摘要:

Chemical Society Reviews Editorial Board Jean-Pierre Sauvage (CNRS, Strasbourg) [Chair] Vicenzo Balzani (Bologna) Ed C. Constable (Basel) Chris Elschenbroich (Marburg) Tim C. Gallagher (Bristol) Editorial Office Martin Sugden (Managing Editor) David Bradley; Peter Whittington (Production) Debbie Halls (Editorial Secretary) http://chemistry .rsc.org/rsc tel: +44 (0)1223 420066 Chemical Society Reviews publishes concise, succinct and lightly referenced articles that provide an introductory overview to topics of current interest in chemistry. The articles appeal to the general research chemist as well as to the expert in the field and provide an essential starting point for further reading. Advanced undergraduates, postgraduates and experienced re- searchers should all benefit from reading Chemical SocietyReviews. Chemical Society Reviews (ISSN 0306-0012) is published bimonthly by the Royal Society of Chemistry, Thomas Graham House, Science Park, Cambridge, UK CB4 4WF.1997 subscription rate: &130 (USA $234). Customers in Canada will be charged the sterling price plus a surcharge to cover GST. Individuals can subscribe for &45 (USA $80) providing their institutional library takes a full price subscription. All orders accompanied by payment should be sent directly to The Royal Society of Chemistry, Turpin Distribution Services Ltd, Blackhorse Road, Letchworth, UK SG6 1HN. (NB Turpin Distribution Services Ltd., distributors, is wholly owned by the Royal Society of Chemistry.) Payment should be by cheque in pounds sterling payable on a UK clearing bank or in US dollars payable on a US clearing bank.Second class postage is paid at Zdenek Herman (Prague) Horst Kunz (Mainz) John P. Maier (Basel) D. Mike P. Mingos (Imperial) Jeremy K. M. Sanders (Cambridge) Royal Society of Chemistry Thomas Graham House Science Park Cambridge UK CB4 4WF csr@rsc .org fax: +44(0)1223 420247 The Editorial Board commissions articles that encourage international, interdisciplinary dialogues in chemical research. The Board welcomes any suggestions for new articles. A guide for authors and synopsis form can be found in the first issue of this year’s volume or on the RSC’s World-Wide Web home page (URL above).Alternatively, they can be requested from the Managing Editor, in paper or electronic form (postal and e- mail address above). Jamaica NY 1141-9998. Airfreight and mailing in the USA by Publications Expediting Services Inc., 200 Meacham Avenue, Elmont, NY 11003 and at additional mailing offices. US Postmaster: send address changes to Chemical Society Review, c/o Publication Expediting Services Inc., 200 Meacham Ave- nue, Elmont NY 11003. All dispatches outside UK by bulk airmail within Europe and Accelerated Surface Post outside Europe. 0 The Royal Society of Chemistry, 1997 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic, mechanical, recording, or otherwise, without the prior permission of the publishers. Typeset and printed in Great Britain by Black Bear Press Limited.

ISSN:0306-0012    

DOI:10.1039/CS99726FX009

出版商:RSC    

年代:1997

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99726BX011

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Roger J. Mortimer Department of Chemistry, Loughborough University, Loughborough, Leicestershire, UK LEI 1 3TU Electrochromic materials are currently attracting much interest in academia and industry for both their fascinating fundamental spectroelectrochemical properties and their commercial applications. A large number of electrochromic materials are available from all branches of syntheticchemistry. In this review some of the most importantexamples from the major classes of electrochromic materials, namely the transition metal oxides, Prussian Blue systems, l,l’-disubstituted-4,4’-bipyridylium salts (the viologens), conducting polymers, metallopolymers and metal phthalo- cyanines are described. Examples of their use in both prototype and commercial electrochromic devices are given.1 Introduction Chemical species that can be electrochemically switched between different colours are said to be electrochromic. Electrochromism results from the generation of different visible region electronic absorption bands on switching between redox states.1 The colour change is commonly between a transparent (‘bleached’) state and a coloured state, or between two coloured states. In cases where more than two redox states are electrochemically available the electrochromic material may exhibit several colours and be termed polyelectrochromic. Likely applications of electrochromic materials include their use in controllable light-reflective or light-transmissive devices for optical information and storage, anti-glare car rear-view mirrors, sunglasses, protective eyewear for the military, con- trollable aircraft canopies, glare-reduction systems for offices, Roger Mortimer is a Lecturer in Physical Chemistry at Loughborough University.After gaining his BSc in chemistry (1977) from imperial College, he remained there to carry out PhD research (1977180) under the supervision of Michael Spiro. As a postdoctoral fellow (1980181) and visiting associate in chemistry (1988) with Fred Anson at California institute of Technology he developed interests in modified electrodes. His interest in electrochromic materials was stimulated on collab- oration with David Rosseinsky at the University of Exeter (198144). After lecturing posi- tions in Physical Chemistry at Cambridgeshire College of Arts and Technology (now Anglia Polytechnic University) and Analytical Chemistry at Shef- field City Polytechnic (now Shefield Hallam University), he took up his present post in 1989.His current research in- terests include electrochro-mism, spectroscopic and elec- trochemical ion recognition, electrochemical sensors and electrocatalysts for fuel cells. and ‘smart windows’ for use in cars and in Of these, electrochromic car rear-view mirrors have already achieved considerable commercial success. These safety de- vices prevent mirror-reflected glare which causes an ‘after image’ to stay on the eye’s retina. Whilst many types of chemical species exhibit electroch- romism, only those with favourable electrochromic perfor- mance parameters1 are potentially useful in commercial appli- cations.Thus most applications require electrochromic materials with a high contrast ratio, colouration efficiency (absorbance change/charge injected per unit area), cycle life, and write-erase efficiency (% of originally formed colouration that may be subsequently electro-bleached). Whereas displays need fast response times, by contrast ‘smart windows’ can tolerate response times of up to several minutes. Electrochromic materials are generally first studied at a single working electrode, under potentiostatic or galvanostatic control, using three-electrode circuitry. Traditional techniques such as cyclic voltammetry, coulometry, chronoamperometry , all with, as appropriate, in situ spectroscopic measurements are employed for characterisation.For electrochromic device (ECD) investigations a simple two-electrode system is con- structed in a sandwich configuration (Fig. 1). Such an ECD can --glasssubstrate IT0 coating -electrochromic electrode -polymer electrolyte -charge-balancing counter electrode -ITOcoating -glass substrate Fig. 1 Schematic diagram of a solid-state electrochromic device (ECD) suitable for a transmissive light-modulation application be thought of as a rechargeable electrochemical cell, in which the ‘electrochromic electrode’ (where the colour switching takes place) is separated from a charge-balancing counter electrode by a solid (often polymeric) or liquid electrolyte. Colour changes in the ECD occur by charging/discharging the electrochemical cell on application of an electrical potential.After the resulting pulse of current has decayed and the colour change has been effected, the new redox state persists, with little or no input of power, in the so-called ‘memory effect’. The electrochromic electrode, which can work in the reflective or transmissive mode, is generally glass coated with an electrically conducting film such as tin-doped indium oxide (ITO), onto which is deposited the electrochromic material. Alternatively, if one or both redox states are soluble, the electrochromic material may be present dissolved in the electrolyte solution. In variable light transmissive devices the counter electrode substrate also has to be transparent IT0 glass, with the counter electrode chemical species being either colourless in both its redox forms or electrochromic in a complementary mode to the primary electrochromic material.For applications that are designed to operate in the reflective mode, such as displays, the counter electrode can be of any material with a suitable reversible redox reaction. The purpose of this review is to give a flavour of the diversity of this fascinating subject by the description of some of the most Chemical Society Reviews, 1997, volume 26 147 important examples from the major classes of electrochromic materials, and to give some examples of their use in both prototype and commercial ECDs.2 Transition metal oxides Many transition metal oxide films can be electrochemically switched to a non-stoichiometric redox state which has an intense electronic absorption band due to optical intervalence charge transfer.l.4 A good example is the tungsten trioxide system, which, since its electrochromism was first reported in 1969, has been the most widely studied electrochromic material.1.4 Tungsten trioxide, with all tungsten sites as oxidation state Wvl, is transparent as a thin film. On electrochemical reduction, Wv sites are generated to give the electrochromic effect. Although there is still controversy about the detailed colouration mechanism, it is generally accepted that the injection and extraction of electrons and metal cations (Li+, H+, .. .) play a key role. In the case of Li+ cations the reaction can be written as eqn. (1). W03+x(Li++ e-) -+ LixWzl- x,WyO, (1)(transparent) (blue) At low x the films have an intense blue colour caused by photo- effected intervalence charge transfer (CT) between adjacent Wv and WVI sites. At higher x, insertion irreversibly forms a metallic ‘bronze’ which is red or golden in colour. Electrochromic tungsten trioxide coatings have been pro- duced by a number of different deposition techniques including thermal evaporation in vacuo, electrochemical oxidation of tungsten metal, chemical vapour deposition (CVD), sol-gel methods and by RF-sputtering. Tungsten trioxide research has always been driven by the many possible commercial applica- tions and prototype alphanumeric displays and electrochromic mirrors were soon reported.A major present aim is the development of ‘smart windows’ for control of thermal conditions within a building, thereby reducing winter heating and summer cooling requirements. Glass manufacturers have recognised this opportunity, with the Pilkington Technology Centre having produced a prototype electrochromic window (dimensions 0.7 X 1 m) which when coloured is capable of reducing light transmission by a factor of four.3 Such electro- chromic systems are of a simple two-electrode sandwich-device construction, as described above. In the development of such variable transmission windows, glass companies favour the sputtering technique because it is already in place for the production of a range of coatings for architectural glazing~.~ Many other thin-film transition metal oxides are electroch- romic,1.4 for example the oxides of molybdenum and vanadium, eqn.(2) and (3). Moo3+x(M++ e-) -+ M,MO:*-(transparent) (blue) .,MoY03 (2) V205+ x(M++ e-) -+ (yellow) MxV205 (pale blue) (3) In these examples, the more intensely absorbing redox state is produced on reduction (cathodic ion-insertion). In contrast, Group VIII metal oxides become coloured on electrochemical oxidation (anodic ion-insertion); as in the case of hydrated iridium oxide (strictly iridium hydroxide). The mechanism of colouration is uncertain, with both proton extraction and anion insertion routes being proposed, eqn.(4) and (5). II-(OH)~ -+Ir02eH20 + H++ e-(4)(transparent) (blue-black) Ir(OH)3+ OH--+ Ir02.H20 + H20 + e-(5)(transparent) (blue-black) Another commonly studied anodic ion-insertion material is nickel oxide, a material more commonly known for its use in secondary batteries. Nickel oxide (strictly nickel hydroxide) in basic electrolytes switches from pale green to brown-black. 3 Prussian blue Prussian blue [PB, iron(m) hexacyanoferrate(~~)] is the proto- type of a number of polynuclear transition metal hex-acyanometallates which form an important class of insoluble mixed-valence compounds.6 They have the general formula M’k[M’’(CN>& (k, 1 integral) where M’ and M” are transition metals with different formal oxidation numbers.The first report7 concerning the electrochemistry and electrochromism of PB prompted numerous investigations into the properties of PB thin films. PB thin films are generally formed by electrochemical reduction of solutions containing iron(rr1) and hexacyanofer- rate(m) ions.8 Reduction of the brown-yellow soluble complex Prussian brown [PX, iron(I1r) hexacyanoferrate(m), present in equilibrium with the iron(m) and hexacyanoferrate(rI1) ions], is the principal electron-transfer process in PB electrodeposition, eqn. (6). [Fe111Fe111(CN)6]+ e--+[FeI1IFe1I(CN)6]-(6)PX PB Charge-compensating cations (initially Fe3+, then K+ on potential cycling in K+-containing supporting electrolyte) are present in the PB film for electroneutrality.9 Partial electro- chemical oxidation of PB in pure supporting electrolyte yields Prussian green (PG), a species historically known as Berlin green (BG), eqn.(7). 3[Fe1I1Fe”( CN)6] --+ PB [Fe1113{ Fe111(CN)6}2{ FeII(CN)6}]-+ 2e-(7) PG Although in bulk form PG is believed to have a fixed composition with anion composition as above, for thin films there is a continuous composition range between PB and PX, which becomes a golden yellow in the fully oxidised form? The latter may be obtained by electrochemical oxidation of a particularly pure form of PB,*,9 eqn. (8). [Fe111Fe11(CN)6]--+ [Fe11JFe111(CN)6]+ e-(8)PB PX Reduction of PB yields Prussian white (PW), also known as Everitt’s salt, which appears transparent as a thin film, eqn.(9). [Fe111Fe11(CN)6]+ e--+ [FeIIFeII(CN)6]2-(9)-PB PW For all the electrochromic redox reactions above, there is concomitant ion ingress/egress in the films for electroneutral- ity. The spectra9 of PX, PG, PB and PW are shown in Fig. 2, together with two intermediate states between blue and green. The intense blue colour in the [Fe111Fe11(CN)6] -chromophore of PB is due to an intervalence charge-transfer (CT) absorption band centred at 690 nm. The yellow absorption band in PX corresponds with that of [FeIIIFeIII(CN)6] in solution, both maxima (h,,, 425 nm) coinciding with the (weaker) [FeIrl- (CN)6l3- absorption maximum. On increase from +0.50 v to more oxidising potentials, the original PB peak continuously shifts to longer wavelengths with diminishing absorption, while the peak at 425 nm steadily increases, owing to the increasing [Fe111Fe111(CN)6]absorption. The reduction of PB to PW is in contrast abrupt, with transformation to all PW or all PB without pause, depending on the potential set.In the CV of a PB- modified electrode, the broad peak for PB-PX in contrast with the sharp PB-PW transition emphasises the range of composi- tions involved. This difference in behaviour, supported by ellipsometric measurements,lO indicates continuous mixed-valence compositions over the blue to yellow range in contrast with the presumably immiscible PB and PW which clearly transform one into the other without intermediacy of compo- sition.Early PB ECDs employed PB as sole electrochromic material. Examples include a seven-segment display using PB- 148 Chemical Society Reviews, 1997, volume 26 0.6 04 0.2 \ 00 400 500 600 700 800 880 960 1040 1120 1200 Wnm Fig. 2 Spectra of iron hexacyanoferrate films on ITO-coated glass at various potentials [(i) +0.50 (PB, blue), (ii) -0.20 (PW, transparent), (iii) +0.80 (PG, green), (iv) +0.85 (PG, green), (v) +0.90 (PG, green), and (vi)+1.20 V (PX, yellow) (vs. SCE)] with 0.2 mol dm-3 KCl +0.01 mol dm-3 HCl as supporting electrolyte (Reproduced with permission from J. Chem. Soc., Dalton Trans., 1984, 2059) modified Sn02 working and counter electrodes at 1 mm separation] 1 and an IT0 I PB-Nafion I IT0 solid-state device.’2 For the solid-state system, device fabrication involved chem- ical, rather than electrochemical formation of the PB on immersion of a membrane of the solid polymer electrolyte Nafion (a sulfonated polytetrafluoroethane polymer) in aqueous solutions of FeC12, then K3Fe(CN)6.The resulting PB-containing Nafion composite film was sandwiched between the two IT0 plates. The construction and optical behaviour of an ECD utilising a single film of PB, without addition of a conventional electrolyte, has also been described.13 In the design, a film of PB is sandwiched between two optically transparent electrodes (OTEs) [Fig. 3(a)]. Upon application of dh -n Prussian white Prussian yellow (a) (b) Fig. 3 Schematic cross section of a single film PB cell.The cell is shown (a) without and (b) with a voltage applied. (Reproduced by permission from J. Electrochem. SOC., 1990, 137,2464.) an appropriate potential across the film, oxidation occurs near the positive electrode and reduction near the negative electrode to yield PX and PW, respectively [Fig. 3(b)]. The conversion of the outer portions of the film results in a net bleaching of the device. The functioning of the device relies on the fact that PB can be bleached both anodically (to the yellow state) and cathodically (to a transparent state), and that it is a mixed conductor through which potassium ions can move to provide charge compensation required for the electrochromic redox reactions. Since PB and tungsten trioxide are respectively anodically and cathodically colouring electrochromic materials, they can be used together in a single device so that their electrochromic reactions are complementary, eqn.(10) and (1 1). [Fe111Fe11(CN)6]-+ e--+ [Fe1IFe1*(cN)6]*-(blue) (transparent) (10) W03 +,(M+ + e-) -+MxWZ1- x,Wy03 (transparent) (blue) (1 1) In the construction of such a device, thin films of these materials are deposited on OTEs that are separated by a layer of a transparent ionic conductor such as KCF3S03 in poly(ethy1ene oxide).l4 The films can be coloured simultaneously (giving deep blue) when a sufficient voltage is applied between them such that the W03 electrode is the cathode and the PB electrode the anode. Conversely, the coloured films can be bleached to transparency when the polarity is reversed, returning the ECD to a transparent state.Several PB analogues have been reported as thin films in modified-electrode studies. l5 Whilst the majority are expected to be electrochromic, absorption spectra as a function of redox state have rarely been reported, and only ruthenium purple [iron(rII) hexacyanoruthenate(~~)] and osmium purple [iron(IrI) hexacyanoosmate(~r)] have been used in prototype seven-segment ECDs. The feld therefore appears to be open for further investigation and exploitation. 4 Viologens Diquaternisation of 4,4’-bipyridyl produces 1,l ’-disubstituted- 4,4’-bipyridylium salts16 (Scheme l), commonly known as Scheme 1 ‘viologens’. The compounds are formally named as 1,l’-di- substituent-4,4’-bipyridylium if the two substituents at nitrogen are the same, and as 1-substituent- 1’-substituent’- 4,4’-bipyridylium should they differ.The prototype, 1 ,l‘-di- methyl-4,4’-bipyridylium, is often known as methyl viologen (MV), with other simple symmetrical bipyridylium species being named substituent viologen. Of the three common viologen redox states (Scheme l), the dication is the most stable and is colourless when pure unless optical charge transfer with the counter anion occurs. Reductive electron transfer to viologen dications forms coloured radical cations, the stability of which is attributable to the delocalisation of the radical electron throughout the n-framework of the bipyridyl nucleus, the 1 and 1’ substituents commonly bearing some of the charge.Electrochromism occurs in viologens because, in contrast to the viologen dications, the radical cations have a delocalised positive charge, colouration arising from an intramolecular electronic transition. Suitable choice of nitrogen substituents to attain the appropriate molecular orbital energy Chemical Society Reviews, 1997, volume 26 149 0’5, levels can, in principle, allow colour choice of the radical cation. Simple alkyl groups, for example, promote a blue-violet colour whereas aryl groups generally impart a green hue to theradical cation. The intensity of the colour exhibited by di-reduced viologens is low since no optical charge transfer or internal transition corresponding to visible wavelengths is accessible.For display applications, the write-erase efficiency of ECDs using short alkyl chain viologens such MV in aqueous electrolytes would be low since both dicationic and radical- cation states are very soluble. Improvements may be made in MV-based ECDs by retarding the rate at which the radical- cation product of electron transfer diffuses away from the electrode and into the solution bulk either by tethering the dication to the surface of an electrode, or by immobilising the viologen species within a semi-solid electrolyte such as Nafion (Fig. 4).17 The solubility4iffusion problem can be avoided by 0.1 0 300 400 500 600 700 800 hlnm Fig. 4 Spectra recorded at t = 0, 10, 20, 30,40 and 50 s in response to a potential step from +1.00 to -0.90 V for an ITO/Nafion/l,l’-dihexyl-4,4’-bipyridylium electrode in 0.1 mM 1,I’-dihexyl-4,4’-bipyridylium dibromide +0.2 mol dm-3 KCl (pH 5.5).The absorbance at all wavelengths increases with time. The colour appears pink, indicating a high proportion of radical cation dimers in the Nafion. (Reproduced by permission from Electrochromic Materials ZZl, ed. K.-C. Ho, C. B. Greenberg and D. M. McArthur, PV 96-24, pp. 3-13, Electrochem. SOC. Proc. Ser., Pennington, New Jersey, 1997). the use of viologens having long alkyl-chain substituents at nitrogen, for which the coloured radical-cation is insoluble. Of this type, l,l’-diheptyl-4,4’-bipyridylium(heptyl viologen, HV) as the dibromide salt has been the most thoroughly studied.l8 HV2+ dication is soluble in water, but forms an insoluble film of crimson radical-cation salt adhering strongly to the electrode surface following a one-electron reduction.An ECD display (Fig. 5) using the HV system was reported to have a response time of 10-50 ms, with a cycle life of > lo5 cycles between redox states.18 The high response times noted above are however not essential in other applications such as electrochromic car rear- view mirrors and ‘smart windows’. In fact, Gentex’s commer- cialised automatic-dimming interior ‘Night Vision Safety’ (NVS) mirror functions wholly by solution electrochromi~rn.~.~ In this system, an ITO-glass surface (conductive side inwards) and the reflective metallic surface, spaced a fraction of a millimetre apart, form the two electrodes of the cell, with a solvent containing two electroactive chemical species that function both as electrochromic materials and supporting electrolyte. The system may be inferred to operate as f0llows.~.3 a = anode c = cathode r = reference electrode Fig.5 Alphanumeric character, from a heptyl viologen display device. (Reproduced by permission from Appl. Phys. Lett., 1973, 23, 64). A substituted (cationic) viologen serves as the cathodic- colouring electrochromic material, with a negatively charged (possibly) phenylene diamine as the anodically colouring electrochromic material. When the mirror is switched on, the species will move by electrical migration to their respective electrodes.After the dual electrochromic colouration process is initiated, the products will diffuse away from their respective electrodes and meet in the intervening solution where a mutual reaction regenerating the original uncoloured species takes place. Thus, in this type of ECD, maintenance of colouration requires application of a continuous small current for replenish- ment of the coloured electroactive species lost by their mutual redox reaction in solution. Bleaching occurs at short or open circuit by homogeneous electron transfer in the bulk of the solution. While not an electrochromic phenomenon, the ingen- ious control system is worth noting. A photosensitive detector is placed facing rearward to monitor any dazzling incident light but, in daylight, it would be triggered, resulting in an unwanted darkening of the mirror.This is avoided by the second forward- looking detector, which on seeing daylight, is programmed to cancel any operation of the controlling sensor, which thus responds only in the dark of night. Gentex Corporation sell their rearview mirrors to 16 major car manufacturers with (in 1996) NVS mirrors being available on 95 vehicle models. The company also manufacture exterior wing mirrors, and from 1987 to the end of 1996 have sold a total of 10 million electrochromic mirrors. 5 Conducting polymers Chemical or electrochemical oxidation of numerous resonance- stabilised aromatic molecules including pyrrole, thiophene, aniline, furan, carbazole, azulene and indole produces novel electronically conducting polymers.19 In their oxidised forms, such conducting polymers are ‘doped’ with counter anions (p-doping) and possess a delocalised n-electron band structure, the energy gap between the highest occupied TC electron band (valence band) and the lowest unoccupied band (the conduction band) determining the intrinsic optical properties of these materials. The doping process (oxidation) introduces polarons (in polypyrrole for example, these are radical cations deloc- alised over ca. four monomer units) which are the major charge carriers. Reduction of conducting polymers with concurrent counter anion exit removes the electronic conjugation, to give the ‘undoped’ (neutral) electrically insulating form. Conducting polymers can also, in principle, undergo cathodic doping with cation insertion (n-doping) to balance the injected charge.However, n-doped forms are less stable than p-doped forms and few reports are available concerning the electrochromism of the n-doping process. All conducting polymers are potentially electrochromic in thin-film form, redox switching giving rise to new optical absorption bands together with transfer of electrons/counter anions. In this review, some examples from the three main classes of conducting polymer that have been investigated for their electrochromic properties will be described. A major advantage of using conducting polymer for electrochromic 150 Chemical Society Reviews, 1997, volume 26 applications lies in the fact that subtle modifications to the monomer can significantly alter the spectral properties of the electrochromic material.Examples will be noted where suitable monomer modification allows ‘tuning’ of colour states. 5.1 Polypyrrole(s) The electrochromic properties of polypyrrole are generally investigated using thin-film polypyrrole prepared by electro- chemical polymerisation of pyrrole from acetonitrile solution. The oxidative electropolymerisation process is initiated by monomer oxidation to yield a radical-cation species. Poly- pyrrole generation, in its oxidised conducting form, then follows via a mechanism19 that is believed to involve either radical-cation/radical-cationcoupling or attack of radical-cation on neutral monomer (Scheme 2).Polypyrrole thickness is QI HI--I I H H I -2H+ H H \ /x Scheme 2 controlled through the charge passed, further film growth occurring at the polypyrrole-solution interface. Doped (oxidised) polypyrrole is blue-violet (Amax 670 nm),20 electrochemical reduction yielding the yellow-green (Amax 420 nm) ‘undoped’ form, eqn. (12). p-doping (12)undoping insulating conductive (yellow-green) (blue-violet) Removal of all dopant anions from polypyrrole yields a pale- yellow film, however, complete de-doping is only achieved if films are extremely thin. This means that polypyrrole of thickness commensurate with device construction (> 1 pm) has a low contrast ratio.The electrochromism of polypyrrole is unlikely to be exploited, mainly due to the degradation of the film on repetitive colour switching. Conducting polymers with im- proved electrochromic properties are however formed on electrochemical polymerisation of 3,4-disubstituted pyrroles.1 5.2 Polythiophene(s) As for polypyrrole, polythiophene thinfilms may be prepared by the electrochemical oxidation of solution monomer via in this case the thiophene radical cation. The electrochromic properties of polythiophene and of the polymers of several substituted thiophenes are reproduced in Table 1 .20 Table 1 Polythiophenes“ Polymer h,,/nm and colour Monomer Oxidised Reduced Thiophene 730 470 Blue Red 3-Methylthiophene 750 480 Deep blue Red 3,4-Dimethylthiophene 750 Dark blue 620 Pale brown 2,2’-Bithiophene 680 460 B lue-gre y Red-orange Adapted with permission from F.Gamier, G. Tourillon, M. Gazard and J. C. Dubois, J. Electroanal. Chem., 1983, 148, 299. Tuning of colour states is possible by suitable choice of thiophene monomer. For example, the electrochromic proper- ties of polymer films prepared from 3-methylthiophene-based oligomers are strongly dependent on the relative positions of methyl groups on the polymer backbone (Table 2). Table 2 Colours of polymers derived from oligomers based on 3-methylthiophene“ Monomer Polymer colour (figure does not represent the molecule’s stereochemistry) Oxidised Reduced Pale blue Purple Violet Yellow Blue Red Blue Orange Blue Yellow Violet Yellow Blue-violet Yellow Blue Yellow-orange a Reproduced by permission from M.Mastragostino, in Applications of Electroactive Polymers, ed. B. Scrosati, Chapman and Hall, London, 1993, ch. 7, p. 244. In research directed to using the same electrochromic material for both the working and counter electrodes, a series of conducting polymer films based on 3-(p-X-phenyl)thiophene monomers (X = -CMe3, -Me, -OMe, -H, -F, 41, -Br, -CF3, -S02Me) have been investigated.2l The presence of electron- withdrawing groups on the phenyl ring serves to assist the stabilisation of the n-doped state and these materials can be both reversibly reduced and oxidized (n- and p-doped).A model ECD using poly(cyclopenta[2,1 -b;4,3-b’]dithiophen-4-(cyano,nonafluorobutylsulfony1)-methylidene),a low bandgap con- ducting polymer which is both p- and n-dopable, as both the anode and the cathode material has been reported.22 Chemical Society Reviews, 1997, volume 26 151 5.3 Polyaniline(s) The electrical and electrochromic properties of polyaniline not only depend on its oxidation state, but also on its protonation state, and hence the pH of the electrolyte used. The electro- chemistry of polyaniline thin films has been extensively investigated in aqueous acid solutions and in organic media, and several redox mechanisms involving protonation-deprotona- tion and/or anion ingress/egress have been proposed. l9 Scheme 3 depicts the mechanism of electropolymerisation of aniline, +H+ 11 -Hi J 1 2a 2b 2a + 2b -2H+ I -H+ 1.1 -H+ Scheme 3 with Scheme 4 giving the composition and redox pathways of the various redox states in the product polyaniline.23 Polyaniline films are polyelectrochromic (transparent yellow to green to dark blue to black), the yellow-green transition being durable to repetitive colour switching.The wavelengths of the absorption maxima of polyaniline and the polymers of two substituted anilines are given in Table 3, with some of the spectra for poly(rn-toluidine) illustrated in Fig. 6.24The two low-wavelength spectral bands observed in the polyanilines are assigned to an aromatic n-n* transition (S330 nm) related to the extent of conjugation between the adjacent rings in the polymer chain, and to radical cations formed in the polymer matrix (S440nm).With increase in applied potential the S330 nm band absorbance decreases and the 440 nm increases (Fig. 6). Beyond +0.30 V the conducting region is entered; the S440 nm band decreases as a broad free carrier electron band ca. 800 nm is introduced. Polyaniline has been combined with PB in complementary ECDs that exhibit deep blue f-, green electrochromism.' Electrochromic compatibility is obtained by combining the coloured oxidised state of the polymer with the blue PB and the bleached reduced state of the polymer with PG: Oxidised polyaniline + PB -+ emeraldine polyaniline + PG (deep blue) (green) Both liquid electrolyte and solid-state configurations have been described.152 Chemical Society Reviews, 1997, volume 26 Table 3 Visible absorption spectra maxima, hm,,(*4)/nm, with applied potentials (EN)parenthesised for the polymer films on IT0 glassa Polyaniline 320, 440 (-0.20) 320, 440, > 820 (+0.20) 330,430, 800 (+0.30) 340, 420, 800 (+0.40) Poly(o-toluidine) 308, 420 (-0.20) 308, 420 (+0.20) 312, 410, > 820 (+0.30) 380, 800 (+0.40) Poly(m-toluidine) 304, 420 (-0.20) 304,420 (+0.20) 304,410, > 820 (+0.30) 306, 390, > 820 (+0.40) Reproduced by permission from R. J. Mortimer, J.Muter. Chem.,1995,5, 969. 0.10 8 me nz 0.05 0.00 I I I I I I 300 400 500 600 700 800 h/nm Fig.6 Visible absorption spectra of poly(m-toluidine) films on ITO-coated glass in 1mol dm-3 hydrochloric acid at the potentials: i, -0.20 V; ii, +O. 10 V; iii, +0.20 V; iv, +0.30 V. (Reproduced by permission from J. Mater. Chem., 1995, 5, 969.) Whilst electropolymerisation is a suitable method for the preparation of relatively low surface area electrochromic conducting polymer films, it may not be suitable for fabricating large-area coatings. Significant effort therefore goes into the synthesis of soluble conducting polymers such as poly(o-methoxyaniline) which can then be deposited as thin films by casting from solution. In a novel approach large-area elec- trochromic coatings have been prepared by incorporating polyaniline into polyacrylate-silica hybrid sol-gel networks using suspended particles or solutions and then spray or brush- coating onto IT0 surfaces.25 Silane functional groups on the polyacrylate chain act as coupling and cross-linking agents to improve surface adhesion and mechanical properties of the resulting composite coatings.6 Metallopolymers Transition metal coordination complexes are potentially useful electrochromic materials because of their intense colouration and redox reactivity. Chromophoric properties typically arise from low-energy metal-to-ligand charge transfer (MLCT), intervalence CT, intraligand excitation, and related visible- region electronic transitions. Because these transitions involve oxidation , emeraidine salt (conductor) /fJDNflDJHI I \ H/n \ I H t /n / H H \ \ /n leucoemeraldine (yellow) -2(H+ + X-) pKa-3-4 +2(H+ + X-)11 deprotonation \ emeraldine base /n (blue) -2 (e-+ H+ + X-) +2 (e-+ H+ + X-)I1 pernigraniline(black) Scheme 4 valence electrons, chromophoric characteristics are altered or eliminated upon oxidation or reduction of the complex.While these spectroscopic and redox properties alone would be sufficient for direct use of transition metal complexes in solution-phase ECDs, polymeric systems have also been investigated. Many schemes have been described for the preparation of thin-film ‘metallopolymers’,26 the reductive electropolymerisation of suitable polypyridyl complexes being a particularly versatile technique.This technique relies on the ligand-centred nature of the three sequential reductions of complexes such as [RuII(~bpy)g]~+ (vbpy = 4-vinyl-4’-methyl-2,2’-bipyridine), combined with the anionic polymerisability of suitable ligands. Vinyl-substituted pyridyl ligands (examples in Fig. 7) are generally employed, although metallopolymers have also been formed from chloro-substituted pyridyl ligands, via electrochemically initiated carbon-halide bond cleavage. In either case, electrochemical reduction of their metal complexes generated radicals leading to carbon-carbon bond formation and oligomerisation. Oligomers above a critical size are insoluble and thus thin films of the electroactive metal- lopolymer are produced on the electrode surface.Fig. 8(a) shows repetitive cyclic voltammograms (CVs) for a typical electropolymerisation pr0cess.~7 The film thickness increases with CV scan number owing to the electrocatalytic reduction of the fresh solution metal complex monomer by the growing electroactive polymer. The steady increase in current is attributable to the combined electroactivity of the orange polymeric film and the inward-diffusing metal complex mono- mer. The electroactivity of the resulting modified electrode [Fig. 8(b)]is stable on transfer to pure supporting electrolyte solution, orange c-) transparent electrochromicity being ob- served. The colour of such metallopolymer films in the MI1 redox state may be selected by suitable choice of the metal (e.g.M = Fe, red; M = Ru, orange; M = Os, green). Electro- chromicity results from loss of the MLCT absorption band on switching between the MI1 and the MI11 redox states. Inter- estingly, the chromophoric properties of the coloured state of such supramolecular metallopolymers are modified on binding of Group I/II metal cations by the crown ether groups.27 In a recent development, spatial electrochromism has been demonstrated in metallopolymeric films.28 Photolysis of poly- [R~~~(vbpy)2(py)2]C12thin films on IT0 glass in the presence of chloride ions leads to photochemical loss of the photolabile pyridine (py) ligands and sequential formation of pol y [Rur1(vbpy)2(py)C1] C1 and pol y [RuII(v bpy)2Cl2] (Scheme 5). Contact lithography can be used to spatially control the photosubstitution process to form laterally resolved bicom- ponent films with image resolution below 10 pm.Dramatic changes occur in the colours and redox potentials of such ruthenium(1r) complexes upon substitution of chloride for the Chemical Society Reviews, 1997, volume 26 153 pyridine ligands (Scheme 5). Striped patterns of variable colours are observed on addressing such films with a sequence of potentials. 7 Metal phthalocyanines Since the first report of the polyelectrochromism of thin films of lutetium bis(phtha1ocyanine) ([LU(PC)~])in 1970, numerous metal phthalocyanines have been investigated for their electro- chromic properties.’ Such compounds have a metal ion either at the centre of a single phthalocyanine (Pc) ring (S), or between 8 two rings in a sandwich-type compound.[Lu(Pc)2], an example of the latter type, is generally formed as a vivid green film on vacuum sublimation. [Lu(Pc)*] films undergo a series of ring- based redox processes. On oxidation, films can be switched first to a yellow-tan form and then to a red form. On reduction the green state can be switched first to a blue redox form and then to a violet-blue form. Although, as described, [Lu(Pc)*] films can exhibit five colours, only the blue-green transition is utilised in most prototype ECDs. Mechanical problems, such as film fracture and/or loss of adhesion to the electrode substrate, arise from anion ingresdegress during colour switching. Despite such difficulties, [L~(Pc)~]-based electrochromic dis- plays with good reversibility, fast response times, and little degradation over > 5 X 106 cycles have been described.29 Although as noted [Lu(Pc)2] films are usually prepared by vacuum sublimation, the practical problems encountered with extending this technology to the manufacture of practical devices have been rec~gnised.~o These include the slow rate of deposition, the partial decomposition of [Lu(Pc)~] under the sublimation conditions and the presence of electrochemically inaccessible sites in the resulting films.Electropolymerisation of [Lu(T4APc)2] (T4APc = 4,4’,4”””’-tetraaminophthalocya-nine) has been reported as a possibly viable route to the fabrication of practical ECDs.30 The oxidative film-formation mechanism is assumed to be analogous to that described for polyaniline, although the poly[L~(T4APc)~] films are believed to be oligomeric in nature rather than truly polymeric.Although loss of electroactivity in such films was found on sweeping to positive potentials (likely to be the result of the loss of electronic and/or redox conductivity), the electrochemistry in DMSO at -l,O} -800 -600 400 -200 ZO 200 400 600 1000 ‘*’ 0.0 -0.4 -0.8 -1.2 -1.6 -2.0 f/Vys. SSCE -1 000 I -800 200 n 4oo[ -400 wN’wN/ A / E/v vs. SSCE Fig. 8 (a)Fifteen sequential CVs at 100 mV s-1 in an acetonitrile solution containing 1.33 mmol dm-3 [Ru1*L3][PF6I2(L = 4-(benzo-l5-crown-3 5)vinyl-4’-methyl-2,2’-bipyridine) and 0.1 mol dm-3 Bu4NBF4.Initial scan direction negative. (b)CVs as a function of scan rate (v/mV s-1 as labelled on each) for the poly[RuL3]*+ modified electrode from (a) in supporting electrolyte. (Reproduced by permission from J. Chem. Soc., Faraday Trans., 1993, 89, 333.) 6 7 Fig. 7Structures of various vinyl-substituted ligands. 3 = 4-vinyl-pyridine (vpy), 4 = 4-vinyl-4’-methyl-2,2’-bipyridine(vbpy), 5 = 4’-vinyl-2,2’ :6’,2”-terpyridine (vtpy), 6 = 4-(benzo-15-crown-5)vinyl-4’-methyl-2,2’-bipyridine, 7 = 4-(aza-15-crown-5)styryl-4’-methyl-2,2’-bipyridine. Scheme 5 154 Chemical Society Reviews, 1997, volume 26 negative potentials is well-behaved, with the observation of two broad quasi-reversible one-electron redox couples. Spec-troelectrochemical measurements revealed switching times of < 2 s for the observed green-grey-blue colour transitions in this region.The electropolymerisation method is also applicable to single-ring transition metal phthalocyanine c~mplexes,~~ the preparation of films with reasonable optical densities requiring minutes compared to the several hours potential cycling that is needed for [Lu(T4APc)Z]. The redox reactions and colour changes of two of the metal complexes studied are summarised in Scheme 6. pol y [Co”T4APcI + ne-+pol y [Co1T4APc]-(blue-green) (yellow-brown) poly[CorT4APc]-+ ne-+ poly[Co1T4APcI2-(yellow-brown) (red-brown as thick films or deep pink as thin films) p0ly[Ni~~T4APc]+ ne-+poly[Ni1*T4APc]-(green) (blue) poly[Nir1T4APc]-+ ne--+ poly[NiI1T4APcl2-(blue) (purple) Scheme 6 The first reduction in the cobalt polymer is assigned to a metal- centred redox process, resulting in the appearance of a new MLCT transition, with the second reduction being ligand- centred.In the case of the nickel polymer, both redox processes are ligand-based. A further alternative to the vacuum deposition method is the Langmuir-Blodgett (LB) technique, which is suitable for both monolayer and multilayer film formation.32 The electro-chemical properties of a variety of substituted and unsubstituted phthalocyanine metal complexes as multilayer LB films have been studied; the first paper on this subject reported the electrochemical study of alkoxy-substituted [Lu(Pc)2].33 LB films exhibited a one-electron reversible reduction and a one- electron reversible oxidation corresponding to a transition from green to orange and blue forms, respectively, with the electron- transport through the multilayers being at least in part diffusion controlled.An explanation of the relatively facile redox reaction in such multilayers is that the Pc ring is large compared with the alkyl tail projected area, with enough space and channels present in the LB films to allow ion transport. Recently, high- quality LB films of MIr tetrakis[(3,3-dimethyl- 1 -butoxy)carbo- nyllphthalocyanine (M = Cu, Ni) have been reported.34 Ellipsometric and polarised optical absorption measurements suggest that the Pc molecules are orientated with their large faces perpendicular to the dipping direction and to the substrate plane.That the LB technique is amenable to the fabrication of ECDs is supported by the recent report of a new thin-film display based on LB films of the praseodymium bisphthalocya- nine complex.35 The electrochromic electrode in the display was fabricped by deposition of multilayers (10-20 layers, ca. 100-200 A) of praseodymium bisphthalocyanine onto ITO- coated glass (7 X 4 cm2) slides. The display exhibits blue- green-yellow-red polyelectrochromicity when a potential rang- ing from -2 to +2 V is applied. After 105 cycles no significant changes are observed in the spectra of these colour states. The high stability of the device was ascribed to the preparaton of well-ordered monolayers (by the LB technique) which seem to allow better diffusion of the counter ions into the film and improve the reversibility and stability of the system.8 Concluding remarks It has been shown in this article that a large number of electrochromic materials are available from all branches of synthetic chemistry. The ultimate usefulness of these fascinat- ing materials hinges on a detailed understanding of their fundamental redox operation and the accompanying physico- chemical-structural changes. Indeed, even in the case of tungsten trioxide the colouration mechanism is still under debate with Deb and co-workers36 recently proposing a new theory based on the intervalence transition between Wv and WIV states instead of the Wv and WVI states as suggested by previous theories.The emphasis on applications of electrochromic materials has in recent years shifted from small-scale display devices to large- scale transmissive and reflective devices. Clearly, the commer- cial opportunities in the development of electrochromic systems are enormous, with electrochromic glazings for cars and buildings being forecast for the near future (three and five years, respectively).3 Widespread application of ECDs, particularly for architectural applications, depends on reducing cost, increasing device lifetime and circumventing the problem of ECD degradation. Of further concern is that existing ‘smart window’ designs require an external power source, making the retro-fitting of existing buildings difficult and expensive.Photoelectrochromic systems, which change colour electro- chemically but only on being illuminated, seem to be the way around this problem. Thus, an important recent development in this area is a thin-layer cell where a light-absorbing layer of RuIrL2L’ (L = 2,2’-bipyridine-4,4’-dicarboxylate,L’ = 2,2’-bipyridine) adsorbed onto 4 pm thick nanocrystalline Ti02 film on IT0 glass is the working electrode in a thin-layer cell.37 Absorption of solar radiation produces a photovoltage sufficient to colour a layer of tungsten trioxide deposited on the counter electrode. The successful fabrication of cells of dimensions 1-25 cm2 suggests that this might be a suitable approach to the development of photoelectrochromic devices for large-area window applications.The acceptable lifetime of a retrofit, self- powered, photoelectrochromic window might be shorter com- pared to a conventional electrochromic window, because the former can easily be replaced. References 1 P. M. S. Monk, R. J. Mortimer and D. R. Rosseinsky, Electrochromism: Fundamentals and Applications, VCH, Weinheim, 1995. 2 H. Byker, in Electrochromic Materials 11, ed. K.-C. Ho and D. A. MacArthur, PV 94-2, pp. 3-13, Electrochem. SOC. Proc. Ser., Penning- ton, New Jersey, 1994. 3 M. Green, Chem. Ind., 1996, 17, 641. 4 C. G. Granqvist, Handbook of Inorganic Electrochromic Materials, Elsevier, Amsterdam, 1995. 5 R.A. Batchelor, M. S. Burdis and J. R. Siddle, J. Electrochem. Soc., 1996, 143, 1050. 6 A. G. Sharpe, The Chemistry of Cyano Complexes of the Transition Metals, Academic Press, New York, 1976. 7 V. D. Neff, J. Electrochem. Soc., 1978, 125, 886. 8 R.J. Mortimer and D. R. Rosseinsky, J. Electroanal. Chem., 1983,151, 133. 9 R. J. Mortimer and D. R. Rosseinsky, J. Chem. Soc., Dalton Trans., 1984, 2059. 10 A. Hamnett, S. Higgins, R. J. Mortimer and D. R. Rosseinsky, J. Electroanal. Chem., 1988, 255, 315. 11 K. Itaya, K. Shibayama, H. Akahoshi and S. Toshima, J. Appl. Phys., 1982, 53, 804. 12 K. Honda, J. Ochiai and H. Hayashi, J. Chem. Soc., Chem. Commun., 1986, 168. 13 M. K. Carpenter and R. S. Conell, J. Electrochem. Soc., 1990, 137, 2464.14 K.-C. Ho, T. G. Rukavina and C. B. Greenberg, in Electrochromic Materials ZI, ed. K.-C. Ho and D. A. MacArthur, PV 94-2, pp. 252, Electrochem. SOC. Proc. Ser., Pennington, New Jersey, 1994. 15 K. Itaya, I. Uchida and V. D. Neff, Acc. Chem. Res., 1986, 19, 162. 16 C. L. Bird and A. T. Kuhn, Chem. Soc. Rev., 1981, 10,49. 17 R. J. Mortimer and J. L. Dillingham, in Electrochromic Materials 111, ed. K.-C. Ho, C. B. Greenberg and D. M. MacArthur, PV 96-24, pp. 3-13, Electrochem. SOC. Proc. Ser., Pennington, New Jersey, 1997. 18 C. J. Schoot, J. J. Ponjee, H. T. van Dam, R. A. van Doom and P. J. Bolwijn, Appl. Phys. Lett., 1973, 23, 64. 19 G. P. Evans, in Advances in Electrochemical Science and Engineering, ed. H. Gerischer and C. W. Tobias, vol.1, VCH, Weinheim, 1990, pp. 1-74. Chemical Society Reviews, 1997, volume 26 155 20 F. Gamier, G. Tourillon, M. Gazard and J. C. Dubois, J. Electroanal. Chem., 1983,148,299. 21 D. J. Guerrero, X. M. Ren and J. P. Ferraris, Chem. Mater., 1994, 6, 1437. 22 J. P. Ferraris, C. Henderson, D. Torres and D. Meeker, Synth. Metals, 1995, 72, 147. 23 F. Rourke and J. A. Crayston, J. Chem. SOC., Faraday Trans., 1993,89, 295. 24 R. J. Mortimer, J. Muter. Chem., 1995, 5, 969. 25 G.-W. Jang, C. C. Chen, R. W. Gumbs, Y. Wei and J.-M. Yeh, J. Electrochem. Soc., 1996, 143, 2591. 26 R. J. Mortimer, in Research in Chemical Kinetics, ed. R. G. Compton and G. Hancock, vol. 2, Elsevier, Amsterdam, 1994, pp. 261-311. 27 P. D. Beer, 0.Kocian, R. J. Mortimer and C. Ridgway, J. Chem. SOC., Faraday Trans., 1993, 89, 333. 28 R. M. Leasure, W. Ou, R. W. Linton and T. J. Meyer, Chem. Muter., 1996, 8, 264. 29 G. C, S. Collins and D. J. Schiffrin, J. Electrochem. Soc., 1985, 132, 1835. 30 D. J. Moore and T. F. Guarr, J.Electroanal. Chem., 1991,314, 313. 31 H. F. Li and T. F. Guarr, J. Electroanal. Chem., 1991,297, 169. 32 L. M. Goldenberg, J. Electroanal. Chem., 1994, 379, 3. 33 S. Besbes, V. Plichon, J. Simon and J. Vaxiviere, J.Electroanal. Chem., 1987, 237, 61. 34 C. Granito, L. M. Goldenberg, M. R. Bryce, A. P. Monkman, L. Troisi, L. Pasimeni and M. C. Petty, Langmuir, 1996, 12, 472. 35 M. L. Rodriguez-MCndez, J. Souto, J. A. de Saja and R. Aroca, J.Muter. Chem., 1995,5,639. 36 J.-G. Zhang, D. K. Benson, C. E. Tracy, S. K. Deb, A. W. Czanderna and C. Bechinger, in Electrochromic Materials 111, ed. K.-C. Ho, C. B. Greenberg and D. M. MacArthur, PV 96-24,pp. 25 1-259, Electrochem. SOC. Proc. Ser., Pennington, New Jersey, 1997. 37 C. Bechinger, S. Ferrer, A. Zaban, J. Sprague and B. A. Gregg, Nature, 1996,383,608. Received, 24th December 1996 Accepted, 5th March 1997 156 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600147

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

James Utley Department of Chemistry, Queen Mary and Wesfleld College, University of London, Mile End Road, London, UK El 4NS Dedicated to Professor Dr Hans J. Schafer on the occasion of his 60th birthday. Electrochemistry is a clean and convenient method for the generation on a preparative scale of many reactive interme- diates (radical-ions, radicals, carbanions, carbocations, quinodimethanes). This forms the basis of organic electro- synthesis and conditions have been devised for selective and useful chemical conversions. Particularly useful and versa- tile are carbon-carbon bond forming reactions such as those involving the Kolbe reaction, electrohydrodimerisationsand N-acyliminium cations. The in situ electrogeneration and regeneration of redox reagents in non-stoichiometric amounts, and the related redox catalysis, is proving to be an attractive, indirect, electrosynthetic approach.And mecha- nistic insight into useful preparative reactions is now at an advanced stage. 1 Introduction Electrochemical conversions of organic compounds have much in common with photochemical conversions. Electrochemical activation to produce reactive intermediates is, like photo- chemical activation, non-thermal. Furthermore, organic com- pounds which are candidates for electrochemical conversion must have electroactive functionality (an electrophore) which is the analogue of the chromophores necessary for photochemical activation. Molecules are activated by the addition or removal of electrons at an electrode.This must involve addition to a LUMO (reduction at a cathode) and removal from a HOMO (oxidation at an anode); it is relatively easy qualitatively to identify which compounds are likely to be reducible or oxidisable. In practice, the limits coincide with the most powerfully reducing and oxidising conditions available by conventional chemistry [ca. +3 and -3V (vs. SCE)]. Benzene can thus be oxidised to benzoquinone and reduced to 1,4-dihy- Jim Utley is Professor of Organic Chemistry at Queen Mary and Westfield College (University of London). His early career was in physical-organic chemistry [with N. B. Chapman, J. Shorter (Hull), B. M. Wepster (Delft) and J. H. Ridd (UCL)] but, under the influence of the natural products chemist B. C.L. Weedon (QMC) and later L. Eberson (Lund), he turned seriously to synthetic, rnecha- nistic and stereochemical as-pects of organic electrochem- istry in ca. 1970 and was awarded the DSc of London University in 1980. He has worked on most aspects of the subject, often in collaboration with major groups in Europe and the USA, and has been a guest Professor at the Uni-versities of Aarhus, Miinster, Texas (at Austin) and the Ecole Normale Supkrieure (Paris). drobenzene. Many relevant books and reviews are available’ which cover these and other basic aspects. The most important reactive intermediates formed by electrochemical reduction and oxidation are radical-ions, radi- cals, carbocations and carbanions; radical-ions are usually the first-formed species but fragmentation can lead typically to radicals which may be further reduced or oxidised (Fig.1). MX mU Fig. 1 Common species formed electrochemically Electron transfer brings about umpolung, electron-rich com- pounds are oxidised to electron-deficient species (electrophiles) and reduction of electron-poor compounds gives nucleophiles. The chemical reactivity of these species is as expected; radical- anions may be nucleophilic and basic, radical-cations may be electrophilic, and both may react as radicals, e.g. to give coupling reactions. Much is now understood about the structural features which determine selectivity between these potentially competing pathways and it is the purpose of this review to demonstrate this understanding rather than to give a compre- hensive survey of all types of useful electrosynthetic reaction.Here, we concentrate on recent examples although they are mostly based on electrosynthetic reactions which have become well-established during the past 30 years or more. Many advances in electrosynthesis have depended on advances made in unravelling mechanisms of electroorganic reactions. Furthermore, the techniques which have been used to study mechanism, mainly electroanalytical, have also been powerful in the characterisation of short-lived intermediates and in the study of such fundamental processes as proton transfer and single electron transfer (SET). The study of electroorganic chemistry has had a considerable impact on organic and physical-organic chemistry in general.Electrochemical conversions are clean. Either no reagent is employed, other than the electron, or electrochemistry is deployed to generate and regenerate chemical reagents in sealed systems, in non-stoichiometric amounts and possibly in benign solvents. An example is the production of adiponitrile by cathodic hydrodimerisation of acrylonitrile (the Monsanto process) which is operated by BASF in the UK at ca. 90000 tonnes p.a. and essentially in aqueous conditions. Other ‘clean technology’ aspects will become apparent. Chemical Society Reviews, 1997, volume 26 157 2 Practical matters Electrochemical conversions may be carried out directly at an electrode (direct electrolysis) or by using the electrode reaction to electrogenerate a chemical reagent (indirect electrolysis) or an electron transfer reagent (mediated electrolysis).In each of these cases the reactors are electrochemical cells which come in many designs and sizes; they are usually simple to construct and are well-described.l.2 They may have cathode and anode compartments separated by membranes or microporous materi- als such as sintered glass (divided cells), or be single compartment (undivided cells). The electrode materials are usually chosen to be inert at the reducing or oxidising conditions being used and the potential at the working electrode may be controlled with respect to a reference electrode (potentiostatic electrolysis).Alternatively, the reactions may be run at controlled (constant) current (galvanostatic electrolysis). In most cases, for an electrosynthetic conversion to compete with a chemical alternative the reaction conditions should be developed to run galvanostatically in undivided cells. The reactions are also usually run in solutions containing ionic salts (supporting electrolytes) which must be inert at the reducing or oxidising potentials involved. Their role, in combination with the polar solvents required, is to carry the main part of the current passing between the electrodes; the (usually) neutral organic substrates are transported by convection (stirring or flowing) or diffusion. 3 Common electrochemical conversions 3.1 Carbon-carbon bond formation Kolbe electrolysis and electrohydrodimerisation are well- established electrosynthetic processes which operate under relatively simple conditions and which give one-step carbon- carbon bond formation in a manner difficult to match by other routes (Fig.2). These reactions, and to a lesser extent the anodic Kolbe Reaction Electroh ydrodimerisation I b Anodic Coupling I b Nu Fig. 2 Coupling of electrogenerated radicals and radical-ions coupling of alkenes, have been prominent in recent electro- synthetic developments. They have been cleverly exploited with regard to interesting target compounds, stereoselectivity and the development of intramolecular variants, e.g. for cyclisations. The carbon radicals produced by anodic oxidation of carboxylate anions are known to be 'free' but, probably because of the high local concentrations formed at the electrode, coupling is the dominant reaction.Similar radicals produced homogeneously do not couple with useful yields and there is only one example3 of a metal-based oxidant (OsCl6-) which gives comparable coupling. Carbon-centred radicals generated by Kolbe electrolysis are highly reactive; different radicals may be generated by co-electrolysis and they combine randomly (the Brown-Walker reaction). They will also add to double bonds either inter- or intra-molecularly. These aspects, applied to useful synthetic conversions, are displayed in Fig. 3. The examples involving the construction of partially fluori- nated compounds4 and C-glycosidess, compounds with inter- esting rheological or liquid crystal properties, are derivative of Weedon's pioneering work6 with the cross-coupling Kolbe reaction [Fig.3(a)]. Many fatty acids were thus synthesised and their stereochemistry established, including absolute configura- tion. Kolbe oxidation is also a convenient method for generating the trifluoromethyl radical7 which may substitute into aromatics or add to double bonds [Fig. 3(b)]. Addition of anodically generated radicals is clearly very fast; the hex-5-enyl radical is known to cyclise with k = 2.9 X 104 s-1 which compares with a value of lo7s-l estimated in a theoretical treatment* based on the ratios of the products formed competitively in tandem cyclisations such as that shown.Anodic generation of carbon radicals will initiate these radical tandem cyclisations in a practically convenient one-pot reaction and with good regio- and stereo-selectivity, as demonstrated in one of several examples [Fig. 3(d)]. Electrohydrodimerisation (EHD) is a common outcome of the cathodic reduction of alkenes activated by electron-withdrawing groups (see Fig. 2). Numerous examples are known1 but here we will concentrate on just two recent reports which illustrate the versatility and scope of the reaction (Fig. 4). It depends on the rapid combination of delocalised anion- radicals. They must be formed under conditions which minimise competing protonation, i.e. hydrogenation of the double bond. Thus, alkenes activated by a variety of electron- withdrawing groups (commonly CO2R, COR, CN, N02, C=NR) will undergo EHD.The ele~trohydrodimerisation9~of dimethyl maleate [Fig. 4(a)] nicely illustrates the simplicity of conditions (constant current, graphite electrodes, MeOH-NaOAc electro- lyte) which may be developed for a reaction which in the laboratory was run on a 100 g scale and later developed to pilot scale. In this case dimerisation of the radical-anions is faster than protonation, even in methanol solution, which is a consequence of the double activation by two C02Me groups which greatly delocalise the negative charge effectively, thus reducing the basicity of the radical-anion and enhancing its radical character. This is an important concept in designing EHD reactions.For the second example given [Fig. 4(b)], the EHD of a series of cinnamate esters, the radical-anion dimerises fastergb the more the charge is delocalised (resulting in a less negative EO). A linear relationship between EO and log k2 is found. The EHD of cinnamate esters proceeds with subsequent in situ Dieckmann condensation to give 3,4-diarylcyclopentan- ones with high stereoselectivity. The all-trans isomer is the only detectable product. Furthermore, chiral esters reducegc with a diastereoselectivity which can reach 95%. Study of this reaction has taught us much about the mechanism of EHD and it will be returned to (section 5). 3.2 Functionalisation In a seminal review10 of the prospects for electrosynthesis in industry, C-C bond formation and functionalisation of alkenes and aromatics were highlighted as likely to be significant.The most important examples of electrochemical functionalisation involve anodic oxidation, probably because hydrocarbon radi- cal-cations often readily deprotonate (X = H in Fig. 1) which paves the way for subsequent functionalisation. Some examples are given in Fig. 5. The 'industrial' use is illustrated by the efficient conversion of substituted toluenes into the correspond- ing benzaldehydes [Fig. 5(a)]. In addition to the example given11 it is understood that several related processes are used by BASF to produce fine chemicals for flavourings and perfumes. 158 Chemical Society Reviews, 1997, volume 26 The other examples12 are recent adaptations of the reaction discovered by Ross-Eberson-Nyberg, and much exploited by Tatsuya Shono, whereby amides are oxidised to N-acyliminium cations [e.g.1 in Fig. 5(b)]. No other route is so direct and this must be the method of choice for preparation of these versatile intermediates. Because they are usually formed in methanol solution it is the a-methoxyamides which are isolated. These are easily converted by treatment with acid into the N-acyliminium B cations which can then undergo reaction with a variety of nucleophiles. A short routel2a to the Geissman-Waiss lactone [2, Fig. 5(b)] and its enantiomer is based on electrochemical formation of an N-acyliminium cation (1) which is functionalised under stereochemical control.More ambitious are those recently reported syntheses lzb which involve intramolecular trapping of N-acyliminium cations; the example given [Fig. 5(c)] is a one--2e, -C02 (4 * (G)-(CH~)~CHS(70%) 6-Deoxy-l,2:3,4-di-Oisopropylidine-6-C-octyl-a-D-galactopyranose -e, -co2 ____) -cop- Fig. 3 The Kolbe reaction revisited dimerisation cC02Me e * hydrolysis ___)___t /CH3’ from CH3C02-J co-electrolysis 42%, isomer ratio ca. 3:l H02C\C02Me H02C Ar , Ar, Fig. 4 Electrohydrodimerisation Chemical Society Reviews, 1997, volume 26 159 pot procedure. This approach has been exploited in several other similar syntheses. 3.3 Other cyclisation reactions The intramolecular reactions of electrogenerated radical-ions, radicals, cations or carbanions are well-established and have featured recently in useful preparative procedures.Three fairly typical examples are shown in Fig. 6. The first [Fig. 6(a)] comes13 from the work of Little et al; it serves to emphasise an important point concerning mechanism. The relevant reaction centres should not only be proximate but also predictions concerning successful cyclisation demand a mechanistic hy- pothesis and knowledge of Baldwin's rules! In this case there is a clear difference in reducibility of the two reactive groups and the first electron transfer is to the activated alkene function. But from that point there are several possibilities: (i) the first- formed radical-anion could add to carbonyl as a radical or nucleophile; (ii) protonation could give a radical as the attacking species; or (iii)further electron transfer could give a carbanion as the key nucleophile.Little et a1have summarisedl3 the electrochemical and chemical arguments and their preferred mechanism is given in Fig. 6. However, each case must be considered on its merits! The coupling involving anodic oxidation of enol ethers [Fig. 6(b)] has two identical electroactive centres and the coupling is likely to be either via two radical-cations combining (radical-ion/ radical-ion coupling) or one radical-cation centre attacking a neighbouring enol ether function (radical-ionhubstrate cou-pling). All that we know about analogous cathodic couplings, involving radical-anions, suggests that radical-ion/substrate routes are rare and radical-ion/radical-ion routes common (see section 5).But the final example [Fig. 6(c)] is best explained in 4-H+OBu' OH OBut -H+ C02Me (R = TBDMS) 1 terms of electrophilic (or radical) attack by the radical-cation from the enol ether adding to the furan ring. There are many similar ring closure reactions old and new,14 typically involving closure onto electron-rich aromatics, e.g. phenols and phenol ethers. 3.4 Polymerisation via electrogenerated intermediates Anodic oxidation of electron-rich aromatics produces radical- cations in a sea of the nucleophilic starting materials and electrophilic addition can induce stepwise polymerisation.The mechanism is similar to that of acid-catalysed polymerisation and early examplesl5 of such anodic polymerisation were the formation of polypyrroles, polythiophenes and polyanilines [Fig. 7(a) and (b). In these cases the polymers are formed as a film on a solid anode, often platinum, and they may be further oxidised (doped) to produce conducting films; the charge carriers are radical-cation centres (polarons). Although soluble polymers may be made anodically, by appropriate substitution with lipophilic groups, they are mostly obtained as intractable films. In contrast, the cathodic formation of some polymers can be carried out at a mercury cathode and stirring prevents film formation. This approach has led to a particularly versatile method for the cathodic preparation of poly-p-xylylenes (PPXs)16 and poly-@-phenylenevinylenes) (PPVs)17-see Fig.7(c). The key intermediates, as highlighted, are quinodimethanes. The parent p-quinodimethane (xylylene) polymerises rapidly when formed cathodically, although not so rapidly as to preclude its detection by fast sweep cyclic voltammetry. Combination of the quinodimethanes (QDMs) is unselective and an unsymmetrical QDM gave the statistical ratio of possible linkages [ref. 17 and Fig. 7(d)]. This means that cross-coupling of different QDMs, produced by co-electrolysis CH20Me CH(OMe), CHO$,repeat - Qacid- hydrolysis OBut OBu' OH CO2Me 602Me oxidation,I separation hydrolysis t ?J Fig. 5 Anodically generated cations 160 Chemical Society Reviews, 1997, volume 26 of different bis-(halomethyl)arenes, leads to random co-polymers, cf.the random coupling in the Kolbe reaction. Furthermore, a considerable range of functionality is tolerated in the cathodic method and by careful choice of functionality and electrolysis conditions it is possible electrochemically to produce PPXs and PPVs which are soluble in organic solvents. This also allows their formation at solid cathodes. 4 The indirect approach The examples dealt with so far involve the reactions of intermediates produced directly at the electrode by hete-rogeneous electron transfer. Disadvantages of this approach can include the inhibition of the electrode reaction by formation of non-conducting films (not all polymer formation is as welcome as that of pyrrole etc.) and for a given conversion the electrode overpotential required may be significantly greater than that expected from theory. This leads to lack of selectivity and unwanted competition from side reactions such as oxidation or reduction of solvent.Much attention has been paid in recent years to electrosynthetic conversions which are indirect, i.e. the electrode reaction is used to generate and regenerate a reagent. Typically, these are generated at potentials which are less cathodic or anodic than those required for the corresponding direct electron transfer. The reagents may be formed in situ (at less than stoichiometric concentration) or ex situ. A special situation arises when the electrogenerated reagent is an electron-transfer reagent, often called a mediator.In many cases the mediator is formed reversibly at the electrode and, after diffusion into the bulk solution, is involved in electron transfer with the target substrate. This may be in an unfavourable equilibrium which is driven forward by a rapid and chemically irreversible follow-up reaction. This situation is illustrated for reductive cleavage in Fig. 8. The practical consequence is that the target substrate is converted at a potential up to ca. 0.5 V less than indicated by its formal potential. And because the mediator is constantly and rapidly regenerated the term redox catalysis is often applied (see later).4.1 Electrogenerated reagents In the simplest applications of this approach quite common reagents are generated electrochemically, normally in situ. Chlorine can conveniently be introduced into a reaction mixture by anodic oxidation of chloride; thus methyl hypochloritel8 is the effective reagent in the practically convenient conversion of 3 into 4 by anodic oxidation of a mixture of starting material and sodium chloride in methanol solution. Similarly, the anodic generation of bromine in NaOAc-HOAc is superior to direct electrolysis for the allylic acetoxylationlg of (3-ionone to give 5. Direct anodic conversion of the less easily oxidised a-ionone in NaOAc-HOAc gives many products but clean allylic acetox- ylation to 6 is achieved by the in situ generation19 of CoIII acetate from the CoII salt which is present at ca.10 mol%. Cathodic reduction is probably the method of choice for preparation20a of solutions of CrII and this method has recently been used20b to induce free radical cyclisations, e.g. in the high yielding conversion of 7 into 8. The examples described above, and collected in Fig. 9, are merely recent representatives of many successful applications of this approach. The success of this approach is not surprising considering that the manufacture of most highly oxidising and reducing species is electrolytic (Li, Na, Al, F2, S2OS2-, CeIV, CrvI and many more)! A process21 for the large-scale production of naphthaquinone, as an intermediate in a route to anthraquinone, depends on the generation and regeneration of non-stoichiometric amounts of CeIV.Rather less common electrogenerated reagents include the hypervalent iodobenzene difluorides, e.g. 4-02NC6HJF2, pre- pared by anodic oxidation22 of the corresponding iodobenzene I OP CH20P 90% isolated yield Med I Fig. 6 Other intramolecular cyclisation Chemical Society Reviews, 1997, volume 26 161 in the presence of Et3N.3HF. This reagent has been used'22 ex situ, for the conversion of 9 into 10. And a convenient and inexpensive method for the preparation of Ruppert 's reagent (CF3SiMe3) has been developed.23 This reagent is commonly used to introduce the trifluoromethyl group via a number of addition or coupling routes. Ruppert's reagent may now be prepared in high yields by simple constant current reduction of CF3Br at a nickel cathode in the presence of Me3SiC1.The correct choice of anode material is crucial to the success of this reduction! A sacrificial A1 anode is used-i.e. one which dissolves to give Al3+ in step with the cathodic reaction. Perhaps the most significant recent development concerning the electrogeneration of reagents at low or catalytic concentra- tions is an electrochemical variantz4 of the Sharpless asym- metric dihydroxylation of alkenes using OsvIII in the presence of a chiral ligand. This has been developed by re~earchers~~a at Engelhard Ltd to the point where it could be used on a commercial scale. Double mediation is used in this reaction [Fig.10(a)]. The original Sharpless method uses stoichiometric amounts of ferricyanide to oxidise Osvl to maintain the required concentration of OsVIII; in the electrochemical version a lower concentration of ferricyanide is used and its level is maintained by anodic oxidation. Thus the ferricyanide acts as the second mediator causing the OsVIII concentration to be maintained electrochemically but indirectly. The point of this is that the process is run in a two-phase system (cyclohexane-water) and the OsvIII/chiral ligand complex reacts with the alkene in the organic phase but can be regenerated with ferricyanide at the interface. An alternative to ferricyanide is iodine in a mixture in which iodide is recycled an0dically2~h to iodine.The Mn"1-mediated electrosynthesis of sorbic acid precur- sors is another, exceptionally well-described,*5 process which was developed to operate in a flow cell on a multi-kilogram scale [Fig. 10(b)]. The reaction is catalytic in Mn" and its success depends on the additional role of CuI1 in promoting the oxidation of an allylic radical to the corresponding cation. X=NH,S -polypyrroles, polythiophenes [oxidise further to conducting forms] H s SflH polyanilines [oxidise further to conducting forms; pH dependent] xYy Y*X YQX a quinodimethane X = Br Y = Br or H q-=y-&$"\/ L(=ych$n\/ R R PPVS PPXS Br\ a,a 1 Fig. 7 Polymers from electrogenerated intermediates 162 Chemical Society Reviews, 1997, volume 26 Mediator IMediator 14 L = hydroquinidine OH 4-chlorobenzoate 90% yield, >90% ee RX+*-Rx[ Mediator ] + Mediator fastRx*--R' + x-And: Eo(1) positive to E0(2)where E0(2) refers to: Fig.8 Mediated electrolysis (redox catalysis) 4.2 Redox catalysis A scheme for the redox-catalysed cleavage of an organic halide has been set out in Fig. 8. It is the clearest case, in which the mediator acts as a single electron transfer (SET) reagent and the follow-up reaction, for alkyl halides at least, is very fast, perhaps even concerted with electron transfer. It is the coupled chemistry which allows direct or indirect reactions, cathodic in these cases, to proceed at more positive potentials. Some recent, preparatively useful, examples are given in Fig.11. Each of these may in some sense be described as redox catalysis but where catalytic amounts of transition metal complexes are the mediators it is not always clear whether SET is involved (outer- sphere electron transfer) or whether intermediate complexes are formed (inner-sphere electron transfer). ortho-Quinodimethanes are much used as synthons in cycloaddition reactions; they are produced by a variety of routes and several of these involve 1,4-elimination. Recently, it was discovered26 that co-electrolysis of 1,2-bis(bromo-methylarenes) and dienophiles (maleic anhydride or quinone based) gave the Diels-Alder adduct expected for reaction involving the o-QDM derived from the dihalide [Fig. ll(a)].3 4 AcO OAc 5 [Brz, NaOAc/HOAc] 6 [Co"', NaOAc/HOAc] 7 a n 9 10 Fig. 9 Conversions with electrogenerated reagents [OS*+/L] [OSC+/L] / [Fe(CN)63-] (maintained by anodic oxidation) ,o Mn3+ Mn2+ p , \i *H&q VO -%O A C 0 \\ 0 0 AcO--CO*H 4 hydrolysis efc,sorbic acid 0 Fig. 10 Electrochemical 'Sharpless' dihydroxylation and radical addition route to sorbic acid Direct reduction of the dihalides is known16 to give o-QDMs and in the absence of a dienophile they polymerise. However, the co-electrolyses are carried out at the reduction potential of the dienophile which is several hundreds of millivolts less negative than that required for direct reduction of the dihalide. Furthermore, cyclic voltammetric experiments (Fig.12) show nicely the characteristic behaviour of redox-catalysed systems. In the absence of dihalide the dienophile/mediator shows chemically reversible reduction with the radical-anion having a significant lifetime. The radical-anion can transfer an electron to the added dihalide which rapidly loses bromide anion (cf. Fig. 8); eventually the o-QDM is formed which can react, probably relatively slowly, with the regenerated mediator [Fig. 1 l(a)]. On the cyclic voltammetric timescale the re-generated mediator is rapidly reduced and it is this process which is responsible for the loss of reversibility and the more than doubling of the peak current (the catalytic current) seen in Fig. 12. The second example27 [Fig. 1 1 (b)] nicely illustrates several issues.It is undoubtedly a preparatively useful electrosynthetic reaction but although it is plausibly portrayed as involving radical cyclisation on to a double bond the mechanism is far from established. The best yields are obtained for closure on to double bonds activated by C02Me. This raises the question of whether it is the nickel catalyst which is reduced or the conversion of the substrate electrophore into radical-anion followed by intramolecular displacement of bromide. No voltammetric data is presented and the choice of reference electrode (Ag/AgCl) makes it difficult to be certain that NiI is involved although there is good precedent for this. Furthermore, the mediator, and electrolyte system (NH4C104, DMF, Et4N- C104) is highly hazardous.The use of perchlorate electrolytes is banned in many electrochemical laboratories and usually tetraalkylammonium tetrafluoroborates are adequate alter-natives. In contrast, much more is known about the nickel complex catalysed reductive cross-coupling28 of alkyl and aryl halides (Fig. 1 1 (c)]. The reactions are carried out at constant current but at potentials (ca. -1.2 V vs. SCE) at which NiO formation occurs. Organic halides are well-known to react with Ni" complexes by oxidative addition. Instances of this have been established29 for the electrochemical formation of poly-(p-phenylene) from 1,4-dibrornobenzene and of poly-p-xylylene from 1,4-bis-(chloromethylbenzene);both processes are catalysed by nickel complexes. The reaction given Chemical Society Reviews, 1997, volume 26 163 [Fig.ll(c)] fits well into this picture; apart from the case carefully balanced. Even so, such modified electrodes lack displayed similar cross-coupling reactions have been demon- long-term stability and are not suitable for larger-scale syn- strated29 between aryl halides and a variety of other organic thesis. halides and allylic compounds. More promising is the application of bulk solution redox mediators combined with the design of special reactors. These 4.3 The SR~~ reaction approaches have been well reviewed,31 and the reviewer (E. One other type of mediated electrosynthetic reaction must be Steckhan) has been responsible for most of the recent mentioned.Under the special reaction conditions pertaining in liquid ammonia (H-atom abstraction precluded) and the use of Cyclic voltammetry of: a magnesium ‘sacrificial’ anode to deliver a stoichiometric amount of Mg2+ cations, it is possible reductively to cleave aryl halides to aryl o-radicals and force in situ reaction with organic nucleophiles. The outcome is a reaction sequence in which the mediator radical-anion is regenerated, effectively by electron 0transfer from the nucleophile. Thus, in principle the reaction is 2.000 1only initiated electrochemically; in practice a small charge is consumed. The reaction scheme, and just one preparative example from many,30 are displayed in Fig. 13. 4.4 Electroenzymatic synthesis The use of redox enzymes as mediators, with the prospect of harnessing their catalytic activity and selectivity, has long been a goal of organic electrochemists.But the practical problems are formidable and a major obstacle is the slowness of electron I I I Itransfer between electrodes and redox proteins. For analytical -0.400 I purposes electrodes modified with other redox functions (e.g. 100.0 -100.0 -300.0 -500.0 -700.0 -900.0 ferrocene) have been developed in these cases the modified EImV electrode promotes indirect electron transfer between the Fig. 12 Characteristic cyclic voltammetric behaviour for redox catalysis electrode and redox protein. The kinetics of the various steps Conditions: Hg bead cathode, DMF-Et4NBr (O.~M), 0.3 V s-l, V vs. Ag/involved (reaction between enzyme and substrate, rate of AgBr: (i) anhydride only, (ii) 0.5 equiv.dibromide, (iii) 1.0 equiv. electron transfer between mediator and cofactor) have to be dibromide, (iv) 2.0 equiv. dibromide 0 ___) reduction at Er.ed(2) ’a::1 aI+ [Dienophile]‘--+ [Dienophile] 2 equiv. needed R’ = H, Ph, Et 16 -88% RZ = H, Me, COzMe cathodic reduction (c) ArX + CICH2COCH3 ArCHZCOCH,C NiBr2bipyX = Br, I Ar = variously substituted 35 -80% phenyl Fig. 11 Some redox catalysed reactions 164 Chemical Society Reviews, 1997, volume 26 significant advances in the field. The flavour of this work is indicated in Fig. 14,which illustrates a model oxidation reaction in which the refreshment of NAD+ is achieved with anodically regenerated ruthenium complex.In this instance, carried out in a phosphate buffer solution, 10 mmol of substrate was converted at a rate of 28 turnovers per hour in the presence of 5 mol% mediator and NADH. The components of such reactions are all in solution which can therefore be carried out in a flow reactor. The key feature of these so-called electrochemical enzyme membrane reactors is a separator equipped with an ultrafiltration membrane; this permits the low molecular mass product to pass out of the reactor but retains the higher molecular mass mediator and redox enzyme. Model oxidations of alcohol to ketone and the reverse reductions have been 1. achieved with good turnover and the high enantioselectivity expected for enzyme reactions.5 Selected mechanistic aspects Modem software-driven electroanalytical equipment allows deceptively precise measurements to be made. Such measure- ments have long been the province of specialist groups, skilled in the art of performing high scan rate voltammetry and related amperometric experiments. Furthermore, the relevant theory is well-described in standard textbooks. Now, commercially available computer simulation software apparently allows one to measure, say, a series of cyclic voltammograms at various sweep speeds and by fitting redox peak shifts and peak shapes I2. kx + [Mediator]-' [Mediator] + Ar' + X-I 3. Ar' + Nu-ArNu" -4. NU-' +--[Mediator] &Nu + [Mediator]-' -----: e-g 'N/ + [ M"]F3C+CI CN [Mediator] 50% Fig.13The SR" reaction 0Ru" complex N N HLADH --horse liver alcohol dehydrogenase Fig. 14 Indirect electro-regeneration of NAD+ Chemical Society Reviews, 1997, volume 26 165 by simulation propose a ‘mechanism’. There are dangers in this approach. However sophisticated the electrochemical work- stations the value and precision of the measurements are usually crucially dependent on the design of cells, the preparation and size of electrodes, and the ‘know-how’ associated with detecting and eliminating disturbing effects, e.g. from adsorp- tion. And data may be collected and processed without appreciation of the assumptions made in deriving the control- ling theory! For example in linear sweep voltammetry (LSV) mechanistic conclusions and associated rate constants can be derived from shifts in measured peak potentials (Ep), for a chemically irreversible system, with scan rate (Y).The slope of the linear plot of logy vs. Ep is different for different mechanistic possibilities. But reliable conclusions can only be drawn if the peak potentials are measured carefully and, crucially, the rate of the irreversible chemical follow-up reaction after initial electron transfer must be quite fast, i.e. the reaction must be completely under kinetic control. To be convincing about mechanistic suggestions it is wise to seek evidence from a variety of methods and determine not only kinetics of follow-up reactions but also reaction orders, coulometry and, where possible, energies of activation.And above all it is essential to characterise all products thoroughly and achieve good material balance. Against this background it is perhaps useful to indicate, very selectively, those types of measurements which give information concerning the mecha- nisms of electrosynthetic reactions and in some cases data of considerable general interest. Because electrohydrodimerisation is such an important electrosynthetic process (cf. section 3.1) much attention has been paid to elucidating the mechanism of key examples. Plausible possibilities are given in Fig. 15. The list is not dim (II e T Fig. 15 Major potential routes for electrohydrodimerisatioil exhaustive but ‘paper’ mechanisms also have to be viewed in the light of general mechanistic common sense. The routes given divide into the radical dimerisation routes (of radical- anion or radical) and the intermediate-substrate routes (addition of radical-anion or carbanion to neutral starting material).It is not easy to distinguish between these possibilities using purely electroanalytical methods; the LSV approach works only for those cases demonstrably under kinetic control (very fast follow-up reaction) and both radical-radical and radical-substrate routes give second-order kinetics. Furthermore, there is always the possibility that the low concentration, non-steady state, regimes used for voltammetric studies give results not relevant to the high concentration, steady state, conditions of preparative electrolysis.The application of a relatively new technique to these issues has been especially helpful. Scanning electrochemical micro- scopy32 is a method whereby a species generated electrochem- ically at an electrode (diameter ca. 60 pm) may be collected at the tip of an ultramicroelectrode (ca. 5 pm) in close but variable proximity. The current measured is a function of the amount of electrogenerated species which survives the journey between the electrodes and the journey time may be varied by varying the interelectrode distance. The theory for these experiments has been establi~hed3~ and application to the electrohydrodimerisa- tion of acrylonitrile, the ‘parent’ EHD reaction, shows33 unambiguously that the reaction is second-order in radical- anion and that the radical-anions dimerise rapidly following initial electron transfer [k2 = (6 f 3) X 107 dm3 mol-1 s-11.The speed of this reaction makes it difficult to maintain one argument used against the radical-anion/radical-anionroutes, i.e. that coulombic repulsion would be prohibitively severe. The same conclusion has been reachedgh for the much slower EHD of esters of cinnamic acid; in DMF solution the radical- anion of methyl cinnamate undergoes second-order reaction at k2 = (7.8 f 0.7) X lo2 dm3 mol-l s-’. Valid LSV experiments give values for the slopes of logy vs.E, plots consistent with the radical-anion/radical-anionroute. This analysis is supportedgh by measurements on the reduction of the trans-cyclohexane- 1,2-diol diester of cinnamic acid which has adjacent activated alkene groups.Cyclic voltammetry shows the two electron transfers to be sequential and for intramolecular reaction both units need to be converted into radical-anions. At the radical- anion/substrate stage no intramolecular reaction is observed at modest sweep speeds despite the favourable disposition of the potentially reactive centres. The question now arises as to whether, in EHD, any clear example of radical-substrate has ever been established. Intramolecular reductive cyclisation has recently been ana- ly~ed~~in detail with emphasis on the conversion of o-keto- a$-unsaturated esters (Fig. 16). The results of LSV measure- IIIF Me02C =L homogeneous ET Fig.16 Reductive cyclisation of o-keto-&$-unsaturated esters ments and chemical arguments were used to reduce the plausible mechanistic possibilities to a handful and the product from the ‘designer’ reaction shown in Fig. 16 strongly indicated the key intermediate to be the carbanion. It was confidently assumed that the intermediate hex-5-enyl radical would have cyclised rapidly to an alternative product had it not been rapidly reduced, presumably by the first-formed radical-anion. It is rare that second electron transfers are heterogeneous (at the electrode) and in most cases initially-formed intermediates diffuse away from the electrode and react in bulk solution; in this case the slow step (protonation) is believed to occur in bulk solution.In this rep0rt3~ the authors recommend adoption of a common mechanistic language based on lower-case letters for elementary steps with upper-case reserved for rate determining 166 Chemical Society Reviews, 1997, volume 26 steps Thus, in the Little-Fry description the reaction in Fig 16 would be an e-P-d-c-p reaction (e = ET at the electrode, d = ET in solution, c = cyclisation, P/p = protonation) Time will tell Finally, an important spin-off of much earlier synthetic work which engendered a need to know more about the redox potentials (EO) of electrogenerated radicals has led to a relatively simple and powerful method for making such measurements This is now well-reviewed 35 The method is based on measuring relative amounts of products from the competing reactions following cathodic generation of aromatic radical-anions (with a known EO) in the presence of alkyl halides (Fig 17) Depending on the reduction potential of the I! E"(A) 1A+ e A-2 A.-+ RX-A+R + X--AR--products 13 A'-+ R A-+ E0(R)Objective R Re R CH, CH,CH, PhCH, PhZCH CH$O PhCO -EO(R) 119 164 167 140 107 170 114 [vs SCE] Fig.17 Measurement of standard potentials for electrogenerated radicals alkyl radical (R) it will either be reduced by the aromatic radical-anion (k4)or couple with it (k3) There is much evidence to support the necessary assumption that the coupling IS fast, ca 109 dm3 mol-1 In these circumstances the measurable product ratios give a 'competition parameter', q, defined as q = k4/(k4+ k3) Plots of q vs the EO values for aromatic radical-anion formation are S-shaped and it can be shown that the mid-point of such curves gives a reduction potential of the radical from which the corresponding EO value can be calculated Thus, the standard potentials for many carbon-centred radicals have been measured, examples, some of which are given in Fig 17, include aliphatic, allyl, benzyl and acyl radicals 6 Dedication This review is dedicated to Professor Dr Hans J Schafer (University of Munster), a pioneer in the field, on the occasion of his 60th birthday 7 References 1 J Utley, Chem Znd ,1994,215,Organic Electrochemistry, ed H Lund and M M Baizer, 3rd edn ,Marcel Dekker, New York, 1991,Synthetic Organic Electrochemistry, A J Fry, John Wiley, New York, 1989 2 F Walsh and D Robinson, Chem Tech Eur , 1995,2, 16 3 L Eberson and M Nilsson, J Chem SOC Chem Commun, 1992, 191 4 H J Schafer, A Weiper, M aus dem Kahmen and A Matzeit, Perspektiven fur die Chemie, VCH, 1993, 97 5 M karenbrock, A Matzeit and H J Schafer, Liebigs Ann , 1996, 55 6 B C L Weedon, Quart Rev, 1952,6,380,B C L Weedon, Adv Org Chem , 1960,l 7 K Uneyama, Tetrahedron, 1991,47, 555, and earlier refs 8 A Matzeit, H J Schafer and C Amatore, Syntheszs, 1995, 1432 9 (a)E A Casanova,M C Dutton, D J Kalotaand J H Wagenknecht, J Elecfrochem SOC, 1993, 140, 2565, (b) I Fussing, M Gullu, 0 Hammerich, A Hussain, M F Nielsen and J H P Utley, J Chem SOC Perkin Trans 2 1996, 649, (c) J H P Utley, M Gullu and M Motevalli, J Chem SOC Perkin Trans 1 1995, 1961 10 D Degner, Top Curr Chem , 1988,148, 1 11 M Barl, D Degner, H Siege1 and W Hoffmann, DE 2 935 398 (BASF) [Chem Abstr , l981,95,24556y] 12 (a)M Thaning and L G Wistrand, J Org Chem , 1990,55, 1406, (b) U Slomczynska, D K Chalmers, F Comille, M L Smythe, D D Beusen, K D Moeller and G R Marshall, J Org Chem , 1996, 61, 1198, and earlier refs 13 R D Little, Chem Rev, 1996, 96, 93 14 N G New, Z Tesfai and K D Moeller, J Org Chem ,1996,61, 1578, and earlier refs 15 Handbook of Conducting Polymers, ed T A Skotheim, Marcel Dekker, New York, 1986 16 J H P Utley, Y Gao, J Gruber and R Lines, J Mater Chem , 1995, 5, 1297 17 J H P Utley, Y Gao, J Gruber, Y Zhang and A Munoz-Escalona, J Mater Chem, 1995,5, 1837 18 A Papadopoulos, J Heyer, K D Ginzel and E Steckhan, Chem Ber , 1989, 122,2159, A Papadopoulos, B Lewall, E Steckhan, K D Gin- zel, F Knoch and M Nieger, Tetrahedron, 1991,47, 563 19 A Guirado, G P Moss and J H P Utley, J Chem SOC Chem Commun , 1987,41 20 (a) D W Sopher and J H P Utley, J Chem SOC Perkin Trans 2, 1984, 1361, (b)C Hackmann and H J Schafer, Tetrahedron, 1993,49, 4559 21 R P Kreh, R M Spotnitz and J T Lundquist, J Org Chem , 1989,54, 1526 22 T Fuchigami and T FUjita, J Org Chem , 1994,59,7190 23 F Aymard, J Y Nedelec and J Penchon, Tetrahedron Lett, 1994,35, 8623 24 (a)A R Amunsden and E N Balko, J Appl Electrochem, 1992,22, 810, (b)S Toni, P Liu, N Bhuvaneswari, C Amatore and A Jutand, J Org Chem , 1996,61,3055 25 J P Coleman, R C Hallcher, D E McMackins, T E Rogers and J H Wagenknecht, Tetrahedron, 1991, 47, 809 26 E Oguntoye (nee Em), S Szunents, J H P Utley and P B Wyatt, Tetrahedron, 1996,52, 7771, E Eru, G E Hawkes, J H P Utley and P B Wyatt, Tetrahedron, 1995, 51, 3033 27 M Ihara, A Katsumata, F Setsu, Y Tokunaga and K Fukumoto, J Org Chem , 1996, 61, 677 28 M Durandetti, J Y Nedelec and J Penchon, J Org Chem , 1996,61, 1748, S Condon-Gueugnot, E Leonel, J Y Nedelec and J Penchon, J Org Chem , 1995, 60,7684 29 C Amatore, F Gaubert, A Jutand and J H P Utley, J Chem SOC Perkin Trans 2, 1996, 2447, and refs therein 30 C Combellas, Y Lu and A J Thiebault, Appl Electrochem ,1993,23, 841, and earlier refs 31 E Steckhan, Top Curr Chem , 1994, 170, 84 32 C Demaille, P R Unwin and A J Bard, J Phys Chem , 1996, 100, 14137,and earlier refs 33 F Zhou and A J Bard, J Am Chem Soc , 1994,116, 393 34 A J Fry, R D Little and J Leonetti, J Org Chem , 1994,59,5017 35 H Lund, K Daasbjerg, D Ochidlini and S U Pedersen, Ele-ektrokhimiya, 1995, 31, 939 Received, 8th January 1997 Accepted, 4th March 1997 Chemical Society Reviews, 1997, volume 26 167

ISSN:0306-0012    

DOI:10.1039/CS9972600157

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Reactions of complex metalloproteins studied by protein-film voltammetry Fraser A. Armstrong, Hendrik A. Heering and Judy Hirst Inorganic Chemistry Laboratory, Oxford University, Oxford, UK OX1 3QR The following review explores applications of voltammetric methods for observing reactions of complex metalloproteins. Attention is focused upon the technique of ‘protein-film voltammetry’, in which the protein molecules under in- vestigation are adsorbed on the electrode surface and electrochemically ‘interrogated.’ The experiments address a minuscule sample with high sensitivity, and optimal control over both potential and time dependence of reactions. Factors governing the voltammetric response are outlined, and particular emphasis is given to the ability to study reactions that are coupled to and may ‘gate’ the primary electron exchange processes.Examples described include proton-transfer and metal-binding reactions of iron-sulfur clusters, coupling of electron transfer in peroxidases, quanti- fying electron-transport pathways in multi-centred enzy- mes, and detection of ‘switches’ that modulate the catalysis as a function of potential. 1 Introduction Dynamic electrochemical methods are an obvious yet largely unexplored approach to studying redox-active sites in proteins. Consequently, we write this Review with the aim of identifying opportunities afforded by techniques such as cyclic voltam- metry for elucidating the complex reactions of biological redox Fraser Armstrong gained his PhD in 1978 at the University of Leeds under the supervision of Professor A.G. Sykes. He then held a Royal Society European Exchange Fellowship with Professor Peter Kroneck at Universitat Konstanz, Germany. After post- doctoral research with Professors R. G. Wilkins (New Mexico State University), Helmut Beinert (Wisconsin) and Allen Hill (Oxford) he took up a Royal Society University Research Fellowship at Oxford in 1983. In 1989 he joined the faculty at the Department of Chemistry, University of California, Irvine. He returned to Oxford in 1993, where he is currently University Lecturer and a Fellow of Fraser A. Armstrong Hendrik A. Heering systems. The approach we will describe stems from the original discoveries of Hill and of Kuwana and their coworkers1.2 who first demonstrated reversible, diffusion-controlled voltammetry of cytochrome c at a solid electrode without mediators and with no sign of denaturation.The all-important feature is application of a controlled electrochemical potential which is varied either continuously or in steps or pulses: this activates specific redox centres, the reactions of which are measured simultaneously through the current response, thereby procuring interdependent thermodynamic and kinetic information from a single set of experiments. The interactive nature of the experiments is optimised by confining the sample under investigation to the electrode surface, in a configuration which we refer to as ‘protein-film voltammetry’ and upon which we will focus our discussion.In this way, limitations due to the complex and sluggish diffusion of irregular macromolecules to an electrode surface are avoided. The cartoon in Fig. 1 depicts the ‘ideal’ case-a homogeneous electroactive monolayer of protein molecules-which, at a coverage of between 10-11 and 10-12 mol cm-2, enables minuscule quantities of sample, often in scarce supply, to be studied. As will be evident, the sharpness and finite nature of the voltammetric response observed for surface-confined redox couples compensates for the resulting low Faradaic currents. The information obtained is wide ranging, beginning most fundamentally with interfacial elec- tron-transfer kinetics, coupled chemical equilibria in labile ~ ~~~~~~~~~~~ St.John’s College. His interests are in bioinorganic redox chemistry, in particular the application of physical methods to detect and measure unusual reactivities and to characterise transient or unstable species. Hendrik (Dirk) Heering received his undergraduate education in Molecular Sciences at the Agricultural University, Wageningen, The Netherlands in 1990. He obtained his PhD in 1995 for studies in bio-electrochemistry under the supervision of Professors W. R. Hagen and C. Veeger at Wageningen. He is currently carrying out postdoctoral research with Fraser Armstrong at Oxford University. Judy Hirst is from Yorkshire and graduated in chemistry from St. John’s College, Oxford in 1994 after working with Professor Richard Compton for a year on photoelectrochemistry.She is currently Senior Scholar at Lin- coln College and is completing a doctorate under the supervision of Fraser Armstrong. Judy Hirst Chemical Society Reviews, 1997, volume 26 169 centres, and kinetics of gated electron-transfer reactions, and, at the most complex level, the organisation of catalytic electron transfer in multicentred redox enzymes. It is easy to forget that one is also able simply to measure reduction potentials! ELECTROLYTE 2 Interpreting voltammograms 2.1 Reversible and coupled electron transfer The majority of studies on adsorbed proteins have been carried out using cyclic (linear sweep) voltammetry which is the most 'visual' of dynamic methods, revealing potential (E) and time perspectives in the simplest way.The scope for studying redox- protons, metal ions, ligands, subtrates, etc, at controlled activity/flux It electrons, at controlled energy ELECTRODE Fig. 1 An idealised configuration for protein-film voltammetry. The protein molecules are arranged as a perfect monolayer, each behaving independ- ently. They interact non-covalently with functionalities on the electrode surface to produce orientations allowing facile interfacial electron exchange and interaction with agents in the contacting electrolyte. 1 -1 active sites and their reactions by analysing the voltammetric waveform is illustrated in Fig. 2. Three situations are con- sidered, each of which is simplified by being reversible in terms of interfacial electron transfer: they differ in how this is coupled to further chemical processes.Fig. 2(a) shows the voltammogram expected for a simple, uncoupled, reversible electron transfer for a surface-confined species. We use the term 'reversible' in the electrochemical sense to indicate that reduced and oxidised forms remain in Nernstian equilibrium throughout the cycle. The reductive and oxidative waves are symmetrical, and currents reach a max- imum value at the formal reduction potential E"', thereafter decreasing to zero as the finite number of redox centres are transformed. The separation between peaks (AE,) is zero, the half-height peak width 6 is 3.53 RTInF (83 mV at 0 "C) and the peak current i, = n2F2vAT/RT (Y is the scan rate, n is the number of electrons being transferred, A is the electrode area and r is the surface coverage).3 Obviously, a co-operative two- electron reaction gives rise to a much more prominent signal since i, is proportional to n2 and 6 varies as lln.Integration gives the number of electrons transferred and hence direct calculation of the electrode coverage (peak area = nFAT). For a protein of molecular mass 100000, maximum (monolayer) coverage is in the region of 3 X 10-l2 mol cm-2. An example of what happens when the electron transfer is coupled to a non-catalytic chemical reaction is shown in Fig. 2(b).This extension is described generally in terms of the square-scheme mode1435 where the electrochemical and chem- 1 0 -2 -3 -4 -5 I -6 e-e-A A.A 1A-A 1A-product substrate Fig. 2 (a) The ideal response expected from a monolayer of adsorbed electroactive species when the electron transfer is reversible. Currents are presented normalised to n2F2vAT/RT. (6) A possible response obtained when the electron transfer is followed by chemical conversion to a more stable electroinactive species. (c) Conversion of a one-electron reversible wave to a one-electron reversible catalytic wave on addition of substrate. 170 Chemical Society Reviews, 1997, volume 26 ical reactions are assumed to occur separately and are represented respectively by horizontal and vertical co-ordinates. A redox reaction is said to be ‘gated’ when its rate is controlled by a chemical reaction (vertical transition) rather than by the mechanics of elementary electron transfer (horizontal transi- tion).The kinetics of these chemical transformations are investigated by varying the experimental timescale, for example by changing the concentration of a species involved in a bimolecular step, or, more importantly, by using different potential scan rates. Voltammetry is therefore the obvious and direct method for detecting and deconvoluting gated redox reactions. The situation depicted is one in which the time constant for interconversion between reduced state A- and the more stable successor species B- is comparable with the experimental timescale, and B-is electroinactive within the experimental potential range.Obviously, the reductive and oxidative scans are no longer symmetrical and the reduction potential now reflects the thermodynamic characteristics of the diagonal interconversion between A and B-instead of the elementary electron exchange A -A-. The reductive wave is shifted to higher potential and sharpened as the Nernstian equilibrium is ‘pulled over’ by removal of A-; conversely the re-oxidation of B- is gated because it must first convert to A-. Species A- is metastable, and although such states may have functional relevance, we note that they might escape detection by conventional (slow) potentiometric methods.6 An example is described later (Fig. 9 and lo). As the scan rate is lowered, the waves eventually become symmetrical and the reduction potential reflects a thermodynamic distribution of species.Fig. 2(c) shows the voltammetry expected when reversible electron transfer is coupled to catalytic turnover. The active site is electrochemically transformed (in this case it is reduced) and then restored (re-oxidised) to the initial state by substrate, the mass transport of which can be controlled hydrodynamically, most obviously by rotating the electrode. Electrons are thus no longer confined to the adsorbed protein film, and the balance produced between electrochemical and catalytic redox trans- formation of the active sites results in a steady-state current- potential response. (Note that rotating the electrode does not affect signals from the adsorbed centres in the absence of catalytic turnover).The current relates to the rate of electron transport through the enzyme, and the standard kinetic parame- ters (k,,,, KM) can be determined from the dependence of limiting currents on rotation rate and substrate concentration.7 For more complex systems, such as those containing several redox sites, the shape and position of the steady-state catalytic wave reveal subtleties of intramolecular electron transport. One item of interest is how the effective ‘n-value’ (the gradient of a plot of log((illrn-i)/i}vs.E) varies with potential and reflects the patterns of redox equilibration along the electron-transfer chain comprising [substrate+nzyme centres-electrode]. The complete catalytic i-E dependence will be referred to through- out as the catalytic waveform and may comprise multiple waves having different waveshapes.Cyclic voltammetry provides a ‘wide-angle’ picture of complex redox chemistry; for example each of the cases A, B and C can be selectively accessed by correct choice of conditions (such as supply and concentration of substrate, and experimental timescale). However, there are drawbacks: for example, the sensitivity of analogue linear-sweep voltammetry is low [i.e.signals for non-catalytic (i.e.un-amplified) reactions are small], and use of a potential modulation, for example as in square-wave voltammetry, can significantly raise detection limits.8 Other methods offer advantages for deconvoluting the potential and time domains.Thus, whereas extraction of transient kinetic data from a cyclic voltammogram requires simulation procedures, potential-step methods, i.e. chrono-amperometry and chronopotentiometry, remove the potential domain and yield direct measurement of rates. These techniques are viable for adsorbed proteins, just as they have been used for small molecule systems (see below).9Jo Methods based on impedance, using frequency to impose the time domain, have also been used to measure electron-transfer rates of adsorbed species, with the advantage that large potential excursions are not involved.’ 3 ‘Simple’ voltammetry: theory and experiment 3.1 Marcus theory and electrochemical rates The ‘reversible’ electrochemistry assumed above is obviously an idealistic notion, and voltammetric waveforms must under certain conditions (e.g.sluggish electron transfer, high scan rates) become sensitive to the kinetics of interfacial electron exchange. Use of protein-film voltammetry to investigate complex redox reactions of proteins necessitates that the interfacial electron-exchange characteristics be defined. Major recent developments in the understanding of interfacial electron transfer between an electrode and adsorbed species stem from studies of systems having acceptable structural definition, and in particular from the application of Marcus theory.13 The more traditionally used Butler-Volmer formulism3 takes no account of the nature of the electronic states involved, and two major assumptions are made.First, it is assumed that the reaction surface is linear, so that potential energy varies linearly rather than parabolically along the reaction co-ordinate, and secondly, all electrode energy levels apart from the Fermi level are ignored. As a result, the variation of electrochemical rate constants with overpotential is incorrect: Butler-Volmer theory predicts an ever (exponentially) increasing electrochemical rate constant, whereas in fact (and correctly predicted by Marcus theory) the rate reaches a constant level (independent of applied potential) at high overpotential, as shown in Fig. 3. This ‘plateau’ is the electrochemical equivalent of the inverted region.9 Rate constants at large overpotential are thus greatly over-estimated by the Butler-Volmer model.25 T n EIV Fig. 3 Comparison of electrochemical rate constants predicted by Marcus and Butler-Volmer theories. (a) and (b): Marcus theory rate constants for small (0.2 eV) and larger (0.4eV) reorganisation energies respectively, the same plateau level (2 X 105 s-1) is achieved in each case. (c): the Butler- Volmer dependence with ko the same as for (b). Butler-Volmer theory thus predicts that even a very sluggish electron-transfer reaction should exhibit a sharp waveform (since the electron-transfer rate increases exponentially), the irreversible limiting value for 6 being 62.5/a mV at 25 OC,i.e. 125 mV for a (transfer coefficient) = 0.5. Broadening arises from non-ideality ,an obvious example being kinetic dispersion in which different orientations of molecules display a spectrum of rate constants.12 However, Marcus theory predicts broad- ening even for an ideal, homogeneous array since the rate of a process having a small reorganisation energy quickly ceases to respond to an increase in driving force.This is illustrated in Fig. 4, which compares waveshapes expected, in such a situation, for Butler-Volmer and Marcus models having the same value of ko. For cyclic voltammetry, the interfacial electron-transfer kinetics can be measured by analysing how peak positions vary with scan rate.l4,I5 The Marcus model Chemical Society Reviews, 1997, volume 26 171 Normalised Current 0-8~ ,,., b.4+ -0.8l Fig. 4 A comparison of the waveshapes predicted by Marcus theory (-) and Butler-Volmer theory (---) for an 'irreversible' reaction with equal ko values.The Marcus reorganisation energy is 0.2 eV and the Butler-Volmer 'a'is 0.5. predicts electrochemical rate constants by summing over all the individual rate constants for each Fermi level in the electrode as given below. The separate rate constants are the standard Marcus long-range electron-transfer rate constants specific to two energy levels: the actual rate of electron transfer from or to a specific level is influenced by the probability of occupancy of that level as predicted by the Fermi-Dirac distribution, eqn. In this equation, kred is the rate constant for reduction, E, is the energy of a specific Fermi level, E is the applied potential, EO is the reduction potential of the adsorbed species, h is the reorganisation energy and VR2 is a constant relating to the degree of electronic coupling.The behaviour of systems having poor electronic coupling and low reorganisation energy (e.g. metalloproteins) is better accounted for compared to the Butler- Volmer model, particularly under conditions of fast scan rates where the peaks occur at higher overpotential and lie more in the plateau region. 3.2 Small molecule model systems Cyclic voltammetry and other methods such as chronoamper- ometry are being used to test these theoretical models using surface-confined redox systems possessing a high level of chemical definition. A general theme has been to derivatise a metal surface with a layer of bifunctional molecules as illustrated by Fig.5. Here, a functional group such as a redox- active entity is linked to the surface via an alkyl spacer (CH;?) terminated by a sulfhydryl. Several groups have studied the electrochemical kinetics of self-assembled monolayer (SAM) structures consisting of ferrocenes linked to gold through alkane thiols of varying chain length. Typically, the electrode is a gold mirror surface prepared by sputtering on silica wafers. Notably, Chidseyg was able to predict the variation of electron-transfer rate with free-energy and temperature, and his conclusions have been supported and extended in studies by Murray and co-workers,l5 and by Weber and Creager.14 Another system investigated in detail features OsI*IJ*complexes adsorbed on platinum.lo Monolayer coverage of [O~(bipy)~Cl(X)] (bipy is 2,2'-bipyridyl; X is 4,4'-bipyridyl or analogues with two or three CH2 groups spacing the pyridyl rings) is achieved at platinum microelectrodes utilising the X is redox group, e.g. ferrocene or functionality interacting with protein xxxxxx 4 I Distance fixed by number (n) of , in spacer ssssss + I II I IIIIII1IIIIIllITrIrTlrl IIII I Ill IIIIII Au electrode Fig. 5 Cartoon showing the structure of electrode surfaces modified with a monolayer of functionalised alkanethiolate. The group X may be redox active or a functionality such as carboxylate that is capable of interacting with the surface of a protein.pendant pyridyl-N atoms from ligands X as anchors. The relatively ideal response allows detailed investigation of the effects of electrolyte, solvent, temperature and electron-transfer distance. At a sufficiently large overpotential, electrochemical rates were observed to become independent of driving force, again indicating the validity of Marcus theory. This work raises several important points relevant to observations we have made with proteins, and indicates the usefulness of such systems as models for protein film voltammetry. First, the response is asymmetric with respect to overpotential and can be modelled as a tunnelling process between electronic manifolds on either side of the interface. Secondly, a finite peak separation remains at lowest scan rates.Thirdly, the rate constant is approximately, but not completely, independent of ionic strength, raising the possibility that ion transportbinding effects are influential. 3.3 Proteins at electrodes: configurations The concept of structurally defined monolayers has been extended to study electron transfer in protein films. Cytochrome c adsorbs strongly at SAMs prepared by treating gold mirror electrodes with u-mercaptocarboxylic acids of the type HS(CH2),COOH, providing what is so far the best achievement of a protein4ectrode interface having a definable struc-ture.11,'2316 The principle is that the carboxylates project out into the solution where they interact with the protein, presumably by salt bridging to lysines (-NH3+) located around the exposed haem edge.Bowden and co-workers have studied interfacial electron transfer in cytochrome c SAMs using cyclic voltam- metry and impedance. Protein coverages were measured independently by X-ray photoelectron spectroscopy. 1 Electro-active monolayers were obtained for films composed with SAMs having n = 15, although the rate constants were small due to poor electronic coupling. As expected, shorter chain lengths gave higher rate constants, but a smaller fraction of the adsorbed cytochrome c molecules were electroactive, suggest- ing incorrect orientation. Niki and co-workers have used a potential-modulated optical method (AC modulated UV-VIS electroreflectance) to calculate apparent electron-transfer rate constants for different chain lengths.16 Although results for small chain length were lower than predicted, those for larger n agreed with distance-dependence expectations for a through- bond tunnelling mechanism. Studies in our laboratory have focused on the use of pyrolytic graphite edge (PGE) electrodes which are prepared simply by polishing with an alumina slurry and cleaning by ultrasound. This produces a fresh, rough surface which is hydrophilic and rich in acidic oxides.17 The roughness is probably important in enabling the irregularly shaped macromolecules to make multiple polar contacts with the electrode. Adsorption and stabilisation are optimised by low temperatures (e.g. 0 "C), choice of electrolyte conditions such as ionic strength, and co- adsorbates such as aminocyclitols that may create ternary salt 172 Chemical Society Reviews, 1997, volume 26 bridges between like-charged regions of protein molecules and T(uelectrode surface.The strategy is empirical but pragmatic. It ‘A’ provides a surface which has a wide potential range in aqueous solutions and is effective for a wide range of proteins, for which we are interested not so much in the mechanism of interfacial electron transfer per se but more in the characterisation of active sites and measurements of redox-coupled activities. We have found that a major obstacle to success is not so much electrode preparation but the failure to have samples of sufficiently high purity-denatured proteins and fragments probably tending to adsorb preferentially and attenuate the response.3.4 The integrity of adsorbed proteins There are several accounts of spectroscopic studies to determine the degree to which proteins retain their native structure when adsorbed at an electrode. Surface-enhanced Raman spectros- copy has been used to study cytochromes adsorbed at sil~er.~z?~9 These experiments have revealed that the native state can be preserved even at an unmodified metal electrode. Cytochrome c3 exhibits particular robustness, while for mitochondria1 cytochrome c, structural changes do occur but depend on the electrode potential that is applied. Hinnen and Niki have used electroreflectance to examine cytochrome c at modified metal surfaces20 while Bowden’s group have studied cytochrome c adsorbed at optically transparent indium tin oxide.21 The intense Soret absorption band provides an excellent spectroscopic handle even though the quantity of sample in a monolayer is so small. It was concluded that little alteration in the haem environment occurs upon adsorption at this hydrophilic sur- face.An obvious key factor is the degree to which the reactivities observed in the adsorbed state compare with those characteristic of the protein when free in solution. These quantities can range from reduction potentials of well-characterised systems, to rate and equilibria data for coupled reactions. Small discrepancies in the reduction potential are easily tolerated given that the more robust and characteristic chemical properties are maintained.In general, a certain degree of alteration is expected because of the special nature, e.g. field inhomogeneity, of an interfacial environment, although we decline to accept that such a medium is non-physiological since so many biological processes occur at comparable interfaces e.g. membranes and large protein surfaces. As an example, cytochrome c adsorbed at a gold/ alkanethiolate SAM surface exhibits a reduction potential of 215 mV compared to the value of 260 mV characteristic of the protein in solution: however the surface-adsorbed value is closer to data determined for cytochrome c bound at biological interfaces, for example 225 mV when complexed with cyto- chrome c oxidase.Il For enzymes, the situation is more easily assessed since one can measure and compare specific catalytic activities.3.5 Even ill-defined protein films produce surprisingly ideal electrochemical characteristics Protein film voltammetry at PGE electrodes produces some excellent results despite the probability that the electrode surface and protein film are very different to the flat ordered array depicted in Fig. 1. Peak widths 8 measured at low scan rates are often observed to be at or just above the theoretical minimum (hence no thermodynamic dispersion). This result is not unexpected because (unlike small molecules) the redox centres are well separated by the insulating protein matrix and their local environments are defined independently of solvent- solute interactions. However, there are deviations from ideal behaviour, such as asymmetry between oxidative and reductive scans, that may be attributable to conformational changes in the adsorbed state, or to asymmetry in the electron-transfer characteristics as has been observed for osmium monolayers at platinum.10 However, this does not greatly detract from the basic results of our experiments; indeed we have been able to achieve cyclic voltammetry of adsorbed proteins at scan rates in Fig.6 Rapid cyclic voltammetry of Pseudomonas aeruginosa Azurin, a ‘blue copper’ protein, at 500 V SKI,pH 8.0,O “C.Baseline subtracted peaks (not to scale) are shown in the centre of the voltammogram. the region of 1000 V s-I.Fig. 6 shows the voltammetry of Pseudomonas aeruginosa Azurin, a ‘blue’ copper protein, measured at 500 V s-l. The timescale is now fast enough to render any coupled reactions on the millisecond timescale as rate limiting and thus gating. Such reactions, if they occurred, would remain unnoticed and unaddressed by static electro- chemical methods or indeed by conventional voltammetry which relies on sluggish protein diffusion. In summary, it is becoming clear that electron exchange between an electrode and an adsorbed protein may be extremely facile, thus enabling investigations to focus upon the characteristics of the protein itself. 4 The film voltammogram as a ‘spectrum’ An obvious and straightforward application of protein film voltammetry is to use the voltammetric signals (i.e.a pair of oxidation-reduction peaks) as electrochemical markers for identifying and quantifying the status of centres present within the protein. In effect, this amounts to an interactive ‘spectrum’, but in stating this we note that voltammograms provide no structural information. Assignment of signals thus depends on the use of true spectroscopic methods such as EPR to examine species generated in solution at corresponding potentials. 4.1 A sensitive method for observing Fe/S clusters and characterising their redox chemistry Iron-sulfur (Fe/S) clusters are well established as electron- transfer centres, but much less is known about their atom- transfer properties, i.e.their abilities, in different proteins, to undergo changes in composition and ligation. Iron-sulfur clusters generally lack characteristic spectral features useful for real-time monitoring, and their reactivities depend critically upon oxidation level, thereby creating problems that can render conventional investigative strategies ineffective. Protein-film voltammetry provides a way to control the chemistry through the applied potential and observe the cluster transformations that are induced.22-24 The spectrum analogy is well demon- strated for proteins that contain several such centres. Fig. 7 shows a voltammogram obtained at 20 mV s-1 for a small negatively charged protein (a ferredoxin from Sulpholobusacidocaldarius) containing a [3Fe-4S] cluster and a [4Fe-4S] cluster.25 The film, displaying activity corresponding to mono- layer coverage, was formed at a PGE electrode, at 0 “C and in the presence of polymyxin.Signals from the two centres are simultaneously visible and their distinctive redox character- istics are apparent from simple inspection. Signals A’ and B’ have been respectively assigned to the redox couples [3Fe- 4S] and [4Fe-4SI2+/1+ on the basis of correspondence with species generated electrochemically as free solution species and Chemical Society Reviews, 1997, volume 26 173 A 0.2 pA V I I 1 I I -0.8 -0.6 -0.4 -0.2 0 EIV vs. SHE Fig. 7 The voltammogram as a spectrum: Film voltammogram of the 7Fe ([3Fe-4S] + [4Fe-4S]) ferredoxin from Sulfolobus acidocaldarius.Signals A’, B’ and C’ correspond to the couples [3Fe-4SlL+/O, [4Fe-4SI2+I1+ and [3Fe-4S]0/2-.characterised by EPR and MCD spectroscopy.25 The voltam- mogram reveals novel redox activity: signal C’ corresponds to the couple [3Fe-4S]0/2- which, as a two-electron reaction involving multiple proton transfer, has no precedent in Fe/S chemistry. The C’ signal appears for other proteins containing [3Fe-4S] clusters: using the ferredoxin from Sulpholobus acidocaldarius, the hyper-reduced form [3Fe-4S]*- has been generated electrochemically in solution and partially charac- terised by adsorption, EPR and MCD spectroscopies.26 Closer inspection shows how well the voltammetry conforms to the expectations outlined earlier.Signal A’ has half-height widths of 90-100 mV at modest scan rates, thus showing that there is little (thermodynamic) dispersion due to differing orientations and environments. Signal C’ is much sharper, with the reductive peak having a half-height width of 50 mV and charge integral twice that of A’. Indeed, as examined by voltammetry, appearance of the two signals, A’ (one-electron) and C’ (intense, narrow two-electron signal) constitutes the characteristic signature of a [3Fe/4S] cluster.26 Signal B’ appears as a shoulder on the side of signal C’ but is nevertheless clearly defined. The surface coverage based on integration of signal A‘ and assuming a eometric electrode area equates to an area of ca. 600-700 12 per protein molecule: perhaps fortuitously, this value is approximately as expected for a flat monolayer.The peak separations associated with signals A’ and B’ remain small even at scan rates in excess of 10 V s-l, whereas signal C’ undergoes complex changes. These observa- tions relate to the rate and extent of coupling of electron transfer. The redox couple [4Fe-4S]2+/’+ is a ‘pure’ electron transfer reaction, while [3Fe-4S] 1+/0 involves rapid coupled proton transfer. By contrast the [3Fe-4S]0/2- couple is compli- cated by coupling to slow, pH-dependent chemical reactions. 5 Coupled reactions-non-catalytic The capability for achieving precise temporal control of electron transfer enables complex redox-coupled chemical 174 Chemical Society Reviews, 1997, volume 26 reactions to be studied.In effect, the experiment involves causing an electron (or hole) to cross the electrode-protein interface and then ‘calling it back’ after a brief period of time. Transient species can be detected and rate constants determined, the upper limits of which will certainly increase as technical developments improve the experimental timescale. The kinetic parameters fit naturally within the overall thermodynamic picture that voltammetry provides. For iron-sulfur clusters, studies have been made of the reversible binding of metal ions, H+, and ligands, each of which display affinities and rates that depend on cluster oxidation level. 5.1 Controlling and visualising cluster transformations The value of the voltammetric ‘spectrum’ is increased for proteins containing a labile centre (an analogy being that a moving object is more visible than a static one).Introduction of Zn2+ ions (10 VM) to the electrolyte contacting a film of the 7Fe ferredoxin from Desulphovibrio africanus results (Fig. 8) in rapid attenuation of signals A’ and C’ and appearance of a new signal *, as the [3Fe-4S] cluster converts to a cubane formulated as [Zn3Fe-4S] (analogous to [4Fe-4S]). Other metals give [M3Fe-4S] products having different reduction potentials. C’ B’ A‘ I I I I I I I I -0.8 -0.6 -0.4 -0.2 0 EIV vs.SHE Fig. 8 Detecting interactions between redox centres and exogenous agents: Voltammograms (direction of increasing potential) obtained for a film of 7Fe ferredoxin I11 (Desulfovibrioafricanus) showing the rapid reaction of [3Fe-4S]O clusters with Zn2+.Signals A’ and C’ disappear, and are replaced by a signal (*) assigned to.[Zn3Fe-4S]2+/1+ partly overlaying the signal (B’) due to the inert [4Fe-4S] cluster. The affinity of the [3Fe-4S] cluster for metal ions depends on its oxidation level. Transfer to a cell containing a complexing agent such as EGTA (or cyanide for M = Cu) and cycling to high potentials causes transformation back to [3Fe-4S]’+. Such transformations are relevant to cluster assembly in proteins and to the emerging role of these centres in cellular sensory functions .26 The system is suitably described in terms of interconnected square schemes, shown in Scheme 1, in which solid arrows indicate the experimentally observed pathways.Since the entire protein sample responds immediately to the electrode potential, specific binding equilibria are addressed by holding the potential at selected values. In this way, the binding of M2+ ions (M = Fe, Zn and Cd) to [3Fe-4S]O producing [M3Fe-4SI2+ could be studied quantitatively, minimising the effects of e--[3Fe-4SIo[3Fe-4SI1+ e-*- - -- - -+ [3Fe-4SI1-I1M"' : M"' t e-e-[M3Fe-4S]("+ l)+ +--+ [M3Fe-4SIn+ -[M3Fe-4S](" -I)+U V M = TI+, CU' M = Fe2+,Cd2+,Zn2+ Scheme 1 Electrochemical relationships between [3Fe-4S] and [M3Fe-4S] clusters, represented by connected square schemes coupling electron transfer to metal binding competing redox reactions. By measuring the magnitude of signals due to the cubane product [M3Fe-4S] as a function of metal ion concentration, dissociation constants & (= { [3Fe-4S]O } { M2+ )/{[M3Fe-4S]2+}) were determined and the affinity order Cd > Zn > Fe was deduced.22 The ferredoxin from D.africanus is unusual in that the [3Fe-4S] cluster lacks a cysteine residue to coordinate the incoming metal ion, as is normally the case for a [4Fe-4S] cluster.22 Instead, an aspartate or perhaps a water molecule completes the coordination in [M3Fe-4S], resulting in a weaker affinity for M. The controlling factor is the poor metal coordinating capability of the oxidised [3Fe-4S] l+ cluster: application of an oxidising potential causes ejection of M, in a reaction that mimics the degradation process known to occur for 3Fe/4Fe proteins such as aconitase.Feinberg and co- workers studied this reaction in aconitase it~elf.~7 Although not configuring the sample as an adsorbed film, they could detect cluster interconversions occurring in solution by monitoring changes in the square-wave voltammogram. By contrast with the relatively slow reactions observed with divalent transition metals, T1+ binds and dissociates much more rapidly, i.e. within the timescale of voltammetry performed up to a scan rate of 1V s-1 .23 During the passage of a voltammetric cycle, this system equilibrates among all the species shown in the left-hand square of Scheme 1 so that just a single signal is observed over a wide range of TI+ concentration. The equilibrium constants for the reactions were determined from the variation in reduction potential with T1+ concentration.The results showed that T1+ binds tightly to [3Fe-4S]O (& = 1 VM) and weakly to [3Fe-4S]l+ (Kd = 0.1 M). 5.2 Rate-limiting proton-transfer reactions Long-range electron-transport systems involving coupled pro- ton transfer may be limited not by electron transfer but by the rate at which a proton can hop between bases within the protein interior. An example is the redox cycling of bound quinone in photosynthetic reaction centres in which rate-determining proton transfer to the cofactor is mediated by bases such as the side-chain carboxylates of aspartate and glutamate.28 These are obviously gated electron-transfer reactions, the details of which can be studied by voltammetry.The [3Fe-4SIo (i.e. one-electron reduced) cluster in various proteins binds a single proton, to produce a spectroscopically distinct form H+-[3Fe-4S]0.25 The reaction is depicted in Scheme 2. For the structurally characterised 7Fe ferredoxin controlling: H' tH4 effect, over short I t timescale or with e-H+-[3Fe-4S]'+ *- - --H+-[3FdSIa D15N mutant Scheme 2 Square scheme representing redox-coupled protonation in the [3Fe-4SI1+/O system. Slow transfer of H+ from the [3Fe-4S]O cluster gates the oxidation to [3Fe-4S]+. from Azotobacter vinelandii, the pK for this process is 7.8, necessitating protonation even at neutral pH.29 The [3Fe-4S] cluster is buried 8 A below the protein surface but an aspartate carboxylate is located close to the surface in a position where it could act to mediate protons.There is no access of H20 to the cluster, thus eliminating the possibility of H+-mediation by water. Voltammetry reveals interesting differences between the native protein and a mutant form (D15N) in which the aspartate is replaced by asparagine. The comparison is shown in Fig. 9. The native protein shows a reversible oxidation and reduction (i.e. conversion between the oxidised and the reduced-proto- nated cluster is facile) whereas the lack of a clear oxidation wave in the D15N voltammogram reveals trapping of the reduced-protonated species by the rate-limiting excape of the proton from the cluster.29 The fourth species, the oxidised protonated cluster, H+-[3Fe-4S]l+ is included in Scheme 2 in order to complete the cycle; however, rapid sweeps to high potentials have revealed no evidence for its participation.The course of the electrochemically driven cycle thus involves reduction of the cluster followed by a reversible protonation and return along the same route. --b r 30 II I I I ! -50 -0.8 -0.6 -0.4 -0.2 0.0 EIV vs. SHE Fig. 9 Proton-gated electron transfer: Voltammograms recorded (1 V s-I, 0 "C)for Azotobacter vinelandii 7Fe Ferredoxin I native and D15N mutant forms, revealing retardation of oxidation of [3Fe-4S]O upon replacement of aspartate by asparagine. D15N (pK 6.9) pH 5.5; Native (pK 7.8) pH 6.5 thus experiments performed at similar protonic driving force.Shown in Fig. 10 is a plot of the peak positions of the oxidative and reductive waves for the D15N mutant as a function of scan rate at two pH values. At pH 8.5 (the pK for D15N is 6.9), simple electron-transfer characteristics are observed [see Fig. 2(a)] which can be analysed by Marcus theory.14-15 In contrast, at pH 6.5 the behaviour is complex: at low scan rates, the voltammetry appears reversible because protonation and deprotonation of the cluster occur within the timescale of the potential perturbation. The increased reduction potential reflects the thermodynamics of the diagonal transition (the redox Bohr effect). As the scan rate is increased (all scans start at the high-potential limit, and only the initial scan is considered) the oxidation peak becomes attenuated, less defined and then vanished [as evident from Fig.9 and depicted in Fig. 2(b)] because the cluster becomes trapped in the protonated form. (The second cycle reveals, as expected, that the reduction wave is also attenuated.) Finally, at high scan rate, the voltammetry reverts to that expected for electron transfer alone: the peak positions now overlay the curves observed at pH 8.5 because the scan is reversed before the reduced cluster is able to Chemical Society Reviews, 1997, volume 26 175 0.15T I m/ 0.01 0.1 100 10001-wI -0.15I Fig. 10 Proton-gated electron transfer: Peak separations as function of scan rate for voltammetry of the [3Fe-4S]l+”J cluster in the D15N mutant of Azotobucter vinefundii Ferredoxin I..--data points at pH 8.5 (no protonation of [3Fe-4S]O). O-data points at pH 6.5. The lines indicate the predictions of the square-scheme model (Scheme 2). The potential scale is with respect to the potential observed in the absence of proton transfer. protonate. The data are modelled using rate constants for protonation and deprotonation in accordance with the pK. In this case, the rate constant for deprotonation of the cluster ( < 5 s-I) is much slower than observed for the native protein. Other coupled reactions of iron-sulfur clusters have been detected, and a detailed study has been carried out on the rapid and reversible binding of an exogenous thiolate ligand (RS-) to the 2+ and 1+ oxidation levels of the transformed [Fe3Fe-4S] cluster in D.africanus ferredoxin 111.30 In this case it was possible to observe all four states of the square scheme, and to determine rates and equilibrium constants.Notably, binding of RS-to the reduced cluster is too weak to permit its measurement by EPR, the obvious method for detecting such interactions. 6 Coupled reactions-catalysis Studying enzymes by protein film voltammetry holds several attractions, a main feature being that kinetic measurements can be varied from steady-state to transient mode, and carried out on a minuscule population of enzyme molecules which can be interrogated at the same time to address the status of active sites. A high degree of control is exercised over the electrochemical potential and hence the driving force, although (as eluded to earlier) care is needed in order to judge that potential- independent ‘limiting currents’ are not the result of reaching the Marcus plateau region.Furthermore, hydrodynamic control, imposed by the rotation of the electrode, allows the substrate to be supplied to the enzyme at a given rate and enforces removal of products. The detailed catalytic waveform (shape, i.e. gradient, halfwave potential), its variation with conditions, and how it relates to the signals obtained in the absence of turnover, together lead to a picture that integrates both thermodynamics (reduction potentials, substrate binding) and kinetics (both of electron transfer and associated chemical events).The follow- ing examples illustrate the kind of information that can be derived. 6.1 Detection of co-operative electron transfer activity at a high-potential catalytic site Cytochrome-c peroxidase (CCP) is a relatively small enzyme (Mw 34 000) which contains a single Fe porphyrin and catalyses the reductive breakdown of H202 by reduced cytochrome c. The 176 Chemical Society Reviews, 1997, volume 26 catalytic cycle (see Scheme 3) involves a highly oxidising intermediate known as ‘Compound I’ formally at the oxidation level of FeV, but which actually contains FeTV and a radical cation located on a nearby tryptophan (W191). Amino acid radicals and FeIV species are each now known to be very important in oxygenative catalysis; however, the high reduction potentials and reactivities of these species have hindered quantitative studies of their redox chemistry.e-, 2Hik i-.-_8 FeIV e-Scheme 3 The peroxidase catalytic cycle, and replacement of the chemical electron donor (cytochrome c) by an electrode At low ionic strength and low temperature, CCP can be adsorbed on a PGE electrode.31 In the absence of H202, a reversible signal is observed at 720 mV (pH 6.1) which transforms into a catalytic wave when H202 is added. Typical results are shown in Fig. 11. The peak signals correspond to reduction and oxidation of the catalytic machinery and provide a new means of studying the high-potential states. The half- height widths of the peaks in both directions are below the theoretical value for a one-electron redox reaction (i.e.y1 > 1) thus identifying them with a co-operative two-electron redox couple. The implication is that the redox activities of the haem FeIVDrr and Tryptophan- 19 1 are tightly coupled. The catalytic wave produced when H202 is added arises [as described in Fig. 2(c)] because the H202 reaction feeds electrons back into the system. Studies carried out by varying electrode rotation rate, pH and substrate concentration show that this is a very active enzyme, equivalent or even superior to the system studied in solution with cytochrome c as electron donor. 20 10 cu Of\ .--I0 -20 0.4 0.6 0.8 1 EIV vs. SHE Fig. 11 Peroxidase electrocatalysis: Cyclic voltammetry of yeast cyto- chrome c peroxidase obtained at a rotating disc electrode, pH 6.1, (20 mV s-l,O “C) showing how the reversible peaks transform to a catalytic wave in the presence of H202 (20 PM). Upper trace (not to scale) shows baseline- subtracted signal in absence of H202. 6.2 Delineation of electron flow patterns in multicentred enzymes For multi-centred enzymes, the ability to control and fine-tune the electrode potential allows different redox centres to be addressed selectively, and their roles in catalysis examined.Fumarate reductase (FRD) is a multicentred enzyme catalysing the reduction of fumarate to succinate in bacterial respiratory chains.7 It is associated with the cytoplasmic membrane and consists of four subunits, of which two form the membrane anchor and quinone binding sites.The other two subunits are membrane-extrinsic and contain three Fe-S clusters ([3Fe- 4S]1+/0, [4Fe-4S]2+/1+ and [2Fe-2S]2+/1+) and a covalently bound flavin (FAD). Reduction potentials of the [3Fe-4S] and [2Fe-2S] clusters (reported values at pH 7 range between -20 and -79 mV) and the flavin (-55 mV) are suited to mediate electrons between menaquinol (-73 mV) and fumarate/ succinate (+30 mV) but the low potential of the [4Fe-4S] cluster (-320 mV) poses a question as to its role. The membrane- extrinsic domain can be isolated and the resulting soluble enzyme catalyses fumarate reduction by artificial electron donors. As depicted in Fig.12, the soluble protein adsorbs at a PGE electrode, whereupon its properties are revealed in more detail. succinate fumarate succinate fumarate k 4 i Enzyme bound to membrane Enzyme bound to PGE via 'anchor' peptides electrode surface Fig. 12 Cartoon depicting fumarate reductase or succinate dehydrogenase. The two catalytic subunits containing three Fe-S clusters (small subunit) and FAD (large subunit) can be separated from the membrane-bound anchor peptides and adsorbed at an electrode. Cyclic voltammograms measured in the absence of substrate reveal signals that can be assigned to reduction and re-oxidation of the three Fe-S clusters and the FAD.7 Those attributable to [2Fe-2S]2+/1+, [3Fe-4S]l+/O and the FAD are enveloped together in a prominent reversible signal in the region of -50 mV, the prominence being due to the n = 2 character of the FAD.A weaker signal appears at ca. -320 mV, corresponding to the potential expected for [4Fe-4SI2+/1+. When a small concentra- tion of fumarate is added, the envelope of signals at high potential transforms to a catalytic wave. At higher fumarate concentrations, this catalytic wave becomes very large and (particularly at higher pH) a second smaller catalytic wave becomes visible close to the potential of the [4Fe-4S] cluster.7 The resulting waveform is shown in Fig. 13. From the potentials, relative intensities and shapes of the two catalytic waves under different conditions, it could be deduced that intramolecular electron transfer proceeds by two pathways (i.e. electrode +2Fe/3Fe +FAD and electrode +4Fe +FAD).The picture is enhanced by transforming the steady-state voltammogram to the derivative form where the two pathways appear as peaks. The simple catalytic wave observed at low substrate concentrations occurs because both electrons for fumarate reduction are supplied at an adequate rate by the high potential clusters. However, at high substrate concentration, the demand for electrons is greatly increased so that the [4Fe-4S] cluster is recruited, thereby resulting in a boost of turnover. Likewise, as the pH is increased, the reduction potential of the FAD decreases relative to that of the high-potential clusters making them poorer electron donors, so that activity becomes more dependent on participation of the [4Fe-4S] cluster.The 0 cu E0 -0.6 -0.4 -0.2 0 0.2 0.4 EN vs. SHE Fig. 13 Detection of electron-transfer pathways in multi-centred enzymes: (-) rotating disc cyclic voltammogram showing the reduction of fumarate (0.8 mM) by fumarate reductase, pH 9.5, 20 "C, 10 mV s-1. (---) the first derivative of the voltammogram generating a 'pathways' spectrum. results illustrate how protein-film voltammetry can be used to quantify the involvement of different centres in intramolecular electron transfer, a task that is not easily accomplished by other methods. 6.3 Detection of electron-transport switches and catalytic bias Succinate dehydrogenase (SDH) is the soluble component of Complex 11, a mitochondria1 enzyme which catalyses the oxidation of succinate to fumarate in the Kreb's cycle.In terms of structure and redox centres, SDH is similar to the soluble domains of fumarate reductase. There are three Fe-S clusters and one covalently bound FAD. When examined as a protein film, several interesting features are observed, although no signals are apparent in the absence of substrate (implying that the enzyme adsorbs to give only a low electroactive coverage) and for reasons as yet unclear, the catalytic voltammetry is unstable. Shown in Fig. 14(a) are successive cyclic voltammo- grams recorded for a SDH film contacting a solution with equal concentrations of fumarate and succinate, i.e. measuring the bi- directional characteristics of the enzyme.32.33 At first glance the voltammetry appears extremely complicated: in fact, however, the electrochemistry is even more amenable to interpretation than the enzymes described above because it is controlled entirely by the enzyme's kinetics, and not by substrate mass transport or interfacial electron exchange.Several subtle features are now revealed. Both oxidation and reduction currents are observed but these decay with time producing isosbestic points that correspond to the substrate reduction potential EFIS. The waveforms are not sensitive to electrode rotation rate or scan rate (at least up to 50 mV s-1). Most noticeably, the fumarate reduction current reaches a maximum value at a certain potential, then de- creases.32.33 Analysis is carried out by computing difference voltammograms-subtracting later from earlier cycles-which effectively removes the background and yields a relative activity profile.From Fig. 14(b) it is seen that the enzyme activity drops to a lower level (but not zero) as the potential is made more negative. We have termed this the 'tunnel-diode' effect due to analogy with the electronic component which displays negative resistance over a limited region of potential bias. In the absence of this switch, the reduction current would continue to increase to a maximum value as indicated. It is immediately evident that, at pH 7, SDH is more proficient at catalysing in the direction of fumarate reduction even with this switch in operation.For the following reasons, the tunnel-diode Chemical Society Reviews, 1997, volume 26 177 effect has been assigned to reduction of the FAD.32 First, the results of experiments conducted over a range of pH can be modelled in terms of two states of the enzyme (active and less- active) which interconvert rapidly through a simple Nernst equilibrium. The characteristics of the redox couple responsible for this interconversion are very similar to those expected for the FAD. Secondly, the magnitudes and positions of the fumarate reduction peaks are sensitive to oxalacetate, a competitive inhibitor which is known to bind close to the FAD, but more tightly to the oxidised form. Thus, the redox state of the FAD acts as a regulator of electron transport, most likely through a subtle conformational change. A similar ‘tunnel diode’ effect has been reported for the voltammetry of gluconate dehydrogena~e.3~ Successive Cycles (b) EIV vs.SHE I -0.4 I / tII I, -3 1-4 Fig. 14 Revealing a redox-state dependent activity switch: (a) Rotating disc cyclic voltammograms (10 mV s-1) observed at a PGE electrode for a 1: 1 mixture of succinate and fumarate in the presence of adsorbed beef heart mitochondrial succinate dehydrogenase. pH 7.0, 38 “C. (b) The catalytic profile or difference voltammogram for succinate dehydrogenase, derived from the voltammograms in Fig. 14(a) and the simulation (smooth curve) based on a cyclic catalytic model incorporating active and less-active states of the enzyme that interconvert according to the oxidation state of the FAD.Dashed line depicts the fumarate reduction current expected in the absence of the tunnel diode effect. 6.4 Proton transfer: the energetics of substrate-solvent isotope effects The final experiments described in this Review concern the effects of isotopic substitutions on voltammetric waveforms. The key issue is the ability to measure, simultaneously, contributions due to kinetics and thermodynamics, and to correlate these with perturbations of the active site as well as substrate transformation. Numerous conventional kinetic stud- ies have utilised H/D substitution, both in substrate and solvent, as a mechanistic tool. These substitutions have a marked influence on the steady-state waveforms displayed by SDH, and show that catalytic electron transport in both directions is controlled by proton tran~fer.3~ First, dealing with the ‘sub- strate’ isotope effect, substitution of succinate and fumarate by their perdeuterated forms produces a large (ca. four-fold) decrease in oxidation rate but no effect on reduction.Second, addressing the ‘solvent’ isotope effect, substitution of H2O by D2O attenuates the reaction rate in both directions. Isotope effects can arise from changes in both thermodynamics or kinetics, and solvent substitutions can influence intrinsic enzyme properties, particularly reduction potentials and pK values, in addition to primary (kinetic) substrate effects.By examining the waveforms obtained over a range of pL (pH, pD) for 1 : 1 fumarate-succinate mixtures either in 100% H or 100% D systems, the energetics of the overall system have been compared. The consensus voltammograms (simulations) are displayed in Fig. 15. Points to note are that in the 100% D system, the potential of the peak (reflecting the reduction potential of the FAD) increases by 20 mV whereas that for fumarate-succinate (the isosbestic potential) is increased by 43 mV. The net result is that SDH becomes even more biased toward being a ‘fumarate reductase’, although this is a relative effect since electron transport in the fumarate reduction direction is also retarded by H/D substitution. -04 -0.2 0 0.2 EIV vs. SHE Fig.15 Isotope effects on catalytic electron transport: consensus effects on the catalytic voltammetry of succinate dehydrogenase, of deuterating both the solvent and the succinate-fumarate substrates: SH/FH,H20 (faint line), SDFD,D20 (bold line). Dashed lines denote the voltammograms expected if activity was not curtailed by the tunnel diode effect. Conditions: pH = pD = 7.4, 1 :1 fumarate/succinate, 38 “C. 7 Summary This Review has outlined some ideas and described a number of experiments illustrating the results that can be obtained with protein film voltammetry. What of the future? We hope to have convinced readers of the merits for fundamental studies. As has been mentioned by numerous authors, applications could include biosensors, although this would necessitate the develop- ment of systems that are stable.With stability and reproducibil- ity established, enzymes linked to medical problems might be screened directly to obtain a ‘direct read out’ of the malfunction: obviously, in this way an enzyme is effectively integrated into 178 Chemical Society Reviews, 1997, volume 26 the electronic circuitry of an analytical instrument. Efforts are needed to understand and improve the structure of the protein- electrode interface. It will be of great interest to obtain direct structural insight into how protein molecules are actually arranged together at electrode surfaces, not only the better- behaved systems that we have focused on above, but also those that give poor or unstable activity.The successful application of scanning probe microscopy to this task will provide a major step forward in understanding and improving the methodology. Certainly, greater use will be made of spectroscopic methods such as electroreflectance whereby a potential modulation is applied to activate specific centres and measure their spectra and kinetics. With regard to subjects for study, there need be no limit, except to restate our earlier point, that it is with the more complex and intractable problems that voltammetry can come into its own. 8 Acknowledgements Research in the authors’ laboratory is supported by the Wellcome Trust and by the EPSRC. We thank Julea Butt, Jill Duff, Sarah Fawcett, Guy Jameson, Lisa Martin, Madhu Mondal and Artur Sucheta for their efforts in carrying out much of the research described.We also thank Brian Ackrell, Jacques Breton, Barbara Burgess, Andrew Thomson and Joel Weiner for their help and collaboration. 9 References 1 M. J. Eddowes and H. A. 0. Hill, J. Am. Chem. SOC., 1979, 101, 446 1. 2 P. Yeh and T. Kuwana, Chem. Lett., 1987, 1145. 3 E. Laviron, in Electroanalytical Chemistry, ed. A. J. Bard, Marcel Dekker, New York, 1982, vol. 12, pp. 53-157. 4 B. S. Brunschwig and N. Sutin, J.Am. Chem. SOC., 1989, 111, 7454. 5 A. M. Bond and K. B. Oldham, J. Phys. Chem., 1983,87,2492. 6 F. A. Armstrong, J. Biol. Inorg. Chem., 1997, 2, 139. 7 A. Sucheta, R. Cammack, J. Weiner and F. A. Armstrong, Biochemistry, 1993,32,5455.8 J. H. Reeves, S. Song and E. F. Bowden, Anal. Chem., 1993,65, 683. 9 C. E. D. Chidsey, Science, 1991, 251, 919. 10 R. J. Forster and L. R. Faulkner, J.Am. Chem. SOC., 1994, 116, 5444; R. J. Forster and L. R. Faulkner, J. Am. Chem. SOC., 1994, 116, 5453; R. J. Forster and J. P. O’Kelly, J. Phys. Chem., 1996, 100, 3695. 11 S. Song, R. A. Clark, E. F. Bowden and M. J. Tarlov, J. Phys. Chem., 1993,97,6564. 12 T. M. Nahir and E. F. Bowden, J. Electroanal. Chem., 1996, 410,9. 13 R. A. Marcus and N. Sutin, Biochim. Biophys. Acta, 1985,811, 265. 14 K. Weber and S. E. Creager, Anal. Chem., 1994,66, 3164. 15 L. Tender, M. T. Carter and R. W. Murray, Anal. Chem., 1994, 66, 3173. 16 2. Q. Feng, S. Imabayashi, T. Kakiuchi and K. Niki, J.Electroanal. Chem., 1995,394, 149. 17 F. A. Armstrong, Struct. Bond., 1990, 72,137. 18 P. Hildebrandt and M. Stockburger, Biochemistry, 1989, 28, 6710. 19 K. Niki, Y. Kawasaki, Y. Kimura, Y. Higuchi and N. Yasuoka, Langmuir, 1987, 3, 982. 20 C. Hinnen and K. Niki, J. Electroanal. Chem., 1989,264, 157. 21 M. Collinson and E. F. Bowden, Anal. Chem., 1992, 64, 1470. 22 J. N. Butt, F. A. Armstrong, J. Breton, S. J. George, A. J. Thomson and E. C. Hatchikian, J. Am. Chem. SOC., 1991, 113,6663. 23 J. N. Butt, A. Sucheta, F. A. Armstrong, J. Breton, A. J. Thomson and E. C. Hatchikian, J. Am. Chem. SOC., 1991, 113,8948. 24 J. N. Butt, J. Niles, F. A. Armstrong, J. Breton and A. J. Thomson, Nature Struct. Biol., 1994, 1,427. 25 J. L. Breton, J. L. C. Duff, J. N. Butt, F. A. Armstrong, S. J. George, Y. PCtillot, E. Forest, G. Schafer and A. J. Thomson, Eur. J. Biochem., 1995,233, 937. 26 J. L. C. Duff, J. L. J. Breton, J. N. Butt, F. A. Armstrong and A. J. Thomson, J.Am. Chem. SOC., 1996,118, 8593. 27 J. Tong and B. A. Feinberg, J.Biol. Chem., 1994, 269, 24920. 28 P. H. McPherson, M. Schonfeld, M. L. Paddock, M. Y. Okamura and G. Feher, Biochemistry, 1994, 33, 1181. 29 J. N. Butt, A. Sucheta, L. L. Martin, B. Shen, B. K. Burgess and F. A. Armstrong, J.Am. Chem. SOC.,1993, 115, 12587. 30 J. N. Butt, A. Sucheta, F. A. Armstrong, J. Breton, A. J. Thomson and E. C. Hatchikian, J. Am. Chem. SOC., 1993, 115, 1413. 31 M. S. Mondal, H. A. Fuller and F. A. Armstrong, J. Am. Chem. SOC., 1996,118,263. 32 J. Hirst, A. Sucheta, B. A. C. Ackrell and F. A. Armstrong, J.Am. Chem. SOC.,1996, 118, 5031. 33 A. Sucheta, B. A. C. Ackrell, B. Cochran and F. A. Armstrong, Nature, 1992,356,361. 34 T. Ikeda, S. Miyoaoka and K. Miki, J. Electroanal. Chem., 1993,352, 267. 35 J. Hirst, B. A. C. Ackrell and F. A. Armstrong, J.Am. Chem. SOC., 1997, 119, in press. Received, 15th January 1997 Accepted, 14th March 1997 Chemical Society Reviews, 1997, volume 26 179

ISSN:0306-0012    

DOI:10.1039/CS9972600169

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Electrochemistry for a cleaner environment Daniel Simonsson Department of Chemical Engineering and Technology, Applied Electrochemistry, The Royal Institute of Technology, S-100 44 Stockholm, Sweden This article reviews electrochemical processes and devices that can contribute to a cleaner environment. Electro- chemical processes for treatment of waste water solutions, flue gases and contaminated groundwater and soil are described, as well as improvements of existing electro- chemical processes or products in order to minimize their environmental impact. Electrochemical power sources for cleaner generation of electricity in fuel cell power stations and for electrically driven vehicles are also discussed. Finally, the important role of electrochemical sensors for monitoring toxic substances is stressed.1 Introduction With a rapidly growing world population and an increasing number of reports on detrimental effects on the environment, its protection has become a major issue and a crucial factor for future technological progress, which will have to meet the requirements for sustainable development. The strategies for environmental protection in industry generally include proc- esses for waste treatment as well as development of new processes or products which have no or less harmful effects on the environment. Electrochemistry has important roles to play in both types of strategies. Electrochemical processes can be used for recovery or treatment of effluents from industrial or municipal plants.Industrial electrochemistry has undergone a development towards cleaner processes and more environ-mentally friendly products. Electrochemical sensors are effec- tive and inexpensive devices for environmental monitoring of an increasing range of toxic substances. A big and important class of environmental problems can be found in the energy and transportation sectors. Electrochemistry offers unique ways to generate pure electric power at high effeciency in fuel cells or to store it in batteries. These power sources may be used in stationary power systems or in electrically driven vehicles. The potential of electrochemistry for environmental protec- tion has been reviewed in many journal articles and monographs (see e.g.ref. 1,2) since the first book devoted to the topic appeared a quarter of a century ago.3 Many of the concepts discussed in that early publication are still of interest today, but progress in materials science (new electrode materials and Daniel Simonsson was born in Arboga, Sweden, in 1943. He received his Master of Science degree and his PhD from the Royal Institute of Technology (KTH) in 1968 and in 1973, respectively. Since 1989, he has been professor ofApplied Elec- trochemistry at KTH. His re-search interests include bat-teries, fuel cells, electrolytic processes and electrochemical environmental engineering. membranes) and electrochemical engineering (new and more effective cell designs) have offered new and more effective solutions.This review aims to provide the same broad perspective on the possible electrochemical future as in ref. 3 by describing electrochemistry being used both in the electrolytic mode (electrolytic processes for purification and recovery) and in the galvanic mode (for electrical energy production and storage in fuel cells and batteries, respectively) but focuses mainly on recent developments. 2 Electrochemical processes for waste treatment Electrochemistry, with its unique ability to oxidize or reduce compounds at a well-controlled electrode potential and by just adding (at the anode) or withdrawing (at the cathode) electrons, offers many interesting possibilities in environmental engin- eering. Anodic processes can be used to oxidize organic pollutants to harmless products and to remove toxic compounds from flue gases.Cathodic processes using effective cell designs can remove heavy metal ions from waste water solutions down to very low outlet concentrations. In both types of electrode processes, the operating conditions must be carefully controlled in order to avoid side reactions. In aqueous solutions, which are most often used, the side reactions are mainly oxygen evolution at the anode and hydrogen evolution at the cathode. These side reactions lower the current efficiency (i.e. the proportion of the total current used in the required rather than loss reactions), thereby increasing the operating costs, and may disturb the process because of vigorous gas evolution or pH changes at the electrodes. Not only can the two electrodes of the electrochemical cell be used in purification processes, but the ion-selective mem- brane(s) that are often placed between the electrodes to have a selective transfer of only anions or cations can also.New electrodialytic processes using such membranes have been developed, which can solve a variety of environmental problems. 2.1 Anodic oxidation of toxic species High electrode potentials can conveniently be set for oxidation of toxic compounds at the anode of an electrochemical cell. The upper limit is set by the stability of the anode material and the onset of side reactions such as oxygen evolution and, in chloride solutions, chlorine evolution, which lower the current effi- ciency.Therefore, the anode material must have a high overpotential for the oxygen evolution reaction. Lead dioxide is well known to possess this property and has been extensively used in oxidations at high potentials. Dimensionally stable anodes of the types used in chlor-alkali and water electrolysis processes do not give high oxygen overpotentials. Platinized titanium is better in this respect but is expensive. On the other hand, doped Sn02-coated titanium electrodes have been developed4 that give significantly higher oxygen overpotentials than both lead dioxide and platinum. Using the oxidation of phenol as a test case showed that the rate of phenol removal was much higher for Sn02than for Pb02 and Pt. Similar results have been obtained for the oxidative treatment of biologically refractory waste water.5 The efficiency (typically between Chemical Society Reviews, 1997, volume 26 181 3040%)was about five times higher than on a platinum anode.In chloride containing media, less chlorine was produced than with platinum anodes. No interference with the cathode was found, which indicates irreversible oxidation to reaction products that cannot be reduced at the cathode. The energy requirement of electrochemical oxidation of organics in waste water could thus be reduced to 30-50 kW h kg-1 of COD (Chemical Oxygen Demand) removed. 2.1 .I Direct anodic oxidation Within the electroplating and surface finishing industry, large amounts of cyanides are still used to obtain very finely grained metal deposits. The waste water must be purified with respect to its cyanide content.The conventional chemical method is to oxidize the cyanide to the less harmful cyanate using hypoch- lorite as an oxidizing agent. An electrochemical process, in which the cyanide is oxidized anodically, is economically more feasible at higher concentrations of cyanide (> 1 g dm-3). At the cathode, the main part of the metal ions may simultaneously be deposited, with hydrogen evolution as a side reaction. The overall reaction at the anode, which is generally made from graphite or stainless steel, is in a first step [eqn. (l)]. The CN-+ 20H--+CNO-+ H20 + 2e-(1) cyanate then reacts further in the alkaline solution to Na2C03, N2, NH3 etc.Oxygen is evolved in a side reaction. The temperature is ca. 50-90 "C, and the current density ca. 500 A m-2. The energy demand is 10-40 kW h kg-1 cyanide. The remaining cyanide in the outlet stream is at low concentration and has to be oxidized by hypochlorite. In order to achieve lower residual concentrations and also a better process economy, sodium chloride may be added to the electrolyte so that chlorine is formed at the anode. Hypochlorite, which is formed in the solution, can then react with cyanide ions homogeneously in the solution. This is an example of indirect or mediated electrochemical oxidation. Further examples of this principle will be shown below. Alternatively, the efficiency of the direct electrooxidation process can be improved by using three-dimensional electrodes of the type that will be discussed later.The treatment of phenol has already been mentioned above with respect to the choice of anode material. Comninellis and Pulgarin6 have studied this process using Pt and Sn02 anodes and proposed a reaction scheme that involves the initial formation of hydroxyl radicals by the electrooxidation of water. These radicals are adsorbed on the electrode surface and react on an Sn02anode with phenol to carbon dioxide, or react to oxygen in a side reaction. On this anode material aromatic intermediates were formed only in very small concentrations, while they were formed in large concentrations on Pt. The rate of phenol removal is almost the same for both anodes, while the Ti/SnO2 anode gives a much higher rate of total organic carbon (TOC) removal.Chlorinated hydrocarbons are used as solvents and disinfec- tants. They may also appear as products in chemical reactions between chlorine and dissolved hydrocarbons in various waste streams, e.g. chlorinated phenols from bleaching in the pulp and paper industry. Combustion is a frequently used destruction method but is not suitable for the treatment of diluted waste waters, because of the high cost of transportation, the high consumption of fuel and, in some cases, corrosion problems. Other treatment options could involve membrane separation, adsorption on activated carbon and stripping, chemical oxida- tion with air, ozone or other oxidants.Chemical reduction techniques, such as catalytic dehalogenation with hydrogen or other reducing agents, are also used. Biological techniques using special microorganisms or enzymes are under develop- ment. Electrochemical dechlorination is an alternative that can be employed either anodically or cathodically. Wabner et al.7 have shown that p-chlorophenol and penta- chlorophenol can be destroyed anodically on lead dioxide 182 Chemical Society Reviews, 1997, volume 26 anodes. The decomposition of p-chlorophenol follow concen- tration-time curves that are similar to those of phenol. Therefore, the mechanism seems to be similar in the two cases involving hydroxyl radicals formed from water. In the first step chlorine is substituted by these radicals.In subsequent steps further oxidation yields quinone, which decomposes into maleic acid, oxalic acid (primarily) and carbon dioxide. Oxygen and significant amounts of ozone were formed as by-products at the anode. There is a risk that the chloride ions formed may be oxidized to hypochlorite, which can then form chlorinated organic compounds with low molecular mass species formed in the reaction sequence. An interesting alternative is to use the dechlorination step only and decompose the less toxic phenolic substances biologically . The electrochemical treatment may also be used for de- chlorination of other chlorinated aromatic compounds in aqueous solutions, such as hexachlorobenzene, PCBs and tetrachlorodibenzodioxin.2 .I.2 Mediated electrochemical oxidation It has already been noted that dilute alkaline cyanide solutions can be oxidized homogeneously by anodically formed C10-. While C10- is an efficient oxidant, it has the disadvantage that it forms toxic chlorinated compounds. Other examples of mediators are Ag", Co"' and Fe"1 for oxidation of organic compounds. The standard potentials vs. SHE are 1.987, 1.842 and 0.77 V, respectively. The higher the redox potential, the higher is the coulombic efficiency. A process using Ag" has been developed by AEA Technol- ogy in Dounreay, Scotlandg. AgI is oxidized anodically to Ag" on platinized titanium in concentrated nitric acid in a mem- brane-divided cell. Some Ag" may oxidize the organic compounds directly in the anolyte, but the main route is believed to be the reaction of Ag" with water to form radicals, which then react with organics.AgI1 is probably stabilised in the solution as a nitrate complex, AgN03+. In the cathode chamber, which is separated from the anode chamber by means of a cation selective membrane, nitric acid is reduced to nitrous acid, which is reoxidized to nitric acid by air in a separate reactor. Water is recycled to the anode chamber to replace water lost due to electroosmosis through the membrane. The overall effect is that dissolved organics are converted to C02 and H20 by oxidation with air. The process has been shown to successfully destroy a variety of organic compounds, including benzene, phenols, oils and chlorinated organic compounds.2.2 Cathodic processes2.2.1 Cathodic Electrochemical Dechlorination Dechlorination of organic toxicants can occur not only anodically, as discussed above, but also cathodically, according to the overall reaction in eqn. (2). Hydrogen evolution is a RCl + H+ + 2e--+ RH + C1-(2) competing reaction which decreases the current efficiency. Schmal et al. 10 have evaluated the feasibility of electrochemical reduction of waste waters containing halogenated organic compounds. As electrode material, they used thin graphite/ carbon fibres which give a high specific surface area and have a reasonably high overpotential for the competing hydrogen evolution reaction. Their experiments showed that it was possible to remove all chlorine atoms from the organic molecules in aqueous solution.The current efficiency was low, of the order of 1%. As the organic substrate concentration was low (ca. 100 ppm), the energy consumption per cubic metre was still acceptable (ca. 10-100 kW h m-3).The dehalogenation results in a decreased toxicity and an increased biodegrad- ability, thus enabling further biological treatment. They con- cluded that energy consumption and conversion rates are such that a technically and economically viable method for the detoxification of waste waters can be developed. The total cost was estimated to be of the same order of magnitude as that of adsorption on carbon. In contrast to the latter method, electrochemical reduction can be used with polar compounds and does not give a concentrated toxic waste that has to be treated further or dumped in special depots.Cathodic destruction of monochloroethane, trichloroethane, epichlorhydrin11 and of aromatic halides (models of dioxin)'2 has also been demonstrated. 2.2.2 Removal of heavy metal ions Waste waters containing heavy metal ions are generated in metallurgical and electroplating industries and in the manu- facture of printed circuit boards. The conventional purification of these streams uses hydroxide precipitation, which gives a voluminous metal hydroxide sludge that has to be disposed of in an environmentally acceptable way. With complexed metal ions in alkaline solutions, hydroxide precipitation is not a viable method.Cathodic removal of heavy metal ions from different waste waters is an attractive alternative process, since the metal can be recovered in its pure metallic form or as a concentrated solution that can be recycled or allow the extraction of the pure metal in an electrowinning process. The basic problem is that the mass transport controlled limiting current density for cathodic metal deposition from dilute solutions is so low that the conventional use of planar electrodes would require too large a surface, unless the mass transfer rate could be substantially improved. This might be achieved in various ways, e.g. with a rotating cylinder cathode or using expanded metal mesh electrodes immersed in a fluidized bed of small glass beads.The more efficient electrochemical engineering solution is the packed bed elec- trode, or, more generally, three-dimensional electrodes, Fig. 1. Purified Anolyte Solution outlet Three-dimensioi cathode 1 tt Dilute waste solution ~~~lyt~ Fig. 1Cell with three-dimensional cathode for heavy metal removal This wider concept also includes fluidized bed and circulating particulate-bed electrodes. However, most three-dimensional electrodes being used for metal removal are of the fixed bed type. The anode may also be a three-dimensional electrode or just a planar electrode for e.g. oxygen evolution. As cathode material, graphite particles, expanded metal, graphite felt, metal wool, graphite fibres and reticulated vitreous carbon have been examined.The engineering design has to consider the potential drop in the dilute electrolyte along the depth of the electrode in the direction of current flow, since this determines the extent of the side reaction, hydrogen evolution.13 If the bed is too thick, operation at the limiting current density across the whole depth of the electrode would give too a high hydrogen evolution rate in the part of the bed facing the membrane or diaphragm, which separates the cathode from the anode. Three-dimensional electrodes are effective in the treatment of dilute solutions since they offer both a high specific surface area and high mass transport rate conditions. In electrodes of this type the metal concentration can be reduced from, say, 100 to 0.1 ppm at a residence time of a few minutes.14 Operational costs are favourable compared with classical waste water treatment systems.In some cases the removal efficiency is higher, and the space required by the process is low. The deposited metal in the cathode may be recovered as a concentrated solution by chemical dissolution with e.g. an acidified solution of hydrogen peroxide, or by switching the polarity of the cell. In the latter case, the cell will function as a concentrator. At the anode, the metal deposited during the preceding cathodic phase is dissolved into a concentrated solution. This method requires robust anion-selective mem- branes to prevent leakage of metal ions from the concentrated anolyte solution to the dilute catholyte solution.In the best case, the concentrated metal ion solution can be recycled to the process which generated the waste solution. An alternative regeneration procedure is to use three-dimensional electrodes made of e.g. foamed graphite or carbon felt as cartridges that may be removed to recover the deposited metal in an external procedure, for example, according to the principles of electro- refining. Processes utilizing packed-bed electrodes have been applied in the chemical industry mainly to waste solutions containing Cu and Hg.11 Outlet concentrations of less than 1 ppm can generally be achieved. The energy consumption is of the order of 1 kW h m-3. For less noble metals, such as Zn and Cd, the side reaction of hydrogen evolution increases and the current efficiency is low, especially at lower concentrations.Therefore, it is not possible to decrease the concentration of these species below 1 ppm at an acceptable overall current efficiency. A further complication is that excessive cathodic hydrogen evolution may increase pH, so that the metal precipitates as hydroxide. This phenomenon can be used to advantage to remove CrVI as Cr(OH)3. 2.3 Electrodialytic processes Cation exchange membranes, introduced commercially in the membrane cells for chlor-alkali production, can also be used in electrochemical processes for environmental protection. Anion selective membranes, which allow only anions to pass, can be used in cells with three-dimensional electrodes for metal deposition as described above.Acid can be recovered from spent pickling baths by using anion-selective membranes with oxygen evolution at a suitable anode. In electrodialytic processes the membranes themselves do the main part of the work by splitting the inlet flow into one more concentrated and one more dilute stream. Electrodialysis has long since been used for salt or brine production and in desalination processes for obtaining fresh water. It can also be used in combination with electrolysis for desalinating waste streams. Aqueous streams containing e.g. NaCl and Na2S04 are generated in many chemical processing operations such as flue gas scrubbing, metal pickling, fermentation and rayon manu- facture.One important example is splitting of sodium sulfate solutions into sodium hydroxide and sulfuric acid solutions. The simplest design for this is to use a three-compartment cell with cationic and anionic selective membranes, as shown in Fig. 2. In such a cell hydroxide concentrations of ca. 10 mass% can be achieved. The cell voltage and thus the energy demand is to a great extent determined by the anode and cathode potentials. The energy requirement can be reduced by using, instead of the oxygen evolving anode, a hydrogen gas diffusion electrode of the type used in fuel cells on the anode side and utilizing the hydrogen evolved at the cathode. This concept can also be used in a two-chamber process,l5 as shown schematically in Fig. 3. Chemical Society Reviews, 1997, volume 26 183 Fig.2 Three-compartment cell for splitting of sodium sulfate into sodium hydroxide and sulfunc acid solutions. AM = anion selective membrane, CM = cation selective membrane. However, in this case a pure sulfuric acid solution is not obtained as in the three-compartment cell. The energy con- sumption is 1600-1 800 kW h tonne-' caustic at a concentration of 13-18%. Energy savings may also be achieved by using bipolar ion exchange membranes, which consist of threee parts: a cation selective region, an anion selective region and the interface between the two regions. When a direct current is passed through the bipolar membrane with the cation selective side toward the cathode, electrical conduction is achieved by the transport of H+ and OH- ions which are obtained from the dissociation of water. A high efficiency requires that the membrane has a high water permeability to provide water from the external solutions to the interface, and a very thin interface between anion and cation regions to allow efficient transport of H+ and OH-.Such a membrane can be combined with anion and cation selective membranes into three-compartment units, which can be repeated a large number of times into a compact stack. Acidified Sulfate Solution NaOH Separator;l--L;I I Diffusion anode Sulfate Solution Fig. 3 Two-chamber cell with gas diffusion anode in which the hydrogen gas evolved at the cathode is oxidized The gas diffusion anode consists of a porous sheet containing a catalytic layer, which is pressed agamt a cation- exchange membrane on the electrolyte side and a suitable porous current collector on the gas side The electrochemical reaction occurs at the membrane/catalytic layer interface, where hydrogen gas, electrolyte and active electrode matenal can form three-phase contact zones After ref 15.The energy requirement for producing sodium hydroxide is in the range 1300-2000 kW h tonne-' at 1000 A m-2 and a 184 Chemical Society Reviews, 1997, volume 26 current efficiency of 80%. This energy demand is lower than that for the conventional chlor-alkali processes. Therefore this technique may also be used to produce sodium hydroxide without the simultaneous generation of chlonne.2.4 Electrochemical remediation of soils There is a great need for cost-effective methods for restoration of contaminated soils. The use of electrochemistry for this purpose simply means putting electrodes in the soil and applying a voltage over them, as shown schematically in Fig. 4. The groundwater or an externally supplied fluid is used as electrolyte. This technique has been reviewed by Acar and Alshawabkeh16 and Probstein and Hicks.17 Fig. 4 Schematic illustration of electrochemical soil remediation The main electrode reactions are generally oxygen evolution at the anode and hydrogen evolution at the cathode, eqn. (3) and (4). H20 + 1/2 02 + 2H+ + 2e-(3) H20 + 2e-+Hz + 20H-(4) As in ordinary electrolysis, ions will move through the soil due to migration, diffusion and convection.If heavy metal ions are present in the ground water they will move to the cathode and become electrodeposited on its surface. The method is not restricted to ionic contaminants but can also be used to extract organic compounds by means of the electroosmotic flow that is generated by the electnc field in fine-grained soils with pores that are of micrometre-size or smaller. Thus, in contrast to forced convection by means of pumping, which will make the pore liquid flow preferentially through the larger pores, electroosmotic flow will be effective in smaller pores. For ions, mass transfer by migration will generally be much higher than that by electroosmosis. The technique can be applied zn sztu, on-site and off-site.The flow around the electrodes and through the soil can be controlled so that the contaminating species in the fluid may be removed either at the electrodes or in an external extraction system (e.g by ion exchange or chemical precipitation). This also allows for a control of the chemical conditions. Without external control of the chemistry of the system, the electrode reactions are generally oxygen evolution at the anode and hydrogen evolution at the cathode according to eqns. (3) and (4).These reactions will make the anode region acidic and the cathode region basic. An acid front will thus move from the anode towards the cathode, while an alkaline front moves in the opposite direction. The penetrating acid solution will release heavy metals and other cations that are sorbed on negatively charged clay surfaces, the cations migrating towards the cathode.The alkaline front causes precipitation of most heavy metals and radionuclides, unless they form negatively charged complexes. These complexes migrate towards the anode and will form free metal ions again, when they meet the acid front. In this way species are concentrated in the region where the two pH-fronts meet and pH changes abruptly. The above problems may be solved by means of purging solutions. For example, the hydroxide precipitation at some distance from the cathode may be avoided by adding an acid solution to the cathode region. Electrokinetic remediation has been tested successfully in field studies in The Netherlands and in the USA. Estimated costs for the technique17 indicate that they may be several times lower than for conventional methods.2.5 Electrochemical gas purification The processes discussed above all deal with pollutions present in aqueous solutions. Gaseous pollutants may also be removed electrochemically, provided that they are first dissolved into an electrolyte. The overall process will then generally consist of at least two steps: absorption of the gaseous species in a liquid and the subsequent electrochemical conversion of them to less harmful products. The two steps can be integrated into one device or are separated in two different devices,18 as shown schematically in Fig. 5.The rate and capacity of absorption will Purified gas Stream Purified gas Stream -+ f Polluted gas Stream Polluted gas Stream Fig. 5 Electrochemical gas purification using integrated absorber/electro- chemical cell (left) or separate absorption column and electrochemical cell (right). After ref. 18. be increased, since the physical absorption process will not be limited by the equilibrium solubility of the absorbed gaseous pollutant. Examples include reduction of chlorine to chloride, oxidation of nitrous oxides to nitric acid and of sulfur dioxide to sulfuric acid. The reduction or oxidation may occur directly at the electrode or indirectly via a redox mediator. In the latter case, the mediator has to be regenerated electrochemically. The modified Mark 13A process19 is an example of an indirect electrochemical process, which uses bromine as a mediator to oxidize SO2 chemically, eqn.(5). SO2 + Br2 + 2H20 +H2S04 + 2HBr (5) The hydrogen bromide formed is transferred to an electro- chemical cell in which bromine is regenerated electrochemic- ally according to the overall reaction (6). 2 HBr +Br2 + H2 (6) Sulfur dioxide can also be removed electrochemically at higher temperatures in a cell with a molten electrolyte, which essentially concentrates the sulfur dioxide for further treat- ment.20 At the cathode SO2 is reacted with oxygen to sulfate ion, which migrates to the anode, where it is oxidized to SO3 and 02. In a similar way hydrogen sulfide, which is hazardous to plant and animal life even at the ppm level, can be removed as elemental sulfur according to the overall reaction20 in eqn.(7). H2S +Hz + 1/2 S2 (7) 3 Improved electrochemical processes and products Electrochemical technology has traditionally relied to a sig- nificant extent on heavy metals. The electrochemist is aware of the great importance of mercury in both fundamental and industrial electrochemistry. From a technical point of view, mercury with its high overvoltage for the hydrogen evolution reaction has offered practical solutions both for the production of chlor-alkali and against self discharge and gassing of primary batteries. The most widely used battery, the lead-acid battery is the main consumer of lead.Similarily, poisonous cadmium is used in nickel-cadmium batteries. Previous efforts to avoid or minimize the environmental impact of heavy metals and to meet the increasingly strict environmental regulations have been successful in many cases. Not so many years ago, alkaline batteries were a problem due to their mercury content, present to prevent the hydrogen evolving during self discharge of the zinc anode. However, fairly soon ‘green’ batteries were on the market with no or very little mercury thanks to the rational application of well known principles for corrosion inhibition, including very pure elec- trodes and electrolyte solutions. Unfortunately, the situation is not that simple for lead-acid batteries, of which more than 100 million per year are sold.Most of them are used in cars, an application in which they are difficult to replace. The key word is recycling. Batteries are collected and the lead is recovered. Similarly nickel-cadmium batteries are collected and the cadmium recovered. In this case an alternative is available, which is more environmentally compatible: the nickel-metal hydride battery. This is being used to an increasing extent in new high tech products requiring cordless power, e.g. cellular phones, camcorders and notebook computers. The present status of aqueous rechargeable batteries has been reviewed in ref. 21. Rechargeable lithium batteries represent a further improvement for these applications, both with respect to energy density and environmental impact.22 In the electrolytic industries the losses of mercury from the mercury process for production of chlor-alkali have been reduced significantly during the last few decades, but new processes of this type will generally not be allowed in Western countries.Diaphragm processes are not a viable alternative, since the the state-of-the-art diaphragms are made from asbestos, which is a health hazard. New processes are instead based upon the use of ion-selective membranes. The perfor- mance of these processes is constantly being improved. They also offer a higher energy efficiency than the two other processes. The increased efficiency is a result of improved membranes, effective catalytic coatings on the electrodes and improved cell design, which contribute to a lower cell voltage.One of the two products in the chlor-alkali process, chlorine, has received increasing concern with respect to its impact on the environment. In some countries a major fraction of the chlorine produced has been used as a bleaching agent in the production of pulp and paper. However, in the late 1980s it was found that dioxins are formed during pulp bleaching with chlorine. Regulations and consumer demand have lead to a decreased demand for chlorine in bleaching, and the use of chlorine in the pulp and paper industry is steadily decreasing. However, there is still a need for the caustic produced at the cathode for the cooking process in the pulp manufacturing process. Therefore it is of interest to develop electrochemical processes that still produce caustic in the cathode chamber but do not produce chlorine at the anode.One example is oxygen evolution or hydrogen oxidation as the anode reaction, using sodium sulfate instead of sodium chloride as electrolyte. This concept has Chemical Society Reviews, 1997, volume 26 185 already been discussed above as a method to purify waste solutions from sodium sulfate. 4 Electrochemical power sources for cleaner electrical energy Thermal combustion of fossil fuels in power plants and vehicles is a major environmental problem in modern society. The immediate damage of air pollution has been estimated to cost about three times more than the fossil fuels themselves. Long- term effects, such as global warming, are not included. The most important gaseous impurities in the flue gas from electricity generation plants are C02, NO,, SO2 and dust particles.C02 is a major contributor to the greenhouse effect and NO, contrib-utes to the acidification of water and soil, eutrophication, and the formation of smog. 4.1 Batteries for electric vehicles Approximately 20% of the primary energy used in the European Union goes to transportation. Road traffic alone generates more than 50% of the total emissions of nitrogen oxides, carbon monoxide and hydrocarbons. The authorities in California, where the transportation fleet is responsible for more than 75% of the air pollution in the Los Angeles basin, have enacted a law that requires that 10%of all new vehicles sold in the state must be so called zero emission vehicles by the year 2003.The only vehicles that are likely to meet these demands are electric vehicles. In order to meet these new regulations ‘The Big Three’, General Motors, Ford and Chrysler in the USA decided in January 1991 to form a consortium, The United States Advanced Battery Consortium (US ABC), for cooperation towards improved power sources for electric vehicles. The joint research is also funded by the Department of Energy and EPRI. The total budget is $260 million for five years. The goals that have been set by the US ABC are shown in Table 1. These performance requirements are compared in the same table with some battery systems of special interest for electric vehicles and load levelling.Table 1 US ABC criteria and performance of selected battery systems Specific energy/ Peak specific Deep-discharge Couple W h kg-’ power/W kg-’ cycles US ABC criteria 80- 100 150-200 600 US ABC, long 200 400 1000 term Pb/Pb02 35-40 150-300 100- 1000 CdINiOOH 50 80-150 1000 MHINiOOH 60-80 200 750 Znlair 100-1 20 150 < 300 N a/S 100 230 760 Na/NiC12 80 I30 1200 Li/pol ymer 150 400 < 100 (projected) LiAl/FeS2 (cells) 18CL200 > 200 1000 Li-ion 80-90 200-300 400-1200 (cells) Some of the systems that have been proposed to meet the mid-term goals include sodium-sulfur, sodium-nickel chloride, lithium-metal sulfide, lithium-polymer and lithium-ion bat- teries, metal hydride-nickel oxide, zinc-air and zinc-nickel oxide.None of these yet meet the long-term goals. Electric Fuel Limited in Israel has developed a refuelable zinc-air energy system for powering electric vehicles.23 In order to recharge the zinc anode rapidly, their system solution includes a refueling system for mechanically exchanging zinc anode cassettes and a regeneration system for recycling depleted cassettes. Zinc oxide is dissolved in a KOH solution, which is then fed into an electrowinning bath where zinc is deposited, collected and later reassembled into fresh cassettes. This system is being tested by Deutsche Post with Mercedes- Benz Vans and GM-Ope1 Corsa Combo Vans. The driving range is claimed to be over 300 km. The overall advantages of electric vehicles with respect to the environment must be considered with respect to the total system design.The electric vehicle itself emits almost no pollutants. On the other hand, the electricity generating plants delivering electric energy to recharge the vehicles give additional pollution if they are based on fossil fuel combustion. However, these facilities are generally located outside the cities, and the control of their emissions is also easier and more cost effective. The efficiency of fuel utilization is also higher. When the electricity is produced by nuclear reactors, windturbines or solar cells the gaseous pollutants are negligible. Electric vehicles will pri- marily be recharged during off-peak hours, minimizing the need for increased utility capacity.Even with an electricity generating mix which is highly coal orientated, such as in the US, a substantial reduction of air pollutants can be achieved in comparison to the standard gasoline driven car. EPRI estimates that together with new clean coal technologies it is possible to achieve reductions of 99% of volatile organic compounds (VOC), 99% of CO and 83% of NO,. Batteries are still too expensive and do not meet the requirements with regard to energy and power densities for electric vehicles, but the intensive research that is going on worldwide will hopefully lead to a breakthrough for the electric vehicle in the future. 4.2 Electrochemical capacitors Batteries can be used in all-electric vehicles or, alternatively, in hybrid vehicles that use also combustion engines for propulsion. The batteries are then mainly used for acceleration and for citydriving, while the combustion engine gives a reasonable range.In this application, a high peak power density of the battery is a major requirement, while the energy density, which determines the all-electric range, is less important compared to all-electric vehicles. An interesting alternative, or complement, to batteries is electrochemical capacitors (also called ultra- capacitors or supercapacitors), which can give peak power densities greater than 1 kW kg-l while the energy density is only 2-10% of that stored in a battery to deliver the same power.24 An electrochemical capacitor stores the electrical energy electrostatically by charging of the electrochemical double layer at the electrode/electrolyte interface.In some systems inter- mediates are adsorbed on the electrode surface or intercalated into the electrode material, which gives an additional so called pseudo-capacitance that may be 10 to 100 times higher than the double layer capacitance.25 In a finely porous electrode with a high specific surface area, fairly high amounts of electrical energy may be stored per unit volume or mass. Research and development has been going on since the early 1990s to develop ultracapacitors using various types of carbon, doped conducting polymers, and metal oxides as electrode materials. The electrolyte may be aqueous, organic or a solid polymer. Ultracapacitors with aqueous electrolyte can store 1.5 W h kg-1 and deliver 1 kW kg-1, while the best values reported for devices using an organic electrolyte are 5-7 W h kg-l and 2 kW kg-1.Future development is expected to result in ultracapacitors with an energy density of 10-15 W h kg-1 and a power density of 3-4 kW kg-1 in the near-term.Z4 With increasing energy density the possibility increases of using only electrochemical capacitors as storage device for electrical energy in hybrid electric vehicles. 4.3 Fuel cells The fuel cell represents another major challenge for electro- chemists, electrochemical engineers and material scientists. Fuel cells are of great interest because of their efficient conversion of the chemical energy in fuels to electricity, as well as the very small pollutions.Therefore atmospheric pollution 186 Chemical Society Reviews, 1997, volume 26 would be lowered and the energy sources used in a more efficient way, if fuel cells were used in stationary power stations and in electrically driven vehicles. Recent monographs on fuel cells are available.26, Z7 The principle of the fuel cell was demonstrated more than 1SO years ago but did not receive significant attention until the American space program in the sixties. In a fuel cell, a fuel (generally hydrogen) is oxidized at the anode, and an oxidant (oxygen in pure form or in air) is reduced at the cathode. The overall reaction in this 'electrochemical combustion' is thus the same as at thermal combustion. The principle is illustrated schematically in Fig.6 for the case of the hydrogen-oxygen fuel cell. H2 -I, Cathode Anode -+2H20 2H2 -+4H++ Electrolvte f Fig. 6 The working principle of a fuel cell with acid electrolyte The advantage of the fuel cell is that chemical energy can be converted directly to electric energy. For a reversible electro- chemical reaction this is expressed in the well-known ther- modynamic equation (8). AG = -nFE,. In a thermochemical process the heat of combustion ( = -AH) is utilized in a cyclic process, the maximum efficiency of which is limited by the Carnot-efficiency (ET = 1 -T1/T2). In fuel cells, on the other hand, the ideal efficiency is given by eqn.(9). From thermodynamic data, we can calculate the ideal efficiency of the hydrogen-oxygen fuel cell to be 83% at 25 "C. The real efficiency is lower because of the overpotentials and ohmic potential losses in the electrolyte, which mean irreversible losses. The cell voltage in working fuel cells is typically ca. 0.7-0.9 V at current densities of 1-5 kA m-2. With E = 0.8 V for a hydrogen-oxygen fuel cell the real efficiency at 25" C is S4%, which is still considerably higher than the efficiencies that can be reached in conventional thermal cycles. A major part of the energy losses is due to the sluggishness of the oxygen reduction reaction at the cathode. Therefore an important issue in the development of fuel cells is the search for efficient and sufficiently cheap electrocatalysts for this electrode reaction.An alternative method is to reduce the overpotentials by operation at higher temperatures, using a molten salt or solid oxide as electrolyte. This introduces other material problems such as corrosion and mechanical stresses. The ideal fuel cell would use hydrocarbons available from their natural sources directly. Even with platinum as catalyst, the direct anodic oxidation of hydrocarbons is a very slow electrode process, which requires a high overpotential. The only fuel which is practically useful today is hydrogen, and to some extent carbon monoxide in high temperature fuel cells. Apart from applications in which fuel cells utilize hydrogen gas obtained as a side product in chlor-alkali and chlorate processes, the first step is to produce hydrogen from suitable primary fuels.In practice this is done by means of partial oxidation and steam reforming of coal, oil or natural gas. Gasification and reforming of biofuels such as wood and peat is also possible. Therefore, to a large extent, a fuel cell power station will be a chemical factory for up-grading of the primary fuel, as shown in Fig. 7. An important aspect is the purity of the gas, since common impurities such as CO and H2S poison the catalyst. Heat to heating, Heat heat engines t Fig. 7 Schematic drawing of a fuel cell power station It is customary to classify fuel cells according to the electrolyte used. Fuel cells under development are alkaline fuel cells (AFC), phosphoric acid fuel cells (PAFC), proton exchange membrane fuel cells (PEMFC), molten carbonate fuel cells (MCFC) and solid oxide fuel cells (SOFC).Characteristic performance data for these different system are shown in Table 2. The fuel cell which is closest to commercialization is the PAFC, which uses concentrated phosphoric acid as electrolyte and works at 160-200 "C. This type of fuel cell has been tested in Japan in an 11 MW power plant. Since natural gas is generally used as the primary energy source, the methane in this gas must first be converted to hydrogen through steam reforming (CH4 + H20 +3 H2 + CO) and shift (CO + H20 + H2 + C02) reactions. The overall energy efficiency of a power plant using PAFC is ca.40%. PAFC is intended for on-site integrated energy systems to provide electricity and heat for space heating and hot water. PAFC has also been considered for propulsion of electrically driven buses in the US. The molten carbonate fuel cell, MCFC, operates with a carbonate melt as electrolyte at about 650 "C. At this high temperature sufficiently low overpotentials on the electrodes can be obtained without the use of expensive noble metal catalysts. The waste heat is also more valuable at the higher temperature, so that higher overall efficiencies can be obtained, especially if the heat produced is used for the reforming process. Estimates have been made that this concept can give an overall Table 2 Characteristic performance data for different fuel cells AFC PAFC PEMFC MCFC SOFC Temp./"C 60-120 180-210 80-100 60&700 900-1000 Anode fuel H2, high purity H2 H2 H2-CO H2-CO Oxidant 02,high purity Air Air CH4 Air + C02 CH4 Air Electrolyte Anode catalyst Cathode catalyst KOH Pt Pt H3P04 Pt Pt Polymer Pt Pt (K,LW07 Ni Ni0 Y201, ZrOz Ni/ZrZOz La-Sr-Mn03 Chemical Society Reviews, 1997, volume 26 187 efficiency of up to 60% with internal reforming of the fuel gas to hydrogen.This is an impressive figure in comparison to conventional thermal engines. On the other hand, the competi- tion will be strong in the future from advanced technologies such as combined cycles, especially for power stations larger than 10 MW.The remaining advantages are high efficiency even in small-scale applications and the low pollution. A 2 MW MCFC power station has been started up in Santa Clara, USA in 1996. The solid oxide fuel cell is an all-solid-state fuel cell that operates at ca. 1000 "C. Yttria-stabilized zirconia is used as electrolyte. Strontium-doped lanthanum manganite and nickel- zirconia cermet are used as cathode and anode materials, respectively. The material currently used for the inter-connector is Mg or Sr-doped LaCr03. A major technical challenge is the fabrication of the component layers of the electrochemical cell stack. The thermal expansion coefficients of the four ceramic layers must be matched. Three different designs of SOFCs have been developed: tubular, monolithic and planar.Tubular SOFCs from Westinghouse Electric Corporation have been tested in 25 kW systems and are presently being built into a 100 kW power generation system. Proton exchange membrane fuel cells (PEMFC) were introduced in the Gemini space project. Following the general working principle of fuel cells, as shown in Fig.6, the working unit of a PEMFC is the membrane electrode assembly (MEA) consisting of a thin, catalytic gas diffusion anode for electro- chemical hydrogen oxidation, a catalytic cathode for oxygen reduction and a proton conducting membrane in between, forming the electrolyte as well as a separator for the two reactant gases. Platinum is used as electrocatalyst in both the anode and the cathode.The electrodes are supported by porous, inert materials. The solid electrolyte allows efficient sealing and safety of the fuel cell stack and permits the use of less corrosion- resistant construction materials. The performance as well as the cost of the PEMFC depend on the membrane and the utilization efficiency of the platinum catalyst. The high power density of the PEMFC, its simple and reliable construction as well as the possibility of rapid start-up even at low temperatures, make this fuel cell particularly suited as a power source for non-polluting vehicles. A vision of the electrochemical future would include fuel cells both for electricity production in stationary power stations and for transportation. The high-temperature fuel cell systems (MCFC and SOFC) with their high system efficiencies would be most suited for the latter application, but also PAFCs and PEMFCs are being developed for stationary power production.For transportation, the PEM fuel cell is generally the preferred choice. It allows rapid start-up from low temperatures and provides high power densities with values presently over 1 kW dm-3. The first fuel cell powered ZEV bus was put into traffic by Ballard Power Systems of Canada in Vancouver 1993 with PEM fuel cells as the power source. A second generation zero emission transit bus was on the road in 1995. The new PEMFC engine produces 200 kW (275 hp), the same power as the diesel engine normally installed in the bus. Furthermore, the fuel cell engine occupies the same space as the diesel engine and gives the same or superior performance.The power density of the new fuel cell stacks is double that of the stacks in the first bus. The compressed hydrogen gas cylinders are placed in the roof space and are sized to give a range of 400 km. Refueling is a matter of minutes.28 The choice of fuel is a key factor. Most fuel cells operate best on pure hydrogen and oxygen. The use of air instead of oxygen means some loss of performance. Pure hydrogen can be carried aboard a vehicle either as compressed gas, as in the Ballard bus, or as a cryogenic liquid. However, the volumetric energy density of pressurized hydrogen gas is poor, which means a short range. The energy density of liquid hydrogen is somewhat better but less than a third that of gasoline, and the technology for storing small volumes of liquid hydrogen is not well developed.Hydrocarbons provide a simple way to store hydrogen at high density. They can be converted to hydrogen by on-board reforming and shifting. The PEMFC can tolerate COZ in the fuel stream, thus allowing the use of reformed hydrocarbon fuel. Methanol is particularly interesting; it is a likely fuel to replace gasoline in the transportation sector, since it is easily integrated into the existing distribution system, and has about half the energy density of gasoline. To obtain hydrogen as a fuel to the fuel cell an on-board reformer must be used that allows rapid start-up and quick transient response.The long-term solution could be direct methanol fuel cells, in which methanol is fed directly to the anode of a PEM fuel cell. 5 Electrochemical sensors An important issue in environmental engineering is the monitoring of toxic compounds. Electrochemical sensors are convenient and effective devices for this purpose, since they produce an electric signal that can be related directly to the concentration of the compound being measured. One of the major achievements within air pollution control is the intro- duction of the lambda sensor for monitoring oxygen in the exhaust gas from vehicle combustion engines, which is a prerequisite for the modern three-way catalytic converter that reduces considerably the emissions of carbon monoxide, unburned hydrocarbons and nitrogen oxides.The solid electro- lyte in this sensor is of the same type as in the solid oxide fuel cell, stabilised zirconia. Electrochemical sensors have been reviewed in e.g. ref. 29. They can be used for sensing pollutants either as potentiometric, amperometric or voltammetric sensors. Ion-selective electrodes work according to the first method and can be used to determine for example pH, fluoride and cyanide concentrations in water. The concentration of toxic gases such as sulfur and nitrogen oxides can also be determined with potentiometric sensors. Amperometric sensors measure a current at a fixed potential, generally in the limiting current region. The current is then proportional to the concentration of the measured species.The mass transfer control is obtained by means of a porous barrier or a gas permeable membrane through which the gaseous species must diffuse in order to react on the sensing electrode. The Clark electrode for measuring oxygen concentration is the classic example.30 Its general principle also works for toxic gases like CO, NO, NO*, SO2 and H2S. A sensor can be made selective by a suitable choice of electrode potential and electrode material. An array of such selective sensors can be built into one device for monitoring flue gases and other gas streams containing several toxic components. 6 Photoelectrochemical methods Recent advances in photoelectrochemistry have led to new, interesting possibilities, both for treatment of pollutants and for conversion of solar energy from light to electricity.In the first case, suspensions of semiconductor particles can be used to harness the light with production of electrons and holes in the solid, which can destroy pollutants by means of reduction and oxidation, respectively. In this way, air or water containing organic, inorganic or microbiological pollutants can be effec- tively treated. These photocatalytic methods have been re-viewed recently.31 Photoelectrochemical cells for electricity production offer a sustainable way to generate electricity, e.g. for charging batteries in electric vehicles. With semiconductor electrodes using dye sensitized nanocrystalline Ti02 films an efficiency of 12% has been reported.32 Compared to conventional photo- voltaic cells, this type of photoelectro-chemical cell is less expensive, since it uses inexpensive raw materials, is easily 188 Chemical Society Reviews, 1997, volume 26 fabricated and does not require expensive crystal purification processes.7 Conclusions In this review, a variety of selected electrochemical processes and devices for environmental protection have been presented, which have been tested successfully on laboratory scale. Some, but not all of them, have also been tested at pilot scale and some have reached commercialization. There are many reasons why not all of them are yet commercial. In some cases, it is only a matter of time, further development work (and investment) being required.In other cases, chemical or biological processes are preferred because they are competitive and do not require expertise in electrochemistry and electrochemical engineer- ing. It may be expected that the number of electrochemical processes for treatment or prevention of pollution will increase in the future due to their specific advantages in a number of applications. A major beneficial impact of electrochemistry on the environment would be the future introduction of fuel cell or battery driven vehicles. References P. Tatapudi and J. M. Fenton, in Advances in Electrochemical Science and Engineering, ed. H. Gerischer and C. W. Tobias, VCH Verlagsge- sellschaft mbH, Weinheim 1995 vol. 4, 363. K. Rajeshwar, J.G. Ibanez and G.M. Swain, J. Appl. Electrochem., 1994, 24, 1077. Electrochemistry of Cleaner Environments, ed. J. O’M. Bockris, Plenum Press, New York, London, 1972. R. Kotz, S. Stucki and B. Garcer, J. Appl. Electrochem., 1991, 21, 14. S. Stucki, R. Kotz, B. Garcer and W. Suter, J. Appl. Electrochem., 1991, 21, 99. Ch. Comninellis and C. Pulgarin, J. Appl. Electrochem., 1993, 23, 108. D. Wabner, C. Grambow and A. Ritter, Vom Wasser, 1985, 64, 269. J. F. Rusling, Acc. Chem. Res., 1991, 24, 75. D. R. Craig, J. D. Quinn, D. Richardson, P. Page and D. F. Steele, Efficient electrochemical destruction of organic wastes, AEA Technol- ogy, Dounreay, Scotland. 10 D. Schmal, J. van Erkel, A. M. C. P. deJong and P. J. vanDuin, in Environmental Technology, Proc. 2nd Eur.Conf. Environ. Tech., Amsterdam, The Netherlands, June 22-26, 1987, ed. K. J. A. deWaal and W. J. van den Brink, Martinus Nijhoff Publishers. 11 K. J. Muller, in Electrochemical Cell Design and Optimization Procedures, DECHEMA Monographien, 1991, 123, 199, Frankfurt, FRG 12 M. Kimura, H. Miyahara, N. Moritani and Y. Sawahi, J. Org. Chem., 1990,55, 3897. 13 G. Kreysa and C. Reynvaan, J. Appl. Electrochem., 1982,12,241. 14 D. Simonsson, J. Appl. Electrochem., 1984, 14, 595. 15 G. Faita, Proc. 7th Int. Forum Electrolysis Chem. Ind., Electrosynthesis Co, East Amherst, New York, 1993. 16 Y. B. Acar and A. N. Alshawabkeh, Environ. Sci. Technol., 1993, 27, 2638. 17 R. F. Probstein and R.E. Hicks, Science, 1993, 260, 498. 18 G. Kreysa and K. Juttner, in Electrochemical Engineering and Energy, ed. F. Lapique, A. Storck and A. A. Wragg, Plenum Press, New York 1994, p. 255. 19 D. van Velzen, H. Langenkamp and A. Moryouseff, J. Appl. Electrochem., 1990, 20, 60. 20 J. Winnick, in Advances in Electrochemical Science and Engineering, ed. H. Gerischer and C. W. Tobias, VCH Verlagsgesellschaft mbH, Weinheim, 1990 vol. 1, 205. 21 P. D. Bennet, K. R. Bullock and M. E. Fiorino, The Electrochemical Society, Interface, 1995, 4 (4), 26. 22 S. Megahead and B. Scrosati, The Electrochemical Society, Interface, 1995, 4 (4),34. 23 NEWS, January 1996, Electric Fuel Ltd., Jerusalem, Israel. 24 A. F. Burke, Ultracapacitors for Electric and Hybrid Vehicles. Report R 1995:44, Swedish National Board for Industrial and Technical Development, Stockholm, 1995. 25 B. E. Conway, J. Electrochem. Soc., 1991,138, 1539. 26 A. J. Appleby and F. R. Foulkes, Fuel Cell Handbook, Van Nostrand, NY, 1989. 27 L. J. Blomen and M. N. Mugerwa, Fuel Cell Systems, Plenum Press, NY, 1993. 28 Ballard Power Systems Inc., Annual Report 1995, Burnaby, BC, Canada. 29 J. Janata, Anal. Chem., 1992, 64, 196R. 30 D. Pletcher and F. C. Walsh, Industrial Electrochemistry, Chapman and Hall, London 1990, p. 619. 31 K. Rajeshwar, J. Appl. Electrochem., 1995,25, 1067. 32 B. O’Regan and M. Gratzel, Nature, 1991,353, 737. Received, 14th January 1997 Accepted, 13th March 1997 Chemical Society Reviews, 1997, volume 26 189

ISSN:0306-0012    

DOI:10.1039/CS9972600181

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

New mass spectrometric methods for the study of noncovalent associations of biopolymers Richard D. Smith, James E. Bruce, Qinyuan Wu and Q. Paula Lei Environmental Molecular Sciences Laboratory, Pacfi c Northwest National Laboratory, PO Box 999, Richland, WA 99352, USA The use of electrospray ionization-mass spectrometry (ESI-MS) for the characterization of noncovalent complexes of biomacromolecules in solution is based upon the gentle nature of the electrospray process that allows a wide range of associations to be transferred intact to the gas phase as fully desolvated complexes. Examples include multimeric pro- teins, oligonucleotide duplexes, DNA-drug complexes and enzyme-inhibitor complexes. Various studies have indicated that at least some qualities of the three-dimensional solution structures are retained in the gas phase.Recent investiga- tions have also shown the relative stabilities of complexes in the gas phase can be very different than the same complexes in solution. In spite of this, the use of very gentle electrospray interface conditions can provide a direct reflection of relative solution abundances for similar complexes. Competitive Richard D. Smith is Leader of the Advanced Separations and Mass Spectrometry Group and Director of the FTICR Mass Spectrometry Facility at the Pacific Northwest National Laboratory (PNNL), located in Richland, Washington. His research involves the development and application of nanoscale separations and mass spectrometry to biological studies. Dr Smith is the author or co-author of more than 300publications, and has been awarded eleven patents and three IR I00 Awards.He is an adjunct faculty member of the Department of Chemistry, Washington State University, and an affiliate faculty member of the Department of Chemistry, University of Idaho. Qinyuan (Quincey) Wu is currently a research scientist in Hoechst Marion Roussel, Inc. Dr Wu received his BSc degree (I 982) in physical chemistry from Xiamen (Armoy) University, China, and MS (1990) and PhD (1993) degrees from The University of Illinois (Chicago). Dr Wu’sPhD research focused on the interactions of hyperthermal organic ions with metal surfaces. As a postdoctoral fellow at Pacific Northwest binding experiments using sets of ligands have been shown to yield insights regarding relative binding affinities in solution.The potential for high throughput affinity screen- ing of combinatorial libraries using ESI-MS is described based upon the multi-stage MS capability of Fourier transform ion cyclotron resonance instrumentation and involving the characterization of components (after dissocia- tion) of the library constituents initially present as complexes with a target biopolymer in the ion trap. This approach combines, in one rapid experiment, both affinity selection by complex formation with a biopolymer and the identification of the ligands selected from combinatorial mixtures, thus providing information on the relative binding affinities of the library constituents. The present status, limitations and promise of these methods are discussed. National Laboratory, his research involved both fundamental and applied aspects of biological mass spectrometry using electrospray ionization Q. Paula Lei is currently a postdoctoral fellow at Pacific.Northwest National Laboratory. Dr Lei graduated with a PhD in Chemistry from Professor Jon Amster’s group at the University of Georgia in 1996. Her research interest focus on bioanalytical mass spectrometry, including non-covalent com- plexes of biomolecules and mass spectrometry coupled with separation techniques as such capillary electrophoresis. James E. Bruce joined Dr Smith’s group in I992 after receiving a PhD in Physical Chemistry at the University of Florida and IS presently a Senior Research Scientist at PNNL.Dr Bruce’s research at the high field FTICR Facility has focused on the development of ion manipulation and detection techniques to enable new biological applications of mass spectrometry, including the direct bioaffinity analysis approach and the analysis of large individual ions using FTICR Richard D. Smith Quinyuan (Quincey) Wu Q. Paula Lei James E. Bruce Chemical Society Reviews, 1997, volume 26 191 1 Introduction Structurally specific noncovalent associations of biopolymers dictate a large fraction of biochemical processes, from the expression of genes to the activity of gene products (i.e. proteins), and often involve complex and competing pathways where small changes to either the physiological milieu or chemical modifications to the biopolymer (e.g.phosphorylation of specific amino acid residues of proteins) can lead to significant biological consequences.Gaining an understanding of the relevant chemical processes and the role of subtle structural modifications is at the heart of much biomedical research, and has practical implications that include the development of gene therapies, new vaccines (e.g. to cancer), and new drug leads. The difficulties involved in unravelling the underlying complexities include the generally large size of the participating biopolymers and the practical limitations upon sample quantities available in most cases for in vitro studies. The availability of a high resolution structure for a biopolymer (from crystallography or NMR) in many ways only serves to elevate the detail of the chemical questions while providing a scaffold for understanding their interactions.The Human Genome Program will increasingly exacerbate the need for faster and better methods for the study of large biopolymers and their interactions due to the rapid increase in the number of human genes available for study. It is axiomatic that scientific progress is closely coupled with the development and application of improved analytical techniques. The development of gel electrophoresis, for exam- ple, provided a tool now utilized in most biological research, with applications ranging from DNA sequencing to binding assays.Electrophoretic separations in gels are based upon molecular size and shape, and in many ways electrophoresis is analogous to low resolution (time-of-flight) mass analysis in a condensed medium. In recent years several developments have dramatically extended the utility of mass spectrometry (MS) for the study of large biopolymers, traditionally the domain of electrophoretic methods. Recent research has demonstrated its application for characterization of biopolymers extending to a mass greater than 200 kDa while providing detection limits in the attomole range (and below) and mass measurement precisions that exceed those provided by electrophoresis by more than three orders of magnitude! These advances have largely been derived from the introduction of electrospray ionization (ESI),’ which has even been demonstrated to allow the ionization of intact viruses2 and DNA segments of more than 100 MDa.3 Recent advances in mass analyser technology have also played an important role.Although a number of details of the ESI process remain a subject of speculation and debate, and there is yet limited information concerning the 3D structural conformation of large ions in the gas phase, it is now incontestable that many weak associations in solution do survive transfer to the gas phase. A substantial literature now demonstrates that a wide range of complexes from solution conditions (that can often mimic those of physiological interest) can survive the ESI pro~ess.~ Examples of noncovalent complexes observed by ESI-MS include enzyme-substrate, receptor-ligand, host-guest, intact multimeric proteins, DNA duplex and quadruplex species, oligonucleotide complexes with drugs and proteins, and protein-drug complexes.Upon first consideration, the study of solution interactions based upon their analysis in the gas phase appears to be an unlikely approach. In this regard, the distinctively ‘gentle’ nature of the ESI process is of particular significance. Numerous experimental studies have now indicated that using appropriate ESI interface conditions, compact structures are retained and, at least in some cases, significant structural features either survive or incur only modest changes. Once in vacuum and fully desolvated, the available results indicate that noncovalent complexes can be stable for indefinite periods.The lowest energy structures in the gas phase will almost certainly be different from those in solution, but the magnitude of both the kinetic and thermodynamic constraints upon transitions between these structures for the multiply charged species in the gas phase are presently unknown. There are also reasons that gross structural features might be retained for multiply charged proteins in the gas phase. In solution, the compact protein core structure substantially arises due to hydrophobic interactions and charge sites are most likely to reside on the surface in contact with the aqueous continuous phase. After removal of the dielectric solvent, Coulombic repulsion for the multiply charged protein will be minimized for charge sites on the surface-the number of which depends upon the ESI charge polarity, solution composition, molecular shape and the chem- ical nature of the charge site in the gas phase.A key question is whether sufficient internal excitation exists in the gas phase to propel conversion of the solution structure to a lower energy (desolvated) structure. The underlying fundamental issue is the role of solvent in maintaining the structural features of biopolymers. The ability to transfer biopolymers to the gas phase presents new opportunities; not only can measurements be made with incredibly high sensitivity (even a single ion can be repetitively manipulated and accurately weigheds) and enormous resolu- tion, but an increasing array of chemical reactions and other processes (e.g., H/D exchange6) can also be invoked.Before we discuss the experimental methods and the potential scope of their application, we first outline our view of the electrospray process and how it allows noncovalent complexes to be transferred to vacuum and desolvated while retaining important structural aspects. 2 The electrospray ionization process Electrosprays are typically produced at near ambient pressure from a liquid stream emanating from a capillary in a strong electric field. The electrostatic nebulization process produces an average droplet size that increases with flow rate, and is in the micron size range for low p1 min-l flow rates.Droplets carry a large net positive or negative charge (depending upon the applied voltage gradient), allowing either positive or negative ions to be produced. The electrospray is directed towards an aperture or capillary inlet, allowing charged particles to be transferred through differentially pumped regions to the mass analyser. During this transfer, heat (particularly at higher pressures) or electric fields (at lower pressures) are the primary variables that determine the species detected by the mass spectrometer. Perhaps the most widely accepted model for formation of small ions by electrospray is the ion evaporation me~hanism.~ In this model, charged species desorb from the highly charged droplet surface due to the locally high electric field.This process remains speculative for larger ions (e.g.biopolymers). We believe an alternative mechanism of ion formation summarized in Fig. 1 more satisfactorily explains key experi- mental observation^.^ The charged droplets initially produced are close to the charge (Raleigh) limit for their size and composition. In this mechanism, which has its roots in the initial ESI studies of Dole nearly three decades ago,g asymmetric droplet fission occurs as droplets shrink to the point where surface tension is inadequate to maintain droplet stability, generating offspring droplets having significantly smaller diameters. The asymmetric fission process facilitates the rapid formation of smaller charged droplets that are close to their respective maximum charge, and only a small amount of additional evaporation can cause a second asymmetric fission event.The larger progeny droplets can undergo additional asymmetric fission steps as solvent evaporation again drives them to instability, but the process will be faster and require less energy input for the smaller droplets. Through one or more such 192 Chemical Society Reviews, 1997, volume 26 Q(a+)I 0"1 10 nm @ U 1 nm U 1 nm Initial electrospray droplets which shrink by evaporation (a+ -105 net charges; 0 desig-nates macromolecules) Droplets break up by one or more asymmetric fission steps to yield smaller droplets having similar surface charge density (i.e.,a greater charge-to-mass ratio) Nano-droplets incorporating isolated macro- molecules (P) at sufficiently low initial concentrations; evaporation and charge loss continues (L = ligand; b+ d a+) Highly solvated macromolecular ion-ligand complex (c+ < b+; = polyelectrolyte charge site; 0= anion charge site or volatile buffer counter-ion) Warm highly solvated macromolecular ion- ligand complex; solution structure probably retained (0 = solvent; d+ d c+) Desolvated complex.Substantial (some?) features of solution structure maintained; electrostatic interactions of increased im- portance Dissociation of complex; solution structure substantially disrupted Dissociation of covalent bonds Fig. 1 Model for formation of macromolecular ions during electrospray ionization, based upon asymmetric droplet fission and the resulting formation of small nanometre diameter droplets.With increasing activation, residual solvent and charge are shed. The loss of higher order structure and noncovalent associations (e.g. ligand, L) is primarily driven by the extent of activation and repulsive Coulombic forces (after removal of the solvent). steps it is feasible to isolate single molecules given the concentrations relevant to most electrospray applications. For example, a 10-5 M solution initially contains approximately three analyte molecules per 100 nm droplet; a 10-6 M solution contains on average only about 0.3 analyte molecules (before any solvent loss by evaporation). Asymmetric fission steps might continue to droplet sizes approaching the dimensions of large proteins (ca.10 nm), and further evaporation yields the molecular ions detected ultimately. The key experimental variable for ESI is the extent to which the charged droplets are heated and the smaller charged species activated by collisions. Sufficient energy may be imparted to just barely cause desolvation or additional activation may be imparted to induce extensive dissociation in the gas phase. Initially, it was thought that since protein ions were generated by ESI with substantial net charge, their more compact structure Chemical Society Reviews, 1997, volume 26 193 in solution ‘stretched out’ in the absence of solvent due to Coulombic repulsion between charge sites.It gradually became evident that essentially the inverse is probably true; electro- sprayed ions are more highly charged because they are denatured and in a form readily extended prior to ESI (or to a lesser extent, when greater heating occurs during ESI).9 Lower 3.2 Distinguishing specific and nonspecific noncovalent associations in ESI-MS The simple observation of peaks by ESI-MS indicative of a complex constitutes insufficient evidence upon which to infer a structurally specific interaction in solution. Distinguishing between structurally specific noncovalent interactions arising charge states are produced from more compact structure~.~~’0 Although additional more closely spaced charge sites give rise to greater long-range repulsive contributions, the energy differences that occur over distances typical of chemical bonds, and available to drive dissociation, are quite small.Thus, although the activation barrier for breaking a weak bond may be lowered due to the increased Coulombic contributions, non- covalent associations based upon multiple weak electrostatic, hydrogen bonding and dispersive interactions can apparently survive in the absence of excessive thermal activation to the molecular ion. (A corollary is that once the gas phase structure is disrupted, and charge sites are moved apart, it is implausible that molecules will refold unless charging is reduced, and even then it is extremely unlikely that they would refold to the same ‘native’ structure as in solution.) As charging increases for a given molecular structure, the ‘effective’ proton affinities of charge sites for positively charged proteins (the basic amino acid residues), for example, decrease to a level at which proton transfer to volatilized solvent in the gas phase can occur.* Thus, this model implicitly predicts that the charge state distribution will depend upon both the detailed three-dimensional molecular structure and the location (and nature) of charge sites for the multiply charged macromolecular ions, consistent with experi- mental observation^.^ The desolvation of electrosprayed ions is typically completed in ca. 1 ms, and the resulting mass spectra are also strongly influenced by the extent of collisional activation and dissocia- tion in the ESI-MS interface.The loss of noncovalent associations during ESI may be kinetically constrained, allow- ing noncovalently associated species to remain associated as residual solvent is removed. The likelihood of this will depend upon the location and number of charges remaining on each of the associated species (i.e. due to Coulombic considerations), the solvent’s role in the noncovalent association, the relative strength and number of attractive interactions between complex constituents compared to solute-solvent interactions, and the extent of complex activation in the interface. Given this complexity, it is not possible at this time to make blanket statements from fundamental considerations; at present, experi- ment is our best guide.3 Experimental considerations and mass spectrometric methodologies 3.1 Electrospray source and interface conditions for observation of noncovalent associations The detection of noncovalent complexes from solution is, of course, of only limited interest unless it can be effectively applied to ‘unknown’ complexes, e.g. where stoichiometry and/ or relative binding affinity is not established. The ESI-interface conditions necessary to preserve noncovalent complexes are generally more ‘gentle’ than those conventionally employed for ESI-MS, although this is not always the case. The presence of excessive amounts of nonvolatile salts will result in adduction to molecular ions, decreasing sensitivity, and can make molecular mass measurements problematic.The use of ‘vola- tile’ buffers, such as ammonium acetate, is preferred, and one must consider the possible effects relative to established ‘physiological’ buffers. In many cases, however, the restriction to volatile buffers is of little importance, since solution pH, temperature and ionic strength are often the primary factors of concern. Low ESI flow rates (~50nl min-1) are also advantageous since the initial droplet size is smaller (<200 nm), desolvation is faster, and less heating is required for desolvation. IZ from solution and nonspecific aggregation in solution (and that might also result from the electrospray process), does appear to be feasible with careful selection of experimental conditions.However, the necessary conditions (e.g. low analyte concentra- tions) may not be achievable with certain instrumentation (due to sensitivity limitations, for example). There are several criteria we have advanced for evaluating whether complexes observed in the gas-phase reflect specific associations in solution:4 Stoichiometry. Generally, complexes (such as A-B) should be observed without comparable contributions which might arise from random aggregation at higher solution concentrations (i.e. A-A, B.B, AmB2, A2.B, A.B3 ,etc.) in order to be confidently attributed to a specific association in solution. Gasphase lability. Complexes due to weak binding forces (e.g. H-bonding, hydrophobic interactions, etc.) are generally more readily dissociated using more severe interface conditions or in tandem mass spectrometry (MS”) experiments.However, complexes having strong electrostatic interactions (e.g.protein-DNA complexes) may be tenacious in the gas phase, and the cumulative contributions of many weak bonds can make complexes having larger binding domains more stable. 13 Dissociation due to modification of solution conditions. Changes in solution temperature, pH, the presence or absence of buffer components, the addition of organic solvents, etc. may greatly weaken or disrupt specific associations in solution and should produce a corresponding change in the ESI mass spectra. Sensitivity to structural modifications. Perhaps the most unambiguous demonstration of a specific interaction from solution is obtained when a variant of one of the complex components produces a substantial change in the relative intensity of the complex in the mass spectrum.These criteria have now been extensively applied to model systems, and many examples convincingly demonstrating the detection of specific solution associations have been reported. At the same time, it is possible to obtain spectra that are not representative of solution. It is important to understand the origin of such observations, how they can be avoided, and the potential limitations to the methodology these observations suggest. At the present time ‘false negatives’, in which a suspected or expected complex is not observed, must be viewed with caution.Unless a complex of similar size and nature can be readily detected, one cannot be completely confident that either complex stability in the gas phase or some aspect of the interface conditions or instrument design does not prevent detection. On the other hand ‘false positives’ are readily avoidable by applying proper experimental conditions (e.g. low ESI flow rates, low analyte concentrations). The fact that mass spectra showing the expected solution stoichiometries for a range of biopolymers have been reported? and without contributions from nonspecific or random aggrega- tion, constitutes strong evidence for the feasibility of studying such interactions by ESI-MS. However, in addition to these structurally specific associations in solution, a broad range of ‘nonspecific’ associations (at undefined or multiple binding sites and of indefinite stoichiometry) can also occur.A class of ubiquitous noncovalent interactions in solution involves water, an essential component of the native structure of most proteins, where numbers of well localized (i.e. ‘specific’ and ‘internal’) water molecules are essential to protein conformation. Water is likely to be most strongly retained in association with charge 194 Chemical Society Reviews, 1997, volume 26 sites in the gas phase, but is also generally quite specific in its association with hydrophilic protein functional groups on the protein surface and may organize into clathrate-like structures around hydrophobic patches in solution.It is possible to retain some water with electrosprayed species using very ‘gentle’ interface conditions; however, the absence of ‘internal’ water molecules in typical ESI-MS spectra would appear to support the view that the protein’s structure has been ‘extended’ from its native tertiary structure. In apparent contradiction, it has also been shown that a compact structure can be retained even after the complete removal of all solvent (Fig. 1). The details as to how large biopolymer ions are ‘vacuum dried’ but can apparently retain substantial ‘memory’ of their solution struc- tures remain to be determined. It is of course impossible to prove beyond all doubt that a specific association exists in solution solely on the basis of ESI- MS results.A recent study for several model proteins indicated that aggregates observed by ESI were largely due to ‘non-specific’ associations already existing in solution, l4 although small contributions due to random associations are often observed for very large solute concentrations. (In some cases, it may be feasible to experimentally distinguish non-specific and specific complexes based upon their gas phase behaviour; see Section 6). Ideally, ESI-MS will be augmented by independent methods, although the extension to less well-characterized systems is increasingly justified as experimental limitations are defined. When appropriate experimental technique is practiced with amenable instrumentation, the sensitivity and speed of ESI-MS methods suggest its utility for effectively resolving questions of relative binding and stoichiometry.Of particular significance is the capability for conducting studies given only extremely small sample sizes (many orders of magnitude less than required for detailed structural studies) and often with the complex or heterogeneous samples that are the norm for larger biopolymers. In this regard recent developments in mass spectrometric instrumentation are of particular significance. 4 The use of advanced methods based upon ESI-Fourier transform ion cyclotron resonance (FT1CR)-mass spectrometry The combination of ESI with the recent rapid advances of FTICR-mass spectrometry is greatly extending capabilities for the study of large biopolymers.lS Ions are radially confined in the FTICR trap by an applied magnetic field and axially confined by a trapping potential which is applied orthogonal to the magnetic field.The frequency of the cyclotron gyration is inversely proportional to the mass-to-charge ratio (mlz) and directly proportional to the strength of the applied magnetic field. 16 The orbiting ion clouds induce image currents on two or more detection electrodes which, when Fourier transformed, provide an extremely precise measurement of the cyclotron frequencies (and thus the mlz values) and can simultaneously yield both ultra-high resolution and high mass measurement accuracy. This non-destructive FTICR detection scheme is exploited for ion re-measurements of the same ion population and in tandem or multi-stage (i.e.MS”) dissociation experi- ments from the introduction of a single group of ions that can be dissociated and remeasured multiple times, providing extensive structural information with minimal sample consumption. The simultaneous combination of extended mass range and unequalled mass measurement accuracy, resolving power, sensitivity, and MS“ capabilities distinguish FTICR. FTICR also affords the capability for selectively accumulating specific species in the ion trap using quadrupole excitation methods.17 The combination of capillary electrophoresis with ESI-FTICR analysis recently allowed the analysis of proteins from a single red blood ce11.18 The combination of capillary isoelectric focusing with ESI-FTICR analysis of a 2% red blood cell lysate has recently yielded high resolution spectra for carbonic anhydrase without interference from the ca.100-fold excess of haemoglobin. 19 Utilizing broadband quadrupole excitation methods for improved trapping of ion populations and sustained off resonance irradiation (SORI) methods for collisional dissociation of ions in the FTICR ion trap, partial amino acid sequence information could be obtained from the injection of ca. 75 erythrocytes into the separation capillary. MS” measure-ments allow the determination of partial amino acid sequences (the basis for rapid protein identification as well as localization of the site and nature of post-translational modifications).The observation of intact noncovalent complexes by ESI-MS presently requires a balance between providing sufficient heating/activation for ion desolvation, and keeping interface conditions mild enough to preserve the complex (Fig. 1). Insufficient desolvation will cause a loss of sensitivity. However, increasing ESI source heating or collisional activa- tion in order to increase sensitivity can induce dissociation of complexes and provide a distorted view of solution associa- tions. Increasing the biopolymer concentration is generally not a satisfactory solution since it will typically increase non-specific aggregation. It is likely advantageous for the final stages of the desolvation process to proceed slowly, something that is most readily accomplished with mass spectrometers based upon ion trapping technology; i.e.either the electrody- namic ‘quadrupole’ ion trap or FTICR instrumentation. Trap- ping methods allow the use of much gentler electrospray source conditions than generally applied, since the ion population initially trapped in the FTICR can be (on average) significantly solvated. The ion population can then be desolvated on a much longer time scale than possible with conventional instrumenta- tion (> 100 ms for FTICR 17s. < 100 ps for conventional instrumentation). This approach allows ions to be ‘vacuum dried’ in the gentlest fashion feasible, does not require ‘tuning’ of source parameters for each system, and provides higher sensitivity. As illustrated in Fig.2(a), the need for greater activation for desolvation of some ions results in excessive activation and dissociation of others. Qualitatively, this behav- iour is rationalized by the dissociation rate vs. 1 /(temperature) behaviour being much ‘steeper’ (i.e. due to a larger energy requirement) than the desolvation rate behaviour: thus desolva- tion dominates at low temperatures and long times while (a)Conventional instrumentation Tetramer activation MoreSolvated tetramers (6) Ion trapping 03 03 Fig. 2 Comparison of the different results obtained for desolvation due to the different characteristic timescales and necessary heating with conven- tional and ion trapping mass spectrometers (e.g. FTICR). A tetrameric complex is indicated having an initially broad variation in the extent of solvation.A slower desolvation avoids excessive activation in the ESI interface, but results in greatly decreased sensitivity if incomplete at the time of analysis. Chemical Society Reviews, 1997, volume 26 195 dissociation becomes dominant at the higher temperatures needed to drive rapid desolvation. The advantage of this approach is that all trapped ions can be desolvated and that the searching for ‘proper’ ESI source and interface conditions will be minimized. Available evidence to date indicates that biopolymer complexes are stable in the trap for periods of at least minutes to hours. 5 Examples of noncovalent complexes subject to study 5.1 Multimeric proteins The intact specific complexes for a number of multimeric proteins have been studied.10,20,21 Initial studies included such tetrameric proteins as haemoglobin, avidin, streptavidin, the chaperone protein SecB, and concanavalin A.The spectra generally show only a few peaks corresponding to different charge states of the same multimeric species and consistent with the known solution stoichiometry. Peaks indicative of trimeric and possibly pentameric species are not observed, consistent with their origin from specific solution associations. The peaks appear at relatively high mlz and have a ‘narrow’ charge state distribution, observations that have been taken to infer a compact shape and minimal structural heterogeneity.10 At higher ESI source temperatures the multimeric proteins readily dissociate.One of the strongest protein-ligand interactions (KD ca. lO-’5 M) found in nature is that of avidin (a 64 kDa glycoprotein) or streptavidin with biotin, which has wide applicability in biochemistry and molecular biology. Avidin is composed of four identical subunits which associate into the active form. Each of the four subunits can accommodate one biotin molecule. The tetrameric complexes of avidin and streptavidin with biotin have been studied by ESI-MS, and are entirely consistent with this ~nderstanding~~J3 The streptavidin complex with iminobiotin (a biotin analogue with KD ca. M) was observed to be weaker than the biotin complex,qualitatively consistent with solution beha~iour.~~ 5.2 Oligonucleotides and oligonucleotide-drug complexes It is now well established that double stranded DNA of > 10-12 base pair (bp) size is stable in the gas phase; stability increases with size and larger duplex double-stranded (ds) DNA ions will undergo dissociation of covalent bonds (e.g.base loss) in preference to dissociation of the two strands. High resolution ESI-FTICR mass spectra have recently been reported for ds PCR products of > 100 bp size.24 Noncovalent complexes formed between various drug molecules (both intercalators and minor groove binders) and oligonucleotides have been re-ported. 5.3 Protein-DNA complexes Recent work has indicated that complexes of proteins with small oligonucleotides having KD as low as 10-4 M are subject to study by ESI-MS,25 an observation that likely reflects a significant role of electrostatic interactions. Cheng et al.examined the binding of the Gene V protein (a dimeric protein) to a variety of small single-stranded (ss) oligonucleotides by ESI-FTICR. l3 Protein-ssDNA complexes having stoichiome- tries consistent with known behaviour were observed; only one dimer was found to bind to a 12-mer oligonucleotide, while two dimers were observed as the primary binding motif to the longer 18-mer (i.e. a 4: 1 stoichiometry consistent with the known stoichiometry of one Gene V dimer for each eight bases). The use of competitive binding conditions (i.e. molar excesses of two or more oligonucleotides relative to Gene V), provided insights into the sequence specificity of complex formation.‘3 These and other observations reflected solution behaviour and indicated that the observed complexes in the gas phase do not arise from non-specific electrostatic interactions (for which little difference would be expected for Gene V interaction with oligonucleotides of different sequence or between ssDNA and dsDNA). Previous studies had suggested that the Gene V protein-DNA complex has a stoichiometry in which each protein monomer binds to 3-5 nucleotides, but were ambiguous as to whether each one of these stoichiometries reflects complexes of specific ratio or an average of stoichiometries. The ESI-FTICR study showed a transition from 2:l to 4:l stoichiometry at ca.15 nucleotides. Importantly, it was also demonstrated that collisional dissociation methods could be used to break-up the complex, and then retain the Gene V product in the FTICR ion trap, after which collisional dissociation of the Gene V protein provided substantial structural information on the protein. Another example involves complexes of the eukaryotic transcription factor, PU. 1, with ds DNA.26 PU. 1 binds with high selectivity to the recognition sequence GGA(A/T). Fig. 3(a) shows that a solution of the PU. 1 protein with an excess of a 17 bp duplex (‘wild type’) having the recognition sequence gives an FTICR mass spectrum in which the proteinduplex DNA complex of expected stoichiometry dominates. Upon adding PU.1 protein to a mixture of ds oligonucleotides containing both the recognition sequence (wild type) and a ‘mutant’ lacking the recognition sequence under competitive binding conditions, the protein showed binding with only the wild type [Fig.3(b)]. Repeating the same experiment with a 20-fold excess of the mutant duplex again showed only complex formation with the (1 :l)w9-, (l:l)w8-1500 2000 2500 3000 3500 m/z 1000 1500 2000 2500 3000 3500 m/z Dm7’ (1 :1)w9-(1 :1)w8-n 1000 1500 2000 2500 3000 3500 m/z Fig. 3 ESI-FTICR mass spectra showing the complex of the PU. 1 protein with a double stranded 17 bp oligonucleotide (‘wildtype’; w) having the GGA(A/T) recognition sequence (a),and in competitive binding experi- ments with (h)a 1.3 fold and (c) a 20 fold excess of a 19 bp oligonucleotide (‘mutant’; m) without the recognition sequence.The experiments show the expected I : 1 complexes, and excess DNA duplex is evident at low mlz as either the duplex (D) or the two individual strands (A and B). 196 Chemical Society Reviews, 1997, volume 26 wild type [Fig. 3(c)].Gel shift assays showed that ESI-MS correctly reflected solution behaviour.26 6 Can gas phase dissociation studies provide information on binding in solution? 6.1 Experimental insights Questions related to the observation of noncovalent complexes by ESI-MS include: (1) how closely does the gas phase structure of the complexes resemble that in solution?; (2) can measurement of the stabilities of the gas phase complexes give insights regarding the stabilities of the corresponding com-plexes in solution?; and (3) if not, how do relative gas phase stabilities differ from those in solution? To probe these and related questions, we have conducted studies in collaboration with Whitesides and coworkers at Harvard University, using the 29 kDa protein carbonic anhydrase (CA) and its complexes with various substituted benzenesulfonamide inhibitors.CA is a roughly spherical ZnI1 metalloenzyme having a conical binding pocket, which cataly- ses the hydration of C02 to hydrogencarbonate, and is an attractive model system due to the stability of CA and its well characterized structure and ligand complexes. A large body of data correlates structures of sulfonamide ligands with their binding constants to CA, providing a basis for inferences regarding the protein structure and its ligand interactions in the gas phase.Fig. 4 shows an ESI-mass spectrum for a mixture of bovine CAI1 (BCAII) and a set of inhibitors in acidic solution where the native structure of the protein is rapidly denatured. In this case, a relatively conventional ESI-mass spectrum was obtained, even under the gentle interface conditions used, showing a range of charge states extending to >20+. The higher charge states indicate an increasingly disrupted solution structure, which can be attributed to acid catalysed denaturation, and consistent with this the Zn is absent. The lower charge states (at higher mlz) show greater retention of Zn, suggesting that ca.20% of the BCAII still retained Zn and a form of intermediate compactness at the time of analysis. Significantly, no com- plexes with inhibitors were observed, indicating that distinctive binding pocket features were absent in acidic solution (con- sistent with known solution behaviour) and that random aggregation with inhibitors during ESI does not occur. (M -zn)15+ (M Zn)14+ M13t inM'-2000 2100 2200 1600 2000 2400m/z Fig. 4 Mass spectrum of an inhibitor mixture (2.0 VM each) with BCAII (1.0 VM) in 10 mM acetic acid (pH 3.4).27No complex ions of inhibitors with BCAII were observed and BCAII has been denatured with substantial loss of Zn"; at lower charge states some Zn retention is evident (top).A dramatic difference is evident for a similar mixture of BCAII and inhibitors in 10 mM ammonium acetate at pH 7 (Fig. 3.27 Two lower charge state species (lo+ and 9') now dominate the spectrum, indicating the structure is much more compact. A series of peaks is observed corresponding to complexes of BCAII with the various inhibitors, and Zn is fully retained in these gas phase complexes. M10+ MIO+ + Inhibitors 'n 2900 2950 2800 3000 3200 3400m/z Fig. 5 Positive-ion ESI-FTICR spectrum obtained for a set of 17 inhibitors (0.05 VM each) with BCAII (1.0 VM in 10 mM ammonium acetate at pH 7). The bottom shows the narrow charge state distribution observed, typical of cases where noncovalent associations and a (presumably) compact three- dimensional structure is maintained using gentle ESI conditions.The top spectrum shows detail of the 10+charge state region, and shows a series of peaks corresponding to the enzyme-inhibitor complexes and the remaining BCAII.27 It was not clear from these results whether the inhibitor remains in the binding pocket after transfer into the gas phase, or whether the binding pocket structure is maintained (2.e. although the association of BCAII and inhibitor is maintained, the structure might be altered greatly). Experiments which compared the relative stabilities of BCAII complexes having both one and two identical inhibitors, show that the first inhibitor binds more strongly in the gas phase, consistent with retention in the binding pocket.28 If structural features are lost, no special gas phase stability would necessarily be expected for the first inhibitor compared to the second.When two inhibitors are associated with the enzyme, at least one of them must be 'non-specifically ' bound outside the binding pocket. A differ- ence in the complex stability in the gas phase might be expected, if a binding pocket were substantially preserved in the gas phase and the inhibitor specifically bound in solution remained associated; it is reasonable to expect that an inhibitor in the enzyme binding pocket will be more strongly bound in the gas phase than an identical inhibitor randomly associated on the protein surface. The relative gas phase stabilities of the desolvated bovine carbonic anhydrase I1 (BCAII)lO+ complexes with both one inhibitor and two inhibitor molecules (formed using large excesses of inhibitor) have been investigated in the FTICR ion trap using the SORI technique.SORI allows selective collisional heating of species in the FTICR trap. By fixing the frequency difference between the SORI and the complexes' cyclotron frequency, the gas pressure, and the length of the SORI event, the amplitude of the SORI irradiation can be used to vary the extent of collisional activation. Thus, the dissociation efficiency as a function of the SORI peak-to-peak amplitude (V,,) provides a measure of the gas phase stability of the complex. These experiments clearly showed that (BCAII + 2Inh)Io+ complexes are less stable in the gas phase, with loss of the second inhibitor being significantly more facile, and suggest that the enzyme-inhibitor complexes in the gas phase retain significant features from solution and that the dehydrated binding pocket likely retains the specifically bound inhibitor from solution.Studies also showed that stability is greatly reduced in the gas phase for inhibitor complexes with BCAII- Chemical Society Reviews, 1997, volume 26 197 Zn (i.e.the apo-protein) at pH 7 where the structure is known to be largely similar to the intact protein. Results from other experiments also established that the para-NO2-benzenesulfonarnide-human CAI1 (HCAII) complex is more stable than its ortho-substituted counterpart in the gas phase.The bulky NO2 group in the ortho position has a sterically hindered interaction of the inhibitor with the binding pocket and displays reduced binding affinity in solution. In these experiments, the 10’ charge state ions of the protein- inhibitor complexes were selectively accumulated in the FTICR and then subjected to SORI collisional activation. The HCAII complexes with the para-N02 inhibitor show greater gas phase stability (Fig. 6). Since solvent does not directly contribute to 100 2 80 v a,0 60 Uc a 40 .-2 .I-(d g 20 0 58 60 62 64 66 Amplitude of irradiation (Vpp) Fig. 6 A plot of the normalized abundances of the para-and orfho-NO2-benzenesulfonamide-HCAII complex ions (filled symbols) and the dis- sociated HCAII product ions (open symbols) vs.the extent of SORI activation. The amplitudes of irradiation at the crossing point of the two curves provide measures of the relative gas phase stabilities of the complexes in the gas phase. These results indicate a similar steric effects between the protein and the ligand exists in the gas phase as in solution, and suggest that features of the binding pocket are retained after desolvation in the gas phase. the steric effect, a similar difference in complex stability in the gas phase would also be expected for these two isomeric inhibitors if the active site structure of the noncovalent complexes were preserved. On the other hand, randomly associated inhibitors (i.e. where either the binding pocket no longer exists or the inhibitors do not remain in the pocket after transfer to the gas phase) would not be expected to result in any difference in the gas phase stability.These results also suggest that distinctive features of the binding pocket are retained in the gas phase. From the experiments described thus far, it appears that carbonic anhydrase can maintain a relatively compact structure in the gas phase, and that the binding pocket retains features that result in stronger inhibitor interaction in the gas phase. A key question is whether measurements of relative stabilities in the gas phase will generally mirror those in solution. Recent studies of BCAII-inhibitor complexes have explored this issue and showed dramatic differences in the relative gas phase stabilities compared with either solution binding constants or solution dissociation rates (‘off-rates’) for an array of complexes.29 The gas phase stabilities of BCAII-inhibitor complexes were found to have no direct correlation with the complex stabilities in solution [Fig. 7(a)].A plot of the calculated polar surface area for the inhibitors (that portion of the total molecular surface which is charged, has a large dipole or contributes to hydrogen bonding) vs.SORI amplitude revealed good (but different) correlations for sets of inhibitors both with and without aromatic amino acid residues [Fig. 7(b)].The stabilities of the complexes were also found to increase monotonically with the 198 Chemical Society Reviews, 1997, volume 26 8 I 7 6 I 0 5 4 0 0 i 1 I .,.,-,-I ’ 66 68 70 72 74 Gas phase stability (Vpp) Non-aromati2 F6 Gas phase stability (Vpp) Fig.7 (a)Plot of liquid phase dissociation constant vs.gas phase stability for a set of para-substituted benzenesulfonamide inhibitor complexes with BCAII measured SORI induced dissociation, showing stability increases with the size of the inhibitor (b)Plot of the polar surface areas of the same inhibitors vs. gas phase complex stability. The inhibitors having aromatic amino acid residues (circles) show stronger binding with BCAII than inhibitors with aliphatic side chains in the gas phase having the same polar surface area. number of inhibitor amino acid residues (or the polar surface area).These correlations suggest that the major attractive forces for noncovalent protein-ligand binding in the gas phase are due to interactions between the polar surfaces through electrostatic, dipole-dipole, or hydrogen bonding interactions, and that the inhibitor tail has collapsed to the protein surface in the gas phase, and contributes significantly to complex stability. For inhibitors having the same polar surface area, an aromatic amino acid side chain results in a stronger binding interaction with the protein in the gas phase than does an aliphatic side chain. Thus, whereas the off-rates of BCAII-inhibitor com-plexes in solution are mainly affected by hydrophobic interac- tions between the inhibitor and the enzyme, their corresponding gas phase stabilities appear to be primarily determined by polar interactions that reflect the extent of contact area between complex constituents.6.2 Implications for ESI-MS studies These results show that attempts to obtain information regard- ing solution stability directly from the corresponding gas phase measurements will often be problematic. In the case of isomeric inhibitors where differences in solution binding with HCAII are attributed to steric effects, the gas phase results correctly mirrored those in solution, and suggest that the binding pocket retains a crucial role in the gas phase stability. The fact that a second inhibitor molecule binds with less strength in the gas phase, as does a single inhibitor to the apo protein, adds further support to this view.However, in comparing inhibitors of differing structure, no direct correlation between gas phase and solution stabilities was obtained. This is not surprising given the different environments. The key implication for applications is that ESI-MS studies should be conducted under the gentlest conditions possible if they are to correctly reflect the relative abundances of the protein-ligand complexes in solution. As we noted earlier, it appears that both the use of low flow rate electrosprays and the slow desolvation (using ion traps) are likely to be advantageous for this purpose. As harsher ESI interface conditions are utilized, the relative abundances of gas phase complexes will be increasingly skewed due to any differences in gas phase stabilities.Thus, studies are more likely to produce misleading results when significant gas phase dissociation of complexes occurs. Although the evaluation of solution binding based upon gas phase stability is generally to be avoided, this does not prohibit the use of ESI-MS for studies of complexes in solution, but does serve to clarify the experimental requirements (e.g. interface conditions affording negligible dissociation of com- plexes). Studies of small molecule associations that are primarily hydrophobic in nature are most likely to be problem- atic. The gas phase stabilities of complexes will likely increase as molecular size increases, or more importantly, the available ‘contact area’ between the constituents increases.Although the generality of these observations remains to be determined, they suggest that ESI-MS may well be broadly applicable to studies involving interactions of larger biopolymers. In the next section, we describe an approach based upon this understanding and the abilities for various ion manipulations provided by FTICR instrumentation, and illustrate its utility for studies involving protein-inhibitor complexes. 7 Application to screening combinatorial libraries 7.1 Bioaffinity characterization mass spectrometry The traditional approach to drug development has involved the synthesis or isolation of individual compounds followed by their screening and evaluation through a series of chemical and biological assays.An initial step in the screening process often involves identification of candidate compounds that display high binding affinity in vitro to the targeted biopolymer. More recently, the selection of pharmacologically active molecules from mixtures, or ‘libraries’, of compounds has been shown effective and has become a major enterprise. The ability to screen very large libraries to identify candidates for more exhaustive study allows a broad range of molecular diversity to be explored. For example, in the case of tripeptides, a library containing all variations for three amino acid residues will consist of 8000 peptides (twenty possible for each of three residues, i.e. 203). The number of possible hexapeptides is 6.4 X lo7 (i.e.209. In the ‘combinatorial’ approach, many com- pounds are synthesized (often deliberately producing a complex mixture), and then subsequently examined for their affinity to the targeted biopolymer (or other property), typically by partitioning subsets of the library so as to facilitate the identification of the most active components. This is often performed by binding either the target or library molecules to a solid support medium in a fashion designed to facilitate identification of the active components. The combinatorial approach has proved attractive since it allows much larger numbers of compounds to be generated and screened more rapidly than by serial approaches involving the individual synthesis and assay of each prospective affinant.The conventional screening of combinatorial libraries is most effective when the synthesis is well behaved (e.g. components are generated in nearly equimolar quantities, unsuspected components are not present, etc.), the general composition is well established, and suitable analytical methods are available. From this viewpoint libraries consisting of polypeptides or oligonucleotides are attractive, since the well-developed tools of biochemistry and molecular biology may be invoked for their synthesis and analysis. Unfortunately, they also have limited utility as drugs due to their generally rapid degradation by enzymatic processes, and most efforts focus on chemically more diverse libraries where the synthetic and analytical advantages that apply for these linear biopolymers do not currently exist.We have developed an approach for the screening of combinatorial libraries based upon the capabilities of ESI-MS and the use of advanced FTICR ion manipulation methods.30 The bioaffinity characterization mass spectrometry (BACMS) approach eliminates the need for distinct separation and/or purification steps involving the original ligand mixture and the problems associated with linking the affinity ligand to a surface since the complex can be formed free in solution at physio- logically relevant pH values. The key to this approach is the ability to directly transfer fragile noncovalent complexes from solution by ESI and into the FTICR mass spe~trometer.~~ The potential advantages of this approach originate from the ability to first recognize the complex in solution (in which it may have only a very low concentration), to separate it in the gas phase from all other ions (including other complexes), and finally to provide structural information on the complex and/or selected biomolecule.Thus, the separation/affinity selection and analy- sis steps are combined in one experiment. A conceptual representation of one implementation of the BACMS approach is given in Fig. 8. The complexes to be investigated are electrosprayed directly from a solution contain- ing both the affinity target and the ligand library, and using Electrospray from solution and ac- cumulate species from solution mixture of target protein and li- gands in the FTICR trap Trap ‘filled’ with selected protein ligand complexes; spectrum re-corded Complexes dissociated and ligands retained (for determination of rela- tive binding affinities); spectrum recorded Ligands dissociated for identifica- tion (if necessary); spectrum ob- tained Fig.8 A conceptual representation of the BACMS technique for the characterization of mixtures of complexes. In this approach the complexes are selectively accumulated in the trap, and then dissociated with retention of the ligands that would then display relative abundances representative of their binding with the targeted biopolymer.30 Since the ligand ions are retained in the ion trap after a spectrum is obtained, subsequent stages of MS can be used for structural characterization of the ligands by exploiting ion dissociation methods.appropriate ESI source conditions, transferred to the gas phase and accumulated in the FTICR ion trap. Typically, the complexes of interest are first identified in the mass spectrum and then isolated in a separate step using selected-ion accumulation (SIA) (recent developments allow the selective accumulation of even very low concentration species,32 which fills the trap to useful capacity of the population of selected ions). This is followed by SORI dissociation of all the Chemical Society Reviews, 1997, volume 26 199 complexes and the retention of the corresponding higher binding affinity ligand species in the trap, after which a high resolution mass spectrum can be obtained.In the same experiment, the inhibitors can be further dissociated to obtain structural information and for definitive identification27 A requirement of this approach is that the inhibitor carry a charge. Since the complexes are highly charged, dissociation will generally result in net charges on both dissociation products if charge-carrying sites are available, and either positively or negatively-charged complexes may be selected for this purpose. Alternatively, inhibitor libraries can also be designed to incorporate a charge-carrying site that does not affect relative solution binding. Since the noncovalent complexes to be investigated with BACMS require that the concentration of the affinity target should be limited to avoid non-specific aggregation during ESI, the complexes of interest may be present only at very low concentrations in solution.Very large libraries would further increase the need for sensitivity, and benefit from the FTICR capability for accumulating trace level constituents (the accu- mulation of noncovalent complexes from solution concentra- tions as low as M has been demonstrated32). Importantly, the wide range of FTICR capabilities for high resolution and multi-stage MS analysis are fully available to be applied to species recovered from the dissociation of noncovalent com- plexes. 7.2 Bioamnity characterization mass spectrometry of protein-inhibitor complexes The initial demonstration of the BACMS approach was conducted in collaboration with Whitesides and coworkers.We initially studied two inhibitor mixtures, of 7 and 18 components, that had well characterized BCA-inhibitor binding constants spanning three orders of magnitude to M).~~One group of inhibitors consisted of compounds having dipeptide substituents with one amino acid invariant and the other position incorporating amino acids chosen to be representative of diverse size, shape, hydrophobicity and acid-base properties. The relative abundance ratios of the various complexes observed by FTICR for two series of inhibitors were consistent with their relative affinities towards BCAII in solution.27 In this work, we also demonstrated the use of multi-stage MS methods for dissociation of the inhibitors to assist their identification (although high resolution analysis was adequate for small libraries).To demonstrate the extension to larger combinatorial li- braries, two mixtures of para-substituted benzene sulfonamide inhibitors were synthesized (using solid phase chemistries) (derived from 4-carboxy benzenesulfonamide, l),in which all C(O)-NH-AA1-AA&(O)-NHCH2CH&@H 0 1 combinations of 17 amino acid residues (AAI and AA2) were incorporated into two positions (17 X 17 = 289 c0mponents).~3 (Cys, Met and Trp residues were omitted due to the possibility of oxidation during synthesis.) When a mixture of BCAII (2.5 p~)and a peptide library (0.5 VM for each inhibitor; 289 and 256 compounds for the L-and D-libraries, respectively) in a 10 mM NH40Ac solution (pH 7.0) was analysed using ESI- FTICR, we observed major peaks in the mass spectra corre- sponding to intact complexes in the 7- to 9-charge states (Fig.9, top). Note that the individual protein-ligand complexes cannot be resolved due to the large and overlapping isotopic envelopes for the different complexes, a complication that results primarily from the size of BCA. Thus, it is apparent that direct mass spectrometric analysis of the complexes is im- practical, regardless of mass spectrometer resolution. On the other hand, collision-induced dissociation of the selectively accumulated 9-charge state complexes primarily produced the 200 Chemical Society Reviews, 1997, volume 26 [CA1l+l]s \ 3000 4000 5000 [CAM]& 400 450 500 550 600 512 513 514 515 m/z Fig. 9 (a)ESI-FTICR mass spectrum from a mixture of the 289-component L-library (0.5 VM each) and CAI1 (2.5 ELM)in 10 mM N&OAc (pH 7.0).33 (b)SORI dissociation of the isolated complex ions of [CAII+1]9-.The dissociation conditions were set such that the complex ions were completely dissociated. The insets (c) and (4 show expanded views of the singly charged inhibitors, [l]-. 8-charge state of the protein and singly charged negative ions for nearly all the inhibitors ([lll-, Fig. 9). The resulting mass spectra allowed the dissociated ligands to be identified based upon their different molecular masses; their ion intensities provide a measure of the relative binding affinities. The correlation is clearly sensitive to the equilibrium concentration of the protein.To obtain quantitative information about binding constants, the equilibrium concentration of the protein must be accurately known, or a correlation curve must be calibrated using internal standards.33 This approach has several implicit assumptions. First, there should be minimal or no dissociation of the complexes during ionization, ion transport, trapping and detection (since the stability of the complex for each ligand in the gas phase may not follow that in solution, as already discussed). Secondly, differences in sensitivity due to ionization, ion transport and detection of the ligands should also be small.The approach assumes similar ionization efficiencies for complexes of different ligands, an assumption justified since these complex ions have very similar formation efficiencies due to the fact that the protein accounts for the bulk of the complexes' mass and charge. Consistent with this assumption, the observed charge states of the BCAII-inhibitor complexes are similar to those of the free BCAII ions. -+ Hydrophobicity -+ Hydrophobicity -FYL IVAHTGONSPKRED FYLVAHTGQN S PKRED F F Y Y L L I V V A H T G 3 2-4 Q N0 2 4-3 S P'qi L -Library D -Library Fig. 10 The dependence of relative ion intensities on the composition of amino acids in the peptide library The structure of the peptide library is shown at the top of the figure The amino acids (AA1and AA2) are arranged in such a way that their hydrophobicity decreases from left to right, and from top to bottom The relative ion intensity for each ligand was obtained by companng the ion intensity of this ligand to that of the Gly Gly compound (present in both libranes) The results from L-and D-library were plotted on the same scale of relative ion intensity (bottom left comer) 33 The one letter codes for amino acids were used to indicate the identity of AAI and AA2 (adapted from ref 33) The ESI-FTICR approach we have described generates relative binding affinities for a large number of ligands simultaneously in one relatively fast expenment Fig 10 gives a representation of the relative abundances for the dissociated ligands in terms of the composition of the amino acids for the two libranes [Note that sequence isomers (-AA,-AA,- and -AA,-AA,-) and structural isomers (e g -Leu-Leu-and -1le- Ile- in the L-library) are not distinguished since their molecular masses are identical 3 Seven individual inhibitors were then synthesized from the two libranes to determine their binding affinities to BCAII in solution using a fluorescence binding assay33 The binding constants of the seven inhibitors in solution correlated well with the relative intensities of the ions dissociated from the BCAII-inhibitor complexes (Fig 11) The Kb for the tightest binding inhibitor identified by ESI-MS (AAI = AA2 = L-Leu, see Fig 10) was 1 4 X 108 dm3 mol-1 The Kb of the compound bound most weakly (AAl = AA2 = Gly) was 4 9 X 106 dm3 mol-1 The data also binding affinities of the tripetide inhibitors-the side chains of L-amino acids interact more effectively with the active site of CAI1 than did D-amino acids The BACMS approach should be most effective for identi- fication of the tight binding ligands, since lower abundances and contributions from nonspecific (random) associations may limit the quality of information for weaker binding ligands, partic- ularly in larger libranes Further study of additional well- characterized model systems will be necessary to define the extent to which quantitative information can be denved from this method In principle, the ESI-FTICR methods should be most useful for determining the relative binding affinities for a library of compounds to a protein when (a)the binding studies can be carried out using 'volatile' buffers (e g ammonium acetate, Tris acetate, and ammonium hydrogencarbonate), (h) all the library components will be charged after dissociation of complexes in the gas phase, and (c) the library components have different masses The design of libranes in a way that ensures indicated that the addition of hydrophobic groups at the para that each component has a different molecular mass can position of benzenesulfonamide increases binding constants eliminate ambiguities in the identification of ligands and reduce The chirality of the amino acids also appeared to influence the the demands for subsequent dissociation and MS stages The fact that relative binding constants can be rapidly denved from these experiments for all or most of the members of a library may provide the basis for facile screening of combinatorial libranes, as well as a basis for the extension of this methodology loo ' +L (Leu-Leu) for qualitative studies using more complex or poorly charac- -#-D-(Leu-Leu)a,$e 40-- a, [I L -(Thr-Thr) 20 - (Gly-Gly+ I+ +D-(Ser-Ser) D -(Thr-Thr) L-(Ser-Ser) o! 8 -'''''I ' ' .'''''I ' a s a , .J lo6 107 108 109 Kb/dm3 mol-' Fig. 11 Correlation of relative ion intensities vs Kb in solution for seven peptide inhibitors from the L-and D-library 33 We selected inhibitors having identical AA, and AA2, since their ion intensities are free of ambiguities that are caused by sequence isomers (adapted from ref 33) terized libraries and even complex fractions derived from natural products The attractions of this approach include speed and the immediate characterization of the most relevant components from complex and otherwise poorly characterized mixtures Less obvious advantages include the potential for establishing difference in binding properties for biopolymer variants and modified biopolymers, in such cases the bio- polymers would also be charactenzed by MS methods after dissociation of complexes 8 Concluding remarks The application of ESI-MS for the study of noncovalent associations in solution constitutes a new capability with potentially broad applications in biochemical and biomedical research Establishing relative binding affinities using compe- titive binding expenments from measurements of relative abundances of the complexes in the gas phase is clearly feasible The detection of noncovalent complexes also provides a basis Chemical Society Reviews, 1997, volume 26 201 for more conventional approaches, such as those involving the variation of ligand concentration and the use of Scatchard plots for determining binding constants.34 These methods increas- ingly benefit from the rapid maturation of capabilities for accurate mass measurements and multi-stage MS methods for obtaining structural information.The full integration of these capabilities and continuing instrumental advances are certain to greatly expand the range of biopolymers and their complexes subject to study.For example, it should be feasible to study individual virus particles or large protein complexes by ESI-FTICR, and multi-stage MS studies of even individual mole- cules recovered from the surface of such complexes will also likely prove to be feasible. To the extent that the three dimensional solution structure is retained in the gas phase, a question clearly demanding additional study, H/D exchange and affinity (i.e. small molecule adduction) experiments conducted in the gas phase offer the potential for probing structure, and a basis for biophysical studies that can contribute to under- standing protein folding.Such previously fanciful experiments are increasingly feasible based upon continuing refinement of electrospray methods and FTICR capabilities for sophisticated ion manipulations, indefinite trapping periods, high sensitivity, and the ability to make series of precise measurements from only a single i0n.33~5 9 Acknowledgments The authors wish to thank Drs H. R. Udseth, C. Liu, D. C. Muddiman, L. Pasa Tolic and S. A. Hofstadler, the early contributions of Dr X. Cheng (Abbott Laboratories) to this work and gratefully acknowledge the fruitful collaboration with Professor George Whitesides and co-workers at Harvard University. We thank the National Institutes of Health (GM53558) for support of this research. Pacific Northwest National Laboratory is operated by Battelle Memorial Institute for the US Department of Energy, through Contract No.DE- AC06-76RLO 1830. 10 References 1 R. D. Smith, J. A. Loo, C. G. Edmonds, C. J. Barinaga and H. R. Udseth, Anal. Chem., 1990, 62, 882. 2 G. Siuzdak, B. Bothner, M. Yeager, C. Brugkdou, C. M. Faquet, K. Hoey and C. M. Chang, Chem. Biol., 1996,3,45. 3 R. D. Chen, X. H. Cheng, D. W. Mitchell, S. A. Hofstadler, Q. Wu, A. L. Rockwood, M. G. Sherman and R. D. Smith, Anal. Chem., 1995, 67, 1159. 4 R. D. Smith and K. J. Light-Wahl, Biol. Mass Spectrom., 1993, 22, 493. 5 R. D. Smith, X. Cheng, J. E. Bruce, S. A. Hofstadler and G. A. Ander- son, Nature, 1994, 369, 137. 6 A. Miranker, C. V. Robinson, S. E. Radford, R.T. Aplin and C. M. Dobson, Science, 1993, 262, 896. 7 J. V. Iribame and B. A. Thomson, J. Chem. Phys., 1976, 64,2287. 8 M. Dole, L. L. Mack, R. L. Hines, R. C. Mobley, L. D. Ferguson and M. B. Alice, J. Chem. Phys., 1968, 49, 2240. 9 B. E. Winger, K. J. Light-Wahl, R. R. 0. Loo, H. R. Udseth and R. D. Smith, J. Am. SOC. Mass Spectrom., 1993, 4, 536. 10 K. J. Light-Wahl, B. L. Schwartz and R. D. Smith, J. Am. Chem. Soc., 1994,116,5271. 11 E. R. Williams, J. Mass Spectrom., 1996, 31, 831. 12 M. S. Wilm and M. Mann, Int. J. Mass Spectrom. Ion Proc., 1994,136, 167. 13 X. Cheng, A. C. Harms, P. N. Goudreau, T. C. Terwilliger and R. D. Smith, Proc. Natl. Acad. Sci. USA, 1996, 93, 7022. 14 S. L. Kaufman, J. W. Skogen, F. D. Dorman, F.Zarrin and K. C. Lewis, Anal. Chem., 1996, 68, 1895. 15 R. D. Smith, J. E. Bruce, Q. Wu, X. Cheng, S. A. Hofstadler, G. A. Anderson, R. Chen, R. Bakhtiar, S. 0. Van Orden, D. C. Gale, M. G. Sherman, A. L. Rockwood and H. R. Udseth, Mass Spectrometry in the Biological Sciences, 1996, ed. A. Burlingame and S. Carr, Hurnana Press, Totowa, NJ, 25. 16 M. B. Comisarow and A. G. Marshall, Chem. Phys. Lett., 1974, 25, 282. 17 J. P. Speir, G. S. Gorman, C. C. Pitsenberger, C. A. Turner, P. P. Wang and I. J. Amster, Anal. Chem., 1993, 65, 1746. 18 S. A. Hofstadler, F. D. Swanek, D. C. Gale, A. G. Ewing and R. D. Smith, Anal. Chem., 1995,67, 1477. 19 J. C. Severs, S. A. Hofstadler, Z. Zhao, R. T. Sehn and R. D. Smith, Electrophoresis, 1996, 17, 1808. 20 K. J. Light-Wahl, R. D. Smith and B. E. Winger, J. Am. Chem. Soc., 1993,115, 5869. 21 R. D. Smith, K. J. Light-Wahl, B. E. Winger and J. A. Loo, Org. Mass Spectrom., 1992,27, 811. 22 B. L. Schwartz, K. J. Light-Wahl and R. D. Smith, J. Am. SOC.Mass Spectrom., 1994, 5, 201. 23 B. L. Schwartz, D. C. Gale, R. D. Smith, A. Chilkoti and P. S. Stayton, J. Mass Spectrom., 1995, 30, 1095. 24 D. C. Muddiman, D. S. Wunschel, C. Liu, L. Pasa-Tolic, K. F. Fox, A. Fox, G. A. Anderson and R. D. Smith, Anal. Chem., 1996, 68, 3705. 25 M. J. Greig, H. Gaus, L. L. Cummins, H. Sasmor and R. H. Griffey, J. Am. Chem. SOC., 1995,117, 10765. 26 X. Cheng, P. E. Morin, A. C. Harms, J. E. Bruce, Y. Ben-David and R. D. Smith, Anal. Biochem., 1996, 239, 35. 27 X. Cheng, R. Chen, J. E. Bruce, B. L. Schwartz, G. A. Anderson, S. A. Hofstadler, D. C. Gale, R. D. Smith, J. Gao, G. B. Sigal, M. Mammen and G. M. Whitesides, J. Am. Chem. SOC., 1995, 117, 8859. 28 J. Gao, G. M. Whitesides, Q. Wu, P. Lei and R. D. Smith, unpublished work, 1996. 29 Q. Wu, J. Gao, G. B. Sigal, J. E. Bruce, G. M. Whitesides and R. D. Smith, J. Am. Chem. Soc., 1997, 119, 1157. 30 J. E. Bruce, G. A. Anderson, R. D. Chen, X. H. Cheng, D. C. Gale, S. A. Hofstadler, B. L. Schwartz and R. D. Smith, Rapid Commun. Mass Spectrom., 1995, 9, 644. 31 J. E. Bruce, S. L. Van Orden, G. A. Anderson, S. A. Hofstadler, M. G. Sherman, A. L. Rockwood and R. D. Smith, J. Mass Spectrom., 1995, 30, 124. 32 J. E. Bruce, G. A. Anderson and R. D. Smith, Anal. Chem., 1996,68, 534. 33 J. Gao, X. Cheng, R. Chen, G. B. Sigal, J. E. Bruce, B. L. Schwartz, S. A. Hofstadler, G. A. Anderson, R. D. Smith and G. M. Whitesides, J. Med. Chem., 1996,39, 1949. 34 H. K. Lim, Y. L. Hsieh, B. Ganem and J. Henion, J. Mass Spectrom., 1995, 30, 708. 35 J. E. Bruce, X. Cheng, R. Bakhtiar, Q. Wu, S. A. Hofstadler, G. A. Anderson and R. D. Smith, J. Am. Chem. SOC., 1994, 116, 7839. Received, 23rd December 1996, Accepted, 14th March 1997 202 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600191

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Some aspects of organic pigments Zhimin Hao and Abul Iqbal Ciba Specialty Chemicals, Pigments Division, CH-1723 Marly, Switzerland Organic pigments are traditionally used in the mass colouration of plastics and synthetic fibres, and in surface coatings such as paints and inks. More recently, they have also found increasing use in a number of high technology industries. Commercial pigments must satisfy many per- formance criteria, which in turn are determined by their molecular, solid state and particle surface characteristics. This review highlights some of the important aspects of organic pigments which are of importance to both producers and users. 1 Introduction Colour is an important and ubiquitous part of our everyday life and ever since pre-historic times, colours, be they inorganic or organic, have had profound anthropological, psychological, aesthetic, functional and economic impact on society.During the past 140years, since the discovery of mauveine by W. H. Perkin, there has been a remarkable development of the synthetic colourant industries. Colourants are either dyes or pigments. To the colour chemist, it is important to distinguish between these two terms. Organic dyes are soluble substances mostly applied to various substrates from solution. A general characteristic of dyestuffs is their ability to provide colour in monomolecular disperse form solely by selective absorption of visible light. Organic pig- ments, on the other hand, are finely divided, coloured particulate solids which are practically insoluble in most solvents and in the media in which they are incorporated by adequate dispersion techniques.Organic pigments impart colour by selective absorption and/or scattering of visible light. In the course of the last few decades, organic pigments have experienced rapid growth to become an industry of significant commercial potential. The traditional function of organic pigments has been to impart colour, and this will remain the Zhimin Ha0 graduated in 1982 from East China Institute of Chemical Technology, Shanghai, and then went on to study at the University of Leeds, England, where in 1985 he obtained a PhD degree in Colour Chem- istry. Following post-doctoral research at North Carolina State University, USA (I 986-1989), he moved to Zurich to work as a wis-senschaftlicher Mitarbeiter in the Industrial Chemistry Lab- oratories at the Swiss Federal Institute of Technology (ETH).In 1992 he joined Ciba-Geigy Marly Research Centre, now a part of the Ciba Specialty Chemicals Inc., where he is a research scien- tist in the Pigments Division. Zhimin Hao major volume use of pigments for many years to come. Coloured organic pigments today are used in many industrial and consumer sectors to satisfy aesthetic needs, to communi- cate, identify, differentiate, or to secure and protect. The aesthetic significance of coloured organic pigments extends from art, fashion to decoration, and more. Their use in cosmetics and mass coloured synthetic fibres is widespread in the fashion world.Decorative uses of organic pigments is abundantly manifest in the vehicle and machine industry, architecture and building, furniture and household, and leisure industries etc., where pigmented paints, coatings and plastics find frequent application. The most widespread use of coloured organic pigments is in the form of printing inks, followed by paints and plastics. The coloured packages, posters, and other diverse products from the printing ink industry are glaring examples of their use in communicating specific messages to the consumer. When thinking of coloured organic pigments for identifica- tion, examples such as, company car fleets, red fire service cars, product brand colours, etc.come to mind. Organic pigments are used to print postage stamps and currency notes, facilitating their identification and differentiation. Using different coloured organic pigments to identify and differentiate cable coating, gas conduits, electric switches, yellow school buses etc. also has an implied safety aspect. In recent times, however, organic pigments are finding increased use in a number of high technology industries, such as photo-reprographics, opto-electronic displays and optical data storage. In some of these applications, the pigment is still employed because of its colour imparting capability, while in other cases it performs a special function which is not based on its colour. Organic pigments are commonly supplied in a number of different commercial forms including powders or granules (either surface treated or untreated), aqueous press cakes, pre- dispersed aqueous pastes, flushed pigments (dispersions in Abul Iqbal was born in Dacca, Bangladesh, and now lives in Switzerland.He has an MSc in Chemistry from the University of Dacca, and a Diploma and PhD in Applied Chemistry from the Technical University of Aa-chen in Germany. After his PhD, he spent six years as a staff chemist with Monsanto Research in Zurich. Since 1975 he has worked with Ciba-Geigy, Basle, in pig-ments research, and currently heads the world-wide re-search activities of the Pig- ments Division of Ciba Spe- cialty Chemicals Inc. Abul Iqbal Chemical Society Reviews, 1997, volume 26 203 viscous, aliphatic hydrocarbon-based media), resin pre-dis- persed pigments, and plastic colour concentrates or master- batches.The commercial performance of a pigment in a vehicle system is defined by a long list of its application properties, such as its colouristic performance, rheological behaviour, durability and ecological compatibility/acceptability. From the end-user’ s point of view, the pigments must fulfil certain requirements with regard to such properties. However, in this article we shall restrict our discussions only to such characteristics as hue, tinctorial strength, transparency/opacity, resistance to heat, light, weather and solvent, and rheology and dispersibility. The physical and chemical characteristics that control and define the performance of an organic pigment include its molecular, solid state and particle surface characteristics.Table 1 summarises the more important molecular and solid state parameters which primarily influence the end-use properties of the organic pigment. The pigment manufacturers must design their products, accordingly, by judicious choice of the chemistry and solid state pigment elaboration techniques, to meet these performance requirements. This review article purports to highlight some of the above mentioned important aspects of organic pigments, without the intention of being exhaustive in its coverage. 2 Commercial performance criteria of organic pigments The most fundamental performance criterion of a commercial organic pigment is its colour as manifested by its reflectance spectrum, which is most strongly influenced by the molecular absorption spectrum.Commercially, the most desirable prop- erty is probably the value in use, or the colour value per kg of the pigment. Colour value generally translates into tinctorial strength and purity of shade, both of which are dependent on the absorption spectrum intrinsic to the pigment structure, high molecular extinction coefficients and absorption spectra with sharp cut-offs representing the ideal profile. Opacity or hiding power of a pigment is a function of its absorption coefficients, the absorption wavelength of light, its particle size, or light scattering coefficient and relative refractive indices of pigment and vehicle (pigment density).Light scattering has a powerful influence on opacity. As a rule of thumb, scattering and opacity are maximum when the pigment particle size equals half the absorption wavelength of absorbed visible light. As indicated in Table 1, colour strength and opacity can thus be best manipulated by controlling the chemical structure, crystal lattice parameters (e.g.modification) and particle morphology (e.g. size) of the pigment. Table 1 Pigment performance and molecular/solid state parameters relationship Commercial performance Molecular and solid state criteria parameters Colouristics: hue/purity of shade Molecular structure (A,,,), crystal lattice properties Colour strength Molecular structure (E,,,), particle morphology Hiding power Molecular structure (A,,,), particle morphology Fastness: Photochemical Molecular structure, stability particle morphology and crystal lattice properties Heat stability Molecular structure, particle morphology and crystal lattice properties Solvent fastness Molecular structure, particle morphology Rheology Particle morphology, particle surface characteristics Dispersibility Particle morphology, particle surface characteristics The fastness of a coloured organic pigment is a measure of its inherent ability to resist the chemical and physical influences to which it is exposed during and after its incorporation into a pigmented system.The most important fastness property of a pigment is its light and weather fastness, as measured by the extent of fading or darkening of a pigment colouration as a function of the duration of exposure to specified light and weathering conditions.The light fastness of an organic pigment is a function of the intrinsic molecular structure, as well as of the solid state properties of the pigment. Heat stability is primarily a function of the pigment crystal lattice energy, as determined by the electronic, atomic and molecular interactions prevalent in the lattice. Under the drastic processing conditions for most plastics, a good organic pigment must feature sufficiently high heat stability, so as not to suffer any consequential change of shade through chemical degrada- tion, change of crystal modification, or change in particle morphology as a result of Ostwald ripening or re-crystallisation of the pigment particles.Poor solvent fastness of an organic pigment is a function of its solubility in the application medium and is manifested by its increased tendency to bleed or migrate from one pigmented substrate to another, or to migrate from within a pigmented medium to redeposit itself at its surface, or to re-crystallise to larger particles in a pigment dispersion medium. Solubility is an intrinsic molecular property. 3 Chemical features of organic pigments The building blocks in organic pigments are molecules that determine, directly or indirectly, important performance proper- ties of the pigments. Such organic pigment molecules are generally characterised by planar conjugated chromophoric systems featuring functional groups such as the C=O and NH groups.In certain cases they may contain acidic and basic functional groups allowing precipitation of soluble dyes via salt and metal complex formation. Organic pigments are classified according to their generic name and chemical constitution. Details of the chemical constitution of organic pigments is given in the Colour Index (C.I.) published by the Society of Dyers and Colourist.l Each chemical type is characterised by a C.I. number, for example, copper phthalocyanine (see Fig. 1) is designated C.I. Pigment Blue 15 (‘C.I. Pigment’ generally being omitted for brevity, and P.B., P.G., P.O., P.R.,P.V. and P.Y. etc. used to designate blue, green, orange, red, violet and yellow pigments, respectively). The development of organic pigments for colouration was triggered off by two landmark events in the history of colour chemistry, namely, the discovery of the first aniline dye, mauveine or aniline purple, in 1856by the 18 year old William Perkin, followed by the discovery of the diazotisation reaction of aromatic amines by Peter Griess a few years later. The first organic pigments to be released on the market at the turn of the century were all pure azo compounds with P-naphthol derivatives as coupling component (e.g. P.R. 1-3, P.0.5). Such products, along with the analogously structured laked azo pigments like P.R.53:l and P.R. 57:1, which are obtained by precipitation of the adequately functionalised azo dyes via salt formation, indeed mark the actual beginning of the era of organic pigments colouration. Replacement of the naphthol based coupling component by acetarylamide led to the first yellow azo pigments (Hansa Yellows, e.g. P.Y. 1). The azo pigments held sway over the organic pigment industry for several years until the advent of copper phthalocyanines in 1935. Many new organic structures, mainly featuring hete- rocyclic chromophores, have since been discovered and in- troduced as commercial organic pigments. In terms of chromophore structures, roughly half the world market volume of synthetic organic pigments is comprised of pigments based on the azo chromophore, while a quarter of the volume is claimed by metal (mainly copper) phthalocyanines. 204 Chemical Society Reviews, 1997, volume 26 ko CO(NH& + CuCh + cat.c HgC' 2 1 ,NO2-JHN,p CI H \/ 3 Fig. 1 Structures of some typical classical pigments. 1 C.I. Pigment Blue 15; 2 C.I. Pigment Yellow 74; 3 C.I. Pigment Red 112;4 C.I. Pigment Red 57: 1; 5 C.I. Pigment Yellow 13 (R = CH3) and C.I. Pigment Yellow 12 (R = H). The remaining 25% are primarily shared by a number of heterocyclic pigments, a few quinoid and indigoid structures and multidentate metal complexes, not taking into consideration the lakes of heteropoly acids with basic dyes. For a more comprehensive list of organic pigments and their constitution, the reader is referred to the recent publications by Herbst and Hunger,* Z~llinger,~ McKay et al.,4Iqbal et aL5 and Jaffe.6 In the light of the diversity of chemical structure coupled with differing technical performance levels, it has become conven- tional to classify organic pigments into so-called classical pigments on the one hand and high performance pigments on the other. As implied by their name, classical organic pigments have been known for many decades.A few examples of their structure are given in Fig. 1. The so-called azo pigments tend to exist as the hydrazone tautomer (as depicted in Fig. l), stabilised by intramolecular hydrogen bonding-9 Such monoazo and disazo pigments are relatively cheap and their technical performance level is not very high.However, copper phthalocyanines, which are also classified under classical pigments, are an exception to the rule, in as much as they have excellent all-round fastness proper- ties. Copper phthalocyanine pigments are manufactured either from phthalic anhydride, urea, cupric chloride and ammonium molybdate (as catalyst) according to the following general Scheme 1 ,2 or from phthalonitrile, obtained from ammoxidation of o-xylene and an appropriate cupric salt leading to the crude pigment3 as shown in Scheme 2. A useful synthetic equivalent of phthalonitrile in copper phthalocyanine synthesis is 1,3-di- iminoisoindoline or its tautomer 1-amino-3-imino-isoindole-nine (Scheme 3).3 The visible absorption spectrum of copper phthalocyanine is mainly determined by the tetraazaporphyrin chromophore Scheme 1 aCH3CH3 Scheme 2 NH 11 CUClZ -NH Scheme 3 structure and is little influenced by the four benzo rings, as was demonstrated by the fact that the spectrum did not change appreciably upon replacement of the benzo by naphtho or pyridine rings.3 The only effective way of implementing a significant change in hue of copper phthalocyanine is by multiple substitution of hydrogen atoms at the benzo rings, for example, by chlorine and/or bromine to yield green pigments (P.G.7 and 36). Although several classical yellow and red monoazo and disazo red pigments may be used to colour paints and plastics, the majority of them are employed in printing inks manufacture. Disazo pigments like P.Y.12 and 13, lakes of sulfonated monoazos like P.R. 53:1 and 57:1, and p copper phthalocyanine (P.B. 15:3) constitute the most important pigments for this purpose. In comparison with classical organic pigments, high perfor- mance organic pigments tend to be of more recent origin and feature excellent all-round fastness properties. They are more costly to manufacture and find use in specialised and more demanding applications, such as in automotive paints and construction plastics, where exceptional light, weather, and heat Chemical Society Reviews, 1997, volume 26 205 fastness are required. Structural examples of high performance organic pigments are shown in Fig.2. CI' 6 7 a Cl 9 10 CI H 9 \'\ N 0 H \ CI 11 12 X O \/ \/ :m13 Fig. 2 Structures of some high performance organic pigments. 6 C.I. Pigment Red 144; 7C.I. Pigment Orange 36; 8 C.I. Pigment Yellow 110; 9 C.I. Pigment Yellow 139; 10 C.I. Pigment Violet 23; 11 C.I. Pigment Red 254; 12 C.I. Pigment Violet 19; 13 C.I. Pigment Red 179(X =NCH3), C.I. Pigment Red 224 (X =0). The two most significant chemical classes of high perfor- mance pigments are azo and polycyclic/heterocyclic pigments. Of the azo pigments, the most important ones include yellow to red disazo condensation products (e.g., P.R. 144), yellow azo metal salts (e.g.,P.Y. 183), and yellow to red pigments based on benzimidazolone (P.O.36). A number of commercially important azomethine type pigments also belong to the high performance pigments; typical examples are isoindolinones (e.g., P.Y. 1 10) and isoindolines (e.g., P.Y. 139). In addition, two other types deserve mention, namely the anthraquinoids (e.g., P.R. 177, P.B. 60) and dioxazines (e.g., P.V. 23). By far the most important heterocyclic pigments in the family of high performance pigments are quinacridones, including its various polymorphic forms (e.g.,P.V. 19), perylenes (e.g.,P.R. 179,224), and the more recently commercialised diketopyrrolo- pyrroles (e.g., P.R. 254). Of these, the diketopyrrolopyrrole pigments will be described in greater detail below. 206 Chemical Society Reviews, 1997, volume 26 The most recent addition to the class of high performance pigments are the 1,4-diketo-3,6-diaryIpyrrolo[3,4-c]pyrroles, also known as 3,6-diaryl-2,5 -dihydro-pyrrolo[4,3 -c]pyrrole- 1,4-diones.The underlying 1,4-diketopyrro10[3,4-c]pyrrole (DPP) chromophoric system in these pigments combine the elements of indigo-like cross-conjugated vinylogous amidic and vinylogous hydrazine units in a rigid planar structural frame, essentially representing the lactam analogue of the 8n-electron fused ring hydrocarbon pentalene. 0 14 DPP: 1,4-diketopyrroIo[ 3,4-c]pyrrole The first synthesis of a compound incorporating the DPP chromophore unit was reported in 1974 by Farnum, et a1.,10 who while attempting to synthesise 2-azetinones according to Scheme 4 quite inadvertently isolated a small quantity of the 15 Scheme 4 diphenyl DPP derivative (15) as a by-product, instead of the target p-lactam.A few years later, reconsideration of structural analogies of the DPP chromophore with several commercially well-known pigments triggered off reinvestigation of the chemistry and solid state properties of this class of compounds at Ciba-Geigy Company.' 1 The investigations at Ciba culmi- nated in the establishment of the pigmentary potential of the DPP class of compounds. A broad spectrum of brilliant shades, ranging from orange- yellow via blue-red to violet could be accomplished by simply exchanging the substituents at the p-and rn-positions of the two phenyl rings attached to the DPP chromophore unit. Fur- thermore, selected members of this family of pigments offered outstanding light and weather fastness.Despite their low molecular masses, the diary1 DPP pigments are highly insoluble and remarkably resistant to chemicals and heat." Such behaviour may be attributed, above all, to the presence of strong intermolecular bonding forces (e.g.,H-bonding, van der Waal's contact, n-n interactions between molecular planes) in the pigment crystal lattice, as was also corroborated by single crystal X-ray structure analyses of individual members of this group of compounds. Unfortunately, however, the Reformatsky synthesis of DPP, as reported by Famumlo gave yields too poor to warrant commercialisation of the process. Keeping in mind the commercially important criteria of availability of startingmaterials, minimum number of synthetic steps, and technol- ogically and ecologically feasible chemistry, the continued search at Ciba for alternative approaches to the DPP system ultimately led to the discovery of the elegant synthesis shown in Scheme 5, which could be tuned by appropriate choice of CN Base Scheme 5 reactants and conditions to provide optimum product yield, consistent with good process economics.l2 Subsequent mechanistic investigations of the process opened an attractive preparative route of access to asymmetrically substituted aryl/alkyl and diaryl DPP derivatives.13,14 Gompper et a1.15 have subsequently disclosed yet another route to diaryl DPP derivatives, comprising the reaction of succinamide with N,N-dimethylbenzamide diethyl acetal. A qualitative inspection of the diaryl DPP molecule (Fig.3) reveals several centres of reactivity in the molecule. Fig. 3 Centres of potential chemical reactivity in the molecule of DPP While the appropriately substituted phenyl rings should be capable of undergoing diverse electrophilic and nucleophilic substitution reactions, the bicyclic lactam chromophore unit incorporates three different functional groups, namely double bonds, carbonyl and NH moieties, each of them being potentially amenable to chemical transformation. In the interest of keeping the DPP chromophore intact, synthetic efforts have focused mainly on electrophilic aromatic substitution, N-substitution and nucleophilic transformations of the carbonyl group without incurring concomitant rupture of the heterocyclic nucleus.Scheme 6 summarises a selection of products that can be obtained by direct chemical transformation of diphenyl DPP. Currently, the most important member of the DPP class of pigments is Pigment Red 254 which is a very opaque yellow-red pigment of outstanding outdoor durability, brightness, and resistance to heat and chemicals. Another is the diphenyl DPP (P.R. 255), which is a high performance orange-red pigment. Both are used in automotive finishes, while a higher strength version of P.R. 254 is available for pigmentation of plastics. 18 X =Xi = S03H 19 X=X1 =Br 20 X = Br; XI= H X 0% 16 21 R = CH3, wl1 Q ItN/(-j HNW N H R = H, p-CI, p-CN, o-Br, p-NO2, o-CO2R" 23 Scheme 6 4 Solid state properties of organic pigments Unlike dyes, which are present in polymeric substrates as either single molecules or small clusters, pigments are applied in the form of discrete crystalline particles well dispersed in the medium.Their overall pigmentary performance therefore cannot be adequately described in terms of the properties of the individual molecules. The colouristic and fastness properties of a pigment, as well as some other application properties, are significantly influenced by its solid state characteristics. The most important solid state parameters include crystal modifica-tion, crystallinity, particle size, shape and state of aggregation.In addition, the formation of pigment solid solutions and mixed crystals can also bring about unexpected effects. 4.1 Crystal formation and growth By virtue of their structural features, pigment molecules exhibit a strong tendency to form highly ordered crystalline assemblies. The main driving forces for such crystal formation are intermolecula hydrogen bonding, n-TCstacking between ad-jacent planar molecules and van der Waal's interactions. The overall effect of such interactions is generally believed to lead to stabilisation of the crystal lattice through lowering of energy. The tendency of crystal formation and growth can be seen clearly when a layer of copper phthalocyanine, obtained via vapour deposition, is subjected to a solvent treatment.When the Chemical Society Reviews, 1997, volume 26 207 Fig. 4 (a)Scanning electron micrograph of a CuPc layer (0.15 pm)before sohlent treatment; (b)optical micrograph of the same layer after one week of xylene vapour treatment, (c) scanning electron micrograph of the layer after four Mreeks of exposure to xylene vapour partially amorphous pigment layer [Fig. 4(a)] is exposed to xylene vapour at room temperature for one week, long needle- shaped crystals are formed as shown in Fig. 4(b). Prolonged xylene vapour exposure eventually converts the entire layer of the less stable a-form into the more stable, highly crystalline B-form [Fig. 4(c)]. In many cases, the absorption spectrum of an organic pigment undergoes a change upon transition from the molecular to the crystalline state.This is illustrated in the case of a DPP pigment.I6 In solution, diphenyl DPP shows a green-yellow fluorescent colour, whereas in the solid state it is bright red. In Fig. 5 the absorption spectra of both states are plotted together, w t s! 300 400 500 600 700 h Inm Fig. 5 Solution spectrum of DPP in DMSO (solid line) and solid state spectrum of the DPP thin film obtained vra vapour deposition (dotted line) (Reproduced by permission from Mizuguchi and Wooden. 16) showing a bathochromic displacement of the maximum absorp- tion of ca. 40 nm from the solution spectrum to the solid state spectrum. The colour shift can be attributed to intermolecular hydrogen bonding in the solid state.4.2 Polymorphism A large number of pigments are known to be polymorphic, in other words, there are different ways in which molecules with the same chemical constitution may be arranged within the crystal lattice, resulting in differently structured crystals, commonly denoted as crystal modifications or polymorphs. Each of the polymorphs has its own characteristic lattice energy 208 Chemical Society Reviews, 1997, volume 26 and hence its stability. The readiness with which a pigment is converted from one modification into another depends upon the energy barriers between such modifications. To be of value, polymorphs should offer useful and distinct property profiles and sufficient stability to avoid unwanted phase transitions.Sometimes an unstable modification, if commercially attrac- tive, can be stabilised by appropriate chemical and phys- icochemical treatment. Crystal modifications are usually characterised by their distinctive powder X-ray diffraction patterns. Crystallograph- ically polymorphs are classed as distinct substances. In general, X-ray diffraction patterns from different polymorphs bear no relationship to each other. Some pigment polymorphs are almost indistinguishable in appearance except X-ray diffraction pattern, whereas others differ so widely in their colouristics and other properties that at first glance they would not be considered even to be of the same chemical type. As a typical example of polymorphism, copper phthalo- cyanine is known to occur in no less than five crystal modifications,* i.e., the a-form (red-shade blue), the (3-form (green-shade blue) and the y-, 6-and &-forms.The a-form per se (C.I. Pigment Blue 15) is not stable, it tends to revert to the (3-form. Of commercial interest are the phase-stabilised a-form (C.I. Pigment Blue 15:l) and the (3-form (C.I. Pigment Blue 15:3). The stabilisation of the a-form is achieved by partial chlorination, about 0.5 C1 per molecule being sufficient to prevent reversion to (3-form. In Fig. 6, X-ray powder diffraction diagrams of the a- and (3-forms of copper phthalocyanine are compared. In the case of DPP pigments, recent studies carried out in our laboratories have led to the identification of a few new crystal modifications, e.g.the (3-fo1-1111~ of C.I. Pigment Red 254. In paint colouration this (3-form is significantly more yellowish as compared with the a-form, as is illustrated in Fig. 7. 4.3 Solid solutions and mixed crystals Analogous to liquid solutions, in the solid state, if a binary (or multi-component) mixture is prepared by methods other than simple mechanical mixing, one of the components may assume the role of the solvent (host) while the other behaves as the solute (guest), thus forming a ‘guest-host’ solid solution. Molecules of the ‘guest’ compound enter the crystal lattice of the ‘host’ and, as a consequence, the X-ray powder diffraction pattern of such a solid solution is either the same as, or very similar to, that of one of the components, the ‘host’.The term solid solution refers to a crystalline phase which exists over a I I 0 5 10 15 20 25 30 35 40 I 0 5 10 15 20 25 30 35 40 Fig. 6 X-ray powder diffraction patterns of copper phthalocyanine: the a-(upper) and the fi-(lower) crystal modifications range of compositions. 18 X-Ray diffraction patterns of solid solutions may undergo a progressive change in one or both of two ways as the composition varies. If the constituent molecules are not identical in size, then the unit cell size may alter which could result in an angular movement of the peaks. On the other hand, if the substituting molecule does not have the same X-ray scattering factor, the intensities of the diffraction peaks may alter.In certain cases a unique diffraction pattern, different from either of the component pigments, is obtained at a specific pigment composition. It is customary to denote such a product as a mixed crystal.'8,19 A dramatic change in performance, resulting from the synergistic effect, often accompanies the formation of such mixed crystals. A loose analogy can be drawn between a mixed crystal and an azeotrope in the liquid state. In contrast to physical mixtures where an additive effect regarding colour and other properties is usually the case, a solid solution or a mixed crystal often shows non-additive, unpredict- 1 j able colouristic and other performance properties. If crystallo-graphically pure, a solid solution or a mixed crystal will behave just like a single pigment.The phenomena of solid solutions and mixed crystals can be advantageously exploited in practice to extend an existing range of commercial pigments, by widening their colouristic scope and at the same time improving fastness and other application properties. Examples of pigment solid solutions are available in the literature. Solid solutions among quinacridones, substituted and unsubstituted, for example, have been prepared with the colours ranging from orange, gold, red to maroon.20,21 Improvement in pigmentary performance has also been noted. More recently, solid solutions involving acetoacetanilide, as well as diaceto- acetanilide azo pigments have been rep0rted.992~ Investigations in our own laboratories have led to the recent discovery of DPP mixed crystals.2' It is found that equimolar amounts of two different DPP pigments can combine and form an entirely new crystal lattice.The X-ray diffraction pattern of the resulting mixed crystal is different from either of the two constituent DPP pigments. This is exemplified by the mixed crystal of the unsubstituted DPP and the tert-butyl substituted DPP. A very significant bathochromic colour shift is seen concomitant to the formation of the mixed crystal. This is illustrated in Fig. 8. 4.4 Isomorphism Pigments of different chemical constitution but which are isostructural, i.e. those which have similar crystal cell dimen- sions and in which the molecules have similar positions with respect to each other, are called crystal isomorphs.It is obvious that if two pigments are isomorphs and if the X-ray scattering of the pigments is similar, this will give rise to very similar diffraction patterns. Isomorphism is generally observed with pigments from the same chromophore class. Some isomorphs of monoazoacetan- ilide pigments have been rep0rted,~4 the monobromo and dibromo analogues of C.I. Pigment Yellow 3 being good examples. C.I. Pigment Blues 15 and 15:1, mentioned earlier, are also isomorphs. More recently, a similar observation was 1 Fig. 7 C.I. Pigment Red 254, the a-(above) and the fi-(below) crystal modifications in an alkyd resin system Chemical Society Reviews, 1997, volume 26 209 Powder X-ray diffraction patterii PVC inasstoile (1%) 0 5 10 15 20 25 30 35 40 45 equimolar mixed crystal 0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45 Fig.8 PVC coloration and X-ray diffraction pattern of a binary DPP mixed crystal, as compared to the component DPP pigments made in the disazo diarylide series of pigments for C.I. Pigment Yellow 14 and C.I. Pigment Yellow 63, the two pigments showing strikingly similar colouristic proper tie^.^ Another example is the crystallographic mimicry of asym- metric DPP pigments by equimolar binary mixed crystals of symmetric DPPs.~~ Thus the X-ray powder diffraction patterns of a mixed crystal, e.g. see above, and the corresponding asymmetric DPP were compared. They have been found to be nearly identical.From colouristic point of view, the two isomorphs are also identical. This is illustrated in Fig. 9. 4.5 Crystallinity Immediately after their formation, either by grinding or by precipitation, organic pigments tend to show low crystallinity, with many internal and external defects. In practice, condi- tioning and finishing treatments are usually applied in order to remove or minimise the defects and hence to improve crystallinity. This leads, in general, to improved performance properties of the pigment. More crystalline pigments have lower lattice energy and therefore they show better fastness properties. Another characteristic of crystalline pigments is the lower susceptibility to aggregation, which makes them easier to disperse.In addition, crystallinity can also affect optical properties. For example, the absorption maximum and fluores- cence of the untreated amorphous phase of C.I. Pigment Red 3 1 is shifted to longer wavelengths and the maximum extinction coefficient increased upon thermal conversion to the crystalline pha~e.2~ 4.6 Crystal size The crystal size of a pigment has an important effect on the colour strength, opacity/transparency , photochemical and ther- mal stability and viscosity. Fig. 10 gives a qualitative correla- tion between particle size and some of the important parameters of a pigment dispersion. 210 Chemical Society Reviews, 1997, volume 26 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45 the binary symmetric DPP mixed crystal Fig.9 Mimicry of an asymmetric DPP (a)by a binary DPP mixed crystal (h):PVC colouration and powder X-ray diffraction patterns tinctorial strength Yviscosity \hopacity I particle size * Fig. 10 Dependence of some pigment performance parameters on particle size Colour strength increases, in general, as mean particle size decreases. This has been firmly established by theoretical elaborations,26 experimental confirmation,27 and practical usage. The highest dependence is found to occur at d 0.1 pm. The shade of a dispersion can be affected to a certain extent by a reduction of pigment particle size. An example of this is C.I. Pigment Yellow 34, which becomes greener as the size is reduced.28 Another important aspect is size distribution.In general, a wide particle size distribution results in lower colour strength and dullness or dirtiness in shade. The opacity/transparency of a dispersion is affected by particle size and size distribution of the pigment. Thus, in a given medium, there is a maximum in opacity at mean particle size (0.2-0.5 ym).29 Opacity at a given size is greater the narrower the size distribution. Similarly, the transparency of a pigment-vehicle system is also dependent upon particle size. A high degree of trans- parency is an essential requirement of inks for the printing industry, and of paints for special effects (metallic, pearlescent, etc.). In such applications, pigments with very small particle sizes are used.A demonstration of particle size effect on transparency/ opacity of the pigment-vehicle system is provided by an experimental DPP pigment, which is prepared in two different sizes. The electron micrographs and PVC colouration over contrast paper are shown in Fig. 11. It is remarkable to note that equal amounts of the same pigment of variable particle size applied under same conditions give significantly different colour appearances. 5 Surface characteristics of organic pigments The (physicochemical) nature of the particle surface of an organic pigment is to a large extent a function of its intrinsic molecular chemistry. Such surface properties, superimposed with crystal lattice properties (such as crystallinity, crystal modification) and particle morphology parameters (such as size and shape), ultimately account for the surface energy of pigment particles.Thus, in general, the more amorphous a particle the higher is its surface energy. Moreover, as the size of the primary pigment particles is reduced, a natural consequence is the rapidly increasing area of surface created, accompanied by a parallel increase of the total surface energy. These are active surfaces and have a tendency to react in any way which reduces the overall energy level. The small particles can therefore aggregate (stick to each other), or adsorb other chemical species such as solvent, surfactant or resin molecules from solution. The phenomenon of aggregation can have multiple det- rimental effects. First, it negatively influences the dispersibility and dispersion stability of the pigment.The effectiveness of organic pigments in imparting colour depends on how well they are dispersed in the application medium. Secondly, increased structure formation in dispersions leads to a reduction in the flow of such systems due to higher viscosity. Sometimes, the surface chemical character can also influence other mechanical properties of the application media, for example, distortion or warpage of moulded plastic articles. Chemical Society Reviews, 1997, volume 26 211 The most efficient way of controlling aggregation and its consequential effects is through effective modification of the interfacial properties by introduction of certain types of additives that adsorb to the pigment particle surfaces and anchor functional groups there.Several surface treatment procedures can be applied to improve the surface characteristics of an organic pigment. Most of them have in common the fact that they promote the formation of an adsorbed layer containing well-solvated, extended-chain molecules. The physicochemical mechanism involved is probably steric stabilisation, although electric charge stabilisation may play some part in aqueous dispersions. Details of these types of mechanism are given elsewhere.30 6 Some high-technology applications of organic pigments Recently, pigments, as well as dyes, are becoming increasingly important in the electronics industry.In some applications, the role of the pigment is the conventional one of imparting colour, such as in colour filter pigments for liquid crystal displays and electrostatically charged toners and inks for non-impact printing technologies.31 In many other applications, it is the special functional properties of the pigment, such as fluorescence, electro-luminescence, photo-conduction and selective absorp- tion of infrared radiation, which are exploited. A novel erasable optical memory device has been recently claimed by Langhas, et. al.32 which is based on the thermal transformation of a fluorescent into a non-fluorescent modification of a dimorphic DPP pigment. Electroluminescent elements containing a fluo- rescent DPP derivative in the luminescent layer have likewise been ~laimed.~3 212 Chemical Society Reviews, 1997, volume 26 Information storage systems utilising solvent and NIR laser induced phase change of thio DPP derivative (16) as the recording medium, have been described.34 The same thio DPP derivative also exhibits interesting photo-conductive properties, and its use as charge generating material in the preparation of organic photo-receptor drums for laser beam printers has been extensively investigated.35 Recent claims have also been made on the use of diphenyl DPP and its phenyl ring-substituted derivatives as charge generating materials for plain paper copiers and laser printers.36 16 thio DPP derivative 7 Summary and outlook The conventional role of organic pigments has been to impart colour to a substrate.In recent years, organic pigments are finding increasing use in the high technology industries of electronics and opto-electronics. In both cases, the commercial success of a pigment is dependent on the fulfilment of a series of end-user demands, which define the technical performance properties of the product. To achieve the desired technical performance profile, the pigment manufacturers need to fine- tune the intrinsic molecular and solid state parameters of their product. In retrospect, there has been a remarkable growth in the pigment industry over the past one hundred and forty years. Tra- ditionally, the main driving forces behind such a development have been the demand for better technical performance in terms of colouristics and stability on the one hand, and economic considerations on the other.In more recent times, additional factors, such as ecological concern and the emergence of novel markets and engineering trends, have gained increasing im- portance and will continue to shape the future of the pigment industry. While such a development may pose several chal- lenges to be mastered, it also creates new opportunities, lending vital impetus to the pigment industry. To meet these challenges, the pigment industry will need to create and develop innovative products, processes, and applications. At the same time, it will be indispensable to invest in longer term research in order to secure continued growth of the industry well into the next century.8 References 1 Coloui Index, Pigments and Solvent Dyes, 3rd edn., Society of Dyers and Colourists, 1982 (updated 1988) 2 W Herbst and K Hunger, Industiial Organic Pigments, VCH, Weinheim, 1993 3 H. Zollinger, Color Chemistry Synthesis, Proper ties and Applications of Oiganic Dyes and Pigments, 2nd rev. edn., VCH, Weinheim, 4 R. B McKay, A. Iqbal and B. Medinger, in Technological Applications of Dispersions ed., R. B. Mckay, Marcel Dekker, Inc., New York, 1994, pp.143-176 5 A. Iqbal, B. Medinger and R. B. McKay, in Advances in Color Chemistry, ed. A. T. Peters and H S. Freeman, Blackie, Glasgow, 1996, VOI 4,pp. 107-144 6 E. E Jaffe, Encyclopaedia of Chemical Technology, vol.19, 4th edn., Wiley, New York, 1996 7 A Whitaker, J Soc Dyers Colourists, 1978, 94, 431. 8 A. Whitaker, J Soc Dyers Colourists, 1988, 104, 294 9 S. Zheng, D Liu and S. Ren, Dyes and Pigments, 1992, 18, 137. 10 D G Farnum, G Mehta, G G. I. Moore and F P. Siegel, Tetrahedron Lett , 1974, 29, 2549. I1 A. Iqbal, L Cassar, A. C. Rochat, J Pfenninger and 0. Wallquist, J Coat Tech, 1988, 60, 37. 12 A. C Rochat, L. Cassar and A Iqbal, (Ciba-Geigy Ltd ), Eur Patent 9 94911, 1983. 13 J. Pfenninger, A. Iqbal, and A. C. Rochat (Ciba-Geigy Ltd.), Eur Pat Appf 184981, 1986. 14 J Pfenninger, A. Iqbal, A. C Rochat, and 0. Wallquist (Ciba-Geigy Ltd.), Eur Pat Appf 184982, 1986. 15 F. Closs and R. Gompper, Angew Chem , 1987,99,564. 16 J. Mizuguchi and G.Wooden, Bei Bunsenges Phys Chem 1991,95, 1264. 17 Z Hao, I Schloeder and A Iqbal (Ciba-Geigy Ltd), Eur Pat Appl 690058 Al, 1994 18 A Whitaker, J Soc Dyers Colour , 1986, 102, 66 19 A Whitaker, in The Analytical Chemirtry of Synthetic Dyes, ed K Venkataraman, Wiley, New York, 1977, p 271 20 D G. Wilkinson (ICI), BP 968,473, 1964. 21 F. F Ehrich (Du Pont), US Patent 3,160,510, 1965 22 A. Whitaker, J Soc Dyers Colourists, 1987, 103(12), 442. 23 Z. Hao, A Iqbal, B Medinger and 0.Wallquist (Ciba-Geigy), Eur Pat Appl 704,497 A 1, 1994. 24 S. J Chapman and A Whitaker, J Soc Dyers Colourists, 1971, 87, 269. 25 C H Griffiths and A. R. Monahan, Mol Cryst Liq Cryst, 1976, 33, 175 26 A. Brockes, Optik, 1964, 21, 550; M. A Maikowski, Bei Bunsenges Phys Chem , 1967, 71, 313 27 P. Hauser, M. Herrmann and B Honigmann, Farbe Lack, 1970,76,545; M. A Maikowski, Prog Colloid Polym Sci , 1976, 59, 70. 28 P. Gunthert, P. Hauser and V. Radtke, Rev frog Coloration, 1989, 19, 41 29 0.Hafner, J Paint Techno1 , 1975, 47, 64. 30 D. H Napper, Polymeric Stabillsation of Colloidal Dispersions, Academic, New York, 1983. 31 P. Gregory, High-Technology Applications of Organic Colorants, Plenum Press, New York, 1991. 32 H. Langhals, Ger Patent 3901988 (1990). 33 Ricoh K. K., Jap Patent 2296-891-A, 1991. 34 J. Mizuguchi and A C Rochat, J Imag Tech, 1991, 17, 123, J. Mizuguchi, G. Giller, and E Baeriswyl, J Appl Phys , 1994, 75, 5 I4 35 A C. Rochat, A Iqbal, R Jeanneret and J. Mizuguchi (Ciba-Geigy Ltd.), Eur Pat App , 187-620, 1986, J. Mizuguchi and A. C. Rochat, J Zmag Sci , 1988, 132, 135; J. Mizuguchi and S. Homma, J Appf Phys , 1989, 66, 3104. 36 Toyo Ink Mfg. K.K., Jap Pat 2055-362-A (1990); Fuji Photo Film K.K., Jap Pat 2039-159-A, 1990; Dainippon Ink Chem K.K., Jap Pat 3011-357-A, 1991; A C Rochat, 0.Wallquist, A. Iqbal and J. Mizuguchi (Ciba-Geigy Ltd.), Eur Par 353-184-A 1990 Received, 14th February 1997 Accepted, 24th March 1997 Chemical Society Reviews, 1997, volume 26 213

ISSN:0306-0012    

DOI:10.1039/CS9972600203

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

into bile duct towam liver and toward small I I Microdialysis sampling coupled on-line to microseparation techniques 66045-0046, USA Microdialysis sampling is a powerful tool for continuously monitoring the extracellular concentration of compounds in tissues in vivo. In order to fully utilize the high temporal resolution but small sample volume of the microdialysis technique, several approaches to coupling microdialysis sampling on-line to microseparation techniques have been developed. This article will describe the analytical chal- lenges of microdialysis sampling. Direct coupling of the microdialysis system to the analytical system can provide many benefits. The application of on-line microseparation techniques will be reviewed. 1 The process of microdialysis sampling Microdialysis sampling is a powerful technique for investigat- ing biochemical events in the extracellular fluid of virtually any tissue, organ or biological fluid.Development of the technique from a long-term dialysis sac implantation1 through push-pull cannulas2 to its present form as a continuous sampling technique3 was accomplished largely by researchers in the neurosciences. Today, microdialysis sampling is a standard technique in the neurosciences and tertiary and secondary literature has begun to appear in the form of books and review~.~-SThe success of the technique in the study of neurotransmitter release has led to the extension of micro-dialysis for general pharmacokinetic, toxicology and ADME studies. Reviews focusing on the use of microdialysis sampling in these area have also been published.9-11 Microdialysis sampling is accomplished by implanting a probe consisting of a hollow fibre dialysis membrane into the organ or biological fluid of interest.The short length of dialysis fibre is affixed to pieces of narrow bore tubing which serve as inlet and outlet tubes. A solution, termed the perfusate, is pumped slowly through the probe. The perfusate is an aqueous solution which closely matches the ionic composition and pH of the surrounding sample matrix. For sampling in vivo from tissue, the sample matrix is the extracellular fluid (ECF). When Craig E. Lunte is Professor of Analytical Chemistry at the University of Kansas, Lawr- ence, Kansas, USA.Professor Lunte received the BS in Chem- istry at the University of Mis-souri-Rolla in 1979 and the PhD in Analytical Chemistry at Purdue University in 1984. His research interest is in bioana- lytical chemistry, with projects in microdialysis sampling, elec- trochemistry, capillary electi-o- phoresis and microbore liquid chromatography. Craig E. Lunte the perfusate is correctly matched to the sample matrix, there should be no net exchange of ions across the membrane. Microdialysis is a diffusion controlled process. The perfusion rate through the probe is generally in the range 0.5-5.0 pl min-1. At this flow rate there is no net transport of liquid across the dialysis membrane. The driving force for mass transport is then the concentration gradient existing between the ECF and the fluid in the probe lumen.Low molecular mass compounds, such as analytes of interest, diffuse into (recovery) or out of (delivery) the probe lumen. Large molecules such as proteins and small molecules bound to proteins are excluded by the membrane. A diagram of the microdialysis process is shown in Fig. 1. Those molecules entering the lumen of the membrane are swept along by the perfusate and exit the probe. The solution leaving the probe, called the dialysate, is collected for analysis. perfusate _____)2:; A IS being recovered P-A from the wmundta matnx by the probe-Fig. 1 Diagram of the microdialysis process. A represents the analyte of interest in the surrounding medium. P-A represents analyte bound to protein and excluded by the membrane. is a small molecule in higher concentration in the perfusate solution than in the surroundings and 0 represents another compound present in both.Malonne I. Davies is a Senior Scientist with Bioanalytical Systems, Inc., West Lafayette, Indiana, USA. She holds a BS degree in Biology and MS de- grees in Biology and Chemistry from Emporia State University and a PhD in Chemistry from the University of Kansas. She is currently the Manager of the BAS-Kansas Research and De- velopment Laboratory in Lawr- ence, Kansas. Her research fo-cuses on in vivo applications of microdialysis. Malonne I. Davies Chemical Society Reviews, 1997, volume 26 215 Malonne I.Daviesa and Craig E. Lunteb a Bioanalytical Systems Kansas Research Laboratory, University of Kansas, Lawrence, KS, USA h Department of Chemistry and the Center for Bioanalytical Research, University of Kansas, Lawrence, KS 1.1 Advantages of microdialysissampling Microdialysis samples represent a local profile of low molecular mass hydrophilic substances in the matrix surrounding the probe. The sampling process excludes small molecules bound to proteins. For compounds of pharmaceutical interest, the dialysate reflects the free fraction of the compound of interest, that is, the therapeutically active portion of the dose. Enzymes are excluded from the dialysate sample so there will be no further enzymatic degradation of the sample.The small size of the dialysis membrane, nominally 300 pm outer diameter (0.d.) and 4-10 mm in length, causes minimum perturbation to the tissue. The technique can be used in awake animals allowing the integrity of tissues, organs and systems to be maintained. Several different probe geometries have been developed to facilitate in vivo sampling from various sites. The diagrams in Fig. 2 illustrate the general probe geometries commonly used. .PE Tubing Cellulose DialysisTubing Into bile duct toward liver and toward small <,,intestine %-Fiber Skeleton Fiber Skeleton the shunt \ Probe not drawn to scale Fig. 2 Typical microdialysis probe geometries.(a) Rigid cannula probe typically used for brain microdialysis.(b) Flexible cannula probe for implantation in a blood vessel of the rat. (c) Linear probe especially suited for peripheral tissue such as skin, muscle or liver. Loop probes, intended for in vitro or subcutaneous sampling, are essentially linear probes with longer membrane windows. (6)Shunt or by-pass probe consists of a linear probe inside a larger tube and is used for sampling from the bile duct of a rat. Since there is no net fluid loss, samples can be collected continuously for hours or days in one animal. The microdialysis probe acts as an artificial blood vessel in that it can both deliver and remove compounds from the local area. Delivery of the parent compound via the probe allows study of local metabo-lism without systemic involvement, Since the membrane excludes proteins and other macromolecules, the dialysate can usually be analysed without further sample clean-up.Each animal serves as its own control and the number of experimental animals needed is reduced. 1.2 Limitations of microdialysissamplingWhile the small size of the microdialysis probe causes minimal perturbation to the tissue, the surgery to implant the probe is invasive. The experimental animals are anaesthetized during probe implantation and the anatomical location of the target tissue dictates the duration of anaesthesia and the severity of the surgical invasion. For example, probe implantations in the dermis or muscle are much less invasive to the animal as a whole than implantation in the liver.Experiments in which the actual tissue concentration of the analyte is desired, as opposed to changes in concentration, require an in vivo determination of the probe’s extraction efficiency, which is generally a time-consuming procedure. Microdialysis typically results in low volume samples. The process inherently dilutes the samples as they are collected. Small volume samples often with low analyte concentration present a considerable challenge to the analytical techniques. 2 Relationship between the microdialysis experiment and the analytical methods At typical perfusion rates, equilibrium is not established across the microdialysis membrane. The concentration of analyte determined in the dialysate is some fraction of the actual concentration in the surrounding sample matrix.The relation-ship between the analyte concentration in the dialysate and that in the sample matrix may be thought of as the extraction efficiency of the probe. Among the parameters that influence extraction efficiency are temperature, perfusate flow rate, chemical and physical properties of the dialysis membrane, probe geometry, membrane surface area and properties of the analyte. Diffusion rate within the matrix also affects extraction efficiency. In vivo uptake into cells, metabolic rate, extent of tissue vascularization and blood flow will influence diffusion through the matrix. Fortunately, under normal conditions of microdialysis sampling these parameters remain constant so that although equilibrium is not established, a steady-state is rapidly achieved.Thus, the extraction efficiency of the probe for a given set of parameters is constant and the direction of net flux of the analyte across the membrane is determined by the concentration gradient of the analyte. From the analytical perspective, there are two important issues with respect to extraction efficiency. First, the dialysisate concentration will be less than the actual tissue concentration of the analyte. Thus, the limit of detection must be lower than the lowest in vivo concentration expected. Secondly, extraction efficiency increases as perfusion rate decreases. The slower the perfusion rate, the closer the dialysate concentration of the analyte will be to that in the tissue surrounding the probe.When microdialysis sampling is applied in vivo, three previously independent systems become interlinked: the ani-mal, the microdialysis sampling system, and the analytical system. The experimentalist must be aware that once these systems are linked, the conditions that were optimal for each independent system must now be considered in relationship to the other systems. Frequently, the analytical method’s sample volume requirement necessitates increasing the microdialysis perfusion rate which in turn lowers the probe’s extraction efficiency and thus provides samples containing lower concen-tration of analyte. Using a lower perfusion rate to increase probe efficiency and analyte concentration results in longer sampling times.Some degree of temporal resolution is lost by this compromise. The increased recovery may deplete low molec-ular mass compounds in the tissue adjacent to the probe in turn perturbing the biological system. The anatomical location and the spatial resolution needed for obtaining the desired informa-tion influence the probe design and active window length. While the implantation of the microdialysis probe may cause little disruption of the target organ, the necessary anaesthesia and extent of the surgical procedure also impact the biological system. The successful use of microdialysis sampling in vivo will depend on achieving a suitable balance among these systems. The trade-offs among perfusate flow rate, concentration detection limit and sample volume requirement will set the temporal resolution that can be achieved for the experiment.An important consideration in balancing these parameters is the in vivo event being investigated and what information about the event is of primary importance. A clear statement of the 216 Chemical Society Reviews, 1997, volume 26 experimental question should dictate the balancing of the microdialysis sampling and analytical method parameters. 2.1 The analytical challenge of microdialysis sampling The analytical method ultimately determines the sensitivity for the substances recovered by microdialysis sampling. Although the sampling step is physically separate from the analysis step, the two are connected by several experimental parameters.Microdialysis is a continuous sampling method in which the response time is determined by the permeability of the microdialysis membrane. On the other hand, most analytical methods require discrete samples of some finite volume. The time needed to collect the discrete sample is far longer than the microdialysis probe’s response time. The dialysate is therefore collected over some appropriate time interval to provide this discrete sample. The sample volume requirement of the analytical method typically determines the overall temporal resolution and not the properties of the sampling itself. For off-line analysis, the injection volume requirement and the perfusate flow rate determine the fastest temporal resolution that can be achieved.Since the small volume of microdialysis samples precludes preconcentration of the sample, the analyt- ical method must have detection limits below the lowest concentration expected in the dialysates. If this condition is not met, samples must be collected over longer time intervals resulting in poorer temporal resolution in the experiment. The analytical method with the lowest concentration detection limit and smallest sample volume requirement provides the best temporal resolution for a microdialysis experiment. Using microbore chromatographic systems injection volumes of 1 ~1 or less are common. Capillary electrophoresis (CE) requires only a few nl be injected. Clearly, a difficulty encountered with microdialysis sampling is the collection, handling and injection of the small volume samples.Any derivatization, extraction or other manipulation of the sample will likely result in consider- able loss of precision in the data. Loss of sample due to transfer and evaporation are reduced by the use of sophisticated fraction collectors and autosamplers. 2.2 On-line approaches Because microdialysis samples are protein-free, the technique is amenable to on-line coupling with the analytical system. The physical devices required to implement on-line analysis vary with the analytical technique used and are discussed below. Regardless of the technique, on-line analysis has several advantages as well as some limitations. A typical on-line system for microdialysis sampling from an awake animal is shown in Fig.3. Such coupling can eliminate problems associated with transferring small volume samples. Direct transfer from the microdialysis sampling system to the analytical system prevents evaporation of samples. When necessary, the dialysate stream can be joined with other solution flow streams to achieve derivatization of the sample prior to injection into the analytical system. In many cases, on-line systems can provide near-real time analysis of microdialysis samples. The temporal resolution of the experiment is a function of the analysis time when microdialysis samples are analysed on-line. The next sample cannot be injected until the previous one has been completely eluted because there is no provision for sample storage. Clearly, a major limitation for coupling microdialysis on-line with an analytical method is analysis speed.One must consider whether the temporal resolution that can be achieved by the analysis is sufficient for the kinetic or physiological events of interest. Continuous analytical techniques such as enzyme reactors and sensors have been coupled to microdialysis sampling to provide near-real time data for analytes such as lactate, glucose, acetylcholine and ethanol. However, as these are generally not separation based methods, they are beyond the scope of this review. 3 Microseparation techniques This review focuses primarily on separation techniques coupled on-line to microdialysis sampling and we have further narrowed the emphasis to microseparations, that is modifications of traditional separation methodologies which accommodate the small volume, low concentration samples typical of micro- dialysis.Microbore liquid chromatography (LC) and open tubular CE, also called capillary zone electrophoresis (CZE) feature most prominently in the microdialysis literature. 3.1 Microbore liquid chromatography The dialysate is a continuous sample stream with low molecular mass compounds in protein-free aqueous solution of high ionic strength. Microdialysis samples are amenable to various analytical chemistry techniques. These include CE, immu- noassay, ion selective electrodes, mass spectrometry and some specific clinical analysers.Liquid chromatography is, however, Fig. 3 An on-line microdialysis-LC system. (Figure used with permission from BAS, Inc. West Lafayette, IN, USA) Chemical Society Reviews, 1997, volume 26 217 by far the most common choice for the analysis of microdialysis samples.I2 A major consideration in favour of LC is its wide availability and its ability to determine more than one analyte at a time. LC is also inherently compatible with high ionic strength aqueous samples such as dialysates. For in vivo experiments, determining multiple analytes provides a clearer picture of the in vivo event, for example, following the changing levels of several neurotransmitters in response to a specific stimulus or providing the concentration time profiles of metabolites in addition to that of the parent compound.The modes of LC most compatible with direct injection of aqueous microdialysis samples are reverse phase and ion- exchange, the choice being dependent on the physicochemical properties of the analyte or analytes of interest. Column parameters (length, internal diameter and particle size) are determined by the sampling interval desired (temporal resolu- tion needed for the experiment) and the sensitivity required. Equivalent separations can, in theory, be obtained with columns of the same length but different internal diameters (i.d.) because column efficiency and analysis time depend on the linear velocity of the mobile phase. Thus, no resolution advantage is gained by the use of narrow bore columns compared to conventional i.d.columns. However, since peak dispersion is proportional to the square of the column diameter, the use of microbore columns yields a tremendous increase in sensitivity. To realize the advantages of high sensitivity and rapid analysis offered by microbore columns, extra column contributions to band-broadening, dead volumes in the injector and flow cell, along with the length of connecting tubing must be minimized. Ref. 13 contains a particularly good discussion of the trade-offs that must be made for any application coupling microdialysis sampling on-line with microbore LC. Other discussions of microdialysis sampling and microbore LC may be found in refs. 12 and 14. On-line injection systems using a six-port valve have been used for monitoring various substances. With this approach, however, valuable information is lost as the dialysate is shunted to waste during the separation.Possible sample carryover due to incomplete flushing of the injection valve is another considera- tion when using on-line analysis. For a conventional LC column with 1 ml min-1 flow rate, a 5 pl sample loop is completely flushed (ten volumes) in approximately 3 s. When a microbore column is used with a flow rate of 50 1.11 min- l, the time required to flush the 5 pl loop with ten volumes of mobile phase will be 1 min. Even if the separation time is not the limiting factor, 1 min of chemical information is lost during the flushing process. Multiple sample loop systems have been described to overcome this limitation.I5 3.2 Capillary electrophoresis Because of the low sample volume requirement of CE, it is particularly attractive for analysis of the small volume micro- A B Syringe Pump#2 S 1-aser dialysis samples.Since injection volumes of 1-10 nl are typical for CE, it is possible to increase microdialysis probe efficiency by lowering the perfusion flow rate while maintaining the temporal resolution of the experiment. CE with UV detection can be used for some analytes at high micromolar concentration, however, optical detection methods suffer from poor concentra- tion detection limits because of the extremely short optical pathlength available with CE. More sensitive methods are generally necessary for many biologically important com-pounds at normal physiological concentrations.In such cases, laser induced fluorescence (LIF) or electrochemical detection are employed for analysis of microdialysis samples by CE. For a recent review of CE separations applied to microdialysis samples see ref. 16. Coupling microdialysis sampling on-line with CE requires an interface to convert the continuous dialysate stream to discrete samples for CE analysis. The interface also serves to protect the experimental animal from the high potential used for CE separations. Additionally, the interface serves to make the microdialysis flow rate compatible with CE. Typical arrange- ments of microdialysis-CE systems are shown in Fig.4. On-line coupling of microdialysis sampling to CE was first described by Hogan et al.17 In this system a commercially available rotary microinjection valve and a specially designed injection interface were combined. The microinjection valve segments the dialysis stream into discrete plugs (60 nl) which are carried to the injection interface by a stream of CE run buffer continuously pumped through a transfer line into the buffer reservoir by electroosmotic flow. A diagram of the injection interface is shown in Fig. 5. The sample plug exits the transfer line into the reservoir directly across from the injection end of the separation capillary. The CE separation voltage is applied continuously throughout the experiment and buffer is con-tinuously drawn into the separation capillary from the reservoir.Electrokinetic injection of the sample plug into the CE takes place as the analyte passes near the orifice of the separation capillary. Any sample not immediately injected is swept away from the injection site by run buffer being pumped through the transfer line behind the sample plug. Careful positioning of the transfer capillary relative to the separation capillary and proper adjustment of the intervening gap are essential for the proper operation of this injection interface. A second microdialysis-CE interface, as reported by Lada and Kennedy, is a modification of the flow-gated interface developed by Lemmo and Jorgensen for coupling capillary chromatography with CE.18 The interface consists of a 75 pm thick PTFE spacer sandwiched between two steel plates (Fig.6). A slit in the spacer produces a small volume flow channel between the plates. The outlet of the microdialysis probe and the inlet of the CE separation capillary are positioned directly opposite one another at a distance nominally defined by the spacer. Two additional ports are used for introduction and High Positive Voltage I Fig. 4 Typical on-line microdialysis-CE systems. A. Arrangement of an on-line microdialysis-CE system with laser induced fluorescence detection. B. On-line microdialysis-CE system with electrochemical detection. 218 Chemical Society Reviews, 1997, volume 26 Fig. 5 Detail of the gap design injection interface used with an electrically actuated valve.A buffer reservoir is machined in a kelef block. A channel drilled the length of the block and threaded for LC fittings contains a length of PEEK tubing with an opening as shown in the close-up. The opening exposes the fracture between the transfer line from the valve and the separation capillary to the running buffer. removal of electrophoretic buffer in a flow pattern perpendi-cular to the dialysis-CE capillary flow. During the CE run, the gating flow rinses away the dialysate from the inlet of the CE and provides the electrophoretic buffer necessary for the separation. Injection of analyte into the CE is accomplished by stopping the gating flow, turning off the separation voltage thus allowing dialysate to build up near the CE inlet.After an appropriate delay period, a low voltage is applied for a few seconds to inject the sample. The voltage is turned off, the gating flow is resumed sweeping away excess dialysate, and the separation voltage is applied to the capillary. The very low dead volume of this interface allows low nanolitre flow rates through the microdialysis probe while maintaining good temporal resolution. As with the microinjection valve-interface coupling, correct positioning of the dialysis and CE capillaries, adjust-ment of the gating flow rate and delay time are necessary for proper operation of the interface. A disadvantage of both designs is that only a portion of the sample stream is utilized in the analysis. As a result, rapid changes in concentration that occur between runs may not be observed.Additionally, the voltage switching required with the flow gated interface can cause baseline drift and could be a major problem for electrochemical detection. 3.3 Other techniques Mass spectrometry (MS) can be a useful tool for quantitative analysis of polar molecules in aqueous environments. However, coupling MS directly on-line with microdialysis sampling presents significant challenges. Microdialysis samples, espe-cially those collected in vivo, are collected in physiological saline or other high salt content solution. This high salt content is not compatible with the MS inlet. The sensitivity of MS may not be adequate for all analytes of interest.Despite these challenges, tandem MS has been used on-line with micro-dialysis sampling. Gating-Flow Dialysis 4 Review of applications in the recent literature The examples of microdialysis sampling coupled on-line to the analytical system cited in this review are intended to provide an overview of the current scope of applications. Most of the work noted here was published between 1993 and 1996. Earlier reviews of microdialysis sampling coupled on-line with analyt-ical techniques can be found in refs. 19 and 9. A searchable bibliography of reports on microdialysis sampling is also available on computer disk.20While microdialysis sampling has proved to be an excellent technique for in vitro applications such as drug dissolution studies, kinetics in enzyme incubations and monitoring bioprocessors, we have further limited the applications discussed here to in vivo studies.We have chosen to separate the applications on the basis of the analytical method used rather than the tissue sample or the analyte. 4.1 Microbore liquid chromatography As noted above, refs, 12-14 contain helpful discussions of the implementation of microbore liquid chromatography for the on-line analysis of microdialysis samples. A summary of applica-tions with information about microdialysis and analytical parameters is provided in Table 1. Two loops and a multifunction ten-port valve were used by Wang and co-workers for on-line LC-UV analysis to study the distribution of zidovudine using microdialysis probes simul-taneously sampling from cerebrospinal fluid and thalamus of rabbits.21 Dialysate from the probes collected into separate sample loops (5 pl) for sequential injection onto a small bore column (2.1 X 200 mm).The sampling interval for each probe was 20 min. The report also compared in vivo calibration of the microdialysis probes using retrodialyis and the zero net flux method. Wang and associates used a small bore column (2.1 X 200 mm) with fluorescence detection for the on-line analysis of microdialysis samples in a study of the effect of cyclosporin A on the distribution of rhodamine-123 in rat brain.22 Perfusate flow rate was 0.5 pl min-1 into the loop and the injection interval was 40 min. In a summary of their recent results, Lambas-Sefias and colleagues characterize on-line microdialysis sampling coupled to LC with electrochemical (EC) detection and in vivo voltammetry as complementary techniques for monitoring monoamine metabolism in rat brain stem.23 The LC separation of DOPAC and catecholamine was performed on a 2,l X 100 mm column with an interval of 15 min.While in vivo voltammetry provided better spatial and temporal resolution than microdialysis, microdialysis sampling with separation based analysis provided more selective and precise identifica-tion of compounds from the extracellular fluid. Michelsen and Pettersson analysed contrast agents, such as those used in X-ray and magnetic resonance imaging, to demonstrate an on-line microdialysis-LC-MS system.24 Dialy-sate from a probe implanted in the carotid vein was collected directly into the loop of the electrically actuated injection valve.Separation was achieved using a narrow bore column (2 X 150 Detector Interface 414 Voltage Fig. 6 Diagram of the on-line microdialysis-CE system based on the flow-gated interface showing the details of the flow-gated interface. (Reproduced with permission from Anal. Chim. Acra., 1995, 307, 217.18 Chemical Society Reviews, 1997, volume 26 219 Table 1 Applications of microdialysis coupled on-line to microbore liquid chromatographya Microdialysis parameters Probe Analytes Tissue Design Rhodamine- 123 frontal cortex of RC rat brain Zidovudine (AZT) rabbit brain, RC AZdU (as an internal standard) thalmus and ventricle DOPAC rat brain RC catacholiamine Omnipaque (350 mg carotid vein of RC iodine/ml) rat dopamine and rat brain RC metabolites dopamine rat brain RC [I] acetylcholine striatum of rat RC brain [2] 8 monoamines serotonin jugular vein of FC (5-hydroxytryptamine) anaesthetized rats APAP jugular vein of FC APAP-0-sulfate awake rats APAP-glucuroinide APAP and metabolites jugular vein of FC awake rats caffeine and metabolites Substance P striatum of rat RC brain (1 Probe Designs: FC-flexible cannula; RC-rigid Membrane Perfusate Flow rate MWCO Length (pl/min) 20 000 3 mm 0.5 20 000 3mm 1 6000 2mm 1 20 000 10mm2 n.g.4 mm vanous n.g.4mm 0.2 20 000 3mm2 20 000 10 mm 0.5 5000 4mm 1 5000 n.g. 1.5 20 000 4 mm 0.3 Analytical system details Column Loop Injection I.D. X length Volume Interval (mm) n.g. 40 rnin 2.1 X 200 2 loops, 20 min 2.1 x 200 each alter- 5 ~1 nating loops 15 pl 15 rnin 2.1 X 100 5 yl 10 min 2X 150 n.g. 5 min 1 X various 0.5 pl 5 min 1 X 100 5 wl 20 min [I] 1 X530 1 x 55 121 I x 100 10 yl 20 min 1 X 100 2 loops, 5 min 1 x 100 each 7 PI 0.5 yl 1 min 1 X 14 internal loop 10~1 30min 0.05 X50 Detection Mode Notes Ref. FL 21 uv Used 2 probes in 22 different location, alternating injections EC 23 MS and 24 photo-diode arrayEC push-pull perfusion 13 of the microdialysis probe EC push-pull perfusion of 25 the microdialysis probe EC 2 injection loops in 26 series injected microdialysis samples simultaneously into two LC systems EC 27 uv Dual 6-port and 8-port 15 valves compared uv APAP and 2 metabolites 28 were resolved in < 1 min Caffeine and 2 metabolites also resolved in < 1 min MS 29 cannula.Detection Modes: FL-fluorescence; UV-ultraviolet; EC-electrochemical; MS-mass spectrometry. mm) with a mobile phase flow rate of 300 pl min-1. Flow from the LC column was split to achieve a flow rate of 80 pl min-1 before entering the ion pneumatically assisted electrospray source. On-line analysis of microdialysis samples using 1 mm i.d. columns and EC detection has been used to quantify neuro- transmitters.*9Wages et al. used various dopamine metabolites to illustrate the impact of microdialysis sampling and chromato- graphic system parameters on experimental results when using on-line microbore systems.13 Their study characterized the system performance both in vitro and in vivo. Church and Justice demonstrated the use of such a system for studying neurochemical responses to pharmacological agents.25 Ex-tracellular dopamine in the brain of awake rats was monitored on-line at 5 min intervals following administration of haloper- idol (increased extracellular dopamine) or apomorphine (de- creased extracellular dopamine). Tsai and Chen measured acetylcholine and monoamines using two microdialysis on-line systems with two analytical systems simultaneously.26 Dialysate samples were collected into two automated injection valves (5 pl loops) connected in series and injected at 20 min intervals onto the two LC systems. Acetylcholine (ACh) was determined using a prepacked BAS ACh microbore column (1 X 530 mm) and post-column reactor (1 X 55 mm) with EC detection at a platinum electrode.Monoamines were analysed on the second system using a microbore column (1 x 100 mm) and EC detection at a glassy carbon electrode. In another study, Tsai and Chen studied collagen-induced serotonin release in the blood of anaesthetized rats.27 The microdialysis probe was implanted in the jugular vein and dialysate collected directly into the 10p1loop of an on- line injector.With a sampling interval of 20 min, serotonin and 220 Chemical Society Reviews, 1997, volume 26 its major metabolite, 5-hydroxyindoleacetic acid, were deter- mined using a 1 X 100 mm column and EC detection. Steele and Lunte demonstrated the potential of microdialysis sampling coupled to on-line microbore LC analysis for pharmacokinetic studies.I5 The separation of acetaminophen and its two conjugated metabolites was achieved using a microbore column (1 X 100 mm). Injector precision, on-line system response time and system delay time were evaluated for two on-line systems, a dual six-port valve system and an eight- port valve system. The eight-port valve system was used on-line with flexible microdialysis probes implanted in the jugular vein of rats to study the in vivo pharmacokinetics of an intravenous dose of acetaminophen.The dialysate was collected directly into the sample loops (7 pl in all cases) for injection into the LC-UV system. The LC injection interval was 5 min. Fast microbore liquid chromatography was coupled on-line with venous microdialysis sampling by Chen and Lunte.28 Using a 1 X 14 mm microbore column, analysis times of less than 1 min were achieved for acetaminophen and its conjugate metabolites and for caffeine and two metabolites. In this case, the electrically actuated injection valve had a 0.5 pl internal sample loop. The chromatographic detector output directly shows the concentration-time profile of the drug with 1 rnin temporal resolution (Fig.7). LC-MS with micro-electrospray ionization has also been coupled to in vivo microdialysis sampling to study the metabolism of Substance P.29 Dialysate samples were collected directly into the loop of a ten-port valve for loading onto a capillary column (0.05 X 50 mm). Following washing with a weak mobile phase to remove most ionic compounds, a stronger mobile phase was used to elute Substance P and its metabolite fragments into the ion source. rats was carried out. Concentration changes of the analytes, triggered by infusing potassium via the probe, demonstrated the feasibility of the system for neurochemical applications. Lada et al., using a flow gated interface, already discussed, coupled microdialysis with CE-UV for the determination of ascorbate in rat brain.'* The migration time for the ascorbate under the conditions of the in vivo experiment was ca.100 s with the experiment having a temporal resolution of 110 s. I 1 ' ' ' Using the same system, Lada and Kennedy studied ascorbate i5'' ' ' 'k' ' ' 'A' ' ' 'A'I ' ' ' '1' ' 'i,, and lactate in vivo in rats monitoring concentration changes in tlmin response to administration of anaesthesia or to elevated.. ... . . . potassium le~els.3~ Incorporating o-phthaldehydelp-mercap-toethanol derivatization with the on-line system and using a micellar electrokinetic chromatography with LIF detection, they have also monitored aspartate and glutamate in rat brain dialysates.32 0 10 20 30 40 50 60 70 80 90 100 Although CE-EC has been used off-line for a variety of tlmin substances in microdialysis samples, the only report coupling Fig.7 Detector output from the on-line microdialysis sampling-fast microdialysis sampling on-line with CE-EC determined nico- microbore LC system following the plasma concentration of caffeine and its tine from a dermal patch in the skin of a rat.33 The system main metabolites. Inset shows expanded scale of output from 55-60 min. requires an interface at each end of the separation capillary to (Reproduced with permission from J. Chrom. A, 1995, 691, 29.28 shield both the experimental animal and the EC detector from the high separation potential. If the separation current is not grounded before the detector, the relatively small analytical 4.2 Capillary electrophoresis signal generated at the electrode will be lost in the separation Table 2 presents a summary of microdialysis and analytical current.parameters for applications coupling microdialysis on-line with Coupling MS detection to CE currently results in substantial capillary electrophoresis. For additional applications using CE loss of sensitivity compared to LC due to the small amounts of analysis with microdialysis studies, including off-line analysis, analyte that are delivered to the MS by the CE. The literature the reader should consult ref. 16. searches conducted for this review found only one report of CE- The first reported on-line microdialysis-CE system was MS for the analysis of microdialysis samples.Although the based on an electrically actuated micr0injector.~7 Using LIF microdialysis system is not on-line with the CE-MS, Takada detection, Hogan et al. investigated a potential antineoplastic et al. recently reported the detection of y-aminobutyric acid agent, SR 4233, and its main metabolite, SR 4317. A high-speed (GABA), a well studied neurotransmitter, in samples collected micellar electrokinetic chromatographic (also called micellar in vivo from rat brain by microdialysis.34 Previous experiments electrokinetic capillary chromatography) separation provided by the group had determined the limit of detection of GABA for 90 s temporal resolution for in vivo experiments. the CE-MS system was 10-5 M using standard solutions. Zhou and colleagues modified the system to include on-line However, GABA in microdialysis samples was not detected NDA/CN derivatization of amino acids.30 Although the free under the same analytical conditions. After modification of the zone electrophoretic separation of glutamate and aspartate analytical conditions they were able to obtain a signal for required less than 2 min, the overall temporal resolution of the GABA.Although this signal was insufficient for quantitation, system optimized for the derivatization was 5-7 min. Con-the results do suggest that coupling microdialysis sampling on- tinuous monitoring of glutamate and aspartate in the brains of line to CE-MS may eventually be successful. Table 2 Applications of microdialysis coupled on-line to capillary electrophoresisu ~~~~~~~~ ~ ~ Microdialysis parameters Analytical system details Perfusate Capillary I.D.X Membrane flow total :working Probe rate/ Injection Injection lengths Detection Analytes Tissue design MWCO Length ~1 min-I Regime Interval (pm X cm :cm) Mode Notes Ref. SR4233, jugular vein FC 5000 n.g. 1 60 nl plug delivered 90 s 50X40:15 LIF 17 a benzotnazine of rat to interface, SR43 17, electrokinetic metabolite injection glutamate hippocampus RC 20000 3mm 1 60 nl plug delivered 2 min 25 X 30: 14 LIF on-line 30 asp art ate of rat brain to interface, derivatization electrokinetic injection ascorbate rat brain linear, U 6000 2 mm 0.1 10 s at 1 kV, 10 s 110 s 25 x 50-60: 15 UV 18 delay ascorbate rat brain linear, U 6000 2 mm 0.079 5 s at 1 kV 50 s to 25 or 50 X 50: 15 uv 31 1actate 2 min aspartate rat brain linear, U 6000 2mm 0.079 5 sat 1 kV or 0.5 to 25 X 20: 15 or 10 LIF on-line 32 glutamate 2sat 1OOV 3 min derivatization nicotine dermis of linear 30000 15 mm 1 60 nl plug delivered 10 min n.g.EC 33 awake rat to interface, electrokinetic injection GABA striatum of RC 20000 3mm 1 5 nl 50X40:40 MS microdialysis to CE 34 rat brain on-line, detection off-line a Probe Designs: FC-flexible cannula; RC-rigid cannula. Detection Modes: LIF-lasar induced fluorescence; UV-ultraviolet; EC-electrochemical; MS-mass spectrometry. Chemical Society Reviews, 1997, volume 26 221 4.3 Other analytical approaches Tandem mass spectrometry (MS-MS) is another separation based analytical approach that has been coupled on-line to microdialysis sampling.36 Caprioli and Lin directed the dialy- sate flow into the MS by way of the continuous-flow-fast-atom bombardment (FAB) interface.35 Dialysate samples from a microdialysis probe implanted in the jugular vein of a rat were analysed on-line to provide a concentration-time profile of penicillin following an intramuscular dose Deterding and co- workers also used a FAB interface to couple microdialysis on- line to MS-MS Using tris(2-chloroethyl) phosphate as the model compound they compared the blood pharmacokinetics obtained by the on-line microdialysis-MS-MS system with those from conventional methods (serial blood sampling) 36 5 References 1 L Bito, H Davson, E Levin, M Murray and N Snider, J Neurochern , 1966,13, 1057 2 J M R Delgado, R V DeFeudis, R H Roth, D K Ryugo and B M Mitruke, Arch Int Pharmacodyn , 1972, 198, 9 3 T Zetterstrom, L Vernet, U Ungerstedt, U Tossman, B Jonzon and B B Fredholm, Neurosci Lett, 1982, 29, 11 1 4 Microdialysis in the Neurosciences, ed T E Robinson and J B Justice, Elsevier Amsterdam, 199 1 5 K M Kendnck, Method Enzymol , 1989,168, 182 6 H Benveniste and P C Huttemeier, Prog Neurobiol , 1990, 35, 195 7 J Kehr, J Neurosci Method, 1993, 48, 251 8 B H C Westerink, Behav Brain Res , 1995, 70, 103 9 C M Riley, J M Ault and C E Lunte, in Pharmaceutical and Biomedical Applications of Liquid Chromatography, ed C M Riley, W J Lough and I W Wainer, Elsevier, Oxford, UK, 1994 10 C E Lunte, in Drug Toxicodynamics, ed D E Johnson, Marcel Dekker, New York, 1995 11 S M Lunte and C E Lunte, in Advances in Chiornatography, ed E Grushka and P Brown, Marcel Dekker, New York, 1996, pp 383-432 12 P T Kissinger, in Microdialysis in the Neurosciences, ed T E Robinson and J B Justice, Elsevier, Amsterdam, 1991, pp 103-1 15 13 S A Wages, W H Church and J B Justice, Anal Chern , 1986, 58, 1649 14 H 0 Pettit and J B Justice, in Microdialysis in the Neurosciences, ed T E Robinson and J B Justice, Elsevier, Amsterdam, 1991, pp 117-153 15 K M Steele and C E Lunte, J Pharrn Biomed Anal, 1995, 13(2), 149 16 D K Hansen and S M Lunte, Am Lab, June 1996,28, 26 17 B L Hogan, S M Lunte, J F Stobaugh and C E Lunte, Anal Chern , 1994,66,596 18 M W Lada, G Shaller, M H Carriger, T W Vickroy and R T Kennedy, Anal Chim Acta, 1995,307, 217 19 W H Church and J B Justice, in Advances in Chromatography, ed J C Giddings, E Grushka and P R Prown, Marcel Dekker, New York and Basel, 1989,pp 165-194 20 Bibliography of Microdialysis on Disk, ed J M Gitzen, Bioanalytical Systems Press, West Lafayette, IN, 1996 21 Y Wang, S L Wong and R J Sawchuk, Pharm Res, 1993, 10, 1411 22 Q Wang,H Yang,D W MillerandW F Elmquist,Biochem Biophys Res Cornrnun, 1995,211,719 23 L Lambas Sefias, C Vachette, F Robert, C Ortemann and B Renaurd, Clin Exp Hypert ,1995,17, 129 24 P Michelsen and G Pettersson, Rapid Cornrnun Mass Spec, 1994, 8, 517 25 W H Church and J B Justice, Anal Chem , 1987, 59, 712 26 T Tsai and C Chen, Neurosci Lett, 1994, 166, 175 27 T Tsai and C Chen, J Chrorn A, 1996, 730, 121 28 A Chen and C E Lunte, J Chrorn A, 1995, 691, 29 29 P E Andren and R M Caprioli, J Mass Spec, 1995,30, 817 30 S Y Zhou, H Zuo, J F Stobaugh, C E Lunte and S M Lunte, Anal Chem , 1995,67, 594 31 M W Lada and R T Kennedy, J Neurosci Method, 1995,63, 147 32 M W Lada and R T Kennedy, Anal Chem , 1996,68, 2790 33 J Zhou, H Zuo, D M Heckert, C E Lunte and S M Lunte, J Pharm Biomed Anal, 1996, in the press 34 Y Takada, M Yoshida, M Sakain and H Koizumi, Rapid Cornmun Mass Spec, 1995,9, 895 35 R M Caprioli and S Lin, Proc Natl Acad of Sci USA, 1990, 87, 240 36 L J Deterding, K Dix, L T Burka and K B Tomer, Anal Chem , 1992,64,2636 Received, 8th October 1996 Accepted, 27th January 1997 222 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600215

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Intramolecular vibrational energy redistribution has been much studied because of the important role it plays in theories of unimolecular reactions and in dictating the feasibility of bond-selective chemistry. In this review we survey some of the more recent experimental studies and hope to provide a general introduction along with some present day conclusions. 1 Introduction Excited molecules are often a prerequisite to chemical reaction, and as a consequence there has been much interest in their behaviour. When molecules are highly excited they defy our attempts to describe their energy levels with simple expressions and exhibit a phenomenon known as intramolecular vibrational energy redistribution, or IVR. Here, the initial localised energy of an isolated molecule that has been excited by an external perturbation redistributes to reside ultimately in other parts of the molecule.This energy redistribution may be studied by monitoring the evolution of the population in the initially prepared state. Experimental studies of IVR generally involve the radiative excitation of gas phase molecules. The choice of the gas phase enables the study of truly isolated molecules, and thus of intramolecular dynamics that are not complicated by solvent effects. If a particular superposition of molecular eigenstates (stationary states) with a particular phase relationship is prepared in such gas phase molecules by radiative excitation it is commonly known as a bright state. Similarly, those superpositions which are not radiatively coupled to the ground state are labelled dark states.The initial phase relationship of the eigenstates forming the bright state may, for example, corre- spond to predominantly the motion of a single excited bond in the molecule. The differing phase dependence of each of these eigenstates causes the bright state to be nonstationary, i.e. to evolve in time. This evolution causes the molecule to behave more like a classical oscillator, with certain bonds stretched or compressed at certain times. If enough eigenstates form the superposition then the initial phase relationship (e.g. as above, Dean Boyall gained a BSc in chemistry from Nottingham University in 1996. He is cur- rently studying for a PhD in organic chemistry.This article emanates from a literature pro- ject that he undertook as afinul year undergraduate under the supervision of Dr Reid. Katharine Reid gained a BSc and PhD in chemical physics from Sussex University. She then spent twoyears as a SERCI Dean Boyall NATO fellow in the group Of single bond stretched) will never be recovered and the excitation energy will appear to be randomized after some time known as the IVR lifetime. Because the forms of the true molecular eigenstates are often unknown, and because most experiments are unable to resolve eigenstates, the bright and dark states are usually expressed in terms of basis functions that are orthonormal in some reason- able first approximation Hamiltonian.These functions corre- spond to excitation in a set of independent normal modes. Articles on IVR therefore often refer to the coupling between these basis functions by anharmonic or Coriolis interactions in the true Hamiltonian as causing IVR, rather than the cause being phase evolution of the true molecular eigenstates, but either language can be adopted. Here, for the most part, we use the former. The existence of IVR as a ubiquitous rapid process is central to the 1952 Rice-Ramsperger-Kassel-Marcus (RRKM) theory of unimolecular reactions. 1 In this theory, it is assumed that any initial energy of excitation is redistributed across the whole molecule so rapidly that if the original excitation were localized, a rupturing bond would display no memory of this localization.Clearly, in a small molecule with well-separated energy levels there will be instances where RRKM theory will be in-appropriate as there is no mechanism by which the energy can flow, but in a large molecule, energy levels are closely spaced and RRKM theory often provides a good description of reaction rates. Following the initial formulation of the RRKM theory there have been many experimental studies of IVR designed to test the predictions of the theory and to search for systems that display non-RRKM behaviour. The first experimental evidence of the existence of IVR was provided by Butler et a1.2 in 1960 and the first direct measurement of an IVR rate was provided by Deutsch et al. on SF6 in 1977.3 The interest in non RRKM systems and IVR rates stems from the long dreamed of possibility of bond-selective (or mode-selective) chemistry in which bond rupture would predominantly occur in the bond where excitation was initially localised.This possibility has led to a quest for molecules which display redistribution rates that Professor R. N. Zure at Stun- ford University where she stud- ied molecular photoionization dynamics. She returned to the UK in 1992 as an SERC Ad- vanced Fellow in the chemistry department at Nottingham Uni- versity, and took up a lecture- ship in physical chemistry in the same department in 1995. Her current research interests in- clude photoionization, intramo- lecular dynamics, and the use of polarization as a tool in dynam- Katherine Reid ical studies.Chemical Society Reviews, 1997, volume 26 223 are slow compared with the interval between molecular collisions. Despite the wealth of experimental data on IVR, some of which can be found in Table 1, its description has for the most Table 1 Representative IVR lifetimes in the ground electronic state of various molecules. All measurements were made by infrared spectroscopy in the frequency domain except where indicated, and the FWHM lifetimes are given. Where the number of vibrational quanta is not given, the lifetime pertains to the fundamental vibration. Molecule Vibration Ref. Lifetime HC=CC=CH CH stretch, v = 2 12a 165 ps HC=CC=CD (v1 + v2) CH stretch, v = 3 12a 50 ps 2.7 ns (CH3)3CGCH Acetylenic CH, v = 1 24 200 ps Acetylenic CH, v Acetylenic CH, v = 2 = 1 24 110 ps 2000 ps Acetylenic CH, v = 2 4000 ps CHSi(CECH)~ Acetylenic CH, v = 1 12e 350 ps vl3 490 ps Acetylenic CH, v = 1 12c 60 ps Acetylenic CH, v = 2 1 PS Acetylenic CH, v Acetylenic CH, v = 1 = 2 12c 40 ps < 20 ps Acetylenic CH, v = 1 12c 850 ps Acetylenic CH, v = 2 140 ps CH~CECH Acetylenic CH, v = 1 12b 110 ps Acetylenic CH, v = 2 320 ps v1+ 2vg 3.2 ns Ethanol OH stretch 21 25 ps Methyl CH 123 ps CF3GCH Acetylenic CH 29 2 ns HC=C=CH 2v, +v5 25 210 ps 2v6 180 ps But-l-yne Acetylenic CH 27 285 ps Methyl CH 320 ps Pent-1-yne Acetylenic CH Methyl CH 27 241 ps < 40 ps Prop-2-ynyl alcohol OH stretchn 27 110 ps Acetylenic CH 27 250 ps But-1-ene Methyl CH 27 37 ps Butane Methyl CH 27 18 ps Methyl formate Methyl CH 27 70 ps trans-But-2-ene Methyl CH 27 132 ps terr-Buty lacetylene Acetylenic CH 27 200 ps 2-Fluoroethanol Methyl CH 27 565 ps 1,2-Difluoroethane Methyl CH 27 490 ps Isobutane Methyl CH 27 < 177 ps trans-1 -Chloro-2- Methyl CH 27 No Decay fluoroethane SFSCZCH Acetylenic CH, v = 1 14 3 ns Acetylenic CH, v = 2 1.5 ns a Time resolved measurement part eluded theoreticians and in many respects it remains poorly understood. A number of simple models have been developed that seem to provide good explanations for some molecules in some states, but none has so far proved universal.There have been many attempts to rationalise IVR rates in terms of the number of the molecular eigenstates per unit energy, or density of states, but these have not been entirely successful (see Section 4). A common theme in IVR that is related to the density of states is the existence of different regimes of excitation which give rise to qualitatively different IVR dynamic^.^ Within this description, a very slow rate (or long lifetime compared with the inverse of the average energy spacing) would correspond to the population of a single eigenstate in a sparse region [Fig. l(a)]. Conversely, a short lifetime compared with the inverse of the average energy spacing would correspond to a rapid statistical redistribution mediated by a very high density of states.In this fast regime of IVR, energy is redistributed to parts of the molecule distant from the region initially excited too rapidly for any recurrence to be possible. This redistribution is usually an 224 ChemicalSociety Reviews, 1997,volume 26 extremely rapid exponential decay of energy but may take the form of a biexponential with fast and slow components [Fig. l(c)]. In between these two, a region exists in which there is a partial recurrence of the population of the initially prepared state at certain well-defined times [Fig. 1(b)].This phenomenon is known as quantum beats and is analogous to vibrational wavepacket motion with the recurrence time giving an indica- tion of the level spacing.J 1 I I I I I I I 0 t Ins 7 1 8 I I I I I 0 t Ins 6 I 7--r I ---0 t Ins 7 Fig. 1 Fluorescence decay of anthracene in its Sl electronic state at vibrational energies of (a)390 cm-I, (b) 1420 cm-1 and (c) 1792 cm-1 with corresponding densities of states of (a) 10, (b)25-40 and (c) 120per cm-I. Clearly there is little redistribution of energy in (a), the emergence of quantum beats in (b)at a higher density of states and a rapid biexponential decay in (c). (Adapted from P. M. Felker and A. H. Zewail, J. Chem.Phys., 1985, 82, pp. 2961 with kind permission of the American Institute of Physics). This review is intended to serve two purposes: (i) to survey, collate and attempt to rationalise some of the mass of experimental observations and (ii) to extract measured IVR lifetimes from a number of sources in the hope of finding an underlying pattern, and systems that display particularly long lifetimes. It is organised as follows: experimental techniques used to probe IVR are reviewed and the origin of IVR is discussed.Possible rationalisations of IVR lifetimes are con- sidered in terms of the density of states and simple models that have been developed. Finally, a brief section focuses on the possibility of mode-selective chemistry. 2 Experimental techniques Numerous spectroscopic methods are employed to probe IVR in polyatomic systems, in both excited and ground electronic states. Furthermore it is possible to investigate IVR in both frequency-resolved and time-resolved domains. The use of supersonic jet-cooling causes spectral simplification.This has revolutionised IVR studies and provides the basis of most recent experiments. Here, brief details of some relevant techniques will be presented which indicate the scope of information obtainable and limitations where appropriate. The techniques discussed will be referred to in the following sections. 2.1 Excited state Although IVR in the electronic ground state is of most chemical interest, IVR was first studied in electronically excited states of molecules. The main reason for this was the availability of powerful, tunable light sources, primarily lasers, operating in the visible region of the spectrum.These light sources enabled the preparation of states in a well-defined energy range, and a built-in clock was provided by the radiative lifetime of the electronic state. These advantages have meant that excited state studies continue to be actively pursued. In principle, it is possible to prepare any vibrational level in a mode that is of a symmetry that can be accessed, but the probability of preparing each level is governed by Franck-Condon factors (vibrational overlap integrals) that usually make the higher levels in- accessible. Typically, the excitation laser prepares a bright state (see Section 1) in the first electronically excited state (where the electronic states are labelled So and S1 for ground and first excited singlet states, respectively).The extent of IVR can be deduced from either the width of the absorption profile, or the appearance of spectral congestion in emission. The full width at half maximum (FWHM) of the absorption band reflects the reciprocal of the IVR lifetime, however there are some disadvantages in attempting to extract this: (i) a known mathematical function must be fitted to the band, and (ii) the broadening of the band may not be entirely IVR related as Doppler and collisional broadening can also occur, both of which would imply a faster IVR process than the true one. It is also possible to resolve the bright state in the time domain by observing the fluorescence at differing time delays following excitation, enabling the evolution of the population of the bright state to be monitored with time.If fluorescence is detected at differing time delays then IVR is manifested by the appearance of spectral congestion at longer time delays. It is possible to monitor only the bright state population by passing the fluorescence through a monochromator and selecting the frequency corresponding to bright state emission. This direct population monitoring is invaluable to IVR studies since it provides a direct probe of how the energy is redistributed and provides a clear indication of whether IVR is in the sparse, intermediate or statistical regime. This type of information is not always inferable from time integrated studies. The lifetime of the bright state can be extracted as the l/e point of a fitted exponential decay curve of the bright state.Time-resolved studies of this nature require manipulation of the time delay between excitation and observation of the fluorescence. Parmenter and coworkers pioneered a method called chemical timing which is a classic way of imposing picosecond timing by reducing the S1 fluorescence lifetimes with an added quencher gas, usually oxygen.5 In the absence of quencher gas the lifetime will be the inherent radiative lifetime. As the pressure of quencher gas is increased, collisions with activated molecules relax them enabling only the fluorescence in a time ‘gate’ between activation and collisional relaxation to be observed. Because the bright and dark states are collisionally relaxed at the same time and rate, the fluorescence will indicate IVR through the emergence of unstructured emission as the pressure of quencher is decreased.Chemical timing has one substantial disadvantage: because not all molecules are quenched at precisely the same time there is a distribution of lifetimes. This will manifest itself as congestion in the spectrum because the molecules that are quenched later will show enhanced unstructured emission as IVR is more complete, compared with molecules that are quenched earlier. Chemical timing was developed for the purpose of providing quantitative data for IVR processes in systems where IVR is the dominant relaxation pathway, and an estimate of the IVR lifetime can be gained from the ratio of the intensity of structured to unstructured emission.An accurate determination of lifetime however requires mathematical modelling and involves the assumption of either statistical or intermediate IVR. Thus, chemical timing is an excellent way to reveal the presence of IVR but limited in its ability to readily provide accurate lifetimes. The advent of ultrafast lasers was a great redeemer as changing the time delay between pump and probe pulses allows controllable time resolution. However, it is not unusual for these lasers to have pulse widths of the order of femtoseconds which severely limits their frequency resolution. This inherent prob- lem makes it difficult to probe sufficiently few selected levels, especially high in the vibrational manifold where there is a high density of states.Although time-resolved studies and lineshape analysis should provide complementary information, discrep- ancies have arisen when comparisons have been performed between the two methods. Zewail and coworkers, the pioneers of ‘real time’ measurements, have used picosecond molecular beam techniques to record the fluorescence spectra of a number of aromatic molecules such as alkylanilines. They found that for large and complex molecules of this type spectral broadening does not necessarily indicate the rate of energy redistribution. This is especially important at high excitation energies where dispersed fluorescence spectra of different bands tend to be similar in appearance and it may be misleading to infer the dissipative nature of an IVR process by consideration of low resolution frequency-resolved spectra alone.Felker and Zewail studied jet-cooled anthracene using both frequency and time resolution to illustrate this point.6 The dispersed fluorescence of unrelaxed and relaxed vibrational states in S were separated with a monochromator. Fig. 2 illustrates that the differences are more apparent in the time-resolved results than in the fre- quency-resolved spectrum. Evlb= 1792 cm-’ J 0 2 4 6 3400 3500 3600 3700 3800 3900 t Ins h/A Fig. 2 Unrelaxed time-resolved (left column) and frequency- resolved (right column) spectra of anthracene for two different S1 vibrational levels. The time-resolved spectra show much clearer differences between the two levels with quantum beats appearing at 1420 cm-’ and an exponential decay at 1792 cm-1. (Adapted from P.M. Felker and A. H. Zewail, Chem. Phys. Lett., 108 (4).Direct picosecond time resolution of IVR in isolated molecules, p. 303, 1984 with kind permission of Elsevier Science, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands). A relatively new technique has been named time-resolved fluorescence depletion spectroscopy (TRFD) and was pio-neered by McDonald and coworkers.7 Two laser pulses of identical wavelength and duration are used to create and stimulate emission from a vibrational state in S1. Stimulated emission intensity is measured as a function of the delay between the two pulses, and by monitoring the efficiency with which the second pulse can stimulate emission the evolution of Chemical Society Reviews, 1997, volume 26 225 the prepared state can be followed.IVR occurring during the delay causes the prepared state to evolve and reduces the ability of the second pulse to stimulate emission, with the resolution only depending upon the bandwidth of the laser pulse. Thus, there is a large decrease in fluorescence intensity if IVR has occurred. The resulting lineshapes are fitted to an empirical form to deduce the overall IVR lifetime. The technique can also reveal quantum beats which are inferred from the appearance of undulations associated with the band. Excited state IVR may also be studied by ionization using a technique based upon photoelectron spectroscopy.The ZEKE (zero kinetic energy photoelectron spectroscopy) method in- volves the ionization of a molecule in its S1electronic state with the selective detection of those electrons that carry no kinetic energy. A pump pulse prepares an intermediate state in S1 and the probe pulse is tuned to a vibronic feature in the cation. The zero kinetic energy electrons are extracted by a voltage applied 1-2 ps after the probe laser excitation. By tuning the probe laser to a specific vibrational band within S 1, population monitoring can be achieved from the observation of the ZEKE signal at different pump-probe delay times. Because IVR occurring during the delay will depopulate the S1 vibronic bright state, a decrease in the ZEKE signal will be observed.This enables a real-time study of IVR and the method has been used to study a variety of molecules including p-difluorobenzene.8 The result- ing spectra are similar to those obtained with the time-resolved studies described previously, and indicate statistical and intermediate IVR at different excitation energies. 2.2 Ground state Historically IVR studies concentrated on electronically excited states, but it is within the electronic ground state that real chemical interest lies. IR spectroscopy is the most obvious method of studying IVR in the ground state. In its crudest form, applying IR spectroscopy to IVR studies involves standard spectrometers and lineshape analysis similar to that described previously.Conventional IR spectroscopy, however, provides restricted resolution, and until good sources of tunable IR radiation were available the study of IVR in the ground state was somewhat limited. The rectification of this has led to extensive ground state IVR studies in the frequency domain by the direct pumping of low vibrational states and the use of detectors that monitor absorption. An advantage of this is that the spectroscopic resolution is limited only by the bandwidth of the exciting laser, whereas excited state studies often rely on collecting dispersed fluorescence with a monochromator. A disadvantage of IR spectroscopy is the severe harmonic selection rule, Av = +I, which means that higher vibrational levels are difficult to populate by direct photon absorption.This problem has been overcome by a technique known as stimulated emission pumping (SEP) which involves promoting the molecule to an excited electronic state and subsequently stimulating it down to a high vibrational level in the ground electronic state. The accompanying selection rules place fewer restrictions upon the vibrational levels accessible in the ground electronic state. This technique is thus an extremely powerful method of studying IVR in the electronic ground state at differing levels of excitation. Knight and coworkers used this method to study high lying vibrational levels in p-difluorobenzene;g the level of interest was prepared by SEP and the populated level probed by laser-induced fluorescence.The study of very high lying levels is of great chemical significance, especially if their energies are above a dissociation limit. Geers et al. performed just such a study on the CH30 radical.1° The detection of molecular ions in the gas phase can be accomplished relatively easily and accurately, for example by mass spectrometry. This provides the basis for an alternative method to study ground state IVR known as stimulated emission ion dip (SEID)." Here the number of molecular ions produced by sequential excitation with two differing laser 226 Chemical Society Reviews, 1997, volume 26 frequencies is monitored as a function of the second laser frequency v2. However, the second laser can also stimulate emission to a vibrational level in the So state if its frequency coincides with an energy gap.Hence, there is a depopulation of the S1 state and a dip in the ionization spectrum. If IVR is significant the second laser cannot induce pumping back to S1 leading to a large dip. IVR rates can be derived from the extent of the dip and appropriate rate equations. A technique which has recently become possible is double- resonance infrared excitation as demonstrated by Gambogi et a1.12 Here, a vibrational state in So is prepared by sequential excitation by two IR photons. For example, the transitions 1 t 0 and 3 t1 were excited within the v3 mode of diacetylene to avoid the 'more forbidden' character of 3 t0 compared with 3 t1.12" This provides considerable advantages: (i) the spectrum can be recorded with less overall power because the transition probability will be greater when Av is small, (ii) it is possible to reach states not accessible by one photon absorption e.g.symmetric stretches within a molecule with a centre of inversion, and (iii) because the rotational angular momentum quantum number, J, is selected in the first absorption step the number of accessible rotational states is reduced compared with those accessible by direct photon absorption, aiding a full assignment of spectra. Double-resonance IR spectroscopy is often used as a method of 'eigenstate-resolved spectroscopy', i.e., using high resolution IR radiation to resolve the rotational fine structure of vibrational bands.The resolution enables coupling strengths of the bright state to nearly isoenergetic states to be deduced, and since the spectra are eigenstate-resolved the perturbing states can be identified. As an illustration the first overtone C-H stretching vibration of diacetylene was studied using this method.12" The spectra revealed three strong lines for each transition of the P and R branch at low rotational angular momentum quantum number, J, with the emergence of a fourth line at higher J indicating a further perturbing state. The perturbing states were deduced to be bending and other stretching vibrations of the molecule, although the exact modes responsible were not determined. An important application of eigenstate resolution is the determination of the density of states which can be 'counted', and would otherwise need to be calculated. This quantity is important for the considerations of IVR and will be discussed later.The double-resonance IR technique has pro- vided a large source of data on IVR lifetimes to date (see Table 1).IVR in the electronic ground state has also attracted considerable theoretical attention but full quantum calculations are only, as yet, possible for rigid molecules. Classical trajectory calculations are often applied to IVR dynamics and rely upon using classical mechanics to calculate the distribution of energy at desired time delays using a set of initial conditions, an appropriate potential energy surface and spectroscopic data.Full details are beyond the scope of this review but interested readers are directed to a series of papers by Budenholzer and coworkers.13 3 Origins of IVR 3.1 Mode specificity in IVR The basis functions corresponding to excitation in independent normal modes (see Section 1) will in general be coupled by anharmonic interactions. Because potential energy functions are totally symmetric, anharmonic interactions will only couple eigenstates that are of the same vibronic symmetry. However, anharmonic coupling is enhanced through the weak interactions of near degenerate states of differing vibronic symmetry, even though they are in principle forbidden, and it is likely that extensive weak interactions of this nature are important for the rapid statistical IVR reached in large polyatomic molecules. These symmetry controlled interactions lead to the possible transfer of vibrational energy between the coupled states, which is manifested in vibrational spectra by the appearance of bands which would otherwise be dark.For example, Lehmann, Scoles and coworkers studied the CH stretch in SFSCCH using eigenstate-resolved IR spectroscopy. l4 The fundamental ab- sorption revealed five bands of which two could only be attributed to dark states that had borrowed intensity from the bright state and so must have been anharmonically coupled to it. The perturbation was thought to be a CCH bending mode, and it is a general observation that the CH chromophore in acetylenic compounds couples the CH stretch and CCH bend.This coupling was shown to be very important in diacetylene with a second overtone lifetime of 165 ps which upon deuteration of CCH increases to 500-600 ps.l*a This suggests that the removal of the CCH group was accompanied by the removal of a very important IVR channel. A similar observation for acetylenic molecules was made by Nesbitt and cowork- ers.lS CH stretches excited in other structural environments are also observed to couple to bends involving motion of the same CH unit. Examples of molecules that exhibit this behaviour are CHY3 where Y = F or CF3, and CHFZ2 where Z = C1 or CH3.16 This coupling of a stretch and bend involving a fundamental and an overtone vibration is known as a Fermi resonance and has been observed for many other sp3 alkyl chromophores such as toluene16 and trideuteroacetaldehyde.l7 A general phenomenon of mode specificity, i.e. the preferred coupling of the bright state to specific dark states is thus illustrated by trideuteroacetaldehyde and the other molecules above. In trideuteroacetaldehyde coupling was observed be- tween the CH stretch and the CH in-plane bend, but not between the stretch and CH out-of-plane bend. It was also observed that there was no coupling between the CO and CH stretches which to a certain extent provides validation of the 'doorway state' model considered in section 4.2. This mode specificity was thought to be important for modes which possess appreciable anharmonicity, for instance vibrations within a methyl group.An IR study of methyl glyoxal revealed a mode specific coupling between the CO stretching mode and bending modes of the methyl group, suggesting that removal of this group would decrease the IVR rate.l8 The C-C skeletal torsion was also thought to be significantly anharmonic but no evidence of mode specificity involving this channel was observable. The fact that coupling depends upon the vibrational nature of the dark states was also shown by Philips and coworkers in an eigenstate-resolved infrared study of 2-fluoroethanol in the CH stretching region.19 Clusters of peaks were observed in the spectrum, with each peak corresponding to a rovibrational state and each cluster corresponding to a bright state. The molecule displayed a large variation in the degrees of coupling within a single cluster of peaks, which could not be explained by the usual random fluctuations in coupling matrix elements.This suggested that the bright state interacts most strongly with specific dark states. This work also illustrates the importance of rotational resolution in identifying coupling pathways. 3.2 Effect of molecular rotation Molecular rotation can also induce vibrational state mixing. As a molecule rotates it experiences a Coriolis force which can couple vibrational modes. For example, in a triatomic molecule rotation couples the antisymmetric stretch and one component of a doubly degenerate bending mode. Such Coriolis coupling manifests itself in eigenstate-resolved spectra through the dependence of coupling strengths or rotational linewidths upon the rotational quantum number, J; the greater the angular momentum the greater the molecular distortions and hence coupling induced. Lehmann, Scoles and coworkers studied three quanta in the acetylenic CH stretch of propyne using infrared double resonance spectroscopy and illustrated the effect of rotational angular momentum.12h The time evolution of the bright state was calculated from frequency-resolved spectra and revealed quantum beats at low J which were not significant at high J. This suggests that Coriolis coupling can be important in mediating IVR. Coriolis coupling is an important consideration for experi- ments employing supersonic jet-cooling. Jet cooling vastly reduces the number of rotational levels populated which may increase the lifetime of extensively Coriolis coupled vibrational modes.Infrared jet-cooled studies of dimethyl ether and 1,4-dioxane showed that the CH stretch displayed limited coupling at a rotational temperature of 2 K compared with 20 K, where there was considerable coupling.20 Similarly, Felker and Zewail found that the decay rates of quantum beat envelopes of jet-cooled anthracene increased significantly as the rotational quantum number increased.4 McDonald et al. made a direct study of the effect of rotational temperature on the lifetime of a vibrational state.7 TRFD scans of four bands of p-cyclohexylaniline were taken with rotational temperatures ranging from 8-1 10 K by using differing He backing pressures in a molecular beam. The results clearly indicate a decreasing lifetime with increasing rotational temperature which is likely to be caused by more extensive Coriolis coupling (see Fig.3). 500 I I I 100 v)n \ a, .-E 50-+ -Ia,'c U1 20 ' 10 ' 5 I I I ! 0 40 T/K 8o 120 Fig. 3 Data from time-resolved fluorescence depletion (TRFD) scans of four bands of p-cyclohexylaniline with rotational temperatures ranging from 8 to 110 K illustrating the resulting change in IVR lifetime. (Adapted from P. G. Smith and J. D. McDonald, J. Chem. Phys., 1990, 93, 6350 with kind permission of the American Institute of Physics). The overall IVR lifetime can also indicate whether Coriolis coupling is mediating IVR.For example, the frequency-resolved jet-cooled spectrum of the OH stretch in ethanol displayed a FWHM lifetime of 25 ps which is faster than the overall rotation of the molecule at low J.21 It was concluded therefore that anharmonic coupling must be dominant. Thus, states with fast IVR rates are likely to be extensively anharmonically coupled. 4 Rationalisationof IVR A natural consequence of experimental work is the attempt to rationalise any trends that may appear and to propose a model Chemical Society Reviews, 1997, volume 26 227 which can be applied generally. IVR studies are no exception with explanations based on principles that involve knowledge of coupling pathways and the general energy level structure of the molecule, through to models based purely on structural features. 4.1 Density of states Due to its central importance in RRKM theory, the density of rovibronic states is the most commonly used factor to rationalise IVR (see Section 1).A change in the density of states affects the number of dark states that are sufficiently near degenerate with the bright state to couple to it. On these grounds an increase in density of states would imply a corresponding increase in the rate of energy redistribution. Thus, the density of states is generally accepted to be responsible for the existence of a threshold energy for the onset of IVR. However, it has been found that excitation to a vibrational level embedded in a higher density of states does not necessarily produce a faster IVR rate, especially when comparing excitations in different molecules (see below).Work that confirms the expected trend in IVR rate with density of states is provided in a series of three papers by Smalley and coworkers who carried out an extensive study of jet-cooled alkylbenzenes excited to their S electronic state.22 Their fluorescence study revealed increasing unstructured emission with increasing alkyl chain length indicating enhanced IVR with increasing density of states. In addition a picosecond time evolution study of the bright state revealed that early members of the series displayed intermediate case IVR with insufficient density of states to permit any dynamical IVR process to occur on a subnanosecond timescale.This time- resolved work also revealed that the longer chains possessed a sufficient density of states to enable complete randomisation of energy within a nanosecond. Further excellent illustrations of the expected effect of density of states stem from the fluorescence study and an independent time-resolved study of jet-cooled anthracene. Fig. 4(a) illustrates an increase in linewidth with vibrational energy while Fig. 4(b) indicates a decrease in lifetime with increasing density of states. The latter time-resolved study adds weight to the idea of IVR approaching the statistical regime with increasing density of states. The authors found that up to 1200 cm-1 of excitation the vibrational levels were almost eigenstates with little evolution of energy, while intermediate case IVR occurred between 1380 and 1520 cm-1, with a lifetime of 200 ps.At energies greater than 1520 cm-l statistical IVR was observed with a fast component of 22 ps and a slow component of 6.4 ns. This fast component of IVR can be extremely rapid at high vibrational energies which is usually explained as being a consequence of the very high density of states. For example, the sixth overtone of the CH stretch of benzene in its SI electronic state which is embedded within a continuum of vibrational and rotational states, displays a lifetime of 50 fs.23 This trend is also observed for IVR in ground electronic states where it is generally observed that the onset of IVR occurs at a density of states of ca.100 per cm-1. This value predicts the onset in many molecules and is independent of molecular structure.24 However, the term onset of IVR must be interpreted with care since it is open to discussion at which point energy redistribution is fast enough to be classed as true IVR, and it should not be assumed that IVR is absent below a tabulated threshold. Using threshold values and calculated densities of states to predict whether IVR will be rapid is not on the whole reliable, because the observed density is often significantly different from that calculated. For example, the eigenstate-resolved infrared spectrum of allene revealed 27 vibrational states per cm-1 for one of the combination bands compared with the calculated value of l2,25 while the high resolution spectrum of ButCrCH suggested a lower limit of 7.8 X lo3 per cm-l compared with the estimated theoretical value of 1.7 X 102.12c 228 Chemical Society Reviews, 1997, volume 26 rn I I 1 I 1 200 -150 -E ,o0CL' 100 -50 ->18 ns7 I1 0 1000 2000 E,,, 1crn-l Fig.4 Illustrations of the effect of density of states on the rate of IVR, as it would usually be expected. (a) Linewidth increasing with vibrational energy and hence density of states. (b) Calculated density of states vs vibrational energy in anthracene. In (b)experimental IVR lifetimes in the sparse (18 ns at 766 cm-I), intermediate (200 ps at ca.1450 cm-1) and statistical (22 ps at 1792 cm-l) regimes are shown. (Adapted from P. S. H. Fitch, C. A. Haynam and D. H. Levy, J Chem. Phys., 1981,74, 6612 and P. M. Felker and A. H. Zewail, J. Chem. Phys., 1985, 82, 2961 with kind permission of the American Institute of Physics). Another discrepancy arose with the high resolution IR spectrum of 2-fluoroethanol.19 The calculated density of states in the range 2980-2990 cm-1 was 63 states per cm-1. Assuming every dark state couples with the bright state this would lead to a minimum spacing of 1/63, i.e. 0.016 cm-1. However, the spacing between states in the experimental spectrum was 0.006 cm-signifying a disagreement between the calculated density of states and the experimental data. Often trends in lifetimes of vibrational states cannot be explained by considering either calculated or measured den- sities of states.Lehmann, Scoles and coworkers used high resolution IR spectroscopy to study the CH stretch of a series of alkynes (see Table 2). Fluorination causes a dramatic increase in the density of states compared with the hydrogen and deuterium species, but the fluorinated molecule has a shorter and longer lifetime respectively. A further discrepancy occurs in an IR study of propyne.12d Here, when the molecule was excited with three quanta in its acetylenic CH stretch the prepared state displayed a lifetime of 3 10 ps while excitation of a combination band gave rise to a lifetime of 3.2 ns. The respective densities of states were 150 and 107 states per cm-1 and this difference was considered to be too small to explain such a dramatic increase in lifetime.An IR study of the second, third and fourth overtone bands of the OH stretch in hydroxylamine provides a more obvious failing of the density of states as a predictor of IVR lifetime (see Fig. 5). The spectra revealed sharp features for the second and third overtones, but the fourth overtone consisted of a very broad band with unresolved P and R branches. The Table 2 Lifetimes of vibrational levels within the acetylenic CH stretch in various alkynes, compared with the density of states with A, symmetry, all taken from Ref. 12~ Density of states Molecule Lifetime/ps per cm-1 v = 1 (CH~)~CCZCH 200 4.9 x 102 (CD3)3CCsCH 40 2.8 x 103 (CF~)~CCZCH 60 4.2 X 106 (CH3)3SiC=CH 2000 1.0 x 104 (CD3)3SiCsCH 850 LO x 105 v = 2 (CH3)3CC=CH 110 6.2 X 105 (CD3)3CC=CH < 20 7.6 x 106 (CF3)3CCrCH 5 1.0 x 10” (CH3)3SiC=CH 4000 2.0 x 107 (CD3)3SiC=CH 140 6.0 X 108 calculated density of states increases by a factor of 4 between the second and third overtones with a resulting 30% increase in linewidth.By analogy, a similar increase between the third and fourth overtones predicts a linewidth of 1.2 cm-l, not the 7 cm-1 observed. These comparisons suggest that the change in linewidth does not come solely from an increase in the density of states (see later). Thus, the density of states is a quantity that can occasionally explain the changes in vibrational lifetimes accompanying a structural modification or a change in the vibrational energy within a molecule.It is often applied in a rather contingent 11 “11 I1I1~llllll*llJ -1 00 -50 0 100 Y Icm-’ 50 Fig. 5 IR spectra of the second, third and fourth overtone transitions of the OH stretch in hydroxylamine illustrating a large increase in linewidth from the third to the fourth overtone which is not consistent with the small increase in the density of states. (Adapted from J. L. Scott, D. Luckhaus, S. S. Brown and F. F. Crim, J. Chem. Phys., 1995, 102, 675 with kind permission of the American Institute of Physics). manner in an attempt to rationalize experimental observations, but its explicit effect is hard to deduce.However, the density of states is almost certainly a major factor in the division of IVR into sparse, intermediate and statistical regimes and initial qualitative inferences may be drawn from consideration of the region of the vibrational manifold into which the original excitation occurs. 4.2 Empirical models The inability of density of states considerations to provide consistency has led to the development of several alternative explanations of relative IVR rates. Perhaps the most promising is the proposed existence of a doorway state by Lehmann, Scoles and coworkers.24 Within this model mode specific coupling of the bright state to one dark state is a prerequisite to redistribution, and the particular dark state can be considered to be a ‘doorway’ which exposes the bright state to the full density of rovibrational states available. It is only at this point that the original bright state population becomes fragmented into many eigenstates.Even though the model ignores the relative strength of vibrational state coupling it was successful in explaining the lifetime dependence of diacetylene on the rotational quantum number, J.12~ The lifetime of this molecule increased with increasing J even though one would expect greater Coriolis coupling with J and a corresponding decrease in lifetime. However, within the doorway model for this molecule, as the energy increases with increasing J there is a larger energy separation between the bright state and the doorway state resulting in a lifetime increase.The doorway state is also central to the tier model proposed by Marcus and Stuchebrukhov.26 This theoretical model is based upon the assumption that relaxation occurs through a series of ‘tiers’ of states connected to the bright state. The model assumes the lifetime of a vibrational state is independent of the nature of the states that receive vibrational energy, and determined only by the coupling strength and positions of the levels in the first few tiers. Each ‘tier’ represents a level coupled to the bright state in a given order, none of which incorporates the total density of available states. Because it is difficult to apply the above models without some knowledge of the energy level structure, correlations between molecular dynamics and the structural features of molecules are desirable, enabling the prediction of the effects on IVR rates when structural modifications are imposed.Numer- ous such models exist and it is enlightening to consider three of the more successful ones that have been proposed. It is generally observed that excitation near to a heavy atom produces a relatively long IVR lifetime and this has led to the concept of the heavy atom effect. This effect is essentially proposed to be caused by a heavy central atom impeding the flow of energy across the molecule due to its inherent relative inability to participate in vibrational motion. This results in the removal of those decay channels which involve the heavy atom, causing a slower decay of the optically active state.van der Waals interactions have also been proposed to be important in controlling IVR lifetimes because these interactions naturally lead to state mixing. Thus, a high degree of steric congestion should lead to an enhanced decay rate of the bright state. The flexibility of a molecule has also been used to rationalise IVR rates. Perry and coworkers collected a number of IVR lifetimes and discovered a correlation with molecular flexibility.27 Essentially it is observed that IVR is faster in flexible molecules when the initially prepared vibration is close to the bond about which groups possess substantial movement as in, for example, trans-gauche isomerisation.Lehmann, Scoles and coworkers have made detailed studies of a whole series of alkynes and substituted alkynes by looking at IR excitation of the acetylenic CH stretchI2,24 (see Table 2). In the course of these studies they have applied the doorway state model and two of the structural models discussed above to rationalise their results. In what follows we illustrate how the Chemical Society Reviews, 1997, volume 26 229 principles were applied to these alkynes and how this work ties in with the attempts of other groups to apply these models to their experimental data. 4.2.1 van der Waals model The van der Waals model is probably the least successful of the structural models but nevertheless provides an explanation for the results of Parmenter and coworkers who compared the fluorescence spectra of p-difluorobenzene and p-fluorotoluene using chemical timing techniques.5 The addition of a methyl group clearly accelerated IVR as inferred through the loss of structured emission.The effect observed was thought to be a consequence of an extensive van der Waals interaction between the methyl group and the ring. Although the methyl group couples only weakly to the ring, its low barrier to internal rotation enables this extensive interaction. A study of 1-chloro-2-fluoroethane revealed that torsional motion was responsible for the bright state decay.** The gauche conformer displayed significantly greater coupling than the trans conformer which is consistent with there being reduced interactions between the two methyl groups which are further apart in the latter.Many of the alkyne studies support the views of Parmenter and coworkers on van der Waals interactions. The increase in the lifetime of an acetylenic CH stretch upon silylation of (CH3)3CCrCH and (CD3)3CC=CH (see Table 2) can be rationalised by a reduction of steric congestion between methyl groups, as a consequence of the longer Si-CH3 bond. This reduced van der Waals interaction would decrease state mixing leading to a suppressed IVR rate. Similarly, an enhanced van der Waals interaction explains the decrease in lifetime upon deuteration for both silicon and carbon compounds. The same trend is not however observed for the acetylenic CH stretch lifetimes in the pair of molecules (CF&CC-CH and (CD3)3CC-CH.Considering solely van der Waals interactions, one would expect a faster decay of the bright state in the fluorinated molecule, but the corresponding lifetimes are 60 and 40 ps, respectively. Because parameters such as normal mode frequencies and bond lengths are unlikely to remain constant upon fluorination, attempts to rationalise lifetimes based on this simple model may be a little optimistic in this case. The van der Waals model does however lead to the prediction that removal of the trifluoromethyl rotors would lead to a substantial increase in lifetime, which is indeed observed. It is also successful when applied to the IR spectra of trifluoropropyne.29 Here, insignif- icant IVR was observed when the fundamental was excited and a lifetime of 2 ns found for the first overtone, consistent with the reduction of van der Waals interactions compared with (CF3)3CC-CH. However this may be fortuitous since the density of states will also decrease and parameters such as the coupling matrix elements are likely to change upon removal of the CF3 rotors.Further scepticism surrounding this model is apparent on consideration of the lifetime measurements for CH3Si(CzCH)3.12~This molecule has only one methyl group but relaxes faster than the structurally similar molecule (CH3)3SiC-CH that has three (see Table 1). These results suggest that although van der Waals interactions between methyl groups do not dominate IVR, they may be an important initial guide for a qualitative understanding of IVR.4.2.2 Doorway state model The alkyne studies also provide some evidence of the existence of a doorway state. (CH3)3CC=C-H and (CH&SiC=C-H have acetylenic CH stretch lifetimes of 400 ps and 2 ns, respectively. As the density of states increases when the carbon is replaced with silicon, it was thought that the change in mass and force constants moved some key doorway state out of resonance with the CH stretch. Direct experimental evidence for the existence of a doorway state stems from the infrared studies of a number of substituted toluenes by Sowa et ~1.16 It was observed that, contrary to density of states considerations, there was line narrowing of the overtone acetylenic CH stretch transition from toluene to penta-fluoro substituted toluenes.Mode specificity between the CH stretch bend was discovered and the coupling thought to fall out of resonance following the modification of the CH stretching frequency in the presence of the fluorine substituents, leading to a longer lifetime as inferred from the decrease in linewidth. It was concluded that this Fermi resonance can be prevented through a judicious choice of substituents suggesting that one can control the vibrational lifetime of an excited CH stretch. 4.2.3 Heavy atom model The alkyne series probably best illustrates the heavy atom effect. The calculated densities of states show a large increase on going from the fundamental to the first overtone for each compound, and yet (CH3)3SiC=C-H displays a longer lifetime and (CH3)3CC+H a shorter lifetime in the first overtone.It was also recognised that the carbon compounds displayed con-siderably shorter lifetimes than the silicon analogues. As this is contrary to density of states considerations it was thought to be a consequence of a heavy atom effect.12e The study was extended to the tin analogue of the molecules where a lifetime of 6 ns was obtained for the acetylenic CH stretching fundamental which agrees with the proposed mode1.24 However this heavy atom effect is unlikely to be solely due to mass because the bond lengths and vibrational frequencies will also change. The model also fails to explain why a factor of 2.3 mass increase from carbon to silicon increases the lifetime tenfold, whereas from silicon to tin the mass increases by a factor of 4.2 with an accompanying threefold increase in lifetime.It is revealing to compare these results with other experi- mental and theoretical studies of the heavy atom effect which were performed prior to the above experiments. Rogers and coworkers studied the reactions of tetraallyl tin and tetraallyl germanium with fluorine atoms.30 Here, the deduced rate constant was 1000 times larger than that predicted from RRKM theory which suggests that, in this case, the heavy atom blocked the statistical redistribution of energy, which is central to RRKM theory. In contrast, the unimolecular dissociation of activated 4-(trimethylstannyl)-2-butyl and related compounds displayed rates which were consistent with RRKM and suggested that the heavy atom did not block IVR to any significant extent.3 Lopez and Marcus used classical trajecto- ries to investigate IVR in a system designed to explore the heavy atom effect.32 They considered a C-C-C-X-C-C-C system (X = Sn, Ge) and found that when the stretches were described by Morse oscillators, Sn inhibits the flow of energy while the Ge does not, which could be due to the greater mass of the tin.However for harmonic stretches both Sn and Ge allowed energy flow across them. This suggests that a purely heavy atom effect is unlikely to predict the changes in lifetime accompanying substitution for a heavier atom, which is further supported by the calculations of Hase and coworkers.33 Here, a potential energy surface for the tetraallyl system [Sn(-C-C=C),] was calculated using Morse oscillators to describe all stretches.The energy flow past the Sn atom increased with increasing energy placed in the C=C stretch, but upon substitution of carbon for tin in the same surface there was no significant change in IVR rate. However, when the surface was modified to give potential parameters that were chosen to match experimental frequencies and bond energies of carbon, IVR was found to be almost complete within a picosecond, which suggests the change in mass is not the dominating factor in this system. Nevertheless, both theory and experiment suggest that heavy atom substitu- tion is a promising structural modification to inhibit IVR.4.2.4 Molecular flexibility model Molecular flexibility has been used by Perry and coworkers to rationalise IVR lifetimes.27 The model focuses on CH stretches with the following conditions: (i) All stretches are equivalent, with relaxation rates attributed to local identity and not chemical reactivity and (ii) proximity to centre of flexibility 230 Chemical Society Reviews, 1997, volume 26 tends to accelerate IVR in the absence of strong coupling matrix elements. Chromophores in rigid molecules should then diplay longer lifetimes than those in flexible molecules. This flexibility may involve internal rotation, as for example about the C-0 bond in ethanol.In but-1-yne the methyl and acetylenic CH stretches have similar lifetimes of approximately 280 ps which is described as a consequence of this molecule having no significant centre of flexibility. However, in pent- 1 -yne the aliphatic CH relaxes faster than the acetylenic CH which is consistent with the proposed model due to trans-gauche isomerisation about the C2-C3 bond in pent- 1-yne: CH3-CH2- CH&=CH. The enhanced relaxation of the aliphatic CH could be due to the closer proximity of the methyl group to the centre of flexibility. Fig. 6 illustates the concept of molecular flexibility, indicat- ing how the IVR lifetime varies as a function of the distance from the centre of flexibility. For example the lifetimes of the C-H stretch in three rigid molecules, isobutane, trans-but-2-ene + Propargyl alcohol 0 Pentyne +Ethanol 0 + d 0 0 1 2 3 4 Number of bonds between chromophore and centre of flexibility Fig.6 IVR lifetimes as a function of the number of bonds (n) between a given chromophore and the centre of flexibility. For ethanol the OH chromophore is at n = 1 and the methyl chromophore is at n = 1. For trans-pentyne the methyl chromophore is at n = 2 and the acetylenic chromophore is at n = 3. For propargyl alcohol the methyl chromophore is at n = 1 and the acetylenic CH is at n = 3. The IVR lifetime increases as the distance from the centre of flexibility increases. (Adapted from G. A. Bethardy, X. L. Wang and D. S. Perry, Can. J.Chem., 1994,72,652 with kind permission of the National Research Council of Canada).and tert-butylacetylene are longer than those in three flexible molecules, but-1 -ene, butane and methyl formate. This model appears to rationalise IVR lifetimes of C-H stretches quite readily and only requires the structure to estimate the region of the molecule which is likely to retain vibrational energy the longest. However, if the bright state is strongly coupled to the surrounding dark states the model breaks down, and since coupling strengths cannot be deduced from the molecular structure it is difficult to apply this model without experimental data. Although at present correlations between IVR rates and the structure of molecules are purely qualitative they still prove invaluable since they may aid the design of molecules that exhibit significantly slow intramolecular dynamics.This is particularly relevant to mode-selective chemistry since such modifications can have dramatic effects on the lifetimes of vibrational states, which in turn may permit selective reac- tions. 5 Mode-selective chemistry IVR is not only important for unimolecular dissociation (RRKM theory) but also for the consideration of bimolecular collision induced reactions, where mode-selective chemistry may be possible if the retention of energy within a specific mode is longer or comparable to the time between bimolecular collisions in the gas phase. Bimolecular mode-selective chem- istry has been achieved with deuterated water, HOD.34 When HOD reacted with fast hydrogen atoms there was a preference for breaking either the OH or OD bond when the respective bond contained quanta of stretching motion.However, this is the only true example of mode-selective chemistry reported to date and was thought to be possible because of the small size and low density of states of HOD. With obvious synthetic applications in mind, it is desirable to extend mode-selective chemistry to larger polyatomic mole- cules. The inherent problem with this is that large polyatomic molecules possess a greater density of states, especially high up in the vibrational manifold, which promotes the rapid randomi- sation of localised energy. Furthermore, to ensure facile bond breakage it would be necessary to work at these high excitation energies where the enhanced IVR rate could lead to the formation of alternative products, rendering mode-selective chemistry in such species impossible to control.Thus, to accomplish mode-selective chemistry, a compromise is re-quired between the energy needed to ensure a selective reaction and an excitation energy which exhibits slow IVR. To ascertain whether mode-selective chemistry is possible it is necessary to determine if the rate of bimolecular collisions is faster than the rate of IVR. Simple collision theory gives an approximate collision interval at room temperature and pressure of 1 ns. The IVR lifetimes listed in Table 1 reveal that only acetylenic CH stretches and combination bands have so far exhibited IVR lifetimes close to this benchmark value.The energy in combination bands however is distributed over several bonds so, although energy is retained in specific parts of the molecule for the duration of a collision any one of several bonds could break leading to a mixture of reaction products. Thus, the attention must be focused on the acetylenic compounds, and especially on the acetylenic CH stretch itself. Mode-selective chemistry with these compounds has not yet been reported, although many authors have stated it to be possible with, for example, trifluoropropyne, which possesses a first overtone lifetime of 2 ns. Lehmann, Scoles and coworkers recently published a paper stating their intention to study the effect of ‘heating the bath’ of vibrational modes which ultimately receive the vibrational energy in the IVR process in an attempt to extend the lifetimes of molecules.l4 Mode-selective chemistry may be achievable in relatively small polyatomic molecules such as the alkynes listed in Table 1, but in other molecules it appears to be only a distant possibility. The structural modifications discussed previously may be an aid to mode-selective chemistry, if the modification suppresses IVR. The structural modification could be the introduction of a heavy atom which would allow the selective reaction, and the molecule could be subsequently converted back. 6 Conclusion In this review we have attempted to give a broad overview of IVR studies highlighting some recently developed experimental techniques, and some attempts to rationalize results.We have concentrated on only a small subsection of the available literature in order to illustrate our arguments. It can be seen that Chemical Society Reviews, 1997, volume 26 231 there are still many active workers in the field, and many questions to be answered. A review by Nesbitt and Field which appeared after the manuscript for this article was written confirms the timeliness of a review on IVR.35 Their article takes a somewhat different approach from ours, being heavily referenced (including mention of a number of earlier reviews) and more rigorous, and is recommended for further reading. 7 Acknowledgement We are indebted to Professors Kevin Lehmann and Ian Mills for their comments on this manuscript.We also thank Professor Peter Sarre for encouraging us to write it. Dean Boyall would like to thank Chris J. Brennan for drawing his attention to some of the references cited here. 8 References 1 P. J. Robinson and K. A. Holbrook, Unimolecular Reactions, Wiley, New York, 1971, pp. 64-92, and references therein. 2 J. N. Butler and G. B. Kistakowsky, J. Am. Chem. Soc., 1960, 82, 759. 3 T. F. Deutsch and S. R. J. Brueck, Chem. Phys. Lett., 1977, 54, 258. 4 P. M. Felker and A. H. Zewail, J. Chem. Phys., 1985, 82, pp. 2961. 5 C. S. Parmenter and B. M. Stone, J. Chem. Phys., 1986,84,4710, and references therein. 6 P. M. Felker and A. H. Zewail, Chem. Phys.Lett., 1984, 108, 303. 7 P. G. Smith and J. D. McDonald, J. Chem. Phys., 1990, 93, 6350. 8 X. Zhang, J. M. Smith and J. L. Knee, J. Chem. Phys., 1994, 100, 2429. 9 S. H. Kable, J. W. Thoman Jr., S. Bearnes and A. E. Knight, J. Phys. Chem., 1987,91, 1004. 10 A. Geers, J. Kappert, F. Temps and J. W. Wiebrecht, J. Chem. Phys., 1990,93, 1472. 11 T. Ebata and M. Ito, J. Phys. Chem., 1992, 96, 3224. 12 (a) J. E. Garnbogi, R. Z. Pearson, X. M. Yang, K. K. Lehmann and G. Scoles, Chem. Phys., 1995, 190, 191; (b) J. E. Garnbogi, E. R. Th. Kerstel, K. K. Lehrnann and G. Scoles, J. Chem. Phys., 1994, 100,2612;(c)J. E. Garnbogi, K. K. Lehrnann, B. H. Pate, G. Scoles and X. M. Yang, J. Chem. Phys., 1993, 98, 1748; J. E. Garnbogi, R. P. Lesperance, K. K.Lehmann, B. H. Pate and G. Scoles, J. Chem. Phys., 1993, 98, 1116; (4 J. E. Garnbogi, J. H. Tirnrnermans, K. K. Lehrnann and G. Scoles, J. Chem. Phys., 1993, 99, 9314; (e) J. E. Gambogi, R. P. Lesperance, K. K. Lehmann and G. Scoles, J. Phys. Chem., 1994,98, 5614. 13 F. E. Budenholzer, M. Y. Chang and K. C. Huang, J. Phys. Chem., 1994, 98, 12501, and references therein. 14 M. Becucci, J. E. Garnbogi, J. H. Tirnrnermans, K. K. Lehrnann, G. Scoles, G. L. Gard and R. Winter, Chem. Phys., 1994, 187, 11. 15 A. McIlroy and D. J. Nesbitt, J. Chem. Phys., 1990, 92, 2229. 16 M. G. Sowa and B. R. Henry, J. Chem. Phys., 1991,95, 3040. 17 A. Arnrein, H. Hollenstein, M. Quack, R. Zenobi, J. Segall and R. N. Zare, J. Chem. Phys., 1989, 90, 3944. 18 S. A. Reid, H. L. Kim and J. D. McDonald, J. Chem. Phys., 1990, 92, 7079. 19 C. L. Brurnrnel, S. W. Mork and L. A. Philips, J. Chem. Phys., 1991,95, 7041. 20 T. J. Kulp, H. L. Kim and J. D. McDonald, J. Chem. Phys., 1986, 85, 211. 21 G. T. Fraser, B. H. Pate, G. A. Bethardy and D. S. Perry, Chem. Phys., 1993, 175, 223. 22 J. B. Hopkins, D. E. Powers and R. E. Srnalley, J. Chem. Phys., 1980, 73, 683, and ref. l(a)and l(b) therein. 23 R. G. Bray and M. J. Berry, J. Chem. Phys., 1979,71,4909. 24 E. R. Th. Kerstel, K. K. Lehrnann, T. F. Mentel, B. H. Pate and G. Scoles, J. Phys. Chem., 1991, 95, 8282. 25 J. H. Tirnrnermans, K. K. Lehmann and G. Scoles, Chem. Phys., 1995, 190, 393. 26 A. A Stuchebrukhov and R. A. Marcus, J. Chem. Phys., 1993, 98, 6044. 27 G. A. Bethardy, X. L. Wang and D. S. Perry, Can. J. Chem., 1994,72, 652. 28 C. C. Miller, S. C. Stone and L. A. Philips, J. Chem. Phys., 1995, 102, 75. 29 B. H. Pate, K. K. Lehrnann and G. Scoles, J. Chem. Phys., 1991, 95, 3891. 30 P. Rogers, D. C. Montague, J. P. Frank, S. C. Tyler and F. S. Rowland, Chem. Phys. Lett., 1982, 89, 9. 31 S. P. Wrigley and B. S. Rabinovitch, Chem. Phys. Lett., 1983, 98, 386. 32 V. Lopez and R. A. Marcus, Chem. Phys. Lett., 1982,93,232. 33 K. N. Swamy and W. L. Hase, J. Chem. Phys., 1985,82, 123. 34 M. J. Bronikowski, W. R. Sirnpson and R. N. Zare, J. Phys. Chem., 1993,97, 2204. 35 D. J. Nesbitt and R. W. Field, J. Phys. Chem., 1996, 100, 12735. Received, 19th December 1996 Accepted, 21 st February 1997 232 Chemical Society Reviews, 1997, volume 26

ISSN:0306-0012    

DOI:10.1039/CS9972600223

出版商:RSC    

年代:1997

数据来源: RSC