摘要:

Techniques for the Kinetic Study of Fast Reactions in Solution By H. Kriiger DEPARTMENT OF CHEMISTRY, UNIVERSITY OF SOUTH AFRICA, P.O. BOX 392, PRETORIA o o o 1, SOUTH AFRICA 1 Introduction Following the earliest investigation in 1850, the rates of chemical reactions in solution were studied by simple methods for many years. The reaction to be studied was initiated by mixing the reagent solutions, and its progress was followed by titrating samples, or carrying out measurements of some physical properties after various time intervals. In this way reactions which last longer than 10s could have been studied. In 1866/67 Harcourt and Essonl published their classical paper ‘on the laws of connexion between the conditions of a chemical change and its amount’, and first showed that the rates of reactions can be interpreted in terms of math- ematical equations.Since those days a vast number of kinetic investigations has been carried out and has led to the elucidation of the mechanisms of all types of chemical processes. For a long time there were essentially two time ranges for kinetic investi- gations. These were the ranges of chemical kinetics, the lower limit of which is at about 1 s, and the range of spectroscopy, essentially the range between 10-10 and 10-15s. This latter range usually does not involve chemical trans- formations, but gives information on physical processes (e.g. electron motion, vibrations, bond stretching, etc.). Actual chemical transformation is expected to lie just within the gap extending from 1 to 10-lo s.Some time ranges in seconds are given below in Figure 1. tc time limit for chemical changes Figure 1 Time ranges (in seconds) A. V. Harcourt and W. Esson, Philos. Trans., R. SOC.London, 1866, 156, 193: 1867, 157, 117. Techniques for the Kinetic Study of Fast Reactions in Solution The time range between 1 and 10-los has been made accessible by a number of new techniques, which enables us to study almost any chemical reaction in solution with half-times between one second and fractions of millimicroseconds.2 Figure 2 below shows some of the methods available and their time ranges. A whole family of methods for studying fast reactions has been developed over the past 60 years. Much ingenuity has been applied to following the details of chemical reactions on an increasingly short time scale.I I I I I Spectroscopic1 rangeRapid flow 1I !Flash photolysis I micro-I Electric field variation wave, i.r., I I u.v.,visible I Sound absorption t IP-jump I IT-jumpI I;r I Table 1 Kinetic methods nnd time ranges (in s-’) These techniques are specifically adapted to the requirements for recording chemical transformations and utilize conductimetric, polarimetric, spectro-photometric, and fluorimetric detectors depending on which one is most sensitive to the particular reaction. Indirect methods such as n.m.r., e.s.r., electrochemical and photostationary techniques may also provide valuable information. A.Difficulties in Measuring Fast Reactions.-We generally describe as ‘fast’ anything that takes place quickly compared to the rate of resolution of our sense perceptions. Since our perceptions are in turn based on chemical processes, these reactions must necessarily be ‘fast’, indeed ‘extremely fast’. The term ‘fast reaction’ is best applied to reactions that cannot be followed kinetically by conventional methods and which proceed much faster than the time required for mixing and observing. A chemical reaction is generally started by mixing, for example, two solutions and then measuring the concentration of one or more of the species present at known intervals of time. This type of procedure cannot be used if the start of a reaction is ill defined as when the time of mixing is compatible with or is * G.G. Hammes, ‘Techniques of Chemistry’, Vol. VI, Investigation of Rates and Mech-anisms of Reactions, Part 11, 1974, John Wiley & Sons. 228 Kriiger greater than the total time required for the reaction to occur. A similar con- straint will apply if the time required to complete a measurement on a reaction mixture is no longer negligible when compared with the duration of the reaction. There is an upper limit for the rate constant of a chemical reaction. If molecules A and B react, then the process proceeds in two steps. First, A and B must meet in solution. Only after having met can they react with each other. The time of meeting in solution is the time of diffusion of the two partners and is 10-l2s. This time cannot be shorter and is also valid for monomolecular reactions. When molecule A decomposes, the time of decomposition is at least s, i.e.the time of vibration of a breaking bond of that molecule. 2 Flow method^^-^ Until 1923 there did not exist any direct, generally applicable method for measuring the velocity of rapid chemical reactions in the liquid phase with half- times of less than about 10 s. In that year, however, Hartridge and Roughton7 extended the observable time range about 104-fold by devising a new method. If one wants to follow the reaction between species A and B in solution, samples of the two solutions can be let to a mixing chamber, specially designed to achieve rapid mixing, and then passed through an observation tube. The concentration of one or more of the species present can then be detected at various points along the observation tube by spectroscopic means and recorded by some form of high-speed recorders.Other properties of the mixture, such as electrical conductivity, which change during the reaction, can also be measured at different points of the tube. This is called the continuous-flow method and is illustrated in Figure 2. solution A ,-I 4 I/' point of 13mixing-] +-Figure 2 The continuous-j7ow method E. F. Caldin in 'Fast Reactions in Solution' 1964, p. 29, Blackwell Scientific Publications. K. Hiromi, Kugaku, Zokun (Kyoto), 1979, 80, 7. G. Czerlinski and K. Tatti, J. Biol. Phys., 1977, 5, 184.New Applications of chemical relaxation spectrometry and other fast reaction methods in solution, W. J. Gettins and E. Wyn-Jones, NATO Adv. Study Inst. Ser., Aberystwyth,1978, p. 13.' H. Hartridge and F. J. W. Roughton, Proc. R. Soc. London, Ser. A., 1923, 104, 376. Techniquesjor the Kinetic Study of Fast Reactions in Solution The flow rate in the tube is known and is approximately 10 ms-1. At distances 11 ...12. ..13 ...from mixing chamber a time of tn =Inv-l s have passed since mixing took place, where v is the flow velpcity. The mixing chamber is specially designed to achieve rapid mixing. In the stopped-flow method,8 which nowadays is used generally, the flow is suddenly stopped and checked and light absorption measured as a function of time.The response may be applied to a cathode-ray oscilloscope. With the help of a time base, a curve representing the extent of reaction against time may be displayed on the screen. The time scale extends from a millisecond to several mi nu tes . Numerous reactions have been investigated by flow-methods, such as the reaction between Fe3+ ions and thiocyanate ions,g or the dissociation of Ni(en)$+ in strong acid medium,I0 in proton transfer reactions and enzyme catalysis3 or in electron transfer reactions,"-'* and many more.15-17 3 Chemical Relaxation technique^^^^^^^^ A. Introduction.-In 1954 Manfred Eigen reasoned that rate constants could be obtained if a mixture of reactants and products in equilibrium was subjected to a sudden change in external parameters, such as temperature, pressure, electric field strength etc.The chemical system could then be monitored by physical means as it underwent equilibration (or 'relaxation') to the new equilibrium state. There is a time lag while the system approaches equilibrium, and this time lag is related to the rate constants of the forward and backward reactions. This basically simple idea provided a new vision into elementary steps of chemical processes and is providing a powerful tool for the detection of fast react ions. In all relaxation methods the rate of approach to the new equilibrium is exponential as shown in equation (I), dAct Ac~-=--dr 7-(1) a (a) Q. H. Gibson, Discuss. Furuduy Soc., 1954, No. 17, p. 137.(b) B. Chance, Rev. Sci. Instrum., 1951, 22, 619. J. F. Below, jun., R. E. Connick, and C. P. Coppel, J. Am. Chem. SOC.,1958, 80, 2961. lo Reference 3, p. 51. C. F. Bernasconi, in 'Relaxation Kinetics', Academic Press, 1976, p. 32. la A. BakaZ and J. H. Espenson, J. Am. Chem. SOC.,1981, 103, 2721. l3 G. H. Jones,J. Chem. Res., 1981, 228. l4 J. D. Clemmer, G. K. Hogaboom, and R. Holwerda, Inorg. Chem., 1979, 18, 2567. l5 H. Orgino, M. Shimwa, and N. Tanaka, Inorg. Chem., 1979, 18, 2497. l8 W. A. Schenk and H.Muller, Inorg. Chem., 1981, 20,6. I' H. Elias, U. Reiffer, M. Schumann, and K. H. Wannowius, Inorg Chim. Actu, 1981, 53, L. 65. la F. Wilkinson, 'Chemical Kinetics and Reaction Mechanisms', Van Nostrand, 1980. l9 H. Strehlow and W.Knoche, 'Fundamentals of Chemical Relaxation', 1977, Verlag Chemie, Weinheim. 230 Kriiger where dci is the difference in concentration of the i-th species between time t and infinite time. The relaxation time T is the characteristic constant for the exponential change. This is illustrated in Figure 3. Time t Figure 3 Relation bet ween time and concentration in relaxution studies The graph contains an idealized relaxation curve indicating the variables discussed above. The relaxation time T is generally taken from a plot of lndci versus time t. The relaxation time T is related in a unique functional way to the rate constants, the equilibrium constant, and to the concentration depending on the reaction mechanisms. In all relaxation studies the rate constants and concentrations are those of the perturbed state.Provided the displacement of the chemical equilibrium is small enough, the rate of restoration of equilibrium will always follow first-order kinetics, irrespec- tive of the kinetics of the forward and reverse reactions. The following conditions for relaxation studies are important. (i) The chemical reaction must take place under constant and defined physical conditions. (ii) It must be possible to observe the relaxation process separately from other changes in the system. (iii) Concentrations of the reacting partners must be low in solution. (iv) The perturbation of the equilibrium state must be sufficiently small, so that all changes can be represented by linear differential equations.The use of the relaxation methods has widened our kinetic horizons enormously. Many reactions of great importance provided basic information and develop- ment of relaxation techniques opened a hitherto inaccessible time range with the measurement of rate of chemical reactions with half-lives of less than a nanosecond (< s). Relaxation rates can be measured for what are probably the fistest of all chemical reactions.20921 lo D. W. Oxtoby, J. Chem. Phys., 1981, 74, 5371. a1 Pror. Indian Acad. Sci., Sect. A, 1978, 44, 252. 231 Techniquesfor the Kinetic Study of Fast Reactions in Solution Some of the most spectacular work in the field of fast reactions has concerned biological phenomena.22~23 Among these are the exact replication of the genetic code by DNA molecules, the action of nerve membranes, enzyme-catalysed biochemical processes in all organisms, the role of chlorophyll in photosynthesis, the interaction of haemoglobin with oxygen in blood, and a variety of aqueous inorganic complex systems.In all relaxation techniques the relaxation time 7has kl to be determined.24 When an equilibrium, such as A + BZC is perturbed, the k-,system adjusts itself to the external parameters. By measuring this adjustment (chemical relaxation) one can obtain information on the evaluation of kl and k-I. When [A:] and [Be] are the equilibrium concentrations of species A and B the relationship given in equation (2) can be derived. -1 = k1 ([Ae] + [Be]) + k-, (2) Plotting 7-l against the total concentration [Ae] + [Be] one obtains k-1 from the intercept and kl from the slope experimentally.For pseudo first-order conditions i.e. when [Ae] % [Be],equation (3) applies. In the case where [Ae] = [Be],equation (4) applies. -1 = k1 [Ae] + k-1 (3) For other single-step systems : kl (a) For 2A + B: 71 = 4k, [Ae] + k-l k-I ki(e) For A + C + B + C: -1 = (k, + k-,)[Ce] (C is a catalyst)k-I k1(f) For A + B + C + D: ; I = k1 ([A][Be] + [AeILCeI + [BeI[CeI)+ k-,k-I In multi-step systems most reaction mechanisms involve several steps and are characterized by more than one relaxation time. If the intermediate states J.-E. Dubois, Pure Appl. Chem., 1978, 50, 801. 43 H. Hatano, Kagaku, Zokan (Kyoto), 1979,80, 1 E.L. King, J. Chem. Educ., 1979, 56, 580. Kriiger accumulate to measurable concentrations, several or all the relaxation times may in fact be observed experimentally. The number of relaxation times is always equal to the number of independent rate equations that can be written for the system. In linear systems, such as A + B + C + D + E + F this number is 3, the number of reaction steps. In cyclic systems, such as the one shown in Scheme 1, this number issmaller than the number of steps. In general the number A+B+C+D+E 11 11 G+H$I+K+L Scheme 1 of relaxation times is always equal to the number of states minus one. Two-step systems’l are among the most frequently encountered multi-step mechanisms in chemistry (see Table 2).Cyclic reaction schemes are quite common in chemistry. The simplest is the triangle, as shown in Scheme 2. The starting material A forming a product D I kl2 rr % kt3 k21 k32 k3, -1 v-D k13 Scheme2 can react by two competing mechanisms, one being a concerted one-step reaction, the other involving the formation of an intermediate I. Familiar examples are SN~ or El versus E2.versus SN~, For 71 a shorter relaxation time than 72 and setting aii = k12 + k13 + k21 a12 = k21 -k31 a21 = k23 -k13 aZ2= k23 + k31 + h2 we obtain the two relaxation times given in equations (5) and (6). If we assume that the k12 and k13 steps are bimolecular, i.e. A + B + D and A + B + I, then we obtain equations (7) and (8) where KIZ= equilibrium constant for the system A + I.233 Techniquesfor the Kinetic Study of Fast Reuctions in Solution Techniques for the Kinetic Study of Fast Reactions in Solution I Y + + 4 Yv I Y + +.z'sziYYI< Y hI n I Y Y YG I Y -+ -+ -+I h h hn -+ 5 d 5 s & Y Yv t %+ I YI n Y Y Yi + + + b Y Y Y Y YI II + .y Y + Kriiger B. The Temperature-jump Perturbation Meth~d.~,~~.~*t~~-Rapidchanges in temperature can be brought about in a system in equilibrium by charging a capacitor to a high voltage and then allowing it to discharge through the equilib- rium mixture in solution. This discharge current heats the mixture rapidly and the temperature rises a few degrees in microseconds.The instrumentation used for T-jump spectrophotometers is illustrated in Figure 4. variable spark gap I I monochromator C?l --1 **LI-2 I F-" Figure 4 Instrumental set-up for T-jump measurements The sample cell is made of a non-conducting cylindrical plastic chamber equipped with optical observation windows in the side and a pair of electrodes fitted into the ends of the chamber. The high voltage power supply charges the condenser, of some 10-8 F, up to 50 kV. When the variable spark gap is fired the condenser (capacitor) is discharged across the cell and the stored electrical energy is converted into Joule's heat. At the same time the scope sweep is triggered. The temperature-jump is undoubtedly the most versatile and useful of the relaxation methods. Since the vast majority of reaction are temperature dependent, i.e.AH f 0, a variation of equilibrium constant Ke with temperature is to be expected. This dependence is expressed by van't Hoff's equation [equation (9)]. For finite but small changes in K,, i.e. dKe Q Ke, equation (10) alnKe AH-=-aT RT2 (9) 26 T.J. Kemp, Prog. React. Kinet., 1980, 10, 301. 235 Techniquesfor the Kinetic Study of Fast Reactions in Solution applies, so that we can re-write equation (9) in the form of equation (11). It is evident that the relative change in Ke for a given temperature-jump AT is proportional to AH. The temperature-jump is given by equation (12). Changes C is capacitance of condenser (0.1pF)U is voltage applied (50kV) cp is specific heat of solution p is density of solution Y is volume of solution (2 cm3) in colour or transparency of the sample solution to be investigated are monitored by a spectrophotometer and recorded on an oscilloscope.The monochromator emits the wave-length of the maximum absorbance of the species. Changes are recorded on the oscilloscope as a function of time. There are a few limitations to the T-jump method : Products and reactants of a process to be studied must be in equilibrium. The reaction system must have an appreciable concentration of ionic species present to conduct the discharge current. The concentration of species in the system must be temperature de- pendent, i.e. AH # 0.Relaxation times of less than 10-7s cannot be observed with this technique. The method can only be applied to reactions in which the rate of concentration-change is slower than the time required to heat solution to new temperature. The method has an extremely wide range of applications. This is particularly as a result of the development work of Leo de Maeyer,", G. Czerlinski,27 and G. G. Hammes.28 Some well-known applications may be mentioned here: (i) Proton-transfer reaction~:~~JOThe dissociation of water and the recombination of hydrogen and hydroxy-ion are important reactions studied by the T-jump method: k,2H20 f H30++ OH-k-1 also the base-catalysed proton-transfer in intramolecularly hydrogen bonded naphthylammonium l6 Lecture held at the 1968 Meeting of the Optical Society of America, Washington. G.Czerlinski and M. Eigen, Z. Elektrochem., 1960, 64, 128. as G. G. Hammes and J. I. Steinfeld, J. Am. Chem. SOC., 1962, 84, 4639. Is M. Eigen and L. De Maeyer, Z. Elektrochem. 1955, 59, 986. 30 G. Ertl and H. Gerischer, 2. Elektrochem., 1961, 65, 629. 31 F. Hibbert and H. J. Robbins. J. Am. Chem. SOC., 1978, 100, 8239. 236 Kriiger (ii) Metal complex formation in inorganic- and bio-chemistry, such as the binding of CoI1 to adenosine 5'-monophosphate,32 or cations derived from phthalein complexone,33 or the formation of iron complexes such as Fe( H20)5Fe(CN)6. 34 (iii) Tautomerizm in organic compounds such as cyto~ines~~ or keto-enol transformations in heterocyclic compounds.(iv) Electron-transfer reactions, such as that between ferro-and ferri-cyanide ions.36 (v) Enzyme catalyses in biochemistry and biology have been investigated through temperature-jump meth0ds.3~-3~ (vi) The reaction between oxygen and haemoglobin has also been studied by this method.40 Much research has been carried out in order to improve these techniques. A highly sensitive microwave T-jump apparatus has been developed at the University of Paris.*l Temperature-jumps of 1.5" are achieved in a flow microcell within 1.5 microseconds at a repetition rate of 50 Hz. A nanosecond temperature-jump apparatus for high pressures has been developed at the University of Bochum, which gives an increase of a few degrees within 50 nano-seconds.42 J.Aubard and co-workers have described a Laser temperature-jump apparatus with highly sensitive and very fast pulsed spectrometric detection.43 The kinetics of the conformation change of adenylyl 3' to 5' adenosine has been studied with a T-jump of 6" within 20 nanoseconds. C. The Pressure-jump Meth0d,~~-47--When pressure is changed rapidly and hence adiabatically in a chemical process, the equilibrium constant Ke will change accordingly. Rapid changes in pressure can be brought about by the use of sound waves of certain frequencies. 38 A. Pegny and H. Diebler, J. Phys. Chem., 1977, 81, 1355. 33 G. Czerlinski and M. Eigen, 2.Elektrochem., 1959, 63, 652. 34 J. E. Finholt, J. Chem. Educ., 1968, 45, 394. 35 M. Dreyfus, 0.Bensaude, G.Dodin, and J. E. Dubois, J. Am. Chem. SOC., 1976,98,6338. H. Diebler, 2.Elektrochem., 1960, 64, 128. 37 Physiol Chem. Phys., 1979, 11, 537. 38 G. G. Hammes, Adv. inorg. Biochem., 1980, 2, 237. 39 E. J. del Rosario and G. G. Hammes, Biochemistry, 1971, 10, 716; G. G. Hamrnes and J. L. Haslam, ibid., 1969, 8, 1591; G. G. Hammes, Acc. Chem. Res., 1968, 1, 321; K. Kirschner, E. Gallego, 1. Schuster, and D. Goodall, J. Mol. Biol., 1971, 58, 29. 'O G. S. Adair, A. V. Bock, and H. Field, jun., J. Biol. Chem., 1925, 63, 529. 41 J. Aubard, J. M. Nozeran, P. Levoir, J. J., Meyer, and J. E. Dubois, Rev. Sci. Instrum., 1979,50, 52. 48 Rev. Sci. Instrum., 1979, 50, 1089. 43 J. Aubard, J. J. Meyer, and J. E. Dubois, Chem. instrum., 1977, 8, 1.44 G. Platz and H. Hoffmann, Ber. Bunsenges. Phys. Chem., 1972, 76, 491. 46 R. Van Eldik, Chemsa, 1980, 6, 46. 413 J. S. Davis and H. Gutfreund, FEBS Lett., 1976, 72, 199. 47 H. Strehlow and J. Jen, Chem. Instrum., 1971, 3, 47. Techniquesfor the Kinetic Study of Fast Reactions in Solution From thermodynamics : d V = change in reaction volume dH = change in reaction enthalpy a = thermal extension coefficient p = density of reaction solution cp = specific heat at constant pressure R = gas constant Normally, in aqueous solution, the first term on the right side in equation (13) is the major contribution (greater than 90 % of total). The instrumental set-up is shown in Figure 5. Figure 5 Instrumental set-up for P-jump measurements Two identical conductivity cells are used.One is filled with the test sample, the other with a solution having equal electrical conductivity but showing no relaxation. By comparing the changes in resistance of the two cells, disturbances caused by temperature fluctuation largely cancel. The cells are closed with diaphragms (polyethylene membrane). The Wheatstone bridge is tuned at the ambient pressure of 1 atm. Pressure is increased slowly to -50 atm until the diaphragms burst spontaneously. The pressure drops to 1 atm within 50 ps. Kriiger The fast pressure change generates a voltage peak across a condenser which triggers the oscilloscope. The trace on the screen shows the solution under investigation regaining equilibrium at 1 atm.Shock-waves have been used in order to perturb the system and shorten the instrument time to 1 ps. The method has found wide applications in the study of the formation of metal complexes,4*-50 the study of biochemical51-56 and 0rganic5~9~~ reactions, and in the study of micelle f~rmation.~~-~l D. The Electtical-field Jump Meth~d.~J~.~~-Anapplied electric field will in-fluence chemical equilibria involving ions, dipoles, and polarizable species. The magnitude of this field required is as high as 100 kV cm-l in liquids of high di- electric constants such as water. The field always induces a shift in the equilibrium towards producing more ions. An electric field applied to a solution of weak electrolyte increases the ion conductivity (first Wien-effect) and the degree of dissociation (second Wien-effect).The former effect must be kept to a minimum, the latter is of importance to our discu~sion.~~ The fundamental thermodynamic quantity for the interaction of electric fields with chemical transformation is given in equation (14). This principle is (g) =Am TP ~ Iwhere A = reaction affinity -,-E = electric field and rn = electric moment applied in the electric-field jump method. Whenever a solution of a weak electo- lyte is subjected to a very large electric field the equilibrium. is disturbed and Ke increases as shown in equation (15). AM AM = change in molar polarization. b8 H. Hoffrnann and E. Yeager, Ber. Bunsenges. Phys. Chem., 1972, 76, 491. Is H.Strehlow and H. Wendt, Inorg. Chem., 1963, 2, 6. *O G. Macri and S. Petrucci, Inorg. Chem., 1970, 9, 1009. 51 K. Heremans and Y. van Nuland, High Temp.-High Pressures, 1977, 9, 539. sa H. Kihara and S. Saigo, Kugaku, Zokun (Kyoto), 1979, 80, 81. 5s P. B. Chock, F. Eggers, M. Eigen, and R. Winkler, Biophys. Chem., 1977, 6, 239. 64 H. R. Halvorson, Biochemistry, 1979, 18, 2480. 55 D. E. Goldsack, R. E. Hurst, and J. Love, Anal. Biochem., 1969, 28, 273. ‘Methods in Enzymology’, Vol. XVI, Academic Press, 1970. 67 D. A. Palmer and H. Kelrn, Austr. J.Chem., 1977, 30, 1229. 68 H. Strehlow and W. Knoche, Ber. Bunsenges. Phys. Chem., 1969, 73, 427. H. Hoffrnann, W. Ulbricht, and B. Tagesson, Z. Phys. Chem., 1978,113, 17. 6o H. Hoffrnann and B.Tagesson, Z. Phys. Chem., 1978, 110, 113. H. Hoffmann, R. NageI, G. Platz, and W. Ulbricht, Colloid Polym. Sci., 1976, 254, 812; H. Hoffmann and W. Ulbricht, Z. Phys. Chem., 1977, 106, 167; H. Hoffrnann, R. Lang, D. PavloviC, and W. Ulbricht, Croat. Chem. Acta., 1979, 52, 87. a’ M. Eigen and J. Schoen, Z. Elektrochem., 1955, 59, 483. Techniquesfor the Kinetic Study of Fast Reactions in Solution The change occurs under a certain time-effect. The characteristic magnitude of this time-effect is the relaxation time T of the system. The experimental arrangements are virtually identical to those employed for the T-jump measurements,63 the output from a charged condenser being applied to the electrodes in a similar cell. A specific high-field Wheatstone-bridge arrange- ment includes the sample cell for the conductiometric detection of the con- centration changes.Eigen and De Mae~er,~~*~ suggested electric pulses and detection systems. A high-voltage capacitor was used to generate the field pulse, which triggers an oscilloscope. Measurements were obtained with pulses of 50-150 ps. There are two methods of measuring the relaxation time T: (a) an amplitude dispersion-method in which single pulses are applied, and (b) a relaxation method in which rectangular charge-pulses are applied. Eigen and co-workers invented and applied this E-jump technique to measure the rate constants for the combination of hydrogen ions with hydroxy-ions (and acetate ions)62965,66 The equilibrium in equation (16) is established very rapidly.29 However k-1, ki2H20 + H,O+ + OH-k-, a second-order rate constant for the combination of hydroxonium ions and hydroxide ions has been obtained by this method, i.e.1.4 x 10” 1 mol-l s-l at 25°C. This is the largest second-order rate constant known for a reaction in solution. The method has been applied in many investigations, all of which show high resistance or low conductivity. Some examples are listed below: (i) Ion pairing proces~es.~~~~~ (ii) Protonation and deprotonation reaction^.^^ (iii) The reaction HgCl+ + C1-+HgCh70 (iv) Many biological processes, e.g. the mechanism of binding oxygen to haemoglobin in blood? (v) The binding of ligands to metal ions, including hydrolysis.70~72 O3 D.T. Rampton, L. P. Holrnes, D. L. Cole, R. P. Jensen, and E. M. Eyring, Rev. Sci. Instrum., 1967, 38, 1637. 64 ‘Techniques in Organic Chemistry’ Vol. VIII part 2, Interscience, 1963, p. 895. 05 M. Eigen, W. Kruse, G. Maass, and L. De Maeyer, Progr. React. Kinef., 1964,2, 285. 60 J. J. Auborn, P. Warrick, jun., and E. M. Eyring, J.Phys. Chem., 1971, 75, 2488. 07 A. Persoons and M. Van Beylen, Pure Appl. Chem., 1979,51,887. Proc. Int. Meet. SOC. Chirn. Phys., !977, p. 345. 69 M. Eigen, Angew. Chem., Inr. Ed. Engl., 1964, 3, 1. ’O M. Eigen and E. M. Eyring, Inorg. Chem., 1963, 2, 636. 71 T. Yasunaga, H. Takenaka, T. Sano, and Y. Tsuji, in ‘Chemical and Biological Applications of Relaxation Spectrometry’, Reidel Publ. Co., 1975, Proceedings of a NATO Advanced Study Institute (1974); G.Ilgenfritz and T. M. Schuster, ‘Probes of Structure and Function of Macromolecules and Membranes, Vol. 11, Academic Press, 1971, Proceedings 5th Colloquium, Johnson Research Foundation, Philadelphia, 1969. 71 H. Diebler, M. Eigen, G. Ilgenfritz, G. Maass, and R. Winkler, Pure Appl. Chem., 1969, 20, 93; D. L. Cole, L. D. Rich, J. D. Owen, and E. M. Eyring, Inorg. Chem., 1969, 8, 682; L. D. Rich, D. L. Cole, and E. M. Eyring, J. Phys. Chem., 1969,73, 713. 240 Kriiger (vi) The formation of macrocyclic and open-chain antibiotics.53 A new method has been developed for the study of ionic and dipolar equilibria by A. Persoons and co-workers,73 and this has found application in many kinetic research projects.74 E. Ultrasonic Measurement~.2~64~75~7~-Sofar we have dealt with single discrete perturbations of a chemical system. In ultrasonic measurements we have to do with oscillating perturbations. Sound waves, by means of their incremental pressure and adiabatic temperature oscillations, are able to induce periodical perturbations of a chemical equilibrium. In aqueous solutions the pressure perturbation is most important, in non-aqueous solutions the temperature effect is paramount. Characteristic phase-shift changes occur between the oscillation of pressure or temperature and the position of chemical equilibrium leading to a typical energy absorption or velocity dispersion of the ultrasonic wave, which can at the same time be used as a probe.for chemical relaxation.For the small amplitudes of these sinusoidal perturbations the corresponding chemical rate equations can be linearized without significant error. U I I AV = half-power bandwidth Urnax \I = frqucncy bandwidth n u = output voltage Vn Figure 6 Relationship bet ween frequency bandwidth and out-put voltage in ulfrasonic relaxation measurements In the jump methods a recording of the signal on an oscilloscope requires that the time constant Te of the equipment is smaller than the relaxation time r, i.e. Te T. In the case of T shorter than a few microseconds difficulties arise. Sound-wave absorption techniques can solve this problem. Te is determined by the frequency bandwith dv of the equipment by the reIationship shown in equation (17).Also, dv is dependent on the signal to noise ratio S/N, where 1Te x -27rAv 73 A. V. Anisimov and F. G. Miftakhutdinova, Biophys. J., 1979, 24, 116. 74 In?. Symp. Specific Interacr. Mol. Ions, 1976, 2, 545. 76 F. Eggers and Th. Funck, Naturwissenschafren, 1976, 63, 280. 7B M. J. Blandamer, ‘Introduction to Chemical Ultrasonics’, Academic Press, 1973. 241 Techniques for the Kinetic Study of Fast Reactions in Solution S/N 01 (dv)-i. For short relaxation times the bandwidth must, therefore, be large in order to make S/N unfavourably small. Frequencies range from 0.1 to 150 MHz corresponding to relaxation times of 2 ps to 1 ns. The chemical relaxation is observed over many periods of the perturbation.That means the equilibrium concentrations fi change during the measurement, in contrast to the conditions of jump methods. In this case, therefore, Fi can-not be used as the reference concentration. The shift in concentration at any chosen time is given by equation (18): xt = cI -Cpef (18) and the shift in equilibrium concentration is given by equation (19): f. = ,zt -&ref (19) The rate of change of shift in concentration is then given by equation (20). dx--dt = ;1 (x -3) dt[Compare with jump-methods: -* = (c -31 The sinusoidal pressure change in sound waves have very often been used. This causes a sinusoidal change of the equilibrium concentration ci and is described by equation (21). Strehlow and Knocke41 discuss the theory in more detail. -ct = ctreK+ &sin wt with w = -1 (21)i-Many different experimental configurations have been adopted, such as (i) pulse technique,77 (ii) ultrasonic interferometer, (iii) reverberation method, etc.The study of sound propagation demands devices for creating sound waves and subsequently for their detection. This is achieved by means of piezo- electrical crystals, which generate an electric signal when subjected to mechanical stress or, conversely, change their shape when an electric field is applied. The sound wave is generated by applying an oscillating electric field of the same frequency. The pressure amplitude of the sound wave is measured as a function of distance d in sample solution.A well-known method of investigation is that suggested by Egger~,~~?~~ which has the big advantage that comparatively small volumes of solution are required for the measurements of T. Usually a frequency range of 103-108 cycles per second corresponding to relaxation times of 10-4-10-9 s. are covered in all these techniques.56*78-80 77 M. A. Collins and J. ROSS,J. Chem. Phys., 1978, 68, 3774. 78 S. Petrucci, J. Phys. Chem., 1967, 71, 1174. ?* P. Hemmes and S. Petrucci, J. Phys. Chem., 1968, 72, 3986. A. J. Campillo and S. L. Shapiro, Top. Appl. Phys., 1977, 18, 317. 242 Kriiger Some of the disadvantages of the ultrasonic technique are: (i) Broad-band measurements must be performed, which require several overlapping techniques.(ii) Sample cells must be inert against many solvents. (iii) There is a lack of commercial, ready-to-use equipment. (iv) There are large variations of ultrasonic absorption coefficients with frequency. (v) High stability requirements for temperature in the cavity are necessary. There are many applications of this technique, such as: (i) Ligand substitution processes in cation complexes. (ii) Ion-pair formation and desolvation of ions in aqueous solution of electrolytes.81+82 (iii) Hydrogen bond association proce~ses.~~-~~ (iv) Proton-transfer reactions or deprotonations, e.g.81y86 H30+ + Soh2-+HS0,-+ H20 or Et3NH+ + OH---+Et,N + H,O or in anti histamines. (v) Rotational isomerism. 889 89 (vi) Helix-coil transition processes.(vii) Studying properties of ionic hydrate melts.92 4 Applications for the Inorganic Chemist19993 In inorganic chemistry, we find many reactions between a positively charged metal ion (Mn+) and a negatively charged ligand (Xm-) to form a neutral complex. This kind of reaction is frequently described in the literature as very fast. A metal ion in aqueous solutions is always considered an hydrated ion, M(HSO)~~+.It is surrounded by one or more shells of co-ordinated water molecules. Interactions between metal ion and water molecules (ion-dipole attraction or dipole-dipole repulsion) occur amongst the co-ordination shells of the complex. If the metal ion (Mnf) is to combine with another ion of opposite G. Kurtze and K. Tamm, Acustica, 1953, 3, 33, ibid., 1954, 4, 380.8a M. Eigen, Z. Elektrochem., 1960, 64, 115. 83 J. Rassing, Adv. Mol. Relaxation Processes, 1972, 4, 55. 84 M. D. Joeston and L. J. Schaad, ‘Hydrogen Bonding’, M. Dekker Inc. 1974. 85 G. Allen and E. F. Caldin, Quarf. Rev. Chem. Soc., 1957, 11, 147. M. Eigen, Z. Phys. Chem., 1954, 1, 176. J. Gettins, R. Greenwood, J. Rassing, and E. Wyn-Jones, J. Chem. Soc., Chem. Commun., 1976, 1030. 88 R. 0. Davies and J. Lamb, Quurf. Rev. Chem. SOC., 1957, 11, 147. 89 J. Lamb, Z. Elektrochem., 1960, 64, 135. J. J. Burke, G. G. Hammes, and T. B. Lewis, J. Chem. Phys., 1965,42, 3520. 91 G. G. Hammes and P. B. Roberts,J. Am. Chem. SOC., 1969,91, 1812. 9a R. Carpio, F. Borsay, C. Petravic, and E. Yeager, J. Chem. Phys., 1976, 65, 29.93 M. Eigen, Angew. Chem., 1968, 80, 892. 243 Techniquesfor the Kinetic Study of Fast Reactions in Solution charge (Xnl-), the latter must penetrate the hydration shells, substituting successive water molecules in the different shells. Since water molecules in the inner co-ordination shell are bound most strongly, their substitution will be the slowest step of the process. Relaxation studies on very widely differing metal ions have confirmed this assumption. The stepwise substitution mechanism reveals itself in a relaxation spectrum with several time constants.94 The inorganic chemist is primarily interested in the substitution in the inner co-ordination + x-ax-inner sphere I IY I11 Figure 7 Substitution of metal ion in inner co-ordination sphere.The rate constants k, and kP3control the whole process, which may be of SN~ type. The former will or SN~ occur when H,O leaves the co-ordination shell prior to entering of X-, the latter occurs when X-enters the co-ordination shell to increuse the co-ordination number sphere, which has been found to be the rate-determining step. All other steps are very fast, i.e. 10-8 s. Let us consider the reaction of a metal ion (Mn+) with a ligand X-: In general, the rate constants are practically independent of the nature of the entering ligand, X-, and are mostly dependent on the nature, charge, and size of the metal ion, Mnf. SNIis therefore the predominant process. The smaller the radius and the higher the charge of the metal ion, the more strongly are the HzO molecules bound, and hence the more slowly does the substitution take place.Most rate constants lie in the range 103 to 109 s-1. Figure 8 shows the relationship between size of metal ion and rate constant in the substitution of HzO in non-9p R. G. W. Norrish and B. A. Thrush, Quart. Rev. Chem. SOC.,1956, 10, 149. Kriiger 1o'O-cst lo8 c I ul> lo6 loL lo2 1 Radius-' /A-' Figure 8 Relationship between size of metal ion and rate constant in H,O substitution Figure 9 Rate constants in H,O substitution (k/ss1).All rure-earth metal M3+ions have k values between lo8and lo7from left to right transition metals. Ligand substitutions of many transition-metal ions have been studied kinetically. Figure 9 shows a number of rate constants for H2O-substitition in the inner co-ordination sphere of metal ions.V2+ and Ni2+ ions have strikingly slow rates of substitution, whereas Cr2+ and Cu2+ ions are extremely liable to substitution. These two effects may be explained by (a) strong crystal and ligand-field stabilization, and (6) Jahn-Teller effects (the octahedral structure undergoes distortions). 5 Flash Photolysis2~94-96 Flash photolysis is a method whereby a non-equilibrium situation can be created in a reaction system in a short interval of time. It is one of the very few 95 R. G. W. Norrish and G. Porter, Discuss. Faraday SOC.,1956, No. 17, p. 40, G. Porter and M. W. Windsor, ibid., p. 178. 96 G. Porter, 2.Elektrochem., 1960, 64, 59.245 Techniquesfor the Kinetic Study of Fast Reactions in Solution methods for preparing and studying high concentrations of electronically excited molecules and radicals. The concentration of intermediates can be directly measured as a function of time, and their physical and chemical properties can be determined. I This technique falls within the category of large perturbations. An intense flash of visible or ultraviolet light over a few micro-seconds is used to initiate a disturbance photochemically. The intensity must be sufficient to produce a change in chemical composition that is measurable, but of short duration compared with that of the ensuing reaction. The time resolution is of the order of a few micro-seconds and could even be pico-seconds by the use of lasers.97 The course of a chemical change resulting from the flash is observed by fast photometric techniques.Reactions with half-times down to 10-6 s can be observed without difficulty.lB The following types of change can be studied: (i) The absorption of light may photolyse a molecule, producing a non-equilibrium state in solution. The reaction proceeds so that equilibrium is re-established. The principle is the same as that of relaxation methods in that a disturbance is initiated by physical means and the return to equilibrium is observed. (ii) The absorption of light may produce an electronically excited molecule, which in time is deviated and then returns to its original state.Here again the principle is that of relaxation. (iii) The excited molecules may react with some other molecules, producing permanent changes. The principle here is not relaxation but photo- chemical initiation of an irreversible reaction. The main difference from ordinary photo-chemistry is that the energy absorption occurs in a very short time and the subsequent fast changes are directly followed. The experimental arrangement can be set up in different ways. Figure 10 shows a typical 0ne.98 The factor which determines the fastest reaction that can be studied is the finite duration of the primary flash. The minimum duration of the flash to study half-times of about 10-5 s must therefore be about 10 ps. Rate constants that can be determined are about loll 1 mol-1 s-1.Only photochemically initiated processes can be studied. The change in the system on absorption of light is generally considerable. Some applications are given below : (i) Pro ton-transfer react ions. (ii) Haemoglobin reactions.99 (iii) Photochemically-initiated irreversible reactions. (iv) Studies of the triplet state.lO0 s7 U. Schindewolf, Chem. unserer Zeit, 1972, 6, 17. g8 N. K. Bridge and G. Porter, Proc. R. SOC.London, Ser. A, 1954, 244, 276. 99 Kugaku,Zokan (Kyoto), 1979, 80, 107. loo W. Potter and G. Levin, Photochem. Photobiol., 1979, 30, 225. 246 Kriiger (v) Iodine dissociation and recombination: I2 +hv -21 (vi) Reactions of the hydrated electron: e- +H2O -H +OH-10 kV 1 ZW kl lens filter Figure 10 Instrumental set-up for gash photolysis measurements Pulse radiolysislooJolcan sometimes also be used.It is the analogue of flash photolysis in radiation chemistry. A pulse of high-energy radiation such as X-rays or electrons takes the place of the light flash. Laserslo2 have also been introduced, replacing the flash lamps. 6 Electrochemical Methods2Jg The electrochemical determination of rate constants requires that at least one of the reactants of the chemical reaction under investigation be coupled to an electrode reaction as in equation (22). k, +e-A+B-+C (22)k-1 where C is the product of the electrode reaction A is electroinactive B is the depolarizer The rate of the overall process is proportional to the current flowing in the electrochemical system.It depends on the transport of A and B to the electrodes by diffusion and convection. A detailed analysis of the combined diffusion- convection reactions allows the determination of kl and k-1. Either the electrode potential U or the current i can be chosen as the parameter forced on to the system. Current or potential is measured as a function of time, or it may be a known function of time. lol P. Wardman, Rep. Prog. Phys., 1978, 41, 259. ' lo* Ref. 18, p. 89. Techniques for the Kinetic Study of Fast Reactions in Solution There are four electrochemical techniques by which rate constants can be determined : (i) The polarographic method. (ii) The rotating disc method.(iii) The potentiostatic method. (iv) The galvanostatic method. The polarographic method2J9J03J04 was the first of the many techniques and is well known for the determination of fast chemical reactions. With this method we are able to detect reaction times of about 10-8 s. A dropping Hg-electrode is used. A relationship between the diffusion current and the reaction current is employed for kinetic studies. In the case of an electrode reaction in which a reducible species B accepts an electron at the cathode, as shown in equation (23), +e-B+C we deal with a rapid and irreversible process. Species B which strikes theelectrode surface is removed from the system immediately. The current in the cell due to the reduction process is simply proportional to the rate at which species B reaches the surface.The motion of B may be caused either by diffusion, con- vection, or migration. Migration is avoided by adding KCI. Convection occurs if the solution is heated or stirred. Its effect is small compared to diffusion. Diflusion to the surface takes place because B is depleted in the vicinity of the electrode and a concentration gradient therefore builds up around it. This gradient will vary with time as indicated in Figure 11. Distance from electrode Figure 11 A plot of concentration against time in the polarogram. The diffusion current which is proportional to the concerttration gradient, is initially at a maximum and then falls steadily as the gradient decreases. The gradient can be visualized as confined to a difision layer of thickness 6.According to Einstein the displacement x varies with time t, i.e. x = 2Dt where D = diflusion coeficient lo8 K. Wiesner, Z. Elektrochem., 1943, 49, 164. lo4 1. M. Kolthoff and J. J. Lingane, ‘Polarography’, Interscience, Chap. 14, 1952. Kriiger According to Fick's law the flux is defined by equation (24): This flux may be converted to a diffusion current as shown in equation (25): iD = (Y)'FAc,, where A = surface and F = Faraday (25) For B participating in a dynamic equilibrium [equation (26)Jthe rate of arrival k,B +A (26)k-1 of B at the electrode surface depends both on diffusion and on chemical reaction, and with the equilibrium well to the right and the reverse reaction relatively slow, the current becomes completely under kinetic control.Here we have a reaction layer and not a diffusion layer. An average life time r for B is the period of time between its formation and its removal in the equilibrium. Species B will diffuse a distance p = (2DT)&during this life-time (Figure 12). .z 5)-e c C z 0 I 0 /-/I1I --CI-8-Distance from electrode Figure 12 A plot of concentration against distance from electrode. For k,s.k-, the relation between ir and iD is given by ir/(iD -it.)= C[(k-lt)/Ke]+where C = constant of unity order All species B which will be less than the distance p from the electrode surface (within the reaction layer) will be removed by reduction, whilst those beyond this distance will be transformed back to B before reaching the surface. The current at the electrode is then determined by the rate at which species B are formed in the reaction layer.Electrochemical measurements may be made in two ways; (i) either holding i or U constant and following the alteration of the other with time; or (ii) creating continuously a 'new' electrode surface to establish a steady-state situation. A time-averaged current is obtained for the applied voltage. 249 Techniquesfor the Kinetic Study of Fast Reactions in Solution Some of the more important applications of the electrochemical method may be mentioned here: (9 Polarography has been used quite extensively to study the dissociation and recombination of weak acids [equation (27)]: HA + H,O +A-+ H,O+ (27) This technique has produced rate constants which vary from lo1 to 1013I mol-1 s-1.For extremely fast reactions the reaction layer may well be less than lOOA thi~k.~05J06 (ii) Mutarotation of glucose (and sugars).107 (iii) Complex forming reactions.lO*-l10 (iv) Redox reactions of transition-metal i0ns.~ll-l14 7 The Fluorescence Method2 1153 116 Fluorescence techniques for the investigation of rapid chemical processes are concerned with the rates of reaction of photochemically-excited molecules. This is usually carried out by adding a substance which reduces the intensity of fluorescence. The attenuation of fluorescence can be used as a sensitive monitor for the progress of the reaction.The ‘quenching’ reaction competes with deactivation by other mechanisms that occur. A steady-state is set up, depending on the relative rates of these processes. The resulting fluorescence intensity can be observed without fast-recording apparatus. From the relation between the intensity and concentration of reagent, the rate constant for the reaction can be derived. The method is applicable only to excited molecules and very fast reactions. Many substances in solution fluoresce when irradiated continuously with ultra-violet light.l17 Examples are anthracene, /%naphthol, and quinine sulphate. Fluorescence occurs because some of the molecules are raised to an electronically excited state, by absorbing quanta of the exciting light, and re-emit light of visible or near-u.v.wavelength as they drop back to the ground state. Io5 R. Brditka, Z. Elektrochem., 1960, 64, 16. loo S. G. Mairanovskii, ‘Catalytic and Kinetic Waves in Polarography’, Plenum Press, 1968. lo’ J. M. Los, 1.B. Simpson, and K. Wiesner, J. Am. Chem. Sac., 1956, 78, 1564. lo8 L. Andrussow, J. Phys. Chem. (Leipzig), 1958, 208, 157. J. Biernat and J. Koryta, Collect. Czech. Chem. Commun., 1960, 25, 38. 110 J. Koryta, Z. Elektrcchem., 1957, 61, 423. *I1 Z. Popifil, Collect. Czech. Chem. Commun., 1953, 18, 337. lla J. Koryta and J. Tenygyl, Collect. Czech. Chem. Commun., 1954, 19, 839. 113 A. Arevalo, C. D. Silgo, J. C. Rodriguez Placeres, and T. Moreno, An. Quim., 1977,73, 923; and A. Arevalo, J.Acosta, J. C. Rodriguez Placeres, and J. R. Canitrot, An. Quim., 1973, 73, 930. 11* N. V. Guzhova, E. M. Kolosova, A. A. Kononenko, Y. U. N. Lejkin. J. N. Novodarova, and M. E. Vol’pin, Inwg. Chim. Actu, 1981, 50, 21. Pringsheim, ‘Fluorescence and Phosphorescence’, 1949. 116 E. F. Caldin, ‘Fast reactions in solution’, 1964, Blackwell Scientific Publications. 11’ E. J. Bowen and F. Wokes, ‘Fluorescence of Solutions’, 1953. 250 Kriiger Fluorescence and quenching is outlined in detail in F. Wilkinson’s book on pages 268-279.18 Relaxation times of the order of s have so far been determined. There are a number of applications of this method: (i) Rate constants have been determined for various proton-transfers, as illustrated by equations (28) and (29):1l89ll9 acridhe*.H20 * acridhe*.H+ + OH-(28) or R.OH*.H20 + RO*-+ H,O+ (&naphthol) It has been concluded that the transfer of a proton during these processes is effected in a time much less than the mean life-time of the excited molecules.(ii) Formation of hydrogen-bonded complexes.120 3-Hydroxypyrene and pyridine form a hydrogen-bonded complex [equation (30)]: ROH + py + ROH.py kr ROH* + py + ROH*.py Thq hydroxypyrene is fluorescent [equation (31)] and ki may be determined. [/to decreases with increasing concen tration. (iii) Dimerization and self-quenching. Solutions of anthracene in benzene are fluorescent121 [equation (32)1. A + hv -+ A -+ A* + hv’ (32) However, they show a marked decrease of fluorescence intensity as the concentration is increased.A stable dimer (dianthracene) is formed photo- chemically. The reaction scheme was found to be as shown in equation (33): A + A* -+AA* --f AA (33) Excited molecule A* reacts on collision with a ground-state molecule A. Rate constants for dimerization are of the order of 1Olo 1 mol-l s-l. 8 Nuclear Magnetic Resonance Techniques19. 22~123 A. Introduction.-Nuclear magnetic resonance (n.m.r.) may be exhibited by any compound whose molecule contains a nucleus with a spin. Such nuclei include the proton, the common isotopes of nitrogen and fluorine, and less common isotopes of carbon and oxygen. A spinning nucleus (like a spinning electron) 118 A. Weller, 2. Elektrochem., 1960, 64, 55. llS W.Biermann and H. J. Oel, Z. Phys. Chem., 1958, 18, 163. lZo K. H. Grellmann and A. Weller, 2. Elektrochem., 1960, 64, 145. 121 E. J. Bowen and A. Norton, Trans. Faraday SOC., 1939, 35, 44; E. J. Bowen and D. W. Tanner, ibid., 1955, 51, 475. 122 J. A. S. Smith, Quart. Rev. Chem. SOC.,1953, 7, 279. lZ3 Inorg. Chim. Acra, 1981, 48, 225. 251 Techniques for the Kinetic Study of Fast Reactions in Solution has a magnetic moment associated with the axis of the spin, and in a magnetic field it will be aligned somewhat like a bar magnet, with its moment at one definite orientation to the applied field. Nuclei may pass from one orientation to another by applying a second magnetic field, alternating with a particular resonance frequency. When the main field is of the order of 104 Gauss, the resonance frequency is in the radio region.The structure of the absorption spectrum and the width of the lines can be determined. These depend on the lifetime of the proton. They are altered whenever the compound undergoes a reaction altering the lifetime. Typical reaction times from 1 s down to 10-3 s have been determined. Rate constants may be calculated and values up to 1011 1 mo1-l s-l have been reported. N.m.r. techniques are discussed in more detail elsewhere,293 including the intensity and structure of absorption bands, line width and shape, and instrumentation used. B. Nuclear-spin Relaxation and Line-width.-The determination of rates of reaction by n.m.r. methods depends on measurements of line-shape and line- width.Energy is exchanged between spinning nuclei and their surroundings. A nucleus in the upper spin state can 'relax' to the lower, or vice versa. This mechanism is known as spin-lattice relaxation and its efficiency can be expressed by relaxation time 7'1, which is for most liquids of the order of 1 to 10 s. A nucleus can also transfer its energy to a similar neighbouring nucleus by a mutual exchange of spin. This is known as spin-spin relaxation and its corresponding relaxation time TZis also of the order of 1 to 10 s. These relaxation times 7'1 and Tz, are commonly measured by the pulse or spin-echo-meth~ds.~~~ The mathematical theory of line-width is discussed in E. F. Caldin's book 'Fast reactions in solution', pp.229-231 .3 Line-widths are commonly in the region of a few tenths of a cycle per second, and are appreciably larger than (nTz)-l. Line-widths are increased by reactions whose relaxation times are comparable with 7'2. When a reaction occurs the line-width is increased by Q'V = (vT)-~and so becomes (7rT2)-l + (~THA)-~ where THA is the mean life-time of the proton in the HA environment that gives rise to the line, and is also the reiaxation time for the forward reaction. The line-width may also be expressed in terms of the measured relaxation time T'2 as (.rrTz)-l. Hence equation (34) : 1 1---nay = ---1 THA T'2 Tt (34) where 2'2 is the relaxation for nuclear spin and THA is the relaxation for the re- action. lid H. Y.Carr, Phys. Rev., 1954, 94, 630. Kriiger C.Determination of Reaction Rates from N.m.r.Spectra.-Consider the proton exchange in a mixture of HA and HB, illustrated in equation (35): HA + H’B + H’A + HB (35) If there is no interaction, the proton resonance spectrum will show two peaks representing the characteristic chemical shifts of HA and HB. Changes in the n.m.r. spectrum occur as the rate of proton exchange increases. The two lines first broaden and approach each other and then coalesce to a single broad line, which subsequently narrows to a sharp line. For example, the proton resonance spectrum of a mixture of acetylacetone with acetic acid at various temperatures is shown in Figure 13. OH-signals are shown and. chemical shifts are given in p.p.m.The initial broadening of each line is due to the shortening of the mean life-times of HA and HB molecules and hence of lH nuclei in a given environ- ment and spin energy-level as a result of migration of protons. The finite lifetime of the nucleus in a given energy level leads to a finite line-width. Shortening of the lifetime by reaction correspondingly leads to an increase of line-width. -10.48 p.p.m. -11.05 p.p.rn. I I P-PJn -13.50 -9.67 Figure 13 Proton resonance spectra of’acetylacetone at various temperatures Theory shows that if the mean lifetime of the proton in a given environment HA is THA, broadening o’v of the HA line is (TTHA)-~ C.P.S. Hence, if the observed broadening is of the order of 1 c.P.s., the mean lifetime T is in the region of 0.1 to 1 s.As the exchange rate rises, the protons in the two environments no longer behave as independent systems. Lines approach each other, overlap, and then merge. The increase of width will not continue indefinitely. Separate lines are not observed unless 7 > k (dvo)-l where AVO is the frequency Techniques for the Kinetic Study of Fast Reactions in Solution difference of the two chemical shifts in absence of exchange. Since dvo is com- monly of the order of 100 C.P.S. for proton resonance spectra, coalescence corresponds to a lifetime of If: s, and measurements of subsequent narrow- ing permit estimates of lifetimes down to s. The fine-structure due to spin-spin interaction can similarly disappear. The proton resonance in liquid ammonia is a typical example.125 The quantitative derivation of rate constants may be obtained from the Bloch equation or its modification126 shown in equation (36), where v is the TZRate of absorption of energy K 1 -k 4,2~z2(V~-V)2 (36) frequency and vo its resonance value.For maximum absorption v = VO. The width at half-height is (xTz)-l C.P.S. Thus it is T2 which primarily determines the natural width of an n.m.r. line. Line shapes are computed for various T values and compared with the observed line shapes until a match is obtained. The mean lifetime is related to a first-order rate constant. The mean lifetime of the protons in the HA environment is given by equation (37). [HA1 1 THA = -d[HA] = -(37)-dt kHA In general, the quantity measured by the n.m.r.techniques is the mean lifetime 7A of some particular molecular species. A first-order rate constant is found as kA = 7.4-l. All measurements are made on systems in chemical equilibrium. The equilibrium, however, is not disturbed. The method is especially useful for investigations of symmetrical exchanges. Reactions that are too fast for isotopic methods can be studied. Very detailed information about the mechanism may be obtained since the n.m.r. spectrum identifies directly the atoms that exchange. Values for 7dVO must be in the region of 10 to 0.01 so that changes in line-shape are noticeable. For nuclei other than the proton, chemical shifts and line-widths are larger, and shorter lifetimes can therefore be determined; lifetimes of about s have been found with 170 and 10-5 s with 63Cu.The n.m.r. technique has applications to a considerable number of reactions, of which a few are mentioned below: (i) proton transfer: (a) in alcohols and aqueous alc~hoIs.~~~J~* (6) in liquid ammonia.lZ9 135 R. A. Ogg,jun., Discuss. Faraday SOC.,1954, No. 17, p. 215. lae H. M. McConnell, J. Chem. Phys., 1958, 28, 430. J. T. Arnold, Phys. Rev., 1956, 102, 136. la* Z. Luz,D. Gill, and S. Meiboom, J. Chem. Phys., 1959,30, 1540. R. A. Ogg, jun., J. Chem. Phys., 1954, 22, 560. 254 Kriiger (ii) Inversion of configuration.130-132 (iii) Hydration of cations.133J34 (iv) Electron transfer reactions.135-138 (v) Complex formation rea~tions.1~~J40 (vi) Exchange processes.141 ~14~ 130 S.L. Miller, J. Am. Chem. Soc., 1960, 82, 1265. 131 L. J. Durham and H. S. M0sher.J. Am. Chem. Soc., 1961,83,4357. 13p A. Loewenstein, J. F. Neumer, and J. D. Roberts, J. Am. Chem. SOC.,1960, 81, 3599. 133 R. E. Connick and R. E. Poulson, J. Chem. Phys., 1959, 30, 759. lJ4 G. M. Rothberg, J. Chem. Phys., 1961, 34, 2069. 136 H.M. McConnell and H. E. Weaver, jun., J. Chem. Phys., 1956,25, 307. 136 C. R. Giuliano and H. M. McConnell, J. fnorg. NK/. Chem., 1959, 9, 171. B. Zaslow and R. L. Miller, J. Am. Chem. Soc., 1961, 83, 4738. 138 A. Schliiter and A. Weiss, Ber. Bunsenges, Phys. Chem., 1981, 85, 306. 13s J. W. Akitt and A. Farthing, J. Chem. SOC.,Dalton Trans., 1981, 1606. D. L. Rabenstein, S. J. Backs, and A. A. lsab, J. Am. Chem. Soc., 1981, 103, 2836. 141 A. Ponpom, E. Picquerard, C. Coupry, J. Belloc, and P. Dizabo. J. Chem. Res., 1981,238. 142 P. Mirti, M. C. Gennaro, and C. Casalino. Inorg. Chim. Acta, 1981, 53, L257. 255

ISSN:0306-0012    

DOI:10.1039/CS9821100227

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

R. A. ROBINSON MEMORIAL LECTURE Thermodynamics of Hydrogen-bonded Liquids By R. H. Stokes* DEPARTMENT OF CHEMISTRY, UNIVERSITY OF NEW ENGLAND, ARMIDALE, N.S.W., 2350, AUSTRALIA 1 Introduction While water is the most important of hydrogen-bonded liquids, it is also the most complex. This is because its molecule has two hydrogen bond acceptor sites and two protons which can participate in other H-bonds. This gives rise to a three-dimensional hydrogen-bonded network structure which is quite regular in ice and less well-defined but still vitally important in the liquid state. The hydrogen-bonding in alcohols is somewhat simpler because there is only one hydroxy-group per molecule, so that the ‘polymers’ are chains and rings rather than three-dimensional structures.For both water1 and alcohols2 there have been striking recent advances in understanding the structure of the pure liquids as a result of computer studies based on the use of potential functions which more or less accurately represent both the charge distributions giving rise to hydrogen bonding, and the other intermolecular effects. These studies, how- ever, remain very demanding on computer time and there is still an important place for the classical approach of measuring solution properties and inter- preting and correlating the results in terms of chemical or ‘pseudochemical’ models of association through hydrogen bonds. In principle, such studies depend on the proposition that at extremely high dilution in an inert solvent the solute consists of single molecules without H-bonds.As the concentration is progressively increased, a variety of hydrogen-bonded clusters appear, accompanied by changes in the thermodynamic, spectroscopic, and dielectric properties. (Experimentally, the reverse process of progressive dilution of the pure liquid is of course adopted). In this context, ‘inert’ implies that the diluent does not modify the association process except by increasing the average separation of the associating molecules. In the case of water, the associa- tion is so strong that the only solvents which will dissolve useful amounts of it cannot be described as inert towards it. Dioxan, for example, which has a low dielectric constant and is often used with water as a mixed solvent for electrolytes, is seen by a water molecule not as non-polar, but as having two oxygen atoms *Emeritus Professor.This review is an expanded version of the inaugural R. A. Robinson Memorial Lecture given at Imperial College London, the University of Newcastle-upon-Tyne, and St. Andrews University in October, 1981. R. 0. Watts and I. J. McGee, ‘Liquid State Chemical Physics’, Wiley, New York, 1976. a W. 1. Jorgensen, J. Am. Chem. SOC.,1981, 103, 335. 257 Thermodynamics of Hydrogen-bonded Liquids which can accept hydrogen bonds. Even benzene, in which water has a modest solubility, must be considered as providing in its .rr-electron system an environ- ment which lowers the energy of a water proton and hence competes with the hydrogen bonding between water molecules.The only solvents which could not be expected to show such effects are those of paraffinic type, and in these water is so insoluble that dilution of a saturated solution can only occur at concentrations too low for accurate thermodynamic study. Spectroscopy can however give some information at these low concentrations. The study of alcohols dissolved in non-polar solvents is much easier. Some of the lower alcohols are only partially miscible with some paraffins at room temperature, e.g. methanol + hexane and ethanol + hexadecane, but even in these cases the range of miscibility at the low alcohol compositions includes the concentrations at which most of the alcohol association occurs. The fact that incomplete miscibility can occur, however, is evidence that interactions between the alcohol and the solvent must be considered, for it is known3 that the associa- tion of a solute in an otherwise ideal solution cannot give rise to phase separation. The proof of this is so neat as to be worth repeating in a paraphrased form: No matter how assorted a set of association products is formed, the equili- brium constants for their formation must all be positive, and the stoicheio- metric concentration of the solute can therefore be expressed as a polynomial in the concentration of the free monomer, with all coefficients positive. Hence the stoicheiometric concentration must be a monotonicially rising function of the free monomer concentration.Now if the various association products and the solvent form an ideal mixture, the stoicheiometric solute activity is equal to the monomer mole fraction (or concentration, if volume-scale ideality is assumed).Since this rises monotonically with the stoicheiometric solute concentration, there can never be two different solution compositions with the same solute activity, and hence never any phase separation. My interest in liquid mixtures arose naturally out of work with Robinson on electrolytes. In 1948 we had proposed an explanation of the increase in the activity coefficient of many electrolytes at high concentrations in terms of a hydrated-ion model.4 We treated the solution as one of hydrated cations, free anions, and free water, and assumed that the solution behaved as an ideal solution of these species except for the effects of ion-ion interactions, which were described by the usual Debye-Huckel formula.This treatment, though it had some defects, did a good deal to make hydrated ions respectable entities again after some decades of ill repute. A valuable modification of the theory was made by Glue~kauf,~ who pointed out that we had implicitly assumed mole fraction ideality for the entropy of mixing of the solvated ions, anions, and free water molecules. He suggested that the entropy of mixing would be more appropriately calculated by the volume-fraction statistics of Flory and Huggins, E. W. Washburn, Trans. Am. Electrochem. SOC.,1912, 22, 330. R. H. Stokes and R. A. Robinson, J.Am. Chem. SOC., 1948, 70, 1870. E. Glueckauf, Trans. Faradav SOC., 1955, 51, 1235. Stokes and showed that this led to a more consistent set of hydration numbers. This argument was based on the fact that the hydrated ion must be considerably larger than a water molecule. It was not clear, however, that the Flory-Huggins formulae developed for high polymer solutions were necessarily right for mixtures of globular molecules such as hydrated ions and water. I therefore suggested to K. N. Marsh at Armidale that he examine experi- mentally the entropy of mixing of unreactive globular molecules of similar chemical nature, but of substantially different sizes. This he undertook with great success, beginning with some highly precise vapour pressure measurements on mixtures of octamethylcyclotetrasiloxane with benzene and carbon tetrachloride.6 Though these were accurate enough to permit a separation into enthalpy and entropy effects via their temperature dependence, we realized that to supplement the vapour pressures by direct calorimetric measurements of enthalpies of mixing would greatly expedite the work.We therefore developed, from H. C. Van Ness's brilliant concept7 of the isothermal dilution calorimeter, a more refined and precise instrument using mercury displacement instead of a solid piston to eliminate the need of a vapour space.8 Subsequently, similar continuous dilution techniques were developed for the measurement of excess volume~,~-l~compressibilities,12 dielectric constants,13 and vapour pre~sures.1~ Marsh and collaborator^^^ continued the study of globular molecules, working especially with cycloparaffins, while my group took advantage of the new techniques to study hydrogen-bonding interactions in solutions of alcohols in non-polar solvents.16-2' 2 Experimental Techniques The thermodynamic properties of the most direct interest for the study of association equilibria are the activity coefficients of the alcohol and the solvent, and the enthalpy of dilution from a given composition to infinite dilution of alcohol.The dielectric constant of the solution, which yields the apparent dipole K. N. Marsh, Trans. Faraday Sor., 1968, 64, 883. 'C. G. Savini, D. R. Winterhalter, L. H. Kovacs, and H.C. Van Ness, J. Chem. Eng. Data, 1966, 11, 40. * (a) R. H. Stokes, K. N. Marsh, and R. P. Tomlins, J. Chem. Thermodyn., 1970, 2, 751; (b) M. J. Costigan, L. J. Hodges, K. N. Marsh, R. H. Stokes, and C. W. Tuxford, Ausr. J. Chem., 1980, 33,2103. @ R. H. Stokes, B. J. Levien, and K. N. Marsh, J. Chem. Thermodvn., 1970, 2, 43. G. A. Bottomley and R. L. Scott, J. Chem. Thermodyn., 1974,6, 973. l1 M. K. Kumaran and M. L. McGlashan, J. Chem. Thermodyn., 1977, 9, 259. I' M. B. Ewing, K. N. Marsh, and R. H. Stokes, J. Chem. Thermodyn., 1972, 4, 637. lS R. H. Stokes and K. N. Marsh, J. Chem. Thermodyn., 1976, 8, 709. R. P. Tomlins and K. N. Marsh, J. Chem. Thermodyn., 1976, 8, 1185. l5 M. B. Ewing and K. N.Marsh,/. Chem. Thermodyn., 1974,6, 1087 (and earlier papers there cited).l* R. H. Stokes and C. Burfitt, J. Chem. Thermodyn., 1973, 5, 623; 1975,7, 803. l7 R. H. Stokes and M. Adamson, J. Chem. Thermodyn., 1976, 8, 683. R.H. Stokes and M. Adamson, J. Chem. Sor., Faraday Trans. 1, 1977, 73, 1232. l@R. H. Stokes, J. Chem. Sor., Faraday, Trans. I, 1977, 73, 1141. loH. T. French, A. Richards, and R. H. Stokes, J. Chem. Thermodyn., 1979, 11, 671. *I H. T. French and R. H. Stokes, 1.Phys. Chem., 1981, 85, 3347. 259 Thermodynamicsof Hydrogen-bondedLiquids moment of the alcohol as a function of composition, is also informative in a more qualitative way. Excess volumes (volume change on mixing) are also of interest, but do not in general add much to the conclusions from the enthalpy measure- ments.Among non-thermodynamic properties, infra-red and n.m.r. spectroscopy have been extensively studied, though the conclusions show a good deal of depen- dence on assumptions about the assignment of frequencies to particular modes. One of the most powerful techniques of the electrolyte chemist is the measure- ment of the e.m.f. of reversible cells (a method extensively and successfully used by Robinson). This technique gives a direct measure of free-energy change, and has provided extremely accurate data on activities for electrolyte solutions. It is not available for non-electrolyte solutions, and we consequently depend for non- electrolyte activities almost entirely on methods involving phase equilibria :vapour pressures, cryoscopy, and gas chromatography are the most familiar.(The qualification ‘almost entirely’ allows for the possibility of obtaining usefully accurate data from light-scattering studies). A. Vapour Pressure Measurements.-Until the last decade or so, activities of liquid mixtures were usually measured in an equilibrium sti11,Z2 where the condensed vapour is recycled to the boiler until a steady state is reached, whereupon both the liquid phase in the boiler and a condensed sample represent- ing the vapour are analysed. The partial pressures of both components can then be calculated from the known external total pressure. This method does not work well at high dilutions of one component, and if a set of isothermal values is required the pressure must be accurately adjusted and controlled.Barker23 pointed out that only the measurement of the total pressure is really needed, provided it covers a range of liquid compositions at one temperature, because the partial pressures of the components are connected via the Gibbs-Duhem relation. This eliminates the need for the analysis of the vapour phase. The method of calculation proposed by Barker was to assume a suitable functional form for the excess free energy of the solution, with a number of adjustable parameters (e.g. coefficients of a polynomial in the mole fraction). Expressions for the component activities and hence partial pressures are then obtained by appropriate differentiation of the free-energy function. The resulting calculated total pressures are then made to agree with the observed values by a least-squares procedure for adjusting the parameters.Although this method works well for most organic liquid mixtures, it proves difficult with alcohol solutions because of the extremely unsymmetrical nature of the partial excess free-energy functions, which arises from the breaking of the hydrogen bonds at high dilutions. An alternative method, successfully used by Marsh6 for simpler cases, depends on the fact that modern designs of vapour pressure apparatus allow the progressive addition of very small amounts of alcohol to a large amount of solvent, with a vapour pressure measurement at each stage. At the lowest concentrations (say below 0.002 mole fraction of alcohol), the solvent activity coefficient is so close 82 1.Brown, Ausr. J. of Sci. Res., 1952, SA, 530. a3 J. A. Barker, Aust. J. Chern., 1953, 6,207. 260 Stokes to unity that the solvent partial pressure can be calculated as if it were ideal; the subtraction of this from the observed total pressure therefore yields the alcohol partial pressure. From this, the alcohol activity coefficient can be calculated. From a few such measurements at high dilutions, an extrapolation can be made to obtain the activity coefficient of the alcohol at infinite dilution. Integration of the Gibbs-Duhem equation then gives a more correct solvent activity coefficient, which is recycled until both solute and solute activity co- efficients are unchanged. The next solvent addition is then made, and the in- tegration and iteration are continued.This method ceases to work if a maximum in the total vapour curve is reached, but this is at higher alcohol concentrations where either the Barker method can be used, or the curve can be completed by the same marching procedure starting from the pure alcohol end. Tucker and Christian24125 have obtained excellent results by the simple device of using the higher paraffins as solvents. Since these have very low vapour pressures, the measured pressure is the alcohol vapour pressure alone, which greatly simplifies the calculations. In converting the partial pressures to activity coefficients, it is necessary to have reasonably good estimates of the second virial coefficients of the pure and mixed vapours at the temperatures of interest.26 Lack of these is sometimes a limiting factor in the accuracy of the results.For the alcohols especially, virial coefficients are difficult to measure because (a) there is still some hydrogen- bonding in the vapour phase, so that the virial coefficients are large and highly temperature-dependent ; (6) virial coefficients higher than the second have to be considered; and (c) the problems of adsorption are enhanced for highly polar molecules. The routine measurement of solution vapour pressure has been greatly facilitated by technological developments such as the differential capacitance manometer and the quartz spiral pressure gauge, which make it possible to measure vapour pressures without the vapour coming into contact with mercury.The continuous dilution method1* has also led to a considerable speeding up of the measurements, avoiding the need for much weighing of liquids in ampoules which are then broken inside the vapour pressure cell. It goes without saying that a high degree of purity is required for the liquids used; purity testing has been greatly facilitated by the commercial availability of gas chromato- graphic analysers. Freedom from dissolved gases is essential in vapour pressure measurements; this is most readily ensured by repeated vacuum sublimation in between passages of the vapour through a molecular sieve. Temperature control to +_ 0.001 K should be aimed at, and, for the sake of cross-checking with other work, the actual temperature should be known in terms of IPTS 68 standards within at worst 0.005 K.Since excess enthalpies can now be measured very easily, there is little point 24 E. E. Tucker and S. D. Christian, J. Phys. Chem., 1977, 81, 1295. E. E. Tucker and E. D. Becker, J. Phys. Chern., 1973, 77, 1783. ‘Chemical Thermodynamics’, ed. M. L. McGlashan, (A Specialist Periodical Report) The Chemical Society, London 1978, Vol. 2. 261 Tiiermodynamics of Hydrogen-bonded Liquids making vapour pressure or other free-energy measurements at more than one temperature. This may be chosen to give the range of pressures best suited to accurate measurement with the available equipment; in the case of alcohol solutions18~20~21we have used 45 and 52°C.Freezing-point measurements provide another source of free-energy in-formation. Provided that the solid phase wjlich freezes out is the pure solvent, its chemical potential-a measurable function of temperature-must be equal to that of the solvent in the liquid phase. Almost all aqueous electrolyte solutidns satisfy this proviso, but the probability of solid solution formation is higher in the case of non-elecpolyte liquid mixtures. Thus work in Armidale has shown,27-3O by comparing the results of freezing-point and vapour pressure studies, that ethanol in p-xylene gives a pure xylene solid phase, but ethanol in cyclohexane appears to deposit a very dilute solid solution of ethanol in cyclo-hexane. The other standard objection to freezing-point measurements, that they yield results only at or near the freezing-point of the solvent, is also a minor restriction, because it is somewhat more difficult.to obtain high precision temperature control below room temperature than above it. B. Measurement of Enthalpy of Mixing.-Results of calorimetric experiments on the mixing of two liquids are usually expressed in terms of the excess enthalpy, HE,which is defined as the heat absorbed when one mole of solution is formed from the appropriate amounts of components. The term ‘excess’ refers to the excess over the ideal solution value, which is of course zero. (Excess volumes are similarly defined, and again the volume change is zero for an ideal solution. The free-energy change and entropy change on mixing are however non-zero for an ideal solution).Regular solution theory indicates that for normal liquid mixtures the excess enthalpy should be of the approximate form [equation (l)]: HE-D(v,v,/Vrn)XlX2 (1) where VI and V2 are the molar volumes of the components, Vm is the mean molar volume, and x denotes mole fraction; D is a parameter of the system. (Strictly speaking, the theory refers to mixing under conditions of constant volume, and to the excess energy UE rather than to the excess enthalpy). The function HE/x1x2 is therefore often used as a convenient one for plotting and tabulation, as it often changes only slowly with composition, and unlike HEitself, it remains finite instead of going to zero at zero XI or x2.In the alcohol solutions of present interest, however, a slightly different function is often more useful. This is the excess enthalpy per mole of alcohol, HE/x~,which can also be called the apparent molar excess enthalpy of the alcohol. It also has a finite value at infinite dilution of alcohol, and its value relative to this limit is the relative apparent molar enthalpy, a quantity more *’R. H. Stokes, J. Solution Chem., 1974, 3, 671. R. H. Stokes and M. Adamson, J. Chem. SOC.,Faraday Trans. I, 1975, 71, 1707. 89 R. H. Stokes and H. T. French, J. Chem. SOC.,Faraday Trans. I, 1980,76, 537- 30 R. H. Stokes and M. Adamson, J. Chem. SOC.,Faraday Trans. I, 1977, 73, 1236. Stokes familiar to electrolyte workers than to liquid-mixture specialists [equation (2)]: 30.25. c L2 75 20. %a W z 15. 10.0.000 0.005 0.010 xA Figure 1 Extrapolation to limiting apparent molar enthalpy. Ethanol in cyclohexane at 10°C. The broken line shows how an erroneous extrapolation could be obtained in the absence of data below 0.002 mole fraction. The importance of this quantity for alcohol solutions is that its main component is the enthalpy of formation of all the H-bonds present per mole of alcohol, at the composition in question. The accurate determination of the limiting value is thus of considerable importance, and, just as with electrolytes, it requires measurements down to quite low concentrations to yield a reliable extrapolation. Figure 1 illustrates this point. If the measurements were made down to a mole fraction of 0.002, there would be an apparently straight extrapolation to approximately 26.7 kJmol-1 as shown by the broken line.In fact, this extra- 263 Thermodynamics of Hydrogen-bonded Liquids polation is badly in error, as the linear behaviour arises because there is a slow inflection in that region. When the measurements are extended down to another order of magnitude dilution, the much lower value of 24.0 kJmol-1 is obtained. It is only since the introduction of the isothermal dilution calorimeter by Van Ness, and its modification for higher precision work in Armidale, that it has been possible to make these extrapolations reliably. Because the amount of alcohol involved is so small at these dilutions, it is not added directly but in rhe form of a solution already diluted to say 0.01 mole fraction in a previous experiment, in which HEwas also measured.Then the quantity HE/x~shows its usefulness in another way, for if final refers to the result of the second dilution, and initial to the first, the relation [equation (3)]: (HE/XA)Iinal = (HE/XA)initial + dH/nA (3) is exact. Here AH is the enthalpy increase in the second dilution, and nA is the number of moles of alcohol which is added (in the form of stock solution) to the pure solvent to make the final solution. In some cases, we have even used a third stage of dilution to give the highest accuracy in the extrapolation. The point of this method is that one need never make calculations involving an injection of less than 2 cm3 of liquid from the piston-burette containing the alcohol or the stock solution, so that the metering error is below 0.1%.By means of a suitable arrangement of micro-ball-valves and mercury burettes, it is possible to transfer a portion of the final solution from one run into a piston- burette without exposure to air and hence possible contamination with moisture, or evaporation losses. This is then used for injection into the solvent in the next stage . One of the features of the isothermal dilution calorimeter is that it has no vapour space. As the alcohol is added, mercury is expelled from the mixing vessel to make room for it. The sensitivity of the thermistor measurement of the solution temperature is such that it is necessary to adjust the mercury levels to a constant head of pressure on the solution, otherwise the adiabatic com- pression of the liquid results in a detectable heating effect, as can be shown in a blank run. The importance of having no vapour space in the calorimeter arises from the fact that the latent heat of vaporization of volatile components may be large compared to the heat of mixing.In this connection it should be noted that the volatility of alcohols at high dilutions is much higher than would be estimated on an ideal solution basis. For example, the activity coefficient of ethanol at x1 = O.OG1 in cyclohexane, at 6.7"C,is 137-times the ideal solution value. This is directly due to the fact that nearly all the H-bonds are broken at this dilution, so the vapour pressure is nearly that of a solution of alcohol monomer, a much more volatile substance than the pure liquid alcohol. C.Dielectric Constant Measurements.-As with other properties, it is necessary to carry these measurements to high dilutions in order to get a true picture of the way the apparent dipole moment varies with concentration. The method we have usedl3 is again based on the principle of continuous dilution. The dielectric Stokes cell itself is the annular space between two stainless steel cylinders.The liquid is circulated by a totally immersed centrifugal pump from the cell to a mixing-vessel and back to the cell. The mixing-vessel is partly filled with mercury, and the alcohol, or a dilute solution thereof, can be added to the vessel from a piston- burette, displacing mercury.In this way the small capacitance changes resulting from the addition of as little as a few mg of alcohol can be measured with good accuracy, as they are not obscured by the circuit disturbances and variable air-contamination which result from making measurements first on the solvent and then on solutions prepared externally. Using an audio-frequency (10 kHz) capacitance bridge, we have a sensitivity of 0.001pF in a total capacitance of about 200 pF. It is, however, not possible to get accurate measurements above about 0.2 mole fraction of alcohol as the increasing dielectric loss leads to difficulties in balancing the bridge.3 Interpretation of the Thermodynamic Data in Terms of Association Marsh’s work6915 indicates that the entropy of mixing for globular molecules of different sizes lies between the values given by the mole fraction ideality expression [equation (4)] ASm/R = -ZXi In Xi (4) and the volume fraction statistics (Fl~ry-Huggins)~~ expression [equation (5)] I-.where Xi denotes the mole fraction of component i, $i its volume fraction, and ASrnis the entropy of mixing per mole of solution. In most cases a good represent- ation of the observed entropy of mixing is given by the statistical-mechanical theories of the mixing of hard spheres (or other convex particles) which have been developed during the last two de~ades.3~ The experiments also show that the enthalpy of mixing is of comparable importance to the entropy in its effect on the free energy.For calculations of association equilibria in alcohol solutions, we need to know the activity of each species involved. The only chemical potential directly accessible from experimental data is that of the monomeric alcohol, which is also the conventional chemical potential of the solute.33 (The molecular reason for this equality is that the vapour pressure of the alcohol, used in obtaining the chemical potential, is corrected by the use of virial coefficients to that of the ideal monomeric vapour). In estimating the activity coefficients of the associated species we must therefore be guided by theory and by analogy with unassociated solutes.Though statistical-mechanical theories of liquid mixtures have had a good deal of SUCC~SS,~~~~the algebraic form of the activity coefficient expressions is 31 P. J. Flory, J. Chem. Phys., 1941, 9, 660; M. L. Huggins, J. Ph-vs. Chem., 1941, 9, 440. N. S. Snider and T. M. Herrington, J. Chem. Phys., 1967, 47, 2248. 33 I. Prigogine, ‘The Molecular Theory of Solutions’, North-Holland, Amsterdam, 1957, p. 313. 94 K. E. Gubbins and C. H. TWU, Chem. Eng. Szi., 1978,33, 863. 35 C. H. Twu and K. E. Gubbins, Chem. Eng. Sci., 1978, 33, 879. Thermodynamicsof Hydrogen-bonded Liquids I 0.0 0.5 1.0 xA Figure 2 Activity coeficients of component A in typical mixtures of unreactive globular molecules. Points, experimental; lines calculated from equation (1) with the indicated value of the interaction parameter D.In order from top the systems are: A B D}J ~rn-~Temperature1OnL 2,2-dimethylbutane cyclo-octane 5.53 25 1 ,Cdimethylbenzene benzene 2.71 35 cyclopentane cyclo-octane 1.23 25 cyclopentane octamethyl-5.75 25 cyclotetrasiloxanecarbon tetrachloride fetrakis (2-butoxy)- 1.83 35 ethyl silane very complex even for a two-component solution, and at least one parameter (connected with the interaction energy) has still to be fixed from the experimental data. It is therefore worthwhile examining how well the classical theory of activity in non-electrolyte mixtures due to Hildebrand36 and Scatchard3’ describe the 38 J. 1%.Hildebrand and S. E. Wood, J.Chem. Phys., 1933, 1, 817. 37 G. Scatchard, Chern. Rev., 1931, 8, 321. Stokes activity coefficients in mixtures of globular molecules. Using the Flory-Huggins expression for the entropy of mixing this theory gives for a binary mixture [equation (611: In a, = In + 42(1 -Vl/Vz)+ VP$~~((~,-6,I2/RT (6) Here a1 is the activity of component 1 (relative to the pure liquid 1 as standard state)l; $1 and $2 are volume fractions; V1and V2are molar volumes; and 612and 622 are the cohesive energy densities of the component liquids. If equation (6) held exactly, it would be possible to calculate activities in any mixture from the properties of the pure liquids, without appeal to any measurements on the mixture. In practice this is not possible with useful accuracy, but if the quantity (61-82)2 is treated as a parameter D for a particular system, to be determined from the activity at one composition, then equation (6) is a good one-parameter representation of the whole activity coefficient curve of mixtures of globular molecules of different sizes.It does not do well with the excess enthalpy (heat of mixing) and excess entropy considered separately. This failure has been explained by Hildebrand and Scott;38 is it due to the fact that deficiencies in the theory, leading to first-order errors in the enthalpy and entropy separately, cancel in the free energy leaving only second-order errors there. This cancellation occurs essentially because the free energy is a minimum at equilibrium.A. Species Activity Coefficients for Associated Solutes.-In using equation (6) to estimate activity coefficients for association equilibria in solutions of alcohols, two further reasonable approximations are required : (a) that the volume of an associated cluster of i molecules is i-times that of a single molecule, and (b) that the non-H-bonding interactions between the alcohol species and the solvent can all be described by a single parameter D [corresponding to (61-82)2 in equation (6)]. Assumption (a)is demonstrably inexact, since the limiting partial volumes of alcohols in non-polar solvents are several cm3 mol-l greater than those of the pure alcohols; nevertheless the resulting error in activity coefficient ratios should be minor.Assumption (b)amounts to saying that the H-bonding and Van der Waals interactions may be treated independently. With these assumptions, the molarity-scale activity coefficient yi of the i-fold associated species can be deduced from the generalisation of equation (6)to several components as: where VAis the (conventional) molar volume of the alcohol and VB that of the solvent; $I?is the volume fraction of solvent. The summation &k refers to all the distinct species present at equilibrium including the solvent. 38 J. H. Hildebrand and R. L. Scott, ‘Solubility of Non-Electrolytes’, 3rd Ed. Dover, New York, 1967 (Chap. VIII). Thermodvnamicsof Hydrogen-bonded Liquids B. Association Equilibria.-The important point about equation (7) for present purposes is that in calculating a stepwise equilibrium at any chosen solution composition (where ci and $B are fixed), we have Kt = (Cl/Ci-, .cA(yi/yi-1 .Yl) (8) and by equation (7) the activity coefficient factor is always unity, even though the separate activity coefficients are not.This vastly simplifies the calculation of the equilibrium properties. The standard state used here is the one used in dealing with electrolyte equilibria, i.e. a hypothetical one molar solution of the species in question, magically prevented from dissociating or associating. The stoicheiometric alcohol concentration CA can be expressed as a polynomial in the monomer concentration c1 and the overall formation constants of the various i-fold associated clusters (i-mers) CA = ~1 + Ziflicit i=2,.. . (9) Here pi is the formation constant for species i. C. The Free Monomer Concentration.-Another equation connects the free monomer concentration ci with the measured alcohol activity. This is based on the the0rem3~ that in an association equilibrium the stoicheiometric chemical potential of the associating substance is equal to the chemical potential of the free monomer. Therefore if we can calculate the activity coefficient of the monomer in the mixture, we can express the monomer concentration ci in terms of the measured activity. Though the activity coefficients given by equation (7) cancel from the equilibrium constant expressions, it is necessary now to know the value of the parameter D in order to calculate the monomer activity co- efficient.Then equation (1) leads to equation (10) wherefAO is the limiting alcohol activity coefficient at zero alcohol concentration. The mole-fraction-scale activity coefficienrs fA and fB of the alcohol and the solvent can also be expressed in terms of c1 and the 'species sum' [equation (1l)] which is summed over all the alcohol species (but not the solvent): In fA -lirn(1nf~)= In(c,/ca) + In(] + CA(VB-VA)) XA+O + VA(CAVA/YB-Zci) + DYA(W -IYRT (12) + CAVA+ DVB+A'/RT (13)In fB = In(1 + cA(vB -YA)}-VBZC~ Equation (10) shows that the vital quantity ci can be evaluated from the activity coefficient data, without any detailed knowledge of the association process, provided equation (7) holds and that D is known.In principle, it is then possible to use equation (9) to determine the p's. In practice, some simplifying Stokes assumption must be made, as two infinite sets of arbitrary equilibrium constants can be neither calculated nor interpreted. D. Relations between Successive Constants.-The simplest assumption that can be made about the association process is that all the stepwise association con- stants are equal. If this hypothesis is to mean anything, it must be true at all temperatures; hence the enthalpies of formation of the successive hydrogen bonds must be the same. From this model one can obtain an expression for @H the apparent molar enthalpy of the alcohol, relative to infinite dilution. It is of course a function of the alcohol concentration, the equilibrium constant, and the enthalpy of formation of the H-bond, and it can be shown that it cannot give rise to an inflexion in the curve of @H versus concentration. Such an inflexion is invariably found, as shown in Figure 4,so the assumption of equal stepwise constants is too simple.E. Cyclic Species.-Probably the most important additional effect is the formation of cyclic associated species. If a strain-free cyclic hydrogen-bonded cluster is formed from an open chain, there is an enthalpy loss corresponding to the extra hydrogen-bond formed, though the number of molecules in the cluster is unchanged. The entropy change arises mainly from the loss of flexibility of the ring compared with the chain, and would not be expected to be as great as for the the addition of another molecule to a chain.Hence the cyclic forms should be favoured once the cluster gets large enough to permit strain-free cyclization. Cyclic forms with as few as two molecules have been suggested, but these must have an 0-H * -0angle much more acute than the 180" which is the lowest energy configuration for the hydrogen bond. (This favoured configuration is seen very clearly in the structure of ice, where the mean position of each proton is on the line of centres of two oxygen atoms). J0rgensen,3~ using an admittedly simplified but still quite reasonable potential function (in which the H-bonding arises from Coulomb interactions of a charge distribution) finds that the anti-parallel configuration of the methanol dimer has an energy some 6k.J higher than the open or end-on configuration.This would suggest that the open form of the dimer would be favoured by a factor of about 10 at room temperature. Since the H-0 --* H angle has approximately the tetrahedral value (109 ") it follows that five or more hydroxy-groups are necessary to give a strain-free ring. A ring of four should not be severely strained, but it seems reasonable to ignore the contributions of two- or three-membered rings to the thermodynamic properties. The clearest indication that cyclic groups are important comes from dielectric constant studies on the solutions. The typical form of the apparent dipole moment of an alcohol in a paraffinic solvent is shown in Figure 3; the important features are clearest at low temperatures.There is a small initial increase in apparent moment, which is ascribed to formation of small non-cyclic associates (dimer and trimer). Still at quite low concentrations, this increase is overwhelmed 3s W. L. Jorgensen, J. Am. Chem. SOC.,1981, 103, 341. Thermodynamics of Hydrogen-bonded L fquids 1.90 1.70 0 \ I 1.50 I130 0.0 1.o Figure 3 Apparent dipole moments of ethanol at 25 “C in 1,4-dimethylbenzene (upper curve) and n-hexane (lower curve). These curves are typical of ulcohols in aromatic and aliphatic solvents respectively by a rapid drop ascribed to the formation of cyclic species, which would be expected to have nearly zero net moment.As the concentration is further in- creased, the apparent moment rises again. This is probably because longer chains are forming, and there are entropy factors opposing the cyclization of long chains. There is a striking contrast with the behaviour in aromatic solvents, also shown in Figure 3. It will be argued later that the aromatic solvents reduce the proportion of rings present by ‘solvating’ the free hydroxy-groups at the ends of chains which would otherwise tend to close to form rings. The activity data cannot themselves provide any means of telling whether a cluster is open or cyclic; both forms occupy nearly the same volume, and their effect on the activity is the same. In principle, however, the difference should show up in the enthalpy of dilution, since the ring has a larger (more negative) enthalpy of formation.F. Equation involving the Entha1py.-The enthalpy of formation of the various H-bonded clusters is most readily considered as a contribution to the relative 270 Stokes apparent molar enthalpy of the alcohol, OH [equation (2)]. To a first approxi- mation, one may expect that the energy of formation of a hydrogen bond should be independent of the size of cluster in which it occurs, provided only strain-free configurations are accepted. There is, however, known to be a co-operative effect in multiply hydrogen-bonded systems, according to which adjacent bonds mutually enhance each other. To allow for this, we assign an enthalpy of forma-tion hz to the H-bond in a linear dimer, and a constant value h in all other cases.The strain-free cyclic groups are assumed to start with five or possibly four alcohol molecules in the ring. The total contribution of the H-bonds to OH in a solution containing CA moles of alcohol per unit volume is therefore as shown in equation (14): ca@H(H-bonds) = c2h2+hzi -l)ci(linear) + hXici(cyc1ic) (14) i= 3,. . . i =(4or5), . .. where ci denotes the equilibrium concentration of associated species i in the solution. Equation (14), through the respective factors (i -I) and i, allows for the fact that a linear i-chain contains (i -1) bonds, but a cyclic i-group contains i bonds. At concentration below about 0.1 molar (typically XA < O.Ol), the H-bonding terms are the only significant contribution to @H.The other contribution is what may be called the 'ordinary' or Van der Waals enthalpy of mixing. Though we have remarked that this is not accurately given by the Scatchard-Hildebrand expression [equation (6)],there is nothing better which can be used in a practical calculation without a lot of parameters. Equation (1) gives rise to a term in the relative apparent molar enthalpy @H(SH)= -DVA$,\ With typicalvalues of Dand VA, this term is unlikely to exceed -0.2 kJ mol-1 at XA = 0.1, in contrast to typical experimental values of approximately -20 kJ mol-1 for @H. Thus, the imperfection of equation (I) is not of great importance as long as we confine ourselves to alcohol solutions of less than say one molar.This does not mean that the H-bonding forces are hundreds of times stronger than the van der Waals forces, but that the latter change little with concentration in dilute solutions, whereas the amount of H-bonding changes rapidly. If, however, we wish to consider properties nearer the pure alcohol composition, and in particular the other limit lirn(H"/x~x~), XA-0 the non-H-bonding term is of major importance. G. Cyclization Equilibria.-The presence of cyclic H-bonded groups may be formally accounted for in many ways. For example, a five-molecule ring could be thought of as formed from five monomers, a linear dimer plus a linear trimer, or a linear tetramer plus a monomer, or a linear pentamer alone. All of these are Thermodynamics of Hydrogen-bonded Liquids thermodynamically equivalent, and the cyclization equilibrium constant could be expressed in terms of any of these processes.However, it is obviously simplest to consider only the last-named one, the cyclization of an open chain to a ring of the same length. This process must clearly have an equilibrium constant of the form [equation (1 5)]: k’i = ci(cyclic)/ci(linear) = yi exp( -h/RT) (15) where qi is an entropy factor and the exponential factor deals with the enthalpy loss in forming the new H-bond. Since h x -24 kJ moi-1 the exponential factor is of the order of lo4 at rbom temperatures, and even a strongly unfavourable entropy factor is likely to leave ki still large enough to cause a significant amount of the cyclic form to be present. To find an exact expression for qi would be a difficult exercise in statistical mechanics.Clearly it will depend for one thing, on the detailed geometry of the alcohol molecule in question; steric effects may restrict the number of con-figurations possible for cyclic groups, and the internal degrees of freedom may change in different ways on cyclization. There is, however, one effect that must always be relevant, and is fairly easily dealt with. This is the dependence of qi on the chain length i. Apart from the fact that we are excluding the possibility of cyclization for i < 4 on the grounds of excessive ring strain, it is clear that in unstrained rings qi should decrease as i increases.There are two ways of justifying this proposition: either a greater chain length reduces the probability of the ends coming close enough to form the new H-bond, or the larger ring has an increasing probability of breaking, because it has more H-bonds to break. Indeed, there seem no good reason why both these processes should not operate at once, but it is probably sufficient to consider only the second, which would lead us to expect a qi inversely proportional to i for a given alcohol. Thus the simplest form one could expect for the cyclization constant ki’ is given in equation (1 6) ki’ = Oji with alnt?/aT= h/RT2 and 8 independent of i. There seems to be no way of estimating 8 from any thermodynamic property except through optimizing the whole fit of the expressions for the activities and enthalpies to the observed data.Likewise, although the cyclizafion process plays a major role in determining the shape of the apparent dipole-moment curve, it requires several further assumptions about the moments of linear species to make possible any quantitative estimate of 8. Nor does there seem to .be any spectroscopic property which is directly related to the presence of cyclic species. There is, however, an infra-red absorption at -3620cni-1 which is certainly attributable to non-H-bonded hydroxy-groups. There is some disagreement as to whether it arises only from free monomeric alcohol molecules, or from all hydroxy-groups in which the proton is not participating in an H-bond, even though the oxygen is the acceptor of another H-bond.That is to say, does the Stokes band come only from the completely free -OH groups of the monomer, or from these as well as from the terminal -OH groups of H-bonded chains? This is the sort of discussion in which a non-specialist hesitates to take part, and I have adopted the monomer-and-terminal view on the basis that it seems to fit in better with the thermodynamic data. However, the spectroscopic data still do not provide on their own a value for the cyclization constant 8; they have to be combined with the thermodynamic data and the association model to yield a value of 8, and even then the best value of 8 is obtained only from a rather flat minimum in a standard deviation.For ethanol in cyclohexane at 25"C, the value of 8 = 20 was so obtained.19 This means that at 25 "C,the cyclic pentamer is four-times as abundant as the linear pentamer; for the icosamer the ratio has fallen to unity. H. Association Constants for the Linear Association Species.-I t is in general more helpful to work with the stepwise equilibria, i.e. those for the addition of a further monomer to an existing chain, Then for the formation of a chain of length i [equation (1 S)] : Ki = Ci/CI . ~i-1 (18) (since we have shown that activity coefficients cancel). Again, since one new H- bond is formed, it is useful to write [equation (19)]: Ki = ai exp( -AH/RT) (1 9) where AH is the enthalpy increase in forming the new bond, and ai is an entropy factor. Since we are proposing that the co-operative enhancement of the H-bond energy is noticeable only between the cases of one and two bonds, the dimer has one bond of strength h2, and all other linear i-mers have (i -1) bonds of the enhanced strength h.Hence the first few stepwise constants should be [equations (20)-(22)] K, = a2 exp(-h2/RT) (20) and K3 = a3 exp[(-2h + h,)/RT)] Ki = a6 exp(-h/RT) for all i > 3 If h is taken as the normal bond strength, then K2 will have a smaller tem- perature dependence than normal, and K3 a larger one; and the temperature coefficient of all the linear steps after the trimer formation should be the same as for the cyclization. (Strictly speaking a small enthalpy term arising from the thermal expansion has to be included in the temperature dependence of the linear species Ks's, but not in that of the cyclization constant 8).This leaves the question of how the entropy factors ai vary with i. In discussion of chainwise association equilibria it is commonly argued that the first step should have a different constant from the rest, because in the first step the two reactants are the same, whereas in all later steps they are different. In some of our applications of this model we have allowed both a2 and a3 to have different values from the Thermodynamics of Hydrogen-bonded Liquids subsequent ones. This, however, does increase the number of parameters to be fixed from the data, and it is not unreasonable to try the approximation of setting a2 = a3 = at (i > 3).This is more reasonable in the alcohol case than in cases of weaker association because it turns out that the proportion of dimer is small in most of the solutions of interest, the higher association products rapidly predominating as the concentration increases. Also a difference between a2 and ai can be at least partially compensated for by ‘sophisticating’ the value of hz (this delicious term for fudging is borrowed from the science of ballistics, with which I was peripherally concerned in my youth). 4 Expressions for the Thermodynamic Properties Using the model outlined above, expressions may be written for the thermo- dynamic properties in terms of the equilibrium constants. The concentration of free monomer, c1, is the natural independent variable to use; its values can be calculated from the activity data, independently of the details of the association, provided that one assumes a value for the non-H-bonding interaction parameter D of equation (6).For notational convenience we put Ki = k (i > 3) (23) and retain K2 and K3 as distinct quantities with temperature dependence given by equations (20) and (21).It is convenient to use the dimensionless argument y = kc, in the various series summations which arise from the theory; the cyclization constant 8 is already dimensionless. Two important quantities in the theory are the species sum S1 and the stoicheiometric sum SZ. These are respectively the total number of solute entities (monomer, dimer, etc.) and the total number of moles of solute (calculated as monomer), in unit volume of solution.Obviously SZis identical with CA, the stoicheiometric alcohol concentration. s, = zci = c1 + K2Ci2 + (K&3/k3) [Y3/(l -y) -d(fn(1 -v) -k y f C~~s, = ca = c1 + K~ + (&K3/k3)y3[(3-2y)/(i -y)2 + ey/(i Equations (24)and (25) are for the case where cyclization is taken to set in at i = 4. If it does not occur until i = 5, equation (24) has a further term +y3/4 fol-lowing y3/3,and in equation (25) the term Oy/(l -y) becomes 8y2/(1 -y). It is not possible to set up expressions in which all the equilibrium constants appear linearly so that linear least-squares methods can be applied directly. The procedure we have found most practical is to assign a trial value to the inter- action parameter D; then the monomer concentration c1 can be calculated from the experimental activities by equation (10).The equilibrium constants k, Kz, and 8 are then adjusted so that the concentrations CA given by equation (25) reproduce the experimental values as closely as possible. Alternatively, one may assume D and a set of equilibrium constants, and using c1 as a parameter cal- Stokes culate both the activity coefficients and the alcohol concentration CA to match the experimental curve. This is most conveniently done using an interactive computer program in conjunction with a plotter or a high-resolution graphic monitor. As is to be expected with any but an exact theory, the parameters obtained show some dependence on the range fitted.If interest is mainly in the hydrogen bonding, it is best to fit only the range up to about I molar. Figures 6-9, however, show that a fairly satisfactory description of the whole composition range up to the pure alcohol can be obtained; the enthalpy function is less well reproduced than the activities. In the case of ethanol in n-hexadecane, the constants which fit the activity data also predict phase separation, which does indeed occur, as shown in Figure 7. The conjugate phase compositions can be found by plotting lnaA against Ina-B, using c1 as a parameter to calculate both activities and the corresponding stoicheiometric composition XA. Where the curve crosses itself, the two XA 25.20 -15 di! -35 10 w s 5 0 3 0.5 0 cA/ mol I-' %' Figure 4 Showing inflexions characteristic of the apparent molar excess enthalpy of alcohols in dilute solutions. (a) Ethanol in cyclohexane and (6) ethanol in 1,4-dimethyl- benzene (25 "C) Thermodynamics of Hjpdrogen-boncled Liquid 4.5 4.0 3.5 3.0 -4 C-2.5 2.0 1.5 cA 1mol L-' Figure 5 Activity coeficients fA of ethanol at 25 "C in (a) Cyclohexane and (b) 1,4-dimethylbenzene values are the predicted compositions of the conjugate phases. These are close to the observed values at 25 "Cbut do not change rapidly enough as the tempera- ture rises towards the observed U.C.S.T.at 52.5 "C. The values of h2 and K2 are subject to large uncertainties, as the amount of dimer is always small. Indeed the only place where the dimer contribution is noticeable is for the apparent molar enthalpy in the region up to about XA = 0.005.Consequently it is better to take as the adjustable linear chain constants G,KKd, and k rather than K,,KB,and k The choice of (K&) is also advantageous in that the theory requires that the ratio (K2K3/k2)should be independent of temperature. Some typical fits to the theory are shown in Figures 6-9. Stokes 5.01 1 4.0 f 20.0 3.0 < iC d 2.0 10.0 1.c 0.c '0.0 0 Figure 6 Activity coeficients and the excess enthalpy function versus mole fraction for ethanol in hexadecane at 25 "C,showing the extreme steepness in the low alcohol region 5 Alcohols in Aromatic Solvents In aromatic solvents and to lesser extent in carbon tetrachloride, and carbon disulphide, the thermodynamic behaviour of alcohols is markedly different (Figures 4-5).The limiting partial excess enthalpy is much lower (in aromatics about 15 kJ mole-1) than in aliphatic solvents; the apparent molar enthalpy of dilution is much less steep in the low-alcohol region; and the limiting activity coefficient fA of the alcohol is substantially lower (Figure 5). The shift in the limiting partial excess enthalpy can also be looked upon as an enthalpy of transfer of a (monomeric) alcohol molecule from infinite dilution in the aliphatic solvent to infinite dilution in the aromatic solvent; the enthalpy change in this process is about -9 kJ mol-l.In an initial survey of ethanol in several solvents, 1 examined whether this could be explained by the move of the hydroxyl dipole Thermodynamics qf Hydrogen-bonded Liquids 5.0 4.c 3.c c, C d 2.o 1.0 0.0 0.0 0.5 1 Figure 7 The same data as in Figure 6, plotted against xi ; to expand the abscissa scale at low alcohol concentrations. The calculated curves are continued through the phase- separation region where there are no points plotted moment to a higher dielectric constant medium, but this attractive idea was invalidated by the observation that in carbon disulphide, with by far the highest dielectric constant of the non-polar solvents used (2.633), the enthalpy of transfer is only -3 kJ mol-l.It therefore appears that the effects are due to a specific interaction between the solvent and the alcohol hydroxy-group. In the case of carbon tetrachloride, the concept of a weak H-bond from the hydroxy-group to a chlorine atom is acceptable. With the aromatic solvents, however, one must think of the somewhat exposed hydroxyl proton as finding a lower-energy environment in the n-electron cloud of the solvent; in fact a solvation of the hydroxy-group occurs. Besides affecting the limiting partial excess enthalpy, this interaction sub- stantially reduces the amount of H-bonding between alcohol molecules as Stokes -LE 75 n %m 0.0 z 0.0 3.0 0.0 0.5 Figure 8 Observed and calculated properties of ethanol in cyclohexane at 25 "C compared with aliphatic solvents at similar concentrations.In effect, the solvent competes with the solute-solute hydrogen bonding interactions ; a hydroxy-group buried in n-electrons is not available for H-bonding with other alcohol molecules. Clearly, if an isolated alcohol molecule can be solvated by a benzene ring, so can the terminal hydroxy-group of an open H-bonded chain, and a chain thus solvated will no longer participate in the cyclization equilibria. This situation can be treated formally by introducing a solvation equilibrium for each of the open-chain clusters, including the monomer solvated open tmer/unsolvated open i-mer = Ks or %' Thermodynamics of Hydrogen-borided Liquids 5.0 4.0 20.0 c I 4t 7-7 0 -%m 2.0 d W z 10.0 1.o 0.0 I A xA Figure 9 Observed and calculated properties for butan-1-01 in cyclohexane at 25 "C unsolvated open i-merltotal open i-mer = 1/(1 + Ks) .A single Ksindependent of i should be a good first approximation. The equili- brium constants relating the unsolvated monomer, open chains, and cyclic forms should on this model be the same as for the aliphatic case, provided that all the other assumptions of the model are satisfied. The detailed working out of the thermodynamic properties of ethanol in p-xylene on this model has been presented. The calculations prove to be extremely recalcitrant when the volume-fraction statistics equations for the species activity coefficients are used; hence the mole-fraction statistics were used, and the calculation confined to alcohol mole fractions up to 0.2, where the effects of the statistics are quite minor.Using only two new parameters, the equilibrium constant and enthalpy change for the soivation process, it proved possible29 to %' Sokes obtain a reasonably good prediction of the properties of the xylene solution from those of the cyclohexane solution. 6 Dielectric Constant The prediction of the dielectric constant of pure polar liquids has been the subject of numerous theoretical papers using advanced mathematical tech-nique~.~~A special difficulty is associated with the long-range nature of dipole- dipole forces, which in some treatments makes the result highly dependent on the size of the body of liquid considered.Also, some simplified models such as point dipole in the centre of a dielectric sphere, though of value as a starting point, cannot be expected to reproduce with accuracy the complexity of behaviour of real hydrogen-bonded liquids. By taking more realistic potential functions for the water molecule, considerable progress has been made with the problem of water itself by computer techniques, but little such work has yet been done with alcohols or their solutions in non-polar solvents. One of the earliest treatments of the dielectric properties of alcohols was due to Lennard-Jones and Pople,4l who proposed an association into unbranched chains through hydrogen bonds.This idea has remained the basis of most later treatments of pure alcohols and their solutions in non-polar solvents. The typical behaviour of the apparent dipole moment of alcohols in paraffinic solvents is shown in Figure 3. There is initially a small rise in the apparent moment, followed by a marked fall to a rather flat minimum followed by a slow rise which continues to the highest concentrations studied. (One should note, however, that the concept of the apparent dipole moment is in any case of decreasing usefulness as the pure alcohol composition is approached). The association model which we have used for the thermodynamic properties would require, for its detailed application to the dielectric properties, the assignment of a net dipole moment to each of the associated species.While bolder spirits have not shrunk from the necessary guesswork, and have even calculated association constants from our dielectric data alone without cross-reference to the thermo- dynamic data, I do not feel confident enough in our knowledge of the charge distributions and the internal molecular motions of the associated species to attempt this assignment more than qualitatively. There is one important case, however, where I believe that no guesswork is needed. This is in the assumption that for any hydrogen-bonded cyclic group, the net dipole-moment is zero. This is automatically true if the dipole moments of the constituent molecules of the ring are vectorially additive, and there seems to be no reason for this not to hold, even though the individual moments may each be enhanced from the free- molecule value by co-operative effects.There is clearly no reason why any particular molecule in a ring should make a different contribution from the rest. The apparent dipole moment of an associated mixture, in which the i-fold associated species i has an effective dipole moment pi and concentration Ci may ‘O G. Still, G. N.Patey, and J. S. Hnye, A&. Chem. Phys., 1981, 48, 183. J. A. Pople, Proc. R. SOC.London, Ser. A, 1951, 205, 163. 28 1 Thermodynamics of Hydrogen-bonded Liquids be written : C ~~~~A = L'cipi2(linear) + L'ci'pi'2~ (cyclic) (I: = 1, 2 . . . .) where CA is the stoicheiometric concentration. If the cyclic species have zero moments, the summation need be taken only over the linear species.Although we still do not know how the pi's for these depend on i, we can estimate the effect of the omission of the cyclic species by calculating a differerit average moment of the non-cyclic species : pnc =zCipi2 (linear)/zicr(linear) = c~ p2app/~ici(linear) If we use the values of ziti calculable from the model which fits the thermo- dynamic data, pnc turns out to be the monotonically increasing function of c that one would expect.19 It is in fact much like the apparent moment found for solutions in aromatic solvents, where cyclization is to a considerable extent replaced by solvation of the hydroxy-groups, a process which is expected to have little effect on the dipole moments.The limiting moments extrapolated to in- finite dilution are not very different between aliphatic and aromatic solvents (or indeed between one alcohol and another). AcknowZedgements.Many of the results presented here are the work of Dr. Han Tin French; I also acknowledge with thanks the help of K. N. Marsh and his students, and the technical staff of the Chemistry Department of the University of New England in both the construction and the operation of the equipment. The Australian Research Grants Council has provided financial support over a long period. Tribute. R. A. Robinson, who gave me my first introduction to physical chemistry in Auckland in 1937, was a firm believer in the value of accurate thermodynamic measurements on solutions, and his own contributions to our knowledge of electrolytes have been of immense value not only to experimentalists, but also to those who have been developing theories of single and mixed electrolyte solutions during the past half-century.Though for the past fifteen years most of my own work has been in the field of non-electrolyte solutions, I can attribute such success as I have had in large measure to the grounding he gave me in thermo- dynamics. I must also acknowledge his personal generosity to me, which made possible the endowment of this memorial lecture. Those who knew him know also that his modesty would not have allowed him to endow it himself.

ISSN:0306-0012    

DOI:10.1039/CS9821100257

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

CENTENARY LECTURE* Hydrocarbon Reactions at Metal Centres By E. L. Muetterties DEPARTMENT OF CHEMISTRY, UNIVERSITY OF CALIFORNIA, BERKELEY, CALIFORNIA 94720, U.S.A. 1 Introduction Hydrocarbon reactions that are mediated by a metal atom or a collection of metal atoms represent an area of considerable scientific as well as technological importance. For simple hydrocarbons, there are only two fundamental classes of reactions namely carbon-carbon and carbon-hydrogen bond-breaking and -making reactions. The latter often have a lower energy of activation, thus facilitating comprehensive physicochemical studies. Carbon-hydrogen bond-breaking and -making reactions at metal centres is the better understood reaction class, at least for metal surfaces, and it is the subject of this analysis.The scope of the analysis is broad and includes the following general categories of metal species: (i) molecular, mononuclear metal complexes, ML, (ii) molecular, polynuclear metal complexes, M,L, (iii) metal surfaces (iv) metal atoms. A delineation of the thermodynamic, electronic, and stereochemical requirements for carbon-hydrogen bond reactions at metal centres will be attempted, although a definitive characterization of reaction-path pkofiles is not feasible yet. Since general reviews of this class of hydrocarbon reactions have been presented for the first,’-4 second,’ and third6 categories, this analysis will be selective, not comprehensive. * The present text is based upon the lecture delivered on 13 May 1982 at a RSC Dalton Division Meeting at the Scientific Societies’ Lecture Theatre, Savile Row, London W.l.G. W. Parshall, Acc. Chem.Res., 1970, 3, 139. G. W. Parshall, Acc. Chem.Res.: 1975, 8, 113. G. W. Parshall in ‘Catalysis’, ed. C. Kemball (Specialist Periodical Reports), The Chemical Society, London, 1977, Vol. 1, p. 335. D. E. Webster, Ado. Organomet. Chem., 1977, 15, 147.’E. L. Muetterties, T. N. Rhodin, E. Band, C. F. Brucker, and W. R. Pretzer, Chem. Reo., 1979, 79, 91. G. A. Somorjai, ‘Chemistry in Two Dimensions: Surfaces’, Cornell University Press, Ithaca, N.Y., 1981, Chapt. 7 and 8. B. C. Gates, J. R. Katzer, and G. C. A. Schuit ‘Chemistry of Catalytic Processes’, McGraw Hill Book Co., New York, 1979. Hydrocarbon Reactions at Metal Centres Schematically, the basic process of carbon-hydrogen bond scission and reformation at metal centres can be set forth as below where the large hatched sphere may represent a bare metal atom, a metal atom in a mononuclear metal complex, or one or more metal atoms in a metal cluster, or a metal surface (Scheme 1).Here, for initial simplicity, it is assumed that the hydrocarbon is Scheme 1 bound relatively firmly through some functionality, for example, an olefinic bond; later, the bonding and the subsequent reactions of saturated hydrocarbons will be explicitly considered. The scheme provides for the possibility of a reaction intermediate but there is no a priori basis for invoking an intermediate. A digression on intermediates uersus transition states is appropriate here to ensure that no misunderstanding is generated in later discussions because the possible significance or necessity of intermediates in C -H oxidative addition reactions T. S.Products Scheme 2 284 Muetterties has been mentioned in many articles3 For a specific system, the existence of a reaction intermediate confers no kinetic advantage; the free-energy difference between reactants and transition state is constant, Scheme 2. An alternative scenario comprises two closely related complexes: system 1 in which there is no possibility of a reaction intermediate and system 2, related to 1, in the context of the free-energy difference between reactants and products, but differentiated in that there is a ‘productive’ reaction intermediate in which the hydrocarbon forms complexes with the metal species (formation of a ML,-RH complex is not necessarily productive with respect to the C-H oxidative addition reaction). A kinetic advantage may accrue to system 2 if the ML,-RH complex more closely resembles products than initial reactants (free-energy diagram of Scheme 3).Specifically, ‘productive’ here implies that the reactant T.S? SYSTEM i T. S.‘ Products‘ Products’ Scheme 3 metal complex has undergone a reorganization in the intermediate whereby the ML, structure in the intermediate more closely resembles that in the product than that in the reactant, rather than an intermediate in which the C-H (or H-H) bond is significantly stretched because presently available data for oxidative addition reactions indicate that there is little stretching of the C-H bond, or H-H bond in H2 additions, in the transition state and hence in a preceding intermediate, if there is one.Also, a productive intermediate ideally should have a triangular, not linear, M-C-H stereochemistry. The overall forward reaction in Scheme 1 is denoted as an oxidative addition reaction foIlowing an established convention’ in organometallic chemistry :there should be some net electron-transfer, probably small in extent, from the metal centre or centres to the o-bound hydrocarbon radical and to the hydrogen atom in the product. Accordingly, there should be some net electron transfer from the metal centre or centres to the carbon-hydrogen o* orbital in the transition state or intermediate-the closer the transition state resembles product, the greater the extent of electron transfer will be in the transition state as will be the degree ’R.F. Heck, Organotransition Metal Chemistry, Academic Press, New York,1974. Hydrocarbon Reactions at Metal Centres of C-H bond stretching. Experimental data' for H2 or RH addition to a metal centre in molecular ML, Complexes show only a small kH:kD isotope ratio of -1.2 to 1.3 indicating a small degree of H-H or R-H stretching in the transition state. Consistently, a theoretical analysis of hydrogen addition to the hypothetical Pt(PH,), molecule has indicated that there is about a 4% stretching of the H-H bond in the transition state and another analysis has made similar conclusion^.^ Analogous data for the other metal regimes do exist.For low- or moderate-temperature oxidative addition reactions at metal surfaces, the kinetic isotope ratio appears to be normal but precise ratios have not been determined. Since the activation energy is very small for H2 dissociative chemisorption, the isotope effect for this surface mediated reaction is probably small. (Parenthetically, it is notable that carbon-hydrogen bond breaking on metal oxide surfaces, as measured for methanol oxidation to formaldehyde and propylene oxidation to acrolein on molybdenum oxide, have large kinetic isotope ratios of 3-4.'' These C-H bond breaking processes are presumed to proceed through C-H-0 rather than C-H-Mo states.It will be informative to obtain further kinetic isotope data for C-H bond breaking on metal oxide surfaces to see if relatively large kH:kD ratios are characteristic of processes that proceed through C-H-0 interactions). The microscopic reverse reaction of Scheme 1, reductive elimination, has many models in molecular R(H)ML, complexes. Measured' ' kinetic isotope ratios, kH:kD,are all normal and range in magnitude from -1.5 to 3.3. For these speciJic systems, their microscopic reverse reactions of oxidative addition should exhibit kinetic isotope ratios that are normal and larger than their measured reverse reactions. This then implies, in apparent contradiction to the previous statements above, that the range of kinetic isotope ratios for oxidative addition reactions could be quite large and that the character of the transition states might vary substantially.However, the one reductive elimination reaction with a truly large kinetic isotope ratio of 3.3, as established by Halpern et ul.," is alkane elimination from cis-R(H)Pt(PR3)2 complexes -a reaction that appears to be fully irreversible. Possibly, a large normal kinetic isotope ratio may be characteristic of such irreversible alkane reductive elimination reactions and may not apply correlatively to the basic reaction, of C-H oxidative addition to metal centres, under consideration in this analysis. In any case, it is essential that a larger data set of kinetic isotope ratios be generated for the forward and reverse reactions of Scheme 1 so that a more precise outline of transition states (or range of transition states) can be realized.Returning to Scheme 1, the overall equilibrium constant K = K,K2, will be a function of temperature, pressure (surfaces), and the electronic and P. B. Chock and J. Halpern, J. Am. Chem. SOC., 1966, 88, 3511; J. M. Brown and D. Parker, Organometallics, 1982,1, 950. K. Kitaura, S. Obara, and K. Morokuma, J. Am. Chem. SOC., 1981, 103, 2891. A. Sevin, Nouueau J. Chim., 1981, 5, 233. lo C. J. Machiels and A. W. Sleight, J. Catalysis, 1982, 76, 238; C. J. Machiels and A. W. Sleignt, ACS Svmp.Ser., 1982. 178. 239. L.Am, A.3cu, aua J. nalpern, J. Am. Chem.SOC.,1978, 100, 2915. Muetterties stereochemical features of the metal system. In all cases, temperature increase will favour the forward reaction of dehydrogenation. On metal surfaces, the reverse reaction of hydrogenation will be favoured by a higher thermodynamic activity of the chemisorbed hydrogen atoms, an activity that is a direct function of the pressure of H2 in the gas phase. To enhance the forward reaction, a reasonable strategy in the design of the metal system is to place a stereochemical constraint on the reactant state such that a C-H hydrogen atom is forced close to a metal centre or centres. Conceptually, this proximity effect must be a critical factor in carbon-hydrogen bond reactions. Electronic effects must also be important.If the earlier characterization of charge transfer for carbon-hydrogen bond breaking is correct, as it must certainly be at least for metal surfaces, then this forward reaction of Scheme 1 should be enhanced by selecting metal systems in which the negative charge density on the metal atom or atoms is relatively high. This would appear to be correct in a gross sense in that metal surfaces which easily effect dehydrogenation of hydrocarbons are good electron donors -the work functions, formally analogous to the ionization potentials of molecules, are relatively small. In contrast, most transition-metal complexes are not effective in hydrocarbon dehydrogenation reactions and also have high ionization potentials. An energy matching of the CT and Q* C-H orbitals, respectively, is therefore important with unoccupied and occupied molecular orbitals of the metal species (all other factors being equal--i.e., if the interaction parameters j3 and j3* are similar).For molecular complexes, the importance of appreciable electron density on the metal centre of the reactant complex has been presumed and has not been established by experiment. Theoretical analyses of alkane reductive elimination from R(H)ML2 complexes (the reverse reaction of Scheme 1) conclude, by corollary, that the microscopic reverse reaction of oxidative addition is more favoured if electron density on the metal centre is increased.I2 However, these analyses of reductive elimination appear to be incomplete: the reverse sequence of oxidative addition must involve population of the CT*C-H orbital, yet these theoretical analyses of reductive elimination do not explicitly correlate the Q* C-H orbital with an appropriate filled metal orbital.The geometric features of the oxidative addition reaction may comprise an initial linear M-H-C interaction but ultimate C-H bond cleavage requires a triangular M -H -C interaction. As documented in the following section, isolable molecular metal complexe8 with multicentre M -H -C interactions are more nearly triangular than linear in character. Sevin’ in a theoretical analysis of H2 oxidative addition concluded that H2 initially approaches /Hthe metal centre end-on and only later tips over to give a triangular M I ‘H configuration.l2 K. Tatsumi, R. Hoffmann, A. Yamamoto, and J. K. Stille, Bull. Chem. Soc. Jpn., 1981, 54, 1857; A. C. Balazs, K. H. Johnson, and G. M. Whitesides, lnorg. Chem., 1982, 21, 2162; J. 0.Noel1 and P. J. Hav. J. Am. Chem. Soc.. 1982, 104,4578. 287 Hydrocarbon Reactions at Metal Centres The importance of stereochemistry and electronic structure in metal-based hydrocarbon reactions will be explored. Each regime of metal chemistry will be separately analysed in these contexts. The issue of reaction intermediates, as depicted in Scheme 1, will be examined before the actual process of bond breaking or making is considered. 1 The Question of Reaction Intermediates in Carbon-Hydrogen Bond Breaking Reactions at Metal Sites A.Mononuclear Metal Complexes.-There is a plethora of molecular mono- nuclear metal complexes which have a multicentre C-H-metal bond as judged by a crystallographically-defined short H-metal separation ; a relatively comprehensive listing has been presented by Ittel et a1.” The range of H-metal separations is large and the composite set of complexes could serve to delineate the geometric pathways for equation (1). However, the accuracy of the hydrpgen atom placement in this extensive set of structurally defined complexes is too low to assess accurately reaction pathways. Ideally, a selected set of complexes that outfine the sequence in terms of H-metal separation should be crystallographically defined by neutron diffraction studies so that the key distance (x,y,z) and angular (a,P,y) parameters, (l), could be M evaluated.However, this set of parameters would ,e a function of stereochemical constraints associated with the ligand in which the C-H-metal bond is generated. The paradigm would be a set of crystallographic data for the ML, complexes with saturated hydrocarbons in which the only connexity between ML, and the hydrocarbon would be the multicentre C-H-metal bond. Unfortunately, no such complex has been detected, even transitionally, much less isolated and crystallographically defined with the exception of the labile Cr(CO)5 complexes with saturated hydrocarbons like cyclohexane.’ There is increasing evidence that benzene and its derivatives can form q2-arene-metal l3 J.A. Welch, K. S.Peters, and V. Valda, J. Phys. Chem., 1982, 86, 1941. M uetterties complexes as shown by the studies of Sweet and Graham.14 These arene complexes, unlike the labile alkane complexes, are sufficiently stable that a precise structural characterization by a crystallographic study is feasible. presumably, these q*-arene-metal complexes will have two relatively short H-M interactions. The only mononuclear metal complexes with C-H-metal bonding defined by neutron diffraction studies are {(q3-cyc10-octeny1)Fe[P(0CH3),], +}(BF4-)' and {[(CH3)3CCH]Ta[P(CH3)3]C12(p-Cl)}2.16 Co-ordination saturation is achieved in the iron complex by the multicentre Fe-H-C interaction via a saturated C-H bond of a carbon atom adjacent to the allylic carbon atoms in the q3-cyclo-octenyl ligand (Figure 1).The carbon-hydrogen distance of Figure 1 In the iron complex, {Fe[P(OCH3)3]3(q3-C8H,3)+}(BF,-),the iron centre is co-ordinated to three phosphite ligands and to the allylic carbon centres of the cyclo-octenyl ligand and to a hydrogen atom attached to a carbon atom adjacent to one of the terminal allylic carbon atom~.'~ The hydrogen-iron distfnce was estimated with precision by a neutron diflraction study. The distance is 1.874 A. This represents the first precisely determined three-centre C-H-Fe bond. The other four carbon atoms of the cyclo-octenyl ligands are not depicted in the drawing l4 J. R. Sweet and W. A. G. Graham, Organometallics, 1982, 1, in press; .J. Am. Chem. Soc., submitted for publication.l5 R. K. Brown, J. M. Williams, A. J. Schultz, G. D. Stucky, S. D. Ittel, and R. L. Harlow, J. Am. Chem. Soc., 1980, 102, 981; R. L. Harlow, R. J. McKinney, and S. D. Ittel, J. Am. Chem. SOC., 1979, 101, 7496. l6 A. J. Schultz, R. K. Brown, J. M. Williams, and R. R. Schrock,J. Am. Chem. SOC., 1981, 103, 169. Hydrocarbon Reactions at Metal Centres 1.16A is significantly stretched, -6 %, compared to that of normal C-H bonds. There must be some bonding interaction of the iron atom with the carbon atom, and the interaction should be characterized as a triangular three-centre bond. Closely related to the iron complex is (~3-6-endo-methylcyclohexenyl)Mn(C0)3. This complex, defined by X-ray crystallography, has a (CH)H-Mn distance of 1.86& a (C-H)C-Mn distance of 2.30 A and a slightly elongated C-H bond of 1.07A.I7 In the tantalum complex, ([(CH,),CCH]Ta[P(CH, )3]C12(,u-C1))2,the or-C-H bond of the neopentylidene ligand forms an angle of 84.8" with the Ta-C bond, and the C-H bond is significantly stretched, 1.13A.16 The Ta-H separation is 2.12A.Thus, this system can be described as a closed or triangular three-centre bond, (2), although the Ta-H'interaction is not especially strong To-C as judged simply by the separation. However, the three-membered ring system in (2) is highly strained and this strain may be the prime cause of the relatively long Ta-H bond. Recently, the X-ray diffraction study18 of C2H5TiC13[(CH3)2PCH2CH2-P(CH,),] established an interesting weak P-hydrogen C-H-Ti three-centre interaction with the following parameters: C-H, 1.02A;H-Ti, 2.29A; and C-Ti, 2.52(1).This structure rather nicely models an intermediate or transition state of a P-hydride elimination process for a transition-metal alkyl complex. For the complex in solution at -9O"C, there was no n.m.r. evidence of the multicentre C-H-Ti bond but this interaction should be in evidence in equilibrium isotope experiments (vide infra). Also, CH3TiC13 [(CH3)2 PCH2CH2-P(CH3),] has been shown by X-ray analysis to have a closed, three centre /H\Ti-C bond with a Ti-C-H angle of 70(2)" and a Ti-H distance of 2,03(4)A.'* Evidence of C-H-metal bonding for the solution state of a molecular metal complex can be obtained by 'H n.m.r.studies. A relatively tight C-H-metal bonding is reflected by a high field proton resonance, however, if the rate of a C-H-metal bond breaking is high the proton resonance is shifted to lower field as a function of the C-H complex to C-H-M complex concentrations. In addition, the l3C-H coupling constant is substantially reduced for the C-H-..M bond and raised (relative to a normal C-H bond) for the remaining, if any, C-H bonds that are not interacting with the metal atorn.I5- l8 Alternatively, the multicentre interaction can be detected by partial deuterium M.Brookhart, W. Lamanna, and M. B. Humphrey, J. Am. Chem. SOC., 1982,104,2117.'* z. Dawoodi, M. L. H. Green, V. S. B. Mtetwa, and K. Prout, J. Chem. SOC., Chem. Commun., 1982, 802; ibid.submitted for publication. Muetterties substitution into the CH, site. This is a much more sensitive probe of C-H-M interactions in d.n.m.r. studies, through an equilibrium isotope effect.lg Such evidence for C-H-M interaction in the solution state has been incisively established for the cyclo-octenyl-iron complex.’ 5920 B. Polynuclear Metal Complexes.-Four polynuclear metal complexes with multicentre C-H-M interactions have been defined by either crystallographic or n.m.r. studies or both. The first example was CH3(H)Os3(CO)lo.21,22 Although this cluster has never been isolated in single crystal form and there are no precise structural parameters available to date, the ‘H d.n.m.r. studies of this complex unequivocally established the presence of 0s-H-0s and 0s-C(H,)-H-0s bonding in the solution state, (3).21*22In the dinuclear iron complex, {Fe2 (p-CH3)(p-CO)(r15-C5 H5)2 (p-CH2 [P(C6 H5)2]2)) + (PFJ, as defined through X-ray diffraction data,23 there is an unsymmetrically bridged CH3 group, analogous to the osmium cluster as shown in Figure 2.Key structural parameters are respectively C-H, H-Fe, and Fe-C distances of 1.06(0.83), 1.64(1.78), and 2.108(2.118)A with a C-H-Fe angle of -102’ (the distances given in parentheses are for the second independent complex in the unit cell). IH\These data incisively characterise a closed three-centre C-Fe interaction in this complex. Conventional low temperature n.m.r. experiments did not show evidence of the C-H-Fe interaction but the presence of this interaction in the solution state structure was established for this complex and a related one, {Fez(p-CH3)(p-CO)(tf-C5H5)2(CO),),by ‘H d.n.m.r.studies using the p-CHzD derivative whereby an equilibrium isotope effectI9 favouring the C-H-Fe interaction was dete~ted.’~.~~ A rigorously defined closed three-centre interaction for both the solid and the solution state of a cluster is found in HFe4(q2-CH)(C0)1z, the Tachikawa l9 M. Saunders, Stereodyn. Mol. Syst. Proc. Symp., 1979, 171; F. A. L. Anet, V. J. Basus, A. P. W. Hewett, and M. Saunders, J. Am. Chem. SOC., 1980, 102, 3945. 2o S. D. Ittel, F. A. Van-Catledge, C. A. Tolman, and J. P. Jesson, J. Am. Chem. SOC., 1978, 100, 1317; S. D. Ittel, F. A. Van-Catledge, and J.P. Jesson, J. Am. Chem. SOC., 1979, 101, 6905. R. B. Calvert and J. R. Shapley, J. Am. Chem. SOC., 1978, 100, 7726. 22 R. B. Calvert, J. R. Shapley, A. J. Schultz, J. M. Williams, S. L. Suib, and G. D. Stucky, J. Am. Chem. SOC., 1978,100, 6240. 23 G. M. Dawkins, M. Green, A. G. Orpen, and F. G. A. Stone, J. Chem. SOC., Chem. Commun., 1982, 41. 24 C. P. Casey, P. J. Fagan, and W. H. Miles, J. Am. Chem. SOC.,1982, 104, 1134. 29 1 Hydrocarbon Reactions at Metal Centres r-+ Figure 2 In the solid state structure of {Fe2(p-CH3)(p-CO)(qs-C5Hs)2-(p-CH,[P(C,H,),],))'(PF,-),the methyl group is in a bridging position between two iron atoms. In addition, one of the methyl hydrogen atoms closely approaches one of the iron atoms so as to describe a closed three-centre iron-cfrbon-hydrogen bond.23 The C-H, H-Fe, and Fe-C distances are 1.06, 1.64, and 2.108 A, respectiuely butterfly.25 Composite X-ray and neutron diffraction data26,27 precisely define such a closed three-centre C-H-Fe interaction for the solid state (Figure 3).Here the essential distance parameters, shown in (4), are illustrated for one of I .75 A/ H\I. 19 Fe-C 1.93 A (41 the two independent molecules.26-28 The very long C-H bond distance of 1.19A is remarkable and is the longest, accurately assessed C-H bond distance of all molecular species. Consistently for the solution state, the nominal C-H hydrogen atom in the Fe-H-C bond has an n.m.r. resonance shifted substantially to high field (--1 ~.p.m.).~~ These complexes constitute the set of clusters in which three-centre C-H-M interactions are unambiguously defined.This number is small but should grow 22 M. Tachikawa and E. L. Muetterties, J. Am. Chem. SOC.,1980, 102, 4541. 26 M. A. Beno, J. M. Williams, M. Tachikawa, and E. L. Muetterties, J. Am. Chem. SOC., 1980, 102, 4542. 27 M. A. Beno, J. M. Williams, M. Tachikawa, and E. L. Muetterties, J. Am. Chem. SOC., 1981, 103, 1485. *' R. M. Gavin, J. Reutt, and E. L. Muetterties, Proc. Natl. Acud. Sci., USA, 1981, 78, 3981; E. L. Muetterties in 'Reactivity of Metal-Metal Bonds' ed. M. H. Chisholm, A. C. S. Symposium No. 155, American Chemical Society, Washington, 1981, p. 273. Muetterties fY,/B.O.(6.0.= 0.25,0.21) [ = 0.84,0.81) 1 9- (8.0.= 0.73.0.71) z,Figure 3 The Tachikawa cluster, (p-H)Fe4(qZ-CH)(CO), has a closed three-centre C-H-Fe bond as established by X-ray and neutron difraction data.25*27 The core structure for the cluster is shown above.The twelve carbonyl ligands are not shown; each iron has three terminal CO ligands in this cluster. All of the core bonding distances are shown in the above drawing. Also shown are the estimated Mulliken bond orders (Hiickel calculations) for some of the core utom interactions” substantially as neutron diffraction studies of clusters increase and as very low temperature n.m.r. studies of clusters and the n.m.r. technique” of searching for equilibrium isotope effects’ for CH,D-M species become more general experimental procedures (specific studies illustrating this application to C-H-M interactions are described in references 20,21,23-25). C.Metal Surfaces.-Chemisorption states of saturated hydrocarbons on metal should be based largely on the interaction of CT C-H and CT*C-H orbitals with appropriate surface metal atom orbital^.^*-^^ En ergetic considerations suggest that electron transfer from the metal surface to c* C-H antibonding orbitals may dominate in such bonding states.29 Filling of the antibonding 29 E. Shustorovich, R. Baetzold, and E. L. Muetterties, J. Phys. Chem., submitted for publication. 30 T. E. Madey and J. T Yates, Jr., Surf. Sci., 1978, 76, 397. 31 J. E. Demuth, H. Ibach, and S. Lehwald, Phys. Rec. Lett., 1978, 40, 1044. 293 Hydrocarbon Reactions at Metal Centres C-H orbitals should reduce substantially the C-H stretching frequency and should enhance the potential for C-H bond cleavage.Hydrocarbons typically chemisorb reversibly on flat metal surfaces under ultra high vacuum conditions. Depending upon hydrocarbon conformation in the chemisorption state and upon the surface topography, the number of effective C-H-M multicentre interactions per carbon atom may vary frorn one to some fraction of one. Hence the hydrocarbon chemisorption heats should be some function of the number of carbon atoms in the hydrocarbon molecule. In fact, chemisorption bond energy appears to be a nearly hear function of the number of carbon atoms in the hydrocarbon as shown in Figure 4 from the classic study by Madey and Yates3' for hydrocarbons chemisorbed on the SATURATED HYDROCARBON DESORPTION FROM Ru (000I1 I I 1 1 1 1 '/ 1 1 /' i /$ C-C8H16 /'C -C6H I2 / -f' /' C2H6 //I I I 1 I I I 1 1 0 2 4 6 8 I0 Number of Carbon Atoms Figure 4 In a study of hydrocarbon chemisorption on Ru(0001), Madey and Yates3' established the desorption energies for a series of hydrocarbons on this basal plane.The magnitude of the desorption energies is a function of the number of carbon atoms in the molecule. To a first approximation, the relationship is linear, however, the value for the cyclo-octane does not follow this relationship; this apparent deviation might be due to conformational eflects that limit the number of C -H -metal surface interacrions per molecule.The hydrocarbons employed in this study were ethane, cyclopropane, cyclohexane, and cyclo-octane M uetterties basal, close-packed plane of ruthenium, (OOO1). Spectroscopic data for cyclohexane chemisorbed on Ru (OOO1) have been interpreted3' in terms of a model, illustrated in Figure 5, consistent with the foregoing postulate. The proposed electronic representation for hydrocarbon chemisorption states, as described above, requires some reduction in the C-H force constant which should be evident in the vibrational spectrum. In fact, vibrational data derived I I I I I I I I I I I I Figure 5 A model proposed by Madey and Yates3' for the cyclohexane chemisorption on ruthenium(0001 )based on a spectroscopic study has a primary interaction of three axial C-H bonds in the chair form of cyclohexane with three-jbld sites in this closed packed plane.At the top is a side view and at the bottom, a ciew looking down on the surface plane (a view that includes only the three axial C-H bonds that stronglj interact with the surface plane) Hydrocarbon Reactions at Metal Centres from high resoluton electron energy loss spectroscopic studies of the cyclohexane chemisorption states on the close-packed planes of Ni(l1 1),31 Pt (1 1 1),31and Ru(0001)32,33 establish that there is a lower energy C-H stretching frequency. This is consistent with the postulate of C-H-M bonding but, of course, the genesis of the lower energy stretching frequency cannot be uniquely established from these data.In a collaborative programme with E. Shustorovich and R. Baetzold, we have attempted to define on theoretical grounds the stereochemistry for methane chemisorbed on a close-packed metal surface. The preliminary results2 suggest that (5) is favoured over (6)at high d level occupancy; (6)tends to be favoured at low d level occupancy. However, the range from high to low level d occupancy may be, in the theoretical calculations, comparable, larger, or smaller than the HHH Iy/ H I I 15’ actual range for transition metals. Only the trend is significant. Energy differences are small so that rotation of methane chemisorbed on a metal surface will probably encounter a small barrier, In any case, it is interesting that configuration (5) is not a directly productive way point for dehydrogenation of methane, a process that has a large activation energy barrier on all metal surfaces.34 The reaction of methane with a clean tungsten surface has an activation energy of 9.3 k1 kcal mole- at 1350-2359 K.There is a large kinetic isotopic effect, kCH4:kCD4= 4.5, that is ascribed to a tunnelling process.34 D. Metal Atoms.-Condensation of gaseous transition-metal atoms into low temperature methane matrices is suspected to yield M(CH4), complexes for some transition metals but there are no definitive supporting data.35-38 Metal atoms like copper or nickel do not appear to interact strongly with methane whereas the electronic spectrum for copper atoms in an ethane matrix at 30 K shows a much larger splitting of the ’P -+ ’S band than for copper atoms in a methane or argon matrix.It has been suggested that this is an indication of a rather strong interaction between excited copper atoms and the surrounding ethane cage. A metal atom-methane complex should encounter a very small barrier to inter- 32 F M.Hoffman, T. E. Fetter, P. A. Thiel, and W. H. Weinberg, J. Vac. Sci. Technol., 1981, 18, 651. ”F. M. Hoffman, T. E. Felter, P. A. Thiel, and W. H. Weinberg, J. Chem. Phys., submitted for publication. 34 H. F. Winters, J. Chem. Phys., 1976, 64, 3495. 35 W. E. Billups, M. M. Konarski, R. H. Hauge, and J. L. Margrave, J. Am. Chem. SOC.,1980, 102, 7393. 36 G. A.Ozin, D. F. McIntosh, and S. A. Mitchell, J. Am. Chem. SOC., 1981, 103, 1574.’’G. A. Ozin, S. A. Mitchell, and J. Garcia-Prieto, Angew. Chem. Suppl., 1982, 369. P. H. Barrett, M. Pasternak, and R. G. Pearson, J. Am. Chem. SOC.,1979, 101. 222. Muetterties conversion of M-H-CH3, M-H2 -CH2, and M-H3 -CH configurations. Indirect evidence for such interconversions for Cu-CH4 matrices is suggested by the selective excited state reactivities. 2 Carbon-Hydrogen Bond Breaking Process A. Mononuclear Metal Complexes.-The first explicit description of C-H oxidative addition to a metal centre in a molecular transition-metal complex was by Chatt and Davidson in 1965.39 Reduction of the octahedral bis(l,2-dimethylphosphino-ethane)ruthenium dichloride complex in the presence of naphthalene or benzene gave the cis-b-naphthyl or phenyl ruthenium hydride complex, equations (2) and (3).The significance of this finding was ignored to [(CH3)2PCH2CH2P(CH3)2]2RuC12-!% [(CH,),PCH2CH2P(CH3)2]2RU (2) some degree. The reaction was remarkable because the strongest type of carbon-hydrogen bond had been cleaved and because it comprised the reaction of an initially free hydrocarbon with a molecular metal complex. Reaction may have been facilitated by an intermediate in which the arene molecule was [(CH3I2 PCH2CH2 P (CH3Ip] Ru + C6H6 '6' co-ordinated, in some (productive) fashion, to the presumed zero-valent sixteen-electron ruthenium intermediate complex. Following the original Chatt-Davidson report,39 an enormous number of C-H addition reactions were discovered for mononuclear transition-metal complexes (an extensive set of reviews' -4 provide a relatively comprehensive analysis of this chemistry up to 1977).The majority of reported oxidative additions were internal additions: co-ordinately unsaturated molecules, in which a C-H bond associated with one of the ligands could approach closely or was forced to approach closely the metal atom, often exhibited this facile internal oxidative addition reaction. Examples of this important proximal effect were ortho-palladation4* [equations (4)and (5)] and ortho-metallation4' 39 J. Chatt and J. M. Davidson, J. Chem. SOC., 1965, 843. 40 A. C. Cope and R. W. Siekman, J. Am. Chem. SOC., 1965, 87, 3272.41 G. W. Parshall, W. H. Knoth, and R. A. Shunn, J. Am. Chem. SOC.,1969, 91,4990. Hydrocarbon Reactions ut Metal Centres PdClt-+ C6H5CH2NR2 4CI-+ -2{PdCI3N(R)&H;} XI’ + 2HCI + in aryl phosphite complexes [equation (6)l.I-In these cases, the stereochemically favoured five-membered ring is formed. Nevertheless, four- and three-membered ring systems’ -can be and are generated from co-ordinately unsaturated metal arylph~sphine~~ complexes, respectively, (7) and (8).and trimethylph~sphine~~ There is no question but that the proximal effect in these internal oxidative addition reactions greatly enhances the facility of C-H bond breaking. Until recently, there was no example of C-H oxidative addition reactions between a molecular metal complex and a hydrocarbon molecule that possessed no functional group or unsaturation (examples of C-H addition with aromatic m01ecules,~~-were established as in the original 49 acetone,44 and a~etonitrile~~ 42 M.A. Bennett and D. L. Milner, J. Am. Chem. SOC., 1969, 91, 6983. 43 J. W. Rathke and E. L. Muetterties, J. Am. Chem. SOC., 1975, 97, 3272. 44 S. D. Ittel, C. A. Tolman, A. D. English, and J. P. Jesson, J. Am. Chem. SOC., 1976, 98, 6073. 45 C. Giannotti and M. L. H. Green, J. Chem. SOC.,Chem. Commun., 1972, 1114. 46 K. L. Tang Wong, J. L. Thomas, and H. H. Brintzinger, J. Am. Chem. SOC., 1974, 96, 3694. 47 E. K. Barefield, G. W. Parshall, and F. N. Tebbe, J. Am. Chem. SOC.,1970,92, 5234. 48 U. Klabunde and G. W. Parshall, J.Am. Chem. SOC., 1972,94, 9081. 49 L. P. Seiwell, J. Am. Chem. SOC., 1974, %, 7134. 298 Muetterties Chatt-Davidson" observations for benzene and naphthalene). A view developed (including the author's5') that there were two factors that lessened the probability of C-H oxidative addition of saturated hydrocarbon and molecular metal complex. One was the proximal effect which would be so predominant that internal oxidative additions would tend to prevail over oxidative addition of a free saturated hydrocarbon molecule (which is an extremely weak ligand) unless the co-ordinately unsaturated molecular complex were very carefully designed so as to constrain all ligand C-H hydrocarbon atoms distal to the metal centre. The other was the thermodynamic factor that stems from the relatively weak alkyl-metal bond strength whereby the equilibrium constant for addition would be rendered very small.The ubiquitous facile reductive elimination of alkane from cis-ML,( H)(R) complexes was a supportive observation. Competition between internal C-H oxidative addition involving the metal centre and a ligand and external C-H oxidative addition involving the metal centre and a hydrocarbon molecule must be evaluated on kinetic and thermo- dynamic grounds. Entropic considerations generally favour the internal oxidative addition reaction; however, enthalpic factors may favour the external addition. Assuming a reversible process for internal C-H addition it follows that, for some molecular species, molecular design to inhibit internal C-H addition may not be a critical factor vis-a-vis external C-H addition from a hydrocarbon molecule if the enthalpy contribution from the carbon-metal bond for the 'external' adduct is substantially greater than that for the 'internal' adduct.Listed in Table 1 are values for C-H bond strengths for a variety of saturated and unsaturated hydrocarbons. The decreasing order of bond strength probably also mirrors the probable trend in bond strengths for carbon-metal bonds, e.g. C6H5-M > CH3-M > a-allyl-M > a-benzyl-M. Since a reasonable estimate Table 1 Curbon-hydrogen bond dissociation energiesa Bond Energylkcal mol -Ph-H 110 vinyl-H 108 Me-H 104 Et-H 98 Pr'-H 95 Bu'-H 91 allyl-H 87 PhCH2-H 87 a Data taken from A.Streitweiser, Jr. and C. H. Heathcock, 'Introduction to Organic Chemistry', MacMillan Publ. Co., Inc. New York,1981 2nd. Ed., p. 1194 50 E. L. Muetterties in 'Fundamental Research in Homogeneous Catalysis', Vol. 2, ed. Y. Ishii and M. Tsutsui, Plenum Publishing Corp., 1978, p. 1. Hydrocarbon Reactions at Metal Centres for the range in M--H bond enthalpic contributions is 50-65 kcal mole-', an observable addition of a saturated hydrocarbon like pentane or cyclohexane to a metal complex would require a C-M bond energy of 40-55 kcal mole- One factor that would favour such an addition would be a 5d metal centre. Stability for an alkyl-metal bond should increase, through a larger ionic contribution, in going from 3d to 4d to 5d metal [the high stability of compounds like cis-CH3 (H)OS(CO)~ probably resides partly in the relatively large CH3-0s (and H-0s) bond energy; for the 3d-metal cobalt, it has been estimated" that metal-alkyl bond energies do not exceed the range of 20-30 kcal mole- at least alkyl-cobalamin complexes.A more extensive discussion of the metal-atom carbon bond enthalpies is presented by ConnorS2]. The first observation of C-H oxidative addition for a saturated hydrocarbon to a metal centre in a molecular complex was by Crabtree and co-workers in 1979.53,54Cyclopentane reacted with the cationic (IrH2[P(C6H5)3]2[solvent]2+) complex in the presence of 3,3-dimethylbut-l-ene to form ((q5-C5HS)IrH[P(C6H5)3]2'>;the solvate molecules were either water or acetone.Also, cyclo-octane reacted under these conditions with the iridium complex to form {Ir(cyclo-octa-1,5-diene)[P(C,H,),l, '}. However, cyclohexane did not yield an (Ir(C6H6)LX+) complex ostensibly for stereochemical reasons. The function of the olefin was as a hydrogen acceptor for the hydrogen atoms excised from the cycloalkane; a less sterically demanding olefin like ethylene was not effective, presumably because it formed too stable a complex with the iridium centre. The metal centre in this complex was described as an electrophile -not a nucleophile. Unfortunately, it is difficult to characterise the reactant complex as either electrophilic or nucleophilic; the reactant complex is cationic and cannot be electronically defined by a gas-phase photoelectron-spectroscopic study.An extension of the iridium reaction to rhenium [equation (7)] has been rep~rted.~ In 1982, three groups56- independently discovered reversible saturated- hydrocarbon addition to molecular iridium and rhodium complexes. Bergman and JanowiczS6 found that irradiation of q5-C5(CH3)51rH2P(CH3)3 in hydro- carbon media, e.g., cyclohexane and neopentane, led to hydrogen elimination ahd formation of the corresponding alkyliridium hydride complex, q5-C5(CH3)51rH(R)P(CH3)3. Elimination of alkane from these complexes occurs thermally at + 1I0"C allowing conversions such as that shown in equation (8). 51 J. Halpern, F. T. T. Ing, and G. L. Rempel, J. Am. Chem. SOC., 1979, 101, 7124.52 J. A. Connor, Topics Curr. Chem., 1977, 71, 71. 53 R. H. Crabtree, J. M. Mihelcic, and J. M. Quirk, J. Am. Chem. SOC.,1979, 101, 7738. 54 R. H. Crabtree, M. F. Mellea, J. M. Mihelcic, and J. M. Quirk, J. Am. Chem. SOC., 1982, 104, 107. 55 D. Baudry, M. Ephiritikhine, and H. Felkin, J. Chem. SOC., Chem. Commun., 1980, 1243; 1982, 606. 56 A. H. Janowicz and R. G. Bergman, J. Am. Chem. SOC., 1982, 104, 352; ibid., submitted for publication.''W. D. Jones and F. J. Feher, J. Am. Chem. SOC.,1982, 104,4240; personal communication. J. K. Hoyano and W. A. G. Graham, J. Am. Chem. SOC., 1982, 104, 3722. 300 Muetterties Thus, the addition reaction does not simply result from some unique photoexcited state. q5-C5(CH3)51rH(cyclohexyl)P(CH3)3+ neopentane---* cyclohexane + ~5-C5(CH3)51rH(ne~pentyl)P(CH3)3(8) Interestingly, the presumed intermediate q5-C5(CH3)sIrP(C6H5)3 , generated also by photolysis of the hydride, reacts with benzene but there is a competition with internal oxidative addition of an orthq C-;H in the phenyl substituent on the phosphorus atom to give q5-C5(CH3)51rP(C,H4)(C6H5)2(H).This competitive internal oxidative addition is not seen for the trimethyl-phosphine analogue. The Bergman-Janowicz reaction system apparently has many desirable features. The attendant ligands in the presumed intermediate C5(CH3)51rP(CH3)3 are good donors and the ionization potential of this intermediate should be relatively low. In addition, the metal is a 5d metal and the Ir-alkyl bond should be and obviously is relatively robust.Nevertheless, the same chemistry can be effected with the rhodium analogue of the iridium complex as shown by Jones and Feher,’ although the resultant alkylrhodium hydride complexes, explicably, are less thermally stable than the iridium analogues. Also, the electron density at the metal atom in such intermediates can be substantially reduced without losing the reactivity toward saturated hydrocarbons: Hoyano and Graham have shown5* that the photofysis of q5-C5(CH3)51r(C0)2 in cyclohexane yields q5-CS(CH,),IrH(cyclohexyl)(CO),ostensibly through the analogous CO intermediate, q5-C5(CH3),IrC0. Interestingly, Jones and Feher” report that q5-C5H5RhP(CH3)3(H)(C6H4R)complexes are converted, on warming, into q5-’C5H5RhP(CH3)3(qz-C6H5R),which is another indication of an intermediate complex in the oxidative addition of arenes to ML, species.To summarise, factors that apparently enhance the probability of oxidative addition of C-H bonds in saturated or unsaturated hydrocarbons to molecular ML, complexes are as follows: (i) The ML, complex should be a high energy species; this is probably the most important single factor. All other factors being equal, the addition reaction should be more favourable with a 5d than a 4d metal centre and least with a 3d metal because in going down a transition-metal group, the M-H and M-C (alkyl) bond strengths increase. The other thermodynamic factor of importance is the proximal or entropic effect.(ii) The capability of a reactant metal species to complex the hydrocarbon prior to the oxidative addition step may convey an entropic advantage to the metal species in oxidative addition reactions. (iii) If the metal complex has ligands with C-H bonds that can, or are forced to, approach closely the metal centre, then internal C-H addition may be favoured for entropic reasons. Thus, a metal complex designed so as to keep all ligand C-H bonds distal to the centre is sterically ideal for the external addition of a saturated hydrocarbon molecule. But this ideal is not necessarily required for external addition 301 Hydrocarbon Reactions ut Metal Centres of a saturated hydrocarbon. An interesting example of competitive internal and external C-H addition was discussed above for the Bergman-Janowicz system.(iv) Co-ordination unsaturation in the reactant ML, complex is essential, of course. The complex should have a low lying unfilled orbital for interaction with the Q C-H bonding orbital. Important also is an appropriate high lying filled ML, orbital for interaction with the o* C-H antibonding orbital. In the initial phase of the oxidative reaction, either of these two interactions could be dominant. Hence, either an electron rich or electron deficient ML, species could be effective in the oxidative addition of a saturated hydrocarbon. Accordingly, there could be a range of transition states in such oxidative addition reactions. As noted earlier, a sensitive probe of the transition state is the measurement of the kinetic isotope ratio, kH :kD.Clearly, many more experimental data are required to assess objectively the key features of saturated hydrocarbon addition reactions to ML, species. The importance of the Q* C-H interaction with a metal orbital should increase in progressing from metal atoms and mononuclear ML, complexes to clusters and finally to metal surfaces because this is the increasing energy level trend for the metal d-band of these species. However, ‘energy matching’ is not necessarily critical here because the interaction parameters B and /?* may in fact be quite disparate in magnitude. (v) Because of the aforementioned energy considerations in the matching of Q C-H bonding and Q* C-H antibonding levels with appropriate unfilled and filled metal orbitals, it may well be that photochemically or thermally excited states of ML, species are better suited to the oxidative addition of saturated hydrocarbons (provided that AG for the reaction is not adversely affected by temperature increase in the case of the thermally excited ML, species).The ongoing studies by the Bergman, Crabtree, Graham, and Jones research groups may soon provide the necessary background to a more incisive and quantitative characterization of this type of hydrocarbon reaction for molecular, mononuclear metal complexes. B. Molecular Metal Clusters.-External oxidative addition of a C-H bond in a saturated-hydrocarbon molecule to metal centres in molecular clusters has not been demonstrated to date but internal oxidative addition reactions have.Electronically, stereochemically, and thermodynamically, the issues are funda- mentally no different than for mononuclear metal complexes except that the d ‘band’ is higher in energy for clusters than for mononuclear metal complexes. The proximal effect also operates in clusters.’ One example is the thermal conversion of a simple Os-NC,H, bond into a bridged a-pyridyl system, (9), in the triosmium carbonyl cluster, [equation (9)],” Many such examples Muetterties (9) of internal oxidative additions of C-H bonds in molecular clusters have been reported. The development of metal cluster hydrocarbon chemistry thus appears to be following that of mononuclear metal complexes.One may anticipate that photolysis of a second- or third-row metal cluster of the form H2MxLyor MxLy+ will generate a M,Ly intermediate which will react directly with a hydrocarbon. One very interesting cluster is (,u-CH,)(~-H)O~~(CO)~~in that it provides information about the transition state (or reaction intermediate) for C-H bond breaking in polynuclear metal arrays and shows reversible C-H bond breaking in the solution As noted earlier, 'H d.n.m.r. studies have shown that this cluster has the methyl group bridging a cluster edge as 0s-CH2-H-0s and the hydrogen atom also bridging the same edge, (3). The bridging C-H bond is reversibly cleaved on a time scale such that in solution, over a wide temperature range, there is an equilibrium of the methyl complex with (~-CH,)(H)(~-H)OS,(CO)~~, Figure 6 (the rate of C-H bond breaking is, however, not fast with respect to the n.m.r. timescale).Notably, the methyl complex does not reductively eliminate methane, rather dehydrogenation is the favoured process [suggesting that OS~(CO)~~,the fragment that would be generated in methane elimination is a relatively high energy species]. In fact, at elevated temperatures, further dehydrogenation occurs converting the methylene complex into a very stable, triply bridging methylidyne cluster, (p-H)30s3(p3-CH)(C0)9.The reversibility of this last step has been demonstrated: either CO or a phosphine converts the methylidyne complex, apparently through a series of methylene and methyl complexes, ultimately into methane and, respectively, Os3(CO)l or phosphine derivatives thereof.60 The beauty of this osmium cluster chemistry is that it is an excellent formal analogy to metal surface chemistry whereby dehydrogenation processes are facile.The dehydrogenation process can be followed stepwise by n.m.r. spectroscopy, and the bridging Os-CH2 -H-0s interaction of the methyl complex is an explicit model of a transition state or intermediate in a C-H bond breaking process. Of substantial merit would be a crystallographic study, ideally "C.C. Yin and A. Deeming, J. Chem. SOC.,Dalton Trans., 1975, 2091. 6o R. B. Calvert, Ph.D. Thesis, University of Illinois, 1978. Hydrocarbon Reactions at Metal Centres CH3(H)Os3(CO),o (CH, H),Os,(CO),, Figure 6 The two triosmium carbonyl clusters shown above, one with an unsymmetrically bridging methyl ligand and the other with a bridging methylene ligand, are in equilibrium with each other in the solution state.21*22 This is a classic example of facile C-H bond breaking centred on metal atoms in a cluster molecule and represents a simple, formal model of C-H bond breaking for metal surfaces based on neutron diffraction, of (p-CH3)(H)Os3(CO),, or a derivative thereof; attempts to obtain single crystals of this complex have been unsuccessful to date.A related and informative system is based on methylrhodium and methyliridium complexes. Reaction of methyl-lithium with the corresponding {(p-Cl)M(cyclo-octa-1,5-diene)}, complexes yields methyl bridged dimers, ((p-CH,)M(cyclo- 0cta-1,Sdiene))~.61 The rhodium methyl dimer displayed no n.m.r.evidence at low temperatures of an unsymmetric methyl bridging interaction, M-CH2 -H-M; however, the more sensitive n.m.r. experiments with CH2D methyl groups has not been completed. An X-ray crystallographic study of the more stable rhodium dimer established a near symmetric methyl bridging unit but hydrogen atom positions were not accurately established, Figure 7.61 Nevertheless, both the rhodium and the iridium complexes undergo facile C-H bond breaking reactions. For the rhodium complex, this reaction is moderately fast at 0-+20 "C. The major overall process is reductive elimination of methane and the formation of rhodium metal.This process probably proceeds first through an a-hydrogen abstraction from a methyl group to yield some ((CH3)(H)(p-CH2)Rh2(cyclo-octa-1,5-diene))2species (monitoring of the overall process by 'H n.m.r. did give evidence for an intermediate complex with a bridging methylene group, Rh-CH2-Rh). In contrast, the major process with the Sd iridium complex was a-hydrogen abstraction, hydrogen elimination, and formation of the stable bridging methylene complex, {(p-CH2)Ir(cyclo- octa-1,5-diene)},. Thus, in the polynuclear metal complexes as in the mono-nuclear complexes, the tendency to eliminate methane when there are cis or vicinal H and CH3 groups appears to decrease sharply in going from 3d to 4d to 5d metal centres.G. F. Schmidt, E. L. Muetterties, M. A. Beno, J. M. Williams, Proc. Natl. Acad. Sci., USA, 1981, 78. 1318. M uettert ies Figure 7 The structure of ((p-CH,)Rh(cyclo-octa-1,5-diene)f2,as established by an X-ray crystallographic study,61 has bridging methyl groups. The core structure is shown above with the exclusion of all hydrogen atoms and also the saturated carbon centres of the cyclo-octadiene ligand. Ignoring the rhodium-rhodium interaction, each rhodium centre has efectively four-coordinate planar geometry involving the two carbon atoms of the bridging methyl groups and the midpoints of the two olefinic bonds of the bound cyclo-octadiene ligand In the Tachikawa butterfly, Figure 3, which has the closed three centre Fe-H-C bond, (4), there are a number of dynamic processes that for the solution state are fast on the n.m.r.time scale as shown in Figure 8. One of these is hydrogen atom site exchange between the Fe-H-C and the Fe-H-Fe multicentre sites.25 Such an exchange is an expected process in that the methylidyne C-H bond is significantly stretched in the ground-state form. The barrier to H atom site exchange in {(p-H)Fe4(p-q2-CH)(CO)12)is-21 kcal mole- '. Substitution of carbonyl groups by more electron donating ligands such as phosphines or phosphites lowers the exchange barrier to values of 12-16 kcal mole- ' Thus, like the methyl triosmium cluster, this methylidyne iron cluster formally models the facile process of C-H bond breaking that is a general property of metal surfaces.Also, the closed three-centre Fe-H-C interaction of the methylidyne group is an energy minimum for this iron cluster system. C. Metal Atoms.-Transition-metal atoms isolated in methane matrices show no tendency to react with the methane up to temperatures of -80K although there is some spectroscopic evidence of a complex between the methane molecule and the metal atoms.35* 36 However, photolysis that leads to excitation of the metal atoms elicits metal-atom insertion into a methane C-H bond to give initially CH3MH35,36 which in a secondary photolysis gives CH3M + H. and HM + CH3.. Curiously, the first row transition metals of titanium through vanadium failed to show this C-H oxidative addition reaction, however, the manganese through to zinc series of metal atoms, with the notable exception Hydrocarbon Reactions at Metal Centres Figure 8 A number of dynamic processes involving substituent ligands occur in the Tachikawa cluster, (p-H)Fe,(q’-CH)(CO), (see earlier description of the cluster structure, Figure 3).Two of the processes involve the hydrogen atom in the multicentre carbon- hydrogen-iron bond. One process is a pip of this hydrogen atom from one side of the cluster to the other, ie., to the other apical iron atom in the cluster. The other process comprises a site exchange of the C-H hydrogen atom with the hydrogen atom that bridges the two basal iron atoms. The activation energy for this site exchange is about 21 kcal mole-’ of nickel, underwent the C-H addition rea~tion.~’ Ethane also adds oxidatively to photo-excited copper atoms to form, apparently, C2H5CuH intermediate species; there was no spectroscopic evidence of CH3, CH,Cu, or (CH3)&u species formed in this photolytic reaction.37 It has been suggested that the failure of a metal like nickel to react with methane may be due to the non-formation of a metal-atom-methane c~mplex.~’ However, it is not at all evident why cobalt and copper atoms would complex methane and a nickel atom would not.Perhaps, the lack of CH4 addition for some of the metal atoms has more to do with the nature of the electronic state of the photoexcited metal atoms than any other factor or set of factors (photoexcited non-metal atoms such as sulphur also insert in the C-H bonds of hydrocarbons.The preferred reaction to explore is the thermal reaction of a metal atom with a saturated hydrocarbon. For this type of study, the matrix isolation experiment has serious temperature constraints because the metal atoms condense to form dimers, trimers . . . and larger aggregates or clusters at relatively low temperatures. Actually, the iron dimer is reported to react with methane at -196 “C presumably to form CH3FeFeH.38 Also, Klabunde6’ has found that nickel aggregates formed from metal atoms react with saturated hydrocarbons 62 S. C. Davis, S. J. Severson, and K. J. Klabunde, J. Am. Chem SOC.,1981, 103, 3024. Muetterties -at the very low temperature of -140 "C. Alternatively, the hydrocarbon reactions of metal atoms, metal dimers, etc., of established electronic states could be studied in the gas phase.Gas-phase studies of this character have been made but not based on hydrocarbon reactants. D. Metal Surfaces.-(i) Introduction. A major objective in our study of metal surfaces has been the identification of the key factors that facilitate carbon- hydrogen bond breaking. To control separately the variables of surface crystallography (topography), surface composition (carbon, oxygen or sulphur impurity), and surface coverage by the adsorbate, ultra high vacuum conditions and techniques were employed. Some of the crystal planes examined are illustrated in Figures 9-1 1. To anticipate the results from all studies to date, the proximal effect appeared to be the most important one in eliciting facile carbon-hydrogen bond breaking reactions.(ii) Saturated Hydrocarbons. Atomically flat metal planes such as the (111) and (100) planes of face-centred cubic metals like and platinum63 and the close-packed (O001) plane of hexagonal close-packed ruthenium metal3' are remarkably non-reactive towards cyclohexane under ultra high vacuum conditions (cyclohexane desorption from the metal surfaces at these low pressures is fast at temperatures in the range of -60 to -0 "C for these flat metal planes). Stepped nickel surfaces such as Ni( 110)and Ni9( 111) x (111) were similarly nonreactive. fcc (111) fcc (100) fcc (110) Figure 9 Shown above are the three low Miller index planes for a face-centred cubic lattice.For the (1 11) close-packed plane, the co-ordination number of surface metal atoms is 9, for the flat (100) plane, the co-ordination number is 8 and for the super-stepped (1 10) plane, the co-ordination number of metal atoms in step 'sites' is 7 (those just below in the next plane have a co-ordination number of 11) M.-C. Tsai, C. M. Friend, and E. L. Muetterties, J. Am. Chem. SOC., 1982, 104, 2539. Hydrocarbon Reactions at Metal Centres Figure 10 This is a depiction of a stepped surface in which terrace atoms (clear circles) are close packed and have a co-ordination number of 9. The steps are 1 atom high and the step atoms (hatched circles) have a co-ordination number of7. In this particular surface, the width of the steps is nine atoms More reactive was the stepped platinum surface, 6(111)x (111) which effected modest conversions of cyclohexane into chemisorbed benzene at -20 0C.63 A general discussion of stepped surface reactivity has been presented by S~morjai.~~ In sharp contrast, the reconstructed Ir(ll0) surface, Ir(ll0)-(1 x 2), is very reactive toward saturated hydrocarbons as shown by Weinberg and co-worker~.~~~~~Saturated hydrocarbons like ethane and n-pentane irreversibly chemisorb -only hydrogen, H2, desorbs as the temperature 1s raised. Spectroscopic studies indicated that dehydrogenation of the adsorbed saturated hydrocarbons begins at temperatures of --140 "C.This extraordinary reactivity cannot be traced solely to unique electronic properties of this surface: the reconstructed surface is of lower surface free energy than the (110) plane.The 64 G. A. Somorjai, 'Chemistry in Two Dimensions: Surfaces', Cornell University Press, Ithaca, N.Y 1981, Chapt. 8. 65 T. S. Wittrig, P. D. Szuromi, and W. H. Weinberg, J. Chem. Phys., 1982, 76, 716. 66 T. S. Wittrig, P. D. Szuromi, and W. H. Weinberg, J. Chem. Phys., 1982, 76, 3305. Muetterties fee (l0,8,7) Figure 11 In this surface there are terrace atoms (clear circles), step atoms (hatched circles) and kink atoms (dark circles). The co-ordination number of the atoms at the kink sires is 6 reconstructed surface is believed to have a saw-toothed form; one possible representation is depicted in Figure 12.Coincidentally, or possibly not, hydro- carbon dehydrogenation on this iridium surface begins at the same temperature observed for saturated hydrocarbon reactions with small nickel particles.62 The low reactivity of atomically flat metal planes (under ultra-high vacuum conditions) may have a largely stereochemical genesis. The primary chemisorption bond@) for the molecular chemisorption state of saturated hydrocarbons on metal surfaces probably is the multicentre C-H-M interaction. On flat metal surfaces, most of these multicentre interactions may be of the open and nearly colinear three-centre form and would not be directly productive with respect to C-H bond breaking (see earlier discussion of alkane chemisorption states). On the more deeply grooved or grossly irregular surfaces, some C-H-M /c\interactions may be forced to take closed form, M,-H, which would be directly productive with respect to C-H bond breaking reactions.Stereochemistry is emphasized here but the actual differences among these classes of surfaces derive from interrelated geometric and electronic factors. A reasonable projection based on the limited data for saturated hydrocarbon reactions on clean metal surfaces is that stereochemistry is important and the Hydrocarbon Reactions at Metal Centres Y Z t t Top View Side View Figure 12 Shown above is a top and side view of a possible for reconstructed Ir(ll0) surface, Ir(ll0)-(1 x 2). For an alternatioe, see the model proposed by H. P. Bonze1 and S.Ferrer (Surf. Sci., 1982 118, L263) key variable is surface topography whereby the extent of closed multicentre /c\M,-H interactions is determined. The proximal effect so evident in C-H oxidative addition reactions in molecular metal complexes may be important also in metal surface reactions of saturated hydrocarbons. (iii) Benzene. Benzene chemisorbs on atomically flat, clean metal surfaces like Ni( 11 l),Ni( 100), Pt( 11l),and Pt( 100) by interaction of the ring z and z* orbitals with appropriate metal surface orbitals; the c6 ring is largely parallel to the surface plane.67- Whether the benzene C-H hydrogen atoms are nearly in the c6 plane or displaced either toward or away from the surface cannot be established from available spectroscopic data (nor is this point likely to be established in the near future). In any case, all the chemical studies suggest that the benzene C-H hydrogen atoms are relatively distal to the surface lane.^'-^^ There is no detectable evidence of C-H bond breaking, reversible or irreversible, for benzene chemisorbed on Ni(111) or Ni(100) up to temperatures of -115 "C and -220 "C respecti~ely.'~These are the respective temperatures at which benzene thermally desorbs from these surfaces; i.e., there is no benzene C-H bond breaking on these flat planes until a temperature is attained where benzene desorption from the surface begins.Similar chemistry was observed for benzene chemisorbed on Pt( 11 1). At temperatures close to molecular desorption, there 67 S.Lehwald, H. Ibach, and J. E. Demuth, SurJ Sci., 1978, 78, 577.'* J. C. Bertolini and J. Rousseau, Surf. Sci., 1979, 89, 467. 69 J. E. Demuth and D. E. Eastman, Phys. Reti. B., 1976, 13, 1523. 70 F. P. Netzer and J. A. Matthew, Solid State Commun., 1979, 29, 209. "E. L. Muetterties and M.-C. Tsai, Bull. SOC.Chim. Belg., 1980, 89, 813. 72 C. M. Friend and E. L. Muetterties, J. Am. Chem. Soc., 1981, 103, 773. 73 M.-C. Tsai and E. L. Muetterties, J. Am. Chem. SOC.,1982, 104, 2534. 74 M.-C. Tsai and E. L. Muetterties, J. Phys. Chem., 1982,86, in press. 3 10 Muetterties Figure 13 On stepped surfaces, the benzene molecule is n-bonded much as it is on atomically flat surfaces. However, because of the steps, some of the benzene molecules have substituent atoms that will closely approach step metal atoms.Because of this proximal effect, some of the benzene molecules should undergo C-H bond breaking at relatively low temperatures. On a stepped nickel surface, these bond breaking reactions are irreversible under ultra high vacuum conditions and do occur at temperatures below 100 "C should be some thermally excited states in which benzene C-H hydrogen atoms approach close to surface metal atoms. This stereochemical feature probably accounts for C-H bond breaking processes competitive with molecular desorption on these flat nickel surfaces. Some irreversible C-H bond breaking for chemisorbed benzene on the stepped nickel surface, 9(111) x (1 1 l), occurs at low temperature^.^, The extent is about 10%.This is attributed to the features of a stepped surface whereby benzene a bound on the flat terraces necessarily has some benzene C-H hydrogen atoms close to stepped metal sites (Figure 13).The proximal effect appears to be a dominant factor for benzene C-H bond breaking processes on metal surfaces. (iv) Toluene. Toluene initially should chemisorb on flat metal surfaces in an electronic and stereochemical fashion analogous to benzene,', but low temperature angle-resolved photoemission studies will be required to support this presumption. At moderate temperatures, toluene appears to be converted into benzyl and H surface-bound species. Labelling studies with C6H5CD3and C6D5CH3 species support this proposal of benzyl formation for Ni( 11 l), Ni(100), and Pt(100) s~rfaces.~~.~~ Conversion of a n-toluene molecule into a n-benzyl species is fully anticipated in terms of the proximal effect: if toluene initially a-bonds to a flat surface with the c6 aromatic ring nearly parallel with the surface plane, then an aliphatic CH3 hydrogen atom will be forced into close proximity to surface metal atoms and the associated C-H bond should cleave to give M(xxx)-CH2C6HS and M(xxx)-H.All seven of the pz orbitals associated with the benzyl radical could then interact with appropriate, syrnmetry-adapted surface-metal orbitals. Spectroscopic (vibrational and photo- emission) tests of the putative n-benzyl surface state are in progress. Toluene chemisorption (submonolayer coverage) is fully irreversible on the low Miller index planes of nickel.Rapid heating of such crystal planes with chemisorbed toluene under ultra-high vacuum conditions yields only H,(g) and a surface nickel plane with chemisorbed carbon atoms. Rate of H, desorption 31 1 Hydrocarbon Reactions at Metal Centres Toluene Decomposition -Ni(ll1) D, Formation Temperature ( "C1 Figure 14 Toluene irreversibly chemisorbs on the Ni( 111) surface. As the 'temperature is raised, C-H bonds are broken and ultimately hydrogen is desorbed from the surface. In the thermal decomposition of perdeuterotoluene, there are two Dz thermal desorption maxi& of relative intensities 3 and 5 at -130 and 185 "C.Separate experiments with the specifically labelled toluene molecules, CD3C6H5and CH3C6Ds, established a regioselective bond breaking sequence.The thermal desorption spectra of deuterium, Dz,from the thermal decomposition of the two labelled toluene molecules on the Ni(ll1) surface are shown here. Deuterium from the CD3C6H5molecule appears only in the low temperature region whereas, as shown in the second experiment with CH3C6D5,there is no deuterium desorption in the low temperature range; deuterium appeared only in the high temperature region characteristic of aromatic C -D bond breaking H Figure 15 Toluene irreversibly chemisorbs on jlat nickel surfaces. The initial chemisorption state probably comprises x-bonding analogous to that proposed for benzene on such surfaces but this necessarily places methyl C-H atoms close to the surface.This should lead to irreversible C-H bond breaking at relatively low temperatures. It has been proposed that one of the first states generated in the dehydrogenation of toluene on jlat surfaces involves the benzyl species which could be very strongly bonded to the metal atoms in such jlat surface planes as shown in thefigure above 312 Muetterties /H I\ Figure 16 On the two Jlat Ni(ll1) and Ni(100) planes, toluene undergoes regioselective C-H bond breaking whereby all aliphatic bonds are cleaved before any aromatic C-H bond are broken (Figure 14). The composite sequences shown in Figures 15 and 16 depict possible models for intermediate states up to the state where all aliphatic C-H bonds have been broken.Spectroscopic studies are required to resolve precisely stereochemical issues under these conditions has two maxima, with a lower temperature maximum of intensity three and a higher one of intensity five for the two flat planes, (111) and (loo), but only a single maximum for the ‘super stepped’ nickel (110) surface. A study of the thermal decomposition of C6H5CD3 and C6D5CH3, monitoring the desorption profiles of H2, D2, and HD, rigorously established a fully regioselective C-H bond-breaking sequence in which all aliphatic C-H bonds were broken before any aromatic C-H bonds were broken for the two flat Ni( 11 1) and Ni(100) planes (Figure 14).72 One possible stereochemical pathway is shown in Figures 15 and 16 for the overall dehydrogenation process.(v) Pyridine.Pyridine chemisorbed on Ni( 100) undergoes two thermal processes: desorption and decomposition of the pyridine molecule to give H2(g), N2(g), and Ni(100)-C.75 In the decomposition of chemisorbed pyridine, there are three H2 desorption maxima of relative intensities, one, two, and two. A study of the thermal decomposition of [2,6-2H2]pyridine, [3,5-’H2]pyridine, and [4-2H,]pyridine established that the low temperature hydrogen maximum of intensity one was basically only composed of D2, H2, and H2, respectively, for the three labelled pyridines. These results incisively indicated facile formation of a 2-pyridyl surface species from an initially molecularly bound pyridine. If some or all of the initially bound pyridine molecules are bonded to the surface through the nitrogen atom, with the ring plane normal to or at least nominally tipped away from the surface plane, then the proximal effect, if operative, should lead to formation of a 2-pyridyl surface species (Figure 17), a feature fully 75 R.M. Wexler, M.-C. Tsai, C. M. Friend, and E. L. Muetterties, J. Am. Chem. SOC., 1982, 104, 2034. Hydrocarbon Reactions at Metal Centres I II I I I I I \\\\\\\\\\\\\\\\\\\\\\\\\\\\\' Figure 17 On the Ni(100) surface, pyridine chemisorbs initially as a molecule and is then readily converted into a 2-pyridyl species as shown in the drawing. This process is reversible and is relatively fast at temperatures at least as low as 70 "C consistent with the experimental results.75 Note that some pyridine molecules, especially at low surface coverages, could be largely n bound in a plane parallel to the surface plane, provided that such a state were in rapid equilibrium with molecular states in which the pyridine bonding were largely, or solely, through the nitrogen atom.Again, careful spectroscopic studies are required to establish the partitioning among q6-NC5HS, q'-NC5H5, and q2-2-pyridyl chemisorption states. The reversible (thermal) desorption of pyridine is preceded by the pre-equilibrium of molecularly bound pyridine with a 2-pyridyl ~tate.~' If pyridine is co-chemisorbed with D2 on Ni(100), the fraction of pyridine molecules that desorb as pyridine in the thermal desorption experiment consist of C5H5N, CSH4DN,and C5H3D2N(but no CSH2D3N)molecules.The stereochemistry in this H-D exchange was incisively established in the H2 and D2 co- chemisorption experiments with the three labelled pyridines as shown in Table 2. Only the a-C-H hydrogen atoms are exchanged in the pre-equilibrium step. All the observable chemistry7' for pyridine on Ni(100) can be summarized as shown in Figure 18; chemistry that is fully explicable in terms of the proximal effect. Table 2 Reaction of Ni(IO0)-pyridine with D2 and H2 Pyridines forme& in the thermal Starting pyridine Hydrogen added desorption experiment I H2 [2H,], [2H,], and ['H2]pyridine [2,6-'H2]pyridine \ D2 ['H2]pyridine only [2H2]pyridine only [2H2], [2H3], and [2H,]pyridine [' H ,]pyridine only I HZ ['HI], [2H2], and ['H3]pyridine I D2 3 14 Muetterties D4$H11 ? IA D2 80 -110 OC [2,6-2H,] PYRlDlNE [ZH,-Z JPYRIDINE IA D* ,HQ H, Figure 18 The dynamic surface behaviour of pyridine on Ni(100) is summarized in the above drawing with the speciJcally labelled molecule, [2,6-’H2]pyridine (vi) Cyclic OIeJins and Cyclic Polyenes.Initial chemisorption of a cyclic olefin such as cyclohexene should be based on a n bound state as shown in (10). Irrespective of the C6 ring conformation in the n bound state, there should be close approach of methylene C-H hydrogen atoms to the surface metal atoms. Ultimately, the sequential dehydrogenation process should generate chemisorbed benzene, a process well established in heterogeneous catalysis.In fact, the chemisorption of cyclohexene on a variety of nickel and platinum surface planes resulted in benzene formation at temperatures ranging from 20 to 100°C, as established by benzene displacement from the crystal into the gas phase by the strong field ligand, trimethylpho~phine.~~ Conceptually, a cycloalkene or cyclic polyene should chemisorb on a flat or nearly flat metal surface and then undergo a series of dehydrogenations to generate a delocalized C,H, chemisorbed species [equation (lo)] M-C,H,--+ H,-,-M-C,H, (10) In support of this generalization is the observation that cyclo-octene and cyclo-octa-1,5-diene are converted into cyclo-octatetraene as demonstrated by trimethylphosphine displacement of cyclo-octatetraene from the metal surface states from cyclo-octene and from cyclo-octa- 1,5-diene.76 Cyclobutene chemisorp- tion did not lead to a cyclobutadiene species that could be either thermally ”M.-C.Tsai, J. Stein, C. M. Friend, and E. L. Muetterties, J. Am. Chem. SOC., 1982, 104, 3533. Hydrocarbon Reactions at Metal Centres (10) desorbed or displaced by strong field ligands. These last experiments were indeterminate; spectroscopic studies are required (angle resolved vibrational and photoemission experiments). Cycloheptatriene appears to chemisorb onto metal surfaces with facile conversion into H-M-C7H7 .76 Cyclopentane probably chemisorbs and reacts to form H-M-(C,H,) but again spectroscopic studies (in progress) are required for verification of this hyp~thesis.~~ (vii) Acetylene.Acetylene chemisorbs initially onto metal surfaces in a 7~ bonded fashion. Subsequent to chemisorption a series of processes may occur. These processes may include dehydrogenation, [1,2] hydrogen atom shifts and hydrogenation. The rates of such processes are a sensitive function of tempera-ture, metal d level occupancy and metal surface t~pography.~~ -93 Chemisorbed 77 J. E. Demuth, SurJ Sci., 1979, 84, 315. J. E. Demuth, Surf. Sci., 1979, 80, 367. 79 J. E. Demuth, Chem. Phys. Lett., 1977, 45, 12. T. E. Fischer and S. R. Kelemen, Surt Sci., 1978, 74, 47. W. J. Lo, Y.W. Chung, L. L. Kesrnodel, P. C. Stair, and G. A. Somorjai, Solid State Commun., 1977, 22, 335.A. E. Morgan and G. A. Somorjai, J. Chem. Phys., 1969,51, 3309. 83 L. L. Kesmodel, P. C. Stair, R.C. Baetzold, and G. A. Sornorjai, Phys. Rev. Lett., 1976, 36, 1316. 84 P. C. Stair and G. A. Somorjai, J. Chem. Phys., 1977, 66, 2036. L. L. Kesmodel, R. C. Baetzold, and G. A. Somorjai, Surf. Sci., 1977, 66, 299. L. L. Kesmodel, L. H. Dubois, and G. A. Somorjai, Chem. Phys. Lett., 1978, 56, 267. L. L. Kesmodel, L. H. Dubois, and G. A. Somorjai, J. Chem. Phys., 1979, 70, 2180. H. Ibach, H. Hopster, and B. Sexton, Appl. Phys., 1977, 14, 21. 89 H. Ibach, H. Hopster, and B. Sexton, Appl. Surf. Sci., 1977, 1, 1. 90 H. Ibach and S. Lehwald, J. Vac. Sci. Technol., 1978, 15,407. 91 A. M.Baro and H. Ibach, J. Chem. Phys., 1981, 74,4194. 92 T.E. Felter and W. H. Weinberg, Surf. Sci., 1981, 103, 265. 93 M. H. Howard, S. F. A. Kettle, I. A. Oxton, D. B. Powell, N. Sheppard, and P. Skinner, J. Chem. Soc., Faraday Trans. 2, 1981, 77, 397. 316 Muetterties species like C=CH2, CH=CH2, and C-CH3 have been proposed for specific stages of the Pt(ll1) reaction with acetylene (and eth~lene).~~-~~ The 'final' state, stable at moderate temperatures, appears to be best characterized as p,-ethylidyne with the C-C vector normal to the surface plane and directed towards t hree-fold sites.' 86*93 Reaction of Pt(100)-C2H2 with D2 at 20-50°C resulted in rapid H-D exchange at the C-H sites with no net reaction of either hydrogenation or dehydr~genation.~~In contrast, Pt(11 1)-C2H2 reacted with D2, but with no H-D exchange between C-H and Pt-H sites, to give a new state that compositionally appeared to be Pt( 11 1)-C2H2D,94 presumably the afore- rnenti~ned~~*~~~~~ state.These results of the isotopic labelling Pt(l1 1)-CCH3 studies can be interpreted in terms of a proximal effect. Apparently on Pt( 11 l), the C-H hydrogen atoms, for all states beginning with n-bound acetylene and ending with the ethylidyne complex, are relatively far removed from the surface metal atoms and undergo no H-D exchange reactions. However, on Pt(100), at least one of the C-H hydrogen atoms in the n-bound acetylene state must closely approach the surface plane. One possible geometric rationalization is as follows. The initial n-bound acetylene can tip with respect to the surface plane if one carbon atom drops into a three-fold site on Pt(ll1) or a four-fold site on Pt(l00).Because of the size difference of these three sites, the C-C tipping with respect to the surface plane can be -45" for Pt(100) but only -13" for Pt( 11 1) (Figure 19).94 Thus, in such a tipped configuration, one C-H hydrogen atom will be quite close to the surface metal atoms on Pt(100)-C2H2 but both C-H hydrogen atoms will be relatively far removed for Pt(ll1)-C2H2 . In fact, a low-energy electron diffraction study of Pt( 100)-C2H2 was interpreted in terms of such a tipped configuration with an angle of -50" between the C-C vector and the surface plane.95 For species like C=CH2 and C-CH3 on the Pt(ll1) surface, none of the C-H hydrogen atoms will be close to the surface if the C-C bond vector is normal or nearly normal to the surface plane, a plausible geometric feature for such chemisorbed species.(viii) Acetonitrile. On the atomically flat Ni( 11 l),*acetonitrile chemisorbs in a nearly fully reversible fashion; under ultra high vacuum conditions acetonitrile, initially chemisorbed at 20 "C, desorbs at -90 C.96 Spectros~opic~~ and diffra~tion~~data for Ni( 11 1)-NCCH3 and Ni(100)-NCCH3 indicate that the nitrile is bridge bonded through the nitrogen atom to two, three, or four surface atoms. In this configuration, the acetonitrile has the C-N bond more or less normal to the surface plane. Accordingly, the methyl hydrogen atoms are relatively far removed from the metal surface atoms.In contrast to Ni( 11 1)-NCCH3, Ni9( 11 1) x (1 1 1)-NCCH3 displays only about 80-90 % reversible thermal desorption of a~etonitrile.~~ The competing 94 E. L. Muetterties, M.-C. Tsai, and S. R. Kelemen, Proc. Natl. Acad. Sci., USA, 1981,78,6571. 95 G.Casalone, M. G. Cattania, and M. Simonetta, Sur$ Sci., 1981,103, L121. 96 C.M.Friend, J. Stein, and E. L. Muetterties, J. Am. Chem. SOC.,1981,103,767. 97 C.M.Friend, E. L. Muetterties, and J. Gland, J. Phys. Chem., 1981,85, 3256. 317 Hydrocarbon Reactions at Metal Centres 318 Muetterties process is acetonitrile thermal decomposition uia C-H bond scission. This difference between these two surfaces is attributed to close approach of methyl hydrogen atoms, of chemisorbed acetonitrile, to the surface metal atoms specifically at step sites on the stepped surface (Figure 20).On the super stepped Ni( 110)surface, acetonitrile chemisorption is nearly irre~ersible.~~ (ix) Diatomic Molecules. Diatomic molecules like HZ,N2, NO, and CO may initially chemisorb on a flat metal surface with the bond vector normal or parallel. Data and theory29 suggest that the dominant state for these diatomic molecules on flat surfaces is with the bond vector normal to the surface. Hence to proceed from a molecular chemisorption state to a dissociated chemisorption state [equation (1l)] requires some activation energy -small for H2and larger for a molecule like CO. Interestingly, there are data that indicate that dissociative chemisorption of diatomic molecules proceeds more readily on stepped metal surfaces than on atomically flat metal ~urfaces.~.~~- This again may reflect the operation of the proximal effect [equation (12)].(x) Conclusions. Essentially all our observations on C-H bond breaking processes on nickel and platinum surfaces can be explained simply by a proximal effect: if hydrogen atoms of a chemisorbed hydrocarbon or hydrocarbon derivative are forced close to the surface metal atoms or readily approach the surface in thermally excited states, then C-H bond breaking will tend to occur. Thermodynamically, these are favourable processes for these transition metals under high vacuum conditions (the reverse reaction requires a relatively high 98 S.L. Bernasek and G. A. Somorjai, J. Chem. Phys., 1975, 62, 3149. 99 M. Salmeron, R. J. Gale, and G. A. Somorjai, J. Chem. Phys., 1977,67, 5324; 1979, 70, 2807. loo M. Balooch, M. J. Cardillo, D. R. Miller, and R. E. Stickney, Surf. Sci., 1974, 46,358. D. A. King and M. G. Wells, Proc. R. SOC. London, Ser. A, 1974, 339, 245. lo' L. R. Clavenna and L. D. Schmidt, Surf: Sci., 1970, 22, 365. Figure 19 Shown opposite are photographs of scale models of three-atom and four-atom sections of Pt(ll1) and Pt(100) with the two carbon atoms (smaller dark spheres) of an acetylene adsorbate that illustrates topographical features for potential tipping of the C-C bond vector with respect to the surface plane. A constraint that both carbon atoms maintain a bonding interaction with surface platinum atoms is imposed here.The maximal tipping angles are approximately 13 and 45" for the (111) and (100) surfaces respectively. These topological diferences between Pt(l1 l)-CzHz and Pt(lOO)-CzH, may account for their qualitatively diflerent chemical behaviour 3 19 Hydrocarbon Reactions at Metal Centres Figure 20 Onflat nickel surfaces, acetonitrile chemisorbs associatively, i.e.,the chemisorption state is molecular in character. On theseflat metal surfaces, very little C-H bond breaking occurs even at temperatures where acetonitrile desorbs molecularly from the surface. In contrast, on stepped nickel surface, there is evidence of C-H bond breaking at low temperatures. This difference can be explained in a stereochemical context in that C-H hydrogen atoms can more readily approach step metal atoms than terrace atoms as schemati- cally outlined above thermodynamic activity of the chemisorbed hydrogen atoms, a feature that is enhanced by higher hydrogen pressure).The observed variation in behaviour of chemisorbed benzene, toluene, and pyridine on flat metal surfaces can be traced to the proximal effect. Similarly, the variation in behaviour of molecules like benzene, toluene, acetylene, and acetonitrile as a function of surface crystallography can also be explained in terms of a proximal effect, e.g., the much greater extent of C-H bond breaking for molecules like benzene and acetonitrile on stepped surfaces can be explained most readily in terms of stereochemical factors.There has been a general ,assumption, often stated in the literature, that surface metal atoms at stepped or kinked sites possess a unique reactivity because of electronic effects and of lower co-ordination numbers. This assumption may well be correct but it has not been established unambiguously by either experiment or theory. Possibly, the proposed proximal effect is a very important factor that embraces both an electronic and a geometric effect in these stepped-surface reactions. Acknowledgement. The research on metal-hydrocarbon chemistry has been supported generously by the National Science Foundation and by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division of the U.S. Department of Energy under Contract Number W-7405-ENG-48.I am also indebted to Professors R. G. Bergman, W. A. G. Graham, W. D. Jones, M. L. H. Green, and W. H. Weinberg for information about unpublished work, Professor R. Hoffmann for a discussion of the oxidative addition reaction, the Miller Institute for Research in Basic Science for a grant in the form of a Miller Professorship, and the students in my research group. Constructive comments concerning isotope effects by Dr. A. Shusterman are especially acknowledged.

ISSN:0306-0012    

DOI:10.1039/CS9821100283

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

The Synthesis of Leukotrienes: A New Class of Biologically Active Compounds Including SRS-A By John Ackroyd and Feodor Scheinmann DEPARTMENT OF CHEMISTRY AND APPLIED CHEMISTRY, UNIVERSITY OF SALFORD, SALFORD, MS 4WT 1 Introduction This review deals largely with the synthesis of the leukotrienes and related compounds and covers the literature up to the end of 1981. To provide an adequate background to this topic the work associated with the discovery of SRS and SRS-A and subsequent structure elucidation is briefly reviewed. The biogenesis, nomenclature, and aspects of the pharmacology of the leukotrienes have also been included since it was the potent action of the metabolites on smooth muscle and as mediators of inflammation that motivated the synthetic work.2 Historical Background In 1938, Feldberg and Kellaway ’injected cobra venom into guinea-pig perfused lungs and observed the release of a substance into the perfusate that differed from histamine and caused a slow contraction, long in duration, of guinea-pig jejunum. The agent responsible was referred to as ‘slow reacting substance’ (SRS).Two years later Kellaway and Trethewie’ demonstrated that a similar substance was pro- duced in the effluent of guinea-pig perfused lung following challenge with an appropriate antigen. This immunologically produced mediator was later termed ‘SRS-A’-‘slow reacting substance of anaphylaxis’ by Brockleh~rst.~ Much of the work in the 1950s and 1960s was motivated by the failure of anti-histamine drugs to control asthma.SRS-A became regarded as an important mediator in asthma since it was produced by lung tissue of asthmatics in vitro as well as in the guinea- pig model of asthma and it was shown to be a potent bronchoconstrictor of human airways. Since SRS-A was also involved in other ‘immediate hypersensitivity’ reactions it became regarded as a pathophysiological product with undesirable effects and no beneficial role and therefore an ideal candidate for chemo- therapy. Progress was frustated by lack of structural knowledge of the SRS group of compounds, which were difficult to purify and relatively unstable. The break-through came in May 1979 when Samuelsson4 announced in Washington, W.Feldberg and C. H. Kellaway, J. Physiol., 1938, 94, 187.* C.H.Kellaway and E. R. Trethewie, Quart. J. Exp. Physiol., 1940, 30,121. W.E. Brocklehurst, Ciba Symposium on Histamine, Churchill, London, 1956, p. 175; J. Physiol., 1960, 151, 416; Progr. Allergy, 1962,6, 540. (a) B. Samuelsson, P. Borgeat, S. Hammarstrom, and R. C. Murphy, in Advances in Prostaglandin and Thromboxane Research, ed. B. Samuelsson, R. Ramwell, and R. Paoletti, Raven Press, New York,Vol. 6, 1980, p. 1. (b) B. Samuelsson in ref. 5b. p. 45. The Synthesis of Leukotrienes at the Prostaglandin Symposium, the lipoxygenase pathway for metabolizing arachidonic acid to give peptidolipids, some of which have the biological properties of SRS. Since compounds of this new class were obtained by metabolism using polymorphonuclear leukocytes (from the peritoneal cavity of rabbits) and each possessed a conjugated triene, Samuelsson proposed the name leukotriene, and showed that SRS from different sources contain varying amounts of different leukotrienes with SRS activity.Morris et a/.,’ however, identified SRS-A from perfused guinea-pig lung as (5S)-hydroxy-(6R)-cysteinylglycinyl-trans-7,9, cis-11,lCeicosatetraenoic acid to provide a structure forty years after Kellaway and Trethewie reported the biological activity of the immunologically generated material. The original structural work of Samuelsson4 and Morris’ and their co-workers did not define the stereochemistry of the leukotrienes, and this follows largely from the synthetic work of Corey et aL6 OH COiH 0 HO OCHZCHCHZO C0,Na 3 Structure and Biogenesis of the Leukotrienes Although SRS activity was first reported in 1938, purification procedures prior to 1976 had not led to homogeneous material.The minute amounts of labile SRSs obtained were monitored by contraction of guinea-pig ileum, which contraction could be reversed by action of the SRS antagonist FPL 55712, discovered by Fisons Pharmaceutical Laboratorie~.~ Little progress had been ’(a) H. R. Morris, G. W. Taylor, P. J. Piper, and J. R. Tippins, Nature, 1980, 285, 104. (b) H. R. Morris, G. W. Taylor, C. M. Jones, P. J. Piper, J. R. Tippins, and M. N. Samhoun, in ‘SRS-A and Leukotrienes’, ed. P. J. Piper, J. Wiley and Sons, Chichester, 1981,p. 19. (a)S. Hammarstrom, B.Samuelsson, D. A. Clark, G. Goto, A. Marfat, C. Mioskowski, and E. J. Corey, Biochem. Biophys. Res. Commun., 1980,92,946.(b) E.J. Corey, in ref. 4a,p. 19.’C. W. Parker in ref. Sb, p. 131. 322 Ackroyd and Scheinmann made concerning chemical structure apart from the suggestion that SRS was a polar lipid or an hydroxy-acid derived from arachidonic acid. The following recent developments led to a rapid advance in the structure elucidation of the SRSs and biogenesis of the leukotrienes. The in uitro generation of SRS from rat peritoneal cavity cells and rat basophilic leukaemia cells after stimulation with ionophore A-23 ~ 7 . ~ The purification of SRS by reverse phase h.p.l.~.~*~ The establishment of SRSs as products from metabolism of arachidonic acid by the 5-lipoxygenase path~ay.~.’ The incorporation of sulphur-containing amino-acids in the structure was suggested by increased yields of SRS in the presence of ~-cysteine.~ Purified SRSs gave diagnostic spectral data.Thus, the ultraviolet spectrum shows a characteristic triene triplet at 280nm and mass spectral analysis of the trimethylsilyl ether N-acetyl-methyl ester of SRS from basophilic leukaemia cells and SRS-A from guinea-pig lung gave fragment ions from which the complete covalent structure could be deduced. Elegant synthetic work by Corey and his co-workers gave the detailed stereochemistry and made the leukotrienes available by total synthesis6 The structure of SRS generated by treating murine mastocytoma cells with ionophore A23187 and L-cysteine, which was elucidated by Samuelsson, is summarized in Scheme 1.The pure spasmogenic material has an absorbance at 280 nm, consistent with a conjugated triene having a non-conjugated sulphur auxochrome. By using labelled precursors it was shown that arachidonic acid, and all three carbon atoms and sulphur from cysteine were incorporated in SRS. Reductive desulphurization with Raney nickel gave 5-hydroxyarachidic acid, which indicated that the sulphur residue was linked to the arachidonic acid derivative by a thioether linkage and that a hydroxy-group was present at C-5. Reductive ozonolysis led to hexan-1-01, suggesting that the Ai4 (a-6)double bond of arachidonic acid had been retained in SRS. The position of the triene system was located by the unique hydroperoxidation with soybean lipoxygenase that converts the cis-lp-diene system at 0-6 and 0-9 to a conjugated system with introduction of oxygen at 0-6.Thus, treatment of SRS with the lipoxygenase gave a tetraene having an absorption maximum at 310 nm, indicating that SRS has a cis double bond at C-14 and C-11 and two additional double bonds at C-9 and C-7.The structural information led to the conclusidn that SRS was a derivative of 5-hydroxy-7,9,11,14-eicosatetraenoicacid with the sulphur auxochromic group at C-6. Acid hydrolysis and amino-acid analysis showed the presence of cysteine, glycine, and glutamic acid. Sequence analysis showed that the tripeptide is y-glutamylcysteinylglycine (glutathione).The synthetic work carried out by Corey et confirmed these deductions and in addition gave the stereochemistry of SRS from murine mastocytoma cells as (5S)-hydroxy-(6R)-S-glutathionyl-truns-7,9,cis-1 1,14-eicosatetraenoic acid, which has been named leukotriene C4 (LTC4). Scheme 2 shows the biosynthetic formation and transformation of the leuko- trienes derived from arachidonic acid. Thus, while cyclo-oxygenase first converts 323 Murine mastocytoma cells Ii NH2 JII H OH H OH HSCH2CHCO2 HCO2H C02H 11H NCH ,C02 H t-+ I CHCONHCH2C02 H H02CCHCH2CH2C02 H LTC4 1I NHCOCH2CHzCHC02HNH2 I NH2 HO 15-HP-LTCd Reagents: i, Ionophore A23187; ii, H+;iii, soybean lipoxygenase; iv, Raney nickel, H,; v, 03,NaBH, Scheme 1 Key experiments in the structure elucidation of SRS generated from murine mastocytoma cells PHOSPHOLIPIDS Phospholipase G z H PROSTANOIDSCycle-oxygenase> G-Arachidonic acid &OH 11-H PETE Lipoxygenase CO2H &COzH CHCONHCH2COzH 5-HPETE I /-NHCOCHzCHzCHCOgH /, 11 1-trans-LTC 44 / NH2 !COZH ;,;cozH, 7 -Leukotriene A4(LTA4) Leukotriene B (LTB4) j,~lutathione-S.transferase CO,H ‘S--CH, CHCONHCHzCOzH CHCONHCHZI I I INHCOCHzCH ZCHC02 H NH2 COzH IW h) Leukotriene C 4(LTC4) NH z Leukotriene D4( LTD4) wl Scheme 2 Biosynthetic formation of the leukotrienes The Synthesis of Leukotrienes arachidonic acid into the 11-hydroperoxy-derivative(Scheme 2; 11-HPETE) lipoxygenase forms (SS)-hydroperoxy-6,8,11,14-eicosatetraenoicacid (Scheme 2; SHPETE), which leads to the major product (5S)-hydroxy-6,8,11,14-eicosate-traenoic acid (5-HETE).However, more polar metabolites were also present and (5S,12R)-dihydroxy-cis-6,trans-8,1O,cis-14-eicosatetraenoicacid [leuko-triene B4 (LTB,)] was identified and isotopic oxygen experiments demonstrated that the oxygen of the alcohol group at C-5 originated from molecular oxygen, whereas the oxygen at C-12 was derived from water. As a result of trapping experiments, an unstable epoxide intermediate (half-life about 3 minutes) was proposed and its structure and stereochemistry was confirmed as the 5,6-epoxy- 7,9,11,14-eicosatetraenoicacid [leukotriene A, (LTA,)] by synthesis. The epoxide undergoes ring-opening enzymatically, by nucleophilic attack of the sulphydryl group of the peptide glutathione (y-glutamylcysteinylglycine) to give leukotriene C4 (LTC,) and 11-trans-LTD,.The former is converted by y-glutamyltrans- peptidase (GGTP) into the biologically more active (SS)-hydroxy-(6R)-S-cysteinylglycine-trans-7,9-cis-1 1,14-eicosatetraenoic acid, known as leukotriene D4 (LTD,). The thioethers have SRS properties. Samuelsson believes that LTD, is the main component of SRS-A from human lung and suggests that the composition of SRS-A will depend on its mode of generation and consists of a mixture of LTC, and LTD4.,* 4 Nomenclature of the Leukotrienes The leukotrienes (LT) represent a new family of acyclic eicosanoids and both systematic and new abbreviated names will replace the terms SRS and SRS-A.The system of nomenclature devised by Samuelsson and Hammarstrom bears some resemblance to that used for the prostaglandins in that numerical subscripts after the generic name are used to denote the total number of double bonds, and an alphabetical sequence refers to the substituents. The leukotrienes from arachidonic acid were originally named LTA [trans-(5S,6S)-epoxy-trans-7,9-cis-l1,14-eicosatetraenoicacid], LTB [5S,12R)-dihydroxy-6,8,10,14-eicosatetraenoic acid], and LTC [(SS)-hydroxy-(6R)-S-glutathionyl-trans-7,9-cis-11,14-eicosatetraenoicacid] following the discovery of conjugated trienes in le~kocytes.~Subsequently it was shown that the glutathione (y-glutamylcysteinylglycine) thioether group at C-6 in LTC is metabolized into the corresponding biologically more potent cysteinylglycine derivative named LTD.Further metabolism gives rise to the cysteinyl derivative LTE. To denote the total number of double bonds present, these metabolites previously referred to as LTA, LTB, LTC, LTD, and LTE, become LTA,, LTB, , LTC, ,LTD,, and LTE, .* The 5,8,11- and 5,8,11,14,17-eicosaenoicacids are transformed into analogous structures e.g. LTCJ and LTC5, respectively, and these are illustrated in Scheme 3. * B. Samuelsson and S. Hammarstrom, Prostaglandins, 1980, 19, 645. Ackroyd and Scheinmann CO, H COP H LTA,(fonnerly LTA) LTB,(formerly LTB) H OH I NHR' LTC,(formerly LTC)iR'= glutamyl, R' = glycine LTD4 (formerly LTD) R'= H, R2 = glycine LTE,(formerly LTE)IR'= H, R' = OH CHCONHCH ,CO, H I NHCOCHZCHzCHC02H I LTC, R=n-C,H,, NH 2 LTC, R= Scheme 3 Nomenclature of the leukotrienes 5 Pharmacological Properties Some pharmacological properties of the SRSs have been briefly mentioned. Isolation and synthesis of pure compounds provide further impetus for more precise studies.Thus, both LTD, and LTC, cause pulmonary changes in anaesthetized monkeys.' These effects were a profound increase in lung resistance and more modest decrease in lung compliance. Simultaneous treatment with FPL-55712 (2) prevented these changes. Although both LTD4 and LTC4 are potent in producing bronchconstrictions and increasing vascular permeability in skin,1° LTD, is the more active by five-to ten-fold.LTD, also causes impaired sputum clearance, thus giving rise to mucous plugging, which is one of the pathological features of asthma. The contractile effects of LTC3, LTD3, LTC,, and LTD, on guinea-pig ileum are similar to those for LTC, and LTD, .ll M. K. Bach, J. Brashler, H. G. Johnson,.and M. L. McNee, in ref. 56. p. 161 lo G. P. Lewis in ref. 5b, p. 227. 'I R. A. Lewis, J. M. Drazen, E. J. Corey, and K. F. Austen, in ref. 5b, p. 101. The Synthesis of Leukotrienes LTB4 stimulates leukocyte formation (chemotaxis and chemokinesis), induces an increase in capillary permeability, and causes smooth muscle contraction.'2 Its chemotactic potency for macrophage and neutrophils at concentrations of -1ngml-' (more potent than any other known lipid chemotactic factor) and its detection in the synovia of patients with rheumatoid arthriti~,'~ implies that it is a primary mediator of inflammatory and allergic states. The synthesis of LTB4 and its analogues has provided the necessary material for detailed structural studies and also for further biological work.6 Approaches to Leukotriene Syntbeses The three approaches that have been largely used for the synthesis of leuko- trienes involve (u) biomimetic syntheses from arachidonic acid, (b) the use of simple unsaturated precursors, and (c) the elaboration of carbohydrates for asymmetric syntheses. The successful use of arachidonic acid necessitated chemical met hods to mimic lipoxygenase-con trolled regioselec tive oxygenation to prepare hydroxy-, hydroperoxy-, and epoxy-derivatives at each of the double bonds, and Ithe 1 racemic products so obtained serve as standards for enzymatic studies.Corey has reviewed this problem and has shown that the carboxy-group in arachidonic acid can control the introduction of an oxygen function, nearest and farthest from it6' Many of the syntheses were aimed at the preparation LTA4, since the epoxide ring can be opened by the thiol of the appropriate peptide to give LTC4, LTD,, and LTE4. Synthesis from simple olefin and alcohol precursors remains most attractive, and requires preparation of chiral epoxides. This was achieved by stereoselective epoxidation of an allylic alcohol in the presence of chiral tartrates, or alternatively, by using carbohydrate precursors which already have the correct chirality for asymmetric synthesis.7 Synthesis of Leukotriene A Precursorsand Their Analogues This section reviews the various methods that have been used to prepare precursors of LTA4 and their analogues. The biogenetic approach involved devising methods for regioselective introduction of oxygen into arachidonic acid by hydroxylation, hydroperoxylation, and epoxidation. The other methods were largely designed to prepare chiral C7 epoxyaldehydes or alcohols for chain extension by a Wittig reaction. The chiral epoxides have been prepared either from allylic alcohols or from sugar precursors. In one case, a chiral hydroperoxyeicosatetraenoicacid has been prepared from D-glyceraldehyde."(a) A. W. Ford-Hutchinson, M. A. Bray, and M. J. H. Smith, 'Inflammation: Mechanisms and Treatment. Proceedings of Future Trends in Inflammation IV', MTP Press, Lancaster, England, 1980. (b) A. W. Ford-Hutchinson, M. A. Bray, M. V. Doing, M. E. Shipley, and M. J. H. Smith, Nature (London), 1980, 286, 264. (c) L. B. Klickstein, T. Shapleigh, and E. J. Goetzl, J. Clin. Inoest., 1980, 66, 1166. (d) R. M. J. Palmer, R. J. Stepney, G. A. Higgs, and K. E. Eakins, Prostaglandins, 1980, 20, 411. (e) M. A. Bray, A. W. Ford-Hutchinson, and M. J. H. Smith, Br. J. Pharmacol. 1981, 73, 259. (J) M. A. Bray, F. M. Cunningham, A. W. Ford-Hutchinson, and M. J. H.Smith, Br.J. Pharmacol. 1981,72,483 l3 (a) P.Sirois, P.Borgeat, A. Jeanson, S. Roy, and G. Girard, Prostaglandins and Medicine, 1980, 5, 429. (b) P. Sirois, J. Roy, and P. Borgeat, Prostaglandins and Medicine, 1981, 6, 153. Ackroyd and Scheinmann A. Chemical and Enzymatic Syntheses of !5-Hydroperoxyeicosa-truns4-ciis-8,11,14-tetraenoic Acid (SHPETE) and 5-Hydroxyeicosa-~run~cis-8,11,14-tetraenoic Acid (SHE"E).-This synthesis (Scheme 4) by Corey et all4 commences from arachidonic acid (3), which is readily converted into the iodolactone (4) in the presence of iodine, potassium iodide, and bicarbonate. This intermediate (4) now has the necessary functionality for introduction of hydroxy-and I H OH 2vii H OOH ' H OOH ez"" iv) (=;L&-H -757-(8) (9) Reagents: i, KI-I,; ii, 1,5-diazabicyclo[5.4.0]undec-5-ene;iii, Et,N in MeOH; iv, LiOH; v, MeS0,CI-Et,N, H,O,; vi, CH,N,; vii, NaBH, Scheme 4 Corey's synthesis of 5-HETE and 5-HPETE l4 E.J. Corey, J. 0. Albright, A. E. Barton, and S.4. Hashimoto, J. Am. Chem. SOC., 1980, 102, 1435. 329 The Synthesis of Leukotrienes hydroperoxy-groups at C-5 and a trans double bond at C-6. Elimination of hydrogen iodide with 1,5-diazabicyclo[5.4.0]undec-5-ene gave the unsaturated lactone (5), which was transformed to the methyl ester of (+)-5-HETE (6) by triethylamine in methanol. Saponification with lithium hydroxide gave ( & )-5-HETE (7) quantitatively. The hydroperoxy-group was introduced at C-5 by treating the ester of (+)-5-HETE (6) with methanesulphonyl chloride followed by hydrogen peroxide in ether at -110 "C.Saponification using lithium hydroxide and hydrogen peroxide gave the required ( f )-5-HPETE (9), which could be reconverted into its ester (8) with diazomethane, or reduced to (+)-5-HETE (7) with sodium borohydride as illustrated by Scheme 4. With the availability of ( f.)-HPETE and ( f )-HETE as chromatographic standards, it was now possible to investigate the ability of various plant lipoxygenases to convert arachidonic acid into (5s)-HPETE (9), and using the enzyme from potato tubers the conversion was achieved in 15% yield. Reduction with sodium borohydride gave (5s)-HETE (9). B. Phenylselenylation of Arachidonic Acid.-In contrast to Corey'~'~ iodolactone route for hydroxylating arachidonic acid at C-5, Baldwin and co-~orkers'~ used phenylselenolactonization and achieved stereoselective syntheses of (f)-methyl-5-hydroxyeicosa-trans,trans,cis,cis-6,8,11,14-tetraenoate(12) and its cis-8 isomer (13) (Scheme 5).Treatment of arachidonic acid with phenylselenyl chloride at -78 "C gave the lactone (lo), which was converted into the ester (11). Oxidative removal of the phenylselenyl group was studied under several conditions and it was observed that isomerization of the conjugated diene system can occur during selenoxide elimination. Thus, oxidative elimination with either sodium periodate or hydrogen peroxide in buffered aqueous methanol (sodium bicarbonate) favoured isomerization at C-8 and gave a mixture of the hydroxy-trans,trans-eicosatetraenoate (12; 85 %) and the hydroxy-trans,cis isomer (13; 15%) from a 70-75 % yield of reaction product.Similar results were also obtained using rn-chloroperbenzoic acid in tetrahydrofuran, followed by addition to acetic acid and treatment with di-isopropylamine in refluxing benzene. However, when the oxidative elimination with hydrogen peroxide was carried out in strong base (potassium hydroxide) isomerization was reduced and the trans-6,cis-8-isomer (1 3) predominated. C. Selective Epoxidation and Hydroxylation of Eicosa-5,8,11,14-tetraenoic (Arachidonic) Acid and Eicosa-cis-8,11,14-trienoic Acid.-Corey and his co-workers' 6,1' have achieved site-selective oxidations of arachidonic acid and related compounds by devising intramolecular reactions that lead to epoxidation of the double bond either farthest or closest from the carboxy-function.Peroxyarachidonic acid (14), prepared from arachidonic acid by reaction with carbonyldi-imidazole and hydrogen peroxide, is transformed on standing at Is J. E. Baldwin, N. V. Reed, and E. J. Thomas, Tetrahedron, 1981, 37, 263. l6 E. J. Corey, H. Niwa, and J. R. Falck, J. Am. Chem. Soc., 1979, 101, 1586. "E. J. Corey, A. Marfat, J. R. Falck, and J. 0.Albright, J. Am. Chem. Soc., 1980, 102, 1433. Ackroyd and Scheinmann HO CO, Me F b : ,=rLC0,Me t Reagents: i, PhSeC1; ii, LiOH, CH,N,; iii, NaIO, Scheme 5 Baldwin’s phenylselenylation of arachidonic acid 20°C to the 14,15-epoxide, which is isolated as the epoxy-ester (15).Space filling models suggest that perarachidonic acid may adopt a &like shape to achieve intramolecular oxygen transfer by a 15-membered cyclic transition state (19), which is energetically more favourable compared to other geometries involving smaller rings. Eicosa-cis-8,11,14-trienoicacid (16) was also converted into the peroxy-acid (17), which rearranged to the Al4,I5 epoxide (18) in high yield (Scheme 6). The A5v6 epoxide of arachidonic acid (20) was obtained from arachidonic acid by first forming an unstable iodo-6-lactone in the presence of potassium bicarbonate and tri-iodide, followed by immediate treatment with lithium hydroxide. Attempts to oxidize the cis,cis-1,4-dienes of C20 polyunsaturated acids with singlet oxygen gave all the expected ‘ene’ oxidation products.In contrast, the magnesium derivative of isopropylcyclohexylamine (MICA), obtained by addition of methylmagnesium bromide in tetrahydrofuran, was found to be remarkably effective for the epoxide + allylic alcohol conversion. Thus, the 14,15-epoxide (15) was transformed to 15-HETE (21) by MICA in 70% yield. Conversion into 331 The Synthesis of Leukotrienes 4' w C O 2 O H i ii1 lii -C02 Me (19) Reagents: i, Carbonyldi-imidazole, H202, lithium imidazolide; ii, r.t. then CH,N2; iii, K13, LiOH, CH2N2;iv, 0 "C, 70 h Scheme 6 Corey's method for epoxidizing arachidonic acid at the double bonds nearest and most remote from the carboxy-group the hydroperoxide 15-HPETE (22) occurs on treating the mesylate of 15-HETE with t-butyldimethylsilylhydroperoxide at -42 "C, followed by hydrolysis to remove the silyl group (Scheme 7).The epoxide of arachidonic acid was obtained from the corresponding A14*15epoxide (15), which was first converted into a mixture of bromohydrins [(23) and (24)]. The 15-bromo-14-hydroxy-isomer (24) underwent epoxidation at C-12 by the procedure of Sharpless'* and the cis double bond was restored at C-13 by a new method involving reductive elimination of hydroxyl and bromine using trifluoromethanesulphonic anhydride and hexamethyl-phosphorous triamide (Scheme 7). Reaction of the epoxy-acid (25) with MICA '* B. E. Rossiter, T. Katsuki, and K. B. Sharpless, J.Am. Chem. Soc., 1981, 103, 464. Ackroyd and Scheinmann Jii (R = H) C0,Me Br H /. OH 111 (R= Me) (23)+ J w C O z Me CO,Me 1v -3-c-bq--HO Br HHO Br Ii' vi f--Vii -OOH (22) Reagents: i, Internal oxygen transfer (98 %); ii, magnesium derivative of isopropylcylohexylamine (MICA); iii, acetic acid-saturated aqueous KBr; iv, anhydrous t-butylhydroperoxide, vanadyl acetylacetonate; v, trifluoromethanesulphonic anhydride; vi, LiOH in DME; vii, methanesulphonyl chloride, Et, N, t-butyldimethylsilylhydroperoxide Scheme7 Corey's route to epoxidizing arachidonic acid at C-11 and preparation of 15-HETE and 15-HPETE The Synthesis of Leukotrienes HO viii >+ Reagents: viii, MICA in THF; ix, K,CO,-MeOH; x, MICA; xi, potassium selenocyanate Scheme 8 Corey’s method for hydroxylating arachidonic acid at C-11 and C-12 gave 11-HETE and 12-HETE (ratio 1:1.5), 12-HETE was also prepared from bromohydrin epoxide (26), which formed the 11,12:14,15 bis-epoxide of methylarachidonate. Saponification to the acid (27) and reaction with MICA gave the 14,150xide of 12-HETE (28), which underwent deoxygenation with potassium selenocyanate to give 12-HETE (29) (Scheme 8).D. Synthesis of an Asymmetric Epoxide Key Intermediate for Leuk0triene.-A new procedureI8 has been developed for preparing chiral epoxides from allylic alcohols and this provides the shortest route for obtaining an asymmetric precursor of LTA, .t-Butylhydroperoxide with tjtanium(1v) isopropoxide in the presence of optically active diethyl tartrate will epoxidize a wide range of allylic alcohols with high asymmetric induction (>90 %e.e.), and enantiomeric tartrates give rise to opposite configurations in the epoxide.Two approaches gave the epoxy-ester (32). The butadiene dimer (30) was epoxidized with t-butylhydroperoxide in the presence of titanium(1v) isopropoxide and (+)-diethy1 tartrate, and the product isolated as the acetyl derivative (31). Cleavage of the double bond was achieved with ruthenium tetroxide and Ackroyd and Scheinmann (30) (31) (32) (33) Reagents: i, Titanium tetraisopropoxide, L-( + )-diethy1 tartrate, t-butyl hydroperoxide; ii, 10; -RuO,, CHzNz, KzC03; iii, titanium tetraisopropoxide, ( + tdi-isopropyl tartrate, t-but ylh ydroperoxide Scheme 9 Chiral epoxidation by Sharpless to provide key synthon for leukotrienes H H viii C02Me HO I I H' H I Reagents: i, t-Butyl-lithioacetate; ii, LiAIH,; iii, p-toluenesulphonyl chloride, pyridine; iv.lithio derivative of propargyl tetrahydropyranyl ether; v, p-toluenesulphonic acid ; vi, t-butyl hydroperoxide, L-( + )-diethy1 tartrate, titanium isopropoxide; vii, Ac,O-pyridine. 03. Jones' reagent, CHzNz;viii, 1.5 equiv. K,CO,-MeOH Scheme 10 Chiral epoxidation by Corey to procide a key intermediate jor leukotrienes The Synthesis of Leukotrienes periodate and the resulting acid methylated with diazomethane and hydrolysed with potassium carbonate to give the epoxy-alcohol (32).In starting from the ester (33) the epoxy-alcohol (32) was prepared by using( +)-di-isopropyl tartrate as the chiral catalyst since it could be separated from the product by chromatography. The work-up procedure also required modification for isolation of the fairly water-soluble product. In a similar approach by Corey and his co-w~rkers'~ the 8-methylnon- 2,7-dien-l-o1 (35) was synthesized starting from 1-bromo-3-methylbut-2-ene(34) as shown in Scheme 10. Chiral epoxidation by the Sharpless method" and acetylation of the alcohol, followed by oxidative ozonolysis of the double bond and methylation of the resulting acid, gave the epoxy-ester (36) with an optical purity corresponding to 93% e.e. which readily hydrolysed to the chiral epoxy-alcohol (32).E. Coben's Synthesis of Optically Active C-7 Leukotriene Intermediates.-In an approach to the leukotrienes and their analogues, Cohen and his collaborators2' at Hoffmann-La Roche prepared the key C7 (5S,6S)-epoxide (37) and its 6-epimer (38) from D-araboascorbic acid and L-diethyl tartrate respectively. 2,3-O-Isopropylidene-~-erythrose,prepared from D-araboascorbic acid, under- went a Wittig condensation with the phosphorane derived from [2-( 1,3-dioxan- 2-yl)ethyl]triphenylphosphonium bromide followed by benzoylation to give the ester acetal (39). Reduction of the double bond by catalytic hydrogenation and exposure of the product (40) to ozone at -78 "C gave the diester (41). The crucial transformation of the synthesis was opening of the acetal with aqueous trifluoroacetic acid to give the 6-hydroxylactone (42) whereby the oxygen functions at C-5 and C-6 have been differentiated. The mesylate (43) formed quantitatively and was converted into the required trans-(5S,6S)-epoxide (37) by opening the lactone ring with methanolic alkali (Scheme 11).A similar sequence of events was employed for the synthesis from (L)-( +)-diethy1 tartrate of the cis-(SS,6R)-epoxide, which was converted into the acetal diol (44). Oxidation of the monobenzoyl derivative (45) gave the aldehyde (46), which was converted into the olefin (47) by the Wittig procedure. Catalytic hydrogenation to (48) and treatment with ozone provided the diester (49), which was transformed with trifluoroacetic acid to the threo-hydroxy-ester lactone (50).The mesylate (51) was converted into the required epoxide (38) as before (Scheme 12). F. Syntbesisof a Leukotriene Intermediate from 2-Deoxy-D-ribose.-An improved synthesis of the leukotriene A4 intermediate (52) was published by Marriott and Bantick* starting with 2-deoxy-~-ribose. 2-Deoxy-~-ribose (53) was converted into the methyl 3,5-dibenzoyloxy-2-deoxy-~-erythro-pentofuranosides l9 E. J. Corey, S.4. Hashimoto, and A. E. Barton, J. Am. Chem. Soc., 1981, 103, 721. 2o N. Cohen, B. L. Barrier, and R. J. Lopresti, Tetrahedron Lett., 1980, 21, 4163. 21 D. P. Marriott and J. R. Bantick, Tetrahedron Lett., 1981, 22, 3657. Ackroyd and Scheinmann 0 0 (43) (3 7) Reagents: i, [2-(1,3-dioxan-2-yI)thyl]triphen.ylphosphonium bromide, n-BuLi; ii, Hz , Pt; iii, 0,; iv, CF3COZH-HzO;v, mesylation; vi, K2C03-MeOH Scheme 11 Cohen's synthesis of the key trans-epoxide, for leukotriene synthesis, from D-araboascorbic acid (54), which on demethylation gave the furanose (55).Without purification the masked aldehyde (55) was converted into a mixture of geometrical isomers (56) (E:Z; 83:17 by 'H.n.m.r.) by a Wittig reaction. Hydrogenation of (56) afforded the ethyl ester (57) and this was transformed to the methyl ester, which is a known precursor of the epoxy-aldehyde (52)26 (Scheme 13; see also Scheme 1926for alternative). 337 The Synthesis of Leukotrienes Ph OH H (45)iii Ph H CHO (47)hv 0 0 (51) viii C02Me HO (38) Reagents: i, PhCOCI, pyridine; ii, oxidation; iii, [2-(1,3-dioxan-2-yl)ethyI]triphenylphosphonium bromide, n-BuLi; iv, catalytic hydrogenation; v, 03;vi, aqueous CF3C02H; vii, mesylation; viii, K2CO3in MeOH Scheme 12 Synthesis of the chiral cis-epoxide (38) Ackroyd and Scheinmann OH (53) (54) lii OCOPh 1 iiiPhCO 1 i---PhCo -ys/OHI II IOH PhCOz OCOPh C02 Et PhCO 2 I OH (57) Reagents: i, 0.05% Methanolic HCI, benzoylation; ii, refluxed in dioxan, water, and concentrated HCl; iii, ethoxycarbonylmethylenetriphenylphosphorane;iv, H2, 10% Pd-C Scheme 13 Marriott and Bantick’s chiral synthesis of a leukotriene intermediate G.Corey’s Stereospecific Total Synthesis of (1 1R)-HETE.-Corey and Kang22 carried out the synthesis of (11R)HETE (Scheme 14) by starting from the acetonide of D-glyceraldehyde to introduce the chiral centre at C-11 and by utilizing a nucleophilic acetylide coupling reaction with an allenic bromide to generate a 1,4-diyne. Thus, the acetonide of D-glyceraldehyde (58) was converted into the trans-enal (59) by reaction with lithium ethoxyacetylide, followed by hydrogenation in the presence of Lindlar catalyst and reaction with a trace of methanesulphonic acid in wet methylene chloride.The Wittig reaction of enal (59) with the ylide from n-hexyltriphenylphosphonium iodide provided the trans-cis diene (60; 79%). The acetylene carbinol (61) was formed by deketalization of (a),monotosylation of the primary alcohol group, and epoxide formation in the presence of DBU and reaction with lithium acetylide- ethylenediamine complex. Protection of the alcohol as the silylether (62) and conversion into a mixed Gilman reagent with n-butyl-lithium and cuprous cyanide led to formation of the Czo diyne (63) by reaction with methyl 5-bromohepta-5,6-dienoate (66).Lindlar reduction of the 1,Cdiyne system and 22 E. J. Corey and J. Kang, J. Am. Chem. Soc., 1981, 103, 4618. The Synthesis of Leukotrienes viiCOZ H + -Me \ \ OH (65) H OH (64) Reagents: i, EtOCGCLi, H,-Pd/CaCO, , H+; ii, n-hexyltriphenylphosphonium iodide; iii, HCl in 4:l MeCN-H20, tosyl chloride, pyridine, 1,8-diazabicyclo[5.4.0]undec-7-ene,lithium acetylide ethylenediamine complex; iv, t-butyldimethylsilyl chloride-imidazole; v, n-BuLi, Gilman reagent-1.1 equiv.cuprous cyanide, methyl S-bromohepta-5,6-dienoate;vi, Lindlar catalyst-H2; vii, 25 “C, absence of air, acidification Scheme 14 Corey’s stereospecific synthesis of (11R)-HETE Ackroyd and Scheinmann removal of the silyl protecting group with tetra-n-butylammonium fluoride gave the chiral methyl ester of (11R)-HETE (64),which can be saponified to the relatively unstable acid (65). 8 Syntheses of LTA4 and its Diastereomers and Subsequent conversions into LTC4 and LTD4, LTEI and Related Compounds.-The synthesis of the leukotrienes and their analogues by unambigous methods was particularly important in order (a) to confirm the stereochemistry of the natural products, (b)to prepare sufficient quantities to make thorough biological studies possible, and (c) to compare the biological activity of the natural and unnatural isomers.The synthetic approach included utilizing the intermediates described in the previous sections and, in addition, new syntheses of LTA, methyl ester whereby the unstable epoxide unit was introduced at the end of the synthesis. A. Corey's Biomimetic Synthesis of LTA4 and LTC4.-( 5s)-Hydroperoxy-trans-6-cis-8,11,14-eicosatetraenoicacid [(SS)-HPETE] (67) in the form of its methyl ester was converted into the trans-epoxide under mild non-acidic conditions (Scheme 15).23 Electrophilic oxygen at C-5 was generated by use of the trifluoromethanesulphonate leaving group and 1,2,2,6,6-~entarnethylpiperidine as base at -110"C to give a mixture of LTA, (68) and the conjugated dienone (69).The mixture was separated by chromatography (p.1.c.) only after reducing the ketone (69) to the hydroxy-ester (70), and the LTA, ester (68) was converted into LTC, (71) by treatment with glutathione followed by hydrolysis. B. Corey's First Synthesis of Leukotriene A4 (LTA4).-The strategy of this synthesis (Scheme 16) involves starting with a C6 fragment that has trans double bonds, which will be at C-7 and C-9 in LTA,, then joining on a C9 synthon by a Wittig process to introduce the cis double bond, which will be at C-11, and introducing the sensitive epoxide in the last step as a C5 fragment. The mono-t-butyldimethylsilyl of trans-hexa-2,4-diene-1,6-diol(72) was converted into the aldehyde by oxidation with pyridinium dichromate and condensed with the ylide derived from non-3-en-1-01 (73) to give the tetraene ether (74).Removal of the silyl group with tetra-n-butylammonium fluoride gave the hydroxytetraene (75), which was converted into the mesylate (76). Treatment with dimethylsulphide and coupling of the resulting sulphonium salt (77) in the presence of lithium di-isopropylamide with methyl 4-formylbutyrate gave the required epoxy-ester (78) and an equal amount of the cis-5,6-epoxide. The ester (78) was saponified with cold aqueous base under argon and the salt could be reconverted to the epoxy-ester (78) with dimethylsulphate. 23 E. J. Corey, A. E. Barton, and D. A.Clark, J. Am. Chem. SOC., 1980, 102, 4278. 24 E. J. Corey, Y.Arai, and C. Miostowski, J. Am. Chem. Soc., 1979, 101, 6748. 341 The Synthesis of Leukotrienes ii J, OH H OH c02 R ~HCONHCH,CO~H I NHCOCH 2CH zCHCO2 H I NH2 (71) R=H Reagents: i, 1,2,2,6,6-pentamethylpiperidine, trifluoromethanesulphonic anhydride; ii, NaBH, iii, MeOH, Et,N, glutathione Scheme 15 Corey’s synthesis of LTCI from arachidonic acid C. Corey’s Syntbesis of trans-11-LTC, and LTD,.-The synthesis of trans-1 1-LTC4 (79) (Scheme 17) and LTD, (80) (Scheme 18) was achieved by studying the factors which controlled the stereochemistry in the Wittig reaction between the Cll trans,trans-dienal ester (81) and the C9 ylide (82).25 Thus, LTA, methyl ester was exclusively formed when the reaction was carried out in tetrahydrofuran 25 E.J. Corey, D. A. Clark, A. Marfat, and G. Goto, Tetrahedron Lett., 1980,21, 3143. Ackroyd and Scheinmann (75) (74) Reagents: i, Pyridinium dichromate; ii, p-toluenesulphonyl chloride; iii, Nal, triphenylphosphine; iv, n-BuLi; v, tetra-n-butylammoniurn fluoride; vi, rnethanesulphonyl chloride; vii, dirnethyl sulphide; viii, methyl 44ormylbutyrate; ix, cold aqueous base; x, MezSOJ Scheme 16 Corey's first synthesis oj LTAI The Synthesis of Leukotrienes OHC CO,Me + J k C0,Me + 1 COz H (83) J R= -CH,I I CHCONHCHzCOzH NHCOCH 2CH zCHCO2 H I (84)+ NH2 CO, Me CO2H (79) Reagents: i, THF-HMPA, -78 "C;ii, LiI; iii, glutathione, Et,N, MeOH; iv, K2C03 Scheme 17 Corey's synthesis of trans-11-LTC, Ackroyd and Scheinmann and hexamethylphosphoric triamide, whereas with an ethereal solution of lithium iodide and tetrahydrofuran a mixture of LTA4 and its trans-11-isomer (83) methyl esters were obtained.The mixture was separated after conversion into LTC4 and trans-11-LTC, methyl esters [(84) and (SS)]. Hydrolysis with 0.1M potassium carbonate gave trans-1 1-LTC4 (Scheme 17). For the synthesis of LTD4 (80) the methyl ester of LTA, was treated with N-trifluoroacetylcysteinylglycine methyl ester (88). The peptide portion was synthesized from N-trifluoroacetyl-L-cystine (87), which was converted into 02$". i-"C0,Me 2 iii H OH ,SHH \\ 1 CO,Me iv -01N-H (89) C02Me H OH CO, H R = CH, R'= CH,I ICHCONHCH,CO, Me CHCONHCHzCO,HI I NHCOCF, NH2 Reagents: i, PCl,; ii, glycine methyl ester; iii, triphenylphosphine; iv, methyl ester of LTA,, Et,N, MeOH; v, K2C03 Scheme 18 Corey's synthesis of LTD4 345 The Synthesis of Leukotrienes the acid chloride (87) and then treated with glycine methyl ester.Reduction of the crystalline cysteine derivative with triphenylphosphine in aqueous dimethoxy-ethane (2:l) gave the required peptide (88). LTD, was obtained from the diester (89) by hydrolysis with aqueous methanolic potassium carbonate (Scheme 18). D. Corey’s Enantiospecific Synthesis of LTA4 from ~-(-)-Ribose.-A key inter-mediate for the synthesis of LTC,, LTD,, and LTE, with the naturally occurring antipodal form is ( -)-methyl trans-(5S,6S)-epoxy-trans-7,9-cis-1 1,14-eicosatetraenoate, LTA, .To introduce the correct asymmetry into the epoxide, Corey used the chirality at C-3 and C-4 in D-( -)ribose, and the functionality at the terminii to introduce the remaining carbon atoms (Scheme 19).26 The tribenzoyl derivative of D-( -)ribose (90) was treated with ethoxycarbonyl-methylenetriphenylphosphoraneto give the a$-unsaturated ester (91).Acetylation of the hydroxy-group to give (92) and removal of the unwanted chiral centre PhCO2 iOCOPh --H 9i HPhCOz CH = CHC02 Et OH PhCd, CH , OCOPh H OCOPh OCOPh I OCOMePhC02 LOCOMe HYH ... ,CH=CHCO,Et rmu2 PhCO,xCH=CI HCH2COZEt H OCOph (93) (92)liv PhCO, phc02 OCOMe H& CO, Et Hkco ,Me PhCOz PhCO 2 (94) (95) 26 E.J. Corey, D. A. Clark, G. Goto, A. Marfat, C. Mioskowski, B. Samuelsson, and S. Hammarstrom, J. Am.Chem. SOC., 1980, 102, 1436, 3663. Ackroyd and Scheinmann PhCO, HO+ HkCO, Me f.--vii C0,Me H PhCO,(97) viii.1 Reagents: i, Ethoxycarbonylmethylenetriphenylphosphorane;ii, Ac20; iii, zinc amalgam; iv, 10 % Pd-C,H,; v, HCl, MeOH; vi, tosyl chloride, pyridine; vii, K2C03, MeOH; viii, Collins reagent; ix, l-lithio-4-ethoxybutadiene, methane sulphonyl chloride, EtJN, pH 7 phosphate buffer; x, n-BuLi; xi, N-trifluoroacetylglutathionedimethyl ester, K2C03-0.03M-KHC03 in 955 H,O-MeOH; xii, glutathione-Et3N-MeOH, K,CO, in H20-MeOH Scheme 19 Corey's enantiospecific synthesis of LTA4 and LTC4from D-( -)-ribose gave the &y-unsaturated ester (93) which was hydrogenated to the saturated ester (94).The acetyl group in (94) was transformed to a tosylate (96) by hydrolysis to the alcohol (95) followed by reaction with p-toluene sulphonyl chloride. The trans-epoxide (97) was cleanly formed by hydrolysis to the benzoyl groups in the presence of potassium carbonate in methanol. The trans double bonds at C-7 and C-9 were introduced by treating the epoxy- aldehyde (98) with 1-lithio-4-ethoxybutadiene at -78 "C, giving a secondary alcohol that was eliminated on treatment with methane sulphonyl chloride to form the resulting dienal ester (99). The synthesis of LTA, methylester (100) was finalized by a Wittig reaction with non-3-enyltriphenylphosphorane.This synthesis was improved by Marriott and Bantick who prepared the key chiral intermediate from 2-deoxy-~-( -)-ribose (see p.336).21 LTA4 methyl ester was converted by sulphydryl compounds and triethylamine into 5-hydroxy-6-thioether derivatives. LTC, was made in this way by reaction with glutathione or N-trifluoroacetylglutathionedimethyl ester followed by hydrolysis. 347 The Synthesis of Leukotrienes The availability of the epoxy-aldehyde (98) enabled Corey and his co-workers also to synthesize truns-(5S,6S)-epoxy-trans-7-cis-9,11,14-eicosatetraenoate(102). The a-lithio-derivative of the t-butylimine of trimethylsilylacetaldehyde was treated with the epoxy-aldehyde (98) to give the enal ester (101).A Wittig condensation with the ylide generated from undec-2,5-dienyltriphenyl-phosphonium mesylate gave a mixture of the required epoxytetraene (102) and LTA4. The mixture proved difficult to separate but reaction with glutathione gave LTC4 (104), and its isomer (103), which were separated by h.p.1.c. (Scheme 20). J, ii H OH CO,H C0,Me SR iii f- H OH COPH C0,Me I CHCONHCH2C02H INHCOCH2CH2CHC02H Reagents: i, t-Butylimine of trimethylsilylacetaldehyde, s-butyl-lithium, pH 7 phosphate buffer; ii, ylide generated from corresponding phosphonium mesylate by reaction with 1 equiv. LDA; iii, glutathione, Et,N; 0.1M-K2C03, H,O-MeOH Scheme 20 Corey’s synthesis of the cis-9-isomer of LTA4 and LTC4 Ackroyd and Scheinmann COz Et + t ph3p c1-DD DD (109) Reagents: i, BuLi; ii, room temperature Scheme 21 Synthesis of’ ethyl pentadeca trans-2-cis-4,6,9-tetraenoateand its subsequent [I,7]-hydrogen migration E.Rokach’s First Synthesis of LTC,.-In the first synthesis by Rokach and his co-workersZ7 of LTA4 and LTC, it was believed that the geometry of the double bonds was cis at C-9, C-11, and C-14 and trans at C-7. The strategy of the synthesis was therefore to prepare a C-15 fragment having the required geometry at the double bonds which could be condensed as the sulphonium ylide with methyl 5-oxopentanoate. Thus the trans,cis-dienal ester (105) was condensed with the ylide from the phosphonium salt (106) to give the tetraene ester (107).However, on standing at room temperature a [1,7]-hydrogen migration took place as in (109) to give (108) (Scheme 21). This spontaneous hydrogen migration cast doubts on the cis geometry at the 9,lO double bond in the leukotrienes since a similar migration would be anticipated. Subsequent work was therefore directed at preparing the trans,trans,cis,cis- tetraene alcohol (111) from the dienal ester (110) as shown in Scheme 22. The alcohol (111) was converted into the mesylate (112) and then the J. Rokach, Y. Girard, Y. Guindon, J. G. Atkinson, M. Larue, R. N. Young, P. Masson, and G. Holme, Tetrahedron Lert., 1980, 21, 1485. The Synthesis of Leukotrienes H SR H SR’ - ~ c oHO z H M e v i i~ c HO oH 2 1 H - - LTC, R= -CH2 R’= -CH2I ICHCONHCH,CO,Me CHCONHCH zC02 H I I NHCOCHzCH zCHCOzMe NHCOCHzCHzCHCOzH I I NHCOCF, NH z Reagents: i, BuLi; ii, AlH,; iii, mesylation; iv, dimethyl sulphide; v, transformed to sulphonium ylide and treated with methyl 5-oxopentanoate; vi, S-trirnethylsilyl glutathione dimethyl ester N-trifluoroacetamide; vii, K2C03 Scheme 22 Rokach’s synthesis of LTC4 350 Ackroyd and Scheinmann H H H I -111I HR \C0,Me HO Ix LTE, R'= -SCH, LTC, R'= -SCHz LTD.R'= -SCH,I I 1-CHC0,H CHCONHCHpC02 H CHCONHCH,CO,HI I I NH2 NHCOCHpCHZCHCOz H NH,I R = -SCH, NH* R = -SCH,I ICHCO, Me R = -SCH, CHCONHCH,CO, Me I I I NHCOCF, CHCONHCH ,CO, Me NHCOCF, NHCOCH ,CH ,CHCO,Me I NHCOCF, Reagents: i, (4Carboxybutyl)triphenylphosphoniurn bromide, dirnsyl sodium; ii, hv; CH2N2; iii, MCPBA; iv, AcOH, NaIO.,; v, formylrnethylenetriphenylphosphorane (84 O0); vi, forrnylrnethylenetriphenylphosphorane (3406); vii, forrnylrnethylenetriphenylphos-phorane; viii, triphenyl[(Z)-non-3en-l-yl]phosphonium chloride; ix, RSH,Et,N.MeOH; x, K2C0,, H20, MeOH Scbeme 23 Rokach's synthesis. of LTA,, C4,0,and E4 from D-and L-glyceraldehpde 351 The Synthesis of Leukotrienes sulphonium salt (1 13) prior to condensation with methyl 5-oxopentanoate to give a mixture of cis and trans epoxides (114a and 114), which were separated by h.p.1.c. Treatment of the trans-epoxide (1 14) with a S-trimethylsilyl derivative of glutatathione gave two diastereomers which were separated by h.p.1.c. After hydrolysis, biological tests showed one of the products to have the biological activity of LTC, (SRS) (Scheme 22).Epoxide openings with other thiotri- methylsilyl derivatives were also examined.*’ F. Rokach’s Synthesis of Natural Leukotrienes From the Acetonides of D-and L-Glycera1dehyde.-The synthesis (Scheme 23)28 involves preparation of the key chiral epoxy-aldehyde (12 l), whereby a chiral centre is introduced into separate diastereomeric products from epoxidation of a chiral alkene [e.g. (1 M)]. The acetonide of D-glyceraldehyde was converted into the cis-olefinic acid (1 15) by a Wittig reaction and then converted into the trans-isomer (116) photo- chemically, in the presence of diphenyldisulphide.Methylation with diazomethane and epoxidation of the ester (1 16) with rn-chloroperbenzoic acid gave a mixture of two epoxides [(117) and (118)] in a 2:l ratio, and these were readily separated by chromatography. By a similar sequence the acetonide of L-glyceraldehyde gave epoxy-esters [(119) and (120)] in a 1:2 ratio. Cleavage ofthe ketals [( 118) and (120)] and oxidation of the diols gave the epoxy-aldehyde (121) which was converted into the dienalepoxy-ester (122) by two successive Wittig reactions. The product (122) was treated with the ylide from triphenyl [(Z)-non-3-en-l-yl]phosphoniumchloride to give LTA, methyl ester, which was subsequently converted into LTC,, LTD4, and LTE, as shown in Scheme 23. G.Rosenberger and Neukom’s Synthesis of LTEI .-This route (Scheme 24)29 to LTE4 involves preparation of the racemic trans-epoxide (130) from a diyne Br CH CC C( CH2)4Me -i -* OCH(Me)OEt -(123) zB J. Rokach, R. N. Young, M.Kakushima, C.-K. Lau, R. Seguin, R. Frenette, and Y. Guindon, Tetrahedron Lett., 1981, 22, 979. 29 M.Rosenberger and C. Neukom, J. Am. Chem. SOC., 1980, 102, 5426. Ackroyd and Scheinmann Br-vii CO,H COZ R I R=Me SCH, R=H I CHCO 2 RI Reagents: i, CuCI, EtMgBr, ethyl vinyl ether adduct of (E)-l-hydroxypent-2-en-4-yne; acetone,ii, 0.2N-HzSO,; iii, pyridinium dichromate; iv, vinylmagnesium chloride; v, PBr, ; vi, tetrahydrothiophene; vii, methyl 4-formylbutyrate, benzyltrimethylammonium chloride; viii, Hz, Lindlar catalyst; ix, methyl ester of L-cysteine Scheme 24 The Hoflmann-La Roche synthesis of LTE, The Synthesis of Leukotrienes (123) whereby the cis double bonds at C-11 and C-14 can be introduced by catalytic reduction with a Lindlar catalyst.Copper-catalysed coupling of 1-bromo-oct-2-yne with the ethyl vinyl ether adduct of (E)-l-hydroxypent-2-en- 4-yne gave the enediyne (123), which was hydrolysed to the alcohol (124) and then oxidized to the C1, aldehyde (125). Conversion of (125) into the vinylalcohol (126) with vinylmagnesium bromide followed by treatment with phosphorus tribromide gave the all trans bromide (127), which yields the sulphonium salt (128) with tetrahydrothiophene. The reaction of the product (128) with methyl 4-formylbutyrate gave the trans-epoxide (129; 38 %) and its cis-isomer (12 %), which were separated by chromatography.Hydrogenation of the trans-epoxide (129) gave the racemic tetraene (130). Addition of the methyl ester of L-cysteine generated a pair of diastereomers which were separated on silica gel. Hydrolysis of the esters gave LTE, and its (SR,6S)-isomer, with LTE, showing the greater spasmogenic activity in the guinea-pig ileum assay. H. Gleason’s Convergent Syntbesis of LTA4 Methyl Ester.-This synthesis (Scheme 25)30 involved coupling a C1 (2,Z)-diene ylide (138), prepared from oct-2-yn-1-01, with a C9 y-epoxy-a$-unsaturated aldehyde (134) prepared from methyl 4-formylbutyrate. The reaction of the stabilized ylide (132) with methyl 4-formylbutyrate (131), followed by epoxidation, gave the epoxy-aldehyde (133) which was then treated with the same ylide (132) to give methyl 9-oxo-trans-5,6- epoxy-(7E)-enoate (134).For the second fragment oct-2-yn-1-01 (135) was converted into 1-bromo-oct-2-yne and treated with the Grignard derivative of propargyl alcohol to give the diynol (136). Catalytic hydrogenation gave exclusively the (2,Z)-diene (137) which was converted into the ylide (138) for coupling with the epoxy-aldehyde (134). The final product was -2:l mixture of E and Z isomers about the C-9 double bond. I. Sib’s Re-investigation of the Biomimetic Synthesis of LTA4, LTC4, and Related Compounds-The group at the University of Wisconsin examined the stereospecificity of converting (-t) 5-HPETE methyl ester (139) into LTA, methyl ester (Scheme 26).31 Treatment of (k) 5-HPETE (139) with methane- sulphonyl chloride and dicyclohexylamine gave not only LTA, methyl ester (140) and the 5-ketone (141) but also the (7E,9,11,14Z)-isomer of LTA, methyl ester (142), which at room temperature is transformed into the known tetraene (143) by a 1,7-hydrogen shift.The reaction of the epoxides [(la) and (142)] with glutathione led to LTC, and the analogues (SR,6S)-LTC, (144) (9Z)-LTC4 (146) and (5R,6S,92)-LTC, (145). In an attempt to find stereospecific conditions for converting 5-HPETE into LTA,, in accordance with a previous report,23 various reaction conditions were studied for the 1,7-elimination 30 J.G. Gleason, D. B. Bryan, and C. M.Kinzig, Tetrahedron Lett., 1980, 21, 1129. 31 V. Atrache, J.-K. Pai, D.-E. Sok,and C. J. Sih, Tetrahedron Lett., 1981, 22, 3443. 354 Ackroyd and Scheinmann OHC CO Me W -(131) i OHC C02 Me _.___) Ph3P= CHCHO (132) lii 0 iv HOCH2CE C(CH,),Me -HO-& (1 35) Reagents: i, Refluxing toluene; ii, H202, NaHCO,, iii, Ph,P=CHCHO; iv, PBr,; HOCH2C=CH, EtMgBr; v, Pd, BaSO,, H,; vi, transformed via the allylic bromide to the phosphonium salt [70% from (135)l; vii, BuLi, aldehyde Scheme 25 Gleason’s convergent total synthesis of LTA, process, and in addition the triflate leaving group was used. Although a trans-epoxide was always formed to the exclusion of the cis-isomer, a mixture of (9E)- and (92)-epoxides [(la) and (142)] and 5-ketone (141) was obtained under various conditions, with a higher dilution of methylene chloride-ether (1:l) solvents giving the epoxides [(la) and (142)] in a 1:l ratio (13% and 15%) with a decreased proportion of the ketone (141; 7 %).Recently, four new dihydroxyeicosatetraenoic acids were isolated from a human leukocyte preparation which had been incubated with arachidonic The Synthesis of Leukotrienes OOH C0,Me + H OH CO, Me (142)tii CO, Me RS H RS H *C02 .OH H RS H t RS H R=- CH2 I 1CHCONHCH ,CO, H NHCOCH,CH,CHCO, H I NHZ Ackroyd and Scheinmann acid.32 These are 14,15-dihydroxy-5,8,10,12-eicosatetraenoicacid (two isomers) and 8,15-dihydroxy-5,9,11,13-eicosatetraenoicacid (two isomers) and all of them probably originate from 14,15-epoxy-5,8,10,12-eicosatetraenoicacid (14,15-LTA4), which in turn would be derived from 15-HPETE. The Wisconsin group therefore re-investigated the conversion of 15-HPETE into 14,15-ep0xides~~ and found that two isomeric epoxides [(147) and (148)] and the ketone (149) were obtained.31 The cis-10 isomer (148) undergoes rearrangement by a 1,7-hydrogen shift to the tetraene (150) (Scheme 27).CO, Me \ \ OOH CO, Me t CO, Me 0 (149) Reagents: i, (CF,S02)20,1,2,2,6,6-pentamethylpiperidine;ii, 1,7-hydrogen shift Scheme 27 Sih's re-investigation of the synthesis of 14,15-LTA4 and its isomers 32 W. Jubiz, 0.Radmark, J. A. Lingren, C. Malmsten, and B. Samuelsson Biochem. Biophys. Res. Commun., 1981.99, 976. 33 E. J. Corey, A. Marfat, and G. Goto, J. Am. Chem. Soc., 1980, 102, 6607. Reagents: i, CH2Cl2,CH3SO2C1,DCMA; ii, glutathione, Et3N; ester cleavage; iii, room temperature, 1,7-hydrogen shift Scheme 26 Sih's biomimetic synthesis of LTA, 357 The Synthesis of Leukotrienes 9 Synthesis of Leukotriene B4 (LTB4) The three syntheses of LTB4 described in this review involve the preparation of two chiral synthons, which are then joined by a cis-Wittig reaction. One synthon comprises C-1 to C-6 of LTB4 with the (5s) stereochemistry while the other segment consists of C-7 to C-20 with the (12R) configuration of the hydroxy-group incorporated.34- 36 A.Corey’s First Synthesis of LTB4.-At the outset of this synthesis the geometry of the conjugated triene system was uncertain but CoreyJ4 considered that, by analysis of the transition state of cation formation from LTA4, LTB,, should have a cis-6,trans-10-triene system (165) and this was confirmed by total synthesis and bioassay of the product. (1 54) (153) 1. H0&c02Me PhC02PhC02 H H+cozMe H (155) (1 56) Reagents: i, 2-Methoxypropene, pyridinium tosylate; pH 5.5 phosphate buffer; ii, methoxycarbonyl- methylenetriphenylphosphorane;iii, H, , Pd-C; iv, tosyl chloride, pyridine; 2 % HCL, MeOH, and ketal; 2 equiv. K2C03, MeOH; v, benzoyl chloride, pyridine; dimethyl carbonate-H,O-70 % perchloric acid; vi, Pb(OAc), Scheme 28 Corey’sjrst synthesis of LTB4 34 E.J. Corey, A. Marfat, G. Goto, and F. Brion, J. Am. Chem. SOC., 1980, 102, 7986. 35 E. J. Corey, A. Marfat, J. Munroe, K.S. Kin, P.B. Hopkins, and F. Brion, Tetrahedron Lett., 1981, 22, 1077. 36 Y. Guindon, R. Zamboni, C.-K. Lau, and J. Rokach, Tetrahedron Lett., 1982, 23, 739. Ackroyd and Scheinmann /, ix Reagents: n-Hexylidenetriphenylphosphorane; viii, tosyl chloride, pyridine: ix, 10 O0 HCI, MeOH; 1.1 equiv. phenyl chloroformate; diazabicyclo[4.3.0]nonene; x. LiOH ; Pb(OAc),; xi. allylidenetriphenylphosphorane;xii, HBr, CH2C12; xiii, triphenylphosphine; xiv. n-BuLi; aldehyde (156); xv, K2C03. MeOH Scheme 29 . Corey's jrst synthesis of LTB, 359 The Synthesis of Leukotrienes The C- 1-C-6 segment was constructed from 2-deoxyribose, whereas the C-7-C-20 portion was constructed from D-( + )-mannose and the correct stereochemistry at C-5 and C-12 was incorporated from the sugars.2-Deoxyribose was converted into the acetonide (151) with 2-methoxypropene and coupled with methoxycarbonylmethylenetriphenylphosphorane to give the a,@-un-saturated ester (152) which was hydrogenated to the hydroxy-ester (153). Tosylation of the primary alcohol group and deketalization led to the formation of the epoxy-ester (154). The (5s)-hydroxyl was protected as the benzoyl derivative and the epoxide ring converted into the glycol (155) and cleaved to the aldehyde-ester (156) with lead tetra-acetate (Scheme 28). The second segment came from the cyclic hemiacetal (157) synthesized from D-( + )-mannose.Chain extension with n-hexylidenetriphenylphosphorane and tosylation produced the cis-olefin (158) which was converted into the cis-epoxide (159) by removal of the protecting groups and treatment with phenyl chloroformate. Hydrolysis of the carbonate group and glycol cleavage gave the aldehyde (160) which was converted into the triene-epoxide (161) on reaction with allylidene triphenylphosphorane. Addition of hydrogen bromide gave the bromo-alcohol (162) which formed the phosphonium salt (163). Conversion of (163) into its ylide followed by coupling with the aldehyde (156) afforded the 5-benzoyl derivative of LTB4 methyl ester (164). Hydrolysis of ester groups gave the desired LTB4 (165) and -15% of the 6-trans-isomer, which were separated by reverse-phase h.p.1.c.B. Corey's Second Synthesis of LTB,.-The second synthesis by Corey (Scheme 30)35 again involves joining C6 and CI4 synthons with the final stage and utilizes a novel internal elimination reaction to form a new double bond and open an epoxide ring by taking advantage of the hydroxy-group at C-5 and the cis-alkene at C-6 to provide the low energy pathway [(178)-+(179)]. HOH~C' II -ti /' OC( Me) OMe Ackroyd and Scheinmann OTos f-- c- CO Me PhCO HO' H H Hgeeviii c--. C02H 0- *-H (179) (CH2),CO, (1 78) Reagents: i, CrO,-ZPyr; ii, 2-methoxy-2-propyl ether of 3-hydroxypropyltriphenylphosphonium bromide, n-BuLi; iii, MeC0,H; iv, tosylation; halogenation; triphenylphosphine; v, methoxycarbonylmethylenetriphenylphosphorane; vi, H2, Pd-C; vii, tosyl chloride-pyridine; viii, K2C03, MeOH; ix, benzoyl chloride, pyridine; x, dimethyl carbonate-H20, perchloric acid; xi, Pb(OAc),; xii, (170), n-BuLi; xiii, K2CO3 Scheme 30 Corey's second synthesis of LTB, The Synthesis of Leukotrienes HO H COzEt HO -OH HO iiiI +ECu mC5Hll C02Et I Li t-BDMSiO t -BDMSiO (185) (184) (183) I" t-BDMSiO CO 2 Et Hoc10CO2Et viii H' -#z (1 89) (1 88)1.4. t-BDMSiO * t-BDMSiO CH 2Pph Br --HcHHc (190) (191) Reagents: i, Ethoxycarbonyltriphenylphosphorane; ii, NaOEt; iii, TosC1, pyr; 2 equiv. t-butyldimethylsilyl chloride; iv, NaI, refluxing acetone; v, heterocuprate reagent (185); vi, n-Bu,NF; MsCl, Et3N; vii, NaOEt, EtOH; viii, t-butyldiphenylsilylchloride;ix, AlH,; x, CBr,, Ph,P; xi, hydrogenation.formation of acetonide: xii, BzCl. Et3N; 1N-HCl in MeOH; xiii, Pb(OAc),; xiv, phosphonium salt (191), BuLi; (194): xv, n-Bu,NF; xvi, K,COJ, MeOH-H20 Scheme 31 Rokach's stereospecific synthesis of LTB4 Ackroyd and Scheinmann xii1 (193) ixiv t-BDPSiO C02Et + CO2Et Jx" LTB4 The Synthesis of Leukotrienes The C,, fragment was synthesized from trans-2,3-epoxy-undeca-cis-5-en-l-ol (166) by oxidation to the aldehyde (167) followed by chain extension with the ylide prepared from the 2-methoxy-2-propyl ether of 3-hydroxypropyl-triphenylphosphonium bromide to give the cis-alkene (168) in 92 % yield. Deprotection of (168) gave the alcohol (169) which was converted uia the corresponding tosylate and iodide into the phosphonium salt (170).The C6 synthon was prepared from 2-deoxyribose which gave, by a Wittig condensation with methoxycarbonylmethylenetriphenylphosphorane, the triol ester (171). Conversion of (171) into the epoxy-diester (174) was achieved by (a) hydrogena- tion to the saturated triol ester (172), (b) monotosylation of the primary hydroxy-group, and (c) oxirane ring closure to (173) with potassium carbonate followed by benzoylation. The key C6 aldehyde (176) was produced by opening the epoxide ring followed by glycol (175) cleavage with lead tetra-acetate. A Wittig reaction of the aldehyde (176) and the ylide from phosphonium salt (170) produced the all cis-triene diester (177).Hydrolysis of the ester (177) with aqueous methanolic potassium carbonate also led to the facile epoxide- allylic alcohol conversion [(178)+( 179)] to give LTB4. C. Rokach’s Stereospecific Synthesis of LTB, .-This synthesis (Scheme 3 1)36 utilises 2-deoxy-~-ribose for the preparation of the C-1 to C-6 and also the C7 to Cz0 fragments of LTB4. For the synthesis of the larger fragment, 2-deoxyribose (180) is coupled with ethoxycarbonylmethylenetriphenylphosphor-ane to give the triol-ester (18 l), which cyclises to the C-glycoside (182). Tosylation of the primary-alcohol group followed by silylation gave the ether (183). Displacement of the tosyl group to give the iodo-C-glycoside (184) followed by chain extension with the cuprate (185) gave the CI4 unit (186).Removal of the silyl protecting group and mesylation gave the key intermediate (187). Ring opening of the glycoside ring with sodium ethoxide gave the hydroxy-triene (188). Some C-glycosides with a tetrahydrofuran ring are masked diene precursors and treatment with base causes b-elimination to give an epoxide which undergoes further ring opening.36 The product (188) was converted into the phosphonium salt (191) by (a) protecting the alcohol group as the ether (189), and (b) converting the ester group into the alcohol (190) and then into the bromide, which was displaced with triphenylphosphine. For the C-1 to C-6 fragment the chirality at C-3 of 2-deoxyribose was incorporated into the hydroxy-group at C-5.The intermediate (18l), used previously for preparation of the CI4 synthon, was hydrogenated and con\l:rted into the acetonide (192). Benzoylation and removal of the ketal group gave the diol (193) which was cleaved to the aldehyde (194) in the presence of lead tetra-acetate. Coupling of the C6 fragment (194) with the CI4 fragment (191) by a Wittig reaction gave a mixture of the cis-ester (196) and its trans-A637 isomer (195) which were separated by h.p.1.c. The cis-isomer (196) was converted into LTB4 by ether cleavage and ester hydrolysis. Ackroyd and Scheinmann 10 Inhibitors of SRS-A Activity From some years prior to the elucidation of the structures of the leukotrienes, sodium chromoglycate (197), also known as Intal, has been used as an anti-allergy agent in the treatment of asthma.Chromone-2-carboxylic acids of this class are not direct end-organ antagonists of putative mediators, including SRS-A, but are believed to act primarily by inhibiting mediator-release in immediate hypersensitivity reaction^.^' A selective antagonist of SRS-A is FPL 55712 (2), mentioned previously, and its properties have been re~iewed.~’ A series of anti-allergic 4-hydroxycoumarins has been prepared by Buckle et and one of the most potent is BRL 19880 (198), which not only antagonises the action of SRS-A but also acts as an inhibitor of histamine relea~e.~’.~’ Another compound of interest is FPL 59257, the chromonepropionic acid (199), which lacks the alcoholic hydroxy-group often present in potent chromones.This compound, while not as potent as FPL 55712, is a selective antagonist of SRS-A with a longer duration of a~tivity.~’ Rotenone (200) the natural chromanochromanone and some related compounds have also been found to be SRS-A antagonist^.^^ With knowledge of the structure of the SRS group of compounds, it is now possible to design and prepare fatty-acid derivatives which may inhibit leukotriene biosynthesis and examples of this approach are now described. 0 OCH2CH(OH)CH20 0 Na02C CO Na OH 37 P.Sheard in ref. 56, p. 209. 38 D. R. Buckle, D. J. Outred, J. W. Ross, H. Smith, R. J. Smith, B. A. Spicer, and B. C. Gasson, J.Med. Chem., 1979, 22, 158. 39 R. J. Ashack, L. P. McCarty, R. S. Malek. F. R. Goodman, and N. P. Peet, J. Med. Chem., 1980, 23. 1022. The Synthesis of Leukotrienes 0 HO OCH 2CH 2 CH 20 CHzCHzCOz H Pr Pr OMe A. Synthesis of Three Cz0 Potential Inhibitors of Leukotriene Bi0synthesis.-Corey et al. synthesized three unsaturated eicosanoic acids as possible candidates to antagonise the action of leukotriene antagonists by inhibiting their bio~ynthesis.~' 5,6-Dehydroarachidonic acid (206) was synthesized from arachidonic acid (201) starting from the 5'6-epoxide of methyl arachidonate (202) which was converted into a mixture of position isomeric bromohydrins (203). Treatment of the mixture (203) with Jones' reagent gave a mixture of the position isomers of bromo-ketones (204a, b).Treatment of this mixture with 2,4-dinitrobenzenesulphonylhydrazine afforded the methyl ester (205). Hydrolysis of the methyl ester yielded the desired 5,6-dehydroarachidonic acid (206) (Scheme 32). To obtain the methyl ester of cis-8-eicosen-5-ynoic acid the following reaction procedure was used. 1-Tridecyne on reaction with iodine and morpholine gave 1-iodo-1-tridecyne which was reduced to cis-1-iodo-1-tridecene (207). Lithiation and treatment with cuprous iodide-dimethylsulphide complex generated the corresponding Gilman reagent which coupled to methyl 7-iodo-5-heptynoate producing the desired methyl ester (208) (Scheme 32). Synthesis of the racemic methyl ester of the thio-analogue of LTA4 (216) started from reduction of methyl 7-hydroxy-5-heptynoate (209) to the cis-olefin (210) followed by oxidation and concurrent isomerization to the trans-a,j- unsaturated aldehyde (21 1).Reduction of the aldehyde gave the 7-hydroxy- trans-Sheptenoate (212), which on epoxidation afforded the epoxy-alcohol (213). Oxidation of the alcohol yielded the epoxy-aldehyde (214) which by 40 E.J. Corey, H. Park, A. Barton, and Y. Nii, Tetrahedron Lett., 1980, 21, 4243. Ackroyd and Scheinmann +CO, Me (204b) Jvii CO Me viii (208) (207) Reagents: i, KHC03, potassium tri-iodide; LiOH, THF-H,O; CHzN2; ii, solid solution KBr in HOAc-H,O-THF; iii, Jones' reagent; iv, 2,4-dinitrobenzenesulphonylhydrazine;v, LiOH, THF-H,O; vi, 12, morpholine in benzene; vii, potassium azodicarboxylate, HOAc.MeOH-pyridine; viii, 2 equiv. t-butyl-lithium; cuprous iodide-dimethyl sulphide; methyl 7-iodo-5-heptynoate Scbeme 32 Synthesis of cis-8-eicosen-5-ynoic acid The Synthesis of Leukotrienes C0,Me iii~ HO-HO*,Me v ,OHC C0,Me C0,Me Reagents: i, H2 Lindlar catalyst; ii. pyridinium chlorochromate; iii, NaBH,; iv, MCPBA; v, Py2-CrO,; vi, triphenyl[(Z)-non-3-en-l-yl]phosphoniumiodide, n-BuLi; vii, 0.05M sodium thiocyanate. Scheme 33 Synthesis of the thio-analogue of ( _+ ) LTA, Wittig reactions was chain-extended to the dienal (215) and then to the methyl ester of ( f)-LTA4. Treatment of ( k)-LTA4 with sodium thiocyanate gave the methyl ester of the racemic thiirane (216) (Scheme 33).Reagents: i, Bu'Ph,SiCl, imidazole, DMF; ii, coupled to (E)-l-bromo-oct-2-ene, n-BuL1; in, HF-pyridine; iv, CBr,-PPh, ;v, (EtO),PO; vi, excess methyl(triphenylphosphorany1idene)acetate. Jones' oxidation; vii, di-isobutyl aluminium hydride; CH2Nz; viii, CH,I,-Zn-CuCI; ix, Cr0,-pyridine; x, (220), LDA; xi, H2, Lindlar catalyst; xii, LiOH Scheme 34 Nicolaou's synthesis of 5,6-methanoleukotriene A, Ackroyd and Scheinmann 369 The Synthesis of Leukotrienes B. Synthesisof S,&MethanoleukotrieneA4,A Selective Inhibitor of Biosynthesis.-Nicolaou et d4'synthesized a stable analogue of LTA4, 5,6-methano-leukotriene A4 (Scheme 34), which after preliminary studies appears to be potent and selective in hi bi tor of leu ko triene bios yn t hesis .4 ' For the preparation of a CI3 synthon, pent-2-en-4-yn-1-01 was protected as its t-butyldiphenylsilyl ether (217) and coupled to (E)-l-bromo-oct-2-ene to afford product (218). Removal of the silyl ether followed by treatment with CBr4-PPh3 led to the bromide (219) which was converted into the required phosphonate (220) by exposure to an excess of triethylphosphite in acetonitrile.The second component required to assemble the leukotriene skeleton was constructed from 6-valerolactone. 6-Valerolactol (22 1) was treated with an excess of methyl(triphenylphosphorany1idene)acetate to afford, after Jones' oxidation, the cq3-unsaturated methylcarbonyloxy-carboxylic acid (222). Treatment of this mono-ester (222) with di-isobutylaluminium hydride followed by esterification with diazomethane gave the allylic alcohol (223), which on cyclopropanation in the presence of methylene iodide, zinc, and cuprous chloride afforded (224).Oxidation of (224) gave aldehyde (225) which was required as the second component. Generation and stereocontrolled coupling of the lithium salt of the phosphonate (220) with the aldehyde (225) produced the compound (226). Selective hydrogenation in the presence of Lindlar catalyst led to the methyl ester (227). Finally, hydrolysis gave the required 5,6-methanoleukotriene A4 (228). 11 Concluding Comments The number of papers that have been published, mostly in preliminary form, on the synthesis of leukotrienes since their structures were announced in 1979, emphasises the importance of this field of research.The contributions from the laboratory of Professor E. J. Corey at Harvard University are outstanding, not only for the asymmetric syntheses of molecules that are often unstable, but also for the new methods that have been devised for the purpose. We may confidently expect further progress in the syntheses of leukotrienes in order to make the compounds more readily available for bio-assay, as well as progress in the design and preparation of analogues which may be of value as anti-allergy drugs. One of us (J.A.) thanks the SERC for the award of a studentship. 41 K. C. Nicolaou, N. A. Petasis, and S. P. Seitz, J. Chem. SOC.,Chem. Commun., 1981, 1195.

ISSN:0306-0012    

DOI:10.1039/CS9821100321

出版商:RSC    

年代:1982

数据来源: RSC