摘要:

QUARTERLY REVIEWS APPLICATION OF ELECTRON DIFFRACTION TO THE STUDY OF THE CHEMICAL BOND IN CRYSTALS By B. K. VAINSHTEIN (INSTITUTE OF CRYSTALLOGRAPHY OF THE ACADEMY OF SCIENCES OF THE U.S.S.R. Moscow) ELECTRON diffraction the first experiments on which were carried out over thirty years ag0,~1~9~ has been the triumph of the wave theory of matter and is now a working method for investigating the structure of matter. It is used to study the structure of molecules in vapours and gases in the phase analysis of oxide and other films on the surface of solids in the study of the shape of submicroscopical crystals and their mutual orientation at epitaxy or in phase transitions. In the post-war years however electron diffraction has found an increasingly independent place as a method for the study of the mutual arrangement of atoms in a crystal.The principal method used for atomic structure i.e. X-ray analysis has in a number of cases proved to be insufficient and to-day along with it extensive use is made of electron- and neutron-diffraction methods. The development of structure analysis by electron diffraction is due primarily to Soviet a u t h o r ~ ~ - ~ and later to workers in many other countries. The improvement of experimental techniques as well as the evolution of the theory and practice of structure analysis by electron diffraction enable us to attack and to solve problems of two types. The first of these is the geometrical problem most important for structure analysis in general i.e. that of establishing the mutual arrangement of atoms their co- ordination lengths of chemical bonds and valency angles.Such work especially if carried out with sufficient accuracy enables us to answer many questions concerning the nature of the chemical bond in a given crystal. The second problem a more complicated and subtle one involves the valency state of atoms and of the field of forces between them and requires determination of the nature of their thermal motion from which Davisson and Germer Phys. Rev. 1927 30 705. Thomson Proc. Roy. Soc. 1928 A 117,600; 119 A 651 ; Thomson and Cochrane “Theory and Practice of Electron Diffraction,” London 1939. Tartakovsky Doklady Akad. Nauk S.S.S.R. 1928 A. Lashkarev “Diffrakzija Elektronov” (Electron Diffraction) Moscow 1933. Pinsker “Diffrakzija Elektronov” (Electron Diffraction) Moscow 1947; London Vainshtein “Structurnaija Elektronographija” (Structure Analysis by Electron 1953.Diffraction) Moscow 1956. 1 105 106 QUARTERLY REVIEWS conclusions may be derived as to the force constants of some particular bond etc. We first give a very brief survey of the experimental techniques of electron- diffraction analysis. The electronograph is a vacuum apparatus. The elec- trons from a hot tungsten wire are accelerated by a voltage of 50-10U kv (corresponding to an electron wave-length A of about 0.06-0.04 A). The electron beam formed with the aid of diaphragms and magnetic lenses is passed through a thin (- 10-6-10-5 cm.) film of the substance being investigated the electrons thus being scattered and yielding a diffraction pattern which is observed on a fluorescent screen and further recorded on photographic plates.Electron-diffraction cameras are usually constructed by the workers themselves but are also produced com- mercially (e.g. in the U.S.S.R. and Switzerland). Devices for observing electron diffraction are supplied with many electron microscopes. In Fig. 1 is shown the “EG” horizontal camera designed for study of the atomic structure of crystals at the Institute of Crystallography of the U.S.S.R. Academy of Sciences.79 * Preparations of thin films are made by applying them on the thinnest possible support (- lo-’ cm.) of celluloid Formvar etc. For this purpose according to the nature of the substance investigated it is precipitated from a solution or suspension or from the gaseous phase in the form of a powder or smoke or by sublimation in a vacuum. In contrast to the view sometimes expressed it should be emphasised that the atomic structure of a given phase in thin films is found in the great majority of cases to be identical with its atomic structure in the bulk specimen as shown by a comparision of data obtained from X-ray and electron- diffraction studies.Owing however to the very small volume of the crystals investigated the temperature range of stability of some phase or other is sometimes found to be displaced. In thin films it is easier to produce the necessary transformations by means of thermal or chemical treatment and this is often used in electron-diffraction analysis. What are then the specific nature of electron-diffraction structure analysis and the features that distinguished it from the X-ray and neutron- diffraction methods ? They are in the first place the peculiar feature just mentioned viz.the possibility of determining atomic structure in thin films. In fact to make a complete structure analysis with the aid of X-ray or neutron diffraction single crystals no less than 0.1 mm. in size are required. Many natural and synthetic substances (e.g. clay minerals) however are found only in a highly disperse state and hence their study by X-ray methods is difficult. On the other hand they are a natural object for electron-diffrac- tion studies. It is also advantageous to use the electron-diffraction method Vainshtein and Pinsker KristuZfograJiya 1958,3 3. For review of modern electron diffraction apparatus see Pinsker Pribori i technika experimentu (Apparatus and experimental technique) 1959 1 3. FIG.1. Electron-difraction camera of the Institute of Crystallography of the Acad. Sci. U.S.S.R. (1) Electron gun. (2) Anticathode. (3) Shutter. (4) Magnetic lens. (5) Central chamber. (6) Specimen. (7) Screen. (8) Photoplate. (9) Electrical panel. (10) Vacuum system. .. 108 QUARTERLY REVIEWS for studying the structure of phases unstable in air (e.g. some oxides or crystal hydrates) but stable in a high vacuum. However the fundamental distinctive feature of the electron-diffraction method and of its application to the analysis of atomic structure and a study of the chemical bond lies in the nature of the interaction between the electrons and the substance. Unlike X-rays which are scattered by the electronic shells of atoms of a particular crystal as well as neutrons which are scattered by the nuclei of atoms electrons are scattered on the electrostatic potential of the crystalline lattice.From this fact which will be presently considered in detail follow several peculiar features in the very principle of the electron-diffraction method. Crystalline lattice potential and Fourier synthesis from electron-diffraction data. The motion of electrons is described by the Schrodinger wave equation 8n2m h2 v 2 + + -(E- V)$ = o . . . . . where +(xu.> is the wave function and E is the total and Y the potential energy of the electron. The last is determined by the electrostatic potential $(xyz) of the scattering object so that V = - 4 where e is the charge of the electron. Hence the scattering of electrons by crystals as well as by isolated molecules or atoms is determined by the potential $(r) of these objects .The crystal potential $(r) is a superposition of potentials of its individual atoms. The potential of each atom is determined by the distribution of charges of the positive charge + Ze concentrated in the nucleus and of the negative charge of the electron shells pat (r). As a consequence of the three-dimensional periodicity of the crystalline lattice the potential of the crystal is a three-dimensional periodic function. The peaks of this function correspond to atoms the positions ofmaxima being determined by the position of nuclei. As compared with the well-known electron-density function p ( r ) obtained in X-ray analysis the potential $(r) has the following peculiar feature^.^ $(r) is a somewhat more diffuse function than p(r) which follows from the relation of the Thomas-Fermi statistical theory of the atom $ - J p 2 / 3 .. . . . . . . . . . (2) From this expression it will be seen that the ratio of the potential peak heights to the electron density is different for crystals with different atomic numbers. The electron-density peak heights are proportional to Z% where a z z l is somewhat greater than unity. The potential peak heights #(O) are proportional to Z% where a c 0.8. Hence in crystals it is Vainshtein Trudy Inst. Krisk. Akad. Nauk S.S.S. R. 1954 9 259. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 109 easier to detect light atoms in the presence of heavy ones by electron- diffraction analysis than by X-ray analysis. Electron-diffraction is extens- ively used in determination of the positions of hydrogen atoms in studying the position of nitrogen and carbon in metal nitrides and carbides etc.2 4 6 8 10 12z He Be C 0 Nc M9 I I I I H L i B N F N a I I 2 I 4 6 8 10 12z I I l L i l B l N I I N I He e c c 0 Nc M9 FIG. 2. Curves of the detectability of atoms in X-ray analysis p(O) in electron difraction Fig. 2 shows curves of the relative detectability of atoms by electron diffraction X-ray and neutron-diffraction methods (points). The less the slope of the curve the easier it is to locate a light atom in the presence of heavy ones. The most advantageous in this respect is the neutron-diffraction method since the scattering of neutrons in general is not obviously dependent on the atomic number while electron diffraction occupies an intermediate position between neutron-diffraction and X-ray analysis.A remarkable characteristic of electron-diffraction as well as of X-ray and neutron scattering is that here the Fourier transform of the corre- sponding function determining the scattering is realised ; that is for the diffraction of electrons on the potential $(r) *(O) in neutron diffraction f, (points). lo Vainshtein and Pinsker Doklady Akad. Nauk S.S.S.R. 1949 64,49. l1 For reviews of Fourier methods in electron diffraction see refs. 6 and 9 and also Cowley and Rees Reports Progr. Phys. London 1958,21 165. 110 QUARTERLY REvlEwS The Fourier coefficients F(hkZ) the so-called structural amplitudes are found directly from the intensities of the scattered beams I h k l [where h k 2 indices of the reflecting crystal plane are also indices of the Fourier series (3)]; V is the unit-cell volume.With kinematic scattering of electrons F(hkl) = K d I h k l ; with dynamic scattering F(hkl) = K d ( l h k l Q ) where Q is a correction function K is a coefficient dependent on the type of electron-diffraction patterns (see electron-diffraction patterns I-IV) and the interplanar distance of the corresponding refle~tion.~~J~** Thus by measuring the intensities of the diffraction spots on the patterns finding from these the structural amplitudes F(hkZ) and summing the Fourier synthesis (3) we obtain the picture of the crystal lattice the individual peaks of which correspond to atoms. The essential principles of the Fourier method which lies at the base of modern structure analysis are easy to understand by analogy with the optics of visible rays.The formation of an image of the object in the optical microscope may be divided into two stages (1) the formation of diffraction beams from the object and (2) the bringing together of these beams into an image by means of lenses. As a result an enlarged image of the object in the microscope is obtained. In the structure analysis of crystals only the first stage-the diffraction-is realised. The second stage cannot be attained for X-rays and neutrons since no lenses for these are available. Lenses for electrons are available (the electron-microscope) but their resolving power is still far from being such as to permit us to see individual atoms. The second stage however i.e. making the diffraction beams con- verge into the image of the object may be realised by calculation the summing of Fourier series thus obtaining a “mathematical microscope” with magnifications of the order of hundreds of millions and a resolving power down to decimal fractions of an A.The Fourier synthesis is carried out in a number of different ways as a projection of the structure on a plane one- or two-dimensional sections of a three-dimensional potential distribution etc. By measuring the peak co-ordinates on the maps of Fourier syntheses the co-ordinates of atoms are found directly and information is obtained concerning the lengths of the chemical bonds the angles between them and other geometrical characteristics. More detailed information may be gained by considering the shape of the atomic potential peaks and the distribution of the potential in inter- atomic space. To achieve this the Fourier series (3) are reduced to the absolute scale and the values of the potential are expressed directly in volts.The potential values in the peaks have magnitudes of the order of ‘*Blackman Proc. Ray. Soc. 1939 A 173 68. Vainshtein Kristullogrufiya 1956,1 17 150; 1957,2 340. * For the theory of intensities the geometrical theory of electron-diffraction patterns and other problems pertaining to the theory and procedures of electron-diffraction structural analysis the reader is referred to the literature.6*6*11-1s VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 111 some tens or hundreds of volts while those in interatomic space are of the order of a few volts. It should be borne in mind that atoms in a crystal are in a state of thermal motion and that the pattern of the Fourier synthesis obtained is an average over time and over all the unit cells of a particular crystal.With mean-square displacements of atoms from the equilibrium position 47 z 0.3 A [which corresponds to a value of the thermal motion parameter or Debye factort (B = 8n2p2) z 2 x and is characteristic for many crystals] the potential peak height of an atom (i.e, of the potential in its centre) may be evaluated by the formula $(O) = 0*5Z0‘8 (volts) . . . . . . . . . (4) where 2 is the atomic number. (The numerical coefficients in this formula change with a change of B.) An important Characteristic of the crystal is its mean inner potential +m = F(OOO)/V determined by the zero term F(OO0) of the Fourier series (3). This value may be either found experimentally from electron refrac- tion5J4 or estimated the~retically.~*~*~~ It is related to such a characteristic of the lattice as the energy of the output upon thermoelectronic emission.At this point the potential in an atom should be considered in greater detail. First we take as an example its simplest model in the shape of a spherical condenser with a charge + Ze in the centre (“nucleus”) and distributed about the sphere with a radius R of the charge - Ze (“electron shell”) [Fig. 3(a)]. The negative charge potential inside the sphere is constant and equal to -Ze/R while outside the sphere it declines as -Ze/r i.e. as the potential of a point source placed at the origin of the co-ordinates. By adding this potential to the positive potential of the nucleus the potential of the “atom” will be obtained as equal to Ze/r - Ze/R inside the sphere and to zero outside it.Consequently the charge of the shell of a neutral atom fully screens the potential of the nucleus outside the shell and diminishes it inside the shell. In reality the electron shell has a complex continuous structure and the sketch in Fig. 3(a) should be “smoothed” as shown in Fig. 3(6). The distribution of electrons of the outer shells of atoms is not necessarily spherically symmetrical which is correspondingly reflected in the shape of the potential peak. Therefore in the general case it could be written as +at(r) = Ze/r - 4 [pat(r)] . . . . . . . . (5) where pat(r) is the atom electron density distribution. Should the atom be ionised the equality of charges and hence of the potential beyond the “boundary” of the electron shell is disturbed and this is revealed by the shape and height of the potential peak [see Fig.3(b)]. Qualitatively the l4 Tull Proc. Roy. SOC. 1951 A 206,219 232. t This factor is used to compensate for the thermal movement of atoms by multiplying the value of the theoretical scattering factorfO calculated for atoms at rest by the term exp. (-B sin z6/Az). 112 QUARTERLY REVIEWS effect of negative ionisation is a decrease and a narrowing of the peak as compared to the neutral atom; and that of positive ionisation is an increase and a broadening of the peak. An important characteristic of the atom is its “full p0tentia1”~J~J~ . . . . . . . . . fe(0) = C +at(r)d V (6) the integral of the potential in the total volume-an analogue to the F(OO0) value for a crystal.It is not difficult to show that the full potential FIG. 3. (a) Potential of the atom superposition of the nucleus potential ++ and the electron-shell potential 4- (Jor a model of atom in the form of spherical condenser). (6) Potential distributions in the neutral atom Cat the cation $cat and the anion dsn. for the model shown in Fig. 3 (a) is fe(0) - ZR2 i.e. is dependent on the radius of the sphere inside which the field of the atom is enclosed. For the real atom . . . . . . . . fe(O) -2;’ (6‘) where r2 is the mean-square radius of the electron she1l.l’ The fe(0) value as will be shown below has a distinct connection with a scattering of electrons. It should be noted that it is proportional to the molar dia- magnetic susceptibility x. Thus by finding the shape of the potential peaks of the atoms from electron-diffraction data a lot of information may be obtained concerning their state and hence concerning the nature of the chemical bond too.It should be stressed that the possibility in principle of obtaining the exact pattern of the potential of a crystalline structure requires for its l6 Vainshtein J. exp. theor. Phys. (U.S.S.R.) 1953 25 157. l6 Idem Acta Cryst. 1958,1I 178. l7 Vainshtein and Pinsker Trudy Inst. Krist. Akad. Nauk S.S.S.R. 1950 6 163. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 113 realisation a particularly high accuracy in experiment and in treating the experimental data. In most cases therefore workers restrict themselves to establishing the co-ordinates of atoms while investigations in which the shape and height of the peaks are analysed are much scarcer.There is no doubt however that studies of this kind are highly promising. It is note- worthy that by combining electron-diffraction data showing the distribu- tion of the electron density of the crystal it should be possible to determine the electronic energy distribution in a crystal. We now discuss the accuracy of structure analysis by electron diffrac- tion. One important property of the Fourier series should be noted. They include a large number of structure amplitudes F(hkZ)-ranging from some tens to several hundreds-i.e. as many of them as there are reflections observed from a given crystal. Owing to the statistical nature of errors in the measurement of the magnitude of amplitudes and the large number of them the resulting function +(r) (see ref.3) is but little susceptible to the errors of measurement of individual values of F(hkZ). Therefore even a visual estimate of the intensities of reflexions of electron-diffraction patterns permits us to determine by employing the Fobrier synthesis the interatomic distances with an accuracy of 0.03-0.05 A. In precision studies with a microphotographic technique this accuracy may be in- creased to 0.01 A and even better and the accuracy in determining the potential values to a matter of only some volts. It should be noted that although the treatment of diffraction data by the Fourier method yields the clearest result i.e. the map of the potential distribution for an analysis of interatomic distances and of the state of atoms in electron-diffraction analysis one may also directly consider the magnitudes of structure amplitudes comparing them with their theoretical values.Examples of this will be cited below. We now pass to a consideration of some examples of structure studies by means of the electron-diffraction method in which problems of the chemical bond in crystals have been considered. Electron-diffraction study of hydrogen atoms in crystals and of the hydrogen bond Since they enter into the composition of a large number of organic and inorganic substances hydrogen atoms largely determine the arrangement of their crystalline structure and a number of their physical properties for instance their property of forming hydrogen bonds plays a particu- larly important part in the structure of living matter e.g. proteins nucleic acids etc. All the three modern diffraction methods permit the direct location of hydrogen atoms in crystals.The accuracy of these determinations is for electron diffraction 0.02-0.03 A ; for neutron diffraction 0-01-0.02 A and for X-ray analysis about 0.1 A. The last method therefore hardly 114 QUARTERLY REVIEWS suffices to detect in the Fourier maps the exceedingly small electron density characteristic of a hydrogen atom. But the electron-density Fourier synthesis as well as the potential Fourier synthesis in electron- diffraction analysis enables conclusions to be drawn regarding the degree of ionisation of hydrogen in some particular compound. The most accurate method neutron-diffraction yields only the co-ordinates of nuclei but it is most convenient for a study of the thermal motion of atoms. The C-H Distance.-For a long time the idea has prevailed that this distance is about the same in all organic compounds and is about 1.08-1.09 A.However in the course of the first study of n-paraffins by electron-diffraction it was discovered that this distance was really greater. Recently,ls a precision electron-diffraction study of n-paraffins C18H38 C28H58 and C3,,Hs2 has been carried out. Electron-diffraction patterns have been obtained from films of paraffin prepared by crystallisation on a support from a solution in toluene (electron-diffraction pattern I). On the basis of a microphotometric intensity measurement a Fourier FIG. 4. Fourier projection of potential of C30HB2 (in volt A). (a) C and H atoms; (6) C atoms subtracted. potential projection has been constructed along the aliphatic chain [Fig.4(u)]. In this projection the C-H distances are displayed in an undistorted way. To eliminate certain effects inherent in the procedure used the so- called “difference syntheses” were employed i.e. the peaks of heavy atoms were removed from the potential map [Fig. 4(b)]. I* Vainshtein Lobatshev and Stasova Kristallografya 1958,3 452. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 115 The average C-H distance has been found to be 1.12 A. The accuracy of this determination estimated by several different methods is & 0.015 A. It is noteworthy that in the lower paraffins (n-pentane -hexane -heptane) as shown by data of electron-diffraction analysis in gases,19 this distance is 1.120 & 0.006 A. An analysis of peak heights has shown that no ionisation of hydrogen atoms is observed.At the same time the shape of the peaks of these atoms indicates that their thermal oscillations take place chiefly in the form of oscillations about carbon atoms as centres while the oscillations along the C-H bond are less pronounced. The same is observed in the CH group of the diketopiperazine structure (see p. 1 16). The increase in the C-H distance above the value of 1.09 is not confined to paraffins. An electron-diffraction structure analysis of hexa- methylenetetramine20 has given a value of 1-14 & 0.10 A for this distance and neutron-diffraction has given a value21 of 1.13 0.02 A. Thus for the tetrahedral atom of carbon there is a distinct tendency towards an increase in the C-H distance to about 1.12-1-13 as compared with the value of 1.08 A. At the same time structure-analysis data for the C-H distance in the aromatic ringe2 show that in this case the distance of 1.08 A is retained.The increase in the C-H distance upon a weakening of the adjacent bonds of the carbon atom has been observed by the spectro- scopic method.23 In this way it may be inferred that the presence of a smaller number of other stronger bonds to the carbon atom (e.g. ,C-H) produces a measurable shortening of the C-H bond and in the tetrahedral configuration this bond is the longest and consequently the weakest. This kind of mutual influence of bonds reminds one of conjugation phenomenon with the distinction that in conjugation the given ordinary bond is surrounded on each side by stronger bonds while in the instance considered the adjacent bonds which exercise an influence are located only on one side of it.It has been already pointed out that electron-diffraction as well as neutron-diffraction studies yield the co-ordinates of nuclei. Remarkably they have diverged from the data of X-ray analysis. X-Ray deter- minations of the C-H distances (as well as of O-H N-H) regardless of their rather low accuracy reveal a systematic shortening of the distance from the carbon atom to the electron-density peak of the hydrogen atom of about 0.85-1.05 A. Theoretical analysis24 shows that such a shorten- ing i.e. a displacement of the electron-density maximum of the bonded hydrogen atom towards the carbon atom may actually take place. It is qualitatively accounted for by movement of a part of the electron cloud lP Bartell Report on the Symposium on Electron Diffraction in Leningrad 1959.ao Lobatshev Trudy Inst. Krist. Akad. Nauk S.S.S.R. 1954 10 167. s1 Andersen Acta Cryst. 1957 10 107. 2a Bacon and Curry Proc. Roy. SOC. 1956 A 235 552. a4 Tomiie J. Phys. SOC. Japan 1958 13 1030. Herzberg and Stoicheff Nature 1955 175 79. 116 QUARTERLY REVIEWS to the bond. The shift increases with increase in the degree of ionic nature of the bond as well as with an increase in the thermal oscillations of the atoms. A lack of coincidence in the positions of electron-density maximum and the nucleus is observed only for hydrogen atoms but not for other heavier atoms. It should also be noted that distances between nuclei in electron-diffraction and neutron-diffraction determinations are found as a rule to be 0.02-0.03 A longer than those given by spectroscopic data.This discrepancy may possibly be due to another cause i.e. to the an- harmonicity of the thermal oscillations of the nuclei of molecules.25 As a result of anharmonicity the distribution function of the probability of the position of nuclei is non-symmetrical and its centre of gravity (mean position of the hydrogen nucleus) as determined by neutron-diffraction and by electron-diffraction does not coincide with the position of the potential-energy minimum determined spectroscopically. This problem however has not yet been fully resolved. The N-H Distance-Fig. 5 shows a three-dimensional potential synthesis of diketopiperazine.2* This structure (I) is rich in hydrogen bonds and FIG. 5 . Three-dimensional Fourier synthesis of potential of diketopiperazine molecule.the object of studying it was to locate hydrogen atoms although it had already been investigated by means of X-rays.27 In a ring the angles of which differ but little from 120° the distances are 25 Ibers Acta Cryst. 1959 12 251. p* Vainshtein Trudy Inst. Krist. Akad. Nauk S.S.S.R. 1954 10 49; J. Phys. Chim. 27 Corey J. Amer. Chem. SOC. 1938 60 1598. Continuous line drawn at intervals of 15 volts; broken lines 7-5 volts. U.S.S.R. 1955 29 327. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 117 (in A) C-N 1.386; N-CH, 140,; C-CH2 1.44,; C-0 1.22 (the mean- square error in these determinations is 0-008 &. The C-H distances in the CH group are 1.09 and 1.1 1 A (& 0.03 A) and the bond angle is near the tetrahedral value i.e. 107”. The hydrogen atom forming the bond between the N and 0 of neighbouring molecules is covalently linked to the nitrogen atom and is 0.98 & 0.03 A distant from it.The hydrogen bonds of NH - * 0 deviate slightly from rectilinearity; their lengths are 2.84 A. The elongated shape of the peaks of hydrogen atoms indicates a different nature for the oscillations in the CH group (around C) and in the NH group (along the hydrogen bond). The values of the potential in the centre of C N and 0 atoms are close to each other being about 160 v. Their equality is accounted for by the fact that although in the C N 0 series the atomic number 2 increases the mean-square radius of the electron shell simultaneously increases that is the screening of the nuclear potential is augmented too [see equation (6)]. The +(O) potential of hydrogen atoms in the CH group is 32 and 33 and in the NH group 36 v .The increase of this potential indicates as discussed above the ionisation of this atom. A theoretical calculation shows that such a potential corresponds to the presence of about 0.85 electron in this hydrogen atom. Thus the concept of a primarily ionic nature of the hydrogen bond suggested by Pauling has been experimentally confirmed in the course of this electron-diffraction study. In a series of structure studies the N-H distance in the ammonium chloride structure has been determined. Since the first electron-diffraction study in which a value of 0.95 & 0.07 A was obtained (1 933) the accuracy of determinations has increased 0.98 & 0.04 A (1956),,* 1.02 0.02 A (1959).29 Neutron diffraction30 yielded 1-03 0.02 A and an X-ray study (cited in ref.29 without any indication of accuracy) gave 0.93 A. During an electron-diffraction study of cryptohalite (NH,),SiF (electron-diffraction pattern 11),31 an interesting effect of reorientation of the NH group was observed. It has been found that these tetrahedral groups surrounded by twelve F atoms from the SiF groups occupy statistically six equally probable orientations forming in each of these weak hydrogen bonds with four F atoms. One of these six orientations is presented in Fig. 6. The N-H distance is equal to 1-03; and H - * F is 1.95 A. It should be noted that as a result of statistical reorientations on the Fourier map only the peak of 1/3 atom is detected (of the six orienta- tions three are close in pairs) only 12 v in height (Fig.7). This precision study in which 1/3 H was detected in the presence of atoms of N (2 = 7) F (2 = 9) and Si (2 = 14) shows that one may hope to determine by electron-diffraction the positions of the “ordinary” hydrogen atoms in 28 Stasova and Vainshtein Trudy Inst. Krist. Akad. Nauk. S.S.S.R. 1956 12 18. ao S. Kuwobara Reports of Electron Diffraction Works in Japan Symposium on *O Levy and Peterson J. Amer. Chem. SOC. 1953 75 1536. Vainshtein and Stasova KristuZZografiyu 1956 1 31 1. Electron Diffraction in Leningrad 1959. 118 QUARTERLY REVIEWS structures with a mean atomic number of the order of 30. As a matter of fact the height of peaks corresponding to statistically distributed hydrogen atoms (12 v) is but twice the value of the potential-measurement error ( f i 6 v).Nevertheless the positions of 1/3 H were ascertained although the accuracy of determination of the N-H distance was inevitably lower (0.1 A) than in the other cases. This study will be further discussed below. FIG. 6. One of six equally probable orientations of the NH4 tetrahedron in the crystal structure of (NHJ2SiF6. 220v n I FIG. 7 . Section of three-dimensional synthesis of the potential of (NHJ2SiF along the (1 10) plane. Lines drawn at intervals of 30 volts. The cross indicates the position of 1/3H as established from differential synthesis. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 119 The 0-H Distance.-The first example of an electron-diffraction deter- mination of this distance was a of boric acid B(OH), in which a value of 1.00-1.05 A was found. Recently electron-diffraction was used for a study of the structure of cubic ice.33 When electron-diffraction studies are carried out at low temperatures ice is readily formed on the support from the water vapour remaining in the vacuum.34 If the condensation temperature is below --8O"c the ice crystallises in the cubic "diamond" modification with a = 3.36 A and an 0-0 distance of 2.75 A.Above -8O"c the common hexagonal ice is formed. Analysis of the intensities and construction of the Fourier potential projection has proved that in the cubic as well as in the common hexagonal modification a statistical distribution of the hydrogen atoms occurs at two equally probable positions on the 0-0 line (Fig. 8). The 0-H distance for each of these positions is 0-97 A. FIG. 8. 0.37 Fourier volt A). projection of the potential of cubic ice on the (100) plane.( U n i t s of scale The instances cited above show the possibilities of the electron-dif€rac- tion method for a study of hydrogen atoms namely the possibility of directly locating them in the lattice and of making an analysis of the thermal motion and of the degree of ionisation. Study of the state of atoms in a crystal The structure of the electron shell of a free atom changes when the atom is incorporated into the crystal lattice. As is known the general 3a Cowley Acta Cryst. 1953 6 516 522 846. 8t Honjo and Shimaoka A d a Cryst. 1957 10 710. s4 For review of electron diffraction investigations at low temperatures see Dvorian- kin KristaZZografiya 1959.4 441. 120 QUARTERLY REVIEWS features of this change may be described as follows.In metals a pait of the electrons go to the conductivity zone ; in ionic compounds the electrons are redistributed from cations to anions; with a covalent bond directed electron “bridges” are formed between valency-bound atoms. These changes affect the potential distribution as has been shown above and hence may be detected by means of electron-diffraction. Electron-diffrac- tion work of this type is analogous to X-ray studies of electron-density distribution initiated by the well-known work by Brill et aZ.35 Some examples may be considered. In the first work of this type,s from experi- mental electron-diffraction data the potential distribution in the crystalline lattice of Al Cu and Ag was found. In Fig. 9 is shown the distribution of the potential along the body diagonal of the face-centred cubic unit cell of Ag.Some unevenness of the distribution obtained is due to errors in the measurement of F(hkZ) and the so-called termination effect in the Fourier series which may be artificially eliminated. In spite of this unevenness however some essential features of the distribution of the potential in metals may be seen in Fig. 9. FIG. 9. Potential of Ag along the diagonal of the unit cell. Analysis of the shape of peaks for all the three metals has shown them to be spherically symmetrical. This is in accord with the concept of an undirected metallic bond. The potential in the interatomic space is *6 Brill Grimm Hermann and Peters Ann. Physik 1939 34 393. Electron diffraction patterns (I) Structure of parafin C,,H,,. Plane of the specimen is perpendicular to the electron beam.(11) Structure of(NH,),SiF,. Angle of tilt of specimen is 55". (HI) Mosaic single crystal film of y-Mo,N. (IV) Structure of kaolinite. Angle of tilt is 60”. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 121 determined by the experimental values of F(hk2) according to eqn. (3) its absolute value however being dependent also on the value of the zero coefficient QOOO) determining the mean inner potential $m = F(OOO)/ V. Without the F(OO0) term the beginning of the scale is the 0’-0‘ line in Fig. lO(a); its introduction transfers the beginning to the 0-0 line. The Reviewers assumed from theoretical calculations that $,(Ag) is 38.8 v. On averaging the background oscillations a value of &in(Ag) = 17 & 4 v was obtained as the minimum potential value in interatomic space.According to other more precise data,l* however &(Ag) is equal to 23 v from which it follows that bin(Ag) = 1 -J= 4v. Thus the potential in the inter- atomic space of metals is in practice very close to zero but still remains positive throughout as would follow from general conceptions. Most sharply pronounced in electron-diffraction is the ionic nature of the bond since there occur a decrease in the potential of anions and an increase in that of cations as compared with neutral free atoms [Fig. 10 - ( b 1 FIG. 10. Schemes of potential distribution (a) in metal; (b) 1 in the crystal structure of neutral atoms; 2 in the same structure at ionisation (C=cation A=anion); 3 the difference between curves 2 and 1. (b)]. The value of the potential in neutral free atoms may be estimated theoretically from the so-called atomic amplitude c u r v e ~ .~ J ~ J ~ Thus in the analysis of the structure of cryptohalite (above),31 it was found that $(O)si = 428 (420) and #(O)F = 278 v (290). A comparison of the theo- retical values given in parentheses shows that in this lattice silicon is positively and fluorine negatively ionised the bond thus having an ionic nature. It should be noted that the fluorine potential peak has proved to be also more “compressed” which may be accounted for by the above- mentioned effect of the electron shells’ being drawn towards the nucleus, 122 QUARTERLY REVIEWS a decrease of r2 [formula (6')] which is particularly marked for elements at the end of the second period of the Periodic Table. In the course of a study of the chemical bond in the LiH structure,36 the following values have been obtained +(O)L~ = 142 (138) +(0)H = 43 ( 5 9 all values being in volts and in parentheses the theoretical values for neutral atoms being cited.In this way in this hydride occurs a marked ionic bond Li+H-. When studying NaF Japanese investigators3' used the procedure of a difference synthesis or subtracting of peaks already referred to in the description of the study of paraffins (cf. Fig. 4). The (3) series is constructed in this case according to the coefficients (@exp.-@calc.) where the experi- mental coefficients Qjexp. carry the effect of ionisation and in general of electron redistribution while the values of Qcalc. have been calculated for neutral atoms. In Fig. 11 is shown the picture of distribution of such a f unit =/.9voltr FIG.1 1. Diference synthesis of potential of NaF along the (100) plane. difference potential in the plane passing through Na and F atoms [see the non-dimensional scheme Fig. lo@)]. The difference distribution in Fig. 1 1 shows that there occurs a shift of the negative charge from the periphery of the Na+ cation to that of the F- anion especially along the line con- necting these two atoms. $(O)F is diminished (by about 13 v) and+(0)Na is slightly raised (by about 2 v). Some details of the picture obtained may be due to errors in the determination of F(OOO) but the general nature of the potential redistribu- tion (and hence of the charge) is not subject to doubt. The lattice consisting of anions and cations is on the whole neutral; the external negative shell of the potential of anions enters into the lattice 86 Pinsker and Kurdyumova Kristallografiya 1958 3 501.Kitamura and Honjo see ref. 29. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 123 potential in a superposition with the excess of positive potential of cations [cf. Figs. 3(6) and 10(6)]. It is not clear a priori whether in the ionic lattice at the boundary of anions there might be sites with a negative potential (in metals and lattices of “neutral” atoms this is altogether impossible). In the synthesis shown in Fig. 11 owing to its different nature the F(OO0) value does not enter and according to other data5 it is equal to +8.3 v for NaF. This is larger than the value of -4 v on the periphery of the atom according to Fig. 11 and hence the potential of this lattice is positive everywhere as has been also found in the investigations of LiH and cryptohalite referred to above.Thus the view as to the non-negative values of the potential in “interatomic” space of crystal structures seems to be of general validity. A qualitative deduction as to the nature of the bond may be made without reference to the potential maps by direct consideration of the course of F(OO0) amplitudes depending on the 8 scattering angle which is characterised by the value sin 6/h. From the values of F(000) amplitudes one may pass to the atomic amplitude fe(sin O/X) characterising the scattering by a particular atom and hence its electron structure fe being equal to fe(0) at sin 6/X = 0 according to eqn. (6’). 0 0.2 0.4 0.6 0.8 /O-‘(sin @/A) FIG. 12.f Curves of oxygen in Li20. (1) Experimental; (2) (3) (4) theoretical (see text for explanation). In Fig. 12 is shown the fe curve (1) for oxygen found in the course of a study of Li,0;38 theoretical curves are also drawn corresponding to the free oxygen atom with a distribution after Hartree-Fock (2) to a bonded oxygen atom (3) and to the 02- anion (4). From this Figure it is seen that ** Vainshtein and Dvoriankin Krisfullogrufiya 1956 1 626. 1 24 QUARTERLY REVIEWS the fe curves at small angles are very sensitive to the nature of the bond. A qualitative inference from a comparison of curves 1-4 is that the bond in Li,O is basically of a covalent nature but has also a very small con- tribution from the ionic state since at low angles curve 1 is somewhat below curve 3. The lowering of curve 1 as compared with the theoretical curves at large angles is due to the thermal motion of atoms in the lattice and is not related to the state of the atom.It should be noted that there is a possibility of finding the structure amplitudes F(hkZ) not from intensities but from the fine structure of the interference spots of electron-diffraction patterns arising from the so-called dynamic birefringence if the microcrystals studied have a regular habit. This possibility has been realised and the Fourier potential series has been constructed for the MgO structure.39 The statements made above may now be summarised. Electron- diffraction data make it possible quantitatively to characterise the distri- bution of the electrostatic potential of atoms in the crystal lattice and to make a number of inferences regarding the chemical bond.These investiga- tions require a high accuracy of experimentation and have as yet been carried out only on crystals of fairly simple structure. From the examples cited it is seen that it is easiest to establish the presence or absence of ionisation of the atom. The identification of a covalent bond in which there is no passage of electrons from one atom to another is more complicated but nevertheless also possible particularly if difference syntheses are used. We may note that in principle (with the aid of the Poisson equation) it is possible to find from the potential distribution that of the charges and to obtain the map of electron-density directly from electron-diffraction and not from X-ray data. Study of the structure of nitrides and carbides of metals Electron-diffraction has been applied to this problem for a number of reasons.One of these is the comparative simplicity of preparing one-phase specimens of metals in thin layers in the gas phase. By sublimation of pure metals onto a heated orienting support [e.g. the (100) face of rock-salt] the metal film may be made a single-crystal one which facil- itates some stages of the study and allows in particular the orientational relations between the phases to be established. X-Ray studies of this kind are carried out on polycrystals by the Debye method which makes it difficult to carry out structure determinations. Moreover not one but several phases are as a rule contained in polycrystalline macro-specimens making it difficult to carry out an X-ray analysis Another advantage of electron-diffraction lies in the possibility of locating light atoms (carbon and nitrogen in this particular case) in the presence of heavy-metal atoms.8Q Cowley Goodman and Rees Actu Cryst. 1957 10 19. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 125 At the same time in analysing such specimens one encounters some specific difficulties in particular in the determination of the composition of the films. Nitrides and carbides forming interstitial structures are generally represented on the equilibrium diagram by several phases these being in most cases phases of variable composition. Chemical analysis of the thin films studied by electron-diffraction methods is very difficult. So an electron-diffraction study of structure is in this case also a determ- ination of composition; i.e.here is an example of chemical analysis by the diffraction method. Nitridation of the films is carried out at temperatures of 3 0 0 - 5 0 0 " ~ in a stream of ammonia which at the metal surface dissociates into nitrogen and hydrogen 2xM + 2NH3 + 2M,N + 3H,. Gas-cementation of films is made in a stream of carbon monoxide. An electron-diffraction study of the systems Fe-C Fe-N Cr-N Mo-N W-N has been systematically conducted by Z. G. Pinsker and his c011eagues.~~-~~ Recently the Ni-C system has been studied by Japanese The struoture of cubic iron nitride Fe,N has been repeatedly subjected to X-ray analysis; but owing to the fact that the scattering of X-rays is Here only a selection of results will be cited. IZSO 1213 FIG. 13. Potential ofFe,N in plane (110) (volts).primarily determined by the presence of four iron atoms it was not pos- sible to obtain direct information on the arrangement of nitrogen atoms in this structure. An idea of their position could be obtained only on the basis of geometrical analysis i.e. by considering the crystal chemical radii of Fe and N atoms. By employing precision electron-diffraction methods and introducing in particular the dynamic correction Q when trying to find the structure amplitudes F(hkZ) Dvoriankina et aL41 calculated the potential dis- tribution in the (111) plane of this structure (Fig. 13). In this map 40 Pinsker and Kaverin Kristallografiyu 1957 2 386. 41 Dvoriankina and Pinsker Kristallografiya 1958 3 438. dm Pinsker and Abrosimova Kristallografiya 1958 3 281. 42 Troizkaija and Pinsker Kristallografiya 1959 4 38.43 Hitrova and Pinsker Kristallografiya 1958 3 545. Nagakura J. Phys. SOC. Jupun 1957 12,484; 1958 13 1005. 126 QUARTERLY REVIEWS the peak of the nitrogen atom potential is distinctively revealed. In viewing this pattern a small difference may be noted in the heights of the potential maxima of iron atoms lying at the origin of co-ordinates (1250 v) and in the centres of faces (1213 v). If a difference in the ionic state exists the profiles of maxima should be different. They wei'e found however to be similar from which fact it was inferred that the difference in heights reflects a small disturbance of the stoicheiometric composition of the phase which corresponds here to the subtraction structure. The positions in centres of faces are occupied by approximately (1213/1250) x 100 = 97% atoms of Fe and the formula of this nitride may be written as Fe,Fe,(,_,)N ( x = 0.03).Probably it is the vicinity of the nitrogen atom to which the distance from the second atom of iron 1.90 A is less than from the first (3.29 A) that explains the partial non-occupation of the positions referred to. A comparison of the heights of maxima with theoretical values indicates a certain diminishing of the potential $(O)N which may be accounted for by an excess of electrons in nitrogen atoms. It should be noted that owing to a smaller amplitude of thermal oscillations of atoms the peak heights of carbon and nitrogen in carbides and nitrides are always greater than those in ionic or organic compounds. Therefore comparison with theory is made with a proper choice of the B parameter of the thermal motion which in this particular case was equal to unity.The inference as to the anionic state of nitrogen was made by constructing fe curves (cf. Fig. 12) in the analysis of the CrN structure also.*la Deviation from the stoicheiometric composition as well as the distribution of nitrogen atoms over several equally probable positions which are characteristic of interstitial phases has been revealed also when analysing the st~xcture~~ of y-Mo,N (electron-diffraction pattern 111). This structure is represented in Fig. 14. @ /oo% 8 roo% 0 33% "'0 6 7 % FIG. 14. Structure of y-Mo,N. X-Ray structure data on nitride phases of tungsten are both indistinct and incomplete. It is altogether hopeless to attempt-even if X-ray photographs be obtained- to reveal the arrangement of nitrogen atoms (2 = 7) in the presence of tungsten atoms (2 = 74).In studying films of tungsten nitrides,43 prepared under different conditions of nitridation in addition to electron-diffraction patterns from the well-known cubic phase of P-W2N with a period a = 4.12 A electron- VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 127 diffraction patterns have been obtained from four hexagonal phases not described previously with a = 2.89 A and c = 15.3 (i) 22.8 (ii) 23.4 (iii) 324 (iv) A. The hexagonal &phase known from X-ray data with the same a period = 2-89 A but a short c period = 2-82 A was not observed. A complete structure study of phase (i) was carried out. This phase was found to be built up of three-layer sheets of tungsten atoms the nitrogen atoms entering into the packing holes-trigonal prisms.The sheets are superimposed upon each other according to the principles of close packing (Fig. 15). The positions in the centre are occupied by tungsten 0,W o,N FIG. 15. Model of structure of W,N4. atoms to 100 % and the external “defective” positions only to 65 % which leads to a formula W,W4(,-,~N4 (x = 0.35) as against the “ideal” formula WBN,. It is noteworthy that for the distances W-W along the ver- tical we have a normal value of 2.80 a and for the “defective” W atoms a diminished one 2.56 A (the positions are occupied statistically). The W-N distance has been found to be 2-23 A. An electron-diffraction of the structure of nickel carbide Ni,C has revealed the conditions of the most rapid formation as well as of the decomposition of this carbide.The electron-diffraction patterns have been treated after introduction of the dynamic correction. Construction of the Fourier potential projection on the basal plane (Fig. 16) has revealed both nickel and carbon atoms. The unit cell has a large period c = 12-92 A. Nickel atoms are arranged within it in closest hexagonal packing while carbon atoms are distributed in one-third of the octahedral holes of this packing according to a rhombohedra1 law. By evaluation of the width of certain diffraction lines the degree of ordering in this structure could be established. It was found that the mean size of the regions in which carbon atoms are regularly positioned is 28 A which corresponds to about 13 layers of packing. 128 QUARTERLY REVIEWS Thus in electron-diffraction studies of interstitial phases it is feasible fully to establish their structure including the position of light atoms (which is difficult by X-ray methods) to reveal the composition of these phases FIG.16. Projection of potential of NiSC on the basal plane. to give a quantitative characteristic of the statistics of arrangement of certain atoms to estimate in some cases the chemical state of the atoms and to analyse the orientational correspondence in a mutual phase transi- tion. Some other applications of structure analysis by electron-diffraction. Conclusion In a number of electron-diffraction studies use was made of the possi- bility of determining the structure of crystals in a highly disperse state. In these studies no qualitatively new relations have been established as compared with data of X-ray analysis.However the crystal chemistry of this kind of substance as a consequence of the specific features of X-ray structure analysis requiring the availability of single-crystal specimens presents a sort of “white spot” against the background of a huge number of other compounds studied by X-ray methods. Each new structure determina- tion is therefore of essential importance. A good example of such a work is furnished by the study of basic lead carbonate.45 The crystals of this basic salt represent minute hexagonal plates about 5p across. This compound is formed in aqueous suspensions of lead oxide by interaction with the carbon dioxide of the air. It has been found that Pb(OH),,2PbC03 has a hexagonal unit cell with a = 9.06 A c = 24.8 A.The structure is built up of three-layer sheets 8.24 A thick in which the layer of Pb(OH) is surrounded by a layer of PbC03 on each side. Fig. 17 shows a difference Fourier potential projection of one sheet of this structure from which atoms of Pb have been subtracted. The light 45 Cowley Acta Cryst. 1955 9 397 399. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 129 atoms of C and 0 are well revealed. The interatomic Pb-0 distances vary over the range 2.5-2.8 A; the C-0 distance in the C03 group is 1-45 A. The structure of an individual sheet is comparatively stable and the interaction between sheets is determined by weak residual forces. In this structure therefore as well as in many other layered structures particu- larly in clay minerals errors in the sheet positions have been revealed.FIG. 17. Projection of potential of basic lead carbonate. Contributions of Pb atoms in the indicated positions have been subtracted. The statistics of these positions may be described by a certain distribution function being directly determined from the electron-diffraction pattern. of a basic salt CuCl,,Cu(OH), a characteristic tendency of the copper atom to form square complexes has been revealed. Four hydroxyl group of the six anions surrounding Cu are spaced at distances of 1.86-1.97 A which points to a covalent nature of these bonds. The other OH groups (or C1 ions) are situated at considerably greater distances viz. 2.37 and 2.86 A respectively. An important field of application of electron diffraction is in the study of clay minerals.47 The specific features of the latter are highly dispersed state ; disturbances of structure perfection ; variations in composition related to isomorphous replacements of cations; and laminated or less frequently fibrous form of particles.Specimens for study are prepared from aqueous suspensions. Most favourable for structure analysis are diffraction patterns from laminated textures of clay minerals. The forma- tion of such textures is related to the layer nature of their atomic structure built up of octahedron networks populated by cations of the Mg Al Fe type and tetrahedron networks of SO4. It should be noted that electron- diffraction data concerning the structure of the swelling clay minerals of the montmorillonite group obtained in conditions of high vacuum refer only to a fully dehydrated state.In another example of structure 46 Voronova and Vainshtein Kristallografya 1958 3 444. 47 Pinsker Works of the Institute of Geochemistry Acad. Sci. U.S.S.R. 1954 p. 116. 130 QUARTERLY REVIEWS In every clay mineral there are specific characteristics due to their being built up in the ideal scheme of networks of octahedra and tetra- hedra. Among these features are nature of linkage of the networks and their mutual orientation; specific configuration of the networks and the interatomic spaces inside them; structure of interlayer spaces; and finally the mutual arrangement of layer sheets. Already in addition to numerous studies on the electron-diffraction determination of the unit cells of clay minerals some examples of full structure determinations are also available. Zvyaghin4* has studied the atomic structure of seladonite-a ferromagnesium mica which is representa- tive of “three-storey” clay minerals the layer of which is built up of an octahedron network with networks of tetrahedra on each side.The same has determined the structure of kaolinite the most import- ant representative of clay minerals built up of “two-storey” layers (one network of octahedra and one of tetrahedra) (electron-diffraction pattern IV). This structure has no centre of symmetry and its layers are polar. The unit cell is triclinic a = 5.13 b = 8.89 c = 7.25 A a = 91”40’ 16 = 104”40’ y = 90”. The construction of the Fourier potential projections (Fig. 18) and the determination of structure because of the absence of a centre of symmetry were considerably more complicated than if 1 \ Fro.18. Sideprojection of kaolinite A12Si2Q5(0H)4. The structure of the sheet built up of Si tetrahedra and A1 octahedra is clearly seen. (Broken lines represent the edges of polyhedra.) a centre of symmetry had existed. In particular the structure factors F(hkZ) have in this case complex numbers. A unique result in calculating a non-centrosymmetrical structure is achieved by a series of successive approximations. The accuracy of determination of the atom co-ordinates is about 0.03-0.04 A. As a result of this study a number of peculiar features in the structure of 4a Zvyaghin Kristallografiya 1957 2 393. 4B Zvyaghin Kristollografiya 1959 4 5. VAINSHTEIN ELECTRON DIFFRACTION AND CRYSTALS 131 kaolinite have been revealed concerning the structure and combination of the tetrahedral and octahedral networks.The 0-0 distances in the com- mon edges of the octahedra are reduced and the octahedra as a whole are somewhat flattened. The A1 atoms are shifted to the lower sides of the octahedra (towards the OH groups) and the Si atoms in the SiO tetra- hedra towards the upper bases of tetrahedra. The atoms of the two external sides of the sheet do not lie in one plane but form a goffered network. The contact between the adjacent layers takes place between the atoms of oxygen and the hydroxyl groups and hydrogen bonds are perhaps formed between them. It should be noted that in recent years additional possibilities of studying the atomic structure of matter (and of clay minerals in particular) have been offered by the use of the microdiffraction method,50 allowing a diffraction pattern to be obtained from an individual microcrystal as well as by the use of superhigh (400 kv) voltages for electron a~celeration.~~ In conclusion the author would like to stress that the present Review does not in any way pretend to give a survey of all the work on electron- diffraction in which problems of the atomic structure of matter and the chemical bond in crystals are attacked in varying detail.The number of papers on the electron-diffraction of solids exceeds 2000 and those on electron-diffraction structure analysis number over 300. The object of the present Review has been a much more limited one namely to give the reader an idea of the principles and possibilties of electron-diffraction structure analysis of crystals and to show some instances of its application to the study of the chemical bond.Many of these examples represent work carried out at the Electron Diffraction Laboratory of the Institute of Crystallography of the Academy of Sciences of the U.S.S.R. At the same time owing to lack of space this presentation has omitted a number of trends of electron-diffraction structure analysis which are of interest for the theory of the chemical bond. Among these are in the first place experimental and theoretical work on the structure and nature of interaction in simple layer lattice^.^ There is in addition work on the structure of crystal hydrates and the crystal chemical functions of water molecules in them.6 In the series of studies on the crystal chemistry of transition metals-Co Ni Mn-peculiar features of octahedral co- ordination and the conditions of a passage to tetrahedral co-ordination have been r e ~ e a l e d ; ~ ~ ~ ? ~ ~ in the work5* on the structure of KPtCI,,NH, the effect of the influence of the NH3 group on the lengths of the bonds Pt-C1 in the square complex of the Pt atom has been observed.Great 8o Honjo and Mihama Acta Cryst. 1954 7 511. O1 Popov and Zwjagin Kristallografya 1958 3 706. 52 Vainshtein J. Phys. Chim. (U.S.S.R.) 1952 26 1774. 55 Tishenko and Pinsker Doklady Akad. Nank S.S.S.R. 1955,100 913. 64 Bokii Vainshtein and Babureko Zzvest. Akad. Nank S.S.S.R. Otdel. Khim. Nauk 1951,6 667. 132 QUARTERLY REVIEWS interest attaches to work on the oxidation of metals,5~32~65~5s and on the structure of metals and A long s e r i e ~ ~ ~ - ~ l of electron-diffrac- tion investigations has been conducted on the atomic structure of semi- conducting materials.(References 47-62 are only individual examples or reviews.) Finally we wish to draw attention to the following points. Electron- diffraction analysis may be regarded as an independent method for atomic structure analysis furnishing in particular much valuable information on the chemical bond. But it would be a serious mistake to consider it as a universal method which may be substituted for X-ray structure analysis. A comprehensive study of the nature and properties of matter requires the use of the most diverse methods including diffraction methods-the X-ray electron-diffraction and neutron-diffraction analyses. Each of them has its own advantages and limitations and there is for each a special field of application.The Reviewer thanks Prof. Z . G. Pinsker for reading the manuscript and making valuable suggestions. 55 Dankov Ignatov and Shishakov “Electronographicheske issledovania okisnich i gidrookisnich plenok na metallah” (Electron diffraction investigations of oxide and hydroxide films on metals) Moscow 1953. 56 Ignatov Kristallografiya 1957 2 4. 57 Trillat and Takahashi Compt. rend. 1953 236 2245. 58 Fujiwara Hirabayashi Watanabe and Ogava J. Phys. SOC. Jupau 1958 13 167. 58 Semiletov Trudy Znst. Krist. Akad. Nank S.S.S.R. 1954 10 189. 6o Vainshtein Nuovo Cim. 1956 Suppl. V 3 773. 61 Semiletov J. Sci. Znd. Res. 1956 16 A 377. 62 Pinsker Adv. Electronics and El. Physics 1959 XI.

ISSN:0009-2681    

DOI:10.1039/QR9601400105

出版商:RSC    

年代:1960

数据来源: RSC

摘要:

ARRHENIUS FACTORS (FREQUENCY FACTORS) IN UNIMOLECULAR REACTIONS By B. G. GOWENLOCK M.Sc. PH.D. (DEPARTMENT OF CHEMISTRY UNIVERSITY OF BIRMINGHAM) IT is well known that the temperature-dependence of the velocity constant (k) of a reaction can be written in the form k = A exp (- E/RT) . . . . . . . (1) where E is the energy of activation and A is termed the Arrhenius factor. ( A is alternatively termed the frequency factor temperature-independent factor non-exponential factor and pre-exponential factor.) It is obviously desirable to be able to calculate Arrhenius factors from first principles and thereby to understand the precise nature of their dependence if any on temperature or pressure. When the transition-state theory is applied to unimolecular reactions two possible cases can emerge.l By this theory where K is the transmission coefficient and F; and Fi are partition functions for the activated and the initial state respectively.This expression can be shown to equal ~ ( k T / h ) (l/fv) exp (- E,/RT) where fv is the partition function for one vibrational mode; fv is equal to [ 1 - exp ( -hv,/k7')]-1 where vo is the vibrational frequency. N7e can distinguish two possible cases on the onc hand the temperature is high and thenf equals kT/hv and the rate constant is reduced to K V exp (- Eo/RT); at relatively low temperaturesfv tends to unity and the rate constant therefore is equal to K(kT/h) exp (- E,/RT). In practice the two cases cannot be distinguished because the value of kT/h is approximately 1013 sec.-l in the temperature range in which most gas reactions are conducted and this value is of the same order as a vibrational frequency (v,).We can therefore assume a value for the Arrhenius factor of about 1013 sec.-l. This approach implies that an equilibrium concentration of activated complexes is maintained and that the frequency of decomposition of the activated complex is equal to some normal frequency of the molecule. In 1928 Polanyi and Wigner2 called attention to the prevalence of values of 1013 sec.-l for the Arrhenius factors of the unimolecular reactions then known. They cited a number of unimolecular decompositions in solution from a previous compilation by Chri~tiansen,~ and also the values obtained from the gas-phase thermal decomposition of dinitrogen pentoxide Szwarc Chem. Rev. 1950 47 75. Christiansen 2. phys. Chem. 1923 A 104 451.a Polanyi and Wigner Z . phys. Chem. 1928 A 139 439. 133 134 QUARTERLY REVIEWS acetone azomethane Prl-N :N-Prl dimethyl ether diethyl ether and propionaldehyde and from the gas-phase racemisation of pinene. We may note that all the gas-phase thermal decompositions cited are now known to involve chain reactions or composite reaction mechanisms and hence it is fortuitous that the Arrhenius factors are in‘the region of 1013 sec-l. These criticisms must not however detract from the importance of Polanyi and Wigner’s suggestion which focused attention on the possibility of classifying reactions according to the “normality” or “abnormality” of the Arrhenius factors. As we shall observe at the conclusion of this Review it implied a close connection between the experimentally deter- mined Arrhenius factor and the reaction mechanism proposed.Different treatments of the Arrhenius factor are given by the various theories of unimolecular reactions. In Kassel’s theory* this parameter is a constant which represents the frequency with which internal transfers carry energy into the critical oscillator ; in Hinshelwood’s earlier appr~ach,~ A is a fixed probability of dissociation for molecules where the total energy E exceeds E, the activation energy. In the variety of formulations em- ployed by Slater,6 the Arrhenius factor is a specially weighted average of the vibration frequencies in the molecules which always lies between the least and the greatest of the fundamental frequencies. The transition-state theory also considers the Arrhenius factor as a vibration frequency.It is possible to use Slater’s theory6 to predict the variation of the Arrhenius factor with pressure from the variation of the activation energy with pressure of reactant. If we define the activation energy derived from the limiting rate constant as E, and that from temperature-dependence of the general pressure rate constant as Ea then we can make use of the relation where 8 varies as collision frequency [Im (0) w 1 when 8 is large] and we derive on differentiation with respect to temperature Ea = E - mkTAm (8) k = k,lm(B) where Am(@ varies from 0 to 1 as 8 varies from 00 to 0. Thus as then k = A exp (- Ea/kT) A = k exp (EaIkT) = A exp (- E,/kT). Im (8) exp (EalkT) = AJrn (8) exp [- mAm (Q] The Arrhenius factor A will therefore fall from the limiting pressure value A as the pressure drops both because I,(@ decreases with decrease in 8 and because Am(@ increases with decrease in 8.Kassel J. Phys. Chem. 1928,32 225. Hinshelwood “Kinetics of Chemical Change,” Oxford 1940. Slater “Theory of Unimolecular Reactions,” Cornell Univ. Press 1959. GOWENLOCK ARRHENIUS FACTORS 135 Essential Experimental Conditions In any evaluation of the Arrhenius factors of unimolecular gas reactions it is primarily important to consider the nature of the experimental evi- dence available. Such evidence may indicate that any one of several possible explanations must be considered before a detailed theoretical assessment of an “abnormal” Arrhenius factor is attempted. The Arrhenius factor is usually determined from extrapolation of the log k-l/T graph to 1/T = 0.Thus the normal experimental uncertainty in the activation energy will be reflected in the Arrhenius factor. For typical gas-phase thermal decompositions an error limit of & 2 kcal.mole-l in the activation energy will result in an error limit of &- 0.7 in log, A. It is necessary also to emphasise that the reaction studied must be predominantly homogeneous and genuinely of first order before attempting to assess the significance of the Arrhenius factor. The following two examples illustrate the importance of showing that chain reactions are absent and that the limiting pressure has been attained. (a) Absence of chain reactions. We have already stated that the reactions cited by Polanyi and Wigner in favour of the 1013 sec.-l Arrhenius factor were in fact decompositions that followed a chain mechanism.It is possible for such reactions to exhibit a “normal” Arrhenius factor. We may illustrate as follows the Rice-Herzfeld7 mechanism for the decomposi- tion of ethane can be written C2H6 -+ X H . . . . . . . (1) CH,+C2H,+CH4 +C2H . . . . (2) C2H,+C2H,+H (3) H +C2H,-+H2 +C2H (4) 2H + H2 ( 5 ) H + C2H5+CZH (6) H + CH -+ CH (7) CH + C2H -+ C3H8 (8) 2C2H5 + C4HIo . . . . . . . (9) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . If we assume this mechanism to be valid and that reactions (9 (7) (8) and (9) are negligible then Thus the Arrhenius factor will in fact be composite and will equal (A1A3A4/ 2A,)t. If A6 = A4 then the experimentally observed Arrhenius factor will be the geometric mean of A l and A and though approximately 1013 sec.-l will have no simple significance and cannot therefore be related to a single reaction step.’ See Steacie “Atomic and Free Radical Reactions,” Reinhold Publ. Inc. New York 1954. 136 QUARTERLY REVIEWS (b) Freedom from fall-oflcharacteristics. We have already mentioned that the theory of unimolecular reactions indicates that the velocity constant will reach a limiting value at any one temperature when a limiting pressure of reactant has been attained and that the activation energy and Arrhenius factor are also pressure-dependent below this limiting pressure. If the reaction is carried out in the pressure-dependent region it is therefore possible that it will be assigned to a class of reaction possessing a low Arrhenius factor when in fact the true Arrhenius factor is normal.There is much published work dealing with unimolecular gas-phase decompositions consequently our selection is not comprehensive. The examples chosen are those that we consider to be well established. In two cases we use apparently reliable results to indicate the need for further investigation. 66Normal” Arrhenius Factors We shall consider that Arrhenius factors lying between and sec.-l are “normal” and subdivide the available results into two maj or groups. (a) Compounds decomposing by a free-radical mechanism. We can represent such decompositions by the scheme A-B-+ A. + .B. Many reactions of this type have been investigated for when the reverse reaction has zero activation energy the activation energy of the decomposition reaction can be identified with the bond-dissociation energy D(A-B).Szwarc et al.l have demonstrated that many such decompositions occur and that the Arrhenius factor is usually “normal”. Before we summarise some of these data it is necessary to make one reservation. On occasion workers have assumed a constant value of the Arrhenius factor for a series of compounds and calculated activation-energy differences from the rate constants obtained at one particular temperature. Butler and Polanyis introduced the use of a 1013 factor in their attempts to determine D(R-I) values in alkyl iodides from first-order velocity constants based upon decomposition of the iodides to give iodine. They acknowledged that their procedure was a considerable over-simplification but considered the de- rived energy value to be better than that obtained from log k-l/T plots.They realised that the activation energy derived from the use of the 1013 factor was relatively insensitive to 50% changes in k. In fact it now seems possible that the good agreement of their D values with modern values is largely fortuitous and that in the pyrolysis of alkyl iodides the first reaction is a dehydrohalogenation step.g Many bond-breaking reactions have been studied and when the Arrhenius factor is normal the activation energy has been identified with the bond-dissociation energy. Thus a pattern of bond-dissociation energies * Butler and Polanyi Trans. Furuduy Soc. 1943 39 19. @ Holmes and Maccoll Proc. Chem. SOC. 1957 175. GOWENLOCK ARRHENIUS FACTORS 137 in organic molecules has emerged (see Sehon and SzwarclO and Cottrellll for detailed considerations) and the self-consistency of the results has been shown for the derivation of heats of formation of free radicals and the prediction of bond-dissociation energies.However it is necessary to add a word of caution in interpreting these measurements. Kinetic schemes can be oversimplified and from apparently small assumptions important consequences can arise. A good example is provided by the controversies over D(C,H,.CH,-H). The original value12 of 77.5 kcal. mole-l has frequently been taken as the basis for a bond- dissociation energy pattern and the fact that the Arrhenius factor for the toluene pyrolysis was 2 x 1013 sec.-l probably served as additional verifica- tion. Blades Blades and Steacie13 obtained however an activation energy of 90 kcal.mole-l and an Arrhenius factor of 5 x 1015 sec.-l on repeating Szwarc’s work.Although we may accept the argument of Sehon and SzwarclO that the discrepancies are due to side reactions consequent upon the greater percentage decompositions employed by Blades et al. yet we must still note that the new revised valuelo of D(C,H,.CH,-H)= 83 kcal.mole-l implies that the Arrhenius factor for the decomposition of toluene (and the xylenes?) is greater than 2 x 1013 and probably about 3 x lo1* sec.-l. Such a value is within the “normal” range. (b) Compounds decomposing by a molecular mechanism. Many gaseous compounds undergo thermal decomposition without production of free radicals. Such a decomposition mechanism may be considered as estab- lished when over a wide range of experimental conditions the composition of the products is constant and the reaction rate and products are un- affected by typical free-radical removers (e.g.nitric oxide toluene propene). Typical examples of a molecular decomposition are provided by dehydrohalogenation reactions investigated by Maccoll et aL9J4 Howlett,15 and Barton? Thus the most probable mechanism is as in scheme (1). The transition state is therefore a four-centre type associated with a bending mode in which a hydrogen and a halogen atom come to- wards one another. The activation energies fall in the 40-60 kcal.mole-l range and the Arrhenius factor is normal (log,,A = 12.6-14.6). lo Sehon and Szwarc Ann. Rev. Phys. Chem. 1957 8,439. l1 Cottrell “The Strengths of Chemical Bonds,” Butterworths London 1958. l2 Szwarc J. Chem. Phys. 1948 16 128.l4 Green Harden Maccoll and Thomas J. Chem. Phys. 1953 21 178. l5 Howlett J. 1952,4487. l8 Barton er uf. J. 1949 165; 1951,2039; Trans. Furuduy SOC. 1949,45 725; 1950 Blades Blades and Steacie Cunud. J. Chem. 1954,32 298. 46 114. 2 138 QUARTERLY REVIEWS Six-centre transition states are exemplified by the thermal decomposi- tion of propyl vinyl ether (log,,,A = 12.6) (see scheme 2).17 H3C- CH=CH Y ? H 3C-..C-C ... r r H3C-y-y \ H. k,:O - + . . . . (2 H H / o - H-C-.C H3C-CHO H,C = c H’ H “Abnormal” Arrhenius Factors We shall consider reactions where the Arrhenius factors are less than 1011s5 or greater than 1014s5 to be “abnormal” and we shall subdivide the available results into four major groups. (a) Entropy of activation less than -7 e.u. The transition-state theory of reaction kinetics can be formulated in a thermodynamic fashion.When this is done the expression for the rate constant for a unimolecular reaction becomes k = e. (kT/h) exp (- E/RT) exp ( AS/R) and thus the Arrhenius factor A is equivalent to e. (kT/h) exp ( A S / R ) Thus if the entropy of activation lies between -7 and +7 e.u. then the Arrhenius factor will be “normal”. This implies that the restrictions on motion in the transition state will be only slightly different in these cases from those in the initial state. However when the entropy of activation is lower than -7 e.u. then we can expect the transition-state configuration to have significantly greater rigidity than the initial state similarly when the entropy of activation is greater than +7 ex. we can expect the transi- tion-state configuration to have much more freedom of vibration than in the initial state.We shall examine some cases for each of these classes where molecular models indicate that such an argument is most plausible. We shall cite a few examples for a negative entropy of activation. Thus the thermal decompositions of many esters have a large negative entropy of H O 0 R’ 0 R’ - ‘c4 + ,c . . . . . . (3) h0.R‘ t0.d activation and therefore a transition state with restriction on freedom of rotation. Coffin et aZ.,18 showed that for esters of the type R-CH(OCO.R‘), whose decomposition mechanism is presumably as scheme (3) the l7 Blades Canad. J. Chem. 1954 31 418. Coffin et al. Canad. J Res. B 1931,5 636; 1932,6,417; 1937 15 229,247,254 260. GOWENLOCK ARRHENIUS FACTORS 139 Arrhenius factor lies between 109m2 and 101l.l corresponding to entropies of activation of -18 to -10 e.u.To form the cyclic transition state a considerable restriction of motion of the reactant is necessary. Reactions with large negative entropies of activation are usually associated with cyclic transition states. Examples are to be found in the work of Murphy who postulates six-membered-ring trznsition states for the reactions shown in Table 1. TABLE 1. Reactions with 6-centre cyclic transition states. Reaction Ref. 19 20 CH,:CH*CH,*O*CH:CH -+ CH,:CH.CH2.CH2.CH0 CH,:CH*O*C2H -+ CH,:CH + CH,*CHO CH,:CH*CH,*CH,*CO.CH 21 C H C H * C H * 0. C( C H 3) C H -+ Transn. state logloA Entropy of activn. (e.u.) -7.7 - 10.2 -7.7 One further example will be mentioned namely the racemisation of 2,2’-diamino-6,6’-dimethylbiplienyl for which the Arrhenius factor is 2.35 x lolo.This is rather low for it is difficult to understand how the transition state can have a restriction on rotation that is not present in the initial state. Kistiakowsky and Smithz2 suggested that the Arrhenius factor was low owing to the rotation of the heavy groups that was necessary to achieve the transition state. It seems more likely that this would be re- flected in the activation energy than in the Arrhenius factor and we shall have occasion to refer to this suggestion for relatively low values again in relation to cis-+trans-isomerisation. (b) Entropy of activation greater than +7 e.u. This group of Arrhenius factors (> 101*s5) indicates that the transition state is “looser” than the initial state and thus allows for example more rotational freedom than is permissible in the ordinary molecule.We can illustrate these statements by considering some decyclisation reactions (which proceed by a molecular mechanism) and by the free-radical decomposition of some compounds l9 Schuler and Murphy J . Amer. Chem. SOC. 1950,72,3155. 2o Blades and Murphy J. Amer. Chem. SOC. 1952 74 1039. 21 Stein and Murphy J. Amer. Chem. SOC. 1952,74 1041. 22 Kistiakowsky and Smith J. Amer. Chem. Soc. 1936 58 1043. 140 QUARTERLY REVIEWS which are known to have hindered internal rotation. We shall select one example of the breaking of a three-membered ring where the Arrhenius factor is 1015.17 s e c . - l ~ ~ ~ It is plausible to suggest that there is greater freedom of torsional movement in one C-C bond in the activated state.There are many kinetic studies of the breaking of four- membered rings. Those which decompose into two fragments are sum- marised in Table 2. It is noticeable that when the molecule contains either a carbonyl or an ether group the Arrhenius factor is lowered. Possibly the freedom of rotation in the transition state is less when an oxygen atom is present. TABLE 2. 4-Centre decyclisation reactions. Compound Products log,& E (kcalmole-l) H2C-CH I I H2$ + p 2 H2C-CH H2C CH 15.6 H2C-CH H2C CH 15.56 Et HY-FH2 EtHf + $H2 62.5 62.0 MeCO.HY-p tvleCO.H$ + FH2 H,C-CH H& CH 14-53 54.5 14.56 H-.fi tS.* H2C CH 52.0 14.79 60.0 H2C-0 I I H2fi +B H2C-CH H2C a 2 15.95 F2C-CF I I F2F +F F2C-CF2 F,C CF2 74.1 Ref. 24 25 26 27 28 29 Examples of high Arrhenius factors in molecules decomposing by free- radical mechanisms have been found for dialkyl peroxides.It is suggested30 that this can be accounted for by the entropy increase consequent upon free rotation about the 0-0 bond in the activated complex such rotation 23 Chambers and Kistiakowsky J. Amer. Chem. SOC. 1934 56 399. 24 Genaux and Walters J. Amer. Chem. SOC. 1951 73 4497; 1953 75 6196. Wellman and Walters J. Arner. Chem. SOC. 1957 79 1542. as DaignauIt and Walters J. Amer. Chem. SOC. 1958 80. 541. Das Kern Coyle and Walters J. Amer. Chem. SOC. 1954 76 6271. Bittker and Walters J. Amer. Chem. SOC. 1955 77 1429. 29 Atkinson and Trenwith J. 1953 2082. so Hoare Protheroe and Walsh Trans. Furaday SOC. 1959,55 548. GOWENLOCK ARRHENIUS FACTORS 141 being impossible in the ordinary molecule.Similarly the Arrhenius factors obtained for the following decompositions CH,*CO.CO.CHs -t 2CHa.CO Ph-CH2-C0.CH3 -+ Ph-CH + CHsCO Ph*CO-O*CH,-Ph -+ Ph.CO.0 + CH2*Ph PhCO-COPh -+ 2PhCO may reflect the freedom of rotation of both radicals in the activated complex.31 Values have been derived for the Arrhenius factors for the thermal decompositions (unimolecular) of ethane32 and neopentane :33 CHS-CHS + CH + CH log,oA = 14.8-15’7 CHa-C(CH3)3 -t CH3 + C(CHa)a logloA = 17 Leigh Szwarc and Bigelei~en,~~ however regard their values for ethane as preliminary and subject to some experimental error and Engel et aLs3 have derived their values from a Rice-Herzfeld mechanism and it may therefore contain errors involved in the determination of the Arrhenius factors of two reactions.(c) Low values for the transmission coeficient. We have already mentioned that the rate constant for a unimolecular reaction can be shown to be exp (- Eo/RT). In most cases K the transmission coefficient is approximately unity but in cases where there is a change from one type of electronic state to another the transmission coefficient is low reflecting the low probability of crossing between energy levels of different multi- plicities. When therefore reaction mechanisms of this type occur it is to be expected that the Arrhenius factor of such reactions will be abnormally low. About 25 years ago a number of gas-phase cis- -+ trans-isomerisations were found to possess low Arrhenius factors and it was consequently suggested that the reaction mechanism was of the following type -t R P‘ R,.2’ H R’ 5-F ,c=c - H H H H Singlet Triplet Singlet In the transition state the planar R-C-H groups are mutually perpen- dicular. As this reaction mechanism requires two transitions between states of different multiplicity it is to be expected that a low Arrhenius factor will be observed. However as Trotman-Dickenson has pointed the experimental evidence for this mechanism is not unambiguous and in one case (but-2-ene) the early work has now been shown to be transit ion state Szwarc Discuss. Faraday SOC. 1953 14 125. se Leigh Szwarc and Bigeleisen J. Amer. Chem. SOC. 1955 77 2193. Engel Combe Letort and Niclause Compt. rend. 1957 244 453. Trotman-Dickenson “Gas Kinetics,” Butterworths London 1955. 142 QUARTERLY REVIEWS invalid.35 If it were necessary to support low values of the transmission coefficient upon the evidence from gas-phase reactions we should have insufficient grounds for this.However the recent solution and liquid- phase evidence for the thermal isomerisation of dimethyl maleate and maleic acid36 suggests that the original singlet -+ triplet -+ singlet mech- anism is in fact the most likely reaction path as log,,A varies between 4.2 and 6.1. This would correspond to probabilities of - or for each of the two transitions. Further evidence for low transmission coefficients in unimolecular reactions is desirable. We may also remark that the majority of cis- -+ trans-isomerisations proceed by a mechanism which does not involve multiplicity changes but presumably occurs by a twisting mechanism about the C=C bond. It is therefore a little surprising that many of the so-called normal isomerisations possess Arrhenius factors which are surprisingly low for gas reactions where the transition state cannot be claimed to have restrictions on rotation that are absent in the initial state.A few examples are cited in Table 3. \ / / \ TABLE 3. Compound log, A (sec.-l) E (kcal.mole-l) Ref. cis-But-2-ene 13.8 62.8 35 cis-Stilbene 12.8 42.8 37 9 9 (liquid) 10.4 36.7 38 Me cis-cinnamate 10.6 41.6 39 cis-/I-Cyanostyrene 11.6 46 40 Kinetic parameters.for cis -+ trans-isomerisation. It has been claimed by Kistiakowsky and Smith*O that theselow Arrhenius factors are due to rotation of the heavy groups from their position in the initial state to the transition state (compare p. 139). If the two possible reaction mechanisms (i.e. “forbidden” and “normal”) were operating together then it is quite feasible for the Arrhenius factor to be lower than the expected 1013.The rate constants for the singlet -+ triplet + singlet mechanism [k - lo5 exp (-25,000/RT)] are approximately the same as those given in Table 3 and a slight curvature of the log k-l/Tplot will result if the two mechanisms operate together. The available data are insufficient to detect such a curvature. (d) Fission into more than two fragments. Until recently it was assumed that when a compound decomposed by a free-radical mechanism the primary reaction involved fission of one bond and the subsequent 35 Rabinovitch and Michel J. Amer. Chem. SOC. 1959 81 5065. 36 Davies and Evans Trans. Faraday Soc. 1955,51 1506. 37 Kistiakowsky and Smith J. Amer. Chem.SOC. 1934 56 638. 38 Taylor and Murray J. 1938 2078. 3D Kistiakowsky and Smith J. Amer. Chem. Soc. 1935,57 269. 40 Kistiakowsky and Smith J. Amer. Chem. Soc. 1936 58 2428. GOWENLOCK ARRHENIUS FACTORS 143 production of two free radicals. However observation of the pyrolysis of di-isopr~pylmercury~~ furnished evidence that two-bond fission into three fragments was a feasible mechanism. This suggestion was made from comparison with the analogous cases of dimeth~l-~ and diethyl-mer~ury~~ and from knowledge of the thermochemistry of mercury dialkyls. Table 4 summarises the data. The approximation equivalence of the activation TABLE 4 R in R,Hg logl,A E (kcal.mole-l) D,+D (kcal.mole-l) Me 13.5 51.5 & 2 59 rf 4 Et 14.1 42.5 & 2 50 & 6 Pri 16.7 40.4 1 41 * 7 energy with the energy required to break both mercury-carbon bonds for the case of R = Pri pointed to this different mechanism.The high value for the Arrhenius factor could not be explained in terms of a chain reaction or in terms of free rotation in the transition state. A suggestion by Hinshel- wood and his co-w~rkers~~ that high Arrhenius factors could correspond to spreading of the energy of activation into more than one bond was therefore adopted as the most likely explanation. This suggestion has also been adopted for a number of reactions where the Arrhenius factors are high. The reactions have a common mechanistic feature which can be represented by the two decomposition possibilities (fast) R2M -+ R e + *MR *MR -+ M + *R followed by . . Mode 1 or R2M -+ Re + M + *R . . . . . Mode2 the functional group M reverting from its bivalent to its zerovalent state.The cases where M = Hg N, and CO have been most extensively investigated. P r i t ~ h a r d ~ ~ has interpreted these reactions on the basis of the Fowler and Guggenheim equation:46 s- 1 k = X [egp (-E/RT)] 2 [(l/r!) (E/RT)‘] r=O which is taken to be the general expression for a unimolecular velocity constant. This rate constant is the product of X (a molecular constant) 41 Chilton and Gowenlock Trans. Faraday Soc. 1953 49 1451. 42 Gowenlock Polanyi and Warhurst Proc. Roy. Soc. 1953 A 218 269. 43 Carter Chappell and Warhurst J. 1956 106. 44 Peard Stubbs and Hinshelwood Proc. Roy. Soc. 1952 A 214 471. 45 Pritchard J. Chem. Phys. 1956,25 267; Clark and Pritchard J. 1956 2136. 48 Fowler and Guggenheim “Statistical Thermodynamics,” Cambridge University Press 1949 p.521. 144 QUARTERLY REVIEWS and the fraction of activated molecules. The condition for a molecule to be activated is that it should have an energy exceeding E distributed over s internal vibrations (or oscillators). On this basis by choosing a value for h of about 1013 sec.-l and using derived values for the critical energy of activation [E = D(RM-R) for mode 1 E = D(RM-R) + D(R-M) for mode 21 it is possible to obtain good agreement between the observed and the calculated value of the rate constant it must be assumed that the effective number of oscillators s contributing to the decomposition by mode 2 increases as the size of the group R increases. Extension to the ketone and peroxide series also resulted in a reasonable agreement between theory and experiment and it was concluded that high Arrhenius factors were therefore explicable in terms of simultaneous decompositions by two or more mechanisms.This initially attractive suggestion has been sub- jected to severe criticism by John~ton,~’ who pointed out that X must be much lower than 1013 sec.-l as the molecular complexity increased and that therefore the increase due to the summation term would be largely (if not completely) counterbalanced. These criticisms have been incorporated into a more developed form of Pritchard‘s theory by who utilises Kassel’s and Slater’s theories of unimolecular reactions. On the alternative bases of either lightly coupled or orthogonal oscillators where the critical energy is localised in only z of the s oscillators both approaches lead to the formula + ( z - l)! (jg)7 where A’ is the frequency of reaction of molecules which contain the critical energy localized with the z “critical” oscillators and will therefore be of the order of 1012-1014 sec.-l.This formula clearly leads to high Arrhenius factors and Steel has further considered the effect on the Arrhenius factors of necessary phase relationships of the critical oscilla- tors i.e. he has considered how the oscillators must not only have the necessary energy requirements but must also be undergoing extension at the same time for reaction to occur. We may therefore conclude that a satisfactory theoretical basis for some observed high Arrhenius factors is now emerging and that localisation of the activation energy in more than one mode is the essential feature of such a theoretical basis.Conclusion It appears from this Review that there is a necessary relation between the Arrhenius factor of a reaction and the reaction mechanism. In general 47 Johnston Ann. Rev. Phys. Chem. 1957 8 249. Steel personal communication; J. Chem. Phys. 1959 31 899. GOWENLOCK ARRHENIUS FACTORS 145 unimolecular reactions will have Arrhenius factors which fall within the range 10f1*5-1014*5 sec.-l. When reactions have Arrhenius factors smaller than 1011*5 sec.-l then an explanation may be forthcoming from either restrictions on motion in the transition state ( A S < - 7 e.u.) or a low transmission coefficient owing to the rate-determining step incorporating a “forbidden” transition. Reactions that possess large Arrhenius factors (greater than 1014a5 sec.-l) may involve either a “loose” transition state ( A S 2 + 7 e.u.) or the participation of a three- or four-fragment decomposition due to the activation energy’s “spreading out” into more than one bond. In all these reactions it is necessary for certain experimental conditions to be fulfilled before any correlation of mechanism with Arrhenius factor can be more than tentative. The Reviewer thanks Drs. N. B. Slater and C. Steel for useful communications and Drs. G. 0. and H. 0. Pritchard for a helpful discussion.

ISSN:0009-2681    

DOI:10.1039/QR9601400133

出版商:RSC    

年代:1960

数据来源: RSC

摘要:

PRIMARY PROCESSES IN PHOTO-OXIDATION By ROBIN M. HOCHSTRASSER PH.D. and GERALD B. PORTER PH.D. (DEPARTMENT OF CHEMISTRY UNIVERSITY OF BRITISH COLUMBIA) THIS Review is an interpretation of photochemical primary processes particularly oxidations by molecular oxygen in terms of simple molecular spectroscopy. It deals only with the intramolecular (unimolecular) and intermolecular (bimolecular) reactions of the electronically excited molecules formed as a consequence of absorption of radiation in the visible and the ultraviolet region of the spectrum. Secondary reactions e.g. reactions of free radicals formed by dissociation are not considered at all. Photochemistry cannot be divorced from spectroscopy and particularly not when primary reactions are considered. The sound principles upon which molecular spectroscopy is based provide a good foundation for the interpretation of photochemistry.Absorption and emission spectroscopy establish the nature and properties of electronically excited molecules. Hence by this method the photochemist can take a major step toward the interpretation of his results. It is our purpose first to present some basic principles of molecular spectroscopy which pertain to the photochemical primary process in general and then to discuss examples of photo-oxidations which have been most widely studied in the gaseous liquid or solid phase. Other reviews in this field include the following Forsterl has reviewed the primary processes generally while No yes Porter and Jolley2 have considered the specific cases of ketones of low molecular weight; Simons3 has discussed the reactions of electronically excited molecules in solution ; and photo-oxidation processes have been reviewed by Etienne4 and by Bergmann and M~Lean.~ 1.The excitation process The essence of a photochemical reaction is that the activation energy of the reaction is supplied by light energy. If visible light is used this energy corresponds to about 40 kcal./mole. The electronic configuration of the excited species is not in general the same as that of the parent unexcited molecule. Molecules may interact with electromagnetic vibrations of a particular frequency which varying from substance to substance depends on the nature of the electronic structure of the molecule. Saturated mole- cules normally do not interact with visible or near-ultraviolet light. Molecules containing n-electrons on the other hand usually absorb throughout this range.Consequently the latter type of molecule which Forster 2. Elektrochem. 1952 56 716. Simons Quart. Rev. 1959 13 3. Etienne ‘‘Trait6 de Chimie Organique,” 1944 Vol. XVIT Masson Paris p. 1299. 146 2 Noyes Porter and Jolley Chem. Rev. 1956 56 49. ti Bergmann and McLean Chem. Rev. 1941,28,367. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 147 includes aromatic hydrocarbons their heterocyclic analogues aldehydes ketones etc. is the type most accessible to experimental study. These molecules have an even number of electrons. All the bonding orbitals are completely filled with electrons according to the Pauli principle when the molecule is in the ground state.* Thus these molecules have no resultant spin angular momentum s which is equivalent to all electrons being paired.The multiplicity of a state is given by the rule Multiplicity = 2s + 1 In the case where s = 0 the multiplicity is one and the state is a singlet state. All the molecules under consideration have singlet ground states. The probability of changing the resultant spin momentum during an electronic transition is zero if spin-orbital interaction is neglected. In discussions of singlet ground states the only other multiplicity which may arise (through violations of the above rule brought about by spin-orbital perturbations) is three and the resulting state is a triplet. This is shown in Fig. 1 where (a) indicates the unexcited state and (b) the excited singlet in ground state 1 + s- 0 s=o S=/ (0) (4 (c 1 FIG.1. Schematic representation of the orbital structure of a molecule (a) in the ground singlet state (b) in thefirst excited singlet state and (c) in the triplet state. The diagram does not give accurate relative energies. The energy of the first singlet state above ground is not simply the diference in energy between the highest occupied and lowest unoccupied orbitals. The interaction of the excited and unexcited configurations will affect the energy of the state. state. Inversion of the spin of the excited electron results in the formation of a state (c) with two unpaired electrons which is a triplet (2s + 1 = 3). The next possible multiplicity the quintuplet never arises in practice as the transition probability is prohibitively low so low in fact that such states have never been observed in emission or absorption for any polyatomic molecules with an even number of electrons.The frequency which excites a molecule to an upper state is predeter- mined by the relative energy spacings of the occupied and the unoccupied orbitals of the molecule. In all the molecules considered in this Review the lowest unoccupied molecular orbital is an anti-bonding .rr-orbital. Excita- tion involves the spontaneous promotion of an electron from an occupied orbital into this anti-bonding orbital. The intensity of the absorption of light by an assemblage of molecules is an experimental measure of the * This statement is true only if the highest occupied level is non-degenerate. 148 QUARTERLY REVIEWS probability of the promotion of an electron to any given level. The more intense the absorption the more probable is the transition between the states.There are three main types of excitation (i) The excited electron may originate from a bonding n-orbital in which case the resulting electronic state is designated a l(r,n*) state (read as “singlet pi-pi star”). In this case the bonding in the upper state will be different from that in the ground state because there is one less bonding electron in the former. The transitions in the ultraviolet and the visible region of the spectrum exhibited by con- jugated aromatic hydrocarbons and polyenes involve excitation to states of this type. (ii) The excited electron may originate from a bonding a-orbital. The resulting state is a l(u,n*) state (singlet sigma-pi star). Such states are not important in photochemistry as they are seldom the lowest singlet electronic states of molecules which absorb in the visible and near-ultraviolet region.(iii) The excited electron may originate by promo- tion of an electron from a non-bonding orbital n into the n*-orbital. In molecules which contain oxygen or nitrogen the non-bonding pairs occupy n-orbitals. These n-orbitals often have higher energy than the occupied n-orbitals and consequently the lowest excited singlet states of such molecules are usually l(n,n*). The lowest excited singlet states6 of all aldehydes ketones quinones and most N-heteroaromatic compounds are l(n,n*).t Excitation of this type leads to a configuration which has n- bonding properties less removed from those of the ground state than arise from n+n*-excitation; little distortion is then to be expected.This fact and other differences to be discussed in the next section lead to fundamental dissimilarities in the photochemistry of molecules whose lowest excited singlet states are l(r,n*) and l(n,n*). It is less convenient to use orbital-energy differences to represent types of excitation than to deal with energy states. The energy of a state above the ground state is essentially the orbital-energy difference. The energy of the electron in the antibonding orbital is lowered by configuration inter- action. The latter will however not be discussed in detail; the electronic interaction between the excited and the unexcited configuration results in a lowering of the energy of the now occupied antibonding orbital. . 2. The properties of (n,n*) and (n,n*) states Associated with each state is a characteristic natural life whose duration is inversely proportional to the strength of the absorption to that state.The strength of absorption is usually measured in terms of the molar 1 cl extinction coefficient E which is defined as E = - log (I,,/I) where c is the Kasha Discuss Faraday Soc. 1950,9 14. t Strictly an n-m* orbital promotion involves the excitation of a non-bonding electron from an orbital which is symmetric to the molecular plane to an antibonding n-orbital which is antisymmetric to this plane. However it is reasonable to extend this useful classification to nearly planar molecules in which the effective skeleton is planar. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 149 concentration in mole@ I is the length of the absorption path in cm.and log (Zo/I) is the absorbance of the solution such that (I/Z,,) is the fraction of the incident light which is transmitted. In general the lifetime can be measured experimentally or calculated from the magnitude of the integrated experimental absorption curve. The following approximation is often useful in the near ultraviolet region 7 0 (in sec.) = 10-4/~max In Fig. 2 typical values of 70 are given for both types of transition. The Singlet -S ing/et lo96 /ogf g Tronritions Forb idden(?/) FIG. 2. The relation between natural (radiative) lifetimes ( T ~ ) oscillator strengths (f) lifetime calculated in this manner is an upper limit since the actual life- time which is influenced by the environment will always be equal to or shorter than the natural radiative lifetime.After the initial excitation to the lowest singlet state the molecule loses its excitation energy in one or more of the following five ways :* and extinction coefficients (c) of electronic transitions. (i) fluorescence; radiative conversion into the ground state; (ii) internal conversion; non-radiative conversion into the ground state; (ii) intersystem crossing; non-radiative transition which involves a spin (iv) photochemical reaction; by unimolecular dissociation or by inter- (v) energy transfer non-radiatively to a neighbouring molecule. The yield of photochemical product depends on the rate of process (iv) in relation to the rates of the other processes. After intersystem crossing (iii) the molecule in a triplet state will follow one of the following courses (a) phosphorescence ; radiative inter-combination with the ground state;? * The terms used in this work to describe intramolecular energy conversions are those defined by Kasha.$ t A combination is taken to mean a spectroscopic transition between two states of the same multiplicity while intercombination is reserved for spectroscopic transitions between states of different multiplicity.These terms may be used with respect to radia- tive and non-radiative transitions. intercombination to the triplet state; molecular reaction; and 150 QUARTERLY REVIEWS (b) internal conversion; non-radiative inter-combination with the ground state (intersystem crossing is retained for use in spin-orbital dependent processes) ; (c) photochemical dissociation or reaction; or (d) triplet-triplet energy transfer ; non-radiative transfer of electronic energy to a nearby molecule.There are two other processes to be considered which arise when the molecules are excited to other than the vibrationless level of the first excited state. At a somewhat higher frequency than that of the zero-zero band the molecules will reach the first excited state with an excess of vibrational energy. In solids liquids and gases at high pressures the perturbing effect of neighbouring molecules rapidly degrades the excess of vibrational energy to thermal energy. Unless there are extremely rapid competing processes vibrationally excited states rapidly revert to the excited state in vibrational equilibrium with the surrounding medium. Excited molecules are isolated from collisional degradation only in gases at low pressures.Dissociative reactions will be enhanced under these conditions when vibrational energy is required for dissociation. At still higher frequencies excitation is into higher electronic states. In most molecules again in solids liquids and gases at high pressure internal conversion between excited singlet states (or in general between excited states of the same multiplicity) occurs very rapidly. The excess of energy is dissipated as thermal energy in the surroundings. Since internal conversion occurs between isoenergetic states the process must involve conversion of electronic energy into vibrational energy followed by rapid degradation of the type discussed in the previous paragraph. Hence even when a molecule is initially raised to a highly excited electronic state it rapidly reverts to the vibrationally equilibrated first excited state." As a consequence of these rapid radiationless processes the molecule usually has only one emitting level of a given multiplicity and that is the lowest excited level of that multiplicity.In the gaseous phase at low pressure it may not be possible to drain off the excess of vibrational energy and only in this case would vibrational energy be able to influence the rate of a photochemical reaction. One of the fundamental problems in chemical kinetics concerns the degradation of vibrational energy i.e. the number of gas-phase collisions required to remove a certain quantity of vibrational energy from a vibrationally excited molecule. All that is known is that more than one collision is necessary to degrade such energy completely.Thus in solutions in gases at pressures greater than about 50 mm. of Hg and in solids and rigid glasses the quantum yield of the unique emission of a molecule is almost invariably independent of the frequency of the exciting light. This implies that the emitting state is the only one which is important in photochemical reactions in such media. * The only confirmed exception is azulene in which fluorescence occurs only from the second excited state (Viswanath and Kasha J. Chem. Phys. 1956,24,574). HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 15 1 In Fig. 3 the ground singlet state first excited state second excited state etc. are designated So S, S2 etc. Consider a substance which has a fluorescence yield of 0-5. When x molecules are raised to S, x/2 of these FIG.3. Electronic energy state diagram for molecules with (a) (~T,T*) states and (6) (n,r*) states. The numbers quoted are unimolecular lifetimes (seconds) under conditions of thermal quenching at room temperature. Those inparentheses refer to theprobable values in the presence of heavv-atom solvents or paramagnetic species. The full lines represent radiative processes and the broken lines non-radiative processes. emit S1+So luminescence. When x molecules are raised to S2 it is still found that x/2 emit S1-+So luminescence and that this emission is identical with the first. Within the limits of experimental error all quanta which are lost during the processes of excitation and emission are lost from the lowest state. In other words the processes responsible for the observation of a luminescence efficiency less than unity occur in times comparable with the fluorescence lifetime but are very slow compared with the internal conversion and vibrational energy redistributions which occur when the excitation is to a higher electronic state than the first excited state.There are a few exceptions to this behaviour. If the molecule contains a group or bond in which the higher excitation energy may be localised and whose absorption is continuous e.g. the absorption of visible light by a molecule with a C-Br bond there will be dissociation if there is mixing of the C-Br continuum with the discrete states of the molecule. This is pre- dissociation. The extent of the mixing is likely to be different in each discrete state. Thus the fluorescence yields of such molecules will not be the same when the excitation is into different states.Photodissociation may compete with internal conversion between excited states. Examples which illustrate the influence of photodissociation on luminescence yields are reported in the section on the photo-oxidation of gases. Recently Ferguson' has discussed the likelihood of intersystem crossing between states other than S and T (the lowest triplet state) (see Fig. 3) in molecules substituted by heavy atoms. Such crossing should result in a variation of the fluorescence yield with excitation energy. A similar variation of the fluorescence yield is observed with some molecules of the Ferguson J. Mol. Spectroscopy 1959 3 177. 152 QUARTERLY REVIEWS ethylenic type. This has been attributed to rapid twisting in the upper states.8 However these examples are definitely exceptions to the rule and unless the molecule under study contains one of these features which causes intersystem crossing in upper states the yield of fluorescence will be independent of the exciting energy in condensed media.Internal con- version between the emitting singlet level and the ground state is a relatively improbable process otherwise fluorescence would seldom be observed. Spin intercombinational processes such as intersystem crossing are of particular importance when photo-oxidations are considered. In an isolated molecule the crossing from one potential-energy surface to another of different multiplicity becomes possible owing to intramolecular perturb- ation of the electronic state.This involves the interaction of the spin and the orbital motion of the excited electron. The details of spin-orbital interactions will not be dealt with here. There are available non-mathe- matical discussions by Kasha,6 by Gilmore Gibson and M~Clure,~ and by Reid,lo while more detailed treatments are those of McClure,ll who deals with aromatic molecules and Clementi and Kasha,12 who treat heterocyclic systems. The reader is also referred to Sidman's review13 where carbonyl compounds are discussed. The rate of intersystem crossing is enhanced by the presence of heavy a t o m ~ ~ J ~ or of paramagnetic ions15 or molecules.l* Oxygen falls in the last class with the result that most substances have a lower quantum yield of fluorescence in the presence of oxygen than in its absence.The paramagnetism of oxygen appears to be responsible for this luminescence quenching since nitric oxide produces comparable effects in the cases which have been so far ex~1rnined.l~ The maximum extinction coefficients for allowed transitions are of the order of lo4 and may exceed lo5 for the more intense absorptions. This implies that the radiative lifetimes for l(n,n*) states are 10-* to second. A photochemical process which involves these radiative states can have an appreciable quantum yield only if the primary process occurs within times of this order of magnitude. For n+* orbital promotions the maximum extinction coefficients range from lo1 to lo3 depending on whether the transition is forbidden or allowed by local symmetry i.e. the orbital symmetry in the region of the hetero-atom.The radiative lifetimes will range from 10-4 to lo-' second. In this case the primary photochemical step can be much slower than that for molecules reacting in l(n,n*) states and yet occur with relatively high yield. Fig. 3 further illustrates these correspondences. Hochstrasser Canad. J. Chem. 1959,37 1367. @ Gilmore Gibson and McClure J. Chem. Phys. 1952,20 829. lo Reid Quart. Rev. 1958 12 205. l1 McClure J. Chem. Phys. 1952 20 682. la Clernenti and Kasha J. Mol. Spectroscopy 1958 2 297. Is Sidman Chem. Rev. 1958 58 689. l4 McClure J. Chem. Phys. 1949 17 905. Is Clementi and Kasha J. Chem. Phys. 1957,26,956. l6 Bowen and Metcalfe Proc. Roy. Suc. 1951 A 206,437. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 153 There are two other differences between (n,n*) and (n,n*) states which are important for an understanding of their behaviour in photochemical reactions.First the energy separation between the l(n,w*) and the 3(n,n*) states is much less than the corresponding separation for (n,n*) states.17 The (n,m*) triplet should lie fairly close to the (n,n*) singlet because of the greater distance between the two unpaired electrons in the triplet state. One of these electrons is in an essentially non-localised antibonding T- orbital while the other is at the heteroatom in the localised lone-pair orbital. This means that although the lowest singlet state of a molecule is l(n,n*) the lowest triplet state may be 3(~,n*). This situation probably exists for most nitrogen-containing heteroaromatic compounds.18 Secondly the probability of intersystem crossing for molecules whose lowest singlet state is l(n,w*) is considerably higher than if the lowest state were l(n,w*).This has been shown to be a consequence of the longer radiative lifetime associated with l(n,n*) states,12 but may be due partly to the proximity of the singlet and the triplet levels. As a result fluorescence is either absent or very weak when the lowest excitation energy involves non-bonding elec- trons. Fig. 3 illustrates these points of difference in terms of the known rate constants for the interstate processes. Although the lowest excited state of a molecule is l(n,n*) the next higher electronic level may be l(w,n*). Thus for example in the vapour-phase photolysis of acetone,19 the products of primary dissociation are different for excitation into different electronic states where the bonding configurations differ appreciably.Considerations of this nature will not be so important in media where the internal con- version processes are facilitated i.e. vapours at high pressure solutions and solids. 3. Primary reactions Detailed examination of the nature of luminescence quenching will aid considerably in the elucidation of the course of primary photochemical processes. The quenching of potentially luminescent singlet states can be subdivided according to the products which result from the quenching process (a) the triplet state; from the enhancement of intersystem crossing; (b) the ground state; from an increase in the rate of internal conversion; and ( c ) photoproducts ; by direct reaction. Oxygen quenches the fluorescence of most molecules efficiently and presumably it is mechanism (a) that is mainly involved.The efficiency of the radiationless conversion from the lowest triplet state into the ground state is usually higher than that of the radiative process as is made evident by the absence of phosphorescence from virtually all molecules in the vapour phase or in solution. The internal conversion process is then markedly viscosity-dependent as illustrated by the appearance in general of l7 Reid J. Chem. Phys. 1953 21 1906. lS Ells and Noyes J. Amer. Chem. SOC. 1938,60.2031. Goodman and Harrell J. Chem. Phys. 1959,30 1131. 154 QUARTERLY REVIEWS phqsphorescence in solutions in rigid glasses. The three known exceptions to this are biacetyl thiobenzophenone,20 and acetone,2 which phosphoresce with relatively high quantum yields even in the vapour phase.Fig. 4 represents the nature of the photochemical primary processes which take place in a system containing oxygen after the absorption of FIG. 4. Primary photochemical processes. The encircled numbers are reactions referred to in the text. R is the molecule in its ground state. Excitation by light of frequency v1 produces lRvib* the molecule in its lowest excited singlet state with vibrational energy in excess of equilibrium. lR* is the same state in vibrational equilibrium with the surround- ing medium. Light of frequency v 2 excites R to 'Rvib** the molecule in a higher singlet electronic state. and 3R* represent the molecule in its lowest triplet state with and without excess of vibrational energy respectively. The values of the lifetimes are in seconds.light. The general scheme relates to gas- or liquid-phase oxidations although the values of some of the rate constants differ considerably in the two cases. In the gas phase the rate of removal of the excess of vibrational energy depends on the collisional frequency and hence on the gas pressure. The absolute rate of a photoreaction such as dissociation may depend on the vibrational-energy content and the nature of the distribution of this energy throughout the many vibrational modes of the molecule. Thus photo- chemical reactions in the gas phase have rates which depend on the excitation energy. Unless dissociation occurs the excess of vibrational energy will be removed after a few collisional periods and the molecule will remain in a nearly vibrationless or thermally equilibrated electronic excited state during the lifetime of that state.Hence in gas phase systems steps 1 and 2 of Fig. 4 are essential primary processes. In solutions the excess of vibrational energy is usually removed by solvent collisions within about second and the probability of photodissociation does not necessarily vary with the energy of excitation. Bimolecular reactions in solution have rates which depend on the number of encounters. The encounter rate is determined by the nature of the Franck-Rabinowitch cage of the solvent and is correspondingly faster than the collision rate between two species at the same temperature and ao Lewis and Kasha J Amer. Chem. SOC. 1944,66,2100. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 1 55 concentration in the gaseous phase.A typical upper limit for a bimolecular rate constant in solution is about 1O1O 1. mole-l sec.-l. At normal solute concentrations the encounter frequency is at least four orders of magnitude less than 1013 sec.-l (Le. >lo9 sec.-l). The result is that in solutions the rate of a bimolecular photoreaction is seldom determined by the amount of vibrational energy which is initially distributed throughout the molecule. Although step 3 may be important in the gas phase this is not the case in solution. However the case may arise where the doubly electronically excited molecule 'Rvib"" dissociates. The actual lifetime of this state is determined mainly by the rate of the non-radiative spectroscopic transition between it and the next singlet state 'Rvib*.This lifetime is of the order of sec. Such a rate constant (10l2 sec.-l) is not necessarily very fast compared with the dissociation rate. Thus one should observe rates of reaction which are constant when the excitation is within any one electronic band system. Steps 4-8 of Fig. 4 cover all the important fates of a singlet excited state. Steps 4 5 and 6 are the three intramolecular processes of internal conversion fluorescence and intersystem crossing respectively. For ~(T,T*) states these changes occur at approximately the same rate ca. sec.; for l(n,?z*) states the fluorescence lifetime is usually much longer ca. sec. Molecules with a lowest-lying l(n,n*) state will seldom show other than a weak fluorescence and the important steps for these molecules are Nos. 4 and 6 internal conversion and intersystem crossing.The oxygen quenching of a singlet excited state may in theory follow two courses. The first of these involves an enhancement of the rate of intersystem crossing through steps 7 and 8. In solution the singlet-triplet crossing rate may be increased by a factor of about a hundred by inter- action with oxygen. It is not known what type of collisions are involved in this catalysis. The intermediate state is written as 3R* - O2 to illustrate the proximity of the two species. It should be possible for this collision complex to give rise to oxidation products. The remaining reaction of the singlet state involves union with oxygen to give instantaneous complexes with the singlet state lR* * - 02. During intersystem crossing the molecule is expected to remain in the triplet state at an isoenergetic line of the two potential surfaces.After the appropriate collisional processes this excess of vibrational energy will be removed and the molecule will revert to a nearly vibrationless level (at normal temperature) of the lowest triplet state. These two processes are represented by steps 6 and 9 in Fig. 4. As before in the gaseous phase photochemical dissociation (step 10) will usually compete with collisional degradation (step 9) while in solutions the thermal equilibration time will usually be very short compared with the time required for dissociation. The effect of oxygen on the lowest triplet state of the molecule may be two-fold (i) photo-oxidation which is chemical union with conservation of spin step 12; (ii) acceleration of the internal conversion rate into the 156 QUARTERLY REVIEWS ground state which would arise because of the paramagnetic influence of oxygen.The latter which is represented by step 1 1 arises because the same perturbation which enhances the intersystem crossing probability may enhance the intercombination probability and hence reduce the lifetime of the triplet state. Internal conversion between the low-lying triplet and the ground state is a particularly difficult process to study experimentally. In order to discover whether this is a true spectroscopic spin-intercombination and hence sensitive to spin-orbital or paramagnetic perturbation a com- plete quantum account is necessary. These perturbations enhance the relative rate of intersystem crossing and at the same time the natural lifetime of the emitting triplet state is reduced owing to breakdown of the spin selection rules.Recent work by Porter and Wright21 has indicated that paramagnetic ions and molecules such as oxygen affect the rate of internal conversion into the ground state although the mechanism of this process is not yet clear. Phosphorescence step 13 usually occurs with a lifetime of sec. or less in the presence of oxygen or heavy atoms. Internal conversion step 14 occurs with high relative probability al- though the rate is very dependent on the properties of the medium e.g. the viscosity. Energy-transfer processes have been omitted from the scheme of Fig. 4. It will be appropriate to discuss them now. This topic has been critically reviewed by Terenin and Ermo1aev,22 although particular reference was not made to photochemical reactions.It has become apparent in recent years that energy of excitation may be non-radiatively transferred from one molecule to another provided the acceptor has energy states lower than the excited state of the donor. Such transfers are known to occur in gaseous liquid ’ and crystalline systems. The sensitisation of the #I- naphthylamine fluorescence by benzene vapour was originally noted by N e p ~ r e n t ~ ~ and detailed studies were later made by Dubois.24 There is little doubt that this system is an example of energy transfer in the vapour phase. Light absorbed by the benzene is emitted as p-naphthylamine fluorescence and both trivial reabsorption and vibrational stabilisation have been shown not to be responsible for the whole effect.In dilute solutions energy transfer from solute to solute and from solvent to solute has been shown to occur. In this respect aromatic hydrocarbons and quinones have been extensively studied.22 It has been demonstrated that such transfer of energy can occur over very large distances (compared with molecular dimensions) and that the transfer is non-radiative. These facts suggest that electronic energy transfer may be extremely important in some photochemical systems in the gaseous phase and in solution. The phenomenon of energy transfer is observed also in crystals. For 21 G. Porter and Wright Discuss. Faraday Suc. 1959,27,18. 22 Terenin and Ermolaev Uspekhi Fiz. Nauk 1956,57 37. as Neporent. Zhur. Fiz. Khim. 1950,24 1219. p4 Dubois J. Phys. Chem. 1959,63 8. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 157 example light absorbed by an anthracene crystal which contains a small amount of (0.01 mole%) of dissolved tetracene is emitted as tetracene l~minescence.~~ Thus the presence of small amounts of impurity in aromatic crystals may radically affect the results of photochemical studies and in particular the primary quantum yields.An impressive demonstra- tion of this occurred in studies of the conductivity of anthracene treated with tetracene.26 It is found that the trapping centres in the anthracene crystal are quenched i.e. the efficiency of trapping is reduced when the crystal contains dissolved tetracene. 4. Photo-oxidation in the gaseous state Of the many gas-phase photochemical oxidations which have been studied most involve primary dissociation into free radicals which sub- sequently react with oxygen.These reactions are outside the scope of a review on primary processes in photo-oxidation and will not be considered further except to note that the dissociation rate (step 1 of Fig. 4) is much faster than any other reaction which involves 'Rvib*. The processes to be discussed in this section include all interactions of molecular oxygen with electronically excited species whether or not such interaction leads directly to chemical reaction. In only two instances is there unambiguous evidence for direct photo- oxidation of excited molecules in the vapour phase namely for b i a ~ e t y l ~ ~ and keten.28 Less certain are the cases of acetone,29 cr~tonaldehyde,~~ and aniline.31 Oxygen quenches at least part of the luminescence of bi- a ~ e t y 1 ~ ~ a ~ e t o n e ~ ~ ~ ~ ~ aniline,35 hexafl~oroacetone,~~ IS-na~hthylamine,~~ anthracene,16 9-phenylanthracene and 9,lO-diphenylanthra~ene,~~ anthra- quinone anthranol anthrone and the 2-amino- 2-methyl- 2-ethyl- and 2-chloro-derivatives of anthraq~inone.~~ In general detailed study of the luminescence and photochemistry of compounds not appreciably volatile at room temperature has been confined to condensed phases which are discussed in the next sections.Studies of both photochemistry and luminescence are rare particularly for the vapour phase. 25 Bowen Mickiewitz and Smith Nature 1947 159 706. 26 Bryant Bree Fielding and Schneider Discuss. Faraday Soc. in the press. 27 G. B. Porter J. Chem. Phys. in the press. 28 G. B. Porter J. Amer. Chem.SOC. 1957 79 1878. 29 Marcotte and Noyes Discuss. Faraday SOC. 1951 10 236; J. Amer. Chem. SOC. so Blacet and Volman J. Amer. Chem. SOC. 1939 61 582. s1 Vartanyan J. Phys. Chem. (U.S.S.R.) 1940,3,467. s2 Groh J. Chem. Phys. 1953,21 674. s3 Groh Luckey and Noyes J. Chem. Phys. 1953,21 115. 34 Heicklen J. Amer. Chem. SOC. 1959,81,3863. 85 Vartanyan Izvest. Akad. Nauk S.S.S.R. Ser. Fiz. 1938 3 341. s6 Okabe and Steacie Canad. J. Chem. 1958,36 137. ST Dubois J. Chem. Phys. 1956 25 178. s8 Stevens Trans. Furaday SOC. 1955,51,610. 1952 74 783. Karyakin J. Phys. Chem. (U.S.S.R.) 1949,23 1332. 158 QUARTERLY REVIEWS Biacety1.-Because it exhibits intense phosphorescence regardless of the state of aggregation biacetyl has been studied There are two principal absorption regions in the visible and ultraviolet spectrum both of which are n+n*-transitions in which an electron is promoted from a non-bonding orbital on oxygen to a carbonyl antibonding n - ~ r b i t a l .~ ~ The extinction coefficients are low; the lifetime of the emission from the lowest l(n,n*) state is about second; hence the fluorescence yield is small. Photochemical dissociation (step 1 of Fig. 4) occurs provided the exciting light is of wavelength shorter than ca. 4000 A.40 The photochemical quantum yield i.e. the yield for the primary dissociation into free radicals decreases with increasing pressure after excitation to either electronic state. Degradation of vibrational energy increases with increasing pressure ; thus fewer molecules can dissociate and more will reach the lower vibra- tional levels from which fluorescence occurs.The “dissociation” reaction is rupture of one of the carbon-carbon bonds which in effect gives a mixture of methyl and acetyl radicals and carbon monoxide. The quantum yield of phosphorescence is quite high 0.145 at 4358 A.43 In this respect the behaviour of biacetyl is unusual compared with the maiority of molecules which luminesce in that phosphorescence is observed in gaseous liquid and solid phases. The molecule in its triplet state which is responsible for the phosphorescence can dissociate if it has vibrational energy equivalent to a temperature well above room temperature. Except at high intensities where triplet-triplet interaction leads to dissociation the triplet dissociation only becomes important above about 50°c and there is at this temperature a corresponding thermal quenching of the phosphorescence.The luminescence and the photochemical primary reactions of biacetyl in the absence of oxygen can be explained by steps 1,2,4 5 6,9 10 13 and 14 of Fig. 4. The fluorescence quantum yield is about 0.0024 at 4358 A where dissociation is negligible independent of temperature and pressure.44 This gives the expected order of magnitude for the apparent lifetime of the singlet state i.e. ca. second. At 3650 A and shorter wavelengths the primary quantum yield for dissociation is small and increases with decreasing pressure and decreasing wavelength of excitation. The data indicate the presence of internal conversion into the ground state from ‘Rvib* step 4 Fig. 4 with a relatively high quantum yield particularly at the longer wavelengths.In the presence of oxygen phosphorescence is quenched but fluorescence is unaffected as shown in Fig 5. The straight portion of the curve for biacetyl luminescence below mole 1.-l of oxygen as well as the entire straight line for anthracene luminescence represents the 40 Sheats and Noyes J. Amer. Chem. SOC. 1955 77 1421 4532. 41 Coward and Noyes J. Chem. Phys. 1954,22 1207. 42 Sidman and McClure J. Amer. Chem. SOC. 1955,77 6461. 48 Almy and Gillette J. Chem. Phys. 1943 11 188. 44 Okabe and Noyes J. Amer. Chem. SOC. 1957 79 801. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 159 quenching effect of oxygen. The luminescence in the second case is fluorescence. The luminescence of biacetyl in contrast to that of anthracene does not decrease without limit as the oxygen concentration o2 ( / ~ - ~ m ~ / e Jl) I I 0 20 40 60 o2 (/o-'mo/e/%) FIG.5. Fluorescence quenching by oxygen for anthracenels and biacetyLsa is increased but instead reaches a small constant value. This shows that the biacetyl luminescence has two components phosphorescence which is quenched by oxygen and fluorescence which is unaffected by oxygen. Oxygen does not interact with the singlet state so that steps 3 7 and 8 are of negligible importance. Oxygen does interact with the triplet state effi~iently.~' This interaction is chemical since photo-oxidation is observed in light of wavelength 4358 A in high yield while under the same conditions in the absence of oxygen there is no dissociation into free radicals. The quantum yields of products formed in the photo-oxidation are of the order of unity or higher under conditions such that no radical-chain reaction is likely.45 The yields are much higher than could be explained on the basis of the phosphorescence yield alone i.e.there must be internal conversion of the triplet state in the absence of oxygen for which the probability is about six times the probability of phosphorescence. Further internal conversion of the nearly vibrationless singlet state must be relatively unimportant compared with intersystem crossing. The effect of oxygen cannot be attributed to enhanced intersystem crossing since the fluorescence yield is unaffected by oxygen. With these data the values for the various specific lifetimes given in the Table can be estimated. Estimated lifetimes in the biacetyl system Step (Fig.4) 4 5 6 12 13 14 Lifetime (sec.-l) 10-7 10-5 10-8 2x (OJ* 2~ * Oxygen concn. measured in moles/l. 46 Taylor and Blacet Ind. Eng. Chem. 1956,48 1505. 160 QUARTERLY REVIEWS Keten.-The interpretation of the photo-oxidation of keten is less certain since no luminescence has been obser~ed.~**~~ The reaction at short wavelengths (ca. 2700 A) is undoubtedly of free-radical nature even in the presence of oxygen because of the rapid dissociation of excited keten molecules into methylene radicals and carbon monoxide. Near the long- wavelength limit of absorption excited molecules are both deactivated and oxidised by molecular oxygen. That the deactivation involves the 'Rvib" state is demonstrated by the negligible amount of primary dissociation in the presence of oxygen.28 The short-wavelength photo-oxidation always leads to formation of ethylene with a quantum yield of 0.67 while in light of wavelength 3650 A this yield is very small above about 50 microns pressure of oxygen.It is possible that the actual reaction of an excited molecule with oxygen which occurs at long wavelengths (and gives carbon monoxide and carbon dioxide among the products but no ethylene) may involve the triplet state of keten but in the absence of luminescence data it is not possible to determine the nature of the reacting species. This emphasises the tremendous advantage inherent in the study of photo- chemical reactions of molecules for which luminescence can also be examined. By analogy with biacetyl and other substances it can be tentatively concluded that this oxidation proceeds via the triplet state and further that the number of excited molecules which reach the triplet state is small at least in the presence of oxygen.Acetone.-The phosphorescence of acetone is strongly quenched by oxygen but fluorescence is ~ n a f f e c t e d . ~ ~ * ~ ~ ~ ~ ' Any interaction between oxygen and excited acetone molecules must involve the triplet state but not the singlet state. The studies of photo-o~idation~~ do not provide un- ambiguous evidence for direct chemical reaction between the triplet state and oxygen since the contribution of primary free-radical dissociation cannot be estimated from the available data. Fig. 6 shows that the reaction at room temperature is quite different from that at higher temperatures (where free-radical dissociation is probably complete).A comparison of the photochemical oxidation at long wavelengths with that at short wave- lengths is required to assess the extent to which oxidation of the excited molecule contributes to the overall reaction at room temperature. In the interpretation of primary reactions it is desirable in each case to have such photochemical-action spectra analogous to action spectra of luminescence. Po1yacenes.-In the examples discussed above optical excitation in the wavelength regions usually studied involves the formation of a l(n,n*) state. Oxygen has little or no detectable effect on these singlet states (except in the case of keten where the assignments are not known) but does interact strongly with the triplet. On the other hand the lowest states of polyacenes (i.e. linear assemblies of fused benzene rings) are (T,T*).Although most 46 Strachan and Noyes J. Amer. Chem. SOC. 1954 76 3258. 47 Heicklen and Noyes J. Amer. Chem. SOC. 1959 81 3858 HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 161 studies of aromatic hydrocarbons have been limited to the condensed phases those which have been studied in the vapour phase show strong oxygen-quenching of the fluorescence.16 As discussed in the previous I 0.1 0-3 0.5 0 (mm.) '2 FIG. 6. the photo-oxidation of acetone at 3130 @pco is the quantum yield ofproduc- tion of carbon monoxide one of the products of reaction. section this effect may be attributed to enhanced intersystem crossing brought about by the paramagnetic nature of oxygen. In these systems phosphorescence is weak and since the photochemical oxidation has not been studied the nature of the interaction between 3(~,7r*) states and molecular oxygen in the vapour is unknown.Even the much studied anthracene-oxygen system where a transannular peroxide is formed in solution has not been examined for photo-oxidation products in the vap our phase. 5. Photo-oxidation in solution So much work has been published in this field during the past decade that this discussion will be confined to aromatic hydrocarbons and their derivatives in organic solvents and to photosynthetic molecules typified by chlorophyll. The reasons for this restriction are two-fold a large volume of work has appeared on these topics whereas other studies have mostly been on isolated topics; and the ultraviolet spectra of these molecules are understood to the extent that we know a considerable amount about the nature of the low-lying energy levels that are important in bimolecular photoreactions.Anthracene and its Derivatives.-The molecule which has been most widely studied from the point of view of photoreactions in solution is anthracene. The linear polyacenes (anthracene tetracene pentacene etc.) combine with oxygen in the presence of light to form transannular peroxides the efficiency of this reaction increasing as the conjugated system is extended. The photo-oxides (bridged oxides) are only very weak 1 62 QUARTERLY REVIEWS peroxides and many of them do not liberate iodine from potassium iodide at an appreciable rate. The quantum yields of photo-oxidation and fluorescence are independent of the wavelength of exciting light over the ranges 2600-3650 and 2200-3650 A respectively.7*48 This indicates that rapid collisional degradation and internal conversion always occur until the thermally equilibrated vibrational levels of the lowest excited state are reached. Thus all the photochemical reactions of anthracene originate from this state and the relative quantum yields of each depend on the magnitude of the unimolecular or bimolecular rate constants for the subse- quent primary processes. The photoreactions of anthracene that have been studied include fluorescence photo-oxidation and dimerisation. Dimerisa- tion gives a compound which consists of two aiithracene molecules joined across the 9,1@positions lacks the typical unsaturation of anthracene and has an absorption spectrum resembling that of a substituted benzene.The lowest excited singlet state of anthracene has known symmetry (lL in the Platt notation or lBpu in group-theoretical notation) and the transition moment dipole is along the short axis of the molecule. The limiting fluorescence yield is 0.24 at very low an thracene concentrations and is not greatly affected by 9- and 10- substit~ents.~~ The phosphorescence from anthracene is extremely weak and appears in the near-infrared region of the spectrum.50 In order that the phosphorescence of anthracene dis- solved in a rigid glass may be observed a very high light-intensity and a long photographic exposure are necessary. In view of this the postulation of triplet states as reacting species present at appreciable steady-state concentrations also implies that the internal conversion from the lowest triplet (T, = second) to the ground state occurs with extremely high efficiency (k = sec.-l).In solvents such as benzene and hexane anthracene shows fluorescence in dilute solution and dimerises in concentrated solution. Photo-oxidation is very inefficient and is a function of the anthracene concentration but not of the oxygen concentration. The product of the photo-oxidation is a transannular peroxide. The quenching effect of oxygen is not directly related to the oxidation process. In solvents such as carbon disulphide and chloroform the fluorescence of anthracene is almost completely quenched and photo-oxidation is rapid. When no oxygen is present dimerisation does not readily occur. It has been shown that the results obtained may be fitted to a kinetic scheme in which (i) photodimerisation results from excited singlet molecules (ii) the photo-oxide is formed from excited molecules in the triplet state and (iii) concentration quenching is caused by the interaction of singlet excited molecules with unexcited molecules of the substance under consideration whether this be anthracene or one of its al kyl derivatives.51 48 Hochstrasser unpublished results. 4Q Popoviko DokIady Akad. Nauk S.S.S.R. 1950,71 453. 5 0 McGlynn Padhye and Kasha J . Chem. Phys. 1955,23 593. 51 Bowen Trans. Faraduy SOC. 1955 51 475 and earlier references therein. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 163 The participation of a triplet state in this reaction is not an arbitrary postulate. The sum of the quantum yields of fluorescence of photo- oxidation and of dimerisation increases with increasing concentration of the solute (except in the case of oxygen-free solutions).At high concentrations i.e. 2 x 10-2~ this sum is greater than the limiting quantum yield of fluorescence attained in oxygen-free solutions at low concentrations i.e. with no concentration quenching or oxygen quench- ing.52 The photo-oxidation of anthracene and some of its derivatives in carbon disulphide solution proceeds with a quantum yield close to unity although the limiting quantum yield for fluorescence in this solvent is very Therefore the processes leading to a low value of the limiting fluorescence yield do not involve direct transition from the excited to the ground state. Instead the presence of an intermediate active molecule incapable of fluorescence but able to react with oxygen is required.Further evidence that this state is in fact a triplet state is found in the flash- photolysis work of Porter and W i n d s ~ r . ~ ~ Irradiation of anthracene in oxygen-free hexane produced a metastable state (lowest triplet) from which absorption to a higher state (upper triplet) was observed. The half-life of the lower triplet state under these conditions 5.8 x sec. decreased enormously in the presence of small quantities of oxygen. The triplet- triplet energy separation agrees well with the calculated value for the anthracene One further instance of diagnostic evidence to support the participation of triplets in these reactions is provided by the nature of the products which arise from the irradiation of anthracene in carbon tetra~hloride.~~ The detection of 9-chloro- and 9,lO-dichloro- anthracene in the products suggests the participation of a diradical which is most likely to be a triplet state.Waters5’ argues that the reactions of anthracene in carbon tetrachloride provide evidence for the diradical nature of the triplet state. It is however not true either that a triplet state is a diradical or that a diradical is a molecule in its triplet state in the general case although in anthracene localisation of electronic charge at positions 9 and 10 satisfactorily accounts for the triplet-state energy derived spectroscopically. There are two facets of this type of photo-oxidation which still remain difficult to explain. The first is that the quantum yield of photo-oxidation is sensibly independent of the oxygen concentration but does depend on the concentration of anthracene.If the stable peroxide were produced by a bimolecular reaction between oxygen and anthracene in an excited singlet state the opposite dependence would be expected. This kinetic result is explained by postulating the existence of a reactive “moloxide” of 62 Cherkassov and Vember Opt. i Spekt. 1959 6 503. 63 Bowen Discuss. Faraday SOC. 1953 14 143. 54 G. Porter and Windsor J. Chem. Phys. 1953 21 2088. 66 Pariser J. Chem. Phys. 1956 24 250. 66 Bowen and Rohatgi Discuss. Faraduy Soc. 1953 14 146. 67 Waters Discuss. Farnda-y SOC. 1953 14 228. 1 64 QUARTERLY REVIEWS anthracene which may dissociate or form the stable peroxide other intermediates have indeed been postulated such as a metastable form of oxygen6* or a labile anthracene dirne~-,~~ but the former is physically im- probable and the latter may be excluded on the grounds of the small dependence of the triplet-state decay time on anthracene concentration.6o The second unexplained factor involves the large dependence of the oxidation rate on the nature of the In fact the mechanism of the photo-oxidation appears to be different in different solvents. It is likely that these differences are due to three factors (i) the viscosity which will determine the rate of bimolecular encounter-controlled reactions as well as influence the rate of internal conversion (ii) the specific quenching ability of the solvent which depends on the presence of heavy atoms in the solvent molecules and (iii) specific solvent-solute interactions which determine the rate of the radiationless internal conversion into the ground state.The photo-oxidation of aromatic hydrocarbons is faster in carbon disulphide than in any other solvent. Carbon disulphide appears to have a profound effect on the mechanism of photo-oxidation and for many years preparative chemists have used it as a solvent to obtain high and rapid yields of photo-oxide from literally hundreds of hydrocarbons. 62 One explanation of this phenomenon is concerned with the relative energy spacings of the excited states of the solvent and solute. The mercury 3650 A lines have been used in nearly all the photochemical studies that have been described; and carbon disulphide has a broad continuum which has an appreciable extinction coefficient at this wavelength.Thus the excited- singlet solute molecule may interact strongly with a carbon disulphide molecule through mixing of the states and the energy transfer which occurs may accelerate the production of anthracene in its triplet state. It is difficult to decide the mechanism of this energy transfer. First we require to know the nature of the interacting states. It appears reasonably certain that the primary process does not involve the electronically and the vibrationally excited molecule. This has been demonstrated by Hochstras~er,~~ who measured the fluorescence yield as a function of the wavelength of the exciting light for anthracene dissolved in solvents such as ethyl iodide and carbon disulphide both of which exhibit a continuous absorption throughout the region examined. It was found that after cor- rection for solvent absorption the fluorescence yield was constant through- out the range 3000-3800A and this implies that the initial vibrational- energy content of the anthracene molecule is not a governing factor and also that the interactions which occur are not dependent on the amount of incident light which is absorbed by the solvent.With regard to the effect of temperature on the fluorescence yields of 68 Kautsky Biochem. Z. 1937,291,271. Dufraisse Bull. SOC. chim. France 1939,6,422. 6o Linschitz and Sarkanen J. Amer. Chem. SOC. 1958 80,4826. Bowen and Tanner Trans. Faraduy SOC. 1955,51,475. 62 Dufraisse Bull. SOC. chim. France 1939,6,422. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 165 anthracene derivatives two distinct types of quenching may occur.In most solvents 9- and 9,lO-substituted anthracenes show high fluorescence yields which are strongly temperature- and viscosity-dependent (E w 4 kcal./mole). Anthracene derivatives having substituents only in the terminal rings show low quantum yields of fluorescence and virtual temperature-independence (E x 0.5 kcal./mole). 63 These results illustrate that there are two types of quenching process. The first involves internal conversion into the ground state and the second is associated with the enhancement of intersystem crossing. This interpretation is supported by the .rr-electron nodal distributions in singlet excited anthracene molecules. Substituents in a terminal ring would be expected to affect the inherent intersystem-crossing rate to a greater extent than peri-substituents do.The parallelism between the amount of quenching by oxygen and the amount of photo-oxide formed which is observed for benzene and some of its methyl derivative^,^^ is not present for the anthracene systems. The sum of the quantum yields of fluorescence and photo-oxidation is not equal to the fluorescence quantum yield in oxygen-free solutions at the same concentrations. For example for 2 x 10-2M-solutions of 9-methyl- anthracene the sum is 0.357 whereas the fluorescence yield in oxygen-free solutions is 0.156. At very low concentrations 10-5~ the fluorescence yield of an oxygen-free solution is 2.5 times greater than for this solution saturated with oxygen.52 This can only inply that oxygen quenching is considerable although at this concentration the formation of photo-oxide is almost absent.Studies of the kinetics of dimerisation of anthracene5’~ 65r 66 have shown that dimers are formed by reaction of singlet excited molecules with molecules in the ground state although collisions between these species need not necessarily result in dimer formation. For anthracene it has been shown that decrease of the fluorescence yield on increase of the anthracene concentration is almost equal to the increased yield of photodimerisation. In this respect an interesting problem is posed by the recent results of Ba~kstrom,~’ which involve the continued irradia- tion of solutions of anthracene and biacetyl with light which is of long enough wavelength (4358 A) to be absorbed only by the biacetyl. After some time a precipitate of dianthracene appears.The assumption made in this case is that energy is transferred non-radiatively from triplet biacetyl to populate the triplet level of anthracene and that collision of triplet anthracene with ground-state anthracene molecules then results in the formation of the photo-dimer. However all the other studies of photo- dimerisation have indicated that only singlet excited anthracene molecules give rise to dimers and that dimerisation is not in competition with photo- tls Bowen and Sahu J. Phys. Chem. 1959,63,4. 64 Bowen and Williams Trans. Furuday SOC. 1939 35 765. 65 Weigert Naturwiss. 1927 15 124. 66 Cherkassov and Vember Opt. i Spekt. 1958 4,203. 67 Backstrom Actu Chem. Scund. 1958 12,823. 166 QUARTERLY REVIEWS oxidation. Since the extinction coefficient of anthracene at 4358 A is less than the maximum amount of dimer formed from anthracene which has absorbed radiation directly is 0.001% while the observed yield of dimer is approximately 1 %.Anthracene has a correspondingly large quenching action on the phosphorescence of biacetyl in solution. The large volume of careful work on hydrocarbons of the anthracene series has enabled many of the primary processes in Fig. 4 to be identified and their probabilities unambiguously computed. However even in these apparently straightforward cases the natures of the solvent and of the substituent in the anthracene nucleus as well as the position of the substituent may drastically alter the importance of any one process. For example 9,lO-diphenylanthracene has nearly a unit quantum yield of fluorescence between temperatures of -70” and +20” and in this case the quantum yield of photo-oxidation depends on the oxygen concentration and on the hydrocarbon concentration.68 This suggests that the oxidation occurs directly by union of oxygen with the singlet state or that oxygen- catalysed singlet-triplet conversion always results in photo-oxidation (cf. step 7 of Fig. 4). The high quantum yield of fluorescence in oxygen-free solutions and the absence of dimerisation suggest that the peripheral phenyl groups which are constrained to be out of the plane of the anthra- cene nucleus sterically prevent solvent interactions that would otherwise result in the internal conversion of the excitation energy into the ground state. This steric explanation is probably the more correct as the rate of the photo-oxidation is considreably enhanced in a heavy-atom solvent.Chlorophyll.-Most of the concepts discussed above are applicable to the primary processes in the photochemistry of photosynthetic molecules. The absorption and emission spectra and the photochemistry of com- pounds such as chlorophyll have been studied in vitro (purified samples in solution) as well as in vivo. In the latter case interpretation is complicated by the uncertain state of aggregation of chlorophyll and by the inter- molecular energy-transfer processes which are part of photosynthesis. Chlorophyll a is discussed in detail here as a representative example of a molecule which is involved in the primary process of photosynthesis. The absorption spectrum shown in Fig. 7 shows two main bands one centred at 6700 A the “red” bands (lQ) and the other at 4200 A the Soret band (1B).6s Although measurements in vivo are somewhat uncertain because of the presence of other pigments the chlorophyll bands are not appreciably shifted.Fluorescence is observed in the red region of the spectrum only as shown in Fig. 7. The fluorescence spectrum and quantum yield are inde- pendent of the exciting wavelength from 4360 to 6980 The absolute 68 Livingston J. Chim. plrys. 1958 55 887. 89 Platt “Electronic Structure and Excitation of Polyenes and Porphyrins,” in “Radia- tion Biology” ed. by A. Hollaender McGraw-Hill Book Co. Inc. New York 1956 ‘O Forster and Livingstan J . Chem. Phys. 1952 20 1315. V O ~ . 111 pp. 71-123. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 167 quantum yield of fluorescence in vitro is 0.24 independent of solvent used e.g.methanol ethyl ether acetone benzene and cyclohexanol except that for dry chlorophyll in a dry hydrocarbon solvent the yield is quite I I 1 I 4000 5000 60000 7000 8000 The absorption and emission spectra of chlorophyll a.73v77 - Absorption spectrum in ether at room temperature - - - - - - - fluorescence spectrum in ether at room temperature ordinate in arbitrary units . . . . . . . . phosphorescence spectrum in a rigid glass (dry hydrocarbon) at - 183 ' ordinate in arbitrary units. low. In vivo the fluorescence quantum yield is very low 0.02. This point is discussed more fully in the following section on the photo-oxidation of solids. As with polyacene solutions internal conversion of excited states above the lowest singlet state occurs very rapidly in chlorophyll solutions (and probably in vivo also) and hence it is unlikely that the higher excited singlet states can play any important role either in the photochemistry of these molecules or in photosynthesis itself.The fluorescence of chlorophyll is quenched in the presence of dissolved oxygen but less than that of other porphyrins or polynuclear aromatic hydrocarbon^.^^ The quenching by oxygen is apparently not related to photo-oxidation since the yield of the latter process is very (ca. sec.) but only in solution in a dm hydrocarbon at low temperature (rigid glass). 73 Its spectrum is shown in Fig. 7. This emission is assigned to the transition from a 3 ( n ~ * ) state to the ground state. Phosphorescence had not pre- viously been observed for chlorophyll a.'* Fernandez and B e ~ k e r ~ ~ attribute the phosphorescence in dry media to a change in the nature of the lowest singlet state from l(7~,7~*) in hydroxylic solvents to l(n,n*) in dry solvents. In the former case the radiative lifetime of the singlet state is short and fluorescence is favoured; in the latter case intersystem crossing to the triplet state is more important and hence phosphorescence is observed. The presence of the triplet state has been demonstrated by flash-photo- lysis. 75 Solutions of chlorophyll show new absorption bands immediately 71 Livingston and Ke J. Amer. Chem. SOC. 1950,72 909. 72 Aronoff and Mackinney J. Amer. Chem. SOC. 1943 65 957. 73 Fernandez and Becker J. Chem. Phys. 1959,31,467. 74 Becker and Kasha J. Amer. Chem. SOC. 1955,77 3669. 76 Livingston and Ryan J.Amer. Chem. Soc. 1953 75 2176. Wuvr/rngth (A) FIG. 7. 5 x 10-4). Chlorophyll a also exhibits strong phosphorescence ( T ~ M 168 QUARTERLY REVIEWS after flash-excitation. Since these bands disappear rapidly they are con- sidered to represent absorption by molecules of chlorophyll in the lowest triplet state i.e. triplet-triplet absorption. The decay characteristics of the triplet state show not only first-order quenching (internal conversion) but also efficient quenching by unexcited chlorophyll molecules and by triplet molecules. Oxygen causes reversible and irreversible photobleaching of chlorophyll and a simultaneous disappearance of triplet molecules. An intermediate “moloxide” has bcen postulated as being formed from triplet molecules and molecular oxygen and able to dissociate (reversible bleach- ing or quenching) or be converted into a photo-oxide (irreversible bleach- ing).In any case the quantum yield of the latter process is small. In the process of photosynthesis (which is yet to be duplicated in vitro) the situationis morenearlyanalogous to that in the solid state than in solu- tions. The chloroplast consists of grana in which the concentration of chlorophyll is high (of the order of 0.3~). 76 Hence the distance between the chlorophyll molecular centres in the grana is about 20 A for completely random orientation or considerably less if the chlorophyll molecules are collected into aggregates or into monolayers. The small red shift in the absorption spectrum of chlorophyll in live cells is evidence against the existence of “crystalline” chlorophyll in the grana but tends to support the concept of unimolecular layers.Intermolecular energy-transfer must be a fundamental part of photo- synthesis. Fluorescence of chlorophyll is excited in vivo by other pigments e.g. phycoerythrin and phycocyanin. Exactly how the excitation energy of the chlorophyll molecules initiates reduction in photosynthesis is un- known. Rabinowitch7* cites three possibilities (i) migration of excitation energy from chlorophyll to the enzyme or photosynthetic unit in the grana (ii) migration of photoproducts formed in the neighbourhood of excited chlorophyll molecules to the enzyme and (iii) migration of the enzyme itself. The known energy transfer from accessory pigments such as caro- tenoids phyobilins and chlorophyll b to chlorophyll a supports mechan- ism (i).Even if further evidence for this is obtained it will prove difficult to determine the extent of the other possible modes of transfer. The problem of energy transfer in vivo is treated further in the following section. 6. Photo-oxidation of solids A considerable amount of work on this topic has been reported for quite complex systems. Most of it has an industrial bias in as much as the workers have been mainly interested in the effect of oxygen on commercial substrates (pigments mordants fibres etc.) under the influence of light. In order to interpret these results in terms of primary process and mechan- ism it is first necessary to study comparatively simple systems under very 76 Rabinowitch “Photosynthesis,” Interscience Publishers London 1956 Vol.11 Part 2 p. 1735. 77 French and Young J. Gen. Physiol. 1952 35 873. 78 See ref. 76 p. 1298. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 169 strict experimental conditions of purity and incident light. This discussion will be confined to molecular crystals. The generality of photo-oxidation of crystalline materials has only recently become apparent through detailed studies of the photoconductivity and emission spectroscopy of molecular crystals. For example an examina- tion of the surface photoconductivity of anthracene single crystals has clearly indicated the formation of chemical adducts of surface molecules of anthracene and molecular oxygen. Oxygen is unable to penetrate deeply into the crystal as bulk photoconductivity is not affected by the oxygen atmosphere of the crystal.It is not known whether the surface “oxide” is formed through the absorption of light initially by the crystal as a whole or by imperfect pseudocrystalline surface layers so it is not possible to say much regarding the primary processes involved in this case. Other examples often arise in crystal spectroscopy. A recent study of the lumin- escence of toluene single crystals has indicated that previous assignments have probably been based on emission by oxidative impurities:80 in this case the impurity was benzaldehyde. The packing in benzene naphthalene? anthracene and tetracene crystals is very close and it is unlikely that oxygen molecules would be able to penetrate the surface layers of the crystals. However when substituted tetracene molecules are examined in the crystalline state one can observe a bulk oxidation effect.s1 The product is the transannular peroxide produced also by photo-oxidation in solution.It has been demonstrated that the rate of production of photo-oxide which is identical with the rate of uptake of oxygen is controlled by the rate of diffusion of oxygen through the surface layers of photo-oxide and not by the intensity or wavelength of the exciting light.82 If this is a true surface effect as is presumed to be the case the primary process involves the transport of excitation energy to the surface layers of the crystal. A consideration of the nature of the energy levels in crystals is necessary at this stage. According to the theory of F~enkel*~ and D a ~ y d o v ~ ~ the energy levels in a molecular crystal bear a definite relation to these levels in the isolated molecule.During light absorption there appear in the crystal waves of excitation (a transfer of excitation from one molecule to another). The solution of the equation that defines the movement of the molecules in the excited states can be obtained for two limiting cases corresponding to the appearance in the crystal of two types of excitation waves (i) the excitation passes from one molecule to another so rapidly that the mole- cules cannot be displaced to new equilibrium positions; and (ii) the excitation passes from one molecule to another so slowly that the molecules 7 9 Chynoweth J. Chem. Phys. 1954,22 1029. 8 0 Kanda and Sponer J. Chem. Phys. 1958,28 798. 81 Hochstrasser and Richie Trans. Furaday Soc. 1956 52 1363. 82 Hochstrasser Canad.J. Chem. 1959 37 1123. 83 Frenkel Phys. 2. Sowjet. 1936 9 158. 84 Davydov J. Exp. Theoret. Phys. (U.S.S.R.) 1948 18 210. 3 1 70 QUARTERLY REVIEWS have time to take up new equilibrium positions. In the latter case there arises a local deformation which travels slowly throughout the crystal. Internal conversion of excitation energy that is the conversion of optical excitation energy into heat manifests itself through case (ii). Transfer of excitation energy between the two types mentioned above is possible and the direction of transfer will depend on the relative frequencies of absorp- tion leading to these modes of excitation. Possibly in the crystal inter- system crossing will manifest itself as localisation of excitation energy. The passage of triplet excitation throughout the crystal will be very slow and depend on the movement of crystal imperfections as in case (ii).The Frenkel-Davydov theory predicts that in the crystal the states of the isolated molecule will be split into a broad band (the exciton band). The width of this band depends on the magnitude of the transition probability between the excited state and the ground state in the isolated molecule. Consequently the triplet levels will not be split into a band as the extinction coefficient for T t S absorption is virtually zero in the molecule. Thus intersystem crossing involves energy transfer from a non-localised state to a localised state (molecular level) which does not move with the exciton wave. Such considerations probably apply not only to molecular crystals but also to dimers and molecular aggregates.McRae and Kashas5 have discussed these cases and have shown the manner in which one would expect the fluorescence and phosphorescence yields to behave in such systems. Although the general features of the Frenkel-Davydov theory have been confirmed by absorption and emission spectroscopy of molecular crystals there has been very little work done on the photochemistry of such systems which might aid in the elucidation of the nature of the primary processes. Measurements of the temperature-dependence of fluorescence in con- junction with quantum yields of photo-oxidation should be made to decide whether or not photo-oxidation and fluorescence are competitive processes. There are three distinct types of luminescence observed from pure crystals. When reference is made to crystal luminescence it is necessary to distinguish between these.The first is the short-lived fluorescence which has a spectral location and a lifetime similar to those of the isolated molecule i.e. singlet-singlet emission. The second is the long-lived phosphorescence which resembles the normal phosphorescence in its approximate spectral location. The third is a luminescence of longer lifetime than the fluoresc- ence the two appearing in the same spectral region. The nature and origin of the third type is not certain although it appears to be quite a general phenomenon.86 Each of these emissions arises initially through the absorp- tion of light into the singlet state of the crystal and no significant direct excitation of a metastable state is inv01ved.~' 85 McRae and Kasha J.Chem. Phys. 1958 28 721. 86 Blake and McClure J. Chem. Phys. 1958 29 722. 87 Belikova Opt. i Spekt. 1959 6 117. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 17 1 During the formation of a transannular peroxide a drastic structural change must occur as is illustrated in Fig. 8 for tetramethylrubrene. A large CH CH CH CH RG. 8. The photo-oxidation of tetramethyZrubrene.82 FIG. 9. A model to represent the photo-oxidation which takes place on the surface of a molecular crystal. amount of lattice distortion is necessary to allow the hydrocarbon to twist about the 5,lZpositions and this can be explained in terms of the scheme shown in Fig. 9. The lines represent a linear aggregate of molecules as for example along a particular crystal dimension.Step (a) is the initial light absorption which gives rise to the excitation of a non-localised crystal excited state. The arrowed arcs represent wave-fronts moving very rapidly through the crystal or along the aggregate when no resultant con- figurational change occurs. Step (b) involves the trapping of a quantum of excitation energy which then moves along the aggregate very slowly and hence separates from the exciton wave. Steps (c) and (d) show the movement of the imperfection until the surface is finally reached; at (e) the reaction with oxygen physically adsorbed on the surface takes place. The variation of the slope of the dislocated molecule in this two-dimensional representa- tion is intended to represent a damping of the localised excitation energy as the excess of energy is taken up in the lattice vibrations.There appear to be three factors which determine the extent of the oxidation if any of a molecular crystal (i) the crystal must not have unit quantum yield of 3* 172 QUARTERLY REVIEWS fluorescence otherwise none of the electronic energy would become local- ised in this sense; (ii) the imperfections can move rapidly enough for the surface of the crystal to be reached before the electronic energy is trans- ferred into the vibrations of the lattice and within the radiative lifetime of the triplet-singlet emission; and (iii) should conditions (i) and (ii) be favourable the oxygen molecules can reach the hydrocarbon crystal surface i.e. permeate the oxide layer during the period that the surface molecules retain some electronic excitation energy.In the photo-oxidation of pure crystals of tetramethylrubrene it was apparent that (iii) was the limiting factor. The measured quantum yield of oxidation in one system was 0.001 but this could be momentarily increased by at least one order of magnitude if the crystals after preliminary oxidation were allowed to stand in oxygen for a few hours and were then re-irradiated.82 Whether or not the model presented here approaches reality still remains to be seen. However when dealing with molecular crystals it is important to realise the scope of the Frenkel-Davydov theory. So far we have used the term exciton only in the sense described by the Frenkel-Davydov theory. Many workers use this term to describe a different type of excitation and although the latter usage is normally confined to ionic crystals the extension to molecular crystals is often ob- vious.The primary act after light absorption by a crystal is the production of mobile excitons which as before are waves of excitation energy. These lose thermal energy to the lattice until they come to rest and the excited electrons revert to the ground state with accompanying phonon emission.* As an alternative the mobile excitons may during their lifetime interact with crystal imperfections where they are ultimately trapped. The resultant species which is a trapped exciton consists of an electron held by its mutual attraction for the trap and the hole. This trapped exciton will have a characteristic lifetime before it finally reverts to a localised exciton. The trapped excitons may also dissociate into an electron and a mobile hole but to effect this dissociation some extra energy is required.The ionisation potential of an aromatic molecule is normally about 7.5 ev. The energy of the exciton cannot be greater than ca. 3 ev if near-ultraviolet light is employed. However such extra energy could be obtained from high-energy trapping centres in the crystal. Another possibility arises and this involves the “bimolecular” reaction between the trapped exciton and a mobile exciton. This interaction may result in chemical reaction as for example in the photolysis of azides.88 In order that a reaction can occur at a surface according to this description traps or imperfections must be present at the surface. The excitation energy *A phonon is a quantum of linear motion of crystal-lattice vibrational energy.In gases and liquids an excess of molecular vibrational-energy is rapidly removed by collisions which lead to the transfer of this energy. In crystals the excited molecular vibrations are only acceptable to the crystal lattice as the much smaller quanta phonons. Thus internal conversions between excited states in crystals of the molecular type occur through phonon-exciton interactions in the lattice. Thomas and Tompkins Proc. Roy. SOC. 1951 A 209 550. HOCHSTRASSER AND PORTER PRIMARY PROCESSES IN PHOTO-OXIDATION 173 is transported to these traps rapidly via mobile excitons and the trapped exciton which results can react with adsorbed gas if this is available. The photochemical yield is then determined by the rate of union of the trapped exciton with oxygen in relation to the rates of the various processes which result in deactivation of the trapped exciton.The extension of such ideas to explain transfer of excitation energy and electronic conduction in aggregates of more complex molecules such as chlorophyll in lamellae has been discussed by Ka~ha.*~ This author presents a simplified though accurate account of the exciton theory with particular reference to the photosynthetic system. In an aggregate such as that of chlorophyll in the chloroplast excitations to the exciton band will have an “allowedness” which depends on the relative orientation of the molecules in the lamellae. The molecular triplet level lies below the exciton band and the probability of populating this level depends on the relative rates of fluorescence and internal conversion compared with intersystem crossing from the thermally equilibrated crystal band to the triplet level.In some aggregates the lowest component of the exciton band combines with the ground state with very low probability. Thus the band remains popu- lated for a long time compared with the intersystem-crossing time. It is to be expected on this basis that the phosphorescence of the aggregate will be enhanced over that of the monomer and in turn the fluorescence is reduced. These considerations could account for the very low quantum yield of fluorescence for chlorophyll in vivo. The Reviewers are grateful for grants from The Petroleum Research Fund administered by the American Chemical Society (to G. B. P.) and from the National Research Council of Canada also to Dr. C. Reid for many helpful discussions and to Dr. L. D. Hayward for reading the manuscript. Kasha Rev. Mod. Phys. 1959 31 162.

ISSN:0009-2681    

DOI:10.1039/QR9601400146

出版商:RSC    

年代:1960

数据来源: RSC

摘要:

PHYSICOCHEMICAL ASPECTS OF SOME RECENT WORK ON PHOTOSYNTHESIS By ROBERT LIVINGSTON (DEPARTMENT OF CHEMISTRY UNIVERSITY OF MINNESOTA) MUCH of our present knowledge of the photochemistry of chlorophyll has been acquired during the last ten years. No other substance has had its photochemical properties studied so broadly and intensive1y.l In spite of the effort which has gone into these studies there are obvious gaps in our factual knowledge of its spectroscopic and photochemical properties and the interpretation of these facts is in part arbitrary and uncertain. Interpretation is rendered difficult by the complexity of the problem. Chlorophyll appears to exhibit every known type of radiative and non- radiative transition between its several electronic states and to undergo or sensitise practically all of the various types of photochemical reactions which have been observed for all other polyatomic molecules.From the viewpoint of photochemistry chlorophyll is an extraordinarily versatile compound. Its photochemistry is interesting not only intrinsically but also because of its relationship to the primary act of photosynthesis. The same amount of study scattered amongst dozens of compounds would have contributed much less to the progress of the photochemistry of polyatomic molecules than has the intensive investigation of this one substance. Changes in the environment of chlorophyll molecules profoundly affect their spectroscopic properties. Measurements have been made with chlorophyll in dilute and in concentrated solutions with microcrystals with unimolecular films and when adsorbed on solid surfaces.Even in dilute solutions the spectroscopic properties of chlorophyll are strongly influenced by the nature of the solvent and by the temperature. In more concentrated solutions new phenomena appear presumably owing to the intermolecular transfer of energy of excitation and to the reversible formation of dimers. Microcrystalline chlorophyll is non-fluorescent and exhibits photoconductivity and related effects. There is much more information available about dilute solutions of chlorophyll than about its other states of aggregation. In the plant cell the pigment is present in a highly concentrated and partially ordered condition. However it is not unreasonable to treat chlorophyll present under these conditions as having properties which are modifications of those observed in dilute solution.For these several reasons emphasis in this discussion will be placed upon the studies of dilute solutions. Material which was published before 1956 has been exhaustively reviewed by E. Rabinowitchl in his three-volume monograph on photosynthesis. Rabinowitch “Photosynthesis,” Vol. I (1945) Vol. 11 Part I (1951) Vol. 11 Part I1 (1956) Interscience Publ. New York. 174 LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 175 In the present article we shall attempt to discuss in detail the results of studies which have appeared in the last five years and to outline those aspects of older studies which are essential to an understanding of the general problem. Molecular Structure Chlorophyll is a substituted magnesium-complexed dihydroporphin.This pigment occurs in Nature either as chlorophyll a whose structural formula2 is given in Fig. 1 or as chlorophyll b in which a carbonyl group is FIG. 1. The structural formula of chlorophylla. substituted for the methyl group marked with an asterisk. In acid solutions chlorophyll is readily hydrolysed ; the magnesium being replaced by two hydrogen atoms. This compound is called phaophytin. The spectroscopic and photochemical properties of chlorophyll are markedly influenced by the presence of the magnesium atom and of the cyclopentenone ring (v). In homogeneous solutions the phytol group CZ0Hs9 has little or no effect upon the properties of the molecule although it undoubtedly plays an important role in determining the spatial arrangement of the pigment molecules in the chloroplast.There are two other naturally occurring chlorophyll-like pigments whose photochemical properties have been extensively studied. One of these protochlorophyll is the precursor of chlorophyll in green plants. Chemically it is similar to chlorophyll a except that it contains two less hydrogen atoms none of the pyrrole rings being reduced. The other bacteriochlorophyll occurs in certain photosynthetic bacteria. It is a tetrahydroporphin ring 11 as well as ring IV being reduced. In place of the red (660 mp) absorption of chlorophyll a its ethereal solutions absorb strongly in the near infrared (770 mp). In addition to these four magnesium-containing compounds and their phzeophytins photochemical studies have been made of some porphyrins of tetraphenylporphin and of phthalocyanine.Both the metal-free com- pounds and their complexes with various metals have been studied. While there are important individual differences the photochemical properties of these several compounds are in general similar. Ref. 1 pp. 438-448. 1’76 QUARTERLY RFWUEWS Absorption Spectra Detailed knowledge of the absorption spectra of chlorophyll has been of value in the determination of the nature of its several electronic states. Spectrophotometric measurements are used in a routine way to measure the concentration of solutions and to estimate the purity of samples of chlorophyll and its derivatives. The first definitive spectra of dilute solu- tions of chlorophylls a and b in the visible and near ultraviolet were published by Zscheile and Comar in 1941 .3 Subsequent measurements,4 using highly purified samples have necessitated only minor modifications of these spectra.Attempts have been made to calculate the energy levels of porphin and dihydro- and tetrahydro-porphins by means of the free-electron model5 and the LCAO-MO approximation.6 Although the agreement between observed and calculated energy levels is moderately good the calculations scarcely permit an unambiguous interpretation of the nature of the several excited states. The infrared spectra of the chlorophylls and of their derivatives and analogues have been H ~ l t ’ ~ has demonstrated that these spectra are particularly useful in identifying the changes in the substituent groups which occur when chlorophyll is allomerised etc. The correspond- ing changes in the electronic spectra are more difficult to interpret and in some cases are scarcely detectable.For example the visible and ultra- violet absorption spectra of chlorophyll a and of the corresponding chlorophyllide (in which the phytol “tail” C20H39 is replaced by an ethyl radical) are practically identical.* The metal-complexed porphyrins and chlorins (including chlorophylls a and b) form stable monosolvates with nucleophilic reagents (i.e. Bronsted bases) such as water alcohols amines ketones and ether^.^ The ordinary absorption spectra of the chlorophylls are those of the solvates. The principal absorption bands of the unsolvated compounds appear broader,1° owing to the superposition of new bands shifted toward the red.ll For example in dry benzene the normal maximum of chlorophyll b at 6450 is accompanied by a second peak at 6650 A.Freed and Sancier have proposed1lU an interesting interpretation of their results ; however Zscheile and Comar Bot. Gaz. 1941 102 463. (a) Smith and Benitez Carnegie Inst. Year Book 1954 53 168; (b) Stoll and Wiedemann Helv. Chim. Acta 1959 42 679. Ref. 1 pp. 1793-1798. * (a) Platt in “Radiation Biology” Vol. 111 McGraw-Hill New York 1956 pp. 94-114; (b) Matlow J. Chem. Phys. 1955,23,673. (a) Ref. 1 pp. 1811-1815; (b) Holt and Jacobs Plant Physiol. 1953 30 553; (c) Holt Canad. J. Biochem. Physiol. 1958,36,439. * Holt and Jacobs Amer. J. Bot. 1954,41 710. Livingston and Weil Nature 1952 170,750. lo Livingston Watson and McArdle J. Amer. Chern. S c . 1949,71 1542. l1 (a) Freed and Sancier J. Amer. Chem. SOC. 1954,76,198; (b) Freed Science 1957 125 1248. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 177 no reasonable explanation has been suggested which is compatible with all of the reported observations.In 1952 Jacobs and Holt12 published an outline of a procedure for the preparation of crystals of chlorophylls a and b. Subsequent experiments4b* l3 have confirmed the validity of the first report. Relative to the solution spectra the principal absorption maxima of suspensions of microcrystals are shifted strongly to the red the shift being about 80 mp for the red band of chlorophyll a. This is much greater than any known solvent effect. The absorption spectra of colloidal chlorophyll and of chlorophyll in vivo resemble more nearly the spectra of solutions than those of the crystalline chlorophyll. Jacobs has attempted to explain the marked effect of crystal- lisation upon the wavelengths of the maxima in terms of Heller and Marcus’s theory14 of the excitation of virtual dipoles in an infinite isotropic lattice.Unimolecular films of chlorophyll have been prepared13b which exhibit the absorption spectra of either amorphous or crystalline chloro- phyll depending upon their surface concentration. The Electronically Excited States of the Monosolvates The photochemical reactions of polyatomic molecules are determined largely by the properties of their electronically excited states. The more important of these properties are the available energy the intrinsic mean life the maximum yield of fluorescence (or phosphorescence) and the probabilities of chemical reactions of the molecule in each of its excited states.These characteristic properties are influenced by the temperature and by the nature of the solvent. The absorption of visible light by chlorophyll corresponds to a transi- tion from the ground to the first (or second) excited singlet (Q) state. Molecules in their first excited (i.e. fluorescent) states have a relatively large probability of undergoing radiationless transitions to their ground triplet state which is known to play an important role in photochemistry. In addition to these n-n (or Q) states n l r (or U or W ) states6= are probably important in determining the spectroscopic and photochemical properties of chlorophyll and related compounds. The formation of a monosolvatelO has a marked effect upon the spectro- scopic properties of chlorophyll probably by changing the energy of its n-7T states relative to the corresponding n+r statesGa Chlorophyll dis- solved in ordinarily pure but not-especially dried solvents exists principally as a monosolvate.Since the great majority of measurements have been made with such “wet” solutions it will be convenient to discuss them separately. The differences between the properties of the solvated and non- solvated molecules are described and interpreted in a later section of this Review. * l2 Jacobs and Holt J. Chem. Phys. 1952,20,1326. l3 (a) Jacobs Vatter and Holt Arch. Biochem. Biophys. 1954 53 228; (b) Zill l4 Heller and Marcus Phys. Rev. 1951,84,809. Colmano and Trurnit Science 1958,128,478. 178 QUARTERLY REVIEWS The Fluorescent State.-For chlorophyll as for practically all poly- atomic molecules the maximum intensity of fluorescence is situated at the long-wavelength side of the first absorption peak.The overlap between the absorption and emission bands is greater for chlorophyll than it is for most dyes and pigments. The red absorption of chlorophyll a has a maximum at 6600 A and extends to 7000 A. The corresponding fluorescence peak is at 6645 A and fluorescence emission is detectable at wave lengths as short as 6300 A. Except for thin layers of very dilute solutions re- absorption greatly distorts the appearance of the fluorescence spectrum.15 The energy corresponding to the zero-zero transition from the ground to the fluorescent state can be estimated by comparing the absorption and emission spectra. It is approximately 43 kcal./mole. The quantum yield of fluorescence q5f1 of a solution is the ratio of the number of quanta emitted to the number absorbed.Except for very dilute solutions the quantum yield is reduced by reabsorption and self-quenching. The most recent and probably the most reliable measurements of the maximum values for the quantum yields of fluorescence were published by Weber and Teale16 in 1957. Some of their results are listed in Table 1. TABLE 1. Maximum quantum yields of fluorescence Solvent Benzene* Benzene Ethyl ether Acetone Ethanol Methanol Cyclohexanol X(ml-4 of exciting light Hg 366 Cd 644 Cd 644 Cd 644 Cd 644 Cd 644 Cd 644 Fluorescent yield +f1 chlorophyll a chlorophyll 6 0-33 0.11 0-32 0.12 0.32 0.12 0.30 0.09 0.23 0.10 0.23 0.10 0.30 * Presumably “wet” benzene was used since chlorophyll is not detectably fluorescent in carefully dried hydrocarbons.1° With the exception of the values for methanolic (and probably ethanolic) solutions these data are consistent with the earlier corrected values of Forster and Livingstonl’ and the uncorrected values of Latimer Bannister and Rabinowitch.l* The correction referred to above was proposed by Shepplg to take care of the reflection of the fluorescent light at the bound- ary of the samples.Although some such correction should certainly be applied to Forster and Livingston’s data and presumably to those of Latimer et al. the 1-20 factor proposed by Shepp may be too large. In l5 Ref. I pp. 603-609 740-747. l8 Weber and Teale Trans. Faraday Soc. 1957 53 646. l7 Forster and Livingston J. Chem. Phys. 1952 20 1315. l9 Shepp J. Chem. Phys. 1956,25 579. Latimer Bannister and Rabinowitch Science 1956 124 585.LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 1 79 contrast to the results of Weber and Teale both of the earlier sets of measurements indicate that the fluorescence yield of chlorophyll b is about two-fold less in methanol than in ether but the yields for chlorophyll a are identical in these two solvents. Tumerman,20 who used a mixture of chlorophylls a and b (probably chiefly component a) reports quantum yields 5 % greater for ethanol than for either acetone or ether. The cause of this disagreement is not known. Apart from this discrepancy the authors’ estimate16 of an uncertainty of &lo% for the quantum yields appears reasonable. For phzophytin a in benzene the quantum yield is 0.18 5 0.02. The intrinsic mean life time TO of an excited state is the value which the mean life would have if the excited molecule could lose its energy only by emitting a photon of fluorescent light.The actual mean life 7 of the state is equal to the product of the intrinsic life and the quantum yield T = $ 1 1 ~ ~ . T and c j f ~ vary with the temperature and the nature of the solvent but their ratio TO is little influenced by these factors. TO can be calculated from the integrated extinction coefficients corresponding to the transition from the ground to the fluorescent state by means of the modified Ladenburg equation. 21 The actual mean life can be determined directly by measuring22 the decreasing intensity of fluorescence as a function of time following excitation by a very brief pulse of light. When the decay of the excited state follows a first-order law it is more convenient to measure T by the phase-shift method.23 It can also be evaluated indirectly from measurements of the degree of polarisation of fluorescence as a function of the viscosity of a series of solvents.24 The mean life of the fluorescent state of chlorophyll has been evaluated by all of these methods and resub representative of the several types of measurements are presented in Table 2.TABLE 2. Actual mean life time of thejluorescent state Pigment Solvent r x lo9 (sec.) Method Ref. Chlorophyll a Ethyl ether 5.1 Direct 22 Chlorophyll a Ethyl ether 4.5-6.3* Landenburg 22 Chlorophyll a Ethyl ether 5.1 Phase shift 25 Chlorophyll a Toluene (wet ?) 4.4 Phase shift 25 * These values were obtained from the calculated values of TO by using & = 0-30 for chlorophyll a and = 0.13 for chlorophyll b.The indicated spread of values is the result of the uncertainty as to what range of frequencies corresponds to this electronic transition. 2o Tumerman Doklady Akad. Nauk S.S.S.R. 1957 117 605; Soviet Physics 1957 2 525. 21 Forster “Fluoreszenz organischer Verbindungen,” Vandenhoeck u. Ruprecht Gottingen 1951 pp. 156-159. ee Brody and Rabinowitch Science 1957 125 555. 23 Rollefson and Bailey J. Chem. Phys. 1953,21 1315. 24 Ref. 21 pp. 168-172. 25 Dmitrievsky Ermolaev and Terenin Doklady Akad. Nauk S.S.S.R. 1957 114 751. 180 QUARTERLY REVIEWS TABLE 2-continued Pigment Solvent T x lo9 (see.) Method Ref. Chlorophyll a Benzene (wet ?) Chlorophyll a Cyclohexanol Chlorophyll a Pyridine Chlorophyll a Ethanol Chlorophyll b Chlorophyll b Chlorophyll b Chlorophyll b Chlorophyll b Chlorophyll b Chlorophyll b Chlorophyll a + b Chlorophyllide a Ethyl ether Ethyl ether Ethyl ether Toluene (wet ?) Benzene (wet ?) P yridine Ethanol Ethanol Ethyl ether Phaeophytin a Ethyl ether Phaeophytin a Toluene (wet ?) 7.8' ' 6.5 5.3 5.0 3 -9 3.0-4.2* 3 *O 4.5 6.3 4.7 3.4 5.3 5.1 5.3 5-6 Direct Polarisation Phase shift Phase shift Direct Ladenburg Phase shift Phase shift Direct Phase shift Phase shift Phase shift Phase shift Phase shift Phase shift 22 26 25 25 22 22 25 25 22 25 25 20 25 25 25 The Triplet State.-The existence of a long-lived energetic form of chlorophyll has been unequivocally established from photochemical kinetic evidence and from a comparison of the quantum yields of fluores- cence with the yields of certain autoxidations photosensitised by chloro- phyll.Presumably this form of chlorophyll is the long-lived species whose properties can be studied by means of the photoflash technique and which is responsible for the weak phosphorescence exhibited by chlorophyll b in rigid solvents at low temperatures. It is commonly assumed that this species is the lowest triplet state. Although this identification has not been proved for chlorophyll in the remaining discussion we shall assume that it is correct. At low temperatures in glassy solvents most aromatic compounds are strongly phosphorescent. Chlorophyll is an exception to this rule. Earlier reports of phosphorescence of chlorophyll a were in error no phos- phorescence of dilute solutions of this compound has been detected although it has been sought for by several investigators.Chlorophyll b phosphoresces weakly2' in ether-pentane-ethanol mixtures at liquid- nitrogen temperatures. The emission has its maximum intensity at about 8650 A; its mean life is approximately 0.03 sec. and its quantum yield is less than 0.01. It is noteworthy that Cu-phzophorbide a which is non- fluorescent phosphoresces strongly in low-temperature rigid solvents. The absorption spectrum of the triplet and the kinetics of its decay can be measured (at ordinary temperatures in fluid solvents) by the use of 26 Weil Doctoral Diss. University of Minnesota 1952. 27 Becker and Kasha. J. Amer. Chem. SOC. 1955,77 3669 LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 18 1 flash-photolytic apparatus. The photochemical formation of the triplet state of chlorophyll is an efficient process.For example over 90% of the chlorophyll molecules present in 15 ml. of a 2 x 10-6M-SOhtiOn in pyridine are raised to their triplet state by a single 50-Joule flash of a few microseconds' duration. Absorption spectra of such solutions have been obtained both by the photog rap hi^^^-^^ and by the phot~electric~l modifications of the photolytic-flash technique. The triplet spectrum (Fig. 2) was obtained by Linschitz and Sa~kanen,~~ using the photoelectric 1.0 0.8 0.6 9 c C Q 0 Q 0.4 T 0.2 n I I 1 1 I I I I i I I I c I b I i 1 - I 400 500 6 0 0 700 W o v o / e n g t h ( m p ) FIG. 2. The absorption spectra of chlorophyll a in its triplet and in its ground state. (Reproduced by permission from J. Amer. Chem. SOC. 1958,80,4826.) method.The other available data,28-30 while probably less reliable are in satisfactory agreement with these curves. Unlike the corresponding spectra of the aromatic hydrocarbons the absorption spectra of the triplet states of chlorophyll and related corn pound^^^*^^ are broad and relatively 28 Livingston J. Amer. Chem. SOC. 1955 77 2179. 2p Livingston and Fujimori J. Amer. Chem. SOC. 1958 80 5610. 30 Claesson Lindquist and Halmstrom Nature 1959 183 661. 31 Linschitz and Sarkanen J. Amer. Chem. SOC. 1958,80,4826. 32 Linschitz and Pekkarinen J. Amer. Chem. SOC. in the press. 182 QUARTERLY REVIEWS structureless extending into the near infrared. Their principal maxima occur at the long-wavelength edge of the Soret bands (of the singlet state). The disappearance of the triplet state of chlorophyll a or b in dilute fluid solutions conforms to the following equation.It is not a simple first order process -d[GH’]/dt = k,[GH’] + k,[GH’]’ + k,[GH][GH’] Table 3 summarises the available data. TABLE 3. Rate constants for the decay of the triplet state of chlorophyll Pigment Chlorophyll a Chlorophyll a Chlorophyll a Chlorophyll a Chlorophyll b Chlorophyll b Chlorophyll b Solvent Pyridine Pyridine Benzene Cyclohexanol Pyridine Benzene Benzene (wet) k x 10’ (sec.-l) 7.3 6.7 4.4 7.7 3.1 3.3 4.0 kB x lo9 (1. mole-1 sec.-l) 1.3 1.5 2.1 0.02 1.6 2.2 2.4 k x lo7 Ref. (1. mole-l sec.-l) 2 33 2 31 5 31 -0.2 33 2 31 9 31 4 30 When chlorophyll in dilute solutions (less than 10-5~) is illuminated with light of ordinary intensity the mean life of the chlorophyll triplet is equal to kA-l approx.2 x sec. The self quenching process GH + GH’-+2GH becomes dominant at high concentrations (greater than TABLE 4. Summary of some properties of the excited states” State AE (kca1.l +(Inax.) T(max.1 (set.) mole) Chlorophyll a First excited singlet 43 ee 0-32 ee 5.3 x py Lowest triplet - g0.01 1.4 x 10-3 py Chlorophyll b First excited singlet 44 ee 0-12 ee 4.7 x py Lowest triplet 33 epa <0.01 epa 3.1 x py First excited singlet 42 ee 0-18 b 5.3 x ee (a + b) Phzophytin a Lowest triplet - < 0.01 - * Solvents b-Benzene; ee-ethyl ether; epa-ether-pentane-alcohol at - 1 8 0 ” ~ ; py-pyridine. (a + b) denotes that a mixture of chlorophylls a and b was used. 33 Fujimori and Livingston Nature 1957 180 1036. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 183 5 x 10-5~) where the lifetime of the triplet state is inversely proportional to the concentration of chlorophyll.The second-order process 2GH' -+ 2GH (or GH + GH') can be safely neglected for all intensities normally used in photochemical experimentation but strongly reduces the mean life observed in flash photolytic studies. Within the limits of uncertainty the value of k is the same for chloro- phylls a and b. While the data are scanty kB appears to be inversely proportional to the viscosity of the solvent and to be not much less than the value of the rate constant k, for a bimolecular reaction whose rate is determined by the number of encounters. Approximately kD = 8RT/30y where y is the viscosity coefficient in centipoises. The Quenching of the Fluorescent and Triplet States by Added Substances Bimolecular interactions of excited pigment molecules with foreign molecules (either added or present as impurities) can reduce the intensity of the fluorescence of the pigment.This process is known as the diffusional quenching of fluorescence. Foreign substances can also quench fluores- cence by forming thermally-stable non-fluorescent addition compounds with the pigment. Diffusional quenching shortens the mean life of the excited state and its efficiency is low in viscous solutions. The second or static type of quenching does not affect the mean life of the fluorescent state and its efficiency is not directly related to the viscosity of the solvent. There is some evidence indicating that fluorescence of chlorophyll is subject to both diffusional and static quenching.However diffusional quenching appears to be the dominant type for the quenchers which have been investigated and to simplify the discussion we shall assume that it is the only type operative. Properly speaking quenching refers to an effect upon the emissionof light by the solution not upon the properties of the excited state. How- ever it is convenient and common practice to say that the fluorescent state is quenched when its mean life is shortened by the addition of a foreign substance. An obvious extension of this usage is to say that the triplet state is quenched when its mean life is shortened by the presence of a foreign compound. This terminology will be employed in the present discussion. The following reaction steps are essential to an analysis of the diffusional quenching process.In some of them the solvent may play a direct role although this is not indicated as they are written. Unimportant steps have been omitted. For example both kinetic and spectroscopic evidence indicate that for chlorophyll the radiationless transition from the first excited to the ground singlet state (GH*+GH) is negligible compared to competitive processes. Since the present section is limited to a discussion of dilute solution the self-quenching of fluorescence has also been omitted. 184 QUARTERLY REVIEWS 1 hv + GH +GH* 2 GH*+GH + hvf 3 GH*-+GH’ 4 Q + GH*+GH + Q (or products) 5 GH’ +GH 6 GH + GH’-+2GH 7 8 2GH’+2GH (or GH + GH’) Q +GH +GH + Q (or products) Excitation Emission of fluorescence Formation of the triplet Diffusional quenching of fluor- Spontaneous decay of triplet Self-quenching of triplet Second-order decay of triplet Diffusional quenching of triplet escence If no irreversible photochemical reactions occur d[GH*]/dt = 0 and d[GH’]/dt = 0 under conditions of steady illumination.It follows that Irtbs = (k2 + k3 + k4 [QI) [GH*I If1 = k2 [GH*] By definition +fl = Ifl/&&bs and therefore + f l = k2/(k2 + k3 + k4 [QI) The (modified) Stern-Vollmer relation follows directly +fl,(max.)/+f1 = 1 + k4[Ql/(k2 + k3) = 1 + k47‘IQI The Stern-Vollmer quenching constant is KQ = kp’ where T’ is the actual mean life in the absence of quenchers. The bimolecular quenching constant k8 can be obtained from flash- photolytic measurements of the rates of disappearance of the triplet in the presence and absence of quenchers - d[GH’l/dt = (krj + k,[GHI + k,[GH’I + k,[QI) [GH’I If the total concentration of chlorophyll is represented by M [GH] = M - [GH’] we may write d[GH’lldt = (k + k s ~ + ks[QI) [GH‘I + (k - ktJ [GH‘I’ = k,[GH’] + k,[GH’I2 Values for k and ku may be obtained from the experimental data by either graphical or analytical calculation; ks is equal to the difference between the values of k, corresponding to the presence and absence of a quencher divided by its concentration.In 1950 Livingston and Ke34 demonstrated that the fluorescence of chlorophyll is strongly quenched by quinones aryl nitro-compounds some nitroso-compounds and azo-dyes in addition to oxygen and nitric oxide. A few amines (notably p-aminophenol phenylhydrazine and dimethyl- aniline) quench weakly. Many active reducing agents (such as aniline 34 Livingston and Key J.Arner. Chem. Soc. 1950,72,909. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 185 quinol and thiourea) are practically without effect upon the intensity of fluorescence. Qualitatively similar results were obtained by Evstigneev Gavrilova and Krasnov~kii.~~ The quenching of fluorescence by oxygen benzoquinone and rn-dinitrobenzene (and presumably other reagents) is not accompanied by an appreciable chemical change. The fluorescence of p~rphyrins~~ and of Mg-phthal~cyanine~~ is quenched by the same types of reagent. It is undeniable for the relatively large number of compounds which were investigated,34* 35 that the effective quenchers are oxidizing agents and that reducing agents are either non-quenchers or quench only weakly.It should not be concluded from these observations that quench- ing is necessarily the result of an electron-transfer process. One possible alternati~e~~ is that the quencher induces a transition of the pigment from its fluorescent to its lowest triplet state and concurrently the quencher molecule goes from its ground to its lowest triplet state. Whether this mechanism is energetically feasible in all cases studied is unknown; however it remains a possible explanation. TABLE 5. Bimolecular constants k, for the quenching of the triplet state of chlorophyll a Quencher p-Benzoquinone Oxygen m-Dinitrobenzene /%Carotene p-Carotene p-Carotene a-Car0 tene Luteol t- Re t inene Cyclo-octatetraene Solvent Benzenea Benzene Toluene Benzene Toluene Benzene Benzene Benzene Benzene Methanol k x (1.mole-lsec.-l) 2.4 & 0.5 1.1 & 0.3 1.2 & 0*4b 1.3 & 0-3 0.7 & 0.2 1.5 & 0.3c 1.6 & 0.3 0.6 & 0.3 0.3 & 0.1 0.0012 0-0002 Ref. 37 37 39 37 39 37 37 37 39 37 a Benzene used in the first series of experiments37 was (intentionally) wet but had been otherwise purified. The benzene and toluene used in the other experiment^^^ did not contain added water but had not been exhaustively dried. b This result was obtained by Ichimura and contradicts the earlier negative observa- ti0n,3~ which was of a preliminary nature and was probably in error. c Chlorophyll b was used in these experiments. The relatively scanty quantitative information regarding the quenching of the triplet state is summarised in Table 5. In addition to the compounds listed therein Fujimori3' investigated the quenching action of phenyl- hydrazine ascorbic3 acid allylthiourea and ethyl N-phenyl carbamate.None of them showed quenching action at concentrations as high as 0*010~; or 36 Evstigneev Gavrilova and Krasnovskii Doklady Akad. Nauk S.S.S. R. 1950 74 36 Livingston Ramarao and Thompson J. Amer. Chem. SOC. 1952 74 1073. 37 Fujimori and Livingston Nature 1957 180 1036. 315. 186 QUARTERLY REVIEWS for phenylhydrazine ascorbic acid and allylthiourea at concentrations of 0.10~. Ascorbic acid and allylthiourea were tested in dry pyridine; the others in methanol. Recently P ~ g h ~ ~ demonstrated that if pyridine con- taining 2% or more of water is used as the solvent ascorbic acid is a moderately effective quencher. He also demonstrated3* that unlike retinene vitamin A is not an efficient quencher.With the exception of oxygen the list of efficient quenchers for the triplet state could not have been predicted upon any simple basis. However some of the results are in good agreement with predictions based upon classical photochemical studies. For example Schenck40 has interpreted his extensive studies of pigment-sensitised photochemical reactions as indicating that the triplet states of pigments are quenched efficiently by benzoquinone and moderately by cyclo-octatetraene. These predictions were confirmed3' by direct measurement at least for the case of chlorophyll. The effects of the carotenoids and retinene were unexpected and remain unexplained. TABLE 6. Comparison of bimolecular quenching constants with kD = 8RT/307.* Quenching of the fluorescent state Quencher Solvent p-Benzoquinone Methanol p-Benzoquinone Methanol Dinitrobenzene Methanol Oxygen Ethanol Quenching of the triplet state Quencher Solvent p-Benzoquinone Benzene Dinitrobenzene Toluene Oxygen Benzene * References in parentheses.&(I. mole-l) ~(sec.) k = K,/r kdk (1. mole-l sec.-l) 104 (34) 7 x 10-9 (28) 15 x 109 1.3 44 (34) 5 x 10-9 (31) 9 x 109 1.5 143 (41) 7 x lop9 (28) 21 x lo9 1-8 86(34) 7 x 10-9(28) 12 x lo9 1.0 k (1. mole-lsec.-l) 2-4 x 109 (37) 1.2 x 109 (39) 1.1 x 109 (37) k8/kD 0.22 0.10 0.10 In terms of the diffusional mechanism the maximum efficiency of quenching is limited (approximately) by the rate of encounters between molecules of the excited pigment and the reagent. Table 6 presents a comparison between the bimolecular quenching constants and the limiting rate constant k = 8RT/30q(l.mole-lsec.-l). For fluorescence quenching the agreement between k4 and k is surprisingly close. The bimolecular 38 Pugh unpublished work at University of Minnesota 1959. 99 Livingston and Pugh Discuss. Faraday SOC. 1959 27 144. 40 Schenck and Ritter Naturwiss. 1954 41 334. 41 Goedheer Biochim. Biophys. Ada 1958 27,478. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 187 constants for the quenching of the triplet state are about ten-fold smaller. This difference in efficiency is not peculiar to the case of chlorophyll; a similar difference was observed42 for the quenching of the triplet and fluorescent states of anthracene by oxygen. Effects of Environment upon the Electronic States In this discussion we have chosen to regard as the normal state of chlorophyll the monosolvated compound in dilute solution at ordinary temperatures.Chlorophyll will be considered to be in a perturbed state when it is absorbed on solids present as crystals in concentrated solutions in solution at low temperatures or dissolved in dry hydrocarbons. Effect of Solvation.-The most striking of *the environmental effects upon chlorophyll are those which accompany exhaustive drying of non- basic solvents such as hydrocarbons halogenated hydrocarbons etc. It has been definitely establi~hed~-~~v~~ that the chlorophylls (and other metal- complexed chlorins and porpliins) form stable 1 1 addition compounds with Bronsted bases (i.e. nucleophilic reagents). These bases include ethers and ketones in addition to hydroxylic solvents and amines.Phzopliytins and metal-free porpliyrins do not form such addition compounds at ordinary temperatures. 9 9 4 3 For either oxygen bases or nitrogen bases considered separately the stabilities of the compounds are symbatic to the basicities of the addenda. However compared to amines oxygen bases form abnormally stable addition compounds ;9 e.g. the stability constant of the methanol complex is about 200-fold greater than that of the aniline complex. When the solvent is a dry hydrocarbon the absorption bands of chlorophyll appear unusually broad. This apparent broadening is due to the occurrence of shoulders on the long-wavelength side of the principal bands.10-11144 Exhaustive drying increases the relative height of the second- ary peak but in no case is the normal band eliminated.The double maxima appear to be characteristic of the unsolvated chlorophylls.ll Lowering the temperature favours the normal band,ll corresponding to standard molar enthalpies of the addition reaction in the range from - 1500 (aniline) to -6000 cal. (heptylamine). Solvation changes the absorption spectra of all of the metal-complexed chlorins and porphins which have been studied but chlorophyll is unique in being strongly fluorescent when solvated and non-fluorescent in the unsolvated state. 9 9 The equilibrium constants for compound formation can be calculated from either spectrophotometric or fluorescence-intensity measurements. In a few cases both methods were usedg and identical values of the equilibrium constants were obtained. This demonstrates that equilibrium is attained in the dark and that either the equilibrium constant 42 Livingston’and Subba Rao J.Phys. Chem. 1959 63 794. 43 Evstigneev,Gavrilova,andKrasnovskii. Doklady Akad. Nauk S.S.S. R. 1950,70.261. 44 Fernandez and Becker J. Chem. Phys. 1959 31 467. 188 QUARTERLY REVIEWS is the same for chlorophyll in its ground and fluorescent states or else the mean life of the addition compound is long compared to the mean dura- tion of fluorescence. In 1949 Livingston et aZ.l0 postulated that the addition resulted from the formation of a hydrogen bond between the labile hydrogens of the activator and the keto-oxygen of ring v of chlorophyll. This postulate was not compatible with the observed activation of fluorescence by simple ethers but the authors dismissed this discrepancy as due to the presence of hydroxylic impurities in the ethers.Very shortly afterwards Evstigneev et aZ.43 pointed out that the change in fluorescence intensity did not occur if phzeophytin was substituted for chlorophyll. They advanced the alterna- tive explanation that the activator is bonded to the central metal atom of the pigment. Although some of the evidence which they was later shown to be in e r r ~ r ~ there can be no reasonable doubt of the validity of their postulate. The addition compounds are formed by metal-complexed pigments regardless of the presence or absence of ring v but are not formed by metal-free pigments. Since chlorophyll dissolved in dry hydrocarbons is not detectably fluorescent (q$1<0.01) the mean life of the first excited singlet state must be very short certainly less than 10-lo sec.A plausible interpretation is that the non-radiative transition from the first-excited singlet to the lowest triplet state is much faster in unsolvated than in monosolvated chlorophyll. If this were true the yield per flash of the triplet should be higher in dry than in wet solvents. Preliminary experiment^^^ indicated that this is not the case. Recently Pugh3* has made a careful flash-photolytic study of dilute solutions of chlorophyll a in very dry and in water-saturated toluene. He found that the yield is four- or five-fold less in dry than in wet toluene and that the rate of decay (more exactly the values of k, kB and k,) is the same in wet and in dry solutions. The triplet-triplet absorption spectrum appears to be the same in dry as it is in wet solutions.The situation is further complicated by the observation of Fernandez and Be~ker-4~ that dry non-fluorescent solutions of chlorophyll a exhibit at liquid-nitrogen temperatures a moderately long-lived luminescence with its intensity maximum at A7550 A. The authors report that the in- tensity of luminescence of these rigid low-temperature solutions is com- parable to that of the fluorescence of ordinary (wet) chlorophyll solutions. They also state that the mean duration of the luminescence at liquid- nitrogen temperatures is equal to or greater than 5 x sec. The same solutions at room temperature likewise luminesce in the near infrared but with an intensity about one-twentieth of the value at -180"~ Al- though in the concentrated solutions which these authors used an ap- preciable fraction of the chlorophyll must be present as dimers or higher aggregate^,^^ it is improbable that the state of aggregation was solely 45 (a) Lavorel J.Phys. Chem. 1957 61 1600; J. Chim. phys. 1958 55 905; (b) Weber and Teale Trans. Faraday Soc. 1958,54,640. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 189 responsible for all of their observations. [The strange results which they report for chlorophyll b were very probably due to the high concentration (5 x 10-5~) and the methylpentane used as a solvent. It is well known46 that chlorophyll b is only slightly soluble in aliphatic hydrocarbons of low molecular weight even at room temperature. At low temperature it is probable that the pigment existed chiefly as a colloidal suspension.] These several observations may be interpreted in terms of a modified form44 of the scheme originally proposed by K a ~ h a ~ ~ and by Platt.6u Fig.3 is an energy-level diagram corresponding to this modified scheme. Singlets Triplets -+- I I 1,2 5 Singlets Triplets 3 Dry FIG. 3. A tentative energy-level diagram for monosolvated and unsolvated chlorophyll. The left half of this diagram represents the energy levels and transitions of the monosolvate ; the right half the corresponding levels and transition of unsolvated chlorophyll a. Non-radiative transitions are indicated by zig-zag vertical lines ; radiative transitions by straight vertical lines. A broken line indicates that the transition is observable but weak. Energy levels corresponding to n-7~ states are represented by solid horizontal lines; those corresponding to y1-n states by dotted lines.For simplicity the overlapping oscillational levels which accompany electronic states have been omitted and only 0-0 transitions are indicated on the diagram. The vertical lines on the extreme left of each half of the diagram represent the absorption of a photon in the region of the Soret band. Since no corresponding fluorescence is observable the arrow is single headed. The next pair of (zig-zag) lines represents the highly probable non- radiative transitions from the second to the first excited singlet states. The 46 Willstatter and Stoll “Untersuchungen uber Chlorophyll,” Springer Berlin 1913 47 Kasha Discuss. Faraky SOC. 1950 9 14. pp. 157-166. 4 190 QUARTERLY REVIEWS normal absorption and emission of red light is indicated by the arrows labelled 1 2 (and 1’).The straight solid vertical lines on the right of each half of the diagram represent the triplet-triplet absorption which is observed in flash-photolytic experiments. The upper triplet level is indi- cated by a series of horizontal lines to suggest the unusually broad struc- tureless nature of the triplet-triplet absorption. The energy of n-n levels is shifted by ~ o l v a t i o n . ~ * ~ ~ We shall assume that for a monosolvate it lies above the FIT singlet and therefore does not interfere with the emission of fluorescence. Absorption corresponding to this n-7r level would lie within the normal rcd band and being weak would be completely masked by the stronger 71-7~ absorption. Molecules which reach the first excited singlet state must either fluoresce or pass by an act of internal conversion to the lowest triplet state; thence they go non-radiatively to the ground state.In dry hydrocarbons we assume that the n l z singlet levels are slightly below the corresponding 7r+- levels. The non-radiative transition 3’ is so fast that it eliminates the competing radiative transition 2 and no fluorescence is observed. The absorption process l’ is responsible for the long-wavelength shoulder of the red band which is observed in dry solutions. To explain the results of Fernandez and B e ~ k e r ~ ~ we shall postulate that a relatively fast radiationless transition 3” leads to the n-7T triplet state. From this state the molecule may undergo non-radiative transitions to either the ground state 5‘ or the n-7-r triplet state 9. Alternatively the molecule may emit 10 a photon of near infrared ( A 7550 A) light.At room temperatures the radiationless processes domin- ate but the radiative process becomes comparable at low temperatures. To fit the scheme to Pugh’s flash-photolytic data we must further assume that at ordinary temperatures process 5’ is four or five times faster than process 9 and that the mean life of the ~T-T triplet is at least 20-fold greater than that of the n-n triplet. Under these conditions the observed mean life and absorption spectrum of the triplet state would be the same in wet and dry solvents but the yield of triplet would be four- or five-fold smaller in dry solutions. Properties of Concentrated Solutions.-Intermolecular transfer of energy of excitation occurs efficiently by inductive resonance,21 in moderately concentrated solutions of chlorophyll monosolvates as is indicated by the self-q~enching~~p~~ and concentration depolari~ation~~**~ of fluorescence and by the sensitisation of the fluorescence of chlorophyll a by chlorophyll b.The self-quenching of the fluorescence of chlorophyll is appreciable in solutions whose concentration is 2 x 10-4~ or greater. It is as efficient in a viscous solvent cyclohexanol as in acetone or ether.2s*48 These data conform to the simple empirical relation +flo/+fl = 1 + 6 2 7 0 ~ ~ but are also consistent with Wavilow’s theoretical equation.49 4e. Watson and Livingston J. Chem. Phys. 1950 18 802. 40 Wavilow J. Phys. U.S.S.R. 1943 7 141. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 19 1 The concentration depolarisation of fluorescence of chlorophyll was studied by WeilZ6 in cyclohexanol and by GoedheeflO in ricinus oil.Weil’s data are represented by the empirical equation Po/P = 1 + 1 7 7 ~ where P represents the degree of polarisation. Goedheer reports the following values for the molar concentrations at which the degree of polarisation is reduced to half of its maximum value chlorophyll a 6 x 10-4~ and chlorophyll b 3.5 x 10-3~. These results are consistent with Forster’s inductive-resonance explanation of concentration depolarisation.21 The sensitisation of the fluorescence of chlorophyll a by chlorophyll b was observed independently by Watson and Livingston4* and by D~yseiis.~~ These authors agree that the efficiency of transfer of energy from chlorophyll b to chlorophyll a (in an equimolar mixture) reaches a maximum of about one half at a concentration of 10-3~.At higher con- centrations of the donor self-quenching reduces the efficiency of transfer. These results agree surprisingly well with Forster’s prediction. 21 In the intact plant cell the fluorescence of chlorophyll a is sensitised by carotenoids. The corresponding sensitisation in vitro was recently observed by Teale,52 using equimolar mixtures of carotenoids and chloro- phyll dispersed in micelles. The local concentration of each of the pigments was about 0 . 1 ~ . Under these conditions when chlorophyll a was the acceptor and p-carotene lutein or fucoxanthol was the donor the probabilities of transfer were 0.1 0.6 and 1.00 respectively. When chlorophyll b was substituted for chlorophyll a the probabilities of transfer were reduced five-fold or more.When phzophytin was the acceptor there was no detectable sensitisation of its fluorescence. These marked individual differences suggest that the transfer is not simply the result of inductive resonance but is chiefly due to the formation of addition compounds between the donors and acceptors. In these micelles and in chlorophyll- free micelles the long-wavelength absorption limit of the carotenoids is shifted strongly toward the red. The chlorophyll absorption spectra are about the same in the micelles as they are in dilute solutions. In general the quantum yield of fluorescence does not depend upon the wavelength of the exciting light. However it has been observed that the yield for solutions of dyes and pigments is less than normal when the wavelength of the exciting light is on the red side of the first absorption band.In the earlier literature this behaviour was accepted as character- istic of such solutions and several ad hoc theories were proposed to explain it. L a v ~ r e l ~ ~ ~ and independently Weber and Teale45b have demonstrated that this decrease in the fluorescence yield occurs only when the con- centration of the pigment is relatively high. It appears to be the con- sequence of the formation of non-fluorescent dimers. In the absorption spectrum of the dimers the single bands of the monomer are split. As a result of partial dimerisation the quantum yield exhibits a minor minimum 6o Goedheer Doctoral Diss. Utrecht 1957. 51 Duysens Nature 1951 168 548. 62 Teale Nature 1958 181 415. 192 QUARTERLY REVIEWS when the wavelength of the exciting light is on the blue side and a sharp decline when it is on the red side of the (monomer) absorption band.In the antistokes region where the total absorption is small relatively low concentrations of dimer can produce a marked decrease in the quantum yield. It has been suggested53 that dimerisation which reduces the efficiency of fluorescence should increase the yield of phosphorescence. Recent observations by B r ~ d y ~ ~ are probably an example of this behaviour. He noted an intense luminescence with a maximum at A715 mp from con- centrated ethanolic solutions of chlorophyll a at - 193 OC. Dilute solutions (10-6~) exhibit neither the luminescence at 715 mp nor the concurrent weakening of the normal fluorescence whose maximum is at 670 mp.Analogous results were obtained with chlorophyll b. The author concludes that the appearance of the long-wavelength luminescence and the weaken- ing of the normal fluorescence are the result of dimerisation which is favoured by low temperatures. Properties of Solid Films.-After five years of intensive investigation in a number of laboratories and much enthusiastic and repetitive discussion it must be admitted that the case for or against the “biological solar battery” is not proved. It was suggested about ten years that the ordered aggregates of pigment and protein molecules which constitute the chloro- plasts might act as an organic analogue to the “solar battery” and so produce an electrochemical “splitting of water” leading to the reduction of carbon dioxide and the evolution of oxygen.This mechanism necessarily involves the following components a light-induced charge separation ; photoconduction of holes and electrons ; something analogous to two spatially separated reversible electrodes ; and a return path for electricity presumably through a continuous ionic solution. This concept has had a strong intuitive appeal to many biologists ; however detailed analyses of the problem56 indicate that it involves serious (although in principle not insurmountable) difficulties. Interesting positive evidence for the existence of some of the components of the biological solar battery have been reported. Electron spin resonance measurements5’ have proved that free electrons are produced by the illumination of chloroplasts and of some other chlorophyll-containing material.Films of chlorophyll and of similar pigments are photocon- ductive the mobile carrier being the positive hole rather than the 63 McRae and Kasha J. Chem. Phys. 1958,28 721. 64 Brody Science 1958 128 838. 66 Katz in “Photosynthesis in Plants” Iowa State College Press 1949 pp. 287-292. 68 (a) Brugger and Franck Arch. Biochem. Biophys. 1958,75,465; (b) Rabinowitch Discuss. Furuduy SOC. 1959 27 161. 67 (a) Commoner Heise and Townsend Proc. Nut. Acud. Sci. 1956 42 710; (b) Sogo Pon and Calvin ibid. 1957,43 387. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 193 e l e c t r ~ n . ~ * - ~ ~ Photochemical charge separation at the boundaries be- tween films of organic compounds has been 63 Several types of long-lived faint luminescence of chloroplasts and intact cells have been observed and sometimes interpreted as the result of the recombination of trapped electrons and holes.This interpretation although possible has in no sense been established. One of the weakest links in this chain of evidence is the lack of reliable determinations of the quantum yields of free electrons etc. While most of the available semi-quantitative information suggests that these yields are very small Tollen et aZ.64 state that their electron spin resonance measure- ments indicate a quantum yield of free electrons greater than 0.1. No details have been published in support of this statement which if con- firmed would be of primary importance. The photoconductivity of organic films appears to be a complex phenomenon58 and its analogy to the photoconductivity of inorganic semiconductors is probably more formal than physically There can be no reasonable doubt of the real importance of the spatial arrangement of the pigment enzymes and other substances present in the chloroplasts.This arrangement must play an essential role in the process of photosynthesis However the “solar battery” hypothesis as originally proposed and as still supported by some biologists appears to be as naive and unreasonable as the view of the opposing extremists viz. that efficient photosynthesis could occur in a homogeneous fluid system. Reversible Photochemical Reactions Reversible Photobleaching or Phototropy.-Under certain specific conditions chlorophyll undergoes reversible photochemical reactions. In 1937 Porret and RabinowitchG6 discovered that air-free methanolic solutions of chlorophyll are reversibly bleached when strongly illuminated.The effect is small; the radiation from a 2000-watt carbon arc produces only a 1 change in the absorption of red light by a 2 x 1 0 - 5 ~ solution of chlorophyll. The absorption spectrum of the reduced form has not been determined reliably but as compared with normal chlorophyll both of the major bands are decreased and the absorption in the intermediate region is increased. The extent of the bleaching is proportional to the square-root of the intensity of the absorbed light. The back reaction is too fast to be measured by ordinary classical methods and only the steady-state changes 68 Weigl J. Mol. Spectroscopy 1957 1 216. 69 Nelson J. Chem. Phys. 1957 27 864. 6o Arnold and Maclay Brookhaven Symposia in Biology 1958,11 1.Terenin and Putzeiko J. Chim. phys. 1958 55 681. 62 Terenin Putzeiko and Akrimov Discuss. Faraday SOC. 1959,27 83. 63 Tollin Brookhaven Symposia in Biology (Bioenergetics) in the press. 64 Tollin Sogo and Calvin J. Chim. phys. 1959,55,919. 65 (a) Kasha Rev. Mod. Phys. 1959 31 162; (b) Garrett Brookhaven Symposia ni 66 Porret and Rabinowitch Nature 1937 140 321. Biology (Bioenergetics) in the press. 194 QUARTERLY REVIEWS have been studied. The effect has been observed in methanol ethanol and acetone but not in dry or wet benzene. It requires the addition of 1 % of methanol to benzene to restore the steady-state change to half of its normal value. The effect is inhibited by oxygen; 5 x lO-%-oxygen in methanol largely eliminates the reversible change. Although there is some irreversible bleaching in the presence of oxygen the action of oxygen is not merely to convert reversible into irreversible bleaching.In an air-saturated solution in methanol the quantum yield of the irreversible reaction is about 4 X but the corresponding lower limit for the yield of the bleached form in an anaerobic solution is 3 x The extent of bleaching is not appreciably affected by moderately high concentrations of allylthiourea or isopentylamine. Certain oxidising agents (notably iodine and Methyl Red) greatly increase the bleaching probably by changing the nature of the reaction. These phenomena were studied intensively by Livingston and his co-workers and their results have been adequately reviewed by Rabinowitch. The measurements were rather poorly reproducible probably owing to undetected impurities in the solvents or pigments.Although no detailed explanation of the results has been established certain general conclusion may be safely stated. The dependence upon the square-root of the light intensity shows that the reverse thermal reaction is bimolecular in respect to the photo-products. This as well as the magnitude of the effect and the influence of solvents demonstrates that the steady-state bleached form is not the triplet state. Presumably the bleached species is a pair of radicals or ions which are formed by the reaction of an excited chlorophyll molecule with the solvent or with an adventitious (oxidising?) impurity. The high efficiency of the inhibition by oxygen suggests that the active excited species is the triplet state.These assumptions being made the generalised mechan- ism may be written as follows where S stands for the solvent or impurity and g + s for the pair of radicals or ions. At the steady state the change in the concentration of chlorophyll should be The values of ka and kb must be consistent with kS and k, as determined from flash-photolytic measurements and +T = 1 - +tl. Within the limits Ref. 1 pp. 4 8 3 4 9 4 1487-1501. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 195 of uncertainty of the available data this equation is consistent with the experimental results. If radicals are formed by the illumination of an air-free chlorophyll solution containing a suitable monomer such as methyl methacrylate they should induce readily detectable polymerisation. Two such studies have been performed,68 and a slow polymerisation was observed in ethanol or pyridine but not acetone or benzene.The polymerisation is greatly accelerated by the presence of reducing agents but is inhibited by oxygen. One authoPa reports that the yield in the absence of added reducing agents was “considerably below unity” and that the average molecular weight of the polymer was approximately lo5. This corresponds to a very inefficient initiation process +i< These results indicate that either the products g + s are not radicals or if they are radicals they do not initiate polymerisation efficiently. Ruppel and WWg demonstrated that chloro- phyll dissolved in hexane is not detectably photoionised. However this is not surprising since chlorophyll is not reversibly bleached in such solvents.Reversible Photo-oxidation.-In the same year that Porret and Rabino- witch discovered the reversible photobleaching of chlorophyll Rabinowitch and Weiss70 observed a rapid reversible reaction between chlorophyll and ferric salts. In anhydrous alcohol chlorophyll can be quantitatively regenerated from the yellowish product of the reaction between ferric salts and chlorophyll by the immediate addition of an excess of ferrous salt or other suitable reducing agent. However if the bleached solution is allowed to stand before the reductant is added or if the solvent contains water allomerised rather than native chlorophyll is formed. 71 In dry alcoholic solutions containing comparable amounts of ferric and ferrous salts an equilibrium between normal and oxidised chlorophyll is attained fairly rapidly.Illumination with red light displaces the equilibrium favouring the bleached form of chlorophyll. It appears probable though it has not been proved that the yellowish product is a reactive oxidised derivative of chlorophyll. A similar but transitory pale yellow solution is formed during the allomerisation of chlorophyll with dilute solutions of iodine or bromine. 71 The alternative interpretation of these observations which was advanced by Ashkinazi et aZ.,72 is not justified by the evidence which they cite. Linschitz et al. 73 studied the reversible photochemical reaction between chlorophyll and benzoquinone. This like the ferric reaction appears to be 68 (a) Uri J. Amer. Chem. SOC. 1952 74 5508; (b) Krasnovskii and Umrikhina Doklady Akad Nauk S.S.S.R. 1955,104 882.6 g Ruppel and Witt 2. phys. Chem. (Frankfurt) 1958 15 21. 70 Ref. 1 pp. 4-66 488 489 1499 1500. 71 Watson J. Amer. Chem. SOC. 1953,75,2522. 72 Ashkinazi et al. Doklady Akad. Nauk S.S.S.R. 1950,74,315; 1951,80 385; 1954 ‘3 (a) Linschitz and Rennert Nature 1952 169 193; (b) Korn Doctoral Diss. 102 767. Syracuse University 1955. 196 QUARTERLY REVIEWS a reversible photochemical oxidation of chlorophyll. If a mixture of chlorophyll and an excess of benzoquinone is illuminated in EPA (a mixture of ethyl ether pentane and ethanol) at -170"~ the chlorophyll is almost completely bleached. At this or lower temperatures the reaction products are stable and the chlorophyll derivative has a spectrum practically identical with that of the Molisch brown pha~e.7~At -75"c or above no appreciable colour change is produced by steady illumination presumably owing to the rapid rate of the reverse reaction.At a still lower temperature - 183 O C the photochemical reaction is inhibited. Either the photochemical reaction has an intrinsic energy of activation or it cannot occur in a rigid medium. No reaction was observed when chlorophyll was replaced by phaeophytin or tetraphenylchlorin. In view of the important role which the reversible oxidation of chloro- phyll may play in photosynthesis and in oxidation-reduction reactions sensitised by chlorophyll in vitro it is most unfortunate that neither of these reactions has been studied intensively or quantitatively. Reversible Photoreduction.-The reversible photoreduction of chloro- phyll was observed by Krasnovskii in 1948.75 Only a limited number of reducing agents (including ascorbic acid dihydroxymaleic acid cysteine hydrogen sulphide and phenylhydrazine) react in this way.The basic reagent phenylhydrazine reacts as efficiently in toluene as in ~ y r i d i n e ~ ~ but the other compounds reduce chlorophyll only in basic solvents. The first experiments were performed in (supposedly) dry pyridine and the reduced form of chlorophyll observed was a pink pigment having an absorption maximum at about 5200 A. The regeneration of the green colour of the solution occurs rapidly if air is admitted to the solution or other oxidising agents are added ; otherwise the green colour returns slowly. Chlorophyll is not completely regenerated; part of it is converted into phaeophytin. The details of these photochemical reactions and of the photoreduction of analogous pigments have been described in a series of papers by Krasnovskii Evstigneev and their co-workers which have been reviewed critically by Rabinowitch.77 In the photoreduction of chlorophyll an excited molecule of this sub- stance must react with one of the reductant. If the reacting molecule were in its first excited singlet state the reaction should quench the fluorescence of chlorophyll. Evstigneev et aZ.35 have shown that no such quenching occurs. It has also been demon~trated~~ that in dry pyridine high concen- trations of ascorbic acid have no effect upon the half-life of the triplet state of chlorophyll. Recent work by Bannister78 seems to render compatible these apparently inconsistent observations. He demonstrated that the 74 Weller J.Amer. Chem. SOC. 1954 76 5819. 7b Krasnovskii Doklady Akad. Nauk S.S.S.R. 1948 60,421. 76 Evstigneev and Gavrilova Doklady Akad. Nauk S.S.S.R. 1953 91 899. 77 Ref. 1 pp. 1501-1507; see also Rackow and Konig 2. Electrochem. 1958 62 78 Bannister Plant PhysioE. 1959 34 246. 482. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 197 photoreduction of chlorophyll by ascorbic acid does not occur detectably in dry pyridine but requires the presence of a moderate concentration of water. In pyridine containing 20 or 30 % of water and 0.1OM-ascorbic acid the pink pigment does not accumulate and reoxidation produces phao- phytin rather than the original chlorophyll. In such solutions the quantum yield of the reduction of chlorophyll is in the range 0.05-0.10. At low concentrations of water (2 or 3 %) the yield of the photoreduction product is small but the process is much more nearly reversible.Presumably the “dry pyridine” used by Krasnovskii in his original experiment^^^ con- tained several units % of water. The publication of Bannister’s results led P ~ g h ~ ~ to re-examine the effect of ascorbic acid upon the mean life of the triplet state of chlorophyll. His results may be summarised as follows In dry pyridine ascorbic acid (even in concentrations as high as 0.10~) has no detectable effect upon the decay of the chlorophyll triplet. In pyridine containing 20 % of water ascorbic acid noticeably shortens the mean life of the triplet and the corresponding value of the quenching constant k, is about 5 x lo5 l.mo1e-1 sec.-l. Quenching is also observable in pyridine containing 2% of water but under these conditions the quenching constant is about 1 X lo5 1.mole-l sec.-l.It should be noted that these rate constants are approximately 10,000-fold smaller than the corresponding constants for quenching by oxygen benzoquinone etc. The flash-photolytic t r a c e ~ ~ ~ - ~ l indicate that a second labile species having a longer life than the triplet state is also formed in wet pyridine containing ascorbic acid. Its mean life is about 0.01 sec. and it absorbs more strongly at 4800 than at 5520 A. These properties demonstrate that this labile substance is not Krasnovskii’s pink pigment but some forerunner of it. It was demonstrated in 1953 by Evstigneev and Gavri10va~~ that the pink pigment is not the primary product of photoreduction.The illumina- tion of suitable solutions of chlorophyll and ascorbic acid containing a reference and an inert electrode changes the potential of the electrode much more rapidly than it produces changes in the absorption spectrum. In the dark the original potential of the cell is restored in 2 or 3 minutes but the reappearance of the green colour is much slower. These differences are especially marked at low temperatures. The same authors observedso photo-conductivity when they illuminated a mixture of chlorophyll and phenylhydrazine in air-free dry pyridine. These and other related measure- mats77 demonstrate that the photoreduction of chlorophyll produces at least two labile intermediates in addition to the pink pigment. The flow sheet shown in Fig. 4 is consistent with the principal observa- tions of Bannister Evstigneev Krasnovskii and Pugh on the photo- reduction of chlorophyll and phaophytin.Its details are by no means 79 Evstigneev and Gavrilova Doklady Akad. Nauk S.S.S.R. 1953 92 381 ; 1954,95 381 ; see also Hendrich Roczniki Chem. 1958,32,107. Evstigneev and Gavrilova Doklady Akad. Nauk S.S.S.R. 1955,103,97. 198 QUARTERLY REVIEWS GH PhH FIG. 4. Intermediates in the reversible photoreduction of chlorophyll and phophytin. established and the process may be more complicated than the diagram indicates. The symbol PhH represents phaeophytin. The “pink pigments” are indicated by GH and PhH,; and the Evstigneev intermediate radicals and radical ions by GH, GH- PhH, and PhH-. Photochemical Reactions sensitised by Chlorophyll The facts presented in the preceeding discussion place severe restrictions upon the mechanisms which otherwise might be advanced in explanation of the role of chlorophyll in sensitised reactions.The results of kinetic measurements of naturally occurring photosynthesis are so complicated that not even the general nature of the primary act of this process can be established. However the accumulated information about the properties of electronically excited chlorophyll has proved usefuI8l in the elimination of mechanisms which previously seemed plausible. For chlorophyll-sensitised reactions occurring in homogeneous solutions the problem is much simpler and it probably is not unduly optimistic to hope that the mechanisms of some of these reactions will be determined with reasonable certainty within the next few years.Chlorophyll efficiently sensitises the photochemical auto-oxidation of a wide variety of reducing agents. In contrast relatively few oxidation-reduction reactions (not in- volving oxygen) are sensitised and these only in hydroxylic solvents. A large number of such reactions have been qualitatively investigated and a few have been studied intensively and quantitatively. The results of these studies and their possible interpretations have been summarised r e ~ e n t l y . * ~ ~ ~ ~ While some of the published interpretations are now out- dated it is doubtful if a complete re-examination of the problem is at present justifiable. The observation that certain photobiological processes occur efficiently at liquid-nitrogen temperatures has been cited as evidence that they are not simple chemical reactions but must involve some special mechanism (a) Brugger and Franck Arch.Biochem. Biophys. 1958,75,465; (b) Franck Proc. Nat. Acad. Sci. 1958 44 941 ; (c) Franck in “Handbuch der Pflanzenphysiologie” Vol. V Springer Heidelberg in the press. 82 Rabinowitch ref. 1 pp. 507-521,545-547,1507-1528. 83 Livingston in “Handbuch der Pflanzenphysiologie” Vol. V Springer Heidelberg in the press. LIVINGSTON PHYSICOCHEMICAL ASPECTS OF PHOTOSYNTHESIS 199 such as the “solar battery”. In view of the fact that the Rose Bengal- sensitised auto-oxidation of terpene has a quantum yield of about 0.25 at -156”~ in a mixture of organic solvent,E4 this interpretation of the biochemical reactions must be regarded with some suspicion. It was reported recentlyE5 that chlorophyll in dilute solution sensitises the photoisomerisation of poly-cis-carotenes. It does not detectably sensitise= the photoisomerisation of trans-retinene. These reactions are worthy of extensive quantitative study. Photoisomerisations sensitised by stable polyatomic molecules have not been previously reported and may be of importance in relation to photobiological processes. 84 Schenck Kinkel and Koch Naturwiss. 1954 41 425; see also Evstigneev and Gavrilova Doklady Akad. Nauk S.S.S.R. 1954 98 1017. Claes and Nakayama Nature 1959 183 1053.

ISSN:0009-2681    

DOI:10.1039/QR9601400174

出版商:RSC    

年代:1960

数据来源: RSC

摘要:

THE BORAZOLES By J. C. SHELDON (WILLIAM RAMSAY AND RALPH FORSTER LABORATORIES UNIVERSITY COLLEGE and B. C. SMITH THE compound B3H3N3H3 (I) is a colourless mobile liquid which was first prepared by Stock and Pohlandl from the addition compound of ammonia and diborane. Electron-diffraction measurements2 showed the molecule to be planar and symmetrical and Kekulk structures (11)3 have become accepted. An interesting comparison with benzene was given by Wiberg and Boh4 who called the compound inorganic benzene or borazol from which the familiar name borazole” is derived. Alfred Stock described the early investigations in his 1932 Baker Lectures,6 a review by Wiberg’ appeared in 1948 and a number of works contain sections on borazole.8 The hydrogen atoms of borazole may be replaced by other groups and the properties of the ring are modified by B- and N-substituents.LONDON) (BIRKBECK COLLEGE UNIVERSITY OF LONDON) H H H R Borazole and its derivatives have attracted much interest because of the resemblance to aromatic organic compounds but six-membered ring systems are quite common in inorganic chemistry. The properties of planar molecules such as the borazoles and boroxoles (111) can be related to dative n bonding between the atoms in the ring and the properties of “saturated” compounds such as B3H6N3H3Me39 and B3H6N3Me6lo0l1 *The name Borazine recommended by an A.C.S. sub-committee,6 is not yet accepted and the name borazole will be used throughout this Review. 1 Stock and Pohland Ber. 1926,59 2215. Stock and Wierl 2. anorg. Chem. 1931,203 228. Stock Wiberg and Martini Ber.1930 63 2927. * Wiberg and Bolz Ber. 1940 73 209. Patterson Chem. Eng. News 1956 34 560. Stock “Hydrides of Boron and Silicon,” Cornell University Press Ithaca N.Y. 1933 Chapter XIV. 7 Wiberg Nuturwiss. 1948 35 182 212. * Schlesinger and Burg Chem. Rev. 1942 31 1; Bauer ibid. p. 43; Lappert ibid. 1956 56 959; Bell and EmelCus Quart. Rev. 1948 2 132; Stone ibid. 1955 9 174; Gerrard J. Oil Colour Chemists’ ASSOC. 1959 42 625; Smolin and Rapoport “s-Triazine and Derivatives,” Interscience Publishers New York and London 1959 Chapter XI. * Bissot and Parry J. Amer. Chem. SOC. 1955,77 3481. 1°Burg J. Amer. Chem. SOC. 1957 79 2129; Campbell and Johnson ibid. 1959 81 3800. l1 Trefonas and Lipscomb J. Amer. Chem. SOC. 1959 81 4435. 200 SHELDON AND SMITH THE BORAZOLES 201 also depend on dative bonding from nitrogen to boron.Nitrogen is a strong electron-donor but is unable to accept electrons whereas tervalent phosphorus forms co-ordinate bonds with borine" and trimethylborine which are strengthened by the back release of electrons from B-H and B-Me bonds to vacant phosphorus d orbitals.12 The remarkable stability of B3H6P3Me23 no doubt depends on such back-donation. Since the molecule has a chair configuration (IV) the angle PBH is approximately 109" and each phosphorus atom can receive electrons from four B-H bonds. Similar compounds B ,Me 6P3Me6 B ,H6P3(CF3) 6 andB3H6As3Me6 have also been-prepared by Burg and his co- w o r k e r ~ ~ ~ ~ ~ ~ but it is of interest that analogues \ k-7~ receive electrons from only two B-H bonds and the angle PBH would be 120" have not yet been Ye M6p-w v Y /" of borazole where each phosphorus atom could Hy/ Me H ( 1 ~ ) prepared.Synthesis of the Borazoles Synthesis of the Borazole Ring.-Methods developed for the synthesis of the borazole ring may be viewed as condensations between amines and boron hydrides or halides at about 200" with the elimination of hydrogen or hydrogen halide 3BX3 + 3NR3 -f B3X,N3R3 + 6RX (e.g. X = H CI Br Alk; R = H Alk) Trialkylborines also condense with amines yielding B-trialkylborazoles and alkanes at temperatures higher than 300". Frequently derivatives of the amines boron hydrides or boron halides are used either because they are more readily available and handled than the simple reagents or because they yield directly borazoles with desired substituent groups. Condensation between boron hydrides and amines.This was the first reaction developed to give products recognised as borazoles. Excess of ammonia reacts with diborane at 190" to give diboron tri-imide B2H +3NH3 -+ B2(NH)3 + 6H2 but equivalent quantities of diborane and ammonia (i.e. B:N ratio 1:l) react at a similar temperature and at atmospheric pressure yielding hydrogen a polymer of composition inter- mediate between (BNH2) and (BNH), and borazole. Diborane and * I.U.P.A.C. makes no recommendation as to a name for BH,. Pending a decision The Chemical Society uses "borine". The series B,H,n are the polyboranes. la Stone Chem. Rev. 1958 58 101; Ahrland Chatt and Davies Quart. Rev. 1958 12 265. 1955 8 199. 1958 80,3198. l3 Burg and Wagner J. Amer. Chem. SOC. 1953 75 3872; Hamilton Acta Cryst.lo Stone and Burg J. Amer. Chem. SOC. 1954 76 386; Burg and Brendel ibid., 202 QUARTERLY REVIEWS tetraborane readily form addition compounds with ammonia at room temperature. B2H6,2NH3 and B4Hlo,2NH315 contain the (NH3-BH2* NH3)+ cation,16 which may well be related to the primary condensation products leading to the formation of borazoles. These ammonia addition compounds undergo the same condensation reaction as equivalent quantities of diborane and ammonia and after a short heating at 20O0 the reaction tube is cooled in liquid air and opened in vacuo. Hydrogen is pumped away and borazole (b.p. 53") remains the only volatile product and is distilled out in high purity. The yield of borazole is 33% from B,H6,2NH31 and 40 % from B,H,o,2NH,.3 Excess of diborane and ammonia (or more specifically diborane and the diborane-ammonia addition compound) yield B,H,N (V)17 which is rather unstable and decomposes slowly even at room temperature giving several products among which are diborane and '\ ' H borazole.B2H,N forms a 1 :1 ammonia addition com- '\ 6 /N\~/ pound which decomposes at 200" producing a 45% H' 'H' 'H yield of borazole. (v) Use of the boron hydride-amine condensation for the preparation of borazoles is severely limited by the difficult and involved preparation of the hydrides themselves. The early investigations relied on the preparation of the hydrides by the action of acids on suitably prepared metal borides a yield of 11 % of boron hydrides is the best recorded for the action of 8~-phosphoric acid on magnesium boride.18 More convenient sources of boron hydrides have since proved to be the hydrogenation of boron trihalides in an electric discharge and the decomposition of metal borohydrides.Even so the use of boron hydrides is unsatisfactory since they require handling in the absence of oxygen moisture and grease in a carefully designed vacuum line and the small amounts of borazoles prepared undoubtedly restricted the early investigations. Nevertheless Schlesinger and his co-workers succeeded in preparing B-methyl- B-dimethyl- and B-trimethyl-borazoles by the condensation of methyldiboranes and ammonia1 and similarly N-methyl- N-dimethyl- and N-trimethyl-borazoles* by the condensation of diborane with mixtures of ammonia and methylamine.20 The usual H l6 Kadoma and Parry J. Amer. Chem. SOC. 1957,79 1007. l6 Schultz and Parry J.Arner. Chem. SOC. 1958 80 4; Shore and Parry ibid. pp. 8,12; Parry and Shore ibid. p. 15; Shore Girardot and Parry ibid. p. 20; Parry Kadoma and Schultz ibid. p. 24; Taylor Schultz and Emery ibid. p. 27; Nordman and Peters ibid. 1959 81 3551. l7 Schlesinger Ritter and Burg J. Arner. Chern. Soc. 1938 60 2297; Hedberg and Stosick ibid. 1952,74 954. 1s Wiberg and Schuster Ber. 1934 67 1805. eo Schlesinger Ritter and Burg J. Arner. Chern. Suc. 1938 60 1296. * In this and subsequent names use of N implies NN'N" and similarly for B. Schlesinger Horvitz and Burg J. Arner. Chem. SOC. 1936,58,409. SHELDON AND SMITH THE BORAZOLES 203 yield of a particular borazole was a few millilitres of vapour requiring careful fractionation from other products. Condensation of a borohydride with an ammonium salt.A major advance in borazole synthesis was the replacement of boron hydrides by lithium borohydride which is readily handled and is now commercially available. The borohydride readily undergoes a partial condensation with ammonium or alkylammonium chlorides usually in some ethereal solvent yielding hydrogen and lithium chloride at room temperature. After removal of the ether the reaction can be completed by pyrolysis of the residual product at 200 to 250" 25" 250" 3LiBH + 3R*NH,CI -+ 3BH,.NHR + 6H + 3LiCI -+ B,H,N,R + 9H + 3LiCI Despite the limited supply of borohydride in the initial work Schaeffer and Anderson21 were able to obtain N-trimethylborazole in gram quantities and in better than 70 yield from methylammonium chloride. The formation of the borazole may be accomplished in one step if a high-boiling ether e.g.dihexyl ether is used as solvent. N-Triethyl- N-tri-n-propyl- and N-tri-isopropyl-borazoles have all been prepared by these methods.22 Borazole itself has been obtained by the pyrolysis at 300" of lithium borohydride and ammonium chloride diluted with powdered "Pyrex" glass.23 The borohydride method has not been widely adopted and is not suitable for the synthesis of B-substituted borazoles. Condensation between boron halides and amines.-This reaction is now accepted as the most convenient preparation of borazoles. The action of boron trichloride and ammonia received early investigation though the only products reported were a boron trichloride-ammonia addition compound triaminoborine and diboron tri-imide.24 As early as 1889 found that aniline and boron trichloride reacted to give hydrogen chloride and a white solid considered to be BClNPh.This work was repeated by Jones and Kinney26 who showed that Rideal's product was a trimer and they suggested its relation to borazole. Jones and Kinney's method which was the first borazole synthesis well adapted to an ordinary laboratory gave an 87-5 % yield of B-trichloro-N-triphenylborazole. Aniline was added to boron trichloride in cold benzene and after 24 hours at room temperature the mixture was filtered hot and the borazole crystallised on cooling 3BCI + 9NH,Ph + B,CI,N,Ph + 6NH,PhCI 21 G. W. Schaeffer and Anderson J. Amer. Chem. SOC. 1949,71 2143. 22 Hough G. W. Schaeffer Dzurus and Stewart J. Amer. Chem. SOC. 1955,77,864. 23 G. W. Schaeffer R. Schaeffer and Schlesinger J.Amer. Chem. SOC. 1951 73 26 Rideal Ber. 1889 22 992. 1612. Martius Annalen 1859 109,79; Joannis Compt. rend. 1902 135 1106. Jones and b e y J. Amer. Chem. Soc. 1939,61 1378. 204 QUARTERLY REVIEWS A1 t hough B-trichlor 0-N- tri-p- tolyl- and B- trichlo r 0-N- trianisyl- borazoles have been prepared in a similar fashion,27 difficulty has been reported in making analogous N-trialkylborazoles. This general synthetic route has received more attention following the report by Brown and Laubengayer28 that ammonia gave only traces of borazole with boron trichloride yet ammonium chloride gave a 50 % yield of B-trichloro- borazole. Two variations of the method are described a dry method involving the passage of boron trichloride vapour over ammonium chloride at 170° and a liquid-phase method employing boiling chlorobenzene 3BC13 + 3NH4CI -+ B,CI,N,H + 9HC1 It is likely that the use of the hydrochloride rather than the free amine moderates the vigour of the reaction and allows the formation of simple products.The use of methylammonium chloride in the above synthesis yields B-trichloro-N-trimethylborazole,2g~30 but higher monoalkyl- ammonium chlorides require the presence of a calculated quantity of tertiary amine to facilitate the elimination of hydrogen chloride :31 R.NH,CI + BCI -+ R*NH,,BCI + HCI 3R.NH2,BCI + 6Me,N + B,CI,N,R + 6Me,N,HCI If a substituted boron chloride is employed (e.g. phenyl- and dialkyl- amino-boron dichlorides give B-triphenylbora~ole~~p~~ and B-trisdi- alkylaminob~razoles~~ respectively) it appears preferable to use the free amine rather than the hydrochloride owing to the decreased activity of the boron chloride species 3BCI,.NR2 + 9NH += B,(NR,),N,H + 6NH4CI The condensation of an alkylamine with a substituted boron chloride requires the presence of a tertiary amine for completion 3R-NH2,BCI,.NR’ + 6Et3N += B3(NR’,),N,R3 + 6Et3N,HCI The condensation of methylamine with fluorodimethylborine also requires vigorous conditions i.e.heating to 400° but gives a B-trifluoroborazole 27 Kinney and Kolbezen J. Amer. Chem. Soc. 1942,64,1584; Kinney and Mahoney J. Org. Chem. 1943 8 526. 28 Brown and Laubengayer J. Amer. Chem. SOC. 1955,77 3699. 29 Hohnstedt and Haworth Abstracts of Papers 132nd Meeting Amer. Chem. SOC. New York N.Y. September 1957 p. 8 s. 30 Ryschkewitsch Harris and Sisler J. Amer.Chem. SOC. 1958 80 4515. 31 Turner and Warne Chem. and Ind. 1958 526. 32 Ruigh 16th International Congress of Pure and Applied Chemistry Paris 1957. “Papers Presented to the Section on Mineral Chemistry,” Butterworths Scientific Publications London 1958 p. 545. Mikhailov and Kostroma Zhur. obshchei Khim. 1959 29 1477; Mikhailov Blokhina and Kostroma ibid. p. 1483. 34 Niedenzu and Dawson J. Amer. Chem. SOC. 1959,81 3561. SHELDON AND SMITH THE BORAZOLES 205 and methane rather than a B-trimethylborazole and hydrogen flu~ride?~ Aubrey and La~pert,~ have reported an interesting variation of borazole synthesis involving the cyclisation of trisalkylaminobcgines and the elimination of alkylamines at 200" 3B(NHR) -+ B,(NHR),N,R + 3NH,R The B-triaminoborazole evolves alkylamine at temperatures of 300" and higher by linking several borazole molecules with B-alkylimine groups.Condensation proceeds until a highly cross-linked polymer is obtained. This is stable up to 600" and is a possible forerunner of some technologically useful materials. It appears that the blocking of the nitrogen atoms in these polymers by alkyl groups is sufficient to avoid total condensation to boron nitride at high temperatures An extension of this method is the cyclisation of amino(butoxy)-borines e.g. (BuO),BNHEt and BuOB(NHEt), to give B-tributoxy-N- triethylborazole with elimination of ethylamine or butanol. Hydrazine hydrobromide does not give an N-triaminoborazole with boron tribromide at 200" but gives B-tribromoborazole with evolution of nitr~gen.~' Substitution of the Borazole Ring.-It is possible to prepare a variety of new borazoles by the substitution of existing borazoles at B-H and B-Cl ring sites.The moderate tendency of amines to undergo substitution NR + AB -+ NR,A + RB is in contrast with the reactivity of boron trihalides borines and a1 kylborines and borazoles usually experience substitution at the boron atom only. The availability of B-trichloro- borazole and the non-existence of N-trichloroborazole have given a further emphasis to reactions at borazole B-positions. Substitution of the B-H group. (i) B-Alkylation and B-arylation. Trimethylborine was among the first reagents found capable of this reaction and gave with borazole or N-methylborazole a mixture of the corresponding B-methyl B-dimethyl and B-trimethyl derivatives :,O 2B,H,N,H + 2BMe -f 2B,H2MeN,H + B2H,Me 35 Wiberg and Horeld 2.Naturforsch. 1951 6 B 338. 37 Ernel6t.u and Videla Proc. Chem. SOC. 1957 288; J. 1959 1306. Lappert Proc. Chem. SOC. 1959 59; Aubrey and Lappert J. 1959 2927. 206 QUARTERLY REVIEWS Alkylation and arylation of borazoles now appears to be a practical procedure since the demonstration that Grignard and organo-lithium reagents react With N-trimethyl- and N-triphenyl-borazole giving satisfactory yields of substituted products :38 B,H,N,Ph + PhMgBr -+ B,H,PhN,Ph + HMgBr B-Mono- and B-di-substituted borazoles can be obtained from appropriate quantities of N-triphenylborazole and alkylating reagent. N-Trimethylborazole however gives mixtures of B-mono- B-di- and B-tri-substituted products. (ii) B-Halogenation. Only one substitution product B-dibromo- borazole appears to have been characterised from the products of direct halogenation of b~razole.~ The action of chlorine on borazole gives a white crystalline sublimable product (of unknown composition) and hydrogen chloride at high temperature.Bromine forms an adduct B,H3N3H3,2Br with borazole at 0" and this complex decomposes at 60-100" into colourless sublimable crystals of B-dibromoborazole and hydrogen bromide. Borazole reacts at room temperature with boron trichloride and boron tribromide giving mixtures of B-mono- and B-di-halogenob~razoles.~~ B-Trihalogenoborazoles can be prepared by the action of hydrogen halides on borazoles at elevated temperatures. At room temperature hydrogen halides form the complexes B3R3N3R3 3HX which eliminate hydrogen4 ( R = H X = Br) at 100" or methane39 (R = Me X = C1) at 450° yielding B-trihalogenoborazoles.(iii) B-Hydroxylation and B-alkoxylation. Water40 and methano141 give 3:l adducts with certain borazoles which decompose on gentle heating into B-trihydroxy- and B-trimethoxy-borazoles. Ethanola1 appears to behave similarly. These substitution reactions have not generally proved convenient preparations of B-substituted borazoles. B-Trichloroborazole however has found favour as a preparative intermediate and the reactions described below find promising applications. Substitution of the E C l Group. (i) Replacement by hydrogen. Ethereal lithium aluminium hydride readily reduces B-trichloroborazole to borazole but difficulty was found in isolating the products and the use of lithium borohydride proved more sati~factory.~~ The reduction of B-trichloro-N-triphenylborazole by lithium aluminium hydride proceeded without complications to give an 85 % yield of N-triphenylboraz~le.~~ Several laboratories concurrently reported the reaction of Grignard and organolithium reagents with (ii) B-Alkylation and B-arylation.Smalley and Stafiej J. Amer. Chem. SOC. 1959 81 582. Wiberg and Hertwig 2. anorg. Chem. 1947 255 141. R. Schaeffer Steindler Hohnstedt Smith Eddy and Schlesinger J. Amer. Chem. 40 Wiberg Hertwig and Bolz 2. anorg. Chem. 1948 256 177. 41 Haworth and Hohnstedt J. Amer. Chem. Soc. 1959 81 842. Soc. 1954,76 3303. SHELDON AND SMITH THE BORAZOLES 207 B-trichloroborazoles to give B-trialkyl(aryl)borazoles.2g~30~a Sisler and his co-workers were able to prepare the intermediate BB’-dibutyl-B”- chloro- and B-butyl-B’B”-dichloro-N-trimethyl-borazoles by adjusting the reagent proportions but found the exchange of ethyl groups between B-chloro-B’B”-diethyl- and BB‘-dichloro-B”-ethyl-N-trimethylbora- zoles too rapid to allow their separation and fractionation of the mixture gave only B-trichloro-N-trimethylborazole in the low-boiling fraction.No compounds containing B-peduoroalkyl groups have been characterised4* and attempts to prepare a B-perfluoroalkyl-borazole by the action of heptafluoro-n-propyl-lithium on B-trichloro-N-triphenyl- borazole were unsuccessful. There is a brief report that B-trichloroborazole undergoes a Friedel-Crafts reaction with benzene and aluminium chloride in boiling chlorobenzene to give a 24 % yield of B-triphenylb~razole.~~ Very pure sodium alkoxides react with B-trichloroborazoles giving B-trialkoxyborazoles *46 (iii) Metathesis with metal salts.Similarly it now appears that CN CNS Br NO2 and NO3 groups may be introduced into the borazole ring by the reaction of sodium or silver salts of the appropriate anion with B-trichloroborazoles.47 (iv) Amination. Primary and secondary amines give B-trisalkylamino- borazoles with B-trichloroborazole but it is doubtful whether ammonia gives an isolatable B-triaminoborazole :3494s B,C13N3R + 3NHR’ -+ B,(NR’,),N,R + 3HCI Amination of the B-chloro-group proceeds so readily that stoicheiometric quantities of the reagents may be employed and even amine hydro- chlorides may be used though yields are lower. Chemistry of the Borazoles The chemistry of borazoles described in the above sections is now considered in detail and with particular regard to postulated mechanisms.Ring Formation.-It is known that hydrogen linked to boron is easily replaced by electronegative groups. Water halogens and hydrogen halides for example react with diborane giving B-0 B-C1 and B-Br bonds. Similarly ammonia and diborane at 400” give boron nitride (BN),,l 48 Groszos and Stafiej J. Amer. Chem. SOC. 1958,80 1357. 44 Lagowski and Thompson Proc. Chem. Soc. 1959 301. 46 Niedenzu and Dawson Angew. Chem. 1959,651. 46 Bradley Ryschkewitsch and Sisler J. Amer. Chem. Sac. 1959,81 2635. 47 R. Schaeffer Brennon and Dahl Abstracts of Papers 133rd Meeting h e r . Gem. SOC. San Francisco Calif. April 1958 p. 3 7 ~ . 48 Gould U.S.P. 2,754,177/1956. 208 QUARTERLY REVIEWS whose structure consists of layers of coplanar condensed six-membered rings which may be regarded as an infinitely large “polyborazole” (VI).49 The reaction of ammonia and diborane at only 200” gives borazole and a poorly characterised polymer of approximate composition (BNH), which can be viewed as a partially cross-linked polyborazole.It therefore seems likely that the condensation of amines with boranes proceeds in a stepwise fashion. Wiberg elaborated this concept in detail and postulated the following condensation chain’ on the grounds of the existence of compounds related to (a) (b) and ( c ) BH,,NH and other compounds of type (a) are well characterised. Compounds of type (b) are well known although the simplest tend to polymerise. For example BH,.NH is reported only as a high polymer,’ BH,.NHMe is a and BH,.NMe is monomeric only at high temperatures and is a dimer7 or trimerll at room temperature.The only examples of type ( c ) compounds BMe:NPh50 and B(0Me) :NH,4 have been shown subsequently to be B3Me3N3Ph,51 and B(OMe),,NHal respectively. It is therefore possible to propose an alternative condensation sequence where cyclisation is not the final step but occurs as an equilibrium between (b) and the trimer ( d ) Compounds related to ( d ) are known to be converted into borazoles on heating and there is evidence that this is possible without the intermediate formation of type (b) and ( c ) compounds. For example borazoles react with three molecules of water to give in the first place adducts closely similar to compound ( d ) which yield on pyrolysis either boroxoles 4vPease Nature 1950 165 722; Acta Cryst.1952 5 356; J. Amer. Chem. Soc. 1952,74,4219. 61 Becher Z. anorg. Chem. 1957 289 262. Wiberg and Hertwig 2. anorg. Chem. 1948 257 138. SHELDON AND SMITH THE BORAZOLES 209 B~R303 on the one hand or B-trihydroxyborazoles and boron nitride on the other H2O H*O B,Me,N,Me -+ B3Me30339 B,Me,N,H -4 B,Me,0,40 H 2 0 Ha0 B,H,N,Me -+ BS(OH)gN3Me840 B,H,N,HS -+ (BN),* It is difficult to rationalise the formation of two kinds of ring borazoles and boroxoles if their formation is consistently systems due to depolymerisation of compounds of type ( d ) and subsequent reiyclisation via species of types (b) and (c). The reaction of boron trichloride and amines yields aminodichloroborines BC1 2.NR2 but trimers of these have not been reported and it remains to be seen if they are unstable intermediates in the formation of B-trichloroborazoles.Borazoles are unstable at high temperatures giving involatile polymers. The extent of this further condensation can be controlled either by blocking the ring positions with alkyl groups or by choice of temperature and is well illustrated by the work of L a ~ p e r t ~ ~ (see p. 205). Controlled pyrolysis of borazoles is certain to yield some interesting polyborazoles and this technique has already been shown to give B,N,H, and B,N5H which are analogues of biphenyl and na~hthalene.~~ Borazole is relatively stable in the gas phase and virtually no decomposition is observed after half an hour at 200". At 500" only 27% decomposes in the same time giving hydrogen and a material best represented by (BNH),.It is notable that liquid borazole is much less stable even at room temperature and this is consistent with the decomposition's being a condensation between several molecules. Addition Compounds.-The borazole ring possesses both electron acceptor (boron) and donor (nitrogen) sites and is capable in principle of giving several types of addition compounds. Class I are considered to be formed by dative 0 bonds whereas Class I1 are viewed as being formed by dative T bonds. D D Class I Class II 62 Moews and Laubengayer Abstracts of Papers 136th Meeting Amer. Chem. SOC. Atlantic City N.J. September 1959 p. 53 N. 210 QUARTERLY REVIEWS a-Bonded compounds. Three possible types of a-complexes are given above (A = electron acceptor D = electron donor). It is to be expected that the aromatic character of the borazole ring would greatly reduce the donor or acceptor potential of the nitrogen or boron atoms and it is not surprising that possible examples of compounds of type (a) are unknown and type (b) are rare.Wiberg4 found that addition of ammonia trimethyl'amine and dimethyl ether to borazole occurred slowly over a period of days to yield amorphous products of no significant composition. Moreover borazole is reported to form no addition compound with or to be appreciably affected at 100" by di-n-butyl ether.42 Hence it is doubtful whether definite addition compounds are given by electron donors with borazole or even whether the addition products are derivatives of borazole rather than linear polymers due to ring cleavage. There are several well-defined addition compounds of the form B3H3N3H3,3AD (AD = H20 MeOH HCl HBr MeI) and although there is no direct evidence about their structures they have been regarded consistently as compounds of type (c).The molecules AD above are known to form adducts in other systems by ionic dissociation and the electronegative fragment of AD is therefore assumed to be the electron donor which adds to boron. Hence B H N H 3H20 is formulated as a trimeric amino- HO;B,+;B<:, 3 3 3 3 9 h y dr ox y bo rine (VII) . Y ,OH H- 4~ ,H H N It is notable that the same ring structure has been established by X-ray diffraction for B,H6N3Me6.11 There are more examples of type ( c ) compounds where six dative CY bonds replace the borazole 7~ bonds than compounds of types (a) and (b) which contain only three dative cr bonds.The compounds B,H3N3H3,3AD eliminate hydrogen on pyrolysis and show no tendency to dissociate into the original components. Hence these addition compounds are probably the intermediates in the substitution reactions of borazole with reagents AD which yield B3D3N3H3 e.g. H'Y+ +Y\H H"H (VII) B,H,N,H + 3HCI 3 B,H,CI,N,H6 3 B,CI,NBH + 3H It is significant that addition compounds are formed with three molecules of AD even in the presence of excess of borazole. The one exception appears to be B3H3N,H3,2Br,.4 It may be supposed that addition of the first AD molecule reduces the borazole resonance energy and this can be offset by the formation of an adduct containing the maximum number of donor-acceptor bonds. Consequently 1 1 and 1 2 addition compounds are unstable and hence rare.n-Bonded complexes. The presence of T molecular orbitals in the borazole ring suggests the possibility that compounds of class I1 are SHELDON AND SMITH THE BORAZOLES 21 1 formed with molecules possessing similar 7~ orbitals but this has not been investigated seriously. Benzene is isosteric with borazole and favourable T orbital overlap may be possible between them. The substituent groups of the borazole and benzene rings will affect the stability of such complexes and one might expect the 1,3,5-trinitrobenzene-hexamethylborazole system to yield a significantly stable addition compound. In this connection it is interesting that B-trichloro-N-tri-p-tolylborazole and B-trichloro- N-tri-p-anisylborazole crystallise from benzene with one molecule of solvent per borazole Exchange Reactions.-It is difficult to view the substitution of borazoles by organometallic reagents alkylborines and boron halides described in earlier sections as proceeding through addition compounds of class I (c) and moreover these reagents appear to react in a stepwise manner giving mono- di- and tri-substituted borazoles rather than the complete substitution shown by water and hydrogen chloride (i) B,H,N,Ph + RM -f B,H,RN,Ph + MH (R = Me Ph; M =.Lf MgCI) B,H,N,Ph + 2RM -f B,HR,N,Ph + 2MH (ii) B,CI,N,Me BunMgB B,BunCI,N3Me3 B,Bun,CIN,Me, B,Bun,N,Me (iii) B,H,N,H Bxa+ B,H,XN,H, B,HX,N,H (X = CI Br) (iv) B,H,N,H,Me BMe3+ B,H#MeN,H,Me B,HMe,N,H,Me B,Me,N,H,Me One possible mechanism involves a bridged complex (VIII). 8-N 8-N R 6- N \ / \ d '6-R t RIM = t( ,B\~M s N/ B - R ' ~ RM 'B-d 8-N R b-d (VI I0 [R R' = H Alk Hal; M = MgCI(Br) BAlk, BHal,] Sisler and his co-worker~~~ postulated such a mechanism for the B-ethyl exchange observed between BB'-dichloro-B"-ethyl- B-chloro-B'B"- diethyl- and B-triethyl-N-trimethyl-borazoles and pointed out that the absence of exchange in the corresponding B-n-butylchloroborazoles may be related to the varying stability of complex (VIII) with methyl or n-butyl groups in the bridging positions.Although there is no direct evidence for this exchange mechanism it is nevertheless useful in correlating a number of observations. For example B-trichloroborazole 212 QUARTERLY REVIEWS may be reduced by lithium aluminium hydride or borohydride in ethereal solvents but the resulting borazole cannot be distilled from the products of lithium aluminium hydride reduction unless excess of lithium hydride is added.42 It seems likely that the borazole may form a complex of type (VIII) with the AlH3 or BH3 species produced in the reaction.However if the borane complex dissociates to any extent diborane will be evolved and the dissociation will become complete allowing isolation of the borazole by distillation. This will not occur with the aluminium hydride complex as this hydride is involatile. Simple Substitution Reactions.-Although many borazole substitution reactions appear to involve one or more types of molecular complex a number of reactions may well be of the familiar kind of nucleophilic substitution of organic chemistry B-CI + EtO- 3 B-0Et + CI- B-CI + BH,- -f B-H B-CI + Br- B-Br + CI- + CI- + i(B,H,) An investigation of the chloride-ion exchange of N-trisubstituted B-trichloroborazoles might confirm the existence of SJ or S'2 borazole reactions and provide the first detailed evidence of ring activation with variation of nitrogen substituent.Although there are no known examples electrophilic substitution is expected to occur at the borazole nitrogen atom. By analogy with other fields of chemistry it is probable that NO+ NO2+ and PhN2+ species would react with borazoles giving products that can be correlated with initial N-substitution. Physical Properties The melting points and boiling points of a number of borazoles are given in Table 1. All are colourless volatile liquids or crystalline solids and use of the letters p d and n indicates where vapour pressure density and refractive index data are available.Despite the chemical differences between borazoles and their benzene analogues properties which depend on structural similarities are frequently compared. For example the ratios of the absolute boiling points of a number of methyl analogues39 are approximately 0.93. Several groups of workers have studied liquid borazole and values of the viscosity obtained by Eddy Smith and Miller53 fall close to the straight line log = 4-45 x lo4 x T-2 - 3.037 (77 poise T'K). This indicates association at low temperatures and provides an interesting difference between borazole and benzene. 68 Eddy Smith and Miller J. Arner. Chem. SOC. 1955 77 2105. SHELDON AND SMITH THE BORAZOLES TABLE 1 Compound M.p. B.p./mm. Data Ref. 213 €3 3F3N3Me3 B 3C13N3H3 B ,C13N 3Et 3 B3C13N3Bun3 B3C13N3(CBH1&3 BsClaNsPhs BSC13NS(CBH4'Me) 3 B3ClSHN3Hs B3ClHzNsHs B 3C13N3Me3 B3C13N3(CsH4.0Me) B3Cl,BunN3Me3 B ,ClBu 2N3Mea B3BraNsH3 B ,Br 3N3Et B ,Br ,N3Ph3 B3Br2HN3H3 B3BrH2N3H3 B 3H3N3H3 B3H 3N3H ,Me B3H3N3HMez B3H3N3Me3 B3HsN3Et3 B3H3N3Pr n3 B3HsN3Pri B3H 3N3(C BH 11) 3 B3H3N3Ph3 B 3H 3N 3(C a 4.OMe) 3 B 3H 2MeN3H3 B 3H2MeN3H,Me B3H,MeN3Me3 B 3H2 MeN3Ph B3HMe,N,H3 B3HMe,N3H,Me B3HMe,N3Me3 B3HMe,N3Ph B3HMeEtN3Ph3 B 3HPh 2N3Ph3 B 3Me3N3H3 B ,Me3N3H2Me B3Me3N3Me3 B ,Me,N 3Ph3 B ,MeEtPrnN3Ph3 B ,Me ,Bu nN,Ph B3Et3N3H3 B3Et3N3Me3 B3Et3N3Ph3 B3Prn3N3H3 B3Pr n3N3Ph3 B,Pri3N3H3 gH 3N3(CBH4'CH 3) 3 B 3H 2PhN3Ph 3 85" 84 162-1 64 57-59 30 21 7-2 19 273-275 308-309 23 3-23 8 3 3.0-3 3 *5 -4 to-6 - 34.6 128-1 29 292-293 - 34.8 - 56.3 -1 t o o -49.1 - 6.5 160-161 149-1 50 137-138 - 59 78-82 49-5-50 98.9 142 21 5 - 48 206 128 207 31.8 99 131 113 - 54 267-269 1-2 169-171 169-171 224" P 192* PY d 1 15-1 20/0*5 152* 101/1.1 109-5* 122/ 1 - 1 167* 122" 53 84* 108" 132 184* 225* 203* 87* 124* 162 107* 139* 187 125* 158; 221 66-67 98/1*8 108/9 70/0.5 P n P n P P P P P d n P d n P d P d n P 4 n P P P P P P P d n 35 28,37 31,34,54 31,54 31 54 26,33,43,54,55 27,54 27.54 23' 30 23 30 37,42 54 33 23 23 4,22,53,54 20 20 9,20,21,22,40,54 22.54 22- 22 54 38,54,55 54 54 19 20 38 38 38 19 20 38 38 38 38 19,40 20 38,39 43,55 38 38 56 30 43 57 43 57 * Estimated from vapour pressure data 54 Hohnstedt and Haworth J.Amer. Chem. SOC. 1960,82 89. 65 Becher and Frick 2. unorg. Chem. 1958 295 83. 5B Zhigach and Krongauz Proc. Acud. Sci. (U.S.S.R.) 1956 111 725. 57 Hawthorne J.Amer. Chem. SOC. 1959 81 5836. 214 QUARTERLY REVIEWS TABLE 1-continued Compound M.p. B.p./mm. Data Ref. 1 1 I I 1 1 I 1 1 1 1 1 1 1 1 1 1 1 6 197-198 -18 to -17 129-1 32 185-1 87 127-128.5 -37 to -39 98-99 184-1 85 270 413415 95-130 112-120 110-111 53-54 84-87 81-84 85-105 112-113 52-55 1 52-1 5 5 > 300 110/0*6 140/1*1 94/0*7 110-112/1.3 62-65/0.07 80jO-10 101-103/0*15 85-87/0* 10 153-1 56/0*4 185-1 87/0*07 130-1 34/0-30 120-1 25/0*52 43 n 32,57 n 30 43 43 57 32 n 30 43 32,45,55 38.55 33; 43 26 27 41 n 46 41 n 46 n 46 n 46 n 46 d n 36 46 46 48 1 lO/0-1 d,n 36 34,48 145-150/0.1 d n 36 164/10 34,48 84 34 106/0-03 d n 36 158/0~005 d n 36 135/0*01 d,n 36 103/0-04 d n 36 36 36 34 >250/0-005 d n 36 94-9613 34 Isomerism.-& and N-Methylborazole are examples of isomers whose structures are known unambiguously.From N-methylborazole Schlesinger Ritter and Burg20 prepared BN-dimethylborazole (IX) BB'N-trimethylborazole (X) and BB'B"N-tetramethylborazole (XI). Each appeared to be a single compound although there are two possible structures of both (IX) and (X). It is not known which of the possible isomers were prepared and the problem of position isomerism has not been investigated since 1938. SHELDON AND SMITH THE BORAZOLES 21 5 Molecular Structure.-Borazole was shown to have a benzene-like configuration by Stock and Wierl,2 and the small dipole moment observed by Rama~wamy~~ was presumably caused by the presence of impurities. The structure was confirmed by BauerSS and structural data for hexagonal boron nitride borazole N-trimethylborazole and B-tri- chloroborazole appear in Table 2.The positive inductive effect of N-methyl and the negative inductive effect of B-chloro-groups both lead to greater double-bonding between boron and nitrogen in the borazole ring. This increase in aromatic character is reflected by shorter boron-nitrogen interatomic distances in N-trimethylborazole and B-trichloroborazole. TABLE 2 Compound B-N (A) Substituents Method Ref. 0” 1 -446 A’-Ray diffrn. 49 B3HsNsHs 1-44 f 0 . 0 2 Electron diffrn. 59 BsH3N3Mes 1-42 f 0.02 N-C = 1-48 f 0.03 A , 9 60 B3Cl3N3HS 1-41 f 0.02 B-Cl = 1.78 f 0.03 A 60 1.413 f 0.010 B-Cl = 1.760 =t 0.015 A &Ray diffr;;. 61 1-415 B-Cl = 1.735 8 9 9 62 It is difficult to obtain reliable values for the lengths of pure single and double boron-nitrogen bonds. If the trigonal covalent radii of boron and nitrogen63 are taken to be 0.8 A and 0.74 A (trigonal being assumed to be similar to tetrahedral nitrogen) and the appropriate allowance 0-09 A is made for the electronegativity differen~e,~~ the estimated boron-nitrogen single bond length is 1-45 A.This value suggests that the bonds in hexa- gonal boron nitride are approximately single bonds which seems plausible in view of the differences in structure electrical properties and colour between boron nitride and graphite. Hexagonal layers occur in graphite but the electrical conductivity and characteristic black lustre have been attributed to the aromatic character of the carbon-carbon bonds. The most acceptable boron-nitrogen double-bond value is probably 1.21 8 A found in the short-lived BN species.65 Application of Gordy’s formula,66 n = aR-2 + b where n is bond order R is bond length and a and b are constants characteristic of any pair of atoms suggests that the boron-nitrogen bonds in B-trichloroborazole have about 20 % of double- bond character; but numerical values of bond order should be accepted 68 Ramaswamy Proc.Indian Acad. Sci. 1935 2 A 364 630. K g Bauer J. Amer. Chem. SOC. 1938 60 524. 8o Coffi and Bauer J. Phys. Chem. 1955,59 193. Coursen and Hoard J. Amer. Chem. SOC. 1952,74 1742. 62 Lonsdale Nature 1959 184 1060. 63 Wells “Structural Inorganic Chemistry,’’ Oxford University Press 2nd Edition 64 Schomaker and Stevenson J. Amer. Chem. SOC. 1941,63 37. 65 Douglas and Hertzberg Canad. J. Res. 1940 A 18 179. 66 Gordy J. Chem. Phys. 1947 15 305. 1950 p. 58 347.216 QUARTERLY REVIEWS with caution. Molecular diamagnetic anisotropy may be taken as a criterion of aromatic character and a comparison of the anisotropies of variously substituted borazoles would be of great interest. Although only one compound B-trichloroborazole has been st~died,~' the approximate value d K = 18 x c.g.s. units at least confirms that B-trichloroborazole has some aromatic character. Coursen and Hoards1 showed that molecules of crystalline B-trichloro- borazole are arranged in layers and the short N-H - . Cl distances between adjacent molecules suggested some type of electrostatic attraction. Lonsdales2 provided further evidence that the hydrogen and chlorine atoms in B-trichloroborazole are oppositely charged by a comparison with the similar crystal structure of 1,3,5-trichlorobenzene.Although intermolecular H * * - C1 distances are less in B-trichloroborazole than in 1,3,5-trichlorobenzene the intermolecular H - * H and C1 - C1 distances are greater in B-trichloroborazole. The short B-N and long B-C1 bond lengths lend support to structure (XIIa) rather than structure (XIIb) for B-trichloroborazole. In PhBCl, where double bonding between boron and chlorine is generally assumed to occur the B-C1 bond lengthss0 are 1.72 A. It is significant that B-trichloroborazole is thermally unstable and evolves hydrogen chloride even at room temperature,68 since this would be characteristic of structure (XIIa) rather than (XIIb). s+lq (4 (b) (Arrows indicate direction of the inductive effect.) Infrared and Raman Spectra.-The Raman spectrum of liquid borazole and the infrared absorption of borazole vapour from 400 to 7000 cm.-l have been described in detail.69 Fundamental frequencies assigned on the basis of DZh symmetry gave a satisfactory correlation with the fundamental frequencies of benzene and force constants were calculated for borazole.The allocation of certain spectral bands in borazole and N-trimethylborazole is shown in Table 3. The natural abundance ratio l1B.l0B is approximately 4:l but identification of the splitting caused by the isotope effect was complicated by the large number of allowed combination^.^^ The intensity of the 1465 cm.-l band in borazole is attributed to charges on the boron and nitrogen atoms and the effect of ring substituents is 67 Lonsdale and Toor Acta Cryst. 1959 12 1048 e8 Hohnstedt and Haworth Chem.Eng. News Sept. 16th 1957 67. 6 g Crawford and Edsall J. Chem. Phys. 1939 7 223; Price Fraser Robinson and Longuet-Higgins Discuss. Faraday Soc. 1950 9 131; Spurr and Chang J. Chem. Phys. 1951 19 518. SHELDON AND SMITH THE BORAZOLES 217 illustrated by Table 4. This characteristic band occurs at 1492 cm.-l in the spectra of B-aminobora~oles,~~* 36 and L a ~ p e r t ~ ~ also considered absorption near 720 cm.-l to be characteristic of B-aminoborazoles and B-alkoxyborazoles TABLE 3 €3 3H 3N3H3 €3 3H3N3Me 3 Y (cm.-l) v (cm.-l) 718 918 1465 2530 3490 6880 1056 BH NH In-plane bending 1107 1425 In-pla;; ring defGmation (E’) 1342 9 3 3 1294 248 5 BH St&ching ” NH Stretching NH Stretching overtone 1 TABLE 4 Compound v (cm.-l) Ref. Compound Y (cm.-l) Ref. Compound Y (cm.-l) Ref 1465 69 B3H3N3Me3 1425 69 B3H3N3Ph3 1401 55 1451* 70 B3Et3N3Me3 1404 30 B3Me3N3Ph3 1380 55 1472 55 B3Ph3N3Me3 1405 55 1445 55 B3C13N3Me3 1392 55 B3C13N3Ph3 1373 55 * Raman frequency.Different assignment from that originally proposed. Ultraviolet Spectra.-The ultraviolet spectrum of borazole has been related to that of benzene by Platt Schaeffer and their co-w~rkers,~~ and the observed spectra were compared with computed energies in a semi-empirical molecular-orbital treatment of the ground states and first excited states by Roothaan and M ~ l l i k e n . ~ ~ The 1700 A band in the borazole spectrum is analogous to the 1800 8 band in the benzene spectrum and the intensity is characteristic of Vt-N transitions. Diffuse bands which occur near 1950 8 in the spectrum of borazole are weaker than the 2000 A bands in benzene but stronger than the 2600 A bands in benzene.Rector Schaeffer and Platt73 also examined three substituted borazoles. B-Trimethyl- and B-trichloro-borazole absorb at similar wavelengths to borazole but N-trimethylborazole absorbs at longer wavelengths and the strongest band occurs at 1900 A. The relative positions of these spectra were believed to indicate the degree of aromatic character and structure (XIIb) was accordingly proposed for B-tri- chloroborazole. However the evidence of bond lengths shows that 70 Goubeau and Keller Z. anorg. Chem. 1953,272 303. 71 Platt Klevens and G. W. Schaeffer J. Chem. Phys. 1947 15 598 Jacobs Platt 72 Roothaan and Mulliken J. Chem. Phys. 1948 16 118. 73 Rector G. W. Schaeffer and Platt J. Chem.Phys. 1949 17,460. and G. W. Schaeffer ibid. 1948 16 116. 218 QUARTERLY REVIEWS B-trichloroborazole has more aromatic character than borazole and an alternative explanation of the spectral data is offered. The replacement of N-hydrogen by methyl as in N-trimethylborazole produces two separate effects. (a) The electron density of the borazole ring is increased and this causes a red shift in the spectrum. (b) The donor properties of nitreen are increased this leads to a greater degree of aromatic character and causes a further red shift. When methyl replaces B-hydrogen as in B-trimethylborazole the two effects are opposed. (a) The increased electron density of the ring would cause a red shift. (b) The acceptor properties of boron are reduced. This leads to reduced aromatic character and would cause a blue shift.Chlorine has a negative inductive effect and in B-trichloroborazole (a) the electron density of the ring is reduced (blue shift) and (b) the acceptor properties of boron are increased and hence the aromatic character of the ring is increased (red shift). These opposing influences may again be roughly equivalent which would account for the similar spectra of B-trichloroborazole and borazole. Steric hindrance. The ultraviolet spectra of four symmetrical triphenylborazoles were investigated by Becher and F r i ~ k . ~ ~ Borazole71 does not absorb between 2000 and 4600 A and bands observed near 2600 A in the spectra of B-trimethyl-N-triphenylborazole and N-trimethyf-B-triphenylborazole are due to unconjugated phenyl groups. Methyl attached to the borazole ring prevents the phenyl groups from achieving co-planarity.Red shifts of increased intensity occurring in the spectra of B- and N-triphenylborazole confirm that there is reduced steric hindrance and greater conjugation in the absence of methyl groups. Thermodynamic Properties.-Crawford and Edsalleg calculated the entropy free energy and enthalpy of gaseous borazole from spectral measurements and used vapour-pressure data to obtain thermodynamic values for the liquid. A heat-capacity equation for gaseous b o r a z ~ l e ~ ~ was derived from these values. The heat of hydrolysis of B-trichloroborazole measured by Van Artsdalen and D w ~ r k i n ~ ~ was 113.8 -+ 0.07 kcal./mole at 25” c. The heat of formation of crystalline B-trichloroborazole and the boron- nitrogen bond energy were calculated from the observed heat of hydrolysis and from the heats of formation of water boric acid and ammonium chloride.A recalculation by C ~ t t r e l l ~ ~ who used more recent thermo- chemical values gave 106.5 kcal. for the boron-nitrogen bond energy in B-trichloroborazole. This was compared with 104 kcal. for the boron-nitrogen bond energy in trisdimethylaminoborine recalculated by 74 Becher and Frick 2. phys. Chern. (Frankfurt) 1957,12 241. 75 Spencer J. Amer. Chern. SOC. 1945 67 1859. ‘13 Van Artsdalen and Dworkin J. Amer. Chem. SOC. 1952 74 3401. ” CottrelI “The Strengths of Chemical Bonds,” Butterworths Scientific Publications London 2nd Edition 1958 p. 240. SHELDON AND SMITH THE BORAZOLES 219 Cottrell from the heat of hydrolysis measured by Skinner and Smith.78 Tri.sdimethylaminoborine is a planar in which some double bonding between boron and nitrogen is also believed to occur. One of the Reviewers (J.C.S.) acknowledges the award of an I.C.I. 78 Skinner and Smith J. 1953,4025. 7 0 Becher 2. anorg. Chem. 1956,287,285. Fellowship by the University of London.

ISSN:0009-2681    

DOI:10.1039/QR9601400200

出版商:RSC    

年代:1960

数据来源: RSC