摘要:

Chemical Society Reviews Vol 13 No 3 1984 Page TILDEN LECTURE The Chemistry and Spectroscopy of Mixed-valence Complexes By Robin J. H. Clark 219 Chemicals from the Glands of Ants By Athula B. Attygalle and E. David Morgan 245 Designing Drugs to Fit a Macromolecular Receptor By C. R. Beddell 279 CENTENARY LECTURE Molecular Ingredients of Heterogeneous Catalysis By Gabor A. Somorjai 321 Corrigenda 351 The Royal Society of Chemistry London Chemical Society Reviews EDITORIAL BOARD Professor K. W. Bagnall, B.Sc., Ph.D., D.Sc., CChem., F.R.S.C. Professor B. T. Golding, B.Sc., M.Sc., Ph.D., C.Chem., F.R.S.C. Professor G. Pattenden, Ph.D., C.Chem., F.R.S.C. Professor P. A. H. Wyatt, B.Sc., Ph.D., C.Chem., F.R.S.C. (Chairman) Dr.D. A. Young, Ph.D., D.Sc., F.R.C.S., M.Inst. P. Editor: K. J. Wilkinson, B.Sc., M.Phi1. Chemical Society Reviews appears quarterly and comprises approximately 20 articles (ca. 500 pp) per annum. It is intended that each review article shall be of interest to chemists in general, and not merely to those with a specialist interest in the subject under review. The articles range over the whole of chemistry and its interfaces with other disciplines. Although the majority of articles are intended to be specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be submitted to the Managing Editor, Books and Reviews Section, The Royal Society of Chemistry, Burlington House, Piccadilly, London, W 1V OBN.Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at €15.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way. All other orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letch- worth, Herts. SG6 1HN England. 1984 annual subscription rate U.K. E43.50, Rest of World E45.50, U.S.A. $87.00. Air freight and mailing in the U.S.A. by Publica- tions Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. Second class postage is paid at Jamaica, New York 11431. All other despatches outside the U.K. by Bulk Airmail within Europe, Accelerated Surface Post outside Europe. 0Copyright reserved by The Royal Society of Chemistry 1984 ISSN 030tGOO12 Published by The Royal Society of Chemistry, Burlington House, London, W1V OBN Printed in England by Richard Clay (The Chaucer Press) Ltd, Bungay, Suffolk

ISSN:0306-0012    

DOI:10.1039/CS98413FP005

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

Molecular By D. E. HathwayDEHATHWAV Toxicology covers a wide field, which comprises chemical toxicology, cosmetics toxicology, environmental toxicology, food and drugs toxicology, industrial toxicology, pesticide toxicology, and the toxicology of atmospheric pollution, and which involves acute and long-term testing as well as special tests for carcinogenicity, irritation, mutagenicity and teratogenicity. In the testing phases, diagnoses are reached through the disciplines of chemical pathology, clinical medicine and pathology, and in the experimental one, explanation is sought through chemical/biochemical and pharmacological methods. In such a fragmented field as toxicology with so many diverse practices and specialisms, it is hard to gain an idea of continuity.This book attempts to unify the subject of toxicology and should be useful to chemists engaged in molecular biology and workers concerned with toxicology and cancer research in particular, as well as to pharmacologists and specialists in drug development and occupational hygiene and medicine in general. Brief Contents: Part I Toxicity of Foreign Compounds Assessment of Toxic Risk; Structure-Activity Considerations; Measurement of a Carcinogenic Exposure; Part II Relation between Dose and Effect and Time Biological Action; Extension of Simple Theory to Toxicology; Part 111 Metabolism Metabolic Pathways for Industrial Chemicals and Pesticides; Kinetic Considerations; Part IV PharmacogeneticsSpecies Differences in Metabolism and Toxicity; Part V Biochemical Lesions Mode of Action Studies; Part VI Chemical Carcinogenesis Importance of Chemical Non-enzymic Reactions in Vivo; Possible Mechanisms of Carcinogenesis and their Biological Significance; Host Factors and Cellular Aspects; Tissue Specificity; Part VII Toxicant Allergy Antigen Formation and lmmunobiological Effects Produced by Foreign Compounds; Hardcover 319pp 0 85186 068 0 Price f27.50($50.00) RSC Members f19.00 -Blackhorse Road, Letchworth, Herts SG6 IHN,England.Oxygen and the Conversion ofqxygen an& the Conversion Future Feedstocks of Future Third BOC Priestley Conference 3rd BOC Priestley Conference This publication contains the proceedings of the third BOC Priestley Conference held in September 1983.This international meeting was sponsored by British Oxygen Company PLC (Gases Division) and organised by The Royal Society of Chemistry, and consisted of a scientific part, on the general theme of the role of oxygen in the conversion of present and future feedstocks, and an historical part, on the various aspects of Joseph Priestley’s life and work. Brief Contents: Introductory Lecture: The Elegant Use of Oxygen; The Catalysis of Synthesis Gas Production; Development of the Shell Coal Gasification Process; Development of the Fixed Bed Slagging Gasifier; The Purification of Synthesis Gas; Transition-metal Peroxides as Reactive Intermediates in Heterolytic and Homolytic Liquid-phase Catalytic Oxidations; Heterogeneous Catalytic Oxidation: A Review of Principles and Practice; Ethylene and Ether from Ethanol; Novel Pathways to Enhance Selectivity in Fischer-Tropsch Chemistry; ln Situ Electron Microscopy Studies of Catalysed Gasification of Carbon; Ethylene Glycol from Synthesis Gas; Chemicals from Coal Gasification; Biomass as a Chemical Raw Material; Gas Separations by Manganese (11) Phosphine Complexes; Engines and Future Liquid Fuels; Concluding Remarks; Historical Sessions Priestley Lecture: The World’s First Chemical Explosive-in China and the West; Priestley and the Dissenting Academies; Priestley in Caricature; Priestley in America; 1794-1804; ’Fresh Warmth to our Friendship’: Priestley and His Circle; From Chaos to Gas: Pneumatic Chemistry in the Eighteenth Century; The Professional Life of an Amateur Chemist: Joseph Priestley; ‘A Sower Went Forth’; Joseph Priestly and the Ministry of Reform; Priestley and the Manipulation of Gases; Special Publication No.48 (1984) Softcover 504 p 085186 915 7 Price f18.00($33.00) RSC Members f13.50 ORDERING RSC Members should send their orders to: The Royal Society of Chemistry, The Membership Officer, 30 Russell Square, London WC1B 5DT. Non-RSC Members should send their orders to: The Royal Society of chemistry, Distribution Centre, -Blackhorse Road, Letchworth, Herts SG6 1HN, England.

ISSN:0306-0012    

DOI:10.1039/CS98413BP007

出版商:RSC    

年代:1984

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS98413FX009

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

Chemical Society Reviews Vol 13 No 3 1984 Page TILDEN LECTURE The Chemistry and Spectroscopy of Mixed-valence Complexes By Robin J. H. Clark 219 Chemicals from the Glands of Ants By Athula B. Attygalle and E. David Morgan 245 Designing Drugs to Fit a Macromolecular Receptor By C. R. Beddell 279 CENTENARY LECTURE Molecular Ingredients of Heterogeneous Catalysis By Gabor A. Somorjai 321 Corrigenda 351 The Royal Society of Chemistry London

ISSN:0306-0012    

DOI:10.1039/CS98413BX011

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

TILDEN LECTURE* The Chemistry and Spectroscopy of Mixed-valence Complexes By Robin J. H. Clark CHRISTOPHER INGOLD LABORATORIES, UNIVERSITY COLLEGE LONDON, 20 GORDON STREET, LONDON WClH OAJ 1 Introduction The considerable current interest in the bonding in, and properties of, mixed- valence complexes, well discussed at a recent NATO ASI’ and derives from the importance of such materials in fields as diverse as solid- state physics, inorganic chemistry, materials science, geology, and bioinorganic chemistry. At least 40 elements in the periodic table form mixed-valence species, and the importance of these species stems as much from their rich and varied chemistry as from their unexpectedly wide range of physical properties. On the geological side many minerals contain an element such as iron in two different oxidation states (e.g.magnetite Fe,O,, biotite KH2(Mg,Fe)3A1(Si0,),, vivianite Fe3P20,4?H20) and this has structural, magnetic, and electronic consequences.On the bioinorganic side, the electron-transport properties of certain metal- loporphyrin and mixed-valence iron-sulphur cluster systems are known to be important in biological processes. The subject of mixed-valence chemistry is now so far-reaching that it deserves substantial treatment in its own right in standard textbooks of Inorganic Chemistry. Although there are many intriguing properties of mixed-valence complexes, it is usually their colours which are their most striking feature. Indeed, this has been recognized industrially since at least the early 18th century6 when Prussian blue, Fe,[Fe(CN),],-14H20 (Figure 1) began to be manufactured as a dyestuff.This complex owes its deep blue colour (ijmax.= 14 100 cm-l, = lo4 M-‘ cm-’)’ not to ligand-field transitions, or to charge-transfer transitions of the * Delivered at the Tilden Symposium, University College London, on 15 November, 1983, and on other occasions at Auckland, Christchurch, Hamilton (Plenary lecture, 1983, NZIC Annual Conference), Oxford, Norwich, Edinburgh, Aberdeen, Hobart (Plenary lecture, 12th COMO Conference of the RACI), Sydney, Brisbane, Canberra, Melbourne, Perth, Leeds, Kings College London, Hull, and Canterbury. ‘Mixed-valence Compounds’, ed. D. B. Brown, Proc. NATO ASI, Oxford, Sept. 1979; Reidel, Dordrecht, 1980.‘Chemistry and Physics of One-Dimensional Metals’ ed. H. J. Keller, Plenum, New York, 1977. ’‘Synthesis and Properties of Low-Dimensional Materials’, J. S. Miller and A. J. Epstein, Ann. N.Y. Acud. Sci., 1978, 313, 1-828. ‘Molecular Metals’, ed. W. E. Hatfield, Plenum, New York, 1979. ‘Extended Linear Chain Compounds’, ed. J. S.Miller, Plenum, New York, Vols. 1 and 2, 1982; Vol. 3, 1983. ‘Miscellanea Berolinensia ad incrementum scientiarium’, Berlin, 1710, p. 377. ’H. J. Buser, D. Schwarzenbach, W. Petter, and A. Ludi, Inorg. Chem., 1977,16,2704. The Chemistry and Spectroscopy of Mixed-valence Complexes sort ligand -+ metal, metal + ligand, or ligand -ligand, but to an intervalence transition of the sort Fe" + Fe"'.Such transitions are often strongly allowed, in which case they dominate the spectrum and thus (if in the visible region) are responsible for the colour of the material. Figure 1 The unir cell of Prussian Blue, Fe,[Fe(CN),],-14H20 0= Fe", 0 = Fe"' The other striking and, at first sight, unexpected property of certain mixed- valence complexes is their electrical conductance. Both 'controlled-valence-semi-conductors' such as Li,Ni,-,O (0 < x < l), with a conductance of 10" times that of NiO, and one-dimensional conductors such as K,Pt(CN),-Br0.,,~3H,O (KCP), with a conductance of -lo9 times that of K2Pt(CN),*3H20 (Table l), display properties vastly different from those of their parent single-valence constituents. These features constitute sufficient reason for carrying out concerted studies of a wide range of mixed-valence materials. However, although many extended-chain structures are potential one-dimensional conductors, their propen- sity towards bond alternation prevents this possibility from being fully realised.A recent analysis of the question of bond alternation in one-dimensional polymers has led to the conclusion" that this feature can be largely circumvented by the appropriate design of off-axis acceptor ligands and of bridging ligands with high ligand-ligand repulsion. The basis of the static model for the classification of mixed-valence complexes S. Shaik and R.Bar, Inorg. Chem., 1983, 22, 735. Clark Table 1 Properties of chain platinum species" Species q(298 K)/n-' cm-' Colour r(Pt-Pt)/A K2[Pt(CN)4]*3H20 5 1~7 white 3.48 3 x lo2 copper 2.89K~[P~(CN)~]B~O.XY~H~O Pt metal 9.4 x lo4 silver 2.77 Ref.17. was laid down in 1967 by Robin and Day and by H~sh.~-l' In short, the basis relates to the degree to which the sites occupied by elements of different valence can be distinguished in their ground state and the ease or difficulty with which an electron can be transferred from one site (A) to another (B). Class I complexes are those in which the A and B sites are very different from one another, and thus are ones in which the valences are firmly localized (trapped-valence species). The properties of such complexes are, to a good approximation, the sums of the properties of the constituent ions.They are insulators (a < R-I cm-' at room temperature and atmospheric pressure), and any intervalence transi- tions A"B"+l -A"+lB" occur at such large energies as to lie well into the ultraviolet region. Accordingly, such complexes are of no particular spectroscopic interest. Class I1 complexes are ones in which the A and B sites are more similar to one another than is the case for class I complexes, but nevertheless ones in which the sites are distinguishable crystallographically. The valences, although localized, can exchange with a relatively small expenditure of energy. Thus intervalence transitions may, and indeed do, occur in the visible region of the spectrum, some- times with considerable oscillator strengths.Many complexes in this category form as linear chains, the prototype being Wolffram's Red, [Pt(etn),][Pt(etn)4C12]-Cl,-4H20 (etn = ethylamine), a complex involving platinum in the (formal) oxidation states II and IV (Figure 2). Such materials have conductances in the axial (chain) direction in the ranges found for insulators to semiconductors (a,,-1(~*~-1@* iz-' cm-' ). The conductance is highly anisotropic (all/a, -300) and very pressure sensitive (a,,increases by ca. lo9 on increasing the pressure to 130 kbar).I2 Class I11 (delocalized valence) complexes are ones in which the A and B sites are indistinguishable and the key element has a non-integral oxidation state. This class is usually divided into two, the class IIIA and IIIB sub-divisions.In the former, the delocalization of the valence electrons is considered (in the first instance on structural grounds) to take place only within a cluster of equivalent metal ions. No individual ion properties can be distinguished, and it is debatable whether or not the low-lying electronic transitions of such species should reason- M. B. Robin and P. Day, in 'Advances in Inorganic and Radiochemistry', ed. H. J. Emeleus and A. G. Sharpe, Academic Press, New York, Vol. 10, 1967, p. 247. lo N. S. Hush, in 'Progress in Inorganic Chemistry', ed. F. A. Cotton, Wiley, New York, Vol. 8, 1967, pp. 357 and 391. l1 P. Day in 'International Reviews in Physical Chemistry', Butterworths, London, Vol. 1, 1981, p. 149. l2 L. V. Interrante, K. W. Browall, and F.B. Bundy, Inorg. Chem., 1974, 13, 1158. 22 1 The Chemistry and Spectroscopy of Mixed-valence Complexes I I I I3.13 8, Figure 2 Structure of Wolffram’s Red, [Pt(etn),][Pt(etn),C1,]C14~4H,0, where etn = ethylamine ably be termed intervalence transitions. The bonding in such complexes is probably best treated on a molecular orbital basis, and the allowed transitions as n: -+ IT* transitions, etc., of the cluster as a whole. A typical example would be the cubic [Nb,C1,J2+ ion, which involves equivalent metal ions each in the +2.33 oxidation state (Figure 3). Class IIIB complexes are delocalized valence ones in which the lattice is continuous. Again, no properties characteristic of the constituent ions can be distinguished.Indeed, many such materials not only have a coppery bronze appearance and strikingly reflect visible light, but behave as metallic conductors (e.g.KCP, Figure 4 and Table 1). The bonding in such complexes is best treated by band theory.’ 33 l4 Representative examples of the very large number of known mixed-valence complexes are given in Table 2. Not included in the table, however, are the l3 M.-H. Whangbo and R. Hoffmann, J. Am. Chern. SOC.,1978,100,6093. l4 M.-H. Whangbo, Acc. Gem. Res., 1983, 16, 95. Clark significant number of complexes, e.g. the Creutz-Taube ion,’ ’ [(NH3)5R~-(pyrazine)Ru(NH,),] ’+,which apparently lie near to the class II/III borderline and which are therefore subject to considerable speculation as to their correct classification.The main features of interest in the rich and varied chemistry of mixed-valence complexes are the relationships between gross physical properties, (e.g. colour, electrical conductance), molecular structure, and the extent of valence-electron delocalization. Indeed, a great many physical techniques have been used in the course of defining the last, and hence the class to which a mixed-valence species belongs. Our spectroscopic interest centres around 16v1 ’ (a) the wavenumbers, intensities, bandwidths, and assignments of the intervalence transitions, (b) the behaviour of Raman bands of mixed-valence species both on and off resonance with intervalence transitions and (c) the nature of the geometric changes undergone by such complexes on excitation to the intervalence state.Information on point (c) derives from the fact that, if there is a change in geometry of a molecule on transition from the ground to an excited state, then the normal co-ordinate responsible for effecting this structural change is found to give rise to a Raman band that is much enhanced (relative to its intensity in the off-resonance situation) when the exciting line wavenumber coincides with that of the electronic transition in question. l8 @=Nb o=Cl Figure 3 Structure of [Nb,C1,2]2+ Is C. Creutz and H. Taube, J. Am. Chem. Soc., 1969,91,3988. l6 R. J. H. Clark, Ref. 1, p. 271; J. Mol. Struct., 1984, 113, 117. R. J. H. Clark, in ‘Advances in Infrared and Raman Spectroscopy’, ed. R. J.H. Clark and R. E. Hester, Wiley-Heyden, Chichester, Vol. 11, 1984, p. 95. Y. Nishimura, A. Y. Hirakawa, and M. Tsuboi, in ‘Advances in Infrared and Raman Spectroscopy’, ed. R. J. H. Clark and R. E. Hester, Heyden, London, Vol. 5, 1978, p. 217. 223 The Chemistry and Spectroscopy of Mixed-valence Complexes b b Figure 4 Structure of KCP, K2Pt(CN),-Br,,,,-3H,0 Also of importance is the connection between the physical properties of mixed- valence compounds and the rates of electron-transfer reactions in solution. The key prediction as far as mechanistic inorganic chemists are concerned is that moderately coupled mixed-valence compounds should possess an intervalence band at an energy, Eop,which is (for one-electron transfer processes) simply related to the barrier for thermal electron-transfer, Eth.For symmetric species, Eop= 4 Eth Further, the full width at half maximum (AG,) of the intervalence band should be a Clark Table 2 Examples of different classes of mixed-valence compound Formal Formal Site A Oxidation Site B Oxidation Class Example Geometry State Geometry State I [ CO(NH~)~]Z[ CoCI4] Tetrahedral I1 Octahedral 111 (high spin) (low spin) i Cu(en)~l[CuBrzIZ* Linear 1 Square planar 11 Ga[ GaC141 (= ‘GaClz’) Dodecahedra1 I Tetrahedral I11 (distorted) Ga-Cl = 3.2-3.3 A Ga-Cl = 2.19 A [Au(dmg)z][ AuCIz] ‘ Linear I Square planar 111 11 Fe4[ Fe(CN)d 3.14Hz0 Octahedral Fe-C = 1.92 A I1 Octahedral Fe-N = 2.03 A 111 [ Pt(etn)4][ Pt(etn)4C1~]- Square planar/ II Octahedral IV C14-4H2Od axially distorted octahedral (linear chain complex) [ NH~Iz[SbBr61 Sb-Br = 2.79 A Octahedral I11 Octahedral Sb-Br = 2.56 A V [ Mo M05 019]3- Octahedral V Octahedral v1 IIIA [ Fe4S4(SCH2Ph)4]’-[ Nb6C112]clz Tetrahedral 2.5 4:4: 1 Co-ordination 2.33 [ (NH~)~OSN~~S(NH~)~]5+ Octahedral 2.5 [ TCzCle] -Square pyramidal 2.5 +s42 Square planar 0.5 [(PEt~Ph)3RuC13Ru-Pseudo-octahedral 2.5 (PEtzPh)3] + IIIB K1.75Pt(CN)4.1.5H20 Square planar 2.25 (linear chain complex, Pt-Pt = 2.96 A) KZP~(CN)~-B~O.~O-~HZOSquare planar 2.30 (linear chain complex, Pt-Pt = 2.89 A) Hgz.86ASF6 Linear 0.35 Na,TiOz (0 < x < 1) Octahedral (111-IV) Na,WO3 (0.4 < x < 0.9) Octahedral (v-VI) Site A is arbitrarily taken to be the one in which the metal ion has the lower valence.en = 12-diaminoethane. ‘dmg = dimethylglyoxime. etn = ethylamine. function of the wavenumber of its maximum absorption, C,,,., which in turn should be a function of the dielectric constant of the solvent. Since 1969, a great many substitution-inert dinuclear complexes related to the Creutz-Taube ion have been synthesized in order to probe the energetics of electron-transfer processes in solution and the mechanisms and magnitudes of electronic interactions between metal centres. This aspect of mixed-valence chemistry will not be developed here since it has been reviewed recently elsewhere.’ 1719,20 The important result, in the l9 R.D. Cannon, in ‘Advances in Inorganic Chemistry and Radiochemistry’, ed. H. J. Emeleus and A. G. Sharpe, Academic Press, Vol. 21, 1978, p. 179. 2o C. Creutz, in ‘Progress in Inorganic Chemistry’, ed. S. J. Lippard, Wiley, Vol. 30, 1983, p. 1. The Chemistry and Spectroscopy of Mixed-valence Complexes context of the present article, is that class I1 mixed-valence species undergo light absorption at an energy related to Eth,whereas class I11 species do so at an energy hv = ~HAB,where HAB is the resonance electronic coupling energy between centres A and B. 2 Class I1 Linear-chain Complexes Since the properties of class I complexes are simply the sums of the properties of the component species, there is little if any new information to be learned from spectroscopic studies of such materials.By contrast, the spectroscopy of class I1 complexes is rich and varied, particularly that relating to halogen-bridged linear- chain complexes. Extensive electronic, infrared, Raman and resonance Raman studies of platinum and, to a much lesser extent, palladium complexes of this sort have yielded a very large amount of intriguing spectroscopic information thereon. The chloro-complexes of this sort may be prepared in a variety of ways, most notably by (a)reaction of equimolar amounts of the platinum(I1) and platinum(rv) constituents in a suitable solvent or (b) partial oxidation of the constituent platinum(1r) species with an oxidizing agent such as chlorine, ammonium persulphate, hydrogen peroxide, or copper(r1) halide. The exact experimental details appropriate to optimum yield and purity are different in each case.The analogous bromo- and iodo-species may be prepared in a similar manner, or from the chloro-species by halogen exchange with, for example, HBr or HI. Such complexes all have the same basic structures (Figure 2), involving five different charge types (Table 3) and in which the equatorial ligand (L) is an amine such as ANH, or EtNH, and X = C1, Br, or I. Bidentate ligands LL can also form complexes Aof this sort, where LL represents (for instance) 1,2-diaminoethane, 1,2-diamino- propane, 1,3-diaminopropane, 1,2-diaminobutane, 1,2-diaminocyclopentane, or 1,2-diaminocyclohexane. (Note, however, that not all possible combinations of L and X are known to yield isolable complexes.) There are also examples where CL represents the terdentate ligand diethylenetriamine (dien), uiz.[Pt"(dien)I]-Table 3 Halogen-bridged mixed-valence complexes of platinum Charge per Pt unit Examples" +2 +1 0 -1 -2 a etn = ethylamine, dien = diethylenetriamine, en = 12-diaminoethane. 226 Clark [Pt1V(dien)13]1221 and, in the case of palladium, where WL represents the tetradentate macrocyclic ligand 1,4,8,1l-tetra-aza-cyclotetradecane(cyclam), uiz. [Pd1'(cyclam)][Pd'v(cyclam)X2]Y~,where X = C1 or Br." Typical counter- cations are K', Cst, or NH4+ and counter-anions are C1-, Br-, I-, [HSOll-, [C104]-, [BFJ-, [NOJ-, as well as polymeric copper halide chains such as [CuX4I3- (X =C1 or Br) and [Cu3Br5I2-. This type of complex can be regarded as the archetypal class I1 or localized valence complex in which the two metal atoms differ in oxidation state by two, just as Prussian Blue is so regarded for those complexes in which the two metal atoms differ in oxidation state by one.The extent of valence-electron delocalization is considered to increase as the bridging atom changes in the order C1 < Br < I. One pointer to this is that, from the large amount of X-ray structural work carried out on these complexes (summarized in ref. 17), the r(Pt'v-X)/r(Ptll-X) ratio increases in the order of C1 < Br < I [the average values being 0.75 (chlorides), 0.79 (bromides), 0.88 (iodides)].This has the consequence that the iodine-bridged species have the highest chain conductance and intervalence bands of the lowest energies (vide infra). A curious structural feature of these complexes, (not shared by other class I1 species such as Prussian Blue) is that, although the two metal-atom sites are structurally distinguishable, they may be interconverted by a concerted movement of the axial halogen atoms in phase, away from platinum(1v) towards platinum(n), i.e. the distinction between platinum(t1) and platinum(1v) is defined only by the position of the bridging atom. The interconversion could be brought about, for instance, by consecutive one-electron jumps Pt" -+Pt" via the intermediacy of the Pt"'-Pt"' state. The ease with which such a transition takes place is connected with the one-dimensional insulator/semi-conductor nature of the complexes.Moreover the electrical conductance in the axial direction is ca. 300-times greater than that in the perpendicular directions. l2 The property of interconversion of oxidation states through a symmetrical intermediate is a feature of certain other mixed-valence systems, uiz. Sb"'.V and Pb"-lv hexahalides, etc. The Pt" --* Pt" intervalence transitions, as judged by transmission spectral measurements, occur in the region 25 -18 200 cm-' for chloro-bridged species, 23 60&14 300 cm-' for bromo-bridged species, and 20 6-7 500 cm for iodo-bridged species, the trend Cl > Br > I being consistent with the reverse trend for the electrical conductances of such complexes.The transition energies for materials that have strong absorbance and high reflectivity are difficult to determine unless single-crystal specular reflectance measurements are carried out. In this case Kramers-Kronig analysis of the data leads to plots of the real (E') and imaginary (E") parts of the dielectric constant as a function of wavenumber, these quantities being related to the transition energy or gap frequency of a semi-conductor. Alternatively, and more easily, the transition energy can be related to the excitation profile of a resonance-enhanced Raman band of the complex (;.c.:i " R. J. H. Clark, M. Kumoo, A. M. R. Galas, and M. B. Hursthouse, J. Chem.Soc. Dnlrun Trm .., 1583.''M.Yamashita, H. Ito, K. Toriumi, and T. Ito, lnorg. Chem., 1983,22, 1566. The Chemistry and Spectroscopy of Mixed-ualence Complexes plot of Raman band intensity uersus excitation line ~avenumber).~~-~~ The wavenumber of the intervalence transition is related to the Pt" PtIV chain distance, as evident from Figure 5 which pertains to the results of a large number of spectroscopic and crystallographic studies on such complexes. ' It is interesting that an extrapolation of the plots for chlorides and bromides crosses the zero for the intervalence transition energy at ca. 4.6 and 4.8 A, respectively. These values correspond approximately to twice the r(PtIv-C1) and r(Pt"-Br) distances, respectively, i.e. to the situation in which the bridging halogen atom is centrally placed between the metal atoms.The slightly erratic nature of the relationship is due to the fact that the intervalence transition energy also depends on the r(Pt'v-X)/r(Pt"-X) ratio, which in turn depends on the extent of hydrogen-bonding between the amines and the counterions. This will vary widely, from zero for neutral chain complexes to very significant values for [Pt(NH3)4][Pt(NH3)412]- [HS04]3[OH]*H20, in which one of the r(N 0)distances is as little as 2.79 A." Chlorides / r L n5 2! r ( Ptn--Ptw)/ A Figure 5 Relationship between the intervalence transition wavenumber (as deduced from the maxima in the v band excitation profiles) for linear-chain halogen-bridged platinum complexes and the Pt" --Pt'" chain distance The Raman spectra of halogen-bridged mixed-valence complexes of platinum, obtained with an exciting line resonant with the Pt" -+ PtIV intervalence band, are characterized by an enormous intensification to the Raman band attributed to the symmetric stretching mode, v1(X-Pt"-X) of the chain-halogen atoms, together with the development of an intense overtone progression u1v1, where u1 is the 23 G.C. Papavassiliou and A. D. Zdetsis, J. Chem. SOC.,Faraday Trans. 2, 1980, 104. 24 G. C. Papavassiliou and C. S. Jacobsen, J. Chem. Soc., Faraday Trans. 2, 1981, 191. 25 R. J. H. Clark and M. Kunnoo, J. Chem. SOC.,Dalton Trans., 1983, 761. 228 Clark vibrational quantum number of this mode. As many as 17 harmonics of the v1 mode have given rise to detectable bands in the resonance Raman spectra of some complexes of this sort.Much weaker, subsidiary progressions also appear in many cases, these mainly consisting of progressions in v1 based upon one quantum of another Raman-active mode. The resonance Raman spectrum of Wolffram's Red (Figure 6), which is typical, clearly indicates the dominant features of the ulvl progression (vl -310 cm-I). "IVI,, , , , , I, , , , , , , , I v1=16 15 14 13 12 II 10 9 8 7 6 5 4 3 2 I 5000 4000 3000 2000 1000 0 Wavenumber /cm-I Figure 6 Resonance Raman spectrum of Wolffratn's Red [P t( e tn),] [Pt(etn),Cl 2] C1,-4H ,O at ca. 80 K using an exciting line of wavelength 514.5 nm A vast amount of spectroscopic information on complexes of this sort has now been gathered and reviewed." From such data it is possible, by standard procedures,26 to calculate approximate values for the harmonic wavenumbers (a1) and anharmonicity constants (xll).Cross terms (xij)may also be determined if subsidiary progressions are observed in the spectra.The average values of o1for a wide variety of bridged complexes are given in Table 4, from which it is clear that, for any given halide, o1falls slightly in the order cation chain >neutral chain > anion chain. The average Pt-X bond lengths increase slightly in this same order, cf: the well known reciprocal relationship between bond lengths and bond stretching frequencies.27 The known xll values, which average -0.98, -0.41, and -0.25 cm-' for chlorides, bromides, and iodides, respectively, are about an order of magnitude 2b R.J.H.Clark and B.Stewart, Struct. Bonding (Berlin), 1979,36,1. 27 R.J. H. Clark, Spectrochim.Acfa, 1965, 21,955. 229 The Chemistry and Spectroscopy of Mixed-valence Complexes Table 4 Average values for ol/cn-'for different types of linear-chain mixed-valence com- plex of platinum a Charge type Bridging group Cationic Neutral Anionic C1 309.1 307.8 297.8 Br 175.7 168.6 172.0 I 122.3 120.8 114.2 "Ref 17 larger than those found for typical single-valent molecular species such as group IV tetra halide^.^^.^^ Nevertheless, they are not large and, curiously, plots of ulvl/vI versus v1 remain linear even out as far as 17v,.Although 17v, occurs at ca. 5 OOO cm-' up thepotptialenergycurve,itmust nevertheless lie below theexpectedcrossing points between the potential energy curves for (II,IV), (III,III), and (IV,II) (Figure 7). leading to no deviations from linearity for the higher harmonics. Presumably, owing to the rather low oscillator strengths (f -10-4)30,31 of the intervalence transitions for linear-chain platinum complexes (high chromophore concentrations in the complexes is responsible for their deep colours rather than high oscillator strengths) the extent of interaction between the different potential energy curves is slight, and they remain nearly parabolic even far from their minima. Note that the position of the minimum in the (111,111) curve relative to the cross-over point between the (IIJV) and (IV,II) points is uncertain.The implication of the long overtone progression observed in the Raman spectrum of such complexes at resonance with the intervalence transition is that the Pt"-X bond lengths must extend very substantially on excitation from the ground to the resonant excited state. It is possible, by way of the theories of Mingardi and Siebrand 32 and Clark and Stewart, 33 to calculate the magnitudes of these bond-length changes on excitation, but usually only for those species for which the resonant electronic transition is vibronically structured. This is not the case here unfortunately, even at liquid helium temperat~re.~' However, by analogy with the results for species which do have structured resonant bands, viz.Mn04-, MoS,~-, WS4 2-7 33-36 the PtIV-C1 bond length increases in the intervalence state by probably 0.14.2 A. Such a large 28 R. J. H. Clark and P. D. Mitchell, J. Am. Chem. SOC.,1973,%, 8300. 29 R. J. H. Clark and T. J. Dines, Inorg. Chem., 1980, 19, 1681. 30 H. Tanino, J. Nakahara, and K. Kobayashi, J. Phys. SOC.Jpn., 1980,49, Suppl.A, 695. 31 R. Aoki, Y. Hamaue, S. Kida, M. Yamashita, T. Takemura, Y. Furuta, and A. Kawamori, Mof.Cryst. Liq. Cryst., 1982, 81, 301. 32 M. Mingardi and W. Siebrand, J. Chem. Phys., 1975,62, 1074. 33 R. J. H. Clark and B. Stewart, J. Am. Chem. Soc., 1981,103,6593. 34 R. J. H. Clark, T. J. Dines, and M. L. Wolf, J. Chem. Soc., Furaduy Trans. 2, 1982,78, 679. 35 R.J. H. Clark, T. J. Dines, and G. P. Proud, J. Chem. SOC.,Dalton Trans, 1983,2019. 36 R. J. H. Clark, Adv. Chem. Ser., 1983,211, 509. Clark Figure 7 Potential energy surfaces for the (I1,IV) and (IVJI) states of a linear-chainplarinum complex, showing the optical (Franck-Condon) intervalence transitions (E, ). Note (a) that the (IIJV) and (IVJI) surfaces do not interact in first order, owing to tXe fact that direct interconversion corresponds to a two-electron jump and (6) that the position of the minimum in the (111,111) curve relative to the cross-over point between the (IIJV) and (IVJI) curves is not known. The dashed lines indicate 'non-crossing ' regions Ground-state geometry ...pt" ...X-Pt'V-X ...pt" ...x-pt'V-x .. . Excited-state geometry ...pp...x.. .pt"'.. .x.. .pp.. .x-pp-x.. . Figure 8 Geometric distortion of linear-chain complexes on excitation to the intervalence state bond-length change is consistent with the various proposed processes (Figure 8) for the axial conductivity of such materials. This structural change is just that expected on excitation to the (111,111) state since, in its relaxed excited-state geometry, the axial halogen atom would have no reason to be other than centrally placed between the two platinum atoms. Consistent with all these results, the large amount of crystallographic data on these complexes l7 reveals that the halogen would need to 23 1 The Chemistry and Spectroscopy of Mixed-valence Complexes move 0.1-0.4A in the chain direction in order to reach the central position between the metal atoms.It is rare that subsidiary progressions in v,, based on one or more quanta of another Raman-active mode, have much intensity, consistent with the resonant transition being z-polarized and localized in the chain direction. However, in the case of the complexes Cs2[Pt(N02)(NH3)X2][Pt(N02)(NH3)X4],X = Br or I, some equatorial modes do give rise to Raman bands enhanced at resonance with the intervalence band.37 Specifically, the resonance Raman spectrum of the bromo-complex consists of eleven progressions of three or more members, in nine of which v1 acts as the progression-forming mode. In some of these progressions the equatorial mode v2, the symmetric Br-Pt"-Br stretching mode of the trans- equatorial bromides, and v(N02), and 6(N02),, the symmetric nitro-group stretching and bending modes, are involved.This indicates that the Pt" -, Pt'" intervalence transition, by altering the effective positive charge on each metal atom, affects the extent of Pt(d,,)+NO,(n*) back bonding and thus the equilibrium geometry of the co-ordinated NO2 group in the intervalence state. The unusually large involvement of modes of the equatorial ligands in this case is partly due to the n-acceptor nature of the nitro-group and partly to the anion-chain (rather than cation-chain) nature of the complex. The relationship between structure and property for linear-chain halogen- bridged complexes has recently been explored by studying the one-electron band structures determined from the tight-binding scheme based on the extended Hiickel method.38 The treatment recognizes that, owing to the chain nature of the complexes, their band electronic structures must be taken into account.Con- sideration has been given to the way in which electrical properties depend on geometric distortion of the repeat unit. The intervalence transition energy is found experimentally to depend slightly on chain length, long crystals (and therefore long chains) having lower transition energies than short chain^.^^-^' This result can also be reproduced by extended Huckel-type calculation^,^^ but only where the chain is terminated by platinum(rv) rather than by platinum(I1). Since the platinum(I1)-bridge bonds are the weakest in the chain, preferential cleavage at this position is to be expected, leading to just the same conclusion as to the nature of the chain terminating group.Many other aspects of the vibrational spectra of these chain complexes have been studied recently in considerable detail. These include 40-43 (a) force constant analysis, (Table 5), based upon the assumption that the normal modes of such complexes may be divided into chain (localized MI") modes and equatorial modes, which are almost independent local modes (note that the progression- 37 R. J. H. Clark and M. Kurmoo, J. Chem. Sor., Chem. Commun., 1980, 1258; R. J. H. Clark, M. Kurmoo, A. M. R. Galas, and M. B. Hursthouse, Inorg. Chem., 1981,20,4206. 38 M.-H. Whangbo and M.J. Foshee, Inorg. Chem., 1981,20, 113. 39 C. G. Barraclough, unpublished results. "C. E. Paraskevaidis and C. Papatrianfillou, Chem. Phys. Lett., 1978,58, 301.'' C. E. Paraskevaidis, C. Papatrianfillou, and G. C. Papavassiliou, Chem. Phys., 1979, 37,389.'' C. G. Barraclough, R. J. H. Clark, and M. Kurmoo, J. Struct. Chem., 1982, 79, 239. 43 S. D. Allen, R. J. H. Clark, V. B. Croud, and M. Kurmoo, Philos. Trans. R.Soc. Lomdon, in press. Clark forming mode, v,, is strictly an even parity longitudinal optical mode of k =0 of the chain), (b) f.t.i.r. studies43 to 20 cm-' (c) studies of the dependence of v, on the wavenumber of the exciting line, a phenomenon which is undoubtedly connected with the degree of electron delocalization in the chain dire~tion,,~ cf: the behaviour of certain modes of the semiconductors Cu,O, CdS, and CdSe.45 E.s.r.measurements, moreover, have demonstrated that only about one platinum atom in lo4 in these complexes has an unpaired electron (and in a d,Z orbital), which suggests that this centre may be located at structure dislocations or at thermal defects in the chain.31 The activation energy to create the Pt"' sites is related to the energy required to displace the bridging halide at the defect site. +Table 5 Pt-C1 Force constants for Complexes containing the tran~-[Pt'~(en)2C12]~ group" Complex f,/102N m-' f,,/102N m-' [Pt(en)2C12]Cl2 2.278 0.228 [Ni(en)z] [Pt(en)zC12] [C104]4 2.062 0.135 [Pd(en)2] [Pt(en)Xlz] [C1O4I4 2.105 0.121 [Pt(en)21 [Pt(en)2C12][C104]4 1.991 0.027 "$ = Pt-CI stretching force constant,$, = stretch-stretch interaction constant (Ref.43). Many fewer mixed-valence halogen-bridged complexes of palladium than of platinum have yet been synthesized, but those that have display broadly similar physical properties to those discussed above.46 The vI band wavenumber of the isolated [M1V(LL)2X2]2 entity is lowered more for palladium than for platinum + on chain formation. Moreover, the intervalence band maxima, excitation profile maxima, and v, values for the palladium complexes are all of lower wavenumber than those for analogous platinum complexes, even in those cases, e.g. [M(en)2][M(en)2C1,][C104]4,for which the r(M1v-Cl)/r(Mu-Cl) values are identical.Since the electrical conductance of these materials is believed to be phonon assisted," low v, values and low intervalence transition energies would point to higher conductance for mixed-valence palladium than platinum complexes, a conclusion in complete agreement with e~periment.~ Such palladium complexes therefore have less localized valences than have the analogous platinum complexes. The reverse is true for mixed-metal mixed-valence chain complexes, some of which have recently been synthesized, uiz. [M11(en)2][Pt'v(en)2C1,]-[ClO,],, M = Ni, Pd, or Pt; en = 1,2-diaminoethane. The intervalence transitions of the mixed-metal complexes occur at higher energies than for analogous Pd"/PdIv or Pt"/PtIv complexes, implying a greater degree of valence-electron localiza- tion and lower chain-conductance for the mixed-metal derivatives.The status of possible linear-chain nickel(Ir)/nickel(Iv) complexes seems uncertain at this stage. A further intriguing feature of this sort of localized-valence complex is the 44 R. J. H. Clark and M. Kurmoo, J. Chem. SOC.,Furuduy Trans. 2, 1983, 79, 519. "P. Y. Yu, in 'Topics in Current Physics', ed. K. Cho, Springer-Verlag, Berlin, Vol. 14, 1979, p. 21 1. 46 R. J. H. Clark, V. B. Croud, and M. Kumoo, Inorg. Chem., in press. The Chemistry and Spectroscopy of Mixed-valence Complexes prediction 47 that, on excitation within the contour of the intervalence band, each should display electronic Raman scattering of intensity comparable to, or possibly even up to 103-times more intense than that of the vibrational Raman scattering already discussed.The electronic Raman scattering corresponds to transitions between the two weakly coupled potential surfaces (Figure 7). Although the analysis fails to take account of the fact that the lowest intervalence transition is from the (I1,IV) to the (111,111) rather than the (IV,II) state, the broad features of the prediction appear to be correct, since an intense band akin to fluorescence has been observed at large (-13 000 cm-') Stokes Raman shifts from the exciting line for complexes such as [Pd(en)2][Pt(en)2Br2][C104]4.48 Both a continuous, z-polarized emission underneath the Raman bands, as well as the emission described above, have alternatively been regarded as photo-luminescence arising from the recombination of electron and hole during and after relaxation of the charge- transfer excited state into the self-trapped (-Pt3 + ---C1----Pt3+---) excited state.49 These phenomena are still under investigation.A final point should be made regarding the difficulty in solving the X-ray crystal structures of this type of chain complex. It appears that, although any individual chain is ordered in terms of being alternately platinum(1v) and platinum(u), adjacent chains may be arbitrarily slipped by c/2; this leads to disorder in the directions perpendicular to the chain (c)axis and to diffuse features in the X-ray pattern. A further difficulty is caused by the fact that many such mixed-valence complexes can be formed as mixtures of conformational isomers unless the resolved diamine is used in their preparation." 3 Class I1 Three-dimensional Complexes The structural changes on excitation to the intervalence state of those class I1 mixed-valence complexes which involve essentially three-dimensional lattices are much smaller than those for linear-chain (essentially one-dimensional) species, and in consequence the degree of resonance enhancement of the appropriate bands is much smaller.Thus, although complexes such as Cs,[Sb"'C1,][SbvC1,], [CO(NH~),]~[P~"C~,][P~~~C~,],and Prussian Blue display Raman bands which are considerably enhanced at resonance with the Sb"' -+ SbV(V,,,~,.. = 17 900 cm-', E,,,. = 110 M-' cm-I), Pb" ---+ Pb", and Fe" -+ Fe"' intervalence bands respectively, the overtone and combination band progressions are short." Clearly, since the structural changes following the electronic transition to the intervalence state are distributed over far more bonds in three-dimensional than in one- dimensional species, each individual change is much smaller in the former case.For Cs,[Sb"'CI,][SbVC1,], the main progression-forming mode is v1 (al,) of the [SbCl,]-ion, the subsidiary progressions being vlvl + vl' (where vl' is the v1 (al,) 47 K. Y. Wong and P. N. Schatz, Chem. Phys. Lett., 1981,80, 172; K. Y.Wong, ibid.,1984,108,484. 48 R. J. H. Clark and M. Kurmoo, unpublished results, 1981. 49 H. Tanino and K. Kobayashi, J. Phys. SOC.Jpn., 1983,52, 1446.50 R. J. H. Clark, M. Kurmoo, D. N. Mountney, and H. Toftlund, J. Chem. Soc., Dalton Trans., 1982, 1851. 51 R. J. H. Clark and W. R. Trumble, J. Chem.Soc., Chem. Commun., 1975,318; R. J. H. Clark and W. R. Trumble, J. Chem. Soc.. Dalton Trans., 1976, 1145. 234 Clark mode of the [SbC16I3- ion), ulvl + v5 (where v5 is the tzg mode of the [SbCls] ion) and ulvl + VL (where vL is a lattice mode at 60 cm-'). The resonance Raman spectrum of the system CS~[S~"'~S~~,S~~ -ZxCl,], x = 0.18, shows host [Sncl6]'- modes as well as those of the [SbC16]3-/[SbC16] chr~mophore.'~Since the exciting lines used had much lower wavenumbers than that of the first charge-transfer band of the [SnC16I2- ion, this result indicates the strong dynamic involvement of the host lattice during the Sb"' -+ SbV dopant electron-transfer process.Moreover, it is regarded as providing the first direct experimental observation of the dynamics associated with outer-sphere chemical activation for electron-transfer processes. The shape of the intervalence band of a single crystal of [CH3NH3]2[Sbr1',- SbV,Snl-2,C16] between 300 and 4 K is Gaussian.53 From the temperature dependence of the second moment of the band, the electron-phonon coupling constant and hence the displacement in the vibrational co-ordinate from the ground to the intervalence state can be deduced, along with the effective phonon wavenumber coupled to the transition. The latter (290 cm-*) is close to the mean wavenumber of the ground state totally symmetric fundamentals of [SbC16I3 (267 em-') and [SbC16]-(327 cm-').Moreover, the displacement in the vibrational co-ordinate is calculated to be ca. 0.19 A, which approaches half the difference between the Sb-C1 bond lengths (0.262/2 = 0.131 A). It is concluded that the electron transfer in this salt is coupled to the antisymmetric combination of the two totally symmetric fundamentals mentioned above, and that the Franck- Condon factors associated with this mode are largely responsible for the breadth of the intervalence band. It is also interesting that the mixed-valence salt Rb2.67SbC16, which must involve a [SbC16]-: [SbC16I3- ratio of 1:5, displays a resonance Raman spectrum closely similar to that of CszSbCl6. Detailed studies of such salts are still in progress.4 Complexes on the Class IIlIII Borderline For a polyatomic molecule in which there are several totally symmetric normal modes, each will involve the motion of a large number of atoms. If the valence electrons are delocalized in the ground or excited states or in both, and the extent of delocalization is changed on electronic excitation, then many atoms need to undergo small displacements in order that the molecule may reach its new equilibrium geometry i.e. there must be a displacement, albeit small, of the excited- state potential minimum along several totally symmetric normal co-ordinates. This situation gives rise to the so-called 'small displacement approximation' in resonance Raman theory, and has the consequence that the Franck-Condon factors are significant for all the totally symmetric fundamentals but not for any overtones. Mixed-valence complexes which lie near to the class II/III borderline have been shown to behave in the manner described above; thus their resonance Raman spectra are very different from those displayed by class I1 linear-chain complexes '' H.W. Clark and B. I. Swanson, J. Am. Chem. SOC.,1979,101, 1604. s3 K. Prassides and P. Day, J. Chem. SOC..Faraday Trans. 2, 1984,80,85; see also P. Day, Inorg. Chem., 1963,2,452 and L. Atkinson and P. Day, J. Chem. SOC.(A), 1969,2423,2432. 235 The Chemistry and Spectroscopy of Mixed-valence Complexes for which any structural change on excitation is confined to being along a single co- ordinate.Good examples of borderline complexes are provided by the cytological dyes Ruthenium Red [Ru302(NH3)14I6+ and Ruthenium Brown [Ru302(NH3)14I7 and by related 1,2-diaminoethane-substitutedspecie^.^^.^^+ These ions are linear, and probably have the eclipsed configuration in solution, uiz. All the totally symmetric skeletal modes of these complexes are enhanced at resonance with the lowest e,* --+e,*, n-type transition (Vmax. = 18 700 cm-', Emax. = 7 x lo4 M-' ~rn-');~~moreover, no overtones of significant intensity are detected. Thus although Ruthenium Red formally involves RU"~RU'~RU~~', it behaves virtually as though it were a delocalized-valence complex with equivalent (and therefore non-integral) metal-ion valences of approximately 3.33.Ruthenium Brown, for which the analogous transition lies at 21 800 cm-' (Emax, = 8 x lo4 M-' cm-l), formally involves RU"'RU'~R~~~ but behaves likewise as a delocalized- valence complex with nearly equivalent metal-ion valences of approximately 3.67. Another complex about which there has been considerable controversy regarding its classification, virtually since its discovery in 1969, is the Creutz-Taube ion,' whose structure (as well as those of its 4+ and 6 + analogue^)^^ is illustrated below: Note that the plane of the bridging pyrazine ligand bisects the angle between the equatorially co-ordinated NH3-Ru-NH3 groups. The controversy relates to whether the complex should be regarded as class 11, trapped valence (Ru"/Ru"') or as class 111, delocalized valence, possible delocalization of the odd d,, electron being effected via the pyrazine n* orbitals.Many studies of different salts of this ion, involving electronic, photoelectron, infrared, Raman, resonance Raman, Mossbauer, and electron spin resonance spectroscopy and magnetic measure- ments, although favouring the delocalized picture, have not provided a definitive answer to this question." Part of the problem relates to the sensitivity of the ion to photodegradation, an effect which seriously compromised many of the early measurements. Moreover, even the most recent study, an extensive collaborative 54 J. R. Campbell, R. J. H. Clark, W. P. Griffith, and J. Hall, J. Chem. Soc., Dalton Trans.,1980, 2228.R. J. H. Clark and T. J. Dines, Mol. Phys., 1981,42, 193. 55 R. J. H. Clark and M. Kurmoo, unpublished work. 56 U. Fiirholz, A. Ludi, H.-B. Burgi, F. E. Wagner,A. Stebler, J. H. Ammeter, E. Krausz,R. J. H. Clark, and M. J. Stead, J. Am. Chem. Sor., 1984, 106, 121. Clark one involving many techniques, was not wholly conclusive on this matter, although the weight of evidence remains in favour of the delocalized-valence description. It is possible that the Creutz-Taube ion is class IIIA on a long time scale (e.g. crystallographic or nuclear magnetic resonance) but class I1 on a short time-scale (e.g. photoelectron). What is certain, however, in view of the conflicting results on this matter, is that the complex must be close to the class II/III borderline.It is also worth noting that there are only slight (2 0.1 A) changes to any of the Ru-N bond lengths on one-electron oxidation or reduction of the Creutz-Taube ion.56 Resonance Raman studies of the Creutz-Taube ion have so far been restricted to the use of exciting lines with wavenumbers falling within the contour of the Ru(d,) -,pyrazine (n*)charge-transfer transition, which reaches a maximum at ca. 20 000 cm-l 51. This leads to resonance enhancement of many skeletal modes of the complex, in a manner similar to that observed for the mononuclear complexes 58 [Ru"(NH,),L][PF~]~ (L = 4-cyano-1-methylpyridiniumor l-methyl-4,4'-bi-pyridinium) and [Ru~~'(NH,),L]C~, (L = 4-dimethylaminopyridine)at resonance with the analogous Ru(d,) ---+ pyrazine (n*)transition.It has not yet proved possible to irradiate within the contour of the intervalence band of the Creutz-Taube ion (qmax. = 6 370 cm-', E,,,, = 5 OOO M-' cm-') owing to the lack of exciting lines in, and detectors for, this low-wavenumber region. This is unfortunate, since intriguing predictions as to the nature of the Raman spectrum of this ion at resonance with the intervalence transition have been mades9 and cannot at this stage be tested. One possibility for a suitable exciting line is the second Stokes hydrogen-shifted emission (ca. -8 OOO cm-' shifted) from Rhodamine 640 (with or without Oxazine 720), the pump line being provided by a frequency doubled Nd:YAG laser (532 nm). This would provide radiation in the 6000 cm-' region; however, the technical problems associated with this experiment have not yet been fully overcome.5 Class In A Simple Radical Species.-The most obvious species which could be regarded as class IIIA, delocalized valence, in type are simple radical cations or anions. It is, of course, a matter of taste as to whether or not such species should be so regarded, but they clearly are limiting versions of this type. Some such species have been first characterized by electronic and resonance Raman spectroscopy, others have had various features of their structures and bonding clarified by such studies. The Xe, ion is an example of an ion first characterized in a condensed phase by + the techniques mentioned above (and others).The ion is prepared by passing xenon gas through the orange solution of 02+[SbF6]-in SbF,, whereupon the colour changes to green (kmaX,= 710 nm, E,,,. =6 O00 M-' cm-'). Use of 530.8 nm excitation, which corresponds to an absorption minimum for the ion, yields an unspectacular Raman spectrum dominated by bands attributed to the SbF stretching modes of the solvent (Figure 9). However, by changmg to 676.4 nm "M. J. Stead, PhD. thesis, University of London, 1983; R. J. H. Clark and M. J. Stead, to be published.''R. J. H. Clark and M. J. Stead, J. Chem. Suc., Dalton Trans., 1981, 1760. 59 K. Y. Wong and P. N. Schatz, Chem. Phys. Lett., 1980,73,456. The Chemistry and Spectroscopy of Mixed-valence Complexes excitation, which falls within the contour of the 710 nm band of the chromophore, a much more intense Raman spectrum is obtained, dominated by a progression of four members (all polarized) in a mode of wavenumber 123 cm-'.The latter is insensitive to "0 substitution, but is lowered by 2.4 cm-' on substituting natural abundance xenon (RAM =131.3) by 136Xe.60 These features allow the chromo- phore to be identified as Xe2+, the first noble gas cation to be characterized in a condensed phase (as Xe,+[Sb,F, 'I-). The transition with which the 676.4 nm line is in resonance is considered to be 'ZfY+-+ 'ride, by analogy with the situation for the isoelectronic ion I, -,for which very similar electronic and resonance Raman spectra have been obtained (Table 6). Both Xe,+ and 1,-have much lower vibrational frequencies than has I,, consistent with the molecular orbital picture of the two ions as being held together by a bond of order one half, whereas I, itself is single bonded.These ions can be regarded as examples of Xe+'.' and I-'.', respectively. Obviously many polyiodides and other related species are of a similar type. X, 676.4nm V I I I I 1 I 1 A, 530.8nm 800 600 400 200 0 Wavenumber / cm-' Figure 9 Raman spectrum of [Xe,+][Sb,F,,]-in SbF, solution both on (A, = 676.4 nm excitation) and off(ho = 530.8 nm excitation) resonance with the R +IC*transition of lowest energy of the cation 60 L. Stein, J. R. Norris,A.J. Downs, and A. R. Minihan, J. Chem. SOC.,Chem. Commun.,1978,502;R. J. H. Clark and D. G. Cobbold, unpublished work, 1978.Clark Table 6 Spectroscopic data on related diatomic species Ground Species Colour State h,,,./nm Overtones o,/cm-' w,x,/cm-' State Xe2+ green 2Ztu+ 710" 4 123 0+1 b 12-green 'Ctu+ 800 6 114.2 0.50 C I2 red-violet 'Cog+ 505 20 214.5 0.61 d E -6 000 M-' cm-I. [ SbzF111-salt, in SbF5 solution. K+ salt, the resonance Raman spectrum being measured on an argon matrix of the salt at ca. 12 K. W. F. Howard and L. Andrews, J. Am. Chem. SOC.,1975,97,2956; L. Andrews, J. Am. Chem. SOC.,1976,98,2152. * Gas phase data: R. F. Barrow and K. K. Yee, J. Chem. SOC.,Faraday Trans. 2, 1973,69,684. Several other simple ions provide examples of elements in the +0.5 oxidation state. Thus the cations S42+, Sed2+, and Te42f involve the square planar M-M2+ 6x-electron system I 67t I ,which has D4n symmetry.They have all recently been M-M characterized in sulphuric acid or oleum media by resonance Raman techniques (which are remarkably adept at probing the nature of chromophores in hostile environments). All exhibit progressions in the v1 (a;,)stretching mode at resonance with the 'A;, --+ 'E,[x(e,) -x*(b2,)] transition, the electric-dipole allowed one of lowest wavenumber, which maximizes at 330 nm for S42+, 410 nm for Se42+, and 510 nm for Ted2+ (Figure Since in these cases there is only a single co- ordinate, Ql,along which any substantial structural change could be effected on excitation, this type of species yields resonance Raman spectra similar to those of linear-chain, class I1 species for which likewise only a single co-ordinate is affected by electronic excitation.Some spectroscopic data on these intriguing ions are given in Table 7. 09 Figure 10 x-Molecular orbital scheme for the 6x-electron species Md2+ 61 R. J. H. Clark, T. J. Dines, and L. T. H. Ferris, J. Chem. SOC.,Dalton Trans., 1982, 2237. 62 R. C. Burns and R. J. Gillespie, Inorg. Chem., 1982,21, 3877. 239 The Chemistry and Spectroscopy of Mixed-valence Complexes Table 7 Spectroscopic data on S4'+, Se4' ', and Te4' 'a s4' + Se4' + Te4" (65% oleum) (25% oleum) (HzS04) 583.6 321.3 219 371.2 182.3 109 598.1 321.3 219 542 302 187 330 410 510 475 5 900 5000 6v1 5V1 6v 1 584.7 321.8 219.5 -0.35 -0.55 -0.30 2.78 2.10 1.45 a Ref. 70.At resonance with the II --+ K* transition indicated; these are analogous transitions since the m.c.d. spectra of the ions in this region are identical to each other in form, magnitude, and sign. P. J. Stephens, Chem. Comm., 1969, 1496. 'f = stretching force constant. Other simple sulphur ions, but anions in this case, have been shown by electron spin resonance, electronic and resonance Raman techniques to be responsible for the intense colours generated in many sulphur-containing media. The blue colour formed when alkali-metal polysulphides are dissolved in highly polar solvents such as dimethylformamide or hexa-methylphosphoramide is now known to be due to the S3-ion, a radical anion with an SSS angle of 105" and hma,.= 600 nm, Emax. = lo4 M-' cm-'. The same species is responsible for the blue colour formed by sulphur in a LiCl-KC1 eutectic, in CsC1-AICl3 or KNCS melts, in the aluminosilicate mineral lapis lazuli, and in its synthetic equivalent Ultramarine For over 5000 years lapis lazuli has been prized as a semi-precious gemstone and as a rich blue pigment for oil-paintings. Ultramarine Blue, which has been manufactured since 1828 for its pigmentary properties, contains the S3-ion trapped at up to 50% occupancy at the cubic sites in the aluminosilicate framework. The radical nature of the chromophore is apparent from the e.s.r. spectrum of Ultramarine Blue, and the chromophore itself can be identified from the Raman band progression (fundamental = 550 cm-') observed at resonance with the 600 nm electronic band.63 W. Holzer, W. F. Murphy, and H. J. Bernstein, J. Mol. Spectrosc., 1969, 32, 13. 64 F. See1 and H. J. Guttler, Angew. Chem., Int. Ed. Engl., 1973, 12, 420. R. J. H. Clark and M. L. Franks, Chem. Phys. Lett., 1975,34,69. 66 T. Chivers in 'New Uses of Sulphur',ed. J. R.West, American Chemical Society, Washington D.C., Adv. Chem. Ser. No. 140, p. 499. 67 R. J. H. Clark and D. G. Cobbold, Inorg. Chem., 1978, 17, 3169. R. W. Berg, N. J. Bjerrum, G. N. Papatheodorou, and S. von Winbush, Inorg. Nucl. Chem. Lett., 1980,16, 201. 69 R.J. H. Clark, D. P. Fairclough, and M. Kurmoo, in 'Time-Resolved Vibrational Spectroscopy', ed. G. H. Atkinson, Academic Press, 1983, p.213. 'O R. J. H. Clark, T. J. Dines, and M. Kurmoo, Inorg. Chem., 1983, 22, 2766. Clark By various modifications to the preparative procedures it is possible to synthesize other pigmentary materials, known as Ultramarine Green, many different shades of Ultramarine Violet, and Ultramarine Pink. These clearly contain additional chromophores. One of these, with its lowest allowed electronic transition at 380-400 nm, has been identified as the S2- ion; this species is present in small amounts in Ultramarine Blue, Violet, and Pink, but in much larger amounts in Ultramarine Green. The third species, principally responsible for the colour of Ultramarine Pink (hmax.= 520 nm) has not yet been certainly identified, Table 8 Spectroscopic properties of diatomic sulphur and selenium species Species Electronic state oe/Cm-' OeXc/Cm-' S/A a 32S2 3c,-725.68 2.852{ o.291 -3cu 434.0 2.75 o.30 S2 -2nts 597.0 2.50 2n.360 -*0se2 1x2 391.77 1.06 ) o.293 281.11 2.65 329.6 0.70 1 o.32 216 -a 6 = bond length change on excitation, as deduced for neutral species by gas phase electronic spectro- scopy, and for ions from Raman band excitation profiles obtained with exciting lines near resonance with the specified excited state (Ref. 70). though it may be S4-. The Se2- ion has likewise been identified as the chromophore present in Selenium Ultramarine (kmax.= 490 nm). Various spectroscopic properties of these diatomic ions are compared with those of the analogous neutral molecules in Table 8.The additional electron in the case of the radicals clearly enters an antibonding orbital, in view of the large drop in vibrational frequency on passing from the neutral species to the anion, but the bond length elongation on excitation to the lowest excited state is in all cases very similar. Thus the key chromophores in all the ultramarines are simple class IIIA radical anions. What feature of the Ultramarine lattice is responsible for its ability to trap and stabilize these normally unstable radicals is not known. Yet the fact that it does is entirely responsible for our interest in these materials as pigments for paints, plastics, rubbers, talcum powders, and eye shadows! Cluste; Sjwcies.-Of the many different sorts of cluster species known, it is arguable that the most important are the cubane-type FeS clusters of the sort [Fe4S,(SR)4]2-, R = alkyl or aryl, since these have been shown to be close representations, or synthetic analogues, of the redox centres present in a variety of different iron-sulphur proteins.Such complexes act as structural models and indicators of the oxidation levels of the protein active-~ites.~~'~~ In particular, these prototypes possess a common oxidation level with oxidized ferredoxin (Fd,,) and 71 R. H. Holm, Endeavour, 1975, 38. 72 R. H. Holm, Chem. Soc. Rev., 1981,10,455. 241 The Chemistry and Spectroscopy of Mixed-valence Complexes chromatium (HPred)in that all contain 2Fe" and 2Fe"'. The key question, in the context of the present review, is whether the valence electrons are localized or delocalized.In fact, a variety of physical techniques, in particular Mossbauer, electron spin resonance and nuclear magnetic resonance spectroscopy but more recently resonance Raman spectroscopy, have demonstrated that all the metal atom sites are equivalent and thus that such clusters are electronically fully delocalized rather than trapped-valence spe~ies.~~,~~ This result ties in well with the most important biophysical property of iron-sulphur proteins, which is that of electron transfer; rapid electron transfer is aided if the geometrical rearrangements around the metal-ion site are small, as implied by the resonance Raman results. The complex in question thus involves Fe+2-5, and it can be one-electron oxidized or reduced to a species involving Fe2.75 or respectively. The [MS4l2- ions, M = Mo or W, which are known to display long pro- gressions in the vl(a1) mode at resonance with the s@,)+M(d,) charge-transfer band,34*35,75do not behave likewise when they are incorporated in dinuclear species such as [(P~S)ZF~S~MS~]~-.~~ At resonance with the corresponding electronic band of such species, all the skeletal fundamentals (and not the overtones) are enhanced, suggesting that the valence electrons are, in this case, largely delocalized over the whole complex ion, a conclusion also suggested by Mossbauer results.The intriguing question as to the degree of delocalization of the valence electrons in cluster species involving a hetero atom, e.g.Fe3Mo, Fe3W, CU~MO, et~.,~~,~~ remains to be investigated. OTTLE (optically transparent thin layer electrochemical) cells may permit the in situ spectroscopic study of electrochemically generated species, a technique of immense but as yet largely unexplored promise. One application already investigated has been to the study of the electronic spectrum of incompletely reduced ruthenium bipyridyl and related species. The new absorption band generated on reduction at ca. 4000 cm-' (E = 100-345 M-' cm-') has been interpreted as arising from the ligand-based intervalence charge-transfer trans- ition, bipy- -bi~y.~~ There are also series of mixed-valence ruthenium complexes, the class of which depends on the closeness of matching of the ligands attached to each metal atom.Thus extensive electrochemical studies have established the existence of several confacial bioctahedral complexes of the sorts [L3RuCl3RuL3l2+ and [L3-,Cl,RuC13RuClyL3_,1"+, where L = PEtZPh, As(tol)3, or PPh3, which display at least one, and generally two, stepwise reversible one-electron transfer reactions without there being any change in gross molecular structure." One 73 D. W. Stephan, G. C. Papaefthymion, R. B. Frankel, and R. H. Holm, Inorg. Chem., 1983,22, 1550. 74 G.D. Friesen,J. W. McDonald, W. E. Newton, W. B. Euler,and B. M. Hoffman, Inorg. Chem., 1983,22, 2202. 75 R. J. H. Clark, ACS Symposium Series, No. 21 1. 'Inorganic Chemistry: Toward the 21st Century', ed.M. H. Chisholm, 1983, p. 509. "R. J. H. Clark, T. J. Dines, and G. P. Proud, J. Chem. SOC.,Dalton Trans., 1983,2299.''C. D. Garner, to be published.''A. Miiller, E. Diemann, R. Jostes, and H. Bogge, Angew. Chem., In&.Ed. Engl., 1981, 20,934. "G. A. Heath, L. J. Yellowlees, and P. S. Bratermann, Chem. Phys. Lett., 1982, 92, 646. G. A. Heath, A. J. Lindsay, T. A. Stevenson, and D. K. Vattis, J. Organomet. Chem., 1982,233,353. 242 Clark important parameter which helps to determine the physical properties of these complexes is the degree of asymmetry, defined to be y -x. Analysis of the wavenumbers and intensities of the intervalence charge-transfer bands of the mixed-valence Ru2I1.II1, versions of this type of complex (bands which are absent from the spectra of their Ru211*" and RU~",''~ congeners) indicates that the extent of valence-electron delocalization depends on the degree of asymmetry (it.on y -x) and on the basicity of the terminal ligands, L. Where y -x is zero, as for [(PE~~P~)~RUC~~RU(PE~~P~)~]~+,[(As to13)2ClRuC13RuCl(As to13)2] and also for the species [(NH3)3RuC13Ru(NH3)3I2+, (thought to be akin to the so-called 'Ruthenium B1ue')81*82 the intervalence band is intense and of low wavenumber (4350,5900, and 7 150cm-', respectively). Such species are regarded as highly delocalized. However, as the degree of asymmetry is increased, i.e. the environments of the two metal atoms become different, the complex ions become trapped-valence systems, as judged by the much lower intensities and higher wavenumbers (e.g.13 500 cm-l for C(PE~~P~)~RUC~~RUC~~(PE~~P~)~])of their intervalence bands. Extensive spectroelectrochemical studies of these and related systems are in hand. Another type of cluster in which the question of electron delocalization in mixed-valence species is important is that of the molybdenum hetero- and iso- polyanions based on the Lindqvist (Mo6019) and Keggin (XM012040) structures (X = Si, Ge, P, or As). The reduced species, the heteropolyblues (containing MoV as well as MoV1) have been studied by both e.s.r. and electronic spectroscopy in order to assess the degree of delocalization of the additional electron. All these species appear to be valence trapped on the e.s.r.timescale at sufficiently low temperatures, even [Mo6019I3- which is thus [MO~MO~'sO~~]~- at 77 K (it exhibits a six-line hyperfine structure at this temperat~re).~~ Owing to the very short timescale of u.v.-visible spectroscopy, the mobile electron appears trapped as assessed by this technique, even at room temperature. The reference materials are [MoVWV'5019I3- and a-[SiMoVWV'l 1040]'-, in which no homonuclear electron transfer process can take place. For some typical homonuclear systems, e.g. [PMo12040]~- and [GeMol 2040]5-, computer simulation of the e.s.r. spectrum over a wide temperature range has allowed the rate of thermal electron-transfer and the corresponding activation energy to be calculated. In particular, the broadening of e.s.r.spectra with increasing temperature indicates that thermal electron-delocalisation occurs when the paramagnetic Mos ion is surrounded by + Mo6+ ions; this is caused by electron hopping between sites. At room temperature it is not possible to detect an e.s.r.-distinguishable MoV site. Analysis of the e.s.r. spectra indicates that the valence electrons are more delocalized in the Keggin than in the Lindqvist structure, presumably due to the corner-sharing mode of junction which is specific to the former. " E. E. Mercer and L. W. Gray, J. Am. Chem. Soc., 1972,94,6426. 82 E. E. Mercer and P. E. Dumas, Inorg. Chem., 1971, 10, 2755. '3 C. Sanchez, J. Livage, J. P. Launay, M. Fournier, and Y. Jeannin, J. Am. Chem. Soc., 1982, 104, 3 194. 243 The Chemistry and Spectroscopy of Mixed-valence Complexes 6 Class IIIB Species Few Raman studies of mixed-valence complexes of this class have yet been attempted. The most interesting candidates in this context, KCP and K1.75Pt(CN)4.1.5H20, are difficult to study owing to their metallic reflectivity and to their tendency to dehydrate on being warmed.Although their Raman spectra have been obtained, these do not show any resonance effects.8L86 Partial oxidation of Magnus Green, [Pt(NH3)4][PtC14], yields a variety of products, some of which apparently involve directly metal-metal bonded species, have room temperature conductance of -1C2 i2-’ cm-’ , and a non-integral platinum oxidation state of 2.41.87 These materials are interesting but, as yet, poorly defined. One of the products of this oxidation is the photochromic salt [Pt(NH3)4][Pt(NH3)4C12][HS04]4, as established by resonance Raman 88 and X-ray 7 Conclusion The very diverse features of mixed-valence chemistry will undoubtedly continue to attract the interest of synthetic chemists, whether based in University or in Industry, for many years to come.” Certainly the search for materials with useful properties is a major challenge to chemists of all persuasions, i.e.not only to those seeking new materials but also to those interested in understanding the basis of, and inter-relationship between, physical properties.” The further application of physical techniques to the study and classification of mixed-valence compounds is a matter of considerable importance to inorganic chemists, geochemists, and bioinorganic chemists alike.Acknowledgement. The research embodied in this article has been supported by the University of London and by the Science and Engineering Research Council. I thank these two bodies as well as my collaborators (in particular Drs. T. J. Dines and M. Kurmoo and Mr. V. B. Croud) whose experiments and comments have contributed substantially to the basis for this lecture. 84 E. F. Steigmeier, R. London, G. Harbeke, H. Anderset, and G. Scheiber, Solid State Commun., 1975,17, 1447. 85 E. F. Steigmeier, D. Baeriswyl, G. Harbeke, H. Anderset, and G. Scheiber, Solid State Commun., 1976, 20, 661. 86 E. F. Steigmeier, D. Baeriswyl, H. Anderset, and J. M. Williams, in Lecture Notes in Physics, ed. S. BarisiC, A. BjeliS, J. R. Cooper, and B. Leontii-, Springer, 1979, p. 229.*’ J.-P. Catinat, T. Robert, and G. Offergeld, J. Chem. Soc., Chem. Commun., 1983, 1310. R. J. H. Clark and M. Kurmoo, J. Chem. Soc., Dalton Trans., 1982, 2515. 89 R. J. H. Clark, M. Kurmoo, A. M. R. Galas, and M. B. Hursthouse, J. Chem. SOC.,Dalton Trans., 1982, 2505. 90 P. E. Fanwick and J. L. Huckaby, Inorg. Chem., 1982,21,3067. 91 P. Day, Chem. Br., 1983, 306.

ISSN:0306-0012    

DOI:10.1039/CS9841300219

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

Chemicals from the Glands of Ants By Athula B. Attygalle" and E. David Morgan* DEPARTMENT OF CHEMISTRY, UNIVERSITY OF KEELE, STAFFORDSHIRE ST5 SBG 1 Introduction Ants are social insects, generally organized into large colonies where the numerous workers, infertile females, do most of the labours except procreation, which is reserved to the exiguous males and queens. All three castes use chemicals for communication, though as yet we know little about the substances specific to males and queens. The ants have inside their bodies a number of tiny glands in which they produce (or sequester) and store a curious variety of natural products. These distinctive and diminutive laboratories have been the subject of study of chemists and biologists in recent years.The substances studied generally have small mole- cules, sufficiently volatile to be studied by gas chromatography, and are presumed to have a pheromone function, though that has not been proven, or even tested in all cases. These glandular substances from ants are the subject of this review. The ants belong to the family Formicidae, which is divided into eleven subfamilies:' Ponerinae, a primitive group common in Australia; Myrmeciinae, the 'bull ants' of Australia; Dorylinae, the Old World army ants; Ecitoninae, the New World army ants; Leptanillinae, Pseudomyrmecinae, and Nothomyrmeciinae, three small groups; Myrmicinae, the largest group, to which our temperate red ants belong; Aneuretinae, a fossil group with only one known species today; Dolichoderinae,a chiefly tropical group; and Formicinae, those that produce formic acid, and to which our black wood ants belong.The exocrine glands (i.e.those secreting to the outside) of ants produce a range of substances for communication that fall under the general heading of 'semio-chemicals' (chemicals which convey information between organisms). Pheromones are only one type of semiochemical and transmit information between members of the same species. Semiochemicals used for interspecific communication are called allelochemics. Three main types of allelochemics are recognized; allomones, kairomones, and synomones. An allomone is a chemical emitted from an insect, which gives adaptive advantage to the odour-releasing individual (e.g.defensive and repellent secretions). In contrast, a kairomone gives an advantage to the receiver of the odour (e.g. substances that enable the predator to locate its prey). The chemical trails of the army ant, Neivamyrmex nigrescens are picked up by predators like the blind snake Leptotyphlops dulcis and the beetle Hellu-morphoides texan~s,~ which feed mainly on the brood of the ants. A synomone I R. R. Snelling, in 'Social Insects vol 11', ed. H. R. Hermann, Academic Press, New York, 1981, p. 369. J. F. Watkins, F. R. Gehlbach, and R. S. Baldridge, Southwest Nat., 1967, 12, 455. R. W. Plesk, J. C. Kroll, and J. F. Watkins, J. Kan. Entomol. SOC.,1969,42,452. 'Present address: Institut fur Organische Chemie, Universitat Erlangen-Nurnberg, D-8520 Erlangen, West Germany.245 Chemicals from the Glands of Ants P pPG’ LG Oe MthG C, \ Figure 1 Idealized section through a typical ant showing the location of the intestinal tract and the known exocrine glands. AG, anal-pygidial gland; C, crop; DG, Dufour gland; HG, hind gut; IS, infrabuccal sac; LG, labial gland; MG, mandibular gland; MthG, metathoracic or metapleural gland; MV, Malpighian vessels; MxG, maxillary gland; Oe, oesophagus; P, pharynx: PoG, poison gland; Po V, poison vesicle; pPG, postpharyngeal gland; PUG, Pavan gland; RG, rectal gland; S, stomach; SG, sternal gland; ST, sting lance; TG, tibia1 gland. (Freely adapted from K. Dumpert, ‘The Social Biology of Ants’, Pitman Advanced Publishing Programme, London, 1981 and Paul Parey Verlag, Berlin) benefits both the producer and recipient (e.g.floral scents that attract pollinating insects). The location of the major exocrine glands in a typical ant is illustrated in Figure 1. In the head region are the mandibular, propharyngeal (maxillary), and the post- pharyngeal glands. The mandibular glands are the major source of pheromones for a number of subfamilies of ants and many of these are described as ‘alarm’ pheromones. These alarm pheromones are released when ants are disturbed and generally increase the rate of movement and aggressiveness of the ants, although the overall behavioural responses produced are complex and not clearly defined. The labial glands associated with the digestive system open into the head region, but are actually situated in the thorax.The metapleural glands are also found in the thorax. Tibia1 glands in the hind legs, described in some species of Crematogaster, provide a unique source of trail pheromones4*5 which are used by an insect to mark a route so that other insects of the same community are able to follow it. The poison and Dufour glands are the main glands in the abdomen. These two glands generally provide defensive allomones and in some cases the trail pheromones and sex attractants. The hind gut is the main source of the formicine trail substances. The supra-anal glands and Pavan’s gland were once thought to be restricted to R. H. Leuthold, Psyche, 1968, 75,233. D. J. C. Fletcher and J. M. Brand, J. Insect Physiol., 1968, 14, 783.Attygalle and Morgan dolichoderine ants only, but Holldobler and Engel have recently reviewed tergal and sternal glands, and they believe that the anal glands are analogous to the tergal pygidial glands which are found in several sub-families. Definitive behavioural activities have been assigned to a number of compounds found in the exocrine glands but many others appear to have no significant behavioural activity. Closely related species are often found to produce similar substances in a particular gland, although the qualitative and quantitative compositions are usually characteristic of the species. Although it is not always easy to explain why these species-specific mixtures of compounds are present, nevertheless they are diagonostically useful, especially for distinguishing between morphologically similar species.Furthermore, the investigations on the composition of exocrine glands can be helpful phylogenetically. 2 Venom Apparatus It is thought that the original function of glands of the venom apparatus, which is attached to the sting lance, was the production of proteinaceous compounds that coated the eggs and allowed them to adhere to a substrate, and from this the venom of the sterile workers developed. The Dufour gland is a sac-like structure and is also attached to the sting. It is found in all Hymenoptera (bees, wasps, and ants) but its primary purpose is unknown. As its contents are oily, it has been suggested that it originally provided a lubricant for the eggs in the ovipositor.All the subfamilies of ants contain species that use venom to subdue their prey. However, in a number of species during phyletic development, an assortment of other functions have become assigned to the glands of the venom apparatus. These secondary functions include the production of communication pheromones, defensive allomones, and other deterrents. A. Poison Gland Substances.-The characteristic chemical components found in the venoms of stinging ants are proteinaceous and alkaloidal. The formicine ants are stingless but their venoms often contain aqueous formic acid in concentrations up to 60%.7The amount of formic acid can occasionally be as large as 2 mg per ant but an amount around 600 pg per ant is considered usual.Formic acid constitutes more than 99% of formicine venom and it is the only volatile compound reported. However, the presence of peptides and free amino-acids in the venoms of Formica polyctena and Camponotus pennsylvanicus * have also been reported. The major function of formic acid is to act as a defensive allomone against predators, although in some formicine ants it also acts as an alarm pherorn~ne.~ Hefetz and Blum have studied the biosynthesis of formic acid in the poison glands of formicine ants. Serine (1) is the major precursor and contributes both its B. Holldobler and H. Engel, Psyche, 1978, 85, 285.’M. F. H. Osman and J. Brander, 2.Nafurforsch., Teil B, 1961, 16, 749. H. R. Hermann and M. S. Blum, Psyche, 1968,75, 216.M. S. Blum and J. M. Brand, Am. Zool., 1972, 12, 553. A. Hefetz and M. S. Blum, Science, 1978, 201, 454. l1 A. Hefetz and M. S. Blum, Biochim. Biophys. Acta, 1978,543,484. Chemicals from the Glands of Ants 248 Attygalle and Morgan a-and P-carbons, but not its carboxyl carbon, to formic acid. The a-carbon of glycine can also be incorporated. The proposed biosynthetic pathway based on the studies of Hefetz and Blum is shown in Scheme 1. Serine (1) is converted into glycine (3) by donating its g-carbon to tetrahydrofolic acid (2). The N5,N10-methylene tetrahydrofolate (4) thus produced is oxidized to the methylidyne form (5) by NADP+. The product (5) is hydrolysed to 10-formyltetrahydrofolate(6), which is further hydrolysed to produce formic acid and regenerate tetrahydrofolic acid. Hefetz and Blum l1 also demonstrated that the enzymes catalysing the reactions of Scheme 1 were present in the poison gland in much higher concentrations that any other tissue that was examined.Proteinaceous venoms appear to be widely spread in the subfamilies, Myrmeciinae, Ponerinae, Dorylinae, Pseudomyrmecinae, and Myrmicinae. No information is available yet about the chemistry of venoms of other subfamilies. The presence of constituents with a wide range of pharmacological activities has been demonstrated in the venom of two species of Myrmecia.'2-'3 The venom of Pogonomyrmex badius l4 contains histamine (7) and a series of free amino-acids, enzymes of six classifications and a number of other non-enzymic proteins.Similar constituents have also been found in Myrmica ruginodis.' H (7) In contrast to the general proteinaceous themes exhibited by the venoms of many subfamilies of Formicidae, a number of species of ants in the Myrmicinae subfamily possess the ability to biosynthesize a variety of alkaloids in the venom gland. All the alkaloids reported in myrmicine venoms are summarized in Table 1. The members of the genus Solenopsis undoubtedly lead the myrmicine ants as alkaloid chemists, possessing the ability to produce alkaloid-rich and low-protein venoms. In Solenopsis geminata around 19 pg of alkaloids are present in the poison gland of a single worker ant.16 The ants belonging to the subgenus Solenopsis of the genus Solenopsis are referred to asfire ants because of the potency of their venoms, which exhibit pronounced necrotic," haemolytic,'* antibiotic," and toxic 2o properties.The venoms of Solenopsis (Solenopsis) species are characterized by a predominance of 2-alkyl-6-methylpiperidines (19, 20). The alkyl group usually contains an odd number of carbon atoms in the C, to C15 range and may l2 J. C. Wanstall and I. S. de la Lande, Toxicon, 1974, 12,649. l3 G. W. K. Cavill, P. L. Robertson, and F. B. Whitfield, Science, 1964, 146, 79. l4 J. 0.Schmidt and M. S. Blum, Science, 1978, 200, 1064. l5 J. Jentsch, Proc. Int. Congr. Int. Union Study Soc.Insects, 6th, 1969, p. 69. l6 A. B. Attygalle, Ph.D. thesis, University of Keele, 1983. l7 D.C. Buffkin and F. F. Russel, Toxicon, 1972, 10, 526. G. A. Adrouny, V. J. Derbes, and R. C. Jung, Science, 1959, 130,449. l9 D. P. Jouvanez, M. S. Blum, and J. G. MacConnell, Antimicrob. Agents Chemother., 1972,2, 291 2o J. R. Joyce, Vet. Med. Small Anim. Clin., 1983,78, 1107. Chemicals from the Glanh of Ants H H (331 (34) Scbeme 2 sometimes contain a double bond. Both cis- and trans-isomers of 2,6-disubstituted piperidines are usually present, with either trans-isomer predominating as found in S. invicta21,22or cis-isomer predominating as in S. xyloni or S. geminata.23 These ring configurational isomers are conveniently separated by g.l.c., with the cis- isomer eluting first on polar phases like Carbowax 20M. This may be because the nitrogen atom is less exposed when the 2,6-substituents are in di-equatorial positions as found in the cis-isomers.The double bonds, when present in the side chains, always appear to have a Z-c~nfiguration.~~ The absolute configurations of the chiral centres of these piperidines still remain unknown. The 2-alkyl-6- methylpiperidines (33) are well suited to be studied by gas chromatography-mass spectrometry, they all show the base peak or an intense ion at m/z 98 (34) due to the cleavage of the long alkyl group (Scheme 2). Baliah et al.24 have provided an excellent review on synthetic methods for 2,6-disubstituted piperidines. The venom of S. xyloni contains, besides the usual piperidines, a 2-alkyl-6-methyl-2,3,4,5-tetrahydropyridine (21).23 Many species of Solenopsis belonging to the subgenus Diplorhoptrum are called thief ants because they steal brood from the nests of other species of ants.The raiding thief ants secrete offensive alkaloidal substances which repel the host ants from defending their brood. Solenopsis (Diplorhoptrum) fugax utilizes 2-butyl-5- heptylpyrrolidine (13) in this context.25 This was demonstrated by applying the venom of S.fugax, synthetic pyrrolidine (13), or mineral oil on the brood of a few other species of ants and placing treated brood in the foraging areas of the respective ants. Ordinarily, if a larva is discovered by a worker it will immediately be transported back to the brood chamber. The worker ants picked up untreated or mineral oil treated larvae in a similar manner but clearly avoided picking up larvae contaminated with either the venom of S.fugax or synthetic pyrrolidine (13).25 In contrast to true fire ants (subgenus-Solenopsis) the thief ants are not noted for their stinging abilities. The subgenera, Diplorhoptrum and Euophthalma produce only very small quantities of alkaloids in their venoms. The venom of species of Diplorhoptrum and Euophthalma subgenera also contain 2-alkyl-6-methylpiperi- dines (19), but only trans-isomers were reported in contrast to both cis- and trans- isomers found in the ants of Solenopsis subgenus.26 Furthermore the novel N-’’ J. G. MacConnell, M. S. Blum, and H. M. Fales, Science, 1970, 163, 840. ”J. G. MacConnell, M. S. Blum, and H.M. Fales, Tetrahedron, 1971, 26, 1129. 23 J. M. Brand, M. S. Blum, H. M. Fales, and J. G. MacConnell, Toxicon, 1972, 10, 259. 24 V. Baliah, R. Jeyaraman, and L. Chandrasekaran, Chem. Rev., 1983,83, 397. 25 M. S. Blum, T. H. Jones, B. Holldobler, H. M. Fales, and T. Jaouni, Nafurwissenschaften,1980,67, 144. 26 T. H. Jones, M. S. Blum, and H. M. Fales, Tetrahedron, 1982,38, 1949. 250 Attygalle and Morgan methylpiperidines (23) reported from S.pergandei and S. carofinensis appear to be unique to thief ants of Diplorhoptrum subgenus.26 A unique mono-substituted piperideine (22) has also been reported from a Solenopsis (Diplorhoptrum) species.26 The South African species Solenopsis punctaticeps is more closely related to thief ants than fire ants.Although S.punctaticeps can sting, the reaction of humans to its venom is mild compared to that encountered with the sting of a true fire ant and its venom shows a marked difference from the fire ant venom by the absence of dialkylpiperidines-instead it is fortified with a number of 2,5-dialkylpyrrolines and -pyrrolidines (9, 10, and 13).27*28 The only bicyclic alkaloid known from Solenopsis is the pyrrolizidine (32). The four possible isomers have been synthesized and the stereochemistry of the ant isomer has been determined by compari~on.~’ A recent enantioselective synthesis of the pyrrolizidine (32) has been reported 30 and the product was spectroscopically congruent with the natural product. The alkaloidal venoms are not restricted only to the genus Solenopsis; many species of Monomorium also have an array of alkaloids in their venoms.Jones et aL3 examined the venoms of a number of species of Monomorium and found that all produce mixtures of different proportions of trans-2,5-dialkylpyrrolidines(13), tr~ns-2-alkyl-5-alkenylpyrrolidines(14), trans-2,5-dialkenylpyrrolidines (15), trans-2,5-dialkyl-N-methylpyrrolidines(16), trans-2-alkyl-5-alkenyl-N-methyl-pyrrolidines (1 7), trans-2,5-dialkenyl-N-methylpyrrolidines(18), 2,5-dialkyl-1-pyrrolines (9, lo), and 2,5-dialkenyl-l-pyrrolines(1 1, 12).26329 Although the purpose of these species-specific mixtures of alkaloids is not clear, at least they are useful for the chemotaxonomist since they provide a distinctive label to a species.The venom of the old world species M. pharaonis is particularly distinctive in containing, in addition to four dialkylpyrrolidines, two indolizidines (29, 31).26.32.33 The indolizidine (Monomorine I) (29), which has an all-cis configuration, and trans-2-pentyl-5-(5’-hexenyl)pyrrolidine(Monomorine 11) (14) were shown to attract Pharaoh’s ant, Monornorium pharaonis, and to have some activity in a trail-following bioa~say.~~.~~ However, the true trail pheromone was later identified from the Dufour gland.34 Methyl 4-methylpyrrole-2-carboxylate(8), a minor constituent identified in the ’’D. J. Pedder, H. M. Fales, T. Jaouni, M. S. Blum, J. G. MacConnell, R. M. Crewe, and M. Robin, Tetrahedron, 1976,32, 2275. J. H. Jones, M. S.Blum, and H. M. Fales, Tetrahedron Lett., 1979, 12, 1031. 29 T. H. Jones, M. S.Blum, H. M. Fales, and C. R. Thompson, J. Org. Chem., 1980,45,4778. 30 S. Takano, S.Otaki, and K. Ogasawara, J. Chem. Soc., Chem. Commun., 1983, 1172. 31 T. H. Jones, M. S. Blum, R. W. Howard, C. A. McDaniel, H. M.Fales, M. B. DuBois, and J. Torres, J. Chem. Ecol., 1982,8,285. 32 F. J. Ritter, I. E. M. Rotgans, E. Talman, P.E. J. Venviel, and F. Stein, Experientiu, 1973,2!3, 530. ”P. E. Sonnet and J. E. Oliver, J. Heterocycl. Chem., 1975, 12, 289. 34 F. J. Ritter, I. E. M. Bruggemann-Rotgans, P.E. J. Verwiel, E. Talman, F. Stein, J. La Brijn, and C. J. Persoons, Int. Congr. Int. Union Study Soc.Insects, 8th, 1977, p. 41 (Chem. Abstr. 1978,89, 176608f). 3s J. H. Tumlinson, R.M.Silverstein,J. C. Moser, R. G. Brownlee, and J. M. Ruth, Nature (London), 1971, 234,348. 36 J. H. Tumlinson, J. C. Moser, R. M. Silverstein, R. G. Brownlee, and J. M. Ruth,J. Insect Physiol., 1972, 18, 809. 37 R. G. Riley, R. M. Silverstein, B. Carroll, and R. Carroll, J. Insect Physiol., 1974,20,651. 251 Chemicals from the Glands of Ants venom of a few species of leaf-cutting ants, Atta texan~,~~,~~ andA. cephalote~,~~ Acromyrmex octospinosus3* is the first substance to be identified as a component of ant trail pheromones. This pyrrole (8) was synthesized by Sonnet 39 and shown to be identical with the natural substance. The compound has a very high behavioural efficiency and the detection threshold is as low as 80 fg cm-' of a trail.Artificial trails of the pyrrole (8) were followed by several leaf-cutting ants of the tribe Attini but two related species, Atta sexdens and Acromyrrnex niger, did not show any significant re~ponse.~' The pyrrole (8) is only one component in the trail pheromone of Atta texana; Tumlinson et al.35 isolated at least four other active fractions but the structures of these other constituents remain unknown. Moser and Silverstein 41 have shown the existence of an active but non-volatile component besides the volatile component in the trail-marking substance of A. texana. Initial laboratory experiments indicated that a practical method of ant control might result by incorporating the pyrrole (8) into toxic bait. In practice this made it easier for the ants to find the baits but did not increase the likelihood of the baits being picked up.42 In field tests, pyrrole (8) was unable to reproduce all aspects of natural recruitment although it did induce trail-following beha~iour.~~ Its addition to current baits would not be cost-effective 44 but it might be a worthwhile addition to the synthetic baits now being developed.45 An understanding of the stereochemistry necessary for a compound to exhibit pheromonal activity is aided by studies using structurally related compounds (sometimes called congeners).The studies of Sonnent and Mo~er~~ on the congeners of the pyrrole (8) show the absolute requirement of the 2,4-substitution pattern and the pyrrolic nitrogen atom. All the other ring-substitution isomers are inactive.N-Methylation also results in the reduction of activity.47 The methyl group at position-4 can be replaced by a chlorine atom without loss of activity, and even when substituted with an ethyl group or a bromine atom the pyrrole shows a substantial degree of activity, indicating that the steric requirement at position-4 is not very stringent. In contrast, for substitution at the 2-position, activity of the pyrrole is retained only when the new substituent group is less bulky than the original. For example, compounds with the carboxy-group esterified with any alcohol higher than methanol were inactive.46 An acetyl group at position 2 showed moderate activity but a free carboxy-group was inactive. Caputo et have calculated the charge densities on the pyrrolic nitrogen for a variety of congeners and show the most active compounds to have the same charge of -0.51 electrons on the nitrogen atom.They have suggested that a close value for this 38 J. H. Cross, J. R. West, R. M. Silverstein, A. R. Jutsum, and J. M. Cherrett, J. Chem. Ecol., 1982,8, 11 19. 39 P. E. Sonnet, J. Med. Chem., 1972, 15,97. 40 S. W. Robinson, J. C. Moser, M. S. Blum, and E. Amante, insectes SOC.,1974, 21, 87. 41 J. C. Moser and R. M. Silverstein, Nature (London), 1967, 215, 206. 42 S. W. Robinson and J. M. Cherrett, Bull. Entomol. Res., 1978, 68, 159. 43 S. W. Robinson, A. R. Jutsum, J. M. Cherrett, and R. J. Quinlan, Bull. Entomol. Res., 1982,72, 345. O4 K. Jaffe and P. E.Howse, Anim. Behav., 1979,27,930. 45 A. R. Jutsum and J. M. Cherrett, Bull. Entomol. Soc., 1981, 71, 607. 46 P. E. Sonnet and J. C. Moser, Agr. Food Chem., 1972,20, 1191. 4' P. E. Sonnet and J. C. Moser, Environ. Entomol., 1973, 2, 851; 1973,56,976. 48 J. F. Caputo, R. E. Caputo, and J. M. Brand, J. Chem. Ecol., 1979,5,273. 252 Attygalle and Morgan charge may be important in a compound with the correct steric properties to show chemorecognition. Atta sexdens, a leaf-cutting ant species related to A. texana and A. cephalotes did not follow an artificial trail made of the pyrrole (8). Subsequently Cross et al.49 identified 3-ethyl-2,5-dimethylpyrazine(24) from the poison glands of A tta sexdens as the major component of its trail pheromone.The pyrrole (8) was also isolated as a minor component but this substance does not evoke trail-following behaviour in Atta sexdens. Evershed et al.50951 identified the same pyrazine (24) as the sole component in the trail pheromone of Myrmica rubra and seven other related species of Myrmica. A quantitative study of the trail pheromone substances in the venom of Attine ants has been made by Evershed and Morgan.” Recently a further pyrazine, 2,5-dimethylpyrazine (26), together with 3-ethyl-2,5- dimethylpyrazine (24) have been identified as the trail pheromone components of Tetramorium caespit~m.~~*~~ A complete identification of a multi-component trail pheromone of ants has been performed only in this case. A 30:70 mixture of 3- ethyl-2,5-dimethylpyrazine (24) and 2,5-dimethylpyrazine (26) constitutes the synergistic mixture that evokes the highest trail-following activity.A concentration of 40 and 90 pg cm-’ trail of the two respective synthetic pyrazines (24, 26) was equivalent in releasing trail-following activity to an artificial trail made of a single poison gland. The ability to release trail-following activity by a number of congeneric pyrazines has been tested on Tetramoriwn cae~pitum.’~ The 2,5-substitution on the pyrazine ring is important because 2,3- and 2,6-dimethylpyrazines are inactive. Furthermore, when the ethyl of pyrazine (24) was replaced by a methyl group, the compound could still show some activity. Although many trail pheromones that originate from the poison glands ’’are not yet chemically identified, they might be expected to be nitrogenous compounds of a similar nature.Skatole (27) and anabaseine (28) are two further alkaloids isolated from two myrmicine ants, Pheidole fallax 56 and Aphaenogaster fulva respectively. Anabaseine, a minor alkaloid in tobacco, acts as a weak attractant to A. fulva, but induces no trail- following beha~iour.’~ In 1980, Vander Meer et al.’* identified the poison sac of queens of imported fire ant, Solenopsis invicta, as the storage site for the pheromone which enables the workers to recognize the queen. This pheromone orients and attracts the worker ants towards the queen. Using a bioassay method, Vander Meer et al. found that 49 J. H. Cross, R. C. Byler, U.Ravid, R.M. Silverstein, S. W. Robinson, P. M. Baker, J. S. De Oliveira, A. R. Jutsum, and J. M. Cherrett, J. Chem. Ecol., 1979, 5, 187. R. P. Evershed, E. D. Morgan, and M. C. Cammaerts, Naturwissenschaften, 1981,67, 374. 51 R. P. Evershed, E. D. Morgan, and M. C. Cammaerts, Insect Biochem., 1982, 12, 383. 52 R. P. Evershed and E. D. Morgan, Insect Biochem., 1983, 13,469. ’’A. B. Attygalle and E. D. Morgan, Naturwissenschafen, 1983,70, 364. “A. B. Attygalle and E. D. Morgan, J. Chem. Ecol., 1984, 10, 1453.’’ A. B. Attygalle and E. D. Morgan, Adv. Insect Physiol., 1984, 18, in press. 56 J. H. Law, E. 0.Wilson, and J. A. McCloskey, Science, 1965, 149, 544.’’J. W. Wheeler, 0.Olubajo, C. B. Storm, and R. M. Duffield, Science, 1981,211, 1051.R. K. Vander Meer, B. M. Glancey, C. S. Lofgren, A. Glover, J. H. Tumlinson, and J. Rocca, Ann. Entomol. SOC.Am., 1980,73, 609. Chemicals from the Glands of Ants 0 (35) (361 0 (37) (38) the characteristic 2-methyl-6-alkyl (or alkeny1)-piperidine alkaloids (19,20) found in fire ant venom to be inactive as the queen recognition pheromones. These piperidines are found also in unmated (alate) queens but the alate queens are not overtly attractive to workers. The queen recognition pheromone is found only in the poison sac of mated queens. When the poison sacs were solvent extracted only the non-alkaloidal fraction was active, and therefore the pheromone was expected to be a minor non-alkaloid constituent in the poison sac. Rocca et aLS9 isolated 5-25 pg of the compounds responsible for activity from 18 OOO fire ant queens.Three components have been chemically identified: (E)-6- (1-pentenyl)-2H-pyran-2-one (39, tetrahydro-3,5-dimethyl-6-( 1-methylbutyl)-2H-pyran-2-one (36), and dihydroactinidiolide (37). The a-pyrone (39, the (&)-&-lactone (36) [together with its 3-epimer (38)J and the dihydroactinidiolide (37) have all been synthesized.” and the first two have been shown to be biologically active. The configurations at chiral atoms were assigned by comparing the i.r. and ‘H n.m.r. spectra with compounds of known configurations. The optical isomeric composition of the natural &lactone (36) is not yet determined. The behaviour of S. invicta workers to the queen recognition pheromone has been studied by Lofgren et a1.62 This pheromone has the potential of being used in fire ant control because the worker ants move inanimate objects treated with pheromone (e.g.pieces of rubber ca.20 mg, ‘surrogate queens’) into their nests. It could be useful to increase the rate of toxic bait pick-up. The venom of the myrmicine ant Myrmicaria natalensis is distinctively aberrant and unusual. The reported 63 presence of monoterpene hydrocarbons, such as a-pinene, camphene, P-pinene, sabinene, P-myrcene, a-phellandrene, a-terpinene, 59 J. R. Rocca, J. H. Tumlinson, B. M. Glancey, and C. S. Lofgren, Tetrahedron Lett., 1983, 24, 1889 and 1892. 6o J. R. Rocca, J. H. Tumlinson, B. M. Glancey, and C. S. Lofgren, Tetrahedron Left., 1983, 24, 1893.61 T. Sakan, S. Isoe, and S. B. Hyeon, Tetrahedron Lett., 1967, 1623. 62 C. S. Lofgren, B. M. Glancey,A.Glover,J. Rocca, and J. Tumlinson, Ann. Entomol. SOC.Am., 1983,76,44. 63 J. M. Brand, M. S. Blum. H. A. Lloyd, and J. C. Fletcher, Ann. Entomol. SOC.Am., 1974,67, 525. Attygalle and Morgan limonene, and terpinolene, in the poison gland of M. natalensis may puzzle anyone interested in the phylogeny of the venom gland or the physiological action of these compounds. A similar mixture of monoterpenes is found in the poison gland secretions of Myrmicaria eumenoides, an opportunistic termite predator in west Africa.64 When the synthetic terpenes were presented experimentally, a-pinene, p-pinene, myrcene, and limonene released ‘alarm behaviour’.The very highly volatile constituents present in nanogram quantities in the poison glands of Myrmica rubra and M. scabrinodis were reported to be simple alkanes, alcohols, and carbonyl compounds containing one to five carbon atoms.65 All reported evidence given so far clearly shows the diversity of ant venoms and probably the venom gland is the most versatile biosynthetic tissue that has been evolved by social hymenopterans. Table 1 Venom alkaloidrs of Myrmicine ants Structure Source Ref: Atta texana 35, 36 MeOK* A. cephalotes 37 OMe Acromyrmex octospinosus 38, 40 1ti Solenopsis punctaticeps 27 S. punctaticeps 27 m = 3,n = 6 Monomorium latinode 26 (9) S. punctaticeps 27 S. punctaticeps 27 Monomorium sp.31 Monomorium sp. 31 (12) 64 P. E. Howse, R. Baker, and D. A. Evans, ref. 34, p. 44. 65 M.C.Cammaerts,M. R. Inwood, E. D.Morgan,K. Parry, and R.C.Tyler,J.Insect. Physiol., 1978,24,207. 255 Chemicals from the Glands of Ants Structure Source Ref: m = 1, n = 6 Solenopsis punctaticeps 27 Me m = 3, n = 4 S. punctaticeps 27..(CH2 m = 3. n = 4 Monomorium sp. (4,26 H m = 3, n = 6 S. fugax 25 m = 3, n = 6 S. punctaticeps 27 (13) m = 3, n = 6 M. latinode 26 m = 4, n = 5 S. molesta 28 m = 4, n = 5 S. texanas 28 m = 5, n = 8 Monomorium sp. 31 n = 4 Monomorium pharaonis (a) n = 6 Monomorium sp. 26 ICH2=CH- (CH,), Q**(cH2)n-Me n = 8 Monomorium sp. 31 H Monomorium sp. 31 CH2=CH -(CH2), '*(CH2)7 -CH=CH2 I H M.latinode 26 Me-( CH2) 00**(CH,),-MeI Me (16) Monomorium sp. 26 (CH,), 0CH,=CH-* 4CH21 -Me I Me (17) Monomorium sp. 26 256 Attygalle and Morgan Structure Me (CH, I,, -Me I H (19) n=6 n=8 n=8 n=8 n = 10 n = 10 n = 12 n = 14 Source S. richteri S. carolinensis Solenopsis sp. S. richteri S. littoralis Solenopsis sp. S. littoralis S. invicta Me I H (CH2),,-CH=CH-(CH2),-Me (20) n=3 n=5 n = 7 Solenopsis sp. S. invicta Solenopsis sp. S. xyloni S. (Diplorhoptrum)? H2H= n n=8 n = 10 S. carolinensis S. pergandei Me (23) Acromyrmex octospinosus Atta sexdens Myrmica sp. Tetramorium caespitum ReJ (426 22 (b), 23 26 22 26 22,23 22,23 22,23 22, 23 23 26 26 26 38 49 50, 51 53 257 Chemicals from the Glandr of Ants Structure Source Ref: Acromyrmex octospinosus 38 Tetramorium caespitum Pheidole fallax aMeI H (27) Aphaenogaster fuloa N’ m = 3 M.pharaonis (29) 32,34 m = 5 Solenopsis sp.(30) 26,29c;3M~ (CHtIm-Me (29 1 (30) M.pharaonis 26 A ttygalle and Morgan Structure Source Re$ H Solenopsissp. 29GQMe (CH,),-Me -. (32 F. J. Ritter, I. E. M. Bruggemann-Rotgans, E. Verkuil, C. J. Persoons, in 'Pheromone and Defensive Secretion in Social Insects' ed. Ch. Noirot, P. E. House, and G. Le Masne, University of Dijon, Dijon, 1975. *J. E. MacConnell, R. N. Williams, J. M. Brand, and M. S. Blum, Ann. Ent. SOC.Am., 1974, 67, 134. 'J. H. Law, E. 0.Wilson, and J.A. McCloskey, Science, 1965,149, 544. B. Dufour Gland Substances.-In 1841, Dufour first described this sac-like gland attached to the poison apparatus in ants and bees.66 The Dufour gland of ants has a remarkable ability to synthesize hydrocarbons. Over 50 alkanes and alkenes have been identified in the Dufour gland and the compounds reported up to 1974 have been reviewed by Blum and Her~nann.~~-~~ Aliphatic hydrocarbons within the range C, to C27are present in the Dufour glands of myrrne~iine,~' ~onerine,~' pse~domyrmecine,~myrmi~ine,~~formi-cine,67 and dolichoderine68 ants. The Dufour glands of ants are typically filled with linear hydrocarbons, but not exclusively so. Although hydrocarbons with an even number of carbon atoms are encountered as minor constituents, the odd- numbered hydrocarbons are always found in much larger quantities.Branched- chain hydrocarbons and many oxygenated compounds can also be encountered. The formicine ants produce hydrocarbons, often undecane and tridecane, as the major class of compounds in their Dufour glands. Some species of Formica and Camponotus contain monomethylalkanes as minor constituents; for example, in Camponotus intrepidus, 3-methylalkanes comprise about 16%and 5-methylalkanes about 2% of the total hydrocarbon^.^' Over 95% of the total secretion of volatiles of C.japonicus and C. obscuripes is ~ndecane.~~ Similarly, for Formica nigri~ans,~~ F. r~fa,~~ and Acanthomyops clauiger 75 undecane accounts for F. p~lyctena,~~,~~ more than 50% of the Dufour gland secretion.In Formica polyctena a number of monomethylalkanes are present in small quantities, with the methyl branching at position 2,3,4,5, or 7.74 The Dufour glands of virgin queens of F.polyctena contain about 74% undecane, and this is also the major component of the worker glands. However, in the glands of mated queens undecane content is less than 1%, indicating that undecane may function as a pheromone for some behaviour related to swarming or pairing. Males did not, however, respond to streams of air laden 66 L. Dufour, Mem. Pres. div. Sav. Acad. Sci. Inst. Fr., 1841, 7, 265. "M. S. Blum and H. R. Hermann, in 'Arthropod Venoms', ed. S. Bettini, Handbook of Experimental Pharmacology, Vol. 48, Springer- Verlag, Berlin, 1978, p.801. M. S. Blum and H. R. Hermann, in 'Arthropod Venoms', ed. S. Bettini, Handbook of Experimental Pharmacology, Vol. 48, Springer-Verlag, Berlin, 1978, p. 871. 69 M. S. Blum, 'Chemical Defences of Arthropods', Academic Press, New York, 1981. "G. W. K. Cavill and P. J. Williams, J. Insect Physiol., 1967, 13, 1097. 71 J. J. Brophy, G. W. K. Cavill, and J. S. Shannon, J. insect Physiol., 1973, 19, 791. 72 N. Hayashi and H. Komae, Biochem. Syst. Ecol., 1980,8,293. 73 G, Bergstrom and J. Lofqvist, J. Insect. Physiol., 1973, 19, 877. 74 J. Lofqvist and G. Bergstrom, J. Chem. Ecol., 1980, 6, 309. 75 F. E. Regnier and E. 0.Wilson, J. Insect Pbysiol., 1968, 14, 955. 259 Chemicals from the Glands of Ants with undecane alone or mixed with formic acid when tested in a climate chamber with high light intensity.74 In a number of species of Lasius also, the major component is ~ndecane.~~ The composition of the glandular secretion of Polyrhachis simplex is very simple and consists of linear alkanes-over 90% tridecane and minor amounts of undecane, dodecane, pentadecane, and heptade~ane.~~ Similarly, tridecane is the major component in the gland of an Australian Polyrhachis species examined by Brophy et al.78 Undecane comprises nearly 50% of the total volatile secretions of P.lamellidens. Besides the alkanes, a number of alkenes are also found in the Dufour glands of formicine ants. Linear monoenes of Cll---C23 range are present in Formica nigricens, F. rufa, and F.p~lyctena.~~It is interesting to find that the pentadecene in an Australian species of Polyrhachis is a mixture of A-6 and A-7 isomers. Similarly, heptadecene is a mixture of A-7 and A-8 isomers but 9-nonadecene, 9-heneicosene, 9-tricosene, and 9-pentacosene were not accompanied by positional isomers.78 Similar mixtures of alkanes and alkenes of Cl0-Cl8 range are found in the Dufour glands of a number of species of Camponotus, Cataglyphis, and Polyrhachis from Israel,79 but in Camponotus intrepidus the alkenes constitute less than 0.2% of the total hydrocarbons 71 and are completely absent from C.japonicus and C. ob~curipes.~~ The formicine ants are notable for their ability to produce a variety of oxygenated compounds together with the hydrocarbons in their Dufour glands.These compounds, produced by a number of species of Formica," La~ius,~~ Camponotus,81,82Notoncu~,~~ include a variety of primary and Gigantiop~,~~ aliphatic alcohols (Cl0-Cl6), simple ketones (C, 3-C19), alkylacetates (C90Ac- C, ,OAc), and a few terpenoid derivatives like or-farnesene,80 farnesyl all-trans-geranylgeraniol (39),73 and geranylgeranyl a~etate.~ The Dufour gland of Formica sanguinea has alkyl acetates as the major components8' (the average amount of decyl acetate is about 100 pg per ant)." Graham et al.84have studied decyl acetate biosynthesis in F. schaufussi to determine whether the ester is formed by the incorporation of molecular oxygen into a 2-ketone or by the esterification of acetic acid with an alcohol.The species F. schaufussi was specially suited because it has several hundred micrograms of decyl acetate in the Dufour gland. The radio- labelling studies of Graham et al. indicate that decyl acetate is synthesized via an esterification reaction.84 Methyl or ethyl ketones are encountered frequently in formicine Dufour glands, e.g. 2-tridecanone is a major constituent in Gigantiops destructor 83 and 76 G. Bergstrom and J. Lofqvist, J. Insect Physiol., 1970, 16, 2353. 77 A. Hefetz and H. A. Lloyd, J. Chem. Ecol., 1982,8, 635. 78 J. J. Brophy,G.W. K. Cavill, J. A. McDonald, D.Nelson,and W. D. Plant, Insect Biochem., 1982,12,215. 79 A. Hefetz and T. Orion, Isr. J. Entomof., 1982, 16, 87. G. Bergstrom and J. Lofqvist, J.Insect Physiol., 1968, 14, 995. G. Bergstrom and J. Lofqvist, Entomol. Scand., 1972, 3, 225. G. Bergstrom and J. Lofqvist, in 'Chemical Releasers in Insects', ed. A. S. Tahori, Gordon and Breach, New York, 1971, Vol. 111, p. 195. 83 M. S. Blum, T. H. Jones, W. I. Overal, H. M. Fales, J. 0.Schmidt, and N. A. Blum, Comp. Biochem. Physiol., 1983, 75B,15. 84 R. A. Graham, J. M. Brand, and A. J. Markovetz, Insect Biochem., 1979, 9, 331. 260 Attygalle and Morgan Acanthomyops claviger 75 and it is also found in Formica rufibarbis." A series of 2- and 3-alkylketones of C13--C19 range is found in Lasius ants.76 The Dufour gland content of L. fravus is unique in containing some hydroxy acids and the corresponding lac tone^.^^ 4-Hydroxyoctadec-9-enolide(40) and its free acid, and in minor amounts, the lower homologue with two carbon atoms less are found in the gland of L.frav~s.~~ OH (40) Among the myrmicine ants the Dufour gland contents of a number of species of the genus Myrrnica have been thoroughly in~estigated.~ '-"The presence of highly volatile oxygenated compounds such as simple alcohols, aldehydes, and ketones in the CIA4 range have also been reported in myrmicine ants of the genus M~rmica.~~Beside the general linear hydrocarbon theme, most species of Myrmica also have some terpenoid hydrocarbons, sometimes even as major components.88 These terpenoid hydrocarbons were identified by Morgan and Wadhams, on mass spectral evidence, as farnesene, homofarnesene, and bi~homofarnesene.~'Subsequently, a trishomofarnesene has also been described from M.scabrinodis.88 Parry 92 identified the farnesene isomer from the Myrrnica ants as (2,E)-a-farnesene (41) by comparison of its mass spectrum and g.1.c. retention times on different phases with those of a mixture of six farnesene isomers prepared from the dehydration of (2)-and (E)-nerolidol(42). The structure of (2, E)-a-farnesene was recently confirmed by total ~ynthesis.~~ On the basis of their mass spectra, structures (43) and (44)have been proposed by Morgan and Wadhams '' for the homofarnesene and bishomofarnesene isolated from the Dufour gland of Myrmica ants. A recent micro-degradation study provided the confirmatory evidence for the structure of homofarnesene and bishom~farnesene.~~ One of the products from the micro-ozonolysis of farnesene was 4-oxopentanal, whereas the homofarnesene and bishomofarnesene isomers from Myrrnica ants gave 4-oxopentanal indicating the A.B. Attygalle, M. C. Cammaerts, and E. D. Morgan, J. Insect Physiol., 1983, 29, 27. a6 A. B. Attygalle, R. P. Evershed, E. D. Morgan, and M. C. Cammaerts, Insect Biochem., 1983, 13, 507. M. C. Cammaerts, R. P. Evershed, and E. D. Morgan, J. Insect Physiol., 1981,27, 59. *' E. D. Morgan, K. Parry, and R. C. Tyler, Insect Biochem., 1979,9, 117. 89 E. D. Morgan, R. C. Tyler, and M. C. Cammaerts, J. Insect Physiol., 1977, 23, 511. 90 M. C. Cammaerts-Tricot, E. D. Morgan, R.C. Tyler, and J. C. Braekman, J. Insect Physiol., 1976,22,927.91 E. D. Morgan and L. J. Wadhams, J. Insect. Physiol., 1972, 18, 1125. 92 K. Parry, M.Sc. Thesis, University of Keele, 1978. "L. J. Thompson, Ph.D. Thesis, University of Keele, 1982. 94 A. B. Attygalle and E. D. Morgan, J. Chem. SOC.Perkin Trans I, 1982,949. 26 1 Chemicals from the Glands of Ants p(411 (421 R’ R2 R3 (43) R’=@=Me R2=Et (44)R1=R2=Et R3= Me presence of an ethyl group at C-7. Similarly, farnesene and homofarnesene gave propanone but bishomofarnesene gave butanone, indicating the presence of an ethyl group at C-11 of the bish~mofarnesene.~~ The configurations of the double bonds of these two isomers are still not determined. A similar homofarnesene isomer has been reported from four species of attine ants95,96 and a-farnesene from Aphaenogaster longicep.~.~~ Hydrocarbon themes, similar to those in Myrmica ants, are shown by the Dufour glands of Pogon~myrmex~~and No~omesser.~~ Dufour gland secretions of dolichoderine ants also show the linear and methyl- branched alkane and alkene pattern,lOO*lol but no terpenoids have been found.Some of the low molecular weight hydrocarbons (out of the large number reported) from the total extracts of a few species of Iridomyrmex may have arisen from the Dufour gland.lo2 No information is available about the chemistry of the Dufour glands of other subfamilies. The primary functions of the Dufour gland appear to be defence and communication. Blum has suggested that the diverse chemical compounds found in the Dufour gland could function to overstimulate the olfactory receptors of predators, and thus act as a deterrent.According to Bergstrom and Lofqvist,82 these compounds are often used as alarm pheromones-as, for example, undecane in Lasius niger 76 and Acanthomyops ~lauiger.’~ Tridecane however, which is almost the only substance in the Dufour gland of Polyrhachis simplex, did not provoke any 95 R. P. Evershed and E. D. Morgan, insect Biochem., 1980,10,81. 96 R. P. Evershed and E. D. Morgan, Insect Biochem., 1981,11,343. 97 G. W. K. Cavill, P. J. Williams, and F. B. Whitfield, Tetrahedron Lett., 1967, 23,2201. 98 F. E. Regnier, M. Nieh, and B. Holldobler, J. Insect Physiol., 1973, 19,981. 99 K. Vick,W. A.Drew, D. J. McGurk, E. J. Eisenbaum, and G.R. Waller, Ann. Entomol. SOC.Am., 1969, 62, 723. loo G. W. K. Cavill and E. Houghton, J. Insect Physiof., 1974,M, 2049. G. W. K. Cavill and E. Houghton, Aust. J. Chem., 1973,26, 1131. J. J. Brophy,G. W. K. Cavill, N. W. Davies,T. D. Gilbert, R. P. Philp, and W. D. Plant, Insecr Biochem., 1983, 13, 38 1. M. S. Blum, Bull. Entomol. SOC.Am., 1974,20, 30. Attygalle and Morgan sustained alarm behaviour.” Bradshaw et allo4 have demonstrated that in the African weaver ant Oecuphylla longinoda, undecane from the Dufour gland and formic acid from the poison gland act synergistically to release a ‘mass attack’ reaction. Similarly, in Formica rufa, some of the hydrocarbons of the Dufour gland as well as the formic acid act as alarm pheromones and the combination of both releases a more intense alarm behavio~r.’~’ The Dufour glands of slave-keeping formicine ants such as Formica subintegra, F.pergandei, and F. sanguinea produce large quantities of C,0-C,4 acetates, which are sprayed during slave raids to excite and attract the slave-maker ants but panic and disperse the slave-defender species.80.’06 The role of Dufour gland hydrocarbons in slave-keeping ants has been demonstrated by Lofqvist.’07 Formic acid from the Dufour gland is the main defence substance in F. rufu and F. sanguinea. However, it is rather harmless by itself because it is hydrophilic and when sprayed on a lipophilic cuticle it forms droplets which can affect only a small area. The hydrocarbons and acetates from the Dufour gland are lipophilic and promote the spreading of formic acid. The experiments of Lofqvist have shown the higher toxicity of formic acid and hydrocarbon or acetate mixtures compared with formic acid alone.Cammaerts et al.87have demonstrated that the ants of the genus Myrmica use their Dufour gland contents as a home-range marking pheromone. The ants move rapidly over any H (45) (49) lo4 J. M. S. Bradshaw, R. Baker, and P. E. Howse, Physiol. Enlomol., 1979, 4, 39. lo’ J. Lofqvist, J. Insect Physiol., 1976, 22, 1331. lo6 F. E. Regnier and E. 0.Wilson, Science, 1971, 172,267. lo’ J. Lofqvist, Oikos, 1977, 28, 137. Chemicals from the Glands of Ants area marked with their own Dufour gland secretions, but more slowly on an alien- marked territory until they have over-marked it with their own Dufour gland contents.85 The evidence so far available suggests that each species of ant has its own characteristic mixture of hydrocarbons and other compounds and that, at least in the Myrmica species which have been studied in detail, they are able to recognize this mixture and distinguish it from other specie^.^' In addition to the roles discussed above, some ant species utilize trace components of the Dufour gland secretion as trail pheromones.The few compounds identified as such include one sesquiterpenoid aldehyde, faranal (45),'08 and a few sesquiterpenoid alkenes (41, 46-49).'09.' lo Although monomorine I (29) and monomorine I11 (14) from the poison glands of Pharaoh's ant, Monomorium pharaonis show some activity in trail following test~,~~.~~ the true trail pheromone originates from the Dufour gland and was identified as (+)-(3S,4R)-3,4,7,11 -tetramethyltrideca-6& 1 OZ-dienal (faranal) (45).34,'O8 The stereochemistry and geometry of faranal have been established by stereospecific synthesis by several groups.' ''-''' Faranal has an interesting structural relationship to juvenile hormones and the farnesene homologue found in Myrmica species.94 In studies of faranal, Kobayashi et al.' ''found that the 3-epimer (3R,4R) is also weakly active when tested separately, though the preference of the ants for the (3S,4R)-optical isomer is unambiguous when tested in a choice test.The (3R)- enantiomer does not interfere with the activity of (3s) since ants follow a trail made of a mixture.The importance of the (4R)-configuration for activity has been shown by Koyama et a1.'18 According to their results the geometry of the C-10 double bond is unimportant; both the cis-and trans-isomers show similar activity. Furthermore, the substitution at C-11 is not so important for the manifestation of trail-releasing activity-an ethyl or a methyl group at C-1 1 has activity, although the 7-methyl cannot be replaced by an ethyl group without losing the activity."8 The trail pheromone from the Dufour gland of the fire ant Solenopsis invicta is definitely multi-component but there is some ambiguity about its composition. According to Williams et al.'".' l9 the major component is (2Z,42,62)-3,7,11- trimethyl-2,4,6,1O-dodecatetraene(Z,Z,Z-allofarnesene) (46). They synthesized the eight isomers of allofarnesene 'l9 and found the Z,Z,Z-isomer to be congruent '08 F.J. Ritter, I. E. M. Bruggemann-Rotgans, P. E. J. Verviel, C.J. Persoons, and E. Talman, Tetrahedron Lett., 1977, 2617. Io9 H. J. Williams, M. R. Strand, and S. B. Vinson, Experientia, 1981, 37, 1159. '' R. K. Vander Meer, F. D. Willams, and C. S. Lofgren, Tetrahedron Lett., 198 1, 165 1. ''I M. Kobayashi, T. Koyama, K. Ogura, S. Seto, F. J. Ritter, and I. E. M. Bruggemann-Rotgans, J. Am. Chem. Soc., 1980,102, 6602. 'I2 K. Mori and H. Ueda, Tetrahedron Lett., 1981,22,461. 'I3 K. Mori and H. Ueda, Tetrahedron, 1982,38, 1227. 'I4 D.W. Knight and B. Ojhara, Tetrahedron Lett., 1981,22, 5101. 'I5 D. W. Knight and B. Ojhara, J. Chem. SOC.,Perkin Trans. I, 1983,955. 'I6 R. Baker, D. C. Billington, and N. Ekanayake, J. Chem. SOC..Chem. Commun., 1981, 1234. R. Baker, D. C. Billington, and N. Ekanayake, J. Chem. SOC..Perkin Trans I, 1983, 1387. T. Koyama, M. Matsubara, K. Ogura, I. E. M. Bruggemann, and A. Vrielink, Naturwissenschaften, 1983, 70, 469. 'I9 H. J. Williams, M. R. Strand, and S. B. Vinson, Tetrahedron, 1981,37, 2763. Attygalle and Morgan with the ant substance. However, all isomers with 2-4 configuration showed trail- following activity. According to Vander Meer et ~1."~four of the components of the trail pheromone from the Dufour gland of S. invicta are (32,6@-3,7,1 l-trimethyl- dodeca- 1,3,6,1O-tetraene (2,E-a-famesene) (4 l), (3E,6E)-3,7,11 -trimethyldodeca- 1,3,6,1O-tetraene (E,E-a-farnesene) (47), (32,62)-3,4,7,1 l-tetramethyldodeca- 1,3,6,10-tetraene (2,Z-homofarnesene) (48), and (3Z,62)-3,4,7,1l-tetramethyl-dodeca-1,3,6,10-tetraene (2,E-homofarnesene) (49).The two farnesenes were obtained by dehydrating (E)-nerolidol (42) and showed activity at a pheromonal level. Z,E-a-Farnesene (41) was the most active component and shows activity even at 100 fg cm-' trail. The other three components were 1&100 times less active.'20 The homofarnesenes have not yet been synthesized. It is interesting to note that four components (41, 4749) were able to duplicate the recruitment response of a Dufour gland when heptadecane (another component of the Dufour gland) was added to the mixture; heptadecane itself is inactive.Barlin et all2' have made a preliminary survey of the trail pheromones of other species of Solenopsis. They report the main component of S. richteri to have a molecular weight of 218 and a molecular formula of C16H26. They assume that the trail pheromones of S. xyloni and S. geminata are similar and possess a molecular formula CI7H28. It is possible to conclude that the Dufour gland has evolved a long way from its suggested original function of providing a lubricant for the sting or for eggs during oviposition,'22 and that it has assumed novel duties acting as an important social organ to carry out a number of functions in defence and communication.3 Mandibular Gland Substances Mandibular glands are found in most insects and probably in all ants. Apart from the variation in size (in Camponotus it is remarkably large and extends up to the abdomen 123), the mandibular glands appear to be similar in anatomy in all ant species. The mandibular gland secretions have both defensive and pheromonal functions and a variety of natural products have been identified in the secretion^.^^-^^^"^ Different subfamilies of ants appear to biosynthesize different classes of compounds. Duffield has made a comparative study of the mandibular gland chemistry of formicine and ponerine ants.'25 A. Su1phides.-Some ponerine ants have the ability to produce alkyl sulphides in their mandibular glands, e.g.Paltothyreus tarsatus secretes dimethyl disulphide and dimethyl trisulphide, both of which release alarm beha~iour.'~~''~' Ants of the 120 R. K. Vander Meer, FI. Entomol., 1983,66, 139. 12' M. R. Barlin, M. S. Blum, and J. M. Brand, J. Insect Physiol., 1976, 22, 839. 122 R. L. Robertson, Aust. J. Zoo[., 1968, 16, 133. 123 U. Maschwitz and E. Maschwitz, Oecologia, 1974, 14, 289. lZ4 K. Parry and E. D. Morgan, Physiol. Entomol., 1979,4, 161. 12' R. M. Duffield, Diss. Abs. Int. B, 1977,37, 3761. 126 G. Casnati, A. Ricca, and M. Pavan, Chim. hd. (Milan), 1967,49, 57. "'R. M. Crewe and D. J. C. Flecher, J. Eniomol. SOC. S. Afr., 1974,37, 291. Chemicals from the Glanak of Ants species Megaponera foetens use dimethyl disulphide and dimethyl trisulphide to co- ordinate attacks on their termite prey.Scout ants, on finding a nest of termites, release these alkyl sulphides from their mandibular glands to attract sister workers who dig into the termite galleries in response to other unidentified pheromones from the mandibular glands.' 28 Benzylmethyl sulphide has also been identified but appears to show no behavioural activity. Biosynthetic studies have shown that the thiomethyl group of methionine is incorporated into these alkyl sulphides.' 29,1 30 B. Pyrazines-Alkylpyrazines have been detected in the mandibular glands of several species of ponerine and dolichoderine ants. Recently pyrazines have also been identified in a few species of formicine and one species of myrmicine ants.Table 2 lists the alkylpyrazines reported from the mandibular glands of ants. All the pyrazines so far identified from the mandibular glands of ants are trisubstituted. Except for the unique trialkylpyrazine (52) recently reported from the myrmicine ant, Aphaenogaster rudi~,'~' all the other pyrazines show 2,5-dimethyl-3-alkyl (50) or 2,6-dimethyl-3-alkyl (51) substitution patterns. Gas chromatography combined with mass spectrometry is an ideal technique for the detection and identification of the pyrazines because the mass spectra of the pyrazines are well characterized. The retention times on gas chromatography and the molecular ion from mass spectrometry are useful for determining the length of the alkyl chain. Brophy and Cavill '32 and Wheeler et all3' provide useful compilations of mass spectral data of pyrazines.The alkylpyrazines from mandibular glands have been reported to act as alarm pheromones.' 33-' 35 In Odontomachus troglodytes, the males retreated from an alkylpyrazine source whereas the workers were attracted to, and attacked, the pheromone source.' 36 Some of the reported alkylpyrazines have interesting side chains like citronellyl 13' and styry1.138 Akita and Ohta 139 have reported a recent preparation of (2)-and (E)-2,5-dimethyl-3-styrylpyrazine.The ability to bio- synthesize alkylpyrazines does not appear to be unique to ants because many other insects including wasp^,'^','^^ beetles,141 and flies '42 also produce a variety of alk ylpyrazines.C. Longhurst, R.Baker, and P. E. Howse, J. Chem. Ecol., 1979,5,703. 129 R.M. Crewe and F. P. Ross, Nature (London), 1975,254,448. 130 R.M. Crewe and F. P. Ross, Insect Biochem., 1975,5,839. 131 J. W. Wheeler, J. Avery, 0.Olubajo, M. T. Shamin,C. B. Storm,and R.M. Duffield, Tetrahedron, 1982, 38, 1939. 132 J. J. Brophy and G. W. K. Cavill, Heterocycles, 1980, 14,477. 133 J. W. Wheeler and M. S. Blum, Science, 1973, 182, 501. 134 R.M. Duffield, M. S. Blum, and J. W. Wheeler, Comp. Biochem. Physiol., 1976,54B, 439. 13' M. V. Brown and B. P. Moore, Insect Biochem., 1979,9,451. 136 C. Longhurst, R.Baker, P. E. Howse, and W. J. Speed, J. Insect Physiol., 1978, 24, 833. 13' J. J. Brophy, G. W. K. Cavill, and W. D. Plant, Insect Biochem., 1981, 11, 307.13* G. W. K. Cavill and E. Houghton, Aust. J. Chem., 1974,27, 879. 139 Y.Akita and A. Ohta, Heterocycles, 1982, 19, 329. 140 A,-K. Borg-Karlson and J. Tengo, J. Chem. Ecol., 1980, 6, 827. 14' B. P. Moore and W. V. Brown, Insect Biochem., 1981, 11,493. 14' R.Baker, R.H. Herbert, and R.A. Lomer, Experientia, 1982,38,232. Attygalle and Morgan Table 2 Alkylpyrazines identijied from mandibular glanak of ants Structure Source-species Sub family Re$ R = ethyl Zridomyrmex purpureus Dolichoderinae 208 R = propyl I. humilis Dolichoderinae 100, 138 I. purpureus Dolichoderinae 208 R = n-butyl I-purpureus Dolichoderinae 208 R = isobutyl Anochetus sedilloti Ponerinae 136 Calomyrmex sp. Formicinae 135 R = s-butyl Anochetus sedilloti Ponerinae 136 R = pentyl A.sedilloti Ponerinae 136 R = isopentyl Odontomachus hastatus Ponerinae 133 0.clarus Ponerinae 133 0.troglodytes Ponerinae 136 Ponera pennsylvanica Ponerinae 134 Hypoponera opacior Ponerinae 134 Iridomyrmex humilis Dolichoderinae 100, 138 Rhytidoponera metallica Ponerinae 137 Calomyrmex sp. Formicinae 135 Notoncus ectatommoides Formicinae 78 R = 2-methyl-Calomyrmex sp. Formicinae 135 butyl R = styryl Iridomyrmex hymilis Dolichoderinae 100, 138 R = citronellyl Rhytidoponera metallica Ponerinae 137 R = ethyl Odontomachus brunneus Ponerinae 133 0. troglodytes Ponerinae 136 R = propyl 0.brunneus Ponerinae 133 R = butyl 0.brunneus Ponerinae 133 0.troglodytes Ponerinae 136 A nochet us sedilloti Ponerinae 136 Brachyponera sennaarensis Ponerinae 136 R = i-butyl Anochetus sedilloti Ponerinae 136 R = s-butyl A.sedilloti Ponerinae 136 R = pentyl Odontomachus brunneus Ponerinae 133 0.troglodytes Ponerinae 136 Brachyponera sennaarensis Ponerinae 136 R = hexyl Odontomachus troglodytes Ponerinae 136 Aphaenogaster rudis Myrmicinae 131 Chemicals from the Glanak of Ants C. Ketones and Alcohols.-Secondary alcohols and their corresponding ketones are the most common chemicals found in the mandibular glands of ants 67*69~ven in those of some ponerine ants where these compounds are found instead of the usual pyrazines. 143,144 In Bothroponera soror different components in the mandibular gland secretion release different behavioural responses.'44 For example, 2- undecanone releases alerting and orientation responses and 2-undecanol an attraction response. The mandibular gland secretions of doryline 67 and pseudomyrmecine 67 ants also contain mainly aliphatic ketones.The mandibular gland secretions of myrmicine and formicine ants have been more extensively studied than those of other subfamilies. Among the myrmicine ants, species which belong to the genus Myrrnica are able to biosynthesize an abundance of homologous 3-alkanones and the corresponding 3-alkanols of the C,-C, 1 range,'45-'48 however, 3-octanone and 3-octanol are the major active components. 14' The 3-octanol of Myrmica ants is chiefly 3-( -)-(R)-octanol with small amounts of the (5')-enantiomer. 149 Only the (R)-enantiomer shows biological activity.The 4-methyl-3-heptanone in the mandibular glands of Atta texana is biosynthesized with stereospecific exactitude to yield the (9-(+)-isomer alone, and only this enantiomer is pheromonally active.' Similarly, only the (3R,4S)-isomer of 4-methyl-3-hexanol is found in Tetrarnorium impurum.'5'*'52 Although it is difficult to generalize on the little information available, it would not be surprising if all 3-alkanols of myrmicine ants are found to have (3R)-configuration. The optical isomeric composition of (4E)-4,6-dimethyl-oct-4-ene-3-one(53) (manicone), an alarm pheromone of Manica rnutica and M. hradleyi, is not yet determined.lS3 A number of syntheses of manicone (53) are available."4-' 57 Usually the myrmicine ants produce 3-alkanones 0nly,145-148*1 58 but rarely, 2-alkanones '58 and 4-alkanones '59 are also encountered in the mandibular glands.On the other hand, the formicine ants produce a number of positional isomeric ketones; those with the carbonyl group at 2,76*77 and 477 position are 3,83*160 frequently found. 143 R. M. Duffield and M. S. Blum, Ann. Entomol. SOC. Am., 1973,66, 1357. 144 C. Longhurst, R. Baker, and P. E. Howse, J. Insect Physiol., 1980,26, 551. 145 M. C. Cammaerts, R. P. Evershed, and E. D. Morgan, J. Insect Physiol., 1983, 29, 659. 146 M. C. Cammaerts, R. P. Evershed, and E. D. Morgan, Physiol. Entomol., 1982,7, 119. 14' M. C. Cammaerts, R. P. Evershed, and E. D. Morgan, J. Insect Physiol., 1981, 27, 225.14' E. D. Morgan, M. R. Inwood, and M. C. Cammaerts, Physiol. Entomol., 1978,3, 107. 149 A. B. Attygalle, E. D. Morgan, R. P. Evershed, and S. J. Rowland, J. Chromatogr., 1983,260,411. R. G. Riley, R. M. Silverstein, and J. C. Moser, Science, 1974, 183, 760. J. M. Pasteels, J. C. Verhaeghe, R. Ottinger, J. C. Braekman, and D. Daloze, Insect Biochem., 1981, 11, 675. J. M. Pasteels, J. C. Verhaeghe, J. C. Braekman, D. Daloze, and B. Tursch, J. Chem. Ecol., 1980, 6, 467. 153 H. M. Fales, M. S. Blum, R. M. Crewe, and J. M. Brand, J. Insect Physiol., 1972, 18, 1077. 154 T. Nakai, T. Mimura, and T. Kurokawa, Tetrahedron Lett., 1978,32, 2895. J. A. Katzenellenbogen and T. Utawanit, J. Am. Chem. SOC., 1974,%,6153. 15' P. J. Kocienski, J. M. Ansell, and R.W. Ostrow, J. Org. Chem., 1976,41, 3625. K. Banno and T. Mukaiyama, Chem. kit., 1976,3,279. 15* C. Longhurst, R. Baker, and P. E. Howse, Insect Biochem., 1980, 10, 107. 0.Olubajo, R. M. Duffield, and J. W. Wheeler, Ann. Enromol. SOC. Am., 1980,73, 93. J. W. S. Bradshaw, R. Baker, and P. E. Howse, Physiol. Entomol., 1979, 4, 15. 268 Attygalle and Morgan D. Aldehydes.-Apart from the terpenoid aldehydes, some simple aliphatic aldehydes are also found in the mandibular glands of ants. Some of the common compounds are 2,6-dimeth~l-5-heptenal,~5*76hexanal,' 6o and 2-hexenal. ' ' E. Terpenes.-A variety of terpenoid compounds are found in the mandibular glands of ants, especially among formicine and some myrmicine ants. Some of the compounds encountered are ~itral,~~*'~~*'~~neric acid,'62 geranic acid,'62 citronellol,' 63 ~itronellal,~5*76geraniol,' 62*' farnesol,' 62 2,3-dih~drofarnesal,'~ ~-pinene,'62geranylcitronellal,76andgeranylgeranial.76Perillene(54)158~' ','64and dendrolasin (55)'65 are two furanoid terpenes that are characteristically found.A convenient method to synthesize these 3-substitute furanoids is available.' 66 (53) (54) (55) Some investigations have been made on the biosynthesis of terpenoids in ants. In Acanthomyops cIaviger the use of acetate and mevalonate as precursors is evident from incorporation Similar biosynthetic studies have been made on dendrolasin.168 F. Lactone.-Massiolactone (56) identified from the workers of carpenter (Camponotus)ants is a powerful skin irritant.'69 Recently a related lactone (35), first identified as a queen recognition pheromone originating from the poison glands of Solenopsis in~icta,'~ has been found in the mandibular glands of male carpenter ants.'" Mellein (57) is another lactone from the mandibular glands of carpenter ants 17' and has also been identified in the gaster of Rhytidoponera rneta//ica.l3' 0L OH 0 (56) (57) 161 C.Longhurst, R. Baker, and P. E. Howse, Experientia, 1979,35,870. 162 H. Schildknecht, Angew. Chem., Int. Ed. Engl., 1976, 15, 214. 163 M. S. Blum, F. Padovani, and E. Amante, Comp. Biochem. Physiol., 1968,26, 291. 164 R. Bernardi, C. Cardani, D. Ghiringhella, A. Selva, A. Baggini, and M. Pavan, Tetrahedron Lett., 1967, 40,3893.16' A. Quilico, F. Piozzi, and M. Pavan, Tetrahedron, 1957, 1, 177. 166 S. P. Tanis, Tetrahedron Len., 1982, 23, 31 15. 16' G. M. Happ and J. Meinwald, J. Am. Chem. Soc., 1965,87,2507. E. E. Waldner, C. Schlatter, and H. Schmidt, Helv. Chim. Acta, 1969, 52, 15. 169 G. W. K. Cavill, D. V. Clark, and F. B. Whitfield, Aust. J. Chem., 1968,21,2819. ''O T. H. Jones and H. M. Fales, Tetrahedron Lett., 1983,24, 5439. 17' J. M. Brand, H. M. Fales, E. A. Sokoloski, J. G. MacConnell, M. S. Blum, and R. M. Duffield, Life Sci., 1973, 13, 20 1. Chemicals from the Glandr of Ants G. Benzenoid Compounds.-Some benzenoid aromatic compounds have been detected in the mandibular glands of myrmicine and formicine ants. Benzaldehyde has been identified as a defensive secretion in an attine ant.'72 Phenylethanol constitutes 15% of the mandibular secretion of Camponotus clarithorux.' 73 o-Aminoacetophenone from Mycocepurus goeldi acts as an attractant for these 'perfume ants','74 so called because of the grape-like fragrance of o-aminoacetophone.A related compound, methyl anthranilate is found in the mandibular glands of some Camponotus '71 ants, Aphaenogaster fuha '75 and Xenomyrmex$~ridanus.~~'Male ants of some species of Camponotus 176 and the workers of Bothroponera soror 144 and Gnamptogenys pleurodon l7 secrete methyl 6-methylsalicylate. H. An Overview.-The diverse groups of chemical compounds mentioned above demonstrates the biosynthetic versatility of the ant mandibular gland.Pheromonal and defensive roles have been attributed to a number of these chemicals but the role of most remains obscure. In a few, the role of individual constituents of the mandibular secretion has been worked out. For example, in Bothroponera soror 144 2-undecanone releases an alerting and orientation response, 2-undecanol an attraction response, and methyl 6-methylsalicylate releases stinging activity. A similar study on weaver ants, Oecophylla longinoda, showed that the components hexanal and 1-hexanol release alerting and attraction responses while 2-butyloct-2-enal and 3-undecanone act as markers for attack.'60" 78 The multifunctional role of mandibular gland secretions of an Australian desert ant, Calomyrrnex,has also been described.'79 The composition of the mandibular gland secretions is species- and sometimes caste-specific 180 and is therefore useful in differentiating between morphologically similar species and in chemosystematics. 4 Postpharyngeal Gland Substances The postpharyngeal glands of the ants are a pair of glove-shaped structures overlying the brain. These glands can occupy a large portion of the head and open separately into the posterior portion of the pharynx ' ' (Figure 1). The function of the postpharyngeal gland remains unknown. It may play a part in larval feeding 182 or have a digestive function lg3 (although the lipase activity has been found to be very low 184). Phillips and Vinson lg5 had claimed that the glands function as a 17' M.S. Blum, F. Padovani, F. Curley, and R.E. Hawk, Comp. Biochem. Physiol., 1969, 29,461. 173 H. A. Lloyd, M. S. Blum, and R. M. Duffield, Insecl Biochem., 1975,5, 489. 174 M. S. Blum, J. M. Brand, and E. Amante, Experientia, 1981,37,816. 17' R. M. Duffield, J. W. Wheeler, and M. S. Blum, FI. Entomol., 1980, 63, 203. 17' J. M. Brand, R. M. Duffield, J. G. MacConnell, M. S. Blum, and H. M. Fales, Science, 1973,179,388. 177 R. M. Duffield and M. S. Blum, Experientia, 1975, 31, 466. 17' J. W. S. Bradshaw, R. Baker, and P. E. Howse, Nature (London), 1975, 258, 230. 179 E. J. Brough, Z. Tierpsychol., 1978, 46, 279. J. W. S. Bradshaw, R. Baker, and P. E. Howse, Physiol. Entomol., 1979, 4, 27."' S. A. Phillips and S. B. Vinson, J. Ga. Entomol. SOC.,1980, 15, 215.E. Bugnion, Bull. SOC.Ent. Egypte, 1930, 40, 85. J. Forbes and A. M. McFarlane, J. New York Entomol. Soc., 1961, 69, 92. B. L. Ricks and S. B. Vinson, Entomol. Exp. Appl., 1972, 15, 329. S. A. Phillips and S. B. Vinson, Ann. Entomol. SOC.Am., 1980, 73, 257. 270 Attygalle and Morgan cephalic caecum and that the major lipid components come from the food but Thompson et a1.'86 have recently discovered that the major class of compounds in the postpharyngeal glands of Solenopsis inuicta queens is hydrocarbons. The gland may have a special function in S. invicta queens, however, as it becomes disproportionately large in virgin queens and is filled with fluid prior to their nuptial flight.'86 Very few chemical analyses have been reported on the composition of the postpharyngeal gland.Usually the contents are simply described as a yellow oil. Vinson et al."' found the composition of the hexane-soluble material of postpharyngeal glands of newly mated S. inuicta queens to be 63% hydrocarbons, 19% free fatty acids, 13% glycerol esters, 6% steroids, and a trace of wax esters. The hydrocarbon fraction was analysed by Thompson et allg6who found four major methyl-branched hydrocarbons of the C28--C29 range. Vander Meer et a1.1g8 found that the total hydrocarbon content of the gland showed a marked increase at 15 days after mating, which suggested that the queen has the biosynthetic capacity to produce these materials herself. The increase in hydrocarbon levels of the postpharyngeal glands coincides with wing muscle histolysis.The hydrocarbon level decreased to the original level after 15 days and during this period the free fatty acid and triacylglycerol concentrations remained the same. The postpharyngeal glands of S.inuicta workers contain microgram quantities of (23-9- tricosene accompanied by tricosane and heneicosane among other minor hydrocarbons.'6 5 Metapleural Gland Substances Metapleural glands, located in the thorax, are found in most ants. Although the functions of this gland are still obscure, its secretions are known to consist mainly of carboxylic acids. At least in Atta sexdens rubropilosa, a leaf-cutting and fungus- growing ant, it has been suggested that the metathoracic gland secretions are involved in the control of fungus gardens.The compounds identified from the metathoracic gland of A. sexdens rubropilosa are phenylacetic acid (58), 3-indoleacetic acid (59),'89 3-hydroxydecanoic acid (60),19*3-hydroxyoctanoic acid, and 3-hydroxyhexanoic acid. The last two acids are minor components. 3- M. J. Thompson, B. M. Glancey, W. E. Robbins, C. S. Lofgren, S. R. Dutky, J. Kochansky, R. K. Vander Meer, and A. R. Glover, Lipids, 1981,16,485. S. B. Vinson, S. A. Phillips, and H. J. Williams, J. Insect Physiof., 1980, 26, 645."'R. K. Vander Meer, B. M. Glancey, and C. S. Lofgren, Insect Biochem., 1982, 12, 123. H. Schildknecht and K. Koob, Angew. Chem., 1970,82, 181. 19* H. Schildknecht and K. Koob, Angew. Chem., 1971,83, 110. 271 Chemicals from the Glands of Ants Indoleacetic acid (59) is one of the widely distributed plant growth substances.Phenylacetic acid (58) can also function as a growth regulator, either alone or even more effectively in combination with indole acid (59). The natural compound myrmicacin (60) is laevorotary and acts as a growth inhibitor at higher concentrations. It is indicated that myrmicacin (60) and the other two related hydroxy-acids are used by the harvester ants (Messor) to prevent germination of grass seeds in their granaries and by the leaf-cutting ants (Atta) to prevent germination of undesirable fungal spores on their fungus gardens. However, phenylacetic acid (58), indoleacetic acid (59),and myrmicacin (60), when present in low concentrations stimulate growth as demonstrated by artificial fungus-growing experiment^.'^' The same acids (58), (59), and (60) were found in Myrmica rubra, but the indole acid (59) was absent in Messor barbarus and phenylacetic acid in Acromyrmex subterraneus.'62 Iwadara and Iwanami '92 have briefly reviewed pollen germination-inhibitory activity, animal cell growth-inhibiting activity, and antimicrobial activity of myrmicacin and other related compounds excreted by ants. 6 Substances from the Hind Gut Although the hind gut is not an exocrine gland in the strict sense, it is the source of several natural products in ants. The hind gut is the source of the trail pheromones for all formicine ants that have been investigated." In the case of Lasius fuliginosus it was first observed that the activity could be extracted from the hind gut into water and this aqueous extract was used to lay artificial trails.'93 The activity disappeared to a large extent when the extract was basified and reappeared at the original level when it was re-acidified.Later Huwyler et af.' 94,195 found the active material to be composed of an acidic and a non-acidic fraction. Six fatty acids, namely, hexanoic, heptanoic, octanoic, nonanoic, decanoic, and dodecanoic acids have been identified as pheromone components in the acidic fraction. Commercial samples of the six acids when tested individually could evoke trail-following behaviour in L.fuliginosus workers. However, the activity towards an appropriate mixture of the acids has not been examined.Furthermore, the composition of the non-acidic fraction of the hind gut material remains unknown. Huwyler et have further reported that the trail pheromone isolated from the rectal fluid of the related species L. niger is non-acidic and can be recovered from the gas chromatographic effluent. A similar mixture of fatty acids has been reported to constitute the trail pheromone of the mynnicine ant Pristomyrmex pungens.' 96,197 The glandular source of the trail pheromone of this species remains unknown, although it is probably not the hind gut because no other myrmicine ant so far investigated has the trail pheromones originating from this organ. The mixture of saturated and 19' H. Schildknecht, P. B. Reed, F.D. Reed, and K. Koob, Insect Biochem., 1973, 3,439. 19' T. Iwadara and Y. Iwanami, Yukugaku, 1979,28, 309. lg3 W. Hangartner, Z. Vergl. Physiol., 1967,57, 103. 194 S. Huwyler, K. Grob, and M. Viscontini, J. Insect Physiol., 1975,21, 299. 19' S. Huwyler, K. Grob, and M. Viscontini, Helv. Chim. Acta, 1973,56, 976. 196 N. Hayashi and H. Komae, Z. Nuturforsch., 1973,28c, 626. 19' N. Hayashi and H. Komae, Experienriu, 1977,33,424. Attygalle and Morgan unsaturated fatty acids of C14-C20 range falls out of line when compared with the chemical structures identified as trail pheromone components of other myrmicine ants. Out of the nine fatty acids reported, three were saturated and identified as tetradecanoic, hexadecanoic, and octadecanoic acids.The remaining five unsaturated acids have been only partially identified as hexadecenoic acid, octadecenoic acid, octadecadienoic acid, octadecatrienoic acid, eicosatetraenoic acid, and eicosapentenoic acid. The positions of the double bonds and the configurations have not been determined. Furthermore, the query as to whether these acids really are the trail pheromone components of P.pungens or not remains because the activity of the synthetic analogues has not been reported. The hind gut is also the source of the trail pheromones of a number of species of ponerine and ecitonine ants but nothing is known yet about their chemistry.55 7 Anal-pygidial Gland Substances All ant subfamilies, except formicine ants, possess an anal-pygidial gland (also referred to as anal or supra-anal glands).Dolichoderine ants are unique in producing in their anal-pygidial glands a group of cyclopentanoid monoterpenes known as the iridoids (Table 3). Iridomyrmecin (69) was the first of these to be di~covered,'~' using conventional large-scale isolation techniques, in the Argentine ant Iridumyrmex humilis, which is a common pest species. The compound was said by Pavan to have insecticidal pr0~erties.I~~ Five years later Cavill et isolated iridomyrmecin from I. nitidus. In all, three iridolactones (66, 67, and 69) and two iridodials (65,68) have been isolated from dolichoderine ants. The substances have now all been known for some time and their chemistry has been thoroughly reviewed by Weatherston201 and there is also a more recent summary6' up to 1975.They are structurally related to nepetalactone, a substance which is physiologically active for cats from the catnip plant Nepeta cataria.202 Robinson, recognizing the probable biochemical precursor of the iridoids, devised a biomimetic synthesis of iridomyrmecin (69) and isoiridomyrmicin (66) from (S)-citr~nellal.~~~.~~~ All the iridoids, including the alkaloid actinidine (70),205 which at one time was thought to be an artefact of isolation, are now proposed to arise from citronellal.206 The four simpler cyclopentanoids listed in Table 3 (61-64) presumably arise by further degradation from the same source. Other terpenoid ketones found in the anal-pygidial gland of dolichoderines are 6- methylhept-5-en-2-one 207 (71) (from 13 species,68 I.purpureus,208and Tapinuma 19' M. Pavan, Ric. Sci., 1949, 19, 1011. 199 M. Pavan, Ric. Sci., 1950,20, 1853. 2oo G. W. K. Cavill, D. L. Ford, and H. D. Locksley, Aust. J. Chem., 1956,9, 288. 201 J. Weatherston, Quart. Rev.,1967, 21, 287. 202 J. Meinwald, Chem. Ind., 1954,488. '03 K. J. Clark, G. I. Fray, R. H. Jaeger, and R. Robinson, Angew. Chem., 1958,70, 704. '04 K. J. Clark, G. I. Fray, R. H. Jaeger, and R. Robinson, Tetrahedron, 1959, 6, 201. 205 J. W. Wheeler, T. Olagbemiro, A. Nash, and M. S. Blum, J. Chem. Ecof., 1977,3, 241. 206 G. W. K. Cavill and D. V. Clark in 'Naturally Occurring Insecticides', ed. M. Jacobson and D. G. Crosby, Marcel Dekker, New York, 1971, p. 271. '07 G.W. K. Cavill, D. L. Ford, and H. D. Locksley, Chem. Ind., 1956,465. 208 G. W. K. Cavill, P. L. Robertson, J. J. Brophy, R. K. Duke, J. McDonald, and W. D. Plant, Insect Biochem., 1984, 14, 505. Chemicals from the Glands of Ants Table 3 Cyclopentanoidr identified from anal-pygidial glands of ants Structure0.. (611 Source-species Azteca instabilis A. nigriven tris A. velox Iridomyrmex purpureus Ref: (4 (a)(4 208 Azteca nigriven tris A. velox (a) (a) Azteca instabilis A. nigriventris Iriabmyrmex purpureus Arteca chartifex (a)(4208 (65) Iridod ia1 A. instabilis A. parzensis A. velox Conomyrma pyramicus Dolichoderus scabridus Iridomyrmex confer I. detectus I. nitidiceps I. purpureus (=detectus) I. pruinosus I. rufoniger Tapinorna nigerrimwn T.sessile T. simrothi Attygalle and Morgan Structure Source-species Re$ Dolichoderus scabridus 21 1 a Iridomyrmex nitidus 200, 207,211 0 Tapinoma sessile I I Isoiridomyrmecin Iridomyrmex nitidus I. nitidiceps I. purpureus Isodi hydro- neptalactone IH Dolichoderus clarki D. dentata D. scabridusq;io Iridomyrmex detectus I. humilis (68) I. myrmecodiae I. nitidicepsDolichodial I. rufoniger Ir idomy rmex humilis (69) I. nitidiceps I. pruinosusIridomyrmecin Tapinoma simrothi (b) (4217 208 (4,211 211 21 1 (e), 21 1 cf),100 21 1 217 21 1 (b), (g),2oo 217 (b) 209 275 Chemicals from the Glands of Ants Structure Source-species Ref , Conomyrma sp.205 Zridomyrmex nitidiceps 217 Z. purpureus (= detectus) 208 (70) Actinidine J. W. Wheeler, S. L. Evans, M. S. Blum, and R. L. Torgerson, Science, 1975, 187,254. * D. J. McGurk, J. Frost,G. R. Waller, E. J. Eisenbraun,K. Vick, W. A. Drew,and J. Young, J. Insect Physiol., 1968,14,841. M. Pavan and R. Trave,fnsectessociuux, 1958,5,299.' G.W. K. Cavill and D. V. Clark, J. Insect Physiol., 1967,13, 131. 'G. W. K. Cavill and H. Hinterberger, Aust. J. Chem., 1961,14, 143. G. W. K. Cavill, E. Houghton, F. J. McDonald, and P. J. Williams, Insect Biochem., 1976,6,483. R. Fusco, R. Trave, and A. Vercellone, Chim. Ind. (Milan), 1955,37,251. simrothi 209), 2-methylheptan-4-one (72) (from Tapinoma nigerrimum,2 ' 7'. sessile,68 and T.simrothi 209), 4-methylhexan-2-one (73) (from Dolichoderus clarki ''), and 4-hydroxy-4-methyl-2-pentanone(74) (from Tapinoma simrothi 209) but linear ketones, e.g. 2-heptanone (seven species 68), 2-pentanone (Azteca sp. and Monacis bispinosa 68), and 4-heptanone (Tapinoma simrothi 'O9) are also found widely distributed. It is curious that citronella1 itself is not found in the Dolichoderinae, though it is found in both formicine and myrmicine species. Iridodials are also found in two other unrelated groups of insects, the phasmids or stick insects (together with neptala~tone)~'~.~'staphylinid beetles (rove beetle^),^'^,^'^ and a longhorn beetle (Aromia moschata).216 The toxicity of iridoids to other insects is not clear, but evidently it is not great.Iridodial (65) and dolichodial (69) are unstable compounds and produce a sticky gel and so probably act defensively. The iridolactones are stable and probably act as alarm substances. In a recent investigation of the Australian cocktail ant I. nitidiceps, Cavill's group found that iridodial (65) and isovaleric acid were the major components of the anal-pygidial gland.21 Behavioural tests showed that isovaleric acid was primarily an alarm substance and iridodial was deduced to be the essential re~ellent.~'~ The body of the Australian meat ant (I. purpureus) contains chiefly iridodial and 6-methylhept-5-en-2-one (7 1) (presumably from the anal-pygidial gland) and smaller quantities of other iridoids and as a minor component, the new compound 1,3,3-trimethyl-2,7-dioxabicyclo-[2,2,l]-heptane (75), evidently derived from methylheptenone (71).208 Some recent reports on anal-pygidial gland show the presence of aromatic 209 A.Hefetz and L. A. Lloyd, J. Chem. Ecol., 1983,9, 607. 210 R. Trave and M. Pavan, Chim. Ind. (Milan), 1956,38, 1015. 211 G. W. K. Cavill and H. Hinterberger, Aust. J. Chem., 1960, 13, 514. 212 J. Meinwald, M. S. Chadha, J. J. Hurst, and T. Eisner, Tetrahedron Lett., 1962, 29. 213 R. M. Smith, J. J. Brophy, G.W. K. Cavill, and N. W. Davies, J. Chem. Ecof., 1979, 5, 727. 214 L. J. Fish and G. Pattenden, J. Insect Physiol., 1975, 21, 741. 215 T. E. Bells, W. V. Brown, and B. P. Moore, J. Insect Physiol., 1974, 20, 277. 216 G.Vidari, M. De Bernardi, M. Pavan, and I. Ragozzino, Tetrahedron Lett., 1973,4065.217 G. W. K. Cavill, P. L. Robertson, J. J. Brophy, D. V. Clark, R. Duke, C. J. Orton, and W. D. Plant, Tetrahedron, 1982, 38, 193I. Attygalle and Morgan (71) (72) compounds. 3-Hydroxybenzaldehyde is found along with other components such as isogeraniol, heptadecane, and heptadecene in the pygidial gland secretions of the ponerine ant Rhytidoponera metaZlica.2 l8 Similarly, methylacetophenone and hydroxymethylacetophenone are reported from the dolichoderine ant Hypo-clinea.2’’ 8 Other Glands Very little information is available on the chemistry of other glands. An enzyme analysis of the labial gland of larvae of Solenopsis invicta has been reported.220 A Pavan’s gland constituent, (9-9-hexadecenal has been identified as a trail pheromone component in Iridomyrmex humiIis.22’,222 The geometry of the C-9 double bond is important for the activity because the (Q-9-hexadecenal could evoke only insignificant trail-following a~tivity.~~~,~~~ The analogues, (9-7-tetradecenyl formate, (E)-7-tetradecenyl formate, and tetradecyl formate were also inactive. Van Vorhis Key and Baker have biologically tested this trail pheromone.225,2 26 Conclusion This review has demonstrated the wide variety of substances that have been found in ant glands.Some of them have provided a challenge and inspiration to the synthetic chemist, some of them are yet to be synthesized, some, to the chemist interested in solving structural and synthetic problems are disappointingly simple.’”J. Meinwald, D. F. Wiemer, and B. Holldobler, Naturwissenschafen, 1983, 70,46. ’I9 M. S. Blum,T. H. Jones, R. R. Snelling, W. L. Overa1,H. M. Fales,and R. J. Heiget,Biochem. Syst. Ecol., 1982, 10, 91. ’’O R.S. Petralia, A. A. Sorensen, and S. B. Vinson, Cell. Tissue. Res., 1980, 206, 145. ’”G. W. K. Cavill, P. L. Robertson, and N. W. Davies, Experientia, 1979,35, 989. ’”G. W. K. Cavill, N. W. Davies, and F. J. McDonald, J. Chem. Ecol., 1980, 6, 371. 223 S. E. Van Vorhis Key and T. C. Baker, J. Chem. Ecol., 1982,8,3. 224 S. E. Van Vorhis Key and T. C. Baker, J. Chem. Ecol., 1982,8, 1057. ’”S. E. Van Vorhis Key, L. K. Gaston, and T. C. Baker, J. insect Physiol., 1981,27, 363. 226 S. E. Van Vorhis Key and T.C. Baker, Entomol. Exp. Appl., 1982,32, 232. Chemicalsfrom the Glanh of Ants Nevertheless their isolation and identification in the tiny quantities available are a triumph of modern micro-chemical methods. To the behavioural biologist, the chemist has thrown down a challenge to explain the purpose of all these diverse substances and to the taxonomist to use the substances to make more certain identification of species and their grouping together into genera and tribes. But all scientists can together marvel at the variety of enzymes that must be available to the insects to make these substances and the great diversity of gland contents that has been thrown up through evolution. There are many other facets to the subject which cannot be covered here. To mention one example, many of these substances (the terpenoids, pyrazines, alcohols, ketones, and aromatic compounds) also have distinctive odours for humans, and seem to be detectable in very roughly the same kind of concentrations in ants and man, while other substances, notably the hydrocarbons, convey no odour message to man. Possibly even the fragrance chemist has something to learn from ant chemistry. Acknowledgement.The valuable assistance of Miho Yamakawa in the preparation of this manuscript is gratefully acknowledged.

ISSN:0306-0012    

DOI:10.1039/CS9841300245

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

Designing Drugs to Fit a Macromolecular Receptor * By C. R. Beddell WE LLCOME RESEARCH LABORATORIES, LANGLEY COURT, BECKENHAM, KENT BR3 3BS 1 Introduction In the search for new and better drugs, many different approaches have been used. The present review is concerned with just one of these; the selection of chemical structures by a relatively new process involving matching of chemical structure to a macromolecular target of known architecture. The relationship of this approach to other approaches in medicinal chemistry is also considered. The development of a series of drugs may start with a compound, available from plant, animal, or synthetic source, which possesses some element of the desired characteristics of a drug. When there does exist a very well defined starting point for series development (e.g.a natural regulator molecule) the options open to a medicinal chemist are usually huge and different approaches towards identi- fying a suitable drug will meet with different levels of success, given the restrictions on time and effort which can be expended.Sometimes there may be no molecule regarded as intrinsically suitable as a starting point, perhaps because no known compound has the appropriate potency, or specificity, or because potential com- pounds pose problems of synthesis. The need to reduce the size of the problem is thus obvious and one of the methods employed to achieve this reduction is that of establishing principles specific to the problem in hand which help rationally to direct attention to certain new chemical entities.For example, there may be established relationships between the biological activity of the compounds and certain physicochemical properties, which in turn may indicate compounds with more appropriate physicochemical (and thus biological) properties. The medicinal chemist’s problem is to optimize within a current series of compounds and to generate new series. To do the former, trends are sought relating the physicochemical properties of the molecules to the required effects, for use as constraints in the choice of further compounds for synthesis. To do the latter, the chemist has often had to rely upon chance observations of chemicals possessing interesting biological activities. The search for trends is a problem in itself, as succinctly described by Ganellin.’ ‘This is the dilemma for medicinal chemists; a simple structural change in a drug molecule has multiple consequences.There is usually no clear indication * This review is based on a paper presented on 20th January 1983 to the Perkin Division and Fine Chemicals and Medicinal Substances Group of the Royal Society of Chemistry, the Fine Chemicals Group of the Society of Chemical Industry, and the Pharmaceutical Society. IC. R. Ganellin, in ‘Quantitative Approaches to Drug Design’. ed. J. C. Dearden, Elsevier Science Publishers B.V.. Amsterdam. 1983. p. 239. Designing Drugs to Fit a Macromolecular Receptor of which chemical properties need to be measured or whether they will have importance in determining the biological activity of a compound.Much effort may be expended in seeking for possible property-activity relationships with no apparent successful outcome. This complexity, which is seemingly unanalys- able, drives chemists to empiricism’.* Suppose a small chemical change is made in a drug. This one change may alter certain attributes of the drug, such as its intrinsic stability, its susceptibility to modification by an enzyme, its ability to partition between different body phases, its tendency to bind to one or more receptors, and its ability to produce a response from each receptor. In each process in which the modified drug behaves different- ly, it may be changes in some specific physicochemical property or combination of properties of the drug which are most influential (for example, molecular orbital energies for stability, steric properties for binding, lipophilicity for partition).Now suppose the overall measured effect (E) from all these processes were a linear function of each property (Pi)in the drug. E = c,P, + c,P,-** + c,P, (1) In principle it would be possible to solve for the n coefficients c, given a knowledge of the n P values and the value of E, provided that there were enough (at least n) independent equations and hence E values. Since many properties vary from position to position in the drug, and some are intrinsically vectorial rather than scalar, a great many parameters and hence equations would be needed. In prac- tice, given also the error usually associated with biological and physicochemical pro- perty estimates, a relationship only becomes apparent when some property comes to dominate in producing the effect in some region of this multi-dimensional property space.This itself implies a non-linear relationship, since the partial deriva- tives (equation 2) are unchanging, and thus the relative importance of property alterations would be invariant, in the linear case. dE = ci (2)a pi A further problem which arises is that of congruence ambiguity. In the use of a set of equations, one equation per molecule, there is an assumption of com- parability between the property value PIi for one molecule and a corresponding value PZifor the next, and so on, but the matching of values of position-dependent properties between molecules presumes a spatial superposition of molecules, achieving a congruence between property values, which is relevant to the func- tionally important processes.Clearly, congruence in one process is established by additional positioning parameters for each drug, and congruence may be different in different processes. Many approaches have been used to search for physicochemical property- activity relationships (P.A.R.). It is not within the scope of this review to detail * Reproduced by permission of the author and Elsevier Science Publishers B.V. from ‘Quantitative Approaches to Drug Design’, 1983, p. 239. (a) ‘Strategy in Drug Research’, ed. J. A. K. Buisman, Elsevier Scientific Publishing Co.,Amsterdam, 1982. (b) ‘Quantitative Approaches to Drug Design’, ed.J. C. Dearden, Elsevier Science Publishers B.V., Amsterdam, 1983. Beddell them, but examples of many are provided in recent conference proceedings.2 The present review is concerned with one such approach, dealing with relationships between the physicochemical properties of the drug and the complementary pro- perties of the receptor. The receptor in this context is any macromolecule with which a series of compounds could potentially interact and, by virtue of this interaction, influence pharmacokinetic or pharmacodynamic aspects of the bio- logical actions of the compounds. Consideration is given to the extent to which, in such an approach, the basic problems in P.A.R.of finding local and global optima are overcome, and to what extent the approach is sensitive to some of the factors which can confound classical P.A.R. approaches, namely the multiplicity and interdependence of molecular properties, congruence ambiguity, and the multiplicity of potential functional forms for the relationship. No review of this length can cover the already substantial literature in this area. Examples of work are chosen with the object of providing a balanced rather than exhaustive view of the field; references are likewise limited, but should provide the reader with entry points to the literature. 2 Fitting to the Macromolecule Beginning with the establishment of the three-dimensional structure of DNA and myoglobin; the X-ray crystallographic technique has provided detailed information on the architecture of many macromolecules, including structural and binding proteins, enzymes, and t-RNAs.There is much known at a qualitative level of the forces which maintain the active conformations of these molecules (van der Waals repulsion and attraction, Coulombic interaction between charged centres, hydrophobic forces in an aqueous environment, hydrogen bonding, for example) and these same types of force, in varying degree, are involved in the interaction between macromolecule and ligand. Indeed, Phillips and his col- leagues’ showed that in the interaction of a chain of sugar residues constituting an oligosaccharide with a chain of binding sites on hen lysozyme, there was a rough correspondence, in the absence of strain, between the unitary free energy for binding of each sugar residue and the total number of contacts less than 4Abetween the residue and the protein.With the possibility that so simple a concept might be more widely applicable, several groups have, during the last ten years, pitted their wits against macromolecules by attempting to design small molecules to fit to them. A. Nonenzymic Proteins.-The following three examples of macromolecule fitting involve non-enzymic proteins, functional through binding with either a small ligand or with a receptor. In each case, the design of chemicals to bind to the macromolecule required fitting to sites in the protein composed of parts of more than one subunit.Compounds were thus designed to bind at intersubunit sites near interfaces between subunits. ’J. D. Watson and F. H. C. Crick, Nature (London), 1953,171, 737. J. C. Kendrew, G. Bodo, H. M. Dintzis, R. G.Parrish, H. Wyckoff, and D. C. Phillips, Nature (London), 1958, 181, 662. L. N. Johnson, D. C. Phillips, and J. A. Rupley, Brookhaven Symp. Biof., 1968, 21, 120. 28 1 Designing Drugs to Fit a Macromolecular Receptor (i) Haemoglobin. Haemoglobin contains two types of peptide chain (aand P) each enclosing haem-bound iron and mutually interacting non-covalently in an ~43~ tetramer. The reversible oxygen binding to the iron can be described by a graph showing saturation with oxygen against oxygen partial pressure. The sigmoid relationship observed (Figure 1) is the result of a transition between two conform- ational states of the tetramer, one with low oxygen affinity (the ‘tense’ T-state) and the other with high oxygen affinity (the ‘relaxed’ R-state)6 (Figure 2).Either state alone would generate a rectangular hyperbola such as shown in Figure 1, the curve to the right relating to the state with lower oxygen affinity. In general a right-shift of the sigmoid curve denotes an overall lowering of affinity for oxygen and this is achieved by a natural regulator molecule, 2,3-diphospho-~- glycerate (DPG) (1) (this and several subsequent compounds are illustrated as the neutral form but exist predominantly as charged species at pH values near to those occurring physiologically) which binds between the P-chains with selectivity for the T-state compared with the R-state.Figure 3 shows how DPG with its rc 0 .-5 c Ue cc Figure 1 Relationship between fractional saturation of haemoglobin with oxygen and oxygen partial pressure. The typical sigmoid binding curve (solid) arises from the transition between two conformational states of the haemoglobin, which individually, in the absence of transition are calculated to bind oxygen according to a rectangular hyperbola such as shown to the left (R- state) and right (T-state) of the sigmoid curve J. Monod, J. Wyman, and J.-P.Changeux, J. Mol. Biol., 1965, 12, 88 Beddell anionic phosphate and carboxylate groups is complementary to cationic imida- zolyl and a-amino-groups in the protein and binds to the T-state by charge-assisted hydrogen b~nding.~ DPG interacts less favourably with the R-state, owing to substantial geometric differences between states.Figure 2 Diagrammatic representation of the a& haemoglobin tetramer, showing intercon- version between the R-state (left) and T-state. The cleft between the P-chains in the T-state in which (1) binds is also illustrated C02H I H2O3P -0-CH2 -CH -0-P03H2 (11 \/CH2 -\/ OOHCO CH2 OCH (31 OCH2COzH HOHyoCH2-CH2&CHOH\ SOgHHO3S Compounds Designed to Lower Oxygen Af3nity. With the help of physical models constructed from co-ordinates provided by Dr. M. F. Perutz 8*9 and representing ’A. Arnone, Nature (London), 1972,237, 146.*M. F. Perutz, H. Muirhead, J. M. Cox, and L. C. G. Goaman, Nature (London), 1968, 219, 131 ’W. Bolton and M. F. Perutz, Nature (London), 1970, 228, 551. Designing Drugs to Fit a Macromolecular Receptor u U QJe h Beddell regions between the P-chains, compounds (2H4) were designed to interact selectively with some of the T-state DPG binding-site residues." In contrast to DPG, which interacts only non-covalently, compounds (2)+4), having alde- hyde groups (or the bisulphite addition complex of aldehyde), can in principle form reversible covalent bonds with the N-terminal amino-groups. In addition, since these compounds could position sulphonate and/or carboxylate groups near to cationic groups in the protein they may also interact with the protein by charge-assisted hydrogen bonding, i.e.ionically. Two particular types of interaction between compound and protein were thus considered, namely ionic and covalent, and it was postulated by molecular modelling that (4) would make most (two covalent, five ionic) and (2) fewest (two covalent, one ionic) interactions (see Figure 4). As shown in Figure 5 each compound produced a right-shift of the oxygen saturation curve and a simplistic count of interactions between compound and protein correlated with the size of the right- shift. N.m.r. studies and crystallographic studies have supported the postulate that such compounds interact with the DPG binding-site, and further evidence has been obtained from the following biochemical experiment.The postulated relationship between compound-protein interactions and binding can in principle be explored by varying the structure of the compound and/or the binding site so that more or fewer interactions are made. The oxygen saturation curves for compounds (l), (2),and (4)each binding to one of six different haemoglobins (Table 1) were used to establish the free energy of binding for each compound to Table 1 Residues in the DPG binding-site of selected haemoglobins p-chain residues Haemoglobin 1 2 143 A1 (human) valine histidine histidine AI c (human) valine-glucose histidine histidine adduct Raleigh (human) N-acetylalanine histidine histidine FI(human) N-acetylvaline histidine serine FII(human) valine histidine serine Horse valine glutamine histidine each haemoglobin (3 x 6 = 18 combinations) and separately, the number of covalent and ionic interactions in each pairing were assessed (Figure 6) by the same simple modelling considerations as above.For example, human haemoglobin 10 C. R. Beddell, P. J. Goodford, F. E. Norrington, S. Wilkinson, and R. Wootton, Br. J. Pharmacol., 1976, 57, 201. 'I F. F. Brown and P. J. Goodford, Br. J. Pharmacol., 1977,60,337.12 A. J. Geddes, personal communication. 13 C. R. Beddell, P. J. Goodford, D. K. Stammers, and R. Wootton, Br. J. Pharmacol., 1979,65, 535. 285 Designing Drugs to Fit a Macromolecular Receptor Figure 4 Diagrammatic representation of the region between T-state haemoglobin P-chains, showing the primary interactions between protein and respectively (1) (top left), (2) (top right), (3) (bottom left), and (4).The manner of binding for (1) has been observed crystallographically, that for the other compounds is postulated. Covalent interactions with residues in each P-chain are shown by solid bonds, charge-assisted hydrogen bonh (ionic) are represented by dotted lines (Reproduced by permission of Macmillan Press Ltd. from Br. J. Pharmacol., 1976,57, 201) has histidine at position p-2 whereas horse haemoglobin has glutamine. DPG paired with human haemoglobin interacts ionically at this position, but with horse, this interaction is absent as glutamine is uncharged. With the inclusion of replicate determinations there were 29 estimates of binding energy and a least-squares analysis revealed that the relationship between binding energy (AG) and the respective number of covalent (nc) and ionic (n,) interactions could be expressed linearly.Beddell 100 100 50 50 I 1 5 10 5 10 Figure 5 Each diagram shows two oxygen saturation curves: The one to the left is for haemoglobin whilst the one to the right is for haemoglobin in the presence of one of the compounds (1)--44) respectively at 2.5mM concentration, compound assignments being as laid out in Figure 4. Note that (2)produced a right-shgt smaller than that of (1) whilst the largest right-shift is produced by (4), correlating with the number of interactions counted in Figure 4.Abscissae, oxygen pressure (kPa). Ordinates, per cent oxyhaemoglobin (Reproduced by permission of Macmillan Press Ltd., from Br. J. Pharmacol., 1976, 57, 201) AG = -3.14~ -6.78nc -8.29 (3) (+0.30) ( 0.68) ( k 1.37) n = 29; r = 0.928; F2,26 = 81.15 The relationship developed shows that the strength of the covalent bond (-6.78 kJ mol-') is about twice that (-3.14 kJ mol-') of the ionic bond. The equation is significant at the 0.1% level and 86% of the variation in the data is explained. Figure 7 shows the approximately linear relationship between the values of binding free-energy estimated from oxygen saturation curves and values calculated by equation (3) for the values of nI and nc shown in Figure 6. 287 Designing Drugs to Fir a Macromolecular Receptor 7 + 2 ..................... ............... ............... ............... .................... 8 6 A1 A~c Ral FI FII Horse Figure 6 Histogram of the postulated number (n)of interactions between various haemoglobins and compountis (1) (top),(2)(middle), and (4) respectively aspostulatedfrom study of molecular models. Bar height represents the total number of interactions (covalent and ionic), and shading denotes covalent interactions. The eighteen combinations of haemoglobin and compound provide a wide variation in total contacts (between 1 and 7) and also in the ratio of ionic to covalent interactions (between 7 :0 and 1:2) The various studies with these right-shifting compounds indicate the following points: (1) It is feasible by modelling studies which highlight qualitatively specific types of intermolecular interaction, to design novel ligands (a) which interact with a protein of defined amino-acid sequence and architecture, and (b) which selectively differentiate between different conformations of the same protein to produce thereby the expected allosteric effect.(2) Binding free-energy can be regarded as a linear function of the numbers of each type of postulated interaction to a first approximation. Compounds Designed to Raise Oxygen Affinity. There exists between the a-chains of haemoglobin a site comparable to that between the 0-chains (since a and p are homologous) but different in detail owing to differences between the amino-acid sequences of the two types of peptide chain.Molecular models of this Beddell -40 *G,eas. (kJ mot-’) -30 -20 -1 0 AGCatc. (kJ mot-l) I -1 0 -20 -30 -40 Figure 7 Relationship between free energy of binding for compound to T-state haemoglobin as estimated from oxygen saturation curves (Acmeas),and as calculated from equation (3)(AGca,c)with the values of n, and n, shown in Figure 6. Compounds are denoted, respectively, (1)by open circles, (2) by solid squares, and (4) by open squares \ CH2-CHZ-CO2H CHO Q-iH2-cHz CHO ‘CH2 -CH2-COzH Q-07Hz-cHz HO CHO (7) region in the R and T states were again constructed from co-ordinates provided by Dr. M. F. Perutz8’9 and molecules were tailored to interact at this site with a selectivity for the R state.I4 Such selectivity should shift the oxygen saturation (a) C.R. Beddell, P. J. Goodford, G. Kneen, R. D. White, S. Wilkinson, and R. Wootton, Br. J. Pharmacol., 1984,82, 397; (b)G.Kneen and R. D. White, Br. J. Pharmacol., 1981,74,965P. Designing Drugs to Fit a Macromolecular Receptor curve to the left. Molecules (5t-(7) were designed to interact covalently (by virtue of the aldehyde function) with one N-terminal group in the protein and to interact ionically via the carboxylate group with the other N-terminal group (Figure 8). Such binding would also position the molecules between the hydrophobic side- chains of two proline residues. Thus the central part of each molecule was designed to be non-polar in character to promote hydrophobic interaction.It was found, as predicted, that molecules (5j(7) could shift the oxygen saturation curve to the left. Furthermore a greater left-shift was obtainable in the presence of the right-shifting compound L-myo-inositol hexaphosphate (8) which binds at the DPG binding-site (saturation curves are shown in Figure 9). Compound (5) showed a small left-shifting effect. It was anticipated that a hydroxy- Figure 8 Diagrammatic representation of the region in R-state haemoglobin between the a-chains, with thepostulatedmanner of binding for compounds (5) (top left). (6) (top right), and(7) Beddell group adjacent to the aldehyde group would promote Schiffs base formation and hence binding, but owing to initial synthetic difficulties a more flexible analogue (6) was made.This was slightly less effective than (5)in these particular experiments, a finding perhaps attributable to its increased flexibility and less suitable hydrophobic moiety. However, addition of a hydroxy moiety produced compound (7), which was substantially more potent. OP03H2I (8) Thus in this study it again proved possible, by consideration of the macromolecule structure, to design selective chemicals to produce a desired effect. The relative potencies of the various compounds can be partially rationalized in terms of the postulated interactions, although it is known that the binding properties of such compounds are more complex than the simple modelling considerations had ~uggested.'~" One of these compounds is now under consideration as a potential therapeutic agent for sickle cell disease.14' In this disease, red blood cells contain S haemoglobin.At low levels of oxygen in the capillaries, the T state polymerizes and distorts the cells to a sickle shape; these distorted cells do not function normally and clinical symptoms arise. Inhibition of the sickling of such cells would be expected of a compound designed to favour the R-state and thereby to lower the level of T-state, and some such compounds have indeed been shown to inhibit and to reverse in uitro sickling. (ii) Insulin. Insulin contains two types of peptide chain, denoted A and B, typically of 21 and 30 residues respectively.These are linked covalently through two disulphide bonds. The hormone circulates in blood primarily at the monomer level, but it is stored intracellularly as a hexamer." Crystals of the hexamer, grown in the presence of zinc ions, have been intensively studied by X-ray crystallography.' By molecular modelling to the insulin hexamer, Andrews and co-workers l7 have designed small organic molecules to bind in a cavity along the central axis of this T. Blundell, G. Dodson, D. Hodgkin, and D. Mercola, Adv. Protein. Chern., 1972, 26, 279. l6 T. Blundell, J. F. Cutfield, S. M. Cutfield, E. J. Dodson, G. G. Dodson, D. C. Hodgkin, D. A. Mercola, and M. Vijayan, Nature (London), 1971,231,506. D. T. Manallak, E. F. Woods, and P. R. Andrews, personal communication.29 1 Designing Drugs to Fit a Macromolecular Receptor 2 4 6 8 100 Figure 9 Each diagram shows oxygen saturation curves for haemoglobin in thepresence of 100 pM compound (8) (solid circles) and for haemoglobin in the presence of both 100 pM compound(8) and 50 pM compound (9,(6), or (7), with compound assignment as laid out in Figure 8. Relative to the control curve, each compound shifts the oxygen saturation curve to the left, and (7) produces the largest shift. Abscissae, oxygen pressure (kPa), Ordinate, per cent oxyhaemoglobin hexamer and to stabilize the hexamer against dissociation by interaction between subunits. One of the molecules designed was found to influence in uitro the distribution of monomer, dimer, tetramer, and hexamer species in the direction of higher molecular weight; the net weight-average molecular weight determined by sedimentation analysis in a solution of insulin was increased by 50%.Beddell (iii) Prealbumin. The hormone L-thyroxine (9) interacts with various proteins, including a cell nucleus receptor and in plasma, a globulin and prealbumin. The crystal structures of human serum prealbumin alone 18,19 and in complex with (9)2o have been used 21 as a basis for macromolecular fitting of thyroxine analogues (lOF(16). Prealbumin is a tetramer of identical subunits which are oriented to form a central channel containing two (9) binding sites. The 222 symmetry of the tetramer requires that not only are the two sites identical, but also that each site itself has on average two-fold symmetry. The binding of (9) to one site presumably alters the other in some way, since the binding of the second molecule is weakened, but statistical averaging in the crystal obscures the finer details.The hormone binds along a two-fold axis with its phenolic hydroxy-group buried deep within the binding channel and its carboxy- and amino-groups ion-paired with side chains of residues lysine-15 and glutamate-54 respectively at the mouth of the binding channel. Figure 10 is a schematic drawing of some of the compound (9)- prealbumin interactions, and shows six pockets capable of binding an iodine substituent, of which four are occupied by the iodines of (9). Of the remaining pockets one is occupied by a crystallographically well-defined water molecule, leaving one pocket empty.Owing to the two-fold symmetry, these six pockets are symmetrically disposed in pairs, denoted in Figure 10 as P(l) and P(l’), P(2) and P(2’), P(3)’and P(3’). The inner ring iodines, 1(3) and I(5), bind to identical hydrophobic pockets, P(l) and P(l’), lined with the methyl groups of leucine-17, threonine-106, alanine-108, and valine-121 and the polymethylene side-chain of lysine-15. The iodine atoms, 3‘ and 5’, of the phenolic ring fit into pockets unrelated by symmetry. The 3’-iodine atom, proximal to the inner ring, binds to pocket P(2), composed of the carbonyl oxygen of lysine-l5A, the side chain of leucine-l7A, the methyl group of alanine-l08A, and the peptide backbone of alanine-109A (the individual protein subunits being denoted, A, B, C, D respectively).The 5’-iodine atom, distal to the inner ring, fits into pocket P(3’) formed by the methyl and HO R(3‘) R(4‘) R(3’)R(5‘4-35 ,NH; RB0+ CHz-CHNH$I1 R(7’)‘’CH2-I:I \/ co, I(3) I(3) (9) R(3‘) = R(5‘) = I (13) R(3’)=Br; R(4’) =OH;R(5’) =R(7’)=H (10) R(3’) = R(5‘) = Br (14 R (3’) = R( 5’) = R(7‘) = H; R(4’)= OH (11) R(3’) = Br; R(5’) =H (15) R(3’) = R(4’) = R(7’)=H; R(S’)= OH (12) R(3’) = R( 5’) = H (16) R(3’) = R(4’)= R15’)=H; R(7‘)= OH l8 C. C. F. Blake and S. J. Oatley, Narure (London), 1977,268, 115. l9 C. C. F. Blake, M. J. Geisow, S. J. Oatley, B. Rerat, and C. Rerat, J. Mol. Biol., 1978, 121, 339.’* C. C. F.Blake, Proc. R. SOC.Lond., Ser. B., 1981,211,413. ” J. M. Blaney, E. C. Jorgensen, M. L. Connolly, T. E. Ferrin, R. Langridge, S. J. Oatley, J. M. Burridge, and C. C. F. Blake, J. Med. Chem., 1982, 25, 785. Designing Drugs to Fit a Macromolecular Receptor carbonyl groups of alanine-l08C, the backbone nitrogen and carbonyl group of alanine-lOgC, the backbone nitrogen and the side chain of leucine-1 lOC, and the hydroxy-groups of serine- 1 17C and threonine- 119C. The water molecule is held in pocket P(3) by hydrogen bonds with the hydroxy-groups of serine-117A and threonine-l19A and possibly with the phenolic hydroxy-group of (9). A possible close contact (<3.2 A) between I(3’) and the carbonyl oxygen of alanine-109A may reflect the polarizability and charge-transfer ability of iodine and may contribute significantly to binding.Figure 10 A schematic illustration of thyroxine (9) in its complex with prealbumin. Some pockets [P( 1jP(3)] in theprotein andsymmetry mates [P( l’jP(3’)l in the neighbourhoodof (9),accommodate iodine atoms; P(3) accommodates water near the phenolic hydroxy-group and P(2’) is unoccupied (Reproduced by permission of the authors2’ and the American Chemical Society from J. Med. Chem., 1982, 25, 785) Molecular modelling of a variety of thyroid hormone analogues by Blaney et al.” led these workers to conclude that analogues with high binding-affinity would occupy at least three of the four outer-ring pockets P(2), P(2’), P(3), and P(3’); the less empty space in the binding site, the more tightly would the analogue bind.This was thought to be due to the stronger van der Waals attraction associated with increased surface complementarity between ligand and protein. During modelling work it was observed that an empty pocket P(2’) could be filled by the presence of an appropriate substituent at the 6’-position in thyroid hormone analogues. Compounds (10)-(16) were therefore modelled to the protein. The a-napthyl analogues were suitable in part, because of their relative rigidity, thus constraining fitting options during molecular modelling. Based on the complementarity of the BeddeII molecular-surface models for the fitting of analogue to protein, the authors predicted that (1 3) would bind more tightly than either (10) or (1 1) since (1 3) is the only analogue which can simultaneously fit all available pockets in the binding site.Similarly (14) was expected to bind more tightly than (12). Comparing (13)-(16), (13) matched the surface of the binding site most closely, with no bad contacts (i.e. contacts shorter than van der Waals). Compounds (14)-(16) all lack the bromine substituent, so that one pocket is empty. The phenolic hydroxy-groups on (14) and (15) appeared to be equally well accommodated by the binding site but the hydroxy-group on (16) would ‘collide’ with the surface of the binding site and (16) might thus bind in a relatively strained manner or in an alternative orientation. Therefore the predicted order of binding affinity for the a-napthyl analogues was K(13) > K(14) N K(15) > K(16).Table 2 Binding to prealbumin of some thyroxine analogues: comparison between modelling predictions and binding constants estimated by competition-binding assay (data from re$ 2 1) Binding Constant Predicted Binding Observed Binding Ratio Constant Ratio Constant Ratio 3.3 17.9 5.3 17.9 0.25 9.0 The match between predicted and measured relative affinity is shown in Table 2. The predicted relative affinities obtained by molecular modelling studies refer to L-enantiomorphs. Likewise the measured affinities also refer to L-enantiomorphs, except for compounds (13), (14), and (1 5) which were racemic. The authors found that the affinity of racemic (9) is 0.59 that of ~-(9), which, if true also for other compounds, would provide the factor necessary to enable the affinity of the L-enantiomorph to be calculated from that obtained with racemic material.The measured relative affinities shown in Table 2 which involve these three compounds incorporate this correction factor. The experimentally determined binding affinities are in the predicted order except for the difference between (14) and (15), which difference was rationalized by more detailed investigation of the fit. It was observed that the hydroxy-group of (15) can interact more strongly with the site than that of (14). The hydroxy-group of (15) in pocket P(3’), comes within hydrogen-bonding distance of the hydroxy-groups of serine-l17C and threonine-l19C.That of (14) is not deep enough to make such interactions directly, though it might do so through an intermediate water molecule. In another paper 22 the estimation of binding enthalpy by molecular mechanics computation for ~-(9), ~-(9), and the deamino (17) and decarboxy (18) analogues was reported. It was concluded that the rank order of binding constants for the four compounds { K( 17) > fl~-(9)] > flD-(9)] > K( 18)>can be matched 22 J. M. Blaney, P. K. Weiner, A. Dearing, P. A. Kollman, E. C. Jorgensen, S. J. Oatley, J. M. Burridge, and C. C. F. Blake, J. Am. Chem. SOC.,1982, 104,6424. 295 Designing Drugs to Fit a Macromolecular Receptor by the calculations, provided in addition an allowance is made for the differences in the free energies of hydration for the complexed and uncomplexed species.Inspection of the modelling work reveals that the analogues are not modelled in a precisely congruent manner. The deamino-compound (17) is modelled with the carboxy-group between symmetrically paired amino-groups of the protein and interacting with both [whereas the carboxy-group of (9) is modelled asymmetrically]. The amino-group of the decarboxy-compound (18) is unable to form good interactions if symmetrically placed and, however placed, makes fewer close interactions than does the carboxy-group of the deamino-compound. Hence it is expected that (17) will bind more tightly than (18) and this was found to be the case. However, it is not immediately obvious why (17) binds more tightly than (9), when (9) has an extra group with which to interact with the protein.The anomaly was attributed to different degrees of desolvation accompanying binding. For (9), with a carboxy- and amino-group, both of which would be desolvated to some extent upon binding, the energetic penalty is greater than for the deamino (or de- carboxy) analogues with one less polar group. This work is of particular importance in that it attempts to incorporate the influences upon binding of hydration. Hydration is ubiquitous and hydration energies are substantial and it might seem surprising that so many other studies should have been successful, despite neglect of hydration. Such success might be hoped for in circumstances in which the hydration changes which accompany the interaction of a series of compounds with the macromolecular receptor are ‘well-behaved’ in the sense that they can be considered to be the sum of a series of components, one for each chemical part of the compounds, and that therefore the hydration-change component for any individual chemical group is constant.This circumstance may well apply when each chemical group, present in a number of different compounds, experiences the same molecular environment regardless of which compound carries it into the binding site in the macromolecule. However, if, for various compounds in the series, the binding site varies in shape, or the compounds bind in different conformations or orientations, it may be essential to consider hydration explicitly before correct predictions can be made of the relative strengths of interaction between compounds and receptor.B. Enzymic Proteins.-The following examples of studies with enzymes all involve modelling to active sites within a single subunit. The object of each study was the production of compounds which would bind at the active site and thereby inhibit enzymic action. (i) Angiotensin-converting Enzyme. Angiotensin-converting enzyme (ACE) cleaves the inactive decapeptide angiotensin I (19) in blood between the eighth and ninth residues to form an active octapeptide angiotensin I1 (20) (Figure 11). The octapeptide can induce vasoconstriction, antidiuresis, and antinatri~resis.~~ Inhibitors of the enzyme may thus produce therapeutically useful effects, including reduction in blood pressure.A compatible influence might in principle arise 23 D. W. Cushman, M. A. Ondetti, H. S. Cheung, E. F. Sabo, M. J. Antonaccio, and B. Rubin, in ‘Enzyme Inhibitors as Drugs’, ed. M. Sandler, Macmillan Press Ltd., London, 1977, p. 231. ANGIOTENSIN I INACTIVE HEPTAPEPTIDE Asp-Arg-Va I-Tyr -I Ie -His -Pro-Phe-H is-Leu Arg-Pro-Pro-GI y-Phe-Ser-Pro ANGIOTENSIN -1 Beddell CONVERTING1 ENZYME AspArg-Val -Tyr-l le-His-Pro-Phe Arg-Pro-Pro-G) y-PheSer-Pro-P he-Arg ANGIOTENSIN I1 I BRADYKININ I i i VAS 0C0NSTRlCTION VASODILATION ANTlDlURESlS DlURESlS ANT 1 NATRl UR ESIS N AT R IURESIS Figure 11 Reactions effected by ACE. Angiotensin Z (19) is converted into active hormone angiotensin ZZ (20), and bradykinin (21) is converted into an inactive heptapeptide (22)(Reproduced by permission of the authors23 and Macmillan Press Ltd.from ‘EnzymeInhibitors as Drugs’, 1980, p. 231) as well from the concomitant protection of bradykinin (21) against hydrolysis by ACE to an inactive heptapeptide (22) with a resultant potentiation of bradykinin-induced vasodilation, diuresis, and natriuresis. Many peptide substrates and inhibitors of the enzyme have been found,23 but of particular relevance in the present context are the molecules shown in Table 3. Biochemical work on the enzyme had revealed that it contains essential zinc and Cushman et al.23postulated that there might be a relationship between the structure of the active site of this zinc-containing carboxydipeptidase and that known to exist in the zinc-containing bovine pancreatic carboxypeptidase A.This latter enzyme hydrolyses the carboxy-terminal residue from a polypeptide in the manner shown diagramatically in Figure 12. Cushman et al. constructed a hypothetical model for ACE in which the hydro-lysed peptide group interacted with the zinc atom, and the terminal carboxy-late interacted ionically with a cationic centre in the protein. The model for ACE differed from that of carboxypeptidase in that the separation between these interaction siteswas increased to change the catalytic activityfrom carboxypeptidase to carboxydipeptidase (Figure 13). The model of the enzyme also incorporated pockets to accommodate amino-acid side-chains.Byers and Wolfenden 24 suggested that benzylsuccinic acid (23) (Figure 12) is a ‘bi-product analogue’ in its manner of binding to carboxypeptidase A and one way in which it might bind, due to Cushman et ~21.~~is shown in Figure 12. Furthermore several peptides with C-terminal proline are inhibitors of ACE, and these two considerations led Cushman et aLt3 to consider succinylproline (24) as a potential inhibitor of ACE, binding as 24 L.D. Byers, and R.Wolfenden, J. Biol. Chem., 1972,247, 606. Designing Drugs to Fit a Macromolecular Receptor Table 3 Activities in vi tro of carboxyalkanoyl and mercaptoalkanoyl amino-acid inhibitors of angiotensin-converting enzyme. 150 is the concentration of a compound producing a 50 per cent inhibition of the activity of rabbit-lung converting enzyme, or a 50 per cent inhibition of the contractile activity of angiotensin I on guinea-pig ileum strip; Asois the concentration producing a 50 per cent augmentation of the contractile action of bradykinin (Reproduced by permission of the authors23 and Macmillan Press Ltd.from ‘Enzyme Inhibitors as Drugs’, 1980, p. 231) 1 RA6BI T- LUNG GUINEA-PIG ILEUM CONVERTING ENZYME A1 BK Kl(yM) Iso(YM) ASO(IJM) (24) 330 -440 37 22 2.5 18 0.87 4.9 0.8 23 4.9 1.1 -0.I9 0.037(27) 0.20 0.012 0.30 0.025(28) 12 5.I (30) 0.023 0.0017 0.023 0.0032 shown in Figure 13. As seen in Table 3, (24) is inhibitory, albeit weakly.Furthermore, the model correctly predicted the enhanced activity seen for analogues of (24) in which a methyl group is attached to the succinyl residue with a configuration*which would match that of the corresponding L-residue [at R(2)] in substrate and product. Thus ~-2-methy1succinyl-t-proline(25) is an analogue of Ala-Pro, the terminal dipeptide in a number of peptide inhibitors derived from snake venom. The model is compatible with the high activity also shown by compounds with a chain length slightly longer than succinyl [e.g. the glutaryl derivative (26)] but derivatives longer than glutaryl are, as the authors expected, less active. A very important advance, made with the aid of the rough hypothetical model of the enzyme, arose when the putative carboxylate zinc ligand was replaced by other functional groups with suitable chemical and positional characteristics for tight binding to zinc.Thus (Table 3) a sulphydryl-containing analogue of (24), i.e. (28), is about a thousand-fold more potent than (24). Shortening or elongation of (28) lowered potency. Captopril (30) is even more potent, and also about a thousand-fold more potent than the corresponding analogue, (25). This study shows how fitting to even a rough model of the macromolecule can be Beddell CARBOXYPEPTIDASE A OF BOVINE PANCREAS I"20 SUBSTRATE-PRODUCTS 0-2-BENZYLSUCCINIC ACID Figure 12 Diagrammatic model of the binding of peptide substrate (top) and products lo the active site of carboxypeptidase.A potential binding mode for the inhibitor ~-2-benzylsuccinic acid (23) is also shown (Reproduced by permission of the authors 23 and Macmillan Press Ltd. from 'Enzyme Inhibitors as Drugs', 1980, p. 231) successfully combined with other sources of knowledge in the design of new drugs. Clearly, the possibility of dramatic improvements in potency through the strengthening of a single putative interaction (with zinc) is a most important factor, but other interactions, such as those mac'e by the methyl substituent, contribute significantly to binding. (ii) Carbonic Anhydrase C. Carbonic anhydrase is inhibited by primary sulphonamides and the utility of acetazolamide (31), both as a diuretic and in the treatment of glaucoma has been attributed to such inhibition.The structure of the C isozyme in the presence and absence of (31) has been determined crystallographically.* The enzyme active-site contains a hydrophilic region, in which a zinc atom is co-ordinated to the imidazoles of three histidines; nearby is a hydrophobic region, provided by residues valine-121, isoleucine-9 1, and phenyl- 25 K. K. Kannan, I. Vaara, B. Notstrand, S. Lovgren, A. Borell, K. Fridborg, and M. Petef, in 'Drug Action at the Molecular Level', ed. G. C. K. Roberts, Macmillan Press Ltd., London, 1977, p. 73. 299 Designing Drugs to Fit a Macromolecular Receptor ,NH-ANGIOTENSIN- CONVERTING ENZYME (HYPOTHETICAL) X X SUBSTRATE-PRODUCTS SUCCINYL -L-PROLINE Figure 13 Diagrammatic representation of a hypotheticalmodel of the active site of ACE with bound substrate (top) and product and with the competitive inhibitor succinyl-L-proline (24).X-H represents a group in the protein h drogen-bonding with the terminal peptide carbonyl group, and residue side chains R’and RYare accommodated in pockets. The zinc provides an important interaction with the carbonyl group of the peptide to be hydrolysed (substrate), or with a carboxyl group in product or inhibitor (Reproduced by permission of the authors 23 and Macmillan Press Ltd.from ‘Enzyme Inhibitors as Drugs’, 1980, p. 231) alanine-131. Compound (31) has been described by Gill et a1.26to be interacting as shown in Figure 14, with co-ordination of the sulphonamide nitrogen and of one oxygen to the zinc.The thiadiazole moiety is placed with the sulphur in contact with the hydrophobic side-chain of valine-121, and the nitrogen atoms may be hydrogen bonded with water molecules; the water molecules in turn could be hydrogen bonded to main-chain carbonyl groups of residues threonine-200 and proline-201. The acetamido-chain carbonyl could accept a hydrogen bond from the side-chain nitrogen of glutamine-92. Thus the interactions between ligand and protein are mainly of polar character. Gill and his colleagues, by inspection of a physical model of the active site and by modelling, devised some alternative sulphonamides with groups which might exploit the hydrophobic pocket. The isobutyl thiophene analogue (32) was modelled with the alkyl side-chain in the hydrophobic pocket, the sulphur atom in contact with valine-121, and the sulphonamide group binding to zinc.It was reported to be twice as potent as (31) at inhibiting the enzyme. (iii) Dihydrofolate Reductase. Dihydrofolate reductase (DHFR) is a widely distributed enzyme which regenerates tetrahydrofolate (33) from folate (34) and 26 E. W. Gill, B. M. Goodall, and P. J. B. Hancock, ref. 1, p. 121. Beddell from dihydrofolate (35). Compound (34) arises from spontaneous oxidation of reduced derivatives, especially in foods. The product from DHFR, (33), acquires a methylene group from serine (Figure 15) to become 5,lo-methylenetetrahy-drofolate (36). This methylene is transferred from (36) to deoxyuridylate and the thymidylate so formed is incorporated into DNA.Concomitant with this transfer is the oxidation of the 5-6 bond, thus generating the DHFR substrate (35) and completing the cycle. Without the active cycle of reactions in which DHFR plays a part, cells will be unable to synthesize DNA, unless they can acquire thymidylate by salvage routes. Thus inhibition of DHFR will stop cell division. By comparing the inhibitory properties of various diamino-derivatives, Burchall and Hitchings *' and others demonstrated that their potencies varied with DHFR origin (vertebrate, trypanosomal, plasmodial, and bacterial) and this variation has now been shown to be related to structural differences between the DHFRs. Indeed, we now know the amino-acid sequence of several DHFRs, the three- dimensional architecture of DHFR from mouse, chicken, E.coli, and L.casei, in q&YL" NIIIII I Il>oN THR \ HIS r Figure 14 Schematic illustration of the active site of carbonic anhydrase to show the binding of acetazolamide (31)(Reproduced by permission of the authors 26 and Elsevier Science Publishers B.V.from 'Quantitative Approaches to Drug Design', 1983, p. 121) *' J. J. Burchall, and G. H. Hitchings, Mol. Pharmacol., 1965, 1, 126. Designing Drugs to Fit a Macromolecular Receptor k (331 IH 1351 (371 complexes with inhibitors and/or cofactors, and something of the dynamics and multiplicity of conformational states which can occur (see reviews 28-3 ’). DHFR is a target for several drugs. Methotrexate (37) inhibits DHFR from most sources and is commonly used in cancer therapy.Pyrimethamine (38), a specific inhibitor of plasmodium DHFR, is used as an antimalarial. Trimethoprim (39), a selective inhibitor of DHFR in many bacteria, is widely used as an ntibacterial, whilst diaveridine (40) has found some veterinary use as an anticoccidial. As a result of widespread crystallographic, binding, and kinetic studies, DHFR is now perceived to be a useful protein for testing fitting methods. Much work is in progress, and one published study sheds some light on the 28 G. H. Hitchings, and S. L. Smith, Adv. Enzyme Reguf., 1980, 18, 349. 29 G. H. Hitchings, and B. Roth, in ‘Enzyme Inhibitors as Drugs’, ed. M. Sandler, Macmillan Press Ltd., London, 1980, p.263. 30 C. R. Beddell, in ‘X-Ray Crystallography and Drug Action’, ed. A. S. Horn and C. J. De Ranter, Oxford University Press, 1984, p. 169. G.C.K.Roberts,in ‘Chemistry and Biology ofPteridines’,ed. J. A.Blair,WalterdeGruyter,Berlin,inpress. 302 Beddell NADPH+H+ NADP+ Pool of Cl-FHa 1 1DNA Figure 15 Some metabolic interrelationships of reduced folate cofactors accuracy of fitting predictions. Following the elucidation of the binding mode for (37) to DHFR32-35 (Figure 16), (39) was modelled to the enzyme.36 The conformation selected for the compound was compatible with the site geometry, the bound position of (37), and with activity data of 6-substituted compounds, in (39) R = OMe (38) (40) R = H 32 D. A.Matthews, R. A. Alden, J. T. Bolin, S. T. Freer, R. Hamlin, N. H. Xuong, J. Kraut, M. Poe, M. Williams, and K. Hoogsteen, Science, 1977, 197,452. 33 D.A. Matthews, R. A. Alden, J.T.Bolin,D. J. Filman,S.T. Freer, R. Hamlin, W. G. J. Hol, R. L. Kisliuk, E. K. Pastore, L. T. Plante, N. H. Xuong, and J. Kraut, J. Biol. Chem., 1978, 253,6946. 34 D. A. Matthews, R. A. Alden, S. T. Freer, N. H. Xuong, and J. Kraut, .I.Biol. Chem., 1979,254,4144. 3s J. T. Bolin, D. J. Filman, D. A. Matthews, R. C. Hamlin, and J. Kraut, J. Biol. Chem., 1982,257,13 650. 36 L. F. Kuyper, B. Roth, D. P. Baccanari, R. Ferone,C. R. Beddell, J. N.Champness,D. K. Stammers, J.G. Dann, F. E. A. Norrington, D. J. Baker, and P. J. Goodford, J. Med. Chem., 1982,25, 1120. Designing Drugs to Fit a Macromolecular Receptor F Figure 16 Schematic illustration of the active site of E.coli dihydrofolate reductase with bound methotrexate.Selected atoms are highlighted, oxygen by stripes, nitrogen in black, and sulphur by hatching. The proximity of N( 1) and the 2-amino-group of methotrexate to the carboxy- group of residue-27 and of the a-carboxy-group to the guanidinium group of residue-51 are examples of charge-assisted hydrogen-bonded interactions. The benzene ring in methotrexate is in hydrophobic contact with the side chains of residue-50 (isoleucine) and -28 (leucine)(Reproduced by permission of the authors3* and Walter de Gruyter from ‘Chemistry and Biology of Pteridines’, 1983, p. 545) which steric hindrance has been inferred.37 This conformation was shown shortly thereafter to be essentially correct 38 (Figure 17).In the molecular model of the enzyme with (39) bound there is an arginine side-chain close by. This interacts with the glutamate a-carboxy-group of methotrexate when this binds, by charge-assisted hydrogen bonding. Molecular modelling considerations led to the replacement of one meta methoxy-group of (39) by a carboxyalkoxy-group, and a chain length (five methylene groups) was selected to optimise interaction between the carboxylate and the guanidinium group of this arginine. This compound (45) and analogues [(41)-(44) and (46)Jwith shorter or longer chains were made (Table 4). Most were more effective than (39) in binding to and inhibiting the enzyme, but the first selected compound (45) was the most potent.Esters (47)-(51) were less potent inhibitors than the respective acid, demonstrating the important contribution to binding made by the free carboxylate group. Furthermore, a crystallographic study36 of (45) confirmed the mode of binding (Figure 18), whilst study of (42) revealed (Figure 18) that lower potency was associated with loss of some hydrogen 37 B. Roth, E. Aig, K. Lane, and B. S. Rauckman, J. Med. Chem., 1980,23, 535. 38 (a)D. J. Baker, C. R. Beddell, J. N. Champness, P. J. Goodford, F. E. A. Norrington, D. R. Smith, and D. K. Stammers, FEBS Lett., 1981, 126, 49. (b) D. J. Baker, C. R. Beddell, J. N. Champness, P. J. Goodford, F. E. Norrington, B. Roth, and D. K. Stammers, in ‘Chemistry and Biology of Pteridines’, ed.J. A. Blair, Walter de Gruyter, Berlin, 1983, 545. Beddell F Figure 17 Schematic illustration of the active site of E.coli dihydrofolate reductase with bound trimethoprim. Like methotrexate, N(1) and the 2-amino-group interact with aspartate-27. As with the benzene ring in methotrexate, the benzene ring in trimethoprim makes hydrophobic interaction with isoleucine-50 and leucine-28 (Reproduced by permission of the authors 38 and Walter de Gruyter from ‘Chemistry and Biology of Pteridines’, 1983, p. 545) bonding, as expected for a compound with a short carboxyalkoxy-group. It has been established that (39) adopts a conformation in the complex with vertebrate DHFR39 different from that which it assumes in bacterial DHFR.Although the architectures of the bacterial and vertebrate DHFRs are rather similar, especially at the active site, (39) is clearly sensitive to the differences that exist. It will be interesting to see whether fitting studies simulate correctly this sensitivity. In conclusion, trimethoprim and many analogues are able to discriminate in their binding between homologous proteins of similar (though not identical) conformation but of different amino-acid sequence, and the increased binding of certain analogues is associated with the acquisition of additional attractive interactions between compound and protein. (iv) Serine Proteases. Crystallographic studies on the homologous serine tryp~in,~~proteases--chym~trypsin,~~ and elastase 42-revealed that these endo- peptidases have a closely similar main-chain fold and a cleft in which various 39 D.A. Matthews and K. Volz, in ‘Molecular Structure and Biological Activity’ed. J. F. Griffin and W. L. Duax, Elsevier Science Publishing Co., New York, 1982, p. 13. 40 D. M. Blow, Acc. Chem. Res., 1976,9, 145. 4’ R. Huber and W. Bode, Acc. Chem. Res., 1978,11, 114. 42 (a)H. C. Watson, D.M. Shotton, J. M. Cox,and H. Muirhead, Nature (London),1970,225,806. (6)D. M. Shotton and H. C. Watson, Nature (London), 1970, 225, 81 I. Designing Drugs to Fit a Macromolecular Receptor Figure 18 Schematic illustration of the active site of E.coli dihydrofolate reductme with two analogues of trimethoprim. Each bears a carboxylate group, but the longer linkage between this group and the benzene ring in (45) (top) permits a closer interaction with the guanidinium group of arginine-57 for this compound than for (42) (bottom)(Courtesy of D. Baker) Beddell Table 4 The relative binding to E.Coli DHFR of analogues of trimethoprim in the absence (binary complex) and presence (ternary complex) of reduced cofactor (data from re$ 36) R Binary complex Ternary complex 1 .o 1 .o 0.8 0.5 3.4 3.5 6.7 37 7.7 20 16 54 7.7 26 0.1 2.8 1.7 1.5 0.7 peptide ligands bind. There exists a bovine pancreatic polypeptide of 58 residues which inhibits trypsin and chymotrypsin and the architecture of this inhibitor is also The inhibitor binds to, but is not hydrolysed by, the enzyme.However, if the disulphide bridge cystine- lkystine-38 in the inhibitor is reduced, trypsin 44 and chymotrypsin 45 hydrolyse the inhibitor between residues lysine- 15 and alanine-16. Blow and his colleagues46 used computer methods to fit the inhibitor to the enzyme, with the susceptible bond in the region of the conserved residues (serine- 195 and histidine-57, chymotrypsin numbering) which had been indicated from various studies to be directly concerned with hydrolysis. A unique fit was derived for the complex between inhibitor and chymotrypsin46 and trypsin 47 respectively. These predictions were confirmed and proved to be accurate when the crystal structure of the trypsin-inhibitor comple~~~,~’ was subsequently determined.The only apparent problem with the computed fit was the unduly close approach of the side chain of serine-195 to the main chain of the inhibitor in the region of lysine-15 and alanine-16. These contacts were compatible with those expected to precede the formation of a tetrahedral adduct between the serine-195 hydroxy-group and the carbonyl carbon of lysine-15, and distortion of this region was reported for the complex. This study illustrates the ability of skilled workers correctly to predict fit between macromolecule and ligand, by optimization of steric fit and complementarity. 43 J. Deisenhofer, and W. Steigemann, Acfa Crystallogr., Sect. B., 1975, 31, 238. 44 L. F. Kress, and M. Laskowski, J. Bid.Chem., 1967, 242, 4925. 45 M. Rigbi, in ‘Proc. Inst. Res. Conf. on Proteinase Inhibitors’, Walter de Gruyter, Berlin, 1971, p. 74. 46 D. M. Blow, C. S.Wright, D. Kukla,A. Ruhlmann, W. Steigemann, and R. Huber, J. Mol. Biol., 1972,69, 137. 47 R. Huber, D. Kukla, A. Ruhlmann, and W. Steigemann, Cold Spring Harbor Symp. Quant. Biol.,1971, XXXVI,141. 48 R. Huber, D. Kukla, W. Bode, P. Schwager, K. Bartels, J. Deisenhofer, and W. Steigemann, J. Mol. Biol., 1974, 89, 73. 49 A. Riihlmann, D. Kukla, P. Schwager, K. Bartels, and R. Huber, J. Mol. Bid, 1973, 77, 417. Designing Drugs to Fit a Macromolecular Receptor Hassall and his colleagues used the results of crystallographic studies on serine proteases to design 50 elastase inhibitors. The motive for the work was the possible use of such inhibitors to treat diseases where tissue degradation by elastase (e.g.pancreatitis, arthritis, emphysema) might be occurring. The study started from N- acylated derivatives of (Ala), or Ala-Pro-Ala. The working hypothesis assumed that these bound with the C-terminal residue in site S(1) adjacent to the active-site serine. By examining a physical model of the enzyme, these workers concluded that the C-terminal residue could be replaced by a carboxylate-free moiety. Substitution by N-cycloalkyl groups improved inhibitory potency, estimated as Ki, by up to nearly two orders of magnitude. However, the authors also indicate that the presumed binding sites S(l), S(2), S(3) for the respective residues, deduced by analogy with those known for trypsin and chymotrypsin, in fact differ from those observed when peptides are crystallographically observed complexed with elastase.The binding of two inhibitors from this study differs from that of the above peptides. Thus, in the case of elastase, which is a peptidase with a shallow cleft, the binding mode is evidently susceptible to modest structural changes in the ligand and direct observation of binding is a much-needed check on the validity of the binding-mode hypothesis which is being used to assist the choice of new structures for synthesis. C. Non-protein Macromolecules.-Nucleic acids are the only non-protein macromolecules reviewed here. Work on, for example, the guest-host complex with crown ethers and cyclodextrins and studies on macromolecular carbohydrates, though adding greatly to our understanding of complementarity, falls outside the scope of the present review.(i) Nucleic Acid. The Watson-Crick double-stranded right-handed B-helix model, for DNA, which is supported by the DNA fibre-diffraction data of Wilkins, Franklin, and co-workers, has hydrogen-bonding between the nucleotide bases, adenine with thymine and guanine with cytosine. It is considered to represent an important form of intracellular DNA, although other right-handed forms (e.g. A), left-handed forms (e.g. Z) and unpaired forms may also be functionally important. Some compounds bind to DNA by interaction with the backbone or by hydrogen-bonding with bases accessed via the minor or major groove.Most compounds which bind tightly to DNA possess a large polarizable aromatic polycyclic system which might in principle interpose in the stack of base pairs, pushing them apart along the helix axis by about 3.4A, the van der Waals thickness of the polycyclic system. Such an intercalation model was proposed over twenty years ago by Lerman (Figure 19), and since then there have been many attempts better to define the extent of DNA elongation, unwinding, base-sequence specific binding, and neighbourhood effects associated with intercalation and also the precise nucleotide geometries at and near the intercalation site. 50 C. H. Hassall, W. H. Johnson, and N. A. Roberts, Bioorg. Chem., 1979,8, 299. L. S. Lerman, J.Mol. Biol., 1961,3, 18. Beddell t Figure 19 Schematic illustration of the DNA duplex (left) and of the intercalation of a compound (shaded) according to the Lerman model. The DNA sugar-phosphate backbone is shown in black with the attached nucleotide bases unshaded discs (Courtesy of S. Neidle) In a recent review of intercalation Neidle 52 concludes: ‘It is now apparent that drug intercalation is a much more complex family of phenomena than was hitherto imagined. Structural studies have started to reveal the extent of these subtleties; many more are needed to clarify questions of, in particular, sequence specificity and intercalation into long sequences of oligonucleotide. Such analyses will undoubtedly enable rational drug design to be conducted to higher levels of sophistication than at the present time’.* Many drugs act at the DNA level (e.g.reviews 53754) and it is perhaps fortunate that the complexities have emerged only recently, as workers in this field might otherwise have been deterred from any attempts at designing drugs to fit DNA.Indeed, in a quite recent review of DNA-directed drug design, Henry 55 concluded that whilst little knowledge exists to support the design of tumour DNA-specific anti-tumour drugs, nevertheless active and novel DNA-directed compounds have been produced which might not have been considered in the absence of receptor concepts. For example, Kundu et al.56attempted to design a non-intercalating DNA binding compound. It was anticipated that a functionalized cyclo- pentaCflisoquinolin-3(2H)-one(52) would hydrogen-bond to a cytosine-guanine base-pair, as shown in Figure 20. This compound showed a low level of cytotoxicity and did bind weakly to calf DNA, with some apparent selectivity for cytosine- guanine base-pairs.* Reproduced by permission of the author 52 and Oxford University Press, from ‘A’-Ray Crystallography and Drug Action’, 1984, p. 129. 52 S. Neidle, in ‘X-Ray Crystallography and Drug Action’, ed. A. S. Horn, and C. J. De Ranter, Oxford University Press, 1984, p. 129. 53 S. Neidle, in ‘Progress in Medicinal Chemistry’, ed. G. P. Ellis and G. B. West, Vol. 16, Elsevier/North Holland Biomedical Press, Amsterdam, 1979, p. 151. 54 ‘Molecular Aspects of Anti-cancer Drug Action’, ed.S. Neidle and M. J. Waring, Macmillan Press Ltd., London, 1983. 55 D. W. Henry, in ‘New Approaches to the Design of Antineoplastic Agents’ ed. T. K. Bardos and T. I. Kalman, Elsevier Science Publishing Co., Amsterdam, 1982, p. 5. 56 (a)N. G. Kundu, J. A. Wright, K. L. Perlman, W. Hallett, and C. Heidelberger, J. Med. Chem., 1975,18, 395. (6)N. G. Kundu, W. Hallett, and C. Heidelberger, J. Med. Chem., 1975, 18, 399. Designing Drugs to Fit a Macromolecular Receptor CH2.C =O N’ k H’ ‘H H Figure 20 Postulated hydrogen bonding between (52) and DNA nucleotide bases The intercalation concept has been incorporated into design, especially in regard to the development of bifunctional intercalators. These are molecules with two planar polycyclic systems, linked with a spacer sufficiently long that the ring systems can both intercalate, at different sites in the DNA helix. Chen et aL5’ prepared bis-9- aminoacridines (53), the rings linked with chains of various lengths.Compounds with a chain length, n, of 6 and greater could double intercalate and showed cytotoxic and/or in uiuo anti-tumour a~tivity.~*,~’ However, the lack of correlation 57 T. K. Chen, R. Fico, and E. S. Canellakis, J. Med. Chem., 1978, 21, 868. 58 R. G. McR. Wright, L. P. G.Wakelin, A. Fieldes,R. M. Acheson, and M. J. Waring,Biochemistry, 1980, 19, 5825. 59 L. P. G. Wakelin, M. Romanos, T. K. Chen, D. Glaubiger, E. S. Canellakis, and M. J. Waring, Biochemistry, 1978, 17, 5057. 310 Beddell between cytotoxic and anti-tumour effects led to investigation of the anti-tumour mechanism 57,60 and revealed that this involved action at the cell surface.Me0 0 2 CI-Me Me Henry and his colleagues 61 designed the 12-aminobenz[zIphenanthridine (54) to provide substantial charge delocalization, on the grounds that such a property was a general feature of intercalating nitrogen heterocycles. This system also appeared from modelling studies to intercalate well and some mono-functional molecules were found to have significant anti-tumour activity, although bifunctional molecules were the most potent and efficacious. Chain lengths of 3 and 10were less effective than those in the range 7-9. Le Pecq and co-workers62 prepared the double intercalator (55) which cured a substantial percentage of animals in the L1210 murine leukaemia experimental anti-tumour system.Small changes in the structure often abolished activity; it may be important for activity in potential bis- intercalators for the structure to resist self-stacking of the aromatic parts. Daunorubicin and adriamycin are intercalative DNA binders used clinically as anti-cancer agents. Several detailed potential binding modes have been reviewed 53 and one was used for further development. In this model, the molecule intercalated. In the wide groove of the DNA double-helix three successive phosphate groups of one DNA chain interacted with, respectively, the 9-hydroxy-, the protonated 3’-amino- and the 4’-hydroxy-groups.The 9-acetyl group of daunorubicin was not involved in interactions, and thus it was felt that coupling of two daunorubicin molecules through the 9-position might generate an active bifunctional molecule. One such series of molecules is (56). Some molecules were indeed bifunctional intercalators and proved to have substantial anti-tumour activity. The crystallographic determination 64 of the structure of a complex between daunorubicin and a self-complementary hexanucleotide has now provided a detailed picture of one way in which such molecules might intercalate in DNA. 6o R. M. Fico, T. K. Chen, and E. S. Canellakis, Science, 1977, 198, 53. W. Fleming, M. Lerom, P. Sturm, C. Mosher, W. W. Lee,D. Taylor, and D. W. Henry, in preparation.D. Pelaprat, A. Delbarre, I. Le Guen, B. P. Roques, and J.-B. Le Pecq., J. Med. Chem., 1980,23, 1336. ”D. W. Henry, in ‘Cancer Chemotherapy’, ed. A. C. Sartorelli, ACS Symposia Series No. 30, 1976, p. 15. 64 G. J. Quigley, A. H.-J.Wang, G. Ughetto, G. van der Marel, J. H. van Boom, and A. Rich, Proc. Natf. Arad. Sci., USA, 1980, 77, 7204. 311 Designing Drugs to Fit a Macromolecular Receptor Other cases of bis-intercalator design are known (see refs. 65-49, for example, and references therein) but will not be reviewed here. Although the detailed models used in these various studies of the intercalation complex may in some instances be wrong in detail, it appears that even gross features inherent in such a complex are sufficient to point the way towards new and useful molecules. y42 OMe 3 Receptor Mapping The present review is concerned with ‘receptor fitting’, defined and exemplified above.It is the author’s experience that such fitting has commonly been confused with ‘receptor mapping’ and even with ‘receptor structuring’. To assist those following the literature a brief comparison of the terms is provided here (see also Gund 70). Receptor mapping is the process of deducing the structure and other properties of the receptor by studying the properties of small ligand ‘probes’, often themselves drugs, which interact to various degrees with &he receptor. Naturally some parts of any one ligand may not interact appreciably with the receptor. Those parts of the ‘’ M.M. Becker, and P. B. Dervan, J. Am. Chem. SOC.,1979, 101, 3664. 66 K. F. Kuhlmann, N. J. Charbeneau, and C. W. Mosher, Nucleic Acid Res., 1978,5,2629. 67 B. K. Sinha, R.M. Philen, R.Sato, and R.L. Cysyk, J. Med. Chem., 1977, 20, 1528. “B. F. Cain, B. C. Baguley, and W. A. Denny, J. Med. Chem., 1978, 21, 658. 69 J. W. Lown, B. C. Gunn, K. C. Majumdar, and E. McGoran, Can. J. Chem., 1979,57,2305. ’O P. Gund, Trends Pharmacol. Sci.,1982, 3, 56. Beddell several ligands which do interact will provide good structural clues and are collectively used to define the ‘pharmacophore’. Thus, part of the receptor- mapping approach involves establishing which parts of ligands comprise the pharmacophore. It is then necessary to deduce the conformation of the receptor- bound ligands and their mutual orientations.The ligands are then considered in superposition and receptor groups are postulated in the space around the ligands in a way which would give rise to interactions between groups in the receptor model and groups in the ligands compatible with the experimental observations of the interaction (as manifest in, for example, binding, agonism, or antagonism). The approach is well illustrated by recent work employing graphics as an aid to three- dimensional repre~entation.~ Receptor structuring is the direct determination of receptor structure with wave/particle probes (e.g. X-rays, neutrons, and electrons in crystallography, radiofrequency electromagnetic radiation in n.m.r., electrons in electron micro- scopy).X-Ray crystallography has been especially valuable in providing a detailed and comprehensive picture of macromolecular architecture. Thus receptor mapping and receptor structuring are, respectively, indirect and direct ways of establishing receptor structure. Receptor mapping is also related in a complementary way to receptor fitting and perhaps it is this complementarity which causes confusion. The receptor-fitting process predicts how a ligand of known chemical structure would interact with a receptor of known architecture; receptor mapping predicts how such ligands would interact with postulated models of the receptor. In receptor fitting, therefore, it is primarily the small molecule which is adjusted to fit the large; in receptor mapping there is much adjustment of the receptor model to fit the ligands.Naturally, considerable skill is needed in both procedures, and in the latter it is particularly difficult to establish internally the accuracy and uniqueness of the model. There may be numerous potential models for the receptor that would lead to predicted interactions with ligands which are in accord with the measurements of the interactions. The predictions will also be sensitive to the assignment of the pharmacophore, conformation, and relative orientation for each ligand. In principle these could all be established by a direct determination of the architecture of the complex between each ligand and receptor. Then ligand conformation and disposition are observed directly and interaction between parts of the ligand and receptor are inferred by proximity and used to define the pharmacophore.From this it would then be possible to map the receptor and compare the map with the directly observed structure of the receptor. Indeed there are already some studies with DHFR 74-76 and with prealb~min,~~ ‘receptors’ of known structure, which are leading towards the production of receptor models by receptor mapping. Similar work, if genuinely performed 71 G. R. Marshall, in ref. 1, p. 129. 72 W. E. Klunk, B. L. Kalman, J. A. Ferrendelli, and D. F. Covey, Mof.Pharmacof., 1983, 23, 51 1 73 J. R. Sufrin, D. A. Dunn, and G. R. Marshall, Mof.Pharmacol., 1981, 19, 307. 74 G. M. Cnppen, J. Med. Chem., 1980,23, 599.75 A. K. Ghose, and G. M. Crippen, J. Med. Chem., 1983,26,996. ”A. K. Ghose, and G. M. Crippen, in ref. 1, p. 99. 77 G. M. Cnppen, J. Med. Chem., 1981,24, 198. 313 Designing Drugs to Fit a Macromolecular Receptor without recourse to crystallographic information on such macromolecules, could be used to evaluate different receptor-mapping methods through comparison with the rapidly increasing structural information. Some idea of the difficulty of meaningful receptor mapping may be gained from the studies of H~pfinger,~~-~~ in which protein X-ray crystallographic and n.m.r. information was explicitly ignored to enable the reliability of indirect methods to be ultimately established. From partially hindered analogues of trimethoprim, taken together with enzyme inhibition data, a conformation for trimethoprim in its complex with E.coli DHFR was inferred and used to calculate shape and potential energy parameters for other analogues, which in turn were tested for correlation with bovine DHFR inhibition data.However, the deduced conformation for trimethoprim in the E.coZi DHFR, as judged by the torsion angles quoted, was not the same as that deduced crystallographically. Similarly n.m.r.'O data for trimethoprim in bacterial or in vertebrate DHFR failed to confirm the calculated conformation. The difficulty of assigning correctly the bound conformation by indirect methods is indeed substantial for most drugs, since these are commonly flexible molecules, and suitable rigid or semi-rigid analogues of most are not available.Establishment of mutual frames of reference may also not be straightforward. The substrate for DHFR differs from such 4-amino-containing inhibitors as aminopterin and (37) by virtue of its 4-0x0-group. There is now good eviden~e~'*~~-~~ that this small difference causes a 180" difference in orientation of the pteridine ring in the DHFR complex. Thus it is clearly difficult to establish indirectly the receptor-bound conformations of individual ligands and to establish their mutual orientations, and the receptor-mapping technique, being crucially dependent on such knowledge, is liable to proceed in error. 4 Characteristics of the Receptor-Fit Approach A. Form of Relationship.-In the receptor-fit approach the derivation of a relationship between properties of the compound and the biological effect occurs in two steps.First, the properties of groups in the compound in conjunction with corresponding properties for the receptor are used in the estimation of the interaction between compound and receptor. To do this, physical or computer- based models of the molecules are manipulated, the interaction being calculated by means of an assumed relationship between interaction strength and molecular properties. The relationship used varies from guesswork, through simple additivity rules, to molecular mechanical and quantum mechanical formulations. The second step relates the calculated interaction to the biological/biochemical effect and will ''A.J. Hopfinger, J. Med. Chem., 1981, 24, 818. 79 A. J. Hopfinger, J. Med. Chem., 1983,26,990. B. Birdsall, G. C. K. Roberts, J. Feeney, J. G. Dann, and A. S. V. Burgen, Biochemistry, 1983,22,5597. P. A. Charlton, D. W. Young, B. Birdsall, J. Feeney, and G. C. K. Roberts, J. Chem. Soc., Chem. Commun., 1979,922. 82 J. C.Fontecilla-Camps, C. E. Bugg, C. Temple, J. D. Rose,J. A. Montgomery, and R.L. Kisliuk, J. Am. Chem. Soc., 1979, 101,6114. 3 14 Beddell of course depend upon the nature of the effect and upon the mechanisms operating in the macromolecular system. Thus for DHFR, inhibitory potency has been related to binding constant (and hence to binding free-energy) by various simple relationship^.^^ For haemoglobin, the two-state mechanism and refinements generate rather more complex equations; nevertheless, from these, a measure of effect, such as a change in the partial pressure of oxygen at a selected degree of saturation, or fractional liberation of oxygen at fixed partial pressure, can be related to the individual affinities of the compound to each of two conformational states.84 There is then, with adequate knowledge of the mechanism by which the macromolecule contributes to effects, ‘a priori’ reason to anticipate the form of relationship.Such ‘apriori’ reasoning is of course also encountered in situations besides molecular fitting. If, for example, partitioning of a compound through lipid is important, a linear free-energy relationship involving partition coefficient may be apparent, or if species of different charge partition and pK, is important, a linear free energy relationship involving Hammett’s CJ might be sought.However, in most instances the form of relationship cannot be anticipated ‘a priori’, as the mechanisms by which the drugs interact in the biological system are not understood in detail. B. Multiplicity of Properties-Many types of physicochemical property might conceivably be involved in producing effects in biological systems, and even more parameters are needed to describe properties which vary as a function of position (position-dependent scalars) and also of direction (position-dependent vectors). Only a very small proportion is ever considered in practice. The primary hope in classical P.A.R.studies is that one or two properties only will be of importance in at least some closely defined region of property-activity space.In the receptor-fitting studies described above, a number of qualitatively different properties, e.g. flexibility, covalent bond geometry, steric repulsion, hydrophobicity, hydrogen- bond potential, and ionic properties, are considered concurrently in a manner which makes allowance for positional dependence and vectorial nature. A drawback is that this has largely been done through interaction with physical models and therefore is subject to problems of incompleteness, some subjectivity, and imprecision. However, this situation is now changing with the introduction of a combination of graphics-docking procedures to generate trial fits and subsequent molecular mechanics energy minimization procedures for the estimation of binding enthalpy.With the aid of modern computer hardware and software, it is feasible to attempt some estimate of the contribution to binding of several different properties, each assigned at numerous different positions in the small molecule. The molecular mechanics approach employed by Kollman and co-workers has been mentioned. This approach has been much used by Scheraga,” Warshel,86 83 D. P. Baccanari, S. Daluge, and R. W. King, Biochemistry, 1982, 21, 5068. 84 P. J. Goodford, J. St-Louis, and R. Wootton, Br. J. Pharmacol., 1980,68, 741 85 M. R. Pincus, and H. A. Scheraga, Acc. Chem. Res., 1981,14,299. 86 A.Warshel, Acc. Chem. Res., 1981, 14, 284. 315 Designing Drugs to Fit a Macromolecular Receptor North 87,88 and co-workers to study protein-ligand interactions and recently also by Neidle 52 and Kollman 89 to study drug-nucleotide complexes. The explicit neglect of entropy remains a potential problem, although where correlations between free energy and interaction enthalpy have been found, entropy neglect has presumably not been a confounding factor. The nature of the algorithms and parameters used in the molecular mechanics approach at present are simple. Sophisticated treatments, taking into account in a more precise way, for example, the influence of solvent, and also property changes associated with polarization, are under study.In some methods of a statistical (Monte-Carlo) or dynamical nature, the estimation of entropy becomes approachable. C. Superposition.-The mutual alignment of different ligands emerges naturally from the process of fitting to the macromolecule. Thus, if several potential fits of a given compound are generated interactively and that which is calculated to bind with minimum binding-enthalpy selected as the most relevant, then the position and orientation of each compound in turn is established in a single frame of reference based upon the macromolecule. This consequently defines the super- position of the various compounds as a by-product. In principle, the alignment problem is solved in this procedure, although it is clear that, owing to the limited radius of convergence of minimization methods, and their inability to overcome large energy barriers in the manner in which they have so far been used in this type of work, it is unlikely that the true fit would be discovered if one fairly close to it in nature were not generated in the interactive docking process.For success therefore, the method requires substantial skill. As mentioned previously, it has been established that the roughly isoelectronic and isosteric substitution of amino for 0x0 at the 4-position of pteridine derivative ligands of DHFR inverts the orientation of the bound pteridine. Hence one must be alert to the possibility of substantial differences in orientation arising from local changes in the structure of the ligand.Although superposition is a byproduct, it is of course true that the receptor itself may adopt different conformations for each compound, and compounds may bind at different sites. The whole concept of superposition underlying classical P.A.R. may be misleading when one attempts to superimpose drugs in order to define properties in common, if the drugs bind in different molecular environments. In these circumstances receptor fitting, being independent of assumptions concerning congruence amongst molecules, may be more robust than P.A.R. methods based on superposition. This fundamental difference of course arises because classical P.A.R. methods are essentially correlation methods, dealing with what the biological system does in response to different drugs. Receptor fit is one of several mechanistic approaches that use physicochemical knowledge of the system, i.e.how it operates, to predict response. ''S. B. Brown, A. A. Chabot, E. A. Enderby, and A. C. T. North, Nature (London), 1981,289,93.''E. A. Potterton, in 'Computer Modelling of DHFR-Ligand Interactions', PhD Thesis, Leeds University, 1983.''A. Dearing, P. Weiner, and P. A. Kollman, Nucleic Acirls Res., 1981.9, 1483. Beddell D. Finding the Local Best Compoud-Consideration of the receptor structure focuses attention on drug action through certain molecularly defined mechanisms. Binding of the drug to the receptor is a prerequisite and binding may be sterically competitive with a natural substrate or regulator, or may selectively favour allosterically a pre-existing or induced state of the receptor.The binding process involves complementarity between ligand and receptor and thus at each position in space in the vicinity of the receptor there is, in principle, an assignable property which must be possessed by the drug for complementarity to exist. The problem then is to optimise complementarity by changing the compound structure, and this is currently done by intelligently devising new structures with locally improved properties. In principle this structural search could be done by computer- automated structure generation with matching to a property field, but the computational problem is not simple in the general case and has yet to be accomplished.However, the illustrative studies described above show that, as with the classical P.A.R. approach, it is not too difficult to select compounds with improved properties, once it is known which properties matter. Furthermore, when structures are generated and fitted, the enthalpy calculation can now provide a relative index of fit and thus a numerical prediction of effectiveness. It is too early to show how reliable this measure is, but preliminary indications are that even the very simple approach based on the totting up of interactions can be useful, so the more detailed calculations of modem molecular mechanics should not fall short of this. E. Generating Structuresin New Areas of Property Space.-In exploring new types of structure, the receptor-fit approach provides guidance and potential binding sites can be explored to which there is no known ligand. The fitting of compounds to such sites requires skill and imagination and will be quite open ended if a great many different types of structure would in fact interact favourably at the site.The medicinal chemist does not need to try to explore all the possibilities however, since a level of fit can be established for a structure which, once achieved by modelling, would establish that structure as the starting point of a new series. The criteria might be based on certain minimum numbers of covalent, ionic, hydrogen-bond, and hydrophobic interactions between compound and protein, or might be based on a calculated binding enthalpy. Potential structures can be quickly eliminated if, for example, during the attempted fit, two groups of like charge or any two atoms are obligatorily so close, that a large repulsive term would be generated.Those not so eliminated might be ranked on, say, estimated binding enthalpy or estimated ease of synthesis. Thus receptor fit differs from classical P.A.R.in that it is well able to generate starting points for optimization in new regions of property space. F. Limitations.-A major limitation of receptor fit at the present time is its scope and lead time. Of the dozens of macromolecules that have now been structured, few have been seen as having potential relevance in therapy. To embark on the ‘denovo’ structuring of a macromolecule demands faith that it is a relevant and important molecule for study.Facilities for the following tasks would be essential: Designing Drugs to Fit a Macromolecular Receptor preparation of substantial quantities of the pure material; determination of residue sequence by direct analysis of protein or nucleic acid; preparation of crystals; structural determination, in general by the preparation of isomorphously prepared crystals containing specifically located heavy atoms. The biochemical procedures for the isolation of protein present in small amounts in organs and tissues have benefitted from selection procedures that induce larger levels of material in cell lines or micro-organisms, and affinity chromatography is now of immense value in the fractionation process.Amplified production by genetic engineering is likely to have substantial impact upon the production of macromolecules by the end of this century. In a few instances, there may be time-saving alternative procedures. In favourable cases the amino-acid sequence can be largely determined from the electron density map determined crystallographically, although there will be numerous instances in which residues will be assigned ambiguously-valine and threonine, for example, are isosteric, and could also resemble aspartic acid and asparagine if there were disorder. Conversely, if the amino-acid sequence of target protein and the crystal structure of a homologous protein are known, a model .of the target protein may be built by intelligent superposition upon the known structure.Another limitation of receptor fit is that it directs attention intrinsically to one facet of the overall process, but it is well known that a compound effective at a receptor in uitro may be less effective in uiuo. The compound may be unfavourable in respect of absorption, distribution, metabolism, or elimination in the more complex bioassays involving cells, micro-organisms, organ systems, and animals. Hence there must in general be a complementary effort, involving P.A.R. at the various levels of biochemical and biological assay, which identifies properties in the compound that come to dominate in the less purified systems. 5 Concluding Remarks Targeted approaches to drug generation have been considered conceptually for many years.Ehrlich, in his search for synthetic drugs early in the present century, came to adopt the belief that specific receptors must exist. His belief was in part founded on Langley’s research on alkaloid^.'^ Advances in our knowledge of function, structure, and mechanism for biological macromolecules arising in part from the advent of new techniques in biochemistry and biophysics have progressively made targeted approaches more realistic. However it is clear that in the past the untargeted approach has been the main influence in the generation of new drugs.9 1*92 The synthesis of diverse analogues of natural biochemicals, and the study of these in diverse biological assays, have been the source of many of today’s drugs.Productive research has often depended upon skilled choice and use of biological assays which allow the signzjicant unexpected event to be spotted, assessed, and exploited, in a manner reminiscent of that associated with the 90 J. Parascandola, Trends Pharmacol. Sci., 1980, 1, 189. 91 T. A. Krenitsky and G. B. Elion, ‘Strategy in Drug Research’, ed. J. A. K. Buisman, Elsevier Scientific Publishing Co.,Amsterdam, 1982, 65. 92 R. Maxwell in ‘Drug Development Research, Alan R. Liss, Inc., New York, in press. Beddell identification of penicillin by Fleming, Florey, and Chain. The tailoring of drugs to cellular targets can of course be approached by correlative methods such as classical P.A.R., but it is now possible to augment this by the receptor-fit method, which provides P.A.R.information at the molecular level through knowledge of the physical chemistry of molecular processes. With such an historical background, conversion to new and relatively untried technology, such as that described in the present review (which deals with only certain aspects of the overall problem) involves commercial risk. However, to ignore new technologies is to give up the hope for a more predictable approach. In practice, new methods are explored tentatively to see what they have to offer and only much later can the main attractions and defects of a new approach to the generation of drugs be assessed. For the macromolecular-fitting approach, it would seem premature to attempt to make such an assessment, as the methodology is little developed.The structural information which provides its foundations must be interpreted with care.93 Even so, it is clear that in some studies this approach has already been of use. Such successes encourage those in the field to continue to develop and apply the technique and with the advanced computer graphics and energy computation methods now under development, rapid progress can be expected. Acknowledgements. I appreciate the interest shown by Professor C. R. Ganellin and by the editorial board which led to this review. I am indebted to Drs. J. N. Champness, L. Kuyper, B. Roth, D. K. Stammers, and R. Wootton for their helpful criticism of the manuscript and to Professors P. R. Andrews, D. M. Blow, F. R. S., and G. R. Marshall, and Drs. J. Blaney, P.J. Goodford, D. W. Henry, R. Maxwell, S. Neidle, S. J. Oatley, G. C. K. Roberts, and L. Sawyer for their advice and information. 93 P. J. Goodford, J. Med. Chem., 1984,27, 557.

ISSN:0306-0012    

DOI:10.1039/CS9841300279

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

CENTENARY LECTURE* Molecular Ingredients of Heterogeneous Catalysis By Gabor A. Somorjai MATERIALS AND MOLECULAR RESEARCH DIVISION, LAWRENCE BERKELEY LABORATORY AND DEPARTMENT OF CHEMISTRY, UNIVERSITY OF CALIFORNIA, BERKELEY, CALIFORNIA 94720, USA 1 Introduction When Professor Thomas asked me to join up with him for a symposium on catalysis, I was happy to oblige. Catalysis science stands on two legs, surface science and the solid state chemistry of metastable materials (Figure 1).In the last ten years a scientific revolution has occurred that has converted catalysis from art to science. From the surface science side, this has been due to atomic level characterization of model, low surface area and real, high-surface area catalysts. As a result, the preparation of catalysts and their application in the chemical and petrochemical technologies have become science driven.That is, the molecular level under- standing that is obtained by research in the laboratory controls the development of chemical or petrochemical catalysis-based technology. Two examples of recently developed high technology catalysts are the catalytic converter and the new generation of zeolites (Figures 2 and 3). These could not have been developed Figure 1 Schematic representation of the close relationship between the materials chemistry of metastable high surface area materials, surface science, and catalysis science * Delivered at the Scientific Societies’ Lecture Theatre, London, November 7th 1983. 32 1 Molecular Ingredients of Heterogeneous Catalysis Figure 2 The catalytic converter for the complete combustion of automobile exhaust.It oxidizes unburned hydrocarbons, and carbon monoxide to carbon dioxide and water, and reduces nitrogen oxides to dinitrogen simultaneously without the molecular level characterization of their structure and composition by electron spectroscopies and solid-state n.m.r. that lead to the establishment of the all important correlation between their atomic structure and composition and performance. Let me review this recent development of the molecular science of heterogeneous catalysis from the surface science side, where much of my research is concentrated, and perhaps also show the direction the field might take in the near future.My investigations utilize mostly small area single crystal catalysts and use the full repertoire of surface science techniques for their characterization. The most ZSM-5 Channel Structure ZSM-11 Channel Structure Figure 3 Two recently synthesized zeolites with large silicon to aluminium ratios Somorjai frequently used tools for surface characterization include low energy electron diffraction (LEED), Auger electron spectroscopy (AES), and high resolution electron energy loss spectroscopy (HREELS).’ The work horse for the catalytic studies was the low-pressure, high-pressure apparatus,2 shown in Figure 4. After appropriate characterization the sample is enclosed in an isolation chamber that can be pressurized to atmospheric pressures (up to 100atm) with the reactant gases.These can be circulated by a pump and when the sample is heated to the reaction temperatures, the rate and the product distribution can be monitored with a gas chromatograph. The high pressure loop acts as a microreactor that can be operated in the flow or in the batch modes. When the reaction is complete, the isolation cell is opened and the sample is now in an ultra-high vacuum environment again. After performing surface analysis to identify the surface composition and structure that was present under the reaction conditions, the surface may be modified and the high pressure reaction study can be continued. Pressure gauge ‘I;ampling valve :hromatograph Gas introduction To mechanical pump needleIt Welded metal To gas‘Hbellows pump manifold Figure 4 Schematic representation of the experimental apparatus utilized to carry out the catalytic reaction rate studies on single crystal or polycrystalline surfaces of low surface area at low and high pressures in the l@’ to torr range When using model catalysts it is essential to establish if close similarity exists between the behavior of the model and the real catalyst systems.This was found to be the case for several reactions that include the hydrogenation of carbon G. A. Somorjai, ‘Chemistry in Two Dimensions: Surfaces’, Cornell University Press, 1981, Library of Congress Catalog Card No. 80-21443. A. L. Cabrera, N. D. Spencer, E. Kozak, P. W. Davies, and G.A. Somorjai, Rev. Sci. Instrum., 1982,53, 1888. Molecular Ingredients of Heterogeneous Catalysis monoxide over rh~dium,~ and the ring opening the hydrogenation of cy~lohexene,~ of cyclopropane.’ We investigated in some detail several important catalytic reactions, including the conversion of hydrocarbons over Pt crystal the synthesis of ammonia over iron and rhenium lo crystals, the hydrogenation of carbon monoxide over iron and rhodium,” and the photodissociation of water over TiO,, SrTiO3,I3 and iron oxides.14 From these studies three molecular ingredients of heterogeneous catalysis were identified. These are: The Atomic Structure of the Surface The Carbonaceous Deposit The Oxidation State of Surface Atoms We shall discuss each one of these ingredients and give examples of their roles in catalytic surface reactions.2 Atomic Surface Structure Figure 5 compares the rates of ammonia production on three single crystal surfaces of body centered cubic iron at high pressures. The (1 11) face is about 430-times more active than the closest packed (110) crystal face while the (100) face is 32- times as active as the (110) face. The rate limiting step in this reaction is the dissociation of N, and it appears that this process occurs with a near zero activation energy on the (11 1) iron surface while there is a larger activation energy on the other iron crystal surfaces. It has been proposed that the active site for breaking the very strong N, bond is a seven co-ordinated iron atom that is present in the second layer under the surface.There is a theory being developed that relates the concentration of nearly degenerate electron vacancy states, the density of hole states near the Fermi level, to the ability of a given site to break and make chemical bonds in a transient manner by charge fluctuations. The sites with the largest number of nearest neighbours (highest co-ordination) have the highest density of electron-hole states and thus, they should be the most active during catalytic reactions. Unfortunately they are located in the bulk and are not accessible to the incoming reactants. However, atoms in the second layer of an open surface structure are accessible but are still surrounded by a large number of neighbours.These are then the most active sites in many catalytic reactions. B. A. Sexton and G. A. Somorjai, f. Catal., 1977,46, 167. S. M. Davis and G. A. Somorjai, f.Catal., 1980, 65, 78.’D. R. Kahn, E. E. Petersen, and G. A. Somorjai, f.Catal., 1974,34, 294. W. D. Gillespie, R. K. Herz, E. E. Petersen, and G. A. Somorjai, f. Catal., 1981, 70, 147. ’S. M. Davis, F. Zaera, and G. A. Somorjai, f.Am. Chem. Soc., 1982,104, 7453. a S. M. Davis, F. Zaera, and G. A. Somorjai, f.Catai., 1984,85, 206. N. D. Spencer, R. C. Schoonmaker, and G. A. Somorjai, f. Catal., 1982,74, 129. lo N. D. Spencer and G. A. Somorjai, J. Catal., 1982, 78, 142. I’ D. Dwyer and G. A. Somorjai, J. Catal., 1979,56, 249. D. G. Castner, R. L. Blackadar, and G. A. Somorjai, f.Catal., 1980, 66, 257.l3 F. T. Wagner and G. A. Somorjai, J. Am. Chem. Sor., 1980,102, 5494. l4 C.Leygraf, M. Hendewerk, and G. A. Somorjai, f.Catal., 1982,78, 341. I’ L. Falicov and G. A. Somorjai, Proc. Natl. Acad. Sci.,1984, to be published. Somorjai (111) C4 (SOLID) C, (SOLID) C, (DOTTED) C, (DOTTED) looor 500- (II I) 100- 50- (100) 10 - 5- Figure 5 The remarkable surface structure sensitivity of the iron catalysed ammonia synthesis Although this theory will have to be tested further and proven by careful experiments, the available experimental data on the structure sensitivity of catalytic reactions can be explained by it. Figure 6 shows the rate of ammonia formation from N, and H, over hexagonal close-packed rhenium crystal surfaces.I6 Again the open (1120) crystal face is about 103-times more active than the closest packed (0001) hexagonal surface, thus exhibiting a profound structure sensitivity that is even more pronounced than that on iron.For catalysed hydrocarbon reactions on platinum, which is an excellent catalyst, M. Asscher and G. A. Somorjai, SurJ Sci., 1984, in press. Molecular Ingredients of Heterogeneous Catalysis Re C,(SOLID) C6(SOLID) C, (SOLID) C, ,(DOTTED) Cl0 (DOT T ED) 10,000 5OOOF I000 500 Figure 6 The structure sensitivity of ammonia synthesis on rhenium single crystal surfaces there are four crystal surfaces with very different atomic surface structures that exhibit very different reaction selectivities.These are the flat (1 11) and (100) surfaces that have hexagonal and square unit cells, respectively. The other two surfaces have ordered steps of atomic heights and one has ledges (or kinks) in the steps as well. These are shown in Figure 7. The test reactions that best demonstrate the structure sensitivity over Pt are the conversions of n-hexane and n-heptane into other organic molecules.6 n-Hexane may be converted into benzene upon dehydrocyclization or into methycyclopentane by a cyclization reaction. These are shown in Figure 8. It may isomerize to branched butanes or undergo C-C bond Somorjai fcc (1001 fcc (1111 fcc (775) fcc (10.8.71 Figure 7 Idealized atomic surface structures for theflat platinum (1 1 1) andplatinum (lo),the stepped platinum (775)and kinked platinum (10,8,7)surfaces breaking (hydrogenolysis) to produce C1-C3 fragments (methane to propane). The first three of these reactions are desirable when the aim is to produce high octane gasoline while the fourth reaction is undesirable as it leads to the production of gases of much less value as fuels.Figure 9 shows that the hexagonal surface produces much more aromatic product than the square (100) crystal face. In fact, a stepped surface with (1 11) orientation terraces that are five atoms wide is perhaps the best catalyst we have found so far to carry out the dehydrocyclization reaction. Conversely, the (100) flat surfaces with the square unit cells are much better isomerization catalysts, as shown in Figure 10, than the hexagonal crystal surface of Pt.’ Thus, depending on the catalyst preparation, one may obtain superior dehydrocyclization or isomerization activity that is certainly well documented in the patent literature. The hydrogenolysis reaction that is also shown in Figure 10is most active on surfaces that contain a large concentration of ledge or kink sites.* It is often necessary to ‘poison’ these sites by the adsorption of sulphur or other strongly bound additives that bind more strongly to the ledge sites than to the other surface sites (step or terrace sites).This way, the ledges cannot participate in Moiecular Ingredients of Heterogeneous Catalysis I CH AROMATIZATION 't4 ;LrHc+c H I I I H2y ,CH CYCLIZATION * H2C,, %H-CH3 + H2 CH2 Platinum r CH3In -hexane ISOMERIZATION I 7 I I propane Figure 8 Skeletal rearrangement reactions of hydrocarbons catalysed by platinum with high activity and unique selectivity. Depicted here are the several reaction pathways which occur simultaneously during the catalysed conversion of n-hexane, C,H ,4.The isomerization, cyclization, and aromalizalion reactions that produce branched or cyclic products are important in the production of high octane gasoline from petroleum naphtha. The hydrogenolysis reaction that involves breaking of C-C bond yields undesirable gaseous products hydrocarbon reactions because they are masked by the selective adsorption of additives while the rest of the higher co-ordination surface sites remain clean and thereby active and selective.3 The Carbonaceous Deposit A catalytically active metal surface is always covered with a carbonaceous deposit. By labelling the reactant organic molecules with 14Cisotope, the residence time of this carbonaceous layer can be m~nitored.'~ It is found that it is usually ten- to fifty- times larger than the turnover time for the catalytic hydrocarbon conversion reactions. This is shown in Figure 11 under the label of the irreversible adsorption along with the hydrogen to carbon ratio of this deposit. As the reaction temperature is increased, the deposit becomes successively dehydrogenated as its stoicheiometry varies from C2H, to C2H and finely it looses all its hydrogen, and becomes graphitic.The metal surface retains its catalytic activity as long as the S. M. Davis, F. Zaera, and G. A. Somorjai, J. Cat& 1982, 77, 439. Somorjai STRUCTURE SENSITIVITY OF ALKANE AROMATIZATION n6 Q)v) E t0 n- hexane benzene n- heptone toluene O 10-d 3 1 a--0 -E Y wF -a 5--' a z I! --I-u b a r w Aa O-'HEXAGONAL SQUARE HEXAGONAL SQUARE Figure 9 Dehydrocyclization of alkanes to aromatic hydrocarbons is one of the most important petroleum reforming reactions. The bar graphs shown here compare reaction rates for n-hexane and n-heptane aromatization catalysed at 573 K and atmospheric pressure over the two $at platinum single crystal faces with different atomic structure.The platinum surface with the hexagonal atomic arrangement is several times more active than the surface with a square unit cell over a wide range of reaction conditions STRUCTURE SENSITIVITY OF LIGHT ALKANE SKELETAL REARRANGEMENT >--c, -c3gases -0.03 cisobutane n- butane isobutane methane, ethane, propane a, 0.2 -0.02E v 11I w I-a oz 0.1 -0.01 z 0 k-0 a W (L 0-0 PLATfNUM SURFACE STRUCTURE Figure 10 Reaction rates are shown as a function of surface structure for isobutane isomerization and hydrogenolysis catalysed at 570 K at atmospheric pressure over four platinum surfaces. The rates for both reaction pathways are very sensitive to structural features of the model single crystal catalytic surfaces.Isomerization of these light alkanes favoured on the platinum surfaces that have a square (100) atomic arrangement. Hydrogenolysis rates are maximized when kinked sites are present at high concentrations, as in the platinum (10,8,7) crystal surface Molecular Ingredients of Heterogeneous Catalysis carbonaceous deposit contains hydrogen, but becomes completely inactive (poisoned) in the presence of the graphitic overlayer. Hydrogen Content %,H,/Pt (Ill1 C .-0 a, L w 0 200 400 Ad so r pt ion Temperature CC 1 Figure 11 Carbon-14labelled ethylene C2H, was chemisorbedas afunction of temperatureon aflat platinum surface with hexagonal orientation, Pt( 111). H/C composition of the adsorbed species was determinedfrom hydrogen thermal desorption studies. The amount of pre-adsorbedethylene, which could not be removed by subsequent treatment in hydrogen at atmospheric pressure represents the irreversibly adorbed fraction.The adsorption reversibility decreases markedly with increasingadsorptiontemperatureas the surface species becomes more hydrogendeficient. The irreversibly adsorbed species have very long surface residencetimes of the order of hYS The sequential dehydrogenation of adsorbed organic monolayers with in-creasing temperature can be readily demonstrated by temperature programmed thermal desorption studies. Figure 12 shows the evolution of hydrogen from adsorbed layers of C,H,, C,H,, and C,H8.18 At well-defined temperatures, hydrogen evolves at a maximum rate until complete dehydrogenation and graphitization of the remaining carbon occurs at the highest temperatures. LEED and HREELS studies reveal the structure of organic monolayers at each stage of chemisorption.At lower temperature (< -300K) the organic molecules exhibit ordered molecular structures. Figure 13 shows one of the ordered surface structures of benzene l9 on the Rh( 11 1) crystal face that was determined by LEED and Figure 14 shows the HREELS spectra of benzene and its deuterated form.*’ M. Salmeron and G. A. Somorjai, J. Phys. Chem.. 1982,86,341. l9 M. A. Van Hove, R. Lin, and G. A. Somorjai, Phys. Rev. Leti., 1983,51, 778. B. E. Koel, J. E. Crowell, C. M. Mate, and G. A. Somorjai, J. Phys. Chem., 1984, in press.330 Somorjai 0.5 L ALKENE ON Pt (111) .. ....pH2 ...... a.....a. .. a.-.:... 112.. 1 1 I I I I I I 200 400 600 800 T (K) Figure 12 Hydrogen thermal desorptionspectra illustrating the sequential dehydrogenation of ethylene,propylene, andcis-Zbutene chemisorbed on R(ll1)at about 120 K (theheating rate is 12 K per second) The ClVsymmetry is clearly compatible with the molecular structure shown in Figure 13 with the molecule lying with its R ring parallel to the surface and the centre of the ring above a three-fold hollow. Figure 15 shows the surface structures of chemisorbed ethylene, propylene, and butene on the Pt(111)crystal face.21These molecules form alkylidyne species upon adsorption near 300 K with the C-C bond that is closest to the metal surface, perpendicular and elongated to a single C-C bond length.Similar alkylidyne structures have been found on other transition-metal surfaces, including Pd, Rh, and Ni.22 The carbon atom that binds the molecule to the metal prefers the three-fold hollow site. Figure 16 compares the molecular structure of the ethylidyne molecule on the Pt(ll1) surface with a structure of ethylidyne containing trinuclear metal cluster compounds. The symmetry, the bond distances, and the bond angles in these clusters are very similar to the molecular structure of chemisorbed ethylene on the transition-metal surface. This similarity indicates the predominance of localized bonding of adsorbed sur-21 R.J. Koestner, J. C. Frost, P. C. Stair, M. A. Van Hove, and G. A. Somorjai, Surf. Sci., 1982, 116,85. 22 R.J. Koestner, M. A. Van Hove, and G. A. Somojai, Surf. Sci., 1982,121,321. 33 1 Molecular Ingredients of Heterogeneous Catalysis Rh(lll)-(: C6Hg Figure 13 The surface structure of benzene as determined from low energy diffraction studies and surface crystallography face species, an important conclusion in our scrutiny of the surface chemical bond.23 23 G.A. Somorjai, ‘Proc.9th Intl. Conf. on Atomic Spectroscopy’, XXII CSI, Tokyo, June 1982, ‘Recent Advances in Analytical Spectroscopy,’ ed. K. Fuwa, Pergamon Press, 1982, p. 21 1. Somorjai x600 300012 (601 C48cni' " I I I 0 1000 2000 3000 ENERGY LOSS (cm-7 Figure 14 The vibrational spectra of benzene and deuterated benzene as determined by high resolution electron energy loss spectroscopy Figure 17 shows the sequential change of the vibrational spectrum of chemisorbed C2H4 on the Rh( 1 11) crystal face as the temperature is increased.24 The molecule decomposes and there is evidence for the presence of CH, CZ, and CZH species on the surface in the spectra.Figure 18 shows schematically many of these species that were detectable by HREELS (not by LEED because these 24 B. E.Koel, J. E. Crowell, C. M. Mate, and G. A. Somorjai, J. Phys. Chem., submitted for publication (1 984). Molecular Ingredients of Heterogeneous Catalysis rpropylidy ne ethyIidyne Pt (Ill) + ethylidyne, propylidyne and butylidyne Figure 15 Surface structures for alkylidyne species formed on platinum (1 11) after the ahorption and rearrangement of ethylene, propylene, and butenes.These structures were determined by LEED surface crystallography fragments are disordered) and also -CH3 that has not been observed as yet.25 It is believed that the location of these organic fragments is governed by the necessity of tetrahedral symmetry for the bonding carbon atoms. That is, a =CH fragment occupies a three-fold site with bonding to three metal atoms; a CH2 fragment has two metal bonds at a bridge site, and by analogy a -CH3 fragment should have one metal bond and be localized at a top site. If this is the desired bonding configuration of the various fragments, it explains the mechanism by which a three- fold strongly-binding sites is freed by successive hydrogenation of the fragments and becomes available to the next incident molecule.The alkylidyne molecules are present only under conditions of catalysed reactions at low temperatures, as they decompose above 400 K on most transition- metal surfaces. This restricts their importance by and large to hydrogenation reactions, which having low activation energies may proceed well below 400K. Recent studies of C2H4 hydrogenation over Pt and Rh( 111) crystal faces indicate that it occurs on top of the ethylidyne layer that remains ordered and its residence time is much longer than the turnover time of C2H4 hydrogenation.26 25 C. Minot, M. A.Van Hove, and G. A. Somorjai, Surf. Sci., 1982, 127,441. 26 F. Zaera and G. A. Somorjai, J. Am. Chem. SOC.,1983, in press. 334 Somorjai Different ethylidyne species: bond distances and angles (rc = carbon covalent radius; rM = bulk metal atomic radius) c [A1 rn rM rc a ["I Cog (CO), CCH, 1.53 (3) 1.90 (2) 1.25 0.65 131.3 H, Ru, (CO), CCH, 1.51 (2) 2.08 (1) 1.34 0.74 128.1 H, Os, (CO), CCH, 1.51 (2) 2.08 (1) 1.35 0.73 128.1 Pt (111) + (2 X 2) CCH, 1.50 2.00 1.39 0.61 127.0 Rh (111) + (2 X 2) CCH, 1.45 (10) 2.03 (7) 1.34 0.69 130.2 H3C -CH3 1.54 0.77 109.5 H,C = CH, 1.33 0.68 122.3 HC = CH 1.20 0.60 180.0 Figure 16 The surface structure of ethylidyne, the bond distances and angles, are compared with several tri-nuclear metal cluster compounds of similar structure For other catalysed hydrocarbon reactions that occur at an appreciable rate only at higher temperatures the organic fragments are the permanent fixture on the surface during the reaction.Their main role appears to be H-transfer to the adsorbed reaction-intermediates as the C-H bonds retain the hydrogen more easily than the bare transition-metal surface. H-D exchange studies using pre-deuterated Molecular Ingredients of Heterogeneous Catalysis (xlooo 023 I ETHYLENEn Vcc 1350 Heating to: x1106 774 500 KI 'C H CH co 3016 (a1 -bx 1238 410 K1 x 1183 I 1 I I I I I 0 1000 2000 3000 cm-' Figure 17 Changes of the vibrational spectrum of chemisorbed ethylene as a function of increasing temperature.Sequential decomposition is clearly visible from the vibrational spectrum obtained by high resolution electron energy loss spectroscopy Somorjai Figure 18 Schematic representation of the various organic fragments that are present on metal surfaces at high temperature. Thepresence of CH, C,, C2H, CH,, and C-CH, species have been detected CO Chemisorption on Carbon Covered Pt (II I ), Pt (1001 and Pt (13,1,1) I.o -I CO chemisorbed after n-Hexane reaction rate studies -CO coadsorbed with Graphitic Carbon, 0.5 n-Hexane deposited 673 K -A A Pt (111) 0 0 Pt (13,IJ) -0 Pt (100) Ul 1 4 0 2.o 4.O 6.0 Carbon Atoms per Surface Pt Atom Figure 19 Fractional concentrations of uncovered platinum surface sites determined by CO adsorption-desorption as a function of surface carbon coverage on the (loo),(1 1l), and (1 3,1,1)platinum crystal surfaces.A comparison is made between the CO uptake determined following n- hexane reaction studies and CO uptake determined when CO was co-adrorbed with graphitic surface carbon 337 Molecular Ingredients of Heterogeneous Catalysis fragments or reactants indicate that the rate of H-D exchange is at least an order of magnitude faster than the turnover rate of most hydrocarbon conversion reactions. Thus, the hydrogen atoms in the C-H bonds of the strongly held organic fragments are readily transferred to the adsorbed intermediates whereas the carbon atom does not exchange easily. Fortunately not all of the metal sites are covered with the organic fragments, although AES studiesindicate that more than amonolayer ofcarbon ispresent on the surface under catalytic reaction conditions.We can titrate the remaining bare metal- sites by the chemisorption of CO at low pressures which, under the same conditions, does not absorb on the carbonaceous deposit6 Figure 19 shows the fraction of the bare metal surface (8/8,), that is present after the reaction where 8, is the concentration of chemisorbed CO on the initially clean metal-surface before the reaction. About 5-20% of the Pt is uncovered, the bare-metal area decreasing with increasing reaction temperatures. Of course at higher hydrogen pressures (all hydrocarbon conversion reactions are carried out in the presence of excess hydrogen) the fraction of uncovered metal increases.MODEL FOR THE WORKING PLATINUM CATALYST :eous Figure 20 Model for the working platinum catalyst that was developed from our combination of surface studies using single crystal surfaces and hydrocarbon reaction rate studies on these same surfaces Somorjai From these studies a molecular model of the working Pt catalyst can be constructed and is shown in Figure 20. There are bare-metal islands whose structure is determined mostly by the catalyst fabrication.6 The incident reactant molecules adsorb and undergo chemical rearrangements on these metal islands. Then the adsorbed intermediates diffuse onto the carbonaceous deposit, pick up one or more hydrogen atoms, and desorb as the products.Once the carbon deposit loses all its hydrogen and becomes graphitic, hydrogen transfer, which is an important part of the catalytic reactions, can no longer occur and the catalyst surface becomes inactive. 4 Oxidation State of Surface Atoms The importance of different oxidation states of transition-metal ions can well be demonstrated through the studies of the CO/H, reaction over rhodium.” The metal produces mostly methane, as shown in Figure 21, because it dissociates CO I:I H2 KO, 30O0C,6atm Clem Rh Preoxidised Rh Rhodium Oxide Lanthanum Rhodote La Rh O3 I-c: too Rh2 03 3 Y m 180 -0+ 3 160 -a n !-v, 5 40 -+ 0 ene$ 20 -a. 0-Figure 21 Product distribution in the carbon monoxide hydrogenation reaction on various rhodium compound surfaces and has superior hydrogenating ability.However, over Rh,O, surfaces a large fraction of oxygenated molecules CH,CHO, CH,OH, and C,H,OH form. When the rhodium ion is incorporated into the crystal lattice of a refractory oxide such as La,O, in the form of LaRh,O, the products of the CO/H2 reaction are exclusively oxygenated hydrocarbon^.^' This drastic change in reaction selectivity has several causes. Figure 22 shows the heats of adsorption of CO and D, on the rhodium metal, Rh,O,, and LaRhO, surfaces. Rh,O, binds CO more weakly and D, more strongly than the rhodium metal. However, the active LaRhO, appears to have at least two binding states, indicating that the transition metal is present on the active surface in at least two different oxidation states.2’ P. R.Watson and G. A. Somorjai,J. Catal., 1982,74, 282. 339 Molecular Ingredients of Heterogeneous Catalysis De sorption Temper atur e ( "C1 0 I00 200 300 400 I1 L0203 Fresh LaRh03 Used La Rho, Used Rh,03 Rh metal 'Ico D2 20 30 40 Heat of Desorption (kcal/mole) Figure 22 Heat of desorption (kcal mole-') of CO and D, from lanthanum oxide, fresh and used lanthanum rhodate, fresh and used rhodium oxide and rhodium metal. The spread of each value represents the variation with surface coverage rather than experimental uncertainty Rhodium oxide has a unique ability to carbonylate olefins, an important step toward the formation of oxygenated species.When C2H, is added to the CO/H2 reactant mixture it is carbonylated quantitatively to propionaldehyde. On rhodium metal, ethylene was hydrogenated fully to ethane. Thus, oxidation of the metal reduces its hydrogenation ability and makes it active for carbonylation. The difficulty is to maintain the higher metal oxidation states that produce desireable products under conditions of the catalytic reactions that occur in a Somorjai highly reducing atmosphere. The refractory oxide support plays a key role in this circumstance. The Auger spectra of Rh203 before and after the CO/H2 reaction indicates that the oxide is reduced to the metal within an hour with the corresponding change of the product distribution from the oxygenated species to methane.However, LaRhO3 does not loose its lattice oxygen and the Rh3+ ion is fixed in the crystal lattice by the large lattice energy. Thus, the higher oxidation state transition-metal ion is kinetically stabilized in the reducing reaction mixture and remains stable indefinitely as long as the temperature is not increased too high. The so-called strong metal support-interaction is often used to stabilize one or more different oxidation states of transition-metal ions. Thus, the catalyst support plays an important chemical role in most catalyst systems in addition to providing high surface areas on which the metal component may be finely dispersed. Another example of the importance of the changing oxidation state of transition- metal ions at the surface is shown by the catalytic cycle leading to the photocatalysed dissociation of H,O or SrTiO, surfaces.28 This is shown in Figure 23.The oxide surface is completely hydroxylated in the presence of water and the Ti ions are in the 4 +formal oxidation state. When the surface region is irradiated with light of energy 3.1 eV or larger, electron-hole pairs are generated. The electron is utilized to reduce the Ti4+ to the Ti3+ formal oxidation state. The electron vacancy induces charge-transfer from the hydroxy-group, producing OH radicals that dimerize to H,O, and split off oxygen that evolve^.^' The reduced Ti3+-containing surface can now adsorb another water molecule that acts as an oxidizing agent to produce Ti4 + again and a hydroxylated surface, evolving hydrogen in the process.Clearly changes of oxidation states of transition metal ions are frequently indispensable reaction steps in catalytic processes. H20 13, 2-13+-Ti-0-Ti-+ H202 I I I I Figure 23 A proposed mechanism for the photodissociation of water over TiO, and SrTiO, surfaces (Van Damme and Hall, J. Am. Chem. Soc., 1980, 101,4373) 5 The Building of Catalysts By giving the examples of surface and catalytic studies on well characterized systems I hoped to demonstrate the understanding that could be achieved of the 28 F. T. Wagner, S. Ferrer, and G. A. Somorjai, in ‘Photoeffects at Semiconductor-Electrolyte Interfaces’, ed. A. J. Nozik, ACS Symposium Series No. 146, 1981, p. 159.29 H. Van Damme and W. K. Hall, J. Am. Chem. SOC.,1979,101,4373. 341 Molecular Ingredients of Heterogeneous Catalysis molecular ingredients of important catalytic systems. We can now utilize this understanding to build better systems by alteration of their structure or their state of surface charge. Below we discuss two examples of deliberate catalyst modifications: the effects of gold and potassium on transition-metal catalysis. A. The Effect of Gold on the Selectivity and Activity of Pt Catalysts.-The influence of gold on Pt hydrocarbon conversion catalysis has been studied by condensing Au on Pt crystal surfaces. Gold forms epitaxial layers on Pt and upon heating it forms an alloy in the near surface region.30 This Au-Pt alloy has a markedly different selectivity and activity for the conversion of n-hexane into other hydrocarbons as shown in Figure 24.The isomerization rate goes up as compared to that on clean Pt AU -Pt (111) Alloys /vv +H2, 573K H2/HC= 10, Ptot = 220 Torr Ill1 IIII A A0 ,-----A' -X ot A '. I!' '. \ a \" \ c! 0\ \ \. \ \ -0 0.5 I Fractional gold surface coverage Figure 24 The rate of formation of various products from n-hexane as a function of fractional gold surface coverage for gold-platinum alloys that were prepared by vapourizing and diffusing gold into Pt(111) crystal surfaces 30 J. W. A. Sachtler, M. A. Van Hove, J. P. Biberian, and G. A. Somorjai, Surf. Sci., 1981,110, 19. Somorjai while the hydrogenolysis and dehydrocyclization rates are reduced exponentially with increasing gold c~ncentration.~ ' This remarkable selectivity and activity alteration can be explained by a change of structure of the Pt(ll1) surface induced by gold alloying.By substitution of a gold atom, the high co-ordination three-fold Pt sites are eliminated much faster than the two-fold and one-fold bridge and top sites. This is commonly called the ensemble effect. As a result, the chemistry that requires the adsorption of molecules and surface intermediates at the three-fold sites is eliminated while the chemical reactions that require adsorption at bridge or top sites are not attenuated. Although subtle electronic changes may also occur at the alloy surface sites, most of the results can be rationalized by this selective high co-ordination site elimination model.Similar observations were reported by Boudart et a/.for the production of water from H, and 0, over Pd-Au alloy surfaces. Small amounts of gold increased the rate of this reaction by fifty-fold. It should be noted that gold is a very poor catalyst for both of these reactions. Nevertheless, its presence as an alloying constituent can beneficially influence the selectivity and the reactivity of transition metal catalysts. B. The Effect of Potassium on the Bonding and Reactivity of Carbon Monoxide and Hydrocarbons.-Potassium has a high heat of adsorption when present in low coverages on transition-metal surfaces (Figure 25). Simultaneously it also reduces the work function of the transition metals, indicating large charge-transfer between the metals.32 A model that assumes that potassium is ionized when adsorbed on the transition-metal surface explains these results.As the potassium concentration Potassium Heat of Adsorption vs. Covero ge , Y = IO~~S~C-~ 4 0-t I I I I Ih 0 0.4 0.8 I. 2" Bulk K Coverage (monoloyers) Figure 25 The heat of adsorption ofpotassium onplatinum single crystal surfaces as a function of potassium coverage 31 J. W. A. Sachtler and G. A. Somorjai,J. Catal., 1983,81, 77. 32 E. L. Garfunkel and G. A. Somorjai, SUP$Sci.,1982, 115,441. Molecular Ingredients of Heterogeneous Catalysis increases the charged species repel each other and depolarization occurs; the potassium layer becomes metallic and its heat of adsorption approaches rapidly the heat of sublimation of potassium metal.Potassium has a strong influence on the heat of adsorption of CO on transition- metal surfaces. This is shown in Figure 26. In the absence of potassium, CO desorbs at a maximum rate from the Rh(ll1) surface at 400K. However, when co-adsorbed with 50% of a monolayer of potassium, it desorbs at 600 K indicating a 12 kcal Soturotion CO Exposure TDS for Various K Coverages0 8,. 300 400 500 60 0 Desorption Temperature (K) Figure 26 CO thermal desorption spectrum from clean platinum and when co-adsorbed with potassium on platinum crystal surfaces increase of its binding energy.33 The HREELS spectra of CO on Pt(ll1) also exhibits major changes that are shown in Figure 27.In the absence of CO, two well- defined CO stretching frequencies are detectable that are associated with CO at a top and at a bridge site adsorbed with its CO bond perpendicular to the surface.33 As the potassium is added to the Rh surface, CO shifts to the bridge site and its stretching frequency decreases by more than 300~m-’.~~ This corresponds to a gradual change of bond order with increasing potassium coverage from 2 to 1.5. This indicates that the electron transferred from the potassium to the transition metal density of states can populate the antibonding molecular orbitals of CO, thereby weakening the C-0 bond. Simultaneously the metalkarbon bond is strengthened as charge density in this bonding orbital must increase.Potassium is often used as a beneficial additive to transition-metal catalysts utilized for the hydrogenation of carbon monoxide. Its presence increases the 33 E. L. Garfunkel, J. E. Crowell, and G. A. Somorjai, J. Phys. Chem. 1982,86, 310. 34 J. E. Crowell and G. A. Somorjai, ‘Summary Abstract for the American Vacuum Society 30th National Vacuum Symposium’, Nov. 14, 1983, Boston, Massachusetts, J. Vuc. Sci. Tech., submitted for publication (1983). 344 Somorjai SATURATION CO COVERAGE (T=mK) ON Pt(III)/K I 1565 eK-oo5 OK = 0 1 I I I I I I I 0 1000 moo 30 00 ENERGY LOSS (cm-1) Figure 27 Vibrational spectra of CO at the saturation coverage when chemisorbed on Pt( 111) at 300 K as a function of a pre-adsorbed potassium coverage molecular weight of hydrocarbon products, as would be expected if the dissociation rate of carbon monoxide is enhanced.Potassium, however, is a non-selective poison for hydrocarbon reactions on platinum surfaces.35 The reason for this is revealed in recent surface studies. The presence of potassium increases the activation energy for the breaking of C-H ’’F. Zaera and G. A. Somorjai, J. Catal., 1983,84,375. Molecular Ingredients of Heterogeneous Catalysis 3c -Q)-28 1 0 V Y v 0 IStPeak W H2 Thermal Desorption After 2 Hours n-Hexane Reaction 26 Pt(II1) 573K PHc=20 Torr, P,, = 600 Torr 2 I I I I 1 0 0.20 0.40 8, Figure 2.8 Activation energy of the hydrogen P-elimination from carbonaceous deposits after n-hexane reactions over platinum (1 1 1) surfaces as a function of potassium coverage bonds, which is an important step in most hydrocarbon conversion reactions.This is shown in Figure 28. Thus, the surface residence time of the molecules increase and this reduces the catalytic turnover rates. There is little doubt that potassium influences the catalytic reaction by charge transfer, that is, by electronicchanges. It has large effects on some molecules when co- adsorbed with them (CO,N,) and virtually noeffects on others(N0, PF3).36It would be of value if we could predict by the use of theoretical techniques whether charge transfer between the molecular orbitals of adsorbates and the charge density that is altered by the adsorption of potassium on the transition-metal surface could or could not take place.6 New Reactions, New Processes As our understanding of how catalysts work increases through the application of molecular surface science, it will be increasingly possible to find catalysts for reactions for which good catalysts do not seem to exist. There are many important reactions of small molecules such as C02, CH,, H,O, and N, that may be investigated. C02 could be well utilized as a source of carbon-containing chemicals if it could be hydrogenated to HCOOH or dissociated to CO. Figures 29-31 36 E. L. Garfunkel. J. J. Maj, J. C. Frost, M. H. Farias, and G. A. Somorjai, J. Phys. Chem., 1983,87,3629.Somorjai 2H+(aq)+2NOJ (oq) 15.99'7 257.115 H20 (I) +N2+ 7 0, HCOOH(aq 1 30.1'T C02(oq)+H2 2 H20 +C H4 Figure 29 Standard free energies for several chemical reactions -200 --I Q,-2-300 7 A U a0ON* a-400 -500,-+ 2 Hfl Figure 30Standard free energies for several chemical reactions 347 Molecular Ingredients of Heterogeneous Catalysis show the free energy changes associated with several reactions. While CO, hydrogenation is thermodynamically feasible, its dissociation to CO and 0, requires the input of excess energy in the form of light or heat. The reaction of carbon with water to produce CH, and CO, is thermoneutral, as shown in Figures 29-3 1. This reaction may represent a desirable alternative for carbon gasification with water to CO and H,, a very endothermic reaction indeed.The partial oxidation of methane to CH,O and CH,OH should be feasible by suitable catalyst surfaces.37 These reactions are shown in Figure 31. Finally, the oxidation of nitrogen to nitric acid (N,O + 5/2 0, + H,O -+ 2H’ + 2NO,-) is thermodynamically feasible, as pointed out by G. N. Lewis in t923.38 ‘Even when starting with water and air, we see by our equations that nitric acid should form until it reaches a concentration of about O.lM where the calculated 30( 2OC Ioc Y 00 80 t 0 --I Q, 75 -lo(E 71 v 0c3 a -20( -3o( -4OC 120,-2+C 2 (graphite) H20-5OC Figure 31 Standard free energies for several chemical reactions 37 M.M. Khan and G. A. Somorjai,J. Catal., to be published (1984). 38 G. N. Lewis and N. Randall, ‘Thermodynamics’, McGraw Hill, 1923. 348 Somorjai equilibrium exists. It is to be hoped that nature will not discover a catalyst for this reaction, which would permit all of the oxygen and part of the nitrogen of the air to turn the oceans into dilute nitric acid.’’* Acknowledgements. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Materials and Chemical Sciences Division of the U.S. Department of Energy under Contract Number DE-AC03- 76SF00098.

ISSN:0306-0012    

DOI:10.1039/CS9841300321

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

Corrigenda Volume 12 No 4 1983 ‘The Glass Transition: Salient Facts and Models’ by R. Parthasarathy, K. J. Rao, and C. N. R. Rao. Line 1, p. 377 should read ‘due to excitation to the firstfew excited states, from Cp,’ Figure 8, p. 378 is to be replaced by the new Figure 8 below. I I II I /I I Ill \ RTlAE Temperature/ OK Temperature /OK Figure 8 Plot of C ‘On‘ versus temperature from the model. Calculations were all performed with a = 1.6 and A4 = 4.184 kJ mol-’. Numbers in theparenthesis are respectively number of initial states ignored as due to SC, and total number of states used in the calculations. The insets refer to variations of 6 as a function of temperature for the particular (n,a) set where n is the total number of states. Figure 8(a) represents calculations without use of ASi while 8(b) corresponds to cblculations performed with ASi = constant = 3.138 J deg-’ mol-’ 35 1

ISSN:0306-0012    

DOI:10.1039/CS9841300351

出版商:RSC    

年代:1984

数据来源: RSC