摘要:

QUARTERLY REVIEWS THE USE OF THE TERMS “ACID” AND “BASE” By R. P. BELL M.A. F.R.S. (FELLOW OF BALLIOL COLLEGE OXFORD AND BEDFORD LECTURER IN PHYSICALL CHEMISTRY) Zdrdwtim-The pursuit of exact verbal definitions of qualitative concepts is rarely of much scientific value provided that the requirements of convenience and consistency are satisfied. However the development and clarification of definitions of acids and bases has been closely connected with experimental and theoretical advances in this field and several new defdtions have been directly fruitful in stimulating fresh work. Since the present usage of these terms is by no means uniform it may be useful to follow the development of the subject. Like most definitions the terms acid and base originated in empirical observatiom of physical and chemical properties rather than in any theoretical interpretation of these properties.They were first regarded as the constituents which react together to form salts usually with the liberation of water. Somewhat later Boyle and others (end of seventeenth century) laid more stress on specific properties of acids e.g. their power of precipitating sulphur from its solutions in alkalis and of changing blue plant-dyes to red. This kind of characterisation was made more definite in the eighteenth century ; for example William Lewis (1746) added sour taste and effervescence with chalk as typical properties of acids. Bases (or alkalis) were characterised chiefly by their negative properties of de- stroying or reversing the effects caused by acids. The fist theory of acidic behaviour which is comprehensible in modern term was that of Lavoisier (end of eighteenth century) who regarded oxygen as the “ acidifying principle ” which converted elements like carbon nitrogen sulphur etc.into carbonic nitric or sulphuric acid. This view led to the assumption that all acids were formed by the combination of a “ radical ” with oxygen. Thus a variety of organic acids were supposed to contain hypothetical radicals combined with oxygen and in particular it was assumed that hydrochloric acid (and hence chlorine) contained oxygen. Davy (1810-15) pointed out that it was wrong to assume that chlorine was a compound of oxygen until its compound nature had been demonstrated and the discovery of hydrobromic hydriodic and hydro- cyanio acids shortly afterwards cast still further doubt upon the oxygen theory of acids.In spite of this and other evidence the latter theory was strongly supported by some chemists notably by Berzelius and Gay-Lussac up to about 1840.l 1 For a detailed account of the early history of this subject see P. Walden “ Salts Acids and Bases ” (New York 1929). 113 H 114 QUARTERLY REVIEWS Davy at first expressed the opinion that “ Acidity does not depend upon any particular elementary substance but upon peculiar arrangement of various substances ” a view which accords well with some current definitions (u. injru). However it soon became clear that all the substances commonly accepted as acids did contain hydrogen and Davy soon recognised hydrogen as the essential element in an acid. Liebig showed that this idea was in harmony with the behaviour of organic acids and in 1838 he defined acids as “ compounds containing hydrogen in which the.hydrogen can be replaced by metals ” a definition which was generally accepted without modification until the advent of Arrhenius’s dissociation theory.It should be noted that this definition denies the name acid to the acidic oxides them- selves although their compounds with water are often ill-defined or unknown (e.g. carbonic acid silicic acid etc.). Bases were still re- gardedas substances which reacted with acids to form salts and there was no theory m to their constitution corresponding to the hydrogen theory of acids. Acids and Bases in tk Clccssical Electrolytic Dissociation Theory.-The replaceable hydrogen definition of acids give no indication of why combined hydrogen could only seldom be replaced by metals and the only criterion which it gave of the relative strengths of acids and bases was the often misleading one of the displacement of one acid or base by another.It was shown at an early date that the catalytic effects of acids in certain reactions showed a general correlation with their generally accepted order of strength but no really quantitative comparison was possible without a knowledge of the laws governing the dissociation of electrolytes. The work of Ostwald and Arrhenius on electrolytic dissociation (1880-90) showed clearly that only those hydrogen atoms which produce hydrogen ions in aqueous solutions can give rise to acid properties and the application of the law of mass action to the dissociation equilibrium givss the disso- ciation constant as a rational measure of the strength of the acid.In fact one of the chief supports of the Arrhenius theory was the close correspon- dence between the electrical conductivities of acid solutions their catalytic effects in various reactions and the (less accurate) estimates of their strengths obtained from measurements of the distribution of a base between two acids. Similarly basic properties were associated with the production of hydroxyl ions in solution and the dissociation of a number of weak bases was shown to conform to the Ostwald dilution law though here the amount of quantitative information and correlation with other data was originally much less than for acids. The definition of acids and baaes as substances giving rise to hydrogen and hydroxyl ions respectively in aqueous solution followed naturally from these considerations and was generally accepted for the next thirty or forty years.It explained their characteristic proper- ties in solution in terms of the ions produced and their neutralisation to give salts by the reaction H+ + OH- + H,O. A great deal of quantitative work was done on the dissociation constants of acids and bases and their application to other types of ionic equilibria such as hydrolysis buffer solutions and behaviour with indicators. The anomalies of strong electro- B E U THE USE OF THE TERMS “ACID” AND “BASE” 115 lytss had relatively little effect on equilibria involving weak acids and bases and a satisfactory account was rendered of a large mass of data. The success of these quantitative developments helped to mask some logical weaknesses in the qualitative definitions of acids and bases.For example it was not clear whether a pure non-conducting substance like anhydrous hydrogen chloride should be called an acid or whether it became one only in contact with water though it was usually considered that the anhydrous compound was an acid in virtue of its latent tendency to split off hydrogen ions. Another type of difficulty arose in solvents other than water where investigation showed that the ions produced by acids and bases were frequently quite different from the hydrogen and hydroxyl ions in aqueous solution. This difficulty appeared in a particularly acute form when it was realised that “ typical ” acid-base reactions such as neutrslisa- tion and titration with indicators could sometimes occur in solvents such as chloroform or benzene where no free ions can be detected by conductivity measurements.There was also a more serious ambiguity in the definition of bases. Most of the substances which would neutralise acids belonged to one of two classes metallic hydroxides and organic amines. Of these only the former could be said to split off hydroxyl ions. The dissociation of the amines in water could be written as NR + H,O + NR,H-OII + NR,H+ + OH- and there was much dispute as to whether NR or NR,H=OH (for the existence of which there is little direct evidence) should be regarded as a base. a distinction was made between “ anhydro-bases ” like NH which neutralise acids by picking up a hydrogen ion and “ aquo-bases ” like KOH which liberate a molecule of water in the process.Dejnitions Reluted to Pmticular Solvents.-If sodium hydroxide is dis- solved in ethyl alcohol hydroxyl ions are produced as in water. On the other hand a solution of still more strongly basic properties is obtained by dissolving sodium ethoxide in alcohol the solution containing the ion OEt- which bears the same relation to EtOH as OH- does to H20. Fur- ther various lines of evidence showed 3 that the proton H+ could not exist in solution being converted completely into solvated species such as H,O+ in water EtOH2+ in alcohol and NH,+ in liquid ammonia. These led to the idea o f “ solvent systems ” of acids and bases in which the acid-base definition is modified according to the solvent being used. This view was originated by E. C. Franklin for the solvent liquid ammonia where the typical acidic and basic anions are respectively NH,+ and NH,-.A typical neutralisation reaction in this solvent is NH,CI + NaNH + NaCl + 2NH No real decision was reached on this point and in some quarters Acid Base Salt Solvent E.g. A. Werner 2. anorg. Chem. 1893,3,267 ; 1897,15 1 ; Ber. 1907,40,4133 ; “ Neuere Anschauungen auf dem Gebiete der anorganischen Chemie ” 2nd edition p. 218 (Braunschweig 1909). For references see R. P. Bell “ Acid-Base Catalysis ” pp. 37-39 (Oxford 1941). J . Amr. Chem. SOC. 1905 27 820 ; 1924 46 2137 ; Amer. Chem. J . 19b.2 47 286 ; “ The Nitrogen System of Compounds ” (New York 1935). 116 QUARTERLY REVIEWS or in terms of ions analogous to H,O+ + OH- + 2H,O There are many other points of analogy for example solutions of ammonium salts in liquid ammonia dissolve metals with the evolution of hydrogen-2NH4+ + Mg + Mg++ + H + 2NH3-and there is a class of " ammono-acids " such as B(NH,), which can be compared with the hydroxy-acids B(OH), etc.In this and similar cases the transfer of a proton is still involved in all typical acid-base reactions. However other workers have gone further and extended the nomenclature acid and base to solvents and solutes con- taining no hydrogen. Thus H. P. Cady and H. M. Elsey 5 have defined an acid as a solute that gives rise to a cation characteristic of the solvent and a base as a solute which gives rise to an anion characteristic of the solvent. This system is exemplified by the work of Janders in liquid sulphur dioxide. Here the solvent is supposed to ionise according to the equation 2S0 -+ SO++ + SO3= ; hence SOC1 is a typical acid and K,SO a typical base.This kind of treatment has been extended to a variety of solvents including COCl, SeOCl, and HCN.7 This kind of definition gives a logical account of behaviour in a particular solvent but has to be modified completely for a change of solvent. It has served a useful purpose in stimulating work in unusual types of solvent but it does not give a clue to several of the general properties of acids and bases (e.g. catalysis action on indicators) and it is inconvenient to use such everyday words as acid and base in a sense which vanes with change of solvent. Hence although this type of definition is still used by a few authors we shall not consider it further here. The Bronsted-Lowry Definition.-This definition of acids and bases was proposed almost simultaneously by J.N. Bronsted 8 and T. M. Lowry 9 in 1923 and is still the one in most general use. It reads An acid i s a species having a tendency to lose a proton and a base i s a species having a tendency to add on a proton. This can be expressed in the scheme A + B + H+ where A and B are termed a conjugate (or corresponding) acid-base pair. The definition places no restriction on the sign or magnitude of the charges on A and By though of course A must always be more positive than B by one unit. It is important to realise that the symbol H+ in this definition represents the bare proton and not the " hydrogen ion " of variable nature which is formed in solution by the addition of a proton to the solvent molecule (H,O+ C,H,*OH,+ NH,f etc.) hence the definition is inde- pendent of the solvent.Further it clearly does not include as acids and J . Chem. Education 1928 1425. * G. Jander and H. Mesech 2. physikal. Chem. 1939 A 183,255 and earlier papers. A. F. 0. Germann J. Amer. Chem. SOC. 1925 47 2275. 2461 ; Science 1925 61 70; a. B. L. Smith Chem. Reviews 1938,23 165 ; G. Jander and G. Scholz 8. physikal. Chem. 1943 192 163. NH*+ + NH,- * 2NH3 Rec. Trav. chim. 1923 42 718. Chem. and In& 1923 42 43. BELL THE USE OF THE TERMS “ ACID ” AND ‘‘ BASE ” 117 bases species derived from a solvent not containing hydrogen (e.g. SOCI in SO,). The Bronsted-Lowry definition of an acid obviously includes tho= neutral molecules known as acids in the classical dissociation theory e.g. HC1 H2S04 CH,CO2H etc.It also includes negative ions like HSO1- CO,H*COO- thus falling in line with the description of salts like NaEISO as acid salts. On the other hand the recognition of cation acids such as NH,+ represented an extension of the older definition. For example the acid reaction of a solution of ammonium chloride is most simply related to the tendency of the ammonium ion to lose a proton (NH44- + NH + H+) the chloride ion playing no part in the process similarly the acidity of solutions of many heavy metal salts can be related to cation acids such as [Fe(H,O)J+++ + [Fe(H,O),OH]++ + H+.IO These explanatiom are simpler and more logical than the older interpretations in term8 of “ hydro- lysis ”. The change in defhition of bases is more radical since no mention is made of the hydroxyl ion or similar ions.Neutral bases will include ammonia and the amines in virtue of reaction schemes like RNHa + H+ + R&*,+. The basic properties of the metallic hydroxides are related to the presence of the hydroxyl ion a typical (but not unique) anion base. A new feature of the definition is the use of the term base for the anions of weak acids in general e.g. the alkaline reaction of a solution of sodium acetate is related to the basic properties of the acetate ion (CH,*COO- + H+ + CH,*COaH). It should be noted that the scheme A + B + H+ used in the mid-base definition is a hypothetical one in so far as the free proton cannot exist in solution in measurable concentrations. On the other hand all typical acid-base reactions can be represented in the form A + Ba s A + B, where A,-B and A,-B are two conjugate acid-baeo pairs.This may be illustrated by some examples (see next page). It will be seen that the “hydrogen ion” H,O+ does not occupy any unique position in the scheme of reactions but behaves in the same way as any other acid. The conception of dissociation as an acid-base reaction gives the clue to the degree of dissociation of acids and bases in different solvents which is determined much more by the acidic or bmic properties of the qolvent than by its dielectric constant. Thus many bases which are weak in water react completely with the strongly acidic solvent anhydrous acetic acid while the “ strong ” acids HCI HBr HClO, etc. which dissociate completely in water give widely Mering electrometric titration and con- ductivity curves in acetic acid indicating incomplete dissociation., 10 A few workers had adopted this point of view much earlier (cf. P. Pfeiffer Ber. 1906 59,1864 ; 1907,40,4040) but without euggeeting any extension of the defhition of acids. Similarly in Franklin’s ammonia system of acids and beees ammonhm salts were regarded as typical acids in liquid ammonia but would not be regarded aa such in other solvents. l1 S e e e.g. J. B. Conant and N. F. Hall J . Amer. Chern. Soc. 1927,49,3047,3062 ; J. B. Conant and G. M. Bramann ibid. 1928 60 2305 ; N. F. Hall and T. H. Werner ibid. p. 2376 ; J. B. Conant and T. H. Werner ibid. 1930,63 4436 ; N. F. Hall and H. H. Voge &id. 1933 66 239. 118 QUARTERLY REVIEWS In solvents unable to accept or donate a proton (e.g. hydrocarbons) there will be no dissociation of acids or bases but typical acid-base reactions can still take place provided that the solution contains more than one acid- base pair e.g.an acid + a basic indicator. The quantitative treatment afforded by the classical dissociation theory remains unchanged when the Bronsted-Lowry definition is used. Thus the ordinary dissociation constant of an acid A in a solvent S is proportional to the equilibrium constant for the reaction A + S + B + SH+. It therefore serves as a method of comparing the strengths of the acid-base pairs A-B and SH+-S. Since the concentration of the solvent is usually omitted from the equilibrium expression this is equivalent to taking the acid strength of the cation acid SH+ as equal to the concentration of solvent molecules in the CH,-CO,H CH,.CO,H HZO NH;’ _____ H,O + NH,+ CH,.CO,H H,O + HzP04- NH,+ H,O + CH,*COO - CH,*COO - OH- NH OH- OH- HPO,- ~~~ Description.Dissociation in water or buffer action in acetic acid + acetate. Dissociation of acetic acid in liquid ammonia or dissocia- tion of ammonia in glacial acetic acid or neutralisation of CH,*C02H by NH, with or without solvent. Hydrolysis of acetate solutions. Hydrolysis of ammonium salts or buffer action in NH + NH,Cl. Hydrolysis of secondary phos- phates. Dissociation of ammonia in water. Dissociation of primary phos- phate or buffer action in mixtures of primary and secondary phosphate. ~ _ _ _ _ pure solvent ; thus the acid strength of H,O+ in water is taken as 55.5 moles/litre. All the usual expressions for hydrolysis buffer action etc. remain unchanged and their meaning is often simplified when they are regarded as examples of the fundamental reaction A + B + A + B,.In particular it is only necessary to specify one constant for each acid-base pair and this is most conveniently taken as the conventional dissociation constant of the acid component. For example the behaviour of ammonia and ammonium salts can be described completely in terms of the acid constant for the ammonium ion [NH,] [H,O+]/NH,+ which is quite analogous to [CH,*COO-] [H30+]/[CH,C0,H]. The classical dissocia- tion constant of ammonia “Ha+] [OH-]/[NH,] is obtained by dividing the ionic product of water by this acid constant. Bronsted himself12 has proposed that the strength of a base should be represented by the reciprocal of the dissociation constant of the corresponding acid. However l2 J.N. Bronsted 2. physikal. Chem. 1934 A 169 361. BELL !CHE USE OF “HE TERMS “ACID” BND “ B U E ” 110 this suggestion has not beeb generally adopted and it t3eex.m more mn- venient to use the dissociation constant itself. This does of corn mean that basic strength is represented by a number which decreaseg with in- creasing basic strength but we already have an example of this in the use of pH for representing the acidity of a solution. The inception of the Bronsted-Lowry definition was closely associated with the discovery of general acid-base Catalys&9.13 Dawson had shown at an early date that the acetone-iodine reaction was catalysed not only by hydrogen ions but also by the undissociated molecules of carboxylic acids. In 1923 J. N. Bronsted and K. J. Pedersen l4 showed that the decompition of nitroamide was catalysed not only by hydroxyl ions and uncharged amine molecules but also by the anions of weak acids.Shortly afterwards it was shown l5 that the mutarotation of glucose is catalysed not only by all the species mentioned above but also by positively charged acids like the ammonium ion. These reactions thus show catalysis by all the types of acids and bases included in the Bronsted-Lowry definition. Many similar cases have been investigated since then and it is also found that catalysis by acids and bases also takes place in solvents such aa benzene in which there is no ionisation or reaction with the solvent. The Special Position of the SoZvent.-While the solvent system definition of acids and bases certainly exaggerated the importance of the solvent in acid-base phenomena the Bronsted-Lowry definition tends to underestimate the special position of the solvent from a practical point of view.Thus although the ions H,O+ and OH- are in principle only particular examples of extended classes of acids and bases they do occupy a particularly im- portant position in the physical chemistry of aqueous solutions. In the same way the ions C,H5*OH,+ and C,H,*O- are particularly important in ethyl alcohol NH,+ and NH,- in ammonia and so on. The solvated hydrogen ions are often referred to as “ hydrogen ions ” in all solvents (though this may cause confusion with free protons) while there is no generic term for the ions OH- C,H,*O- NH,- etc. N. Bjerrum l6 has therefore proposed that the ions derived from the solvent by the addition and subtraction of a proton should be known as the Zyonium and Zyate ions respectively.In particular salts derived from the solvent anion (hydroxides for water ethoxides for ethyl alcohol etc.) would be known as metallic lyates. This suggestion has a good deal to recommend it but it has not been generally adopted and will not be considered further. Pseudo-acids and Pseu&o-bases.-These terms have been used in a confusing variety of ways and there is still no general agreement about their exact meaning. The example usually given of a typical pseudo-acid is nitromethane which has a dissociation constant in water of about l3 For a full account and references see R. P. Bell “ Acid-Base Catalysis ” (Oxford l4 2. physikal. Chem. 1923 108 185. l6 J. N. Brijnsted and E. A. Guggenheim J.AWT. Chem. Soc. 1927 40 2554; l6 Fy8bk Tid%skr. 1931 Nos. 1-2 ; Chem. Reviewe 1935 16 207. 1941) especially Chapter IV. T. M. Lowry and E. A. Smith J. 1927 2539. 120 QUARTERLY REVIEWS and reacts slowly with hydroxyl ions to give an ion whom structure is presumably CH,:N+ . Even very weak acids and brtses normally react together instantaneously (as far as can be ascertahed) and this slow- ness of reaction was originally taken as the characteristic property of a pseudo-acid. A. Hantzsch,l' who discovered this phenomenon supposed that the slow change was the transformation of the pseudo-acid CH,*NO into the true acid CH,:NO*OH which then dissociated rapidly. However it has been shown by K. J. Pedersen that there is no need t o assume the intermediate formation of CH,:NO*OH and the modern view is that the actual loss of a proton from the methyl group to the hydroxyl ion is a slow process.The reason for this slowness is connected with the reorganisation of electronic structure on the loss of a proton and this reorganisation provides the best criterion to use in defining the term pseudo-acid since it is only rarely that a measurably slow reaction occurs. The change of electronic structure is also associated with changes in absorption spectra which have been used widely by Hantzsch as a criterion for pseudo-acids. I n an extreme case like nitromethane the charge on the ion resides entirely on an atom other than that from which the proton has been removed. This is so in a number of other molecules for example acetylacetone CH,*CO*CH,*COCK, which has a dissociation constant of about 10- Q gives an ion which is best represented by the mesomeric structure 0- 6- CH,*C :CH*C*CH the charge being distributed between the two oxygens.In other cases the charge may be only partly displaced from the atom originally bearing the proton for example phenols having a nitro-group in the ortho or para position give coloured ions which may be written as In these ions the greater part of the charge probably remains on the phenolic l7 Ber. 1899 32 676. Kgl. Danske Vid. Selsk. -Math.-fys. Medd. 1932 12 No. 1 ; J . P h y h l Ohem. 1934 38 681. BELL THE USE OF THE TERMS “ACID” AND “BASE” 121 oxygen but there is no certain method of determining the charge distri- bution. It is clearly difficult to decide what degree of charge displacement is newsmy before the term “ pseudo-acid ” shall be used.A. Hantzsch 10 concluded from slight optical changes on ionisation that almost all acids (e.g. halogen hydrides nitric acid sulphuric acid carboxylic acids) are pseudo-acids but his interpretation has been challenged by a number of authors.20 In any case such a wide extension of the term would destroy its usefulness and it is best to reserve it for molecules in which there are strong experimental or theoretical grounds for believing that the charge on the ion is concentrated chiefly on atoms other than that from which the proton has been removed. From the point of view of the Bronsted-Lowry definition the anions derived from pseudo-acids will naturally be described as pseudo-bases. There are also examples of uncharged pseudo-bases in the same sense; for example derivatives of y-pyrone have strong basic properties but the positive charge in the cation produced is probably not situated on the oxygen atom to which the proton adds on e.g.0 CH CH // ‘CH / \ \ / CH CH CH CH CH II II +El+-+ I It \ / c C 0 OH II I The change in electronic structure is reflected in a change of absorption spectrum and similar changen are responsible for the colour change of many plant pigments (anthocyanins and flavones) with pH. The cations of these bases will naturally rank as pseudo-acids. The original discussions of Hantzsch assumed that every pseudo-acid had a “ true acid ” corresponding t o it e.g. CH,:NO*OH for nitromethane the enol form of acetylacetone etc. However these alternative forms cannot often be isolated and it is not now believed that they play any part in the typical reactions of pseudo-acids.Further pseudo-bases would differ from the corresponding “ true bases ” only in electronic arrangement and it i s now generally held that electronic isomerism of this kind has no real existence. The above examples illustrate ths most logical way in which the terms pseudo-acid and pseudo-base can be used in conjunction with the Bronsted- Lowry acid-base definition. However these terms have at different times been used in a variety of ways which are at variance-with these ideas. The most important of these is the use of the term pseudo-base to describe compounds formed by the addition of hydroxyl ions to certain organic 1°Z. Ebktrochem. 1923 29 244; 1924 30 202; Ber. 1925 58 953. ‘O K. Fajens Naturwisa. 1923 11 179 ; H.von Halban 2. Elektrochmn. 1923 29 443; H. Ley and H. Hunecke Ber. 1926 69 510. 122 QUARTERLY REVIEWS cations accompanied by a structural rearrangement. examples are the “ carbinol bases ” of various triphenylmethane dyes. example the ion of crystal violet haa the structure (NMe,-C,H,) &<=>he2 - where the positive charge and quinonoid structure can be associated with any one of the three benzene rings. This ion reacts slowly with hydroxgl ions to give the carbinol base (NMe,*C,H,),C*OH and the reverse change is brought about by the addition of acid. Similar behaviour is met with in simpler compounds sech as the pyrazines and acridines.21 These carbinol compounds would not be classified as bases at all in the Bronsted-Lowry nomenclature since they react with acids by splitting off a hydroxyl ion to give water and not by accepting a proton.It seems undesirable to describe them as pseudo-bases since this term applies more logically to compounds like ypyrone. The names pseudo-acid and pseudo-base have also been applied to molecules which are not themselves acids or bases but which give acids or bases on dissociation or reaction with the solvent. Thus COz has been termed a pseudo-acid (CO + H20 + H,CO,) and NH,*OH (if it exists) a pseudo-base (NH,*OH + NH + H,O). By an extension of this argument a covalent halide hydrolysed by water could also be called a pseudo-acid (e.g. XC1 + 2H,O + XOCl + 2H,O+ + 2C1-). This usage again seems undesirable. The Lewis Dejinition of Acids and Bases.-This definition was first put forward by G. N. Lewis z2 at about the same date as the Bronsted-Lowry definition but it did not attract much attention until the last ten years or so.It has recently come into a good deal of prominence especially in the United States and a recent book on the subject 23 compares its importance in chemistry to that of the theory of relativity in physics. However opinions have been sharply divided on this subject and an attempt will be made in this section to give a fair estimate of the position. Lewis aims at broadening the basis of the acid-base definition from both the experimental and the theoretical standpoint. 24 From the experimental point of view he defines as acids and bases all substances which exhibit “ typical ” acid-base properties (neutralisation replacement effect on indicators and catalysis) irrespective of their chemical nature or exact mode of action.On the theoretical side he relates these properties to the acceptance (by acids) and the donation (by bases) of electron pairs to form covalent bonds irrespective of whether the transfer of protons is involved. The list of compounds classed as bases by Lewis is substantially identical The best-known For *l A. Hantzsch and M. Kalb Ber. 1899 32 3116 ; J. G. h t o n J. Amsr. Chem. 28 “Valency and the Structure of Atams and Molecules” (New York 1923). *a W. F. Luder and S. ZufYanti “The Electronic Theory of Acids and Bases” See particularly G. N. Lewis J. Franklin Inat. 1938 220 293; also W. F.‘ SOC. 1930 52 5254; 1931 53 1448. (New York 1946). Luder Chern. Reviews 1940 27 547. BELL THE USE OF THE TERMS “ ACID ” AND “ BASE ” 123 with the Bronsted-Lowry classification since those species which can accept a proton contain an unshared pair of electrons and will also combine with other electron-acceptors.Thus the base NH combines with BF to give H3N-BF,. On the other hand the list of acids is radically altered Lewis himself writes “ . . . any similar valuable and instructive extension of the idea of acids has been prevented by what I am tempted to call the modern cult of the proton.’’ The typical acids of the Lewis definition are molecules like AlCl, BF, SO, etc. which do not contain a hydrogen atom but which are capable of expanding their shell of valency electrons by receiving one or more electron-pairs. These molecules are not acids in the Bronsted-Lowry sense and most of them are at most “ pseudo-acids ” in any of the earlier definitions.Classical acid-base reactions as usually written do not reveal any electron-deficiencies and in fact the acids of the older definitions (HCl H,SO, CH,*CO,H etc.) can only be included in the Lewis scheme by rather indirect means. Thus reaction between an acid HX and a base B is supposed to be initiated by the formation of a complex XH . . . B in which the hydrogen accepts extra electrons from the base. Although there is evidence of the existence of hydrogen bonding in a few cases this bonding is not now believed to involve the formation of a covalent link. As Lewis himself has said 26 “ Evidently what has become known as the hydrogen bond differs not only in degree but in kind from a true chemical bond ” and for this reason the proton acids are sometimes referred to as “ secondary acids ” by Lewis and his school.They also use the term secondary acid in a different sense to describe some acidic oxides and other substances which as usually written cannot expand their electron shell but which nevertheless’ give rise to acidic properties (e.g. CO, RCOCl). Thus CO can only be logically termed an acid in the Lewis sense by writing it in the form O=C-0. On the experimental side it is certainly true that the Lewis definition correlates a wide range of phenomena in the qualitative sense. For example solutions of BF or SO in inert solvents bring about colour changes in indicators very similar to those produced by HC1 and these changes are reversed by adding bases so that a titration can be carried out. Similarly the same substances catalyse a large number of organic reactions some of which are also catalysed by proton-acids.The definition also includes the solvent systems not involving a proton e.g. SOC1 in SO,. However the wider scope is obtained at the cost of some lack of definiteness. For example in the Bronsted-Lowry nomenclature every acid will react with the ammonia molecule to produce the conjugate acid NH,+ but on the Lewis scheme the original product X-NH will be different in every reaction and may or may not dissociate further. Consequently the similar catalytic effects and colours with indicators represent a similarity in electronic displacements rather than a strict parallelism in the nature of the reactions. The major disadvantage of the Lewis definition compared with those 25 G.N. Lewis T. T. Magel and D. Lipkin J . Amer. Chem. SOC. 1942 64 1774. f - + - - + 124 QUARTERLY REVIEWS depending on proton kransfer is on the quantitative side. In the latter system the knowledge of one constant for each acid-base pair (for a given solvent and temperature) is s&cient to determine the position of equilibrium in any dilute mixture of acids and bases. (The same kind of prediction can be made to a limited extent about reaction velocitie8 since there is com- monly a quantitative relation between catalytic effect and mid-base strength.) In fact the most valuable contributions of both the classical and the Bronsted-Lowry concepts have been in the field of quantitative relationships. This kind of quantitative treatment is quite inapplicable in the extended system envisaged by Lewis.Although relatively little quantitative work has yet been done with Lewis acids not containing protons many facts are already known which show that “the relative strengths of acids and bases depends not only upon the chosen solvent but also upon the particular acid or base used for reference”.26 For example in the classical sense ammonia is a much weaker base than the hydroxyl ion but when referred to the Lewis acid Ag+ the order of strengths is reversed since AgOH is completely dissociated while [Ag(NH,)J+ is a stab3e complex. This absence o€ any simple system of acid-base strengths is a high price to pay for an increased descriptive scope. Genera2 Cmlusions.-The practical point at issue is whether the Bronsted- Lowry definition should be adhered to in scientific writing and teaching or whether the term acid should be used in the wider sen80 suggested by Lewis.The latter definition includes more completely those substances showing the qualitative attributes usually associated with acids. On the other hand the Bronsted-Lowry acids form a group of much greater uniformity obeying quantitative relations confined to this group moreover they can only be included somewhat artificially in the electronic definition of acids proposed by Lewis. It therefore seems desirable to distinguish them in some way and three alternatives are possible 27 (a) To use the term acid in the most general sense but to distinguish “proton-acids” or “ hydrogen-acids” as a special class. This has the disadvantage of altering a time-honoured usage and of introducing a cumbrous name for common substances.(b) To use the term acid in the most general sense and to distinguish by a prefix those acids which cannot give up a proton. Kdthoff (bc. cit.) has suggested “proto-acids” but this seems an unfortunate name for acids which do not contain a proton. It would be more scttisfactory to describe them as “ secondary acids ” or “ pseudo-acids ”. Unfortunately we have seen that the former of these terms is already employed by Lewis in almost the opposite sense while the latter also haa a variety of uses. (c) To restrict the term acid to those species covered by the Bronated- Lowry definition and t o use a different name for the Lewis acids. The term m p t o r or meptor molecule has long been used in this sense e8 and suggests immediately the chemical properties of the molecules concerned.2a G. N. Lewis loc. cit. 27 Cf. I. M. Kolthoff J . Physical Chem. 1944 48 61. ee Cf. N. V. Sidgwick “ The Electronic Theory of Valency ” (Oxford 1929). BELL THE USE OF THE TERMS “ACID” AND “BASE” 125 In the opinion of the writer this proposal represents the most satisfactory alternative since it does not interfere with established usage and it preserves the distinction between the two classes of molecule. It does not automatic- ally imply the qualitative resemblances stressed by Lewis but it is quite natural that proton-donors and electron-acceptors should frequently produce similar effects. The nomenclature of bases offers less difficulty since a proton-acceptor will always contain an unshared electron-pair which it can donate to acceptors other than the proton.However it is probably advisable to use the term “ base ” only in contexts involving the transfer of a proton since it is only here that the correspondence of acid-base pairs and the quantitative aspects of acid-base strength can be applied. In other contexts the term donor forms a natural complement to acceptor. Finally it should be stressed that the questions involved in the use of the terms acid and base are concerned essentially with convenience and consistency and not with any fundamental differences in the interpretation of experimental facts. It is therefore misleading to elevate to a matter of principle controversies about these definitions or to speak of an “ electronic theory ” of acids and bases. The chief importance of revised definitions lies in their stimulating effect on experimental work. Just as the Bronsted- Lowry definition initiated many investigations of acid-base equilibria and kinetics in different solvents so the Lewis definition has led to much valuable work on the reactions of acceptor molecules which will retain its importance even if (as suggested in this review) the Lewis definition is not a convenient one for general use.

ISSN:0009-2681    

DOI:10.1039/QR9470100113

出版商:RSC    

年代:1947

数据来源: RSC

摘要:

THE SEPARATION OF THE LANTHANONS (RARE-EARTH ELEMENTS) By J. K. MARSH D.Sc. F.R.I.C. Terminobgy.-The rare-earth group has at no time been a well-defined body of elements. Some present-day physicists confine the term to the elements Nos. 58-71 (Ce Pr Nd 61 Sm Eu Gd Tb Dy Ho Er 'I'm Yb Lu) for here the 4felectron shell comes into being and fills up while the mat of the electron structure remains unchanged. Some chemists wouId inolude also the elements of Group IIIa viz. scandium yttrium lanthanum and actinium but it is better to include only lanthanum (No. 57). Chemically yttrium certainly falls within the group though its chemical position varies 80 that the term " yttrium group " is unfortunate since yttrium may on occaaion associate with elements outside the " yttrium group ". The term " heavy " earths is to be preferred and for the cerium and terbium groups the terms " light " and " middle " earths but without defining exact boundariea.The appelltttion " rare " is also now known to be inapproprhte. According to V. M. Goldschmidt 1 yttrium and three members of the group are more abundant in the crust of the earth than lead while thulium the rare& member is estimated to equal iodine in abundance and to be three times more plentiful than silver. They are normal constituents of igneow rocks but they show very poor power of segregation. Only where large pegmatite formation has occurred is there any high degree of concentration. Thus they are ordinarily concealed and the term " lanthanon " (Ln) is proposed to denote any element of the group from lanthanum to luteciurn inclusive and to replace such objectionable terms as " lanthanate " or " lanthanide " which have recently had some currency and to bring them into conformity with their new analogues the " actinons ".Scoipe.-Previous accounts of the separation of the lanthanons have been written by L. M. Dennis,2 G. Urbain,3 C. James,* and W. Prandtl,6 who during the present century have each added substantially t o our knowledge of the group and whose accounts are based upon personal work. The author likewise during the pa& eighteen years has added to knowledge affecting the separation of each of the fourteen members of the group with the exception of Cerium.ma-61 The reviews of D. W. Pmrce 7 on bi- and J. 1937 655. a L. M. Dennis and B. Dales J . Amer. Ohm. Soc. 1902 24 401. a J . Chirn.physique 1906 4 31 105 232 321. ' J . Am?. Chem. SOC. 1908 30 979; 1912 34 767; ref. 37. '2. anorg. Chem. 1938 238 321. ao J. 1929 2387 (11) ; 6b J . 1934 1972 (Eu Gd Tb) ; 6c J. 1937 1367 (Yb Lu) ; J . 1942 398 (Sm Eu Gd) ; sf J. 1942 523 (Nd Sm Gd); Q J . 1943 8 (Tm Yb Lu); eh J. 1943 631 (Eu Sm); 66 J. 1946 16 (Pr Tb) ; Y J. 1946 17 (La Pr) ; ek J . 1946 20 (Pr Tb) ; 61 t. 1947 118 (Er Y). 126 J. 1939 664 (Tb Dy Ho) ; chem. Reviews 1935 16 121. MARSH THE LANTHANONS 127 quadri-valency and of T. Moeller and H. E. Kremers 8 on basicity in the group will also be found useful. No attempt has been made to catalogue recent work but it is hoped that no important recent method applicable on a moderate scale as distinct from analytical procedure has been overlooked. Any complete account must take into consideration methods of testing purity.These however must now be left with the passing remark that modern spectrophotometric methods make possible rapid complete analysea of the light and heavy groups respectively in regard to the coloured con- stituents of each.9 HistmicaZ.-In 1794-97 yttria was recognised in gadolinite and in 1804 ceria was recognised in cerite but not until the work of C. G. Mosander (1839-43) were these two earths shown to be complex. Ceria was split by the action of chlorine on the suspended hydroxide whereby a part lanthanum hydroxide dissolved. Lanthana in turn by extraction with 1% nitric acid left a residue of didymia. Mosander also split yttria into three fractions true (colourless) yttria and the weaker bases erbia and terbia. Fractional precipitation by ammonia or oxalic acid was used.Thus there came to be recognised six elements. No true new member was added to the group till 1878 when J. C. G. de Marignac using N. J. Berlin’s nitrate fusion process (1860) separated ytterbia from erbia. Between 1878 and 1880 holmia and thulia (P. T. Cleve) samaria (Lecoq de Boisbaudran) gadolinia (Marignac) and scttndia (L. F. Nilson) were recognised though most of them were not prepared pure till the twentieth century. In 1885 C. A. von Welsbach introduced the use of double ammonium nitrates for fractional crystallisation of the light group and split didgmia into neo- and praseo-didymia. E. Demargay in 1900 introduced the use of the double magnesium nitrate method of fractionation. This marks the next and almost final step forward in the separation of the first half of the lanthanon group.Possible replacement of magnesium by other bivalent metals followed quickly but magnesium is the most generally useful. All the members of the light group (freed from cerium) can be purified by double magnesium nitrate fractionation. Addition of the isomorphous bismuth magnesium nitrate assists the separations a t the more soluble (gadolinium) end of the series. Demarqay prepared pure europia by this method in 1900. In 1898 G. Urbain introduced the use of ethyl sulphates for the fractional crystallisation of the heavy group. His important work on this group waa presented in a completed form in 1906.3 The most notable achievement waa the isolation of dysprosium. Since the ethyl sulphates have the disadvantage of being prone to hydrolysis the introduction of bromates by C.Jamee in 1908 gave a more stable salt and one which is capable of separating each of the heavy mrths from gadolinium to lutecium in a state of purity except that the overwhelming abundance of yttrium which interpolates in the series between holmium and erbium prevents the purification of these two by bromates alone. Since 1908 the most notable developments have been in use of the bivalent * Ibid. 1945 37 97. 0 C. J. Rodden J . Res. Nat. BUT. Stand. 1941 26 657 ; 1942 28 266. 128 QUARTERLY REVIEWS state for the separation of the three elements samarium europium and ytterbium which alone have any significant bivalent stability and the use of the quadrivalent oxides of praseodymium and terbium for the isolation of these two elements.Holmium and erbium are thus now the only lantha- nons together with yttrium which have not at least one neighbour which can be separated in one or two operations. Lanthanum neodymium gadolinium and lutecium have no neighbours not thus separable. Sub-atomic Structure.-The lanthanon series of elements arise from the growth of an inner electron shell 4f while the outer valence shells remain fixed. A tendency appears for this shell to contain 0 7 or 14 electrons the normal tervalent quotas of lanthanum gadolinium and lutecium respec- tively. Multivalence is due to this tendency. Thus quadrivalent cerium arises from the single electron present on the 4f level in tervalent cerium migrating to an outer or valence level. The stability of two electrons in the 4f level is not complete though much greater than one thus arises the formation of praseodymium dioxide.Seven electrons in the 4f shell appear to constitute some sort of sub-group for in many respects gadolinium repre- sents a turning-point in the series. Thus several series of salts have a maxi- mum or minimum solubility a t or near this point and in general the differ- ences in chemistry become less when the electrons exceed seven. The elements of the heavy group are much more difficult to purify than those of the light group. The average frequency of the absorption bands is a t a maximum with gadolinium all the bands of which lie in the ultra-violet and shift on either side of gadolinium (No. 64) towards the red so that the salts of the whole series in number sequence pair off in visual colour-64 j 1 64 The strong attraction to complete the 4b septet is seen in the tendency of europium (No.63) to lose an outer valence electron to this level. Samarium (No. 62) has a lesser tendency of the same sort. Terbium (No. 65) on the other hand tends to lose the eighth electron by migration to the valence level as exemplified by the formation of it higher oxide. Ytterbium with thirteen 4f electrons tends to complete the shell by migration inwards of a valence electron. The building up of electrons at the 4.f level does not entirely compensate for the growing nuclear charges so that there is a progressive fall in ionic radius with increasing atomic number. The fall is particularly large from lanthanum to cerium and large between cerium and praseodymium and is accompanied by chemical differences of more than the usual magnitude.Frequently the early lanthanon salts carry more water of crystallisation than the later members. The ionic radius of yttrium is reached about the position of holmium in the series and gives rise to the extraordinarily close chemical likeness of these two elements. However the writer 6z has shown that when the non-ionised states are compared though there is still a contraction with increasing atomic number yttrium now appears to interpolate between neodymium and samarium. 1ZZinium.-Something like seventy false reports of new members of the lanthanon series have been made but with the advent of Moseley’s demon- stration of the regular decrement of X-ray wave-lengths with rising atomic number it became apparent that a t that time (1913) the complement of the 2 etc.MARSH THE LANTHANONS 129 series was correctly made up except for the long suspected missing member between neodymium and samarium. Confirmation of a place here for a possible element was welcome and three claims to have found it were staked in 1926,lO but they do not appear to be well established. According t o J. Mattauch’s l1 empirical rule two isobars differing by one unit in nuclear charge are never both stable. Now the isotopic constitutions of Nos. 60 and 62 are known to include the following mass numbers l2 Neodymium 142 143 144 145 146 148 150 Samarium 144 147 148 149 150 152 154 Unless therefore No. 61 is to prove to be an exception to the Mattauch rule its mass number must be below 142 or above 150. These possibilities both seem unlikely.If it existed in Nature as a radioactive element it would almost certainly have been detected. S. Takvorian has discussed the evidence for the existence of i1lini~m.l~ Several artificial radioactive isotopes of No. 61 have been made of which 14’61 with a half-life of 3.7 years is the most stable. It should be producible in tangible q~antities.1~~ We may note here l2 that the even numbered lanthanons average more than 6 stable isotopes each and the odd numbered only 1.1. The approxi- mately six-fold greater abundance of the even numbered elements is thus seen to coincide largely with a corresponding abundance of isotopes. All mass numbers from 139 to 176 are represented by stable lanthanon bodies. Sources of the Lanthnons.-Tho Travancore monazite deposits are of unparalleled richness and constitute the only important source in the British Empire of the light lanthanons.The composition of the lanthanon oxide from monazite approximates to CeO, 50% ; La203 16% ; Pr,O, 8% ; Nd,O, 18% ; Sm,O,-Gd,O, 77’’ ; Tb,O,-Lu,O + Yz03 1%. In view of the industrial utilisation of monazite considerable amounts of the middle earths (samarium to holmium) and yttrium may be obtained as by-products but the heaviest earths (erbium to lutccium) are scarce in Travancore sand. Monazite is usually a very poor source of europium and the Travancore deposits are no exception. Indeed lanthanon minerals are generally poor in europium and on account of its strong bivalence it is often relatively abundant in alkaline earth minerals and minerals of secondary origin like pitchblende l4 and scheclite.l5 Heavy earth minerals are scarce by comparison with monazite but deposits in pegmatite formations in North America and Scandinavia are well known.The commonest mineral with a high content is probably gadolinite a basic silicate containing beryllium and iron. The lanthanon lo J. A. Harris L. F. Yntema and B. S. Hopkins J . Amer. Chem. SOC. 1926 48 1585 1694; J. M. Cork C. James and H. C. Fogg Proc. Nut. Acad. Sci. 1926 12 696 ; L. Rolla and L. Fernandes Gazzettu 1926 4 498 ; 2. unorg. Chem. 1926 157 371. l1 Z. Phyaik 1938 91 361. l 3 Phyaikal. Z. 1940 41 1. lSa “ The Plutonium Project ” J . Amer. Chem. SOC. 1946 68 2411 l4 J. K. Marsh Phil. Mug. 1929 7 1005. l 3 Ann. Chim. 1945 20 113. l6 Idem J. 1943 677. I 130 QUARTERLY REVIEWS content and composition both vary widely,16 but gadolinite and the phos- phate xenotime and the tantalo-niobate fergusonite may be expected to yield more than 40% of lanthanon oxide.Ths average atomic weight of the heavy earth group from Scandinavian gadolinite varies between 96.5 and 109:17 about 106-107 has been the author’s experience. Thus yttrium is usually about four times more abundant than the combined heavy earth elements. No minerals rich in lanthanons are known in which the middle earths normally predominate. The Treatment of Monazite.-Formerly the lsnthanons were of little economic value and thorium oxide for gas mantles was the commercial objective in working monazite ; but ceramic metallurgical and pyrotech- nical uaes are now important. The sand a phosphate is always heated with sulphuric acid but thewafter various processes have been applied.The anhydrous. sulphatcs if treated with cold water are readily soluble but hydrate and precipitate on warming. If half the lanthanons are precipitated as oxalates all the thorium comes down simultaneously ; when thorium was the objective the remainder could be rejected but a t the present time thia rejection is uneconomic. The oxalate precipitate is contaminated with phosphate and in presence of sufficient sulphuric acid to hold all the lantha- non phosphate in solution the oxalate precipitation is not quantitative. treats the pasts of sulphates sulphuric acid and phosphoric acid with a little water and renioves a liquor by filtra- tion or centrifuging consis ting approximately of 12% phosphoric acid 55% sulphuric acid and 33% water. This liquor is valuable for preparing fertilisers.The anhydrous sulphat cs are then dissolved in cold water and the solution is neutralised with previously prepared lanthanon oxide. By this means thorium zirconium titanium and the like are precipitated. The filtered solution is nest treated with alkali sulphate and the light-earth doiilde sulphates are thcrebx thrown down while yttrium and the heavy earths done winain in solution. Alternatively nearly a11 the lanthanon s u l p h te may be precipitat ctl siiiiply by boiling. Hot alkali transforms eithcr precipitate to hydrous oxide or hydroxide the alkali su1pli;ite being rccovcrable. The hydroxide is dried in air a t a temperature in the neigh- hourliood of 100”. Thereby owr 07:/ of the cerium is oxidised and remains insoluble when the product i.; treated cautiously with dilute nitric acid.The leached product noif almost free froni tervdent earths is treated with nitric acid (d 1.375) to give a. solution which is next added slowly to about five times its volume of water containing some sulphuric acid. There is thus formed the iwll-knon 11 basic nitrate-sulphate precipitate of pure cerium. The mother liquor i:i used again for leaching niore dried hydroxides. The dilute lanthanon solution from the leaching of the hydroxides is treated with sulphuric acid and sodium sulphate. The precipitate of alniost A modern monazite process 1 6 I. and \V. Noddack “ Dns Rhenium ” Verlag v. Voss Leipzig 1933. 1 7 Dana ‘I System of Mineralogy ” 6th edition I<. Paul Trench Trubner & Co. l* A. R. Powell and Johnson Matthey & Co.Ltd. B.P. 510,198 1938 ; J. Rlumen- Ltd. London 1892. field U.S.P. 2,387,993 1943. MARSH THE LANTHANONS 131 cerium-free lanthanons is recovered and the liquor is distilled for recovery of nitric acid. The recovery and utilisntion of all products is thus complete. Further separation of the light earths will be dealt with later. This process is applicable to other minerals which are attacked by sulphuric acid such as xenotime and some tantalo-niobates and silicates. The Separation of Cerium from other Lanthanons.-Use is made of the fact that cerium alone of the lanthanons gives a quadrivalent ion in solution. The quadrivalent solutions however are very weakly basic and readily undergo hydrolysis. It is usually best to eliminate cerium as a first step from all earths which are to be purified since cerous salts undergo slow oxidation and unless con- centrating at the tail cerium will spread all down a fractionation series e.g.a double magnesium nitrate series. In the monazite earths ceria will amount to nearly 50% of the total. The basic nitrate-sulphate process as used above is very satisfactory. The basic nitrate-bromate process in which a 20y0 neutral nitrate solution is boiled with potassium bromate and near- neutrality maintained by addition of marble is a convenient laboratory process.19 Mosander’s original chlorine or bromine treatment of the hydroxides whereby Ce( OH) alone remains insoluble is now obsolete. Formation of ceric ammonium nitrate Ce(N0,),,2NH4N0,,4H,0 and its crystallisation from the mixed nitrate solution has been used industrially but does not give a sharp separation.The basic broinate process is said to leave a cerium-free solution if carried to a point a t which a little tervalent earth is precipitated. The process should be interrupted in order to remove pure cerium before the last contaminated precipitate forms. NO process will give a perfectly clean separation of cerium and the other lanthanons. A cerous-ceric equilibrium always exists in acid solutions and only if the solution is sufficiently alkaline to precipitate some tervalent earth can all the cerium be oxidised. For removing completely small amounts of cerium from other earths addition of KMnO + 4Na,CO to bhe boiling faintly acid and finally neutral nitrate solution until the perinanganate colour persists is excellent. Ammonia and hydrogen peroxide give a yellowish precipitate so long as cerium is present but bluish white in its absence.Anodic oxidation of an acid phosphatic solution results in the precipitation of ceric phosphate from a solution too acid to permit tervalent phosphate t o pre- cipitate such as is obtained after “ breaking ” monazite with sulphuric acid.20 recom- mended starting by fractionation of the double magnesium nitrates 2Ln(N0,),,3Mg(N0,),24H20 but the strong basicity of lanthanum allows of its ready separation by preferential precipitation of the weaker bases. The ratio of the solubility products for yttrium praseodymium and lanthanum hydroxides have been determined as being 1 80 1300 and Separation of the Light Earths (freed from Cerium).-C. Janies l9 C. James et al. J . Amer. Chem. SOC.1911,33 1326 ; 1912 34 757 ; 1916,38,41. ‘O J. W. Neckers and H. C. Kremers ibid. 1928 50 955 ; 1. A. Atanasiu and M. 21 G. Endres Z. unorg. Chem. 1932 205 321. Babor Bull. Acad. Sci. Rournuine 1939 20 27 32. 132 QUARTERLY REVIEWS 1 333 1235.22 W. Prandtl and his collaborators 23a-23g have carefully examined the basic precipitation of the light earths. The solubility of the hydroxides is increased by the addition of ammonium chloride or nitrate when ammonia is used as precipitant. The presence of cations like Cd" or Zn" capable of binding a part of the ammonia is favourable. The optimum conditions for lanthanum are represented by the reaction conducted at 100" in S~-ammonium nitrate solution the ammonia being added slowly a t 1% strength. The solubility of lanthanum oxide is 1.65 g./100 ml.and of praseodymium oxide 0-75 g./100 ml. under these con- d i t i ~ n s . ~ ~ d The method has been reported upon by J. Wierda and H. C. Kremers z4 who say it is more effective but less convenient than the magnesium oxide precipitation procedure. It does not appear to be suffici- ently effective with other pairs of earths t o be of practical value. When magnesium oxide is added gradually to a boiling nitrate solution of lanthanons there is a preferential precipitation of didymium and weaker bases leaving lanthanum in solution. The separation is easy judged by standards applicable in lanthanon chemistry. Probably however the magnesium process is best used only to prepare a crude lanthanum which is subsequently purified by fractional crystallisation of the double ammonium nitrate La(N0,),,2NH,N0,,4H20.Lanthanum tungstate of good quality is now in demand as an optical glass. Praseodymium.-The best crystallisation methods for praseodymium are the double magnesium or manganese nitrates for elimination of neo- dymium and the double ammonium nitrate for the elimination of lanthanum. The manganese salt is said t o surpass the magnesium salt in speed but magnesium i s cheaper and more easily eliminated when no longer wanted. Praseodymium and neodymium are the most difficult pair of the light earths to separate by crystallisation but the anhydrous nitrates have been found to differ widely in solubility in ethyl ether a t low temperat~res.~s Praseo- dymium nitrate is insoluble at 22" a t which temperature 6 g./l. of neodymium nitrate dissolve. Two low-pressure extractions with ether boiling a t 10" raised 55% neodymium t o 85% ; while 94% praseodymium was rendered substantially pure when treated in a modified Soxhlet apparatus.One shaking of nitrate in ether a t 10" raised 83% neodymium to 97%. A large plant would appear capable of making a rapid separation by this means. G. Beck 26 has found a very successful method of separating praseo- dymium by solution of the hydroxides in fused KOH,H,O. With potassium chlorate or anodic oxygen present after a time a precipitate forms which can be largely left on decantation and which is mainly praseodymium dioxide. The residue is extracted with acetic acid and only the neodymium 2 3 b Z. ariorg. Chem. 1922 120 120 ; 23e ibicl. 122 169 ; 23g ibid. 1924 136 289. 22 T. Moeller and H. E. Kremers J .PhysicaZ Chern. 1944 48 395. 23rl Uer. 1920 53 513 ; 23d ibid. p. 311 ; 23e ibid. 1923 127 209 ; 2 4 Trotis. Amer. Electrochem. Soc. 1'32.5 48 159. 2 5 B. 8. Hopkins and L. Quill Proc. hraf. Acad. Sci. 1933 19 64. 2 6 Angew. Chem. 1939 52 536. 23f ibid. 129 176 ; MARSH THE LANTHANONS 133 is dissolved. Starting with materials containing 10% of praseodymium there was obtained a product with over 50% of praseodymium and also pure neodymium. Further comments on the process will be found later. W. Prandtl and K. Huttner 23g have separated lanthanum from praseo- dymium by a process similar to that used for tho purification of lanthanum. No details of any commercial process for preparing pure praseodymium based on higher oxide formation have been published but such a process would appear feasible and may perhaps have been used for praseodymium dioxide was marketed a t a very low price in 1939.Neodymium.-It is doubtful if Demarqay's double magnesium nitrate fractionation can be surpassed on an economic basis but this again must depend upon the applicability of a process for the separation of praseo- dymium by means of liighcr oxide formation. The magnesium nitrate process quickly expels earths of higher atoiiiic weight than neodymium at the soluble end. The soliibility of saniariuin magnesium nitrate is 2.5 times as great as that of the iwodyniiiitn salt in nitric acid (d 1.325) but neo- dymium is only 36:/ grcntcr in solubilit'y than praseodymium. Perfectly pure neodymium should begin 1 o collect after 500-1000 crystdlisations. As soon as neodymium is fret from praseodymium it may be removed and any samarium present extracted by amalgam formation.6t G.Beck's potash-fusion method is cxccllcnt for separating praseodymium and neo- dymium and will save much fractionation but the writer feels some doubt as to whether it can be relied upon to give a really clean removal of praseodymium. Samarium and Europium .-These are best purified by amalgam forma- tion.6fp 6h The process will be dealt with later. The amount of europium in Travancore monazite is only about 1/80th of the amount of samarium. The double magnesium nitrate fractionation will very rapidly give a samarium salt of 95% purity using nitric acid (d 1.3) as solvent. Care must be taken that magnesium nitrate does not crystallise out but in presence of a seed of double salt it will redissolve and the double salt appear.Addition of bismuth magnesium nitrate in liberal quantities aids the separation of samarium and gadolinium and also of gadolinium and terbium.6b Gadolinium.-This element i s easily purified from a little samarium or europium by causing them to form amalgam.6f Purification from terbium is satisfactory by double magnesium nitrate fractionation aided by bismuth magnesium nitrate.Cb Bismuth is easily removed by fusing the salt a t the conclusion at 220" whereby tho bismuth is converted almost entirely into oxynitrate and recovered on lixiviation. The solution of gadolinium and magnesium nitrates will thcn give only a slight precipitate with hydrogen sulphide.Cb Gadolinium inagncsium nitrate is much more soluble than magnesium nitrate and in order to obtain satisfactory crystallisation nitric acid is used as solvent and bismuth magnesium nitrate continuously added at tho tail of the series.The separation of terbium is followed by observing the tint of the oxide. This may be done by direct ignition of the double salt except in the most critical cases. Gadolinium oxide with O - O l ~ o of terbium oxide will not be pure white. The tinting is more delicate a8 8 test 134 QUARTERLY REVIEWS than arc spectroscopy but white gadolinium oxide is readily obtainable. Alternatively a dimethyl phosphate fractionation gives a rapid method of purification.6dS 27 The Use of DoubZe AZkaZi 8uZphates.-Potassium or sodium sulphate has been in use since 1803 as a reagent to separate the light and heavy earths yet little is known of how best to conduct the separation.The sulphates and double sulphates are much given to forming supersaturated metastable solutions so that phase studies are difficult and have afforded little practical guidance towards effecting separations. C. James and H. C. Holden,,8 for instance found that nine months of continuous shaking was insufficient to establish complete equilibrium in a study of the system Y2( SO4),-Na2SO4-H,O at 25". The work of W. Schroder on Ce,(SO,),-K,SO,-H,O 29a and Ce,(SO,),-(NH,),SO,-H,O 2 Q b systems eovers a complete range of temperatures but F. Zambonini and his collaborators 3O have studied many of'the phases of the sulphatos of La Ce Pr and Nd with NH, Na K Rb Cs and Tl(1) a t 25". As many as six phases are reported for the Nd-K-H,O sulphate system alone at this temperature.The data on middle and heavy earth double sulphates are very scanty and some were obtained before materials of sufficient purit'y were available. It may be said that for sLn,(S0,),,yM,S04,zH20 the precipitate formed froiii solutions low in y is very often x y x = 1 1 2 ; that y/x = 1-6 ; that as temperatures rise solubilities fall and also often the value of x in t'lic solid phase ; and t,hat high y values are associated wit>h prwipit8ates which arc produced by addition of much alkali sulphate to hot solutions. In other words the lanthaiion double sulphates are most fully precipitated by abundant alkali sulphate and high temperature. So far little evidence in favour of one or the other of these variables has bccn produced but it may be surmised that slow precipitation will be tJhe most sclcctivc.%'here the double sulphates have a inoderate solubility this is likely to be attained with moderate alkali sulphate in solution by slow warming. However a reported csperience is that the temperat'ure influence is not large.31 The earths probably separate by double sulphate precipitation in serial order with yttrium interpolating between holmium and erbium. Cerium is best removed or reduced to the tervalent form before applying the process. Only mixtures with more than 20-30% of heavy earths are suitable for treatment. For others a double magnesium nitrate fractionation may prove more effectivo till the simple heavy earth nitrates collect a t the soluble end. The separation of tho light and hcavy earths by double 27 C. Jamcs and J. C. Morgan J .Anter. Chon. SOC. 1914 36 10 ; L. Jordan and B. S. Hopkins ibid. 1917 39 2614. 28Ibid. 1!113 35 559. 29n 2. u'jorq. Chem. 1931 220 389; 29b ibid. 1938 238 209 305. 30 Atti 12. Accad. Lincei 1924 [v] 33 301 308 ; 1925 [vi] 1 278 ; 2 153 300 374 ; 1926 [vi] 3 178 ; 4 5 86 175 424 ; 1927 [vi] 5 630 828 832 ; 1928 [vi] 7 449 ; 1929 [vi] 9 131 ; 1930 [vi] 11 771. 31 T. Moeller and H. E. Kremers Ind. Eng. Chem. Anal. 1945 17 44. MARSH THE LANTHANONS 135 sulphate formation is remarkably sharp but it is of course only fractional. Some heavy earths precipitate before all neodymium is removed but in the initial precipitation of crude heavy lanthanons e.g. from gadolinite a yield of two-thirds of the heavy earths may be expected perfectly free from neodymium. It has been found recently that carried out fractionally the double sodium sulphate process rapidly separates the middle lanthanona from yttrium and the heavy earths in monazite residues.3' The author believes that the process is capable of even further development and that the heavy group may be separated advantageously into pre-yttrium crude yttrium and post-yttrium fractions.He found that a single precipitation of gadolinite earths occupying one month gave a good concentration of ytterbium in the last precipitate and mother liquors.6c He has since treated some crude gadolinite earths with sodium sulphate splitting them into eleven fractions in ten days the final being a hydroxide precipitate. These (a) had weights ( b ) of siilphate precipitate (unwashed) of average atomic weight (c) as shown,32 ( n ) .. l 2 3 4 5 6 7 8 9 1 0 1 1 ( b ) . . 30 32 18 38 32 120 32 81 118 66 (18) ( c ) . . 135 138 128 122 112 106 102 97 102 107 127 Lanthanum and praseodymium wero maximum in 1 noodymium in 2 and yttrium in 8. Nos. 1 and 2 were practically free from heavy earths 7 was free from neodymium 10 from terbium and 11 from holmium. It is usual to start with a nitrate or chloride solution containing about 100 g./l. of oxide but a solution of anhydrous sulphates may be used if convenient. It is essential to successful use of bromate fractionation later that the double sulphate precipitation be carried a t least to the point of complete precipitation of neodymium. Lanthanum and praseodymium will then also have been eliminated. Otherwise these three earths spread among the heavy earth bromates.Omission to remove praseodymium caused it to be the last impurity which H. C. Fogg and C. James had to remove in preparing yttrium for an atomic weight determination after a brornate fractionation. 3 3 J. Kleinberg W. A. Taebel and L. F. Audrieth 33a propose the addition of aodium nitrite to a lanthanon sulphamate solution (67 g./l. of oxide). A controlled rate of precipitation by sulphate is obtained SO,NH,- + NO,- + SO,- -+ N + H,O. A slight advantage is observed over the classical double sulphate method but the value of the process would appear to depend chiefly upon cost and convenience. The Elimination of Yttrium.-The crude heavy earths from an yttrium mineral like gadolinite after elimination of the light earths will generally be found to consist of about 75 molecular yo of yttria.The first aim must therefore be to eliminate ar much of this as possible in order to reduce bulk 39 J. K. Marsh unpublished. 3 3 J . Anzer. Chem. SOC. 1922 44 307. 33a Ind. Eng. Chem. Anal. 1939 11 368. 136 QUARTERLY REVIEWS labour and cost of operations generally. Use is made of the fact that yttria has a greater apparent basic strength than the heavy earths. The separa- tion of yttrium from the earths holmium to lutecium by basic processes is satisfactory but the parting from terbium and dysprosium is less easy. The further concentration of these two earths which probably occur chiefly in fractions 4-7 of the above double sodium sulphate fractionation would therefore appear desirable. For this purpose a further short double sulphate fractionation is applicable.31 When dysprosium and terbium have been reduced to small amounts in the crude yttrium a choice must be made from several good processes for parting the heavy earths and yttrium.34 The so-called basic nitrite and the ferricyanide sz 35 precipitations are capable of placing yttrium right up the series between neodymium and samarium and may therefore be expected to give a good separation from terbium and dysprosium.Some other processes also give rapid separation of yttrium but it is less certainly known how suitable they are when applied to terbium-containing earths These include the use of urea decomposition as a source of amm0nia,~6 and basic nitrate precipitation by addition of sodium hydroxide to boiling nitrate solution till a small precipitate persists.37 On cooling a crystalline crop of basic nitrate Ln60,(N03)8,20H20 37a (or 18H20) 38 forms which is rich in the weakly-basic heavy earths.After a few crops taken thus recourse must be made to the basic nitrite process. This precipitates Ln,0,,(N0,),,17H20.33 Chromate,2 34 ph~sphate,~ 34 or cobalticyanide 34 precipitations may be equally good but are less pmc- ticable. The subject has indeed been little studied from the point of view of quick elimination of yttrium. James did not practise this but converted the total earth into bromate. W. Prandtl 5 35 recommends the ferricyanide process and this is indeed excellent,6z but the nitrite method employs a cheaper reagent ; it has the disadvantage however that with accumulating sodium nitrate the precipitate tends to become colloidal and progress stops but by this time the crude yttrium should be sufficiently extracted to make further recovery of the heavy earths from it uneconomic.The ferricyanide method may be reserved for dealing with fairly rich heavy earth concen- trates. Prandtl prefers to start with a ferricyanide treatment and to apply the double sulphate separation subsequently. The Separation of the Heavy Earths.-These having undergone a double sodium sulphate treatment for the removal of the light earths and having been concentrated by the elimination of as much y%trium as possible are converted into bromates. C. James (1908) used double decomposition between barium bromate and lanthanon sulphate. More recently lanthanon perchlorates and potassium bromate have been recommended,39 while 34 C. James et al. J . Amer. Chem. SOC. 1914 36 638 1418 ; 1915 37 1198 2643 ; 35 U‘.Prandtl and S. Mohr 2. anorg. Chem. 1938 236 243; 237 160. 36 H. C. Fogg and L. Hess J . Amer. Chem. SOC. 1936 58 1751. 37 C. James and A. J. Grant ibid. 1916 38 41. 37a C. James and L. A. Pratt ibid. 1910 32 873. 38 W. Feit 2. anorg. Chem. 1940 243 276. 39 H. E. Kremers and T. Moeller J . Amer. Chem. SOC. 1944 66 1795. 1917 39 933. MARSH THE LANTHANONS 137 Prandtl recommends preparing bromic acid from barium bromate and sulphuric acid. On conclusion of the fractionation most of the bromate can be recovered as barium bromate by treatment with concentrated barium chloride solution. On fractional crystallisation of the bromates samarium europium and gadolinium soon collect at the head. Solubility 4O increases steadily from samarium or europium down to luteciurn with yttrium falling between holmium and erbium.These last three cannot conveniently be prepared pure by bromate fractionation alone though W. Feit 3* prepared good holmium after daily crystallisations of a bromate series for four years followed by a basic nitrate fractionation. The other heavy earths-terbium dysprosium thulium ytterbium and lutecium-can probably all be obtained pure by prolonged bromate fractionation alone but more convenient methods are available. They all either can undergo a valence change or have one neighbour which does. It is likely t,hat some of the late double sodium sulphate fractions after yttrium elimination will consist largely of ytterbium in which case they are treated with sodiuni amalgam and the ytterbium 6g removed before the remainder is converted into bromate.The middle fractions of the whole bromate series which may spread to 70-80 fractions will soon become rich in yttrium. At the tail thulium more ytterbium and lutecium will collect. As soon as tail fractions are free from any absorption spectrum they consist only of lutecium ytterbium and im- purities. They are then removed and these two separated by sodium amalgam. Further up the series erbium and thulium will be separating. Thulium-containing fractions are ready for withdrawal as soon as they are free from erbium absorption bands and are found to show no lutecium arc spectrum lines. The thulium and ytterbium are then separated by sodium amalgam. The erbium-thulium separation is however very slow. Gradu- ally holmium will have been collecting above the yttrium present and erbium below it.The fractionation must proceed till the absorption spectra show that each is free from the other. This is likely to require daily bromate fractionation for a year. The separation of erbium and thulium will require even longer. C. James (1911) crystallised thulium fractions 15,000 times but the modern method of removing ytterbium should enable this number to be reduced. When erbium fractions are free from both thulium and holmium they are converted into chloride and treated with potassium ferricyanide. This is a most effective method for obtaining pure erbium.6' Unlike the basic nitrate process it does not slow up when the amount of yttrium gets small. Starting with an oxide which we may represent as Y60Er,oO160 and splitting it into six fractions these had respectively the com- positions (l) Y33Er670150 Y41Er590150 (3) Y49Er510150 (4) Y56Er440160 (5) Y69Er310150 and (6) y90Er100150* As soon as the main bulk of holmium is free from absorption bands of erbium the bromate fractionation may stop.There will then be a quantity of yttrium containing small amounts of both holmium and erbium which may serve as a source of very pure yttrium when these two are removed go 0. James et at. zbid. 1927 49 873. 138 QUARTERLY REVIEWS and set aside for further treatment with any future batch of earths. The holmium-yttrium fractions now require to have nearly all the yttrium removed by the ferricyanide process since the next step is conversion into the dimethyl phosphate and yttrium dimethyl phosphate has an incon- veniently small solubility (about the equivalent of 7 g./l.of oxide a t 25"). The ferricyanide process for holmium 61 is only a little less satisfactory than for erbium but probably a purer product can be obtained by final application of dimethyl phosphate fractionation. The dimethyl phosphates 6d are interesting salts with several peculiarities. They are the only simple salts of the lanthanons to crystallise from water anhydrous. They are prac- tically insoluble at the lutecium end of the series and increase by a factor of about 1.6 for each unit decrease in atomic number till gadolinium dimethyl phosphate is reached with a solubility a t 25" of 24 g./100 g. water after which the increase is less rapid. This means that the rate of change of solubility is very much greater than is ordinarily encountered among the lanthanons.Further it is most unusual if not unique for the rate of change to be greater for the heavy earths than for the light earths. Solu- bility also usually risos from gadolinium to lutecium whereas in the dimothyl phosphates it falls steeply but this is paralleled in the ferricyanides ferro- cyanides tartrates and perhaps other series. The solubilities like those of the sulphates and double sulphates decrease with rising temperature. The solubility is too small and the degree of hydrolysis too great for yttrium together with the earths nearer the lutecium end of the series than holmium to allow of dimethyl phosphate fractionation for their purification ; but the separation of holmium dysprosium terbium and gadolinium is good and very rapid. Since slight hydrolysis passes back basic products to the head of the series counter to the main separation the removal of the last of dysprosium from holmium or of terbium from dysprosium is not quite complete but with the less hydrolysed gadolinium t-he separation from terbium is very good.So too is the separation of the last traces of yttrium from holmium for long considered the most difficult separation in inorganic chemistry also of holmium from dysprosium and of dysprosium from terbium. Dys- prosium oxide prepared by this process always had a light buff colour but t'his tint due to terbium can be eliminated quickly by a potash fusion,eh 26 which concentrates terbium in the first oxide to precipitate. If solutions are refrigerated and crystallisat'ions conducted below 50" hydrolysis is kept to a minimum.It is possible that some double sulphate fractions may be suitable for direct conversion into dimethyl phosphate without passing t'hrough a bromate fractionation. recommends changing from bromate to double ammonium oxalate for the last stage of the holmium-erbium separation. The oppor- tunity may probably be taken with advantage to eliminate yttrium once again by ferricyanide precipitation of the richer yttrium fractions. The double ammonium oxalates are less soluble and so better suited for handling small quantities of material and give a more rapid separation than bromates. (Some double alkali sulphates do the same.) Thus terbium of the highest purity is obtainable. W. Prandtl MARSH THE LANTHANONS 139 Prandtl recommends them for the final stage in the purification of terbium holmium erbium thulium and lutecium.The writer has no experience of the salts and 110 experimental particulars with regard t o their use have been published. Separation Processes based on Reduction to Bivalent States.-It is remark- able how slow chemists have been to exploit bivalence in the lanthanon group. The field of “anomalous ) ) valence has been reviewed by D. W. Pearce up t o 1935. The then current use of the term “ anomalous ” illustrates the hypnotic effect that a chorus of fourteen tervalent elements had upon chemists for a number of years. Samarium although the least stable in the bivalent state was in 1906 the first t o be so prepared. Europium chloride was reduced in 1911 and ytterbium chloride in 1929.40a L. F. Yntema 41 in 1930 was the first to us0 reduction to a bivalent state as a means of purifying a lanthanon.He reduced EuC1 at a mercury cathode in presence of sulphate ions and obtained insoluble EuSO,. The next decade saw rapid dovelopments. It was found that the tcrvalent europium chloride solution could be reduced by metals.42 Electro-reduction of ytterbium solutions was also found to be an effective means of separating ytterbium and of obtaining approximately purc thulium and lutecium but the solubility of YbSO is higher than that of E u S O . ~ ~ ~ 6g H. N. McCoy 4 4 has done inuch successful work in tlhis field. Hc treated a chloride solution of monazite residues (386 kg. of oxalates) containing about 0.50/ of europium with zinc dust (5 kg.) and acetic acid and aftcr reduction had taken place with magiicsium sulphate (30 kg.) and barium chloride (8 kg.).The forma- tion of barium sulphato helped to carry down the isomorphous but slightly soluble enropous sulphate. Treatment of the precipitate with sodium carbonate and hydroxide yielded 7.2 kg. of 20% europous hydroxide. Two more treatments yielded a 70% product and this in concentrated chloride solution (d 1-35)? treated with concentrated hydrochloric acid precipi- tated EuC1,,2H,07 giving a very perfect separation from tervalent chlorides. McCoy 45 also studied the electrolytic preparation of amalgams of europium and ytterbium using acetate solutions in presence of potassium citrate. Yields were not unsatisfactory but for samarium his maximum yield was 13.6%. Some interesting observations were also made by German chemists. It was found by using strontium amalgam on sulphate solutions of the light lanthanons that these in addition to europium ytterbium and samarium were to some extent reduced and found stabilised as bivalent sulphates mixed wit,h strontium sulphate.46 I n purifying samarium by reduction 1 % calcium 400 1%‘.Klemm a i d \V. Schiith 2. anorg. C‘hena. 1929 184 352. 4 1 J . Amcr. C h m . Svc. 1930 52 2782. 4 3 W. R. Pearce Thesis University of Illinois 1034. 43 R. IV. Ball and L. F. Yntemn J . -41nsr. Chenz. Soc. 1930 52 4264 ; W. Prandtl 41 J . iln2er. Chcm. Soc. 1935 57 1756 ; 1936 58 1577 ; 1937 59 1131 ; 1939 4 5 Ibitl. 1941 63 1622 3432; 1942 64 1009. 46 1,. Holleck aid W. Noddack Angew. Chem. 1937 50 819 ; L. Holleck Atti X %. coiorg. Chein. 1932 209 13. 61 2153. G ‘ O H ~ . Interti. Chim. 1938 2 671.140 QUARTERLY REVIEWS amalgam waa prepared in a special steel bomb and used to treat anhydrous chlorides in dry ethanol containing some hydrochloric acid. 55% Eu-Sm was raised to 92% in one precipitation and to 99.8% on a second treatment and recovered as red samarous chloride by centrifuging. 1-1.5% of samarium was left with the tervalent earths. SmSO is red but the bivalent carbonate and hydroxide are reported to be gree11.4~ It was early evident that bivalent lanthanons resembled alkaline earth elements. The sulphates are isomorph~us.~~ EuSO is more soluble than BaSO and possibly than SrSO,. YbSO dissolves to the extent of 4 g./l. in O.%-sulphuric acid. The addition of freshly precipitated SrSO has been used as a means of more fully precipitating YbS0,.,7 Now the alkali and alkaline earth metals in amalgam if treated with a different alkali or alkaline earth solution all undergo reciprocal replacement following the law of mass action though the equilibrium may be greatly in favour of certain metals remaining in amalgam and others in solution.The displacement order beginning with the most positive amalgam is Mg Li Ca K Na Sr Ba.48 Since lanthanon(I1) is virtually an alkaline-earth metal these con- siderations lead the Reviewer to argue that treatment of a lantlha8non solution with a suitable amalgam should bring about reduction at the amal- gam face followed by replacement giving rise to a lanthanon(I1) amalga~n when the relative positions in the amalgam series were favourable. It should be noted that the amalgam series is not the same as the electropositive series on account of compound formation.McCoy's use of acetate solutions for electro-reductions suggested their use here but the presence of an alkali in solution was considered to be contra-indicated. It was found a t once that very full and perfect exchange between sodium amalgam and europium samarium or ytterbium acetate solutions took placeq6e Samarium reduces and passes into amalgam with an ease comparable to europium and probably more readily than ytterbium. Heretofore it had proved much less tractable to reduction and was never successfully obtained as SmSO on account of the great instability of the Sm" ion. The removal of the last traces or down to 0.01% of the reducible lanthanon in a preponderating quantity of a tervalent earth is possible provided that the solution is rendered free from sodium and that fresh amalgam free from lanthanon is finally used.Only dilute sodium amalgam can be employed. With over 0.3% sodium on exchange partially solid lanthanon amalgam may result particularly with europium and cause trouble in the mechanical separation of amalgam and solution the ease and perfection of which is one of the advantages OP the method. Furthermore strong sodium amalgams tend to react with water liberating hydrogen and causing accumulation of sodium in the lanthanon solution. The efficiency of the reaction which should amount to SO% is lowered. In order to prevent hydroxide formation the solution 47 A. Brukl Anyio. Chem. 1937 50 26 ; 1938 51 192 ; 1939 52 151. 48 E. F. Smith J . Amer. Chem. Xoc. 1905 27 540 ; J . Physical Chem.1905 9 13; Amer. Chena. J . 1907 37 506; E. F. Smith and J. R. Withrow J . Amer. Chem. SOC. 1907 29 321 ; S. B. Frank and J. R. Withrow ibid. 1920 42 671. MARSH THE LANTHANONS 141 is kept weakly acidic with acetic acid during the course of the reaction which may take 1-2 minutes. In preparing samarium amalgam some of the tervalent earths (chiefly neodymium) may also enter the amalgam but they appear t o be held in the amalgam in a different combination and much more firmly. Thus from 70% neodymium the 30% of samarium was extracted as amalgam and the greater part of this was obtained with O.Olyo of neodymium by decomposing the amalgam with weak acid but the last small acid extract contained 50% of Samarium and europium being neighbours are extracted together as amalgam but they can be separated by acid attack on the amalgam.From the atomic volumes of the metals we know that neodymium is tervalent samarium apparently partially tervalent and europium largely bivalent.*Q This probably also holds in amalgams for europium is more readily attacked by acid than samarium. Dilute acetic acid mixed with a little dilute sulphuric acid yields with the mixed amalgam samarium(II1) acetate and a mixed (SmEu)SO precipitate containing perhaps 20% of EuSO,. Since europium accumulates in early fractions of the sulphate precipitate europium amalgam must be more readily attacked than samarium amalgam. The mixed sulphate is then selectively oxidised by dilute nitric acid giving a 90% europous sulphate. This is converted via acetate into amalgam once more and is then subjected t o attack by concentrated hydrochloric acid.There results a precipitate of pure EuC12,2H20 which is filtered off and washed with concentrated acid. The filtrate contains all the samarium with a little europium. I n order to obtain samarium free from europium samarium acetate solution is treated with a little dilute sulphuric acid and dilute sodium amalgam. A lanthanon(I1) sulphate precipitate is formed in which europium rapidly concentrates. Thus the precipitate may be pale a t first but later yields are red and mainly SmSO,. These are examined spectroscopically and further crops taken till the arc lines of europium are no longer found. The samarium acetate solution is then absolutely free from europium. Eight precipitates amounting in all to 804 of the samarium were required to prepare samarium of a purity never before obtained (absence of Eu line A 3819-66 intensity 500).6h It is interesting to noto that high concentrations of europium(I1) solu- tions may stabilise other very weakly bivalent lanthanons.Thus a little neodymium in concentrated Eu" solution may not be detectable by its absorption spectrum. This is predictable from the law of mass action. Similarly it is difficult to extract the last of a potentially bivalent earth from a predominantly tervalent solution. This difficulty is however lessened in the case of samarium and europium since their neighbours are capable of some bivalent ion formation. Nevertheless in the cam of ytterbium-lutecium mixtures lutecium having no bivalent stability it was found possible to reduce ytterbium in lutecium to a concentration believed to be below 1 in 10,000.W. Klemm and H. Bommer 2. unory. Chenz. 1937 231 138 ; H. Bommer and E. Hohmann ibid. 1939 241 268. 142 QUARTERLY REVIEWS The amalgam methods just described give fuller separation and purer products than the older methods based on bivalent sulphate formation. Where sulphate formation is used as in the sepa.ration of samarium and europium the solubility is lower in the acetate solutions employed than in chloride solutions. A more theoretical investigation of the formation of ytterbium amalgam has confirmed the use of acetates as the salts of greatest practical value yet found.50 These sodium amalgam methods supersede all previous methods for separating the lanthanons by means of bivalency . Separations based on Oxidation to Quadrivalent State.6i 6j 6k-Cerium has already been dealt with.It is the only lanthanon which gives rise to a quadrivalent ion. Praseodymium and terbium however give oxides in which quadrivalence is shown. Unlike CeO the oxides are dark or black and it is difficult to maintain that there is full valence bonding of the extra oxygen. The structure of the lanthanon oxides from samarium to luteciurn is cubic of a peculiar type the unit cell of which may be regarded as com- prised of eight modified fluorite type cells. Cerium dioxide has a true fluorite type structure but in these others Ln203 1/4th of the fluorine positions are unfilled by oxygen. It is thus seen that very little structural change is required to fit in the oxygen required to give Pro or TbO,. The amount of dioxide formation is much affected by oxygen gas pressure.51% 6k The extra oxygen appears to be not fully valence bound but on the other hand not entirely in solid solution.It was pointed out earlier how quadri- valence was related to atomic structure. Praseodymium sesquioxide usually has a hexagonal form similar to La,O,. In this form it is incapable of dioxide formation but the transition to cubic Pr,O is ready and by oxidation this passes to R stable cubic structure When light lanthanon nitrates are fused with mixed alkali nitrates cerium dioxide is precipitated a t temperature from 200" upwards and is largely removed from effective participation in the system. Neo- dymium and praseodymium oxides tend to separate only at temperatures above 300". If a cerous salt is thrown into the melt at those temperatures it gives of necessity at once a cubic oxide which tends to stabilise cubic neodymium and praseodymium oxides and thus t o promote the clean separation of these elements from lanthanum the oxide of which has little cubic stability and which therefore stays in the melt as nitrate since also it is the strongest base.The lanthanon hydroxides have considerable solubility in fused potash KOH,H,O at temperatures of 300-320" but as G. Beck 26 has found after a time with oxidising conditions produced either by the presence of potassium chlorate or anodic oxygen a precipitation of praseodymium or terbium dioxide takes place. Beck believed this to be due to these elements giving a quadrivalent anion but the author who has pointed out that water is necessary for the solution in potash to take place believes that loss of water is the primary cause of precipitate formation.If a tervalent 6 o T. Moeller and H. E. Kremers I d . Eng. Chem. Anal. 1945 17 798. 6 1 W. Praridtl and G. Riedcr Z. ccttory. Cliem. 1938 238 225. MARSH THE LANTHANONS 143 hydroxide can pass to a quadrivalent oxide it ceases to be a competitor in the system for the limited amount of water and therefore given available oxygen the quadrivalent oxide will be the first to precipitate as the tem- perature is raised too high for water to be retained. The higher oxide of praseodymium or terbium thus heavily concentrated in the early precipitate settles so that much of the melt can be decanted. The higher oxides are not easily attacked by acetic acid and use is made of this fact to separate them from hydroxides.The process in its present form unfortunately appears too costly for commercial application but the exploitation of valence changes is not yet complete. Other Modes of Fractionation Extraction Processes.-Fractional partition of lanthanum and neodymium thiocyanates occurs between water and n-butyl alcohol the Nd/La ratio in the alcohol being 1.06 of the water value.62 Though the separation is small it is proposed by the use of a suitable counter-current extraction apparatus to exploit it rapidly. The extraction by ether of scandium thiocyanate 53 from aqueous solutions has been found to be very rapid and to give better separation from Ln Al Fe Th and Zr than older methods. A variety of zeolitic materials and methods were used by R. G. Russell and D. W.Pearce 54 and aluminium oxide by 0. Erametsii et The small lanthanon ions were found to be preferentially held and pH has an influence. The addition of citric acid reverses the absorption series in the light earths. The results so far give little promise of successful application on a practical scale but t’he methods may prove useful in anal~-tical operations. Zeolite and chromatographic methods have been tried. This review should serve t o show that the arduous task which the chemist once undertook in starting to separate the rare earths has now been much lightened. Yet it is still formidable particularly for t,he scarce heavy-group coloured earths found mixed with much yttrium. With their persistent tervalency only laborious crystallisation will separate holmium erbium and thulium. Knowledge of the close resemblance in physical properties like ionic radii gives no ground for optimisni that easy methods of separation will ever be found. Yet given the incentive nothing stands in the way of large-scale production of any member of the group. 6* D. B. Appleton and P. W. Selwood J . Amer. Clmn. SOC. 1941 63 2029. 63 W. Fischer and R. Bock 2. airory. Chem. 194’3 249 14G. J . Amer. Chem. SOC. 1943 65 595. 6b Bull. comm. geol. Finlunde 1941 14 36 ; Ann. Acad. Sci. Fennicz 1943 A 57 Nos. 3 5.

ISSN:0009-2681    

DOI:10.1039/QR9470100126

出版商:RSC    

年代:1947

数据来源: RSC

摘要:

REPRESENTATION OF SIMPLE MOLECULES BY MOLECULAR ORBITALS By C. A. COULSON M.A. D.Sc. F.R.S.E. (PHYSICAL CHEMISTRY LABORATORY OXFORD) Introduction THIS account of the method of molecular orbitals is divided into five distinct sections (1) The method of atomic orbitals for the study of individual atoms. (2) Diatomic molecules leading to the characteristic differences between (3) Polyatomic molecules involving localised bonds leading to the theory (4) Polyatomic molecules involving non-localised bonds leading t o the (5) Some general related topics including hyperconjugation. In an account of this longth it is not possible to report anything like all the work that has been published in the last twenty years since the subject began ; and t,he situation is made worse by the fact that no satis- factory simple account of tho method of molecular orbitals by itself has ever been given.Our mothod of approach thereforo will be first to explain the general character of the theory and then to illustrate its power by selecting a series of typical applications. For the benefit of those who wish to study any section in more detail wo include references t o suitable literature. Like all problems in theoretical chemistry tho detailed working out of any part of the theory requires a certain familiarity with mathematical mothods. I n this case thc essential technique is that of solving the Schrodingor wave equation under certain given conditions ; but the fundamental ideas behind the theory are easily understood by experimental workers with little or no mathematical experience. For that reason we have avoided all mathematics and have insisted everywhere upon the pictoria'l character of each step in the argument.It is the visual char- acter of these steps that makes the method of molecular orbitals so funda- mentally simple but this situation must not blind us t o the fact that in practically every case quanti tativo as well as qualitative predictions have been made. Indeed ;t study of recent literature shows that this method originally rather neglected is now receiving more attention than the alter- native method of valence-bond structures first introduced by W. Heitler and F. London and so lucidly developed by L. Pauling in his book " The Nature of the Chemical Bond ".2 A comparative study of the two methods single and double bonds. of stereochemistry and valence angles.phenomenon sometimes referred to as resonance. Cornell Univ. Press. Z. Physik 1927 44 455. 144 COULSON MOLECULAR ORBITALS 145 to which the present account is much indebted has been given by J. H. Van Vleck and A. Sherman.3 1. Atomic Orbitals General PrincipEes .-Before we can profitably deacribe the motions of molecular electrons we must briefly review the simpler problem of how electrons move in isolated atoms. We need to study this not only because molecules are built out of atoms but also because the method of molecular orbitals runs closely parallel t o the atomic problem and a8 we shall see in Section 2 molecular orbitals are themselves compounded out of atomic orbitals. If we make use of the self-consistent-field theory introduced by W. R. Hart~-ee,~ we may summarise the behaviour of the electrons in at!oms by the following three principles In the older quantum theory of Bohr this orbit was as precise and clear-cut as the orbit of a planet round the sun ; but in the newer wave-mechanical theory where such precise measurements of position are not possible and a statistical description has to be used the orbit is described by a wave function y .Not all wave functions are possible but only those which are found as solutions of the Schrodinger wave equation. As y defines the orbit we call it an atomic orbita2 (A.O.). The value of y for any electron varies from point to point and y 2 at any place measures the probability that the electron will be found a t that place. Thus if y is large anywhere irrespective of its sign which has no physical meaning the electron is likely to be found there.For many purposes a more pictorial (though less strictly accurate) interpretation of the wave function may be given as follows the electron may be regarded as spread out in the form of a cloud (charge-cloud) the density of this cloud at any point being proportional to y 2 . In places where y2 is largest the charge-cloud is most dense and most of the negative charge is to be found. The objection to this interpretation is that a single electron cannot possibly be distributed over regions of the size of one or two Bohr radii. Only the statistical or probability interpretation is really valid but the charge-cloud picture is a useful one. Thus we shall often have occasion to represent atomic and molecular orbitals diagrammatically by boundary surfaces; these surfaces are such that if they are drawn to the right scale almost all the electronic charge ( e .g . 90yo) lies within the contour drawn and for most purposes we may say that the electron is confined within this boundary. Examples of these contours are given in Figs. 1-4. ( b ) Each wave function has its own appropriate energy and if this is suitably determined the total energy is approximately the sum of the energies of the constituent orbitals. ( c ) I n addition to its space wave function y each electron has a spin which must have one of two values (& 8 in normal units) and the Pauli Proc. Camb. Phil. SOC. 1928 24 111 426. (a) Each electron is assigned to a definite particular orbit. a Rev. Mod. Physics 1935 '7 168. 146 QUARTERLY REVIEWS " exclusion principle " tells us that in no case may two electrons have both the same y and the same spin.Two electrons may have the same y~ only by having opposed spins. Such electrons generally exert a repulsion upon other electrons near them. In this case we speak of them as paired. For our later purposes it is absolutely essential to have in our minds a clear picture of the appearance of the more common atomic orbitals. This picture must above all reveal the symmetry properties of the A.O. In ordinary atoms the A.O. are classified as s p d . . . Fig. 1 shows the boundary contours for some of these orbitals. The s-type is spherically symmetrical and the significance of the diagram is that the electron is COULSON MOLECULAR ORBITALS 147 almost certainly to be found inside the sphere shown.There are three p-type orbits in each of which the electron is practically confined to two regions together resembling a " dumb-bell ',. In one half of the dumb-bell y is positive and in the other it is negative as shown. There is a very marked directional character in these orbitals which we exhibit by means of the suffixes p, p, p,. An important fact about these A.O. is that the regions where y is of opposith sign are separated by a " nodal plane " over which y = 0. For example in the orbital pz this nodal plane is the yx plane. The three p-type orbits are entirely equivalent except for their directional property. I n a similar way there are five d-type orbitals two of which are shown. In some respects a d orbit is like the superposition of two p orbits for there are four similar regions of alternating sign separated by two nodal planes.In the dw d,, d, orbitals these are two of the co- ordinate planes. I n the so-called d,,-,, dZ1-,a orbitals which are obtained by rotating the previous dq . . . orbitals through 45" the nodal planes are of course also rotated. One might have expected a d,L-yt orbital as well as a d,,-,, d,,-, ; and indeed it does exist. But it may be verified from a superposition of the appropriate diagrams or else from the analytical forin of the wave functions themselves that the sum W(&-y') + w(d,+4 + y(4+d is identically zero. This means that only two of these three are independent. Any pair of them together with the previous d,, . . . make it total of exactly five independent and equivalent d-type orbitals.* More exact diagrams of these orbitals may be found in a paper by H.E. White.5 The full description of an A.O. requires in addition t o its s p d . . . character a knowledge of its principal or total quantum number and its spin. 1s <2tt < 2 p <3s < 3 p < 3 d . . . Tn giving the electronic state of an atom we adopt an " aufbau " or building- up process in which electrons are fed one at a time into the allowed levels beginning with the Is and satisfying the exclusion principle by allowing only two to each of the orbitals just described. Thus hydrogen in its ground state is represented by (Is) helium by ( l ~ ) ~ nitrogen by (l~)~(Ss)~(2p)3 oxygen by (l~)~(zs)~(223)~ etc. In the cases of nitrogen and oxygen however and all other atoms containing incomplete groups or sub-groups this still leaves undecided just which of the equivalent orbitals (here the 2p orbitals) are filled.For this we make use of Hund's rules.6 These are that for equivalent orbits (i) electrons tend to avoid being in the same space orbit so far as is possible and (ii) two electrons each singly occupying a given * For the purposes of this account we have chosen always to use purely real wave functions In such a case the five d orbitals are equivalent and so are the three p orbitals. But for some purposes (e.g. magnetic) a different representation is wed in which the set of d orbitals splits into three groups according to the component of angular momentum in a particular direction (compare the 0 n . . . classification in Section 2). The order of energies is well known to be Phy&al Rev.1931 37 1416. 6 See e.g. Herzberg " Atomic Spectra and Atomic Structure " Blackie Chap. 3. 148 QUARTERLY REVIEWS pair of equivalent orbits (e.g. 2pz 2p,) tend to have their spins pardel in the state of lowest total energy. For instance nitrogen in the ground state has the 18 and the 29 orbitals doubly filled and each of the 2pZ 2p, and 2pz orbits singly filled the three electrons concerned having parallel spins; and oxygen with one more p electron has one orbital (say 2pJ doubly filled the other two being singly occupied with parallel spins as in Fig. 6. The (2pJa group which as we shall me in Section 3 make no contribution to the bivalent character of the oxygen atom are sometimes called the " lone pair " electrons. A similar name is used for the nitrogen (2~)2 electrons.Yet we have mentioned it because it is not so well known that precisely equivalent rules describe the behaviour of electrons in molecules ; but before we can deal with this Hybrid~aatbn.-~l this is very well known. ty I FIG. 2 A single tetrahedral orbital. there is one new factor to be intro- duced. As it is of greatest importance in carbon let us describe it for this atom. According to what we have just said the normal state of carbon is (l~)~(Zs)~(Zp)2 in which there are two unpaired electrons (e.g. 2pz 2pJ and the spectroscopic state is SP. This corresponds to bivalency ; the characteristic quadrivalency can only be obtained if we start from a state with four unpaired electrons. A suitable state may be obtained if we excite one of the 2s electrons to the empty 2pe orbit making a sS state.Approxi- mately 65 k.-cals./mole are required to do this ; 7 but even now the four valence electrons are not equivalent being 29 Zp, 2pV Zp,. In order to get equivalent bonds we must mix these " pure " orbitals together and form " hybridised " orbitals. As Pauling 8 has shown there are at least three such linear combinations that are of the utmost importance. They may be called tetrahedral trigonal and di-gonal (or diagonal but neither word is particularly attractive !). In the tetrahedral hybridisation we replace the four orbitals 29 2p, 2py 2pz by four others t, t, t, t ; y(tl) is shown diagrammatically in Fig. 2. [With these axes y(tl) = $y(Zs) + $y(2p,).] Evidently the electron cloud is greatly concentrated along a particular direction here the positive x axis and the wave pattern is cylindrically symmetrical around this direction.The other three orbitals t, t, and t, are entirely equivalent to t, except that they point in the three remaining directions of the vertices of a regular tetrahedron surrounding the carbon atom. We may anticipate that the characteristic quadrivalency of carbon is associated with the fact that L. H. Long and R. G. W. Norrish Nature 1946 157 486 ; 158 237 ; Proc. Roy. ( a ) J . Amer. Chem. SOC. 1931 53 1367 ; ( b ) ibid. 1932 54 988 3670. ROC. 1946 A 187 337. COI'LSON MOLECULAR ORBITAL5 149 when it is prepared for the formation of a s<iturated ~noleculc like CH or C,H, there is one electron in each of the four orbits t . . . t,. I n the trigonal hybridisation 28 2pz and 2py are compounded to give 111 soniewhat similar in shape three equivalent coplanar orbitals I 11 to t, and pointing a t angles of 1200 in the zy plane as in Fig.3. [Here T h e remaining orbital is the undisturbed 2pz generally referred to in this con- nection as the n orbital. This form of hybridisation is evidently associated with the aromatic state of the carbon atom (e.g. ethylene benzene) where the three valence angles are known to be cxactly or approximately 120". The third or di-gonal type leaves 2py and 2p2 unchanged increly mix- ing the 2.s and 2p orbitals in the y(b) = 2/4(y(Zs) - y ( 2 p ) ) . As shown in Fig. 4 these hybrids point in opposite directions along a straight linc. y(1) = dQy(24 + 2/$y(2Pz)*l f o m s y ( 4 = \'B (VJZ(2*S) 1- y(2pz)) FIG. 3 Trigom1 orbitals [mi? imluding $(pZ) which. i0 us in Fig.I]. They iLre obviously related to the state of the carbon atom prepared to form compounds such as C,H and CO,. Small variants of these three fundarnent,al types occur in molecules FIG. '4 [)lot includitty $(pll) uiul +(pz) which (ire as in Fig. 1J. D i -gonal hyDridi.sd i u t L such as CH,Cl where all four bonds are not quite equivalent but they do not differ appreciably from the above and we shall not discuss them here. When electrons of the carbon atom (or other atom) have been placed in these new hybridised orbits which replace the old s and p orbits we speak of the atom as being in a '. valence state " (see particularly J. H. Van Vleck 9 and R. S. Mullikeri lo). The advantage of the valence state 9 J . Chem. Physics 1933 1 1'77 219; 1934 2 20. lo Ibid. p. 782. 150 QUARTERLY REVIEWS is that the various orbitals have a very strongly directed probability func- tion.The disadvantage is that energy is required to form the hybridisation though this energy is less in the case of carbon where ~ ( 2 s ) and ~ ( 2 p ) have more nearly equal energy than in oxygen where the difference is grater. But we reap our reward later when we find a correspondingly greater gain in energy through the formation of stronger bonds. 2. Diatomic Molecules; Single and Double Bonds General Principles .-There is absolutely no reason why we should not use exactly the same principles for describing the ebctron states in a diatomic molecule as in a single atom. However on account of the fact that the electrons move in the presence not of one but of two nuclei there ;tw certain cxtra features to be considered.We may therefore start as f f d l 0 J f s In) Ehch clcctrori in it inoleculc is described by a wave function y and t h e value of y12 ; i t any poiiit rcpresents the probability of finding the electron in mitt voliiine around that point. These wave functions may appropriately bc callccl molecular orbitcrZs,ll abbreviated t o M.O. to distinguish them froin atomic orbitals (A.O.) and to remind us that they are no longer rnonoc~ntric but are polyccntric since an electron which takes part in molecule formation inust not be confined to one nucleus alone. ( b ) Each y has its own appropriate energy and if this is suitably deter- mined the total electroiiic energy is approximately the sum of the energies of the occupied orbitals. ( c ) Each electron has a spin and Pauli’s exclusion principle must be satisfied just as in the case of an atom.Characteristic molecular features occur in the following way. A strict calculation of y ought to be made by solving the appropriate Schrodinger wave equation. However except for H, this is a mathematical task of too great complexity; but we avoid t,ho difficulty by noting that when an electron is in the neighhourhood of any one nucleus the forces on it arise mainly from that nucleus and the other electrons on the same nucleus ; this means that both the wave equation and its solution must resemble t,hs corresponding equation and solution for an isolated atom. The molecular orbitlad therefore resembles a series of superposed atomic orbitals one at each of tihe constituent nuclei. We may describe this by a new principle This fundamental idea on which the whole of the structure of molecular orbital theory ulti- mately rests is due t o J.E. Lennard-Jones,12 who applied it to the case of homonuclear systems H, N2 etc. I n the form just stated which Mulliken 13 calls the L.C.A.O. approximation (linear combination of atomic orbitals) this principle is not an exact truth ; for the best possible M.O. for a diatomic moleculo are not simply combinations of two A.O. However as a detailed ( d ) Each M.O. is compounded out of atomic orbitals. R. S . Mullikeii. P i i y s i d Rev. 1932 41 49. l 3 Trans. Fnraday SOC. 1929 25 668. l 3 J . Chem. Physics 1035 3 375. COULSON MOLECULAR ORBITALS 151 etudy for the more manageable case of H shows,l*s 1 5 the approximation is pretty good and without some sort of approximation no progress could be made.Consequently this technique of representing M.O. as L.C.A.O. has now become perfectly standard and we shall adopt it throughout. There is in fact a particular significance in the L.C.A.O. relation between a molecular orbital w in EL molecule AB and its component atomic orbitals yA and v,. This is seen most clearly when the two electrons that occupy the orbital ly are valence electrons contributing to the bond A-B. Normally one of these electrons comes from each atom and if the interaction between the atoms could be prevented these electrons would be in orbitals yA and yB. Thus the molecular orbital y which will accommodate two electrons is correlated with the two atomic orbitals into which we might expect the molecular electrons to go when interaction between the atoms was arti- ficially prevented as e.g.by soparating them to infinite distance. I n a loose kind of way we could therefore speak of '' the electron originally in state vA " and " the sloctron originally in state yB " being paired to form molecular electrons in state y~ for the bond A-B. So we might imagine ourselves bringing up the atoms A and B towards one another without allowing any interaction and then pairing together suitable electrons in A and B. This rather crude and certainly not completely accurate de- scription of molecule formation reveals a link between M.O. theory and both the Lewis shared-electron-pair-bond and the Langmuir octet theory. But the L.C.A.O. theory is not miich use to us unless we have some criterion for deciding which of the possible orbitals ?pA is to be combined with Y,.This brings us to anothcr principle ( e ) The energy of a M.0. is lowest ( i . e . binding energy is greatest) when the component atomic orbit& overlap one another as much as possible. This is the " criterion of mnxiiniiin overlapping " originally introduced by J. C. Slater and L. Paiiling so to explain the origin of directional valence. Although these writers developed the criterion for use in a different kind of treatment from ours we shall find in Section 3 that it applies beautifully to M.O. theory and provides an ultimate explanation of all the fundamental rules of stereochemistry. A proof of the validity of the criterion is given by C. A. C0u1son.l~ I n HCl there are two unpaired electrons whivh form a bond.I n the separate atoms they are the H(1s) and thc C1(3p,) atomic orbitals. The molecular orbital will be found by a combination of these but the overlap of the two is greatest when the x direction froin C1 points directly towards the H atom. This means that the H atom will lic along this direction and the two sets of lone-pair electrons on thc C'1 atom (3p,,)2((3p;)Z will point in directions at right angles to thci bond H-Cl. (f) M;ltheinatic;-11 ctllcllli1I ion I 7 shows that other things being equal An example will explain this principle most easily. l p C. A. C:oulsoii T r u m Faraday ,Yoc. 1937 33 1479. l 5 Idcut Proc. Ca,itb. Phil. Sot. 1938 34 201. l6 Physical Rct). 1031 37 481 ; 38 1109. l7 Proc. Ccrnib. Phil. SOC. 1937 33 111. 152 QUARTERLY REVIEWS the binding energy is greatest if the component A.O.have as nearly equal energies as possible. Indeed if these energies (for a diatomic or a polyatomic molecule) are not of comparable magnitude no significant combination occurs. Thus in the HCl molecule just discussed the binding energy of H(1s) is vastly less than that of any of the K- or C-shell electronb in C1 so that these latter are practically unaffected by the association with H. Only the 3pz 3py and 3p are of approximately equal energy with the H(1s) and as we have seen the criterion of maximum overlapping picks out which of these co-operates with v(H 1s) in the formation of a M.O. For this reason we need only consider together atomic orbitals from the valence shells of the various atoms the inner-shell electrons retaining atomic character.The same is true even when as in homonuclear molecules such a8 Cl, the inner electrons of the one atom have identical energies with those of the other. Here it may be shown on account of the fact that the A.O. for these inner-shell electrons scarcely overlap a t all that the result of compounding them to form M.O. gives a final distribution of charge effectively indistinguishable from that when they are not com- pounded. This clears up an ea,rlier difficulty l2 regarding the use of atomic or molecular orbitals for inner-shell electrons. Our conclusion is that only the valence-shell electron orbitals are combined together all the others being given their non-bonding atomic chara,cter. This is of course in complete accord with chemical behaviour. Homonuclear Diatomic Molecules.-Let us consider the application of the above principles to diatomic molecules in which the two atoms are identical -homonuclear diatomics as we may call them.If the nuclei are A and B then molecular orbitals may be obtained by combining together atomic orbitals yA and yB one from each atom. But the condition (f> of equal or nearly equal energy makes it clear that yA and yB must refer to orbitals of the same kind around their respective nuclei. Thus they may both be 1s orbitals as in H, or 2s orbitals as in Li or both may be Zp or both be 2p orbitals etc. Suppose for the moment that they are both 1s orbitals y(A 1s) and y(B 1s). This will make our discussion apply particularly to the case of H2. Then the L.C.A.O. principle ( d ) provides us with the molecular orbital Y = y)(A 19) + hp(B IS) .* ( 1 ) where 3 is a constant. Quantum theory interprets such a wave function by saying that the relative probabilities of y ( A Is) and y,(B 1s) are in the ratio l 2 L2. Now by symmetry (Pauling l8 seems to have been the first to notice this particular point but the more general discussion is due to Lennard-Jones 12) the electron must be equally divided between A and B so that A2 = 1 . There are t'herefore two possible L.C.A.O. which we can form and they are Y g = y(A IS) + y(B IS) . - (2a) Y . = v!(A IS) - y(B IS) . * ( 2 b ) Chem. Reviews 1928 5 173, COULSON MOLECULAR ORBITALS 153 The subscripts g and u are short for gerade (even) and ungerade (odd). All M.O. with subscript g are such that y has the same value a t pairs of points diametrically opposite with respect to,the centre of symmetry in the molecule.We may say that U is even for reflection in the origin (or mid-point) ; ?Pu is odd so that it changes sign on reflection in the origin. These subscripts may be applied either to individual M.O. as here or to the whole molecule. The importance of such a classification may be seen from the fact that the only allowed transitions are such as change the character ; ie. u + g or g -+ u. With heteronuclear molecules such as HCl where the two nuclei are different this type of symmetry about the mid-point no longer exists and separation into g and u orbitals is no longer possible. Fig. 5 (a) shows the two boundary surfaces (somewhat schematically) for the two cases. On the left we show the two separate atomic orbitals and on the right .the two 31.0.Yg and !PI compounded from them. Both boundary surfaces have coniplete symmetry around the axis (this corresponds to zero component of angular momentum in this direction) ; when this happens we call the molecular orbitals a-type. But Fig. 5 shows very clearly that in the !Pg M.O. the electrons are drawn into the region between the atoms and in the !PI' case they are thrown away from the bond. Accurate calcu- lations 15* 20 can be made both of the shape and of the energy of these two orbitals; and it appears that the symmetrical Yg orbital gives rise to bonding the antisymmetrical Yu to antibonding. These two particular orbitals are usually denoted als and a*ls (sometimes lsa and lsa*) a star denoting the antibonding character of the orbital. We are now in a position to apply the '' aufbau " principle to molecules just as we have seen it applied to atoms.Thus the ground state of H would be and that of He,+ would be ( ~ l s ) ~ ( a * l s ) . Similarly the lowest state of He would be ( ~ l s ) ~ ( a * l s ) ~ but here the antibonding power of (a*ls) cancels the bondiqg power of with the familiar result that two unexcited helium atoms do not combine together. It may be regarded as a general rule that a combination of two unstarred and two starred M.O. compounded out of the same atomic orbitals will give no appreciable net binding. This type of argument is easily extended. If we use two s-type atomiG orbitals we form a as and CT*S M.O. with patterns substantially the same as in the als and a*ls cases of Fig. 5 (a). If we combine together two p z A.O.we get a somewhat similar pair of M.O. As these have the same symmetry about the axis of the molecule we call them ap and a*p. However a new feature comes in if we combine a pair of p A.O. The general shape of the p A.O. in Fig. 1 makes it easy to see that the corresponding M.O. shown in Fig. 5 ( b ) are of rather different appearance from our previous as and ap M.O. It is still true that out of the two atomic orbitals shown on the left we are able to form two molecular orbitals. The top one on Let us look a little more closely at tlhe nature of (2a) and (2b). 0. Burrau Kgl. Danske V i d . Selskab. 1927 7 1. 2o E. Teller 2. Phyeik 1930 61 468. 154 QUARTERLY REVIEWS the right which is bonding consists of two ribbon or streamer-type regions in which Y has different signs and the original nodal plane (zy plane) remains a nodal plane'for the molecular orbital.It is important to realise that the two parts of this streamer go together; they are one unit and quite inseparable; together they represent the M.O. But there is no longer symmetry around the bond direction ; there is in fact a component of angular momentum around the bond direction amounting to one unit. Such M.O. are called n orbitals. In the case of the upper bonding M.O. f i,) (3 d Atomic o r b i t a l s Molecular orbitals Wave function Symbo FIG. G (a) Porination of u and u* inoleculrir orbit&. ( h ) Formation of x and T* molecular orbitals. of Fig. 5 ( h ) it would be written np (e.g. n22p or 2p7cz or sometimes just pn) ; and the lower antibonding orbital in this same figure is written n*p If we combined together two p y atomic orbitals we should get two molecular orbitals entirely equivalent to those shown in Fig.5 (b) except that they are rotated through 90" around the axis. The superposition of a ny and a xz M.O. is symmetrical around the bond axis. Similar schematic diagrams of several other molecular orbitals as depicted by Weizel are given by G. Herzberg 21 (but care is needed because the naming of the states is somewhat different from the more modern system described above). *l " Molecular Spectra and Molecular Structure " Prentice-Hall 1939 p. 348. COULSON MOLECULAR ORBITALS 155 The “ aufbau ” principle of Section 1 may now be used to describe the states of homonuclear diatomic molecules as soon as we know the relative order of the various M.O.energies. These can be determined 22 from a study of molecular spectra. The lower molecular orbital energies arising from K- and L-shell atomic orbitals are normally in the order shown below ; but the 02p and n2p M.O. have rather similar energies and occasionally interchange places in the table. Note also the degeneracy between 75,2p and n,2p which have of course identical energies. Full notation Abbreviated notation By this stage we have completed the second molecular shell. A somewhat similar scheme holds for the M.O. formed from atomic 3-quantum orbitals. In the second row of the table above we have shown an abbreviated notation introduced by Mulliken 22 the advantages of this notation are that it mamy be made to apply to heteronuclear molecules where any particular M.O.may be compounded from A.O. of different atomic shells [ e . g . H(1s) and C1(3p,) of HCl] and for which the former notation is not adequate and it also enables us to discard the assumption that for example the 02s molecular orbital is entirely composed of atomic 2s orbitals. Thus the notation za means the lowest M.O. of a-type which will in fact normally be compounded almost entirely but not completely of 2s A.O. The most serious disadvantages of the abbreviated notation are that the x y 2 . . introduced in it have no relation whatever to the Cartesian x y x co-ordinates used in the top row of the table and the new notation disguises the simple pictorial relationship between the M.O. and the two A.O. into which as a rule it reverts on separating the two atoms to infinite distance.For that reason we shall generally use the older notation in the top row of the table. With this order of molecular energies it becomes fairly easy to use tlhe “ arifbau ” principle to describe the electronic states of whole molecules. Thns Li has two valence electrons and they occupy the 02s (or za) M.O. allowing us to write Li[ls22s] + Li[ls22s] --+ Li2[KK(a2s)2] The notation KK implies that the two K-shells of the atoms are filled by non-valence electrons. (i) F[ ls22s22p5] + I?[ ls2Zs22p5] + F2[KK(za) 2(yu)2(za) 2( ~ n ) ~ ( v x ) ~ ] Here the ( ~ a ) ~ and practically cancel one another’s bonding being 02s and a*% and so do the ( w x ) ~ and ( V Z ) ~ . This implies that although all 14 electrons take part tho bond is effectively due to the ( m ) 2 pair. As two electrons are primarily responsible it may reasonably be called a single bond.(ii) 01s < o*ls c U28 < a*2s < a2p < n,2p = nz2p < n;2p = 4 2 p < a*2p zu yo xu WJC VIZ Ucl Both notations are given in Fig. 6 for 0,. Other examples are On account of its symmetry it is a a-bond. ~ [ l s 2 8 ~ 2 2 ~ 3 ] + N[ls2;ls22p3] + N2[I<K(z~)2(y~)2(~a)2(wn)”] 2a Rev. Mod. Physics 1932 4 40. 156 QUARTERLY REVIEWS Here the binding is effectively due to a a-bond plus two n-bonds a t right angles t o each other. are involved we may call this a triple bond. (iii) O[ ls22~~2p~l + O[lss2sa2p4] + 02[KK(za)B(ya)~((za)*(wn)*(vn)2] This is a double bond O=O but the interesting point about 0 is that there are two electrons in the (un) or (n*2p) molecular orbital. Now on account of the degeneracy between ndy and n this M.O.will accommodate four electrons ; it is therefore only half-full. So we appeal to Hund's rule (p. 147) indicating that these electrons will go one each in (ny*2p) and and (w1c)*. This amounts to As six electrons Atomic orbitals Molecular orbitals Atomic orbitals Fra. 6 ilgolccular orbitals for 0,. (n,*2p> state of was onc and will have parallel spins. In this way we see that the ground 0 should be a triplet (3.2g) and should show paramagnetism. It of the earliest triumphs of the M.O. theory that it accounted for this somewhat unusual situation so very neatly. Tho combination of A.O. into M.O. may be exhibited in a convenient pictorial way as follows. Let us take 0 + 0 -+ 0 (Fig. 6) as an example. On the left and right of the figure we show the various L-shell orbits of the separate atoms a,nd in the centre are the allowed molecular levels.Lines are drawn connecting the molecular orbitals to the atomic orbitals from which they are compounded. Energy is plotted upwards so that 02s is the lowest and a*2p the highest of the molecular levels. If a particu1a.r orbital is occupied by one electron we put one arrow in the corresponding COULSON MOLECULAR ORBITALS 157 cell parallel arrows denoting parallel spins. In a doubly-occupied orbital the two arrow8 must point in opposite directions on account of the Pauli principle. In an unoccupied cell there are no arrows. Fig. 6 reveals a t once how in the 0 molecule all the levels up to wn are completely filled and it also shows why the parallel spins in the degenerate m level lead to a triplet ground state.Similar diagrams to this can of course be drawn for all molecules. When as is Li, the binding is,due to two electrons in a a-type M.O. we call it a a-bond similarly two electrons in a n-type M.O. constitute a n-bond. Consideration of the examples (i) (ii) and (iii) above enable UB to eay that (a) a single bond is normally a-type ( b ) a double bond is normally a a-bond and a n-bond together (c) a triple bond is normally a a-bond and two n-bonds together. It is of great importance to recognise that a double bond is not merely two single bonds in a state of some strain.23 Our analysis shows that a double bond is obtained from a single bond by superposing two electrons in a quite different type of orbit. Incidentally as is a reasonable deduction from Figs.5 (a) and 5 (b) the overlap of the component A.O. in a n-type M.O. is less than in a a-type M.O. According to the criterion of maximum overlapping this implies that the n-bond is weaker than the a-bond and it gives us a theoretical explanationof the high reactivity of a double bond. For it is easier to disengage the n electrons from each other and link them up with other approaching atoms. H&ronuckar Diatomic Molecules.-There is not much difficulty in generalising these arguments to deal with heteronuclear molecules. Exactly the same principles operate but the criterion of nearly equal energies implies that the two atomic orbitals which combine to form a M.O. will usually be different though they must always have the same symmetry (component of angular momentum) around the bond axis.Thus for the bonding electrons in HCl which we have seen are compounded out of y(H 1s) and y(C1 3pJ the M.O. is by analogy with (1) = y(H 16) f hfJ(c1 3px) . (3) but now the constant A no longer satisfies A2 = 1 and must be calculated mathematically. There are two values of A as in (2a) and (Zb) one of which gives a bonding orbital a i d the other an antibonding orbital. We could if we wished call these 03p and ~$313 remembering all the time that they are combinations of H(1s) and C1(3pz). If 1 > 1 the M.O. has more of the chlorine-atomic orbital in it and may be said to lean on to the one side of the bond. This is exactly what does happen in HC1 for the greater electro-negative character of C1 than of H attracts the valence electrons towards the C1. The corresponding boundary surface is sketched in Fig.7. Accurate calculations of the shape of this M.O. (particularly its nodes) have not been made and so Fig. 7 must be regarded as largely schematic. It is however sufliciently accurate for our purposes and shows how closely the M.O. resembles a 3pz atomic orbital for a hypothetical atom in which the H and the C1 nucleus have coalesced. This is an important idea known Not more than two electrons may occupy any cell. es W. G. Penney Proc. Roy. SOC. 1934 A 144 166. 158 QUARTERLY REVIEWS as the united-atom viewpoint .ll Together with the separated-atom view- point in which we consider what happens to the M.O. as we separate the nuclei from one another it has proved most useful in correlating molecular orbitals and energies with known atomic ones.(A full account is given in ref. 22 pp. 19-73 ; see particularly the diagrams on pp. 40 41 ; or else ref. 21 pp. 350 351.) Because of their shape one sometimes says col- loquially that M.O. such as in Figs. 5 (a) and 7 are “ sausage type ” but the sausage is “ fatter ” at the C1 end than at the H end. This drift of electrons to the C1 atom gives rise to the dipole moment of the HCI bond. This moment depends almost entirely on A and Mulliken z4 (but see Coul- son 25 for a correction in certain cases) has shown how an experimental determination of the dipole moment may be used to estimate A. According to thia argument the full description of HCl in its ground atate is HCWL ( 39 1 a(,323) 2( 3py 1 2( 3232) The orbitals are arranged in ascending order of energy so that K and L denote chlorine K- and L-shells and only the a313 orbitals are genuinely molecular ; the others are atomic in character though presumably the 39 FIU.7 F m t i o n of molecular orbitals for HCl. will partake very slightly of the H(1s) and the 3par and 3p2 of the H(2py) and H(2p,) atomic orbitahes This will imply a small drift of electrons towards the H nucleus and partly cancel the dipole moment of the (a313)2 group. All this is of course simply the wave-mechanical description of the polarisation of non-bonding electrons. Here is to be noticed however one of the strongest virtues of the M.O. viewpoint. The wave function (3) is a particular case of the general L.C.A.O. type of molecular orbital p = 9yA -!- A9yB * (4) The coefficient 1 which measures the polarity of the orbit may have any value ranging from 0 to infinity according to the nature of the combining atoms.Thus the decreasing dipole moment in the series HE’ HCI HBr HI is automatically included right from the start in the corresponding values of 1. Compare this with the alternative valence-bond viewpoint where we are required to introduce it so-called pure-covalent wave function !Pcov,(H-Cl) and a pure-ionic wave function !Pion(H+Cl-) and then to superpose them in the complete wave function. Neither of these two p4 J . Chem. Physics 1935 3 673. 26 Trans. Parachy Soc. 1942 38 433. 16 R. S. Mulliken J . Chem. Physics 1935 3 614. COULSON MOLECULAR ORBITALS 159 pure-wave functions corresponds to anything that can be found experi- mentally so that their introduction is in a sense redundant. With bhese principles as a background it is not difficult to write down the M.O.representation of the electrons in most diatomic molecules (see refs. 21 p. 366 and 22 p. 78) ; and by the same method it is possible to discuss the possible excited states and deduce selection rules very much as for atoms. In many cases these predictions may be verified by observation of ultra-violet spectra band spectra and ionisation potentials. In most cases this qualitative description of the M.O. presents no great difficulty but there are occasions when all available resources of know- ledge must be used even for a qualitative discussion. Such a case is carbon monoxide CO. It is perhaps not surprising that this molecule which has evaded a precise formulation for so long provides a good example of the type of reasoning common to many M.O.arguments. In view of its intrinsic importance and its illustration of the way in which experimental evidence is absorbed into the M.O. description we close this section by an account of the electronic structure of CO similar in many ways to that of A. D. Walsh.27 Since carbon monoxide has the same number of electrons as N2 we might expect by the isoelectronic principle,22 that it would formally be described just as in (ii) above (p. 155); i.e. we should write Let us see in more detail what the individual M.O. look like. It is con- venient to regard the CO molecule as developed from N by taking one positive charge from the first N nucleus and putting it on the second we ask how the M.O. change in this process. It is quite clear that m must represent very largely the O(2s) electrons since these have an energy much lower than any others present; ya yvould then be a M.O.formed from C(2s) and 0(2pz) so that would provide one a-bond. This uses up almost all the allowed a-type orbits round 0 leaving xcr as mostly but not entirely C(2pz) electrons. There still remain the (wn)* orbits; now a study of the spectrum of CO shows that if one of these n electrons is removed by ionisation the binding is slightly increased (as judged by inter- nuclear distance and force constant of COf relative to CO). In this it differs from N2 and suggests that a t least two of the n electrons are prac- tically non-bonding. Call them ng though there is no way of telling if they are nv or na indeed we should expect resonance between the two possibilities.Presumably the two remaining n orbitals are binding being compounded out of C(Zp,) and 0(2pz). These two electrons will convert the single bond into a double bond. It seems likely that the ycr and um M.O. are approximately equally divided between C and 0 so that there is only a small resulting dipole moment in agreement with observation. It will be noticed how much more reasonable this is than the alternative valence-bond treatment which regards CO a~ a hybrid between C=O C+-O- and C-=O+ in approximately equal proportions. It will also be c [ i S w z ~ 2 1 + o[iS22S22p41+ ~ 0 [ ~ ~ ( z o ) 2 ( y a ) a ( s u ) 2 ( w ~ ) 4 ~ 27 Trans. Paraday Soc. in the press. 160 QUARTERLY REVIEWS observed that the precise meaning to be attached here to the words “ double bond ” is rather hazy since a total of 10 electrons contribute in various degrees to the binding between C and 0 ; it is to this fact that the confusion between double bond triple bond and semi-polar double bond has arisen.3. Simple Polyatomic Molecules Localised or Non-lomlised Bonds.-When we pass to polyatomic molecules we find ourselves in an immediate difficulty. We can best illustrate it by means of an example. Consider therefore the molecule of methane CH,. If we attempt to apply the principles of the last section which have proved so effective for diatomic molecules we shall argue that the ten electrons of methane are distributed so that two of them complete the carbon K-shell and the remaining eight occupy molecular orbitals embracing all five nuclei. It is possible to calculate qualitatively 9 28 z 9 and quantitatively 30 the forms of the allowed M.O.but the difficulty lies in this fact-it is well- known that the C-H bond has charactleristic properties such as its length force constant and polarity which are effectively constant with but small changes from molecule to molecule. Since these properties depend upon the detailed distribution of electric charge it is hard to see why the C-H bond should be so reproducible in character and so independent of other substituents around the carbon atom when the M.O. must themselves be quite sensitive to these substituents. The quandary is resolved3I by a mathematical analysis which shows that in the ground state an alternative and equally valid set of M.O. can be written down supposing that each single bond is due to two paired electrons in an orbital almost completely localised between the two nuclei concerned.Following F. H ~ n d ~ ~ we may call these localised molecular orbitals. I n this way the bonds of a polyatomic molecule each separately resemble a bond such as those discussed in Section 2 for :L diatomic molecule. The distinction that we have made between localised and non-localised M.O. is of great importance. If localised M.O. can be used it is better to do so for they are vastly easier to imagine and handle than are non- localised M.O. and in addition they preserve the idea of a bond connecting two of the atoms in a polyatomic molecule. If we do not use them for those simple cases where they can be used we do violence to the long chemical tradition dating from G. N. Lewis’s famous 1916 paper on electron- pair bonds.Whenever possible therefore we shall use localised M.O. though the discussion a t the end of Section 4 shows how if we wish we may translate the whole problem into the language of non-localised M.O. This language is essential whert we come to aromatic and cbnjugated com- pounds as the early part of Section 4 shows ; and it is to all intents and purposes equally essential to any discussion of the excited states (as opposed R. S. Mulliken J . Chem. Physics 1933 1 490. Idem ibid. 1935 3 586. 30 C. A. Coulson Trans. Furaday SOC. 1937 33 388. 31 Idem unpublished. 2. Phpik 1931 73 1 665; 1932 74 1 429. COULSON MOLECXJLAB ORBITALS 161 fo the ground s t a h ) of molecules such as H,O NH, and CH where there is a certain amount of geometrical symmetry. It is possible to give a plausible justification for the use of localised M.O.if we invoke the criterion of maximum overlapping (p. 151). Let us illustrate it in terms of the water molecule H,O. The available atomio orbitals for formation ofaM.0. are the oxygen 2pz 2py and the two hydrogen 1s. Let us place the H atoms as shown on the left in Fig. 8 directly along the directions of the 2pz and 2py orbitals of 0. Then there is strong overlap between H and 0(2p,) and very little overlap between these and any of the others. The theorem referred to earlier shows that we may use H and O(2pJ to form a localised M.O. of the form y(1) = y(H,) + Ay(0 2p,) and similarly H and 0(2pv) form a M.O. y(I1) = y(H,) + Ay(0 2pr). These two M.O. are shown on the right of Fig. 8 and to a first approximation are quite independent of one another.It is now fairly clear that if we replace the hydrogen atom H by some other group e.g. CH, we should change y(I1) but make little alteration in ~(1). In other words the electrons in an 0-H bond have characteristic wave functions; but this general situation would not hold if the angle between the bonds was sub- stantially altered for if we moved H towards H, their atomic orbitals would start overlapping with each other and with the other oxygen orbit [e.g. y(H,) would start overlapping with y ( 0 2pr)]. Fortunately we know from the criterion of maximum overlapping that the binding is strongest when the H6H angle is go" or thereabouts it is precisely in such a con- dition that each 0-H bond has a characteristic individual property. Valence Angles of N and 0.-The example of H,O discussed above shows that if an atom has two p electrons with which to form bonds these bonds will in general be at an angle of about 90".This is the simplest example of the way in which stereochemical rules follow from the criterion of maxi- mum overlapping ; but it is not the only example for the nitrogen atom in its ground state has three unpaired p electrons and these point in mutually perpendicular directions. If we wish we can call them the x L 162 QUARTERLY REVIEWS y z directions. Then there is best overlap and strongest binding in the NH molecule if the three hydrogen atoms are placed along these x y z directions showing that the ammonia molecule is a pyramid whose apical angle is in the region of 90”. The same should be true of phosphorus and arsenic.The molecular orbitals formed by combination of a central p atomic orbital and an attached group such as H(ls) are symmetrical around the direction of the bond. Thus we could describe NH aa and H,O as They are therefore a-type. N(~s)~(~~)~CN(~P) + H(ls) aI6 o( + H( is) 4 4 Valence Angles of Carbon.-It is not surprising that carbon introduces new features. In its ground state the C atom is (ls)2(2s)2(2pz)(2py) so that we might expect valence angles of the same kind as in oxygen but such a scheme is obviously unable to account for the characteristic quadri- valence of a saturated carbon atom. To do this we must have four unpaired electrons ; and the importance of the hybridised orbitals introduced in Section 1 is nowevident. If we mayimagine a carbon atom to be “prepared” so that its four valence electrons are in the tetrahedral orbitals t, t, t, t, then we may mix these with the orbitals of four attached groups to form four localised M.O.Thus in this scheme methane would be described as CH . . . C(~S)~ [ C ( t ) + H(ls) aI8 where [C(t) + H(ls)] indicates that a M.O. is formed by linear combination of a carbon t orbital and the hydrogen 1s orbital chosen so as to ensure the maximum overlapping but we saw in Section 1 that the four orbitals t . . . t were very strongly directed tetrahedrally so the valence angles in methane should have the characteristic angle 109” 28’. Needless to say this conclusion which finds such a neat explanation in the theory is in complete accord with the experimental investigation of infra-red and Raman spectra.A careful theoretical investigation 9 shows that no other type of hybridisation yields such a good binding energy and even in molecules like CH,Cl where the four tetrahedral bonds are not quite equivalent deviations from the tetrahedral angles are probably not very large.33 The notation [C(t) + H(ls)] used above must not be construed to imply that equal amounts of C ( t ) and H(1s) are used in forming the M.O. in question ; it simply means that these are the A.O. from which the M.O. is compounded. With non-polar or slightly polar bonds to be sure the amounts of the two components will be exactly or approximately equal; but as we have seen a characteristic feature of the M.O. description is that it allows for any degree of relative importance in the two component atomic orbitals.This gives a straightforward account of all the saturated carbon com- pounds. For example in ethane C,H, each carbon atom is in the tetra- hedral state and the carbon-carbon bond is formed by two electrons sharing 3 3 W. G. Penney Trans. Farudug SOC. 1935 31 734. We may refer to these as tetrahedral bonds. COULSON MOLECULAR ORBITALS 163 a M.O. compounded from two t-atomic orbitals pointing directly towards one another. Each of the six C-H bonds is similar to a C-H bond in methane just described (for a diagram of this and of a few other molecules see ref. 34). In this way we can see why the C-C and C-H bond energies are more or less constant and we have an explanation of the additivity of bond energies on which the whole basis of the calculation of resonance energies (Section 4) rests.It is true that the situation is not quite so simple as this present account might lead one to belie~e,~5 but this is the essence of the theory. We are now in a position to hpproach the ethylene molecule.23 369 37 If the two carbon atoms are prepared in a trigonal state (cf. Fig. 3) we can form localised C-H bonds by combination of a H(1s) A.O. and one of the trigonal orbitals pointing towards it and a C-C localised M.O. arises from the Combination of two of the trigonal orbitals directly facing each other. The criterion of maximum overlapping makes all the angles 120" in agreement with experiment but so far says nothing about the relative orientation of the planes containing the two CH groups. This orientation is determined in such a way that the remaining Zp A.O. of each carbon atom join to form a n-bond as in Fig.5 ( b ) . If the two p orbitals are t o overlap as much as possible they must point in parallel directions and work must be done to turn them away from this situation when we rotate one CH group around the bond. I n this way the normal double bond >c=< is seen to consist of a a-bond and a n-bond superposed. We also see that the origin of the potential restricting internal rotation lies in the decreased overlap of the p orbitals when one end group is rotated wit of coplanarity with the other. Now the n-bond has a lower binding energy than the a-bond ; and it may therefore be asked whether it would not be wiser to retain the tetra- hedral hybridisation and let the carbon-carbon bond arise from the some- what lcbss efficient overlap of two pairs of tetrahedral orbitals rather as in the pictorial scheino where two adjacent atomic tetrahedra share an edge when forming a double bond and a vertex when forming a single bond in the manner described by Langmuir.W. G. Penney 3* has shown that such a scheme yields a smaller binding energy than that based on trigonal hybridisation a result already anticipated by E . Huckel.39 This conclusion fits better with observations of valence angles since tetrahedral hybridisation would favour HcH angles of 109" 28' considerably different from the angle of 120" predicted for all aromatic carbons (except for possible strain) and verified in many distinct ways. This means that each C-H bond is of a-type the M.O. being a linear combination of H(1s) with a carbon orbital y ( a ) or y(b) pointing directly towards it.3 4 W. C. Price Ann. Reports 1939 36 47. 36 R. Serber J . Chm. Physics 1935 3 81. 36 R. S. Mulliken Physical Rev. 1933 43 279. 37 Idem J . Chem. Physics 1935 3 517. 38 Proc. Roy. Soc. 1934 A 144 166. In acetylene the hybridisation is of the di-gonal type of Fig. 4. 39 2. Physik 1930 60 423. 164 QUARTERLY REVIEWS The C r C bond arises from two di-gonal A.O. pointing a t each other (a a-bond) supplemented by two n-bonds. These latter are formed by com- bination &st of the two 2pu and then of the two Zpz orbitals of the carbon atoms. This makes the whole molecule linear and cylindrically symmetrical around the axis. Hybridisation in Water Ammonia and Other Systems.-Hybridisation occurs to some extent in other atoms though carbon is its supreme example.Let us return for a moment to the water molecule (p. 161). The HOH angle is not exactly equal to the value 90" predicted for pure p binding because (i) the H atoms repel one another-and ordinary long-range repulsion between two separately bonded atoms is here increased on account of the fact that the O-H bonds are polar so that there are positive charges on the two hydrogens-and (ii) a small amount of 8-p hybridisation takes place. Both of these effects which are small conduce to an increase of the central angle which becomes about 103". I n NH3 the angle is increased to 109" 34 for a similar reason. During the " inversion " type of vibration however where the N atom passes right through the plane of the H group the hybridisation changes gradually from almost pure p to the sp2 trigonal orbitals previously dealt with for carbon.I n this way we see that there may be a gradual transition from one type of hybridisation to another. A second example of such a gradual transition occurs in the Walden inversion when the sp3 tetrahedral orbitals change to ap2 trigonal ones a t the half-way stage. Hybridisation occurs in other compounds and particularly in the transition group. For instance Pauling and others gas 41n 42 have shown that sd3 will give four tetrahedral hybrids sp3d2 will give six orbitals of octahedral type ( L e e directed to the six face-centres of a cube) and <?p2d will give four square-type orbitals directed to the vertices of a square and lying in one plane. In this way the octahedral character of FeF6-- and the plane tetragonal character of Ni(CN)4- are nicely explained.Bonds of a-type which involve hybridised A.O. instead of pure s or piire p have sometimes been called q-type orbitals ; 32 ho-wever in this account we shall continue to call them hybridised M.O. Degree of Localisation.-We have assumed that the M.O. used above are all localised and may be represented as L.C.A.O. of the type yA+ AyR. This means that Mulliken's rules 2 4 9 25 mag be used to correlate A with the dipole moment. If the bond in question is highly polar then A will differ considerably from unity and the valence electrons will tend to congregate on either A or B. This will occur for example -in the carbonyl group >-0 of formaldehyde treated by Mulliken and M~Murry.~~ 44 45 A further correlation with experiment is now possible for the increased charge (For further work see refs.9 and 40.) 40 R. S. Mulliken J . Che~n. Physics 1935 3 506. 41 R. Hultgren Physical Rev. 1932 40 891. 4 2 G. E. Kimball J . Chem. PIysics 1940 8 188. O3 R. S. Mulliken ibid. 1935 3 564. 4 4 H. L. McMurry and R. S. Mulliken Proc. Nut. Acnd. Sci. 1940 26 312. 4 5 H. L. McMurry J . Chem. Physics 1941 9 231 241. COULSON MOLECULAR ORBITALS 165 011 the 0 atom will repel the lone-pair 0(2p,) electrons already there and thus reduce their ionisation potentials. There ought therefore to be some simple relationship between the coefficient 1 and the ionisation potential as suggested by Walsh. In the case of formaldehyde the reduction in ionisation potential 34 is of the order of 2 volts. It is very important to know just how reliable is the assumption of complete localisation.The author has made some unpublished calculations for the C-H bond in methane. Y = y(C t l ) + Aw(H,) is not a completely satisfactory description of an electron in the bond between C and Ha ; for we ought to include in the wave function a small amount of y(H,) + y(H,) + y(H,) to allow the electron a small chance of migrating from the region C-Ha to the hydrogens H, H, and H,. Numerical calculation shows that the chance of being found on H is about 1 in 50. We might describe this by saying that there is a small debuzlisution and it is likely that the order of magnitude of the effect is much the same in the other molecules of this section. To the extent that we may neglect this delocalisation our previous descriptions are adequate. These show that the wave function 4.Non-localised Molecular Orbitals Non-Zocalised Bonds.-In our earlier work wo have tacitly assumed that t,he valence electrons can be represented by localised M.O. ix. molecular orbitals chiefly localised between two nuclei. This is almost always true for the ground states of molecules containing not more than one double bond but it ceases to be true for excited states and for systems containing conjugated double bonds. We can illustrate the first of these situations for the particular case of methane. In the ground state (p. 162) there are four M.O. each doubly filled and each localised in the region of one of the C-H bonds. Suppose one of these electrons is excited; how can we say in which bond the excitation is to be found ?-for there is nothing to dis- tinguish one bond from another.We might say in different words that the excitation (or “ exciton ”) is able to move to any of the four bonds. If this is so our molecular orbitals must allow for the possibilityof the excited electron’s being on any of the five nuclei and we are compelled to abandon our previous description in terms of localised M.O. Our molecular orbitals (see later) must now exhibit the complete symmetry of the whole molecule. For we have seen that localisation is possible if when we form our M.O. from various atomic orbitals (including hybridised ones) each A.O. overlaps significantly with only one other A.O. In that case those two A.O. form a M.O. and the bond is localised. In fact it is clear that one great merit in the various s-p hybridisations of carbon is that they provide A.O.from the carbon atom which are so strongly directed that they overlap very little with any A.O. other than those immediately in their direction. How- ever this pairing of A.O. is not by any means always possible. Consider There are however other cases where localisation breaks down. 166 QUARTERLY REVIEWS as an example the M.O. which could be compounded out of four equivalent 2p2 A.O. on four adjacent atoms which we suppose (Fig. 9) to be collinear and equally spaced ; this makes all the 2pz A.O. point in parallel directions a t right angles t o the common nuclear axis. If we call the atoms A B C D and the 2p2 A.O. ya etc. it is perfectly obvious that v b overlaps y a just as much as it overlaps vC. Evidently we cannot form a localised M.O. from va and v b alone we must include yc and for the same reason yd.TES means that the M.O. which are compounded from these four A.O. must be of the form where c1 . . c4 are certa-in constants. There are definite rules for finding these constants,46 47 which we shall not reproduce here ; what is important is that these electrons cannot be localised they are therefore " mobile " electrons 48 or- " unsatxiration '' electrons.49 It turns out that the four A.O. va . . . vd give rise to four M.O. of type ( 5 ) two of which are bonding and two anti-bon&ng. The wave functions have measurable values over the whole system a typical boundary surface is shown very schematically in Fig. 9 ( b ) . The double streamers that we described in Section 2 have now spread out to cover four atoms This delocalising may be expected % q% pc qd to produce two new effects.I n the (a) ( b ) first place if the electrons " swarm " over the complete molecule some- Butadiene T - orbitals. what like the conduction electrons of a metal any electrical influence in one part of the system is easily propagated to any other so that we have a basis for discussing directional substitution in conjugated compounds. 50 5l In the second place the fact that the electrons now have greater space in which to move implies that in general they have lower total energy i.e. greater binding energy than when paired in localised bonds. This increase of binding energy could very properly be called the " delocalisation energy " for that completely describes its origin. But on account of the fact that the M.O. theory grew up rather later than the valence-bond theory it is more usual to call it the " resonance energy ".This energy is calculated by first determining the total energy of the mobile electrons in their com- pletely delocalised orbits and then subtracting from this the energy that would have been expected if the electrons had been paired in the most sensible way possible to give localised bonds and we were able to use the principle of the additivity of bond energies. If as a result of delocalisation y = ClWa f c2Wb + c3Wc + C4Vd - ( 5 ) A ~ ~ ( f j l f i j - - - - -jr- c73 A T L ' i t - - D instead of two. FIU. 9 4 6 E. Huckel Iut. Conf. Phys. London Physical SOC. 1934 p. 9. 47 J . E. Lennard-Jones and (1. A. Coulson Trans. Fnraday SOC. 1939 35 811. 48 J. E. Lennartl-Jones Proc. Roy. SOC. 1937 A .158 280. 4D R. S. Mulliken C. A. Rieko and W. G. Brown J . Amer. Chem. Soc. 1941 63 41. 6o G. W. Wheland and L. Pauling ibid. 1935 57 2086. 5 1 G. W. Wheland ibid. 1942 64 900; J . Chem. Physics 1934 2 474. COULSON MOLECU1,AR ORBITALS 167 the bond lengths in the actual moleciile differ substantially from those associated with the pure single or double bonds t o which the additivity rule applies we oughtl to make a further allowance for " compression energy " due to changes in bond length ; but such a correction is seldom 1nade.~89 49 show that t8he resonance energy is considrrablj- greater for cyclic systems than for open-chain ones in complete agreement with experiment (see ref. 2 Chap. 4). In our discussion of the non-localised M.O. for four atoms A . . . D we made the assumption that thc nuclei were collinear ; but this is obviously it more stringent condition than is necessary for in order to get non-localised 11.0.and the corresponding resonance energy all that is needed is that y b should overlap both y ~ ( ~ ;i~id ylC and similarly for yc with yb and yd. This niay be achieved by making all the atoms coplanar and not necessarily collinear with the ;3-dircction perpendicular t o the plane. It is precisely this factor which wuses the cwplanarit ,v in larger conjugated systems such as benzene and naphthalene. 111 the c t w of four atonis we could make the angles and Bc'D cach c y i i a l t o 120" and we should then be discussing the hutidiene inolec.ulc H,C=CH-CH=CH,. In fact a complete description of b11tittlientb woultl he that eacli of the four carbon atoms was prepared in the t r i p i i ; i I state with it8 valmve iingles all eqiiul to 120" and two electrons wcli I\ ere allottcd to localised bonds (I-H and C-C.This leavesfour electrons t h e 2pz chlcctrons of the carbon ittollis from which we compound the non- locdiwcl 31.0. in the manner previously described. Two of these lattor- the two most strongly bonding ones-are each doubly filled giving a dielocalisation or resonance energy of between 5 and 8 k.-cals./mole. Benzene.-It is worth while discussing the benzene molecule a little iiiore fully siiice this is the classic example of non-localised bonds. A variety of evidence e g l 539 5 4 tells us that all twelve atoms six carbon and six hydrogen lie in a plane and that the carbon skeleton is a regular hexagon. This implies that the carbon atoms are in the trigonal state where the bonds are directed in the plane of the molecule at angles of 120".Herein as W. G. Penney has s ~ o w - ~ ~ ~ lies the peculiar stability of benzene conipared with other cyclic molecules C,H, for if the carbon atom is t o provide three bonds in a plane there must necessarily be sp2 hybridisation and the strongest binding in the resulting o-bonds occurs when the angles are each 120' a situation that can only arise when n = 6. The trigonal >4.O. and the hydrogen 1s atomic orbitals pair up to form localised bonds as indicated in Fig. 10 ( a ) (after A. L. Sklar 56) and each pairing of two overlapping A.O. provides a M.O. that will accommodate two electrons. Simple counting shows that this leitves us with six unused electrons namely the 2pz electrons from each of the six carbon atoms.These are shown somewhat diagrammatically in Fig. 10 ( b ) where for pictorial convenience Explicit calculations 46 51 5 2 .T. K. Syrkin a i ~ l M. T". Diatkitia RitlT. .4cnd. ,Yci. iJ.R.S.S. CI. Sci. C'hint. 19461 63 C. K. Ingold et al. J . 1936 912. 6 5 Proc. Roy. SOC. 1934 A 146 223. 5 6 J . Chem. Physics 1937 5 669 ; M. G. Mayer and A. L. Sklar ibicE. 1938,6 646. 1 5 3 . 5 4 C. K. Ingold et al. J . 1946 222. 168 QUARTBBLY REVIEW8 the “ dumb-bell ” orbitah have been drawn quite separate from each other they really overlap more than the diagram shows. Indeed thb overlapping leads to completely non-locrtlised M.O. in which the double streamers formerly drawn for ethylene [Fig. 5 ( b ) ] and butdiene [Fig.9 (b)] stretch right round the ring one streamer above and the other below the phne of the nuclei as in Fig. 10 (c). Out of the six 2pl A.O. we can compound six distinct molecular orbitala of the form where w1 is the A.O. at nucleus 1 etc. Three of these M.O. are bonding and three are anti-b0nding.~7 68 As we have exaotly six mobile electrons Y = ~ 1 ~ 1 + CaYa + - * + ~ 6 ~ 6 FIG. .-. .. . .-... ( t I ..-. ( C ) 10 (a) (I Orbitals; ( b ) n atomic orbitals; (c) n molecular orbitals. to allot to tho molecular lovels they will complotely occupy the bonding orbits and make their contribution to the energy of binding. The delocalisa- tion or resonance energy is large and amounts to rather less than 40 k. -cals./mole. It is obvious that all this throws new light on the OM idea of Kekul6 in which the benzene ring was regarded as being in a state of dynamic oscillation between structures containing alternate single and double bonds.ti7 E. Huckel 2. Physik 1931 70 240; 2. Elektrochem. 1937 43 762 827. 68 C. A. Coulson Proc. Roy. SOC. Edin. 1941 A 61 115. COULSON MOLEUU'LAB OEBITALS 169 There is nothing in the M.O. account that remobly reeembles the separate Kekul6 structures indeed every bond is always entirely equivalent to every other bond and explicit calculation shows that the electrone are uniformly distributed on all the nuclei and bonds. We have in fact neither single nor double bonds but a state in between the two (mesomeric state). This accomt of benzene may be tmtd in three wap. Fir& all the bonds are found to be equal in length by X-ray anal- and the length agrees with that calculated theoretically.69 Seoond electronic tramitions among the mobile levels called by Mulliken N --+ V tramitions give rise to characteristic absorption which may be calculated and for which quite good agreement with experiment is found.*O# Third there are anomalous magnetic properties for the mobile electrons may be regarded as tiny currents flowing round the ring.Normally they flow equally in both directions but under the influence of a magnetic field there is a preference for one direction rather than the other giving rise to a large diamagnetism when the magnetic field is perpendic~r to the plane of the molecule. This is precisely what is found experimentally. (For further work on this subject see refs. 62-64.) Other Molecules especiaUy CO,.-The calculations for benzene are typictll of many others and they explain the characteristic properties of naphthalene diphenyl and many other similar condensed and aromatic systems (see e.g.refs. 46-48 51 52 59). They may also be used for systems such as pyridine containing a hetero-atom.6'1 66 Here also the mobile electrons flow round the ring but the greater electronegativity of the N atom is shown by a greater concentration of charge there than on the other atoms of the hexagon. They may also be applied to CO, and here because of the intrinsic interest of this molecule it is worth giving the description rather more fully. We follow fairly closely but not exactly the account given by R. S. Mu1liken.M Let us call the molecule which is known to bb linear A-C-B where A and B are oxygen atoms and imagine ourselves to be progressively filling the allowed electronic levels.First we complete the three R-shells and then the (29)2 groups on A and B. The carbon atom-C is prepared in the di-gonal form of hybridiaation (Fig. 4) and overlap between one of these orbitals and the 2pz orbital of the appropriate oxygen atom gives two almost localised a-type bonds which we might write aA2p and aB2p These use four electrons and leave us eight. These eight fit into four M.O. two of which are composed of atomic 2pt orbitals and the other two entirely equivalently of atomic 2pz orbitals. These four M.O. are non-localised 6B J. E. Lennard-Jones and J. Turkevich Proc. Roy. SOC. 1937 A 158 297. R. S. Mullikon and C. A. Rieke Rep. Prog. Physics 1941 8 231. A. Maccoll this vol.p. 16. 62 F. London J . Physique 1937 8 397. 63 H. Brooks J . Chem. Physice 1940 8 939. 6 5 H. C. Longuet-Higgbis and C. A. Coulmn Tram. Furaday Soc. 1947 48 87 ; 56 J . Chern. Physic8 1936 3 720. Idem ibid. 1941 9 463. M. J. Dewar ibid. 1946 42 764. 170 QUARTERLY REVIEWS very much as in the case of butadiene (Fig. S) except that we have three nuclei instead of four the more bonding (written 7t + 7t + n) is a double- streamer extending over ad1 three nuclei and probably fairly equally divided among them the less bonding (written 7t - n) is associated solely with the oxygen atoms A and B and is very nearly without bonding power at all. Thus the molecule is cylindrically symmetrical around tlhe axis and apart from its K-shell electrong is (OA 28)2(0B 28)'(aA2~)'(clI32~)'(Z~ Zy + 712 f 7721- ,l~)~(.Zy - Zy)'((Xt - &) Group Theory Non-localised M.O.in General.-We conclude this section with an explanation of the way in which the symmetry properties of a molecule determine the possible types of 31.0. when we remove the restriction to being localised betwecn two adjacent nucltii. We have already seen that the excited states of molecules compel us to think in terms of M.O. covoring the whole nuclear framework. Tndeed it is in the study of excited states transition moments (i.e. intensities) and selection rules that the work we are about to describe is particularly valuable. It will be simplest to describe it in terms of an esnmplc. Consider therefore the allowed 31.0. for tctlrahedrnl nicthmc. CII,. The M.O. for the valence electrons of nietjh;tne must be compounded from carbon 2s 2p2 Zp,, 2p2 (this autoinntically inclndes hybrids of these if they arc necessary) anti tho four hydrogen I n A.O.Let 11s t)hink first of tho hydrogen orbitals. If one of thest. (call it y,) forms part of a M.O. then by symmetry all the others must be involvod equally. Thus one possible combincition is Y ' t r 1- Y b -1 Y'c -k '+'d * - ( 6 ) since this obviously gives equal weight to all hydrogen atoms. This is a particularly symnietrical combination or grouping of the H group. We may say that it possesscs the coniplthte symnietry of the molecule since any interchange of nuclei leaves it unaltered but it is not the only possible grouping ; for cxample ?/'(l ~ 7ptJ - l/'c - Y'd . * ( 7 ) is acceptable since the weights (or probabilities) of the component hydrogen orbitals am in the ratio l 2 l 2 (- 1)2 (- 1 ) 2 ; that is they are all equal.It is not difficult t o see that all permitted groupings of this kind are included in tho expressions y~~ y b & yC & ytl. Now only four of these c i ~ n possibly bo independent for w x only startcd \!.it 11 four atomic orbitals yn . . . yd. hlullikcn 369 67 has shown that thf. iiiost sensible combinations to choose arc (6) and tho three of type (7) u-herc there are two positive and two ncbgative signs. The latter have less fiindaincwtal symnietry than (6) but they are evidently cntirely equivalent among themselves and may he shown to bo approxiinatdy of the sniiic symmetry type as the three co- ordinates y y z . For vw~iipk as Fig. I 2 shows tlw wave function (7) changes sign wlwn w c h i ~ ~ f l e c ~ t it i t ) t 1~ y; p l m ~ and so does t h e co-ordinate .7*.A convenient n a n i c b for thcw cwnibinat ions of the H atomic orbitals is composite orbitals or group orbitals. I t is important to realise tthatt these 13' R. S. Midliken J . Chem. Physics 1933 1 492. COULSON MOLECULAR ORBITALS 171 particular groupings are forced upon us by the symmetry of the molecule. In NH, for example where there are only three hydrogens the groupings are not the same as in (6) and (7) ; but they are the same in CCI, where ya etc. now refer to the chlorine 3p A.O. directed towards the central carbon atom. Mulliken 36 has shown how the mathematical apparatus of group theory may be employed to sort out the permitted composite orbitals and arrange them in so-called " symmetry classes ".These symmetry classes include at the same time both the composite orbitals of the attached atoms and also t'he A.O. of the central atom (ref. 3 p. 219). This analysis is somewhat complicated and we shall not develop it here suitable accounts of the theory which is useful in studying vibration frequencies as well as electronic levels are found in refs. 68-72. The importance of the sub-division into symmetry classes cannot be exaggerated. It enables us to introduce a seventh fundamental principle to be added to the six enunciated in Sect'ion 2 (a)-(f ) ; for it may be shown that (9) M.O. can only be formed by linear combination of atomic and group orbitals of the same class a'nd allowed transitions are only from one class to another class. It will be recalled that when we were discussing homonuclear diatomic molecules such as H-H (p.152) we introduced the g and u classification and we stated that allowed transitions w e r e necessarily g -+ u or u -+ 9. This is of course merely an example (actually the simplest possible molecular example) of the way in which the geo- metrical shape of the molecule allowed M.O. FIG. 11 Methane. determines symmetry classes for the In the case of CH, the group orbital (6) and the carbon orbital C(2s) are the only ones in their particular class so that they alone combine to form a M.O. Linear combinations of this kind which have very great symmetry are called a-molecular orbitals. They have the analytical represent ation [a] = Ac(2s) f P(Ya -k Yb -t- Vc + Y J d ) 68 H. Margenau and G. M. Murphy " M&hemat,ics of Physics and Chemistry " ~39 J.Rosenthal and G. M. Murphy Rev. Mod. Physics 1936 8 317. 70 H. Eyring J. Walter and G. E. Kimball " Quantum Chemistry " Wiley 1944. 7 1 A. G. Meister F. F. Cleveland and M. J. Murray Amer. J . Phyeice 1943 11 239. 73 A. G. Meister and F. F. Cleveland ibid. 1946 14 13. Van Nostrand 1943 Chap. 15. 172 QUARTERLY REVIEWS where A and p are certain constants. Similarly the type (7) group orbitals of H combine only with C(2p2) etc. and a glance at Fig. 11 will show that these combinations (called t-type) must be ;I' and p' are two new constants and [tJ and [t,] are just like [t,] apart from the change in direction. Values of the constants A p A' p' have been calculated by Couls~n.~o The ground state of the molecule omitting the carbon K-shell would be simply [~]~[t,]~[tJ~[tJ2 and there would be excited states where for example one of the [a] electrons moved to another more excited M.O.of type [t]. A similar analysis could be used for any other symmetrical molecule the greater the symmetry the greater is the value of this group theory analysis in sorting out the allowed types of M.O. and the allowed tramitions between them. It will be recognised that division into symmetry classes is the molecular parallel to the division of atomic orbitals into 8 p d . . . types. Indeed the likeness is sometimes very close as in [a] and [t,] above which are very similar to atomic s and pz states. But there is one important comment to be made-group theory tells us what types of M.O. may occur but it does not tell us the coefficients (e.g.A p above) that occur in them or the relative order of the energies of various M.O. of different symmetry t'ypes. It is possible however to estimate these by various empirical 29 so that reliable descriptions of the ground state and the allowed transitions may be obtained qualitatively a quantitative study is usually quite prohibitively complicated as some recent work on diphenyl shows.73 [tz] = A'c(2pz) + I.'(ya -k YYb - Yc - y d ) 5. General Topics Hyperconjugation.-When we were discussing the M.O. for methane (p. 165) we showed that it was not quite correct to describe a particular bond (e.g. C-Ha in Fig. 11) by two electrons having a localised M.O. !P = Y(C tl) + &P(&) * - (8) Thus (8) was replaced by = Y(C tl) + Mzd + I.[Y(Hb) + Y(&) + Y(Ha)l (9) in which ,u was a constant considerably smaller than A.At the time we interpreted (9) to mean that electrons in the bond C-Ha had a small chance of about 1 in 50 of being found on one of the other hydrogens such as Hb. Our introduction of group orbitals (p. 170) however shows that another interpretation could be given as follows If it were not for the H a a d group the bond C-H would indeed be localised with a wave function (8) ; but the other three hydrogens are able to attract the electrons from the region C-Ha behaving for this purpose not primarily as individual atoms but as a group. That is because the last terms of (9) are a group combina- tion. In fact we could say that the combination [y(H,) + y(H,) + ly(Hd)] was of the right symmetry to interact with y(H,) and y(C tl). It is obvious but a small degree of delocalisation was necessary.73A. London J . Chern. Physics 1946 13 396. COULSON MOLECULAR ORBITALS 173 from what we have said in earlier paragraphs that the last two atomic orbitals are symmetrical around the bond direcfion C-Ha; but a glance at Fig. 12 (a) shows that this is also approximately true for the group orbital [y(H,) + y(H,) + y(Hd)]. This figure shows contours of constant y for this group orbital the plane of the paper being the plane through the three nuclei b c d shown in ,the diagram. The direction of the C-Ha bond is perpendicular to this plane through the point marked with a cross. It is clear that there is approximate symmetry around the bond direction. [More precisely we should say that all three parts of (9) were unchanged by a rotation of & 2n/3 around the axis of symmetry ; they are therefore in the same symmetry class for a rotation described as C,.] Thus all three parts of (9) are needed in the M.O.for the bond C-Ha. Fortunately there is relatively little overlap between the composite orbital for H,HPd and y(C tJ for the tetrahedral orbital points directly away from the carbon towards Ha. This type of interaction has been called a-hyperconjugation. In this way we may infer yet another principle-the eighth and last one-to guide us in the formation of molecular orbitals; this principle shows the effect of a group of atoms or a radical upon an adjacent bond. (h) When a group of atoms adjacent to a bond but not actually part of it is able to provide a composite M.O. of the right symmetry to combine with the electrons of the localised bond a measure of delocalisation takes place.This delocalisation like all delocalisations is associated with a lowering of the total energy i.e. increased binding. It is difficult to estimate the magnitude of this delocalisation energy; but there is evidence that in diamond where every C atom is surrounded by four other C atoms occupying the positions taken up by the H in methane the delocalisation energy amounts to about 5 k.-cals./mole out of a total binding energy of 124 k.-cals./mole. This if correct is a reasonably important fraction of the whole. Its importance however lies more in this conclusion-the fundamental properties of any one bond as e.g. C-H or C-CI are not entirely independent of the adjacent bonds but influences are propagated from one bond to another.In recent years there has been accumulating a large variety of experimental evidence for these small but significant changes. Now we can understand them a little more clearly from the theoretical side. There is incidentally another way 'in which the H,H& group that we have described above can function. In Fig. 12(a) we showed the cr-type behaviour of this group leading to interaction between it and C-Ha; but it may also function with a group wave function [ W ( H b ) - HY(Hc) + W(&))I ' (10) Contours of this function are shown in Fig. 12 (b) which should be compared with Fig. 12 (a). Evidently there is now a line of nodes running acros~ from left to right separating regions of positive and negative y. This combination is clearly of the same general symmetry (n-type) as the dumb- bell orbital of an isolated carbon atom shown on the right [Fig.12 (c)]. Thus the delocalisation is not large. 174 QUARTERLY REVIEWS In this way the HbHcH group can behave as a pseudo-atom with a n-orbital perpendicular to the C-Ha bond. Such a composite orbital could enter into combination with other orbitals on adjacent atoms to form M.O. It is true that in methane such interaction will not take place for the simple reason that there is nothing in the C-Ha bond of the proper symmetry; but if we have a -CH group attached to a benzene nucleus as in toluene C,H,-CH, the double-streamer orbitals of the benzene ring are of the required symmetry type n and there will be interaction between them and the group orbital (lo) leading to M.O. in which electrons from the methyl group can migrate into and out of the ring system.It is the result- ing delocalisation of the electrons that provides the dipole moment of toluene and contributes and additional stability of about 1.5 k.-cals./mole 49 found in branched-chain paraffins as well. It is also the cause of the so-called “ alkylation red-shift ” in the absorption spectra of many dye molecules but as the relation between colour and chemical constitution is dealt with elsewhere in this volumepl we shall not discuss this aspect further. In a formal sort of way we could write a bond structure for toluene such as /“=% C H,_C-C This explains why the American workers refer to this phenomenon as n-hyperconjugation.49~ 74 In the case of toluene we may expect the C-CH bond to be somewhat strengthened by becoming an “acceptor” bond drawing its new strength from the “ donor ” bonds on either side.There are experimental grounds for believing that its length-and particularly t’hat in the somewhat similar dimethyldiacetylene 754s reduced below the value appropriate to a normal single C-C bond. Hyperconjugation of this 7 4 R. S. Mulliken Re?. Mod. Physics 1942 14 265. 7 b G. W. Wheland “ Theory of Resonance ” Wiley 1944 p. 286. COULSON MOLECULAR ORBITALS 175 kind between a methyl group and an aromatic system may possibly be significant in the action of carcinogenic substances (for full references see ref. 76). The remainder of this report is devoted to a series of very brief intro- ductions to other work involving molecular orbitals ; it is intended to help those who wish to follow any particular point in more detail by indicating where such work may be found.Partial Bond Order.-As a result of delocalisation in the electrons of aromatic and conjugated systems the bonds are neither pure single nor pure double but must be described in terms of a fractional bond order. Each one of the mobile electrons contributes a partial bond order to each of the bonds the total order of any bond being the sum of all such con- tributions (e.g. four in butadiene and six in benzene). Coulson 77 has shown how these bond orders may be calculated from the molecular orbitals and finds values of 18 for all the bonds in benzene and 1.894 and 1.447 for the so-called double and single bonds respectively in butadiene =-=. Other bond orders are in refs. 47 49 65 78. The idea of fractional bond order had earlier been introduced under the title " percentage double bond character " by L.Pauling and his c ~ l l a b o r a t o r s ~ ~ ~ ~ ~ and by W. G. Penney and others.S2 83 If we know the fractional bond order of a bond we can calculate its length as was first shown by J. J. Fox and A. E. Martin ; 84 these workers drew a smooth curve showing how the C-C bond length and bond energy vaned with bond order. In this manner the theory may be tested experimentally. 47 85 An alternative calculation of bond lengths by direct methods and without reference to bond order has been given by J. E. Lennard- J0nes.~8 Substantially similar results are obtained in all cases. The partial bond order is also important when discussing vibrational fre- quencies in ground and excited states since evidently there must be a smooth relation between bond order and force constant (see e.g.ref. 58 and unpublished work of Longuet-Higgins and the writer). Partial bond order also enables us to discuss the time-honoured question of bond fixation though on account of the fact that bonds are no longer pure single or pure double (in conjugated compounds) we can no longer think of complete fixation ; but if a particular bond order is high we may say that a double bond is more nearly " fixed " in this position than if the order is low. In this way bond fixation in naphthalene,W in q~inones,~' 88 and in nitrogen compounds such its pyrrole and carbazolegS has been investigated with 76 R. Daudel; Rev. Sci. 1946 37. 77 Proc. Roy. SOC. 1939 A 169 413. H. C. Longuet-Higgins and C.A. Coulson Trans. Faraday SOC. 1946 42 756. i 8 L. Pauling L. 0. Brockway and J. Y. Beach J . Amer. Chem. SOC. 1935,57 2705. L. Pauling H. D. Springall and K. J. Palmer ibid. 1939 61 927. 81 Proc. Roy. SOC. 1937 A 158 306. B 2 W. G. Penney and G. J. Kyndi ibid. 1938 A 164 409. 83 C. V. Jonsson Arkiv Kemi Min. Geol. 1942 15 A No. 14. 86 Idem J . Chem. Physics 1939 7 1069. 87 Idem Trans. Faruduy Soc. l N t i 42 106. J . 1938 2106. 85 C. A. Coulson Nature 1944 154 797. 88 31. G. Evans ibid. p. 113. 176 QUABTERLY REVIEWS conclusions in substantial agreement with experimental evidence ; and some di&ulties in the interpretation of the Mills-Nixon effect 78 have been elucidated. Charge Distribution and Free V&nce.-There are several calculations of the charge distribution and electronic energies in systems containing atoms other than carbon and hydrogen though certain fundamental points still require t o be cleared up thus there seems no reasonably convincing method for calculating the absolute values of some of the resonance and Coulomb integral8 that appear in the course of the work.These integrals have either to be guessed or alternatively values are chosen t o fit some molecules enabling us to predict properties of other molecules and to correlate different properties of the same molecule. We have already referred to calculations for the carbonyl TOU UP,^^^ 44* 4 5 the quinones,8’* 88 pyridine etc. ; 60s 61. 66 t o this list we ought t o add the discussion of Wurster’s salts,8Q the chloro- henzenes,QO and naphthyls,Ql and a whole series of substituted benzenes.Q2-Q5 H.C. Longuet-Higgins 96 has shown how the charge distribution in hetero- molecules may be used to predict the so-called resonance-dipole moment (ref. 75 ; Section 5.4). Unfortunately although experimental evidence is becoming increasingly available a complete correlation between theory and flxpwimentJ has not yet been achieved. A completely general molecular- orbital treatment of conjugated systems by which it is possible t o study both the bond orders and charge densities and to estimate how these are changed when the environment of any part of the molecule is altered has h e n made by C. A. Coulson and H. C. Longuet-Higgins and is in course of publication. It seems likely that this type of treatment will prove useful i n a study of rtlnctions anti reactive centres along the lines indicated by M G.Evans and E. Warhurst 97 ant1 by G. W. Wheland and L. Paiiling.60 5 1 To m n e extent the reactivity of a particular atom in a molecule rriust be cwnditioned by the number and strength of the bonds that it is already forming. Indced this view is already latent in Wwnds conception of ‘ * rcxicliinl affinity ” and it is implkit in ?‘hielc*’s faniouu “ Theory of Partial VttIvi1c.c. ”. The author (ref. ! f H and later work in publication) has shown how the idea of fractional h n d order may be used to give a quantum- mec*hanicul version of Thiele’s theory and to provide an absolute measure of the partial valence at each of the atoms in a conjugated moleculc. This theory is of particular importance for deciding at what points polymcrisation is likely to take place and it may be used to justify R.Lutz’s principle 99 of conjugate addition. Nurntiricd rcsults obtained for the partial free v&nw at cliffcront centres agrce calosely with those obtained in an alternative a9 M. Goopprt-May’r and K. J . McCallum 12~~1. Mod. f’hysic.v 1942 14 248. u 1 J . A. Kcxtc.ltutr micl G. W. Van Oostwhout J . Chem. Z’hysics 1945 13 448. y 2 A. I>. Sklar R v v . Jlorl. l’hysic.q 19-12 14 232. n5 15. Huckel %. l’hysik 1931 72 310. u7 II’r(m.v. Farachy SOC. 1938 34 614. eLI C. A. Coulson ibitl. 1946 42 205. 9e J . dmer. Chem. SOC. 1919 51 3008. . I . Sliormn~i I L I ~ ( I J . A. Kotclttur Z’hysicn 1939 6 572. Ithmi J . ( ’ l i r ~ i . l’h!j.qics 1942 10 135. 9 4 I h m iDid. 1939 7 985. e6 Unpublished calculations. COULSON MOLECULAR ORBITALS 177 treatment of this problem using the valence-bond method and developed by C.V. J O ~ S S O ~ ~ ~ N. Svartholm,127 and by R. Daudel and his collaborators.128 The energies bond orders electron affinities and free valences of several of these systems have been calculated by G. W. Wl~eland,~~ L. Pauling and G. W. Whelaiid,121 C. A. Coulson,loo W. E. Moffitt and C. A. Coulson,lO1 W. J. C. Orr (quoted by J. L. Bolland and G. Gee,lo2) E. Huckel,4% 57 1039 104 and by Penney and Kynch.82 The relation between the energies of free radicals and certain dissociation products has been discussed by these writers and by E. C. Baughan M. G. Evans and M. Polanyi,lo5 and by G. W. Wheland (ref. 75 Section 7.6). In a similar way one may discuss di-radicals,126 particularly the so-called meri-quinones of which a typical example is Chichibabin's hydrocarbon Systems with an odd number of electrons are usually free radicals.A. J. Namiot M. E. Diatkina and J. K. Syrkin lo6 have shown that at sufficiently low temperatures substances of this kind exist in a singlet diamagnetic state but that as the temperature rises some of them may acquire a paramagnetic condition to be interpreted as a di-radical state. Exact CaZculations.-Practically all the calculations described 80 far are approximate i.e. certain assumptions are made which although physically reasonable are not exact or the values of certain quantities which occur in the analysis are obtained by correlation with experiment. So far as the author knows the only complete straightforward M.O. calculations are for H2+ (a singularly simple case since an exact solution may be obtained if the correct types of co-ordinates are used ; see e.g.refs. 18-20 107-log) H2,14 l5 H 3 9 + l10 HeH+,ll1 LiH+,l12 Li 2 9 + 113 Li2,114 and CH4.30 This list makes no attempt to include a large number of other calculations in which ionic terms or polar terms are added for this while it certainly improves the accuracy of the result destroys the genuine M.O. character of the final wave function. It is very doubtful whether exact calculations of other molecular systems are worth making by this method. Electron Velocities.-Most of the work previously described has been concerned with the space distribution of molecular electrons but in a series loo Proc. Roy. SOC. 1938 A 164 383. lol Trans. Faruday SOC. in the press.lo2 Ibid. 1946 42 244. l o 3 2. Physik 1933 76 628; 83 632. lo4 Trans. Farachy SOC. 1934 30 40. 105 Ibid. 1941 37 377. l o 8 Acta Physicochim. U.R.S.S. 1946 21 2 3 ; Compt. rend. Acad. Sci. U.R.S.S. 107 B. M. Dickinson J . Chem. Physics 1933 1 317. 108 E. A. Hylleraas 2. Physik 1931 71 739. l o 9 G. Steensholt ibid. 1936 100 547 ; Norske V i d . Akad. A v h . 1926 No. 4. l10 C. A. Coulson Proc. Camb. Phil. SOC. 1935 31 244. l 1 1 C. A. Coulson and W. E. Duncanson Proc. Roy. Soc. 1938 A 165 90. 112 J. K. Knipp J . Chem. Physics 1936 4 300. 119 H. M. James ibid. 1935 3 9. 114 C. A. Coulson and W. E. Duncanson PTOC. Roy. SOC. 1943 A 181 378. 1945 48 267. M 178 QUARTERLY REVIEWS of papers 115 C. A. Coulson and W. E. Duncanson have shown how the M.O. method may be used to describe electron velocities ; and they have related the resulting velocity distribution function to the shape of the Compton X-ray line by means of which tlhc theoretical predictions may be checked.The agreement is quite satisfactory. Similarly C. A. Coulson and R. P. Bell 116 have shown that the virial theoremis satisfied by M.O. wave functions if the screening constant is suitably chosen. Improved Calculations for Conjugated Systems.-Improvements in accuracy upon the early work on hyperconjugation 49 have been made by R. S. Mulliken and C. A. Rieke,l17 and have been applied to more complex con- jugated systems and free radicals by G. W. Wheland.ll8 The self-consistence of the method of M.O. has been verified by C. A. Coulson and G. S. Rush- brooke,119 and the author 120 has shown how the energy of such systems may often be evaluated wit'h some facility by the use of complex integration.Group Theory.-Group-theory relations between the M.O. method and the valence-bond method have been considered by J. H. Van Vleck.122 A further comparison of the two methods which shows that there are cases (e.g. cyclobutadiene) when the simple M.O. theory is inadequate has been given by Wh~!land,l~~ and there are other comparisons by H ~ n d ~ ~ ~ and Van Vleck.9 Group theory has been applied to study the relation between diatomic and polyatomic molecules and to indicate some of their magnetic properties by J. E. Lennard-Jones.12* A discussion of the diamagnetism of methane has been given by C. A. Coulson.125 I n conclusion the author wishes to acknowledge permission from the Council of the Royal Society of Edinburgh to reproduce certa,in figures from a paper (ref.58). 116 Idem Proc. Camb. Phil. SOC. 1941 37 55 67 74 397 406; 1943 38 100; 1943 39 180. 116 Trans. Faraduy SOC. 1945 41 141. 117 J . Amer. Chem. SOC. 1941 63 1770. ll@ Proc. Camb. Phil. SOC. 1940 36 193. 120 C. A. Coulson ibid. p. 201. 121 J . Chem. Physics 1935 3 315. lZ3 Proc. Roy. SOC. 1938 A 164 397. 124 Trans. Faraday Xoc. 1934 30 70. l Z 5 Proc. Physical SOC. 1942 54 51. 126 E. Huckel 2. physikal Chem. 1936 B 34 339. 12' Arkiv Kemi Min. Geol. 1942 15 A No. 13. la8 R. Daudel and A. Pullmann Compt. rend. 1945 220 889 ; 221 201 and other 11* Ibid. p. 2025. la8 Ibid. p. 803. papers in this journal and J. Physique in 1945 and 1946.

ISSN:0009-2681    

DOI:10.1039/QR9470100144

出版商:RSC    

年代:1947

数据来源: RSC

摘要:

ASPECTS OF IMMUNOCHEMISTRY By MAURICE STACEY PH.D. D.Sc. F.R.I.C. THE science of immunology is mainly concerned with the action of extraneous high-molecular chemical substances on animal cells and tissues and with the mechanisms of the resistance of the host to such foreign sub- stances. From the practical point of view we are mainly concerned in dealing with the effects on thg human and animal body of agents of disease and with the studies of the tissue immunity so necessary for our survival which it is possible to induce against these infective agents in order to keep the body cells in a healthy condition. Immunity phenomena are one manifestation of the so-called “ detoxi- cation mechanisms” by which the body attempts to deal with toxic sub- stances and it is now firmly established that most defence reactions can be interpreted on a definite chemical basis.Numerous manifestations ‘of immune states of the body are commonplace in everyday life. We have for example the relatively permanent immunity which follows an attack of one of the infectious childhood diseases such as measles grid chicken-pox we have the prophylactic methods of dealing with diseases such as immun- ization against diphtheria aad tetanus and vaccination against smallpox and typhoid. It has long been apparent that an attack of one infectious disease does not confer any immunity towards another disease and it is now realised that this high specificity of the disease process can only be interpreted in terms of chemical reactions. Chemistry is now contributing so much to immunology that we now recognise the branch of medical science which deals with the immune state as “ immunochemistry ”.The object of this review is to outline the main advances in the subject and to indicate the special progress which has arisen from the more purely chemical approach. For those who may be stimulated to inquire further there are numerous excellent summaries of the truly vast literature. For the backgroundthere are the essentially immunological approach and bacterio- logical aspects in the well-known work by W. W. C. Topley and G. C. Wilson as well as in American publications.2 A fascinating contribution is R. Dubos’s recent book? The chemical approach is dealt with in the classic but highly specialised book of K. Landsteiner 4 (whose contributions to immunoohemistry have been so great) with its recent valuable addendum by Linus Pauling ; J.R. Marrack’s lucid monograph gives probably the most concise account of (a) “ The Newer Knowledge of Bacteriology and Immunology ” E. 0. Jordon and J. S. Falk Chicago 1928 ; (b) “ Agents of Disease and Host Resistance ” F. P. Gay and associates Springfield Baltimore. (PROFESSOR O F CHEMISTRY UNIVERSITY O F BIRMINGHAM) l “ An Outline of Immunity ” Arnold London 1946. “ The Bacterial Cell ” Harvard University Press 1945. “ The Specificity of Serological Reactions ” Harvard University Press 1943. “ The Chemistry of Antigens and Antibodies ” Medical Research Council 1938. 179 1 80 QUARTERLY REVIEWS its subject while the books of W. C. Boyd are most useful. Other text-books monographs etc. are listed at the end of Land- steiner's 4 book.Progress in the field has been summarised in a valuable wag by M. Heidelberger's numerous reviews together with those of his brilliant pupils H. P. Treffers @ and E. A. Kabat.lo With Dr. Kabat's permission the writer has quoted liberally in this article from his recent and authoritative summaries. Another bold approach which may appeal to some readers is the book by M. G. Sevag.ll Concise popular abstracts have been given in lectures by C. R. Harington l2 and L. Pa~1ing.l~ These various publications carry valuable references to the literature so that in the following account only relatively few key references will be given. Owing to the prolonged wartime diversion of the activities of our scientists into other channels progress in immunochemistry in Great Britain has suffered a setback the severity of which can be gauged by reference to recent publications listed by Kabat lo and Treffer~,~ nearly all of which are American.In America immunochemical studies have been pursued with great vigour since many of the war problems on infectious diseases were concerned with the preparation of effective immunising agents from tissues and bacterial constituents. The enthusiastic activities of Pauling and his group during the last decade in the immunochemical field have been watched with interest by chemists everywhere. Although some of Pauling's conclusions and methods have occasionally met with sharp criticism l o from some immunochcmists (see later) there is no doubt that his experience and outlook in such a comparatively new field are greatly to be welcomed for the greatest possible efforts and energy in many direcbions are needed to obtain satisfactory explanations in chemical terms of some of the phenomena of specificity.Definitions of Immunologiml Terms.-The ailimal body possesses a certain degree of innate resistance to infectious disease this is a variable type of immunity which is probably hereditary. In general however immunity is an acquired characteristic. Two types " active " immunity and " passive " immunity are well recognised. Activc immunity is gained in several ways ; for example during recovery from an infectious disease or as a result of inoculations by means of suitable constituents of the agent of disease or by the injection of an appropriate artificially pre- pared complex. In the animal body active immunisation is of relatively lasting duration and in a limited number of specific diseases such as small- pox and diphtheria practical application of the knowledge of active immunity has provided us with an ideal method of preventing the disease.and F. M. Burnet " Fundamentals of Immunology " Interscience New York 1943. (a) Ann. Rev. Biochem. 1932 655; 1933 503; 1935 569 ; ( b ) Bact. Rev. 1939 Advances in Protein Chemistry 1944 1 69. l 0 ( a ) J . Immun. 1943 47 513; ( b ) Ann. Rev. Biochem. 1946 505. l 1 " Immunocatalysis " C. Thomas 1945. l e Chent. and Ind. 1944 87. 'I " The Production of Antibodies " Melbourne 1041. 3 45. Chem. Eng. Newt? 1946 24 1064. STACEY ASPECTS OF IMMUNOCHEMISTRY 181 Passive immunity is obtained by the injection of serum from immunised animals and occasionally to a limited extent congenitally through the colostrum.The degree of passive immunity induced depends a good deal upon the amount and quality of the immune serum injected so that it is of a relatively transient type. This immune state however is of value in cams of critical illness and in the past has been invaluable in treatment of pneumonia and more particularly of influenza1 meningitis in the young. Acquired active immunity exists in various grades and its acquisition depends upon the nature of the Mectiveness of the invading organism a,s well as upon the state of the tissues of the host. The degrees of immunity shown by any animal to infectious diseases can be classified as follows (1) Complete lack of immunity-in such cases as in meningitis recovery (2) Low or medium grade immunity-here there are a few recoveries.(3) High grade immunity-here there is a high percentage recovery. (4) Complete immunity-in this case the disease does not gain a hold. In immunotherapy the aim is of course t o induce (3) and (4) in the individual. Most of the reactions concerning immunological specificity are carried out in vitro on the constituents of serum and it is upon these “ serological ” specific reactions that the attention of chemists is now directed in order to gain information upon the wider problems of immunology. The whole sthdy is bound up with questions regarding the structural chemistry of macro-molecules e.g. of the proteins fats nucleic acids and carbohydrates which go to make up cell tissues of all types. Before dealing with the main advances it will be appropriate to define some of the terms used by immunologists t o describe the various manifesta- tions of the immune state.The term “ antigen ” denotes any substance foreign to the blood which when introduced into an animal parenterally (i.e. outside the digestive tract) stimulates in the serum the formation of new “antibody” proteins. The antigen always reacts in a visible way with its homologous antibody and “ antigen-antibody ” reactions take place in a variety of forms. In the past the word antigen has been used in a rather loose kind of way to denote any substance which will react specifically with antibody and for this reason other terms e.g. “ complete antigen ” or “ immunogen ” are sometimes used to describe an antigen possessing the widest immunising properties. In order to be retained in circulation in the animal body for a sufficiently long period to produce antibodies an antigen must be of a high molecular weight (e.g.> 10,OOO for proteins) and it must be injected beyond the epithelial tissues. The majority of antigens are undegraded proteins though it appears likely that other undegraded macro-molecules of biological origin e.g. mucopoly- saccharides mucolipoids etc. can also behave as antigens. Further a8 will be described later certain proteins of low molecular weight can be rendered antigenic by combination with a polysaccharide or by adsorption on a colloid carrier such as collodion kaolin or even charcoal. is rare. 182 QUARTERLY REVIEWS The route of injection of antigens may exert a quantitative influence on the degree of antibody stimulation.Small molecules can occasionally “ sensitise” an animal and thereby give rise to an “ anaphylactic” or “ allergic ” state and this phenomenon can be studied as described later by the same methods as in immunity. When part of an antigen can react in some specific way with homologous antibody but cannot itself stimulate antibody production it is known as a “ hapten ”. The term “ toxin ” from the older literature may now be regarded as being generally synonymous with antigen and “ antitoxin ” as being synonymous with antibody. These terms are used in special cases; e.g. most toxins are poisonous substances produced by micro-organisms and are usually classified as exo-toxins and endo-toxins though in addition there is the important class of plant and animal venoms such as abrin snake and scorpion venoms etc.It is fortunately now possible to produce antibodies (antitoxins) which will neutrglise the effect of most toxins. Exo-toxins e.g. diphtheria toxin are isolated from the metabolism solution in which the organisms are grown and they appear to be mainly proteins. The endo- toxins e.g. from Bad. shigz appear to be contained in the somatic parts of certain cells and are phosphorus-containing mucolipoids. The toxicity of a toxin does not appear to depend upon any particular prosthetic group in the molecule and this toxicity can be diminished e.g. by formaldehyde treatment without loss of antigenic properties. Such products are used commercially and are termed “ toxoids ”. Individual proteins vary in their power to produce antibodies or anti- toxins and there are often wide differences in the response of the animal injected the horse and the rabbit apparently being the best antibody pro- ducers.There is good evidence that the largest and least undegraded molecules provide the most complete antigens as exemplified by whole blood serum and bacterial cells. Gelatine and the protein hormones are non-immunogenic. Antibodies (sometimes called immune bodies) are altered or unusual globulins found in the serum of an injected or a disease-infected animal. The presence of the antibodies confers on the animal a certain degree of immunity towards the infection and the serum is known as an immune or “ anti-serum ” and it can react visibly with the infective agent. There are various methods of detecting antibodies all of which depend upon the reaction between the antigen and the antibody-a reaction which is remarkably specific-and generally the methods of detection of speci- ficity are known as “ serological reactions ”.When the antigen is insoluble in physiological saline e.g. as with bacterial cells blood cells etc. then the addition of a specific immune (or anti-) serum to it in stable suspension will cause a visible coalescence of the particles and this is termed the “ agglutination ” reaction. When the antigen is in solution or colloidal solution e.g. a soluble protein then addition of an appropriate amount of antiserum results in the formation of a flocculent precipitate and the reaction is known as the “ precipitin ” reaction. It has been found by experiment that in all antigen-antibody systems there are definite pre- STACEY ASPECTS OF lMMUNOCHEMISTRY 183 cautions which must be taken in order to demonstrate a true reaction; some aspects of the immune reaction are described later.A haptene can frequently combine with part or all of an antibody and if added initially to the antibody solution can inhibit the normal pmcipitin or agglutinin reaction between an antigen and its “ homologous ” antibody. The combination is mainly due to the presence in the haptene of certain structures including polar grpups which are known as “ determinant ” groups. The property of blocking the reactivity of some structures in the antibody has been adapted for specificity detection and is known as the “ specific inhibition ” reaction ; this is of particular value when used in blood grouping. Some serological reactions are extraordinarily sensitive and c&n be carried out with substances in solution diluted to 1 in 20,000,000 for frequently it needs but a small amount of antigen to precipitate relatively large amounts of antibody.Frequently the antigen-antibody complex can remove from serum one or more of a group of normal serum components termed “ haemolytic com- plement ” (earlier known as “ alexin ”). This was the term applied to the thermolabile factor which was necessary for the lsftic action on cells sensitised with antibody. Anaphylaxis. The anaphylaxis reaction depends on the fact that an animal can be sensitised with an initial injection (a sensitising dose) of an antigen and shocked (often fatally) by a second injection (a shocking dose) of the same antigen. Certain skin reactions e.g.Tuberculin Schick and certain allergic states such as hay fever and asthma may be manifestations of anaphylaxis and are discussed later. Serological reactions may be either “ homologous ” i.e. when an antigen reacts with the antibody it has engendered or “ heterologous ” when it reacts with an antibody produced by a different antigen. In general antigens having closely related chemical structures may react with each other’s antisera and when an antigen does react with heterologous antieera it is said to give a “ cross-reaction ”. The cross-reaction may depend on the presence in the antigen of common or closely related determinant chemical groups. Frequently specific parts of the antiserum can be removed by an ’‘ absorption ” method. As an example hen-egg albumin cross reacts strongly with antiserum to duck-egg albumin.If however the antibodies which react with the hen-egg albumin are precipitated (or absorbed) by addition of-a slight excess of this protein to the antiserum it can be shown that the supernatant fluid will still precipitate with the homologous duck- egg albumin. This specific absorption method is of great value in dealing with the components of antiserum produced by the injection of whole bacterial cells. The Chemical Basis of Specijicity The major problems still engaging the attention of immunochemists concern the explanation of the unique specificity of every antigen in terms of chemical structural differences. One aim is of course to correlate immuno- 184 QUARTERLY REVIEWS logical differences with protein structure. A good deal of headway in this direction has already been made especially by Landsteiner and his schooL4 They have shown particularly by adopting one main procedure-namely that of taking a known protein modifying it by chemical means in some clear-cut manner and then using it as a new antigen-that it is possible to produce antisera which react in a homologous manner with the chemically altered antigen and cross-react with the heterologous original protein and with proteins containing known related groupings.More precise knowledge on specificity is now available from studies of the haptene properties of carbohydrates. The great dependence of specificity on chemical structure has been proved in several ways e.g. (a) chemically different proteins can always be differentiated by a sero- logical reaction ; ( b ) conjugated carbohydrates known t o contain related structures give typical and frequently predictable cross-reactions and ,other simple chemical structures when tested as haptenes react similarly ; ( c ) chemical alteration of antigens changes specificity in a manner which to some extent can be controlled.Functionally and structurally related corresponding proteins of different species cross-react very closely. These facts are illustrated in numerous examples below. Almost any chemical change in a protein alters the nature of its serological reactions so that two main lines of approach to the problem of specificity were possible (1) the alteration of certain parts of a protein structure b y chemical action (2) the coupling of different groups and molecules of known chemical structure i.e." determinant " groups to a common protein. In (2) the methods must of necessity be somewhat drastic and thereby cause some unwanted degradation and denaturation of the molecules but never- theless this line has been very valuable. Under (1) the following methods have been used. Digestion. Breakdown of proteins by means of enzymes generally destroys rapidly the antigenic power with complete loss of specificity. There are some exceptions however for it has been found as mentioned later that antitoxin protein may retain some antibody activity after treatment with pepsin. Denaturation. Denaturation usually by heat does not appear to cause complete loss of antigenic power though this of course depends on the degree of denaturation and antisera to native proteins react t o a much less extent with the same proteins after denaturation.By reversal of denaturation there is some evidence that the original immunological speci- ficity can to some extent be restored though it has been shown that a lowering of the antigenic power of horse and bovine serum albumins denaturated with urea or guanidine was obtained after regenerating them. In general denaturation gives a moderate decrease in species specificity. Oxidation of proteins with potassium permanganate gave a product which was still antigenic but its homologous antiserum would not cross-react with the original protein or with other similarly oxidised Oxidation. STACEY ASPECTS OF IMMUNOCHEMISTRY 185 proteins. This was an example of the retention of species specificity after alteration of the antigenic nature of the protein.Reduction. D. Blumenthal l4 found that reduction of egg albumin with thioglycollic acid did not affect its serological behaviour but that there was some reduction of the reactivity of serum albumin after a similar reduction. A further oxidation did not restore the reactivity of egg albumin and it is likely that -S-S- + -S-H changes occur. Various keratins oxidised with copper-oxygen and then reduced with thioglycollic acid cross-react with antisera to each other. Degrcldcttion. Treatment of proteins with acids and alkalis decreases antigenic activity alkalis generally being the more effective. Acid-treated proteins lose some species specificity but not all their antigenicity. This effect has been used l5 to study the size of serologically active units in silk.Silk was partly hydrolysed and then dialysed for varying periods to give fractions which were further purified. The fractions were used in the inhibition test with the native silk and its homologous antibody reaction. Complete inhibition could be obtained with a chain of 7 amino-acids (mole- cular weight 600) while t'he strongest inhibition produced in greater dilution was obtained with a product having a chain of 12 amino-acid units. Treatment of proteins with formaldehyde (a reagent which may achieve more than the blocking of -NH2 groups) was shown by Landsteiner to give but little effect on species specificity. I n this con- nection it is of interest to note that rabbit serum could be altered and rendered antigenic for the same animal by formolisation.Benzaldehyde treatment of proteins caused little effect while the ninhydrin reagent caused some alteration in cross-reactivity. Treatment of casein with 7% acetic acid and sodium nitrite to remove free amino-groups gave a protein with altered physical properties but there was no apparent alteration of antigenic properties. Esterijication. Methylation gave more profound changes in the mole- cule. Two main methods of esterification were employed by Landsteiner -an acid-alcohol treatment which esterifies carboxyl groups and a diazo- methane treatment which methylates -OH -NH2 and >NH groups. In both methods insoluble proteins were formed and the esterified proteins behaved in a manner generally similar to that of the xanthoproteins and iodoproteins which are mentioned later. The capacity of the esterified proteins to react with immune sera against unchanged protein is mainly lost while their homologous reactions were quite strong.Their homologous sera reacted with other proteins similarly esterified. It is clear that the process produced new groups which were strongly determinant and accounted for the. new cross-specificity. Acetylation. Acetylation by means of acetic anhydride caused a specificity change analogous to that produced by esterification. Land- steiner considered that changes were effected by altering the -NH and -OH groups since some cross-reactions with other proteins containing Mild substitution. Deamination. l4 J. Biol. Chem. 1936 113 433. 1 5 K . Landsteiner J . Ezp. Med. 1942 75 269. 186 QUARTERLY REVIEWS acetyl groups were given. The significance of naturally occurring acetyl groups in pneumococcus polysaccharides is important and is discussed later.Acetylation with keten CH,:CO also caused some serological changes. Graded benzoylation of proteins imparted a new common specificity the degree of which was determined by the number of benzoyl groups preeent and was a maximum wjth 5% of benzoyl groups. It was noted too that cross-reactions were obtained with proteins containing m- and p-nitro- benzoyl groups and also p - bromopropionyl groups. Generally similar results were obtained with the action of benzyl chloroformate C,H,*CH,*O*COCl which acted on free -NH groups. The original species specificity was almost completely destroyed while there were again cross-reactions with other proteins similarly treated. There was a specific inhibition of the reactions by carbobenzyloxy-amino-mids.This probably involves an action on free amino-groups. I. Berenblum and A. Wormall l6 treated proteins with “ mustard gas ” to give “ H ” proteins and with 2 2’-dichlorodiethyl sulphone to give ‘‘HO2” proteins. A new specificity was conferred although the species specificity was retained. This may have been due to unchanged protein. Recently the results of a very comprehensive investigation have beer published by A. Wormall and his collaborators 5’ on numerous aspects of the reactions of “ mustard gas” and related substances with proteins some of the work involving the use of radioactive sulphur. Earlier findings were confirmed i.e. antisera to “ H ” proteins were obtained by the action of “ H ”-treated horse serum.The precipitin reaction between “ H ” rabbit serum proteins and antisera to “ H ” horse serum was only partially inhibited by an “ H ” glycine derivative although this hapten completely inhibited the reaction between “ HO ” proteins and their antisera. This suggested that the action of “ HO ” on proteins involved an action on -NH groups. The absence of serological cross-reactions between “ H ” proteins and “ HO ” proteins together with the results of inhibition tests gave strong indication that “ H ” and “ HO ” differed profoundly in their action on proteins. “ H ” has a strong inactivating action on complement affecting all the components though not a t the same rate. With this reagent at pH 8 there was an attack on free amino-groups or on -SH groups possibly as follows Cross-reactions with untreated protein were considerably reduced and a new specificity was conferred.Inhibition was obtained with lysine or 1 -aminopentane- 1 -carboxylic acid treated with phenyl isocyanate. It has been claimed that it is possible to confer antigenicity on a degraded protein protamine with this reagent. F. Obermayer and E. P. Pick,l* A. Wormall,lD and others treated proteins with nitric acid tetranitromethane etc. to give nitro- or “ Mustard gas ” treatment. Phenyl isocyanate. -CHz.NHz + 0 :C‘ :NPh -+ -CH,*NH.CO*NHPh. Nitration. l6 Biochem J . 1939 33 75. “ I b i d . 1946 40 734-774. la Wien Klin. Wochemchr. 1904 17 265. J . Exp. Med. 1930 51 295. STACEY ASPECTS OF IMMUNOCHEMISTRY 187 xantho-proteins. The effect of the reagent was mainly on the aromatic rings and the yellow products had acquired another new common property which caused them to cross-react serologically.Specific inhibition was obtained with mononitrotyrosine and other similar compounds and it was shown that the reactivity depends on the presence of -NO2 and -OH groups in the aromatic ring and on the presence of free -C02H groups in the molecule. It is noteworthy that gelatine was not rendered antigenic by nitration but it could thereby be made t o act as a specific inhibitor. Halogenation The pioneers in this field were Obermayer and Pick,18 who studied bromination and iodination. Iodinated proteins acquired a new specificity and antisera to them reacted with other iodoproteins. Brominated proteins differ but slightly and cross-react with sera to iodo- proteins. Wormall found 3 5-di-iodo- or -dibromo-tyrosine and to a less extent dichlorotyrosine would specifically inhibit the iodoprotein homo- I logous reaction.Any compound with <->OH groups would also act I as an inhibitor. From the work of A. Kleczkowski 2O on quantitative iodination and serological studies it appears that tyrosine forms an essential part of the determinant groups in native horse serum globulin. Conjugation of Proteins This method gives a means of studying antibodies in minute detail for it is possible to control very precisely the determinant groups of an artificial or “ synthetic ” antigen. In a chemically conjugated protein a new type of specificity can be created which will stimulate the production of anti- bodies capable of reacting with proteins unrelated except for the presence in them of the same new determinant groups.Numerous methods of conjugation are now available none of them can however be claimed to be a mild treatment of the protein. Perhaps the most useful in the past has been the method of coupling the protein with a diazo-compound to form a so-called “ azoprotein ” a valuable example of which was Landsteiner’s “ atoxyl azoprotein ” shown approximately as follows CH,*CHCO *NH-protein residue OH Such coloured products formed by coupling proteins with diazonium compounds when prepared in a suitable manner gave but weak reactions with the immune sera for the unchanged protein but did elicit readily homologous antibodies. The original protein specificity was altered to 20 Brit. J . Exp. Path. 1940 21 98. 188 QUARTERLY REVIEWS some extent the degree of change depending upon the nature of the new prosthetic group in the azoprotein and upon the degree of coupling.The coupling probably took place through tyrosine residues as shown in the atoxyl example. Some of the conclusions drawn from numerous precipitin tests with a wide range of azo-proteins were (1) by coupling proteins gained new specificity and the homologous reactions were always strongest ; (2) newly- introduced arsinic acid groups gave the strongest altered specificity and there was but little crossing with -CO,H and -SO,H groups; (3) methyl halogen methoxyl and nitrogen in a nucleus had a relatively small influence on the specificity as compared with acid groups. Bromine and methyl groups which are approximately equivalent in molecular " bulk " are serologically equivalent ; (4) the relative position of -C02H to -NH has a pronounced influence on specificity ; ( 5 ) the position in the aromatic ring of neutral substituents was of greater effect than the nature of these sub- stituents ; (6) strongly basic groups are as powerful as strongly acid groups in directing specificity ; (7) pairs of cyclic compounds-e.g.benzene and thiophene pyridine and thiazole-which are related in chemical pro- perties were serologically equivalent. The Influence of Side-chain Aliphatic Groups on Speci$city.-Investiga- tions were carried out on coupled homologues of p-aminophenylacetic acid. It was found that the lengthening of a side chain produced a profound dif- ference in reactivity with loss of the sharpness in specificity as the chain length increased.Very important experiments were conducted with coupled synthetic peptides in order to throw light on the specificity of proteins. Compounds were made analogous to p-aminobenzoylglycyl- leucine with variations such as glycyl-glycine leucyl-glycine and leucyl- leucine. It was clearly shown that the specificity was determined by the nature of both the amino-acids but the most determinant factor was the terminal amino-acid. This important effect was confirmed by use of synthetic peptides containing five amino-acids. Alteration by Coupling through the -SH Group of Proteins.-L. Pillemer E. E. Ecker and E. W. Martensen 21 reduced -S-S- groups to -SH and allowed these to react with organic halogen compounds. The method was applicable to proteins which contain l0-15% of cystine.The authors studied carboxy-alkyl and carboxy-aryl derivatives of proteins and found that a new immunological character was conferred. There was little crossing with proteins differently treated and the original specificity was almost completely abolished. Coupling via Azides.-R. F. Clutton C. R. Harington and T. H. Mead 22 considered it unlikely that azo-groups would occur in Nature and they developed a method which could link a determinant group to a protein through a peptide link. They used the azide of c,H,,o,.o~cH,.cH(NH,).co- ~ ~~~~~ 21 J . Exp. Med. 1039 70 387. 2 2 Biochem. J . 1937 31 764. STACEY ASPECTS OF IMMUNOCHEMISTRY 189 and R. F. Clutton C. R. Harington and M. E. Yuill 23 found that such a group completely masked the original specificity and the new Specificity was dependent upon the determinant group.They were able to introduce a tyrosine and a carbohydrate group into gelatine and thereby render it slightly antigenic. In the same way these authors introduced the thyroxyl group into proteins and obtained reactions between the new antisera and thyroglobulin. Such immune sera protected animals against the normal physiological effects of thpoglobulin and thyroxin. C. R. Harington’s 24 approach offers great possibilities in immunochemistry. Effect of Spatial Differences in Haptens .-Landsteiner has obtained most striking results by conversion of &- I-; and meso- tartaric acids into arninotartranilic acids then diazotising these and coupling them to the same protein. The serological cross-reactions showed distinct differences between the &- I- and meso-forms as shown diagrammatically as follows Homologous antiserum.1 Synthetic antigen. 1 . . . . . . . +++ d . . . . . . m e s o . . . . . . 1 I +:+ +..+ (where + + + = very strong precipitin reaction) Thus the stereochemistry of a prosthetic group has an important deter- minant influence. “ Tartaric acid antisera ” had an important reactivity with similarly prepared “ malic acid antisera ” the homologous reactions i.e. d- with d- etc. being the strongest. A specificity due to cis-trans isomerism with antigens from maleic and fumaric acids coupled t o proteins was demonstrated using a specific inhibition reaction. Thus it will be seen that serum tests roughly analogous to enzyme reactions may be used for determination of spatial configuration. It was readily apparent from the knowledge of the high serological activity of the hapten specific polysaccharides that application of the azoprotein method to the study of carbohydrate groups would be of high significance.The advances in this field are mainly due to Avery Goebel and their associates (see reference 5). The monosaccharides D-glucose and D-galactose were converted into p-aminophenyl-p-glycosides and then after diazotiation were coupled to serum globulin. These on injection produced antisera which differentiated the two substances quite sharply showing that a difference on C of the sugar part conferred a serological specificity. cc- and @-Glucosides could be distinguished in the same way but they exhibited it stronger cross-specificity. In this work it was noted that the homologous reaction could be inhibited only by the homologous simple glucoside but the cross-reactions were inhibited by both homologous and heterologous glucosides.2sI€rid. 1938 32 1111 1119. 24 J . 1944 193. 190 QUARTERLY REVIEWS Acetylation of a glucoside e.g. on C, sharply changed its specificity but did not cut out reactivity in the homologous sense. The inhibiting cross-reactions were similar to those of the a- and #l-glucosides. A brilliant piece of work was carried out using azoproteins containing the four disaccharides lactose gentiobiose cellobiose and maltose and the mono- saccharides glucose and galactose. It was not possible to account for the occurrence and degree of all the cross-reactions which were strong but it could readily be shown that the terminal hexose exhibited the most dominant determinant effect.The whole disaccharide molecule determined speci- ficity but even the position of linkage between the hexoses had a distinct effect. The specific inhibition of these reactions by the mono- and di-saccharide glycosidic haptens was studied with care and was most informative. This work was extended by W. F. Goebel and his associates to include the hexuronic acids because it was already known as discussed later that these occurred in the soluble specific substances from pneumococci Fried- H AH H*OH H OH lander’s bacillus etc. and that plant gums containing uronic acid gave reactions with pneumococcal antisera. Azoproteins were made by coupling the p-aminophenylglycosides of D-glucuronic acid (I) and D-galacturonic acid (11) which are related on C as glucose is to galactose.These investigations are important since the pneumococcus Type I specific polysaccharide contains a galacturonic acid constituent while many others contain glucuronic acid. Antigens containing galacturonic acid did indeed precipitate with Type I pneumococcus antisera but also in a rela- tively non-specific manner with Types I11 and VIII antisera. This rela- tively non-specific nature of some cross-reactions was shown by W. F. Goebel and R. D. Hotchkiss 25 to be due to the reactions between the acidic groups of the antigen and the basic groups of the antibody. They showed for example that Type 111 pneumococcal antisera would react with antigens containing unrelated organic acid residues such as p-amino-carboxylic and -sulphonic acids. W. F. Goebel 26 prepared the aldobionic acid cellobiuronic acid (111) from the Type I11 pneumococcus specific polysaccharide and synthesised from it an azoprotein antigen in the usual way.Antisera to it would pre- cipitate the Type I11 polysaccharide when it was combined to a heterologous protein. Artificial antigens containing cellobiuronic acid reacted with * I J . Exp. Med. 1937 66 191. ae Ibid. 1938 68 469. STACEY ASPECTS OF IMMUNOCHEMISTRY 191 Types 11 111 and VIII antisera. Even more striking was the finding 27 that antisera to cellobiuronic acid antigen conferred passive protection to mice against infection with virulent Types 11 111 and VIII pneumococci. This method of immunisation against an actual disease by an antigen con- taining a synthetic haptene has not yet been developed despite its immense potential importance.It was found that antisera to glucose and glucuronic acid azoproteins showed no cross-reaction with each other while the derivatives of glucuronic and galacturonic acid cross-reacted to the same extent as did the derivatives of glucose and galactose. The very powerful effect of acid very sharply in high dilution c$T:ko*-O) OH H He OH groups was shown e.g. glucu- ronic acid protein cross-reacted with Types 11 111 and VIII etc. antisera and a Type VIII CO2H H OH with Type I11 antiserum. W. F. Goebel 28 has prepared a synthetic antigen containing gentio- biuronic acid (IV) which differs from cellobiuronic acid in that glucose is attached through its C group to,the glycosidic group of the glucuronic acid moiety. The gentiobiuronic acid antigen formed precipitates with antisera to gentiobiose and to cellobiuronic acid antigens while gentiobiuronic acid antisera gave precipitin reactions with gentiobiose and cellobiuronic acid antigen.The gentiobiuronic acid antigen gave a precipitin test with Types I11 and VIII antipneumococcal sera and reacted slightly with Type 11. Rabbit antisera to p $ O $ ; f > H.OH the plex gentiobiuronic did not agglutinate acid Types com- 0 11 or I11 pneumococci but OH H OH protected mice against Type I1 (IV.) pneumococcal infection. This irnmunisation could also be ob- tained by glucuronic acid and cellobiuronic acid antigens and so showed that the in vivo protective action is due to the glucuronic acid determinant group. CH2* OH glucuronide-protein c r o s s e d (111.) Chemical Nature of Antibodies Antibodies are serum proteins and their physical properties show that they are globulins closely allied to the y-globulins.They differ from the normal globulins only in the respect that they possess the special specificity not possessed by the normal ; in other properties except perhaps isoelectric point l1 (pH 44-57) they are apparently indistinguishable. Some differences have been claimed between certain antibodies. In their reactions to denaturing agents and protein destroying agenta W. F. Goebel ibid. 1940 72 33. %* Ibid. p. 37. 192 QUARTERLY REVIEWS they resemble proteins in general. The correlation between heat denatura- tion of proteins and antibody destruction is not complete because the lability of antibodies is variable though generally as serum proteins become insoluble the antibody content disappears.Cold alcohol and ether do not in general denature proteins and treatment of antibodies with cold organic solvents is not harmful providing that the solubility of the proteins is main- tained. Diazonium compound formation and formaldehyde keten iodine and other substances which effect considerable substitution in serum pro- teins all reduce the activity of antibodies to a degree which depends on the extent of the treatment. Concerning the action of enzymes some antibodies have been reported to be destroyed by pepsin and less rapidly by trypsin. It is to be noted however that action of pepsin e.g. on diphtheria antitoxins gave con- siderable purification with production of a stable product that was specific- ally antitoxic. The initial agglutinin antibodies of the flagella or " H " antigen of Bact.typhosum were destroyed in vitro by pepsin trypsin and papain but after several immunising injec- tions the antibodies became resistant to pepsin and trypsin but not to papain. Horse antipneumococcal serum when treated with pepsin lost its mouse-protective power but not its precipitating power. There are state- ments in the literature that some samples of digested horse antipneumo- coccus sera combine with twice as much specific homologous polysaccharide as does normal antibody and that the product has a constant molecular weight of 100,000. These results point to the fact that non-protein prosthetic groups in antibodies may prove to be of high importance. Normal Serum Proteins.-It will be of interest to note at this stage some properties of the normal blood serum proteins.The following account is modified from that of Kabat.lOa Serum is an extremely complex mixture of proteins of which the main components have been widely studied under both normal and pathological conditions. Among the numerous methods devised for their estimation purification and characterisation the most important are fractional pre- cipitation by salts diffusion ultracentrifugal and electrophoretic analysis and immunological studies. Serum from different animal species is essentially the same according to the usual methods of chemical fractionation or even by electrophoretic or ultracentrifugal analysis. Thus sera from various animal species may be fractionated into albumins and globulins by the use of ammonium or sodium sulphate all show an albumin and two globulins by the ultracentri- fuge technique and usually four components-an albumin with a- b- and y-globulim-in the Tiselius electrophoresis apparatus.However the delicate techniques of serological methods have demonstrated that the serum components of various animal species are structurally different. Antisera prepared by immunising an animal with serum (or plasma) from one species will react in the highest dilution in the precipitin reaction with its homologous antigen but will also react less intensely with sera from closely related animal species and not at all with sera from species of more It is difficult to generalise however. STACEY 1 ASPECTS OF IMMUNOCHEWSTRY 193 remote zoological relationship. This phenomenon of species-specificity in the precipitin reaction which is shown by albumin globulin fibrinogen and mrum mucoid was established largely through the studies of G.H. F. Nuttall ao on whole serum and has been confirmed and extended by other workers using purified serum-proteins. Albumin- and globulin-fractions of serum can readily be separated by the u m of concentrated solutions of ammonium or sodium sulphate. The globulin-fraction has been generally considered to be that portion of serum- protein precipitated by half-saturated ammonium sulphate and the protein remaining in solution has been designated aa albumin. Both these fraction8 are relatively crude and each has been further fractionated by dialysis of the globulin-fraction against distilled water followed by separation of the water-insoluble globulins or by further use of ammonium sulphate.The albumin-fraction can readily be obtained in crystalline form and can be purified by repeated recrystallisation. Cohn and his co-workers 30 followed each step in the ammonium sulphate fractionation of serum-proteins in the Tiselius apparatus and were able to correlate solubility with the size and charge of proteins which were pre- cipitated by the ammonium sulphate under controlled conditions of pH and temperature. The percentages of water-soluble and insoluble protein were determined ; each of the fractions could be further purified and char- acterised by various well-known methods such as alcohol-precipitation at low temperatures (- 5") followed by purification by electrophoretic and ultracentrifugal methods. It was shown that the percentage composition and the electrophoretic mobilities of the various proteins of serum vary from species to species whilo the sedimentation-constants of each component 88em to be about the same for the various species.It has been found that crystalline serum-albumins are not necessarily pure proteins. Thus N. E. Goldsworthy and G. V. Rudd 31 demonst,rated by precipitin tests that even thrice-crystallised horse serum-albumin prepared according to G. S. Adair and M. E. Robin- son 32 contained as much as 2% of globulin impurity. E. A. Kabat and M. E. Heidelberger 33 showed in the precipitin reaction that this was respon- sible for the occurrence of a broad " zone" in which both antigen and antibody could be detected in the supernatant liquid in quantitative studies on horse serum-albumin system.Other workers demonstrated the presence of two components in both horse and human serum-albumin by electro- phoresis at pH 4.0. L. F. Hewitt 34 by repeated crystallisation of horse aerum-albumin separated from the crystalline albumin about 15% of a protein " seroglycoid " of high carbohydrate-content (8-6%) while R. A. Kekwick 35 crystallised two fractions one of high and one of low carbo- Serum-a2bumins. " Blood Immunity and Blood Relationship " Cambridge 1904. 30E. J. Cohn T. L. MacMeekin J. L. Oncley J. M. Newell and W. L. Hughes 32 Biochem. J . 1930 24 993. J4Bioche?n. J . 1934 30 2229; 1937 31 360; 1938 32 26; 1939 33 1496. 361bid. 1938 32 552 560. J . Amer. Chern. SOC. 1940 82 3386 3396. 31 J . Path. Buct. 1935 40 169. 33 J . Exp. Med. 1937 06 229. 194 QUARTERLY REVIEWS hydrate-content from the original crystalline horse serum-albiimii.Both fractions were antigenic. Other workers have also prepared crystalline carbohydrate-free and carbohydrate-containing albumin-fiactiom of horse serum and demonstrated that they were homogeneous in size and in electrophoretic mobility. a- and /?-GbbuZins. The electrophoretic and ultracentrifugal properties of normal a- and /?-globulins have been determined. It would appear that these products may be mixtures of proteins for a-globulins have been shown to have components with different sedimentation-constanto. Because of the diificulty of obtaining sufficient amounts in relatively pure form little immunochemical work has been carried out on these fractione. There have been obtained from normal human serum several globulins (" euglobulina ") which are immunologically different and it is probable that some of these correspond to the a- and @-globulins but as yet no direct comparison has been reported.Considerable new data on human and bovine serum-proteins have accumulated owing chiefly to the interest in these substances because of their value in combating shock. Janeway et aLS6 studied the immunological specificity of horse human and bovine serum-fractions prepared by repeated precipitation with ammonium sulphate or by alcohol fractionation. Albu- mins could be clearly distinguished serologically from the y-globulins but a- and /?-globulins gave cross-reactions with the other fractions. Antisera prepared against pure preparations of bovine albumin gave much weaker precipitin-reactions which may be due to a cross-reaction between home 1-globulin and bovine @-globulins present as impurities in the albumin preparations.It is to be noted that horse bovine and human albumim were found to be serologically distinct. The y-globulin of serum hm perhaps been the most exten- sively studied and is readily separated from the other serum-protein con- stituents in the Tiselius apparatus. Some fractions have been found to be homogeneous both in electrophoretic mobility and in sedimentative velocity though since they failed to give the type of solubility curve required by a single chemical entity they may be mixtures. Horse y-globulin frequently contains small amounts of an additional component which is similar to some pneumococcal antibodies. A y-globulin preparation hrts proved to be very valuable in treatment and prophylaxis of measles and of potential importance in therapy of scarlet fever.T. Harris and H. Eagle 3' studied euglobulins and pseudogloblllins from horse and human sera and concluded that euglobulin and pseudoglobulin were antigenically different proteins. J. R. Marrack and D. Duff 38 reported differences in the behaviour of eu- and pseudo-globulins with an antiserum to whole serum-globulin. F. E. Kendall 39 hasprepared the water-soluble euglobulin originally termed s6 G. A. Janeway S. Mudd and W. Thalkisner " Blood Substitutes and Blood 87 J . Urn. Physiol. 1935 19 383. seBrit. J. E x p Path. 1938 19 171. y-Globulin. "ramfusion" Chap. 21 C. Thomas New York 1942. 8@ J . Clin. Inveat. 1937 16 921. STACEY ASPECTS OF lMMTNOCHEM1STRY 195 an a-globulin from human serum and has shown that it is immunologically homogeneous and equivalent to the y-globulin fraction of Tiselius.By preparing antisera to this fraction he was able to determine the y-globulin content of serum and to obtain valuable results by measuring the increase in the amount of this component in certain pathological conditions. Antibodies have been known to be associated with the globulin fraction of serum and some antibodies produced in certain animal species such as the rabbit have recently been shown to be y-globulins. The general course of immunisation is characterised by well-defined changes in the electro- phoretic pattern of the serum globulins the proportion of y-globulin gener- ally increasing. In the horse injection of pneumococci often results in the production of a new globulin component which has been but little investi- gated.Its mobility in the Tiselius apparatus lies between the /3- and y-com- ponents and it is termed the " p2 " or " T " globulin. It has a higher molecular weight (ca. 1,000,000) than that of y-globulin (150,000). Van der Scheer and his colleagues 40 have carried out extensive investigations on the T globulin particularly in antitoxins from tetanus. The toxins and toxoids from the diphtheria bacillus have also been extensively studied ; the first effect during the immunisation of the horse is an increase in the y-globulin content. R. A. Kekwick and B. R. Record 41 found two types of antitoxin separable by electrophoresis in the same serum one being associated with the p-globulin and the other with the y-globulin.The /I-component which may be identical with a p2 or T component is produced more slowly and also flocculates much more slowly than the y-component. A. Tiselius and E. A. Kabat 42 in a most important study showed that rabbit- and monkey- (pneumococcal) antibody and rabbit-anti-egg-albumin (antibody) were quantitatively contained in the y-globulin fraction that removal of t h e antibody by adding antigen produced a decrease in the y-globulin and that this decrease corresponded quantitatively to the amount of antibody removed. The pH-mobility curves and isoelectric curves and isoelectric points of rabbit antibodies purified by salt dissociation or .barium hydroxide treatment were found to be identical with those of an electrophoretically separated y-globulin containing 76b/0 of anti-egg-albumin.These preparations were also found to be identical in sedimentation- and diffusion-constants molecular weight and frictional ratio. Samples of human pneumococcal antibody purified by salt-dissociation and of normal human y-globulin were also identical in these properties. Owing to the sharpness of the specific antigen-antibody reaction there are now available numerous methods for the assay of antibodies and it is possible especially with the modern methods of quantitative assay easily to follow their concentration. We have the non-specific methods available from colloid protein chemistry and the more specific methods from the antigen-antibody precipitation and regeneration of both components of the precipitate. Very valuable work has been done in this field by Felton (see reference 4).Isolation and Purification of Antibodies. 40 J. Van der Scheer R. W. G. Wyckoff and F. H. Clarke J . Immun. 1940,39,65. 4 1 Brit. J . Exp. Path. 1941 22 29. 4 2 J . Exp. Med. 1939 69 119. N* 196 QUABTERLY REVIEWS Nm-spcifi metha&. Whatever method of separation is used there is abundant evidence that antibodies can be precipitated in the globulin fraction of mrum. The recognition that globulins can be separated into several fractions led to the expectation that antibodies would be found to predominate in one of these though this has not k n entirely borne out by experiment. Fractionation by salt (e.g. ammonium sulphate) precipi- tation haa been used satisfactorily as well as dialysis electro-dialysis dilution methods pH changes etc.This method was investigated chiefly by Felt on with and withaut salt addition. Cold alcohol precipitation removed 90% of inactive protein ; in Felton’s important investigations the method was improved by use of alumina. Adsorption method. Aluminium hydroxide has been used followed by elution with N/lOO-sodium hydroxide. This was successful only with some anti bodies. In this group the most satisfactory preparations have been obtained with use of the pneumococcus polysaccharides as antigens. These can be obtained free from extraneous protein and good analytical control can be achieved since some polysaccharides are essentially free from any nitrogenous constituent. L. D. Felton 43 was able to separate off the polysaccharide haptene portion of the antigen-antibody complex by making use of the fact that this portion can be precipitated with calcium or strontium hydroxide.Thus he precipitated the homologous antibodies with Type I and I1 poly- saccharides and dissolved these precipitates with calcium hydroxide. Dialysis of the resulting solution gave the antibody protein in the form of an insoluble precipitate and this could be redissolved and reprecipitated. On reprecipitation with specific antigen upwards of 90% of the protein could be thrown down showing that this amount of antibody was equal to that given by any other method. It was mainly the studies on the mechanisms of precipitin- and agglutinin- reactions using the quantitative techniques described later which led. to the development of immunochemical methods for the purification of anti- bodies.M. Heidelberger F. E. Kendall and T. Theorell 44 found that the same amount of pneumococcal polysaccharide precipitated less antibody- nitrogen as the salt-concentration was increased from 0.151 M- to 1.79 M- sodium chloride. Making use of this observation M. Heidelberger and F. E. Kendall45 obtained antibody-solutions of high purity by extracting carefully washed pneumococcal polysaccharide-antibody specifk precipitates with 15% sodium chloride solution centrifuging off the remaining precipi- tate and dialysing the 15% extract against 0.85y0 saline solution. This principle was extended to the purification of antibody by salt dissociation of washed specifically agglutinated pneumococci. By its use antibody- solutions in which up to lOOyo of the total nitrogen was “immune” nitrogen were obtained from a number of animal species including horse Alcoholic precipitatwon.Specific precipitation. 43 J . Immun. 1932 22 453. 461bid. 1936 64 161. 4 4 J . Exp. Med. 1936 63 819. STACEY ASPECTS OF IMMUNOCHEMISTRY 197 cow pig rabbit monkey and man immuniaed against pneumococci. By following each step in the process of purification with the ultracentrifuge it was possible to show that the salt dissociation method did not produce any alterations in the size of the antibody rpolecules. The amount of antibody recovered using this method varies from 5 to 30% depending on the species of antibody and tb individual serum. The purity of the recovered antibody also seems to vary with different samples of serum. An additional amount of antibody may be recovered from the specific (insoluble) precipitate or specific agglutinate after the 15% salt extraction by suspending in water adding barium hydroxide and barium chloride in the cold centrifuging making faintly acid with dilute acetic acid and dialysing against 0.9yo sodium chloride until free from barium.Antibody- solutions prepared in this way from horse antisera were shown to have a high degree of purity by quantitative methods but they contained in- homogeneous components of molecular weight lower than the homogeneous preparations obtained by salt dissociation. Homogeneous products were however obtained by the barium hydroxide method with rabbit anti- pneumococcal antibody. The amount of antibody recovered by the barium hydroxide method was much higher than that obtained by salt extraction some yields being up to 40%.Studies by M. Heidelberger P. Grabar and H. P. Treffers 46 have shown that antibody prepared by these methods reacts with polysaccharide almost as does the antibody in the original sera and that the methods appear to yield a portion of all the anticarbohydrates of different reactivities present in the original serum. B. F. Chow and H. Wu 47 purified antibody by dissolving washed specific precipitates in alkali and neutralising ; a portion of the antibody then remained in solution. Some slight alterations in the immunological properties of the antibodies must occur during the process of purification since the purified antibody- solut,ions invariably contained a slightly larger amount of “ agglutinin ” than of “ precipitin ” although in the original serum agglutinin and pre- cipitin were identical.This alteration was also found to occur when horse pneumococcal antibody was concentrated by Felton’s method of pouring immune serum into twenty volumes of slightly acidified water. Other chemical methods not based on dissociation of specific precipitates have also been used. The methods for purification of antibody are not applicable to some specific protein-antiprotein precipitates such as egg-albumim-anti-egg- albumin. Methods have also been developed for the purification of antibody by dissociation using heat ether extraction and strong salt dissociation of Wassermann type antigen-antibody aggregates obtained by flocculative methods. Small amounts of antibody with considerable activity have been thus recovered. ‘13 Ibid. 1938 68 913. 47 Science 1936 84 316.196 QUARTERLY REVIEWS Quantitative Immunological Methods for Antibody Assay There is no doubt that the development by Heidelberger and his school of quantitative methods for antigen-antibody estimation and the steady insistence on their application t o every type of immune reaction has been responsible for the most outstanding advances as well as the steady progress in immunology. The first comprehensive theories were that of Ehrlich which emphasised the chemical nature of the reaction between antigen and antibody and that of Bordet which held that adsorption of one by the other explained the phenomenon. The gap between these two theories was not bridged by what Heidelberger aptly calls the “ mysticism of colloidal reactions ”. The more recent trends towards the interpretation in terms of chemical reactions especially in regard to protein studies place the matter in a new light.Landsteiner’s work on chemical specificity tlhe newer knowledge of immuno- logically specific polysaccharides and the artificial protein-carbohydrate antigens provided a ground-work for relating antigen-antibody reactions to chemical structures. When Heidelberger began his work several important facts were known ; thus it had been customary for serologists in assaying the potency of an antiserum (a) t o keep the volume of antibody constant and decrease the quantity of antigen ( b ) to keep the quantity of antigen constant and to decrease the amount of antibody or ( c ) to vary both and to determine the point of ’‘ optimal proportions ” of each. H. R. Dean and R.A. Webb 48 recognised the significance of the optimal proportions method and on it devised the useful serological method known by their name. T. Danysz 49 showed that when toxin is added t o anti- toxin the mode of addition of the toxin greatly influences the nature of the toxin-antitoxin complex. Thus if an equivalent of toxin is added in one portion the complex is non-toxic whereas if the same amount of toxin is added in fractions then the complex is toxic. An explanation of this phenomenon is connected with the ability of the toxin to combine with the antitoxin in multiple proportions. This effect is shown when a pneumo- coccus specific polysaccharide is added to homologous pneumococcus anti- body ; we generally can say that when an equivalent amount of an antigen is added in small portions t o a constant volume of antiserum then a greater total amount of antibody is precipitated than when the same equivalent amount of antigen is added in one lot.Analytical chemical methods were applied to the study of the precipitin reaction by Wu and his collaborators 50 who studied the hzemoglobin- antibody and iodo-ovalbumin-antibody systems and attempted to estimate ‘‘ marked ” antigens colorirnetrically in order t o compare their amounts with the total nitrogen content of the precipitate. ‘8 J . Path. Bact. 1926 29 473. 4@ Ann. Inst. Pasteur 1902 16 331. 60 (a) H. Wu L. H. Chang and C . P. Li Proc. SOC. Exp. Biol. Med. 1987-28 25 852 ; ( b ) H. WU P. P. T. Snh and C. P. Ti ibid. 1928-29 26 737. STACEY ASPECTS OF IMMUNOCHEMISTRY 199 M. Heidelberger and F. E.Kendall 51 then took advantage of the fact that by using a nitrogen-free .Type I11 specific pneumococcus polysaccharide as an antigen it was possible by determining the total nitrogen (estimated by the Kjeldahl method) in such an antigen-antibody precipitate to assay the antibody directly and to express the antibody content of a serum in terms of “antibody nitrogen ”. The authors have since applied their methods widely to other systems. In an early study increasing amounts of antigen were added to a constant volume of antibody and the amount and composition of each precipitate were determined. Four zones were recognised ( a ) that of antibody excess in which the antibody is detect- able in the supernatant fluid ( b ) an “ equivalence ” zone in which neither antigen nor antibody could be detected ( c ) a zone of antigen excess when the amount of precipitated antibody protein just begins to diminish and (d) an inhibition zone of antigen excess when antibody remains in the supernatant fluid as a complex no longer detectable as antibody.With this knowledge available and by correct choice of the reaction range i.e. working in the range of antibody excess Heidelberger and Kendall worked out their theory which was based on the following assump- tions (a) antigen and antibody are “ muhivalent ” in respect to one another (i.e. contain two or more combining groups and obey the law of mass action),* ( b ) combination of antigen and antibody proceods by a series of bimolecular reactions the initial compounds formed being soluble but owing to the multivalency of antigen and antibody combination con- tinues until large aggregates are formed and are precipitated ( c ) the dis- sociation of the initial compounds is negligible and (d) when the antigen is in considerable excess a soluble compound is formed which has a ratio of antigen to antibody about half that of the ratio at the equivalence point.The authors arrived at the formula R8x2 Antibody in the precipitate (in mg.) = 2Rx - - A antigen antibody where R is ratio of ~ at the antigen-antibody equivalence point ; A is total antibody ; x is the antigen added in milligrammes. By a slight modification of this equation it was possible to get a linear relationship between the factors and by making two or three analyses (in duplicate) to characterise an unknown antiserum. In most of the vast number of cases studied the formula has given satisfactory agreement between observed and calculated values in the range considered * and has given a sound and mathematical expression to Marrack’s “ lattice ” hypothesis.In general the theory has raised some controversy and criticisms.s Pauling 62 in par- ticular considered that Heidelberger’s assumptions were unlikely and 51.J. Exp. Med. 1935 61 559 563; 62 467 697. * The writer is of the opinion that the application of the term “ valency ” as applied in immune reactions is unfortunate and could well be replaced by the term “sero- valency ”. b3L. Pauling D. H. Campbell and D. Pressman Phyeiol. Rev. 1943 @la No. 3. 200 QVARTERLY REVIEWS arbitrary and welcomes the improvements by F. E. Kendall,63 who more recently made a statistical approach avoiding some of the oversimplified assumptions of the older theories and by assuming antibodies to be bivalent arrived at the same equations as above.He was able to account quantita- tively for the course of the precipitin and toxin-antitoxin reactions by assuming that precipitating antibodies have two similar reactive groups per molecule while antitoxins have two different reactive groups per molecule. Pauling and his colleagues s 2 studied the specific precipitation of azo- dyes and other simple compounds as well as specific inhibition of such reactions by univalent haptenes and applied in an advanced form the laws of chemical equilibrium but they arrived a t the same general formulae as that of Heidelberger. It is considered that the theory applies only to bivalent antigens and antibodies univalent antigens and certain soluble complexes.A. D. Hershey 54 has put forward an involved theory of antigen-antibody equilibrium based in part on probability considerations and on the multivalency of antigens and antibodies. The theory is in approximate agreement with experiment a1 findings. The quantitative methods can be applied to the agglutinin reaction 86 by making use of the fact that " agglutinin nitrogen ) ' is obtained by sub- tracting the nitrogen content of the bacterial suspension from the total nitrogen content of the same volume of agglutinated washed bacteria. Numerous applications of quantitative methods are now being applied e.g. for the estimations of haemolysis by use of a washed suspension of stromata for measuring the amount of Wassermann antibody in human syphilitic sera and for the estimation of the specific polysaccharide content of an unknown solution etc.Further there is the important application as mentioned elsewhere for the estimation of " complement nitrogen ". This is based on the fact that combinations of certain antigens with homo- logous antibodies e.g. in the rabbit will " bind " complement. Thus by adding antigen and antibody to a serum containing complement the esti- mation of the increased nitrogen in the centrifuged washed precipitate as compared with controls gives a direct measure of complement. Quantitative immunochemical methods will be increasingly applied for determination of the homogeneity of proteins. Specific antisera to a sus- pected contaminant will detect it particularly in virus and tissue proteins In this way it has been shown that considerable amounts of normal tissue constitxents are contaminants of supposedly pure influenza1 virus and that all known virus preparations are highly contaminated with closely related tissue products.This is of importance in tumour studies. Heidelberger and his colleagues 55 found that antibody molecules were not always uniform some for example having but weak combining power with the antigen and being brought down only in the presence of more potent or " avid " antibodies. An elegant justification of Heidelberger'a faith in his power to get pure antibody protein in his washed precipitates 6 3 Ann. N.Y. Acad. Sci. 1942 93 85. s c J . Imrnun. 1942 4 3 9 ; 1943 46 249; 1944 48 381. ssM. Heidelberger F. E. Kendall and H. W. Scherp J .Exp. ikled. 1936 64 559. STACEY ASPECTS OF IMMUNOCHEMISTaY 201 haa been g h e d from a study of labelled nitrogen 68 in amino-acids fed t o animals paasively immunised with Type I pneumococcus antibody. The complete absence of 16N in the washed antigen-antibody complex precipi- tated under the standard conditions afforded a rigorous test of the specificity of immune precipitation and justified the previous assumption that only antibody nitrogen was meaaured. A most valuable practical application of the quantitative methods is the growing realisation of the truth of the demonstration that there is a parallel between mouse protection and the amount of “ antibody nitrogen ” precipitated from a homologous reaction. Despite chemotherapy passive immunotherapy will still play a part in the conquest of disease e.g.as with influenza1 meningitis in which it still is the sole remedy while the realisation of an immune state against all infectious diseases is an ideal always to be striven for. The Size of Antibody Molecules (see Sevag,ll p. 10).-Some studies have been made on the molecular size of immune and y-globulins. Pneumo- coccal antibody from the horse cow and pig have been shown to have the same molecular weight and frictional ratio and the same variation of the sedimentation constant with concentration. Thus horse diphtheric anti- toxin was found by A. M. Pappenheimer junr. H. P. Lundgren and J. W. Williams 57 to have a molecular weight of 184,000 and to correspond in size to the normal horse y-globulin. Highly significant estimations of the molecular weight of antibodies have been obtained by M.Heidelberger and K. 0. Pederson,58 and by E. A. Kabat 69 in Svedberg’s laboratory. It was apparent to these workers that knowledge of such molecular weights would give valuable information regarding the relationship of antibodies to normal serum proteins and the mechanism of antibody formation. Further it might be possible to learn something of the limiting compounds involved in antigen-antibody forma- tion and to decide whether antibodies from the same infectious agent of disease were the same in different animals. Heidelberger and Pederson 44 found that highly potent rabbit Type I11 pneumococcus anticarbohydrate (antibody) was homogeneous in the ultra- centrifuge and that its sedimentation constant did not differ from that of the principal component of normal rabbit globulin or of immune rabbit globulin containing up to 50% of antibodies to egg albumin.The molecular weighta of all were of the order of 150,000. Antibodies are generally assumed to be ellipsoid in shape. Type I pneumococcus anticarbohydrate was homogeneous in the ultra- centrifuge only when prepared from sera stored without preservation. Its mdimentation constant was comparable with that of most purified anti- bodies. The molecular weight of pneumococcus anticarbohydrate in the home is three or four times that of the principal normal globulin. K a b t 60 extended the work and found that antibodies from various a* M. Heidelberger H. P. Treffera R. Schoenheimer S. Katner and D. Rittenberg &7 J. Zxp. Me&. 1940 71 247. 3. B b l Uhem. 1942 I* 666.68 IW. 1937 86 393. u Ibid. 1939 Q9 103. 202 QUARTERLY REVIEWS animals which could be obtained in homogeneous condition by salt dis- sociation methods fell into two groups. I n one group-cow horse and pig- a giant molecule of molecular weight of almost 1,000,000 was formed. I n another group-human monkey and rabbit-the molecular size was lower being that of normal y-globulin (150,000). Kabat found that with pH change horse antibodies tended to aggre- gate at higher acid pH and to break down at more alkaline pH. Some horse antibody preparation showed evidence of breakdown during extended immunisation periods. L. Pauling and D. H. Campbell6(’ believe that they have “manufac- tured antibodies in vitro ” by mild denaturation and “ regeneration ” of bovine y-globulins in the presence of certain dyes and of antigens including the Type I11 pneumococcal polysaccharide.Precipitates formed during these procedures are said to contain antibodies and these authors state that on separating the antigen the “ antibody ” solutions obtained reacted specifically with the “ antigen” used to produce them. Kabat 9a has criticised the technique of this work. Origin of Antibodies The problem of t’he origin and specificity of antibodies is most intriguing. The earliest theory held that tJhey were prpteins containing a constituent fragment of the antigen which served to determine specificity. The idea was abandoned when it was shown that antigens containing such readily detectable elements as arsenic gave antibodies which were proteins con- taining no demonstrable arsenic.Antibodies are modified serum globulins so that interest is now centred in the mechanisms responsible for normal globulin synthesis. The reticulo- endothelial cells may be involved and much effort has been expended in proving this. The literature on the subject has been critically reviewed by F. P. Gay.61 The most reasonable theories of antibody formation are those of F. Breinl and F. Haurowitz,62 J. A l e ~ a n d e r ~ ~ and S. Mudd,‘j4 which in common assumed that antibodies are new globulins synthesised under the directing influence of a molecule or a determinant fragment from the antigen. As the new molecule of globulin is formed in the cell the arrangement of the chemical groups on part of its surface will be partly determined in template fashion by the proximity of the antigen molecule.The configuration of part of the surface of the antibody molecule is thereby mirrored from a portion of the surface of the antigen molecule. The cell system remaining fixed will continue t o pour out many similar molecules. The compatible relationship in structure leads to a strong attraction between the antigen and the antibody the shapes being such that the two molecules can come 60 J . Exp. Med. 1942 76 211. 61Medicine 1929 18 211 (see also Physiol. Rev. 1931 11 277). 6 2 2 . physiol. Chem. 1930 192 45. 64J. Imntun. 1932 23 423. 63 Protoplasm 1931 14 296. STACEY ASPECTS OF IMMUNOCHEMISTRY 203 in close contact with one another and thus increase the intermolecular force of attraction between them. The high complexity of the protein surface structure makes readily apparent the reasons for the great specificity shown by these reactions.These theories of antibody formation have been given a physiological basis by F. P. Sabin 65 who followed the fate of a dye protein “ R-salt- azobenzidine-azo-egg-albumin ” introduced in several ways into the animal and traced its disappearance in various tissues. She observed during antibody formation macrophages in the milk spots of the omentum and cells of the reticulo-endothelial system which appeared to lose surface films by extrusion a t an abnormal rate. With the shedding of the layers normal and antibody globulins were carried into the blood plasma and it was postulated that during the increase in the synthesis of normal globulin by the modified macrophages the antigens were able in some way t o alter some of the globulins into the approximate antibody.Other workers have produced considerable evidence that lymphocytes form antibodies which along with y-globulins have actually been isolated from extracts of lymphocytes. does not believe that an antigen modi- fies an antibody by altering the order of the amino-acids in the globulin chain. He assumes that all antibody globulins contain essentially the same long polypeptide chains as those in normal globulins and differ only in the manner in which the chain is coiled in the molecule. He regards it as possible that a relatively stable middle part of an antibody remains con- stant and identical with the same part of a normal long chain globulin. During antibody formation however the relatively labile chain-ends of the globulin can uncoil and then under the influence of the antigen can fold up in a modified way to give a different configuration possessing nearly the same energy.The modification takes place in such a way that owing t o their structural complementary nature there will be an attraction between the newly coiled chain-end and the antigen. The newly acquired con- figuration may be any one of a large number according to which part of the antigen happens t o exert its influence on the chain-end and upon the size of the surface covered by the antigen. Thus all antibodies should have a good deal in common with globulins and if used as antigens ought to give essentially complete cross-reactions with normal globulins. It should however be pointed out that J. H. N ~ r t h r o p ~ ~ using as an antigen a crystalline antitoxin prepared by digesting a diphtheria antitoxin obtained an immune serum which reacted specifically with the antitoxin but failed to react with normal serum proteins.This may throw doubt on the pres- ence of Pauling’s “ mid ” and “ end ” pieces in antibody. Regarding the continuity of antibody production Pauling states that “ the antigen molecule after its desertion by the newly formed antibody molecule may serve as the pattern for another ”. Based on his conception of antibody formation Pauling and his col- 66J. Exp. Med. 1939 70 67. 6’J Gen Phy8wl. 1942 25 465. Pauling’s theory. L. Pauling e6 J . Arner. Chem. SOC. 1940 62 2640. 204 QUARTERLY RJEVIEWS leagues 60 made efforts to denature globulins and to “ refold ” the ends in the presence of antigen in uitro and indeed he has claimed to have done this.As mentioned previously the methods of detecting these antibodies have been sharply criticised by Kabat,lW and it k perhaps significant that no further developments of such a remarkable method have yet been reported. I n order to account for the changes in the nature of antibody during extended immunisation Burnet suggested that antibodies are syntheeised by intracellular proteases altered by the antigen which even when destroyed still leaves its impress upon these proteases. I n the writer’s view Pauling’s theories on the mode of antibody forma- tion are unlikely for antibody formation must be considered in closer regard to the enzyme systems responsible for serum protein synthesis and in regard to the fact that y-globulins are mucoproteins containing a significant carbo- hydrate residue.It is clear from the work of Schoenheimer Heidelberger and their col- leagues that antibodies are being continuously synthesised a t the same rate as other plasma proteins. These workers used dietary amino-acids containing 15N and showed that plasma proteins incorporated the labelled element a t the same rate as did those of the liver and kidney and that the half-life of an antibody molecule was the same as that of other serum proteins i.e. about 2 weeks. Moreover they showed that passively intro- duced antibodies do not play a part in the metabolic cycle. It is most unlikely that globulins once formed ~ l l be modified in vivo in any way. From our knowledge of the powerfully determinant nature of carbohydrate residue of other complexes it would seem that the precise configuration of all y-globulin merits urgent investigation.I n the writer’s view in order to discover the origin of antibodies one will need to refer to those enzymes which are responsible for the synthesis of mucoproteins of the y-globulin type and moreover to those synthesising enzymes which are carried in the chromosome and gene systems of cells. We now have the knowledge from the work of Avery 68 and his colleagues that capsular pneumococcal polysaccharide synthesis can be induced by the introduction of deoxyribonucleic acid. Thus a rough form of Type I1 pneumococcus was converted permanently into a Type I11 smooth type by means of a minute amount of a Type I11 deoxyribonucleic acid ; this was the first instance of the alteration of a genic system by a chemical substance.The writer believes that on injection an antigen can combine with or modify permanently part of the enzyme system which is responsible for mucoprotein synthesis of the globulin type. These mucoprotein-synthesising enzymes are most likely to be part of the chromosome system of the cell and belong to thd class of self-perpetuating or autosynthetic enzymes. Thus once altered (and the alteration will be such that the synthesised macro- molecule will be complementary to some part of the “altering” factor namely the determinant part of the antigen) there will be for a considerable 0. T. Avery C. M. Mrtcleod and M. McCarty J . Eq. Med. 1944 79 137. STACEY ASPECTS OF IMMUNOCHEMIBTRY 205 period-perhaps indeed for the lifetime of the animal-some enzymes con- tinually produced which always synthesise antibody globulins and give rise to identical " daughter :' enzymes.The problem as to whether the carbohydrate residue or the polypeptide reeidue of the y-globulin mucoprotein is the part most readily m W d remains for future study. Antigen-Antibody Reactions.-There would now a p p r to be no doubt that the reaction between antigen and antibody is chemical in nature although interpretation of some observations is often dScult because of the size and complexity of the reactants. In many ways the reactions are similar to those between enzymes and their substrates.ll There are several salient facts regarding the reaction. Thus (a) it is specific ( b ) both antigen and antibody enter into the specific precipitate or agglutinate ( c ) the whole antigen and the whole antibody react (d) the antigen-antibody complex appears to behave like a rigid ellipsoid (e) there is no degradation of any part of the complex it being possible to recover unaltered antibody from it (f) the combination of antigen and antibody takes place at the surface of the molecules (9) the combination between antigen and antibody although firm is reversible and (h) antigen and antibody can combine in varying proportions and in general they are both at least bivalent and probably multivalent.The quantitative studies of Heidelberger and his rjchool are discuseed in another section and similar work has been done by F. Haurowitz 69 and others. There has more recently been conducted an extensive investiga- tion on the serological properties of polyhaptenic substances by L.Pauling and his co-~orkers,~~ who prepared 27 simple compounds containing the phenylarsonic group as the haptenic principle [of the types (V) to (VIII)] apd used them in precipitin tests with certain antisera which were obtained from rabbits which had been injected with azo-phenylarsonic-acid-sheep- serum. R R (V. 1 Me HOOR R (VIII.) H O O - < - > O H OH <->Aso,H HO-R ROR (VII.) OR It was found that twenty simple antigens containing two or more hap- tenic groups per molecule gave precipitates with antisera whereas seven monohaptenic compounds [of the type (IX) to (XI)] failed to do so. OeZ. physiol. Chem. 1936 295 23; Bull. SOC. Chim. bid. 1937 19 1453. 7O J . A w . Chem. Sw. 1942 64 2994 3003 3010 ; 1943 85 728 ; 1944 86 330 1731; 1945 67 1003 1219 1602.206 QUARTERLY REVIEWS Pauling concluded that these results were in accordance with the multi- valent antibody concept indicated by the Marrack-Heidelberger frame- work theory. The failure of monohaptenic compounds to form precipitates with homologous antisera was ascribed by Pauling to the solubility of such an antigen-antibody complex. He maintained that polyhaptenic substances reacted with antisera to form infinite aggregates of visible dimensions. From quantitative investigations of these reactions under various con- ditions (pH temperature time etc.) an expression was deduced for the rate of combination of antigen and antibody based on the assumption of the bivalency of the haptenic compound. The bivalent nature of the antigen was further substantiated by determination of the composition of antigen- antibody complexes.The composition appeared to be constant for all poly- haptenic compounds [such as (VI) (VII) or (VIII)] and this was explained The inhibition by a variety of substituted phenylarsonic acids of the action between poly- haptenic antigens (containing phenylarsonic acid groups) and homologous antiserum was then studied. The results indicated that the strength of t,he hapten-antibody bond was related to the position of a substituent in a phenylarsonic-acid-hapten molecule (rather than to the change in the degree of dissociation of the -As0,H2 group). It was found that o-substituted haptens formed weaker bonds with antibodies than m- or p-compounds this was ascribed to the minimum steric hindrance arising in p-substituents. The -NO2 group produced a striking increase in the '' bond strength constant " of the hapten group.Further credence in favour of the " framework " theory was gained from the study of simple substances containing two different haptenic groups vix.,~ -N=N-AsO,H (" R ") and -N=N-CO,H (" X '7. Such simple substances form precipitates not with either anti-R serum or anti-X serum but with a mixture of the two. This furnishes proof of the effective bivalency of the dihaptenic antigen. Many of the investigations using phenylarsonic acid as the haptenic group have been repeated using p-substituted azobenzoic acid derivatives and it has been confirmed that bivalent simple substances gave specific precipitation with homologous antiserum and that p-substituted azobenzoic acid haptens were endowed with greater inhibiting power than 0- or m-derivatives.Pauling and his collaborators then proceeded to re-examine the behaviour of -0- m- and p-substituted haptens in contact with homologous and heterologous antiserum as first investigated by Landsteiner. Experiments were made with antisera hom,ologous to 0 - m- and p-azo- phenylarsonic acid groups (prepared by injection of rabbits with sheep serum coupled with diazotised 0- m- and p-arsanilic acids) in reaction with azo-ovalbumins containing these groups. It was found with substi- tuted azophenylarsonic acids that the values of hapten " inhibition con- stants " could be largely accounted for by consideration of the operative intermolecular forces van der Waals forces attraction of substituent groups and antibody formation of hydrogen bonds and steric hindrance.steric interference of attached antibody molecules. STACEY ASPECTS OF IMMUNOCHEMISTRY 207 Some evidence concerning the molecular asymmetry of antibodies was next adduced. An antiserum was prepared by injection of rabbits with an azoprotein consisting of sheep serum coupled with diazotised p-amino- succinanilic acid ; the antiserum so obtained gave specific precipitations with a homologous azoprotein antigep (consisting of ovalbumin coupled with diazotised p-aminosuccinanilic acid). The precipitation it was observed was inhibited in differing degrees by D- and L-isomers of N-(at-methylbenzy1)succinamic acid the L-isomer having the greater in- hibitory power. This behaviour was ascribed to the presence of optically active amino-acid residues in the antibody niolecule.An invegtigation has been carried out c m the power of various haptens related to phenylarsonic acid in which t lie molecular structure had been modified to inhibit the 0- m- and p-azophenylarsonic acid honiologous reaction. Replacement of As by P Sb and S in the modified liapten wits effected. Whilst the p-phosphonic acid showed siiiiilar (or enhanced) inhibitory power to that of the p-arsonic wid the antimony and sulphur derivatives were almost devoid of such activity. Fkplacenieiit of plienyl by methyl caused greater loss of inhibitory power than replacenlent by benzyl. The strength of the antigen-antibody bond was accordiiigly correlated with structural similarities of the hapten molecule. The Specijc Forces holding together the A ntigen-A ntibody Complex It must be emphasised that antigen-antibody reactions are invariably performed in weak salt solutions and in general though the interactions are highly specific they are weak and.reversiblo and there is no doubt that antigen-antibody attraction is not due to ordinary chemical bonds.Pauling has drawn attention to the fact that these interactions may bch classified as electronic van der Waals attract ion Coulomb forces attraction of electric dipoles or multipoles hydrogen bond formation etc. The shape of the constituent smaller molecules determines tho ways in which the macromolecules can be packed together. and since the forces of attrac- tion increase rapidly as the molecules approach more closely to one another it is clear that those molecules which can bring large portions of their surfaces in close-fitting juxtaposition will generally show relatively strong mutual attraction.The most important intermolecular forces in the antigen-antibody complex are probably the van der Waals though the almost equally important structural feature termed the hydrogen bonds which involves the attraction of polar groups is now rapidly being recog- nised. The intermolecular forces do not in themselves account for the specificity of the immune reactions. This depends much more on the actual shape of the interacting molecules which must possess on their surfaces relatively large regions showing mutually complementary configurations with consequent mutual attraction actively electrically charged groups and hydrogen-bond forming groups. This close proximity of " opposit,e type )' groups and the complementary nature of structure allows more specific intermolecular forces to come more fully into operation than would be possible with other less specific structures.This importance of con- 208 QUARTERLY REVIEWS figuration and molecular size has an analogy in crystallisation where examples of close molecular packing of similar shapes are well known. In both immune reactions and crystalline state the equivalence in size and shape of unrelated groups such as the methyl group and the bromine atom are well recognised and are important. I n most immunity reactions it is usually considered tthat the initial specific union is followed by a second stage which results in the precipita- tion of the high molecular complex. Pauling l3 draws attention to the fact that it would appear that each antigen molecule attaches itself to two antibody molecules the process continuing until a framework has been built up of such a size that it no longer can stay in solution and therefore precipitates much in the same way as in the precipitation of silver cyanide.Pauling recalls that the silver ion has the property of forming two covalent bonds with cyanide ion the ion attaching itself to two cyanide ions which stick out on either side of it so that cyanide ion has the property of forming two covalent bonds one by C and one by N. I n this way long chains of alternating bi-covalent silver and bi-covalent cyanide groups are formed and ultimately these long chains arrange themselves side by side to form t h e silver cyanide precipitate. Pauling has suggested other similarities b 4 ween the specificity of serological reactions and crystallisations.In regard to the second stage in the formation of the antigen-antibody complex there has been some controversy as to whether or not it is specific like the first stage. One of the first theories considered the precipitation as being due t o the neutralisation of opposite electrical charges but it is now known that under ordinary conditions in salt solution both antibodies and most antigens are negatively charged. Another theory was that there was initial formation of a hydrophobic colloid which was later precipitated in the presence of electrolytes but this does not stand up t o close scrutiny particularly since it is knovn that cells can be agglutinated by an amount of antibody insufficient to cover the cell surface.S. B. Hooker and W. C. Boyd 71 favour the idea that the particles grow to visible size by a process of indiscriminate aggregation of different sized particles though Pauling disputes this. Another idea is that polyhaptenic compounds can bond the antibody molecules so tightly that the polar groups are masked and solvent molecules fail to penetrate. In general the experimental results show that the reaction is specific but there still remains a big field for research regarding the mechanisms of the aggregation of giant molecules. The most reasonable theory regarding the mechanism of both precipita- tion and agglutination is the lattice or framework theory of Marrack 5 which is supported by the schools of both Heidelberger and Pauling. If we consider the agglutination of bacterial cells must accept the possi- bilities that the same forces which act in the first stage of the specific reaction must come into play in the second stage.When the antibody has the power of attaching itself to two cells (A) it could form two bonds 'l J . Imrnzin. 1931 21 1 I:!; 1932 23 446; 1933 24 141. STACEY ASPECTS OF IMMUNOCHEMISTRY 209 and hold them toget4her thus A-B-A. Repetition of the combining process in a linear direction would lead to a larger complex such as -A-B-A-B- and in three dimensions to still larger framework ; thus (in its simplest form) I B I I3 I I -A-B-A-B- I I B A I I R A and so on until the clumps become macroscopic in size and visible. These ideas require that antigens must be multivalent and antibodies in general at least bivalent and there is now a mass of evidence to supporh this although some special antibodies are most probably univalent.It is of interest here to note the significance of the “ equivalence zone ” for it is known that serological precipitates dissolve in presence of ex,cess of antigen. This phenomenon closely parallels the dissolution of silver cyanide by excess of cyanide ion and in a similar way is due to the formation of soluble complexes. The specific inhibition reaction by haptenic substance will be seen to be due to the blocking of combining groups on the surface of the antigen and the prevention thereby of formation of a large framework. The writer has recently observed a phenomenon involving the biological synthesis of macromolecules which undoubtedly is generally analogous to the precipitin reaction.The observation emphasised the possibilities that some of the far-reaching theories set out in Sevag’s “ Immunocatalysis ” may be correct. Thus when glucose-1 phosphate is set up at pH 6 with an active phosphorylase enzyme extracted from peas there is a gradual increase in opalescence followed by a deposition of amylose-type granules. In presence of excess of glucose- 1 phosphate the precipitation continues until all the enzyme has been removed from solution and begins again when more enzyme is added at the correct pH. In the writer’s view the precipitate consists of amylose-type chains “ cemented ” together by pros- thetic groups originating from the enzyme and it would appear that just as a minute amount of antigen can precipitate a relatively large excess of antibody so a minute amount of enzyme prosthetic group can unite a large excess of synthetic polysaccharide.Boyd brings forward some objections t o the framework idea (which he terms the “ alternation theory ”) and has offered an alternative theory of the precipitin reaction which he calls the “ occlusion theory ”. He explains the initiation of precipitation as being due t o the fact that when antigen- antibody molecules are brought into close apposition there are steric effects and a mutual neutralisation of the solubilising polar groups ; these thereby become too few t o attract water molecules and there is a consequent lowering of the solubility of the complex in the salt solution. It is claimed 210 QUARTERLY REVIEWS that the theory is supported by experiments involving the precipitation of antibody by bi- ter- and multi-valent haptens but no explanation is given of the method whereby the insoluble complexes unite to give the visible precipitates.Complement Complement is umally detected by adding to a system sheep-blood cells sensitised with specific antibody ( e .g. anti-sheep-cell rabbit serum). When complement is absent no lysis of the cells occurs. Complement in addition to the property of being able to lyse antibody- sensitised red cells can lyse some sensitised bacteria and can kill other bacteria; further it can add on to many antigen-antibody complexes. The well-known " Wassermann reaction " used in the test for syphilitic infection is analogous to complement fixation. The work done on fractionation of serum in respect to the components responsible for complement activity has led to the characterisation of four specific factors each of which is necessary for complement action.T. W. B. Osborne 7 2 and L. Pillemar 73 have summarised the experimental evidence on which the separation of complement into four components is based; these may now be briefly characterised as follows. This factor is precipitated from guinea-pig serum by passing carbon dioxide through a solution of serum diluted t o 1 in 10 with distilled water or by dialysis against distilled water ; C'1 is destroyed by being heated to 56" for 30 minutes. It is of the globulin type possibly associated with a phosphatide. This factor remains in solution when carbon dioxide is bubbled through the 1 in 10 aqueous solution and is also of the globulin type containing 10% of polysaccharide.This is specifically inactivated by yeast zymin or cobra-venom but is unaffected by being heated at 56" for 30 minutes. It is to be noted that the entire complementary activity of zymin-treated serum can be restored by addition of heat-inactivated serum though it has also been shown that an insoluble-carbohydrate fraction of yeast specifically inactivates this third component. This is specifically inactivated by treating serum with dilute ammonia hydrazine or viper-venom or by shaking it with chloroform or ether. It is unaffected by being heated a t 56" for 30 minutes. The addition of 10% sodium chloride to complement markedly diminishes the thermost ability of the C'4 component. Pilleniar and his colleagues 7 4 have reported on the effect of numerous agents on complement.Reactivation of sera from which any one com- ponent is missing can be achieved by adding separately this missing component. C'I or mid-piece. C';? or end-piece. C'3 or third component. C'4 or fourth component. 7 2 " Complement or Alexin " Oxford University Press London 1931. 7 3 Bact. Rev. 1943 33 1. 74 L. Pillemar S. Seifter C. L. San Clemente and E. E. Ecker J . Immun. 1943 47 5. STACEY ASPECTS OF IlKMUXOCHEMJSTRY 211 In whole serum or in any artificial mixture of the four components of complement the complement-titre is determined by the component present in the smalIest number of units and it has been demonstrated that sera of Werent animal species vary in their relative content of the various com- ponents. Thus for example in guinea-pig serum the component usually present in lowest effective concentration has been found to be C’l whereas in human serum it is usually C’2. Heidelberger and his colleagues 76 have been able by quantitative methods to show that absorption of complement by the antigen-antibody complex gives an appreciable and measurable increase in the nitrogen content of the precipitate. A possible mechanism for the mode of incorporation of com- plement into the precipitate has been discussed and the signjficant observa- tion made that only a very small portion of the surface of red blood cells need be covered in order for haeniolysis to occur. 76 (a) M. Heidelberger J:Exp. Med. 1941 73 681 ; ( b ) M. Heidelberger and M. Mayer ibid. 1943 75. 285. (To be conclude&.)

ISSN:0009-2681    

DOI:10.1039/QR9470100179

出版商:RSC    

年代:1947

数据来源: RSC