|
1. |
The mechanism of electrode processes in aqueous solutions |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 95-125
A. Hickling,
Preview
|
PDF (2802KB)
|
|
摘要:
QUARTERLY REVIEWS THE MECHANISM OF ELECTRODE PROCESSES IN AQUEOUS SOLUTIONS. By A. HICKLING D.Sc. PH.D. F.R.I.C. (SENIOR LECTURER IN PHYSICAL CHEMISTRY UNIVERSITY OF LIVERPOOL) I. Introduction THE electrolysis of aqueous solutions may involve a wide range of electrode reactions and consequent variation in the products of electrolysis. Thus a t an anode there may occur evolution of oxygen from the discharge of the hydroxyl ion discharge of other anions to give specific products dissolution of the anode material to give metallic ions and electrolytic oxidation of ions or un-ionised substances in the solution. Similarly a t a cathode there may occur evolution of hydrogen from the discharge of the hydrogen ion deposition of metals from the discharge of metallic ions and electrolytic reduction of ions or un-ionised substances in the solution.A satisfactory general theory of electrolysis should make it possible to predict what will be the nature and current efficiencies of the electrode reactions in any par- ticular set of experimental conditions. So far as any such general theory has yet been approached it suggests from simple energy considerations that the anodic process requiring the least positive potential and the cathodic process requiring the least negative potential will normally take place. Since however the potential a t which an electrode process occurs may depend very largely on its mechanism and be quite different from the calculated thermodynamic value a complete understanding of the mechan- ism of the possible electrode reactions is a necessary preliminary to the application of any such general theory of electrolysis.In this Review therefore attention will be primarily directed to the present state of our knowledge of the mechanism of the reactions occurring a t working electrodes,* i.e. electrodes a t which current is passing and elec- trolysis is continuously occurring. In contrast to the behaviour of reversible electrodes in equilibrium with ions in solution which is relatively simple and amenable to general quantitative treatment by thermodynamic methods working electrodes exhibit highly complex specific phenomena and the elucidation of the mechanism of any particular anodic or cathodic process * The reactions occurring at the dropping-mercury electrode in polarography COn6titUte a particular specific field and will not be included here.95 U 96 QUARTERLY REVIEWS constitutes a formidable physicochemical problem. The main features which a complete understanding of the mechanism should explain are (a) The potential a t which the process occurs a t a given rate and its dependence on experimental conditions ; this aspect embraces the wholo field of the kinetics of electrode reactions polarisation and overpotential. (b) The current efficiency of the process and how it is affected by experi- mental variables ; this is of interest where alternative electrode reactions are possible and is particularly important in technical applications of electrolysis. (c) The specific phenomena associated with the anodic dissolution and cathodic deposition of metals such as anodic passivity and the physical properties and structure of electrodeposited metals.Although these aspects of electrode phenomena have been the subject of continual investigation over the past 50 years since the pioneer work of Le Blanc and Caspari and a wealth of empirical data has been accumulated it cannot be claimed that any very satisfactory or complete theoretical interpretation has yet been achieved and there exists a diversity of ideas which still await a general synthesis. 11. Experimental Methods Voltammetric Methods.-The classical method of investigating reactions a t working electrodes involves the determination of the current density (c.d.)-potential relations under various experimental conditions and their correlation with the processes taking place as ascertained by chemical analysis. In general the current passing through the electrolytic cell is raised in stages by the application of an increasing e.m.f.and the potential of the electrode under investigation a t each c.d. is measured against a standard reference electrode in the solution. Where a stationary state is rapidly reached a t the electrode the instantaneous c.d.-potential curves are usually reproducible and provide a variety of information. For instance the potential a t which any particular process begins is indicated by a sudden increase in the current passing and the rates of the reaction per unit area of the electrode surface a t different potentials are measured by the corre- sponding c.d.s; furthermore if there exists a limiting rate a t which the reaction can occur this is indicated by the presence of a limiting c.d.and a characteristic step in the curve By making observations a t different temperatures the dependence of the rate of reaction on temperature as indicated by the variation of c.d. a t a fixed potential can be found and the energy of activation of the reaction deduced. Similarly the influence of concentration of electrolyte pH value solvent etc. on the rate of the electrode reaction can be investigated. The nature and current efficiency of the process occurring in any particular section of the c.d.-potential graph may be ascertained by carrying out electrolyses a t appropriate fixed c.d.s and determining the products formed. Where a stationary state at the electrode is not rapidly achieved and the potential of the electrode varies with time a t constant c.d. the position is rather more complicated.I n this case starting with the electrode in a ELICKLING ELECTRODE PROCESSES 97 reproducible virgin state it may be polarised at constant c.d. and the varia- tion of potential with time observed ; by carrying out a series of experiments of this type at different c.d.s a c.d.-potential graph may then be constructed the potentials being taken either when the electrode has reached a steady state on prolonged polarisation or after comparable periods of polarisation. Alternatively it may often be of interest to study the variation of the rate of an electrode reaction with time by following the variation in c.d. at a fixed potential. The exploration of c.d.-potential relations and the identifi- cation of the reactions occurring at particular potentials have been greatly simplified in recent years by the introduction of instruments with the aid of which the potential of a working electrode may be fixed at any arbitrary value the current being automatically adjusted to maintain this potential.Thus Hickling has described an electronic device of this type termed a potentiostat which is suitable for use with small currents and is specially adapted to the rapid determination of c.d.-potential curves and the direct observation of the rate of an electrode reaction at fixed potential under various experimental conditions and J. J. Lingane and others have described mechanical potentiostats suitable for large currents which are particularly useful in carrying out electrolyses at fixed potentials and determining the products formed. Rate of Growth of Polarisation Potentials.-The data given by the methods discussed above refer to a working electrode when a steady state has been attained or approached and it is apparent that valuable additional information should be provided by the study of the variation of potential with the quantity of electricity passed in the earliest stage of polarisation i e .immediately following the completion of the electrical circuit and before any products of electrolysis have left the electrode. Two main methods have been developed suitable for high and for low c.d.s respectively. In work a t high c.d.s the electrode in a reproducible virgin state is polarised at constant c.d. and the very rapid variation of the potential with time is followed oscillographically. Pioneer experiments in this field which met with only limited success were made by Le Blanc and Reichin- stein about 191OY3 but it is only in the last 20 years that valuable results have been achieved.In the early successful experiments 4 of Bowden and Rideal and of Baars the voltage developed between the working and the reference electrode was applied preferably after amplification to some form of string galvanometer the movement of which was photographed on a film moving at constant speed. After suitable calibration measurement of the track produced permitted construction of a graph of potential against quantity of electricity passed. Many later workers have used a similar lTrans. Paraday SOC. 1942 38 27. 21nd. Eng. Chem. Anal. 1945 17 332; Faraday SOC. Discussion 1947 1 203; C. W. Caldwell R. C. Parker and H. Diehl Ind.Eng. Chem. Anal. 1944 16 532. a M. Le Blanc Abh. Bunsen cfes. 1910 No. 3 ; D. Reichinstein 2. Elektrochem. 1909 15 734 913; 1910 16 916. ' F. P. Bowden and E. K. Rideal Proc. Roy. SOC. 1928 A 120 59 ; E. Baars Sitzungsber. Qes. Bef6rd. Naturw. Marburg 1928 63 213 ; seo also H. D. Holler Bur Stand. Sci. Papers 1925 20 153. 98 QUARTERLY REVIEWS techniq~e,~ but there has been a general tendency in recent years to avoid the difficulties attendant on the inertia of any mechanical oscillograph by using the cathode-ray tube as the indicating instrument.6 A further development 7 avoids the inherent difficulties of single sweep oscillographic methods in the observation and recording of high-speed transients by the use of a repetitive method which gives a stationary picture on the oscillograph screen showing directly the variation of potential with quantity of electricity passed.At very low c.d.s the change of potential with time occurs relatively slowly and can be followed over a period of seconds or minutes by taking successive readings from a suitable form of electrometer. This method thus has the advantage of simplicity. It was first used in the investigation of the anodic polarisation of silver,8 and has subsequently been employed by many worker^.^ The main difficulty encountered arises from the great effect of small amounts of depolarisers in the solution and the danger of loss of electrolytic products from the electrode surface. To minimise this special types of cell have been devised in which only a very small volume of electro- lyte is in contact with the electrode,lO and in recent work 11 on the cathodic polarisation of mercury at very low c.d.s the interesting device has been used of completely isolating the cathode and catholyte in a sealed glass vessel contact being made with the anolyte and reference electrode through the glass walls.These charging-curve methods have so far been mainly employed in the investigation of the processes occuring before the cathodic evolution of hydrogen and the anodic evolution of oxygen. Where a linear relation between potential and quantity of electricity passed is found the process is generally identified as the charging of a double layer and from the slope of the graph the capacity can be simply obtained since C = 6q/SV where C is the capacity in microfarads q the quantity of electricity in micro- coulombs and V the voltage developed in volts.Where discharge of ions occurs to give adsorbed products or compounds with the electrode material this is indicated by a step in the charging curve from which the potential range over which the reaction takes place and the quantity of electricity 5 H. Brandes 2. physikal. Chem. 1929 142 97 ; T. Erdey-Gr6z and M. VoImer ibid. 1930 150 213 ; T. Erdey-Grfiz and G. G. Kromrey ibid. 1931 157 213 ; W. J. Shutt and V. J. Stirrup Trans. Faraday SOC. 1930,26,635 ; W. J. Shutt and A. Walton ibid. 1932 28 740. 6 J. A. V. Butler and J. D. Pearson ibid. 1938 34 806 1163 ; J. A. V. Butler and I. M. Barclay ibid. 1940 36 128. A. Hickling ibid. p. 364; 1945 41 333. J. A. V. Butler and G. Armstrong Proc. Roy. SOC. 1932 A 137,604 ; G. Armstrong F.R. Himsworth and J. A. V. Butler ibid. 1933 A 143 89 ; J. A. V. Butler and G. Drever Trans. Paraday SOC. 1936 32 427 ; A. Slygin and A. Frumkin Acta Physico- chim.. U.R.S.S. 1935,3,791 ; 1936,4,911 ; 1936,5,819 ; B. Ershler and M. Proskurnin ibid. 1937 6 195 ; B. Ershler ibid. 1937 7 327 ; B. Ershler and G. Deborin ibid. 1940 13 347 ; B. Ershler Faraday SOC. Discussion 1947 1 269. 1oB. Ershler G. Deborin and A. Frumkin Trans. Faraday SOC. 1939 35 464. l l F . P. Bowden and K. E. W. Grew Faraday SOC. Discussion 1947 1 86 91. *R. Luther and F. Pokorny 2. anorg. Chem. 1908 57 290. HICKLING ELECTRODE PROCESSES 99 involved in it can be measured. Hence the number of atoms of hydrogen or oxygen participating in the process can be found and if the number of metal atoms in unit area of the electrode surface is known the nature and thickness of the film formed can be precisely worked out.The method has already proved very valuable since the information given is of a very direct and unequivocal nature and it seems likely to be of wide application to other electrode reactions. The main difficulty in the interpretation of the data provided is a t present the determination of the number of metal atoms in unit area of the electrode surface; this involves a knowledge of the ratio of the real to the apparent area of the electrode surface about which there is considerable uncertainty. A measure of this ratio can be obtained by Bowden and Rideal’s method,12 in which the cathodic capacity of the electrode in acid solution is measured and compared with that of mercury for which the ratio is taken to be unity but an alternative method for checking these figures would be very useful.Rate of Decay of Polarisation Potentials.-The study of the decrease of the potential of a working electrode with time after interruption of the polarising current has attracted considerable attention largely from the point of view of separating the various factors contributing to overpotential. Early workers l3 made use of commutator methods in which the polarising current was mechanically interrupted for varying times and the potential measured in these intervals. Later workers l4 have generally employed mechanical or cathode-ray oscillographs for following the decay process the general technique being similar to that involved in the study of charging curves. A method has also been described l5 in which an electronic interrupter is used in combination with a special type of potentiometer ; the polarising current is interrupted for adjustable intervals of time varying from about second upwards and the potential a t the end of each interval is measured.The information furnished by the study of decay curves is not so simple or direct as that provided by charging curves. Arrests correspond- ing to particular depolarisation reactions can sometimes be distinguished and some indication of double layer capacities obtained but in general the interpretation of a decay curve involves a knowledge of the mechanism of the polarisation and hence the method has been mainly used in testing various theories of overpotential. la Proc. Roy. SOC. 1928 A 120 80.13E. Newbery J. 1914 105 2419; 1916 109 1051 1066; 1922 121 7 ; 1924 125 511 ; S. Glasstone J. 1923 123 2926 ; 1924 125 250 ; 1925 127 1824 ; M. Knobel J . Amer. Chem. SOC. 1924,46 2613; H. J. S. Sand J. Grant and W. V. Lloyd J. 1927 378 ; A. L. Ferguson and G. Van Zyl Trans. Electrochem. SOC. 1925 47 227 ; A. L. Ferguson and G. M. Chen J . Physical Chem. 1932 36 1156 ; A. L. Ferguson Trans. Electrochem. SOC. 1939 76 113. lP F. P. Bowden and E. K. Rideal ; E. Baars locc. cit. ref. (4) ; A. L. Ferguson and G. M. Chen J . Physical Chem. 1934 38 1117 ; A. L. Ferguson and S. Kleinheksel ibid. 1938 42 171 ; A. L. Ferguson Trans. Electrochem. SOC. 1939 76 113 ; Faraday SOC. Discussion 1947 1 50; E. Newbery Proc. Roy. SOC. 1925 A 10’7 486; Trans. Electrochem. SOC. 1930 58 187 ; Trans. Faraday SOC.1947 43 127. 15A. Hickling Trans. Faraday SOC. 1937 33 1540; A. Hickling and F. W. Salt ibid. 1941 37 450 ; F. W. Salt Fazaday Soo Discussion 1947 1 169. layer a t the electrode ; this provides a very con- venient method for the measurement of double- layer capacities. l8 Where discharge of ions can where D is the diffusion coefficient of the potential-determining ion x its valency co the concentration in moles per c.c. CL) the frequency of the alternating current C' the capacity of the double layer and R T and F have their usual significance. Where the discharge of the ion is itself slow these expressions have to be modified and an additional resistance term introduced which by making certain assumptions can be correlated with the reaction rate. Thus P. Dolin and B.Ershler 2o have deduced that the measured total resistance should decrease with increasing frequency and approach a residual value a t high.frequencies equal to RT/zFJ, where J is the c.d. a t which the reaction can occur at the mean potential under l6 G. Jones and S. M. Christian J . Amer. Chem. SOC. 1935 57 272 ; P. D o h and B. ErshIer Acta Physicochim. U.R.S.S. 1940 13 747 ; D. C. Grahame J . Amer. Chem. Soc. 1941 63 1207 ; see also B. Breyer and F. Gutmann Trans. Faraday SOC. 1946 42 650; 1947 43 785. 17T. Borissova and M. Proskurnin Acta Physicochim. U.R.S.S. 1936 4 819; 1940 12 371 ; J. E. B. Randles Faraday SOC. Discussion 1947 1 11. 1*Cf. A. Frumkin and M. Proskurnin Trans. Faruduy SOC. 1935 31 110; M. Proskurnin and M. Vorsina Compt. rend. Acad. Sci. U.R.S.S. 1939 24 915 ; D.C. Grahame Zoc. cit. ref. (16). ID F. Kriiger 2. physikaz. Chem. 1903 45 1 ; T. R. Rosebrugh and W. Lash Miller J . Physical Chem. 1910 14 816. 2o LOC. cit. ref. (16) ; B. Ershler Faraday SOC. Discussion 1947 1 269. HICKLING ELECTRODE PROCESSES 101 equilibrium conditions (exchange current). J. E. B. Randles 21 has con- sidered the particular case of the discharge of metallic ions at amalgam electrodes in a similar way and has shown how the rate constants for the discharge reactions can be derived from the measured capacities and resist- ances. Although the results so far obtained in this way should be inter- preted with caution since the correlation of reaction rate with observed resistance involves a number of assumptions the method seems likely to be of great value in the study of the kinetics of electrode reactions.111. Electrode Kinetics and Polarisation Phenomena When ions are discharged at an electrode at an appreciable rate the electrode potential may be displaced from the reversible value. The elec- trode is then said to be polarised and to exhibit an overpotential or over- voltage which is defined as the potential difference between such a working electrode and a similar reversible electrode in the same solution. Two main factors may contribute to this overpotential (a) Changes in concentration of the ions near the electrode as a result of electrolysis may set up a potential difference ; this is usually referred to as concentration polarisation or concentration overpotential. ( b ) Thermodynamic irreversibility in the electrode reaction may necessi- tate an increased potential; this is the factor generally implied when the term overpotential is used without qualification and it is sometimes distinguished as activation overpotential.22 In addition if there is any appreciable resistance between the working and the reference electrode the measured potential difference may include an ohmic fall of potential equal to the product of the resistance and current strength. This has sometimes been termed resistance overpotential,22 but since it is not characteristic of the particular system being studied but depends largely on the particular experimental arrangement adopted and since experimental technique is generally directed to eliminating or minimis- ing this component it is perhaps better regarded as an undesirable resistance error.Concentration Polarisation.-As a result of electrolysis the concentration of the ions taking part in the reaction at a working electrode usually tends to diminish at the electrode surface. This tendency is opposed by the bringing up of ions to the electrode by diffusion convection and migration and a steady state is eventually reached when a concentration gradient has been established and the ionic concentration at the electrode surface has a value (c,) less than the value (c) in the bulk of the electrolyte. Hence by the Nernst equation the reversible potential of the working electrode is changed by an amount RT c 'Ic = logeCs As the reference electrode is usually outside the region of these concentration alLoc. cit. ref. (17). aaF. P. Bowden and J. N. Agar Ann. Reports 1938 35 90.st A. Hickling loc. cit. ref. (16). 102 QUARTERLY REVIEWS changes qc is included in the measured total potential difference and constitutes the concentration overpotential. Concentration polarisation is liable to occur with all electrode processes and its magnitude and dependence upon experimental conditions are best considered 24 by reference to the empirical concept of the diffusion l a ~ e r . ~ 5 It is supposed that a working electrode is covered by a thin layer of solution within which convection is negligible) and across which the ions pass by diffusion and migration. If it is assumed as a first approximation that the concentration gradient across the layer is linear then application of Fick’s law t o the diffusion of the relevant ionic species of diffusion coefficient D (cm.2 sec.-l) gives Rate of diffusion g.-ion/sec.= AD(c - ce)/6 . - (2) where 6 is the thickness of the layer A its area (equal to the area of the elec- trode) and c and c are the boundary concentrations in g.-ion/c.c. In the presence of excess of indifferent electrolyte so that migration of the ions in carrying the current is negligibly small this can be equated to the current passing and hence where I is the c.d. and substituting in AD(c - Ce)/S = IA/zF in amp./sq. cm. Hence on equation (1) ce = c - I ~ / D z F . 1 * (3) * (4) rearrangement From this relation it is seen that concentration overpotential will be mini- mised by high diffusion coefficient concentration and valency of the relevant ion while it will be increased by high c.d. and conditions which favour a thick diffusion layer.In unstirred aqueous solutions a t room temperature 6 has a fairly constant value of about 0.03 cm. ; 26 it is appreciably decreased by rise of temperature and very considerably by agitation) and both these factors therefore tend to reduce concentration overpotential. It is apparent from equation (5) that for a given set of conditions there will exist a limiting value of I a t which the concentration overpotential may become infinitely large when the second term in the bracket approaches unity. This is termed the limiting diffusion c.d. for the process under consideration and corre- sponds to a state of affairs in which ce approaches zero ; further increase of c.d. will then cause the potential to change to such a value that another 24 Cf. J. N. Agar and F.P. Bowden Proc. Roy. SOC. 1938 A 169 206. as A. A. Noyes and W. R. Whitney 2. physikal. Chem. 1897 23 689 ; W. Nernst ibid. 1904 47 52 ; W. Nernst and E. S. Merriam ibid. 1905 53 235 ; E. Brunner ibid. 1904 47 56 ; 1906 58 1 ; F. Weigert ibid. 1907 60 513 ; T. R. Rosebrugh and W. Lash Miller loc. cit. ref. (19) ; R. E. Wilson and M. A. Youtz I n d . Eng. Chem. 1923 15 603 ; S. Glasstone Tram. Electrochem. SOC. 1931 59 277. 2sEarly work suggested a value of 0.05 cm. (see S . Glasstone and A. Hickling “ Electrolytic Oxidation and Reduction ” 1935 Chap. 3 for review). More recent studies have given appreciably lower values (cf. I. M. Kolthoff and J. J. Lingane ‘‘ Polarography ” 1941 Chap. 32 ; J . Physical Chem. 1941,45 1062 ; J . Amer. Chem. Soc. 1939 61 3344 ; A. Hickling and W H.Wilson Nature 1948 162 489). HICKLING ELECTRODE PROCESSES 103 electrode process can occur. limiting diffusion c.d. ( I d ) is given by Equation (5) can therefore be rewritten From equation (5) it is apparent that the I d =cD.zF/G . * (6) Limiting c.d.s are readily measured experimentally from observations on c.d.-potential curves and hence the accuracy of the relation expressed in the form of equation (7) can easily be tested; it appears to hold satisfac- t ~ r i l y . ~ ' In the above deduction it has been assumed that the migration of the ions to the electrode is negligibly small; where this is not so the contribution made to the c.d. by transport can be allowed for and equa- tion (5) then becomes RT 1 qC = xF loge (1 - Idn/cDzF)' ' where n is the sum of the transport numbers of all the ions in the diffusion layer other than the species being discharged.It is thus apparent that the migration of the ions will tend to reduce concentration polarisation and increase the limiting c.d. With concentrated well-stirred solutions of simple salts concentration polarisation is usually very small except at extremely high c.d.s and from equation (7) it is seen that in general the working c.d. must exceed about one-tenth of the limiting value before the overpotential can become sub- stantial ; thereafter it increases very rapidly with rise of c.d. If however the substance to be liberated at the electrode is present in the solution as a complex ion in equilibrium with a very small concentration of simple ion the latter being discharged the position is rather different.Here the limiting c.d. is controlled by diffusion of the complex ion and may be quite high but the potential is governed by the minute concentration of simple ion and this may be markedly affected by electrolysis even a t very low c.d.s ; hence considerable overpotential may develop under these con- ditions. Abnormal concentration polarisation of this kind is frequently observed in the electro-deposition of metals from cyanide solutions.28 The above discussion of concentration overpotential has been based on the empirical concept of the diffusion layer since this provides a simple and direct approach. Attempts to treat the problem more fundamentally have until recently been unsuccessful. If the transfer of ions from an unstirred solution to a plane electrode is treated by ordinary diffusion theory then no indication is obtained that a stationary state should be set up in a short time and a diffusion layer of definite thickness established.Since these are characteristic features of the behaviour of a working electrode I. M. Kolthoff and J. J. Lingane op. cit. ref. (26) Chap. 9 ; I. M. Kolthoff and H. A. Laitinen J . Physical Chem. 1941 45 1062. 28 Cf. F. Foerster 2. Elektrochem. 1907 13 561 ; S. Glasstone J. 1929 690 702 ; 0. A. Essin and E. Alfknova J . Physical Chem. Russia,'1936 8 137 ; 0. A. Essin and T. Beklemysheva J. Ben. Chem Russia 1936 6 1602. 104 QUARTERLY REVIEWS it is apparent that some further factor must enter into the transfer process and this is provided by convection due to density differences which serves to keep the concentration of the solution uniform a t a small distance from the electrode.B. Levich 29 has recently succeeded in developing a funda- mental theory of convective diffusion which leads to the conclusion that a diffusion layer will in general be set up and permits the calculation of its thickness in some special cases and J. N. Agar 30 has used the method of dimensional analysis combined with data obtained from the analogous problem of heat transfer and arrives a t similar results. The problem of the variation of concentration overpotential with time at constant c.d. has received some attention. H. J. S. Sand 31 considered the case of an electrode in an unstirred solution and utilising Fick’s diffusion equations deduced that the time necessary for the concentration of ions at the electrode surface to be reduced to zero (transition time z) should be given by z = L(c/I)B .* (9) where k is a constant c the bulk concentration and I the c.d. This relation has been verified under special conditions such that convection is small.32 Where no special precautions are taken to avoid convection it would not be expected that the equation should hold accurately sinco a t all c.d.s below the limiting value the transition time will be infinite; Z. Kara0glanoff,~3 however showed that it applied fairly well to the oxidation of ferrous and reduction of ferric ions in these circumstances provided the c.d. used was appreciably higher than the limiting value. J. A. V. Butler and G. Arm- strong 34 have investigated a number of cathodic reduction processes and find that in general for short transition times the results can be represented by the relation which approximates to equation (9) when the c.d.used is much higher than the limiting value ; for long transition times however it appeared that the data were more in accordance with a relation of the form ( I - I& cc c . - (11) Activation Overpotential.-If any stage in the overall electrode reaction in between the arrival of the ions at the surface of the electrode and the ultimate formation of the stable electrolytic product is slow and limits the rate a t which the electrode reaction can occur then the electrode will exhibit ( I - I d ) d C c c . * (10) 2s Acta Physicochim. U.R.S.S. 1942 17 257 ; 1943 19 117 133 ; Faraday SOC. 80 Faraday SOC. Discussion 1947 1 26. 81 Phil. Mag. 1901 1,45 ; 2. physikal. Chem.1900 35 461 ; Trans. Faraday SOC. 1905,1 1 ; see also T. R. Rosebrugh and W. Lash Miller Zoc. cit. ref. (19) ; J. N. Agar and F. P. Bowden Zoc. cit. ref. (24). 3 2 H. J. S. Sand Eoc. cit. ref. (31) ; T. G. Cottrell 2. physikal. Chem. 1903 42 385 ; I. M. Kolthoff and H. A. Laitinen J. Amer. Chem. SOC. 1939 61 3344. s3Z. Elektrochem. 1906 12 5. 34 Proc. Roy. Soc. 1933 A 139 406 ; 1933 148 89 ; Trans. Faraday SOC. 1936 Discussion 1947 1 37 ; see also A. Eucken 2. Elektrochem. 1932 38 341. 32 989; 1934 30 1173. HICKLING ELECTRODE PROCESSES 105 activation overpotential. In general two main theoretical methods have been used in considering and treating this type of overpotential. The modern view which has been developed in various ways 35 in recent years has focussed attention particularly on ion-discharge as the likely rate- determining process and regards the electrode potential as directly affecting its speed.Thus if it is supposed that there is an energy barrier to be surmounted such that an energy of activation W is required at the reversible potential then this energy may be regarded as dependent upon the electrode potential. If a cathodic reaction is under consideration the energy of activation will be decreased by making the electrode more negative and for a displacement of potential q equal to the activation overpotential the new energy of activation will be W - ayF where a is a coefficient the value of which must lie between zero and unity. Thus at the reversible potential the rate of ion-discharge will be proportional to kla+e-w/RT where u+ is the activity of the cation at the surface of the electrode and E is a con- stant and since no net current is here flowing this must also be the rate of the reverse ionisation process.For the polarised cathode the rate of the forward reaction will be increjsed and will be given by klafe-(W-aqP)/ICT it being assumed for simplicity that the activity of the cation is not changed and the rate of the reverse reaction will be decreased and will be given by E,a’e-[~+(l-u)’iPl/RT where a‘ is the activity of the discharged ions and W’ is the activation energy of the ionisation reaction. Hence for a univalent cation the current I flowing per unit area at a working cathode will be given by I = Fk,a+e-(w- a‘iF)/RT - Fk,a‘e-[W+ (1 - E)VPI/RT . * (12) For appreciable displacements from the reversible potential the rate of the ionisation reaction is very small and hence I rr FkIa+e-(w’-d9/RT .(13) which reduces to 7 = a + (RT/aF) loge I . ’ (14) where a is a constant including the activation energy W. Alternatively equation (13) may be written 7 = (RT/aF) log I / I o . * (15) where I is the c.d. corresponding to the rate of the electrode reaction at the reversible potential. From equation (14) it is seen that the plot of q against log, I should be a straight line of slope 2*303RT/aF and this type of relation is often found to hold experimentally. The variation of overpotential with temperature at constant c.d. is then given by * (16) (W - wm = - ____.-_ aFT and of c.d. with temperature at constant overpotential by a1og I w - ccqF (e)~ = RT2 - (17) 85 See e.g. F. P. Bowden and J.N. Agar Zoc. cit. ref. (22) ; J. A. V. Butler “Electrocapillarity ” 1940 Chap. 6; S. Glasstone K. J. Laidler and H. Eyring “The Theory of Rate Processes” 1941 Chap. 10. 106 QUARTERLY REVIEWS from which expressions given suitable experimental data W can be found. For small polarisations where the electrode potential is close to the reversible value the rate of the ionisation process cannot be neglected. Simplifying equation (12) and expanding the exponential terms this reduces to r] OC (RT/F)I . (18) i e . in these circumstances the overpotential is directly proportional to the c.d. This approach to the problem of activation overpotential is very attrac- tive in its simplicity and directness and it provides a useful basis for the interpretation of experimental data. It is important to recognise however that it can only be applied if the slow stage responsible for activation over- potential concerns the discharge of ions.Although the nature of the energy of activation is not specified and might be involved either in the neutralisa- tion of an ion or in its dehydration the fundamental postulate is that it will be affected by the electrode potential. If however the slow stage in the electrode reaction occurs after the ions are discharged and does not involve charged particles it is not clear why the eleqtrode potential should exercise any direct effect on the rate of the electrode reaction. In this case the older approach to the problem of overpotential seems more useful. This postu- lated that owing to the slowness of the electrode reaction some unstable intermediate electrolytic substance regarded as more electromotively active than the ultimate product would tend to accumulate at the electrode surface and under a given set of experimental conditions would attain a stationary concentration and set up its own electrode potential.Where adequate thermodynamic data exist the potential of the intermediate in a standard state can be calculated and if the activity of the substance at the electrode surface can be inferred from the postulated mechanism of the electrode reaction the potential it should set up can be calculated from the Nernst equation and its dependence upon c.d. temperature and other experi- mental variables can be deduced. This quasi- thermodynamic method has been used by a number of authors,36 particularly in connection with the problem of hydrogen overpotential and often leads to an equation for the dependence of overpotential upon c.d.of a form similar to equation (14). These two theoretical approaches to activation overpotential do not seem to be in any way necessarily opposed; each would appear to be of value in its own field. Where the process of ion-discharge is slow the second method obviously breaks down since it assumes a thermodynamic equilibrium between some chemical product and its ions to be instantane- ously set up Similarly where ion-discharge is not the slow process in an electrode reaction the first method seems equally inapplicable since it is difficult here to see how the electrode potential can then directly affect the kinetics of the electrode process. The methods are therefore rather com- plementary and in any particular case it is the agreement of the inferences which result from the two treatments with experimental observations which must decide which is the more useful method and simultaneously will throw 38 See e.g.A. Hickling and F. W. Salt Tram. Faraday SOC. 1942 38 474. HICKLING ELECTRODE PROCESSES 107 light on the probable location of the rate-determining stage in the electrode re act ion. The main electrode processes in which activation overpotential is con- spicuously shown are the cathodic liberation of hydrogen the anodic evolution of oxygen and the electrodeposition of nickel iron and cobalt and these are discussed in subsequent sections. There has recently been a tendency 37 to cite as examples of activation overpotential any rsactions in which the working potentials differ from the values calculated from free- energy data and processes such as the anodic evolution of nitrogen in the electrolysis of azide solutions and the Kolbe and the Hofer-Moest reactions have been quoted.Although such reactions are indubitably irreversible in the thermodynamic sense it may perhaps be doubted whether this extension of the idea of overpotential is a useful one. The commonly accepted dehition of overpotential contains the implication that a reversible elec- trode for the system under consideration can be experimentally realised or approached under special conditions and this is certainly not the case with the reactions mentioned. Furthermore it is quite possible that in many electrolytic oxidation and reduction processes which would be included in the extended concept the products formed do not arise by direct discharge of ions but by the action of chemical oxidising and reducing agents first produced a t the electrodes.IV. The Cathodic Evolution of Hydrogen The formation of gaseous hydrogen from the discharge of hydrogen ions is under most conditions a highly irreversible process characterised by a marked activation overpotential and it is one of the most thoroughly investigated of electrode reactions. In early work great difficulty was encountered in obtaining reproducible values of hydrogen overpotential a t most cathodes and although considerable improvement has resulted from modern techniques in which the greatest possible care is taken to eliminate impurities from electrode and electrolyte and to allow for the influence of time of polarisation published values still have a relative rather than an absolute significance.It has however been possible to establish the influence of most experimental factors. The most striking feature of hydrogen overpotential is its marked dependence upon the nature of the cathode material. This is illustrated by the values summarised in Table I which are the measured overpotentials at various cathodes in N-hydrochloric acid at a c.d. of 1 ma./sq. cm.38 I n general as was pointed out by W. R. M ~ t t ~ ~ metals of low melting point have high overpotentials and vice versa although the correlation is by no means exact ; as would be expected from this overpotential like melting point seems to be a periodic function of atomic n~rnber.~O From some 37 Cf.J. N. Agar Ann. Reports 1947 44 5 ; H. P. Stout Trans. Faraday SOC. 1946 41 64. 38 Selected values taken from the work of A. Hickling and F. W. Salt ibid. 1940 36 1226 and J. O’M. Bockris ibid. 1947 43 417 carried out under similar conditions. 3D Trans. Electrochem. SOC. 1909 15 569. 40 A. J. Allmand and H. J. T. Ellingham Trans. Furaday SOC. 1924 19 748. 108 QUARTERLY REVIEWS theories of hydrogen overpotential it might be expected that there should be a parallelism between overpotential and the thermionic work function of the electrode material; this has recently been investigated by J. O'M. Bockris,41 who has reached the interesting conclusion that the relation is of an inverse type an increase in work function from one metal to another leading to a decrease of overpotential.T1 . . . 1-05 . . . 1.04 gf . . . 0.99 Pb . . . 0-88 Sn . . 0.85 * 1 TABLE I Hydrogen overpotentials (volts) in acid solution In . . . 0.80 Cu . . . Bi . . . 0.69 Ag . . . Nb . . . 0.65 Ta . . . Be . . . 0.63 1 Fe . . . Al . . _~_.__________ ~- 0.17 ~- rise of c.d. and over a considerable range the relation is a logarithmic one and as first noted by J. Tafe1,42 can be expressed by the equation q = a + b log, I where a and b are constants depending upon the nature of the cathode material and temperature. This relation has been the subject of much study and certainly holds with considerable accuracy over the c.d. range to amp./sq. em. for most cathodes. Since b varies with temperature and nearly all theories of hydrogen overpotential lead to its expression as a factor embodying RT/F the equation is now frequently written in the form q = a + (2*303RT/aF) log, I (cf.equation 14) and particular attention has been directed to the value of a According to those theories of hydrogen overpotential which regard the ion discharge step as the slow stage in the electrode reaction it seems probable that a should be 0-5 which would give b a value of 0.116 a t 17". For mercury cathodes in carefully purified acid solutions values very close to this have been observed and for many metals b is in the vicinity of 0.1 ; wide variations are however also encountered and values as high as 0.3 for lead and as low as 0.02 for platinised platinum have been reported.*3 There has been a marked tendency in recent years to suggest 22 that b always has the theoretical value (corresponding to a = 0.5) for pure metals and that discrepancies are due to contamination of the surfaces studied and in agreement with this it is found that traces of catalytic poisons tend to increase b .This view would not however account for the values of b lower than 0.116 which are often found and furthermore the high values even if due to contamination remain experimental facts which require explanation. The investigation of hydrogen overpotential a t c.d.s below amp./sq. cm. is difficult owing to the great effect of minute amounts of depolarisers but 41 LOC. cit. ref. (38). 4 2 2 . physikal. Chem. 1905 50 641. 43For lists of values of 6 see F. P. Bowden and J. N. Agar Zoc. cit. ref. (22); A. Hickling and F. W. Salt ; J. O'M. Bockris Zocc. cit. ref. (38). HICKLING ELECTRODE PROCESSES 109 it has been shown 44 that Tafel's equation holds for a mercury cathode down to 10-9 amp./sq.cm. while for catalytically active metals such as platinum there is some evidence that q is proportional to I a t very low ~ . d . s . ~ ~ The study of hydrogen overpotential a t high c.d.s above amp./sq. cm. involves experimental difficulties due t o the magnitude of the resistance error which may be included in the measured potential differences. B. Kaban0~,~6 using the direct method of measurement and attempting to correct for resistance error claimed that Tafel's equation applied accur- ately a t silver amalgamated silver and platinum cathodes up to c.d.s greater than 10 amp./sq. cm. On the other hand Hickling and Salt,3* working in the c.d. range to 1 amp./sq.cm. found that for some metals there were negative deviations from Tafel's equation and there was frequently a tendency for the overpotential to approach a specific limiting value ; they used an electronic interrupter method in which the potential was measured after interruption of the polarising current for varying periods and the values extrapolated to zero time of interruption. These results have been criticised by A. Frumkin 47 on the ground that the graphical extrapolations are unreliable but it may be noted that B~ckris,~S using the direct method of measurement under conditions such that resistance error is minimised has observed similar tendencies with some metals. Tafel 4 2 noted that hydrogen overpotential at most cathodes tended to rise with time of polarisation and this has been confirmed by many sub- sequent workers.Mercury tungsten and platinised platinum give over- potentials which reach constant values within a very short period of polarisation and most common metals attain a stationary state within about one hour but with some electrodes notably platinum and palladium the potentials may increase over a prolonged period ; with electrodeposited metals these times are usually reduced. In some cases the deposition of impurities from the electrolyte upon the cathode may contribute to this time variation but in general there appears little doubt that the phenomenon is a characteristic feature of hydrogen overpotential. Its importance in determining the conditions necessary for obtaining reproducible measure- ments of hydrogen overpotential has been stressed in recent pa~ers.3~ On very prolonged electrolysis roughening of the cathode surface may some- t,imes cause a decrease of overpotential.A large number of investigations of the influence of hydrogen-ion con- centration on hydrogen overpotential have been made,48 and although the results are not completely satisfactory the bulk of the evidence seems to 44 F. P. Bowden and K. E. W. Grew Faraday SOC. Discussion 1947 1 86. 45 J. A. V. Butler and G. Armstrong Trans. Paraday SOC. 1932,28,380 ; M. Volrner 46 Acta Physicochim. U.R.S.S. 1936 5 193. " I b i d . 1943 18 23. 48 See e . g . J. Tafel Zoc. cit. ref. (42) ; S. Glasstone J. 1924 125 2646; F. P. Bowden Trans. Paraday SOC. 1928 24 473; Proc. Roy. SOC. 1929 A 126 107; S. Lewina and V. Sarinsky Acta Physicochim. U.R.X.S. 1937 0 491 ; 1937 7 485 ; S.Sofa and B. Kabanov ibid. 1939 10 616 ; A. Hickling and F. W. Salt Trans. Faraday SOC. 1941 37 333. and H. Wick 2. physikal. Chem. 1935 172 429. 110 QUARTERLY REVIEWS indicate that in dilute solutions of pure acids the overpotential is practically independent of pH. In concentrated acid solutions it has been reported that the overpotential decreases with increasing concentration. 4B Accord- ing to some theories the overpotential might be expected to increase with decreasing hydrogen-ion concentration in acid solutions containing an excess of added salts and this variation has been observed by some workers although the change is small.50 At high c.d.s in dilute acid solutions con- centration overpotential may occur and has recently been investigated ; 51 where the electrolyte contained excess of neutral salt the polarisation was in general that to be expected from the change in hydrogen-ion concentration a t the cathode surface but in dilute acid solutions not containing any added electrolyte very high overpotentials exhibiting many peculiar features were observed.Few measurements have been made 52 in alkaline solution and these generally a t a mercury cathode where amalgam formation may cause complication but it appears that Tafel’s equation is approximately obeyed b having however a value considerably greater than in acid solution. Variation of the nature of the solvent might be expected to have a consider- able effect on hydrogen overpotential if the hydrogen ion is involved in the rate-determining step. Measurements have been made 53 in methyl and ethyl alcohol acetic and formic acid ether ethylene glycol dioxan cyclo- hexanol and liquid ammonia.So far as the results can be readily summar- ised i t would appear that some lowering of high overpotentials takes place in the anhydrous solutions although with low overpotential cathodes the influence is much smaller ; the solvent effect is however very complex and in mixed aqueous-non-aqueous solutions maxima are frequently observed a t intermediate compositions. Investigation of the rate of decay of hydrogen overpotential on inter- ruption of the polarising current suggests that two depolarisation processes are inv0lved.5~ One which becomes apparent only at high c.d.s greater than 10-3 amp./sq. em. leads to a rapid drop of potential in the fmt thousandth of a second and seems to depend primarily on the polarising 49 S.Jofa Acta Physicochim. U.R.S.S. 1939 10 903; A. J. de Bethune and G. E. Kimball J. Chem. Physics 1945 13 53 ; A. Rius and J. Llopis Anal. Pis. Quim. 1946 42 897. 5 0 s . Lewina and V. Sarinsky loc. cit. ref. (48); C. Wagner and W. Traud 2. Elektrochem. 1938 44 391 ; S. Jofa and A. Frumkin Acta Physicochim. U.R.S.S. 1943 18 183. 5 1 G. E. Coates J . 1945 484 ; P. M. Bryant and G. E. Coates Faraday SOC. Dis- cussion 1947 1 115. 5 2 F . P. Bowden and H. F. Kenyon Nature 1935 135 105. 53 G. Carrara 2. physikal. Chem. 1909 69 75; S. Swann and E. 0. Edelmann Trans. Electrochem. SOC. 1930 58 75 ; S. Lewina and M. Silberfarb Acta Physicochim. U.R.S.S. 1936 4 275; I. S. Novoselski J . Physical Chem. U.S.S.R. 1938 11 369; V. Pleskov Acta Physicochim.U.R.S.S. 1939 11 305 ; A. Hickling and F. W. Salt Trans. Paraday SOC. 1941 37 224 ; J. O’M. Bockris Faraday SOC. Discussion 1947 1 95; J. O’M. Bockris and S. Ignatowicz Trans. Faraday SOC. 1948 44 519; J. O’M. Bockris and R. Parsons &id. p. 860. 5 4 See e.g. E. Baars loc. cit. ref. (4) ; G. Armstrong and J . A. V. Butler Trans. Paraday SOC. 1933 29 1261 ; A. L. Ferguson Trans. Electrochem. SOC. 1939 76 113 ; A. Hickling and F. W. Salt loc. cit. ref. (15). HICKLING ELECTRODE PROCESSES 111 c.d. used ; the second which is observed at all c.d.s leads to a slower decay and follows a course such that the overpotential ultimately becomes pro- portional to the logarithm of the time and the rate of decay is then primarily dependent upon the nature of the cathode material.The rate of build-up of hydrogen overpotential has been studied by charging-curve methods. At a mercury cathode which has been very fully in~estigated,~~ the potential varies linearly with the quantity of electricity passed until it is close to the hydrogen evolution value and the process appears to correspond simply to the formation of an ionic double layer having a capacity of about 20 ,m./sq. em. in pure acid solutions. At platinum cathodes the formation of an approximately monatomic layer of adsorbed hydrogen atoms can be detected. Hydrogen overpotential is lowered by increase of temperature and for high-overpotential metals the temperature coefficient is some - 0.002 to - 0.003 v./degree whereas for low overpotential metals it is very much smaller.56 Heats of activation calculated by equations (16) and (17) from the rather limited experimental data available range from about 6 to 20 kg.-cals.for different cathodes.22 Early experiments on the effect of pressure on hydrogen overpotential led to uncertain results but more recent investigations indicate that 7 is slightly decreased by rise of pres~ure.~' The addition of catalytic poisons such as arsenious oxide mercuric chloride and carbon disulphide markedly increases hydrogen overpotential at catalytically active cath0des,~8 and similar effects have been reported 69 for colloidal substances such as gelatin and gum arabic which are often used as addition agents in electroplating ; alkaloids of the quinoline group have been observed 6o to bring about a lowering of overpotential at mercury cathodes although they raise the overpotential a t platinised platinum electrodes.Capillary-active substances usually have some influence on hydrogen overpotential but any general relation is so far obscure.61 The overpotential of deuterium is definitely higher than that of hydrogen 5 5 F. P. Bowden and E. K. Rideal; E. Baars locc. cit. ref. (4) ; H. Brandes T. Erdey-Grfiz et al. locc. cit. ref. ( 5 ) ; A. Hickling loc. cit. ref. (7) ; F. P. Bowden and K. E. W. Grew loc. cit. ref. (11). 5 6 E. K. Rideal J . Amer. Chem. Soc. 1920 42 9 4 ; M. Knobel and D. B. Joy Trans. Electrochem. Soc. 1923 44 443 ; G. M. Westrip J . 1924 125 1112 ; S. Glass- stone ibid. p 2651 ; W. D. Harkins and H. S. Adams J . Physical Chem. 1925 29 205 ; F. P. Bowden Proc. Roy. SOC. 1929 A 128 107 ; N. Kobosew and N.I. Nekras- sow 2. Elektrochern. 1930 36 529 ; A. Hickling and F. W. Salt loc. cit. ref. (45). 57 See e.g. S. J. Bircher and W. D. Harkins J . Amer. Chem. SOC. 1923 45 2590 ; H. M. Cassel and E. Krumbein 2. physikal. Chem. 1934 171 70 ; V. Schischkin et al. 2. Elektrochem. 1934 40 713 724 ; 1936 42 631 693 ; G. Schmid and E. K. Stoll ibid. 1941 47 360. 58 M. Volmer and H. Wick loc. cit. ref. (45) ; von Naray-Szabo Naturwiss. 1937 25 12 ; A. Hickling and F. W. Salt loc. cit. ref. (48). 59 C. Marie Compt. rend. 1908 147 1400 ; J. N. Pring and U. C. Tainton J . 1914 105 710 ; N. Isgarischew and S. Berkmann 2. Elektrochem. 1922 28 47 ; G. M. Westrip loc. cit. ref. (56). 6o J. O'M. Bockris and B. E. Conway Nature 1947 159 711. 61 See e.g. A. Thiel and E. Breuning 2. anorg. Chem. 1913 83 329 ; S.Glasstone Trans. Faraday Xoc. 1925 21 3 6 ; T. Onoda 2. anorg. Chem. 1927 165 93. 112 QUARTERLY REVIEWS under comparable conditions 62 implying that the speed of the rate-deter- mining reaction is less and this factor must therefore be operative in the separation of the isotopes by electrolysis. Cathodes appear to fall into two groups 63 as regards the separation factor ; for platinum nickel gold silver and copper the factor has values ranging from 5 to 7 while for tin mercury and lead it has a value of about 3. A satisfactory theory of hydrogen over- potential might therefore be expected to throw light upon these differ- ences and upon the dependence of the separation factor on experimental conditions. Theories of Hydrogen Overpotential.-Few problems have attracted as much theoretical speculation as that of the mechanism of hydrogen over- potential and the literature offers a bewildering complexity of alternative views.A very thorough and detailed account of current theories has recently been given by J. O'M. Bockris.64 It is proposed here to summarise the present position in general terms. It is now widely agreed that three main reactions may be involved in passing from the hydrated hydrogen ion to hydrogen gas and may be intimately concerned in the development of hydrogen overpotential. There is &st the discharge of the hydrogen ion involving its dehydration to give hydrogen atoms which in most cases will be adsorbed upon the metal cathode surface wiz. H,O+ +Me + e + H,O + MeH . - (1) This may be followed by the combination of hydrogen atoms in pairs accord- ing to the Tafel 4 2 catalytic mechanism thus 2MeH -+ 2Me +H .* (W or by discharge of a further hydrogen ion upon an adsorbed hydrogen atom according to the electrochemical mechanism first suggested by Heyrovsky thus H,O+ + MeH + e -+ H,O + Me + H,. . (111) Any one of these stages may be the rate-determining step responsible for the observed overpotential. The idea that the discharge of the hydrogen ion might be a slow process (reaction I) first attracted favourable attention when developed mathematic- ally from different standpoints by Erdey-Grbz and Volmer and R. W. Gurney,65 and it has been adopted in various forms by many subsequent workers ; 66 its later development is particularly due to A. Fr~mkin.~' In its extended form in which allowance is made for the activity of the hydrogen 6 2 F.P. Bowden and H. F. Kenyon loc. cit. ref. (52) ; J. Heyrovsky and 0. H. g3 J. Horiuti and G. Okamoto Sci. Papers I n s t . Phys. Chem. Res. Tokyo 1936 28 Muller Coll. Czech. Chem. Comm. 1935 7 281. 231 ; H. F. Walton and J. H. Wolfendcn Trans. Paraday SOC. 1938 34 436. Chem. Reviews 1948 43 525. 6 5 Proc. Roy. SOC. 1931 A 134 137. 66 See e.g. R. H. Fowler Trans. Paraday Soc. 1932 28 368 ; J. A. V. Butler ibid. p . 379; Proc. Roy. SOC. 1936 A 157 423; J . Horiuti and M. Polanyi Acta Physicochim. U.R.S.S. 1935 2 505. 67 Ibid. 1937 7 475; 1943 18 23; Faraday SOC. Discussion 1947 1 57. EIICKLING ELECTRODE PROCESSES 113 ions in the double layer adjacent to the cathode if the coefficient cc is given the probable value of 0.5 it leads to an equation for the overpotential of the form where y is related to the electrokinetic potential.this reduces to In pure acid solutions which is an equation of the Tafel type b being 0.116 at 17" in close agree- ment with the value found for a mercury cathode. Addition of neutral salts to the electrolyte would be expected to decrease y and hence increase the overpotential while if y is kept constant and the pH value varied should increase with decreasing hydrogen-ion concentration in agreement with some reported observations. Furthermore the theory would predict that in the initial growth of hydrogen overpotential the rate-determining process should be the building-up of an ionic double layer and hence the potential should be directly proportional to the quantity of electricity passed which again is found experimentally for a mercury cathode.The slow- discharge theories therefore account satisfactorily for some aspects of hydrogen overpotential particularly a t high overpotential cathodes but they seem defective in many other respects. For instance they have provided no real explanation of the marked dependence of overpotential on cathode material ; in a qualitative manner it can be seen that where a metal is a good adsorbent for hydrogen atoms with a high heat of adsorption the energy of activation of the neutralisation process may be reduced and hence the overpotential lowered but no detailed treatment has yet been worked out. Furthermore they give no explanation of such characteristic features of hydrogen overpotential as the variation with time the influence of' catalytic poisons and the solvent effect.The initial postulate of the slow-discharge theories vix. that the neutralisation of the hydrogen ion is slow is a concept which is not altogether easy to accept and H. Eyring S. Glasstone and K. J. Laidler 68 have put forward a new theory of over- potential in which it is supposed that the rate-determining stage is the transfer of a proton from a water molecule in the solution to an adsorbed water molecule on the cathode. By application of the theory of absolute reaction rates to this process the usual equation for the dependence of overpotential upon c.d. can be deduced with the coefficient a having the value of 0 - 5 ; the variation of overpotential with electrode material can also be qualitatively accounted for in terms of the strength of the Me-H bonds formed.The theory is also considered to be applicable to the explana- tion of oxygen overpotential the rate-determining stage here being the reverse of that at a cathode i.e. the transfer of a proton from a molecule of adsorbed water on the anode to one in the solution and this is supposed 6a J . Chem. Physics 1939 7 1053 ; Trans. Electrochem. Soc. 1939 76 145 ; G. E. Kimball S. Glasstone and A. Glassner J . Chem. Physics 1941 9 91. 114 QUARTERLY REVIEWS to account for certain apparent similarities between hydrogen and oxygen overpotentials. Several workers have criticised the theory adversely on various grounds,69 and from the experimental standpoint it might be expected that according to it change of solvent would have a much larger effect on hydrogen overpotential than is in fact found.The view that hydrogen overpotential is due to the accumulation of atomic hydrogen at the cathode appears to have originated with Tafe1.42 In its simplest form it supposes that the rate of combination of hydrogen atoms (reaction 11) is slow and that therefore at any appreciable rate of electrolysis there will exist a small concentration of atomic hydrogen at the electrode and this as may readily be shown thermodynamically will give rise to a potential substantially more negative than the reversible hydrogen potential. Those metals which are good catalysts for the combination of hydrogen atoms might then be expected to give low overpotentials while those with little catalytic power would have high overpotentials and K. F. Bonhoeffer 7O has shown that this correlation does hold experimentally.Other evidence such as the ready diffusion of cathodic hydrogen through the action of catalytic poisons,58 the marked reducing power of high overpotential cathodes,72 and the transference of overpotential from the polarised to the unpolarised side of iron and palladium diaphragms separating two ele~trolytes,7~ lends further qualitative support to the general theory. If however the kinetics of the process are considered and it is assumed that the potential is proportional to the logarithm of the adsorbed hydrogen atom concentration the dependence of overpotential upon c.d. is given by the equation q = const. + (RT/ZF) loge I . ' (21) This is of the correct form but with the coefficient a = 2 giving b a value of 0.029 a t 17" ; this is very much lower than is observed experimentally except for very active electrodes such as platinised platinum.It has however been pointed out 36 that there is no theoretical basis for considering the potential as directly dependent upon the logarithm of the adsorbed hydrogen concentration but rather the pressure of free atomic hydrogen in equilibrium with it should be considered. If this is done then for poor adsorbents or for surfaces approaching saturation much larger values of b are obtained.74 The possibility of the electrochemical mechanism (reaction 111) being involved in the determination of the rate of the overall electrode eg A. Frumkin Acta Physicochim. U.R.S.S. 1940,12,243 ; J. A. V. Butler J . Chem. ?OZ. physikal. Chem. 1924 113 199. 71 See C. J. Smithells " Gases and Metals " 1938 p.77 for a full account of this Physics 1941 9 279. phenomenon. See S. Glasstone and A. Hickling op. cit. ref. (26) Chap. 5. 73 H H. Uhlig N. E. Cam and P. H. Schneider Trans. Electrochem. SOC. 1941 79 111 ; A. Frumkin and N. Aladjalova Acta Physicochim. U.R.S.S. 1943 19 1 ; see also H. P. Stout Faraday SOC. Discussion 1947 1 107. 7 4 A. Hickling and F. W. Salt Eoc. cit. ref. ( 3 6 ) ; J. Horiuti G. Okamoto and K. Hirota Sci. Papers In&. Phys. Chem. Res. Tokyo 1936 29 223. HICKLING ELECTRODE PROCESSES 115 reaction has been particularly studied by Horiuti and his co-workers ; 75 results in moderate quantitative agreement with experimental observations have been obtained on this basis. If on the other hand I11 is assumed to be a fast reaction and to become of importance a t surfaces largely covered with adsorbed hydrogen then by making fairly plausible assump- tions as to the relative participation of reactions I1 and I11 in the removal of adsorbed hydrogen it has been found possible to account for negative deviations from the Tafel equation at high c.d.s and the approach to constant values of overpotential for the general form of hydrogen overpotential decay and for the observed separation coefficients of hydrogen and deuterium.36 It is of course entirely possible that the mechanism of hydrogen over- potential may be different a t different metals a point of view which has been emphasised by a number of authors.77 So far as the present position can be summarised it would appear that there is a fair general measure of agreement among workers in this field that at low overpotential cathodes the rate-determining stage responsible for overpotential is the formation of molecular hydrogen from adsorbed hydrogen atoms (reaction I1 and/or 111).At high overpotential cathodes such as mercury there is little evidence of the presence of adsorbed hydrogen and a considerable body of opinion holds that the slow-discharge mechanism (reaction I) is in closest agreement with the experimental observations. This dual view which is in many ways very attractive meets however with a fundamental difficulty which has not so far been resolved. If the formation of molecular hydrogen is accepted as the rate-determining step at low overpotential cathodes which are known to be good catalysts for the combination of hydrogen atoms it is not easy to see why it should not also apply t o high overpotential metals which are known to be poor catalysts for the reaction.V. The Anodic Evolution of Oxygen It has generally been considered that an oxygen electrode consisting of oxygen gas a t a passive electrode in contact with an aqueous solution should give rise to a potential corresponding to the equilibrium 0 +2H,O f 4 e + 40H- and the potential of such an oxygen electrode against a hydrogen electrode (both at 1 atm. pressure) in the same solution can be calculated from thermo- dynamic data in a number of ways 78 and is found to be 1.23 v. at 17"; this leads to a value of + 0.40 V. for the standard potential of oxygen. 7 5 Idem ibid. ; J . Horiuti ibid. 1940 37 274 ; G. Okamoto J . Fac. Sci. Hokkaido Imp. Univ. 1937,. 2 115 ; see also A.Frumkin Zoc. cit. ref. (67) ; P. Lukowtsew S. Lewina and A. Frumkin Acta Physicochim. U.R.S.S. 1939 11 21 ; A. Legran and S. Lewina ibid. 1940 12 243 ; P. D o h and B. Ershler ibid. 1940 13 747 ; K. Rosenthal P. D o h and B. Ershler ibid. 1946 21 213. 76 N. Kobosew and N. I. Nekrassow Zoc. cit. ref. (56). 77 L. P. Hammett Trans. Paraday SOC. 1933,29,770 ; F. P. Bowden and J. N. Agar 78 See S . Glasstone " The Electrochemistry of Solutions " 1945 p. 334. Zoc. cit. ref. (22) ; J. A. V. Butler Zoc. cit. ref. (35). 116 QUARTERLY REVIEWS Attempts to realise such a reversible oxygen electrode experimentally a t room temperature have been unsuccessful ; 79 in practice the potential reached is always lower than the theoretical value and drifts with time and it does not vary with alteration of pressure of oxygen gas in the way to be expected from the Nernst equation.The potentials at which oxygen is formed at a working anode have therefore usually been measured relatively to a hydrogen electrode in the same solution and the calculated value for the reversible oxygen electrode has been used in assessing the oxygen overpotential. TABLE I1 Oxygen overpotentials (volts) in alkaline solution ~ ~~ AU . . . . . . . Pt . . . . . . . Cd . . . . . . . CU . . . . . . . Pd . . . . . . . Ag . . . . . . . 1 c.d. (amp./sq. cm.) = 1 I 10-I. I/ c.d. (amp./sq. cm.) = 1 I lo-'. 0.93 0.80 0-67 0.49 0-48 0.45 1.63 1.50 1-21 0.73 1-12 0.94 Ni . . . . . . . 0.45 0.91 Fe . . . . . . 0-37 0.56 Co . . . . . . . 0.32 0.54 C (graphite) . . . . 0-37 1-12 Platinised Pt . . .. 0.32 0.89 Early measurements so indicated that oxygen overpotentials were sub- stantial and varied considerably with the nature of the anode material. It was also found s1 that the overpotential varied very markedly with time of polarisation and this factor which has been ignored by many later workers contributes to the great difficulty which has been met in obtaining reproducible values of oxygen overpotential. It has recently been shown,82 however that starting with a clean metal anode and polarising a t constant c.d. until the potential is steady oxygen overpotentials reproducible to about & 0.02 v. can be obtained and in Table I1 are given the values obtained by this procedure at a number of anodes in N-potassium hydroxide at 20" ; e2 the dependence of overpotential on c.d.varies considerably with different electrode materials so that it is impossible to arrange them in 79 See F. Foerster " Elektrochemie wiisseriger Losungen " 1922 p. 200 for account of early work ; N. H. Furman J . Amer. Chem. Soc. 1922 44 2685 ; G. Tammann and F. Runge 2. anorg. Chem. 1926 156 85; W. T. Richards J . Physical Chem. 1928 32 990 ; H. V. Tartar and V. E. Wellman ibid. p. 1171 ; T. P. Hoar Proc. Roy. Xoc. 1933 A 142 628 ; H. G. Bain Trans. Electrochem. SOC. 1940 78 173. See e.g. A. Coehn and Y. Osaka 2. unorg. Chem. 1903 34 86; F. Foerster and A. Pipet 2. Elektrochem. 1904 10 714 ; J. B. Westhaver 2. physikal. Chem. 1905 51 65; E. Miiller and F. Spitzer 2. anorg. Chem. 1906 50 321. *IF. Foerster loc. cit. ref. (80) ; 2. physikal. Chem. 1909 69 236. 82 A. Hickling and S.Hill Faraday SOC. Discussion 1947,1,236. For other measure- ments on the influence of electrode material see T. Onoda J . Chem. Xoc. Jupan 1922 43 782; 2. anorg. Chem. 1927 165 79; M. Knobel P. Caplan and M. Eiseman Trans. Electrochem. SOC. 1923 43 55; M. Knobel ibid. 1925 47 131 ; M. de K. Thompson and A. L. Kaye ibid. 1931 60 229 ; A. D. Garrison and J. F. Lilly ibid. 1934 65 275 ; H. Hunt J. F. Chittum and H. W. Ritchey ibid. 1938 73 299; G. Grube and W. Gaupp 2. Elektrochem. 1939 45 290; M. de K. Thompson and G. H. Sistare Trans. Electrochem. SOC. 1940 78 259. HICKLING ELECTRODE PROCESSES 117 any order of decreasing overpotential which will be generally valid and hence two sets of values are given for moderate and for high c.d.s respectively. It may be noted that oxygen overpotentials tend to be even higher than those observed with hydrogen; gold and platinum anodes show the highest values while those at cobalt and iron are relatively low and there appears to be no relation whatever to the corresponding hydrogen overpotentials.Oxygen overpotential increases with rise of c.d. and in acid solutions the dependence seems to follow a Tafel equation b for platinum having a value in the vicinity of 0.12.83 On this slender basis it has been supposed 22 that there is a fundamental similarity between the kinetics of oxygen and of hydrogen evolution but it is very doubtful whether this claim can be substantiated. Thus in alkaline solution it has been shown 82 that although over limited ranges of c.d. the q-logI graphs for cobalt iron nickel platinum platinised platinum palladium silver and copper all show agreement with an equation of the Tafel type b varying however between 0-07 and 0.30 yet when the graphs are considered in their entirety a number of discrepancies are apparent.The most striking feature is that at a number of anodes notably gold palladium and graphite the graphs have a step- like character the overpotential changing abruptly in certain specific c.d. regions; furthermore at low c.d.s there appears to be a general tendency for the overpotentials to approach constant minimum values. The influence of pH value and nature of electrolyte on oxygen over- potential has not been very fully investigated and the results reported are not in agreement. studied a platinum anode in sulphuric acid neutral phosphate buffer and dilute sodium hydroxide solution over the c.d.range 10-7 to amp./sq. cm. and found that a Tafel relation was followed in each case the overpotentials being nearly the same in acid and in alkali but somewhat higher in phosphate buffer. F. P. Bowden and H. W. Keenan 84 found that in alkaline solutions the overpotential decreased as the hydroxyl-ion concentration increased while in sulphuric acid it was independent of pH value. More recently A. Hickling and S. Hill,85 working with fully polarised platinum gold and palladium anodes in N-potassium hydroxide neutral phosphate buffer and N-sulphuric acid over the c.d. range to 1 amp./sq. cm. have found that at low c.d.s the overpotential is almost independent of pH value but at high c.d.s increases considerably with increasing pH; this is attended by an increase in the slopes of the 7-log I graphs and a t gold and palladium with the appearance of the step- like dependence which is largely absent in acid solution.J. O’M Bockris 86 has compared oxygen overpotentials in M-sulphuric acid solutions in water and in acetic acid-water and dioxan-water mixtures ; he finds that addition of either non-aqueous component causes an increase of overpotential rapidly 83 Cf. F. P. Bowden Proc. Roy. SOC. 1929 A 126,107 ; T. P. Hoar loc. cit. ref. (79). s4Quoted in ref. (22). 85 Unpublished work ; see S. Hill Thesis Liverpool 1948. B6Faraday SOC. Discussion 1947 1 229. 118 QUARTERLY REVIEWS a t first and then more slowly the slopes of the Tafel lines also being increased. E. Muller 87 found that addition of sodium fluoride to an acid electrolyte considerably increased the oxygen overpotential a t a platinum anode the effect increasing with increasing fluoride concentration.This has been con- firmed by more recent work; a5 the effect occurs in both acid and alkaline electrolytes but is only apparent a t relatively high c.d.s and appears to be specific to platinum no influence being apparent a t gold cobalt or palladium anodes. A number of observations of the influence of temperature on oxygen overpotential have been made.88 In general the overpotential is decreased by rise of temperature and temperature coefficients of - 0.003 and - 0.002 v./degree a t platinum in acid and in alkaline electrolytes have been rep0rted.8~ Calculation of the corresponding energies of activation in the conventional way gives values of 18.7 and 25.3 kg.-cals.respectively,89 although in view of the different slopes of the Tafel lines reported by various workers it is doubtful how reliable these values are. Limited observations on the decay of oxygen overpotential mainly a t platinum anodes have been made,g0 and it appears that in the long-term decay 7 becomes propor- tional to log t as for hydrogen overpotential. It has been variously reported that oxygen overpotential is diminished by the superimposition of an alternating current,g1 and by light and X-rays,92 but fuller investigation seems to be required before the exact significance of these results can be assessed. Theories of Oxygen Overpotential.-From the preceding account it is apparent that speculation as to the mechanism of oxygen overpotential must be deemed premature until fuller and more reliable experimental observa- tions a,re available ; in particular it is very desirable that in future experi- mental work due allowance should be made for the great influence of time of polarisation and that observations should be made over extensive ranges of c.d.Such suggestions as to the origin of oxygen overpotential which have been made have been put forward largely by analogy with the more thoroughly investigated case of hydrogen overpotential ; thus it has been suggested that the rate-determining stage is the neutralisation of the hydroxyl i0n,~3 the heterogeneous combination of oxygen atoms to give 87 8. EZektrochem. 1904 10 776 ; E. Muller and A. Scheller 8. anorq. Chem. 1905 48 112 ; see also N. A. Isgarischev and D. V. Stepanov 8.Elektrochem. 1924 30 138. 88 F. Foerster loc. cit. ref. (81) ; T. Onoda Zoc. cit. ref. (82) ; F. P. Bowden Zoc. cit. ref. (83) ; V. A. Roiter and R. B. Yampolskaya 8. physikal. Qhem. 1937 9 763 ; H. P. Stout Faraday SOC. Discussion 1947 1 246. B9 F. P. Bowden and H. P. Stout Zocc. cit. ref. (88). 90 E. Newbery Proc. Roy. Soc. 1927 A 114 103 ; V. V. Picheta J . Gen. Chem. Russia 1931 1 377 ; J. A. V. Butler and G. Armstrong Trans. Faraday SOC. 1933 29 1261 ; A. L. Ferguson and H. Bandes Tram. Electrochem. SOC. 1942,81 preprint 11. 01 See e.g. G. Grube and B. Dulk 2. Elektrochem. 1918 24 237; S. Glasstone J. Amer. Chem. SOC. 1925 47 940. 92G. Grube and J. Baumeister 2. Elektrochern. 1924 30 322; F. P. Bowden Trans. Faraday SOC. 1931 27 505 ; J. P. E. Duclaux Compt. rend. 1935 200 1838.93 R. W. Gurney Zoc. cit. ref. (65) ; see also R. Audubert and E . T . Verdier Compt. rend. 1941 213 870; R. Audubert Faraday SOC. Discussion 1947 1 72, HICKLINGF ELECTRODE PROCESSES 119 gaseous oxygen,Q4 the evaporation of oxygen dipoles from the anode sur- f a ~ e ~ 5 and the transfer of a proton from a water molecule on the anode surface to one in the electrolyte.68 Until further experimental data are available any detailed treatment of mechanism would seem futile but some general points appear worthy of consideration. Thermodynamic data now available from spectroscopic studies permit the ready calculation of the standard potentials of the hydroxyl radical and of atomic oxygen which may be intermediates in the anodic reaction and for these species at 1 atm. pressure the values found are + 2-01 and + 1*60v.respectively on the hydrogen scale in the vicinity of room temperature.85 The maximum overpotentials to which these species could give rise at the anode would therefore be about 1.6 and 1.2 v. respectively. The former value is adequate to cover the range of overpotentials experi- mentally observed but the latter would not account for the very high overpotentials a t gold and platinum at high c.d.s. Hence it would appear that if the slow stage in the anode reaction is the combination of hydroxyl radicals the magnitude of the observed overpotentials could be explained. The obvious way in which it might be supposed that such radicals react a t the anode is 20H --+ H20 + 0 Experimental observations on the disappearance of hydroxyl radicals in dissociated water vapour suggest however that this does not take place to any appreciable extent in the gaseous state since attempts to detect atomic oxygen have been unsuccessful,g6 and it seems likely therefore that it would not occur at an anode.In this connection the work of W. G. Berl 97 is of great importance. He has shown that a completely satisfactory rever- sible oxygen electrode can be set up experimentally by bubbling oxygen through a porous carbon electrode immersed in alkaline solutions of hydrogen peroxide. O2 + H,O + 2e + OH- + H0,- and the electrode satisfies all the criteria of reversibility ; for instance the potential varies with the activities of the reacting species in accordance with a Nernst equation passage of quite large currents does not displace the potential and the electrode reaction is in accordance with the postulated equation and the standard potential of - 0.04 v.is in fair agreement with that calculated from free energy data. Since Berl's oxygen electrode involves in one direction the formation of hydroxyl ions from oxygen and water and in the other direction the discharge of hydroxyl ions and since it is completely reversible it suggests that there is no intrinsic slowness in the neutralisation or formation of the hydroxyl ion. It may well be there- fore that at an anode it is the difficulty of formation of the HO radical followed by 2 0 -+ 0 > . ' (1) The potential corresponds to the equilibrium 94 See F. Foerster op. cit. ref. (79) p. 335. 9 5 F. P. Bowden loc. cit. ref. (83). 96 Cf. W. H. Rodebush et al. J . Chem. Physics 1933,1 696 ; 1936,4,293 ; J .Amer. 37 Trans. Electrochem. SOC. 1943 83 253. Chem. Soc. 1937 59 1924; 0. Oldenberg J . Chem. Physics 1935 3 266. 120 QUARTERLY REVIEWS with which the hydroxyl radical can react which is the cause of over- potential and it has been found experimentally 85 that the addition of small amounts of hydrogen peroxide to the electrolyte causes an immediate drop in oxygen overpotential. Two ways in which hydroxyl radicals might interact at an anode are 1 - * (11) OH + O H -b H,O OH + H,O -b H,O + HO OH +HO -b H,O + O or reaction might take place between a free hydroxyl radical and one adsorbed upon the oxidised metal surface thus 1 . * (111) Me0 + OH MeO-*OH MeO...-OH + OH -+ Me + H,O + 0 followed by re-oxidation of the metal but these mechanisms must at present be regarded as entirely speculative.VI. Electrolytic Oxidation and Reduction Processes Anodic oxidation and cathodic reduction reactions are very numerous and varied in character and it is possible here only to indicate some of their outstanding feat~res.~8 They may be divided into two classes involv- ing reactions which are thermodynamically reversible and irreversible respectively. In the first category the process takes place at a definite potential almost the same as the reversible oxidation-reduction potential of the system and provided this is not close to the oxygen or hydrogen evolution potential and that the c.d. is below the limiting value as deter- mined by diffusion of the depolariser to the electrode the current efficiency is largely independent of experimental variables such as the nature of the electrode material temperature and the addition of foreign substances to the electrolyte.Most oxidations and reductions of inorganic ions involving a simple one- or two-electron transfer fall into this category as illustrated by the Fe ++/Fe+++ Fe( CN) --/Fe( CN) -- and Tl+/Tl+++ systems and the working electrodes may here be conveniently regarded as merely removing or supplying electrons respectively. When however the electrode reaction is thermodynamically irreversible as seems to be the case with some inorganic and organic ions and most undissociated organic substances the position is very dserent. In this case the potential of the working electrode may not be dependent in any simple way upon the depolariser which is being oxidised or reduced and the current efficiency of the process may vary in a very complex manner with alteration of experimental factors ; in many of these reactions it seems very probable that some chemical oxidising or reducing agent primarily formed at the electrodes brings about the process observed.A particular type of anodic oxidation which has attracted much atten- tion in recent years is that which involves the polymerisation of anions. 98 See S. Glasstone and A Hickling op. cit. ref. (26) for full account and complete references up to 1935. HICKLING ELECTRODE PROCESSES 121 Examples of this kind of reaction are the formation of tetrathionate from thiosulphate ions 2s20,- - - 2e + S,O,- of persulphate from sulphate ions 2SO,- - 2e + S20,- of ethane and carbon dioxide from acetate ions (Kolbe reaction) ZCH,*CO*O- - 2e + C,H + 2C0 and of diethyl succinate and carbon dioxide from ethyl malonate ions (Crum Brown-Walker synthesis) Such reactions can be formulated electronically as above and at first sight it seems reasonable to suppose that they take place by complete or partial neutralisation of the ions at the anode followed by interaction of the radicals produced.Where the anodic oxidations can be brought about in non- aqueous solutions 99 the phenomena observed are in reasonable agreement with this view although the anodic potentials are much higher than those calculated from free-energy data implying that some stage in the reactions is slow. In aqueous solution however a very different state of affairs is manifest ; here the current efficiency is very markedly dependent upon the nature of the anode material and upon temperature and can be greatly affected by the addition of foreign substances to the electrolyte; further- more side reactions leading to different products frequently occur.S. Glasstone and A. Hickling loo studied the thiosulphate oxidation and found that small amounts of substances sharing the common property of being catalysts for hydrogen peroxide decomposition if added to the electro- lyte brought about a very great diminution in the efficiency of tetrathionate formation in some cases reducing it to zero and in general there appeared to be a close correlation between the current efficiency and the likely stability of hydrogen peroxide under the experimental conditions. Mainly on this basis they put forward a theory of electrolytic oxidation which has been applied to a large number of anodic oxidations.101 According to this view in aqueous solution hydroxyl ions are considered to discharge at low potentials to form hydroxyl radicals which combine irreversibly to give hydrogen peroxide.This may then react with a depolariser if present bringing about its own characteristic oxidation or it may decompose to give oxygen and water thus 2C2H,*O*CO*CH2*COO- - 2e --+ C2H,*O*CO*CH2*CH2*CO*OC2H + 2C0 Oxidation of depolariser H20 + *O2 20H- - 2 e -+ 20H + H202 7 Different types of oxidation may be brought about by products arising from the decomposition of hydrogen peroxide such as oxygen or metallic oxides Og Cf. S. Glasstone and A. Hickling J . 1936 820 ; A. Hickling and J. V. Westwood J. 1939 1109; S.Glasstone V. V. Barr and B. 0. Heston Trans. Electrochem. Soc. 1943 84 preprint 10. lol See S. Glasstone and A. Hickling Chem. Reviews 1939 25 407 for general account ; also Trans. Ekctrochem. Soc. 1939 75,333 ; A. Hickling and 8. H. Richards J . 1940 256 ; A. Hickling and F. Rodwell J . 1943 90. loo J . 1932 2345 2800. 122 QUARTERLY REVIEWS formed with the anode material. The anode potentials which often display peculiar features,lo2 are considered to be set up indirectly by the accumula- tion of oxygen a t the anode or in some cases by the formation of electro- motively active radicals in the oxidation reaction. This hydrogen peroxide theory has been very successful in giving a detailed interpretation of the features of a large number of electrolytic oxidations but it meets with a number of general difficulties and has been adversely criticised by several authors.103 For instance it is not always possible to imitate the results of electrolytic oxidation by chemical oxidation with hydrogen peroxide and hydrogen peroxide itself tends to be destroyed a t an anode ; furthermore it is not clear how the theory in its original form can be connected with the mechanism of oxygen overpotential.Other suggestions which have been made are that the effective oxidising agent a t an anode is the hydroxyl radical,1O4 or atomic oxygen,lo5 but little attempt has yet been made to apply these ideas to the detailed explanation of electrolytic oxidation. Irreversible eIectrolytic reduction processes mainly involve undissociated organic substances although some inorganic reactions such as the reduction of nitric acid and nitrates to hydroxylamine and ammonia and of sulphites to hyposulphites (dithionites) are probably irreversible.In general it is found that the most important factor in determining the course of a cathodic reduction is the cathode potential,lo6 and usually reduction occurs most readily and is most intense a t cathodes of high overpotential such as mercury and lead and is least a t cathodes of low overpotential such as platinum and nickel although this is the reverse of the catalytic influence of these metals in chemical hydrogenation processes. It has also been shown that poisoning of low-overpotential cathodes will sometimes improve their efficiencies as reducing agents. lo' The mechanism of irreversible electrolytic reduction has generally been regarded as involving the formation of atomic hydrogen which reacts chemically with the depolariser and this simple view seems to be in good agreement with experimental observations in most cases.Thus if the substance to be reduced is represented as RO and the reduction consists in the removal of oxygen the possible cathodic reactions would be 2H++2e -+ 2H Hence the current efficiency of the reduction process would be determined 102 Cf. A. Hickling Faraday SOC. Discussion 1947 1 227. 103 0. J. Walker and J. Weiss Trans. Farday Soc. 1935 31 1011 ; J. A. V. Butler and W. M. Leslie ibid. 1936 32 435 ; W. D. Bancroft Trans. Electrochem. SOC. 1937 '41 195; M. Haissinsky Faraday SOC. Discussion 1947 1 254. 104 Cf. A. Klemenc 2. physikal. Ghem. 1939 185 1 ; S. Glasstone Trans.Electro- chem. Xoc. 1943 84 preprint 10. 105 Cf. F. Fichter J . Xoc. Chem. Id. 1929 48 3 2 5 ~ 341r 3 4 7 ~ . 106 See e g . F. Haber 2. Elektrochem. 1898 4 506 ; F. Haber and K. Schmidt 2. physikal. Chem. 1900 32 271 ; N. Kobosew and N. I. Nekrassow loc. cit. ref. (56). lo' W. M. Monblanove N. Kobosew and P. S. Phillippovioh Acta PhySicochim. U.R.S.S. 1939 11 317. HICKLING ELECTRODE PROCESSES 123 by the competition between the latter and hydrogen evolution and any factor tending to raise the hydrogen overpotential would favour the reduc- tion reaction. Thus a t high overpotential cathodes the concentration of atomic hydrogen is high according to this view and the combination of hydrogen atoms is slow and hence reduction tends to be more efficient and intense. Most of the phenomena of cathodic reduction can be qualitatively explained on this theory.This mechanism however presupposes that it is tho combination of hydrogen atoms which is the slow stage responsible for hydrogen overpotential. If the rate-determining stage is that involving the discharge of a hydrogen ion or the transfer of a proton to the cathode surface it is by no means clear why high overpotential cathodes should be such good reducing agents. The reduction processes can of course be formally represented without involving atomic hydrogen as RO +2H+ + 2 e + R +H,O and it might be suggested that the depolariser is adsorbed on the cathode surface and becomes negatively charged and then reacts directly with hydrogen ions thus RO- +2H+ + R +H,O The competing processes might then be the transfer of a proton to the metal cathode leading to hydrogen evolution and the transfer to an adsorbed molecule of depolariser resulting in reduction.Although there is nothing intrinsically improbable in such a mechanism it can only be regarded at present as highly speculative and the atomic hydrogen view undoubtedly affords the more generally useful practical theory. VII. Cathodic Deposition and Anodic Dissolution of Metals It is possible here only to touch very briefly upon certain limited aspects of metal deposition and dissolution. The mechanism of metal deposition has generally been considered 108 as consisting of two main stages ( a ) the discharge of the metal ions and ( b ) their incorporation in the metal lattice although some authors 109 have regarded these processes as occurring simultaneously the ions being dis- charged a t favourable points on the cathode surface requiring the least discharge potential.If either of the processes (a) or ( b ) is slow then over- potential might be expected to occur. In practice it is found that when concentration polarisation is allowed for the cathodic deposition of a metal usually takes place very close to the reversible potential implying that all stages in the process are rapid and occur without difficulty. An exception to this is afforded by the transition elements iron cobalt and nickel where appreciable overpotentials are observed even at very low c.d.s.110 This is lo* For useful reviews of the mechanism of metal deposition see L. B. Hunt J. Physical Chem. 1932 36 1006 2259; F. Muller 2. Elektrochem. 1937 43 812; J.A. V. Butler op. cit. ref. (35) p. 166 ; G. E. Gardam Faraday SOC. Discussion 1947 1 182. lo9Cf. W. Blum and H. S. Rawdon Trans. Electrochem. Soc. 1923 44 397; K. Frolich and G. L. Clark 2. Elektrochem. 1925 31 649. lroA. Schweitzer ibi&. 1909 15 602; R. Schildbach ibid. 1910 16 967; F. Foerster Abh. Bunsen Ges. 1909 No. 2 ; S. Glasstone J. 1926 2887; F. Foerster 124 QUARTERLY REVIEWS shown by the figures in Table I11 for the deposition overpotentials from N-SOhtiOnS of the metal sulphates a t various temperatures. TABLE 111 Metal deposition overpotentials (voEts) F e . . . 0.22 0.03 0.00 Co . . . 1 0.28 Ni . . . . 1 0.34 I The overpotentials are very markedly reduced by rise of temperature but do not appear to vary very much with c.d. Various suggestions have been made as to the nature of the slow stage responsible for these over- potentials.For instance it has been suggested that it may be the initial discharge of the ions,ll2 the conversion of an unstable form of the metal initially deposited into the stable form,113 and the incorporation of the metal ions after discharge into the crystal lattice,l14 and many of the theories regard the hydrogen which accompanies the deposition of these metals as having an important retarding influence. In the present state of the subject it does not appear possible to arrive at any final conclusion. Very small overpotentials have been reported 115 in the deposition of metals such as silver copper lead cadmium and zinc from their salt solu- tions and'in general it has been found that the overpotential is proportional to the c.d.while this is low ; with higher c.d.s the overpotential increases less readily and may tend to a logarithmic dependence upon c.d. Attempts have been made to correlate the dependence with particular slow stages in the electrode reaction but it appears difficult to arrive at a definite conclu- sion.116 Where the cathodic deposition of a metal takes place on an electrode of a different kind initially a substantial overpotential may be necessary,l17 but this usually decreases greatly as soon as a thin film of the electrodeposited metal is formed. When a metal is made anodic in a suitable aqueous solution dissolution and K. Georgi 2. physikal Chem. Bodenstein Festband 1931 453 ; N. Thon Compt. rend. 1932 197 1312; V. Roiter and V. Jusa Acta Physicochim. U.R.S.S.1936 4 135; V. Roiter V. Jusa and E. S. Poluyan ibid. 1939 10 389. 111 S. Glasstone Zoc. cit. ref. (1 10). 112 See e.g. F. Foerster 2. Elektrochem. 1916 22 85 ; N. Thon Zoc. cit. ref. (110). 113 See e.g. V. Kohlschutter Trans. Electrochem. Soc. 1924 45 229 ; S. Glasstone Zoc. cit. ref. (110). 114 T. Erdey-GrGz and M. Volmer 2. physikal. Chem. 1931 157 165; see also L. B. Hunt Trans. Electrochem. SOC. 1934 65 413. 116 T. Erdey-Griiz and M. Volmer Zoc. cit. ref. (114) ; T. Erdey-Griiz 2. physikal. Chem. 1935,172 157 ; T. Erdey-Grbz and E. Frankl ibid. 1936,178 266 ; T. Erdey- Gruz and R. Kardos ibid. p. 255; 0. A. Essin and A. Levin J. Gen. Chem. Russia 1936 6 1539; 0. A. Essin L. Antropov and A. Levin Acta Physicochim. U.R.S.X. 1937 6 447. 117 T. Erdey-GrGz and M. Volmer 2. physikal.Chem. 1931 157 182 ; T. Erdey- Gruz and H. Wick ibid. 1932 162 63. 116 Cf. J. A. V. Butler Zoc. cit. ref. (108). HICRLING ELECTRODE PROCESSES 125 usually takes place at potentials only slightly more positive than the revers- ible value when concentration polarisation is avoided and the reaction can be regarded simply as the passage of metallic ions from the electrode to the solution; iron cobalt and nickel anodes are again exceptional in showing appreciable irreversibility in the ionisation process. Under appropriate conditions however most metallic anodes may become passive and metal dissolution may be largely replaced by some other process usually oxygen evolution. The conditions governing the onset of passivity have been largely clarified by the systematic investigations of W.J. Miiller and his co-workers and of Hedges.118 In general it appears that in the initial stages of passivity the corrosion product usually a metallic normal or basic salt tends to accumulate in the vicinity of the anode and eventually may separate out upon the electrode surface; when this happens the covered portions of the anode are shielded and the effective c.d. at the exposed parts is raised with corresponding increase of potential until hydroxyl-ion dis- charge can occur. This may lead to the production of a thin invisible oxide film if the metallic oxide is a protective one and the film impermeable to metallic ions which grows under the previously formed salt film the latter stripping off when oxygen evolution commences leaving the anode in a clean but passive state.The presence of an oxide film upon passive anodes has been confirmed in more recent work using charging-curve methods under conditions such that passivity sets in almost at once. Thus with platinum and gold anodes,l19 which appear inactive in most solutions it has been shown that oxide formation begins a t the reversible Pt/PtO and Au/Au,O potentials and oxygen evolution only commences when an approximately unimolecular film of oxide has been formed ; with gold anodes in chloride solutions the onset of oxide formation is apparently dependent upon the depletion of chloride ions in the vicinity of the anode.120 With anodes of nickel silver copper and iron in alkaline solution again the passivity seems to be accompanied by the formation of an oxide film the initial thickness when oxygen evolution first begins varying with the metal and with experi- mental conditions.121 An interesting anodic phenomenon which has attracted much attention is that of electrolytic polishing.This can be brought about at many anodes in the unstable potential region immediately before oxygen evolution ; it is apparently a differential dissolution effect projecting points upon the metal surface undergoing particularly rapid corrosion but in spite of much investigation its precise mechanism is still obscure. 1z2 l1* For reviews and references see W. J. Miiller Tmns. Farachy Soc. 1931 27 736 ; E. S. Hedges " Protective Films on Metals " 1932 Chap. 6 ; U. R. Evans " Metallic Corrosion Passivity and Protection " 1946 Chap. 1. ll9 Cf. A. Hickling Trans. Farday Soc. 1945, 41 333; 1946 42 518.120 Cf. J. A. V. Butler and J. D. Pearson ibid. 1938 34 806. 121See e.g. A. Hickling and J. E. Spice ibid. 1947 43 762; A. Hickling and D. Taylor ibid. 1948 44 262 ; Faraday SOC. Discussion 1947 1 277 ; B. Kabanov R. Burstein and A. Frumkin ibid. p. 259. 12aFor full reviews with references to the extensive literature see P. A. Jacquet Proc. 3rd Int. Electrodep. Conference 1947 3 ; R. E. Halut ibid. p. 16.
ISSN:0009-2681
DOI:10.1039/QR9490300095
出版商:RSC
年代:1949
数据来源: RSC
|
2. |
Kinetics of thermal addition of halogens to olefinic compounds |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 126-145
P. B. D. de la Mare,
Preview
|
PDF (1618KB)
|
|
摘要:
ICINETICS OF THERMAL ADDITION OF HALOGENS TO OLEFINIC COMPOUNDS By P. B. D. DE LA MARE M.Sc. PH.D. (TEMPORARY ASSISTANT LECTURER IN CHEMISTRY UNIVERSITY COLLEGE LONDON) THE literature concerning the mechanisms of addition of halogens to olefins is involved and confusing. Complex kinetics and the occurrence of side reactions have considerably hindered experimental study but a certain clarification has of recent years become apparent since it has been realised that various mechanisms are available for such additions. The contrasting and to some extent contradictory opinions expressed in recent reviews 1s emphasise however the- lack of general agreement. It has been thought useful therefore to attempt to survey the situation as it now appears. The discussion will be confined mainly to ‘‘ thermal ” (non-photochemical) additions in hydroxylic solvents ; for although the products of the reaction are simpler in non-dissociating solvents such as carbon tetrachloride the kinetics are on the whole more difficult to interpret and furthermore the experimental work is less detailed and therefore does not permit extensive generalisation.1. The General Nature of the Addition Frocess.-The most familiar addition processes with which we shall be concerned involve electrophilic attack by the halogen which must become a t least partly polarised in the sense Br- - -Br. Two theoretical considerations would lead one to expect that the positively rather than the negatively charged end of the halogen molecule would usually initiate addition. First carbonium cations seem to be more stable entities than carbanions whereas the bromide ion is better characterised than the Br+ ion.The energy of the transition state leading to addition should therefore be made less (and the addition should be facilitated) if in the transition state any incipient carbanionic centre is effectively destroyed by the developing positive bromine. Secondly nucleophilic attack by negative ions on the ethylenic carbon atoms is hindered by the screen of unsaturation electrons which are themselves vulnerable to electrophilic reagent^.^ The experimental evidence in favour of the electrophilic character of halogen additions has been sumrnarised by Williams. Various workers have shown 49 5 6 that halogens in aqueous and alcoholic solutions react with olefins by a two-stage process wherein either the anion derived from the reagent or any extraneous nucleophilic anions or even the weakly d+ d- 1G.Williams Trans. Faraday Soc. 1941 37 749. S . V. Anantakrishnan and R. Venkataraman Chem. Reviews 1943 33 27. Cf. A. G. Catchpole E. D. Hughes and C. K. Ingold J. 1948 8. A. W. Francis J. Amer. Chem. SOC. 1925 47 2340. E. M. Terry and L. Eichelberger ib.id. p. 1067. * P. D. Bartlett and D. S. Tarbell ibid. 1936 58 466. 126 DE LA MARE THERMAL ADDmON OF HALOGENS ETC. 12 7 nucleophilic solvent molecules may complete the addition thus CH,:CH + Br -+ CH,-CH,Br + Br- ; + f CH,-CH,Br + Br- -+ CH,Br*CH,Br , + C1- --+ CH,Cl*CH,Br , + H,O -+ CH,(OH)*CH,Br + €€+ Yet it might be argued 7 that the reaction may proceed in the reverse way e.g. in the presence of sodium chloride by a nucleophilic attack by the chloride ion followed by interaction with a bromine molecule with the resulting liberation of a bromide ion.Thus the orientation of the products of the unsymmetrical addition of iodine chloride (e.g. Me*CHCl*CH,I from propylene and CH,Cl*CHI*CO,K from acrylic acid 8 ) is not an unambiguous diagnostic of the course of the reaction. On the other hand the principles which govern the production of isomeric (1 2 and 1 4) mono-addition products from halogens and conjugated dienes are intelligibly correlated with the facts known about the mobility of anionotropic lo suggesting that similar intermediates are involved and therefore that addition of the anionic part of the reagent is the second rather than the first atage of the addition process. Proof that the electrophilic attack by the reagent usually determines the course of the reaction is derived from the effects of substituent groups on the rates of addition to substituted ethylenes.Ingold et ~ 2 . ~ ~ 9 l2 showed by a competition method that electron-donating groups (such as CH or C,H,) favour the addition of bromine which is inhibited by electron-withdrawing groups (such as Br or C0,H). Cor- respondingly an actual charge on the molecule is important in determining the rate. Thus the ion CH,:CH*CO,- is more reactive than the undis- sociated acid l3 ; and when an otherwise reactive unsaturated group is part of a positive ion as in R,N*CH,*CH:CH-NR, no addition of bromine will take place.ll l4 Further discussion and exemplification of the detailed theory due to Ingold and Ingold l1 can better be given below since the reagent and mechanism operating in their experiments are not yet certain.2. Second-order Halogen Addition.-A. Berthoud and M. Mosset showed l5 that in water bromine addition to maleic and fumaric acids proceeds with rate proportional to [A][X,] where A is the unsaturated compound and X the halogen. Kinetically similar are the additions of iodine to ally1 alcohol in water,15 of bromine to stilbene in methyl alcoho1,a and of bromine to cis-cinnamic acid and to acrylic acid in aqueous acetic acid and in water.13 Trihalide ions were found to be less reactive than 3- + I. Cf. R. A. Ogg ibid. 1935 57 2727; 1939 61 1946. H. Burton and C. K. Ingold J . 1928 910. * C. K. Ingold and H. G. Smith J . 1931 2742. lo P. B. D. de la Mare E. D. Hughes and C. K. Ingold J .1948 17. l1 C. K. Ingold and E. H. Ingold J . 1931 2354. la S. V. Anantakrishnan and C. K. Ingold J . 1935 984 1396. l3 P. W. Robertson N. T. Clare K. J. McNaught and G. W. Paul J . 1937 335 ; K. Walker and P. W. Robertson J. 1939 1515. l4 C. K. Ingold and E. Rothstein J . 1931 1666. J . Chim. physique 1936 33 272. I 128 QUARTERLY REVIEWS the parent halogen molecules ; in studying additions to organic acids it was necessary to ensure by adding mineral acid in order to suppress ionisa- tion that the addition did not proceed via the very reactive unsaturated carboxylate ion. I n one of the above investigations that of Bartlett and Tarbell careful kinetic study eliminated the possibility that in methyl alcohol as solvent the true brominating agent was MeOBr an erroneous view held by K.Meinel ; l6 the evidence that such compounds formed by equilibrium with the solvent (X + R*OH + H+ + X- + RO*X) are not generally intermediates in addition reactions in hydroxylic solvents is given by Williams. E. P. White and P. W. Robertson l7 investigated the kinetics of chlorine addition to cis-cinnamic acid and other olefinic compounds in acetic acid as solvent. The reaction followed a second-order course as has been confirmed in later investigations by Robertson and co-workers for olefins of widely varying reactivity. The second-order velocity constants obtained in these investigations are collected in the Appendix and are quoted without further reference hereafter. Ingold and Ingold l1 had observed that when two or more polar groups are attached to an ethenoid centre their interaction may complicate the resulting effect on the rate.When one such group only is varied however the changes in rate are consistent with the view that electron accession to the double bond increases the rate of reaction. Thus the methyl group when attached to the ethylenic carbon either directly or through a conjugated system increases the rate of chlorine addition by a factor of more than 10; similarly electron-with- drawing groups reduce the rate the nitro-group producing a change in velocity of addition by a factor of lo4 Parent compound. R = p-Me. H. p-c1. m-NO,. P-NO*. R*C,H,*CH:CH*COPh * k = ca. 800 61 23 0.23 - R*C,H,*CH :CH-CO,H k = 103 4.9 I 0.01 1 0.0049 Two methyl groups are superior to a phenyl group in promoting reactivity the latter being more effective than a single methyl group CMe :CH*CO,H trans -CHP h :CH*CO,H trans- CHMe :CH*CO,H k - 51 4.9 0.62 CH*CO,H CH*CO,H CH :CH*CO,H II 0.018 0.000 1 1 l6 Annalen 1935 516 231.18 I. D. Morton and P. W. Robertson J. 1945 129. 19 P. B. D. de la Mare and P. W. Robertson ibid. p. 888. 20 I. Hartmann and P. W. Robertson ibid. p. 891. 21 I. Ting and P. W. Robertson J . 1947 628. 2 2 B. E. Swedlund and P. W. Robertson ibid. p. 630. 23 H. P. Rothbaum I. Ting and P. W. Robertson J. 1948 980. * Substituents in the PhCO group as compared with the Ph’. group of CHPh’:CH*COPh are much less effective in altering the rate of addition-as would be expected since the carbonyl group breaks the conjugation between the substituent and the double bond. l7 J. 1939 1509. DE LA MARE THERMAL ADDITION O F HALOGENS ETC.129 The types of electron displacement which favour halogen addition to a double link are those which would favour aromatic substitution R--CH=CH Ty /-A RY/>-L- CH=CH /-x RN_/zTy \=/- \== (1.) (11.) (111.) It is not surprising therefore that the effect of R in promoting addition of chlorine to (I) and (11) is similar to that of the same substituent on the chlorination of (111) as shown by the following values of E C1 addition t o C1 substitution in p-Me0 .C,H,R relative rate 24 29 19 16 10 - - 0.07 - Sufticient evidence has been given to illustrate that electrophilic attack by the halogen molecule is the rate-determining stage of these additions. It has already been pointed out however that the intervention of solvent and of added ions in the completion of the reaction demonstrates that the whole process of addition involves a t least two stages.Neither the kinetics nor the effect of structure on the rate throws light on the exact nature of the intermediates involved. In particular it is not known whether the halogen-halogen link in the molecule which initiates reaction is broken in the first (rate-determining) stage of the process. It is commonly supposed that such a structure as (Ia) is formed by attack of halogen on the olefinic compound ; and it is now usually thought that structure (Ib) 25 contributes to the resulting resonance hybrid and accounts for the familiar trans-addition of halogen.* R = COPh. C0,Et. C0,Me. COaH. CHO. CN. NO,. S0,CI. Ph*CH:CH*R 61 10 10 4.9 1.8 0.022 0.020 0.001 f CH2*CH*CH :CH2 CH2-CHCH :CH *-:< Br Yvc< Br + Br I v Br + (IIb.) + CH,CH :CH*CH2 Br I CH=CH Br (IIIa.) (IIIb.) a4A. E. Bradfield and B. Jones Trans. Paraday Xoc. 1941 37 726. 26 I. Roberts and G. E. Kimball J. Amer. Chem. SOC. 1937 57 947. * The n-complex formulation (cf. Dewar z e ) is perhaps an equivalent met,hod of writing a complex between a Brf ion and an olefm ; but the former type of representa- tion appears to indicate more clearly that the two alternative forms Rl R3 Rl R3 \+ / and \ / c-c do not in general contribute equally to the resulting resonance hybrid. 130 QUARTERLY REVIEWS I n an extension due to Burton and I n g ~ l d ~ ~ ( ~ . lo) of the above theory the coexistence of 1 2- and 1 4-addition to conjugated systems is associated with the distribution of charge due to mesomeric electron shift in the cation of an anionotropic system.Thus the mesomeric cation involved in the reaction between butadiene and bromine would be written as a resonance hybrid of structures ( I I a ) and ( I I I a ) ; and this can then take up an anion a t either the 2- or the 4-position. Either of the ring structures (IIb) and ( I I I b ) could theoretically contribute also but since they involve different geometrical relationships of atomic nuclei they cannot both contribute appreciably to the same resonance hybrid. If however the facts neces- sitated the postulation of all four types of structure (IIa) (IIb) (IIIa) (IIIb) then this might be accommodated by assuming a tautomeric equili- brium between a hybrid of (IIa) (IIb) ( I I I a ) and one of (Ira) (IIIa) (IIIb). Although these true carbonium ions are plausible intermediates in addition reactions the known facts could equally be explained in terms of intermediates such as (IVa) and (IVb) cf* 1s 27 which also constitute a possible > y Y Br- resonance hybrid.Although a carbonium ionic centre is present in such it structure thus permitting in an ionising solvent stabilisation by solvation forces the resulting complexes are dipolar rather than fully ionic. Such a view would seem to approach that held by Sir R. Robinson,28 but is as yet neither necessitated nor disproved. A more detailed study of the products of addition as related to reaction kinetics in the presence of added salts may help to elucidate this problem. 3. Third-order Brumine Addition. -It was first established by Robertson Clare McNaught and Pau1,13 and has since been confirmed for olefins of widely different reactivity,21 23 that in the concentration region ~/40 in pure acetic acid addition of bromine to unsaturated compounds proceeds by third-order kinetics the rate being proportional to [A][Br,12.Dilution increase of temperature or dilution of the solvent with water was shown to result in the gradual dominance of a second-order reaction with a higher heat of activation. Similar behaviour was observed 29 for bromination of aromatic substances in the same solvent. The effects of substituents on the rates of reaction by this mechanism are collected as relative rates in the Appendix. For several compounds comparison may be made of the rate ratio of second-order chlorine to third- order bromine addition as follows M. J. S. Dewar J.1946 406 777. 27 H. D. C. Waters A. R. Caverhill and P. W. Robertson J. 1947 1168. as Ibid. p. 1294. P. W. Robertson P. B. D. de la Mare and W. T. G. Johnston J. 1943 276. DE LA MARE THERMAL ADDITION OF HALOGENS ETC. 131 Ph H Me Me Ph H Ph H Small variations in the rate ratio are not unexpected ; for in third-order bromination when as compared with second-order chlorination one additional molecule must be accommodated in the transition state in the vicinity of the double bond larger steric effects might be anticipated. In fact although the absolute resctivities of these compounds cover a large range the rate ratios differ by a factor of less than 2 illustrating that the electronic require- ments of the two reactions are similar. Table I exemplifies that electron- withdrawing substituents (R) retard whereas the phenyl group and more powerfully two methyl groups accelerate the addition.TABLE I Effects of substituents o n the rate of third-order bromine addition Parent compound. Relative rate of addition. __ R = CHSCl. 1 R = Hal. 1 R = COaH. CMe :CHR . . trans-CHPh:CHR . . CH,:CHR. . . . . 0.017 0.001 I I 1 At this stage attention should be drawn t o an apparent irregularity. The rate of bromine addition t o stilbene (trans-CHPh:CHPh) as compared with stFene (CHPh :CH,) is unexpectedly small although several examples have already been given (e.g. in Table I) of the considerable effect of the phenyl group in facilitating addition. Ingold and Ingold 11 gave an explana- tion of this point referring particularly to the small reactivity of tetra- * tx = t h e for 20% disappearance of halogen (reactants initially ~ / 8 0 at 24' in acetic acid).132 QUARTERLY REVIEWS phenylethylene. The second phenyl group in stilbene is conjugated with the first phenyl group which in styrene makes electrons available at the attacked olefinic carbon atom (Ph-CH=CH,). Electrons supplied by one phenyl group therefore are absorbed by the other (structures such as +<=)=CH-CH=(-=> - making a substantial contribution to the stability of the molecule) instead of becoming available for co-ordination with the attacking halogen. This effect will contribute to the relative unreactivity of any olefin in which an electron-donating group is similarly conjugated with one which can withdraw electrons. The groups CH,C1 and CH,Br are electron-attracting in character as shown for third-order bromine addition by the following comparison CH :CH*CH,*O*COPh CH :CHGH,Cl CH :CHCH,Br CH :CH(CH,Cl) CH :CHBr 0.019 0.00 11 /y /?A 14 1.6 1.0 The greater reactivity of allyl chloride than of allyl bromide (the opposite would be expected because chlorine is more powerfully electron-attracting than bromine) is not a t present understood and awaits more detailed study of the effect of halogen substituents on the rate of addition.The inclusion of vinyl bromide in the above series emphasises the importance of the inductive effect in deactivating the unsaturated compound even though as in the halogenobenzenes the electron-releasing potentiality of the halogen controls the orientation of attack by electrophilic reagents. 30 The effects of substituents (Me Br C0,H) on the rate of addition as determined by the competition method of Ingold et aZ.ll l2 were much smaller than those indicated by the above kinetic experiments.The difference is partly explained by the difference in conditions ; it is probably also relevant that bromine addition to allyl benzoate in chlorobenzene as solvent may induce addition to the much more slowly reacting cinnamic a ~ i d . ~ 1 Indeed it appears in the comparison on p. 129 that substituents influence the rate of halogen addition to ethylene more powerfully than they affect the rate of aromatic halogen substitution. The following are rates of uncatalysed addition to cis-ethylenic derivatives relative to the rate for the corresponding trans-isomer the halogen and solvent being given in parentheses cis-Dichloroethylene .. 3.0 (Clz aqueous acetic acid) cis-Cinnamic acid . 3-8 (Br2 acetic acid) 3-0 (C12 acetic acid) Maleic acid . . 2.3 (Br2 aqueous acetic acid) Citraconic acid . . 2.8 (Brz aqueous acetic acid) A greater reactivity of cis- than of trans-compounds is therefore,character- istic of these reactions in which halogen is an electrophilic reagent ; though the same is not true of addition by other mechanisms. 2521 2531. 90 M. S. Kharasch M. C. McNab and F. R. Mayo J . Amer. Chem. SOC. 1933 55 a1 P. B. D. de la Mare R. A. Scott and P. W. Robertson J. 1945 509. DE LA MARE THERMAL ADDITION OF HALOGENS ETC. 133 It is also noteworthy that electrophilic attack by halogens on olefins appears to be more rapid than similar attack on the corresponding acetylenes. Thus we have the following relative rates Second-order chlorine addition in acetic acid 4.9 2.4 Third-order bromine addition in acetic acid 0.017 0.0053 trans-Cinnamic acid.Phenylpropiolic acid. Similarly phenylacetylene as compared with styrene is very much less reactive towards bromine. No satisfactory explanation of this result seems to have been given ; it is unexpected on various grounds namely (a) in acetylenes as compared with ethylenes there must exist a greater density of unsaturation electrons available for attack ; (6) steric considerations must if anything favour the reaction of the acetylene rather than that of the olefin; (c) the addition of one halogen molecule to an acetylene is probably more exothermic than the addition of one halogen molecule to the corresponding olefin as judged at least from the related heats of hydrogenation 32 ; ( d ) the triple bond is apparently more polarisable than the double bond since in vinylacetylene attack by hydrochlor-ic acid is initiated by Hf a t the acetylenic rather than the olefinic terminal carbon atom.l0* 33 As compared with second-order halogen addition (rate cc [A][X2]) even less is known about the intermediates concerned in third-order bromine addition (rate GC [A][Br212).There is complete lack of evidence concerning whether third-order addition always produces a dibromide or whether instead the mechanism is so complex that anions or solvent molecules may still participate in a stage of the reaction which is not rate-determining. It is unlikely that Brf formed in a pre-equilibrium (e.g. Br + Br + Br+ + Br;) takes part in the reaction since then the addition of negative ions would greatly reduce the rate of addition (e.g.Brf + OAc- + BrOAc). The kinetics make it clear that the presence of hydrobromic acid or a similar catalyst is not essential for the addition of bromine despite the contrary earlier opinion held by Anantakrishnan and Venkataraman. Thus the presence of sodium acetate which should in acetic acid act as a base and reduce the rate of acid-catalysed addition actually slightly facilitates the addition e.g. of bromine to w-bromostyrene. It is not intended to imply that hydrobromic acid cannot intervene in addition; but merely that the reactions which have been described above are essentially those found in the absence of hydrobromic acid. Two possible two-stage mechanisms would lead to third order kinetics as found namely A + Br + A,Br ; A,Br + Br -+ products .* (i) Br + Br + Br ; A + Br -+ products . . (ii) The first was that favoured by White and Robertson,17 and accepted by Williams ; but the kinetics do not distinguish between the two since either J. B. Corm G. B. Kistiakowsky and E. A. Smith J. Amer. Chem. SOC. 1939 61 1868. as W. H. Carothers G. J. Berchet and A. M. Collins $bid. 1932 54 4066. 134 QUARTERLY REVIEWS would be expected to have a low temperature coefficient and would be supplanted by a concomitant second-order process on dilution or a t high temperatures. 4. Third-order Addition of Other Halogens.-The additions of iodine chloride and of iodine bromide to double bonds are the bases respectively of the Wijs and the Hanus method of determining quantitatively the unsaturation of fats oils and related compounds.Few measurements of rates and kinetics of these reactions of interhalogen compounds have however been made. White and Robertson l7 showed that with reactants in the concentration region ~ / 4 0 iodine chloride addition is a reaction of the third order in acetic acid. Hydrochloric acid retarded addition pre- sumably by formation of the unreactive HClI,. The compounds examined were undecenoic acid ally1 acetate cinnamic acid and o-methoxycinnamic acid ; extension of the measurements to include less reactive olefins would be instructive. Bromine chloride and iodine bromide additions were also found to be kinetically of the third order but the dissociation of these compounds is sufficient to necessitate correction for the concomitant addition of the more reactive halogen component.An estimate was made of the relative reactivity of the various halogens by the third-order mechanism ; it is as follows Halogen . . I IBr Br IC1 BrCl Relative rate of addition . . 1 3 x 103 104 106 4 x 108 The increasing reactivity in the series I, IBr IC1 may be associated with the increasing electron withdrawal from the point of initiation of addition ( L e . with increasing polarity in the sense 1-1 < I-& < I-Cl). It is difficult to avoid the conclusion however that the stability of the final product is partly responsible for the rate differences particularly in the unpolarised halogens themselves the additive reactivities of which increase in the order I < Br,. Quite a different situation arises in the substitution reactions between halide ions and alkyl halides; in acetone for example alkyl iodides are attacked more rapidly by iodide ions than by bromide The addition of iodine to ethylenes differs from that of other halogens in that it is reversible.N. J. Bythell and P. W. Robertson 36 isolated the forward reaction by keeping the olefin in excess and measuring only the initial stages of the reaction. The kinetics appeared to be similar to those of the addition of bromine except in that water added to the acetic acid solvent had a markedly less powerful effect in accelerating the rate of the reaction. Constitutive effects indicated that iodine acts as an electrophilic reagent ; the work of K. Nozaki and R. A. Ogg,37 however suggests that a re-examination of iodine addition in acetic acid would be desirable.5. Fourth-order Bromine Addition.-The large variation in rate with 84L. Fowden E. D. Hughes and C. K. Ingold unpublished work. sf E. D. Hughes F. Juliusberger S. Masterman B. Topley and J. Weiss J. 1935 36 J . 1938 179. 6+ 6- 6+ 8- 6+ 6- ions. Cf. 34 35 1625. 8' J . Amr. Chm. Soc. 1942 64 697 704 709. DE LA MARE THERMAL ADDITION OF HALOGENS ETC. 135 initial concentration of reactants makes it difficult to follow these reactions over a wide concentration range. Nozaki and Ogg 37 produced evidence for the existence of a reaction of the kinetic form dx/dt = EIA][Br,ln in the addition of bromine to allyl halides though B. E. Swedlund and P. W. Robertson 38 were unable quantitatively to confirm their findings. The latter authors however,22 later showed that in addition of bromine to slowly reacting olefins (e.g.m-bromostyrene ~ / 5 - ~ / 1 0 ) a reaction of kinetic order higher than 3 appears. A similar phenomenon was observed in an examination of the kinetics of bromination of aromatic compounds.29 39 The following sequence of reactions was regarded as providing a plausible explanation of these unusual kinetics Br + Br + Br ; A + Br + A,Br + Bra ; A,Br + Bra --+ products The possibility was considered that the previously described third-order additions in the concentration region ~/40 could be explained as Nozaki and Ogg apparently believed by a roughly equal contribution of second- and fourth-order addition. If this were so however there would be expected a much more rapid variation in reaction order with concentration than is actually observed.d* 29 39 6.Bromine Addition Catalysed by Halide Ions.-(a> CutuZysis by chlor- ide ions. Nozaki and Ogg observed that halide ions mag act as catalysts for bromine additions in acetic acid ; for instance addition to allyl chloride and to vinyl bromide is considerably catalysed by lithium chloride. The results were consistent with the contribution to the rate of a term - d[Br,)/dt = E[A][Br,)[Cl-] and these authors proposed that such a mechanism is of general availability for bromine addition. Subsequent investigations particularly by Swedlund and Robertson 22 38 have oonfirmed that lithium chloride does in general catalyse bromine addition to olefins as shown by the results summarised in Table 11. TABLE I1 Halide-ion catalysis of bromine addition in acetic acid Compound.CHPh :CHPh . . . . . . . CH,:CH*CR,CI . . . . . . CH,:CH*CH,Br . . . . . . CMe :CHCl . . . . . . . CHPh :CHBr . . . . . . . CH,:C(CH,Cl) . . . . . . CHBr :CH . . . . . . . . I Ratio of catalvsed rate to 1 Relative rate 1 uncatalysed 'rate reactants of third-order initially ~/80 x = 20% reaction at 24' with catalyst uncatalysed bromine i addition. - 1 LiC1 ~ / 2 5 . HBr M/20. I I I 18 1-6 1.0 0.5 0.1 0.019 0.0011 1.6 6-3 5.5 2.2 1.5 7.0 18 0-43 2.4 2.2 0-64 0.64 8.9 10.5 I Analogy with the two possibilities considered for third-order un- s8 J. 1946 131. 89 P. B. D. de la Mare and P. W. Robertson J . 1948 100 136 QUARTERLY REVIEWS catalysed addition [equations (i) and (ii)] suggests the two alternatives for additions catalysed by chloride ions A + Br + A,Br ; A,Br + C1- +products .. (iii) Br + C1- + ClBr,- ; A 4 ClBr,- + products . . (iv) Equation (iv) can the Reviewer believes hardly be significant in addition to allyl chloride for the following reason. A negative ion such as the C1Br2- ion should act as a nucleophilic reagent ; yet the relative rates of the catalysed additions to allyl chloride and to vinyl bromide require this anion to be acting as an electrophilic reagent seeing that the extent of the catalysis is much the same for these two compounds whose reactivities towards electro- philic halogen differ by three powers of 10. It is not meant to imply that the mechanism represented by equation (iv) is never available for addition of bromine. In a later section evidence will be presented of a rapid nucleo- philic attack by C1Br2- on olefinic cations such as R*CH*CH:CH*OH ; and some evidence exists 40 that attack by ClBr,- is also possible in additions to such olefins as maleic acid in which electron regression from the double link is considerable.Equation (iii) as the mechanism of additions catalysed by halide ions is consistent with the observation that chlorine addition for which the third- order uncatalysed mechanism does not seem to be available is little influenced by the presence of lithium chloride.* A difficulty arises however in that the extent of the catalysis seems to be selectively dependent on the nature of the olefin. It would have been expected that the rate of such a catalysed addition would follow and decrease less rapidly than the rate of uncatalysed addition as the structure of the olefin became progressively less favourable towards electrophilic attack by halogen and correspondingly more favourable toward attack by a nucleophilic ion.I n other words catalysis should be greatest for the least rapidly reacting olefins. The data of Table I1 indicate considerable deviations from such a simple hypothesis for example catalysis of addition to m-bromostyrene is less than that of addition either to stilbene (which without catalysts is very much more reactive) or to vinyl bromide (which is many times less reactive in addition reactions). Such considerations led Swedlund and Robertson 2 2 p 38 to propose a special mechanism for these additions; it may be on the other hand that some special steric or polar consideration affecting the rate of the attack by halide ions is causing deviations from the simple predictions of equation (iii).Further examination over a wider range of compounds of the catalytic effects of halide and other ions together with detailed know- ledge of the equilibria present in solutions of bromine in the presence of these ions seems necessary to clarify the position. The last column in Table I1 gives the rate ratios for hydrogen bromide-catalysed to uncatalysed addition of bromine in acetic acid.21 22 23 Since these are very near the limiting values for 40 P. W. Robertson and co-workers unpublished work. * Iodine addition on the other hand is catalysed by halide ions but the reaction + (b) Catulysis by bromide ions. is complicated by the considerable stability of such complex ions as CI1,-. DE LA MARE THERMAL ADDITION OF HALOGENS ETC.137 excess of hydrogen bromide it is presumed that they include negligible contributions from uncatalysed third-order addition the rate of which ( K[A] [Br,] ,) is very critically determined by the concentration of free bromine. The results are therefore consistent with the rate expres- sion - d[Br,]/dt = E[A] [HBr,] = E[A] [BrJ [Br-].37 The equilibrium HBr + Br + HBr in acetic acid lies well to the right but the equilibrium constants as determined by W. J. Jones 41 vary with concentration though they correspond roughly with the value obtained by Nozaki and Ogg 37 for the equilibrium between lithium bromide and bromine in the same solvent. The presence of hydrobromic acid or of lithium bromide in a solution of bromine in acetic acid will therefore considerably reduce the concentration of free bromine in the mixture.Bromide ions however like chloride ions should act as catalysts for additions of bromine to olefins. Accordingly as shown by the last column of Table 11 the presence of excess of hydrobromic acid actually retards the rate of bromine addition to those olefins addition to which is not much catalysed by lithium chloride ; for the other compounds both lithium chloride and hydrobromic acid act as catalysts. It would seem reasonable then to suppose that for addition to these olefins lithium chloride and hydrobromic acid have a similar catalytic function. Further evidence was given by Nozaki and Ogg who showed that hydrobromic acid and lithium bromide are equally effective as catalysts for addition to ally1 chloride and to vinyl bromide.Swedlund and Robert- son 2 2 s ,* proposed the alternative view which is also permitted by the kinetics that in addition catalysed by hydrogen bromide the electrophilic reagent is the hydrogen tribromide molecule. Their reason for adopting this view is essentially as follows. The basicity of the reagent is usually (cf. E. D. Hughes 42) regarded as the most important factor determining the nucleophilic reactivity of ions. One would therefore be tempted to predict that since the dissociation of acids increases in the order HOAc < HC1 < HBr,4 the corresponding ions should decrease in the order OAc- > C1- > Br- in their efficiency as nucleophilic catalysts for addition. The opposite is in fact the case ; for when allowance has been made for the disturbance of the concentration of free bromine in solutions containing hydrobromic acid bromide ions appear to be about 20 times as effective as chloride ions in catalysing addition of bromine ; and the catalytic power of acetate ions though perhaps not zero is certainly almost negligible as compared with that of chloride ions.There exists however some evidence that basicity does not always give a satisfactory measure of the relative nucleophilic power of anions. For example measurements of the rates of bimolecular nucleophilic reactions with bromide ions and with chloride ions of a number of substituted alkyl halides 44 45 46 indicate that bromide ion is at least 5 times as reactive as 41 J. 1911 99 392. 43 I. M. Kolthoff and A. Willmann J. Amer. Chem. SOC. 1934 56 1007. 4 4 T. A. Bither J. M. Sturtsvant and H.C. Thomas ibid. 1945 67 1562. 46 A. R. Olson and F. A. Long ibid. 1936 58 393. 413 M. J. Young and A. R. Olson ibid. p. 1167. 4 a Trans. Paraday SOC. 1938 34 185. 138 QUARTERLY REVIEWS chloride ion. Similar comparisons and conclusions can be reached in other cases,*7 though data for reactions in acetic acid do not seem to be available. The factors determining the rates even of these simple substitutions are not yet fully understood but it is satisfactory that the order of reactivity for nucleophilic substitution Br- > C1- appears to be the same as the order of catalytic efficiency in addition. The fact 37 that water added to the acetic acid solvent reduces the efficiency of halide ions as catalysts for bromine addition is t o be expected and is of no help in further elucidating the mechanism.An increase in the solvating power of the solvent should stabilise the polarised transition state of uncatalysed addition relative to the formally neutral reactants thus accelerating addition by this mechanism ; 48 but such a solvent change should still more stabilise the fully ionic halide ions and should thus diminish their catalytic activity. As compared with addition in acetic acid addition in strongly solvating solvents such as methanol should not be greatly catalysed by halide ions. 7. Nucleophilic Halogen Addition.-(a) Catalysis by sulphuric acid and by perchloric m i d . In the reactions considered so far the effects of structure on the rate of halogen addition indicate that the halogen molecule acts as an electrophilic reagent. It was first predicted by Ingold and Ingold,ll however that if the ethylene link were polarised by a sufficiently strongly electron-demanding group an addition involving nucleophilic attack by halogen might be realisable.Experimental evidence for the existence of such a mode of addition is most strikingly given by the reactions of ap-un- saturated aldehydes and ketones in acetic acid.l99 23 Addition of bromine to cinnamaldehyde (reactants M/SO) in the absence of catalysts is very slow a t first but after about 10 minutes begins to increase autocatalytically and reaches 70% in about 30 minutes. In the presence of sodium acetate (M/20) the autocatalysis is repressed and the reaction proceeds slowly without an induction period at a rate less than that of addition to cinnamic acid under the same conditions.In the presence of perchloric or sulphuric acid however addition is rapid and with acid concentration constant the initial rate of addition is proportional to the concentration of olefin and of bromine. Water reduces the rate of the acid- catalysed reaction. Perchloric acid a stronger acid in this solvent is more effective as a catalyst than sulphuric acid and nitric acid relatively a very weak acid has very little catalytic power. Corresponding results have been obtained for addition to such ap-un- saturated ketones as benzylideneacetophenone CHPh :CH*COPh 23 the structure of which precludes substitution adjacent to the carbonyl group. The reactions of crotonaldehyde and of acraldehyde are similar but are too rapid for detailed investigation. Addition of bromine to @?-unsaturated acids such as acrylic and maleic acids to maleic anhydride to w-cyanostyrene and to w-nitrostyrene is also catalysed by these mineral acids but to a much smaller extent.Sulphuric acid does not catalyse addition t o ally 47 Cf. P. B. D. de la Mare Ph.D. Thesis University of London 1948. 48 Cf. E. D. Hughes Traw. Faraday SOC. 1941 37 763. DE LA MARE THERMAL ADDITION OF HALOQENS ETC. 139 benzoate and only slightly catalyses addition to ethyl cinnamate and to ally1 chloride. TABLE I11 Acid-catalysed addition of bromine in acetic acid (reactants M/SO) Compound. H,SO,-catalysed bromine addition kB (HBS04 = M/80). H,SO,-cstalysed chlorine addition kl (HsS04 = M/320). Uncatalysed chlorine addition Benzylideneacetophenone . Ethyl cinnamate . . . Cinnamaldehyde . . . trans-Crotonic acid.. . Crotonaldehyde. . . . o-Nitrostyrene . . . . 32 27 > 1000 + 0.013 * ca. 0.009 * ca. 0.005 * 61 10 1.8 0.62 0-41 0.020 * Corrected by deduction of the uncatalysed rate. In Table 111 some data for acid-catalysed bromine and chlorine additions are compared with velocity constants for uncatalysed chlorine addition to emphasise that the effect of structure on the rate of addition is quite different for the two reactions. The very rapid rate of addition to unsaturated aldehydes and ketones as compared with the corresponding acids is suggestive of addition initiated by nucleophilic halogen. The initial polarisation of the olefin is however clearly not the only factor concerned in determining the reaction rate since co-nitrostyrene as compared with cinnamaldehyde is less reactive under conditions of acid catalysis despite the greater electron-withdrawing capacity of the NO as compared with the CHO group.The essential rate-determining stage of acid-catalysed addition is thought therefore to involve nucleophilic attack by halogen on the ion t produced by addition of a proton t o a group conjugated with the double bond as in the following sequence l9 0 OH OH -+ products 4- / Br / I \ + R*CH*CH:C / \ R-CH:CH*C + H+ + R*CH*CH:C +Br-Br R’ \ R’ R’ The stage subsequent to proton-addition is thought to be complex since the rate of addition is not proportional to the stoicheiometric concentra- tion of acid catalyst. The proposed mechanism accounts for the observed kinetics (including the effect of water which acts as a base and thus reduces the proton-donating power of the acid catalyst) and is also consistent with the effect of structure on the rate of addition.The electronic requirements of the preliminary addition of a proton to the olefinic derivative are clearly opposite to those of the subsequent nucleophilic attack by halogen and p Such an ion as an intermediate in addition was first suggested by Nozaki and 0gg,s7 but the subsequent stages of their mechanism involved a halide ion which the Reviewer believes to have as shown below quite a different function in such reactions. 140 QUARTERLY REVIEWS therefore the effect of structure on the rate of acid-catalysed addition is complicated. The small magnitude of catalysis in addition to m-nitro- styrene and to ap-unsaturated acids as compared with addition to the corresponding oc#l-unsaturated aldehydes and ketones is almost certainly determined by the greater basicity of the latter compounds.The effect of changing R and R’ (in R-CH:CH-COR’) has been investigated 23 and reveals further the balanced electronic requirements of the two important stages of this type of addition. In certain cases however the stage involving Qucleophilic attack by halogen predominates resulting in a reversal of the normal order of reactivity found when halogen acts as an electrophilic reagent. Thus in Table I11 it is shown that acid-catalysed addition to crotonaldehyde is more rapid than similar addition to cinnamaldehyde. Correspondingly p-xyloquinone as compared with benzoquinone with sulphuric acid as a catalyst adds bromine much less rapidly.23 Chlorine addition as is shown in Table 111 is less sensitive than bromine addition to the presence of acid catalysts.The chlorine molecule is less polarisable than the bromine molecule and since nucleophilic attack involves polarisation of the halogen in the sense opposite to that usually effected in this type of reaction it is perhaps natural that nucleophilic as compared with electrophilic addition is more critically determined by the polarisability of the halogen. Although sodium acetate usually eliminates autocatalysis of halogen addition in acetic acid its presence does not always ensure as low a rate of addition as would be expected from the structure of the olefin. Particularly is this the case for such olefins as acraldehyde addition to which is very sensitive to acid catalysis. The reason for this anomaly is not known for certain.It is possible that the solvent itself acts as a source of protons to catalyse the addition ; alternatively anion catalysis may be important in electrophilic addition to these compounds. (b) Catalysis by haZogen acids. As catalysts for bromine addition hydrobromic and hydrochloric acids are much more efficient than would be expected from the order of acid strengths in acetic acid namely 43 HClO > HBr > H,SO > HC1 > HNO Thus the following are relative rates of addition of bromine to cinnamal- dehyde reactants M/SO catalyst ~ / 3 2 0 HClO HBr H2SO4 HC1 mo 7.3 ca. 300 3.4 24 ca. 0-8 It seems probable that the great catalysis by hydrobromic acid and by hydrochloric acid of halogen addition is attributable to the formation of complex ions (e.g.Brg ClBr; Cl;). These act as nucleophilic reagents to effect addition of halogen to the olefinic cation (e.g. CHPh*CH:CH*OH) for which task they are as would be expected more active than halogen molecules. The possibility is not excluded that undissociated hydrogen tribromide and similar molecules are the effective reagents. It is unlikely however that the olefin itself (rather than the carbonium ion formed by addition of a proton) is attacked ; for Nozaki and Ogg 37 showed that for + DE LA MARE THERMAL ADDITION OF HALOGENS ETC. 141 this type of addition (exemplified in their experiments by bromine addition to maleic acid) the proton plays a specific part in the reaction hydrobromic acid being much more effective than lithium bromide as a catalyst.* Furthermore as in addition catalysed by sulphuric acid addition to w-nitro- styrene catalysed by hydrogen bromide is much less rapid than similar addition to cinnamaldehyde though the former should theoretically be more susceptible to attack by a nucleophilic reagent.The effects of structure on the rates of addition reactions catalysed by hydrobromic acid have confirmed and extended the conclusions obtainable from the similar reactions catalysed by sulphuric acid. Because the addition of a proton to a basic centre in the olefin is favoured by the presence of suitably placed electron-donor groups in the molecule whereas the second stage of the reaction namely nucleophilic attack by Brg may be decelerated by the same structural influences the effects of substituents on the rates of these additions are often complex.There are however a number of examples cf. l8 which demonstrate that the normal order of reactivity found when electrophilic attack by the halogen is the dominant mechanism may become reversed under conditions of catalysis by hydrobromic acid. Thus we have the following rates CHMe :CH-GOSH. CH :CH*CO,H. c12 W 2 ) ' . 0.62 0.018 Br + HBr (as k ; reactants ~ / 8 0 excess HBr) . 0.051 0.44 In a similar way it is found that introduction of a methyl group into maleic fumaric or glutaconic acid (thus forming citraconic mesaconic or P-methylglutaconic acid respectively) results in a deceleration of bromine addition in acetic acid when catalysed by hydrobromic acid but an accelera- tion of addition in aqueous acetic acid in which solvent the normal electro- philic reaction of bromine is favoured.Another striking example is the following CHPh:CHCO,H. CHPh:C(CO,H),. Br (aqueous HOAc) relative rates. . . 100 1 Br + HBr (HOAc) relative rates . . 1 250 In benzylidenemalonic acid it is seen that the presence of two carboxyl groups on the same carbon atom as compared with that of a single carboxyl group considerably favours acid-catalysed addition. The presence of two electron-demanding groups on different carbon atoms usually however has the opposite effect on addition by this mechanism. Thus bromine in the presense of hydrobromic acid adds to maleic acid at almost the same rate as to acrylic acid. With the same reagent Ph-COOCPh:CH*COPh is less reactive than Ph*CH:CH*COPh. It is possible that steric influences are important in these reactions ; but a very plausible explanation due to Hartman and Robertson,20 of such abnormally low reactivities is as follows.The normal state e.g. of the maleic acid molecule involves the contribution of polar structures such as (I). These are unfavourable to the stability of ion (11) as compared with (111) since the presence in (TI) of adjacent * A part of the LiBr-catalysed addition may however in this case involve attack by Br; on the olefinic compound. 142 QUARTERLY REVIEWS positive poles involves considerable destabilisation. These ions (11) and (111) are involved in acid-catalysed addition to maleic and acrylic acids respectively and hence the second carboxyl group in maleic acid hardly facilitates this type of addition. When however the second electron- attracting group is separated from the ethylenic carbon atom by a - 0 0 C-CH :CHC \+ +/ / \ (1.1 OH KO OH + / \ CH,*CH :C OH (111.) - 0 OH \+ + / C*CH*CH :C / HO OH \ C*OH - / \ / OG CHCH \\ methylene group this condition no longer applies and the expected order (e.g.CO,H*CH,*CH:CH*CO,H > CHMe:CH*CO,H) of reactivity is obtained. Similarly though the p-benzoquinones contain the structure CO-CH :CH*CO the formation of adjacent positive polarities in the ion may be avoided by resonance as in (IV) and these compounds add bromine very rapidly under conditions of acid catalysis. Description of acid-catalysed addition as in the above discussion has been challenged by S. V. Anantakrishnan and R. Venkataraman.2149 The mechanism proposed by these authors was originally believed to apply equally to catalysed and uncatalysed addition and involves the following stages CH,:CH + Br CH,:CH + HBr -+ CH,:CH +CH,:CHBr + HBr ; i ' 1 Br-H HBr + Br -+ HBr + Br,* CH :7H2 + 1 + Br,* -+ Br*CH,*CH + Br- + HBr Br-H Br*CH,*CH + Br- -+ Br*CH,CH,Br + It has been shown above that such a mechanism is not responsible for those addition reactions in which bromine acts as an electrophilic reagent ; and indeed the existence of uncatalyse'd third-order bromine addition (rate E [A][Br2]2) has recently been admitted by Anantakrishnan and his co-workers.50 A scheme such as the above still seems however to be supposed to explain acid-catalysed addition ; in particular these authors 4Q Proc.Indian Acad. Sci. 1946 23 A 307 312 319 Ibid. 1948 27 A 184. DE LA MARE THERMAL ADDITION OF HALOGENS ETC. 143 still believe that a proton catalyses addition by co-ordination with the ethylenic electrons.The evidence indicates strongly in the Reviewer’s opinion that such an attachment though in some circumstances possible in no case so far investigated results in catalysis of halogen addition. Two types of addition catalysed by hydrogen bromide have been recognised. In the first (anion- catalysed addition e.g. to vinyl bromide) the proton plays no part in the reaction. In the second (acid-catalysed addition e.g. to cinnamaldehyde acrylic acid co-cyanostyrene or wnitrostyrene) an alternative basic centre exists in the molecule which provides as compared with the double bond a much more probable site for addition of a proton and the presence of which in the molecule is essential for substantial catalysis to occur.A third type probably exists but has not yet been fully established in which the Br ion acts as a nucleophilic brominating agent for olefinic compounds containing strongly electron demanding groups. 8. Addition in Non-hydroxylic Solvents.-In hydroxylic solvents the kinetics of addition reactions though often complicated have been shown above to be consistently intelligible and the effects of structure on the rate of addition can be interpreted in terms of the electronic theory of organic reactions. The exact nature of the intermediates involved how- ever remains to be elucidated. It is known that addition is usually initiated by an electrophilic reagent probably by the halogen molecule ; but details of the subsequent stages are incompletely understood and an experimental study of the relation between products (determined by these final stages) the presence of added nucleophilic reagents and the kinetics should prove of value in extending knowledge of the detailed mechanism of this complex reaction.Non-hydroxylic solvents on the other hand are often used for pre- parative additions since the product of the reaction is then substantially the dihalide. Under these conditions however the kinetics are often intractable. For instance in carbon tetrachloride addition of bromine to ally1 benzoate can proceed by a heterogeneous reaction on a polar surface or at higher concentrations by a homogeneous reaction of high kinetic order. The rate of addition is greatly increased by the presence of small amounts of polar catalysts-water pyridine hydrobromic acid or iodine.Dissolved oxygen can also affect the rates of these additions. A kinetic study of iodine-catalysed addition of chlorine and of bromine to ethyl cinnamate has shown that iodine is not used up in the course of the reaction and that though the uncatalysed addition of chlorine is very slow and that of bromine is not reproducible the presence of iodine renders chlorine addition faster than that of bromine. The kinetics of the two reactions when the amount of added iodine is small are consistent with the rate equations -d[Cl,]/dt = L[A][Cl,][ICl] and - d[Br,]/dt = k[A][Br,l2[IBr] but the order with respect to iodine halide increases as the amount of added iodine becomes greater. Similar though less pronounced catalytic effects persist in the more polar solvents chloroform and chlorobenzene in which the uncatalysed K 144 QUARTERLY REVIEWS addition of bromine appears to be similar in kinetic form to that obtained in acetic acid.The effects of structure on the rates of these additions have not however been extensively examined by a satisfactory technique though it is probable that electrophilic attack by halogen is the commonest way in which reaction is initiated. The experiments of Anantakrishnan and Ingold l2 indicate further that nucleophilic attack by some halogenating agent can also be important when hydrobromic acid acts as a catalyst in addition of bromine to or#l-unsaturated aldehydes. The tendency towards over-simplification in discussions of addition in non-ionising solvents is exemplified by the recent work of R. W. Taft,51 who attempted to explain the varying proportions of substitution and addition in the gas- and liquid-phase chlorination of olefins.Thus iso- butylene is supposed to react with C1+ to form a carbonium ion which can decompose as follows \ CH3 No evidence exists however so far as the Reviewer is aware for such an intermediate in additions under the non-ionising conditions considered. It will be remembered for example that chlorine addition to ethyl cin- namate in carbon tetrachloride is certainly more complicated. 27 Detailed deductions from such a mechanism should therefore be treated with reserve. The present review has attempted t o indicate both what is known and also what is still uncertain in the detailed mechanism of halogen addition reactions and it will be valuable if some new experimental approach can be found to throw further light on the actual course or courses of these reactions.61 J. Amer. Chem. SOC. 1948 70 3364. The Reviewer has had both valuable comments on this manuscript and access to unpublished experimental findings and theoretical discussions provided by Professor P. W. Robertson of Victoria University College Wellington New Zealand whose assistance and encouragement are gratefully acknowledged. Advice and help given by Professor C. K. Ingold D.Sc. F.R.S. and by Professor E. D. Hughes D.Sc. have also been of great benefit ; and further the Reviewer thanks Drs. A. Maccoll R. J. Glillespie and D. P. Craig for discussion and comment. APPENDIX Relative rates* of addition reactions in acetic acid at 24". - Compound. CHPh :CMe . . . . . . . CHPh :CH . .. . . . . CHPh:CHkH,Cl . . . . . CHPh:CHPh . . . . . . CH,:CH.CH,-OAc . . . . . CH,:CHCH,*OBz . . . . . CH,:CH*CH,*O*CO*CH,Cl . . . CH,:CH*CH,Cl . . . . . . CH,:CH*CH,Br . . . . . . CMe :CHC1 . . . . . . . CHPh :CH.COPh . . . . CHPh:CH*CO*C,H,Me-p . . . CMe,:CH*CO,H . . . . p-C1*C,H4-CH:CH-C0.CMe . . CHPh :CHBr . . . CHPh:CH*CO,H (cis) . . . CHPh:CH*CO*C,H,(NO,) (m) . . CHPh:CH.C,H,(NO,) (2 4) . . CHPh:CHCO,Me . . . . . CH,:C(CH,Cl) . . . . . CHPh:CH*CO,H (transj . . . CPhiC*CO,H . . . . . . CHPh:CHGHO . . . . . CHMe:CH*CO,H (tranttj . . . CHMe:CH-CI-10. . . . . . CH :CHBr . . . . . . . CH,:CH*CO,Et . . . . . . CHPh:CH.CN . . . . . . CHPh:CH.NO . . . . . . CH,:CH-C0,H . . . . . . m-NO,*C,H4*CH :CH*COz1Cl@ ( ~.TYz?%s) p-NO,*C,H,*CH :CH*CO,Me (trans) Coumarin . . . . . . . o-NO,.C,H,*CH :CH*CO,Me (trans) HO,C*CH:CH*CO,H (cis) .. . p-Me*C,H,*CH :CH*COPh . . . p-Me*C,H,.CH :CH*CO,H (trans) . p-C1*C6H,*CH:CH*CbPh . . . m-NO,*C,H,*CH :CH*COPh . . Chlorine addition. - - - - - - ca. 800 - - - - 103 61 61 51 36 30 23 22 15 13 10 - 4-9 2-4 1.8 0.62 0.41 0.28 0.23 0-026 0.022 0.020 0.018 0.01 1 0.0049 0.0030 0.001 1 0*0001 I Bromine addition v. fast v. fast 77 18 10 14 7.8 -/- 6 1.6 1.0 0.5 0.33 0-33 -/- 0.15 0.11 0.063 0.028 0.019 0.017 0.0053 - - - - - - - - 0.001 1 - - - - - - - - - - Bromine addition catalysed by excess @/20 or ~ / 1 0 ) HBr. - - 16 8 9 - - - 3.8 2.2 0.3 - > 1000 - 0.18 0.07 0.07 0.03 0.1 7 0.03 0.051 0.012 - - - - - > 1000 > 10,000 - - 4.0 ca. 1.0 0.44 0.006 1 0.003 - - 0.56 * As given by Robertson and co-workers.19 17-2s~ 381 40 Selection has been made in certain cases between slightly diverging values of different experimenters. The values for chlorine addition are bimolecular velocity constants (min.-l g.-mol.-ll.). Those for bromine addition have been calculated as " bimolecular constants" i.e. if t (mins.) is the time for 20% disappearance of bromine with olefhic compound and total bromine initially each ~ / 8 0 then t h e quoted figure is 20/t. -/-Determined in the presence of excess of sodium acetate. $These values are for the corresponding acids. 146
ISSN:0009-2681
DOI:10.1039/QR9490300126
出版商:RSC
年代:1949
数据来源: RSC
|
3. |
Some aspects of the organic chemistry of derivatives of phosphorus oxyacids |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 146-159
F. R. Atherton,
Preview
|
PDF (951KB)
|
|
摘要:
SOME ASPECTS OF THE ORGANIC CHEMISTRY OF DERIVATIVES OF PHOSPHORUS OXYACIDS By F. R. ATHERTON B.Sc. M.Sc. PH.D. (RESEARCH DEPARTMENW ROCHE PRODUCTS LIMITED WELWYN GARDEN CITY) THE only naturally occurring organic derivatives of phosphorus are the esters and amides of phosphoric acid and the esters of polyphosphoric acids these compounds being present in the living organism as structural units co-enzymes and enzyme substrates. During the present century numerous compounds of these types have been isolated and identified thus enabling the establishment of the r81e played by phosphorus-containing materia.ls and in particular the elucidation of the pattern of carbohydrate metabolism. It is now known that the phosphate residues introduced into metabolites serve not only to facilitate breakdown? but also as prospective carriers of energy metabolic processes being so designed as to accumulate the energy of individual stages in " energy-rich " phosphate bonds.These '' energy- rich" phosphate bonds are distinguished by the large amount of energy liberated by bond-fission (ca. 11,000 cals.) in comparison with fission for a normal phosphoric ester (ca. 3000 cals.). The types of compounds possess- ing such bonds are those containing the linkages carboxyl-P enol-P N-P and P-0-P. The first two types are formed in metabolic processes by reactions which occur with little energy change but involve a redistribution of the energy of the molecules so that a considerable part becomes associated with the phosphate bonds. The other types which are con- cerned with the storage and utilisation of phosphate bond-energy do not arise directly but are the result of trans-phosphorylations.Enzymes acting as catalysts in these and other biological processes have in many cases been found to possess co-enzyme components which are phosphorus- containing materials. Since compounds of phosphorus occur ubiquitously and play such varied dies in the living organism it is only to be expected that related types of compounds would possess physiological activity. Indeed research during the war-years led to the discovery of phosphorus compounds of extreme toxicity which act by inhibiting choline-esterase the enzyme which hydrolyses acetylcholine. The difference between a toxic compound and one with therapeutic action being of degree and not of type it is likely that many new drugs will be found among phosphorus compounds.There is already some evidence that this is the case. Much of the present interest in the organic compounds of phosphorus lies in the development of methods for the synthesis of the more complex naturally-occurring derivatives of phosphoric acid and the search for compounds with physiological activity. It is however by no means restricted to these aspects for in addition to the many interesting theoretical points connected with their chemistry the patent literature indicates the most diverse uses for phosphorus compounds. 146 ATHERTON PHOSPHORUS OXYACID DERIVATIVES 147 The present article deals only with certain aspects of the chemistry of the substituted acids esters and chlorides of phosphorus. As the available space doesnot permit a full discussion of all classes of compounds con- taining these groupings the subject-matter presents some of the reactions characteristic of the individual types of groupings.The nomenclature used is that recently put forward by the Chemical Society in which compounds are derived by the replacement of hydrogen in the following parent acids HO*PH2 (HO )2PH HO *PO -H2 (HO),PO*H Phoephinous Phosphinio Phosphonous Phosphonic acid. acid. acid. acid. In American usage the terms phosphinic and phosphonous are inverted. 1. Hydroxy-compounds of Quinquevalent Phosphorus Compounds containing the characteristic grouping ‘PO*OH exhibit / well-marked acidic properties forming stable salts with both inorganic and organic bases. The most important reactions of the group are the replace- ment of the hydroxyl by halogen and its conversion into esters.(a) Formation of Halides.-This reaction does not seem yet to have been widely employed chiefly because many halides are more easily obtained by other methods. It is most likely to be of use for those compounds which do not contain ester groups. The reaction can be effected by treating the acid with phosphorus pentachloride under mild conditions ‘POeOH + PCl -+ >OCl + HCl + POCI In this manner alkylphosphonic and dialkylphosphonous acids have been converted into the corresponding chlorides. The above reaction proved less successful when applied to dibenzyl hydrogen phosphate,2 but in this case conversion has been achieved by the action of thionyl chloride on the potassium salt 3 (C,H,-CH,*O),PO*OK + SOCI -.+ (C,H,*CH2*O),PO-C1 + KC1 + SO ( 6 ) Formation of Esters.-The direct esterification of acids derived from phosphorus with alcohols though one of the earliest known reactions 4s 6 can be accomplished only under drastic conditions and has little application.On the other hand the formation of esters by the reaction of halides with silver salts has been so widely employed as to merit no further mention. Among the other methods of producing esters from the acids are the reactions occurring with diazo-compounds and olefin oxides. (i) Reaction with substituted diazomethanes. Diazomethane and sub- stituted diazomethanes react smoothly to give the corresponding esters. / 1A. W. Hofmann Ber. 1873 6 306. SA. Deutsch and 0. Ferno Ncctupe 1945 156 604. 4 T. J. Pelouze Anmkn 1833 6 129. =L. Zervas Naturwhs. 1939 27 317.5 J. yon Liebig ibid. p. 149. 148 QUARTERLY REVIEWS Phosphorous acid which contains only two true acid groups gives the corresponding diesters 6~ 'I BRR'CN + (HO),PHO -+ (RR'CH*O),PHO + ZN whilst phosphoric acid has been preferentially monosubstituted 8 RaCHN + HO*PO(OH) -+ R*CH,*O*PO(OH) + N Fully substituted phosphoric esters have been obtained by the action of diazoalkanes on monoesters and die~ters.~ (ii) Reaction with o l e h oxides. The oxide ring in olefin oxides can be opened by the acids of phosphorus to produce hydroxyalkyl esters. In order to effect mono-substitution of phosphoric acid 0. Bailly 10 treated disodium hydrogen phosphate with glycidol to give the disodium salt of a-glycerol phosphate /O\ CH,-CH-CH,*OH + HO*PO( ON&) + OH*CH,*CH( OH)*CH2*O*PO*( ON&) Other workers have treated olefin oxides with free phosphoric acid to get partial 11 l2 or complete sub~titution.~~ The hydroxyl groups of the products may react further with excess of olefin 0xide.13 As would be expected phosphorous acid gives only diesters,l3 e.g./"\ (HO),PHO + BCH,-GH + ( OH*CH,*CH,*O),PH*O Olefin imines undergo analogous reactions 139 l4 as do olefin sulphides.13 (c) Formation of Amides.-Amide formation from oxyacids of phos- Dianilinodithiophosphonous acid phorus and amines has not been reported. however reacts at high temperatures with aniline 15 (C6H5NH),*Ps*SH + C,H5*NH -+ (C,H,NH),PS + H2S ( d ) Formation of Anhydrides.-(i) With carboxylic acids. Monoacyl phosphates may be prepared by the action of acyl chlorides on monosilver phosphate 16* l7 R-COC1 + AgO.PO(OH) -+ R*CO*O*PO(OH) + AgCl or of a keten on excess of phosphoric acid 18 e.g.CH :CO + HO*PO(OH) -+ CH,*CO*O*PO(OH),e The dibenzyl esters of acyl phosphates are similarly prepared from dibenzyl hydrogen phosphate.l8. l9 * F. C. Pallazo and F. Maggiacomo Gaxzetta 1908 38 11 115. 7 F. R. Atherton H. T. Howard and A. R. Todd J. 1948 1106. * T. Reichstein and W. Schindler Helv. Chim. Acta 1940 23 669. BE. Muller A. Langerbeck and W. Riedel 2. physiol. Chem. 1944 281 29. 10 Ann. Chirn. 1916 6 133. 11F. Zetzche and F. Aeschlimann Helv. Chim. Acta 1926 9 708. l*E. Eiderbenz and M. Depner Arch. Pharm. 1942 280 227. 18 C. E. Adams and B. M. Shoemaker U.S.P. 2,372,244. 1 4 H. N. Chiistensen J. Biol. Chem. 1940 135 399. 1sA. C. Buck J. D. Bartleson and H. P. Lankelma J.Amer. Chem. Soc. 1948 17 A. L. Lehninger ibid. 1946 162 333. 1eR. Bentley J. Amer. Chm. Soc. 1948 70 2183. 1eF. Lynen Ber. 1940 73 367. 70 744. 16F. Lipmann and L. C. Tuttle J. Biol. Chem. 1944 153 571. ATHERTON PHOSPHORUS OXYACID DERIVATIVES 149 (ii) With acids of phosphorus. Polyphosphoric esters may be syn- thesised by condensation of the silver salt of an acid with a phosphorus halide. Thus chlorophosphonic esters react with the silver salt of a diester of phosphoric acid 2o to give the substituted pyrophosphate 0 0 + AgCl ""\f ?/OR RO 0 0 OR + P-0-P OR' / \ \Pf + \I?/ RO / \ RO ' 'Cl Ago OR' and with the silver salts of triesters of pyrophosphoric acid to give the substituted triphosphate' 21 0 0 0 0 0 ""\f t L O R + \P' -+ P-0-P-0-P ""\t f/ OR 0 OR P-0-P OAg C1 ' \OR R'O OR OR / \ R'O 2.Hydroxy-compounds of Tervalent Phosphorus The compounds which might, from their mode of synthesis be expected to be hydroxy-derivatives of tervalent phosphorus are not acidic and do not form stable salts although the hydrogen may be replaced by alkali metals. Their reactions which have been mainly studied with diesters of phosphorous acid are best interpreted as those of the tautomeric form The most important of these reactions are those occurring with halogens and halogen-containing materials. ( a ) Formation of Halogenophosphonates.4hlorophosphonates are formed from dialkyl 22* 23 and diaralkyl 24 phosphites by the action of chlorine a t low temperatures RO 0 RO 0 \I?' +HCl RO / \Cl / \ RO H or of sulphuryl chloride at room temperature 7 RO 0 RO 0 \Pf +HCl +so GI / \ RO / \H RO 2O J.Baddiley and A. R. Todd J. 1947 648. *1 J. Baddiley A. M. Michelson and A. R. Todd Nature 1948 161 761. 2aH. McCombie B. C. Saunders and G. J. Stacey J . 1945 380. **H. G. Cook H. McCombie and B. C. Saunders ibid. p. 873. %4F. R. Atherton H. T. Openshaw and A. R. Todd ibid. p. 382. 150 QUARTERLY REVIEWS Although the intermediate product cannot be isolated chlorination almost certainly takes place in the reaction which occurs when diesters of phos- phorous acid are treated with a polychlorinated hydrocarbon and an amine,25* 26 e-g- RO 0 RO 0 \I?/ +CHCl c1 / \ RO / \ RO H RO 0 R’ RO 0 -+ \/. + HCl / \ \Pf +HN / \ R’ RO NR’R” c1 / \ RO The formation of bromophosphonates can be effected by the action of bromine on diaralkyl phosphites 27 but iodine does not give iodophos- phonates,28 probably because the hydrogen iodide reducing the iodo p hosp hona t e .RO H RO 0 \PP \ / / 4 p +I + I / \ RO RO 0 (b) Formation of Phosphonates.-The sodium produced is capable of + HI derivatives of dialkyl phosphites which may be produced by the action of metallic sodium or sodium ethoxide react with halogen compounds when heated with them under reflux for several hours in an inert solvent to form phosphonates RO 0 RO 0 R‘ / \ RO / RO m a This reaction originally observed with ethyl iodide,29 has been extended to many other alkyl halidesY3O aralkyl halides,31 halogeno-carboxylic esters,32* 33 cyanoalkyl halide~,~3 and diaryl- arylalkyl- and dialkyl-arsine halides.34 It is of no use when applied to trialkyl- and triaryl-tin halides35 and proceeds only very poorly with dialkyltin ~lihalides.3~ Abnormal reactions occur a5 F.R. Atherton H. T. Openshaw and A. R. Todd J. 1945 660. asF. R. Atherton and A. R. Todd J. 1947 674. 27F. R. Atherton F. Bergel A. Cohen J. W. Haworth H. T. Openshaw and zsH. McCombie B. C. Saunders and G. J. Stacey J. 1945 921. zBA. Michaelis and T. Becker Ber. 1897 30 1003. 3oG. M. Kosolapoff J. Amer. Chem. Soc. 1945 67 1180. 311dern ibid. p. 2259. 3aP. Nylen Ber. 1924 57 1023. 3(L G. Kamai and 0. N. Belorossova Bull. Acd. Sci. U.R.S.S. Classe sci. chim. 35 B. A. Arbusov and A. N. Pudovick J. Gen. Chem. Russia 1947 17 21 58. s6B. A. Arbusov and N. P. Grechkin ibid. p. 2166. A. R. Todd B.P. 593,480. 331dern ibid. 1926 59 1119. 1947 191. ATHERTON PHOSPHORUS OXYACID DERIVATIVES 151 when the halogen is " positive " ; thus the reaction of ethyl bromomalonate and sodium dibutyl phosphite 37 gives tetraethyl ethanetetracarboxylate.Although no explanation has been published it would appear that the ethyl bromomalonate halogenates the sodium derivative i.e. RO 0 C0,Et RO 0 C0,Et / \ -+ \Pf +Na}CH / \ + Br*CH C0,Et Br / \ RO \P7 C0,Et / RO p a and further reaction then occurs among the components of the reaction mixture leading to the desired product tetraethyl ethanetetracarboxylate and probably tetrabutyl hypophosphate. This would account for the high phosphorus content of the phosphorus-containing fraction which was observed. An abnormal reaction also takes place with 9-chloroacridine to give acridme. s* 3. Esters of QuinquevaIent Phosphorus These esters can be converted into the acids by acid or alkaline hydrolysis or in the case of phenyl and benzyl esters by hydrogenolysis.The aryl esters are comparatively inert and will not be considered here. The most interesting reactions of the alkyl and aralkyl esters which are in many respects similar to esters derived from the sulphur acids are alkylation and the reactions they undergo in the presence of phosphorus halides. (a) Alblating Properties.-Although in published work at the present time these properties have been demonstrated only with esters of phosphoric acid they are not restricted to these compounds. (i) Alkylation of phenols. The alkylation of phenols by alkyl phosphates was first demonstrated 50 years ago when A. Morel 39 observed the production of phenetole in the reactions of triphenyl and diphenyl ethyl phosphate with sodium ethoxide e.g.(PhO),PO + 3EtONa -+ (EtO),PO*ONa + ZPhONa + PhOEt This reaction undoubtedly proceeds by a trans-esterification followed by alkylation of the phenoxide by an ethyl phosphate. It has been shown more recently 40 that trialkyl phosphates will alkylate phenols and that not all of the alkyl groups are utilised. At temperatures over 160° high-boiling alcohols are converted into their methyl ethers by trimethyl p h ~ s p h a t e ~ ~ whilst in addition to trans-esterification ether formation has been observed when trialkyl phosphates react with sodium b ~ t o x i d e . ~ ~ (EtO),PO + BuONa + (EtO),PO*ONa + BuOEt (ii) Alkylation of alcohols. (iii) Alkylation of amines. At high temperatures many trialkyl phos- 37 G. M. Xosolapoff J . Amer. Chm.SOC. 1946 68 1103. 3s1dem ibid. 1947 69 1002. 40C. R. NoIler and G. R. Dutton J . Amer. Chern. Soc. 1933 55 424. 41 A. D. F. Toy ibid. 1944 66 499. e2W. H. C. Rueggeberg and J Cbemaok ibid. 1948 70 1802. 39 Compt. rend. 1899 128 507. 152 QUARTERLY REVIEWS phates convert arylamines into the tertiary aminesY43 all three alkyl groups being utilised in the reaction 3NH& + 2(RO),PO -+ 3NArR + 2H,P04 Triisopropyl phosphate however gives only the secondary amine. 4-Ethyl- morpholine is produced by the action of triethyl phosphate on morpholine a t 150°.42 It should be emphasised that as with dialkyl sulphates there is con- siderable difference in the alkylating powers of successive groups the neutral esters being far more powerful alkylators than the salts produced by the primary alkylation.The aralkyl esters of phosphates and pyrophosphates are as would be expected considerably more powerful alkylating agents than albyl esters-with tertiary amines they form quaternary salts under comparatively mild conditions. These properties have been used to achieve the selective debenzylation necessary for the synthesis of the biologically important adenosine tri- phosphate.21 / \ +/ \ / \ / \ PO*O*CH,Ph + N- -+ PO*O-{CHzPh*N- (b) Interchange Reactions with Phosphorus Halides.-(i) Direct replace- ment. On heating with phosphorus pentachloride ester groups may be replaced directly by chlorine. When this reaction is applied to a dialkyl alkylphosphonate,44 either one or both of the ester groups may be replaced. 0 0 0 t/”’ or -P L O R -+ -P -P t/”” OR \ CI \ c1 \ (ii) Redistribution.It has been observed that trialkyl phosphates react when warmed with phosphorus oxychloride. Thus W. Gerrard 45 demon- strated the successive replacement of the butyloxy-groups in tributyl phosphate the extent of the replacement being dependent on the excess of phosphorus oxychloride present. (BuO),PO + POCI --+ (BuO),POCl or BuO*POCl Since the reaction will not be discussed elsewhere it is relevant to point out that similar redistributions occur with trialkyl phosphites and phosphorus trichl0ride.4~ (c) EIimination Reactions with Phosphom Halides.-Those compounds which contain both a halogen and an ester group are capable of eliminating the alkyl halide and forming a polymeric compound J. K. BilImann A. Radike and B. W. Mundy J . Amer. Chem. SOC. 1942 84 44 M. I.Habachnik and P. A. Roasiiekaya BulI. Acad. Sci. U.R.S.S. C b s e sci. 2977. c h h . 1946 615. ‘ 6 J . 1940 1464. ATHERTON PHOSPHORUS OXYACID DERIVATIVES 7 0 + [-P-O]n + RCI -P f/OR I I c1 \ 153 The reaction probably occurs in many of the thermal decompositions observed with compounds of this type. In thermal decompositions of the ethyl dial kylaminoc hlor o p hosp honi t e s which occur a t about 1 50 O the phosphorus compounds produced have been shown to be cyclic t r i m e r ~ . ~ ~ 0 0 T -+ [R,N-P-O-] + 3EtC1 7 7 3R,N*P I OEt \ This type of reaction may also be brought about by the action of tertiary bases at room temperature. Thus 4-methylmorpholine reacts with dibenzyl chlorophosphonate to give a high yield of the quaternary salt with benzyl chloride the other product presumably being benzyl metaphosphate.' (CH,Ph*O),POCl -j.[CHzPh.O*PO,]z + CHzPhCl There is also evidence that the reaction can occur between dissimilar molecules as the new insecticide tihe so-called " hexaethyl tetraphosphate " which is prepared by the action of phosphorus oxychloride on triethyl phosphate is now known to be a mixture of ethyl metaphosphate and tetraethyl pyrophosphate. G. M. Kosolapoff 47 has suggested rather a com- plex mechanism for this reaction although the formation of the products is simply explained by an initial redistribution reaction to form diethyl chlorophosphonate which could then undergo an elimination reaction either with itself to give ethyl metaphosphate or with triethyl phosphate to give tetraethyl pyrophosphate. 4. Esters of Tervdent Phosphorus The most characteristic reactions of this class of compounds occur in the presence of halogens or halogen-containing materials and result in a change of valency.(a) Action of Halogens.-Halogens react at low temperatures to split out alkyl halides and effect the reaction 0 In this manner trialkyl phosphites are converted into dialkyl chlorophos- phonates by chlorine 22 4 5 3 48 dialkyl bromophosphonates by bromine,4* dialkyl iodophosphonates by iodine 28 and dialkyl cyanophosphonates by 40 A. Michaelis Annakn 1902 826 129. r8 H. Wichelhaus AnnizZen Suppl. 1867 6 266. 47 Science 1948 108 486. 154 QUARTERLY REVIEWS the pseudo-halogen cyanogen iodide. 49 Diethyl chlorophosphinate reacts with chlorine to give ethyl dichloroph~sphonite.~~ (EtO),PCl + C1 + EtO*POCl + EtCl Although of little practical importance the conversion of alkyl dichloro- phosphinites into phosphorus oxyhalides and alkyl halides by the action of halogens 489 50 shows the generality of the reaction still further.The action of bromine on 1 1 1-trifluoro-2-propyl dibromophosphinite 51 is interesting in that these materials give an additive compound in the cold which decomposes on warming partly into its components and partly in the usual manner. CF CF3 CF3 \ / CHBr + POBr \ / CH*O*PBr + Br =J CH*O*PBr + \ / CH3 (b) Action of Halides.-Many halogeno-compounds react with tervalent compounds of phosphorus containing an alkoxy-group in accordance with the equation CH3 CH3 0 P-OR+XY -+ \pi’ +RX \ / / \ Y This reaction which usually occurs on heating but may sometimes take place on mixing the components is known as the Arbusov reaction.It has been demonstrated with a variety of types of tervalent esters but does not appear to occur if the phosphorus atom is halogen-substituted. Alkyl halides react with trialkyl phosphites to give dialkyl alkylphos- phonates ,52 dial k yl arylphosp hinat e s to give alky 1 ar ylal kylphosphoni t es 53 and alkyl diarylphosphinites to give diarylalkylphosphine oxides. 53 The ethyl esters of bisdialkylaminophosphinous acids also undergo the reaction,46 but in thiskinstance an unstable adduct may be isolated which eliminates alkyl halide on standing e.g. (R@),P*OEf + Me1 -.+ [(R2N),P<M]1 + (R,N),Pf + EtI Moreover the reaction is not restricted to the oxygen esters since alkyl diarylthiophosphinites form diarylalkylphosphine sulphides when treated with alkyl halides 54 Ar,P*SR + RI -+ Ar,R*PS + RI OR 0 Me \ ~ ~~~~ 4s B.C. Saunders G. J. Stacey F. Wild and I. G. E. Wilding J . 1948 699. 50N. l\denschutkin Anmkn 1866 139 343. 5rF. Swarts Bull. SOC. chim. Belg. 1929 38 99. 5aA. Arbusov J . Rum. Phys. Chem. SOC. 1906 38 687. 531dem ibid. 1910 42 396. sp Idem <bid. p. 549. ATHERTON PHOSPHORUS OXYACID DERIVATIVES 155 It has been shown that the reaction occurs with a wide variety of halogen- containing compounds. Thus trialkyl phosphites have been treated success- fully with polyhalogenoalkanes 38 49 55 halogenocarboxylic esters,56 aralkyl halides 31 49 57 halogen derivatives of heterocyclic halides of disubstituted ar~ines,3~ trialkyltin halides 35 dialkyltin dihalides and alkyltin trihalides.36 Acid chlorides such as acetyl chloride and benzoyl chloride react readily even at room temperature to give esters of acylphosphonic acids.58 The above reactions are undoubtedly typical of the corresponding aralkyl esters. 5. Chlorides of Quinquevalent Phosphorus Compounds of this type exhibit most of the characteristic properties of normal acid chlorides. They are decomposed by water to give the corre- sponding acids with varying ease and exhibit the usual formation of esters and amides and replacement reactions of the halogens. (a) Ester Formation-(i) With sodium alkoxides. The reaction takes place smoothly alkyl dichlorophosphine dialkylaminodichloro- phosphine oxides 46 and aryl dichlorophosphonites 59 undergoing replacement of both halogens and bisdialkylaminochlorophosphine oxides 46 and diary1 chlorophosphonates 59 replacement of the one halogen atom.A r ~ l - ~ o alkyl-,61 and anilino-dichlorophosphine oxides 62 and aryl dichlorophosphonites 63 usually react with both halogens but it has been shown that when quinoline is used as the tertiary base it is possible to achieve selective replacement of one chlorine in phenyl dichloropho~phonite.~~ Diary1 63 and diaralkyl73 24 chlorophosphonates and dianilinochlorophosphine oxide 62 react satisfac- torily and removal of the protecting groups from the products may be achieved to yield pure monosubstituted phosphoric esters. (b) Replacement Reactions.-(i) By amino-groups. All types of com- pounds containing halogen atoms may be caused to react successfully with amines to give the corresponding amino-derivatives. A.Michaelis 4~ gives a wide range of examples of both complete and partial substitution. Alkyl dichlorophosphonites,65~ e6 dialkyl chlorophos- phonates 66 dial k y laminodichlor o - and bis dial k ylamino - c hlor o - p hosp hin e (ii) In the presence of organic tertiary bases. (ii) By fluorine. 55A. J. Arbusov and N. P. Kuschkowa J . Gen. Chem. Rwaia 1936 6 283. 56A. Arbusov and A. Dunin J . Rws. Phys. Chem. SOC. 1914 46 295. 67 B. P. Lugovkin and B. A. Arbusov Doklady Akad. Nauk. S.S.S.R. 1948,59,1301. M. I. Kabachnik and P. A. Rossiiskaya Bull. Acad. Sci. U.R.S.S. Classe sci. chim. 1945 364. 59 A. Morel Bull. Soc. chim. 1899 [iii] 21 491. 6o A. D. F. Toy J . Amer. Chem. SOC. 1948 70 186. 61 M. I. Kabachnik P. A. Rossiiskaya and N. N. Novikova Bull. A d . Sci. U.R.S.S. 62 F. Zetzche and W.Buttiker Ber. 1940 73 47. 63 P. Brig1 and H. Miiller {bid. 1939 72 2121. 64E. Baer and M. Kates J . Amer. Chem. SOC. 1948 70 1394. 6 5 B. C. Saunders and G. J. Stacey J . 1948 695. Classe sci. chirn. 1947 97 B.I.O.S. Final Report No. 714 Item 8, 156 QUARTERLY REVIEWS oxides and sulphides 66 are converted into the corresponding fluoro-com- pounds by the action of sodium fluoride in an inert solvent 0 0 4 3. \I \I P-Cl + NaF + P-F fNaCl The complete and partial fluorination of ethyl dichlorothiophosphonite may be accomplished by the action of antimony trifl~oride.~’ (iii) By the thiocyanato-group. Dialkyl chlorophosphonates 49 react with potassium thiocyanate to give dialkyl thiocyanatophosphonates. The reac- tion of diethylamino-dichlorophosphine oxide 46 with silver thiocyanate is said to give a thiocyanato-derivative though from general experience it would not be surprising if this were actually an isothiocyanato-compound.The action of an alcoholic solution of potas- sium cyanide on diethylaminodichlorophosphine oxide 46 gives simultaneous replacement by the cyano-group and esterification (iv) By the cyano-group. 0 LCN Et,N*POCl + Et,N*P OEt \ Diethyl chlorophosphonate however reacts poorly with potassium cyanide.49 (v) By acyloxy-groups. Dialkylaminodichlorophosphine oxides react with the sodium salts of carboxylic acids to replace both halogens,66 but when the reaction is carried out in alcoholic solutions an ester group is also . introduced. 0 0 t/OR’ R,N*P --+ R,N*P 0.OC.R” \ c1 \ ( c ) Reaction with diazoa1kanes.-Although only one observation of this type of reaction has been made to date namely that a dialkyl fluorophos- phonate reacts with diazomethane to give the fluoromethylphosphonate,49 RO 0 RO 0 \Pf + CH,N -+ + NB CH,F / \ F RO / \ RO it is very probable that this will prove to be a general reaction applicable to all types of halides.67 H. S . Booth D. R. Martin and F. E. Kendall J . Amer. Chern. Xoc. 1948,70,2523. ATHERTON PHOSPHORUS OIZI?ACfD DERIVATIVES 157 6. Chlorides of Tervdent Phosphorus Compounds of this type are as would be expected very similar in many respects to the corresponding quinquevalent compounds giving esters with sodium alkoxides or alcohols in the presence of tertiary bases and under- going replacement of the halogen by amino- and other groupings. These reactions of the compounds will not therefore be discussed further although it is relevant to point out that irregularities are more.likely to occur in this series.Thus whilst diphenylchlorophosphine reacts normally with most sodium alkoxides to give the corresponding esters sodium methoxide and benzyloxide give the phosphine 0xide,~3 e.g. PPh,C1 + NaOMe + PPh,MeO + NaCl Reactions peculiar to this class of compound are those which they undergo with carbonyl compounds and olefin oxides. (a) Reaction with Carbons1 Compoun&Although earlier workers had found that mixtures of phosphorus trichloride and carbonyl compounds gave a-hydroxy-phosphonic acids when treated with water our knowledge of this type of reaction is largely due to the studies of J. B. Conant and his co-workers. The initial reaction of phosphorus trichloride with carbonyl compounds was found to be reversible,6** G9 giving an additive product which reacted more readily with acetic acid or anhydride than did phosphorus trichloride.In this way it was possible to disturb the equilibrium and make the reaction proceed virtually to completion. The mechanism of the reaction is postu- lated by these workers as an extension of valency of the phosphorus tri- chloride 0 PCI \ / \ &o +PCl + c- \ / / as had already been suggested,'O followed by a disturbance of the equilibrium by interaction of the intermediate with acetic anhydride (or acid) e.g. 0 PCI + Ac,O C- \ / \ / Hydrolysis of the reaction mixture 0 0 \ / \ P / + c--- P-Cl + 2AcC1 would then give the hydroxy-phosphonic acid OH 0 0 0 C- P-CI \ / \ P / \I t/OH + c-P OH / \ 68 J.B. Conant and A. D. Macdonald ibid. 1920 42 2337. 70 A. Michaelis and A. Fleming Ber. 1901 34 1291. J. B. Conant A. D. Mscdonald and A. McB. h e y {bid. 1921 43 1928. 158 QUARTERLY REVIEWS Similar reactions occur with phenyldichlor~phosphine,~~ diphenylchloro- ph~sphine,~~ phenyl and alkyl dichlorophosphinites and diphenyl chloro- ph~sphinate.~~ The general overall reaction may therefore be expressed OH 0 \I 7 C=O +Cl-PRR’ -+ C-PRR’ \ / / When ap-unsaturated ketones are treated with phosphorus tri- The production of this class of compounds is postulated to occur in a similar manner the initial addition being of the 1 4-type 75 the product is a p-keto-phosphonic acid. CHR-CH*CO.R’ + PCl + CHR-CH=CR’ I 0 I C1,P to give an adduct which then reacts with acetic anhydride (or acid).On hydrolysis the enol-phosphorus link is severed to give the /?-keto-phosphonic acid CHR-CH=CR’ CHR*CH,*CO OR’ I -+ //OH 0 OtP O I \P OH / \ Cl Here again similar reactions occur with phenyldi~hlorophosphine,7~ T7 diphenyl~hlorophosphine~7 phenyl dichlorophosphinite and diphenyl chloro- ph~sphinate,~~ the general reaction therefore being CHR==CH*COR’ + PR”R”’C1 + CHR-CH,*CO*R’ I 0 CPRNR”’ Although the overall reactions as described above are correct the mechanisms are very probably incorrect? since other experiments 7* suggest that the initial reaction between phosphorus trichloride and benzaldehyde is a more conventional type of addition. Ph*CHO + PCl + CHPhCl*O*PCl A further point against the above mechanisms is that the reaction can be carried out successfully with alkyl dichlorophosphinites which would necessitate the production of an intermediate compound with an alkoxy- group and a halogen on a phosphorus atom with five substituents.71A. Michselis Annalen 1896 293 222. 7% J. B. Conant J. B. S. Braverman and R. E. Hussey J . Amer. Chem. SOC. 1923. 79 J. B. Conant V. H. Wallingford and S. S. Gandheker ibid. p. 762. 74 J. B. Conant ibid. 1917 39 2679. 75 J. B. Conant and A. A. Cook {bid. 1920 42 830. 76 J. B. Conant and S. M. Pollack ibid. 1921 43 1665. 77 J. B. Conant A. H. Bump and H. S. Holt ibid. p. 1677. 78F. R. Atherton and A. R. Todd unpublished. 45 165. ATHERTON PHOSPHORUS OXYACID DERIVATIVES 159 (b) Reaction with Olefin Oxides.-Although halides of quinqevalent phosphorus compounds do not react with olefin oxides in the absence of hydrogen halide tervalent compounds react very readily to form esters e.g.0 /-\ \ P-Cl + CH,-CH2 + P-O*CH,*CH2CI \ / / In this way ethylene oxide reacts with diphenyl chlorophosphinate and phenyl dichlorophosphinite to give triesters of phosphorous a~id.7~ With phos- phorus trichloride ethylene oxide reacts progressively to give 2-chloroethyl dic hlor o phosp hinite di - (2 - chlor oe t h yl) c hlor op hos phinat e and tri - (2 - chlor o - ethyl) phosphite 8o ; propylene oxide behaves similarly.78 It is hoped that the above review will serve to give an insight into the chemical behaviour of the more common types of phosphorus compounds. The material represents however only one facet of the organic chemistry of phosphorus of which the pioneer A. Michaelis is quoted by G Schrader 8* as having written “ Even if at this present moment no special possibilities are apparent yet there will of that I am sure be a future for this subject surpassing even its great past.’’ 7sM. I. Kabachnik Bull. Acad. Sci. U.R.S.S. C h s e sci. china. 1947 631. M. I. Kabachnik and P. A. Rossiiskaya Bull. Acad. Sci. U.R.S.S. Classe sci. china. 1946 295. L
ISSN:0009-2681
DOI:10.1039/QR9490300146
出版商:RSC
年代:1949
数据来源: RSC
|
4. |
Carbides, nitrides, and carbonitrides of iron |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 160-172
H. L. Riley,
Preview
|
PDF (1189KB)
|
|
摘要:
CARBIDES NITRIDES AND CARBONITRIDES OF IRON By H. L. RILEY D.Sc. A.R.C.S. P.R.I.C. (DIRECTOR OF CREMICAL RESEARCH AND DEVELOPMENT UNITED STEEL COMPANIES LTD.) Introduction o-.- METALLIC interstitial solid solutions are alloys in which the solute atoms are situated in the interstices formed by the atoms of the metal solvsnt. Our knowledge of interstitial alloys is fragmentary because those atoms which are small enough to occupy the interstices of a metal-atom lattice viz. hydrogen boron carbon nitrogen and probably also oxygen have low scattering powers for X-rays and in many cases their location in the alloy is conjectural. Nevertheless the results which the further study of these alloys is likely to yield are of importance in several fields the elucida- tion of the finer features of their crystal structure will advance our funda- mental knowledge of the metallic state ; they play an essential but obscure part as catalysts in such technically important reactions as ammonia synthesis and oxidation the Fischer-Tropsch reaction etc.; and the inter- stitial carbides and nitrides are of importance in general ferrous metallurgy and particularly in the .case-hardening and heat treatment of steel The numerous interstitial solid solutions formed by the transition metals on the one hand and hydrogen boron carbon or nitrogen on the other have been investigated in considerable detail by G. Hagg,l who has classified them on the basis of their crystal structures. Most of the transition metals have close-packed structures ; if we picture the metal atoms as rigid spheres then the h I I !(3 __-___- * ----- --o 0.) b.,,’ i ‘ close-packed hexagonal structure can be regarded as being made up of triangularly close-packed layers of spheres stacked in a close-packed sequence abababab the a layers being normally over each other and similarly the b layers. The close- packed face-centred structure similarly consists of close-packed layers but these lZ. physikal. Chem. 1929 B 6 221 ; 1931 B 12 33. 160 RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 161 in which the metal atoms are not close-packed but have body-centred or simple hexagonal structures. In both the close-packed structures there are two tetrahedral interstices and one octahedral per metal atom. Not all the interstices need be occupied but only one type of interstice either tetrahedral or octahedral is occupied in a particular interstitial compound or intermediate phase.It is believed that in occupying an interstice the atom must be in contact with the surrounding metal atoms a condition which limits the radius ratio of the atoms concerned. The condition for contact in the tetrahedral interstice is that the ratio of the radius of the interstitial atom R, to that of the close-packed metal atom R, shall ba 0.22 and for an octahedral interstice 0.41. Interstitial carbides are of two kinds, viz. the refractory carbides such as those of titanium zirconium tantalum etc. on the one hand and carbides such as those of iron cobalt and nickel on the other. Carbides of the former group are chemically inert and melt only a t exceedingly high temperatures. C. Agte and H. Alterthum give the following melting points ZrC 3530" ; NbC 3500" ; Mo,C 2690" ; MoC 2690" ; HfC 3890" ; TaC 3880" ; W,C 2860" ; WC 2870".E. Friederich and L. Sittig report the following melting points for corresponding nitrides TiN 2930" ; ZrN 2930" ; VN 2050" (decomposition) ; NbN 2050" ; TaN 2800" (decom- position) ; ScN 2650". Carbides of the second group are chemically reactive being readily decomposed by dilute acids and are only metastable at temperatures in the neighbourhood of 1000". The corresponding nitrides are even less stable. Hagg has pointed out that inert carbides have the radius ratio R R, less than 0.59 whilst the ratio for the unstable carbides exceeds this value. Interstitial alloys are metallic in character opaque good conductors of heat and electricity lustrous and like metallic alloys show variable composition ; in contrast to pure metals however the inert interstitial alloys melt only at exceedingly high temperatures.Their electrical conductivities decrease with rise of temperature. C. Agte and I(. Moers 5 have described the preparation in the pure state and the properties of a large number of high-melting carbides nitrides and borides; using a filament technique I(. Moers has measured their elec- trical conductivities over a wide range of temperature ; the borides had the highest specific conductivities even higher than those of the pure metals. Atomic diffusion in solids can occur by two processes where the diffusion involves the formation of primary substitutional solid solutions an interchange of lattice sites by the participating atoms must take place ; where interstitial alloys are formed the penetration of the smaller atoms from interstice to interstice in the lattice of the larger atoms is involved.The latter process probably involves a much smaller sctivation energy than the former. According to M. Paschke and A. Hauttmann the rate of 2 See M. von Stackelberg ibid. 1934 B 27 53 ; G. Hagg loc. cit. and A. Wsstgren 3 2 . tech. Physik 1930 11 183. J . Fyanklin I n s t . 1931 212 577. 4Z. anmg. Chem. 1925 143 293. Ibid. 1931 198 233. Arch. Hisenhiittenw. 1935-36 9 305. 162 QUARTERLY REVIEWS diffusion of manganese into iron a t 1400" appears to be only about 1% of that of carbon. The Reaction 2CO = C + CO Metallic interstitial carbides are usually prepared by one or other of two methods the interaction of elementary carbon with the metal or its oxide at high temperatures or the interaction of a hydrocarbon gas or carbon monoxide with the metal or its oxide at much lower temperatures.The mechanism of the direct interaction of metal and carbon is obscure; in some cases it undoubtedly involves the intermediate formation of a gaseous phase carbon monoxide hydrocarbon or even metal vapour. The high temperature necessary when elementary carbon is employed however is undoubtedly connected with the high thermal stability of the graphite lattice. Because of its technical importance the interaction of carbon monoxide with transitional metals and oxides particularly with iron and its oxides has been the subject of numerous investigations ; the literature concerning the cleavage of carbon monoxide 2CO = C + CO, AH = - 39 k.-cals.on metals of the iron group is extensive. There are however few papers dealing with the reaction in presence of other catalysts. G. Fester and G. Brude,' using palladium deposited on activated charcoal or activated silica have observed some reaction at relatively low temperatures (240'). A. Foixs stated that at > 1100" carbon monoxide deposited amorphous carbon on diamond the latter undergoing no change; equili- brium had not been reached even after three hours. J. Cleminson and H. V. A. Briscoe9 found that no reaction takes place below 400" when glass and mercury are the only contact surfaces; they report however slow reaction a t temperatures below 300" in the presence of solid carbon magnesia or alumina. A. R. McKinneylO has investigated the decom- position of ethylene and carbon monoxide on metallic catalysts and con- cluded that decomposition of carbon monoxide is catalysed only by metals capable of forming carbonyls.R. L. Burwell (jun.) and H. S. Taylor l1 have shown that when carbon monoxide is passed over zinc oxide a t > 184" the cleavage reaction occurs at the surface and that the deposited carbon presumably in the atomic form is removed by hydrogen at the same temperature. Finely divided iron cobalt and nickel and/or their oxides exceed greatly in catalytic activity all the above-mentioned solids and there is little doubt that the formation of interstitial carbides plays an important and probably essential part in the catalytic mechanism. The activity of these iron-group metals and oxides varies over a wide range and is influenced greatly by the previous history of the catalyst.In spite of the large number of experi- mental studies particularly of iron and its oxides there is still disagreement as to the mechanism of the reaction. 0. Boudouard l2 was of the opinion that the oxides of iron catalysed the deposition of carbon from carbon 8 Bull. SOC. chim. 1922 [iv] 33 678. lo J . Physical Chem. 1943 47 152. l2 Compt. rend. 1899 128 98 307. 7 Ber. 1923 56 2245. Q J . 1926 2148; see also ref. (62). 11 J . Amer. Chem. SOC. 1937 59 697. RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 163 monoxide whereas R. Schenck and F. Zimmerman l3 believed that the metal itself and not its oxides was the catalyst. S. Hilpert and T. Dieck- mann suggested without experimental proof that the iron itself is not the specific catalyst but rather certain carbides including a higher carbide formed from the metal.It has been known for some time that in the decomposition of carbon monoxide on iron or its oxides carbide formation occurs.15 W. Gluud K. V. Otto and H. Ritter l 6 referred to the existence of a new carbide of iron Fe,C when Fe,03 was heated in carbon monoxide a t 275" rapid reduction to Fe,O occurred and was followed by the simul- taneous deposition of carbon and formation of carbide; the existence of Fe,C was argued from the kinetic study of the reaction. U. Hofmann and E. Groll l 7 obtained indications from X-ray-diffraction photographs of the existence of a new iron carbide mixed with the carbon deposited from carbon monoxide; they supposed it to be the new carbide previously reported by Gluud and co-workers.About the same time F. Fischer and H. A. Bahr l8 reported indications of the formation of a higher carbide (Fe,C,) by the prolonged low-temperature (270") interaction of carbon monoxide and ferric oxide. found that the long continued treatment of iron or iron oxide with carbon monoxide a t 225" gave a new carbide for which the formula Fe,C was suggested ; the diffractions in its X-ray powder photograph were listed but not analysed. H. A. Bahr and V. Jesson 198 thought it probable that pure Fe,C containing 9.7% of carbon could be prepared from iron and carbon monoxide by employing the lowest possible temperature of interaction 225" or lower ; between 230" and 400" they concluded that besides free carbon a mixture of cementite and Fe,C is formed and over 400" only cementite with 6.68% carbon.H. Tutiya l5 inferred from his own experimental results that iron itself functions not as the catalyst but as a supporter of autocatalytic dissociation of carbon monoxide ; iron carbide is formed as soon as the iron comes in contact with the carbon monoxide the two reactions G. Hagg and 3Fe + 2CO = Fe,C + CO 2co = c + co occurring simultaneously ; it is so he states the Fe,C which plays the chief part in the catalytic dissociation of carbon monoxide. Hofmann and Groll l7 claimed to have established with certainty that when carbon is deposited from carbon monoxide cementite is always present in the solid phase; the formation of elementary carbon occurs however not by the decomposition of the cementite for it was found that the latter formed l3 Ber.1903 36 1231 ; Stahl zc. Eisen 1905 25 768. l4 Ber. 1915 48 1281. l5 K. Stammer Pogg. Ann. 1851 82 136; I. L. Bell J . 1896 209; A. Gautier and P. Clausmann Compt. Tend. 1910 151 16 335 ; S. Hilpert and T. Dieckmann loc. cit. ; H. Tutiya Sci. Papers I n s t . Phys. Chem. Res. Tokyo 1929 10 69. 17Z. anorg. Chem. 1930 191 414. l6 Ber. 1929 62 2483. l98 Ber. 1933 66 1238. Ges. Abh. Kennt. Kohle 1924/1927 8 255. 2. Krist. 1934 89 92. 164 QUARTERLY REVIEWS in this manner could be heated in nitrogen for two hours at 450" without appreciable decomposition whereas when heated in carbon monoxide at the same temperature and for the same time it deposited ten times its own weight of carbon. The X-ray diffraction diagrams of the higher carbide were obtained only from the products of experiments carried out below 400" and it was therefore concluded that the higher carbide decomposed quickly a t temperatures above 400".Hufmann and Groll suggested that when free carbon is formed by the interaction of carbon monoxide with iron or iron oxide it is through the agency of the formation and decom- position of this higher carbide that free carbon and in addition iron and cementite are formed. The following two well-established and significant experimental facts have an important bearing upon any theory claiming t o describe the mechanism of this reaction. (i) The carbon formed from carbon monoxide is relatively highly graphitic ; although microcrystalline the crystallites present are much larger than those prepared by carbonising a carbonaceous solid a t the same temperature ; 2o whatever the reaction mechanism there- fore it must involve a certain amount of mobility in the carbon atoms in order that crystal growth may take place.(ii) The iron catalyst is dispersed in an extremely fine state throughout the whole of the deposited carbon and the reaction continues until the concentration of iron in the carbon has been. reduced to about 1yo.21 The iron is present in the carbon largely as carbide and is exceedingly difficult to remove; prolonged treatment with boiling hydrochloric and nitric acid is necessary for its elimination. The work of W. Riidorff and collaborators 22 on the ferric chloride-graphite complex explains why attempts to eliminate this iron by treating the product with chlorine at high temperatures were only partly successful ; under these conditions the ferric chloride formed would tend to become intercalated in the graphite crystal lattice.F. Fischer and H. A. Bahr l8 showed signi- ficantly that if an iron-copper catalyst interacts with carbon monoxide at 500° the carbon formed contains highly dispersed copper in addition t o iron and this suggests that carbonyl formation is not responsible for the dispersion of the catalyst. That the formation and decomposition of carbides or percarbides are not essential parts of the catalytic mechanism is indicated by the fact that the reaction can be carried out at high temperatures a t which cementite is unstable. According to H. A. S ~ h w a r t z ~ ~ in high-purity iron-carbon alloys containing 0.03% of silicon the reaction Fe,C +3Fe + C proceeds to the right at all temperatures from 630" to the melting point of the eutectic.G. Naeser 24 states that the decomposition of cementite which was followed by measurements of magnetic susceptibility takes place a t 1050-1060". The 20 U. Hofmann 2. Elelctrochern. 1936 42 504. 2 1 Cf. e.g. von Wangenheim Qes. Abh. Kennt. Kohle 1924/1927,8 227 ; F. Fischer and P. Dielthey Brenmtoff-Chern. 1927 8 388 ; 1928 9 24 ; Qes. Abh. Kennt. Kohle 1924/1927 8 234. 23 W. Riidorff and H. Schdz 2. anorg. Chem. 1940 245 121. 28 Trans. Arner. SOC. Met. 1935 23 126. 24 Mitt. Kaiser Wilh. Inst. Eisenforsch. 1934 16 211. RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 165 cementation of steel rods in a stream of pure carbon monoxide at temperatures up to 1000" has been investigated in detail by A.Bramley and A. J. J i n k i n ~ . ~ ~ Under these conditions carbon atoms penetrate into the y-iron lattice to form a solid solution (austenite) and cementite is formed only when the steel cools. This may take the form of inter-granular cementite and pearlite according to the concentration of the carbon. It has been shown 28 that if the cementation reaction is prolonged even at temperatures as high as 1 loo" considerable quantities of highly graphitic carbon containing dispersed iron are formed ; the iron is in the form of cementite indicating that this reaction is essentially similar to that occurring at lower tem- peratures. Occasionally the presence of the higher iron carbide mixed with excess of carbon was observed in samples which had been formed a t temperatures as high as 900" ; however the conditions necessary to bring about the formation of this carbide were critical and were not fully eluci- dated.Carbon is appreciably soluble in the face-centred cubic y-iron and the following appears a likely explanation of the reaction mechanism. Presumably the carbon monoxide molecules first condense on the surface of the steel [formation of a surface carbonyl (McKinney lo)] and two of the condensed molecules react to form a molecule of carbon dioxide (which is desorbed) and an adsorbed carbon atom (surface carbide); the carbon atom then penetrates by way of the interstices into the y-iron lattice leaving the surface free for the condensation of further carbon monoxide molecules. If a high concentration of carbon atoms in the iron is built up in this way there will arise an increasing tendency for the formation of graphite crystals.How the first crystal nucleus is formed we do not know but the diffusion experiments of Bramley and Jinkins 26 and others leave no doubt as to the relatively high mobility of carbon atoms dissolved in y-iron. Evidence of this mobility is also apparent in partly graphitised alloys ; 27 photomicrographs show regions of carbon impoverish- ment immediately surrounding the graphite crystal flakes. During the carburisation of steel the highest concentration of carbon in the y-iron will always occur in the surface layers and it is therefore there that the formation of graphite crystals is most likely to occur. P. R. Marshall 28 carburised a small piece of steel to saturation in carbon monoxide at 1000" ; a photo- micrograph of the product showed a relatively large graphite crystal immediately below the surface ; the crystal was surrounded by a eutectoid zone which in turn was surrounded by a cementite network.Electron- microscope photographs of carbon monoxide-carbon 29 indicate that the iron impurity must be present in an extremely fine state of division. It appears probable therefore thaf the contamination of the elementary carbon by iron and cementite is due to the mechanical dispersion of the solid catalyst brought about by the graphite crystal-building forces acting 26 Mem. Iron Steel Inst. Carnegie Schol. 1926 15 17. 2a J. Taylor and D. Laidler unpublished. 27 H. Carpenter and J. M. Robertson " MetaIs " Oxford University Press London 28Ph.D. Thesis Univ. Durham 1944 p. 82. 1939 vol. 2 p.1172. 20 Ibid. p. 67. 166 QUARTERLY REVIEWS in its surface layers.30 A similar view has been advanced by L. J. E. Hofer 81 to explain the catalyst dispersion which occurs at much lower temperatures. If however we attempt to extend this view to explain the mechanism of the carbon-deposition reaction to lower temperatures a difficulty arises for carbon is only very sparingly soluble in the body-centred a-iron lattice. At 1130" y-iron can dissolve 1.7% of carbon; a t 725" this has fallen to 0.87%. At 725" the body-centred a-iron dissolves only 0-035y0 of carbon which decreases to 0.007% at room temperature. Below the a-y transition temperature iron formed by the reduction of iron oxide will presumably be in the a-form the penetration of carbon atoms into which would be expected to follow a somewhat different course from that obtaining in the case of y-iron.Carbon monoxide does not react with massive iron a t an appreciable velocity at temperatures below 750". It might be supposed that the fine state of sub-division in which iron is formed by the reduction of oxide powder is responsible for the relatively high reaction velocities observed with some iron oxides iH the temperature range 400-550". The formation of cementite and iron percarbide and the dispersion of the catalyst in the deposited carbon in both the high- and low-temperature reactions suggest that the reaction mechanisms are fundamentally similar. We shall return to this point later. Iron Nitrides Metallic iron does not react with molecular nitrogen but when the metal is heated in gaseous mixtures containing ammonia it forms a t relatively low temperatures a number of nitride phases.The interest in these nitrides was increased by the discovery by A. FryYs2 that when special steels containing aluminium and chromium (nitralloy) are heated in ammonia at 500" hardening of the surface occurs without any further heat treatment. Case-hardening by nitriding is now carried out by subjecting the machined steel to the action of ammonia for seventy to ninety hours a t 500-540". Various iron-nitrogen phases are formed by the progressive penetration of nitrogen into the iron lattice. The ammonia is dissociated a t the iron surface and nitrogen atoms from the adsorbed layer diffuse into the iron. At 500" the iron is in the a-form the solubility of nitrogen in which is of the order of 0.3% at 500-540".Above this concentration of nitrogen a second phase y' is formed; it has a face-centred cubic structure and is homogeneous for the range 5.7 to 6.1% of N ; it is usually referred to as the nitride Fe4N.3S The next phase (E) which appears as the nitrogen con- centration increases has a close-packed hexagonal structure and a range of homogeneity of 7-3 to 11.1% of N ; 34 this range includes the stoicheiometric formula Fe,N (N 7.72%). A phase still richer in nitrogen the [-phase (Fe2N contains 11.14y0 of N) also exists ; it has a base-centred orthorhombic 30 See W. Baukloh and B. Edwin Arch. Eisenhiittenw. 1942 16 197. 31 U.S. Bur. of Mines Rep. Invest. No. 3770 July 1944. 32Krupp Monatsh. 1923 4 137; Stahl u. Eisen 1923 43 1271. G. Hiigg Nova Acta SOC. Sci. Upsal. 1929 [iv] '7 1.34 G. Hligg loc. cit. ; A. Osawa and S . Iwaizumi 2. Krist. 1928 69 26. RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 167 structure which can be formed by a slight distortion of the close-packed hexagonal structure of the &-phase; it was first reported by G . Hagg.35 Recently K. H. Jack 36 has obtained evidence for the location of the nitrogen atoms in the y'- and c-phases from the positions of super-lattice lines in their respective X-ray powder diffraction diagrams. These iron nitride phases are formed successively as more and more nitrogen atoms penetrate into the interstices of the iron lattice ; the nitriding reaction is reversed if the phases are heated in a vacuum or in an inert gas. The nitrides are dull grey powders which dissolve in dilute hydrochloric acid with the formation of ammonium chloride.Because of the lower temperature at which it is carried out nitriding is a slower process than carburising and various attempts have been made to accelerate it ; these have been critically reviewed by E. K~nze,~' who states that phosphatising before nitriding is the only method of increasing both the depth of penetration and the concentration of nitrogen at the surface of the steel. H. Bennek and 0. Rudiger 38 have studied the nitriding of steel in a glow discharge in nitrogen and have reported greater hardness and a slightly deeper case than were obtained in the normal nitriding process. The difference in the stability of nitrides and carbides is probably not unconnected with the diatomic character of elementary nitrogen and the polyatomic character of elementary carbon.Carbonitrides In addition to carburising and nitriding steel can also be case-hardened by immersion in a bath of molten sodium cyanide.3g A bath made up of sodium cyanide 30-45% sodium carbonate 40-37% and sodium chloride 30-18y0 is maintained a t about 870° and both carbon and nitrogen atoms penetrate into steel immersed in the molten mixture. There is no reason to believe that the carbon and nitrogen penetrate into the steel as cyanide ions as these would tend to form an ionic lattice with the iron ; it is probable that they penetrate as individual atoms forming a metallic interstitial alloy by simultaneously occupying different interstices. Still another process of case-hardening termed " dry cyaniding " or " carbonitriding " has been developed recently 40 ; this consists in heating the steel in an atmosphere (H2 40 ; CO 20 ; and N, 20%) to which controlled amounts of ammonia and methane have been added.It is carried out at a temperature somewhat lower than that used in carburising and it is claimed produces a case which is deeper and of greater wear-resistance than a carbide case. Presumably both carbon atoms from the methane and nitrogen atoms from the ammonia penetrate into the steel and simultaneously occupy different interstices. Until recently little fundamental knowledge of ternary interstitial solid solutions of this kind was available. G. J. Fowler 41 reported that when carbon monoxide was passed over 35 Nature 1928 122 962 Nova Acta SOC. Sci. Upsal. 1929 [iv] 7 1. 36 Proc. Roy. SOC. 1948 A 195 34. 3BIbid.p. 61. 3s H. Carpenter and J. M. Robertson op. cit. p. 1123. 37 Arch. Eisenhiittenw. 1944 18 57. P1 J. 1901 79 285. W. H. Holcroft Metal Progr. 1947 52 380. 168 QUARTERLY REVIEWS iron nitride (Fe,N) carbon dioxide but no cyanogen was formed ; this result indicated that the carbon deposition reaction 2CO = C + CO, took place on the surface of the nitride. A. Fry 32 suggested the possibility of the existence of carbonitrides. A. Bramley 42 studied the simultaneous diffusion of carbon and nitrogen into steel by carrying out gaseous cementa- tions in the vapours of pyridine and methyl cyanide ; he found that nitrogen diffuses into steel in much the same manner as carbon. W. Koster 43 has studied the properties of iron supersaturated with both carbon and nitrogen with particular reference to theories of age-hardening.Several carbides take up nitrogen to form complex compounds containing both carbon and nitrogen. Unless nitrogen is rigorously excluded in the preparation of alkaline-earth carbides carbon-nitrogen complexes are formed. According to T. A o ~ o ~ ~ with calcium the cyanide is first formed ; it then decomposes to form the cyanamide and free carbon. C. H. Prescott jun. and W. B. Hincke 45 studied the interaction of aluminium nitride and solid carbon in the temperature range 1774-1909" K. 4A1N + 3C = Al,C + 2N M. von Stackelberg E. Schnorrenberg R. PauIus and K. F. Spiess 46 found that in the presence of excess of nitrogen aluminium and carbon interact a t 1800" to form the compound AI,C,N the appearance and properties of which scarcely differ from those of the carbide AI,C ; they consider that this carbonitride is an intermediate stage in the complete nitriding of the carbide 5A1,C3 + 10N = 4AI,C,N + 3C + SN = 20AlN + 15C M.von Stackelberg and K. F. Spiess 47 investigated the crystal structure of this carbonitride by X-ray diffraction ; they suggested that the lattice consists of five planes of aluminium atoms arranged hexagonally with three planes of carbon atoms and one plane of nitrogen atoms alternately between them. This view is speculative ; the crystallographic similarity of nitrogen and carbon atoms suggests that a more random distribution of these atoms would obtain. I n the ternary system titanium-carbon-nitrogen compounds are formed usually termed titanium cyanonitrides which are highly refractory and show an intense metallic lustre.Well-developed copper-coloured cubic crystals of this material are sometimes found in blast furnaces which have been smelting titaniferous iron ores. The substance has also been reported in meteorites. F. A. Bannister 48 has examined by X-rays speci- mens of both meteoric and terrestrial origin and confirmed that they are alloys of titanium carbon and nitrogen ; he concluded however that there is as yet insufficient evidence to prove the existence of a complete series of 42 Mem. Iron Steel Inst. Carnegie Schol. 1926 15 174. 43 2. Metallk. 1930 22 289 ; Arch Eisenhuttenw. 1929-30 3 553 637 ; Stahl 44 Bull. Chem. SOC. Japan 1941 16 91 106. 45 J . Amer. Chem. SOC. 1928 50 3228. 4 8 2 . physikal. Chem. 1935 175 127. u. Eisen 1930 50 254 629. 47 Ibid. p. 140.Min. Mag. 1941 26 36. RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 169 mixed crystals. C. Agte and K. Moers 5 have shown by X-ray diffraction the existence of mixed crystals in the systems TiC-TiN and TaC-TaN. H. Ste.4. Deville 49 claimed to have prepared niobium carbonitride by heating Nb,05 with sodium carbonate at 1200" in a graphite crucible ; he gave it the formula mNbN,nNbC. By a similar method of preparation A. Jolly 5O obtained the same product and stated that he obtained niobium oxycarbonitride by heating the pentoxide with ammonium cyanide in a graphite crucible. 0. Heusler 61 reported that when uranium carbide UC, was heated in nitrogen a t 1100" it was eventually converted into nitride; at higher temperatures (1500") mixtures of carbide and nitride were formed.The characterisation of compounds of this type without the use of X-ray- diffraction methods is unsatisfactory and it is possible that carbonitride phases were formed in these reactions. The above evidence suggests the close crystallographic similarity of carbon and nitrogen atoms and the possibility of the existence of a large number of hitherto unknown compounds and intermediate phases inter- stitial ionic and covalent containing both carbon and nitrogen. The existence of nitrogenous carbons 52 is further evidence of this similarity ; in fact the replacement of carbon by nitrogen or vice versa may occur in any crystal in which electronic compensation is possible. A detailed study of the formation and properties of iron carbonitrides has been carried out by K. H. Jack.53 Pure carbon monoxide was circulated over iron nitride powders heated a t various temperatures (450" 470° 500" and 600").The reaction 2CO -+ CO + C~atm~cl occurred a t the surface of the nitride; a t the lower temperatures there was apparently little tendency for the formation of graphitic carbon suggesting that the presence of nitrogen atoms in the interstices of the iron atom lattice tends to inhibit this reaction and thus lending support to the view advanced above that carbon formation is not a surface reaction but occurs within the iron lattice. The iron &-nitrides containing 31-33 atom-% of nitrogen treated in this manner at 450" gradually lost nitrogen and gained carbon at approximately the same rate ; the solid phase remained homogeneous until about three- quarters of the nitrogen atoms originally present had been replaced by carbon atoms.Throughout this interchange the structure of the solid phase remained similar to that of the parent &-nitride ; its iron-atom lattice showed the base-centred orthorhombic structure of the 5'-nitride. As the interaction with carbon monoxide was continued beyond the above stage a second solid phase appeared ; this proved to be an iron percarbide which was obtained in the pure state when all the nitrogen had been eliminated from the iron nitride. The percarbide ha.s a small range of composition (30-4-32.4 atom-% of carbon) which includes Fe,,C (C 31.0%). Although this percarbide is probably identical with that reported by previous workers (see above) SaH.L. Riley Quart. Reviews 1947 1 63. B9 Compt. rend. 1868 66 183.51 2. anorg. Chem. 1926 154 333. 63 Proc. Roy. SOC. 1948 A 195 41. Bull. Xoc. chim. 1868 [ii] 25 606. 170 QUARTERLY REVIEWS Jack was the first to prepare this compound in a pure state free from elementary carbon iron and cementite. He has suggested 54 that it has either an orthorhombic or a hexagonal crystal structure. The ?'-iron nitrides also interacted with carbon monoxide at 450" and 470" and were finally converted into the pure iron percarbide ; in these experiments how- ever the intermediate homogeneous carbonitride phase showed the &-iron nitride hexagonal structure. At 600" the &-iron nitrides reacted with carbon monoxide to give finally a mixture of iron percarbide and cementite and at 700" only cementite and graphitic carbon could be detected. The c-iron carbonitrides decomposed in a vacuum a t about 350" yielding E-carbonitrides which decomposed further at 450" to form 7'- or &-nitrides and iron percarbide or cementite according to the initial composition of the carbonitride.Iron percarbide or cementite heated in ammonia at 450" yielded c-carbonitride phases. During the interaction of the iron nitrides with carbon monoxide a t 450" a proportion of the nitrogen was eliminated from the solid phase as cyanogen and the remainder as elementary nitrogen. The variation of the lattice dimensions as interstitial nitrogen is replaced by interstitial carbon in the homogeneous phases is of considerable interest and importance. The carbon atoms in the c-carbonitrides almost certainly occupy the position vacated by the nitrogen atoms in the parent nitride ; this is suggested by the position of the superlattice lines in the respective X-ray powder diffraction diagrams.The anisotropic distortion of the iron- atom lattice which is characteristic of the formation of the [-nitride from the &-nitride shows an interesting modification in the analogous carbonitride conversion. As the ratio of carbon to nitrogen increases in the carbonitride phases a slight expansion of the lattice along the b-axis occurs; there is however a progressive decrease along the a and c axes which results in an overall decrease in volume. This result indicates that in these carbonitrides the atomic volume of the carbon is less than that of the nitrogen. Assuming the radius of the iron atom RFe to be 1.260 EX Jack finds the radii of the nitrogen atom R, and the carbon atom R, to be respectively 0.677 and 0.663 kX in the &-phases.These values are substantially smaller than the covalent radii of carbon (0.771) and nitrogen (0.70) atoms.55 H. Lipson and N. J. Petch 56 have attempted to locate the positions of the carbon atoms in cementite and given the following values as the Fe-C distances 2.15(2) 2.06(2) 1.89 and 1-85 EX i.e. an average of 2.03 EX which Lipson and Petch consider has more significance than any of the separate values. Taking 1-260 JcX as the radius of the iron atom the average gives 0-770 kX as the radius of the carbon atom i.e. a value ident*ical with the covalent radius. If however the smaller Fe-C distances are significaiit then values of Rc even less than those reported by Jack are indicated. Jack's values for Rc and RN and also the fact that he finds RN > R in these alloys are consistent with the view that the carbon and nitrogen are present in the 54 Proc.Roy. Soc. 1948 A. 195 56. 65 L. Pauling " Nature of the Chemical Bond " Cornell Univ. Press 1944 p. 164. 68 J . Iron and Steel Inst. 1940 142 95. RILEY CARBIDES NITRIDES AND CARBONITRIDES OF IRON 171 interstices not as neutral or negatively charged atoms but as positive ions and that they have lost electrons either to the incomplete 3d level in the iron atom or to an incompletely filled band in the alloy crystal. The higher electronegativity of nitrogen compared with carbon suggests that this transfer of electrons will occur less readily and therefore to a smaller extent with nitrogen than with carbon which may account for the smaller observed value of R,.The above value of Rc gives a radius ratio R R, of 0.54 which is distinctly less than Hagg's critical value of 0.59. W. Seith and 0. Kubaschewski 57 have shown that carbon in a steel wire heated to 1000" and under a potential gradient diffuses towards the cathode much more rapidly than towards the anode a result in keeping with the above views. Conclusion The covalent radius of the boron atom is 0-88 kX; its smaller electro- negativity however may bring about a loss of electrons and consequently a smaller effective radius in an interstitial alloy; in this respect the high electronic conductivity of interstitial borides reported by M ~ e r s ~ is sug- gestive. The existence of borocarbides boronitrides and perhaps boro- carbonitrides appears possible.The position of oxygen is also of interest ; its covalent radius is 0.66 k X Le. less than those of carbon and nitrogen. Its electronegativity is however much greater and the tendency will therefore be for it to form an ionic lattice rather than an interstitial alloy. Nevertheless it has been suggested 58 that solid solutions of ferrous oxide in metallic iron exist. R. Schenck 59 has gone further and postulated the existence of oxoaustenite a solid solution of carbon and oxygen in iron in equilibrium with a CO-CO atmosphere. H. Diinwald and C. Wagner,60 however from a study of the CO-CO equilibrium over carbon dissolved in iron found that at 800" a- and y-iron can co-exist in equilibrium and that the concentration of carbon in the a-phase is about 0.025~0 ; they calculate that the solubility of oxygen in a-iron a t 800" or in y-iron a t 1000" is less than 0.01%.It has been shown 61 that preliminary surface oxidation or nitriding of a steel specimen increases the velocity of its carburisation in carbon monoxide. This may be due merely to the " opening up " of the surface. The carbonitride results however suggest that the increase in velocity may be due to the initial presence of oxygen or nitrogen atoms in the interstices of the iron lattice facilitating the entry of carbon atoms. If this is so then an explanation of the catalytic activity of certain iron oxides in the deposition of carbon is suggested during the progressive reduction of the oxide by carbon monoxide a stage will be reached at which residual interstitial oxygen will probably still be present in the newly- formed iron lattice and this residual oxygen possibly facilitates the entry of carbon atoms.57 2. Elektrochem. 1935 41 551. J . Amsr. Chem. Xoc. 1924 46 892. A. Matsubara 2;. anorg. Chem. 1922 124 42 ; E. D. Eastman and R. M. Evans 5sZ. anorg. Chem. 1927 167 254 315. 6o Ibid. 1931 199 321. H. L. Riley's co-workers unpublished. 172 QUARTERLY REVIEWS The deposition of carbon from carbon monoxide is highly susceptible to catalytic influences ; e.g. W. Bauklok and G. Henke 62 have shown that the deposition can be decreased by as much as 95% by the addition of 1% of ammonia cyanogen or hydrogen sulphide ; it appears possible that interstitial phenomena are concerned in these inhibitions. The dispersion of the metal catalyst in deposited carbon the pitting of platinum catalysts in ammonia oxidation and the use of metal catalysts in reactions involving hydrogen all point to the importance of interstitial reactions in catalysis.S. Weller L. J. E. Hofer and R. B. Anderson 63 concluded that bulk cobalt carbide is neither an intermediate nor a catalytically active substrate in the Bischer-Tropsch synthesis. The interaction of carbon monoxide with iron nitrides suggests similar reactions with other metal nitrides and the existence of a large number of new carbonitride phases. Apart from the technical interest of the above carbonitride phases in the heat treatment and case hardening of steel their detailed crystallographic study offers a new approach t o the investigation of the metallic state. The quantum theory of solids has not yet been applied to interstitial alloys but there is little doubt that accurate data such as that given in K.H. Jack’s papers will open up this field. It is significant that both interstitial carbon and nitrogen have a great effect upon the a-y change point of iron ; it therefore appears possible that the measurements of the electronic specific heats 64 of carbides nitrides and carbonitrides would yield further valuable results. These possibilities are being investigated. Ba Illetallwirt. 1940 19 463. 64 See F. Seitz “ The Modern Theory of Solids ” McGraw-Hill New York 1940 6s J. Amer. Chem. Soc. 1948 70 799. p. 487.
ISSN:0009-2681
DOI:10.1039/QR9490300160
出版商:RSC
年代:1949
数据来源: RSC
|
5. |
Ionic solvation |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 173-180
J. O'M. Bockris,
Preview
|
PDF (692KB)
|
|
摘要:
By J. O’M. BOCIIRIS PH.D. (LECTURER IN ORCIANIU AND PHYSICAL CHEMISTRY IMTERIAL COLLEGE S.W.7) IN spite of much experimental work the significance of the term “ solvation number” has remained indefinite. This is partly because the various experimental methods of determination apparently give considerable differences in numerical values (Table I) and so have provided but little incentive to the calculation of absolute solvation numbers. Consequently more recent theoretical work on solvation has been directed principally to the calculation of solvation energy because this quantity is open to less ambiguous experimental determination than are the solvation numbers. Knowledge of absolute solvation numbers is of great importance in several ways. Such values are necessary to further the model of an ionic solution and for many detailed purposes ; for instance the theory of con- centrated electrolytic solutions awaits a knowledge of solvation numbers of definite significance.The large range of experimental phenomena which have been examined in attempts to determine solvation numbers experi- mentally illustrates the widespread influence of ion-solvent interactions. These phenomena include refractivity freezing and boiling points vapour pressure surface tension density specific heat and compressibility of ionic solutions transport number of ions and solubility of non-electrolytes in the presence of electrolytes. Lastly a need exists for the calculation of solvation numbers in the theory of reactions in solution. The object of the present work therefore is more closely to define the term “ solvation number ” and to attempt from a critical discussion of the results of the literature to find some order in the apparent chaos of conflicting results existing there.Former Concepts of Ionic SoIvation I n the earliest concepts of ionic solvation ions were regarded as chemically bound to water molecules forming hydrates in aqueous solution. This concept was supported by observations quoted by S. Aschkenasi upon the sudden changes of conductivity cryoscopic constants etc. with solution composition which he regarded as corresponding to the formation and decomposition of various definite hydrates. I(. Pajans’s consideration 2 of the polarisability of water molecules and the difficulty of substantiating the existence of definite hydrates in solution led to the concept that the interaction energy in solvation was entirely coulombic and did not involve covalent links.The molecules in the solvation sheath were regarded as continually interchanging with those in the bulk of the solution. M. Born showed that the value of the solvation energy arising from a very simplified interpretation of this model for univalent 1 2 . EZektrochem. 1922 38 106. lDeut. physikd. ##. 1919 21 709. 2. Physik 1920 1 45. 173 174 QUARTERLY REVIEWS salts was in substantial agreement with experiment. Some objection is made to the Fajans-Born concept of electrostatic solvation on the grounds that it would imply that the sequence of ionic velocities of a series of ions would be the same in different solvents whereas this is not always the case.4 This criticism was made before the possession of a “ structure ” by most polar solvents was realised; a consequence of this structure is that the steric properties of an ion become important in determining its solvation SO that the order of the ionic velocities for a given series of ions might be different in different solvents owing to the difficulty of an ion having a radius greater than some critical value “ fitting into ” the structure of certain ‘solvents.So long as the simpler ions of Groups IA and IIA of the Periodic Table and the halogens are considered it seems probable that the success of the electrostatic theory of solvation energy implies that the ion-solvent inter- action is almost entirely coulombic and that other types of interaction are not important in this connection. This conclusion is less certain for more complicated ions e.g.for those of Group VIII where strong co-ordinating power tends to favour covalent attachment between ions and solvent. There exists little quantitative evidence on the solvation of these ions. Acceptance of a coulombic view of the solvation of simple ions does not suffice for a satisfactory definition of solvation. It follows from the Born- Rajans concept that a certain number of solvent molecules will be firmly attached to the ion and that a further ion-dipole interaction takes place involving solvent molecules outside the inner ion-solvent entity. It may be suggested that the term ‘‘ primary solvation ” is used in reference to the attachment of solvent molecules to ions in solution for which (i) the ion and its attached solvent molecules move together as one entity during electro- lytic transport ; (ii) the solvent ions have completely lost their own separate translational degrees of freedom.‘‘ Secondary solvation ” may be suggested as the term referring to all electrostatic interactions which are not included in the definition of primary solvation given above. The term “total solvation” would then refer to the sum of these two types of solvation. It has been suggested 5.6 that “ chemical ” and “ physical ” should be respectively the description of ‘‘ closely ” and ‘‘ loosely ” attached solvent molecules. The term “ chemical ” appears however to be misleading when referring to coulombic solvation. A quite different theory of ionic solvation from the general electrostatic theory of Fajans was proposed by I?.A. Lindemann,7 who suggested that no permanent solvation sheath existed in solution and that the solvent transport during movement of an ion is caused by the transfer of momentum from the ion to the solvent molecules with which it collides. Greater transfer of momentum would be expected with smaller than with larger ions as is experimentally found. On this basis it wits possible to explain the depen- 4 C. A. Krass “ Properties of electrically conducting systems ” 1922 p. 203. H. Ulich “ Hand- und Jahrbuch der chemischen Physik ” 1933 6 ii-iii 186. 6 Idem 2. Elektrochem. 1930 36 504 ; cf. also J. Baborovskjr Chem. Listy 1934 28 242 ; 2. physikal. Chem. 1934 188 A 135. Trans. Paraday Soc. 1927 23 387. BOCKRJS IONIC SOLVATION 175 dence of the temperature coefficient of conductance upon ion size.No other independent evidence supports Lindemann’s theory which probably concerns a secondary effect (see below). Choice of Concordant Experimental Methods €or the Investigation of Primary Solvation The discrepancy between the results of various experimental methods is exemplified in Table I by values from the literature for the sodium ion. TABLE I Values of the hydration number of the sodium ion at inJinite dilution by various experimental methods Method Mobility (H. Ulich 8 ) . . . . . . . . . . Mobility (E. H. Riesenfeld and M. Reihoid lo) . . Dialysis (H. Brintzinger 11) . . . . . . . . Partial molar volume (E. Darmois 12) . . . . . Polarimetric ( 0 . 5 ~ ) (M. Cordier 13) . . . . Water transport (M. Brull s . . . Hydration number. 4 3 71 17 1 700 approx.* These typical large discrepancies imply that different experimental methods give a measure of different types of solvation.As the methods are roughly divisible into those giving lower (< 10) and higher values (> 10) of solvation numbers it is probable that the former give a measure of primary and the the latter a measure of some part of total so1vation.f To elucidate this supposition critical aspects of the main methods will be discussed; for which purpose they may be divided into the following groups according to the property measured (i) Ionic and solvent transport (ii) activity of solvent (iii) activity of non-electrolyte solute (iv) ionic entropy (v) density of solutions (vi) miscellaneous properties. Difficulties in Methods.-A difficulty common to all methods except Ulich’s ionic mobility method 8 and P.Debye’s proposed supersonic e.m.f. method l4 is that of dividing up the solvation of the salt between its ions. Some workers l5 have assumed that large organic ions mch as Ph-NH,+ possess zero solvation numbers. Less justifiable assumptions (e.g. that 8Trans. Faraday Soc. 1927 23 392. l 1 2 . anorg. Chem. 1935 223 101. l3 J . Chim. physique 1946 43 54. l* J . Chem. Physics 1937 5 13 ; cf. also Okay Proc. Math. Phys. SOC. Japan 1933 l5 E.g. H. Remy Trans. Paraduy Soc. 1927 23 381. Gazzetta 1934 64 624. loZ. physikal. Chem. 1909 66 672. la J . Phys. Radium 1941 2 2. 15 415. * From the value 1111 for NaCl (see p. 179). Methods which measure more than the primary solvation would be expected to show less mutual agreement than those which measure the primary solvation because of the less definite nature of secondary solvation.Moreover some methods e.g. the salting-out method doubtless depend upon other factors than the interaction between ion and solvent. M 176 QUARTERLY REVIEWS solvation of NO,- l6 and of I- l7 is zero) have been made by other workers. The most reasonable method appears to be division in the ratio of the ion sizes based upon some salt such as potassium fluoride where the ionic radii are nearly equal. Principal Methods .-The method of ionic m solvent transport includes several distinct approaches. The inert-reference-substance method l** l9 is the best known. The fundamental assumption of this method is that the reference substance does not move under the influence of the applied field. P. 2. Fischer and T.E. Kova1,20 however have recently reported the tram- port of raffinose under these conditions. It also follows that a local change in concentration of the reference substance would occur in the vicinity of the ion (salting-out). I n the Remy-Baborovsky method 21 of water transport the liquid is intended to pass through a thin membrane between catholyte and anolyte and thus indicate the net volume change arising from the different solvation of cation and anion. The principal objection to this method is the probable inclusion in the measured water transport of water which is transported by a transfer of momentum from ions to molecules in the manner suggested by Lindemann 7 (streaming effect cf. Ulich 6). Values of solvation number resulting from the method would therefore be too high.Other difficulties of the method exist. For instance there is a considerable volume change in the catholyte and anolyte caused by reactions at the electrodes. Some correction for this term has been attempted by H. C. Hepburn.22 Further the method is experimentally of doubtful validity because below concentra- tions of 1 ~ . the effect of electroendosmosis at the membrane is considerable. It follows from the above that the Remy-Baborovskf method measures some part of the total rather than the primary solvation and this conclusion is doubtless also valid for Jander’s diffusion method 23 and Brintziger’s 16 F. Bourion et al. Compt. rend. 1937 204 1420. 1 7 M. Cordier ibid. 1942 142 707. 18G. Buchbock 2. physikal. Chem. 1906 65 563. 19E. W. Washburn J . Amer. Chem. SOL 1909 31 322. 20 Bull.Sci. Univ. Kiev 1939 No. 4 137. 21H. Remy 2. physikal. Chem. 1915 89 529; J. Baborovskjr Rec. Tyav. chim. 1923,42,229 ; H. Remy 2. physikal. Chem. 1925,118,161 ; 1926,124,394 ; Baborov- skj. ibid. 1927 129 129 ; J. Baborovskf and J. VelGek Chem. Listy 1927 6 227 ; H. Remy and H. Reisener 2. physikal. Chem. 1927,126 161 ; A. Wagner Chem. Listy 1927 21 543 ; Baborovsk9 Velsek and Wagner J . Chim. physique 1928 25 452 ; Baborovskjr and Wagner ibid. p. 452 ; Chem. Listy 1929 23 97 ; Wagner ibid. p. 349 ; 1931,25 349 ; H. C. Hepburn Proc. Physical SOC. 1932,44,267 ; Bab.orovskjr and Wagner Coll. Czech. Chem. Comm. 1931 3 5 3 ; Baborovskj. 0. Viktorin and Wagner ibid. 1932 4 200 ; Baborovskj. Arhiv Hemiju 1932 6 85 ; Baborovskj. and Viktorin Coll. Czech. Chem. Comm. 1932 4 155’; 1933 5 518 ; Wagner Chem.Listy 1933 27 481 ; Baborovskjr and Viktorin &id. 1934 28 168 ; Baborovskj. ibid. 1936 30 5 ; Baborovskj. and G. Koudela ibid. 1938 32 5 ; Baborovskj. Trans. Electrochem. Sac. 1939 35 13. 22 Phil. Mag. 1938 25 1074. 23 G. Jander and A. Winkel 2. physikal. Chem. 1930 149 97 ; H. Brintziger and C. Ratanarat 2. anorg. Chem. 1935 222 113 ; G. Jander and H. Mohr 2. physikaE. Chem. 1942 190 A 81. BOCKRIS IONIC SOLVATION 177 dialysis method,ll for both these methods give high results usually of the same order as those obtained from Remy’s method. Ulich’s ionic mobility method8 appears to measure a more defkite quantity than the above methods because it obtains directly the volume of the aolvated ion from an application of Stokes’s law to ionic mobilities at very low concentrations so that the number of water molecules which migrate permanently attached to the ion (primary solvation number) is obtained.The method is somewhat limited by the non-applicability of Stokes’s law to some ions in aqueous solution.24 Other objections recorded in the literature appear to have little force ; for instance the use of crystal radii for calculating the volume of the ion in the solvent sheath involves some loss of accuracy but the difference between crystal ionic radii evaluated by several methods are not large enough to make this a serious one. The appropriate volume of the solvent molecule to use in the evaluation of the solvation number is somewhat more indefinite. Owing to the high pressure in the vicinity of the ion and the inhibition of rotational degrees of freedom of the dipoles attached to the ion the molecular volume calculated in the usual way is too large.Calculation of the solvation number on the basis of an arrangement of close-packed spheres with the usual (uncompressed) volume for the solvent molecule was taken by Ulich to give a lower limit for the solvation number. The true value should be higher than this because of electrostriction of the solvent molecules in the neighbourhood of the ion and assuming arbitrarily that this caused the solvent volume to be reduced to about half it’s original value Ulich calculated an upper limit for the solvation number. The values accepted here from the method are the mean of these two estimates. Methods depending upon Activity Determinations e.g. cryoscopic,26 vapour pressure,26 and ~alting-out,~’ all give relatively high results but are similar in representing a “ total interaction effect ” and would therefore be expected to measure both primary and secondary solvation which view accords with the high results obtained.The cryoscopic method 25 appears to be of doubtful validity because the results depend upon the non-electrolyte used ; e.g. acetone and paraldehyde show unreasonable results whilst resorcinol appears to function satisfactorily. The theory by means of which the results are evaluated is extremely inadequate and a satisfactory theory for the complicated systems used would be very difficult. The change in solubility of gas or solid upon the addition of 24 Cf. G. Kortiim “ Lehrbuch der Elektrochemie ” Leipzig 1948 p. 130. 2 5 F. Bourion and E.Rouyer Compt. rend. 1933 197 52 ; Bourion Rouyer and 0. Hun ibid. 1933 196 1015 ; Bourion and Rouyer ibid. p. 1111 ; 1933 197 52 ; 1935 201 65 ; Rouyer ibid. 1934 198 742 1490 ; Hun ibid. p. 740 ; Rouyer ibid. p. 1156 ; Bourion and Rouyer ibid. 1935 201 65 ; Bourion and Hun ibid. 1936 202 2149 ; Hun ibid. p. 1779 ; Bourion Rouyer and Hun ibid. 1937 204 1420 ; F. H. Getman Rec. Trav. chim. 1937 56 1025 ; Bourion and Rouyer Compt. rend. 1937 205 662 ; Bourion and Hun ibid. 1937 204 1872 ; Bourion and Rouyer {bid. 1938 206 1967. 26 B. H. Van Ruyven Rec. Trav. chim. 1937 56 111. 2T P. P. Kosakewich and N. A. Ismailow 2. physilcal. Cham. 1930 150 A 308. 178 QUBTERLY REVIEWS electrolyte has been used to obtain a measure of hydration on the assumption that the decrease in solubility on addition of a salt is due to a reduction in the activity coefficient of water.27 This analysis is also over-simplified ; a formally correct picture for very dilute solutions was given by P.Debye and J. McAulay.28 Essentially the simpler analysis omits to take account of the effect of the non-electrolyte solute on the dielectric constant of the medium. Values calculated from this method also depend upon the nature of the non-electrolyte 27 and cannot therefore be considered as valid for determination of ionic solvation. However it is sometimes possible to make qualitative inferences regarding solvation from salting-out studies. E’or example J. O’M. Bockris and H. Egan 29 found that for concentrations of sodium chloride between ~/1000 and N (non-electrolyte solute benzoic acid) the salting-out appears to decrease anomalously with decreasing dielectric constant in alcohol-water mixtures containing an excess of water.This fact can be explained if it is assumed that the sodium chloride is pre- ferentially solvated with water. N. Bjerrum 30 calculated solvation numbers from the activity of the solvent in a given solution. His method was considered by Ulich 5 to give primary solvation numbers but it does not seem possible to draw this definite conclusion owing to the small number of ions to which the method has been applied. E. S ~ h r e i n e r ~ ~ using this method found the improbable result of 11 for hydrogen chloride in this way. Further R. H. Stokes and R. A. Robinson’s use 32 of an improved form of this method to derive an empirical relation between the activity coefficient and concentration for electrolytes in relatively concentrated solutions involves some consequences concerning hydration which are contrary to the general trend of results obtained by other methods in dilute solutions.The entropy decrease occurring when gaseous ions undergo solution in water a t infinite dilution can probably be used to calculate primary solvation numbers because it depends upon the “freezing u p ” of the degrees of freedom of the water molecules near the ion.33 The assumption is made that the entropy decrease when water is transferred from the bulk of the solution to the primary solvation sheath is the same as that on freezing water. Values from this method agree very well with those calculated from Ulich’s ionic mobility method.8 The entropy method can also be applied to the calculation of the solvation of non-electrolytes.34 Lastly special interest has been attached to methods of determining solvation numbers from measurements of the partiaZ molar volume of elec- trolytes a t infinite dilution. J. D. Bernal and R. H. Fowler 35 refer to this method and its results as being particularly satisfactory but give no e8 2. physikal. Chem. 1925 26 23. 2D Experientia 1947 3 11 ; Trans. Paraday Soc. 1948 44 151. 30Z. anorg. Chem. 1920 109 275. 31 Ibid. 1924 135 333. a 2 J . Amer. Chem. Soc. 1948 70 1870. 33 H. Ulich 2. Elektrochem. 1930 36 497; 2. physikal. Chem. 1934 168 141. 34 J. Kielland J . Chem. Educ. 1937 14 412. 35 J . Chem. Physics 1933 1 515. BOCKRIS IONIC SOLVATION 179 details as to how it might be applied.Darmois l2 describes the procedure and results of this method for a large number of simple aqueous electrolytic solutions. It is clear that as the method is concerned with the volume of the ion-solvent complex itself and not with the surrounding water molecules which are not firmly attached to the ion it should measure the primary solvation. However the results obtained are substantially lower than those of the other concordant methods for obtaining primary hydration numbers. Other Methods.-These involve measurements of reaction velocity the variation of the distribution of an electrolyte between two immiscible solvents the compressibility of ionic solutions and the use of supersonic e.m.f.s. The first method has been used recently by Cordier l3 and yielded extremely high results (of the order of 1 0 2 ) .Such values sometimes imply that more water is used up in the hydration sheath than exists in the solution e.g. this is so for a solution containing 0.5 g.-ion/l. of sodium chloride for which substance a t this concentration the solvation number is 1111 according to Cordier.l3 The distribution method 31 is connected with several unsolved difficulties the results of J. N. Sugden 36 and of E. Angelescu and 0. Dutchievici 37 are discrepant and Sugden’s values yield negative solvation numbem for anions which were supposed to exert a “depolymerising effect ” upon water regarded by him as (H,0)3 in the liquid state. Measure- ment of compressibility 38 appears to be a satisfactory method of obtaining solvation numbers results from it agree quite well with those of ionic mobility * and entropy 33 methods and it seems probable therefore that this method also yields primary solvation numbers.This is supported by the concept that the solvent sheath in primary solvation is under very large pressure owing t o the field near the ion and therefore reaches a limiting value beyond which further compressibility is due to that of the bulk of the solvent. Lastly Dehye’s theoretical treatment l4 of the p.d. produced in an electrolytic solution when it is subjected to supersonic vibrations appears to provide a method without objection in principle and yielding individual solvation numbers as the e.m.f. produced is a function of the ratio of the masses of both positive and negative ion-solvent complexes in the electrolyte. The e.m.f. involved is of the order of volt; so that measurements which do not yet seem to have been attempted are likely to be difficult.The above survey shows therefore that the following methods probably measure primary solvation Ionic mobility ionic entropy compressibility and density. The agreement between these methods is relatively satis- factory compared with the very considerable disorder which is apparent when the results of various methods are taken without discrimination (see Table I). Numerical results of these four methods are shown in Table 11. There is little experimental information concerning the concentration The reason for this is not clear at present. 36 J . 1926 1742. 37 Bull. SOC. chim. Romania 1932 14 46. 38A. L. T. Moesveld and H. T. Hardon 2. physikal. Chem. 1931 155 A 238 ; A. Passinski Acta PhysicOchimica 1938 8 385.180 QUARTERLY REVIEWS variation of solvation which is small for methods measuring primary solvation (contrast the large concentration variation given by the oryoscopic 25 and the water-transport methods 21). TABLE I1 Results of experimental methods giving primary solvation numbers at 2 1 0 0 0 0 in$nite dilution 4 3 2 1 3 2 0.7 Ion. Li+ . . . Na+. . . K + . . . Rb+. . . F - . . . c1- . . * 1- . . . Mobility.$ (1.) 6 4 4 * 4" 0.7 * Entropy (2.) 5 4 2 2 5 2 0.5 Compressibility.? (3.) 3 4 3 5 3 2 Density. Mean of (4.) 1 (1 2 3 4). Mean of (1 2 3). 4 4 3 2 5 3 0.6 * Extrapolated for aqueous solutions from Home results in alcoholic solutions t Calculated for individual ions on the approximate basis that K+ and C1- are 3 Mem of upper and lower limit vaIues. Stokes's law not being applicable in aqueous solution of these ions. equally hydrated. Summary.-( 1) Previous concepts of ionic solvation are discussed and the de6nition of two types of solvation primary and secondary is given. (2) A critical discussion of the experimental methods for the determina- tion of solvation numbers show8 that there are several methods of measuring '' primary hydration numbers " which give concordant and rational values.
ISSN:0009-2681
DOI:10.1039/QR9490300173
出版商:RSC
年代:1949
数据来源: RSC
|
6. |
Some aspects of pyrimidine and purine chemistry |
|
Quarterly Reviews, Chemical Society,
Volume 3,
Issue 2,
1949,
Page 181-207
B. Lythgoe,
Preview
|
PDF (2055KB)
|
|
摘要:
SOm ASPECrS OF PYRIMIDINE AND PURINE CHEMISmY By B. LYTHGOE M.A. PH.D. (LECTURER IN ORGANIC CHEMISTRY UNIVERSITY OF CAMBRIDGE) THE chemistry of pyrimidine (I) and of condensed systems containing the nucleus such as purine (11) quinazoline (III) pteridine (IV) and the hypothetical isoalloxazine (V) has been developed largely because certain derivatives are important constituents of living organisms. Amongst natural pyrimidines are orotic acid (2 6-dihydrouypyrimidine-4-carboxylic acid) which occurs in milk and vitamin B,.1 The naturally occurring purines such as uric acid and the adenine and caffeine groups were among the earliest investigated natural products ; more as the co- dehydrogenases and co-phosphorylases recently coenzymes such have been identified as adenine derivative^.^ The alkaloid vesicine is a quinazoline compound and vitamin B (riboflavin) It is only since 1940 that the chemistry of the pteridine system has become known; 4 its derivatives occur in insect wing-pigments and representatives such as pteroylglutamic acid (VI) have vitamin activity and function in haemato- poiesis.The complex cell-constituents known as nucleic acids have a biological significance scarcely inferior to that of proteins and polysac- charides; they are built up from pyrimidine and purine compounds. In view of the part played in cellular economy by such compounds it is not surprising that pyrimidine derivatives should be known which have important pharmacological activity. The barbiturate hypnotics which 1 See H. R. Rosenberg " The Chemistry and Physiology of the Vitamins " New York 1942.* See Mayer and Jacobson " Lehrbuch der Organischen Chemie " Berlin and Leipzig 1923. 4 See J. C. E. Simpson ibid. 1946? 43,250 ; M. Gattea Chem. Reviews 1947,41,.63 ; T. D. Spies Ann. Rev. Biochm. 1947 16 387. contains the isoalloxazine system. See B. Lythgoe Ann. Reports 1945 42 175. 181 182 QUARTERLY REVIEWS have been known for many years are used very widely in clinical practice. Sulphadiazine (VII) and its 4-methyl and 4 6-dimethyl derivatives are out- standingly valuable amongst sulphonamides 5 in the treatment of bacterial infections. During the war years an intensive study of pyrimidine deriva- tives was carried out with the object of obtaining new antimalarials ; some representatives with marked activity were found though none with the outstanding value of the diguanide derivative " Paludrine " discovered as a sequel to this work.* The antithyroid activity of thiouracil (6-hydroxy-2- mercaptopyrimidine) and the action of alloxan in causing diabetes through the destruction of the pancreatic islets are recently investigated topics.' On the chemical side a major part of our knowledge of the group con- sists of synthetic methods.This term is intended to convey not only the methods by means of which the pyrimidine nucleus can be built up but also the numerous transformations which substituent groups can undergo ; much of this knowledge has been gained in the work on the various biolo- gically important compounds mentioned above. Apart from synthetic and biochemical work there has been little important activity in the purine field during the last twenty years.Although the systematic chemistry of the pyrimidine nucleus has never been given the attention which it deserves sufficient information has now been accumulated to make a logical treatment possible ; the main deficiencies concern the simpler compounds ; for example pyrimidine itself has probably never been obtained in amounts of more than a gram and its 5-hydroxy-derivative is unknown. The practice of interpreting the behaviour of related heterocyclic systems in terms of modern theories of tautomerism and resonance has made progress in recent years and there are signs that pyrimidine compounds are amenable to similar treatment. The way in which substituents interact with each other and with the nucleus presents some unusual features and in hydroxy- and See E.H. Northey " The Sulphonamides and Allied Compounds " London and New York 1948. 6 See K. C. Blanchard Ann. Rev. Biochem. 1947 16 595. 7 See W. T. Salter ibid. 1945,14 570 ; C. F. Cori and G. T. Cori ibid. 1946 15 203; D. Stettm ibid. 1947 16 136. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 183 amino-derivatives interesting structural problems arise. These aspects will be given prominence in the present survey. Relationship of Pyrimidines and Purines to Nucleic Acids.-Some of the recent work on pyrimidine and purine derivations owes its initiation to the biological importance of nucleic acids and related compounds. Some aspects of the biochemical investigations on nucleic acids have been col- lected in a recent symposium ; 8 of these only one or two can be mentioned here.0. T. Avery C . M. McCarty and M. McLeod have shown that the type-transforming factor of pneumococci is a nucleic acid of the deoxyribose type. J. Brachet and T. Caspersson lo independently brought forward evidence implicating nucleic acid in protein synthesis. Nucleic acids are present in combination with proteins in self-duplicating systems such as chromosomes and viruses ; l1 C. D. Darlington l2 has described the acids as having the function of “ the molecular midwife of all reproductive par- ticles ”. The association of nucleic acids with proteins in these particles is especially significant in view of the fact that closely related nucleotides and dinucleotides act in conjunction with phorylase and dehydrogenase systems. specific proteins in intact phos- r0- c H - & H( o ti8 2s c H c H,.o H The nucleic acids l3 are macromolecules ; they are polynucleotides built up from nucleotide units in a way not yet fully understood. There are two main types namely ribonucleic acids which are cytoplasmic constituents and deoxyribonucleic acids ; it is the latter type which occurs in cell nuclei. The nucleotides obtained by the breakdown of both types appear to be the 3’-phosphoryl derivatives of compounds known as nucleosides ; the latter are N-glycosides of purine or pyrimidine derivatives. Those from ribo- nucleic acid contain ribose as the sugar component and are guanosine (VIII ; R = NH, R’ = OH) adenosine (VIII ; R = H R’ = NH,) cytidine (IX ; R = NH,) and uridine (IX ; R = OH). The nucleosides from deoxyribonucleic acid are probably similarly constituted but the sugar component is 2-deoxyribose and thymine (5-methyluracil) is present instead of uracil as the aglycone in one of them.Clarification of the structures of the nucleoside units was clearly a necessary preliminary to constitutional work on the macromolecular acids. Much work directed to this end was done before 1940 using degradative Nucleic Acid. Symposia SOC. Exp. Biol. 1947 Vol. I. lo Ref. 8 p. 127; ibid. p. 207. l3 See J. M Gulland ref. 8 p. 1. * J. Exp. Med. 1944 79 137. l1 See A. E. Mirsky Adv. Enzyrnol. 1943 3 1. l2 Ref. 8 p. 267. 184 QUBTERLY REVIEWS methods which have been reviewed elsewhere. l4 Synthetic approaches 15 have recently made it possible to prepare the four ribonucleosides from compounds of known structure and as a consequence their constitutions can now be regarded as established.The extension of similar methods to nucleotides and dinucleotides should in the future help to throw light on the way in which the nucleotide units are united in the polynucleotide molecule. It is probable that the internucleotidic linkages in the latter are of the phosphate ester type but the molecular position linked in this way to the sugar hydroxyl at position 3’ of an adjacent nucleotide is not yet known. When this has been ascertained the problems remaining for solution are similar to those arising in other macromolecules whether a straight chain or branched structure is present ; the relative number and disposition of the different nucleoside units ; the identity of end groups ; and so on. It would be digressing too far from pyrimidine and purine chemistry t o consider here the various chemical and enzymatic degradation methods by which these problems are now being studied but one aspect may be mentioned.The advances made recently in the investigation of protein structures owe much to the development of convenient methods for the separation and determina- tion of the component amino-acids and the solution of similar problems in the field of pyrimidines purines and their nucleosides would clearly be of equal value for constitutional research on nucleic acids. These problems are now being studied and the techniques emerging seem of considerable promise. A few years ago it was assumed as a result of earlier inexact work that equimolecular amounts of each of the four nucleosidea were present in ribonucleic acid (the “ statistical tetranucleotide ” concept) and in the absence of criteria for their identification it was impossible to say with confidence whether we should speak of one ribonucleic acid or of many.The preliminary results from the newer methods of estimation indicate that samples of acid from different sources may contain different proportions of the four nucleosides.18 It seems in fact that the acids will eventually be found to have a manifold and highly organised nature. Biogenetic Aspects.-Cellular activities are so dependent upon purine and pyrimidine derivatives that the fundamental question of their origin is provoked what powers of synthesising them do various organisms possess and what compounds are involved as intermediates in these biosyntheses ? Experiments using compounds marked with isotopic nitrogen (I5N) have now shown that in mammals and birds very considerable powers of syn- thesis are present.In the rat tissue purines and pyrimidines are probably mainly of endogenous origin ; thus although dietary adenine is used to some extent for nucleic acid synthesis guanine is not used in the same way.17 14 See B. Lythgoe Ann. Repurts 1944 41 200. l5 B. Lythgoe and A. R. Todd ref. 8 p. 15; A. R. Todd J. 1946 647. l6 H. S. Loring G. L. Ordway and J. G. Pierce J. Biol. Chem. 1948 1’76 1123 ; l7 G. B. Brown P. M. Roll and A. A. Plentl Pd. Proc. 1946 6 517 ; A. A. Plentl E. Vischer and E. Chargaff ibid. p. 715. and R. Schoenheimer J Bid. Chem. 1944 153 203. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 185 F. W. Barnes and R. Schoenheimer 18 have found that when 16NH is administered to rats and pigeons the marked nitrogen atom is incorporated rapidly into the tissue nucleic acids and the excreted uric acid; that is the purines and pyrimidines like the proteins are in a state of continual degradation and resynthesis.It had been suggested in the past that the ureide systems in purines such as uric acid might originate in urea and that amino-acids such as histidine might provide a source of the iminazole ring of the purine nucleus. The work with isotopically marked materials has refuted these suggestions and points to the view that purines are built up in vivo from small metabolic units. In pigeons and humans uric acid is synthesised in part from glycine the atoms of which are incorporated as N of the glyoxaline nucleus and C and C of the pyrimidine nucleus respectively ; glycine is similarly utilised for the synthesis of the nucleic acid purines in yeast.We now know some- thing of the origin of the remaining carbon atoms of uric acid; pigeons incorporate the labelled atoms of H*13C0,H and CH,*WO,H but not that of WO, into both the ureide systems (i.e. C and C,) ; C of the pyrimidine ring is probably derived from C0,.19 For the success of studies of this kind convenient synthetic methods are needed to prepare labelled pyrimidine and purine compounds from the available isotope source and convenient degradative methods are necessary so that a t the conclusion of the bio- chemical work the locus of entry of the isotopes into the heterocyclic mole- cules may be determined an interesting example of the dependence of biochemical work on earlier chemical studies.For securing a more detailed insight into the processes of nucleic acid biosynthesis a t least two other techniques are available. One of these the study of isolated enzyme systems may be illustrated by the work of H. M. Kalckar.20 He has shown that in mammalian tissue a mechanism exists for the. phosphorolytic synthesis of purine nucleosides from the purine aglycones. Rat liver contains an enzyme which catalyses the reversible change Inosine + inorganic phosphate + hypoxanthine + ribose- 1 phosphate By using the reverse reaction a physiological synthesis of inosine and guanosine has been effected. A related enzyme has been isolated from muscle by S. P. Colowick and W. H. Price,21 and it is probable that enzymes capable of synthesising pyrimidine nucleosides may also be present in mammals.The second method attempts to explore the synthetic pathway between the small metabolic units and the nucleic acid end-products by a study of micro-organisms which lack the enzymic equipment required to carry out one or more of the necessary synthetic steps. Work of this kind is expand- ing rapidly a t the present time and the results obtained to date should ls Im. 1943 151 123. l o J. C. Some J. M. Buchanan and A. M. DeUuva ibid. 1948 173 69 81 ; *O Ref. 8 p. 38. R. A b r m E. Hammarsten and D. Shemin ibid. p. 429. *l Fed. Pros. 1946 5 130. 186 QUARTERLY REVIEWS perhaps be regarded as suggestive rather than conclusive until the complete picture emerges in sharper definition. They do however justify us in tak- ing it as a working hypothesisrthat in nucleic acid biosynthesis there is a pathway whereby sugar-free pyrimidines and purines are first synthesised from as yet unidentified intermediates and then converted into nucleosides ; in the first of these steps and possibly in both vitamins of the B-group are implicated as coenzymes or coenzyme precursors.On this hypothesis the fact that certain micro-organisms require for growth purines such as adenine or pyrimidines such as thymine or uracil is interpreted as indicating that the equipment necessary to synthesise these aglycones has been lost although the power to transform them into nucleosides remains. One very interesting case of this type is an induced mutant of E. coli which requires p-aminobenzoic acid for growth. This requirement can be dispensed with if an adequate supply of thymine purines and methionine is provided; growth under these conditions is subculturable and is not inhibited by sulphonamides.22 In view of the competitive relationship known to exist between p-aminobenzoic acid and sulphonamides and of other circumstantial evidence such as the fact that yeast grown in the presence of sulphonamides has a reduced nucleic acid c0ntent,~3 the behaviour of E.coZi appears significant. It is suggested that p-aminobenzoic acid or some related compound functions as a coenzyme in the synthesis of purines and thymine ; there is some evidence that the active form of p-aminobenzoic acid may be a compound of the pteroylglutamic acid type.24 W. Shive and his co-workers 25 point out that if sulphonamides act by inhibiting purine synthesis the intermediate compound whose further transformation is prevented may accumulate in the culture medium.Under such conditions E. coli synthesises a base 4-aminoglyoxaline-5-carboxy- amide and it was suggested that p-aminobenzoic acid or a related compound functions in converting this base or one of its close relatives into purine compounds. A physiological synthesis of this kind would be paralleled by chemical syntheses of the pyrimidine ring of Type 111 which are discussed in a later section. Other micro-organisms are known which fail to grow satisfactorily when pyrimidines such as uracil and cytosine are provided but respond well to the intact nucleosides uridine and cytidine ; Neurosporu mutants provide examples of this class.26 This behaviour may mean that the organisms in question lack the power to convert the pyrimidines into their glycosides; alternatively it might be that they are derived from types in which uracil and cytosine are not normal intermediates in nucleic acid biosynthesis.Thymidine is required for growth by several Lactobacilli; the aglycone thymine is ineffective. In a strain of L. bctis which requires vitamin B, * 2 2 J. 0. Lampen R. R. Roepke and M. J. Jones J. Biol. Cl~em. 19-16 164 789. 23 W. H. Schopfer Experientia 1946 2 188. 34 See D. D. Woods Ann. Rev. Biochem. 1947 16 613. 25 J . Amer. Chem. SOC. 1947 69 726. 26 H. K. Mitchell and M. B. Houlahan Fed. Proc. 1947 6 506 ; H. S. Loring and J G. Pierce J . Biol. Chem. 1944 153 61. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 187 (the anti-pernicious anzmia factor of liver 27) for growth this factor can be dispensed with if adequate amounts of thymidine are supplied and this has been interpreted as indicating that vitamin B, functions as a coenzyme in the conversion of thymine into thymidine.2s It is clear that the views tentatively expressed above may require some revision or expansion as further experimental evidence accumulates but the methods described obviously have an important contribution to make to biogenetic studies.They hold out the promise that it may ultimately become possible to exercise some control over the activities of the cell nucleus with important implications for the treatment of bacterial and virus-borne infections perhaps even of cancer. Synthetic Methods Pyrimidines.-The earlier methods for the preparation of pyrimidine compounds are given in Mayer and Jacobson’s text-book,2 and need there- fore no detailed discussion here.Most of them involve two separate stages namely a synthesis of the ring-system followed by a process in which the substituent groups are transformed into those present in the desired com- pound. These processes can be illustrated by the following preparations of pyrimidine itself from barbituric acid (X) and of cystosine (6-amino-2- hydroxypyrimidine) (XI) N (X.) E t S C y NH CHO ___ E ~SCJ (a) POC 13 ~ Ets?) aq.HCL H O ~ J (b) NH3 N\ I I N H2 ,CHNa C0,Et OW PJ H2 NH2 (XI.) The principal transformations open to various substituents in the nucleus willlbe mentioned later ; the methods available for the formation of the nucleus can be classified into three main types according to the distribution of nitrogen atoms in the two components used Type I.Type n. Type IU. 27 E. Lester Smith Nature 1948 162 144 ; E. L. Rickes et al. Science 1948 107 396. 2 * E. E. SnelI E. Kitay and W. S . McNutt J . Biol. Chem. 1948 1’95 473 ; L. D. Wright H. R. Skeggs and J. W. Huff ibid. p. 475 ; W. Shive J. M. Ravel and R. E. Eakin J . Amer. Chem. SOC. 1948 ‘SO 2614. 188 QUARTERLY REVIEWS Examples of Type I are involved in the syntheses of pyrimidine uracil and cytosine already mentioned. The nitrogenous (left -hand) components can also be compounds such as thiourea guanidine and amidines. The second (right-hand) component may be ethyl malonate ethyl cyanoacetate malono- nitrile a @-diketone a @-keto-ester or an ab-unsaturated ketone. This type of synthesis was and remains the most versatile of all the methods.An example of Type I1 is the preparation of the derivative (XII) from phenyl isocyanate and aminomethyleneacetoacetic ester ; none of the methods of this type achieved much practical importance. Type I11 is illustrated by the formation of hexahydropyrimidine (XIII) from formaldehyde and 1 3-diaminopropane ; such syntheses also had little practical value. These three types of synthesis have all been used and extended in more recent work and have proved adequate for most needs. No revolutionary methods have been introduced but it is possible that in the future methods for the synthesis of hydropyrimidines 29 may play a more important part than formerly ; their dehydrogenation to true pyrimidines has not yet been given much attention. Some of the more useful extensions of the general methods will now be mentioned.The value of Type I methods has been increased by the use of com- ponents containing aldehyde groups or their equivalent. Such components allow a direct synthesis of derivatives containing unsubstituted 4 or 6 posi- tions which formerly had to be prepared by indirect methods. Components of the type mentioned include the nitro- and chloro-derivatives of malon- dialdehyde which give 5-nifro- and 5-chloro-pyrimidines ; 30 p-ethoxyacral- dehyde acetal which serves in place of malondialdehyde for the prepara- tion of 2-aminopyrimidine (XIV),31 and malic acid and its methyl derivative which are used in the presence of fuming sulphuric acid and give direct and 2gCf. M. Senkua J . A m r . Chem. Soc. 1946 68 1611.30 W. J. Hale and H. C. Brill ibid. 1912 34 82 ; R. 0. Roblin P. S. Winnek and 31 R. W. Price and A. Moos ibid. 1945 67 207. J. P. English ibid. 1942 64 567. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 189 easy access to uracil (XV) isocytosine (2-amin0-6-hydroxypyririmidine)~ and thymine. 32 Syntheses of Type I have been used extensively in the preparation of pyrimidine compounds related to vitamin B ; these compounds belong to the class in which position 6 is unsubstituted and accordingly ethoxymethyl- ene derivatives were found useful as starting materials. Thus the important nitrile (XVI) which is used in F. Bergel and A. R. Todd's synthesis of the vitamin,33 was prepared from acetamidine as shown below.34 It has also been obtained by a synthesis of Type I1 from acetiminoether and amino- methylenemalononitrile ; 35 this represents one of the few important modern applications of this type of synthesis Recently formamidine has been introduced as a component for syntheses of Type I ; its interest lies in the access which it gives to pyrimidine deriva- tives containing an unsubstituted position 2 which are otherwise difficult t o prepare.4 6-Dihydroxypyrimidine can be obtained with its aid but not 4 6-diaminopyrhidine since with malononitrile the initial reaction takes place a t the reactive methylene group ; the resulting aminomethylene- malononitrile (XVII) then undergoes a Type 11 reaction with a second molecule of formamidine to give 4-amin0-5-cyanopyrirnidine.~~ With benzeneazomalononitrile formamidine condenses normally and the resulting 4 6-diamino-5-benzeneazopyrimidine (XVIII) has been used as the starting material for a very convenient synthesis of adenine (p.192). 4 6-Diamino- pyrimidine (XIX) formerly obtained by indirect methods from barbituric acid is a compound of value for the synthesis of adenine nucleosides. It 32 D. Davidson and 0. Baudisch ibid. 1926 48 2379 ; R. 0. Roblin et a,Z. ibid. 1940 62 2002; 1946 68 2339. 33 A. R. Todd and F. Bergel J . 1937 364. 3 p R. Grewe 2. physiol. Chem. 1936 242 89. 35 0. Hromatka D.R.-P. 670,635. 38 G. W. Kenner B. Lythgoe A. R. Todd and A. Topham J . 1943 388 ; J. Baddiley B. Lythgoe and A. R. Todd {bid. p. 386. 190 QUARTBRLY REVIEWS has now become readily accessible by a synthesis of Type 111 in which ethyl formate and malondiamidine are condensed together. 37 More direct methods of preparing compounds such as sulphamethazine are now available.Sulphamethazine was first synthesised 3s by a procedure involving interaction of 2-aminopyrimidine with an arylsiilphonyl chloride. It has since been prepared directly by a Type I synthesis from sulphaguani- dine and acetylacetone. 39 Finally two rearrangement reactions for the preparation of uracil derivatives deserve mention. The hydantoin (XX) which is best prepared from the product of a Wohler synthesis on aspartic acid undergoes ring- enlargement on treatment with hot alkali giving orotic acid.40 The preparation of Uracil (XXI) by a Hofmann rearrangement of maleic diamide is not important as such but it forms the basis of a method for the synthesis of condensed pyrimidine systems to be mentioned later.Pyrimidine Nuc1eosides.-In 1930 G. E. Hilbert and T. B. JohnsonQ1 reported a preparation of 3-glucosidouracil by heating acetobromoglucose and 2 6-diethoxypyrimidine followed by removal of the protecting acetyl and ethoxy-groups by acid hydrolysis O E t O E t OH A number of pyranose analogues of uridine and cytidine have been obtained by this method ; the preparation of one of the natural nucleosides awaited only a method for the preparation of the furanose acetobromo-compound of D-ribose. This has recently been provided by G. A. Howard B. Lythgoe and A. R. Todd,42 whose synthesis of cytidine is shown on page 191 G. W. Kenner B. Lythgoe A. R. Todd and A. Topham J. 1943 574. sB W. T. Caldwell E. C. Kornfeld and C. K. Donell J. Amer. Chem. SOC. 1941 39 F. L. Rose and G. Swain J.1945 689. 40 H. K. Mitchell and J. F. Nyc J . Amer. Chem. SOC. 1947 65 1382. 41 Ibid. 1930 52 2001 ; G. E. Hilbert and E. F. Jansen ibid. 1936 58 60 ; G. E. 42 J . 1947 1052. 63 2188. Hilbert ibid. 1937 59 330. LYTEEGIOE PYRIMIDINE AND PURINE CHEMISTRY 191 I01 I01 c H [c H ( OAC 1; c H - c H 2+ o AC I B r .C H +c H (oA~I - c H c H,. OAC Eto9 N\ OEt O E t k H Purine Derivatives and Purine Nuc1eosides.-General methods of purine synthesis have undergone some improvements as a result of the interest aroused by the problems involved in the synthesis of purine nucleosides. W. Traube’s 43 method of synthe- sising purine derivatives from these intermediates has long been regarded as the most valuable and general method for obtaining such compounds. It will be recalled that in this method a 4-aminopyrimidine is converted through its 5-nitroso-derivative into the 4 5-diaminopyrimidine which is then cyclised by treatment with formic acid at elevated temperatures.One drawback to this method was the inaccessibility of the starting materials required for the preparation of purines unsubstituted in position 2 ; thus adenine had to be obtained by an unsatisfactory desulphurisation procedure from 2-mer captoadenine. An examination of this and other steps in Traube’s method has been made the object of which was to adapt the latter for the synthesis of adenine and hypoxanthine glycosides ; for this purpose mild procedures would be imperative in order to avoid the risk of hydrolysing sensitive intermediates. As a result the following improvements have been introduced.(i) By methods indicated in an earlier section pyrimidines bearing a hydrogen atom at position 2 have become accessible. (ii) The value of 2-methyl- thiopyrimidines for purine synthesis has been demonstrated ; after com- pletion of the synthesis the methylthio-group can be replaced by hydrogen by means of Raney nickel containing adsorbed hydrogen.44 (iii) A very satisfactory alternative to the introduction of the 5-amino-group by nitro- sation and reduction is to employ coupling with a diazonium compound; the 5-arylazo-group introduced in this way is readily reduced to a &amino- group.45 (iv) 4 5-Diamino-compounds are most easily cyclised to purines by treatment with sodium dithioformate to give 4-amino-5-thioformamido-corn- pounds which lose hydrogen sulphide readily to give the required purine.When in this cyclisation amino- and alkylamino-groups at 4 and 6 compete (a) From 4 5-diaminopyrimidines. 43 Annalen 1904 331 64 ; cf. C. 0. Johns J. Biol. Chem. 1911 9 161. 4 4 G. A. Howard B. Lythgoe and A. R. Todd J. 1945 656. 4 5 B. Lythgoe A. R. Todd and A. Topham J. 1944 315. N 192 QUARTERLY REVIEWS for reaction with the thioformamido-group the superior reactivity of the alkylamino-group leads to the formation of a 9-albyladenine derivative. 46 One outcome of this work was the convenient synthesis of adenine 47 shown below which starts from 4 6-diamino-5-benzeneazopyrimidine (p. 189). It has been used for the preparation of adenine containing isotopic nitr0gen.l’ N H I NH2 I NH2 B. Lythgoe and A. R. Todd15 have reviewed the way in which the experience gained in the above work has been applied to the synthesis of adenine and hypoxanthine nucleosides.The following example of the preparation of the pyranose analogues of adenosine and inosine illustrates the methods used although a number of variations in the general procedure each possessing advantages in particular cases are possible. A H $4 H A particularly valuable feature of this method O H of synthesis is that the producf must contain the sugar residue at N of the p&ne system and not at N,. Since 9-D-mannopyranosidoadenine prepared in this way is degraded by periodate to the same dialdehyde that is obtained from the degradation of adenosine a direct chemical proof i s available that the latter is a 9-glyco- side >N-CH.[CH(OH)]; I01 CH.CH,.OH ‘*.., \ >N.CH.CHO 1°1 OHC.CH.CH,.OH I O - 7 >N-CH-[CH(OH~ - CH-CH,.OH 2 This unambiguous synthesis of 9-glycosides can be extended to compounds of the furanose series but so far no naturally occurring compound has been ** J.Baddiley B. Lythgoe D. McNeil and A. R. Todd J . 1943 383. 47 J. Baddiley B. Lythgoe and A. R. Todd J. 1943 386. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 193 obtained by its application. However by making use of the method of preparation of acetohalogenoribofuranoses employed in the synthesis of cytidine mentioned earlier and by applying the method of E. Fischer and B. Helferich 48 for the introduction of a sugar residue into a preformed purine skeleton the naturally occurring purine nucleosides adenosine and guanosine have been synthe~ised.~~ The method used is shown below.T O 1 c L c H fi H(oA~~J 2 c H c H ,-OAC ?9 N H (a)NH3-MeOH (b) HZ-Pd N H OH H2- Pd N H N H OH The glycosides prepared in this way e.g. adenine glucopyranoside are undoubtedly 9-glycosides since the same adenine glucopyranoside is obtained by the unambiguous synthetic route mentioned above. This second method of synthesis has the advantage that it allows the correlation of the glycosidic configurations in guanosine and adenosine ; these and the natural pyrimidine ribonucleosides are now known to belong to the @-series. The purine syntheses discussed above start with the pyrimidine ring preformed and proceed to build up the glyoxaline nucleus. This order of events can be reversed as first shown by J. Sarasin and E. Wegmann in 1924 ; 50 they used a synthesis of Type I11 for building up the pyrimidine nucleus on to a preformed glyoxaline nucleus so obtaining 7-methylxanthine (xanthine is 2 6-dihydroxypurine).Xan- thine itself has been obtained similarly by a synthesis of the pyrimidine nucleus belonging to Type 11,51 and very recently R. H. Baxter and F. S. (b) Syntheses from glyoxaline derivatives. 48 Be?. 1914 47 210. 49 J. Davoll B. Lythgoe and A. R. Todd J. 1948 967; idem in the press. 50 Helv. Chim. Acta 1924 7 713. 5 1 W. E. Allsebrook J. M. Gulland and L. F. Story J . 1942 232. N* 194 QUARTERLY REVIEWS Spring s2 have employed the Hofmann rearrangement method (p. 190) to prepare derivatives of xanthine and alloxazine ; derivatives of the pteridine system had been obtained in the same way earlier. Their methods can be illustrated by the following preparation of 9-methylxanthine The fact that this cyclisation leads to 9- and not to 7-substituted xanthines permits its application to the synthesis of 9-glycosidoxanthines ; thus by using the preparation of acetochlororibofuranose mentioned earlier the nucleoside xanthosine identical with material from the deamination of natural guanosine was obtained 53 N H3 - MeOH E tO,C E t 0,C I KOBr HC)fyN>" N' N OH The synthetic methods mentioned in this and the previous section clearly represent a considerable advance and in the near future we may expect to see this work extended to clarify the structures of the naturdly occurring purine deoxyribosides.The Chemical Behaviour of Pyrimidines Since the behavionr .of pyrimidine derivatives has not been reviewed for many years the following pages will be devoted to mentioning briefly the more important transformations to which the different nuclear substi- tuents can be subjected (since these play such an important part in prepara- tive work in the series) and to a discussion of the various theoretical points of interest.Three prominent features of pyrimidine chemistry are (i) In simple derivatives containing alkyl aryl or nitro-groups or halogen atoms but no hydroxy- or amino-groups the nucleus has aromatic character and behaves like that of pyridine. 52 J . 1945 229; 232; 1947 378. 63 G. A. Howard A. C. McLean G. T. Newbold F. S . Spring and A. R. Todd in the press. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 195 (ii) Nuclear substituents vary in their behaviour according to the position which they occupy.At position 5 the properties of a group can be loosely described as similar to those which it normally possesses when attached to an aromatic nucleus ; at 2 4 and 6 marked deviations from the normal behaviour are observed. The contrast is parallel to that between p-substituted pyridines and their a- and y-isomers. (iii) The aromatic behaviour mentioned in (i) diminishes progressively as hydroxy- or amino-groups are introduced into positions 2 4 and 6. This effect is seen in uracil and barbituric acid into which substituents are readily introduced at position 5 even by mild reagents such as diazonium compounds ; simple pyrimidines such as those mentioned in (i) appear to be very resistant to electrophilic substitutions. As the simpler compounds are much less well known than the highly hydroxylated or aminated members a rather distorted impression of pyrimidine chemistry has grown up much as if the behaviour of benzene were known only through the reactions of compounds like phloroglucinol.The groups which give rise to this atypical behaviour (OH SH NH,) have been termed somewhat loosely “tauto- rneric ” substituents ; the structural problems arising when they are present will be discussed later. Each of these three features will now be considered separately in more detail. The Analogy with Pyridine.-The pyrimidine nucleus can be regarded as a resonance hybrid of the structure (XXII-XXV) with four equivalent structures. In the classical structure (XXII) the nitrogen atoms are doubly linked and their electronic effects which are the same as that of the doubly-linked nitrogen in the pyridine nucleus co-operate since the key atoms are situated 1 3 to each other.Pyrimidine would thus be expected to have aromatic characteristics similar to those of pyridine or nitrobenzene,64 with the differences from benzene even more strongly marked. There should be a marked electron-deficiency at positions 2,4 and 6 and a similar though smaller deficiency at 5 which although insulated from the ring- nitrogens will be affected by induction. The general electron-deficiency is borne out by the fact that when quinazoline is oxidised it is the pyrimidine nucleus rather than the benzene nucleus which survives ; the resulting 4 5-dicarboxylic acid gives pyrimidine-5-carboxylic acid on being heated. 64 For a discussion of the analogy between pyridine and nitrobenzene see T.W. J. Taylor and W. Baker “ Sidgwick’s Organic Chemistry of Nitrogen ” Oxford 1937 Chap. XVIII. 196 QTTARTERLY REVIEWS These changes are analogous to those whereby quinoline is degraded to nicotinic acid. It is the stability of the nucleus towards oxidising agents which allows pyrimidinecarboxylic acids to be prepared by oxidising alkylpyrimidines. On the other hand such stability is often accompanied in aromatic nuclei by lability to reducing agents ; the ready reduction of pyridine is well known. Pyrimidine has not been reduced catalytically but 4-methylpyrimidine is reduced by sodium and alcohol with the formation of 1 3-diaminobutane. This is presumably due to the instability under these conditions of the hexahydro-compound first formed which contains a carbon atom (C,) directly united to two nitrogen atoms.Another effect of the electron deficiency at positions 2 4 and 6 is to encourage their attack by nucleophilic reagents. Thus 4-methylpyrimidine can be aminated with sodamide (the Tschitschibabin reaction familiar in the pyridine series) giving Z-amino- and 2 6-diamino-derivatives. 5 5 Nuclear substitution by electrophilic reagents such as nitric acid and halogens on the other hand would be expected to be difficult as it is in pyridine and indeed substitutions of this kind have only been recorded where one or more " tautomeric " substituents are present in the nucleus ; the discussion of such reactions is accordingly deferred till later in this review When they do take place the substituent is invariably introduced at position 5 the position where the electron density is least reduced by the ring-nitrogens.It would be interesting to know at what positions radical substituting agents attack the nucleus but such substitutions have not yet been attempted. One of the most characteristic reactions in heterocyclic compounds similar to pyridine is the conversion of their quaternary compounds into the dihydro-derivatives known as pseudo-bases a reaction which reflects the susceptibility of the positions a or y to the positively charged nitrogen atom to attack by negative ions. The quaternary salts of the pyrimidine series are not well known but a phenomenon analogous to pseudo-base formation exists in the behaviour of 2 6-diethoxypyrimidine with alkyl and glycosyl halides.41 These compounds react with the elimination of ethyl halide and the formation of N-substituted ketoethoxydihydropyimidines a reaction clearly due to the attack at position 2 by the halide ion of the quaternary pyrimidinium compounds (XXVI) B NO B r O ' I + R B ~ .-> EtocJ d o < J + E t B r N\ N\ OEt 0lE t OEt (=w 9 As already mentioned these analogies with pyridine lose some of their 5 5 E.Ochiai and M. Karii J. Pharm. SOC. Japan 1939 59 18. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 197 validity where two or more hydroxyl or amino-groups axe present in the pyrimidine nucleus ; thus in uracil and barbituric acid the nuclear stability towards oxidising agents is completely lost. Vestiges of the behaviour of the parent compound however remain; the 2 4 and 6 positions retain their electrophilic character as shown by the reactivity of halogen atoms located there ; and substitution by electrophilic reagents is still confmed to position 5.Properties of Substituent Groups.-In this section the properties associated with some of the substituent groups will be surveyed with the object of bringing out the contrast between their behaviour at position 5 and that at 2 4 and 6 which was mentioned on p. 195. Since the behaviour at position 5 is of the kind normally found in groups attached to aromatic nuclei no special comment is needed concerning them. Groups at positions 2 4 and 6 show in general an abnormal or reactive behaviour which is due to resonance with or within the nucleus. There is one aspect of this reactive behaviour which has very important practical consequences ; the replace- ability of groups such as C1 OEt and SEt to mention only the more im- portant is used extensively in the preparation of those substituted pyri- midines which cannot be obtained directly by building up the pyrimidine nucleus.Chloro- alkoxy- and alkylthio-compounds therefore occupy a key position in preparative operations for which they are well fitted by virtue of their ready accessibility. Chloro-compounds are obtained by heating the corresponding hydroxy-compounds with phosphoryl chloride ; it has recently been found that where this reaction is difficult it can often be promoted by the addition of dimethylaniline. 56 Alkylthio-derivatives are mainly of value in cases where the alkylthio-group is present at position 2. Such compounds can be obtained in two ways; S-alkylthioureas can be condensed with a second component in a synthesis of Type I ; or thiourea can be used and the resulting 2-thiol compound afterwards subjected to X-alkylation.Both these methods are convenient since Type I syntheses with thiourea or its S-alkyl compounds proceed very readily in comparison with those where the less basic urea is employed. A few examples will make clear the use of the chloro- alkoxy- and alkyl- thio-compounds ; for others the reader is referred to a review by T. B. Johnson and D. Hahn in 1933,57 and to the more recent literature citations. The most important replacements of chlorine atoms are by amino- or substituted amino-groups usually by heating the chloro-compound with alcoholic ammonia or with the appropriate amine,5* and by alkoxy-groups by the action of sodium alkoxides.Cytosine 59 is prepared from uracil by the following route. Uracil + 2 6-dichloropyrimidine + (6-chloro-2- aminop yrimidine) + 2- chloro-6 - aminopyrimidine ++ 6-amino-Z-ethoxypyri- midine -+ 6-amino-2-hydroxypyrimidine. The syntheses of vitamin B 5c3 Ref. 36 ; J. Baddiley and A. Topham J. 1944 678. 57 Chem. Reviews 1933 13 193. 58 F. 33. S. Curd and F. L. Rose J . 1946 343 ; J. P. English et al. J . Amer. Chem. 59 G. E. Hilbert and T. B. Johnson ibicl. 1930 52 1154. SOC. 1946 68 1039. 198 QUARTERLY REVIEWS due to R. R. Williams and his co-workers 6O and to H. Andersag and K. Westphal 61 make use of ammonolyses of 4-chloropyrimidines. Ethoxyl groups are replaced by hydroxyl groups by treatment with dilute acids or by amino-groups by the action of hot alcoholic ammonia ; these reactions are made use of in Hilbert’s method for the synthesis of pyrimidine nucleo- sides (p.190). Ethylthio-groups can similarly be replaced by hydroxyl amino- and substituted amino-groups.62 All these replacements and many others such as the exchange of amino- groups for hydroxyl groups on boiling with dilute acids fall into a common pattern. In their reactive behaviour halogen atoms at positions 2 4 and 6 are analogous to those in l-chloro-2 4-dinitrobenzene or cc- and y-halo- genopyridines. The halogen replacements are due to the electrophilic character of the nuclear carbons at these positions (p. 196) and their attack by the nucleophilic reagents (NH, OEt@ etc.) is facilitated by the low energy of transition states such as (XXVII). A similar explanation applies to the R replacement of OEt and SEt groups in the presence of strong bases such as ammonia.It has been mentioned however that certain replacements such as those of OEt SEt and NH groups by OH require acid as the re- agent ; this presumably acts by permitting a neutral transition state such as (XXVIII). A similar effect has been found in the replacement of halogen atoms by feebly basic arylamines ; the reaction can be conducted more easily in the presence of a ~ i d . ~ 3 That this is not the case with more strongly basic amines is no doubt due to the latter undergoing salt formation. (a) AZEyZpyrimidines. The most striking property of methyl groups at positions 2 4 and 6 is their reaction with benzaldehyde in the presence of zinc chloride to give styrylpyrimidines 64 (e.g. XXIX) ; &methyl deriva- tives do not behave in this way.These reactions recall those of the methyl groups in cc- and y-methylpyridines and no doubt the reactive entity is a zinc chloride complex such as (XXX). In the light of the behaviour of the analogous methyl group in acetophenone it is perhaps not surprising that such methyl groups are catpable of side-chain bromination 65 and 6o J . Amer. Chem. SOC. 1937 59 1052. 6 2 Y . F. Chi and Y. S . Kao J . Amer. Chem. SOC. 1936 58 772; F. H. Case and 63 C. K. Banks J . Amer. Chem. SOC. 1944 66 1127 ; F. H. S . Curd et al. J. 1946 64 S. Gabriel and J. Colman Ber. 1903 36 3383 ; E. Ochiai and M. Yanai J . 6 5 C. C. Price N. J. Leonard and R. L. Whittle J . Org. Chem. 1945 10 327. 61Ber. 1937 70 2035. A. J. Hill ibid. 1929 51 1590 ; F.H. S. Curd and F. L. Rose J . 1946 343. 343 370. Pham. SOC. Japan 1938 58 397. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 199 the formation of pyrimidylfuroxans (XXXI) on treatment with nitric acid. (b) Halogenopyrimidines. Members in which the halogen atom is present in position 5 have the expected inertness paralleling that of @-chloropyridine or bromobenzene. The halogen can be replaced catalytically by hydrogen,66 and exchange for a cyano-group on heating with cuprous cyanide in quinoline has been but exchange for amino- or alkoxy-groups is not gener- ally practicable. The presence of two " tautomeric " groups in the nucleus modifies this behaviour ; thus 5-bromouracil and 5-bromocytosine can with some difficult,y undergo ammonolysis to give 5-amino-derivatives The ready reactivity of chlorine atoms a t the remaining positions has already been discussed.4-Halogenopyrimidines react with phenylmag- nesium bromide or with benzene in the presence of aluminium chloride to give 4-phenyl derivatives,68 and they have been used in Ullmann reactions to give dipyrimidyl compounds. 69 The reductive replacement of halogen atoms at positions 2 4 and 6 is very frequently used for preparing pyrimi- dines with unsubstituted positions. Recently this has been effected by catalytic methods ; 70 examples are given in the nucleoside syntheses mentioned on p. 193. (c) Nitropyrimidines. 5-Nitropyrimidines are the only known repre- sentatives of this class ; in early work they were used extensively for the preparation of 5-amino-compounds. The nitro-group exerts an influence on groups situated o- and p - to it just as in the benzene series.Thus the reactivity of the 6-chlorine atom in 2 6-dichloro-5-nitropyrimidine is so enhanced that it is replaced rapidly by ammonia at room temperature and the amino-group of 5-nitro-2-aminopyrimidine is removed as ammonia on heating with aqueous alkali. (d) 2-Mermpto- and 2-alkylthio-compounds. The importance which these compounds have in preparative work and some of their replacement reactions have been noted earlier. It is also possible in some cases to W. Huber and H. A. Hiilscher Ber. 1938 '71 87 ; J. P. English et al. J. Amer. Chem. SOC. 1946 68 1039. 67 C. C. Price N. J. Leonard and R. H. Reitsema ibid. p. 766. 68 M. Anker and A. H. Cook J . 1941 323; E. Ochiai J. Pharm. SOC. Japan 1940 60 164. 70 R. 0. Roblin J.H. Williams P. S. Winnek and J. P. English J. Amer. Chem. Soc. 1940 62 2002; H. J. Backer and A. B. Grevenstuk Rec. Trav. chim. 1942 61 291. 6 9 M. Yanai and T. Naito ibid. 1941 61 99. 200 QUARTERLY REVIEWS replace alkylthio-groups at position 2 by the action of chlorine in aqueous solution,71 but this method has little preparative value. A useful reaction of the thiol group at position 2 in both pyrimidines and purines is the replacement by hydrogen which it undergoes when treated with hydrogen peroxide ; 7 2 it is used in Traube's synthesis of adenine and hypoxanthine and is interesting enough to require comment although it has never been examined very closely It undoubtedly occurs by oxidation of the thiol t o a sulphonic acid group which is then split off by hydrolysis as sulphuric acid ; similar behaviour is observed in 2-mercaptoglyoxalines.The structural requirement is clearly adjacence of the thiol group to a doubly linked heterocyclic nitrogen but the question arises as to why the sulphonic acids which are first formed should undergo such ready hydrolysis ; the electrophilic character which is the outstanding feature of such nuclear positions seems inadequate to provide an explanation. The answer is probably to be found in the observation of D. L1. Hammick and his colla- borators 7 3 that the ready decarboxylation of pyridine- and quinoline-2- carboxylic acids is due to the intervention of a rather stable pyridyl or quinolyl anion in which the negative charge is carried at position 2 of these heterocyclic systems; they point out that the stability of such anions is probably due to their cyanide-ion-like structure.Positions 2 4 and 6 in the pyrimidine nucleus are fully analogous to the 2-position in pyridine and the ready decarboxylation of a carboxyl group at position 4 compared with that at position 5 has already been noted. It seems probable that the ready loss of the sulphonic acid group from position 2 in the pyrimidine nucleus is due to the operation of similar factors. A desulphurisation method which is more convenient than that just mentioned applies to 2-alkylthio-compounds which can be converted into derivatives with an unsubstituted position 2 by treatment with Raney nickel containing adsorbed hydrogen. This method which is an application of a procedure first used by J. B~ugeault,'~ has recently been used t o prepare adenine- 9-glycosides from their 2-methylthio-derivatives .Hydroxy- and Amino-compounds.-(a) Hydroxypyrimidines. Neither 5-hydroxypyrimidine nor any well-authenticated alkyl or aryl homologue is known. These compounds should like P-hydroxypyridine have well- defined phenolic properties ; the only evidence supporting this view is the colour given by isouracil 75 (2 5-dihydroxypyrimidine) with ferric chloride. Hydroxypyrimidines bearing up to three hydroxyl groups at positions 2 4 and 6 are well known; they include uracil thymine and barbituric acid. Their properties will be considered later ; here a few features of their behaviour may be noted which throw doubt on the view that their struc- tures are those of simple hydroxyl derivatives. They show no phenolic 7l J. M.Sprague and T. B. Johnson J. Amer. Chem. SOC. 1935,57,2252 ; 1936 58 423; 1938 60 1622. 73 Nature 1948 162 7 3 ; J . 1939 809; K. Mislow J . Amer. Chem. Xoc. 1947 69 2559. 7 * J. Bougeault E. Cattelain and P. Chabrier Bull. SOC. chim. 1939 6 34 ; 1940 7 781 ; Cornpt. rend. 1939 208 657. 75 J. Tafel and P. A. Houseman Ber. 1907 40 3743. p 2 H. L. Wheeler J. Bid. Chem. 1907 3 285. LYTHGOE PYIZIMIDME AND PURINE CHEMISTRY 201 behaviour. The “hydroxyl groups” are replaced by chlorine atoms on heating with phosphoryl chloride. They are difficult to acylate and their acyl derivatives are readily hydrolysed. The action of alkylating agents on them varies with the compound and the reagent used. Thus uracil is converted into its N1 N3-dimethyl derivative either by methyl sulphate and alkali or by ethereal diazomethane 76 ; 6-hydroxy-2-phenacylthio-4-methyl- pyrimidine gives the corresponding O-ether on treatment with sodium ethoxide and ethyl bromide.The O-ethers have the interesting property of rearranging on heating ; 2 6-dimethoxypyrimidine gives N1 N3-dimethyl-uracil. 5-Aminopyrimidine and its 4-methyl derivative have been prepared? but not examined closely. We are thus not in a posi- tion to say what properties are associated with an isolated 5-amino-group. Such knowledge of 5-amino-derivatives as exists comes from the well- known 4 5-diaminopyrimidinesY which are important as intermediates from which purine and pteridine derivatives can be synthesised. Apart from the cyclisation reactions involved in these syntheses only the amino-group a t position 5 in these compounds shows normal reactivity.It alone can be acylated (e.g. with acetic anhydride or sodium dithioformate) or made to undergo Wohler syntheses (with potassium cyanate or alkyl isothiocyan- ates). With aldose sugars too N-glycoside formation takes place only a t the amino-group a t position 5. With nitrous acid it is not clear if this amino-group can give true diazonium compounds since by secondary reaction with the 4-amino-group cyclisation to pyrimidotriazole derivatives takes pIa~e.7~ I n pyrimidines bearing amino-groups a t 2 4 or 6 anomalous properties arise which justify doubts as to their structures. They are difficult to acyl- ate ; reaction with dithioformates which the 5-amino-group undergoes readily fails completely. The reaction with aldose sugars which has been examined in connection with the synthesis of purine nucleosides is difficult? and requires special structural features for its success.78 With nitrous acid rather sluggish deamination occurs apparently without the intervention of diazonium compounds.It has already been mentioned that deamination can also be effected by the action of hot acids. The anomalous proper- ties of compounds containing “ hydroxyl ” or “ amino- ” groups a t 2 4 or 6 suggests that they may in reality be derivatives of imino- or keto- dihydropyrimidines (e.g. ? XXXIII XXXV). Similar problems of course (b) Aminopyrimidines. (c) Xtructure of amino- and hydroxy-compounds. 7~ F.H. Case and A. J. Hill J . Amer. Chem. SOC. 1930 52 1536. 77 R. 0. Roblin et al. ibid. 1945 67 290. 78 J. Baddiley B. Lythgoe and A.R. Todd J . 1943 571. 202 QUAITERLY REVIEWS are met with in other heterocyclic systems the so-called '' tautomerism " of isatin and a- and y-hydroxypyridines being long-known examples. In the following discussion it will be simplest to consider the hydroxy-deriva- tives of pyrimidines for illustration purposes but much of what will be said applies also to the amino-compounds. 2-Hydroxypyrimidine can be written in two classical forms the lactim (XXXVI) and lactam (XLI). These represent quite distinct compounds ; but each of them possesses various resonance possibilities and is to be regarded as a hybrid (XXXVI-XL) and (XLI-XLV). When two such (xLr.) (XLIC.) (BCLX) (XLIP.) (XLP) hydroxyl groups are present in the nucleus as in uracil four distinct forms are possible each a hybrid derived from one of the classical structures (XLVI-XLIX).It is important to decide for any particular derivative (XLP[.)* (XLrn) (XLPIII.) (XLIX.) whether it shall be represented by the lactim or lactam structure or whether one of these is present in the solid state and an equilibrium mixture in solution with proportions controlled by the environmental conditions. No evidence has been obtained of the latter possibility which would imply that these compounds were truly tautomeric. The lactim-lactam question has been aired most extensively for uracil and barbituric acid. One approach to a problem of this kind is to use ultra- violet spectroscopy a technique which has already shown its value in other branches of pyrimidine and purine chemistry. It is being increasingly used for the quantitative determination of small quantities of pyrimidine and LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 203 purine derivative^.'^ I n the hands of T.Caspersson 80 it has proved of the utmost value for the detection and investigation of the behaviour of the nucleic acids present in chromosomes. H. M. Kalckar *l has combined the use of specific enzymes for transformations such as adenine --+ hypoxanthine + xanthine -+ uric acid with ultra-violet measurements to give a '' dif- ferential ultra-violet spectroscopy " which should be of great value in deter- mining these derivatives in the presence of each other. On the constitutional side the natural purine nucleosides were first diagnosed as 9-glycosides by ultra-violet spectroscopy,82 and the spectrum of one of the fission products of vitamin B gave the clue to its identity as a 4-aminopyrimidine deri~ative.~3 In an attempt to clarify the structure of uracil by spectroscopic methods J.E. Austins4 has investigated the ultra-violet absorption of a number of its derivatives. She finds that uracil does not behave as the form (XLVI) since the spectrum is different from those of 2 6-dichloro- and 2 6-diethoxy- pyrimidine which must possess this type of structure. Uracil differs also from 6-methoxy-3-methy1uraci1 which is a derivative of the form (XLVIII) and also from 3-methyluracil and 1 3-dimethyluracil; the last two com- pounds have the same spectrum and the second of them is certainly the first of them in all probability a derivative of the form (XLIX). Uracil has the same spectrum as l-methyluracil in which the atomic arrangement at positions 1 and 6 is stabilised in the lactam form.Austin concludes and to all appearances justifiably that in solution in alcohol in which the measure- ments were made uracil is to be represented by the form (XLVII). F. Arndt,85 on the other hand considers that the lactam structure is preferred by hydroxypyrimidines except where such a structure would increase the energy of the system by removing the resonanace energy derived from the aromatic character of the ring ; he formulates uracil as (L) but barbituric acid is given structure (LI) in which aromatic character is still possible by virtue of resonance rather than structure (LII) where it is forbidden. Arndt attaches importance to the fact that with ethereal diazomethane uracil gives 1 3-dimethyluracil ; if as he maintains this E.Vischer and E. Chargaff J . Biol. Chern. 1947 168 781; J. F. Tinker and J . B. Brown ibid. 1948 173 585; R. D. Hotchkiss ibid. 1948 175 315. 81 J . Biol. Chern. 1947 167 429 445 461 477. 80 Ref. 8 p. 127. 8% J. M. Gulland E. R. Holiday and T. F. Macrae J . 1934 1639 ; J. M. Gulland and E. R. Holiday J. 1936 765 ; J. M. Gulland and L. F. Story J. 1938 259 692. a 3 R. R. Williams J . Amer. Chern. Xoc. 1935 57 229; R. R. Williams E. R. Buchman and A. E. Ruehle ibid. p. 1093. 84 Ibid. 1934 56 2141. 85 Rev. Fm. Sci. Istanbul 1944 A 9 19. 204 QUARTERLY REVIEWS reagent acts by replacing the most acidic hydrogens of a molecule by methyl groups without alteration of structure then solid uracil would be a fully lactam compound. While perhaps no final verdict should be given at present the difficulties inherent in chemical methods of deciding this sort of structural problem will be remembered and it seems likely that a complete clarification may ultimately come from the application of physical techniques.Infra- red spectroscopy appears particularly promising for this purpose especially if technical difficulties such as those involved in the use of hydroxylic media can be overcome. In the case of barbituric acid good chemical reasons for preferring structure (LI) exist ; they are provided by the contrast between barbituric acid and its 5-monosubstitution products on the one hand and the 6 5-disub- stitution products on the other ; the latter must of course be derived from structure (LII). Thus barbituric acid and its ðyl derivative are strong acids (Ku 1051 and 383 x loA7) ; the 5 5-diethyl derivative is very much weaker (Ku 0.37 x The feeble acidity of ethyl malonate suggests that the difference is not likely to be due to the possibility of direct C-H dissociation in the f i s t two compounds but rather that an enol system as in (LI) contributes to their acidity.A similar point occurs in considering the bromo-derivatives. One of the two bromine atoms in 5 5-dibromo- barbituric acid which is the product of direct bromination of the parent acid is readily displaced by ammonia or sodium acetate giving 5-bromo- barbituric acid; the bromine atom in the latter is hard to replace and replacement of both halogens of the dihalogeno-compounds occurs only in isolated cases for example in condensation with phenylhydrazine to give alloxan phenylhydrazone and in the formation of riboflavin from the 5 5-dichloro-compo~nd.~~ The 5-halogeno-5-alkylbarbituric acids like the 5 5-dibromo-compounds contain a reactive halogen which is replaceable by ammonia to give 5-alkyl- uramils.Clearly there is a difference in type between the 5 5-dihalogeno- compounds and 5-bromo-5-alkyl compounds which must both be derived from structure (LII) and 5-bromobarbituric acid ; the last and the parent acid are presumably structures of the type (LI). (d) Injuence of structure on properties of amino- and hydroxy-compounds. Many of the properties of hydroxy- and amino-pyrimidines are to be regarded as consequences of their hybrid structures irrespective of whether these are of the lactim or of the lactam type.Taking first the physical properties it is known that the polyhydroxy-compounds behave quite differently from pyrimidine and its simple alkyl and halogeno-derivatives ; they are soluble in water rather than in organic solvents and have high melting points (above 300"). These properties are due to hydrogen-bonding favoured by the contribution of Zwitterion structures such as (XXXVIII-XL ; XLII- XLV). The amino-compounds are usually more soluble in alcohol melt lower and are often volatile enough to be purified by sublimation ; here the 86 J. C. Wood J . 1906 89 1831. 87 M. Tishler and J. W. Wellmen U.S.P. 2,261,608. LYTHGOE PYRXMIDINE AND PURINE CHEMISTRY 205 hydrogen- bonding is weaker because nitrogen is less electronegative than oxygen. Removal of a proton from either lactim or lactam forms gives an anion with effective resonance possibilities ; the acidity of the hydroxy-compounds depends on this.Introduction of an amino-group on the other hand increases the basicity of the molecule ; thus 2-hydroxypyrimidine is amphoteric ; uracil has no basic properties ; cytosine has well-marked basic properties. The separation methods which are becoming of major im- portance in nucleic acid chemistry make use of these principles of solubility and acid-base behaviour. Cytosine and uridine are readily separated using ion-exchange resins ; 88 filter-paper chromatography and counter-current extraction methods using buffers permit the separation and identification of purine and pyrimidine derivative^.^^ Hydroxyl and amino-groups at positions 2 4 and 6 can influence pro- foundly the chemical properties of compounds containing them.A n important example of this is the behaviour of pyrimidines towards electro- philic substituting agents ; it has already been noted that in the absence of at least one " tautomeric " group halogenation of pyrimidine derivatives has not been effected. It becomes possible when one such group is present as in 2-amino~yrimidine,~~ and is easy in uracil barbituric acid and their amino-analogues. The way in which substitution is facilitated by the pro- gressive introduction of hydroxyl or arnino-groups will be clear from the structures (XL) and (XLV). Nitration of the nucleus appears to be rather more difficult than halo- genation and only when two or more " tautomeric '' groups are present has it been used for preparative purposes.Given these structural features it is possible to effect nuclear substitution by more weakly electrophilic reagents ; nitrosation and coupling with diazonium compounds take place at position 5 and some applications of these useful reactions have already been mentioned. There is little information as to the mechanism whereby these two substi- tutions occur in pyrimidine compounds but the structural requirements necessary for their success are known to some extent.45 With sodium nitrite and acetic acid the introduction of a nitroso-group requires " tauto- meric " groups at both 4- and 6-positions. It is facilitated by a group such as SMe at position 2 which can act as an electron source in the conventional manner ; where less activating groups (H or Me) are present a t position 2 in a 4 6-diaminopyrimidine the presence of mineral acid is required to promote the nitrosation.Pyrimidine derivatives will couple with reactive diazonium compounds derived from chloro- or nitro-anilines when '' tauto- meric " groups are present at either the 2 and 4 or at the 4 and 6 positions ; when three such groups are present as in barbituric acid coupling occurs even with such a mild reagent as benzenediazonium chloride. The coupling reac- tion presents some interesting and as yet unexplained features. Thymine reacts with diazosulphanilic acid although no hydrogen is available for 88 D. T. Elmore Nature 1948 161 931 ; R. J. C. Harris and J. F. Thomas ibid. J. P. English et al. J . Amer. Chem. Xoc. 1946 68 453. 206 QUARTERLY REVIEWS replacement at position 5.The observation 90 that both uracil and l-methyl- uracil react under these conditions whilst 3-methyluracil does not probably has an important bearing on the lactam-lactim structure of uracil ; it may be compared with Austin's data on the ultra-violet absorption of these compounds. Among the other substitution reactions of hydroxypyrimidines hydroxy- methyl and chloromethyl groups can be introduced into the uracil molecule ; as in the benzene series the 5-chloromethyl compounds so obtained contain a replaceable halogen atom. The many substitution reactions which bar- bituric acid derivatives can undergo are probably not all of the same type as those described for the less heavily hydroxylated compounds. Among them are the introduction of an hydroxyl group at position 5 by use of hydrogen peroxide the alkylation with bases and alkyl halides to give 5-alkyl- and 5 5-dialkyl-barbituric acids and the formation of 5-arylidene derivatives when 1 3-dimethylbarbituric acid reacts with aromatic alde- hydes.These reactions are more reminiscent of the behaviour of malonic acid derivatives than of true pyrimidine derivatives ; much of the aromatic character of the nucleus is lost in barbituric acid. The loss of aromatic character resulting from the presence of hydroxyl groups is shown by the resonance formulations given on p. 202 ; it is already apparent when two such groups are present in the nucleus and is then manifested by loss of stability to oxidising agents and by the addition reac- tions which can take place at the 4 5-double bond. Hypohalous and nitric acids both undergo addition to this link in thymine ; oxidising agents prob- ably act by adding two hydroxyl groups to it The addition compounds so formed are unstable to alkali; this affords a method of degrading the nucleus.Addition reactions often intervene to complicate the substitution of uracil and its relatives ; uracil and bromine water give 5 5-dibromo-4- hydroxydihydrouracil. In monoamino-compounds the resonance energy of the nucleus is sufficient to forbid the addition reactions and with aqueous halogens only substitution occurs. Rr7=NH Me- <JJ-c L M e - ( p k 3 <J=NH H N O N / 0 N / NH NH N H2 (L m.) ( L W (L y.) ( L W One further effect of the presence of " tautomeric " groups in the pyri- midine nucleus calls for comment namely their capacity for influencing decisively the behaviour of another group situated m- to them.The replace- ment of the halogen atoms by amino-groups in a di- or tri-halogenopyrimi- dine can be accomplished step by step a fact of some importance in prepara- tive work. Clearly the reactivity of the second halogen atom is depressed by the introduction of an amino-group m- to it. Taking 4 6-dichloro-2- 90 T. B. Johnson and S. H. Clapp J . Biol. Chem. 1908 5 49. LYTHGOE PYRIMIDINE AND PURINE CHEMISTRY 207 methylpyrimidine as an illustration this effect can be explained by reference to formula (LIII) for the 4-chloro-6-amino-2-methylpyrimidine obtained by mild treatment with ammonia. The tendency of the latter to behave as the iminodihydro-form reduces the chances of forming the transition state (LIV) necessary for the replacement of the second halogen atom.Similarly it has been suggested ' 8 that the failure of 4-aminopyrimidine derivatives to yield 4-glycosidamino-compounds on treatment with aldose sugars is due to their tendency to react in the iminodihydro-form (LV). I n 4 6-diaminopyrimidines bearing a t position 2 a " non-tautomeric " group such as H Me or SMe it is clear that both of the amino-groups can- not exist in the imino-form so that one of them should show the characteris- tics of a true amino-group ; such compounds e.g. (LVI ; R = H Me SMe) do in fact show the expected capacity for reaction. It is to be expected that in the future many other reactions of pyrimidine derivatives will find an interpretation along similar lines.
ISSN:0009-2681
DOI:10.1039/QR9490300181
出版商:RSC
年代:1949
数据来源: RSC
|
|