapped by the polymer film, it was possible to monitor the current flowing as the polymer was redox-cycled, which is related to the conductivity; it was found that the conductivity of the film was almost completely suppressed in the presence of Na+ (Fig. 5).Even though Li+and K+had no significant effect upon either the electronic spectrum of the polymer or its redox potential, they also significantly suppressed film conductivity. It was sug-gested that these results could be accounted for by localisation of charge carriers via electrostatic repulsion from occupied binding sites. In this respect, the in situ monitoring of polymer conductivity provided the best detection route for cation binding. The suppression of film conductivity on cation binding has been termed a ‘chemoresistive’ response.Unlike the behaviour of polythiophenes with pendant crowns, the binding of Na+ was reversible; holding the polymer in its oxidised form resulted in expulsion of the Na+, and the voltammogram and Chemical Society Reviews, 1997, volume 26 253 12 10 8 a6 \ -4-1.1.1.1.1.1.1-1. 0 0.2 0.4 0.6 0.8 EIV Fig. 5 (Bottom) Cyclic voltammogram (drain current, 20 mV offset, 10 mV s-I, 0.1 M Bu4NPF6, CHJCN) of a film of AB copolymer of 22 with 3,3’-bis(2-methoxyethoxy)-2,2’-bithiophenein the absence and presence of 0.5 mM Na+ recorded, using a bipotentiostat, on two interdigitated microelectrodes 10 ym apart, connected by the film. (Top) Current flowing between the microelectrodes, due to the potential difference (20 mV) maintained between them (200mV s-1 0.1 M Bu4NPF6, CH3CN).Note that while the presence of this small concentration of Na+ has some effect on the voltammogram, it completely suppresses the conductivity of the film as determined by the latter technique. Potentials are vs. a silver quasi-reference electrode. (Reproduced from ref. 26. Copyright 1995, American Chemical Society). conductivity response of the original polymer film were restored. Recently, there has been much interest in the complexation of viologen-related molecules (which are electron-poor aromatic molecules) by electron-rich 1,4-dialkoxyaryls incorporated in polyether rings. Interlocked rings (so-called catenanes) can be synthesised utilising the self-assembly of such individual electron-poor and -rich building blocks, and a new sub-field of supramolecular chemistry has developed recently from this finding.28 Thiophenes are also suitably electron-rich; an AB copolymer of 23 with 3,3’-bis(2-methoxyethoxy)-2,2’-bithiophene has been successfully demonstrated as a sensor for the paraquat (4,4’-dimethylbipyridinium) cation using a very similar approach to that described for the alkali metal-sensing polythiophenes.26 Once again the most sensitive form of detection of paraquat binding in the receptor sites of these polymers was monitoring, in situ, the conductivity of cast films using interdigitated electrodes.254 Chemical Society Reviews, 1997, volume 26 Another class of interlocked molecules which has attracted interest are those prepared from reactive 1,lO-phenanthrolines by formation of a tetrahedral metal complex, followed by ring closure with oligooxyethylene chains, to give complexes of the general structure 24.The electropolymerisation of the pyrrole- 24 25 R= -(CH,),-N bearing copper complexes 25 has been used to modify electrode surfaces with similarly interlocked metal coordination sites, the interlocking in this instance being due to polypyrrole cross- linking.’9 That these sites, once formed, are rigidly anchored within the polymer matrix was demonstrated by removal of the copper(1) by treatment of the polymer film with CN- or SCN-, followed by treatment with a different metal ion [e.g.cobalt(n), zinc(11)1; the voltammetry of the film changed to reflect whichever rigid tetrahedral bis( 1,lO-phenanthroline) metal complex was present within the film. Moreover, when the free bis-(pyrroly1)phenanthroline ligands themselves were electro- polymerised, then soaked in solutions of metal ions, no metal ion complexation could be detected voltammetrically. Clearly, a metal ion must be present to preserve the tetrahedral cavity during electropolymerisation for the films subsequently to undergo metal ion exchange. The approach has since been extended to 3-functionalised pyrroles and to pre-formed rotaxane complexes.29 These systems could be utilised for transition metal ion sensing, as the later 3d metal ions all form tetrahedral complexes with the intertwined ligands that have distinctive redox properties which can be distinguished vol- tametrically.Particularly with nickel and cobalt, the redox properties are also interesting in that the rigid enforcement of tetrahedral geometry by the interlocked 1,lO-phenanthroline ligands stabilises the unusual MI oxidation state, and electro- catalytic applications can also be forseen for these modified electrodes. Related work, with polythiophenes, has focussed on more direct electronic coupling between the metallorotaxane receptor site and the conjugated polymer backbone. By combining a preformed cyclo-oligooxyethylene- 1,lO-phenanthroline ligand, the ligand 26 and CuI or Zn” as the ‘ternplating’ metal ion, rotaxane complexes with two pendant bithiophene units were prepared, and these have been electropolymerised to afford alternating quaterthiophene-bpy complex copolymers with threaded cyclic phenanthroline units.30 Nature is still supreme in the area of host-guest chemistry.Various dipeptides and tripeptides, known to be specifically recognised and bound by appropriate proteins, have been attached to pyrrole, via the reaction of pyrrol-3-ylacetic acid with the amine-terminal end of the peptide in the presence of dicyclohexylcarbodiimide.31 Some of the peptides were used with the carboxy terminus in the free acid form, and some as the methoxy esters. The peptide-functionalised pyrroles were electropolymerised and the cyclic voltammograms of the polymers were examined in the presence of varying concentra- tions of the proteins concerned.The selective binding of a given SA 26 27 = R~= 28H R’ = H, R2= 2-thienyl ~1 = R* = 2-thienyl protein to its specific peptide could be detected by a shift to more positive potential and a diminution of current response for the redox process of the polymers. Moreover, once the protein was bound by the peptide, it could subsequently be released (if the peptide was in the free acid form) by electrooxidation of the polypyrrole backbone. This occurred because on oxidation, protons were released from the carboxy terminal groups to balance the positive charge of the polymer backbone; this pH change causes peptide binding to weaken.6 Attaching biomolecules to conjugated polymers, via post-polymerisation modification Many attempts to fix redox enzymes to electrode surfaces using polymers of various kinds have been described. Although the idea of ‘wiring’ a redox enzyme to an electrode via a conducting polymer, as in Fig. 6, is attractive from the point of view of 0 0 0 0 0 / 0 0 0 0 0 0 Metal Polymer Solution Fig. 6 ‘Wiring’ an enzyme (shaded circle) to an electrode via a conducting polymer matrix. The substrate (S) partitions into the polymer from solution and is oxidised by the enzyme to the product (P); the electrons released are shuttled to the electrode by the conducting polymer backbone. sensor construction, the conflicting requirements of preserving enzyme activity while preparing a truly conductive polymer mean that it is extremely difficult to achieve in practice.The difficulties are well illustrated by the following example. Glucose oxidase oxidises fi-D-glucose to &gluconolactone, at the same time reducing O2 to H202. This enzyme has been functionalised by attaching 3-carboxymethylpyrrole to protein surface lysyl (i.e. pendant NH2) groups by amide bond f0rmation.3~ This did not impair enzyme activity. The pyrrole- functionalised enzyme could only be attached to an electrode surface by co-polymerisation with pyrrole itself. Moreover, the voltammetry suggested strongly that the electroactivity, and therefore the conductivity, of the ‘polypyrrole’ is destroyed as the polymer forms, as a consequence of the concentrations of monomers used (1 mM) and the necessity for using buffered media (pH 7) to preserve enzyme activity.Therefore ‘wiring’ as in Fig. 6 has not been achieved-the enzyme is merely anchored to the electrode surface. Nevertheless, the activity of enzyme films formed by covalent anchoring in this way (maintained at +0.7 V vs. SCE) was over six times as high as films formed by simply entrapping glucose oxidase, as a charge-balancing anionic species, within a growing polypyrrole film. If it is desired to prepare a polymer-modified electrode bearing a particularly sensitive or expensive functional group, it may be preferable to attach this after the electrode modification step, and methods for doing this are now being developed.These are particularly appropriate for the attachment of biomolecules to conducting polymers. For example, glucose oxidase has been attached to the surface of conducting polymer films formed by the electrooxidation of N-(4-aminophenyl)- 2,5 -di(2- thieny1)pyrrole and N-(2-aminoethyl)-2,5 -di( 2-thie- nyl)pyrrole, using protein surface carboxy groups to form amides in this instance.33 The activity of the resulting polymer- anchored enzyme electrodes was measured as a function of the number of scans in the RSCV experiment to grow the polymer; the optimum number of scans was around 50. Beyond this, activity fell off, due to diffusion limitations within the polymer. The mechanism was thought to involve formation of H202 and its diffusion through the polymer layer to become reduced at the underlying Pt electrode, rather than charge transport via the polymer film, however.The response time of the electrode was fast as the enzyme was only attached at the polymer-electrolyte interface, rather than dispersed within the film. Polythiophenes have been derivatised similarly. Activated ester-functionalised polythiophene films made by electro-polymerisation of the monomers 27 have been shown to react with the aminoethoxymethyl28 to give polythiophenes bearing the corresponding pendant amidoferrocene groups, using cyclic voltammetry.33 The technique was extended to the attachment of amine-bearing biomolecules. 7 Self-assembled monolayers and conjugated polymers Much of the chemistry covered in this brief review relies on an electropolymerisation step to fabricate a modified electrode.The morphology of electrodeposited conducting polymers, and how well they adhere to the electrode surface, are important questions likely to have a bearing on the utility of polymer- modified electrodes, and it seems appropriate to end this article by discussing a class of functionalised monomer whose purpose is to enable some control to be exercised over the electrode- polymer interface. Interest in the monolayer-modified electrodes referred to in the Introduction revived recently with the discovery that very well-ordered, close-packed monolayers could be fabricated by the adsorption of alkanethiols on to gold electrodes.These are referred to as self assembled monolayers (SAMs) since they form spontaneously if a gold electrode is soaked in a solution of the alkanethiol RSH (or dialkyl disulfide RSSR). Many different redox-active molecules, from simple ferrocenyl groups to organometallic clusters or enzymes, can thus be attached a fixed distance from a gold surface by an alkylthio pendant group of known length. Several studies have been reported recently on the behaviour of pyrrole- or aniline-terminated SAMs; thiophenes are not useful here, as their oxidation potentials are well positive of that for the oxidative desorption of thiols from gold surfaces ( > ca. +1.3 V vs. SCE). Simultaneous in situ ellipsometry and electrochemical quartz crystal microbalance (EQCM) measurements have been used to examine the growth of polyaniline films on gold electrodes, either bare, or pre-coated with a SAM of either 4-aminothiophe- no1 (4-ATP), or other alkane- or arene-thi0ls.3~ Ellipsometry provides a measurement of the complex refractive index and thickness of surface films.EQCM, in its simplest manifestation, measures changes in the mass of a film on an electrode surface as a function of potential. The results showed that on bare gold, or on gold coated with alkane- or arene-thiols, an optically diffuse film (i.e. one containing much solvent) is deposited in the early stages of growth. On gold pre-coated with 4-ATP, Chemical Society Reviews, 1997, volume 26 255 optically much denser films formed, although the amount of material deposited as a function of time was not greatly different.Subsequent voltammetry and AC impedance meas- urements showed that films grown on 4-ATP-coated gold switched more rapidly between the conducting and insulating states. The 4-ATP film evidently aids the process of nucleation, the initial event in the deposition of conjugated polymer films. It may also, as evidenced by the switching experiments, impose some order on the initial film formation process. fi +1.4 +0.7 0.0 EIV Fig. 7Cyclic voltammograms of (A) gold electrode coated with a SAM of o-pyrrol-l-ylhexanethiolscanned between 0 and +0.75 V (vs. sodium standard calomel electrode); (B)between 0 and +1.1 V showing the charge due to the irreversible oxidation of the pyrrole moieties (shaded area); (C) between 0 and +0.75 V after the potential excursion shown in B, showing the charge due to the electrochemistry of the thiol-anchored polypyrrole strands formed in B (shaded area).Area of electrode 1.96 X 10-3 cm2, scan rate 100 mV s-1, current scale S = 5 PA cm-2 for (A)and (C), 13 pA cm-2 for B. (Reproduced from ref. 35. Copyright 1994, American Chemical Society). Similar conclusions were reached when poly-3-ethylpyrrole was grown on gold pre-coated with monolayers of 29, 30 or 31.35 Much more adhesive, denser and more conductive films were produced on the SAM-coated gold than on bare gold, and poly-3-ethylpyrrole grown on gold coated with a simple alkanethiol was similar to that grown on bare gold.Again, this suggests that the presence of the pyrrole-terminated SAM results in the growing polypyrrole becoming chemically bonded to the surface, as the terminal pyrrole is involved in the electropoly merisation. H H H 29 30 31 Attempts were also made to electrooxidise these pyrrole- terminated SAMs in the absence of additional pyrrole monomer 256 Chemical Society Reviews, 1997, volume 26 in solution. Although the SAMs of 29-31 became irreversibly oxidised at the expected potentials, the products were electro- inactive, suggesting that the electrogenerated pyrrole radical cations react with trace nucleophiles rather than undergoing coupling to form polymer. However, similar monolayers of N-(o-thioalky1)pyrroles can be electropolymerised, to give an electroactive, polypyrrole-like material (see Fig.7).35 Pre-sumably, the pendant pyrroles knit together to form strands of polymer held parallel to the surface by the alkanethiol tethers. Evidence for this comes from competitive displacement experiments. Alkanethiol SAMs undergo exchange with thiols in solution. When the unpolymerised N-(o-thioalky1)pyrrole SAMs were treated with solutions of ferrocenylalkanethiols, they were displaced; the increasing surface concentration of ferrocenylalkanethiol as a function of time could easily be monitored using the surface-anchored ferrocene redox chem- istry. However, once the pyrrole groups had been electro- oxidised, the ferrocenylalkanethiol failed to displace the ‘polypyrrole-knitted’ SAM.The use of monolayers of anchored heterocycle monomers to fix conjugated polymer layers to electrode surfaces predates the discovery of alkanethiol SAMs, however. As long ago as 1982, when polypyrrole was being examined as a protective coat for n-doped silicon in photoelectrochemistry, it was found that it did not adhere well to the surface. Silicon is inevitably covered by a thin layer of oxide. The surface was therefore first treated with 1-pyrrol-1-yl-3-(trimethoxysilyl)propane;trialkoxysilanes are known to anchor firmly to oxides by reaction with surface hydroxy groups with elimination of alcohol. Subsequent electropolymerisation gave a polypyrrole layer which was so adherent that it could not be peeled off the surface with adhesive tape.35 8 Conclusions and prospects This review commenced with conjugated polymers incorporat- ing metal complexes. The range of metal complexes which has been anchored to electrode surfaces by the electropolymerisa- tion of a pendant pyrrole group has been extended over the last few years to include coordinatively unsaturated or organome- tallic species,17-20,22 which are likely to prove more interesting as electrocatalysts. Developments in applications, such as in sensing, may be forthcoming for these systems, but these will probably be in ‘niche’ areas.That there is a role for well- designed, chemically-prepared conjugated materials in sensor applications has been well-demonstrated for calixarene-, crown ether- and rotaxane-containing polythiophenes.26 Chemical synthesis avoids some of the additional variables inevitably encountered in electropolymerisation experiments, but at a cost-the synthetic effort involved is substantial. Another important pointer from this work is that changes in the voltammetry of polymer films may not be the most sensitive parameter for detecting a response to binding by the polymer; conductivity and optical properties offer other possibilities.It may be useful to see if the same concepts can be applied to the detection of enzymes or oligonucleotides using polypyrroles bearing short polypeptide or RNA side-chains re~pectively.~~ Finally, the use of self-assembled monolayers of pyrrole- terminated alkanethiols offers a good way of preparing extremely thin yet dense conjugated polymer films by electro- polymerisation.It will be interesting to examine such processes using in situ techniques such as scanning tunnelling micros- copy, to see if more ordered polymer chains also result, and to apply this idea to the preparation of thin, functionalised conjugated polymer films. 9 Acknowledgements I thank the EPSRC, the Nuffield Foundation and the Royal Society for funding, and co-workers whose names appear in some of the references for their hard work, collaboration and valuable discussions. 10 References 1 T A Skotheim, Handbook of Conducting Polymers, Marcel Dekker, New York, 1986 2 H D Abruiia, Coord Chem Rev, 1988,86, 135 3 J Roncali, Chem Rev, 1997, 97, 173, and references therein 4 J Roncali, Chem Rev , 1992, 92, 71 1, and references therein 5 P N Bartlett, L Y Chung and P Moore, Electrochim Acta, 1990,35, 6 Comprehensive Heterocyclic Chemistry, ed A R Katntzky, C W Rees and C J Drayton, Pergamon, Oxford, 1984 7 D Delabouglise, J Roncali, M Lemaire and F Gamier, J Chem SOC , Chem Commun , 1989, 475 and references therein 8 T Inagaki, M Hunter, X Q Yang, T A Skotheim and Y Okamoto, J Chem SOC , Chem Commun , 1988, 126 9 R D McCullough and R D Lowe, J Chem SOC Chem Commun , 1992,70, A Iraqi, J A Crayston and J C Walton, Covalent binding of redox-active centres to preformed regioregular polythiophenes, patent no UK 96016340, 1996 10 M Berggren, 0 Inganas, G Gustafsson, J Rasmusson, M R Anders-son, T Hjertberg and 0 Wennerstrom, Nature, 1994, 372, 444 11 P A Christensen and A Hamnett, Techniques and Mechanisms in Electrochemistry, Chapman and Hall, London, 1994 12 D Curran, J Grimshaw and S D Perera, Chem Soc Rev , 1992, 20, 39 1 13 F Bedioui, J Devynck and C Bied-Charreton, Acc Chem Res , 1995, 28,30 14 A Hamnett, P A Christensen and S J Higgins, Analyst (London), 1994, 119, 735 and references therein 15 P A Chnstensen, A Hamnett and S J Higgins, J Chem Soc ,Faraday Trans, 1996, 92, 773 and references therein 16 S Gould, T R O’Toole and T J Meyer,J Am Chem SOC , 1990,112, 26 and references therein 17 S Cosnier, A Deronzier and J F Roland, J Electroanal Chem , 1990, 285, 133, C Lopez, J -C Moutet and E Saint-Aman, J Chem Soc , Faraday Trans, 1996,92, 1527 18 W F D Giovani and A Deronzier, J Chem SOC , Chem Commun , 1992, 1461, A R Guadalupe, X Chen, B P Sullivan and T J Meyer, Inorg Chem , 1993,32, 5502 19 S Chardon-Noblat, S Cosnier, A Deronzier and N Vlachopoulos, J Electloanal Chem , 1993,352, 213 20 H C Y Bettega, J C Moutet and S Tingry, J Electroanal Chem, 1995,391,51 21 S J Higgins, M K McCart, M McElhinney, D C Nugent and T J Pounds, J Chem SOC ,Chem Commun ,1995,2129and references therein, S J Higgins, M K McCart and T J Pounds, unpublished work 22 M N C Dunand-Sauthier, A Deronzier, J C Moutet and S Tingry, J Chem SOC,Dalton Trans, 1996,2503 23 P N Bartlett, A C Benniston, L Y Chung, D M Dawson and P Moore, Electrochim Acta, 1991, 36, 1377 24 H K Youssoufi, M Hmyene, F Garnier and D Delabouglise, J Chem Soc , Chem Commun, 1993, 1550 25 P Bauerle and S Scheib, Adv Muter, 1993, 5, 848 26 M J Marsella and T M Swager, J Am Chem Soc , 1993,115,12214, M J Marsella, R J Newland, P J Carroll and T M Swager, J Am Chem Soc, 1995, 117, 9842, M J Marsella, P J Carroll and T M Swager, J Am Chem SOC, 1995,117,9832 27 I Levesque and M Leclerc, Chem Muter, 1996,8,2843 28 A C Benniston, Chem SOC Rev, 1996,25,427 29 G Bidan, B Divisia-Blohorn, M Lapkowski, J M Kern and J P Sauvage, J Am Chem SOC,1992, 114, 5986, J M Kern, J P Sauvage, G Bidan, M Billon and B Divisia-Blohorn, Adv Muter , 1996,8,580 30 S S Zhu, P J Carroll and T M Swager, J Am Chem SOC, 1996,118, 8713 3 1 F Garnier, H K Youssoufi, P Srivastava and A Yasser, J Am Chem Soc, 1994,116, 8813 32 B F Y Yon-Hin and C R Lowe, J Electroanal Chem, 1994, 374, 167 33 H Rockel, J Huber, R Gleiter and W Schuhmann, Adv Muter , 1994, 6,568, P Bauerle, M Hiller, S Scheib, M Sokolowski and E Umbach, Adv Muter, 1996, 8, 214 34 E Sabatini, Y Gafni and I Rubinstein, J Phys Chern , 1995,99, 12305 and references therein 35 C N Sayre and D M Collard, Langmuir, 1995,11,302and references therein, R J Willicut and R L McCarley, J Am Chem SOC, 1994, 116, 10823 Received, 28th April 1997 Accepted, I9th May I997 Chemical Society Reviews, 1997, volume 26 257

ISSN:0306-0012    

DOI:10.1039/CS9972600247

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

John R.Owen Department of Chemistry, University of Southampton, Southampton The market for lithium batteries is undergoing a rapidexpansion as new applications demand higher densities of energy and power storage. Simple theoretical estimates show that lithium and lithium ion cells can reach specific energies of 880 and 500 W h kg-1 respectively. With an electrolyte conductivity above 3 x 10-4 S cm-1 and thickness below 0.01 cm, a power density of 300 W dm-3 can be obtained without excessive energy losses. Diffusion in porous or polymer composite electrodes is enhanced by an inter-penetrating electrolyte provided the electrode particles are small. Batteries using transition metal oxide positive elec- trodes and carbon negative electrodes are expected to give practical specific energies up to 180 W h kg-1 including packaging and other essential additional materials in the near future.1 Introduction 1.1 Market requirements At the time of writing there are many new electrical technolo- gies which require rechargeable batteries, for example: information technology-e.g. camcorders, digital cameras and personal computers; portable machine tools-e.g. power drills, saws, sanders etc.; electric traction-e.g. various vehicles from golf carts to electric buses. Apart from a rapidly expanding market, what these have in common is that the ultimate performance is now restricted by the power capability and energy storage capacity of the battery. The performance of the product will be determined by the specific power (W kg-I) or power density (W dm-3) of the battery. A high specific power is required to meet the power demand of the product, within the battery mass or volume constraint, whichever is limited.The specific energy (W h kg-1) or the energy density (W h dm-3) of the battery, along with the power consumption of the product, determine the period of useful use before recharge is required. In addition, a John Owen graduated in chemistry at Imperial College, London, where he subsequently obtained his PhD for studies of the crystal growth of organic electro-optic materials. After a brief period working on solar energy in Iran he returned to Imperial College in 1979 as Wolfson Fellow. In 1984 he began lecturing in electrochem- istry at the University of Salford and founded the Solid State Electrochemistry Group.Since 1991 the group has been at the Department of Chemistry at Southampton, specialising in the study and application of insertion electrodes and polymer electrolytes (http:l/www.soton.ac.ukl -jro). UK SO17 IBJ reasonable lifetime is expected of the battery; this is usually defined in terms of the cycle life, i.e. the total number of charge- discharge cycles which may be obtained before a significant degradation occurs in the energy and power values. Specific energy, power and cyclability values of most commercial rechargeable batteries have been of secondary importance to the unit cost; for example the relatively cheap lead-acid system delivers less than 50 W h kg- I and the cycle life is low for deep cycling.Furthermore, the new environmen- tal legislation which will require recycling of heavy metals is now beginning to increase the real cost of batteries containing lead or cadmium. Some main targets for a high energy density battery have been described as:' Specific energy-180 W h kg-1; Energy density-360 W h dm-3; Cycle life- >500. The situation is therefore quite different from that of batteries for automotive engine starting, where improvements in the batteries offer only a marginal benefit. In the case of portable computers, tools and electric vehicles, any improvements in the battery will translate directly into increased profit. High-energy and -power batteries are therefore a unique opportunity for investment.1.2 New developments As a consequence of the situation described above, a rapid expansion in research on rechargeable batteries has occurred over the last two decades. Traditional batteries such as lead- acid and nickel oxide-cadmium have benefited from physical design improvements to improve energy and power density and incorporate new features such as overcharge protection. Con- currently, some new systems have developed which are based on entirely new chemistry. Perhaps the greatest conceptual leap in the battery design was the use of non-aqueous electrolytes to allow the use of lithium or sodium metal as the negative electrode. This has two effects: first, the mass and volume of the negative electrode (e.g.Li) can be very small in comparison with the positive electrode (e.g. Mn02), resulting in almost double the specific charge (A h kg-1); secondly, the energy per unit charge stored (i.e. the potential difference) is also doubled, or almost tripled in comparison with the nickel oxide-cadmium system. A comparison between some traditional and novel battery systems is shown in the Ragone plot of Fig. 1. It may be seen that lithium batteries occupy the prime position. 1.3 Lithium batteries A schematic diagram of a single cell of a rechargeable lithium battery is shown in Fig. 2(a). Compared with some aqueous systems, the electrochemistry should, in theory, be refreshingly simple; the electrolyte takes no part in the reaction except for conveying the electroactive lithium ions during discharge from a high energy state in the negative electrode to a low energy state in the positive electrode while the electrons pass through the external circuit with a release of energy.The opposite reaction occurs on charge, so that rechargeability depends on the reversibility of the reactions at the electrodes. Chemical Society Reviews, 1997, volume 26 259 100 2 Theory In the following sections we shall review some candidate materials with reference to the properties which are required for high energy and high power. F 50 Ni-Metal Hydride W .-0 .ead-aci d 'c0B cn 10 20 40 100 200 400 1000 Specific PowerNV kg-' Fig.1 A Ragone plot of energy and power density for various battery systems, showing the deterioration of energy density with increasing power demand Fig. 2 Schematic diagrams of (a) a lithium cell, with a lithium metal negative and an insertion positive electrode, (b)a lithium ion cell which exchanges lithium between two insertion electrodes At the positive electrode, reversibility is a consequence of the use of an insertion electrode material.2 This is a solid which can incorporate the electroactive material into a solid solution with a wide stoichiometry range as in the example of titanium disulfide [eqn. (l)]. 6Li++ 6e-+ LiyTiS2 Li, + hTiS2 (1) The electrode undergoes a reversible topotactic redox reaction, meaning that the electrode material acts as a host structure which accommodates guest ions and electrons without destruction of the lattice.Many compounds have been selected for the positive insertion electrode and these will be reviewed below. A second type of lithium cell, the lithium ion or rocking-chair cell is shown schematically in Fig. 2(6). Here the negative electrode is also composed of an insertion electrode, with advantages of dimensional stability and improved chemical stability as detailed in a later section. No lithium metal need exist in this cell-the lithium is always held as a guest in one of the electrodes depending on the state of charge. However, the specific charge is decreased, although the overall specific energy obtained is still impressive and an excellent cycle life may be obtained.The literature on lithium batteries is already extensive, and the reader may find more detailed information in a recently published book on lithium batteries,3 a more general handbook of batteries," and a series of lithium battery conference proceedings.5 260 Chemical Society Reviews, 1997, volume 26 2.1 The cell potential The maximum available energy is simply the free energy of reaction, AG. Therefore, high energy is a consequence of the choice of electrode materials. In the ideal case where there are no energy losses, the cell potential can be defined as in eqn. (2), negative electrode -ppositive electrode )IFE = (PLI L1 (2) where plLI indicates the values of the chemical potential of lithium metal at the electrode surfaces and F = Faraday's constant, 96500 C mol-1.From the point of view of maximising the cell potential, lithium metal as the negative electrode with pL1= 0 is of course the best choice. In the lithium ion cell, small negative values of pLIcan be obtained using host materials where lithium exists at an activity close to that in the metallic state, e.g. graphite, for which the electrode potential is no more than 0.25 V positive with respect to lithium metal over the range 0.1 < y < 1 for Li,C6. Conversely, in a positive electrode a highly negative value of pLIis obviously desirable so that a highly oxidising host is required. Lithium nickel oxide, Li,Ni02, has a stoichiometry range of 0 < y < 1 with an electrode potential of between 4.2 and 3.5 V vs.lithium metal. Therefore, with a suitable choice of both materials, a minimum cell potential of about 3 V at the end of discharge is obtainable. 2.2 Theoretical cell capacity and specific energy The theoretical cell capacity in A h g-1 or A h cm-3 can be estimated from the stoichiometry ranges and molar masses of the electrode materials. Thus LiC6 and Li,Ni02 are each required in quantities of about 80 g per Faraday of charge stored as compared with only 7 g of lithium metal. The theoretical maximum capacities of the Li/Ni02 and LiC6/Ni02 cells are therefore about 300 A h kg-I and 170 A h kg-1 respectively. With a useful cell potential of 3 V the specific energies become 880 W h kg-1 and 500 W h kg-1.2.3 Power losses in the electrolyte High power requires fast transport of lithium ions, which involves energy losses at several points in the cell. The most obvious loss is the Ohmic loss in the electrolyte, which is Fig. 3 Calculation of the Ohmic loss in the electrolyte according to eqn. (3) calculated here for the cell shown in the scheme of Fig. 3 [eqn. (3)],where i = the cell current; I = the electrolyte thickness; A AV(ohmic loss) = iR = il/Ao = (i/vk)I2/o (3) = the cross-sectional area of the electrolyte perpendicular to the current path; o = the electrolyte conductivity; k = the ratio of the separator electrolyte volume to the cell volume; v = the cell volume. [Lit] 0 distance along cell b negative electrolyte positive electrode electrode 0 .---_.PLi -P0 Li APL~at equilibriumIw-distance dong cell b Fig. 4 Concentration polarisation of the electrodes is a consequence of changes in the surface concentrations of lithium during discharge. The difference between the chemical potentials of lithium at the two electrode surfaces, A~L~,determines the cell potential according to V = ApL,/F -iRe1ectrolyte. The power density, P, is obtained by multiplying the current per unit cell volume by the cell voltage, V [eqn. (4)]. P, = V(i/v) = VkoAVI12 (4) Taking, for example, V = 3 V, k = 0.3, o = 3 X S cm-I, AV = 0.1 V, 1 = 0.01 cm, we would have P, = 300 W dm-3. The calculation emphasises the needs for a high conductivity and minimisation of the electrolyte thickness as far as possible.2.4 Concentration polarisation in the electrode Discharge at a finite rate causes depletion of lithium at the surface of the negative electrode and accumulation at the positive electrode as shown in Fig. 4. Since it is the composition of the surface of the electrodes which determines the cell potential, the concentration changes can result in premature discharge if the concentration gradients become large. The temporary loss in cell potential due to the difference between the surface and average compositions of the electrodes is termed concentration polarisation. The effect can only be relieved by efficient transport of lithium in the electrodes. Lithium transport in the electrodes has generally been treated as simple diffusion in a homogeneous phase.The minimum time taken for the discharge of a single electrode at constant potential (vs. a reference electrode) can be estimated from eqn. (5),6 where Mo is the initial lithium content; Mt is Mt -M, = 1--exp{ (5)8 $}M, -M, n2 the amount of lithium in the electrode at time t; M, is the amount of lithium in the electrode at equilibrium; D is the diffusion coefficient of lithium; 1 is the electrode thickness. A 90% discharge would be obtained, for example, in a time t = 0.85 12/D. Diffusion coefficients of lithium in pure electrode materials vary by many orders of magnitude, and much early work emphasised the need for insertion electrodes with diffusion coefficients of the order 10-8 cm2 s-1 in order that electrodes of about 0.01 cm thickness could be discharged in about one hour.2.5 Porous and composite electrodes In reality, electrodes are formed into porous microcrystalline structures which can be permeated by a liquid electrolyte or formed into composite electrodes by blending with a polymer electrolyte7 (Fig. 5). In these circumstances, problems due to Fig. 5 (a)A composite electrode structure reduces the length of the diffusion path within the electrode material to about half the particle radius. Mass transport is provided by the interpenetrating electrolyte. (b)An effective diffusion coefficient for the structure is given by o'/C,. concentration gradients within the electrode particles them- selves can be alleviated by reducing the particle size, typically to 10-50 pm in diameter. The problem which remains is equalisation of concentration across the electrode thickness, which generally occurs by lithium ion conduction and diffusion in the intergranular electrolyte phase.Although the electric potential, @, is almost constant in the regions which have a good electronic conductivity, an electric field a@/& exists in the intergranular electrolyte because of the gradient in the inter- facial potential arising from a gradient in lithium chemical potential in the surrounding electrode grains [eqn. (6)]. a@/ax= -(apLl/ax)/F (6) Provided that the particle size is sufficiently small the effective diffusion coefficient may be shown to be given by eqn. (7), D = d/Cv (7) where o' is the effective ionic conductivity due to the electrolyte in the composite and C, is the pseudocapacitance per unit volume of electrode material as shown by the equilibrium variation of electrode potential with inserted charge per unit volume [eqn.(S)], where Vele&.& is the electrode volume. Cv = (m/dQ)-l/Velectrde (8) As an example, the value for the pseudocapacitance can be estimated for a non-graphitic carbon composite electrode; assuming a capacity of 1000 C ~m-~available over a potential range of, say 1 V, we obtain a capacitance of 1000 F cm-3. The effective ionic conductivity is reduced from the value for the electrolyte according to the volume fraction and another factor due to the tortuous conduction path.With an effective ionic conductivity of 10-5 S cm-1 the effective diffusion coefficient for the composite electrode would be around lo-* cm2 s-l. The above calculation emphasises the need for a highly conducting electrolyte within the electrode regions as well as the interelectrode gap. It also shows that high power density also requires a laminate structure with electrodes as thin as 0.01 cm. To conclude this section, the estimations above may be related to the Ragone plot of Fig. 1. Lithium batteries may be expected to reach above 200 W h kg-1 and 100 W kg-1 provided the following conditions are met: an electrode couple with a potential difference of at least 3 V throughout discharge; electrode materials with (molar) masses below 100 g per Faraday of charge stored; electrolyte conductivity at least 3 x 10-4 S cm-1; a thin laminate structure with electrode and electrolyte thicknesses around 0.01 cm.Chemical Society Reviews, 1997, volume 26 261 3 Non-aqueous electrolytes 3.1 Some alternative approaches A prime consideration for an electrolyte in a battery is chemical stability to both the electrodes in the charged state. With a lithium electrode this is a difficult problem, although in some rare cases thermodynamic stability in contact with metallic lithium is indeed possible, e.g. the lithium halides. One of the first lithium cells to be manufactured was the lithium-iodine pacemaker cell in which the electrolyte, lithium iodide, is formed in situ at the iodine positive electrode surface during discharge [eqn.(9)]. 2Li++ 2e-+ I2 +2LiI (9) Although this is an ingenious scheme for the prevention of self discharge, lithium iodide is a brittle solid at room temperature and has a low conductivity, about S cm-1. In order to achieve useful rates of discharge in the commercial devices the lithium surface is first coated with poly(2-vinylpyridine) to increase the conductivity of the lithium iodide, once formed, to a maximum of about S cm-l. Even so the applications of this cell are limited to those requiring very low discharge rates, so that rechargeability is not required. A high temperature rechargeable cell has been developed based on molten lithium aluminium chloride as the electrolyte.9 Because the melting point of the salt is higher than that of lithium metal, a lithium-aluminium alloy is used as the negative electrode.The insertion electrode FeS2 is used as the positive electrode, so that the cell reaction is given by eqn. (10). LiAl + FeS2+LiFeS2 +A1 (10) Solid electrolytes other than LiI have also been considered. For example, a lithium conducting glass has been used as the electrolyte in a small thin film cell.l0 However, a general problem with brittle solid electrolytes is that on cycling large strains occur at the electrode-electrolyte interface which cause disintegration of the structure if the film thickness exceeds a few vm. 3.2 Organic solvents Most viable rechargeable lithium cells resort to electrolytes which are composed of solutions of appropriate lithium salts in organic solvents which are liquid at room temperature.The solvent should be aprotic, as active protons react with lithium to give hydrogen gas, e.g. acids and alcohols are prohibited [eqn. (1111. 2ROH + 2Li +2ROLi + H2 (11) No organic solvents have been reported to be thermodynam-ically stable against lithium metal. However, varying degrees of kinetic stability have been reported in aprotic solvents. In addition to stability, a high conductivity is required, which means that lithium ions must be mobilised in solution and separated from associated anions. Hydrocarbons, although quite stable to lithium, can neither dissolve nor dissociate lithium salts effectively-these properties require groups which can solvate the lithium ion (a Lewis acid) through the acid-base reaction in eqn.(12). Li+X-+ n(So1vent)+Li+ (Solvent), + X-(12)Lewis base The solvent basicity may be quantified by the donor number." A high relative permittivity also aids the separation of ion pairs. As well as a high degree of salt dissolution and dissociation, a low viscosity is also required for high conductiv-ity. Some values of these properties for common solvents are given in Table 1. After more than 25 years of study, involving many linear and cyclic ethers, linear and cyclic alkyl carbonates and other esters, a few systems have emerged as having the optimum combina-tion of high conductivity, stability, low viscosity, low cost and Table 1 Properties of some organic solvents used in electrolytes12 Relative Solvent Mp,tl"c permittivity, E Donor number Viscosity/ CP 2-Methyltetrahydrofuran -137 6.2 18 0.47 Diethyl carbonate -43 2.3 15 0.75 Diethyl ether Hexamethylphosphoric -116 4.3 19 0.24 triamide 7 30 39 3.2 Ethylene carbonate 36 95 16 1.9 (at 36 "C) low toxicity.Suitable mixtures can be made which combine the favourable properties of the constituents and modifications can be made to suit particular system requirements such as low-temperature operation. 3.3 New lithium salts The choice of lithium salt is also important. Early work centred on lithium perchlorate as a salt which dissociated readily and showed surprisingly good kinetic stability against lithium metal. Unfortunately, however, certain conditions led to unpredictable and catastrophic breakdown with explosive consequences in prototype batteries so that perchlorates are now avoided in most products destined for public use.Substitute salts have been found in the hexafluoroarsenate and hexa-fluorophosphate, which form solutions with conductivities around S cm-l . The former is now regarded as being too toxic for general use whereas the latter has been, at least temporarily, accepted despite problems of stability against dissociation into LiF and PF5. The continuing search for stable and easily dissociated salts has promoted research into new anions based on perfluoroal-kanesulfonates. The first of these, lithium trifluoromethane-sulfonate, abbreviated to lithium triflate, gave modest properties and is now challenged by specifically designed salts such as the 'TFSI' and 'methide' salts.0 F Li+ -C-F]I F Li TFSI Li+ Li Methide 3.4 Polymer electrolytes A liquid electrolyte must be contained in a porous solid or elastomeric separator which prevents direct contact between the electrodes. A stack pressure is generally applied to ensure good interparticle contact within the electrode material while dimen-sional changes in the electrode particles occur during cycling. Problems arise because separators containing liquid electrolytes 262 Chemical Society Reviews, 1997, volume 26 tend to dry out, particularly with gas evolution during the first cycle.Containment of the liquid electrolyte in a separator can also cause problems due to non-uniformities in the stack pressure and the current path. The problems of mechanical mobility in electrolytes are offered a solution through the use of polymer electrolytes with elastomeric qualities. Armand13 described two ways in which elastomeric and electrolytic properties could be combined. The first and most elegant method was to introduce a solvating character into the polymer molecule. Poly(ethy1ene oxide), PEO, also named poly(oxyethylene), was found to solvate most metal salts with easily dissociatable anions to form elastomeric solutions with conductivities up to S cm-1 above the melting point of 65 "C.The field of polymer electrolytes has developed extensively since Armand's first work and polymeric solutions with lower melting points have been subsequently found by the depression of melting point and suppression of crystallisation using the following strategies: use of high concentrations of salts with large anions, e.g. Li TFSI; randomisation of chain sequences by introduction of oxy- methylene units; 14 introduction of solvating groups into the side chains of low melting and low glass transition polymers such as poly- (phosphazene);15 addition of liquid plasticisers;16 cross-linking, which also improves mechanical strength. 17 Considering that one of the major advantages of a polymer electrolyte is that it presents a mechanically stable barrier between the electrodes, some efforts have concentrated on the second type of conducting elastomer suggested by Armand, the gelled polymer electrolyte.This is essentially a liquid electro- lyte of the type described above which is immobilised in a polymer matrix which does not necessarily have inherent solvating properties. Another important area under development is that of polymer electrolytes having a transference number of one for the lithium ion. Progress in this direction has been made by binding the counterion to the polymer chain. However, the conductivities obtained are disappointingly low, possibly because the triple ions and higher aggregates which are normally responsible for high conductivities in liquid electrolytes are also immobilised in this approach.4 Positive electrode materials 4.1 Models for the electrode potential It has already been stated that the potential of an insertion electrode with respect to lithium metal is -pL, /F. The value of the chemical potential of lithium may be written in terms of the chemical potentials of its constituents, the lithium ion and the electron [eqn. ( 13)]. PL1 = PL1+ + Pe-(13) The two terms are illustrated in Fig. 6. The first term represents solvation of the lithium ion by the host lattice. A Lewis base character of the environment of the lithium ions, as provided by an oxide lattice, for example, is advantageous here. However a greater variation over different materials is caused by the electronic term which contributes to a negative chemical potential through a large negative energy EFof an electron at the Fermi energy, with respect to the vacuum level [eqn.(14)]. In qualitative chemical terms this simply means the electrode should be a strong oxidising agent. As the cell discharges, the average concentration of lithium in the positive electrode increases. Two contrasting situations are shown in Fig. 7. In the first case shown in Fig. 7(a),lithium is inserted to form a single- Fig. 6 Ionic and electronic contributions to the electrode potential. Ape-depends mainly on the Fermi energy of host structure, whereas ApL,+vane$ with lithium ion site occupancy according to p = pe + ky + RTln bl(ymax -y)] (see text).phase solid solution so that the electrode potential falls gradually as the chemical potentials of both ions and electrons are increased. However, within the phase stability region the increase in Fermi energy due to electron population is largely compensated by a lowering of the band energy due to screening of the nuclear charge.*8 The most significant change in the equilibrium potential with composition is therefore due to the change in the chemical potential of the lithium ions [eqns. (1 3, (16)], where EO and pe are standard potentials due to the E = Eo -ky/F -RT/F In Iy/(ymax-y)] guest-host interaction, ymaxis the maximum value of y allowed by the number of available lattice sites; k represents the repulsion between lithium ions in close proximity, R In IY/(ymax)] is the configurational entropy of the inserted ions.(The configurational entropy term applies only to the ions if electrons are delocalised; a similar term for electrons may be added in the case of localised electronic states). The simple model breaks down when the electron energy levels are almost filled and a new level or band must be occupied, or when the given lithium ion sites are almost filled, and an alternative structure becomes energetically favourable. In this case [Fig. 7(b)]the inserted lithium is partitioned between two phases. The discharge then proceeds by growth of the concentrated phase and a near-equilibrium discharge occurs at an almost constant potential until one phase is consumed.4.2 Estimations of electrode capacity Whether discharge occurs via a single phase or a succession of phases, the capacity of an electrode over an acceptably small potential range has been estimated as follows for ternary transition metal compounds Li,MX,.19 The following assump- tions were made: the oxidation state of the host metal changes by no more than one; the anions, X, are close-packed to give one octahedral cation site per anion. In this case the guest ion sites for Li+ are the octahedral sites which remain after occupation by M. phase boundary motion I I Fig. 7 Near-equilibnum potential-composition relation and composition- distance profiles (schematic) during discharge of a positive electrode: (a) without and (b)with phase conversion on insertion Chemical Society Reviews, 1997, volume 26 263 Following this argument it may be seen that MX has no available sites for lithium and therefore has no reversible capacity.Discharge occurs via destructive phase transformation to give the metal and a binary lithium compound [e.g. eqn. (1711. 2Li + CuO 4Liz0 + Cu MX3 would seem to have a high capacity due to the large number of vacant octahedral sites; however, the formation of Li2MX3 requires a double oxidation state change with a correspondingly high reduction in potential over the discharge range. Also, at high ratios of lithium to the transition metal, the structure can become unstable with respect to the irreversible precipitation of the binary lithium compound [e.g.eqn. (18)]. LiMX3+ 2Li +Li2X + LiMX2 The maximum specific capacity (stored charge/mass) is therefore expected for MX2 where reversible insertion reactions may occur according to eqn. (19). yLi + MX2 f LiyMX2 (0 < y < 1) (19) The argument presented above seems to hold for the smaller transition metal oxides which are well described by oxide close packing. However, capacities greater than one lithium per transition metal have been reported for layered and open framework structures. For example, the hexagonal phase of W03, stabilised by a small amount of added sodium oxide, has been reported to accommodate up to 2 Li per W.20 Un-fortunately, the additional capacity is of limited value because the potential decreases to below 2 V as the oxidation state of W approaches +4. Nevertheless, this finding challenges the general assumptions made above. To maximise the potential of a positive electrode the electron energy must be considered.Generally, a low Fermi energy requires a very high oxidation state of an early transition metal, e.g. V5+, or a moderate oxidation state of a late transition metal, e.g. Ni4+, where poor d-electron screening of the nucleus increases the effective positive charge. 4.3 Inorganic electrodes designed for high energy Titanium disulfide was one of the first compounds suggested for use as a lithium insertion electrode. Although insertion proceeds at rather low potentials, between 2.4 and 1.8 V vs.lithium metal, its redeeming features include: a good electronic conductivity over the range 0 < x < 1 in Li,Ti S2; a high diffusion coefficient for lithium ions; a reasonably low mass and volume per Faraday of charge stored; an excellent reversibility of lithium insertion. A substitution of a later transition metal for Ti and an increase in the oxidation state may be expected to increase the potential. However, a consideration of the electron energy levels involved shows that high oxidation states of the metal cannot be accessed without oxidation of the sulfide ion to disulfide, higher sulfides and eventually sulfur. In the case of a transition metal oxide, however, very high oxidation states of the metal can be achieved where the Fermi energy is highly negative.However, a thermodynamic limita- tion occurs due to the possibility of oxygen evolution by oxidation of the oxide ion. The equilibrium reaction of lithium in solution in the ternary oxide with oxygen gas can be written as in eqns. (20), (21). 2Li(M02) +4 02 *Liz0 (20) AG = AGO (Li20) + p(Li20) -2 p(Li) -$ p(02) where AGO (Li20) = ca. -600 kJ mol-1 At equilibrium, AG = 0, ~(02)= 0 and p(Li) = 0.5 p(Li20)+ 0.5Af GO(Li20). 264 Chemical Society Reviews, 1997, volume 26 This equation means that given an almost continuous distribution of electron energy states and lithium sites, the minimum potential, corresponding to saturation with Li20, would be about 3.1 V. The additional potential available in an unsaturated case can be estimated by realising that the change in p(Liz0) during insertion is mainly due to the Li+-Li+ repulsion term ky.This varies typically by about 0.5 V over the range 0 < y < 1 in simple compounds and hence the potentials of such oxides approach 3.6 V for low y values.An ingenious method of simultaneously raising the electrode potential beyond the thermodynamic prediction and increasing the discharge capacity is via synthesis of a precursor to the electrode material in the discharged state, followed by electro- chemical extraction of lithium2' at low temperature, where the electrode may be kinetically stable against oxygen loss. This was reported for LiyCo02, which is currently used as the positive electrode in a commercial cell.The precursor is synthesised by high temperature reaction of the two carbonates to form LiCoO2. Charging the electrode in situ within a cell gives a positive electrode with a specific capacity of up to 200 mA h g-1, corresponding to the stoichiometry Lio3CoO2. At this point the potential vs. a lithium electrode would be about 4.5 V, where parasitic reactions of electrolyte oxidation become excessive and prevent further extraction of lithium. Lithium nickel oxide and lithium cobalt-nickel oxides have been shown to have a greater stoichiometry range due to slightly lower potentials for lithium extraction. It should be noted that this route would give greater scope for the improvement of energy density if new electrolytes could be found to withstand the highly oxidising conditions on charge.The same principle as that described above has been used to synthesise the spinel-related Li,MnzO4 electrode.22 In this material the value of y can be reduced from one, as prepared, to almost zero because the potential is lower, about 4.15 V. Because the lithium-transition metal oxide ratio is smaller in this case the specific energy is correspondingly smaller. LiMn204 can also be reduced electrochemically to Li2Mn204. In this case the reaction involves an abrupt phase change at a constant potential of about 3 V. 4.4 Organic electrodes Conducting polymers have also been considered as positive electrodes for lithium batteries. Polypyrrole, in combination with a graphite negative electrode, was recently shown23 to give a cell with a specific energy of 300 W h kg-1.Poly(carbon disulfide), PCS, has been reported to give a specific capacity approaching 1000 A h kg-1 when used as the positive electrode.24 Although the average cell potential with metallic lithium electrodes was only 2 V, this gives an impressively high theoretical specific energy, close to 2000 W h kg-1. However, a substantial capacity loss occurs on extended cycling. 5 Negative electrodes 5.1 Lithium metal A negative electrode consisting of metallic lithium should be ideal from a consideration of the specific energy because of its small molar mass of 7 g mol-1, theoretically providing almost 4 A h of charge storage per gram of material.However, complete reversibility of the negative electrode reaction, lithium dissolution and plating, is practically unobtainable because of corrosion and dendritic plating25 which cause a decrease in the specific capacity and a safety hazard due to micronisation of the lithium (Fig. 8). It is probably inevitable that some reaction occurs between lithium and any electrolyte which has a sufficiently high conductivity at room temperature. Kinetic stability thus exists due to a passivation layer of reaction products. Successful discharge is, however, possible because the passivation layer ndrites isolated before charge after charge after discharge lithium Fig. 8 Dendrite formation on a lithium electrode causes capacity loss by redistributing lithium from the bulk to isolated islands in the electrolyte itself can have a sufficient conductivity to act as an auxiliary electrolyte layer, or solid electrolyte interface (SEI).26 Problems arise on charge, when a non-uniform current density causes the lithium to plate in a dendritic form, by- passing the original passivation layer and causing the deposition of a secondary passivation layer on the dendrites themselves.Further problems arise on the next discharge half-cycle, when some of the newly-plated lithium becomes isolated and full discharge can only be obtained by removing lithium from beneath the first passivation layer. The end result is a composite mass of finely divided lithium interspersed with the reaction products.Under these circumstances a significant corrosion of the lithium electrode occurs so that a large excess of lithium has to be used to compensate. Furthermore, the finely divided lithium is extremely reactive and presents a serious fire hazard especially when lithium dendrites cause a short-circuit. The above problems do not, however, necessarily preclude the use of lithium metal as the negative electrode because it has been found that the quality of lithium plating reaction depends strongly on the electrolyte composition. Active research on the lithium metal negative electrode is still being pursued in many laboratories where the following approaches are used: the use of more inert solvent systems27 and salts;28 increasing the stack pressure on the electrode;29 the addition to the electrolyte of compounds which improve the plating morphology;3'J the use of polymer electrolytes.3I 5.2 Negative electrodes other than lithium metal Lithium-aluminium alloys have already been mentioned as the negative electrode material in high temperature cells.These cannot be used at low temperature because of the very low diffusion coefficient of lithium in the a A1 phase which forms on discharge. However, a number of other alloy negative electrodes have been proposed, which include Li,Sn, Li,Bi, and Li,Si. Graphite was one of the first known examples of an insertion electrode, with an ability to accept lithium up to a maximum stoichiometry of LiC6, corresponding to a theoretical specific capacity of about 370 A h kg-l.Fortunately for the lithium-ion battery, the electrochemical insertion of lithium occurs reversi- bly between 0.2 and about 0.05 V vs. lithium. Phase transitions due to increasing orders of lithium insertion between the graphene sheets have been observed by plateaux in the near- equilibrium discharge curve and in situ X-ray diffractograms taken during discharge32 (Fig. 9). Graphite insertion is often accompanied by side reactions involving electrolyte decomposition, enhanced by the large surface area compared with a metallic lithium electrode and, in extreme cases, exfoliation of the structure. Suitable choice of electrolyte reduces this problem by surface passivation during the first charge cycle. The situation is different from that of a plated lithium electrode because the dimensional changes are less and more evenly distributed-dendrite formation cannot occur.However, the additional charge required to form the passivating film is substantial, requiring the addition of excess positive electrode material and a consequent reduction in the energy density. Non-graphitic forms of carbon have shown more efficient passivation, possibly because exfoliation is less likely in a cross-linked network structure. In the case of 'hard' carbons 0.60 0.60 0.50 0.50 0.40 0.40 3 0.30 t-J 9.3 0.20 0.30 0.20 0.10 0.10 0.00 0.00 -0,lO -0.10 22 23 24 25 26 27 8 20/" (b) Fig. 9 (a)Discharge of a graphite electrode as observed by current-pulsed coulometric titration.Potentials in lower and upper curves correspond to 300 s periods during insertion (including iR and concentration polarisation) and 750 s relaxation, (near-equilibrium values) respectively. The time axis also represents the total inserted charge. (h)X-Ray diffractograms showing the position of the (002) peak, which reveals the various stages (phases) produced during lithium insertion. such as various types of coke, however, discharge occurs over potential range between 0 and 1 V vs. lithium and the capacity is less than that of graphite. 'Soft' carbons are partially graphitised by heating above 2000 "C and heating to 2600 "C produces electrode materials with capacities approaching that of natural graphite.33 Carbons with capacities much higher than that of LiC6 have been rep0rted.3~ However, problems remain with irreversible capacity on first charge and the optimisation of the carbon negative electrode is under intensive investigation. 6 Current manufacturing technology and future prospects A theoretical specific energy of 500 W h kg-1 was calculated for the LiCfli02 cell. However, the specific energy of commercial lithium ion cells is presently around 110 W h kg-1 for the C/LiCoO2 system.29 This is partly due to the chem- istry-electrolyte instability prevents full extraction of lithium from the positive electrode, and requires the addition of excess LiCo02 to provide for the initial passivation of the negative electrode during the first charge. However, the remainder of the Chemical Society Reviews, 1997, volume 26 265 shortfall is due to the packaging and the presence of additional components for fast and efficient extraction of electrochemical energy from the active materials.Efficient current collectors are required to act as substrates for the electrode films and to carry the cell current to and from the electrodes. Since the mass and volume of these materials are included in the practical energy specifications, it is important that the current collectors are made in very thin film form and that they have sufficient strength to withstand stresses generated in the cell fabrication and in subsequent cycling. Metals are normally used, so that corrosion is an important issue, especially in view of the reactivity of the materials and extremes of electrochemical potential applied.The negative electrode current collector must resist lithium insertion to form alloys, because alloy formation would deplete lithium from the electrode and also degrade the mechanical properties of the current collector. Thermodynamic data are a good guide in this case, and alloy phase diagrams can be used to find which metals have a low lithium solubility and no compound formation. Copper is a popular choice since it is relatively cheap, less harmful than nickel, highly conductive and easily rolled into thin foils. The positive electrode current collector must be resistant to oxidation in the presence of the anion present in the electrolyte. This is a difficult condition to satisfy, since the potential of the positive electrode is usually above the dissolution potential of all the metals. Also, anions designed for high solubility and dissociation of their lithium salts usually form highly soluble salts with other metals.The same is true for the solvent-a high donor number stabilises not only lithium complexes, but metal salt complexes in general. Therefore a passivating film is essential for a metallic positive electrode current collector and aluminium has been chosen in the first generation of products. Table 2, taken from data for a lithium ion cell,35 shows the components of a lithium ion cell of the type shown in Fig.10. The low thicknesses of the electrolyte and electrode layers are noteworthy, and contribute to a high power density. Table 2 Typical design parameters for a C/LiCo02 cylindrical cell29 Thickness/ Function Material Wn Negative electrode active material non-graphitised carbon 90 each side Negative electrode current collector copper 25 Positive electrode active material LiCoO;! 80 each side Positive electrode current collector aluminium 25 Electrolyte/separator PC/DEC/LiPF&elgard 25 Improvements will be sought in the fabrication methods for materials already in use in order to increase the energy storage and power specifications, e.g. to give an important increase in charge rate which may be required in the electric vehicle application.The density of active material on the electrodes is presently far below the theoretical value because of the porosity required to accommodate electrolyte and possibly some decom- position products. Nanostructural and microstructural control of materials during synthesis and fabrication of electrodes are of obvious relevance here. A decrease in the laminate size may also contribute to high charge rates, as will control of the electrolyte path within the electrode composite. Safety, cost and environmental acceptability are of param- ount importance in the choice of the battery system and any associated materials. Thus carbon and manganese dioxide are considered to be important materials for lithium ion systems.It is generally accepted that lithium itself is plentiful, e.g. in ocean water. Regarding other components, it is easy to estimate the costs of simple materials according to the presence of any 266 Chemical Society Reviews, 1997, volume 26 Positive cap ,Positive tab Gaske 3arator Insulator f 'NegativeCan Centre pin Fig. 10 Schematic construction of a typical cylindrical-type lithium cell rare elements, but it is difficult to estimate the ultimate cost of a speciality chemical in tonnage quantities. Therefore changes in the choice of materials can be expected with the discovery of a cheap synthetic methods. Polymer electrolytes should ultimately be superior to liquids for many reasons, including durability, safety, flexibility and ease of manufacture.Solvent-free polymers also offer the possibility of increasing the lifetime beyond the two years typically obtained at present. However, improvements in room temperature conductivity are required for high power. In conclusion, it may be said that rechargeable lithium batteries are now well established and set to take over a large fraction of a rapidly expanding market. Projected specifications for future generation products are: 180 W h kg-1, 360 W h dm-3 with a 500 cycle life, or more than 120W h kg-', 240 W h dm-3 with a 3500 cycle life.' Given the appropriate investment in research and development, improvements in materials and cell design may be expected to continue for many decades ahead.7 References 1 S. Yoda, Extended Abstract, 8th International Meeting on Lithium Batteries, Nagoya, Japan, June 1996. 2 B. C. H. Steele, Phil. Trans. R. Soc. Lond. Ser. A, 1981,302, 361. 3 Lithium Batteries, ed. Pistoia, Elsevier, Amsterdam, 1994. 4 Handbook of Batteries, ed. D. Linden, 2nd edn., McGraw Hill. 5 Proceedings of the 1st to 7th International Meetings on Lithium Batteries, Elsevier Sequoia. The latest volume also appeared in J. Power Sources, 1995, vol. 54. 6 J. Crank, The Mathematics of Diffusion, Clarendon, Oxford, 2nd edn., 1975, p. 48. 7 J. R. Owen, Solid State lonics, 1981, 5, 343. 8 C. F. Holmes, ch. 10 of ref. 3. 9 G. A. Hendriksen, ch. 39 of ref. 4. 10 S. D. Jones and J. Ackridge, J. Power Sources, 1995,54,63. 11 V.Gutmann, Coordination Chemistry in Non-Aqueous Solvents, Springer-Verlag, Vienna, 1968. 12 L. A. Dominey, ch. 4 of ref. 3. 13 M. Armand, Proceedings of the Second International Meeting on Solid Electrolytes, St Andrews, Scotland, 1978. 14 J. R. Craven, R. H. Mobbs, C. Booth and J. R. M. Giles, Makromol. Chem., Rapid Commun., 1986,7, 81. 15 P. M. Blonski, D. F. Shriver, P. Austin and H. R. Allcock, J. Am. Chem. Soc., 1984, 106, 6854. 16 I. E. Kelly, J. R. Owen and B. C. H. Steele, J.Electroanal. Chem., 1984, 168, 467. 17 J. R. McCallum, M. J. Smith and C. A. Vincent, SolidStateZonics, 1981, 11, 307. 18 W. R. McKinnon, in Solid State Electrochemistry, ed. P. G. Bruce, Cambridge University Press, 1995. ch. 7, p. 187. 19 T. Ohsuku, ch. 6 of ref. 3. 20 N. Kumagai, N. Kumagai and K. Tanno, J. Electrochem. Soc., 1992, 135, 3553. 21 K. Mizushima, P. C. Jones, P. C. Wiseman and 3. B. Goodenough, Mater. Res. Bull., 1980, 15, 783. 22 M. M. Thackeray, W. I. F. David, P. G. Bruce and J. B. Goodenough, Mater. Res. Bull, 1983, 18, 461. 23 S. Panero, E. Spila and B. Scrosati, J. Electrochem. Soc., 1996, 143, L29. 24 A. B. Gavrilov, I. P. Kovalev and T. A. Skotheim, Electrochem. Soc. Proc., 1996, 97-17, 204. 25 I. E. Eweka, J. F. Rohan, J. R. Owen and A. G. Ritchie, Power Sources, ed. A. Attewell and T. Keily, International Power Sources Symposium Committee, 1995, vol. 15, no. 241. 26 E. Peled, J. Electrochem. SOC.. 1979, 126, 2047. 27 D.Aurbach, I. Weissman, A. Zaban and 0.Chusid, Electrochim. Acta., 1994, 39, 51. 28 C. Frignant, A. Tranchant and R. Messina, Electrochim. Acta., 1995,40, 513. 29 T. Hirai, J. Yoshimatsu and J. Yamaki, J. Electrochem. Soc., 1988,135, 2422. 30 T. Hirai, J. Yoshimatsu and J. Yamaki, J. Electrochem. Soc., 1994,141, 2300. 31 T. Osaka, T. Momma, H. Ito and B. Scrosati, Electrochem. SOC.Proc., 1996,97-17, 1. 32 A. H. Whitehead, K. Edstrom, N. Rao and J. R. Owen, J. Power Sources, 1996, 63, 4 1. 33 J. R. Dahn, A. K. Sleigh, H. Shi, B. M. Way, W. J. Weydanz, J. N. Reimers, Q. Zhong and U. von Sacken, ch. 1 of ref. 3. 34 G. Sandi, R. E. Winans and K. A. Carrado, J. Electrochem. Soc., 1996, 143, L95.. 35 S. Hosain, ch. 36 of ref. 4. Received, 24th March 1997 Accepted, 23rd May 1997 Chemical Society Reviews, 1997, volume 26 267

ISSN:0306-0012    

DOI:10.1039/CS9972600259

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Oxaziridine rearrangements in asymmetric synthesis Jeffrey Aube Department of Medicinal Chemistry, University of Kansas, Lawrence, Oxaziridines undergo photochemical rearrangement reac- tions to afford chiral lactams with high levels of enantio- or regio-selectivity. This reaction was applied to the synthesis of targets such as carnitine, the yohimbine alkaloids, and several classes of peptidomimetics. In addition, single- electron transfer reactions can be elicited from oxaziridines using CuI. These reactions afford nitrogen and carbon radicals that add to an appended olefin, with the final products depending on both the substitution and stereo- chemistry of the starting oxaziridine. 1 Introduction to oxaziridines The oxaziridine, that three-membered ring containing nitrogen, oxygen and carbon, has been of mechanistic and structural interest since its introduction in a classic paper by Emmons in 1957.1 This paper not only describes the preparation and structural proof of a new heterocycle, it also contains the basis for most oxaziridine chemistry known to the present day, including a rich rearrangement chemistry and the decomposi- tion of oxaziridines by low-valent metal salts.2 In the years that followed, the oxaziridine was mostly a curiosity that found little synthetic utility.This was despite significant efforts to devise nice methods for the synthesis of oxaziridines-ultimately settling on the oxidation of imines with peracids as the best method-and the attention of stereochemists interested in the configurational stabilization of a potential nitrogen stereocentre [eqn.(l)].3 It turns out that the R R, N N*.' (1) R'&O Rq2CYO combination of placing a nitrogen in a three-membered ring (which destabilizes the 120"angle of the sp2-like transition state involved in pyramidal inversion) with attaching the electron- withdrawing oxygen atom (which opposes the increased s-orbital character of nitrogen during inversion) makes simple oxaziridines the most easily prepared and synthetically useful class of compounds containing a bonafide nitrogen stereogenic reffrey Aube' attended the Uni- Jersity of Miami, where he did indergraduate research with 3rofessor Robert Gawley. He *eceived his PhD in chemistry n 1984 from Duke University, vorking with Professor Steven 3aldwin, and was an NIH post-ioctoral fellow at Yale Uni- 7ersity with Professor Samuel >anishefsky. In 1986,he moved o the University of Kansas, vhere he is now a professor in he Department of Medicinal ?hemistry.KS 66045-2506, USA centre. More recently, N-sulfonyl oxaziridines ('Davis rea-gents') have become popular oxygen-transfer agents for the oxidation of olefins, sulfides and especially enolates. Asym- metric induction is often possible and practical in this chemi~try.~?~A less-explored area that nonetheless has con- siderable promise is the use of oxaziridines as nitrogen-transfer reagents for the synthesis of aziridines and hydrazine deriva- tives.5 Our own interest in oxaziridines began with the rearrange- ment chemistry of N-alkyl oxaziridines, especially those containing chiral substituents.In particular, we wished to exploit the nitrogen stereocentre as a controlling feature in ring- expansion chemistry. This review will describe the develop- ment of ring-expansion reactions that proceed via oxaziridines and lead to the asymmetric synthesis of chiral lactams. This reaction has been used in the synthesis of a variety of interesting natural and unnatural products, including GABOB and carni- tine, the yohimbine alkaloids, and several classes of peptidomi- metics. In addition, we have recently become interested in the use of oxaziridines as precursors to nitrogen and carbon radicals, and our continuing efforts to develop useful synthetic techniques from this chemistry will be discussed. 2 Stereoselective synthesis and rearrangements of chiral oxaziridines 2.1 The concept of a group-selective ring expansion In the modern era of asymmetric synthesis, many methods for the stereodifferentiation of prochiral faces have been intro- duced.A fundamentally different mode of asymmetric induc- tion involves the differentiation of enantio- or diastereo-topic groups. Many enzymes operate in just this way [Scheme l(a)]. In this example, the two acetate esters present in the starting molecule are enantiotopic because they can be interchanged by a mirror-plane operation. Selective hydrolysis of this type of ester can be accomplished by a variety of different enzymes, including such popular 'reagents' as pig liver esterase or electric eel acetylcholinesterase.Inspired by nature, chemists have begun to develop abiotic reactions that can achieve this kind of selectivity; several processes particularly relevant to this review involve cyclic ketones containing a plane of symmetry [Scheme l(b)]. Note, that the two methylene groups adjacent to the carbonyl moiety in a 4-alkylcyclohexanone are enantiotopic, as are the axially orientated hydrogen atoms shown. Stereodiffer- entiation of these protons can be accomplished by reaction with some external chiral reagent. For example, various chiral lithamide bases have been used to generate scalemic mixtures of axially dissymmetric enolates [Scheme 1 (h)].6 We decided to explore another variation on this theme.The ring expansion of the 4-alkylcyclohexanone in Scheme 1(h) involves the insertion of a group X between the carbonyl and either enantiotopic methylene group; depending on which group migrates, one or another enantiomeric product results. This means that if a way could be found to differentiate these adjacent methylene groups in the migration step, it would be possible to enact an asymmetric synthesis of a seven-membered hetero-or carbo-cyclic product. Once again, enzymes have proved useful: Taschner and coworkers examined cyclohex- anone oxygenase, which can effect enantioselective Baeyer- Chemical Society Reviews, 1997, volume 26 269 R+OH RSok (a) R<oAc- OAc OAc OH enantiotopic esters enantiomers OLi OLi (b) asymmetric/ deprotonation 8 00 R R R \ asymmetricring, ~3 7')expansion the methylene groups YY adjacent to the carbonyl R R group are enantiotopic X = 0 (Baeyer-Villiger) X = NR (Beckmann, Schmidt) Scheme 1 Villiger chemistry over a range of symmetrical cyclohex- anones.7 Non-enzymatic methods of carrying out this reaction have been relatively slow in coming and do not yet approach the efficiency of the biochemical process.8 In contrast, the possibil- ity of a corresponding nitrogen insertion chemistry, i.e.asymmetric equivalents of the Beckmann or Schmidt rearrange- ment reactions, was recognized rather early but did not enter the synthetic mainstream until much later.Thus, Lyle and Lyle9 and Todalo were able to resolve axially dissymmetric oximes and carry out their rearrangement to enantiomerically pure lactams, but neither a generally useful asymmetric nitrogen ring expansion chemistry nor its systematic application to problems in asymmetric synthesis were in place. It should also be noted that the case corresponding to Scheme l(b) where X = CH2has received some recent attention as well. 2.2 Synthetic methodology development In the meantime, Lattes, Riviitre, and coworkers at the University of Toulouse had been investigating the photochem- ical rearrangement reactions of oxaziridines to lactams.12 A few examples are cited out of many because they demonstrate some puzzling aspects of the regiochemistry of this reaction (Scheme 2).In Scheme 2(a), ring-contraction to the azetidinone product shown was observed instead of migration of the methyl group, which would have led to a considerably less strained pyrrolidi- none lactam. The outcome of the rearrangement of a spirocyclic oxaziridine derived from 2-methylcyclohexanone [1, Scheme 2(b)] highlights one of the most confusing aspects of the reaction mechanism as originally proposed. Photochemical promotion of the oxaziridine to a singlet excited state would afford an N/O diradical set up for subsequent C-C bond cleavage. Carbon-nitrogen bond formation would then afford lactam product. However, the main product of this reaction- favoured 95 :5-results from migration of the less-substituted a-carbon, which corresponds to carbon-carbon cleavage af- fording the less stable carbon-centred radical intermediate.Lattes hypothesized that the bond antiperiplanar to the lone pair in each oxaziridine-signified by a bold bond in the Scheme 2-underwent preferential cleavage; this idea was convincingly demonstrated by the ingenious experiment shown in Scheme 2(c).'2 In this case, the cleavage of either C-C bond of the oxaziridine ring would afford similarly substituted primary carbon radicals, presumably of nearly equal energy. Thus, photolysis of spirocyclic oxaziridine 2 afforded a single lactam 3a in 70% yield after recrystallization. Since the complete structures of both oxaziridine and lactam were secured by X-ray crystallography, this experiment unambiguously demonstrated 7 notMeYN observed: me.^M hv 0 0 more strained product predominates U U U 1 95 : 5 less stable radical predominates Butv Y Y (2) X-ray Scheme 2 that the nitrogen stereocentre was responsible for the re-giochemistry of the carbon-carbon cleavage event.Despite the utility of this observation and continued mechanistic interest,13 many questions regarding the reasons for this selectivity and other details of oxaziridine photochemistry remain. In partic- ular, the existence of radicals along this reaction pathway has still not been demonstrated. Despite such nagging mechanistic issues, the stereoselective conversion of 2 +3a captured our imagination as a bona fide example of a group-selective nitrogen insertion reaction, especially considering that oxaziridines like 2 could be readily prepared from the corresponding substituted cyclohexanone.Before one could use this chemistry in asymmetric synthesis, it was necessary to quantitatively assess the level of stereochem- ical control possible for oxaziridine synthesis and rearrange- ment (recall that Lattes' experiments all involved recrystallized material, so that the actual diastereomeric ratios obtained in each reaction were unknown). Oxaziridines derived from six-membered cyclic imines tend to form via equatorial attack14 and are also subject to control by a chiral group on nitrogen substituent.15 A useful way of applying these precepts to the second-order problem of inducing axial chirality is shown in Scheme 3.First, condensation of a 4-substituted cyclohexanone generates a mixture of diaster- eomeric imines. However, the imine oxidation step is generally considered to be a two-stage process, although this has never been rigorously proved. Accordingly, addition of the peracid to either imine can afford products of equatorial attack or axial attack, with the former being favoured. At this point, the configuration of the imine has been lost and is no longer relevant. Now, N-O bond closure occurs such that the newly generated nitrogen stereo- 270 Chemical Society Reviews, 1997, volume 26 centre emerges with the unlike relative configuration when a-methylbenzylamine (a-MBA) was used for imine formation.The best results are usually obtained when a bulky peracid, like monoperoxycamphoric acid (MPCA), is used; for 4-phenyl- cyclohexanone, the ratio of the four possible products is 83 : 12 :3 :2 (87% combined yield).16 Incidentally, the fact that the oxidant is chiral has no bearing because any given stereocentre on MPCA is too far away from the forming bonds to have an effect on the face selectivity of oxygen addition. So far, in fact, no generally useful reagent for the asymmetric synthesis of oxaziridines exists. Taken together, the combination of the stereoselective synthesis of an oxaziridine (Scheme 3) with its regiospecific rearrangement reaction [Scheme 2(h)] results in the enan-tioselective synthesis of a lactam from a symmetrical ketone.Careful examination of both the oxaziridine-forming and -rearrangement steps established the efficiency of each reaction and the overall sequence. The proportion of the oxaziridine mixtures comprising equatorial/unlike product [e.g. 2 in Scheme 2(c)] was 59-91% when MPCA was used as the oxidant.16 The best ratios were obtained in imines prepared from cyclohexanones containing bulky substituents, presum- ably due to their superior ability to prejudice the six-membered ring into one chair-like conformation over the other. Fur- thermore, photolysis of samples of all four purified oxaziridine diastereomers showed that the group anti to the nitrogen lone pair predominantly migrated in each case; ratios ranged from 7.3-24: 1.[These experiments proved that the result in Scheme 2(c) was not just a fluke resulting from some unexpected stability of lactam 3a over 3b.l When one considered the entire sequence from ketone to lactam, however, the practical ratios of chiral lactams generally plateaued at ca. 7.3 : 1, although this was mitigated by the easy separability of the products. Although other amines have been investigated, a-MBA is preferred for simple ring-expansions due to its low cost and ease of removal under dissolving metal conditions. As an aside, the structural determination of these spirocyclic oxaziridines has proved very difficult throughout our work 4Ph Me Me because the quaternary and nitrogen stereocentres lack protons useful for coupling constant determinations.Accordingly, we have had to rely on NMR trends, chemical interconversions, and the occasional X-ray structure. A useful byproduct of all of this has been a greater appreciation for the solution state conforma- tions of the spirocyclic oxaziridines (Fig. l).l7 In this case, the axial hydrogen atom depicted is observed in the 'H NMR spectrum at 0.22 ppm, a value consistent with a highly populated conformation in which the phenyl group is poised over the six-membered ring. Fig. 1Proposed predominant conformation of spirocyclic oxaziridines 2.3 Synthetic applications of the oxaziridine -+lactam rearrangement reaction 2.3.1 From amino acids to peptidomimetics Our first synthetic goals were modest: directly apply the asymmetric ring expansion reaction to some simple precursors of biologically relevant heterocycles (Scheme 4).The ring expansion of the phenyl-substituted oxaziridines shown in Scheme 3 came with a ready-made application to morphinoid synthesis as racemic 5-phenylcaprolactam had been previously converted to the benzomorphinan ring system [Scheme 4(a)].l6,I* On the other hand, the application of oxaziridines to five-membered lactam synthesis was not selective, affording an equimolar mixture of lactams from 3-substituted cyclobut- anones regardless of substitution. Still, the alkoxy-substituted version shown in Scheme 4(b) was used in a very brief synthesis Ph unlike 83 Me Ph 12 3 Ph Ph 2 Ph Scheme 3 Chemical Society Reviews, 1997, volume 26 271 of enantiopure carnitine, an essential amino acid involved in fatty acid metabolism.19 76% overall yield benzomorphinan 7.3 : 1 selectivity ring system OH 58-66% overall yield R = H, GABOB 1.1 : 1 selectivity R = Me, carnitine Scheme 4 Another direct application of asymmetric synthesis has permitted the synthesis of a new and exciting class of peptidomimetics.Peptidomimetics have biological activity similar to naturally occurring peptides (e.g. hormones) but dissimilar chemical structures.20 Lactams are especially good substructures for peptidomimetic design because the amide linkage is subject to a variety of conformational constraints imposed by the ring.Molecules in which the side chain of one amino acid is formally connected to the a-carbon of the next (going from N to C terminus) are colloquially called Freidinger lactams, after a key inventor of this strategy for peptide constraint [Scheme 5(a)] .2 We have reported a new and potentially general synthesis of Freidinger lactams and applied it to the preparation of some inhibitors of angiotensin-converting en~yme.2~7~3 Accordingly,2-N-Boc-aminocycloalkanonesvarying from 5-7 membered rings were synthesized and converted to the corresponding oxaziridine by condensation with an amino ester like L-phenylalanine [Scheme 5(b)]. In this case, the face selectivity of the oxidation step is of little importance; instead, the key feature is the generally exclusive formation of the oxaziridine in which the group on nitrogen is trans to the more highly substituted carbon substituent.Photorearrangement in which the carbon anti to the lone pair on nitrogen predominates gives the desired peptidomimetic molecule. Note, that the issue is no longer stereoselectivity, but rather regioselectivity. Typical Beckmann and Schmidt rearrangements occur so that the more substituted carbon migrates to nitrogen.24 However, formation of the trans isomer coupled with the stereoelectronic preference of the photochemical rearrangement lead to migra- tion of the less substituted carbon in Scheme 5(b). Thus, the Beckmann/Schmidt chemistry and oxaziridine chemistry com- plement one another (if the former worked at all: substrates containing a-heteroatoms often run into problems under normal Beckmann or Schmidt conditions). This is a general and most useful feature of oxaziridine chemistry.An even trickier situation arises in unsymmetrical ketones where the substitution is not adjacent to the carbonyl group. For example, the two a-carbons of the bicyclic ketone in Scheme 6 have similar migratory aptitudes; in addition, there are no steric interactions that could be expected to lead to stereoselectivity in the derived oxime needed for Beckmann chemistry. However, since the ketones are (1) chiral and (2) give imines that should each undergo equatorial attack (i.e., P-attack as drawn), one can use a chiral amine to direct the absolute configuration in the derived oxaziridines.25 Thus, merely switching from (S)-to (R)-a-MBA in the imine-forming step, followed by m-CPBA oxidation, affords oxaziridines in which the nitrogen stereo- centres are of opposite configuration.Photolysis and N-a-methylbenzyl reduction lead to regioisomeric lactams as shown. This strategy is one of very few in which a technique normally associated with asymmetric synthesis is coupled with reaction on an enantiomerically pure substrate to afford regioisomeric products. The need for enantiomerically pure starting materials arises from the fact that the racemic ketone would couple with enantiomerically pure amine to give an unavoidably 1 : 1 mixture of diastereomeric oxaziridines.In fact, racemic ketones can be ‘resolved’ into two lactams of opposite (1) re-giochemistry and (2) absolute stereochemistry using this chemistry; the details of this are left to the interested reader as an exercise.25>26 It turned out that all of these techniques were needed for the preparation of a series of stereoisomeric (3-turn mimics.27 (3-Turns [see Scheme 7(a) for an example of this type of peptide secondary structure, including a ball-and-stick model of a ‘Type 11’ Cj-turn] are known to be involved in a wide variety of biologically important binding events; because of this, they have been some of the most studied archetypes for peptidomi- metic design. We have been interested in making (3-turn peptidomimetics by the straightforward strategy of constraining a dipeptide unit into a cyclic molecule using an aminocaproic acid linker [Scheme 7(b)] .27 We discovered that variously disubstituted linkers were able to bias the conformation of the macrocyclic peptidomimetics into different subclasses of P-turns.This study was made possible by preparing the appropriate aminocaproic acids in enantiomerically pure form and then coupling them to an appropriately protected dipeptide using standard chemistry [Scheme 7(c)]. For the cis (meso) isomers, application of our asymmetric ring expansion protocol gave the desired syn-dimethyl linker without incident (not shown). However, the trans-dimethylcyclohexanone starting material provided an opportunity to conduct a simultaneous ring expansion/resolution procedure [Scheme 7(d)].Here, the N-a-methylbenzyl group permitted an easy chromatographic separa- tion of the diastereomeric lactams prior to nitrogen ‘deprotec- tion’. An alternative route would have been to prepare trans-3,5-dimethy lc yclohexanone in enantiomerically pure Scheme 5 rn-CPBA = rn-chloropenbenzoic acid 272 Chemical Society Reviews, 1997, volume 26 form and carry out a standard Beckmann rearrangement, of several members of the yohimbine alkaloid family.28.29 As recognizing that the two a-carbons are now identical. shown in Scheme 8,we hoped to maximize the convergency of our syntheses by installing a nitrogen atom already bearing an 2.3.2 Yohimbine alkaloids indole group. An early route to alloyohimbane used a-MBA in Some more sophisticated applications of the oxaziridine -+ the ring-expansion step; not only was this route poorly selective lactam rearrangement reaction were carried out in the synthesis but several low-yielding steps were encountered between removing the N-a-methylbenzyl group and attaching the indole (sequence not shown).Here, the stereochemistry of the emerging oxaziridine would be controlled by the chiral amine, .+RN$0 in direct analogy to the examples above, and by the tendency of cis-bicyclic ketones to undergo attack by oxidant from the less- hindered ex0 (convex) face of the molecule. In contrast, the normal yohimbine series has a trans C/D ring fusion, meaning that its synthesis can begin with a Cz symmetrical ketone, in which the two a-carbons are rendered identical [Scheme 8(b].The direct use of a tryptophan ester in the oxaziridination was o& H not much better in terms of selectivity-the lactam shown in Scheme 8(a) was obtained as a 2.2 :1 mixture of stereoisom- ers-but the overall efficiency was much improved. (The survival of the oxidation-sensitive indole ring is itself note- Scheme 6 worthy.) The conversion of the lactam to alloyohimbane Q Q H CH3 H-GI y-Ala-Gly -Gly-OH racernic separable by chromatography Scheme 7 achiral ketone c-* H chiral ketone yo himbane Scheme 8 Chemical Society Reviews, 1997, volume 26 273 entailed Bischler-Napieralski cyclization, isomer separation, and removal of the methoxycarbonyl group.For comparison, the chiral ketone in Scheme 8(b)(prepared in high enantiomeric purity using a sequence featuring an asymmetric Diels-Alder reaction) was readily converted to yohimbane in just six steps and 31% overall yield. The ability to directly introduce a substituted nitrogen is a real bonus of oxaziridine-based insertion chemistry as compared to classical reactions. To date, the most ambitious synthesis that features an oxaziridine rearrangement reaction has been the total synthesis of (+)-yohimbine, the first non-formal asymmetric synthesis of this famous target (Scheme 9).28 Of several possible ap-proaches, we chose to begin with the same chiral ketone used for the yohimbane synthesis; osmylation and acetylation of its alkene installed a protected diol that would later be processed to the E-ring functionality of the natural product. Experimentally, it proved possible to prepare the oxaziridine shown as a complex mixture of stereoisomers that led to two isomeric lactams upon photolysis in 55 and 25% yields, respectively.The major lactam underwent Bischler-Napieralski cyclization to give a pentacyclic diacetate that was converted to yohimbine. Two points stand out. First, the conversion of the E-ring functionality to that of yohimbine was possible because of the rigid nature of this multicyclic compound and the fact the one ester occupied an equatorial position and the other was axial. Secondly, the minor isomer of the ring-expansion sequence, which is not shown, carried two (3lcisacetoxy groups that could also have been differentiated in later chemistry and converted to yohimbine as well.Even if it did not really matter here (except in terms of convenience), the modest regioselectivity of the ring expansion (ca. 2.2 : 1) was remarkable given the distance between the imine and the resident acetoxy groups. 3 Single-electron transfer reactions of oxaziridines: nitrogen and carbon radical generation As the photochemical rearrangement reactions of oxaziridines were beginning to achieve some notoriety as a synthetic method, it occurred to us that it would be nice if we could replace the photochemical reaction conditions with another, preferably catalytic method. Recalling the early postulates regarding the mechanism of oxaziridine photochemical rear- rangements, we wanted to compare the features of the light- induced reactions-especially regioselectivity-with 'real' rad- ical conditions.Once again, it was only necessary to return to Emmons' original paper' to note that low-valent metal salts like FeII could reduce the oxaziridine ring via single-electron transfer (SET); these observations were later refined through AcO important work by Minisci30 and Black.31 The resulting nitrogen radical-oxygen anion pairs could then lead to a variety of products depending on oxaziridine substitution. As applied to the present circumstance, a diastereomerically pure spirocyclic oxaziridine was converted to lactam by using [Cu(PPh3)C1]4, an easily prepared form of CuI that is soluble in tetrahydrofuran (THF) [Scheme lO(a)].However, although the reaction was reasonably efficient, this rearrangement method was not pursued because the reaction was poorly selective compared to the photochemical version described above. Other workers have since investigated SET-induced oxaziridine rearrangement ~hemistry,3~but we were intrigued by the possibility that the putative nitrogen radical could undergo addition to an appended olefin and chose to briefly investigate that possibility. To this end, the alkene-bearing oxaziridines shown in Scheme 10(b)were synthesized as a mixture of stereoisomers and separated by column chromatography. Each isomeric oxaziridine was individually submitted to the same conditions as used for the previous reaction.33 We were unprepared for the products obtained in these experiments, and especially for the fact that two entirely different compounds were obtained from isomeric oxaziridines! The isomer in which the oxygen atom was (3 as drawn in Scheme 10(b) gave the pyrroline shown in 63% yield and, more interestingly, high enantiomeric purity.A change in the relative stereochemistry between the oxaziridine C-2 and N positions (such as the a-0x0 isomer shown) resulted not in a stereoisomer of the pyrroline (as we expected) but rather an aziridine. After carrying out the appropriate experiments to completely ascertain the stereostructures of the products and to show that the aryl group was transferred intramolecularly, one possible explanation for these results was crafted (Scheme 11).SET from the copper to the oxaziridine makes a nitrogen- centred radical that adds to the olefin as expected. In one case [Scheme 1 l(a)],the resulting radical attacks the aromatic ring, which is transferred, leaving behind a stabilized radical a to the nitrogen group. The formal loss of acetaldehyde now occurs; although several mechanisms to account for this are possible, we have been unable to obtain conclusive experimental evidence in favour of any one of these. The failure of some stereoisomeric oxaziridines to afford pyrroline was rationalized as shown in Scheme 1l(b):the necessary conformation needed for radical attack on the aryl group leads to an unfavourable steric interaction between the a-methyl group and the phenyl substituent on the five-membered ring, and so the alternative path shown was taken, leading to aziridine.We have begun the considerable work of amassing experimental evidence pertain- ing to these mechanisms. 75% I"1.. Scheme 9 274 Chemical Society Reviews, 1997, volume 26 (a) Me Me 5 mIY0A$Ph [Cu( PPh3)Clh -CU' ___) THF reflux But Bu' ca. 50% ye same<::Ph 3 : 2 mixture of isomers ,Ph >95% ee 0 Ph 62"/0 Ph >95%de NYMe Ph Scheme 10 (a) ye Ph Ph *C,H2 ___ccu' MeANd MetNG SET Phi#* Ph cu"-0 cu"-0 (Ph Metn-cu' -CH&HO c ___c Me:p "c:u"-o Ph Ph' Ph' cu"-0 Scheme 11 Still more complexity was introduced through the examina- tion of alternative substrate types.It appears, for example, that an aromatic group on the oxaziridine is necessary for either of these reaction pathways to succeed [Scheme 12(a)]. Other groups, particularly ones that can readily form a relatively stable radical, undergo the cleavage reaction shown to afford amide. To circumvent this problem, an oxaziridine containing two chemically differentiable womatic rings was subjected to the cyclization-aryl transfer protocol to afford a pyrroline in high enantiomeric purity and 74% yield [Scheme 12(b)]. Ster-eoselective reduction of the pyrroline and nitrogen protection was followed by the chemoselective oxidative degradation of the more electron-rich aromatic ring to afford the proline derivative sh0wn.3~ Even more promising is the possibility that this cleavage reaction can be used to generate carbon-centred radicals under very mild conditions; the cyclization reaction shown in Scheme 12(c) is a preliminary indication of the potential of this chemistry.35 The mechanistic and stereochemical complexities of the reactions of oxaziridines with low-valent metal salts would seem to limit the synthetic utility of this reaction for the time being, but the same could once have been said about the photochemistry that has become such a recognized tool of substantial utility in nitrogen insertion chemistry.What is certain is that oxaziridine chemistry will continue to provide mechanistic and synthetic challenges for some time to come.Chemical Society Reviews, 1997, volume 26 275 CU'PhAN, R& R = -CH2Ph, CH(CH3)2, but-3-enyl Me0 RuO,. NalOl 69% --T Me0 Scheme 12 4 Acknowledgments I would like to thank all of the collaborators, named in the references, who have made our contributions to oxaziridine chemistry possible. Their work has been funded by the University of Kansas, the donors of the Petroleum Research Fund as administered by the American Chemical Society, the National Institutes of Health, the National Science Foundation (Epscor grant NSFEHR-92-55223), the Alfred P. Sloan Foun- dation, and the American Heart Association-Kansas Affiliate. The generous support of these agencies is greatly appreciated. 5 References 1 W.D. Emmons, J. Am. Chem. SOC., 1957,79,5739. 2 F. A. Davis and A. C. Sheppard, Tetrahedron, 1989, 45, 5703. 3 F. A. Davis and R. H. Jenkins, Jr., in Synthesis and Utilization of Compounds with Chiral Nitrogen Centers, ed. J. D. Morrison and J. W. Scott, Orlando, 1984. 4 F. A. Davis and B.-C. Chen, Chem. Rev., 1992,92, 919. 5 J. Vidal, J. Drouin and A. Collet, J. Chem. SOC.,Chem. Commun., 1991, 435. 6 P. J. Cox and N. S. Simpkins, Tetrahedron: Asymmetry, 1991, 2, 1. 7 M. J. Taschner, D. J. Black and Q.-Z. Chen, Tetrahedron: Asymmetry, 1993, 4, 1387. 8 C. Bolm, G. Schlingloff and K. Weickhardt, Angew. Chem., Int. Ed. Engl., 1994, 33, 1848. 9 R. E. Lyle and G. G. Lyle, J. Org. Chem., 1959, 24, 1679.10 F. Toda and H. Akai, J. Org. Chem., 1990,55,4973. 276 Chemical Society Reviews, 1997, volume 26 11 H. Nemoto, H. Ishibashi, M. Magamochi and K. Fukurnoto, J. Org. Chem., 1992,57, 1707. 12 A. Lattes, E. Oliveros, M. Rivikre, C. Belzecki, D. Mostowicz, W. Abramskj, C. Piccinni-Leopardi, G. Germain and M. Van Meerssche, J. Am. Chem. SOC., 1982, 104, 3929. 13 A. J. Post, S. Nwaukwa and H. Morrison, J. Am. Chem. Soc., 1994,116, 6439. 14 E. Oliveros, M. Rivikre and A. Lattes, Org. Magn. Reson., 1976, 8, 601. 15 C. Belzecki and C. Mostowicz, J. Org. Chem., 1975, 40, 3878. 16 J. AubC, Y. Wang, M. Hammond, M. Tanol, F. Takusagawa and D. Vander Velde, J. Am. Chem. SOC., 1990,112,4879. 17 Y. Usuki, Y. Wang and J. AubC, J.Org. Chem., 1995,60, 8028. 18 K. Mitsuhashi, S. Shiotani, R. Oh-uchi and K. Shiraki, Chem. Pharm. Bull., 1969, 17, 434. 19 J. Aubt, Y. Wang, S. Ghosh and K. L. Langhans, Synth. Commun., 1991, 21, 693. 20 A. Giannis and T. Kolter, Angew. Chem., Int. Ed. Engl., 1993, 32, 1244. 21 R. M. Freidinger, D. F. Veber, D. S. Perlow, J. R. Brooks and R. Saperstein, Science, 1980, 210, 656. 22 J. AubC and M. S. Wolfe, Bioorg. Med. Chem. Lett., 1992, 2, 925. 23 M. S. Wolfe, D. Dutta and J. AubC, J. Org. Chem., 1997, 62, 654. 24 R. E. Gawley, Org. React., 1988, 35, 1. 25 J. AubC and M. Hammond, Tetrahedron Lett., 1990, 31, 2963. 26 J. AubC, M. Hammond, E. Gherardini and F. Takusagawa, J. Org. Chem., 1991,56,499. 27 0. Kitagawa, D. Vander Velde, D. Dutta, M. Morton, F. Takusagawa and J. AubC, J. Am. Chem. SOC.,1995, 117,5169. 28 J. AubC, S. Ghosh and M. Tanol, J. Am. Chem. SOC.,1994, 116, 9009. 29 J. AubC and S. Ghosh, in Asymmetric Approaches to the Yohimbine Alkaloids: Development of a Group-Selective Ring Expansion Process and Its Application to Total Synthesis, W. H. Pearson, Greenwich, 1996. 30 F. Minisci, R. Galli, V. Malatesta and T. Caronna, Tetrahedron, 1970, 26, 4083. 31 D. St. C. Black, N. A. Blackman and L. M. Johnstone, Aus. J. Chem., 1979,32,2041. 32 K. Suda, M. Sashima, M. Izutsu and F. Hino, J. Chem. Soc., Chem. Commun., 1994,949. 33 J. AuM, X. Peng, Y. Wang and F. Takusagawa, J. Am. Chem. SOC., 1992,114, 5466. 34 J. AuW, B. Gulgeze and X. Peng, Bioorg. Med. Chem. Lett., 1994, 4, 2461. 35 Y. Usuki and J. AuW, unpublished results. Received, 21st January I997 Accepted, 15th April I997 Chemical Society Reviews, 1997, volume 26 277

ISSN:0306-0012    

DOI:10.1039/CS9972600269

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

MELDOLA LECTURE: understanding the properties of urea and thiourea inclusion compounds* Kenneth D. M. Harris School of Chemistry, Unirwsity of Bir-minghuni,Edghaston, Birmingham, UK B15 2TT Much of the intrinsic appeal of structural science arises from the fact that structural behaviour at the molecular level often resembles macroscopic structures that we can see in the world around us. In the same way that we perceive beauty in the symmetries and forms of macroscopic objects, there is an equally enthralling beauty in the way that nature fashions symmetry and diversity within the architectures of crystal- line solids. In the field of inclusion chemistry, for example, many direct analogies can be drawn between the concepts of inclusion in the microscopic and macroscopic worlds, but the scientific interest and importance of inclusion chemistry extends far beyond such structural comparisons.As this article demonstrates, solid organic inclusion compounds can exhibit a diversity of interesting and important fundamental properties, which can form the basis of a range of important applications. 1 Introduction In general, inclusion compounds can be defined as systems in which one species (the ‘guest’) is spatially confined within another species (the ‘host’). Inclusion phenomena are wide- spread throughout chemistry (see ref. 1 for a comprehensive survey of this field), and can be subdivided into two types. The first type comprises molecular host-guest complexes, in which the host is a molecule possessing an appropriate binding site or cavity for inclusion of the guest.Such complexes can usually exist as associated entities both in the solid state and in dispersed phases (for example, in solution). Examples of these hosts are crown ethers, cyclodextrins, cryptands, rotaxanes and catenanes. In the second type of inclusion compound, guest molecules are located within the architecture of a solid host material, and in these cases the association of host and guest components is strictly a solid state phenomenon. The ‘inclusion spaces’ within these solid hosts encompass a wide variety of topologies, such as linear tunnels, isolated cages, networks of intersecting Kenneth D. M. Harris was born in Alexandria (Scotland) in 1963 and educated at the Uni- versity of St Andrews (BSc, 1985) and the University of Cambridge (PhD, 1988),carry-ing out research for the PhD under the guidance of Professor Sir John Meurig Thomas.He then held lectureships at the University of St Andi-ews and University College London, be- fore taking up his present posi- tion as Professor of Structural Chemistry at the University of Birmingham in 1995. * This article IS based on the Meldola Lecture delivered at the Royal Institution, London on 25th February 1993. tunnels and/or cages, and two-dimensional inter-lamellar re-gions within layered hosts. Well known examples of solid host materials are aluminosilicates (such as zeolites and clay minerals), aluminophosphates, graphite, layered metal chalco- genides and layered metal phosphonates, as well as crystalline organic hosts such as urea, thiourea, tri-ortho-thymotide, perhydrotriphenylene and deoxycholic acid.In these host solids, the smallest dimension of the ‘inclusion spaces’ is of the order of molecular dimensions, and these host structures are therefore able to include individual guest molecules in a manner in which the spatial constraints on the properties of the guest molecules are imposed primarily by their host environment. Within the broad range of solid inclusion compounds, there is an important subdivision between those for which the host structure remains stable when the guest component is removed and those for which the host structure undergoes substantial reorganization when the guest component is removed.For convenience, the terms ‘hard’ host (for the former category) and ‘soft’ host (for the latter category) may be used to distinguish these different types of behaviour. In the case of the soft hosts, the structural reorganization generally involves collapse of the low-density ‘empty’ host structure, with recrystallization to a more compact structure of higher density. Thus, for inclusion compounds of the soft hosts, the guest component generally acts as an essential template for the formation of the host structure as well as an essential buttress for maintaining the stability of the host structure; the collapse of the host structure on removal of the guest component is often an irreversible process.The exact structural nature of the soft hosts often varies substantially depending on the structural and chemical properties of the guest molecules. There is a greater synergy in properties between the host and guest components in the case of the soft hosts, and it is generally not satisfactory to attempt to rationalize the properties of these inclusion compounds in terms of the separate behaviour of the host and guest components. This article describes some of the fundamental scientific issues associated with two particular families of solid organic inclusion compounds-the urea and thiourea inclusion com- pounds. These inclusion compounds exhibit a broad range of fundamental phenomena, and progress in understanding these phenomena has resulted from a substantial amount of work by many scientists over many years, as discussed in a recent comprehensive review .’2 Underpinning the research strategy that has allowed a detailed understanding of these solids to be established, has been the recognition that to understand fully the behaviour of a solid requires the combined application of a wide range of experimental, computational and theoretical ap-proaches, each providing information on a different aspect of the solid.The present article aims to highlight some of the issues of contemporary interest for urea and thiourea inclusion compounds, with some emphasis on selected results from our own recent contributions to this field. 2 Urea inclusion compounds 2.1 An introduction to urea inclusion compounds Urea inclusion compounds were first discovered in the 1940s by Bengen, who found by chance (while studying the effects of Chemical Society Reviews, 1997, volume 26 279 urea on proteins in pasteurised milk) that octanol forms a crystalline adduct with urea.Subsequently, it was found that a wide range of long-chain molecules can form similar adducts with urea, and X-ray diffraction studies carried out by Smith3 provided direct evidence that these materials are based on a tunnel host structure. In the structure of the conventional urea inclusion compounds,334 the urea molecules form an extensively hydrogen-bonded arrangement (Fig. 1) containing linear, paral- lel tunnels; the guest molecules are densely packed along these tunnels. The host structure is hexagonal (P6,22;a = b L-8.2 A; c = 11.O .$) at ambient temperature, with the effective tunnel ‘diameter’ between ca.5.5 and 5.8 A. It is important to note that P6122 is a chiral space group, and aspects of the chirality of urea inclusion compounds are discussed in Section 2.5. Fig. 1The hexadecane-urea inclusion compound at ambient temperature, showing nine complete tunnels with van der Waals radii, viewed along the tunnel axis. The guest molecules have been inserted into the tunnels, illustrating orientational disorder (the positions of the guest molecules are not actually determined from X-ray diffraction data at ambient tem-perature). Structural compatibility between host and guest components is fundamental to most inclusion phenomena, and as a consequence, urea only forms inclusion compounds with guest molecules that are based on a sufficiently long alkane chain with only a limited degree of substitution of this chain allowed.Examples of appropriate guest molecules (Fig. 2) are alkanes and derivatives such as a,o-dihaloalkanes, diacyl peroxides, carboxylic acids, alkanones, a,o-alkane dicarboxylic acids, (a+ l),(o -1)-alkanediones and carboxylic acid anhydrides. In general, molecules containing a benzene ring or a cyclohex- ane ring do not form inclusion compounds with urea, pre- sumably because these structural components are too wide to fit inside the urea tunnel. On the basis of empirical observations, generalizations on the characteristic molecular features in the guest molecules that form inclusion compounds with urea have been establi~hed.2~596 As a direct consequence of the require- ment for size and shape compatibility between the host and guest components, urea inclusion compounds may be used in applications based on molecular separation, such as the separation of linear and branched alkanes from mixtures.Indeed, this was the motivation for much of the early research (particularly within the petrochemicals industry) on urea inclusion compounds, before the realization that zeolitic materials offer several advantages in such applications. Never- 280 Chemical Society Reviews, 1997, volume 26 theless, the use of urea inclusion compound formation as a method for isolating linear molecules is still used on the laboratory scale by synthetic organic chemists.XXfi X = F, CI, Br, I, CN, C02H 00 Fig. 2 Representative examples of guest molecules that form inclusion compounds with urea. The urea tunnel structure is an example of a soft host structure, and it has been shown (both by experiment and computer simulation) that the tunnels collapse if the guest molecules are removed from the inclusion compound; the urea then recrystallizes in its ‘pure’ crystalline phase, which does not contain empty tunnels. Clearly the instability of the ‘empty’ urea tunnel structure provides some limitations on the scope for applications of urea inclusion compounds.While we focus here on urea inclusion compounds that have the conventional urea tunnel structure (shown in Fig. 1) at ambient temperature, it is important to note that certain guest molecules induce significant changes in this host structure; in general, such changes occur when there is a commensurate relationship (see Section 2.2.1) between the host and guest substructures in the inclusion compound. Examples are the urea inclusion compounds containing 1,6-dibromohexane7 and seba- conitrile8 guest molecules. In a comprehensive series of investigations, Hollingsworth has shown9 that urea inclusion compounds containing (a+ l),(o -1)-alkanedione guest mol- ecules are also of this type, and exhibit an interesting diversity of commensurate superstructures. 2.2 Periodic structural properties 2.2 .I One-dimensional properties: the incommensurate structural nature An important fundamental property of solid inclusion com- pounds is the degree of structural registry between the host and guest substructures.In general, the guest molecules in conven- tional urea inclusion compounds are arranged in a periodic manner (repeat distance cg) along the host tunnels, with an incommensurate relationship between cg and the repeat distance (ch) of the urea molecules along the tunnel. In classical terms, the inclusion compound is incommensurate if there are no sufficiently small integers p and q for which p ch = q cg, and commensurate if sufficiently small integers p and q can be found to satisfy this equality. One consequence, with important physico-chemical implications, of an incommensurate struc- tural relationship between the host and guest components is that different guest molecules within a given tunnel sample a range of different environments with respect to the host structure.For several reasons, the classical (structural) definition of incommensurate and commensurate systems given above is far from satisfactory (a detailed discussion of this issue, and of commensurate and incommensurate behaviour in one-dimen- sional inclusion compounds in general, is given in ref. 10). To understand the commensurate versus incommensurate nature of one-dimensional inclusion compounds at a more fundamental level, new directions have led to the development10 of a commensurate/incommensurate classification that reflects a division in the energetic behaviour of the inclusion compounds within each category.Specifically, the classification (Fig. 3) is based on the magnitude of fluctuations in the average host- guest interaction energy per guest molecule as the guest substructure is moved along the tunnel (keeping the guest periodicity cg fixed). If these fluctuations are sufficiently small (i.e. within +E, where E is some physically meaningful energy term) the inclusion compound is considered to exhibit incom- mensurate behaviour, whereas if these fluctuations are suffi- ciently large (i.e. larger than +E) the inclusion compound is considered to exhibit commensurate behaviour.In the com- mensurate case, a significant energetic ‘lock-in’ between the host and guest substructures will occur for a specific position of the guest substructure relative to the host substructure, whereas for the incommensurate case, the energy of the inclusion compound is essentially independent of the position of the guest substructure relative to the host substructure. w Fig. 3 Schematic graphs illustrating the fluctuation in average host-guest interaction energy per guest molecule on moving the guest substructure along the host tunnel for: (a) an inclusion compound that exhibits incommensurate behaviour; and (b)an inclusion compound that exhibits commensurate behaviour. See ref. 10 for full details. Methodology has been developed’ for applying these new concepts to predict structural properties of one-dimensional inclusion compounds from knowledge of potential energy functions for the inclusion compound (with known host structure and fixed ch).Fundamental to this approach is the definition of an appropriate energy expression-the ‘charac-teristic energy ’--that directly indicates the relative energetic favourability of inclusion compounds yith different guest periodicities. The characteristic energy E(a,n) is defined by eqn. (I), where: n is the number of guest molecules within the host tunnel; a is the scaled guest periodicity cg/ch;the first guest molecule is located at position t = h along the tunnel; Eh(t) is the host-guest interaction energy for _an individual guest molecule at position t along the tunnel; Esuest(a)is the guest- guest interaction energy per guest molecule whec the scaled periodicity of the guest molecules is a;and E,,,,, is the intramolecular potential energy of the guest molecule.The optimum guest structure for the inclusion compound corre- sponds to minimum characteristic energy, and the methodology allows the following structural properties to be est+blished from t_he computed potential energy functions &(t), E,,,,,(a) and El,,,, for the inclusion compound of interest: (i) the optimum guest periodicity (c,); (ii) whether this value of cg corresponds to commensurate or incommensurate behaviour; (iii) the optimum conformation of the guest molecules within the host structure.Importantly, the methodology can handle tunnels of finite length, allowing the properties of ‘real’ one-dimensional inclusion compounds to be predicted directly. The methodology has been applied successfully’* to predict structural properties of alkane-urea inclusion compounds, giving results in excellent agreement with experimental ob- servations, and leading to new insights concerning the energetic properties of these inclusion compounds. Inter alia,the results demonstrate that, in the optimum structure of these incommen- surate inclusion compounds, the interaction between neighbour- ing guest molecules in the same tunnel is repulsive, in agreement with inferences from X-ray diffraction data. l3-12 We now consider the experimental investigation of incom- mensurateness in solid inclusion compounds, recalling that comparison of values of c, and ch measured from diffraction data is generally not a satisfactory approach.An alternative approach is based on recognizing that conventional crystals (including commensurate inclusion compounds) have three translation invariances, whereas an incommensurate one-dimensional inclusion compound has four translation invari- ances; the extra translation invariance corresponds to the shift of the guest substructure relative to the host substructure along the incommensurate direction (as discussed above, the energy of an incommensurate inclusion compound is, in principle, independ- ent of the shift of the guest substructure relative to the host substructure along this direction). Corresponding to each translation invariance in a crystal there is an acoustic phonon, and therefore an incommensurate one-dimensional inclusion compound should have four acoustic phonons and a com-mensurate inclusion compound should have three acoustic phonons.The additional acoustic mode in the incommensurate system is called the ‘sliding mode’, and observation of the sliding mode can be taken as direct experimental evidence for incommensurate behaviour of the inclusion compound. With this motivation, Brillouin scattering investigations l4 of the heptadecane-urea inclusion compound have provided direct evidence for a fourth acoustic mode, assigned as the sliding mode, thus substantiating the incommensurate nature of this inclusion compound. It is interesting to reflect that the new (energetic) definition of commensurate versus incommensurate behaviour discussed above is directly akin to the concept of a sliding mode for an incommensurate material.Before discussing the three-dimensional structural properties of urea inclusion compounds (Section 2.2.2), it is relevant to consider some of the wider consequences of the incommen- surate structural relationship between the host and guest substructures along the tunnel. Although the host and guest substructures possess different structural periodicities (as a consequence of the incommensurate relationship), these two substructures are not independent, since each substructure will exert an incommensurate modulation upon the other.The three- dimensional host substructure is best considered in terms of a ‘basic structure’ which is subjected to an incommensurate modulation through its interaction with the guest substructure; the basic structure can be described using conventional crystallographic principles (e.g. three-dimensional space group symmetry). In a similar way, the guest substructure can be considered in terms of an incommensurately modulated ‘basic structure’. The incommensurate modulations describe perturba- tions to the basic structures that arise as a result of host-guest Chemical Society Reviews, 1997, volume 26 281 interaction. A full discussion of these structural issues for the urea inclusion compounds is given else~here.~J~?~~ 2.2.2 Three-dimensional ordering of guest molecules We now consider the ordering of guest molecules in three dimensions within the urea host structure, focusing first on the positional relationship between guest molecules in adjacent tunnels.The inter-tunnel ordering of the guest molecules in urea inclusion compounds is conveniently described in terms of two parameters (Fig. 4): cg and A, (A, is the offset, along the tunnel axis, between the centres of mass of guest molecules in adjacent tunnels). Importantly, it is found that the nature of the inter- tunnel ordering depends critically on the functional groups present on the guest molecule, with different families of guest molecule exhibiting different characteristic modes of inter- tunnel ordering.Results for selected families of guest molecule at ambient temperature are summarized as follows: (i)alkane-urea inclusion compounds17- A, = 0 (independent of the value of c,), with c, increasing linearly with the number of CH2 groups in the alkane molecule; jzi) diacyl peroxide-urea inclusion compoundsI8-A, = 4.6 A (independent of the value of c,); (zii)a,o-dibromoalkane-urea inclusion compounds19- Ag depends on the value of c,, with A, and c, related by the exact relationship A, = c,/3; (zv) carboxylic acid anhydride- urea inclusion compounds20-A, = 0, with the exception of heptanoic anhydride-urea, for which A, = 2.3 A. It is important to emphasize that these well-defined correlations between the positions of guest molec$es in different tunnels (which are separated by more than 8 A) exist despite the fact (arising from the incommensurate relationship between the host and guest substructures) that there is no well-defined position- ing of guest molecules relative to the host substructure. Fig.4 Schematic two-dimensional representation of a urea inclusion compound, viewed perpendicular to the tunnel axis, indicating the definitions of cg, ch and Ag The complete three-dimensional packing arrangement of guest molecules within the urea tunnel structure can be understood by extending the A, concept into three dimensions. Thus, at ambient temperature, the basic guest structure in diacyl peroxide-urea inclusion compounds is monoclinic (probable space group C2), the basic guest structure in a,m-dibromoalk- ane-urea inclusion compounds is rhombohedra1 (probable space group R32) and the basic guest structure in alkane-urea inclusion compounds is hexagonal (probable space group P622).In some cases, these symmetries require disorder of the guest molecules. In the case of the diacyl peroxide-urea and a,o-dibromoalkane-urea inclusion compounds, the symmetry of the basic guest structure is lower than the symmetry of the basic host structure, and generally a given single crystal of these inclusion compounds contains different domains of the guest substructure---each domain has an identical packing of guest molecules, but has a different (although equivalent) orientation relative to the host structure, with the different domains related by rotation about the tunnel axis.It should be emphasized that while the diffraction data allow the average periodicity and symmetry of the basic guest structure to be determined, disorder of the guest molecules (see also Section 2.4) has made it impossible to actually solve the basic guest structure for any 282 Chemical Society Reviews, 1997, volume 26 conventional urea inclusion compound at ambient tempera- ture. As a consequence of the incommensurate relationship between the host and guest substructures in urea inclusion compounds, the symmetry of the composite inclusion com- pound cannot be described by a three-dimensional space group, but instead requires a four-dimensional superspace group.De- scriptions of the symmetry properties of urea inclusion compounds in superspace groups have been developed. 15 The development of a fundamental understanding of the factors that control the three-dimensional ordering of guest molecules in urea inclusion compounds is an issue of particular importance at present. The incommensurate modulations are undoubtedly important in establishing well-defined positional correlations between the guest molecules in adjacent tunnels, with the relative positioning of guest molecules in adjacent tunnels thus controlled by their mutual interaction with the urea molecules in the ‘tunnel wall’ between these tunnels. However, so far it has not been possible to determine the extent of the modulations in the host and guest substructures (which would require the structure of the composite inclusion compound to be solved in a superspace group as discussed above). Thus, at present, our understanding of the structural properties of the urea inclusion compounds is confined to the separate knowl- edge of the basic structures of the host and guest subsystems.2.2 3 Structural properties at low temperature All structural properties described so far have been at ambient temperature. At sufficiently low temperature, most conven-tional urea inclusion compounds undergo a phase transition which is associated, inter alza, with a change in symmetry of the basic host structure [hexagonal in the high-temperature phase, usually becoming orthorhombic in the low-temperature phase (Fig.5)] and a change in the dynamic properties of the guest molecules (see Section 2.4). These phase transitions have been investigated extensively for alkane-urea and a,o-dibromoalkane-urea inclusion com- pounds, both with regard to structural21-24 and dynamic25-29.23 aspects; in qualitative aspects, the behaviour of the alkane-urea and a,o-dibromoalkane-urea inclusion compounds with re- spect to these transitions is very similar. There have been various attempts30-32 to rationalize the phase transition in the alkane-urea inclusion compounds. The most recent of these approaches32 embodies certain crucial features of the experi- mental behaviour, and draws an analogy between the phase transition in alkane-urea inclusion compounds and the order- disorder phase transitions in alkali cyanide crystals.Specifi- cally, it has been proposed that, in the alkane-urea inclusion compounds, coupling between transverse acoustic phonons of the host structure and the orientational order of the guest molecules provides an indirect mechanism for orientational ordering of the guest molecules in the low-temperature phase. In spite of this recent progress, however, several aspects of these phase transitions remain to be understood, and the development of a fundamental understanding of the mechanism of these phase transitions is still one of the major challenges in this field. While structural aspects of the low-temperature phase are essentially the same for alkane-urea and a,o-dibromoalkane- urea inclusion compounds (with the low-temperature orthor- hombic basic host structure based approximately on the orthohexagonal description of the high-temperature phase), we have found that other urea inclusion compounds, such as decane- 1,lO-dicarboxylic acid-urea33 (Fig.6), exhibit more complicated superstructures in the low-temperature phase. Clearly, the exact nature of the structural distortion in the urea inclusion compounds depends critically on the type of guest molecule. Although the phase transitions studied so far for alkane-urea and a,o-dibromoalkane-urea inclusion compounds involve a distortion of the host structure, they are not associated with Fig.5 Structure of the 1,lO-dibromodecane-urea inclusion compound in the low-temperature phase at 108 K, viewed along the tunnel axis. There is a comparatively narrow distribution of guest molecule orientations, which correlates well with the distortion of the tunnel (see ref, 24 for full details). changes in the mode of three-dimensional packing of the guest molecules. Other urea inclusion compounds exhibit more complicated behaviour, and as an illustration, heptanoic anhydride-urea exhibits two phase transitions on cooling below ambient temperature.34 The first transition (at ca. 179 K on cooling) is associated only with a change in the three-dimensional packing of the heptanoic anhydride guest mole- cules (Ag = 2.3 A above 179 K; A, = 1.5 8, below 179 K), whereas the second transition (at ca.122 K on cooling) is associated with a distortion of the host structure as well as a further change in th? three-dimensional packing of the guest molecules (Ag = 0 A below 122 K). Extrapolating from these preliminary observations, the structural characterization of urea inclusion compounds containing different families of guest molecules below ambient temperature is expected to reveal a great diversity of structural and phase behaviour. In summary, the phase transitions described above reflect the cooperative behaviour of the host and guest components in the urea inclusion compounds. The guest molecules are usually dynamic at ambient temperature, and the average host structure (as determined from diffraction data) has a high symmetry that reflects the time-averaged distribution of guest molecules within it.At sufficiently low temperature, the extent of the dynamics of the guest molecules diminishes, and the guest molecules adopt a well-defined orientation with respect to the host; concomitantly, the host structure distorts to a lower symmetry that reflects the static distribution of the guest molecules (which may or may not be disordered). The intimate interplay of both host and guest components is crucial in controlling the overall behaviour of these phase transitions. 2.3 Local structural properties The discussion in Section 2.2 assumed the 'periodic approxi- mation' for the structural properties of crystalline solids.However, the periodic structural description (as probed by hr Fig. 6 Structure of the decane-1,IO-dicarboxylic acid-urea inclusion compound, determined at 173 K (low-temperature phase), viewed along the tunnel axis. There are four independent types of tunnel, with different modes of distortion-within each type of tunnel, the distribution of guest molecule orientations correlates well with the distortion of the tunnel (see ref. 33 for full details). diffraction-based investigations) is only an averaged represen- tation of the true system, as periodicity arises only on averaging the true structure over both space and time. To extend our understanding of the structural properties of a crystalline solid, it is necessary to go beyond this periodic description by investigating the distribution of local (spatial and/or temporal) structural features about this periodic average.In this regard, several experimental and computational approaches have been used to probe local structural aspects of urea inclusion compounds. Dynamic properties (i.e. temporal behaviour) are discussed in Section 2.4. Local structural properties that have been investigated include the conformational properties of the guest molecules and the interaction between adjacent guest molecules within the urea tunnel. Bromine K-edge EXAFS experiments have been carried out35 on urea inclusion compounds containing a,w-dibromoal- kane guest molecules [Br(CH2),Br; n = 7-11] with the principal aim of determining the Br...Br distance between adjacent guest molecules in the urea tunnel (motivated by the expectation that the repulsive interaction between adjacent guest molecules within the tunnel predicted in Section 2.2.1 should lead to an uncharacteristically short intermolecular Br.-Br distance).However, an accurate determination of the Bre-Br distance was not possible from data collected at ambient temperature and at 77 K (and also at 9 K for 1,lO-di-bromodecane-urea), as a consequence of dynamic disorder at high temperature and static positional disorder at low tem-perature. It is interesting to note that, as a consequence of the incommensurate relationship between the host and guest substructures in these inclusion compounds, no well-defined features arising from backscattering by atoms in the host substructure were observed in the EXAFS spectra. Chemical Society Reviews, 1997, volume 26 283 Raman spectroscopy has been used36 to probe conforma- tional properties of a,o-dihaloalkane guest molecules [X(CH2),X; n = 8 for X = C1; n = 7-11 for X = Br; n = 8 for X = I] in urea inclusion compounds.In particular, the C-X stretching vibrations were used to assess the relative amounts of trans and gauche end-groups as a function of: (i) the length (n) of the guest molecule; (ii) the identity of the terminal substituent X; (iii) temperature; and (iv) pressure. Inter alia, these investigations have shown: (i) there is no well-defined relation- ship between the proportion of end-groups in the gauche conformation and the length of the Br(CH2),Br guest molecules (the proportion of gauche end-groups is in the range 7-13% for n = 7-1 1 at ambient temperature); (ii) the proportion of end-groups in the gauche conformation at ambient temperature is ca.5 1% for Cl(CH2)&1-urea, ca. 7% for Br(CH2)8Br-urea and ca. 1% for I(CHz)gI-urea-thus, the proportion of gauche end- groups decreases as the size of the terminal substituent increases; (iii) the proportion of gauche end-groups [for Br(CHz),Br-urea inclusion compounds] increases slightly with increasing temperature; (iv) the proportion of gauche end- groups [for Br(CH2)1 lBr-urea] increases markedly with an increase in applied pressure. A molecular dynamics simulation of the 1,lO-dibromodec- ane-urea inclusion compound37 has investigated several local structural properties of the 1,l O-dibromodecane guest mole- cules at 300 K, providing results that corroborate well with the results from bromine K-edge EXAFS spectroscopy and Raman spectroscopy discussed above.The bromine radial distribution function (Fig. 7) determined from the molecular dynamics simulation indicates a broad distribution for the intermolecular Br-eBr distance, but considerably narrower distributions for intramolecular Br-C distances, in support of the conclusions from the bromine K-edge EXAFS results.35 The results from the molecular dynamics simulation also provide direct evidence that a small proportion of the 1,lO-dibromodecane guest molecules contain a gauche end-group, and indicate that the interconversion between gauche and trans end-group con-formations occurs on a timescale of the order of picoseconds within the urea tunnel structure.el !I Br ---Br !II' ; :, ,. I '. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 (Distancefrom Br) / A Fig. 7 Bromine radial distribution function for the 1,lo-dibromodecane-urea inclusion compound, determined from the molecular dynamics simulation discussed in ref. 37. Note the broad distribution of inter-molecular Br-Br distances. The conformational properties of the guest molecules in alkane-urea inclusion compounds have also been probed using vibrational spectroscopy (ref. 2 contains a comprehensive review, with detailed referencing).IR spectroscopy of the isolated CD2 rocking vibrations for selectively deuterated samples of tridecane-urea and nonadecane-urea suggest that the extent of gauche end-groups is below 3%.Similarly, Raman spectroscopy of the methyl group rocking modes [for alkanes CH3(CH2),CH3 with even values of n between 12 and 201 suggest that the concentration of gauche end-groups is low (ca. 5%). A recent Raman study of the methyl group rocking modes for alkane guest molecules (n = 6-10 and 17) has extended this work to encompass the low temperature phases, and has shown that the concentration of gauche end-groups (gt-conformation) is of the order of 5%, both at 298 and 90 K, independent of the length of the guest molecule. Evidence was also obtained for the existence of some amount (less than 5%)of end-groups with the tg-conformation.This paper also reported that the proportion of gauche end-groups increases significantly with increase of pressure. Interestingly, 13C NMR and 2H NMR studies of alkane-urea inclusion compounds have led to substantially higher estimates of the proportion of gauche end-groups in the alkane guest molecules. In a general sense, the fact that guest molecules trapped within solid host structures may be constrained to exhibit unconventional conformational properties can be exploited as a means of spectroscopic characterization of these conformations. For example, the recent discovery that the guest molecules in the 1,6-dibromohexane-urea inclusion compound exist ex-clusively with the bromine end-groups in a gauche conforma- tion has allowed the definitive characterization38 of the vibrational properties of this end-group conformation.Similar examples for thiourea inclusion compounds are discussed in Section 3.2. Finally, we consider end-group interactions for unsymmetric guest molecules [X(CH2),Y] in urea inclusion compounds. In principle, three different types of interaction between the end- groups of adjacent guest molecules are possible (Fig. 8): X.--X (head-head), X...Y (head-tail) and Y...Y (tail-tail). For two linear guest molecules of the type X(CH2),Y constrained to approach each other along the one-dimensional tunnel, it may be assumed that the guest-guest interaction is dominated by the interaction between the end-groups (i.e.X and/or Y); thus, as proposed by Holling~worth,~~ experimental measurements of the relative numbers of these different types of end-group interaction in a given inclusion compound can provide funda- mental information on the relative preferences for different types of functional group interaction. For many unsymmetrical guest molecules of the type X(CH2),Y in urea inclusion compounds,2939 the ratio of the number of X--.X interactions to the number of X--Y interactions and/or the ratio of the number of Y-.Y interactions to the number of X...Y interactions in the X(CH2),Y-urea inclusion compound can be determined from high-resolution solid state l3C NMR spectroscopy (in some cases both ratios can be measured from the NMR spectrum of a given inclusion compound, whereas in many cases only one of these ratios can be measured).An important advantage of this experimental strategy for deriving fundamental information on functional group interactions is the fact that the terminal functional groups on guest molecules in urea inclusion compounds are constrained to approach each other in a well- defined and controlled geometry, allowing interactions between the different types of functional groups to be compared on a systematic geometrical basis. -X(CHZ),Y ---X(CH,),Y ---Y(CH*),!X ---X(CH*),Y ---Y(CHz),X -Fig. 8 Schematic illustration of the X...X, X...Y and Y.-Y intermolecular interactions for unsymmetric guest molecules of the type X(CH*),Y inside a host tunnel structure.2.4 Dynamic properties A wide range of techniques have been applied to investigate the dynamic properties of urea inclusion compounds, including solid state NMR spectroscopy, incoherent quasielastic neutron scattering, EPR spectroscopy, molecular dynamics simulation, Raman spectroscopy, IR spectroscopy, dielectric loss spectros- copy and X-ray diffraction. The vast majority of these investigations have probed the dynamic properties of the guest 284 Chemical Society Reviews, 1997, volume 26 molecules, although some attention has also been given recently to the dynamics of the urea molecules. Early studies of guest motion in alkane-urea inclusion compounds by solid state NMR focused on 1H linewidth and second moment measurements and measurement of 1H spin lattice relaxation times.While these methods yielded con-siderable insights into the guest mobility and its temperature dependence, it is difficult from these techniques to derive well- defined and unambiguous information relating to the mecha- nism for the motion. For this reason, subsequent NMR studies have focused on 2H NMR spectroscopy of urea inclusion compounds containing fully deuterated or selectively deut- erated guests. These experiments probe the 2H quadrupole interaction parameters, and the technique can provide detailed mechanistic information for motions with characteristic time- scales between ca. 10-3 and 10-8 s. From variable-temperature 2H NMR investigations of the [2H34]hexadecane-urea inclusion compound,26 dynamic prop- erties of the guest molecules have been established over a wide temperature range, with the following mechanism deduced at ambient temperature: (i)rapid (K >, lo7 s-I) reorientation of the whole molecule about its long molecular axis (which is coincident, on average, with the tunnel axis); (ii)rapid torsional libration (with approximate amplitude k 25") about the penultimate C-C bond; (iii) rapid rotation of the CD3 group about the C-CD3 bond.There is a substantial change in the 2H NMR spectrum on crossing the phase transition temperature, suggesting that the phase transition is associated with an abrupt discontinuity in the motional freedom of the guest molecules; nevertheless, there is evidence for some amount of motion even below the phase transition temperature. Similar conclusions have also been reached from an independent 2H NMR investigation25 of [2H40]nonadecane-urea.For the [2H20]-I, 10-dibromodecane-urea inclusion compound, the re- sults from a 2H NMR study23 are in good agreement with those obtained for alkane-urea inclusion compounds with regard to the reorientational motion of the guest molecules about the tunnel axis; however, there are subtle differences with regard to both the change in dynamic properties on crossing the phase transition temperature and the extent of motion of the end- groups. The NMR techniques discussed above provide information on reorientational motions of the guest molecules, but do not yield direct information on translational motions.Considerable progress in understanding translational motions (in addition to reorientational motions) of alkane28 and a,~-dibromoalkane29 guest molecules in urea inclusion compounds has been made using incoherent quasielastic neutron scattering (IQNS), which probes motions with characteristic timescales between ca. 10-10 and 10-12 s. As a consequence of the large incoherent neutron scattering cross-section for 'H, IQNS studies of urea inclusion compounds containing [2H4]urea and guest molecules with natural isotopic abundances ensure that the incoherent neutron scattering arises predominantly from the guest mole- cules. Translational motions along the tunnel axis have been investigated separately from reorientational motions about the tunnel axis by studying semi-orientated polycrystalline samples in which the tunnel axes of all crystals are aligned parallel to each other.Separate IQNS experiments performed with the neutron momentum transfer vector parallel (QII spectra) or perpendicular (QL spectra) to the urea tunnel axis selectively probe translational motions of the guest molecules along the tunnel axis and reorientational motions of the guest molecules about the tunnel axis, respectively. For alkane-urea28 and a,o-dibromoalkane-urea29 inclusion compounds, quasielastic broadening is evident in the QI spectra in the high-temperature phase, implying that the guest molecules undergo rapid reorientational motions.This reor- ientational motion is diffusive in character (rather than a discrete jump motion), and can be modelled as uniaxial rotational diffusion in a onefold cosine potential. Rotational diffusion coefficients (ca. 0.3 X 10-12 s-1 for nonadecane- [2H4]~rea at 306 K) and other parameters relating to this dynamic process have been elucidated as a function of temperature. The QII spectra also exhibit substantial quasielastic broadening in the high-temperature phase, assigned to transla- tion of the alkane molecules along the tunnel. This motion has been modelled successfully as translational diffusion between rigid impermeable boundaries, and the diffusion coefficient and translation length have been determined as a function of temperatuFe.For nonade~ane-[~~H4]urea, the translation length is ca. 2.7 A at 306 K and ca. I. 1A at 160 K (just above the phase transition temperature); the translational diffusion coefficient at ambient temperature is ca. 1.5 X 10-5 cm2 s-1. Quantitative details relating to this translational motion are in good agreement with information on longitudinal motions of alkane guest molecules determined from analysis of X-ray diffraction intensitiesl7 (specifically, it was proposed that the alkane molecules undergo large amplitude motions $long the tunnel axis, with an average displacement of over 2 A for hexadecane at ambient temperature). It is perhaps remarkable that, despite the appreciable amount of translation of the guest molecules along the tunnel at ambient temperature, X-ray diffraction patterns nevertheless indicate both long-range intra- and inter- tunnel ordering of the guest molecules; it is thus very likely that the translations of neighbouring guest molecules within a given tunnel are highly correlated and that the translations of guest molecules in adjacent tunnels are also highly correlated.A detailed assessment of the dynamic properties of the guest molecules in the nonadecane-urea inclusion compound has been achieved recently from molecular dynamics computer simulations.40 Inter alia, the results of these simulations indicate that the interaction between adjacent guest molecules in the tunnel exerts an important influence on the translational and reorientational motions of the guest molecules, and demonstrate that the reorientational motions of the guest molecules about the tunnel axis are coupled with movements of the host structure.An important strength of this work has been the establishment of links between the results from the molecular dynamics simulations and results from the IQNS experimentsZ8 on nonadecane-urea described above. Although the a1 kane-urea and a,cu-di bromoal kane-urea inclusion compounds exhibit very similar dynamic behaviour, and apparently undergo the same type of phase transition, the dynamic behaviour of other urea inclusion compounds can differ substantially. Thus, the guest molecules in diacyl peroxide-urea inclusion compounds undergo a substantially different dynamic mechanism, and exhibit no evidence for a low temperature phase transition, at least down to liquid nitrogen temperature.The vast majority of research on the dynamic properties of urea inclusion compounds has focused on the motion of the guest molecules, although some studies of the dynamics of the urea molecules have also been carried out. 2H NMR investiga- tions of alkane-[2H4]urea41 and a,w-dibromoalkane-[2H4]urea42,23 inclusion compounds have demonstrated that, at sufficiently high temperature, the urea molecules undergo I 80" jumps about their C=O axes. For 1,lO-dibromodecane-[2H4]urea23 the jump frequency is ca. 5 X 106 s-I at 293 K. These 2H NMR investigations provide no evidence for reorientation of the NH2 groups about the C-N bond (on the 2H NMR timescale).The timescale for the 180"jump motion of the urea molecules is substantially longer (at the same temperature) than the timescale for the motions of the guest molecules described above, confirming that the 180" jumps of the urea molecules and the reorientational and translational motions of the guest molecules are not correlated. 2.5 Host-guest chiral recognition As discussed in Section 2.1, the host structure in the conventional urea inclusion compounds comprises a spiral hydrogen-bonded arrangement of urea molecules. The sym- Chemical Society Reviews, 1997, volume 26 285 metry of the basic host structure in any given single crystal is either P6122 (the inclusion compound contains only right- handed spirals of urea molecules) or P6522 (the inclusion compound contains only left-handed spirals of urea molecules).This chirality of the urea tunnel structure is generated spontaneously during crystal growth of the inclusion com-pound, and represents an example of chirality being introduced into a crystal by spontaneous assembly of achiral molecules into a chiral packing arrangement. Clearly, chiral host structures can exert an important influence on the structural and chemical properties of chiral guest molecules. As the R-guest/ (P6122)-host and S-guest/(P6,22)-host inclusion compounds have a diastereoisomeric relationship, they should generally differ in energy, and a given crystal of a chiral host should therefore have a preference for incorporating one particular enantiomer of a chiral guest.An extensive series of experimental investigations by Schlenk43 has demonstrated that inclusion of chiral guest molecules within the urea tunnel structure can be associated with a significant degree of chiral recognition. In addition to experimental investigations of this phenomenon, computational investigations can provide detailed insights into the character- istics of host-guest interaction that underlie this chiral recogni- tion in urea inclusion compounds. A recent computational of 2-bromoalkane-urea inclusion compounds has demonstrated (Fig. 9) a clear preference for the R-enantiomer of 2-bromoalkane guest molecules within the P6122 urea tunnel structure, with the proportion of R-2-bromoalkane guest molecules at 300 K predicted to be ca.0.75 for 2-bromo- tridecane-urea and ca. 0.82 for 2-bromotetradecane-urea. Interestingly, (see Fig. 9), for the lowest energy conformation (Br trans-CH3 gauche) of the 2-bromotridecane guest molecule within the urea tunnel, the same enantiomer (R) of the guest is preferred at all positions along the tunnel [note that for other conformations (for example, Br gauche-CH3 trans) of the 2-bromoalkane guest molecules, the R enantiomer is preferred at some positions and the S enantiomer is preferred at others (Fig. 9)]. It is important to note that, in assessing the enantiomeric excesses for incommensurate inclusion com-pounds, such as 2-bromoalkane-urea inclusion compounds, it is necessary to consider the characteristics of the host-guest interaction as a function of the position of the guest molecule along the host tunnel.ZIA Fig. 9 Host-guest interaction energy [&~,r(z)] for 2-bromotridecane guest molecules as a function of position (z) along the tunnel of the P6,22 urea host structure (for 0 G z < 46). The R,, R,, S, and S, types of 2-bromoalkane molecule are defined in the Newman projections [R and S enantiomers; Br gauche-CH3 trans conformation (8) and Br truns-CH3 gauche conformation (t)]. 2.6 Generation of orientationally well-ordered molecular assemblies Inclusion within one-dimensional tunnel structures may be exploited as a means of generating an orientationally well- 286 Chemical Society Reviews, 1997, volume 26 ordered ensemble of molecules, which may be difficult to achieve in other phases.An example based on this fact concerns measurement of the orientation of the electronic transition dipole moment for conjugated polyenes; this property is important in relation to the use of these molecules in non-linear optoelectronics and other applications (including their use as probes of biophysical systems). Simple theoretical approaches to predict this property have generated differing results (some suggesting that the transition dipole moment is essentially parallel to the long axis of the molecule, others suggesting an angle of 30" with respect to this axis), and experimental verification of these predictions has been hindered by the difficulty of preparing perfectly orientated samples of the polyenes.This problem has been addressed45 by constraining these molecules as guests within the urea tunnel structure, thus ensuring that the molecular axes of all guests in a given single crystal are parallel and orientationally well-defined with respect to the external morphology of the crystal. Specifically, octadeca-9,11,13,15-tetraenoicacid was considered as a dilute 0 HO guest within the hexadecane-urea inclusion compound (dilution ensures that absorbance is low and that exciton effects are eliminated). Polarized fluorescence excitation spectra of a single crystal of this material have shown that the transition dipole does not lie strictly along the molecular axis, but at an angle of ca.20 k 1" with respect to this axis. On taking into account the effects of the surrounding medium for the guest in the urea inclusion compound, this value is modified to ca. 15 & 1O for the isolated molecule. This result has important implications with regard to applications of these chromophoric materials. In general terms, it is clear that there are important prospects for exploiting uni-directional tunnel host structures in applications of this type, in which a highly anisotropic (uni- directional) orientation of guest molecules is required for the measurement of electronic or other properties. 3 Thiourea inclusion compounds 3.1 An introduction to thiourea inclusion compounds Shortly after the discovery of urea inclusion compounds, it was found that thiourea also forms a tunnel host structure (Fig.10) in the presence of appropriate guest molecules. The thiourea tunnels have a larger cross-sectional area than those in urea inclusion compounds, so the urea and thiourea host structures tend to incorporate different types of guest molecule. For example, thiourea forms tunnel inclusion compounds with cyclohexane and some of its derivatives, ferrocene and other organometallics, and certain compounds containing a benzene ring. Such guest molecules do not generally form inclusion compounds with urea. In general, the host structure in thiourea inclusion compounds is either rhombohedral or monoclinic (structural properties of thiourea inclusion compounds are discussed in detail in ref.2). For guest molecules that are substantially isotropic in shape [for example cyclohexane (see Fig. lo), chlorocyclohexane and ferrocene], the host structure is typically rhombohedral at ambient temperature, and the guest molecules usually exhibit substantial disorder. In many of these cases, the rhombohedral structure transforms to a monoclinic structure at low tem- perature. Planar guest molecules (for example, 2,6-diethyl- naphthalene and cyclo- 1,5-0ctadiene), on the other hand, tend to favour the monoclinic host structure at ambient temperature; within this structure, the guest molecules are constrained to adopt an ordered arrangement (lowering the symmetry from rhombohedral to monoclinic is associated with a deformation of the tunnel, which restricts the orientational freedom of the guest molecules).Fig. 10 The cyclohexane-thiourea inclusion compound at ambient tem- perature, showing ten complete tunnels with van der Waals radii, viewed along the tunnel axis. The guest molecules have been inserted into the tunnels, illustrating orientational disorder (the positions of the guest molecules are not actually determined from X-ray diffraction data at ambient temperature). In contrast to the tunnel in conventional urea inclusion compounds, which is relatively cylindrical in the sense that there is only a small fluctuation in tunnel diameter on moving along the tunnel, the thiourea tunnel has prominent bulges (diameter ca.7.1 A) and constrictions (diameter ca. 5.8 A) at different positions along the tunnel (Fig. 11). As a consequence, it is often more appropriate to regard the thiourea tunnel structure as a ‘cage’ type host rather than a ‘tunnel’ type host, and indeed many properties of thiourea inclusion compounds can be interpreted more directly on this basis. The guest molecules in thiourea inclusion compounds generally occupy preferred sites along the tunnel, corresponding to one guest molecule per cage (i.e. two guest molecules per unit repeat distance of the thiourea structure) and a stoichiometric guest/ thiourea molar ratio of 1/3. This leads to the commensurate structural relationship cg/ch = 1/2 (see Section 2.2.1 for definitions of cg and ch).7.5 T 0.’50 2.50 4.50 6.50 8.50 10.50 12.50 ZIW Fig. 11Minimum tunnel diameter d,,, as a function of position (z) along the tunnel for the urea and thiourea tunnel structures. In both cases, the range of z shown in the graph represents just over one lattice period of the host structure along the tunnel. 3.2 Conformational properties of guest molecules As a consequence of the structural selectivity between the host and guest components, inclusion within a solid host structure can often serve to select an uncharacteristic conformation of the guest molecules. In general terms, this can be important in allowing spectroscopic characterization of conformations that may not be significantly populated in dispersed phases or in the ‘native’ crystalline state of the molecule.In addition, the attainment of uncharacteristic conformations may open up reaction pathways for constrained guest molecules that may be improbable for the same molecules in their normal conforma- tional state. A dramatic illustration of the constraints that a host structure can impose on the conformational properties of guest molecules is provided by monohalocyclohexane (C6H11X; X = C1, Br, I) guest molecules in the thiourea tunnel structure. For mono- halocyclohexanes in the liquid and vapour phases, the dynamic equilibrium between the equatorial and axial conformations favours the equatorial conformation, and in the solid state (at sufficiently low temperature or high pressure) these molecules exist almost entirely as the equatorial conformation.On the other hand, when included as guest molecules within the thiourea tunnel structure, C6H1 lC1, C6H1 lBr and C6H1 1I exist predominantly in the axial conformation; these results have been established from IR, Raman and high-resolution solid state l3C NMR techniques (ref. 2 contains a comprehensive list of references for this work). From 13C NMR results,46 the mole fractions of the equatorial conformations of these guest molecules in the thiourea tunnel structure are in the range 0.05-0.15, in contrast to the corresponding values (0.75-0.8 1) observed in CFC13-CDC13 (3 : 1) solution (the quoted values refer to temperatures of 159-220 K). The l3C NMR results also demonstrate that a ring inversion process occurs for these guest molecules inside the thiourea tunnel structure at sufficiently high temperature.Bromine K-edge EXAFS spectroscopy, which provides a direct measurement of the intramolecular Br.-C(3) distance (ca.3.27 A) [see Fig. 12(a)],confirms that the axial conformation of bromocyclohexane predominates within the thiourea tunnel structure.47 la 1.. 4.3wMr---Br axial equatorial (b)@14i-CI aFi3.4A di axi al diequatorial Fig. 12 (a) Comparison of Br-C(3) distances in the axial and equatorial conformations of bromocyclohexane; (6) Comparison of Br-Cl distances in the diaxial and diequatorial conformations of trans-1 -bromo-2-chloro-cyclohexane [note that measurement of the Br-C(3) distance, as in (a), also allows the diaxial and diequatorial conformations to be distinguished].The quoted distances have been computed for idealized molecular geometries. For guest molecules C6H11X with X = CH3, NH2, OH, the equatorial conformation is preferred (mole fraction ca. 0.824.97) inside the thiourea tunnel ~tructure.~6 The confor- mational properties for these guests resemble those for the same molecules in solution, and contrast markedly with the behav- iour, discussed above, for monohalocyclohexane guest mole- cules in the thiourea tunnel structure. There is also a marked contrast between the conformational properties of mono-halocyclohexane guest molecules in thiourea and in various zeolitic hosts, within which the equatorial conformation predominates.Certain disubstituted cyclohexanes also exist in uncharac- teristic conformational states within the thiourea tunnel struc- ture. For the trans-1-bromo-2-chlorocyclohexane-thioureain-clusion compound [Fig. 12(b)], the intramolecular Br.-.Cl and Br--C(3) distances of ca. 4.50 and 3.27 A determined from bromine K-edge EXAFS spectra47 demonstrate clearly the preference for the diaxial conformation of the guest molecule. Chemical Society Reviews, 1997, volume 26 287 In contrast, the diequatorial conformation is preferred in dispersed phases. The theoretical approach discussed in Section 2.2.1 for investigating the structural properties of one-dimensional inclusion compounds has also been applied48 to assess, from first principles, the preferred conformation of chlorocyclohex- ane guest molecules within the thiourea tunnel structure.For axial-chlorocyclohexane-thiourea, the optimum guest period- icity corresponds to a = +, representing commensurate behaviour and corresponding to a lower characteristic energy than any guest periodicity for equatorial-chlorocyclohexane-thiourea, This predicted preference for the axial conformation is in direct agreement with the experimental results discussed above. In essence, axial-chlorocyclohexane can be packed more efficiently (smaller a) than equatorial-chlorocyclohexane within the constrained environment of the thiourea tunnel, and this contributes [through the factor l/ain eqn.(l)] to the more favourable characteristic energy for the axial conformation. The optimum guest period (cg = ach = c& = 6.24 A) predicted for axial-chlorocyclohexane-thiourea is in good agreement with information inferred from X-ray diffraction data. 3.3 Application in non-linear optics The potential to exploit inclusion phenomena in the field of non- linear optics has received considerable attention in recent years. We focus here on second harmonic generation (SHG), which involves doubling the frequency of light as it passes through a material. Materials that exhibit SHG are important in many device applications (including extending the frequency range of lasers) and have an important role in the field of optoelectronics. For a material to exhibit SHG, the component molecules must have a high value of the second order molecular hyper- polarizability (@),and in addition the molecules must aggregate in a non-centrosymmetric arrangement.Molecules with large fi often have large degrees of intramolecular charge transfer and usually possess a large dipole moment in their ground state; however, there is a strong tendency for the crystal structures of such molecules to be centrosymmetric. There is therefore substantial impetus to develop ways to induce these molecules into non-centrosymmetric environments, and an attractive prospect is to include them as guest molecules within appro- priate host materials. Thus, parallel alignment of guest mole- cules (with high values of 6) within tunnel host structures has been particularly exploited in this regard, and has included successful applications involving thiourea inclusion com-pounds.49 In this work, thiourea inclusion compounds contain- ing appropriate organometallic guests [for example, (rf-C6Hs)Cr(CO),] were shown to exhibit pronounced SHG.These organometallics possess large values of @, but their ‘native’ crystalline phases are centrosymmetric and are there- fore inactive for SHG. For the thiourea inclusion compounds, it was shown that the SHG arises predominantly from the organometallic guests rather than the thiourea molecules (which also have significant p). The structures of the inclusion compounds are non-centrosymmetric, in accord with the idea that dipole organization of the guest molecules should be favoured both within and between tunnels.It is interesting that the host structure in many other thiourea inclusion compounds (for example, cyclohexane-thiourea) is centrosymmetric, and the results here illustrate that the structure of a given host material can differ, often substantially, depending on the identity of the guest molecules within it. Such observations are particularly prevalent for hosts of the soft type. 4 Concluding remarks It is clear that urea and thiourea inclusion compounds exhibit a wide range of interesting and important fundamental phys- icochemical phenomena, and that the application of a wide range of experimental and computational techniques has been essential in the endeavour to understand these properties.288 Chemical Society Reviews, 1997, volume 26 However, although significant progress has been made in recent years in expanding our fundamental understanding of these inclusion compounds, there is still a great deal to be learned. It is now generally accepted that the properties of urea and thiourea inclusion compounds depend critically on the nature of both the host and guest components, and it is therefore not expected that universality will be observed in any particular property across all urea inclusion compounds or across all thiourea inclusion compounds. At best, a set of inclusion compounds with a given host and a closely related family of guest molecules (e.g. a homologous series, in the case of urea inclusion compounds) may be found to exhibit common behavioural trends, although even in these cases anomalous members of such families are often observed. When studied in sufficient depth, each particular inclusion compound of urea or thiourea is best regarded as an individual entity with its own characteristic set of properties, even though the host structures of many urea and thiourea inclusion compounds are essentially identical.At present, materials applications based upon urea and thiourea inclusion compounds (and solid organic inclusion compounds in general) are comparatively scarce in comparison, for example, to the wide range of applications that exploit the properties of microporous inorganic materials.Nevertheless, the development of a fundamental understanding of the structural, dynamic and chemical properties of urea and thiourea inclusion compounds will lead the way towards the future design and development of applications of these materials. With the realization that urea and thiourea inclusion compounds exhibit a wide range of interesting properties and phenomena (with the interplay between these properties only at the very earliest stages of being understood) and the recognition that a detailed understanding of these systems will emerge only from the combined knowledge acquired from a wide range of experimental and computational techniques, there are exciting prospects for the continued study of these materials long into the future.5 Acknowledgements I am grateful to Professor Sir John Meurig Thomas for introducing me to this fascinating subject, and to many other collaborators, particularly Professor Mark Hollingsworth and Dr FranGois Guillaume, for many stimulating discussions. The contributions of my own research group to this field pay tribute to the endeavours of the many research students, postdoctoral colleagues and collaborators, mentioned in the references, with whom it has been a pleasure to be associated. The EPSRC, the Nuffield Foundation, NATO, the British Council and Ciba Speciality Chemcials Ltd are thanked for financial support. 6 References Comprehensive Supramolecular Chemistry, ed. J.-M. Lehn, J. L. Atwood, J. E. D. Davies, D. D. MacNicol and F.Vogtle, Pergamon Press, Oxford, 1996, vol. 1-11. M. D. Hollingsworth and K. D. M. Harris, Comprehensive Supra- molecular Chemistry, ed. D. D. MacNicol, F. Toda and R. Bishop, Pergamon Press, Oxford, 1996, vol. 6, p. 177. A. E. Smith, Acta Crysrallogr., 1952, 5, 224. K. D. M. Harris and J. M. Thomas, J. Chem. Soc., Faraduy Trans., 1990, 86, 2985. R. W. Schiessler and D. Flitter, J. Am. Chem. SOC.,1950, 74, 1720. L. C. Fetterly, in Non-stoichiometric compounds, ed. L. Mandelcom, Academic Press, New York, 1964, p. 491. M. D. Hollingsworth, K. D. M. Harris, J. D. Chaney, U. Wemer-Zwanziger, J. C. Huffman, S. P. Smart and B. D. Santarsiero, manuscript in preparation. M. D. Hollingsworth, B. D. Santarsiero and K. D. M. Harris, Angew.Chem., Int. Ed. Engl., 1994, 33, 649. M. E. Brown and M. D. Hollingsworth, Nature, 1995, 376, 323; M. E. Brown, J. D. Chaney, B. D. Santarsiero and M. D. Hollingsworth, Chern Mater, 1966, 8, 1588, M D Hollingsworth, M E Brown, A C Hillier, B D Santarsiero and J D Chaney, Science, 1996, 273, 1955 10 A J 0 Rennie and K D M Hams, Proc Roy Soc A, 1990, 430, 615 11 A J 0 Rennie and K D M Harris, J Chern Phys , 1992, 96, 71 17 12 I J Shannon, K D M Hams, A J 0 Rennie and M B Webster, J Chem Soc ,Furuduy Trans, 1993,89, 2023 13 F Laves, N Nicolaides and K C Peng, 2 Kristallogr , 1965, 121, 258 14 D Schmicker, S van Smaalen, J L de Boer, C Haas and K D M Harris, Phys Rev Lett, 1995,74, 734 15 S van Smaalen and K D M Harris, Proc Roy Soc A, 1996, 452, 677 16 R Lefort, J Etnllard, B Toudic, F Guillaume, T Breczewski and P Bourges, Phys Rev Lett, 1996, 77, 4027 17 K Fukao, H Miyaji and K Asai, J Chern Phys , 1986, 84, 6360 18 K D M Hams and M D Hollingsworth, Proc Roy Soc A, 1990,431, 245 19 K D M Harris, S P Smart and M D Hollingsworth, J Chem Soc, Faruduy Trans , 199 1,87, 3423 20 I J Shannon, N M Stainton and K D M Hams, J Muter Chern , 1993,3, 1085 21 Y Chatani, Y Taki and H Tadokoro, Actu Crystullogr , Sect 5,1977, 33, 309, Y Chatani, H Anraku and Y Taki, MoZ Cryst Liq Cryst, 1978,48,219 22 K D M Harris, I Gameson and J M Thomas, J Chern Soc ,Faraday Trans, 1990, 86, 3135 23 A E Aliev, S P Smart, I J Shannon and K D M Harris, J Chern Soc ,Faruduy Trans, 1996,92, 2179 24 L Yeo and K D M Hams, Actu Crystullogr ,Sect B, in press 25 H L Casal, D G Cameron and E C Kelusky, J Chem Phys , 1984, 80, 1407 26 K D M Harris and P Jonsen, Chern Phys Lett, 1989, 154,593 27 A El Baghdadi, E J Dufourc and F Guillaume, J Phys Chem , 1996, 100, 1746 28 F Guillaume, C Sourisseau and A -J Dianoux, J Chzrn Phys (Paris), 1991,88, 1721 29 S P Smart, F Guillaume, K D M Harris and A -J Dianoux, J Phys Condens Matter, 1994,6, 2169 30 N G Parsonage and R C Pemberton, Trans Faraday SOL , 1967,63, 311 31 K Fukao, J Chern Phys , 1990,92,6867 32 R M Lynden-Bell, MoZ Phys , 1993,79, 313 33 L Yeo, K D M Harris and F Guillaume, J Solid State Chem , 1997, 128, 273 34 I J Shannon, K D M Hams, F Guillaume, E H Bocanegra and E J MacLean, J Chem Soc ,Chern Cornrnun , 1995,2341 35 I J Shannon, K D M Hams, A Mahdyarfar, P Johnston and R W Joyner, J Chern Soc ,Faraduy Trans, 1993,239, 3099 36 S P Smart,A ElBaghdadi,F GuillaumeandK D M Harris,J Chem Soc ,Faruduy Ttuns , 1994,90, 1313 37 A R George and K D M Harris, J Muter Chern , 1994,4, 1731 38 L Elizabe, S P Smart, A El Baghdadi, F Guillaume and K D M Har-ris, J Chern Soc ,Faruday Trans, 1996,92,267 39 M D Hollingsworth and N Cyr, Mol Cryst Lzq Cryyt , 1990, 187, 135 40 M Souaille, F Guillaume and J C Smith, J Chern Phyr , 1996, 105, 1516, 1966, 105, 1529 41 N J Heaton,R L VoldandR R Vold,J Am Chem SOC,1989,111, 321 1, J Magn Reson, 1989, 84, 333 42 A E Aliev, S P Smart and K D M Harris, J Muter Chem , 1994,4, 35 43 W Schlenk, Justus Liebigs Ann Chern , 1973, 1145, 1156, 1179, 1195 [see also R Arad-Yellin, B S Green, M Knossow and G Tsoucans, Inclusion Compounds, ed J L Atwood, J E D Davies and D D MacNicol, Academic Press, New York, 1984, vol 3, p 2631 44 L Yeo and K D M Hams, Tetrahedron Asymmetry, 1996, 7, 1891 45 Q Y Shang, X Dou, and B S Hudson, Nature, 1991, 352,703 46 A E Aliev and K D M Hams, J Am Chern Soc, 1993, 115, 6369 47 I J Shannon, M J Jones, K D M Harris, M R H Siddiqui and R W Joyner, J Chem Soc ,Faraduy Trans, 1995,91, 1497 48 P A Schofield, K D M Harris, I J Shannon and A J 0 Rennie, J Chern Soc ,Chern Cornmun , 1993, 1293 49 W Tam, D F Eaton, J C Calabrese, 1 D Williams, Y Wang and A G Anderson, Chem Mater , 1989, 1, 128 Received, 2I st November 1996 Accepted, 25th March 1997 Chemical Society Reviews, 1997, volume 26 289

ISSN:0306-0012    

DOI:10.1039/CS9972600279

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Speciation of trace metals in the environment With the recognition of the importance of metal speciation in many areas of environmental study, there has been a tremendous increase in research to aid our understanding of the role of specific metal species in terms of toxicity, bioavailability, bioaccumulation, mobility and persistence in the environment. In turn this has led to the development of analytical techniques that are both sensitive and specific to meet this challenge. This review provides an overview of trace metal speciation in the environment, an insight into some of the analytical techniques that are suitable for metal speciation studies on environmental samples, and a more detailed look at some selected elements for which the environmental chemistry has now been better charac- terised.1 Introduction The determination of distinct chemical species, often referred to as speciation analysis, is now widely acknowledged to be of vital importance in environmental chemistry. The term chem- ical speciation may be used to encompass both functionally defined speciation, that is, the determination of species that are, for example, available to plants or present as 'exchangeable' forms, and operationally defined speciation which refers to the determination of 'extractable forms' of an element. Whilst it is often possible to define a particular compound or oxidation state when dealing with solutions, for example, natural waters, it is far more difficult to characterise the actual chemical form of an element in solids such as soils and sediments.Thus, speciation tends to be defined somewhat differently by workers to reflect their field of study. However, one of the most comprehensive formal definitions of speciation is the one recommended by the International Union of Pure and Applied Chemistry (IUPAC) which states that speciation is 'the process yielding evidence of the atomic or molecular form of an analyte'. The determination of such specific chemical entities is of course not new to Steve Hill is Professor of Analytical Chemistry at the University of Plymouth where he is currently Head of the Department of Environmental Sciences. Until recently he was Chairman of the Atomic Spectroscopy Group of the RSC and currently serves on several of the Society's commit- tees.In addition he is on the Editorial Boards of The Ana- lyst, JAAS, and Analytical Communications and is Gen-eral Editor of Atomic Spectrom- etry Updates. His research in- terests are varied and to date he has published well over 100 scientific publications and pre- sented over 150 conference presentations, many in the area of metal speciation. NO;-, NH4+ and NH3 where ihe nitrogen is characterised into its most environmentally important forms is long established. However, the characterisation of metals does not have the same history, although with the increased awareness of the im- portance of speciation in terms of toxicity, bioavailability, bioaccumulation, mobility and persistence in the environment, a range of sensitive yet specific analytical techniques have now been developed to address a wide range of complex prob- lems.It should be stressed that the identification and quantification of specific metal species in environmental samples is no longer an academic curiosity or indeed confined to a few specialist laboratories. As more information is collected on the im- portance of specific species, the determination of total concen- trations of the metal in a sample is increasingly inadequate for many purposes since this information alone will not reflect, for example, the toxicity of the sample. Thus, the importance of metal speciation is now incorporated in new legislative requirements which often list specific species.The EC Council Decision 75/437/EEC (Marine Pollution from Land Based Sources) specifically includes As, Cd, Cr, Cu, Pb, Ni and Zn and their compounds in the list of substances requiring strict control. Other Council Directives mention Cd and Hg and their compounds (80/68/EFC-Groundwater) and As, Be, Cd, Cr, Hg, Sb and T1 and their compounds (78/319/EEC-Toxic and dangerous wastes).' In some cases individual species are listed, for example, in the UK tributyltin is listed specifically in terms of water quality. Further details on some important elements are given below, although clearly the introduction of such legisla- tion has provided further impetus to develop reliable techniques to detect and quantify these species and presented a consider- able challenge for the analytical chemist.2 The practice of metal speciation There are a number of important environmental factors which may affect the speciation of metals in the environment and these should be borne in mind during speciation studies. One of the most important of these is the prevailing redox conditions which not only determine the oxidation state of some metals, but may also influence the bioavailability and toxicity of the element. For example, Fe" and MnI1 are soluble in natural waters deficient in oxygen but will precipitate out at higher oxidation states. In other cases photoreduction may be important, and changes in pH may shift the acid-base equilibrium and redox conditions. In addition there are also a number of practical difficulties associated with metal speciation studies.The first, and perhaps most obvious problem, is to obtain a representative sample in which the integrity of the species of interest remains intact. Thus the collection of samples can lead to errors if suitable precautions are not taken. It is obviously necessary to ensure that the container used is not made of the same material as the analyte and some plastics are more suitable than others depending on the metals to be determined. Mercury, for example, is known to escape from some types of plastic container. Although the preservation of samples is frequently achieved by the addition of acid, as mentioned above, this may not be appropriate for speciation studies since it may lead to Chemical Society Reviews, 1997, volume 26 291 changes in oxidation state and alterations in the speciation.Filtration may also alter the species present and lead to a redistribution between the mobile and stable metal forms, as for example in the case of aluminium. Storage of some samples may add to the problems and it is normally recommended that analysis should take place as soon after sampling as possible. Some species, e.g. the alkyltin species and arsenobetaine are photosensitive and so if exposed to light following collection a radical induced change in the speciation may occur. In other cases specific precautions may aid the collection of samples for speciation work. Chloride, for example, should be added to solutions where selenium speciation is to be performed since this will prevent oxidation of SeIV to SeVI. In most cases samples should be kept at low temperature since this will reduce the risk of thermal rearrangement or thermal degradation.Often placing the sample in a refrigerator at 4 "C is sufficient, but occasionally freezing may be required. The next stage to consider is the preparation of the sample for analysis. Quantitative extraction of the species without altera- tion to the speciation can prove to be very difficult. For liquid samples, the problem is not so bad. The sample may be centrifuged or filtered and if a pre-concentration procedure is not required, it may be analysed directly. For solid samples, the extraction may be far more problematic.Clearly an acid digestion is inappropriate and so the species must be taken into solution by another method. There are a number of different approaches in common use. For biological samples, an extraction using an organic solvent (often methanol or toluene) is fairly common. The use of an ultrasonic bath increases the efficiency and speed of the extraction. Soxhlet extraction is a more lengthy method and care must be taken not to lose labile species or to change the speciation at elevated temperatures. An alternative method that has been employed by some laboratories is the use of enzymes and a Potter homogeniser to disrupt the samples. Trypsin has been used successfully to extract arsenic species from proteinaceous materials.The use of ammonium carbonate to obtain the optimum pH for trypsin ensures the best possible extraction efficiency. For cereals or plant materials, an alternative enzyme is required. The use of cellulase has been found to be efficient at digesting this type of material. To prepare soil and sediment samples for speciation studies, 5-20 g (to obtain a representative subsample) of an unground < 2 mm particle size air dried sample should be used. The use of fresh, field moist soils is not considered a practical possibility for general use.* However, the use of dried material may affect the soil speciation and this should be considered when interpreting the results. In addition, storage of dried samples may be a problem, particularly if anoxic sediments are being used.Separation by sieving may also discriminate against some elements that are associated with the larger particle size fractions and so the use of a 'conserved' element such as aluminium may be required to act as an internal standard.2 Once a representative sample has been obtained different extractants can be used not only to assess the bioavailability of particular metals but also to isolate and determine elements bound or associated with defined soil fractions or phases. An improve- ment in the specificity of the extraction can be achieved by using a sequential extraction scheme in which the residue from one extract is extracted by the next extractant in the ~equence.~ Typical extracting reagents include calcium chloride, ammo- nium acetate, sodium nitrate and ammonium nitrate solutions.To extract organometallics organic solvents are often the method of choice. Occasionally tropolone may be used since it will aid very efficient extractions although this in itself may be problematic since it may also extract unwanted concomitants that could affect the analytical methodology, if for example, chromatography is used. Supercritical fluid extraction is also gaining in popularity for organometallic extractions in specia- tion studies since it yields high extraction efficiencies under relatively mild conditions. Thus, although relatively new, this approach would seem to offer much potential for the future. Solid phase extraction has also achieved great popularity with some laboratories.One of the major challenges in speciation studies is that often the analytical scientist is being asked to quantify species present at very low concentrations, well below the detection limits of current techniques. Under such circumstances, it is often necessary to perform some kind of preconcentration procedure. A variety of methods are available including the use of liquid- liquid extractions, freeze drying, gentle evaporation (assuming none of the species are labile) and preconcentration using an ion exchange resin. Of these different methods, the liquid-liquid extractions probably yield the best preconcentration factors, with factors of over 100 sometimes being possible.This is especially useful for liquid samples such as seawater. In addition, the process removes the vast majority of the matrix which could otherwise interfere with the determination. As with all sample preparation procedures, care must be taken that analyte loss does not occur. More accurate results may be obtained if standard solutions of the analytes of interest are treated in an identical manner, since this should overcome many of the problems associated with different species of an analyte having a different affinity for the extractant. A final practical consideration associated with speciation studies is that of quality control and quality a~surance.~~~ Most competent laboratories will use either certified reference materials (CRMs) or spiking/recovery experiments to ensure the accuracy of their determinations.However, for speciation studies this is not straightforward. At the moment very few CRMs with species specific data are available and so finding one that closely matches the matrix of interest is difficult. This is an area where the agencies producing CRMs are paying increasing attention and a number of new programmes to produce a range of reference materials specifically for specia- tion work are now under way. These include roadside dust and rainwater for lead species, fish tissues for mercury species, various fish and mussel tissues and algae for arsenic species and fish tissue and sediments for tin species. A summary of available reference materials for speciation studies is presented Table 1 Certified reference materials available for metal speciation analysis DBT and TBT in marine sediment (PACS-l)-NRCCa TBT in fish tissue (NRES 1I)-NIESh Hg and MeHg lobster hepatopancreas (TORT 1)-NRCC MeHg in dogfish muscle (DORM I)-NRCC DBT and TBT in coastal sediment (CRM 462>-BCRc MeHg in tuna fish (CRM 463/464)-BCR Other materials in preparation: CP and CrV1 in freeze dried solution and welding dust (BCR) SeIV and SeV' in solution (BCR) As in fish (NIES and BCR) MeHg in sediments (BCR) Organoleads in rainwater and urban dust (BCR) a NRCC-National Research Council, Canada.b NIES-National Institute for Environmental Studies, Japan. BCR-Bureau of Community Reference (Standards, Measurements and Testing Programme). in Table 1.Although this list is not comprehensive, it does reflect the increasing demand for CRMs to validate speciation data. However, the preparation of such materials is not trivial. For instance, many CRMs produced for use in other areas are sterilised by gamma irradiation. However, this is not always appropriate in the case of CRMs certified for metal speciation since in some cases, for example, tin, dealkylation may take place, thereby making valid speciation impossible. The im- portance of matrix matching should also be emphasised. For example, if a sediment is to be speciated for butyltins (mono-, di-and tri-butyltin), the CRM PACS-1 (Harbour Marine Sediment, National Research Council, Canada) is available.Unfortunately, not all sediments have the same chemical composition. A procedure using glacial acetic acid is very 292 Chemical Society Reviews, 1997, volume 26 efficient at extracting the butyltin species from PACS- 1, which has a very high organic composition, but it is considerably less efficient at extracting them from sediments that contain significant proportions of silica or alumina. In such cases, it is impossible to assume that because a procedure works for one reference material, that it will work for all such samples and serious underestimates of the true content can be made. The efficiency of an extraction procedure can be measured by first determining total analyte concentration by conventional means, e.g.acid digestion of a sub-sample. An appropriate extraction procedure for the species of interest may then be used on a second sub-sample. After extraction, the species specific analysis is performed. The concentration of each of the species is then summed and if the sum of the concentrations of the species is the same as the total concentration, the extraction efficiency may be assumed to be 100%. If the extraction efficiency is reproducible e.g. 80 f 5%, the method is under control and may still be valid. If, however, the extraction procedure is not reproducible, i.e. an extraction efficiency of 80 k 20% is obtained, the method is not under control and will not provide accurate results. As mentioned previously, the alter- native method of quality control is the use of spiking/recovery experiments. Again, however, care must be taken in speciation studies since different species may have different extraction efficiencies.If a species is contained within fatty tissue, it may be extracted less efficiently than species that are simply bound to more hydrophilic materials. This means that even if a small volume of an aqueous standard is allowed to soak into the sample for 24 h (the recommended time for spiking experi- ments), the spike may not penetrate the lipid layers. The spike will therefore be more readily extracted and have a better extraction efficiency than the analyte in the sample (assuming methanol or some other polar solvent is used for the extraction). A further complication in many cases is that the standards to perform such spiking experiments are not always commercially available.For example, for arsenic speciation most of the species of interest are now available, although standards for some arseno-sugars are hard to obtain. 3 Analytical approaches to speciation studies In terms of analysis, it is possible to identify species in a specific fraction separated during the sample preparation stage, or as is often desirable, by direct measurement from the sample matrix. Speciation studies on environmental samples are often per- Table 2 Analytical approaches to metal speciation formed using a chromatographic method to separate the species of interest directly interfaced with an element specific de- te~tor.~-~Electroanalytical methods, for example, anodic strip- ping voltammetry (ASV) in the differential pulse stripping mode may also be used but such studies are generally confined to ‘labile’ versus ‘strongly bound’ metals in aqueous samples.’ Such methods have, however, proved particularly popular for use on board ships for selected elements such as Cd, Cu, Pb and Ni.4 A comparison between electroanalytical techniques and those utilising atomic spectroscopy has been made Similarly, ion selective electrodes may be used to detect a specific chemical form although electrodes are not available for organometallic cations.A range of other approaches have also been used including techniques such as neutron activation analysis, radioanalytical methods l and more exotic methods such as photoacoustic spectroscopy and thermal lens spectrom- etry but their use is not widespread.Tables 2 and 3 summarise the more common approaches to speciation studies and provide a number of examples of their use. 4 Chromatography Although gas chromatography (GC) is an extremely common analytical technique for volatile analytes, its use for speciation of inorganic analytes has been relatively limited. This possibly reflects the fact that most GC detectors are not element specific, and that there may be problems associated with transferring the analyte from the end of the GC column to the atom/ion source when atomic spectroscopy is used for detection. The GC eluent is obviously in the gas phase and is usually at an elevated temperature.This temperature must be maintained all the way along the transfer line to the detector since failure to do this leads to cool spots and condensation of the analyte. A heated transfer line is therefore obligatory. For many analytes that are volatile, e.g. organolead species, the transfer line may be kept relatively simple as the temperatures required need not exceed 200-250 “C. For species that have been separated using high temperature GC techniques, e.g. metalloporphyrins, the transfer line may need to be heated to temperatures of approximately 400 OC.12 Ensuring that no cool spots exist in such a transfer line is much more complex. In addition, coupling GC with plasma instruments may lead to the metallic transfer line acting like an aerial and thus coupling to the radio frequency (RF).This is potentially dangerous and may cause instability of the plasma, and instrument failure. The majority of work performed Analytical examples Technique Gas chromatography High performance liquid chromatography Polarograph ylanodiclcathodic stripping voltammetry Ion selective electrodes Auger electron spectroscopy and electron paramagnetic resonance Nuclear magnetic resonance Comments GC has been used with a number of different detectors to The determination of organometallic compounds of tin, effect speciation. Examples include the use of electron arsenic, lead and mercury.23 capture or flame photometric detectors. These tend to be non-specific, and so peaks from interfering matrix constituents may cause confusion.UV-VIS detectors have been used occasionally, but suffer Speciation of aluminium in waters. Speciation of from a lack of sensitivity. Electrochemical and vanadium or chromium using HPLC with fluorescence detectors have also been used. spectrophotometric detection.24 These have been used for differentiating between different Electrochemical behaviour of metallothioneins.*5 oxidation states of species. They have also been used to Speciation of butyltin compounds by CSV.26 Such determine speciation in the presence of humic and fulvic methods have proved popular for work on board acids. ships. Although they have found use in differentiating between Speciation of cadmium or copper in the presence of different oxidation states, they tend to suffer from a lack humic or fulvic acids.27 Determination of fluorides of sensitivity and so are used only for polluted waters.in the presence of aluminium. These are rarely used techniques. Iron speciation at physiological pH in the presence of ascorbate and oxygen.2R This has found several different uses, but the sensitivity is Aluminium complexation with nucleosides. limited. Determination of A13+,hydroxy and other A1 species of geological interest. Actinide speciation using complexation.29 Chemical Society Reviews, 1997, volume 26 293 coupling GC with atomic spectrometry has so far been achieved using flame spectrometry as a detector.5 This approach has been used most commonly for analytes that are present in relatively high concentration e.g.determining organolead species in fuel or roadside dust. A potential limitation of using GC is that often the sample must be derivatised to make it voltaile enough for analysis. This can greatly increase sample preparation time, may cause loss of analyte and uncertainty about the identity of the original species in the sample. High performance liquid chromatography (HPLC) is one of the most common methods used for the separation of non- volatile analytes and has been extensively coupled to atomic spectroscopy for detection. There are a large number of resin supports that may be used as the stationary phase with anion exchange, cation exchange, size exclusion, chelating and reversed phase functionality.The reversed phase resins e.g. octadecylsilane (ODS) are often used in conjunction with an ion pairing reagent, e.g. tetrabutylammonium phosphate, diethyl dithiocarbamate or 8-hydroxyquinoline. In this way, analytes with different oxidation states, e.g. Cr"' and CrVI may be separated. Arsenic has been speciated into AsT", AsV, mono- methyl arsonic acid, dimethyl arsinic acid and arsenobetaine using an anion exchange resin with an ammoniacal potassium sulfate mobile phase. Other approaches have used a reverse phase column with a mobile phase of sodium dodecyl sulfate in 5% methanol and 2.5% acetic acid or cation exchange columns. More recently, the HPLC separation of arsenic species has been coupled with on-line microwave digestion and hydride genera- tion to facilitate the direct determination of both reducible and non-reducible forms of arsenic.13 In order to allow the optimum conditions for the separation of species to be maintained it is important that the detector facilitates as wide a range of solvents as possible. Since the majority of analyses using HPLC are undertaken at room temperature there are no real problems associated with the transfer line coupling the chromatograph to the detector.Thus, couplings with HPLC can often be made cheaply and quickly without major modification to either in~trument.~,~ Table 3 Overview of coupled techniques used for metal speciation Technique Ease of coupling Gas chromatography- A heated transfer line is usually atomic spectrometry required.Holes must be machined in ICP torch boxes to accommodate the transfer line. High performance liquid Easy. The end of the column may chromatograph y-atomic simply be attached to the capillary of spectrometry the nebuliser, although an air bleed may be necessary to overcome differences in flow rate. For ICPs a desolvation device may be necessary if organic solvents are employed. Flow injection-atomic Very simple. FI has been coupled with spectrometry numerous detection systems. Capillary zone Difficult. The low flow rates of CZE are electrophoresis-atomic not readily compatible with atomic spectrometry spectrometric methods. Hydride generation-atomic Usually quite easy.spectrometry The use of capillary electrophoresis is rapidly growing in popularity. At present the flow rate used (10 1.11 min-1) is not readily compatible with atomic spectrometric detection since there are fundamental problems associated with sample trans- port. Conventional nebulisers for example would be starved of liquid and thus not function properly. It is, however, probable that a union will eventually be developed which will facilitate a successful coupling between the two techniques. Hydride generation, although not a chromatographic tech- nique, can be used in conjunction with a cryogenic trap to effect speciation.l4 As"', AsV, monomethylarsonic acid and dimethyl- arsinic acid all form hydrides. If the hydrides formed are flushed from the gas-liquid separator into a liquid nitrogen trap, the hydrides freeze.Removal of the transfer line from the liquid nitrogen trap will allow the hydrides to boil off at their own respective boiling points, and subsequently be swept to the atomic spectrometer for detection. This technique has found a niche for certain applications since it offers a means of improving sensitivity. It is not used very frequently, but does demonstrate that chromatography is not always necessary to obtain separation. 5 Element specific detection Flame atomic absorption spectrometry (FAAS) is one of the most common techniques employed for trace inorganic determi- nations, although it is often not sensitive enough for environ- mental work.Flame AAS is extremely tolerant of organic solvents, indeed the nebuliser efficiency may be increased when compared with water and this in turn may lead to increased sensitivity and better detection limits. Clearly, this is an advantage for some HPLC applications, although care must be taken when using organic solvents to prevent a build up of carbon on the burner. This may be readily removed but failure to do so will adversely affect the flame and result in a decrease in sensitivity. For HPLC, in which mobile phases with high dissolved solids may be employed, the nebuliser and burner system of flame Potential problems Analytical examples Analytes must be volatile or derivatised Determination of organolead into a more volatile form. Analytes species in petrol.5 may condense on any cool spots, Determination of organotin hence losing sensitivity.Transfer compounds in waters and lines may pick up RF. molluscs. Determination of metalloporphyrins in crude oils.12 Some eluents contain high dissolved Determination of organotin solids that may block the nebuliser. compounds in water, molluscs Eluents containing organic solvents and sediments.3O Determiantion may thermally decompose and block of arsenic species. Speciation the burner head with carbon when of antimony, selenium,3' using flame AAS. Organic solvents mercury and lead.32 may extinguish ICPs and also clog cones and the ion lens stack in ICP-MS. Simialr to those described for HPLC. Determination of inorganic selenium and chromium species in waters.33 Determination of Cr species using XRF (a batch method).Difficulty in coupling the two together. A few research papers have been published.34 Cryogenically trapped hydrides of Speciation of reducible arsenic species are slowly ramped to room species.35 temperature, volatilisng individual hydrides at their respective boiling point. 294 Chemical Society Reviews, 1997, volume 26 atomic absorption spectrometers may occasionally block, especially after extended use. It may therefore be necessary to have a wash out period at the end of each chromatographic run. A small air bleed into the nebuliser is frequently used to compensate for the different flow rates of the HPLC (1-2 ml min-l) and the uptake rate of nebulisers used in flame AAS (5-8 ml min-l).This air bleed often improves peak shape and decreases noise. The coupling of GC with atomic spectrometric detectors also needs care; the importance of maintaining the interface temperature has been discussed above. Most reports in the literature where GC has been used as the method of separation in speciation studies have used flame AAS for detection. The sensitivity of the analysis may be increased if the interface leads directly to the atom source rather than going through the conventional nebuliser/spray chamber arrangement. The use of a quartz or ceramic tube placed in the flame to increase the residence time of the atoms in the flame is a convenient way of accomplishing this.l5 Electrothermal atomic absorption spectrometry (ET-AAS) may also be used although the batch nature of this technique precludes its use for routine on-line speciation analysis despite its increased sensitivity when compared with flame AAS. On- line couplings can be made but these require the use of either an electrospray or thermospray device, and even then, running the spectrometer isothermally at the atomization temperature will lead to rapid tube wear. Chromatographic fractions may also be collected off-line and then analysed, but this is laborious and prone to error. Inductively coupled plasma atomic emission spectrometry (ICP-AES) has also been used as the detection method for speciation analysis. Such instruments have better sensitivity and longer dynamic ranges than flame spectrometers.In addition, they also allow alternative spectroscopic lines to be monitored either simultaneously or at least far more rapidly than AAS instrumentation. The use of an appropriate type of nebuliser/ spray chamber configuration rather than simply using the type supplied with the instrument will lead to greatly improved results in terrhs of peak shape and signal to noise properties. However, the effects of organic solvents on the plasma can be substantial. Some instruments operate readily with close to 100% methanol without having an excessively high reflected power. Other instruments have very high reflected powers under the same conditions, thereby risking damage to the (RF) generator.In extreme cases, the plasma may be extinguished by even relatively small proportions of organic solvent (e.g. 30% acetonitrile). It is worthwhile remembering that if a mobile phase containing a high percentage of organic solvent is to be used, it should be introduced in stages to give the matching network time to compensate, as for example when using gradient elution chromatography. This is less likely to lead to plasma extinction. Typical flow rates of HPLC pumps and the uptake rate of an ICP nebuliser are compatible (1-2 ml min-l) and since both usually operate at room temperature, the interface coupling can theoretically be very simple. There are occasions, however, when an on-line solvent removal system incorporating membrane drier tubes, Peltier coolers or a similar device is necessary between the spray chamber and the plasma to prevent problems due to excess solvent.16 GC is less simply interfaced with ICP-AES instruments requiring custom built interfaces to be built.Many of the same considerations are required with induc- tively coupled plasma-mass spectrometry (ICP-MS). These instruments offer detection limits as low as 1 ng dm-3 and have a linear range of five to six orders of magnitude. Therefore, such instruments are particularly attractive for speciation studies although they do have additional drawbacks. For example, the interface between the plasma and the mass spectrometer can be seriously affected by organic solvents. The sample and skimmer cones on the interface region between the atmospheric pressure plasma and the mass spectrometer, can become clogged with carbon which in turn leads to excessive signal drift until the cones become completely clogged, whereupon no signal is obtained.The problem can be overcome in most cases by the addition of oxygen (ca. 3% v/v) into the nebuliser gas flow although excess oxygen will lead to rapid ablation and to substantially reduced cone lifetime. The cones themselves are usually made of nickel, although platinum ones are available (but more expensive) and are not attacked by HPLC/ion chromatography mobile phases containing sulfate or phosphate. The opportunity to utilise isotope dilution in conjunction with speciation studies is also possible using ICP-MS and again offers potential for the future.3 Several other element specific detectors have also been used to good effect and a number of commercial instruments have been produced for specific applications.Mercury, for example, has been speciated using GC coupled with an atomic fluorescence detector producing limits of detection below the ng ml-1 level. Microwave induced plasmas (MIP-AES) have been linked to GC. The helium MIP has an extremely high ionisation energy and may therefore be used to determine analytes such as the halogens, sulfur, nitrogen, oxygen and phosphorus. The use of MIP has been limited to analysing gaseous samples since even a few microlitres of solvent tend to extinguish the plasma, although research continues into the introduction of liquid samples using direct injection nebulisation, 'thermospray' and 'particle-beam' approaches.l7 Finally, new approaches are also being developed such as the use of liquid chromatography-mass spectrometry and low power ICP-MS which when coupled to GC has the potential to determine both atomic and molecular species.'* 6 Selected elements of special interest Although the above text has stressed a number of precautions that must be taken when performing speciation analysis, if appropriate care is taken in terms of sampling, sample preparation and choice of analytical technique, very good results may be obtained on a routine basis.Many elements exist as different species; however, this final section will consider only a few examples where there has been extensive environ- mental interest: arsenic, mercury, tin and lead.For many elements the organometallic forms tend to be substantially more toxic than the inorganic forms. This is true for tin, where inorganic tin is virtually non-toxic, but where the toxicity of the tin species increases as the percentage of organic moiety in the molecule increases. Thus, in the marine environment where organotin compounds have been widely used as the active ingredient in antifouling paints on boats, compounds such as tributyltin (TBT) and triphenyltin (TPhT) are known to have toxic effects on a number of organisms. Eventually these compounds are known to degrade to their progressively less toxic di-, mono- and inorganic derivatives.There are, however, cases where the inorganic forms of the analyte are more toxic than the organometallic forms. Examples include arsenic and selenium, where the inorganic forms (arsenite and arsenate or selenite and selenate) are extremely toxic whereas many of the organic forms, e.g. arsenobetaine or selenomethionine are regarded as being non-toxic. Inorganic arsenic has been cited as a poison in numerous murder cases; however, the organic forms are found at relatively high concentrations in some fish and may be so stable that they do not degrade to the inorganic form even when treated with concentrated acid. Some analytes, e.g. lead and mercury, are poisonous regardless of the chemical form in which they are present; but the degree of toxicity will again depend on the species. Methylmercury for example is substantially more toxic than inorganic mercury.The increased toxicity of the organic form in this case is due to the increased efficiency with which it crosses the blood-gut barrier. The following sections give a brief overview of some of the better known species of each of the selected metals and the methods that have been used to separate and determine them. Chemical Society Reviews, 1997, volume 26 295 7Tin Organotins (mono- and di-organotins) are widely used as stabilizers for rigid PVC. However, it is their use as biocides (triorganotins) that has given rise to environmental concern. Triorganotin biocides (mainly triphenyltin) are used in pesti- cides and tributyltin (TBT) and triphenyltins in antifouling paints.It is this latter usage that has received most attention with regard to the effects of organotin compounds in the environ- ment. In particular TBT has been the subject of many studies. For example, the concentration of TBT has been found to reach 1.5-2 pg 1-1 in seawater, approximately 7 vg1-1 in fresh water, 25-30 mg kg-1 in marine sediments and 3.54.0 mg kg-1 in fresh water sediments. In addition, TBT is also known to enter the flora and fauna of these environments and in many cases lead to severe deformity and in some cases death. Concentra- tions of TBT of 6, 2 and 11 p kg-1 have been reported for bivalves, gastropods and fish, respectively.As TBT is relatively insoluble in water and is lipophilic, it readily adsorbs onto particulates and bio-accumulates in the fat of organisms. Accumulation factors of several thousand have been observed. It is also present in seawater as either the oxide, the hydroxide or the carbonate. It degrades through various mechanisms (e.g. photolysis, hydrolysis and by the action of micro-organisms), where it is slowly de-butylated in a stepwise manner by breaking of the tin-carbon bond. It is known that dibutyltin (DBT) is more readily degradable than TBT, and the DBT then forms monobutyltin (MBT), which is slowly transformed to inorganic tin. Under natural conditions, hydrolysis and photol- ysis are limited and the bio-degradation by the micro-organisms predominates. This bio-degradation is important environmen- tally since the toxicity of the species decreases in the order: TBT Air light light light Me2SnH2 I Me4Sn 1 Sny 1 \ \\ \ I I 7,e.g.TBT microlayer \ \ / Water Influencing factors \\ I/Volatilisation (i) suspended matter (ii) salinity (iii) pH MeSSn+ BunSn3+ Me2Sn2+ bacterial methylation Sn", SnIV degradation Buf12Sn2+ redistribution /MeSn3+ Biota Bun3Sn+ Sediment \/ Biogeochemical factors MeBuSn2+ Fig. 1 Biochemical pathway (simplified) of tin compounds in the environment > DBT > MBT > inorganic tin. Fig. 1 shows some of the biochemical pathways that have been suggested for tin compounds in the environment. A large number of research papers have been published that determine different tin species in a variety of matrices. The environmental aspects of non- biocidal organotin compounds have also been reviewed.19 In terms of analysis, the tin compounds are often extracted from the matrix, derivatised into a more volatile form (usually either by a Grignard type reaction or by the formation of a hydride) and then separated using GC.Detection by flame ionisation detection (FID), flame photometric detection (FPD), AAS, GC-AED and ICP-AES is possible. The detection limits reported have been very impressive, with levels for different tin species at the ng dm-3 level being obtained. Other methods have coupled HPLC with ICP-MS with success. Here, the extraction methods employed will differ depending on the matrix.For example with sediments, glacial acetic acid or a method involving the use of tropolone may be used whilst for mussels enzymatic digestion followed by extraction with organic solvents is preferable. 8 Arsenic As described above, there are a large number of arsenic species present in the environment. These range from the extremely toxic inorganic forms arsenate and arsenite through the mildly toxic species such as dimethyl arsinic acid and monomethyl arsonic acid to the harmless arsenobetaine and arsenocholine. A number of relatively harmless arseno-sugars are also known. Fig. 2 shows the structures of some of the better known arsenic species. Arsenic has found a variety of uses including being an active ingredient of herbicides, fungicides, insecticides, wood preservatives and growth promoting agents for poultry (al- though this use has now ceased).Once in the environment, arsenic and its compounds may undergo physical, chemical and biochemical transformations which may or may not involve a change of oxidation state, and/or mineralization, adsorption and precipitation processes. It has been demonstrated conclusively that arsenic can be methylated in the environment. Mould, for example, has been shown to produce trimethylarsine from arsenite. Bacteria and algae have also been known to methylate arsenic from sediments and waters. It is also known that some fish and higher animals can readily synthesize methylated forms of arsenic.It is assumed that this process occurs in the liver and is a result of de-toxifying the inorganic forms. Arsenobetaine is the end product of arseno-riboside metabolism and is the principal form of arsenic found in many aquatic organisms and sediments. It is worth stating that the total levels of arsenic can be very high for some types of sample although the arsenic content differs markedly between samples. For sediments and soils the level can be as high as 1 g kg-', although a concentration of 7-20 mg kg-' is more normal. In extreme cases, e.g. from a spoil tip close to an arsenic mine, arsenic can be at the percentage level. Natural levels in waters are usually very low (0.1-0.5 pg 1-I), but this again can depend upon a number of other factors.Some vegetation will take in arsenic from soil (and possibly from insecticides). It has been shown that spinach contains approximately 10 mg kg-l arsenic and that radishes can contain levels far in excess of this (up to 100 mg kg-1). However, such total levels reveal little about the potential toxicity of the sample. The main source of arsenic in the human diet, for example, is seafood. Benthos feeders and shellfish can contain extremely high levels of arsenic (mussels and prawns 50-200 mg kg-1) although fortunately this is present mainly as arsenobetaine and is thus non-toxic and readily excreted from the body if consumed by man. Plankton are also known to contain arsenic at concentrations up to 2.4 g kg-1.Fishmeal fed to farm animals has increased the arsenic levels in these species but to a far lesser extent than for the fish themselves. A large amount of literature has been devoted to arsenic speciation.3 Much of this has concentrated on the analysis of seafood samples, although some workers have analysed poultry or urine in diet trials. The vast majority of methods utilise HPLC coupled with a sensitive detector such as ICP-MS or ICP-AES. Other laboratories do partial speciation. For example, hydride generation is useful for determining the reducible forms of arsenic, e.g. the inorganic and methylated forms, but will not give a response for arsenobetaine or arsenocholine. In this way 296 Chemical Society Reviews, 1997, volume 26 a distinction between the reducible 'toxic' forms and the non- toxic non-reducible forms can be made.Extraction methods used have again concentrated very much on the use of organic solvents such as methanol, although enzymes are increasingly being used for biological samples. As always, the use of acids for sample destruction is precluded because although species such as arsenobetaine are stable, the methylated forms will become de-methylated. 9 Lead Lead is not only ubiquitous in the environment but is also a well-known poison. Although used in numerous industrial applications the concentration of lead in the environment has increased substantially over the last fifty years largely because of its use in antiknock additives in petroleum.The alkylated lead compounds used for this purpose are extremely toxic and therefore legislation has been introduced in many countries to reduce the amount of leaded petrol. The toxicity of alkyllead species increases in the order Pb2+ < RzPb2+ < R3Pb+ < R4Pb where R is a methyl- or ethyl-group. In the environment the ionic forms of lead have been found to be the most persistent. In algae and higher plants alkyllead compounds have been shown to inhibit growth and cause disturbances of mitosis and ultrastructural alterations. Fish have also been shown to be affected by organoleads and fatalities in man through chronic exposure are also known. Between 0.1-2.0% of the organolead (principally tetramethyllead and tetraethyllead) in petrol pass- ing through the engine is not combusted and thus ends up in the environment. Natural bio-methylation of lead also occurs, but only very slowly and it must be noted that, like tin, lead in the environment can also become de-alkylated. This is again due to factors such as photolysis.In the absence of light and at reduced temperature the decomposition of tetraethyllead is very slow (2% over 77 days) but under atmospheric conditions it is likely to be decomposed rapidly (half-life of 2-8 h). However, due to the non-polar nature, high vapour pressure and lipophilic H~AsO~ H~AsO~ Arsenous acid Arsenic acid (As"') (As") 0 .OH I1 CH~ASZO (CH3)2A7 'OH OH Momomethylarsonic acid Dimethylarsinic acid (MMAA) (DMAA) CH~ASH~ (CH3)2AsHMeth ylarsine Dimethylarsine 0 I/CH,-OCRI /?CH-OCRI //oCH~-OPOCH~CH~AS+(CH~)~ '0-character of tetralkylleads they are likely to volatilize from water, be absorbed into organisms (e.g.fish tissue) or directly absorbed onto sediments. Many different procedures have been used for the analysis of organoleads in a variety of different samples. Gas chromat- ographic procedures have often been used because of the volatility of many lead species and this technique has been linked to numerous types of detector including atomic absorp- tion and atomic emission spectrometers. More detailed reviews of lead speciation including the occurrence, chemical trans- formations, sampling, storage and pretreatment for water, air and solids together with an overview of the analytical methods that have been used are available.20.21 10 Mercury Mercury is an unusual metal with a number of unique properties and a wide range of industrial applications.There are three main species of mercury: inorganic (Hg"), methylmercury (MeHg) and dimethylmercury (Me2Hg). All forms of mercury are considered poisonous, although methylmercury is of particular concern since it readily undergoes biomagnification in food chains. Indeed, one of the best known cases of poisoning involving metal speciation was that in Minamata, Japan where methylmercury was accidentally released into the sea and was consequently taken up by fish. The fish were later ingested by local people including some women who were pregnant, and resulted in severe abnormalities in the newly born children.In a second case, this time in Iraq, seed grain was sprayed with methylmercury-containing fungicides and again resulted in large-scale poisoning since the seed was eaten and not planted as intended. In many cases of mercury poisoning micro- organisms are responsible for the natural formation of the methylated species, although there are also micro-organisms that de-methylate. The methylmercury balance is therefore a result of two competing reactions. The monomethyl species is soluble in water whereas the dimethyl species is not. However, the pH of the sample can also have an effect on the formation of ASH^ Arsine Trimethylarsine oxide (TMAO) (CH313ASTrimethylarsine Arsenolecithin Carboxymethyl(trimethy1)arsonium zwittenon 2-Hydroxyethyl(trimethyl)arsonium salt (arsenobetaine) (arsenocholine) 0cAS-CH2kH3&/o>ocH2yH2R where X and R can be OH OH OH SO3H OH OS03H NH2 S03H Dimethyl(ribosy1)arsine oxides OH OP03CH2(0H)CH20H Fig.2 Structures of arsenic species Chemical Society Reviews, 1997, volume 26 297 the species since at low pH the formation of methylmercury is favoured in preference to the dimethylmercury. The organic forms of mercury are far more easily absorbed than the inorganic form so the methylated species constitute 60-90% of the total mercury in fish. However, the interconversion between the different forms of mercury is complex and it has now been suggested that various mercury cycles may operate in the environment.For example, in sediments the biomethylation of mercury is obviously important whereas in fish methylmercury may result from an atmospheric depositional flux of me-thylmercury.3 Due to the long established importance of mercury specia- tion, the element has received considerable attention. A review by Puk and Weber22 has described many of the analytical approaches applied to mercury speciation. However, it is worth remembering that many analysts stabilise samples to be analysed for mercury with dichromate. Although this is extremely effective at preventing loss of total mercury, it will substantially affect the speciation, since the dichromate con- verts the volatile species to inorganic mercury.Various hyphenated techniques or derivatization methods may be employed successfully for the determination of mercury species although the use of CRMs (certified reference materials) is to be highly recommended. 11 Conclusions Metal speciation is now well established in many areas of chemistry. Although a lot has been achieved in terms of increasing our knowledge of the behaviour of some elements in the environment, much still needs to be done. Whole areas of environmental chemistry can now be readdressed to provide more insight into the mobility of metals in the environment and how these pathways link with the bioavailability of the metals, bioaccumulation and possible toxicity. Certainly there is parallel interest in metal speciation in the area of clinical chemistry. At the forefront of such developments is the analytical chemist who must find new ways of extracting the metal species from complex matrices and further develop analytical tech- niques to perform both qualitative and quantitative analysis at lower and lower levels.Thus, the whole area of metal speciation continues to provide an exciting and challenging arena for research in the future. 12 References Research trends in the field of environmental analysis, ed. P. Quevauviller and E. A. Maier, European Commission, Brussels, 1994. Chemical Speciation in the Environment, ed. A. M. Ure and C. M. Davidson, Blackie Academic and Professional, 1995. P. Quevauviller, E. A. Maier and B.Griepink, Quality Assurance for Environmental Analysis, Elsevier, 1995. 4 Element speciation in bioinorganic chemistry, ed. S. Caroli, Wiley, 1996. 5 L. Ebdon, S. J. Hill and R.Ward, Analyst, 1986, 111, 1113. 6 L. Ebdon, S. J. Hill and R. Ward, Analyst, 1987, 112, 1. 7 S. J. Hill, M. J. Bloxham and P. J. Worsfold, J.Anal. Atomic Spectrom., 1993, 8, 499. 8 R.M. Harrison and S. Rapsomanikis, Environmental Analysis using Chromatography Interfaced with Atomic Spectroscopy, Ellis Harwood, 1989. 9 C. M. G. van den Berg, Anal. Chim. Acta, 1991, 250,265. 10 P. M. Bersier, J. Howell and C. Bruntlett, Analyst, 1994, 119, 219. I1 Determination of trace elements, ed. Z.B. Alfassi, VCH, Weinheim, 1994. 12 A. Kim, S. J. Hill, L. Ebdon and S.Rowland, J. High. Rex Chromatogr., 1992, 15, 665. 13 K. Lamble and S. J. Hill, Anal. Chim. Acta, 1996,334, 261. 14 J. Dedina and D. L. Tsaleu, Hydride generation atomic absorption spectrometry, Wiley, 1995. 15 L. Ebdon, R. W. Ward and D. A. Leathard, Analyst, 1982,107, 129. 16 W. Cairns, L. Ebdon and S. J. Hill, J. Fresenius, Anal. Chem., 1996,355, 202. 17 G. K. Webster and J. W. Karnahan, Element specific. Chromatographic detection by atomic emission spectroscopy, ed. P. C. Uden, ACS Symp. Ser. No. 479, 1992, 218. 18 G. O’Connor, L. Ebdon, E. H. Evans, H. Ding, L. K. Olsen and J. A. Caruso, J. Anal. Atomic Spectrom., 1996, 11, 1151. 19 D. C. Baxter and W. Frech, Pure Appl. Chem., 1995,67, 615. 20 R. J. Maguire, Water Poll. Res. J. Can., 1991, 26, 243. 21 Hazardous Metals in the Environment, ed. M. Stoeppler, Elsevier, 1992. 22 R. Puk and J. H. Weber, Appl. Organornet. Chem., 1994,8,293. 23 I. L. Marr, C. White, D. Ristau, J. L. Wardell and J. Lomax, Appl. Organomet. Chem., 1997, 11,11. 24 S. J. J. Tsai and S. J.Hsu, Analyst, 1994, 119, 403. 25 J. Schwarz, G. Henze and F. G. Thomas, J. Fresenius, Anal. Chem., 1995,352,474. 26 A. Munoz and A. R. Rodriguez, Electroanalysis, 1995,7, 674. 27 H. M. V. M. Soares and M. T. S. D. Vasconelos, Anal. Chim. Acta, 1994, 293, 261. 28 C. Dorey, C. Cooper, D. P. E. Dickson, J. F. Gibson, R. J. Simpson and T. J. Peters, Brit. J. Nutrition, 1993, 70, 157. 29 D. L. Clark, S. D. Conradson, S. A. Ekberg, N. J. Hess, D. R. Janecky, M. P. Neu, P. D. Palmer and C. D. Tait, New J. Chem., 1996, 20, 211. 30 C. Rivas, L. Ebdon, E. H. Evans and S. J. Hill, Appl. Organomet. Chem., 1996, 10, 61. 3 1 L. Pitts, A. Fisher, P. Worsfold and S. J. Hill, J. Anal. Atomic Spectrom., 1995, 10,519. 32 A. A. Brown, L. Ebdon and S. J. Hill, Anal. Chim. Acta, 1994, 286, 391. 33 A. M. Naghmush, K. Pyrznska and M. Trojanowicz, Anal. Chim. Acta, 1994,288, 247. 34 M. J. Tomlinson, L. Lin and J. A. Caruso, Analyst, 1995,120, 583. 35 A. G. Howard and C. Salou, Anal. Chim. Acta, 1996,333, 89. Received, 17th October I996 Accepted, 25th March 1997 298 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600291

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Developing the physical organic chemistry of Fischer carbene complexes Claude F. Bernasconi Department of Chemistry and Biochemistry, University of California, Santa Cruz, California 95064, USA Fischer carbene complexes have over the last 25 years become a major focus of many synthetic groups. On the other hand, until recently there have been only a few kinetic and/or thermodynamic studies of even the simplest reactions of these complexes. A number of such studies have now been carried out in my laboratory. This review summarizes the results obtained thus far and focuses on three areas: (1)thermodynamic and kinetic acidities of carbene complexes that contain an ionizable proton on the carbon adjacent to the carbene carbon; (2) mechanism and structure-reactivity relationships in the reactions of carbene complexes with nucleophiles; (3) mechanism of hydrolysis of carbene com- plexes with and without ionizable protons.1 Introduction Fischer carbene complexes are compounds of the general structure 1-M.1 M is a transition metal, L are ligands while X /X ,om3 ,om3 L.,,M=C, (COkM=C\ (CO)SM=C\ Y CH3 Ph 1-M 2-Cr(M = Cr)2-W (M= W) 3-Cr (M= 0)3-w (M = W) and Y represent a variety of groups including alkyl, aryl, vinyl, alkynyl, amino, alkoxy, alkylthio, halo and others; typically one of these groups is a n-donor such as alkoxy, amino or alkylthio which provides stabilization of the electron deficient carbene carbon. Since the first targeted synthesis of (methoxymethylcar- bene)pentacarbonyltungsten(O), 2-W,by Fischer and Maasbo1,z Claude F.Bernasconi was born in Zurich, Switzerland. He received his undergraduate and PhD degrees (with Heinrich Zollinger) from the Swiss Federal Institute of Technology (ETH). Following a postdoctoral year with Manfred Eigen at the Max Planck Institute for Biophysical Chemistry in Gottin- gen, he joined the chemistry faculty at the University of California, Santa Cruz, in 1967, where he has been a Professor of Chemistry since 1977. His current research in- terests are in physical organic chemistry with particular atten- tion to proton transfers, nucleo- philic addition to electrophilic alkenes, nucleophilic vinylic substitution, and reactions of Fischer carbene complexes. He is the author of over 150 re-search publications and of the book 'Relaxation Kinetics.' the chemistry of Fischer carbene complexes has developed in an explosive fashion and become an important branch of organo-metallic chemistry.The major thrust of this development has been the synthesis and characterization of hundreds of such complexes and their use as synthons and catalysts3 This intensive synthetic activity is in sharp contrast to the modest efforts toward kinetic and thermodynamic studies of even the simplest reactions of prototypical Fischer carbene complexes such as 2-Cr, 2-W, 3-Cr and 3-W. For example, the substitution of the methoxy group in complexes such as 2-Cr or 3-Cr by amines,1,4 thiolate ions*,5 carbanions1 and other nucleophilesl is well documented, yet only very few kinetic investigations of such processes have been reported, e.g.the reaction of amines with 2-Cr6 and with (dithiomethylcar- bene)pentacarbonyltungsten(O).7 Examples of other processes that have been the subject of kinetic studies include the thermolysis of 2-(oxacyclopentylidene)pentacarbonylchro-mium(0),8 CO-exchange reactions,g substitution of CO ligands by alkynes,l* the addition of amines to the triple bond of an a$- acetylenic pentacarbonylchromium carbene complex and the reaction of several carbene complexes with tertiary phosphines and phosphites.12 A few years ago we started a research program aimed at studying the kinetics of reactions of 2-M, 3-M (throughout this review we use the symbols 2-M, 3-M, etc.when more than one derivative is meant, e.g. the Cr or W derivative) and related carbene complexes with nucleophiles and bases in polar solvents, mainly aqueous acetonitrile. Our objective is to firmly establish the mechanisms of these reactions and to develop a better understanding of their structure-reactivity behaviour. This review summarizes our results obtained thus far and focuses on three areas: (i) thermodynamic and kinetic acidities of complexes such as 2-M that contain an acidic proton on the a-carbon (carbon adjacent to the carbene carbon); (ii) mecha-nism and structure-reactivity relationships in the reactions of complexes such as 3-M with nucleophiles; (iii) mechanism of hydrolysis of carbene complexes with and without an acidic proton on the a-carbon.2 Kinetic and thermodynamic acidities 2.1 General features and methods The first indication that carbene complexes of the type 2-M are relatively strong carbon acids was Kreiter's 13 observation of the rapid conversion of 2-Cr to (CO)~CI=C(OCH~)CD~ in dilute NaOCH3-CH30D solutions. Casey and Anderson14 subse- quently showed that in tetrahydrofuran (THF) the acidity of 2-Cr is approximately the same as that of p-cyanophenol. In aqueous solution acidity measurements are more difficult because of rapid hydrolytic decomposition of the conjugate anion. This is probably the reason why an acidity constant in water was not reported until 1989 when a kinetic study of the deprotonation of 2-Cr yielded a pEH of 12.3 (the superscript CH will be used to distinguish pE" from the pK, of other acids).15 This value was obtained from the relationshipgH= (kyH/k?fo)K, where kyH and k!fo) are the rate constants for the proton transfer reaction [eqn.(l)] and K, is the ionic product of the solvent; kyH and k??O were determined as Chemical Society Reviews, 1997, volume 26 299 the slope and intercept, respectively, of a plot of the observed pseudo-first-order rate constant for equilibrium approach vs. [OH-] [eqn. (2)]. The kobs values were obtained in a stopped- flow spectrophotometer. Applying the same methodology, kobs = kyHIOH-] + k??O (2) kyH, k?:O and pEH were determined for a number of carbene complexes, mainly in acetonitrile-water (50 :50);1618 the results are summarized in Table 1.In cases where the pEH was substantially lower than 12, i.e. for 8-Cr and 8-W (for numbering of compounds, see Table l),the intercepts of the plot of kobs vs. [OH-] were too small for a reliable determination of k??O. In these cases the peH values were determined based on the rate of deprotonation of the carbene complex by butylamine and reprotonation of 2-Cr-by BuNH$.16,I7A different problem prevented application of eqn. (2) for the determination of k??O for 6-Cr and 7-Cr. In these cases the rate constants for hydrolysis of the respective anions (g2O in Scheme 3, see below) are of the same order of magnitude as k??O, which, at low [OH-], leads to strong coupling between proton transfer and hydrolysis.l9 As a result, eqn. (2) breaks down and kH?O had to be determined by a method which is based on measuring hydrolysis rates under two extreme conditions. One was at high [OH-] where kyHIOH-] >> k"_Oand kyHIOH-] >> q2Oso that kF20 could be obtained directly. The other was at low enough [OH-] so that the anions (e.g. 2-Cr-in Scheme 3) are steady state intermediates and the observed rate constant for hydrolysis is given by kyH @20[OH-]/(kH_To + @*O); k??O was then obtained as the only unknown in this steady state expression. Rate constants for proton transfer to various amines were also measured. Because of the relatively high pg" values of the carbene complexes the deprotonation is thermodynamically unfavourable in most cases and hence the rates had to be determined by generating the anion in KOH solution and then reacting it with the protonated amine.In order to prevent hydrolysis, the anion was generated in a double-mixing stopped-flow spectrophotometer and reacted with the proto- nated amine buffer within 50-100 ms after its formation. Some representative results are summarized in Table 2; ky refers to the deprotonation of the carbene complex by the amine (B) and kB_" to the protonation of the anion by the respective ammonium ion (BH+), as illustrated for 2-Cr in eqn. (3). k? 2-Cr + B 2-Cr-+ BH+ (3)k!?? p3 k FH -,OCH3 (CO)50=C, + OH--(co)scr-C\ +H20 (1) HOCH3 k-l' CH2 2-Cr 2-Cr-Table 1 Summary of pKFH values and rate constants for proton transfer to hydroxide ion in MeCN-water (50 :50) at 25 "C Carbene Complex (co)5cr a (4-Cr)a PGH 14.47 pGH( corr) 14.77 kpHf 74 kYH(carry 37 kHzOR 14 log k:H(corr) 1.36 (5-Cr)b 12.98 13.46 24 1 80.3 1.48 1.03 (2-Mo)' 12.81 13.29 181 60.3 0.75 0.82 (2-Cr)c 12.W 12.98 456 152 0.91 1.07 (2-W)b 12.36 12.84 284 94.7 0.42 0.80 (6-Cry 12.32 12.62 46.8 23.4 0.064 0.09 (7-Cr)e 12.27 12.27 2.97 2.97 3.57 x 10-3 -0.99 OCH3 (CO)5Cr=( CH2Ph (8-Cr)c .10.40 10.70 115 57 1.90 X -0.48 (8-W)b 10.18 10.48 140 70 1.35 X 10-3 -0.36 a Ref. 18. b Ref. 17. L Ref. 16. d pEH in water is 12.3, ref. 16. Ref. 19./In units of dm3 mol-I s-I. R In units of s-l. The log k:H (corr) values have an estimated uncertainty of k0.25 log units. 300 Chemical Society Reviews, 1997, volume 26 Table 2 Summary of representative rate constants for proton transfer to amines in MeCN-water (50 : 50) at 25 "Ca B pKFH ky 4-Crh(peH = 14.47) Piperidine 11.01 47 1.36 X lo5 ca.3.40 2-Crc (pGH = 12.50) Piperidine 11.01 906 2.77 x 104 1Piperazine 9.97 246 8.23 x 104 3.52 BuNH2 10.40 122 1.52 X 104 1 MeOCH2CH2NH2 9.39 22.6 2.88 X 104 2.85 H~NCOCHZNH~ 8.14 5.01 1.14 x 104 1 6-Crd(pcH = 12.32) Piperidine 11.01 48.2 9.82 X 102 ca. 2.19 7-Crd (pEH = 12.27) Piperidine 11.01 0.194 3.55 ca. -0.1 1 8-Crc (pe" = 10.40) Piperidine 11.01 225 55 1 Piperazine 9.97 74.8 200 1Morpholine 8.70 19.5 978 1.70 BuNH2 10.40 75.5 MeOCH2CH2NH2 9.39 16.1 16575S 1.371H2NCOCH2NH2 8.14 2.30 413 NCCH2NH2 5.29 0.126 1.61 X 104 ~~~~~~~~~~ (1 In units of dm3 mol-1. h Ref.18. ' Ref. 16. d Ref. 19. In most cases the range of amines amenable to study was limited by the fact that protonation of the anion by BH+ was too fast for the stopped-flow technique. This problem was generally more severe for carbene complexes with the highest pe" values, especially 4-Cr (pEH = 14.47) which only allowed measurements with piperidinium ion. 18 This contrasts with 8-Cr (pe" = 10.40) for which k!!? with buffers as weakly basic as aminoacetonitrile could still be measured.16.17 As discussed in Section 2.4, factors other than a low peH contribute to the low kB_): values for 8-Cr and some other complexes, e.g.6-Cr and 7-Cr. 2.2 Effect of structure on pK:H values The pEH values of the carbene complexes listed in Table 1 range from 10.18 for 8-W to 14.47 for 4-Cr. This compares with a pEH of 25.6 for ethyl acetate in aqueous solution,20 indicating that the electron withdrawing effect of the (CO)5M- moieties is much stronger than that of the ester carbonyl oxygen. This may be attributed to charge dispersion into the CO ligands, a point to which we will return below. Regarding the variation of the peH values with structure the following features are noteworthy. (i) The peH values are not very sensitive to the metal. This is seen from a comparison of 2-Mo, 2-Cr and 2-W, and of 8-Cr with 8-W. This result implies that the stabilization of the anion by the (CO)5M moiety is similar for the three metals; it is in marked contrast to the strong metal dependence of the pK, values of metal hydrido complexes of the type (q5-CSHS)M(C0)3H.In acetonitrile these pKa values are 13.3 for M = Cr, 13.9for M = Mo and 16.1 for M = W; in methanol they are 6.4, 7.2 and 9.0, respectively.21 This trend for the hydrido complexes would be difficult to reconcile with the results for the carbene complexes if it were assumed that the pK, values of the hydrido complexes refleg a strong increase in the stabilization of the anions, (T~-CSH~)M(CO)~, along the series W < Mo < Cr. However, Norton has shown that the differences in the acidities of the hydrido complexes are not due to differences in the stabilization of the anions, but can be traced to differences in the M-H bond dissociation enthalpies,21" i.e.there is no contradiction between Norton's and our results. (ii) The effect of changing the alkoxy group can be seen by comparing 4-Cr, 5-Cr and 2-Cr. The change from methoxy (2-Cr) to ethoxy (5-Cr) leads to an increase in pEH of 0.48 units. Inasmuch as the resonance structure a appears to play a dominant role in the stabilization of carbene complexes,l the higher peH for 5-Cr is most plausibly attributed to increased stabilization of a by stronger electron donation when R = ethyl instead of methyl.17 In the case of 4-Cr, the stabilization of the carbene complex by n-donation from the oxygen (b) is even more effective because, by virtue of the cyclic structure of 4-Cr, the oxygen is locked into a position for better n-overlap with the carbene carbon.'* 53Cr NMR data are in agreement with this notion.22 The result is a substantial further increase in pG".(iii) Substituting one of the a-hydrogens by a phenyl group increases the acidity by more than two pK,-units (8-Cr vs.2-Cr and 8-W vs. 2-W). This acidifying effect is mainly the result of additional resonance stabilization of the anion provided by the phenyl group. (iv) Substituting one or two a-hydrogens by methyl groups (6-Cr and 7-Cr vs. 2-Cr) has an acidifying influence which is particularly apparent when considering the statistically correc- ted peH, pEH(corr), which are included in Table 1.They are pKa(corr) = 12.98, 12.62 and 12.27 for the CH3, CH2CH3 and CH(CH3)2 derivatives, respectively.This acidity enhancement may be attributed to an increase in stability of the respective anions (7-Cr-> 6-Cr-> 2-Cr-) caused by the methyl groups. Assuming that the dominant resonance structure of the anions has the negative charge mainly delocalized into the (CO)SCr moiety [see, e.g. 2-Cr-in eqn. (l)], the increased stability in the order 7-Cr-> 6-Cr-> 2-Cr-simply reflects the well-known stabilization of alkenes by methyl groups.2' This stabilization is commonly interpreted in terms of hyper- conjugation23 although other factors may contribute to it. 19,23 The increase in acidity of nitroalkanes in the order CH3N02 (peH = 10.22) < CH3CH2N02 (PE" = 8.60) < (CH&CHNO2 (pe" = 7.74) has been explained in similar terms.24 2.3 Kinetic acidities: comparison with other carbon acids Proton transfer rate constants for the reactions with OH- are summarized in Table 1, those for the reactions with amines in Table 2; the tables include statistically corrected (for the number of protons on the carbene complex) rate constants [(kpH(corr), k,B(corr)].Intrinsic rate constants, k,(corr), based on the statistically corrected eH and k: values are also reported; they refer to deprotonation of the carbon acid by a hypothetical base such that the statistically corrected equilibrium constant is equal to 1. For the reactions with OH-, kzH(corr) was estimated based on the relationship log k:"(corr) = log kyH(corr) -0.5 log eH(corr) [~"(corr) = E"(corr)/K, with K, = 6.46 X mo12 dm-6 being the ionic product of the solvent16] which corresponds to the simplest version of the Marcus25 equation.The log kzH(corr) values are included in Table 1; because they are estimates, there is probably an uncertainty of about ca. k0.25 log units in these values. For the reactions with amines, log k:(corr) was obtained by extrapolation or inter- polation of Bronsted plots to pK?" -pEH(corr)+ log(p/q) = 0 where peH is the pKa of the protonated amine and p and q are the statistical factors for B and BH+, respectively. The log k:(corr) values are reported in Table 2. Because they are independent of the thermodynamic driving force of the proton transfer the intrinsic rate constants provide a useful measure of the purely kinetic barrier (intrinsic reactivity) of the reaction.This intrinsic reactivity can be related to salient structural differences between the transition state and carbon acid and/or its conjugate anion. Indeed, there exists an inverse relationship between log k, for proton transfer from carbon Chemical Society Reviews, 1997, volume 26 301 acids activated by n-acceptors and the degree of resonance stabilization by charge delocalization in the carbanion26 Table 3 summarizes log k, values for a number of representative examples; for a more complete list see ref. 26(b). The reason for this inverse relationship is that the transition state is imbalanced in the sense that resonance development in the incipient carbanion lags behind proton transfer.26 This means that at the transition state the negative charge is closer to the a-carbon than to the n-acceptor [e.g.the (C0)SM moiety], as illustrated in exaggerated form by structure c (Y = -1 for B = OH-, Y = 0 for B = amine). As a result, the stabilization of the transition state is disproportionately weak compared to that of the carbanion which depresses the intrinsic rate constant, the more so the greater the resonance stabilization of the anion. This effect is an example of a much more general phenomenon known as the principle of nonperfect synchronization or PNS;26 this principle states that whenever in a reaction the development of a product-stabilizing factor (in our case resonance) lags behind the main bond changes (in our case proton transfer), there is a reduction in the intrinsic rate constant.The log eH(corr) and kE(corr) values for the deprotonation of the carbene complexes fall into the range between -0.99 and 1.36 for log eH(corr) (Table 1) and between -0.1 1 and 3.52 for log kE(corr) (Table 2). These ranges are below the log eHand log kE values of the deprotonation of most carbon acids except the nitroalkanes (see Table 3). The implication is that the conjugate anions of the carbene complexes derive a substantial fraction of their stabilization from resonance, probably mainly by virtue of charge dispersion into the CO ligands of the (C0)SM moieties.This interpretation is consistent with IR data of anions such as 2-Cr-or 4-Cr-in THF (where the anions are stable) which show a substantial reduction in the CO stretching frequency upon deprotonation of 2-Cr or 4-Cr, respec-tively. 1427 2.4 Kinetic acidities: dependence on carbene complex structure We first consider deprotonation of the carbene complexes by OH-. Except for the trend towards higher rate constants with increasing thermodynamic acidities in the series 4-Cr < 5-Cr < 2-0, the kyH values do not correlate well with the peH values. For example, the most acidic compounds (8-Cr and S-W) have kyH(corr) values that are less than twofold higher than for the least acidic carbene complex (4-Cr), despite a difference in pGH(corr) of more than four units; or kpH(corr) for 2-Cr is 2.7-fold higher than for 8-Cr yet its pEH(corr) is 2.3 units higher than for 8-Cr; or koH(corr) for 7-Cr is 7.9-fold lower than for 6-Cr yet the p&(corr) of 7-Cr is 0.35 units lower, etc.The poor correlation between kYH(corr) and pEH(corr) is the result of significant variations in the intrinsic rate constants. Table 3 pK$H and log kg values for representative carbon acidsa ~~~~~~ ~ ~~ Carbon acid Solvent PEH HCN H20 9.0 CHz(CNI29-Cyanofluorene H20 50% DMSO 11.2 9.53 Meldrum's acid 50% DMSO 4.70 CH3COOEt H20 25.6 4-NOzC6hCHzCN 50% DMSO 12.62 1,3-1ndandione 50% DMSO 6.35 Acetylacetone 50% DMSO 9.12 2,4-(NO2)2C6H3CH2CN CH3N02 50% DMSO 50% DMSO 8.06 11.32 PhCH2N02 50% DMSO 7.93 ~~~~~~~~ ~ ~ ~ ~ There is only one group of carbene complexes for which the log g"(corr) values are all approximately the same.They are 5-Cr (1.03), 2-Mo (0.82), 2-Cr (1.06) and 2-W (0.80); considering that the log @H(corr) values are estimates, the slight variation in these numbers is too small to warrant any attempt at interpretation. Even 4-Cr (1.36) probably belongs to this group as will be shown below when comparing intrinsic rate constants for the reactions with piperidine. We conclude that changes in metal or alkoxy group have no or, at best, only a small effect on the log eH(corr) values. These are the cases where kinetic acidity correlates reasonably well with thermodynamic acid- ity. In contrast, substituting one or two a-protons by a phenyl or by methyl groups results in a substantial lowering of log eH(corr).For 8-Cr and 8-W this reduction can be mainly attributed to the resonance effect of the phenyl group which stabilizes the anions 8-Cr-and 8-W-;8 this resonance effect is in addition to the resonance effect of the (C0)5M moiety and further reduces the already relatively low kzHvalues of typical carbene complexes such as 2-Cr or 2-W. The eH-reducing effect of methyl groups is the result of a combination of several factors.lg The first is that, due to the imbalanced nature of the transition state, the stabilizing effect of the methyl groups that manifests itself in the anions (6-Cr-and 7-Cr-) and is responsible for the increased thermodynamic acidities of 6-Cr and 7-Cr is only minimally developed at the transition state.By virtue of the PNS,26 this factor alone would lead to reduced eHvalues. However, there are additional factors that actually destabilize the transition state and further decrease eH.One is the repulsive interaction between the partial negative charge on the a-carbon (see c) and the inductive/field effect of the alkyl group. Another is the repulsive field effect between the partial negative charge on the hydroxide ion (Bv = OH- in c) and the alkyl groups. The third factor is steric crowding which is particularly significant in the reaction of 7-Cr. The latter three factors lead to a decrease in kYH(corr) in the series 2-Cr > 6-Cr > 7-Cr; since the thermodynamic acidities increase in the order 2-Cr < 6-Cr < 7-Cr the result is a negative Bronsted a value which is reminiscent of similarly negative a values for the deprotonation of CH3N02, CH3CH2N02 and (CH&CHN02 (nitroalkane anomaly).24 Turning to the proton transfer reactions involving amines, we note that kE(corr) shows the same qualitative dependence on carbene complex structure as eH(corr).In particular, k?(corr) is essentially independent of the metal and the alkoxy group, while substituting a-hydrogens by phenyl and methyl groups de- presses k:(corr). These reductions in kE(corr) are somewhat larger than the corresponding reductions in kEH(corr). This is the result of greater steric crowding in the transition state because of the larger size of the amines compared to OH-; the effect is seen to be particularly important with 743, the bulkiest carbene complex.19 ~it$~(corr) log kFHh log kF(corr)L ~ ~~~~~~~~~~~~~~ log k:(corr)d 9.0 ca.8.60 11.5 ca. 6.95 9.53 ca. 2.69 4.58 3.76 5.00 3.75 26.08 > 2% 12.92 2.62 3.62 2.88 6.65 0.91 2.95 2.27 9.42 1.25 2.58 1.90 8.36 1.49 2.60 1.70 11.80 0.04 0.53 8.23 -0.91 -0.41 a Taken from ref. 26; note that in ref. 26 the pEH and log k: values were not corrected for the number of acidic protons on the carbon acid. The log kzH values have an estimated uncertainty of k0.25 log units. B = Secondary alicyclic amines. d B = Primary aliphatic amines.e Estimated based on data in ref. 20. 302 Chemical Society Reviews, 1997, volume 26 3 Reactions of nucleophiles at the carbene carbon 3.1 Methoxide ion as nucleophile Fischer carbene complexes show a strong qualitative similarity to carboxylic esters with respect to their reactions with nucleophiles, i.e. they undergo nucleophilic substitution by an addition-elimination mechanism involving a tetrahedral inter- mediate (T -). However, there is a major quantitative difference in that the tetrahedral intermediates derived from carbene complexes are much more stable than T-(ester). pcH3 -PcH3 (c0)5M--F-ph Nu "TPhNu T-T-(ester) In favourable cases this allows a direct detection of the tetrahedral adduct, e.g.in the reaction of 3-Cr and 3-W with MeO- in methanol, eqn. (4).*8The rate constants kl and k-l for 'Ph 3-M eqn. (4) are summarized in Table 4 along with the equilibrium constants K1 = kl/kWl.For the corresponding reaction of methyl benzoate a K1value of 1 X 10-7 to 5 X 10-7 dm3 mol-* has been estimated.28 This value implies that K1 for methoxide ion addition to the carbene complexes is 2 X 108 to 109-fold higher than for addition to the corresponding ester; it demon- strates that the (C0)5M moiety is a much stronger electron acceptor than the carbonyl oxygen in the ester, presumably, at least in part, because of a resonance effect brought about by delocalizing the negative charge into the CO ligands of T6Me. This is the same effect that is largely responsible for the much higher acidity of 2-Cr (pK, = 12.5) compared to ethyl benzoate (26.5) discussed earlier. Table 4 Rate and equilibrium constants for methoxide ion addition to 3-Cr and 3-W in methanol at 25 "0 3-Cr 3-W kl/dm3 mol-1 s-I 77.1 186 k-l/s-' 1.10 I .68 Kl/dm3 mol-1 70.1 111 1% ko 0.96 1.25 a Ref. 28.Further evidence for the importance of resonance in the stabilization of ToMe comes from the intrinsic rate constants, k,, for nucleophilic addition. They have been estimated in a similar way as eH(corr) for the deprotonation of acidic carbene complexes by OH-, i.e.log k, = log kl -0.5 log K,;for both 3-Cr and 3-W log k, ca. 1 (Table 4). This compares with log k, ca. 3.1 estimated for the reaction of methyl benzoate with methoxide ion.28 The substantially lower log k, for the carbene complexes is most easily understood as the result of the resonance contribution to the stability of TcMe.As in the case of resonance stabilized carbanions generated by deprotonation of carbon acids activated by n-acceptors, the delocalization of the charge in T,-,, presumably lags behind bond formation at the transition state and leads to a reduction of k, (PNS effect).26 Regarding the dependence of reactivity on the metal, it is noteworthy that the rate and equilibrium constants are quite similar for both complexes (Table 4). This is consistent with the very small metal dependence of pcH and proton transfer rate constants when comparing 2-W with 2-Cr or 8-W with 8-Cr (Tables 1 and 2).3.2 Amines as nucleophiles Aminolysis of Fischer carbene complexes with NH3, primary amines and unhindered secondary amines is a facile reaction that leads to the corresponding amino carbene complex, e.g. eqn. (5).1,4,6 A kinetic study of this reaction with 3-Cr and primary amines in decane, dioxane, methanol and dioxane- methanol mixtures was reported by Werner et a1.6 However, in these solvents complications arise due to the low polarity which makes mechanistic interpretations difficult. More definite results were obtained from a kinetic investigation of the same reaction in acetonitrile-water (20 :80 v/v).29 Even though the presumed tetrahedral intermediate does not accumulate to detectable levels, the kinetic data provided compelling evidence that the reaction involves such an intermediate and proceeds through the mechanism shown in Scheme 1.The evidence for the involvement of T$, was based on 3-Cr fTA kYlOH-1 9-Cr Scheme 1 the observation of base catalysis by the amine and by OH-, and specifically on how the second-order rate constant, kA [see eqn. (5)], depends on the amine and hydroxide ion concentration. The dependence on OH-is shown in Fig. 1 for the reaction with 3000 0 aa. a 0 3000 2000 4) 0 0.03 0.06 0.09 I I 1 1 I I butylamine; it indicates a change from rate-limiting base catalysed conversion of the intermediate (T3 to products at low base concentrations to rate-limiting nucleophilic attack on 3-Cr at high base concentrations.This is consistent with eqn. (6) for kA based on the steady state approximation. Note that k2 for the spontaneous conversion of TZ to 9-Cr has been omitted from eqn. (6) because this term was negligible under the experimental conditions. Analysis of the data according to eqn. (6) Chemical Society Reviews, 1997, volume 26 303 provided the kl ,k$/k-1, kyH/k-and kyH/k$ values summarized in Table 5. kl(kf[RNH2]+ ky[OH-]) kA = ~k-~(kt[RNH~l+ky[OH-]) (6) Table 5 Kinetic parameters of the reaction of 3-0 with primary amines in MeCN-water (20 :80) at 25 OCu RNH2 pKtH klb @/k-Ic kyH/k-L ky"/@ BuNH~ 10.67 2900 33.7 8.1 X lo3 2.40 X 102 CH30CH2CH2NH2 9.52 400 21.1 8.4 X 104 3.98 X 103 ClCH2CH2NH2 8.61 91 19.5 2.8 X 105 1.44 X 104 H2NCOCHZNH2 8.03 100 17.3 1.25 X 106 7.22 X 104 EtOCOCH2NH2 7.70 36 9.7 1.49 X 106 1.54 X lo5 0 Ref.29. b In units of dm3 mol-l s-1. L In units of dm3 mol-1. Regarding the mechanism of the base catalysed conversion of TA to products, there are two reasonable possibilities. The first involves rate-limiting deprotonation of Ti followed by rapid methoxide ion expulsion, eqn. (7). For this mechanism, one can &RN-H2I om3* (CO)sM-(!!-Ph fast_ 9-Cr + CH30H(CH30-) (7)IRNH k3OH-l ".-'A equate kt and kyH with the respective rate constants for deprotonation of TA by the amine (@J and hydroxide ion (ky.). In the second mechanism proton transfer occurs as a rapid pre- equilibrium step followed by general acid catalysed leaving group departure, eqn.(8); here kA corresponds to ktHHK$/KtHand kyH to e2OKz/Kw, with K$d being the acidity constant of RNH; and K, the ionic product of the solvent. kHZo 3 In the aminolysis of carboxylic esters, the deprotonation of the corresponding zwitterionic intermediate is usually rate limiting.30 However, for the reaction of 3-Cr with primary amines conversion of TA to products occurs by the mechanism of eqn. (8).29 This conclusion was based on the following analysis of kyH/k$ values (Table 5).If proton transfer were rate limiting, the kyH/k$ ratios should be equal to kyF/k$,. The rate constant ep"refers to a diffusion controlled proton transfer and should have a value of ca. 1Olo dm3 mol-1 s-1, independent of the amine.31 k3"p should also be independent of the amine since the pK, difference between Ti and RNH; should be roughly constant; it was estimated to be in the order of 4 X lo8 to 2 X 109 dm3 mol-1 ~-l.3~Hence the kyH/k$ ratio should be independent of amine and have values of the order of 5-25.This contrasts with the experimental ratios that increase with decreasing pKfH from 240 to 1.54 X 105. This increase which corresponds to a Bronsted coefficient, d(kyH/kt)/dpK$H, of -0.92 is easily accounted for by the mechanism of e n. (8) according to which the kyH/k$ ratios correspond to 8KfH/kPHKw.The Bronsted coefficient of -0.92 indicates that the kyH/k$ ratios are not quite proportional to KfH because of a small compensating effect by the J$z0/ktH ratios which decrease slightly with decreasing amine basicity, reflecting the increasing catalytic effect of RNHZ on ktH with decreasing pKtH.Because the rates of interconversion between TZ and Tx by proton transfer should be about the same in the aminolysis with 3-Cr and that of the corresponding methyl ester, the change from rate-limiting proton transfer for the ester to rate-limiting methoxide ion departure for 3-Cr must be the result of lower kAHand J$zo values with the carbene complex. These reduced k$ and gz0 values reflect the combined effects of the increased stability of TA derived from 3-Cr and the lower intrinsic rate constant for leaving group departure, both a consequence of the strong delocalization of the negative charge into the (CO)5Cr moiety of TA.3.3 Hydroxide ion and water as nucleophiles Until recently, the hydrolysis of carbene complexes has received very little attention. The first study was that reported in 1993 by Aumann et al.33 who investigated the reaction of several substrates of the type 10-Cr (R = Ph, CH=CHPh, ,OEt (CO)s(3r=C, + H20 + C&12N4+ RCH=O + (CO)~C~C~H~ZN~+EtOH (9) R 10-Cr C4H3S, CH=CHC4H3S and C-CPh) in THF containing small quantities of water. In the presence of urotropine (hexamethy- lenetetramine, C6H12N4) the aldehyde RCH=O is formed in 290% yield with all R groups except when R is CECPh; in this latter case the triple bond undergoes nucleophilic attack by the amine. We recently published a kinetic study of the hydrolysis of 3-Cr, 3-W, 11-Cr, 11-W and 12-Cr in acetonitrile-water /OEt /om3(CO)sM= C, (CO),Cr =c, Ph CH=CHPh 11-Cr (M = 0) 12-Cr11-W (M = W) (50 : 50).34 Two processes were observed.The first, faster one, corresponds to conversion of the carbene complexes to (hydroxyphenylcarbene)pentacarbonylchromium(O) or the tungsten analogue, while the second process leads to the formation of benzaldehyde (from 3-M and 11-M) or cinnamal- dehyde (from 12-Cr), consistent with the findings of Aumann et al.33 The rate law for the first process examined over a wide pH range is given by eqn. (10) with kB[B] representing buffer base catalysis. kobs = kH20 4-kOHIOH-1 i-~B[BI (10) The data were interpreted in terms of Scheme 2 (shown for the specific case of 3-M) which is an elaboration of the mechanism proposed by Aumann et al.33 and reminiscent of the mechanism of carboxylic ester hydrolysis.Scheme 2 takes into account all potential pathways under basic as well as acidic conditions. Specifically, Poand kPHIOH -1 represent nucleo- philic addition of water and OH -, respectively, while kElaH+ and kFjo refer to H+-catalysed and spontaneous loss of OH- from TGH, respectively. Regarding product formation, the scheme allows for the possibility of either direct conversion of TGH to products by unimolecular (@2O), H+-catalysed (ky aH+) or intramolecularly acid catalysed (k;) expulsion of CH30-, or of reaction via the dianionic form of the intermediate (T@, again with either spontaneous (q2O) or H+-catalysed (ky)loss of CH30-.Note, that for simplicity the buffer catalysed pathways have been omitted from the scheme. The question whether TGH accumulates to detectable levels at high pH is an interesting one. It was estimated that the equilibrium constant for OH- addition to 3-M in acetonitrile- water (50 : 50) is at least as large as that for MeO- addition to 3-M in methanol (K1 ca. 100 dm3 mol-l, Table 4), suggesting that TGH might be detectable. However, TGH was not observable 304 Chemical Society Reviews, 1997, volume 26 \ Ph T& 13-M- Scheme 2 even at the highest OH- concentrations (0.1 M) used. This implies that conversion of TGH to products is much faster than its formation, i.e.kY2O + k; + cky >> kyHIOH-], turning TGH into a steady state intermediate whose formation (kyH) is rate limiting. Further analysis indicated that it is k; and/or el$ rather than k?*O that greatly exceed kyHIOH-]. In acidic solution water attack on the carbene complex (qZo)is rate limiting. The kyH and kyZo values are summarized in Table 6. The following points are noteworthy. (i) The rate constants depend little on the metal as seen by comparing 3-Cr with 3-W or 11-Cr with 11-W.This is consistent with observations made for the reaction of 3-Cr and 3-W with MeO- in methanol.2s Table 6 Summary of kY2O and kYH values for nucleophilic attack on various carbene complexes in MeCN-water (50 :50) at 25 "Co Carbene complex kYIO/s-l kyH/dm3mol-I s-I (CO)SCpC(OMe)Ph(3-Cr) 2.9 X 26.6 (CO),Ct=C(OEt)Ph (11-Cr) 4.5 x 10-4 10.5 (CO),Cr=C(OMe)CH=CHPh (1243) 1.8 x 10-4 14.6 (CO)5W=C(OMe)Ph(3-W) 2.8 x 10-3 26.3 (CO),W=C(OEt)Ph (11-Cr) 4.7 x 10-4 17.6 From Ref.34. (ii) Substitution of a methoxy for an ethoxy group (11-Cr vs, 3-Cr and 11-W vs. 3-W) lowers the reactivity. This is the result of more effective reactant state stabilization by n-donation by the ethoxy group, the same effect which decreases the acidity and rate of deprotonation of 5-Cr com ared to 2-Cr. This effect manifests itself more strongly in the kfZ0values than in the kyH values and suggests that the transition state is relatively late for water addition and relatively early for OH- addition, consistent with the Hammond postulate35 or reactivity-selectivity princi- ple.36 (iii) The reactivity of OH- towards 12-Cr is marginally lower than that towards 3-Cr but attack by water on 12-Cr is 16 fold slower than on 3-Cr.The lower reactivity of 12-Cr may be attributed to extra resonance stabilization of 12-Cr by the styryl group.34 The larger effect on kY2O than on kyH may again be the result of a later transition state for the water reaction. 4 Hydrolysis of ionizable carbene complexes 4.1 Thermal reaction The hydrolysis of 2-Cr,37 4-Cr,38 5-Cr,37 8-Cr39 and 8-W39 (for structures see Table 1) has been thoroughly investigated in acetonitrile-water (50 :50).The organic products of the hydrol- ysis were: acetaldehyde and methanol for 243, acetaldehyde and ethanol for 5-Cr, (3-methoxystyrene for 8-Cr and 8-W, and 2-hydroxytetrahydrofuran in equilibrium with small amounts of 4-hydroxybutanal for 4-Cr.Fig. 2 shows a typical pH-rate profile. The points below pH 8.5 represent a 'water reaction', 0 8 0-1 I 0aoCT)g-2t 0 -0 -3 0 0 I I I 1 I I 0 2 4 6 8 10 12 14 PH Fig. 2 The pH-rate profile of the hydrolysis of 2-Cr in MeCN-water (50: 50) at 25 "C those above pH 8.5 an OH-catalysed reaction while the levelling off at pH > 12 is due to the ionization of the carbene complex (see Section 2). The hydrolysis is subject to general base catalysis; the points on the pH-rate profile between pH 4.8 and 11.3 were obtained by extrapolation of buffer plots to zero buffer concentration.The pH dependence of the hydrolysis of all compounds studied is, in principle, consistent with the mechanism of Scheme 2 for the hydrolysis of 3-M, and so are the products of the reactions of 243, 4-Cr and 5-Cr. However, the products obtained in the hydrolysis of 8-Cr and 8-W and the fact that in basic solution the hydrolysis of all the compounds is subject to a substantial kinetic solvent isotope effect are inconsistent with Scheme 2, at least at pH > 8.5. The mechanism that accounts best for all experimental observations at pH > 8.5, including the isotope effect, is shown in Scheme 3 for the example of 2-Cr. It involves fast deprotonation of 2-Cr followed by rate limiting reaction of 2-Cr-with water to form 15-0; complexation between (CO)SCr and the vinyl ether activates the latter towards basic hydrolysis which rapidly leads to the vinyl alcohol and tautomerization to the aldehyde.Control experi- ments demonstrated that the kind of complexation indicated by 15-Cr indeed promotes rapid hydrolysis of the vinyl ether.37 In the reaction of 8-Cr and 8-W complexation of the vinyl ether ((3-methoxystyrene) appears to be weak, presumably because of steric crowding, and hence the reaction essentially stops at the (3-methoxystyrene stage although small amounts of PhCH,CH=O could be detected with 8-W. Regarding the conversion of 2-Cr-into 15-Cr, there are two possible pathways, stepwise and concerted. The stepwise pathway consists of rate-limiting protonation on the metal Chemical Society Reviews, 1997, volume 26 305 15-Cr (CO)sCflH-+ CH3CH =O + CH3OH Scheme 3 followed by rapid reductive elimination while the concerted pathway involves protonation on the carbene carbon which is simultaneous with bond cleavage between the metal and the carbene carbon.No firm distinction between these two pathways is possible but the concerted alternative is preferred because the stepwise mechanism requires the PKa of 14-Cr to be unrealistically high.37 The fact that the base catalysed hydrolysis of ionizable carbene complexes proceeds by the mechanism of Scheme 3 instead of the nucleophilic mechanism of Scheme 2 implies that the former is energetically more favourable.For the nucleo- philic mechanism to be competitive with Scheme 3 in the case of 2-Cr the rate constant for OH- attack on 2-Cr would have to be approximately three times higher than for OH- attack on 3-Cr.34 This contrasts with the expectation that it is 3-Cr which should be the more electrophilic carbene complex, due to the electron withdrawing inductive effect of the phenyl group. The potential m-donor effect of the phenyl group is negligible because this group was found to have an orthogonal orientation both in the solid state40 and a sol~tion.~l Regarding the ‘water reaction’ (pH < 8.5),whether it follows Scheme 3 with conversion of 2-Cr- to 15-M involving H30+ instead of water, or Scheme 2 with the q2O-step being rate limiting, remains an open question.38 4.2 Light-induced reaction The hydrolysis of 8-Cr and 8-W is catalysed by light.39 Accelerations in the presence of high light intensities of the order of five- to six-fold have been observed.A possible mechanism for this photochemical process is shown in Scheme 4.39 This mechanism is consistent with the fact that PhCH=CHOCH3 is the main organic product of the reaction, just as for the thermal reaction, and with a kinetic solvent isotope effect that is similar to that for the thermal process and suggests that reaction of 8-M-with water is rate limiting. The results did not, however, allow a distinction between the two pathways that lead to 8’-M-. Note that the increased reactivity of 8’-M-compared to 8-M-is consistent with a higher electron density induced by replacing one of the strongly electron withdrawing CO ligands by acetonitrile.The hydrolysis of 7-Cr19 is also catalysed by light although the effect is smaller than for 8-Cr and 8-W and the phenomenon was not studied in detail. On the other hand, no light-induced rate accelerations have been observed for the hydrolysis of 2-Cr,37 4-Cr,38 5-Cr37 or 6-Cr.19 It is unclear at this point what structural factors are responsible for inducing catalysis by light. 306 Chemical Society Reviews, 1997, volume 26 5 Conclusion Until recently there has been a dearth of kinetic and thermody- namic data on fundamental reactions involving Fischer carbene complexes. Over the last few years, we have begun to fill this void by studying the thermodynamic and kinetic acidities of ionizable carbene complexes, and the kinetics of nucleophilic addition and substitution at the carbene carbon of phenyl carbene complexes.In these processes Fischer carbene com- plexes show a behaviour that is qualitatively similar to that of carboxylic esters but differs dramatically in its quantitative manifestations because of the much stronger electron with- drawing effect of the (C0)5M moiety compared to that of a carbonyl oxygen. For example, the thermodynamic acidities of the ionizable carbene complexes are 11-15 pKa units higher than that of ethyl acetate, and the equilibrium constants for MeO- addition to 3-M are ca. 109-fold higher than for MeO- addition to methyl benzoate.On the other hand, the rate constants for proton transfer and nucleophilic addition do not show as dramatic an increase; this is because the strong electron-withdrawing effect of the (C0)SM moieties is mainly the result of m-delocalization which lowers the intrinsic rate constants of these processes, thereby attenuating the effect of the stronger thermodynamic driving force on the actual rates. Regarding the hydrolysis of ionizable Fischer carbene complexes, our studies have shown that a reaction pathway FH3 FH3(CO),M=C + OH-Kl (CO),MS + H20 \CH$’h -%IPh8-M 8-M-8’-M 8’-M- through the ionized carbene complex is more favourable than the classic nucleophilic mechanism governing hydrolysis of the non-ionizable complexes.For this reaction there is no counter- part in the hydrolysis of esters. In some cases the pathway through the ionized carbene complex is further enhanced by light. 6 Acknowledgments I am deeply grateful to my numerous students and postdocs who did all the work. The continuous financial support by the National Science Foundation and the Petroleum Research Fund administered by the American Chemical Society is also greatly appreciated. 7 References 1 K. H Dotz, H. Fischer, P. Hofmann, F. R. Kreissl, U. Schubert and K Weiss, Transition Metal Carbene Complexes, Verlag Chemie, Deerfield Beach, Flonda, 1983. 2 E. 0.Fischer and A. Maasbol, Angew Chem , Int Ed Engl , 1964, 3, 580 3 (a) K H. Dotz, Angew Chem, Int Ed Engl, 1984, 23, 587; (b) W D.Wulff, in Advances in Metal-Organic Chemistry, ed. L S. Liebeskind, JAI Press, Greenwich, CT, 1985,vol. 1; (c)W D Wulf, in Comprehensive Organic Synthesis, ed. B. M. Trost, Pergamon, Oxford, 1991,vol. 5, p. 1065;(6)H. G. Schmaltz, Angew Chem ,Int Ed Engl , 1994, 33, 303, (e)L. S. Hegedus, Transition Metals in the Synthesis of Complex Organic Molecules, University Science Books, Mill Valley, CA, 1994, ch. 6. 4 U Klabunde and E. 0.Fischer, J Am Chem Soc , 1967,89,7141. 5 R. Aumann and J. Schroder, Chem Ber , 1990,123, 2053 6 H. Werner, E. 0. Fischer, B. Heck1 and C. G. Kreiter, J Organomet Chem , 1971,28,367. 7 A L Steinmetz, S. A. Hershberger and R. J. Angelici, Organometallics, 1984,3,461 8 C P Casey and R L.Anderson, J Chem SOC, Chem Commun , 1975, 895 9 C P. Casey and M. C. Cesa, Organometallics, 1982, 1,87 10 H. Fischer, J. Muhlemeier, P. Mark1 and K. H Dotz, Chem Ber , 1982, 115, 1355. 11 R. Ripoh and R. van Eldik, Organometallics, 1993, 12, 2668 12 H. S. Choi and D. A. Sweigart, J Organomet Chem, 1982, 228, 249. 13 C. G. Kreiter, Angew Chem ,Znt Ed Engl, 1968, 7, 390. 14 C P. Casey and R. L. Anderson, J Am Chem SOC, 1974,96, 1230. 15 J. R. Gandler and C. F. Bernasconi, Organometallics, 1989, 8, 2282. 16 C F Bemasconi and W. Sun, J Am Chem SOC, 1993, 115, 12526. 17 C. F. Bemasconi and W. Sun, Organometallics, 1997, 16, 1926 18 C. F. Bernasconi and A. E. Leyes, J Am Chem SOC,1997, 119, 5169. 19 C. F. Bemasconi, L.Garcia-Rio, W. Sun, K. Yan, and K. W. Kittredge, J Am Chem Soc, 1997,119,5583 20 T. L. Amyes and J. P. Richard, J Am Chem Soc , 1996, 118, 3129. 21 (a)R. F. Jordan and J. R. Norton, J Am Chem SOC, 1982,104, 1255; (b)S. S. Knstjansd6ttir and J. R. Norton, in Transition Metal Hydrides, ed. A. Dedieu, Verlag Chemie, New York, 1992, p. 309. 22 A. Hafner, L. S. Hegedus, G. de Weck and K. H. Dotz, J Am Chem Soc, 1988,110, 8413. 23 (a) J. McMuny, Organic Chemistry, Brooks/Cole, Pacific Grove, California, 3rd edn., 1992, p. 190; (b) K. P. C Vollhardt and N E Schore, Organic Chemistry, Freeman, New York, 2nd edn , 1994, p. 394 24 A J. Kresge, Can J Chem , 1974, 52, 1897. 25 R A. Marcus, J Chem Phys , 1965,43, 679 26 (a)C. F Bernasconi, Acc Chem Res , 1987,20,301, (b)C. F Bernas- coni, Adv Phys Org Chem , 1992, 2, 119. 27 C P. Casey and W. R. Brunswold, J Organomet Chem , 1976, 118, 309 28 C F Bernasconi, F. X. Flores, J. R. Gandler and A. E Leyes, Organometallics, 1994, 13, 2186 29 C. F. Bemasconi and M. W Stronach, J Am Chem Soc , 1993, 115, 1341. 30 (a)A. C. Satterthwait and W. P. Jencks, J Am Chem Soc , 1974, 96, 7018, (b)M. J. Gresser and W. P. Jencks, J Am Chem SOC, 1977,99, 6963, (c)M M. Cox and W. P Jencks, J Am Chem Soc , 1981,103, 580. 31 M. Eigen, Angew Chem Int Ed Engl , 1964,3, 1. 32 M -L Ahrens and G. Maass, Angew Chem , Int Ed Engl , 1968, 7, 818. 33 R Aumann, P. Hinterding, C Kruger and R Goddard, J Organomet Chem , 1993,459, 145 34 C F Bemasconi, F. X. Flores and K W Kittredge, J Am Chem Soc , 1997, 119,2103 35 G S. Hammond, J Am Chem Soc, 1955,77, 334. 36 A. Pross, Adv Phys Org Chem , 1977, 14,69 37 C. F. Bernasconi, F. X. Flores and W. Sun, J Am Chem Soc , 1995, 117,4875. 38 C. F. Bernasconi and A. E Leyes, J Chem SOC,Perkin Trans 2, 1997, 1641 39 C. F. Bernasconi and W. Sun, Organometallics, 1995, 14, 5615. 40 0.S. Mills and A. D. Redhouse, J Chem Soc (A), 1968,642. 41 S. R. Amin, K. N. Jayaprakash, M. Nandi, K. M. Sathe and A Sarkar, Organometallics, 1996, 15, 3528. Received, 21st January I997 Accepted, 16th April 1997 Chemical Society Reviews, 1997, volume 26 307

ISSN:0306-0012    

DOI:10.1039/CS9972600299

出版商:RSC    

年代:1997

数据来源: RSC