摘要:

Quarterly Reviews No2 Vol21 1967 Molecular Hyperpolarisabilities By A. D. Buckingham and B. J. Orr Page 195 Photochemical Behaviour of Transition-metal Complexes By E. L. Wehry 21 3 The Cephalosporin C Group By E. P. Abraham 23 1 Hydrogen Abstraction in the Liquid Phase by Free Radicals By R. S. Davidson 249 Grignard Reagents. Compositions and Mechanisms of Reaction By E. C. Ashby 259 The Chemical Society London Quarterly Reviews contains articles by recognised authorities on selected topics from general physical inorganic and organic chemistry. The Journal and Annual Reports interest primarily the research worker Quarterly Reviews is designed for a wider range of readers. It is intended that each review article shall be of interest to chemists generally and not only to workers in the particular field being reviewed. The submission of reviews for publication is welcomed but intending authors are advised to write in the first place to the Editor The Chemical Society Burlington House Piccadilly London W. 1. Such preliminary communications should be accompanied by an outline of the ground to be covered rather than by the completed manuscript. Price to non-fellows &4 10s. Od. per annum @ Copyright reserved by The Chemical Society 1967 Published by The Chemical Society Burlington House London. Printed in England by The Thanet Press Margate.

ISSN:0009-2681    

DOI:10.1039/QR96721FP003

出版商:RSC    

年代:1967

数据来源: RSC

摘要:

Molecular Hyperpolarisabilities By A. D. Buckhgham and B. J. Orr BRISTOL 8 DEPARTMENT OF THEORETICAL CHEMISTRY UNIVERSITY OF BRISTOL 1 Introduction The idea that an atom or molecule may be distorted by a uniform electric field so that a dipole moment is induced in direct proportion to the field strength has long been a basic postulate in physical and chemical theories. This postulate may be expressed algebraically in the form m = aF (1) where m (a vector) is the dipole moment induced in the system by the uniform electric field F (also a vector) and the proportionality constant a (a second-rank tensor) is known as the polarisability of the system. Equation (1) is satisfactory for normal laboratory fields but fails when the induced moment is comparable with a permanent molecular dipole. In strong fields eqn.(1) becomes m=arF+#F2+$yF3+ . . . (2) where the coefficients fi and y of the higher powers of F are referred to as the first and second hyperpoZarisabiZities respectively. As shown in Section 2A fi is zero for systems with inversion symmetry and the term involving y then represents the initial deviation of m from a linear dependence on F. Figure 1 illustrates the dependence of m on F obtained by substituting typical values of I 2 3 F OO*v~tts/cmI) Fig. 1 Typical dependences of the induced moment m on electric jieId strength F. Values of a p y (in e.s.u.) are respectively curve ( A ) + 0 + 10-36; curve ( C ) + - + 10-ss. 0 0 ; curve (B) + 195 Quarterly Reviews a p and 7 in eqn. (2). Other varieties of hyperpolarisability describing the distortion of molecules by magnetic fields and electric field-gradients exist but these are not considered in this Review.Quantitative calculations of the magnitude of y were first made by Sewel12 for atomic hydrogen and by Coulson Maccoll and Sutton? who referred to earlier suggestions that eqn. (1) might fail at high field strengths. Allowances for hyperpolarisabilities have been incorporated in theories of a number of physical effects including electric birefringence (the Kerr effect),4s5 dielectric satura- t i ~ n ~ ’ depolarisation of scattered light,8,9J0 the pressure-dependence of molar refractivity;lJ2 and intermolecular forces.13 The topic has been surveyed briefly by Le F&vre.14 The development of lasers has recently brought into prominence the subject of ‘non-linear optics’. Here hyperpolarisabilities play an important r6le since they are responsible for the phenomenon of optical harmonic generation (fre- quency doubling and trebling) and other associated effects.There are several reviews of this rapidly expanding s u b j e ~ t ? ~ - ~ ~ ~ In this Review we discuss the properties and measurement of hyperpolaris- abilities and attempt to suggest future lines of research in this field. 2 Properties of Hyperpolarisabilities A. Symmetry Properties.-Equation (2) can be rewritten in Cartesian tensor notation where ma is the a-component of m aaB the ap-component of a and so on; the Greek suffixes denote Cartesian components and a repeated suffix implies a summation over the X Y and 2 components.19 It may be assumed that the A.D. Buckingham Quart. Rev. 1959,13 183; Chem.in Britain 1965 1 54. G. L. Sewell Proc. Cambridge Phil. Soc. 1949 45 678. C. A. Coulson A. Maccoll and L. E. Sutton Trans. Faraday SOC. 1952 48 106. A. D. Buckingham and J. A. Pople Proc. Phys. SOC. 1955 A 68,905. S. Kielich Acta Phys. Polon. 1958 17 239. A. D. Buckingham J. Chem. Phys. 1956 25 428. A. D. Buckingham and M. J. Stephen Trans. Faraday SOC. 1957 53 884. A. L. Andrews and A. D. Buckingham Mol. Phys. 1960 3 183. ’ S. Kielich and A. Piekara J. Chem. Phys. 1958 29 1297; Acta Phys. Polon. 1959,18,439. lo S. Kielich Acta Phys. Polon. 1960 19 149; 1963 23 321 819. l1 A. D. Buckingham Trans. Faraday SOC. 1956 52 747. l2 S. Kielich Acta Phys. Polon. 1962 22 477. l3 A. D. Buckingham Discuss. Faraday Soc. 1965 no. 40 232. l4 R. J. W. Le Fevre Adv. Phys. Org. Chem. 1965,3 1. l5 P.A. Franken and J. F. Ward Rev. Mod. Phys. 1963 35,23. l6 N. Bloembergen Proc. I.E.E.E. 1963 51 124; ‘Nonlinear Optics’ Benjamin New York 1965. l7 R. W. Terhune Solid State Design 1963 4 38; R. W. Terhune and P. D. Maker in a review entitled ‘Nonlinear Optics’ Prog. in Lasers vol. 11 in the press. l8 P. S. Pershan Progress in Optics 1966 5 85. l8 (a) J. A. Giordmaine Scientific American 1964 210 (4) 38. l9 H. Jeffreys ‘Cartesian Tensors’ Cambridge Univ. Press 1931 ; G. Temple ‘Cartesian Tensors’ Methuen London 1960. 196 Table 1 Symmetry properties of hyperpolarisabilities Quarterly Reviews tensors aaB Pagy and yagys are symmetric in all suffixes that is aaB = alga is strictly valid for static fields F; for optical fields the influence of dispersion may invalidate the assumption.15 However it represents a reasonable approxima- tion as long as the frequencies of the optical fields are not close to absorption bands (see Section 5).The number of independent constants required to specify the tensors p and y for various molecular symmetries has been tabulated,8 and a group-theoretical study of p has recently been reported.20 The non-zero components of B and y for several symmetries are listed in Table 1 where symmetry groups are described in the Schonflies notation.21 The subscripts 1 2 3 refer to molecular axes the assignment of which is illustrated in Figure 2. For the groups Dmh Cmv D3h 8apy = Pyga = Pgya = Pay@ - Pya8 = Pgay and Yagys = Ysyga etc* This Fig. 2 Symmetry notation and designation of molecular axes for typical molecules C3, and C,, the 3-axis is along the major axis of rotation and the 13-plane coincides with a plane of symmetry.For the group T, the axes 1 2 3 are the edges of the cube containing the tetrahedron. It can be shown that all components of fl are zero for molecules possessing a centre of inversion and that all polar molecules have a non-vanishing p. The second hyperpolarisability y is not subject to the same symmetry restrictions and is non-zero for all atoms and molecules. For any molecule the mean first hyperpolarisability can be defined as and the first hyperpolarisability anisutrupy as 2o S . J. Cyvin J. E. Rauch and J. C . Decius J . Chem. Phys. 1965 43 4083. 21 A useful tabulation of symmetry symbols and elements is to be found in 'Tables of Inter- atomic Distances and Configuration in Molecules and Ions' Chern.SOC. Special Publ. No. 11 ed. L. E. Sutton et al. 1958 pp. 18-26. 198 Buckingham and Orr where the 3-axis is that of the permanent dipole. The parameters p and dfl specify completely the tensor of a molecule with C, symmetry. For C3 symmetry pill completes the specification of fl and for C2 an additional parameter is (8113 - p223). The mean second hyperpolarisability can be defined as which completely specifies y for a spherical system. For lower symmetries additional anisotropy parameters would be required. B. The Bond-additivity Approximation.-The usefulness of a bond-additivity scheme for molecular polarisability a is well established,14 and it is reasonable to suggest an analogous approximation for hyperpolarisabilities. This approx- imation should however be applied with caution since the molecular internal field at a particular bond could cause deviations from additivity.can be written in the form For a molecule with N bonds the approximation for If the ith bond is symmetric about its axis ,8!iv can be expressed in terms of its mean and its anisotropy where la(i) is the cosine of the angle between the axis of the ith bond and the a-direction. The substitution tensor a, equals unity if a = and is zero other- wise.lg To illustrate the usefulness of the bond model consider a series of substituted methanes with tetrahedral bond angles. The only non-zero component of fl for CX is 18123(cx4) = (4/3 d3) 'pc-x (8) where d!c-x is the anisotropy of for a C-X bond. For CY3X which has C3V symmetry 199 Quarterly Reviews 4(Cy2X,) = -(4/343) (APC-X - Al3C-Y) (Pll - P 2 2 J (CY2X2) = (4/343) (ABC-X + APC-Y) (13) (14) The application of the bond-additivity approximation to y defined in eqn.(6) is even more straightforward since y is independent of the bond directions. Hence for a substituted methane CY,X,,, where n can take values from 0 to 4 y(CYnX4-n) = nyC-Y + (4-4yC-x (15) The parameters yc-y and yc-x can be evaluated from y for CY and CX4. Bond-additivity rules could also be stated for the anisotropic components of y. Applications of the bond-additivity approximation are mentioned in sub- sequent sections. C. An Electrostatic Model.-It is easy to understand qualitatively why a molecule should have a second hyperpolarisability y since each of its component atoms has a non-vanishing y. In the case of Is however the isolated atoms do not possess a so that like the dipole moment it must arise from interactions within the molecule.In order to investigate this a simple electrostatic model of a polar diatomic molecule such as HCl is now considered. The model consists of a pair of oppositely charged spheres A (charge - q) and B (charge + q) with centres separated by the internuclear distance R of the molecule which they represent as A 0 illustrated in Figure 3. It is assumed that the Fig. 3 Electrostatic model of a polar diatomic molecule polarisability aA and second hyperpolarisability yA of A are much greater than those of B which are negligible. Such a model might represent HCl but it would be a poor representation of CO where the two atoms are of comparable size. The charge q of B produces an internal field F = - q E 2 along the 3-axis at A.From consideration of the two situations in which a uniform external electric field is applied parallel and perpendicular to the 3-axis the two com- The excited H atom is an exception for spatial degeneracy leads to energy terms in odd powers of the electric field strength; see E. U. Condon and G. H. Shortley ‘Theory of Atomic Spectra’ Cambridge University Press 1935 ch. XVII. 200 Buckingharn and Orr 1 3 ponents of fi emerge as /3333 = yAF and Bsll = -yAF. Hence the first hyper- polarisability of the model arises from the internal field F and the second hyper- polarisabilities of the component atoms. The anisotropy LlP of the model is zero and when reasonable numerical parameters (q = ca. cm. y4 = ca. e.s.u.) are inserted the mean hyperpolarisability p of a molecule such as HC1 should be ca.e.s.u. A negative sign for /?333 implies that an external field acting along the dipole (from - to +) induces a moment (proportional to the square of its strength) which is opposed to the permanent moment. Similarly a negative p311 implies that a field perpendicular to the axis induces a dipole component antiparallel to the permanent moment. 3 Determination of First Hyperpolarisabilities Information concerning the first hyperpolarisability fi of a number of simple molecules has been obtained by a variety of methods. Mean hyperpolarisabilities and other parameters are listed in Table 2. It should be stressed that some of the values are tentative. The methods by which the results have been obtained are discussed below.A. Quantum Mechanical Calculations.-Third-order perturbation theory13 for a molecule in a uniform static electric field gives for the ground state e.s.u. R = ca. - sc < $0 I Pa l$o > < $0 I P/3 19 I > < $ i. I Pyl$o> (16) i+O w - WOY where S denotes a sum of all permutations of the subsequent vector components Wo and Wi are the unperturbed energies of the ground and ith states (I)o and $ *) of the system and pa is the a-component of the electric dipole operator. Matrix elements < $ I pa I$,> have been written in the Dirac bracket notation. Approximate quantum mechanical calculations have been carried out8 for methane using eqn. (16) to yield P i 2 3 = -0.21 x e.s.u. Estimates of #3 and A/3 for a number of polar diatomic molecules have recently been obtained from Hartree-Fock wave B.Second Harmonic Scattering.-When a very intense beam of light such as that produced by a pulsed ruby laser passes through a fluid composed of mole- cules lacking inversion symmetry a small fraction of the light is scattered at a frequency twice that of the incident radiation. The origin of the scattering may be understood in terms of a simple classical treatment (more rigorous theories have been given by Kieli~h,2~ Bersohn Pao and Fri~ch,2~~ and others). If the incident radiation propagating along the X-direction in space is polarised with 22 (a) J. M. O'Hare and R. P. Hurst J. Phys. Chem. 1967 46 2356. a3 S. Kielich Bull. acad. polon. Sci. 1964 12 53. 23 (a) R. Bersohn Y.-H. Pao and H. L. Frisch J . Chem. Phys. 1966,45 3184. 201 Table 2 Values of first hyperpolarisability parameters Other parameters Source 1030/3123 12 -0.21 e.s.u.1030/&2,( N 0.01 e.s.u. 10301p1231 < 0.03 e.s.u. Approximate quantum mechanical calculations Second harmonic scattering of gas Second harmonic scattering of gas - Refractivity virial data - Refractivity virial data - Refractivity virial data - Refractivity virial data - - - - - Kerr effect of dilute solutions Kerr effect of dilute solutions Kerr effect of dilute solutions Kerr effect of dilute solutions Kerr effect of dilute solutions Rayleigh scattering of liquid lO3OdP 21 0 (assumed) Ref. 8 25 26 32 32 32 32 R R R R R 8 R a Values in parentheses are approximate only; * Reference numbers refer to those in the text. R denotes this Review. Buckingham and Orr its electric vector in the 2-direction then eqn.(3) shows that an oscillating dipole moment with frequency 2 0 is produced by the electric field Fo cos wt of the radiation. The amplitude of the a-component of this oscillating dipole is $Pa Zo2. Now according to classical radiation theory the intensity of light scattered from the medium is proportional to the mean square oscillating dipole moment. The depolarisation ratio for second harmonic scattering defined as the ratio of the intensity S of light of frequency 2w scattered in the Y-direction and polarised in the X-direction to the intensity S of that polarised in the 2-direction is where the angular brackets denote an average over all orientations in space of a molecule of the scattering medium provided that there is no correlation between the positions and orientations of the molecules.Evaluation of the average by use of standard methodsg gives po(2~) = (gt2 + 7)/(12t2 + 63) (1 8) where t2 = (5PapyPapy - 3Pa~,8ayy)/2Pa/1pBayy In the case of molecules with symmetries Td and D3h it follows that pO(2w) equals #.20 For molecules with C, symmetry e2 = 9(4m2/25P2. Corresponding relations can be obtained for molecules of other symmetries. Expressions can also be written for the individual scattered intensities S and S, which are proportional to ( 2 ~ ) ~ and to the square of the intensity of the incident radiation. Few second harmonic scattering results have so far been reported; the experi- ment was first described in 1965 by Terhune Maker and Savage,24 who examined several liquids. Recent measurement^^^ of the total scattered intensity (S + SJ indicate that lp1231 for gaseous methane is ca.e.s.u. with an estimated experimental uncertainty of a factor of three. This differs considerably from the approximate theoretical estimate* of -2.1 x e.s.u. (Section 3A). Absolute intensity measurements on carbon tetrachloride as a give Ip1231 = 3.5 x an interaction-free model of the liquid being assumed. However in recent experiments26 no second harmonic scattering could be observed from carbon tetrachloride in the vapour phase from which it has been deduced that [P1231 is less than 3 x e.s.u. and the comparatively strong scattering from the liquid has been attributed to intermolecular interactions. This is sup- ported by the fact that p0(2w) observed for liquid CCI is 0.34 & 0.03 whereas that expected for isolated tetrahedra is 2/3.23a The difference between the observed magnitudes of second harmonic scattering from liquid and vapour is surprisingly large; some of the discrepancy may be due to uncertainties in the measured intensities of the scattered beam and of the focused laser.Second harmonic 24 R. W. Terhune P. D. Maker and C. M. Savage Phys. Rev. Letters 1965,14,681. es P. D. Maker personal communication. P. D. Maker to be published. 203 Quarterly Reviews scattering from water and acetonitrile has also been measured,a4 and depolarisa- tion ratios of 0.12 f 0.01 and 0.10 f 0.01 obtained. Detailed analysis of these results is not justified in view of the uncertainties associated with interpreting harmonic scattering by liquids but the qualitative conclusion that the parameter f 2 is ca.lo- or less may be drawn in both cases. This implies that for H,O lpl is at least an order of magnitude greater than ldfll and [(pll3 - pZ,,>l; likewise for CH3CN The bond-additivity approximation (Section 2B) shows that j6123 for a tetrahedral molecule is proportional to and of the same order as LIP for its bonds. The measurements on gaseous CH and CCI therefore indicate that ldpl for C-H and C-CI bonds is ca. e.s.u. which is orders of magnitude less than lpl for typical bonds (see Section 3F). The scattering results for the three liquids are in conformity with this difference as is the model of Section 2C (for which Op/p = 0). Apart from scattering at frequency 20 inelastic scattering at frequencies (2w f a$ where wt is the frequency of a molecular transition may also occur.A few observations of this second harmonic analogue of the Raman effect have been rep~rted;~,-~~ expressions for the intensities of inelastic scattering have been derived,25 and selection rules deduced from the symmetry properties of p.20 is apparently much greater than [dgl and lpllll. C. Electric Birefringence (the Kerr Effect).-When a uniform electric field is applied to any material it becomes birefringent causing a beam of plane- polarised light passing through it to emerge elliptically polarised. Measurement of the ellipticity for gases or for solutes at high dilution in non-polar solvents leads to a molar Kerr constant mK characteristic of the molecules of the gas or solute. It has been shown4 by classical statistical mechanics that provided a 8 and y are symmetric (19) where aaB(0) represents the static polarisability aaB the optical (high frequency) polarisability pa the permanent molecular dipole moment and a ( = $app) the mean polarisability.For is0 tropically polarisable molecules (e.g. Ar CH, SF& only the temperature-independent term survives (Section 4B). For aniso- tropic molecules with no permanent dipole moment (e.g. N, BF3 C,H,) mK is normally dominated by the second term in T-l. If there is a dipole the final term which is proportional to the difference between the polarisability in the dipole direction and a is generally predominant. However in certain polar molecules the symmetry is such that this difference is small and the term involving 6 is dominant. The series of substituted methanes CH,X, where C-X is an axially symmetric polar group represents such a case.Molar Kerr constants of three methylene dihalides2' and of malononitrile 27 D. Izsak and R. J. W. Le Fevre J . Chem. SOC. (B) 1966 102. 204 Buckingharn and Orr (dicyanomethane),28 measured as solutes at S i t e dilution in carbon tetra- chloride and benzene are much more negative than expected from the Langevin- Born theory of the Kerr effect which neglects hyperpolarisability contributions. A number of rather unattractive explanations of the apparent anomalies have been advanced,2'~~~ but when the contributions of p and y are included experi- mental and calculated results are readily reconciled. The bond-additivity approxi- mationl for a with the assumption that C-H bonds are isotropically polaris- able,29 gives mK(CH&) = (4rN/405)(5y + (1/kT) [10/~/3/3 + 4(a33 - a11)2c-x (3 COS,~ - 3 COS~+ + 1) 1 + (1/k2T2> p2(a33 - all)c-x (3 cosy - 1)) (20) where (as3 - all)C-X is the C-X bond anisotropy and 24 the XCX bond angle.If 2$ is tetrahedral cos24 equals 9 and the term in F2 vanishes. In practice when 24 is not exactly tetrahedral the T2 term is non-zero but still small in comparison to the T-l terms. Values of p are k n ~ w n ~ ' ~ ~ and estimates of (as3 - aIl)c-x for aliphatic C-X bonds have been r e p ~ r t e d . ~ ~ ~ ~ ~ ~ From measurements of y for CH, CCI, and CBr (Section 4) and with use of eqn. (15) values of 1V6y emerge as 15 and 40 e.s.u. for CH2Cl and CH,Br, respec- tively; for CH212 and CH,(CN), 1036y may be ca. 50 and ca. 15 e.s.u. With these data eqn. (20) can be used to calculate @; the results are in Table 3.Varia- tions in /3 are particularly pronounced in the case of CH,Cl and may indicate that fl is more sensitive than 01 to changes in state. Despite the uncertainties these estimates of are the most accurate now available for any molecule. Unfortunately the method is limited to polar molecules in which the F2 term happens to be comparatively small. Another molecule in which this occurs is dimethyl ether and its /3 derived from Kerr effect measurements in carbon tetrachloride,3l is approximately - 5 x e.s.u. Temperature studies of mK for selected gaseous polar molecules would increase the precision of 16 and experiments of this type are currently being organised in Bristol. D. Refractivity Virial Data.-The earliest concerted effort to evaluate /3 experi- mentally involved the measurement of the pressure-dependence of the molar refraction mR of polar which can be written as a virial-type expansion in the molar volume V where n is the refractive index and AB BR CR .. . are the first second third . . . refractivity virial coefficients. The coefficient A R is 4rNa/3 and B is related to the mean contribution to the refraction due to an interacting pair of 28 R. J. W. Le Fhvre B. J. Orr and G. L. D. Ritchie J. Chem. SOC. 1965 2499. 2a R. J. W. Le Fkvre B. J. Orr and G. L. D. Ritchie J . Chem. SOC. (B) 1966 273. so R. J. W. Le Fbvre and B. J. Orr J. Chem. SOC. 1965 5349; J. Chem. SOC. (B) 1966,37. s1 M. J. honey R. J. W. Le F&vre and J. D. Saxby J. Chem. SOC. 1963 2886. s3 A. R. BIythe J. D. Lambert P. J. Petter and H. Spoel Proc. Roy. SOC.1960 A 255,427. 205 Table 3 Calculation of fi from Kerr egects of CH2X2-type niolecules Statea Temp. 1@2mK(e.s.u.) lOl*p(e.s.u.) 1030/3(e.s.u.) CT 25 O - 12.7 1.59 - 6.6 Molecule 2+ CH2C12 112" 1 B 25 " - 21.0 1.59 G 83.5 O - 12.1 1.58 -12 rfI 2 cr 25 O - 17.6 1.40 - 10.6 B 25 -23.6 1.41 - 13.3 CH,Br 112" { -163:4 - 8 + 2 a 25 " - 15.2 1 *08 - 16.0 B 25 O - 14.9 1 -08 - 15.8 CH2I2 114" { CH,(CN)2 114" B 25 " - 72 3.56 a,= B G indicate respectively that measurements were made on solutions in carbon tetrachloride solutions in benzene or the gas. The gas measure- ment is that of H. A. Stuart 'Die Struktur des Freienmolekiils' Springer-Verlag Berlin 1952. Buckingh and Orr molecules. In a polar gas BR can be related approximately to p a 18 and y.ll Measurement of BR however involves finding the small difference of two large quantities so the precision obtained is low.Estimates of 18 obtained in this way are in Table 2 but they may be unreliable to the extent of an order of magnitude. Similar measurementsa for triethylamine yielded 18 = ca. + e.s.u. which is inconsistent with the results in Table 2. Much more precise estimates of BR might be expected from a ‘differential’ experimental method rather than one involving absolute measurement of n as a function of pressure. E. Rayleigh Scattering by Polar Liquids.-When a beam of unpolarised light passes through a fluid a fraction of the light is scattered with the same frequency as the incident radiation. This is known as Rayleigh scattering and its depolarisa- tion ratio is p(o).= A theoretical treatments of depolarisation in polar liquids using the Onsager model has yielded information about fi for chloroform.However the estimates of the anisotropy of a for CHC1 used in this calculation are wrong because of an error which has since been corrected.35 Using correct polarisabilities,36 the depolarisation ratio of liquid CHCl3,% and the Onsager model (p3, - 0.8 p113) is +18 x e.s.u. If dp is vanishingly small as seems likely from Section 3B and the bond-additivity model 18 is approximately +26 x e.s.u. It should be noted that 6 for CHC1 is positive whereas for all the other molecules listed in Table 2 it is negative; this difference in sign may be understood in terms of eqn. (9). A similar theory due to Kielich combining the depolarisation ratio and Rayleigh ratio of a polar liquid has yielded3’ for chloroform p = +203 x e.s.u.and = -267 x This p is an order of magnitude greater than any of the results in Table 2 and the very large is inconsistent with other estimates (Section 3B). F. Comparison of Hyperpolarisability Estimates.-The values of listed in Table 2 for substituted methanes can readily be analysed in terms of the bond- additivity approximation by use of eqn. (9) and (12). Table 4 displays values of (Pc-x - ~ C - H ) deduced in this way. For CH,X and CHX it has been assumed that ~ c - H . is vanishingly small. 4 Determination of Second Hyperpolarisabilities The range of sources from which information about y has been derived is smaller than that for p. Estimates of the mean second hyperpolarisability y for a number of atoms and molecules are listed in Table 5 and the methods employed discussed hereunder.33 D. H. Everett and R. J. Mum Trans. Faraday SOC. 1963 59 2486. 34 R. C. C. k i t e R. S. Moore and S. P. S. Porto J . Chem. Phys. 1964 30 3741. 36 C. G. Le Fkvre and R. J. W. Le Fkvre J . Chem. SOC. 1953 4041 ; R. J. W. Le F&we and B. P. Rao J . Chern. SOC. 1957 3644. 36 N. J. Bridge and A. D. Buckingham Proc. Roy. SOC. 1966 A 295 334. 37 S. Kielich Acta Phys. Polon. 1962 22 299. 207 Quarterly Reviews Table 4 Values of hyperpolarisabilities for polar bonds in substituted methanes Bond Source C-F c-Cl C-Br CH2Br c-I (33212 C-CN CH2(CN)2 I@OP(e.s.u.)Q (-2) (- 8) - 7 f 2 (+ 26) -12 f 2 - 1 6 f 4 - 8 * 2 1@OCBc-x - pC-H)a (-ab (-8Y - 6 f 2 -10 f 2 -14 f 3 - 7 f 2 (- 15y =Values in parentheses are approximate; Walculated on the assumption that pC-H =ca.0. A. Quantum Mechanical Calculations.-For the hydrogen atom y has been evaluated exactly by Sewell,2 giving 1P6y = 0.6714 e.s.u. Recent calculation^^^ employing a perturbation procedure give for helium 1036y = 0.0157 e.s.u. which compares with the experimental result of 0.026 e.s.u. ;38 further calculations currently being carried out at Bristol employ a variational method and are expected to yield an improved theoretical result. Langhoff Lyons and H ~ r s t ~ ~ have calculated y for a variety of S-state atoms and ions by a Hartree-Fock perturbation procedure; their estimates for helium neon and argon are in reasonable agreement with experiment (see Table 5). Further calculations of y for several S-state atoms have been made by C~hen.~O For an atom in its ground state # the static hyperpolarisability y is given by fourth-order per- turbation theory as - c ~ * o l P ~ l * i ~ ~ * f I C L ~ I * O > < ~ O I ~ y l * ~ < * ~ l P 8 l * O ~ iJS0 (W - wo)2(w - W O ) (22) where notation is as for eqn.(16). For most systems y is positive; for two-level systems however the first term in eqn. (22) is zero (provided that the excited state is non-dipolar) so that y is negative. In their early investigation Coulson Maccoll and Sutton3 determined the effect of field strength on the mean polaris- ability of an anisotropic two-level model and found two contributions in F2; these terms are those which determine the Kerr constant. The temperature- independent contribution corresponding to y was found to be negative as expected for a two-level system.Mention is made in Section 5 of calculations of for hydrogenic and alkali-metal atoms at frequencies close to electronic transitions. 3* L. L. Boyle A. D. Buckingham R. L. Disch and D. A. Dunmur J. Chem. Phys. 1966 45 1318. 39 P. W. Langhoff J. D. Lyons and R. P. Hurst Phys. Rev. 1966,148 18. 40 H. D. Cohen J. Chem. Phys. 1965 43 3558; 1966,45 10. 208 Table 5 Values of mean second hyperpolarisabilities Molecule 1P6y(e.s.uJU Sourceb H 0.6714 C He 0.026 G He (0.016) C He (0.026) C Ne (0.055) G Ne (0.062) C Ar (0.73) G Ar (1.16) C Kr (1.6) G Ref. 2 38 38 39 41 39 41 c 39 41 Molecule Xe CH4 CCl CCl CBr CWO,) cs2 Mesitylene 2-6 G Ref. 41 C d e d f 42 42 =Values in parentheses are approximate or preliminary results; bThe symbol C indicates that y was obtained by quantum mechanical calculation; G L and S refer to experimental determination in gas liquid and solution phases respectively; CE.Kuss and H. A. Stuart Phys. Z. 1941 42,95; d C . G. Le FBvre R. J. W. Le Fbvre and D. A. A. S. N. Rao J . Chem. SOC. 1956 708; eC. G. Le F h e and R. J. W. Le Fkvre,J. Chem. SOC. 1953,4041;fC. G. Le FBvre R. J. W. Le FBvre and M. R. Smith J. Chem. SOC. 1958 16. Quarterly Reviews B. Electric Birefringence (the Kerr Effect).-The most important source of information about y has been the Kerr effect. For non-interacting isotropic molecules mK is directly proportional to y (see eqn. 19) The use of a phase-sensitive detection system and a gas laser light source has recently enabled hyperpolarisabilities as small as that of helium to be determined accurately from the Kerr e f f e ~ t .~ ~ ~ ~ Earlier evaluations of y from the Kerr effect have been mentioned by Le F 6 ~ r e . l ~ From eqn. ,(19) it can be seen that y for non-spherical systems is determinable at least in principle from studies of the temperature-dependence of mK. Approximate values of y for carbon disulphide and mesitylene have been obtained in this way;42 the reliability of these estimates has however been q~esti0ned.l~ Temperature studies of the Kerr effect for a number of non-polar gases and vapours currently being undertaken in Bristol should provide additional information about y. Measurements of the Kerr effect of solids in which molecules are not free to be oriented by the applied field might also yield estimates of y ; however suitable experimental results are not available.5 Hyperpolarisabilities at Optical Frequencies A. frequency-dependence of fl and 7.-In the foregoing discussion it has been assumed that hyperpolarisabilities possess symmetry properties and magnitudes which are independent of frequency. Detailed quantum mechanical theory of the frequency-dependence of hyperpolarisabilities has been formulated with use of time-dependent perturbation t h e ~ r y . ~ ~ ~ ~ ~ ~ If the frequencies of all the optical fields impinging on the hyperpolarisable system are much smaller than those of the significant absorption bands 6 and y are approximately symmetric in the Cartesian suffixes a p y . . .I5 This is supported by second-harmonic genera- tion experiments on piezoelectric solids.45 The frequency-dependence of y is further complicated by resonance at the difference frequencies of the optical fields inv01ved.l~ B.Non-linear Optical Effects.-A number of the experiments which have been performed in the new field of non-linear optics involve hyperpolarisabilities ; all the physical effects discussed in Sections 3 and 4 may be classed as non-linear optical e f f e ~ t s . ~ ~ ~ ~ ~ The hyperpolarisability fl gives rise to the following non- linear phenomena second-harmonic generation by solid^,^^,^' second harmonic scattering by f l ~ i d s ~ ~ ~ ~ (Section 3B) optical rectification,15 two-wave mixing,17 and the linear electro-optic (Pockels) effect.15 Non-linear effects in which y is 41 D. A. Dunmur D. Phil. Thesis Oxford 1965. 42 A. D. Buckingham and R. E. Raab J. Chem. SOC. 1957 2341. 43 J. A. Armstrong N.Bloembergen J. Ducuing and P. S. Pershan Phys. Rev. 1962 127 1918. 44 J. F. Ward Rev. Mod. Phys. 1965,37 1. B. J. Orr and J. F. Ward to be published. 45 R. C. Miller Phys. Rev. 1963 131 95. 210 Buckingharn and Orr involved include third harmonic generation ,17,46 electric-field-induced second harmonic the Kerr effect4 (Section 3C) intensity-dependent refractive inde~,4~9~' and three-wave mixing.46 For a number of these phenomena only macroscopic susceptibilities characteristic of the bulk medium have been evaluated; their interpretation in terms of molecular hyperpolarisabilities may be complicated by co-operative effects. The frequency-dependence of fl and y can conveniently be represented in a n ~ t a t i o n ~ ~ ~ in which a dipole oscillating at frequency w1 induced by electric fields F(w *) of frequency w is m,(wl) = aap(-W1,wl) F,&w,> + Kpapy(-wl,w2,w3) F7(w3) + hyapy &(-wl,w2,w3,w4) Fp(w'2) Fy(U3) F&(04) (24) The sign attached to a frequency is positive if the appropriate photon is absorbed and negative if it is emitted.The sums of the frequencies attached in parentheses to the tensors are zero so in &-wl,w2,w3) and y(-w1,w2,w3,wq) it is necessary that w1 equals (w2 + w3) and (w2 + w3 + w4) respectively. The numbers K and h take particular values for each non-linear process according to the number of frequencies which are equal or zero.44 For example K = 1/4 if w2 = w3 and = 1/24 if w2 = w3 = w4; K = 1/2 and h = 1/6 if all frequencies are zero. Thus the tensor for second harmonic generation is /3apy(-2w,w,w) which is symmetric in the suffixes Py.Those appropriate to the Kerr effect are asp( -w,w) paPy( -w,w,O) and yaBr 6( -w,w,O,O) which are symmetric in a/3; y is also symmetric in 78. Recently calculations have been made of y( -w,w,O,O) for hydrogenic atoms4s and alkali-metal v a p o u r ~ ~ ~ ~ ~ at frequencies w close to electronic absorption bands using formulae derived by double perturbation theory.61 The results for sodium vapour in the vicinity of the d i n e s are to be in satisfactory agreement with the early experiments of Kopfermann and Ladenburg.62 6 Conclusion This Review has dealt with two new electronic properties of molecules fl and y. As fundamental constants determining the distortion of a molecule by a strong electric field they are intimately connected with intermolecular forces electronic interactions within molecules and chemical rea~tivity.~,~~ But quite apart from these applications fl and y are of interest in themselves for they provide new information about the electronic structure of the molecule; presumably they are particularly sensitive t o the precise nature of the outer reaches of electronic wave-functions.46 P. D. Maker and R. W. Terhune Phys. Rev. 1965 137,801. 47 A. D. Buckingham Proc. Phys. SOC. 1956 B 69 344; G. Mayer and F. Gires Compt. rend. 1964 258 2039; M. Paillette Compt. rend. 1966 262 264. 48 L. L. Boyle and C. A. Coulson Proc. Phys. SOC. 1966 89 499. 49 L. L. Boyle and C. A. Coulson Mol. Phys. 1966 11 165. 50 M. P. Bogaard A. D. Buckingham and B. J. Orr Mol. Phys. 1967 in the press. 51 A. D. Buckingham Proc. Roy. SOC. 1962 A 267,271.52 H. Kopfermann and R. Ladenburg Ann. Physik 1925,78 659. 21 1 Quarterly Reviews Much of the information now available about and y should find application in assessing hyperpolarisability contributions to the Kerr effect thereby increas- ing the reliability of a well-established method of structural ana1y~is.l~ The constants fl and y are also at the heart of some of the important new non-linear optical effects; it has been suggested that optical harmonic generation could be used in characterising m a ~ r ~ m ~ l e ~ ~ l e ~ ~ ~ ~ ~ ~ ~ and in studying molecular inter- actions in liquid^.^^^,^^ The authors gratefully acknowledge valuable communications with Dr. R. P. Hurst Professor R. J. W. Le Fkvre F.R.S. Dr. P. D. Maker and Dr. R. W. Terhune and the award (to B.J.O.) of an 1851 Exhibition Overseas Scholarship. 53 R. Bersohn J . Ainer. Chem. SOC. 1964 86 3505. 212

ISSN:0009-2681    

DOI:10.1039/QR9672100195

出版商:RSC    

年代:1967

数据来源: RSC

摘要:

Photochemical Behaviour of Transition-metal Complexes By E. L. Wehry DEPARTMENT OF CHEMISTRY INDIANA UNIVERSITY BLOOMINGTON INDIANA U . S . A . There has recently developed a surge of interest in photochemical research especially with regard to organic molecules. The photoresponsive behaviour of inorganic species particularly co-ordination compounds has received consider- ably less attention. It is hoped that the present Review will stimulate more extensive quantitative work in this area. Acquaintance with the basic vocabulary of photochemistry1y2 will be assumed. 1 Electronic Spectra of Co-ordination Compounds3~* Before one can commence to understand the photochemical behaviour of a species he must have available detailed information regarding the nature of its electronic energy states.Electronic absorption spectra of metal complexes are frequently complicated and spectral assignments difficult to establish. A co- ordination compound may exhibit ultraviolet and visible absorption due to any (or all) of the following types of electronic transitions A. Transitions Involving Electrons of the Central Metal Ion.-Of particular importance are transitions within the partially-filled d shells of transition-metal ions. The five d orbitals of an isolated metal ion are degenerate. However the electrostatic crystal-field theory predicts that complex formation will partially remove the d-orbital degeneracy so transitions within the d levels become possible. For structures in which there is an inversion centre d+d transitions are forbidden by the Laporte rule4s5 and can occur only because of transitory ‘loss’ of the inversion centre via molecular vibrations.Consequently d-+d transitions are invariably weak (~<200); they generally occur in the visible or near-infrared regions and are responsible for the colours of many complexes. The d+d absorp- tions are commonly referred to as ‘ligand-field’ bands. E. J. Bowen ‘The Chemical Aspects of Light’ Oxford Univ. Press London 1946 ch. 6. J. G. Calvert and J. N. Pitts jun. ‘Photochemistry’ Wiley New York 1966. C. K. Jarrgensen ‘Absorption Spectra and Bonding in Complexes’ Pergamon London 1962; B. N. Figgis ‘Introduction to Ligand Fields’ Interscience London 1966. T. M. D u n in ‘Modern Co-ordination Chemistry’ ed. J. Lewis and R. G. Wilkins Inter- science London 1960. G. Herzberg ‘Spectra of Diatomic Molecules’ 2nd edn.Van Nostrand New York 1950 pp. 28-29. 21 3 Quarterly Reviews B. Transitions Involving Promotion of Ligand Electrons.-Co-ordination compounds frequently exhibit electronic absorption which is characteristic of the uncomplexed ligand. Significant ultraviolet or visible absorption occurs only for ligands with large r-electron systems especially aromatic and heteroaromatic molecules. Allowed transitions in aromatic molecules are often extremely intense ( E = ca lo5). Energies and molar absorptivities of ligand transitions will usually be altered by complex-formation with a transition-metal ion although the magnitude of the perturbation is frequently rather small. Detailed treatments of electronic transitions in organic molecules have been given by MurrelP and Jaff6 and Orchin.’ Unfortunately the photochemical behaviour of complexes of highly conjugated ligands has received virtually no attention.It is therefore impossible to discuss the photochemical significance of ligand-electron transi- tions; accordingly they will not be considered further here. C. Charge-transfer Spe~tra.~~*-Irradiation of a co-ordination compound with light of appropriate frequency may induce transfer of an electron from ligand to metal e.g. (1). Ligand-.metal charge-transfer processes occur quite frequently [Fe3+(H20),Br]s+ -+ [Fe2+(H20)5Br]2+* [Fe2+(H20),l2+ + Br (1) because most strong complexing ligands are Lewis bases. Occasionally metal- ligand electron-transfer absorption may occur especially in metal carbonyls and nitrosyls. Charge-transfer absorption bands nearly always occur at higher energies than d-td transitions and are frequently found in the blue or ultraviolet regions.Charge-transfer absorption followed by chemical steps generally results in oxidation of one or more Iigand molecules with simultaneous reduction of the metal ion. hv 2 Energy Dissipation in Electronically Excited Metal Complexes Electronically excited metal complexes are highly energetic species and they will invariably seek to divest themselves of the excess of energy as rapidly as possible. There are several pathways by which excitation energy can be dissipated includ- ing those in (A)+) following. A. Internal Conversion.-An electronically excited complex may return directly to its ground state without emission of a photon converting the excitation energy into heat.This radiationless process called ‘internal conver~ion’,~ is not well understood and its efficiency can be measured only with difficulty and by indirect means. Internal conversion from the first excited state to the ground J. N. Murrell ‘The Theory of the Electronic Spectra of Organic Molecules’ Methuen London 1963. 7 H. H. Jaff6 and M. Orchin ‘Theory and Applications of Ultraviolet Spectroscopy’ Wiley New York 1962. L. E. Orgel Quart. Rev. 1954 8 422. M. Kasha Discuss. Faraday SOC. 1950,9,14; Radiation Research 1960 Suppl. 2 243; M. Gouterman and P. Seybold Chem. Rev. 1965,65,413. 214 Wehry state is believed to be a significant process in co-ordination compounds.1° Of greater importance is the fact that internal conversion between upper excited states and the first excited state of a given multiplicity is invariably much more rapid than any competing process.(‘Multiplicity’ = 2s + 1 where S is the total molecular spin quantum number.) Hence whenever the second (or higher) excited state of a given multiplicity is populated by absorption rapid internal conversion occurs to the lowest excited state of that multiplicity from which virtually all photochemical processes rig in ate.^ B. Intersystem Crossing.-Radiative transitions between electronic states of different multiplicity are forbidden (highly improbable). There is however a form of radiationless transition which can lead to population of spin-forbidden states. This radiationless process is called ‘intersystem cro~sing’;~ it is a very important phenomenon in transition-metal complexes because the rate constant for intersystem crossing is greatly enhanced in paramagnetic species.Since excited states populated by intersystem crossing are spin-forbidden they are characterised by relatively long radiative lifetimes. For this reason they are often termed ‘metastable’ states; in many cases their lifetimes are sufFiciently long to enable excited molecules to engage in chemical reactions which are only modera- tely rapid. In contrast spin-allowed excited states usually have lifetimes no greater than lop8 sec. so only very rapid chemical reactions can compete with other decay mechanisms. Consequently intersystem crossing is an important precursor to many photochemical processes. C. Photoluminescence.11*12-An electronically excited metal complex can return to the ground state by emission of a photon.If the multiplicities of ground and excited states are equal the emission is termed ‘fluorescence’; if different ‘phosphorescence’. Fluorescence is relatively uncommon in transition-metal complexes. Particularly if the metal ion is paramagnetic intersystem crossing will lead to significant population of one or more spin-forbidden excited states which can subsequently decay to the ground state via phosphorescence. Most luminescent complexes involve highly conjugated ligands and the transitions responsible for emission frequently are slightly perturbed ligand 7P-v transi- tions. Occasionally d+dl3,l4 or metal-ligand charge-transfer15 photolumines- cence is observed in transition-metal complexes. D. Intermolecular Energy Transfer.-An electronically-excited transition-metal complex may transfer its energy to other molecules in solution by any of several lo J.B. Allison and R. S. Becker J. Chcrn. Phys. 1960 32 1410. l1 G. B. Porter and H. L. Schlafer Ber. Bunsen Gesellschaft Phys Chcm. 1964 68 316. l2 W. E. Ohnesorge in ‘Fluorescence and Phosphorescence Analysis’ ed. D. M. Hercules Interscience New York 1966 ch. 4. lS K. De Armond and L. S. Forster Spectrochim. Acta 1963 19 1393 1403 1687. l4 K. K. Chatterjee and L. S. Forster Spectrochim. Acta 1964 20 1603. l6 J. P. Paris and W. W. Brandt J. Amer. Chem. SOC. 1959 81 5001. But see also G. A. Crosby W. G. Perkins and D. M. Klassen J. Chem. Pliys. 1965 43 1498. 21 5 QuarterZy Reviews rather complicated processes.l6 This is believed to constitute the fundamental r6le of the porphyrin complexes in photosynthesis; however the significance of intermolecular energy-transfer processes in simpler co-ordination compounds is not very clear.E. Photochemical Reaction.-If the above processes are not exceedingly rapid it is possible for an excited complex to engage in chemical reaction. Whenever photochemical reactions occur they tend to compete with one or more of the other decay mechanisms adding enormously to the complexity of the problem. Consequently in dealing with photochemical reactions of co-ordination com- pounds it is necessary to make frequent reference to the other decay processes. 3 General Survey of Co-ordination Compound Photochemistry h4any co-ordination compounds are characterised by a high degree of structural complexity and the number of photochemical reactions in which they might engage may appear virtually limitless.In fact however virtually all reported photoreactions of complexes fall into one of three categories (A) (B) or (C). A. Photochemical Oxidation-Reduction Reactions.-In a large number of photochemical processes the overall result is reduction of the metal ion with concomitant oxidation of one or more ligands; examples are (2) and (3). hv Fe3+[(C204)2-]3 -+ Fe2+[CzO4)”I2 + C204- The primary photochemical process is thought to be homolytic fission of a metal-ligand bond1’ (see p. 218). Such processes are probable whenever the absorption spectrum of a complex includes at least one moderately intense ligand-to-metal charge-transfer band. The oxidised ligands are usually released from the complex subsequent to charge transfer; they are often unstable and engage in secondary chemical steps.Consequently the stoicheiometry of photo- chemical redox reactions is frequently complicated and the primary process difficult to elucidate. B. Photosubstitution Processes.-Excitation of a co-ordination compound may lead to ligand-exchange reactions such as (4) and (5). hv Cr(CO) + py -+ Cr(CO),(py) + CO (‘py’ = pyridine) (4) 16 F. Wilkinson Adv. Photochem. 1964 3 241. 17 A. W. Adamson and A. H. Sporer J. Amer. Chem. SOC. 1958,80,3865. 21 6 Wehry Photochemical ligand-exchange reactions are particularly common for metal carbonyls and their derivatives; invariably the entering ligand is a strong electron donor.18 Excepting metal carbonyls most co-ordination-compound photo- chemistry has been performed in aqueous solution; consequently a large majority of reported ligand-exchange processes have been aquations.Photochemical aquation seems to occur only for complexes susceptible to thermal (‘dark’) aquation and the principal effect of light frequently appears to be acceleration of the thermal reaction. Experimental information at present available is in- sufficient to permit correlation of the degree of photochemical enhancement of substitution processes with metal-ion ligand-field parameters. C. Photochemical homerisations and Racemisations.-A number of optically active complexes exhibit a tendency to racemise photochemically; most cases studied have involved octahedral complexes of bidentate ligands. Photochemical isomerisation generally is an important process only for complexes which undergo thermal isomerisation.In addition most complexes exhibiting photo- isomerisation or photoracemisation also undergo photoaquation reactions. These facts strongly suggest that the primary processes for photochemical aquations and isomerisations are identical or very similar. Primary processes and subsequent chemical steps in the three major types of photoreaction can best be described by means of specific examples. Three important examples will be considered the photochemistry of Complexes of tripositive chromium tripositive cobalt and the oxalate ion. Much of the experimental work in co-ordination-compound photochemistry has concerned one or more of these three systems. In addition the photochemical behaviour of metal carbonyls and the various complexes in the photosynthetic chain have been extensively investigated.The photoreactions of metal carbonyls have recently been reviewed,l* and several general discussions of primary photo- synthetic processes are available,19 so they will not be considered further here. 4 Photochemistry of Cobalt(m) Complexes A wide variety of cobalt(m) complexes undergo photochemical reactions ; before describing them it is necessary to consider briefly the electronic absorption spectra of tripositive cobalt complexes. There are generally three principal regions of absorption; two of these are ligand-field (d-td) bands one located in the range. 475-625 mp and the second located between 345 and 415 mp.20 A third region of absorption in the vicinity of 230 mp has been assignedz1 as a ligand-metal charge-transfer transition. Most photosensitive cobalt(m) complexes undergo both oxidation-reduction and ligand-exchange reactions.Adamson and his co-workers have examined the lS W. Strohmeier Angew. Chem. Internat. Edn. 1964 3 730. l9 Particularly useful are J. B. Thomas ‘Primary Photoprocesses in Biology’ North-Holland Amsterdam 1965 and M. D. Kamen ‘Primary Processes in Photosynthesis’ Academic Press New York 1963. 2o Ref. 4 p. 291. A. Linhard and M. Weigel Z . anorg. Chem. 1951 266,49. 217 Quarterly Reviews photoresponsive behaviour of Co(NH3),X2+ complexes (X = C1- Br I- SCN-) in detai1.l'~~~ As one would expect the relative importance of photoredox to photoaquation becomes greater with increasing ease of oxidation of the ligand X. Also the total photosensitivity is significantly greater when X is easily oxidisable.This fact suggests that the primary process is the same in both photoreduction and photoaquation involving homolytic cleavage of the Co-X bond17,21,22 as in (6) where d represents excess of energy which is presumably dissipated vibrationally coiicurrently with recoil of X from the complex. The detailed course of the photochemical reaction is then dependent upon the fate of the intermediate [Co2+(NH3)5,Xl2+. The pentamminecobalt(u) complex and X may tend to diffuse away from one another but such diffusion will be inhibited by collisions of the two species with the surrounding sheath of solvent molecules; this is the 'Franck-Rabinowitch cage effect'.23 Hence one reaction of the inter- mediate is simple recombination with X yielding the original complex. The cage effect should increase in significance as the d term of reaction (6) becomes smaller because the collision of the pentamminecobalt(I1) intermediate and X with the solvent cage will be less energetic if d is small.The d term generally decreases with decreasing energy (it?. increasing wavelength) of absorbed light. Therefore if the cage effect is important in the photoreactions of cobalt(II1) complexes one would predict that quantum yields would tend to decrease with increasing wavelength of incident light provided of course that the same electronic transition were excited in all cases. Indeed Adamson22 noted that the total quantum yield for photolysis of a number of tripositive cobalt complexes does decrease with increasing wavelength. This implies that the cage effect is operative in these systems and provides support for Adamson's postulate that metal-ligand homolytic bond cleavage is the primary process.If the products of homolytic cleavage do manage to diffuse from each other reduction of the metal ion will be the net result. The fate of the pentammine- cobalt(@ complex is dependent solely upon thermal reactions; in the absence of extraneous oxidising agents a stable hexaco-ordinated cobalt(@ complex should be the final product. Thus the Adamson mechanism can quite easily rationalise photoreduction of cobalt(& complexes. If the homolytic cleavage mechanism is operative significant steady-state concentrations of X atoms should be present during photolysis. That Br and I atoms are actually formed in appreciable quantities upon photoreduction of Co(NH3),Br2+ and Co(NH3),12+ respectively is indicated by observation of halogen-atom transients subsequent to flash photolysis of the complexes.24 For the homolytic cleavage mechanism to be of value it must rationalise the fact that photoreducible cobalt(m) complexes also undergo photosubstitution 22 A.W. Adamson Discuss. Faraday SOC. 1960 29 163. 23 J. Franck and E. Rabinowitch Trans. Faraday SOC. 1934,30 120. 24 S. A. Penkett and A. W. Adamson J. Amer. Chem. SOC. 1965 87,2514. 218 reactions. Considering once again the c0(NH3)5x2+ complexes we can conceive that the intermediate of reaction (6) could undergo thermd oxidation-reduction if the solvent cage prevented the fragments from separating as in (7). Clearly [CO~+(NH~)~,XI~+ -+ [Co2+(NH,),l2+ + X [C03+(NH3),]3+ + x - t I (7) process (7) would be more probable for X = C1 than for X = Br or I.What is important to note is that the product of reaction (7) is a pentamminecobalt(m) complex which can co-ordinate with a molecule of solvent water to yield the aquation product of the original complex. The aquation itself is thermal so the mechanism implies that significant photoaquation will occur only if the complex exhibits some propensity to undergo dark aquation. This has indeed been the case in most systems examined so far and may indicate that the homo- lytic cleavage mechanism is a reasonably accurate description. An impression of the complexity of events which can follow the postulated photochemical homolytic cleavage may be gained by considering the photolysis of Co(en),s+ (‘en’ = ethylenediamine). Photolysis products include Co(H,0),2+ ammonia formaldehyde and eth~lenediamine.~~ Presumably the primary process is homolytic cleavage of one Co-N bond.The ‘semico-ordinated’ ethylenedia- mine molecule may reco-ordinate producing the original complex (or its enantiomer if several similar intramolecular steps occur). Alternatively it is possible that the partially-displaced ethylenediamine may fragment yielding a co-ordinated NH,CH radical and a methyliminiwn ion which is unstable in aqueous s0lution,2~ as in (8)-(11). This series of reactions rationalises the + CH = NH + H2O + CH2O + NH,+ NH,+ + H,O + NH3 + H30f etc. (1 1) production of formaldehyde and ammonia in the photolysis. It remains to ascertain the fate of the cobalt(n) complex which is the product of reaction (9). That complex would surely be unstable in aqueous solution and one might expect it to dissociate yielding Co(en),(H,O)Z+ and methylamine.However methy- lamine is not detected as one of the products of the photoreaction possibly D. Klein and C. W. Moeller Inorg. Chem. 1965 4 394; W. C. Taylor jun. and C. W. Moeller ibid. 1965 4 398. 219 QuarterlyRevie ws because of polymerisation reactions with formaldehyde. To further complicate the system cobalt(I1) complexes of ethylenediarnine are not stable so Co(en),(lH,0),2+ undergoes thermal dissociation to yield hexa-aquocobalt(u) and ethylenediamine. Under such circumstances identification of all photoproducts becomes a triumph of analytical chemistry and postulation of a mechanism one of ingenuity. A principal shortcoming of the homolytic cleavage argument is that it provides little insight concerning the nature of excited states involved in the photoreac- tions; indeed Adamson ‘assigns no r61e to the fact that the nature of the excited state is different at different wavelengths’., A major difficulty in this respect is serious overlap between charge-transfer and ligand-field absorption bands in the spectra of many cobalt(II1) complexes.It is evident that a high redox quantum yield would be expected for irradiation in the charge-transfer absorption region. When cobalt(II1) complexes are photolysed with the 254 mp mercury line corresponding to ‘pure’ charge-transfer absorption only photoreduction is observed; there is no evidence for aquation.26 If homolytic fission is a common primary process for both reduction and substitution reactions it is not obvious why charge-transfer excitation should lead solely to reduction.The situation becomes even less well-defined when long-wavelength irradiation is performed owing to spectral overlap. The clearest evidence so far is that of Klein and M~eller,,~ who studied photoreduction of trisethylenediaminecobalt(u1) ; they noted a redox quantum yield of 0-070 for 254 mp (charge-transfer) excitation but a yield of only 0.0005 upon 366 mp (ligand-field) irradiation. Perhaps the non-zero reduction yield for ligand-field excitation results from a tail in the charge-transfer absorption overlapping the d-td bands. It appears then that only redox occurs under the influence of charge-transfer excitation while solely (or almost exclusively) aquation results from ligand-field absorption.This is not necessarily inconsistent with the homolytic cleavage picture but in its present state of development the mechanism does not predict or even convincingly rationalise the observation. Attempts have been made to elucidate the nature of excited states involved in photoreactions of cobalt(II1) complexes by investigating the effect of pH upon quantum yields.26 The variation of yield with pH is complicated but the results seem to indicate that cobalt(rI1) complexes containing amine ligands undergo excited-state protolytic dissociation before reaction as in (12). Cleavage of a li v Co(en)z+ + [Co(en)?+]* + [CO(~~),(HNCH,-CH,.NH,)~+]* + H+ (12) Co-N bond and subsequent chemical steps as in reactions (8)-(11) are pre- sumed to follow excited-state dissociation. It is inferred that homolytic cleavage must occur while the complex is in a spin-forbidden (‘metastable’) excited state since the lifetime of allowed charge-transfer states might be too brief to allow both protolysis and bond cleavage to occur.26 Intersystem crossing is therefore 26 J.F. Endicott and M. Z. Hoffman J. Amer. Chem. SOC. 1965 87 3348. 220 Wehry assumed to play a r6le in the photolyses. While this may well be true present evidence is unconvincing. It is known that excited-state proton transfers may be exceedingly rapid with second-order rate constants on the order of 1O1O 1.mole-l sec.-l not at all uncommon.27 Consequently it is not inconceivable that both proton transfer and homolytic metal-ligand bond rupture could occur while the complex is in a spin-allowed excited state.The hypothesis of excited-state protolysis is itself quite reasonable for it is known that many uncomplexed amines are stronger acids when electronically excited by as much as six to eight pK units compared with the same compounds in their ground states.28 Hence such processes may well play an important r6le in the photochemistry of amine complexes of cobalt(m) and other transition-metal ions. It can be concluded that although a considerable body of chemical informa- tion is now available excited-state processes in the photochemistry of complexes of tripositive cobalt are not well understood. Flash-photolysis experiments should aid in establishing the importance of intersystem crossing and changes in medium (pH solvent polarity solvent viscosity) may lead to fuller understanding of the chemical pathways of photoreduction and photoaquation of cobalt(m) complexes.For example it has recently been demonstrated that cobalt(rI1) complexes can react with photochemically generated hydrogen atoms and solvated electrons.29 Experiments of this nature may assist in unravelling the photo- chemical behaviour of the complexes themselves. 5 Photochemical Behaviour of Chromium(m) Complexes Many complexes of tripositive chromium undergo photochemical substitution and isomerisation reactions but unlike their cobalt(u1) analogues do not generally exhibit pho t oredox. Quan t um yields for phot osubs ti tu tion processes are appre- ciably greater for chromium(n1) complexes than for those of cobalt(n1); this can be rationalised by crystal-field arguments.30 The resistance of chromium(II1) complexes to photoreduction is easily understood since chromium(n) is an extremely strong reducing agent which is immediately reoxidised to chromium(1rr) in aqueous media.Electronic absorption spectra of chromium(n1) complexes are rather com- plicated and the d-+d transitions have been subjected to considerable study. Ligand-field states relevant to consideration of photochemical processes are represented in Figure 1. The ground state is a quartet and three quartet+quartet transitions can usually be observed in the vicinity of 250 400 and 580 mp31 The multiplicity-forbidden *A,,j2E transition occurs in the vicinity of 650 mp and is quite weak (E = ca. 1). 27 A. Weller Progr. Reaction Kinetics 1961 1 189. 28 E. L. Wehry and L. B. Rogers in ‘Fluorescence and Phosphorescence Analysis’ ed.D. M. Hercules Interscience New York 1966 pp. 125-135. See also G. Jackson and G. Porter Proc. Roy. Soc. 1961 A 260 13. 29 J. F. Endicott and M. Z. Hoffman Abstracts 151st American Chemical Society Meeting Pittsburgh Pennsylvania March 1966 No. ~ 8 8 ; J . Phys. Chem. 1966 70 3389. 30 H. L. Schlafer J. Phys. Chem. 1965 69 2201. 31 H. L. Schlafer ,Z. phys. Chem. (Frankfurt) 1956,11 65; Z. Elektrochem. 1960,64,887. 221 Quarterly Reviews 1 1 Ivtersystem crossing Fluorescence - Phosphorescence Fig. 1 Schematic diagram of the ligand-field states of photochemical significance in chromium (m) complexes. One of the most thoroughly studied photosubstitutions of chromium(1n) is the exchange (13) of co-ordinated water with solvent enriched with oxygen-18. hv Cr(H,O),S+ + 6H,180 -+ Cr(H,1*0),3+ + 6H20 Low quantum yields of the order of 0.01 are noted; they are sensibly independ- ent of wavelength in the quartet-quartet absorption region (260-570 m ~ ) .~ ~ The low quantum yields are certainly not due to cage effects for the solvent cage is one of the reactants. Calculations indicate that the lifetimes of the second and third excited quartet states of chromium(m) are of the order of lO-llsec. so any molecules excited to these levels will undergo internal conversion to the lowest excited quartet. Fluorescence is generally a very minor means of energy dissipa- tion in most excited chromium(m) complexes. Hence the two processes most likely to occur from the lowest excited quartet are internal conversion to the ground state or intersystem crossing to the lowest doublet (spin-forbidden) excited state.That intersystem crossing is a significant process is indicated by numerous reports of 2E,,-+4A2 phosphorescence observed from complexes of tripositive c h r o ~ n i u m . ~ ~ - ~ ~ ~ ~ The doublet is relatively long-lived (ca. sec.) so chemical reaction may well compete with intersystem crossing and phos- phorescence. In contrast the lifetime of the lowest excited quartet is certainly no greater than lo-' sec. which is probably too brief to enable reaction to compete effectively with other decay processes. It therefore has been p o ~ t u l a t e d ~ ~ ~ ~ ~ that 32 R. A. Plane and J. P. Hunt J. Amer. Chem. SOC. 1957,79 3343. 33 G. B. Porter and H. L. Schlafer 2. phys. Chem. (Frankfurt) 1963 37 109; 1963 38 227; 1964,40,280. 34 M.R. Edelson and R. A. Plane J. Phys. Chem. 1959 63 327. 222 Wehry hv OW3):+ -+ 4[Cr(NH3),3-t-l* ---+ S[Cr(NH3)63+]* intersystem crossing t +I + L the photoreactions proceed through the lowest doublet as in (14). The double arrow signifies that more than one chemical step may follow the primary photo- chemical act. It is presumed that photochemical quantum yields are limited by the efficiency of intersystem crossing rather than by cage effects or related factors. It is very tempting to ascribe special photochemical significance to the lowest doublet state but supporting evidence is far from conclusive. One would expect the difference in energy between the first excited quartet and lowest doublet state to be a function of the field strengths of the ligands and to vary from one chromium(rz1) complex to another.The smaller this difference the greater the probability that thermal excitation will raise some doublet molecules back into the first excited quartet. Consequently if the doublet but not the quartet is photochemically active substitution quantum yields should tend to decrease as the energy difference between the 4T2s and 2Eg states becomes smaller. Edelson and PlaneM have presented data consistent with this prediction. However Adamson% has pointed out that such comparisons involving quantum yields for reactions of different complexes are difficult to interpret unequivocally and they cannot by themselves be regarded as demonstrating involvement of the doublet state. Direct evidence concerning the r6le of the doublet state could be obtained by measuring quantum yields for direct excitation to the doublet state comparing them with yields in the quartet region.This constitutes a rather formidable undertaking. Quartet-doublet absorption is forbidden; it can be observed but E is seldom greater than unity. Further the transition occurs at a wavelength of ca. 650 mp in which region both chemical actinometers and physical methods for measuring light intensities tend to be unreliable. Edelson and Plane have reported results for such an experiment.= The stepwise photoaquation of was reported to proceed with a quantum yield of approximately 0.3 in the quartet region but with a yield approaching unity in the doublet absorp- tion band. If reliable these data provide highly convincing evidence for direct doublet-state involvement in the photoaquation.Recently however Wegner and Adamson3' have restudied the Cr(NH3),3+ photoaquation; at 452 mp (quartet absorption) they obtained an aquation yield of 0-26 while at 652 mp (doublet excitation) a yield of 0.29 was observed. In these later measurements the doublet yield was equal within experimental error to the quartet yield; thus they cast serious doubt on the importance of intersystem crossing in the photo- chemistry of chromium(m) complexes. More extensive quantum-yield data are needed to settle the question. 35 A. W. Adamson J. Inorg. Nuclear Chem. 1960 13 275. a6 M. R. Edelson and R. A. Plane Inorg. Chem. 1964 3,231. 37 E. E. Wegner and A. W. Adamson J. Amer. Chem. SOC. 1966,88,394. 223 Quarterly Reviews At this point it is relevant to ask whether or not the photoreactions involve electronically excited states at all especially since most (but not all) photo- chemistry of tripositive chromium complexes involves acceleration of known thermal reactions.It is not inconceivable that the excited electronic states act merely as convenient vehicles for storage of energy until such time as collision of a molecule of complex with the solvent cage releases the energy and makes it available for reaction.31 Presumably in this situation the actual reactions would involve vibrationally excited complexes in their ground electronic states. It would however seem rather likely that in many cases the energy would be conducted away from the complex by motion of solvent molecules before re- action could take place. Furthermore A d a m ~ o n ~ ~ has recently reported on several chromium(m) complexes in which the thermal and photochemical reactions are markedly different.The photochemistry of the isomeric Cr(en),(OH),+ complexes is stereospeczjic ; the cis isomer undergoes predomi- nantly aquation while the trans form favours photoisomerisation. In contrast the thermal reaction chemistry of the isomeric complexes is not at all stereospecific and involves almost exclusively isomerisation. Apparent activation energies for the thermal and photochemical reactions are quite different. It is difficult to understand these experimental facts without assuming that the photochemical reactions proceed through electronically excited states. Adamson’s results also seem to indicate that Cr(en),(OH)$ complexes exhibit different photoresponses depending on whether the quartet or doublet states are excited.The photo- isomerisation and photoaquation reactions do not seem to involve a common precursor so the mechanistic details of the system are likely to be highly com- plicated. Further experiments (especially flash photolysis) are needed before detailed pathways can be charted but it appears safe to assert that electronically excited states do in fact take part in the reactions. Two possible mechanisms for the chemical steps in photosubstitutions of chromium(m) complexes have been suggested by Adam~on.~~ One postulated mechanism is essentially equivalent to reaction (14) except that the doublet state is not specifically implicated as an intermediate. The electronically excited complex is assumed to undergo a nucleophilic substitution; no attempt is made to specify whether the actual process is ,S”1 5”2 or a variant thereof.Indeed such labels are probably inappropriate when applied to photochemical pro- cesses; a more adequate description would involve a continuum of excited states rather than that described by one of the common ‘SN’ labels. A second chemical mechanism apparently favoured by Adamson assumes that ion-pair formation precedes excitation and that the attacking species displaces one of the original hv Cr(H20)63+ + x- [Cr(H20)63t- x-] [cr(H20)63+ x-]* (1 5 ) KIP Cr(H20),X2+ + H20 +- +- I ssA. W. Adamson in ‘Mechanisms of Inorganic Reactions’ ed. R. F. Gould American Chemical Society Washington 1965 pp. 237-248; J . Phys. Chem. 1967 71 798. 224 Wehry ligands in the electronically excited ion as in (15).Several steps un- doubtedly ensue between excitation of the ion-pair and appearance of final products. The ion-pair precursor mechanism has two attractive aspects. First by this mechanism the quantum yield of the photoreaction must be dependent upon the ion-pair formation constant; if KIP is small the quantum yield for substitution cannot be large. Quantum yields for photosubstitution of chromium(m) complexes are frequently quite small even when the solvent is the attacking ligand in which case the cage effect obviously cannot be yield-limiting. In the above view the factor which defines the maximum possible quantum yield is KIP. A second advantage of the ion-pair mechanism is that formation and dissociation of ion pairs is often extremely rapid and could occur even in short-lived excited quartet states of chromium(II1) complexes.30 It is therefore unnecessary to ascribe special significance to excited doublet states if the ion-pair mechanism is assumed.It must fairly be said that there is little evidence to sub- stantiate the importance of an ion-pair intermediate. One would for example expect that a decrease in the dielectric constant of the solvent would increase the extent of ion-pairing and hence the quantum yield. No report of this experi- mental test has appeared. Undoubtedly such data would be very difficult to interpret since it is impossible to alter the dielectric constant without simul- taneously modifying other solvent properties such as cage structure. One significant by-product of photochemical research involving tripositive chromium complexes is the development of promising new chemical actino- meters.Present actinometer systems (potassium ferrioxalate and uranyl oxalate) tend to be unreliable at wavelengths greater than 500 mp. It appears that two complex salts KCr(NH,),(CNS) and Cr(urea),CI, can serve as useful actino- meters at wavelengths as long as 750 mp.,’ Since metal complexes usually have long-wavelength ligand-field absorption bands the development of suitable actinometers should serve as a stimulus for intensive quantitative research in the area. 6 Photochemistry of Oxahto-complexes of Transition Metals There is a serious dearth of systematic information concerning the photochemical behaviour of complexes of a single ligand with various metal ions. The only ligand for which an appreciable body of such information exists is the oxalate ion.Absorption spectra of transition-metal oxalato-complexes consist of two main regions a fairly strong ligand-tmetal charge-transfer absorption at about 420 mp and one or more weak d-td bands at wavelengths greater than 600 mp. As one would expect from the nature of the spectra oxidation-reduction is the principal photochemical reaction undergone by most oxalato-complexes. For complexes of tripositive metal ions the usual stoi~heiometry~~ is as shown in (16). The quantum efficiency of the reaction can be correlated with ease of T. B. Copestake and N. Uri Proc. Roy. SOC. 1955 A 228,252. 225 2 Quarterly Reviews reduction of the metal ion. The most thorough study is that of Porter Doering and Karanka,4O who considered M = Crm Mnm F&II and CoLI1.As expected the chromium(m) complex exhibits no photoredu~tionl~~~~ while the others do; the order of decreasing reduction quantum yield is F S > Mnm> Corn which is precisely the order of increasing energy of the charge-transfer excited state. In the absence of light the complexes undergo thermal reduction; the order of decreasing dark reactivity is Mnm> Com>Fem. Evidently the photochemical yields do not correlate with the thermodynamic E" values for the complexed metal ions but must instead be compared with the energies of the charge- transfer excited states. Hence the relative oxidising powers of complexed metal ions may change considerably upon electronic excitation an important effect which has been more than occasionally overlooked. The variation of redox quantum yield with wavelength for oxalato-complexes is quite interesting.As expected yields are greatest in the charge-transfer region and gradually decrease with increasing wavelength (Figure 2). The Figure shows I I I I 3 0 0 0 4000 5 0 0 0 600 A (A) Fig. 2 Variation of quantum yield with wavelength for photoreduction ofi - 9 Fe(Ca03ss-; . . . . Mn(CaOJsS-; - - - Co(CaOJs&; - - - 9 Cr(CaOJsJ-. [Reproduced by pedssion from G. B. Porter J. G. W. Doering and S. Karanka J. Amer. Chem. SOC. 1962 84,4027.1 that in the charge-transfer region the yields are virtually constant but appreci- ably smaller than unity.40 The fractional yields cannot be rationalised by invoking the cage effect for its importance would increase with increasing wavelength (see p. 218). That the yields are constant but fractional may signify that they are limited by the efficiency of intersystem crossing to a metastable excited state before reaction.There are additional reasons for postulating intersystem crossing as a prerequisite for reaction in these systems as seen below. Figure 2 shows that the reduction quantum yield decreases as expected in the ligand-field absorption region but the yields remain quite appreciable even at very long wavelength^.^^^^^ This fact might suggest some form of radiationless 40 G. B. Porter J. G. W. Doering and S. Karanka J. Arncr. Chem. SOC. 1962,84,4027. 226 Wehry crossover from the ligand-field to chargetransfer states before reaction. There may be mixing of states such that the ligand-field levels possess considerable charge-transfer character i.e.net electron transfer from oxalate to metal ion is a highly probable event even though the long-wavelength absorption is nominally due to d-+d promotion. Alternatively a second discrete charge-transfer transition or a long tail of the first may be submerged in the ligand-field band. Low tem- perature spectroscopic studies and molecular orbital calculations on oxalato- complexes might serve to elucidate the nature of the long-wavelength excited states. It is clear from reaction (16) that the series of chemical steps following photo- activation is quite complicated. Most experimental work has concerned ferri- oxalate ion because of its great utility as an a~tinometer.~~ An important empirical observation is that two moles of bivalent-metal complex are produced per einstein of absorbed light.A plausible mechanism which rationalises that fact is shown& in (17)-(19). In step (17) excitation to the lowest charge-transfer Fest(C20f-) + [Fe2+(C202-) 2(C204-) ] * -+ [Fe2+( C202-)2(C204-) I* hv (17) (1 8) (19) intersystem crossing (1) [Fe*(C,O,Z-) (C204-) I* ' -+ Fe(C20a)2- + C204- Cz04- + Fe3+(C20,2-) __+ 2C02 + Fe(C20,,) + 2C,O4+ state perhaps followed by intersystem crossing to a metastable state is indicated. The excited complex then releases a Cz04- radical anion which attacks a molecule of the original complex resulting in production of two molecules of CO and two iron@)-oxalato-complexes per photon. It is unnecessary to assume that dissociation of the excited complex to yield C204- actually takes place; the same quantum dependence would result if the excited complex itself directly reacted with a ground-state molecule.It is however attractive to include the release of C204- in the mechanism because of the obvious analogy between that step and the homolytic fission processes which are believed to be important in photo- reactions of cobalt(m) and chromium(m) complexes. Further paramagnetic resonance indicates that radicals (presumably C02- and/or C204-) are formed in appreciable quantities during photoly~is.~~ It might reasonably be expected that the rate-limiting step in the above series of events would be reaction of C204- with ground-state complex reaction (19). If that were true flash photolysis should produce results consistent with second- order kinetics. In fact however the rate-determining step in flash photolysis of ferrioxalate comes closest to obeying first-order kinetics,"2 and the limiting reaction is evidently not that between ground-state complex and oxalate anion- radical.Perhaps the photochemical reaction rates and presumably the quantum 41 C. A. Parker Proc. Roy. SOC. 1953 A 220 104; C. G. Hatchard and C. A. Parker ibid. 1956 A 235 518. 42 C. A. Parker and C. G. Hatchard J. Phys. Chem. 1959 63,22. 48 D. J. E. Ingram W. E. Hodgson C. A. Parker and W. T. Rees Nature 1955,176,1227. 227 Quarterly Reviews yields as well are limited by intersystem crossing to a metastable charge-transfer excited state before reaction. That would lead to first-order flash kinetics and would rationalise the constant but fractional yields in the charge-transfer region noted above. Unfortunately most oxalato-complexes exhibit neither fluorescence nor phosphorescence so there exists no direct evidence for the significance of intersystem crossing.An alternative possibilitfe is that C204- and ground-state ferrioxalate form some type of intermediate species and the rate-determining step is dissociation of that intermediate. Several possible intermediates are illustrated in Figure 3 which also depicts the other pertinent stages in ferri- * 3- hu @ Excitation Trioxalato fenic iron p e 2 0 x ‘7 of D i s z t i o n metastable Dioxalato ferrous iron ox - Oxalate radical ox- + Oxalate radical Trioxalato femc iron metastable state 3- Possible structure of metastable state A free radical? A quartet? Or a fern-oxalate ion attached to an oxalate radical? (compare brown compound between NO and Fe2+) L- A v ox2- + @ Oxalate ion Muxalato femc iron attached to oxalate radical Dioxalato ferrous iron Fig.3 PossibZe stages in the photochemicul reduction of Fe(C,0Ja8-. [Reproduced by permission from C. A. Parker and C. G. Hatchard in ‘Photochemistry in the Liquid and Solid States’ ed. F. Daniels Wiley New York 1960 p 44.1 oxalate photolysis. Presumably the other photoreducible oxalato-complexes react via similar mechanisms. 228 Wehry As noted above Cr(C20,),S- is not photoisomerisation a process which photoaquation reactions discussed chromium(m) undergoes a thermal photoreduced. However it does undergo exhibits a number of similarities to the in the previous section. Trisoxalato- racemisation and light serves only to accelerate the reaction. Also the racemisation quantum yield is virtually wave- length-independent throughout the quartet-quartet absorption region ; there is no firm evidence for yield enhancement in the quartet-+doublet region.These facts suggest that the photoisomerisation of Cr(C20,)33- and the photoaquation of various chromium(I11) complexes proceed through primary processes that are identical or highly similar. In the case of the chromium(rI1)-oxalato-complex a plausible primary process is homolytic cleavage (20) of one of the oxalate co-ordinate bonds. It is assumed that the vacated co-ordination site is occupied hv by solvent water. The ‘displaced’ oxalate can subsequently eject the water molecule re-establishing the oxalate co-ordinate bond; when this occurs the product is simply the original complex. Alternatively the other end of the ‘half- co-ordinated’ oxalate may be displaced thermally followed by co-ordination of a second water molecule the net result being aquation.There exists a third possibility; if the displaced oxalate reco-ordinates by displacing another oxalate rather than ejecting the water molecule and the second displaced oxalate then reco-ordinates and ejects the water molecule the final product will be the enantiomer of the original complex.a4 The above mechanism though based on circumstantial evidence is useful because it emphasises that photoisomerisation and photoaquation reactions of the complex should exhibit very similar charac- teristics. The extent to which photoisomerisation competes with oxidation-reduction in other oxalato-complexes has not received much attention ; only the trisoxalato- complex of tripositive cobalt has been studied in any detail.Unlike the chromium(1u) system the total photochemical quantum yield for CO(C,O,)~- is very wavelength-dependent ;37943 the more important photoprocess is oxida- tion-reduction but Spees and Adamsonu estimate that about 15 % of the total quantum yield can be assigned to photoracemisation. It is probable that the two processes have common primary steps probably homolytic cleavage of a cobalt- oxalate bond analogous to reaction (20). One would then expect isomerisation to compete quite generally with photoreduction in oxalato-complexes ; this point does not appear to have been investigated although photoracemisations in 40 S. T. Spees and A. W. Adamson Znorg. Chem. 1962,1 531. 229 Quarterly Reviews several other oxalate-containing cobalt(rrr) complexes have been reported.45 It has been noted that the importance of racemisation (relative to other photo- processes) is significantly greater for the trisoxalato-complex of rhodium(n1) than for its congener of cobalt(rr~).~~ This represents only a qualitative observa- tion but it raises the interesting possibility that rather noteworthy photochemical differences may exist between the complexes of first-row transition metals and those of second- and third-row elements.More thorough comparative studies must be made before anything further can be said about this point. It is clear that in oxalato-complexes the sequence of ‘chemical’ steps following photoexcitation is quite complicated. That statement is likely to be generally applicable to co-ordination-compound photochemistry. Utilisation of flash photolysis magnetic resonance and mass spectrometry has revolutionised organic photochemistry; undoubtedly the application of these techniques to co-ordination-compound photochemistry will have a similarly beneficial effect. This work was supported in part by the National Institute of Health. l6 F. P. Dwyer and F. L. Garvan J. Amer. Chern. SOC. 1961 83 2610; F. P. Dwyer I. K. Reid and F. L. Garvan ibid. 1961 83 1285. 230

ISSN:0009-2681    

DOI:10.1039/QR9672100213

出版商:RSC    

年代:1967

数据来源: RSC

摘要:

The Cephalosporin C Group By E. P. Abraham SIR WILLIAM DUNN SCHOOL OF PATHOLOGY UNIVERSITY OF OXFORD 1 Introduction Cephalosporin C (1) was discovered at Oxford during a chemical study of the structure of penicillin N (2) a penicillin with a 8-(D-a-aminoadipoyl) side-chain produced by a species of CephaZosporium which had been isolated near a sewage outfall off the coast of Sardinia.ls2 When the penillic acid from a partly purified preparation of penicillin N was chromatographed on an anion-exchange resin it was followed from the column by a second substance that was detected by its ultraviolet absorption at 260 mp. This substance which was given the trivial name Cephalosporin C was readily isolated as a crystalline sodium salt and assigned the molecular formula C1,H,10,N,S.3 It was subsequently found to have antibacterial activity to resist hydrolysis by penicillinase from B.cereus,3 to show an extraordinary lack of toxicity to mice and to protect mice from infection with a penicillin-resistant strain of Staph. aureus which produced penicillinase? It closely resembled the penicillins and penicillin N in particular in some of its chemical and biological properties but differed from them strik- ingly in others. Cephalosporin C was formed in too small an amount by the Sardinian Cephalosporium sp. to have been detected in culture fluid by its antibacterial activity. Its presence was only revealed after it had been concentrated with penicillin N during the purification of the latter. The production of cephalo- sporin C in sufficient quantity for detailed chemical study was greatly facilitated by the isolation at the Medical Research Council’s former Antibiotics Research Station of a mutant strain from which much higher yields could be obtained.Culture fluids from this strain were incubated at pH 3 to convert penicillin N into its penillic acid and the unchanged cephalosporin C then purified by chromatography in volatile buffers on anion-exchange resins. The results of chemical investigations led to the suggestion of structure (1); and the validity of this structure was soon demonstrated by an X-ray crystallographic analysis.6 The antibacterial activity of cephalosporin C irz vitro was low. But the apparent relationship of the substance to the penicillin family coupled with its resistance to penicillinase from Staph. aureus gave it at once a potential clinical interest.G. Brotzu Lav. Ist. Igiene Cagliari 1948. E. P. Abraham and G. G. F. Newton Biochem. J. 1954,58,266. a G. G. F. Newton and E. P. Abraham Biochem. J. 1956,62 651. H. W. Florey Ann. Internal Med. 1955 43 480. E. P. Abraham and G. G. F. Newton Biochem. J. 1961 79,377. D. Hodgkin and E. N. Maslen Biochem. J. 1961,79 393. 23 1 Quarterly Reviews Before cephalosporin C could be assigned a definitive structure the possibility was envisaged of obtaining more active compounds by exchanging its ~-(D-W aminoadipoyl) group for other types of side-chain since benzylpenicillin was known to have a much higher activity than penicillin N against most gram- positive bacteria. This possibility was realised when the nucleus of cephalosporin C (7-arninocephalosporanic acid) was obtained in low yield by mild acid hydro- lysis and converted into N-acyl derivatives.' A further series of active compounds became accessible when it was discovered that the acetoxy-group of cephalo- sporin C and 7-aminocephalosporanic acid could be displaced by pyridine and other weak heterocyclic bases to yield betaines.* From an early stage work in this field was supported by the National Research Development Corporation which made agreements with a number of pharma- ceutical companies.Efforts by the pharmaceutical industry resulted in the isola- tion of more highly yielding strains and in the production of cephalosporin C on a large scale. A chemical method for removing the side-chain of cephalosporin C to yield 7-aminocephalosporanic acid in good yield was discovered in the Lilly Research Laborat~ries.~ In consequence two derivatives of cephalosporin C with approved names cephalothin and cephaloridine have been introduced into medicine by Eli Lilly and Company and Glaxo Laboratories respectively.The total synthesis of cephalosporin C and of cephalothin has recently been achieved at the Woodward Research Institute in Basel.lo 2 The Chemistry of Cephalosporin C A. Nomenclature.-The name cepham has been suggested for the l-aza-5- thia-6R-bicyclo[4,2,O]octan-8-one system by analogy with penam for the bicyclic system (3) in penicillanic acid.9 The systematic nomenclature of compounds containing the /3-lactam-dihydrothiazine ring system of cephalosporin C (1) L (4) JJ-9 (3) 0 'I B. Loder G. G. F. Newton and E. P. Abraham Biochem. J. 1961 79,408. C.W. Hale G. G. F. Newton and E. P. Abraham Biochem. J. 1961 79,403. R. B. Morin B. G. Jackson E. H. Flynn and R. W. Roeske J . Amer. Chem. SOC. 1962 84,3400. lo R. B. Woodward K. Heusler J. Gosteli P. Naegeli W. Oppolzer R. Ramage S. Rangana- than and H. Vorbriiggen J Amer. Chem. SOC. 1966,88 852. 232 Abraham is thus based on the trivial name d3-cephem for the bicyclic system (4) numbered as shown. B. Structure and Conformation.-Cephalosporin C sodium salt (Amax 260 mp log t 3.95; [aJ;* + 103") gave a positive ninhydrin reaction and behaved as an aminodicarboxylic acid on electrometric titration showing ionisable groups with PKa values <2.6 3.1 and 9.8 respectively.2s11 Its infrared absorption spectrum showed a band at 5.61 p a position at which the penicillins show absorption associated with the stretching vibration of the C = 0 of the /3-lactam ring in the fused p-lactam-thiazolidine ring system.It also showed bands at 5.77 p and 9.7 p which could be attributed to an ester grouping. Hydrolysis of cephalosporin C with hot acid yielded C02 and D-a-amino- adipic acid products which had been obtained earlier on hydrolysis of penicillin N. But cephalosporin C yielded two mol. of ammonia whereas only one was obtained under similar conditions from penicillin N. After hydrogenolysis of cephalosporin C with Raney nickel hydrolysis gave D-a-aminoadipic acid and L-alanine and partial hydrolysis gave a dipeptide of D-a-aminoadipic acid and ap-diaminopropionic acid (5). In neutral aqueous solution at 37" cephalo- sporin C was partially converted into ~-2-(4-amino-4-carboxybutyl)thiazole-4- carboxylic acid (6).12 The formation of these products and the infrared absorption spectrum could be accounted for by the partial structure (7) the thiazole (6) arising from opening of the /3-lactam ring fission between sulphur and the remainder of the molecule and a nucleophilic attack of sulphur on the amide carbon of the side-chain.An acetoxyl group accounted for two of the unplaced carbon atom in structure (7). The remaining five atoms like those of a corresponding fragment of the penicillins formed the carbon skeleton of valine since DL-valine and some a-oxoisovaleric acid were among the products formed when cephalosporin C was hydrogenolysed with Raney nickel. Under similar conditions D-valine was obtained from penicillin N by removal of sulphur from the penicillamine frag- ment of the molecule.However cephalosporin C unlike the penicillins did not yield penicillamine on hydrolysis and its nuclear magnetic resonance spectrum which did not give a signal at 7.9 p.p.m. indicated that agem-dimethyl group was not present in the molecule. Moreover cephalosporin C was much more stable at pH 3 than most of the penicillins and did not undergo a penicillin- penillic acid type of rearrangement. l1 E. P. Abraham and G. G. F. Newton Biochem. J. 1956 62 658. l2 J. d'A. Jeffery E. P. Abraham and G. G. F. Newton Biochem. J. 1960 75,216. 233 Quarter& Reviews Two sulphur-containing lactones were obtained by hydrolysis of cephalo- sporin C with 1*25~-HC1 at 100". On the basis of their physical and chemical properties one of these compounds appeared to have the a-tetronic acid structure (8) and the other to be a corresponding thiolactone (9).When treated with Raney nickel both compounds gave /I-methyl-a-tetronic acid. They were clearly formed by condensation of the five-carbon fragments from two molecules of cephalosporin C. In 0-1N-HCl at room temperature cephalosporin C itself lost an O-acetyl group and yielded a lactone which was named cephalosporin Cc (Amax 257 nip). Treatment of this lactone with Raney nickel gave a-amino- P-methylbutenolide (lo) which could be hydrogenated in the presence of Adams catalyst to y-hydroxyvaline lactone. The formation of compounds (8) (9) and (10) could be accounted for by the presence of the grouping (11) in cephalosporin C . The position assigned to the double bond was consistent with the isolation of hydroxyacetone 2,4-dinitro- phenylosazone after ozonolysis of cephalosporin C and treatment of the product with Raney nickel.Hence the structure (1) was suggested on chemical grounds for cephalosporin C and (12) for deacetylcephalosporin C lactone (cephalo- sporin Cc).5 Oxidation of cephalosporin C under mild conditions with hydrogen peroxide yielded a sulphoxide. The ultraviolet absorption of cephalosporin C in relation to structure (1) has been the subject of some speculation. It was suggested that the lone pair of electrons of the P-lactam nitrogen atom are involved in the chromophore because amide resonance is suppressed in the fused ring system5 and also that the sulphur is involved in resonance as shown for example in (13).13 The carboxyl group at C(4) is not an essential part of the cephalosporin C chromophore since (14) shows hmax 256 mp.14 Is A.G. Long and A. F. Turner Tetrahedron Letters 1963 7,421. l4 R. B. Morin B. G. Jackson R. A. Mueller E. R. Lavagnino W. B. Scanlon and S. L. Andrews J. Amer. Chcm. SOC. 1963 85 1896. 234 Abraham An X-ray crystallographic analysis of cephalosporin C sodium salt has given positions for the atoms which correspond with structure (1) and the correct absolute configuration of the molecule.6 The region comprising the #?-lactam ring and the atoms directly attached to it is closely similar to the corresponding region in benzylpenicillin. But the carboxyl group attached to C(6) of the rather flat dihydrothiazine ring lies much more in the plane of the ring than does the carboxyl group of benzylpenicillin which is turned at about right-angles to the thiazolidine ring.An X-ray crystallographic analysis of deacetylcephalosporin C lactone gave atomic positions which were defined with greater accuracy than in the earlier analysis of cephalosporin C and corresponded with (12). There was no significant difference in the /%lactarn ring of the cephalosporin and that of 6-aminopenicil- lanic acid but the angle between the p-lactam ring and the dihydrothiazine ring was less acute than that between the p-lactam and the thiazolidine ring.15 C. Other Chemical Reactions.-In boiling aqueous solution cephalosporin C is converted partly into a compound containing a fused imidazole-piperidine- 2-carboxylic acid ring system (15) which has been given the trivial name cephalo- sporidine.l8,l7 This compound is derived from the a-aminoadipoyl side-chain and C(6) and C(7) of the p-lactam ring.It is formed from penicillin N as well as from cephalosporin C. Many of the other chemical reactions of cephalosporin C involve changes in its ring system which can also be made with analogues in which the D&(a-arninoadipoyl) side-chain has been replaced by other groups. The infrared and proton magnetic resonance spectra of these analogues and their derivatives have been described in detail.18 3 Biosynthesis The cephalosporin C structure can be formally dissected into residues of D-a- aminoadipic acid L-cysteine a@-dehydro-y-hydroxyvaline and acetic acid whereas a dissection of the penicillins produced by P. chvysogenum gives a monosubstituted acetic acid L-cysteine and D-valine.When [2-14C]~~- a- aminoadipic acid [3-14C]~~-mesocystine and [ 1-l4C]~~-valine were added to fermentations with a CephaZosporiuin sp. the carbon-1 4 was incorporated mainly into the side-chain the carbon of the p-lactam ring and the valine-yielding fragment of cephalosporin C respe~tive1y.l~ [ 1 -14C]Acetate labelled the 0-acetyl l5 R. D. Diamond D.Phil. Thesis Oxford 1963. l6 E. P. Abraham and P. W. Trown Biochem. J. 1963,86,271. l7 E. 0. Bishop and R. E. Richards Bioehem. J. 1963 86,277. l9 P. W. Trown B. Smith and E. P. Abraham Biochem. J. 1963,86,284. G. F. H. Green J. E. Page and S. E. Staniforth J. Chem. Soc. 1965 1595. 235 Quarterly Reviews group and the D-a-aminoadipoyl side-chain. The distribution of carbon-14 in the side-chain of cephalosporin C labelled from [l-14C]acetate and from [5-14C]- a-oxoglutarate indicated that the formation of the D-a-aminoadipic acid involved the condensation of acetyl coenzyme A and a-oxoglutarate by reactions analogous to those by which a-oxoglutarate is formed from acetyl coenzyme A and oxaloacetate in the citric acid cycle.20,21 This is also the mechanism by which L-a-aminoadipic acid is formed in yeast.22 L-a-Aminoadipic acid is used by the Cephalosporium sp.for the synthesis of lysine saccharopine [~-N-(2-glutaryl)lysine] being a probable intermediate.23 Despite the presence of a D-a-aminoadipic acid residue in the side-chains of cephalosporin C and penicillin N free a-aminoadipic acid extracted from the mycelium has almost entirely (>99 %) the L-configuration. In suspensions of washed mycelium L-a-aminoadipic acid is a more efficient precursor of the side- chain than the isomer.^^ Thus the stage at which the D-configuration arises remains to be determined.Carbon-14 from both L- and ~-[~~C]valine is incorporated into the 5-carbon fragment of cephalosporin C by mycelial suspensions but D-valine is converted into L-valine in the r n y ~ e l i u m . ~ ~ ~ ~ No evidence has been obtained that free y-hydroxyvaline is an intermediate and the oxidation of a valine methyl group may occur at a later stage. A variety of penicillins with different side-chains can be obtained by the addition of appropriate side-chain precursors to fermentations with Penicillium chrysogenum but no analogue of penicillin N and cephalosporin C with side- chains other than D-a-aminoadipoyl has been obtained from the Cephalosporiurn sp.However isopenicillin N with an L-a-aminoadipoyl ~ide-chain,~~~~’ and a peptide which appears to be 6-(a-aminoadipoyl)cysteinylvaline28 are present in P . chrysogenum. It may be that L-a-aminoadipic acid is involved in the synthesis of all penicillins by P. chrysogenum but that no mechanism exists for the replace- ment of the D-a-aminoadipic acid residue by other groups in the synthetic path- ways which occur in the Cephalosporium sp. 4 Derivatives and Analowes of Cephalosporin C Both the a-aminoadipoyl and O-acetyl group of cephalosporin can be removed selectively and replaced by other groups. 2a P. W. Trown E. P. Abraham G. G. F. Newton C. W. Hale and G. A. Miller Biochem. J. 1962 84 157. p1 P. W. Trown M. Sharp and E. P. Abraham Biochem.J. 1963 86,280. 2a M. Strassman L. W. Ceci and B. E. Silverrnan Biochem. Biophys. Res. Comm. 1964,14 268. E. P. Abraham G. G. F. Newton and S. C. Warren I.A.M. Smyp. Appl. Microbiol. (Tokyo) 1964 6 79. 24 S. C. Warren G. G. F. Newton and E. P. Abraham Biochem. J. 1967 103 891. 26 A. L. Demain Biochem. Biophys. Res. Comm. 1963 10 45. 26 E. H. Flynn M. H. McCormick M. C. Stamper H. DeValeria and C. W. Godzeski J. Amer. Chem. SOC. 1962 84,4594. 27 M. Cole and F. R. Batchelor Nature 1963 198 383. ** H. R. V. Amstein M. Artman D. Morris and E. J. Toms Biochem. J. 1960,16,353,357. 236 Abraham A. 7-Aminocephalosporaniic Acid.-Hydrolysis of cephalosporin C with aqueous acid under mild conditions resulted in the removal of the a-aminoadipoyl side- chain and the production in very small yield of a compound in which the remainder of the molecule was unchanged.' This compound [3-acetoxymethyl-7- aminoceph-3-em-4-oic acid (1 6)] was given the trivial name 7-aminocephalo- sporanic acid by analogy with 6-aminopenicillanic acid in the penicillin series.It readily yielded N-acyl derivatives with acid chlorides in aqueous acetone containing bicarbonate. But competing reactions such as the removal of the 0-acetyl group and subsequent lactonisation prevented this hydrolytic method from being a practical one for the preparation of 7-aminocephalosporanic acid in quantity. Attempts to find an enzyme which will catalyse the removal of the a-aminoadipoyl side-chain from cephalosporin C have hitherto been unsuccess- ful. In contrast enzymes which will remove the phenylacetyl and other non- polar side-chains from both the penicillins and analogues of cephalosporin C are widely distributed in micro-organisms.The specificity of these acylases is associated with the nature of the N-acyl side-chain rather than with the structure to which the latter is a t t a ~ h e d . ~ ~ ~ ~ ~ However chemical methods have been devised by which 7-aminocephalosporanic acid can be obtained from cephalo- sporin C in good yield. Treatment of cephalosporin C with nitrosyl chloride in anhydrous formic acid gave an intermediate iminolactone (17) which was hydrolysed to 7-amino- cephalosporanic acid and a-hydroxyadipic acid when dissolved in water.g Cephalosporins which have found general clinical use have so far been prepared from 7-aminocephalosporanic acid obtained in this way31 and have the general X R x.OCOMe (9) R SEt S,O,Na (k) PhCH-NH 0-COMe b (1) PhCH O-COMe a9 M. Cole Nature 1964 203 519. 30 W. Kaufmann and K. Bauer Nature 1964,203 520. 31 R. R. Chauvette E. H. Flynn B. G. Jackson E. R. Lavagnino R. B. Morin R. A. Mueller R. P. Pioch R. W. Roeske C. W. Ryan J. L. Spencer and E. van Heyningen Antimicrobial Agents and Chemotherapy Amer. Sac. Microbiol. Ann Arbor Michigan 1962 p. 687. 237 Quarterly Reviews structure (18). For example cephalothin (18a) is the sodium salt of 7-(thiophen- 2-ace t amido)cephalosporanic acid. In another procedure for the preparation of 7-aminocephalosporanic acid cephalosporin C is first converted into a diester by protection of its free NH with a benzylcarbonyl or other suitable group esterification and removal of the protecting group.An ester of 7-aminocephalosporanic acid is obtained in good yield together with the ester of 6-oxopiperidine-2-carboxylic acid when the diester of cephalosporin C is kept for several days in methylene dichloride containing acetic acid at room tem~erature.~~ In a-third procedure the diester of an N-acyl derivative of cephalosporin C is converted into an iminochloride by treatment with phosphorus oxychloride and then to an amino-ether (19). The latter yields 7-aminocephalosporanic acid ester on hydrolysis in dioxan-aqueous phosphoric acid.33Q The use of benzyl and diphenylmethyl esters from which the corresponding acids can be obtained by hydrogenolysis and by hydrolysis with trifluoroacetic acid in anisole respectively enables free 7-aminocephalo- sporanic acid to be prepared by these procedures.Hydrolysis of the t-butyl ester of 7-aminocephalosporanic acid is also readily achieved with trifluoroacetic acid.33b 7-Chloroacetarnidocephalosporanic acid reacts with thiourea in water to give 7-aminocephalosporanic acid and an iminothiazolidone presumably by the intramolecular displacement shown in (20).= €3. Deacetyl and Deacetoxy-cepha1osporins.-Treatment of cephalosporin C with an acetyl esterase yielded deacetylcephalosporin C (1 8b) which readily Iactonised in acid solution.36 Hydrolysis of the O-acetyl group of other N-acyl derivatives of 7-aminocephalosporanic acids has been carried out similarly and the resulting deacetyl compounds [(18) X = OH] have been assigned the trivial name cephalosporadesic acids. O-Aroyl derivatives of cephalosporadesic acids have been prepared by a Schotten-Baumann reaction with aroyl chlorides and soditun hydroxide in aqueous acetone.36 Hydrogenation of cephalosporin C in the presence of a large amount of palladium catalyst results in hydrogenolysis of the allylic acetoxy-group and the 32 Ciba Belg.P. 645,157/1964. 33 (a) Ciba Belg. P. 643,899/1964; (b) R. J. Stedman J. Med. Chem. 1966,9,444. 34 J. D. Cocker B. R. Cowley J. S. G. Cox S. Eardley G. I. Gregory J. K. Lazenby A. G. Long J. C. P. Sly and G. A. Somerfield J. Chem. SOC. 1965 5015; H. Fazakerley D. A. Gilbert G. I. Gregory J. K. Lazenby and A. G. Long Autumn Meeting of the Chemical Society Nottingham Sept. 1965. 35 J. d’A. Jeffery E. P. Abraham and G. G. F. Newton Biochem. J. 1961 81 591. 36 E. van Hayningen J. Med. Chem. 1965 8,22.238 Abraham formation of deacetoxycephalosporin C (18~).1~1~~ Removal of the acetoxy- group from 7-aminocephalosporanic acid has been reported to occur under similar conditions. C. Displacement of the Acetoxy-group by Nucleophiles.-During the purification of cephalosporin C in pyridine acetate buffer a second active compound was encountered which showed no net charge at pH 7 and was given the trivial name cephalosporin CA (pyridine) (1 8d). This compound was shown to be a pyridinium betaine formed by displacement of the acetoxy-gro~p.~~ Similar displacements were found to occur with surprising facility with a variety of other heterocyclic weak bases and with 7-aminocephalosporanic acid as well as with cephalosporin C itself. With sodium thiosulphate cephalosporin C yielded a Bunte salt (1 8e)?9940 The displacement of the acetoxy-group in a variety of 7-acylaminocephalo- sporanic acids has been studied in detail.Substitution occurred with azide (1 8f) ethanethiol (1 8g) thiobenzoate (1 8h) thiourea (1 89 dithiocarbamates and xanthates as well as with pyridine and other heterocyclic base~.~,*l The reaction proceeds by an S,l With certain bidentate nucleophiles such as pyrid-Zthione displacement by the sulphur is followed by an internal Michael addition by the nitrogen atom to the double bond of the thiazine ring and the formation of spirocyclic compounds.3P However the methyl esters hydroxy-acids and lactone derived from (1 8) did not undergo analogous replace- ment reactions and little substitution occurred with a l-oxide. It has been sug- gested that the l-sulphur atom is implicated in the displacement reaction and that alkyl-oxygen fission of the acetyl ester is facilitated by resonance forms &$.0 - 0 indicated in (21).= Cephaloridine (l8j) which is used clinically is the pyridinium betaine derived from cephalothin [7-(2-thienyl)acetamido-3-( l-pyridylmethyl)-3- cephem-karboxylate betai~~e].~ D. Isomerisation to 7-Acylaminoceph-2-em-4-carboxylic Acids.-In the presence of pyridine and acetic anhydride 7-acylaminocephalosporanic acids undergo an isomerisation involving a shift of the double bond to the 2,3 position. These compounds isomerise slowly in the presence of pyridine alone but their esters 37 R. J. Stedman K. Swered and J. R. E. Hoover J. Med. Chem. 1964 7 177. 38 C. W. Hale G. G. F. Newton and E. P. Abraham Biochem.J. 1961 79 403. 39 A. L. Demain J. F. Newkirk G. E. Davies and E. R. Harman Appl. Microbiol. 1963,11 58. 40 A. L. Demain Trans. N. Y. Acad. Sci. 1963 25 731. 41 E. van Heyningen and C. N. Brown J . Med. Chem. 1965,8 174. Q3 P. W. Muggleton C. H. O’CalIaghan and W. K. Stevens Brit. Med. J. 1964,2 1234. A. B. Taylor J. Chem. SOC. 1965 7020. 239 Quarterly Reviews do so more readily. An equilibrium is reached when the proportion of d2 ds is about 7 :3. The A2 compounds absorb strongly between 220 and 300 mp and their esters show Amax or an inflexion between 245 and 250 m,u.44 The acetate group in the ceph-2-em-4-carboxylic acids is replaced by pyridine thiourea and thiobenzoate but the reactions are slower than the corresponding ones with the d3 compounds. Hence the A2 compounds are presumably not involved in the displacements with 7-acylaminocephalosporanic acids.Con- formatioils such as (22) with the 4-carboxy-group pseudo-axial to the six- membered ring have been suggested for the A2 compounds.44 E. Base- and Enzyme-catalysed Hydrolysis of Derivatives of 7-Aminoceph- alosporanic Acid.-When the @-lactam ring of the penicillins is opened by mild alkaline hydrolysis or with enzyme penicillinase the immediate products are penicilloates which are well defined and relatively stable compounds. Early attempts to characterise the products of a similar hydrolysis of cephalosporin C indicated that the corresponding compound (23) had no more than a transitory C existence and was rapidly fragmented in aqueous s01ution.l~ Opening of the p-lactam ring of cephalosporin C and cephalothin by a /3-lactamase from Pseudomonas pyocyanea at pH 7 is accompanied by the spontaneous expulsion of acetate and a similar hydrolysis of cephaloridine by the expulsion of ~ y r i d i n e .~ ~ Deacetylcephalosporin C also undergoes further fragmentation when its p- lactam ring is opened but the corresponding lactone yields a relatively stable product (Amax 265 mp) which presumably has the structure (24). Expulsion of the acetoxy-group probably as part of a concerted reaction involving a prototropic rearrangement occurs on fission of the @-lactam ring of 7-phenylacetamidocephalosporanic acid [cephaloram (1 81) ] with sodium benzyloxide in benzyl alcohol but in this case a major product of the reaction Scheme I 44 J. D. Cocker S. Eardley G. I. Gregory M.E. Hall and A. G. Long J. Chem. Soc. (C) 1966,1142. 45 L. D. Sabath M. Jago and E. P. Abraham Biochem. J. 1965,96 739. 240 Abraham has been isolated and shown to be the 6H-1,3-thiazine (25) or a tautomer. The latter map6 be formed as shown in Scheme 1. The p-lactam rings of the ceph-2-ems are much more stable to alkali than the corresponding d3-compounds.44 5 Chemical Routes from the Penicillin to the Cephalosporin Series Two chemical methods have been reported for the conversion of the fused p-lactam-thiazolidine ring system into p-lactam-dihydrothiazines. The sulphoxide of phenoxymethyl penicillin ester [ (26) R = C6H,0CH2] gives a compound containing the 3-methylceph-3-ern-4-carboxylic acid ring system (27) in 15% yield on refluxing in xylene in the presence of traces of toluenesulphonic acid.14 The transformation (Scheme 2) is presumed to involve the intermediate (28).Activation of the carboxyl group of a penicillin followed 0- Scheme 2 by refluxing in the presence of a base yields an anhydropenicillin (29). It has been stated that anhydropenicillins can be converted into compounds of the cephalosporin C series either by allylic bromination with N-bromosuccinimide to give (30) and treatment of the latter with base or by hydroxylation with microbial enzyme~,4~ but no details of these processes have been given. 6 Synthesis of Degradation Products A. Compounds (6) and (15).-The DL form of the thiazole (6) was synthesised by condensation of the thioamide (3 1) with methyl bromopyruvate and subse- 46 S. H. Eggets V. V. Kane and G. Lowe J. Chem. SOC. 1965 1262.47 S. Wolfe J. C Godfrey C. T. Holdrege and Y. G. Perron J. Amer. Chem. SOC. 1963 85 643; Belg. P. 621 452/1963. 24 1 Quarter& Reviews quent treatment with ~ N - H C ~ at 110'. It was identical with the compound obtained by racemisation of the thiazole from cephalosporin C with acetic anhydride.12 The DL form of cephalosporidine (15) was synthesised by heating DL& (waminoadipoy1)aminoacetaldehyde diethylacetal in aqueous acetic acid at 100". It was identical with the compound formed by racemisation of cephalosporidine at 19O0.l6 Cephalosporidine is presumably formed from the D-8-(a-amhoadipoyl) aminoacetaldehyde fragment of penicillin N and cephalosporin C by intra- molecular condensation. B. Compounds (8) and (lo).-Synthetic routes to the sulphur-containing lactone (8) and aminobutenolide (10) start from /3-dimethylaminomethyl-a-tetronic acid hydrochloride first obtained in 1924 from pyruvic acid dimethylamine hydrochloride and f~rmaldehyde.~~ The free base (32) reacted with the sulphy- dry1 anion in dimethylformamide to yield (33).It appeared that the enolate salt (33) formed initially equilibrated with the S-anion (34) and that this was followed by a nucleophilic attack of the latter on the f ~ r r n e r . ~ ~ ~ ~ ~ ~ + - O D 0 (33) HofiO 0 (34) Reaction of (32) with toluene-w-thiol gives p-benzylthiolmethyl-ctetronic acid from which the enamine (35) is obtained by fusion with ammonium acetate. Treatment of (35) with Raney nickel gives The latter has also been obtained by fusion of /%rnethyl-a-tetronic acid with ammonium acetate.PhC H?S p 2 C. Compound (25).-Synthetic routes udc . have been found to t,,e d4-dihydro- 6H-1,3-thiazine system (36) and to some 6H-1,3-thiazines depend on the addition of a thioamide to an ap-unsaturated ketone.52 The thioamide (37) was prepared by reduction of the hydroximino-nitrile (38) acylation of the amine with phenylacetyl chloride and treatment of the product with hydrogen sulphide. 48 C. Mannich and M. Bauroth Ber. 1924,27 1108. 49 (a) E. Galantay H.Enge1 A. Szabo and J. Fried J . Amer. Chem. SOC. 1964 29 3560; (b) J. C. Sheehan and J. A. Schneider J. Org. Chem. 1966 31 1635. (c) R. HeymBs G. Amiard and G. Nomine Compt. Rend. 1966 263 170. 50 A. G. Long and A.F. Turner Tetrahedron Letters 1963 421. 51 D. M. Green A G. Long P. J. May and A. F. Turner J. Chem. SOC. 1964 766.52 G. C. Barrett S.H. Eggers T. R. Emerson and G. Lowe J. Chem. SOC. 1964 788. 242 Abraham HO*N=FC02CH2 Ph (38) CN I (37) / CH2=Ctvk COCO,*CH,Ph (39) When (37) reacted with the vinyl keto-ester (39) in dioxan saturated with hydro- chloric acid dehydration as well as addition occurred and the product (A,, 342 and 285 mpu) was the 6H-1,3-thiazine (40) or a tautomer. Hydrolysis with aqueous ethanolic sodium carbonate converted (40) into a half-ester identical with (25) obtained from 7-phen ylacetamidocep halosporanic 7. The Total Synthesis of Cephalosporins Several possibIe synthetic routes were envisaged which might lead to the 3,6- dihydro(2H)-l,3-thiazine-P-lactam ring system of the 7-acylaminocephalo- sporanic acids. One which resembled a route followed successfully in the penicillin field and has been used as a synthetic approach to the cephams from homocy~teine,4~~ would have involved the condensation of a phthalimidomalon- aldehydate ester with the butenolide (41) followed by subsequent closure of the p-lactam ring and re-opening of the lactone ring.49a Another involved the addi- tion of a thioamide to a vinyl keto-ester for the construction of the dihydro- thiazine ring.52 A third envisaged ring expansion of a thiazolidine to give the appropriate dihydrothiazine and was analogous in some respects to the route from the sulphoxide of phenoxymethyl penicillin ester (26) to (27).In a model experiment the methanesulphonate (42) appeared to be converted in high yield into (43) when refluxed in dioxan with anhydrous sodium acetate.53 None of these suggested procedures has yet been brought to a successful conclusion.However a synthesis of D,L-deacetylcephalothin lactone has recently been reported which involves condensation of the unstable /3-thiolmethyl-or-tetronic acid (33) with the enamine of phthalimidornalonaldehydate t-butyl ester the subsequent stages being similar to those in the penicillin series.49c A major difficulty confronted attempts to synthesise cephalosporin C by a route similar to that which was successful with the penicillins and involving closure of the p-lactam ring as the final step. The immediate product of cleavage 53 G. Stork and H. T. Cheung J. Amer. Chem. SOC. 1965,87 3783. 243 Quarterly Reviews of the /?-lactam ring of a 7-acylaminocephalosporanic acid unlike the penicil- loates was highly unstable.The first total synthesis of cephalosporin C and cephalothin has now been achieved by a new and ingenious approach to the problem and the discovery of a remarkable series of reactions.1° L- C y s t eine was converted into L-N- t - bu t yloxycarbonyl-2,2-dime t hyl- thiazolidine-4-carboxylic acid [(44) R = HI. The methyl ester [(44) R = CH,] reacted with an excess of dimethyl azodicarboxylate at 105" to give the hydrazo- diester (45). Me Me Me Me Bu'OCO*N S A wl'OCO*N A R02c PS H (44) M 2 c ~ . y . N H * c 0 3 . l 0 CO-Me 2 Me ,Me Me Me Me Me A 6u'OCO.N A A Bu'0.CO.N S (46) (47) (48) When (45) was oxidised with lead tetra-acetate in boiling benzene and the reaction mixture treated with sodium acetate it was converted into the trans- hydroxy-ester (46). Treatment of (46) with excess of di-isopropylethylamine and methanesulphonyl chloride led to an intermediate from which the methane- sulphonate group could be displaced by azide ion with normal inversion to yield the cis-azido-ester (47).The latter was reduced with aluminium amalgam to the cis-amino-ester (48). The structures of the surprisingly stable trans- hydroxy-ester and the cis-amino-ester were rigidly established by X-ray crystal- lographic analysis at Harvard. Thus by these novel reactions a properly oriented nitrogen atom was introduced into the /?-position of a cysteine residue to produce an arrangement of atoms corresponding to one moiety of cephalosporin C. The cis-amino-ester (48) yielded the /?-lactam (49) on treatment with tri- isobutylaluminium in toluene. For reaction with the p-lactam a novel dialdehyde was synthesised in which a CHO carboxyl function was covered by a new protective group that could be removed by reduction.Di-P/?fl-trichlorethyl D-tartrate was oxidised with sodium meta- periodate to /3/3/3-trichloroethyl glyoxylate (50). Condensation of the latter in aqueous solution with the sodium salt of malondialdehyde gave the aldol(51) 244 Abraham which lost water on heating in n-octane with the formation of the dialdehyde (52). When (52) was heated with the p-lactam (49) in n-octane at 80" for 16 hours the adduct (53) was formed This yielded the aminoaldehyde (54) in trifluoro- acetic acid at room temperature the reaction involving the formation of a six- Me .Me - C02CH;CCL3 - . (54) -2u 2 LbL3 membered ring by attack of the electrophilic carbon atom of a protonated carbonyl group on the nucleophilic sulphur atom and the removal of the N-t- butyloxycarbonyl group.Acylation of the aminoaldehyde (54) with thiophen-2-acetyl chloride and reduction of the aldehyde group with diborane in tetrahydrofuran yielded an alcohol which gave the d2 isomer of cephalothin ppp-trichloroethyl ester (55) on acetylation with acetic anhydride and pyridine. In anhydrous pyridine (55) equilibrated with the ppp-trichloroethyl ester of cephalothin and the two isomers were separated by chromatography on silica gel. Reductive removal of the ppp-trichloroethyl group with zinc dust in 90% acetic acid yielded cepha- lothin itself (1 8a). The aminoaldehyde (54) was also acylated with N-/3#-trichloroethyloxy- carbonyl-D-a-aminoadipic acid and the N-acyl derivative esterified with /3p/3- trichloroethanol in the presence of dicyclohexylcarbodi-imide and pyridine.Reduction of one of the products followed by acetylation isomerisation and reductive removal of the trichloroethyl groups from the d3 isomer yielded cephalosporin C (1). 8 Structure-Activity Relationships A. Antibacterial Activity in Vitro.-Early studies with cephalosporin C revealed that this antibiotic had a broad antibacterial spectrum but that its activity was relatively low. It showed only about 0.1 % of the activity of benzylpenicillin in vitro against a number of gram-positive bacteria. The activity of penicillin N against these organisms was about 1 % of that of ben~ylpenicillin.~~~ However cephalosporin C unlike benzylpenicillin and penicillin N was as active against penicillinase-producing staphylococci as against non-penicillinase producers.M Its activity against the penicillinase-producing organisms cobild be correlated with its resistance to hydrolysis by staphylococcal penicil1ina:;e.This resistance to the enzyme was associated with the ring system of the molecule and not with the nature of the N-acyl side-chain since penicillin N which .also contained a &(D-a-aminoadipoyl) sidechain was rapidly hydrolysed. Comparison of the 54 H. W. Florey Giorn. Microbiol, 1956 2 361. 245 Quarterly Reviews properties of cephalosporin C penicillin N and benzylpenicillin therefore suggested that replacement of the Nacyl side-chain of cephalosporin C by phenylacetyl or a related group would lead to compounds with a much higher activity against gram-positive bacteria which retained a resistance to penicil- linase.This was shown to be so when 7-phenylacetamidocephalosporanic acid was first ~repared.~ The activity of this compound against the Oxford strain of Staph. aiireus was several hundred times that of cephalosporin C and of the same order as though less than that of benzylpenicillin. Deacetylcephalosporin C shows about 25 % of the activity of cephalosporin C deacetoxycephalosporin C appears to be less active than the deacetyl derivative and the activity of 7-amino- cephalosporanic acid itself is extremely low. In contrast the betaine obtained on replacement of the acetoxy-group in cephalosporin C by a pyridiniuni group was about ten timss as active against the staphylococcus as cephalosporin C itself. The discovery of a procedure for the production of 7-aminocephalosporanic acid in quantity has resulted in the production by pharmaceutical companies of many hundreds of cephalosporins with the general structure (18).The prepara- tion of these derivatives has followed that of a series of new penicillins obtained in a similar manner from 6-aminopenicillanic acid.55 In general the relative changes in activity against a non-penicillinase producing strain of Staph. aureus which occw when one N-acyl side-chain is replaced by another are similar in the cephalosporin and penicillin series of compounds. As with cephalosporin C other N-acyl derivatives of 7-aminocephalosporanic acid yield compounds with different activities when changes are made in group X (18).31,5s It is not easy to predict the effect of changes in R and X together from the effects observed on changing each group separately.But cephalothin (1 8a) and cephaloridine (1 8j) have high activities against many gram-positive bacteria and useful activities against a number of gram-negative bacteria.57 Both compounds have been found to have clinical value for the treatment of a variety of infections. Important differences between the antibacterial activities of the cephalosporins and penicillins respectively are associated with differences in the behaviours of the two groups of compounds to p-lactamases. It is now clear that p-Iactamases from different bacteria may differ greatly in their behaviour to a given member of the cephalosporin or penicillin family and in particular that staphylococcal penicillinase differs from the 13-lactamases produced by gram-negative organisms.Compounds with the /%lactam-dihydrothiazine ring system are resistant to hydrolysis by the staphylococcal enzyme even when they have a high affinity for the enzyme.58 Thus cephalosporins with N-acyl side-chains such as phenylacetyl and thiophen-2-acetyl which confer a high intrinsic activity on the molecule 55 F. R. Batchelor E B. Chain T. L. Hardy K. R. L. Mansford and G. N. Rolinson Proc. Roy. SOC. 1961 B 154 498. 56 R. R. Chauvette E. H. Flynn B. G. Jackson E. R. Lavagnino R. B. Morin R. A. Mueller R. P. Pioch R. W. Raeske C. W. Ryan J. L. Spencer and E. van Heyningen J. Amer. Chem. SOC. 1962 84 3401. 57 M. Barber and P. M. Waterworth Brit. Med. J. 1964 2 344. B. Crompton M. Jago K. Crawford G. G. F. Newton and E. P. Abraham Biockaem.J. 1962,84,52. 246 Abraham against most gram-positive bacteria are also highly active against the penicil- linase-producing staphylococcus. In the penicillin series resistance to staphy- lococcal penicillinase depends on the side-chain and not on the ring system. Penicillins with certain side-chains such as 2,6-dimethoxybenzoyl have a very low affinity for the enzyme59 but a considerable intrinsic activity against the staphylococcus. In contrast to the staphylococcal enzyme 8-lactamases from some gram- negative bacteria may hydrolyse cephalosporin C cephalothin and cephal- oridine at a higher maximum rate than benzylpenicillin or than ampicillin which has a D-phenylglycyl side-chain. p-Lactamases from other gram-negative bacteria appear to show penicillinase activity but little cephalosporinase activity.60 Penicillins and cephalosporins with a 2,6-dimethoxybenzoyl side- chain show a very high affinity for some of these p-lactamases (in contrast to their low affinities for the staphylococcal enzyme) and are resistant to enzymic hydrolysis but they have only a very low activity against the organism by which the enzymes are pr~duced.*~,~l It appears that cephalothin and cephaloridine are more active than ampicillin against some gram-negative bacteria but not against others.The relative importance of the r6les played by /3-lactamase production and by the inlierent resistance of bacterial cells in this context has still to be evaluated. However ampicillin is normally the penicillin of choice for the treatment of infections by gram-negative bacteria The question therefore arises whether the respective advantages of ampicillin and of a cephalosporin can be combined in a compound having the side-chain of the former and the ring system of the latter.This compound named cephaloglycin [7-(~-a-amino- pheny1acetarnido)cephalosporanic acid (1 8k) 1 has shown interesting biological properties in a preliminary study.62 The available evidence strongly supports the view that the mode of action of the cephalosporins is essentially the same as that of the penicillin^.^^,^,^^ The latter inhibit the cross-linking by a transpeptidation reaction of mucopeptide formed in the synthesis of the bacterial cell-wall. It has been pointed out that there are resemblances between a face of the penicillin molecule and certain faces of N-acetylmuramic ~-alanyl-~-alanine:~ and L-alanine-D-glutamic acid6* respectively which are components of the mucopeptide network.Hypo- theses about the mode of action of penicillin have invoked each of these com- ponents in turn as one with which penicillin competes for attachment to an enzyme surface by hydrogen and ionic bonds. This primary combination of 59 R. P. Novick Biochem. J. 1962 83 229. 6o P. C. Fleming M. Goldner and D. G. Glass Lancet 1963 1399. 62 W E. Wick and W. S. Boniece Appl. Microbiol. 1965 13 248. 63 E. P. Abraham and G. G. F. Newton Ciba Found. Symp. Amino-acids and Peptides with Antimetabolic Activity 1958 p.205. 64 T. W. Chang and L. Weinstein Science 1964 143 807. 65 J. Bond R. W. Brimblecombe and R. C. Codner J . Gem Microbiol. 1962,27 11. 66 J. F. Collins and M. H. Richmond Nature 1962 115 142.67 E. M. Wise and J. T. Park Proc. Nat. Acad. Sci. 1965 54 75. J. M. T. Hamilton-Miller J. T. Smith and R. Knox Nature 1964 201 867. D. J. Tipper and J. L. Strominger Proc. Nat. Acad. Sci. 1965 54 1133. 247 Quarterly Reviews penicillin with the enzyme may be followed by opening of the /?-lactam ring and formation of a covalent bond. X-Ray crystallographic analyses have shown that the relative positions of the atoms in the #?-lactam ring of cephalosporin C and the atoms immediately attached to the ring are similar to though not identical with those in benzyl- penicillin and 6-aminopenicillanic acid. But the orientation of the carboxyl group attached to C(4) in (18) which lies in the plane of the six-membered ring is somewhat different from that of the corresponding carboxyl group in the penicillins.Moreover deacetylcephalosporin lactones (23) may have high anti- bacterial activity.31 It seems therefore that the presence of a carboxylate ion is not essential for antibacterial activity but that if this group is present its orientation may be varied within certain limits. When more extensive changes in configuration occur as in synthetic enantiomorphs of the natural penicillin~,6~ or when structural changes are made which result in a substantial decrease in the sensitivity of the /?-lactam ring to attack by nucleophiles activity may be lost. Either or both of these factors could be responsible for the inactivity of the d2 isomers of cephalosporins in a conformation such as that shown in (22). B. Absorption and Hypersensitivity.-Two other biological properties appear to be associated specifically with the p-lactam-dihydrothiaine ring system.One is the failure of many cephalosporins to be absorbed as efficiently from the gastro-intestinal tract as are penicillins with the same N-acyl side-chains. The precise features of the two ring systems which are responsible for this difference have not been established. A second property is the failure of cephalosporins in most cases to produce a reaction in patients reported to be hypersensitive to penicillin. Hypersensitivity to the penicillins is a complex phenomenon and the nature of the antigen responsible for its most serious manifestation anaphylactic shock remains to be establi~hed.'~ But it is a reasonable assumption that this antigen like those responsible for milder reactions is associated with the open- ing of the penicillin p-lactam ring and linkage of a transformation or degradation product of the molecule to tissue protein. The opening of the /?-lactam ring of the cephalosporins in aqueous solution is associated with an extensive fragrnen- tation of the molecule. This could well result in the formation of antigens from the cephalosporins which would not react with antibodies to conjugated proteins derived from the penicillins. 69 J. C. Sheehan and K. R. Henery-Logan J. Amer. Chem. SOC. 1959 81 3089. 7 0 B. B. Levine Fed. Proc. 1965 24 45. 248

ISSN:0009-2681    

DOI:10.1039/QR9672100231

出版商:RSC    

年代:1967

数据来源: RSC

摘要:

Hydrogen Abstraction in the Liquid Phase by Free Radicals By R. S . Davidson UNIVERSITY OF LEICESTER One of the most frequently encountered reactions of free radicals is that of hydrogen abstraction. The purpose of this Review is to draw together informa- tion on the abstraction of hydrogen from a variety of hydrogen-containing bonds in a variety of environments X-H + R. -+ X. + R-H General Principles Bond Strengths.-The significance of the strength of the bond broken (X-H) and of the one formed (R-H) in the reaction has been considered in a previous Review. Polar Effects.-On the basis of Hammonds’ postulate2 (see Figure) we can say that for a highly exothermic reaction (A) the transition state will resemble the reactants i.e. very little rupture of bond X-H will have occurred. For a less exothermic reaction (B) the transition state will bear more resemblance to the products i.e.more bond rupture will have occurred. From this consideration we would expect polar effects within the substrate molecule which affect the stability of the incipient radical to be more important in reactions of high Reaction co-ordinote __t Energy diagram for the X-H + R. + X. + H-R reactions A Reaction with reactive free-radical B Reaction with stable free-radical J. M. Tedder Quart. Rev. 1960 14 336. G. S. Hammond J. Amer. Chem. SOC. 1955,77 334. 249 Quarter& Reviews activation energy. This has been demonstrated in the abstraction of hydrogen from benzylic C-H bonds.3 The order of reactivity of C-H bonds has been found to be tertiary> se~ondary>primary.~ This difference in reactivity is dependent upon the difference in stability of their respective incipient radicals.It is not surprising therefore that a reactive radical (e.g. Cl. CH,.) shows little dis- crimination between the different types of bond whereas less reactive radicals (e.g. Br. CCl,) show much greater selectivity in their attack. In the course of attack upon a bond (X-H) by a radical (R.) electron transfer to or from the radical may occur and to signify the preferred direction of move- ment the terms acceptor and donor radicals have been suggested.s The direction of the transfer depends on the relative stabilities of the R+ and R- ions. Alkyl radicals are classed as donor radicals and in certain cases this has been verified experimentally.s The donor property is to be expected from a consideration of the stability of carbonium ions relative to carbanions.The majority of radicals halogen alkoxyl peroxy etc. have been found to behave as acceptor radicals. The preferred direction of electron transfer is particularly relevant to the type of polar structures which can be written as contributing to the transition state e.g. X-H.+ R. -+ [Xa+--H-R8- X+H* R-] -+ X. + H-R R. - an acceptor radical Substituents in the substrate molecule which favour the polar structures prekrred by the attacking radical wiil obviously facilitate the reaction however little bond rupture has occurred in the transition state. Russell’ has determined the reactivity of a number of nuclear-substituted cumenes towards peroxy- radicals (reactive acceptor radicals) and has found that electron-releasing sub- stituents (e.g.alkyl groups) facilitate attack. Electron-attracting substituents (e.g. halogen nitro- or cyano-groups) had a deactivating influence even though some of them can stabilise the incipient radical. Electron-releasing groups (e.g. RO R,N) are particularly effective in activating a hydrogen atom attached to the same carbon atom as themselves since they can also stabilise the incipient radical by resonance. Electron-withdrawing groups on the other hand de- activate similar C-H bonds to attack by acceptor radicals since they do not favour the development of the required polar character in the transition state. Such an effect does not operate in reactions with donor radicals since the con- tributing polar structures to the transition state are of a different type.It has been shown for instance that the carboxyl group is particularly effective in de- activating adjacent C-H bonds towards chlorine radicals* whereas no such effect is found9 in the reaction with methyl radicals. In some cases the deactivating R. F. Bridger and G. A. Russell J. Amer. Chem. Soc. 1963 85 3754. A. F. Trotman-Dickenson Quart. Rev. 1953 7 198. W. A. Waters Tetrahedron 1959 Suppl. No. 3 151. A. L. Buley and R. 0. C. Norman Proc. Chem. SOC. 1964,225. G. A. Russell J. Amer. Chera. SOC. 1956 78 1047. C. Walling ‘Free Radicals in Solution’ John WiIey and Sons New York 1957. C. C. Price and H. Morita J. Amer. Chem. SOC. 1953,75 3686. 250 Davidson influence of an electron-withdrawing substituent may be offset by its ability to stabilise the incipient radical. There is very little difference in the activation energies for the abstraction of hydrogen from methane and methyl chloride by chlorine radicals whereas the reactions with methylene dichloride and chloro- form show a large increase.l Thus as the number of chlorine substituents increase the deactivating influence increases relative to the stabilising effect.Solvent Effects This subject has been reviewed recently.1° Abstraction from Alkanes and Aralkanes The reactions of these compounds with methyl phenyl t-butoxyl p e r o ~ y ~ ~ ~ chlorine,ll bromine and trichloromethyl radicals12 have been studied and many of the results collated in a recent paper.3 The attack upon the C-H bonds becomes more selective as the stability of the radical increased. The selectivity of the chlorine radical is13 unusually high in the reaction with aralkanes.This was found to be due to an interaction between the aromatic system and the radical. Extrapolation of the results obtained from reactions run at high dilution in inert solvents to infinite dilution overcame this effect and normal selectivity values were obtained. It has been shown14 that in the bromination of aralkanes the reactivity of the C-H bonds is the same whether the source of radicals be N-bromosuccinimide or bromine. This has been taken as further evidence in support of the in which the bromine radical is the species which abstracts hydrogen. Abstraction from Cyclic Alkanes The reactivities of a number of cyclic alkanes towards methyl,16 phenyl? chlorine,17 trichloromethyl,l* and trichloromethylsulphonyl18 radicals have been determined.The order of reactivity was C,<C,<C,<G. In some cases very little difference in reactivity between cyclopentane and cyclohexane was de- tected.l* The order of reactivity parallels the stability of their respective cyclo- alkyl radicals which suggests that the amount of C-H bond rupture in the transition state allows some relief of ring strain. The reactivities of the cyclo- alkanes C to C3 towards t-butoxyl radicals decreaselg as the ring becomes lo E. S. Huyser Adv. Free Radical Chem. 1965 1 77. l1 G. A. Russell and R. C. Williamson jun. J. Amer. Chem. SOC. 1964 86 2357. l2 (a) E. S . Huyser J. Amer. Chem. Soc. 1960 82 394; (b) R. E. Pearson and J. C. Martin J. Amer. Chem. SOC. 1963 85 3142. l3 G. A. Russell A. Ito and D. G. Hendry J. Amer. Chem. SOC. 1963 85 2976.l4 C. Walling A. Rieger and D. D. Tanner J. Amer. Chem. SOC. 1963 85 3129; see also G. A. Russell and K. M. Desmond J. Amer. Chem. SOC. 1963 85 3139. l5 J. Adam P. A. Gosselain and P. Goldfinger Nature 1953 171 704; Bull. SOC. chim. belges 1956 65 523. l6 A. S. Gordon and S. R. Smith J. Phys. Chem. 1962 66 521. l7 G. A. Russell J. Amer. Chem. SOC. 1958 80 4997. l9 C. Walling and P. S . Fredricks J. Amer. Chem. SOC. 1962 84 3326. E. S. Huyser H. Schimke and R. L. Burham J. Org. Chem. 1963,28,2141. 25 1 Quarterly Reviews smaller. Cyclopropyl C-H bonds were particularly unreactive. It has been sug- gested that the high degree of ‘s-character’ of the bonds does not favour the appropriate polar contributions to the transition state which an attack by an acceptor radical requires.The high reactivity of methyl groups attached to a cyclopropane ring is believed to be due to the ability of the ring to stabilise the incipient radical. The reactivity of bridgehead C-H bonds towards chlorine radicals has been investigated. These bonds are particularly unreactive in norbornane20 but in bicycl0[2,2,2]0ctane~~ they showed some reactivity. The difference in reactivity probably reflects the greater flexibility of the latter ring system which allows the development of ionic character in the transition state. Abstraction from Cycloalkenes The high degree of reactivity of allylic C-H bonds may be ascribed to the ability of the double bond to stabilise polar structure which contribute to the transition state and also the incipient radical. This stabilisation effect will be affected by the amount of orbital overlap between the double bond and the reaction centre.The amount of overlap in cycloalkenes will be regulated by the geometry of the ring system. The planar cyclopentene system for instance is much more reactive than the skewed cyclo-octene system.22 The reactivities of a number of cyclo- alkenes (C to C,) towards phenyl? bromine,l* and peroxy-radicals22 have been determined and it was found that the order of reactivity varied accord- ing to the radical used. There has not been a satisfactory explanation for these observations. Tntramolecular Hydrogen Abstractions of Synthetical Importance The Hofmann-Loeffler Barton and Related Reactions.-The essential steps of these reactions are shown in Scheme 1. Photon or thermal tf ,c i 7 c c c ,= o( ‘c’ 8 C homolysis of X-A bond X’ + ’+ 4 X ’ c ,c C Scheme 1 + A = RNH; X = Halogen; Hofmann-Loeffler reaction A - 0; A = 0; X - Halogen NO; Barton reaction X = H; Lead tetra-acetate used to generate alkoxyl radicals (ref.24) 2o E. C. Kooyman and G. C. Vegter Tetrahedron 1958 4 382. 21 A. F. Bickel J. Knotnerus E. C. Kooyman and G. C. Vegter Tetrahedron 1960 9 230. 22 D. E. Van Sickle F. R. Mayo and R. M. Aluck J. Amer. Chem. SOC. 1965,87,4824. 23 J. Gresser A. Rajbenback and M. Szwarc J. Amer. Chem. SOC. 1961 83 3005. 24 K. Heusler and J. Kalvoda Angew. Chem. Internat. Edn. 1964 3 525. 252 Davla3on The synthetical importance of the reactions lies in the fact that the substitution which is brought about at the 6 carbon atom may be very difficult to obtain by other chemical methods.The specific attack at the 6 carbon atom is due to the necessity for a six-membered cyclic transition state. The steric requirements of the reaction have been reviewed.M The product of the Hofmann-Loeffler reaction a &halogenoamhe has been found to cyclise readily under basic conditions to give pyrrolidine derivatives. The scope of the reaction has been reviewed.26 The Barton reaction which has been reviewed,26 has been extensively used in the steroid field. Barton and his co-workers have illustrated the usefulness of the reaction in their brilliant syntheses of aldesterone2’ and connessine.28 Radical Cyclisations.-Julia has recently reviewed29 his work on the peroxide- initiated intramolecular cyclisation of ethylenic a-cyano-esters. The reaction has been employed to synthesise cyclopentane cyclohexane indane decalin and some tricyclic systems.Di-t- butyl peroxide L The propogation reaction is the radical (b). Role of Hydrogen Abstraction in abstraction of hydrogen from ester (a) by the Homolytic Aromatic Substitution The substituted cyclohexadienyl radical which is an intermediate in homolytic aromatic substitution may dimerise disproportionate or have a hydrogen atom abstracted by a radical with formation of a substituted aromatic product Scheme 2 (Scheme 2).30 If the last reaction is relatively inefficient the yield of dimerisation and disproportionation products is increased. For example it has been observed31 in the reaction between amino-radicals and substituted benzenes that the yields of tetrahydrobiaryls are high unless the substituents have strong electron-releas- 25 M.E. Wolff Chem. Rev. 1963 63 55. 26 M. Akhtar Adv. Photochem. 1964 2,263. 27 D. H. R. Barton and J. M. Beaton J. Amer. Chem. SOC. 1960 82 2641. 28 A. L. Nussbaum F. E. Carlon E. P. Oliveto E. Townley P. Kabasakalian and D. H. R. Barton Tetrahedron 1962,18 373. 29 M. Julia Rec. Chem. Progr. 1964 25 3. 30 For a more detailed account see G. H. Williams ‘Homolytic Aromatic Substitution’ Pergamon Press London 1960. 31 F. Minisci R. Galli and M. Cecere Tetrahedron Letters 1965 No. 51 4663. 253 Quarterly Reviews ing properties which facilitate hydrogen abstraction by the acceptor amino- radicals. Bryce-Smith and his c o - ~ o r k e r s ~ ~ observed in the reaction between phenyl radicals and isopropylbenzene that the lower the concentration of radicals in the solution the lower the yield of phenylated cumenes.An increased yield of the latter product is obtained when the concentration of radicals is increased which favours the hydrogen-abstraction process. Hey and his co-workers,= in their detailed studies of the reactivity of sub- stituted benzenes towards substituted phenyl radicals found that electron- releasing substituents activate the molecule towards acceptor radicals and electron-withdrawing substituents favour attack by donor radicals. It was found,= for instance in the reaction between substituted phenyl radicals and toluene that substituents which increased the donor properties of the radical brought about an increased amount of hydrogen abstraction from the benzylic C-H bonds ; i.e. the electron-releasing properties of the methyl group deactivated the nucleus to attack by donor radicals.Hydrogen Abstraction from Various Compounds Carboxylic Acids and Deriva- tives.-There are a few examples of abstraction of hydrogen from a carboxyl group. Norman and his co-workers have by e.s.r. spectroscopy that methyl radicals are produced in the reaction between hydroxyl radicals and acetic acid. It has been suggested that the methyl radicals were formed by de- carboxylation of acetoxyl radicals. The formation of benzyl radicals in the reaction between hydroxyl radicals and phenylacetic acid would appear to be an analogous Decarboxylation has also been found3' to occur when mandelic acid is &irradiated. In the reaction between N-iodosuccinimide and trifluoroacetic acid trifluoromethyl iodide and succinimide are formed.38 It is believed that a protonated succinimide radical is responsible for the hydrogen abstraction.There are several examples which illustrate that C-H bonds adjacent to carboxyl or ester groups are deactivated towards acceptor radical attack. For example hydrogen is abstracted preferentially from the /?-position of propionic acid by hydroxyl radicals;35 the chlorination of cyclobutanecarboxylic acid produces only 2- and 3-chloro-compounds ;39 ethoxyl radicals abstract hydrogen from the benzylic position in p-phenylpropionic acid;40 and methoxyl radicals react at the benzylic position in benzyl dimethylmalonate.4° The deactivating effect decreases as the electron affinity of the radical decreases. Methyl radicals for instance abstract hydrogen from C-H bonds adjacent to the ester groups of 32 J.McDonald Blair D. Bryce-Smith and B. W. Pengilly J . Chem. SOC. 1959 3174. 33 D. H. Hey S. Orman and G. H. Williams J. Chem. SOC. 1965 101 and previous papers cited. 34 J. K. Hambling D. H. Hey and G. H. Williams J. Chem. SOC. 1962 487. 35 W. T. Dixon R. 0. C. Norman and A. L. Buley J. Chem. SOC. 1964,3625. 36 H. Fischer 2. Naturforsch 1965 20a 488. 37 S. A. Barker J. S. Brimacombe and D. J. Milner J. Chem. SOC. (0,1966,511. 38 D. D. Tanner J. Amer. Chem. SOC. 1964 86 4674. 39 W. A. Nevill D. S. Frank and R. D. Trepka J. Org. Chem. 1962,27,422. 40 H. C. McBay 0. Tucker and A. Milligan J. Org. Chem. 1954 19 1003. 254 Davidson benzyl dimethylmalonate:O In the reaction between the t-butoxyl radicals and propionic acid attack on the a-C-H bonds was sixteen times more rapid than that on the fl-C-H bonds.4l This suggests that the t-butoxyl radical has a rela- tively low electron affinity.Hey and his co-workers foundd2 that the alkylation of diethyl malonate and related compounds can be brought about by the peroxide-initiated addition of the parent esters to olehs. Aldehydes.-The radical-initiated decarbonylation of aliphatic aldehydes has been extensively studied. The t-butoxyl radical is particularly effective as a chain initiator X* or RCHO -+ RC = O+R. + co R. X- initiator radical. The alkyl radical produced by decarbonylation of the intermediate acyl radical continues the chain reaction by abstracting hydrogen from an aldehyde group. Stable radicals (e.g. Me2CC0,Me) are less efficient as initiators since an initiator and an acyl radical combine so terminating the ~hain.4~ The amount of decarbonylation in such reactions has been found4 to be significantly in- creased by the addition of a thiol which acts as a hydrogen carrier i.e.a thiyl radical abstracts hydrogen from an aldehyde group efficiently and a hydrogen from the thiol group so formed is easily abstracted45 by an alkyl radical. The generation of acyl radicals from aldehydes by means of free radicals has found application in the study of the stabilities of cyclopropyl,4s bridgehead4' and other cyclic radicals:* radical rearrangement^,"^ and intramolecular acylation~.~~ The peroxide-initiated reaction of aldehydes with carbon tetra- chloride51 or sulphuryl chloride52 produces acyl halides. and conjugated unsaturated aldehydes are relatively stable and do not normally decarbonylate.Difluoramino-= and diphenylphosphino-radicals56 have been found to The acyl radicals obtained by hydrogen abstraction from A. L. J. Beckwith Austral. J. Chem. 1960 13 244. 42 J. C. Allen J. I. G. Cadogan B. W. Harris and D. H. Hey J. Chem. SOC. 1962 4468. 43 E. F. R. Harris and W. A. Waters J. Chem. SOC. 1952 3108. 44 K. E. J. Barnett and W. A. Waters Discuss. Faraday SOC. 1953 14 221. 45 L. H. Slaugh J. Amer. Chem. SOC. 1959 81 2262. 46 D. I. Schuster and J. D. Roberts J. Org. Chem. 1962 27 51. 47 D. E. Applequist and L. Kaplan J. Amer. Chem. SOC. 1965 87 2194. 48 J. W. Wilt and A. A. Levin J. Org. Chem. 1962 27 2319. 49 R. Kh. Freidlina Adv. Free Radical Chem. 1965 1,211. 50 W. H. Urry D. J. Trecker and H. D. Hartzler J. Org. Chem.1964 29 1663; R. Dolon Y. Chretien-Bessiere and H. Desalbres Compt. rend. 1964 258 603. 51 S. Winstein and F. H. Seubold J. Amer. Chem. SOC. 1947 69 2916. 52 M. Arai Bull. Chem. SOC. Japan 1965,38,252. 53 F. R. Rust F. H. Seubold and W. E. Vaughan J . Amer. Chem. Soc. 1948,70 3258. 54 R. C. Petry and J. P. Freeman J. Amer. Chem. SOC. 1961 83 3912. 55 R. S. Davidson R. A. Sheldon and S. Trippett unpublished results. 255 Quarterly Reviews abstract hydrogen from the aldehyde group. Amides were obtained as products from the reaction with the nitrogen radical whereas the phosphorus radical gave products derived by the addition of diphenylphosphine to the aldehyde. Ethers-C-H bonds adjacent to the oxygen atom in ethers are particularly re- active towards free radicals and this is due to the ability of the oxygen atom to stabilise polar contributions to the transition state and also the incipient radicals.C-H bonds flanked by two oxygen atoms show an enhanced rea~tivity.~~ The reactions of ethers with methyl,56 phenylY2 alk0xy1,~~ peroxy-,ll a~yloxy-,~* and halogen radicals5D have been investigated. The decomposition of acyl peroxides in ethers is particularly rapid. Apparently the intermediate a-alkoxyl- alkyl radicals attack the peroxide which brings about induced decomposition. The abstraction of hydrogen from cyclic ethers has received attention.s7 Usually if the intermediate a-alkoxylalkyl radical cannot combine with another radical to form a stable product ring-opening occurs with the formation of a carbonyl compound In the reaction between t-butyl hypochlorite and substituted ethylene oxides ring-opening and the formation of chloroethylene oxides were foundlS to be in competition.Abstraction of hydrogen from acetals by the t-butoxyl radical was foundso to produce both esters and aldehydes RC(OCH,R) + RCO.OCH,R' + RCH,. + RCH*O.CH,R' + RCHO In an analogous reaction with 2-methoxytetrahydrofuran esters were formed.s1 In the reaction between bromine radicals and benzyl ethers alkylated benz- aldehyde bromohydrins were formed,@ which readily decomposed to benzalde- hyde and alkyl bromide. In the reactions of nuclear-substituted benzyl ethers with bromines3 and peroxy-radicals,ll the substituents had little effect upon the reaction. C. L. Aldridge Z. B. Zachry and E. A. Hunter J. Org. Chem. 1962,27,47. 57 T. J. Wallace and R. J. Gritter J.Org. Chcrn. 1962 27 3067. 58 W. E. Cass J. Amer. Chcm. Soc. 1947 69 500. 59 H. Bdhme and A. DCirries Chern. Bcr. 1956,89 723. 6o L. P. Kuhn and C. Wellman J. Org. Chem. 1957 22 774. 62 R. L. Huang and H. K. Lee J. Chem. SOC. 1964 5957 5963. 63 R. E. Lovins L. J. Andrews and R. M. Kelfer J. Org. Chem. 1965.30 1577. E. S. Huyser J. Org. Chcm. 1960 25 1820. 256 The oxidation of benzaldehyde acetals by N-bromosuccinimide has been founds4 to produce alkyl benzoates and alkyl bromides. t-Butoxyl radicals have been useds5 to abstract hydrogen from benzaldehyde cyclic acetals. The intermediate radical decomposes to give an ester. If the ether ring contains substituents a mixture of isomeric esters is obtained. The reactions of a number of nuclear-substituted dibenzyl ethers with t-butoxyl and bromine radicals have been reported.66 The observation that a larger polar effect operates in the reaction with the more reactive t-butoxyl radical than with the less reactive bromine radical is rather surprising.The oxidation of substituted dihydroisobenzofurans by N-bromosuccinirnide has been founds7 to produce substituted o-benzdialdehydes. MeC Me0 H,C @$H2 -0 (ii) Hydrolysis CHO Alkylphenols and Alky1thiophenols.-In reactions between these compounds and free radicals substitution competes with hydrogen abstraction. There is evid- enceSs that t-butoxyl radicals abstract hydrogen from anisole with the formation of phenoxymethyl radicals which under thermal conditions (140") undergo substitution reactions and under photolytic reactions dimerise. The reaction between t-butoxyl radicals and methyl phenyl sulphide producedag thiophenoxy- methyl radicals which dimerised under thermal conditions (1 40") and were deme t hylated under pho t ol ytic conditions.Me / (QH (i)N-Bromoruccinimide Me t Alcohols Phenols and 0ximes.-A wide variety of radicals e.g. alkyl,7O a l k ~ x y l ~ ~ hydr~xyl,~~ nitrogen,73 and sulphur74 radicals abstract hydrogen from alcohols. In general the a-C-H bond (D 90 kcal./mole) and not the 0-H bond (D 108 kcal./mole) is attacked. This was demonstrated by Kharasch Rowe and Urry70 who studied the reaction between diacetyl peroxide and alcohols containing a-C-D bonds. In all cases a high yield of deuteriomethane (CH,D) was obtained. The inertness of the 0-H bond was shown by studying the analogous reaction with deuterio-t-butyl alcohol [(CH,),COD] in which deuter- iomethane was not produced.E.s.r. spectroscopic evidence has since been 64 S. 0. Lawesson and T. Busch Arkiv Kemi 1961 17,421. 65 E. S. Huyser and 2. Garcia J. Org. Chem. 1962 27 2716. 66 R. L. Huang H. H. Lee and S. H. Ong,J. Chem. SOC. 1962,3336; R. L. Huang H. H. Lee and M. S. Malhotra ibid. 1964 5947. 67 J. Blair W. R. Logan and G. T. Newbold J. Chem. SOC. 1956,2443. H. B. Henbest J. A. W. Reid and C. J. M. Stirling J. Chem. SOC. 1961 5239. 69 H. B. Henbest J. A. W. Reid and C. J. M. Stirling J. Chem. SOC. 1964 1220. 'O M. S. Kharasch J. L. Rowe and W. H. Urry J. Org. Chem. 1951,16,905. 71 D. C. Neckers A. P. Schaap and J. Hardy J. Amer. Chem. SOC. 1966,88 1265. 72 L. J. Leyshon and D. H. Volman J . Amer. Chem. SOC. 1965 87 5565. 73 A.Schiinberg and A. Mustafa J. Amer. Chem. SOC. 1951 73 2401; B. R. Cowley and W. A. Waters J. Ckem. SOC. 1961 1228. 74 M. Nakasaki J. Chem. SOC. Japan 1953,14,405. 257 3 Quarterly Reviews obtained76 which suggested that attack occurs at the a-C-H bonds. It was found76 that acyl peroxides undergo induced decomposition in alcohols (cf. the reaction with ethers). A few casFs have been reported of abstraction from the hydroxyl group. The Me& = CH radical abstracts77 deuterium from deuter- ioethanol (EtOD). It has also been found7s that some cyclopropanols on oxida- tion give products consistent with the formation of alkoxyl radicals. Recently it has been suggested70 that diphenylphosphino-radicals abstract the hydroxylic hydrogen by way of an initial attack upon the oxygen atom.The formation of aryloxy-radicals from phenols has recently been reviewed.e0 Some of the reagents which bring about these oxidations have been founde1 to oxidise oximes with the formation of iminoxyl radicals. N-H P-H and S-H Bonds.-These bonds have been shown to be particularly reactive towards free radicals. This undoubtedly reflects the ability of the hetero-atom to stabilise polar structures which contribute to the transition state. The N-H bond. Alkyl,sa p e r o ~ y ~ ~ and stables4 free radicals have been found to abstract hydrogen. The N-H bond in diphenylamine is 3.3 times more reactive towards phenyl radicals than the benzylic C-H bonds in diphenylmethane. In some abstraction it is possible that radical ions are intermediates e.g. in the reaction between primary aromatic arnines and diphenylpicrylhydrazyl radicals.e4 The P-H bond.The free radical-initiated reaction in which a primary or secondary phosphine is added to an olefin has been reviewed.s5 Peroxides acyl peroxides and azonitriles act as initiators. The P-H bond in diphenylphosphine is thirty times more reactive towards phenyl radicals than diphenylamine. The stable diphenylamino-radical can abstract hydrogens6 from P-H bonds. The S-H bond. Abstraction of hydrogen from S-H bonds has received particular attentione7 since it is the first step in the addition of thiols to olefins. Peroxides and azo-compounds are efficient initiators. The reactivity of S-H bonds towards alkyl radicals has been commented uponp4 (p. 255). Difluoramino-,88 dipheny- lamino-,8s and triphenylmethyl radicalsso abstract hydrogen from thiols.J. F. Gibson D. J. E. Ingram M. C. R. Symons and M. G. Townsend Trans. Furuday SOC. 1957 53,914. 76 E. S. Huyser and C. J. Bredeweg J. Amer. Chem. SOC. 1964,86,2401. 77 F. Glockling J. Chem. SOC. 1956,3640. 78 C. H. Depuy G. M. Dappen and J. W. Hausser J. Amer. Chem. SOC. 1961,83 3156. 79 R. S. Davidson R. A. Sheldon and S. Trippett J. Chem. SOC. (0 1966 722; Chern. Comm. 1966,99. 8oA. I. Scott Quurt. Rev. 1965,19 1. 8s0. A. Chaltykyan and E. N. Atanasyan Doklady Akad. Nauk Armyun 1953 17 65; Chem. Abs. 1955,49 11374d. 84 J. C. McGowan T. Powell and R. Raw J. Chem. SOC. 1959 3103. 85 G. Sosnovsky ‘Free Radical Reactions in Preparative Organic Chemistry’ Collier- Macmillan Ltd. 1964 London p. 153. 86 H. Low and P. Tam Tetrahedron Letters 1966 No. 13 1357. 87 See ref. 85 p. 62. 88 J. P. Freeman A. Kennedy and C. Colburn J . Amer. Chem. SOC. 1960,82 5304. J. R. Thomas J. Arner. Chern. Soc. 1964 86 1446. J. Hutton and W. A. Waters J. Chem. SOC. 1965 4253. T. J. Wallace J. J. Mahon and J. M. Kelliher Nature 1965 206 709. A. Schonberg and G. Aziz J. Org. Chem. 1957,22 1697. 258

ISSN:0009-2681    

DOI:10.1039/QR9672100249

出版商:RSC    

年代:1967

数据来源: RSC

摘要:

Grignard Reagents. Compositions and Mechanisms of Reaction By E. C. Ashby ATLANTA GEORGIA 30332 U.S.A. SCHOOL OF CHEMISTRY GEORGIA INSTITUTE OF TECHNOLOGY 1 Introduction Without doubt one of the most fascinating and fundamental problems in organic chemistry today concerns the composition of Grignard reagents in ether as solvent. A different but closely related problem involves the mechanism or mechanisms by which these Grignard reagents react with organic functional compounds such as ketones nitriles etc. The solution to the second problem can only be forthcoming when the solution to the first problem is in hand. The importance and basic nature of these problems are well recognised yet there has been no full scale assault for any length of time by anyone to bring about a complete solution.Many workers have become interested in the composition problem from time to time but efforts have been somewhat sporadic and no real progress has appeared until recently. In retrospect it appears almost inconceivable that such a fundamental prob- lem could have remained in so confused a state for so long while much more seemingly complex problems were being solved every day. For example the structures and absolute configurations of cholesterol and chlorophyll are known and yet organic chemists do not know the composition in solution of a class of compounds that bears the simple empirical formula RMgX. Part of the problem seems to be that it takes an organic chemist to recognise the importance of this problem a physical chemist to make the type of measurements which could be informative in what has become a complex physical-chemical problem and an organometallic chemist with the background and experience to handle studies involving such sensitive organometallic compounds.Each person having faced the Grignard problem soon recognises his weakness in at least one of the above areas and is discouraged to make more than a token contribution in what has turned out to be a very elusive and complex problem. Another problem seems to have been that many of the contributions that have been made were a result of preconceived notions and hence the conclusions were not always justified by the results. As a result workers entering this area become quickly confused because of so many conflicting conclusions presented by so many different contributors. Only if one studies the approximately 300 contributions in this area over and over again until all the conflicting facts are well recognised can one then decide what to believe and what might be poor work that should be repeated.If one is 259 Quarterly Reviews not hopelessly discouraged by such a large mass of data which does not appear to prove much of anything then one can begin to try to decide first just what is the problem and secondly what is the best thing to do to solve it. 2 Composition of Grignard Compounds A. Early History.-In order to appreciate this problem fully one should go back to the very beginning of this story-the year 1900. Grignard then a gradu- ate student was working in the laboratory of Professor Barbier at the University of Lyon in France. His problem was to optimise conditions for what is known today as the Barbier reaction.The specific system involved is shown in (1). ,. OH Grignard thought the intermediate in this reaction to be RMgX and thus con- cluded that yields might be improved by preparing this compound first and then adding it to the ketone. He found that alkyl halides do react readily with mag- nesium in ether as solvent and that the resulting reaction mixture reacts with aldehydes and ketones in higher yields than when the Barbier procedure is used to produce the corresponding addition pr0duct.l Grignard represented the com- position of the reaction product of an alkyl halide and magnesium in ether as RMgX and represented the reaction of this reagent with ketones as a simple addition reaction. For Grignard's great discovery and subsequent development of this finding he was awarded the Nobel prize in Chemistry in 1912.The first serious suggestion since Grignard's initial one concerning the com- position of Grignard compounds in ether solution was made by Baeyer and Villiger.2 These workers suggested representing Grignard compounds in ether solution as an onium compound. The structure suggested for methylmagnesium iodide is shown in (I). Et Mg Me Et 1 (I) 'd / \ Et ,Me 0 E( h g I (D) This suggestion was followed by a similar one from Grignard3 depicting the onium compound differently as 0. Although Standnikov4 appeared to have evidence to support Grignard's suggestion it was later proved by both Gorskijs and Chelhtzev and Pavlov* that the evidence presented was not unequivocal. Shortly thereafter Thorp and IV.Grignard Compt. rend. 1900 130 1322. * A. Baeyer and V. Villiger Ber. 1902,35 1201. 9V. Grignard Compt. rend. 1903 136 1260. 5A. 1. Gorksij J. Russ. Phys.-Chem. SOC. 1912 44 581. 6V. V. Chelintzev and B. V. Pavlov J. Russ. Phys.-Chem. SOC. 1913,45,289. G. L. Standoikov J. Russ. Phys.-Chem. SOC. 1911,43 1235. 260 Ashby K a m 7 demonstrated conclusively that Grignard compounds could not be represented by the onium structure. They did this by demonstrating that the products of reactions (2) and (3) are not identical. While the onium com- (2) Ha0 Et,O + PhBr + Mg + C6H6 (60%) (3) Ha0 EtOPh + EtBr + Mg -+ C2H (99.7%) position was being debated Abegg* in 1905 suggested a polar composition for the Grignard reagent (R-[MgXl+) and even suggested the possibility of an equilibrium (4) to describe the system.2RMgX + R,Mg + MgX (4) Jolibois9 in 1912 was the fist to suggest what is referred to as the unsym- metrical dimeric structure (R,Mg.MgX,) to represent Grignard compounds in ether solution. This suggestion was based on the facts that (1) Grignard com- pounds in ether were believed to be dimeric (2) Et,Mg and Mg12 in ether had the same physical properties as a solution prepared from C2H,I and Mg and (3) electrolysis of EtMgI under certain conditions deposited magnesium at the cathode without evolution of gas. In rebuttal GrignardlO suggested that the observations of Jolibois could just as easily be explained by a symmetrical dimeric (RMgX) composition. Thus the controversy began concerning the description of Grignard compounds by the symmetrical or unsymmetrical dimeric structure.ll Although GrignardlO and Terentievl reported dimeric association for methyl- magnesium iodide Meisenheimer and S~hlichenmaierl~ later showed that over a wide concentration range the molecular association of this compound varies with the concentration.This information was partly the reason for a sudden surge of interest in explaining the composition of Grignard compounds by what is known today as the Schlenk equilibrium. Schlenk and Schlenk14 found that essentially all of the halogen as MgX could be removed from certain Grignard compounds in ether solutions by the addition of dioxan. Based on this information and the new association data they suggested the equilibria ( 5 ) and (6) to explain the composition of Grignard reagents. 2RMgX + R2Mg + MgX2 RgMg + MgX + R2Mg.MgX2 (5) (6) L.Thorp and 0. Kamm J. Arner. Chem. SOC. 1914 36 1022. R. Abegg Ber. 1905 38 4112. P. Jolibois Compt. rend. 1912 155 353. loT. Zerevitinov Ber. 1908 41 2244. 11 M. S. Kharasch and Otto Reinmuth ‘Grignard Reactions of Nonmetallic Substances’ Prentice-Hall Inc. New York 1954 p. 103. l3 J. Meisenheimer and W. Schlichenmaier Ber. 1928 61 B 720. l4 W. Schlenk and W. Schlenk jun. Ber. 1929 62 B 920. A. Terentiev 2. anorg. Chem. 1926 156,73. 261 Quarterly Reviews Assuming that the equilibrium composition of (5) could be determined by the precipitation of the MgX from the mixture by dioxan Schlenk16 used this method to determine equilibrium values for several Grignard compounds. Further determinations of equilibrium compositions by the same method were reported by several other workers.l6,l7 A few years later in 1937 Noller and White18 cast doubt on the validity of this method to determine equilibrium com- positions by showing that the amount of MgX precipitated as the bis-dioxan complex is a function of time.In 1950 Kullman19 confirmed the original findings of Noller and White that the amount of MgX precipitated was a function of time and thus if an equilibrium did exist it was being shifted. For example Kullman showed that on treating ethylmagnesium bromide with dioxan 55 70 or 93 % of the total MgBr was precipitated depending on the mode of addition and the length of time the mixture was stirred. An argument was presented by Aston and Bernhard20 in an attempt to show that methylmagnesium iodide in dibutyl ether contains no dimethylmagnesium.Their evidence was based on the difference in relative rates and heats of reaction of solutions of methylmagnesium iodide and dimethylmagnesium with acetone and ethyl acetate. Further support included a comparison of the pyrophoric nature of a solution of methylmagnesium iodide and dimethylmagnesium. Clearly this conclusion was based on the lack of interaction between (CH,),Mg and MgI or else the comparisons were not valid. Dialkylmagnesium compounds are now known to react rapidly with magnesium halide so that one cannot validly compare these compounds on the bases cited. Because of the limited solubility of MgCI in diethyl ether (ca. 1 x 10-3~) a lack of precipitation of MgCl from alkylmagnesium chlorides was cited as an indication of the Schlenk equilibrium ( 5 ) lying predominantly to the left.21 Once again these conclusions are of questionable validity on the basis that it is now known that alkylmagnesium chlorides are dimeric even at low concentration thus lack of precipitation of MgCI is inconclusive.Many other reports could be cited attempting to prove that there is or is not a Schlenk equilibrium or that the Schlenk equilibrium (5) lies to the right or left. They fall in the category of the above cited cases namely the conclusions are not valid either on the basis of information known at the time or more recently. In order to simplify further discussion only those contributions which were once thought to be very im- portant will be considered. B. Ionic Nature.-There exists considerable evidence concerning the electrolytic nature of Grignard compounds in ether solution.Joliboiss was the first to report W. Schlenk Ber. 1931 64 B 734. l6 A. C. Cope J. Amer. Chem. SOC. 1934,56,1578; C. R. Noller and F. B. Hilmer ibid. 1932 54,2503. l7 P. D. Bartlett and C. M. Berry J. Amer. Chem. SOC. 1934 56 2683. l8 C. R. Noller and W. R. White J. Amer. Chem. SOC. 1937 59 1354. l9 R. Kullman Compt. rend. 1950 231 866. 2o J. G. Aston and S. A. Bernhard Nature 1950 165 485. 21 C. R. Noller and D. C. Raney J. Amer. Chem. Sac. 1940,62 1749. 262 Ashby that Grignard compounds in ether are electrically conducting. This fact has been confirmed by several other research groups2 and a great deal of infomation has been acquired through such experimentation. Both Evans and P e a r s ~ n ~ ~ and Zei1,24 from conductance measurements have pictured the ionisation of Grignard compounds in ether solution as producing large mobile highly asso- ciated anions and small relatively immobile ether co-ordinated cations.DCcombe and D ~ v a 1 ~ ~ had suggested earlier from much less information that phenyl- magnesium bromide be represented by the ionic composition represented by (111). These workers suggested that if the composition below is correct the Mg++Ph2MgBr2(Et,0)212- @I) anionic magnesium should be replaceable by a less reactive metal such as zinc. Reaction of methyl iodide and magnesium-zinc alloy in diethyl ether produced what was thought to be Mgtf-[Me2Zn12(Et20)2]2-. Elemental analyses were con- sistent with the above formula; however more important were the observations that (1) hydrolysis of the product produced methane zinc iodide and magnesium hydroxide and (2) electrolysis produced magnesium at the cathode and zinc at the anode.Although not conclusive the evidence is consistent with the suggestion and an earlier one by Evans and that the products of elec- trolysis of ethylmagnesium bromide can be represented by equation (7). Since the initial flurry of conductance and electrolysis experiments in the 1930’s little has been done along these lines. Dessy and Jones2’ confirmed the most important finding initially reported by Evans and Lee that in the electro- lysis of n-butylmagnesium bromide magnesium-containing species migrate to both the anode and the cathode. Although the number of possible ionic species described by all the contributors in this area are numerous it appears that the most prevalent simple species in solution are RMg+ and RMgX2-.The work of Evans and Lee which shows an increase in conductivity with concentration at concentrations greater than 0-5 M indicates that the system is past the conductivity minimum. Thus it appears that in the higher concentration ranges studied (+03 M) one is dealing with ion-pair formation. Applying the Harned and Owen symbolism to this system (1) RMg+ and X- would be single ions (2) RMgX would be an ion pair (3) RMgX2- and (RMg),X+ would be ion triplets (4) (RMgX) would be a quadruple ion and so on. RMgX and (RMgX) should have zero conductance and hence could be in large concen- tration. The concentration of the ionic species (1 and 3) cannot be very great 22 J. M. Nelson and W. V. Evans J . Amer. Chem. SOC. 1917,39,82; F. Kondyrew Ber.1925 58 B 459; L. W. Gaddum and H. E. French J Amer. Chem. SOC. 1927,49,1295. 23 W. V. Evans and R. Pearson J. Amer. Chem. SOC. 1942 64 2865. 24 W. Zeil 2. Elektrochem. 1952,56 789. 26 J. Decombe and C. Duval Compt. rend. 1938,206 1024. 26 W. V. Evans and F. H. Lee J. Amer. Chem. SOC. 1933,55 1474. 27 R. E. Dessy and R. M. Jones J. Org. Chem. 1959,24 1685. 263 Quarterly Reviews owing to the low conductances reported for Grignard compounds in ether solution.27 Since high ionic mobilities would be expected in a medium of low viscosity such as diethyl ether high conductances would be expected if dissocia- tion into ions were extensive. Therefore ionic species of the type suggested are probably present only to a small extent in solution. Although these species may not be important in describing the composition of Grignard compounds owing to their low concentration they could be important in describing the mechanism of Grignard reactions since reaction rate is not only a function of concentration but of reactivity of the various species in solution.C. Recent Work.-From the late 1930's to the late 1950's little was done to relieve the confusion concerning the nature of the composition of Grignard compounds in diethyl ether. Some workers used the RMgX formulation to represent Grignard compounds some used the R,Mg.MgX formulation some gave up the solution to the problem as hopeless. It was not until 1957thatDessy and his co-workers28 presented the first convincing evidence permitting a clear- cut choice between the RMgX and R,Mg.MgX formulations. These workers found no exchange between 28MgBr2 and Et,Mg and presented evidence that an equimolar mixture of MgBr and Et,Mg has the same characteristics as the Grignard reagent prepared from ethyl bromide and magnesium.Thus it was concluded (1) that alkyl exchange does not take place in ether solution (2) the RMgX species does not exist in solution and therefore Grignard compounds are best represented by the structure first suggested by Jolibois namely R,Mg.MgX,. Owing to the work of Dessy and his co-workers the representation of Grignard compounds by the R,Mg-MgX formulation was widely accepted. Thus M0sher,2~ BeCker,3O and their co-workers and others postulated that the reaction of Grignard compounds with ketones involves a six-centre transition state (IV) in which the Grignard compound is represented by the unsymmetrical dimer.Although there has been much difficulty in rationalising all of the kinetic data in terms of reaction order and reaction mechanism with respect to an attacking dimeric species certainly the mechanism presented was the most logical at the time and well accepted. In spite of all the experiments to determine the nature of Grignard compounds in ether solution none appeared significant beside the isotopic labelling experi- ments. Several additional contributions by Dessy3l and his co-workers after 1957 only served to support the equivalency of a mixture of diethylmagnesium- 28 R. E. Dessy G. S. Handler J. H. Wotiz and C. A. Hollingsworth J. Arnet. Chem. SOC. 1957,79 3476. 29 J. Miller H. S. Mosher G. Grigorian J. Amer. Chem. SOC. 1961 83 3966.30 W. M. Bikales and E. I. Becker Canad. J. Chem. 1962,41 1329. 31 R. E. Dessy J. Org. Chem. 1960 25,2260. 264 Ashby magnesium bromide and the Grignard compound prepared from ethyl bromide and magnesium. Attempts by J. D. Roberts and his c o - w ~ r k e r s ~ ~ ~ ~ to verify the conclusions of Dessy and his co-workers by examination by nuclear magnetic resonance (n.m.r.) of Grignard solutions and mixtures of R,Mg and MgX were not fruitful. However Kirrmann Hamelin and Hayes,% by crystallisation studies involving several Grignard compounds showed that the crystalline fractions always contained more MgX than R,Mg. Unable to isolate RMgX they concluded that the Grignard composition is expressed best by a mixture of R,Mg and MgX and associated forms. D. R,Mg-MgX Composition Questioned.-The composition of Grignard compounds represented by the unsymmetrical dimer R,Mg-MgX, although accepted for several years was first questioned by Ashby in 1961 and again in 1963 on the basis of evidence establishing the existence of RMgX species in tetrahydrofuran ~ 0 1 u t i o n .~ ~ ~ ~ ~ ~ ~ It was shown that in tetrahydrofuran alkyl exchange in Grignard compounds does take place and that the species RMgX does indeed exist in solution. These conclusions were based on three observations. (1) Grignard compounds are monomeric in tetrahydrofuran over a wide con- centration range (2) crystallisation of Grignard compounds produced the compounds of stoicheiometry RMg,X and MgR in essentially quantitative yield and (3) in tetrahydrofuran EtMg,Cl is soluble and its molecular associa- tion was determined to be 0.62.(Molecular association = observed molecular weight -+ formula weight = i.) Since the specific conductance of ethylmagnesium chloride in tetrahydrofuran was shown to be very small the association factor 0-62 indicates extensive dissociation of EtMg,Cl into two species in tetra- hydrofuran according to (8). Thus the presence of EtMgC1 Et,Mg and MgCI EtMgSC1 -+ EtMgCl + MgCl in tetrahydrofuran was demonstrated leading to the conclusion that ethyl- magnesium chloride in tetrahydrofuran is best described by the formulations first reported by Schlenk,15 namely (9). The concentration of RMgX was 2RMgX $ R,Mg + MgX (9) suggested to be significant on the basis that MgCl is soluble in tetrahydrofuran 3e M. S. Silver P. R. Shafer J. Eric Nordlander C.Ruchardt and John D. Roberts J. Amer. Chem. SOC. 1960 82 2646. 33 G. M. Whitesides F. Kaplan K. Nagarajan and John D. Roberts Proc. Nut. Acad. Sci. 1962,48 No. 7 1112. 34 G. M. Whitesides F. Kaplan and John D. Roberts J. Amer. Chem. SOC. 1963 85,2167. 35 A. Kirrmann R. Hamelin and S . Hayes Bull. SOC. Chim. France 1963 1395. 36 Discussion of paper presented by R. E. Dessy Organometallic Symposium Vancouver British Columbia 1961. 37 Panel discussion concerning Grignard Compound Composition H. S. Mosher E. I. Becker A. Frey and E. C. Ashby 1 st International Organometallic Symposium Cincinnati Ohio 1963. 38 E. C. Ashby and W. E. Becker. J. Amer. Chem. Soc, 1963 $5 118. Quarterly Reviews only up to 0 . 5 ~ ~ whereas a 2.0~ solution of ethylmagnesium chloride in tetra- hydrofuran deposited no precipitate after 3 months.It was also reported33 that a study of the molecular association of ethyl- magnesium chloride in diethyl ether over a concentration range indicated asso- ciated species in solution. This fact complicated the simple equilibrium suggested for Grignard compounds in tetrahydrofuran in that in diethyl ether solution dimeric species would have to be included. A\ /X\ R-MCJ,~,M~-R S 2RMgX RzMg 4- M9Xz E>%,M9 W) (VO (VI 0 (VIII) (10) The equilibrium (10) was suggested on the grounds that the difference between tetrahydrofuran and diethyl ether is one of degree rather than of kind and one would only expect to find a difference in the association reflected by the difference in basicity of the two solvents. The structures for diethyl ether (IX) and tetrahydrofuran (X) are drawn only to focus attention on the structural similarity of the two compounds.lsolation of RMg,X3 compounds from diethyl ether solution was also reported but these compounds would not redissolve in diethyl ether and therefore their molecular association could not be studied. For this reason the possibility of these compounds being represented by struc- ture (XI) does exist although structure (XI) would appear more reasonable. Although the existence of RMgX species had been demonstrated in tetra- hydrofuran it was not until later that evidence became available to indicate that such was also the case in diethyl ether. Almost simultaneously two labora- t o r i e ~ ~ ~ ~ ~ s ~ ~ reported that ethylmagnesium bromide in diethyl ether is mono- meric at low concentration ( < O .~ M ) . Vreugdenhill and Bl~mber$~ with unusual care determined the molecular association of ethylmagnesium bromide diethyl- magnesium magnesium bromide and a mixture of diethylmagnesium and magnesium bromide as 1-00 f 0.02 1-00 f 002 1.13 f. 0.04 and 1-05-1.06 respectively. They concluded from these measurements that at low concentra- tions there esists no equilibrium of the type represented by (11). The results of Et,Mg + MgBr + Et,Mg.MgBr (1 1) Ashby and Smith37,40 showed clearly that in diethyl ether organomagnesium bromides and iodides are monomeric at low concentrations (ca. 0 . 0 5 ~ ) and associate only at higher concentrations whereas organomagnesium chlorides are associated even at low concentration. 39 A. D. Vreugdenhill and C. Blomberg Rec. Trav. chim.1963,82,453. 40 E. C. Ashby and W. B. Smith J . Amer. Chem. SOC. 1964,86,4363. 266 Ashby 2.2 I I I I i 1 0 0.1 0.2 0.3 0.4 Molarity Fig. 1 Association of Grignard compounds in diethyl ether. The difference between chlorides and bromides-iodides can be explained easily in terms of inductive and steric effects. These results establish without question that in diethyl ether the composition R,Mg.MgX is not adequate to describe Grignard compounds. Since Grignard compounds are normally employed in solution between 0.1 and 0 . 5 ~ concentration it appears that one must consider reaction of both monomeric and associated species at these con- centrations. The conclusion drawn from this work was that the composition of Grignard compounds in diethyl ether should be represented by a monomer- dimer equilibrium (2 M + D) and more specifically could be expressed by the equilibria (10) although no concrete evidence for the existence of the RMgX species was presented.The reaction between R,Mg and MgX to form 2RMgX was suggested40 to proceed via a mixed alkyl-halogen bridge intermediate (XIIT) as in (12). This is not unreasonable since alkyl-bridged compounds are well known in the chemistry of related aluminium compounds. The species (XIII) would not be expected to have more than a transitory existence since dimeric structures containing halogen atoms in both bridging positions should be thermodynamically favoured. E. Consequence of Association Data.-The knowledge of how the degree of association of Grignard compounds in diethyl ether depends on concentration has turned out to be very important.This information has been used40 to over- come evidence that was thought to argue against the presence of RMgX species in diethyl ether solution. For example Evans and Maher?l on the basis of n.m.r. studies of Me,Mg and ‘MeMgI’ in ether suggested that the complex Me,Mg-MgI, a dimeric species is stable in highly dilute solution. (The quota- O1 D. F. Evans and J. P. Maher J. Chern. SOC. 1962 5125. 267 Quarterly Reviews tion marks as in ‘RMgX’ are used not to indicate the RMgX species but merely the Grignard compounds formed from RX and Mg.) The molecular weight studies (Figure l) however indicate that ‘MeMgI’ is essentially mono- meric at the concentration of the measurements (0.06~). Since according to Evans and Maher Me,Mg and MgI interact the product of this interaction must be MeMgI.The .r-values reported by Evans and Maher for dimethyl- magnesium in diethyl ether show an appreciable concentration-dependence whereas the concentration-dependence for methylmagnesium iodide is very small over a considerable concentration range. These authors attributed this difference to the dissociation of polymeric dimethylmagnesium species on dilution. This appears unlikely since diethylmagnesium is monomeric at this concentrati~n.~~ It appears that these data support the conclusion that ‘MeMgI’ does not contain an appreciable amount of Me,Mg rather than the opposite conclusion drawn by the authors. Other studies by Evans and Maher in diethyl ether showed that the proton resonance spectra of ‘EtMgBr’ closely resembles that of Et,Mg for the concen- tration range 0.040-0.296~ (with respect to the ethyl group).They interpreted this as indicating the absence of a significant amount of EtMgBr and supporting the representation of Grignard compounds in solution as R,Mg-MgX,. How- ever the molecular weight studies indicate that at the highest concentration studied (0-296~) about 64% of the ‘EtMgBr’ exists as dimeric species either (V) or (VIII) or both; n.m.r. was therefore unable to distinguish between Et2Mg and structures (V) or (VIII). Since each of these structures contains Mg bonded to both carbon and halogen it follows that n.m.r. could not be expected to distinguish between Et,Mg and EtMgBr. It was later shown40 that in tetrahydrofuran ‘EtMgCl’ and Et,Mg exhibit essentially identical n.m.r. spectra yet in tetrahydrofuran the RMgX species is present in solution.Apparently n.m.r. is unable to distinguish between Et2Mg and EtMgCl in tetrahydrofuran or diethyl ether solution. Recently Fraenkel and his co-worker~~~ reported little if any difference between the n.m.r. spectra of ‘MeMgI’ and MeLi. Once again this points out the inability of n.m.r. in certain cases to differentiate between two different chemical species. As a second example Dessfl reported dielectric constants for 0.1 39~-MgBr in Et20 to which was added varying amounts of Et,Mg. The plot of dielectric constant against the ratio Et,Mg/MgBr showed a distinct break at the 1 :1 ratio indicating compound formation. The dielectric constant for the 1 :1 mixture was identical with that measured for ‘EtMgBr’ at the equivalent concentration indicating that the mixture and the ‘EtMgBr’ contained the same species.According to Figure 1 the i-value for ‘EtMgBr’ (at 0.278~) is 1.43 so that about 60 wt. % of the ‘EtMgBr’ is present as monomer. Since this monomeric ‘EtMgBr’ is a compound rather than a mixture of Et,Mg and MgBr, it must consist pre- dominantly of the species EtMgBr. The existence of the species EtMgBr in dilute ether solution is also indicated 43 G. Fraenkel D. Adams and J. Williams Tetrahedron Letters 1963,767. 268 AsMy 0 0.2 06 1.0 1.4 1.8 2.2 Et,Mg/MgBr ratio Fig. 2 DieIectric constant as a function of the ratio Et,Mg/MgBrs by the data of Vreugdenhill and Bl~mber$~ who reported the association factors listed earlier at concentrations of 10-3-10-2~. A possible interpretation of their results follows.If the ‘EtMgBr’ had consisted of Et2Mg + MgBr, it should have had an i-value of 1.06 [calculated from the i-values for Et2Mg and MgBr,] as was actually found for the mixture. The i-value of 1-00 determined for ‘EtMgBr’ did not change within 72 hr. indicating that the monomeric Grignard compound must have consisted of EtMgBr and that the latter had no measurable tendency to disproportionate to Et2Mg + QMgBr,. The i-value of 1.05-1.06 for Et2Mg + MgBr indicates that very little reaction took place. Perhaps this reaction is quite slow particularly at these low concentrations unless a suitable catalyst is present. The argument based on the data of Vreugdenhill and Blom- berg is not strong and is only valid if the small association differences reported by these authors is real. The possible existence of the symmetrical dirneric structure (V) for ‘EtMgBr’ was argued40 by comparison of the association phenomena for ‘mesityl magnesium bromide’.‘Mesitylmagnesium bromide’ in diethyl ether solution showed at least as much association as EtMgBr’ (Figure 1) in the same medium If the dimeric species in diethyl ether solution exists as the unsymmetrical dimer it seems that ‘mesitylmagnesium bromide’ should be more dissociated than ‘EtMgBr’ over a wide concentration range because of its greater steric require- ment. Since ‘mesitylmagnesium bromide’ shows a slightly higher degree of asso- ciation over the same concentration range as ‘EtMgBr’ it was suggested that the dimer formed would be predominantly the symmetrical one which could only originate from monomeric RMgX species.Stucky and Rundle recentlf3 found by X-ray studies that phenylmagnesiurn bromide is composed of units containing the phenyl group a bromine atom 43 G. D. Stucky and R. E. Rundle J . Amer. Chem. SOC. 1963 85 1002. 269 Quarterly Reviews \-/Me (XV) and two diethyl ether molecules bonded tetrahedrally to a single magnesium atom (XVI). Although the structure of phenylmagnesium bromide in the solid state cannot be extrapolated to solution without causing some concern this work did add to the evidence supporting the existence of RMgX species in diethyl solution of Grignard compounds. As another example Dessy and his ~ o - w o r k e r s ~ ~ ~ ~ ~ described experiments in which equimolar amounts of M a r (labelled with radioactive magnesium) and Et,Mg were dissolved in diethyl ether to give a solution 1 .0 ~ in Mg. In all the experiments in which 25Mg was used as the tracer complete exchange occurred between Et,Mg and 25MgBr2. Evidently the Et,Mg and 25MgBr reacted according to the Schlenk equilibrium to form EtMgBr + Et25MgBr. On the other hand when 2sMg was used as the tracer only 6-10% exchange was reported to occur even after contact times as long as 36 hr. This difference in behaviour of 25Mg and ,*Mg was attributed to impurities in the 25Mg which was concluded to have catalysed the exchange. The possibility that some trace impurity in the ,*Mg may have inhibited exchange was not considered. The association studies in diethyl ether solution have shed light on the validity of much of the work reported in the 1930’s concerning the determination of equilibrium constants for the Schlenk equilibrium.Noller and Raney21 and othersl69l7 attempted to determine equilibrium compositions of Grignard com- pounds by shaking Grignard solutions with excess’of MgCl in order to pre- cipitate any MgCl present in what other workers described as super-saturated solutions. The equilibrium composition of n-butylmagnesium chloride at con- centrations ranging from 0.4 to 1 . 8 ~ shows that in spite of the low solubility of MgCI,-ether complex in ether only 3-10% of the halogen originally present precipitated as magnesium chloride. Precipitation with dioxan indicates that at least 88% of the halogen should be present as MgCl in solution. These workers concluded either that precipitation with dioxan does not give a correct picture of the composition of Grignard solutions and the reagent is almost entirely in the form RMgX or that the solubility of MgCl is increased from essentially zero to the extent of approximately one mole per mole of R,Mg.Of course the solubility of MgCI can be increased from essentially zero to one mole of MgCI per mole of R,Mg. This however does not necessarily mean conversion into RMgX but could represent association to R,Mg.MgX,. It appears that MgCl is held in solution by both of the equilibria suggested by Schlenk. 44 R. E. Dessy and G. S. Handler J. Amer. Chem. SOC. 1958 80 5824. 270 Ashby The lack of acceptance of (14) was based on the fact that both the dioxan p r e cipitation method and the MgCl supersaturation method produced equilibrium values independent of Grignard concentration. Since it is now clear that alkyl- magnesium chlorides (those studied) are associated even at low concentration it does not appear that these conclusions are valid or that compositions of Grignard compounds can be determined by precipitation of MgC12 incurred by addition of dioxan or by adding an excess of MgCl,.Since equilibria are involved this approach can only be valid if it is possible to freeze the equilibria and prevent RMgX from producing more R,Mg and MgX with time as the MgX is recovered as a precipitate. Although the arguments presented by Ashby and Smith3's40 were somewhat revolutionary in 1964 they are not nearly so revealing now since Dessy and his c o - ~ o r k e r s ~ ~ have repeated the exchange experiments they initially reported in 1957. In their most recent publication these workers report statistical exchange in the system 2sMgBr2 and Et,Mg with two out of the three grades of magnesium used.Exchange took place with the purest forms of magnesium used (triply sublimed and Grignard grade turnings) whereas no exchange was observed with the most impure sample @ow atomised shot). Soon thereafter Cowan Hsu and Roberts4s reported statistical exchange in the system 25MgBr and Et,Mg. There now appears no doubt that there is exchange in Grignard solutions and that RMgX is an intermediate in this exchange even if not present in appreciable concentration. Thus it appears that the equilibria represented by (10) is a reason- able representation of Grignard compounds in diethyl ether solution. It is entirely possible that small traces of some metal impurity (in p.p.m.) in the several grades of magnesium used for the exchange experiments could have catalysed or inhibited the exchange of R,Mg and MgX,.It is also possible as suggested by Vre~gdenhill,4~ that the exchange is catalysed by oxygen and therefore the extent to which oxygen was excluded in the experiments becomes very important. Thus one must be cautious in generalising the exchange pheno- mena. The conclusions are however that exchange has been observed on mixing R,Mg and MgX under the best possible conditions obtainable. Although it may be possible in the future to obtain non-exchange results with ultrapure magnesium in a completely oxygen-free atmosphere the point remains that under the normal conditions of Grignard formation alkyl exchange takes place and RMgX species in some concentration are present.Even though the position for the monomer-dimer equilibrium is known it would be of more importance to know the concentration of RMgX with respect 45 R. E. Dessy S. Green and R. M. Salinger Tetrahedron Letters 1964 1369. 46 D. 0. Cowan J. Hsu and J. D. Roberts J . Org. Chem. 1964 29 3688. 47 A. D. Vreugendhill opinion expressed at the 3rd International OrganometalIic Symposium Madison Wisconsin September 1965. 271 Quarterly Reviews to R,Mg + MgX,. The best way of course to establish equilibrium constants would be to observe the concentration of the undisturbed species in solution by n.m.r. or infrared analysis. Unfortunately attempts by Roberts3 and Fraenke14 and their colleagues to do this by n.rn.r. analysis were not successful. Attempts by Salinger and Moshefi8 to detect RMgX species in diethyl ether solution by high-resolution infrared analysis also were not successful.However Mosher and his co-workers were able to detect the RMgX species in tetrahydrofuran solution reaffirming the existence of a Schlenk equilibrium (9) established earlier.38 Mosher and his co-workers assigned an equilibrium constant K = 4 for ethylmagnesium bromide in tetrahydrofuran indicating statistical dis- tribution. F. Evidence for RMgX Composition.-The fist direct observation of RMgX species in diethyl ether other than as an intermediate was provided recently.49 The same report also claimed that RMgX is the initial species formed when an alkyl halide and magnesium react. It was argued that the difference between the composition of Grignard compounds in tetrahydrofuran (9) and diethyl ether (10) is explained by the difference in basicity of the two solvents.Ebullio- scopic measurements show a monomer-dimer equilibrium in diethyl ether but only monomer present in tetrahydrofuran. Thus tetrahydrofuran co-ordinates with magnesium more strongly than diethyl ether and a stable halogen-bridge compound is not formed. The exchange of alkyl groups in either tetrahydrofuran or diethyl ether can be explained by an intermediate mixed alkyl-halogen bridge structure of the type suggested earlier.40 Grignard compounds co-ordinated to a more basic solvent than diethyl ether or tetrahydrofuran might not form such intermediates if the magnesium orbitals are strongly bonded to the basic solvent. Thus disproportionation might be prevented and the initial species formed by reaction of RX and Mg could be isolated.y 3 - M3-X R- ilk (XVll) In order to test this hypothesis ethylmagnesium bromide was prepared from ethyl bromide and Mg in triethylamine. The reaction product (XVIII) was fractionally crystallised into seven fractions. Each fraction had a Mg :Br :N ratio of 1.O:l.O:l.O within experimental error. (Although C,H,MgBr crystallises 48 R. M. Salinger and H. S. Mosher J. Amer. Chem. SOC. 1964,86,1782. 49 E. C. Ashby J Amer. Chem. Soc. 1965,87,2509. 272 Ashby from triethylamine as the bis-solvate the monosolvate is isolated on drying under high vacuum.) Molecular association measurements of the crystallised fractions in triethylamine at 35" showed the presence of only monomeric species over a wide concentration range.Because of the highly insoluble nature of MgBr in triethylamine and the soluble nature of Et,Mg precipitation of MgBr from solution would have occurred if an unassociated mixture of these two products were present. These data prove that the reaction product is a single species and not a mixture. The product EtMgBr.NEt, did not disproportionate in boiling triethylamine during 24 hr. nor was it formed by redistribution of Et2Mg and MgBr in triethylamine. Thus the indication is that EtMgBr is the initial product formed by the reaction of EtBr and Mg and in triethylamine solution the composition can be represented by this single structure. Similarly the existence of RMgX in diethyl ether solution was established. When a diethyl ether solution of ethylmagnesium bromide prepared from EtBr and Mg in diethyl ether was added slowly to a large rapidly stirred volume of triethylamine EtMgBr.NEt was isolated in over 90% yield by fractional crystallisation.In the Schlenk equilibrium (16) magnesium bromide is the 2EtMgBr + Et,Mg + M a r 2 (16) strongest Lewis acid and diethylmagnesium is the weakest. It is not reasonable that triethylamine would solvate only the species of intermediate acidity (RMgX). The fact that no MgBr,.Et,N was isolated although it is the most insoluble of the possible products leads to the conclusion that the rate of solvation is greater than the rate of equilibration and therefore in diethyl ether solution ethyl- magnesium bromide consists mainly if not entirely of RMgX species (as monomer or dimer). Since RMgX is the only product isolated in triethylamine when RX and Mg are allowed to react it must be the initial and only magnesium compound formed in the reaction.If any other magnesium compounds such as R,Mg or MgX had been formed they would have been solvated and protected from redistribution. Diethyl ether and tetrahydrofuran being more weakly basic solvents than triethylamine are not sufficiently strong complexing agents to prevent the disproportionation of the initially formed RMgX and therefore in these solvents redistribution takes place. The position of the equilibria (10) in diethyl ether was further verified by Smith and Becke9O in a more exact and convincing way than previously re- ported?' They reported that 0 . 1 ~ solutions of diethyl magnesium and mag- nesium bromide reacted instantaneously in 1 :1 molar ratio on mixing to evolve 3.6 kcal/rnole of heat.Since only monomeric species are involved at this concen- tration and heats of dilution were found to be negligible this evolution of heat was attributed to the formation of EtMgBr. Figure 3 represents the heat evolved on addition of MgBr to one mol. of Et2Mg. 50 M €3. Smith and W . E. Becker Tetrahedron Letters 1965 3843. 273 Quarterly Reviews Moles MgBr Fig. 3 Heat of reaction ns a function of the ratio Et2Mg/MgBr Equilibrium constants for Grignard compounds R,Mg 4- MgX + 2RMgX Grignard compound K (in Et,0)50 K (in THF)51 EtMgBr 480 4 EtMgI 630 PhMgBr 55 PhMgI 15 The equilibrium constants shown in the Table were obtained by measuring the heat evolved when the R,Mg compound was added to the MgX and vice versa. The curves of AH plotted against MgBr,/Et,Mg or Et,Mg/MgBr were superimposable.These results indicate that in diethyl ether solution the equili- brium R,Mg + MgX + 2RMgX lies predominantly to the right. The results are in agreement with the earlier work involving quenching experiments in tri- eth~lamine.~~ Similar thermochemical experiments were conducted by Smith and Becker51 in tetrahydrofuran. In this solvent the instantaneous reaction of R,Mg with MgX occurs with absorption of heat. The position of the equilibrium R,Mg + MgX + 2RMgX is not nearly as far to the right as it is in diethyl ether. The equilibrium constants (Table) do not differ greatly from the value K =4 which corresponds to statistical distribution in agreement with the earlier work of Salinger and Moshefl8. In the case of the system Et,Mg-MgCI, the curve of AH against MgX,/R,Mg is different from the curve of AH against R,Mg/MgX,.This was shown to indicate the existence of appreciable quantities of the species EtMg,CI in tetrahydrofuran solution. This result is 61 M. B. Smith and W. E. Becker Tetrahedron 1966 22 3027; 1967 23 in press. 274 Ashby consistent with the work of Ashby and Be~keI.3~ who obtained crystals of the compound EtMgaC1 by fractional crystallisation of the corresponding Grig- nard reagent from tetrahydrofuran solution. Recent n.m.r. studies by Roberts and his c o - ~ o r k e r s ~ ~ and Fraenkel and his c o - ~ o r k e r s ~ ~ on Grignard reagents in diethyl ether have established the ionic nature of the C-Mg bond and rapid inversion at the a-carbon atom. The n.m.r. work was not informative in establishing the nature of the associated species in solution or the magnitude of equilibrium constants for the Schlenk equilibrium.Evidence was presented in favour of a bimolecular mechanism to explain the alkyl transfer from one magnesium atom to another involving inversion at the a-carbon atom by means of a mixed alkyl-halogen bridge intermediate (XVIII). Thus some evidence is now available to verify alkyl transfer in the Grignard system by a mixed-bridge intermediate suggested earlieFO strictly on an intuitive basis. In 1965 Hashimoto and his co-workersM reported results which appeared to confuse the evolving picture of Grignard compound composition in solution. They reported that EtMgBr in diethyl ether is associated in solution past the dimer state (i = 2-4 at 1 .0 ~ ) and then decreases in association below the dimer stage with an increase in concentration (i = 1-6 at 2 . 0 ~ ) . They also report that EtMgBr in tetrahydrofuran is associated at low concentration (i = 1.6 at 0 . 1 ~ ) and then decreases in association to a monomer only at high concentration (i = 1.0 at 1 . 2 ~ ) . It does not seem reasonable that association phenomena should decrease with an increase in concentration and indeed this does not appear to be the case. A recent report by Ashby and Walker55 sustantiates earlier data and con- c l u s i o n ~ ~ ~ concerning association of Grignard compounds in both diethyl ether and tetrahydrofuran. They were attempting to establish the exact nature of the associated magnesium species in solution For example if 2 molecules of the RMgX species associate to a dimer in solution the resulting structure can be represented by (XIX) (XX) or (XXI).Although structure (XIX) should be the most stable there exists no evidence anywhere to substantiate such a choice. Likewise association of the R2Mg and MgX species can be described by means of a double halogen-bridge a double alkyl-bridge or a mixed alkylhalogen- bridge species. Since RMgX is the predominant species in diethyl ether only the association of this species will be considered here. The data (Figure 4) show quite clearly that association of magnesium through the halogen in M a r and MgI (MgCI is insoluble in diethyl ether) is much 52 G. M. Whitesides and J. D. Roberts J. Amer. Chem. SOC. 1965 87 4878. 5a G. Fraenkel and D. T. Dix J. Amer. Chem. SOC. 1966,88,979.54 H. Hashimoto T. Nakano and H. Okada J. Org. Chem. 1965.30 1234. 55 E. C. Ashby and F. Walker J. Organometallic Chem. 1967,7 P17. 275 Qwrterly Reviews Concentration (M) Fig. 4 Association of Grignard species at high concentrations. stronger than through the alkyl group as in EtzMg or Me,Mg. Since methyl and ethyl are optimum bridging groups and dimethylmagnesium and diethyl- magnesium are only weakly associated in diethyl ether it appears that associa- tion of the Grignard compounds is predominantly through the halogen. The association of all of the compounds studied can be explained purely on inductive grounds i.e. association is proportional to the positive character of the mag- nesium atom. The association of some Grignard compounds past the dimer stage points out the need to consider the possibility that different Grignard compounds are asso- ciated differently.For example it appears that PhMgBr is linearly associated whereas the ethyl Grignard compounds appear to be either linearly associated (curvature due to deviation from ideality at high concentration) or are trimeric. On the other hand the association of t-butylmagnesium chloride appears to level off at the dirner stage. The representation of associated molecules of PhMgBr does not appear to be tenable on the basis of structure (XIX) since association of this type would predict gross curvature of the association line at i = 2 owing to the change in the nature of the bonding which must take place at this point. Structures (XXII) and -1) are compatible with the type of association exhibited by PhMgBr.Since solid magnesium halides and Grignard compounds hold one molecule of the 276 ether tightly per magnesium atom structure (XXDI) is preferred. The X-ray data reported by Stucky and Rundle@ would also favour structure (XXIII). In contrast to the report of Hashimoto Nakano and Okada,54 these results confinn an earlier reports8 that ethylmagnesium bromide is monomeric in tetrahydrofuran showing only a low degree of association at high concentration. The anomalies exhibited in the data of Hashimoto et al. could be due to any of several problems which make measurements on such sensitive systems very difficult. G. Grignard Reagents prepared in Hydrocarbon Diluent.-Some investigations into the preparation of Grignard compounds in hydrocarbon solvent have been reported by Bryce-Smith and his co-workers,56 who reported that aryl and alkyl primary chlorides bromides and iodides react with magnesium in hydrocarbon solvents at elevated temperature (> 100"~).The solvents employed were deca- or tetra-hydronapthalene and paraffin oil while the choice of optimum reaction temperature was a compromise of many factors such as sufficient reaction rate Wurtz-type coupling Friedel-Crafts-type alkylation of aromatic solvents etc. The products of these reactions are white amorphous non-volatile solids. The solubility of these products is quite small in hydrocarbons the solubility in- creased with the size of the halogen (Cl<Br<I) the length of the R group and the type of solvent (aromatic > aliphatic). An interesting feature of this work in addition to providing reaction of an alkyl halide and magnesium in a non-polar solvent in up to 95 % yield is that in freshly prepared solutions analysis has detected compounds of the empirical formula R,Mg,X.Unfortunately the concentration is low and solids precipitate from solution on standing until products of indiscriminate empirical formulae remain. It has been suggested that R,Mg,X is associated in hydrocarbon solution and that predominantly MgX precipitates on standing. The solid products produced in the reaction simply appear to be a mixture of R,Mg MgX, and associated species thereof. Thus one is not sure that the name 'Grignard Reagent' should be applied to the solid mixture of compounds produced when an alkyl halide is allowed to react with magnesium in hydrocarbon diluent.Zakharkin and his co-~orkers~~ later expanded the work in hydrocarbon diluent demonstrating higher yields (70-95 %) for most primary alkyl and aryl halides using diluents such as dodecane 2,2,3-trimethylpentane and sometimes no diluent at all. Essentially all of the magnesium-containing product was pre- cipitated from solution and no attempt was made to determine the nature of the products other than to determine the empirical formula. More recently it has been found58 that Grignard compounds can be prepared in aromatic hydrocarbons as a solvent rather than a diluent. In this connection 2~ solutions of several typical alkylmagnesium halides (e.g. C,H,MgBr) were prepared. The difference in this report and those of Bryce-Smith and Zakharkin D. Bryce-Smith and G. F. Cox J. Chern. SOC.1961 1175. 57 L. I. Zakharkin 0. Yu. Okhlobystin and B. N. Strunin TetruhedronLetters 1962,631. 58 E. C. Ashby and R. Reed J. Org. Chem. 1966,31,985. 277 Quarterly Reviews and their colleagues is that one mole of triethylamine as a complexing agent was employed per mole of magnesium so that the actual product was RMgX.NEt,. It is interesting that use of one mol. of diethyl ether or tetrahydrofuran did not impart such solubility. The composition of the Grignard compounds reported in benzene solvent is believed to be described by the singular species RMgX.NEt for reasons presented earlieP (fractional crystallisation and molecular associa- tion). H. Conclusions concerning the Composition of Grignard Reagents.-It appears that the following conclusions are justified. (1) RMgX species exist in diethyl ether tetrahydrofuran and triethylamine.(2) The compositions in the above solvents are represented as follows Diethy2 ether. Monomeric species consist largely of RMgX along with a small amount of R2Mg + MgX2. Association is extensive above about 0 . 3 ~ etc.+Trimer+Dimer+2RMgX+R2Mg + MgX,+Dimer+Trimer+etc. Tetrahydrofuran. Relatively little association occurs in this solvent. The pre- vailing monomeric species consist of substantial quantities both of RMgX and of R,Mg + MgX2. The Grignard compound EtMgCl also contains the species EtMg2CI in appreciable quantity. Triethylamine. Only one species (RMgX) is observed for simply alkylmagnesium bromides and chlorides. In solutions of alkylmagnesium iodides and aromatic Grignard compounds the situation is not so simple.59 Hydrocarbon.The work of Bryce-Smith and his co-workers indicates that when RX and Mg are allowed to react in hydrocarbons as diluents a mixture of insoluble organomagnesium and inorganic magnesium compounds is produced (R2Mg + MgX and highly associated combinations). This mixture of insoluble compounds will behave similarly to ether solutions of Grignard compounds toward some organic functional compounds. The species R,Mg,X is soluble to a small extent; however the solutions are not stable and precipitation results in time. Benzene-soluble Grignard compounds have been prepared but the solubility and composition (RMgXaNR’,) is a result of complex-formation with 1 mole of a tertiary amine for each mole of organomagnesium compound. 3 Mechanisms of Grignard Reactions The mechanism of the addition of Grignard reagents to ketones has been a s9 E.C. Ashby and T. Bickley unpublished work. 278 Ashby subject of much controversy for some years.Bo The first serious mechanistic suggestion (1 8 19) concerning this reaction was made by Swain and Boyles61 in 1951. They suggested the mechanism to be third order; first order in ketone and second order in Grignard reagent. This mechanism was criticised after 1957 when the Grignard reagent was reported to have the R2MgMgX2 structure. Almost simultaneously Miller et aZ.,62 Bikales and Becker,B3 and Hamelin and Hayess4 suggested a bimolecular mechanism involving one molecule of ketone and one molecule of unsymmetrical dimeric Grignard reagent (20). In the years ensuing since 1957 an apparent impasse was reached concerning the two different mechanistic descriptions of this reaction.On the one hand Mosher Becker and Hamelin upheld the conclusion that the mechanism of addition of Grignard compounds to ketones is best represented in terms of a six-centre transition state involving one molecule of ketone and one molecule of Grignard dimer R,Mg.MgX,. On the other hand Anteunise5 has held that the mechanism originally suggested by Swain and Boyles correctly represents the course of this reaction. It is clear that neither group of workers possessed kinetic data which were consistent with their suggested mechanism. This was because although each of these workers recognised that the mechanism was complex the kinetic data were analysed as if the reaction was simple; more specifically as if the reaction should be either first or second order in Grignard reagent.The second-order kinetics reported by Bikales and Becker actually represented only the first 30 % of the reaction. After 30 % reaction the simple second-order plot showed a serious deviation. On the other hand Becker has criticised the kinetic data presented by Anteunis for several reasons the M. S. Kharasch and Otto Reinmuth Grignard Reactions of Nonmetallic Substances Prentice-Hall Inc. New York 1954. 61 C. G. Swain and H. B. Boyles J. Amer. Chem. SOC. 1951 73 870. 62 J. Miller G. Gregarion and H. S. Mosher J. Amer. Chem. SOC. 1961 83 3955. 63 N. M. Bikales and E. I. Becker Canad. J. Chem. 1962,41 1329. 64 R. Hamelin and S. Hayes Compt. rend. 1961 252 1616. 6s M. Anteunis J. Org. Chem. 1961 26 4214 1962 27 596; Bull. SOC.chim. belges. 1964 73 655. 279 Quarterly Reviews most important of which is that the actual data presented when calculated correctly and plotted do not substantiate a third-order mechanism. In the mean- time Mosher quoted Swain in a person communication footnotes2 saying that Swain no longer maintains that the Grignard-ketone mechanism proceeds by third-order kinetics. Thus until recently it appeared that there were two distinct views concerning the mechanism of this reaction; on the one hand Mosher Becker and Hamelin maintain that the Grignard-ketone addition reaction proceeds by second-order kinetics and on the other hand Anteunis maintains that the reaction proceeds by third-order kinetics. While it appeared that the controversy had died down because of the difficulty in obtaining better results Smith and Sus6 suggested a quite different mechanism.This was that the ketone reacts with monomeric Grignard compound in a fast step (21) to produce a complex and then the complex rearranges in a first-order K K + G + C k C+P K = Ketone G = G-rignard C = Complex P = Product fashion (22) to produce the product. Smith and Su reported some very important observations. When methylmagnesium bromide and 2,4-dimethyl-4'-methyl- mercaptobenzophenone were allowed to react two bands were observed in the ultraviolet spectrum. One band (315 mp) was the v-n* carbonyl band of the ketone and the other band (355 mp) was attributed to the shifted carbonyl band due to the formation of a complex between ketone and Grignard reagent. The two bands were found to decrease concurrently establishing that an equilibrium does exist between ketone and complex.In a pseudo-first-order kinetic study disappearance of the ketone and complex took place in a first-order fashion. Attempts by Smith and Su to establish the mechanism presented earlier (eqns. 21 22) meet with much difficulty. The kinetic data were consistent with the suggested mechanism up to a Grignard concentration of 0 . 3 ~ . At higher concen- trations serious deviation from expected results in the plot of kobs against [GI was observed. Smith suggested an effect due to the medium to explain the deviation. This is not unreasonable and there is precedence for the mathematical forms (23) put forward. The weakness in this interpretation is that the major effect of [GI on the rate is exerted via the medium effect rather than on how G enters the mechanism.Thus the mechanism is not demonstrated in a positive way. 66 S. G. Smith and G. Su J. Amer. Chem. Soc. 1964,86,2750; 1966,88,3995 280 Ashby One problem which appears regularly in a study of this type is the uncertainty of the aggregation of Grignard reagents at the concentrations being studied. Realising this Smith and Su suggest an alternative mechanism at Grignard con- centrations > 0 - 3 ~ which approaches in essence the mechanism originally reported by Becker Mosher and Hamelin. As a matter of fact with the constants (Kl and K,) that Smith and Su use to fit their mechanistic scheme most of the reaction proceeds via the path involving reaction of the dimer G2.66 Smith and Su concluded their very detailed study by saying that many other reaction mechanisms could be written which with suitable choice of numerical parameters are consistent with their experimental results.Ashby and his co-workerss7 studied the kinetics of the reaction of benzo- phenone with methylmagnesium bromide in diethyl ether using concentrations of the latter so small that the reagent is predominantly monomeric and at the same time sufficiently greater than the ketone concentration that the concentra- tion of Grignard remains constant during a given experiment. Each reaction was then pseudo-first order; and the variation of its rate constant kobs was observed as the Grignard concentration [GI was varied from one experiment to another. The kinetic results were obtained by quenching individual samples of the reaction mixture at appropriate intervals and following the disappearance of ketone (250 mp) by ultraviolet analysis.The functional dependence of kobs on [GI was used as a test of possible mechanisms. If the reaction were second order overall (26) first order in Grignard and first order in ketone the simplest interpretation of such a law would be a Rate = k,[K][G] (26) birnolecular reaction (27). In such a case kobs should be given by (28). A plot k4 K + G + P of kobs against [GI was not linear. kobs = kJG1 (28) If the mechanism suggested by Smith and Su were correct then kobs is related to [GI by the expression (29). Quantitative adherence to this expression was tested graphically by use of eqn. (30). 67 E. C. Ashby R. Duke and H. M. Neumann J. Amer. Chem. SOC. 1967 89 1964. 281 Quarterly Reviews A plot of [GI against [G]/kobs produced a line through the data points with a negative slope which is meaningless physically.A mechanism (31)-(33) which did fit the experimental results is similar to that originally suggested by Swain. K G+K"C k C + G + P ' P + P + G In this case and a quantitative test of this mechanism was a plot of [GI against [GI2/kobs since Such a plot is shown in Figure 5. Fig. 5 Graphical test of suggested termolecular mechanism. Within experimental error the points fall on a line and the slope and intercept have signs required for meaningful interpretation. The mechanism suggested by Becker and others (36)-(38) is somewhat similar to the one suggested by Ashby and his co-workers except that the components are brought together in a different manner. Ashby Kg 2G + G2 K K+Gz--+C k3 C + P In this case k3K9[GI2 - k3K[GI2 1 + &K[GI2 - 1 + K[GI2 kobs = (39) This is true for conditions such that the concentration of dimer G2 is small with respect to that of monomer G.The test of the mechanism is whether or not a plot of kobs against / C ~ ~ / [ G ] ~ is hear which it is not. The fact that at low Grignard-to-ketone ratios the rate of reaction decreases markedly after 50 % of the available R groups are utilised can readily be explained by assuming that P' (eqn. 33) does not readily regenerate the active Grignard species as originally suggested by Swain. The detailed mechanism suggested by Ashby and his co-workers is shown in (40)-(43). R-c=O***Mg'' + R'MgX - R' 'X When benzophenone and methylmagnesium bromide were allowed to react in 1 :2 stoicheiometry the products were (XXIV) and (XXV).sa This does not exclude the possibility that (XXVI) may be an intermediate in the reaction since the alkoxymagnesiumalkyl and magnesium bromide were shown to redistribute rapidly to form the alkoxymagnesium halide.sa -.F;h Ph Ph-7-Mg-Br MeMgBr Ph-C-Mg Mc Me (XXIV) ow) Me (XXVI) 68 E. C. Ashby and R. Amott unpublished results. 283 Quarterly Reviews The ultimate test of the suggested mechanism is whether the rate behaviour in solutions where Grignard and ketone are in comparable concentration is consistent with the numerical values of k3 and K (eqns. 34 and 35) obtained under pseudo-first-order conditions. Ashby and his co-workers claim this to be so. Yh Th 7h Me Me Mg-Br Me + (XXVIII) Ph-7-OMgMe + MgBr L- Ph-y-Pv-Me = Ph-y-mBr (XXVI) 8 (XXVII) MeMgBr (43) It should be noted that the suggested mechanism is compatible with the existence of the Schlenk equilibrium but that the data cannot indicate which of the species (RMgX MgX, or R,Mg) may be involved in the equilibrium step or the rate-determining step of the mechanism if specificity indeed occurs.This is because the fraction of the Grignard reagent present in any one specific form is constant under the conditions employed. Thus if @ is the species of Grignard involved in the equilibrium step @I the species in the rate-determining step and FI and FII the corresponding fractions eqn. (44) follows. The functional depend- ence of kobs on [GI the total concentration of Grignard compound is thus no different than it would be if the Grignard existed solely in the RMgX form.From eqn. (34) it is clear that the reaction can exhibit simple second- or third- order kinetics depending on the magnitude of K[G]. If K[G] > > 1 then the reac- tion should exhibit simple second-order kinetics. If K[G] < < 1 then the reaction should exhibit simple third-order kinetics. Thus discussions as to whether the reactions is second- or third-order are meaningless unless the nature of the ketone and Grignard are known so that the magnitude of the equilibrium constant ( K ) is known. For example K can be made very small by placing a strong electron-withdrawing group in the para position of benzophenone or K can be made very large by placing a strong electron-donating group in that position. K>> I K K + G + C (45) In the case report by Ashby and his co-workers K is 1040; however [GI = lo4 therefore K[G] is neither large nor small compared with unity. Thus a more accurate picture of the mechanism is expressed in terms of molecularity rather than reaction order. Thus it appears that the mechanism of Grignard addition to ketones is much 284 better understood especially in terms of the suggested mechanism’s being con- sistent with the available data. The termolecular mechanism is consistent with the kinetic data and the isolated intermediates and also explains the much lower reactivity of the second 50% of the R groups in the Grignard compound. All the questions concerning Grignard compound composition and reaction mechanisms certainly have not been answered however it looks as if chem- ists are pointed in right direction. 285

ISSN:0009-2681    

DOI:10.1039/QR9672100259

出版商:RSC    

年代:1967

数据来源: RSC