摘要:

Chemical Society Reviews Vol 7 No 3 1978 Page LIVERSIDGE LECTURE The Surface of a Liquid By J. S. Rowlinson 329 Meldrum’s Acid By Hamish McNab 345 Bond Valences-A Simple Structural Model for Inorganic Chemistry By I. D. Brown 359 Monoalky1triazenes By K. Vaughan and M. F. G. Stevens 377 Individual CH Bond Strengths in Simple Organic Compounds:Effects of Conformation and Substitution By D. C. McKean 399 The Chemical Society London Chemical Society Reviews Chemical Society Reviews appears quarterly and comprises approximately 25 articles (ca. 500 pp) per annum. It is intended that each review article shall be of interest to chemists in general, and not merely to those with a specialist interest in the subject under review. The articles range over the whole of chemistry and its interfaces with other disciplines.Although the majority of articles are intended to be specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be sub- mitted to The Managing Editor, Books and Reviews Section, The Chemical Society, Burlington House, Piccadilly, London, W 1V OBN. Members of the Chemical Society may subscribe to Chemical Society Reviews at E6.00 per annum; they should place their orders on their Annual Subscrip- tion renewal forms in the usual way. Non-members may order Chemical Society Reviews for E16.00 ($33) per annum (remittance with order) from: The Publications Sales Officer, The Chemical Society, Distribution Centre, Blackhorse Road, Letchworth, Herts., SG6 lHN, England. 0 Copyright reserved by The Chemical Society 1978 Published by The Chemical Society, Burlington House, London, W1V OBN Printed in England by Eyre & Spottiswoode Ltd, Thanet Press, Margate

ISSN:0306-0012    

DOI:10.1039/CS97807FP007

出版商:RSC    

年代:1978

数据来源: RSC

摘要:

LIVERSXDGE LECTURE* The Surface of a Liquid By J. S. Rowlinson PHYSICAL CHEMISTRY LABORATORY, SOUTH PARKS ROAD, OXFORD, OX1 3QZ 1 Introduction Until about ten years ago it was common to read in books and reviews that little was understood about the structure and properties of liquids, in contrast with our detailed understanding of gases and solids. The position has now changed dramatically; quantitatively acceptable theories of the structure and equilibrium properties of liquids have been developed and tested exhaustive1y.l There are still problems of relating structure to intermolecular forces in liquids composed of polar, flexible, or ionized molecules, and in understanding the dynamic properties, but we now understand well the relation between structure and inter- molecular forces for liquids and liquid mixtures composed of simple moiecules.Our understanding of the molecular structure of the gas-liquid surface is, how- ever, still at the primitive state of our understanding of bulk liquids fifteen years ago. There are several reasons for this. First, the diffraction and scattering experiments that have proved so powerful in determining the structure of bulk matter and of the gas-solid surface have been much less effective for probing the structure of the gas-liquid surface. Visible light is reflected from the surface and can reveal structure on a scale down to about 10-5 cm, but this is still in the continuum or hydrodynamic regime-nothing molecular can be seen on such a scale.The technique of ellipsometry can in principle extend this range to 10-6 cm, or even lower for the gas-solid surface,2 but this is still a large distance when compared with the thickness of the surface layer separating a vapour from its iiquid near its triple point, which is generally about 10-S-10-7 cm. Optical reflectance is useful only at higher temperatures, close to the critical point, where the thickness of the surface layer grows to 10-5 cm and bey~nd.~ Shorter wavelengths, X-rays or neutrons, have been little used, since it is difficult to distinguish between the scattering from the surface and that from the bulk. If, however, the refractive index of the liquid is less than unity then the rays *Delivered at Imperial College, London on 16th February 1978, and on other days in Aston, Leeds, Oxford, Loughborough, Hull and Aberdeen. For recent reviews see J.P. Hansen and I. R. McDonald, ‘Theory of Simple Liquids’, Academic Press, London, 1976; J. A. Barker and D. Henderson, Rev. Mod. Phys., 1976,48, 587.* For the study of surface films by ellipsometry see Sj~mposioFuraday Soc., 1970, No. 4. J. S. Huang and W. W. Webb, P/ij.s.Rev.Lrttcrs, 1969, 23, 160; E. S. Wu and W. W. Webb, Ph>.s.Rv. (A), 1973, 8, 2065. The Sirrface of a Liquid that are totally reflected are on the gas side of the surface not, as is usual, on the liquid side. Rice and Lu4 are exploiting this fact to study the total reflection of X-rays from a clean mercury surface. The refractive index is only a little less than unity (H z 1-10-6) and so the angle for total reflection is close to the grazing angle.The experiment is difficult, but is a promising way of determining surface profiles when the refractive index permits. The second difficulty in determining the molecular structure of the gas-liquid surface is that of developing a molecular theory-t hat is, a self-contained theory which allows us to deduce the structure from a knowledge of the intermolecular forces.5 The statistical mechanics of inhomogeneous systems is a difficult and, until recently, little studied branch of theory. There are two ways of making progress when faced with such experimental and theoretical difficulties. As a substitute for experiment we can turn to computer simulation, and as a substitute for the statistical theory of realistic systems we can turn to the study of model systems that are theoretically tractable but physically less realistic than we would wish.This review describes some recent work along both these lines. 2 Computer Simulation of the Surface There are two kinds of computer simulation, conventionally called Monte Carlo (MC) and molecular dynamic (MD) simulation,6 and in each we model a small portion of the system of interest by establishing within the computer store a set of co-ordinates to describe the positions of (usually) between 100 and 1000 molecules, and a specification of the intermolecular energy between each pair. In a MC simulation we specify the temperature, and then move the molecules in such a way that each configuration appears with a probability proportional to its Boltzmann factor, exp (-Uconfig/kT), where Uconfig is the sum of all the inter- molecular energies.Averages over many such configurations give the equilibrium but not the dynamic properties of the system. In a MD simulation, one gives each molecule a randomly chosen velocity and then solves Newton’s equations of motion to find out how the system evolves in time. After a system of fixed initial energy has come to equilibrium the average temperature is steady, but subject to statistical fluctuations, and equilibrium properties can be found from an average over time. Here dynamic properties can also be calculated since the successive states of the system are generated in their correct time sequence.It is only recently that these techniques have been applied to the gas-liquid surface. The main difficulty is in devising constraints that maintain a plane surface fixed in ‘laboratory’ (i.e. computer) co-ordinates without distorting the molecular structure of the surface. In the real world a weak gravitational field, acting on a system large compared with the range of the intermolecular forces, gives this stability. This will not do for a system of lo00 molecules, since over a S. A. Rice and B.-C. Lu, personal communication. For a recent review of statistical theories of the gas-liquid surface see S. Toxvaerd, ‘Statistical Mechanics’, ed. K. Singer (Specialist Periodical Reports), The Chemical Society, London, 1975, Vol.2, p. 256. I. R. McDonald and K. Singer, Quart. Rev., 1970, 24, 238. Rowlinson distance of about 10 molecular diameters the change in the earth's gravitational potential is only about 10-6 of the intermolecular energy, and so without effect. To use a much more powerful gravitational potential would risk distorting the delicate molecular structure of the surface. The problem can be overcome to a large degree by using a device that is, in any case, necessary on other grounds. We assume that the rectangular prism that contains the molecules is surrounded on each of its four vertical sides by an infinite set of replicas of itself. Each molecular x-and y-co-ordinate is thus repeated in each cell, but the molecules can pass freely through these vertical walls.In this way a molecule leaving the cell through, say, the right-hand wall is deemed to re-enter through the corre-sponding point on the left-hand wall; the fluid can be regarded as having an infinite extent in the x-and y-directions. Once a planar surface has been estab-lished, the repetition of the x-and y-co-ordinates ensure that it remains planar on the square grid formed where this plane cuts the vertical walls. Most of our work has been carried out in a cell 5o x 50 in the x-and y-dimensions and 250 in vertical or z-direction, where o is the molecular diameter (Figure 1). The liquid phase is anchored to the bottom half of the cell by placing \ 9 Z ........... .................. ........... ... ............... ............. .......... .......... ...... .... ......... ......... ........ .... ................ ................ ......................... 2 c X7\ $ @Figure 1 Sketch of the cell usedfor a computer simulation. The gas is in the upper half and the liquid is anchored to the lower half by the intermolecular forces emanating from a simulated homogeneous liquid phase below the bottom wall of the cell (shown shaded). The top and bottom walls are reflecting boundaries, but the molecules pass freely through the four vertical walls The Surface of a Liquid beneath the bottom wall a crude representation of the bulk liquid which interacts with the molecules in the cell through an averaged intermolecular potential.' The top and bottom walls are impenetrable to the molecules and this constraint induces a distortion of the local fluid densities near these boundaries, z = 0 and z = 250; that at the bottom is clearly visible in the graphs shown below, but is unimportant if we are interested only in the fluid near the gas-liquid surface at about z = 130.Other workerss have used similar constraints, but Barker and his colleaguesg rely solely on the repetition of the x-and y-co-ordinates to keep a two-sided semi- infinite slab of liquid near the centre of the cell, with two planar surfaces, and gas in the bottom and top thirds of the cell. This arrangement avoids the distortion of impenetrable boundaries but allows the slab of liquid to move as a whole in the z-direction.Almost all work up to n0w7-11 has used a Lennard-Jones intermolecular potential u(r) = 44(u/r)12 -(~/r)~] (1) where u(o)= 0, and -E is value of u at its minimum. (The potential is sometimes put equal to zero beyond Y = 2.50 or 30 for computational convenience). This is, in fact, a simple effective potential for the liquid inert gases, but an assembly of Lennard-Jones molecules is properly regarded as a realistic model, rather than as the best representation of a real system. The properties of the bulk gas and liquid phases of this model fluid are now well established by computer simulation, and we show below how this technique can be used to study the properties of the gas- liquid surface 3 The Structure and Properties of the Liquid Surface A.The Surface Profile.-The density of a vapour near its triple point is generally only or less of that of the liquid in equilibrium with it. This great change is believed to take place over a distance comparable with the size of a molecule or with the effective range of its intermolecular potential. The difference in density decreases at high temperatures, vanishing at the critical point, and the thickness of the interface increases, becoming infinite at the critical point. The first question of interest is therefore, what is the function p(z) which represents this change of. density, p = N03/Y with height, z. Figure 2 shows the computed surface profiles for three reduced temperatures r = kT/E for a Lennard-Jones fluid, whose triple point is at r = 0.68 and whose critical point at r = 1.34.As the ' G. A. Chapela, G. Saville, and J. S. Rowlinson, Faraday Discuss. Chem. SOC.,1975, No. 59, p. 22.* A. C. L. Optiz, Phys. Letters (A), 1974, 47, 439; K. S. Liu, J. Chen.1.Phys., 1974, 60, 4226. F. F. Abraham, D. E. Schreiber, and J. A. Barker, J. Chem. Phys., 1975,62, 1958. lo M. Rao and D. Levesque, J. Chem. Phvs., 1976, 65, 3233. l1 M. H. Kalos, J. K. Percus, and M. Rao, J. Stat. Phys., 1977, 17, 11 1. Rowl’inson A Figure 2 Three density profiles. The local density p is shown as a function of the height z for three reduced temperatures, r = 0.701, 0.918, and 1.127. The oscillations for z < 40 are artefacts induced by reflection of molecules at the bottom wall, z = 0 temperature rises the liquid density falls, the gas density rises, and the thickness increases. The last is conventionally defined by d = (PI -Pdl P’(z*)Il-’ (2) where z* is the height at which the profile is steepest and Ip’(z*)l the absolute value of the slope at that point.The thickness calculated from these runs increases from 1.50 just above the triple point (T = 0.70) to 2.20 at T = 0.92 and 3.10 at T = 1.13. These results agree with our expectations, but when we turn to the effect of the size of the system, in the x-and y-directions, on the shape of the profile p(z), then the position is not so happy. It is our hope that, as the area d of the system increases indefinitely, such functions as p(z) quickly settle down to a steady value that is independent of d.Such limits undoubtedly exist for the properties of homogeneous fluids, although their existence is often difficult to prove, and the steady values of the functions are said to be those of the thermodynamic limit.To study the existence of such a limit for p(z) we used not only cells of 50 x 50 in cross-section (containing 255 molecules, see Figures 1 and 2), but also 100 x 100 (1020 molecules) and 200 x 200 (4080 molecules). The computing runs12 take much longer with these large samples and only sparse information could be obtained. The thickness (Figure 3) is, however, found to increase with dat all temperatures. It may be approaching a finite limit, or it may be diverging weakly to infinity-the evidence is inconclusive-but at least no limit has been reached at an area of 4000~.There are recent theoretical arguments which suggest l2 G.A.Chapela, G.Saville, S. M. Thompson, and J. S. Rowlinson, J.C.S.Furaduy ZZ, 1977, 73, 1133. 333 The Surfuce of a Liquid 3.0 b\ b 2.0 -1.0 I 1 I 1 I 1 0.8 1.0 1.2 Figure 3 The thickness of the surfNce, as defined in equation (2), as a function of the reduced temperature T,for three diflerent sirrface areas: 0,d = 250~;A,d = 1000~;0,d = 40002 that the correct result is a slow divergence.11913 Even a weak gravitational field probably suppresses the divergence, which arises from the build-up of long but feeble capillary waves, so if we are to look for the divergence experimentally it will have to be on the surface of a large sphere of liquid in a space laboratory.The existence of such a divergence is therefore of little practical importance, but its theoretical significance is considerable. If we cannot go to the thermo- dynamic limit, without applying a gravitational field, then the development of a rigorous statistical mechanics of the gas-liquid surface is going to be much more difficult than might be expected. Indeed the status of p(z) calculated for theoretical models (see e.g. Section 4) is called in question. This is not a problem that can be l3 M. S. Wertheim, J. Chem. Phys., 1976, 65, 2377; J. D. Weeks, ibid., 1977, 67, 3106; H. T. Davis, ibid., 1977, 67, 3636.334 Rowlinson pursued here except to remark that the calculated profiles may represent an average local profile at a fixed time, in which the long capillary fluctuations are ignored. B. Temperature, Pressure, and Chemical Potential.-Once the profile p(z) has been determined it is natural to ask about the behaviour in the surface of the tempera- ture, pressure, and chemical potential, i.e. about the functions, T(t),p(z), and p(z). These each have the same value in the two bulk phases, for this is a con- dition of equilibrium. How do they behave in the surface region where p(z) is a rapidly changing function of z? Before this question can be answered we need an operationally meaningful definition of the local values of these functions.A local temperature can be defined in terms of a local density of kinetic energy. In a classical, but not in a quanta1 system, the kinetic energy of the monatomic molecules is strictly separable from all other energies and so its local density can be used to define a local temperature. Figure 4 shows that molecules in the 2 Vh Figure 4 The distribution function of molecular speeds, n(v), in arbitrary units, for molecules in the surface layer (T = 0.708).The segmented curve is the result of the computer simulation and the smooth curve is the Maxwell-Boltzmann function for this temperature The Surface of a Liquid 0.8 Figure 5 Temperature and chetnicalpotential in the surface layer. The two functions plotted are the ratios T(z)/Tl and p(z)/pgwhere T(z) and p(z) are the local values of the tewrgerature and the configurational cheniical potential.The statistics of the computation are such that TI and pg are relatively easy to estimate but T, and 111 are dificult. The dashed curve in the lower part of the diagram is a sketch of p(z) from Figure 2, which serves to show the position of the gas-liquid surface surface layer have a Maxwell-Boltzmann distribution of speeds, and Figure 5 shows that T(z)is, within the statistical error, a constant independent of z. A local pressure can be defined, although not entirely without ambiguity.14 in terms of the local strength of the intermolecular forces. It is not a constant scalar quantity, independent of z, although this is the expected behaviour of its component normal to the surface; px(z) = pg = pl ES p.The transverse com- ponent of the pressure tensor ~T(z)is not a constant, but reduces to p in the homogeneous bulk phases. It is the integrated difference between these two components that is measured by the surface tension [The value of the integral is independent of any local ambiguity in defining pT(Z)]. For a computer calculation this equation is more conveniently re-written in a form, due to Kirkwood and Buff,15 to which we turn in the next Section. The constancy of the chemical potential is a more subtle question. Gibbs16 asserted that p(z) was a constant, equal to ,ugand PI, but did not define exactly what he meant by ,u(z).The question has been debated many times since in classical thermodynamic terms,I7 but it is one which can be resolved only in l4 A.Harasima, Adv. Chem. Phj,s., 1958, 1, 203. J. G. Kirkwood and F. P. Buff, J. Cheni. Phys., 1949, 17, 338. l6 J. W. Gibbs, 'Collected Works', Longmans, New York, 1928, Vol. I, p. 219. J. C. Eriksson, Adv. Chem. Phj,s., 1964, 6, 145. Rowlinson terms of statistical mechanics. This was not done until 1963 when there was derived1* what is now usually called the potential distribution theorem. This allows us, by means of a thought-experiment, to measure chemical potential at a point by repeatedly adding an additional or test molecule at that point, observing the change in the Boltzmann factor, and then withdrawing the test molecule. The experiment is one more easily made on the computer than in the laboratory, but even there it is statistically difficult to get a good result.Figure 5 shows that p(z) is, approximately, independent of z; to be more precise our sampling technique shows a small mismatch between pugand p1, and that p(z) in the surface layer is, at least, not behaving in some quite unforeseen way. We return to this point in the discussion of a model system in Section 4. C. The Calculation of Surface Tension.-The mechanical definition of surface tension, equation (3), can be transformed into one relating the tension to a lack of balance of the intermolecular forces in the surface layer, where the sum is taken over all molecular pairs, i and j, and where this pair is separated by rij of which the vertical component is zij.The angle brackets denote an average over a system at equilibrium. By symmetry, (rij2 -3zij2j vanishes in a homogeneous fluid so that the whole of the effective contribution to the sum in equation (4) comes from the surface layer. Figure 6 shows the results obtained by our and other computer simulations, and the line is the result of a perturbation theory developed by T~xvaerd.~ This calculation assumes a particular functional form of p(z) and so is not an a priori calculation, but it is seen to be an accurate one. The possible conclusion is that the surface tension, although experimentally important, is not very sensitive to the precise form of p(z) or to the molecular arrangements that give rise to it.The surface tension vanishes at the critical point, but computer simulation cannot be used when the thickness of the surface becomes as large as the x-or y-dimensions of the cell. D. Molecular Diffusion in the Surface Layer.-The properties of the surface layer discussed so far have all been equilibrium properties. Molecular dynamic simulation can also tell us of the dynamic or transport properties of a system, and the simplest of these is diffusion. In the laboratory we can measure a coefficient of self-diffusion if we can label a molecule isotopically in a way that has no effect on any relevant physical property. In a computer we can achieve the same end either directly, by keeping a record of each molecule's identity and position, or indirectly by studying the persistence of molecular velocities; the latter is technically the easier.B. Widorn, J. Chem. Phys., 1963, 39, 2808; J. L. Jackson and L. S. Klein, Phys. Fluids, 1964, 7, 228. The Surface of a Liquid 0 \ \ \ \ \ \ I I I I I I \ I 0.8 1.o 1.2 1.4 Figure 6 The reduced surface tension, ya2/c,as a function of reduced temperature 7. 0,ref. 12; A,ref. 9; 0,ref. 8. The surface tension vanishes at the critical temperature, TC = 1.34, but computer simulation cannot be used near this temperature Let v(0)be the velocity of a molecule at a time, taken to be zero, and v(t)its velocity at time t later. The average value of the scalar product of these two velocities (v(O).v(t))is a measure of the persistence of velocities, for this function of time starts at ([~(o)]~)= 3kT/m, and goes to zero as t becomes infinite.At very long times the average square of the speed of a molecule will still be of the order of 3kT/m, but the direction of its velocity will be quite Rowlinson uncorrelated with that at t = 0; hence the average value of the scalar product is zero. Einstein showed that the coefficient of diffusion is determined by an integral over all time of this velocity auto-correlation function,lg which we denote $(t): ICl(t> = <m.v(t))/< 40)2> (5) In a dense homogeneous fluid the correlation function falls from #(O) = 1 to zero at a time of about 10-13 s, and is small and negative for longer times, becoming indistinguishable from zero after about s.The negative region arises from the probable reversal of the direction of motion of a molecule in a liquid after it has moved for a distance comparable with the mean molecular spacing. This result is confirmed in our computer simulations20 for those molecules which start at r = 0 in the middle of the liquid zone, i.e. at a height of around z = 7a. If, however, we take the average in (5) only over those molecules which start in the surface layer, 12.4~< z < 12.60, then we obtain different preliminary resylts. It is instructive to analyse separately the motions in the z-direction and in the x-and y-directions. Neither fiZ(t)nor #L(t)= qhy(t)has a negative region, and the latter has a particularly pronounced long positive tail. The area under the curve for z,hL or #y is greater than that under t,hZ.We deduce that molecular motion parallel to the surface shows a greater persistence of velocity than that normal to the surface and so also it must have a greater coefficient of diffusion.Molecules moving along the surface seem almost to 'skate' without experiencing the hindrances of those molecules moving in or out of the liquid. E. Adsorption.-In a mixture of two components, a and b, the surface structure can be defined by means of the two functions pa(Z) and pb(Z). These will generally be different functions since we know16 that the component of lower surface tension is present in excess in the surface layer. A measure of this excess is provided by the quantity r where r = xa'rb -xblra.(7) and I'a and I'b are the numbers of molecules of each component per unit area of the surface phase.21 Gibbs showed by thermodynamic arguments that r is proportional to the rate of isothermal change of surface tension with com- position of the liquid phase.16 If the liquid mixture is ideal, then (ay/Zxb')T = -r(kT/xa'Xb') (8) It is difficult to test this equation since r is not easily measured. The classic method of McBain was to use a microtome to slice off the surface layer, and l9 J. S. Rowlinson and M. Evans, Ann. Reports (A), 1975, 72, 5. 2o S. M. Thompson, G. Saville, and J. S. Rowlinson, unpublished results. 21 A. W. Adamson, 'Physical Chemistry of Surfaces', 3rd Edn., Wiley, New York, 1976, Section IT, p.6. The Surface of a Liquid then to analgse it.21 More accurate is the use of p-active tracers whose emission has so short a range that the activity of the surface layer can be measured without too much contamination from the bulk liquid.21 Computer simulation allows a third method of test with the added bonus of telling us not only how much is adsorbed, but exactly where it is adsorbed; that is, we can study p,(z) and pb(Z), and not merely integrals of these functions. The system chosen for simulation12 is one of two Lennard-Jones molecules with E parameters chosen to represent rather crudely either argon (substance a) or krypton (substance b). The results (Figure 7) show that the lighter component 0.4 0.2 I I 1 I 1 1 I , L‘-L! ,6 8 10 12 14 16 z/(T Figure 7 The surface profile for a binary mixture.The full ciirve, a, is the cotnponent of lower surface tension (burgon’) mid the dashed curIie, b, that of higher surface tension (‘krypton’) is adsorbed at the surface, and that the form of this adsorption is a monatomic layer of almost pure a; that is the profile pa(z) extends into the gas phase by about one more diameter u than the profile pb(Z). Figure 8 shows that the integrated amount of this excess is consistent, within the rather large errors, with the excess calculated from Gibbs’s equation (8). The agreement is about as good as can be obtained by the microtome method, but rather less than the best P-isotope experiments, which are, however, restricted to surface-active materials.4 The Penetrable Sphere Model There is a long tradition in physics of devising models to represent phase transitions which are constructed in such a way that the theoretical difficulties are less than they would be for a real system. The classic case is the Ising model of ferromagnetism, and its vast literature is a measure of its value in studying the statistical mechanics of phase transitions. The penetrable sphere mode122 is particularly suitable for studying the gas- liquid transition, and the structure of the surface phase, since it is one in which 22 B. Widom and J. S. Rowlinson, J. Chem. Phys., 1970, 52, 1670. 340 Rowlinson // 5 0 0 0 0 /0 0 0 0 0 1.0 .o/ 0.5 0 Figure8 Reducedsurfuce tension as a function of composition in a binary mixture.The circles are for a reduced tempernture of ra8 = 0.701, and for a computer simulation by the Monte 788Carlo method. The triangles are for a reduced temperature of = 0.713, and for a Molecular Dynamic simulation. The straight lines are the slopes calculated from the adsorption bv means of the Gibbs equation the molecules can move freely through space without restriction to lattice sites or cells (as in the king model). It is described by specifying the energy of inter-action of a set of N molecules. Around each molecule draw a sphere of volume VO, and allow the molecules to occupy any set of positions r1 . . . rAvin a volume V. Let W(r1, . .rN) be the total volume covered by the Nfreely interpenetrating spheres (Figure 9) and let the energy of this configuration of the molecules be U(r1 . . . r,~)where U(r, . . . YN) = [W(Y, . . * YN) -NV,](€/V,) (9) where E is a characteristic energy. Clearly if no spheres overlap W = Nvo, and U is zero. As the molecules become more closely packed overlapping will occur and W < Nvo, so that U < 0. Hence the intermolecular energy is either zero or negative; that is, it is an ‘attractive’ intermolecular potential. At sufficiently low values of kT/E such a system can be shown to form two phases, a gas and a The Surface of a Liquid Figure 9 A typical cluster of N = 11 tnolecules in the penetrable sphere niodel. The shaded volume (area in this two-dimensional sketch) is that denoted W(r,.. . rly) in equation (9),and the energy U(r, . . . rN) is proportional to the diflerence between this volwne and that of the eleven separate spheres, Nv, liquid. The critical temperature is not known exactly but is of the order of kTc z E. The intermolecular potential is quite unrealistic, but it has the two virtues of leading to a realistic phase transition, and, because of a hidden symmetry analogous to the spin-up-spin-down symmetry of the Ising model, of leading to tractable equations for p(z) in the gas-liquid surface. In the simplest case this function is found to be a hyperbolic tangent.23 The successful calculation of p(z) allows us to use the model as a test-bed on which to try out ideas which seem to be plausible but which cannot be tested rigorously for more realistic systems.These have included : (i) An explicit demonstration that the local chemical potential &), as defined by means of the potential distribution theorem of Section 3B, is indeed a con- stant,23 equal to pgand and independent of z. (ii) A demonstration that the concept of a local Helmholtz free-energy density, a concept which has played an important role in theories of surface tension since the time of van der Waals, is useful near the critical temperature, but leads to inconsistencies at lower temperature~.~s?~~ (iii) A demonwation that three apparently quite different theoretical routes to the calculation of the surface tension [one of which is equation (4) above] do, in fact, all lead to the same result.25 This conclusion is well worth having since there had been some doubt of their eq~ivalence,~ and although a proof for a model is not a proof in general, it goes a long way to setting such doubt at rest.23 C. A. Leng, J. S. Rowlinson, and S. M. Thompson, Proc. Roy. Soc., 1976, A352, 1. 24 J. M. Harrington and J. S. Rowlinson, unpublished results. 26 C. A. Leng, J. S. Rowlinson, and S. M. Thompson, Proc. Roy SOC.,1978, A358, 267. Rowlinson A mixture of penetrable sphere molecules is also a tractable mode126 and preliminary results27 indicate that it will provide a possible theoretical test of Gibbs’s adsorption equation at a molecular level. Such results are an indication of the value of model systems in supplementing the few explicit theoretical results for realistic systems, and for testing the validity of theoretical ideas.5 Conclusion This review has not been a balanced review of recent work on the gas-liquid surface, but has had the aim of showing how two particular techniques can be used to increase our understanding of a surface at a molecular level. The first of these is computer simulation, which is justified when (and perhaps only when) the equivalent laboratory experiments are too difficult to carry out. The second is the use of a simple model system to answer questions on the statistical mech- anics of a surface that are too difficult to be answered for more realistic repre- sentations of an actual surface. There is clearly much more work to be done along both lines. This research could not have been carried out without the collaboration of those named in the references, and in particular without the computing skill of my former colleague Dr. Graham Saville of Imperial College. To these I owe my thanks. 26 N. Desrosiers, M. I. Guerrero, J. S. Rowlinson, and D. Stubley, J.C.S. Furaday ZZ, 1977, 73, 1632. 27 C. A. Leng, P. Turq, and J. S. Rowlinson, unpublished results.

ISSN:0306-0012    

DOI:10.1039/CS9780700329

出版商:RSC    

年代:1978

数据来源: RSC

摘要:

Meldrum’s Acid By Hamish McNab DEPARTMENT OF CHEMISTRY, UNIVERSITY OF EDINBURGH, WEST MAINS ROAD, EDINBURGH EH9 355 1 Introduction In 1908, A. N. Meldrum reported that the condensation of malonic acid with acetone, in acetic anhydride containing a small amount of sulphuric acid, gave a white crystalline solid, C6H~04, which titrated as a monobasic acid and lost carbon dioxide on heating.l He reasonably stated that ‘there can hardly be a doubt’ that the compound was the carboxylic acid (1). Sadly, this structure proved to be quite wrong, though forty y:ars were to elapse before it was deduced that the condensation must involve only the carboxyl groups of the malonic acid. On this basis, Davidson and Bernhard2 correctly assigned the structure of ‘Meldrum’s acid’ as 2,2-dimethyl-1,3-dioxan-4,6-dione(2), whose properties therefore relate to those of other cyclic 1,3-diones such as dimedone (3) and barbituric acid (4).0 (1) (2) (3) (4) 2 Preparation The original method is invariably used for the preparation of 1,3-dioxan-4,6 diones2 (Scheme 1). The procedure has some generality, and derivatives have HO,CXCO,H + RlYR3 AcZO x WO,R R‘ 0 o+ R R‘Scheme 1 been made by condensing aliphatic3 or aromatic4 ketones, or aryl aldehyde^,^ A. N. Meldrum, J. Chem. SOC.,1908, 93, 598. a D. Davidson and S. A. Bernhard, J. Amer. Chem. SOC.,1948, 70, 3426. B. Eistert and F. Geiss, Chem. Ber., 1961, 94, 929. J. Swoboda, J. Derkosch, and F. Wessely, Monatsh., 1960, 91, 188. A. Michael and N. Weiner, J.Amer. Chem. SOC.,1936, 58, 680. Meldrum’s Acid with un-, mon0-,3?~ malonic acids. The reaction is apparently or di-s~bstituted29~ ineffective for all alkyl aldehydes. Despite the novelty of the reaction (Scheme l), its mechanism remains un- investigated, though it has long been recognized that the first step involves acid catalysed formation of a mixed anhydride of malonic and acetic An alternative-and unusual-source of Meldrum’s acid is from the action of carbon suboxide on acetone in the presence of oxalic acid.g The function of the latter seems to be to supply the elements of water needed for the stoicheiometry.1° 3 Physical and Chemical Properties By the standards of organic chemistry 1,3-dioxan-4,6-dione derivatives are indeed strong acids.Meldrum’s acid itself (pK 4.83)11 is comparable in strength to acetic acid (pK4.76) and is some ten pKunits more acidic than acyclic malonate esters. The explanation of this facile proton loss-as with dimedone (pK 5.2)2-lies in the stability of the resultant anion (5), in which the .rr-orbitals are rigidly held in the ideal configuration for overlap. The meneidicl2 properties of this profoundly stable structure rationalize much of Meldrum’s acid chemistry. Surprisingly, the tautomeric properties of dimedone and Meldrum’s acid are quite different. The former exists predominantly in the mono-enol form, whereas Meldrum’s acid is overwhelmingly ( > 99.5 %) the diketo tautomer.13 It seems that no special explanation may be required for this anomaly, since esters are generally enolized to a much smaller extent than ketones.l* The problem of the conformation of the 1,3-dioxan-4,6-dione system has been approached by n.m.r.15 and dipole-momentl6 methods.Although there is no E. Ott, Annalen, 1913, 401, 159. P. J. Scheuer and S. G. Cohen, J. Amer. Chem. SOC., 1958, 80, 4933. A. Michael and J. Ross, J. Amer. Chem. Soc., 1933, 55, 3684. 0. Diels, R. Beckrnann, and G. Tonnies, Annalen, 1924, 439, 76. lo T. Kappe and E. Ziegler, Angew. Chem., 1974, 86, 529; Angew Chem. Internat. Edn., 1974, 13, 491. l1 (a)K. Pihlaja and M. Seilo, Acfa. Chem. Scand., 1968,22, 3053; (b)K. Pihlaja and M. Seilo, Acta. Chem. Scand., 1969, 23, 3003. l2D. Lloyd and D. R.Marshall, Angew.Chem., 1972, 84, 447; Angew. Chem. Internat. Edn., 1972, 11, 404. l3 M. Eigen, G. Ilgenfritz and W. Kruse, Chem. Ber., 1965, 98, 1623. l4 A. Cero, J. Org. Chem., 1954, 19, 1960. l5 P. Ayrtiis and A. Partanen, Finn. Chem. Letters, 1976, 110. lo (a) D. Korberl and P. Schuster, Monatsh., 1972, 103, 1483; (6) D. Korberl and 0. E. Polansky, Monatsh., 1973, 104, 1421 ; (c) E. N. Klimovitskii, L. K. Yuldasheva, and B. A. Arbuzov, Izvesf. Akad. Nauk. S.S.S.R.,Ser. khim., 1973, 1527. McNab overall agreement as yet, the consensus of opinion favours a boat structure for most derivatives, with a chair structure for 2,2,5,5-tetrasubstitutedcom-pounds. This last conclusion is confirmed by the X-ray structure of an esoteric derivative.17 Other physical properties of the system which have been discussed include u.v.,~ i.r.,I* mass,l9 lH-,20 and 13C-n.m.r.21 spectra.The chemistry of Meldrum’s acid is dominated by its susceptibility to nucleo- philic attack at positions 4 and 6 and to electrophilic attack (via the anion) at position 5. Simple hydrolysis to the malonic acid is a common example of nucleo- philic attack, and may be accomplished under acidic or basic c0nditions.l The reaction mechanism has been studied.11 Use of alcoholic hydrogen chloride yields the malonate diester,3*4 while ‘solvolysis’ by phenols gives monoaryl esters which can be easily converted to diary1 esters.22 Ketones react with Meldrum’s acid by displacement of acetone to give 2,2-disubstituted-l,3- dioxan-4,6-diones.23 Nitrogen nucleophiles give a possible route to monoamides of malonic acid,24 although the cleavage of the ring may be followed by decarb- oxylation, as is the case with aniline,‘ and Hector’s base25 (Scheme 2).Frag-mentation of the ring may also be accomplished directly by pyrolysis. RNH, Scheme 2 In contrast to these examples, the reactions of Meldrum’s acid with electro- philes generally leave the ring intact. Scheme 3 gives the flavour of the chemistry involved, which is considered in detail in later sections. Reaction of electro- philes at the carbonyl oxygen does not give stable cyclic prod~cts.~~~~~~~ 4 5,5-Dialkyl Derivatives These derivatives may be made by standard condensation methods, but an important alternative route is by the reaction of Meldrum’s anion (5) with alky16 l7 P.G. Jones and 0. Kennard, Cryst. Struct. Comm., 1977, 6, 97. (a) R. A. Abramovitch, Canad. J. Chem., 1959, 37, 361; (6) E. E. Ernstbrunner, J. Mol. Structure, 1973, 16, 499. H. Egger, Monatsh., 1967, 98, 1245. eo I. Schuster and P. Schuster, Tetrahedron, 1969, 25, 199. 21 P. Ayras, Acta. Chem. Scand., 1976, B30, 957. 22 (a) H. Junek, E. Ziegler, U. Herzog, and H. Kroboth, Synthesis, 1976, 332; (b) G. Uray,H. Junek and E. Ziegler, Monatsh., 1977, 108, 423. *3 E. Ziegler, H. Junek, and H. Kroboth, Monatsh., 1976, 107, 317. 24 (a) C. D. Hurd and S. Hayao, J. Amer. Chem. SOC.,1954, 76, 5563; (6) P. Crooy, R. De, Neys, J. Eliaers, R. Liveyns, G. Simonet, and J. Vandevelde, Bull.SOC.chim. belges., 1977, 86, 991. 26 A. R. Butler, J. Chem. Res. (S), 1978, 50; (M),0855. S. I. Zav’yalov, Izvest. Akad. Nauk S.S.S.R., Otdel. khim. Nauk, 1961, 2042. G. A. Bihlmayer, P. Schuster, and 0. E. Polansky, Monatsh, 1966, 96, 145. 347 Meldrum’s Acid 0x0 0vo N* 0XOtT.:;,,NNHAr + SMe, Ac,O Scheme 3 (or activated heterocyclic28) halides [Scheme 3, equation (A)]. This almost in- variably29 gives the dialkyl derivative, and the procedure has been logically extended to the preparation of spiro compounds from dibrom~alkanes.~~ Since the formation of anions such as (5) cannot compete, hydrolysis of 5,5-disubstituted-l,3-dioxan-4,6-dionesbecomes almost instantaneo~s3~ under basic conditions. Malonate mono- and di-esters are also available, by standard methods.28 (a) F. X. Smith and G. G. Evans, Tetrahedron Letters, 1972, 1237; (6) F. X. Smith and G. G. Evans, J. Heterocyclic Chrm., 1976, 13, 1025; (c) F. X. Smith and A. Scoville, J. Heterocyclic Chem., 1977, 14, 1081. 29 R. B. Mane and G. S. K. Rao, Chem. and Ind., 1976, 786. 30 J. A. Hedge, C. W. Kruse, and H. R. Snyder, J. Org. Chem., 1961, 26, 992. 31 K. Pihlaja and J. Ketola, Finn. Chem. Letters, 1976, 123. McNab Of particular interest are the spiro cyclopropyl derivative~3~ e.g. (14) whose Michael-type reactions with nucleophiles are reminiscent of those of the 5-methylene compounds (Section 5). Intramolecular decomposition of the ring-opened product may lead to lac tone^^^.^^ (Scheme 4) or lac tarn^.^^ 5 5-Methylene derivative^^^ In some instances, reaction of a carbonyl compound (or its equivalent) with malonic acid proceeds directly by a double condensation to give 5-methylene derivatives of 1,3-dio~an-4,6-diones~~~~and a corresponding procedure using carbon suboxide is known.36 However, the preparative method of choice involves Knoevenagel condensation of Meldrum's acid with a carbonyl component [Scheme 3, equation (B)].This reaction proceeds easily for aromati~~~l~~ or hindered aliphatic aldehydeP or aliphatic ketone~,3~*~0 though aromatic ketones require activation either by the use of a catalyst,41 or by their prior conversion to a ketimine.42 Simple aliphatic aldehydes often give the 2:l condensation product, by Michael addition of Meldrum's acid to the intermediate 5-methylene compound3~3g-a process which is of paramount importance also in dimedone chemistry.In these examples, the 5-methylene compound is available by the cunning use of methoxide as a competitive nucleophile in the Michael addition. The adduct (6; Nu = OMe) so obtained may be decomposed with acid to the required product.43 The activation of a carbonyl compound by 0-methylation is particularly important when the resulting salt is resonance stabilized. Base catalysed con- densation of such cations with Meldrum's acid has afforded 5-methylene 32 (a)S. Danishevsky and R. K. Singh, J. Amer. Chem. SOC.,1975,97,3229; (b)S. Danishevsky and R. K. Singh. J. Org. Chem., 1975, 40, 3807; (c) R.K. Singh and S. Danishevsky, J. Org. Chem., 1976, 41, 1668. 33 F. P. Schmook and 0. E. Polansky, Monarsh., 1969, 100, 1631. 34 For a review, see F. J. Kunz, P. Mzrgaretha, and 0. E. Polansky, Chinria (Swirz.), 1970, 24, 165. 35 D. T. Mowry, J. Airfer. Chenr. SOC.,1947, 69, 2362. 38 H. Hopff and G. Hegar, Helv. Chim. Acta., 1961, 44, 2016. E. J. Corey, J. Amer. Chem. Soc., 1952, 14, 5897. 38 P. Schuster, 0. E. Polansky, and F. Wessely, Munatsh., 1964, 95, 53. 39 J. A. Hedge, C.W. Kruse, and H. R. Snyder, J. Org. Chem., 1961, 26, 3 166. ho G. Swoboda, J. Swoboda, and F. Wessely, Munatsh., 1964, 95, 1283. 41 G. J. Baxter and R. F. C. Brown, Austral. J. Chetn., 1975, 28, 1551. 42 G. A. Bihlrnayer, F. J. Kunz, and 0. E. Polansky, Munatsh., 1966, 97, 1293.43 P. Margaretha and 0. E. Polansky, Tetrahedron Letters, 1969, 4983. 349 Meldrum’s Acid pyryli~m,~Scompounds derived from cyclopr~penium,~~ and other hetero-cyclic46 salts. Eistert has made similar products from chloropyrylium salts,47 and even, in a peculiar redox reaction, from the unsubstituted pyrylium salt itself.48 0’XT,0XO 0x0 0x0 (9) The parent 5-methylene compound (7) is apparently highly reactive, and cannot be prepared by standard methods. The sequence which was ultimately successful (Scheme 5) employs a unique oxidative process in the key ~tep.~g 0XOoQ.o Me. SePh 8) Scheme 5 A number of functional derivatives of 5-methylene compounds are known. Thus (9; X = H, Y = OEt) is readily available from Meldrum’s acid and triethylorth~formate~~[Scheme 3, equation (C)], while addition of an amine to this reaction mixture gives the amino compounds (9; X = H, Y = NHR) in a one-pot synthesis of wide general it^.^^^^^ The diamino compound (9; X = Y = NHC6-H11), which is an extended urea, can be made from dicyclohexylcarbodi-imide,52 and the ‘amides’ (9; X = OH, Y = NHR) from is~cyanates.~~ A remarkable property of 5-methylene derivatives in general is that like the parent 1,3-dioxan-4,6-dione, they are unexpectedly strong a~ids.~~~~~ In this 44 T.Eicher, T. Pfister, and N. Kreuger, Org. Prep. Proc. Internat., 1974, 6, 63 (Chem. Abs.. 1974, 81, 3403). I6 (a)J. A. Van Allan and G. A. Reynolds, J. Hetrrucyclic Chem., 1971, 8, 803; (6) J.A. Van Allan and G. A. Reynolds, J. Heterocyclic Chem., 1972, 9, 669; (c) J. R. Wilt, G. A. Rey- nolds, and J. A. Van Allan, Tetrahedron, 1973, 29, 795. R. Neidlein and M. H. Salzl, Annalen, 1977, 1938. “B. Eistert and T. J. Arackal, Chem. Ber., 1975, 108, 2397. 4a B. Eistert, A. Schmitt. and T. J. Arackal, Chem. Ber.. 1976, 109, 1549. K.F. C. Brown, F. W. Eastwood, and G. L. McMullen, Austral. J. Chem., 1977, 30, 179. 50 G. A. Bihlmayer, G. Derfiinger, J. Derkosch, and 0.E. Polansky, Monatsh., 1967, 98, 564. 51 Sterling Drug Inc., Brit. 1,147,759 (Chem. Abs., 1969, 71, 70 125). 52 A. Stephen, Munatsh., 1966, 97, 695. j3 U. Herzog and H. Reinshagen, Eur. J. Mrrl. Cheni-Chim. Ther., 1975, 10, 323. .54 P. Schuster, A. Stephen, 0. E.Polansky, and F. Wessely, Monotsh., 1968, 99, 1246. 350 McNab case, it is a manifestation of Lewis acidity,55 due to the formation of pseudobase adducts or anbadons (6) for which the driving force is clearly the meneidic ten- dency of the stable anion. The whole range of hard and soft nucleophiles, from alkoxides55 to phosphines56 show this behaviour. The pseudobase may be quenched with iodomethane to give the alkylated products (10).40956 In some cases where the 5-methylene compound has a ,8-hydrogen available, proton loss (i.e. Bransted acid behaviour) can compete with pseudobase formation [e.g. (1111.57 Clearly the pseudobases (6; Nu = OH) are simply intermediates in the hydro- lysis of 5-methylene derivatives to Meldrum’s acid and the carbonyl component, and the kinetics and mechanism of this process have been ~tudied.~8 Under certain conditions, the heterocyclic ring may be preferentially hydrolysed, pro- viding a route to 2-methylenemalonate~.~~*~~Reduction of 5-methylene-l,3- dioxan-4,6-diones to the corresponding 5-alkyl compound is possible catalyti- cally,30 or using lithium aluminium h~dride.~~ Reaction with Fez(C0)g gives a n-cornplex760 while the labile formyl derivative (9; X = H, Y = OH), produced by hydrolysis of the ethoxy compound, gives a chelate with copper acetate.50 The 5-methylene derivatives of Meldrum’s acid are excellent dienophiles, and give a potential route to cyclohexenes with unusual substituents.49~61 This reaction has recently been exploited by two groups in natural product synthesis62963 and further developments in this area are awaited with interest. The 173-dipolar cycloaddition of dia~oalkanes~~has been studied in great detail.In all cases, cyclopropane derivatives are obtained, although the reaction may proceed twice65 (Scheme 6; R1 = P5, R2 = H) and traces of further alkylation can sometimes be detected.33 That the initial reaction is indeed a standard 1,3-dipolar cycloaddition, was proved by an elegant kinetic study,66 whose results were vindicated by the isolation of an intermediate pyra~oline~~ (Scheme 6; R1 = Pri, R = C02Et). 6 5-Halogeno Derivatives Although the 5-chloro compound has been prepared,G8 the known chemistry of 5-halogeno derivatives of Meldrum’s acid is almost wholly associated with brominated materials.Both the 5-bromo and the 5,Sdibromo compounds are formed by direct halogenation of Meldrum’s acid in the presence of one6 or 56 P. Schuster, 0. E. Polansky, and F. Wessely, Tetrahedron Suppl. 8 Part 2, 1966, 463. 58 P. Margaretha and 0. E. Polansky, Monatsh., 1969, 100, 576. 67 J. Leitich, P. Schuster, and A. Eitel, Tetrahedron, 1967, 23, 2221. 58 P. Margaretha, P. Schuster, and 0. E. Polansky, Tetrahedron, 1971, 27, 71. 68 A. Stephen and F. Wessely, Monatsh., 1967, 98, 184. Eo E. Korner von Gustorf, 0.Jaenicke, and 0. E. Polansky, Z. Natrirfursch. B., 1972, 27, 575. F. J. Kunz and 0. E. Polansky, Monatsh., 1969, 100, 920. 6z W. G. Dauben, A. P. Kozikowski, and W. T. Zimmermann, Tetrahedron Letters, 1975,515.83 G. A. Mock, A. B. Holmes, and R. A. Raphael, Tetrahedron Letters, 1977, 4539. O4 F. Wessely and A. Eitel, Monatsh., 1964, 95, 1577. 65 G. Swoboda, A. Eitel, J. Swoboda, and F. Wessely, Monatsh., 1964, 95, 1355. e6 F. Nierlich, P. Schuster, and 0. E. Polansky, Monatsh., 1971, 102, 438. 87 H. Kisch, F. Mark, and 0. E. Polansky, Monatsh., 1971, 102, 448. O8 0. Y. Neiland and B. Y. Karele, Zhur. org. Khim., 1971, 1674. Meldrum's Acid R' = Ph, R' = H: -70 C r 1 RTHN,0X O I 4 O 2R2 l RI 1 \ -J f RZ R' = Ph, R' = H:+ 20 C Scheme 6 moles of base respectively [Scheme 3, equation (D)]. A unique situation pertains to iodine-containing derivatives, since a phenyliodonium ylid is well characterized68 yet no classical iodo compound has ever been reported.Treatment of the 5-bromo-derivative (12; R = H) with nucleophiles can give rise to three possible modes of reaction (Scheme 7). The trivial case, in which UU + Br-Br R Br R Nu Scheme 7 the nucleophile acts as a base, is the dominant route even for such mild bases as aniline.6g The ubiquitous anion may also be formed by nucleophilic attack at the halogen atom (Scheme 7, route 2). This route is particularly important 69 H. R. Snyder and C. W. Kruse, J. Anirr. Chem. So<..,1958, 80, 1942. McNab for (12; R =Me), the use of which as a brominating agent for carbanions has been recommended.70 Trost’s process of ‘transfer alkylation’ is a subtle applica- tion of this principle, in which the anion leaving group re-attacks the brominated substrate at a remote site, to complete a chain extension sequence.71 Although there is no unambiguous example of a simple SNreaction from (12; R =H), this mechanism can apparently give rise to products with hard nucleophiles and (12; R =Me).72 The 5,5-dibromo compound (12; R =Br) is also a useful brominating agent,73 but little else is known of its chemistry save for two astonishing reactions.69 Basic hydrolysis gives carbon tetrabromide in a process which, for its total destruction of a stable system to C1 fragments is reminiscent of the bromopicrin reaction.74 Secondly, simple dissolution in dimethylformamide yields the coupled derivative (13) in 30%yield.The mechanisms of these reactions are overdue for investigation.x Ox0 oL)(Jo H NH2 (13) (14) 7 5-Nitrogen-containing Derivatives Although Davidson and Bernhard reported a colour reaction of Meldrum’s acid with sodium nitrite,2 it was not until 1961 that two groups independently isolated the oxime as an unstable yellow solid3926 [Scheme 3, equation (E)]. Its reduction over platinum dioxide provides the only known route to the 5-amino compound3 (,l4), but otherwise the properties of this compound are unknown. Similarly little studied are the hydrazones, prepared by coupling with appropriate diazo compounds75 or diazonium salts3 [Scheme 3, equation (F)]. The parent 4,5,6-trione of which these compounds are formal derivatives, remains un-prepared, despite a number of attempt~.~~*~~ The most important nitrogen-containing derivative of Meldrum’s acid is the diazo-compound (15), formed by diazotization of the amir~e,~ or, more con- veniently, by direct reaction with p-toluenesulphonyl azide75 [Scheme 3, equa-tion (G)].Photolysis or thermolysis of (1 5) generates ‘Meldrum’s ~arbene’~~ which io (a) J. P. Marino, J.C.S. Chem. Conzm., 1973, 861 ;(6) E. Vedejs, W. R. Wilber, and R. Twieg, J. Org. Chem., 1977, 42, 401. i1 B. M. Trost and L. S. Melvin jun., J. Amer. Chem. SOC.,1976, 98, 1204. i2 R. F. C. Brown, F. W. Eastwood, S. T. Lim, and G. L. McMullen, Austral. J. Chem., 1976, 29, 1705. 73 R. Bloch, Synthesis, 1978, 140. “A. R. Butler and H. F. Wallace, J. Chem. SOC.(B), 1970, 1758.75 (a) M. Regitz and D. Stadler, Annalen, 1965, 687, 214; (6) M. Regitz, A. Liedhegener, and D. Stadler, Annalen, 1968, 713, 101. 76 M. Regitz and H-G. Adolph, Annalen, 1969, 723, 47. 7i A. Hochrainer and F. Wessely, Monatsh., 1966, 97, 1. M. Jones Jr., W. Ando, M. E. Hendrick, A. Kulczycki jun., P. M. Howley, K. F. Hummel, and D. S. Malament, J. Amer. Chem. SOC.,1972,94,7469. Meldrum’s Acid may be trapped with soft nucleophiles to give ylide~,~~q~O or with olefins in the presence of triplet sensitizer to give the predicted trans cyclopropane derivatives.78 In the absence of trapping agent, the normal ketocarbene Wolff ring-contraction takes place (Scheme S), though under extreme conditions, the system may be totally destroyed, with acetone and carbon monoxide as dominant products.81 (15) Scheme 8 8 Ylides Meldrum’s acid ylides (1 6) of the sulphonium,77~79 selenonium,82 arsonium780 and iodonium68 series have been prepared.Their syntheses require either condensation of Meldrum’s acid itself with a highly oxidized hetero-component (possibly generated in sit^),^^ or the decomposition of the diazo-compound (1 5) in the presence of the hetero-component itself (Scheme 9). In addition, the 0x0 lor OX0 + R,,M0,+RnMX2 + 0 a0 -oXo 7oJ+o + MR, N* a; M = S, R = dialkyl, n = 2 (X = OAc) b; M = Se, R = Me, ti = 2, X = Br C; M = AS, R = Ph, n = 3 (X = OAC) d; M = 1, R = Ph,II = I, X = OAc Scheme 9 dimethylsulphonium ylid (16a) is available from the bromo compound77 (12; R = H) and from the iodonium ylid68 (16d).The most noteworthy property of these ylides is their remarkable stability, inevitably associated with the dipolar canonical form (16). All are crystalline solids which may be handled without special precautions. The dimethylsulphonium and phenyliodonium ylides (16a) and (16d) show 78 W. Illger, A. Liedhegener, and M. Regitz, Annalen., 1972, 760, 1. I. Gosney and D. Lloyd, Tetrahedron, 1973, 29, 1697. slS. L. Kammula, H. L. Tracer, P. B. She\lin, and M. Jones jun., J. Org. Chem.,1977, 42, 293 1. 8a E. Ernstbrunner and D. Lloyd, Annalen, 1971, 753, 196. 83 J. I. G. Cadogan and 1. Gosney, JCS Perkin I, 1974, 466. McNab pronounced thermal stability by comparison with related systems, while the stability of the selenonium and arsonium ylides (16b) and (16c) is reflected in their lack of reactivity with aldehydes in attempted Wittig reactions.The iodonium ylid (1 6d) gives 5-chloro-l,3-dioxan-4,6-dioneon treatment with hydrogen chloride.68 9 Pyrolysis The pyrolysis of Meldrum's acid derivatives has long been known to give ketenes.6 The mechanism presumably involves homolysis of a C-0 bond, followed by cleavage of acetone and carbon dioxide, both of which are found in the pyrolysate along with the ketene.84 (Scheme lo). Using flash vacuum L, y +co,+ RyR0 IIoJ$o 0 RR Scheme 10 pyrolysis, this route may be used to prepare ketenes which are difficult to obtain by conventional met hods-no tab1 y cyclopropyl ke tenes5-bu t the technique has been applied most particularly to the generation of methyleneketenes from 5-methylene derivatives.Thus phenylmethyleneketene (1 7) has been identified by its i.r. spectrum at -196"C,by its dimerization to the cyclobutandione deriva- tive (18), and by various trapping experimentss4 (Scheme 11). Methyleneketenes 0 Scheme 11 are available thus, in two steps, from aryl aldehydesS4 and ketoness6, or alkyl ketones9 those derived from aliphatic aldehydes tautomerize if possible to vinyl ketenes.84 Other more interesting intramolecular reactions which may take place in the presence of a suitable trapping group, include an elegant phenol 84 R. F. C. Brown, F. W. Eastwood, and K. J. Harrington, Austral. J. Chem., 1974, 27,2373.G.J. Baxter, R. F. C. Brown, F. W. Eastwood, and K. J. Harrington, Tetrahedron Letters, 1975, 4283. R. F. C. Brown, F. W. Eastwood, K. J. Harrington, and G. L. McMullen,Austra/.J. Chem., 1974, 27,2393. Meldrum’s Acid synthesiss7 (Scheme 12). Recently, the parent methyleneketene has been identified from the pyrolysis of the very reactive compound (7), itself generated in sit[/ by thermal retrocyclization from the adduct (S)49 (Scheme 5). 0 OHA6 q)E!!5+Q)A II 4 0% Scheme 12 Pyrolysis of Meldrum’s acid derivatives at high temperatures causes decar- bonylation of the intermediate ketene to give a carbene, which then rearranges to valence-satisfied products. Reactions of this type include the generation of acyloxycarbenes and their rearrangement to a-diketones in excellent yield (Scheme 13).72 Similarly, methylenecarbenes generated from methylneeketenes I co R’ Scheme 13 show hydrogen migration to give acetylenes, although intramolecular cycliza- tion products may also be detected in favourable instances.B6 10 Some Synthetic Uses As with simple malonic ester syntheses, the presence in Meldrum’s acid of a highly active methylene group adjacent to an ester function, allows the molecule to be manipulated to give a wide range of possible products.At the simplest level, Meldrum’s acid can act as a methylene synthon by hydrolysis and exhaustive decarboxylation of its derivatives. For certain heterocyclic systems, the conditions needed for this sequence are surprisingly mild28*45 (Scheme 14).Intermolecular cleavage by other nucleophiles can also give useful products; a particularly beautiful example is Dauben’s two step synthesis of 6-damascone62(19) Scheme 15). This is the only known example of a potentially general ketone synthesis from Meldrum’s acid. We have already met examples of intramolecular nucleophilic attack (Scheme 4). IJnder extreme conditions, cleavage of acetone may be followed by decarboxylation to give lactamss8 (Scheme 16); in these cases, Meldrum’s acid acts as the synthetic equivalent of ketene. 87 (a)R. F. C. Brown and G. L. McMullen, Austral. J. Chem., 1974, 27, 2385; (6) G. J. Baxter, R. F. C. Brown, and G. L. McMullen, Airstral. J. Chetn.. 1974, 27, 2605. 88 G.Y. Lesher, U.S. 3,907,798 (Chrm.Ah., 1976, 84, 44130). 356 McNab AcOHIHCIO, 100,lo min Ar Ar02:r Scheme 14 Scheme 15 0x0 H Scheme 16 0 0 OH XO 0Uo+ mo Scheme 17 357 Meldrum’s Acid Finally, a widespread use of Meldrum’s acid in heterocyclic chemistry is in the preparation of a-pyrones and 2-pyridones. A variety of such reactions is known, and some are of formidable complexity, but a common feature is that the Meldrum’s acid behaves as a C302 synthon45~89~90(Scheme 17). I am grateful to Professor J. I. G. Cadogan for his encouragement, and to Dr. D. Lloyd for his invaluable comments on the manuscript. 89 (a) E. Ziegler, H. Junek, and U. Herzog, Monatsh, 1971, 102, 1096; (6) E. Ziegler, H. Junek and U. Herzog, Monatsh, 1971, 102, 1626. Bo 0. S. Wolfbeis, Monarsh, 1977, 108, 499.

ISSN:0306-0012    

DOI:10.1039/CS9780700345

出版商:RSC    

年代:1978

数据来源: RSC

摘要:

Bond Valences-A Simple Structural Model for Inorganic Chemistry By I. D. Brown INSTITUTE FOR MATERIALS RESEARCH, MCMASTER UNIVERSITY, HAMILTON, ONTARIO, CANADA L8M 4S1* 1 Introduction In spite of the advances made in quantum chemistry during the past sixty years, organic chemists still describe many chemical structures in terms of the nine- teenth century (as opposed to the twentieth century) concept of valence. In this familiar ‘ball and stick’ model, atoms are regarded as connected together by bonds which are either single, double, triple, or ‘delocalized’. The number of bonds each atom can form is determined by its valence, an intrinsic property that is determined by the position of the atom in the periodic table. The resulting picture of a molecule not only accounts for many of its chemical properties, but also gives a description of its three-dimensional structure that is hardly less accurate than that provided by X-ray diffraction.Unfortunately, the inorganic chemist has been less well served by chemical models, both those of the siniple valence type and those based on electrostatic or quantum principles. Like organic compounds, inorganic compounds can be treated as atoms linked by bonds, but the bonds can vary greatly in character and their number does not correspond in any obvious way to the atomic valence. Around some atoms, notably hydrogen and the alkali metals, the bond distances vary so widely that it is often difficult to decide which neighbours are to be considered bonded and which are not.Inorganic structures can rarely be pre- dicted with confidence in spite of the large number of accurately determined structures now known. Indeed, to give even a qualitative description of the bond- ing in such structures it is often necessary to use several chemical models. As a result inorganic structures are frequently described in purely geometric terms such as the packing of polyhedra or the close packing of spheres. In this review I shall describe how the old ‘ball and stick’ valence model can be extended to inorganic compounds. The inorganic version differs from that used by organic chemists, but is equally powerful in accounting for structure and reactivity. The basic ideas are not new. They were described by Pauling fifty years ago,l but since he presented them in terms of an electrostatic model it was not *This review was prepared during a sabbatical leave spent at the University Chemical Laboratory, Lensfield Road, Cambridge.L. Pauling, J.Amer. Chem. Soc., 1929, 51, 1010. 2 359 Bond Valences-A Simple Structural Model for Inorganic Chemistry realized how widely they could be applied.” To avoid the limitations implied by such physical models it is better to develop the theory in a purely empirical way. Only then is it profitable to compare it with the more familiar physical theories of covalence and ionicity. 2 Definition and Use of Bond Valences We shall assume that chemical structure can be completely described as a net- work of bonds connecting neighbouring atoms.The network may be finite, as in the case of a molecule, or infinite, as in the case of most inorganic crystals. We shall call the number of bonds formed by an atom its co-ordination number and the bonding power of an atom its valence. The valence is determined by the position of the atom in the periodic table and can be thought of as the number of electrons that the atom uses for bonding. For the compounds considered here the valence is synonymous with the formal oxidation state. In organic compounds the valence is usually equal to or greater than the co-ordination number giving rise to single or multiple bonds, but in inorganic compounds the opposite is usually true, so that the valence to be associated with each bond is frequently less than one.We define the valence of a bond (sij) between atoms i andj such that for all i, where Vi is the valence of atom i. This is called the valence sum rule. It is the same rule which, combined with the restriction that sij is an integer, allows one to write down unique structures for organic molecules. In inorganic com- pounds it is not so easy to assign bond valences npriori since they are usually non- integral and range widely in value. On the other hand, in inorganic compounds the bond valences correlate much better with the bond lengths than they do in organic compounds. As a result they can be accurately determined from the experimentally measured lengths. Our knowledge of the correlation between bond length (R)and bond valence (s) has increased over the past fifty years as the accuracy of crystal structure deter- minations has improved.The relation is of an inverse type (Figure 1) that can usually be fitted over the range of interest by one of a number of analytical expressions, the most commonly used ones being those in (2) and (3), where RO R = R, -Alog,,s (2) s = (R/R,)-” (3) is the length of a bond of unit valence and N (-5) and A ( N 0.8) are parameters, which determine the slope of the curve [equation (2) was proposed by Pauling2 on the grounds that the electron density in an atom drops exponentially with distance, It has the advantage that A = 0.85 for many bonds. Equation (3) is *Pading uses the term ‘electrostatic bond strength’, but the term ‘bond valence’ is preferred since it emphasizes the empirical nature of the concept.* L. Pauling, J. Ai7wr. Ch~nrSoc., 1947, 69, 542. Brown 1.5 2.0 2.5 3.0 DI STANCE / 6 Figure 1 Correlation between valence and distance for bonds between 0 and atoms of the second row (Composite after I. D. Brown and R. D. Shannon, Acta Cryst., 1973, A29, 266, and G. Donnay and R. Allmann, Arner.. Miner.., 1970,55, 1003) related to the Born3 equation with N + 1 being the Born exponent]. Ro and N or A are chosen to satisfy equation (1) in a large number of accurately deter- mined structures and, once found for a given pair of elements, can be used for any further bonds between them. Tables of R9 and N or A for bonds between metals and oxygen have been given by Allmann using equation (2) and Brown and Wu using equation (3).* General methods for obtaining the parameters have been described by various a~thors.~ Donnay and Donnay5 have compared a series of different correlations and conclude that most of them work equally well.Brown and Shannon6 have observed that the same curve (Figure 1) can be used for all the bonds between 0 and Na, Mg, Al, Si, P, and S, making for a great simplification in treating the structures of these important elements. In the form of Pauling’s electrostatic valence rule,l the bond valence model has been used for many years by mineralogical crystallographers, who have found it ideal for confirming structures that contain a complex mixture of ionic and covalent bonds.Using the present approach bond valences can be calculated directly from the observed bond lengths and can be summed around each atom to give an experimental measurement of the atomic valence. If this agrees with the theoretical valence, one has confidence that the structure has been correctly determined and correctly interpreted. In this case the bond valences give a quantitative measure of the relative strength of the various bonds and can be used to describe the structure. If the experimental and theoretical valences do not agree the cause can usually be traced to one of the following: M. Born, ‘Atomtheorie des festen Zustandes’, Teubner, Leipzig, 1923. R. Allmann, Monatslz., 1975, 106, 779; I. D. Brown and K. K. Wu, Acta Crjvst., 1976; B32, 1957, and references therein.G. Donnay and J. D. H. Donnay, Actu Crj,sf., 1973, B29, 1417. I. D. Brown and R. D. Shannon, Aclu Crjist., 1973, A29, 266. Bond Valences-A Simple Structural Model for Inorganic Chemistry (i) the material is one (e.g. a metallic solid) to which the model does not apply, (ii) the structure has not been correctly determined, or (iii) the structure is correct, but has not been correctly interpreted, for example bonds have been overlooked or atoms wrongly assigned (e.g. the ordering of Si and A1 in minerals). These points are discussed and illustrated below. Although the bond valence model applies to the majority of inorganic struc- tures it does not hold universally. It clearly cannot be used for compounds such as metal alloys that do not combine according to the stoicheiometric proportions of classical valence theory and the correlation between bond length and valence does not hold well for bonds between atoms of the same element, particularly those in aromatic rings.As a general rule the model can be used for any structure or part of a structure, in which bonds only occur between atoms of the opposite formal charge." It applies to such compounds as Na2S04 and even to acetic acid, but not to the hydrazinium ion (H2NNM3),? nor the carbonyl complexes of transition metals. Calcium chromium pentafluoride (CaCrFs) provides a good example of the use of bond valences. The crystals contain distorted CrF6 octahedra linked by fluorine bridges into parallel chains bonded to each other by calcium atoms as is shown schematically in Figure 2.Although the Ca-F interactions can reasonably be described as ionic, the nature of the Cr-F bonds is more difficult to divine. However, one only needs to know the bond lengths, not the bond character, in order to calculate bond valences and hence experimental atomic valences. The structure was determined by X-ray diffraction and refined in the space group Cc.8 Since the atomic valence calculated for one of the F atoms in this structure was only 0.76 valence units (v.u.), the structure was rerefined in space group C2/c.9 The correctness of this latter refinement was confirmed not only by an improvement in the X-ray diffraction agreement index (Rfactor) from 0.10 to 0.08, but also by the improved atomic valences, none of which deviated by more than 0.08 V.U.from the theoretical value. The environment of the Cr atom in the second refinement is still slightly distorted (Figure 2), but instead of the four short terminal and two long bridging bonds that might have been expected, the co-ordination around Cr contains only two short terminal bonds; the other two terminal bonds and the two bridging bonds are almost the same length. The bond valence model explains this unexpected distortion. The F atom that bonds to only two cations [F(3)] forms stronger (i.e. shorter) bonds than the other two, each of which forms bonds to three cations. Detailed examination of the bond valences *It is therefore convenient to use the terms cation and anion without necessarily implying any separation of charge.+However it can be used for hydrazinium salts providing that the N,H, group is treated as if it were a single atom.' S. Vilminot, M. R. Anderson, and I. D. Brown, Acfa Crj-sf.,1973, B29, 2628. D. Dumora, K. Von der Muhll, and J. Ravez, Materials Res. Bull., 1971, 6, 561. K. K. Wu and I. D. Brown, Materials Res. Bull., 1973, 8, 593. 3 62 Brown Figure 2 Scheniatic diagrnni oj' the bond network in CaCrFj. The numbered circles are F atonis ad the (CrF,)7_chains run hi-izontnllj,. Observed bond lengths are given in A, bond valencies determined fi.oni these lengths are enclosed in parentheses in Figure 2, shows that not only can all the distortions be understood in detail but, given the bonding network, it would not have been difficult using equation (1) to estimate the bond valences a priori. This kind of valence description of chemical bonding has all the advantages of the traditional valence-bond (V-B) description'" but, by taking explicit account of the environment of the molecule or ion, it is able to give a more accurate picture of the real structure.Figure 3a shows the V-B structure of (4 (b) (c) Figure 3 a, Valence-bondstructure of H,SO, ;b, valence-bond structure of HS0,-; c, bond valence structure of HS0,-lo F. A. Cotton and G. Wilkinson, 'Advanced Inorganic Chemistry', Second Edition, Inter- science, New York, 1966, pp. 57ff. Bond Valences-A Simple Structural Model for Inorganic Chemistry H2S04 which is the same as the bond valence structure of the isolated molecule.Removal of one H ion leads to the anion HS04-whose average V-B structure is given in Figure 3b. In such a structure the environment of the ion is ignored, the unsatisfied valences on the oxygen being formally treated as residual charges. But the environment, whether it be the hydration sphere in aqueous solution or adjacent ions in crystals, has an important influence on the structure of the ion. For instance, the hydrogen atom, if not completely detached from the ion, will inevitably form a hydrogen bond to an adjacent anion or other electron donor, so that the actual valence structure will be closer to that shown in Figure 3c.* It is this structure, rather than the one implied by Figure 3b, that is found in the crystalline salts.’l 3 Applications of Experimental Atomic Valences and Bond Valances There are many ways in which the experimentally determined atomic valences can assist in the determination and interpretation of a crystal structure. As in the case of CaCrFs discussed above, poor agreement with the theoretical atomic valences may indicate an incorrectly determined structure.An experimental valence of 0.70 for Rb and 0.87 for two of the I atoms in RbCdI3 first drew our attention to the water of crystallization in this compound.12 Low values of the valence can also indicate that one or more bonds have been over10oked.l~ Hydrogen bonds can be reliably located by noting which anions still have unused valence after account has been taken of all the other bonds.14 Experimental valences are particularly useful in distinguishing between different elements with similar atomic numbers and similar crystal chemistry. Bond lengths have been used for some time to distinguish between A1 and Si atoms in rninerals,l5 but the distinction can be made simply and elegantly by Table 1 Experimental Atomic Vulences itz V407a Average experimental V valence 3.48 k 12 V valence from stoicheiometry 3.50 “(Structure determined by H.Horuichi, M. Tokonami, N. Morimoto, and K. Nagasawa, Acta Cryst., 1972, B28, 1404.) *The structure of the hydrogen bond and the rational for splitting the valence around H in the ratio 0.8 to 0.2 are discussed in Section 6.Once the valence of this bond has been assigned, the other valences all follow from equation (I). l1 R. J. Nelmes, Acta Cryst., 1971, B27, 272. l2 M. Natarajan Iyer, R., Faggiani, and I. D. Brown, Acta Cryst., 1977, 33, 129. l3 E. Kostiner and J. R. Rea, fnorg. Chem., 1973, 13, 2876; A. D. Baranyi, M. Onyszchuk, Y. LePage, and G. Donnay, Canad. J. Chem.. 1977, 55, 849. l4 G. Donnay and R. Allmann, Amer. Miner., 1970, 55, 1003; M. Catti and G. Ferraris, Acta Cryst., 1974, B30, 1789; F. C. Hawthorne, Acta Crysr., 1976, B32, 2891. l5 M. W. Phillips and P. H. Ribbe, Contr. Miner. Petrol., 1973, 39, 327. 364 Brown calculating experimental cation valences, particularly as A1-0 and Si-0 have the same bond valence-bond length curve.If the observed valence at a particular site is 3.0 it is occupied by Al, if 4.0it is occupied by Si and if it is 3.5 the site is occupied equally by A1 and Si atoms.16 The method is illustrated in Table 1, which gives the experimental atomic valences of the mixed valence oxide V407. The correctness of the valence assignment at the vanadium sites is confirmed by the calculation of correct experimental oxygen valences and the excellent agreement between the average experimental vanadium valency and the average vanadium valence determined from the stoicheiometry. Valences can be useful in predicting the structure of materials under the extremes of pressure and temperature that are found in the centre of the earth.Bad7 has used a related technique to discuss the structure of possible unknown high pressure forms of MgzSi04 that are thought to be a major constituent of the mantle. Kahn and Hazen and Prewittl8 have shown that the coefficient of thermal expansion of individual bonds varies inversely with their valence and have used this relationship to predict high temperature structures. Brown, Howard- Lock, and Natarajanlg have suggested that valences might be used to predict stretching force constant and hence i.r. and Raman spectra. 4 Prediction of Structure The bond valence concept treats a chemical compound as a network which must obey the valence sum rule [equation (l)]. This law is formally equivalent to Kirchhoff’s current law in electrical network theory.20 For compounds that have the stoicheiometry predicted by the classical valence rules there is necessarily at I Figure 4 Boiid valence structrrre of (H602)2S04 l6 H.D. Grundy and J. Ito, Amer. Miner., 1974, 59, 1319; M. Calleri and G. Gazzoni, Acfa Cryst., 1976, B32, 1196. li W. H. Baur, Amer. Miner., 1972, 57, 709. la A. A. Khan, Acta Cryst., 1976, A32, I1 ; R.M. Hazen and C. T. Prewitt, Anier. Miner., 1977, 62, 309. l9 I. D. Brown, H. E. Howard-Lock, and M. Natarajan, Canad. J. C~PI?~.,1977, 55, 1511. *O A. L. Mackay and J. L. Finney, J. Appl. Crj.st., 1973, 6, 284. Bond Valerices-A Simple Strirctirral Model for Ittorgarlic Chemistry least one bond valence structure that obeys the valence sum rule. In cases where only one such structure is possible the predicted bond lengths are very close to those observed.Figure 4 shows the valence structure of (Hj0.).SO-r.21 All the S-0 bonds are equivalent by crystallographic symmetry and the H502 ion lies on a crystallographic two-fold axis. The bond valences can be assigned either by dividing the valence of S (V = 6) equally between the four S-0 bonds or the valence of the acidic proton (V = 1) equally between the two bonds that it forms. The other valences follow by application of the valence sum rule to the 0 and H atoms. The agreement between the predicted and observed bond lengths is excellent (S-0, 1.476, 1.479 A; 0-0,. 2.63, 2.66 A; 0,-0, 2.41, 2.43 A respectively). In cases where more than one valence structure can be written, one needs a second theorem.A good solution to the network can be found if one requires the valences of all the bonds formed by each atom to be as nearly equal as is con- sistent with the valence sum rule. An algorithm based on this condition gives predictions of bond valences that are accurate to 0.05 V.U.~~and lengths that are generally accurate to within a few hundredths of an Angstrom provided that the compound does not contain distortions that arise from steric or electronic effects. An alternative procedure based on the same principle has been described by Ba~r.~~ The predictions made by the valence sum rule have for obvious reasons mostly been tested on crystalline structures, but the technique is expected to work equally well in amorphous networks and should therefore prove valuable in the development and testing of models of amorphous structure.Taylor and Brown2* have used it to discuss which impurities will stabilize domain boundaries in CaTiSiO5. The effects of the non-bonding interactions are important in determining co- ordination numbers and thus in determining the structure of the network itself. In most compounds the anions are larger than the cations and the co-ordination numbers are determined by anion-anion contacts. They can be predicted following Pauling25 by considering the ratios of the cation and anion univalent radii but such a 'hard sphere' model does not give a realistic measure of the actual 0-0 non-bonding distances, which are found to vary from 2.2 8, in NO3~-to over 2.8 8, for the 0 atoms surrounding Na. The difference arises from the greater strength of the N-0 bond (5/3 v.u.), which can pull the 0 atoms closer together.We can allow for this effect in the following way. We assume that for a given valence (s) of an A-0 bond there exists a minimum allowed 0-0 distance (Rm),which is given by the relation 21 T. Kjallman and I. Olovsson, Acfn Cryst., 1972, B28, 1692. 22 I. D. Brown, Actu Cryst., 1977, B33, 1305. 23 W. H. Baur, Trans. Anier. Cryst. Ass., 1970, 6, 129. 24 M. Taylor and G. E. Brown, Anicr. Miner., 1976, 61, 435. 25 L. Pauling, 'The Nature of the Chemical Bond', Third Edition, Cornell University Press. Ithaca, 1960, pp. 543ff. 366 Browir where 8 is the angle between that A-0 bond and the 0-0 vector.Tf the values of the constants Ro and N' are taken to be 2.22 A and 7.4, respectively, one gets predictions for the maximum possible co-ordination numbers for 0 around various cations that are as good as those of Pauling's hard sphere model but have the advantage that they involve realistic estimates of the 0-0 distances. This allows the model to be extended to problems of irregular co-ordination such as the bonding around hydrogen dealt with in Section 6. The best scos8 vRm curve for 0-0 interactions is shown in Figure 8. 5 Distorted Cation Environments The valence sum rule and the form of the bond valence-bond length correlation can be used to prove an interesting theorem concerning distorted environments, namely that any deviation of the bond lengths from their average value will increase the average bond length.6 This can be demonstrated as follows: let a cation in regular co-ordination form bonds of length ~2 and valence s2 (Figure 1).Suppose that the environment distorts into two equal groups of bonds with lengths YI and r3 with corresponding valences s1 and s3. Since, by the valence sum rule sl + s3 = 2s2 it follows from the shape of the valence-length curve (Figure 1) that r1 + r3 > 2r2. The effect is quite general and is well attested.26 For example increasing the degree of distortion of VO6 groups can increase the average V-0 distance by as much as 0.1 A. A similar distortion occurring on cooling CaTiSiO5 through 220 "Cis responsible for increasing the crystallographic a axis by 0.01 A.24 A number of corollaries follow from this theorem.The first is that since the application of pressure tends to decrease the average bond length it will also tend to make the co-ordination more regular. An interesting example is provided by C~Te03~~ where both cations have irregular environments. If the co-ordination number of the cations are taken as six, the range of cation-oxygen distances for both Cu and Te changes from 1.9-3.2 8, at low pressure to 1.9-2.6 A at high pressure. Another corollary to this theorem is that if the environment of a cation distorts in such a way that the average bond length is forced to remain constant (e.g. by the cation moving off centre in a rigid polyhedron of anions), then the experi- mental atomic valence of the cation will increase. This effect can be used to discuss the distortions found in AB03 crystals with a perovskite-like structure.In the ideal structure all the atom positions are fixed by symmetry and the only free parameter is the length of the cubic unit cell (Figure 5a). Since the crystals contain both A-0 and B-0 bonds, at least two free parameters are needed if the valence sum rule is to be obeyed. The result is generally a compromise with either the valence of A being too large and the valence of B being too small or vice versa. The cation with the low valence (i.e. that occupies a cavity that is too 26 R. D. Shannon, P. S. Gumerman, and J.Chenavas, Amer. Miner., 1975, 60, 714; R. D. Shannon and C. Calvo, Am Crysr., 1973, B29, 1338; see also ref. 6. 27 E. Philippot and M. Maurin, Rev. Chim. Miner., 1976, 13, 162. Boiid Valences-A Simple Strirctirral Model for Inorganic C'hemistry 00 (b) (aFigure 5 ABO, perovskite structures: a, undistorfed; b, distorted by B cation moving 08 centre in its octahedron; c, distorted by the collapse of the co-ordination polyhedron around A large for it) will tend to distort its environment to increase its valence. Two types of distortion are thus observed: (i) when the octahedral B atom has the low valence and moves off-centre in its octahedron (Figure 5b), and (ii) when the twelve-co-ordinated A atom has the low valence and the octahedra twist to give four shorter and four longer bonds (Figure 5c).In both cases the valence sum rule applied to the 0 atoms requires that the distortions occur co-operatively in adjacent Figure 6 shows the deviations \ \ \ o \o -1 0 I Figure 6,4 plot ofthe deviutions bet ween the experimental and theoretical atomic valences that would be expected for a number of ABX, compounds if they had the ideal perovskite structure. Circles denote those that have the undistorted pervoskite structure (Figure 5a).Squures denote Jtructures distorted 0s in Figure 5b and diamonds structures distorted as in Figure 5c Brown between the experimental and theoretical cation valences that would be expected for a number of ideal perovskite structures (Figure 5a).Only when the experi- mental valences of both cations are within 0.1 V.U. of the theoretical value does the ideal structure occur at room temperature. In all other cases the crystal is found to be distorted. For example, the rare earth ferrites, LFe03, have a structure like that of Figure 598 with observed atomic valences within 0.05 V.U. of the expected values, even though in the ideal structure (Figure 5a) the differ- ences would be as large as 1 V.U. There are some distortions that cannot be predicted from the bond valence model alone since they arise from steric or electronic effects. Typical are those found around H, CuII, SbIII, and TeIv. Even in these cases the valence sum rule holds29 and the bond valences calculated from the bond lengths still provide a good description of the bonding.They provide a natural interpretation of the weak secondary and bridging bonds that are such a conspicuous feature of the oxide and fluoride complexes of atoms of the series SnI1 to XeVL,30 providing a complementary treatment to the VSEPR approach of Gillespie and Nyh01m.~~ Brown3O has also shown that there is a simple correlation in these compounds between the valence and the angles formed by the bonds. Similar correlations between bond angle and length have been noted in the tetrahedral oxyanions such as P043-23 and S042-.32 By studying the distortions found in a large number of crystals, Murray-Rust, Burgi, and dun it^^^ have shown that the angles in most tetrahedral groups change in the same way when one of the ligands is gradually removed. Two extreme geometries define three angles 109" (tetra-hedron), 1 20°, and 90" (triangle with one ligand removed to infinity).Intermediate 8 = 109.5 + 180(d~,'V)-652(d~/V)~ (5) values can be interpolated using equation (5), where V is the cation valence and As is the average deviation from V/4, the ideal valence of the four bonds defining the angle. This relation describes most of the angular distortions found in so42-32 and c104-34 groups and is expected to apply to many other tetrahedral ions where the bonding is str0ng.~3 6 Hydrogen Bonds Few bonds are more important and more difficult to describe theoretically than the hydrogen bond, the system formed by two anions linked by a hydrogen atom.Yet this bond is one that is described well by bond valences. Since the most important and best studied hydrogen bond system is the one where both anions 28 M. Marezio, J. P. Remeika, and P. D. Dernier, Acta Cryst., 1970, B26, 2008. 29 H. Mayer and G. Pupp, Monatsh., 1976, 107, 721. 30 I. D. Brown, J. Solid State Chenr., 1974, 11, 214. j1 R. J. Gillespie, 'Molecular Geometry', Van Nostrand Reinhold Series in Inorganic Chemistry, London, 1972. 32 I. D. Brown, Acra Crysf.,1973, 29, 1979. 33 H-B. Biirgi, Angew. Cheni. Internat. Ed/?.,1975, 14, 460; P. Murray-Rust, H.-B. Biirgi, and J. D. Dunitz, J. Amer. Chetn. Soc., 1975, 97, 921. 34 I. D. Brown, Acta Crjtst., 1976, A32, 786. 369 Bond Vulences-A Simple Structural Model for Illorganic Chemistry are oxygen, the following discussion will, for the purposes of illustration, be confined to the 0-H -.* 0 bond but the results apply mutatis mutandis to other systems. Characteristically the hydrogen bond is asymmetric with the hydrogen atom closer to one 0 (the donor) than the other (the acceptor). The hydrogen bond is usually characterized by its length (the 0-0 distance), which has been shown to correlate with the asymmetry of the bond, short hydrogen bonds (2.4 A) being symmetric and longer ones increasingly more asymmetric. Bonds between 2.4 and 2.75 A are essentially linear (angle 0-H * -* 0 > 170") but thereafter an increase in the length is accompanied by a decrease in the 0-H * 0 angle.35 It is therefore convenient to divide hydrogen bonds into two classes, short (or strong) bonds which are linear and long (or weak) bonds which are more or less bent. The linear hydrogen bond found in ice (0-0 = 2.76 A) can be taken as the dividing point.The bond valence-bond length curve for 0-H bonds is shown in Figure 7*.3*935 1.o 15 2.o 25 30 0-H DISTANCE /A Figure 7Bond valence-bond length correlation for H-0 bonds The asymmetry of the hydrogen bond results from the repulsion between the two 0 atoms.35 According to the distortion theorem (Section 5) this repulsion can be reduced, i.e. the 0 atoms can move further apart, if the H atom moves off the centre of the bond. The increase in the 0-0 distance with the weakening of the hydrogen bond is indicated by the broken curve in Figure 8.The solid curve in Figure 8 is the minimum allowed non-bonded 0-0 distance (Rm from equation 4, in which SCOS~is taken to be the valence of the weaker H * * 0 bond recognizing that for a linear bond cod = 1). *The central section of this curve has been interpolated. If one uses the distances obseried in strong bonds one obtains a curve with discontinuities at s --0.8 and s = 0.2 V.U. The reason for this is discussed below. 35 I. D. Brown, Acta Cryst., 1976, A32, 24 Brown From Figure 8 it is seen that a linear, symmetric 0-H-0 bond with 0-H valences of 0.5 V.U. should have a length of 2.30 A. However, the minimum possible 0-0 distance is 2.42 A. Symmetrical two co-ordination is therefore not normally possible, but by allowing the H atom to move off-centre the sum of the two 0-H bond lengths is increased without increasing its atomic valence.2.0 2.5 3.0 ANGSTROM Figure 8 Solid curve: eflective vnlence s’(= scos0) versus Rm equation (4) indicating the closest possible approach of 0 atoms when bonded to a common cation (e.g. H) by bonds of strength s. Broken curve: sum of the 0-H and H * . * 0 distances in a hydrogen bond whose weak H . . . 0 bond has valence s’ However, the minimum 0-0 distance also increases though not as rapidly, so that the two 0-0 distances become equal at around 2.8 A, with valences for the two 0-H bonds of 0.81 and 0.19. This is the most symmetrical bond that can normally occur and it is the type most commonly found.Shorter hydrogen bonds are formed when the donor 0 cannot provide 0.81 v.u., e.g. when it is a water molecule co-ordinated to a small highly charged cation such as H+ or Be2+. Such short hydrogen bonds have lengths that follow the R(0-0)min curve rather than the broken curve given in Figure 8. Less symmetrical bonds can be found either when the acceptor 0 cannot provide 0.19 v.u., e.g. when it belongs to a strong acid like C~OC.or when the crystal packing requires a bent bond, since for a bent bond R(0-H) + R(H . * * 0) > R(0-0), thereby forcing the bond to become more asymmetric and longer. The geometric constraints implied by Figure 8 suggest that 0-H . * 0 bonds will be linear for 0-0 distances less than 2.8 A and generally bent for larger distances with the 0-H * .. 0 angle determined by R(0-H), R(H . * . 0),and R(0-O)min, all of which are known. The agreement between the predicted and observed geometries is shown in Figure 9. Bond valences are particularly useful in predicting hydration structures around anions. If we assume that in neutral aqueous solution each 0atom of an oxyanion Bond Valences-A Simple Structural Model for Inorganic Chemistry 0 Figure 9 Geometries of 0-H * * * 0 hydrogen bonds. The circles are observed acceptor 0 positions [relative to the O(donor)-H bond]. The solid line gives the locus of predicted acceptor positions either accepts three hydrogen bonds or donates one and accepts one bond, and that these bonds will, where possible, be typical of those found in the adjacent water, we can predict hydration structuresthat agree with crystal structure studies and acid pK values.Figure 10 shows some of the species that are to be expected in neutral water. The hydration structure around C104-will contain hydrogen bonds that are weaker than those found in bulk water in agreement with observa-\\ I I’-‘b\ &:o-i I ’ I .I c --07\ /‘\--0 --0 0-H--’/ I/’ I /’ ‘ /II I‘ //-/-I I I‘ /I I (a)(a) (b)(b) (4 Figure 10 Expected bond valence structures of a, CiO, ,-; b, SOA2-;andc, H,PO,-in aque-ous solution. Broken lines indicate the weaker (H . . . 0)end of a hydrogen bond tion,36 but PO& will act as a donor to one or two hydrogen bonds so that the species HP042-and H2P04-will be the ones found at pH 7.Figure 11 shows the hydration structures predicted for three cations. Beryllium can co-ordinate four 0 atoms with bonds of 2/4 = 0.5 v.u.; Mg can co-ordinate six 0 atoms with bonds of 2/6 = 0.33 V.U. The difference in the strength of the 36 G. Brink and M. Falk, Canad. J. Cherii., 1970, 48, 3019. 372 Brown metal-oxygen bonds is reflected (through the action of the valence sum rule at 0 and H) in the differences in strength of the hydrogen bonds formed by the water molecules. These simple valence structures give predictions of bond lengths that agree within 0.02 8, with those found in Experimental results for solution structures of hydrated cations are more difficult to obtain, but the agreement between the predicted and observed solution structures of hydrated Co3-(Figure 11~)~~ -indicates that bond valences work as well in liquids as in I---/ / --1 / './ \ 050 / / \yo04 0 96/ H '\ I& cm~L04 09h ' HI0.54', / ,\o 04 \ /'\ (a) (b) (c) Figure 11Predicted bond valence structures for a, Be(H,0),2+; b, Mg(H20)62+;and c, CO(H,O),~+in aqueous solutions. Broken lines indicate the weaker (H . . 0)end of a hjsdrogen bond. Part of'the second hydration sphere is shown in c solids. The bond valence structure accounts for such diverse properties as the stability of the Co(Hz0)~~moiety (resulting from the large Co-0 valence), the strong bond between the first and second hydration spheres (predicted 2.63 A, observed 2.74 A) and the failure of the first hydration sphere to act as a hydrogen bond acceptor.7 Chemical Activity For some purposes it is convenient to consider the bonds as directed from the more electronegative to the more electropositive atom. An arrow placed on the bond then indicates the direction of net electron flow into the bond as the isolated ions are brought together or as an electron donor forms a bond with an electron acceptor. This convention is particularly useful when considering the properties of atoms or groups of atoms that can act both as electron donors and electron acceptors, since the sum of the bond valences between the group and its environ- ment (taking into account the direction of the bonds) must equal its formal charge.Thus water (Figure 12a) uses its hydrogen atoms to form two hydrogen bonds (electron acceptor bonds) with typical valences of 0.2 V.U. each and it must therefore act 8s an electron donor to the extent of 0.4 v.u., in order to maintain its valence neutrality. The difference in acidity between acetic acid and trifluoro- acetic acid is illustrated by the observed bond valence structures given in Figures 12b and 12c. The methyl protons act as electron acceptors (0.04 v.u.), but 3i S. K. Sikka and R. Chidambaram, Aria Cr~~st.,1969, B25, 310; G. Ferraris, D. W. Jones, and J. Yerkess, J.C.S. Dulion, 1973, 816. 38 R. Caminiti, G., Licheri, G. Piccaluga, and G. Pinna, J. Chein. Phps., 1976, 65, 3134. 373 Bond Valences-A Simple Structural Model for Inorganic Chemistry Figure 12 Typical bond valence strricture observed in crystals for a, H,O; b, CH,COO-;and c, CF3COO-the fluorine atoms act as electron donors (0.08 v.u.).In both cases the sum of the directed bond valences around the ion is close to one, but the ability of the 0 atoms to bond hydrogen, as indicated by its available valence, is very different, 0.52 against 0.36 V.U. Bond valence structures of this kind not only bring out the essential similarities between ionic and charge transfer bonding, but the out- wardly directed available valence of an atom provides a quantitative measure of its basicity at that site. It has been used by Rutherford and Robertson39 to demonstrate the highly basic nature of the nitrosy10 in 03S(NO)S033-.Figure 13 shows the correlation between Gutmann’s electron donicity40 for various organic bases and the available valence at the base group.In this figure the available valence has been calculated by assuming that H atoms bonded to the C adjacent to the base act as electron acceptors to the extent of 0.05 v.u., H atoms bonded to C two atoms from the base accept 0.025 v.u., all other H atoms bonded to C accept 0.01 V.U. and H atoms bonded to 0 or N accept 0.20 V.U. Bond valences not only give a measure of chemical activity but can be used in the prediction of reaction mechanisms. By studying the molecular distortions found in a large number of crystals, Murray-Rust, Biirgi, and dun it^^^ have indicated that the valence sum rule is satisfied by the series of intermediate configurations that two molecules pass through when they react.Bond valences can therefore be used to test the intermediate states of proposed reaction mechanisms. When two ions are brought together their bonding electrons move towards the centre of the bond, so that the shorter a bond becomes the more covalent it is. Brown and Shannon6 have shown that the valence of a typical bond between a 39 J. S. Rutherford and B. E. Robertson, Inorg. Chetti., 1975, 14, 2537. 40 V. Gutmann, Angew. Chenr. Internot. Edn, 1970, 9, 843. 374 Brown 0 0.50 AVAILABLE VALENCE Figure 13 Coniparison of rhe donicity and available valence of various organic bases cation and 0 is a direct measure of its covalent character as defined by Pauling41 and they suggest that the covalent character of an individual bond can be deter- mined from its valence.The V-0 bonds, which in the V1oOzs6-ion range from 1.593 to 2.313 8, have valences ranging from 1.82 to 0.28 V.U. and covalent characters ranging from 70 % to 25 %. Klemperer and Sh~m~~ have used the sums of covalences at the terminal 0 atoms in this ion to predict correctly which 0 atoms are most easily protonated in acid solution. All the external 0 atoms have the same available valence which is normally used to accept hydrogen bonds in solution, but those with the lowest covalence sum will be the most ionic, carry- ing the largest charge, and hence the most susceptible to electrophilic attack.8 Conclusions The chemical structure of inorganic compounds is at best explained only qualitatively by physical theories. The bond valence model provides a framework for discussing structural correlations without implying the physical mechanism that produces them. The ultimate understanding of chemical bonding must be sought not in simple covalent or electrostatic models but in complete quantum 41 L. Pauling, ‘The Nature of the Chemical Bond’, Second Edition, Cornell University Press, Ithaca, 1942, pp. 69ff. 42 W. G. Klemperer and W. Shum, J. Amer. Chcm. SOC..1977, 99, 3544. Bond Valences-A Simple Structural Model for Inorganic Chemistry mechanical solutions. It follows that the simple empirical laws described here must be hidden within these solutions.We ought to be able to find a quantum mechanical theorem that corresponds to the valence sum rule and this theorem must be relatively rigorous at least for those structures in which the rule works. Future work should involve the extension of the empirical rules on one hand and a search for simplifying theorems of quantum mechanics on the other. In the meantime, the bond valence model, even without a sound physical basis, is proving a powerful tool in the analysis and description of inorganic structure and, because of its simplicity, provides a useful vehicle for the teaching, understanding, and predicting of structure and reactivity in inorganic chemistry. I would like to thank Drs. M. Falk and K. Rubenson for their helpful discussions. 376

ISSN:0306-0012    

DOI:10.1039/CS9780700359

出版商:RSC    

年代:1978

数据来源: RSC

摘要:

Monoalky1triazenes By K. Vaughan DEPARTMENT OF CHEMISTRY, SAINT MARY’S UNIVERSITY, HALIFAX, NOVA SCOTIA, CANADA B3H 3C3 M. F. G. Stevens DEPARTMENT OF PHARMACY, UNIVERSITY OF ASTON IN BIRMINGHAM, GOSTA GREEN, BIRMINGHAM B4 7ET 1 Introduction Triazene, HN=N.NH2, is an important member of the family of open-chain nitrogen compounds and, although not isolable, has been postulated as an intermediate in the one-electron oxidation of hydrazine to ammonia and nitrogen during pulse radiolysis.1 Many derivatives of triazene have been synthesized, the most notable being the monoaryl-, 1,3-diaryl-, 1,3-dialkyl-, l-aryl-3-alkyl-[ArN=N. NHR; ‘monoalkyltriazenes’ (l)], l-aryl-3,3-dialkyl-[ArN=N .NR2; ‘dialkyltriazenes’ (2)], and N-hydroxy-[ArN=N .N(OH)R] triazenes.2 Recently reported novel triazenes are the lY2,3-triazabutadienes R ‘=N .N=N-Ar )334 and the closely related azidinium salts.5 Monoalkyl- (d triazenes merit separate consideration not only because of their abundant and distinctive chemistry, but also because of their important biological properties.The monoalkyltriazenes may be the active metabolites of the anti-tumour dialkyltriazenes (2) (Section 5). Synthetic applications of monoalkyltriazenes (Section 4) include esterification, deamination of primary amines, and the ability to act as bridging agents in co-ordination chemistry. The first sections, however, consider the synthetic and structural aspects of monoalkyltriazene chemistry. 2 Preparation A. Grignard Method.-The formation of a monoalkyltriazene by the reaction of a Grignard reagent with an aryl azide was first demonstrated by Dimroth6 E.Hayon and M. Simic, J. Amer. Chem. SOC.,1972,94,42. a H. Zollinger, ‘Azo and Diazo Chemistry’, Interscience, New York, 1961, 179; P. A. S. Smith, ‘Open-Chain Nitrogen Compounds’, Vol. 2, 1966, Benjamin, 336343. R. Ebisch and P. Niedermayer, 2.Chenr., 1975, 15, 143. ’V. E. Fanghanel, R. Hansel, W. Ortman, and J. Hohlfeld, J.prukt. Chem., 1975, 317,631 ; ibid., 1976, 318, 671. B. von Hellrung and H. Balli, Helv. Chim.Am, 1976,59, 140. ‘0.Dirnroth, Ber., 1903,36,909. Monoalkyltr iazenes HO ArN, + RMgX -[ArN(MgX).N=N.R] -k-Ar.NH.N=N.R (1) [equation (l)] and the importance of the reaction is well recognized.7 The Grignard method is quite versatile with respect to variations in the alkyl group and has been applied recently to the preparation of monoalkyltriazenes labelled with 14C in the alkyl group,* a novel ferrocenylmethyltria~ene,~and monovinyltriazenes labelled with 14C in the vinyl group.lO An unusual variation on the Grignard theme is the reaction of the azides (3a) and (3b) with the sulphur ylide (4) to give vinyl triazenes.ll Although mechanistically intriguing, this reaction is severely limited in scope since other azides give rise to dihydro- triazoles, not triazenes.0 IR-”, + 2CI-l2=SMe,-R-N=N.NH.CH=CH, + 2Me,SO (3) (4) a; R = p-0,N .C,H, b; R = 02 Unlike monoalkyltriazenes prepared by other routes, compounds generated from Grignard reagents are generally free from contamination by higher homologous open-chain nitrogen compounds.However, the Grignard method cannot be used when the aryl group contains substituents, e.g. CO2R and CN, which themselves react with the Grignard reagent. Fortunately in the majority of such instances the diazonium coupling method is a practical alternative. B. Diazonium Coupling.-N-Coupling of an aryl diazonium salt with a primary aliphatic amine is an obvious method of generating a monoalkyltriazene [equation (2)]. However, this apparently simple method, described in Organic Syntheses,12 is fraught with difficulties, not the least of which is the propensity of the triazene (1) to react further with diazonium ion to give the penta-azadiene (5).l3 The R ArN;--RNH,+ ArN=N.NHR ArN2+ I+ArN=N.N.N=NAr (2) (1) (5) recommended 12 method of separation of (1) from (5) is vacuum sublimation, the penta-azadiene usually being involatile, although exploitation of fractional solubilities in methanol has been suggested as a satisfactory a1ternati~e.l~ V. Zverina and M.Matrka, Chem. listy, 1969, 63, 51. A. N. Lotsova, T. N. Shatkina, and 0. A. Reutov, Doklady Akad. Nairk S.S.S.R., 1968, 183,1091. D. E. Bublitz, J. Organometallic Chem., 1970, 23, 225. lo C. C. Lee and E. C. F. KO, Canad.J. Chem., 1976,54, 3041. l1 G. Gaudiano, C. Ticozzi, A. Umani-Ronhi, and P. Bravo, Gazzetta, 1967, 97, 1411. 12 E. H. White, A. A. Baum, and D. E. Eitel, Org. Synrh., 1973, Coll. Vol. 5, 797.l3 G. F. Kolar, in ‘Mass Spectroscopy in Biochemistry and Medicine’, ed. Frigerio and Castagnoli, Raven Press, New York, 1974, 267. l4 C.S. Rondesvedt and S. J. Davis, J. Org. Chem., 1957, 22, 200. Vaughan and Stevens Proponents of sublimation appear to have overlooked the danger of explosion due to the thermal instability of triazenes and penta-azadiene~.~~ The tendency towards penta-azadiene formation is dependent on the nature of substituents in the aryl group; it has been shownl5 that diazonium salts with strongly electron-withdrawing substituents in the aryl group react with methyl- amine to give penta-azadiene-free triazenes. This specificity has been attributed to the resonance effect of the substituent in Ar, which reduces electron density at the N-3 nitrogen in (1) and prevents further coupling with diazonium ion.Notwithstanding the difficulty involved, the N-diazo coupling method has been widely used to prepare a number of monomethyltriazenes with different aryl groups.16J7 The method is also amenable to variations in the alkyl group, e.g. cyclohexyl-,ls glucopyranosyl-,19 galactopyranosyImethyl-,20and xylopyrano-syl-triazenes;21 significantly in all these cases the aryl group contained an electron-attracting substituent at the para-position. Although triazenes of type 1, with ortho-substituents in the aryl group, are susceptible to cyclization, a number of triazenes of this type have been isolated and characterized, such as the esters (6a),22 the nitrile (6b),23 and the analogous irnida~ole-(7a),~~ 1,2,3-tria~oIe-(7b),~~and pyrazole-carboxamides.26 (6) a; X = C0,R’ (7) a; Y = CH b; X = CN, R = CH,Ph b;Y =N C; X = COCH, A novel diazo-coupling of this type is the reaction of the tetrazotized biphenyl (8) with methylamine, which initially generates the triazenido diazonium salt (9).27This triazene undergoes two distinct cyclizations to give the unstable penta-azadienes (10) and (11) (Scheme 1).N-Diazo-coupling has also been used to prepare the novel 1,3-dialkyltriazene l5 7;. P. Ahern and K. Vaughan, J.C.S. Chern. Cotnni., 1973, 701. l6 M. Remes, J. Divis, V. Zverina, J. Marhold, and M. Matrka, Crsk. Farin., 1972, 21, 442; J. Arient and K. Panus, Coll. Czech. Chem. Conini., 1972, 37, 1900.T. P. Ahern, H. Fong, and K. Vaughan, Canad. J. Chem., 1977, 55, 1701. H. Maskill and M. C. Whiting, J.C.S. Perkin I/, 1976, 1462. 0.Larm, K. Larsson, and M. Wannong, Acta Chent. Scand. B, 1977, 31,475. 2o M. L. Sinnott and P. J. Smith, J.C.S. Cheni. Conini., 1976, 223. J. M. J. Tronchet and F. Rachidzadeh, Helv. Chint. Acta, 1976, 59, 2855. 22 R. J. LeBlanc and K. Vaughan, Canad. J. Chew., 1972, 50, 2544. 23 H. N. E. Stevens and M. F. G. Stelens, J. Chem. SOC.(C),1970, 765. 24 Y. F. Shealy and C. A. Krauth, J. Mediciti. Cheni., 1966,9, 34; Y. F. Shealy, C. A.Krauth, and C. A. O’Dell, J. Pharni. Sti.,1975, 64, 177. 25 Y. F. Shealy and C. A. O’Dell, J. Medicin. Chem.. 1966, 9, 733. 26 Y. F. Shealy and C. A. O’Dell, J. Phariv.Sci., 1971, 60, 554. 27 S. F. Gait, M. E. Peek, C. W. Rees. and R. C. Storr. J.C.S. Perkin I, 1974, 1248. 379 Monoalkyltriazenes (12) from the unusually stable cyclopropyldiazonium ion,28 and in the synthesis of triazabutadienes derived from iminoheterocyclic systems.4 Diazotization of aromatic amines, in the presence of a primary aliphatic amine, and coupling with phenolic couplers, is reported to give azo-pigments with improved dispersal characteristics; presumably a monoalkyltriazene is generated insif~.~~ R“ R“IDN=N--NHE~ArN=N-NHCH,Y R’-C-N=NNHAr ‘GoH (12) (13) a; Y = CO,H R-C=O $,/N\ArI b; k = CO,R c;Y = CN (144 (14b) Diazo-coupling with a-substituted alkylamines (NH~CHQY,where Y is electron-withdrawing) gives rise to triazenes of type (13), which vary widely in stability.Thus, coupling with a-amino acids initially gives the unstable triazenes (13a), which hydrolyse spontaneously to give the a-hydroxy acid, N2 and the associated arylamine [equation (3)], providing ArNs-N=&-CH .CO,H + OH, ArNH, + N, + HOCH.C02H (3)I I R R a valuable method of deamination of the amino a~id.~O A recent study 31 of the kinetics of the reaction has shown that there is no correlation between the rate of deamination and the pKa of the amino acid. By contrast the triazenes (1 3b) and (13c), derived from a-amino acid 28 W. Kirmse and U. Siepp, Chem. Ber., 1974, 107, 745. 2B M. Rajewski, J. Korzeniowska, and W. Siwek, Pol. P., 72853 (Chem. Abs., 1977, 86, 18349f). 30 H.Zahn, B. Wollerman, and 0.Waschka, Z. Physiol. Chem., 1953, 294, 100. 31 M. Remes, J. Divis, V. Zverina, and M. Matrka, Coll. Czech. Chem. Comnz., 1976,41,2566. 32 R. J. Baumgarten, J. Org. Chem., 1967,32,484. 33 J. F. McGarrity, J.C.S. Chem. Cornnt., 1974, 558. 380 Vaughan and Stevens and nitriles34 respectively, are quite stable and readily prepared. However, the keto-triazenes (14a), produced either by N-diazo-coupling with w-aminoaceto- phenone33 or by reaction of an aryl azide with a ketone,35 prefer to exist in the 5-hydroxy-d2-1,2,3-triazolineform (14b). Olsen has carried out a careful study of the triazene+triazoline tautomerism using i.r. and n.m.r. spectroscopy36 and has shown that some of the triazolines exist in solution as an equilibrium mixture of diastereomers, most likely interconvertible via triazenes.Although the triazene-esters (13b) have not shown any tendency to cyclize, the related dialkyl triazene (I 5) cyclizes in thionyl chloride to give a mesoionic triazolium- oxide.37 Conflicting claims have been made as to whether N-diazo-coupling occurs when a diazonium salt is treated with an amide; such a reaction, although inhibited by the weak nucleophilic character of the amido-nitrogen, would give rise to a monoacyltriazene (an oxidized form of a monoalkyltriazene). Odd0 and Algerin03~ reported the synthesis of 1 -phenyl-3-acetyl triazene by reaction of benzene diazonium chloride with acetamide in the presence of sodium hydroxide, whereas the same reaction in the presence of sodium acetate gave only 1,3-diphenyitria~ene.~~ Failure to react with acetamide was also recorded in the case of the diazonium salt from sulphanilic acid (-03s.C6H4N2+).39 Nevertheless, the formation of the N-formyltriazenes (p-X.CeH4N=N.NH . CHO) by N-diazo-coupling with formamide appears to be unequivocal, since +-these triazenes could be dehydrated to give diazoisocyanides (A~N=N-NEC)~@ Indeed, a number of monoacyltriazenes have been isolated in crystalline form by partial hydrolysis of hexa-azadienes41 and are quite stable. 1 -Phenyl-3- benzoyl-triazene has been obtained by reaction of phenylmagnesium bromide with benzoyl a~ide~~ and, more recently, by base-catalysed condensation of nitrosobenzene with ben~hydrazide.~3 34 T.A. Daniels, S. Sidi, and K. Vaughan, Canad.J. Chem., 1977, 55, 3751. 36 C. E. Olsen, Angew. Chem. Internat. Edn., 1974, 13, 349. 36 C. E. Olsen, Acta Chem. Scand., 1973,27,1987; C. E. Olsen and C. Pedersen, Acta Chem. Scand., 1973, 27, 2271. 37 T. L. Gilchrist and G. E. Gymer, Adv. Heterocvclic Chenr., 1974, 16, 58. 38 G. Odd0 and A. Algerino, Chem. Ber., 1936, 69, 279. 39 H. Eagle and P. Vickers, J. Biol. Chem., 1936, 114, 193. 40 T. Ignasiak, J. Suszko and B. Ignasiak, J.C.S. Perkin I, 1975, 2122. J. P. Horwitz and V. Grakauskas, J. Amer. Chem. SOC., 1957,79, 1249. IaA. Bertho, J. prakf. Chem., 1927, 116, 101. 43 S. Ito and T. Fukuyama, J. Org. Chem., 1971,36,2008. Monoalky ltriazenes Ne~ynov~~reported the reaction of diazonium salts with secondary amides and thioamides and found that 0-or S-arylation occurred, together with N-coupling to afford triazenium salts [e.g.ArN-N-N+ (Arl)=C(Me)OAr. AcO-1, whereas the analogous reaction with selenourea45 resulted in N- and Se-coupling giving triazenes of type, ArN--N. NHC(-=NH)Se. N--N. Ar. The novel cyanotriazenes (16), obtained by similar N-diazo coupling with cyanoguanidine, were found to be explosive when dry46 and did not lend them- selves to cyclization to iminobenzotriazines. This area of triazene chemistry continues to be of interest, as indicated by a recent comm~nication,~~ which describes new sulphonyl triazenes (ArN=N . NHS02Arl) as the N-diazo-coupling products of sulphonamides.3 Structure:Application of Physical Methods Detailed studies of the mass spectra of monoalkyltriazenes are few.13J7 Frequently observed fragmentation pathways (Scheme 2) are (a) loss of alkyl radical; (b) loss of alkylamino radical; and (c) loss of a diazoalkane moeity. Surprisingly direct loss of N2 (path d) is not a normal mode of fragmentation unless the aryl group contains an electron-withdrawing group.17 The presence of these groups also appears to promote breakage of the N-alkyl linkage (path a), which is not prominent in the spectra of 3-alkyl-1 -phenyltriazenes;13 the outstanding feature of the spectrum of 3-methyl-I-phenyltriazeneis a very intense peak at m/e 93 which has been assigned to the radical ion of aniline (17), arising by fission of the triazene group between N-1 and N-2, followed by hydrogen transfer.d ------__---R' [ArNH-R] f [Ar-N-N-N-R]:---a'[ArN,H]+--N* A[ArNH,] t -diazoalkane -RNH [ArNJ-(17) Mass Spectral Fragmentation of Monoalkyltriazenes Scheme 2 Applications of i.r. and n.m.r. spectroscopy to the structural elucidation of monoalkyl triazenes are more numerous and these techniques have proved invaluable in the elaboration of the tautomeric equilibrium (Ia) s(Ib). Indeed, the tautomeric structure of triazenes in general has been the subject of a long-standing debate.48 44 E. P. Nesynov and M. M. Besprozvannaya, Ukrain. khim. Zhur., 1974, 40, 182. (Chem. Abs., 1974, 80, 133326j); E. P. Nesynov, M. M. Besprozvannaya, and P. S. Pel'kis, Dopovidi Akari.Nairk Ukrain. R.S.R.,Ser. B., 1969, 31, 27 (Chem. Abs., 1969, 71,30134e). E. P. Nesynov and T. F. Aldokhina, Zhur. obshchei Khim., 1976, 46, 1098. (Chem. Abs., 1976,85,77803b). 46 S. M. MacKenzie and M. F. G. Stevens, J. Chem. SOC.(C),1970, 2298. R. Kreher and R. Halpaap, Tetrahedron Letters, 1977, 3147. ** T. W. Campbell and B. F. Day, Chem. Rev., 1951,48,299. 382 Vaughan and Stevens The i.r. spectra of monoalkyltriazenes generally show two NH stretching vibration bands, one at 3480-3440 cm.-l assigned to tautometer (Ia), and the other near 3338 cm-1 assigned to (Ib).49 Unambiguous assignment of these bands was made possible by observing the shift of the low frequency band to lower frequency when l5N is introduced at N-1.The relative intensities of these bands are influenced by the substituent (X), implying that the substituent influences the tautomeric equilibrium; electron-withdrawing substituents shift the equili- brium in favour of form (Ib). A similar substituent effect has been observed, also by i.r. measurement, in the tautomerism of 1,3-diaryltriazene~.~~ From the temperature dependence of the intensity ratio of the i.r. bands, AH of the tautomeric reaction (Ia)+(Ib) was estimated to be 0.3 kcal mol-1,49 which is in agreement with the vaIue obtained for diaryltriazenes. Low temperature p.m.r. spectra of the p-tolyltriazenes (I; X = p-Me, R = Me or PhCHz) in [2Hs]acetone solution51 exhibit two resonances for the a-protons in the R-group; a low-field singlet was assigned to form (Ib) and the high-field doublet, arising from coupling of a-protons with the adjacent NH proton, was assigned to form (Ia).By careful measurement of integrations of N-alkyl group proton signals, KT values for these triazenes were evaluated as 0.46 (R = PhCH2) (-40 "C)and 0.1 1 (R = Me) (-55 "C); thus the proportion of tautomers (Ia) and (Ib) of 3-methyl-1-p-tolyltriazeneis approximately 9:l at -55 "C in acetone solution. AH for the tautomeric equilibrium of the methyl-p-tolyltriazene was estimated to be 1.1 kcal mol-l; the analogous measurement for thep-nitro- phenyltriazenido-sugars (I; X = p-NO2, R = Xylopyr.mosy1) was even greater, 3.03 kcal mol-1.21 These AH values are significantly greater than those estimated from i.r.measurements, but no explanation has been offered for the discrepancy. Further lH n.m.r. studies with monoalkyltriazenes have shown that the tautomerism is both solvent-dependent and influenced strongly by substituents in the aryl group. For example, the lH n.m.r. spectrum of 3-methyl-1-p-tolyl- triazene, measured in dichloromethane at -65 OC52 or in chloroform at -55 OC,53 exhibits only one resonance, a doublet, for the N-methyl protons, indicating that only tautomer (Ia) is detectable under these conditions. However, tautomer (Ib) of the same p-tolytriazene has been detected in CDC13 by l3C 4B D. Hadzi and J. Jan, Spectroscopy Letters, 1968, 1, 139. S. Weckherlin and W. Luttke, TetrahedronLetters, 1964, 171 1 ;T. Mitsuhashi, Y.Osamura, and 0. Simamura, Tetrahedron Letters, 1965, 2593. 51 R. Curci and V. Lucchini, Spectroscopy Letters, 1973, 6, 293. st N. S. Isaacs and E. Rannala, J.C.S.Perkin II, 1974, 899. 63 K. Vaughan, J.C.S.Perkin II, 1977, 17. 383 Monoalkyltriazenes n.m.r. spectroscopy5* and the failure to detect the lH n.m.r. signals of form (Ib) of this triazene has been ascribedS5 to low intensities of these signals which may well be considerably broadened. However, clear distinction of tautomers (Ia) and (Ib) can be seen in the 1H n.m.r. spectrum when the aryl group contains electron-withdrawing groups.53 Thus the lH n.m.r. spectrum of l-(g-cyanophenyl)-3-methyltriazenein CDC13 at -30 “C shows the singlet methyl resonance (8 3.62) and the doublet methyl resonance (8 3.22) in approximately equal intensities, indicating that the presence of the cyano-group shifts the tautomeric equilibrium towards (Ib), in agreement with the i.r.results previously discussed. The 1H n.m.r. spectrum of 1-(o-ethoxy- carbonylphenyl)-3-methyltriazene(6a; R = Me) exhibits only one N-methyl resonance, which is a singlet (8 3.57). The e~planation~~ of this observation is that intramolecular hydrogen bonding with the o-ethoxycarbonyl group holds the triazene in form (Ib) (see 18). The assignments of singlet and doublet methyl resonances in the spectra of monoalkyltriazenes to tautomers (Ib) and (Ia) respectively has been confirmed by chemical shift correlation with model compounds. The well known 3,3- dimethyltriazenes (2; R = Me) serve as an excellent model for tautomer Ia of the monoalkyltriazenes; the observed chemical shifts are 83.23j6 and 83.17- 3.22.S2e53 l-Methyltriazenes (e.g.19), which would be useful model compounds for tautomer (Ib), were until recently unreported; however Pochinok57 has re- ported a chemical shift at 83.98 for the l-methyl-group in 1,3-dimethyl-3-phenyI-triazene (19) although the source or method of preparation of (19) is not clear. Other mode compounds of potential application are the 1,3-dimethyl-3-silyltri-azenes (2O),5* which undergo silyl-group migration analogous to the prototropy in monoalkyl- and diaryl-triazenes. A surprising result was obtained when optically active a-methylbenzylamine was coupled with a benzene diazonium salt;59 the resulting triazene (21) was 64 K.Albert, K-M. Dangel, A. Rieker, H. Iwamura, and Y. Imahashi, B/(//.Ctieriz. SOC. Japan, 1976,49,2537. 55 H. Iwamura, K. Albert, and A. Rieker, Tetrahedron Letters, 1976, 2627. 56 Sadtler Index of N.M.R.Spectra, 1968,4499M. 37 V. Ya Pochinok, Ukrain. khirn. Zhur., 1977, 43, 180. (Cherii.Abs., 1977, 86, 188927j). 58 N. Wiberg and H. J. Pracht, Chern. Ber., 1972, 105, 1377. 1388. 58 A. P. Terent’ev, E. G. Rukhadze, T. V. Ershova, and S. A. Burkhova, Zhur. org. Ktiirii., 1968, 4, 125. (Chern. Ah., 1968, 68, 86923s). 384 Vaughan and Stevens optically inactive. This observation was accounted for by the extended tauto- merism [(21a)+(21 b) +(21c)] whereby the mobility of the a-proton causes racemization.The driving force for the shift to tautomer (21c) may be the conjugation of the C=N group with the benzene ring. Analogous reaction of optically active N-ethyl-a-methyl-benzylaminegave the optically active dialkyl- * triazene [PhN=N .N(CzH5). CHMe .Ph], which cannot undergo prototropy. A similar phenomenon has been observed when diazotized methyl anthranilate was coupled to L-phenylalanine ethyl ester to afford an optically inactive 3-alkylbenzotriazinone.60The optical inactivity of the product was rationalized by the formation of the intermediate triazene [6a; R = -CH(C02Et)CH2Ph], in which the a-CH group is acidic relative to that in simple 3-alkyl triazenes (e.g. 21) and which could readily ionize under the conditions used to give a trigonal planar, resonance stabilized anion.Reprotonation and cyclization would then afford the racemic benzotriazinone. hl e Me c PhCHMe.NH, PhN,-I 4PhN=N-NtlCHPh PhNH-N=N-CHPh I (2I a) (Zlb) (2I c) 4 Reactions An aspect of triazene chemistry of quite recent origin, but of growing importance, is their ability to act as ligands in co-ordination compounds, although one could look upon this property as an extension of their nucleophilic character, and this topic is treated as such in this review. However this Section is primarily con- cerned with the synthetic applications of monoalkyltriazenes, particularly with respect to the deamination of primary amines, the esterification of carboxylic acids and the formation of 1,2,3-benzotriazines. The first two types of reaction are manifestations of the facile release of the alkyl group during degradation of the triazene, whereas cyclization reactions require strict control over degradative side-reactions to optimize the yield of cyclic triazine.Thus, a sound knowledge of the degradative properties of monoalkyltriazenes will aid the synthetic chemist and we begin this section with an appraisal of recent studies of the acid-catalysed and thermolytic reactions of monoalkyltriazenes. A. Acid-catalysed Degradation.-(i) Protolysis. Degradation of monoalkyl-triazenes in protic media leads to products arising from protonation mainly at O0 A. C. Mair and M. F. G. Stevens, J. Clzeiit. SOC.(C),1971, 23 17.Monoalkyltriazenes N-3 of the tautomeric form Ib (Scheme 3, path ii). Thus hydrolysis of l-phenyl- 3-methyltriazene in water gives a mixture of aniline and methanol, but not pheno1.6?61" Significant evidence for the involvement of both tautotners (la, Ib) is adduced by the formation of diaryltriazenes during degradation of mono-meth~1-l~and monobutyl-61b triazenes in protic media; the diaryltriazene can arise by coupling of the diazonium ion Ar.N2+ (path i) with the arylamine ArNHz (path ii). The formation of 1,3-diaryl-3-alkyltriazenesin these degrada- tions is evidence for the formation of secondary amines by loss of nitrogen (path iii) before diazo-coupling. Matrka et established a significant cor- relation between the rate of protolysis and the o-value of the substituent in the aryl group; in a predictable manner, electron-withdrawing groups slow down the protolysis.The greater reactivity of the non-conjugated tautomer (Ib) has been suggested by a kinetic study of the hydrolysis of the vinyl ethers of hydroxy- ethyltriazenes (PhN=N . NHCH~CH~OCHTCHZ).~~ Evidence of carbonium-ion involvement (path iia, Scheme 3) in the mechanism of protolysis has been obtained from the decomposition of 1-phenyl-3-n-pro-pyltriazene in aqueous sulphuric acid; analysis of the resulting mixture of alcohols showed a significant amount of 2-propanol, in addition to I-propanol, implying formation and rearrangement of the n-propyl cation.8 Indeed acidic hydrolysis of monovinyltriazenes has been used as a method of generating and Ar-N=N-NHR<-Ar-NH-N=N-R iii/-N2 HtIr ii"if t t Ar -N =N -N H ,R Ar-NH, -N =N-R ArNHRb 41 YArNH, + J R-N, ---Hiib +>diazoalkaneArN,? + RNH, ArN =N -N-Ar iia 1 -N,I R R-LR-X ArN=N-NHAr Scheme 3 studying vinyl cations.1,2-Aryl shifts across the double bond of the vinyl group were observed in the case of unsymmetrically substituted vinyltriazenes, e.g. 3-(1-phenyl-2,2-di-p-tolylvinyl)-l-phenyltriazenedecomposes in acetic acid at 61 (a) V. Y. Andakushkin, B. A. Dolgoplosk, and I. I. Radchenko, Zhur. obshchei Khim., 1956, 26, 2972 (Chem. Ah., 1957, 51, 8674a); (b) V. H. Oelschlager and H. Blume, Arzneim. Forsch., 1976, 26, 303. V. Zverina, M. Remes, J. Divis, J. Marhold, and M.Matrka, Cull. Czech. Chem. Confm., 1973,38,251. 63 V. S. Sukhinin and A. P. Kozlov. Zhw. ~bshcheiKhint., 1975, 45, 145. (Cheni. Abs., 1975, 82,97332~). Vaughan and Stevens room temperature in a matter of seconds giving a 20% yield of the rearranged decomposition product 2-phenyl-l,2-di-p-tolylvinyl Recent studies by a~etate.6~ Lee and Ko10 with 1*C labelled vinyltriazenes have shown that about 38% scrambling of the label, attributable to 1,2-anisyl shifts, can occur but that analogous 1,2-phenyl shift does not operate. Triazenes of type (13b) do not degrade in acid to give carbonium ions but instead give stable diazoacetates [path (iib), Scheme 31, e.g. NKHzC02Et ;32 this reaction is also ba~e-catalysed.~~ (ii) Reaction with Lewis Acids.Monoalkyltriazenes degrade when in contact with Lewis Acids; consequently, purification by standard chromatographic methods is not always possible. The reaction of 1-phenyl-3-alkyltriazeneswith aluminium trihalides in benzene affords N-alkylanilines, with nitrogen evolution, together with alkylbenzenes from alkylation of the solvent,65 whereas the esters (6a) decomposed over alumina in benzene, to give the primary arylamine (o-ROK. CsH4NH2) in almost 80 yield.22 The facility with which the esters (6a) convert to the arylamine is possibly a consequence of the preference of these triazenes to exist as the unconjugated tautomer (ArNH. N=NR).j3 The specific involvement of this tautomeric form is also implicated by the products observed when triazene decomposition is initiated by silica gel in benzene.66 These authors favour a carbonium-ion pair mechanism [equation (4)] on grounds that: ArNH.N=N-R-ArNH N,R -ArNH.R-i N, (4)I (ion pair) (a) R migrates to N with retention of configuration; (b)The reaction is intramolecular and no crossed products are formed; (c) The reaction is suppressed entirely in ethanol. Furthermore, analogous 1,3-diphenyl- and 3,3-dialkyltriazenes do not decompose under the same con- ditions, suggesting that the formation of tautomer (Ib) is essential for reaction to occur.B. Thermolysk-The formation of secondary amines in the reaction of diazomethane with primary amine~~~ may well be the first reported example of thermal breakdown of a monoalkyltriazene; although this reaction has found little synthetic application, the possible intermediacy of a triazene in the con- version [equation (5)] has not been discounted.CHZN, + RNH, +[RNH-N=N-CHJ -RNHCH, + N, (5) 6-1 W. M. Jones and F. W. Miller, J. Anier. Clic~tii.Soc., 1967, 89, 1960. 65 R. Kreher and K. Goth, Z. Natiirforscfi., 1976, 31b, 217. 88 M. Kawanisi, I. Otani, and H. Nozacki, Tetralietlron Letters, 1968, 5575. 67 L. I. Smith, Cheiii. Rev.. 1938, 23, 202. 387 Monoalkyltriazenes Cserhegyi et a1.68followed the kinetics of the pyrolysis of 3-methyl-1-phenyl-triazene in unlabelled and in tritiated cyclohexane by measurement of the CH4:N2 ratio in the gaseous products. Although a kinetic isotope effect was observed for hydrogen abstraction, no firm conclusions were apparent regarding the mechanism of the reaction.Recent ‘H- and 13C-CIDNP investigation^^^ of the thermolysis of 3-alkyl-1 -p-tolyltriazenes at 125-1 35 “C have shown that although tautomer (Ia) is predominant in the tautomeric mixture, the major products (ArNHz, ArNHR, and RH) arise from homolysis of tautomer (Ib) [equation (6)]: A N2 (1)(Ib) +ArNH t N=N.R-ArNH.R +ArNHR (6) (22) (jj)i ArNH, + R-H The CIDNP results show polarization of the nuclei in ArNH2, ArNHR, RH, and in minor products, in agreement with the formation of an intermediate arylaminyl-alkyl radical pair (22), which may collapse (i) to the cage recom- bination products, ArNH R and ring-alkylated anilines, or undergo H-abstraction (ii) to give the escape products, ArNHz and RH.Thermolysis of dialkyltriazenes, which cannot tautomerize to a form like (Ib), follows a different route; for example, thermal breakdown of the novel azo-aziridines (23) gives phenyl azide and alkenes, specifically with retention of configuration in the alkene.69 solvent (24) C. Deamination ofPrimary Amines.-Conventional methods for the deamination of primary aliphatic amines include the nitrous acid-amine reaction and the decomposition of nitrosoamides. The use of triazenes as intermediates in deamination, which is a consequence of the facile degradation of monoalkyl-triazenes discussed earlier, has been recognized for some time30p70 and has been used with sufficient frequency to be regarded now as almost conventional.The triazene method is useful for the conversion of amines into esters, alkyl halides, ethers and related derivatives, and yields are usually superior to other methods. Although the triazene method of deamination leads to considerable race-mization in the product RX, suggesting that the carbonium ion mechanism 6* A. Cserhegyi, G. Szentgyorggyi, and 0. Bobis, Magy. Kem. Foiy., 1971, 77, 607 (Chern. Ah., 1972, 76, 71793r). 89 R. E. Clark and R. D. Clark, J. Org. Chem., 1977,42, 1 136. 70 E. H. White and H. Scherrer, TefralzeclronLefters, 1961, 758. Vaughan and Stevens (Scheme 3, path iia) is dominant, later studies by White et al.71 showed that the carbonium ion could be intercepted by the arylamine.The arylamine substitution products, e.g.RNHC6H4Y and NHzCsH3Y R, retain configuration in the R-group which suggests that the carbonium ion is formed in the same solvent cage with the counterion and the aromatic amine, i.e. [24; R = -CH(CHs)Ph], related to that proposed for the nitrosoamide decompo~ition.~~ The triazene method does offer some advantage over other methods of deamination. For example, Storesund and Whiting73 were unable to prepare the nitrosoamide derived from 2-amino-adamantane. However, the related triazene (25), although unstable when crystalline, could be obtained with difficulty and was used to study the acetolysis of the 2-adamantyl group. Simil- arly cis-and trans-4-t-butylcyclohexyltriazeneswere used in the deamination of cyclohexylamines in order to study the derived carbonium ions.18 A more exotic application of the method is the use of the P-D-galactopyranosylmethyl-p-nitrophenyl-triazene (26) to generate ‘hot’ carbonium ions which can act as ‘affinity labels’ in blocking the active site of liganding enzymes (in particular E.coli fl-galactosidase).20 HO(CH,),CH( N HCO.,CMe,)CO,CH,Ph (28) D. Esterification.-By virtue of the facility with which monoalkyltriazenes release the alkyl fragment, alkylation of organic species is perhaps their most important synthetic function. In particular, the esterification of carboxylic acids has been applied to derivitization of benzoic acids,l2 fatty acids,7* insect hormone derivatives [(e.g. (27)3,75 biiir~bin,~6pyrrole carboxylic acids,77 N-blocked a-amino a~ids,~8 polymethacrylic acid,79 and other polymeric carboxylic acids. *O The use of a monobenzyltriazene for the synthesis of an N-blocked homoserine benzyl ester (28) was found78 to be the best compromise to overcome the 71 E.H. White, H. Maskill, D. J. Woodcock, and M. A. Schroeder, Tetrahedron Letters, 1969, 1713. 72 E. H. White and C. Aufdermarsh, J. Amer. Chem. Soc., 1961, 83. 1179. 73 El. J. Storesund and M. C. Whiting, J.C.S. Perkin If, 1975, 1452. 74 C. A. Henrick, U.S. P. 3873586. (Chem.Abs., 1975,82, 170098~). 75 C. A. Henrick, W. E. Willy, J. W. Baum, T. A. Baer, B. A. Garcia, T. A. Mastre, and S. M. Chang, J. Org. Chem., 1975,40, 1. 76 D. W. Hutchinson, B. Johnson, and A.J. Knell, Biochem. J., 1973, 133, 493. 77 E. Campaigne and G. M. Shutske, J. Heterocyclic Chem., 1975, 12, 67. 76 T. Fickel and C. Gilvarg, J. Org. Chem., 1973,38, 1421. R. Blumstein, G. J. Murphy, A. Blumstein, and A. C. Watterson, J. Polymer. Sci., Polymer Letters Edn., 1973, 11, 21. L. H. Cohen, J. Polymer Sci., Polymer Letters Edn., 1976, 14,7. 389 Monoalkyltriazenes problems of the interfering lactonization of N-blocked homoserine in acid media and of the benzyl ester in the presence of base. However the aromatic amine, which is a by-product of esterification by triazenes, can induce base-catalysed lactonization and the neutral conditions employed in the triazene method do not entirely eliminate cyclization. Application of the triazene method to the esterification of polymethacrylic acid79 eliminated the possibility of polyethylene contamination, thus leaving the polymer tactility unchanged.Other advantages of this method, compared to the conventional method using diazomethane, were seen to be: (a) the triazenes are non-explosive under the conditions used; (b)they are readily available (some commercially), are reasonably stable and do not have to be freshly prepared (cf. diazomethane), and (c) the triazene method is faster. However the advantages of monoalkyltriazenes must be weighed against their potent carcinogenic activity (see Section 5). Moreover, comparisons of the triazene method with other alkylation procedures are not always favourable.80 A kinetic study of the reactions of a series of 3-alkyl-1-aryltriazenes with ben- zoic acids supports a mechanism in which proton transfer and departure of an alkyl cation are rate determining and synchronous [equation (7)].52 0 II Ar-N= N-N H--C-ph --+-ArN=N-NH, + (R+.PhCO,-) L'R (29)1 IArNH, + N, PhC02R (7) The occurrence of Wagner-Meerwein rearrangements, the extent of which were dependent on solvent polarity, in all cases examined, supported the carbonium ion-carboxylate ion-pair (29) hypothesis, whereas White et al.71 dismissed the suggestion of intimate carbonium ion-carboxylate ion-pairs on the basis of stereochemical studies (see earlier discussion, Section 4-C).Clearly there is an unresolved contradiction here, which may have arisen because of the different solvents employed.Alkylation of other organic functional groups by triazenes has not been so thoroughly studied. Succinimide and phenols are methylated by monomethyl- triazenes, but the alkylation of alcohols and mercaptans requires aluminium alcoholate catalysis.81 Evidently, alkylation by monoalkyltriazenes requires acid catalysis, but further studies appear to be necessary in this area. V. Ya. Pochinok and L. P. Limarenko, Ukrain. khim. Zhirr., 1955, 21, 496. (Chem. Abs., 1956, 50, 5549i); V. Ya. Pochinok and 0. I. Shevchenko, Ukrain. khim. Zhur., 1954, 20, 289. (Chem. Abs., 1956, 50, 272e). Vaughan and Stevens E. Cyclization Reactions--Although monoalkyl triazenes possess the necessary structural requirement to act as precursors for cyclic 1,2,3-triazines, there are no reported examples of the conversion of a monoalkyltriazene to a monocyclic triazine.However the synthesis of 1,2,3-benzotriazine derivatives from ortho-substituted 3-alkyl-1-aryltriazenes (6) is well documented and quite versatile. Although the o-alkoxycarbonyl derivatives (6a) can be isolated,22**2 cyclization of these esters to benzotriazinones (30a) is extremely facile and is sometimes observed spontaneo~sly.~~~83 The conversion (6a)+(30a) is catalysed by protic media or by alumina in organic solvents. Cyclization of the a-acetylphenyl- triazenes (6c) is also catalyseci by alumina and affords 4-methylene-benzo- triazines (30~).*~ Whereas many 1-(o-cyanophenyl)-3-aryltriazenes undergo cyclization to 4-iminobenzotriazines,*5the only unequivocal example of a monoalkyltriazene undergoing such cyclization is that of the benzyl derivative (6b) which produces the unusually stable 4-imino-3-alkylbenzotriazine(30b).23 The analogous mono- methyl triazenoimidazole (32) is possibly an intermediate in the conversion of the chlorotriazine (31) to the iminotriazine (33) by the successive action of water and me thy lam i ne.86 X c1 (30) a;X =O b; X = NH, R = CH2Ph C; X = CH, F.Nucleophilic Properties.-The nucleophilic character of a monoalkyltriazene is at once evident from the fact that formation of penta-azadienes often accom- panies triazene formation during N-diazo coupling. Indeed, Dimroth87 showed that the same unsymmetrical penta-azadiene (34) arises from two different coupling reactions of diazonium salt with monomethyltriazene [equation (S)]: W.F. Gilmore and R. N. Clark, J. Heterocyclic Chem., 1969, 6, 809. 83 E. van Heyningen, J. Amer. Chem. Soc., 1955,77, 6562. 84 H. Fong and K. Vaughan, Cunud.J. Chem., 1975,53,3714. M. S. S. Siddiqui and M. F. G. Stevens, J.C.S.Perkin I, 1974, 609. V. S. Mokrushin, V. I. Ofiiserov, T. V. Rapakova, A. G. Tsaur, and 2. V. Pushkareva, Khim. geterotsikl. Soedinenu, 1976, 556. (Chem. Abs., 1976, 85,46590a). 87 0.Dimroth, Ber., 1907,40,2390. 3 391 Monoalkyltriazenes MeI ArN,+ + Ar1N=NNHMe+ ArN=N.N.N=NAr'+ ArN=N.NHMe + AT",+ (34) (8) Significantly, diazo-coupling occurs at the N-atom adjacent to the methyl, not the aryl, group.The greater nucleophilicity of N-3 is also evident during acetyl- ation with acetic anhydride, which affords 3-acyl derivatives.** Recent studies reinforce these observations. Reaction of 3-methyl-1-p-tolyl- triazene with thioketenes affords the S-aryl thioimide (36), via the thioacylated triazene (35)89 (Scheme 4;Ar = p-MeCsH4). An analogous reaction takes place when a thioketene is treated with a diaryltriazene but not with l-aryl-3,3-dimethyltriazenes, which instead induce dimerization of the thioketene. Although the thioacyl triazene (35) is not stable, the analogous reaction of monoalkyltri- azene with is0 thiocyana tes affords stable N-arylazo thioureas (3 7), which have useful miticidal properties.90 Me n I Ar-N=N-NHMe + (CF,),C=C=S +Ar-N=N-N-C-CH(CF,), II S (35) NArN =N-N /R II /Me Ark/ 'C-NHR C4 /\S Ar-S' CH(CF,), (37) (36) Scheme 4 G.Complex Formation.-The nucleophilic character of triazenes lends itself well to the formation of triazenido-metal bonds, which offer a variety of bonding arrangements in transition metal (M) complexes. Three fundamental co-ordi- nation modes have been identified. Although the monodentate mode (38) has been disputed, two instances of complex formation by diaryltriazenes with platinumg1 a* 0.Dimroth, Ber., 1905,38,670. Is M. S. Raasch, J. Org. Chem., 1972, 37, 1347. so S. Kano, E. Takeuchi, T. Noguchi, and M. Asada, Jap. P. 74/28188. (Chem. Abs., 1975, 82, 155909r). Dl L. D. Brown and J.A. Ibers, J. Amer. Chem. SOC.,1976,98, 1957. Vaughan and Stevens and palladium92 in this mode have been verified by X-ray analysis. The bidentate mode of complexation (chelation) (39) occurs more commonly, as in the ruthenium91 and cobalt93 complexes of lY3-diphenyltriazene. The third type of triazene complex (40) has the triazenido-group acting as a bridging ligand between two metal atoms, which are sometimes bonded to each other; this mode has been characterized for complexes of 1,3-dimethyltriazene with rhodium/copper94 and for diaryltriazenes with copper95 and palladium.96 I /MI-N M-N-N=N-M' i'N )" Not all of these co-ordination modes have been observed in complexes of monoalkyltriazenes. Chelation of molybdenum and tungsten by 1-(p-chloro- phenyl)-3-iso-propyltriazene has been rep0rted,~7 and these complexes were unusual in being non-fluxional, unlike diary1 complexes.3-Methyl-1 -p-tolyl- triazene serves as a bridging ligand in complexes with silver/rhodium, silver/ indium, and mercury/iridium.gY Complexes of 3-t-butyl-1 -aryltriazenes with AgI, HgI, HgII, and CuII are known,99 but their structures have not been defined. The in situ formation of an aluminium complex of 3-methyl-1 -phenyltriazene has been reported,lOo but this 'complex' may be simply a salt of A13+. The monodentate mode of complex formation has not been observed for mono- alkyltriazenes. 5 The Role of Monoalkyltriazenes in the Metabolism of Anti-tumour Dialkyl- triazenes Monoalkyltriazenes are cytotoxic agents and have carcinogenic,lo1 mutagenic,lo2 teratogenic,l03 anti-fungalYl04 and anti-tumour activity.lo5 The intense interest in B2 G.Bombieri, A. Tmmirzi, and L. Toniolo, Inorg. Chem., 1976, 15, 2428. 93 M. Corbett and B. F. Hoskins, Chem. Cornm., 1968, 1602. @( J. Kuyper, P. I. van Vliet, and K. Vrieze, J. Organornetallic Chem., 1975, 96, 289. M. Corbett, B. F. Hoskins, N. J. McLeod, and B. P. O'day, Austral. J. Chern., 1975, 28, 2377. Be S. C. de Sanctis, L. Tonido, T. Boschi, and G. Deganollo, Inorg. Chim. Acta, 1975, 12, 251. s7 E. Pfieffer, J. Kuyper, and K. Vrieze, J. Organometallic Chem., 1976, 105, 371. s8 P. I. van Vliet, J. Kuyper, and K. Vrieze, J. Organomrtnllic Chern., 1976, 122,99. gg E. G. Rukhadze, T. V.Ershova, S. A. Fedorova, and A. P. Terent'ev, Zhur. obshchei Khim., 1969,39, 303. (Chem. Abs., 1969, 71, 3451~). loo F. E. Brinkman, H. S. Haiss, and R. A. Robb, U.S. P. 3386985. (Chem. Ah., 1968, 69, 434 1Sf). lol R. Preussman, A. von Hodenberg and H. Hengy, Biochem. Pharmacol., 1969, 18, I. lo2 T. Ong and F. J. de Serres, Mutation Research, 1971, 13, 276. H. Druckrey, S. Ivankovic, R. Preussmann, and U. Brunner, Experientia, 1967, 23, 1042. lo4 I. V. Zlochevskaya, E. G. Rukhadze, T. S. Bobkoval, and L. N. Chekunova, Vestn. Mosk. Univ., Biol. Pochvoved, 1973, 28, 42. (Chern. Abs., 1974, 81, 21539~). lo' F. A. Schmid and D. J. Hutchinson, Cancer Research, 1974, 34, 1671. Monoalky ltriazenes these triazenes revolves around their (possible) role as the active metabolites and proximate carcinogens of the anti-tumour 3,3-dialkyl triazenes and interpretation of their mode of action offers a challenge to the chemist interested in reactive chemical species.Anti-tumour activity in dimethyltriazenes was first reported more than twenty years ago,lo6 but their clinical use originated from the search for inhibitors of de novo purine synthesis.lo7Jo8 5-(3,3-Dimethyl-l -triazeno)imi- dazole-4-carboxamide (DTIC; ‘Dacarbazinem’) (4 1a), a product of this search, is clinically useful in the treatment of malignant melanoma.10g It has been suggested that DTIC is a stabilized form of Diazo-IC (42) and that the diazo derivative is liberated in target cells.l1° The inhibitory activity of DTIC against bacterial systems’ll and against Chinese hamster ovary cells and malignant melanoma cells in is certainly enhanced ir.the presence of light which transforms DTIC to Diazo-IC and thence to 2-azahypoxanthine (43). The comparative biological activities of these three imidazole derivatives have been reviewed.115 However, simple 3,3-dialkyl-l-aryltriazenes(44)not bearing an imidazole moeity are equally efiective as DTIC as anti-tumour agents. The diazonium ions derived from these compounds are not always active116 although CONH, 0 NfNX ,CH3 (qcoNH*H N=N-N N N,+ ‘R (41) a; R = Me (DTIC) (43) b; R = H (MTIC) QN=N--N\R, /R X (44) a; R = R‘ = Me b; R = Me, R’ = Et C;R = R’ = Et d; R = Me, R’ = But looD. A. Clarke, R. K. Barclay, C.C. Stock, and C. S. Rondestvedt. Proc. Soc. Exp. Biol. Med., 1955, 90, 484. Io7 Y.F. Shealy, J. A. Montgomery, and W. R. Laster, Biocheni. Pharmacol., 1962, 11, 674. Y.F. Shealy,J. Pharm. Sci., 1970, 59, 1533. S. K. Carter and M. A. Friedman, European J. Cancer, 1972, 8, 85. 110 K. Hano, A. Akashi, I. Yamamoto, S. Narumi, and H. Iwata, Gann., 1968,59, 207. ll1 P. P. Saunders and G. A. Schultz, Biochem. Pharmacol., 1972, 21, 2065. 11* A. H. Gerulath and Ti Li Loo, Biochern. Pharmacol., 1972,21, 2335. 113 A. H. Gerulath, S. C. Barranco, and R. M. Humphrey, Cancer Research, 1974, 34, 1921. 114 P. P. Saunders and L.-Y. Chao, Cancer Research, 1974, 34, 2464. 115 M. F. G. Stevens, Progr. Medicin. Chem., 1976, 13,205. R. C. S. Audette, T.A. Connors, H. G. Mandel, K. Merai, and W. C. J. Ross, Biochem. Pharmacol., 1973, 22, 1855. Vaughan and Stevens this has been attributed to unfavourable transport characteristics of the diazonium compounds.l17 The recognition of a correlation between the rate of hydrolysis of dimethyl- triazenes and the ability to induce local tumours at the site of administration,l18 lends support to the hypothesis that triazenes may be transport forms of reactive diazonium species [Scheme 5, route (i)]. ,CH3 ,CH,OH ArN,-' + HN(CH3)2,iAr-N=N-N Oxidation Ar-N=N-N \ 'CH, " CH,(45) -CH20 NU-H \ ArNH, + N, + CH,Nu ArNH-NN= N-CH, L-7. Ar-N= N-NHCH, Scheme 5 Set against these results, there is now compelling evidence that an alternative mechanism-enzymic activation-is responsible for the systemic carcinogenic activity and, possibly, anti-tumour activity of dialkyltriazenes, and that alkylating metabolites formed following oxidative N-dealkylation are the bio- active species [Scheme 5, route ($1.Considerable support for this hypothesis has been obtained from experiments with isotopically labelled triazenes. Thus incorporation of radioactivity in the DNA of human melanoma cells was observed after incubation with DTIC labelled with 14C at the side-chain methyl groups.llg The primary site of methyl- ation has been identified as N-7 of guanine residues from in vivo studies with labelled DTIC in man and 1-phenyi-3,3-[14C]-dimethyltriazenein rats,120 in vitro a1 kylation of nucleic acids and nucleotides by phenylmonomethyltriazene,121 and by in vitro treatment of calf-thymus DNA with tritium-labelled MTIC (41 b), the proposed active metabolite of DTIC.122 In the latter case, however, less than 0.5% of the added 3H was recovered in DNA because most of the tritiated MTIC was hydrolysed rapidly in the aqueous medium to form methanol, which contained 96:< of the added label. It is also significant that MTIC inhibits DNA and protein synthesis in cultures of Novikoff hepatoma cells.123 The [Me2H3]- guanine isolated after incubation of [Me-2H3]-MTIC with calf thymus DNA in vitro displayed (chemical ionization mass spectrometry) a quasi-molecular ion (MH') at rn/e 169 which was three mass units higher than the comparable Ili W.J. Dunn, h.1. J. Greenberg, and S. S. Callejas, J. Medicin. Chem., 1976, 19, 1299. G. F. Kolar and R. Preussmann, Z. Naturforsch., 1971, 26B, 950. 'ID J. L. Skibba arid G. T. Bryan, Toxicd. Appl. Pharmacol., 1971, 18, 707. lZo P. Kleihues, G. F. !<olar, and G. P. Margison, Cuncrr Research, 1976, 36,2189. lZ1 R. Preussmann and A. von Hodenberg, Biochrm. Pharmacol., 1970, 19, 1505. lZ2 N. S. Mizuno and R. W. Decker, Biochetn. Pharmacol., 1976,25, 2643. D. D. Beal, J. L. Skibba, K. K. Whitnable, and G. T. Bryan, Cancer Research, 1976, 36, 2827. Monoalkyltriazenes ion for an undeuterated sample of 7-methylguanine; major fragment ions also occurred at m/e values three mass units higher than fragment ions in the standard. These important observations confirm that methylation of biological macro-molecules by MTIC must involve the transfer of an intact methyl group and exclude the involvement of diazomethane (or methylene).l24 As a consequence of the methylating reaction of MTIC the imidazole fragment is liberated as the primary amine 5-aminoimidazole-4-carboxamide.~~~~~~~ Although many of the biological effects of DTIC can be satisfactorily accounted for by invoking an MTIC intermediate127 monomethyltriazenes are, in general, no better as anti-tumour agents than their dimethyl counterparts. This is due to the greater toxicity and shorter biological half-life of the monomethyl-triazenes:128 direct administration of the monomethyltriazene is less effective than using the dimethyltriazene because the latter serves as a stabilized slow- release (‘prodrug’) form of the active metabolite. A monomethyltriazene (p-CH3S02.CsH4N=N. NHMe) with a relatively long half-life has comparable anti-tumour activity to that of its dimethyl homologue.128 A classical study by Connors and his co-workers on structure-activity relationships for the dialkyltriazenes (44), with variations in X, R, and R’, suggested that only chemicals which can be converted enzymatically to a mono- methyltriazene have anti-tumour properties.l16J28Thus the methyl-ethyl analogue (44b) is active because de-ethylation is preferred to demethylation and the active monomethyltriazene is formed. Although the diethyl analogue (44c) is dealkyl- ated as readily as the dimethyl (44a), the former is inactive, clearly because it cannot metabolize to monomethyltriazene.Mixed methylalkyltriazenes are active only if the higher alkyl group can suffer a-hydroxylation and removal to leave the monomethyltriazene; this cannot occur with the t-butylmethyltriazene (44d) since the higher alkyl group has no a-hydrogen and the compound is preferentially demethylated by liver microsomes to give the mono-t-butyl- triazene. Even though the methylating activity of monomethyltriazenes is conceded, the mechanism by which they methylate components of DNA is still obscure. From chemical studies (Section 4-D) it is apparent that monomethyltriazenes alkylate acidic sites more effectively and that acid catalysis is often necessary.Significantly the in vitro alkylation of guanosine by phenylmonoalkyltriazenes requires catalysis by acetic acid.121 What is certain is that a free methylcarbon-ium ion cannot be involved in the process although this has been repeatedly claimed in the literature. Apart from strictly chemical objections to such a fugitive intermediate, there is no possibility that such a reactive species could lap H. T. Nagasawa, F. N. Shirota, and N. S. Mizuno, Chem.-Biol. Interactions, 1974, 8,403. lLS J. L. Skibba, D. D. Beal, G. Ramirez, and G. T. Bryan, Cancer Research, 1970, 30, 147. lZ6 N. S. Mizuno and E. W. Humphrey, Cancer Chemother. Rep., Part 1, 1972, 56, 465. (Chem. Abs., 1973,78, 1141lw). IZ7 N. S. Mizuno, R. W. Decker, and B. Zakis, Biochem.Pharmacol., 1975, 24, 615. lf8 T. A. Connors, P. M. Goddard, K. Merai, W. C. J. Ross, and D. E. V. Wilman, Biochem. Pharmacol., 1976, 25, 241. Vaughan and Stevens achieve the necessary biological selectivity. Furthermore the methylation hypo- thesis of anti-tumour activity can be criticized on other grounds: (i) The hypothesis does not explain why dimethyltriazenes are selectively cytotoxic in vivo (i.e. after metabolism to a monomethyltriazene) to a tumour cell line (TLXS lymphoma) which has natural resistance to classical alkylating agents such as cycloph~sphamide.~~~ (ii) Diethyltriazenes are not anti-tumour agents despite the fact that they are metabolized to monoethyltriazenes which have alkylating capability.121J28 It is possible that the methylating activity of monomethyltriazenes is respon- sible for their carcinogenic properties and general cytotoxicity, and that different reactive metabolites are involved in selective anti-tumour action : the hydroxy- methyltriazenes (45) are obvious candidates since oxidative metabolic demethyl- ation of dimethylamino compounds generally is known to proceed via hydroxy-methyl intermediatesl3O Studies of the time course of oxidative N-demethylation of DTIC and dimethyltriazenes showed that the process levels off within five minutes of incubation and indicated the occurrence of concomitant metabolic conversion to compounds other than mon~methyltriazenes.~~~ Traces of water soluble triazenes tentatively identified as conjugates of 3-hydroxymethyl-3-methyl-1-phenyltriazene (45; Ar = Ph) have been isolated from the urine of rats treated with 3,3-dimethyl-1 -phenyltria~ene.l~~?l~~ When DTIC was adminis- tered intravenously to man 43% of the dose was recovered unchanged in the urine after 6 hours.One unidentified metabolite which gave a positive Bratton- Marshall test (i.e. contained an intact triazene linkage) was detected in trace am0unts.l3~ Possibly this metabolite is a hydroxymethyltriazene or a conjugate thereof. Future synthetic studies in the triazene field might be aimed at the synthesis of hydroxymethyltriazenes of type (45). It is possible that such compounds may be more stable than hitherto assumed, and have interesting biological properties. 12@ T. A. Connors and J. R. Hare, Biochem. Pharmacol., 1975,24,2133. lJo H. B. Hucker, in ‘Drug Metabolism Reviews’, ed. F. J. Di Carlo, Marcel Decker, Inc., New York, 1974, vol. 2, pp. 34-42. lS1 T. Giraldi, C. Nisi, and G. Sava, Pharmacol. Res. Comm., 1976, 8,25. 132 G. F. Kolar and J. Schlesiger, Chem.-Biol. Interactions, 1976, 14, 301. lJ3 G. F. Kolar, Xenobiotica, 1977, 7, 100. IS‘ Ti Li Loo, 3. K. Luce, J. H. Jardine, and E. Frei, Cancer Research, 1968,28,2448.

ISSN:0306-0012    

DOI:10.1039/CS9780700377

出版商:RSC    

年代:1978

数据来源: RSC

摘要:

Individual CH Bond Strengths in Simple Organic Compounds: Effects of Conformation and Substitution By D. C. McKean DEPARTMENT OF CHEMISTRY, UNIVERSITY OF ABERDEEN, ABERDEEN, AB9 2UE 1 Introduction:‘Isolated’ CH Stretching Frequencies, V(CH)~~ The purpose of this article is to review the recent evidence for the strengths of individual CH bonds which has been obtained from a study of CH stretching frequencies in incompletely deuteriated compounds, and to discuss its implications in the fields of molecular geometry, chemical kinetics, bonding theory, and substituent effects. The data obtained indicate the strength of the CH bond in its eqirilibrirrm state, as does the bond length, in contrast to the dissociation energy 0’298, which measures the diference between the energy of the equilibrium state, and that of its dissociated products.As will be seen later, these two kinds of ‘bond strength’ are often, though not always, related. The data are largely new, and precise to an extent rarely possible from con- ventional studies of CH stretching frequencies of normal, undeuteriated species. In the latter, the vibrational spectrum is highly complicated, mainly due on the one hand, to effects of coupling between CH bonds in CH3 and CH2 groups to give symmetric and antisymmetric stretching modes, and on the other, to Fermi resonances between these and overtones of CHs and CH2 bending modes. These difficulties are almost always removed by the observation of CH stretching frequencies in which all hydrogens are deuteriated except for one.For example in a methyl compound, the CHDz derivative is studied. In such compounds the CH bending modes are much lower than in the CH3 one, and Fermi resonances become insignificant in the CH stretching region. Moreover the coupling to other motions of the CH stretching is minimal, probably amounting to less than 10 cm-I in v(CH). Such frequencies have been termed ‘isolated’ oqes, [symbol V(CH)’~].‘While they may be used to calculate individual CH bond stretching force constants, the variation in coupling to other motions is so small, probably less than 5 cm-I in v(CH),= that the V(CH)~~values may be used directly as a measure of strength. One of the most pleasing results of CHD2 substitution in a methyl compound is the demonstration of the presence of more than one type of CH bond, in that D.C. McKean, Spectrochim. Acta, 1975, 31A, 861.’D. C. McKean, Spectrochim. Actn, 1973, 29A, 1559. Individual CH Bond Strengths in Simple Organic Compounds group, from a single spectrum. This was first carried out by Saur, Lavalley, and Romanet in CH3CH2X compounds, where X was C1, Br, or COOH.3 The potential barrier opposing internal rotation of the CH3 group is sufficiently high that a solitary CH bond in a CHDz group produces two conformers, one in which the hydrogen is trans to a CH bond in the CH2 group, the other in which it is trans to the CX bond. The barrier is of course low enough to ensure that bothconformers are always present.Two bands are then seen in the i.r. spectrum.* In the spectrum of (CHD2)2NH4 three i.r. bands are seen, as in Figure 1. Here the solitary CH bond can be trans either to H, or to the other methyl group, or to the lone pair of electrons.? 1 1 1 I I I 3000 2950 2900 2850 2800 2750 crn I Figure 1 Injrared spectrum in gas phase of (CHDZ)~NH.~~CHx signifies the CH bond trans to group X *The success of this experiment depends on the short time scale of the i.r. absorption process. The longer time scale required for the n.m.r. experiment means that only an average signal from both kinds of proton is observed in the latter. ?The spectrum is nof complicated by the presence of two CHD, groups. Coupling between two CH bonds separated by another atom such as C, N, or 0 has been shown to be negli- gible.4 ' 0.Saur, J. C. Lavalley, and R. Romanet, Compf.rend., 1969. 269, B, 816. D. C. McKean, Chem. Comm., 1971, 1373. T The first appreciation of the quantitative value of such CHDz frequencies appears to have been made by Allan and McKean, in a study of (CHD2)20.5 The foundation for the use of CH stretching frequencies for the prediction of bond lengths and dissociation energies, however, was laid in 1962 by Bernstein,6 who correlated 'average' stretching frequencies for CH3, CH2, and CH groups with bond length and dissociation energy, respectively. These 'average' v(CH) data however are only physically meaningful if all the CH bonds in the one group are identical, and even then they are subject to error due to Fermi resonances, being derived from frequencies of the normal, undeuteriated species.2 The Quantitative Use of v(CH)'S Values A. v(CH)iS Values and Predictions of Molecular Geometry.-v(CH) isvalues may be used generally for the prediction of CH bond lengths, and, in the case of certain CH3 groups, for the prediction of HCH angles. For bond lengths, the correlation curve of Figure 2 is empl~yed.~ The experimental bond lengths are confined to spectroscopic ro or rs values, for maximum compatibility.* Between 3050 and 2800 cm--l, the data are well fitted by the equation ro(A) = 1.402 -0.000 103 54CH)'". Two pairs of points, those for Me20 and Me3N, are particularly pleasing, since in each compound both the i.r.v(CH)iS bands and the microwave structures indicate the presence of two types of CH bond, the resulting pair of points lying close to the correlation curve. Another encouraging feature has been the number of occasions on which improvement in the precision or in the interpretation of a microwave spectrum has resulted in moving a point much nearer the curve, e.g.for HCN0,8 CH3CN,S CH3C= CH,'" and CH3F.l' Two notable anomalies remain, however, as previously noticed by Bernstein, viz. CHF3 and CHC13. For these the micro- wave CH bond lengths are greater by 0.0148, and by 0.015 8,than those predicted from v(CH).+ Apart from these, one can predict bond length differences with an accuracy of aoout kO.0005 A, since the slope of the curve is approximately 10 cni-l per 0.001 A, and the uncertainty in v(CH)~Sis less than 5 cm-l (due mainly to variations in coupling, not to experimental inaccuracy).This accuracy of *In the original plot, 0.003 A was added to each rs value, as for some molecules there is reason to suppose ro % rs + 0.003 A. In olefinic CH bonds, however, there is little evidence to support this. iAssumingr, = rs + 0.003. A smaller discrepancy of 0.009 A, possibly having the same origin, occurs in Me,CH, for the tertiary CH bond.la A. Allan and D. C. McKean, 'International Congress on Molecular Spectroscopy', Liege 1909; A. Allan, D. C. McKean, J. P. Perchard, and M. L. Josien, Spectrochim. Acta., 1971, 27A, 1409. H. 5. Bernstein, Spectrochim. Acra, 1962, 18, 161.D. C. McKean, J. L. Duncan, and L. Batt, Specrrochivn. Acta, 1973, 29A, 1037. 9.P. and M. Winnewisser, and F. Winther, J. Mol. Specfroscopy, 1974, 51, 65. J. L. Duncan, D. C. McKean, and N. D. Michie,J. Mol. Srructure, 1974, 21, 405. lo J. L. Duncan, D. C. McKean, P. D. Mallinson, and R. D. McCulloch, J. Mol. Spectro-scop~.,1973, 46, 232. l* D. F. Eggers, J. hlol. Strwfim, 1976, 31, 367. l2 D. C. McKean, S. Biedermann, and ti.Biirger, Spccfroc,Aim. Acra, 1974, 30A, 845. Individual CH Bond Strengths in Simple Orgunic Compocmds k0.0005 8, greatly exceeds the precision of most microwave determinations. It has therefore been of considerable interest to discover that ub initio calculations, aided by the force method of Pulay for optimizing geometry,I3 reproduce many of the CH bond !ength variations to 0.001 8,or better.Table 1 shows some recent 0 z m I II 0E m 0 ‘E 0 I 0 0x 0 0 0, (u 0 0 Q)Iu Figure 2 Correlation between V(CH)~~and ro(CH).7The dotted line for CH3CN shows the improvement due to a recent redetermination of geometryg lS P. Pulay, Mol. Phys., 1969, 17, 197. 402 Table 1 V(CH)~~arid predictioris of To, DO298 in some hydrocarbons v(CH)ib/cm-L ro(pred.)/8i re(4-3 I G)/A yo -r,/A D"298(pred.)/kcalinol-I 0'298 (obs.)/kcal mol-l CH4 2992" 1.092 1.081 0.01 1 103.2 104.0" C2H6 2950" 1.096(5) 1.083 0.013(5) 99.6 98" (CH3)2CMg S-H 292OC 1.099(5) 1 .0856 0.014( 5) 97.0 95a p-H.\I" 2950" I .096(5) I .083 0.013(5) 99.6 p-HH 2936" 1.098 1.084!' 0.014 98.4 98" (CH3)sCH t-H 2894" I.I02(5) 94.8 92" p-.H 11 e 2943 1.097 99.0 p-H" 2924 1.099 97.3 (CH3hC 2934" 1.098 98.2 99.3* PhCH3 Me 2947" 1.097 99.4 88.3" CYCIO--C~HG 3056" 1.085(5) 1.072" 0.013(5) 108.7 100.4" CYCIO-C~H 2956(l 1.096 100.1 96.gay C2H4 3055" 1.085(5) I .073 0.012(5) 108.7 108" CH3CH=CH 2 Ha 3013f 1.090 1.076b 0.014 105.0 Ht 3064f 1.084(5) 1.072" 0.012(5) 109.4 HC 3039f 1.087 1.074O 0.013 107.3 (CH3)zC -CH2 3050f 1.086 108.2 CHg-C CH2 3049" 1.086 108.1 CsH 6 3065" 1.084(5) 109.5 110.5" C2H2 3336" 1.060 1 .O51 0.009 128" CH3-CCH ac-H 3 3 34" 1.060 1.0509 0.010 128N Me-H 2958" 1.095(5) 1.083s 0.013 100.3 93.9h 5 Uref. 7; "refs. 21, 220; Cref. 12; dJ.Lascombe, personal communication; see also text, Section 3G; eref. 23; fref. 1; gref. 24; 25; p = primary. 3 ~s secondary, t -= tertiary, ac --acetylenic, Ha,t,c -hydrogen a, frans,or cis to methyl, Hx = H frans to X. For data for Mein propene, but-1-ene 58 see Section 3F. !2w Individual CH Bond Strengths in Simple Organic Compounds examples of this amongst hydrocarbons, other examples being found in other tables. It may be concluded that a very simple i.r. experiment yields at least one geometric parameter, r"CH, more accurately than any other known method. The correlation between V(CH)~~and the HCH angle for methyl groups is shown in Figure 3.14 For the line drawn, CL'HCH = 0.0471~(CH)~" 31.1. It-appears to hold for symmetric CH3 groups, excluding CH3F, and also, rather well, for CH3XY2 type compounds in which internal rotation is essentially free.For these last, the average V(CH)'~frequency is used (see further, below). For asymmetric CH3 groups it enables a rough estimate to be made of average HCH angles, except where CH3N or CH30 are present. One success of the correlation was the prediction of 107.0' for the HCH angle in Me4C. A recent experimental determination yields 106.6" for this quantity.'5 No such correlation exists for HCH angles in methylene groups in general, but a rough one for C --CH2 systems is evident in the data compiled by Duncan.I6 It is not linear. 111.0' llO*O"' ~HCH 109.0" 108.6 107.0' 2900 20 40 60 80 3000 20 40 v is ( crn-1) Figure 3 Correlation between v(CH)~~and aHCH in symmetric CH3 groups (points denoted I+ ).Points denoted 0are for asymmetric methyl groups which undergo free ititernal rotation, using v(CH)av.is = $[v(CH),, + v(CH)l] (noerror indicated) l4 D. C. McKean, J. Mol. Strlrrtirre, 1976, 34, 181. l5 L. S. Bartell and W. F. Bradford, J. Mof.Structirre, 1977, 37, 1 13. J. L. Duncan, Spectrochim. Acta, 1970, 26A, 429. Mc Kean B. V(CH)~~Values and Predictions of Dissociation Energies : Applications in Chemical Kinetics.-Predictions of 0'298 for the CH bond may be made using the correlation curve of Figure 4, based on experimental D'PM values, for the most 0 8 m 0 z m 0 9 m .-I: 840 m 0 0 0, (u 0 043 (u -J Y ~t 3 01 0 Figure 4 Correlutiori betweeri ~(cI-4)~~ariddissociutiori energy, D0298,in kcal rn01-l.~ The newer experimental Dovalue of 119.2for HCNL7clearly coriflicts with that for CzHz arid with the general trend of the lower- points Individual CH Boiid Stretigths i/i Simple Orgrrtiic Compounds part derived from kinetic The curve is drawn so as to lie close to points for compounds such as hydrocarbons which are likely to dissociate into radicals in which stabilization energy is negligible.Between 2700 and 3 100 cm-1, D3m(kcal) = 0.0861~V(CH)~"-154.6.The upper end of the curve was originally fixed by an earlier experimental D^value for HCN of 130.8 kcal mol This is now thought to be somewhat lower, at 119 kcal,17 which may indicate that a little curvature is needed at the upper end of the plot. All the deviant points lie below the plot, the lowest being those in which stabilization energy, accompanied by marked change in geometry, has long been associated with the radical formed, e.g.CHKN, CHZPh. For each of four compounds, (CH3)2CO, CH30H, CHsNHs, and (CH3)20, two points are plotted, thus G-U. This is because they each have two v(CH)iS values, but only one experimental DO298 kalue. It may be significant that the point for the lower v(CH)lS value in each case lies closer to the correlation curve, which is understandable if in the dissociation process it is only the weaker CH bond in the methyl group which is broken. The relevance of this to kinetic processes in which the rate-determining step is the breaking of a CH bond, seems obvious. The gradient of the curve, approximately 0.9 kcal mol-l per 10 cm-l, implies that in the same methyl group, differences in dissociation energy, and therefore of activation energy, of the order of 3-5 kcal mol-l are common, and for CH3-0 and CH3-N compounds are much greater, the highest difference found being 13 kcal mol in trimethylamine.These differences are quite large enough to imply that only the weaker bond stands a chance of being broken in the kinetic process. This conclusion of course assumes an 'adiabatic' process in which the conformation of the methyl group is maintained, and that no stabilization energy is involved. It may be argued that the dis- sociation process should be considered to involve some kind of averaging of the separate strengths of the bonds in an asymmetric methyl group, particularly if internal rotation is essentially free.Against this, it must be stated firstly that there is clear evidence (see Section 3G) that CH bond strengths vary during a free internal rotation, secondly that in any event free rotation in the transition state seems highly unlikely, and thirdly that in a bimolecular breaking of the bond, the conformation of the CH bond, relative to the framework which causes the asymmetry, will probably be determined by the incoming trajectory of the colliding species. Thus for CH3NO2, it should be much easier for a radical to remove a hydrogen atom if it impinges at right angles to the skeleton plane, than if it strikes within that plane.* Both activaiion energy and pre-exponential factors should therefore be affected.What will happen in a unimolecular dis- sociation is harder to imagine, but it would be surprising if the fact that a CH bond may more easily be broken in one direction than in others does not play some part. Stubilizc-rtiorrEtiergy. The correlation plot of Figure 4offers an improved means of measuring this quantity, as the difference between the DOmvalue predicted *In addition to the predicted difference in D",,,of 5 kcal mol-' (Section 3B), stabilization energy should further lower the activation energy for the approach at right angles. V. H. Dibeler and S. K. Liston, J.Chctn. Phj,.~.,1968, 48, 4765. McKean from the Y(CH)~Sand the experimental one. One such application throws light on an old controversy, whether the weakness of a tertiary CH bond is due to a large stabilization energy of the t-butyl radical. The data in Table 1 indicate that the latter cannot exceed 2-3 kcal mol-l. The tertiary CH bond is weak in the equilibrium state. By contrast, halogenated methanes, other than fluorides, appear to involve significant amounts of stabilization energy. This may help to explain why the attack of chlorine atoms on ethyl chloride occurs almost wholly on the CH2 group,l8 where the bonds have a 0'298 of 102.5 kcal mol-l, in contrast to the weaker bond in the CH3 group which has a DO298 value of only 99.2 kcal m01-~.~~* We note finally that the stabilization energy of 8 kcal mol-1 for the cyclo- C3H5- radical indicated by the present results seems rather high, and possibly the existing experimental value of 0'298 is in err0r.~1-~~ C.v(CH)iS Values and Force Constant Calculations.-It must be stated cate- gorically that v(CH)iS values should be the sole sources of individual bond stretching force constants, unless there is greater error in them, due to Fermi resonances in the partially deuteriated species, than there is in the normal, undeuteriated ones. The latter should normally be used only to determine stretching interaction force constants and/or Fermi resonance shifts.2,26 This approach has been very seldom adopted, except in the treatment of vibrations of CH3 or CH2 groups which consider only CH or CD stretching (e.g.refs.1, 2, and 12). It is of course easy to calculatef(CH) in the diatomic approximation.+ Such constants should normally be higher than those from a complete treatment by about 0.03 mdyn A-1.t In one comprehensive treatment of CH stretching frequencies,2s V(CH)~~ values *That the CH dissociation energy can be important in hydrogen abstraction reactions is suggested by the preferential attzck of CI on the CH, a to 0 in ethers,2@which is easily understood on the basis of the lone pair truns effect (see Section 3A). tf(CH)diat. =[V(CH)'~/I302.8312/1 .0756. Normally variations in f(CH)dlat larger than 0.015 mdyd-' should be significant. $In H,CO, however, owing probably to the abnormally high CH/CO stretch-stretch interac-tion, 'f( CH)diat.< f(CH)(g .f. f. ). la D. C. McKean and B. W. Laurie, J. Mol. Strucfirre, 1975, 27, 317; see also F. F. Rust and W. E. Vaughan, J. Org. Chem., 1941, 6, 479. D. C. McKean. 0. Saur, J. Travert, and J. C. Lavalley, Specfrochim. Acta, 1975, 31A, 1713. 2o H. Singh and J. M. Tedder, J. Chenz. Soc (B), 1966, 612. 11 R. Ditchfield, W. J. Hehre, and J. A. Pople, J. Chem. Phys., 1971, 54, 724. 22 (a)C. E. Blom, P. J. Slingerland, and C. Altona, Mol. Phys., 1976, 31, 13'59; (6) C. E. Blorn, L. P. Otto, and C. Altona, Mol. Ph~s.,1976, 32, 1137. a3 J. L. Duncan, personal communication. 24 D. C. McKean, unpublished work. 25 R. Walsh, Trans. Faraday Snc., 1971, 67, 2085.26 (a) D. C. McKean, J. C. Lavalley, 0. Saur, and J. Travert, Spectrochim. Acta, 1977, 33A, 865; (6) D. C. McKean, J. C. Lavalley, 0. Saur, H. G. M. Edwards, and V. Fawcett, Spectrochim, Acta, 1977, 33A, 9 13. 27 J. L. Duncan and P. D. Mallinson, Chem. Phys. Letters, 1973, 23, 597. 28 (a) H. Perrin, A. CossCe-Barbi, N. Brigot, and S. Qdio:, Mol. Phys., 1975, 29, 1261 ; (b) H. Perrin, J. Mol. Structure, 1975, 28, 305; (c) A. Cossee-Barbi and H. Perrin. J. Mol. Structure, 1976, 30, 69. 407 Individual CH Bond Strengths in Simple Organic Compounds have been used in several cases and differences here from ‘diatomic’f(CH) values of the order of 0.07 mdyn A-1 found. It is not clear however in this work, what assumptions were made concerning CH stretching interaction constants.These are in general very hard to determine. The same work is notable for a CNDO treatment of CH stretching force constants. While CNDO calculations alone do not produce meaningful values of force constants, or of bond lengths, they can be made to predict variations in f(CH) on a semi-empirical basis with a substantial degree of success, although CH bonds affected by lone pair trans effects require a different approach to those in hydrocarbons. 3 Effects of Conformation and Substitution These will be considered in seven groups: A. the trans effect of lone pairs of electrons; B. the systems H-C-0 and H-C-N; C. a-substitution effects of Me, halogens, CF3, and CC19 in saturated com- pounds ; D.p-substitution effects of Me and halogens in saturated compounds; El. substituent effects on olefinic CH bonds; F. methyl groups adjacent to multiple bonds : hyperconjugation ; G. methyl groups attached to planar XY2 skeletons. Where substitution of a group X for hydrogen may be studied, one can define a substituent effect Sx as the difference in V(CH)~~ caused by that substitution. The evidence below suggests that it is rarely profitable to do this, in view of the non-additivity of such parameters. A. The trans effect of lone pairs.-The presence of two types of CH bond in CH30 or CHzO compounds was first recognized by Fre~mann~~ in studies of the overtone region of v(CH). The specific role of the lone pair was identified through the disappearance of the weak CH bond frequencies when the lone pairs were involved in complex formation.* Rather better known is the post-war work which characterized CH30 and CH3N groups by their low-lying bands near 2800 cm-1, which reflect the presence of weak CH bonds.31t Bohlmann first recognized that a lone pair weakens a CH *It is interesting to observe that these earlier conclusions have been recently vindicated by contemporary work in the overtone region, which has resulted in the ‘local mode’ description of CH stretching in CH, or CH, +The suggestions in these papers that the 2800 cm-I bands are due to ‘symmetric’ CH, stretching fundamentals are now seen to be erroneous.The more appropriate is the word ‘fundamental’, the less appropriate is the word ‘symmetric’. This conclusion follows from quantitative descriptions based on the CHD, data.35 2y R.Freymann and J. Gueron, Compt. rend., 1937, 205, 859; M. et R. Freymann, Compt. rend., 1944, 219, 515. 30 B. R. Henry, Accounfs. Chetn. Res., 1977, 10, 207. 31 H. B. Henbest, G. D. Meakins, B. Nicholls, and A. A. Wagland,J. Chetn. Soc., 1957, 1462; J. T. Braunholtz, E. A. V. Ebsworth, F. G. Mann, and N. Sheppard, J. Chem. Soc., 1958, 2780; R. D. Hill and G. D. Meakins, J. Cfienz.Soc., 1958, 760. McKean bond in the trans position, in studies of quinolizidine alkal0ids.3~ Much work has since been done, both in n.m.r.33 and in the i.r., in the latter notably by Krueger and co-workers.34 However, v(CH)~S data were not obtained, so that a quantita- tive treatment of the bond strength was not possible.* The first v(CH>is values were reported by the author in the simplest CH3N and CH30 compounds in 1971.4,35 Table 2 shows Y(CH)~~ data for these and some CH3P and CH3S corn pound^.^^^^^ It is convenient to identify each CH bond by the group X lying trans to it, thus, v(CH)~.It is seen that v(CH)lp is considerably lower than v(CH)~ in the N and 0compounds, its lowness increasing or V(CH)”~ with the number of methyl groups, and decreasing with the number of lone pairs.+ Where there are two of the latter, as in CH30 species, there are two CH bonds trans to lone pairs, if these are visualized as in a Nyholm-Gillespie model, and accordingly two weak CH bonds are found.Extension of this idea to CH3F leads us to expect three weak CH bonds here, and indeed the CH bond in this molecule is anomalously weak compared with CH4 or CHSCI.~ Although CH3P and CH3S compounds are also electron donors, the larger size of the P or S atom and their differing valence angles lead one to anticipate smaller truns effects. Such effects are indeed fomd to be small, zero, or even reversed,37 as the data in Table 2 show. Mechariism of the Lone Pair trans Efect. Insofar as ab initio calculations reproduce the bond strength variations very well, it may be said that the computer knows the physical cause of this effect, the difficulty lying in translating information concerning MO eigen-functions and values into intuitive pictures.Two prominent features of the population analyses are the higher charge densities found on the trans hydrogens, and much larger off-diagonal antibonding terms between these same hydrogens and the nitrogen, than for hydrogens gauche to the lone pair. These results are in keeping with the n.m.r. evidence that the trans hydrogen is more screened than the gauche one and with the suggestion of Hamlow ef af.33 that the lone pair is partly delocalized into an antibonding 0orbital in the trans CH bond. In VB terms this would be equivalent to a repulsion between the CH bonding pair and the lone pair of electrons. A similar explanation, in terms of a Nyholm-Gillespie picture, has been given for the tilt of the methyl group, away *e.g. derivatives such as (CH,),CDNH, were studied and low v(CD) bands found where CD was trans to the lone pair.CD stretching modes however are not good characteristic group frequencies, and in addition are often subject to Fermi resonances. See also ref. 42. TWe note an ambiguity in defining a lone pair trans effect. It could refer to the absolute weakness of v(CH)’P, or to the difference between v(CH)’P and v(CH)~ Theand/or v(CH)~~. distinction is clearly unimportant in CH3N and CH30 compounds. 32 F. Bohlmann, Chem. Ber., 1958, 91, 2157. 33 H. P. Hamlow, S. Okuda, and N. Nakagawa, Tetrahedron Letters, 1964, 2553. 34 e.g. P. J. Krueger, J. Jan, and W. Wieser, J. Mol. Sfrirctiire, 1970, 5, 375; P. J. Krueger and J. Jan, Canad. J. Chem., 1970, 48, 3229, 3236.35 D. C. McKean and I. A. Ellis, J. Mol. Strircfirre, 1975, 29, 81. 36 J. Travert, 0. Saur, A. Janin, and J. C. Lavalley, J. Mol. Structure, 1976, 33, 265. 37 D. C. McKean and G. P. McQuillan, submitted for publication in J. Mol. Strucfiire. 2 g and lone-pair trans eflects 4Table 2 V(CH)'~ Are(ab.in.)a/A A Do29ab/ 5kcal mol-I F 2955 2880 0.008 0.009d 6.5 2 2962 293 7 2834 0.013,O.Oll 0.010,0.009e 11 .O, 8.9 $ 2952 2799 0.016 0.015f 13.2 h 2979 2920 0.006 0.006h 5.1 2 2984 2884 O.OlOC5) O.OOgh 8.6 20 2964 2979 -0.001(5) 1.3 2962 2952 2949 0.001(5),0 1.1 2. 2954 2919 0.003(5) 3.1 2985 2982 -0 -0 2979 2947 0.003(5) 2.8 nkr -rc-lp -~CH~LS'II; D'CHvr,Ii-DrCHip; rref. 35; "ref. 38; Pref. 39; fref. 37; gref.4; href. 22b; 'ref. 37; Jref. 36; v(CH)'S data for ~/JD~ CH,OPCI,, CH,COOCH, and HCOOCH, are quoted in ref 7, p. 1049. ?* P. Pulay and F. Torok, J. hfol. Structure, 1975, 29, 239. 39 (a) S. Skaarup, L. L. Griffin, and J. E. Boggs, J. Atner. Chem. Soc., 1976, 98, 3140; (h) E. Flood, P. Pulay, and J. E. Boggs, J. Amer. Chem. Soc., 1977, 99, 5570. McKean from the CN direction, in (CH~)ZNH.~~ The author has suggested35 that since a non-bonding repulsion is a mutual effect, the decrease in ionization potential of the lone pair in the series MeNH2, Me2NH, Me3N may be due to the cumulative effect of repulsions from the tram CH bonds in each methyl group. However it is then difficult to see why the v(CH)lpvalue falls with increasing methyl substitution unless (i) there is a direct ‘7’repulsion effect between the methyl groups, as suggested by Bellamy and may^,^^* or (ii) the character of the lone pair orbital is altered by replacement of the OSH bond by the USC bond.The latter could then alone be responsible for the change in ionization potential. However it would seem prudent to include a ‘back’ trails effect as contributing to the changes in 1P observed. Wider Implications of the trans Efect. These have been well reviewed by Bellamy and May0.41~ trans Effects of lone pairs should be, and can be found, affecting not only CH but NH and OH bonds. There seems no reason also why they should not likewise affect C-C or C-X bonds generally, and indications may be found that this is ~0.4~ If however C-C bonds are less affected than C-H, then the tram effect will have a significant role to play in determining the relative stabilities of the conformers of pyridine-type molecules.The effect will then tend to favour conformers in which the lone pair is trans to C-C (NH axial), contrary to the preferred conformation in which NH is eq~atorial.~? Lone pair trans effects should also occur across C -0 and C -N bonds (see Section 3B) and various author^^^?^^ p43 have suggested that the well-known weakness of the aldehydic bond arises from this cause. The lone pair trans effect is also a useful tool in conformational problems, e.g. the studies of dioxalane, dioxane, and other -CHRO- systems by Lavalley and co-workers ;44145 also, of pyridine~.~~ Another example, of different type, involves the disappearance of the effect when the lone pair is removed from the nitrogen ititerriallq>,as for instance when CH3 is replaced by SiH3 or SiF3.35 The low i.r.band at 2865 cm-1 (gas), (2845 cm-l, liquid) in B[N(CHD2)?]3 suggests that the BNCZ group is pyramidal, whereas the corresponding band at 2924 cm-l (gas) in C12BN(CHD+ is in keeping with a planar skeleton.46 B. HC-0 and HC=N Systems.-For many CH=O systems, the observed v(CH) values of the normal species, which are well known, are sufficiently low to be regarded as v(CH)is, although Fermi resonances with 28(CH) may remain. Table 3 lists the most reliable of these, to illustrate several notable features. Firstly, the a effect of methyl is a depression of about 43 cm-1, as found in *See Section 3D.However the 4-31 G calculation for (CH,),N shows positive overlap between each of the HIP atoms.24 +A possible application in the chlorination of ethersz0 was noted earlier (Section 3B). L. Bellarny and D. W. Mayo, to be published. 41 L. J. Bellamy and D. W. Mayo, J. Phys. Chem., 1976, 80, 1217. 4z I. D. Blackburne, A. R. Katritzky, and Y.Takeuchi, Accounts. Chern. Res., 1975, 8, 300. 43 D. C. McKean, Spectrochim. Actu, 1975, 31A, 1167. 0.Saw, A. Janin, J. C. Lavalley, and N. Sheppard, Compt. rend., 1973, 276, B, 725. 45 0. Saur, A. Janin, A. Vallet, and J. C. Lavalley, J. Mol. Strircture, 1976, 34, 171. 48 Data from H. J. Becher, personal communication. 41 1 Individual CH Bond Strengths in Simple Organic Compounds Table 3 V(CH)~~in some H-C=-0 and H-C=N systems V(CH)~~ V(CH)'~ u(CH)is HCHO 2813" FCHO 29810 CHOCHO 2850* CH3CHO 2770a HOCHO 2943" HSCHO t, c 2843,2845C CH3CHO,BF3 2940b CH30CHO 2930" HCzzCCHO 285@ PhCHO 2770a CHz=CHCHO 2800b CC13CHO 2856d PhCHO,BF3 2930b NH2CHO 2852' CBr3CHO 2845d HCOO- (2816)b CH3CH=NCH3 2839e CH2=NH 3018f 2936f aref.7; bref. 28c; cref. 47; dref. 48; efrom the species CD,CH=NCD,;2* see also ref. 50; fref. 49. Whether these frequencies are truly v(CH)'S, seems doubtful. alkanes (see below). Secondly, certain substituents, viz. F, HO, CH30, produce very large increases in v(CH), in the range of 120-170 cm-l.Thirdly, complexing with BF3 also produces large upward frequency shifts. A fourth observation of considerable interest is that u(CH)is in HCOSH is almost identical in its cis and trans isomers, this apparently indicating the absence of any effect on v(CH) of H-H interaction in this molecule.47 The success of semi-empirical CNDO calculations has already been mentioned ;28 the physical reasons for the variations however remain obscure. Ab initio calculations on H2CO and CH3CHO show clearly an unusually high negative charge on the aldehyde hydrogen, and similarly high negative overlap populations between the latter and the oxygen. The weakening effect of the methyl group appears to arise from greater Hald ---0 and Hald * * -ChTe antibonding (the latter compared with Hald * -* Hald in HzCO) in CH3CH0, the overlap between Hald and all three methyl hydrogens being positive.24 In CH=N systems, Bellamy and Mayo4I found good evidence for low v(CH) values associated with the expected lone pair trans effect in a number of cases, e.g.CH2=NOH.However the only secure value of U(CH)~~ known so far is that for MeCH=NMe (2839 cm-l), which exists only in the conformer in which the CH bond is trans to the lone pair. C. a Effects of Methyl, Halogens, Cc13, and CF3, in Saturated Compounds.- These are shown in Table 4. The a effect of methyl is always a depression of v(CH)iS, by an amount varying from 42 to 19 cm-l. It diminishes with an increase in the number of methyl groups, as seen in the series CH4, MeCH3, Me2CH2, and Me3CH, and it is reduced by the successive addition of chlorine atoms.The a effect of C1 has been most studied.18.19 This is a rise in frequency which 47 W. H. Hocking and G. Winnewisser, Z. Natiirfursch., 1976, 31a, 422, 438, 995. 48 G. Lucazeau and A. Novak, Spectrochim. Acta, 1969, 25A, I6 15. 49 C. B. Moore, G. C. Pimentel, and T. D. Goldfarb, .I.Chem. Ph!*s., 1965, 43, 63. 50 H. Hollenstein and Hs. H. Giinthard, Chern. Phys., 1974, 4, 368. 412 Table 4 Some &-substitution eflects on v(CH)iS in saturated compounds v( CH)'S SaF v(CH)~" SaF Y(CH)~~ SaF CH4 2992 MeCH3 2950 MezCH2 2920 CH3F 2976 -16 MeCHzF 2950 0 Me2CHF 2931 11 CHzFz 2984 8 CHF3 2990" 6a Sac' s, c1 Sac' CH3CI 3012 20 MeCffzCI 2983 33 MezCHCl 2962 42 CHzCIz 3025 13 MeCHCl2 3006 23 CHC13 3034 9 SUB' SaBr SaBr CH3Br 3027 35 MeCHzBr 2996 46 MezCHBr 2975 55 CH2Br2 3040 13 CHBr3 3050 10 'a1 'a1 Sal CH31 3029 37 MeCH21 2998 48 MezCHI 2974 54 CHzI2 3037 8 CHI3 3026 -11 S,CF3 Sac% SSC* CH3CF3 3010 18 CH3CCI3 2988 -4 CH3CN 2985 -7 CH2(CF3)zb 3008 -2 CH2CICC13 3000 -12 CH2(CN)2 2963 -22 CH(CF3)3' 2992 -16 CHClzCC13 3002 -23 Sax values are the increments in Y(CH)~~ due to successive a subsitution of X for H; Sable can be obtained from horizontal differences in v(CH)'B.5* Data from refs. 7, 18, and 19 except where otherwise indicated. a Qf: aconsiderable uncertainty due to Fermi resonance, ref. 5 1 ; OH. Burger, personal communication. 3 Individual CH Bond Strengths in Simple Organic Compounds is progressively smaller the more chlorines that are added.The a effect of C1, however, is enhanced by methyl substitution. It seems as though the initial depression by a methyl makes possible a higher subsequent elevation by chlorine, in contrast to the effect noted above of the effect of chlorine on a methyl sub- stitution. It is also in contrast to the effect of cc13 which greatly reduces the effect of a CI. The effects of a bromine and a iodine are elevations which initially are greater than those of chlorine, but which thereafter are similar, or as in the case of CHI3, reversed. Fluorine behaves anomalously amongst the halogens in that the first fluorine gives a marked depression, attributed to a lone pair affe~t,~ whereas the second and third give only small increases.* The variations in these a effects on CH bonds are in contrast to the a effects found in SiH bonds.In the latter, S,hfe is essentially constant, at about -20 cm-1, while SaFand Sac1 appear to increase marginally with successive addition of halogen, SaFbeing positive from the beginning.ls The 01 effect of a CX3 group can also be considered as the effects of the CX bonds concerned and it is in this context that models for the mechanism of the a effect have been discussed (see Section 3D). D. p Effects of Methyl and Halogens, in Saturated Compounds.-A group placed in a p position may in general produce different effects depending on whether the CH bond affected is trans or gauche to the substituent.It is not easy to decide how best to describe the effects found. In some cases the simplest method is to quote v(CH)iS values classified according to the group lying in the trans position, as in Table 5 which summarizes the data for all the ethyl, isopropyl, and t-butyl halides. Three main features are seen. The value for V(CH)'~ for a CH bond trans to halogen falls in the sequence F > C1 > Br > I, the extent of the fall increasing with the addition of methyl groups. The starting point of each series, however, differs, thus StFtakes the values + 7 (EtF), +22 (i-PrF), + 36 (t-BuF). Again, StC1 starts negative (-5, EtCI) and then becomes positive (+5 in i-PrC1, + 12 in t-BuC1). The second feature is the constancy of the V(CH)~~ values tram to CH3 or H as the halogen is varied.Since these are all gauche to halogen, this could be described as a constant gauche effect of halogen, which is a consistent rise of 20-24 cm-1 compared with the corresponding hydrogen compound. The third feature is that the bonds tram to CH3 are consistently 14-19 cm-1 stronger than those trans to H, in the same molecule. A larger difference (25 cm-I) is found in CH3NHMe. This could be termed a gauche effect of methyl, providing it is realised that this difference is in fact SP-SyAlein terms of the definition of S. The value of StJIe-SgJIe is more constant than is that of St31ealone. Thus Sti\Ie from a comparison of propane with ethane is 0, but from the pair MesCH2/ Me3CH it is +7 cm-I, and from MeK/MesCH, + 10 cm-l.Sg3Ie is more constant: -14 cm-1 from the pair CHKH.Me/CH3CH3, -12 cm-1 from CH3- CHMez/CH3CHzMe and -13 cm-1 from CH3NHMe/CHsNH.. *The value of V(CH)'~for CHF, is uncertain due to a strong Ferrni resonance.61 jLC. C. Costain, J. Mol. Spectroscop~',1962, 9, 317. 414 McKean Table5 p-substitution effects of Me and halogen on v(CH)~Sin saturated compoundsa groups lying trans CH3 H X CH3CH2-H 2950 F 2973 2957 c1 2972 2945 Br 297 1 2936 29 69 2928 CH3MeCH-H 2950 2936 F 2974 2958 2958 C1 2976 2959 294 1 Br 2976 2958 2929 I 2973 2955 2918 CH3Me2C-H 2943 2924 F 2967 29 60 C1 2968 2936 Br 2971 2925 I 2968 2909 CH3CHC12 2984 297 1 CH3CC13 2988 CH3CMeC12 2993 2966 aThe v(CH)’s values listed are for CH bonds in the methyl group in italics, data from ref.I8 and 19; j3 effects of methyl in CH,-N, P, 0,and S compounds can be seen in Table 2. The reliability of the value of StJIe-SgnIeleads to the prediction that in cyclohexanes, the equatorial CH bond will be stronger than the axial one, by 2-5 kcal mol-1, since the former is trans to two C-C bonds, the latter trans to two CH ones.* A tabulation of Sg and St effects for both methyl and chlorine for chloro- alkanes is included in ref. 18, mainly to draw attention to the high degree of non- additivity they exhibit. A general conclusion for Scl values is that both SaC1and SBCl effects diminish with increasing numbers of chlorine atoms.It is plain that the trans and gauche effects of halogen cannot be easily related to the normal pattern of inductive effects. Chemical models for ,&substitution eflects. Two models have been suggested, both based on the premise of repulsions between the terminal atoms, or bonding pairs of electrons involved, in the p bonds. The Reporter, starting from the above VB description of the lone pair trans effect, has speculated that such repulsions also occur between two bonding pairs of electrons, similarly related. In this way, the decreasing strength of the CH bond with a methyl substitution in the series CHI, CH3Me, CH2Me2, CHMe3 is attributed to the increasing number of repulsions involving the trans CH bonds in each methyl group.12 The varying trans effect of *This has now been confirmedj5 (J.Caillod, 0.Saur and J. C. Lavalley, to be published). The difference in strength is 31 cm-l 5 2.7 kcal mol-I; see also ref. 45. 415 Individual CH Bond Strengths in Simple Organic Coriipoirnds halogen in the alkyl halides is then attributed to a progressive 'loss of control' by the halogen of the CX bonding electrons, in the order F > C1 > Br > I, with a resultant increase in the repulsion between this pair and the pair in the trans CH bond.43* Bellamy and may^,^^ by contrast, suppose gauche H . . * H (or bonding pair) repulsions to dominate both CH bond strength, and also CH3 HCH angles, gauche H * -X repulsions being held to be insignificant. The fall in v(CH)~S of 42 cm-l from CH4 to CH3Me is then due to the gauche effect of two hydrogens.This then leads to an explanation of the constant guuche effect of halogen of about +20 cm-l, since a CH bond gauche to halogen in CH3CHzX is now gauche to only one CH bond, and the repulsion from the gauche CX is considered insignificant in all cases. The v(CH)~~ consequently rises one half of the way back to the value in CH4. If two or three phalogens are present, the CH bond may be gauche only to two halogens and a v(CH)'" value close to that in CH4 is then expected, and indeed found. The varying trans effect of halogen is attributed in this work40 to changes in HCX angles, with changes of halogen. Bellamy and Mayo extend their invocation of repulsions between hydrogen atoms to explain the gauche effect (or rather, (SPe-SSMe) of methyl described above.Here the repulsions invoked involve CH bonds y to each other. Slight support for part of the Bellamy-Mayo model comes from studies of p CH-CH stretch-stretch interaction force constants.26 The constant involving gauche bonds is positive, as expected for a repulsion between the hydrogens. By contrast, the constant for trans bonds is negative, in agreement with population analyses from ah initio calculztionsZ4 which show positive (bonding) terms between trans hydrogens, but negative (antibonding) ones between gauche ones. These overlap populations are of course connected with the ab.initio force constants, which for ethane have the same signs and fairly similar magnitudes to those found by experiment.266 However, the overlap populations for y hydrogens in close proximity, are positive,24 as noted also by Cremer et al.52 in connection with barriers to restricted rotation, so that no support is given to the Bellamy-Mayo explanation of the gauche effect of methyl.A further consideration is that these por y overlap populations are all smaller by a power of ten than the a ones, acting between the H and X atoms in the system H-C-X, which are always negative. Moreover the correlation between CH bond strengths and HCH angle works equally well for CH3X compounds in which X is halogen, CN, CrCH, NOz, BFz, BC12, Ph, as it does for CH3Me.14 These represent a wide range of V(CH)~~ values in situations where no p CH bonds are present.The implications seem to be that it is the nature of the X group and the CX bond which determine the HCH angle and perhaps also the CH bond strength. Exploration of the substitution effects of halogen by ah initio calculations is *Supposing [he carbon 2pcharacter to be responsible for the repuI~ion,~~~this may be expected to increase in the series F, CI, Br, or 1. D. Cremer, J. S. Binkley, J. A. Pople, and W. J. Hehre, J. A1iii.r. C/irl?i.Soc., 1974, 96, 6900. 416 McKean restricted to fluorine and chlorine compounds.24 The results have been dis- appointing. The 4-31G basis set describes very poorly both the a effect of F or C1, and the combined a effects of halogen and methyl. The p effects of halogen are somewhat better reproduced, the constant gauche effect exactly so.The difference in re for CHx between fluorine and chlorine is somewhat greater [0.003(4)A] than the V(CH)~~ data imply (O.OOl(2) A). E. Substitution ERects on Olefinic CH Bonds.-v(CH)iS data are available for the effects of methyl groups1 and halogen atoms43 on CH bonds in positions cis, trans, and a to the substituent, although the data are less accurate than those for saturated compounds since partial deuteriation does not in general remove resonances in the CH stretching region arising from v(C=C) and 8(=CH) levels. The reference compound, ethene, has a current v(CH)i* value of 3055 crn-l.,3 Table 6 Siibstitiieiit eflects oti olrfinic CH hoiids(lyh V(CH)~~ HC Ht Ha scx StX sax CH, -CHMeC 3039 3064 3013 -16 9 -42 CH, CHCld 3074 3072 3082 19 17 27 CH, CHBrd 3074 3057 3085 19 2 30 "reference compound C,H,, 3055 cm-1;21"for other halogenated ethylenes, see ref.43; Cref. 1 : "ef. 43. Table 6 shows experimental v(CH>~S values for propene, vinyl chloride, and bromide, and the S values obtained from them. S, for methyl is -42 cm-l, identical with its value in ethane.* The trans and cis effects of methyl are positive and negative, respectively, rather like the trans and gauche ones in saturated compounds. It is interesting to note that they predict, rather well, the observed frequency of 3050 cm-1 in but-1-ene [SCnfe+ Stsre = -7, V(CH)~~(pred.) = 3048 cm-I]. The a and cis effects of chlorine and bromine are also like the a and gauche effects in saturated compounds.The trans effects are however both positive. in contrast to the alkyl halides, although, as in the latter, StC1> StBr. Since no v(CH)iS data are available as yet for vinyl fluoride and iodide, the CH stretching frequencies of di- and tri-halogenated ethylenes have been examined for indirect evidence of substituent eirects.43 It seems very probable that the sequence StF> Stc1> StHr> St1is even more marked in ethylenes than it is in alkyl halides; also that fluorine markedly strengthens CH bonds in the a, cis, and trrins positions. The cis effect in an ethylene would of course be expected to resemble an ovtho effect in an arornitic compound. Bellamy and Mayo40 quote v(CH) data for benzenes which suggest a strong similarity with the ethylenes: e.g.for 1 : 3: 5 benzene-X3, X = D, 3053 cm-l ;F, 3108; C1, 3089; Br, 3090; Me, 3020 *With reference to the Bellamy-Mayo model for Sabre, the gauche H-H distance in propene is 0.035 A greater than that in ethane. 417 Individual CH Bond Strengths in Simple Orgunic Compounds cm-l. In these compounds the u(CH) modes are essentially isolated; however, each CH bond is subject to one para and two ortho effects of X. The similarity to Scis effects therefore suggests that Spars is either negligible, or similar to Sortho. The success of ah initio calculations of re in predicting the effect of methyl on =CH in propene can be seen in Table 1.However these calculations exhibit one curious feature; if the remaining geometrical parameters are constrained to their spectroscopic values, no differences in re(CH)emerge. Only when the C=C bond in particular is allowed to refine to an optimum 0.02 A less than the rs value, do the expected differences appear. Preliminary calculations on the vinyl halides do not merit space here. F. Methyl Groups Adjacent to Multiple Bonds : Hyperconjugation.-The data available so far are shown in Table 7, where Hs and Ha refer to atoms in and out of the skeletal plane respectively. In these four C=X compounds, CH, is always weaker than CHs. This might be attributable to hyperconjugation, since the latter, as normally conceived, involves a movement of charge out of an MO involving the Ha hydrogen into the rest of the molecule.Mul!ike~i~~ has estimated that this should weaken the CHa bonds by about 0.001 A, corresponding to about 10 cm-1 in v(CHa)iS.The Hs-HH. difference, however, is considerably more than this, especially in the C=O compounds. Ah itiitio calculations suggest that part of this may arise in a different manner. On the one hand, the net atomic popula- tion on Ha in propene is appreciably greater than that on Hs24although the gross populations are similar. On the other, the dipole moment ill propene is said to arise largely from a polarization of the 7~ bond away from the methyl groupS54 Together these findings suggest that the main effect in propene is a non-bonding electron pair repulsion between the 7~ electrons and the CHa bonding pairs.In acetaldehyde, both the net and gross populations are larger on Ha than on Hs, although hyperconjugation would have been expected to be greater than in propene. The tentative conclusion is that hyperconjugation, if it exists, is masked by other, stronger, influences. Loss of charge from the CHs atom seems to be correlated with a rise in CH bond strength. Where a methyl group is attached to a triple bond, all these CH bonds will be involved in either loss of charge, due to hyperconjugation, or in repulsion invol- ving the 7~ bonds. It is therefore interesting to observe that the CH bond is markedly stronger in CH3CN than it is in CH~CECH.The 4-31G basis set can reproduce the difference in bond length (Table 8) but only provided all the bonds are allowed to vary in length.Both net and gross populations on the hydrogens are larger in the acetylene. As would be expected intuitively, the CH bond is slightly weaker in CH3-C=C-CH3 than in CH3-C=CH. Returning briefly to the C=X compounds, we note that the effects of the second methyl group on the trutis and gauche CH bonds of the first one. are 53 R.S. Mulliken, C. A. Rieke, and W. G. Brown, J. Amer. Chem. SOC.,1941, 63, 41. L. Radom, W. A. Lathan, W. J. Hehre, and J. A. Pople, J. Amer. Chem. SOC.,1971, 93, 54 5339. 418 Table / v(CH)i-values jor methyl groups adjacent to rnirltiple borids A. double bonds v(CH)is/cm-ro(pred.)/8i r44-3 1G)/A n.a.p."/e g.a.p .(I /e D"298(pred.)/ kcal rnol-' CH3CH CH2 HS 2959 I .095(5) I .082' 0.540 0.838 100.4 HE& 2931 1.098(5) 1.085 0.555 0.836 98.0 Hs-Ha 28 -0.003 --0.003 -0.015 + 0.002 2.4 (CH3) 2C =CH 2 HS 2965 1.095 100.9 Ha 2924 I .099 97.4 Hs-Ha 41 -0.004 3.5 CH3CHO H5 3002 1.091 [ 1.07% 5)pd [0.4911" [0.794]" 104.1 H a, 2945" 1.097 [ 1.0835)]"J [OS 1 813 [0.8 1 3Id 99.2 Hs-Ha 57 -0.006 -0.006 -0.027 -0.109 4.9 CCH3)KO Hs 3 oQ4" 1.09 1 104.3 Ha 2946 1.097 99.3 Hs-Ha 58 -0.006 5.O B.triple bonds CH~CECH 2958" 1.095(5) 1.082( 8), 0.515 0.807 100.3 [1.082(3)]'*d 2950f 1.096(5) 99.6 2985s I .093 1.080(5), 0.476 0.780 102.6 [l .082]a3d 2989" 1.092(5) [ 1.08I (3)]a4 [0.498Id [0.795Id 103.0 aref. 24; n1g.a.p. =-net/gross atomic population; bref.I ; Cref.21 ; "unbracketed values: optimization of all bond lengths, bracketed values: optimization only of ~CH;eref. 7; fref. 55; gref. 9; "ref 56. 0 55 J. C. Lavalley, personal communication. 2;f: 56 D. C. McKean. J. L. Duncan, and M. Mackenzie, J. kfo/. Structure, 1977, 42, 77. 3\D Table 8 v(CH)~Sand CH bond strength variations in CH3-planar XY2 compoundso h v ib/cm-l v1b/cm-l A vc/cm-l AroC/A Ar,c/A A D029gc/ 3 kcal mol-1 3 CH3N02 3065 3006 59 -0.006 -0.005 5.1 5 CH3BF2 2997 2937 60 -0.006 -0.006 5.2 2 (CH3B0)3 2987 2927 60 -0.006 -0.006(4) 5.2 40 CH3BC12 2989 291 6 73 -0.007(5) 6.3 g(CH3)3B 2980 2884 95 -0.010 -0.007 8.2 2 Odata from ref. 57, 58; *v,, vl signify v(CHP with hydrogen respectively in skeletal plane, and at right angles to it; CAY= v//-vl,Ar = 711 -rL 9 etc.2;$ $" McKean normal for but-1-ene, (SPe= +6, Sg-\Ie= -7 cm-1) but that in acetone they are both zero within the experimental error. G. Methyl Groups Attached to Planar XY2 Skeletons.-A methyl group so attached experiences a very low potential barrier to its internal rotation, and the latter is essentially free at ordinary temperatures. The vibrational spectra of such molecules present unusual features, not all of which are understood, but the studies of CHDz substituted MeN02,57 MeBF2, MeBC12, (MeB0)3, and Me3B5* have left no doubt that the strength of the CH bond varies during the internal rotation, such that it is a maximum when in the plane of the skeleton, and a minimum when at right angles to it, with a total variation of V(CH)~~ between 60 and 95 cm-l, as shown in Table 8.While no CHDz data in the gas phase are yet available for toluene, the existing spectra59 give a sure indication of a similar alteration in strength with torsional angle. Here again, as in CH3-C=X compounds, hyperconjugation may be invoked to explain the weakness of CH, (out-of-plane), and it is certainly true that the difference v(CH) ,/ -v(CH), is greatest in BMe3, where the boron 2p, orbital is essentially empty. The population analyses from the ab iiritio studies of MallinsonG0 on MeN02, MeBF2, and Me3B are illuminating. In all these cases, HI,has a gross population 0.008-0.009 e greater than that of H,.Ther orbitals of the NO2, BF2, and B groups have gained from eachmethyl group, respectively, 0.021,0.032, and 0.032e. These, however, must be contrasted with gains in the (7 orbitals of 0.45 e for the NO., 0.43 e for BF., and a loss of 0.033 (per Me) for the boron in Me3B. Clearly the 7~ effect can be insignificant compared with the 0 one. Moreover, the CH overlap population is higher for H, than it is for HI in MeNOz and MeBF2, so that some other reason for v(CH),, > v(CH), must be found. One factor in MeN02 may be the marked positive overlap between H,, and the cis oxygen. This is almost completely absent in MeBF2, although the cis and trans effects of fluorine found in ethylenes (see Section 3E) should be present, affecting H,,here. However, for Me313, the CH overlap is slightly lower for H, than it is for HI,,so that it remains possible to postulate hyperconjugation as a factor contributing here to the largest v,,-vL difference found.4 Conclusion 'Isolated' CH stretching frequencies constitute a unique source of quantitative information on the strengths of individual CH bonds. The present survey, limited as it is," reveals patterns of conformational and substitutional effects which present new problems of chemical interpretation, although in part they are *The vast fields of aromatic and organometallic compounds are only represented by a single paper in each bi D. C. McKean and R. A. Watt, J. ,4401. Spectroscopj?, 1976, 61, 184. jMD. C. McKean, H. J. Becher. and F. Bramsiepe, Spectrochim.Acta, 1977, 33A, 951. A. B. Dempster, D. B. Powe!l, and N. Sheppard, Spectrockirn. Acta, 1975, 31A, 245. 6o P. D. Mallinson, personal communication. 61 E. D. Schmid and F. Langenbucher, Spectrochim. Acfu, 1966, 22, 1621; see also ref 40. 62 H. Burger and S. Biedermann, Spectrochini. Acta, 1972, 28A, 2283. 421 Individual CH Bond Strengths iii Simple Organic Conipoiiizds successfully predicted by quantum-mechanical methods. The new data have considerable implications also in the fields of chemical kinetics and of molecular geometry. I am greatly indebted to Dr. P. D. Mallinson for advice and assistance in the use of the Gaussian 70 programme 236 (W. J. Hehre, W. A. Lathan, R. Ditchfield, M. D. Newton, and J. A. Pople) Quantum Chemistry Package exchange, Indiana University, 1971, as modified by himself and Dr. H. B. Schlegel, and to Dr. L. J. Bellamy for stimulating discussions and communication in advance of his own work. 1 thank Professor N. Sheppard, F. R. S. for his interest and encouragement during this work, Dr. J. L. Duncan for helpful comment on the manuscript, and the University of London Computer Centre for a generous allocation of comput-ing time.

ISSN:0306-0012    

DOI:10.1039/CS9780700399

出版商:RSC    

年代:1978

数据来源: RSC