摘要:

QUARTERLY REVIEWS THE RATES OF SIMPLE ACID-BASE REACTIONS By R. P. BELL F.R.S. (BALLIOL COLLEGE OXFORD) IT is common experience that reactions between acids and bases in solution take place extremely rapidly and would often be loosely described as “instantaneous”. It might be thought that a reaction between a suffi- ciently weak acid and a sufficiently weak base would necessarily be slow since the activation energy cannot be less than the endothermicity. However in any direct kinetic observation the rate measured refers to the approach to equilibrium and this involves the velocity constants of both the forward and the reverse reaction for example for two opposed first- order reactions the approach to equilibrium follows a first-order law with a velocity constant equal to the sum of the velocityconstants for the forward and the reverse reaction.If the forward reaction is endothermic the reverse reaction will be exothermic and the observed change is again likely to be a very fast one.* Moreover if both the acid and the base are very weak the extent of their reaction at equilibrium will be small and it may not be easy to follow the change in the system. It is possible that the relatively high velocity of acid-base reactions is inherent in their nature as proton-transfer processes. The proton is a bare nucfeus without any electrons attached to it and in simple cases its transfer is attended by the minimum of disturbance in the rest of the molecule in particular it avoids the repulsion of unshared electrons which accom- panies the transfer of other ions and groups There are however a number of ways in which the velocities of acid-base reactions can be measured either by indirect means or by making use of various techniques which have been developed recently for the study of fast reactions.This Review deals with the different techniques employed and comments on some of the results obtained. The ionisation of carbon acids This subject will be treated briefly and separately since the velocities involved are comparatively low and most of the measurements have been made by older techniques. The group f C H possesses no detectable acidic properties in simple hydrocarbons and only very weak ones in acetylenic or aromatic molecules. However if it is combined with certain * In other words in the reaction Al f B2 + B1 + A2 %Al and B2 are both very weak the conjugate species B1 and A2 will be strong and the reverse reaction will be fast.1 69 . 1 70 QUARTERLY REVIEWS activating groups of which the most important are >C=O -NO2 and -CN the resulting molecules are acids of appreciable strength usually having pK values in the range 7-20. This increased acidity arises because the negative charge in the anion can now reside on an oxygen or a nitrogen atom rather than on the carbon atom from which the proton has been detached. Thus the groups (A) (B) and (C) )CH.C:O (4 give respectively ions )c k.0- ( A l l >HA<; )CH.C;N having the structures ( A l ) (BI) and (CI). )C:C:N- When this type of acid reacts with the base there is therefore a considerable degree of electronic rearrangement and it is not surprising that these reactions are much slower than those of simpler acids of the same strength.There are a few systems in which the rate of reaction of a carbon acid with a base can be observed directly by conventional methods. The classical example is the reaction of nitro-paraffins with hydroxyl ions first observed by Hantzsch,l and later studied by other workers.2 The reactions can be readily followed by the change in electrical conductivity and more recently a thermal method has been employed3 to extend the range of investigation to half-times of only a few seconds. The con- ductivity method was also used to study the neutralisation of nitro- paraffins by solutions of ammonia and amines,* and it was found that the observed velocity cannot be attributed entirely to reaction with the hydroxyl ions present but must also involve direct reaction with amine molecules e.g.CH3.N02 + R-NH -+ CH2:N02- + R.NH3+. This finding is of course analogous to general catalysis by acids or bases in catalysed reactions. Similar behaviour is found with some other nitro-compounds provided that the reaction is slowed down by reducing the temperature. Thus tri-p-nitrophenylmethane reacts at a measurable rate with ethoxide ions in alcohol at -6O" the neutralisation being readily followed by the A. Hantzsch and A. Veit Ber. 1899,32,615. Hantzsch supposed that the observed rate represented the change of the normal form of nitromethane (a pseudo-acid) into its mi-isomer (a "true acid") which then reacted rapidly with hydroxyl ions. It is not now believed that the aci- form plays any part in the change which is regarded as a direct reaction between hydroxyl ions and the nitromethane molecule.W. F. K. Wynne-Jones J. Chem. Phys. 1934 2 381; S. €3. Maron and V. K. LaMer J. Amer. Chem. Soc. 1938,60,2588; R. P. Bell and A. D. Norris J. 1941 118. R. P. Bell and J. C. Clunie Pruc. Roy. Soc. 1952 A 212 16. R. G. Pearson J. Amer. Chem. Soc. 1948,70 204. BELL RATES OF SIMPLE ACID-BASE REACTIONS 171 colour of the anion prod~ced.~ The charge distribution in this anion will be a complicated one but no doubt most of the negative charge resides on the six oxygen atoms. If a weak acid is added to a solution of the sodium salt the anion is reconverted into the hydrocarbon at a measurable rate and the dependence of the reaction velocity upon the composition of the solution shows that most of the anions react with undissociated acid molecules rather than with hydrogen ions.This is again analogous to general acid-catalysis and the rates of reaction with different acid molecules are closely related to their acidic strengths6 just as in the well-known Bronsted relation for catalysed reactions. Other reactions of similar types have been studied over a wide temperature range by Caldin and his collaborators. By far the commonest method of measuring the rate of ionisation of carbon acids is the study of base-catalysed reactions. Since the anions of these acids differ considerably in electronic structure from the undis- sociated molecules they can often undergo rapid decomposition re- arrangement or reaction with other species and the rate of this reaction is then a measure of the rate at which the substrate reacts with the basic catalyst to form the anion.This field of investigation has been fully documented8 and will not be dealt with here. The further reaction of the anion makes the ionisation process irreversible so that the slow forward reaction can be measured without complication by the rapid reverse reaction. The reaction of the anion with halogens (kinetically of zero order with respect to the halogen) has been widely used for this purpose and a compilation by Pearson and Dillong gives the rates of the reaction RH + H20 -+ R- + H,O+ for 37 carbon acids mostly determined by this method. When the equilibrium constant of this reaction is known it is possible to deduce the velocity constant of the very fast reverse process. Rates of hydrogen isotope exchange If the ionisation of a weak acid takes place in deuterium oxide or a similar deuterated solvent the reconversion of the anion into the acid will result in the introduction of deuterium into the molecule and the initial rate of deuteration is equal to the rate of ionisation.This method is particularly suited to slow rates of ionisation and in comparing it with other methods allowance must be made for the effect of isotopic substitu- tion in the acid or the solvent upon the rate. Many substances which undergo exchange at a negligible rate in neutral solution do so more G. N. Lewis and G. T. Seaborg J . Arner. Chem. Soc. 1939,61 1894. These authors gave a more complex explanation of the observed phenomena which seems very unlikely cf. M. Kilpatrick ibid. 1940 62 1094. M. Kilpatrick ref.5. E. F. Caldin and J. C. Trickett Trans. Faraday Soc. 1953 49 772; E. F. Caldin and G. Long Proc. Roy. Soc. 1955 A 228,263. * See e.g. R. P. Bell “Acid-Base Catalysis” Oxford 1941 ; Adv. Catalysis 1952 4 151. R. G. Pearson and R. L. Dillon J. Amer. Chem. Soc. 1953,75,2439. 172 QUARTERLY REVIEW§ rapidly in the presence of alkali and the acid-base reaction involved is then RH + OD- -+ R- + HDO. Rough measurements on a large number of carbon acids were made several years ago by Bonhoeffer,lO and recent work is exemplified by the investigations of Hine and his collaboratorsll on the rates of ionisation of the haloforms. In this last work the deuterium compound CDXYZ (where X Y and Z are halogens) was dissolved in an aqueous alkaline solution and the extent of conversion into CHXYZ determined from time to time by extraction with octane and examination of the infrared spectrum.As might be expected protons attached to oxygen undergo exchange with hydroxylic solvents too rapidly for measurement by isotopic ex- change methods but nitrogen is probably intermediate between carbon and oxygen in this respect and measurable exchange rates have been observed with the cations of ammonia and some amines. In the first work on this subject12 solutions of ND,NO in 54% aqueous nitric acid were precipitated after a short time by adding acetone and the isotopic composition of the precipitate was determined. At 0" c the exchange had half-times of 1-10 minutes and half-times up to 10 hours were observed in the later work1 in which the rate of exchange of alkylammonium salts with butanol in chloroform solution was studied by a similar method.A more detailed study of the same kind has been carried out by Swain14 on the rate of exchange of deuterium or tritium between alkylammonium ions and methanol or ethanol both species being dissolved in the non- exchanging solvent dimethylformamide. The observed rate was pro- portional to the concentrations of alkylammonium ion and alcohol and inversely proportional to the hydrogen-ion concentration showing that a proton has been lost in the transition state. This cannot be caused by a rapid pre-equilibrium NR3H+ + NR + H+ since this would itself cause isotopic exchange. The simplest assumption is that the proton is lost from the alcohol molecule the reactants being NR3H+ and MeO- (or EtO-); however Swain produces some evidence that the free ions MeO- and EtO- are not present in kinetically significant amounts and prefers the mechanism S + MeOH + NRSH+ + SH+ + MeO--H---NR,(Fast) MeO---H---NR -+ MeOH + NR (Slow) where S is the solvent and the intermediate complex is held together by hydrogen-bonding.A mechanism of this kind receives some support from the studies of proton magnetic resonance described in a later section. lo K. F. Bonhoeffer Trans. Faraday Soc. 1938,34 252. l1 J. Hine N. W. Burske M. Hine and P. B. Lanford J. Amer. Chem. SOC. 1957,79 l2 A. L. Brodskii and L. V. Sulima Doklady Akad. Nauk S.S.S.R. 1950 74 513. l3 L. Kaplan and K. E. Wilzbach J. Amer. Chem. SOC. 1954,76 2593. l4 C. G. Swain and M. M. Labes J. Amer. Chem. Soc. 1957,79 1084; C. G. Swain 1406 and earlier papers.3. T. McKnight and V. P. Kreiter ibid. p. 1088. BELL RATES OF SIMPLE ACID-BASE REACTIONS 173 Slow acid-base reactions involving N-H bonds are presumably also responsible for the change of conductivity with time which is observed when amines are dissolved in alcohols,15 though the interpretation of these results is not fully understood. The contrast between 0-H N-H and C-H acids is illustrated by some investigations by Bell and Pearson.16 With ethylenedinitramine the rate of the reaction CH2*N.N0,- CH2*N H-NO C H 2* N * N 0,- C H,*N.N 0,- + NH8-t I + NH,+ I is just measurable by flow techniques giving a velocity constant of about lo5 1. rnole-lsec.-l compared with 7 x 1. mole-lsec.-l for the corres- ponding reaction of nitroethane. On the other hand the neutralisation of the anion of p-nitrophenol by hydrogen ions was immeasurably fast (ta < lo-* sec.) even at low temperatures contradicting an earlier state- ment17 which has been widely quoted.The exchange of phosphine with deuterium oxide is a slow process which is catalysed by both hydrogen ions and hydroxyl ions.l* Phosphine has extremely weak acidic and basic properties in aqueous solution and there is little doubt that the velocities of two exchange processes are those of the reactions PH + D,O+ -j PH3D+ + D20 and PH + OD- -+ PH2- + HDO since PH,D will be formed when either of these reactions is reversed in a medium consisting mainly of D,O. By making reasonable assumptions about the velocities of these reverse processes the authors were able to estimate values of and for the otherwise unobtain- able acidic and basic dissociation constants of phosphine.On the whole isotopic exchange rates have not been used very extensively for measuring the rates of acid-base reactions since more convenient and accurate methods are often available. The same applies to the use of rates of racemisation because of the difficulty of preparing optically active compounds. On the other hand a comparison of the rate of isotopic exchange with that of other processes such as racemisation or chemical reaction has often given information about the details of reaction mechan- i s m ~ . ~ ~ The use of relaxation methods This type of method has been applied recently to a number of acid-base reactions especially by Eigen and his collaborators.20 It is necessary that the system should contain a state of physical or chemical equilibrium the position of which can be shifted by a change in some external parameter lli A.G. Ogston J. 1936,1023; J. R. Schaefgen M. S. Newman and F. H. Verhoek l6 R. P. Bell and R. G. Pearson J. 1953 3443. l7 G. E. K. Branch and J. Saxon-Deelman J. Amer. Chem. SOC. 1927 49 1765. R. W. Weston and J. Bigeleisen J. Arner. Chem. SOC. 1954 76 3074. l9 See e.g. C. K. Tngold “Structure and Mechanism in Organic Chemistry” London 2o For a general account see M. Eigen Discuss. Faraday SOC. 1954,17 194. J . Amer. Chem. SOC. 1944,66 1847. 1953 pp. 569-575. 174 QUARTERLY REVIEWS (electric field pressure or temperature). If the system is perturbed by a rapid change in this parameter the rate at which it changes to its new equilibrium state can be studied either by direct observation or by its interaction with the perturbing agent.Such systems are usually character- ised by a relaxation time T which is the time needed for the system to traverse a fraction l/e of its path to the new equilibrium. For chemical changes T is equal to the reciprocal of a first-order velocity constant and is of the same order of magnitude as the half-time of the reaction. The principles of the relaxation method can be illustrated by the application of the electric impulse technique for measuring the rates of acid-base reactions. The equilibrium degree of dissociation of a weak electrolyte is increased by the application of a strong electric field. This phenomenon is known as the dissociation field effect or the second Wien effect its magnitude can be calculated theoretically and has been con- firmed by experiment.If the field is changed suddenly the degree of dis- sociation a will not change immediately to its new value and Fig. 1 shows t how a will change during square-wave electric impulses of varying dura- tion. If the duration of the impulse t is much smaller than the relaxation time T the average value of a during the impulse will differ little from the low-field value while if t $ r the average a is close to the high-field value. A measurement of a when t -rr T should therefore make it possible to estimate T and hence the velocity constant of the dissociation process. Since the conductivity depends upon a the method used is to measure the high-field conductivity as a function of t . In practice it is often convenient to use a sine-wave impulse instead of a square one and the conductivity is measured by comparison with a strong electrolyte so as to eliminate that part of the Wien effect which is due to the ionic atmosphere.In order to obtain an appreciable change in a fields of the order of lo5 volt cm.-l must be used together with impulse times of 10-5-10-7 sec. which is comparable with the relaxation times involved. At such high field strengths it is in any case necessary to use very short impulses in order to avoid undue heating of the solution. The electric-impulse method has been used to measure the rates of the reactions MeC02H s Me.C02- + H+ and NH + H20 + NH,+ + OH-,21 and also to the self-dissociation of water.22 The results of these and 21 M. Eigen and J. Schon 2. Elektrochem. 1955 59 483.22 M. Eigen and L. de Maeyer Z. Elektrochern. 1955 59 986. BELL RATES OF SIMPLE ACID-BASE REACTIONS 175 other measurements are given in the last section of this article. Similarly measurements of the high-field conductivity of very pure ice gave values for the equilibrium and rate constants of the process H,O + H+ + OH- in the solid phase.23 Under these conditions the ions are removed so fast at high field strengths that the conductivity is determined by the rate at which fresh ions are formed by dissociation. Another application of high- field conductivities is to elucidate the state of carbon dioxide in aqueous In these solutions the equilibrium H2C03 + H+ + HCO,- will be affected by the field-strength but not the equilibrium H2C03 + H20 + CO,; moreover during the short time of the impulse the second equilibrium does not have time to readjust itself so that the observed effect relates solely to the dissociation of H2C03.The magnitude of the field effect can be predicted theoretically in terms of the ionic mobilities and the dissociation constant so that the observed effect leads to a value for the "true" dissociation constant of carbonic acid K(H,CO,) = [H+][HC03-]/[H2C0,] = 1-3 x at 25". Since the usual methods of measurement give the apparent constant K(C02) = [H+][HC0,-]/([C02] + [H2C03]) = 4.5 x lo-' the ratio of the two constants is equal to [H,C03]/[C02] = 0.0037 in satisfactory agreement with earlier less accurate values. Similar principles apply when a high-frequency alternating field is applied to a solution of a weak electrolyte. This is illustrated in Fig.2 I I ' I e) where the upper curve represents the variation with time of the field strength and the lower three plots the corresponding variation in the degree of dissociation a for different reaction velocities. Plot (a) represents a low velocity so that a differs negligibly from the value corresponding to zero field while for curve (b) the velocity is so high that o( follows the instantaneous field strength and varies in phase with it. The case of most interest is when the half-time of the reaction is of the same order of 23 M. Eigen and L. de Maeyer 2. EEektrochem. 1956 60 1037. 2c D. Berg and A. Patterson J. Amer. Chem. SOC. 1953,7§ 5197; D. M. French and A. Patterson J. Phys. Chem. 1954,58 693. 176 QUARTERLY REVIEWS magnitude as the periodic time of the field we then obtain a curve such as (c) which has a smaller amplitude than (b) and is now out of phase with the applied field.As was first pointed out by P e a r ~ o n ~ ~ the conductivity should vary with frequency in this region but the effect will be a very small one since only low field strengths can be employed if heating effects are to be avoided. A more promising experimental approach is to measure the dielectric loss which arises from the phase difference between a and the applied field this makes the process partially irreversible and leads to a dissipation of electrical energy as heat in excess of the normal power loss due to ionic migration. This method has recently been used to measure the rate constants for the reaction H,BO + H2B03- + H+ and should be widely applicable.26 Analogous considerations apply to the effect of ultrasonic vibrations on a solution of an incompletely dissociated electrolyte.The primary effect here is due to the displacement of the equilibrium by the oscillating pressure of the sound wave according to the equation a In K/ap = &RT but since the conditions are adiabatic rather than isothermal the temperature oscillations may also contribute to the observed effects. In principle it would be equally informative to study the frequency dependence of the velocity of sound (usually observed by determining the wavelength corresponding to a given frequency) of the heat produced or of the attenuation of the sound-waves in practice the last is the most convenient. It is of course important to eliminate other factors which can lead to ultrasonic dispersion such as the finite rate of energy exchange between different degrees of freedom which can be done in principle by investigat- ing a sufficiently wide range of frequencies and concentrations.Not many acid-base systems have been studied but the method has been used to obtain a value of 4 x 1O1O 1. mole-lsec.-l for the velocity constant of the reaction H+ +- -+ HSOp.27 It is noteworthy that this reaction is much faster than reactions of the type M2+ + S042- -f MS04 where M is a metal which have velocity constants in the range lo4-lo6. This is probably because the association of the sulphate ion with a metal cation requires the removal of one or more water molecules from its hydration shell while in the formation of HS04- from SO,2- the proton is derived from one of the water molecules originally in contact with the anion the charge being handed on by the same mechanism which is responsible for the abnormal mobility of the hydrogen ion in water.Polarographic measurement of reaction rates The polarograph can be used in some cases for the study of a variety of rate processes.28 Under ordinary conditions the limiting current observed 25 R. G. Pearson Discuss. Faraday Soc. 1954 17 187. 26 W. R. Gilkerson J. Gem. Phys. 1957,27 914. 27 M. Eigen G. Kurtze and K. Tamm 2. Elektrochem. 1953 57 103. 28 For a general account of reaction kinetics in polarography see P. Delahay “New Tnstrumental Methods in Electrochemistry” New York 1954 Chapter V. BELL RATES OF SIMPLE ACID-BASE REACTIONS 177 at a dropping or rotating cathode (corresponding to the flat part of the polarographic curve) is controlled by the rate at which the reducible species can diffuse to the electrode and is proportional to the concentra- tion of this species.However if a small quantity of a reducible species is in chemical equilibrium with a second species which is not reducible then under suitable conditions of drop-rate concentration etc. the observed current may be controlled by the rate of the chemical process producing the reducibfe species. Because of the complicated diffusion conditions it is difficult to obtain an exact mathematical solution of the kinetic problem involved but approximate absolute values of velocity constants can be easily obtained and also fairly accurate relative values for a series of similar reactions. The first use of this method was for aqueous solutions of f~rmaldehyde,~~ in which the equilibrium CH2(OH)2 + CH20 + H20 is far over to the left and only the unhydrated formaldehyde is reducible.It is usually found that only one member of an acid-base pair is reducible at the cathode for example undissociated pyruvic acid MeCO-CO,H is reducible while its anion is not. Polarographic m e a s ~ r e m e n t s ~ ~ ~ ~ therefore make it possible to measure the velocity of the very fast reaction MeCOC0,- + H+ -+Me-COCO,H. Similar measurements have been made with phenylglyoxylic acid and its derivative^.^^ By working over a range of conditions values were obtained for the rate of reaction of the anion with hydrogen ions and also for its much slower reaction with weakly acidic species such as H,O and H3B03.The above procedure is limited to the rather small class of reducible acids and bases but the scope of the polarographic method can be greatly extended by using a somewhat different principle. For example the cathodic reduction of azobenzene takes place according to the scheme PhN :NPh + 2H+ + 2e -+ PhNH-NHPh the hydrogen ions being commonly supplied by a buffer system A + B + H+. Under suitable conditions the polarographic current is determined by the rate at which hydrogen ions are produced and since the equilibrium constant is known the velocity constant of the reverse reaction B + H+ -+ A can also be determined. In principle this method is applicable to any non-reducible buffer system A-B (though in practice there are quantitative limitations) but so far only formic and acetic acid have been s t ~ d i e d .~ ~ ~ ~ Because of the low solubility of azobenzene in water measurements were made in 50 % alcohol but presumably other reducible substances could be used. 29 R. Brdicka Coll. Trav. Czech. Chim. 1947 12 213. 30 P. Ruetschi and G. Triimpler Helv. Chim. Acta. 1952 35 1957. 31 P. Delahay and T. J. Adams J. Amer. Chem. SOC. 1952,74 1437. 32 K. Wiesner M. S. Wheatley and J. M. Los J. Amer. Chem. Soc. 1954,76 4858; 33 P. Delahay and W. Vielstich J. Amer. Chem. SOC. 1955 77 4955. 34 P. Ruetschi. 2. phys. Chem. (Fr~nkfurt)~ 1956,5 323. M. S. Wheatley Experientia 1956,12,339. 178 QUARTERLY REVIEWS Nuclear magnetic resonance35 This method of investigation depends essentially on measuring the frequency of the radiation absorbed in transitions between different orientations in a magnetic field of a nucleus possessing a magnetic moment for example the proton.The exact frequency absorbed and also the fine structure of the absorption line depend upon the environment of the nucleus and although the differences are extremely small the resolving power in the radio-frequency range is ample for detecting and measuring them. If a solution contains two sets of protons in different environments the observed spectrum depends upon the rate at which the two sets can interchange. If the interchange is slow compared with the frequency used for making measurements two distinct peaks will be observed in the spectrum but if it is fast there will be only a single peak at a position between the two. If either the rate of interchange or the observing fre- quency is varied continuously there will be an intermediate range in which the two separate peaks broaden and coalesce and in this range the mean life of the proton in one of its situations is of the same order of magnitude as l / u where u is the observing frequency.Similarly the splitting of the lines into a number of components (which is due to the interactions between the observed nucleus and other magnetic nuclei) is lost if the mean life-time is much smaller than l / v and will be restored if either the life-time or the frequency is increased. In principle therefore the velocity of the interchange can be determined by observing the changes in the spectrum of a given solution when the frequency is progressively varied or (which is experimentally more convenient) when the frequency is held constant and the reaction velocity varied by altering some property of the solution such as the pH or the concentration.This method offers great possibilities for measuring the velocities of rapid acid-base reactions and in particular makes it possible to determine the rate of symmetrical proton-transfers such as NH4+ + NHs + NH + NH4+. So far however most of the kinetic information obtained has been qualitative or semiquantitative in nature. This is partly because of experi- mental difficulties but also because the relation between the observed pattern and the velocity constants involved is complicated to derive and difficult to apply in practice. A recent paper by McConnelP* gives a simple derivation which should be useful in interpreting future observations.Some of the earliest applications to acid-base kinetics are due to Ogg3' He showed that in very pure liquid ammonia the proton-resonance line shows a triplet structure due to interaction with the nucleus l*N but that this structure disappears on introducing very small amounts of the ions NH2- or NH,+ the latter by adding a trace of water. This is undoubtedly 35 For a general account of this technique see J. A. S. Smith Quart. Rev. 1953 7 279; R. E. Richards ibid. 1956 10 480. 36 H. M. McConnell J. Chem. Phys. 1958,28,430. s7 R. A. Ogg J. Chem. Phys. 1954,22 560; Discuss Faraday SOC. 1954,17 215. BELL RATES OF SIMPLE ACID-BASE REACTIONS 179 due to the exchange reactions NH + NH2- + NH2- + NH3 and NH + NH4+ NH4+ + NH, but only a very rough estimate of the velocity constants could be obtained partly because the very low ionic concentra- tions were not well defined.Similar results were obtained by observing the nuclear resonance of the 14N nucleus rather than of the proton. Ogg also showed that an acid solution of an ammonium salt gave a triplet due to 14NH4+ and a singlet due to H20 but that in neutral solution these were merged into a single peak by the occurrence of exchange reactions with NH or OH-. A quantitative study of the last system has been recently carried out by Meiboom Loewenstein and their collaborators who observed the proton magnetic resonance spectra of solutions of ammonium salts over a range of pH and c~ncentration.~~ The reactions considered (after excluding some improbable ones) were the following (1 1 NH,+ + H,O + NH + H,O+ k (2) NH4+ + OH- -+ NH + H20 k2 (3) NH4++ NH + NH,+ NH,+ k3 (4) NH4+ + 0-H + NH -+ NHg + H-0 + NH*+ k I H I H In strongly acid solution there is a sharp triplet due to NH4+ and a singlet due to H20 and the width of these lines gives a maximum value kl < 6 x 1.mole-lsec.-l this is consistent with the approximate values obtained from isotopic exchange experiments. In the range pH = 15-2-5 the triplet broadens and coalesces and the dependence of the broadening upon the ammonia concentration shows that this must be caused by reactions (3) or (4) rather than by NH4+ + OH-. The concentra- tion of hydroxyl ions is actually so low that the last reaction would not be detected unless k2 were greater than 10l2 1. mole-lsec.-l this is greater than the maximum value expected theoretically and the electrical impulse method (ref.21) gives k2 = 3 x 1O1O 1. mole-lsec.-l. Most of the NH4+ broadening is attributed to reaction (3) for which k3 = 1.1 x lo9 but since it is accompanied by some broadening of the water line reaction (4) is believed to occur simultaneously with about one-tenth of the velocity. The same authors have studied the three methylamines in the same way,39 and for these systems confirmatory evidence is obtained by observing the splitting of the peak from the CH8 protons which is due to the protons attached to the nitrogen atom.* The results obtained are similar to those for ammonia but there are some interesting quantitative differences. Thus the relative importance of reaction (4) involving a water molecule 38 S. Meiboom A. Loewenstein and S. Alexander J. Chem.Phys. 1958 29 969. 3s E. Grunwald A. Loewenstein and S. Meiboom J. Chem. Phys. 1957 27 630; A. Loewenstein and S. Meiboom ibid. p. 1067. * It is not immediately obvious why the CH peak is not further split by the 14N nucleus but a quantitative theoretical treatment shows that this further splitting should not be detectable. 180 QUARTERLY REVIEWS increases with methyl substitution and for trimethylamine the direct exchange reaction (3) is undetectable. It seems likely that solvent molecules may take part in inany other acid-base reactions but this cannot be de- tected by any of the usual means of observation. Trimethylamine also differs from ammonia and the other methylamines in that reaction (1) now makes a detectable contribution with kl = 6 x 1. mole-lsec.-l. A small amount of information has also been obtained about the ex- change of protons attached to oxygen.Thus found that in “pure” ethyl alcohol the life-time of the -OH proton was about 1 sec. but that this was reduced to about 0.01 sec. in either 10-5~-a~id or 10-4~-alkali. This suggests velocity constants in the range 106-107 1. mole-lsec.-l for exchange reactions such as EtOH + EtOH2+ e EtOH,+ + EtOH and EtOH + EtO- + EtO- + EtOH. More quantitative information has been obtained by Loewenstein and Meiboom for aqueous solutions of hydrogen peroxide,41 in which the main process leading to exchange is a reaction between H02- and H20 probably also involving a second water molecule. The study of nuclear magnetic resonance clearly offers great possibilities in acid-base kinetics as it does in so many other fields and only a small fraction of these have been explored.Acid-base reactions of excited species The fast acid-base reactions of electronically excited species have been recently investigated by Weller,42 using measurements of fluorescence. Most of these measurements relate to /3-naphthol which is a weak acid (pK 10) but is converted into a much stronger acid (pK 3) on absorbing ultraviolet radiation. Both the excited naphthol molecule and the anion formed from it can lose energy as fluorescent radiation and the relative intensities of the two types of fluorescence depend upon the rate at which the molecule is converted into the anion. If A-B is any acid-base pair present in solution the complete kinetic scheme is as follows (i) (ii) ( i i i ) ROH* -+ ROH + hv’ (Ultraviolet fluorescence) (iv) RO-* -+ RO- + hv (Blue fluorescence) ROH + hv 4 ROH* ROH* + B + RO-* + A (v) ROH” -+ ROH } Quenching fvi) RO-* ,+ RO- In practice the reverse of reaction (ii) can be neglected and the velocity constant for ROH* + B can be evaluated in terms of the observed in- tensities of the two kinds of fluorescence and the mean life-times of the excited species.The latter are of the order sec. and are either obtained *O J. T. Arnold Phys. Rev. 1956 102 136. *l M. Anbar A. Loewenstein and S. Meiboom J. Amer. Chem. SOC. 1958,80,2630. O2 A. Weller 2. Elektrochem. 1952 56 662; 1954 58 849; 1956,60 1144; 1957,61 956; 2. phys. Chem. (Frankfurt) 1955 3 238. BELL RATES OF STMPLE ACID-BASE REACTIONS 181 from other experiments or estimated theoretically from the extinction coefficients.The base B can be either a water molecule or a constituent of a buffer solution (e.g. acetate ion) and the velocity constants are about lo9 1. mole-lsec.-l. Similar measurements were made for excited acridine and its cation. Some experimental results In contrast to the wealth of information for carbon acids (ref. 9) our knowledge of the more rapid reactions of nitrogen and oxygen acids is still fragmentary and the accuracy of the results is not high. Some of the values are given in the Table together with a few results for carbon acids for comparison. The first part of the Table shows that although the strengths of the acids vary over 14 powers of ten the velocity constant k for the reverse reaction B +- H30+ remains almost constant at about 3 x 1O1O 1. mole-lsec.-l.(It may be significant that the largest deviations from this mean value are shown by the polarographic data for which there is the largest uncertainty in the absolute values of the constants.) This suggests strongly that none of these reverse reactions possesses any appreciable energy of activation the rate being determined by the rate at which the reacting species can come together. The value of 3 x 1O'O is in fact in good agreement with theoretical expressions for the rate of a diffusion- controlled reaction between two ions.43 Velocities of acid-base reactions (aqueous solutions at 18-25 O C) kl A (+ H,O) + B + H30+ b Acid Method Ref. pK iogk log k 2 H2O Electric impulse 22 15.7 -4.6 11.1 resonance 39 9.8 1.0 10.8 Me,NH+ Proton magnetic H,BO Dielectric loss 26 9.1 1.0 10.1 MeC0,H Polarograph y 33,34 5.4" 5.4* 10.8" H-CO ZH Polarograph y 33 4*3* 4.7" 9*0* Electric impulse 21 4.8 5-9 10.7 CloHS.OH* Fluorescence 42 3.1 7.6 10.7 MeCO-CO,H Polarograph y 30,31 2.6 6-3 8.9 HSO4- Ultrasonic 27 1.6 9.0 10-6 Ph*CO*CO .H Polarograph y 32 1.4 10.2 11.6 MeCOMe Bromination 9 20 -9.3 10.7 MeCO-CH2-N02 9 CHdNO,) >* kl in sec.-l k in 1.mole-lsec.-l 9 13.3 -4.6 8.7 9 11.2 -1.8 9.4 9 9.0 -1.8 7.2 9 8.6 -7.4 1.2 9 5.1 -1.4 3.7 9 3.6 -0.1 3.5 * Denotes value in 50% alcohol. 43 P. Debye Trans. Electrochem. SOC. 1942,82,265; L. Onsager J. Chem. Phys. 1934 2 599. 182 QUARTERLY REVIEWS Since k is roughly independent of the acid-base strength of the pair A-B the forward rate kl must be approximately proportional to the acid dissociation constant of A. These two types of behaviour correspond respectively to /3 = 0 and a = 1 in the equations of the type k = G P kb = G,(l/K)B which relate reaction velocity to acid-base strength in reactions catalysed by acids and bases.It is noteworthy that when these relations were first put forward in 1924 Bronsted and Pedersen predicted that these extreme values would occur in very fast reactions although it is only recently that experimental evidence has been f~rthcoming.~~ The position is different for the carbon acids in the second half of the Table. Although k approaches the maximum value for the very weakest acids it is usually at least several powers of ten below this and its value depends not only on the pK of the system but also on the chemical nature of the activating groups in the molecule. This is undoubtedly because of the considerable and varied electronic rearrangements which occur in the ionisation of this type of acid in contrast to the simple processes involved in the first half of the Table. It would be of interest to extend rate measure- ments for acid-base reactions to a wider range of chemical types especially since there has recently been renewed interest in the theory of diffusion- controlled reactions.45 44 J. N. Bronsted and K. J. Pedersen 2. phys. Chem. 1924,108 185. 45 See e.g. R. M. Noyes J. Chem. Phys. 1954,22 1349; J. Amer. Chem. SOC. 1956 78,5486; A. Weller 2. phys. Chem. (Frankfurt),l957,13,335.

ISSN:0009-2681    

DOI:10.1039/QR9591300169

出版商:RSC    

年代:1959

数据来源: RSC

摘要:

MOLECULAR QUADRUPOLE MOMENTS By A. D. BUCKINGHAM M.SC.(SYDNEY) PH.D.(CAMBRIDGE) M.A.(OXFORD) (INORGANIC CHEMISTRY LABORATORY OXFORD) Introduction.-Any system of electric charges has associated with it a set of electric multipole moments (charge dipole quadrupole octopole . . . 2”-pole. . .). The potential of the electric field at any point outside the distribution of charges and arising from it is simply related to these moments and so is the energy of interaction of the system with an external field. Multipole moments have proved useful in investigations into the nature of intermolecular forces and have therefore helped in the search for an understanding of the properties of imperfect gases liquids and solids. Measurements of the dipole and quadrupole moments of molecules are also very important from the structural point of view.The interpretation of dipole moments is well known (see for example the books by Debye,l Smyth,2 Le F i ~ r e ~ and Smith,4 and the chapter by Sutton in Braude and Nachod5). Similarly molecular quadrupole moments can lead to important structural information; thus just as the absence of a dipole must mean that C 0 2 is a linear molecule so the absence of a quadrupole must mean that CHI is tetrahedral; also just as the knowledge that the dipole moment of H20 is 1.84 x e.s.u. (Le. 1.84 D) leads to conclusions about the charge distribution in the molecule so too does the quadrupole moment of C 0 2 (-3 x e.s.u.) tell us much about the C-0 bonds in this mole- cule. The Electrostatic Potential of Two Point Charges.-Consider the poten- tial 4 at an arbitrary point P (represented by the polar co-ordinates R 6) due to charges el and e2 at points distant z1 and z2 from an origin 0 (see Fig.1) P c FIG. 1 The position of the arbitrary point P relative to the origin 0 and the point charges el and ea. Debye “Polar Molecules” Chem. Catalog Co. New York 1929. Le Fkvre “Dipole Moments” Methuen London 1953. Smith “Electric Dipole Moments” Butterworths London 1955. Braude and Nachod “Determination of Organic Structures by Physical Methods” a Smyth “Dielectric Behaviour and Structure” McGraw-Hill New York 1955. Academic Press Inc. New York 1955 chap. 9. 183 184 QUARTERLY REVlEWS el[R2 - 2Rz1 cos 6 + ~ ~ 2 1 - 4 $- e,[R2 - ~Rz cos 8 +-+-= el e r1 r2 . . . . . . . + z,2]-6 (1) If R is greater than z1 and z, then the expressions for rl-l and r2-l can be expanded in powers of z,/R and z,/R leading to + + .. . * (2) el + e2 cos e 3 cos2 e - I $== R + (‘1’1 + e2z2> F+ + e2z22) - 2R3 5 cos3 6 - 3 cos e 2 ~4 . . . + (e,z,3 -t- e,z,3) (el + e,) is the zeroth moment (the charge) of the system of two charges (e,z + e,z3 is its first moment (the dipole moment) (e,z12 + e,z,,) its second moment (the quadrupole moment) and (e,z13 + e ~ ~ ) its third moment (the octopole moment); all are defined relative to the origin 0. $ is seen to be equal to the sum of the potentials of a point charge [whose magnitude is (el -k e,)] and a point dipole quadrupole octopole and higher moments; thus the name “quadrupole” does not necessarily imply that four or more charges are present. As is well known the dipole moment of an uncharged body can be thought of as being formed by separating positive and negative charges the magnitude of the dipole being the product of the charge and the separation.Similarly the quadrupole moment of a system with zero dipole moment can be thought of as arising from a separation of equal and opposite dipoles the magnitude of the quadrupole being proportional to the product of the dipole moment and the separation. Thus the linear carbon dioxide molecule which may be supposed to tend towards the structure 0-C-0 possesses a quadrupole moment the two C-0 dipoles being opposed to but separated from one another. Unfortunately several different definitions of the quadrupole and higher moments are in use so that care is needed in applying formulae in the literature. To clarify the multipole-moment concept the zeroth first and second moments of mass and of charge of a linear array of particles will now be discussed.Let mi and ei be the mass and charge of the ith particle distant zi along the z-axis from the origin 0. The zeroth moments are mi = M and < ei = q (where stands for a summation over all the particles of the array) and are the total mass and total charge of all the particles. The first moments about 0 are and qeizl = p,. If 0 is the centre of gravity of the system then mizi vanishes. To determine the effect on the moments of a change of origin consider the dipole moment t ~ ’ ~ relative to an origin Of where 0‘ is the point - 2. Clearly ... at 0. It should be noted that the charges el and e have a quadrupole - ++ - + - pLLlZ = C ei (zi + 2) = p z -f qZ .(3) so that the dipole moment is independent of the position of the origin only BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 185 if y is zero that is if the system is electrically neutral. The reason why a centre of mass but not a centre of charge can always be found is that negative masses do not exist so that no real system can have M = 0 however eqn. (3) shows that ions will have a centre of charge but if this 0' 0 ei mi i-- I _I__--__ - - - @ t- Z 3t - zi -+ FIG. 2 The co-ordinate system for the linear array of particles of mass m and charge e . does not correspond to the centre of mass 0 then the ion will experience a torque (as well as a force) in a uniform electric field. For convenience the centre of mass is normally chosen as the origin; thus the hydroxide ion OH- and the acetate ion CH,*C02- will possess dipole moments although NO,+ NH4+ and S042- will not.mizi2 = r,, and is the well-known moment of inertia. Similarly the second moment of charge about 0 is f eizi2 = O, and is the quadrupole moment. Only the first non-zero electric multipole moment is independent of the origin SO that O, will be independent of 0 only if both q and pa vanish. The Interaction of a Charge Distribution With an External Fie1d.- Consider a distribution of charges ei at points (xi,yi,zt) represented by the vectors ri (corresponding to scalars ri) from the origin 0 to el. Suppose that the distribution is in an external field produced by distant charges. We wish to find the energy of interaction u of the distribution with the external field. If the potential of the external field at rl is +$ then The second moment of mass of the linear array about 0 is u = C e & c .. . . . . . . . . (4) +# can be written in terms of the potential and its derivatives at 0; thus The subscript 0 denotes a value at 0. Eqn. ( 5 ) can be put into a short-hand tensor notation as follcws Greek suffixes denote tensor components (thus ria can be xi yi or zi) and 186 QUARTERLY REVIEWS repeated suffixes imply a summation over all components (thus riurf = tis = xi2 + yi2 + q2). On introduction of the parameters q = pi 7 eqn. (6) becomes where Thus Fuis the a-component of the external field at 0 FUB the a/3-component of the field gradient at 0 etc. Clearly FUp F’’cray . . . and Qa, &By . . . are symmetric in all suffixes (that is they are unaffected by interchanging suffixes).Some authors (for example Condon and Shortley in “The Theory of Atomic Spectra” Cambridge University Press 1953 p. 85) call Qap RUpy . . . the quadrupole octopole . . . moments of the charge distri- bution but we shall not do so and our moments describe departures from spherical symmetry (Q, is not zero for a sphere). Eqn. (8) can be simplified by using Laplace’s equation F’uu = F’, + FIWv + = 0 and u = q$o - paFa - &Ou,F‘, - ~Q,~yF1‘,BY - . . . . (10) and q is the charge of the distribution pu is its dipole moment 6, its quadrupole moment and QUBy its octopole moment. These multipole- moment tensors are symmetric in all suffixes. As with moments of inertia it is always possible to find three mutually perpendicular axes such that O, = 0 if a # 18. There are therefore three principal quadrupole moments O,, Oyy O,, but since [from eqn.(9)] O, = 0 only two are independent and need be specified to describe the BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 187 interaction of the system with a field gradient. [Eqn. (9) also requires that QaaS = 0.1 From the definitions (7) and (9) it is clear that * (11) I O, = 8 + ei(3zi2 - r t ) = +'eiri2P2(z,/r,) O, = +'eiri2P,(xi JrJ O, = +'eiri2P2(yi/ri) Q z z z = 9 Pi(5zi3 - 3zir:) = Ceir?P3(zl/ri) where P is the nth Legendre polynomial. Some authors (for example Hirschfelder Curtiss and Birds) define a quadrupole moment as 2 &'eiri2P2(zi/ri) and in nuclear physics tbe conventional definition is 2e-'piri2P2(zJrd) where e (= 4.803 x 10-lo e.s.u.) is the protonic charge. For a continuous charge distribution O, = 4 (3z2 - r2)p d v = r2P2(z/r)p dv .. . (12) I 1 where p dv is the charge in the volume element dv at r. - becomes If the charge distribution is symmetric about the z-axis @, = @, = = - 80 and Q,, = Qsuu = - $Qzzz = - +Q and eqn. (10) u = & - pF - Q O F f Z E - & Q F",,z - . . . . (13) Thus only one dipole one quadrupole one octopole and one multipole of any order' are required to specify the interaction of an axially sym- metric charge system with an arbitrary electric field. From eqns. (10) and (13) it can be seen that the energy of a charge system in an electrostatic field is the sum of the energies of a point charge in a potential a dipole in a uniform field a quadrupole in a field gradient and an octopole in the gradient of a field-gradient. In general there will be a couple acting on a rigid charged body in an electric field the torque about any point being proportional to the multipole moments about that point as origin.The Potential produced by a Charge Distribution.-Again consider a distribution of charges ei at points (xz yr zi) represented by the vectors re from an origin 0. We wish to find the potential 4 produced by the charges at an arbitrary point P at (X Y 2) denoted by R where R > ri for all i. Then 4 = zy$ = piRi-1 = P J ( X - xiy + ( Y - yJ2 + 6 (2 - Zi)"]-* . . . . . . . . . (14) * Hirschfelder Curtiss and Bird "Molecular Theory of Gases and Liquids" Wiley New York 1954 p. 839. Jansen Physica 1957,23 599. 188 QUARTERLY REVIEWS where Ri is the distance between the charge ei and P. We can expand Ri-l in terms of its derivatives with respect to ria at 0.Thus 1 ( a3(1'Ri) ) ' 8riaariBariy a B y Ri2 = ( R - ria) (R - ria) ri ri ri + . . . and since then From (7) (9) and (15) +=x+ ' EIaRu - + 5 ' ( 3 R a R - R2SSmp) + R3 3R5 For an axially symmetric charge distribution each multipole moment is determined by a single scalar quantity (viz. q p 0 Q . . .); for example 0 can be expressed in terms of its principal moments O, = 0 @, = a, = - +@ as where I is a component of the unit vector along the axis of symmetry (the z-axis) and ma and n are unit vectors in the x and y directions. The multipole moment tensors for axially symmetric systems are @,p = 4 @ [31aI - ( 1 J p + mamp + n,ns)] = 4 0 [31,1B - 4 = 4 PCY = Plu 1 . . . (17) o, = + 0 (31Jp - Sup) Qapy = +Q[51,Zply - la Soy - lfi Sy - I y 6,,] so that eqn.(16) becomes y pcos 8 0 I-2 2 R 2114 +x +Ra+-3 (3 cOs2 o - 1 ) +- (5 cos3 e - 3cos 8) +. . . . . (18) where P is now the point (R 0) relative to an origin on the axis of the distribution. Eqns. (16) and (18) show that the potential at any point BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 180 outside an arbitrary charge distribution is the sum of those due to a charge a dipole a quadrupole an octopole etc. located at the origin. Actually gaseous molecules cannot be completely orientated; if the rotational quantum number of a linear molecule is J and if the component of its angular momentum in a fixed direction is Mh/2n then there is a further component (J2 + J - M2)*h/2n at right-angles to the fixed direc- tion but its orientation in this plane is uncertain. The potential +J,M at the point distant R from the molecule’s centre of gravity and at a polar angle a with the fixed direction is where cos 0 = cos G cos ,$ + sin (T sin ,$ cos 7 and #J,M = PIMIJ (cos nexp (iMq) is the rotational wave function.In the states M = J where J is large the quadrupolar potential of eqn. (1 9) is approximately minus one-half of the “classical” value given by eqn. (18) with 8 replaced by a; this is because when it is rapidly rotating the molecule effectively becomes symmetric about its axis of rotation and this is at right angles to its own axis of symmetry . The electric field F at P can be obtained by differentiating eqn. (18). The radial component is Fr and the components at right angles to R are Fs and Ft (see Fig. 3) where FIG. 3 The components of the field at a point P due to an axially symmetric charge distribution.1 I 2 SZ (5 C O S ~ 0 - 3 cos 0) + . . . R5 +- 3~(5cos28-1) --+...I I +- 2 ~ 5 I J The fields arising from the various multipoles are illustrated in Fig. 4 where lines of constant F are shown. 190 QUARTERLY REVIEWS 8 83 (4 (6) (c) (4 FIG. 4 Lines of constant electric field strength emanating from (a) a charge their equation being RW2 = constant; (b) a dipole ) , , (3 cos2 tl + l)B/R3 = constant; (c) a quadrupole ) , (5 C O S ~ 9 - 2 cos2 8 + l)+/R4 = constant; ( d ) an octopole I , (175 cos* e - 165 COP e + 45 cOS~ e + 9 ) q ~ 5 = constant. The field gradient FfaS at P has the following components FfTT == - - - . . . - 2q 6 ,u cos 6 R3 R4 R5 + + 12 0 sin 8 cos 8 6 0 (3 cos2 8 - 1) q R3 R3 F',.s = - Fist = FftT = 0 3 p cos 6 3 0 (7 cos2 6 - 3 ) 2 ~ 5 2 ~ 5 - + .. . + . . . R4 3 ,U cos 6 R4 3 p sin 8 F'ss = -+ 3 0 ( 5 C O S ~ 6 - 1) F'tt = -+ . . . - - R4 R5 The Interaction Energy of Two Charge Distributions.-Consider two axially symmetric charge distributions 1 and 2 in the relative configuration illustrated in Fig. 5. The angle fs is that between the planes formed FIG. 5 A configuration of an interacting pair of axially symmetric charge distributionr (the arrows indicate the dipolar directions). by the axes of molecules 1 and 2 with the line of centres. If the centres of mass are sufficiently far apart (R must be greater than the sum of the distances of the furthest charges from their origins) the interaction potential energy u12 is given by a relation similar to (13) where q p 8 Q . .. are the multipole moments of the charge distribution 2 and+ Fs, BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 191 F'zz F"azz . . . are the potential and its derivatives at the centre of mass of 2 due to the charges of 1; alternatively the rales of 1 and 2 could be re- versed. Hence From (18) (20) (21) and (22) U l 2 = q 2 4 2 - p,F,r - 8 @ 2 F ' 2 Z z - & Q2F"2z1z - . . . ' (22) q1q2 1 1 u12= -R + R2 [q1p2 cos e,+ q2p1 cos ell + 2R3 [ql 0 (3 cos2 e - 1) + q2 o (3 cos2 el - 111 + - p1 p 2 [2 cos el cos e + sin el sin e cos 51 + - 2R4 [ql Q (5 cOs3 13 - 3 cos 0,) + q2 Q~ (5 cos3 el - 3 cos el)] + R3 I 3 2 R4 - [p102(c0~ 8 (3 cos2 0 - I) + 2 sin 8 sin 19 cos 8 cos 5 ] + p 2 O,{cos 8 (3 COS 8 - 1) + 2 sin 0 sin 19 cos 8 cos 5 >] 3 olo [I - 5 cos 2 el - 5 cos2 e + 17 cos2 el cos2 e -k 4R5 + 2 sin2 8 sin2 8 cos2 5 + 16 sin 8 sin O2 cos 8 cos 8 cos 51 + .. . (23) The favourable relative orientations for charge-dipole charge-quadrupole dipole-dipole dipole-quadrupole and quadrupole-quadrupole inter- actions are illustrated in Fig. 6. The structures of molecular crystals are normally determined by the leading multipole moments. The two types of t (i) -+ I .t. .1 f-3 I (ii) t- c--f J. (i) --f ---f 3 .f - J. + I (ii) < + - -+ (a) (b) (c) (4 (4 Charge-d i pole Charge-quad ru pole Di pole-d i pole Di pole-quad. Quad-quad. 7r e = e = 2 (i) el = 0 e = (i) el = 0 e2 = 9 el = 0 e = 5 ?r (ii) el = z' en = (ii) el = r e = 0 g = x FIG. 6 Favourable configurations for various multipolar interactions. favourable dipolar interaction are found in (i) ferromagnetic and (ii) antiferromagnetic materials (magnetic dipoles interact similarly to their electric counterparts).There is some evidences that dipole-quadrupole forces are important in the solid hydrogen halides. In molecules where the leading multipole is a quadrupole (e.g. H, N, CO,) the structure of the solid may be influenced by quadrupole-quadrupole forces ; thus solid CO,Q and N2Io (in its low-temperature form) have face-centred cubic Cole and Havriliak Discuss. Furahy Soc. 1957 23 31. * Keesom and Kohler Physica 1934 1 167 655. lo Vegard 2. Physik 1929 58 497. I92 QUARTERLY REVIEWS structures in which the orientations of the molecular axes are such that the quadrupole-quadrupole energy is a minimum-the axes are directed towards the body centres so that in CO, each oxygen atom approaches three nearest-neighbour carbon atoms (see Fig.7). The orientations of the molecular axes in solid nitrous oxidell and in the a-form of COI2 (both of which possess small permanent dipole moments) are apparently also determined by quadrupolar forces. FIG. 7 The molecular arrangement in solid carbon dioxide. The height of the potential barrier opposing free internal rotation in molecules has been attributed13 to the electrostatic interaction between dipoles and quadrupoles in the bonds. In ethane the dipole moment of the C-H bond is small (ca. 0.4 D) and its quadrupole moment may be the chief cause of the hindered rotation but owing to the proximity of the interacting C-H bonds there is some doubt14J5 about the convergence of the multipole expansion. The relative orientations of the NH,+ ions in the solid ammonium halides is probably determined by the octopole moments of the tetrahedra but in solids comprised of regular tetrahedral molecules (e.g.CH4 CCl,) the packing of nearest neighbours is presumably the most important orientational effect although the interaction energy of the octopoles will doubtless contribute significantly to the heat of sublimation. Multiple Moments and Symmetry.-A charge distribution of any shape can be ionic the magnitude of the net charge being ne where e = 4.803 x 10-lo e.s.u. is the protonic charge and n is a positive or negative integer. l1 Vegard 2. Physik 1931 71 465. l2 Jansen Michels and Lupton Physica 1954 20 1235. l3 Lassettre and Dean J. Chem. Phys. 1949 17 317. l4 Oosterhoff Discuss. Faraday SOC. 1951 10 79. l5 Wilson Proc.Not. Acad. Sci. 1957 43 816. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 193 Molecules with centres of symmetry cannot have permanent dipole or octopole moments (or any inultipole of order 22n+Z) but many possess quadrupoles ; thus linear molecules (e.g. diatomic ones C02 C2H2 etc.) or planar ones (BF, C6H6) are quadrupolar. Regular tetrahedral molecules like CH4 Ni(CO)4 etc. have octopoles as their leading multipoles and octahedral ones like SF6 possess hexadecapoles (at long-range the field of a hexadecapole varies as P ) . One can easily determine whether a molecule of known structure has a non-zero multipole of a given order say a 2"- pole by evaluating 5 eirinP,(cos ei) for an ionic model of the molecule; if it is not zero then the molecule possesses a 2"-pole.Thus for the symmetrical planar molecule BF, the structure could be written as inset; clearly Ze,r,nP,(cos Oi) is zero for n = 0 I and 1 s i BF3 does not have a charge or dipole moment but L? ei(3zi2 - ri2)/2 = 31e12/2 where the z-axis is at right- angles to the paper and I is the B-F bond-length. Since 6e is of the order of 10-lo e.s.u. and cm. molecular quadrupole moments will normally be of the order of e.s.u. Molecular octopole moments are ca. e.s.u.; / F-6e \ thusCH has a single octopole moment component 52,,, calculated16 to be -1.1 x e.s.u. where the x- y - and z-axes are the sides of a cube at whose centre is the carbon atom and at four of whose corners are hydrogen atoms. Induced Multipole Moments and Energies of Induction.-It is well known that a uniform field F induces a dipole moment of magnitude CCF where a is the polarizability of the molecule.The induced moment will not in general be parallel to F for anisotropic molecules are more easily polarizable in some directions than in others-= is actually a second-order tensor aaB. In strong fields the induced dipole moment proportional to F2 and F3 can be i r n p ~ r t a n t . ~ ~ J ~ The field I; can also induce higher-order moments ; thus polar molecules will have induced quadrupoles propor- tional to F while spherical molecules can only have quadrupoles propor- tional to F2 and higher even orders of F. If it is supposed that the energy u of an uncharged molecule in an external electrostatic field can be written as a power series in Fa Fa, . . . where Fa is the a-component of the field at the centre of mass of the molecule and FaB is the field gradient u = U(') - paFa - QaapFaFp - QpaayFaFpFy - &yasy6FaFBFyFa - LO 3 ai3 F'aS-$Aa:py F,F'sy-~Ba8:yGFaFi3Ftys-~Cai3:~6F'a~Ff~6-.. . (24) Buckingham and Stephen Trans. Faraday SOC. 1957 53 884. l7 Coulson Maccoll and Sutton Trans. Furaday SOC. 1952 48 106. la Buckingham and Pople Proc. Phys. SOC. 1955 A 68 905. 194 QUARTERLY REVIEWS where do) is the energy of the molecule in the absence of the field then it is clear from eqn. (10) that pa is the permanent dipole moment and 0 the permanent quadrupole moment of the molecule; the nature of the other coefficients can be understood by differentiating u with respect to Fa and to FaB. Now where ma is the a-component of the total dipole moment and Tap the ap-component of the total quadrupole moment of the molecule (a/aFu means a differentiation with respect to Fa keeping Flap and higher derivatives of F constant)? whence from eqn.(24) mx p.cL + X a p F p + + PaPyFpFy + i y a ~ y s F s F y 6 + Q A a ByF’sy + BaB:yaFpF‘ys+. . . . (25) Tas = @a + Ay:apFy .t + Bys:a@’yFa + Cap:ysF‘ys + * (26) From eqn. (25) cxaB can be seen to be the usual polarizability tensor; PaBy and yaBys are “hyperpolarizabilities” describing departures from a linear law (they play a part in the Kerr effect1*). The tensors Aa:By and describe the quadrupole moment induced in a molecule by a uniform electric field (alternatively 8 F’, is the dipole moment in- duced by a field gradient). is the “field gradient quadrupole polarizability” discussed by Mayer and Mayer,l9 SternheirnerY2* and Dalgarno and Lewis.21 Mayer and Mayer extended Born and Heisenberg’s2* idea and showed that the “spectral defects” of the alkali-metal atoms (that is the splitting of the energies of the states with different values of the orbital quantum number I ) were partly dependent on Cae:ys.The tensors BUS aaB /Iaay and yaBrs are totally symmetric as can be seen from eqn. (24); A,:, is symmetric in /3 and y and Bag;us and in a,#l and y,8; Cap:yG is also unaffected by interchanging the pairs a/3 and y8. For molecules with elements of symmetry relations between the tensor components e x i ~ t . l ~ * ~ ~ For molecules with centres of inversion pa t This can be proved as follows The Hamiltonian of a molecule in the field is H = Ho - rn&~ - +TupF‘aB where ma and Tas are the molecular dipole and quadrupole moments for a particular configuration of the nuclei and electrons.Thus if 4 is a normalized eigenfunction the eigenvalue is u = (t,b*IHJ #) whence the matrix elements involving first derivatives of 4 and $* with respect to Fa and F’aB must vanish for the derivatives are orthogonal to 4 and to $*. la Mayer and Mayer Phys. Rev. 1933 43 605. 2o Sternheimer Phy.s. Rev. 1957 107 1565. 21 Dalgarno and Lewis Proc. Roy. Soc. 1957 A 240 284. 22 Born and Heisenberg 2. Physik 1924 23 388. 23 Buckingham J. Chem. Phys. 1959,30 1580. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 195 and As:By must vanish (for u must be unchanged on inverting the molecule) and for spherical molecules in a uniform field 1 . . . (27) m = aF + &yF3 + . . . T = &BF2 + . . . a is easily measured through dielectric-constant or refractive-index measurements and the Kerr constant (being a measure of the difference between the polarizabilities of the sphere in the direction of and at right- angles to the field F ) is proportional to y.The quadrupole polarizability B of the ground state of the hydrogen atom has been exactly calculated2* to be 213/2 atomic units (or -0.487 x e.s.u.). The constant B may become measurable through an experiment in which the strong uni- form field of the Kerr effect is replaced by a field gradienP3 (see p. 198). The energy of a hydrogen atom in its ground state interacting with a proton has been evaluated for large R as a power series in 1/R by C o ~ l s o n ~ ~ and is e2[l 9 ( 5 ) 4 +- 1 5 ( a ) 6 0 $- 2 1 3 ( a ) 7 2 + - 7755(a)* 0 + ...I (28) u=-a j+4 2 R 4 R 64 R where a is the Bohr radius (0.5292 A).It is interesting to compare this expansion with (24). The field F at the hydrogen atom is e/R2 and the field gradient FfZz = - 2e/R3 F', = FIuv = e/R3 FraS = 0 if 01 # /3. Thus the term in eqn. (28) in R-4 is - &aF2 = - 1 2cce2/R4 whence Q = 9aO3/2. Also the term in R-g in (28) is - F'aBF'y6 = - &C (F'zz)2 = - Ce2/R6 whence C = 15a,5/2 where C = C,,:, - Czz;, = 2C,,,, (see ref. 23). Similarly the term in R-7 in eqn. (28) is the interaction energy of the quadrupole induced in the hydrogen atom by F with the field gradient thus - 2$a e2 ao6R-7 = - $ BF2 FIZZ = &Be3/R7 so that B = -213a,6/2e. This term-by-term interpretation of (28) is in fact well based quantum-mechanically.24 Little is as yet known of the tensors /3aB and A,,,,; the former probably plays a part in the density-dependence of the Lorentz-Lorenz function of a polar substance26 and possibly in the depolarization of the light scattered by a dense polar fluid,16 as well as in intermolecular forces.The energy of induction of a molecule has been equated to - &xaBFaFp (see for example M a r g e n a ~ ~ ~ and p. 984 of ref. 6) but by comparing this expression with (24) it is clear that it is only a part of the dipole induction energy-hyperpolarizabilities and induced quadrupoles octopoles etc. have been neglected. In the absence of detailed knowledge about the magnitudes of /3 y A B and C it is not easy to determine how good an p4 Buckingham Coulson and Lewis Proc. Phys. SOC. 1956 A 69 639. 16 Coulson Proc. Roy. SOC. Edinburgh 1941 A 61 20.*' Buckingham Trans. Faraday SOC. 1956 52 747. Margenau Rev. Mod. Phys. 1939 11 1. 196 QUARTERLY REVIEWS approximation - $a,@’$, is in a particular case. However for guidance it could reasonably be estimated that for H20 p r= 1-84 x 10-ls e.s.u. ,8 = 1 x e.s.u. a = 1-5 x e.s.u. ( ~ r n . ~ ) y = 1 x e.s.u. (the value for y is indicated by the fact that for CH a = 2.6 x and y = 2.6 x 10-36;1s the /3 is the component of /3usu along the dipolar axis). Thus for a water molecule whose centre is 2 A from an A13+ ion and where the dipole p is directed away from the ion the dipole energy is - pF = -6.6 x 1O-l2 erg (10-l2 erg per molecule is equivalent to 14 kcal. mole-I) while - *aF2 = - 9.7 x 10-l2 - 8 /3F3 = - 8 x lo-” and - -1 F* = 2 -aY - 7 x erg. Thus it is probable that the hyperpolarizabilities 18 and y contribute significantly to heats of hydration of aluminium salts (possibly the terms in A B C etc.are also important). Little is known about /3 and it may be negative in which case the induction energies - &/3F3 and - &yF4 would tend to cancel. It is clear that much more work is needed before we can understand energies of induction when large fields or large field gradients are involved. Multipole moments can be induced in interacting groups of spherical molecules by the intermolecular perturbations. Thus a pair of atoms in S-states will have an induced quadrupole moment proportional to RV when R is large R being the distance between the nuclei; its magnitude is 25eccC 4 R6 approximately 0 = _- - . Each of the atoms will possess a dipole pro- portional to R-’ but unless they are dissimilai these will cancel.In the hexagonal arrangement of close-packed spheres (e.g. solid helium) each of the spheres becomes quadrupolar and the crystal as a whole anisotropic; the magnitude of the induced quadrupoles could possibly be measured by the method discussed on p. 21 2 in connection with macroscopic quadrupoles. Methods of Measuring Quadrupole Moments.- Since nuclear dimensions are ca. e.s.u. (for the deuteron 0 = 0.66 x e.s.u.). Nuclei with spin quantum numbers I > 1 possess quadrupole moments and these give rise to hyperfine structure in the atomic spectrum. Observations of this hyperfine structure28 or of nuclear resonance spectraz9 lead to information about the magnitude of nuclear quadrupole moments. The field gradient F f z at the centre of a hydrogen atom in the 2pz state is 1-08 x 1014 e.s.u.; thus the nuclear quadrupole contribution to the difference between the energies of the hyperfine states of the 2p deuterium atom in which the z-components of the nuclear spin are 0 and 1 is ca.lo4 cycles sec.-l. For atoms with complete inner electron shells it is not possible to deter- mine the nuclear quadrupole moment with precision for the field of the ?* Townes and Schawlow “Microwave Spectroscopy” McGraw-Hill New York 1955 p. 130. 2s Ramsey “Molecular Beams” Ciarendon Press Oxford 1956 p. 213. of atomic ones nuclear quadrupole moments are ca. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 197 quadrupolar nucleus induces a substantial but uncertain moment in the neighbouring electrons.30 Important information about the electronic structure of molecules can be obtained through measurements of nuclear quadrupole resonances; the eigenstates are those corresponding to the various permissible orientations of the nuclear spin relative to the mole- cular axes.Although molecular quadrupole moments are normally ca. lO1O times nuclear moments they are not easy to measure directly owing to our inability to produce sufficiently large macroscopic field gradients (the biggest values of PZz obtainable are ca. 1000 e.s.u.). Interaction energies of about erg can be obtained while for dipoles (p-10-18 e.s.u.) in fields of 10 e.s.u. (3000 volts cm.-l) the energy is ca. 1O-l’ erg. At a point a few Angstrom units from an ion or polar molecule the field gradient is ca. 1014 e.s.u. (its value 2 A from a univalent ion) whereas the field is ca.lo6 e.s.u. Thus molecular quadrupoles will contribute significantly to intermolecular forces and most of the molecular quadrupoles so far “measured” have been determined indirectly by studying the interactions of molecules; however the values deduced are uncertain for they depend upon doubtful assumptions concerning the nature of the intermolecular force-field. The principal methods of this type as well as others are discussed below. A few molecular quadrupole moments have been computed from accurate wave functions. However as 0 is sensitive to the outer reaches of the electron cloud this approach is reliable only in a few cases but conversely if 0 were known then it would provide a severe test of the accuracy of a trial wave function and would thus lead to structural information of importance.( i ) Direct calculation from known wave functions. For a linear molecule where is a summation over the nuclear charges en and -e#*$ is the electronic charge density; the origin is at the centre of gravity of the molecule. # is accurately known for one-electron atoms so that eqn. (29) can be used to compute exact values of 0 for excited hydrogen-like atoms. If the atom’s nuclear charge is Z e n2 [5n2 + 1 - 31 ( I + l)] [l(Z + 1) - 3m2]eao2 . . (30) @n.lrm = - 222721 - 1 ) (22 + 3 ) where n 1 m are the usual quantum numbers. For np states - n2 (n2 - 1) eao2 z2 - - 2 0 n l + l . . . @n,r.a = - Sternheimer Phys. Rev. 1957 105 158. 198 QUARTERLY REVIEWS Hence the well-known dumb-bell shape of a p z orbital causes the mean value of (z2 - x2) to be positive and 0 to be negative.For the 2pz state of hydrogen 0 = - 16.14 x e.s.u. Eqn. (30) shows that one-electron atoms with 1 = 3 and rn = 2 have zero quadrupole moment. Exact electronic wave functions are known31 for one-electron molecules such as H$. With the mass centre as origin OH,+ = 2-1 X e.~.u.,~l while the best approximate LCAO wave function so far obtained leads to OH,+ = 1-7 x e.s.u31 A reliable wave function is known for the ground state of H, and for e.s.u. Less precise calculations one based on a self-consistent field wave function33 leading to 0 = 0.78 x e.s.u. and another on a Wang-type function giving 0 = 0.34 x have also been performed.34 Elaborate self-consistent LCAO wave functions have been determined for N and CO, and these lead to 0~~ = - 2.5 x 35 and to 0c02 = - 8.6 x ~ .s . u . ~ ~ These numbers are probably not accurate 0 being even more difficult to compute than p. The calculated quadrupoles of H20 NH3 and HF3' should also be considered to be only very ap- proximate. (ii) A direct method involving induced optical anisotropy. In the well- known Kerr effect a gas or liquid becomes doubly refracting in the presence of a strong field F. F acts on the molecular dipoles and anisotropic polarizabilities partially orienting them and thereby leading to an induced difference (n - ny) between the refractive indices in the directions parallel and perpendicular to F. Fig. 8 shows a cross-section of a Kerr cell; clearly a reversal of Fcauses no change in (n - ny) so that (n - nY) must be of the second order in F; it has been shown1* that hyperpolarizabilities being neglected 0 = 0.60 x where N is the number of molecules in unit volume and (azz - am%) is the difference between the polarizabilities along and at right angles to the axis of a molecule.If the uniform field of the Kerr cell is replaced by the field gradient F' = - FYY of a four-wire condenser whose cross-section is shown in Fig. 8 then an anisotropy is induced such that23 31 Bates and Poots Proc. Phys. SOC. 1953 A 66 784. James and Coolidge Astrophys. J. 1938 87 438. a3 Coulson Proc. Camb. Phil. SOC. 1938 34 204. 34 Massey and R. A. Buckingham Proc. Roy. Irish Acad. 1938 A 45 31. s6 Hamilton J. Chem. Phys. 1956,25 1283. 37 Duncan and Pople Trans. Faraday Soc. 1953 49 217. .5 Scherr J. Chem. Phys. 1955 23 569. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 199 so that (n - ny) is of the first-order in the applied field gradient; this can be seen by noting that if the voltages applied to the wires are reversed (n - nv) must change sign arid iiiiist therefore be an odd function of Both pF and @F‘, are necessarily small compared with kT for eqns.(32) and (33) to be valid so that although OF‘,,< pFfor reasonable values of F and F x x (see p. 197) OF‘,,/kTand ( P F / ~ T ) ~ are more nearly com- parable. For non-polar anisotropic molecules (e.g. C02 C2Hz C,H,) eqns. (32) and (33) lead when BarS:aB is negligible to F I X * (34) 2OF‘, . . . . L-* = (nx - ny)F ( a z z - axx)@ and this ratio can be about 0.02 under realistic conditions.23 Thus there is an excellent chance that this direct method of measuring 0 will lead to accurate quantitative results; both the sign and the magnitude of 0 are obtainable through eqn.(33). Unfortunately the experiments have not yet been performed although the relevant equipment is now being assembled in Oxford. ;t 3 G T + ) e- + + + + + + + $ + X t Lf 0 0 - - - - - - - -- 04 + (4 FIG. 8 Cross-sections of (a) a Kerr cell producing a uniform field F and (b) the four- wire condenser yielding a field gradient Ffxx = - FfYY at its centre. (nx. - nv) i8 measured by determining the ellipticity induced in a beam of linearly polarized hght travelling in the z-direction and with the electric vector at 45” to the x-axis. (iii) Magnetic-susceptibility measurements. The most accurate measure- ment of @ is that on H2 by Harrick and Ram~ey.~* They utilized the fact that the diamagnetic anisotropy of an axially symmetric molecule is s9 e2 4m.2 x z z - x x x = - F (2i2 - xi2) - XHF .. . (35) where xzz and xzz are the molecular susceptibilities parallel and perpen- dicular to the axis and the summation is over all the electrons (of mass m); XHF the “high-frequency” contribution to xsx is only a small part of xzx in most molecules (in H, it is - 3.7% of xxx and xxs = - 6.9 x erg gauw2). There is no high-frequency contribution to xZZ (as for atoms). Wick40 and Ramsey41 showed that XHF is related to the rotational magnetic 38 Harrick and Ramsey Phys. Rev. 1952 88 228. 39 Van Vleck “Theory of Electric and Magnetic Susceptibilities” Oxford Univ. 40 Wick 2. Physik 1933 85 25. 41 Ramsey Phys. Rev. 1940 58 226; 1952 87 1075. Press 1932 p. 275. 200 QUARTERLY REVIEWS moment and this has been measured (it is of the order of a nuclear magneton) for certain states of 13 and D2 by observing the radio- frequency spectrum of a molecular beam in a strong magnetic field.XHF can sometimes be obtained in other ways,42 and if it is known (or negligible) measurement of xz - xxx leads to 0. Thus where Z,e is the charge on the nth nvcleus at the point (xn,yn,zn). Harrick and R a m s e y ’ ~ ~ ~ observations lead to 013~ = 0.63 -J= 0.05 x e.s.u. for the ground state of orthohydrogen and to OD = 0.63 5 0.07 x Diamagnetic anisotropies are known for a number of rnolecule~,4~ but the difficulties involved in measuring XHF limit the usefulness of this method of determining 0. Anisotropy in the molecular electronic polarizability aa8 can lead to 0 after some doubtful approximations have been madz.Thus 1 (0 IZi I4 (0 IZi Im> h vv I .2 111 J LL YY iJ 111 where-e(O Izi Im) is the dipole matrix element for the ith electron involving the ground and mth excited states and hv is the energy difference between these states; z‘ is a summation over all excited states (it excludes the ground state). If hv is approximated by a constant U (actually the best U for a, need not be the best for a,,) then For non-polar molecules (0 Izi 10) = 0 and for the independent-electron (molecular-orbital) model (OjzizilO) = (01zd210) and this is presum- ably an upper limit since electron correlation will make C (O1zizjlO) negative. With these simplifications eqn. (38) leads to b J * i c ( O l Z i 2 - Xl2]0) . . . * (39) a z z - ass = 7 * 2e2 Now e 2 x = & ( x z z + 2 x x x ) = - G 2 c(0 lri2 10) $XIIF (40) 42 Weltner J.Chem. <$v.F. 2958 28 477. 43 Landolt-RBmstein Zahlenwerte und Functionen” 1 Band 3 Teil Springer 1951 p. 536. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 201 and if XHF is negligible relative to #x (but not necessarily compared with xaz - xxz) then U = - 4mc2x/a and e2a 2mc2x azz - a, = - ~ F ( O [ z i 2 - x,’]O) . . . (41) whence where the anisotropy K = (azz - az,)/3a; K is known in many instances from light-scattering or Kerr-constant data (see ref. 43 p. 509); for Ha K = 0.117 44 and x = - 6-6 x erg gauss-, 45 so that eqn. (42) leads to 0 = 0.64 x e.s.u. in agreement with the above experimental value. Two approximations were made in deriving eqn. (42). The first involves the use of a constant excitation energy U = hv, in eqn.(37) but thisis probably not serious for compensation occurs as the same U is used for 01 in eqn. (41). The second consists in putting $(O [zizj [ 0) = F(0 [zi2 10) and $0 [ x,xj 10) = +‘(O [ xi210) and therefore neglects electron correlation. If C (OIzizj - xixj]O) = - g J (Olzi2 - xi210) then g is a number between 0 and 1 describing the extent of electron correlation and eqn. (42) becomes i J % i If 0 were known g could be determined and knowledge thereby obtained about the extent of electron correlation. For “inner” electrons g must be ca. 1 for these will nearly balance their equivalent nuclear charges leading to a zero “inner-shell” quadrupole moment. g computed through eqn. (44) would be an average for all electrons. For CO, using 0 = - 3 x e.s.u.K = 0.24 x = - 34.5 x erg gausr2 and omitting the K-shell electrons [that is 2 = 6 for each oxygen atom for which Zn = & 1.163 A x = 0; actually mean square values of z and xn should be used but for heavy atoms these are approximately equal to the square of the equilibrium values although for protons significant differences exist. Thus Hirschfelder Curtiss and Bird (ref. 6 p. 1028) incorrectly use the square of the equili- brium nuclear separation in H2 (0.550 instead of the mean square value (0.594 A2) and this leads to an underestimation of 0 of 17 %I we obtain g = 0.896 indicating that the one-electron molecular-orbital approxima- tion (for which g = 0) is limited in its validity in this case. If 0 were zero g would be 0.891. (iv) Indirect methods involving molecular interactions.Since quadrupole moments contribute significantly to intermolecular forces estimates of 0 can sometimes be made through observations of the effects of collisions 44 Ishiguro Arai Mizushima and Kotani Proc. Phys. Suc. 1952 A 65 178. 45 Havens Phys. Rev. 1933 43 992. 2 202 QUARTERLY REVIEWS involving quadrupolar molecules. The measurements may be of several types. Second virial coeficients. Experimental data on the equation of state of gases at low pressures have been widely used to determine the force constants occurring in approximate intermolecular potential energy functions (see ref. 6 chap 3). In the virial equation of state . . . . (45) the second virial coefficient B(T) (a function of T but not of the molar volume V ) represents an initial deviation from the ideal-gas law and hence is related to the interaction of molecules in pairs.If u12 is the inter- action potential energy of two axially-symmetric molecules 1 and 2 in the configuration of Fig. 5 the classical statistical-mechanical expression for B(T) isg6 B(T) =- Nr 4 R2 dR 1 sin 8 d8 1 sin 8 do2 j d C [ I - 0 0 0 0 exp (- ul,/kT)] . . . . . (46) If u12 is of the central-force type that is if u12 = u1J0)(R) a function of R only then eqn. (46) can be evaluated analytically for a number of simple forms of ul2(O)(R). For the hard-sphere model (see Fig. 9) u12(0)(R) = UHS = 00 for R < R, and = 0 for R> R . . . (47) B(T) = $ nNR2 = 4 x actual volume of the molecules . . (48) For the Lennard-Jones 6-12 potential (see Fig. 9) where E and Ro are parameters having the dimensions of energy and B(T) = 8 d R O 3 F(y) = f ~ T N R ~ ~ ~ - ~ [ I Y ~ ( ~ ) - &H6(y)] .where y = 2(~/kT)* and F(y) is a function that has been tabulated (see ref. 6 p. 1114) and is expressible in terms of the functions H,(y) in- troduced by P ~ p l e ~ ~ defined by (50) a3 H&) = 12y"R,n-3 I R-" exp [ - y2{ (2) l2 - ( R2 dR 0 Fowler and Guggenheim "Statistical Thermodynamics" Cambridge Univ. Press 1953 chap. 7. 47 (Jxnnard-) Jones Proc. Roy. SOC. 1924 A 106 4 4 1 . 48 Buckingham and Pople Trans. Furuduy Soc. 1955 51 1173. 4B Pople Proc. Roy. SOC. 1954 A 221 508. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 203 These functions have been relations) and for 0.6 < y < 3.2. for all n (since there are recurrence Observed values of B(T) lead to the “best” force constants E and R FIG.9 The hard-sphere and Lennard-Jones intermolecular potential energy functions. by a method due to R. A. Bu~kingharn,~~ and Lennard-Jones potentials are now known for many non-polar molecules (see ref. 6 p. 1 1 10); the potentials obtained through measurements of B(T) are usually in reason- able agreement with those derived from other observables such as viscosity. Keesom51 evaluated eqn. (46) for potentials of the type ‘12 == ‘HS + ‘dipole-dipole } . . . . (52) ‘12 = ‘HS + ‘9uadrupole-quadrupole where Udipole-dipole and Uquadrupole-quadrupole are the energies proportional to p2/RS and to 02/R5 in (23). His result for the quadrupolar case is 2 [ 3 ( o2 )2 18 ( 02 )3 3 5 Ro5kT 245 RO5kT B(T) = - 77NR03 1 - - -~ + - -_ - 4165 639 ( @%-T O2 )”+ . . *] . . . * . . . . (53) Keesom erroneously supposed that the forces between inert-gas atoms were of the types (52); the quantum mechanical nature of dispersion forces was proposed some years later by London.52 Second-virial data for Cog have been obtained over the wide temperature 6o R.A. Buckingham Proc. Roy. SOC. 1938 A 168,264. 61 Keesom Physikal. Z. 1921,22 129. 61 London Trans Faraday Soc. 1937,33 8 . 204 QUARTERLY REVIEWS range 0-600”~ by MacCormack and S~hneider,~~ who showed5* that single values of E and R in uw cannot explain their observations over the full range of temperature. They proposed “high” and “low” temperature force-fields but P o ~ l e ~ ~ found that a potential of the form ‘12 ‘LJ+ ‘~undrupole-~iundrupole . * ’ . (54) (55) leading to the approximate formula (higher powers of O2 being neglected) 2 7 3 B(T) = - 7r NR,3[F(y) - To A2 H,,(y)] .. . where X is the dimensionless parameter O2/eRO5 adequately explained the data over the full range of T. In (54) there are three adjustable parameters namely E R, and 0 and Pople found that 0c02 = & 5.73 x e.s.u. gave the best agreement [(54) is independent of the sign of 0 but the negative sign is probably the appropriate one]. Second-virial and crystal data. The molar heat of sublimation H of a crystal at O’K and the equilibrium molecular separations are potential sources of information about intermolecular forces. H is equal to -E minus the zero-point energy where E is the molar intermolecular poten- tial energy when the molecules are in their equilibrium configurations. The zero-point energy can be obtained through low-temperature specific-heat measurements and is 9 N k 8 ~ / 8 where 8D is the Debye temperature of the crystal.For a face-centred cubic crystal in which all molecules interact with all others according to the Lennard-Jones potential (49),65 E = - H - 9 N k 8 ~ / 8 = 2Ne [I21318 (:)12 - 14.4539 (%)”I (56) and for the equilibrium separation Re = 0. These equations adequately describe the properties of the solid inert gases but for carbon dioxide the calculated H is only 3480 cal.mole-l for the “high” and 3210 cal. mole-l for the “low” temperature potential of MacCormack and S~hneider;~~ the experimental value is 6440 cal. m ~ l e - ~ . * ~ Pople’s potential (54) leads to 49 E = ~ N E [ 12.1318 ( i 2 ) 1 2 - 14.4539 (g!?)6 - 5.3533 h ($71 (57) and his parameters ~ / k = 160.0”~ R = 3.80 A 0 = e.s.u.to H = 7330 cal. mole-l which although high is closer to the experimental result than those obtained from a pure Lennard-Jones potential. There is therefore little doubt that a substantial part of the binding energy of solid carbon dioxide is due to quadrupole- quadrupole 5-73 x 65 MacCormack and Schneider J. Chem. Phys. 1950,18 1269. 6p MacCormack and Schneider J. Chem. Phys. 1951,19 849. 55 Lennard-Jones and Ingham Proc. Roy. SOC. 1925 A 107 636. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 205 interactions. Pople's overestimation of H may be due to neglect of induction effects. The principal contributor to the induction energy for large R corresponds to the interaction between the permanent quadrupoles and the dipoles induced in the other molecules; it takes the form 5 .. . . . . . . . 3- cos4 e,] (58) where & = +aBB is the polarizability (assumed isotropic) of one of the molecules. In the crystal the symmetry is such that there is no electric field acting on any molecular centre and Uinduction does not contribute to E but in a gaseous collision all configurations are possible and Uinduction will tend to diminish B(T). If Uinduction is added to the potential (54) so that then E is given by (57) and 2 7 3 B(T) = - d R O 3 [F(y) - - h2 Hlo(y) - 16 3 320 The parameters E R, and 0 in (59) can be determined from the following three observations E = - 6530 cal. mole-l Re = 3.917 a,9 B (60"~) - 96.9 ~111.~ m01e-l.~~ 0 = 4.60 x e.s.u. About 54 % of E arises from the quadrupole- quadrupole interactions. This potential leads to B (0"c) = - 146.0 B (200"~) = - 41.8 B (600"~) = 13.0 ern? mole-l and these compare favourably with the observed values53 of - 156.4 - 34.1 and 12.1 cm.8 mole-l.The apparent failure56 of the well-known combination rules c12 = (el c2)* and ROl2 = 8 (Rol + Ro2) for the Lennard-Jones parameters corresponding to the interaction of dissimilar pairs of molecules is at least partly due to the neglect of quadrupolar effects. In a collision between a quadrupolar molecule I and a spherical molecule 2 '12 = 'u 'quadru~ole-quadrupole+ 'induction ' (59) hy- H&)] . (60) - - They lead to ~ / k = 195.7 OK Ro = 3.768 A The above combination rules together with the parameters for argon 56 Cottrell Hamilton and Taubinger Trans. Faraday SOC. 1956,!52 1310, 206 QUARTERLY REVIEWS Calculated ( ~ m .~ mole-l) - 40.1 - 31.3 - 24.2 (see ref. 6 p. 1110) e2/k = 119*8"~ R, = 3.405 A a2 = 1.63 X ~ m . ~ and the above constants for carbon dioxide lead to '' Experimental "b6 ( ~ r n . ~ mole-l) - 32.3 - 26.0 - 19.2 and this comparison is much more favourable than that obtained on the basis of the pure Lennard-Jones potential.66 dispersion force potential energy is an important contributor to B(T). However two opposing effects are involved in both B(T) and E namely a coupling between the anisotropic dispersion force and the quadrupole- quadrupofe interaction [yielding a positive term proportional to HI&) in B(T)] and the pure anisotropic dispersion effect [leading to a negative term in H12(y)]. Thus the compfications caused by using this more accurate form for the dispersion force are probably not justified.The compressibility of a single crystal being proportional to It has been suggested5' that the orientation-dependence of the ("") dRa R=R is also a potential source of information about inter- molecular forces ; presumably the elastic constants relating to shear stresses and strains are closely related to the orientational intermolecular forces- one would expect significant differences between solid argon and solid carbon dioxide ; unfortunately measurements of these kinds do not seem to have been made. The dielectric constant of an imperfect quadrupolar gas. The total dielectric polarization .P of a perfect gas of non-dipolar molecules is independent of T. At higher densities interacting pairs of molecules (which may possess induced dipoles) become significant and =P will be temperature-dependent.If the orientation polarization .P is expanded as a power series in 1/V similarly to (49 where E is the static dielectric constant and d ( T ) a ( T ) %(T) . . . are dielectric virial coeficients then d ( T ) = 4 ~ N u ~ / 9 k T where p is the per- manent dipole moment of the molecule and B(T) is proportional to the mean square dipole moment of an interacting pair of 57 Castle Jansen and Dawson J. Chem. Phys. 1956,24 1078. ij8 Buckingham and Pople Trans. Faraduy Soc. 1955,51 1029. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 207 where m12 is the scalar corresponding to the vector m, = ml -+ m2 the dipole moment of the interacting pair of molecules 1 and 2 and the angular brackets denote an average over all configurations. Thus for axially symmetric molecules with relative co-ordinates as in Fig.5 03 x x an n N2 9kT g ( T ) = -I R2 dR sin 8 d8 1 sin 8 do 1 d5[~2,,~ - 2p2] 0 0 0 0 exp ( - ul,/kT) . . + . . . (65) For a pair of molecules quadrupole p = 0 and 7T%l can be ap- proximated by the product of the polarizability of molecule 1 and the electric field Fl (scalar value F,) at its centre due to the quadrupole moment of molecule 2. Thus from eqn. (16) and if aaP is isotropic (aafl = d,,) 9202 mi$ = 4 ~ 8 [4 - 8 COS 8 - 8 cos2 8 + 5 COS* + 5 C O S ~ 8 + 18 C O S ~ 8 COS 8 - 8 sin 8 sin 8 cos el cos 8 cos 51 . (66) If uI2 is of the form of (54),59 If K = ( M ~ ~ - a,,)/3a is not zero thenaoVa1 <rnI2> = < r r ~ ~ > = 3a2 (1 -k 2 K,) o2 <R-8> 18 <m,-m2> = - 5 a2 K~ O2 <R-8> whence and <ml:> = 6 a* (1 + K ~ ) @ <R-s> Buckingham Trans.Furaduy SOC. 1956 52 747. [Eqn. (3.11) of this paper 6o Buckingham and Pople J. Chem. Phys. 1957 27 820. Jansen personal communication 1958 pointing out that < ml*rnz> does not erroneously omits a term in he arising from the cross-term < ml-np > .] vanish if K % 0. 208 QUARTERLY REVIEWS For carbon dioxide E/k = 195.7 OK R = 3-768 A 0 = & 4.60 x e.s.u. 01 = 2.97 x lo-,* ~ r n . ~ K = 0-24 so that 9(100°c) should be 31-9 ern? mole-,; the experimental value deduceds9 from the E and n data of Michels et aLs2 is 13.6 cm.6 mole-,. In compressed carbon dioxide microwave absorption proportional to p2 where p is the gas pressure has been observed and interpreteds3 in terms of transient dipoles induced by the permanent quadrupoles of neighbouring molecules.If the full frequency range of the absorption can be covered then E - n2 can be directly measured and [if d ( T ) = 01 from (63) . . . . (69) where R = Nk is the universal gas constant. Induced vibration and rotation spectra. The general theory of pressure- induced infrared absorption spectra has been developed by van Kranen- donkmsq The induced fundamental vibration-rotation band of H in the pure gas and in mixtures with inert gases has been inter~reted~~ in terms of transient dipoles induced both by short-range overlap forces and by quadrupolar fields. At low pressures only binary interactions are im- portant and by introducing reasonable assumptions van Kranendonkea deduced from Chisholm and Welsh’sa6 spectra for H,-N mixtures that ON = & 1.64 x In the case of pressure-induced rotational transitions in H2,67 the observed absorption proportional to p2 can be explained by assuming that the transition dipoles are due to quadrupolar fields; OH = 0.6 x e.s.u.leads to good agreement with experiment. In hydrogen-argon mixtures the induced rotational absorption ( AJ = 2) is proportional to the pressure of the inert gas and to the square of its polarizability as well as to (OH2)2. Spectral lines have a finite width dv (the width in cycles sec.-l at half the maximum intensity) for three reasons namely because of the natural line width (related to the uncertainty principle) the Doppler effect (connected with the thermal velocities of the molecules) and pressure broadening (see ref. 28 p. 336 and ref. 6 p. 1020).Only the last is normally significant 62 Michels and Kleerekoper Physica 1939 6 586; Michels and Hamers Physica 1937 4 995. 6a Birnbaum and Maryott Phys. Rev. 1954 95 622; Maryqyt Wacker and Birn- baum, N.B.S. Report (Project 0536) June 1957; Buckingham PropiktCs optiques et acoustiques des fluides comprimCs et actions intermolkculaires” C.R.N.S. Bellevue France July 1957 p. 57. e.s.u. Pressure broadening in the microwave spectral regions. 64 van Kranendonk Physica 1957,23,825. 65 van Kranendonk Physica 1958,24 347. O6 Chisholm and Welsh Canad. J. Phys. 1954 32 291. Colpa and Ketelaar Molecular Physics 1958 1 343. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 209 in the microwave region of the spectrum and measurements of d v lead to “collision cross-sections” (3 by means of the formula a = 2n AvjNv .. . . . . . . . (70) where N is the number of molecules in unit volume and v the mean relative velocity of colliding molecules. Andersons and Leslies have derived quantum-mechanical formulae relating a to the intermolecular forces. The most-studied spectral line is the inversion of the J = 3 K = 3 state of N14H at 23,870 Mc./sec. (0.796 cm.-l) and the self-broadening of this line has been satisfactorily interpreted in terms of dipole-dipole interactions (see ref. 68 and ref. 28 p. 361). Smith and Howard70 were the first to suggest determining molecular quadrupole moments by measuring (T for the broadening of microwave resonance lines (notably the ammonia inversion spectrum) by foreign non-dipolar gases. An approximate theory has been developed by Smith;71 if the observed CT is considerably larger than the kinetic-theory collision diameter it is probable that the dipole-quadrupole interaction is dominant in determining (T for collisions between an ammonia molecule and one whose leading multipole is a quadrupole for its range is greatest but other effects such as the quadrupole-quadrupole interaction (F) are probably also significant.For this reason the quadrupole moments deduced from pressure-broadening data are not beyond suspicion. Recent measurement~~~ of the broadening effect of foreign gases on the J = 3 K = 3 inversion line of ammonia have led to @CO = & 1.7 x and ON^ = & 0.80 x The broadening effect of argon and other spherical molecules on this NH3 inversion line has been inter~reted~~ as arising from the interaction proportional to between the permanent NH quadrupole and the dipole induced in the sphere by the field of the permanent NH dipole.Measurements of this effect are consistent with an assumed quadrupole moment ONH3 = 0.66 x e.s.u. but this figure should be regarded as only approximate since all shorter-range interactions (such as exchange and overlap effects and permanent moment-induced quadrupole terms proportional to Cp2P8) have been neglected. Direrences in the heats of solvation of oppositely charged spherical ions of the same size. Although the heats of solvation of salts can easily be meas- ured the contributions due to the positive and negative ions have not been separated experimentally. However different 75 are agreed that e.s.u. Anderson Phys. Rev. 1949 76 647. Leslie Phil. Mag.€951 42 37. 70 Smith and Howard Phys. Rev. 1950 79 132. ‘l Smith J. Chern. Phys. 1956,25 510. 71 Feeny Madigosky and Winters J. Chem. Phys. 1957 27 898. 73 Anderson Phys. Rev. 1950,80 511. 74 Latimer Pitzer and Slansky J . Chem. Phys. 1939 7 108. 75 Verwey Rec. Trav. chim. 1942 61 127. 210 QUARTERLY REVIEWS although K+ and F- ions are similar in size their heats of hydration A H are different and are about - 75 and - 122 kcal. mole-l respectively at 25”c and at infinite dilution. If it is assumed that the first shell of water molecules is tetrahedrally disposed with dipoles pointing away from or directly towards the ion’s centre and that more distant water molecules interact with the ion as if they formed a continuum having the bulk properties of water then the difference between K+ and F- hydration energies can be explained in terms of the interactions between the ion and the quadrupoles of the water molecules.The charge-dipole energy ( R-2) will presumably dominate and since the field of the ion at a water molecule’s centre will change sign on reversing the ionic charge the molecule will invert in order to preserve the negative sign of the ion-dipole energy (thus in K+ the “complex” is of the type K+ dH,while in F- it is F- \O). However O, is un- affected by an inversion so that the ion-quadrupole energy (R-3) will be equal but opposite in the two cases (it is assumed that the centre of mass of the water molecule is equidistant from the centres of the two ions). There is another small contributor to the hydration energy difference and this arises from the dipole-quadrupole interactions among the “adsorbed” water molecules ; the other significant contributions to A H should be identical for the two ions (the ion-octupole interaction will be the same in the two cases).The final result is 76 H ‘H H’ where 2 is the modulus of the charge on the ion (2 = 1 for K+ and F-) R is the distance from the ion’s centre to the centres of gravity of the four nearest solvent molecules (R = 2-73 A for K+ and F- in water) p is the permanent dipole moment of a solvent molecule and O, its quadrupole in the direction of the axis of p. If ( AH+ - AH-) for the K+ F- pair in water is 47 kcal. mole-l then (@,,)H,o = 2.0 x e.s.u. Similarly Coulter’s figures7’ for the heats of ammonation of the same pair lead to O N H ~ = 1.3 x e.s.u. It is not easy to assess the accuracy of these 0s; probably the best attitude to adopt is to look upon the model as providing an explanation for any difference that exists between the solvation energies of oppositely charged spherical ions of the same size and if reliable 0 s should become available then they could be used with the model to determine absolute solvation energies from the accurately known relative values.Eflects of interactions on nuclear quadrupole resonance frequencies. In the gas phase the energy differences between the states corresponding to the 76 Buckinghain Discuss. Furaduy Suc. 1957 24 151. 77 Coulter J. Phys. Chem. 1953 57 553. BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 21 1 allowed orientations of a quadrupolar nucleus in the field gradient of the other charges of the molecule can be determined through observations of the hyperfine structure of the pure rotational spectrum (see ref.28 p. 149). In solids pure quadrupole resonance spectra can be obtained and the hyperfine energy levels determined through observations of the absorption of radio-waves in the s ~ l i d . ~ * * ~ ~ The coupling constant obtainable from both sources is eQq where eQ/2 = @ is the nuclear quadrupole moment (actually 0 is the moment for the state M = I where I is the nuclear spin and A4 its component in a fixed direction so that Q = p(3zn2- rn2)dv where z is measured along this fixed direction and p dv is the nuclear charge in the volume element dv) and - q = FZz is the electric-field gradient along the molecular axis. Differences between qgas and qeolid have been observed; 8 s 8 0 in solids comprised of polar molecules neighbouring dipoles may contribute significantly to qsolid,*’ but in other cases (e.g.solid Dz and N,) it should be possible to use the known crystal structure to relate (qgas - qsolid) to the molecular quadrwpole moment @. In a face-centred cubic lattice of the N and C 0 2 type (see Fig. 7) the quadru- pole moments of neighbouring molecules produce a field gradient FZ2 = 42-83 @/Re5 at each molecular centre [it is F’zz that yields the quadrupolar contribution to E in (57)] and since F”,, = 0 at this point the field gradi- ent acting on nuclei that are slightly removed from the centre (such as the N nuclei in solid N,) will be approximately equal to F’,, so that (qgas -qsolid) = 42.83 @/Re5. Since @/Re5 is ca. lof1 e.s.u. the quadrupole coupling constants eQq may differ by up to a few kilocycles sec.-l in gases and solids of this type.If a quadrupolar molecule containing a nucleus with I > 1 is frozen into a matrix of spherical molecules such as CCl, then there will be a “reaction field gradient” FZz at the quadrupolar nucleus arising from the polarization induced in neighbouring molecules by the field of the mole- cular quadrupole @. This effect can be evaluated approximately by supposing that FZz is the field gradient existing at the centre of a spherical cavity of molecular size in a continuum of dielectric constant E when the cavity contains a quadrupole at its centre. Arguments similar to those used by OnsagerS1 lead to ‘J where a is the radius of the cavity. Hence F’,2 is ca. lof2 e.s.u. so that changes in coupling constants of the order of tens of kilocycles sec.-l could be expected in these cases; if a change of this kind were observed Dehmelt Discuss.Faraday SOC. 1955 19,263. 7s OrvilIeThomas Qmrf. Reviews 1957,11,162. ** Duchesne paper presented at the symposium on “Applications of Electron and Nuclear Resonance in Chemistry” Bristol 1958; Chem. SOC. Special Publ. No. 12 p. 235. Onsager J. Arner. Chem. Soc. 1936 58 1486. 212 QUARTERLY REVIEWS eqn. (72) would lead to an approximate value of 0 if Onsager's formula 4nNa3/3 = Mfd where Mis the molecular weight and d the density of the pure quadrupolar liquid or solid were used for determining a. (v) Macroscopic quadrupoles. Just as electric and magnetic dipoles can be aligned in ferroelectric and ferromagnetic solids so too can molecular quadrupoles be oriented so that a crystal possesses a large permanent quadrupole moment.Calcite potassium nitrate hexamethylbenzene polyacetylenes etc. have their molecular quadrupoles parallel and a piece of any of these crystals will behave as a macroscopic quadrupole. If the molecules are uncharged and non-dipolar (or alternatively if the unit cell has a centre of inversion) then the total quadrupole moment is . . . . . . . . . 0(0),p = c O(Jl,p (73) where the summation extends over all molecules in the crystal. For axially symmetric crystals comprised of axially symmetric molecuIes eqn. (73) becomes O(0) = $0 ?(3cos2Bj - 1) . . . . . . . (74) where Bj is the angle between thejth molecular axis and the axis of the sample. For parallel quadrupoles 6 = 0 and @(O) = NO.A needle-shaped macroscopic quadrupole will experience a torque in a field gradient which could be produced either by another macroscopic quadrupole or by a condenser of a suitable shape. If the needle's moment of inertia about an axis is I its period for small oscillations in the xy plane about its equilibrium position in a field gradient F'xx = - FZz = 0 (such a field gradient would exist on the line midway between two long parallel wires at a positive potential relative to a large cylinder surrounding them) is r = J(gX) . . . . . . . . (75) Calcite has a density of 2-71 1 g. ~ r n . - ~ at 25" and the unit cell (see Fig. 10) contains two CaCO units. The quadrupole moment of a unit cell is approximately 0 = [- 146 x + 2 0 ~ 0 ~ 2 - 1 and for an ionic structure -0-C+ (which is isoelectronic with BF,) 0 ~ 0 ~ 2 - = 12 X e.s.u.as the C-0 bond length is 1-31 I$; thus the total quadrupole moment of a unit cell is probably about - 120 x e.s.u. and the moment of a crystal weighing 1 g. is ca. 3-6 x lW3 e.s.u. For I - 1 g. cm.2 and FZx - 100 e.s.u. eqn. (75) leads to a period T - 7 seconds. This should be a reasonable period to measure so that O(O) and hence 0 (or in the case of calcite OCO,~-) might be measurable by this means. /O- \O- BUCKINGHAM MOLECULAR QUADRUPOLE MOMENTS 213 FIG. 10 The structure and dimensions of the unit rhombohedra1 cell of calcite. Quadrupole Radiation.-Allowed transitions take place between eigen- states $(l) and if a non-zero electric dipole moment matrix element (+(l) lp exists and the transition probability is proportional to the square of this transition moment.The symmetry of #(l) and #t2) is often such that the matrix element vanishes and the transition is then said to be forbidden. However spectral lines corresponding to forbidden transitions are sometimes observed (usually they are weak) and they may arise from (a) perturbation of the wave functions by external influences such as electric fieldsa2 or collisions with other molecules (b) a non-zero magnetic dipole transition moment (c) a non-zero quadrupole transition moment (+(l)l@l@c”>>. Effects (b) and (c) normally lead to intensities only - and - respectively of that of an allowed transition but they have been observed. 83 For the electronic ground state of H, all dipole vibrational transition moments vanish but 0 will vary with the internuclear distance R (James and Coolidges2 calculated that = 0.8 x 10-l8 e.s.u.) and vibration-rotation bands have been observed by Her~berg,~* using a multipath infrared spectrometer of total path length 5.5 kilometres at 10 atm.The selection rule for vibration-rotation transitions in homo- nuclear diatomic molecules is AV = & 1 AJ = 0 & 2 (except that J = 0 + J = 0 is f ~ r b i d d e n ) . ~ ~ It was suggested86 that these H spectra were Herzberg “Spectra of Diatomic Molecules” D. van Nostrand New York 1950 82 Condon Phys. Rev. 1932,41 759. 84 Herzberg Nature 1949 163 170. *6 Crawford Welsh and Locke Phys. Rev. 1949,75 1607. p. 277. 214 QUARTERLY REVIEWS in fact due to intermolecular collisions (and hence proportional to the square rather than to the first power of the H pressure) but owing to the sharpness of the lines this suggestion was later shown to be in- correct.86 Bond Quadrupole Moments.-When values of molecular quadrupole moments become known there will be a need for the development of the relationships between the moments and molecular structural features. Empirical bond dipoles have proved very useful and so too have bond polarizabilitie~.~~ Similarly “bond quadrupole moments” should also be of considerable importance. If the bonds are axially symmetric (most are approximately so) the molecular quadrupole has components xz ,- f c Oj ( 3 cos2 &> - 1) Oj (3 cos2 O j - 1) O YY = & X Oj ( 3 C O S ~ O, - 1 ) [Or -1- O, 3- O, = 01 . . (76) o, = 4 where Oj is the quadrupole moment of thejth bond and Oia is the angle between its axis and the a-axis of the molecule.Thus bond quadrupoles are simpler to handle than bond polarizabilities for the latter have com- ponents along and perpendicular to the bond; if the perpendicular components were always zero the rules for deriving molecular quadrupoles from bond parameters would be identical to those used for obtaining molecular polarizabilities from the appropriate bond constants.87 A Table of Molecular Quadrupole Moments.-The Table lists the most probable values (deduced from the meagre present data) of the molecular quadrupole moments of simple molecules. The last seven entries are based on microwave pressure-broadening data (incorporating the correc- tion recently deduced by Smith7’) but as explained on p. 209 these may sometimes be unreliable. Numerical values of quadrupole moments Substance 1 10260, (e.s.u.) I Substance I 10280, (e.s.u.) 1 0.66 x - 16-14 - 48.4 0.63 0.63 & 1.5 - 3 2 1 >-0*3 <0.3 z t 1 I t 1 - 2 - 2.5 3 Z t 1 *a Welsh Crawford and Locke Phys. Rev. 1949,76 580. Le Fhre and Le F&vre Rev. Pure Appl. Chem. 1955,5,261.

ISSN:0009-2681    

DOI:10.1039/QR9591300183

出版商:RSC    

年代:1959

数据来源: RSC

摘要:

THE MASS SPECTROMETRY OF FREE RADICALS By J. CUTHBERT M.A. D.PHIL. (INORGANIC CHEMISTRY LABORATORY OXFORD) A STUDY of the free radicals which are intermediates in many gas-phase reactions is of great importance in understanding the mechanisms of these reactions. There are at present only a few experimental methods by which the chemical nature of free radicals can be studied; of these methods optical spectroscopy has been very successful with certain radicals,l while microwave spectroscopy and paramagnetic resonance have as yet found little application to gaseous radicals. Mass spectrometry has been of increasing importance during the last two decades since it affords a simple method of identifying a radical by its molecular weight. The experimental difficulties of transferring a sample from relatively high pressure reaction systems to the very low pressure mass spectrometer without losing any radicals through collisions in the gas phase or on surfaces and of prevent- ing the formation of radicals by surface reactions in the mass spectrometer itself have been a severe limitation on the method.The mass spectrometer (a) Instrumental.-Mass spectrometry began in J. J. Thompson’s experiments with “anode rays” or “positive rays” and in F. W. Aston’s separation of gaseous ions of different masses (or rather mass/charge ratios) by focusing beams of these ions with electrostatic and magnetic fields. The modern in~trument,~,~ however shows more clearly its descend- ance from that of Nier,4 and it is this type which has been used to study gaseous radical^.^,^ Fig. 1 shows schematically a Nier-type mass spectrometer.An electron beam ( E ) produced by a heated filament (F) ionizes gas molecules in a metal box (B) which is at a high positive potential (usually 2 kv) with respect to earth. The positive ions ( J ) are drawn through a slit in the box by a potential field formed by a stack of plates the last of which is at earth potential; some of these plates are used as an electrostatic lens to focus the ion beam on a small slit in the last plate (D). Having left the source the ions travel down a metal tube (T) through a magnetic field ( M ) which causes them to move in a circular path; the magnetic field is uniform and extends over a 60” or 90” sector and acts to focus an ion beam of uniform Norrish and Thrush Quart. Rev. 1956,10 149. Dunning Quart. Rev.1955,9 23. Duckworth “Mass Spectroscopy” Cambridge University Press Cambridge 1958. Nier Rev. Sci. Instr. 1947 24 1 19. Lossing Ann. New York Acad. Sci. 1957 67 499. Beckey Angew. Chew. 1958,70 327. 215 216 QUARTERLY REVIEWS -<- - I - ,+-J F-&==- J+FG - It- .7 I - - - FIG. 1 Nier-type mass spectrometer. The N pole of the magnet ( M ) is above the paper and the S pole below. The inset shows the electron gun at right angles to the plane of the main diagram with the filament ( F ) and the electron collector (G). energy on a collecting slit (C). Ions of different mass/charge ratios require different magnetic fields to focus them on C. Behind the collecting slit a metal cup collects the ions and the current so formed is amplified by an electrometer. In order to prevent ion-ion and ion-molecule collisions and also space-charge effects in the ion beam the pressure in the instrument is kept as low as possible (usually mm.) The condition for focusing is that C 0 and D are collinear with OC = OD (Fig.1). This instrument gives constant deviation direction focusing with the mass of a singly charged ion in focus related to the magnetic field ( H ) and the ion accelerating potential ( V ) by mV cc H 2 to A mass spectrum can be scanned with either V or H constant the other being varied; in practice magnetic scanning is more usual. (b) Uses.-The probability of occurrence of the process M + e- -f Mn+ + (n + 1)e- is proportional to the electron beam intensity the pressure of M and the ionization cross section of M; the last is zero for an electron beam whose energy is less than the nth ionization potential (In) of M and it increases with the excess of electron energy for a beam of greater energy than I,.The probability of forming M2+ is always much less than the probability of forming M+ and only singly charged ions need be considered here. The use of a mass spectrometer for qualitative and quantitative analysis is based on the fact that other parameters being kept constant the presence CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 217 of an ion beam M+ is diagnostic for the presence of My and the relative intensities of different ion beams from a mixture are proportional to the products of the ionization cross sections and the pressures of the parent molecules. One major complication must be considered here; this is that a polyatomic molecule gives not only parent ions formed by abstraction of an electron but also fragment ions formed by processes of the type where A and B may be polyatomic.Thus ethane gives ions C2H6+ C2H5+ C,H,+ CH,+ etc. The energy required for the process is the dissociation energy of the C2H5-H bond plus the ionization potential of the C2H radical i.e. D(C2H5-H) + 1(C2H5) which is greater than the energy l(c2H6) for the process so that at low electron energies (ca. 1 5 ~ ) the parent peak (ii) predominates but the spectrum is very dependent on electron energy. As the electron energy is increased the spectrum changes process (ij becoming more important and at high energy (50-70 v) the parent peak is quite small and the relative heights of the peaks will be much less dependent on the energy of the electrons.Quantitative analytical applications require that the spectrum or “cracking pattern” obtained from a pure compound at the appropriate electron energy be known accurately; analysis of a mixture is then a question of solving several simultaneous linear equations. If the cracking pattern of some of the components is not known or cannot be determined a complete quantitative analysis is not possible. The electron energy at which an ion current first appears is called the appearance potential (A) for the ion in question; it is equal to the ioniza- tion potential (for a parent ion) or the fragment ionization potential plus the energy of the bond broken in fragmentation plus any kinetic energy of fragments; a plot of ion current against electron energy should therefore AB + e- -+ A+ + B + 2e- .. . . . . . . . C2H6 + e- -+ C2H5+ + H + 2e- (9 C,H + e- -+ C,H,+ + 2e- . . . . . . . . . . . (ii) €/ecfron energy FIG. 2 Ionisation eficiency curves 1 monoenergetic electrons; 11 electrons with Rnltzmann enerpv distribution. 218 QUARTERLY REVIEWS enable one to determine these quantities. For constant energy electrons the curve is shown at I in Fig. 2; but the electrons are obtained from a heated filament and therefore they have a Boltzmann distribution of energy centred on the measured electron energy. Electrons at the high- energy end of the Boltzmann distribution will cause ionization when the measured electron energy is below the appearance potential and the curve obtained is therefore like I1 in Fig. 2. Various empirical methods have been used for obtaining the appearance potential from such curves.These include extrapolation of the more or less linear upper part of the curve to the energy axis,’ estimation of the “initial upward break” of the curve from the energy axis,8 application of a tangent of “critical slope’’ to the plot of log (ion current) against electron energy,g and extrapolation of the energy differences between the curve in question and that for a standard gas to zero ion current.1° The obvious differences between these methods are largely removed in practice by meas iring appearance potentials relative to that of a standard gas (e.g. argon ) whose spectroscopic ioniza- tion potential is accurately known; this is necessary also because irre- producible effects due to contact potentials in the electron gun and field penetration in the ionization chamber make the measured electron energy different from the true one.Three experimental methods exist for minimising the effect of the Boltzmann distribution of energy in the electron beamll or for minimising the distribution itself;12 these methods suffer from the disadvantage that the overall sensitivity of the instrument is much reduced and it has not yet been possible to apply them to the study of free radicals. Application to free radicals (a) Principles of the Method.-The appearance potential of the ion CH3+ from CHI is equal to the dissociation energy of the CH3-H bond plus the ionization potential of the CH3 radical formed plus the kinetic energy of the fragments i.e. An empirical rule by Stevenson13 states that the kinetic energy is zero if the ionization potential of the non-ionized fragment is greater than that of the ionized fragment which is usually the case I(H) > I(CH3).The appearance potential of the ion CH,+ from CH is simply the ionization potential of the radical A(CH4 -+ CH3+) = D(CH,-H) + I(CH,) + kinetic energy A(CH3 -+ CH,+) = I(CH,) Mariner and Bleakney Phys. Rev. 1947,72 807. Honig J. Chem. Phys. 1948 16 105. * Smith Phys. Rev. 1937,51 263. lo McDowell and Warren Discuss. Faraday Soc. 1951 10 53. l1 Morrison J. Chem. Phys. 1954,22 1219. l2 Fox Hickam Grove Kjeldaas Rev. Sci. Instr. 1955 26 1101; Hutchison J. Chem. Phys. 1956 24 628. Stevenson Discuss. Faraday Soc. 1951,10 35 CUTTTBRRT MASS SPFCTROMETRY OF rREC RADICALS 219 which is clearly less than A(CH -+ CH,+). If the electron energy (E) were such that A(CH3 -+ CH3+) < E < A(CH4 -+ CH,+) the ion beam CH3+ would only be formed from methyl radicals; an ion beam with mass 15 is then evidence for the presence of methyl radicals in the gas sample.At high electron energy however there will be a peak at mass 15 due to methane (and other molecules containing a methyl group); detection of the radicals in the sample is therefore most conveniently done at low electron energies. Unfortunately the rapid change of cracking patterns with electron energy and the relatively low ionization efficiency at low energies makes it necessary to use beams of high-energy electrons for measurement of concentrations even for radicals. The utility of the mass spectrometer in observing free radicals is due to several factors (i) the method is more specific than Paneth and Hofeditz's the catalytic probe,15 or Wrede-gaugels methods since it is diagnostic for molecular weight; (ii) mass spectrometry is less selective than optical spectroscopy,l for this is dependent on the radical's having a considerable absorption in a suitable energy range; for example the HOz radical has not yet been detected spectroscopically although its detection was one of the early triumphs of the mass-spectrometric method;17 (iii) although the ion source must operate at very low pressures it is possible to sample through a pinhole leak (see below) from much higher pressures and thus bridge the gap between the low pressures necessary for Paneth techniques and the rather high pressures desirable in spectroscopy; (iv) continuous records of concentration changes can be obtained by focusing on one peak; (v) the mass spectrometer can be used as a fast scanning instrument to record rapid changes in concentration of several substances at once; (vi) the presence of oxygen in the mixture which makes the mirror technique impossible and often renders spectroscopic techniques difficult owing to Schumann absorption is no disadvantage to the mass spectrometer.(21) Experimental Difficulties.-(i) Sampling. The essential nature of free radicals viz. their reactivity especially their tendency to recombine or react on surfaces makes it diacult to transfer radicals from the gas system under investigation into the electron beam of the mass spectro- meter. The gas system in which the radicals are produced e.g. a homo- geneous gas reaction is often at too high a pressure for direct coupling to the ion source and some form of sampling leak is required.A common type of leak used in simple analytical mass spectrometry is a sintered disc this is of no use in free-radical studies as it presents to the gas a large surface area thus giving radicals a high chance of recombining. The idea form of leak is a single pinhole in a very thin diaphragm so that moleculei may pass straight through without collision with the walls of the leak Paneth and Hofeditz Ber. 1929 62 1335. Wrede 2. Physik 1929 54 53. l5 Tollefson and LeRoy J. Chem. Phys. 1948,16,1057. l7 Foner and Hudson J. Chem. Phys. 1953,21 1608. 220 QUARTERLY REVIEWS The types of leak most often used are a hole sparked in the very thin end of a quartz or glass thimble,18 or a hole drilled in gold leaf.lg A serious source of error is the possibility of a molecule's colliding with the hot filament which is usually a tungsten wire heated electrically to about 2000"~; there is a high probability that any molecule hitting this filament will dissociate thereby forming radicals which may be subse- quently ionized.This effect can be lessened by differential pumping of the filament region which to be effective requires that the filament be in an enclosure separate from the rest of the ion source and with its own pump the only connection between the two regions being the slit by which the electrons enter the ionization chamber. This system was used in the first paper on this subject by Eltenton;ls Lossing and Tickner18 enclose the filament in a chamber which is pumped out by the pump on the analyser tube via a series of holes in the six ion-accelerating plates (see Fig.5) the sides of this pumping path being closed by glass spacing rings between these six plates. (ii) Calibration. It has already been stated that if the cracking patterns of all the components of a mixture are not known it is not possible to make a full quantitative analysis of the mixture; knowledge of the cracking pattern of a pure component means not only knowledge of the relative heights of the various peaks but also the absolute sensitivity under the experimental conditions used. In the case of free radicals this presents some formidable problems for it is not possible to introduce into the mass spectrometer pure free radicals at a known pressure. In the rare cases where it is possible to say that only one free radical contributes to the mass spectrum the spectrum of this radical can be found by subtracting the spectra due to all other components but it is still necessary to measure the sensitivity for this one radical before its absolute concentrations can be measured.LeGoff and Letort20 obtained the spectrum of the methyl radical having made the radical by pyrolysis of methyl iodide or tetra- methyl-lead on a tungsten filament. They were able to show that peaks at masses 13 and 14 were not due to CH or CH2 radicals since the ap- pearance potentials were higher than those of peaks 15 and 16; after subtraction of contributions due to methane and ethane there remained the following spectra due to the methyl radical Mass From methyl iodide From tetramethyl-lead 15 100 14 32 5 4 13 9 r t 1 12 3.6 & 0.3 Lossing and Tickner J.Chem. Phys. 1952,20,907. LeGoff and Letort J. Chim. phys. 1956,53,480 lo Eltenton J. Chem. Phys. 1947,15 455. CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 22 1 In simple cases such as these it is possible to obtain a value for the sensitivity for the radical by assuming a 100% material balance; e.g. in the pyrolysis of dimethylmercury the pressures of methane and ethane and the reactant can be obtained from their (known) mass-spectrometric sensitivities under the conditions of the experiment and a perfect carbon balance being assumed the pressure of methyl (the only other product) can be calculated. In this way Lossing and TickneP found the sensitivity of their instrument for methyl radicals to be 0.47 & 0.07 of the sensitivity for methane.The difficulty of such a calibration is increased when as is often the case the reaction producing the radicals is at a high temperature; the magnitude of a mass-spectral peak decreases with increasing temperature. This decrease is largely due to a change in the flow through the sampling leak caused by the temperature coefficient of viscosity of the gas but there is 20 - - Temperature FIG. 3 The temperature coefficient of sensitivity. A Methane; B ethane; C di- methylmercury. (Reproduced by permission from J. Chem. Plays. 1953 20 907.) also an instrumental effect. Fig. 3 shows the variation of sensitivity with temperature for methane ethane and dimethylmercury;18 the sudden drop in the curve for dimethylmercury is due to its disappearance by decomposition the broken curve being an extrapolation of the Iow- temperature part of this curve obtained from the almost linear plot of sensitivity against 1/T.The temperature sensitivity of methyl which cannot be measured was assumed to be the same as that for methane; since their masses are nearly equal the temperature variation in viscosities would be expected to be very similar. ( c ) Apparatus.-In this section three apparatuses will be described which illustrate many of the general principles outlined above; they are 222 QUARTERLY REVIEWS the first apparatus used in studying free radicals in chemical reactions by Eltenton,19 a more typical later apparatus developed by Losing5 and extensively used throughout Canada and the molecular beam sampling system of Foner and Hudson.21 P I FIG.4 Eltenton's reaction vessel and ion source. (Reproduced by permission from J . Chem. Phys. 1947,15,455.) Eltenton's apparatus is shown in Fig. 4; a considerable limitation in this case was imposed by the use of a 180" Dempster-type mass spectro- meter22 in which the whole of the ion source is contained in the magnetic field and is less accessible. This apparatus is nevertheless noteworthy for the fact that the ionization and ion-accelerating chambers were evacuated by two separate large diffusion pumps (P, P3) while the filament (Pa) and analysing chambers were separately evacuated by smaller pumps. The reaction vessel consisted of a spiral heater wound in a double-walled quartz tube (Q) down the centre of which the reactant was introduced.The reaction zone was sampled through a gold-leaf diaphragm (D) the remaining gases being pumped out by P1 through the annular space between Q and the water jacket (W). The ion-source pressure was main- tained at 1 micron which is considerably higher than would be the case in a more modern instrument and dangerously high if one wishes to avoid all ion-molecule reactions; the pressure in the reaction vessel could be as high as 10 cm. Low electron energies were used for detecting radicals. t1 Foner and Hudson J. Chem. Phys. 1953,21,1374. 2z Dempster Phys. Rev. 1918 11 316. CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 223 FIG. 5 Lossing's reaction vessel and ion source. (Reproduced by permission from Ann. New York Acad. Sci. 1957 67 499.) Fig. 5 shows a recent version of the apparatus developed by Lossing and Tickner,l* which differs from Eltenton's mainly in being built around a 90" Nier-type mass spectrometer.The filament pumping via a series of holes in the ion-accelerating plates has already been described. The sample leak consists of a quartz thimble through the thin top of which a hole is sparked by a Tesla discharge; the holes used are about 30 microns in diameter and 10-20 microns long. The simple pinhole leak gives a beam of molecules distributed over a very wide angle many of which will therefore collide with the walls of the ion source before being ionized; Foner and Hudson's molecular beam systemz1 (Fig. 6) greatly reduces the beam spread and increases the chance of a molecule's entering the electron beam before striking a wall. Slits 1 and 2 are circular holes; the space between slits 1 and 2 was maintained at about 5 x 10-3 mm.pressure and the ion source at about mm. The molecular-beam intensity was a maximum when the mean free path in the first (scattering) region was equal to the distance between slits 1 and 2. The molecular beam passed through the ion source along the path of the electron beam instead of at right angles to it. 224 QUARTERLY REVIEWS nt Pump Pump Pump FIG. 6 Molecular beam sampling system. (Reproduced by permission from J. Chem. Phys. 1953 21 1374.) The sensitivity of the instrument to the sample was greatly increased by “chopping” the sample beam in the first region by means of a vibrating reed driven electronically and by amplifying the oscillatory component of the ion current at this same frequency.A discrimination of 10,OOO in favour of the sample beam rather than the background was obtained; if the background arose from beam scattering an additional tenfold discrimination was obtained by using the molecular beam system since the density in the beam was about ten times that in the rest of the apparatus. The sensitivity was still further increased by the use of an electron multi- plier on the ion collector and by using pulse-counting techniques for very weak signals. The majority of experiments on free radicals with mass spectrometers have employed a flow reaction system and slow scanning of the mass spectrum. A few investigators have made use of the fast scanning possibili- ties of the mass spectrometer. With a conventional instrument this can be achieved with electrostatic scanning by applying a sawtooth voltage wave- form to the ion-accelerating plates the resulting modulated ion current being displayed on a cathode ray tube whose time-base is derived from the sawtooth wave; such an arrangement was used by Foner and Hudson.21 Fast scanning is also a feature of the time-of-flight mass spectrometers which are based on the principles of the cyclotron or the drift Kistiakowsky and have used a drift tube type of spectrometer to scan a spectrum every 50 microseconds but the reproducibility of the spectrum is sacrificed for speed owing to statistical variation in the number of ions formed in each pulse.Examples Mass spectrometers have been used to study free radicals in pyrolyses photolyses combustion reactions including flames and heterogeneous reactions.The identification of free radicals in these processes has proved of great assistance in determining the mechanisms of the reactions but since the failure of the mass spectrometer to detect a particular radical 23 Ref. 3 pp. 74-85. 24 Kistiakowsky and Kydd J. Amer. Chem. SOC. 1957 79 4825, CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 225 cannot be taken as evidence that the radical is not present in the reaction it is rare for any reaction mechanism to be unequivocally determined by this technique. In some favourable cases it has been possible to obtain an estimate of the rate of an elementary radical reaction but considerable difficulty is found not only in the measurement of absolute concentrations in the mass spectrometer but also in the estimation of contact times and temperatures in the flow types of reaction vessel usually used.These points will be illustrated in the following sections of this Review; the measurement of ionization potentials of free radicals and the subsequent estimation of bond dissociation energies will also be discussed. (a) Detection of Radicals and Measurements of Concentration.- (i) Pyrolyses. The first study of radicals formed by pyrolysis was made by EltentonlS in the pyrolysis of some simple hydrocarbons. Using a quartz vessel he found methyl radicals to be present in the pyrolysis of methane at 10-300 microns pressure and 800-1100" c. The production of methyl radicals had an activation energy of 46 kcal./mole; methylene was not found although it could easily be detected in the same apparatus from the pyrolysis of diazornethane.This was interpreted in favour of a split of the methane molecule into a methyl radical and a hydrogen atom; the latter species was not found being presumed to recombine rapidly to form molecular hydrogen which was detected. The ethyl radical proved difficult to detect in Eltenton's instrument owing to the fact that ethylene which was almost always present with ethyl has an ionization potential only about two volts higher than that of ethyl and the isotope 13CH2-12CH2 at mass 29 interferes with the parent peak of C2H,. Some radicals which might confidently have been expected to be present were undetected e.g. vinyl in the pyrolysis of propene which yielded abundant methyl radicals. Lossing and TickneP studied the pyrolysis of dimethylmercury and di-t-butyl peroxide using a few microns of reactant in 5 to 10 mm.of helium; the carrier gas reduces the probability of wall reactions by hinder- ing diffusion to the walls and also assists in obtaining a high flow rate. The instrument was calibrated for methyl radical concentrations as described above and it was then possible to study the first-order decompo- sition rate of the di-t-butyl peroxide; the decomposition was found to have an activation energy of 37 kcal./mole in good agreement with previous All the ethane formed in the decomposition of dimethyl- mercury being assumed to come from the recombination of methyl radicals the collision efficiency of this process at 859" was of the order of this calculation neglected concentration gradients in the furnace. I ngold and Lossing,26 who studied this radical recombination more extensively found it to consist of a second-order homogeneous reaction a5 Raley Rust and Vaughan J.Amer. Chem. SOC. 1948 70 88; Szwarc J. Chem. *6 Ingold and Lossing J . Chem. Phys. 1953,21,1135 ; Ingold Henderson and Lossing Phys. 1951 19 698. ibid. p. 2239. 226 QUARTERLY REVIEWS and a first-order reaction presumed to be heterogeneous. Considerable difficulty was encountered in the measurement of contact times in the furnace; by taking measurements with a retractable furnace at several distances from the sampling leak the amount of reaction corresponding to the reaction zone formed by retracting the furnace could be measured and the unknown concentration gradients in the furnace itself could be ignored. From a knowledge of the temperature gradient in the rdevant zone the appropriate contact time could be calculated.The recombina- tion of methyl radicals was then found to be proportional to the total pressure and had an activation energy of -1-5 kcal./mole attributed to the decrease in the lifetime of the activated C2H,* complex with increasing temperature. Lossing Ingold and Henderson2' studied the pyrolysis of some simple compounds containing the C-0-C linkage. They found methyl radicals in the decomposition of ethylene oxide but very few in the case of dioxan; no other radicals were detected. In the case of ethylene oxide they suggested a mechanism based on the primary reactions (CH2)20 + CHI + CO . . . . . . . . . (iii) (CH,),O + CH + CHO . . . . . . . . (iv) followed by CHO + C O + H CH f (CH2)2O -f CH4 $- C2H.90 H + (CH2)20 -f H + C,H,O and rapid decomposition of the C2H30 radical plus various radical- radical reactions.Failure to detect H CHO or C,H,O implies that these species if present react very rapidly under the conditions used (ca. 10 microns of reactant in 15 mm. of helium at about 900"). No stable products of higher molecular weight than (CH2),0 were found implying that the C2H30 radical if it exists decomposes. It being assumed that there are no reactions for the formation of methane or the removal of hydrogen atoms other than the above reaction (iv) occurred in at least 37 % of the molecules decomposed; the absence of methylene was regarded as evidence against the reaction (CH,),O -f CH20 + CH proposed by Fletcher and Rollefson.2a The pyrolysis of several aromatic compounds has been studied at over lOOO" and many aromatic radicals were detected together with considerable free carbon and hydrogen suggesting the occurrence of heterogeneous processes.29 (ii) Photolyses.The study of high-temperature reactions entails the 27 Lossing Ingold and Henderson Discuss. Furuday Soc. 1953,14 34. 28 Fletcher and Rollefson J. Amer. Chem. Soc. 1936 58 2135. 29 Ingold and Lossing Canad. J. Chem. 1953,31 30. CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 227 disadvantage that the more unstable radicals will themselves be pyrolysed. For example the 2- and the 4-methylbenzyl radical formed almost quantitatively from their iodides at 825" are themselves pyrolysed at any higher temperatures being completely dissociated at 1OOO" ; dissociation of the 3-methylbenzyl radical only becomes appreciable above 1050°.30 Photolysis offers the chance of producing at low temperatures radicals which are unstable at the high temperatures necessary for pyrolysis.Unfortunately in the optimum conditions for mass spectrometric observa- tion of the radicals in a homogeneous system i.e. at a low pressure of reactant in a carrier gas most reactants have such low absorption coeffici- ents that sufficient light cannot be absorbed with the available path lengths; it is therefore necessary to resort to photosensitisation or to flash- photolytic techniques. Mercury-photosensitised reactions have been studied by Lossing and his co-workers in the apparatus shown in Fig. 7. The mercury lamp is of Reaction zon FIG. 7 Reaction vesselfor photofytic studies. (Reproduced by permission from Canad.special design,31 the discharge being constrained in a zig-zag path which almost fills the annular lamp body; the body of the lamp and the mercury J. Chem. 1957 35 305.) 30 Farmer Marsden and Lossing J. Chem. Phys. 1955 23 403. 31 Lossing Marsden and Farmer Canad J. Chem. 1956 34 701. 228 QUARTERLY REVIEWS saturator were surrounded by water at 55-60' so that the partial pressure of mercury in the gas system was 18-25 microns; the absorption of 2537 radiation in the reaction volume of 1.39 C.C. was calculated to be greater than lo1* photons/sec. As in pyrolytic studies the various reactants at partial pressures of a few microns were carried in a stream of helium at a pressure of about 10 mm. The detection of free radicals at low electron energies could con- ceivably be falsified in the case of photolyses if any excited molecules were present ; the ionization potential of an excited molecule e.g.C3H6* would be lower than that of the molecule in the ground state and the appearance potential of any ion e.g. CH3+ from it might also be lower than the normal appearance potential from the ground state molecule. In fact no such lowering of the ionization potentials of stable molecules was observed the lifetimes of any excited molecules being probably too short .31 In the mercury-photosensitised decomposition of ethylene only acetylene and hydrogen were detected supporting the idea of a molecular rearrange- ment. Propene gave large quantities of allyl radicals allene and hydrogen; when the contact time was decreased by lowering the shutter (see Fig.7) the ratio of allene to allyl decreased rapidly indicating that the allene is a secondary product from allyl. In view of the high concentration of allyl radicals and Hg atoms compared with that in conventional photo- sensitization experiments the allene was presumed to have been formed by The reaction would be too slow at 60" to form the large amounts of allene found and similarly a unimolecular split of the excited allyl formed in the initial reaction of propene with an Hg atom into allene and a hydrogen atom would be expected to have an activation energy of at least 30 k cal ./mole. A mass-spectrometric study of flash-photolytic reactions has been made by Kistiakowsky and K ~ d d ~ ~ who used a time-of-flight mass spectro- meter. It was found that the photolytic flash suppressed the mass spectrum for several hundred microseconds unless a negatively charged grid were placed between the lamp and the sampling leak; the disturbance was shown to be photochemical rather than electrical and may be connected with a photoelectric effect's forming ions in the reaction cell which in some way interfere with the working of the ion source.In the flash photolysis of keten methylene radicals could not be detected but the formation of ethylene was almost complete in 50 microseconds; this was thought to be due to the rapid reaction CH,:CH*CH,. + Hg -+ CH,:C:CH + H + Hg (IS,) CH,:CH.CH,. + H -+ CH,:C:CH + H CH + CH,:CO -+ C,H + CO CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 229 (iii) Combustion reactions. Flames have been studied by Eltentonlg and by Foner and Hudson.21 The former claimed to observe the HO radical (amongst others) in methane propane and carbon monoxide flames but surprisingly could not find OH.Foner and Hudson found some H OH and 0 radicals in a hydrogen-oxygen flame but the maximum radical concentration was of the order of 1 % and H 0 2 could not be found even up to a burner pressure of half atmospheric. In a methane-oxygen flame they measured the spectrum with 15 18 and 48 v electrons but were unable to identify all the components because of the complexity of the spectrum and the possible occurrence of excited molecules or metastable species (not radicals) of unknown cracking patterns; the only radical definitely identified was methyl which was shown by the very large peak at mass 15 and by appearance-potential measurements.Identification of HO was complicated by the occurrence of methyl alcohol in unknown amounts since this has a peak at mass 33 due to l3CH,-OH. confirmed the presence of HO and OH in the hydrogen-oxygen reaction at 10oOo but were unable to find any H or 0 atoms. They also investigated the reactions of methyl radicals from the pyrolysis of dimethylmercury with oxygen3 and with nitric oxide:33 collision efficiencies of to were reported in each case and a large number of products including free radicals was observed. In these experiments the heating element was exposed to the gas stream and heterogeneous reactions were thus highly probable. The HO radical was first conclusively demonstrated by Foner and Hudson17 in the reaction of hydrogen atoms from a Wood’s discharge tube with molecular oxygen carried in an inert gas; the peak at mass 33 increased and decreased as the discharge was switched on and off although that at mass 32 was unaffected.Large yields of H02 were later obtained from the reaction between hydrogen peroxide and the products of a discharge in water v a p o ~ r . ~ ~ Robertson35 also observed H02 in the reaction of hydrogen atoms with oxygen; in this case the reaction products from the main flow line diffused down a side tube at the end of which was the leak to the mass spectrometer. Under the conditions used a radical would make many thousands of collisions in the gas phase and on the walls during this diffusion process; it is therefore not surprising that OH and 0 could not be detected in the reaction. From the fact that the HO concentration increased with increasing pressure in the reaction vessel Robertson deduced that the excited H02* formed in the reaction Ingold and H + 02+ HO,” .. . . . . . . . . . (4 32 Ingold and Bryce J. Chem. Phys. 1956,24 360. 33 Bryce and Ingold J. Chem. Phys. 1955,23 1968. 34 Foner and Hudson J. Chem. Phys. 1955,23,1364. 35 Robertson “Applied Mass Spectrometry” Institute of Petroleum London,1954 p. 112. 230 QUARTERLY REVIEWS was stabilised by gas-phase collisions H02* + M -f HO + M . . . . . . . . . . (vi) and he showed that a lifetime of to 10-l2 seconds for the excited complex was sufficient if reactions (v) and (vi) occurred at every collision. (iv) Heterogeneous reactions. In studying cracking reactions which occur on heated metal filaments L e G ~ f f ~ ~ and Robertson3' placed the heated filament in the mass spectrometer's ion source above the top plate or ion repeller which was replaced by a wire gauze.The reactant flowed over the catalytic filament at a low pressure to mm.) so that homogeneous reactions were a minimum. The catalytic filament and the surroundings need to be at the same potential as the ionization chamber to prevent electrons from the catalytic filament being accelerated so that they might ionise gas molecules in the reaction vessel (i.e. above the ion ~epeller.~~ Robertson37 found methyl radicals in the decomposition of methane on a platinum filament at about 1000"; the formation of ethane propane and butane was attributed to secondary reactions on the cold walls of the mass spectrometer the overall reaction scheme being CH -f CH + H filament C,H -+ C,H4 + 2H filament C,H4 + H -+C,H filament CH + CH -+ C,H6 walls CH 4- C2H -+ CH + C,H walls CH + C2H5 -+ C3Hs walls C2H5 f C2H5 C4H10 walls Ethyl radicals were not detected in the decomposition of ethane only ethylene and hydrogen being observed; this was attributed to a molecular dehydrogenation process on the filament.Blanchard and L e G ~ f f ~ ~ showed that there was a difference in the mechanisms of pyrolysis of some substances on clean and carburised tungsten filaments. On clean tungsten only H2 and S2 were detected in the decomposition of H,S whereas on carburised tungsten CS and CS2 were also found being formed by the reactions Recently some attempts have been made to use field emission as a source of ions for a mass-spectrometric study of species adsorbed on metal surfaces; this method of ionization has the great advantage that fragment ions are produced only in very small quantities compared with electron-impact ionization; it is therefore much easier to identify free 36 LeGoff J .Chim. phys. 1953 50 423. 37 Robertson Proc. Roy. SOC. 1949 A 199 394. 38 LeGoff J. Chim. phys. 1956,53 369. 3s Blanchard and LeGoff Canud. J. Chem. 1957 35 89. CUTHBERT MASS SPECTROMETRY OF FREE RADICALS 23 1 radicals. Gomer and InghramgO detected the CH,.O radical from methyl alcohol adsorbed on the surface of tungsten and Beckey41 has found the C2H,-0 radical from ethyl alcohol on tungsten. (b) Ionization Potentials and Bond Energies.-The experimental difficult- ies in measuring ionization potentials have already been described; even if they can be overcome electron-impact ionization potentials are thought to correspond to “vertical” Franck-Condon transitions and may be appreciably higher than the “adiabatic” spectroscopic values.For free radicals apart from methyl spectroscopic values are not available and there is no alternative to electron-impact data; for the methyl radical the electron-impact value36.42 of 9.85 v is only 0.01 v above the spectroscopic value.43 It should be noted that the errors quoted by most workers for electron-impact values represent limits of reproducibility and are not estimates of the absolute errors which are often much larger. measured the ionization potential of the H02 radical as 11.53 & 0.1 v and its appearance potential from H202 as 15.41 & 0.1 v; the difference between these values is the 0-H bond energy in H202 if the fragments have zero kinetic energy in the process H,O + e- 3 HO$ + H + 2e- Since I(H) > I(HO,) Stevenson’s rule13 is satisfied and the assumption of zero kinetic energy is most probably satisfactory.This gives the bond energy D(H02-H) = 3.88 v = 89.5 kcal./mole. From the (known) heat of reaction H + H + O2 -+ H202 + 136.6 kcal. (= 5.92 v) Foner and Hudson derived the bond energy D(H-02) = 2-04 3 0.1 v = 47 2 kcal./mole. The ionization potential of the NH2 radical is 11.4 rfI 0.1 v,44 and its appearance potential from ammonia is 16.0 & 0.1 v giving a bond energy D(NH2-H) = 4.6 3 0.15 v = 106 & 3 kcal./mole in excellent agreement with thermochemical and spectroscopic data.45 Similarly investigation of hydrazyl N2H3 from hydrazine gave a bond energy D(N,H,-H) = 76 & 5 kcal./m Use of ionization and appearance potentials to estimate bond energies depends on a knowledge of the nature of the non-ionized undetermined fragments. In the above cases there was no difficulty as the radical formed by electron impact was simply a hydrogen atom. If the non-ionized fragment is polyatomic there is always the possibility of its dissociating in which case the relevant bond dissociation energy must be taken into account; there is also a possibility of the formation of negative ions which would normally be undetermined but which could in principle be Foner and 40 Gomer and Inghram J. Arner. Chem. Soc. 1955 77 500. 41 Beckey Naturwiss 1958 45 259. 42 Langer Hipple and Stevenson J. Chem. Phys. 1954 22 1836. 43 Herzberg and Shoosmith Canad.J. Phys. 1956 34 523. 44 Foner and Hudson J. Chem. Phys. 1958 29 442. 46 Altshuller J. Chem. Phys. 1954 22 1947. 232 QUARTERLY REVIEWS detected. In an attempt to estimate the bond energy in the CCI radical Blanchard and L e G ~ f f ~ ~ measured the ionization potential of the CCI radical formed by the pyrolysis of carbon tetrachloride on tungsten. In this case the required bond energy is very dependent on the processes assumed for the appearance of CC1,+ and CC1+ from CCI,; in the former case for instance the possibilities are CCI -+ e- -+ CC12+ + CI + 2e- A = 0.0 d = 2.47 v A = 3-8 v CCI + e- -+ CCl,+ + CI + CI + 2e- CCI + e- -+ CCl$ + CI + C1- + e- CCI + e- -+ CCl,+ + CI- + CI- A = 7.6 v The values of d show the magnitude of the energy differences between these processes.The authors were able by considering also the appearance of the ion CCl+ to restrict the possible solutions to two only viz. D(CC1,- C1) = 62 A list of electron-impact ionization potentials of free radicals is given by Los~ing.~ 7 or 93 & 7 kcal./mole. Conclusion The mass spectrometer has made available considerable information on the radicals present in a wide variety of homogeneous and hetero- geneous gas reactions. Many of the possibilities of the method have not yet been exploited owing to the formidable experimental difficulties. A greater use of the molecular beam sampling system21 seems to be desirable to overcome the sampling problem and electron multipliers or the use of pulse counting techniques would increase the sensitivity of the instruments used.Time-of-flight mass spectrometers have yet to prove their value especially as fast scanning can be obtained on a Nier-type instru- ment without the disadvantages of the time-of-flight instrument. An extension of the principle of low electron energy detection of free radicals should allow one to detect the presence of excited molecules since their ionization potentials are lower than those'of ground-state molecules by the amount of the excitation energy. More accurate measurements of ionization and appearance potentials will be required and it is desirable to use a technique which eliminates the wide Boltzmann energy distribu- tion of the electron beam12 or alternatively photo-ioni~ation.~~ Excited molecules will also present more acute sampling difficulties since they are in general more unstable towards wall collisions than the radicals so far detected. With these improvements in technique which may confidently be expected within a few years mass spectrometry should prove one of the most useful methods of following unstable intermediates in gas-phase reactions. The Reviewer thanks Dr. J. W. Linnett F.R.S. for his advice and Imperial Chemical Industries Limited for a Fellowship during the tenure of which this Review was written. 46 Lossing and Tanaka J. Chem. Phys. 1956,25 1031 ; Hurzeler Inghram and Morrison J . Chem. Phys. 1958,28,76.

ISSN:0009-2681    

DOI:10.1039/QR9591300215

出版商:RSC    

年代:1959

数据来源: RSC

摘要:

PERFLUOROALKYL DERIVATIVES OF METALS By J. J. LAGOWSKI PH.D. (UNIVERSITY CHEMICAL LABORATORY CAMBRIDGE*) AND NON-METALS SINCE the conclusion of World War 11 perfluoroalkylt derivatives of many elements have been prepared and characterised. This Review deals with the chemistry and properties of the known perfluoroalkyl derivatives of the elements other than carbon oxygen nitrogen and the halogens. Many of the physical and chemical properties of trifluoromethyl derivatives of the elements can be correlated with the dual nature of this group. As an organic radical it gives compounds formally related to normal alkyl compound and many of the reactions of perfluoroalkyl derivatives are analogous to those of the corresponding covalently bonded alkyl compounds. Replacement of the hydrogen atoms in an alkyl group by fluorine atoms produces a group with a substantial inductive character i.e.a highly electronegative group and a number of observations suggest that the effective electronegativity of a trifluoro- methyl group lies between that of fluorine and of chlorine. Effective Electronegativity of Perfluoroalkyl Groups.-Values for the electronegativity $ of an atom have been principally obtained from (a) thermochemical data (b) ionization potentials and electron affinities and (c) stretching force constants ; considerations of other molecular properties have also been used. Methods (a) and (c) have been extended to the determination OF the electronegativity of organic radicals but method (b) does not appear practicable at present since so little is known of the electron affinity of organic groups.Qualitative electronegativity scales for organic radicals are usually based on a comparison of the relative re- activities of various molecules containing the radical under observation or on the relative reactivities of molecules possessing a particular structure with varying substituents. The results of such studies are often conflicting and it has been concluded that other factors besides the electronegativity of the radical are operative.a An attempt is made here to establish the effective electronegativity of the trifluoromethyl group by using methods (a) and (c). The empirical observation has been made that the energy of a bond A-B [D(A-B)] is generally larger than the additive mean of the energies of the bonds A-A [D(A-A)] and B-B [D(B-B)] and the difference d * Present address Department of Chemistry The University of Texas Austin.t In this Review the terms fluoroalkyl and pertluoroakyl denote partially fluorinated Electronegativity values are those given by Paulingl unless otherwise stated. Pauling "The Nature of the Chemical Bond" 2nd edn. Cornell University Press Pritchard and Skinner Chern. Rev. 1955,55 745. and fully fluorinated al ky1 groups respectively. Ithaca New York 1948. 233 3 234 QUARTERLY REVIEWS (defined by equation l) was found to be proportional to the difference between the electronegativities of the atoms (equation 2) :l A == [D(A-B)] - &[D(A-A) + D(B-B)J . . . (1) Bond dissociation energies of a number of simple trifluoromethyl compounds of the type CF3X have been reported and the calculated values of d and the electronegativity differences for these compounds TABLE 1.The electronegativity of the CF group estimated from bond dissociation energies. Cl 79.5 a 17;5 0.88 3.9 Br 64.5 9.0 0.62 3.4 I 57.0’ 6-6 0.53 3.3 H 1 02-0 18.0 0.88 3.0 Calculated from equation (1) by using the following values of D(X-X) (CF& 65 (ref. 118); H2 104; C1,58.0; Br 46-1 ; I2 36.1 (ref. 1) (all in kcalJmofe). Calculated from equation (2). Calculated by assuming the “best values’’ of the electronegativity of X as given in ref. 1 viz. H 2.1 CI 3.0 Br 2.8 I 2.5. Rabinovitch and Reed J . Chem. Phys. 1954,22 2092. Sehon and Szwarc Proc. ROJJ. Soc. 1951 A 209 110. f Farmer Henderson Lossing and Marsden J. Chem. Phys. 1956,24 348. Q Pritchard Pritchard Schiff and Trotman-Dickenson Trans. Faraday Suc.1956 52 849. appear in Table 1. The values for A parallel the trends observed in the hydrogen halides but unlike those of the alkyl halides the values for the electronegativity of the trifluoromethyl group apparently do not require a correction term2 There is fair agreement between the values calculated for the electronegativity of the trifluoromethyl group from the data for CF,Cl CF3Br CF3T and CF3H and it is significant that all are equal to or greater than the electronegativity of chlorine. The shift which occurs in the characteristic infrared absorption fre- quency of a bond with changes in the nature of a substituent originates in the altered electron distribution in the bond brought about by the overall electron-withdrawing ability (which is the sum of the inductive and mesomeric effects) of the substituent.It is often difficult to separate these effects in systems where the bonds in question are coplanar. Non-planar structures should be comparatively free from mesomeric effects and the substituent’s effect on the group frequency should be almost wholly due to an inductive effect operating along the bonds independent of the molecular geometry and dependent on the electronegativity of the substituent. A tabulation of effective group electronegativities has been LAGOWSKI PERFLUOROALKYL DERIVATIVES 235 made by observing the change in the position of the P-0 frequency in a series of phosphoryl cornpound~.~ The P-0 bond can be described as the sum of the structures +P=O and 3 P++O-. The former structure will be more important when strongly electronegative substituents are attached to the phosphorus atom and hence the frequency of the P-0 absorption band should increase with respect to that for similar compounds carrying less electronegative substituents.A similar study has been made by using the C=O frequency of a series of carbonyl compounds with varying substituents and the effective electronegativity of the trifluoroniethyl group has been calculated to be 3.20 on the Gordy scale of electro- negati~ity.~ The planar structure of the carbonyl group permits a greater degree of mesomerism than is possible in the tetrahedral phosphoryl group and it would be expected that the effective electronegativity of a substituent as determined by its influence on the position of the C=O vibration in a carbonyl compound will be 1 s than that determined from the shift in P-0 vibration frequency.The position of the characteristic group frequencies in several other systems has been shown to be dependent on the electronegativity of the ~ubstituents,~ and the approach used to determine the effective electro- negativity of the trifluoromethyl group is similar to that used in references 3 and 4 but with some modifications. The dependence of the phosphoryl carbonyl and HX vibrations on the sum of the Pauling electronegativities of the substituents is shown in Fig. 1. Substituents were chosen whose f800 7850 7300 C=Q 4000 H-X ' ' ' ' ; ' ' ' ' ; ' ' " ' 7350 1400 1450 P=O 3000 4 k 9 A ' 2k10 1300 ~ c m - 9 FIG. 1 The PO CO and HX stretching frequency as the sum of the electronegativities of substituents in A X&O; B X3PO; and C HX respectivel-v.* Bell Heisler Tannenbaum and Goldenson J. Amer. Chern. Soc. 1954 76 5185. Kagarise J. Amer. Chem. Soc. 1955,77 1377. Daasch S'ectrochim. Actu 1958 13 257. 236 QUARTERLY REVIEWS electronegativities had been determined by other methods i.s. hydrogen and the halogens (Tables 2 3 4). By using the carbonyl or phosphoryl absorbing frequencies for compounds containing the trifluoromethyl group the effective electronegativity of this group was estimated graphic- ally. The P-0 frequency in tristrifluoromethylphosphine oxide occurs6 at 1327 cm.-l which corresponds to an effective electronegativity of 3.3. The results of a similar treatment of trifluoroacetyl halides appear in Table 5. The values obtained for the acid chlorides being excluded the effective electronegativity of a trifluoromethyl group lies between those of chlorine and fluorine.TABLE 2. Phosphoryl absorption frequencies for X3P0 compounds. X in X3P0 F c1 Br CF3 =xx 12.0 9.0 8-4 9.9 vpo(crn.-l) 1415" 1290" 1266" 1327 Data taken from ref. 3. Determined graphically. Data taken from ref. 6. TABLE 3. Carbonyl absorption jrequencies for X,CO compounds. X,CO COF COC1 COBr COHF COClF COHz =XX 8.0 6.0 5.6 6.1 7.0 4.2 vcO(cm.-l) 1928" 1827" 1826b 1834* 1868" 1745" a Data taken from ref. 4. Bellamy and Williams J. 1957 4294. TABLE 4. The electronegativity of the CF3 group estimated from the position of the carbonyl absorption band for compounds of the type CF,*COX. XinCF,.COX F c1 Cl Br 1 CF3 X(X+-CF3) a 7.5 5.6 5.8 6.2 5.7 5.9 vco(cm.-l) 190lc 1810" 1820d 1838c 1812c 1825c XCFQ 3.5 2.6 2.8 3.4 3.2 3*Ob Estimated as 4 x(x+cP,).Estimated from Fig. 1. See ref. 9. Bellamy and Williams J. 1957 4294. TABLE 5. Absorption frequency of the HX vibration. X in HX F c1 Br 1 CF3 x 4.0 3.0 2.8 2.5 3*2b vHX(cm.-l)a 3875 2856 2553 2230 3021 See ref. 8. Determined graphically. The HX stretching frequency of the hydrogen halides corrected for mass effects has been shown to be a linear function of the electronegativity Emelkus Haszeldine and Paul J. 1955 563. LAGOWSKI PERFLUOROALKYL DERIVATIVES 237 TABLE 6. Absorption frequency of the SO vibration in compounds of the type S02X2. vs02(cm.-1> Compound - Sym. Asym. SO2FZa 8.0 1269 1502 SO,Cl,b 6-0 1205 1434 S0,CF3C1 6.2 ‘ 1240 1439 S02CF3F 7.3 ’ 1 240 1478 Martz and Lagemann J. Chem. Phys. 1954 22 1193. See ref. 148.See ref. 137. Estimated by the method outlined in the text. of X.’ The mass-corrected hydrogen stretching frequency in trifluoro- methane has been calculated to be 3021 crn.-li8 by using this value and the data for HX (Fig. 1) the effective electronegativity of the trifluoro- methyl group can be estimated as 3-3. Substituted sulphuryl compounds SO 2XH2 and S02XX’ should be essentially free from mesomeric effects on structural grounds and the variation of the S-0 stretching frequency with the substituent should be primarily due to inductive i.e. electronegativity effects of the substituents. Although the necessary infrared values are known for only two sulphuryl halides (Table 6) it is apparent that both the symmetric and the asym- metric stretching frequencies increase with increasing electronegativity of the substituents.If the trifluoromethyl group has an effective electro- negativity of 3-0 or greater the corresponding frequencies for trifluoro- methanesulphonyl chloride or fluoride should be between that of S02C1 and S02F2 and this is observed experimentally. The symmetrical stretching frequencies have been reported to be the same for both CF,.S02C1 and CF3-SO2F but there is a difference in the asymmetrical frequencies. Assuming that there is a linear relation between the sum of the electro- negativities of the substituents and the :SO2 asymmetrical frequencies the slope of which can be obtained from the experimentally determined frequencies of sulphuryl chloride and fluoride we can estimate the sum of the effective electronegativities of the substituents in trifluoromethane- sulphonyl chloride and fluoride as 6.2 and 7.3 respectively; it follows that the estimates for the effective electronegativity of the trifluoromethyl group are 3-2 and 3.3 respectively.The methods used to determine the effective electronegativity of a trifluoromethyl group give consistent results in that all of the estimated values lie between that of fluorine and of chlorine; the “best value” is apparently 3.3. The effective electronegativities of other perfluoroalkyl Bellamy “The Infra-red Spectra of Complex Molecules” Methuen and Co. Ltd. London 1958 pp. 377-409. a Perkins and Wilson J. Chem. Phys. 1952 20 1791. * Bellamy and Williams J. 1956 2753. 3* 238 QUARTERLY REVIEWS groups might be expected to be similar to that of the trifluoromethyl group.If the position of the C=O frequency is taken as a measure of the overall electron-withdrawing ability of the substituents then the position of this band for compounds of the type RC0.X (Rf=perfluoroalkyl X=halogen) should be essentially constant and independent of Rf for a given halogen. This effect is observed in the perfluoroacyl fluorides and chlorides (Rf = CF, C,F, C3F, C5Fll) where the C=O frequency is 1900-1891 and 1809-1815 cm.-l re~pectively.~ Preparative Methods.-Four general methods have been employed for the preparation of perfluoroalkyl derivatives of the elements (a) reactions involving the interaction of perfluoroalkyl halides with metalloids or metals (b) exchange reactions (c) the use of reactive perfluoroalkyl intermediates and (4 direct fluorination of alkyl compounds.Method d has been very successfully applied to the preparation of perfluoroalkyl derivatives of carbon nitrogen and oxygen but reactive intermediates such as RfLi and RfMgI (method c) have been used relatively little undoubtedly because of the difficulties involved in their preparation. High-temperature exchange reactions and exchange reactions in solutions between perfluoroalkyl iodides and alkyl derivatives have been employed in only a few instances. Method a has been the most successful for the preparation of organometallic compounds containing perfluoroalkyl groups. Perfluoroalkyl iodides react directly with metalloids and some metals at high temperatures to form perfluoroalkyl derivatives but the formation of the more reactive perfluoroalkyl derivatives (with respect to alkylation) requires the use of a solvent which aids in the stabilisation of the perfluoroalkyl compound (see p.239). Group 1.-Lithium is the only element in Group I which has been shown to form relatively stable and reactive perfluoroalkyl derivatives. Although perfluoroalkyl iodides do not react directly with lithium,1° perfluoroalkyl- lithium compounds can be prepared by a halogen-metal interchange with an alkyl-lithium (eqn. 3) ; trifluoroiodomethane however does not under- go this conversion.ll Trifluoromethyl-lithium has been reported but no experimental details were given.12J3 RLi + C,F,I -+ C,F,LI + RI . . . . . . . . . (3) Syntheses involving perfluoroalkyl-lithium compounds are best con- ducted by adding the alkyl-lithium and the other reactant simultaneously to an ether solution of the perfluoroalkyl iodide at -40” to -50”.Perfluoroalkyl-lithium reagents undergo the displacement and the addition reactions common to Grignard reagents and lithium alkyls and WeibIen in Simons “Fluorine Chemistry” Vol. 11 Academic Press Inc. New York 1954 p. 449. Pierce McBee and Judd J. Amer. Chem. Soc. 1954,76 474. Haszeldine Angew. Chem. 1954 66 693. lo Emeleus and Haszeldine J. 1949 2948. l2 Haszeldine Nature 1951 168 1028. LAGOWSKI PERFLUOROALKYL DERIVATIVES 239 are unstable with respect to hexafluoropropene and lithium fluoride ; in some instances the perfluoroalkyl-lithium reagent causes reduction of ketones to occur.14 Although the perfluoroalkyl-lithium compounds have not been investigated extensively as a synthetic route for prepara- tion of perfluoroalkyl derivatives of elements other than carbon hepta- fluoro-n-propyl-lithium reacts with diethylsilicon dichloride to yield diethylbisheptafluoro-n-propylsilane and chlorodiethylheptafluoro-n- propylsilane.ll Group 11.-Perfluoroalkyl derivatives of magnesium zinc and mercury have been prepared by the action of a perfluoroalkyl iodide on the element.The mercury derivatives are most satisfactorily prepared without a solvent but a basic solvent is necessary for the formation of the magnesium and zinc compounds. The nature of the solvent used is important and no reaction occurs in solvents with little or no basicicity. It has been suggested that basic solvents form molecular addition compounds with perfluoroalkyl iodides in which the nature of the C-I bond approaches that in the alkyl iodides because of a partial transfer of charge from the base to the initial- ly positive iodine atom.15J6 Under these conditions perfluoroalkyl iodides should undergo heterolysis more readily than in the absence of a basic solvent.A perfluoroalkylcadmium compound has been reported,12 but no details were given concerning its preparation. Although the alkyl derivatives of the Group I1 elements have been classically used as alkylating agents for the preparation of other organo- metallic compounds perfluoroalkyl-zinc or -mercury compounds have not yet proved useful for this purpose. Perfluoroalkylmagnesium deriva- tives have been suggested as a route to perfluoroalkyl compounds of elements other than carbon but no detailed investigations have appeared. 2~13~17 Magnesiurn. Perfluoroalkyl Grignard reagents can be prepared from perfluoroalkyl iodides and magnesium *in basic solvents in the usual manner.Although heptafluoro-n-propylmagnesium iodide has been the most widely investigated because of its ~ t a b i l i t y l ~ J ~ - ~ ~ there are indications that the less stable trifluoromethylmagnesium iodide can be prepared by using special p r e c a ~ t i o n s . ~ ~ ~ ~ ~ * ~ ~ The nature of the solvent the reaction temperature the dilution of the perfluoroalkyl iodide and the purity of McBee Roberts and Curtis J. Amer. Chem. Soc. 1955 77 6387. l5 Haszeldine J. 1952 3423. l6 Haszeldine J. 1953 2622. l7 Haszeldine Abstracts of Papers 120th Meeting of the American Chemical Society l8 Henne and Francis J. Amer. Chem. Soc. 1951,73 3518. lD Haszeldine J. 1953 1748.2o Henne and Francis J. Amer. Chem. SOC. 1953 75 992. 21 Pierce and Levine J. Amer. Chem. Soc. 1953 75,. 1254. 22 Brice Pearlson and Shnons J . Amer. Chem. Soc. 1946 68 968. 23 Haszeldine J. 1954 1273. 24 Haszeldine Abstracts of Papers 122nd Meeting of the American Chemical Society 1951 p. 6 ~ . 1952 p. 1 3 ~ . 240 QUARTERLY REVIEWS the magnesium used in the reaction have all been investigated as variables in the preparation of perfluoroalkyl Grignard reagents. Tetrahydropyran tetrahydrofuran tertiary amines ethyl ether and n-butyl ether have all been used as solvents but the aliphatic ethers are less effective for the stabilisation of the Grignard reagent than of the other compounds. The stability of perfluoroalkyl Grignard reagents increases but the rate of their formation and the ease of initiation of the reaction decreases with de- creasing temperature.As a practical compromise between these tempera- ture effects the reaction is initiated at room temperature or above and then the reaction temperature is lowered. Grignard reagent formation and reaction has been observed down to -80"; the optimum temperature appears to be about -20".15 The purity of the magnesium employed in the reaction has been cited as a critical factor in the formation of perfluoroalkyl Grignard reagents,15 but there appears to be no general agreement on this point. Although heptafluoro-n-propylmagnesium iodide reacts with aldehydes acid chlorides and esters to form the expected secondary alcohols ketones and tertiary alcohol^,^^^^^ there are no recorded attempts to extend these reactions to the preparation of other perfluoroalkyl- metallic compounds.Trifluoromethylmagnesium iodide undergoes similar reactions but the conditions for its formation are difficult to reproduce precisely.23 Better overall yields are obtained in reactions employing perfluoroalkyl Grignard reagents if the Grignard reagent is formed in the presence of the substance with which it is to react. An exchange reaction between perfluoroalkyl iodides and phenyl- magnesium bromide (eqn. 4; Rf = C2F5 C3F7) occurs at normal tempera- tures in ether solution and may prove to be more practicable from a manipulative standpoint :25-27 Zinc. Although trifluoromethylzinc derivatives have not been reported higher perfluoroalkylzinc halides are known.28 Heptafluoro-n-propylzinc iodide and bisheptafluoro-n-propylzinc have been isolated from dioxan or 1,2-dimethoxyethane solution as stable solvates ;2892g the removal of solvent molecules was reported to occur during high-vacuum s~blimation,~~ but this has not been ~onfirrned.~~ The ether solvate molecules are readily displaced by stronger bases and the 1 :1 pyridine addition compounds of heptafluoro-n-propylzinc iodide and bisheptafluoro-n-propylzinc are formed from the 1,2-dimethoxyethane solvates of these substances in this 25 Pierce Meiner and McBee J.Amer. Chem. Soc. 1953,75 2516. 26 McBee Meiner and Roberts Proc. Indiana Acad. Sci. 1954 64 112; Chem. Abs. 27 McBee Roberts and Meiner J. Amer. Chem. Soc. 1957 79 335. 28 Miller Bergman and Fainberg J. Arner. Chem. Soc. 1957 79 4159. 1956 50 5546h. Miller and Bcrgrnan Abstracts of Papers 126t h Meeting of the American Chemical Haszeldine and Walachewski J.1953 3607. Society 1954 p. 3 5 ~ . LAGOWSKI PERFLUOROALKYL DERIVATIVES 24 1 manner. Zinc alkyls form complexes of the type MZnR (M = Na Li; R = C,H5) but no complexes have been described with amines or ethers. On the other hand zinc halides form complexes with these Thus the zinc atom in perfluoroalkylzinc compounds acquires some of the properties of that atom in zinc halides because of the presence of the perfluoroalkyl groups. Attempts to convert perfluoroalkylzinc iodide into bisperfluoroalkylzinc by heat led to decomposition. The major product is hexafluoropropene which presumably arises from the reaction C,F,Znl -f ZnlF + C,F small amounts of bisheptafluoro-n-propylzinc are also f ~ r r n e d .~ ~ ? ~ There is no indication that the equilibrium 2C3F,Znl + (C,F,),Zn + Znl exists in dioxan solution. The yield of heptafluoro-n-propylzinc iodide is reported to be temperature-sensitive,30 but other investigations have shown that dioxan solutions of this compound are stable at reflux temperatures for long periods.28 The chemical reactivity of heptafluoro-n-propylzinc iodide apparently lies between that of the very reactive perfluoroalkylmagnesium halides and the relatively unreactive perfluoroalkylmercury compounds. Neither heptafluoro-n-propylzinc iodide in dioxan solution nor its dioxan adduct reacts with carbonyl compounds but aldehydes ketones and acid anhydrides containing enolisable hydrogen atoms liberate heptafluoro-n- propane with the formation of condensation products.28 Acid chlorides and perfluoro-acid chlorides yield the expected k e t ~ n e .~ ~ ? ~ ~ Attempts to carry out typical organometallic syntheses in dioxan solution were unsuccessfu1.28~29 Oxygen had little effect on heptafluoro-n-propylzinc iodide over prolonged periods and water decomposed it only slowly. Mercury. Perfluoroalkylmercuric iodides can be prepared by the action of perfluoroalkyl iodides on mercury under the influence of ultraviolet radiati~n,,~-~~ and the bisperfluoroalkylmercury compounds are best prepared by the reaction of perfluoroalkyl iodides with cadmium Bisheptafluoro-n-propylmercury could not be prepared by this method;36 only an oil consisting of a mixture of fluorocarbons and a substance with a mercurial odour was obtained (cf. p. 247). Fluoro-olefins either partially or fully fluorinated react with mercuric fluoride to form fluoroalkyl- or perfluoroalkyl-mercury derivatives ; bispentafluoroethyl- mercury can be prepared in good yields from tetrafluoroethylene in this 31 Gmelin’s “ Handbuch der Anorganischer Chemie Zinc” Verlag Chemie G.m.b.H.Weinheim 1952 pp. 875 877. 32 Miller Fainberg and Bergman Abstracts of Papers 122nd Meeting of the American Chemical Society 1952 p. 1 4 ~ . 34 Emelkus and Haszeldine J. 1949 2953. 36 Emelkus and Lagowski J. 1959 1497. 38 Lagowski Ph.D. Diss. Cambridge 1949. Banus Emelkus and Haszeldine J. 1950 3041. 242 QUARTERLY REVlEWS manner.37 Conversion of perfluoroalkylmercuric iodides into other perfluoroalkylmercuric salts can be effected by treatment with moist silver oxide and subsequent neutralisation of the resulting perfluoroalkyl- mercuric hydroxide with an appropriate acid.Bisperfluoroalkylmercury derivatives and perfluoroalkylmercuric halides react with excess of halogen to give quantitative yields of per- fluoroalkyl halides and presumably mercuric halides ; these reactions are similar to those of dialkylmercury compounds. There is no indication that a reaction occurs between bispentafluoroethylmercury and cyanogen under the influence of ultraviolet radiation except that the latter is polymerised to para~yanogen.~~ This is to be contrasted with the formation of phenyl- mercuric thiocyanate from diphenylmercury and thiocyanogen in ether solution.39 Bistrifluoromethylmercury reacts with mercuric halides to yield trifluoromethylmercuric halides ; the reaction occurs readily at the melting point of the mercurial but in the presence of a solvent e.g.acetone trifluoromethane is liberated. The reaction of bisperfluoroal kylmercury derivatives with metal or metalloid halides does not yield the perfluoroalkyl derivatives expected by analogy with the corresponding alkylmercury compounds. Only de- composition products arising from the mercurial i.e. mercury silicon tetrafluoride and carbon dioxide (the last two presumably arise from the decomposition of trifluoromethyl radicals on the walls of the reaction vessel) were obtained on reaction of trifluoromethylmercury derivatives with zinc magnesium amalgams of either of these aluminium halides,40 or the halides of arsenic and antimony.36 The principal reaction of trifluoromethylrnercury derivatives at high tempera- tures appears to be their decomposition but no fluorocarbons are formed.34 In contrast to these results hydrocarbons are formed when dialkyl- mercury compounds are pyrolysed or irradiated.The photochemical or pyrolytic decomposition products of bistrifluoromethylmercury have been reported to initiate the polymerisation of ethylene tetrafluoroethylene and other unsaturated compounds but no details were given.41 The higher perfluoroalkylmercurials however appear to undergo the expected radical reactions. Bispentafluoroethylmercury when irradiated in a silica tube decomposes to give free mercury and perfluoro-n-b~tane.~~~~~ Small amounts of carbon dioxide silicon tetrafluoride and hexafluoro- ethane were formed also; the last two compounds were formed in equiva- Ient amounts as would be expected from the disproportionation of penta- flu oroe t hyl radicals.Possibly the most intriguing property of bistrifluoromethylmercury is its solubility in water (437 g./l. i.e. 1 . 3 ~ 1 ) ; ~ ~ aqueous solutions give no 37 Krespan U.S. P. 2,844,614. 38 Banus Ph.D. Diss. Cambridge 1949. 39 Soderbach Annalen 1919 419 266. 40 Pugh unpublished results University of Cambridge. 41 Haszeldine. J.. 1949. 2856. LAGOWSKI PERFLUOROALKYL DERIVATIVES 243 indication of the presence of mercuric ions and the mercurial can be recovered unchanged. On the other hand dimethylmercury is insoluble in water as would be expected for a covalent compound.The solubility of trifluoromethylmercuric iodide (61 -4 g./l. i.e. 0-16~) although lower than that of bistrifluoromethylmercury is greater than that of methyl- mercuric iodide (1 x l o - ~ ) .~ ~ Aqueous solutions of trifluoromethyl- mercuric salts conduct a current but the slight conductivity of aqueous solutions of bistrifluoromethylmercury is wholly ~nexpected.,~ If the relative electronegativity of a trifluoromethyl group is 3.3 the electronic environment of the mercury atom in bistrifluoromethylmercury is similar to that in mercuric chloride and the properties of these two com- pounds should be similar. Ionic mercuric salts are extensively hydrolysed in aqueous solution but the more covalent compounds e.g. mercuric halides undergo hydrolysis to a smaller extent. The evidence available indicates that the conductivity of bistrifluoromethylmercury can be attributed to hydrolysis without the displacement of a trifluoromethyl group :36 Hg(CF,) + H20 + Hg(CF,),OH + Hg(CF,),OH- + H+ In analogy with the mercuric halides trifluoromethylmercurials can form addition compounds with halide Conductometric titrations of aqueous solutions of bistrifluoromethylmercury trifluoromethyl- mercuric iodide or heptafluoro-n-propylmercuric iodide with potassium halides indicate the existence of ionic species of the types Hg(CF,),X- Hg(CF,),X,2- Hg(Rf)IX- and Hg(Rf)IX,2- (X = C1 Br I; Rf = CF, C3F7).The complex ions become less stable in passing from the iodo- to the chloro-type. Anionic complexes of mercury-containing organic radicals other than cyanide have not been reported although compounds of the type LiM(C6H5) have been characterised for the other elements of Group 1 1 . ~ The relative stabilities decrease with increasing atomic weight of the central atom; no compound is formed with diphenylmer~ury.~~ The complex anions containing two moles of iodide per mole of perfluoroalkylmercury derivative Hg(CF3)13,- Hg(CF,),I :- and Hg(C3F7)Is2- were isolated by precipitation from aqueous solution with salts of ethylenediamine-transition metal complexes e.g.Cu(en),2+ Ni(en),U- Cd(en),2+ (en = eth~lenediamine).,~ Group In.-Attempts to prepare peduoroalkyl derivatives of this group of elements have been unsuccessful and there are indications that these compounds may be unstable. Only indirect evidence has been obtained for their existence. Boron. The synthetic methods used to prepare boron alkyls have not been successfully extended to the preparation of the analogous perfluoro- alkyl compounds.Bisperfluoroalkylmercury reacts readily with either 42 Waugh Walton and Laswick J. Phys. Chem. 1955,59 395. 43 Emelkus and Lagowski Proc. Chem. Soc. 1958 231. 44 Wittig and Hornberger AnnaZen 1952 557 11. 244 QUARTERLY REVIEWS boron halides or alkylboron halides to yield boron trifluoride or its derivatives as the major volatile products ; under certain conditions a very small amount of a substance which has the properties expected of a compound containing a B-Rf bond is formed.45 The low-temperature reaction of heptafluoro-n-propyl-lithium with boron chloride or bromide leads to extensive decomposition with the formation of tetrafluoroborate ions.45 In contrast the reactions of dialkylmercury compounds or reactive metallic alkyls with boron halides give good yields of the corresponding t r i a l k y l b o r ~ n .~ ~ ~ ~ ~ Although diborane reacts with non-fluorinated olefins to form a tri- alkylbor~n,~~ yet with tetrafluoroethylene trifluoroethylene 1,l-difluoro- ethylene and vinyl fluoride it yields boron trifluoride substituted boron fluorides and triethylb~ron.~~ The distribution of products is in the order BF > C2H;BF2 > (C,H,),BF > (C2H5),B with tetrafluoroethylene but the order changes as the fluorine content of the olefin decreases becoming C2H;BF2 > (C2H5)2BF > BF - (C2&)$ in the case of vinyl fluoride. Only one fluoroalkylboron compound F2BCH2F has been described,,O and its physical and chemical properties are of interest since they would be those with suitable modifications of compounds containing a B-CF bond.Fluoromethylboron difluoride is a liquid between -47" and +7" has a value of 30.5 for Trouton's constant and has an observed molecular weight 3.5 % higher than that calculated suggesting that association occurs in the vapour and the liquid phase. Spectral data support this conclusion and indicate that association occurs through the fluorine atom substituted on the alkyl group (structure I). The fact that neither methylboron difluoride nor trimethylboron is associated also supports this suggestion. Liquid fluoromethylboron difluoride attacks glass forming silicon tetra- fluoride and boron trifluoride; the formation of the latter can be easily visualised from structure (I). By analogy the isolation of boron trifluoride or its derivatives from the reaction of boron halides with bisperfluoro- alkylmercury suggests that an unstable pedluoroalkylboron compound is formed as an intermediate.Recent data indicate that stable compounds containing B-Rf bonds can be prepared by using reactive perfluoroalykl- metal compounds if the acceptor properties of boron are eliminated or 46 EmelCus and Lagowski unpublished results. 46 Michaelis Annalen 1901 315 19. 47 Krause and Nitsche Ber. 1921 54 2784. 48 Hurd J. Amer. Chern. SOC. 1948,70,2053. 49 Bastocha Graham and Stone J . Inorg. Nuclear Chern. 1958 6 119. Goubeau and Rohwedder Annalen 1957,604 168. LAGOWSKI PERFLUOROALKYL DERIVATIVES 245 substantially reduced e.g. by employing stable boron halide' addition compounds or B-halogenoborazole derivative^.^^ Aluminium. Perfluoroalkylaluminium derivatives have not been isolated but the existence of a perfluoroalkylaluminium complex in ether is indicated by studies of the reaction of lithium aluminium hydride with perfluoroalkyl iodides.5a The products stoicheiometry and hydrolytic behaviour of ether solutions observed in the stepwise reaction can be interpreted by assuming the formation of complexes of the type shown in equations (5)-(7) (Rf = C,F,) Rfl + LiAIH -+ LiAIRfH,I + H .. . . . . (5) Rfl + LiAIRfHJ -+ LiAIRfHI + RfH . . . . . . (4) Rfl + LiAIRfHI -f LiAI(Rf),l + HI . . . . . . . (7) Equation (5) represents the stoicheiometry observed for trifluoroiodo- methane but solutions containing the complex LjAl(CF,)H I decomposed to a mixture of trifluoromethane and methane on the addition of water. In contrast solutions of LiAI(C3F7)H12 yielded heptafluoropropane quantitatively under the same conditions.Attempts to isolate the complex LiAl(C3F7)H21 led to extensive decomposition forming aluminium lithium iodide hydrogen and probably heptafluoropropane. Group 1V.-With the exception of carbon more is known about the fluoroalkyl derivatives of silicon than of any other element in this group. Attempts to prepare trifluoromethyl derivatives of germanium by reaction of sodium germanyl derivatives with trifluoroiodomethane were un- successful ; trifluoromethane was the only volatile product Trimethyl(trifluoromethy1)tin has recently been obtained by reaction of trifluoroiodomethane and he~amethylditin.~~ Silanes substituted with fluoroalkyl groups can be prepared by the interaction of (a) a reactive fluoroalkyl-metal compound Ce.g.LiRf RfMgI) and silicon tetrahalides (b) a fluoroalkyl halide and elemental silicon Cc) silanes and unsaturated compounds containing fluorine or (d) fluoroalkyl iodides or bromides and an alkene-silicon It should be noted that fully fluorinated alkylsilanes cannot be prepared by methods c and d and that all four methods can be used in preparing partially fluorinated silanes. The use of a perfluoroalkyl Grignard or lithium reagent for the preparation of perfluoroalkyl derivatives of silicon has been described in only one instance,ll although its use has been reported without experimental details.13J7 Partially fluorinated alkylsilanes have also been prepared by this m e t h ~ d . ~ ~ ~ ~ ~ Perfluoroalkyl halides when passed 51 Lagowski and Thompson Proc. Chem Soc. 1959,301.52 Haupschein Suggeomo and Stokes J. Amer. Chem. SOC. 1956,78 680. 53 Cullen Ph.D. Diss. Cambridge 1958. 54 Clark and Cullen personal communication. 55 Geyer Haszeldine Leedham and Marklow J. 1957,4472. s6 Pierce McBee and Cline J. Amer. Chem. Soc. 1953,75 5618. 57 McBee Roberts Judd and Chao J. Amer. Chem. Soc. 1955,77 1292. 246 QUARTERLY REVIEWS over a heated mixture of silicon and copper metal yield bisperfluoroalkyl- silicon dihalides (eqn. 8; X = Cl Br I; Rf ’= CF, C2F5 C,F,);58 by altering the conditions trifluoro(trifluoromethy1)silane and bromodifluoro- (trifluoromethy1)silane were obtained. 2RfX + Si -+ (Rf),SiX . . . . . . . . . (8) Method c has been the most fruitful for the preparation of partially fluorinated alkylsilanes. The reaction of a silane with a fluoro-olefin under the influence of ultraviolet radiation or in the presence of organic peroxides yields partially fluorinated alkylsilanes (eqn.9). Partially or fully fluorinated olefins as well as halogeno- and alkyl-silanes have been employed in this r e a c t i ~ n . ~ ~ > ~ ~ - ~ ~ In the reactions of tetrafluoroethylene with silanes the length of the partially fluorinated alkyl chains attached to the silicon atom is determined by the ratio of the reactants. If the silane contains two hydrogen atoms the products of the reaction with tetrafl uoroethylene include substances which can be represented as H- [CF,-CF2] ,.SiR2. [CF,CF,-] y-H arising from secondary reactions. 6o The Si-H bond in silanes containing highly fluorinated alkyl groups (e.g. R,SiH-[CF,-CF,];H) is not attacked by water but these com- pounds are immediately cleaved by bases with the formation of hydrogen and the partially fluorinated alkanes He [CF,-CF2] n*H.60 This behaviour parallels that of silanes substituted with highly electronegative groups.6fi Fluoroalkylsilicon trihalides hydrolyse to yield polysiloxanes of the type (RrSiOl. 5)12 ; the rate of hydrolysis decreases as the length of the fluoroalkyl chain increases. 61 Alkyl(fluoroalky1)silicon dihalides exemplified by CHF,CF2-Si(CH,)C1, undergo hydrolysis to form fluorine-containing silicones [ -Si.(CH,).(CF,CF,H).O -3 n which are high-boiling oils stable to water but hydrolysed by base yielding fluorocarbons of the type HCF,-CF,H. Group V.-The reaction of a Group V element (M = P As Sbj with CF,I under pressure produces a mixture containing (CF,),M 68 Passino and Rubin U.S.P. 2,686,194; Chem. Abs. 1955 49 1363h. 69 Simons and Dunlap U.S. P. 2,651,651; Chem. Abs. 1954 48 10056a. Geyer and Haszeldine J. 1957 1038. 61 Haszeldine and Marklow J. 1956 962. 62 Hazeldine and Marklow Nature 1956 178 808. 63 McBee Roberts and Puerckhauere J . Amer. Chem. SOC. 1957 79 2329. 64 Geyer and Haszeldine J. 1957 3925. 65 Frost US. P. 2,596,967; Chem. Abs. 1953 47 4365h. 66 Wagner U.S. P. 2,637,738; Chem. A h . 1954,48 82546. 67 Tarrant Dyckes Dunmire and Butler J. Amer. Chem. SOC. 1957 79 6536. 68 Hurd “Chemistry of the Hydrides” John Wiley and Sons New York 1953 p. 58- LAGOWSKI PERFLUOROALKYL DERIVATIVES 247 (CF,),MI CF,-MI, and iodides of the element.69-72 Reaction occurs in the same temperature range 200-220° for phosphorus and arsenic but antimony reacts at a substantially lower temperature 165-175 O.The lower limit in the optimum range gives a larger percentage of the iodo- compound and the addition of metal tri-iodides also increases the propor- tion of these compounds. There are indications that a series of equilibria (10-12) is established during the reaction . . . . . . . . . 2(CF3),MI + (CF,),M + CF,-MI (1 0) Z(CF,)MI + (CF,),MI + MI (11) (CF,),M + MI + (CF,),MI CF,.MI, CFJ etc. . . . . . (12) . . . . . . . . . . If the temperature of the reaction is allowed to rise above the optimum decomposition occurs producing fluorocarbons as well as the Group V metal fluorides. Some anomalies have been observed in the preparation of higher perfluoroalkyl-phosphorus and -arsenic compounds although the preparation of (CzF,)3A~ by this method appears to have been succe~sfu1.~~ The interaction of C3F,I and phosphorus produced only two iodohepta- fluoro-n-propylphosphorus compounds and an attempt to prepare heptafluoro-n-propylarsenic compounds gave only fluorocarbon^.^^ The molecular structures of compounds of the type (CF3)3M (M = P As Sb) have been determined by electron-diffraction methods,75 and the data indicate that the CMC angles are normal but that C-M bonds are systematically longer for the trifluoromethyl compounds than for the corresponding methyl derivatives.Phosphorus. The equilibrium mixture of tristrifluoromethylphosphine di-iodotrifluoromethylphosphine and iodobistrifluoromethylphosphine obtained by reaction of trifluoroiodomethane with phosphorus is the starting point for all other trifluoromethylphosphorus compounds which have been prepared.Iodoperfluoroalkylphosphorus compounds can be converted into the corresponding cyano- or chloro- (and presumably bromo-) derivatives by using the appropriate silver salt (eqns. 13 14; Rf = CF3,69 C3F77*) ; fluorobistrifluoromethylphosphine has been (Rf),Pi + AgX -f (Rf),PX + Agl RfPI + 2AgX -f RfPX + 2Agl prepared by the action of antimony trifluoride on the corresponding iodo- Phosphorus acids containing perfluoroalkyl groups can be prepared by the hydrolysis of various halogenoperfluoroalkylphosphorus compounds 69 Bennett EmelCus and Haszeldine J. 1953 1565. 70 Brandt EmelCus and Haszeldine J. 1952 2552. 71 Dale Emelkus Haszeldine and Moss J. 1957 3708. 72 Burg Mahler Bilbo Haber and Herring J .Amer. Chem. Soc. 1957,79,247. 73 Ayscough and EmelCus J. 1954 3381. '* Emelhs and Smith J. 1959 375. 76 Bowen Trans. Faraday SOC. 1954 50 463. 78 Burg and Brendel J. Amer. Chem. SOC. 1958,80 3198. . . . . . . . (1 3) (1 4) . . . . . . . 248 Acid pK pK2 Ref. - CF,*PO(OH) 1-16 3-95 ‘77 (CF,),PO(OH) 1.0 - 6 (C,F,)PO(OH) 0-9 3.96 74 QUARTERLY REVIEWS Acid pK pK2 Ref. - Cl,C.PO(OH) 1.63 4.81 Me,PO(OH) 3.08 - a H3P03 1-41 6.7 (Cf&PC13 (CFaC12 (CF3l2PX (O3I3P \ H 2 0 1 - T = “‘q (CF,)2P(ONa) -2CF,H CF3 PX (x = Cl 1) P *Corresponding CBF7 acids FIG. 2 FIydroIysis of perfluoroalI~y(ph0sphorus compounds to form phosphorus acids containing per-iroroalkyl groups.6*77~78 (Fig. 2). The fact that these acids are all stronger than their alkyl counter- parts can be attributed to the strong inductive effect of the perfluoroalkyl groups present (Table 7).As would be expected trifluoromethylphos- phonic acid is a stronger acid than trichloromethylphosphonic acid. The loss of one equivalent of trifluoromethane during the neutral aqueous hydrolysis of halogenobistrifluoromethylphosphines indicates the instability of bistrifluoromethylphosphinous acid UI) and the latter has not been well characteri~ed.~~$~~ This behaviour is in agreement with the relative instability of the dialkylphosphinous acids (eqn. 15) ; however they usually yield the corresponding phosphine on hydrolysis. 2R,P*OH 4 R,PH + R,PO-OH . . . . . . . . (15) “ Bennett Emelkus and Haszeldine J. 1954 3598. Bennett Emelkus and Haszeldine J. 1954 3986. LAGOWSKI PERFLUOROALKYL DERIVATIVES 249 Trifluoromethylphosphonous acid (111) has not been isolated as the free acid since it is volatile in steam at reduced pressure;77 it behaves as a monobasic acid (pK = l.Ol) liberating trifluoromethane at pH > 11.Infrared studies of the mono-potassium and -sodium salts of (111) suggest that the structure is best represented as the keto-form (VI) rather than R ?H (VI) CF3-T-O- No' H CF,-P-O- Na+ (VII) as the enol form (VII). Aqueous solutions of the free acid are most con- veniently prepared by distillation of the alkali-metal salts with dilute sulphuric acid. Potassium iodate dichromate and permanganate and mercuric chloride (at 60") are slowly reduced by aqueous solutions of (III) but molecular iodine resists reduction.6 Aqueous solutions evolve tri- fluoromethane and form phosphorous acid at 140"; a side reaction (8% of the total reaction) which is dependent on the acid concentration produces trifluorome thylphosphonic acid (V) and trifluoromethyl- phosphine.This reaction is similar to the decomposition of alkylphos- phonous acids (eqn. 16) and even phosphorous acid is known to undergo a similar disprsportionntion (eqn. 17). 3RP(OH) + RPH + ZRPO-(OH) . . . . (1 6 ) 4H,PO -f 3H,PO + PH . . . . . . . (1 7) Bistrifluoromethylphosphinic acid (IV) is also steam-volatile and aqueous solutions of it have been prepared in a manner analogous to that used for the preparation of trifluoromethylphosphonous acid. The pure acid can be distilled from the silver salt by heating the latter with con- centrated sulphuric acid. It is a strong monobasic acid in contrast to the very weak dimethylphosphinic acid and liberates trifluoromethane at pH > 8.7.Trifl uoromethylphosphonic acid (V) the most stable of the trifluoro- methylphosphorus acids which have been studied is dibasic and gives a sparingly soluble barium Infrared absorption studies on the mono- sodium and -potassium salts indicate the presence of strong hydrogen bonds which could be either inter- (VIII) or intra-molecular (IX) in nature. Heptafluoro-n-propylphosphonic acid has been prepared by a similar method and according to the value of its first ionisation constant it appears to be slightly stronger than trifluoromethylphosphonic acid;74 however the reverse effect is observed in the acidity of primary and secondary perfluoroalkyl alcohols and has been attributed to the replace- ment of a fluorine atom with a less electronegative perfluoroalkyl group.79 7B Haszeldine J.1953 1257. 250 QUARTERLY REVIEWS The precision with which ionisation constants of acids can be determined decreases as the acid strengths increase and this could account for the apparent anomaly in the values observed for the first ionisation constant. On the other hand the values of pK (given to a higher degree of precision) for trifluoromethyl- and heptafluoro-n-propyl-phosphonic acids indicate that the latter is slightly weaker than the former which is in agreement with the data for the ionisation of primary and secondary alcohols. Bistrifluoromethylphosphinic acid has been shown to be stronger than trifluoromethylphosphonic acid by conductivity measurements in anhydrous acetic acid.6 This is the order which would be expected from a consideration of the inductive effect of a second trifluoromethyl group.The slight increase in acid strength in going from trifluoromethylphos- phonic acid to trifluoromethylphosphonous acid parallels the increased acid strength of phosphorous acid compared with phosphoric acid. Bistrifluorornethylph~sphine~~~~~ and trifl~oromethylphosphine~~ can be prepared by the reduction of the corresponding iodo-compounds with zinc and hydrochloric acid by hydrogenation in the presence of Raney nickel by lithium aluminium hydride or by mercury and an- hydrous hydrogen chloride. Trifluoromethylphosphine has also been obtained by the decomposition of concentrated aqueous solutions of trifluoromethylphosphonous acid;78 bistrifluoromethylphosphine is pre- pared by acid hydrolysis of tetrakistrifluoromethyldiphosphine.78~80 The trifluoromethylpolyphosphines (CF,.PH) and H,(PCF,) have been obtained on hydrolysis of tetrakistrifluoromethyltetraphosphine and pentakistrifluoromethylpentaphosphine respectively.81 Although no study of the co-ordinating ability of the trifluoromethylmonophosphines has been reported there are indications that they do interact with transi- tion metals.Reduction of di-iodotrifluoromethylphosphine with hydrogen in the presence of Raney nickel failed to yield the expected phosphine (or indeed any unchanged starting material) although this method has been used to prepare bistrifluoromethylphosphine.78 The suggestion has been made that compound formation with the nickel could explain the failure of the reduction; the difficulty of removing bistrifluoromethyl- phosphine from the reduction catalyst has also been attributed to com- pound formation.78 This behaviour is consistent with the observation that a trifluoromethyl group has an electronegativity greater than that of chlorine but less than that of fluorine.The trifluoromethylphosphines should behave in a manner analogous to phosphorus trichloride or trifluoride ; both of these phosphorus halides form addition compounds with transition elements i.e. palladium,82 p l a t i n ~ r n ~ ~ ~ ~ ~ and nickel 8 5 3 8 6 (see p. 254). Burg and Mahler J. Amer. Chem. SOC. 1957;79 4242. Mahler and Burg J. Amer. Chem. SOC. 1958 80 6161. 82 Fink Compt. rend. 1692 115 176. 83 Schutzenberger Compt. rend. 1870,70 1287 1414. 84 Chatt and Williams J. 1951 3061.85 Irvine and Wilkinson Science 1951 113 742. 86 Wilkinson J. Amer. Chern. Soc. 1951,73 559. LAGOWSKI PERFLUOROALKYL DERIVATIVES 25 1 The reaction of iodotrifluoromethylphosphine or iodobistrifluoro- methylphosphine with mercury gives rise to a series of compounds con- taining P-P bonds (eqns. 18 1 9 ) . 6 9 y 8 1 9 8 7 Higher polymers are formed in . . . . . . . . . . . (CF&PI -+ (CF3)2P-P(CF3) ' (18) (CF,)PI -+ (CF,P) + (CF,P) (1 9) . . . . . . . . . . . . (CF,),P-P(CF& -+ (CF3P)a + (CFSP) + (CFaP), . . . . . . . . (20) addition to the cyclic polymers (CF,P) and (CF,P) when either tetrakis- trifluoroniethyldiphosphine or bistrifluoromethylphosphine is pyrolysed (eqn. 20).81787 Pentakistrifluoromethylpentaphosphine which is thermally less stable than the corresponding tetramer decomposes into the tetramer in addition to tetrakistrifluoromethyldiphosphine and tristrifluoro- rnethylphosphine on heating which suggests that reaction (20) is reversible to some extent.The ultraviolet spectra of the cyclic polymers indicate that the phosphorus lone-pair electrons are delocalised and are supplementing the normal o-bonding. Basic hydrolysis of most trifluoromethylphosphorus compounds yields trifluoromethane q~antitatively,'~ but those containing P-P bonds give a mixture of trifluoromethane and fluoride ions.78-81 The formation of fluoride ions in the hydrolysis of trifluoromethylpolyphosphines exempli- fied by tetrakistrifluor~methylphosphines,~~ can be explained if the initial step is cleavage of a P-P bond followed by decomposition of the products (eqn.21). It was shown independently that bistrifluoromethylphosphinous (CF,),P-Pp(CF3)2 + H,O -+ (CF,),PH + (CF&P*OH . . . (CF,),PH + H,O -+ CF,H + F- + CO acid gives only trifluoromethylphosphine and that bistrifluoromethyl- phosphine yields a mixture of trifluoromethane and fluoride ion on hydrolysis under similar condition^.^^ Basic hydrolysis of the trifluoro- methylphosphorus cyclic polymers can be visualised as an initial opening of the phosphorus ring followed by equal and .quantitative conversion of trifluoromethyl groups into trifluoromethylphosphine and trifluoromethyl- phosphinous acid which then decompose to fluoride ions and trifluoro- methane respectively.81 Basic hydrolysis of trifluoromethylpolyphosphine chains appears to occur in such a manner that the second group either H or OH substituted on a phosphorus atom is identical with the previously substituted group.Solutions of the tetramer in fluorocarbon solvents absorb gaseous oxygen in two distinct steps each corresponding to the addition of one oxygen atom per atom of phosphorus to yield a substance with the empirical formula (CF3.POz)z. The observed induction period suggests a chain reaction the first step of which is probably the formation of a peroxy-phosphorus compound. The oxidation product which appears to be 1 (21 1 (CF,),P*OH + H20 -+ 2CF,H J Mahler and Burg J. Amer. Chem. Soc. 1957,79 251. 252 QUARTERLY REVIEWS a mixture of polymeric anhydrides of trifluoromethylphosphonic acid is slowly converted into this acid on reaction with water; titration curves of incompletely hydrolysed aqueous solutions of the polymeric oxide suggest the presence of the strong acid CF,*P(0)(OH)-P(O)(OH)CF,.81 Polymers containing B-P bonds and trifluoromethyl groups have been prepared by the reaction of diborane with either fluorobistrifluoromethyl- phosphine or bistrifluor~methylphosphine.~~ The major product is the air-stable trimer [(CF,),PBH,] , but traces of the tetramer [(CF3)2PBH2] were also formed in the reaction with bistrifluoromethylphosphine.Dimethyl ether not only catalyses the reaction of the latter compound with diborane but it also attacks the phosphine to yield methylbistrifluoro- methylphosphine. There was no evidence for the existence of the borine adducts (CF,),P,BH or (CF,),PH,BH, but there were indications that a very uiistable (CF,),PF,BH adduct was formed.The phosphorus atom in (CF,),PF,BH has an electronic environment similar to that in the unstable adduct BH,,PF3ss if the electronegativity of a trifluoromethyl group is taken as 3.3. It may be significant that although phosphorus halides (the donor properties of the phosphorus atoms have been essentially removed) do not form adducts with boron halides (which should be better acceptors than borine) yet unstable addition compounds of borine are formed with both phosphorus trifluoride and fluorobistrifluoromethyl- phosphine. It has been suggested that the inertness of the hydrogen atoms in the polymer [(CH,),P*BH,] towards protonic reagents may be due to weak multiple n-bonding in which the B-H electrons enter the phosphorus dlevels thus lowering the electronic density about the hydrogen atoms.89>90 Electronic drift of this type should occur less easily when both boron and phosphorus carry highly electronegative groups.Halogenobisperfluorophosphines react with ammonia and primary or secondary amines to yield aminotrifluoromethylphosphines and hydrogen halides (eqn. 22).74-91392 Only one hydrogen atom in ammonia could be (Rf),PX + 2R2NH -+ (Rf),P-NR + R,NH,HX . . . . . (22) replaced by a -P(CF,) group which is consistent with the suggestion that reaction occurs via the intermediate (X) with subsequent elimination of hydrogen halide. 92 The electron-withdrawing effect of the trifluoromet iyl groups which facilitates the formation of (X) decreases the availability 88 Parry and Bissot J. Amer. Chern. SOC. 1956,78 1524. 89 Burg and Wagner J. Amer.Chem. Soc. 1953,75 3872. 92 Harris Proc. Chem. Soc. 1957 118. Graham and Stone J . Inorg. Nuclear Chem. 1956,3 164. Harris J. 1958 512. LAGOWSKI PEWLUOROALKYL DERIVATIVES 253 of the lone-pair electrons on the nitrogen atom in (CF3)2P.NH2 for dona- tion to another molecule of (CF3),PX. These factors should also make the hydrogen atoms in aminobistrifluoromethylphosphine more acidic than those in the methyl analogue. The loss of basic nitrogen properties is exemplified by the fact that amiiiobistrifluoromethylphosphine does not form a quaternary salt with methyl iodide; attempts to prepare the silver and sodium salts were unsuccessful although pyridine solutions of aminobistrifluoromethylphosphine do conduct a current more readily than the pure solvent. The high values of Trouton’s constants for arninobis- trifluorome thylphosphine and met hylamino bis trifluor ome thylphos- phine suggest that association occurs in the liquid phase possibly as hydrogen bonds -F * * H-N-.92 The aminoperfluoroalkylp hosphines are unstable to hydrolytic agents as well as to hydrogen chloride and chlorine; fission of the P-N bond occurs in each case. Tristrifluoromethylphosphine is an inflammable liquid which is stable to water and aqueous hydrochloric acid but it reacts with aqueous alkali yielding trifluoromethane by a series of reactions involving bistrifluoro- methylphosphinous acid (11) and trifluoromethylphosphonous acid (111) as intermediates (see Fig. 2).78 A mixture of iodotrifluoromethylphosphines phosphorus iodides and trifluoroiodoinethane is formed when tris- trifluoromethylphosphine reacts with iodine but tristrifluoromethyl- phosphine readily adds chlorine to form dichlorotristrifluoromethyl- phosphorane which in turn yields trifluoromethylphosphonic acids on hydrolysis.Tristrifluoromethylphosphine oxide was not obtained by atmospheric oxidation of the phosphine but was successfully prepared by using anhydrous oxalic acid.93 In contrast to the alkyl- or aryl-phosphine oxides which form stable hydrates tristrifluoromethylphosphine oxide and dichlorotristrifluoromethylphosphorane react with water to form bistrifluoromethylphosphinic acid and trifluoromethane.6 The electrolytic behaviour of dichlorotristrifluoromethylphosphorane in acetonitrile is similar to that of phosphorus pentachloride. Acetonitrile solutions of dichlorotristrifluoromethylphosphorane exhibit a greater molar conductance at infinite dilution than solutions of either phosphorus trichloride or tribromide.94 It has been suggested that this conductance behaviour is due to an equilibrium (23) which is analogous to that found 2P(CF,),CI + P(CF,),CI+ + P(CF,),CI,- . . . . . . . . (23) in similar solutions of other phosphorus pentahalides. 95*96 Trichlorobis- trifluoromethylphosphorane is a non-conductor under these conditions and infrared data suggest that this compound is a trigonal bipyramid with trifluoromethyl groups at the apices. If the mechanism of ionization involves the apical bond trichlorobistrifluoromethylphosphorane would O3 Paul J. 1955 574. O6 Harris and Payne J. 1956,4617. EmelCus and Harris J. 1959 1494. Payne J. 1953 1052. 254 QUARTERLY REVIEWS not be expected to conduct since the trifluorornethide ion has not been observed in any system.Mixed methyltrifluoromethylphosphines can be prepared by the action of methyl iodide on tristrifluoromethylphosphine at 240" ; dimethyltrifluoromethylphosphine reacts rapidly with additional methyl iodide to form the quaternary phosphonium iodide.97 The reaction is reversible to some extent since trimethylphosphine reacts with trifluoro- iodomethane to form dimethyltrifluoromethylphosphine but neither the latter compound nor methylbistrifluor ome thylphosp hine reacts further with trifluoroiodomethane. The reaction with trifluoroiodomethane has been considered to involve the formation of an unstable phosphonium salt of a trifluoromethide ion which then rapidly rearranges to a more stable phosphonium salt The mixed methyl trifluorome t hylp ho sp hi nes yield trifluoromethane quantitatively on basic hydrolysis and the rate of hydrolysis increases as the number of trifluoromethyl groups attached to the phosphorus atom increases.This is consistent with the hypothesis that the hydrolysis proceeds via a nucleophilic attack on the phosphorus atom since the basicity of the phosphine should decrease with an increase in the number of trifluoromethyl groups present. Apparently the introduction of two or more trifluoromethyl groups into the phosphine is sufficient to prevent quaternary salt formation. No reaction was observed between methyl iodide and either tristrifluoromethylphosphine or methylbistrifluoro- methylphosphine but dimethyltrifluoromethylphosphine reacted to form trimethyltrifluoromethylphosphonium iodide.The tervalent phosphorus compounds PX3 (X is any group) can form co-ordination compounds with electron-acceptor molecules which have filled or partially filled d orbitals. If the d orbitals of tbe acceptor are filled the stability of the compound is governed mainly by the basicity of the tervalent phosphorus compound. When the phosphorus atom carries strongly electronegative groups as for example in phosphorus trifluoride or trichloride either no or very unstable addition compounds are formed with strong electron-acceptors like boron t r i f l u ~ r i d e . * ~ ~ ~ * - ~ ~ ~ If quaternary salt formation is taken as a rough guide to the basicity of the phosphorus atoms in different trifluoromethyl compounds only dimethyltrifluoro- methylphosphine will form complexes with electron acceptors which have no electrons in their d orbitals.This suggestion is supported by the isolation of the boron trifluoride addition compound of dimethyltrifluoromethyl- s7 Haszeldine and West J. 1956 3631. 98 Chatt Nature 1950 165 637. ) @ Chatt J. 1949 3340. loQ Baumgarten and Brune Ber. 1947 80 517. LAGOWSKI PERFLUOROALKYL DERIVATIVES 255 phosphine (m.p. -9") and the failure to isolate similar compounds of methylbistrifluoromethylphosphine or tristrifluoromethylphosphine.lOl If the acceptor has available d electrons the bond formed has two components the normal (T bond formed by donation of the lone-pair electrons and the so-called 7~ component formed by back-donation of the acceptor's d electrons to vacant d orbitals of the phosphorus atom.g8 As the electronegativity of the groups attached to phosphorus increases the availability of the phosphorus lone pair of electrons for a-bond formation will decrease but this will be partially compensated by a strengthening of the v component.The stability of complexes of tervalent phosphorus compounds is not greatly affected by the electronegativity OF the substitu- ents on the phosphorus atom and phosphorus halides as well as alkyl- phosphines form stable addition compounds with transition metals. Thus tristrifluoromethylphosphine reacts with nickel carbonyl to form (CF,),P.Ni(CO) and [(CF,),P] zNi(CO),1°2J03 in a manner analogous to the replacement of carbon monoxide from nickel carbonyl by phos- phorus trihalide~.*~J~~ Direct reaction with phosphorus trifluoride or tristrifluoromethylphosphine does not replace all of the carbon monoxide from nickel carbonyl ; with tristrifluoromethylphosphine only two carbon monoxide molecules could be replaced.Trifluoromethylpolyphosphines also react with nickel carbonyl.lo2 The compound obtained from the reaction of tetrakistrifluoromethyl- diphosphine with nickel carbonyl has been formulated as (XI). The reaction with tetrakistrifluoromethyltetracyclophosphine yields a mixture containing compounds of the type (CF,),[Ni(CO),] (n = 1 2 3) and chains of the type (XII) with terminal Ni(CO) groups and Ni(CO) units cross-linking the chains through phosphorus. The formation of platinous chloride co-ordination compounds with dimethyltrifluoromethylphosphine and methylbistrifluoromethylphos- phinelo1 is consistent with the formation of similar compounds with phosphorus trichlorideS3 and alkylphosphine~.~~~J~~ The mixed methyl- trifluoromethylphosphines can thus be treated as mixed halogenomethyl- phosphines with respect to the electronic environment on the phosphorus atom.It is surprising that tristrifluoromethylphosphine did not form co- ordination compounds with platinous chloride under the same conditions lol Begg and Clark Abstracts of the International Conference on Co-ordination Chemistry London 1959 Abstract No. 62. lo2 Burg and Mahler J . Amer. Chem. Soc. 1958 80,2334. lo3 EmelCus and Smith J. 1958 527. lo4 Wilkinson J. Amer. Chem. SOC. 1951,73 5501. lo6 Cahours and Gal Compt. rend. 1870,70 1380 71 208. lo6 Klason and Wanselin J . prakt. Chem. 1903,67,41. 256 QUARIERL.Y REVTEWS since both phosphorus trichloride and trifluorides4 form stable compounds and since the electronegativity effect in tristrifluoromethylphosphine should be intermediate between those of phosphorus trifluoride and trichloride.Arsenic. The reactions and chemical properties of trifluoromethyl- arsenic compounds very closely parallel those of the corresponding phosphorus compounds. Halogenotrifluoromethylarsines can be pre- pared by the action of an appropriate silver salt on the iodotrifluoro- methylarsines or by the reaction of tristrifluoromethylarsine with ha10gens.l~~ In the latter reaction only chlorine gave a quinquevalent arsenic compound (CF,),AsCl, but by altering the reaction conditions bis trifluoromethylarsenic trichloride was also formed and could be thermally decomposed into chlorobistrifluoromethylarsine and dichloro- trifluoromethylarsine.Only fluorobistrifluoromethylarsine was obtained from the reaction of elemental fluorine with tristrifluoromethylarsine. Methylbistrifluoromethylarsine and trifluoromethylarsine can be prepared from the corresponding iodoarsines by the action of either lithium aluminium hydride or zinc and hydrochloric acid;lO* the latter reagents gave better yields in each case. Tetrakistrifluoromethyldiarsine was obtained admixed with bistrifluoromethylarsine by using a modification of the lithium aluminium hydride procedure described for the correspond- ing diphosphine. Mixed methyltrifluoromethylarsines can be prepared by the action of methylmagnesium iodide on the appropriate iodotrifluoromethyl- arsine,lo8 by the interaction of tristrifluoromethylarsine and methyl i ~ d i d e ~ ~ J ~ ~ J ~ * or by the reaction of trifluoroiodomethane with trirnethyl- ar~ine.~~JO~ Trifluoroiodomethane will replace two methyl groups step- wise from trimethylarsine to yield first dimethyltrifluoromethylarsine and then methylbistrifluoromethylarsine ; in contrast the analogous reaction with trimethylphosphine yields only dimethyltrifluoromethyl- phosphine.The co-ordinating ability of the mixed methyltrifluoromethyl- arsines is markedly influenced by the presence of trifluoromethyl groups. Dimethyltrifluoromethylarsine does not form addition compounds with mercuric halide~,lll-~~~ and no reaction occurs between tristrifluoro- methylarsine and nickel carb0ny1.l~~ These results are in accord with the facts that addition compounds of arsenic trihalides with transition metals apparently are not formed and that arsenic trichloride does not displace carbon monoxide from nickel carbonyl.lo4 From the data available it appears that the stability of the complexes formed by transition metals and lo' Walachewski Chem.Ber. 1953 86 272. lo* Emelbm Haszeldine and Walachewski J. 1953 1552. loo Haszeldine and West J. 1957 3880. 110 Emelkus Haszeldine and Paul J. 1954 881. ll1 Challenger Higgenbottom and Ellis J. 1933 35. Challenger and Ellis J. 1935 398. 113 Challenger and Rawlings J. 1936 264. 114 Smith Ph.D. Diss. Cambridge 1958. LAGOWSKI PERFLUOROALKYL DERIVATIVES 257 tervalent arsenic compounds unlike those formed with tervalent phos- phorus and antimony compounds is dependent on the relative electro- negativity of the substituents on arsenic.Dimethyltrifluorometlzylarsine forms a quaternary salt with methyl iodide indicating that the donor properties of the arsenic lone-pair electrons are not completely lost by the introduction of a trifluoro- methyl group. This quaternary compound appears to be less stable than the corresponding phosphoniuni iodide. 97 Bistrifluoromethylarsine or tristrifluoromethylarsine shows no tendency towards complex formation with mercuric chloridelos or towards quaternary-salt formation with either trifluoroiodomethane or methyl iodide although methyl iodide apparently forms a 1 1 azeotrope with tristrifluoromethylarsine.ll0 All halogenotrifluoromethylarsines yield trifluoromethane or a mixture of trifluoromethane and fluoride ions on alkaline hydrolysis.The latter products are formed in the hydrolysis of bistrifluoromethyl- arsine trifluoromethylarsine and tetrakistrifluoromethyldiarsine and this behaviour parallels that found in the coresponding phosphorus compounds. Cyanobistrifluoromethylarsine is the only arsine which is hydrolysed by water alone ; the products are trifluoromethane and hydrogen cyanide. The instability of the resulting bistrifluoromethylarsinous acid (CF,),As-OH which this behaviour implies parallels the instability of bis- trifluoromethylphosphinous acid although the latter appears to be more stable than the former. The silver salt of bistrifluoromethylarsinous acid can be isolated but attempts to liberate the acid in aqueous solution lead to decomposition with the formation of trifluoromethane.lo8 Bistrifluoromethylarsinic acid can be prepared by the oxidation of iodobistrifluoromethylarsine with aqueous hydrogen peroxide.l1° The solid acid has an empirical formula corresponding to (CF,),AsO.OH but in aqueous solution it behaves as a dibasic acid; two points of inflection occur in the titration curve corresponding to (CF,),As(OH),.ONa and (CF,) ,As(OH)(ONa),.The acid decomposes rapidly in solutions of pH > 7 liberating trifluoromethane. When heated in vacuo bistrifluoro- methylarsenic acid does not form an anhydride quantitatively but under- goes intramolecular dehydration. Trifluoromethylarsonic acid (XIII) can be prepared from di-iodotri- fluoromethylarsine by an extension of the techniques used to prepare the corresponding phosphorus acid.l1° The acid undergoes stepwise dehydra- tion to give the pyro-form (XIV) and the anhydride (XV).The ortho-form (XIII) is a dibasic acid and one hydrogen atom is almost completely (XI I I) OH (XIV) ionised in aqueous solutions (a = 0.96 at 0.01~). Conductivity measure- ments in anhydrous acetic acid indicate that bistrifluoromethylarsonic 4 258 QUARTERLY REVIEWS acid is stronger than trifluoromethylarsonic acid which is to be expected from the inductive effect of the trifluoromethyl group; both are stronger than their methyl analogues. Attempts to prepare tristrifluoromethyl arsine oxide the anhydride of the hypothetical acid (CF,),AS(OH)~ the acid itself or its ethyl ester were unsuccessful. The interactions of halogenobistrifluoromethylarsines with liquid ammonia follow essentially the same course as the corresponding phos- phines.lf5 Whereas only one (CF,),P.group can be substituted on the nitrogen atom in ammonia it is possible to replace two hydrogen atoms with (CF,),As. groups. Chlorobistrifluoromethylarsine reacts with gaseous ammonia to give (CF,),As-NH and [(CF,),As] ,NH but reaction in liquid ammonia yields only the latter. The corresponding compound in the water system [(CF,),Asl20 was prepared by the reaction of iodobistrifluoromethylarsine with mercuric oxide.lo7 The reaction of chlorobistrifluoromethylarsine with primary or secondary amines produces products containing only one (CF,),As* group attached to a nitrogen atom. There are no indications that tristrifluoromethylarsine forms addition compounds with amines although arsenic trichloride is known to form addition compounds with organic nitrogen b a ~ e s .~ l ~ J l ~ Kinetic data on the pyrolysis of trisperfluoroalkylarsines indicate that the primary reaction is the formation of perfluoroalkyl radicals which then combine to form fl~orocarbons.~~ Tristrifluoromethylarsine gives hexafluoroethane in 60-90 % yield in the temperature range 350-410"; in addition up to 10% of perfluoropropane and perfluorobutane are formed but no unsaturated fluorocarbons were detected. The remainder of the volatile products 20--30% consists of silicon tetrafluoride and carbon dioxide. If the pyrolyses are conducted in a platinum vessel the reaction is still of the first order but the rate constant increases by about SO% and the composition of the products is greatly changed. A larger proportion of higher fluorocarbons is detected and although silicon tetrafluoride and carbon dioxide are not formed the trifluoromethyl radicals appear to decompose to give fluorinated products as indicated by the formation of appreciable amounts of arsenic trifluoride.Increasing the surface area of the silica reaction tube by adding powdered silica not only increases the rate constant for the decomposition of tristrifluoro- methylarsine by 25 % but also increases the ratio of carbon dioxide plus silicon tetrafluoride to fluorocarbons tenfold. Although the mechanism for the formation of silicon tetrafluoride and carbon dioxide is uncertain these products probably arise from the attack of trifluoromethyl radicals on silica; such attack is indeed known to occur at lower temperatures than those used for the pyrolysis of tristrifluoromethylarsine.Trispenta- fluoroethylarsine decomposes to a mixture containing 92 % of fluoro- 116 Cullen and Emelkus J. 1959 372. Grossman 2. phys. Chem. 1906,57 545. 11' Dafert and Melinski Bet-. 1926 59 789. LAGOWSKI PERFLUOROALKYL DERIVATIVES 259 carbons and only 8% of silicon tetrafluoride and carbon dioxide. It thus appears that pentafluoroethyl radicals are more stable in the presence of silica than trifluoromethyl radicals. The first-order decomposition of tristrifluoromethylarsine has an activation energy of 57.5 k~al./mole,~~ and if this is assumed to be the dissociation energy of the C-As bond it appears that replacement of a hydrogen atom by a fluorine atom has little effect on the bond energy; the decomposition of trimethylarsine has an activation energy of 54.6 kcal./mole.A similar effect has been noted in a comparison of the dis- sociation energies of the C-H bond in methane and trifluoromethane;lls both bond energies were determined to be 102 kcal./mole. Trispentaethyl- arsine decomposes with an activation energy of 48.0 .k~al./mole.~~ Antimony. The halogen o trifl uoromethylstibines a1 t hough they do undergo many of the reactions of the corresponding phosphines and arsines are difficult to handle in the laboratory since they readily disproportionate to yield tristrifluoromethylstibine and antimony trihalides as the major As is the case with most of the trifluoromethyl compounds of the Group V elements the trifluoromethylstibiiies are quantitatively hydrolysed with aqueous base to yield trifluoromethane.In contrast with the hydrolysis of the corresponding arsenic and phosphorus compounds tetrakistrifluoromethyldistibine gives almost quantitative yields of trifluoromethane and only about 1-2% of fluoride ion. If the reactions postulated for the basic hydrolysis of tetrakistrifluoromethyldi-phos- phine and -arsine are correct the results for the distibine suggest that bistrifluoromethylstibine should yield all of its fluorine as trifluoro- methane. Of the halogens only chlorine gives stable quinquevalent com- pounds with tristrifluoromethyl-phosphine and -arsine but both chlorine and bromine react with tristrifluoromethylstibine to form compounds of the type (CF,),SbX,. Only one mixed alkylperfluoroalkylstibine Sb(CH3) ,CF, has been prepared. 97 The properties of tristrifluoromethylstibine appear to be more similar to those of the antimony trihalides than the alkylstibines.When similarities occur between the alkyl- and perfluoroalkyl-stibines the same properties can also be observed for antimony trihalides. This is consistent with the value calculated for the electronegativity of the trifluoromethyl group. Antimony trichloride and trifluoride have very slight donor properties but they can act as acceptors as exemplified by the formation of the 1:l adduct of these trihalides with ammonia and organic nitrogen bases.119J20J21 Tristrifluoromethylstibine forms a 1 1 pyridine ad- d ~ ~ t l ~ ~ J ~ ~ whereas the trialkylstibines behave only as weak donors as is 118 Pritchard Pritchard SchifF and Trotman-Dickenson Chem. and Ind. 1956 896. 119 Ruff Ber.1906 39 4310. lZo Ephraim and Weinberg Ber. 1909 42 4447. 141 Biltz and Rahlfs 2. anorg. Chem. 1927 166 351. 123 Rose Compt. rend. 1901 132 204. lS3 Vincent 2. analyt. Chem. 1880,19,479. 260 QUARTERLY REVIEWS illustrated by the formation of quaternary stibonium compounds. Trialkyl- stibines add sulphur quantitatively to form compounds of the type R,SbS,124 but neither tristrifluoromethylstibine nor antimony trihalides does so. Both antimony trichloridelo4 and trialkyl~tibinesl~~J~~ form complexes with transition metals which presumably are stabilised through dn-dn bonding in a manner similar to that which occurs in the phosphorus- transition metal complexes (cf. p. 254). In view of this it is surprising that a reaction does not occur between tristrifluoromethylstibine and mercuric chloride or palladium chloride.71 Trifluoromethyl compounds of quinquevalent antimony are also more nearly comparable to antimony pentachloride or pentafluoride than to the corresponding alkyl compounds.Both antimony pentach10ridel~~J~~ and tristrifluoromethylantimony dichloride71 form 1 1 complexes with water and pyridine. A second stable hydrate of tristrifluoromethylantimony dichloride Sb(CF,),C1,,2H20 can be isolated and bas been formulated as [Sb(CF,),Cl,OH]- H,O+ in analogy to the dihydrate of antimony pentafl~0ride.l~~ Quinquevalent antimony halides can also accept halide ions to form compounds containing ions of the type SbX,-. Tristrifluoromethyl- antimony dichloride reacts with nitrosyl chloride yielding a compound which has been formulated as NO+[(CF,),SbCl,]- in analogy to the compound formed between antimony pentachloride and nitrosyl chlor- ide.130 Pyridinium tristrifluoromethylantimonate is readily converted into the corresponding pyridinium trihalogenotristrifluoromethyl- antimonate C5H5NH+[(CF,),Sb(OH),1- + 3HX -+ C,H,NHf[(CF,),SbX,]- + 3H,O .(25) In contrast to the weakly acidic properties of alkylantimonic acids an aqueous solution of [Sb(CF,),(OH),]-H,O+ [prepared by the action of silver oxide on solutions of tristrifluoromethylantimony dichloride] is a strong acid (pK = 1.85). The silver salt has been characterised as the benzene addition compound Ag(CF3),Sb(OH),,C6Hs and salts of larger cations as well as of organic nitrogen bases can be readily isolated. The salts of the heavy metals and of most simple inorganic cations appear to be soluble.J~~ and selenium133 have been prepared by the direct action of trifluoroiodomethane on the Group V1.-Perfluoroalkyl derivatives of 12* Landolt Annalen 1852 84 44. lZ5 Morgan and Yarsley J. 1925 127 184. Jensen 2. anorg. Chem. 1936 229 225. lZ7 Anschutz and Evens Annalen 1887 239 291. Williams J. 1876 30 463. Emeleus and Moss 2. anovg. Chem. 1955 282 24. 130 Gall and Mengdehl Ber. 1927 60 86. 131 Brandt EmelCus and Haszeldine J. 1952 2198. 132 Hauptschein and Grosse J. Amer. Chern. Soc. 1951 73 5461. 133 Dale Emeleus and Haszeldine J. 1958 2939. LAGOWSKI PERFLUOROALKYL DERIVATIVES 26 1 free element at high temperatures. The reaction with sulphur yields mainly bistrifluoromethyl disulphide together with small amounts of bistrifluoro- methyl tri- and tetra-sulphides but no monosulphides are formed.In contrast the reaction with selenium gives bistrifluoromethyl mono- selenide and diselenide only. A more convenient method for the prepara- tion of bistrifluoromethyl disulphide is the fluorination of carbon disulphide with iodine pentafluoride ;134 sulphur tetrafluoride as well as small quantities of bistrifluoromethyl trisulphide are also formed in this reaction. The polysulphides have been shown by a combination of chemical and physical The electrolytic fluorination of alkyl sulphur compounds in anhydrous hydrogen fluoride solution has been used to prepare perfluoroalkyl derivatives of sulphur. Complete decomposition with fluorination of the fragments is often observed and the sulphur-containing fragments generally appear as sulphur hexafluoride or its derivatives.13' Thus the electrolytic fluorination of dimethyl sulphide or carbon disulphide yields the extremely inert compounds CF3.SF, (CF,),SF, CF,(SF,), and CF,(SF3),.138 Trifluoromethylsulphur pentafluoride is the major reaction product; it is also formed in the fluorination (with cobalt trifluoride) of carbon disulphide methanethiol and carbonyl ~ulphide.l~~J~* Attempted electrolysis of solutions of dimethyl selenide or carbon diselen- ide in anhydrous hydrogen fluoride leads to extensive decomposition and deposition of elemental selenium.The electrolytic fluorination of alkane- sulphonyl halides yields perfluoroalkanesulphonyl fluorides which give rise to a series of perfluoroalkanesulphonic acids on h y d r ~ l y s i s . ~ ~ ~ J ~ The properties and some of the reactions of these perfluoroalkanesulphonic acids have been described.143 Sulphur.Of the bistrifluoromethyl polysulphides which have been prepared the disulphide is the most valuable from a synthetic viewpoint. The chemical inter-relation of bistrifluoromethyl disulphide bistii- fluoromethylthiomercury trifluoromethanethiol and trifluorornethyl- sulphenyl chloride are shown in Fig. 3. Heptafluoro-n-propyl derivatives undergo similar rea~ti0ns.l~~ Bistrifluoromethylthiomercury can also be prepared by the high-temperature reaction of carbon disulphide with mercuric f l ~ 0 r i d e . l ~ ~ Trifluoromethanesulphenyl chloride reacts with a number of hydrogen- 134 Haszeldine and Kidd J. 1953 3219. 135 Brandt EmelCus and Haszeldine J. 1952 2549. 136 Bowen Trans. Faraday SOC. 1954,50 452.13' Gramstad and Haszeldine J. 1956 173. 138 Clifford El-Shamy EmelCus and Haszeldine J. 1953 2372. 130 Silvey and Cady J. Amer. Chem. SOC. 1950,72 3624. I4O Silvey and Cady J. Amer. Chem. SOC. 1952 74 5792. 141 Gramstad and Haszeldine J. 1957 2640. 142 Brice and Trott U.S. P. 2,732,398; Chem. A h . 1956 50 13982h. 143 Gramstad and Haszeldine J. 1957 4069. 144 Haszeldine and Kidd J. 1955 3871. 145 Muetterties U.S. P. 2,729,663; Chem. A h . 1956 50 11362~. to contain linear S-S bonds. 262 QUARTERLY REVIEWS FIG. 3 Inferconversion of trifuoromefhyl sulphur compounds. U.V. = Ultraviolet irradiation. containing compounds with the elimination of hydrogen chloride (reactions 26);134J46 it can be oxidised with chlorine water to trifluoromethane- sulphonyl chloride,14' which is very slowly hydrolyzed by water at room 1 ' (24) H,S + 2CF3.SCI MH + CF3*SCI -+ M-SCF + HCI (M=N P J -+ CF3-S-S.SCF3 + 2HCI CFS'SH + CFa*SCl + CFa.S*S*CF + HCI temperature and more rapidly at higher temperatures to trifluoromethane- sulphonic acid CF3*S03H.This acid can also be prepared by the action of aqueous hydrogen peroxide on bistrifluoromethylthiomercury but attempts to oxidise bistrifluoromethyl disulphide directly led to extensive decomp~sition.~~~ As would be expected trifluoromethanesulphonic acid is a strong acid in aqueous solution and has many of the properties of the mineral acids; a monohydrate H,0+CF3.S03- has been is01ated.l~~ The free acid boils considerably lower (162") than its methyl analogue (165"/8 mm.) thus paralleling the decrease in the boiling point of the trifluoromethyl derivatives with respect to their methyl analogues noted for other compounds.The resistance of trifluoromethanesulphonic acid to alkaline hydrolysis is similar to that of the trifluoromethyl-phosphonic and -arsonic acid. Thus trifluoromethyl-oxy-acids appear to be most resistant to hydrolysis when the central element exhibits its maximum valency . Trifluoromethanesulphinic acid CF3.S02H has not been isolated but its sodium salt CF3*S0,Na,H20 has been prepared by the action of zinc dust on trifluoromethanesulphonyl chloride in air-free water.148 Basic hydrolysis of trifluoromethanesulphinic acid yields trifluoromethane quantitatively; this is the only trifluorometliyl sulphur derivative which does so since the other sulphur compounds are either stable to hydrolysis or decompose completely to yield fluoride sulphide and carbonate ions.Neither trifluoromethanesulphenic acid nor any of its salts has been isolated,14* but its formation has been assumed in the initial step of the hydrolysis of its acid chloride CF3.SC1.131J48 Trifluoromethanethiol the sulphur analogue of the unknown trifluoro- methanol is a stable gas at room temperature. It is decomposed slowly by 146 Nabi unpublished results. 14' Haszeldine and Kidd J. 1954 4228. 14g Haszeldine and Kidd J. 1953 2901. LAGOWSKI PERFLUOROALKYL DERIVATIVES 263 water and rapidly by aqueous base yielding fluoride sulphide and carbonate ions quantitatively. There is some evidence that the decomposi- tion occurs through the formation of thiocarbonyl fluoride and hydrogen fluoride (eqn. 27).144 This reaction is favoured in ionising solvents and in those which can act as acceptors of hydrogen fluoride suggesting that the CF3*SH + H20 -+ H,O+ + CF3*S- -+ CF2S + F- .. . . * (27) trifluoromethyl sulphide ion is unstable. Alkanethiols are more acidic than their oxygen analogues and on the basis of electronegativity considerations trifluoromethanethiol should exhibit marked acidic properties in aqueous solution. Thus trifluoromethanethiol follows the decomposition pattern of compounds containing the CF,-MH portion. Bistrifluoromethylthiomercury is substantially different from its methyl analogue; it is a low-melting solid (m.p. 37.5") which is soluble in water and in most organic solvents and it can be used to introduce trifluoromethanethio groups into molecules containing an acidic halogen (reactions 28-31 ; M = As P).40J44J49 There are indications that this may be a general reaction.Preliminary results150 indicate that bis(trifluor0- methy1thio)mercury can form 1 1 addition compounds with halide ions MX3 + Hg(S-CF,) MX3 + ZHg(S.CF.42 -+ 2Hg(SCF3)X + MX(SCF3) 1 . (28) -+ Hg(S.CF,)X + MX2SCF3 MX3 + 3Hg(S-CF3) 3 3Hg(S*CF3)X + M(S*CF,) J C12C:S + 2Hg(S*CF3) -+ (CF,*S),C:S + 2CF3.S*HgCl . . . . (29 CF,*S*C(:S)-F i- Hg(S-CF,) -+ (CF,*S),C:S + CF3.S.HgF . . . . . . (30) R-CO-CI + Hg(SCF,) -+ R-C(:O).S*CF + Hg(SCF,)CI . . . . (31) in a manner analogous to bisperfluoroalkylmercury compounds ; attempts to prepare higher addition compounds gave decomposition products.* Other metallic derivatives containing the trifluoromethylthio- group have been prepared. Bistrifluoromethylthiomercury precipitates trifluoromethylthiosilver from aqueous solutions of silver nitrate and trifluoromethylthiocopper(1) can be obtained by interaction of bistri- fluoromethylthiomercury and copper powder at 100°.145 Seleni~n2.l~~ Bistrifluoromethyl diselenide trifluoromethylselenium chloride trifluoromethaneselenol and bistrifluoromethylselenomercury undergo reactions analogous to those of the corresponding sulphur compounds (cf.Fig. 3). Their physical and chemical properties show a general resemblance to those of the sulphur compounds and the differences 14s Downs unpublished results. 160 Jellinek and Lagowski J. in the press. * Recent indicates that bistrifluoromethylthiomercury is associated in ben- zene as well as in the pure state. In addition this compound dissolves with evolution of heat in solvents which exhibit donor properties and the relative stabilities of the solvate appear to be PR3 > NR3 > SR > ORz in agreement with the that bistrifluoromethylthiomercury can act as an acceptor towards halide ions aqueous solution.laoa Man Coffman and Mueterties J. Arner. Chern. SOC. 1959 81 3575. 264 QUARTERLY REVIEWS which appear follow the trends noted in Group VI. Trifluoromethy derivatives of selenium are more reactive than the corresponding sulphur compounds and this is attributed to the weaker C-Se and Se-Se bonds. The presence of the very electronegative trifluoromethyl groups in bistrifluoromethyl selenide makes the unshared electrons on the selenium atom unavailable for donation to form a selenonium compound; com- pound formation did not occur with either methyl iodide or mercuric chloride.Similar results were obtained with bistrifluoromethyl diselenide. Alkyl selenides readily form addition compounds with both of these reagent~,l~l-l~~ but the donor ability of selenium dihalides has not been investigated. Bistrifluoromethyl selenide reacts with chlorine to give selenium tetrachloride chlorotrifluoromethane and trifluoromethyl- selenium trichloride. The last compound presumably arises from the addition of chlorine to trifluoromethylselenium chloride. Although alkylselenium compounds readily form dihalides (R,SeX& there has been no indication that bistrifluoromethylselenium dichloride can be prepared. Bistrifluoromethyl diselenide reacts with excess of chlorine to form trifluoromethylselenium trichloride or with a limited amount of chlorine to form trifluoromethylselenium chloride.Reaction with bromine yields only trifluoromethylselenium bromide while only trifluoroiodomethane was formed with iodine. Trifluoromethylseleninic acid CF,.SeO.OH was obtained from both the oxidation of bistrifluoromethyl diselenide with concentrated nitric acid and the action of water on trifluoromethyl- selenium trichloride. The increased acid strength relative to that of the methyl analogue parallels the effect noted with other acids containing the trifluoromethyl group. Attempts to oxidise trifluoromethylselenic acid to the selenonic acid were unsuccessful although trifluoromethylsulphonic acid can be prepared in this manner. The Reviewer thanks Professor H. J. Emeleus for his interest in and encourage- ment during the preparation of this Review which was written during the tenure of a Marshall Scholarship at Cambridge. Morgan and Burstall J. 1929 1096; 1930 1497; 1931 173. 16* Cam and Pearson J. 1938 282. 153 Jackson Annalen 1875 179 1.

ISSN:0009-2681    

DOI:10.1039/QR9591300233

出版商:RSC    

年代:1959

数据来源: RSC

摘要:

NEWER ASPECTS OF THE STEREOCHEMISTRY OF CARBOHYDRATES By R. J. FERRIER PH.D. and W. G. OVEREND D.Sc. (BIRKBECK COLLEGE UNIVERSITY OF LONDON) ALTHOUGH the term “conformation” was first used by W. N. Haworth in 1929l in connexion with sugars the principles of conformational analysis were developed later mainly in the realm of alicyclic chemistry. Nowadays interest is being focussed to an increasing extent on the detailed stereochemistry of carbohydrates and the application of these principles has proved fruitful. For discussions of stereochemistry it is important that the formulae used to depict compounds should give as accurate a representation as possible of the molecular shape. The earliest formula (e.g. I) for sugars gave only an indication of the functional groups present in the molecules.Later developments took into account classical stereochemical features but only recently has serious consideration been given to the graphical representa- tion of the detailed shapes of sugars. Historical aspects of the conventions adopted for the Fischer (straight- chain) formulae for sugars (e.g. 11) have been adequately discussed by Hudson.2 Such formulae are still of value as the simplest representation of aldoses and ketoses and probably cause the student least difficulty. Early in the development of carbohydrate chemistry it became apparent that sugars had properties which could only be reconciled by postulating cyclic structures of the hemiacetal type (e.g. 111); the existence of Q- and ,&forms was explicable owing to asymmetry at C(l). In the D- series the more dextrorotatory sugar of an anomeric pair was designated as the tc-f~rm,~ whereas in the L-series such a compound was of the ,&type.It is possible for the cyclic structures to be either five- or six- membered-the so-called furanose (e.g. TIIb) and pyranose (e.g. IIIa) s ~ g a r s ~ and Drew and Haworth5 introduced the now well-known perspect- ive formulae (e.g. IV) to give a more accurate representation of these cyclic forms. Formula= (I-IV) show the progressive development in the representation of D-glucose. For many years Haworth formula proved to be satisfactory and served for the development of descriptive carbohydrate chemistry as it is known today but they are not so useful in discussions of mechanisms of some reactions and of the detailed stereochemistry of sugars. In the hope that a more thorough understanding of these problems would result the ideas of conformational analysis (see ref.6) have been Haworth “The Constitution of the Sugars” E. Arnold and Co. London 1929 p. 90. Hudson Adv. Carbohydrate Chem. 1948 3 1 . Idem J. Amer. Chem. SOC. 1909,31 66. Drew and Haworth J. 1926 2303. Barton and Cookson Quart. Rev. 1956 10 44. * Goodyear and Haworth J. 1927 3136. 265 266 QUARTERLY REVIEWS HO,/H P HO,,H P H ,OH cc ( 1 ) YHO YHO HYoH OC (2) THOH H-?-OH H - 4 - 7 1 H - T J o (3) VHOH HO-7-H (4) CHOH H-?-OH 'EIPEH 7 H-{ (5) ~ H O H H-C-OH H-C H-$-OH Ho-cY HO*H,C HO*H2Y HO-CH applied. Some consequences of this development will be discussed in this Review. Six-membered Rings Six-membered (pyranose) ring systems will be considered in greatest detail because sugars and their derivatives exist and react most frequently in this form.Extensive investigations with cyclohexane7 have shown that it exists preferentially in the chair (V) rather than the boat form (VI). In this (V> (v I> (VIO preferred form the non-bonded interactions of the axially (ax) and equatorially (eq) disposed hydrogen atoms are minimised. It has been demonstrated by Hassel and his colleagues7 that the preferred conforma- tion of substituted cyclohexanes is that which has most of the substituents in equatorial positions. (The free-energy difference between the cyclo- hexanols with an axial or an equatorial hydroxyl group is about 0.8 kcal . /mole.*) The substitution of an oxygen atom for carbon in the cyclohexane ring leading to the basic skeleton (VII) of pyranose sugars causes only minor distortions in the ringYs and so the general conformational features are retained.It might however be expected that the heterocyclic ring will be slightly more readily distorted than the cyclohexane ring owing to the Hassel Quart. Rev. 1953 7 221; see also ref. 6. * Cookson Ann. Reports 1957 54 170. a Hassel and Ottar Acfa Chem. Scand. 1947 1 929. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 267 elimination of interactions of two hydrogen atoms which are replaced by lone pairs of electrons. Unfortunately we have to consider for the hetero- cyclic ring more strainless forms than for cyclohexane. ReeveslO has designated the possible regular forms as in Fig. 1 but more recently he has pointed out that there is an infinite number of skewed conformations a11 devoid of angle strain.ll Although it is convenient to discuss pyranose conformation in terms of regular geometrical structures it is necessary to note that distortions are extremely likely especially in the course of reactions.Other conventions have been proposed12 but for the purpose of this Review Reeves’s symbols are adopted. Formulz (VIII) and (IX) will be used throughout to represent the pyranose ring in the Cl and 1C conformations as they are most easily related to the usual Haworth formula. Chair forms Boat forms B 3 3 m 0 38 Ow3 FIG. 1. The eight regular strainless conformations of the pyranose ring with Reeves’s symbols ReeveslO concluded that whenever possible a chair form is preferred to a boat form. He assigned additive instability units of arbitrary value to the following features (i) axial groups other than hydrogen atoms (ii) an axial 2-hydroxyl group of which the C-0 bond bisects the angle between the two Co)-O bonds r‘A2 effect” caused probably by unfavourable dipolar interactions (cf.p. 268)] and (iii) an axial 5-hydroxymethyl group on the same side of the ring as another bulky axial substituent (“Hassel- Ottar effect”). Thus he was able to predict the favoured conformation of pyranose sugars of various configurations. By detailed study of the complexes which cuprammonium forms with diol groups in the molecules he showed the validity of his predictions. By using molecules containing a 1,2-diol system with the hydroxyl groups lo Reeves A h . Carbohydrate Chem. 1951 6 107. l1 Idem Ann. Rev. Biochem. 1958,27,15. l* Isbell J.Res. Nut. Bur. Stand. 1956,57 171; Guthrie Chem. andlnd. 1958 1593. 268 QUARTERLY REVIEWS held in fixed relationslo it was found that the cuprammonium reagent formed complexes only when the hydroxyl groups had a projected angle of 0" (i.e. were truly cis) or & 60". If the projected angle between the hydroxyl groups was greater than 60° complex formation did not occur. By working with suitably substituted derivatives of methyl p-D-glUC0- pyranoside ReeveslO was able to demonstrate that a laevorotatory complex is formed at the 2,3-position whereas with the 3,4-diol the complex is dextrorotatory. Little or no complex-formation occurs elsewhere in the molecule. The only possible conformations which would permit the forma- tion of these complexes are C1 and 3B. Since the 4,6-O-ethylidene deriva- tive which cannot assume the 3B conformation behaves like the 4-0- methyl ether it is believed that all the glucosides react with cuprammonium in the C1 chair form.A range of methyl glycopyranosides was examinedlO by this procedure and excellent correlation was noted between the con- formations deduced from the complex-forming reactions and those pGdicted on the basis of the assigned instability units. In the C1 conformation (X) /h-glucopyranose has all the hydroxyl groups and the hydroxymethyl substituent in equatorial positions. According to Reeves's ideas it has no instability factors and so this would 00 be expected to have the most stable ring. Glucose derivatives are found to be more stable than derivatives of other sugars. For example the free- energy difference between a-D-glucose 1-phosphate and a-D-galactose 1 -phosphate was found to be 0.7 kcal./mole.13 Rather surprisingly however 1 -substituted derivatives of a-D-glucose generally are more stable than the corresponding /3-analogues e.g.equilibrium mixtures of methyl ~-glucopyranosides,~~ penta-O-acetyl-~-glucopyranoses,~~ and acetohalo- geno-D-g1ucopyranosesfS all contain more of the M- than of the /I-form. This apparent anomaly is explained by Edward" in terms of dipole- dipole interaction between the lone-pair electrons on the ring-oxygen atom and the polar group on C(l); this favours the a-form. The representation of any sugar in the C1 conformation is probably best done by relating it to fl-D-glUCOSe in either the Fischer or the Haworth formula and making the necessary alterations to the all-equatorial model.The D-aldo-pentoses and -hexoses in the C1 conformation are shown in Hansen and Craine J. Biol. Chem. 1954 208 293. Bollenbeck "Methyl Glucoside" Academic Press Inc. New York 1958 p. 12. l5 Bonner J. Amer. Chem. Soc. 1951,73 2659. l6 Lemieux A h . Carbohydrate Chent. 1954 9 1; Haynes and Newth ibid. 1955 l7 Edward Chem. and Ind. 1955 1102. 10 207. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 269 Fig. 2. The more stable chair forms (as determined by Reeves and in- dicated in parenthesis) are not rigidly fixed. Instances are known in which molecules react in other conformations but in the sugar series little is known about the energy differences between conformations or their energies of interconversion. Angyal and McHugh,18 working with inositols recently calculated interaction energies of neighbouring groups in cyclo- hexane rings but we find that application of their values to the pyranoses does not give results in good agreement with those of Reeves.Barker and Shawlg have approached the problem from a purely geometrical view- point and after calculating total atomic overlaps for each chair form of the pyranoses have predicted the preferred conformations. The results obtained in many cases agree with those of Reeves but their approach is probably oversimplified in that it deals solely with steric factors in non- solvated molecules. Reeves’s method on the other hand suffers from the drawback that no alteration in shape of the sugar ring is envisaged during w Ribose (a -CI P-c I) Al lose (o( - c I,b- a> Arabi nose (a - IC ,/-I c) A1 trose (jS-Clor ICY /s <I or IC) Xylose (Oc-Cl,~-Cl) Glucose (oc - CI ,p c I) Gulose (d -CI or IC P -CII ldose (d-IC J-Ct or IC) Galactose (6 - CI p - c I) Lyxose (d-CI or IC A - C l .or lc) rn Mannose (d-Cl /-Cl or ic) Ta lose (O(-CI or IC P - C i or IC) FIG.2. The D-aldo-pentoses and -hexoses represented in the C1 conformation. The stable conformation of the anomers as predicted by ReeveslO are indicated in parenthesis. (Substituents other than hydrogen are denoted at C(2)-C(s).) complex-formation. It must be borne in mind that the non-formation of a complex is important in Reeves’s derivation of conformation and the above criticism does not apply in these circumstances. Lately infrared spectroscopy20 and proton magnetic resonance studies21 have provided alternative experimental approaches to the assignment of stable conformations.By these methods equatorial and axial glycosidic bonds can be differentiated and results indicate that the common sugars l8 Angyal and McHugh Chem. and Ind. 1956 1147. lQ Barker and Shaw J. 1959 584. 21 Lemieux Kullnig Bernstein and Schneider J. Amer. Chem. SOC. 1958 80 6098. Brock-Neely Adv. Carbohydrate Chem. 1957 12 13. 270 QUARTERLY REVIEWS exist in the conformations predicted by Reeves. Moreover Lemieux et aL21 have demonstrated that the configurations assigned to the a- and the /3-anomers of sugar acetates on the basis of Hudson's rules22 are correct. In 1957 Reeves and B l o ~ i n ~ ~ as a result of the study of the effect of alkali on the optical rotation of solutions of glycopyranosides were led to reconsider the importance of boat forms.They concluded that although chair forms are generally of lower energy in certain circumstances the difference may be offset when a boat conformation allows all bulky substituents to be equatorial. Rather than geometrically symmetrical structures it is suggested that distorted boats occur in which the tendency for large groupings to become equatorial is balanced by non-bonded interactions brought about by groupings approaching eclipsed positions. Previously Reeves24 had suggested in agreement with Freudenberg and Cramer,25 that amylose the linear component of starch contains CC-D- glucopyranose units in the Bl or 3B conformation which allows the polymeric 1- and 4- substituents to assume equatorial positions. Green- wood and Rossotti26 on the other hand have decided on the evidence of infrared absorption studies on the amylose-iodine complex that the pyranose units of amylose have chair (Cl) rather than boat forms.X-Ray crystallographic measurements have shown that a-D-glUCOSe,27 methyl /3-D-xylopyranoside,28 the pyranose ring in sucrose,29 a-D-glucos- amine hydr~bromide,~~ cc-~-rhamnose,~~ and P-~-arabinose~~ all exist in the crystalline form in a chair conformation and that this conformation is in all cases that predicted by Reeves. This cannot however be taken as proof that these compounds assume in solution exclusively the same chair conformations as the nature of inter- and intra-molecular bonding forces differ in the solid state and in solution. Hydrogen-bonding would be expected to influence the fine structure of sugars particularly in solution.There is good evidence for hydrogen-bonding to water rn01ecules,~~ and recent work with simple model compounds has indicated that intra- molecular hydrogen bonds may be significant in fixing conformation. Evidence which clearly demonstrates this has been given by Brimacornbe Foster and S t a ~ e y ~ ~ who showed spectroscopically that in dry carbon az Hudson J. Amer. Chem. SOC, 1916 38 1566. 23 Reeves and Blouin J. Amer. Chem. SOC. 1957,79 2261. 24 Reeves J. Amer. Chem. SOC. 1954 76 4595. 25 Freudenberg and Cramer Chem. Ber. 1950,83,296. 28 Greenwood and Rossotti J. Polymer Sci. 1958 27 481 McDonald and Beevers Acfa Crysf. 1952 5 654. 28 Brown Ph.D. Thesis Birmingham 1939. 2 9 Beevers and Cochran Proc. Roy. Soc. 1947 A 190 257.30 Cox and Jeffrey Nature 1939,143 894. 31 McGeachin and Beevers Actu Cryst. 1957 10,227. 33 Furberg and Hordvik Acra Chem. Scand. 1957,11 1594. a3 Kabayana and Patterson Canad. J. Chem. 1958,36,563 and refs. therein. Brimacombe Foster and Stacey Chem. and I d . 1958,1228; see also Brimacornbe Foster Stacey and Whiffen Tetrahedron 1958 4 351. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 271 tetrachloride solution cis- 1,3-O-benzylideneglycerol (XI) is completely hydrogen-bonded whereas the trans-isomer (XII) exists as an equilibrium mixture of the bonded (a) and the non-bonded (b) form. The appreciable amount of (XIIa) and the 20-30 % of form (XIII) found in 1,5-dideoxy-2 4-0-methyleneribitol indicate that in carbon tetrachloride solution bond- ing of this type could be sufficiently strong to hold pyranoid rings in what otherwise would be very unfavourable conformations.It would be F4 Ph (XI I> p?& Me (XIII) expected however that in aqueous solution the intermolecular bonding to solvent molecules would significantly diminish the effect of intra- molecular hydrogen-bonding of this type. Half-chair Forms.-It has been shown by electron-diffraction that cyclohexene oxides5 (XIV) exists like cycl~hexene~~ (XV) in the half- chair form Le. Ctl) C(z) C(,) and c(6) are coplanar and the other methylene groupings are in staggered positions. The 4- and 5-hydrogen atoms are in true axial and equatorial positions but those at C(,) and c(6) become displaced and so are termed quasi-equatorial (eq') and quasi-axial (ax').36 It is possible that these molecules could exist in half-boat forms (XVI) but physical and chemical evidence36 together with thermodynamic data3' suggest that the half-chair conformation is the more stable.Structures of similar shape occur in sugar derivatives which have a feature constrain- ing four consecutive atoms of the pyranoid ring in a plane e.g. a double bond as in glycals or an epoxide ring. Intermediates of the type (XVII) occur in some sugar reactions. Although these do not necessarily take up a half-chair they probably are in the form of the half-chair oxonium ion (XVIII). (XVI I) (XVIII) Half-chair conformations can be named by reference to the chair forms; thus a glycal derived from a sugar in a C1 conformation could we propose be considered to be in the H1 conformation. The alternative half-chair conformation is the 1H.35 Ottar Acta Chem. Scad. 1947 1 283. 36 Barton Cookson Klyne and Shoppee Chem. and Ind. 1954 21. 87 Beckett Freeman and Pitzer J. Amer. Chem. Soc. 1948 70 4227. 272 QUARTERLY REVIEWS Five-membered Rings on cyclopentane and tetrahydrofuran it has been deduced that although the rings are not quite planar the puckering is so small that the molecules are essentially flat. An indication that furan- oid rings are also flat was given by Barker and Stephens3* who found that a- and /3-furanosides could not be differentiated by infrared spectroscopy ; both anomers had the same projected angle from the ring for the C(,,-H bond. In the fructofuranosyl portion of crystalline sucrose29 and in the ribofuranosyl part of crystalline ~ y t i d i n e ~ ~ four of the ring atoms are coplanar and the fifth is 0.5 out of this plane.In both cases the atom which is displaced occupies the same relative position in the furanosyl ring i.e. 4 in the ketofuranoside and 3 in the aldofuranoside. It is considered that a rotation about the axis shown dotted in (XIX) facilitates lattice packing by bringing the larger substituents at positions A and B nearer to the plane of the ring and the hydroxyl at C almost exactly into that plane. There is evidence that in solution the ring is flexible. Thus irrespective of con- figuration furanosides are oxidised at C(2)-C(3) by the periodate ion via a cyclic intermediate the formation of which must cause considerable From infrared distortions particularly when the diol is trans. Such distortions are im- possible when the ring is locked by a bridge system (cf.1,6-anhydro- glucofuranose which is not oxidised). Relative Stability of Pyranoid and Furanoid Rings.-By analogy with results obtained in other branches of aliphatic chemistryPO (see Table 1) it would be expected that free sugars would exist preferentially in the six-membered pyranose ring form. On the other hand a five-membered ring would be predicted for lactones derived from sugar acids. These expectations are in accord with the well-established facts of carbohydrate 38 Barker and Stephens J. 1954 4550 and refs. therein. 39 Furberg Acta Cryst. 1950 3 325. 40 Brown Brewster and Shechter J. Amer. Chem. Soc. 1954 76,467. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 273 chemistry. That glycosides also are more stable in the six-membered ring form is illustrated by the fact that an equilibrium mixture of pyranosides and furanosides contains only small amounts of the latter.41 Brown Fletcher and Johann~en~~ have discussed the reactivity of groupings TABLE 1.Compound y-Hydroxy - butyraldehyde 6-Hydroxy- valeraldehyde Derived cyclic structure % of cyclic form at equilibrium 88.6 93.9” 72.8 y -Hydroxy- butyric acid lactone 6-Hydroxy- valeric acid lactone 9.0 (“Sugars have much higher proportions of cyclic structures. This conforms with the generalisation that substitution favours ring-f~rmation.~~) attached to five- and six-membered rings and have attributed the relative stability of the cyclohexane and the cyclopentanone derivatives to internal ring strains which arise during the formation of reaction intermediates.Acyclic Sugars A carbon chain takes up a zig-zag conformation with large substituents in staggered position^.^>^^ Consequently the acyclic form of D-glucose would be expected to exist as (XX; R=H) a representation which obviates the misconceptions which can arise with the Fischer formula. Working with hexitols Scl~warz~~ has shown that preferential attack by periodate occurs at diol systems which are represented as trans (threo-configuration) in the Fischer projection formula see (XX1)-(XXIII). Periodate normally oxidises preferentially a cis-a-glycol (erythro-configuration) but the apparent anomaly is readily explicable by reference to formulse (XX1a)- (XXTIIa). 41 Hammond in “Steric Effects in Organic Chemistry” ed. Newman Wiley and Sons Inc. NewYork 1956 p. 425.p 2 Brown Fletcher and Johannsen J. Amer. Chem. Soc. 1951,73,212. 43 McCoubrey and Ubbelohde Quart. Rev. 1951,5 364. 44 Schwarz J. 1957 276. 274 QUARTERLY REVIEWS y OH H OH .,I s I HO.N2C CHjOH HO (xx) (xx I a) SHiOH HO-$-H HO-$-H -_--_ ---- H-$-OH H -5- OH Manni to1 CH,-OH (xx I 1) (XXI I a) (XX I I I a) (--- Position of preferential attack by periodate) These results however do not prove that such conformations are adopted in solution. The phenomenon may be due to differences in stability of the cyclic intermediates derived from the threo- and the erythro-systems. In the same way the results obtained during studies of cyclic acetal formationa5 and borate compIex-formation,46 which have been interpreted as indicating that acyclic carbohydrates take up zig-zag conformations may be equally readily explained in terms of stability of the cyclic structures involved.Less ambiguous proof is given by Hough and Bragg4' who showed that a large substituent R in (XX) hinders approach of reagents to position 1 and by Littletonas who showed by X-ray analysis that in a crystal the gluconate ion has a nearly planar six- membered zig-zag carbon chain. Stereochemistry of Cyclic Intermediates and Derivatives The most common type of cyclic structure derived from sugars is that obtained by the reaction of an a-diol in the pyranose ring with a molecule or ion to give a system of fused five- and six-membered rings. cis-a-Diols 45 Barker Bourne and Whiffen J. 1952 3865. O6 Foster Adv. Carbohydrate Chem. 1957 12 81. 47 Hough and Bragg J. 1957,4347. O8 Littleton Act# Cryst.1953 6 775. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 275 (axial equatorial hydroxyl groups) or trans-a-diols (equatorial equatorial) may react in this way. The oxygen atoms of the hydroxyl groups are the same distance apart in both cases and the relatively high reactivity of the cis-diol is usually ascribed to the lower energy required to bring these oxygen atoms closer together i.e. to produce flat five-membered rings. In the case of a cis-a-diol a reagent which tends to produce flat rings will decrease the buckling of the pyranose ring while the puckering will be increased in a molecule having a trans-di01.~~ Periodate Oxidati~n.~~-Buist Bunton and their co-workers51 concluded that periodate cleavage of diols proceeds through a five-membered cyclic intermediate.Working with cycl~hexanediols,~~ they have shown that when the oxidising agent is the H310s2- ion the greater reactivity of the cis-compound may not be due to the above-mentioned “buckling effects” but rather to steric interactions which reduce the stability of the five- membered ring formed. Reaction with the ion H3102- occurs essentially without distortion of the pyranose ring and a buckled ring containing iodine is formed.52 It is probable that at the pH normally used for periodate oxidations this ion is unimportant and other species which lead to a less buckled five-membered ring and a distorted pyranose chair must be con- sidered These factors are important in discussions of the details of oxidation of glycosides and free sugars which are also oxidised in the In pyranoside rings cis-diols are also more reactive than trans-diols so that mannosides and galactosides are oxidised more rapidly than glucoside~.~~ A spectacular demonstration of the difference in rates of oxidation of galactosides and of glucosides and fructofuranosides (which have only trans-or-diols) has been provided by Mitra and P e r l i ~ ~ ~ ~ who preferentially oxidised the galactoside units in the tri- and tetra-sacchar- ides raffinose and stachyose and isolated sucrose from the oxidation products .The presence of substituents on a ring is important in determining the rate and site of primary oxidation. While it is uncertain where some pyranosides with small aglycone groupings suffer primary attack it has been that the 3,4-bond in phenyl p-D-glucoside is cleaved pre- ferentially.The rate of oxidation is less than for the or-anomer. It was assumed56 that these effects arise because of steric hindrance by the equatorial phenoxy-group particularly felt at the C(2)-C(3) position. Jackson in “Organic Reactions” Wiley and Sons Inc. New York 1944 Vol. II. cyclic f01111.53 49 Angyal and Macdonald J. 1952 686. 61 Buist and Bunton J. 1954 1406; 1957,4567,4580. 63 Buist Bunton and Miles J. 1959,743. 63 Hough Taylor Thomas and Woods J. 1958 1212. 64 Halsall Hirst and Jones J. 1947 1427; Jackson and Hudson J. Amer. Chem. 66 Mtra and Perlin Canad. J. Chem. 1957,35 1079. 66 Garner Goidstein Montgomery and Srmth J. Amer. Chem. SOC. 1958,80 1206. p. 341; Bobbitt Adv. Carbohydrate Chem. 1956,11 1. SOC. 1937,59 994. 276 QUARTERLY REVIEWS Examination of models shows that this explanation is not entirely satis- factory and we believe that the large equatorial aromatic substituent may cause partial locking of conformation particularly at the end of the sugar ring thereby reducing the tendency of the 2- and 3-hydroxyl groups to form the planar intermediate in the oxidation; the phenoxy-group in an axial position has an influence on the conform- ation and flexibility of the ring which is quite different.A 6-trityl group was shown to hinder attack at position 3 4 and the rate of oxidation at the 2,3-diol position in methyl 4-chloro-4-deoxy-~-glucoside is reduced by the presence of the chlorine atom.57 It is not known whether these effects are caused by steric hindrance to approach of the attacking ion or to alterations in the flexibility of the pyranose rings caused by the intro- duction of large substituents.The periodate oxidation products of glycosides have been studied5* and six-membered ring structures with appropriate conformations have been proposed. Barker and Shawl9 encountered a novel feature in the oxidation of D-ribose when a complex with the periodate ion involving three-point attachment was observed. Subsequently other compounds containing cis-cis-triols were found to form such complexes e.g. 1,6-anhydro-~- allose gives the complex (XXIV). This indicates that D-ribose may react in the ZC conformation in which one hydroxyl group of the reacting trio1 is equatorial and the other two are axial. (XXIV) Borate Interaction and Ele~trophoresis.~~--It is well known that borate anions form complexes with diols of suitable stereochemical form59 and it is believed that equilibria of the type are involved.The reaction is much used in carbohydrate chemistry to provide sugar molecules with an electric charge by virtue of which they can move across a potential gradient. The rate of migration (MG value) is dependent upon the position of the equilibria which is governed to a 57 Buchanan J. 1958 995. 68 Cadotte Dutton Goldstein Lewis Smith and Van Cleve J . Amer. Chem. Soc. 1957,79 691 ; Goldstein Lewis and Smith ibid. 1958,80,939; Guthrie and Honeyman Chem. and Ind. 1958 388. 5u Boeseken Adv. Carbohydrate Chem. 1949 4 189. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 277 large extent by the stereochemistry of the diol. Favourably disposed hydroxyl groups form stable complexes and confer high MG values on the sugar.The strongest complex is formed with a 1,2-cis-diol although the borate ion will form complexes to a smaller degree with other diols. By determining the MG values of a range of substituted glucoses FosterGo has shown that at pH 10 there are three main sites of combination in glucose. These are the 2,4- and 4,6-positions of the open-chain form and the 1,2-cis-diol in the a-form of either glucopyranose or glucofuranose. Although there are other possible sites of interaction e.g. the 2,3-diol it is believed that these are relatively unimportant. It has been shown similarlyG1 that the main sites of borate interaction with D-galactose are 1,2 and 3,4 in the ring form and less important 4,6 in the acyclic form. Consequently the MG value of a sugar or a sugar derivative depends on the sites available for complex formation the stability of the different complexes formed and presumably the ease with which the open-chain structures can be attained.A study of the MG values of glycosides shows that the extent of complex- formation in some cases depends upon anomeric configuration and also confirms the observation that the stereospecificity for complex-formation is not great. The borate complex formed by the methyl glucopyranosides is located at positions 4 and 662 and is more stable in the case of the 8- derivative. has suggested that the relative instability of the complex obtained from methyl a-D-glucoside is due to steric interaction between the axial glycosidic methoxyl group and the axial hydrogen atoms located at positions 3 and 5. The methyl xylopyranosides do not migrate in borate buffer and so the high MG values of the methyl arabopyranosides can be attributed to complex-formation by the 3,4-diol.Reaction at a cis-2,3-diol is hindered in pyranosides which have an aglycone grouping in a cis-disposition; e.g. methyl 13-D-lyxopyranoside methyl 8-D-mannopyranoside and methyl a-D-gulopyranoside all form complexes less readily than do their anomers. In pentofuranosides62 a cis-arrangement of the 3-hydroxyl group and the 4-hydroxymethyl group such as is present in methyl D- (XXVI) xylofurano sides (XXV) permits strong borate interaction whereas a trans-relation such as occurs in the L-arabofuranosides (XXVI) allows only a weak one. Comparison of the MG values of methyl a-D-glucofuranoside 1,2-O-isopropylidene-a-~-glucofuranose and the anomeric methyl D- Foster J.1953 982. 81 Bouveng and Lindberg Actu Chern. Scand. 1956,10 1283. 62 Foster and Stacey J. 1955 1778; Foster J. 1957 1395. 278 QUARTERLY REVIEWS galactofuranosides shows that the 2-hydroxyl substituent is not involved in the complex formed by methyl glucofuranosides. As the 3- 5 and 6- hydroxyl groups are important has suggested that a tridentate complex (XXVII) of the type proposed by Angyal and McHugh18 is formed. ;;-2zQ H ,OW t.JOL)B (XXVI I) AcetaZs and KetaZs.-For the purposes of this Review attention will be confined to the acetals and ketals formed by aldoses and aldosides. The chemistry of the acetals and ketals of glycitols has been dealt with in detail by Barker et and by MilkG3 Generally aldehydes condense with cyclic and acyclic polyhydroxy-systems to form preferentially six-membered rings but ketones more often give rise to five-membered rings.64 It has been suggested4o that factors important in this connexion are the size of the substituents in the ketone used and the fact that if a six-membered cyclic ketal is formed one of these substituents must assume an unfavourable axial disposition.Both factors tend to make a six-mem- bered ring less likely for ketals. Five-membered substituent rings. (a) On furanose sugars. When a 1,3-dioxolan ring is fused to a furanose ring the union is generally cis; isopropylidene derivatives arc the usual examples studied e.g. 1,242- isopropylidene-a-D-glucofuranose. Condensation may also occur with exocyclic hydroxyl groups but on partial acidic hydrolysis the fused ring system is always the most stable part of the molecule; e.g.1,2:5,6-di-0- isopropylidene-a-D-glucofuranose yields the 1,2-linked ketal. The bicyclic system assumes a V shape substituents taking up endo- or exo-positions and the favoured course of reaction is that which leads to a product with a minimum number of endo-substituents. For example D-ribofuranose can afford theoretically either 1,2-O-isopropylidene-a-~-ribofuranose (XXVIII) or 2,3-O-isopropylidene-cc- or -P-D-ribofuranose (XXIX) or (xxx>. It is clear that (XXX) with the fewest large endu-substituents will be favoured and it is indeed f0rmed.~5 When aldehydes are used to produce such bicyclic systems isomerism at the acetal carbon atom is theoretically possible but interactions caused when the large alkyl or aryl grouping is in an endo-position ensure that only one isomer is obtained; e.g.1,2-O-benzylidene-~-~-g~ucofuranose has the structure (XXXI). (b) On pyranose sugars. &Fusion of a five-membered to a six-mem- bered ring will cause alterations in the shape of the larger one which depend upon the tendency of the smaller one to be planar. Only when this tendency Mills Adv. Carbohydrate Chem. 1955 10 2. *4 Barker and Bourne J. 1952 905. 66 Levene and Stiller J. Bid. Chem. 1933 102 187. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 279 is overwhelming will the chair of the six-membered ring be distorted into a half-chair conformation,66 and it is usually considered that five-membered acetal or ketal rings do not alter the shape of the pyranose ring radically (cf. the cis-perhydroindane systemse7).(XXVIlI> (xx I X I Ho.H 2c%?H CMe I I1 (xxx I) Pyranose rings on which there are two cis-a-diol systems one on each side of the ring e.g. a-D-galactose readily form diacetals in which the ring fusion is cis-anti-cise8 as in 1,2 :3,4-di-O-isopropylidene-or-~-galacto- pyranose (XXXII). It is uncertain why this structure should be more stable than the system possessing two fused five-membered rings as in the isomer 1,2 5,6-di-O-isopropylidene-~-~-galactofuranose~ In this connex- ion it is noteworthy that fructose gives rise to the 2,3-0-benzylidene and 2,3:4,5-di-O-benzylidene-~-~-fructopyranoses in preference to 2,3-0- benzylidene-/h-fructofuranose. Possibly the endo-substituents present in all these furanose derivatives hinder their formation or perhaps this is another example of the preference of the sugars to assume pyranose ring forms.Isomerism occurs at the acetal carbon atom when a five-membered acetal ring is fused to a pyranoid ring and no preferred form can be predicted. Isomeric di-0-benzylidene derivatives of methyl a-D-mannopyranoside have been obtained and have been assignede9 structures which differ only in stereochemistry at C* in (XXXIII). 66 Cf. Lemieux and Cipera Canad. J. Chem. 1956,34,906. 67 Eliel and Pillar J. Amer. Chem. SOC. 1955 77 3600. 68 Turner in “Natural Products Related to Phenanthrene” Fieser and Fieser 69 Mills Chem. and Ind. 1954 633; see also Dobinson Foster and Stacey Tetra- New York 3rd edn. 1949 p. 620. hedron Letters 1959 1 1. 280 QUARTERLY REVIEWS Me MU (xu1 I) Six-membered substituent rings.(a) On furanose sugars. Although a molecule comprising a six-membered ring fused to a furanose ring is formed less readily than a system of fused five-membered rings several examples are known e.g. 3,Sacetals or -ketals of 1,2-0-isopropylidene- a-D-glucofuranose l,~-O-methylene-cc-~-glucofuranose 1,2-O-isopropyli- dene-a-D-xylofuranose. An explanation of the difference in the ease of acetal formation at positions 3,5 of 6-deoxy- 1,2-O-isopropylidene-6-nitro-cc- D-glucofuranose and -P-~-idofuranose~~ (which differ only in the configura- tion of the 5-hydroxyl group) has b:en offered on conformational In the idose derivative the six-membered ring has no interfering axial substituents whereas in the glucose compound both possible confor- mations of the largest ring have two large axial substituents on one side.It would be expected that the bulky grouping on the acetal-carbon atom would assume an equatorial position. (b) On pyranose sugars. Six-membered acetal or ketal rings may be fused to a pyranose ring by either a cis- or trans-junction. In the latter case provided the chair forms remain intact the compounds are held in a specific conformation (cf. trans-decalin) and the hydroxyl groups of the pyranose ring are sharply defined as axial or equatorial. It is believed however that the end of the pyranose ring furthest from the ring junction is somewhat flexible and may be distorted during a reaction.71 4,6-Cyclic derivatives of pyranosides which have a trans-relation for the 4-hydroxyl and the 5-hydroxymethyl group fall into this class e.g. methyl 4,6-0- benzylidene-or-D-glucopyranoside (XXXIV).Isomers differing in the stereochemistry at the acetal carbon atom are not found and it is assumed that the bulky group always takes up an equatorial position. Complications arise in cis-fused six-membered rings. Two conforma- tions are possible for methyl 4,6-O-benzylidene-~-~-galactopyranoside and these are termed “0 inside’’ (XXXV) and “H inside” (XXXVI) respectively. Fusion of the acetal ring to the galactoside in the C1 con- formation would give (XXXV) whereas reaction in the 1C form leads to (XXXVI). Less non-bonded interaction is involved in the “0 inside” form which provided other factors remain unaltered would be expected to be more stable. The behaviour of methyl 4,6-0-benzylidene-a- and -#h-galactoside in cuprammonium solution suggests that both exist in ‘O Grosheintz and Fischer J.Amer. Chem. Soc. 1948 70 1476. 71 Korytnyk and Mills J. 1959 636. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 28 1 this form.72 It would be assumed that the phenyl residue would take up an equatorial position in both the “0 inside” and the “H inside” form (as shown in XXXV and XXXVI) but there is some doubt whether this applies in every case.73 (xxxrv) (XXXV) (XXXVI) Sugar Anhydride~.~*-All hexoses when heated in acid solution form 1 ,&anhydrides to some extent. Until recently only the pyranose forms of the anhydrides had been encountered but 1,6-anhydro-~-~-galactofuranose has now been obtained by the treatment of galactose with To form the anhydropyranose the D-hexose adopts the 1C conformationlo and a relation is noted between the conformational stabilities and the position of the equilibrium.lOJ For example D-glucose and D-mannose form only very small amounts (ca.1 %) of anhydrides presumably because these derivatives have respectively three and two axial hydroxyl groups. On the other hand D-idose and D-altrose which have few instability units in the 1 C conformation exist preferentially in the anhydro-form in acid solution. Pratt and Ri~htmeyer~~ recently concluded that substituents at position 3 are important in determining the position of equilibrium and their predic- tion that a 3-deoxy-sugar would form an anhydride intermediate in amount between those produced by sugars isomeric at was confirmed by experiment. Foster et al. 77 have considered the stereochemistry of some 3,6-anhydro- pyranosides and have decided that chair (XXXVII) and boat (XXXVIII) forms may both be important.3,6-Anhydropyranosides are extremely (XXXVl I) (XXXVII I) strained and in non-aqueous acidic solution this strain is relieved by a rearrangement to the 3,6-anhydrofuranosides e.g. methyl glucopyrano- sides rearrange to the furanosides with retention of configuration at 72 Reeves J. Amer. Chem. SOC. 1949 71 1737. 73 Fletcher Diehl and Ness J. Amer. Chem. SOC. 1954 76 3029. 74 Peat Adv. Carbohydrate Chem. 1946 2 37. 75 Richtmeyer Arch. Biochem. Biophys. 1958 78 376. 76 Pratt and Richtmeyer J. Amer. Chem. SOC. 1957 79 2597. 77 Foster Overend and Vaughan J. 1954 3625. 282 QUARTERLY REVIEWS position 1 (scheme A). Alternatively dialkyl acetals of the 3,6-anhydrides are produced if the ring-transformation is sterically impo~sible;'~ e.g.methyl 3,6-anhydro-P-~-galactoside affords 3,6-anhydro-~-galactose di- methyl acetal (scheme B). Substitution of a phenyl for the methyl group @Me + ;Me - QHOH 'CH-OMe - H + C - t i 0 HO OH / Ho OH OH H o H i z e M e + ........ (A) ........ (B) precludes such a transformation as the electrophilic character of the aromatic ring prevents the formation of the required carbonium ion.77 Recently the converse type of reaction has been reported,78 namely an acid-catalysed ring expansion of methyl glycofuranosides into the pyrano- sides without inversion at position 1 a derivative (XXXIX) of methyl p-D-xylofuranoside was converted into the methyl p-D-xylopyranoside derivative (XL). The mechanism proposed is as shown. Ms = Me-SO (x L> One of the most important classes of sugar anhydrides is the epoxides; but these have been adequately described in this series recently79 and so are not considered here.78 Schaub and Weiss J. Amer. Chem. SOC. 1958 80,4683. 79 Newth Quart. Rev. 1959 13 30. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 283 Some Applications Mutarotation.-Reevesl0 has pointed out that in the equilibrium mixture of a- and fl-pyranose sugarsso in aqueous solution the predominant isomer is that expected from a consideration of instability factors. Furthermore the equilibrium constants lie near to a value of 2 which implies that the free energy difference between the anomers is of the order of 0.5 kcal./mole. The full implications of this are not apparent. It is interesting to note that 2,3,4,6-tetra-O-methyl-~-glucose and 2,3-di-O-methyl-~-xylose as well as lactose conform with these observations.Selective Reactivity u f Hydroxy I Grozrps. 81-Aft er protection of the highly reactive 1-hydroxyl group in sugars by glycosidation it is found as expected that a primary hydroxyl group is most reactive. Of the secondary hydroxyl groups that at position 2 often reacts before the remainder. Tlius methyl a-D-glucopyranoside yielded the 6-0-tosyl derivativesa and the 2,6-dibenz0ate.~~ In alicyclic chemistry it is well known that equatorial hydroxyl groups undergo esterification more readily than those which are axial.6 Recently examples of such a difference towards esterifying agents have been furnished in the carbohydrates series. Aspinall and Zweife18* have demonstrated this most clearly in their experiments on mannose derivatives.For methyl 4,6-O-ethylidene-a-~-m annoside (X L’z ) and 4- 0-met hy l- 1,6-anhydro-P-~- mannose (XLII) the conformations are held so that the 2- and the 3- hydroxyl groups are axial-equatorial and equatorial-axial respectively. Selective esterification (60-70 yicld) took place at the equatorial positions. The situation is more complex in the case of methyl 4,6-0-benzylidene- cc-D-glucoside in which both free hydroxyl groups must take up equatorial positions. It has been shown that the position of monoesterification is dependent upon the nature of the reagent.85 Acid chlorides of carboxylic and sulphcnic acids give in preponderant yield a product with the ester at position 2 anhydrides of carboxylk acids give esters at position 3 and sulphonic anhydrides again give the ester in position 2.The results listed in Table 2 were not obtained under standardised conditions but nevertheless 8o “Polarimetry Saccharimetry and the Sugars” U.S. Government Printing Office Sugihara Adv. Carbohydrate Chem. 1953 8 1 . 82 Compton J. Ainer. Chem. SOC. 1938 60 395. 83 Lieser and Schweizer Annulen 1935 519 271. 84 Aspinall and Zweifel J. 1957 2271. 85 Jeanloz and Jeanloz J. Amer. Chem. SOC. 1957 79 2579. 1942. 284 QUARTERLY REVIEWS indicate that the products are largely dependent upon the reagents. It i s apparent therefore that esterification of a hydroxyl group in a pyranoside ring is particularly facilitated if the group is in an equatorial position but that other factors presumably electronic also operate.TABLE 2. Reagent Crystalline products (%) 2,3- 2-Es ter 3-Es ter Starting Diester material BzCl . . . . 35 24 6 12 Me-SO,Cl . . . . 16 68 p-C,H,Me.SO,Cl . . - Ac20 . . . . 26 3 42 7 (p-CGH,Me*SOz)zO. . 15 80-85 - - - - - 60-7086 - 22 - (Me . SO 2) ,O . . 5 36 Relative Rates of Acidic Hydrolysis of Pyranosides.-The mechanism of acidic hydrolysis of glycosides has been showns7 to follow one of the pathways A and B. There are no results available to distinguish the a 0 r-c O H % Products ....... (A) O t R tlc (XLI I I) H aoR GgR % eCr:2 Products ._... ... (8) 'OR mechanisms. Shafizadehsa has pointed to the circumstantial evidence indicating that scheme B might be operative but Edward1' discusses the relative rates of acidic hydrolysis of pyranosides in terms of scheme A i.e.he assumes the rate-controlling step to be formation of a half-chair intermediate cation (XLIIT). Formation of this H 1 conformation involves small rotations about the 2,3- and 4,Sbonds. The ease of these rotations will depend in two ways upon the configuration and size of substituents on the ring (i) it will be decreased by the opposition between substituents at positions 2 and 3 and at positions 4 and 5 and (ii) it will be assisted by 8o Bolliger and Prins Helv. Chirn. Acta 1945 28 465. Bunton Lewis Llewellyn and Vernon J. 1955 4419. Shafizadeh Adv. Carbohydrate Cherrt. 1958 13 9. FERRIER AND OVEREND STEREOCHEMISTRY OF CARBOHYDRATES 285 recession of the axial substituents at positions 2 and 3 from those at positions 4 and 5 respectively. The former effect explains why heptosides are more stable than hexosides which are more stable than pentosides and why 2,3-dideoxypyranosides are more labile than 2- or 3-deoxy- pyranosides which are more sensitive than normal glycosides.It would be expected as a result of the second effect that glycosides with few axial hydroxyl groups in the stable conformation would be more stable than those with several and the determined ordersg is glucosides > mannosides > galactosides > gulosides. In the pentopyranoside series the order is xylosides > arabinosides > lyxosides. Chromatographic Behaviour.-The mobility of carbohydrates and their derivatives on cellulose paper is dependent upon adsorption on the cellulose support and on partition between the stationary aqueous phase and the mobile organic phase.It is influenced by the fundamental pro- perties of the molecules such as relation between the molecular weight and the number of free hydroxyl groups. In addition other more subtle factors must operate as in practice it is a simple matter to separate closely related isomers. Isherwood and Jermyng0 have discussed structure and chromatographic behaviour and have indicated some correlations e.g. the mobilities of a homomorphous series of pyranoses show regularities so that changes in the 5-substituent (ie. H Me CH,.OH) cause similar effects in each configurational series. Recently33 it has been predicted on theoretical grounds that equatorial hydroxyl groups in sugars will be considerably more strongly hydrated than axial groups. Therefore it might be expected that sugars which have few axial hydroxyl groups will be more soluble in the aqueous phase than those with several and will consequently travel more slowly on partition chromatograms.Similarly it might be expected that the adsorption factor would reduce the relative mobility of flat molecules. Pairs of sugars differing only in configuration at a single carbon atom were c ~ r n p a r e d . ~ ~ It was shown that for hexoses a change in configuration of a hydroxyl group from above the plane of the ring to below (Haworth perspective formulz) causes a shift in the RF value in one direction only and that this direction is not the same for changes at each carbon atom. The glucose-galactose pair was the only one of 12 examples which did not conform with this observation. Reconsideration of the subject shows that in every case (with the glucose-galactose excep- tion) the change in configuration causing a decrease in RF is an axial- to-equatorial shift when the hexoses are considered in the C1 conform- ation.Further we have shown that the RF value of a free sugar can in most cases be calculated from a formula of the type RF = K + (n + a)a where K and a are constants for one series (e.g. hexoses) n is the number of axial hydroxyls in the molecule and a is the fraction in the 8 9 Pigman “The Carbohydrates” Academic Press Inc. New York 1957 p. 209. Isherwood and Jermyn Biochem. J 1951 48 515, 286 QUARTERLY REVIEWS equilibrium mixture of anomeric form of the pyranose sugar which has the 1-hydroxyl group in an axial position (n and a depend upon conformation and it is of interest that the conformations of the sugars which give suitable n and a values are generally those predicted by Reeves).The homomor- phous series arabinose 6-deoxygalactose galactose (fructose) appears to be exceptional in that the members all have lower Rp values than would be predicted on this theory. Gulose too has a lower mobility than would be calculated and Jager et aLgl have offered an explanation for this fact. in terms of a shielding of the ring by an axial hydroxyl group which prevents organic solvent molecules from approaching the ring and thus effectively increases hydration. This concept is diametrically opposed to the ideas of preferential hydration and it is difficult to see how of the hexoses galactose and gulose alone are subject to this effect. BioZogicaZ Considerations.-Although it is apparent that those sugars which occur most frequently in Nature have conformationally stable pyranose ring forms and that those which are unstable conformationally do not exist extensively there appears to be only an indirect connexion between the two features.An explanation of the phenomenon is to be found in biosynthetic terms. 92 Aldolases from plants and animals catalyse the condensation of a system having a primary alcoholic grouping with aldehydes to give a product with a diol having the D-threo-configuration e.g. dihydroxyacetone phosphate 3. D-glyceraldehyde + D-fructose I-phosphate. As a result of reactions of this type glucose and xylose occur commonly. The occurrence of other sugars depends largely upon the ease with which they can be derived from such precursors.The selectivity of enzymic reactions need not be stressed here; it sufltices to mention the classical work of E. F. Armstrongg3 who related methyl a- and p-glucoside with a- and fl-glucose by the specific use of maltase and emulsin and the recent studies by Mayer and Larnerg4 who investigated the hydrolysis of glycogen by a- and /%amylase and discussed the mechanisms in terms of conformation of the a-glucopyranose rings. It has been demonstrated that stereochemical features are also very important in the reactions between antisera and artificial antigens having different carbohydrate determinants. 95 Such reactions are becoming increasingly useful in structural analytical studies of the polysaccharides. g6 The authors are grateful to Drs. A. B. Foster C. W. Rees and J. C. P. Schwarz for valuable discussi’ons and suggestions. Jager Ramel and Schindler Helv. Chim. Acta. 1957 40 1310. ga Hough and Jones Adv. Carbohydrate Chem. 1956 11 185. gs Armstrong J. 1903 83 1305. 94 Mayer and Larner J. Amer. Chem. SOC. 1959,81 188. O6 Marrack and Orlans “Progress in Stereochemistry,” Vol. 11 ed. KIyne and de la O8 See Heidelberger in “Polysaccharides in Biology” ed. Springer Josiah Macy Jr. Mare Butterworths London 1958 p. 228. Foundation New York 1956 p. 271.

ISSN:0009-2681    

DOI:10.1039/QR9591300265

出版商:RSC    

年代:1959

数据来源: RSC

摘要:

ERRATUM 1959 Vol. XIII No. 3 Page 186 Eqn. (7) For p read pa 188 Eqn. (15) Second line right-hand side For - R2,p read -R28,p Third line right-hand side should read 3[5R,RpRy - RZ(R,%3y + R,8, + RySad1/R7 189 Eqn. (19) First line For $p read 4 193 Line 17 For 6e read 8e 194 Eqn. (25) For A py read Aa:py 198 Eqn. (33) For /3cr#:cr~ read B,,x,~ 198 Footnote references For read 35 200 Eqn. (40) For ~ X H F read + ~ X H F 207 First line after eqn. (65) FUF molecules quadrupole read quadrupole molecules 209 First line after eqn. (70) For v read v

ISSN:0009-2681    

DOI:10.1039/QR9591300374

出版商:RSC    

年代:1959

数据来源: RSC