摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Dr. A. R. Butler Professor E. C. Constable Dr. T. C. Gallagher Professor D. M. P. Mingos FRS Professor J. F. Stoddart FRS Consulting Editors Dr. G. G. Balint-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor B. T. Golding Professor M. Green Professor A. Hamnett Dr. T. M. Herrington Professor R. Hillman Professor R. Keese Dr. T. H. Lilley Dr. H. Maskill Professor A. de Meijere Professor J. N. Miller Professor S. M. Roberts Professor B.H. Robinson Professor M. R. Smyth Dr. A. J. Stace Staff Editor Mr. K. J. Wilkinson University of Sussex University of Leicester University of St.Andrews University of Basel, Switzerland University of Bristol Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich, Switzerland University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Bath University of Newcastle upon Tyne University of Reading University of Leicester University of Bern, Switzerland University of Sheffield University of Newcastle upon Tyne University of Gottingen, Germany Loughborough University of Technology University of Exeter University of East Anglia Dublin City University, Republic of Ireland University of Sussex Royal Society of Chemistry, Cambridge It is intended that Chemical Society Reviews will have the broad appeal necessary for researchers to benefit from an awareness of advances in areas outside their own specialities.Deliberate efforts will be made to solicit authors and articles from Europe which present a truly international outlook on the major advances in a wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca. 30 references), but should act as a springboard to further reading.In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context. Review articles must be short, around 8-1 0 journal pages in extent. In consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Information to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis (including a selection of the literature references that will be cited in the review and a brief academic CV of the author), rather than the completed article, should be submitted to the Senior Editor (Reviews), Books and Reviews Department, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 4WF. @ The Royal Society of Chemistry, 1994 All Rights Reserved ’ No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic or mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Typeset by Servis Filmsetting Ltd. Printed in Great Britain by Blackbear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99423FX009

出版商:RSC    

年代:1994

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99423BX011

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

ISSN 0306-001 2 CSRVBR 23(3) 147-226 (1994) Chemical Society Reviews Volume 23 Issue 3 Pages 147-226 June 1994 w n I Polyradicals: Synthesis, Spectroscopy, and Catalysis By J.A. Craysron, A. Iraqi, and J. C. Walton (pp. 147-1 53) Polyradicals are polymers with saturated or conjugated backbones containing a succession of paramagnetic centres. Synthetic strategies and methods of characterization are described with examples of each. Unconjugated polyradicals, e.g.poly(TEMPOacrylate), are useful catalysts for the oxidation of amines and alcohols. Conjugated polyradicals include polypyrroles and polythiophenes functionalized with nitroxide, quinone, and viologen units. Polyradical-coated electrodes catalyse redox reactions of organic substrates. Spin cooperation may be induced by certain structural features.Developments in the search for organic ferromagnets are surveyed. Possibilities for superconductivity in polyradicals are briefly treated. Chemistry in Near-critical Fluids By Roberto FernBndez-Prini and M. Laura Japas (pp. 155-1 63) With the increasing popularity of supercritical fluids as media for physical and chemical processes, much effort has been devoted to understand the peculiar behaviour shown by near-critical fluids. Their observed macroscopic behaviour shifts with density from that typical of a liquid to that typical of a gaseous state, passing through a marginally stable region. Microscopic analysis is mainly focused in the interplay of a long- range effect (enhanced solvent susceptibility) and short-range intermolecular contributions as the fluid density varies.1 HOD I Electrophoretic NMR By Manfred Holz (pp. 165-1 74) NMR experiments in the presence of the sample of an electric direct current in (DCNMR) are now feasible. Thus in complex electrolyte solutions or in fluid macromolecular mixtures distinct charge-carrying species are observable and the sign of their charge and their electrophoretic mobilities can be measured. The author, one of the pioneers in this area, describes the principles and methods of electrophoretic NMR (ENMR), discusses practical experimental problems and their solutions, and gives an overview of actual measurements, applications, and possible future developments of this new class of NMR experiments. The Hydrides of Aluminium, Gallium, Indium, and Thallium: A Re-evaluation By Anthony J.Downs and Colin R.Pulham (pp. 175-1 84) Unlike boron, the Group 13 metals aluminium, gallium, indium, and thallium form few well authenticated hydrides, the status of which has suffered over the years from either neglect or controversy. Recent experimental advances (leading, for example, to the synthesis and characterization of gallane) have now stimulated, and in turn profited from, relatively sophisticated quantum mechanical analyses. We draw on a combination of experiment and theory to re-assess these compounds, to review their properties, as well as some of the issues remaining unresolved, and to signal ways in which such studies may be turned to account. Trimetallic Units as Building Blocks in Cluster Chemistry By D.Imhof and L. M. Venanzi (pp. 185-1 93) Metal centres with dlO-electron configuration readily form trimetallic units, which can add one or more metal atoms or ions. Thus, preformed Pt,-units {Pt,), react with metal cations, and even metal atoms, forming tetrametallic, pentametallic, or heptametallic clusters. The {Pt,)-units are generally of the type [Pt3(p2-L),L',], e.g. [Pt3(p2-CO),(PR3)J. Trimetallic units containing the coinage metals {M,}, + can be assembled by reacting transition-metal hydrides of the type [M'H,L'',,] (M' = Ru, Os, Rh, Ir) with the corresponding metal cations, e.g. [RhH,(triphos)] (triphos = CH3C(CH2PPh,),) and one equivalent of Ag(CF,SO,) give [Ag3(p-H)9Rh3(triphos),l(CF3S03)3.Some of the homo- (Pt,) and heterometallic (Pt,Cu and Pt,Ag) clusters, supported on, e.g.,alumina, have been used as catalyst precursors for the dehydrogenation of methylcyclohexane to toluene.Towards a Laboratory Strategy for the Study of Heterogeneous Catalysis in Stratospheric Ozone Depletion By Martin R. S. McCoustra and Andrew B. Horn (pp. 195-204) Recent observations suggest that stratospheric ozone depletion over the poles in wintertime may be linked to heterogeneous processes occurring upon low temperature particle surfaces, in addition to homogeneous gas- phase reactions. In situ measurements are difficult, necessitating the development of laboratory methods for the accurate determination of heterogeneous reaction parameters for a wide variety of atmospheric constituent gases with stratospheric particle surfaces. In this review, some techniques currently being applied to the simulation of representative surfaces and techniques for probing chemistry thereon will be discussed.Affinity Biosensors By Dbnal Leech (pp. 205-21 3) Recent advances in the development and application of affinity biosensors are presented in this review. Current assay technology for the detection of ligand binding to antibodies, receptors, DNA, and other binding proteins and selected approaches to the develpment of reversible, non-destructive affinity biosensors are discussed. The problems to be overcome for the commercialization of practical affinity biosensors are examined and future trends in affinity biosensor research are predicted. Homo- and Hetero-metallic Alkoxides of Group 1,2, and 12 Metals By R.C. Mehrotra, A. Singh, and S. Sogani (pp. 215-225) The preparation, properties, and recent structural elucidations of homo- and hetero-metallic alkoxides as well as 0x0-alkoxides of Group 1, 2, and 12 metals are critically reviewed. The applicability of new soluble alkoxy derivatives in the preparation of ultrahomogeneous materials by the sol-gel process is discussed. The potential future areas of developments are exemplified by uniquely stable heterotrimetallic coordination systems, opening up possibilities for the molecular design of single source precursors. Articles that will appear in forthcoming issues include Syntheses, Structures, and Properties of Methanofullerenes F. Diederich, L.Isaacrs, and D. Philp Crystal Engineering of Diamondoid Networks M. J. Zaworotko Solution Chemistry of Lanthanide Macrocyclic Complexes F. Arnaud-Neu Microelectrodes: New Dimensions in Electrochemistry R. J. Forster Propagation of Interfacial Waves in Microgravity F. Quirion, M.4. Asselin, and G. G. Ross Linear Free Energy Relationships and Pairwise Interactions in Supramolecular Chemistry H.J. Schneider Some Aspects of the Metal-Insulator Transition J. K. Burdett Oxidation of Some Organic Compounds by Aqueous Bromine Solutions J. Palou Biological Activity, Reactivity, and Use of Chromotropic Acid and its Derivatives J. Duda Non-ideality in Isotopic Mixtures G. Jan&, L. P. N. Rebelo, and W. A. Van Hook Protein Structure from Linear Dichroism Spectroscopy and Transient Electric Birefringence Michael Bloemandal Chemical Society Reviews (ISSN 03060012) is published bi-monthly by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge, CB4 4WF, England.All orders accompanied with payment should be sent directly to The Royal Society of Chemistry,Turpin Distribution Services Ltd., Blackhorse Road, Letchworth, Herts., SG6 IHN, U.K. NB Turpin Distribution Services Ltd., distributors, is wholly owned by The Royal Society of Chemistry. 1994 annual subscription rate E.C. €99.00, U.S.A. $186.00, Canada €1 1l.OO+ GST, Rest of World f 106.00. Customers should make payments by cheque in sterling payable on a U.K. clearing bank or in U.S. dollars payable on a U.S. clearing bank. Second class postage is paid at Jamaica, N.Y. 11431. Air freight and mailing in the USA. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. All other despatches outside the U.K. by Bulk Airmail within Europe and Accelerated Surface Post outside Europe. PRINTED IN THE U.K. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at f30.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.

ISSN:0306-0012    

DOI:10.1039/CS99423FP015

出版商:RSC    

年代:1994

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99423BP017

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Polyradicals: Synthesis, Spectroscopy, and Catalysis Joe A. Crayston, Ahmed Iraqi, and John C. Walton Department of Chemistry, University of St. Andrews, St. Andrews, Fife, KYl6 9ST, Scotland 1 Provenance of Polyradicals Polymers have always been useful as chemically robust stuctural components and such applications are expected to continue for most bulk commodity polymers. Alongside these engineering materials, designed for their mechanical properties, specialized polymers have been developed in the past few decades which contain periodic functionality suiting them for specific physico- chemical tasks. Examples include electrode coatings, catalysts, photo-resists, and light-emitting polymers. Polyradicals com- bine together the high molecular weights and structural regulari- ties of common solid-state polymers, with the unique character- istics associated with the unpaired electrons present in free radicals.In the production of virtually all polyradicals advan- tage is taken of the ease with which chains and networks can be built up by the stepwise repetition of sequences of reactions like addition, coupling, condensation etc. from the standard meth- ods of polymer synthesis. The product polyradicals contain a succession of paramagnetic centres, usually stable free radicals or radical ions. These radical centres occur as repeating side chain units, or as repeating electronic structures within the polymer backbone, and they are ordered by the recurrent polymer structure. Two major classes of polyradicals can be distinguished.First, those with unconjugated backbones to which stable radicals are attached as pendant groups, i.e. (1). The second class contains materials with conjugated backbones of the polyacetylene type, linked either to free-radical side-groups, i.e. (2), or incorporat- ing repeating paramagnetic centres within the backbone. Technically, lightly doped poly(acetylenes), poly(thiophenes), polypyrroles) etc. belong to this second class, because in poly- pyrrole, for example, the doping process (oxidation) introduces Joe Crayston is a Lecturer in Inorganic Chemistry at the University of St. Andrews, Scotland's oldest university. After gaining hisfirst degree in chemistry from Cambridge in 1981 he moved to Nottingham to carry out Ph.D.work on the matrix isolation of 0x0 and oxide species under the supervision of J. J. Turner, M. Poliakofl, and G. Davidson. Duringpostdoctoral work with M. S. Wrighton (MIT) he developed interests in electrochemistry and inorganic materials. Since his appointment at St. Andrews in 1986 he has focused his research on modijied electrodes, including conducting polymers, inorganic redox polymers, and electrochromic thin films prepared by sol-gel processing. In 1990 he was awarded the Harrison Memorial Prize 147 $qqkR' R' R' R' R' R' polarons (radical cations delocalized over ca. four monomer units) which are the major charge-carriers. However, conduct- ing polymers have received excellent coverage elsewhere and, with a few exceptions, are excluded from the present review.Polyradicals contain ordered arrays of unpaired electrons and their most important chemical properties stem from this. Because the incorporated free radicals usually have accessible oxidation or reduction potentials, polyradicals can act as cata- lysts, or as full reactants, in a variety of electron transfer processes. Depending on the nature of the electron array, ferromagnetic, ferrimagnetic, or antiferromagnetic behaviour may be displayed. Polyradicals, in contrast with conducting polymers, may have appreciable conductivity before doping. It is also possible that, in conjugated polyradicals containing two different kinds of non-bonding electrons, enhanced conductivity could be produced and that pairwise mobility of the electrons could even give rise to superconductivity. In this review we shall discuss the synthesis, characterization, and spectroscopy of polyradicals.We shall also survey the chemical reactivity, including the electrochemistry and catalysis of, for example, single electron transfer reactions. Physical properties will not be discussed in detail, except where it is relevant to the design of polyradical structures for the goals of ferromagnetic and superconducting materials. from the Royal Society of Chemistry. Ahmed Iraqi received his B.Sc. (Chemistry) from the University of Fes in 1984. He then moved to the Universitt Paul Sabatier in Toulouse where he obtained M.Sc. (1985) and Ph.D.(1988) degrees. He took up an appointment as a Research Fellow at St. Andrews in 1988 and has since worked on organometallic chemistry, homogeneous catalysis, and synthetic and electroche- mical aspects offunctionalized conducting polymers. John Walton graduated from Shefield University in 1963 and studied with Lord Tedder for his Ph.D. which was awarded by St. Andrews in 1970. He is now a Reader in Organic Chemistry at St. Andrews. He has been a frequent Visiting Scientist at the National Research Council of Canada in Ottawa and, in collaboration with Keith In-gold, has examined the struc- tural and kinetic behaviour of many free radical types. His current research interests include synthetic and mechan- istic studies of reactive inter- mediates, organic and bioche- mical aspects of dioxygen and nitric oxide behaviour, and functionalized conducting polymers.2 Synthetic Strategies Four basic strategies have been utilized for the production of various types of polyradicals. (i) A monomer containing a vinyl, diacetylene, thiophene, pyrrole, etc. unit, to which is attached a stable free radical, is polymerized by free radical, anionic, topochemical photolytic, or electrochemical means, to give a polyradical directly. This approach is severely limited by the fact that the radical centre in the monomer often inhibits, prevents, or diverts the polymerization process. (ii) A similar monomer, but functionalized with a free radical precursor is polymerized.Subsequent treatment of the polymer, usually oxidation or reduction, converts the precursors into free radicals. For exam- ple, polymers containing hindered secondary amine groups may be oxidized to poly(nitroxides), polymers containing quinones may be reduced to poly(semiquinone radical anions), and polymers with diazo groups may be photolytically cleaved to give poly(trip1et carbenes). (iii) Stable free radicals are chemi- cally attached (grafted) to a preformed polymer. The advantage of this approach is the good control which can be exercised over the molecule weight by selection of an appropriate starting polymer such as a poly(butadiene) or a poly(acry1ic acid). (iv) Free radical precursors are chemically attached to a preformed polymer which is subsequently oxidized, reduced, or otherwise chemically treated to give the polyradical.The main drawback to strategies (ii) and (iv), which give precursor polymers, is the difficulty of ensuring complete conversion of the precursor groups, particularly if, as often happens, the polymer is not very soluble in inert reaction solvents. For example, synthesis of copolymers, (3), which were prepared by irradiation of an oxallyl ester with an attached phenol, were recently reported.* 1-BUY B ” 4 Allowing for the co-radicals produced in the photochemical reaction, only about 25% of the theoretical concentration of polymer-bound spins was observed. The disruption of the radical array in incompletely converted polyradicals can have drastic effects on any desired cooperative activity of the final material, for example, the electrical conductivity or ferro-magnetism. Consequently, optimization of the yield of radical centres is a major concern.3 Specialized Methods of Characterization 3.1 EPR Spectroscopy of Polyradical~~ Spin-spin dipolar interactions in solid state polyradicals lead to a broadening of the EPR lines compared with solution spectra. In favourable cases the degree of broadening can be calculated from known or estimated spin densities and the distance the spins are apart. Dipolar interactions are typically on the order of to 10-l cm-l (10 to 100 mT). A biradical, for example, generally exhibits a six-line spectrum in frozen glasses; from this spectrum two parameters D and E may be extracted.The value of D is related to the distance the two spins are apart, and the degree of interaction of the spin-containing orbitals. A conju-gated, or folded over, biradical has a D of ca. 0.06 cm- l, while a non-conjugated biradical has a much lower value of D. On the other hand, exchange interactions (J)can be larger than the dipolar interaction by up to two orders of magnitude. They can bring about exchange-narrowing of linewidths to approach those of the monomer radicals in solution. If the observed linewidth of the solid polyradical is the same as that for the polymer in a dilute solution or in a dispersed solid, then it is reasonable to assume that the intrachain exchange interactions CHEMICAL SOCIETY REVIEWS, 1994 dominate.If the observed linewidth is broader than in solution, then interchain exchange interactions are important. As the temperature is lowered the dipole-dipole linewidth increases, while the exchange-narrowed linewidth is unchanged. The integrated signal intensity is proportional to the magnetic susceptibility, x: This may be useful when the contributions of two different signals to the bulk magnetic susceptibility are under considera tion. 3.2 Electrical Conductivity Polymer conductivities span a wide range. For example, Teflon has a conductivity of < lo-’* S/cm (Q-l cm-l) and is an insulator whereas polythiophene, which has a conductivity of lo2 S/cm in its oxidized (‘doped’) form, is a good conductor.These differences reflect the number and mobility of itinerant electrons or holes. The electrical conductivity of a polymer sample indicates whether the charge carriers are delocalized and mobile, giving rise to high conductivities, or localized and immobile, leading to low conductivities. Many novel electrically conducting polymers are prepared as powders, and must be pressed into pellets before the conductivity is measured, gener- ally by attaching four fine gold wires or by pressing four graphite contacts onto the sample. These contact methods prevent cracking of brittle pellets. Two of the contacts are used to pass a known current, the other two measure the voltage drop across the sample: this is called the four-point probe method, a procedure which eliminates the effect of contact resistance.A pressed powder sample, however, may have a lower conductivity than the intrinsic conductivity because of the resistance between the powder particles and the inherent defects in the polymer chains. Defect-free, oriented-chain doped-polyacetylene sam- ples have been produced with conductivities of up to lo5 S/cm, approaching those of metals. Turning to polyradical samples, since they contain a large number of unpaired electrons which have singly-occupied (or vacant) orbitals in the adjacent radical centre to jump to, we should expect the conductivity to reflect the mobility of the electrons. A high conductivity for a polyradical indicates that the unpaired electrons are mobile and metallic; this will tend to cause the electrons to pair up.However, a low conductivity would indicate immobile, localized electrons and the possibility of unpaired spins. 3.3 Electrochemistry Many radical groups are redox-active, and hence electrochemi- cal measurements can confirm that the redox potential of the polyradical is similar to that of the monoradical redox unit; a broad redox couple indicates extensive hydrophobic or electro- static interactions between groups. The percentage redox activity (charge associated with each monomer unit) of a sample indicates whether the radical groups are accessible to the elec- trode and whether fast electron transfer between redox groups is possible. Immobilization of the polymer onto the electrode surface is very convenient for making these measurements, as well as measuring DCt,the diffusion coefficient of charge trans- port within the polymer, which is a sensitive function of polymer solvation, flexibility, and structure.There is also a great deal of interest in measuring the rate of redox reactions mediated by the polymer interior and outer surfaces. In many cases electroche- mical oxidation is a convenient way to synthesize conjugated polyradicals, or to generate in situ radical groups from non- radical precursors. 4 Unconjugated Polyradicals The first rather ill-defined polyradicals were prepared in the mid ’50s by incorporating DPPH (diphenyl picryl hydrazyI), triaryl- methyl, or verdazyl radicals into poly(styrene) by irradiating with 7-rays and subsequently oxidizing the polymer^.^ Subse-quently, the most popular unconjugated polyradicals have POLYRADICALS: SYNTHESIS, SPECTROSCOPY, AND CATALYSIS-J.A. CRAYSTON, A. IRAQI, AND J. C. WALTONRx0 0' 0vAlBNlAcOH ___)A -,cJ,H A-H A-X=O,NH A = CI,HSOd contained 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO) as the pendant stable radical. For example, poly(TEMPOacrylates), (3,have been made by anionic polymerization of monomer (4) prepared from TEMPOL and a substituted acrylic acid chlor- ide.s The extent of polymerization depends critically on reaction conditions, but molecular weights up to 140K have been reported. Free amine monomers inhibit polymerization with free radical initiators.However, the analogous hydrochloride (6) or hydrogen sulfate salts, gave good yields of polymer with AIBN as initiator, provided acetic acid was used as the solvent. Conversion of the polymer (7) into polynitroxide (8) was effected by oxidation with hydrogen peroxide. The free radical route via amines enabled copolymers incorporating unsubsti- tuted styrene, methyl methacrylate, and divinylbenzene units to be assembled. The polynitroxides (5) and (8) are orange-brown paramagne- tic powders which are insoluble in hexane but sparingly soluble in acetonitrile. Their redox potentials and chemical reactivity are similar to those of monomeric nitroxides.s Apart from the possibility of using them as minimum disruption anti-oxidant additives for bulk polymers, the main interest in these materials is associated with their use as catalysts for the oxidation of organic substrates.For example, a catalytic amount of (8) in acetonitrile was employed, in conjunction with a second aqueous phase containing alkaline K,[Fe"'(CN),], to oxidize benzyl alcohol to benzaldehyde in yields of > 50% These polynitroxides were also effective as coatings on elec- trodes for the electrochemical oxidation of amines. For exam- ple, an electrode made by dip coating (5)(R = H) onto platinum was effective in the oxidation of a range of amines.s The oxidation was also successful on a preparative scale using a carbon fibre cloth electrode coated with a monolayer of (5)(R = H). Electrolysis at + 0.85V of benzylamine in anhyd- rous acetonitrile yields benzonitrile; addition of water gives instead a good yield of benzaldehyde. The oxidation takes place viaa catalytic cycle involving the oxoammonium ion, nitroxide, and hydroxylamine.In general, the efficiency of this electrosyn- thesis depends on the structure of the amine, but catalytic currents and turnover numbers are high for most primary amines. A large number of papers have reported electrochemical studies on redox polymers but we shall mention only two. (i) AWN -0 (ii) hv Ph Unconjugated polyradicals containing viologens have been made by AIBN-catalysed copolymerization of 1-propyl-1'-vinyl-4,4'-bipyridinium diperchlorate (9) and vinyl benzophe- none., The resulting copolymers were converted into deep blue poly(vio1ogens) (10) in DMSO solution by photo-induced elec- tron transfer.The ease of this conversion suggested that poly- (viologens) might be used as photochromic materials. Nambu et al. showed that intermolecular charge-transfer complexes, formed between viologen and benzophenone groups inside a copolymer, could mediate the photo-reduction of viologen by DMSO. Evidence for this interaction was also obtained from electrochemical studies.6 Wrighton et al. described a redox polymer containing quinone groups sandwiched by viologen units. The quinone was inaccessible to the electrode and could only be reduced via electron transfer from viologen groups.' 5 Conjugated Polyradicals 5.1 Functionalized Polypyrroles and Polythiophenes Pyrrole, thiophene, aniline, and several other monomers can be polymerized either by electrochemical oxidation or by chemical oxidizing agents to give black polymers with backbones contain- ing long sequences of conjugated double bonds.These materials become electronic conductors when oxidized or reduced (doped). This is explained by the creation of delocalized charge centres which may be singly-charged (polarons) or doubly charged (bipolarons), both of which can migrate along the conjugated chain under the influence of an applied potential. If the monomers have chemically bound side groups containing redox centres then the resulting polymers are conjugated polyra- dicals of type (2). Most work has been carried out with pyrrole and thiophene derivatives using electrochemical polymeriza- tion.The polymers form as layers on the anode. Modified electrodes made in this way can be activated by applying the appropriate potential, and used for the catalysis of electron transfer initiated chemical transformations. The polymerization process links pyrrole nuclei via their 2-and 5-positions so that any functional groups must be attached at the pyrrole nitrogen atom (1 -position) or the 3-position of the pyrrole nucleus. The latter derivatives are quite difficult to prepare so that only a few such polymers have been studied. These include poly( 11) which was readily formed by electroche- mical oxidation; films were highly electroactive. 0 Many different polyradicals have been made from N-substi- tuted pyrroles. A thorough exposition of this area has been given in recent interesting review^.^ The chain linking the substituent to the pyrrole nitrogen must be at least two methylene units long for electrochemical polymerization to be successful. Polypyr- roles with reducible substituents include the quinone-containing poly( 12), which catalysed the reduction of oxygen, and viologen containing-poly( 13).O Platinum electrodes coated with violo- gen-containing poly(pyrro1es) were shown to catalyse the reduc- tion of organo-halogen compounds. 0 For pyrroles linked to oxidizable substituents the potential needed to cause polmerization may be greater than that required to oxidize the substituent.Regardless of this, several such monomers have successfully been polymerized by cyclic voltam- metry, provided that the potential is swept sufficiently positive for initiation of the oxidative pyrrole coupling process -as we have shown for a pyrrole monomer linked to a TEMPO group. Electrodes coated with poly( 14) exhibited a reversible couple due to oxidation of nitroxide to oxoammonium cation. In the presence of an added alcohol the oxoammonium ions were reduced to the hydroxylamine and the alcohol was oxidized to aldehyde or ketone. Then, reoxidation of the electrode-bound hydroxylamine to oxoammonium cations prepared the system for another catalytic cycle. Although this system successfully converted alcohols into aldehydes/ketones on an analytical scale, problems with the electrode coating stability were encoun- tered on scaling up the process.Ruthenium complexes made from pyrroles N-functionalized with pyridine and bipyridine ligands such as poly(l5) and poly(l6) were found to be better electrocatalysts for alcohol oxidation. l Many other poly(pyr- roles) with attached metal complexes of pyridines, porphyrins, and cyclams have been inve~tigated.~ Thiophenes functionalized at the 3-position can be polymer- ized by controlled potential oxidation only if electron withdraw- ing groups are remote from the thiophene. Usually the redox functional group must be separated from the thiophene by a methylene chain of at least two units. Poly(thiophenes) with quinone side chains such as (1 7)' and viologen side chains such CHEMICAL SOCIETY REVIEWS, 1994 (17) (18), (X = 4-10) as (18)16 have been made and studied electrochemically. Elec- tronic conductivities up to 6 S/cm have been recorded and these materials are being tested for electrochromic devices and as microelectrochemical transistors.5.2 Co-operative Magnetic Interactions in Conjugated P~lyradicals~~ For solids composed of ordered arrays of atoms or molecules, each containing one or more unpaired electrons, there are several possibilities for intermolecular cooperative behaviour amongst the spins. Most commonly, the exchange interaction between the unpaired electrons is negative so that antiparallel coupling occurs and results in overall antiferromagnetic behav- iour.To achieve macroscopic ferromagnetism the exchange interaction must be positive so that parallel alignment of spins is obtained over extended regions of the solid. The goal of creating an organic ferromagnet has stimulated a lot of attention from innovative scientists. Indeed it seems to exercise the same attraction as the proverbial lodestone. Several ways whereby ferromagnetism might be achieved have been suggested by theoreticians. Of the three main models, the Topological Model, has provided greater inspiration in the design of polyradical magnets than the well known McConnell (I)and McConnell(I1) mechanisms. In the Topological Model ferromagnetic coupling (strictly, intramolecular high spin states arising from Hund's rule), occurs in very large organic species which contain bands of degenerate, or nearly degenerate, orbitals as a consequence of their molecu- lar topology.For example, in polymeric triplet carbenes (19), or radicals (20)(R = H, Ar), in which the aromatic substitution is meta, the frontier m-orbitals constitute three distinct bands. In polymeric species this topological arrangement leads to the formation of a band of non-bonding MOs (NBMOs) in addition to the full or valence band and the vacant or conduction band. The poly(phenylcarbenes)( 19) have additional 0-non- bonding MOs which are of similar energy to the n-non-bonding MOs and contribute to the NBMO. If the energies of the non-bonding orbitals are all the same, or nearly so, they constitute a family of superdegenerate orbitals in which all the unpaired electrons will have parallel spin.Polyradicals of this type have the potential to wqQ$n-2 0. n-2 0 R R (19) (20),(R = H, Ar) POLY RADICALS: SYNTHESIS, SPECTROSCOPY, AND CATALYSIS-J. A. CRAYSTON, A. IRAQI, AND J. C. WALTON behave as linear ferromagnets. Magnetic resonance and magne- tic susceptibility measurements on several oligomeric carbenes of type (19) in the solid state showed that they do indeed exhibit the predicted high spin multiplicities. For example, the penta- carbene, (19)(n = 6) was found to have an undecet electronic configuration. However, these oligocarbenes were only stable at cryogenic temperatures and so far there has been no success in making polymeric analogues.Building on this same basic concept, Fukutome and co- workers mapped out the design of 'polaronic ferromagnets' which consist of lengths of conjugated polyene (polyacetylene) chain connected by n-systems with an odd number of sp2-carbons.20 Unpaired electrons are then created by doping the conjugated blocks. The linking odd n-systems, e.g. meta-substi- tuted aromatic rings, build in the correct topology to induce ferromagnetic coupling of the spins. For example, oxidation of (21) should create polarons in the tetraene blocks which will be separated by meta-substituted aromatic rings. The latter intro- duce three (or five) sp2-carbon atoms into the n-system so that the polymer will possess a band of non-bonding superdegenerate orbitals which will constrain all the polaron spins to be parallel.Similarly,if blocks of thiophenes of sufficient length to support polarons were linked by carbonyl groups, which introduce one spz-carbon into the chain (22), they should ferromagnetically couple the induced polarons. R A polymer of type (2l), with octadecyloxyl substituents (R = OC, 8H3,)to promote solubility, was made by Dougherty and co-workers.z This material, when oxidatively doped with AsF,, showed weak ferromagnetic coupling. The small magni- tude of the ferromagnetism was accounted for by intermolecular antiferromagnetic coupling between segments of the chain situated close together in the solid. Moreover, the doping was inhomogeneousso that the number of spins introduced was too small for unifo- interaction to develop.Further design is needed -for example, more controlled doping could be achieved by electrochemical methods -but this is obviously a promising strategy in the search for organic ferromagnets. A further attempt to arrange the radical units in accord with the topological model was based on the oxidative coupling of 1,3,5-triaminobenzenes to provide polymers conjugated in two dimensions. Polymer (23), for which ferromagnetic interactions have been claimed, was prepared as a precipitate from a solution of triaminobenzene in acetic acid.2 A one-dimensional version of this polymer, i.e. (24), was prepared by the Ullmann coupling of rn-bromoaniline.However, doping of this material led to a spin concentration of only 1 in 60 monomer Ovchinnikov showed that alternant hydrocarbons (and .*-ON+.ON+.. heteroatomic analogues), for which the number of atoms in the two sets was unequal, would have a finite spin multiplicity. He identified several series of molecules, with members increasing by a regular repeating unit, for which the resultant spin would be proportional to the size.24 Other theoretical studies have shown, however, that pendant radical groups may interact with each other via a conjugated backbone.2 These examples included hypothetical polyacetylenes and polyacenes as well as conju- gated polyradicals such as (25) in which the unpaired electron would be conjugated with the backbone n-system.Several polyradicals of this general type, e.g. (26)26 and (27)27 and the galvinoxyl derivative (28)2 8, have been prepared, but the solids did not display bulk ferromagnetism; probably because pairing of adjacent spins took place. The surprisingly stable (27) with up to 38% of the theoretical spins was prepared from the parent phenol by oxidation with either Pb02 or alkaline ferricyanide; addition of base produced the polyanion. The visible spectrum of (27) indicated a fairly long n-system, and the EPR spectrum of the polyradical in solution implied coupling to the backbone protons. This suggested that spin was delocalized onto the chain and a quinone group was formed (27b). The reversible decrease in the EPR signal at T< 50 K indicated an antiferromagnetic interaction.00. 0. 0 t-vu Bu-t 0 Bu-t Bu-t Polyradical (28) was prepared in a similar way, giving up to 80% of the theoretical number of unpaired spins. The EPR spectrum in solution showed evidence for intrachain dipolar interactions. In the solid state the EPR linewidth of the 77% spin polymer indicated both intrachain exchange and interchain dipolar interactions. There was no evidence, however, for ferro- magnetic spin ordering in this sample. It is interesting to contrast the essentially localized spins of (28) with the chain-delocalized spins of (27), which unfortunately couple antiferromagnetically . Other polyradicals with polyacetylene and polyimine backbones containing nitroxide, nitronylnitroxide, and verdazyl units have been examined but without finding evidence of ferromagnetic behaviour.Perhaps the system to attract most attention has been that based on 4,4'-(butadiyne- 1,ddiyl)-bis(TEMP0L) (29). Follow- ing the original work with (29), numerous attempts have been made to prepare ferromagnetic polymers from diacetylenes substituted with nitroxides, or other stable free radicals, either by topochemical photolysis, pyrolysis, or mechanical means. This research has been critically reviewed recently.29 Although claims of ferromagnetic behaviour have been made, the reported saturation magnetizations have been very low, and the polymers have been chemically ill-defined. It seems probable that in most, if not all cases, the magnetism is due to trace amounts of iron- containing impurities. 5.3 Polyradicals: Electronic Conductivity and Superconductivity According to the classic BCS theory, superconductivity is main- tained by mobile pairs of non-spin-paired electrons (Cooper pairs) in which the electrons are coupled to each other via phonons, i.e.vibrational modes of the lattice material. Even for crystalline solids, theories explaining how Cooper pair forma- tion is related to short and long range molecular and crystal structure are rudimentary, and theoretical guide-lines of this type are non-existent for organic or organometallic materials. It has been pointed out that the doubly charged bipolaron, thought to be present in doped conjugated polymers, is related to the Cooper pair, but so far no reports of superconductivity have been published.Data from the new Cu-oxide and related high T, superconductors indicate that although the charge carriers are electron pairs, the force that keeps them together is not always phonon-coupling. 30 Conjugated polyradicals typi- cally contain two types of unpaired electron which could, depending on the details of the structure, move in concert over long distances through the infrastructure. Poly(su1fur nitride), (SN),, which possesses one unpaired electron in each SN unit, becomes superconducting below 0.3 K.3 The doped polynitroxide (30) potentially has mobile polarons in the delocalized backbone in addition to the unpaired electrons of the nitroxide groups. Experimental investigati~n~~ showed that (30) has low electronic conductivity probably because electron pairing via species such as (30b) is important. In addition, oxidation of the poly(N-hydroxypyrrole) precursors was accompanied by significant disruption of the chains.More promising candidates appear to be poly(thiophenes) or poly- (pyrroles) like (31), or more generally (32), where L are non- conjugated linkages and R’are stable free radicals. Polarons created in the conjugated backbone may be able to form Cooper pairs with unpaired electrons of the nitroxide groups. Transfer of the nitroxide electron to an adjacent oxoammonium site will convert the donor nitroxide into oxoammonium, and the accep- tor oxoammonium into nitroxide.Potentially this could occur in response to the passage of a polaron along the backbone, depending on the nature of the linkage L, and hence electron- pair mobility along the chain would be actuated. Polyradicals of this type with nitroxide, quinone, viologen, and complexed transition metal side-chains are being studied. 6 Future Developments Electrodes coated with polyradicals containing nitroxides show good promise as catalysts for the electrochemical oxidation of OH CHEMICAL SOCIETY REVIEWS, 1994 amines, alcohols, and other organic substrates. Similarly, it is likely that polyradicals functionalized with quinones or violo- gens will be developed for the electrochemical reduction of a range of substrates.There is a need for better methods of permanently attaching the polyradical films to the metal sur- faces so that more robust electrodes with longer active lifetimes can be produced. Modified electrodes of this type will find use as probe elements in sensors. The polaronic magnets are obviously one of the most fascinating emergent concepts. A wide range of structural types have still to be investigated. There is a conspi-cuous need to improve chemical stability, perhaps by using oligothiophene rather than polyene blocks for housing the polarons. To prevent the interchain interactions which antifer- romagnetically couple spins, structures with two- or three- dimensional regularities need to be designed and synthesized. There seems to be no limit to the functional groups which can be incorporated into polymers, or to the physical and chemical behaviour that such materials can mimic.7 References 1 ‘Handbook of Conducting Polymers’, ed. T. A. Skotheim, Vols. 1 and 2, Marcel Dekker, New York, 1986. 2 F. C. Rossitto and P. M. Lahti, J. Polym. Sci., Polym. Chem. Ed., 1992,30, 1335. 3 A. Bencini and D. Gatteschi, ‘EPR of Exchange Coupled Systems’, Springer-Verlag, Berlin, 1990, pp. 135-166; pp. 187-192; pp.253-265. 4 A. Henglein and M. Boysen, Makromol. Chem., 1956, 20, 83; D. Braun, I. Loflund, and H. Fischer, J. Polym. Sci., 1962,58,667. 5 0.H. Griffith, J. F. W. Keana, S. Rottschaefer, and T. A. Warlick, J. Am. Chem. SOC., 1967,89,5072; F. MacCorquodale, J. A. Crayston, J.C. Walton, and D. J. Worsfold, Tetrahedron Lett., 1990,31, 771. 6 Y. Nambu, Y. Gan, C. Tanaka, and T. Endo, Tetrahedron Lett., 1990,31,891; Y.Nambu, K. Yamamoto, and T. Endo, Macromole-cules, 1989,22, 3530. 7 D. K. Smith, G. A. Lane, and M. S. Wrighton, J. Am. Chem. SOC., 1986,108,3522. 8 C. P. Andrieux and P. Audebert, J. Electroanal. Chem., 1989, 261, 443; C. P. Andrieux, P. Audebert, and J.-M. Savkant, Synth. Met., 1990,35, 155. 9 A. Deronzier and J.-C. Moutet, Acc. Chem. Res., 1989,22, 249; D. Curran, J. Grimshaw, and S. D. Perera, Chem. SOC. Rev., 1991,20, 391. 10 P. Audebert, G. Bidan, and M. Lapkowski, J. Electroanal. Chem., 1987, 219, 165; P. Audebert and G. Bidan, J. Electroanal. Chem., 1987,238,183; J. Grimshaw and S.D. Perera, J.Electroanal. Chem., 1990, 281, 125; L. Coche, A. Deronzier, and J.-C. Moutet, J. Electroanal. Chem., 1986, 198, 187. 11 L. Coche and J.-C. Moutet, J.Electroanal. Chem., 1988,245,313; L. Coche and J.-C. Moutet, J. Electroanal. Chem., 1987,224, 11 1. 12 J. A. Crayston, A. Iraqi, P. Mallon, J. C. Walton, and D. P. Tunstall, Mol. Cryst. Liq.Cryst., 1993, 236, 231. 13 W. F. De Giovani and A. Deronzier, J.Chem. SOC., Chem. Commun., 1992, 1461; W. F. De Giovani and A. Deronzier, J. Electroanal. Chem., 1992,337,285. 14 S.Cosnier, A. Deronzier, and J.-C. Moutet, J. Electroanal. Chem., 0’ 0- POLYRADICALS: SYNTHESIS, SPECTROSCOPY, AND CATALYSIS-J. A. CRAYSTON, A. IRAQI, AND J. C. WALTON 1985,193, 193; S. Cosnier, A. Deronzier, and A.Llobet, J. Electroa-nal. Chem., 1990,280,213. 15 J. Grimshaw and S. D. Perera, J.Electroanal. Chem., 1990,278,287; J. A. Crayston, A. Iraqi, P. Mallon, J. C. Walton, and D. P. Tunstall, Synth. Met., 1993,55-57,867. 16 C.-F. Shu and M. S. Wrighton, in ‘Electrochemical Surface Science’, ACS Symp. Ser., 1988,Chapter 28, p. 408; P. Bauerle, K.-U. Gaudl, and G. Gotz, Springer Ser. Solid State Sciences, 1992,107, 384. 17 For reviews see: ‘Magnetic Molecular Materials’, ed.D. Gatteschi, 0.Kahn, J. S. Miller, and F. Palacio, Kluwer, Dordrecht, Nether- lands, 1991; F. Wudl and J. D. Thompson, J. Phys. Chem. Solids, 1992,53, 1449. 18 N. Mataga, Theor. Chim. Acta, 1968,10, 372. 19 Y. Teki, T. Takui, K. Itoh, H. Iwamura, and K. Kobayashi, J. Am. Chem. Soc., 1986,108,2147;A. Izuoka, S. Murata, T. Sugawara, and H. Iwamura, J. Am. Chem. Soc., 1987,109,2631; D. A. Dougherty, Ace. Chem. Res., 1991,24, 88. 20 H. Fukutome, A. Takahashi, and M. Ozaki, Chem. Phys. Lett., 1987, 133, 34. 21 D. A. Kaisaki, W. Chang, and D. A. Dougherty, J.Am. Chem. Soc., 1991,113,2764. 22 I. Johannsen, J. B. Torrance, and A. Nazzal, Macromolecules, 1989, 22, 566. 23 K. Yoshizawa, K. Tanaka, T. Yamabe, and J. Yamauchi, J. Chem. Phys., 1992,92, 5516. 24 A. A. Ovchinnikov, Theor. Chim.Acta, 1978,47, 297; Dokl. Akad. Nauk SSSR, 1977,236,957. 25 P. M. Lahti and A. S. Ichimura, J. Org. Chem., 1991, 56, 3030; J. Chandrasekhar and P. K. Das, J. Phys. Chem., 1992,96,679. 26 C. Alexander and W. J. Feast, Polym. Bull., 1991, 26, 245; C. Alexander, W. J. Feast, R. H. Friend, and L. H. Sutcliffe, J. Mater. Chem., 1992,2,459. 27 H. Nishide, N. Yoshioka, K. Inagaki, T. Kaku, and E. Tsuchida, Macromolecules, 1992,25, 569. 28 N. Yoshioka, H. Nishide, T. Kaneko, H. Yoshiki, and E. Tsuchida, Macromolecules, 1992,25, 3838. 29 J. S. Miller, Adv. Muter., 1992,4, 298; 435. 30 H. Jaeger, Adv. Mater., 1990,2, 16. 31 R. L. Greene, G. B. Street, and L. H. Suter, Phys. Rev. Lett., 1975,34, 577; M. M. Labes, P. Love, and L. F. Nichols, Chem. Rev., 1979,79, 1. 32 C. Kakouris, J. A. Crayston, and J. C. Walton, Synth. Met., 1992,48, 65.

ISSN:0306-0012    

DOI:10.1039/CS9942300147

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Chemistry in Near-critical Fluids Roberto Fern6ndez-Prini1n2 and M. Laura Japas" 'Departamen to Quimica de Reactores, Comisih Nacional de Energia Atomica, A v. Libertador 8250,1429-Capital Federal, Argentina 21nquimae, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Pabellbn 11, Ciudad Universitaria, 1428-Capital Federal, Argentina I Introduction The physico-chemical description of the fluid state of matter has been accomplished in two steps separated by a significant time- interval. Gases were among the first systems to be studied in Physical Chemistry; it was possible to obtain an explanation of many of their properties in terms of intermolecular parameters -a well known case is the van der Waals equation of state.' On the other hand, the liquid phase proved more elusive and in fact the properties of solids had a firmer theoretical basis much earlier than those of liquids. It was only in the second half of this century that the theory of high density fluids showed significant progress, and this has been strongly accelerated in the past thirty years by the increasing capacity of computer simulation to deal with more complex sy~tems.*3~ In spite of this progress the behaviour of fluids in the thermodynamic region which surrounds the critical point are not so well understood.This thermodynamic region is essential in order to bridge completely the gaseous and liquid states of matter, and fluid systems in the neighbourhood of critical conditions are now attracting increasing interest.Critical phe- nomena had been studied before any successful molecular theory for liquids was a~ailable,~ but widespread study of near- critical behaviour has only occurred quite recently. The reason for its increasing popularity is related to the interest in using supercritical fluids as suitable reaction media for many pro- cesses, in particular for extraction and purification of thermally labile substances; they have also found wide acceptance as mobile chromatographic phases. The most appealing feature they exhibit is that their density may be varied in a continuous manner without the occurence of a liquid-vapour phase transi- tion. This feature allows the control of their density providing a means offine tuning intermolecular interactions, thus helping to optimize chemical processes.M. Laura Japas obtained her B.Sc. (1978) and her Ph.D. (1986) degrees from the University of Buenos Aires. Since 1979 she has been a stafmember of the Department of Reactor Chemistry of the National Commission of Atomic Energy (Argentina). She was a research fellow with Professor E.U. Franck at the University of Karlsruhe and with J.M.H. Levelt Sengers at the Thermophysics Division of the National Institute of Standarh and Technology. She has been research assist- ant at the Chemical Engineer- ing Department, University of Delaware. Her research inter- ests include physicochemical behaviour of fluid systems under near critical conditions. 155 It is also true that fluids close to a critical point show a behaviour which is different from that of ordinary fluids, thus increased scientific interest has accompanied the recent interest in the application of these fluids.For the purpose of clarifying the scientific aspects of their behaviour, it is useful to consider the properties of subcritical fluid systems which are close to the critical region, but may undergo phase transition. Their study has shed light on the characteristic features of near-critical behaviour. The near-critical state of fluid matter is affected by an enhanced susceptibility (long-range effect) typical of the proxi- mity to the critical region, and by the fluids' lower density (short range effect). The main interest for the purpose of their appli- cations resides in understanding dilute solutions which are the most frequently used and which exhibit the most unusual behaviour.While this article was being written Tom and Debe- nedetti published5 a study of supercritical mixtures with a very similar approach which inspired and influenced the final version of the present review. We shall attempt to rationalize some of the observations in terms of intermolecular interactions and the resulting thermodynamic properties; this will be illustrated with some examples selected from the great number of practical applications. Some critical phenomena which occur very close to the critical point, e.g.critical opalescence, density stagnation by gravitational effects, etc.will not be considered in this work -they are relevant when inquiring into the behaviour observed in a thermodynamic region which is much closer to the critical points than the one considered in the present work. 2 Critical State of Matter It is convenient to start by illustrating the critical phenomenon with the relatively simpler behaviour observed in a pure fluid. A Roberto Fernandez-Prini graduated and obtained his Ph.D. from the University of Buenos Aires where he was appointed full Professor in 1971. From 1983 to 1993 he acted as head of the Department of Inorganic, Analytical, and Physical Chemistry and now heads the Institute of Chemical Physics for Environment, Materials and Energy. He is senior scien- tist at the Comisibn Nacional de Energia Atbmica. His re-search interests cover the field of chemical thermodynamics of fluids, extending from non-ionic to ionic systems and in- cluding critical behaviour, with particular emphasis in the relation between molecular properties and macroscopic beha viour .CHEMICAL SOCIETY REVIEWS, 1994 Figure 1 Liquid-vapour equilibrium illustrating the difference in par-ticle densities between equilibrium liquid and vapour phases. substance in the gaseous state at very low pressure is slowly compressed at temperature T, (Figure l), when its density reaches a certain value, p(g, T,),two equilibrium states having very different densities become possible: (i) a low-density vapour having a large entropy content, but a relatively small (absolute) enthalpy since intermolecular interactions are weak because the average distance between molecules is relatively long;* (ii) a liquid phase of higher density p(1, Tl),having a consequently smaller entropy, the decrease in fluid entropy upon condensa-tion is compensated by a larger enthalpic contribution due to stronger interactions among molecules which are now much more closely packed.The substance is in two equilibrium states which are a compromise between the greater number of spatial Configurations (vapour phase) and an increase in attractive interactions (liquid phase). As the temperature increases it will be necessary to compress the vapour more before it can coexist with another fluid phase having higher density.Increasing the temperature reduces the enthalpic contribution because the ratio of intermolecular energy to average thermal energy decreases; hence it becomes more difficult for the dense phase to compensate the loss of entropy due to condensation with an increase in intermolecular interactions, and the difference in densities between the two coexisting phases is reduced. A temperature will be reached, called the critical temperature, T,, above which the system will not be able to generate another stable phase by a discontinuous change of its density -phase separation will not be possible. For vapour-liquid equilibrium at constant (T,p,,) the fluid density (or its molar volume V = p-l) has only two possible values which correspond to the densities of the phases in equilibrium,i.e. p(g, T,)and p(1, Tl).In Figure 2 the isotherm T, illustrates the dependence of the Gibbs energy, G,with the molar volume at a temperature much lower than T,and atpSat.The two equilibrium states correspond to minima in G =A V) which are of the same depth; its curvature is a measure of the mechanical stability of the equilibrium states of the fluid, and is equal6 to =-(d~/aE')~(k'df)-l, where KF is the isothermal compressibi- lity (susceptibility). At low temperatures the minima are sharp (small susceptibiIity) -the probability ofexistence of states close to, but different from, the minima is very low. The particle density exhibits only small deviations from its mean equilibrium values (fluctuations) which are only appreciable within the molecular range; they are averaged out when larger (macro- scopic) scales are considered.As shown in Figure 2, closer to the critical region the minima in G(V) become closer to each other and the height of the barrier separating them and the curvatures at the minima are significantly reduced. As a consequence, density fluctuations become more important and they persist beyond the immediate molecular region. The size of the Auctuat- ing domains, the correlation length 5, becomes increasingly * The fluid packing fractions, Le. the fraction of space occup~edby the molecules, is typically 0.5 for iiquids near their triple points and 50.05 for vapours, Figure2 Gibbs energy of a pure fluid as function of its molar volume at constant T,p.The minima correspond to the states of equilibrium. Temperatures as in Figure 1. bigger as thecritical point is approached.+ In Figure 2 the critical isotherm illustrates the fact that at the critical point the curva- ture of G is zero, its susceptibility becomes infinite, and the state has only marginal stability. The important role of long-range fluctuations in critical phenomena cannot be overemphasized. A spontaneous local density fluctuation in a fluid may be construed as the response of that region of the fluid to a local pressure perturbation. Since at the critical point the fluid's susceptibility is infinite, the system responds strongly to perturbations. Thus a local pressure change will affect all the system, because at the critical point fluctuations have an infinite range, i.e.all the molecules in the fluid are correlated and the correlation length becomes infinite. Far from the critical point a perturbation which takes the fluid away from an equilibium state will bestrongly resisted according to the laws of Thermodynamics (Le Chatelier principle). Close to the criti- cal region the fluid will be in a state highly indifferent to molecular configuration, this at the expense of thermodynamic stability which is strongly reduced, hence intermolecular inter- actions will be expected to affect the macroscopic properties differently. Macroscopic properties in this region reflect the presence of all-scale fluctuations. Analytical mean-field equations are unable to describe the observed critical behavi~ur.~ The strategy to formulate the state of the system is a step-by-step renormali- zation which leads to a rescaling of all spatial vectors averaging out fluctuations over increasing length scales, from molecular microscopic to macroscopic.The dimensionalities of the spatial and order parameters enter into the theoretical calculations determining what is known as the universality class of the critical phenomena. One-component systems at the liquid-vapour criti- cal point and binary mixtures at liquid-vapour or liquid-liquid consolute points, belong to the same universality class, hence these systems exhibit the same behaviour while approaching a critical point.3 Mixtures For the description of the critical behaviour of mixtures it is convenient to classify the thermodynamic variables into: densit-ies, which are extensive quantities having different values in the two coexisting phases, like entropy, volume, enthalpy, density, orcomposition; and fields, which are intensive quantities having the same value in both phases at equilibrium, like pressure, temperature, and chemical potential. Griffiths and Wheelers established a convenient generalization of the divergent behav- The phenomenon knownascritical opalescence, which occurs very CIOW to T,, indicates that the fluctuating domains have the size of the wavelength of visibie hght. CHEMISTRY IN NEAR-CRITICAL FLUIDS-R.FERNANDEZ-PRINI AND M. L. JAPAS iour of thermodynamic quantities in terms of fields and densit- ies. Thus, at the critical point the partial derivative of a density with respect to a field will diverge strongly if the path along which the differentiation is made is defined only in terms of constant fields; it will diverge weakly (i.e. noticable only very close to the critical point) if one density is kept constant; it will not diverge if more than one density is fixed along the path. Other differential quotients can be reduced to the above in order to ascertain whether they are bound when approaching a critical point. The thermodynamics of solutions at temperatures close to the solvent’s triple point is usually described in terms of partial molar properties Xi,i.e.derivatives of a density Xwith respect to the number of moles of component i at constant fields (tempera- ture and pressure).The solute’s partial molar properties at infinite dilution may be expressed by, where Xt is the molar property of the pure solvent. For dilute solutions, which are the most important systems for applications of near-critical fluids, partial molar quantities are dominated by the properties of the highly susceptible near-critical solvent. According to the criterion of Griffiths and Wheeler, (dX?/dp),in equation 1 will diverge strongly, while the other factor will tend weakly to zero as the critical point of the solvent is approached. Consequently all infinite dilution (standard) partial molar properties of the solute will diverge strongly at the solvent’s critical point; this is a universal feature of dilute solutions in the neighbourhood of the critical point.On the other hand, the sign and amplitude of the divergence are determined by the solute- solvent molecular interactions relative to the solvent-solvent interactions as given by (ap/a~)?~. The origin of this divergence can be better understood by remembering that represents the change in X at constant (p, 7)when one solute molecule is added to the solvent. Equation 1 divides this process into two steps, which we shall illustrate for the case X = V.First an exchange of solvent by solute molecules occurs at constant V,T which will give rise to a change of pressure [term (ap/dx)ETin equation 11, in the second step the pressure is returned to the initial value.If the system is very close to the critical point, whilst the quantity (ap/ax)FTwill not be substantially different from the values it had far from the critical region, the volume change required to bring the system back to the initial pressure will be very large because (aVf/ap)Tin equation 1 will diverge near the solvent critical point according to the rules of Griffiths and Wheeler. Figure 3 depicts schemati- cally the divergence observed when X= V for a non-volatile solute; it also gives (ap/ax)zTas function of p. One of the most unusual features of near-critical solutions is the path dependence of the partial molar properties. It was shown experimentallyQ that the limiting values of the partial molar volume of SF, (solvent) is -230 cm3 mol- when its critical point is approached along the critical curve of the binary mixture with co,(solute), while the critical volume of pure SF, is 198 ~m~mol-~.Moreover, when the extrapolation to the critical point was made along the isothermal-isochoric path, it was found that V, = -40 cm3 mol- l.The path dependence of the limiting value of the solvent’s partial molar volume may be understood, within the classical description, if the Helmholtz energy of the system, A( V,TJ),is expanded in a Taylor series in ST, 6 V,and x, the differences of the variables from the values at the solvent’s critical point. Thus, O Figure 3 Near-critical behaviour of V: (full curve) and (ap/ax):y (dashed curve) as function of the fluid density.Here we have adopted an abridged notation for the derivatives of A( T,V,x),the subscripts indicate the variables of differentia- tion. From equation 2 it is obvious that Vl -+ V,, whenever the limit x-+ 0 is taken first. But, along the isothermal-isochoric path the limiting value V,(T,,,pcl,x-+O)= V,, + A$JA;V, which is different from the critical molar volume of the solvent, VC1.In general, only those paths that approach the critical point of the solvent with an xdependence weaker than x1will lead to non-anomalous limiting values of solvent’s partial properties. All other paths, including the critical line and the critical isotherm-isochore, will show different limiting values of the solvent’s partial molar properties from Xt...To illustrate this characteristic of near-critical behaviour, Figure 4 depicts the dependence of the molar volume on the solute’s concentration x at T,, and different pressures. The example corresponds to a solution of a volatile solute in a near- critical solvent. The path [(ST;6V) = 01 corresponds to the horizontal line V = V,,. For each isobar the tangent of the (V, x) curve at the point 6V = 0 is shown in the figure, the intersections of these lines with the two pure component axes give the partial molar volumes at (x,6T = 0,s V = 0). Figure 4 illustrates the fact that when x -+ 0 the partial molar volume of the solute diverges, while Vl(x)reaches a constant value that is different than the critical molar volume of the pure solvent.V Vcr Figure 4 Molar volume of a dilute binary mixture at T,,as function of the composition. The curves represent isobars. A is the value obtained for V: at the critical point following an isothermal-isochoric path. 4 The Dissolution Process The dissolution process of solids, liquids and gases may be conveniently discussed jointly if solutes in condensed phases are first considered to vaporize at the equilibrium pressure, p;. The first step is accounted for by the enhancement factor, 8,which gives the ratio of the actual solubility in the fluid to that in an ideal gas phase (ideal solubility). Thus, Perturbation models2*’ provide a very convenient physical description of the dissolution of ideal gaseous substances in fluids because they relate, in a simple way, molecular parameters to thermodynamic properties; however they do not always give the most accurate quantitative description.According to pertur- bation theory12 (cf: scheme in Figure 5) the dissolution of gaseous substances in fluids may be divided in two steps: (i) A cavity is opened in the fluid to host the solute molecule, considered at this stage to be a hard body having no attractive interactions with the solvent. (ii) The attractive solute-solvent interactions are switched on. Adding both contributions, the change in chemical potential when the solutes go from ideal gas to infinitely dilute solution may be calculated.Step (i) makes a positive contribution to the change of solute chemical potential upon dissolution, while step (ii) contributes a negative term. -I’ .. 11 Figure 5 Schematic representation of the dissolution process. Steps (1) and (ii) illustrate the perturbational model. The circles represent solute particles; they are shaded when attractive interactions are switched on. Thus the perturbation model separates the repulsive contribu- tion due to the solvent structure [expressed in terms of its packing fraction, 7, in step (i)], from that due to solute-solvent attractive interactions described in step (ii). Both contributions to dSo1Hare the product of the solvent’s expansivity and a function of 7; while d,H > 0 is a strong function of 7, d,,H < 0 depends directly on the solvent-solute interactions and less strongly on 7.It is interesting to compare the features exhibited by the dissolution process when the fluid density (i.e.7)changes from that of a liquid to that of a gas through the near-critical state. For liquid and solid solutes dSolHcontains the contribu- tion of the enthalpy of vaporization which is the dominant contribution when the solvent density is very low. 4.1 Solvents at their Triple Point Liquids typically have 7 N 0.5 close to their triple point, water being a notable exception. The perturbation model tells us that: Step (i) Local density fluctuations are very small and opening a cavity in the liquids demands the expenditure of much work. This step leads to a large positive contribution to the solute’s chemical potential (resistance to dissolution). Thus Alp > 0 and A,H > 0.Fluids like water, having a very weak depen- dence of density on temperature close to the triple point, are exceptional because then the enthalpic contribution of step (i) is very small. Step (ii) For all solutes the attractive interactions involved in this step contribute terms which facilitate the dissolution (i.e. Allpe< 0 and d,,H < 0); however their magnitude depends critically upon the intermolecular solute-solvent interactions. The triple point dissolution picture is the one that we are most familiar with. It seems natural to expect the solubility of a solid solute not having very strong interactions with the solvent, e.g.naphthalene or Cl, in CCl,, to increase with temperature CHEMICAL SOCIETY REVIEWS, 1994 because in these fluids d,,lH is dominated by the positive term A,H, to which for solids and liquids dVapH,another positive term, should be added. 4.2 Solvents with Gas-like Density These systems typically have 7 < 0.05. The perturbation model gives: (i) Since the fluid has very low density the probability of opening a cavity to locate the solute will be large, consequently this repulsive term will be very small. (ii) Although this contribution is smaller than for triple-point liquids, it is of the same nature; the number of solvent molecules surrounding the solute has decreased significantly and the average solute-solvent distance has increased. Both effects will reduce the magnitude of this term compared with that in triple-point fluids.Thus at gas-like density the dissolution of gaseous solutes and the enhancement factor for liquids and solids will be dominated by step (ii), i.e. Alp+ dllpwill decrease with increasing tempera- ture. For solid solutes, the term, dva,Hwill cause a change in the sign of the enthalpy of dissolution at low densities, generally around 7 N 0.1.13 4.3 Near-critical Solvents If the solvent were not in a region of high susceptibility, the expected behaviour would be intermediate between the two cases discussed previously (7 is close to 0.15, the critical packing fraction). The perturbation model indicates: (i) Its contribution to the chemical potential of the solute will be similar to that in a fluid far from the critical region having the same packing fraction.d,H, although positive as in the two other cases, will be dominated by the divergence of the solvent’s expansivity and will tend strongly to infinity. (ii) dnpwill be also intermediate between that for the high and that for the low density fluids, but d,,H will diverge to -co due to the dependence of the solvent density on the temperature. For near-critical fluids d,,lH = dVapH+ d,H + A,,H N A,H + d,,Hwill depend on the relative weight of the attractive and repulsive interactions: typically solids and liquids will show negative values dsolH in near critical solvents, while for gases they will be positive.In either case the absolute values of AH are anomalously large. Figure 6 depicts the change with temperature of the isobaric solubility for a solid in the three density regions. When p: is close to pCl the temperature dependence of the isobaric solubility is In x Gas-like behaviour Near-critical behaviour T Figure 6 Scheme of the temperature dependence of the isobaric solubi- lity of a solid solute throughout the three fluid regions. CHEMISTRY IN NEAR-CRITICAL FLUIDS-R. FERNANDEZ-PRINI AND M. L. JAPAS complicated because a change in temperature at constant pres- sure will strongly affect pT and the solute properties are domi- nated by the solvent density, thus the solubility will decrease with increasing temperatures for near-critical conditions imply- ing that d,,,H < 0.The change of sign of dSo1Hwith pf does not imply a change in the nature of the solvent-solute interactions with density, it is due to the strong density dependence of the enthalpy of cavity formation. On the other hand, for a solid solute an increase of temperature at very low density must lead to an increase of solubility because the solute's vapour pressure increases. According to equation 1, the thermodynamic description of near-critical behaviour in dilute solutions may be conveniently formulated in terms of a contribution related to the solvent's susceptibility which tends to diverge, and contributions arising in the specific solute-solvent interactions. The latter are gov- erned by (ap/ax);, a quantity whose sign reflects the volatility of the solute compared to that of the solvent and does not diverge; moreover this important quantity is directly related to VF and SF; thus equation 1 gives for X = A, which by proper differentiation yields, where a;1 is the isobaric expansivity of the solvent.Equation 5 shows that it is important to measure Vy since it is only related to the two factors which model the behaviour of dilute near- critical solutions. According to equations 4-6 thermodynamics leads naturally to a separation of the diverging factor from the well-behaved factors which contribute to the properties. Figure 7 illustrates the effect of intermolecular parameters upon Vy/ V, at different densities calculated with the Percus-Yevick approximation.l4 It may be observed that the sign of the partial molar volume of the solute changes with solvent density, more- over at low density VT becomes more negative for bigger solutes, as discussed below (cf equation 10) this is a typical gas- like feature. 0 * 2 8%LL \ -50 I I I 2 4 Figure 7 (V,"/V3against the ratio of solute to solvent intermolecular energies (e2Jc1 (p:/p, 2.4, full curves; 1.05, dashed curves. (o~z/all):1.5,0;1.0,Ai0.5, 0. 4.4 Microscopic Interpretation An imaginative, albeit somewhat misleading terminology has been coined to describe the not-to-be-expectedbehaviour of solutions, including near-critical behaviour, e.g. hydrophobic hydration, frustration effect, charisma, critical clustering, and critical local-density enhancement.This terminology frequently reflects our lack of a thorough understanding of the phenomena. In particular, the unusual behaviour of near-critical systems have stimulated speculation about the meaning of partial molar properties diverging to plus or minus infinity; a dramatic change of solvent density around the solute molecules has been pro- posed to explain it. In order to inquire into the structure of the solutions it is important to use the pair distribution function go{r;T,p), a quantity which is central to modern theories of liquids.lS giJ{r;T,p) is proportional to the probability of finding a particle i at a distance rfrom particlej. As indicated in Figure 8 it contains the short-range structural information typical of a liquid.This figure also shows gj,{r;T,p) for a gas and for a crystalline solid. r/a Figure 8 Pair distribution functions for gases (a), liquids (b), and solids (c>. The gas has no structure whatsoever, while the solid shows a structure extending throughout the crystal. The solute-solvent pair distribution function may be divided into two terms. The first factor is related to the interaction of the two particles in V~CUOwhile y12, known as the cavity distribution function, reflects all indirect interactions between particles 1 and 2, and will be strongly dependent on fluid density. When the fluid density goes to zero, (cf. Figure 8a) and the cavity distribution function becomes unity, the whole process of dissolution from the gas phase will then be dominated by the attractive solute-solvent interaction energy. For dilute gaseous mixtures this is conveniently reflected by the cross second virial coefficient, B12,which is related to the intermolecular solvent-solute interaction energy by, Hence B, is related to the pair correlation function in the limit of vanishing total density where indirect correlations between molecules 1 and 2 are negligible.Second virial coefficients may be expressed in terms of the reduced temperature T, = kT/clZby,16 1 60 CHEMICAL SOCIETY REVIEWS, 1994 For slightly volatile solids Bf2 will be negative because of the strong solute intermolecular energy,i.e.low TR.Since according to equation 10, B,, is proportional to the cube of the solute- solvent contact length, uI2, it will be more negative for larger solutes. This gas-like feature is already observed at near-critical densities, as illustrated by Figure 7. Pair distribution functions facilitate distinguishing between long-range effects, which cause the enhanced susceptibility in the near-critical fluid, and short-range effects shaped by the inter- molecular interactions plus the decreasing fluid density. Many thermodynamic properties are related to integrals which contain the pair correlation function,l7 so they will tend to diverge when gJr) has a long-range tail. This situation has frequently been interpreted as a very large excess or defect of solvent molecules around the solute particle, but really there is no evidence of critical clustering around the solute.Figure 9 shows the total correlation function h,,{r) = (g,(r) -1) for a model Lennard- Jones mixture according to the Percus-Yevick approximation. 1 2 3 4 Figure 9 Total correlation functions against the distance from a solute molecule. Lower curves, i = 1; higher curves, i = 2. The inset gives the quantity I(r) = r2h12(r). (p*/PCl) is 0: -, 0.33: -.-.--, 1.02: ----, 1.33:- - - - -. c2Z/elI=2.0;uz2/u11= 1.5; kT/cll= 1.36; TIT,, = 1.031 It is evident that the number of first neighbours of the solute is not too much affected by the vicinity of the critical point, its long range tailing-off being its most notable feature and the reason for the divergence of the integral of h,{r) taken over all the system.For the case of gases, i.e. (p: +0),h,,-(r)has a single peak and is short-ranged, according to equation 9 its integral is equal to -2B,. As the density increases, indirect interactions become more important, giving rise to an incipient increase of h,jr) beyond the first peak, as shown in Figure 9. The fact that the first peak of h22(r)is higher than that of h12(r)is due to the stronger solute-solute interactions. The relative excess fraction of particles i surrounding a central j molec~le,fe~,~is a relevant quantity to establish the existence of critical clustering, it is defined by, 0.0 -r----l 0.0 -0.0 0.5 1.0 1.5 PI'% Figure 10 fii as function of the reduced density. Full symbols, first neighbours; open symbols, second neighbours.Figure 10 shows the effect of solvent density uponfy for two different choices of R,,,: (a) corresponding to first neighbours (I?,,, = u,/2 + uJ, and (b) also including neighbours separated by a solvent molecule. As illustrated in Figure IO,f',' passes through a sharper maximum for the (2-2) case than for (1-2); the (1-2) interaction does not even show a maximum for the smallest value of R,,,. For the (2-2) case, the maximum occurs at a density lower than the critical density, reinforcing the conclusion that the short-range microstructure is not coupled with the long- range critical fluctuations. The maxima in the curves are due to indirect interactions, hence an incipient manifestation of a characteristic feature of dense fluids.Since the solute or co-solvent+ molecules have E,~> c12, at low density fi; as the solvent density increases, its molecules will occupy positions close to the central solute. The solvent moiety surrounding the central solute parti- cle provides a region where more solute or cosolvent may be preferentially attached. This incipient microheterogeneity can explain the role of co-solvents in enhancing the solubility of solutes. 5 Near-critical Behaviour of Dilute Solutions From the discussion in the preceding sections, a general picture of the critical state of dilute solutions emerges. Anomalies in the susceptibility of the pure solvent, (e.g.~f~ or C;,)driven by long-ranged, long-lived fluctuations which reflect the solvent's critical state, produce diverging partial molar properties of dilute solutes. A short-ranged microstructure, which depends upon the solute-solvent interactions is superimposed on the long-range fluctuations. Statistical thermodynamic models, which were very successful in describing the properties of liquids and gases with the proper physical insight, have difficulties in including the effect of long- range fluctuations for the correct description of the thermodyna- mic properties of near-critical systems. However, for the basic understanding of the processes which occur and for the design and control of several technological applications, a correct description of supercritical fluids is of fundamental importance. 5.1 Asymptotic Behaviour Most of the processes involving supercritical fluids occur in dilute solutions subjected to phase and/or chemical equilibria: + The co-solvent is a substance added to increase the solubility of the solutes at concentrations which usually are close to 1% CHEMISTRY IN NEAR-CRITICAL FLUIDS-R.FERNANDEZ-PRINI AND M. L. JAPAS extraction, chromatography and chemical reactions in supercri- tical fluids. Equation 4 shows that the solute and the solvent chemical potentials are related through the derivative of the Helmholtz energy with respect to composition, A,. The solute's chemical potential may be expressed by, p,(T,p,x) = p:(T,p) + RTlnxy, = &(T) + RTlne P" (12) where G2and y2 are the solute's fugacity and activity coefficients respectively.The solute's standard state may be chosen either as the ideal gas, p;( T), or as the infinitely dilute (Henry) solution, p; ( T,PI. Since near-critical systems show unusual behaviour it is extremely helpful to have exact asymptotic relationships as guides for the description of the properties in the neighbourhood of the critical point. For this purpose A, in equation 4 is expanded in a Taylor series around the solvent critical point (classical description). O Thus pF( T,p)is given by, Even in the critical region pt is a well-behaved function, hence when the path is isothermal or corresponds to the coexistence of two phases, the leading term in the expansion is the 6 Y term and an asymptotic linear dependence of p; on solvent volume or density results, The validity of this expression for the description of the liquid- vapour phase equilibrium in dilute solution has been tested with the solute partition constant, Kz, defined by, where y is the solute mole fraction in the vapour phase.The condition for liquid-vapour equilibrium p2(1)= p2 (v), deter- mines that RTln K: = pT(1) -pT(v) which, combined with equation 14, yields RTlnK: = -*[pf(l) -p:(v)]Pc 1 or the asymptotically equivalent expression, RTln K," = -?!k[p:(l) -pel] Pfl (15) For several binary systems, equation 16 was shown to apply over an unexpectedly wide temperature range. Figure 11 shows that the linear relationship between TlnKg and [pf(l) -pel] holds over 150 K for some aqueous systems.The range of validity of the linear equation 16 will depend on the magnitude of the solvent-solute interactions, i.e. on A:, = -(C?~/C?X);~; the approximation of using only the first term in expansion 13 will be more valid when the pressure change with composition is bigger. It should be noted that equation 14 correctly predicts that the standard partial molar volume and entropy of the solute (equa- tions 5 and 6) will diverge at the critical point of the solvent similar to the derivatives of pf with pressure and temperature. The sign of the divergence will be that of -A:,. Equation 14 can also be used to describe the asymptotic solubility of solids in a near-critical solvent.In that case, phase equilibrium implies that, PHS) = P2W) 0 10 20 30 Figure 11 T In K," against the (pl(f) -pel) for the partition of N, between liquid and vapour HzO. Experimental points from ref. 19. and p; + RTlnf!) +I" V:(s)dp = p:" -*Cp(f) -p,,] + RTln(x7,)P: Pc1 wheref: is the fugacity and V:(s) the molar volume of the pure solid. For dilute solutions (sparingly soluble solids) y2(x -+ 0) = 1 and the last term in the left-hand side member may be neglected, and the solubility can therefore be expressed as RTlnx = (& -p;") + RTln 1+ !$b(l) -pel] Pel where byflT) we denote a well-behaved and slowly varying function of the temperature. On the other hand, from equation 3 we have, RTlnd = (p; -p;") f *b(f) -pel]+ RTln PCZ Once more the second term dominates and for non-volatile solid solutes in common solvents A:v > 0, hence the solubility and d increase with increasing solvent density.This is illustrated in Figure 12 for the system Xe-I,.20 Solutes with large ,4iVappear to be more soluble, but the contribution of its vapour pressure, given in equation 17 byf:, may dominate the magnitude of the relative solubility of different solutes. This is not the case for d which will always increase with A&. Equations 17 and 18 show that in a near-critical solvent the solubility and the enhancement factor are well-behaved functions of the solvent's density albeit strongly affected by changes in pressure and temperature.In other words, because supercritical fluids are very compressible, small changes in pressure result in large variations of their solvent power. A relatively high dissolution capacity and strong response to external perturbations are two important features of solutes in supercritical fluids. Figure 12 In 8for I2 against the reduced density of xenon. Full curve, isothermal run (TIT,,) = 1.031; dashed curve, isobaric run (PIPCl) = 1.23. 6 Chemical Reactions in Near-critical Fluids Critical phenomena, characterized by a high susceptibility due to long-lived, long-ranged fluctuations, can strongly influence chemical kinetics and equilibrium. The effect on the kinetics of reaction could be expected to be twofold. Critical phenomena are characterized by a slowing down of most transport proper- ties,, 1,22 hence reaction rates may drop to zero at a critical point.On the other hand, near-critical dilute solutions exhibit anoma- lies in the temperature and pressure dependence of the reaction rate constant k which are related to the enthalpy and volume of activation respectively. The first effect21 would be related to the fact that near a critical point the rate of equalization of the density (concent- ration) gradients by diffusion is markedly reduced. Since the driving force for diffusion, (dplax)),, is very small (it actually equals zero at the critical point) the system becomes indzflerentto changes in density (concentration) near a critical point.,, When the reaction occurs at constant fields this would reduce the rate of reaction, which is proportional to the restoring force (ad/ag)defined in terms of the affinity at’= -vlpl.The second effect has been observed in many dilute near- critical systems.Examples are the decomposition of a-chloro- benzyl methyl ether in 1-1 difl~oroethane,~where, even 16 K away from the solvent critical point, the rate constant increased 20 times when the pressure changed from 4.5 to 6.9 MPa, implying an activation volume of -6.0 dm3mol- l; or the spin- exchange reaction between very dilute nitroxide free radicals in C2H, close to the diffusion limit, where the rate constant decreases by a factor of 4to 5 when pressure increases from 4to 5 MPa (activation volumes of 7.5 dm3mol-1).24 The effect of criticality on chemical equilibrium is better understood.For example, the chemical equilibrium condition when p changes along an isothermal path is, where, together with T, other variables denoted by Y may be kept constant. The pressure derivative of the equilibrium extent of reaction is then, Equivalently, for the isobaric path There are two possible reasons for (dg/ap)& (or (d(/dT)i+) to diverge: cvlV,= d,V(~v,S,= d,S) may diverge. If the reactants and the products are at infinite dilution in a near-critical, almost pure, solvent, V,and S,will diverge to plus or minus infinity (depending on the interaction with the solvent), making d,Vand d,S diverge. This effect has been reported extensi- vely.As an example, the equilibrium constant for the tautomerization of 2-hydroxypyridine into 2-pyridone in 1,l -difluoroethane at 6T = 16K increased by a factor 2 due to a pressure change from 4.5to 5.9 MPa (volume change of reaction -1.4dm3mol-1).2* If the change in pressure occurs while none of the other Y variables that are kept constant are densities, then (ad/tends to zero while the critical conditions are approached. In the case of binary mixtures in a single phase having 3 degrees of freedom with frozen chemical equili- brium or 2 degrees of freedom if equilibrium is allowed, only one variable needs to be fixed, the temperature. In that case, divergence of the isothermal pressure dependence of the extent of reaction should be observed only under conditions where criticality can be encountered for this system, i.e.at the critical point.For systems having three or more components, additional variables need to be fixed; CHEMICAL SOCIETY REVIEWS. 1994 only those paths corresponding to constant activities (or chemical potentials) are expected to exhibit strong anoma- lies. To date there is no reliable evidence of strong anoma- lies in multicomponent systems. Weak anomalies have been reported, e.g. for the degree of dimerization of NO, near the liquid-liquid critical point of the solvent system per- fluoromethylhexane-carbon tetrachloride, the observed effect being very small (about 4%). 7 Some Applications of Near-critical Fluids The peculiar combination of equilibrium and transport proper- ties which has been described, constitutes the reason for the successful applicaton of supercritical fluids in many processes.7.1 Supercritical Fluid Chromatography (SFC)27 Gas chromatography is not well-suited for the separation of thermally unstable compounds or of non-volatile solids which have a very low concentration in the gas phase. On the other hand, high-performance liquid chromatography has the disad- vantage of a poorer resolution because the higher liquid visco- sity restricts the length of the columns and the low diffusivity implies wider chromatographic peaks. In SFC, the solute parti- tions between a mobile supercritical fluid phase and a stationary phase. The distribution coefficient, and therefore the retention time, will depend smoothly on the density of the supercritical fluid and the separation of solutes can be performed by controll- ing this variable easily.Other chemical parameters, like the presence of co-solvents, may be manipulated to improve the separation performance in SFC. 7.2 Supercritical Fluid Extraction (SFE)** This is another technological application that takes advantage of the strong Tandp dependence of the solubility in near-critical solvents. In this process the solvent fluid changes its density and consequently its capacity for dissolution. The adjustment of the thermodynamic variables (temperature, pressure, or co-solvent concentration) can result in changes of orders of magnitude in the solubility as the density changes between the stages of dissolution and that of separation, thus improving extraction yields.Reprecipitation of extracted products can be achieved easily and with fine control of the variables to facilitate the separation steps. The two best known examples of technological application of SFE, are the extraction of caffeine from green coffee beans and of hops in the beer industry. Also becoming increasingly import- ant are the extraction of essential oils (terpenes among the more soluble), natural flavours, and fragancies. These are mostly em- ployed by the food and pharmaceutical industries. It is also used to eliminate impurities and regenerate activated charcoal filters. In most of these processes of SFE the solvent employed is CO, since it is neither toxic for industrial manipulation nor contains hazardous impurities which might be incorporated into the purified products. Moreover CO, is environmentally safe.Its critical temperature is close to room temperature (304.2K), an important advantage for processes applied to thermolabile substances, it is inexpensive and non-flammable. The use of co- solvents, can greatly improve the selectivity and efficiency of the SFE process. 7.3 The Synthesis of Fine and Ultrafine Powders This process takes advantage of the very large difference in solvent power of near-critical fluids produced by moderate changes in its temperature and pressure. It is similar to SFE, but the interest is now centered in the reprecipitation stage: a supkrcritical solution is suddenly expanded as it flows through a nozzle and during its expansion the fluid may attain supersonic velocities.The abrupt reduction of the medium’s solvent capa- city leads to the formation of very small monodisperse particles CHEMISTRY IN NEAR-CRITICAL FLUIDS-R FERNANDEZ-PRINI AND M L JAPAS of solid solute A good control of the pre- and post-expansion variables (pressure and temperature) and of the solute concent- ration determine the size and morphology of the crystalline or amphous solids that precipitate The technological appli- cation~~~of this new process are oriented towards very specific areas, like biotechnology (production of bioerodible polymeric microspheres and microparticles for controlled drug release) or materials science (production of ultrafine ceramic precursor powders, pure or intimately mixed) 7.4 Use of Near-critical Fluids as Media for Chemical Reactions Diffusion-controlled reactions exhibit reaction rates several orders of magnitude higher than in normal liquids because in supercritical fluids typical viscosities are around 0 01 mPa s, I e one hundred times smaller than that of typical liquids The advantage of supercritical fluids arises from the combination of liquid-like solvent capacity (higher concentration of reactants) and gas-like transport properties (higher rate constants) More- over rate constants may be strongly modified by moderate changes of the state variables Not only the kinetics of the reactions can be precisely controlled in supercritical fluids, equilibrium properties aIso are strongly affected by the solvent conditions For reactions having more than one possible reaction path involving different pro- ducts, the change In the solvent properties will affect the type of intermediate species which are formed by reaction, thus modify- ing the yield of the various possible products This feature is dramatically enhanced when the supercritical fluid solvent is water 30 as solvent density is increased, the increasing dielectric constant preferentially stabilizes ionic intermediate species over radicals, with the concomitant change in the reaction mecha- nism Examples can be found in the pyrolysis of coal model compounds and in the processes of catalytic and non-catalytic oxidation and dehydration in supercritical water Other technologically interesting processes of chemical reac- tions in supercritical fluids include enzymatic reactions (like the oxidation of steroids, virtually insoluble in liquid water, in supercritical CO,), oxidation of hazardous materials in super- critical water, and catalysed heterogenous reactions, where the catalyst deactivation can be avoided by the interplay of solva- tion power and mass-transfer properties 8 References 1 J S Rowlinson, ‘J D van der Waals On the continuity of the Gaseous and Liquid States’, 1988, North Holland 2 D Henderson and J A Barker, Rev Mod Phys ,1976,48,587 3 K E Gubbins, K S Shing, and W B Street, J Phys Chem ,1983, 87,4573 4 H E Stanley, in ‘Introduction to Phase Transition and Critical Phenomena’, Oxford University Press, Oxford, 197 1 5 J W Tom and P G Debenedetti, Ind Eng Chem Res , 1993,32, 21 18 6 I C Sanchez, Macromolecules, 199 1 ,24,908 7 K G Wilson, Rev Mod Phys , 1983,55,583 8 R B Griffiths and J C Wheeler, Phys Rev ,1970,2, 1047 9 A M Rozen, Russ J Phys Chem , 1976,50,837 10 J M H Levelt Sengers, in ‘Supercritical Fluid Technology Reviews in Modern Theory and Applications’, ed J Bruno and J F Ely, CRC Press, Boca Raton, 1991 11 L R Pratt and D Chandler, J Chem Phys , 1977,67,3683 12 R Fernandez-Prim, H R Corti, and M L Japas, in ‘High Tempera- ture Aqueous Solutions Thermodynamic Properties’, CRC Press, Boca Raton, 1992 13 E Marceca and R Fernandez-Prim, J Chem Thermodyn ,1993,25, 237, 1994,26, in press 14 R Fernandez-Prini and M L Japas, J Phys Chem ,1992,96,5115 15 J P Hansen and I R McDonald, in ‘Theory of Simple Liquids’, Academic Press, New York, 1976 16 J 0 Hirschfelder, C F Curtiss and R B Bird, in ‘Theory of Gases and Liquids’, Wiley, New York, 1966 17 J G Kirkwood and F Buff, J Chem Phys , 1951,19,744 18 C Carlier and T W Randolph, AIChE J ,1993,39,876 19 J Alvarez and R FernandezIPnni, Fluid Phase Eq , 199 1,66,609 20 D Fernandez and R Fernandez-Prini, J Chem Thermodyn , 1992, 24,277 21 I Procaccia and M Gitterman, Phys Rev, 1982, A25, 1137, S C Greer, Int J Thermophys ,1988,9,761 22 J M H Levelt Sengers, U K Deiters, U Klask, P Swidersky, and G M Schneider, Int J Thermophys , 1993,14,893 23 K P Johnston and C Haynes, AIChE J, 1987,33,2017 24 T W Randolph and C Carlier, J Phys Chem ,1992,96, 5146 25 D G Peck,A J Mehta,andK P Johnston, J Phys Chem , 1989, 93,4239 26 J L Tveekrem, R H Cohn, and S G Greer, J Chem Phys , 1987, 86,3602 27 M D Palmien, J Chem Educ, 1988,65, A254 28 S Peter, Ber Bunsenges Phys Chem , 1984,88,875 29 P G Debenedetti,J W Tom,X Kwauk,andS D Yeo,FluldPhase Eq ,1993,83,311 30 R Narayan andM J Antal, J Am Chem SOC, 1990,112, 1927

ISSN:0306-0012    

DOI:10.1039/CS9942300155

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Electrophoretic NMR Manfred Holz lnstitut fur Physikalische Chemie, Universitat Karlsruhe, Kaiserstrasse 12, D-76128 Karlsruhe, Germany 1 Introduction For more than thirty years nuclear magnetic resonance (NMR) techniques have proved to be among the most sucessful tools in the study of local structures and microdynamics in electrolyte solutions. 1,2 Moreover, NMR spin-echo experiments with pulsed magnetic field gradients (PMFG) have allowed the measurement of an important transport quantity, the translatio- nal diffusion coefficient D, for solvent molecules as well as for ionic species. The outstanding advantage of the spectroscopic NMR methods lies in their selectivity, allowing the observation of distinct species in complex systems by using selected nuclei of different elements or signals of the same nucleus but with different chemical shifts, corresponding to distinct molecules or molecular groups.Since electrolyte solutions contain at least three different constituents and are thus typical multi-compo- nent systems, it is easy to see why NMR techniques can contribute so much to our understanding of the microscopic properties of these solutions. Bearing in mind the numerous successful applications of NMR to various problems of the physical chemistry of electro- lyte solutions, it is remarkable that until the early 1980s practi- cally all these NMR experiments had been performed on electrolyte solutions at equilibrium, 1,2 which means in the absence of an external electric field and thus in the absence of an electric current.This is astonishing because electric conductivity is the most important and the characteristic property of electro- lyte solutions. Moreover, from a theoretical point of view, it is easy to see that in the study of mixtures of macromolecules, e.g. in the field of molecular biology, electrophoretic experiments combined with NMR could offer very promising new possibilities. The reason why in the literature up to 1980 only two NMR experiments on electrolyte solutions in the conducting state were des~ribed~?~lies in the fact that serious experimental difficulties are to be expected when an electric current flows in the solution during an NMR experiment, in which the sample is exposed to a strong magnetic Zeeman field Bo.Clearly, there are two main problems, namely undesired additional magnetic fields and heat Manfred HoIz was born in LudwigshafenlRhein, Germany in 1938. He attended the University of Karlsruhe, where he obtained his ‘Diplom’ in physics in 1966 under the supervision of G. Laukien. From 1966 to 1970 he worked for the Bruker company, where he was member of the team that developed the first commercial pulsed NMR spec- trometers. He then moved back to the University of Karlsruhe to carry out doctoral research under Professor H. G. Hertz. He is now lecturer and ‘Aka- demischer Direktor’ at the In-stitute of Physical Chemistry. His research interests are in the development and application of NMR techniques, such as inter- molecular relaxation, NMR difusion measurements, and electrophoretic NMR, to the study of liquid systems.165 generation in the sample cell, both arising from the electric current. Further difficulties could be expected to arise from electro-osmosis, from the fact that electrodes have to be placed in direct neighbourhood of the NMR receiver coil, and from mechanical disturbances connected with switching of electric direct currents in strong magnetic fields. A first attempt to observe the migration of ions in an external electric field by NMR was undertaken in 1972 by Packer and co- worker~.~However, the above mentioned problems produced unsuccessful experiments. In 1980 the present author and his co- workers examined the feasibility of conducting NMR experi- ments in the presence of an electric current.First we studied5 the influence of internal field gradients, caused by thecurrent, on the NMR signal. Encouraged by the results we systematically investigated in the following years the applicability of certain NMR methods to measurements on ionic solutions in the conducting state.6-10 The general goal of these investigations was the introduction of spectroscopic NMR techniques into the important area of electrochemistry and into its fields of appli- cation. Our first and most promising special aim was the measurement of mobilities u& of charged species in liquid solutions by applying known NMR flow measuring techniques. We also intended to examine whether and how magnetic relaxa- tion of nuclei in ions, moving in an external electric field, can be influenced by the migration process.In 1982 in our laboratory the first successful mobility measurements by ‘H NMR were performed on (C,H,),N+ ions in ~ater,~,~ demonstrating that such measurements were indeed possible and that the data obtained agreed with results from classical methods. After further technical and methodical deve- lopment~,~,~we showed that ’Li NMR can be used to observe Li+ mobilities9 Finally in 1987 we succeeded in demonstrating that the quadrupolar relaxation of ionic nuclei, e.g..23Na+ and 87Rb+, can be influenced if the ion moves in an external electric field.1° In the following years two further groups entered this new field of NMR application. Johnson and co-workersll conducted mobility measurements for the first time under high- resolution ‘H NMR conditions and they introduced the expres- sion ‘electrophoretic NMR’ (ENMR) for the method.Their group went on to develop interesting experimental and methodi- cal improvements, e.g. a two-dimensional (2D) representation of the ENMR results.12.13 In 1989 Strange and co-worker~’~ finally demonstrated that the new technique can be used success- fully for electrophoretic mobility studies on surfactant systems. Meanwhile the potential of NMR studies in the presence of an electric direct current (DCNMR) had been impressively demon- strated and a new and promising field of NMR thus opened. The basic technology for DCNMR has now been developed, but is not yet commercially available.It should be possible for the instrumental technique to be markedly improved in the future since a number of difficulties still remain. Thus, DCNMR may be regarded as still in its infancy and electrophoretic NMR experiments, as the main application of DCNMR, do not yet have a routine character. It is the purpose of the present review to give a brief introduction to the principles and methods of electrophoretic NMR, to discuss in general the practical problems of DCNMR (and the solutions found so far), and to give an overview on measurements, applications, and possible future developments of this new class of NMR experiments. It should be pointed out that in this article the term ‘electrophore- sis’ is used in a general sense, namely as the term for the migration of all kinds of particles in an electric field.2 Principles of DCNMR Experiments 2.1 Measurement of Ionic Drift Velocities by NMR In an electric field E between two electrodes in an electrolyte solution, ions move with a drift velocity v+ along the direction of the electric field. In electric fields of 10 to 100 V cm-these drift velocities are in the range of 10-4-10-2 cm s-l. That means we are dealing with very slow motions, which can be considered as a field-assisted diffusion process. The drift velocity v+ is related to the ionic mobility p+ by: v* =p*.E (1) This means that a knowledge of E and the measurement of vi allows the determination of the electrophoretic mobility p+.If we further measure the electric current I flowing in the electro- phoresis cell during the experiment, we obtain an important single ion transport quantity, namely the transference number Tiof the ion of interest in strong electrolytes: cAFT, =--’v* I where c is the salt concentration in moles per litre, F is the Faraday constant, and A is the area of cross-section of the cell.The drift velocity of single charged constituents in an electrolyte solution delivers important characteristic quantities of the system of interest. In the early years of NMR it had been discovered that a translational displacement of particles carrying a nucleus with spin Ican be detected by the NMR spin-echo technique, when a linear magnetic field gradient G is applied.1s The spin-echo is observed at the time 27 after a T/~-T-Tradiofrequency (r.f.) pulse pair.Since the NMR precession frequency w,, is strictly field-dependent, according to w0 =yB, (y =gyromagnetic ratio; B, =magnetic Zeeman field), a translational motion of the spin-carrying particle along an imposed field gradient is connected with a change of the NMR frequency of the nucleus. The displacement of the nucleus in the direction of G during the time 27 results in a loss of nuclear spin phase coherence, compared with a non-moving nucleus, and in a change of the echo amplitude at 2T compared with a spin-echo experiment without an imposed field gradient. The physical origin of the displacement may be different.Incoherent translational motions, like translational diffusion, produce a random dephas- ing of the spin precession during 27,resulting in a damping of the echo amplitude M(2T).’ Hence, to the exponential decay of the transverse nuclear magnetization by spin-spin relaxation a second term due to the diffusion is added and we obtain: hf(2~)=M(0)eXp[-2.r/T2-bZGZD~3] (3) with T, =spin-spin relaxation time, D =self-diffusion coefficient. Coherent motion along the field gradient, e.g.due to plug flow or due to a constant drift velocity, produces for all spins during 2T the same frequency change, that is the same phase shift with respect to a reference frequency. Consequently, a constant drift velocity causes no dephasing but aphase shift of the echo signal, and this phase shift can be easily derived.Assume we apply a magnetic field gradient along the z-direction, that is G =aBZ/az. If the nucleus e.g. in an ion moves with constant velocity v along the z-direction, if the electric field E is applied in this direction, the nucleus experiences a time- dependent magnetic field B(z(t)), since its position z is a function of time; namely ~(t) =v.t. From: B(z(t))=Bo.z(t)G (4) it follows: w(t) =yBo +yGvt =wo +yGvt (5) The frequency difference h(t) with respect to w0 is: CHEMICAL, SOCIETY REVIEWS, 1994 This frequency difference leads to an accumulated phase differ- ence A+21 at the time 2T, when the echo maximum appears:” =~GvT’ (6) It is in principle this phase shift which has to be measured in order to determine the velocity v by NMR.The shift causes a cosine modulation of the echo amplitude at 2T when G or v are changed. Therefore, if both diffusion and a drift velocity control the translational motion, the transversal magnetization M,,(21) is given by: hf,(2T) =hf(0)cos(yGvT2)exp[-21,T2 -~Y~G~DT~](7) In modern NMR diffusion and flow studies, pulsed magnetic field gradients (PMFG) are used instead of a steady gradient. as shown in Figure 1 and as introduced and analysed by Stejskal and Tanner.’* The two gradient pulses with a duration 6, an amplitude G, and a distance A are mainly applied for practical experimental reasons. In such a PMFG experiment, equation 7 is slightly modified to be~ome:~ d2 x I PMFG I PMFG 1... .......... .................IE I DAT Figure 1 Pulse sequence of the standard pulsed magnetic field gradient (PMFG) spin+cho experiment for the measurement of drift velocities of charged species. n/2and n denote the corresponding rf pulses; G is the strength of the magnetic field gradient; E is the strength of the pulsed electric field, and DAT is the data acquisition trigger. In practical electrophoretic NMR experiments one can deter- mine the ratio hf,(2T)/hf(2T) =COS(~GV8d)=COS(A$Z,) (8) by measuring My(&) in presence of an electric field E (v+ # 0) and measuring M(2T) in absence of E (v+ =0). For reasons of measuring accuracy the amplitude ratio, or directly A+27 in equation 8, is commonly determined as a function of one of the quantities 6,G, or v where v is proportional to the applied electric field E and to the current Iin the sample.2.2 Measurement of Relaxation Times in the Presence of an Electric Current For electrolyte solutions, the case where magnetic relaxation of nuclei residing in ionic species, which move in an external electric field, is changed, has been theoretically analysed for the first time by Atkins and Clugston. l9 These authors considered dipole- dipole relaxing nuclei such as ’H and 19Fand a counterion carrying an electron spin. Their calculations revealed that only extremely high electric fields (2lo7 Vm-l) would produce remarkable changes in the nuclear magnetic relaxation times caused by the change in the dynamics of the interacting partners.The present author pointed out that the quadrupole relaxation of ionic nuclei might be the more interesting relaxation mechan- ism2S1O for investigations in conducting electrolytes, where at ELECTROPHORETIC NMR-M. HOLZ lower electric fields structural effects could be expected, which could alter the relaxation times. The experimental method is not of a special nature. The measurement of NMR relaxation times is performed with the known pulse sequence e.g. for T, by the inversion-recovery (T-T-712) method. The only difference from a common NMR relaxation experiment is that during the time T, in which the relaxation process is going on, an electrical field is applied between the two electrodes and thus a current flows in the electrolyte solution (Figure 2).Consequently, these relaxation measurements can be made with any probe which is suited for DCNMR experiments. E ...................................... IDAT Figure 2 Pulse sequence for a DCNMR T, relaxation measurement by the inversion-recovery method. The pulses are denoted as in Figure 1. 3 DCNMR: Technical Requirements, Problems, and their Solutions We saw that NMR in the presence of an electric current in principle only requires the application of an electric field between two electrodes within the sample cell. Thus for DCNMR experiments a constant current power supply (e.g.0-500 mA, 0-1000 V) is needed. Since for practical reasons the electric field is only applied for certain time intervals (see Figures 1 and 2) a gateable supply is used.Gate pulses for the power supply are usually derived from the computer system of the NMR spectrometer, and they are then part of a synchronized pulse sequence consisting of r.f.-pulses, magnetic field gradient pulses, electric field pulses, and data acquisition pulses. For electrophoretic measurements, where the PMFG-technique is applied, it is clear that the NMR probe must be equipped with a field gradient coil system. Modern shielded gradient coil systemsz0 are recommended because they avoid problems with magnetic field disturbances due to eddy currents following the gradient pulses, allowing for example short gradient pulse distances A and pulses with high gradient strength G to be applied.These two conditions are particularly important if electrophoretic measurements are performed on systems with small T,-values. The electrodes used are in most cases made of platinum wires or sheets. If solutions of halides are to be investigated Ag/AgX (X = C1, Br, I) electrodes are advantageous. More technical and methodical details may be found in the original literature5-I4 and in a comprehensive review on electrophoretic NMR which appeared some years ago.21 The special problems connected with the presence of an electric current in the sample during the NMR experiments, and the technical solutions of these problems, are now briefly discussed. 3.1 Magnetic Fields Induced by the Current From classical physics we know that an electric current produces a magnetic field in the plane perpendicular to its direction of flow.Hence, in DCNMR experiments also, additional magnetic field gradients are produced in the sample which can interfere with the Zeeman field B,. However, as our investigations sh0wed,~~~~9the expected difficulties can be overcome in a 167 relatively easy way. The simplest solution of the problem is to choose as the current direction the direction of the magnetic main field Bo (2-direction). Consequently, the magnetic field lines produced by the current lie in the x,y-plane. There are no field gradients in the z-direction, which is the relevant direction in NMR, and therefore no signal disturbances can occur.In modern superconducting magnets the probe and the sample tube axis lie in this z-direction and therefore it is in practice very simple to achieve a parallel alignment of current and magnetic field. IG ,rf-coil I-(a) (b) (a Figure 3 Three basic DCNMR cell geometries, (a), (b), (c), with the corresponding electrode arrangements. The rf coil is a saddle coil as normally used in superconducting magnets. The direction of the Zeeman field B, and of the imposed field gradient G = 8B:jaz is also shown. In Figure 3 three general electrophoretic cell geometries are shown together with the corresponding electrode positions and r.f. saddle coil arrangements. In all these cells the current Iflows parallel to B,.However, we recognize that there are important differences. In the two U-tube-type cells in Figure 3(a and b) the current Iflows parallel to B, (and in the direction of the applied gradient G) within the r.f. coil region, but in the range of the knee of the U-tube the direction of I is perpendicular to B,. Thus the disadvantage of cells (a) and (b) lies in the fact that z-field gradients can act within the coil volume, where the NMR signal is generated. This undesired effect may be reduced if the knee of the U-tube cell is not too near to the r.f. coil. The advantage of cells (a) and (b) is that both electrodes are in the upper part of the cell and are vented to the atmosphere, a fact which is important if the electrodes are gassing and bubbles occur.In order to minimize the induced field gradients, the best solution is the cylindrical cell (c), but gassing of the lower electrode has to be avoided. Finally, one has to keep in mind that the electrical connections to the electrodes should be kept, as far as possible, parallel to the B, direction. It should be pointed out that the cell shown in Figure 3b is an example of the counterflow technique, previously developed by my group,' where within the measuring volume, the sample coil, the current flows in both directions. A disadvantage of the counterflow arrangement is the loss of information with respect to the sign of the mobilities. Ifelectromagnets must be used or if from a theoretical point of viewlg an ionic flow perpendicular to the B, field is to be observed, internal field gradients in z-direction will be induced.However, this problem can be solved. As shown in a previous work these induced gradients are linear5 and, as demonstrated experimentally,s the current through the electrophoretic cell can be led through shim coils outside the sample in such a way that a self-compensating system is achieved, where the external gradi- ents from the shims compensate the internal gradients. In summary, the problem with the additional magnetic field gradients in a DCNMR experiment is practically solved and today is of minor importance. However, in a general context it should be pointed out that field gradients produced by a current in the sample may be utilized, for example if current distribu- tions in complex conductors are the subject of intere~t.~~~ Meanwhile, NMR imaging experiments with the aim of map- ping electrical circuits have been successfully commenced.22 3.2 Problems Caused by Resistive Heating, Mechanical Disturbances, and Electro-osmosis Serious problems in electrophoretic NMR can arise from heat- ing of the sample by the electric current.There are two possibili- ties for reducing resistive heating: (i) The use of pulsed electric fields. This means that the current is applied only during the actual measuring time 27 in the spin-echo experiment or during the time interval 7in a T, experiment. (ii) The use of an efficient cooling system. The quantity (see equations 6 and 8) is proportional to v*.Thus, for reliable measurements of mobilities, high v+ values and therefore high electric fields E are desirable. On the other hand, without effective cooling the temperatures increase d Tby resistive heating is proportional to E2.Therefore one must find a compromise with respect to these two effects. In practice, electric fields of about 1-100 V cm-are applied for a duration from 50 ms up to ca. 1 s, depending mainly on T, and T2of the nucleus under observation. The cell volumes used are typically 0.5-5 cm3. Depending on the conductivity and the heat capacity of the sample, theoretically the temperature in the cell can increase by ca. 0.01 to 1 "C under the above conditions during one current pulse. Since in most cases signal accumulation is performed, accumulation of resistive heating can also result.These facts clearly reveal the importance of the thermostatting system in practical DCNMR. A temperature rise has two consequences. First, through the temperature dependence of translational molecular motions in liquid systems the mobilities and relaxation times are changed. Secondly, a much more serious problem can arise through convection. Any macroscopic motion in a liquid system (convec- tion, vibrations, or shock waves) can cause dramatic measuring errors in diffusion or mobility measurement by the spin-echo method. The reason lies in the fact that with ionic mobilities in the range of uf = 5 x lod8 m2s-lV-l for example, typical velocities of vh -ms-l are measured and thus with A = 300 ms in a PMFG experiment an actual displacement of the ionic species of 3 pm is measured! This figure clearly demonstrates why convective motions or mechanical distur- bances must be strictly suppressed.In order to stabilize the solution against 'macroscopic' motions like convection, agar-agar and some other gelling agents were tested and successfully used in aqueous solutions by my grou~.~-~ The investigation showed for example that for (C,H,),N+ ions in aqueous solutions with 1 wt% agar an obstruction effect of only 1% is observed. Since, as we will see, most mobility and diffusion measurement are performed rela- tive to a standard system under identical stabilization con- ditions, the obstruction effects can be neglected at reasonable agar concentrations.We recently succeeded in applying a porous sinter glass as stabilizer for a non-aqueous electrolyte However, this stabilizer can only be used if no high resolution spectra are required. Johnson and co-workers2 proposed using fibrin, a material with large pores, which might be. ideally suited for medium-sized proteins. However, it should be emphasized that a number of successful measurements, for example in microemulsions with low conductivities, have been performed in free solution without a ~tabilizer,~ showing1724 that with low currents (0-10 mA) the above mentioned problems are markedly reduced. We saw that mechanical disturbances can also cause serious problems in ENMR.One source of those disturbances is DC pulses, which are switched on and off during the experiments. If the electrical connections to the electrodes and electrodes them- selves are not exactly parallel to the Bofield, strong forces can act on these components and therefore the DC switching can produce shock waves in the solution under investigation. Special care has to be taken therefore rigidly to fix these electrical connections at the probe. The same is true for the gradient coils CHEMICAL SOCIETY REVIEWS, 1994 and their connecting wires. The effect of small vibrations of the electrodes alone is avoided by the gel stabilization, but shocks arising from the other electrical DC connections can only be eliminated by a good mechanical stability of the whole probe system.A further possible difficulty for electrophoretic NMR measurement can come from electro-osmosis. If the walls of the electrophoretic cell are untreated and therefore charged in the presence of an electrolyte, this electrokinetic effect has to be taken into consideration. Then, as discussed in detail else- where,2, in most electrophoretic experiments and also in ENMR, a diffuse layer, including the solvent, flows along the cell walls and generates a counterflow in the centre of the cell. Similar problems to those encountered with convection can then result -but by using stabilizers these electro-osmosis effects can also be circumvented. Johnson and He2 described how to solve the electro-osmosis problem in non-stabilized solutions by care- fully coating the cell surface with a polymer like methylcellulose.In this way, in their ENMR experiments in free solutions, the electro-osmosis flow could be reduced to zero. Thus this problem may also be regarded as basically solved. Finally, we come to a special problem in connection with DCNMR relaxation studies. Macroscopic movements in the solution caused by convection or electro-osmosis can also play a role in measurements of the relaxation time T,. As can be seen from Figure 3 for example, in DCNMR cells there is always electrolyte solution above and below the r.f. coil volume. This fact has to be taken into account if relaxation measurements are performed. The first n or ~/2 pulse in the inversion-recovery or saturation-recovery pulse sequence, respectively, which changes the population of the Zeeman levels, acts only inside the r.f.coil volume. Thus, for example, after a ~/2-pulse, the nuclear magnetization inside the coil volume is saturated, but outside it is not. If during the relaxation delay 7in a v/~T-T/~sequence, convection or electro-osmosis transports species of interest from outside the coil volume into this volume, unsaturated magneti- zation flows into the r.f. coil and is detected by the second n/2- pulse, resulting in an apparently shorter relaxation time T,. o.2 This effect can cause considerable measuring errors, if the relaxation time T, is relatively long and therefore long values have to be applied. With a typical coil length of 1 cmand a convection velocity of 0.25 cm s-l, a T,of 50 ms is decreased by 3% and a T, of 500 ms by 23%.26These effects have been experimentally observed and it has also been shown that this undesired influence on the NMR TI measurement can be suppressed by using a porous sinter glass,26 as described above for the ENMR experiment.In DCNMR relaxation experi- ments, two glass filter discs at the immediate upper and lower ends of the coil are sufficient to eliminate inflow and outflow of magnetization by convection. It should be mentioned that in every DCNMR experiment, due to the drift velocity in the electric field, magnetization transport into the coil and out of the coil does occur. However, at the electric fields normally applied the drift velocities are so small that this effect can be neglected.3.3 DCNMR Cells In Figure 3 we saw the possible basic geometries of DCNMR cells. Three cells in practical use today are now described. In the early days of DCNMR, U-tube cells with counterflow were ~referred~-~3J 2, but now the tendency is towards sample cells of cylindrical geometry (Figure 3c). In Figure 4a a cell is shown, designed by Coveney et aZ.,24 which has easily detachable electrodes. These are Pt-blacked electrodes, in order to prevent gas evolution. For reduction of electro-osmosis effects the perspex tube has a glass sleeving. This electrophoresis cell has been used in connection with a custom-built probe. Another interesting technical solution, by Morris and John~on,~ is shown in Figure 4b.It is a concentric cylindrical electrophoresis chamber having the advantage that both electrodes are in the upper part of the cell and thus vented to the atmosphere. The actual sample of interest is in the inner tube, separated by a gel ELECTROPHORETIC NMR-M. HOLZ (4 Platinum -blacked Glass sleevin *-PEF unit 1Electrodes Teflon spacer Electrolyte solution Sample volume Gel plug +! DC connection Glass filter disc Figure 4 Examples of’the ENMR cells in practical use. (a) ENMR cell, by Coveney et al.,24made of a perspex tube of 8 mm diameter. The electrodes are Pt discs of 3 mm diameter. ‘PEF unit’ means the pulsed electric field generator. (Reproduced by permission from Mol.Phys., 1992,75, 127.) (b) Concentric ENMR cell by Morris and Johnson2’ made of glass. The inner tube is held by a Teflon plug with ventilation holes. The 0.d. of the inner tube is 3-5 mm. (Reproduced by permission from J. Mugn. Reson., 1993, A101, 67.) (c) Probe-head insert for DCNMR, consisting of a Plexiglas body on which the gradient coils are mounted. The electrophoresis cell in the centre is made ofglass and is surrounded by a temperature bath liquid. (The coolant inlet can be seen at the bottom). The cell is easily exchangeable. (Reproduced by permission from J. Mugn. Reson., 1993, A105, 90.) 169 plug from a salt solution in the annulus between the tubes. This salt solution, with high conductivity, serves simply as an electric conductor.Despite the fact that just below the gel plug the current flows perpendicularly to the B, direction, magnetic field gradients in the z-direction are not to be expected since the outer electrode has the shape of a ring and therefore in the x,y-plane currents appear in all directions compensating for the induced magnetic field gradients. This cell design has the further advantage that it might be the best suited for use in commerical NMR probes for cryomagnets. However, there is also a disadvantage. The filling factor is relatively poor, since the sample of interest fills only the volume of the inner tube, and this probably restricts its application to electrophoretic H NMR, where normally no serious sensitivity problems are to be expected.The third example (Figure 4c) is a special probe insert for DCNMR experiments which is in use in my laboratory and which has recently been described in detail elsewhere.28 A Plexiglass body carries the gradient coils and also allows inspec- tion of the electrophoresis cell under working conditions.The cylindrical cell is made of glass and is easily removable for filling and cleaning purposes -the electrodes are fixed on detachable plastic plugs. An important aspect of this cell arrangement is that it is located in a liquid temperature bath. The coolant is pumped through the interior of the Plexiglass body and results in a direct and effective cooling of the electrophoretic chamber. Temperature gradients along the cell axis are also completely avoided.All three cells shown in Figure 4 allow the mobility sign of charged species to be obtained from the direction of the phase shift A&,.9324327 4 Experiments and Applications 4.1 Which Nuclei can be used in ENMR? The measuring principle of ENMR is, as we saw, based on the spin-echo experiment. The higher the maximum A$2T values which can be measured in an experiment, the greater the accuracy of the method. We recognize from equations 6 and 8 that large 7 and A values, respectively, are the most important conditions for successful ENMR measurements. In a standard spin-echo experiment (see Figure 1) the transverse relaxation time T, determines the intensity of a spin-echo at 27 (also see equation 7), hence the longer the T2the better are the conditions for ENMR.One can estimate that under the extreme narrowing condition, where T, = T2is valid, a value for T,, T2> 50 ms is required. This means that in diamagnetic solutions all spin-1/2 nuclei, which relax by the dipole-dipole interaction (e.g. ‘H, 13C, I9F, 29Si, 31P, 205Tl), are nuclei which could be utilized in ENMR experiments. Of course, among these spin-1/2 nuclei, H plays an outstanding role owing to its NMR sensitivity and because of its general importance in chemistry and life sciences. Nuclei with 1> 3 relax almost exclusively by quadrupolar relaxation, which is a strong relaxation mechanism, and these nuclei often have very short relaxation times. However, there remain a number of possible candidates for ENMR experiments and we mention here 2H, 7Li, 23Na, 27A1, 35Cl, and 133Cs. It should be emphasized though that the relaxation times of these nuclei can vary strongly depending on the electronic surround- ings of the nucleus. With a high symmetry of the charge distribution around a quadrupolar nucleus we can expect longer relaxation times.For example 7Li in Li+, with an electronic noble gas configuration, has a much longer relaxation time than 7Li in a covalently bound Li atom. With spin- 1/2 nuclei, e.g. ‘H, residing in molecules with a high molecular mass such as proteins, the molecular correlation times T~ in aqueous solution are typically 10-8-10-9 s and therefore the extreme narrowing condition o~-T~<< 1 is no longer ful- filled. This results in a shortening of T2 and consequently TI > T2is valid. TImight then be one order of magnitude longer than T2,and T2can be too short to achieve the required A-value in the conventional spin-echo experiment.However, there is a modified spin-echo technique, 'the stimulated-echo (STE) tech- nique' which can be applied in this case.21929 In this well-known NMR pulse sequence, magnetization is stored for a time in the range of Tl by two ~/2pulses and after that it is detected relatively shortly after a third ~/2pulse. Therefore the measuring parameter A is only limited by the longer relaxation time Despite the fact that there is a considerable number of nuclei that might be utilized for ENMR, the only successful experi- ments conducted so far have been with 'H, 'Li+, and 133Cs+.This may be regarded as a further indication that ENMR is still in its infancy. Up to now in this section we have spoken solely of electrophor- etic NMR. The situation with respect to the above given conditions for TI and T2 is different for DCNMR relaxation measurements, where quadrupolar relaxing ionic nuclei are of interest.2.10 In these experiments the absolute values of T, and T, do not play such a restrictive role as in ENMR experiments. 4.2 Measuring Methods The practical NMR diffusion measurements and NMR flow measurements are of a very similar nature because in both cases the spin-echo is observed in a PMFG experiment.They differ only in the dependence of M(2r) on the gradient strength G or the pulse length 6, respectively, for v and D, (see equation 7a). Therefore the simplest ENMR experiment is analogous to a standard NMR self-diffusion experiment, where the dependence Of M(27)upon 6 or G is measured. Because the effective gradient G acting over the sample volume is very sensitive to the sample position and geometry, the absolute value of G is normally not known with high accuracy. For reliable and precise NMR diffusion and velocity measurements it is therefore recommend- able to measure relative to a reference ample.^^^^^ For the determination of absolute mobilities in an ENMR experiment, a knowledge of the acting electric field E in the NMR coil volume, the actual measuring volume, is required.This can be obtained by mounting two Pt probe electrodes in the cell just above and below the NMR Using a current- stabilized power supply as the electric field generator so that the value of the current Iflowing in the cell is known then allows the accurate determination of transference numbers T*. In the standard ENMR experiment M(27) is observed in a measuring sequence for every G or 6 value, with and without the applied electric field, giving the amplitude ratio from equation 8. Since we also obtain in this procedure the quantity M(27) in absence of the current, these data can be evaluated with respect to the translation diffusion coefficient D. This fact reveals an interesting advantage of ENMR, since if we measure the mobility p& of an ion, we can always gain, as a 'by-product', the equilibrium transport quantity D*, which delivers further valu- able information on the system of interest.Since in ENMR a phase shift A+ is the quantity of interest, the pure amplitude measurement of M,(~T)in the standard spin- echo experiment can be replaced by measuring procedures where the phase shift is determined more directly. If simple NMR instruments are used, in an 'off-resonance' experiment the phase shift can be directly determined from a phase shift of the time-domain ech~-signal,~*' a method which has certain advantages over amplitude measurements. With a FT-NMR system it is better to make the phase-shift measurement, after Fourier-transformation, in the frequency domain,, 3124 since the signal-to-noise ratio is better in the frequency domain.More- over, in solutions with a number of different ionic species containing the same nucleus, the FT-NMR experiment allows the simultaneous mobility measurement of the various charged species, under the condition that their chemical shift difference is resolved. As an illustration of these direct measurements as a function of gradient pulse length 6, we show in Figure 5 data of 'Li+ drift velocities in aqueous solution at different LiCl con- centrations and electric fields E9A simple slope comparison yields the vf values of interest. The above procedures can also be used with stimulated echoes. CHEMICAL SOCIETY REVIEWS, 1994 60 O.50 o. 40 O-30 O-0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 8(ms) Figure 5 Experimental data of A$*,, the phase shift of the 'Li+ spin- echo signal, as a function of PMFG duration 6 in a standard spin- echo experiment as shown in Figure 1. System: LiCl in water. The slopes of the straight lines are proportional to the drift velocities v+ at different salt concentrations. (0:0.15 M LiC1, E = 12.5 V cm-l; : 0.2MLiCl,E=9.3Vcm-'; A:0.3MLiCI,E=7.1 Vcm-*; 0:0.2 M LiCl + 0.1 M CsCI, E= 6.4 V cm-l) (Reproduced by permission from J. Mugn. Reson., 1986,69, 108.) From the relation A+2T = yGvsA, it can be seen that in ENMR with high-y nuclei, particularly with H, relatively large values of can be achieved. Thus, varying the drift velocity v+ by increasing the electric field (increasing the current I), the value of can be varied up to a value of 3 to 4 n and the cosine modulation of the FT NMR signal intensity can be evaluated.In Figure 6 an example from the work of Saarinen and Johnson' is given, where the (CH,),N+ drift velocity has been observed for the first time in a high resolution experiment. Note that the resonance line of the stationary solvent HDO does not show the cosine modulation and thus moving and non-moving species can be easily differentiated. HOD .a 111111111111 I I I I I I I l l 4.5 4.0 3.5 3.0 PPm Figure 6 The first 'H ENMR experiment under high resolution con- ditions. The dependence of resonance lines of HOD and (CH,),N+ upon the current I, and hence drift velocity v, is shown.(Reproduced by permission from J. Am. Chem. SOC.,1988,110,3332.) A further nice example is given in Figure 7 where the ENMR spectrum of a mixture of ethylendiamine and an amino acid in water is shown,21 and where two charged species move in the electric field with different mobility. In Figure 8 the cosine peak height modulation of the lines in Figure 7 is plotted, showing the different drift velocities of the two species. In FT-NMR the time domain signal which is acquired is normally the second half of the spin-echo; the time axis is called the t,-axis and the Fourier transformation is performed with respect to this axis. The cosine modulation of the signal intensity in electrophoretic NMR described above can also be produced with constant v (constant I) but by an incremented variation of A, the gradient pulse distance.If we denote the time interval A as tl, we have a second time axis and we can perform a second ELECTROPHORETIC NMR-M. HOLZ Figure 7 250 MHz 'H ENMR spectra of a mixture of 1 mM ethylene diamine with an amino acid (1 mM L-alanylglycylglycine) in DzO. (Reproduced by permission from Adv. Mugn. Reson., 1989, 13, 131.) g5.-ar YO aa a -5 -1 0 -15 !0.0 0.1 0.2 0.3 0.4 I0.5 I (mA) Figure 8 The intensity of the 'H NMR lines in Figure 7 versuscurrent I, showing the different cosine modulation for the amine line and the amino acid lines resulting from their different mobilities.(Reproduced by permission from Adv. Magn. Reson., 1989,13, 131.) Fourier transformation with respect to t,, yielding a second frequency axis fl, and hence a two-dimensional ENMR spec- trum can be obtained as in conventional 2D NMR. He and Johnson12'21 introduced such 2D NMR experiments into the field of electrophoretic NMR. These were important in two respects: NMR spectra could now for the first time be resolved on the basis of electrophoretic mobilities, and the chemical shift dimension could be introduced into electrophoretic studies. The f,-frequency in the 2D NMR experiment can, with known electric field, be easily transformed in a mobility p+ sincef, is directly proportional to v+ .Morris and have recently presented ENMR spectra where one axis is the chemical shift axis and the other is the mobility axis, as shown for example in Figure 9. They designated the experiment 'mobility-ordered 2D NMR spectroscopy' (MOSY) and implemented a linear predic- tion analysis of the data with respect to the mobility axis,27 thus replacing in this dimension the common Fourier transformation and thus avoiding truncation broadening of the observed lines. The authors claim a resolution of approximately one per cent in the mobility dimension.Gradient pre-pulses Figure 9 Mobility-ordered 2D ENMR 'H spectrum of 2 mM (CH,): (TMA) in the presence of mixed micelles [1.5 mM sodium dodecyl sulfate (SDS) and 4 mM octaethylene glycol dodecyl ether (C12E8)]. Note the different sign for the mobilities of TMA and C12EB.(Reproduced by permission from J. Mugn. Reson., 1993, A101,67.) In recent 2D experiments, an incrementation of the amplitude of E is preferred over of an incrementation of A, since short and constant A -values reduce diffusional broadening. In 'H NMR, where homonuclear J-coupling plays a role and where often molecules with relatively high mass are investigated, the stimulated-echo (STE) technique has advantages. In Figure 10 a sophisticated pulse sequence is shown, which includes gradient pre-pulses (recommendable in all PMFG experi- ments!), and in which a modified STE sequence is applied where instead of the echo a free induction decay after a further ~/2 pulse is observed. The delay T, is introduced in order to wait for the decay of undesired eddy currents and mechanical distur- bances after the last gradient pulse.3 2D ENMR experiments hold promise for a wide field of 'H ENMR applications.The same might be true for 19Fand 31P, nuclei with high 7-values. However, where nuclei with low y-values have to be utilized, for example in the field of inorganic chemistry, the measurable A4,, values are typically smaller than T,or even n/2, and therefore standard 1D experiments must be used, as shown for example in Figure 5. But this does mean that there is no limitation with respect to the resolution since often single line spectra, or well resolved spectra of only a few lines, are observed with heteronuclei, as large chemical shifts are typical.As mentioned in Section 3.2, the STE technique is also useful in heteronuclear ENMR if measurements have to be conducted in heterogeneous media, as for example in sinter glass. 4.3 Applications of ENMR Nuclear magnetic resonance in the presence of an electric current allows the introduction of NMR techniques into the general field of electrochemistry, and therefore also into applied electrochemistry. The DCNMR methods described above are Figure 10 Modified stimulated echo (STE) pulse sequence for 2D ENMR, after Morris and Gradient pre-pulses are used before the three n/2 rf pulses of the standard STE sequence are applied. The last two ~/2-pulses with the waiting time T, produce a free induction decay (FID), which is observed.(Reproduced by permission from J. Magn. Reson., 1993, A101,67.) nf2 n12 nf2 nf2 nl2 84 flow the first steps towards these new NMR applications. Let us briefly summarize the general advantages which are connected with the NMR method. First, NMR is a noninvasive technique and to a large extent independent of the mechanical and optical properties of the system of interest. It can thus be applied to a great variety of systems. Secondly, and perhaps of more rele- vance, is that by using NMR the selectivity of a spectroscopic method can be included in electrochemical studies. This selecti- vity on a molecular or atomic level allows the observation of distinct particles in a system of interest. The NMR diffusion and flow measuring techniques described above even allow particles to be distinguished from their physically and chemically identi- cal neighbours by means of the phase of their nuclear spin precession.In this way mobile particles can be distinguishd from immobile particles of the same species. It is clear that the selectivity of NMR is an outstanding advantage in the study of complex multicomponent systems, which are in many cases inaccessible to classical methods. (For example, mobility data of single ionic components in mixed electrolytes are rarely found in the literature.) On the other hand, in molecular biology and in applied chemistry electro- phoretic analyses play an important role, and here the combi- nation of electrophoresis with spectroscopic information from NMR is a very tempting approach.Finally, it should be emphasized that owing to the inherent selectivity of NMR, molecular labelling (as required in certain conventional electro- phoresis methods) is not needed, and therefore the properties of the system of interest are not affected by the ENMR experiment. 4.3.1 Application with Heteronuclei in Ionic Solutions Naturally, the first comparisons of ionic mobilities obtained by the novel ENMR method with literature data from classical methods have been made with data from simple binary electro- lyte sol~tions.~-~ The ions considered were (C,H,),N+ 7,8 and Lif in aqueous solution9 and the agreement was found to be within 1to 5%. Afirst systematic study on ternary electrolyte solutions of simple ions has been performed very recently in my lab~ratory.~~?~~The systems were y x 0.5m LiX + (1 -y) x 0.5mCsX in H20(X = C1, Br, I); these are mixtures of lithium and caesium halides at constant ionic strength.Owing to the experimental difficulties associated with the classical methods, it is characteristic that literature data for ionic mobilities, diffu- sion coefficients, and transference numbers are sparse, even for such a simple mixed electrolyte system as this. The ENMR measurements were performed by utilizing the 7Li+ and 133Cs+resonance. These two ionic nuclei are very favourable for electrophoretic NMR studies because of their long relaxation times and from a physical-chemical point of view they are interesting species in a mixture, since their ionic diameters differ appreciably.In order to demonstrate the results that can be obtained in such a system, the composition depen- dence of the ionic mobilities, diffusion coefficients, and transfer- ence numbers of the LiBr-CsBr system is shown in Figure 1I. Acomparison of the composition dependence of D+ and p+ in Figure 11 shows that the cationic diffusion coefficients, which are equilibrium quantities, change linearly with the composition in the mixed electrolyte, whereas the mobilities of Li and Cs +,+ the non-equilibrium quantities, show a non-linear behaviour with opposite curvature. This latter fact indicates that cross- correlation effects influence the ionic mobilities in the ternary system.It is clear that the measurement of single ionic mobilities in mixed electrolytes with more than three constituents should also be easily feasible by ENMR, and this opens new possibilities in the study of mixed simple electrolytes. Mobility studies in non-aqueous systems by conventional methods, e.g. the moving-boundary method, are often beset with difficulties. Gel stabilization of non-aqueous systems is often impossible. However, the use of a sinter glass stabilizer has proved to be successf~1,~~ as demonstrated in Figure 12. In this LiCl in methanol system the mobility of Li+ could be accurately measured as a function of the salt concentration. It can be seen CHEMICAL SOCIETY REVIEWS, 1994 t I I I I2.00 0.2 0.4 0.6 0.8 1 y(Li Br) t 0 cs+ --I1.7 0 Li+1 -7-0.91 1 a I0.2 0.4 0.6 0.8 y(LiBr) C 0.6 -0.6 0.4 0.2 -0.2 0"0 0.2 0.4 0.6 0.8 1 y(LiBr) Figure 11 Results of ENMR experiments on a ternary mixture of simple ions at constant ionic strength.System:y x 0.5~LiBr + (I -y) x0.5mCsBr in H20. (a) The mobilities Y+ of the cations Li+ (0) as aand Cs+ (0) function of the mixture composition. (b) The translational diffusion coefficient D, of Li+ and Cs+ in the same system. (c) The transference numbers T* of the three ionic constituents Li + , Cs+, and Br- in the mixed electrolyte system. (Reproduced by permission from J. Magn. Reson., 1993, A105,90.) that the ENMR studies could be performed up to comparatively high salt concentrations and that the agreement with literature data is excellent.4.3.2 Applications of IH ENMR In electrophoretic NMR, as in conventional NMR, the hydro- gen nucleus 'H plays the most important role. Because of its sensitivity, measurements on charged solutes in the mM con- ELECTROPHORETIC NMR-M. HOLZ of I I I I I 0.00 0.15 0.30 0.45 0.60 0.75 c/rnot I-' Figure 12 Dependence of Li + mobility in methanol upon LiCl concent- ration c at 25"C.23 Open circles: data of ENMR in sinter glass stabilized solutions. Filled circles: literature data, obtained with conventional techniques. centration range are easily possible and there is a huge number of hydrogen-containing charged species which are of interest in many fields of chemistry and biology.We saw that the first lH ENMR measurements were per- formed on binary electrolyte solutions of tetraalkylammonium salts. The next application of the technique was to the investi- gation of mobilities in mixtures of amino acids21 (see for example Figure 7). It can be stated that ENMR may be applied, without difficulty, to investigations of mixtures of simple ions, amino acids, and peptides, where the proton T, values are in the favourable range of some hundred milliseconds. However, the most interesting applications of lH ENMR are expected in the important field of supramolecular chemistry; most examples of applied ENMR come from this branch of chemistry. Coveney et uI.14*24have observed the mobilities of charged micelles, namely the anionic sodium dodecyl sulfate (SDS) and the cationic cetyltrimethyl ammonium bromide (CTAB) in water.In these charged surfactant systems the electrophoretic mobilities at various concentrations were mea- sured (see Figure 13), allowing extrapolation to the critical micelle concentration, where for SDS p-= 4.49x lov8 m2V-ls-l and for CTAB p+ = 3.58 x lo-* m2V-'s-' were found at 25"CZ4 These values agree within 1% and 2%, respectively, with Dye tracer measurements. In the same the mobility of SDS was measured after the ionic environment of the micelles had been changed by the addition of salt. Mixtures of surfactants have also been studied. An impressive example of a 2D mobility-ordered spectrum of negatively charged micelles I 0 SDS I x cI3 2 0 2 4 6 8 10 Surfactant conc.1 Oh wlw Figure 13 Mobilities of charged micelles in water measured by ENMR as a function of surfactant concentration.SDS: sodium dodecyl sulfate, CTAB: cetyltrimethyl ammonium bromide. (Reproduced by permission from Mol. Phys., 1992,75, 127.) in a mixture with positively charged tetramethylammonium ions has already been given in Figure 9.In both the examples of micelle studies given here, the information about the sign of the mobility has been derived from the corresponding phase shift direction. The power of electrophoretic NMR in microemulsion systems has been demonstrated by Johnson and co-workers.21 In micro- droplets of oil in water (o/w) it allows the drift velocity of surfactants and of the hydrocarbon in the droplets to be compared. These interesting applications have been extensively reviewed in the literature.21 An example is given where modern multiquantum (MQ) NMR is combined with electrophoretic techniques in a MQ ENMR experiment.21 This might offer certain advantages, namely simplification of spectra by multi- quantum filtering and increased values, which are import- ant when low mobilities are of interest.However, MQ ENMR suffers from sensitivity problems. Finally, studies on phospholipid vesicles are further interest- ing examples of the application of electrophoretic NMR. These systems consisting of a phospholipid shell around an aqueous inner region may serve as models for biological cells.H signals can be observed from the shell and from the molecules entrapped within. A question of interest here is the charge and size distribution of these polydisperse systems. This polydispersity means that for the same type of species there is a velocity distribution g(v) in the electric field. Under certain conditions this velocity distribution can be derived from electrophoretic NMR data, as shown by Johnson and He2 and as demonstrated practically in a 2D STE NMR e~periment.,~ In this connection it would be interesting to compare the distribution of the translational diffusion coefficients g(D), which is solely depen- dent upon size and mass, with gb), the mobility distribution, which depends additionally upon the charge distribution. ENMR measurements can be made of slowly diffusing supra- molecular systems in a rapidly diffusing solvent such as water and in the presence of other small molecules by the incorpora- tion of 'diffusion filters' as used in modern NMR spectroscopy.By using relatively strong magnetic field gradient pulses, the signals from the highly mobile components in the mixture are strongly quenched and the signals from slowly diffusing particles can be monitored without problems of overlap. 4.4 Relaxation of Quadrupolar Nuclei in Migrating Ions The area of application for DCNMR relaxation studies is different from that just discussed for ENMR and up to now there has only been one publicationlo in this field. This kind of DCNMR represents an attempt to contribute to the basic understanding at a microscopic level of the mechanism of motion of ions in liquid solutions under the influence of an external electric field.It seems to be possible to obtain infor- mation about the behaviour of solvent molecules around the migrating ions and about the relative motion of ions. The first experiments on quadrupolar ionic nuclei (23Na +,87Rb+,and 3sCl-) showed that there are indeed several systems where the quadrupolar relaxation of these ionic nuclei is changed when the ion migrates in the electric field.1° The relaxation rates in the presence of a current I, (I/Tl)Ipo and in the absence of I, (l/Tl)I= have been measured, allowing the determination of o.the ratio RI = (I/Tl),+ o/(l/Tl)I=In Figure 14 the behaviour of this relative relaxation rate is given as a function of the current I, which means as a function of the drift velocity, for 23Na+ in acetonitrile and in methanol.RI can increase or decrease, and this behaviour has also been found in other systems. The increase of the relaxation rate is explained2J0 by a deformation, and thus by a symmetry distortion, of the solvation shell of the migrating ion, e.g.as discussed theoretically in terms of a 'kinetic depolarization effect'.32 Such a structural symmetry distortion increases the electric field gradient at the nuclear site and would increase the relaxation rate. The decrease observed in methanol can be interpreted as a result of a decreasing ion-ion contribu- tion to the quadrupolar relaxation rate2 by an alteration of the 174 1.2 1.1 -1m Nal in AN, 25 “C 1.o 0.9 0.8 I I I I I I 0 50 100 150 200 250 Il mA Figure 14 23Na+ relaxation rate ratio R1 as a function of the applied current lin solutions of lm NaI in acetonitrile (AN) and in methanol (MeOH).(Reproduced by permission from Chem. Phys. Lett., 1987,142,492. cation-anion pair correlation function in the electrolyte solu- tion, when electric field forces are acting on the ions. In other words, the latter effect would then reflect the asymmetry of the ion cloud according to the Debye-Huckel theory. 5 Concluding Remarks and Outlook It is hoped that the present review has shown that DCNMR, a relatively young and promising NMR technique, offers new possibilities for investigating and analysing complex fluid systems with charge-carrying species, such as simple ions, charged macromolecules, and charged supramolecular systems.Up to now, because of the technical problems that have been described, DCNMR in general and electrophoretic NMR in particular have not found extensive application, although it has been shown that many very interesting applications are feasible. The main reason for this may be that purpose-designed equip- ment is not yet commercially available. If so, a most interesting future development will be a DCNMR cell which can be easily incorporated into conventional NMR probes. As a result of the recent development of NMR ‘gradient accelerated spectro- scopy’ (GRASP), more commercial probes are now equipped with gradient coils and therefore in the near future it may be easier to perform ENMR experiments with conventional instrumentation. With respect to future applications, it is expected that IH ENMR will be used to investigate the wide area of polyelectro- lytes and colloidal systems.An interesting general application of ENMR could lie in its use as a ‘mobility filter’, i.e.NMR signals from charged species could be identified when an electric field is applied and the charged particles migrate in this field. We saw that heteronuclear ENMR can be performed with a number of nuclei and it is clear that the future development will also go in this direction. The systematic investigation of mix- CHEMICAL SOCIETY REVIEWS, 1994 tures of simple electrolytes has just begun and will surely be continued.The combination of DCNMR with NMR imaging methods is very tempting and has been proposed previously.2J The future developments of DCNMR relaxation studies cannot be judged so easily. In the reviewer’s opinion, generally very interesting information can be expected; the interpretation of quadrupolar relaxation data is, however, a difficult task.2Jo It is clear that more experimental data are required in order to judge the full power of DCNMR relaxation studies. The application of alter- nating currents, leading to ACNMR relaxation experiments might be a very interesting approach, in particular if high frequencies could be used which would modify the microdyna- mics in electrolyte solutions in a distinct frequency range of interest.6 References I C. Deverell, Prog. NMR Specrrosc., 1969,4,235, 2 M. Holz, Prog. NMR Spectrosc., 1986, 18, 327. 3 D. Geissler and H. Pfeifer, 2.Naturforsch., 1957, A12, 70. 4 K. J. Packer, C. Rees, and D. J. Tomlinson, Adv. Mol. Relax. Proc., 1972,3, 119. 5 M. Holz and C. Muller, J. Magn. Reson., 1980,40, 595. 6 M. Holz and J. Radwan, 2.Phys. Chem. N.F., 1981,125,49. 7 M. Holz and C. Miiller, Ber. Bunsenges. Phys. Chem., 1982,86, 141. 8 M. Holz, 0.Lucas, and C. Muller, J. Magn. Reson., 1984,58,294. 9 M. Holz, C. Miiller, and A. M. Wachter, J. Magn. Reson., 1986,69, 108. 10 R. E. S. Hofmann and M. Holz, Chem. Phys. Lett., 1987,142,492, 11 T. R. Saarinen and C. S. Johnson, Jr., J.Am. Chem. Soc., 1988,110, 3332. 12 Q. He and C. S. Johnson, Jr., J. Magn. Reson., 1989,81,435. 13 K. F. Morris and C. S. Johnson, Jr., J. Am. Chem. SOC.,1992,114, 776. 14 F. M. Coveney, J. H. Strange, A. L. Smith, and E. G. Smith, Colloids and Surfaces, 1989,36, 193. I5 See e.g. J. R. Singer, J. Phys. E: Sci Instrum., 1978, 11,281. 16 C. P. Slichter, ‘Principles of Magnetic Resonance’, 3rd Edn., Springer, Berlin, 1989, p.597ff. 17 K. J. Packer, Mol. Phys., 1969,17, 355. 18 E. 0.Stejskal and J. E. Tanner, J. Chem. Phys., 1964,42,288. 19 P. W. Atkins and M. J. Clugston, Adv. Mol. Relax. Proc., 1975,7, 1. 20 See e.g. S. J. Gibbs, K. F. Morris, and C. S. Johnson, Jr., J. Magn. Reson., 1991,94, 165. 21 C. S. Johnson, Jr. and Q. He, Adv. Magn. Reson., 1989,13, 131. 22 See e.g. Y. Manassen, E. Shalev, and G. Navon, J. Magn. Reson., 1988,76, 371. 23 S. Heil, Diplom Thesis, University of Karlsruhe, 1994. 24 F. M. Coveney, J. H. Strange, and E. G. Smith, Mol. Phys., 1992,75, 127. 25 T. M. Plantenga, H. A. Lopes Cordozo, J. Bulthuis, and C. Maclean, Chem. Phys. Lett., 1981,81,223. 26 B. Straub, Diplom Thesis, University of Karlsruhe, 1989. 27 K. F. Morris and C. S. Johnson, Jr., J.Mugn. Reson., 1993, A101,67. 28 M. Holz, D. Seiferling, and X. A. Mao, J. Magn. Reson., 1993, A105, 90. 29 Q. He, D. P. Hinton, and C. S. Johnson, Jr., J. Mugn. Reson., 1991, 91,654. 30 M. Holz and H. Weingartner, J. Mugn. Reson., 1991,92, 115. 31 S. J. Gibbs and C. S.Johnson, Jr., J.Magn. Reson., I991,93,395. 32 J. P. Hubbard, J. Chem. Phys., 1978,68,1649.

ISSN:0306-0012    

DOI:10.1039/CS9942300165

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

The Hydrides of Aluminium, Gallium, Indium, and Thallium: A Re-evaluation Anthony J. Downs Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QR, U.K. Colin R. Pulham Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3JJ, U.K. 1 Introduction No class of compounds exemplifies better than the hydrides the individualism of the Group 13 elements. In the diversity and profusion of its hydride boron gives little hint of the comparative wasteland making up much of the hydride estate of the heavier Group 13 elements., At a recent count2 about 100 binary boranes are now known, typically as discrete molecules remarkable for their stoicheiometries and structures which have done much to challenge and reshape our under- standing of chemical bonding at large.By contrast, aluminium forms only one binary hydride stable under normal conditions as a polymeric solid, [AIH,],, the a-form of which is isostructural with AlF,, featuring 6-coordinate aluminium atom^.^^^ Attempts to prepare the analogous gallium compound have a chequered hist~ry,~ and it has taken nearly 50 years from the first reported sighting to establish the true credentials of gallane, [GaH,],,6 which now emerges as showing obvious affinities to diborane in the vapour state (i.e. n = 2) while being relatively short-lived under normal conditions. Despite some claims, however, it is unlikely that the hydrides [JnH,], and [TlH,], have yet materialized. In this account we review the current status of the hydrides formed by the Group 13 metals aluminium, gallium, indium, and thallium.Coordinatively saturated derivatives like MH, (M = Al, Ga, In, or T1) and Me,N. MH, (M = A1 or Ga) having been known for some year^,^?^ we are concerned primarily with the parent hydrides, [MH,], (m = 1, 2, or 3; n = 1, 2...), and related unsaturated derivatives. The last category includes species with more than one Group 13 element, for example tetrahydroborate derivatives like Al(BH,), and H,Ga(BH,),-, (m = 1 or 2) and tetraborane(l0) derivatives like 2-R2MB,H, (M = Al, R = Me; M = Ga, R = H or Me). It is appropriate first to consider the physical properties of the binary hydrides [MH,],. Hence it is possible to identify not only Tony Downs has been at the University of Oxfordsince 1966,first as a Senior Research Oficer then as a Lecturer in Inorganic Chemistry; he has been concurrently a Tutorial Fellow at Jesus College.His research interests combine synthetic studies of reactive-hydrido- and organo- derivatives of typical and tran- sition elements with matrix isolation and various spectro- scopic techniques. Following his Ph.D. research at Cam-bridge on perfluoroorgano-sulphur compounds, supervised by the late Professor H. J. Emelkus, F.R.S., he held a Salters’ Fellowship in 1961-2. Prior to the move to Oxford, he was a Senior Demonstrator (1962-3) then Lecturer (19634) at the University of Newcastle upon Tyne. 175 feasible methods of synthesizing compounds with M-H bonds, but also the origins of the thermal lability and reactivity besetting such compounds.1.1 Theoretical Modelling Exploration of the Group 13 metal hydrides has been spurred by the greatly enhanced sophistication of modern computational methods which now admit the use of relatively elaborate basis sets, as well as making due allowance for factors like configu- ration interaction and relativistic correction^.^ Where compari- sons can be made, such calculations typically yield dimensions and energetics which reproduce closely the experimental find- ings, and in some cases improve upon those findings. Such is the case, for example, with the monohydride molecules MH (M = B, Al, Ga, In, or Tl), which are short-lived under normal condition^.^,^ Accordingly we can place some trust in such results to anticipate the likely equilibrium molecular structures, vibrational properties, and binding energies of Group 13 hyd- rides, including numerous species whose existence has yet to be authenticated.Just what inferences are to be drawn will be discussed in Section 2. 1.2 Experimental Methods In many cases, then, theory appears to lead experiment. In fact, the relation is more nearly a synergistic one, for the recent flurry of theoretical enquiries’O has been stimulated to a large extent by experiments, and notably by the synthesis and characteriza- tion of gallane, [GaH3]n.5.6 The practical side is apt to be a Sisyphean task. The hydrides of the Group 13 metals are, without exception, unusually susceptible to attack by air or moisture.None is thermally robust and most decompose at ambient temperatures. That the compounds are nearly always handled in vacuo is a minimum, not a sufficient requirement. Among the special procedures which it has been necessary to Colin Pulham began his career in researchfirst as a Part IIstudent and then as a D.Phi1. student at Oxford under the supervision of Tony Downs. He remained in Oxford for a further two years as a postdoctoral research assistant and as an EPA Ce- phalosporin Junior Research Fellow at Linacre College. In 1992 he was awarded a Royal Society University Research Fellowship and moved to the University of Edinburgh. His current research interests include structural studies of the Group 13 hydrides in the con- densedphases and the develop- ment of the hydride chemistry of tin and lead.adopt is the use of all-glass apparatus individually constructed for each experiment and with provisions (i) for rigorous pre- conditioning (ideally by heating under continuous pumping) and (ii> for close temperature control extending over all the exposed surfaces to which the hydride has access. Manipulation and characterization of the product then depend on keeping the temperature and vapour pressure down to the levels compatible with the experimental needs. As to the detection and specifica- tion of a Group 13 metal hydride, there are four aspects which stand out.1.2.1 Vapour Studies at Low Pressure Vapour samples including low partial pressures of the molecules may be generated, usually by high-energy methods involving, CHEMICAL SOCIETY REVIEWS, 1994 and intensities of the symmetnc and antisymmetric stretching vibrations of an M-H-M bridge give a useful indication of the M-H-M bond angle. 54 Other questions of interpretation affecting, say, the molecular structure can be addressed by examining the response of the spectrum to changes of isotopic composition. Thus, the infrared spectra associated with the different isotopomers MH,D, -,(M = A1146J5a or Ga;' sb x = 0-2) and AlH,D,-, (x= 0-3)14a leave little doubt about the equivalence of the hydrogen atoms in each of these molecules. 1.2.4 Structure Determination Such is the thermal instability of most of the hydrides formed by the Group 13 metals that it would be a brave, and possibly vain venture to attempt to grow single crystals.Preliminary studies of for example, the action of an electric discharge on M/H, deuteriated gallane do show encouraging signs that crystalline mixtures(as with MH, where M = Al,llaGa,llbIn,llcorTl,lldpowder samples at low temperatures are amenable to neutron and A1H2'le), but in the case of Ga2H6 simply by careful vapori~ation.~.~The sample can then be interrogated in various ways. For example, the electronic emission spectrum may admit the characterization of various electronic states of MH1 Id and AlH2 le molecules. The infrared spectrum measured ideally at high resolution, as in studies using a diode laser, may afford a detailed picture of the dimensions as well as the vibrational and rotational properties of MH molecules in their electronic ground states.Mass spectroscopy is a third option, which has been exploited to test for the molecules MH, (M = Al, Ga, or In) and Al,H, in high-temperature vapours.12 Yet another method of attack is to appeal to the electron-diffraction pattern of the vapour, as in the case of Ga2H6,536 to determine the vibration- ally averaged structure of the molecule. I .2.2 Trapping Experiments The tracking and identification of Group 13 metal hydrides have frequently called for trapping. This may be achieved physically by quenching the vapour on a cold surface either alone or with an excess of a suitable diluent, or by photolysis of appropriate precursors also trapped in a low-temperature matri~.~ For example, the reactions 1 and 2 have been shown to occur on broad-band irradiation of matrix samples.GaCl HX Ar mafnx, hv ClGaHX (X = H13nor CllJh)+ (1) Ar or Kr matnx, hv + 3H* + AlH3 (ref. 14) Alternatively the experimenter may resort to chemical trapping of the hydride by treating it with a compound likely to undergo a facile and quantitative reaction yielding a known product. An illustration is provided by the addition reaction 3 which gives rise to the known adducts H,Ga(NMe,), (m = 1 or 2) and so attests to the identity of the hydride precurs~r.~Jj I .2.3 Infrared Spectroscopy Foremost among the properties used to identify and character- ize a Group 13 metal hydride is its infrared spectrum.As with the study of metal carbonyls, it is the stretching vibrations which by their energies and intensities are likely to offer the most telling commentary on the molecular identity. Where u(C-0) modes differentiate between terminal and bridging carbonyl functions, so u(M-H) modes differentiate between terminal and bridging M-H functions. For example, the stretching fundamentals of terminal Ga-H bonds give rise to strong infrared absorptions in the range 1720-2050 cm-l, whereas the corresponding vib- rations of bridging Ga-H-Ga units absorb with variable intensity in the range 900-1720 cm-l. In addition, the energies diffraction.However, there is ample circumstantial evidence that aggregation is a primary motif and that the form of such compounds is liable to vary substantially from one phase to another. In practice, definitive measurements of the structures and dimensions of discrete hydride molecules have depended mostly on detailed analyses of the rotational lines associated with specific vibrational transitions. These may appear either in infrared absorption (e.g.for molecules of the type MH1 lb*c)or in the optical emission spectrum characterizing one or more excited electronic states of the molecule (e.g. MH1ld and AlH,' le). The only other method of stucture determination to be enlisted for discrete gaseous molecules is electron diffraction.This has been the principal agent of characterization not only for the binary gallium hydride Ga2H6,6 but also for mixed hydrides like GaBH,," HGa(BH,),,'* and 2-GaB,Hl,.1g 2 Physical Properties of the Hydrides The hydrides of the Group 13 metals carry bonds which are polarized in the sense M*+-H*-. The pseudo-anionic hydrogen ligand is thus susceptible to electrophilic attack, and the metal centre to nucleophilic attack, this partly by virtue of the positive charge and partly by virtue of the vacant orbital which it bears. Herein lie the seeds of aggregation and other distinctive properties. Table 1 lists thermodynamic and vibrational properties, together with dimensions, for Group 13 hydrides of the types MH, MH,, and [MH,], (n = 1 or 2).The parameters elicited from ab initio calculations have been augmented, where poss- ible, by experimental results so that there are some opportunities for testing how well theory imitates nature. All but one of the species are gaseous molecules; the sole exception is the solid polymer [AlH,], which appears to be unique among the binary hydrides of Group 13 in being formed exothermally from the constituent elements in their standard states. Analysis of the properties en bloc brings out a number of features germane to this class of compounds at large. 2.1 Heats of Formation All the gaseous molecules are endothermic with respect to the elements in their standard states; allowance for the effects of entropy indicates that none of them is thermodynamically stable at normal temperatures with respect to the decomposition reaction 4.On the other hand, the unsaturated valence shell of the species MH,, allied to the relatively electron-rich nature of the hydrogen ligands, means that substantial stabilization is to be found through aggregation with the formation of M-H-M bridges. Thus, the best estimates of the standard enthalpy change associated with reaction 5 are -82, -64.5, and -46 kJ mol-l for M = B, Al, and Ga, respectively.loc In the case of aluminium such aggregation can continue to give the polymer [AIH,], with each A1atom bound to six bridging H atoms;4 that AHfor reaction 6 is estimated8uJ0 to be -169kJ mol-l must be a major factor governing the stability of the solid hydride.Table 1 Physical properties of the simplest binary hydndes of the Group 13 elements Dimensionsu Thermodynamic propertiesb Vibrational properties,c theor exP theor exP theor exP re(M-H)/pm (angle/") r,(M-H)/pm (angle/") B(M-H)/kJ mol-' A*,, K/kJ mol-1 B(M-H)/kJ rno1-I dfHT,, K/kJ mol-i v(M -H)/m-' 122 5,d 123 0,' 124 1,f 122 78 123 24h 347 4d +435 6d 333 9' +449 6' 2513,*d 2532,g 2366 90" 118 5 (126 5): 119 4 (127 1)8 ro 118 (131)r 344d +313d ca 4ood ca +201d 2867,* 2728*d -I118 8,d 119 Y(120) 379d + 82 4d 373' + 100 0' 28 13,* 2693*d 2808,* 2623*k 118 9(t), I32 4(b), If8 4(t), 131 4(b), -+ 19 6' -+ 17 8' 2785-1758' 26 I 3-I 6 15*" LB-Hb-B 84 8' LB-Hb-B 83 1" 165 2,d 166 5,' 167 4,f 165 lg 164 738" 303d + 245d 289' + 259 2' 1771,*d 1738g 1799,* 1766*q159 5 (118 O),d 159 0 (118 7)g ro 159 (119)p 249d + 268d --1976,* 1954*d 1682 43" 158 4,d 158 6,' 159 3,f 157 1' -287d + 123d -2024,2021' 1882 I*' ( 120)156 3(t), 171 9(b), +58 5d' 2047-1368' -LAl-Hb-A1 97 9' r, 171 5, LAl-Hb-Al -46 0' -ca 1600*' 141 2s 167 5,' 169 6,f166 2," 167 2,v 166 21 18" 271" + 224" 270" + 225" 1612," 1605g 1603 9566 (69GaH)w 167 7g 158 0 (120 3)," 160 0 (119 5)," re -(136)" 220" + 274" -1799,* 1728*x 158 0 (1 19 7)g 157 7,' 158 6,J 156 5,' 155 7," 260" + 151" 2055,2028' -158 2'(120) 155 2(t), 175 3(b), ra 152(t), 171(b), + 105'" 2 10 1- 1294' 1993--1202*Y LGa-Hb-Ga 96 1' LGa-Hb-Ga 9&y 183 8,' 185 0,f 182 3,2184 99 183 77630"" 2502 +211' 2Ybb 1469,2 1539 1475 4343 (IlsInH)On I178 2 (119 7),bb 175 5 (119 0)e 191bb + 296bb --172 5,' 173 4,f 175 3bb (120) 22jbb + 222bb --189 9,' 191 2,f 188 3," 189 9g 187 02h 1 97bb cc + 203bbcc 199bb 1413,cc 1300,& 1383g 1391 2681 (20STlH)ce I -I185 4 (121 5),bb 177 6 (119 9):~ -143bb + 332bb 176 0 (121 7)g 174 5,' 175 6,f 178 8,bb 173 gCc 1 80bb + 297bb (120) b bridgmg atom, t terminal atom B refers to the actual M-H bond energy for MH, but to the mean M-H bond energy for molecules containing more than one M-H bond = Vibrational wavenumbers are for a harmonic oscillator unless labelled with an astensk which denotes an anharmonic oscillator d Ref 8a ab initio MO theory, Msller-Plesset theory to full fourth order with a senes of extended basis sets c Ref 9 ab inrtio relativistic MO theory, Msller-Plesset theory to second order allowing for spin-orbit coupling f Ref 9 ab initio relativistic MO theory with quadratic configuration interaction and allowing for spin-orbit coupling 8 Ref 8e ab initro SCF + CI calculations using effective core potentials and taking account of relativistic effects for the heaviest atoms ' h Ref 1Id Ref 20 G Herzberg and J W C Johns, Proc Roy SOC, 1967,2%A, 142 k A Kaldor and R F Porter, J Am Chem SOC, 1971,93,2140J Ref 1Oc ab initio MO theory, single and double excitation coupled cluster (CCSD) method with DZP basis sets m J L Duncan and J Harper, Mol Phys , 1984,51,371 n J L Duncan, J Mol Spectrosc , 1985,113,63 Ref 1 la p Ref 1 le 4 Ref 15a Ref 14a * Ref 4 H Rosnnski, R Dautel, and W Zed, Z Phys Chem (Frankfurtam Main), 1963,36,26 u Ref 8b ab iniiio MO theory, complete active space MCSF (CASSCF) followed by full second- order configuration interaction (SOCI) calculations " Ref 8d ab inrtio MO theory, Msller-Plesset correlation corrections to fourth order, MP2 = Full optimization -Ref 1lb x Ref 15b " Ref 6 Ref 8c ab initio MO theory, complete active space MCSCF (CASSCF)/second-order configuration interaction/relativistic CI calculations Ref 1 lc bb K Balasubramanian and J X Tao, J Chew Phys , 1991,94,3000 ab initio MOtheory, complete active space multiconfiguration SCF/second-order configuration interaction/relativisticCI calculations P Schwerdtfeger, P D W Boyd, G A Bowmaker, H G Mack, and H Oberhammer, J Am Chem SOC, 1989, 111, 15 relatlvisticcc pseudopotential MO calculations dd P A Chnstiansen, K Balasubramanian, and K S Pitzer, J Chem Phyf , 1982,76, 5087 relativistic calculations including spin-orbit coupling and configuration interaction n R -D Urban, A H Bahnmaier, U Magg, and H Jones, Chem Phyf Lett, 1989,158,443 MH,(g)-+ M(s) +If H2(g)2 (4) MH,(g)-+tH,M(I*-H)zMH,(g) (5) 1AIH,(g) -+-IA1H3lAs) (6) How the gaseous MH, molecules compare in their heats of formation with the hydrides formed by adjacent elements in the Periodic Table is illustrated in Figure l(a).Hence it is apparent that the Group 13 hydrides are peculiarly high-energy species, the heat of formation of the characteristic hydride formed by each element in a given period rising to a maximum at Group 13.This position is modified somewhat when the hydrides assume their natural forms at room temperature. Thus, the hydrides of the alkali and alkaline-earth metals are typically exothermic compounds by virtue of the large (mainly coulombic) stabiliza- tion energies which accrue through hydrogen-bridging on lat- tice-formation. The corresponding energies, although substan- tial for AIH, (as noted above), are smaller for the other MH, species in Group 13, and assume still less importance to the gross stabilities of the hydrides formed by the elements in succeeding Groups. +3MH(g) 2M(s) + MH3(g) (74 3MH2(g)+ M(s) + 2MH3(g) (7b) The heats of formation of the different Group 13 hydrides, MH,, can also be compared, in juxtaposition to those of the solid chlorides, in a thermochemical oxidation state diagram [see Figure l(b)].Under normal conditions and for all M, the results show that MH and MH, molecules are both highly susceptible to disproportionation in accordance with equations 7a and 7b. By contrast, although solid MCI, is invariably unstable with respect to disproportionation, solid MCl is prone to dispro- portionate when M = A1 but not when M = TI. In this respect, CHEMICAL SOCIETY REVIEWS, 1994 . (M = Al, Ga, In,or Ti) -400 --600 1 Y1' AICi3 (5) -800 0s(s' i +1 +2 +3 Oxidation state (b) ROWin the Periodic Table 1 2 3 4 5 +loo '\ 2 3 4 5 Row in the Periodic Table (C) Figure 1 Thermodynamic properties of Group 13 metal hydrides: (a) standard enthalpies of formation, Afsg8 (in kJ mol-I), ofs-andp-block hydrogen compounds; (b) oxidation state diagram giving experimental (0)or estimated (A)values for the standard enthalpies of formation of compounds of the types MX, MX,, and MX, (M = Group 13 element; X = H or C1); and (c) standard enthalpy 1changes for the reaction -MH, + +Cl,(g) +-1 MCI, + fH,(g), where n n M is an s-or p-block element.then, hydrogen is akin to CH,, being a 0-type ligand with little potential to bring out oxidation states lower than + 3, even in an element like thallium. At high temperatures, where entropy changes are likely to have a major voice, however, the thermody- namic balance will inevitably tilt in favour of MH. THE HYDRIDES OF ALUMINIUM, GALLIUM, INDIUM, AND THALLIUM-A.J. DOWNS AND C. R. PULHAM Bond enthalpy A comparison of the energetics of formation of hydrides and chlorides is revealing too for what it tells us about the feasibility of metathesis reactions of the type shown in equation 8. Figure l(c) plots periodic variations in the enthalpy change attending 1 1the transformation -MH, + )C12(g) -+-MCI, + +H,(g). The n nmore exothermic the process, the greater is the potential of MH, -and its derivatives, presumably -as a precursor to the hydrides of other metals. In this respect, as Figure I(c) demonstrates, the more-or-less ionic hydrides formed by the metals of Groups 1 and 2 are unsurpassed. In practice, for reasons of low solubility, purity, and state of subdivision, they do not always live up to their potential.The Group 13 elements clearly occupy a position intermediate between the s-block metals and the less metallic p-block elements of Groups 14 and 15. Aluminium stands out in Group 13 as the element which emulates most closely the metals of Groups 1 and 2. This reduces severely the synthetic options for the preparation of aluminium hydrides; for example, reac- tion 9 which works well as a source of gallium hydrides with n = 1 la or 2,, lb has no useful counterpart for aluminium, once it is appreciated that the reagent Me,SiH is itself made by chloride-hydride exchange normally at the expense of the AI-H bonds of LiAlH,. Ga,CI, + 2nMe3SiH -t [H,GaCl,-,], + 2nMe3SiC1 (9) 2.2 Bond Enthalpies The Group 13 metal hydrides, [MH,],, owe their elusiveness, so it is commonly suggested, purely and simply to the weakness of the M-H bonds, as opposed to the strengths of the M-M and H-H bonds in the elements.That ‘the truth is rarely pure and never simple’ appears to be the principle upheld by the mean bond enthalpies displayed in Figures 2(a) and 2(b). Weak the M-H bonds may be on average, certainly compared with their M-C1 and M-0 counterparts (thereby accounting in part for their vulnerability to attack by air and moisture), and also with the bonds to hydrogen formed by adjacent p-block elements of Groups 16-17. Yet they are stronger than the metal-hydrogen bonds incorporating neighbouring s-block metals (notwith- standing the durability of the solid hydrides formed by these metals); they are also marginally stronger than the analogous M-C bondsz2 (notwithstanding the position vis-u-vis com-pounds of the type MMe, which, unlike the hydrides, are well authenticated and long-lived at room temperature).HF 571dl B Al Ga L 400 -$-300 --2B 5 200 -s loot 0 B Al Ga In TI (b) Figure 2 Mean bond enthalpies (in kJ mol-l) (a) for hydrogen com- pounds of the s-andp-block elements, and (b) for the bonds formed by Group 13 elements to H, C1,0, and C in comparison with dfflg8 for the gaseous metal atoms. The explanation of these anomalies, which is partly thermo- dynamic and partly kinetic, hinges on the capacity of the metal M and the ligand X to enter into the formation of M-X-M bridges.Reliable thermodynamic data are sparse but some idea of the strengths of M-X-M bridges for the Group 13 elements B, Al, Ga, and In may be gained from the estimated mean bond enthalpies listed in Table 2. In the case where X = H, the bridges vary in strength in the order B > A1 > Ga, but for X = C1or Me the order changes to B << A1 > Ga; with M = A1 or Ga the bridges become weaker as X changes in the sequence C1 > H > Me. The metals of Groups 1 and 2 are also electroni- cally unsaturated in discrete molecules like MH or MH, and they too have the ability to aggregate through M-H-M bridges. The signs are that the individual M-H-M bridges are somewhat weaker than those formed by the Group 13 metals, but the greater degree of unsaturation of the s-block metals CHEMICAL SOCIETY REVIEWS, 1994 Table 2 Strengths of bridging bonds M-Xb-M, where M = B, Al, Ga, or In and X = H, C1, or Me (mean bond energies, B, in kJ mol-l) Group 13 element, M B(M-Xb-M) X = HU B(h4-Xb-M) -B(M-X,) B(M-Xb-M) x = CP B(M-Xb-M) -B(M-X,) B(M-X,-M) X = Meu B(h4-xb-M) -B(M-X,) B 451h 76h 444' ca.0 370d ca. 0 A1 351h 64h 575' 195' 324d 43d Ga In 30tih ? 46h ? 407' 394h 51' 66h ca. 25gdJX ca. 233df ca. 3fg ca. 19 B(M-X,-M) = B(M-X,) + fdH[M,X,(g) -+ 2MX,(g)]. No allowance has been made for the effects of 'reorganization' (see K. Wade, 'Electron Deficient Compounds', Nelson, London, 1971, pp.128-131). Subscripts: b bridging; t terminal. See Table 1. Ref. 20. Ref. 22. C. Chatillon and C. Bernard, J. Cryst. Growth, 1985,71,433. Estimate based on the thermodynamics of vaporization. 8 D. S. Matteson, 'Organometallic Reaction Mechanisms', Academic Press, New York, 1974, p. 38. h F. Defoort, C. Chatillon, and C. Bernard, J. Chem. Thermodynamics,1988,20, 1443. causes this factor to be outweighed by the number of such bridges (6 or 8) filling out the coordination shell of the metal in the extended three-dimensional array of the crystalline solid. The same principle is presumably at work with solid [AlH,],, which owes its unusual thermodynamic stability ultimately to the propensity of aluminium to rise to six-fold coordination., By contrast, gallium tends to follow boron in remaining four- coordinate. Indium and thallium are more prone to assume high coordination numbers but the weakness of In-H and Tl-H interactions then militates against the formation of a solid stable at normal temperatures.Insofar as decomposition of a discrete molecular hydride MH, may be initiated by homolytic dissociation of an M-H bond, it is not obvious why a Group 13 hydride like gallane should be so much more labile than, say, stannane [see Figure 2(a)]. Presumably, however, there are circumstances in which an associative mechanism, possibly preceding concerted elimina- tion of H,, offers a lower barrier to reaction than does a purely dissociative one. In that case the tendency of the hydride to associate or dissociate and the strength of M-H-M bridge bonds are likely to be crucial to the kinetics of decomposition.By contrast, decomposition of the methyl derivatives Me,M appears typically to be dissociatively activated2, and so to be opposed by a relatively large energy barrier; in this case associat- ive activation is made ineffectual by the weakness of M-Me-M bridges. It would be fascinating to know more about the kinetics of decomposition of a molecule like Ga,H, in the gas phase, particularly in the light of the complex behaviour of on pyrolysis,1+2but it will be a tall order to prevent heterogeneous surface reactions from playing the dominant r6le. 2.3 Structures On the accumulated evidence of experimental and theoretical enquiries, the M-H bonds in monomeric Group 13 metal hydrides vary in length in the order MH > MH, 2 MH,, the trihydride typically displaying M-H bonds up to 10 pm shorter than that in the monohydride.The MH2 molecules in their electronic ground states are bent with bond angles close to 120°, as attested experimentally in the case of A1H2.11e The MH, molecules in their electronic ground states are planar with DJh symmetry, in keeping with the prognosis of VSEPR theory, and that too finds support from the first definite sighting of the AlH, as generated in a low-temperature matrix and identified by its infrared spectrum. Perhaps the most distinctive property of the hydrides in physical terms is their proclivity for aggregation. With regard to the dimer of MH,, a single configuration seems to hold sway, namely the familiar diborane-like one, H,M(p-H),MH,, as deduced experimentally for Ga,H, (Figure 3) on the grounds of its infrared spectrum and electron-diffraction pattern.5,6 For the dimers [MH], and [MH,],, however, the situation is evidently more complicated, although this is an area where hypothesis leads and experiment lags far behind. Diborene, B2H2, the boron analogue of acetylene, has a linear triplet Distances given in pm, angle in O Figure 3 The structures of the gaseous molecules (a) Ga,H, (ref. 6) and (b) 2-GaB3H,, (ref. 19) as deduced by electron diffraction. ground-state, with no experimental evidence for significant B-B double-bonding.8e On the other hand, a bis(p-hydrido) struc- ture (1) with Dzh symmetry is found to be the most stable form of M2H, for all the other Group 13 elements.8e Instead of the linear form, which is now only a transition state, a trans-bent isomer (2) (C,h) appears as a second minimum on the potential energy surface for M,H, where M = Al, Ga, or In.There are two other possible isomers, namely the asymmetric M-MH, (3) (C,")and the mono-H-bridged HM(p-H)M (4) (Q, which typically give minima at energies not far above the global minimum for M,H,. Configuration (3) is relatively accessible to all the Group 13 metals, (4) to all but T1. Infrared measurements suggest that the dimer of elemental gallium, Ga,, reacts spontaneously with H, in an argon matrix to form Ga(p- H),Ga (1) and that this can be converted photoreversibly into a second isomer HGaGaH @).Isb The dimer Tl,H, may be unsubstantiated by experiment, but it has attracted particular attention as a model compound in connection with the vexed question of whether Tl-TI bonding interactions of significant strength can be e~tablished.~?~~ THE HYDRIDES OF ALUMINIUM, GALLIUM, INDIUM, AND THALLIUM-A. J.DOWNS AND C. R. PULHAM H X=400--520nrnGa' 'Ga H'G"-y(10)I X132O4Wnrn H Theoretical calculations2 3u--c also find quite different proper- ties for molecules of the type M2H4 according to whether M = B or M = A1 or Ga. Boron favours a classical structure H,B-BH, with D2d symmetry, whereas Al,H, and Ga,H, prefer a tri-H- bridged structure (5) with C,, symmetry, and studies of the mixed hydrides BGaH, and AlGaH, lead to analogous global minima.23d Such a structure can be described in terms of more- or-less coulombic interactions between an M+ cation and a tetrahedral MH$ anion, thereby betraying disproportionation of the divalent species./"? But dimerization need not be the limit to association, what- ever the experience of BzH6 might suggest. The trimers [MH,], (M = B, Al, or Ga) are calculated to have substantial binding energies with respect to the monomer MH,;lob the energetically favoured structures of and Ga3H, involve planar six- membered [MH], rings with D3hsymmetry, (6), but these are only slightly more stable than the acyclic structures, (7), con-forming to C, symmetry and featuring a pentacoordinate central M atom.Whereas the hypothetical reaction 11is endoer- gic for M = B (+ 121 kJ), it turns out to be exoergic for M = A1 (-32 kJ) and M = Ga (-42 kJ).lob No homonuclear molecule of the type M3H, is known, but gallium does form the mixed hydride GaB,H, with the acyclic structure (7) based on a pentacoordinated central Ga atom.' Although the compound undergoes some form of weak association in the condensed phases, it maintains a terminal Ga-H unit. By contrast, the corresponding aluminium compound is a strongly associated, involatile liquid in which the unique AI-H bond is involved in bridging, probably to complete six-fold coordination of the Regarding the formation of still larger oligomers, there is as yet no guidance to be had from theory.What is known experi- mentally about aluminium trihydride indicates that polymeriza- tion to a three-dimensional network ultimately offers optimum stability. Preliminary studies of the corresponding gallium com- pound leave little doubt that the Ga2H6 molecules present in the vapour also aggregate in the condensed phase^.^.^ However, the properties of solid gallane (its infrared spectrum, volatility, and solubility in solvents like toluene) point to the formation not of an extended polymer but of a discrete oligomer such as the tetramer [GaH3],, (8), which, unlike solid AlH,, retains terminal M-H bonds. Direct metal-metal bonding has yet to be established as a significant principle in the hydrides of the Group 13 metals, and there are no known homonuclear analogues of B-B bonded polyborane species like B4H10, B10H14, and B1,H1,2-.1 Tempting though it may be to invoke M-M bonding in molecules (M = A1 or Ga) in view of the short M **a M distance, current theoretical opinion gives little or no weight to such an interaction. O The mixed hydride GaB3Hl has been described -ud (8) recentlylg and its structure is akin to that of B4H10 but with gallium replacing boron at the 2-position so that interaction between the heavy atoms is confined mainly to the B-B hinge of the GaB, skeleton (Figure 3).That compounds containing M-M bonds can be made has been demonstrated unequivocally with the isolation and structural characterization of species like [(Me,Si),CH],M, (M = Al, Ga, or In),2sa [(2,4,6-Pr\- C6H,),M,]'-(M = A1 or Ga),2s6 [(~S-CsMe,)A1]4,2sc and [Al12Buil,]2-.25d Evidently, however, large substituents are needed at the metal centres to inhibit more extensive aggrega- tion, usually at the expense of rupturing the metal-ligand bonds.Hence it is far from certain that hypothetical species like Al,,H,, and [Gal,Hl,]2-, with the potential for kinetic and thermodynamic stabilization of the M-H bonds, will enjoy more than a transient existence under normal conditions, although they may have a part to play in the thermal decompo- sition of hydrido- and organo-derivatives of the metals. Group 13 metal clusters are already known in intermetallic com- pound~.~Typically they are not discrete but interlinked, as with the Ga,, and In,, units found in the phases Rb0,6Na6.25 Ga20.02260and Na,In, 1.8,26b respectively.3 Chemical Aspects 3.1 Synthesis Most of the routes likely to lead to metal hydrides have been tried at one time or another with the Group 13 metals. Perhaps the most conspicuous exception is the action of a proton source on an electron-rich form of the Group 13 metal (as in an intermetallic compound in combination with a Group 1 or Group 2 metal, q.v.), there being no reported parallels to Stock's classical synthesis of the boron hydrides from a metal boride and a protonic acid. On the other hand, compounds containing bonds between hydrogen and a Group 13 metal have been prepared with varying degrees of success by the following methods: (i) addition of H, or a hydrogen-containing molecule like HCl to an unsaturated form of the metal; (ii) metathesis involving a hydride source and a halide or other derivative of the metal; (iii) decomposition or elimination reactions; and (iv) acid-base reactions.3.1.1 Addition of H, and Related Reactions Direct synthesis of a Group 13 metal hydride from the elements is a high-energy process and even the metal atoms seem to require electronic or some other form of excitation before they will add to H,. For example, the results of matrix experiments involving photolysis of A1 atoms in the presence of H, argue for the following mechanism (equation 12). l4*ls' Similar insertion reactions occur -sometimes spontaneously, more often with photolytic initiation -between a Group 13 metal atom and a hydrogen-containing molecule HX to give the MI1 hydride HMX (e.g.X = NHZ2, or Me' sb).Given the appropriate stimu- lus, molecular derivatives of the univalent metals, such as AlH, AlCl, and GaCl, will also add to H, and HCl to give the corresponding MI1' hydride, as in equations 1 and 12. So far this strategy has been confined to the small scale of matrix-isolation experiments but it could be viable as a method of synthesis on the larger scale, at least for products with the thermal stability to survive at or near ambient temperatures. An important recent important recent advance in this direction has been made through the preparation of metastable AlClZ8a and GaC1286 in cooled solutions. + I"2 h~3 AlH, 3.1.2 Metathesis This has been one of the principal agencies of synthesis, with hydride-halide or hydride-methyl exchange being exemplified in reactions 9,2113,6 and 14,246J9The choice of hydride ion source is dictated by its activity and by the possibility that exchange may also lead to mixed hydride derivatives, as in reactions 15'' and 16.5The reactions are normally carried out between the neat reagents and in the absence of a solvent, a feature prescribed by the need to avoid contamination of the free hydride, notably by basic impurities.%H,GaCl], + LiGaH, 243-250K 1'-[GaH,], + LiGaH,Cl n (13) -!-[MMe,l, + M'MH4--[Me,MH], + M'MH,Me1 m n (M = A1 or Ga; M' = Li or Na) (14) 250K 1+[H2GaCl], + LiBH, --[GaBH6], + Licl (15)n MMe, + B2H6-Me2M@-H),BH2 + f[BMeH,], (M = Ga or In) (16) 3.1.3 DecompositionlEliminationReactions An aluminium hydride is often the first product, accompanying elimination of the appropriate alkene, in the thermal decompo- sition of a trialkylaluminium compound, although it usually takes a 2-substituted alkene to make this a workable synthetic route.22 Controlled pyrolysis of triethylgallium, brought about by infrared laser radiation, has also been shown to proceed via p-elimination (equation 1 7).,O f Ga + gH2 \ toluene &In: -3o oc/ CHEMICAL SOCIETY REVIEWS, 1994 Et,Ga -+ +[Et,GaH], + H,C=CH, (17)1 )[EtGaH,], + H2C=CH, 3.1.4 Acia'Base Reactions For many years the displacement reaction 18 was believed to afford a route to gallane,5 being driven by the preference of Me,N for coordination to BF, rather than GaH,.Unfortuna- tely halide-hydride exchange and not displacement turns out to be the dominant pathway. On the other hand, the action of the unusually 'soft' base CO, with its marked affinity for coordina- tion to BH,, has been turned to account in reaction 19.246 258 K 1Me,N.GaH,(s) + BF,(g) --[GaH,],( 1) + Me,N- BF,(s) n (18) 3.2 Reactions Some of the chemical properties of gallane have now been ~harted,~.~with the results summarized in Scheme 1. Apart from the obvious parallels with diborane,' the reactions seem to typify the behaviour of Group 13 metal hydrides. Reaction pathways of particular note with regard to the wider significance of these compounds are: (1) thermal decomposition to the metal; and (ii) complex-formation with Lewis bases.In addition, the Group 13 metal hydrides are powerful reducing agents, the M-H bonds being highly susceptible to a wide range of oxi- dation, addition, exchange, or solvolysis reactions. There is the potential appeal of unusual selectivity in the reduction of organic substrates, as shown by AlH, and Bu',AlH for exam- ple,,l and which may repay the formidable investment which the preparation and handling of such compounds normally demand. 3.2.1 Thermal Decomposition A recurring theme of this account is the thermal frailty of the known binary hydrides of the Group 13 metals which decom- pose to the elements at or near ambient temperatures. In fact, it is the ability to deliver the pure metal which confers considerable advantages on the hydrides over more conventional organome- + 3H2 />-20°C NMe,HI, I,Ga-H HI NMe3 Scheme 1 Preparation and some reactions of gallane.THE HYDRIDES OF ALUMINIUM, GALLIUM, INDIUM, AND THALLIUM-A. J. DOWNS AND C. R. PULHAM I83 tallic precursors for the creation of thin films of 111-v compounds or of the Group 13 metal, as in the growth of GaAlAs and the ‘metallization’ of semiconductor devices., In addition, the hydrides may be either active intermediates or parents to such intermediates in the thermolysis reactions associated with useful CVD processes and our current knowledge of which owes more to conjecture than to well substantiated fact.That they may serve as instructive models for surface reactions is borne out by very recent spectroscopic studies of the adsorption of hydrogen atoms on the (1 x 6) reconstruction of GaA~(100);~~ unmistak-able evidence has thus been found for the formation of gallium hydride surface species bearing both terminal and bridging Ga-H functions. 3.2.2 Complex-formation One of the most characteristic reactions of a Group 13 metal hydride is that with a Lewis base. This may proceed with homolytic, symmetrical cleavage of MG-H),M bridging units to give the corresponding molecular adduct in which the hydride is coordinated by one or more base molecules, e.g. (Me,N),GaH,(n = 1 or 2) and H,P*GaH, (see Scheme l).536 More or less tetrahedral coordination of the metal centre is the norm in discrete 1:1 complexes like Me,N -MH, (M = A1 or Ga) in the gas phase.On the other hand, the greater predisposition of aluminium to rise to coordination numbers greater than 4 is manifested in the dimeric structures assumed by I :1 complexes like C4H80*A1H334a in the solid state (see Figure 4). Here we find an A12H6 unit with highly unsymmetrical Al-H-**Al bridges, evincing a relationship with the elusive dialane, A12H6. Complexation can also evolve through an alternative channel, with heterolytic, unsymmetrical cleavage of M(p-H),M bridg-ing units to give a salt-like product most aptly formulated in terms of cationic and anionic metal hydride moieties, 596e.g.[H,Ga(NH,),]+[GaH,]-and [H,AlL]+[AlH,]- (L = N,N,N‘,N”,N“-pentamethyldieth ylenetriamine). 4b Complexes of all kinds are easier of access, more stable and, usually, more tractable than the parent hydrides. Accordingly they are preferred in most of the applications which the Group 13 metal hydrides have found, although they do not necessarily duplicate the behaviour of the parent hydrides. Most familiar is the use of alanes -notably LiAlH, and its derivatives -as powerful and selective reducing and hydrogenating agents.,., Key transformations involve the reduction of organic >c=C(, )C=O,and -CrN groups. In inorganic chemistry, too, adducts of alane and gallane have considerable synthetic potential not only as reducing agents, but also as precursors (i) to other compounds of the metals, the generation and survival of which requires that strongly reducing conditions be maintained, as in reaction 2O,, and (ii)to hydride derivatives of other metals, as in reaction 21.36In addition, compounds like Me,N*AlH, are attractive as agents for chemical vapour deposition of the Group 13 metal in thin film te~hnology.~~ Here they offer the advantage over more conventional organometallic sources of affording carbon-free deposits. Figure 4 Molecular structure of the adduct [C,H,O.AlH,], in the crystalline solid.34u Et 0($-C5H,),TiCl, + Et,O*AlH,L .0Et2 + +H, (21)(~5-C5H5)2Ti(p-H)2AlC12 4 Conclusions Unlike the boron hydrides, which have waxed more or less strongly since their discovery, the hydrides of the heavier Group 13 elements have suffered fluctuating fortunes.From the first fine careless rapture of the energy-hungry 1950s and 1960s, aluminium hydride has lapsed into semi-obscurity, even while some of its derivatives have gained prominence as commercial and/or laboratory reagents. Gallane, [GaH,],, has had an even more vexed history, with its very existence befogged for nearly 50 years by claims and counter- claim^.^ However, the recent experimental advances that have led to the synthesis and charac- terization not only of gallane,5*6 but also of various novel adducts of alane and gallane, and to the isolation of discrete molecules like AlH, and GaH, in solid matrices at low tempera- tures have breathed new life into an otherwise faltering chemical body.One of the principal consequences has been to stimulate theoretical enquiries at a relatively sophisticated From such calculations have come results which have served the dual purpose of testing and supplementing experimental find- ings on known compounds, and urging the performance of new experiments to seek out other, hitherto undisclosed hydrides whose properties have been foretold. Such is the weakness and reactivity of M-H and M-M bonds that the chemistry of the Group 13 metal hydrides is never going to rival that of the corresponding boranes. On the other hand, this rather thinly populated borderland between the electron-precise regions of conventional ionic and molecular species still raises fundamen- tal issues of structure and bonding, while the very reactivity and thermal instability of the compounds, as exemplified by the adduct Me,N*AIH,,31~32~35 may yet be virtues.5 References 1 ‘Gmelin Handbook of Inorganic Chemistry, 8th edn., Boron Com- pounds’, Syst. No. 13, Parts 14, 18 and 20, 1977-79; 2nd Supple- ment, Vol. 1, 1983; 3rd Supplement, Vol. 1, 1987; Springer Verlag, Berlin and Heidelberg. 2 N. N. Greenwood, Chem. SOC.Rev., 1992,21,49. 3 M. J. Taylor and P. J. Brothers, in ‘Chemistry of Aluminium, Gallium, Indium and Thallium’, ed. A. J. Downs, Blackie, Glasgow, 1993, p. 11 1. 4 J. W. Turley and H. W.Rinn, Inorg. Chem., 1969,8, 18. 5 A. J. Downs and C. R. Pulham, Adv. Znorg. Chem., 1994,41, 171. 6 C. R. Pulham, A. J. Downs, M. J. Goode, D. W. H. Rankin, and H. E. Robertson, J. Am. Chem. SOC.,1991,113,5149. 7 See, for example, W. J. Hehre, L. Radom, P. v. R. Schleyer, and J. A. Pople, ’Ab Initio Molecular Orbital Theory’, Wiley, New York, 1986. Q Distances given in pm, angle in O 8 (a)J. A. Pople, B. T. Luke, M. J. Frisch, and J. S. Binkley, J. Phys. Chem., 1985,89,2198. (b)K. Balasubramanian, Chem. Phys. Lett., 1989, 164, 231. (c) K. Balasubramanian, J. Phys. Chem., 1990, 94, 6582.(d)C. W. Bock, K. D. Dobbs, G. J. Mains, and M. Trachtman, J.Phys. Chem., 1991,95,7668. (e)G. Treboux and J.-C. Barthelat, J. Am. Chem. SOC., 1993,115,4870. 9 P.Schwerdtfeger, G. A. Heath, M. Dolg, and M. A. Bennett, J.Am. Chem. SOC.,1992,114,7518. 10 (a)C. Liang, R. D. Davy, and H. F. Schaefer,111,Chem. Phys. Lett., 1989,159, 393. (b) B. J. Duke, C. Liang, and H. F. Schaefer, 111,J. Am. Chem. Soc., 1991,113,2884.(c)M. Shen and H. F. Schaefer, 111, J. Chem. Phys., 1992,%, 2868. 1I (a)J. L. Deutsch, W. S. Neil, and D. A. Ramsay, J. Mol. Spectrosc., 1987, 125, 115. (b) R.-D. Urban, U. Magg, and H. Jones, Chem. Phys. Lett., 1989,154, 135. (c)A. H. Bahnmaier, R.-D. Urban, and H. Jones, Chem. Phys. Lett., 1989,155,269. (d)K. P. Huber and G. Herzberg, ‘Molecular Spectra and Molecular Structure. IV. Con- stants of Diatomic Molecules’, van Nostrand Reinhold, New York, 1979. (e)G. Herzberg, ‘Molecular Spectra and Molecular Structure.111. Electronic Spectra and Electronic Structure of Polyatomic Molecules’, van Nostrand, Princeton, N. J., 1966, p. 583. 12 P. Breisacher and B. Siegel, J. Am. Chem. Soc., 1965,87,4255. 13 (a)R. Koppe and H. Schnockel, J. Chem. SOC., Dalton Trans., 1992, 3393. (b) R. Koppe, M. Tacke, and H. Schnockel 2.Anorg. Allg. Chem., 1991,605,35. 14 (a)F. A. Kurth, R. A. Eberlein, H. Schnockel, A. J. Downs, and C. R. Pulham, J. Chem. SOC., Chem. Commun., 1993, 1302. (b)G. V. Chertihin and L. Andrews, J. Phys. Chem., 1993,97, 10295. 15 (a)J. M. Parnis and G. A. Ozin, J.Phys. Chem., 1989,93,1215,1220. (b)Z. L. Xiao, R. H. Hauge, and J. L. Margrave, Znorg. Chem., 1993, 32,642. 16 C. R. Pulham, K. S. Knight, and A.J. Downs, unpublished results. 17 C. R. Pulham, P. T. Brain, A. J. Downs, D. W. H. Rankin, and H. E. Robertson, J. Chem. Soc., Chem. Commun., 1990, 177. 18 M. T. Barlow, C. J. Dain, A. J. Downs, G. S. Laurenson, and D. W. H. Rankin, J. Chem. SOC.,Dalton Trans., 1982,597;A. J. Downs, L. A. Harman, C. R. Pulham, D. W. H. Rankin, H. E. Robertson, and P. F. Souter, unpublished results. 19 C. R. Pulham, A. J. Downs, D. W. H. Rankin, and H. E. Robertson, J. Chem. SOC.,Dalton Trans., 1992, 1509. 20 ‘CRC Handbook of Chemistry and Physics’, ed. D. R. Lide, 74th Edn., CRC Press, Boca Raton, 1993-1994. 21 (a)H. Schmidbaur, W. Findeiss, and E. Gast, Angew. Chem., Int. Ed. Engl., 1965, 4, 152; H. Schmidbaur and H.-F. Klein, Chem. Ber., 1967,100,1129.(b) M. J. Goode, A. J. Downs, C. R. Pulham, D. W. H. Rankin, and H. E. Robertson, J. Chem. SOC., Chem. Commun., 1988,768. 22 ‘Comprehensive Organometallic Chemistry’, ed. G. Wilkinson, F. G. A. Stone and E. W. Abel, Pergamon Press, Oxford, 1982, Vol. 1, Chapters 1, 6, 7, and 8. CHEMICAL SOCIETY REVIEWS, 1994 G. A. Stone and E. W. Abel, Pergamon Press, Oxford, 1982, Vol. 1, Chapters I, 6, 7, and 8. 23 (a) R. R. Mohr and W. N. Lipscomb, Znorg. Chem., 1986,25, 1053. (b)K. Lammertsma, 0.F. Guner, R. M. Drewes, A. E. Reed, and P. v. R. Schleyer, Znorg. Chem., 1989,28,313. (c)K. Lammertsma and J. Leszczynski,J.Phys. Chem., 1990,94,5543. (d) J. Leszczynski and K. Lammertsma, J. Phys. Chem., 1991,95, 3941. 24 (a) P.R. Oddy and M. G. H. Wallbridge, J. Chem. SOC., Dalton Trans., 1978, 572. (6) L. A. Jones, D.Phi1. thesis, University of Oxford, 1993. 25 (a)W. Uhl, Z. Naturforsch., 1988,43b, 11 13; W. Uhl, M. Layh, and T. Hildenbrand, J. Organomet. Chem., 1989,364, 289; W. Uhl, M. Layh, and W. Hiller, J. Organomet. Chem., 1989,368, 139. (6) R. J. Wehmschulte, K. Ruhlandt-Senge, M. M. Olmstead, H. Hope, B. E. Sturgeon, and P. P. Power, Inorg. Chem., 1993, 32, 2983. (c) C. Dohmeier, C. Robl, M. Tacke, and H. Schnockel, Angew. Chem., Znt. Ed. Engl., 1991,30,564. (d)W. Hiller, K.-W. Klinkhammer, W. Uhl, and J. Wagner, Angew. Chem., Znt. Ed. Engl., 1991,30, 179. 26 (a)M. Charbonnel and C. Belin, J. SolidState Chem., 1987,67,210. (6) S. C. Sevov and J. D. Corbett, Znorg. Chem., 1992,31, 1895. 27 J. A. Howard, H. A. Joly, P. P. Edwards, R. J. Singer, and D. E. Logan, J. Am. Chem. SOC., 1992,114,474. 28 (a)M. Tacke and H. Schnockel, Znorg. Chem., 1989,28,2895. (b)M. Tacke, H. Kreienkamp, L. Plaggenborg, and H. Schnockel, Z. Anorg. Allg. Chem., 1991,604,35; D. Loos, H. Schnockel, J. Gauss, and U. Schneider, Angew. Chem., Znt. Ed. Engl., 1992,31, 1362. 29 P. L. Baxter, A. J. Downs, M. J. Goode, D. W. H. Rankin, and H. E. Robertson, J. Chem. Soc., Dalton Trans., 1990, 2873. 30 D. K. Russell, Coord. Chem. Rev., 1992,112,131; A. S. Grady, R. D. Markwell, and D. K. Russell, J.Chem. SOC., Chem. Commun., 1991, 14. 3 1 See, for example, E. R. H. Walker, Chem. SOC. Rev., 1976,5,23; J. A. Miller, in ‘Chemistry of Aluminium, Gallium, Indium and Thal- lium’, ed. A. J. Downs, Blackie, Glasgow, 1993, p. 372. 32 See, for example, A. T. S. Wee, A. J. Murrell, N. K. Singh, D. OHare, and J. S. Foord, J.Chem. SOC., Chem. Commun., 1990,Il; F. M. Elms, R. N. Lamb, P. J. Pigram, M. G. Gardiner, B. J. Wood, and C. L. Raston, J. Chem. SOC.. Chem. Commun., 1992, 1423. 33 P. E. Gee and R. F. Hicks, J. Vac.Sci. Technol. A, 1992,10,892; H. Qi, P. E. Gee, and R. F. Hicks, Phys. Rev. Lett., 1994,72,250. 34 (a) I. B. Gorrell, P. B. Hitchcock, and J. D. Smith, J. Chem. SOC., Chem. Commun., 1993, 189. (b)J. L. Atwood, K. D. Robinson, C. Jones, and C. L. Raston, J.Chem. SOC.,Chem. Commun., 1991,1697. 35 F. G. N. Cloke, C. I. Dalby, M. J. Henderson, P. B. Hitchcock, C. H. L. Kennard, R. N. Lamb, and C. L. Raston, J. Chem. SOC., Chem. Commun., 1990, 1394. 36 B. M. Bulychev, Polyhedron, 1990,9,387.

ISSN:0306-0012    

DOI:10.1039/CS9942300175

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Trimetallic Units as Building Blocks in Cluster Chemistry D. lmhof and L. M. Venanzi Laboratorium fur Anorganische Chemie, ETH Zurich, Switzerland 1 Introduction Over several decades cluster chemistry has produced a bewilder- ing variety of complex structures. Fortunately, in more recent times, theoretical studies have provided useful rationalizations for the existence of such compounds and for their structural characteristics. Particularly important for the development of cluster chemistry has been the recognition that even com- pounds with very complex structures can be described in terms of assemblies of ‘building blocks’, often clearly recognizable as molecular fragments, using the isolobal analogy. The most recurrent structural feature in cluster chemistry is the trimetallic unit M,.A variety of metal centres, particularly those of Groups X and XI with d’O-electron configuration, build molecular compounds containing such units, generally held together by bridging ligands, and/or direct metal-metal bonds. A striking property of many trimetallic fragments is their reactivity, which can, at least formally, be described in terms of Lewis acidic or basic character and, therefore, the ability to interact with conventional Lewis acids or bases to produce more complex structures. This account provides a brief summary of the chemistry of the best known M,-fragments and of the types of larger clusters they can produce. The potential uses of such species will also be briefly mentioned.2 Clusters of the Group X Elements containing M3-Units The elements of the nickel triad readily form complexes contain- ing discrete M,-units or, even more readily, clusters which can be considered as being made up of triangular units joined either through an M,-edge or a single M-atom. Although the most extensive range of compounds is given by platinum (Pf,), palladium (Pd,) and nickel (Ni,) also form such clusters (Figure Pt,R/Pt Pd, ,Pd Ni, /Ni Pd Ni Figure 1 The most marked feature of these compounds is their readi- ness to form larger heterometallic clusters. The known skeletal units containing Pt, or Pd, ( ) and fragments containing other metals (0)are shown in Table 1. ~~~~ ~~ ~ Daniel Imhof was educated at the Swiss Federal Institute of Technology (ETH), Zurich where he received his Dipl.S. Nut. in 1990. He is currently completing work for his Ph.D. at the ETH Zurich in the labor- atory of Professor L. M. Venanzi. His fields of interest include cluster chemistry and its application in homogenous and heterogenous catalysis. Table 1 Pt,- and Pd,-Heterometallic clusters Z3l-T0 Ir Sn U I I I I 3 Platinum Compounds The best known Pt,-clusters, of the type [Pt,(p-CO),(PR,),] (3:3:3),are readily obtained from a variety of complexes, as can be seen in Scheme 1. There are also many other related trimetallic clusters, with bridging ligands other than CO and terminal ligands other than PR,. Some of these are shown in Figure 2.4---7 Furthermore, compounds with different Pt-to-ligand ratios are known, viz.[Pt3(p-C0)3(PR3)4] (3:3:4),4.s [Pt4(y-CO),(PR,),] (4:5:4),9 and [Pts(p-C0)6(PR,),] (5.~5:4).~Their structures are shown in Figure 3. The richest heterometallic cluster chemistry is given by com- pounds of the 3:3:3-type. This arises mainly because these units can act either as Lewis bases or as Lewis acids, as can be clearly seen in the frontier orbital diagram shown in Figure 4. The HOMO, C,-symmetric a;-orbital is doubly occupied and its electrons can be used to form a ‘dative’ bond. Furthermore the LUMO a>-orbital, which is also C,-symmetric, can be used Luigi M. Venanzi received his Dipl. Chem. from the University of Kiel in 1952.He then joined the Butterwick Research Laboratories of I.C.I. Ltd., working in J.Chatt j. research group. In 1956 he was appointed I.C.I. Research Fellow at the University of Oxford where he was awarded his D.Phi1. degree in 1958. He then became a Lecturer at the Inorganic Chemistry Laboratory of that Univer- sity and a Fellow of Magdalen College, Oxford, positions he held until 1968 when he joined the State University of New York at Albany as Professor of Chem-istry. In 1971 he became E.I. du Pont Professor and Chairman of the Chemistry Department at the University of Delaware. Professor Venanzi has been Professor of Inorganic Chem- istry at the Swiss Federal Insti- tute of Technology since 1973. His research interests range from homogeneous catalysis to cluster chemistry, with a strong bias towards the coordination chemistry of the platinum-group metals and polydentate p hosp h ines .185 CHEMICAL SOCIETY REVIEWS, 1994 (3:3:3) Scheme 1 Synthetic routes for (3;3:3)-clusters. P" p" 3:3:4 4:5:4 964 Figure 3 Orbital Energy (eV) LUMO a2' HOMOal' Figure 4 A simplified LCAO-MO scheme of the frontier orbitals in a cluster of the type Pt,L,. TRIMETALLIC UNITS AS BUILDING BLOCKS IN CLUSTER CHEMISTRY-D. IMHOF AND L. M. VENANZI as 'acceptor' orbital, imparting Lewis-acidic character to the Pt,-unit. It should also be noted that the a-character of both orbitals renders them best suited to combine with totally sym- metric acceptor or donor orbitals. The metal centres with which they can interact, either as Lewis bases or acids, are summarized in Figure 5.Figure 5 Metal centres forming heterometallic clusters with [Pt, (p-L),(L),]-units. 3.1 Pt,-Clusters as 'Electron Donors' The 3:3:3-complexes readily form heterometallic clusters of several types. The most common, with half-sandwich structure, has the composition [{3:3:3}ML]+, where M is copper(I), sil- ver(r), and gold(1) and L a ligand such as phosphine. The 3:3:3-complexes also bind to the copper, silver, and gold monohalides, the zinc and cadmium dihalides, and the indium trihalides.lOJ Furthermore, the cations copper(I), silver(I), gold(r), and cad- mium(rr) even form sandwich-type complexes of the compo- sition [(3:3:3},M]+ (Figure 6).12 Generally, the preparative methods for complexes of the above types are extremely simple, requiring only the addition of the appropriate reagents in stoichiometric ratios.Some specific reactions are represented in Scheme 2. I+ I+ 6r I Scheme 2 Reactions of (3:3:3)-clusters with Lewis acids However, phosphine ligand exchange is possible and, there- fore, isomerically pure complexes of the type [{3:3:3)ML]+ can be obtained only when phosphine equilibriation of the reagents can be prevented. O Other Pt,-species, most notably those having SO,-or RNC-bridging groups, can also form heterometallic clusters of half-sandwich 3~14A representative selection of such com- pounds is shown in Figure 7. As might be expected, the 'donor ability' of these triangular 0 -PR3 I' X , PR3i' Figure 6 M = CU+,Ag', Au+ L = so2,cNxyl.cl Figure 7 units is affected by the nature of the bridging and terminal ligands.Preparative studies show that the tendency to form 'half-sandwich' compounds decreases in the order: [Pt3&- co)3(PR3)31 > [Pt3&-SOz 2&-CO)(PR3 31 'lPt 3&-so2 3 (PR3 )31 > [Pt,(,-CNR)z&-Co)(CNR)(PR3) 21 > [Pt3 b-CNR) 3 (CNR),(PRs)] > [Pt3&-CNR)3(CNR)3].' O>l3 Even complexes of the 3:3:4-type can form half-sandwich complexes (Figure 8). Figure 8 3.2 Electron-transfer Reactions Particularly interesting are the redox processes which take place between the 3:3:3-clusters with mercury(r1) compounds. As can be seen in Scheme 3, a platinum(0) 3:3:3-species donates electrons to a mercury(r1) halide giving a platinum@) complex of the type [Pt,X,(CO),(PR3),] and a bicapped [{3:3:3)[HgX),] cluster containing, at least formally, mercury(1).X '-i 'PR3 -rX Scheme 3 Reactions of (3:3:3)-clusters with mercury halides. Indeed, Pt3Hg,-clusters can be directly obtained by reacting one equivalent amount of a 3:3:3-compound with an Hg,X, salt. Furthermore, the redox potentials are such that any remaining HgX, will oxidize platinum(1) intermediates to plati- num(I1) complexes, once again with formation of 'HgX'-frag- ments which are stabilized as pentametallic clusters. Interest- ingly, in the solid state, these compounds are dimeric with a structure of the type shown in Figure 9.These electron-transfer reactions are not limited to mer- cury(I1) halides: they occur even with organomercury species, as indicated in Scheme 4.If one could find inexpensive clus- ters, which gave analogous reactions, one could use them for the detoxification of solutions containing organomercury compounds! CHEMICAL SOCIETY REVIEWS, 1994 9 Figure 9 A schematic representation of the crystal structure of [{Pt,~-CO),(PCy,),},(HgBrj,](0 = Hg; 0 = Pt; 0= Br; 0 = P; 0 = 0; 0 = C). Ph-Ph + R3P\ .. cis-[Ptph2(C0)(PPhPr',)J .,I ,*' cis-[PtXPh(CO)(PPhPi2)] Xi' Scheme 4 Reactions of (3:3:3)-clusters with organomercury compounds. 3.3 Pt,-Clusters as 'Electron Acceptors' As mentioned earlier, the 3:3:3-clusters can also act as 'electron acceptors'.In this context, it is noteworthy that the 'electron donors' that bind readily to the Pt3-units are heavy B-metal atoms or ions with d1Os*-electron configuration, i.e.,mercury(o) and thallium(I), as can be seen from the examples shown in Figure 10.1731s An interesting feature of these compounds of the first two types is that they tend to associate in the solid state, in the case of mercury forming weak Hg-Hg bonds, and in the case of thallium, through electrostatic interactions with coordinated halides. The preparation of complexes of the above types is also remarkably simple, as shown in Scheme 5. The mercury-containing complexes have very intense colours, ranging from blue to violet. Thus, when a solution of a 3:3:3-cluster is exposed to traces of mercury vapour, its surface becomes blue.It appears likely that the formation of these mercury addition compounds could be used to construct a sensitive detector for mercury vapour. Clusters in which the basic Pt3-unit is bicapped by heterome- tals other than mercury are also kno~n.'~.'~ However, these appear to be exclusively formed (1) by Pt3-fragments having SO,-bridging units or (2) where the triangle of Pt-atoms is held together only by metal-metal bonds, as in the examples shown in Figure 11. 4 Palladium Compounds The cluster chemistry of palladium is poor compared with that of platinum. This is likely due to the low stability of Pd-Pd bonds and the lability of the species formed. Although a variety TRIMETALLIC UNITS AS BUILDING BLOCKS IN CLUSTER CHEMISTRY-D.IMHOF AND L. M. VENANZI XylNC\ NXvl "p I 8 8 r . 'i PR3 Figure 10 npF6 f benzene pz 8,* -88 .a' (3:3:3) 6omin m2a2 I in solution Scheme 5 Reactions of (3:3:3)-clusters with Lewis bases. Au l+ -PR3 Figure 11 of homometallic clusters have been identified, they are of varying nuclearities and structures, e.g. of the types [Pd4(p- )4I,' iPdS h-C0)6(PRd41, 'CO)5 (PR3)419' O [Pd4(~c-CO)6(PR3and [Pd7(p-CO)7(PR3)7]23 shown in Figure 12. The compo- sition of the isolated product appears to depend on the relative solubilities of the species present in solution as well as the electronic and steric properties of the phosphine.Furthermore, no trimetallic complexes of the type [Pd,(p-CO),(PR,),] appear \ PR3 co R8\ " PR3 Figure 12 T ol+ Scheme 6 Reactions of palladium clusters with silver cations. to be known. However, the reaction of tetrametallic clusters with silver salts has produced heterometallic species of the types shown in Scheme 6. 5 The 'Chini Clusters' The most intriguing series containing triangular M,-units, formed by clusters of the Group X metals, are those obtained by CHEMICAL SOCIETY REVIEWS, 1994 insoluble olive green yellow green blue green violet green W -orange red Scheme 7 Reaction pathway for the formation of [{Pt3(p-CO),(CO),}x]z-clusters. Chini and co-workers. The series of compounds given by n platinum is shown in Scheme 7.24 A schematic representation of the crystal structures of two members of this series are shown in Figure 13.24 The formation of the trimetallic anion [Pt3(~-C0),(CO),l2 -can be easily rationalized in terms of the MO diagram shown in Figure 4.The two additional electrons occupy the a;-orbital, its high electron affinity being a consequence of the low 0-donor and high n-acceptor capacities of the terminal CO-ligands. However, this dianion can be considered as a strong Lewis base and, therefore, can associate with [Pt,(p-CO),(CO),] acting as a Lewis acid. Furthermore, the electron-donor capacity of the resulting hexametallic dianion remains sufficiently high to allow its association with one additional, uncharged Pt,-unit forming a ‘triple-decker’ unit.Up to ten triangular fragments can be Figure 13 Schematic representations of the crystal structures of [{Pt, assembled through such interactions (see Scheme 7). (p-CO),(CO),},]*- and [{Pt3(p-CO)3(C0)3}3]z- (0= Pt; 0= C; The Chini-type compounds also provide one of the rare 0 = 0). examples of clusters containing triangular nickel units, i.e., [{Ni3(p-CO),(CO),},]2 -,which has the same basic structure as Table 2 Cu,-, Ag,-, and Au,-Heterometallic clusters the corresponding platinum complex shown in Figure 13.2s 6 Trimetallic Copper, Silver, and Gold 1Compounds Fe 0s Nb Ta Nb Ta Many compounds containing M,-fragments, where M is a coinage metal, are known. However, the basic structural units forming such species differ from those described earlier for platinum.Here, the M,-fragments are held together by polyhyd- rido-complexes of the later transition elements of composition ‘M’H,L,’ (e.g., M’ = Re, Ru, Os, Rh, and Ir; m = 2 and 3; L = tertiary phosphine and CO). The heterometallic com-pounds thus formed are mainly of the types {M,}{M’H,L,},, {M3}{M’HmL,,}2,and {M,}{M‘H,L,} shown in Table 2 (M = , M‘ = 0). TRIMETALLIC UNITS AS BUILDING BLOCKS IN CLUSTER CHEMISTRY-D. IMHOF AND L. M. VENANZI I 12+ Scheme 8 Reactionof M'H,(CH,C(CH,PPh,)), (M' = Rh and Ir) with one, two, and three equivalent of Au(PPh,)+. Many polyhydrido complexes readily assemble up to three AuL -cations in a stepwise manner. The products obtained by + successive addition of Au(PPh,) to the hydrido-complexes + [M'H,{CH,C(CH2PPh2),}] (M' = Rh and Ir) are shown in Scheme 8.26 It is noteworthy that the tetrametallic M'Au,-cluster is formed only after deprotonation of the above hydrido-complex, i.e.the 'assembling unit' is the anionic species '[M'H,{CH,C (CH,PPh,),}]-' containing M' in the formal oxidation state of (0.A particularly interesting feature of clusters of this type is the strict C3-symmetry of the metal and P-atoms, despite the fact that only two of the three M'Au,-faces contain bridging hydro- gen atoms, a feature confirmed by a neutron diffraction study of [{Au(PPh,)},H,{Ir{CH,C(CH,PPh,),}}][PF6], (see Figure 14)., Thus, these M'Au,-clusters show the hitherto unique feature of having trimetallic faces with bond distances and angles which are not affected by the presence of bridging hydrides. P2 a4 H An LCAO-MO study of complexes containing M'Au,-units has been carried out and the stability of clusters of this type has been correlated with the number of bridging H-atoms present.,* Related compounds containing tung~ten,,~rhe-nium, and ruthenium (M' = Re, L = PMe,Ph, z = l;,l M' = Ru, 3L = CH,C(CH2PPh2),, z = 2,,), as well as rho- have also been reported, and one example of each type is shown in Figure 15.1 PPh3 'PPhj Figure 15 The assembly of Ag,- and Cu,-triangles requires higher metal to hydride ratios, the best combination being 1 :1.34 However, in these cases, the hydrides can preserve the formal oxidation state, e.g.(111) for rhodium and iridium. The largest class of heterometallic hydrides has the compo- sition [{M}3{M'H3L3)3]3+(M' = Rh, M = Cul and Agl; M' = Rh or Ir, M = Cul, Agl, or Aul; L = phosphine) and can be obtained as shown in Scheme 9. These clusters contain planar hexametallic units. The crystal structure of the rhodium-silver complex has been determined., Although the hydride ligands could not be located on the Figure 14 An ORTEP drawing of the cation [(AU(PP~~)}~H~{I~(CH,C (CH,PPh,),))][PF,], obtained by neutron diffraction. (The crystallo- Fourier map, the position of the Rh-P vectors and the Rh-Ag graphic C,-axis of the cation generates three bridging hydride ligands. distances indicate that the Rh- Ag edges are alternatively singly Each of these is to be taken as having3 occupancy.) Ir-H = 1.77(4) A; and doubly hydrogen-bridged.However, in solution, all the Au-H = 1.95(4) A. hydride ligands appear as equivalent on the NMR time-scale. CHEMICAL SOCIETY REVIEWS, 1994 Scheme 9 Planar hexametallic clusters from M’H,{CH,C(CH,PPh,),} and one equivalent of M+ (M’ = Rh, M = Cu, Ag; M’ = Rh and Ir, M = Cu, Ag, Au). Although all the above compounds were obtained with plati- num-metal hydrides having tripod-like tritertiary phosphines as co-ligands, these are not necessary, as shown by the formation of [{Cu3}H,{os(PMe,Ph)3)}31*+ .36 Furthermore, one can also obtain clusters with a coinage metal-trihydride ratio of 3:2, e.g. [{Cu(MeCN},H, {Ir(PMe,Ph)3},]3+.Its crystal structure has been determined and a schematic representation of the cluster core is shown in Figure 16.37 Figure 16 A schematic representation of the crystal structure of [{Cu3(MeCN),}H6{1r(PMe,Ph),),l3+ (0= Ir; 0 = Cu; 0 = P; 0 = N; 0= H). While the clusters described above provide fascinating mater- ial for theoretical studies, as well as aesthetically pleasing examples of molecular architecture, up to now, the potential for practical applications of these classes of compounds has scarcely been explored. An exception is provided by the recent study by S. Miiller and E. Newson of some homo- and heterometallic clusters as catalyst precursors for a number of heterogeneously catalysed dehydrogenation reactions (equations 1 and 2).0 0+3H2 (1) Methylcyclohexane(MCH) Toluene Catalyst 0+ 4H2 (2) n-Heptane(n-C,) Toluene Different types of compounds were used, mainly [Pt3(p- CO)3(PR3)31, [{pt,O,-Co),(PR3),},Agl[CF,SO,I,and [{Pt3(~- CO)3(PR,),},Cu][PF,].The most extensive series of tests was carried out on reaction 1. The results are summarized in Figure 17. 46 Conversion methylcyclohexene Figure 17 Platinum content: 0.8% w/w; impregnation: CH,CI, as solvent; activation: in situ 1 h under H, at 400 ‘C; reaction conditions: 0.05 gplatinumcatalyst, 30ml H,/min, 0.2~1MCH or n-C,/pulse, 2.0 bar abs. As can be seen, the most efficient catalyst is obtained starting from a homometallic (3:3:3)-cluster supported on alumina.Given their preliminary nature, these results can be considered quite promising. 7 Conclusions The dlO-metal centres of the Group X and XI elements are practically unique in forming a wide variety of heterometallic clusters. This versatility is likely due to the ease with which these centres form 16-and even 14-electron species, i.e. ML,- and ML,-fragments, respectively. This tendency is retained by their trimetallic clusters, which then readily add electron-rich atoms, or ions. Furthermore, filled HOMO orbitals of suitable energy are often present and can donate electrons to suitable Lewis acids, generating the rich synthetic and structural chemistry described above. It is expected that the Lewis base/Lewis acid idea, despite its formal nature, will be useful not only in rationalizing a wealth of synthetic chemistry, but also in predicting the existence of new classes of clusters.However, it is also clear that the platinum systems discussed above are uncommon cases, as their pre- TRIMETALLIC UNITS AS BUILDING BLOCKS IN CLUSTER CHEMISTRY-D IMHOF AND L M VENANZI formed trimetallic clusters can be directly used to build up more complex molecular structures Finally, it should not be ruled out that LCAO-MO examin-ation of other oligometallic building blocks could lead to the discovery of new families of heterometallic clusters with features of interest for solid-state chemistry and catalysis 8 References 1 D M P Mingos and T Slee, J Organomet Chem, 1990,394,679 2 R B King, Inorg Chzm Acta ,1986,116,119, R G Woolley, Chem Phys Lett, 1988, 143, 145, A Dedieu and R Hoffmann, J Am Chem SOC,1978, 100, 2074, K Wade, Adv Inorg Chem Radzo- chem ,1976, 18, 1 3 R Hoffmann, Angew Chem ,1982,10,725, T S A HorandA L C Tan, J Coord Chem , 1989,20,311, F G A Stone, Angew Chem , 1984,%, 85 4 J Chatt and P Chini, J Chem Soc(A) ,1970, 1538 5 A B Goel and S Goel, Inorg Nucl Chem Lett, 1980,16,397, S G Bott, M F Hallam, 0 J Ezomo, D M P Mingos, and I D Williams, J Chem Soc , Dalton Trans, 1988, 1461, R Bender, P Braunstein, and A T Camellini, Angew Chem ,1985,97,862, P W Frost,J A K Howard, J L Spencer,D L Spencer,D G Turner, and D Gregson, J Chem Soc , Chem Commun , 198 1, I 104, G Ferguson, B R Lloyd, and R J Puddephatt, Organometallics, 1986,5, 344 6 G Ferguson, B R Lloyd, L Monojilovic-Muir, and R J Pudde- phatt, Inorg Chem , 1986,25,4190 7 M Green, J A K Howard, M Murray, J L Spencer, and F G A Stone, J Chem Soc ,Dalton Trans, 1977, 1509 8 A Albinati, G Carturan, and A Musco, Inorg Chim Acta , 1976, 16, L3 9 A Moor, P S Pregosin, L M Venanzi, and A J Welch, Inorg Chzm Acta ,1984,85,103, R G Vranka, L F Dahl, J Chatt, and P Chini, J Am Chem Soc, 1969, 91, 1574, R F Klevtsova, E N Yurchenko, L A Glinskaya, E B Burgina, N K Eremenko, and V V Bakakin, Zh Strukt Khim, 1985, 84, P Chini, J Organomet Chem ,1980,200,37, C J Wilson, M Green, and R J Mawby, J Chem Soc ,Dalton Trans , 1974,421 10 D Imhof, K -H Dahmen, and L M Venanzi, unpublished work 1 1 A Stockhammer, K -H Dahmen, T Gerfin, V Gramlich, W Peter, and L M Venanzi, Helv Chim Acta, 1991, 74, 989 12 K -H Dahmen, D Imhof, and L M Venanzi, unpublished work, A Albinati, K-H Dahmen, A Togni, and L M Venanzi, Angew Chem ,1985,97,760, M F Hallam, D M P Mingos, T Adatia, and M McPartin, J Chem Soc Dalton Trans, 1988,355, C E Bnant, R W M Wardle, and D M P Mingos, J Organomet Chem ,1984, 267, C49 13 D Imhof, U Burckhardt, K -H Dahmen, H Ruegger, T Gerfin, and V Gramlich, Inorg Chem , 1993,32,5206 14 D M P Mingos and R W M Wardle, J Chem Soc ,Dalton Trans , 1986,73, C M Hill, D M P Mingos, H Powell, and M J Watson, J Organomet Chem , 1992,441,499 15 P Braunstein, S Freyburger, and 0 Bars, J Organomet Chem , 1988, 352, C29, S Bhaduri, K Sharma, P G Jones, and C F Erdbrugger, J Organomet Chem ,1987,329, C46, J J Bour, R P F Kanters, P P J Schlebos, W P Bosman, H Behm, P T Beurskens, and J J Steggerda, J Organomet Chem ,1987,329,405 16 A Albinati, K -H Dahmen, F Demartin, J F Forward, C J Longley,D M P Mingos,andL M Venanzi, J Chem SOC,Chem Commun ,1992,924 17 A Albinati, A Moor, P S Pregosin, and L M Venanzi, J Am Chem Soc , 1982,104,7672 18 0 J Ezomo, D M P Mingos,and I D Williams, J Chem SOC, Chem Commun, 1987, 924, Y Yamamoto, H Yamazaki, and T Sakurai, J Am Chem Soc ,1982,104,2329 19 D M P Mingos, P Oster, andD J Sherman, J Organomet Chem , 1987,320,257, N C Payne, R Ramachandran, G Schoettel, J J Vittal, and R J Puddephatt, Inorg Chem , 1991,30,4048 20 R D Feltham, G Elbaze, R Ortega, C Eck, and J Dubrawski, J Inorg Chem ,1985,24,1503, E G Mednikov, N K Eremenko, and S S Kurasov, Usp Khzm , 1985,54,671 21 E G Mednikov, N K Eremenko, S P Gubin, Y L Slovokhotov, and Y T Struchkov, Organomet Chem ,1982,239,401 22 E Leber and L M Venanzi, unpublished work 23 R Goddard, P Jolly, C Kruger, K P Schick, and G Wilke, Organometallics, 1982, 1, 1709 24 G Longoni and P Chini, J Am Chem Soc , 1976,98,7226 25 D A Nagaki, L D Lower, G Longoni, P Chini, and L F Dahl, Organometallics, 1986, 5, 1764 26 A Albinati, F Demartin, P Janser, L F Rhodes, and L M Venanzi, J Am Chem Soc, 1989,111,2115 27 A Albinati, W Klooster, and T F Koetle, unpublished work 28 A Albinati, J Eckert, P Hofmann, H Ruegger, and L M Venanzi, Inorg Chem , 1993,32,2377 29 J E Ellis, J Am Chem Soc , 1981,103,6106 30 A Berry, M L H Green, J A Bandy, and K Prout, J Chem Soc , Dalton Trans, 1991,2185 31 B R Sutherland, K Folting, W E Streib, D M Ho, J C Huffman, and K G Caulton, J Am Chem Soc , 1987,109,3489 32 A Albinati, L M Venanzi, and G Wang, Inorg Chem , 1993,32, 3660 33 P D Boyle, B J Johnson, A Buehler, and L H Pignolet, Inorg Chem ,1986,25,5 34 U Stadler and L M Venanzi, unpublished work (see also U Stadler, Ph D Thesis ETH Zurich, 1989, Nr 9067) 35 F Bachechi, J Ott, and L M Venanzi, J Am Chem Soc ,1985,107, 1760 36 T H Lemmen, J C Huffman, and K G Caulton, Angew Chem , Int Ed Engl , 1986,25,262 37 L F Rhodes, J C Huffman, and K G Caulton, J Am Chem Soc , 1985,107, 1759 38 S Muller and E Newson, Heterogenised Homogeneous Cluster Compounds for Dehydrogenation and Aromatisation Reactions, 1992, PSI Wurenhngen, Annual Report, 6

ISSN:0306-0012    

DOI:10.1039/CS9942300185

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Towards a Laboratory Strategy for the Study of Heterogeneous Catalysis in Stratospheric Ozone Depletion Martin R. S. McCoustra and Andrew B. Horn School of Chemical Sciences, University of East Anglia, Norwich, NR4 TTJ, U.K. I Introduction In 1985, Farman, reporting on a series of measurements of austral spring average stratospheric ozone concentrations made at the British Antarctic Survey (BAS) Halley Bay facility and stretching back to International Geophysical Year (1957), clearly demonstrated a decline in these levels. 1,2 The decline appeared to be accelerating in the late 1970s and early 1980s, as shown by the data reproduced in Figure 1. The trend exhibited by Farman's more recent data was soon to be corroborated by a re-analysis of historical data from the Total Ozone Mapping Spectrometer (TOMS) mounted on the Nimbus 7 satellite.2 Subsequently, a number of large-scale coordinated field campaigns such as the Airborne Antarctic Stratosphere Experi- ments (AASE I and 11) and the European Arctic Stratospheric Ozone Experiment (EASOE) have made measurements of glo- bal stratosphere ozone concentrations using a variety of comple- mentary techniques and have highlighted a small but significant decline in recent years in both the Antarctic and the Arctic.These studies have also served to enhance our understanding of the composition of the polar stratosphere and its perturbation by pollutants. Although Arctic and mid-latitude perturbations are not as drastic as the Antarctic phenomenon, depletion of stratospheric ozone levels on a global scale is now known to be occurring and is clearly of significant concern to mankind.While the chemistry of stratospheric ozone formation and destruction in the homogenous gas phase is well establi~hed,~-~ sophisticated atmospheric models, constructed in an attempt to explain the geochemical field observations and including all known homogeneous chemistry and appropriate meteorologi- cal processes of relevance, do not appear to do so. New chemistry or new meteorology must be proposed to develop these models. Since a discussion of atmospheric physics is beyond the scope of this review, we will concentrate on possible new chemistry of relevance to global stratospheric ozone depletion.Field observations in the Antarctic indicate that the enhanced ozone depletion each austral spring is a result of the ultra-low temperatures (<200 K) attained within the trapped airmass of the polar vortex and the consequent formation of Polar Stratos- Andrew Horn obtainedhis BSc. (Chemistry) and Ph.D. (Surface Science) from the University of East Anglia in 1986 and 1989, respectively, before holding a postdoctoral fellowship there studying reactions on ruthe-nium surfaces. In 1990, he moved to the Physics Depart- ment of the Loughborough Un- iversity of Technology where he undertook studies of mixed metal alloys. Since returning to EUA in 1992, he has developed techniques for the laboratory study of heterogeneous reac-tions relevant to ozone chemistry, and was appointed to a lectureship in physical chemistry in 1993.pheric Clouds (PSCS).~Such clouds consist of either small ice particles (Type I1 PSC) or solid nitric acid trihydrate (NAT) particles (Type I PSC).The new chemistry that has come to light is the catalytic regeneration of ozone depleting species from stable reservoir materials on cold solids. In global terms, the r61e of PSCs is clearly negligible as the low temperatures necessary for their formation are rarely achieved outside the polar vor- tices. However, the low temperature (190-230 K) sulfuric acid (40--60% H2S04)aerosol, always present in the stratosphere, may provide a suitable surface for such chemistry at other latitudes. The study of cold, acidic solid and liquid surfaces will therefore be of relevance in widening our understanding of stratospheric ozone depletion and it is the aim of this brief review to highlight some of the techniques that have been brought to bear on this problem.2 Background 2.1 The Chemistry of Stratospheric Ozone The chemistry of ozone formation and destruction in the homogenous gas phase is well doc~mented.~-~ Oxygen is known to photodissociate at all wavelengths shorter than 240 nm, producing both ground state O(3P)and excited O(lD) atoms. The former are produced both by the photodissociation of the A 3ZTstate populated by excitation in the weak, forbid- den Herzberg I band system in the 195 to 240 nm region, and by excitation into the B3CJ state via the well-known Schumann-Runge bands at wavelengths below 195 nm.Dr. Martin McCoustra was born in September 1962 in Falkirk, Stirlingshire. He gained his Ph.D., through a scholarship from the Carnegie Trust for the Universities of Scotland, on the study of the photodissociation dynamics of NO-containing molecules from Heriot-Watt University in 1987. In April 1988, after a short postdoctoral period with his former supervisor (Professor J. Pfab), he was appointed to a lectureship in physical chemistry at the School of Che- mical Science of the University of East Anglia. There he has developed new research inter- ersts in the study of dynamics at the gas-solid interface and in the heterogeneous aspects of the chemistry involved in stratospheric ozone loss as well as retaining an interest in free jets and their use in spectroscopy.195 I96 CHEMICAL SOCIETY REVIEWS, 1994 ~ F1l F12 00 100 200 200 400 h 5 200E E 1 8 ~ " " " ' " ~ ' ' ~ 1960 1970 1980 Figure 1 (a) Austral Spring (October) and (b) Summer (February) monthly means of total ozone (crosses) at Halley Bay, and Southern Hemisphere measurements of CFCl, (FI 1, filled circles) and CF,Cl, (F12, open circles) in p.p.t.v. over the period 1957-84. Note that the F11 and F12 amounts increase down the figure to emphasize the correlation with ozone loss. (Reproduced with permission from Nature, 1985,315,207.) The excited O(l0) atoms are produced only by the latter mechanism.These excited-state atoms are rapidly relaxed colli- sionally. The subsequent ozone formation step, Mo+o,-03 where M is a third body required to remove excess internal energy in the nascent O3molecule by collisional energy transfer, involves only the ground-state O(3P)atoms. The primary destructive mechanism is photochemical. Ozone itself is known to be highly photochemically active, absorbing light at all wavelengths below 1180 nm. It absorbs most strongly in the near-UV region below 350 nm, producing oxygen atoms and molecules in a variety of electronic states dependent upon the excitation wavelength. 03 -hv 0 + 0, As such, ozone effectively blocks all UV radiation in the range 240 to 290 nm.It is this, in combination with the blocking of shorter wavelengths by molecular oxygen absorption, that prevents any significant flux of UV radiation of wavelength shorter than 290 nm reaching the surface of the Earth. In addition, however, the reaction of free 0 atoms with ozone leading to the formation of molecular oxygen is known to occur. In 1930, the combination of the ozone-forming process and the two ozone destroying processes discussed above, 0,-hv o+o Mo+o,-03 03 -hv 0 + 0, 0+03-0,+ 0, was proposed by Chapman to be primarily responsible for the formation of the thin, stable ozone layer in the stratosphere' and now bears his name -The Chapman Cycle. However, simple atmospheric models using only this chemical mechanism with an appropriate incident solar UV flux and suitable thermal and mass balance relationships overestimate the concentration of ozone in the ozone layer by a factor of as much as five.Clearly, even in the unperturbed stratosphere, there must be additional ozone-loss mechanisms in operation. The additional loss mechanisms can in part be summarized by the simple catalytic cycle, x+o,+xo+o, xo + 0 + x + 0, This has the net effect of removing two odd oxygen species (0 and 0,)and producing two even species (two 02), O+O,+O,+O, Various catalysts X are available within the stratosphere. At altitudes above 50 km, the dominant catalysts are OH radicals and H atoms. This is the HO, Catalytic Cycle. The OH radicals are produced by reaction of O(l0) with water and methane present in the stratosphere, OH + OH"'O and the H atoms by the subsequent reaction, OH + O+H + 0, In addition to the simple HO, cycle, more complex odd oxygen-removing cycles involving HO, species are known to occur such as, OH+O- H + 0, H+O,- M HO, HO,+O- HO + 0, which has the net effect, Mo+o-02 and, OH + O3 + HO, + 0, HO, + O3 + OH + 0,+ 0, which has the overall effect of removing two ozone molecules and generating three oxygen molecules.At somewhat lower altitudes, around 40 to 45 km, the dominant catalyst is NO, giving the NO, Catalytic Cycle. NO is produced from the reaction of nitrous oxide with O(lD), O(l0) + N,O +NO + NO In the late 1960s and early 197Os, considerable concern was expressed about the potential perturbation of the NO, cycle by the injection of substantial quantities of NO directly into the stratosphere by high-flying supersonic stratospheric transport (SST)aircraft.The failure of the envisaged fleets of these aircraft to appear alleviated the concern. However, this potential HETEROGENEOUS CATALYSIS IN STRATOSPHERIC OZONE DEPLETION-M. R. S. McCOUSTRA AND A. B. HORN problem may yet return as there appears to be growing interest in the further development of such stratospheric SST vehicles. In addition, increasing N,O concentrations as a consequence of the increased use of nitrogen-based fertilizers to maintain high crop yields across the globe is of concern, since this may subsequently lead to an increase in the stratospheric NO concentration.NO from this indirect source is not as readily controlled as NO emission from SSTs and may therefore be more of a problem. At lower altitudes still, above 30 km, the catalytic cycles involving chlorine and bromine atoms, the Hal, Catalytic Cycle, become important. At low temperatures and high halogen monoxide concentrations additional steps involving the so-called Dimer Cycle, c10 + CIO -M (ClO), (ClO), -hv c1+ClOO ClOO -M c1+0, 2 x (C1+ 0, -c10 + 0,) and the reactions, C10 + BrO C1+ Br + 0, C1+ 0, + C10 + 0, Br + 0,--+ BrO + 0, both of which have the net effect of converting two molecules of ozone into three oxygen molecules, must be considered. In the clean stratosphere, the only source of C1 and Br atoms is the photolysis of the corresponding methyl halide which originates from biogenic sources.Under such conditions, the concent- rations of C10 and BrO are too low for the C10 dimerization and C10-Br0 reactions to be significant. Any additional anthropo- genic source of halogen atoms will clearly perturb these catalytic cycles. This was recognized in the early 1970s by Molina and Rowland in their consideration of the potential impact of the increasing use of chlorofluorocarbons (CFCs; CC1,Fy and C,Cl,F,) and their brominated counterparts.8 Whilst these materials are photochemically stable in the troposphere, they undergo rapid photodissociation in the presence of the shorter wavelength UV radiation present above the ozone layer.This releases a substantial and growing quantity of halogen atoms directly into the stratosphere. It is significant that the increase in atmospheric loading of CFC over recent decades almost mirrors the ozone depletion data reported by Farman as shown in Figure 1. Quite clearly, the additional halogen loading and pertur- bation of the Hal, cycle is of significance in explaining both the Antarctic ozone depletion and also more global depletion phenomena. In addition to the ozone-destroying catalytic cycles, a number of reaction cycles exist that simply interconvert 0, molecules and 0 atoms. Consequently, these do not affect the overall balance of the ozone chemistry but simply trap the destructive catalytic species in a non-destructive channel.This is exemplified by the process, NO+O,-NO, + 0, NO, -hv NO+O which has the net effect, hv 03 -0 + 0, i.e. simply photosensitizing the decomposition of 0,. In addi- tion to such null cycles, the catalytically-active species can be trapped into relatively stable and unreactive reservoir species.For the NO, cycle, up to about 10% of the NO, is trapped as the stable reservoir species N,05, produced by the sequence of reactions, NO,+O, -NO3 + 0, M NO, + NO, -KO5 These become important at night when substantial concent- rations of NO, can be built up. A further 50% of the NO, is trapped in the form of nitric acid (HONO,), OH + NO, --t HONO, which, in addition, traps a substantial proportion of the active HO, and provides a link between two of the important destruc- tive catalytic cycles. Similar links between the destructive Hal, and other destructive cycles are to be found in the reservoir species hypochlorous acid, HOCl, formed by reactions such as C1+ OH -P HOCl C10 + HO, --+ HOCl + 0, and chlorine nitrate, ClONO,, C10 + NO, -,ClONO, The principal reservoir for the Hal, cycle is the corresponding hydrogen halide, formed by reaction of the halogen atom with stratospheric methane, C1+ CH4 -+ HCl + CH, Any new chemistry which affects the stability of these reservoir species will obviously therefore alter the overall balance of the chemistry within the stratosphere.It has been demonstrated that the presence of the cold, acidic surfaces described above (PSC particles and sulfate aerosol particles) plays the major r6le in perturbing the steady state by more rapidly regenerating the destructive halogen atoms from their relatively stable reservoirs than by normal gas-phase photochemical routes.6 This is believed principally to involve the reaction HCI + CIONO, PSCs and sulfate aerosol * C1, + HONO, which is thought to occur in two steps, ClONO, + H,O PSCs and sulfate aerosol HOCl + HONOz HC1+ HOCl PSCs and sulfate aerosol -C1, + H,O generating molecular chlorine, which is rapidly photolysed to yield free chlorine atoms.Additionally, this process traps nitric acid within a solid or liquid matrix from which it is unlikely to escape and therefore will subsequently be unable to participate in the normal photoinitiated gas-phase stratospheric chemistry of nitric acid.This effectively reduces the overall concentration of trapping partners (HO and NO,) for the active C1 and C10 in the stratosphere, enhancing the effect of the destructive catalytic cycles involving these species. 2.2 Databases of Relevant Chemical Information Many of the important chemical species mentioned in the previous section have been identified by field studies as being present in the stratosphere and have been characterized in the laboratory by a number of techniques in the course of other diverse investigations. We are therefore already provided with a large database of chemical and physical information which may be exploited.Perhaps the largest database that exists is that of spectroscopic information. Techniques such as infrared, micro- wave, and UV/visible spectroscopy are frequently applied as a matter of course during chemical experiments in virtually every laboratory. Whilst the conditions under which these spectra have been obtained are generally not directly relevant to the chemistry of the stratosphere, they do serve to provide an extensive reference library for the assignment of spectra obtained in laboratory studies of ozone loss in the presence of cold surfaces. 2.2.1 Matrix Isolation Techniques The low-temperature chemistry of the stratosphere involves many species which are extremely unstable at room tempera- tures.One of the most significant techniques for the stabilization of reactive compounds and photolytic intermediates for spectro- scopic study is matrix isolation (MI). A compound (or its precursor) is mixed in the gas phase with a supporting gas such as argon, nitrogen, or oxygen and is then condensed on a cold window. Typical matrix ratios are in the range 1:100to 1: 10,000, depending upon the nature of the compound and its tendency to aggregate. Window temperatures of between 4.2 and 20 K are easily obtained using a helium cryostat. At these ultra-low temperatures and with high dilution ratios, the trapped mole- cules are sufficiently far apart to be unable to react with each other. By decreasing the matrix ratio, we can also adjust the conditions to investigate the formation of dimers and higher aggregates. The species trapped in the matrix can be studied using a variety of methods including infrared, UVIvisible, and Raman spectroscopy .9 As an example of potential relevance, the formation and stability of HCl:H,O aggregates has been studied extensively in low temperature matrices.O Below 50 K, the number of water molecules available per HCl unit determines whether or not proton transfer occurs to produce an ionic hydrate: the limit appears to be three water molecules. Warming above‘60 K results in the formation of amorphous ionic hydrates for all water:HCl ratios. Low-temperature matrices therefore appear to assist in the stabilization of an unstable molecular complex. Extrapolating such measurements to higher temperatures, the adsorption of HCl on ice at stratospheric temperatures will initially involve the arrival of an intact molecule of HCl at the surface, followed by the formation of a weak bond to the surface and subsequent dissociative adsorption.Whilst the resulting surface species can be analysed using the spectroscopic methods described below, the weak initial interaction occurs on too short a timescale for this approach to be useful. Experimentally, MI studies at low temperatures have shown that the interaction is a weak hydrogen bond between the HCl molecule and the oxygen atom of the surface water.” Theoretically, and in excellent agreement with detailed experiments, Clary and co-workers’ 2713 have modelled the interaction of HCl with an ice crystal by the use of classical trajectory calculations, employing both ab initio potentials for the interactions of HCl with a single H,O molecule in the ice surface and molecular dynamics potentials for the behaviour of the ice, assuming both dissociative and non-dissociative adsorption mechanisms.Their results suggest that the molecular interaction is very weak, leading to a low equili- brium surface concentration of physically adsorbed molecular HCl. 2.2.2 Condensed Phase Vibrational Spectra Classical methods for the characterization of molecular vib- rations, such as Raman and infrared spectroscopy, provide perhaps the most useful information in the context of PSC chemistry. Vibrational spectra are extremely sensitive to both inter- and intra-molecular bonding in the condensed phase, and can provide a great deal of information about molecule-mole- cule and molecule-surface interactions.There are in the litera- ture many studies, some dating back to the early days of infrared and Raman measurements, of the structure and bonding of important stratospheric species such as HCl, HBr, and HONOz with water. These studies cover a broad range of conditions, from gas-phase molecular complexes to solid films of ionized polyhydrates14J 2.2.3 Photolysis and UV/ Visible Studies As discussed above, short wavelength UV photolysis plays an important role in the homogeneous chemistry of the strato- sphere. Determination of the yields of products from such photolyses and the variation of these yields with wavelength are CHEMICAL SOCIETY REVIEWS, 1994 crucial to our understanding of photo-initiated chemistry.Whilst such measurements are routine in the gas phase, analo- gous measurements on the potentially perturbed species adsorbed on cold surfaces have yet to be made in detail and are likely to be highly significant. Photochemical studies in low- temperature matrices have gone some way to providing this information and have indicated that alternative reaction chan- nels may become available to species trapped at or near a surface. 2.2.4 Crystallographic Met hods X-Ray crystallography of the structure of ionic hydrate single crystals can provide information about the nature of the hydro- gen bonding in H30+ ions.” This information can be used in the spectroscopic identification of a number of acid polyhyd- rates, for which the infrared and Raman spectra indicate a number of 0-H bond lerigths and environments. Additionally, the structure of adsorbates such as N,Os and N,O, in the solid state can be used to interpret trends in their surface packing and ordering.2.2.5 Other Techniques Other analytical techniques can provide useful information for the interpretation and analysis of iceladsorbate chemistry. Weak hydrogen-bonding interactions have been studied extensi- vely using methods such as microwave (rotational) spectroscopy to investigate the bonding to water in molecular complexes18 and NMR to study hydroxonium ion structure.Molecular beam methods for the study of molecule-ion interactions and laser- based spectroscopy for the investigation of state-resolved mole- cular energy partitioning also provides insights into the internal energy distribution of atmospheric species and its possible influence on reaction mechanisms. 3 Theoretical Modelling of Surface Processes of Relevance to Stratospheric Particles Interpretation of results in science frequently involves the use of models. Experimental observations are compared against theor- etical predictions based upon a particular model, with discre- pancies being used to further refine the model until a better fit is achieved. This phenomenological approach has been used very effectively in the prediction of the homogeneous gas-phase chemistry of the stratosphere, and its logical progression into heterogeneous processes is described below.Additionally, the inclusion of accurate heterogeneous chemistry models into global stratospheric ozone prediction programs have dramati- cally improved their accuracy in predicting ozone trends. 3.1 Definition of Terms In order for information gathered in separate experiments to be consistent, it is necessary to define the terms in which adsorp- tion, reaction, and release of product can be quantified. Many studies have used terms to express the amount of material which either remains adsorbed on the surface, the amount of material removed from the gas flow, describing such processes using ‘mass accommodation’, ‘uptake’, and ‘sticking’ coefficients.Such terms frequently have established definitions in classical surface science, although the quantities actually measured usually depend upon the nature of the experiment. The follow- ing definitions may be of use. The accommodation coeficient simply reflects whether a molecule can be accomodated by the surface or not.19 For a molecule in collision with a surface with which it does not interact, then the accommodation coefficient is zero and the molecule rebounds with its original energy: conse- quently, no adsorption has occurred. If an attractive force exists between the surface and the molecule such that the molecule is accommodated at the surface for sufficient time for energy exchange to occur, then the accommodation coefficient is unity HETEROGENEOUS CATALYSIS IN STRATOSPHERIC OZONE DEPLETION-M.R. S. McCOUSTRA AND A. B. HORN and either physical or chemical adsorption has occurred. The sticking coeficient is usually defined for a specific experiment. Generally, it is based upon the ratio of the number of molecules which stick to the surface and equilibrate per unit time and the total number of molecules incident per unit time. In a system where the rate of desorption is low (i.e.most of the equilibrated molecules remain attached to the surface), then the sticking coefficient can be determined using the technique of King and Wells.20 The partial pressure of the reagent of interest in the reaction chamber is measured with an unreactive flag between the gas jet (molecular beam) and the sample.The partial pressure is then monitored as the flag is withdrawn. The adsorption of gas results in a second pressure which is lower than the first, and the ratio can be related to the sticking probability. This technique works best for chemisorption processes with a relatively high sticking probability. The sticking coefficient is often a function of coverage in a chemisorption system because the number of molecules adsorbing decreases as the number of available sites falls. Other effects may be observed if the rate of desorption is high, in which case the sticking coefficient may be high with an equilibrium surface concentration close to zero.The extent of a reaction is often quantified using an uptake coeficient or a reaction probability. The existence of these two extremes for the sticking coefficient can cause confusion when the detailed nature of the surface processes in the system under study is not known. For example, in a system where the sticking coefficient is high and the rate of desorption is low, the uptake coefficient would be high at first and would decrease with exposure, leading to a correct assumption about the nature of the sticking coefficient. However, at the other extreme with a high sticking coefficient and a low equilibrium concentration, the uptake coefficient would be observed to be low for all exposures and the sticking coefficient may incorrectly be inferred to be low.This may well be the case, but if a small surface concentration of one reactant is required for another reaction channel to become favourable, such as appears to be the case for the HCl/ice system, then incorrect assumptions may be made about the surface processes. In other words, the use of such macroscopic parameters to describe the amount of reaction which is observed to occur should be treated with caution because they do not take into account any of the microscopic details of the reaction scheme. 3.2 Processes Involved in Gas/Surface/Bulk Chemistry The mechanisms involved in the transport of reactants from the gas-phase to the surface, the surface adsorption/desorption equilibria, and the subsequent desorption of reaction products and their transport away from the surface are often considered to be the most important factors in heterogeneous catalysts. An additional, equally important factor to be considered in the case of reactions which occur between adsorbates and a surface which may be mobile, and which consume some of the surface material, is the rate at which these surface species can be replaced from either other surface sites or from the bulk.This behaviour necessarily also includes the transport of reaction products away from the reaction site. Figure 2 shows a schematic of all the processes to be considered. These diffusion processes can be divided into a number of types. Lateral diffusion may occur along a surface where residence times are long.Surface/bulk diffusion is expected to play an important role in stratospheric heterogeneous catalysis, particularly in liquid and quasi-liquid systems such as aerosols. In more solid systems such as micro- crystalline ice and NAT, where the surface is rough and pitted, pore diffusion and grain boundary diffusion may also be important. The development of a comprehensive model including all of the processes observed to occur on and in PSC particles has been addressed by Tabazadeh and Turco.21 Although this introduces a much greater degree of complexity into the overall models for global stratospheric ozone, and increases computation time, the results already appear to vindicate this approach. e 0 lo1 Bulk\I Figure 2 Gas-surface interactions in heterogeneous chemistry model.Molecular species are indicated by circles. The processes depicted are (1) molecular adsorption, (2) molecular desorption, (3) surface diffu- sion, (4) pore diffusion, (5) bulk crystalline diffusion, (6) surface activation and reaction, (7) reactive product desorption, (8) secondary reaction, and (9) surface poisoning. (Reproduced with permission from J. Geophys. Res., 1993,98, 12727.) 4 Composition and Characterization of Stratospheric Substrates Although the approximate composition of particles found in the stratosphere is known, the detailed nature is the subject of some conjecture. We do however have an accurate idea of the pres- sure, temperature, and component partial pressure conditions in the stratosphere.Using this information, we can exploit widely available thermodynamic information such as phase diagrams and vapour pressure/composition data to extrapolate likely PSC composition. This approach can also be used to estimate and measure the extent of acid absorption (HCl, HONO, etc.) into bulk particles by measuring the vapour pressure of a particular component over the sample. Since no experiment has yet been devised for either the production of polar stratospheric clouds in the laboratory or in situ surface spectroscopic measurements in the stratosphere, it is necessary to devise a technique to mimic the relevant aspects of the physics and chemistry of such particles in the laboratory. For processes which are primarily concerned with gas/surface inter- actions, studies of heterogeneous stratospheric chemistry are most easily performed using apparatus in which only the surface functionality of PSC particles is simulated.As in the use of all model, idealized systems, the nature of the mimic will obviuusly affect the experimental results. As discussed in the previous section, the nature of the PSC particle composition and struc- ture can be inferred from phase diagrams of the relevant constituents and the surface functionality can be inferred from what is known about the physical chemistry of ice and solid acid polyhydrates. The task which remains is to develop experimen- tal apparatus to generate and maintain PSC mimics in a controlled laboratory environment, within which meaningful measurements can be made.4.1 Stratospheric Particle Compositions from Phase Diagrams There is a great deal of speculation as to the nature of the formation of stratospheric particles from a wide variety of stratospheric species. The condensation at different tempera- tures of various sulfuric acid/nitric acid/water mixtures as solids has been suggested as the most likely scenario. This rationale is supported by a number of studies of the phase diagrams of binary systems such as HON0,/H20 and H,SO,/H,O and ternary systems such as HONO,/H,SO,/H,O. Perhaps the most convincing evidence for this is presented in a recent report by Molina et a1.2z 4.2 Vapour Pressure Measurements of LiquidlSolid Systems The uptake of HCl by ice, NAT, and sulfuric acid aerosol has also been studied using vapour pressure measurements and related to bulk concentrations by the Gibbs-Duhem equation.23 A number of conclusions can be drawn from trends in vapour pressure/composition measurements, with the most noticeable being that the bulk solubility of HC1 varies quite widely in all of these systems, although it is never particularly high.NAT appears to have a significantly higher affinity for HCl than does ice. 4.3 Spectroscopic Characterization of Water Ices Perhaps the simplest system upon which representative measure- ments can be made in the laboratory is that of type I1 PSCs, which consist of ice particles of between 1.0 and 10 pm dia- meter.6 Two possibilities exist for generating ice surfaces: thin films and particles.The literature is rich with examples of spectroscopic and crystallographic studies of the structure of many of the phases of solid water. Small clusters of ice can be generated in an infrared cell using the method of De~lin.,~ Ice clusters are formed by introducing a mixture of water and either argon or nitrogen into a pre-cooled optical cell, mounted in the sample compartment of a spectro- meter. At 77 K, aggregation of the water into small clusters (lo2-lo3 molecules) occurs whilst the carrier gas remains unaffected. These clusters remain stable and in suspension for several hours. Infrared spectra of these clusters has led to the identification of a particular feature that has been assigned to the vibration of surface water molecules, relatively unaffected by bulk hydrogen bonding.This ‘dangling’ OH bond is particularly prominent because of the small bu1k:surface water molecule ratio. The generation of thin solid films of ice at low temperature is of more widespread applicability in many of the techniques discussed below, and which we have adopted in our own studies25926 We have concentrated on producing a well charac- terized, easily reproducible substrate for such studies. The experimental apparatus consists of a circular stainless steel vacuum system, capable of evacuation to pressures of less than 1 x torr. Ice films are grown on a thermostatted flat gold foil, upon which infrared spectroscopic measurements can be made simultaneously. The substrate can be held at any tempera- ture in the range 80-500 K with a stability of better than 1 K.The substrate, held at 80 K, is exposed to gas-phase water at a pressure of 1 x torr for 60 seconds from a precision leak valve. This generates an amorphous film of approximately 50 nm thickness. A surface ‘dangling’ OH bond feature similar to that observed in small clusters can be discerned in the infrared spectrum. When this film is then heated in vacuo, the partial pressure of water in the chamber starts to increase, as measured by a quadrupole mass spectrometer. This water desorption corresponds to the low-temperature shoulder in the thermal desorption spectrum (Figure 3, inset), which is due to an ordering of the amorphous film into a polycrystalline structure.The substrate temperature is then stabilized at this value until the desorption ceases. The substrate can then be cooled to the required temperature. Figure 3 shows a typical infrared spec- trum of an ice film grown by this method. Ice films for use in flow tube experiments are normally deposited by flowing the carrier gas (generally He) through water and passing the resultant mixture over the cold walls of the reaction chamber. There is some debate in the literature con- cerning the physical state and surface area of ice films deposited in this manner. Many early studies have used the geometric surface area of the film, but recent evidence suggests that, because of the highly porous nature of the layer, the actual area may be as high as fifty times this val~e.~~?*~ The reasonable degree of reproducibility observed using this method, however, suggests that such films are a valid substrate for measurement.CHEMICAL SOCIETY REVIEWS, 1994 0.028 0.026 0.024 0.022 0.020 0.018 $ 0.016 c2 0.014 09 0.012 0.010 0.008 0.006 0.004 0.002 0.000 41 10 3500 3000 2500 2000 1500 1000 Wavenumbers Figure 3 RAIR spectrum of a typical thin ice layer at 80 K, formed by low temperature water deposition from the gas phase onto a cold gold substrate and annealing to the low temperature desorption shoulder (inset TDS). (Reproduced with permission from J.Phys. Chem., 1994,98,946.) 4.4 Spectroscopic Characterization of Nitric Acid Hydrates Examination of the HONO,/H,O phase diagrams under strato- spheric conditions suggests that type I PSC particles are com- posed of nitric acid trihydrate (NAT), as discussed pre- viously. A number of authors have spectroscopically character- ized thin films of nitric acid hydrates, condensed from the vapour above aqueous nitric acid solutions of various compo- sition, and have identified three distinct hydrates; nitric acid monohydrate (NAM), dihydrate (NAD), and NAT.29 Qn annealing to 180 K, typical of the polar stratosphere in winter, the lower hydrates appear to convert into the stable trihydrate. Recent evidence suggests that there are two structural modifica- tions of NAT, a-NAT and /I-NAT, and that p-NAT is the more stable high temperature form.Transitions between the phases appear to be irre~ersible.~~ The simulation of sulfuric acid aerosol surfaces has recently been addressed by a number of groups. Molina et al. have investigated the inclusion of HONO, into aerosols of varying H,SO, composition and subsequent crystallization.22 The gene- ration of such surfaces in flow tubes and Knudsen cells has also been performed. 5 Probing Stratospheric Reactions Once a mimic for a particular stratospheric substrate has been generated, a technique for the investigation of its surface chemistry needs to be devised. The popular methods available for this can be divided into two types. Non-surface-specific techniques generally tend to probe the gaseous composition over a substrate or ‘downstream’ from it in a gas flow.Surface- specific techniques are generally spectroscopic probes which have been adapted to be surface sensitive rather than surface specific in the sense of classical surface analysis. 5.1 Non-surface-specific Techniques All the techniques currently in use for the investigation of HETEROGENEOUS CATALYSIS IN STRATOSPHERIC OZONE DEPLETION-M. R. S. McCOUSTRA AND A. B. HORN 201 heterogeneous stratospheric chemistry can roughly be divided into two groups: those which use a surface-sensitive probe, and those which do not. The use of mass spectrometry as a detection method in fast-flow systems we class as a non-surface method, whereas its use in thermal desorption spectroscopy we class as a surface-sensitive method.In this section, the two major non- surface-specific techniques using flow tubes and Knudsen cells are reviewed. These techniques rely on the use of models for the Reservior Gas inlet &( chamber pQMsinterpretation of data; consequently, the quality of the result depends upon the accuracy of the model. 5.1.I Flow Tube Experiments Figure 4, reproduced from reference 31, shows a schematic diagram of a modern flow tube experiment. The flowing gas composition is measured using a differentially pumped quadru- pole mass spectrometer (QMS). Such reactors are generally constructed from glass. The principle of operation is a simple modification of the chemical gas kinetic technique.32 A carrier gas, usually an inert gas such as helium, is passed through the yeatin; 1,coils reaction cell.A high flow rate is maintained with a large capacity vacuum pump. The substrate of interest is deposited onto the walls of the cell, which can be cooled to the required tempera- ture. Reactant gases are introduced into the carrier flow at the upstream end of the cell via a moveable injector and the composition of the eluted gas is measured by the QMS at the downstream end. The detected composition is compared to that measured in the absence of a substrate to yield information about the reaction probability on the substrate. ROTARY PUMP PREAM Figure 4 Schematic diagram of a typical flow tube system.Gases are introduced via the moveable injector and the composition of the eluted gas is measured using the differentially pumped quadrupole mass spectrometer. (Reproduced with permission from J. Phys. Chem., 1993,97,7779.) Flow tube measurements provide the bulk of laboratory studies of heterogeneous stratospheric chemistry to date. The interactions of single reactants such as ClONO,, C10, N205, HC1, HBr, HF, and HOCl with ice and various nitric acid hydrates have been studied under a variety of pressure and temperature regimes applicable to the stratosphere. The infor- mation obtained using this technique is useful for the determi- nation of reaction probabilities and reaction rates. Reactions such as ClONO, + HC1+ Cl,(g) + HONO,(s) have also been studied on ice and NAT.5.1.2 Knudsen Cell Partial Pressure Methods Uptake measurements can be made using a static reaction chamber comprising two separate compartments, separated by a valve (Figure 5).33 In one of the compartments, a cold reaction surface is deposited. The reactive mixture of interest is held in the second compartment. The composition of this mixture is mea- sured using a differentially pumped QMS. When the valve is opened, the gas mixture comes into contact with the cold surface Temperature Cooled ethanol bath sensor Figure 5 Schematic diagram of a typical Knudsen cell gas uptake system. The composition of the gas in the reservoir chamber is measured using the quadrupole mass spectrometer as a function of time after the separating valve has been opened.(Reproduced with permission from Geophys.Res.Lett., 1993,20,1191.) and the evolution of the reaction between the gas components and the surface can be monitored using the QMS. An example of the use of such a chamber is in the evaluation of reaction potentiality. It has been suggested that the reaction of NO and NO2 with sulfuric acid aerosol may lead to the formation of nitrosylsulfuric acid (NOHSO,), with a potential for further chlorine reservoir reacti~ation.~~ Knudsen cell stu- dies however, indicate that the reaction probability for this reaction is less than 5 x thereby eliminating it as an important process. 5.2 Surface-specific Methods By far the most informative method for the investigation of PSC surface chemistry is spectroscopic probing.Common techniques such as UV/visible (electronic), infrared, and Raman (vibratio- nal) spectroscopy can be adapted to probe the surface and to provide direct information about the electronic states of adsor- bates, details of the bonding environments, and direct surface fragment identification. Surface sensitivity can be attained by either reducing the amount of material present whilst retaining the surface area; e.g. by the use of ultrathin films, or by adapting the optics of the system using waveguides or underlying metal substrates.34 5.2.1 In situ Surface Techniques A number of spectroscopic techniques are currently in use for the study of ice and NAT surface chemistry using thin solid films.Figure 6 shows a schematic of ouf:experimental appara- tus. This consists of a cylindrical stainless steel chamber pumped by an oil-vapour diffusion pump. The vacuum system is opti- cally coupled to a Bio-Rad FT infrared spectrometer (model FTS60A/896) and the IR beam is focused onto the substrate at an angle of 75" to the surface normal. The reflected beam is detected using a HgCdTe photoconductive detector cooled to 77 K. The substrate upon which ice films are condensed consists of a flat foil of gold or nickel mounted between a pair of tungsten supports. These supports are in thermal contact with a liquid nitrogen reservoir. The temperature is controlled by balancing the reservoir cooling against the resistively heated foil.Sample gases are dosed onto the vacuum chamber from one of three independent gas reservoirs to ensure sample purity. Gas samples CHEMICAL SOCIETY REVIEWS, 1994 UHV chamber From optical bench Dosing lines Figure 6 Schematic diagram of the UEA RAIRS/ultra-high vacuum system. Ice films are deposited upon the cooled gold substrate. Reactions between impinging gases and the ice film are monitored using the infrared spectroscopic probe. inside the chamber are directed at the sample position by glass guide tubes. The spectra recorded in this configuration are reflection-absorption infrared (RAIR) spectra. An alternative reflection technique, used by Schrems et al.,29 utilises a normal incidence reflection from a cold substrate.This has the advantage of a more convenient optical arrangement for use with more conventional cryostat cold fingers. This method, which requires a much thicker film, can be used to provide complementary information to grazing incidence reflection. In a grazing incidence RAIRS experiment, only vibrational modes polarized perpendicular to the metal surface are detected, by virtue of the metal surface selection rule. The origin of this phenomenon is beyond the scope of this review, but a compre- hensive treatment can be found in reference 35. In the normal incidence RAIR configuration, the electric field vector of the incident radiation is confined to the surface plane. Species close to the metal surface cannot be detected because of the metal surface selection rule, hence the requirement for thick films, for which parallel-polarized modes can be obtained.Transmission infrared spectroscopic studies of thin films have been performed using thin, transparent optical windows as a substrate, in a similar manner to matrix isolation spectroscopy. Although this method samples both the bulk and the surface of the film simultaneously and is relatively insensitive to monolayer quantities, it has the benefit of experimental simplicity. Good quality spectra can be obtained for reasonably thick surface layers on thin substrates, with several examples in the literature of its application to the characterization of NAT films and HCI uptake.36 5.2.2 Applications to Surface Reactions: HCl and N,O, on Ice Our own studies have concentrated on the infrared spectro- scopic study of the adsorption and reaction of stratospheric species such as HCl, Cl,, HONO,, and N205 on thin film ice surfaces.Initial experiments in which the adsorption of HCl on ice at 100 K was monitored indicated that there is little or no molecular adsorption at this temperature. This is in agreement with the results of MI studies and theoretical predictions. What is observed is the formation of a thin surface film of an amorphous ionic hydrate. Figure 7 shows the result of exposing a D,O ice film at 100 K to a pressure of 1 x lo-' torr DCl for 10 seconds. The positive absorbance bands are characteristic of the hydroxonium ion, D,O +, whereas the negative absorbance features are due to both the removal and conversion of water from the original ice film.Studies of the uptake of DCl as a function of exposure (Figure 8) suggest that the adsorption increases to a saturation level, due to the depletion of surface water to such a level that no further dissociative adsorption can occur. Studies of the rate of uptake as a function of temperature lo8i gained DsO'CI-98-\ /I96-2800 2600 2400 2200 2000 180096g8\ I I I I I 2800 2600 2400 2200 2000 1800 Wavenu mberkm-' Figure 7 RAIR difference spectrum of DCI uptake by a D20 ice film at 90 K. The spectrum shows upward bands due to the loss of water from the film and downward bands from the formation of D,O+.(The behaviour of HCl and H20 ice at 90 K is qualitatively similar.) (Reproduced with permission from J.Chem. SOC.,Faraday Trans., 1992, 88,1077.) 200 1801 8 8 c 160j 3 .8 4 1404 n '""1 . 401m20 0 50 100 150 200 250 DCI Exposure/arb. units Figure 8 Intensity of the D30+peak from Figure 7 as a function of exposure to DC1. It can clearly be seen at this temperature that the uptake falls off as exposure increases, indicating a saturation of the ice film. This is believed to be due to the consumption of all the available surface water. (The behaviour of HCl and H,O ice at 90 K is qualitatively similar.) indicate that the adsorption is likely to be controlled by a precursor-mediated scheme such as that shown below. HCl(g) =HCl(ads) -,H,O+CI -(s) The reaction requires a pre-equilibrium of physically adsorbed HCl with a sufficiently long surface residence time for reaction into the ionic hydrate to occur.This process does not require a high instantaneous surface concentration, and indeed, the molecular, physically adsorbed state cannot be detected by reflection absorption infrared spectroscopy, despite its sub- monolayer sensitivity. This is consistent with calculations by Clary and co-workers. 2~13 The reaction of increasing doses of gas-phase covalent N,O, with a thin ice film at 140 K is shown in Figure 9. At low doses, the first observation is a water loss, evidenced by the negative bands at ca. 3400, 1600, and 800 cm-I. Concurrently, strong broad features centred on ca.2900 and 1750 cm-I and sharper bands at ca. 1450, 1300, 1040, and 810cm-l start to grow in, labelled species A and B respectively. Increasing doses result in HETEROGENEOUS CATALYSIS IN STRATOSPHERIC OZONE DEPLETION-M. R. S. McCOUSTRA AND A. B. HORN t I 1 I I I I 4000 3500 30002500200015001000 Wavenumbers Figure 9 RAIR difference spectra of successive doses of N,05 onto an ice film at 140 K. The spectra show, from the bottom, the build-up of amorphous H,O +NO;, the onset of molecular nitric acid and nitro- nium ions, and the eventual domination of the spectrum by a mixture of H30+,NO;, NO;, and HONO,. (Reproduced with permission from J. Phys. Chem., 1994,98,946.) further water loss and new features at 2380 cm-(speciesC) and 1682,13 18,965, and 778 cm-(species D).By comparison with literature spectra of hydrated acid ices,14 species A can be identified as being a hydroxonium ion. Species B is assigned to a solvated nitrate ion for which the bands at ca. 1450 and 1300 cm-l are the two components of the normally degenerate asymmetric stretching mode, where the degeneracy is lifted due to the low symmetry of the planar nitrate environment. The weak feature at 1040cm-' is the formally forbidden symmetric stretch which is often observed in disordered films. The weak feature at 810 cm-is the NO; bending mode. The presence of species A and B are evidence for the formation of an amorphous nitric acid hydrate, for which we propose the reaction scheme: The number of water molecules, n, involved in the reaction will be determined principally by the preferred stoichiometry of the hydrate formed, i.e.NAT, NAM, etc. and the availability of water in the region of reaction. Studies of the formation of nitric acid hydrates as a function of temperature suggest that the monohydrate and the dihydrate may be favoured at 140 K. However, we are unable unequivocally to assign our spectra as anything other than an amorphous hydrate. As the amount of surface water available for reaction is reduced on further dosing of N,O,, a change occurs with the simultaneous growth of species C and D. By comparison-with the solid N205film spectra, species C can be assigned to linear NO;. The remaining bands (species D) are due to molecular nitric acid, confirmed by comparison with matrix isolated nitric acid at 4.2 K.37 The formation of molecular nitric acid as the amount of available surface water decreases is not surprising. From the point at which the formation of the amorphous nitric acid hydrate reaches a saturation coverage, further reaction requires that either diffusion of the water to the surface occurs to release more water or that the reaction scheme must change.Our spectra suggest that the latter is the case. Since the bands due to the nitronium ion and molecular nitric acid grow in together, it seems likely that they are formed concurrently from the partial hydrolysis of gas phase N,O, as it impinges upon the surface. Considering the likely surface species in the saturated layer (NO:, H,O+, HONO, etc.) and the scarcity of water, it seems likely that the hydrolysis occurs by reaction of N,05 not with water but rather with the hydroxonium ion.The products from such a reaction would be HONO,, water and NO;. N,05 + H30++HONO, + NO: + HZO This generation of more surface water, able to undergo further reaction with N,O, to produce more partially hydrated nitric acid and therefore more nitrate ions, accounts for the growth of the nitrate bands during the formation of molecular nitric acid. Annealing this composite film to 160 K results in the produc- tion of a thicker film of molecular nitric acid and a small amount of nitric acid hydrate. This occurs because the increase in mobility of the ice film allows water to diffuse.As more water becomes available for reaction with the molecular nitric acid and NO;, further changes occur. NO; and water are both consumed to produce more nitric acid and a proton. The proton becomes hydrated and this can be seen in the growth of the broad H30f absorption in the difference spectrum (Figure 9). The charge associated with this ion is balanced by the negative charge on the nitrate ion originally associated with the NO;, thus maintaining overall neutrality. The fact that there is only a slight increase in the nitrate ion absorptions indicates that the rate of solvation of molecular nitric acid is slower than that for reaction of the nitronium ion. Annealing above 180 K results in the loss by evaporation of this film.To summarize, the reaction of gas-phase N,OS with a thin ice film occurs through a complex interplay of ionic chemistry involving nitrate, nitronium, and hydroxonium species, for which we suggest the following scheme: N,05 + 3HzO -P 2NO; + 2H3O+ (1) H30++ N,O, -+ HONO, + NO: + H,O (2) 2H,O + NO; + HONO, + H30+ (3) HONO, + H,O -,NO; + H30 (4) overall, 2N,0S + 6H,O + 4NO; + 4H30+ During the initial uptake, reaction 1 prevails until saturation occurs. Beyond this limit, dictated by the water availability which prevents complete hydration of NO: and nitric acid and effectively eliminates steps 3 and 4, there is a competition between nitrogen-containing entities for the surface water.NzOSarriving from the gas phase forms molecular nitric acid and NO; from the hydrolysis of N205 on the hydroxonium-rich surface, 2. The charge balance is maintained by the presence of the NO; counter-ion released by the reaction of its accompany- ing hydroxonium ion. On annealing, more water becomes available as the ice film becomes mobile and is consumed by the above reaction scheme to reduce the concentration of NO: which produces more molecular nitric acid and hydroxonium species. There is a competition between the NO; and the molecular nitric acid for the water in steps 3 and 4, and our observation that the NO; band is depleted and more HONO, is formed suggests that 3 occurs in preference to 4. The pH of the surface layer is likely to control the relative contributions of these four surface reactions and further investigations are in progress.6 New Directions There are a large number of classical surface science techniques that are of possible application to the study of stratospheric chemistry.34 Although many electron-based spectroscopies such as X-ray photoelectron spectroscopy, Auger electron spec- troscopy, and ultraviolet spectroscopy are extremely surface- specific and sensitive, they are limited in their application to higher pressure systems by the short inelastic mean free path of electrons in high pressure gases. However, these techniques may be rendered useful by the application of differential pumping and cryogenic trapping. Nuclear magnetic resonance, which can be extremely sensitive to proton environments, may also be of use in the determination of the mobility of acidic protons and their role in bulk PSC structure.An example of the recent application of a classical surface analysis to the study of PSC chemistry which shows great promise is the use of secondary ion mass spectrometry (SIMS).38 In this technique, inert ions (Ar, Ne) are fired at the surface. The sputtered material is analysed using mass spectro- metry. Modern instrumentation makes it possible to observe negative ions and neutrals as well as positive species. This technique can be used in two regimes. Static SIMS uses a low power beam to probe the surface layer with little disruption. Dynamic SIMS uses higher powers to ablate the surface, and can be used to extract bulk concentration gradients.This is currently being applied to the determination of chloride ion diffusion in ice by Vickerman et al. Recent trends in ozone depletion in mid-latitudes have been linked to rising concentrations of stratospheric particulates, particularly in the Northern hemisphere. There is mounting evidence that low-temperature heterogeneous reactions can occur on the surface of sulfuric acid aerosol droplets and volcanogenic dust particles in addition to PSC particles. New techniques are required to investigate the surface chemistry of such systems and some of the techniques discussed above will find application in this area. Clearly, there is scope for the use of other surface analytical techniques to be applied to this problem.Acknowledgements. Support from the SERC Atmospheric Chemistry Initiative and the Commission of the European Communities STEP program (contract CT90-007 1) is gratefully acknowledged. We would also like to thank our co-workers M. A. Chesters, J. R. Sodeau, T. Koch, and S. F. Banham. 7 References 1 J. C. Fannan, B. G. Gardner, and J. D. Franklin, Nature, 1985,315, 207. 2 R. S.Stolarski, Sci. Am., 1988,258,20. 3 R. P. Wayne, ‘Chemistry of Atmospheres’, 2nd Edn., Oxford University Press, Oxford, 1991. 4 F. S. Rowland, Annu. Rev. Phys. Chem., 1991,42,731. 5 H. S.Johnston, Annu. Rev. Phys. Chem., 1992,43, 1. 6 0.B. Toon and R. P. Turco, Sci. Am., June 1991. 7 S. Chapman, Memoirs Roy.Meteorological SOC., 1930,3, 103. CHEMICAL SOCIETY REVIEWS, 1994 8 M. J. Molina and F. S. Rowland, Nature, 1974,249, 810. 9 ‘Spectroscopy of Matrix Isolated Species’, ed. R. J. H. Clark and R. E. Hester, Advances in Spectroscopy, Vol. 17, John Wiley, Chichester, 1989. 10 C. Amirand and D. Maillard, J. Mol. Struct., 1988, 176, 181. 11 L. Delzeit, B. Rowland, and J. P. Devlin, J. Phys. Chem., 1993,97, 10312. 12 G.-J. Kroes and D. C. Clary, J. Phys. Chem., 1992,96, 7079. 13 S.H. Robertson, G.-J. Kroes, and D. C. Clary, J. Phys. Chern., in press. 14 A. S.Gilbert and N. Sheppard, J. Chem. SOC., Faraday Trans., 1973, 69, 1628. 15 C. C. Ferriso and D. F. Hornig, J. Chem. Phys., 1955,23, 1464. 16 J. R. Sodeau, in ‘Low Temperature Chemistry of the Atmosphere’, ed.A. J. Barnes, G. LeBras, G. K. Moortgat, and J. R. Sodeau, Springer-Verlag, Berlin, 1994. 17 R. G. Delaplane, I. Taesler, and I. Olovsson, Acta Cryst., 1975, B31, 1486. 18 A. C. Legon and D. J. Millen, Chem. SOC. Rev., 1992,21,71, 19 A. W. Adamson, ‘Physical Chemistry of Surfaces’, 4th Edn., Wiley, New York, 1982. 20 D. A. King and M. G. Wells, Surf. Sci., 1972, 29,454. 21 A. Tabazadeh and R. P. Turco, J. Geophys. Res., 1993,98, 12727. 22 M. J. Molina, R. Zhang, P. J. Wooldridge, J. R. McMahon, J. E. Kim, H. Y. Chang, and K. D. Beyer, Science, 1993,261, 1418. 23 D. R. Hanson and K. Mauersberger, J. Phys. Chem., 1990,94,4700. 24 B. Rowland, M. Fisher, and J.-P. Devlin, J. Phys. Chem., 1991,85, 1378. 25 A. B. Horn, M. A. Chesters, M. R. S.McCoustra, and J. R. Sodeau, J. Chem. SOC., Faraday Trans., 1992,88, 1077. 26 A. B. Horn, T. Koch, M. A. Chesters, M. R. S.McCoustra, and J. R. Sodeau, J. Phys. Chem., 1994,98,946. 27 R. D. Kenner, I. C. Plumb, and K. R. Ryan, Proc. 2992 Quadrennial Ozone Symp. 28 L. F. Keyser and M. T. Leu, J. Colloidlnterface Sci., 1993,155, 137. 29 0. Schrems, S.Peil and S. Seisel, Proc. 9th Int. Conf. on Fourier Transform Spectroscopy, SPIE, 1993. 30 M. A. Tolbert, B. G. Koehler, and A. M. Middlebrook, Spectrochi-mica Acta, 1992,48A, 1303. 31 L. T. Chu, M. T. Leu, and L. F. Keyser, J. Phys. Chem., 1993, 97, 7779. 32 G. Hancock, in ‘Modern Gas Kinetics’, ed. M. J. Pilling and I. W. M. Smith, Blackwell, Oxford, 1987, p. 137ff. 33 0.W. Saastad, T. Ellermann, and C. J. Nielsen, Geophys. Res. Lett., 1993,20, 1191. 34 M. A. Chesters and A. B. Horn, in ‘Low Temperature Chemistry of the Atmosphere’, ed. A. J. Barnes, G. LeBras, G. K. Moortgat, and J. R. Sodeau, Springer-Verlag, Berlin, 1994. 35 A. M. Bradshaw and E. K. Schweizer, in ‘Spectroscopy of Surfaces’, ed. R. J. H. Clark and R. E. Hester, John Wiley, London, 1988, p. 41ff. 36 B. G. Koehler, L. S. McNeill, A. M. Middlebrook, and M. A. Tolbert, J. Geophys. Res., 1993,98, 10563. 37 W. A. Guillory and M. L. Berstein, J. Chem. Phys., 1975,62, 1058. 38 D. Briggs and M. P. Seah, ‘Practical Surface Analysis, Volume 2: Ion and Neutral Spectroscopy’, Wiley, London, 1992.

ISSN:0306-0012    

DOI:10.1039/CS9942300195

出版商:RSC    

年代:1994

数据来源: RSC