|
1. |
Contents pages |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 007-008
Preview
|
PDF (204KB)
|
|
摘要:
ISSN 0306-001 2 CSRVBR 21 (3) 147-214 (1992) Chemical Society Reviews Volume 21 Issue 3 Pages 147-21 4 September 1992 Solvatochromism, Thermochromism, Piezochromism, Halochromism, and Chiro- Solvatochromism of Pyridinium N-Phenoxide Betaine Dyes By Christian Reichardt (pp. 147-1 53) The position of the longest-wavelength, intramolecular charge-transfer absorption band of the solution UV/ Vis spectra of pyridinium N-phenoxide betaine dyes depends on the polarity of the solvent, the temperature of the solution, the external pressure, the nature and concentration of added salts, and possibly, in the case of homochiral betaine dyes, on the use of chiral solvents. A survey of these peculiar spectral properties of pyridinium N-phenoxide betaine dyes is presented. The reasons for this extraordinary behaviour and possible applications (e.g.empirical determination of solvent polarities) are discussed.0 Molecular Dynamics Simulations of Surface Chemical Reactions By Barbara J. Garrison (pp. 155-1 62) Molecular dynamics simulations provide a means to examine the atomistic details of chemical reactions and at the same time yield information which can be compared directly with experimental data. This review presents the basic ideas behind molecular dynamics simulations and interaction potentials. Results of molecular dynamics simulations on the probing of surface processes by keV particle bombardment, the molecular beam epitaxial growth of Si, and the F atom etching of Si are discussed. Magic Numbers in Molecular Clusters: A Probe for Chemical Reactivity By M.Todd Coolbaugh and James F. Garvey (pp. 1 63-1 69) When neutral van der Waals clusters produced in a supersonic expansion are ionized, the resulting cluster ion distribution can exhibit dramatic intensity anomalies for very specific cluster sizes. These ‘magic numbers’ typically result from a particularly stable cluster ion geometry generated by the completion of a full solvent shell around a central ion core. This paper reviews the authors’ current work which employs magic numbers as a probe, not only of cluster ion structure, but also of chemical reactivity within these species. That is, the appearance of certain magic numbers can be attributed to unique ‘cluster reactions’ occurring between the ion and the solvating molecules within the cluster.Binuclear Iron Centres in Proteins By Ralph G. Wilkins (pp. 171-1 78) Binuclear iron sites with bridging ligands feature in a number of important non-haem iron proteins. Haemerythrin is the respiratory protein in marine worms. In addition, there are the enzymes ribonucleotide reductase, important in DNA synthesis; purple acid phosphatases with as yet unknown function; and methane monooxygenase, from methanotropic bacteria, which catalyses the insertion of 0 into C-H bonds. The variety of techniques used in their structural characterization is described from an historical viewpoint. Finally, the reactivity and mechanisms of action of these proteins are briefly discussed. r Ruthenium 0x0 Complexes as Organic Oxidants By W.P. Griffith (pp. 179-1 85) A summary is given of the applications of oxoruthenium complexes as catalytic oxidants for organic substrates, mainly but not exclusively for the oxidation of primary alcohols to aldehydes or carboxylic acids and secondary alcohols to ketones. Complexes containing ruthenium(vrI1) to (111) inclusive are covered, with an emphasis on recent developments in oxoruthenates(vI1) and (vI). Particular attention has been paid to those species which have been structually or spectroscopically characterized. Molecular Fluorescent Signalling with ‘Fluor-Spacer-Receptor’ Systems: Approaches to Sensing and Switching Devices via Supramolecular Photophysics By Richard A. Bissell, A. Prasanna de Silva, H.Q. Nimal Gunaratne, P. L. Mark Lynch, Glenn E. M. Maguire, and K. R. A. Samankurnara Sandanayake (pp. 1 87-1 95) The established principle of photoinduced electron transfer is combined with the modular system ‘Fluor- Spacer-Receptor’ in order to develop rationally an approach to the phenomenon of guest-responsive fluorescence. This approach is highlighted with regard to its flexibility, expandability, and its ability formally to unify a wide variety of fluorescent signalling strategies. Its value for the development of molecular photoionic devices with digital action as well as for the design of sensors for chemical species and properties is pointed out. Electrochemical Aspects ofSTM and Related Techniques By P. A. Christensen (pp. 197-208) The application of Scanning Tunelling Microscopy, Scanning Tunelling Spectroscopy, and Atomic Force Microscopy to electrochemistry is reviewed, with particular reference to the practical aspects. A selection of recent work emphasizing the kind of information that can be obtained using these novel techniques is also discussed.Synthetic Amphiphile Vesicles By A. M. Carmona-Ribeiro (pp. 209-21 4) At the border between colloids and biomimetic systems, dioctadecyldimethylammonium chloride and dihexadecylphosphate vesicles have had their properties scrutinized during the past decade. Physical and functional properties, responsiveness to the environment, aggregation and fusion, and new uses for synthetic amphiphile vesicles are critically overviewed. Articles that will appear in forthcoming issues include The Construction of a Molecular Lego Set J.F. Stoddart Caged Explosives: Metal-Stabilized Chalcogen Nitrides J. D. Woollins et al. Peptide Structure from NMR M. P. Williamson and J. P. Waltho Zero Oxidation State Compounds of Scandium, Yttrium, and the Lanthanides F. G. N. Cloke Motion of Sorbed Gases in Polymers W.-Y. Wen Individual Solvated Ion Properties and Specificity of Ion Adsorption Effects in Processes at Electrodes B. E. Conway Transition Metal Complexes of Silylenes, Silenes, Disilenes, and Related Species P. D. Lickiss Hi in Space S. Miller and J. Tennyson Ion Pairing and Radical Processes C. I. F. Watt and K. J. Msayib Thermodynamic and Related Studies of Amphiphile + Water Systems M.I. Davis Selectivity and Mechanism in Catalytic Asymmetric Synthesis J. M. Brown Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E25.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way. All other orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letchworth, Herts. SG6 1HN England. 1992 annual subscription rate E.C. 272.00, Overseas &81 .OO, U.S.A. $158.00. Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A.Postmaster: Send address changes to Chemical Society Review’s, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. Second class postage is paid at Jamaica, New York 11431. All other despatches outside the U.K. by Bulk Airmail within Europe, Accelerated Surface Post outside Europe.
ISSN:0306-0012
DOI:10.1039/CS99221FP007
出版商:RSC
年代:1992
数据来源: RSC
|
2. |
Front cover |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 009-010
Preview
|
PDF (667KB)
|
|
摘要:
Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Dr. A. R. Butler Dr. E. C. Constable Professor B. T. GoldingProfessor M. Green Professor D. M. P. Mingos FRS Professor J. F. Stoddart Consulting Editors Dr. G. G. Balint-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor A. Hamnett Dr. T. M. Herrington Dr. R. Hillman Professor R. Keese Dr. T. H. LilleyDr. H. Maskill Professor Dr. A. de Meijere Professor J. N. Miller Professor S. M. Roberts Professor B. H. Robinson Dr. A. J. Stace Staff Editors Mr. K. J. Wilkinson Dr. J. A. Rhodes University of Sussex University of Leicester University of St. Andrews University of Cambridge University of Newcastle upon Tyne University of Bath Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Reading University of Bristol University of Bern University of Sheffield University of Newcastle upon Tyne University of Gottingen Loughborough University of Technology University of Exeter University of East Anglia University of Sussex Royal Society of Chemistry, Cambridge Royal Society of Chemistry, Cambridge It is intended that Chemical Society Reviews will have the broad appeal necessary for researchers to benefit from an awareness of advances in areas outside their own specialities.Deliberate efforts will be made to solicit authors and articles from Europe which present a truly international outlook on the major advances in a wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca. 30 references), but should act as a springboard to further reading. In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context.Review articles must be short, around 6-8 journal pages in extent. In consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Instruction to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be submitted to the Senior Editor (Reviews), Books and Reviews Department, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 4WF. @ The Royal Society of Chemistry, 1992 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/CS99221FX009
出版商:RSC
年代:1992
数据来源: RSC
|
3. |
Back cover |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 011-012
Preview
|
PDF (1123KB)
|
|
ISSN:0306-0012
DOI:10.1039/CS99221BX011
出版商:RSC
年代:1992
数据来源: RSC
|
4. |
Solvatochromism, thermochromism, piezochromism, halochromism, and chiro-solvatochromism of pyridiniumN-phenoxide betaine dyes |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 147-153
C. Reichardt,
Preview
|
PDF (1023KB)
|
|
摘要:
Solvatochromism, Thermochromism, Piezochromism, Halochromism, and Chi ro-Solvatochromism of Pyridin ium N-Phenoxide Betaine Dyes C. Reichardt Department of Chemistry, Philipps University, Hans- Meerwein -Strasse, D-3550 Marburg, Germany 1 Introduction Whenever a chemist wishes to carry out a given chemical reaction, not only has he to decide which reaction partners, reaction vessel, reaction temperature, and reaction time should be applied -he also has to select carefully the appropriate solvent for the planned reaction. As a result of the pioneering work of Berthelot and Saint-Gilles in Paris in 1862 (discovery of the solvent influence on the esterification of acetic acid by ethanol) and Menschutkin in St. Petersburg in 1890 (solvent dependence of the alkylation of triethylamine by iodoethane), and the host of papers on solvent effects published since then, every chemist is nowadays well aware of the fact that solvents can have a strong influence on reaction rates and chemical equilibria as well as on the position of spectral absorption maxima. I At present, there exists a supply of about 300 common organic solvents, apart from the infinite number of solvent mixtures.A chemist therefore needs, in addition to his exper- ience and intuition, some guidelines for the selection of the proper solvent. In principle, there are two main objectives: (a) the prediction of rate or equilibrium constants as well as absorption maxima in other solvents, and (b) to reach some understanding of the various solute-solvent interactions that might effect rate and equilibrium constants as well as absorption spectra.Usually, chemists have tried to understand solvent effects in terms of the so-called solvent polarity, which is not easy to define precisely or to express quantitatively. Seduced by the simplicity of electrostatic solvation models, attempts at expressing solvent polarity quantitatively involve mostly physical solvent para- meters such as dielectric constant, dipole moment, refractive index, or functions thereof. However, this procedure is often inadequate because it does not take into account specific solutc- solvent interactions such as hydrogen-bonding, electron-pair donor (EPD) -electron-pair acceptor (EPA), and solvophobic interactions. Hence, from a more practical point of view, it seems reasonable to understand solvent polarity in terms of the overall solvation cupability of a solvent for reactants and acti- vated complexes as well as for molecules in the ground and excited states, excluding solute-solvent interactions such as protonation, oxidation, reduction, complexation, etc., which may lead to a chemical change of the solute.' This overall solvation capability depends on all possible -specific and non- specific intermolecular forces between solute and solvent.~ Christian Reichardt was born in 1934, in Ebersbach, Saxony, Germany. He studied chemistry at the 'Carl Schorlemmer ' Tech-nical University for Chemistry in Leuna-Merseburg, GDR, and -after moving illegallj~ to West Germany in I955 -at the Philipps Universitjl in Marburg, FRG, w)here he obtainedhis Ph.D.in I962 under the tutelage of Prof. K. Dimroth, and completed his Habilitation in 1967. Since I971 he has been Professor of Organic Chemistry at Marburg. He has authored and co-authored more than 125 papers and patents, and a book 'Solvents and Solvent Efects in Organic Chemistry'. His research interests are in sjnthetic organic chemistry (chemistry of aliphatic dialdehydes; sjnthesis qf'polymethine dyes) and in physical organic chemistry (solvatochromism qf organic dyes; solvent effects in organic chemistrji; empirical parameters ojsolvent polarity). Obviously, solvent polarity thus defined cannot be described by a single physical solvent parameter. The lack of comprehensive theoretical expressions for the calculation of solvent effects, and the inadequacy of defining solvent polarity in terms of simple physical characteristics, has led to the introduction of empirical parameters of solvent polarity.Based on the assumption that particular, carefully selected and well understood solvent-dependent chemical reactions or spectral absorptions may serve as suitable model processes, various empirical solvent polarity scales have been developed.'T2 The first empirical parameter of 'solvent ionizing power' was the Y-scale introduced by Winstein in 1948 and derived from the SN1solvolysis of t-butylchloride.2c The first to suggest that solvatochromic dyes might be used as indicators of solvent polarity was Brooker in 195 1, but Kosower was the first to set up a real spectroscopic solvent polarity scale in 1958.This was called the Z-scale and used the intermolecular charge-transfer absorption of 1-ethyl-4-methoxycarbonylpyridiniumiodide as solvent-sensitive reference process. The pyridinium N-phe- noxide betaine dye (1) was proposed by Dimroth, Reichardt, et al.1,3 in 1963 as a new UV/Vis-spectroscopic indicator of solvent polarity which, by virtue of its exceptionally large negative solvatochromism (i.e. a hypsochromic shift of the UV/Vis absorption band with increasing solvent polarity), overcame some practical limitations of other solvatochromic reference molecules. 1,2 (1) R'=R2=H;X=0 (2)R' = R2 = CMe,; X = 0 (3) R' = S02Me;R2 = H; X = 0 (4) R' = R2 = H; X = S In applying solvent polarity scales based on a single empirical parameter, it is tacitly assumed that the combination of solute- solvent interactions between the reference solute and the solvent is the same as with the particular substrate under consideration.It turns out that, in many cases, this is an oversimplification. Therefore, multiparameter correlation equations have been introduced, which consist of up to four single empirical para- meters, each of them measuring a certain aspect of the overall solvation capability of a given solvent (e.g.polarizability, dipo- larity, Lewis acidity, Lewis This kind of procedure, i.e. the use of standard or reference compounds in order to establish empirical solvent parameters, is quite common in chemistry.For example, the well-known 147 Hammett equation, used to estimate substituent effects on rates, equilibria, and absorptions, is likewise based on an empirical reference process -the ionization of substituted benzoic acids in water at 25 "C. The correlations between these empirical substi- tuent or solvent parameters and the substituent- or solvent-dependent processes under study take usually the form of a linear free-energy (LFE) relationship. 2 Solvatochromism of Betaine (I) and the E~(30)Scale The negatively solvatochromic pyridinium N-phenoxide betaine dye (1) exhibits one of the largest solvatochromic shifts ever observed: its longest-wavelength intramolecular charge-transfer (CT) absorption band is hypsochromically shifted by 9730 cm-(375nm) on going from diphenyl ether (A,,, = 810 nm) to water (A,,, = 453 nm).Solutions of (1) are red-coloured in methanol, violet in ethanol, blue in isoamyl alcohol, green in acetone, and yellowish-green in ethyl acetate, thus covering the whole visible region and allowing even a visual estimate of the solvent polarity (Figure 1). CH30H C2H5OH i-C5H1 ,OH CH3COCH3 CH~CO~C~HS CH30H Figure 1 Solution colours of the negatively solvatochromic pyridinium N-phenoxide betain dye (1) in five solvents of different polarity and in the two-phase solvent system methanol/isoamyl alcohol (red/blue). This outstanding negative solvatochromism stems from the unequal, differential solvation of the electronic ground and excited state of (1) with increasing solvent polarity (Scheme 1). The solvent-mediated stabilization of the highly dipolar, zwit- terionic ground state of (I), relative to its less dipolar excited state, results from the following properties of the betaine molecule: (a) it exhibits a large permanent dipole moment, suitable for the registration of dipole-dipole and dipole-induced dipole interactions; (b) it possesses a large polarizable 7~ -electron system, suitable for the registration of dispersion interactions; and (c) with the phenoxide oxygen atom it has a highly basic EPD centre, suitable for interactions with Brmsted acids (through H-bonding) and Lewis acids (through EPD-EPA bonding).The positive charge of the pyridinium moiety is delocalized and sterically shielded. Therefore, the CT absorp- tion of (1) depends strongly on the electrophilic solvation power of the solvents, i.e.on their hydrogen-bond donor (HBD) ability and Lewis acidity (= EPA behaviour), rather than on their nucleophilic solvation capability ( = EPD behaviour). In stronger acidic solvents, (1) is protonated and the solvatochro- mic CT absorption band disappears (reversibly).',3 In analogy to the Z-values of Kosower, the empirical solvent polarity parameter ET(30)is simply defined as the molar electro- CHEMICAL SOCIETY REVIEWS, 1992 k=6D Ph Solvent Polarity -I Ph I -PhQPh I01 h = 81Onn h = 453nn +E~(30)= h&NA + hvCT i in Ph20 in H20 = molar Transition Energy Ph Ph0Ph I101-Non-polar Polar Solvent JL~=15D Scheme 1 nic transition energy of dissolved (I), measured in kcal / mol -', according to equation A high ET(30) value corresponds to high solvent polarity.In order to avoid the dimension kcal/ mol-', normalized Ep values can be used instead of ET(30).36 They are defined according to equation 2, using water and tetramethylsilane (TMS) as, respectively, extreme polar and non-polar reference solvents. Hence, the Ey-scale ranges from 0.0 for TMS, the least polar solvent, to 1.0 for water, the most polar solvent. -ET(solvent) -ET(TMS) = ET(solvent) -30.7 ET -E,(water) -ET(TMS) 32.4 (2) Since betaine (1) is insoluble in aliphatic hydrocarbons and only sparingly soluble in water, the more lipophilic penta(t- buty1)-substituted dye (2)3bas well as the tris(methanesulfony1)- substituted dye (3)6bhave been used as secondary standard betaine dyes, the ETvalues of which correlate linearly with the ET(30)values of (I).In that way, ET(30)values are presently known for more than 300 organic solvent^^.^.^ and for numerous binary solvent mixture^.'.^ lo In addition, ET(30)values have been determined for room-temperature liquid salts, ' supercriti-cal-fluid (SCF) solvents,12 as well as polymers 13a and aqueous polymer solutions. 3h Only for perfluorohydrocarbons [insolu- bility of (1) and (2)] and acidic solvents [protonation of (1) and (2)] are ET(30)values not directly measurable. The border between acidic and less acidic solvents for which ET(30)values are available is determined by the acidity constant of the corresponding acid of (I): the pKA of protonated (1) is 8.65 f0.05,14 8.63 f0.0318' in H,O. The synthesis of the standard betaine dye (1) has been simplified recently,' so and the dyes (1) and (2) are now commercially available.5d In addition to the characterization of solvent polarities, the ET(30)values have found many other applications, e.g. (a) LFE correlations with other solvent-dependent reactions and absorp- tions;'.' (b) quantitative analysis of binary mixtures of solvents with different p~larity;'.~ 'O (c) characterization of mobile phase polarities in reversed-phase liquid chromatography (RPLC)' 7a and counter-current chromatography (CCC);7h (d) determination of the polarity of alumina surfaces in adsorption liquid chromatography (ALC);' 7( and (e) description of the interfacial microenvironment of micro-heterogeneous solutions (e.g.microemulsions, micelles, vesicles, phospholipid bilayers, SOLVATO-, THERMO-, PIEZO-, HALO-, AND CHIRO-SOLVATOCHROMISM OF BETAINE DYES-C. REICHARDT surfactants, tensides).* Applications of ET(30)values in analy- tical chemistry have been reviewed.' su For example, they can be used for rapid UV/Vis-spectroscopic determination of water in organic solvents.sh Zwitterionic betaine dyes of type (1) have been used to construct a molecular battery (utilizing reversible one-electron redox reactions of a certain betaine dye),' and as potential materials with non-linear optical properties for fre- quency doubling of laser radiation.,O In order to unravel the combination of specific and non- specific betaine-solvent interactions, the pyridinium N-thio- phenoxide betaine dye (4)(see p.147) has been synthesized recently.2 Since sulfur is known to be a much poorer hydrogen- bond acceptor than oxygen, the homomorphic thiobetaine (4) should be less sensitive (or even insensitive) to specific hydrogen- bond interactions with HBD solvents. Therefore, a correlation line between ET(30) of (I) and EK30) of (4)should be linear for non-HBD solvents, but non-linear with deviations for HBD solvents. Surprisingly enough, there is an excellent linear corre- lation between ET(30) of the oxygenbetaine (1) and Es(30)of the thiobetaine (4)including both HBD and non-HBD solvents, as shown in equation 3 (number of solvents n = 60; correlation coefficientY = 0.993).The reasons for this unexpected behaviour of (4)in comparison with (1) have been discussed.2 The slope of equation 3 is larger than unity. That means that thiobetaine (4)is more strongly solvatochromic than (1) and could serve as a new, more sensitive solvent polarity indicator. Unfortunately, solu- tions of (4)are rather unstable because the thiobetaine is easily oxidized to the corresponding, sparingly soluble disulfide. E+(30)/(kcalmol-') = 1.25ET(30)/(kcalmol-I) -10.81 (3) A recent analysis of the ET(30)values of (1) for 100 solvents by means of the multiparametric Kamlet-Taft equation20 had led to equation 4 (n = 100; Y = 0.984),22in which T* is an index of solvent dipolarity/polarizability, a a measure of the solvent HBD ability and 6 a polarizability correction term for the solvent being aromatic (6 = 1.O), polyhalogenated aliphatic (6 = 0.5), or neither (6 = 0.0).Equation 4 clearly shows that ET(30)values represent a combination of non-specific (n*)and specific (a) solute-solvent interactions. The ratio of the coeffi- cients of a and T* is 1.1, hence the sensitivity of ET(30) to a is slightly larger than to 7~*for HBD solvents22 (cf. also reference 4c). E,(30)/(k~almol-~)= 13.68 (r*-3.456) + 14.51a (4)+ 30.8 The position of the long-wavelength UV/Vis absorption band of dissolved (1) depends not only on the solvent polarity (solvatochromism), but also on the solution temperature (ther- mochromism), on external pressure (piezochromism), and on the type and concentration of added salts (halochromism).3 Thermochromism of Betaine (1) A solution of (1) in ethanol is blue-violet (A,,, = 568 nm) at + 78 "C and red-coloured (A,,, = 513 nm) at -78 "C. This corresponds to a hypsochromic shift of the CT band of (1) by -55 nm with decreasing temperature (AT= 156"C).23a This temperature-dependent behaviour of solutions of (1) represents a new type of thermochromism ('negative thermo-solvatoch- romism'). It is obviously caused by the increased differential stabilization of the dipolar electronic betaine ground state, relative to its less dipolar excited state, with decreasing solution temperature (cf.Scheme 1). At low temperatures, the specific and non-specific intermolecular interactions between betaine and solvent molecules are strengthened, without any alterations of the chemical structure of (1). This is in contrast with already known types of thermochromism, which in most cases result simply from the temperature-dependent shift of a chemical equilibrium between a coloured and non-coloured species. In conclusion, the lower the solution temperature, the higher the corresponding ET(30)values, and the better the solvation capability of the solvent. That is, solvent polarity is temperature- dependent. This is a rather trivial statement, but thermochromic solutions of (1) allow a simple quantitative determination of this temperature dependence.Recently, the thermochromism of solutions of (1) in some binary solvent mixtures has been In addition, the thiobetaine dye (4)exhibits an even larger thermochromic range than (1).2 Since the thermochro- mic (and solvatochromic) UV/Vis absorption band of (4) [and (l)] lies within the visible range, the thermochromism of (4)[and (l)] can be easily visualized by means of simple test-tube experiment^.^',^^ 4 Piezochromism of Betaine (1) Application of external pressure to solutions of betaines (1) and (2) produces in all solvents used a hypsochromic shift of their long-wavelength CT absorption band.24 For example, the UV/ Vis absorption maximum of (1) in ethanol is shifted hypsochro- mically by -27 nm (from A,,, = 547 to 520 nm) by increasing the external pressure up to 10 kbar (9869 atm).24c This corres- ponds to an increase in the ET(30)-value of ethanol by 2.7 kcal mol- l.Thus, solvent polarity is pressure dependent. The reason for this new type of piezochromism ('negative piezo- solvatochromism') lies again in the differential solvent-mediated stabilization of the dipolar electronic betaine ground state, relative to its less dipolar excited state (cf. Scheme I), with increasing external pressure. The range of piezochromism is different from solvent to solvent. The pressure-dependent absorption energies of (1) and (2), dissolved in one solvent under study, correlate well with the dielectric function (E -l)/(~+ 2) of this solvent, which also increases with compression.26c This seems to indicate that only non-specific betaine-solvent interac- tions contribute to the observed piezochromism, whereas any specific betaine-solvent interactions that occur are, to first order, constant with pressure -at least for the alcoholic HBD solvents used.,,' 5 Halochromism of Betaine (1) and its Crown Ether Substituted Derivatives Addition of salts (electrolytes, ionophores) to solutions of betaine dye (1) leads in most cases to a hypsochromic shift of its long-wavelength CT absorption band, depending on the nature and concentration of the salt added.1.6,25 For example, the addition of KI, NaI, LiI, BaI,, Ca(SCN),, and Mg(ClO,), in excess to solutions of (1) in acetonitrile causes a differential hypsochromic band shift.256 This shift increases with this elec- trolyte order, that is with the increasing effective cation charge (i.e.ion charge/ion radi~s),~~.~~~ whereas the anions seem to have little or no influence on this halochromism.26 There exists even an almost linear correlation between the salt-induced hypsochromic shifts of (1) and the effective charge of the cations of the added alkali and alkaline earth salts, as shown in Figure 2.19 Generally, a CT absorption [as that of (l)] depends on the electron affinity of the electron-acceptor part (here the pyridi- nium moiety) and on the ionization energy of the electron-donor part (here the phenoxide group) of the light-absorbing molecule.Ion-pair association between the phenoxide group of (1) and the cations of the salts added increases the ionization energy of the electron-donor part for electrostatic reasons. This corresponds to an increase in CT excitation energy and is in agreement with the observed salt-induced hypsochromic shifts. A particularly strong betaine-ion association is to be expected with the lithium cation, the alkali ion with the highest effective charge, This is indeed the case, as has been shown recently by addition of LiCIO, to solutions of (1) in diethyl ether or tetrahydrofuran.2 5d The lithium ions act like Lewis acids towards betaine dye (l), making its CT absorption extremely sensitive to low concent- rations of lithium ion.Because of this strong betaine-Li+ CHEMICAL SOCIETY REVIEWS, 1992 PhAN Ph ,.'Sr 2+ I/ Ba2+ Ph-0-Ph I /lQ -/ 111, 11 1 I1 1 1 5 10 15 20 (n+/rM,n +) x 1031pm-' Figure 2 Correlation between the salt-induced change in the molar transition energy ET(30)of betaine dye (l), dissolved in acetonitrile at 25 "C,and the effective cation charge ( = ion charge/Pauling radius) of the added alkali and alkaline earth salts (i.e. LiI, NaI, KI, RbI, and CsI, as well as CaI,, Sr12, and BaI,). 19.29hEffective cation radii: R. D. Shannon, Actu Crystullogr., Sect. A, 1976, 32, 751. Correlation equation: ET(30) = 0.992[(n+/rv,n + ) x lo3]+ 41.6; pairs of values n = 8; correlation coefficient r = 0.989; standard deviation of the estimate .F = 0.85.interaction in non-polar solvents, one can consider the lithium ion essentially titrating the betaine dye, similar to its behaviour against Brransted acids2 5d Are aqueous salt solutions more polar than pure water? The availability of the more hydrophilic, better water-soluble betaine dye (3) makes investigations of such aqueous electrolyte solutions possible.6h Addition of up to one mole salt to an aqueous solution of (3) results in small hypsochromic CT band shifts of 8-28 nm, which correspond to an increase in the ET(30) values of 0.8-3.9 kcalmol-*, as shown in Figure 3. Obviously aqueous electrolyte solutions behave as more polar solvents than water against betaine dye (3). From the limited data shown in Figure 3, it is at present difficult to draw general conclusions about the influence of the various cations and anions on the extent of halochromism found in aqueous solu- tion.It seems that in aqueous solution, cations with higher effective charges such as Mg2 + induce larger hypsochromic KI KF KCI 65.1 CSI 63.9 Pure Water (E~(30)= 63.1 kcal mol-') Figure 3 Increase in ET(30)of (1) for 1M aqueous salt solutions, relative to the ET(30)-value of pure water, determined by means of the secondary standard betaine dye (3).6h shifts than cations of lower charge density. Unfortunately, it is not possible to investigate aqueous solutions of (3) with salt concentrations higher than 1M because the betaine dye is salted out.Better water-soluble pyridinium N-phenoxide betaine dyes are therefore needed. Plots of ET(30) values for binary solvent mixtures as a function of their composition are mostly not linear -in some cases they exhibit even maxima and minima.7 -lo For the whole concentration range, the two-parameter equation 5 has been developed, describing the entire ET(30)/concentration curve in closed form.8 ET(30) and ET(30)" are the empirical solvent polarity parameters of the binary solvent mixture and of its pure, less polar component, respectively; cpis the molar concentration of the pure, more polar component of the binary solvent mixture, and EDas well as c* are adjustable parameters specific for the binary mixture under study.8 Equation 5 has been found valid for about 60 different binary solvent mixtures8 E,(30) = EDln(c,/c* + 1) + ET(30)" (5) ET(30) = Aln(c/c* + 1) + ET(30)" (6) It is now of particular interest to note that the polarity of electrolyte solutions can be treated in the same way as the polarity of binary solvent mixtures, the salt added being equiva- lent to the more polar solvent ~omponent.~.~~~ The polarity of an alcoholic electrolyte solution as a function of the salt concentration can be analogously described by equation 6, in which ET(30) and ET(30)" are the polarity parameters of the salt solution and of the pure solvent, respectively; c is the molar salt concentration; A and c* are again adjustable parameters specific for the electrolyte solution under Equation 6 has been applied successfully to twenty combinations of salts and alco- hols25c That is, added inorganic salts behave as more polar 'co- solvents', at least in alcoholic electrolyte solutions.The validity of equation 6 for other, non-alcoholic salt solutions has still to be proven. Ph Ph PhI (5)n = 0 (6) n = 1 (7)n =2 The halochromism-producing intermolecular betaine-cation interaction should be strengthened utilizing crown-ether substi- tuted betaine dyes such as (5)-(7), which have been synthesized very recently.27 This is indeed the case. For example, addition of potassium iodide to an acetonitrile solution of the [18lcrown-5-betaine (6) causes a distinct colour change from violet to dark- red, which corresponds to a hypsochromic CT band shift of -54 nm (Scheme 2).27The crown-ether substituted betaines (5)-(7) constitute a very sensitive, new class of so-called chro- moionophoric dyes.Chromoionophores are dye molecules which contain chromophoric and ionophoric subunits. By the way, the chromoionophoric dyes (5)-(7) exhibit also a pro- nounced negative solvatochromism: a solvent-change from chloroform (A,,, = 668 nm) to water (A,,, = 433 nm) shifts the long-wavelength CT band of (6) hypsochromically by 235 nm!27 151SOLVATO-, THERMO-, PIEZO-, HALO-, AND CHIRO-SOLVATOCHROMISM OF BETAINE DYES-C. REICHARDT Ph Ph PhI h = 583 nm (MeCN) h = 529 nm (MeCN + KI) Ah = -54 nm (!)1 Colour change from violet to dark-red I t (18-Crown-5)-Betaine (6) A chromo-ionophoric dye in MeCN (c = moll-’) Scheme 2 25 20 0 A .4 \ \O l5 E5 10 Q 5 (7)n = 2 0 Figure4 Cation-selective halochromism of (5)-(7) in methanol at 25 “C as function of the Pauling cation radius r of the alkali metal iodides NaI, KI, RbI, and CsI (betaine concentration ca.M; salt concentration M). AX = A,,, (without salt) -A,,, (with salt).” Depending on the size of the crown-ether ring, the betaine dyes (5)-(7) show a definite cation-selective UV/Vis behaviour, as illustrated in Figure 4. From the dependence ofthe &,,-shifts of (5)-(7), measured in methanol, on the size of the alkali metal ion of the iodide salt added, it follows that [ 151crown-4-betaine (5)complexes preferably with Na+,[181crown-5-betaine(6) with + +K , and [2llcrown-6-betaine (7) with Cs .2 Treatment of Na+-selective(5)in methanol with a mixture of Na, K, Rb, and Cs iodide, results in a hypsochromic band shift as if NaI alone were added.Addition of lithium and alkaline earth metal salts to solutions of (5)-(7) in methanol or acetonitrile leads to the disappearance of their long-wavelength, halochromic CT absorption band due to stronger betainexation complexation -similar to the effect of protonation. The halochromism of the betaine dyes (l), (3), (5)-(7), as mentioned before, constitutes a new type of ‘genuine’ halo-colourless yellow hmX = 265nrn (AcOH) = 431nrn (AcOH + H2SO4) yell0wish orange-red h,, = 327nm (MeOH) Scheme 3 ~hromism.~.~~The term ‘halochromism’ was first introduced by Baeyer and Villiger in 1902,26and was simply defined as the colour change of a dissolved compound on addition of acid or base, during which a chemical reaction transforms a colourless compound into a coloured one.Typical examples of this ‘trivial’ halochromism are the acid-base reactions given in Scheme 3.26 In contrast to this trivial halochromism, the genuine haloch-romism of pyridinium N-phenoxide betaine dyes refers to a colour change on addition of an electrolyte to the dye solution, not accompanied by chemical change of the dissolved, halochro-mic compound. We have suggested the designation ‘negative (positive) true halochromism’ for a hypsochromic (bathochro-mic) shift of the UV/Vis absorption band of a dissolved sub-stance on increasing electrolyte concentration (or increasing ionic strength), provided this shift is not caused by chemical alterations of the ~hromophore.~,~~ 6 Attempts at the Detection of Chiro-Solvatochromism The extreme spectral sensitivity of pyridinium N-phenoxide betaine dyes to small changes in the surrounding solvation shell should make it possible to detect a phenomenon called by us chiro-solvatochromism, using chiral derivatives of these dyes, dissolved in pairs of enantiomeric, homochiral sol~ents.~~~~~ That is, the formation of diastereomorphic solvates between chiral solvatochromic solutes [e.g.an (9-configured chiral betaine dye] and pairs of enantiomorphicsolvent molecules [e.,g. (R)-and (9-configurated alcohols such as octan-2-01]ought to result in a shift of the solvatochromic UV/Vis absorption band of the chiral solute, as compared to its position in the corres-ponding racemic solvent.A recent review describes first attempts to differentiate UV/Vis-spectroscopically between enantiomorphic guest molecules by means of chiral crown-ether substituted dyes as optical sensors.28 Ph Ph Ph I Ph Ph I Ph In order to be get such chiro-solvatochromic dyes, we have recently synthesized the chiral betaine dyes (8) and (9).29The negatively solvatochromic (S,S)-configurated s-butyl substi- CHEMICAL SOCIETY REVIEWS, 1992 tuted dye (8) was prepared from homochiral (s>-( -)-2-methyl-second-harmonic generation (i.e.frequency doubling of laser radiation).*Ob Finally, the polymeric betaine dye (16) has been butan-1-01 (optically active amylalcohol from fermentati~n),~~~ and the (R,R)-[l8]-crown-5-betainedye (9) was obtained in a 10-step procedure from natural (5‘)-(+ )-~aline.~~~ Disappointingly, solutions of these chiral betaine dyes in pairs of enantiomeric homochiral solvents such as butan-2-01, octan- 2-01, neomenthol, 2-amino-butan- 1-01, and 1-phenylethylamine do not exhibit any chiro-solvatochromism, at least at room temperature. The small band shifts observed are within the limits of experimental error.29 Further attempts to get chiro- solvatochromic betaine dyes are in progre~s.*~~,*~~ 7 Concluding Remarks The outstanding sensitivity of the long-wavelength, intramole- cular charge-transfer UV/Vis absorption of the pyridinium-N- phenoxide betaine dyes to small changes in solvent polarity, solution temperature, external pressure, as well as nature and concentration of added salts, makes these dyes a very useful class of compounds -useful not only for the study of more theoretical concepts, but also for very practical applications such as the construction of an empirical solvent polarity scale.The behav- iour of these betaine dyes may be compared to that of the Princess in Hans Christian Andersen’s fairy-tale ‘The Princess and the Pea’. As one perhaps remembers, the Princess was so sensitive to her surroundings that she was able to feel a pea through twenty mattresses and twenty eider-down quilts on her bed.Ph Ph Ph CIQCI Ph I R (13) R = Me 1 (14) R = OMe Over the past few years attempts have been made to obtain further pyridinium N-phenoxide betaine dyes with special and improved properties. Other zwitterionic dyes such as (lo)-( 16) have been synthesized recently and investigated spectroscopi- cally.14,20,30 The chloro-substituted betaine dye (10) is less basic than (1) (pKA= 4.78 for the corresponding acid) and is suitable as a solvent polarity probe in more acidic solvents.’4 The negatively solvatochromic dyes (1 l)30u and ( 12)306seem to have an even larger solvatochromic range than (I), whereas the likewise negatively solvatochromic ayes (14)30‘and (15y0exhi-bit a weaker solvatochromism than (1).Betaine dyes such as (1 5) have been tested for applications in non-linear optics such as prepared, by polymerization of the corresponding isocyano- substituted monomer with NiCI,, and is a promising candidate for the construction of polymeric photo~ond~~tors.~~~ Acknowledgements. I thank all my co-workers mentioned by name in the various citations. I appreciate their dedication, enthusiasm, and hard work. In addition, I wish to thank the Deutsche Forschungsgemeinschaft, Bonn, and the Fonds der Chemischen Industrie, Frankfurt(Main), for their continued financial support of this research work. 8 References 1 C. Reichardt, ‘Solvents and Solvent Effects in Organic Chemistry’. Second, completely revised and enlarged edition.VCH Publishers, Weinheim, 1988 (in particular Chapter 6, pp. 285-305, and Chapter 7, pp. 339405). 2 Recent reviews on solvent polarity scales: (a) R. W. Taft, J.-L. M. Abboud, M. J. Kamlet, and M. H. Abraham, J. Solution Chem., 1985, 14, 153; (6)0.Pytela, Collect. Czech. Chem. Commun., 1988, 53, 1333; (c) T. W. Bentley and G. Llewellyn, Progr. Phys. Org. Chem.. 1990. 17, 121; (4E. Buncel and S. Rajagopal, Ace. Chem. Res., 1990, 23, 226; (e) M. Sandstrom, I. Persson, and P. Persson, Acta Chem. Scand., 1990,44,653. 3 (a) K. Dimroth, C. Reichardt, T. Siepmann, and F. Bohlmann, Liebigs Ann. Chem., 1963,661, 1; (6)C. Reichardt and E. Harbusch- Gornert, Liebigs Ann. Chem., 1983, 721; (c)S. Spange. M. Lauter- bach, A.-K. Gyra, and C.Reichardt, Liebigs Ann. Chem., 1991,323. 4 (a)C. Laurence, P. Nicolet, M. Lucon, and C. Reichardt, Bull. Soc. Chim. Fr., 1987,125, 1001; (6)C. Laurence, P. Nicolet, M. Lucon, T. Dalati, and C. Reichardt, J. Chem. Soc., Perkin Trans. 2, 1989, 873: (c) C. Laurence, in ‘Similarity Models in Organic Chemistry, Bio- chemistry, and Related Fields’, ed. R. I. Zalewski, T. M. Krygowski, and J. Shorter, Elsevier, Amsterdam, 1991, Chapter 5, pp. 23 1-281. 5 In the first publication, reference 34 the betaine dye (1) had by chance the formula number (30). Therefore, the number 30 was added in order to avoid confusion with ET used in photochemistry as the abbreviation for triplet energy. 6 (u)C. Reichardt, E. Harbusch-Gornert, and G. Schafer, Liebigs Ann.Chem., 1988, 839; (b) C. Reichardt, P. Milart, and G. Schafer, Collect. Czech. Chem. Commun., 1990,55, 97. 7 (a)K. Dimroth and C. Reichardt, 2.Analyt. Chem., 1966,215,344: (b)Z. B. Maksimovii, C. Reichardt, and A. SpiriC, Z. Anal. Chem., 1974, 270, 100; (c)T. M. Krygowski, P. K. Wrona, U. Zielkowska, and C. Reichardt, Tetrahedron, 1985, 41, 4519. 8 (a) H. Langhals, in ‘Similarity Models in Organic Chemistry, Bio- chemistry, and Related Fields’, ed. R. I. Zalewski, T. M. Krygowski, and J. Shorter, Elsevier, Amsterdam, 1991, Chapter 6, pp. 283-342; (b)H. Langhals, Anal. Lett., 1990,23,2243; GIT Fachz. Lab., 1991, 35, 766. 9 I. A. Koppel and J. B. Koppel, Org. React (USSR), 1983,20, 523. 547. 10 (a) J. G. Dawber, J. Ward, and R.A. Williams, J. Chem. Soc., Faraday Trans. I, 1988,84,713; (b)J. G. Dawber, S.Etemad, and M. A. Beckett, J. Chem. Soc., Faraday Trans., 1990,86, 3725. 1 1 (a)W. Schroth, H.-D. Schadler, and J. Andersch, Z.Chem., 1989,29, 56, 129; (6)S. K. Poole, P. H. Shetty, and C. F. Poole, Anal. Chim. Acta, 1989,218,241; (c)W. B. Harrod and N. J. Pienta, J. Phys. Org. Chem., 1990,3,534; (d)1.-M. Herfort and H. Schneider, Liebigs Ann. Chem., 1991,27. 12 (a)J. A. Hyatt, J. Org. Chem., 1984, 49, 5097; (h)J. F. Deye, T. A. Berger, and A. G. Anderson, Anal. Chem., 1990, 62, 615; (c) Y. Ikushima, N. Saito, M. Arai, and K. Arai, Bull. Chem. Soc. Jpn., 199 1,64,2224. 13 (a)M. S.Paley, R. A. McGill, S.C. Howard, S.E. Wallace, and J. M. Harris, Macromolecules, 1990,23,4557; (b)B.Yu. Zaslavsky. L. M. Miheeva, E. A. Masimov, S.F. Djaforov, and C. Reichardt, J. Chem. Soc., Faraday Trans., 1990.86, 519. 14 M. A. Kessler and 0.S. Wolfbeis, Chem. Phys. Lipids, 1989.50, 5 1. 15 (a) B. P. Johnson, B. Gabrielsen, M. Matulenko, J. G. Dorsey, and C. Reichardt, Anal. Lett., 1986, 19, 939; (h)M. A. Kessler and 0.S. Wolfbeis, Synthesis, 1988, 635; (c) M. C. Rezende and C. M. Radetski, Quim. Nova, 1988, 11, 353 (Chem. Abstr.. 1989, 111, 8876~);(d)Aldrich Chemical Company, MilwaukeeKJSA, AIdrichi-mica Acta. 1987, 20(2), 59 {cf. also Aldrich Catalogue 1990i1991, order numbers 27,244-2 [betaine (I)], 27,305-8 [betaine (2)], and 33,418-8 (unsubstituted betaine);. SOLVATO-, THERMO-, PIEZO-, HALO-, AND CHIRO-SOLVATOCHROMISM OF BETAINE DY ES-C.REICHARDT 16 (a)C. Reichardt, in ‘Molecular Interactions’, ed. H. Ratajczak and W. J. Orville-Thomas, Wiley, Chichester, 1982, Vol. 3, Chapter 5, pp. 241-282; (b)C. Reichardt, Pure Appl. Chem., 1982,54, 1867. 17 (a)J. J. Michels and J. G. Dorsey, J. Chromatogr., 1988, 457, 85; ibid., 1990, 499, 435; (b)T. P. Abbott and R. Kleiman, J. Chromu- togr., 1991, 538, 109; (c)J. J. Michels and J. G. Dorsey, Lungmuir, 1990, 6, 414. 18 K. A. Zachariasse, N. Phan Phuc, and B. Kozankiewicz. J. Phys. Chem., 1981, 85, 2676; (6) P. Plieninger and H. Baumgartel, Ber. Bunsenges. Phys. Chem., 1982, 86, 161; (c) C. J. Drummond, F. Grieser, and T. W. Healy, Faraday Discuss. Chem. SOC., 1986,81,95; (6)I. A. Koppel and J.B. Koppel, Org. React. (USSR), 1989,26,78; (e)M. B. Lay, C. J. Drummond, P. J. Thistlethwaite, and F. Grieser, J. Colloid Interface Sci., 1989, 128, 602. 19 H. Bock and H.-F. Herrmann, Helv. Chim. Acta, 1989,72, 1171. 20 (a)M. S. Paley, E. J. Meehan, C. D. Smith, F. E. Rosenberger, S. C. Howard, and J. M. Harris, J. Org. Chem., 1989,54, 3432; (b)M. S. Paley and J. M. Harris, J. Org. Chem., 1991, 56, 568. 21 C. Reichardt and M. Eschner, Liebigs Ann. Chem., 1991, 1003. 22 Y. Marcus, J. Solution Chern., 199 1,20, 929; personal communica- tion from Prof. Marcus, Jerusalem, to C. R. from November 15, 1990. I thank Prof. Marcus for this information prior to publication. 23 (a)K. Dimroth, C. Reichardt, and A. Schweig, Liebigs Ann. Chem., 1963,669,95; (6)R. I.Zalewski, I. Adamczewska, and C. Reichardt, J. Chem. Rex (S), 1990, 280; J. Chem. Res.(M), 1990, 2157. 24 (a)K. Tamura and T. Imoto, Bull. Chem. Soc. Jpn., 1975,44369;(b) J. v. Jouanne, D. A. Palmer, and H. Kelm, Bull. Chem. SOC. Jpn., 1978,51,463; (c)W. S. Hammack, D. N. Hendrickson, and H. G. Drickamer, J. Phys. Chem., 1989, 93, 3483; (6) Y. Ikushima, N. Saito, and M. Arai, J. Phys. Chem., 1992,96, 2293. 25 (a)I. A. Koppel, J. B. Koppel, and V. 0.Pihl, Org. React. (USSR), 1984,21,98, 144; (6)G. Hollmann and F. Vogtle, Chem. Ber., 1984, 117, 1355; (c) M. C. Rezende, Tetrahedron, 1988, 44,3513; (d) Y. Pocker and J. C. Ciula, J. Am. Chem. SOC., 1989, 111,4728. 26 A. Baeyer and V. Villiger, Ber. Dtsch. Chem. Ges., 1902,35, 1189. 27 C. Reichardt and S. Asharin-Fard, Angew. Chem., 1991, 103, 614; Angew. Chem., tnt. Ed. Engl., 1991,30, 558. 28 F. Vogtle and P. Knops, Angew. Chem., 1991, 103, 972; Angew. Chem., tnt. Ed. Engl., 1991, 30, 958. 29 (a)C. Reichardt and M. Wilk, Liebigs Ann. Chem., 1990, 189; (b)C. Reichardt, S. Asharin-Fard, and Th. Niem, unpublished results. 30 (a)R. Gompper, V. Figala, R. Kellner, A. Lederle, S. Lensky, and W. Lipp, Lecture at the 1 1 th International Colour Symposium, Mon- treux/Switzerland, September 23-26, 1991; (6) H. Takeshita, A. Mori, N. Kato, E. Wada, S. Kanemasa, A. Mori, E. Fujimoto, and N. Nishiyama, Chem. Lett., 1991, 721; (c)C. Reichardt, P. Milart, and G. Schafer, Liebigs Ann. Chem., 1990, 441; (d)A. J. M. van Beijnen, R. J. M. Nolte, and W. Drenth, Red. Trav. Chim. Pays-Bas, 1986, 105,255.
ISSN:0306-0012
DOI:10.1039/CS9922100147
出版商:RSC
年代:1992
数据来源: RSC
|
5. |
Molecular dynamics simulations of surface chemical reactions |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 155-162
Barbara J. Garrison,
Preview
|
PDF (1735KB)
|
|
摘要:
Molecular Dynamics Simulations of Surface Chemical Reactions Barbara J. Garrison Department of Chemistry, The Pennsylvania State University, University Park, PA 16802, U.S.A. I Introduction Within the framework of classical mechanics, molecular dyna- mics simulations yield an essentially exact method for observing the dynamics of atoms and molecules during complex chemical reactions. Consequently, this technique can be used to study a wide range of dynamical events which are associated with surfaces. For example, the atomic motions which lead to the ejection of surface species during keV particle bombardment (sputtering) have been identified using molecular dynamics (MD) simulations, and these results have been directly corre- lated with various experimental observations.Often such simu- lations provide the only direct link between experimentally measured macroscopic quantities and microscopic chemical dynamics. In its pure classical form, molecular dynamics is straightfor- ward to carry out. For a system of interest, one starts with a given set of initial conditions, which include the atomic positions, velocities, and forces which are derived from interac- tion potentials. For example, to simulate a sputtering event, the initial conditions might correspond to a collection of atomic coordinates which comprise the solid surface and the incoming energetic particle (Figure 1). The dynamics are then determined by numerically solving a set of classical equations of motion. Various aspects of the dynamics, such as reaction mechanisms and product distributions, can then be determined by examining the motions of the atoms during the simulation. In this process, all that is assumed is the validity of classical mechanics and the quality of the interaction potential being used.The former approximation, although never totally true, is reasonably well understood. An indicator of the validity ofclassical mechanics is the deBroglie wavelength, h = h/Jm, of the moving parti- cle. Here h is Planck’s constant, m is the mass of the particle, and Eis its kinetic energy. For a particle moving with a kinetic energy of 1 eV, the deBroglie wavelength of the particle is 12 8, if it is an electron, 0.3 8, if it is a H atom, and 0.05 8, if it is a Si atom.Consequently in solid state materials, where typical interatomic spacings are 1-3 A,the wave nature of the electron will dominate over the particle behaviour. The electron interacts with many atoms at the same time and can easily tunnel in and out of many energy wells in the system. The motion of a 1 eV electron in a lattice, therefore, can only be described by including the quantum effects in the dynamics. On the other hand, the deBroglic wavelength of a 1 eV Si atom is so small that it can be correctly considered as a ‘point’ particle. The Si atom moves mainly in the classically allowed region of the configuration space. Therefore its motion is often well described by classical mechanics. The H atom is somewhere in between the two, and Barbaru J.Garrison received her B.S. Degree in Physics from Arizona Stute University in 1971 andher Ph.D. in Chemistry jrom the Universitj? of California, Berkelej? in 1975. A,fter a postdoc- toral position at Purdue University and u Lectureship at UC Berkelej., she joined the Chemistry faculty at Penn State Univer- sit])in 1979 us an Assistant Projessor. She is now Professor of Chemistrj? arid Head of the Department. Her research interests include computer simulutions of chemical reactions at surfaces such as ke V particle bombardment, molecular beam epitaxy, diamondfilm groir?th, Si-Fetching; many-body interactionpoten- tial development; and bonding at surfaces. quantum corrections are often superimposed on the classical description of the motion. Potential surfaces or solutions to the electronic Schrodinger equation within the Born-Oppenheimer approximation, on the other hand, are only well known for a few systems.Recently, though, there has been a significant increase in the number and quality of interaction potentials for systems of interest to surface scientists. The examples of molecular dynamics calculations of surface reactions that we could give in this article are ever increasing2 We have chosen to present work from our own group, and as a matter of convenience, have chosen examples that involve only one crystallographic arrangement of atoms -Si(OOl] and GaAs(O01) surfaces. The first chemical reaction to be discussed is keV particle bombardment.With this approach, the kinetic energy of the primary particle exceeds the binding interactions normally present in chemical bonds. Because of this energy difference, a variety of novel and intriguing chain of events is rapidly set into motion subsequent to the impact event. Atoms may be significantly displaced from their equilibrium positions with old bonds broken and new ones formed. These phenomena have led to important applications. A major impetus for this research has been, of course, in the microelectronics area, where ion implantation of dopant ions and reactive ion etching of semiconductors are hot topics. There is also interest in evaluat- ing surface processes which occur when energetic ions and molecules present in the space environment impinge on a diverse array of synthetic materials ranging from light metallic compo- sites to protective polymer overlayers deposited on non-linear optical materials.The morphology of extraterrestial surfaces is believed to be influenced to some degree by interaction with the solar wind or other energetic ions. Of interest here is the use of the angular distributions of Gaf ions ejected from molecular beam epitaxially (MBE) grown crystals as a measure of the surface structure. This leads naturally into a discussion of how the microscopic reaction mechanisms of MBE growth proceed. In MBE, atoms or molecules effuse from an oven source of material and impinge on a constant temperature ~ubstrate.~ Depending upon growth conditions the film can have the same well defined crystalline structure as the underlying substrate (epitaxial) or it can have a random bonding structure (amor- phous).The usual goal is to obtain layer-by-layer crystalline growth, for example, alternating bands of two layers of Si and two layers of Ge. Depending on the number of layers of each material the electronic properties such as the band-gap can be tailored to a desired value for a specific purpose. It is this control over the electronic properties that intrigues people in the electro- nics industry. To achieve this control, the layer growth must be almost perfect and thus it is of great interest to understand the fundamental reaction mechanisms so that we can propose new ideas for making higher quality films.The construction of electronic devices not only involves growth processes such as MBE, but also involves etching of the material. One method of doing this is by keV particle bombard- ment, but one can also etch Si by reaction with F atoms. It is easy to imagine that the F atoms would react with unsaturated Si atoms on the surface because the Si-F bond strengths are quite strong at 5.0-7.0 eV. (Si-Si bonds are 2-3 eV.) After this step, it is unclear as to how and which silicon bonds will be broken in order to produce the gas-phase product^,^ predominantly SiF,. Experimentally, it is found that during the initial exposure, the F atoms adsorb to the surface atoms forming the reaction interme- diates -SiF,(x = 1,3).5 As the F exposure is increased, the 155 intermediates form a fluorosilyl adlayer 1&20 8,thick.The -SiF and -SiF, species are thought to be located deeper in the adlayer near the silicon substrate with the -SiF, species termi- nating the surface.5 The absolute amounts of the intermediates are still unresolved and the atomic structure of the adlayer has not yet been determined. Some very interesting chemical reac- tion questions are involved with this important processing procedure. Our focus in the following sections will be on explaining the MD technique, exemplary interaction potentials, and the micro- scopic mechanisms of surface reactions of keV particle bom- bardment, MBE growth of Si, and F atom etching of Si. 2 Description of Molecular Dynamics Calculations The details of performing molecular dynamics simulations can be subdivided into two basic parts.The first is the numerical solution of the classical equations of motion. Techniques for doing this are well developed so we will only briefly mention them here.6 The other aspect of interest is the interaction potential for describing the forces among all the atoms. The development of these many-body interaction potentials is a current forefront research area. We will discuss two potentials that illustrate the different ways in which the problem can be approached. 2.1 Equations of Motion Over 300 years ago Newton formulated the basic laws of classical mechanics. Of interest here to surface chemical reac- tions is the second law, or Figure 1 Sample keV particle bombardment event of a GaAs(001) surface.The Ga and As atoms are represented by blue/purple and yellow/green spheres, respectively. The incident Ar+ ion is grey. Four frames are shown at 0, 50, 200, and 300 fs where 1 fs = 1 x 10-l5 seconds. Initially the Ar+ ion is moving towards the surface with 3 keV of kinetic energy. CHEMICAL SOCIETY REVIEWS, 1992 or mld2ri/dt2= Fi with Fi = -ViV(r1,r2,.. .r,) where for y1 atoms, mi,ri, vi, ai, and Fi are the mass, position, velocity, acceleration, and force of the thparticle. The force can be obtained from the interaction potential V(rI,rz,...r,) for all the particles. To obtain the positions and velocities as a function of time from equation 1, it is necessary to integrate the differential equations. The algorithms for doing this for the most part assume that the forces are constant over some timestep At.The solution is then propagated through time by using this approxi- mation. To start the integration, initial conditions, in this case the positions and velocities, must be specified. In the example shown in Figure 1, we start all the atoms in the solid at assumed lattice positions and give them zero initial velocities. The kinetic energy and angle of approach of the incoming particle are known from the experimental conditions, thus we know the particle’s initial velocity vector. The initial position of the incoming particle is a bit trickier to define. The coordinate above the surface must be large (usually 5-10 A) so that it mimics a real experimental source a long distance from the surface.Unfortunately, the experimentalists cannot aim the primary particle ‘head-on’ towards a surface atom or directly between two surface atoms. Experimentally then, any datum is an average over many different aiming points on the surface. Consequently, in order to compare with experiment results, it is necessary to rerun the calculations many times with different aiming points on the surface. Once the initial conditions and the interaction potential are known the classical equations of motion can be directly inte- grated. The ‘results’ of this integration are the positions and velocities of all the atoms as a function of time. The position data really represent a movie of all the atomic motions, snapshots of which are shown in Figure 1.All reaction mechanisms or atomic motions can be followed in exquisite detail. Moreover, the final positions of the atoms tell us, in particular, which ones have escaped into the vacuum and can subsequently be detected and which ones have rearranged in the solid. In the example shown in Figure 1, four particles eject, one from the first layer (yellow) and MOLECULAR DYNAMICS SIMULATIONS OF SURFACE CHEMICAL REACTIONS-B. J. GARRISON three from the second layer (blue). From the velocity vectors the energy (E= tmvov = )mv2) and direction of motion of each particle can be computed. Energy and angular distributions are often measurable in experiments.’ Of note is that both macro- scopic quantities such as energy and angular distributions and microscopic quantities such as reaction mechanisms viewed as movies can be extracted from the classical picture.The real strength of the molecular dynamics simulations comes, how- ever, when a correlation is found between the microscopic mechanisms and the quantities observed in experiments as will be discussed in Section 3. There are three additional aspects of molecular dynamics simulations that should be mentioned before discussing the interaction potentials. These are time-step limitations, tempera- ture control, and ‘how to mimic an infinite solid?’ The first two are features of all molecular dynamics simulations and the third is peculiar to the systems discussed here.As mentioned above, the critical assumption in numerical integrations is that the forces are constant over some At. The constancy of the force is determined by the interaction potential of choice. For those that apply to reactions, the acceptable timestep is about 1 x seconds or 1 femtosecond (fs). Due to the speed of computers and ultimate round-off error problems, it is only feasible to calculate 1O6 to 1 O8 timesteps, thus we are limited to simulating processes that occur shorter than 1-100 nanoseconds. This limitation is of no hindrance for keV particle bombardment (Section 3.1) but it is a problem for molecular beam epitaxy studies (Section 3.2). Mimicking an infinite solid and temperature control can be approached in many ways.Only the methods used for the examples given here are discussed. For keV particle bombard- ment a slab of arbitrary size is chosen for a few test trajectories. Particles that want to leave through the sides or bottom are allowed to do so. In the real system they would go deeper into the solid taking their energy with them. Next, we test to see if this size is sufficiently large by making the slab bigger, rerunning the same trajectories and comparing the results. We finally choose a crystal size such that the quantities of interest are converged. In the case of the keV particle bombardment, the quantities of interest are the amount and identity of the particles that eject and their energy and angular distributions. Substrate damage has not been of interest to us so it is only necessary to choose a crystal size that is sufficiently large to mimic ejection but not damage. Temperature is not a significant perturbation to this system since the ejection process is over 100-500 fs whereas vibrational periods are -1000 fs or -1 ps.A popular scheme to mimic an infinite solid is to use periodic boundary conditions with a slab of atoms as shown in Figure 2. Any particle that ‘leaves’ the left of the crystal then ‘enters’ on the right. The top surface towards the gas phase is of course free and generally the bottom surface is held rigid to keep this surface from reconstructing or rearranging. Periodic boundary con- ditions are more appropriate for processes that occur on longer timescales since small thermal motions must be taken into account. To maintain temperature control there are several layers in the slab.The top few layers are subject only to the forces from the interaction potential. The next few layers experience these same forces in addition to a force that maintains the system temperature. For example, one might include a friction term (-Pv) where the friction coefficient, P, can be positive or negative depending on whether the system is too hot or too cold relative to the desired temperature.’ This temperature control is critical for the exothermic deposition of atoms on the surface. For a real system of lo2, atoms, the excess energy will be dissipated.However, if there are only a few (several hundred) atoms, then the system will overheat or melt if some form of temperature control is not included. 2.2 Interaction Potentials The validity of the simulation rests ultimately on the ability of the interaction potential to represent accurately the true chemi- Figure 2 Top and side views of the (2 x 1) dimer reconstructed Si{OOl) surface. The Si atoms that are fourfold coordinated are depicted as light spheres. The surface atoms are represented by the dark spheres. cal interactions. For many years two-body potentials were used for simulations. The assumption in two-body potentials is that the interaction between two atoms is independent of the pres- ence of a third atom. A two-body description for H, would predict that it is three times more stable than H, whereas in reality H, is unstable.It is the development of many-body potentials for systems with hundreds of particles that makes these molecular dynamics simulations possible. For comparison purposes the two-body Lennard-Jones and Morse potential functional forms are given. In each case the total energy is written as where rliis the scalar distance between atoms i andj and I/, is the two-body term. For a Lennard-Jones potential (3) where E and u are adjustable parameters. For a Morse potential V2= D,{exp[ -2a(r -re)]-2exp[-a(r -re)]} (4) where D,,a, and re are the adjustable parameters. In both of these cases the atoms are represented by spheres and like a box of marbles, these Lennard-Jones or Morse atoms will pack in the closest array possible -a structure similar to copper metal or solid argon. Silicon and gallium arsenide, on the other hand, crystallize in a diamond lattice with each atom having four neighbours in a tetrahedral arrangement. It is mandatory then that many-body potentials be used for a realistic description of such systems.As two examples of many-body potentials we will discuss the Stillinger-Weber (SW) potential for Si/F and the Tersoff potential for Si. The SW potentials is based on the complete expansion of the total interaction potential in 2,3,4... body terms as where V3is the three-body term and four and higher body terms are omitted. In the SW work the following form is used for V,, where A, B, c,p,and q are parameters.The first term in equation 6 has a Lennard-Jones form (equation 3) and the second term is a cut off function that smoothly terminates the potential distance r,. This cutoff makes the computations more tractable than an infinite ranged potential as there are fewer interactions to evaluate. The three-body term is written as where with A, y, a,p, and p as parameters and ejik the angle centred on atom i. There are many words that can be used to describe the two- and three-body interactions but the simplest is that V, is a bond stretch and V3is an angle bend in the SW potential. The V3 term is designed to maintain the tetrahedral angle of the Si crystal structure (diamond lattice) as the value of ,8 is cos( 109.47").Of note is that this is a potential for reactions, that is, Si atoms can be added and subtracted from the surface and the potential and forces are continuous and smooth. For determining the Si-F interactions, Stillinger and Weber fitted the parameters in V,and V,to experimental data such as the lattice constant, the heat of sublimination, the elastic con- stants, and the melting temperature of Si; the energetics and geometries of SiF, molecules; and whatever else they could find. The major portion of the creative work is to obtain reasonable functional forms. Fitting the parameters can be, however, a computational challenge. In determining his potential for group IV elements (Si, Ge, and C), Tersoff9 attacked the problem quite differently.His potential does not build on the expansion in many-body terms but is in actuality explicitly a three-body potential with implicit many-body interactions. He writes the interaction potential as (9) where VRand VAare the repulsive and attractive portions of the 'pair-like' interaction. The difference between this potential and a Morse potential (equation 4) is that the coefficient of the attractive term, b,, depends on the neighbours of atom i. Tersoff assumes with and In these equations p, 6, A,, c, d, and h are parameters. The angular dependence is in the b, term but not in a straightforward manner. The difference between the Tersoff and SW angular dependences is that the Tersoff dependence does not pay a penalty for going away from a tetrahedral angle.This potential is still a functional form with parameters that must be fitted to experimental data. The Tersoff functional form allows one to fit Si, energetics as well as bulk Si properties which is not possible with the SW potential. Note that there are cutoff functions in the Tersoff potential which allow for speedier computations but these are not explicitly shown in equations 9 and 11 in order more clearly to illuminate the functional forms. CHEMICAL SOCIETY REVIEWS, 1992 The ultimate test of any potential function is how well it predicts experimental data to which it was not fitted. Tests of these and other interaction potentials are continually in progress. The challenge ultimately will be to find more flexible functional forms, more experimental data to use in fitting the parameters, or perhaps a totally different manner in which to determine the interatomic forces on a computational timescale such that the forces can be evaluated during the simulation.O 3 Dynamics of Surface Reactions The goal of using the molecular dynamics calculations is to better understand physical and chemical processes that occur in nature. Our contact to the real world is experimental data. Because of the space limitations we will concentrate here on the microscopic pictures that result from the MD calculations and will briefly mention the comparisons to experimental data. This priority is chosen to emphasize the aesthetic value of being able to visualize the microscopic world.Three MD simulations will be discussed. The first is the use of keV particle bombardment to determine structures of GaAs surfaces. The second is the molecular beam epitaxial growth of Si and the third is the F atom etching of Si surfaces. 3.1 Surface Structure Determinations of GaAs(001)(2 x 4) The bombardment of solids by particles with keV of kinetic energy provides a dramatic example of how molecular dynamics simulations can be used to understand the basic reaction process as well as aid in the determination of surface structures. To put this energy regime in perspective, remember that bond strengths are usually in the order of 3-5 eV. The energetic particle, thus, has much more energy than the attractive interaction that binds atoms in the solid.Given this imbalance in energies, at first glance it is suprising that this process yields any useful infor- mation about the surface structure. However, we feel that the combination of MD simulations along with the experimental angular distributions of the ejected particles has been one of the major accomplishments in the use of MD for modelling surface reactions. The system of interest is GaAs(OO1). There are over 30 structures of this surface depending upon the relative coverage of Ga and As. * The GaAs (001 ) surface is the template which is most widely used in the construction of microelectronic devices and has considerable technological relevance. Thus it is import-ant to be able to characterize these structures in order to gain an atomistic picture of thin film growth.The approach that the group of Winograd at Penn State has used to examine these structures is to grow the GaAs(001) surfaces in a molecular beam epitaxy (MBE) chamber. They then measure the angular distributions of the particles that are ejected as the result of keV particle bombardment. With MBE they force the top layer to be As atoms and thus the second layer contains Ga atoms. The hypothetical or bulk-terminated unreconstructed GaAs(O01} arsenic-terminated surfaces consists of a square array of As atoms (yellow) bonded to two Ga atoms (blue) in the layer below, as shown schematically in Figure 3. (Both Si and GaAs have structures in which each atom is tetrahedrally bonded to four other atoms.) Each As (or Si) atom possesses two partially filled 'orbitals' pointing upwards.Thus the As (or Si) atoms react to form dimers along this direction, doubling the lattice periodicity, as shown in Figures 2 and 3. In the labora- tory, the (2 x 1) dimer reconstruction is observed for Si but not GaAs. A number of other reconstructions appear for GaAs, but we will concentrate on the (2x4). This structure has been suggested', by other techniques to be a dimerized surface with every 4thAs-surface dimer missing, as shown on the right in Figure 3. Since the focus of this paper is on the MD simulations, we will describe this project from the theoretical rather than the experi- mental perspective.In order to determine the effect of surface structure on the keV particle bombardment process, MD simu- MOLECULAR DYNAMICS SIMULATIONS OF SURFACE CHEMICAL REACTIONS-B. J. GARRISON Figure 3 Top panel: Representations of three hypothetical GaAs(OO1) surfaces. The yellow spheres represent lSt layer As atoms; the blue spheres represent 2ndlayer Ga atoms; the green spheres represent 3rd and Shlayer As atoms; and the purple spheres represent 4thlayer Ga atoms. Bottom panel: Calculated angular distribution of 10-30 eV second layer atoms (Ga ions in the experiment) desorbed by 3 keV+ Ar+ bombardment of the corresponding surface shown in the top panel. The polar angle of emission is proportional to the distance of a spot from the centre of the circle.Ion ejection mechanisms (a), (b), and (c) are discussed in the text. lations were performed for the three surfaces of GaAs as shown in Figure 3. One of the advantages of the MD simulations is that surface structures that cannot be fabricated in the laboratory can be constructed in the simulation. The incident Ar + ion beam had 3 keV of kinetic energy aimed perpendicular to a crystal with 2184 atoms. For each surface we ran loo@--3000 different aiming points in order to obtain statistically significant results. After running the MD simulations, the angular distributions of the particles that were ejected were computed from the known velocities. Of note is that the experimentalists measure only the Ga+ ion distributions and that they know that the Ga is in the 2nd layer, thus only the angular distributions of atoms that originated from the 2nd layer are examined.14 The angular distributions of the 2nd layer atoms are shown as spot patterns in Figure 3.The polar angle of emission is proportional to the distance of a spot from the centre of the circle. The orientation is the same as the surface shown above it. The major features of the angular distributions of the Ga+ ions (or second layer atoms) appear stepwise with the com- plexity of the reconstruction. For the bulk terminated surface there are two symmetrically equivalent directions of high inten- sity (spots) in the angular distribution. These appear almost identically along the direction between the 3rd layer As atom (green) and the 2nd layer Ga atom (blue) as is shown by the arrow labelled (a).As determined from the MD simulations, the dominant mechanism of ejection is a 3rdlayer atom hitting a 2nd layer atom from behind and ejecting it. Note that there is plenty of room for the 2nd layer to escape from the surface. When the surface is dimerized (middle of Figure 3) many of the atom positions are shifted. There is now a trough between the dimer rows and the Ga atom direction is not as well collimated upon exit. Sidewings develop on the spots in the direction labelled by the (b) arrow. These sidewings become even more intense when there is a missing As surface dimer row (right side of Figure 3). In this (2 x 4) reconstruction the Ga+ ions can now escape in the direction labelled by the (c) arrow at 90" relative to the major spots.The experimental measurements of these angular distri- butions agree quite nicely with the calculated distributions and show conclusively that the spot along the (c) direction is a signature of a missing surface As dimer.14 The goal now is to start examining some of the other 29 proposed structures and also surface structures with small amounts of A1 on the surface (precursors to Schottky barriers). There is one issue that has not been mentioned. The experi- ments were performed on a GaAs crystal. The calculations were performed using Tersoff s interaction potential for Si. Why can we compare (successfully!) these two seemingly different sets of data? It is our belief, based on years of experience, that many of the processes and consequently the experimental results in the keV particle bombardment process are dominated by the rela- tive positions of the atoms.This highly energetic process is not so sensitive to the precise chemical interactions. Silicon and GaAs have virtually identical structures and thus very similar angular distributions to keV particle bombardment. The same conclu- sions hold true for metals. The angular distributions for all face- centred-cubic metals display the same predominant features. Details such as peak positions, intensities and widths do vary, however, with the chemical nature of the system. Chemical effects show up much more dramatically in processes such as MBE growth of Si and F atom etching of Si which will be discussed next.3.2 Molecular Beam Epitaxial Growth of Si{001)(2x 1) Molecular-beam epitaxy (MBE) is widely employed for semi- conductor epitaxial film gr~wth.~ The quality of the electronic device is related to the perfection of the layers of material. One of the problems with obtaining layer-by-layer growth is the fact that atoms in the surface layer of most semiconductors are significantly displaced from the bulk configuration. As discussed above, the surface atoms on the bulk terminated Si(OO1) face (Figure 2) tend to move closer to the neighbouring atoms and form rows of stable dimer structures. These rearrangements do not only constitute the basis of esoteric surface physics experi- ments, but are also extremely important from a microelectronic device point of view.In order for epitaxial growth to occur on a reconstructed surface (Figure 2), the atoms in the dimerized surface layer must reorder back to their original bulk positions, so that the atoms in the deposited layer are in the expected bulk positions for the next higher layer. It is the question of the dimer opening that we wish to assess with the MD technique. The molecular dynamics simulation’ consisted of 10 layers of silicon with 32 atoms per layer as shown in Figure 2. As mentioned above, the atoms in the bottom layer were anchored in position with the atoms in the next four layers f0rrning.a stochastic region.The atoms in the top five layers and all the deposited atoms were considered ‘real atoms’ and moved only under the influence of the interaction potential. The deposition rate of thermal energy Si atoms, at I atom per 2-3 ps (1 ps = second), was sufficiently slow as to allow equilib- ration of the system before the arrival of the next atom. It has been pointed out in an earlier study16 that if the deposition rate is too fast to allow for complete dissipation of the kinetic energy from exothermic adsorption reactions, the excess energy in the interaction region effectively destroys the structure of the inter- face between the originally reconstructed surface and the layers of the deposited material. The effect of long-time equilibration after the deposition, in that case, is like the simulation of crystallization from a liquid state.After each deposition of 1.5 monolayers of atoms the full system was equilibrated for 0.5 to 1 ns. The dynamic behaviour of the deposited and the substrate atoms revealed microscopic and macroscopic features of the growing film. The predominant adsorption process is the attachment of the incoming Si atom to one of the radical orbitals on the surface dimer Si atoms. In the left frame of Figure 4a, Si adatoms are shown on the radical orbital of each of the surface dimer atoms. (The third Si adatom is attached to an adjacent Si dimer atom and is not in the gas phase.) The microscopic mechanisms of dimer openings can be broadly categorized into two types.I5 First is the diffusing adatom induced mechanism, typical examples of which are shown in Figures 4(a,b).The Si adatom moves as shown by the arrows in Figure 4a during the equilibration period (0.5-1.0 ns). Initially both the radical orbitals of an isolated dimer are saturated with two adatoms. A third adatom diffuses over a period of hundreds of picoseconds, to move closer to the dimer, and pushes the adatom on the radical orbital into the epitaxial position. The diffusing adatom itself then occupies the vacant radical orbital of the open dimer. The three adatoms finally occupy the epitaxial positions, whereas the surface dimer atoms have relaxed back to their bulk positions. Another observed mechanism of the diffusing adatom induced dimer opening is shown in Figure 4b.Instead of pushing the adatom on the radical orbital into the epitaxial position, in this case, the CHEMICAL SOCIETY REVIEWS, 1992 Figure 4 Diffusing adatom motion induced mechanisms of dimer open- ings. Only a subset of the atoms are shown. The same colour scheme as in Figure 2 is used. The deposited Si adatoms are shown as various shades of peach spheres. (a) Epitaxial growth. (b) Defect formation. (c) Direct dimer opening from adatom deposition. diffusing adatom can move underneath the adatom on the radical orbital and push the dimer atom into the epitaxial position. Here the diffusing adatom has moved into the top substrate layer whereas the substrate dimer atom has moved up into the epitaxial layer.The end products, as shown in Figures 4a and 4b, in the homoepitaxy of Si or Ge are identical. However, in case of the heteroepitaxy of Ge on Si this second mechanism will cause the formation of a defect at the interface. This defect is formed during the growth event and is not due to interlayer diffusion during an annealing process. In other words, the final configuration in Figure 4a does not convert into the one in Figure 4b and vice versa during annealing. The atoms in a surface dimer are constantly vibrating about their equilibrium positions with amplitudes determined by the temperature of the substrate. The second mechanism of dimer opening occurs when an incoming adatom is directly inserted into the available epitaxial position, thus stabilizing the opening.We found that the direct insertion of the adatom into epitaxial positions can occur on a bare dimer (Figure 4c), on a dimer with one radical orbital occupied, and also on a dimer with both of its radical orbitals occupied. Most of such direct insertion mecha- nisms occurred during the deposition process, i.e. the surface was relatively clean. The number of such occurrences increased with an increase in the temperature of the substrate. The computations described in this section are quite extensive by today’s standards as they took the equivalent of over a month of dedicated time on an IBM 3090 computer. There were several hundred atoms in the system and a total time of 3.5 x seconds was simulated for the growth of 4-5 layers of material.However, the experimental conditions are such that a layer is grown in 1 second to 1 minute! For the near future these times will be impossible to simulate with MD because in MD we must use timesteps on the order of 10-seconds. The main physical process that is missing in the simulation is diffusion of the MOLECULAR DYNAMICS SIMULATIONS OF SURFACE CHEMICAL REACTIONS-B. J. GARRISON deposited atoms on the surface. In fact, the atoms do move many lattice positions before finally incorporating into the crystal. Other techniques such as transition state theory are now being applied to examine these long term effects." These results of the transition state calculations compare favourably with the observed activation energies of diffusion and the direction anisotropy.18 3.3 F Atom Etching of Silicon The atomic mechanisms responsible for the formation of the gas-phase products observed during the etching of silicon by fluorine atoms and the structure of the surface during the etching process are quantities of interest in the Si-F study.The Still- inger and Weber potential energy function which describes the interaction between Si and F atoms has been used.s For the purpose of simulating the initial stages of fluorine adsorption on the Si{ 100)(2x 1) surface, a microcrystallite of ten layers of silicon atoms with 32 atoms per layer and horizontal periodic boundary conditions was used. The surface atoms were arranged into the dimer reconstructed surface and a F atom was attached to each dangling bond.Estimates of the kinetic energy of F atoms in plasmas are between thermal and 8 eV,20 thus in order to overcome the barriers for reaction, F atoms with 3 eV of kinetic energy were used. A total of 200 F atoms were eventually adsorbed onto the surface. At this point a steady-state condition is reached such that the rate of deposition of F atoms equals the rate of F atoms leaving the surface in products such as SiF,. To our knowledge this is the first molecular dynamics calculation in which a near steady-state reaction condition of an observed macroscopic process has been simulated. Figure 5 F atom etching of Si. The yellow spheres represent non-fluorinated Si atoms, the green, mauve, red, and light blue spheres represent mono-, di-, tri-, and tetrafluorinated Si atoms.The dark blue and black small spheres represent F atoms. (a,b) Top and side views of the surface at the end of the simulation. (c-f) Formation of a gaseous SiF, molecule from the reaction between the incoming fluorine atom (black sphere) with an SiF, adspecies. The surface configurations during the simulation were exa- mined in order to elucidate the possible structure of the experi- mentally observed fluorosilyl layer. In Figure 5(a,b) we display the surface configuration after the last fluorine adsorption event. At this time almost two of the original layers of silicon atoms are etched. Only the top five layers of the crystal are depicted. Upon examination of this surface, there are many scattered vacant regions which are created by the etching of silicon atoms.The vacancies are scattered inhomogeneously across the surface due to the random nature of the etching process. Moreover, the surface configuration has evolved such that it is difficult to locate the position of the dimer rows of the initial defect free surface. Some interlayer mixing has also occurred. More information regarding the typical surface structure is revealed when the surface configuration is examined for local structure. For instance, some pairs of adspecies dimerize while others retain approximately the bulk geometry of the silicon lattice. In addition to the bonding between pairs of adspecies, there are groups of three adspecies where the bonding direction is nearly perpendicular to the surface plane, forming a tower-like struc- ture.These structures are attached to the solid through the SiF and SIF, moieties and can either still be in the solid or protrude into the vacuum. One tower with the composition SiF-SiF-SiF, (green-green-red spheres) appears in the middle of Figure 5b. For all the tower structures, the SiF and SiF, are adspecies which lengthen the tower whereas the SiF, are adspecies which terminate the tower at the vacuum-surface interface. This observation would suggest that a type of layering occurs while the surface is being fluorinated. The fraction of fluorine in the different layers at the end of the simulation suggests that the first layer silicon atoms are mainly SiF, adspecies, the second layer silicon atoms are mainly SiF adspecies while the third and fourth layers are just beginning to be fluorinated.Thus, a sequential fluorination appears to be occurring in the crystal, starting with the first layer and progressing deeper into the crystal. This observation of the SiF, layering agrees with conclusions drawn from surface science experiments. The results from the MD calculations illustrate a mechanism responsible for the major gas-phase product, SiF,, as is shown in Figure 5. The initial configuration is a SiF, adspecies with its Si-F bonds directed into the vacuum. As the fluorine atom approaches the silicon atom, the Si-F bonds invert from extending into the vacuum (Figure 5c) to being planar with the silicon atom (Figure 5d) to being directed towards the surface (Figures 5e,f).During the reaction the Si-Si bond lengthens and breaks. This umbrella type motion of the Si-F bonds typifies the SN2 reaction as suggested by ab initio electronic structure calculations.2 As a final note, we have proposed that the tower- like features are the precursors to the etch products Si,F, and Si,F,. 4 Conclusion Molecular dynamics simulations have an attractive appeal in that they can supply pictures of microscopic reaction events that seem eminently plausible. The ultimate challenge is to determine their connection to nature. Consequently there should be conti- nual interaction of the computer simulators with the experimen- talists.Where possible, calculated results should be compared with experimental data. The experimentalists should look upon the results of the calculations as challenges to prove or disprove. Over the years we have found that this synergistic approach has led to the understanding of surface reactions to a greater depth than would have been achieved if either the theorists or experi- mentalists were working in isolation. The use of molecular dynamics’and other types of computer simulations will continue to grow. Computers are becoming even faster. The graphics for visualization of the results are becoming better and easier to use. The interaction potentials are becoming more sophisticated, and, as they are tested and improved, their reliability becomes even better.Finally, we have chosen three examples from our own work to highlight in this article. We by no means wish to imply that others are not pursuing similar directions, either for surface reactions or reactions in other media such as liquids, biological compounds, and polymers. Acknowledgements. The ventures into the detailed understand- ing of the keV particle bombardment process were initiated in 1977 in collaborations with Nicholas Winograd of Penn State University and the late Don E. Harrison, Jr. of the Naval Postgraduate School. It was the interaction with these two people that propelled the combined theoretical/experimental approach towards a deeper understanding of the fundamental processes than either approach could have achieved in isolation. Our research programme ventured into the lower energy pro- cesses through the initiative of Donald w.Brenner who is now at the Naval Research Laboratory. To these three people I owe a big debt of gratitude. Throughout the years there have been numerous students and postdoctoral associates that have contri- buted to the research projects. The people whose work is present here include Deepak Srivastava, Tracy Schoolcraft, Kevin Caf- fey, and Rik Blumenthal, all of Penn State, and Roger Smith (University of Loughborough). Four current collaborators, Ramona Taylor, Eric Dawnkaski, Recaldo Carty, and Dan Bernard0 have been especially helpful with the colour graphics. Finally I am grateful for the financial support of the National Science Foundation, the Office of Naval Research, the IBM Programme for the Support of Materials and Processing Sciences, and the Camille and Henry Dreyfus Foundation. Two people at these agencies, Larry Cooper (ONR) and Henry Blount (NSF), have been particularly supportive over the years.CHEMICAL SOCIETY REVIEWS, 1992 Penn State University has supplied a generous grant of computer time for these studies. 5 References 1 N. Winograd and B. J. Garrison, in ‘Ion Spectroscopies for Surface Analysis’, ed. A. W. Czanderna and D. Hercules, Plenum Press, 1991, pp. 45-141 and references therein. 2 For example, U. Landman, W. D. Luedtke, N. A. Burnham, and R. J. Colton, Science, 1990, 248,454; R.A. Stansfield, K. Broomfield, and D. C. Clary, Phys. Rev., 1989, B39, 7680; T. J. Raeker, D. E. Sanders, and A. E. DePristo, J. Vuc.Sci.Technol., 1990,AS, 3531;J. C. Tully, J. Chem. Phys., 1980,73,6333; M. S. Daw and M. I. Baskes, Phys. Rev. Lett., 1983,50, 1285; D. W. Brenner, B. I. Dunlap, J. W. Mintmire, R. C. Mowrey, and C. T. White, ‘Proceedings of the Second International Conference on New Diamond Science and Technology’, ed. by R. Messier and J. T. Class, MRS International Conference Proceedings Series, Materials Research Society, Pitts- burgh, PA, 1991, p. 39. 3 J. C. Bean, Phys. Toduy, 1986,39, 36. 4 F. A. Houle, J. Appl. Phys., 1980,60,3018; H. F. Winters and F. A. Houle, J. Appl. Phys., 1983,54, 1218; F. A. Stevie and M.J. Vasile, J. Appl. Phys., 1982, 53, 3799; and H. F. Winters and I. C. Plumb, J. Vuc. Sci.Technol., 1991, B9, 197. 5 F. R. McFeely, J. F. Morar, N. D. Shinn, G. Landgren, and F. J. Himpsel, Phys. Rev. B, 1984, 30, 764; F. R. McFeely, J. F. Morar, and F. J. Himpsel, Surf Sci., 1986,165,277;C. W. Lo, D. K. Shuh, V. Chakarian, K. A. German, and Y. A. Yarmoff, submitted to Phys. Rev. B. 6 W. G. Hoover. ‘Molecular Dynamics’, Lecture Notes in Physics #258, Springer-Verlag, Berlin, 1986; D. W. Heermann, ‘Computer Simulation Methods in Theoretical Physics’, Springer-Verlag, Ber- lin, 1986. 7 H. J. C. Berendsen, J. P. M. Postma, W. F. van Gunsteren, A. Dinola, and J. R. Haak, J. Chem. Phys., 1984,81, 3684. 8 F. Stillinger and T. Weber, Phys.Rev., 1985, B31, 5262; F. H. Stillinger and T. A. Weber, Phys. Rev. Lett., 1989, 62, 2144; T. A. Weber and F. H. Stillinger, J. Chem. Phys., 1990,92, 6239. 9 J. Tersoff, Phys. Rev. Lett., 1988, 61, 2879; J. Tersoff, Pliys. Rev., 1989, B39, 5566. 10 R. Car and M. Parrinello, Phys. Rev. Lett., 1985, 56, 2471. 1I J. H. Neave and B. A. Joyce, J. Cryst. Growth, 1978, 44, 387. 12 R. Blumenthal, S. K. Donner, J. L. Herman, R. Trehan, K. P. Caffey, E. Furman, and N. Winograd, J. Vuc. Sci.Technol., 1988, B6, 1444; R. Blumenthal, Thesis, The Pennsylvania State University, 1990. 13 M. D. Pashley, K. W. Haverern, W. Friday, J. M. Woodall, and P. D. Kirchner, Phys. Rev. Lett., 1988, 60, 2176; D. K. Biegelsen, R. D. Bringans, J. E. Northrup, and L. E. Swartz, Phjs. Rev. B. 1990,41, 5701. 14 R. Blumenthal, K. P. Caffey, E. Furman, B. J. Garrison, and N. Winograd, Phys. Rev. B, 1991,44, 12 830; R. Smith, D. E. Harrison, Jr., and B. J. Garrison, Php. Rev. B, 1989,40, 93. 5 D. W. Brenner and B. J. Garrison, Surf. Sci.. 1988, 198, 151; D. Srivastava, B. J. Garrison, and D. W. Brenner, Phys. Rev Lett., 1989, 63,302; D. Srivastava, B. J. Garrison and D. W. Brenner, Langmurr, 1991, 7, 683. 6 R. Biswas, G. S. Grest, and C. M. Soukoulis, Phys. Rev. B, 1988,38, 8 154. 7 D. Srivastava and B. J. Garrison, J. Chem. Phys., 1991,95,6885; A. F. Voter and J. D. Doll, Phys. Rev. B, 1986, 34, 6819. 18 R. J. Hamers, U. K. Kohler, and J. E. Demuth, Ultrarnicroscupj., 1989,31, 10; M. G. Lagally, R. Kariotis, B. S. Swartzentruber, and Y. W. Mo, Ultramicrascopj, 1989, 31, 87; A. J. Hoeven, J. M. Lenssinck, D. Dijkkamp, E. J. Van Loenen, and J. Dieleman, Phys. Rev. Lett., 1989, 63, 1830. 19 T. A. Schoolcraft and B. J. Garrison, J. Vuc. Sci. Technol. A, 1990,8, 3496; ihid.. J. Am. Chem. Soc., 1991, 113, 8221. 20 T. J. Sommerer and M. J. Kushner, J. Appl. Phj..~.,1990, 70, 1240. 21 B. J. Garrison and W. A. Goddard 111, Phys. Rev. B. 1987,36,9805.
ISSN:0306-0012
DOI:10.1039/CS9922100155
出版商:RSC
年代:1992
数据来源: RSC
|
6. |
Magic numbers in molecular clusters: a probe for chemical reactivity |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 163-169
M. Todd Coolbaugh,
Preview
|
PDF (954KB)
|
|
摘要:
Magic Numbers in Molecular Clusters: A Probe for Chemical Reactivity M.Todd Coolbaugh and James F. Garvey” Acheson Hall, Department of Chemistry, State University of New York at Buffalo, Buffalo, NY 14214, U.S.A. 1 Introduction Clusters, which are gas phase aggregates consisting of two to several thousand atoms or molecules, have attracted a great deal of attention in recent years. This is due to the fact that these weakly bound species exist as a state of matter intermediate between the gas and condensed phases (solid and Much of the recent activity in cluster science has been directed towards measuring and rationalizing the evolution of various physical properties, such as ionization potentials (IP), as a function of cluster size. The purpose of such studies is to develop a better understanding of the relationships between the proper- ties of the isolated (gas phase) molecules and the bulk properties of liquids and solids.The overall goal of much of this research is the development of a microscopic understanding of solvation effects. As a result, cluster research is attracting a growing number of researchers and cluster science is a rapidly developing discipline. Studying chemical reactions in clusters allows one then, in principle, to study the transition from bimolecular reactivity to bulk phase reactivity, by observing reactions in successively larger clusters. At the same time an understanding of the factors which govern the structures of finite clusters may provide insight into the microscopic structure of bulk solvent medium.It is becoming clear that clusters present the experimental chemist with an unparalleled opportunity to investigate rather complex chemical processes in environments of greatly reduced complexity. This review will discuss several aspects of novel chemistry within van der Waals clusters. We will first briefly outline the experimental methodology utilized in the production and detec- tion of clusters, emphasizing the importance of various experi- mental parameters. Following this, a section will be devoted to a discussion of the origins and importance of ‘magic number’ clusters. Finally, several examples of the use of magic numbers in elucidating cluster chemistry taken from our own work will be presented. 2 Experimental Methodology 2.1 The Production of van der Waals Clusters Adiabatic expansions are the most widely utilized method for the generation of weakly bound van der Waals clusters.In this technique the species to be clustered is allowed to expand from a region of high pressure into a region of low pressure through a small orifice, i.e. a molecular beam nozzle. The species to be M. Todd Coolbaugh was born in Elmira, N Y in 1962. He received his B.A. from AIfred University in 1985 and is currently complet- ing his Ph.D. work at SUN Y/%uflalo. James F. Garvey was born in Passaic, NJ, in 1957. He received his B.S.1M.S. from Georgetown University in 1978 andhis Ph.D from the California Institute of Technology in 1985 under A ron Kupper- mann.He then went on to become apost-doctoral fellow at UCLA under Richard Bernstein, where at that time he became interested at looking at ion-molecule reactions within clusters. In 1987 he joined the faculty at SUNYIBuflalo. This past year he was appointed as an associate professor and was awarded an AIfred P. Sloan Fell0 wsh ip. clustered is often seeded into an inert carrier gas and clusters are formed in the low temperature, high collision frequency environment found in the early stages of the expansion, as discussed by Kappes and Leutwyler. lo These molecular beam cluster sources have the advantage of producing very intense beams of clusters but also suffer from the disadvantage of producing a wide distribution of clusters of various sizes.The distribution of clusters generated is highly dependent on the experimental conditions of the expansion. The present understanding of clustering in adiabatic expansions is such that no more than qualitative conclusions concerning the cluster distributions produced under any given set of experimen- tal conditions can be made. The most important experimental parameters are the nozzle orifice diameter (d), expansion (or stagnation) pressure (Po),and expansion/stagnation tempera- ture (To).In general, larger nozzle diameters, higher expansion pressures, and lower stagnation temperatures all shift the overall cluster distribution towards larger cluster sizes. It is also expected that the overall width of the cluster distribution is proportional to the average cluster sizelo (i.e., the larger the average cluster size, the broader the overall distribution of sizes).2.2 Cluster Detection At present there is no generally applicable, convenient method of selectivity detecting neutral clusters by size. As a consequence of this limitation, the great majority of cluster studies have employed mass spectrometric detection which offers the advan- tages of high sensitivity and size selection following the initial ionization event which produces the cluster ion. Electron impact, single photon (e.g. synchrotron radiation), and multi- photon (laser) ionization are among the more common modes of ionization currently utilized in cluster mass spectrometry.It is now recognized that ionization of a distribution of neutral clusters leads to a distribution of substantially smaller cluster ions and this evaporative process will be discussed in the next section. Since neither the neutral cluster distribution, nor the cluster ionization cross sections are generally available, the cluster ion distributions measured via the spectrometer cannot be quantitatively related to the original neutral cluster distributions. Ionization of a neutral molecule within the cluster may also trigger complex bond cleavage and reformation reactions. Although this serves further to complicate the already ‘delicate’ relationship between the generated ion clusters and the original neutral clusters, these ‘intracluster ion-molecule reactions’ are proving to be of great interest to the chemical community.’- Many of the ions produced in molecular clusters bear close resemblance to intermediates encountered in condensed phase reactions so that a study of these reactions serves to provide a direct link between gas and solution phase chemistry.3 The Cluster Mass Spectrum and Magic Numbers A cluster mass spectrum (CM will normally consist of one (or more) series of evenly spaced m ss peaks, with the peak spacing 3 corresponding to the monomer mass. The most basic piece of information available from the CMS is the mass to charge ratio (m/z)of the cluster ion. From this (and a general knowledge of the composition of the neutral clusters) it is normally possible to assign accurate empirical formulae to the observed cluster ions.163 The chief drawback to any mass spectral experiment, however, lies in the fact that the m/z ratio does not directly provide information concerning the structure of the ions, or their origins. Fortunately, valuable insight concerning the structures, stab- ilities, and sometimes the processes giving rise to particu- lar cluster ions may often be obtained from examination of the cluster ion intensity distributions, i.e. the abundances of the cluster ions as a function of the cluster size. For this reason characterization of cluster ion intensity distributions has been a central theme in much of the cluster literature. In general, cluster ion intensity distributions are found to vary rather smoothly as a function of cluster size (with the overall intensities dropping off in an exponential fashion as one goes to larger cluster sizes).Anomalous intensities or abrupt changes in the forms or shapes of cluster ion intensity distributions which occur at specific clusters sizes have been termed 'magic numbers' and have provided the key to understanding a number of cluster systems. 3.1 The Origin of Magic Numbers The origins of magic numbers have been the subject of much discussion over the years. In some early reports it was suggested that magic numbers reflected abundance anomalies in the neutral cluster distributions. This viewpoint has been aban- doned since it is now recognized that ionization of neutral clusters nearly always leads to extensive fragmentation. 423 424 425 426 427 428 rn Iz Figure 1 A representative portion of a typical 70 eV electron impact mass spectrum of the ammonia cluster beam. This portion of the spectrum corresponds to the four different cluster ion species contain- ing 25 nitrogen atoms.This sequence of four peaks is observed throughout the entire cluster mass spectrum as a function of n,spaced by 17 amu. The abscissa is scaled to the nominal masses of the clusters. The fourth peak (at mjz = 427, not labelled) is attributed to l5N contribution from peak 1, water impurity, and/or intracluster reaction. (Reprinted with permission from reference 1 1. Copyright 1989, Ameri- can Institute of Physics.) It is now generally accepted that the cluster ion distributions of weakly bound clusters are a reflection of the stabilities of the cluster ions.* Magic numbers therefore are nearly always asso- ciated with some abrupt change in the stepwise binding energies of individual monomer units to the cluster ion.Perhaps the most cited example is ammonia clusters where the CMS consists of a group of four peaks each separated by one mass unit, as shown in Figure 1. This set of four peaks is repeated throughout the CHEMICAL SOCIETY REVIEWS, 1992 CMS, regularly spaced by 17 mass units. The peak labelled P corresponds to the parent cluster ion {NH3)i, while the peaks one mass unit above and below (a and 1) correspond respectively to the loss or gain of a hydrogen atom.Peak a corresponds to simple fragmentation of an excited NHS ion within the cluster to form a solvated NH; cluster ion. Peak 1 comes about from a exoergic intracluster ion-molecule reaction between the NHS ion and one of the NH, monomers to form NH, and a solvated protonated ammonia ion as shown in reaction 1. It was long suspected that the ion 'core' of these protonated ammonia clusters should be the ammonium ion, NH:. Figure 2a shows a plot of (NH,),H + as a function of n. A prominent magic number is observed at n= 5 which corresponds to the (NH,),NH: cluster ion. The magic number could be explained in terms of the completion of the first solvation shell about the NH; ion by four NH, molecules directly hydrogen bonded to a central NH: ion as indicated in Figure 2b.In fact, most magic numbers in hydrogen bonded cluster ions have now been shown to arise as a consequence of solvation shell closures. t= 0 5 10 15 20 25 n Figure 2a A log plot of the intensity of {NH,},H+ clusters obtained from 70 eV electron impact ionization of a neat ammonia cluster beam:Po = 2.2 atm., To= 293 K. 1 / (Z) Figure 2b Structure for the {NH,},NHt cluster ion. The shaded circles correspond to nitrogen atoms while the open circles correspond to hydrogen atoms. This structure represents a protonated ammonia ion surrounded by a complete solvation shell of ammonia molecules. MAGIC NUMBERS IN MOLECULAR CLUSTERS-M. T. COOLBAUGH AND J. F. GARVEY The physical origin of magic numbers may be traced to the kinetics of the various processes taking place subsequent to the ionization event.8 Ionization of cold neutral clusters leads to the production of internally excited, i.e.'hot', cluster ions. This excess energy results from differences in the structures of the neutral and ion clusters; exoergic intracluster ion-molecule reactions may also contribute. This excess internal energy is dissipated by the loss of monomers from the cluster in a process which may be likened to evaporative cooling generating a smaller, cooler cluster ion. The appearance of magic numbers is a direct consequence of the kinetics of the fragmentation reactions following ionization and any subsequent ion-molecule reactions. It is now believed that many monomers are lost from the cluster following ioniza- tion.The kinetics of these monomer evaporations are, as a result, quite sensitive to variations in the binding energies within the cluster and are therefore the size determining reactions, on the timescale of mass spectroscopic detection. Essentially, clus- ters with lower binding energies will be characterized by faster dissociation rates than those of higher binding energies and will thus be observed with lower intensity. Magic numbers thus signal the existence of particularly stable cluster ions, or sudden changes in the stepwise binding energies. 4 Results In the following sections we will discuss several examples of the types of insight provided by the observation of magic numbers which have been taken from work conducted in Buffalo.Magic numbers observed in the CMS of ammonia and dimethyl ether will provide examples of magic numbers as well as demonstrat- ing the way in which these observations can provide insight into the reactive processes initiated by ionization of a cluster. The CMS of ethene and 1,l-difluoroethene also provide interesting examples of magic numbers which suggest the possibility that clusters may provide useful media in which to study the basic processes of ionic polymerization. The experimental setup has been described in detail pre- viously" and is shown schematically in Figure 3. Briefly, it consists of a continuous molecular beam cluster source of the Campargue type.A 250 pm sonic nozzle was employed in all the experiments reported below. The nozzle assembly is con- nected to a circulating chiller to allow control and variation of the temperature of the nozzle and the gas stagnation region Figure 3 Schematic side view of the differentially pumped cluster beam apparatus and quadrupole mass spectrometer. The temperature of the nozzle and stagnation region is regulated by a circulating chiller. (Reprinted with permission from reference 1 1. Copyright 1989, Ameri- can Institute of Physics.) immediately behind it. The mass spectrometer (Extrel C50 unit mass resolution up to 1500 amu) is equipped with an electron impact ion source and a channeltron particle detector. 4.1 Ammonia Clusters Ammonia is one of the most extensively studied cluster ion systems and represents one of the first cluster ion systems for which the link between the observation of magic numbers and cluster ion stabilities was established.The CMS of ammonia are dominated by the protonated cluster ions, {NH,},H +. Figure 2a displays the ammonia cluster ion intensity distribution obtained with expansion conditions favouring extensive cluster forma- tion. The magic number at n = 5 is clearly evident. Spectroscopic studies' have quite convincingly demonstrated that this ion may be regarded as an NH,f ion to which four NH, molecules are directly hydrogen bonded as shown in Figure 2b. Additional spectroscopic studies14 have been interpreted in terms of an additional solvation shell closing at n = 9, i.e.{NH,),NH;.This interpretation is supported by the mass spectral data of Figure 2a where an additional large drop in intensity is observed at n = 9. In addition to the protonated clusters, several other series of cluster ions are observedg,' as demonstrated in Figure la. The cluster ions labelled a in Figure 1a are particularly interesting in that the m/z ratios allow the empirical formulae {NH,), -,NH; to be assigned to these ions. The NH; ion may be formed via fragmentation of an ammonia cation within the cluster, but is it correct to view these ions as NH; ions solvated by ammonia molecules? Consideration of the IPS of NH, (1 1.14 eV) and NH, (I 0.16 eV) makes this highly unlikely. {NH,}, -,NHi must be rejected as representing the structure of the cluster ions under consider- ation.Stephan et a1.I6 reported the observation of an ion at m/ z = 33 (NH,NH:) in an electron impact ionization study of ammonia clusters and reported an appearance energy for this ion of 15.6 f0.3 eV. It was suggested that this ion represented an N,H: ion produced by an intracluster ion-molecule reaction of an electronically excited NH:* ion. Figure 4a provides evidence, in the form of cluster ion intensity distributions, which supports this suggestion. These distributions show a prominent magic number at n = 7 for (NH,),- ,NH;. Figure 4b displays the structure we propose for the magic number cluster which may be thought of as a protonated hydrazine molecule with a complete solvation The prominence of the magic number can be rationalized in terms of the highly exoergic nature of the ion molecule reaction which forms the protonated hydrazine (approximately 4.5 eV exothermic). This is reflected by the tendency of the highly excited N,HZ ion to 'shake off' NH, solvent molecules which are outside of the first solvation shell. I I 20cm ).I I I I I -+-+electric Iilower package turbo-pihip gate 1 tt1rbo-purn p 1000 m3-h 360 1-s va 1ve 360 1-s 166 Electron Energy Dependence 40000 A --70eV 100eV 4-9 -40eV 30000 20000 h(I) 10000 c.-C3 0 I ., ~ , ~ , . ,'I' .,2 v 5 7 9 11 13 15 17 19 2..-c (I) Nozzle Temperature Dependence Q) C-c 1 t -40000 * ---293K 253K 273K -+ 313K 30000 20000 10000 Figure 4a A plot of the intensities of the {NH,), -,NHz cluster ions as a function of electron impact energies (top panel, T = 273 K, and as a function of nozzle temperatures (bottom panel, electron impact energy = 70 eV).(Reprinted with permission from reference 1 1. Copyright 1989, Ameri-can Institute of Physics.) Figure 4b Structure proposed for the {NH,j,NH: cluster ion. The shaded circles correspond to nitrogen atoms while the open circles correspond to hydrogen atoms. This structure represents a proto- nated hydrazine ion (within the circle) surrounded by a complete solvation shell of ammonia molecules. 4.2 Dimethyl Ether Clusters Figure 5 displays a portion of a typical CMS of dimethyl ether (DME; rnjz = 46 amu) clusters.The dominant cluster ions correspond to the protonated clusters, DME,H . Several series + of 'fragment' ions are also observed in the CMS. The two most important of these are found at masses given by 46n + 45 CHEMICAL SOCIETY REVIEWS, 1992 M,CH,OCH~ M,H+ n h c c I I I I I I I I I I I 1 I I I I I I I I 100 110 120 130 140 mlz Figure 5 A representative portion of the electron impact mass spectrum of a dimethyl ether cluster beam. (Reprinted with permission from reference 19. Copyright 1990, Ameri-can Chemical Society. j and 46n + 15. These may be assigned empirical formulae (DME),C,H,O+ and {DME),CH:, respectively. Based on the known gas phase ion chemistry of DME" the most likely assignments of these ions are {DME},CH,OCH'; and {DME}, -[(CH,),O+].The expected relationship between these ions is shown in equations 2 and 3: {DME}, + e---* {DME},_ ,CH,OCH; + H + 2e-(2) ~{DME},-,CH,OCH: -+{DME),, 2(CH3j30+ (3)+ CH20 The bimolecular reaction between the methoxymethyl cation CH,OCHl and DME giving rise to the trimethyloxonium ion (CH3)30+ is well known from high pressure mass spectrometric studies.' Both of these cluster ions have been observed in SIFT studies of clustering reactions in DMEls and the relationships depicted in equations 1 and 2 have been confirmed by appear- ance energy measurements. In addition to the 'fragment' ions just discussed, another series of cluster ions are observed at masses given by 46n + 33.The most reasonable empirical formulae for these ions are given by (DME},,CH,O+. Figure 6 displays the intensity distribution for these ions observed, in this case, at several different electron impact energies. A magic number is observed in these distribu- tions at n = 2. On this basis, we have propo~ed~?'~ that the 46n + 33 ions may be assigned as methanol ions solvated by DME ions, i.e. {DME),CH,OHt. The observation of these ions is particularly intriguing since there are no reports of any gas phase reactions of DME with DME+ or any of its fragment ions giving rise to methanol ions. The reactions(s) giving rise to the {DME),CH,OH: ions may be related to the catalytic conversion of methanol into hydrocar- bons.It is known that the initial step in this process involves dehydration of methanol to give DME but to date, the mecha- nism of the initial C-C bond formation remains to be unequivo- cally established.20 The initial step of the catalytic conversion, i.e. dehydration of methanol to DME, has been observed in methanol cluster ions.21 It may be noted that acidic sites on the zeolite catalyst are believed to play an important part in the catalytic process and thus it may not be surprising that similar reactions are observed in protonated cluster ions. Several ion-molecule reactions in the DME system have been reported in which methanol was suggested as the neutral pro-duct. Methanol was a suggested product arising from the collisional activation of the trimethyloxonium ion:22 (CH,),O+ +C,Hl+ CH,OH (4) We have observed that the intensity of the IDME},,.(CH,),Of ions fall off very rapidly with increasing cluster size MAGIC NUMBERS IN MOLECULAR CLUSTERS-M. T. COOLBAUGH AND J. F. GARVEY evM,CH~OH~+ ;;ev ---t 30 eV ev eV eV ev - - 0 2 4 6 8 10 n = # of DME molecules M2CH30Hzf Q W Figure 6 Top panel: Plot of the intensities of the {(CH,),O),CH,OH$ clusters at several electron energies. Bottom panel: Structure pro- posed for the {(CH,),O),CH,OH: cluster ion. The shaded circles correspond to oxygen atoms while the open circles correspond to hydrogen atoms and the black circles correspond to carbon atoms. This structure represents a protonated methanol ion solvated by two dimethyl ether molecules.(Reprinted with permission from reference 19. Copyright 1990, Ameri- can Chemical Society.) whereas the intensity of the { DME},CH,OH; ions increase with increasing cluster size, particularly at the lower electron energies where fragmentation is less pronounced. It is possible these trends reflect the consumption of trimethyloxonium ion by reaction 5 within the cluster^'^ (DME),(CH,),O+ -,{DME},CH,OH: + C,H, (5) M = CH~OCHQ L CH3 _I r’ cnt ta, mlz 105 115 125 c -0 Figure 7 Schematic representation of possible ion-molecule reactions within dimethyl ether clusters leading to the observed CMS. (Reprinted with permission from reference 9.Copyright 1991, American Chemical Society.) {DME},-,[CH,OCH:* + CH,OCH,] + {DME), -,CH,O(H)CH,CH,OCH~ + {DME), -,CH,OH: + C,H,O (6) A number of the other ions observed in the fragment ion regions are consistent with this hypothesis including the fairly strong {DME}, + ICH30f ion and the minor ions correspond- ,;. ing to (DME),CH Glyme formation from DME in superacid and would require that C,H, be produced instead of C2Hf. Both CH,OH and DME have higher proton affinities than C2H4 and the presence of the solvent molecules may be respon- sible for this apparent change of reaction products; e.g. because of hydrogen bonding, solvation of a CH,OH; in a DME cluster might be expected to be more favourable than that of C2Hf. The reaction mechanism outlined in the paragraph above and shown in Figure 7, implies that the (CH,),O+ ions are destabi- lized by the presence of solvent molecules, i.e.the trimethyloxo- nium ions are consumed in the production of protonated methanol ions. It is also possible to propose a reaction mechan- ism based on a competition between production of the trimethyl- oxonium and methanol ions. The methyl cation transfer reac- tion, reaction 3, has been shown to be a SN2reaction and its very slow rate in the gas phase the result of severe steric requirements (approach angles and orientation^).^ This suggests that the rapid quenching of reaction 3 in the clusters is related to the fact that proper orientations of the reactants are more difficult when solvent molecules are present -this is common to many SN2 reactions.In this case it is not clear what reaction gives rise to the protonated methanol ions within the DME clusters. We are currently investigating the possibility that the production of methanol within DME clusters arises as a result of a bond formation reaction between an excited CH,OCH: ion and DME giving rise to a protonated 1,2-dirnethoxyethane (glyme) molecule. media has also been reported.24 It is perhaps of interest to note that reaction 6 could be considered as generally analogous to the reaction of NH; ions in ammonia clusters discussed above. In both cases reactions may take place in clusters which do not occur under bimolecular conditions. The presence of ‘solvent’ molecules serves to stabi- lize the products of extremely exothermic reactions allowing formation of covalent bonds.4.3 The Olefin Clusters Figures 8 and 9 display the pressure dependence of ethene and 1,l -difluoroethene (1, I-DFE) cluster ion intensity distributions, respe~tively.~~~5,26 The striking feature of these distributions is the rapid increase of the intensity of the n = 4 and 5 ions observed with increasing stagnation pressures which is accom- panied by a concomitant decrease in the intensities of the n = 2 and 3 ions. This behaviour is in marked contrast to that associated with typical magic numbers in that it is dependent on expansion conditions. High pressure mass spectrometry has established that sequen- tial ion molecule reactions within ethene lead to the formation of covalently bonded molecular ions.27 The reactivity of the grow- ing ion was found to decrease dramatically as a function of the ion size.Kebarle and co-workers have found that the C,H:, and C, ions were the largest ions produced with high intensity. 168 (CH2 = CHZ}~’ h u) .-c C 30000@ 2.20000 x c.-g 10000 Q,c =o t 0 10 20 0 10 20-0 n n 30000 Po = 2.5 atm 20000 10000 c 0 10 20 0 10 20 -0 n n h I.--v, C 50000 40000 f 40000 30000230000 .% 20000 20000 z 10000 10000 c -= oE0 10 20 0 0 10 20-n n Figure 8 Plots of the stoicheiometric ethene cluster ion intensities (C,H,},+, at several different expansion pressures.The expansion temperature was maintained at 253 K. The electron impact energy was 13 eV. [Reprinted with permission from reference 25. Copyright 1990, Elsevier Science Publishers B. V. (North-Holland).] This effect was attributed to steric hindrance due to the structure of the larger ions. Cationic polymerization was also suggested as one of the processes taking place in solid ethene following radiolysis.2 Once again the ionic reactions were found to be quite inefficient, mainly producing molecules only up to about C12H24. It was unclear in these experiments whether the extent of reaction was limited by the kinetics of the ionic reactions or by neutralization by recombination with geminate electrons. The sharp drop observed beyond n = 5 in Figures 8 and 9 indicates that the probability of observing ions larger than C,,H~, is low regardless of the starting size of the cluster ion initially formed.The behaviour of the cluster ion distributions may then be explained as follows for ethene clusters: The C,H: ion within the cluster reacts with one of the neighbouring monomers to give an internally excited C4Hi* ion. Under bimolecular conditions this intermediate rearranges and frag- ments via CH, or H In clusters of sufficient size, this intermediate may be stabilized by transfer of its internal energy into the cluster modes, probably resulting in the ‘boiling off’ of additional ethene monomers. Such processes may be expected to become more efficient in larger clusters which explains the observed rapid decrease in the intensities of the non-stoicheio- metric ‘fragment’ ions with cluster size.The stabilized C4H; ion may then react with another monomer molecule with the excess energy again being dissipated by ‘boiling off’ monomers, etc. These ionic addition reactions proceed until an ion is formed which is characterized by extremely low reactivity towards CHEMICAL SOCIETY REVIEWS. 1992 chemistry and are often associated with the formation of cyclic ions.3o Such ions often possess low reactivities because of the high activation barriers associated with the breaking of the C-C bond which is necessary for further reaction. The magic numbers observed in the ethane and 1,1 -DFE CMS most likely arise as a consequence of the formation of cyclic (probably five- and/or six-membered rings) molecular ions.The similarities between the high pressure mass spectrometric and cluster mass spectrometric results suggest that the qualita- tive trends in the kinetics of the ion-molecule reactions are very similar in both a high collision frequency gas phase environment and the interior of a cluster. These results also suggest that the solid state polymerization is indeed limited by kinetic effects associated with the growth of the polymer ion. Overall, then, investigations of ionic polymerization reactions within cluster ions may provide valuable insight into the early stages of ionic polymerization. 5 Future Directions The study of clusters will certainly continue to provide insight into the nature of solvation effects in chemistry and the local structure of the solvent medium.One may also expect to see an increasing number of investigators taking advantage of the simplified medium of the finite cluster to study chemical pro- cesses such as ionic polymerization. Observation of magic numbers will undoubtably continue to play an important part in understanding the structure and reactivities of cluster ion sys tems. Several lines of research are currently being pursued in this laboratory. Among these is an investigation of alcohol-water and other mixed protonated cluster ion systems. Alcohol-water clusters are rather interesting since it is known that clusters of the type {ROH),(H,O)H+ preferentially lose H20when n is small (n<ca.8-10), while large clusters preferentially lose ROH. No generally satisfactory explanation of this behaviour has been postulated. Ligand preference switches have been noted in several other molecule-water cluster ion systems and have been linked to formation of cluster ions in which (H,O),H + clusters form the ion core of the clusters even though the molecular components of the cluster possess much higher gas phase proton affinities. It is our contention that the energetically favoured 30000 20000. 10000’ n .,.,.... . , -0 1 3 5 7 9 1113 13 5 7 91112 n n Po = 2.2 atm 10000.lj0 -13 5 7 91113 13 5 7 91113 n n additional monomer molecules, i.e. until a ‘kinetic bottleneck’ is Figure 9 Plots of the stoicheiometric 1,l -difluoroethene cluster ion reached.The resulting molecular ion may be expected to contain intensities, {C,H,F, I:, at several different expansion pressures. The considerable internal energy which is dissipated by boiling off a expansion temperature was maintained at 247 K. The electron impact number of the remaining monomers giving in many cases the energy was 14 eV. bare molecular ions. (Reprinted with permission from reference 26. Copyright 199 1, Ameri-Kinetic bottlenecks are often encountered in polymer can Chemical Society.) MAGIC NUMBERS IN MOLECULAR CLUSTERS-M. T. COOLBAUGH AND J. F. GARVEY {CH~OH}~H~O+ Figure 10 Proposed structure for the {CH,OHi,(H,O)H cluster ion. + This particular species is the most prevalent of all cluster ions in the series {CH,OH},,{H,O}H + (which starts at n = 7).The dark circles are carbon atoms, the shaded circles oxygen atoms, and the open circles hydrogen atoms. Chemical bonds are indicated by ‘sticks’ while hydrogen bonds are indicated by thin lines. This structure is somewhat ‘flattened’ in order to highlight the three ‘5-membered’ hydrogen bonding rings. structure for these clusters is thus the one that (1) maximizes the number of hydrogen bonds and (2) minimizes the distances between the alcohol molecules and the ion core. Figure 10 shows our postulated structure for the (ROH),{ H,O}H magic+ n~mber.~The structure consists of a central H,O + ion comple- tely solvated by a ring (or chain) of hydrogen bonded alcohols generating three fused five-membered rings, each consisting of four CH,OH molecules and a H30+ ion hydrogen bonded together and is the one particular structure which maximizes the number of possible hydrogen bonds in the cluster ion.We are currently investigating the possibility that this is a general model which may explain the magic numbers observed in all distribu- tions of (ROH),{H,O},H heteroclusters.+ We are also continuing our investigation of ionic polymeriza- tion reactions within cluster ions. We have begun to expand our studies to include neat acetylene clusters and clusters composed of acetylene and molecules containing carbon-oxygen double bonds (e.g. acetone) and carbon-nitrogen triple bonds (e.g.acetonitrile). Our preliminary findings suggest that ring forma- tion reactions in acetylene clusters give rise to the very stable benzene ion. The intensity distributions of the mixed cluster ions show features which can be explained by the formation of six-membered ring heterocyclic compounds, e.g.the methylpyridine ion in acetylene-acetonitrile clusters.32 Acknowledgements. The support by the Office of Naval Research for this research is gratefully acknowledged. We also acknow- ledge the work of Dr. W. R. Peifer, Gopalakrishnan Vaidya- nathan, William J. Herron, and Stephanie G. Whitney on various aspects of this research. JFG recognizes the Alfred P. Sloan Foundation for a Research Fellowship (1991-1993). 6 References 1 T.D. Mark and A. W. Castleman, Jr., Adv. At. Mol. Phys., 1986,20, 65. 2 A. W. Castleman, Jr. and R. G. Keesee, Annu. Rev. Phys. Chem., 1986, 37, 525. 3 T. D. Mark, Int. J. Mass Spectrom. Ion Processes, 1987, 79, 1. 4 T. D. Mark, in ‘Electronic and Atomic Collisions’, ed. W. R. Gilbody, W. R. Newell, F. H. Read, A. C. H. Smith, Elsevier, Amsterdam, 1988, p. 705. 5 J. Jortner, Ber. Bunsenges. Phys. Chem., 1984,88, 188. 6 ‘Physics and Chemistry of Small Clusters’, NATO AS1 Series B, Vol. 158, ed. R. Jena, B. K. Rao, and S. N. Khanna, Plenum, New York, 1987. 7 A. W. Castleman, Jr. and T. D. Mark, in ‘Gaseous Ion Chemistry and Mass Spectrometry’, ed. J. H. Futrell, Wiley Interscience, New York, 1987. 8 ‘Elemental and Molecular Clusters’, ed.G. Benedek, T. P. Martin, and G. Pacchioni, Springer-Verlag, Berlin, 1988. 9 J. F. Garvey, W. R. Peifer, and M. T. Coolbaugh, Ace. Chem. Res., 1991, 24,48. 10 M. Kappes and S. Leutwyler in ‘Atomic and Molecular Beam Methods’ ed. G. Scoles, Oxford University Press, New York, 1988, p. 380. 11 W. R. Peifer, M. T. Coolbaugh, and J. F. Garvey, Chem. Phys. Lett., 1989, 156, 19. 12 R. Campargue, J. Phys. Chem., 1984,88,4466. 13 J. M. Price, M. W. Crofton, and Y. T. Lee, J. Chem. Phys., 1989,91, 2749. 14 J. M. Price, M. W. Crofton, and Y. T. Lee, J. Phys. Chrm., 199 1,95, 2182. 15 W. R. Peifer, M. T. Coolbaugh, and J. F. Garvey, J. Chem. Phys., 1989,91,6684. 16 K. Stephan, J. H. Futrell, K. I. Peterson, A. W. Castleman, Jr., H. E. Wagner, N.Djuric, and T. D. Mark, Int. J. Muss Spectrom. Ion Phys., 1982,44, 167. 17 A. G. Harrison and A. B. Young, Int. J. Ma.ss Spectrom. Ion Processes, 1989,94, 32 1. 18 A. J. Illies, Org. Mass Spectrom., 1990, 25, 73. 19 M. T. Coolbaugh, W. R. Peifer, and J. F. Garvey, J.Am. Chem. Soc., 1990, 112, 3692. 20 C. D. Chang and C. T.-W. Wu, J. Catal., 1982,74, 203. 21 S. Morgan and A. W. Castleman, Jr., J.Am. Chem. Soc., 1987, 109, 2867. 22 M. L. Sigsby, R. J. Day, and R. G. Cooks, Org. Mass Spectrom., 1979, 14, 273. 23 S. Okada, Y. Abe, S. Taniguchi, and S. Yamabe, J. Am. Chem. Soc., 1987, 109,295. 24 H. Choukroum, D. Brunel, and A. Germain, J. Chem. Soc., Chem. Cornmun., 1986, 6. 25 M. T. Coolbaugh, W. R. Peifer, and J. F. Garvey, Chem. Phys. Lett., 1990, 168,337. 26 M. T. Coolbaugh, G. Vaidyanathan, W. R. Peifer, and J. F. Garvey, J. Phys. Chem., 1991, 95, 8337. 27 P. Kebarle, R. M. Haynes, and S. Searles, in, ‘Ion-Molecule Reac- tions in the Gas Phase’, Advances in Chemistry Series No. 58, ed. P. Ausloos, American Chemical Society, Washington, D.C., 1968. 28 C. D. Wagner, J. Am. Chem. Soc., 1962,66, 1 158. 29 P. R. LeBreton, A. D. Williamson, J. L. Beauchamp, and W. T. Huntress, J. Chem. Phys., 1975, 62, 1623. 30 See for example, H. R. Allcock and F. W. Lampe, ‘Contemporary Polymer Chemistry’, Prentice-Hall: Englewood Cliffs, NJ, 198 1. 31 W. J. Herron, M. T. Coolbaugh, G. Vaidyanathan, W. R. Peifer, and J. F. Garvey, J. Am. Chem. Soc., 1992, 114, 3684. 32 S. G. Whitney, M. T. Coolbaugh, G. Vaidyanathan, and J. F. Garvey, J. Phys. Chem., 1991,95,9625.
ISSN:0306-0012
DOI:10.1039/CS9922100163
出版商:RSC
年代:1992
数据来源: RSC
|
7. |
Binuclear iron centres in proteins |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 171-178
Ralph G. Wilkins,
Preview
|
PDF (1146KB)
|
|
摘要:
Binuclear Iron Centres in Proteins This Review is dedicated to Dennis Darnall, who introduced the Author to haemerythrin. His recent tragic death saddens us. Ralph G. Wilkins Department of Chemistry, University of Warwick, Coventry CV4 7AL I Introduction Iron is the fourth most prevalent element and the most abundant transition element in the Earth's crust. It is not surprising then that it occurs widely in biological materials which furthermore display a large variety of functions.' Familiar to the reader is the iron-haem moiety which, in proteins, can transport and store dioxygen (haemoglobin and myoglobin). activate oxygen for insertion into a C-H bond (cytochrome P-450), and act as a conduit for electron transfer (cytochromes). It is probably less appreciated that there are a whole array of iron proteins, without a porphyrin ring, which can function in a similar way, albeit in simpler organisms. These are an emerging class of proteins which contain an oxygen bridged binuclear iron unit, designated Fe-O-Fe.The promi- nent members of this group and their general features are shown in Table 1. Intense interest in these proteins in the past twenty years is shown by a recent spate of reviews., There is, in addition, a large number of enzymes containing iron bound to sulfur donors in bridging arrangements (the iron-sulfur pro-teins),' but we shall not be concerned with these in this Review. The iron contents of all the proteins in Table 1 have been subjects of confusion and controversy, but the values are now established except perhaps for the case of methane monooxygen- ase.All contain at least one binuclear iron centre for which three states of oxidation might be anticipated, designated Fe2ll1, Fe211, and FelllFeI1. These entities are all known for each protein, although they are of variable stability. Most of the early investigations employed the fully oxidized species since these are usually the (stable) isolated forms. The proteins containing one and (especially) two FeI1 ions have been examined later because of their instability (sensitivity to air) and their lack of obvious spectral properties. Haemerythrin is the most studied and best characterized protein of the group in Table The various techniques which have been used to characterize the iron sites in these proteins were often first used with haemerythrin and so they will feature heavily in the discussions of that protein.Their applications to the other proteins will be detailed in the subse- quent sections. Generally an historical sequence will be used in Section 2. Original literature references, except to the latest work, can be found in the reviews cited. 2 Structural Aspects 2.1 Haemerythrin Haemerythrin is the oldest member of the group, having first been described in 1823. It usually occurs as an octamer in the coelomic fluid of the sipunculid (peanut worm) and as a monomer (myohaemerythrin) in the retractor muscle of the Ralph G. Wilkins was born in Southampton and obtained his degrees iivhile attending Southampton University College at the end of World War II.He has had a very enjoyable academic life in chemistrjl departments at the University of Shefield, State University of New, York at Buflalo, New Mexico State Universitjt, and the University of Warwick. He is Emeritus Professor at New Mexico State University and Honorary Professor at the Univer- sity of Waruyick. He has had over forty years experience in research and teaching in the area of mechanisms of inorganic and bioinorganic reactions and some I50publications. worm. Myohaemerythrin (13.9 kDa) is very similar to the octamer subunit, both structurally and in its properties. The relationship of the iron to the peptide chain has been known for some time and is shown diagrammatically in Figure 1.In deoxyhaemerythrin, both irons are in the + 2 oxidation state and this is the species which binds dioxygen reversibly to form oxyhaemerythrin. The latter slowly autoxidizes to methaemer- ythrin, in which both irons are in the + 3 oxidation state. This form, and a recently discovered mixed oxidation state contain- ing an FelIIFeI1 unit do not bind 0, but, typical of iron(rrr), they combine with a number of anions, the azide derivatives being particularly useful for e~amination.~ The first inklings of the presence of a binuclear iron site with an oxide ion bridge in methaemerythrin and derivatives, as well as, significantly, in oxyhaemerythrin, came from an examination of their electronic spectra.The peak positions and relative intensities (M-'cm-') for azidomethaemerythrin at 325 nm (E = 6.8 x lo3), 380 nm (sh) (E = 4.3 x lo3) and 446 nm (E = 3.7 x 1 03)closely resemble those of an iron(rrr)*dta complex which has an X-ray crystallographi- cally established 0x0 bridge [Fe,(edta),O], E = 1.06 x lo4 at 335 nm, 8.1 x 10, at 405 (sh) and 2.0 x 10, at 475 nm. Two antiferromagnetically coupled high-spin iron(Ilr j ions have low magnetic moments at 298 K (1.8-3.0 BM) and the magnetic susceptibility of the entity decreases with a decrease in temperature (anti-Curie behaviour). Such properties in simple metal complexes had been interpreted by Lewis and his co- workers in the ~OS.~Their emergence in methaemerythrin (as well as the absence of an EPR signal) afforded further evidence for a bridged diiron site in the protein.A useful measure of the interaction between the two spins is the exchange coupling constant, J, defined by the spin-exchange Hamiltonian, Hexch= -JS,S,. Its value can be estimated from the tempera- ture dependence of the magnetic susceptibility and is roughly a measure of the energy required to convert the system from a spin-paired into a spin-free state., A negative value of Jindicates antiferromagnetic coupling and the large value determined for azidomethaemerythrin and oxyhaemerythrin is typical for binuclear iron(r1r) complexes with p-0x0 bridges (Table 2). It was already clear, therefore, how valuable a comparison of the spectral and magnetic characteristics of this protein with iron complexes could be.These observations initiated the design and production of a very large number of model c~mplexes~.~ with interesting inorganic chemistry. These could be used also for comparisons with this and the other proteins of the group. Mossbauer spectroscopy was another tool that was early used to probe the nature of the iron site in haemer~thrin.~.~ The 57 Fe isotope is particularly suitable for such study. The important parameters are the isomer shift (a), which is the displacement of the doublet centre from zero, and the quadrupole splitting (LIE,), which is the energy difference between the two peaks (Figure 2).8 Isomer shifts of about 0.5 rnm s-and quadrupole splittings < 2 +mm s-are typical of Fe3 (Table 2).Lack of broadening of the spectra by an applied field at 4 K indicates that the irons are antiferromagnetically coupled. In addition two quadrupole doublets in oxyhaemerythrin (Figure 2 and Table 2) indicate that the two irons are inequivalent and that the species is best formulated as a met derivative containing the Fe,lllO: -unit. The larger values of the isomer shift and quadrupole splitting for deoxyhaemerythrin (Figure 2 and Table 2) confirm that we are dealing with Fe2+ centres. A single quadrupole doublet does, however, falsely suggest that there are equivalent irons, which we now know is not the case. CHEMICAL SOCIETY REVIEWS. 1992 Table 1 Proteins containing a binuclear iron unit and 0-containing bridge(s) Protein Occurrence Function Ref. Haemerythrin Several phyla of marine invertebrates.Those from sipunculids (peanut worms) Phascolopsis gouldii, Themiste zostericola, and Themiste dyscrita most examined. Stores and transports diooxygen 2,3 Ribonucleotide reductase Animals, bacteria, and virus-infected mammalian cells. Iron enzyme one of four types. That from Escherichia coli best characterized. Catalyses formation of deoxyribonucleotide di- or tri- phosphates (first step in DNA synthesis). 2,4 Purple acid phosphatase Glycoproteins from mammalian, plant, and microbiol sources. Those from bovine spleen and porcine uterine fluid (uteroferrin) most studied. Unknown physiological role. Catalyses hydrolysis of phosphate esters in pH -5-6.Methane monooxygenase Methanotropic bacteria. Those from Roman Baths (Methylococcus capsulatus (Bath), and Methylosinus trichosporium Catalyses the oxidation of CH, to CH,OH. Can also insert 0 into C-H bond of large variety of substrates. 296 most studied. Figure 1 Tertiary structure of haemerythrin showing the relationship of the Fe, unit with the four a-helices in myohaemerythrin. (Reproduced by permission from J. S. Richardson, Adv. Protein Chem., 198 1, 34, 167.) In principle, infra-red measurements might yield information on vibrational modes at the chromophore site. Unfortunately there is substantial background due to water and protein so that resonance Raman is preferred, this being less sensitive to water and in addition resonance enhancement of vibrational modes is observed.* Resonance Raman peaks near 500 cm-' and 800 cm-were assigned to v,(Fe-0-Fe) and v,,(Fe-0-Fe) in methaemerythrin and oxyhaemerythrin.These assignments have been subsequently used to detect the presence of the Fe-0-Fe bridging unit in other proteins. An 0-0 stretch at 844 cm-in oxyhaemerythrin could be identified with that in peroxide. By using 601*O it could be further shown that the two oxygen atoms were inequivalent, thus ruling out a symme- trically bound 0, moiety. Spectral examination of polarized single crystals of oxyhaemerythrin and azidomethaemerythrin showed a similar position and mode of bonding for the 0, and N; moities, respectively. This as a timely finding which was shortly to be substantiated by X-ray crystallography.The determination of the structure of oxidized haemerythrin by X-ray crystallography featured relatively early in the chemi- cal history of this protein. This was a fortunate circumstance since it aided considerably the interpretation of the spectro- Table 2 Magnetic parameters for iron-oxo complexes and proteins" ' Species -Jjcm - Gh/mms- A Eq/mms - Oxidized forms Methaemerythrin 134 0.46" 1.57l Oxyhaemerythrin 77 0.51 1.96 0.52 0.95 Rubrerythrin - 0.52 1.47 Ribonucleotide 108 0.55 1.62 reductase (Protein B2) 0.45 2.44 Uteroferrid 640 0.55 1.65 0.46 2.12 Methane monooxygenase 32 0.50 1.07 Fe20(OAc),(Me3TACN)$+'' 115 0.47 1SO Reduced forms Deoxyhaemerythrin 13 1.14 2.76 Rubrerythrin - 1.30 3.14 Ribonucleotide reductase 5 1.26 3.13 (reduced Protein B2) Uteroferrin (semi-reduced 10 0.531 1.781 form)d 1.22R 2.63R Methane monooxygenase - 1.30 3.14 Fe,O(OAc),( Me,TACN);' 15 0.48' 0.421 1.19R 3.38n Fe,(OH)(OAc),(Me,TACN)+ 13 1.16 2.83 Mainly at 4 K.Data from L. Que, Jr., and A. E. True, Prog. Inorg. Chem., 1990, 38, 97, Tables I and 11. All isomer shifts are quoted relative to iron metal at room temperature. Azide derivative shows two AEQvalues (1.47 and 1.95) and one 6 value (0.51 mms-l). d These values are modified on addition of phosphate. Me,TACN = 1,4,7-trimethyl-l,4,7-triazacyclononane. 1 High spin Fe3+. fl High spin Fe2'.scopic data with this protein and furthermore laid a firm basis for their applications to the other diiron proteins. The methaem- erythrin azide adduct was chosen for study because of its stability. Early disagreement between two groupsg working independently gave way in the early 80s to a consensus and the details of the iron site, roughly shown in Figure 1, now became clear (Figure 3).9 Both irons are octahedrally coordinated and bridged, not only by the expected 0x0 group, but, quite surpris- ingly, also by an aspartate and a glutamate residue. These features are retained in oxyhaemerythrin, with the dioxygen bound to one iron(Ir1) in an end-on-fashion, and with one oxygen close enough to the bridging oxygen atom to be hydrogen bonded to it, i.e.it is present as HO; (Scheme I). The lower value BINUCLEAR IRON CENTRES IN PROTEINS-R. G. WILKINS 0 +2 -2 0 +2-2 I I I I I I I II I 1 I I I I -2 0 +2 -2 0 +2 Velocity(mm s-’1 Velocity(mm s-’1 Figure 2 Mossbauer spectra of dinuclear iron proteins. Left, oxidized forms, from top, Phuscolosomalurcooxyhaemerythrinandazidohaem-erythrin; Escherichiu coli ribonucleotide reductase subunit B2; pig allantoic fluid purple acid phosphatase; Methylosinus trichosporium OB3b methane monooxygenase hydroxylase component. Right, cor- responding fully reduced forms except pink (semi-reduced) acid phosphatase. Left ordinate represents absorption. All determinations at -4 K except pink acid phosphatase (185 K). [Adapted by permission from P.E. Clark and J. Webb, Biochemistry, 198 1,20,4628 (haemerythrin); J. B. Lynch, C. J.-Garcia, E. Miinck, and L. Que, Jr., J. Bid. Chem., 1989,264,8091 (ribonucleotide reductase); P. G. Debrunner, M. P. Hendrich, J. De Jersey, D. T. Keough, J. T. Sage, and B. Zerner, Biochim. Biophys. Actu, 1983, 745, 103 (purple acid phosphatase); B. G. Fox, K. K. Surerus, E. Miinck, and J. D. Lipscomb, J. Bid. Chem.. 1988, 263, 10553 (methane monooxygenase)]. of v,(Fe-O-Fe) for oxyhaemerythrin, by about 20cm- than for the other met derivatives, had already been ascribed to hydrogen bonding of the 0, moiety. The structure of methaemerythrin (Figure 3) also produced a big surprise. It might have been anticipated that the azide in azidomethaemerythrin would be replaced by the water in anion- free methaemerythrin.In fact, the one iron(m) associated with N; or 0, in the met derivative becomes five-coordinated (Figure 3). The structure of deoxyhaemerythrin at 2.0 8, has been recently reported.’O The longer Fe-0 bond lengths and larger Fe-O-Fe angle (Table 3)support the idea, broached previously on the basis, for example, of a low -J value (Table 2), that the bridging 0x0 group in deoxyhaemerythrin is protonated. The five- and six-coordinated irons in deoxyhaemerythrin and met- haemerythrin confer bridge asymmetry (Figure 3). Bond dis- Hi Azidomethaemerythrin Methae meryt h rin H &His-25 %His-25 Oxy h aemeryt h ri n Deox y hae meryt h ri n Figure 3 The binuclear iron complex in azidomethaemerythrin, methae- merythrin, oxyhaemerythrin, and deoxyhaemerythrin.In all cases octameric protein from Themiste &scrita was used. H H (Serni-met)o (Semi-rnet)R Meta Scheme 1 Suggested mechanisms for interconversion of various forms of haemerythrin. (Based on R. E. Stenkamp, L. C. Sieker, L. H. Jensen, J. D. McCallum, and J. Sanders-Loehr, Proc. Nutl. Acad. Sci.USA, 1985,82,7 13 and J. M. McCormick, R. C. Reem, and E. I. Solomon, J. Am. Chem. Sue., 1991, 113, 9066.) “Species isolated in pH 6-7 region. ox = Fe(CN)i- and red = S,O:-. Within the bridge, Fe=Fe represents a glutamate and an aspartate linkage. Outside the bridge, Fe- represents a histidine residue. tances and angles for all forms of haemerythrin are contained in Table 3.1° The application of electron paramagnetic resonance (EPR) to these systems has been very rewarding. New types of EPR signals were first observed for both the mixed oxidation state’ and the fully reduced forms’ of haemerythrin.As well as their intrinsic theoretical interest,’ such signals have become import- Table 3 Crystallographic data for various forms of haemeryt hrin" Parameter AzidomeP Met Oxy Deoxy 3hi~Fe-0~-(A)2hisFe-O2-(A) Fe * -Fe(A) Fe-O-Fe(") 1.80 1.79 3.23 130 1.92 1.66 3.25 127 1.88 1.79 3.27 125 2.15 1.88 3.32 111 From M. A. Holmes, I. Le Trong, S. Turley, L. C. Sieker, and R. E. Stenkamp, J. Mol. Biol., 1991,218, 583. Similar values for myohaemerythrin. I gz = 1.96 I v gx = 1.66 I-...I I g = 1.95 I g = 1.72 I I H-,u I I g = 1.69 I (Semi-met)o Figure 4 EPR spectra of (semi-met), and (semi-met), haemerythrin from Phascolopsis gouldii. Temperature = 10 K. (Reproduced by permission from B. B. Muhoberac, D. C. Wharton, L. M. Babcock, P. C. Harrington, and R. G. Wilkins, Biochim. Biophys. Acta, 1980, 626, 337.) ant clues for the existence of this type of binuclear iron unit in other proteins. The species containing the mixed oxidation states (the so-called semi-methaemerythrin) escaped detection until the late 70s. One-electron reduction of methaemerythrin and one-elec- tron oxidation of deoxyhaemerythrin gave two distinct species, termed (semi-met)R and (semi-met)o respectively, with different spectral characteristics and kinetic behaviour.These showed rhombic and axial EPR respectively only at liquid He tempera- tures with g,, -1.7-1.8 (Figure 4 and Table 4). These arise from an S = $ state with antiferromagnetic coupled high-spin Fe"' and Fe" and J--8 (Table 2), consistent with an endoge- nous bridging OH group. These signals resemble those found in iron-sulfur proteins which have a higher gav,and g values greater than two (Table 4). Variable-temperature magnetic circular dichroism (MCD) and EPR spectroscopies have been interpreted in terms of structures for the two forms shown in Scheme 1.l4 Quite surprisingly, strong adducts of deoxyhaemerythrin with certain anions are formed. The interaction has been detected by competition experiments with Oi5and directly by MCD as well as by the observation of a new type of EPR signal at g = 16(with the azide adduct). Weak ferromagnetic coupling between the CHEMICAL SOCIETY REVIEWS, 1992 Table 4 Values of g for the EPR spectra near liquid helium temperatures of the mixed oxidation states of binuclear iron proteins Species g values Themiste zostericola haemerythrin (semi-met), 1.96, 1.88, 1.66 (semi-met)o 1.95, .72, 1.69 (semi-met)N; 1.90, .82, 1.50 Desulfovihrio vulgaris rubrerythrin (semi-met), I .98, .76, 1.57 Escherichia coli ribonucleotide 1.93, .85.1.64 reductase Beef spleen acid phosphatase ?pi&) low p~ form 1.94, 1.78, 1.65 high pH form 1.85, 1.73, 1.58 Methylococcus capsulatus (Bath) hydroxylase semi- reduced 1.92, 1.86, 1.71 Bacterial type ferredoxins Fe\"Fe"forrn 2.12, 2.04, 2.03 irons and transitions within doublets of an integer spin system give rise to this signal, which is not observed in deoxyhaemeryth- rin but has since been detected in the reduced forms of ribonuc- leotide reductase and methane monooxygenase( vide iqfra) and other iron complexes and proteins.2 Finally X-ray absorption fine structure (EXAFS) has been the technique most recently applied to these The determination of accurate bond distances is possible and since concentrated solutions (mM) may be used, the method has particular value when the X-ray crystallographic approach has problems (lack of suitable crystals, high molecular weight protein). The iron-containing protein is subjected to high energy monochromatic X-rays.At a certain energy of incident radiation (around 7140 ev), transition of 1selectrons to unfilled 4p and 5p levels occurs and results in a pronounced absorption edge which is -24 ev lower for Fell than Fe"'. At higher energies, photoelectrons are ejected and an absorption continuum results. Because of interference between the outgoing photoelectron wave from the iron and the returning wave from the backscat- tered ligand atom, or another iron, modulation of the conti- nuum occurs. The degree and form of modulation can be used to give information on the distances and the types of neighbouring atoms involved,2u.16 The use of simple model complexes for comparative purposes is essential and the selection of inappro- priate models early resulted in, for example, longer values for the Fe * * -Fe distances for methaemerythrin than measured by X-ray crystallography.These difficulties have now been mainly over- come. Examination of Tables 3 and 5 shows generally good agreement (except for deoxyhaemerythrin) in the distances and angles at the diiron site measured by the two techniques. EXAFS experiments on semi-methaemerythrin show lengthening of the Fe-p-,O and Fern-. Fe bonds, (Table 5) which, taken in conjunc- tion with data from model complexes, suggests conversion of the bridging 0x0 into an hydroxo group as in deoxyhaemerythrin. Detailed information on the sites in purple acid phosphatase and methane monooxygenase has also been obtained from EXAFS measurements (vide iqfka).2.2 Ribonucleotide Reductase Ribonucleotide reductase from E. coli (Table 1) consists of two dimeric non-identical proteins az and p2 known as protein B1 and protein B2 re~pectively.~ The large a subunits contain the nucleotide binding site as well as redox-active cysteines. Protein B2 contains a tyrosyl radical near the diiron centre, the first BINUCLEAR IRON CENTRES IN PROTEINS-R. G. WILKINS Table 5 EXAFS data for various forms of haemerythrin" Parameter Azidomet Met Oxy Deoxy Azidosemimet Fe-p-O(A) 1.80 1.82 1.82 1.98 1.87 Fe.*.Fe(A) 3.13-3.19 3.13 3.24 3.57 3.46 Fe-O-Fe(") 126 118 128 128 135 From L. Que. Jr. and A. E. True, Prog.Inorg. Chrm., 1990.38, 91 *Tyr-122 a Tyr-122 Tyr-122 Fe m...Fem Fe =...Fern .-Fen...Fen (active) B2 met 82 reduced 82 t O2(via radical intermediates)' I Scheme 2 Redox chemistry of ribonucleotide reductase protein B2. Mild reduction (hydroxyurea or enzymatically). Stronger reductants (S20: -//methyl viologen or enzymatically). Stopped-flow and rapid freeze-quench EPR indicate two intermediates, J. M. Bollinger, Jr., D. E. Edmondson, B. H. Huynh, J. Filley, J. R. Norton, and J. Stubbe, Science, 1991, 253,292. example of several proteins now known to contain a stable free radical.' This gives rise to an EPR (g = 2.0047) signal and a sharp absorption bond at 410 nm. These features disappear if B2 is reduced by hydroxyurea (or the tyrosine is replaced by phenylalanine by site-specific mutagenesis).The material, thus treated, termed met B2 (Scheme 2) now closely resembles methaemerythrin with iron-related spectral bands at 325 nm, 370 nm, 500 nm, and 600 nm.' 'The likelihood that B2 contained an iron site similar to that in oxy- and met-haemerythrin was strongly supported by variable-temperature magnetic suscepti- bility measurements (strong antiferromagnetically coupled high spin Fell1 signals) and resonance Raman spectra [v,(Fe-O-Fe) at 492 cm-1].4 It was long believed that there was only one binuclear cluster at the interface of the two identical polypeptide chains of protein B2. This would mean that the two irons would be expected to be indistinguishable. The observation of two pairs of Mossbauer quadrupole doublets of approximately equal intensity (Table 2 and Figure 2) was difficult to rationalize on this basis.This fact, together with improved iron analyses, led eventually to the courageous suggestion that the B2 dimer contained two diiron centres, bound to each of the polypeptide chains.ls The 6 and AEa values (Table 2) suggested that there was highly distorted high-spin Fe"', while EXAFS experiments supported a coordination environment for the irons similar to that in haemerythrin, but with fewer nitrogen (histidine) ligands. Short Fe-0 (1.8 A), and Fe..*Fe (-3.2 A) bonds probably meant additional bridge or bridges to that provided by an 0x0 group.2 All these features were satisfyingly confirmed in the recently reported three-dimensional structure of B2 at 2.2 8, resolution.'9 The iron site is shown in Figure 5.There are indeed two diiron centres about 25 8, apart, each centre resting in the middle of four a-helices, reminescent of that of haemerythrin (Figure I), although the Fe... Fe axis is parallel (not perpendicular) to the helices. The two irons are octahed- rally coordinated but the irons have marked differences in the ligands used. There is an 0x0 and a sin le carboxylate bridge. The radical provided by Tyr- 122 is -5 x from the closest iron and -10 8, from the surface. Protein B2 can be reduced to a form containing two Fell ions (Scheme 2). Susceptibility and Mossbauer parameters of reduced B2 (Table 2) point to a structure similar to that in deoxyhaemerythrin, i.~.weak cou- pling of two high-spin Fell ions.It has been a puzzle for some time why a semi-reduced form of the protein has not been reported. Characteristic EPR signals (Table 4) have now, how- ever, been observed when B2 is reduced anaerobically with diimide at pH 6.5 2ou or reduced by y-irradiation at 77 K.20h -0 Fe-1 O--(Glu-238 iN Glu-115 'N -/-His-241 H H H---OG NE-' --O\/OH I-N"O ' Ser-114 I ITrp-48 Asp-237 Gln-43 Figure 5 The coordination at the diiron site in Escherichiu coli ribo-nucleotide reductase B2 protein. (Reproduced by permission from P. Nordlund, B.-M. Sjoberg, and H. Eklund, Nuture, 1990,345,593.) 2.3 Purple Acid Phosphatase Acid phosphatases have maximum hydrolytic activity towards orthophosphate monoesters in the slightly acid region (equation 1).The metal-dependent subclass most studied contains iron and in particular, proteins isolated from porcine uterine fluid (uteroferrin, 35 kDa) and from bovine spleen (35 kDa). At first, as was often the case, the iron was not detected. Later, the iron content of uteroferrin became a controversial value. It was finally established when it was found that phosphate, which is tightly bound to the iron centre, interferes with the iron analy- S~S.~Both proteins contain two irons per molecule. There is 90% homology in the structures of the beef spleen and porcine proteins and their properties are very ~imilar.~J~ ROPOi-+ H,O +ROH + HPOi-(1) The oxidized (purple) form is reduced in a one-electron process to the pink form, which is the enzymatically active mixed-valence form.Both forms have high absorptions (C = 2 x lo3 M-lcm-' /Fe) at 550 nm (purple) and 510 nm (pink). This is ascribed to one tyrosine coordinated to one of the Fe"' ions. This must therefore be the iron in the purple form which remains unreduced in the production of the pink form. Resonance Raman spectra of the purple and pink forms using visible excitation showed four resonance-enhanced peaks between 1600 cm- and 1160 cm-' due to a coordinated tyrosine ring, but there is no evidence for an Fe-O-Fe sym-metric stretch at -500 cm-' (Figure 6). Magnetic susceptibility and Mossbauer spectroscopy showed antiferromagnetically coupled iron centres, strong in the oxidized form and weak in the reduced species (Table 2).One of the most telling pieces of evidence for a haemerythrin-like binuclear iron site in the pink form is the low temperature EPR spectrum. The rhombic signal results from a mixture of a low pH form and a high pH form (Table4) with a pK, -4.4. It is interesting that there are also two forms of semi-methaemerythrin interconverted by pH (Scheme 1, vide infra). Since suitable crystals of the protein have yet to be obtained for X-ray crystallographic examination, knowledge of the active site is still conjectural. Based on the experience with haemerythrin, and the EXAFS data (which suggest, because of a short Fe... Fe distance, that there is multiple bridging) as well as other spectral information, one that has been suggested is shown in Figure 7.21 The Fe, unit may also occur in some plant acid phosphata- ses.2h An interesting enzyme is present in red kidney beans.It contains one Zn2 ion and one Fe3 + and has a similar site to + that in the acid phosphatases. This is suggested by a visible absorption peak at 560 nm (C = 3.4 x 103M-'cm-') and an EPR signal at g = 4.3, characteristic of high spin Fell'. More significantly, replacement of (labile) Zn2 + by Fe2 + leads to an +Fe3 +Fez form with a characteristic EPR (g = 1.88, 1.76, I .62, and 1.49). Conversely, replacement of one iron(l1r) in purple 176 I 7 03 (uT Purple FePase II WJ? 0a 200 500 800 1100 1400 1700 Frequency (cm-') Figure 6 Resonance Raman spectra of purple and pink forms of beef spleen purple acid phosphatase, at pH 5.0 and 5"C, using 514.5nm excitation.(Reproduced by permission from B. A. Averill, J. C. Davis, S. Burman, T. Zirono,J. S.-Loehr,T. M. Loehr, J. T. Sage, and P. G. Debrunner,J. Am. Chem. SOC.,1987, 109,3760.) 0'\9 ;c--R h I R Figure 7 Proposed active site structure of purple acid phosphatase(phosphate complex). (Reproduced by permission from J. B. Vincent and B. A. Averill, The FASEB Journal, 1990,4, 3009.) uteroferrin by Zn2+ gives a protein which retains activity and has spectral characteristics (electronic, EPR) similar to those in kidney bean acid phosphata~e.~,~ 2.4 Methane Monooxygenase The three types of respiratory proteins, which combine reversi- bly with dioxy gen (haemoglobin, haemocy anin, and haemery t h- rin)' have corresponding oxygenases which use dioxygen (e.g.cytochrome P450, tyrosinase, and methane monooxygenase respectively). There are two kinds of methane monooxygenase, the mem- brane bound enzyme containing copper and a soluble enzyme containing non-haem iron (s-MI'A0).6 The soluble enzyme from Methylococcus capsulatus (Bath) and from other sources (Table 1) consists of three components: A, an hydroxylase (250 kDa) containing binuclear iron and substrate binding sites; B, a small (16 kDa) regulator protein with no metal chromophores; and C, CHEMICAL SOCIETY REVIEWS. 1992 a reductase (40 kDa) which accepts electrons from reduced nicotinamide adenine dinucleotide (NADH) and passes them to component A, one at a time.The whole entity catalyses reaction 2. NADH + CH, + O2 + Hf -+ NAD+ + CH,OH (2)+ H,O Protein A has been most studied but details of the iron site are the least understood of the four proteins considered in this section. Protein A has no absorption much above 350 nm and this poses a severe problem in its investigation. The magnetic parameters (Figure 2 and Table 2) hint at an antiferromagneti- cally coupled high spin Fell' component with bridging provided by an OH on an OR group, but probably not an 0 group, because of the low Jvalue. The best evidence for a binuclear iron site of the type considered in this Review is the appearance (low temperatures) of a rhombic EPR signal (Table 4) due to the production of the Fe1l1Fe" form on mild reduction of methane monooxygenase,22 and an integer spin EPR (g = 16) on strong reduction giving the FelIFe" form.23 A recent EXAFS study of protein A yields an Fe.*.Fe distance of 3.4 8124 Further clues as to the nature of the iron site in the hydroxy- lase component of MMO emerge from a comparison of the amino-acid sequence for MMO and ribonucleotide reductase protein B2.As might be anticipated, the six amino-acids involved in the diiron site (Figure 5) are amongst the few amino- acid residues conserved in nine different sources of the B2 protein. In two separate subregions, the ligands Glu-115, His- 1 18 and Glu-238, His-241 in B2 protein can be aligned with Glu- 144, His-147 and Glu-243, His-246 in MM0.6 It is not unreason- able then to suppose that these amino-acids in MMO are involved in metal binding in a similar type of diiron site as established in ribonucleotide reductase.3 Reactivity of the Binuclear Iron Sites This section must, of necessity, be brief but an attempt will be made to emphasize the important aspects. 3.1 Haemerythrin The reaction of physiological importance is the reversible bind- ing of dioxygen to deoxyhaemerythrin (equation 3, X = 0,), and this has been the focus of both structural and kinetic studies. The reaction is pictured as shown in Scheme 1. The high formation rate constants (kf,Table 6), particularly for deoxy- myohaemerythrin, prompted the investigators (Table 6, foot- note a) to suggest that there was addition of 0, to one of the Fell atoms and that Fe"-H,O replacement was not involved.This idea is supported by the now known structure of deoxyhaem- erythrin (Figure 3). Binding by NO, but not CO, also occurs (Table 6). Table 6 Kinetic and thermodynamic data for reaction of deoxyhaemerythrin with ligands X at 25 "C X x kf kd x K M-1s-1 S-1 M-' 0," NOh 7.4(78) 4.2 5 l(3 1 5)0.84 0.1q0.25) 5.0 HN,l 0.03 0.1 0.3 HCNO' 0.058 0.012 4.8 HFi 0.005 0.01 0.50 A. L. Petrou, F. A. Armstrong, A. G. Sykes. P. C. Harrington. and R. G. Wilkins, Biochim. Biophj.5. Acta, 198 I, 670, 377. Values in parenthesis for reaction of Themiste zos[ericoludeoxymyohaemerythrin.J. Springborg, (P. C. Wilkins, and R. G. Wilkins, Arm Chem. Scand., 1989. 43, 967. P. C. Wilkins and R. G. Wilkins, Biochim. Biophys. Acta, 1987, 912. 48. BINUCLEAR IRON CENTRES IN PROTEINS-R. G. WILKINS A,deoxyhaemerythrin + X +deoxyhaemerythrin -X (3)hd It was noted quite early that oxyhaemerythrin could be rapidly and reversibly bleached by anions such as azide, cyanate, and fluoride. This effect results from a surprisingly strong interaction of these anions with deoxyhaemerythrin. This has also been detected by MCD and EPR experiments.12 The kinetics of formation and dissociation of the colourless anion adducts have been measured by the stopped-flow technique using the coloured oxyhaemerythrin as a competitive probe.The interesting results that emerge are that (a) binding is quite strong, comparable to that of 0,,(b) anion is introduced as the neutral HX adduct, and (c) kf and kd values (Table 6) are very small for an FelI-ligand interaction. These points emphasize the importance of the hydrophobic nature of the site on the dyna- mics of reactions which occur there. The physiological significance of the semi-methaemerythrins, if any, is unknown. Some very interesting chemistry involving these forms has been reported. The two semi-met forms are in a pH-dependent equilibrium which is believed to involve a ligand- induced intramolecular electron transfer (Scheme 1). The inter- conversion is slow, and a first-order conformational change (k-lop3s-I) is believed to control this and a number of other reactions involving haemer~thrin.~ The structures proposed explain the early observations of rapid interconversions of (semi-met)o and deoxyhaemerythrin and of (semi-met)R and methaemerythrin.3 3.2 Ribonucleotide Reductase It is established that the tyrosine radical plays an essential role in the overall catalysis.The tyrosine radical might generate a protein radical on the B1 subunit and this in turn abstract a C- 3'H for the nucleoside diphosphate substrate and initiate the deoxygenation reaction. What is quite unclear is the function of the iron centre, although it is suggested that it is necessary for the generation and stabilization of the radical. The redox chemistry of ribonucleotide reductase protein B2 (Scheme 2) shows that the tyrosyl radical, once reduced, can only be regenerated via the FeIIFe" form reacting with 0,.The reduced B2 form thus mediates the oxidation of Tyr-122. An oxidoreductase which may produce reduced B2 in vivo has been identified.," 3.3 Purple Acid Phosphatase The physiological role of these proteins is uncertain. A number of oxyanions are inhibitors of the pink (reduced) enzyme. That most studied is phosphate (Pi) which is a competitive inhibitor and is indeed isolated with the enzyme. It is believed that a weak complex (Uf,,d'P,) is reversibly formed and that this is then converted into the oxidized strong adduct (Uf,;P,) with loss of activity (equation 4).Ufred+ PI*Uf,,, * PI-+ Uf,, * P, (4) Evidence for the initially formed adduct was obtained by Mossbauer spectroscopy.* The EPR signals of uf,,d * Pi are broad and were finally detected only with difficulty, in the presence of those due to free Ufred.26 The proposed interaction of phosphate with the suggested diiron site in uteroferrin is shown in Scheme 3. Similar behaviour of a substrate (ROP0,H -) would facilitate hydrolysis and promote phos- phate and phenolate f~rmation.~~.~~ 3.4 Methane Monooxygenase Many of the reactions catalysed by methane monooxygenase are similarly catalysed by cytochrome P450. Parallelisms in their mechanism of action have been sought therefore, despite the profound differences in the coordination sites of the two pro- teins.Thus it is suggested that 0, binds at, and substrate binds near, the iron centre with both proteins. A suggested substrate Fem-Fen -i""""e" o\p/o o\do // \0 OH J\OH A Scheme 3 Proposed interaction of HPO, with pink acid phosphatase. Production of A goes via a unidentate HPO, binding to Fell'. radical mechanism is shown in Scheme 4.6,28.29 A high-valent iron intermediate is postulated which abstracts a H atom from, for example, CH, to produce the methyl radical. Evidence for the 'CH, radical has been obtained by recent exciting spin- trapping experiments. 30 Radicals have also been observed from the substrates methanol ('CH,OH) and acetonitrile ('CH,CN).30 A similar type of mechanism to that shown in Scheme 4 has been suggested for the reaction of the diferrous forms of ribonucleotide reductase Protein B2 with 0, and the concomitant production of tyrosine radicals. Scheme 4 Proposed mechanism for s-MMO-catalysed conversion of CH, into CH,OH.An OH or OR bridge is represented by Fe-S-Fe. The Fe1l1.*.FeV = 0 entity may be in equilibrium with an FeIV ...FeiV= 0species. It is likely that CH, is attached near the Fe-.. Fe entity during the whole sequence. 4 Replacement and Removal of Iron The replacement of the metal in the native form of metallopro- tein by a different metal is an engineering process which has long fascinated bioinorganic chemists. Invariably this conversion requires the production of the apo form, that is the protein stripped of the metal. This has to be effected sufficiently gently that the addition of the native metal ion to the apo form regenerates the original protein with substantially unchanged properties.This process has been achieved with all the proteins listed in Table 1. The apo proteins are formed by dialysis of the metallo- protein against a chelating agent sometimes (haemerythrin and purple acid phosphatases) in the presence of a reducing agent, in which case it is the Fe," form which is being treated. With uteroferrin, treatment with dithionite alone (no chelating agent) leads to the rapid loss of one iron and a slower (hours) removal of the second iron. The production of the apo form of uterofer- rin containing no iron can be speeded up by the addition of denaturing agent.The presence of denaturing agent is also required for the successful preparation of apohaemerythrin,, while imidazole (probably) induces conformational changes and thus exposes the iron to attack (by 8-hydroxyquinoline) in the production of aporibonucleotide reductase protein B2. The chelating agents which are usually employed are catechols, 8- hydroxyquinoline, and 4,4'-bipyridine. The revival of the native from the apo form can be quite difficult.32 Usually the anaerobic addition of iron(I1) salts in the presence of reducing agents produces the iron(I1) form which can then be oxidized, usually by oxygen to the iron(m) form, which must have the native protein characteristics. If Fe-enriched iron salts are used in the regeneration process, it is possible to produce 57Fe-enriched protein which is useful for more sensitive Mossbauer measurement^.^ * Other metallo-derivatives may be prepared by treatment of the apo protein with the appropriate metal salt.In this way it may be determined how essential is the iron in the function of the protein. For example, the addition of Zn2+ to the ‘half-apo’ form of uteroferrin (in which only one of two irons is removed) produces an Fe3+Zn2+ form. This remains active and resembles the phosphatase from red kidney bean (Section 2.3). The Zn2 +Zn2 + form generated from apo- uteroferrin and zinc salts is however inactive, thus demonstrat- ing the necessity of one iron in the native protein. The distinctive properties of metal ions may also be exploited.Thus, the apo site in methane monooxygenase may be probed by titration with Mn2+ using the characteristic EPR spectra of Mn2 +,which is different from the Mn”-containing enzyme.33 It is clear that there are potential values in the use of the apo-and metallo-derivatives. 5 Future Developments The detection of the binuclear iron site in biological materials has come slowly since its establishment in haemerythrin. The question arises as to whether there will be yet more examples. A realisation of the disparate nature of the proteins shown in Table 1 suggests that a diiron site should always be considered in newly found iron-containing proteins or, indeed, existing ones. Thus a novel non-haem protein isolated from the periplasmic fraction of Desulfovibrio vulgaris was recently shown to contain two rubredoxin-like Fe(SR), centres.In addition, a binuclear iron centre was found. The latter must resemble that in haemerythrin since it has similar properties ascribable to that site (Tables 2 and 4)as well as those expected for a rubredoxin,2htc. The physiologi- cal role for the protein (termed rubrerythrin) is unknown, a situation we have noted for the purple acid phosphatases. Ferritin is an iron storage protein with a large number of irons, bridged by 0x0 and phosphate groups, inside a protein Ferritin, without the iron core, can be treated with a mixture of Fe2 + and Fe3 + ions in 0, to regenerate the protein. In the early stages there is evidence for binuclear iron cluster formation, judging from Mossbauer, EPR, and EXAFS stu- die^.^^*^^ Values of g for one of the initial transients, for example, are 1.95, 1.88, and 1.77 implicating a Fe1l1Fe1I oxo- bridged species.The structural characterization of these proteins proceeds well apace. The mechanism of their action and particularly the varying roles of the diiron sites in these proteins are much less understood and it will require much ingenuity for their solution. Acknou9ledgements. The author is grateful to Drs. H. Dalton and E. I. Solomon for preprints of publications and to Dr. R. E. Stenkamp for furnishing Figure 3. References J. J. R. Frausto da Silva and R. J. P. Williams, ‘The Biological Chemistry of the Elements’, Clarendon Press, Oxford, 199 1, Chapter 12and 13.(a)J. Sanders-Loehr, in ‘Iron Carriers and Iron Proteins’, ed. T. M. Loehr, VCH Press, New York, 1989, 375; (b)J. B. Vincent, G. L. Olivier-Lilley, and B. A. Averill, Chem. Rev., 1990, 90, 1447; (c) L. Que, Jr. and A. E. True, Prog. Inorg. Chem., 1990,38,97. (a)R. G. Wilkins and P. C. Harrington, Adv. Inorg. Bioclzem., 1983, CHEMICAL SOCIETY REVIEWS, 1992 551; Coord. Chem. Rev., 1987,79, 195; I. M. Klotzand D. M. Kurtz, Jr., Acc. Chem. Res., 1984, 17, 16. 4 B.-M. Sjoberg and A. Graslund, Adv. Inorg. Biochem., 1983, 5, 87. 5 B. C. Antanaitis and P. Aisen, Adv. Inorg. Biochem., 1983,5, I 1 1; K. Doi, B. C. Antanaitis, and P. Aisen, Struct. Bonding (Berlin). 1988, 70, 1.6 H. Dalton, D. D. S. Smith, and S. J. Pilkington, FEMS Microbiol. Rev., 1990,87,201; H. Dalton in ‘Methane and Methanol Utilizers’, ed. J. C. Murrell and H. Dalton, Plenum Press, New York. 1992, pp. 85. 7 D. M. Kurtz, Jr., Chem. Rev., 1990,90,585 discusses in detail the 0x0 and hydroxo-bridged diiron complexes. 8 A. Trautwein, Struct. Bonding (Berlin), 1974, 20, 101. 9 (a) R. E. Stenkamp, L. C. Sieker, and L. H. Jensen, J. Am. Chem. Soc., 1984, 106, 618; (6) S. Sheriff, W. A. Hendrickson, and J. L. Smith, J. Mol. Biol., 1987, 197, 273. 10 M. A. Holmes, I. L. Trong, S. Turley, L. C. Sieker, and R. E. Stenkamp, J. Mol. Biol., 1991, 218, 583. 11 B. B. Muhoberac, D. C. Wharton, L. M. Babcock, P. C. Harrington, and R. G.Wilkins, Biochim. Biophys. Acta, 1980, 626, 337. 12 R. C. Reem and E. I. Solomon, J. Am. Chem. Soc., 1984,106,8323; 1987, 109, 1216. 13 M. P. Hendrich, L. L. Pearce, L. Que, Jr., N. D. Chasteen, and E. P. Day, J. Am. Chem. SOC., 1991,113,3039;P. Bertrand, B. Guigliarelli, and C. More, Nouv. J. Chem., 1991, 15,445. 14 J. M. McCormick and E. I. Solomon, J. Am. Chem. Soc., 1990,112, 2005; J. M. McCormick, R. C. Reem. and E. I. Solomon, J. Am. Chem. Soc., 1991, 113,9066. 15 P. C. Wilkins and R. G. Wilkins, Biochim. Biophys. Acta, 1987,912, 48. 16 C. D. Garner. Adv. Inorg. Chem., 1991,36, 303. 17 C. L. Atkin, L. Thelander, P. Reichard, and G. Lang, J. Biol. Chem., 1973,248, 7464. 18 J. B. Lynch, C. J.-Garcia, E. Munck, and L. Que, Jr., J.Biol. Chem., 1989,264,8091. 19 P. Nordlund, B.-M. Sjoberg, and H. Eklund, Nature (London), 1990,345, 593. 20 (a)C. Gerez, M. Atta, M. Fontecave, J. Gaillard, C. Scheer, and J. M. Latour, J. Znorg. Biochem., 1991, 43, 536; A. Ehrenberg, R. Davydov, P.Allard, and S. Kuprin, J. Znorg. Biochem., 1991,43,535. 21 J. B. Vincent and B. A. Averill, The FASEB Journal, 1990,4, 3009. 22 M. P. Woodland and H. Dalton, J. Biol. Chem., 1984, 259, 53. 23 B. G. Fox, K. K. Surerus, E. Munck, and J. D. Lipscomb, J. Biol. Chem., 1988,263, 10553. 24 J. G. Dewitt, J. G. Bentsen, A. C. Rosenzweig, B. Hedman, J. Green, S.Pilkington, G. C. Papaefthymiou, H. Dalton, K. 0.Hodgson, and S. J. Lippard, J. Am. Chem. Soc., 1991, 113,9219. 25 J. W. Pyrz, J. T. Sage, P. G. Debrunner, and L. Que, Jr., J. Biol. Chem., 1986,261, 11015. 26 S. S. David and L. Que, Jr., J. Am. Chem. Soc., 1990, 112, 6455. 27 M. Dietrich, D. Munstermann, H. Suerbaum, and H. Witzel, Eur. J. Biochem., 1991,199, 105. 28 J. Green and H. Dalton, J. Biol. Chem., 1989,264, 17698. 29 B. G. Fox, J. G. Borneman, L. P. Wackett, and J. D. Lipscomb. Biochemistry, 1990, 29, 6419. 30 N. Deighton, 1. D. Podmore, M. C. R. Symons, P.C. Wilkins, and H. Dalton, J. Chem. Soc., Chem. Commun., 1991. 1086. 31 L. Que, Jr., Science, 199 1,253,273. See however, J. M. Bollinger, D. E. Edmondson, B. H. Huynh, J. Filley, J. R. Norton, and J. Stubbe, Science, 1991, 253, 292. 32 J. H. Zhang, D. M. Kurtz, Jr., Y.-M. Xia, and P. G. Debrunner, Biochemistry, 1991, 30, 583. 33 M. Fontecave, P. C. Wilkins, and H. Dalton. unpublished experiments. 34 E. C. Thiel, Adv. Enzymol. Relat. Areas Mol. Biol., 1989, 63, 421. 35 P. M. Hanna, Y. Chen, and N. D. Chasteen, J. Bid. Chem., 1991, 266, 886.
ISSN:0306-0012
DOI:10.1039/CS9922100171
出版商:RSC
年代:1992
数据来源: RSC
|
8. |
Ruthenium oxo complexes as organic oxidants |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 179-185
William P. Griffith,
Preview
|
PDF (995KB)
|
|
摘要:
Ruthenium 0x0 Complexes as Organic Oxidants William P. Griffith Department of Chemistry, Imperial College of Science, Technology, and Medicine, London SW7 2AY 1 Introduction Over the past decade the use of 0x0 complexes of ruthenium to effect catalytic regioselective homogenous oxidations has become increasingly important, and this review particularly emphasizes recent developments in the area of the oxidation of primary and secondary alcohols (RCH,OH and R,CHOH) to aldehydes, carboxylic acids, or ketones (RCHO, RCOOH, or R,CO). With the ruthenium catalysts so far developed, the oxidation of alcohols is the reaction effected with the greatest efficiency. Although such oxidation is one of the commonest of organic transformations, few reagents will effect such reactions catalytically and without attacking sensitive linkages in the R group; most of the complexes described herein are sufficiently selective to tolerate such linkages.Ruthenium and osmium are unique in at least two respects: they are the only elements in the Periodic Table to exhibit octavalency, and in their complexes they both cover the entire range of eleven oxidation states theoretically available to a transition metal, here VIII to -11 inclusive, corresponding to electron configurations of do to d'O inclusive. The 0x0 ligand 0;-is a strong u and T donor and so would be expected to favour high metal oxidation states. With ruthenium and osmium it stabilizes the VIII and VII states in the homoleptic species [MO,] and [MO,]-, and when accompanied by other ligands (usually though not always 'hard' donors) it coordinates to ruthenium and osmium(vI), (v), and (IV) and also (in a bridging rble) to ruthenium( 111).Because of their diversity in oxoruthenium complexes it seems logical to use oxidation states as a primary classification for the material of this review. The coverage here -which is very much inorganic in its emphasis -has been limited to reactions of those 0x0 complexes which have been reasonably well characterized, so that the very considerable body of often excellent work in which lower-valent ruthenium complexes have been used in conjunction with a variety of co-oxidants to effect specific oxidations is not covered. The latter subject has however been well reviewed elsewhere.2 Ruthenium(vllt) The only well-defined complex is the tetraoxide [RuO,]. Although this was known for over a century to be a very powerful oxidizing agent it was not until 1953that its properties as an organic oxidant were first systematically explored. It is a rather non-selective oxidant, reacting wifh alkenes, alkynes, alcohols, diols, aromatics, and ethers, often causing multiple bond or aromatic ring rupture. Nevertheless it is easy to generate catalytically (hypochlorite, bromate, or periodate are com- monly used co-oxidants), and it normally functions at room Dr. Bill Grifjith is Reader in Inorganic Chemistry at Imperial College. He obtained his Ph.D. at Imperial with Professor Sir GeofSre?? Wilkinson FRS and then went as a postdoctoral to Chicago and Stanjbrd Universities to work with Professor Henry Tuube.His research interests are principally in the coordination chemistry qj'the platinum group metals. He has written a number qf'hookson the platinum metals, mostly in the Gmelin series; other interests include peroxide chemistry and the applications of Raman spectroscopy to inorganic chemistry. temperature. Its main use is probably for the conversion of hydroxyl groups on carbohydrates into carbonyl functions and for the oxidation of steroidal alcohols. Its applications to organic chemistry have recently been reviewed. The mechanism of the stoicheiometric oxidation of propan-2- 01 by [RuO,] in aqueous perchloric acid has been investigated.In 1 to 6.5M HCIO, a hydride transfer appears to be the rate- determining step, while at higher acid concentrations the rate- determining step probably involves formation of carbonium ions. The overall reaction is thought to involve five electrons with Ru"' as the final product, and the involvement of radicals was thought to be unlikely since cyclobutanol is not cleaved in the reaction.2 Attempts to modify the oxidative ability of [RuO,] by attach- ment of, for example, N-donor groups have not been successful; ruthenium(v1) complexes are formed.3 3 Ruthenium(vl1) The tetrahedral perruthenate ion, [RuO,]- has long been known; in 1984 we studied its properties as a stoicheiometric oxidant towards alcohols in aqueous solution and also as the 18-crown-6 salt in benzene.In aqueous M NaOH solution primary alcohols are converted into carboxylic acids and secondary alcohols into ketones, but double bonds were cleaved (thus cinnamyl alcohol gave benzoic acid)., Kinetics of the stoicheio- metric oxidation of propan-2-01, mandelic acid and cyclobuta- no1 by [RuO,]- in aqueous base have been studied. It was concluded that the most likely mechanism involved formation of a radical intermediate and RuV1 with substrate oxidation by the latter [presumably present as ruthenate(vr)]. The oxidation of cyclobutanol by [RuO,] -under such conditions is attended by much C-C bond cleavage, the yield of cyclobutanone being only 33%, suggesting that one-electron processes are involved.2 The isolation of an organic-soluble salt of perruthenate, (Bu;N)[RuO,] ('TBAP') showed that, in non-aqueous rather than in aqueous media, [RuO,]- is a remarkably gentle oxidant, stoicheiometrically oxidizing primary alcohols to aldehydes and secondary alcohols to ketones without competing double-bond atta~k.~The discovery6 that this reaction could be made cata- lytic by using N-methylmorpholine N-oxide (NMO) as a co- oxidant (a co-oxidant previously and effectively used with lower- valent ruthenium complexes by Sharpless et al.') and isolation of the more easily prepared tetra-n-propylammonium salt (Pr;N)[RuO,] ('TPAP') has made this one of the most useful and versatile selective catalytic oxidants for alcohols available.8 The salt is simply and inexpensively made from solutions of RuCl,.nH,O and NaIO, in aqueous solution; the [RuO,] thus produced is swept out by air into a solution of molar aqueous (Pr;N)OH and K2C03,whereupon the deep green reagent is precipitated in good yield.Salts with other organic cations, uiz. PPhz and (PPh3)2N+, may similarly be ~btained.~ An even simpler, recently developed procedure is to add (Pr2N)OH to an aqueous solution of [RuO,]-generated by adding sodium bromate to ruthenium dioxide in molar carbonate. O In a dichloromethane-acetonitrile solution the reagent, with N-methylmorpholine N-oxide (NMO) as co-oxidant, is an excel- lent oxidant of primary alcohols to aldehydes and of secondary alcohols to ketones; sensitive units such as allylic, epoxy, lactone, indole, silyl ether, acetal, and tetrahydropyranyl func- tions are unaffected by it.It has also been used for the oxidation of homoallylic alcohols to dienones and for selective oxidation I79 of primary-secondary diols to lactones; la since its inception in 1987 there have been some sixty publications concerning its applications as an lh and it is increasingly being used in the fine organic chemicals industry. Typically it is used in 5 mol% quantities with 1 equivalent of the alcohol and 1.5 equivalents of NMO in dichloromethane-acetonitrile in the presence of 500 mg mmol- of powdered 4A molecular sieves, the latter to remove water formed during the reaction. All oxidations are effected at room temperatures and are usually complete within 30 minutes.8 Catalytic turnovers (i.e.moles of product/moles of catalyst) are high, in excess of 500 in some cases. An example of a reaction which would be difficult to carry out with conventional reagents is the oxidation of the secondary 5-hydroxy group of the 16-membered ring macrolide (1 a) to the aldehyde (1 b) without affecting the allylic 13-hydroxy group or the ether linkages; this step was an essential one in the total synthesis of the insecticide Avermectin Bla.l lC CHEMICAL SOCIETY REVIEWS, 1992 green colour of [RuO,]- reappears when the reaction with the substrate is complete. lo 4 Ruthenium(v1) A number of ruthenium(v1) oxidants have been studied, mostly but not exclusively for the oxidation of alcohols.4.1 Complexes with 0-Donor Ligands 4.1.1 With Hydroxo and Periodato Ligands The simplest of these is the 'ruthenate' ion, long thought to be tetrahedral ([RuO,], -) but now known, from X-ray studies of the potassium and barium salts, to contain the trigonal bipyra- midal tr~ns-[RuO,(OH),]~ -anion (2). r -2-This functions stoicheiometrically in aqueous base at room temperature as a two-electron oxidant, converting primary alcohols into carboxylic acids and secondary alcohols into a...ketones, the reaction being catalytic if excess persulphate is A OH The mechanism of oxidation by TPAP is not clear, however, and this aspect needs investigation. It oxidizes cyclobutanol to cyclobutanone in good yield (unlike the situation with [RuO,]- in aqueous solution2) suggesting that one-electron processes do not play an important part, though the overall reaction from ruthenium(vI1) to (IV) involves three electrons.It is possible that a sequence of two-electron steps similar to that proposed for chromium(v1) oxidation of alcohols is involved, e.g. RuV" -RuV RuV+ RuV"-2RuV1 RuV' -Ru" Recent work in these laboratories shows that [RuO,]+ can be generated in aqueous solution from RuCl,.nH,O and excess sodium bromate in aqueous molar carbonate solution at pH 11; the resulting catalytic [RuO,]- -BrO, system oxidizes, with good yields and turnovers, primary alcohols to carboxylic acids, secondary alcohols to ketones, activated primary alkyl halides to carboxylic acids, secondary alkyl halides to ketones, and primary nitroalkanes to carboxylic acids.Double bonds are, however, cleaved by the reagent (eg. cinnamyl alcohol and cinnamyl halides are converted into benzoic acid). As with the tran~-[Ru(OH),O,]~-S,O; -and trans-[R~O,(I0,(OH)),1~ -~ 10, reagents discussed below it is self-indicating: the yellow- introduced as co-oxidant. Although this catalytic tvans-[Ru(OH),0,I2 --S20i-reagent converts cinnamyl alcohol into cinnamic acid in high yield other allylic alcohols such as crotyl alcohol give only a low yield of P-methylacrylic acid, and clearly competing double-bond cleavage occur^.^ This limits the utility of the reagent, as does the fact that it is usable only in aqueous base, in which many organic materials are insoluble and may also undergo condensation reactions. It does have attractive features, though: it is a self-indicating reagent, the initial orange-red colour becoming black (probably due to colloidal RuO,) when the alcohol is added, but resuming its orange colour when the reaction is over.Reactions take about an hour at room temperatures with catalytic turnovers in the region of 450;, sonication of the reaction mixture considerably decreases the reaction times in some cases. Recently we have found that this trans-[R~(OH),O,]~ --S,O; reagent will also ~ oxidize activated primary alkyl halides to carboxylic acids and secondary halides to ketones; it is more effective (in rcspect of yields and catalytic turnovers) than the [RuO,] ~ -BrO, system referred to above, perhaps because the higher pH (pH 14 for the ruthenate as against pH 11 for the perruthenate catalytic reagent) helps removal of halide more effectively.Moreover, the double bond in cinnamyl chloride (and bromide) is not cleaved by trans-[R~(OH),O,]~ -;cinnamic acid is formed in good yield from both. It is likely to be more effective at halide oxidations than the traditional Kornblum or Sommelet procedures. Prim- ary nitroalkanes are also oxidized to carboxylic acids (a form of the Nef reaction). O The insoluble barium salt trans-Ba[RuO, (OH),] can be used as a stoicheiometric reagent for the oxidation of activated (benzylic) alcohols to aldehydes or ketones., The kinetics of the stoicheiometric oxidation of propan-2-01, mandelic acid, and cyclobutanol by trans-[R~O,(OH),]~ in M-aqueous NaOH have been studied.It was suggested that addi- tion to the a-C-H bond occurs to give an organometallic RuV intermediate which then reacts via a single-electron transfer reaction to give RuIV as a resonance-stabilized a-hydroxycarbo- cation. Unlike oxidations with aqueous [RuO,] -in base under similar conditions it was found that cyclobutanol is oxidized to cyclobutanone in good yield (as we have noted for the catalytic tr~ns-[RuO,(OH),]~--S,O; -system) suggesting that a two- electron oxidation has occurred.2 Another reagent which functions as an oxidant for alcohols in aqueous base is trans-[R~O,(I0,(OH))~]~~ (3).The X-ray crys- tal structure of the salt trans-NaK,[R~O~{I0,(0H))~]:8H,O RUTHENIUM OX0 COMPLEXES AS ORGANIC OXIDANTS-W.shows the trans 0x0 ligands (Ru=O 1.732 A) and monoproto- nated periodato ligands (Ru-0 2.003 A). It is unusual in that both the ligands and the metal centre function as oxidants: stoicheiometrically it is an overall six-electron oxidant, two electrons going to the {IV"0,(OH)J4 -ligand (which is reduced to (IvO, -)j and two to the metal Ru vlO$ unit which is reduced + to RuO,. Its action is catalytic with excess periodate (104-) as co-oxidant, converting primary alcohols into carboxylic acids and secondary alcohols into ketones. Double bonds are cleaved (thus cinnamyl alcohol gives cinnamic acid), as are diol linkages.Although this trans-[R~0,{1O,(OH)),]~--10,-reagent suffers from the same disadvantages as the trans-[RuO,(OH),I2 --S,O; -reagent (i.e. it functions in aqueous base and attacks some double bonds) it is, like the latter, self- indicating in colour (red -black -red) and gives high catalytic turnovers (ca.600).14It is also a good catalytic reagent for the oxidation of activated alkyl halides; primary benzylic halides are oxidized to acids and secondary halides to ketones. * Presuma-bly, as with the [RuOJ -BrO, -and the tran~-[RuO,(OH),]~ --S,O; -reagents the basic medium assists removal of halide. (3) 4.I .2 With Carboxylato Ligands We have recently isolated complexes of the form trans-[RuO,(OCOR)CI,]-(R = Me, Prn, Bun, CF,H) and find these to be excellent two-electron oxidants in organic solvents for the conversion of primary alcohols into aldehydes and secondary alcohols into ketones, the action being catalytic in the presence of NMO as co-oxidant.The acetato complex has been the most intensively studied: as with TPAP there is no competing double- bond cleavage with substrates such as geraniol, citronellol, chrysanthemyl, and cinnamyl alcohols. Phosphines are oxidized to phosphine oxides, sulfides to sulfoxides and sulfones' 6,1 (but our observation that the reagent oxidizes alkyl halides is in error: it is the NMO co-oxidant which does this17). The X-ray crystal structure of the complex (4) shows the anion to contain the unusual cis RuO, grouping (Ru=O 1.67 A, ORuO angle 120.1 "), and the acetato group is symmetrically bonded in bidentate fashion (Ru-0 2.13 A) so that the overall structure is that of a distorted octahedron (4).However, if it is supposed that the midpoint of the two oxygen donor atoms of the acetate is the effective site of coordination of this ligand then the structure is that of an essentially undistorted trigonal bipyramid' and so can be regarded as being structurally related to the ruthenate(v1) anion. (4) Preliminary investigations of the kinetics of the oxidation of propan-2-01 by [RuO,(OCOMe)CI,]- in dichloromethane at room temperatures have been carried out in the presence of excess NMO, and the stoicheiometry of the overall reaction determined. Surprisingly it appears that two moles of N-methyl- morpholine (NM) are produced per mole of propan-2-one rather than the expected 1:l ratio, so presumably the overall P.GRJFFJTH reaction involves production of oxygen or, possibly, of hydro-gen peroxide: (CH,),CHOH + 2NMO -+ (CH,),CO + 2NM + H202 The reaction appears to be first order in propan-2-one and in N-methylmorpholine. The precise r6le of the catalyst is not yet established. * Another acetato complex, this time containing the more usual trans-RuO, unit, is RuO,(OCOMe)(py),, made from trans-Ba[RuO,(OH),], pyridine, and acetic acid; the complexes RuO,(OCOR)(py),, (R = Et, Pr, C6H,) were also made. The X-ray crystal structure (5) shows the trans configuration [Ru=O 1.726(1) A]. It is a rather unselective stoicheiometric oxidant: E-and Z-/3-methylstyrenes give the cis and trans epoxides, tetra- hydrofuran gives y-butyrolactone, and 2,6-di-t-butylphenol gives the t-butylated quinone and diphenylquinone, while PPh, is oxidized to Ph3P0.l9 0O' 4.2 Complexes with Halide Donor Ligands We isolated (Ph,P)[RuO,CI,] some years ago by reaction of trans-[R~O,(OH),]~- with HCl and PPh4CI and found it to be an efficient stoicheiometric two-electron oxidant for alcohols, again converting primary alcohols into aldehydes and second- ary alcohols into ketones without competing double bond attack. Thus geraniol, E-cinnamyl, and chrysanthemyl alcohols were converted into their aldehydes in high yield,4 and the reagent was subsequently used in the total synthesis of the natural product isodrimeninol, an insect antifeedant.,O More recently we have shown that its action may be rendered catalytic by the use of NMO as a co-oxidant,21 its efficacy in this respect being similar to that of the [RuO,(OCOMe)C1,] -NMO rea- gent.Such comparability of reaction would be explicable if both reagents contained a cis RuO, grouping; that this was the case for [RuO,Cl,]- was suggested by our earlier Raman and infrared measurements4 and has been confirmed by Kochi and Perrier who have shown by X-ray structural determinations that the anion in [(Ph,P),N][RuO,Cl,] is trigonal bipyramidal (6) with the 0x0 ligands in the equatorial plane (Ru=O 1.694 A, ORuO angle 127.1"). In (Ph,P)[RuO,CI,] the anion is disor- dered between trigonal bipyramidal and square-based pyrami- dal, the 0x0 ligands being trans in the latter case.,, It is interesting to note that Kochi and Perrier find that [RuO,CI,]- is, as a stoicheiometric reagent in dichloromethane at room temperatures, reactive towards double bonds and a variety of substrates (thus cyclohexene gives a mixture of cyclohexene- oxide, 2-cyclohexenone, 2-chlorocyclohexanone, and cyclohex- enechlorohydrin; PPh,.gives PPh,PO, and the hindered phenol 2,6-di-t-butylphenol gives 2,6-di-t-butyl-p-benzoquinoneand I82 CHEMICAL SOCIETY REVIEWS, 1992 R2C-t + I OH R&=O + RUO~+ HCI + CI-Scheme 1 3,5,3',5'-tetra-t-butyldiphenoq~inone.~~Despite the apparent similarity in reaction conditions we find no competing double- bond attack in the reactions of [RuO,Cl,]-with unsaturated alcohols either under stoi~heiometric~ or catalyticZ conditions (i.e. with NMO as co-oxidant); perhaps the rate of oxidation of the hydroxyl group is greatly in excess of that for double bonds.The [RuOzC1,(OPPh,)]- ion, made from RuO, and HCl with PPh,, is an oxidant towards alcohols comparable in efficacy with [RuOzCl,]-., However, the red trans-[RuO,C1,I2 -ion (gener- atedzz in solution from [RuO,Cl,]- and excess Cl-), in the presence of excess C1- with NMOas co-oxidant, is substantially less effective in its oxidative abilities towards alcohols, as is [RuO,Cl,(PY)l-: Clearly it is not external charge which is of primary importance. It is more likely that the first two members of this series have in effect a vacant coordination site (triphenylphosphine oxide is an easily displaced ligand) whereas the last two do not.The trans-[RUO~C~,]~-anion was used in the presence of excess of C1- (to reduce dissociation of C1-), and the pyridine molecule, unlike PPh,PO, is firmly attached to the metal in [R~0,Cl,(py)]-.~~ A possible outline mechanism for oxidation of secondary alcohols by [RuOzC1,]- is given in Scheme 1. 4.3 Complexes with N-Donor Ligands 4.3.1 With Pyridine (py) The complex [R~0~(0COMe)(py)~] has already been men-tioned. Recently we synthesized a wide variety of neutral, cationic, and anionic oxoruthenium(v1) complexes containing coordinated pyridine, substituted pyridines (R-py), or related ligands, in most cases from with RuO, or trans-[R~O,(OH),]~ -and the ligand with appropriate pH contr01.~~~~ The neutral complexes take the form [Ru206(R-py),] where (R-py) is pyridine, 4-t-butylpyridine (4-Butpy), +bipy, nicotinic acid, or pyridine-2-carboxylic acid.These were made from RuO, and the ligand but also, in the case of [Ru206(py),], from trans-[R~O,(OH),]~ and pyridine with HC1 or HPF,. The-pyridine complex was shown by an X-ray crystal structure determination to have structure (7) in which the Ru,Oz bridge is planar (Ru-0 1.91 A, RuORu angle 100") but the trans 0x0 ligands are significantly distorted from linearity (Ru=O 1.72 A, ORuO angle 160.5"). This bending occurs away from the RuzOz bridge and towards the pyridine molecules, and is likely to arise from electron pair -electron pair repulsions between these 0x0 ligands and the oxygen atoms in the tightly bound bridge moiety.,, The same basic type of structure is likely for the other dimeric [Ru,O,(R-py),] complexes.The neutral species trans-[R~0~(py)~Cl,]and trans-[RuO,(bipy)Cl,] are made from RuO,, HCl, and the ligand,,, and trans-[RuO3(OH),l2 -, R-py, and HCl yield trans-[RuO2(R-py),C1,1 (R-py = py, 4-Butpy, and 4-~hloropyridine).~ The cationic trans-[RuO,(py>,](BF,), and anionic species (RH)[RuO,Cl,(R)] (R = py, 4-Bu'py, 3-methylpyridine, 3,4-dimethylpyridine) were made by similar procedures., In all cases the trans arrangement of 0x0 ligands was established by the Raman and infrared spectra of the solids and, in some cases, of the solutions.00 PY'l I'0'lJiPY0 (7) All these complexes oxidize primary alcohols to aldehydes and secondary alcohols to ketones in good yield; their action is catalytic in the presence of NMO as co-oxidant. Double bonds or allylic double bonds are not attacked (e.g. in citronellol and geraniol); catalytic turnovers of up to 135 with NMOand up to 85 with (BuiN)[IO,] as co-oxidant have been achieved., As oxidants however they differ in several respects from the oxo- ruthenium(vi1) and other oxoruthenium(v1) reagents so far described. First, they function stoicheiometrically as overall four-electron oxidants [rather than the two-electron oxidations given by the oxoruthenium(vi) species described above].Attempts to isolate oxoruthenium(1v) species such as the known +trans-[RuO(py),Cl] from reactions of these pyridine com-plexes have so far failed, but it is likely that the reactions do proceed via oxoruthenium(iv) species.,, The fact that rutheniu- m(n) species are the final products of these reactions presumably arises because, as a mild n-acceptor, pyridine and its analogues stabilize the d6 configuration of ruthenium(ri), whereas there is no such stabilization by the co-ligands (hydroxide, carboxylate etc.) in the other oxidants so far discussed. The second unusual aspect is that there is definite, though unspectacular, aerobic catalysis of the oxidations of alcohols by [Ru,O,(py),], trans-[Ru0,(py),12 +,and trans-[RuOzC1,(4-Butpy)]-at room tem- peratures in the absence of NMO but in the presence of air or dioxygen, whereas there is no such aerobic catalysis whatsoever for the other oxoruthenium(vr1) and -ruthenium(vI) complexes discussed so far.In acetonitrile solution catalytic turnovers of 12 were observed for [Ru206(py),] and for trans-[RuOzC1,(4- Bu'py)]- when dioxygen was bubbled through the solutions. The use of dioxygen under pressure did not increase the turn- overs nor did addition of copper(i1) acetate (commonly used to assist aerobic oxidations), but at 50°C the use of copper(I1) acetate with dioxygen at ambient pressure gave turnovers of up to 30 for primary and secondary It is not clear why coordinated pyridine helps such reactions, nor why the turn- overs are so low (formation of an inert p-0x0 dimer of ruthen- ium(1v) is a possibility).Finally, in connection with these pyridine complexes, it is of interest to note that pyridinium or substituted pyridinium salts of trans-[R~O,Cl,]~ -will abstract one pyridine ligand from the cation in solution: RUTHENIUM OX0 COMPLEXES AS ORGANIC OXIDANTS-W. where (R-py) is 4-Butpy, 3-methylpyridine, or 3,4-dimethyl- pyridine., 4.3.2 With 2,2'-Bipj)ridyl (bipy) and 1,IO-Phenanthroline (phen) Ligands We have already mentioned [Ru,O,(bipy),]; it has limited solubility and so is not suitable for oxidation^,^^ and surpris- ingly [R~O,(bipy)Cl,]~~ seems to function only as a stoicheio- metric oxidant4 though this really needs reinvestigation.Lack of solubility is also a problem with tr~ns-[RuO,(LL),]~+ (LL = phen, bipy), made from [Ru(OH)H,O)(LL),I2 and+ cerium(lv), but the more soluble trans-[R~O,(drnbipy),]~ + (dmbipy = 5,5'-dimethyl-2,2'-bipyridyl) will effect, albeit stoi- cheiometrically, a wide range of oxidations, e.g. of alkenes to epoxides, secondary alcohols to ketones, and cyclohexene to cyclohexenone. For the alcohol oxidations a hydride abstraction mechanism was proposed, [Ru(OH)(H,O)(LL),I2+ being formed., The complex ci~-[RuO,(dmp),]~ +,made by oxi- dation of [Ru(H2O),(dmp),l2+ with cerium(1v) (dmp = 2,9-dimethyl-I, 10-phenanthroline) will catalytically epoxidize nor- bornene, cyclohexene, and trans-/3-styrene under 3 atm.of dioxygen at 50°C, and Drago has sought to rationalize the oxidizing behaviour of cis and trans dioxoruthenium(v1) com- plexes towards alkenes by the use of INDO/l semi-empirical MO models.26 4.3.3 With Porphyrin and Macrocyclic N-Donor Ligands Most, though not all of the oxidation work in this area is stoicheiometric rather than catalytic. The 5,10, I5,20-tetramesit- ylporphinato (TMP) complex trans-[RuO,(TMP)] can be made by oxidation of [Ru(CO)(TMP)] with PhIO or peroxy acids, or by treatment of [RuL,(TMP)] (L = THF, CH,CN) with dioxy- gen. The complex will, at ambient temperatures and pressures, catalyse the aerobic epoxidation of a number of alkenes (e.g. cyclooctene, cis and trans /3-methylstyrene, norbornene) over a 24 hour period; it functions as an overall four-electron oxidant.2 Much work has been carried out on macrocyclic N-donor tetra-aza complexes of trans-dioxoruthenium(v1) species trans- [RuO2(R-TMC)I2+, e.g.1,4,8,11-tetramethyl-l,4,8,1l-tetra-P. GRIFFITH iety of a-hydroxy carboxylic acids H0CR'R"COOH (R' = R" = Me; R' = R" = Et; R' = Me, R" = Et; R' = Ph, R" = Me) gave [RuO(O,COCRr-R")]-. The X-ray crystal struc- ture of the 2-hydroxy-2-ethylbutyrato complex (Pr;N)[RuO (O,COEt,),] shows the anion of this to have a trigonal bipyra- midal structure with the 0x0 ligand (Ru=O 1.697 A) and the two deprotonated hydroxo ligands (Ru-0 1.860A) in the equatorial plane, the axial positions being occupied by the deprotonated carboxylato oxygen atoms (Ru-0 2.008 A) (9).29 The 2- hydroxy-2-ethylbutyrato complex is, in the presence of NMO, a catalyst for the oxidation of activated (benzylic) alcohols to aldehydes or ketones and of PPh, to PPh,O, but the reactions are slow and the turnovers for alcohols undistinguished (ca.25); the reagent is in this respect much inferior to most of the ruthenium(v1) complexes discussed above and to [RuO,]-.The oxoruthenium(v) species trans-[RuOX(R-TMC)I2+ (X = CI-, NCO-, N, -) have been generated electrochemically and these are catalysts for the electro-oxidation of benzyl alcohol to benzaldehyde, though activity is gradually lost over a few cycles. A salt trans-[RuVO,( 14-TMC)](C1O4) has also been isolated.2s The [RuVO(EDTA)]- complex is likely to be poten- tially useful as an oxidant.,O +The mono-oxo species [Ruv0(N4O)l2 {N40= 2-hydroxy-2-(2-pyridyl)ethyl[bis(2-(2-pyridyl)ethyl)]amine~, made by oxi- dation of [Ru(H,O)[N40)I2 + with cerium(Iv), is an effective though non-specific stoicheiometric oxidant, oxidizing alcohols to aldehydes or ketones, cleaving double bonds and oxidizing cyclohexane and adamantane.For the oxidation of alcohols mechanisms involving a two-electron hydride transfer or a one- electron hydrogen abstraction process were ~uggested.~ 6 Ruthenium(1v) All the reported oxoruthenium(1v) complexes are mono-oxo species, most are paramagnetic and all contain at least one N- donor ligand. Relatively few of these mild oxidants function catalytically however. The complex [Ru0(py)(bipy),l2 +,made by the oxidation with cerium(1v) of [RuO(py)(bipy),12 +,has been much studied; in dichloromethane solution it will catalytically epoxidize alkenes with hypochlorite as co-oxidant.Thus styrene is oxi- dized to styrene oxide, trans-stilbene to trans-stilbene oxide and cis-stilbene to a mixture of 95% of the cis and 5% of the trans isomers. With periodate or hypochlorite as a co-oxidant it will also catalytically oxidize benzhydrol to benzophenone, but there is considerable double-bond cleavage side reaction to give benzaldehyde. Similar reactions are observed with [RuO(bipy) (terpy)], +,and the latter has been used as an electrocatalytic oxidant for a variety of substrates including alcohols. Thus propan-2-01 gives acetone and ethanol gives a mixture of acetaldehyde and acetone.,, The study by Meyer et al. of the mechanism of the stoicheiometric oxidation of alcohols by the +paramagnetic ci~-[RuO(py>(bipy),]~ is probably the most detailed yet reported for such oxidations by oxoruthenium species.,, The studies were carried out both in aqueous solution and in acetonitrile with a range of primary and secondary alcohols, and a large C-H kinetic isotope effect observed.It was shown that the redox step almost certainly involves a two- azacyclotetradecane (14-TMC), 1,4,8,12-tetramethyl-l,4,8,11-tetra-azacyclotetradecane (1 5-TMC), and 1,4,8,13-tetramethyl- 1,4,8,13-tetra-azacyclotetradecane(1 6-TMC) (8). 14-TMC 15-TMC 16-TMC A general method of preparation for these is the oxidation of [Ru(H,O),(R-TMC)]~+ with H,O,; the X-ray crystal structures for the perchlorate salts of the 15- and 16-TMC complexes show the Ru=O distances to be 1.7 18 and 1.705 A respectively.These species exhibited slight aerobically assisted catalysis of benzyl alcohol to benzaldehyde (e.g. catalytic turnovers of 2.5 to 2.8 over 18 hours) under ambient conditions. Stoicheiometrically they oxidize cyclohexene at its allylic carbon atom to give 2- cyclohexen-1-one as the only product.2s 5 Ruthenium(v) Ruthenium(v) complexes are rare, and this is particularly true of oxoruthenium(v) species. Rea~tion~,,~ of [RuO,]- with a var- electron hydride transfer, the overall reaction for a secondary alcohol being [RurvO(py)(bipy),12 + R,CHOH -P+ ~~~"~~~2~~PY~~~~PY~zIZ+ + RZCO There is probably an initial pre-association step (py)(bipy)2Ru1v(=0)2 + R,CHOH*+ (py)(bipy),R~~"(=O)~+ ,HC(OH)R, followed by a hydride transfer redox step: (py)(bipy),R~~"(=O)]~+,HC(OH)R: ++ [(py)(bipy),R~-0---H---C(OH)R~]~+ [(py)(bipy),R~-0---H---C(OH)R,]~+ -P (py)(bipy),RuIl-OH ,R,COH+ + which is then followed by separation and rapid proton equilibration: 33 The intermediates in this mechanism were examined by an INDO/l MO calculation of the epoxidation catalysed by ruthen- ium(rv) 0x0 Meyer et al.have recently studied the oxidation by [RuO(py)(bipy)]$ + of hydroquinone to p-benzo- quinone and provide evidence that the reaction proceeds via a proton-coupled electron tran~fer.~ The species [R~O(bipy),(LR,)1~ + (LR, = tertiary phosphine or arsine) are made by oxidation of [R~(H~0)(bipy)~(LR,)]~ + with cerium(1v) and are effective stoicheiometric oxidants for alcohols to aldehydes and ketones, sulfides to sulfoxides, and phosphines to phosphine oxides.Studies were made in aqueous and non-aqueous media and, for the former, it was found that the rate of oxidation was dependent on the nature of LR, and on the hydrophobic nature of the target alcohol. For the oxidation of alcohols a hydride transfer step was suggested in which there is a synchronous transfer of a proton and two electrons from the alcohol followed by a fast proton transfer.36 +Oxidation of tran~-[Ru(NO)Cl(py),]~ with hypochlorite yields the paramagnetic trans-[RuOCl(py),12 +,and the X-ray crystal structure of the perchlorate shows the cation to contain a long Ru=O distance of 1.862 A.With alcohols ROH it gives [R~~~~(0R)Cl(py),]~Me, Et, Prn).37 We find that it also + (R = acts as a catalytic oxidant, with NMO or PhIO as co-oxidants, for the oxidation of benzylic alcohols to aldehydes. However yields are only moderate and turnovers low, up to 8.23 The macrocyclic complexes trans-[RuOX(R-TMC)I2+ (X = C1-, NCO-, N,) all have Ru=O distances of 1.765 A, substantially shorter28 than the 1.862 8, found in trans-[RuOCl(py),12 ., ,These macrocyclic complexes will oxidize + benzyl alcohol to benzaldehyde, there being slight aerobic catalysis (as there is with trans-[RuO,(R-TMC)I2 +) with cata- lytic turnovers of up to 6 over a 21 hour period at room temperatures. A two-electron process was suggested.,* 7 Ruthenium(l1i) The trinuclear species [Ru,O(OCOR),L,]~+ (R = Me, Et; L = H,O, PPh,; n = 0, 1) will catalyse the oxidation of primary alcohols to aldehydes and of secondary alcohols to ketones at 65°C under 3 atm.of dioxygen; under such conditions high catalytic turnovers were observed (near 1000 over 143 hours for [Ru,O(OCOEt),(PPh,),]). With the perfluorobutyrato (pfb) complex [Ru,o(pfb>,(Etzo),](pfb) aerobic catalysis of alkene oxidation occurs at 65 "C.,* Recently electrochemical evidence has been obtained for the existence of {[Rulll,lll(bipy), (OH)] 0}+ , { [RU~~'I'~( bipy)bipy) (OH)] 0}+ , { [RU'~,~( OI2O),+, and {[R~~~~(bipy),O]~O}~+.~~ CHEMICAL SOCIETY REVIEWS, 1992 8 Conclusions and Future Perspectives This review has sought to show that well-defined, easily prepared oxoruthenium complexes can be used to carry out a number of useful organic transformations.In a catalytic rijle some of them are particularly efficient for the mild and selectivc oxidation of alcohols, tolerating the presence of sensitive func- tional groups often attacked by other oxidizing reagents. There is much scope for advance in the field: synthetic procedures for the design of new oxoruthenium complexes are now well deve- loped and it should be possible to refine further the selectivity of oxidations, to extend the number of functional groups oxidized, and to effect chiral oxidations. Acknowledgements.I gratefully acknowledge my past and pres- ent students who are named in the appropriate references, and in particular thank Steve Ley for his enthusiastic collaboration, David Williams for carrying out many X-ray studies, and Ed Smith for helpful discussions. 9 References 1 R. Hartley in 'Chemistry of the Platinum Metals, Recent Develop- ments', ed. F. R. Hartley, Elsevier, Amsterdam, 1991 p. 180; J. L. Courtney in 'Organic Syntheses by Oxidation with Metal Com- pounds', ed. W. J. Mijs and C. R. H. I. de Jonge, Plenum, London 1986, p. 445; W. P. Griffith, Truns. Met. Chem., 1990,15,251; Plut. Met. Rev., 1989, 33, 181. 2 D. G. Lee and L. N. Congson, Can.J. Chem., 1990,68, 1774; D. G. Lee, L. N. Congson, A. Spitzer, and M. E. Olson, ibid., 1984, 62, 1835; D. G. Lee and M. van der Engh, ibid., 1972,50,2000. 3 A. M. El-Hendawy, W. P. Griffith, M. N. Moussa, and F. I. Taha, J. Chem. Soc., Dalton Trans., 1989, 901. 4 G. Green, W. P. Griffith, D. M. Hollinshead, S. V. Ley, and M. Schroder, J. Chem. Soc., Perkin Truns. I, 1984,681. 5 A. C. Dengel, W. P. Griffith, and R. A. Hudson, Trans. Met. Chem., 1985, 10, 98. 6 W. P. Griffith, S. V. Ley, G. P. Whitcombe, and A. D. White, J. Chem. Soc., Chem. Commun., 1987, 1625. 7 K. B. Sharpless, K. Akashi, and K. Oshima, Tetrahedron Lett., 1976, 29, 2503. 8 W. P. Griffith and S. V. Ley, Aldrichim. Actu, 1990,23, 13. 9 A. C. Dengel and W. P.Griffith, Inorg. Chem., 1991,30, 869; A. C. Dengel, J. F. Gibson, and W. P. Griffith, J. Chem. Soc., Dalton Trans., 1991, 2799. 10 W. P. Griffith, S. I. Mostafa, and P. A. Sherwood, tg be published. 1 1 (a)R. Bloch and C. Brillet, Synlett., 1991,829;M. J. M. Moreno, M. L. Sae Melo, and A. S. Campo Neves, Tetrahedron Lett., 1991, 32, 3201; (6)S. V. Ley, S. C. Smith and P. R. Woodward, Tetrahedron, 1992, 48, 1145; (c) S. V. Ley, A. Armstrong, D. Diez-Martin, M. J. Ford, P. Grice, J. G. Knight, H. C. Kolb, A. Madin, C. A. Marby, S. Mukherjee, A. N. Shaw, A. M. Z. Slawin, S. Vile, A. D. White, D. J. Williams, and M. Woods, J. Chem. Soc., Perkin Trans.. 1991, 667. 12 D. Fischer and R. Hoppe, Z. Anorg. Chem., 1991, 601, 41; M. 0. Elout, G. Haijie, and W.J. A. Maaskant, Inorg. Chem., 1988,27,610; G. Nowogrocki, F. Abrahams, J. Trihoux, and D. Thomas, 1976. 32, 2413. 13 M. Schroder and W. P. Griffith, J. Chem. Soc., Chrm. Commun., 1979, 58. 14 A. M. El-Hendawy, W. P. Griffith, B. Piggott, and D. J. Williams, J. Chem. SOC.,Dalton Trans., 1988, 1983. 15 A. C. Dengel. A. M. El-Hendawy, W. P. Griffith, S. I. Mostafa, and D. J. Williams, J. Chem. Soc.. Dalton Trans., to be submitted. 16 W. P. Griffith, J. J. Jolliffe, and S. V. Ley, J. Chem. SOC.,Chem. Commun., 1990, 1219; W. P. Griffith and J. J. Jolliffe, unpublished observations. 17 W. P. Griffith and J. J. Jolliffe in 'Dioxygen Activation and Homo- geneous Catalytic Oxidation', ed. L. I. Simandi, Elsevier, Amster- dam, 1991, p.395; W. P. Griffith, J. Jolliffe, S. V. Ley. and K. Sprinhorn, Synth. Lett., in press. 18 W. P. Griffith, J. J. Jolliffe, S. V. Ley, and A. Lucy. unpublished observations. 19 S. Perrier, T. C. Lau, and J. K. Kochi, Inorg. Chem., 1990,29,4190. 20 D. M. Hollinshead, S. C. Howell, S. V. Ley, M. Mahon, N. M. Ratcliffe, and P. A. Worthington, J. Chcm. SOC.,Perkin Trans. I, 1983, 1579. 21 A. C. Dengel, A. M. El-Hendawy, W. P. Griffith, and J. J. Jolliffe, Pol-vhedron, 1990,9, 175 1. RUTHENIUM OX0 COMPLEXES AS ORGANIC OXIDANTS-W. 22 S. Perrier and J. K. Kochi, Inorg. Chem., 1988, 27, 4165. 23 A. C. Dengel, A. M. El-Hendawy, W. P. Griffith, C. A. O’Mahoney, and D. J. Williams, J. Chem. Soc., Dulton Trans., 1990, 737. 24 W. P. Griffith and D.Pawson, J. Chem. Soc., Dalton Truns., 1973, 1315. 25 C. M. Che, W. H. Leung, C. K. Li, and C. K. Poon, J. Chem. Sue., Dalton Truns., 1991,379;C. M. Che, K. Y. Wong, W. H. Leung, and C. K. Poon, Inurg. Chem., 1986, 25, 345. 26 T. R. Cundari and R. S. Drago, Inorg. Chrm., 1990,29,2303and 487; C. L. Bailey and R. S. Drago, J. Chem. Soc., Chem. Commun., 1987, 179. 27 J. T. Groves and K. H. Ahn, Inorg. Chem., 1987, 26, 3833; J. T. Groves and R. Quinn, J. Am. Chem. Sue., 1985,107, 5790. 28 C. M. Che. T. F. Lai, and K. Y. Wong, Inorg. Chem., 1987,26,2289; K. Y. Wong, C. M. Che, and F. C. Anson, ibid., 1987,26, 737. 29 A. C. Dengel, W. P. Griffith, C. A. O’Mahoney, and D. J. Williams, J. Chem. Sue., Chem. Commun., 1989, 1720. 30 M.M. Taqui Khan, D. Chatterjee, R. R. Merchant, and K. N. Bhatt, J. Mol. Cuzal., 1991, 67, 309; M. M. Taqui Khan, D. Chatterjee, S. Kuma, R. R. Merchant, and K. N. Bhatt, J. Mol. Catal., 1991, 67, 317; M. M. Taqui Khan, D. Chatterjee, A. P. Rao, and S. H. Mehta, Ind. J. Chem., 1992, 31A, 146. P. GRIFFITH 31 C. M. Che, C. Ho, and T. C. Lau, J. Chem. Soc.,Dalton Trans., 199 1, 1259; C. M. Che, V. W. W. Yam, and T. C. W. Mak, J. Am. Chem. Soc., 1990, 112, 2284. 32 J. C. Dobson, W. K. Seok, and T. J. Meyer, Inorg. Chem., 1986,25, 1513; M. S. Thompson, F. D. G. Wagner, B. A. Moyer, and T. J. Meyer, J. Org. Chem., 1984,49,4972. 33 L. Roecker and T. J. Meyer, J. Am. Chem. Soc., 1987,109,746;J. C. Dobson, J. H. Helms, P. Doppelt, B. P. Sullivan, W. E. Hatfield, and T. J. Meyer, Inorg. Chem., 1989,28,2200. 34 T. R. Cundari and R. S.Drago, Znorg. Chem., 1990,29,487and 3904; Int. J. Quantum Chem., 1989,23,489. 35 R. A. Binstead, M. E. McGuire, A. Dovletoglou, W. K. Seok, L. E. Roecker, and T. J. Meyer, J. Am. Chem. Soc., 1992, 114, 173. 36 M. E. Marmion and K. J. Takeuchi, J. Chem. Soc., Dalton Trans., 1988,2385; J. Am. Chem. Soc., 1988,110, 1472; 1986,108. 510. 37 H. Nagao, K. Aoyagi, Y. Yukawa, F. S. Howell, M. Mukaida, and H. Kakihana, Bull. Chem. Soc. Jpn., 1987, 60, 3247; K. Aoyagi, Y. Yukawa, K. Shimizu, M. Mukaida, T. Takeuchi, and H. Kakihana, ihid. 1986,39, 1493. 38 D. Davis and R. S. Drago, Inorg. Chem., 1988,27,4759; C. Bilgren, S. Davis, and R. Drago, J. Am. Chem. Soc., 1987, 109, 3786. 39 J. K. Hurst, J. Zhou, and Y. Lei, Inorg. Chem., 1992,31, 1010.
ISSN:0306-0012
DOI:10.1039/CS9922100179
出版商:RSC
年代:1992
数据来源: RSC
|
9. |
Molecular fluorescent signalling with ‘fluor–spacer–receptor’ systems: approaches to sensing and switching devicesviasupramolecular photophysics |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 187-195
Richard A. Bissell,
Preview
|
PDF (1102KB)
|
|
摘要:
Molecular Fluorescent Signalling with ’Fluor-Spacer-Receptor‘ Systems:Approaches to Sensing and Switching Devices via SupramolecularPhotop hysics Richard A. Bissell, A. Prasanna de Silva, H. Q. Nimal Gunaratne, P. L. Mark Lynch, Glenn E. M. Maguire, and K. R. A. Samankumara Sandanayake School of Chemistry, Queen’s University, Belfast B T9 5AG, Northern Ireland I Why Bother with Molecular Fluorescent SignaII ing? Occasionally, along comes a discipline which can be useful to a wide spectrum of sciences and technologies. Supramolecular photophysics’ is one such central research area. It arises from the coincidence of the photon-molecule encounter and the molecule-molecule association. The latter is relevant since reversible binding between the signalling agent and the species being signalled is a prerequisite for molecular signalling.With regard to the former, molecular fluorescence is one of the most visually powerful phenomena in photophysics/chemistry. The researcher in this field has the privilege of literally seeing the outcome of molecular experiments, i.e. molecular fluorescence is a natural interface between human and molecular domains via the intermediacy of the photon. The phenomenon of molecular fluorescence possesses many features which make it particularly suitable for real-time and real-space monitoring ofiresponding to atomic and molecular specie^.^,^ The molecular basis means that: (Ia) it can report on nanometre spaces ifproperly targetted. This is real-space monitoring according to all but the most exacting criteria of molecular scientists.(Ib) The active mole- cules can be easily smuggled into dynamic/living systems which then act as unwitting hosts. Feature (Ha) suggests that the low concentration of active molecules need not poison the host. The photonic basis means that: (IIa) It has high sensitivity of detection approaching the single molecule, i.e. human address- ing of molecules is possible. Even routine experiments only need micromolar levels of active molecules. (IIb) Remote ‘wireless’ communication is available between the scientist and the active molecule. If necessary, active molecules can be immobilized on fibre optic tips for improved directionality, but at the expense of the freedom and flexibility of the free-swimming molecules.(IIc) It has natural imaging capability, especially with confocal microscopy. When coupled with its real-time capability [feature (IITa)], this results in cinematic representation of the micro- scopic world. Micrometre visualization is routine and near-field microscopy could reduce this limit further in certain situations. The excited state basis means that (IIIa) it has nanosecond response time, though the time constants of the necessary molecular associations can increase practical response times to milliseconds in the present supramolecular contexts. (IIIb) It can be easily ‘switched off’ at all wavelengths, unlike optical absorption. This is because the radiative deactivation of excited states is slow enough to suffer competition from chemical processes.Since excited state quenching mechanisms are reaso- nably well understood, it is also possible to arrange for ‘switch- ing on’. Hence, two-state digital action is feasible. Given all these features, it is easy to see the value of molecular fluorescent signalling systems for (a) sensing species and properties in the ~ ~ ~~~~~~ While originating jrom backgrounds which differ by as much as 5000 miles and 15 years, the authors share the common factors of Ph.D. stud?/ at the School of Chemistry in the Queen’s University of Belfhst and a strong interest in Photophysicsichemistry and Supramoleculur science. I87 physical and life sciences and, (b) data processing via molecular optoelectronics/ionics. Before concluding this section it is worth pointing out that fluorescent sensors and switches are examples of signalling systems which allow continuous monitoring/res- ponse which are distinct from reagents which irreversibly respond to a chemical stimulus and from labels which produce a chemically unresponsive signal for purposes of tracking a spe- cies.The development of fluorescent reagents and labels are also vibrant research areas in their own right.4 2 The Photoelectrochemical Approach to the Design of Fluorescent Signalling Systems It is natural that a fluorescent ion signalling system should include a fluor and an ion receptor as critical components. In the present design, these are distinct modules separated by an all-0- bonded spacer, i.e.the only communication between the optical and ligating modules is via relatively long-range forces. In their electronic ground states, these optical and ligating subunits show some features of a supramolecular assembly with only weak interactions between their 7~-and n-electron systems, even though they are covalently connected via the spacer. However, in properly designed systems, a strong long-range interaction develops in the form of an electron transfer from the ion-free receptor to the fluor when the fluor module is photoexcited. Photoinduced electron transfer (PET)5 is the essence of photo-electrochemistry which came to prominence with its application to solar energy conversion. However, the exploitation of light- driven redox reactions for signalling purposes is still at an early stage.The signalling possibilities arise from the fact that the ‘fluor-spacer-receptor’ system in the cation-free situation has been chosen such that its fluorescence is ‘switched off’ by the PET process3 (Figure la). The PET process, in turn, can be suppressed by the entry of a cation into the receptor by the cation-induced increase of the ionization/oxidation potential of the receptor. At the simplest level this is an electric field effect. However, other ways of inhibiting electron transfer are avail- able. Conformational changes, local polarity modulations, and hydrogen bonding are three approaches discussed in this article. Such a suppression of the PET process means that fluorescence again becomes the dominant decay channel of the excited fluor (Figure I b).Further clarification of the situation is possible in terms of the frontier orbital energy diagrams in Figures 2a and b. Thus, cation entry is signalled by photon output when interro- gated by excitation photons. For example, a proton signalling system, structure (1 a), may be designed by combining an organic base, i.e. an amine with the common fluor anthracene via a methylene spacer.6 Spectro- scopic details for related cases [Structures (1 b) and (Ic)] may be found in reference 3. The feasibility of PET within this system can be assessed by means of the Weller equation (equation The thermodynamic driving force for PET (A GET)in polar media is found to be between -0.1 and + 0.1 eV.8 Being at most only slightly endergonic, PET can therefore be expected to be rapid compared to fluorescence whose rate is 5 x I O7 s-as estimated CHEMICAL SOCIETY REVIEWS, 1992 I I I I Figure 1 Schematic representation of photoinduced processes in a ‘fluor-spacer-receptor’ signalling system (a) when cation-free and (b) when cation-bound.,LUMO+ Potential Energy Excited fluor Cation-free receptor (4 HOMO +’ -H-HOMO Excited fluor Cation-bound receptor (b) Figure 2 Frontier orbital energy representation of photoinduced pro- cesses in a ‘fluor-spacer-receptor’ signalling system (a) when cation- free and (b) when cation-bound. A 380 Wavelength Figure 3 pH-Dependence of fluorescence spectra of 10-6M (la) excited at 364 nm in methanol: water (1 :4,v/v).The pH values are, in order of decreasing intensity, 2.9, 7.2, 8.3, 8.9, 9.4, 9.8, and 10.8. sigmoidal behaviour can be biased in either direction by choos- ing the pH and the magnitude of its variation. A further contrast of PET signalling systems with respect to conventional fluorescent pH indicators emerges when the pK, values are closely examined. The present systems show experi- mentally indistinguishable values whether examined under excited state (fluorimetry) or ground state conditions (pH- dependent solubility).6 Conventional counterparts are dis-tinguished by large differences in excited- and ground-state pK, valuesg Thus, in well-designed cases, PET signalling systems possess the detection sensitivity of excited state experiments while maintaining thermodynamically valid binding constants characteristic of ground state measurements.FLUOR SPACER RECEPTORL-.?-~ I I FLUOR SPACER RECEPTOR :$:from the value for 9-methyl anthracene. Protonation of the I!! I amine moiety raises its oxidation potential to> + 2.5 eV. A fresh application of equation 1 yields A GET> + 1.4 eV. Hence PET should be strongly inhibited, resulting in a revival of fluorescence. Figure 3 confirms this expectation of proton- induced fluorescence recovery in (la). It is notable that the spectral shape and position of fluorescence are pH independent. Additionally all of the features (position, shape, and height) of the So+S, absorption spectra are also essentially pH invariant.Thus we have the simplest possible signalling action in that only one parameter, i.e. fluorescence quantum yield is proton controlled. Conventional fluorescent pH indicators based on the different design principle of photoinduced proton transfer exhi- bit more complex signalling behaviour which includes shifts of band positions in absorption and emission ~pectra.~ The fluores- cence intensity (IF)measured at any suitable wavelength varies with pH in a sigmoidal manner and satisfies the Henderson- Hasselbach-type mass action equation (equation 2), where K, is the acid dissociation constant. The fluorescence intensity is only dependent upon pH over a range of ca.2 pH units as is common for conventional absorptio- metric and fluorimetric pH indicators and is essentially pH invariant outside of this pH window. It is interesting that the value of such signalling systems f3r sensing purposes is due to the presence of this window, whereas the existence of the two pH-invariant plateaux (i.e.a bistable system) hold the appeal for switching applications.The analogue-digital duality of such Figure 4 Schematic representation of photoinduced processes in a ‘fluor-spacer-receptor’ signalling system with reversed logic (a) when cation-free and (b) when cation-bound. An attraction of a simple logic system is the ease with which it can be modified, even reversed. Such a reversal of the logic given in Figure 1 is shown in Figure 4.Now the PET process only occurs upon cation binding and does so from the fluor to the cation-bound receptor.Compound (2)1° was an early example of molecular engineering of a fluorescent ion sensor, i.e. the detailed plans on the designer’s drawing board are directly and quantitatively translated into experimental reality (Table 1). This was expected because the modular construction of fluor- escent PET signalling systems implies that the properties of the ‘parent’ fluor and receptor units will be largely preserved in the assembled system. In (2), simple thermodynamic arguments via equation 1 with redox potentials of the ‘parent’ units allow us to predict whether pH signalling is possible or not. Then, the entire set of absorption/emission spectral features (band position, shape, and height) and acid-base behaviour [applicability of a version of the mass action equation (equation 2),pK, values] are predictable within experimental error, except the maximum fluorescence quantum yield (&max) which deviates negatively by MOLECULAR FLUORESCENT SIGNALLING WITH 'FLUOR-SPACER-RECEPTOR' SYSTEMS-A. P.DE SILVA ET AL. 189 /NRR' Ph H02CI Ph (4)x (3) la; X=H;R=FY=Et b; X=CH2NRR' c; X=CI d; X = CHzNRR'; R' = Table 1 Comparison of predictions and observations in fluorescent PET signalling systems Property Found in (2) Predicted from Found in (1) Predicted from NEt, (3) or (4) or 9-methylanthracene Signalling action possible? Yes Yes Yes Yes AAbs.rnax(nm) 3 59" 358h 365".d 366".d km-') 2.2h 2.1" 1.04.d 0.91'.d~,,,~,(lO~drn~mol-AFlu.rnax(nrn) 479 479 420" 416" 0.13h 0.19h 0.69" 0.29".'@Fmax @Fmin 0.003h 0.003hf 0.004" -Henderson-Hasselbach equation obeyed? Yes Yes Yes Yes PKa 4.2U.g,4.4".h 4.5" 9.1" 10.7' Solvent: methanol-water (1:4 v/v).h Solvent: methanol-water (1: 1 v/v). ' Solvent: methanol. d 0-lvibrational band. 9-Anthracenylmethyl iminodiacetate in 11 c fpH7 water gives a value of 0.40. Value for methyl ester of (2),which is locked in the 'switched off' state. R pK,(S,) value. pK,(S,) value. Solvent: water. a factor of 1.5. Such nearly complete quantitative predictability The provision of the spacer module is instrumental in preserv- is due to the heart of the proton receptor being a low-charge- ing the identities of the fluor and receptor units within the density carboxylate which is held remote and rigidly away from signalling system, which in turn gave rise to many of the special the fluor module along a plane approximately bisecting the features discussed above.The presence of the spacer also alerts excited state dipole. Compound (l), while possessing a substan- us to the importance of the relative spatial disposition of the tial fraction of quantitatively predictable signalling parameters, fluor and receptor units. Several formats in which the fluor and shows a notable breakdown of predictability in its pK, value receptor units can be arranged are shown in Figure 5. When (Table 1). This is at least partly because of steric inhibition of ordered according to a scale of decreasing time-averaged proxi- solvation of the protonated receptor unit due to the non-mity/interaction, we see that the 'fluor-spacer-receptor' format adjacent but proximal 9-anthracenyl fluor being separated only occupies a central position in the spectrum of fluor-receptor by a methylene group.configurations that are useful for fluorescent signalling. On a formal level, Figure 5 unifies a substantial fraction of fluorescent Figure 5 The spectrum of 'fluor-receptor' configurations useful in signalling research and arises from a simple appreciation of the fluorescent signalling research. role of the spacer module. (I) Integral systems form the vast PROXIMAL, NON-ADJACENT PSEU DO-INTRAMOLECULAR FLUOR SPACER RECEPTOR n FLUORI gi 0RTHOGONAL rn I RECEPTOR 01 e.g./miRECEPTOR 01 e.g.(6)" INTERMOLECULAR proximitylinteraction INTEGRAL e.g. (5)' CHEMICAL SOCIETY REVIEWS, 1992 PhNMe2 and pNMe2?I majority of known fluorescent indicators for protons and other ions. Optical excitation leads to internal charge transfer (ICT)14 in these fluors and ion binding to a site on the fluor itself will naturally modulate these charge shifts and hence the fluores- cence band wavelengths. (11) Orthogonal systems known thus far are twisted biaryls where the 7~ molecular orbitals of the fluor and receptor are separated because of a steric clash of their (T frameworks, i.e. orthogonality of molecular orbitals is due to geometric orthogonality. The signalling behaviour of these systems can be interpreted as a PET process in a 'fluor-spacer- receptor' assembly with a virtual C, spacer.The alternative viewpoint is that these systems produce non-emissive twisted intramolecular charge transfer (TICT)' states upon excitation. However, the photoelectrochemical criteria for the occurrence of both these processes are virtually identical and cation entry blocks off either the PET or the TICT process. (111) Proximal, non-adjacent systems arise from the 'fluor-spacer-receptor' format with C,,C,, and C, skeletons as common spacers. The difference in orthogonal systems is that the spacer is C,, i.e. virtual. (IV) Pseudo-intramolecular systems result from spatial constraints imposed by non-covalent interactions, e.g.hydro-phobically driven co-complexation of fluor and receptor units by a cyclodextrin host.', Such a strategy is appealing because organic synthesis is traded off against supramolecular self- assembly. This enforced proximity of the receptor to the fluor results in fluorescence quenching. Upon ion binding, however, the receptor unit can be expected to lose its quenching ability. Further, in suitable cases, the decreased hydrophobicity of the ion-bound receptor would increase its escape probability from the cyclodextrin. (V) Intermolecular systems rely on diffusive processes to set the stage for the fluor-receptor interaction and therefore are rather inefficient for signalling purposes.On the and NMe3 / (9) in P-cyclodextrin (8) other hand, the modular nature of the fluor and the receptor are most pronounced in this class. Thus, it is natural that the wealth of photophysical information gathered on this class can allow the designer rationally to plan signalling systems in categories 11, 111, and IV. 3 The Scope of the PET Design Logic for Fluorescent Signalling The simplicity of a concept aids the development of examples illustrating its applicability. This has certainly been true regard- ing the PET design logic for fluorescent signalling with nearly two dozen examples being available even by late 1990 from laboratories on four continents. This body of literature is catalogued in reference 16.For example, Table 2 lists the families of fluorescent signalling systems for protons which have been constructed and examined by ourselves and colleagues. Table 2 also summarizes useful optical spectroscopic parameters and acid-base properties of these signalling systems. The gener- ality of the PET design logic for proton signalling via fluores-cence is thus clearly established, since significant proton- induced enhancement (or attenuation in cases with reversed PET logic, i.e. Figure 4)of fluorescence (FE) is found in most cases. The few cases with FE -I were investigated to test the range of applicability of a given family and these thresholds are predictable by simple theoretical considerations outlined in Section 2. It is notable that all but two of the major types of excited states with significant light emission found in organic/ inorganic chemistry are represented in Table 2.They are: (a) pure T~T*states characteristic of aromatic hydrocarbons; (b) internal charge transfer (ICT)' states found in heteroatom containing donor-acceptor or push-pull 7~ systems; (c) essen- Table 2 Fluorescent PET signalling systems for protons and their salient features System Fluor Anthracene (10y7 2,4,6-Triphenylpyridinium (2)'O 1,3-Diaryl pyrazoline (I 1117 1,3-Diaryl pyrazoline (12a)l8 (Anthracene 1 ( I 2b) (1,3-Diaryl pyrazoline ) (I 2~)' (2,7-Bis(dimethylamino)7 (xant heni um 1 (13)17 1,6,7,12-Perylene tetra carbox-diimide (14)" 6-Amino-3H-benzimidazo [2,1-a]benz[d,e]-isoquinolin-3-01~ (15)" 4-Chloro,- 1,s-Naphthalimide (16)17 7-Methoxy coumarin (17)19 meso-Tetraphenylporphyrin Sn'" (I ,)I9 Tris(2,2'-bipyridy1)Ru1' fl These are fluorescence attentuation factors.Excited state hab,(nm) &rn(nm) FE Spacer Receptor PK, mr* 370400 414425 30-1200 CH, Amine (aromatic & Aliphatic) 1.2-9.1 ICT ICT 295 354-398 405 470-5 10 30 1.0-50" (CH,),CHCH, Benzoate Diethyl amine 3.84.6 < 2.0 ICT 35-20 470-542 30-60" CHCH, Pyridine 3.24.7 TT* 370-546 416-575 20-230 CH, (Alkoxyphenyl imino) 3.0-5.7 ICT CHCH, (diacetate ICTh C,(virtual) -TT* 491-528 530-545 5.3-50 (CH,), Amine (aliphatic) 3.8-6 .O n = 2,3 ICT 41 1454 500-545 6-200 (CH,), Amine (aliphatic) 5.8-9.2 n = 2,3 -m* 345 409 25-90 (CH,), Amine (aliphatic) 6.0-8.6 ICT 32&333 414 3040 CH, Amine (aliphatic) 4.5-7.I TT* 423, 550-596 600-650 1.1-14 CH, Amine (aromatic & aliphatic) 2.1-8.0 (ligand) MLCT 444450 653 30-150 CH, Amine (aromatic) I .9-2.4 This is an orthogonal system (Figure 5) which can be rationalized vici a non-emissive TICT excited state. MOLECULAR FLUORESCENT SIGNALLING WITH 'FLUOR-SPACER-RECEPTOR' SYSTEMS-A. P. DE SILVA ET AL. N Et2 R=qp X NMe, 12a; X = Me, Y = R b; X=Me,Y=R' c; X=Me,Y=R" d; X=Y=H ?"' CJ [NRR' tially ligand-localized 7~7~*states in complexes with p-block metal ions; and (d) metal-to-ligand charge transfer (MLCT) states in complexes with d-block metal ions. The two classes unrepresented so far are: (a) twisted intramolecular charge transfer (TICT)' states seen in certain amino-substituted elec- tron-deficient aromatics and biaryls and (b) metal-localized states in complexes with d orf-block metal ions. Examples of these two categories are eagerly awaited.Table 2 gives an indication of the scope of proton signalling with fluorescent PET systems. Their modular construction allows wide variation of optical and ligating units within the design constraints. The range of fluor modules employed gives both the designer and the end-user much flexibility with respect to: (a) Hydrophobicity and polarity (varied from polycyclic aromatic hydrocarbon to ionic heterocyclic) for targetting to microenvironments of given polarity.(b) Absorption and emis- sion wavelengths (varied from 295 to 596 nm and from 405 to 653 nm respectively) for use in variously pigmented microsys- tems. Within certain individual families, simple substitutional tuning of these wavelengths can be achieved over a wide range, e.g. 70 nm each for (1 1) without compromising signalling action (FE> > 1). (c) Stokes shifts (varied from 2 to 200 nm) of which the latter would find use in turbid/scattering microenviron- ments. (d) Emission lifetimes (varied from ns to ps) of which the latter, e.g. (1 8), would be suitable for use in microenvironments with native fluorescence of any wavelength (cases with ms lifetimes will be addressed in Section 4). (e) Chemical stability and immunity towards a given intermolecular quencher (varied from electron rich to electron poor).Such immunity would also be enhanced in those cases with short excited singlet state lifetimes. The spacer module has been varied as well. This was usually a flexible oligomethylene unit, though rigid CHCH, spacers have been employed within small rings. Rigid and multiple spacers can increase the usually weak coupling between the terminii, i.e. the fluor and receptor modules. However, the CH, spacer was our favourite since it allowed fast PET rates (because of the small distance of separation of the terminal modules) and was most convenient for synthesis via benzylic functionalities. Vir- tual C, spacers have also been used successfully in the orthogo- nal system (12c).All but one of the common classes of organic proton receptors are represented in Table 2. They are (a) amine, (b) carboxylate, and (c) pyridine. The one not examined yet, phenolate, deserves attention. Availability of choice of the receptor module can be crucial for microenvironmental studies since these classes have different changes in charge type and hydrophobicity upon protonation. A wide range of pK, values, currently from 1.2 to 9.1, are available. Thus, pH values from 0 to 10 are addressable with the present signalling systems. The simplicity of a design logic can occasionally appear as naivety in the cold light of continued experimentation. Such occasions have arisen during our investigations of proton sig- nalling PET systems.These, in turn, have clarified the role of the spacer. For instance, (16, R = Et)' shows many features of the expected signalling behaviour, but the So+S, absorption spec- tra are distinctly pH dependent with a limiting red shift of 9 nm in acidic solution. This is understandable since the ammonium centre of rather high positive charge density preferentially interacts across the short methylene spacer with the closer negative terminus of the ICT excited state of the 7-methoxy coumarin located in the lactone ring. In this instance the methylene spacer is unable to isolate the terminal modules from their electric poledipole coupling. Nevertheless, the dampening effect of the spacer is clear since the observed absorption band CHEMICAL SOCIETY REVIEWS, I992 f-ono7 I (19) shift is smaller than those seen in proton-responsive integral signalling sys tems.A tougher test of the generality of fluorescent PET signalling was its extension to other cations besides protons. The behav- iour of (19),O and (20),’ demonstrated that such extension was feasible. Compound (20, n = 1) was remarkable in having (a) a K +-induced FE of 47, (b) quantitatively predictable binding constants (log /3) for K+ and Naf and, (c) extreme synthetic accessibility (one-step reaction of known components). How- ever (19) and (20) shared the problem of pH-responsive fluores- cence. This problem was solved simply by constructing (21)22 which possessed no basic nitrogen centres.The response of (21, n = 0) showed good selectivity towards Na+ (FE = 15) versus the other alkali cations while displaying excellent H rejection.+ Again, it is notable that (21) was exactly designed employing electrochemical data of model components and their experimen- tal log /3 values for alkali cations were quantitatively predictable from those of the ‘parent’ benzocrown ether receptors. The optical properties were also predictable except that the experi- mental QFmax values were 30-fold smaller than expected. This poor performance of @Fmax was subsequently corrected by optimizing the signalling action by structural and solvent varia- tion. Then the QFmax values only differed by factors of 1.25- 2.5 from the expected value.Compounds (21, n = 0 and 1) operate best in alcoholic media and their use would be more suited to membraneous, rather than wholly aqueous, micro- environments. Compound (22)23, which employs a cryptand of reduced basicity as the cation receptor, was constructed as a forerunner of alkali-cation monitoring systems for intracellular applications. Though insoluble in water, (22) binds alkali cations avidly in methanol and gives a maximum FE of 11 with Rb . Water-soluble versions of (22) would be useful since some + of the elegant sensors presently available24 (which are integral signalling systems) for intracellular Na + monitoring suffer from the problem of short wavelength excitation (ca.340 nm), though longer-wave versions are bound to appear. Compound (22) is well-excited at 383 nm and fluor modules responsive to even longer wavelengths have already seen service in other PET signalling systems (e.g.Table 2). Ca2 +-induced FE values of 16 for (1 2a), 92 for (1 2b), and 25 for (12~)’ were attained by attaching Tsien’s selective calcium receptor25 via a spacer (CH,, CHCH,, and Co respecti- vely) to anthracene, 173-diaryl pyrazoline, and rhodamine fluors respectively. These examples nicely complement Tsien’s fluor- escent Ca2 sensors2 which are integral signalling systems and + which are popular for intracellular investigations. It is particu- larly notable that Tsien has reported a sensor with a structure identical to (12c) (though synthesized by a different route) but with FE = 3.25 Compound (12c) is a clear example of an orthogonal signalling system. Currently, (12b) has the largest Ca2+-induced FE value known in the literature and relatives with even larger FE values are now available.’ These dramatic examples of fluorescence ‘off-on’ signalling are due to an amplification of the mechanism outlined in Figure 1.The special ability of a metal ion to organize an acyclic, flexible ligand around itself was exploited by Tsien to create a Ca2 +-induced conformational change in the receptor (12d)25 We noted that this Ca2 +-induced conformational change which decouples the iminodiacetate moieties from the alkoxyphenyl units caused a large increase in oxidation potential of the receptor (beyond that caused by simple proximity of dication) which drastically inhi- bits the PET process in e.g.(1 2b). Such powerful signalling of Ca2 in the physiological range (10-7-10-6M) with essentially + no response to protons around pH7 and Mg2+ around 10- 3M has much potential for intracellular monitoring. These fluor- escent signalling systems (12a), (12b), and (12c) cover a wide range of excitation/emission wavelengths (370-546 and 4 16- 575 nm respectively) for the convenience of the end-user. Again, almost all of absorption/emission spectral features and ion- binding behaviour (for CaZ +,Mg2+,and H +) are reasonably predictable -the exception being QFmax values which deviate negatively by a factor of 5-25. The development of fluorescent PET signalling systems for other non-transition metal ions and organic cations can be expected to follow.For instance, London’s Mg2 +-selective receptor (23),26 from which integral fluorescent signalling systems have been built,26 is a likely candidate for conversion into analogues of (12). The incorporation of d-block metal ions, with their redox activity, would block off the designed PET channel (Figure 1b) but could simultaneously open up new ones, i.e. no fluorescence signal would result. On the other hand, the binding of d-block metal ions into ‘fluor-spacer-receptor’ systems which are designed to be initially fluorescent would cause efficient quenching, e.g. (24)27 with Ag’. This is reminis- cent of, but distinct from, the reversed-PET signalling systems, e.g.(2). The current paucity of PET signalling systems for anions is due to the relative unavailability of selective anion receptors with suitable electroactivity and a window of optical trans- parency. This difficulty has been neatly circumvented in one instance by Czarnik who employed a partially protonated polyamine to provide both the electroactive amine group and the anion-binding polyammonium array.28 The fluorescence ‘switching on’ is due to hydrogen bonding of the amine lone electron-pair by HPOi- (rather than an electric field effect) as depicted in (25). A feature of the PET design logic is that it is possible to design fluorescent signalling systems for nett microenvironmental properties to complement the species-signalling systems dis- cussed above. The well-known polarity-dependence of PET rates was exploited in (Id) to yield simple but sensitive signalling of solvent polarityz9 which could be of use in monitoring phase transitions.The methylene spacer and secondary amine units in (Id) contribute to the suppression of exciplex emissions, thus simplifying the fluorescence behaviour. 4 Higher Generations of PET Signalling Systems A special feature of modular systems is that they can be progressively increased in sophistication in a controlled and stepwise manner by the designer in different directions for different purposes. Some of the directions under current con- MOLECULAR FLUORESCENT SIGNALLING WITH ‘FLUOR-SPACER-RECEPTOR’ SYSTEMS-A. P. DE SILVA ET AL.sideration are formalized in Figure 6 in terms of modification of the first generation ‘fluor-spacer-receptor’ system. (I). The addition of two terminal targetting/anchoring modules allows the exploitation of the molecular nature of the PET signalling systems for the investigation of microhetero- geneous media with high spatial resolution. Membrane-bounded protons lie at the heart of most energy transduction processes in biology and so we considered targetted pH-signall- ing systems, e.g. (26),17 in detergent micelles, which are the simplest model membranes. While the sensor is anchored at the micelle-water interface, the microlocation of the proton-recep- tor module can be substitutionally tuned by variation of the hydrophobicity of either targetting module.Such depth-depen- Figure 6 Formal extensions of the first generation ‘fluor-spacer- recep- tor’ signalling system. Key: F = fluor, S = spacer, R = receptor, A/ T = anchor/targeting entity, L = lumophore (phosphor or environ- ment-sensitive fluor), TS = transparent shield, F = family with addi- tive behaviour, i fj. =I \ ‘ dent pH measurement, reminiscent of a submarine periscope, allows the spatial mapping of pH near the membrane surface. (11). The regioselective provision of a transparent shield module around the lumophore (as a generalized fluor) module allows the exploitation of phosphors with long-lived (ms) emis- /v \ Br &NEt2 sion for interference-free time-resolved sensing in host systems with native fluorescence.Usually, phosphor excited-states are easily quenched by molecular oxygen and other triplet states and therefore are difficult to observe in fluid solution at room temperature. Steric protection of the phosphor module to prevent material contact with its environment while allowing access to communication photons is achieved by the transparent shield. Sensing remains viable because the receptor module is not encapsulated. It is notable that a regioselective self-assembly process is required here. This may remind the reader of a 'message in a bottle' which, however, differs from the classic case because the cork is responsive to its environment and controls the message being read through the bottle. Compound (27) in p-cyclodextrin is an experimental realization of this scenario in aqueous (111).An important outcome of the modular construction of fluorescent PET signalling systems is that, in suitable cases, all the members of a given family with a common fluor unit, e.g. (I b), will have essentially identical and pH-independent optical properties. Of course, the fluorescence intensity (IF)will be strongly pH-dependent (according to equation 2) over a range of cu. 2 pH units and the pKa value derived therefrom will be substituent dependent. Thus the IrpH profile of an equimolar mixture of n members can be computed by assuming additive behaviour and can be optimized towards linearity by choosing the relationship between pKai values. Such a quasilinear IrpH function with a wide dynamic range due to a molecular fluor- escent system is a glass pH electrode mimic" and possesses the advantages of high spatial and temporal resolution for microen- vironmental research applications.(IV). The presence of a second identical receptor module can result in signalling systems which can selectively respond to homobifunctional guests. Since fluor modules are rigid almost by definition, the separation between the two receptor modules is approximately constant. This leads to fluorescence signalling upon recognition of the length of the guest molecule. Compound (28)3 is a fluorescent sensor with quantitatively demonstrable atomic resolution of linear recognition. The guests putrescine and (less so) cadaverine, H,N+(CH,),,N +H3,n = 43, which are 'the molecules of death' are selectively signalled rather than the shorter and longer a,w-alkanediyl diammonium and alkyl monoammonium ions.On one hand, optimized versions of (28) should allow remote signalling of cell death which is of use in early warning of pollution damage and food spoilage. On the other hand, (28) is an intelligent molecular device which commu- nicates molecular geometric information directly to the most powerful of human senses, i.e. vision. So (28) is particularly CHEMICAL SOCIETY REVIEWS, 1992 ready for use in molecular optoelectronics/ionics. The one-step synthesis of (28) from known components is another positive aspect of its application to the two areas mentioned. It is also profitable to think of linear recognition asa rudimentary form of geometric recognition.From this viewpoint, (28) becomes the forerunner of fluorescent signalling systems which recognize various patterned arrays of functional groups in guests by means of a complementary set of receptor modules preorganized on a fluor framework. (V). The presence of a second receptor module which is different from the first creates at least two possibilities. (a) The simplest logical outgrowth from class 4.1V is the development of systems for the linear recognition and fluorescent signalling of heterobifunctional guests. Both receptor modules should be capable of PET activity with the fluor for efficient signalling of linear recognition.However, the designer's task is simplified, at the expense of operational efficiency, if the requirements of PET activity is relaxed for one receptor module. Compound (29)17 illustrates this approach with regard to the fluorescent signalling of the brain neurotransmitter y-amino butyric acid and relatives. (b) A separate direction involves the coincidence signalling of two separate and different guests. This has important ramifi- cations for molecular optoelectronics/ionics since molecular devices with self-selection of input signal channels become feasible. (VI). The consideration of homobifluorophoric signalling systems introduces the phenomenon of excitation energy mig- ration. If suitably efficient, this process can statistically enhance the probability (and hence the rate) ofelectron transfer from (to) the receptor to (from) the fluor.Thus, the guest-induced fluores- cence enhancement can be larger than that for the corresponding first-generation signalling system. Homobifluorophoric versions of (1 2) indeed show this effect to a significant degree. (VII). Heterobifluorophoric signalling systems bring exci- tation energy transfer processes into operation. While the inter- play of electron- and energy-transfer is interesting in its own right, it also creates the opportunity for PET activity in one fluor and not the other. This leads to self-calibrated signalling with one sensory channel and another for internal referencing. Finally, it must be stressed that each of these classes 4.1-VII are potentially generalizable as was demonstrated for the first generation signalling systems in Section 3.Therefore we can believe that the PET design of 'fluor-spacer-receptor' signalling systems, though simple and innocent in concept, is quite general in scope and applicability. Further growth of this research front, both at conceptual and applied levels, should be expected. AcknowMgements. This article is dedicated to Errol Fernando and Vincent Arkley who 'switched on' so many of us into chemistry. We are grateful to S.E.R.C., D.E.N.I., The Queen's University of Belfast, The University of Colombo, and The Nuffield Foundation for support of our research. We owe a special debt to Amila Norbert, Annesley Peiris, Dayasiri Rupas- inghe, Lalith Silva, Thilak Fernando, Shantha Samarasinghe, Dayal de Costa, Ranjith Jayasekera, Saliya de Silva, Nalin Goonesekera, Aruna Dissanayake, and Swam Patuwathavith- MOLECULAR FLUORESCENT SIGNALLING WITH ‘FLUOR-SPACER-RECEPTOR’ SYSTEMS-A.P. DE SILVA ET AL. ana who got us started on this road. Space limitations prevented us from detailed discussion of many interesting results from many laboratories. Though many of these are cited within the references to this article, we hope to give them the attention they deserve elsewhere. A special thank you is due to Irene Campbell for her cheerful assistance in preparing this and other manus- cripts. Finally, where would we be without Providence and Ser~ndipity~~? 5 References 1 V.Balzani and F. Scandola, ‘Supramolecular Photochemistry’, Ellis-Horwood, Chichester, 1991. 2 A. S. Waggoner in ‘Applications of Fluorescence in the Biomedical Sciences’, ed. D. L. Taylor, F. Lanni, R. Murphy, and A. S. Waggoner, Liss, New York, 1986, p. 3. 3 A. J. Bryan, A. P. de Silva, S. A. de Silva, R. A. D. D. Rupasinghe, and K. R. A. S. Sandanayake, Biosensors, 1989,4, 169. 4 R. P. Haugland in ‘Biosensors with Fiberoptics’, ed. D. L. Wise and L. B. Wingard, Humana Press, Clifton, New Jersey, 1991, p. 85. 5 R. S. Davidson, Adv. Phys. Org. Chem., 1983, 19, 1; ‘Photoinduced Electron Transfer’, ed. M. A. Fox and M. Chanon, Parts A-D, Elsevier, Amsterdam, 1988; Top. Curr. Chem., 1990,156, 158; 1991, 159. 6 A. P. de Silva and R.A. D. D. Rupasinghe, J. Chem. Soc., Chem. Commun., 1985, 1669. 7 A. Weller, Pure Appl. Chem., 1968, 16, 115. 8 For feasibility assessment of PET processes we find it convenient to use the expression E,, fluor -Ered = E, along with equation 1. Figure 2 also suggests this simpler criterion. E, = singlet energy of the fluor. For 9-methyl anthracene E,, = +0.96 V and for triethylamine E,, = + 1.19or 1.OO V (H. Siegerman in ‘Techniques of Electroorganic Synthesis. Part II’, ed. N. L. Weinberg, Wiley, New York, 1975, p. 667; C. K. Mann and K. K. Barnes, ‘Electroche- mical Reactions in Nonaqueous Systems’, Dekker, New York, 1970). Since E,,, Pd,r = 0.1 eV (Z. R. Grabowski and J. Dobkowski, Pure Appl Chem., 1983,55,245) we get AGFT= +0.13 or -0.06 eV.9 J. F. Ireland and P. A. H. Wyatt, Adv. Phys. Org. Chem., 1976, 12, 131. 10 A. P. de Silva, S. A. de Silva, A. S. Dissanayake, and K. R. A. S. Sandanayake, J. Chem. Soc., Chem. Commun., 1989, 1054. 11 H. Shizuka, T. Ogiwara, and E. Kimura, J. Phys. Chem., 1986,89, 4302. 12 B. K. Selinger, Ausf.J. Chem., 1977,30,2087. I3 K. Kano, I. Takenoshita, and T. Ogawa, J. Phys. Chem., 1982,86, 1833. 14 H. Shizuka, Acc. Chem. Res., 1985, 18, 141. 15 W. Rettig, Angebv. Chem., Int. Ed. Engl., 1986, 25, 971. 16 A. P. de Silva and K. R. A. S.Sandanayake, Tetrahedron Lett., 1991, 32, 42 1. 17 Unpublished results obtained at Queen’s University. 18 A. P. de Silva and H. Q. N. Gunaratne, J. Chem. Soc., Chem. Commun., 1990, 186. 19 R.Grigg and W. D. J. A. Norbert, unpublished results. 20 J.-P. Konopelski, F. Kotzyba-Hibert, J.-M. Lehn, J.-P. Desvergne, F. Fages, A. Castellan, and H. Bouas-Laurent, J. Chem. Sac., Chem. Commun., 1985,433; F. Fages, J.-P. Desvergne, H. Bouas-Laurent, P. Marsau, J.-M. Lehn, F. Kotzyba-Hibert, A.-M. Albrecht-Gary, and M. Al-Joubbeh, J. Am. Chem. Soc., 1989,111,8672. 21 A. P. de Silva and S. A. de Silva, J. Chem. SOC.,Chem. Commun., 1986, 1709. 22 A. P. de Silva and K. R. A. S. Sandanayake, J. Chem. Soc., Chem. Commun., 1989, 1183. 23 A. P. de Silva, H. Q. N. Gunaratne, and K. R. A. S. Sandanayake, Tetrahedron Lett., 1990,31, 5193. 24 G. A. Smith, T. R. Hesketh, and J. C. Metcalfe, Biochem. J., 1988, 250,277; A. Minta and R. Y. Tsien, J. Bid. Chem., 1989,264, 19449. 25 R. Y. Tsien, Biochemistry, 1980, 19, 2396; R. Y. Tsien, Ann. Rev. Neurosci., 1989, 12, 227; R. Y. Tsien and A. Minta, Eur. Pat. Appl. EP, 1990,314480. 26 R. E. London, Ann. Rev. Physiol., 1991, 53, 241. 27 M. Gubelmann, A. Harriman, J.-M. Lehn, and J. L. Sessler, J. Chem. SOC.,Chem. Commun., 1988, 77. 28 M. E. Huston, E. V. Akkaya, and A. W. Czarnik, J. Am. Chem. Soc., 1989, 111,8735. 29 R. A. Bissell, A. P. de Silva, W. T. L. M. Fernando, S. T. Patuwathavithana, and T. K. S. Samarasinghe, Tetrahedron Lett., 1991, 32, 425. 30 R. A. Bissell and A. P. de Silva, J. Chem. Soc., Chem. Commun., 1991, 1148. 31 A. P. de Silva and K. R. A. S. Sandanayake, Angewi. Chem., Int. Ed. Engl., 1990,29, 1173. See also F. Fages, J.-P. Desvergne, H. Bouas- Laurent, J.-M. Lehn, J.-P. Konopelski, P. Marsau, and Y. Barrans, J. Chem. Soc., Chem. Commun., 1990, 655 for related work with a different logical basis. 32 Ancient shipwrecked mariners referred to Sri Lanka as Serendib.
ISSN:0306-0012
DOI:10.1039/CS9922100187
出版商:RSC
年代:1992
数据来源: RSC
|
10. |
Electrochemical aspects of STM and related techniques |
|
Chemical Society Reviews,
Volume 21,
Issue 3,
1992,
Page 197-208
P. A. Christensen,
Preview
|
PDF (2264KB)
|
|
摘要:
Electrochemical Aspects of STM and Related Techniques P. A. Christensen The Chemistry Department, Bedson Building, The University, Newcastle upon Tyne NE1 7RU 1 Introduction The importance of acquiring a detailed understanding of the solid/liquid interface has long been recognized with respect to its role in catalysis and electrochemistry. This interest has resulted in an explosion in the number and variety of surface analytical techniques, both in situ and ex situ, over the past twenty years. More particularly, the past decade has seen the increasingly routine application of in situ infrared, Raman, EPR, and, X-ray spectroscopies, ellipsometry, etc.1~2These techniques have proved to be extremely valuable in helping to elucidate the complex processes that occur at the electrode/electrolyte inter- face even though they can only provide information averaged over the electrode surface.In contrast, it is widely accepted that a high proportion of the chemical or electrochemical phenomena that take place at solid surfaces do so at ‘active sites’: defects, adatoms, kinks, or particular arrangements of a few atoms. Whilst there are techniques, such as electron bombardment methods, capable of providing structural information on such sites, down to ca. 30-50 8,resolution, they require Ultra High Vacuum conditions to operate; and the application of emersion techniques carries with it the possibility of, for example, surface reconstruction during the transfer between the electrochemical cell and the UHV chamber.The advent of Scanning Tunnelling Microscopy (STM) filled a major gap in the surface scientist’s armoury and was rapidly seized upon by electrochemists eager to exploit its unique capability of providing real space images of the surface of a conducting sample with atomic resolution. The technique of STM was first reported by Binnig, Rohrer, and colleagues in the early eightie~.~ STM relies upon quantum- mechanical tunnelling of electrons between a sharp tip and a conducting sample, maintained by the presence of a potential difference between the two, to obtain an atomic resolution ‘picture’ of the sample surface. Topographical images of the surface are obtained by monitoring the tipsample distance (at constant tunnelling current) or the magnitude of the tunnelling current (at fixed tip-sample distance) as the tip is drawn across the sample.Under UHV conditions, STM has a lateral resolution (i.e. parallel to the surface) of ca. 5 A, and a vertical resolution of 0.1 A; the impact of a structural technique having such a resolution can best be conveyed in a picture. Thus, Figure 1 shows an STM image of the surface of Highly Ordered Pyrolytic Graphite, (HOPG), obtained by Cataldi and Brigg~.~ The individual atoms can clearly be seen. At first sight, STM might be thought to be limited to the study of surfaces under UHV conditions, but STM has had a major impact on electrochemistry, and continues to provide exciting insights into the atomic-scale processes responsible for the phenomena we usually measure.Dr. Paul Christensen was educated at South Shields Grammar School and Exeter College, Oxford. He carried out the research tobturds his Ph.D. at the Royal Institution under the supervision of Lord Porter. Dr. Christensen has been a Lecturer in Physical Chemistry ut Newcastle University since October 1989, and his research interest lies primarily in the use of in-situ techniques such as STM, FTIR, Ellipsometry etc. to study the electrodelelectro- lyte interjace. This paper is not intended to be an exhaustive review of electrochemical STM, or related techniques; there are many excellent such reviews in the literature, (see, for example, refer- ences 5 and 6).Instead, it is intended to provide an introduction to STM, Scanning Tunnelling Spectroscopy, and Atom Force Microscopy, with particular respect to their strengths and weaknesses when applied in situ in electrochemistry. 2 Scanning Tunnelling Microscopy 2.1 Mechanics As was mentioned above, the essence of STM lies in the quantum-mechanical tunnelling of electrons between a sharp metal tip and the surface of a conducting substrate, maintained by a bias potential between the two; tunnelling occurs when the wavefunctions of the tip and sample overlap.Thus, the tip is first brought near the sample via a coarse z positioner, such as a 10:1 reducing lever, or a controlled-approach piezo-electric motor, (InchwormTM). The latter provides a 10 A step size, and a movement of up to 0.5 mm s-l.As soon as tunnelling current is detected, the tip is stopped, and the fine control system switched in. Thus, the actual STM tip is suspended over the sample surface by a piezo-electric mount that consists of either three linear crystals arranged orthogonally, or a single crystal ‘tube’ scanner; in either case allowing movement in the x,y, and z directions, see Figures 2a and b. This system takes over the cautious advance of the tip until the desired tunnelling current is achieved, usually corresponding to a tipsample separation of ca. 5-10 A, and then is used to raster the tip across the surface and obtain the image. The data collection is generally performed in one of two ways: 2.1.I Constant Current The tip is moved along a fixed trajectory in the x direction, and the tunnelling current, it, monitored.A feedback loop is then used to move the tip up or down in order to maintain a constant current, (see Figure 3a). At the end of a single scan, the tip is displaced a single step in the y-direction, and the scan repeated; the image obtained is thus a linescan plot of (x,y,z). The imaging speed is severely limited by the response time of the feedback loop and z piezo; too fast and the tip risks crashing into an atomic mountain, a typical scan rate is < 5000 A s-l. 2.1.2 Constant Height In this approach, the tipsample distance is fixed during the scan, and the variation in tunnelling current is plotted as a function of lateral position, see Figure 3b.Thus, only the electronics have to respond, rather than the piezoelectric crystal. As a result, the imaging speed can be increased, by a factor of up to two orders of magnitude. The constant height method can only be used to image very flat areas of the sample, and requires the relationship between it and the tipsample separation if images of the sample surface are to be obtained in terms of the z displacement. This is usually achieved by calibration with a surface of known geometry. Images from both methods are in the form of line scans; image enhancement procedures may then be employed to convert the raw data into grey-scale or colour images. 197 CHEMICAL SOCIETY ,REVIEWS. 1992 Figure 1 STM image of HOPG in air. Tungsten tip, bias voltage 5 mV, ntunnelling current 5 nA.(Reproduced with permission from Dr. G. A. D. Briggs.) X X1 b b Figure 2 Schematic representations of (a) three-way piezo-electric Figure 3 Schematic representation of the (a) constant current, and (b) scanner and (b) a tube scanner. constant height modes of operation of an STM. ELECTROCHEMICAL ASPECTS OF STM AND RELATED TECHNIQUES-P. A. CHRISTENSEN 2.2 Principles The basis of all tunnelling microscopies is electron tunnelling through a potential barrier. This process has no chemical analogy, and it is most easily understood by considering the progress of a free electron wave from left to right in Figure 4a. For simplicity, we will only consider tunnelling between metals. In regions I and I11 the electron has zero potential energy, all of its energy is kinetic.In region 11, the electron experiences a potential U,, over a region from z = 0 to z = d. In regions I and 111, the Schrodinger equation for the electron takes the simple form: [ -h'/2tw,].d2 Y/dz2 = EY (1) where meis the mass of the electron, and Eis the energy as shown in the figure. This can be solved immediately to give, in region I: where k = [2rn,E/h2]11'2,A and Bare constants, and the momen- tum of the electron is just hk. The first term represents the electron transition from left to right; the second allows for interference between the incident wave and the wave corres- ponding to reflection from the barrier. Evidently, in region 111: where F is a constant, since only the transmitted wave is of interest.In region 11, by contrast, the appropriate form of the Schrod- inger equation is: [h2/2m,]d2Y/dz2= [U, + E] (4) and provided U, > E, has the solution: where k' = [2rn,(U0 + E)/h2]1'* The dominant term, save close to the II/III boundary, is the second, and physically this corresponds to the exponential decay of the wavefunction as the barrier is traversed. This is a remarkable result: classically, we would expect the electron to be completely reflected at the 1/11 interface, but quantum mecha- nics shows that this will not happen unlessU, co.At finite --f values of U,, the wavefunction decays exponentially into the barrier, with I Y I xe-2k'z. It is instructional to evaluate k':for E= 2 eV and U, = 6 eV, we have k' = loto m-l.Hence,I YyIII decays by a factor of ca. 7for every I A thickness of the barrier. The tunnelling probability, PT(E), through the rectangular barrier of Figure 4a can be obtained approximately by calculat- ing the total decay in the wavefunction over the entire barrier. This is clearly given by: where d is the sample-tip separation. If the barrier is not rectangular, and the potential energy is now a function of z, then it can be readily shown that: -r PT(E) = exp{ -21 [2m,{ U(z)+ E)/h2]+dz) (7) which is termed the WKB approximation, and zi and z, are the classical turning points, i.e. the values of z on the left and right of the barrier for which U(z)= E. For a trapezoidal barrier, of the type shown in Figure 4b, we can replace 21 and z, by 0 and d, respectively. Figure 4b represents the form of Figure 4a for the tip and sample in an STM, with the tip and sample at equili- brium, i.e.joined only by a wire. Figure 4c represents the case when the sample is made positive with respect to the tip: i.e. a bias potential of Vb is applied, where Vb = Vsample-Vtip.Under these conditions, tunnelling occurs from tip to sample, as shown. At low bias voltages, such as those typically employed in an STM experiment (i.e.for the case given, of tunnelling from tip to sample, 0 < Vb<0.6 V), the tunnelling current is given by: Where D,(EF,,) and Dt(EF,t)are the density of states at the Ferrni level of the sample and tip, respectively. Thus, at a fixed (positive) bias, tunnelling is from filled states near the Fermi level of the tip, to empty states on the sample.D,(EF.,) and Ds(EF,s)are constant, and the tunnelling current is effectively a function only of the tipsample separation. This correlation between it and topography breaks down when the surface is not electronically homogenous, e.g. when adsorbed species are present, or the sample is a semiconductor. This gives rise to spatially-dependent Local Density Of States, LDOS, and hence Ds(EF,s)shows a strong spatial distribution. Thus, a more correct representation of Figure 4c is as shown in Figure 4e; the particular LDOS depicted in the figure will only be true at the particular position of the tip over the surface of the sample.Figure 5 shows a schematic representation of the tip and sample in an STM experiment. A typical tip diameter is ca. lpm, consequently, if the figure was drawn to the scale of the 'atoms' on the surface, the tip would ca. be 10 metres in diameter. Clearly, if the situation was as depicted in Figure 5, the STM would not be capable of atomic resolution. The reason it is lies in the exponential dependence of the tunnelling current on the sample-tip separation, as given in equation 8. Hence, it is thought that tunnelling occurs via the selection of one or two atoms lying proud of the mean surface of the tip. Providing that d is large compared to k'-l, then tunnelling occurs from this same atom, or atoms, irrespective of the lateral position of the tip above the surface; the tip density of states can thus be considered as constant.In general, k'x 1 A-whilst d> 5 A. If d< 1 A, then the precise definition of 'd'is lost, and the tip LDOS show a dependence upon tip position. Thus the vertical resolution of STM is ca. 0.1 A; the lateral resolution is set by the intersection of the solid cone of current with the surface, and is typically 4-7 A.7 The above empirical treatment really only applies to experi- ments performed in a vacuum, with no account taken of the effect of placing an electrolyte solution above the sample. Soon after the publication of the initial reports on STM, some controversy existed amongst (increasingly interested) electro- chemists as to whether or not the technique would function when tip and sample were immersed in even a weak electrolyte. The concern was that ions of opposite charge from the tip would cluster around it, under the influence of the large electric fields across such a sharp point, perhaps providing numerous tunnel- ling pathways to degrade the lateral resolution.This was not found to be the case; one explanation given was that electrolytes are ionic conductors but electronic insulators. However, the presence of the electrolyte does have an, as yet unquantified, effect upon the images obtained. Thus, discrepancies have been observed between images obtained under UHV conditions, and those obtained in the presence of electrolyte.* It is only recently that workers such as Sass and colleagues9 have started systema- tically to investigate the effect upon the tunnelling barrier height, and shape, of the intricate interactions between electrons and a dense ensemble of polar molecules.One approach to such questions, suggested by these workers, is that the trajectory of the tunnelling electrons is not simply described by a solid cone between tip and sample. Instead, the orientational relaxation of particular solvent molecules, or clusters of molecules, may open up particular lower energy routes between tip and sample. These relaxation times will vary with the solvent, and may determine the variation in lateral resolution when the tip-sample spacing is changed. It is clear that our understanding of the physics of CHEMICAL SOCIETY REVIEWS, 1992 Total energy of electron h -- E incident & reflected uo transmitted .V Z Energy,E ,+-d--+, *Z Tip Sample (b) w Sample Figure 4 (a) Schematic representation of the tunnelling of an electron through a rectangular barrier. (b) The energy levels of tip and sample when joined by a wire. EF is the Fermi level, +t and +s are the work functions of tip and sample, respectively, and d is the tipsample separation. (c) As for Figure 4b, except showing the effect of an imposed positive bias voltage. The bias voltage is defined as Vsample-VtlP.(d) Variation in log, &, (itin picoamps), as a function of bias voltage, calculated on the basis of the principles in equation 8.(After C. R. Leavens and G. C. Aers in reference 7b, p. 32). (e) Schematic representation of the energy levels of tip and sample, when sample shows some spatially-dependent Local Density Of States, LDOS. ELECTROCHEMICAL ASPECTS OF STM AND RELATED TECHNIQUES-P. A. CHRISTENSEN 20 1IT Sample Figure 5 Schematic representation of the STM tip and sample. electron tunnelling across a liquid gap, particularly for polar solvents, is still incomplete, as is our ability to interpret correctly the images obtained from in situ STM. However, for the purposes of this article, the treatment given above is sufficient. 2.3 Practicalities 2.3.1 Tips Tips, typically Pt/Ir, Au, W, or Ta, are sharpened, either simply by fracture or via electrochemical etching.One such method involves forming a ‘soap bubble’ of 2M NaOH within a Pt/Ir loop, and inserting a length of tungsten tip wire. An a.c. potential of ca. 1-1 5 V peak-to-peak is then applied between tip and loop, at a frequency of 50 Hz. The wire is etched until breaking point, at which time the weight of the wire helps to pull it out into a very fine point. 2.3.2 Samples The samples should be as flat as possible, if atomic resolution images are required, otherwise a great deal of time is lost searching for atomically flat areas; common samples include Pt or Au evaporated or sputtered onto glass or mica. Layered materials such as Highly Ordered Pyrolytic Graphite, (HOPG), or MoS, can be cleaved to reveal fresh, clean surfaces simply by placing sellotape onto the old surface and peeling it away.2.3.3 Piezo-elec t ric Crystals The original STM instruments employed three orthogonal piezo elements, one to control scanning in each direction, (see Figure 2a). The device favoured more recently is the tube scanner, as shown in Figure 2b. This still allows fine control in all three directions, but consists of only a single unit. 2.3.4 Cleanliness An essential practical point is to keep the sample clean, i.e. protected from airborne contamination, since adsorption can cause noise on the tunnelling current, as well causing the surface to reconstruct. One way to maintain a clean surface is to hold it in a UHV chamber, the other is to cover it with a highly purified liquid, the latter indicating a possible advantage of operating in situ.However, a common contaminant of aqueous electrolyte is C1-. Even very low concentrations of chloride ion have been shown to increase substantially the mobility of noble metal ions at anodic potentials, giving rise to an unstable surface. As a result, great care must be exercised when employing a SCE reference electrode to prevent leakage of the C1- electrolyte from the reference into the cell. 2.3.5 Potentiostatic Control of the Sample For STM experiments performed in vacuo, it is sufficient simply to apply a bias voltage across the tip and sample in order to obtain tunnelling; and this approach was also employed by Sonnenfeld and Hansmal O in their early experiments intended to show that STM images could be obtained with a layer of electrolyte above the sample.However, in these experiments, the potential of the sample is undefined with respect to any fixed reference, an unsatisfactory scenario for an electrochemist. Electrochemical control of a sample is usually provided via a three-electrode potentiostat and the potential of the sample or working electrode is maintained and monitored very precisely with respect to a non-current carrying reference electrode. Faradaic current flows only between the working and counter electrodes, the latter being usually a Pt loop or gauze. Providing such potentiostatic control of the sample, as well as maintaining a fixed bias potential between sample and tip, is a tricky problem.The essence of this problem is that Vb must be held constant when the sample potential is changed, otherwise spec- troscopic information becomes mixed in with the topographic data (see section on Scanning Tunnelling Spectroscopy). Thus, on stepping the sample potential, E,, prior to collecting an image, then the tip potential Et must also be changed so as to maintain a constant Vb. Most of the in situ electrochemical experiments reported of late (see, for example, references 11-15) employ independent control of Et and E, with respect to the reference electrode. The two most popular methods in this respect are: A three-electrode potentiostat employing a current-carrying refer- ence electrode. Such a reference can support the passage of small currents without altering its potential.The tip acts as a fourth electrode held either at a constant potential with respect to the reference, or at a constant bias voltage with respect to the sample. Bipotentiostatic control. In effect, the tip and the sample are controlled as two separate and independent working electrodes: this allows true independent control of Et and E, with respect to a non-current carrying reference electrode. In both of the above methods, the bias voltage is given by 2.3.6 Noise There are four principal sources of noise in STM: Imposed vibrations. The tipsample separation must be stabi- lized on the sub-8i scale. Hence, the system must be isolated from any external vibrations, particularly those of the building itself.The first STM system achieved this by levitating the entire UHV arrangement on superconducting magnets. Over the few years since, it has been found that much simpler (and cheaper) methods can be just as effective. Thus, a common arrangement used to isolate the STM from vibration is shown in Figure 6. The STM ‘head’, including the inchworm/tube scanner/tip mount assembly and electrochemical cell, sit upon four or five heavy metal plates; the plates are separated by halved rubber ‘0’ring dampers. The vibrational mismatch between each plate/damper pair gives a total reduction in vibration of at least three orders of magnitude. The whole instrument is then suspended via four elasticated ropes from the ceiling, or from a suitable frame.In order to prevent acoustic noise, the instrument is covered with a bell jar; this arrangement also allows for some degree of environ- mental control. Electronic noise. Careful shielding of the electronics is required, including placing the STM in a Faraday cage. Thermal drft and piezo-electric creep. These effects render it essential that images are collected quickly, e.g. in ca.610 seconds. Uncontrolled Faradaic reactions at the tip. These are a potentially major source of noise in electrochemical STM, and to avoid such complications, the potential of the tip should be maintained within the double-layer region (the region of constant current in Figure 7) wherever possible. However, even if the tip potential CHEMICAL SOCIETY REVIEWS, 1992 Flrl 111 8 Figure 6 Schematic representation of the STM head and electrochemi- cal cell assembly.(1) Inchworm motor, (2) Inchworm, (3) Faraday cage shield around tube scanner, (4) Teflon pot electrochemical cell, (5) working electrode sample, (6) stainless steel plates, (7) halved Viton ‘0’rings, (8) elasticated ropes attached to baseplate. The counter and reference electrodes, and the electrical connection to the sample, are not shown for clarity. can be maintained in its double layer region, Faradaic reactions can still occur, and it is thus essential that the tip wire be insulated up to as near to the tip itself as possible, in order to minimize the available area accessible to the electrolyte.The insulating materials that have been employed include nail var- nish, as well as various waxes. Figure 7 Typical electrochemical response of an STM tip. The Faradaic current flowing in the cathodic region is due to H, evolution and the anodic current is due to 0, evolution or tip dissolution. Once it was realised that STM could be performed in situ in an electrochemical environment, the technique was seized upon by electrochemists who rapidly developed the necessary sophisti- cated potentiostatic control, and proceeded to produce increas- ingly exciting and elegant work5J’ [references 11-15 and refer- ences therein]. An example of the potential of STM to provide a causal link between observed electrochemistry and the topogra- phical behaviour of the surface at the atomic level is the work of Itaya, Sugawara, Sashikata, and Furuya’ which is now described.2.3.7 The Potential-Dependent Reorganization of Pt(l1I) The report by Itaya and colleagues concerns the application of STM to study the effects of potential cycling on the topography of a Pt(l11) electrode, and follows the general interest over recent years in the study of well-defined highly-ordered crystals, and of platinum in particular because of its fundamental techno- logical importance as the foremost electrocatalyst. Polycrystal- line surfaces are too complicated for such studies, whereas the well-defined, homogeneous, and essentially flat nature of a single-crystal surface provides a readily-understood starting point.An important observation which links single-crystal Pt( I 1 1) electrochemistry with that of the polycrystalline mater- ial, arises from cyclic voltammetry. Cyclic voltammetry involves the repeated linear ramping of the potential of the sample electrode between two preset limits and simultaneously moni- toring the current that passes. If Pt( 11 1) is cycled in this manner between a potential above that at which hydrogen evolution takes place and one in the double layer region, as shown in Figure 8a, then the surface is stable; the various features in the voltammogram merely indicate the formation and removal of surface hydrides. However, if the electrode is cycled up to a potential where surface oxides are formed, and back, then the cyclic voltammogram is observed to revert rapidly to that expected for polycrystalline Pt, see Figure 8b.1 0 0.5 V vs. RHE (4 vs. RHE -1oc Figure 8 Cyclic voltammograms for a Pt( 1 1 1) electrode immersed in 0.05 M H,SO,. (a) 0.1 V to 0.65 V versus the Reversible Hydrogen Electrode, (RHE); (b) 0.05 V to 1.5 V versus RHE. Itaya and colleagues decided to employ STM to see if any correlation existed between the surface topography and the electrochemistry in Figures 8a and b. The images were obtained at constant current, with the tip and sample controlled independently. Figure 9 shows a token STM image of a 1000 8, x 10008, area of the Pt( 11 1) surface. The image was collected at 0.95 V versus the Reversible Hydrogen Electrode, RHE, i.e.in the double layer region, (see Figures 8a and b). As can be seen, the surface is ELECTROCHEMICAL ASPECTS OF STM AND RELATED TECHNIQUES-P. A. CHRISTENSEN dominated by steps of ca. 2.3 A, consistent with monatomic height, (the diameter of Pt is 2.38 A). The steps cross at ca. 60°, such that the terraces so formed are triangular in shape, as expected for a surface with threefold symmetry. 7 1 nm 100Ir I I / I/,,x,, , I Y" 50 100 nm Figure 9 STM image of a 100 nm x 100 nm Pt(ll1) facet surface obtained in 0.5 M H,SO,. The electrode potentials of the Pt sample and Pt tip electrodes were 0.95 V and 0.9 V, respectively. The tunnelling current was 2 nA. Scan speed was 200 nmjs. (With the permission of Professor K.Itaya and the Journal of Vacuuni Science and Technology. taken from K. Itaya, S. Sugawara, K. Sashik-ata, and N. Furuya, J. Vuc. Sci. Tcdinol., 1990, AS. 515.) Figure 10 shows an image of a 1000 8, x 1000 8, area of the surface collected at 0.95 V after five potential cycles between 1.5 V and 0.05 V. After one potential cycle, noisy signals appeared on the terraces; although even after 5 cycles, the original monatomic steps can still clearly be distinguished, and the position of these steps remain unchanged. This observation is very important as it indicates that the only change induced by the potential cycling is the formation of the disordered struc- tures on the terraces. From Figure 10, it can be seen that the disordered structures on the terraces are randomly oriented islands, ca.30-50 8, in diameter, and one or two atoms high. nm Z 11 nm Figure 10 STM image of a 100 nm x 100 nm area of the Pt( 1 1 1) surface obtained after 5 potential cycles between 1.5 V and 0.05 V. The conditions were as in Figure 9. (With the permission of Professor K. Itaya and the Journal of Vacuum Science and Technology, taken from K. Itaya, S. Sugawara, K. Sashik-ata, and N. Furuya, J. Vuc. Sci Technol., 1990, AS, 5 15) The authors observed that neither the location of the steps nor the disordered structures on the terraces changed with time in the absence of further potential cycling. However, they did observe that an image collected at 1.5 V (i.e. well into the region where the surface is covered with oxide) during the first cycle shows no differences from that in Figure 9, strongly suggesting that it is the reduction of this oxide surface, rather than its formation, that produces the structures observed in Figure 10.The authors noted that the islands formed on the terraces are only one or two atoms high, strongly suggesting that formation of the oxide is accompanied by place-exchange, see Figure 1 1. Hence, they concluded that the roughening of the Pt( I1 1) surface is via this place-exchange mechanism: once the oxide layer is stripped off, the Pt atoms left behind as a result of the place-exchange (in the form of adatoms) do not go back into their original positions, and this results in the observed topogra- phical changes.It does not seem unreasonable to correlate the formation of these randomly-oriented small islands with the appearance of the polycrystalline platinum electrochemistry. Step Figure 11 Schematic representation of island formation on a Pt( 11I) surface, induced by electrochemical cycling. The 'eyewitness' nature of the work of Itaya and colleagues speaks volumes for the potential of STM in electrochemistry. 3 Scanning Tunnelling Spectroscopy The images obtained in STM are often strongly dependent upon the magnitude and sign of the bias voltage, as indicated by equation 8. Equation 8 assumes that D,(EF,,) and Dt(EF,t) are uniform across the surface. As was discussed above, this is a reasonable approximation for a clean metal surface (hence the tip density of states can be regarded as spatially independent). However, for a metal upon which is an adsorbate, or for non- metals, D,may show a marked spatial variation.As a simple example of this, consider Figure 12a, which shows a homo- geneous metal surface having constant density of sites, except for the presence of a single impurity atom. Let the alien atom be more electronegative than its neighbours, such that the energy diagram at the alien atom resembles that shown in Figure 12b. If the surface is scanned in constant current mode, at a bias potential between vbl and vb2, (see Figure 12b), then the trajectory followed by the tip will be as shown in Figure 12a. On passing over the alien atom, the tunnelling current from the tip will be reduced, and so the feedback loop will have to move the tip closer to the surface to maintain the current.Hence, electro- nic information has been mixed in with the topography. The presence of the alien atom could be verified by holding the tip at a constant height above the surface, ramping the bias potential, and monitoring the tunnelling current. For locations other than that over the alien atom, we may expect a smooth increase in i, with bias potential, (for a homogeneous and uniform density of states). However, over the alien atom, this smooth increase will have a ripple superimposed upon it where the tunnelling current is reduced due to the filled LDOS, see Figure 12c. Thus, as far as STM is concerned, it is essential when imaging electronically in- homogeneous surfaces to separate electronic from topographic effects.Fortunately, as was mentioned above, for STM on clean metal surfaces the STM image is a fairly close reflection of the surface topography. However, the surface of a semiconductor may be totally different, as a given wavefunction is often preferentially localized on specific atoms or bonds. As a result, in order to understand STM images of a semiconductor, a detailed understanding of the surface electronic structure is required. ’\-_ ,‘ Tip Sample (b) Figure 12 (a) Schematic representation of the trajectory followed by an STM tip in constant current mode over a metal sample. The surface has one alien atom, more electronegative than its neighbours. (b)Energy level diagram corresponding to case in Figure 12a.(c) A schematic representation of the’STS spectrum expected if tip is poised over alien atom depicted in Figure 12a. CHEMICAL SOCIETY REVIEWS, 1992 STS is a means of mapping out the surface electronic structure with atomic resolution. As we have seen, (Figure 4e), the LDOS of non-metal surface need not be constant. As tunnelling passes through the LDOS in Figure 4e, the tunnelling current increases or decreases slightly; this ‘ripple’ in the current is due to the tunnelling passing through the states comprising the LDOS and is superimposed upon the increasing background described above. This is the basis for Scanning Tunnelling Spectroscopy; the measurement of it versus Vb at fixed lateral and vertical position.In practice, in order to be able to clearly identify the energy of the ‘onset’ of these ripples (and hence the energies of the various states comprising the LDOS) the i/ Vplot is generally differentiated and the spectrum is thus presented as di,/dVb versus Vb. In order to obtain as much information as possible, Vb should be ramped between the widest possible limits; consequently, the majority of STS studies so far reported have been performed under UHV conditions, or in air [reference 7, and references therein]. However, useful data can still be obtained in electrolyte solutions, as will be seen below. There are essentially four common approaches to STS reported in the literature.Of these, Atom Selective Imaging,16 Tunnelling Spectroscopy, ’ and Current Imaging Tunnelling Spectroscopy (CITS)’* have only been reported under UHV conditions, and so will not be discussed here. The forms of Tunnelling Spectroscopy that have been employed in situ are simple variations of the above techniques. Tomita, Matsuda, and Itaya [see, for example, reference 19 and references therein] have employed STS in situ to the study of ZnO, Si, and TiO, in aqueous solution. The initial experiments on ZnO and TiO, involved monitoring the potential applied to the 2-piezo, (i.e.the movement of the tip in the :-direction), as the bias potential was ramped at constant tunnelling current. The study on p- and n- Si(100) single-crystal electrodes’ involved the more conven- tional method of monitoring it as a function of Vb at a constant tipsample separation. The tip and sample potentials were controlled separately with respect to a SCE electrode via a bipotentiostat.Thus, Figure 13 shows the variation in the tunnelling current as the electrode potential of the n-Si sample was ramped between -0.5 V and 0.4 V versus SCE; the tip potential was maintained at 0.45 V versus the reference, i.e. in the double layer region of the platinum tip. As can be seen from the plot, the tunnelling current decreased as the potential on the sample was increased, but did not fall off to zero near the flat band potential, EFB,of the n-Si, as would be expected in the absence of any surface states.The shape of the plot was explained by the authors in terms of the existence of surface states. Thus, Figures 14a to c show schematic representations of the tunnelling process from sample to tip, (Et>EJ, for the cases where (a) E, < EFB,(b) E, = EFB, and (c) E, > EFB, at constant tip potential; in the presence of surface states. In Figure 14a, tunnelling can occur from both the conduction band and the surface states; in Figure 14b, tunnelling can occur from all of the surface states. However, any surface states can only be popu- lated up to a maximum energy less than or equal to the Fermi level and, as can be seen from Figure 14c, as the potential on the sample increases fewer states are populated; but the surface states are filled with electrons from the conduction band, hence the observed plateau in the tunnelling current.Once the Fermi level energy is lower than the lower limit of the surface state energy, the tunnelling current decreases to zero as E, is increased to Et.In the absence of such states, tunnelling could only occur when Es> EFB,(see Figure 14a). The authors commented that extending the range of vb, by moving the tip potential to more cathodic values, would require the use of non-aqueous solvents owing to problems with hydrogen evolution at the tip. 4 Atom Force Microscopy 4.1 Introduction The enormous success of STM has encouraged the development of a wide range of high resolution probes sensitive to force, ion ELECTROCHEMICAL ASPECTS OF STM AND RELATED TECHNIQUES-P.A. CHRISTENSEN flow, magnetic fields, capacitance etc. The resolution of these probes varies according to the property measured and the size of the probe. In STM, the atomic resolution is a result of the /-<0.5 0 -0.5 V vs. SCE Figure 13 Tunnelling current versu.Y electrode potential of n-Si at fixed tip-sample separation. The electrode potential of the tip was 0.45 V versus the saturated calomel electrode (SCE). The electrode potential of the silicon was scanned from -0.5V versus SCE, at a scan rate of 1 v s-1. (With the permission of Professor K. Itaya and the Journal of Vacuum Science and Technology, taken from E. Tomita, N. Matsuda, and K. Itaya, J. Vac. Sci. Technol., 1990, AS, 534.) Figure 14 Schematic representations of the tunnelling process from an n-type semiconductor sample to the tip, for the cases where: (a) the sample potential, E,, is less than the flat band potential, EFB,(b) E, = E,,, and (c) Es> EFB.The tip potential is constant and the semiconductor has surface states (SS on the figure).exponential dependence of the tunnelling current on the tip sample separation, and the consequent selection of one or two atoms on the tip for tunnelling. The atomic resolution of Atom Force Microscopy, AFM, is a result of the same form of selection, due to the inverse power law dependence of force on distance. In contrast, the magnetic force sensor depends upon several thousand magnetic moments in the tip to give measur- able force; as a result, the resolution of such approaches is limited to several hundred A.The most common force microscopy so far employed is AFM, and it is this approach that we will concentrate upon. AFM was first reported by Binnig, Quate, and Gerber in 1986,O and senses the van der Waals forces between the tip and sample; as a result, this technique does not require either the tip or sample to be conducting. The dispersion forces between two bodies a distance R apart are most commonly described by the differentiated form of the Lennard-Jones potential: F(R) = [(l2Cl/Rl3)-(6C,/R7)].R(R), (10) where C,and C, are constants and R(R)is the unit vector in the R direction. The second term in equation 10 describes the long range attractive force, and the first term is the short-range repulsion experienced when the electron clouds on two atoms come into contact.AFM has been used to image surfaces UHV,,I under liquid,,, and, more recently, in an electrochemical cell with the electrode under po ten tios ta tic control. 4.2 Mechanics The exact approach employed depends upon whether the attractive or repulsive forces between tip and sample are to be probed. In both attractive and repulsive AFM, the tip is mounted on a cantilever spring, see Figure 15, with the ampli- tude or frequency of the spring deflection being used to obtain the surface topography as the tip is rastered across the surface. The tip is either in direct contact with the sample, (repulsive mode), or maintained above the sample, (attractive mode).In general, in both modes, the sample is moved rather than the delicate tip mount. In attractive AFM, the tip is poised above the sample, and the cantilever vibrated using a piezocrystal. The fine (x,y,z)move-ment of the sample is achieved by mounting the sample on a piezo-electric stage. The tip is then vibrated by a second piezo- crystal at an imposed frequency near resonance, and with an amplitude of ca. 20 A. This large vibration amplitude requires a correspondingly large tip/sample separation, and in conse-quence leads to the reduced lateral resolution of this approach. The attractive force experienced by the tip is detected as the force gradient, dF(R)/dR, which is given by:24 dF(R)/dR = Av/[v,/~~] (1 1) where dv is the change in vibration frequency caused by the attractive force, vR is the resonance frequency of the spring, and k is the spring constant of the cantilever.dv is measured via the determination of the change in amplitude of vibration as the tip is rastered across the surface. The image is then a plot of force versus lateral position. In the repulsive mode of operation, the tip is in actual contact with the surface. An image can then be obtained in one of two principal ways; in the simplest approach, the sample is moved beneath the tip, exactly as in a record-player, the vertical deflection of the tip following directly the corrugation of the surface. This method has the great advantage of being simple, and therefore potentially cheap. The second approach involves vibrating the tip, which is still in contact with the sample, at a 2--2 I Figure 15 Schematic representation of an AFM electrochemical cell.(1) photodiode, (2) electrolyte solution inlet/outlet, (3) spring clip, (4) cantilever holder, (5) glass cell body, (6) '0' ring, (7) sample, (8) s,j.,r translator, (9) mirror, (10) tip. frequency of several hundred Hz and an amplitude at the tip of' 5-20 A, the exact conditions depending upon the resonance frequency and spring constant of the cantilever. The amplitude of vibration is reduced on contact with the sample, this reduc- tion being proportional to the corrugation of the surface. Profiling is achieved by moving the sample up or down during a scan to maintain a constant vibration amplitude; the image is then simply a contour map of (.Y,,Y,z), and AFM should not, in theory, suffer from the interpretational difficulties experienced in STM.4.3 Practicalities The original method of detecting the deflection of the tip in the early (repulsive) AFM experiments, was via a STM tip mounted behind the AFM tip, the latter having a metal contact on the rear of the cantilever. By applying a bias voltage across the two, the tunnelling current can then be used to determine the vertical displacement of the AFM tip. This is rather a complicated approach; in essence, all that is actually required is a means of producing, and monitoring, a signal that varies rapidly with the movement of the AFM tip.More recent approaches commonly employ optical techniques, the most simple of these involves measuring the deflection of a laser beam from a mirror mounted on the cantilever spring. The deflection is monitored by a position-sensitive detector (see Figure 15). The use of optical techniques as the means of monitoring spring deflection has several advantages, including the fact that photons impose only a tiny force upon the tip for transmission to the sample. This has allowed AFM to be used non-destructively to image adsorbed organics, for which the force exerted on the sample must be < lo-" N. CHEMICAL SOCIETY REVIEWS, 1992 It is known that the amplitude, A, of a vibration induced in a spring in response to an imposed vibration of amplitude A,, and frequency v,, is: A = A,[v/v,]' where the resonance frequency of the cantilevcr is given by: where m is the mass of the spring.Thus, in order for the vibrational noise on the tip to be as low as possible, uR must be as high as possible; in turn, this indicates that the mass of the spring should be low. Tips with high resonance frequencies are now commercially available. Typically, a building vibration has A, -lpm and v = 20 Hz; for a common cantilever spring resonance frequency of 10 kHz, this gives a noise amplitude of 0.04 A. This will cause very little interference and, as a result, atomic resolution is possible with very little vibrational isolation, in complete contrast to STM.4.3.1 The Underpotential Deposition qf'Copper on Gold( 111) Underpotential deposition (upd) is a much-studied phenome- non in electrochemistry, and is the electrochemical reduction of a metal cation to form a monolayer or submonolayer of the corresponding metal at the surface of an electrode. The critical point is that deposition occurs at a potential higher than that dictated by the reversible potential of the metal/metal cation couple, suggesting that such a upd layer is energetically quite different from the bulk metal. However, subsequent deposition on a upd monolayer occurs at the expected potential and the resulting surface is typical of the bulk metal. The upd process at noble metal surfaces is of considerable interest for several reasons: (i) as a model of deposition in general; (ii) upd surfaces show increased resistance to poisoning by oxidation products and thus are of interest with respect to fuel cell and biosensor development; (iii) the distinctive chemical properties of the upd surface.A representative example of the upd process is copper on gold and an extremely illuminating study upon this system using repulsive AFM was reported by Manne, Hansma, Massie, Elings, and Ge~irth.~~ The authors employed a commercially available AFM, the essentials of which are shown in Figure 15. The reference electrode was a copper wire in contact with the electrolyte at the outlet of the cell. The counter electrode was the stainless steel spring clip holding the AFM cantilever in place.The working electrode was a 100 nm-thick evaporated Au film [which is known to expose mainly the Au( 11 I) surface] mounted upon an (XJ))translator. Figure 16 shows the Au( 11 1) surface in 0.1 M HClO,,/ 1 x lop3M CuClO, at + 0.7 V, prior to the deposition of the Cu. The (x,y)displacement ofthe translator was calibrated using the known atom-atom spacing of the close-packed Au( 1 11) lattice of 2.9 A.The image was the same when 0.1 M H,SO,/ 1 x 10 M CuSO, was employed as the electrolyte; a surprising observation in the light of the fact that SO;-is thought to be strongly adsorbed at this potential. The authors postulated that either sulfate is not as strongly adsorbed as was previously thought, or the AFM tip 'pushes away' the adsorbed SO:-during its passage across the surface.Figure 17shows an image collected at -0.1 V in the perchlor- ate solution after the bulk deposition of several monolayers of Cu. The Cu-Cu distance in the figure was found to be 2.6 A, which was taken as evidence of the Cu atoms simply sitting 'on top' of the underlying gold atoms. Sweeping the potential up to + 0.1 1 V stripped off all but the initially deposited upd layer. At this point, the authors observed that the nature of the electrolyte had a major effect upon the structure of the adsorbed Cu. Thus, Figure 18a shows a schematic representation Au( 1 1 1):upd Cu surface in the perchlorate electrolyte. The Cu-Cu spacing was the same as for the Au( 11 1) surface at 0.7 V. 2.9 A.The Cu ELECTROCHEMICAL ASPECTS OF STM AND RELATED TECHNIQUES-P.A. CHRISTENSEN Figure 16 AFM image of the Au( 11 I) surface in 0.1 M HC10,/1 x lop3 M CuClO, at 0.7 V. prior to copper deposition. The reference electrodc was a copper wire in contact with the same electrolyte. The AU-AUspacing is 2.9 A. (By pern~ission of Professor A. Gewirth and Science, taken from S. Manne. P. K. Hansnia, J. Massie, V. B. Elings, and A. A. Gewirth, Scicuico, 1991, 251, 185. Copyright 1991 by the AAAS). Figure 17 AFM image of Au(1 11) surface after bulk deposition of several monolayers of Cu. The potential of the gold is -0.1 V, and cucuspacing is 2.6 A. lattice, however, was no longer exactly overlying, i.e. commen-surate with the underlying gold lattice.It appeared that the Cu lattice direction was rotated by 30" f10" relative to the gold. The authors attributed this re-arrangement to the strain of incorporating the larger copper atoms into a close-packed structurc cxactly over the Au (1 1 1) surface (Cu radius 1.75A,Au I .42 A). In 0.I M H,SO, at + 0.144V, the upd Cu layer formed a much more open lattice (see Figure 18b) with a Cu-Cu spacing of 4.9 A; an observation supported by STM.24 The authors postulated that the more open structure found in the sulfuric acid electrolyte was due to the stabilization of the Cu monolayer by co-adsorption with SO:-, although they had not observed this co-adsorbed sulfate directly. This was supported by radio-chemical data which showed that, prior to Cu upd on gold, there is little or no sulfate on the gold surface.Immediately following the deposition, substantial sulfate adsorption occurs. Figure 18 (a) Schematic representation of the incommensurate close- packed overlayer of Cu on Au formed in the perchlorate electrolyte. The open circles are the gold atoms. Only part of the monolayer is shown in order to exhibit the overlayer-underlayer orientation. (b) Schematic representation of the more open lattice formed in the sulfuric acid electrolyte. On sweeping the potential back to + 0.7 V, the upd Cu was removed and the original Au( 1 11) surface regained, showing unequivocally that the upd of Cu on Au( 1 1 1)in either electrolyte is completely reversible, contrary to some previous speculation.Thus, AFM is a very powerful complement to STM and has the great advantage of not involving quantum-mechanical tunnelling. This means that non-conducting samples can be imaged, both in terms of the surface and any adsorbate, and that images can be obtained when Faradaic current is flowing, as in the study discussed above. The latter could not be done with STM, as a Faradaic current can have a disastrous effect upon the tunnelling current. Hence AFM offers the very real possibility of real-time imaging, and is certain to become an important in situ tool of the future in electrochemistry. Acknoit>/t.dgernmts.I would like to thank Professor A. Hamnett for helpful consultations, and both he and Mr J. Brooker for critical appraisal of the manuscript.5 References 1 'Comprehensive Chemical Kinetics', ed., R. G. Compton and A. Hamnett, Elsevier, Amsterdam, Vol. 29, 1989. 2 'Spectroelectrochemistry: Theory and Practice', ed. R. J. Gale, Plenum Press, New York, 1988. 3 G. Binnig, H. Rohrer, Ch. Gerber, and E. Weibel, Phys. Rev.Lett., 1982,49, 57. 4 T. R. I. Cataldi and G. A. D. Briggs, personal communication. 5 R. Sonnenfeld, J. Schneir, and P. K. Hansma, 'Modern Aspects of Electrochemistry', ed. R. E. White, J. O'M. Bockris, and B. E. Conway, Plenum Press, New York, 1990, p. I. 6 T. R. I. Cataldi, I. G. Blackham, G. A. D. Briggs, J. Pethica, and H. A. 0.Hill, J. Electround. Chem., 1990, 290, 1. 7 (a)R. J. Haniers, Am. Rev. Phys. Chem., 1989,40,531: (b)'Scanning Tunnelling Microscopy and Related Methods', ed.R. J. Behm, N. Garcia and H. Rohrer, NATO AS1 Series, Series E: Applied Sciences Vol 184, Kluwer Academic Publishers, Dordrescht, 1990. 8 J. V. Coe, G. H. Lee, J. G. Eaton, S. T. Arnold, H. W. Sarkas, K. H. Bowen, C. Ludewigt, H. Haberland, and D. Worsnop. J. Ckeni. Phjvs., 1990,92, 3980. 9 W. Haiss, D. Lackey, J. K. Sass, and K. H. Besocke, .I. Chtw. Phys., in press. 10 R. Sonnenfeld and P. K. Hansma, Scitwce, 1986,232, 21 1. I1 F.-R. Fan and A. J. Bard, J. Electrochmi. Soc., 1989, 136, 3216. 12 R. Christoph, H. Siegenthaler, H. Rohrer, and H. Wiese, Elcclro-diiw.Actcr. 1989,34, 10I 1 . 13 D. J. Trevor, C. E. D. Chidsey, and D. N. Loiacono, Phys. Rev. Lett., 1989, 62,929. 14 H. Honbo and K. Itaya, Nouv. J. Chim., in press. 15 K. Itaya, S. Sugawara, K. Sashikata, and N. Furuya, J. Vuc. Sci. Technol., 1990, AS, 515. 16 R. M. Feenstra, J. A. Stroscio, J. Tersoff, and A. P. Fein, Phys. Rev. Lett., 1987,58, 1192. 17 R. S. Becker, J. A. Golovchenko, D. R. Hamann, and B. S. Swartzentruber, Phys. Rev. Lett., 1985, 55, 2032. 18 R. J. Hamers, R. M. Tromp, and J. E. Demuth, Phys. Rev. Lett., 1986,56, 1972. 19 E. Tomita, N. Matsuda, and K. Itaya, J. Vuc. Sci. Technol., 1990, AS, 534. CHEMICAL SOCIETY REVIEWS, 1992 20 G. Binnig, C. F. Quate, and Ch. Gerber, Phys. Rev. Lett., 1986, 12, 930. 21 J. A. Stroscio, R. M. Feenstra, and A. P. Fein, Phys. Rev. Lett., 1987, 58, 1668. 22 0.Marti, B. Drake, and P. K. Hansma, Appl. Phys. Lett., 1987,51, 484. 23 S. Manne, P. K. Hansma, J. Massie, V. B. Elings, and A. A. Gewirth, Science, 1991, 251, 183. 24 0.M.Magnussen, J. Hotlos, R. J. Nichols, D. M. Kolb, and R. J. Behm, Phys. Rev. Lett., 1990,64,2929. 25 P. K. Hansma, V. B. Elings, 0.Marti, and C. E. Bracker, Science, 1988, 242, 209.
ISSN:0306-0012
DOI:10.1039/CS9922100197
出版商:RSC
年代:1992
数据来源: RSC
|
|