摘要:

QUARTERLY REVIEWS ASPECTS OF IMMUNOCHEMISTRY (concluded) By MAURICE STACEY PH.D. D.Sc. F.R.I.C. (PROFESSOR OF CHEMISTRY UNIVERSITY OF BIRMINGHAM) Serological Properties of Antibodies. Formation of '' Anti-antibodies '' AN obvious result of the recognition of the globulin nature of antibodies is to study their immunological behaviour when they are used as antigem for injection into animals to give " anti-antibodies " which may be of some importance in therapy. Preliminary data were obtained by K. Land- steiner and E. Prasek 76 who demonstrated that precipitins for normal horse serum would remove the agglutinin from anti-typhoid horse serum. Instead of using purified antibody solutions as antigens specific antigen- antibody precipitates are quite satisfactory and are more accessible. When the antigen part of the specific precipitate is a polysaccharide haptene it will obviously not give rise to antibodies on injection and the antibody response will be due solely to the antigenic power of the protein antibody.In an extensive series of experiments on normal and immune horse serum fractions injected into rabbits Ando 77 concluded that antigenically there was a relation between normal horse globulin and diphtheria antitoxin and that this was distinct from the specificity of various antibacterial antibodies. These results were extended by H. P. Treffers and M. Heidelberger 78 by an adaptation of their quantitative agglutinin methods. They showed that anti-pneumococcus Types I and 11 and anti-C (group specific) and anti-H influenza antibodies from horse antisera were quantitatively identical in their antigenic reactions towards a rabbit antiserum to one of them (Type I pneumococcus horse antibody).It appeared that those chemical groups in the antibody protein molecule which were responsible for antibody proper- ties were not involved in the antigenicity of the antibody. Different characteristics were found with diphtheria antitoxin-toxin floccules which on the other hand were found to remove only one-half of the precipitable antibody from this anti-antibody serum. The results were confirmed with a similar anti-antibody produced in the chicken. Use of the latter species also permitted the examination of'the antigenic behaviour of rabbit anti- bodies which were found to be identical as far as studied. In a later paper by H. P. Treffers D. H. Moore and M.Heidelbe~-ger,~~ comparison was made using the same rabbit anti-antibody serum (to a specific precipitate from horse antiserum) of the antigenic properties of salt-dissociated anti- body solutions with various fractions isolated by electrophoresis an& ultra- 7aZ. Immunforsch. 1911 88 10. '7 ( a ) K. Ando J . Immm. 1937 33 41 ; ( b ) K. Ando R. Keg and T. Komhyana ibid. 32 181. J . Exp. Med. 1941 78 125. 79 J . Exp. Med. 1942 76 135. 213 0 214 QUARTERLY REVIEWS centrifugation of normal horse and goat sera. It was shown that the two antibody solutions as well as the y-globulin fraction from an antipneumo- coccus horse serum all had the same quantitative antigenic properties. There was some evidence that denaturation had occurred during the preparation of the purified antibody solutions.Fractions having the same sedimentation constant found for anti- bacferia.1 antibodies in horse sera were prepared from two samples of presumably normal horse serum in which they occurred in small amounts. One such fraction approached the antibody solutions in antigenic be- haviour ; possibly it represented a so-called “ normal antibody ” to an inapparent sub-clinical infection. The other fraction similarly prepared was quite different in this property and no satisfactory explanation is as yet available. Natural Bacterial Antigens including Immuno-pohpaccharides The agglutination of bacterial cells is generally considered t o involve mainly those chemical groupings situated at the cell surface. Since the antibodies are not directed to the cell as a whole the immunologist has come to ascribe highly specific immune reactions to different components of the bacterial cell.Extracellular products capsular substances flagella and somatic substances have all been allocated definite places in the immunity pattern and the chemical composition of each major component of the cell is now being worked out so that some explanations of specificity in terms of chemical groupings are already forthcoming. In general whole bacterial cells are not perfect antigens particularly in regard t o their power of eliciting potent protective antibodies. It is of course the ideal aim to separate the antigens in a state of purity and it is clear that injection of a whole bacterial cell may mean that one is injecting a wide mixture of antigens some of which niay be highly toxic to the animal body.Since the aim in producing a good antigen is the removal of substances giving deleterious reactions in the body numerous methods for the solubil- isation of bacterial cells have been worked out. Following this it has been possible to separate cell proteins polysaccharides lipoids etc. in essentially purified forin and to determine their antigenic power. Formerly it was assumed that only the proteins were antigenically active but now it is known that mucopolysaccharides mucolipoids and other complexes may be good antigens. Pneumococcus. The existence of pneumococcus types was shown by F. Neufeld and L. Handel 8o who found that antisera which would protect mice against some strains of pneuniococci would not protect them against others. These differences could be shown by serological reactions.The truly remarkable story of the achievements of the Rockefeller Institute group notably Avery Heidelberger Goebel Dubos and their colleagues on the pneumococcus has been told by B. White.8’ Arb. Kais. Gesdhamt. 1910 34 166 293. ‘’ The Biology of the Pneumococcus ” Commonwealth Fund New York 1938. STACEY ASPECTS OF IMMUNOCHEMISTRY 215 The immunological significance of bacterhl polysaccharides may be b i d to have originated by the discovcry of 0. T. Avery and M. Heidelberger B2 which revealed that the mucinous capsular substances of the pneumococci contained polysaccharides which gave to strains of each type their special serological characters. These studies were epoch-making in that they abolished the older view that proteins only were of significance as antigens and introduced a new biological concept namely that of the directive inhence of carbohydrate residues in the antigenic sense.I n pneumococci at least it would seem that the function of the protein has been relegated to the position of a vehicle for carrying the carbohydrate structure which decides the immunological character of the mucopolysaccharide antigen. In 1917 A. R. Dochez and 0. T. Avery83 had isolated from various sources-e.g. cell-free filtrates of pneumococcus Types I 11 and I11 ; human serum ; urine of patients ill with lobar pneumonia ; etc.-a soluble substance which gave a specific precipitin reaction with homologous anti- pneumococcal serum. Zinsser and his colleagues 84 studied the immuno- logical properties of this and various other constituents of the pneumocoacal cell and then in 1923 Heidelberger and Avery gave the first detailed des- cription of the polysaccharide or " soluble specific substance " of pneumo- cocci and thereby established a chemical basis for the interpretation and understanding of many hitherto obscure problems of immunity.The soluble specific substance from the Type I1 pneumococcus was the first to be investigated. The general method of obtaining the substance was t o concentrate an autolysed broth culture of the organism and then precipitate the crude polysaccharide with alcohol or acetone. Purification was effected by fractional precipitation with organic solvents and with ammonium sulphate followed by dialysis removal of protein etc. I n more recent methods as described below care is taken to avoid heat acid or alkali treatment which would degrade the molecule while deproteinisation is effected by shaking with chloroform.Heidelberger and his colleagues s l y 8 5 examined the polysaccharides of Types I 11 and I11 and found marked differences between them. Thus the Type I11 was the soluble salt of a strong acid and contained a negligible amount of nitrogen while the Type I liberated a lower amount of reducing sugar after hydrolysis and contained a relatively high percentage of nitrogen as an essential constituent. Avery and Heidelberger 82 in 1923 were able to state '' The ectoplasmic layer of the cell is composed of carboliydrate-material which is identical in all its features with the type specific substance. On the other hand the endoplasm or somatic substance consists largely of protein which is species and not type specific.This protein is possessed in common by all pneumo- cocci while the carbohydrate is chemically distinct and serologically specific J . Exp. Med. 1923 38 81. 8 3 PTOC. SOC. Exp. Bbl. Med. 1927 37 275. E4 ( a ) H. Zinsser and J. T. Parker J . Exp. Med. 1927 37 275 ; ( b ) H. Zinsser and * 6 For summary see M. Heidelberger PhySioL. Rev. 1927 7 107 ; Chem. Rekw8 T. Tamiya &id. 1925 42 311. 1927 3 403. 216 QUARTERLY REVIEWS for each of the three fixed types. The cell therefore may be conceived of as so constituted that there is disposed at its periphery a highly reactive substance upon which type specificity depends.” More than forty types of capsular polysaccharides are now known and in some the differences in the chemistry of their monosaccharide units and in their physical properties are striking as shown by examples taken from the listing by Boyd.6 The various specific carbohydrates of the pneumo- coccus appear to fall broadly into two groups-the nitrogen-free aldobionic acid-containing group as typified by Types 11 111 VIII etc.and the acetylated nitrogen-containing group as typified by Type IV the XIV and the somatic or group polysaccharide. It was early realised particularly from the failure of the polysaccharides to stimulate antibody formation that the methods of isolation of the specific polysaccharide as used by Heidelberger were very drastic and gave degraded products with molecular weight probably not greater than 10,000. With gentler methods products of much greater molecular weight and with wider serological properties were discovered.Thus J. F. Enders in 1930 s6 dis- covered a Type I polysaccharide substance which appeared to be of a new antigenic type which he termed the “ A ” substance. It is a more powerful antigen than the soluble polysaccharides which for example would not remove all the antibodies from a Type I antiserum. The investigation was extended to obtain analogous substances from Types I1 and I11 pneumococci. Since the products gave rise on injection to homologous antisera which possessed type specific antibodies after the type specific anti-carbohydrate had been removed it was suggested that in pneumococci there existed a type specific antigen or “ agglutinogen ” distinct from the specific poly- saccharide. in 1931 obtained from Type I pneumococcus a product which appeared to combine the properties of the type specific polysaccharide and Enders “ A ” substances It was a poly- saccharide which gave a homologous precipitin reaction with antipneumo- coccal serum but unlike the soluble specific polysaccharides stimulated the production of protective antibodies in mice.In 1933 the problem was clarified by the discovery by 0. T. Avery and W. P. Goebels8 that in the Type I intact cell the specific polysaccharide occurred in an acetylated form ; these workers were able to say that “ so distinctive are the immunological reactions of the acetyl polyssccharide and those of the deacetylated derivative that it is now possible to clarify many of the conflicting views . . . concerning the nature and properties of the specific carbohydrates of Pneumococcus Type I ”.Further extensive studies by these authors and by J. F. Enders and J. Wug9 revealed that the acetylated Type I polysaccharide and Enders “ A ” substance were essentially identical and a t the time appeared to represent very closely in antigenic function the type specific antigen as it occurs in the cell. L. D. Fulton and B. Prescott however presented evidence which challenged A. Wadsworth and R. Brown 86 J . Exp. Med. 1930 52 235. J. Imnun. 1931 21 345. 88 J . Exp. Med. 1933 50 731. Ibid. 1936 60 127. Bull. Johns Hopkim Hosp. 1930 59 114. STACEY ASPECTS OF IMMUNOCHEMISTRY 217 the validity of the theory that the specific antigenic properties of the polysaccharides were due to presence of acetyl groups. M. G. Sevag 91 thought that the acetyl groups might have been intro- duced from acetic acid into the molecule during isolation and devised a very gentle method of obtaining the polysaccharide.This involved dis- integrating the cells in liquid air and coagulating the protein by shaking the mass with chloroform and amyl alcohol etc. By avoiding the use of acetic acid a product was obtained which still contained approximately 8% of acetyl residue. Heidelberger and his co-workers 9 2 ( a 3 h ) then carried out a careful study on the Types I 11 and I11 polysaccharides prepared from autolysed cultures using a modification of Sevag’s method for deprotein- isation. The method is a standard one and gives undegraded products closely resembling the form in which they occur in the pneumococcal cell. In the main such specific polysaccharides are powerful haptens though in the white mouse they can also often act as antigens doses as low as 0.0001 mg.protecting these animals against a thousand fatal doses of virulent pneumococci. The last word has still to be said on the immunising properties of the specific polysac~harides.9~(~~~) In the writer’s view the growing appreciation of the significance of nucleic acids of both ribo- and deoxyribo-types is pointing in the right direction. As already stated Dubos has hinted a t a possible r61e for nucleic acid in the pneumococcal cell and the writer believes that one will need to examine the polysaccharide complexes of cells which have undergone the minimum amount of autolysis during isola- tion. In particular polysaccharide-nucleic acid complexes should now be studied.Chemical Structure of the Pneunwcoccal Po1ysacchurides.-With the excep- tion of the Type I11 polysaccharide all investigations on the various types are pre-structural. It is probably true to say that most constituent mono- saccharides already identified represent only part of the complex molecules. The Type I polysaccharide contains an amino-sugar but presents interesting problems since some of its nitrogenous constituents are still not yet charac- terised. The identification of methylgalacturonoside methyl ester among the products of methanolic hydrogen chloride@ treatment of Type I polg- saccharide is of great interest.93 The writer has identified Z-rhamnose as a constituent of the Type I1 polysaccharide and from both types has obtained a methyl ether by methylation with methyl sulphate and sodium hydroxide.The methylation technique in the hands of Goebel and his colleagues Q4 O 1 Biochem. Z . 1934 273 419. 94 ( a ) 31. Heidelbcrger F. E. Kendall and H. W. Sdierp J . Exp. Med. 1936 64 550; (t) M. Heidelberger nncl F. E. Kendall J . Biol. Chem. 1932 95 127; (c) M. Heidelberger C. M. RlacCleod S. J. Kaiser and B. Robinson J . Exp. Med. 1946 83 303 ; ( d ) M. Heidelberger C. M. MacCleod R. G. Hodges W. G. Bernhard and M. M. De Lapi ibid. 1947 85 227. 93M. Heidelberger W. F. Goebel and 0. T. Avery J . Exp. Med. 1925 4% 701. O a R. E. Reeves and W. F. Goebel J . Biol. Chem. 1941 139 511. 218 QUARTERLY REVIEWS has revealed structures of striking interest in the Type 111 specific poly- saccharide. They &sign the structure (XII). 4-070 CHiOH < - o ~ * ~ c Y I F -0 COzH (XII.) The molecule is made up from cellobiuronic acid units (consisting of glucuronic acid 1 4 glucose) which are mutually joined through the reducing group of the glucose unit to C of the glucuronic acid constituent the whole forming a long chain structure.There would appear to be no doubt that the hexuronic acid constituents in particular the aldobionic residues of these specific polysaccharides account for the cross-reactions among themselves and with other polysaccharides such as those from the plant gums like gum acacia and cherry gum (XIII) from Rhizobium radicicolum etc. With the latter there is a clear relation- ship95> 96 between structure and immunological character which is extended to the Type VIII pneumococcus polysaccharide and t o oxidised cotton (XIV).Q6 - o f q J L o { y j r o < \ ~ j - 2 -o<~:&o~/l~*<-.o& - CHZ*OH COzH (XIII.) CHZ*OH COzH (XIV.) The cross-reaction between Types I11 and VIII antipneumococcal sera gave a met,hod for determining the fate of oxidised cotton 97 (made up for surgical use) when injected intravenously.Precipitin tests showed that about 80% of the injected material disappeared from the blood stream within 24 hours and appeared in the urine within 3 hours of injection. From further quantitative 98 studies on the cross-reaction between Types I11 and VIII pneuinococcal polysaccharides in horse antisera it was 95 M. Stscey and E. Schliichterer J. 1945 776. Q6 M. Heidelberger and G. Hobby Proc. Nut. A d . Sci. 1942 28 516. 97 E. A. Kabat G. G. Hennig and J. Victor Federation Proc.1945 4 93. D8M. Heidelberger E. A. Kabat and M. Mayer J. Exp. Med. 1942 75 35. STACEY ASPECTS OF IMMUNOCHEMISTRY 219 possible to interpret the cross-reaction in terms of similarities and differ- ences in chemical composition of the polysaccharide and to calculate the minimum molecular weights as 62,000 and 140,OOO respectively. Numerous different kinds of anticarbohydrate antibodies could be demonstrated. Dr. B. R. Record (private communication) has made a study of the be- haviour of undegraded specimens of Types I 11 and I11 pneumococcus specific polysaccharides prepared by the writer. He finds the following properties PNEUMOCOCCUS POLYSACCHARIDES __ - - - - -~ ~ - Type. 1 Szo x ~ Dzo x lo7 Molecular (c-0). (c-0). weight. - _I-..-_~- -____ I -~ _ _ 6.5 170,000 500,000 I .. . . I1 . . . I11 . . 1.60 I Sedimentation and Diffusion Data _ _ Frictlonal ratio. - - .- 3.2 6.0 4.3 A departure from linearity in the shape of the molecules was the large frictional ratios. shown by Aldobionic constituents also appear to relate the cross-reactions which are known to occur between the various types of Friedlander’s bacillus and Types I11 and VIII pneumococcus polysaccharides but there is less apparent structural relationship between the dextrans and the pneumococcus polysaccharides. Dextrans are polyglucoses characterised by possessing a large number of 1 6-glucosidic linkages. When nitrogen-free they are not antigenic but claims have been made that they can be made antigenic by adsorption on t o a colloidal carrier such as c~llodion.~~ Dextrans including those synthesised enzymically give sharp precipitin cross-reactions with antisera from pneumococcus Types 11 XII and XX.100 Precise knowledge of the types of glucose linkage in these pneumococcal and dextral1 polysaccharides is needed in order to explain the immunological relationship but since nitrogen-free dextran behaves only as a weak hapten one considers some of the relationship will perhaps be found in the prosthetic groups on which the mucopolysaccharide macromolecular structure depends.The deoxyribonucleic acid responsible for the transformation of pneumo- coccal types may be important in this connection. The antigenic relationship lol which has been observed between the Type XIV specific polysaccharide and certain human red cells is of some practical interest. Type XIV antisera from the horse contains agglutinins against all four main blood groups.These are anticarbohydrates which *s J. Zozaya J . Exp. Med. 1922 55 353. lo0 J. Sugg and E. J. Hehre J . Immun. 1942 43 119. lol W. F. Goebel P. B. Beeson and C. L. Hoagland J . BioZ. Chem. 1939 129 455. 220 QUARTERLY REVIEWS appear to agglutinate the erythrocyte carbohydrates as well as the Type XIV polysaccharide and it is believed that certain fatal reactions observed with use of Type XIV antiserum were due to the unusual cross-reactions. A purified blood-group A polysaccharide removed all the Type XIV anti- carbohydrate and the structural relationship between the A substance and the Type XIV capsular polysaccharide both of which contain galactose and an acetamido-sugar was very close.The antigenic properties of this group have been widely studied particularly by Linton and his colleagues,lo2 and they have been classified into pathogenic and non-pathogenic types. Early studies by K. Landsteiner and P. A. Levene lo3 had shown that a polysaccharide could be extracted from vibrio-strains while Linton isolated three chemically different polysaccharides from vibrio-strains by a method involving extraction with boiling N/ZO-acetic acid and alcohol- precipitation. These polysaccharides like the pneumococcal Type I polysaccharide contained acetyl groups readily split by alkali. They were similar to “ globulins ” and were prepared by extraction with 1% sodium hydroxide neutralisation with acetic acid and repeated precipitation by half-saturated ammonium sulphate. B. N. Mitra lo4 isolated some of the amino-acids from vibrio-proteins racemised by alkali and found differences in the degree of optical activity aome being active and some not.Encapsulated strains of B. anthracis were found by J. Tomcsik and H. Szongott lo5 to contain a substance called “ P ” which was precipitable by copper from broth-cultures. ‘‘ P ” fixed complement with anti-anthrax sera and minute amounts were able t o produce fatal anaphylaxis in a passively sensitised guinea-pig. Non-capsulated strains did not contain “ P ”. G. Ivanovics,106 following up the earlier work of others isolated the capsular substance from capsulated strains of B . anthracis and some related organisms in a state approaching purity the material was shown to have the properties of a hapten giving precipitates with antisera to cap- sulated strains of B.anthracis but failing to induce antibody formation or to confer protection against infection with B. anthraxis when injected into animals. On hydrolysis the capsular substance gave good yields of d( + )-glutamic acid enantiomophous with the I( -)-acid normally encoun- tered in proteins and G. Ivanovics and U. Bruckner 107 concluded that it was a polypeptide built up solely of d(+)-glutamic acid residues. None of the usual proteolytic enzymes will attack it. W. E. Hanby and H. N. Rydon lo8 described methods for the isolation and purification without undue degradation of the capsular substance Cholera vibrios. Vibrio-proteins have also been studied by Linton. Anthrax bacillus. lo2R. W. Linton and B. N. Mitra Indian J . Med. Res. 1937 25 466; 1936 24 lo3 J .Exp. Med. 1927 46 213. Io4 Indian J. Med. Res. 1936 23 573 579. lob 2. Immunforsch. 1933 78 86. lo’ Ibid. 1937 90 304. 323 ; etc. lo6 Ibid. 1940 97 402 443 ; 98 373. lo8 Bwchem. J. 1946 40 297. STACEY ASPECTS OF IMMUNOCHEMISTRY 22 1 from two strains of B. anthracis and showed that there were no chemical differences between the capsular substances obtained from these two strains. The molecular weight of the native material was greater than 50,000 and was thus of the same order of size as many proteins. Structurally the capsular substance is a long-chain molecule made up of or-peptide chains of 50-100 d( + )-glutamic acid residues joined together by y-peptide chains of d( + )-glutamic acid residues. This polypeptide which is homogeneous in the Tiselius apparatus was common to a large number of strains of Gram-positive spore-formiw organisms including B.subtilis and B. mesentericus. Ivanovics has used the quantitative precipitin-methods to measure the antipolypeptide content of rabbit- and horse-antisera. The anthrax bacillus also contains a somatic carbohydrate which has been puri- fied and found to be related to the Type XIV pneumococcal polysaccharide and the blood-group A substance and to contain D-ghcosamine and D-galactose in equimolecular proportion. The r81e of the capsular polypeptide in relation to immunity to anthrax seems rather anomalous. Tomcsik and his co-workers have regularly and repeatedly found that mice could be protected against 20 to 100 lethal doses of virulent anthrax bacilli by injection of rabbit-antisera containing antipolypeptide.However these same sera failed completely in protect- ing guinea-pigs and rabbits. Immunity in the rabbit and in the guinea- pig was found by Ivanovics to be unrelated either to the polypeptide or to the somatic " C " carbohydrate of the anthrax bacillus. From the existing data it seems possible that the immunity produced by anthrax vaccines involves some as yet unknown antigen. Strains of Bact. typhosum occur which contain various antigens the most important being the somatic O-antigens described later. A highly labile antigen-the " Vi "-antigen-was discovered by A. Felix and R. M. Pitt lo9 who found that strains inagglutinable by O-antiserum were more pathogenic for mice than agglutinable strains and that on immunisation with vaccines prepared from these inagglutinable strains antibodies apparently unrelated to the O-antibodies were formed It was found that live cultures could not be agglutinated by O-antiserum and it was suggested that it was the presence of the Vi-antigen which pre- vented this.Antisera containing antibodies (agglutinins) to the Vi-antigen can only be prepared by injection of living strain of Bact. typhosum and the labile nature of the antigens is shown by the fact that the organisms after having been heated at 60" for one hour can no longer be agglutinated by Vi-antiserum but can be agglutinated by O-antisera. Vi-substance is rapidly released into solution when cells are suspended in saline. Extraction methods generally give Vi reactive substance contaminatad with 0-somatic antigens but such material gives reactions which differ from those of the purified 0-antigens and claims have been made to have separated the Vi-material by fractionation with ammonium acetate.It was a mucolipoid apparently less toxic than the O-antigen. The relatively non-toxic nature of the Vi-antigen was demonstrated by D. W. Hender- The Vi-antigen. loQ J . Path. Bact. 1934 38 409. 222 QUARTERLY REVIEWS son,110 who extracted a rough Vi strain which contained Vi-antigen only. The Vi-antigen rapidly loses its power to produce protective antibodies on injection. One important fact resulting from the various studies is that only those strains of Bact. typhosum which possess the maximum amount of 0- and Vi-antigen have the highest degree of pathogenicity of which the micro-organism is capable. There is some evidence that the Vi-antigen is a surface antigen.Further work on the chemistry of this antigen is awaited with interest. Immunological Behaviour of Lipoids The r61e of lipoids in some immune reactions is of considerable prac- tical importance but bhe interpretation of some observed phenomena is complicated by the fact) that lipoids usually occur in admixture with pro- teins and are exceedingly difficult to obtain pure. The relatively unreliable complement fixation method is often used to detect specificity while in flocculation reactions it is difficult to decide whether the lipoid component of the precipitate arises from the antigen or is carried down from solution with some other complex. There have been numerous claims that lecithin and disteargl lecithin are antigens but these lipoids generally need to be injected in admixture with a protein such as serum.Different lecithins injected in this way can be distinguished serologically. From cephalin and cerebrosides antisera have been obtained which give specific complement fixation with the corre- sponding antigens. I n the same way various sterols have been shown to possess different immunological characters but on the whole specificity is much less sharp than with carbohydrates. The most important members in this group are those compounds of high practical immunological signi- ficance as particularly exemplified in the case of the Wassermann substance the reaction of which is employed in the diagnosis of syphilis and as a guide to the progress of therapy. This is the most widely used of all serological tests ; there are also the Forsmann antigen and the Salmonella antigens.While therehis an extensive literature dealing with the immunological proper- ties of such mucolipids chemical knowledge is very scanty. It does appear however that the biological specificity of members of the group is mainly determined by the presence in them of a carbohydrate hapten. The Wassermann substance is prepared by extracting various animal organs particularly beef heart with alcohol and its lipoidal nature was early recognised. M. C. Pangborn ll1 described the preparation of a new L‘ phospholipid ” termed “ cardiolipin ” from beef heart and claimed that it was the essential constituent of the Wassermann substance. On hydro- lysis it gave a fatty acid and a phosphorylated polysaccharide. In a later communication however the carbohydrate constituent is stated to be due to an impurity.A. J. Weil 112 has ably reviewed the literature on the M7assermann and related antigens. llOBrit. J. Exp. Path. 1939 20 1 11. ll1 J. Biol. Chem. 1942 143 247 ; 1944 153 343. 112 Bact. Rev. 1943 5 293. STACEY ASPECTS OF IMMUNOCEIEMISTRY 223 The Forsmann antigens may be regarded as heat-stable substances which when injected into rabbits can evoke sheep-cell hsmolysins. Unlike most serological reactions which are amazingly specific the interactions of these so-called " F " or heterophile antigens shows a high degree of cross- specificity which can now be traced to the presence in them of chemically similar constituents. The " F " antigens are widely distributed in nature being present in the tissues of mammals birds fishes yeasts and bacteria.They have been classified into four groups. Studies on the " F " antigens as they occur in animal tissues suggest that they are carbohydrate-lipo- protein complexes and F. E. Brunius 113 considors that the carbohydrate part contains D-glucosamine units. More knowledge is available on the heterophile antigens from bacteria particularly those in the Gram-negative group. The 0- or somatic antigens are together with the less well-defined Vi-ant'igen undoubtedly the impor- tant protective antigens in this group. The carbohydrate nature of the O-antigens which are endotoxins was shown by J. Furth and K. Land- steiner,ll* who obt,ainod them by dissolving various species of Salmonelk in sodium hypochlorite solution and then precipitating the carbohydrates with alcohol.Several other methods were also used in their preparations. These carbohydrqtes showed cross-reactions in accordance with the distri- bution of somat'ic antigens in tho Kaufniann-White l1 classification and removal of the O-agglutinins with alcohol-extracted bacilli removed all of the precipitin for these carbohydrate-fractions. P. B. White 116 was also able to inhibit agglutination of O-organisms by previous addition of carbo- hydrate. With the O-somatic antigen from Bact. dysenteriz (Shiga) it was possible to demonstrate that on injection into rabbits it would evoke the formation of heterophile antibodies and moreover that only the poly- saccharide-peptide part of the mucolipoid was necessary for this antibody production. It was shown simultaneously by A.Boivin and L. Mesrobeanu 117 and H. Raistrick and W. W. C. Topley 11* that the O-antigens of certain Gram- negative organisms can be isolat,ed in a relatively pure condition and unchanged in their specific immunological properties. The latter authors provided good evidence that though their antigens were essentially free from intact protein they were highly toxic and also efficient immunising agents. Work along similar lines has been carried out by numerous investi- gators. The material of Boivin and Mesrobeanu was highly toxic and contained lipoid and carbohydrate ; the carbohydrate alone was not anti- genic nor toxic but reacted with O-antisera ; the lipoid fraction alone was also non-antigenic. By tryptic digestion and fractional precipitation with alcohol Raistrick and Topley isolated a similar material from Bact.aertryclce. 1 1 3 " Chemical Structures of the True Forsmann Haptene " Stockholm 1936. 114 J . Exp. Med. 1929 49 727. 116 See discussion in " An Outline of Immunity " W. W. C. Topley and G. C. Wilson llS J . Path. Bact. 193-1 39 529 530 ; 1 1 7 Compt. rend. SOC. Bwl. 1934 115 304 309. 118 Brit. J . Exp. Path. 1934 15 113. Arnold London 1946. 1937 44 706. 224 QUARTERLY REVIEWS This phospholipoid material produced somatic Kaufmann-White ‘( 1 2 ” antibodies. Another antigen cross-reacting to Bact. aertrycke Bad. enteritidis and Bact. typhosum was also found in this fraction. The preparations yielded about 35% of reducing sugar on hydrolysis and contained N 3.6 to 6.5% ; P 1.8 to 3.4% ; and S 0.69 to 1.03y0. Raistrick and Topley believed it to be essentially similar to that described by Boivin but also thought it contained some material of a peptide nature.The various methods which have been used to extract the O-antigens have involved extraction of cells with trichloroacetic acid trypsin diethylene glycol phenol guanidine etc. followed by precipitation of the antigens by alcohol or acetone. All methods appear to give closely related complexes of high molecular weight. The Bact. typhimurium (a~rtrycke) somatic ant>igcn contains a complex constituted of four components a specific polysaccharide a polypeptide an acetyl polysaccharide and a pho~pho1ipoid.l~~ These constituents which are produced on gentle hydrolysis have been examined in some detail by Freeman and his coll,ihorators.120 The complex contained 69yo of poly- saccharide 16% of conjugated protein 3-40; of lipoid and so/ of an alcohol-soluble polysaccharide.Thc alcohol-insolubl(~ polysaccharide which readily combined with protein was considered t o be built up from units of D-glucose (19yo) D-mannose (2 L.Fio/,) and D-galactose ( l O Y & ) . The Bad. typhosum antigen is a similar complex containing 50-60y0 of a poly- saccharide ([.In + 114” in water) of an insoluble polypeptide 10-ZO~O of a soluble nitrogenous constituent and 3-4% of a lipoid component. The polysaccharide yielded n-galactose n-manno~e and D-glucose on hydrolysis. It appears that the Bact. typhosum polysaccharide contains D-glucosamine also as a constituent unit. A. J. Weil‘s I2l review surnmariscd the recent immunochemical work on the problem of dysentery. Most of the studies have been carried out with the Shiga bacillus the somatic antigen of which has bmn prepared in an actively antigenic forin by the same methods as those described for the isolation of the Salmonella O-antigens i.e.extraction with trichloro- acetic acid or diethylc.nc glycol. The purified products have regmibled the salmonella O-antigens and have been shown to consist of toxic phospholipoid polysaccharide-protein complexes. W. T. J . Morgan and S. AT. Partridge 12z have bmn able to separate the phospholipoid from the polysnccharide- protein complex and to demonstrate that the polysaccharide-protein com- plex alone was both antigenic and toxic. The polysaccharide-protein complex could be dissociated with 90% phenol to give the non-antigenic polysaccharide with a high viscosity.It could however be recombined with the protein part in slightly alkaline solution to form a (‘ complete ” 119 M. Stacey S. W. Chnllinor nntl 13. Raistricllc Proc. I n t . Cong. Microbiol. 1937 333. 120 ( ( 1 ) G. G. Frocmnn Biocheiri. J . 1943 37 601 ; ( b ) G. C. Freeman S. IV. Chdlinor and J. Wilson ibid. 1010 34 307 ; ( r ) G. (2. F~*cemnn ant1 T. H. Anclcrson ibid. 1911 35 561. lZ1 J . Intmun. 1943 46 13. lZ2 ( a ) Biochenz. J . 1940 34 169 ; 1941 35 1140 ; Brit. J . E s p . Path. 1910 21 180; 1942 23 8 4 ; Ghem. and Ind. 1041 60 72.2. STACEY ASPECTS OF IMMUNOCHEMISTRY 225 antigen. The protein part was a conjugated protein very similar to that obtained from the O-antigens of other SaZmoneZZu. Morgan and Partridge 122 showed that agar gum acacia and the blood-group A substance could be combined with it to produce immunising antigens.Rough variants of Bact. shigz yielded negligible amounts of these complex antigens on extraction with diethylene glycol but similar materials were obtained from other varieties of dysentery bacilli (Flexner). The dysenterg-antigens have not yet been characterised as to homogeneity in the Tiselius apparatus or in the ultracentrifuge nor have detailed quantitative precipitin-studies been carried out to determine the relation of the isolated antigenic materials to the intact cell. The r61e of phospholipoid polysaccharide and conjugated protein in the reaction with antiserum also have not yet been studied quantitatively. W. T. J. Morgan 123 carried out some quanbitative pre- cipitin studies on the amount of antibody precipitated by the dysentery polysaccharide and by the polysaccharide-protein complex from antisera to the polysaccharide-protein complex.Both materials precipitated the same aniqunt of antibody from the sera within the error in determining the point of maximal precipitation ; about 40% as much polysaccharide was needed for complete precipitation of antibody as was needed of the polysaccharide-protein complex. Numerous complexes from Bact. &sen- terize (Shiga) have been the subject of very extensive studies by Morgan and his colleagues. The non-toxic Bact. shiga O-hapten polysaccharide- component yielded on hydrolysis D-galaCtOSe L-rhamnose and N-acetyl- D-glucosamine. The toxicity of these antigenic preparations appears to be largely due to the integrity of undegraded complex.Examination of the individual components revealed many interesting facts. Thus with Buct. shigz the specific polysaccharide failed to induce any demonstrable anti- bodies while the polypeptide component engendered homologous precipitins of low titre but no agglutinins against Bact. shigze ; however the “ recon- stituted ” polysaccharide-polypeptide complex 122 (made by mixing the two substances in formamide) induced the formation of specific immune serum of high titre. W. F. Goebel F. Binkley and E. Perlman 12* have confirmed and extended some of Morgan’s findings on B. dysenteriz (Shiga) From a practical point of view it is important to note that purified somatic antigens from typhoid and paratyph0id-~4 and -B organisms pro- duce in humans local and general reactions which are less severe than those produced by the stttnclard T.A.B.vaccine but also give equally good protective antibodies. A most important property shown by Grarn-negative O-antigens is their power to produce tumour haemorrhage in animals with transplantable tumours and also to give a haemorrhagic action on the placenta and thereby interrupt a t certain stages pregnancy in mice. It is considered that the substance accounting for the toxicity of these endotoxins is also responsible for the haemorrhage induction in tumours. The most striking results have (Type V). 123 Biochem. J. 1937 31 2003. 124 Science 1944 99 412; J . Exp. Med. 1945 81 315. 226 QUARTERLY REVIEWS been obtained by Shear and his colleagues 12s with a mucolipoid from S. marcescens which may be related to the somatic antigen described above.Goebel and his colleagues 126 have examined the group specific or C-carbo- hydrate of the pneumococcus discovered by W. S. Tillet and T. Francis junr.,127 who showed that it possessed a high phosphorus content and that it exhibited no type specificity but would give a precipitin reaction with individuals ill with most types of lobar pneumonia and also with some patients ill with streptococcal and staphylococcal infections. W. F. Goebel and M. H. Adams 12* have differentiated between the pneumococcus C-substance and the F or heterophile antigen inasmuch as the latter is shown to contain 6% of a fatty acid which is liberated on hydrolysis. Further the heterophile antigen can stimulate in rabbits the production of precipitins antibodies and sheep-cell hamolysins whereas the C-substance is non-antigenic in rabbits.It would appear that the heterophile antigen consists of the C-substance in combination with a fatty acid. Im?rzunologicwl Behaviour of Acid-fast Organisms It has been long recognised that the occurrence of considerable amounts of specific lipoidal substances in the cells is a characteristic property of the acid-fast group of organisms and that consequently these organisms are endowed with unique metabolic processes. The processes have a direct bearing on the antigenic structure of the bacilli. The problems of the lipoidal components have been investigated at length by It. J. Ancior~on,~*~ who found that in addition t o phosphatides the bacterial cells contained wax-like substances which were described as esters of polysaccharides combined with characteristic branched-chain fatty acids.Earlier workers have established the fact that injection into experi- mental animals of thoso wax-like substances or of the characteristic fatty acids caused the for.mat,ion of typical tubercle lesions and that the phos- phatides also caused severe toxic symptoms. Whether or not such lipoidal substances can entor into antigen-antibody reactions has not definitely been established though some evidence is available to suggest that such a possibility exists. The chemical at'ructure of two notable fatty acids from M . tuberculosis has now been determined. Tuberculostearic acid was found to be 10-methyl- stearic acid ; this has bwn synthesised chemically and the acid so obtained possessed physicit1 arid biological properties apparently identical with those of the naturally occurring coinpound.Tho acid-fast property of this group of bacteria has been assigned to a lipoidal constituent (mycolic acid) the structure of which is not known. It possesses wcuk optical activity ( [ a ] D - 4.8") and has the iriolecular 136 ( a ) M. J. Shoar I?. (!. 'l'urmr A. Perrault imd T. Shovelton J . Not. Cancer Inst. 1943 ; (0) J. L. Hiirtwcll RI. J. Shear J. R. Adttms junr. A. Perrault ibid. 4 81. See also M. J . Slioar and nunwrous colleagues ibid. 1946 6 488 489 490 491. 12* W. F. Goobel 7'. Sliecllu;.sky S. I. Lttviii uiitl &I. H. Adams J . Biol. Chem. 1943 148 1. 1*7 J. Exp. filed. 1030 52 561. lZ8 J . Exp. Med. 1943 77 435. 1 * 9 Chem. Reviews 1941 29 225. STACEY ASPECTS OF IMMUNOCHEMISTRY 227 formula C&&& or C&7@3.Dubos has pointed out however that the acid-fast staining properties of these organisms probably depends on factors other than the mere existence of mycolic acid. There is good evidence that carbohydrate-lipoid complexes play an important part in the serological reactions of acid-fast micro-organisms. Thus M. Heidelberger and A. hlenzel 130 have prepared from the somatic part of the tubercle bacillus several specific polysaccharides which were shown t o be constituted of D-mannose D-arabinose and D-galactose units. Some of the polysaccharides were esterified with palmitic acid. The pres- ence of D-glucosamine in tubercle polysaccharides has been reported by R. J. Anderson.131 A hapten polysaccharide from the somatic part of M . tuberculosis has been investigated in some detail 133 and has been shown to possess a structure of the type shown in (XV).R = L-Rhamnopyranose ,4 = n-Ai-abofnranose M = D-Mnnnopyranose Tu' = Glucosamine (possiblc 1 2 1 2 1 " posit ioii) 2'yf 1 1 2 6 1 A = D-habofuranose G = D-Galactopyranose M = D-Mannopyranose A-M -G--A A-M-A N = Glucosamine ' I N (XVI.) [Tlie numerals show the points of attachment.] It has also been suggested that the wax from the tubercle bacillus may contain fatty esters of carbohydrates. A polysaccharide associated with the " n-axes " of M . tuberculosis has also been investigated structurally ; 133 D-niannose u-arabofuranose D-galactose and glucosamine were identified as sugar units coinprising the polysaccharide molecule which it was considered has a highly branched structure (XVI). The polysaccharide behaved as a hapten and in the precipitin test with homologous serum reacted in a dilution of 1 2,000,000.The high antigenicity of proteins of both intra- and estra-cellular origin derived from acid-fast types has been long known. The protein constituents of cell-free filtrates of M . tuberculosis have been rigorously investigated by 3'. Seibert,134 who showed that whilst the proteins werc frequently present 130 J . Biol. Cherr-L. 1937 118 79. 131 Ann. Rev. Tuber. 1930 22 664. 132 W. N. Haworth Y. W. Kent and M. Stacey Abs. Chicago Meeting. Amer. 133 Idem J . Amer. Chem. SOC. 1947 in the press. 134 Chern. Reviews 1944 32 107. Chem. SOC. 1946. 228 QUARTERLY REVIEWS as nucleoproteins or in association with specific polysaccharides it was only the proteins (in particular the " P.P.D." or purified protein derivative) which possessed antigenic power and were able to elicit skin-sensitivity reactions upon subcutaneous injection.In the precipitin test the reaction with tuberculin proteins and antiserum was not sufficiently specific for immunological investigations. By use of the ultracentrifuge and the Tiselius electrophoresis technique Seibert and Nelsonl35 were able to separate crude tuberculin proteins into two protein fractions '' A " and " B " a poly- saccharide and deoxyribonucleic acid. Proteins " A " and " B " were both antigenically active (" A '' rather more than " B ") and were both found in culture filtrates of human and bovine strains of M. tuberculosis but in avian strain preparations only a single protein constituent was observed.Certain immunological differences between " A " and " B " regarding pre- cipitin titre and skin sensitisation in homologous or heterologous reactions in sensitised guinea-pigs were observed. Earlier work of Seibert l36 showed the existence of certain specificity- differences between the proteins of the various strains of M . tuberculosis and M . phlei though no marked difference in chemical properties could be detected. Numerous complex fractions showing varSfing degrees of anti- genic activity have been isolated from other acid-fast types. It is claimed that a complex isolated from defatted cells of M . tuberculosis exerfed a stabilising effect on the progressive disease in experimental animals. The nature of this complex was not established but it seems probable that it was a phospholipoid or lipoprotein.Further evidence of the part played by lipoidal constituents of M . tuberculosis in antigenic reactions has been adduced by means of the complement -fixation reaction. A lipo-poly- saccharide antigen capable of reacting with immune serum has been des- cribed by Siebert. Nucleoprotein complex antigens were studied bg R. D. Coghill 13' who isolated these substances by extraction of lipoid-free tubrcle bacilli with water and dilute alkali. The two fractions so obtained Wered in both chemical and antigenic properties. The water-soluble complex was an albumin containing a considerable amount of basic amino-acids and having pronounced " tuberculin '' activity ; the alkali-soluble substance was devoid of such tuberculin activity and had a nitrogen content much greater than that of the former complex.M. Heidelberger and A. Menzel 13* confirmed the view of Coghill that the water-soluble and the alkali-soluble complex were serologically distinct since antisera to these fractions showed only slight cross-reactions with heterologous antigens. Similar nucleoprotein fractions have been isolated from avian tubercle bacilli and M . phlei and a considerable degree of immunological specificity was detected as compared with absence of specificity found in nucleo- proteins from human and bovine strains of M . tuberculosis. There had been isolated a water-soluble nucleoprotein from Bacillus Calrnette Guerin though E. Chargaff and W. Schaefer 139 failed to obtain from this source 136 J . Amer. Chem. SOC. 1943 85 272. 138 Ibid. 1934 104 655. J . Biol.Chem. 1933 101 763. 13' Ibid. 1921 70 439 449. Ibid. 1936 112 393. STAGEY ASPECTS OF IMMWXOCHEMISTRY 229 any water- or alkali-soluble complexes corresponding to those obtained by Heidelberger and Menzel from other members of the acid-fast group. Such nucleoprotein complexes did in general exhibit tuberculin activity. Although deoxyribonucleic acid has been identified in many complex fractions from acid-fast bacteria no immunological significance has yet been ascribed to it. I n all these cases antigenic fractions have been assessed by tuberculin activity (i.e. skin-sensitivity reaction) precipitin titre or modifications of these tests depending on circulating antibodies. Indications exist that tissue antibodies may play a fundamental part in tuberculosis infection but as yet this aspect has been little developed.Immunological Properties of Nucleic Acids and Nucleoproteins Nucleic acids and nucleoproteins were a t one time thought to occur mainly in the somatic portion of bacterial cells but now it is clear that in Gram-positive bacteria the ribonucleic acid occurs in the Gram complex which is mainly located at the cell surface. M. Heidelberger and F. E. Kendall 140 extracted nucleoproteins from defatted cells of a scarlatina1 strain of Sir. h~molyticus by alkali at room temperature a t pH between 6.5 and 11 after a preliminary extraction of defatted ground organisms with acetate buffer a t pH 4 to remove carbo- hydrate. This procedure was based on the assumption that the proteins in bacteria might vary in their acidic strength. Each protein was further purified by repeated isoelectric precipitation.The fraction extracted at pH 6.5 was a labile dextrorotatory nucleoprotein having a high phosphorus content and from which a considerable amount of nucleic acid could be split off. The labile nucleoprotein reacted readily with sera of patients suffering from different streptococcal infections. Certain type-specific fac- tors were found in the viscous residues which did not pass through Berke- feld filters. All fractions contained " C " carbohydrate substance bound to the protein in an antigenically active form. Those fractions extracted a t successively more alkaline pH were more lanorotatory and contained less phosphorus. M. Heidelberger and H. W. Scherp 141 subsequently found that only a small portion of the nucleic acid in some fractions could be removed by ammonium sulphate fractionation or with barium acetate.S. Mudd and M. Wiener l42 confirmed an observation by Heidelberger and Kendall that hEmolytic streptococci group A contained nucleoproteb more closelg related to those of the pneumococci than to those of streptococci group B. From the immunological point of view the classical work of Mrs. R. Lancefield 143 has revealed a good deal regarding the antigenic picture of the hamolytic streptococci. The significance of the chemistry of the nucleo- proteins in these studies is now being revealed by Mudd and Zittle and their colleagues in numerous papers in the Journal of Immunology. Lance- 140 J . Exp. Med. 1931 54 515. 142 S. Mudd and M. Wiener ibid. 1942 45 21. 14s Harvey Lect. 1941 XXX 36 251. 1 4 1 J .Immun. 1939 37 563. P 230 QUARTERLY REMEWS field discovered several kinds of streptococcal " nucleoprotein ' ' (Le. materid containing both nucleic acid and protein); one was group-specifk (the " P "-protein) and one was type-specific (the " M "-substance). The M-substance was originally obtained by Lancefield by hot hydro- chloric acid extraction and was rapidly destroyed by pepsin and trypsin. It is an antigen which gives rise to type-specific protective antibodies ; it is absent from the sobcalled " glassy " forms of hamolytic streptococci but is present in all mucoid and matt forms irrespective of their origin. The chemical and physical properties of the M-substance have been extensively studied by Zittle and his colleagues and its nucleic acids and protein com- ponents separated and examined.The properties of these were very similar to those of other bacterial nucleoproteins. C. S. Zittle and F. Seibert 144 showed that in the M-substance " nucleoprotein " the niicleic acid was free and its mobility was sharply similar to that of both yeast and thymus nucleic acids. In the streptococcus group Lancefield 145 has made observations of fundamental importance which illustrate well the highly complex nature of immunity phenomena. Thus she has shown that both matt and glossy strains of hamolytic streptococci the latter of which do not contain the M-substances and therefore do not give rise to protective antibodies con- tain another antigen (the " T "-antigen) which gives rise to antibodies responsible for the type-specific agglutination of hamolytic streptococci.Such " T "-antigens may be used for type classification of streptococci but do not stimulate the production of any protective antibodies. Thus it is clear that mere agglutination of bacterial cells by an antiserum as in tuberculosis does not necessarily mean that the serum has any protective antibodies in it. There have been numerous observations which would ascribe an immuno- logical r61e to the nucleic acids themselves. Heidelberger and his col- leagues have noted that nucleic acids from acid-free bacilli could be pre- cipitated with homologous serum and that certain streptococcal nucleic acids could be precipitated by antisera though they probably regard such reactions as being due to impure preparations. Other workers deacribe a specific reaction with nucleic acid and its degradation products in certain ragweed pollen sensitive individuals.A serological study was made of various types of nucleic acid by Lack- man and others.146 The samples used were ( a ) a streptococcal ribonucleic acid (N 156% ; P 8.87y0) ( 6 ) a streptococcal nucleic acid containing both ribo- and deoxyribo-types (N 16-2y0 ; P 9.12yo) ( c ) yeast nucleic acid (N 14.6% ; P 8.0y0) ( d ) nucleic acid from the thymus gland (not analysed) (e) nucleic acid from the tobacco mosaic virus (not analysed) and (9) nucleic acid from bull sperm (not analysed). The nucleic acids were set up with rabbit sera against streptococci pneumococci and acid- The liberated protein was only a weak antigen. 144 J . Immun. 1942 43 47. lo6 (a) R. C. Lancefield J . Exp. Med. 1943 78 465; ( b ) R. C. Lancefield and W.A. Stewart ibid. 1944 79 79. 14* D. Lackman S. Mudd M. G. Sevag J. Smolens and M. Wiener J . Immun. 1941 40 1. STACEY ASPECT$ OF IMMUNOCHEMISTRY 231 precipitated pneumococcal nucleoproteins and with horse sera against streptococci pneumococci and Jeveral other organisms. Also various hydrolysis products of nucleic acids were tested for their power t o inhibit the specific reaction between nucleic acids and horse antipneumococcal sera. The general conclusions were that nucleic acids precipitate specific- ally with certain antisera particularly horse antipneumococcal sera. The specific antibodies appear t o be located in the euglobulin fraction of the serum and the specific reaction is very sensitive to pH change. The specific reaction can be inhibited by purine nucleosides purine nucleotides or purine bases.Pyrimidine bases show weak inhibition only while pentoses and phosphates give no inhibition. In his recent book Dubos hints a t a possible antigenic r61e €or nucleic acid when combined with specific capsular polysaccharide. Thus he states that the pneumococcus capsular antigen possesses a complex structure one part of which is present in all pneumococci while another part present only in encapsulated cells varies in composition from one type to another and confers upon each type the immunological specific character of the pobsaccharide. He points out that all types of pneumococci whether encapsulated or not contain an enzyme capable of inactivating the capsular antigen of all types without destruction of the polysaccharide. This enzyme has an action similar to that of the ribonucleinase of animal tissues which also can destroy the activity of the capsular antigen and which decreases the affinity of the cell for basophilic dyes.Moreover the capsular antigen is extremely resistant to proteolytic enzymes so that it may well be that it consists of a ribonucleic acid in combination with a specific polysaccharide. Such complexes have not yet been studied in any cells. In view of the increasing interest in bacterial nucleic acids of both ribo- and deoxyribo-types-e.g. Avery Macleod and McCarty 68 ; H. Henry and M. Stacey 147-it is clear that immunological studies will need to be made on undegraded nucleic acids and nucleoproteins in order to deter- mine something of their possible specific nature. The immunology of viruses will need t o be related to their nucleoprotein struct,ure.Anaphylaxis Allergy and Sensitisation It has often been observed that an animal when injected with a non- toxic innocuous protein may actually become highly suscept#ible to further injections rather than immune to it. Similarly an animal may be immunised against an infective disease agent and although resistant to subsequent infection by the living micro-organism may be less resistant than the normal animal to injection of some substances derived from the infective agent. These facts manifest themselves in immunological phenomena which are of high practical significance and include various states of hypersensitivity especially in humans. The most extreme case is manifested in the biological reaction known as anaphylaxis while other sensitised or allergic states such as asthma hay fever skin rash serum sickness etc.are well known in man. When an animal is given a small “ sensitising ” dose of an antigen 14’ Proq. Roy. SOC. 1946 B 133 391. 232 QUARTERLY REVIEWS followed 10 days later by a second (" shocking ',) injection (often a minute amount) of the same antigen it may pass into an anaphylactic state the outstanding symptom of which is a bronchial spasm. This in a sensi- tive animal like the guinea-pig leads to death from asphyxia. All antigens can under certain conditions produce these specific effects and in true anaphylaxis there is a complete absence of protection ; this is an express recognition of the fact that the " sensitivity " of the animal is more important than the toxicity of the antigen in determining the reaction.Many of the distressing allergic states in man originate from a hyper- sensitivity not necessarily due to infection but often to the effect of a foreign protein. Patients suffering from these complaints find that they have acquired an altered way of responding to certain everyday substances. Individuals suffering from hay fever are highly sensitive to pollen proteins ; others may react to dandruff and hdr of certain animals or t o certain food proteins and the symptoms often involve attacks either of asthma analogous to the bronchial spasm of guinea-pigs or of urticaria1 skin eruptions. An exactly similar state of hypersensitivity can originate not only from con- tact with a protein but also from continuous contact with simple chemical substances.The condition sometimes arises through slow absorption of the chemicals through the skin as with workers in laboratories and explosive factories so that the effect is manifested on the skin by contact with the provoking chemical-the " allergen " Other patients develop an increased susceptibility to drugs i e . a " drug allergy " and the reaction is caused usually by some sharply determinant group in the chemical molecule. Nearly all the exciting agents are themselves non-antigenic and their effects are due to their conjugation with proteins. All cases of allergy have an immunological bassis but there is very little knowledge concerning the nature of the " niitibodies " involved. One thing does seem clear and that is that these antibodies do not circulate in the blood stream but attach themselves to the tissue cells in highly specialised 1ocalit)ies such as the bronchial tract in the sensitised guinea-pig.Land- steiner has made important contributions to this study by using e.g. picryl chloride (XVII) and several important facts were revealed by his work for example (a) intradermal injection of the substance into guinea- pigs leads to the production of skin sensitisation and anaphylaxis but not t o precipitating antibody production (b) injection of picryl chloride intra- peritoneally does not lead to skin sensitivity when there bas been no contact of picryl chloride with the skin ( c ) injection of picryl chloride in combina- tion with a protein does not lead to skin sensitivity but does give rise to specific precipitin formation. Landsteiner's studies were carried a good deal further by Harington and his colleagues 148 who have studied the antigenic function of simple chemical compounds.The most important of these was trinitrophenyl- methylnitroamine (" tetryl " ; XVIII) which during the recent war caused a high incidence of contact dermatitis among workers in explosives factories. 14* P. G. H. Gell C. R. Harington and R. P. Rivers Brit. J . Exp. Path. 1946 27 267. STACEY ASPECTS OF IMMUNOCHEMISTRY c1 233 ozNoNo2 NO (XVII.) (XVIII.) Various methods were worked out whereby guinea-pigs were sensitised to tetryl and these investigations were conducted on a wide series of com- pounds of structure related to that of tetryl in order to decide which compounds or groupings could elicit a similar skin reaction. The results clearly indicate rather surprisingly that the actual haptene grouping was the 2 4 6-trinitrophenyl residue which also would arise from the reaction of picryl chloride with a protein.This conclusion was confirmed by the following facts (a) tetryl reacts readily in vitro with amino-compounds to form 2 4 6-trinitrophenyl derivatives ; ( b ) good skin reactions were obtained in tetryl-sensitised guinea-pigs on application of picramide di- and mono-methylpicramide or 2 4 6-trinitrophenyl derivatives ; (c) unlike picric acid 2 4 6-trinitrophenetole was active in eliciting the skin reaction. B. Landsteiner and M. W. Chase 149 had transferred anaphylactic anti- bodies by passive sensitisation of normal antibodies and now Gell Haring- ton and Rivers 14* have endeavoured to bridge the gap existing between the action of skin-sensitising compounds in causing ‘‘ fixed ” or localised tissue antibodies and the action of those sensitisers when combined with proteins in stimulation of circulating antibodies.In general it seems fairly well established from P. G. H. Gell’s work 1m that the presence of circulating antibodies in the blood specifically directed against the sensitising group has no effect whatever on the skin sensitivity produced by intradermal injection of this sensitising group. The work has emphasised the growing appreciation of the fact that under certain con- ditions an animal’s own proteins may be rendered antigenic and so stimulate production of “auto-antibodies”. From a study of the injection skin test and precipitin response of a wide range of compounds including iodine trinitrophenetole and 4-( 3’ 5’-di-iodo-4‘-acetoxybenzyl)-2-methgloxazolone which are known to react strongly with proteins it was possible t o show for the first time that precipitins can be formed in response to the injection of simple chemical compounds which apparently cause homologous proteins to be rendered antigenic in vivo.Thus all the compounds studied when injected intradermally produced skin sensitivity and an anaphylactic shock and when injected intraperitoneally gave rise to the formation of specific precipitins. The simple sensitising compounds were designated as “ pro- antigens ” for i t wag considered that they acted upon a body protein to give a furl protein antigen. The knowledge obtained constitutes a funda- 14s (a) J. Exp. Med. 1937 66 337 ; 1940 71 237 ; 1941 73 431 ; ( b ) Proc.SOC. Exp. Bwl. 1942 49 668. lSo Brit J. Exp. Path. 1944 25 174. 234 QUARTERLY REVIEWS mental approach towards the systematic attack on which a solution of these difficult problems must depend. Anaphylaxis Desensitisation.-When an animal recovers froin a shock- ing dose of an antigen it is in a refractory state and is no longer susceptible to similar shocking doses. In this state which is temporary the animal is said to be " desensitised ". In order to avoid or postpone the sensitised state injections of the antigen can be given in moderate quantities just before the usual development of hypersensitivity. Other methods make use of procedures which allow gradual penetration to the sensitised tissues and also methods by which minute doses too small to sensitise are adminis- tered.During the refractory state it may well be that tissue antibodies are saturated for the time being with antigen thereby blocking the effect of further amounts. One can often stimulate a non-specific type of desensitisation bg injection of large quantities of either related or unrelated antigens. Desensitisation has also been obtained by injection with histamine. In general however anaphylaxis is highly specific for after sensitisation to one antigen an animal reads only to this antigen or to a very closely related one. More- over after desensitisation with an antigen an animal will still remain reactive to other antigens. This fact is made use of in detecting antigen particularly some polysaccharide haptens. Treatment of '' Atopic '' AZlergic States by Desensitisation Methods.-Some types of hypersensitivity in humans are hereditary and are known as atopic diseases ; their treatment is of great importance.The most obvious method is to avoid all contact with the allergic-producing substance the " allergen ". This cannot always be done especially in cases of sensitisation by pollens and other substances which are frequently inhaled. The usual method of specific desensitisation is to introduce very small amounts of the allergen at frequent intervals until the hypersensitive state is diminished. In the past mixed pollens for injection have been much used but there is a ten- dency now to use a single member of a plant group for desensitisation. A good deal of activity is going on in order to determine the chemical nature of the determinant groups in natural allergens.Pollens appear to contain polysaccharide and nucleoprotein but the work on the subject is still only in its infancy. Ragweed pollen and cotton seed provide useful substances for investigating hypersensitivity of the allergic type. In an investigation on cotton-seed hypersensitivity a mucoprotein which appeared to contain all the allergenic factors 151 was isolated from cotton seed. The procedure necessitated aqueous extraction heat and fractional precipita- t'ion with solvents and basic lead acetate. It was possible to remove the carbohydrate from the complex by use of picric acid and chromatography including adsorption and elution from alumina. The active allergen was a peptide-like substance and gave powerful anaphylactic reactions in guinea- pigs.By electrophoresis of the picrates of this allergen two different specific 151 ( a ) J. R. Spies and E. J. Coulson J . Amer. Chem. SOL 1943 65 1720 ; ( b ) J. R. Spies E. J. Conlson and H. Stevens J . Immun. 1941 41 875 ; 1943 46 347 367 3 7 7 ; and later papers. STAGEY ASPECTS OF IMMUNOCHEMISTRY 235 substances were separated from each electrode end the anodic fraction being more specific in sensitisation. Acid treatment of both fractions did not remove the power to elicit a skin reaction in cotton-seed sensitive individuals but did reduce their capacity to sensitise and shock guinea-pig$. The mucoprotein appeared to be contained in cotton-seed globulin though it is likely that cotton seed contains numerous other allergens. A peptids- like sensitising compound has been obtained from ragweed pollen.Antitoxins In some respects there is a sharp distinction between antibacterial and antitoxic immunity though both phenomena have some manifestations in common. In the former we are concerned with the reactions of the body in attempting to eliminate or a t least prevent the propagation of the invading organism. I n addition to the immune reactions and the action of complement previously outlined the body reactions concern also the lytic action of the phagocytes. In studying antitoxic immunity we are required to determine the factors which will neutralise the toxic products of the agent of disease or the toxicity of the injected chemical. The antitoxic immunity appears to involve only the reaction between toxin and ant'itoxin. It is direct' in its action so that when the toxin is present in low concentration the antitoxin is generally highly efficient as with snake antitoxins.The necessity of having available suitable antitoxic sera against the venoms of the various poisonous snakes has stimulated numerous laboratories to produce uni- and multi-valent " antitoxins " directed against the snakes particularly prevalent in the respective coun- tries. Among the more important distributing centres are the Pasteur Institute in France the Butantan Institute in Brazil and the Antivenin Institute of America in Pennsylvania. In general the neutralising activity of an antisnake serum prepared against the venom of one poisonous snake is sharply specific and only with those snakes which are closely related zoologically is there any overlap in cross-reactivity.Thero does not appear to be any international agreement on methods of standardisation of the antivenoms but in general the com- mercial preparations provide a valuable antidote to snake bites and there seems to be no doubt regarding the efficiency of serum therapy. Antitoxic immunity is often insufficient to stem the invasion of some micro-organism such as haemolytic streptococci and other curative methods have to be attempted. Gay 2(b) has discussed critically the merits of vaccination in the cure and prevention of disease while Jungeblutt 2(b) reviews the proven efficacy of serum prophylaxis and serum therapy. The situation is very encouraging but much remains to be done on the chemical side. It is highly likely that the capacity of the body to produce antibodies and antitoxins is not unlimited.Since bacteria contain a great variety of antigens which are useless for protective antibody production it is apparent that in order to assist in the economy of the antibody-synthesising mechan- ism one will need to isolate purified antigens for use in immunisation. 236 QUARTERLY RHVIEWS Species Specificity and the Blood Groups As a result of many years of serological inquiry on the behaviour of natural components of all parts of the body Landsteiner came to the important conclusion that there existed two systems of species specificity in the animal kingdom-the specificity of proteins which might undergo gradual changes in the course of evolution and the specificity of cell hap- tens which could undergo rapid profound change.The antisera against tissues display " organ " specificity and " species " specificity and despite the immense wealth of immunological knowledge in this field chemical knowledge lags very far behind. One fact that stands out is that chemical degradation of any giant molecule can alter profoundly its antigenic proper- ties. The extreme case is that of the degraded protein gelatin which is not antigenic. An important factor regarding the antigenic behaviour of tissue antigens is the correlation with natural biological function. Those proteins which display the highest immunological specificity are those which play the most important rhle in metabolic process e.g. blood proteins. All the blood proteins of an animal can be differentiated serologically from one another and moreover corresponding blood proteins of different but closely related species can also be sharply distinguished from one another.When there is qualitative overlapping between such closely related species as the dog and the wolf or man and the anthropoid ape the precise methods of quantitative immunochemistry readily differentiate them. Structural proteins such as keratins and storage proteins such as casein do not shorn such a sharpness of immunological specificity. They generally give rise to antibodies but these give a wide range of cross- react'ions with similar proteins from unrelated species. The protein hor- mones such as insulin which possess a highly specialised biological function do not behave as antigens unless altered chemically. Blood cells of closely related species can be distinguished either by quantitative agglutination estimation or better by absorption experiments.Work in this field has been of the highest significance in blood transfusion and in legal and forensic medicine and has been ably summarised by M. Wiener.lS2 Basis of the Four Classical Groups.-There are four main blood groups 0 A B and AB which can be distinguished because the group-specific substances (sometimes termed " agglutinogens ") A and B may be present in the erythrocytes either singly or together or may both be absent. In the normal human serum there are naturally present isoant'ibodies or " agglutinins " which can also be prepared artificially in rabbits for example by injection of the appropriate human erythrocytes. The A cells on injec- tion give rise to serum containing a (or anti-A) agglutinins while the B cells likewise give /3 (or anti-B).In normal human serum these a- and p-agglu- tinins are regularly distributed and in general the serum contains the agglutinin for the absent A or B factor. The combinations are shown in the following table. ss2 I' Blood Groups and Transfusion " Thomas Baltimore 1943. 237 Of the whole blood. STACEY ASPECTS OF IMMUNOCHEMISTRY st~hz y :k Agglutinin in the serum. red cells. o . . . . . A . . . . . B . . . . . AB . . . . - A B- B a- A + B - a (anti-A) + #I (anti-B) In routine blood grouping reactions occur as shown in the following table Serum. 1 Agglutination. Unknown cells + normal human B serum (a) . . Unknown oells + normal human A serum (@) . . Group . . . . ly‘ + + AB Since stocks of a- and p- are now available it is easy to type the blood.The significance of these reactions has been cleady set out by Boyd,5 who gives diagraminaticnlly the transfusion possibilities (XIX) (in which the arrows indicate the direction of transfusion). U J A B 2 AB (XIX.) In order to avoid reaction between the donor’s cells and the recipient’s serum it is usual to transfuse from a donor in the same blood group of the patient ; though in t,he past donors in group 0 have been used as ‘’ universal ” donors. The A and B groups have now been sharply subdivided and two further groups M and N have been characterised. These are of no significance in human blood grouping because of the general absence of agglutinins to them in normal human serum they were discovered by K. Landsteiner and P. Levine,l5s and they appear t o be confined to the erythrocytes.Since isoagglutinins for these have not been found in humana it is neces- 153 ( a ) J . Immun. 1920 12 415 i ( b ) PTOC. SOC. Exp. BWE. Med. 1927 24 600 941. 238 QUARTERLY REVIEWS sary to detect them by the use of agglutinins immune to them obtained by injection of red cells into animals. They are of importance in forensic and anthropological studies. The relatively new Rh or Rhesus factors are discussed later. Blood Group “ Factors ” from Tissues other than Blood CeZLs.-As in immunity so also one can prepare haptens capable of specifically inhibit- ing isoagglutination between heterologous red cells and sera. These hap- tens are termed “ blood group specific substances ” which cannot func- tion as antigens but from which antigens can be prepared artificially.In addition t o their ability t o inhibit isoagglutination blood group specific substances can inhibit the haemolysis of sheep erythrocytes by rabbit sera under certain conditions though use of this property in assessing the activity of preparations has largely been replaced by the more reliable isoagglutination inhibition technique. The two types of activity do not seem to be related in any way inasmuch as it is possible to destroy one without interfering with the other.187 During the last two c1cc:tdes thero has been a large amount of work done regarding the superficial chemistry of blood group factors and on the reactions underlying the processes involved in isoagglutination and the other manifestations of blood group activity. There is yet no precise knowledge of the chemical structure of the blood group substances which are largely carbohydrate in character.The most obvious material for investigations of the nature of agglutinogens is the erythrocytes though so far such studies have been unsatisfactory. F. Schiff and L. Adelsberger l54 and K. Landsteiner and Vuii der Scheor 155 showed that a specific substance could be extracted with a’lcohol from red cells. C. Hallauer 156 was able to extract serologically activc material from human red cells of groups A By and 0 but apart from showing that in every case the active material gave positive tests for carbohydrates and that its elementary composition was C 43-46y0 ; H 7-8-50/ ; N 6.8-7-Oyi ; P 15-21y0 he was able to draw no conclusions as to its chemical nature. More recently Koss- jakow and Tribuleur 15’ claimed to have isolated “ A ” “ B ” “ M )’ and “ N ” factors from red cells and suggested that the factors were poly- saccharides though so far no confirmation of these results has appeared.Erythrocytes undoubtedly present a fruit,ful field for future chemical investigation. From the chemist’s point of view there are fortunately sources of blood group polysaccharides more readily accessible than those associated with the erythrocytes and it is with these that most chemical investigations have been concerned although the relationship between them and the blood group substances proper is by no means clear. Indeed the relationship may be no more than a close similarity in chemical structure of parts of the molecular complex. Blood group substances from erythrocytes.Blood group substances from other sources. l a r Z . Immunforach. 1924 40 335. 155 J. Exp. Med. 1925 41 123. l m Z. I.mmunfvrsch. 1936 63 287 ; 1934 84 114. lS7P. N. Kossiekow and G P. Tribuleur ibid. 1940 98 2 6 ; 1941 99 221. STACEY 1 ASPECTS OF IMMUNOCHEMISTRY 239 In 1926 Landsteincr and Levine 153 showed that substances possessing “ blood group activity ” occur in cells other than erythrocytes and later supporting evidence was obtained. I n 1931 F. Schiff l58 reported that group-specific substances could be extracted from tissues in a water-soluble form and in a form soluble in organic solvents e.g. alcohol. I n addition to those occurring in tissue the water-soluble form was found to be present in most body fluids e.g. saliva urine digestive juices cyst fluids etc.Not every individual however secretes blood group substances in this way and on this basis individuals of all groups may be divided into .‘ secretors ’’ and ” non-secretors ”. The most significant investigations reported hitherto have used as sources of blood group active material commercial pepsin gastric much saliva and ovarian cystic fluid. Other sources which have proved useful include urine pcptone and gastric juice. Investigations of the active material obtainable from human urine have been crtrricd out by K. Freudenberg and H. Ei~hel.15~ The aniount present is very small and urine cannot be regarded as a con- venient source from which to obt(nin the quantities necessary for extensive investigations. These workers were ablc to show that i he material from the urine of individuals of all groups was largely carbohydrate in nature and contained N-acetyl-glucosamine and galactose.Peptone was used as a source of a blood group sub- stance A by Goebel who showed that the active material it contained was principally carbohydrate in nature and contains galact ost and glucosamine constituents. The gastric juice of .. secretors ” is an important gource of human blood group active material siiice it is comparatively easy of access and the yields obtained are reasonably high. E. Witcbsky and N. C. Klendshog 160 obtained active substances from the gastric juice of “secretors ” of groups B and 0 and have reported briefly upon their properties. They appeared to be carbohydrates containing nitrogen. I n the writer’s laboratory it has been recently shown that human gastric juice is a useful source of blood group substances for chemical studies.Even when comparatively drastic methods of isolation e.g. evaporation of solu- tions at 60-70” with calcium carbonate are used the products often retain considerably serological activity 0.1 mg./c.c. being detclctnblo by the isoagglutination technique and the yields are high enough to riiakc its accumulation in useful amount possible. These have proved invaluable sources of blood group A substances and many workers have studied active material isolated therefrom. K. Meyer E. Smyth and J. Palmer 161 obtained a polysaccharide from hog stomach which was very active and contained N-acetyl-glucosamine and galactose. K. Landsteiner ( a ) From urine. ( b ) From peptone. ( c ) From gastric juice. (d) From pepsin hog stomach and hog mucin.158 “ Uber die gruppenspezifischen Substanzen des Menschen Korpers ” G. Fischer 150 Annulen 1934 510 240 ; 1935 518 97. l e 0 J. Exp. Med. 1940 72 663 ; 1941 73 655. l e l J . Biol. Chenb. 1937 119 73. Jena 1931. 240 QUARTERLY REVIEWS and P. A. Hartel62 obtained similar substances from hog stomach. A detailed investigation of the material from hog stomach has recently been reported by E. Kabat.ls3 W. T. J. Morgan and H. K. Kingls4 have investigated the isolation of active material from hog mucin in great detail and have obtained very active preparations. Their investigations of the stability of these will be referred to later. (e) From pseudo-mucinous ovarian cyst fluids. Morgan and King 164 obtained a very active blood group substance from these fluids of indi- viduals of group A and were able to show that it was very similar in properties to the substance previously obtained from hog mucin.More recently W. T. J. Morgan and M. B. R. Waddell165 obtained the corre- sponding material from cyst fluids of individuals of group 0. The active substances of the two groups were very similar in their properties differing mainly in their specific optical rotation. The saliva of " secretors " has been used in several investigations as a source of blood group substances. Landsteiner obtained very active material from horse saliva and was able to show that it con- tained galactose and a hexosamine. More recently the same worker com- pared the blood group substances from the saliva of individuals of groups A B and 0 and found that there was very little difference in their properties.Properties and Structure of Blood Group Substances.-Although no com- plete investigations of the structure of the blood group substances have yet been reported it is evident from the information available that all the material so far studied is predominantly carbohydrate and that the active materials belong to the class of mucoproteins. Preparations from widely different sources have many properties in common-e.g. elementary analysis specific rotation-but they must be still regarded as relatively crude products. Where constituent sugar units have been identified D-galaCtOSe and N-acetyl-glucosamine have been found. In addition D-mannose and Z-fucose may be present in the active material from commercial pepsin.166 It would appear that the structure of the stable carbohydrate from pepsin is of a branched chain type in which the ends of the branched chains are constituted of E-fucose units.Dr. G. Bray with the writer has recently shown that in the molecule there is also a relatively stable fragment of a polyglucosamine possibly closely related to chitin. W. T. J. Morgan 167 and his co-workers have studied in some detail the properties of prepara- tions made by using relatively mild conditions of isolation. These may be regarded as being very closely related to if not identical with the blood group substances as they are found naturally in secretions etc. They retained much of the viscosity of the original mucin and it was possible to correlate this property with the ability to inhibit isoagglutination. These (f) From saliva.18* J . Exp. Med. 1940 71 551. 18*Biochem. J. 1943 37 640; 1944 38 X . 16s Brit. J . Exp. Path. 1945 26 387. l 8 O H. G. Bray H. Henry and M. Stacey Bwchem. J. 1946 40 130. Brit. Me&. Bulletin 1944 2 165. lE3 Ibid. 1946 83 477 485. STACEY ASPECTS OF IMMUNOCHEMISTRY 241 authors have shown that treatment with N/lO-sodium carbonate at 100" for three hours causes some degradation of the molecule the activity being reduced to 1% of the original value and the optical rotation changing from + 10" to - 20". This change is accompanied by the splitting off of some of the hexosamine residues which can be separated by dialysis and can be shown to contain free reducing groups although the non-dialysable complex is still non-reducing. This suggests that some of the hexosamine molecules are linked to the main complex through their reducing groups which in some inexplicable way are alkali-labile.Kabat 163 has recently reported a careful study of the properties of blood groups substances prepared by Morgan's phenol method from hog stomach linings. He showed that they are serologically stable for 2 days at 37" and pH 1.02-10-7 and that treatment for 2 hours at 100" does not inactivate them between pH 2.97 and 7.58. This is confirmation of Morgan's findings on alkali-lability. He also isolated substances from indi- vidual linings and found that constancy of analytical properties gave no information as to their purity and activity since the chemical properties of active preparations were very similar to those of inactive ones. Witebsky and his colleagues 168 have introduced an important innova- tion in blood transfusion technique based on the knowledge that haptenes can inhibit specific reactions.They have suggested that the specific blood group factors be added to group 0 blood (" universal donor type ") in order to neutralise any isoagglutinins before intergroup transfusion. Clinical studies have shown the value of this and purified A and B factors are available commercially for the purpose. It is now recognised that amino-acids are important constituents of blood group substances. Landsteiner and Harte 162 showed that amino- acid nitrogen accounted for the bulk of the non-hexosamine nitrogen present and suggested that amino-acids play a part in the serological specificity of. these haptens. It was then realised that the slightly positive protein colour react'ions given by all preparations described must be due to these acids and not necessarily to small amounts of protein impurity present.Some detailed investigations of the amino-acid content of various prepara- tions have now been reported. K. Freudenberg H. Walsh and H. Molter 169 have isolated threonine from an A substance and Morgan 16' reports that by using the paper chromatography method at least 15 amino-acids includ- ing threonine and hydroxyproline in comparatively high concentrations have been found cystine being absent. In an appendix to Kabat's second paper Brand and Saidel 163 report the presence in the substance from hog stomach linings of glycine (1.6%) valine (0.7%) isoleucine (o.3y0) proline (3.3y0) phenylalanine (0.1%) tryptophan (0.2%) histidine (0.6%) lysine (0.6%) aspartic acid (0-8y0) glutamic acid (1.3%) serine (1-9yo) and tyrosine (0.3%).Thus the details of structure of these interesting sub- 16* ( a ) E. Witebsky N. C. Klendshog and P. Swanson J . Amer. Med. ASSOC. 1941 116 2654; ( b ) E. Witebsky N. C. Klendshog P. Swanson and C. McNiel Internal Med. 1942 70 1. 160 NaturwiSs. 1942 30 87. 242 QUARTERLY REVIEWS stances are gradually becoming known but until the modes of linkage between the sugar units and amino-acids and in the polysaccharide portion itself are determined it will be impossible t o integrate all the results which have been obtained and difficult to arrive at generally applicable con- clusions as to the determinant groups which decide specificity and degree of activit,y. Apart from elucidating the structure of individual blood group sub- stances investigation may reveal the causes underlying specificity itself.It is tempting to speculate that the specific substances of all groups con- tain tho same " coro "-possibly polysaccharide-and that the attached amino-acids arc responsible for conferring group specificity. We have observed that several polysaccharide-containing materials e.g. frog spawn mucin niay cxliibit somo degree of activity of inhibiting isoagglutination. I t will be of great intcrcst to determine which groupings presumably carbo- Iivdrnte-amino-~Lcid and others are responsible for blood group activity. ('onipounds of this type havo not beon much studied and it is possible that investigations of this nature may yield important results applicable to the wholt3 ficltl of inimunochemistry.Xaturally-occurring Polysaccharide Complexes related to Blood Group Sub- stances.-During investigations by several workers it has been noted that blood group substances have some properties very similar t o or the same as those of somo other naturally occurring carbohydrate complexes. The fact that blood group substances from certain sources especially erythro- cytes wcrc soluble in aqueous alcohol suggested a similarity t o the Fors- rnnnn polysaccharide haptens which are found in many mammalian and avian tissues (though not in those of man or the rabbit) and in some bac- terial " F "-antigens nicmtioncd previously. Thc specific polysaccharide of Type XIV pneumococbcus is of particular interest in this connection. It may prove t o be significant that agglutinin 2nd precipitin cross-reactions also occur between blood group A substances and the antibodies engendered to Pnoumococc1113 Type I and 8.A'chottmullwie. A rtijlcial Antigens from Blood Group Substances.-It has already been stated that blood group substances are not complete antigens but hap- tens. The possibility of converting them into true antigens in a manner similar to that in which specific bacterial polysaccharides and similar com- plexes have been convcrted into antigens has been investigated by Morgan ant1 Partridge. 122 These workers were able to prepare an artificial complex by mixing in formamide aqueous solutions of the blood group substances (A-substance from hog niucin) and the purified protein component of the 0-somatic antigen of Bact. typhosum or Bact.shiga. Other workers have coupled an A-substance with egg albumin using Landsteiner's azo-method involviiig the preparation of a p-nitrobenzyl ether of the A-substance and its reduction to the corresponding amino-compound which is then coupled with the protein. The antisera to which the injection of these artificial antigens givos rise are of considerable value in blood grouping. The Rhesus Factor.--In addition t o the blood groups and sub-groups related to the agglutinogens A and B and the other types of blood groups, STACEY ASPECTS OF IMMUNOCHElKfSTRY 243 e.g. those related to the antigens M and N there is the Rhesus (Rh) factor. All the antigens appear t o occur mainly in association with erythrocytes though claims have been made that their presence in tissues has been shown.Kossjakow and Tribuleur state that they are insoluble in organic solvents but stable only in chloroform and ether. The Rhesus factor was discovered by K. Landsteiner and M. Wiener 170 in the course of attempts to find new agglutinogens by injecting rabbits with red cells of various animals and testing the antisera obtained against human red cells. It was found that the sera of rabbits or guinea-pigs injected with the blood of a Rhesus inonkey (Macacus rhesus) after suitable absorptive treatment to remove known agglutinogens agglutinated the red cells of approximately 85% of the individuals tested. Such people am designated " Rh + " and the agglutinogen responsible for this property is termed the " Rhesus " (Rh) factor. The Rh factor is a powerful antigen comparable with agglutinogens A and B.Soon after its discovery in the laboratory a correlation was observed between haemolytic reactions occur- ring in blood transfusion in post-partum cases (which could be said to be due to the Rh factor) and certain pregnancy complications especially erythroblastosis fcetalis (haemolytic disease of the new-born). The hamo- lytic reactions which may occur on transfusion are due to the fact that introduction of Rh + blood into the circulation of a Rh - individual induces the formation of Rh antibodies which will react with any Rh + blood subsequently transfused (" second transfusion reaction "). The aetiology of erythroblastosis fcetalis may be somewhat similar. The f&us produced by a Rh + father and a Rh - mother may be Rh +. Under some conditions it is possible that fcetal blood containing Rh + red cells may find its way into the maternal circulation and stimulate the formation of Rh antibodies.These may cause a reaction if the mother is transfused with Rh + blood (" first transfusion reaction "). The antibodiee may also find their way back into the foetal circulation and cause a reaction with the Rh + cells present. Such a reaction may cause the death of the fetus in severe cases but in others may be relieved by transfusion of Rh + blood until the Rh antibodies ar0 eliminated. Thai; such an explanation is sub- stantially correct (though it may prove t o be oversimplified) is strongly supported by statistical investigations into the Rh grouping of parents and children in cases of erythroblastosis foetalis. The importance of Rh typing is clearly indicated.It is not known with certainty to what extent the Rh factor occurs in cells other than erythrocytes or in secretions. It appears to be absent from semen and to be present in saliva in only very small amounts and it has been stated that its quantitative distribution in organs is similar to that of A and B substances. Anti - enz ymes Bacterial enzymes in particular carbohydrases proteases and nucleases play a very important r61e in bacterial autolysis. Since many vaccines in present-day use are chemically crude substances it is likely that they con- l 7 0 Proc. Soc. ,Ezp. Biol. Med. 1940 43 223. 244 QUARTERLY REVIEWS tain lytic enzymes so that their fate in the body and the possibility of the formation of antibodies to them presents an important research problem for the future.Some enzymes have been investigated in regard to their antibody production ; for example bovine ribonuclease is an antigen i.e. the enzyme is precipitated by the homologous antibody but the complex still retains 80-90% of its ribonuclease activity. Anticatalase similarly has been prepared by injecting bovine catalase into rabbits. The anti- catalase cross-reacts with catalases from other animals while the washed antigen-antibody precipitates possess the original enzymic activity un- changed. Crystalline urease has been extensively studied and on injection gives rise to antibodies of the anti-enzyme type the immunising effect being to make rabbits tolerant to relatively high doses of the toxic enzyme. Specific antibodies to ty-rosinase from mushroom have been obtained both in humans and in the rabbit while a serological study of pepsins and pepsinogens has also been carried out with useful results.Of possible practical value are some investigations on the lecithinase from the toxin of CZ. welchii. Antitoxin specifically reduced up to 90% of the enzyme action and an inhibition reaction could be used as a means of measuring the approximate antitoxin content of serum. Antibacteriophuges.-Closely allied to lytic enzymes of bacteria and also to viruses are bacteriophages which are potentially of great therapeutic value. Phages belong to the class of autosynthetic molecules and some have been obtained in a highly purified state. The crystalline staphylo- coccus phage of Northrop appears t o be a homogeneous nucleoprotein with a molecular weight of the order of 300,000,000.Many phages behave as antigens on injection and give rise to antibodies which in high dilution inhibit the specific lytic action of the phage. Growth of a phage under various conditions does not appear to alter its antigenic specificity and it has been found possible by making use of serological methods to demon- strate the presence of different " receptor " (i.e. determinant) groups on the phage molecule. The chemistry of these aspects of phages and their substrates is as yet unexplored. The Signi$cance of Immunological Methods of Virus Studies.-Immuno- logical methods have been of great value in the early work on viruses. It was possible to show that various strains of the tobacco mosaic virus were closely related but not identical and further work has revealed that tho Holmes rib-grass strains and cucumber viruses " 3 " and " 4 " which differ from the other members of the group in their content of certain amino-acids also differ in a parallel way in their serological activities against tobacco mosaic antivirus.Significant observations regarding the shape of viruses and their power to precipitate antivirus antibodies have been made by A. K l e c z k ~ w s k i . ~ ~ ~ He found that the use of a constant amount of tobacco mosaic virus and the aucuba mosaic virus both of which are rod-shaped would throw down four times as much precipitate "lBrit. J . Exp. Path. 1941 22 188 192. STACEY ASPEUTS OF IMMUNOCHEMISTRY 245 as the bushy stunt virus which is spherical. A number of complex studies have been made on vaccinia antibodies some of which are involved in agglutination reactions with the vaccinia elementary bodies) though an antibody protective action against the infection does not yet appear t o have been characterised.Perspectives.-It will be apparent to the reader of this review that although chemistry has made some progress in the attack upon problems of immunity the surface of the subject has only been scratched. Many questions regarding determinant groups would appear t o have been answered but most of our knowledge regarding the chemistry of natural antigens and antibodies is undoubtedly " pre-structural ". Almost all the major successes achieved both by immunotherapy and chemotherapy have been mainly due to purely empirical methods. The rational basis of both methods of coinbating infectious disease is undoubtedly chemical specificity which concerns the shape and size of macromolecules so that in the future the two met'hods will need t o be studied and applied in closer regard t o one another. The next problems for the chemist con- cerning our defence against disease agents must be the working out on a firm basis of the fundamental and detailed chemical structure of those macromolecules from which are built ~p cells and tissues of all kinds.

ISSN:0009-2681    

DOI:10.1039/QR9470100213

出版商:RSC    

年代:1947

数据来源: RSC

摘要:

BASIC SALTS By HENRY BASSETT D.Sc. PH.D. D.-13s-Sc. F.R.I.C. (EMERITUS PROFESSOR OF CHEMISTRY UNIVERSITY OF READING ; DIRECTOR OF RESEARCH AND DEVELOPMENT MESSRS. PETER SPENCE AND SONS LTD. WIDNES) OUR knowledge of the structure of inorganic compounds has increased enormously during the past thirty years. This has been due chiefly to the application of X-ray diffraction methods to the study of crystalline solids.' Before this method was developed chemists endeavoured to deduce the structure of inorganic compounds from studies of gases liquids and solutions assuming that indications so obtained were applicable directly to the solid compound from which the gas liquid or solution was obtained. This method works well with organic compounds but for relatively very few inorganic ones. We know now that this is because the majority of organic compounds are composed of discrete molecules even in the solid state whereas very few inorganic compounds are so composed.Far that small number the old methods are satisfactory up to a point but for the majority of inorganic compounds they are of little help. Even for compounds either organic or inorganic composed of discrete molecules the older methods of study gave information only about single molecules and told little or nothing about their arrangement in the crystals. That arrangement existed only in the crystal and was dest,royed on vaporisation fusion or solution but insofar as the same molecules were present in all four physical states the compound could be said to exist in all these states. The study of inorganic crystals by X-ray methods has made it quite clear that the majority of these compounds exist only in t'he crystalline state (or in several crystalline states).In the majority of cases no molecules can be distinguished in the crystals which are built up of atoms or ions arranged in a comparatively limited number of patterns. On vaporisation fusion or solution the crystal structure breaks down into atoms ions or molecules. The only similarity to the original solid lies as a rule in the proportion of the various atoms or ions being the same in the new physical state as it was in the original solid. Even this similarity is often absent when for example a solid on heating gives a mixture of a gas and a new solid. With few exceptions therefore the study of the structure of inorganic compounds had to wait until methods were available for studying crystals in the solid state.The case of the silicates is of course classical in this connection. There had been little progress in the understanding of this complex group of compounds in a structural sense until a systematic study of the diffraction of X-rays by silicate minerals had been made largely by W. L. Bragg and 246 W. H. and W. L. Bragg " X-Rays and Crystal Structure" 1st edn. 1915. BASSETT BAS10 SALTS 247 his schooL2 That work gave an enormous impetus to the similar study of other compounds and very many inorganic compounds of all types have been investigated by X-ray methods during the last twenty years. The basic salts form one group of compounds upon which much light has been shed by such methods. Views as to the nature of the interatomic links in the compounda of the metals have undergone a considerable change as a result of these X-ray studies more particularly in regard to the differentiation between ionic and covalent links.Such work has shown that there is no sharp line of division between compounds which are truly ionic in nature and those which are entirely covalent. There appears to be every shade between these two extremes. When a structure is fully ionised the constituent atoms (or in some cases groups of atoms forming a complex ion) are held together by electrostatic forces and the charge on each atom can be regarded as being uniformly dispersed. If linkages are covalent then the charges become localised and the links are directional. This difference has an impofiant effect on the type of packing of the atoms in the crystal.In the ionic structure the ions pack together as closely and in such a manner as is deter- mined chiefly by their relative sizes. An ionic crystal tends therefore to have a close-packed lattice. Where the links are covalent and directed this is not usually the case. Thus when covalent links are present an oxygen atom cannot have more than four neighbours in a tetrahedral configuration whereas if the linkages are ionic with close packing six neighbows may be present in an octahedral arrangement. The co-ordination of the atoms in a lattice (Le. the number of atoms surrounding another atom) is a very important criterion in assessing the types of linkage present. Comparison of ionic radii with covalent radii or metallic radii shows that a positive ion is appreciably smaller than the neutral atom of the same element owing to the excess of nuclear charge over that of the orbital electrons whereas the radius of a negative ion is much larger than the covalent radius Li .. . . . 1.52 A. C1- . . . . . . 1-81 A. Lil . . . . . 0.60 A. c1 . . . . . . 0.99 A. The interatomic distance in a covalent link is longer than that of an ionic link for the same pair of elements. Thus for a covalent link between Li and C1 the distance would be 2.51 A. while for an ionic link it would be 2.41 A. For a covalent link between Si and 0 the distance would be 1.91 A. while for an ionic link it would be 1-81 A. For N-0 the covalent link is 2.00 A. and the ionic link 1-90 A. Now in a covalent crystal the individual links are longer than in an ionic crystal and these linkages are directed whereas in an ionic crystal the atoms are much more close-packed.For these reasons a crystal formed by covalent links has a more open structure and sometimes a much more open structure than one formed only by ionic links. X-Ray observations enable the interatomic distances and relative arrangements of the atoms in a crystal to be deduced. The decision as t o * " The Structure of Silicates " 2nd edn. 1932 ; Trans. Faraduy Soc. 1929,2!5,291. 248 QUARTEBLY BEVIEWS whether a given structure is covalent or ionic or to what extent it has either characteristic is based mainly on such determinations. The links between metal atoms and oxygen or fluorine appear to be essentially ionic. As one goes from fluorine to iodine the links between metal and halogen become increasingly more covalent in character though even the iodides especially of the more electropositive metals are mainly ionic.Even in the case of silicon and oxygen the links appear to be about 50% ionic. At one time it was thought that a link must be either ionic or covalent and the experimental evidence that this is not the case has been accounted for by the principle of resonance. The binding of atoms is a very complex process and is brought about by the interaction of the valency electrons of the combining atoms (or their corresponding orbitals). Several different types of link could originate as a rule between any two atoms according to just how these interactions took place. These several types of link are compounded by resonance into the one which actually occurs which is of higher stability than any of the possible types individually.This recalls the well-known case of benzene where the distance between each pair of carbon atoms is not that characteristic of either a single (1.54 A,) or a double (1.34 A.) bond but has an intermediate value (1.39 A.) less than the mean of the two. It is easy to see that covalent linkages can lead to crystal lattices showing either isolated groups of atoms or to chain band sheet or meshwork structures. It is not so readily appreciated that similar types of structure can occur in ionic crystals. If an ionic crystal has a layer structure then the oppositely charged ions in the layers will be held together by strong electrical forces each ion with its appropriate co-ordination number. These layer structuresare built up into crystals by much weaker van der Waals forces.The more purely ionic the linkages in a crystal the more close-packed its structure and the more this corresponds to a three-dimensional mesh- work. Formation of sheet and chain structures is generally an indication of increasing covalency of the atomic linkages but a three-dimensional meshwork is not necessarily a closely paoked or an ionic one and this is true of the silicate meshworks. The presence of hydroxyl groups or ions in a crystal leads to other complications due to the formation of what are known as hydroxyl bonds. As these bonds are largely responsible for the formation and peculiarities of many though not all basic salts a short account of their nature is here given. The Hydroxyl Bond. 3-AbnormalIy short 0-0 distances in certain acid salts arise from the presence of an H atom between two 0 atoms.Such short hydrogen bonds -0-H-O- of length about 2-55 A. are symmetrical owing to resonance between the extremes -0-H . . .O- and -0 . . . H-O-. The H atom is not attached preferentially to either of the 0 atoms as is shown by the infra-red absorption spectra. The absorption band at 3 p characteristic of the OH group is absent in the spectrum of H,SO and of Cf. A. F. Wells " Structural Inorganic Chemistry " Clarendon Press 1945 pp. 350 351. BASSETT BASIC SALTS 249 compounds with intra-molecular H bonds such as salicylaldehyde. Bonds of a somewhat similar nature are formed between hydroxyl groups. In this case as distinct from the above there is an H atom attached to each 0 atom and in some cases a t least there is good evidence that the H atoms of these longer " hydroxyl " bonds (2.7-2-8 A .) remain associated with their o m 0 atoms. I n other words the hydroxyl group retains its identity when forming an hydroxyl bond so that compounds with such bonds still give the absorption band a t 3 p. The packing of the OH group in certain hydroxides indicates that directed bonds are formed between OH groups attached t o different metal atoms leading to a more open packing of these groups than is found for the halogen atoms in the corresponding halides. Since the OH groups concerned are attached on one side to metal atoms the bond between the OH groups cannot be any simple dipole attraction since similar ends of the dipole would be in contact. therefore suggested that in a suitable environment the oxygen atom of the OH group develops the same tetrahedral character as in the water molecule.I n the OH group the H atom lies about 1 A. from the centre of the 0 atom the effective radius of which ranges from 1.3 to 1.8 A. and we may assume a similar position for the H in the OH ion. The O-H bond is polar so we may picture the OH ion as a dipole with a charge of about 1.5e near the centre of the 0 atom and a charge of + he at the H atom. This ion has cylindrical polar symmetry. I n the vicinity of the more highly charged positive ions however the electron distribution is distorted. Bernal suggests that the OH ion passes through the following stages. If the polarisation is small the group retains its cylindrical symmetry and merely increases its dipole moment.Further distortion leads to a separation of charges arranged tetrahedrally. The negative charge on one 0 can attract the H atom attached to another forming a hydroxyl bond. With still further polarisation the H atom can leave the 0 and migrate to a neighbouring 0 atom. This corresponds to our picture of the short H bond for the two states of the system have identical energies and the H may be regarded as belonging equally to both atoms. This final stage in the polarisation of the OH group will occur only in oxy-acids and acid salts. In hydroxides such a transfer of H lvould lead to the formation of a water molecule. The old view of basic salts was that they resulted when some only of the hyctrosyl groups or oxygen atoms of a polybasic hydroxide or oxide reacted with acid.I n this way such basic salts as Cd(0H)Cl or BiOCl were formed. They were regarded as the opposite of acid salts such as NaHCO or CaHPO,. Simple compounds of this kind could be readily explained in this way but it was more difficult to account for more complex basic salts such as CuC12,3Cu(OH), of which many were known. A. Werner attempted to deal with some of these in terms of his co-ordination theory by supposing that the1 contained complex kations in which molecules of the metal hydroxide were co-ordinated through the hydroxyl groups t o the metal ion in much Proc. Roy. SOC. 1935 A 151 384. " Neuere Anschauungen auf dem Gebiete der Anorganischen Chemie " Zurich J. D. Bernal and H. D. Megaw 1st edn. 1905; 2nd edn. 1908. 250 QUARTERLY REVIEWS the =me way as water molecules in hydrated ~alts so thGt CuC1,,3Cu(OH) became a copper ion was at the centre of an octahedron with three Cu(OH) molecules straddling three edges.This view was supported by G. T. Morgan 6 and N. V. Sidgwick.' There were many basic salts known however especially among minerals which could not be accounted for in such ways. Basic salts will be defined for the purposes of this article as " salts in which the proportion of base to acid is greater than in the normal salts ". This includes a small number of compounds such as BiOCl which are excluded by an alternative definition which only includes those compounds coming under the- above general definition which contain OH groups. The majority of metal oxides and hydroxides are either very insoluble or only sparingly so so it is not surprising that basic salts and especially the very basic ones tend to be very insoluble even when the corresponding normal salts are easily soluble in water.This makes the study of basic salts by phase-rule methods rather difficult. Each basic salt is generally stable only in contact with very dilute solutions and often over only a very short range. For these reasons basic salts are often difficult .to prepare in a pure state or well crystallised. X-ray examination is of great help in sorting out the various basic salts in such cases quite apart from any attempts at determinations of structure. Many well-crystallised basic salts occur as minerals and many of these as well as others not known as minerals have been prepared artificially. Some of them crystallise with surprising readiness but rather special conditions are frequently necessary.The flocculent precipitates first ob- tained by adding alkali to heavy-metal salt solutions are as a rule basic salts. Prepared in this way they may look amorphous but may give an X-ray diagram characteristic of a crystalline solid. Further action of alkali will convert them into the metal hydroxide. I n some cases metal hydroxide is readily converted into basic salt by the action a t room temperature of a solution of the normal salt. Possible Types of Basic Salts I. Salts derived from Poly-acid Bases in which only some of the 0 or OH Groups have reacted with Acid and in which the Remainder are covalently attuched to the Metal.-Uranyl chloride UO,Cl, appears definitely to contain the bivalent UO ion. It is noteworthy that this and the nitrate sulphate and acetate are all readily soluble and stable towards water and probably the UO kation is present in all these compounds.The complete elucidation of their crystal structure may be a matter of great difficulty however.8 It does not follow that all urnnyl compounds are of the same character and " A Survey of Modern Inorganic Chemistry " p. 47 ; Institute .of Chemistry Lecture 1933. " Electronic Theory of Valency " 1927 (Clarendon Press). * Cf. the case of [U02][N0,],,6H,0 L. Pauling and R. G. Dickinson J . Amer. Chem. SOC. 1924 46 1615. BASSETT BASIC SALTS 25 1 in the case of the mineral autunite Ca[(U02),(P0,),],nH20,Q the structure recalls that of the micas inasmuch as PO and UO groups are linked together into two-dimensional sheet anions.The three atoms of the UO group form a line perpendicular to the plane of the linked PO groups and the uranium atoms are large enough to cause an oxygen atom of the UO groups to project alternately above and below the PO plane. These projecting oxygen atoms form cavities between the planes in which are situated the calcium ions which hold the anionic sheets together and the water of hydration. This water is zeolitic in character and the calcium also as it is readily replaceable by other metals by the action of salt solutions (see Fig. l).Qu n n 1 ! x x jc FIG. I Elevation of the structure of a u t u ~ i t e Ca(U02)2(P04)Z,nH,0 (after Beintern). (Reproduced by permission from Wella.’s “ Structural Inorganic Chemistry ”.) So far as the X-ray evidence goes autunite is to be regarded as a salt of a complex uranic-phosphoric acid and this shows how artificial is the division of salts into basic normal complex etc.So much depends upon the sizes of the atoms and how they can be packed together. Even aluminium which occurs as kations in most of its solid compounds can enter readily into very stable complex anions as is well known to all students of chemistry. It is by no means necessary that metals behaving in this way should be of an “ acidic ” nature though it is especially those of low electro-affinity which tend to form negative complexes. For instance in salts containing the OsO J. Beintema Rec. Trav. chim. 1938 57 155. J. G. Fairchild Amer. Min. 1929 14 265. 252 QWaaTHsLY BlnvIEw8 or oemyl group the latter ia usually preeent aa part of a complex negative ion aa ia %he m e with &[Os0,,C14].In the crystals of this compound the two KO ions and the complex [Oa0,C14]" ion are arranged in a manner closely similar to that of the two F and the Ca" ions in fluorite.lO It is clear that all sdts containing negative complex ions composed partly of rnehl+xygen group can be regarded aa a apecid type of baaic salt but it is customary and usu&uy more convenient fo regard such compounds aa dts of substituted oxy-acids. There are a few compounds at the other end of the smle which although they contain no metal have some claim to be regarded as a kind of basic salt. Some of the nitrosyl compounds-such as (NO)ClO and (NO)BF4 ll-have been shown to have similar structures to the corresponding ammonium salts so that they probably contain (NO)' kations.Formally at least the NO' kationis basic just as UO," although nitrosyl salts are not generally thought of from this angle. Several basic salts seem to have kations of the Werner co-ordination pattern with covalently attached hydroxyl groups though here also the X-ray evidence is either incomplete or entirely lacking at present. The simplest of these is the least basic of the basic aluminium sul- phates Al,03,2S03,1 1H20. This is well crystallised and readily soluble in cold water (with slow further hydrolysis) and the X-ray evidence though not yet completed shows that it probably is to be regarded as [Al(OH)(H20),]"[S04]".12 Werner prepared a very large number of cobaltic and chromic salts in which the cr"' or Co"' ions were co-ordinated to a variety of groups in octahedral complex kations.A number of these were found to exist in two forms in agreement with the geometrical requirements of an octahedral configuration which required the existence of cis- and trans-isomers in certain cases. A few of the compounds occurring in cis- and trans-forms were basic salts inasmuch as an OH group was present in the complex ion. Examples of these are [ :$ co en,] X 13 The one form in each case was readily converted into an oxalato-compound [C204Co(or Cr) en,]X and was therefore considered to be the cis-form. This form readily yields the diol on heating or OH Fz co OH co en,]^ 16 OH [en Cr OH cr en,]^ 17 10 J. L. Hoard and J. D. Grenko 2. Krist. 1934 87 100. l1 L. J. Klinkenberg Rec. Traw. chim. 1937 50 749. l2 H. Bassett work not yet published. 1s A. Werner Ber.1907 40 272 ; J. Meisenheimer and E. Kiderlen Annalen 1924 1 4 P. Pfeiffer 2. anorg. Chem. 1907 50 279. l5 A. Werner Ber. 1907 40 4434. l6 Idem Annalen 1910 375 83. 17 P- Pfeiffer W. Vorster and R. Stern Z. a w q . Chem. 1908 58 272. 438 252 (en = ethylenediamine ; X = univalent negative ion). BASSElT BUIU SALTS Compounds claimed to have the structure (NH,),Co OHCo(NH,) X [ OH OH 1 h ave also been prepared.lB The still more complex compound is an important one in the history of stereochemistry. It was prepared by Werner and resolved by him into its optically active isomers being about the first compound containing no carbon whatever to be so resolved.1B It was the existence of compounds of this type which led Werner to the view that in the basic salts of other metals such as copper zinc etc.the metal hydroxide was co-ordinated to the kation of the normal salt. The stereochemical evidence quoted is strongly in favour of these com- pounds having essentially the structures assigned to them by Werner. Many other closely related basic complex salts of Co Cr and Pt have been prepared but there is not always much stereochemical evidence in support of their supposed structures. The fact that no cases of polynuclear kations linked together by hydroxyl groups appear to have been found among the numerous basic salts which have been examined by X-rays makes the need for such examination of the cobalt and chromium complexes all the more urgent. 11. Basic &‘ah with an Injinite Three-dimensional Lattice of an Ionic Type. -Many basic salts are best regarded as a special type of double salt con- taining 0” or OH’ anions in addition to typical anions of the acidic type (F’ Cl’ SO4” etc.).This group of basic salts is probably a large one and contains some well-known and important minerals. Lanthanum oxyfluoride 2O LaOF prepared by heating a finely powdered mixture of La,O and LaF to 900” for 36 hours in a high vacuum is the simplest fully ionic basic salt known. It is cubic and has the fluorite structure. It is able to take a large amount of LaF into solid solution (up to about 62 mols. yo). This is because LaF also has a lattice closely related to that of fluorite. Up to about 2 mols. yo of La,O can go into solid solution in the LaOF the smaller proportion in this case being due to the fact that the crystal structure of La,O differs considerably from that of LaOF.This capacity for solid-solution formation is a characteristic of many basic salts and helps to make their study more difficult. The well-crystallised minerals Zibethinite 21 Cu2( OH)PO, olivenite 22 Cu,(OH)AsO, and &mite 23 Z%(OH)AsO are isomorphous and all have three-dimensional structures of the ionic type. Half of the metal ions are surrounded approximately octahedrally by four oxygen atoms of AsO (or PO,) ions and by two OH ions while the other half are surrounded by l* A. Werner Ber. 1907 40 4834. 2O W. Klemm and H. A. Klein 2. czn.org. Chem. 1941 248 167. 21 H. Heritsch 2. Krkt. 1939 102 1. 22 H. Strunz ibid. 1936 94 60. l9 Idem Ber. 1914 47 3087. a3 P. Kokkoros ibid. 1937 96 417. 254 QUARTERLY REVIEWS four oxygen atoms of AsO (or PO,) ions and one OH ion the five oxygens being a t the corners of a deformed trigonal bipyramid (see Fig.2). There is a possibility of hydroxyl bonds between the OH ions and the oxygen atoms of the PO or AsO ions but any effect of this is so small as to have little influence on the packing of the ions. It is thus found that the structure of the orthosilicate andahsite 24 A12(0)Si04 is essentially the same as that of olivenite etc. The other two polymorphic forms of Al,(O)SiO have closely related structures. I n sillimanite half the A1 ions are 6 co-ordinated and the other half 4 co-ordinated while in cyanite z 5 they are all 6 co-ordinated in all cases to oxygen. The density of these three polymorphic forms increases with increase in the co-ordination number of the A1 ions. Higginsite Fra.2 Plan of the structure of CuZ( OH)AsO ( o h e n i t r ) . Large circZes represent 0 (light) or OH (heavy) si~mll shuded circles Cu and sniull hluck circlcs AH. Shuding indicates heiqht abore the plrrne of the pciprr ; iinshncled atoms (ire ( i t lieiglit 0 those shaded horizontally at height c / 2 and those shuded horizontnlly and vertically at h e i g h t s c / 4 and 3 ~ 1 4 . The 6 co-ordinatrd Cu atoms are those (it the cornzprs and centre of the unit cell. (Reproduced by permission firm f e l l s ’ s “ Strztclurnl Inorganic Chemistry ”.) CaCu( OH)AsO and desdoisite (Pb ,Zn)Pb( OH)VO probably have similar structures to adamite with the Cu in higginsite and half the Pb in descloisite in thc 5 co-ordinated condition referred to above but tilasite CaMg(OH,F)AsO appears to have the same structure as titanite (sphene) &Ti( O)SiO4 in which both Ca and Ti have the usual octahedral co-ordination.26 When it is found that I?’ can replace OH’ isomorphously as in the apatites or 0” as in the durnngite-titmite pair this can be taken as strong presumptive evidence that the OK or 0 is ioiiic and that for example in titanite one has Ti”” and 0” ions and not (rl’iO)**.This deduction depends upon the fact that in it’s compounds with metals fluorine is almost always ionised. Dqlrangitc 27 NnAlFAsO also has the same structure as titanite. 2 4 W. H. Taylor ibid. 1920 71 203. a6 S. NQray-Szsbo W. H. Taylor and W. W. Jackson ibid. p. 117. 26 H. Strunz ibid. 1937 96 7. 27 P. Kokkoros ibid. 1938 99 39. BASSETT BASIC SALTS 255 Topaz 28 Al,(F,OH),SiO also contains individual SiO tetrahedra and has in a general sense a structure somewhat similar to that of the above minerals.Each Al ion is at the centre of an octahedron composed of four oxygen atoms of SiO ions and two F or OH ions. Linarite CuPb(OH),SO possibly belongs t o this type of basic salt also. Malachite Cu,(OH),CO is another important mineral with a structure similar to that of olivenite in the sense that some of the copper ions are 6 co-ordinate being surrounded by two oxygens of CO ions and four OH ions while the others are 5 co-ordinate and surrounded by three oxygens of CO ions and two OH ions. The anhydrous aluminium silicates are very important in connection with refractories and they have been studied considerably. If the formulz of sillimanite Al,O,,SiO and mullite 3Al20,,2SiO are written in this way the)- suggest that mullite is a much more basic aluminium silicate than is sillimanite.Optical and X-ray data show that the two minerals must have almost identical structures. We have seen that sillimanite is A126+02-(Si04)4- and it seems fairly certain t,hat mullite differs from this in having one quarter of the silicon replaced by aluminium. This requires either the provision of one extra positive charge for every Si so replaced or what comes to the same thing the removal of one negative charge. Mullite 29a can then be written The mullite lattice is then similar to that of sillimanite with one oxygen ion position vacant and a quarter of the (SiO,) ions replaced by (A10,) ions. Looked a t from this angle mullite is actually less basic than sillimanite since it contains a smaller proportion of oxygen ions and its greater A1,0 content is due to the presence of much of this in the complex negative ions.Such replacement of silicon by aluminium gives rise to compounds which are often called “ more basic ” by the mineralogists. Thus anorthite Ca2+[Al,Si,0,]2- is commonly called a more basic felspar than albite Na+[AlSi,O,]- with which it is isomorphous. This use of the term “ basic ’’ must not be confused with the use made of it in the present review. From the view point of the latter both anorthite and albite are “ normal ” salts. Hydroxyapatite is generally written as [Ca,(PO,),],,Ca(OH),. As a mineral this is of minor importance in comparison with the isomorphous jluorapatite [Ca,( PO,),],,CaF,. Biologically hydroxyapatite is of great importance as it is the major constituent of bone.A. Werner 3O considered apatite to have the structure 28 J. Leonhardt ibid. 1923 59 216 ; N. A. Alston and J. West Proc. Roy. SOC. 1928 A 121 358 ; 2. Krist. 1928 69 149 ; L. Pauling Proc. Nut. Acad. Sci. 1928 14 603. 29 H. Brasseur and J. Toussaint Bull. SOC. roy. sci. LiQe 1938 213. 30 Ber. 1907 40 4447. W. H. Taylor 2. Krist. 1928 68 503. 260 gUaaTHuY REvInw8 with a very complex kation but without the slightest evidence. Only quite recently haa the structure been fully elucidatd,*l although the new results agree in all important points with the older determination by S. Nhray- Szabo.81 The structure ie a three-dimensional ionic complex of a remark- able type. Along each of the three-fold axes is a chain of Ca ions Q of the c-axis apart each of which is bonded to its neighbours above and below by mans of three oxygen atoms thus forming a continuous chain -Ca-3O-C~3O-Ctc.Each Ca is also linked to three oxygens almost at the same level on the c-axis as itself 80 that each Ca ion in the chain is surrounded by 9 oxygens. All these oxygens belong to PO ions and the P atoms link the chains together producing a hexagonal network with channels running down the c-axis of the prism. In the walls of these channels are six hollows per unit length three at one level and three a t another and into these six other calcium ions fit with just enough room in the centre of each group of three for one F or OH ion to fit. This gives a very unusual planar arrangement of three calcium ions at the corners of an equilateral triangle around the central fluorine or hydroxyl ion.These planar Ca3F groups are arranged one above the other in the hexagonal channels in such a way that the Ca ions of alternate planes are directly above one another while those in two adjacent planes form a six-fold arrangement. An interesting feature of this arrangement is that the OH ions of hydroxyapatite are adjacent to one another which seems remarkable in view of the fact that a temperature of about 1200" is needed for complete expulsion of the water. It is true that they are some distance apart (3.4 A.). Similar proximity of hydroxyl ions and difficulty in expelling water is found in other cases such as topaz and especially in basic complex silicates such as the amphiboles including Gcsbestos and mim and may be usual.It is possibly associated with great stability of the lattice structure as a whole and the fact that this must collapse when water is driven off. The stability of the lattice as a whole the exact distance apart of the OH ions or groups the presence or absence of hydroxyl bonds and the extent of the ionic character of the OH in any given case must all affect the ease with which elimination of water will occur on heating. The ionic strength is perhaps very important as judged from the relative behaviour of the members of the series NaOH or KOH LiOH Ca(OH), Mg(OH), Zn(OH), where the ease of elimination of wafer increases greatly with decrease in ionic character. All the indications appear to be that the apatite structure is highly ionic. It should of course be written Ca,F(PO,) or Ca,(OH)(PO,) and not in the conventional manner [Ca3( PO,),],CaF or [Ca,( PO,),],Ca( OH) which suggests that Ca,(PO,), CaF, and Ca(OH) are present in the compounds.The simplest formula which gives the chemical composition should be used for all double salts. Pb,(OH)(PO,) is said to show great similarity to calcium hydroxy- apatite and presumably has similar structure.33 31 C. A. Beevere and D. B. McIntyre Min. Mag. 1946 27 254. 32 2. Krist. 1930 76 387. sa R. Klement 2. a m g . Ch. 1938 257 161. BASSErP BASIU SALTS 257 111. Basic S'ah with Layer Lattices of the Ionic Type and without Hydroxyl. -The majority of metal oxyhalides which have had their structures fully determined are less fully ionic than LOF. An indication of this is given by their layer-lattice structures. FeOCl 34 and lepidocrocite y-FeO-OH are both orthorhombic and have essentially the same structure in the individual layers of which the crystal lattices are composed.The Fe ions are a t the centres of somewhat distorted octahedra of oxygen and OH ions or of oxygen and chlorine ions in the case of FeOC1. These octahedra are linked together in such a way that each C1 or OH ion is common to two octahedra while each oxygen ion is shared by four octahedra. An electrically neutral layer structure results which has a series of ridges and valleys with OH or C1 ions at the tops of each ridge. The ridges are alternately on the one side or the other of the layer. In FeOCl the outermost atoms on both sides of the layers are chlorine and the layers are stacked in the crystal so that there is close-packing of the chlorine ions-the ridges on one layer fit into the valleys in the next.In y-FeO*OH the OH ions are outermost on both sides of the layers. These cannot fit together so as to give close-packing of the oxygen atoms owing to the inter- vention of hydroxyl bonds. These result in directed links of a tetrahedral nature between the hydroxyl groups of neighbouring layers and a rather more open structure in which the outer oxygen atoms of one layer lie directly upon those of the layer below instead of fitting in between two oxygens of the lower layer. The bismuth oxyhalides BiOCl BiOBr and BiOI are tetragonaL35 They too,. form layer lattices with the halogen ions on both sides of the layers. The covalent character of the linkages becomes more marked as the atomic weight of the halogen increases.This is shown by the cleavage parallel to the plane of the layers becoming more and more marked. It is also shown by the distortion of the halogen ions due to the small bismuth ions. In the case of the oxychloride and oxybromide this distortion or polarisation is more marked along the c-axis i.e. between contiguous layers than along the a-axis. In both compounds the distance between two halogen atoms in contiguous layers is somewhat less than the theoretical diameter of the halogen atom but distances between haIogen atoms in the same layer indicate no compression. In the case of the oxyiodide there is marked compression in both directions ; the distance between two iodine atoms in the same layer is 4-01 A. and in contiguous layers 4-17 A. as compared with the theoretical diameter 4-40 A.The red colour of the oxyiodide is also an indication of marked polarisation. IV. Basic Salk with Layer Lattices of the Ionic Type and containing HydroxyL-A great deal of work has been carried out on the basic salts of bivalent metals especially magnesium zinc and cadmium which belong to this category. The salts examined have been mainly halides but some nitrates sulphates phosphates etc. have also been dealt with. Apart from 34 S. G . Goldaztaub Cmnpt. rend. 1934 198 667 ; Bull. Soc. fraw. Min. 1935 58 6. 35F. A. Bannister and M. H. Hey Min. Mag. 1936 24 49. 258 QUARTERLY REVIEWS the fluorides which are fully ionic with either the fluorite or ruffle struoture the halides and hydroxides of these metals form similar layer-lattice struc- tures. Each metal ion is surrounded octahedrally by six halogen or hydroxyl ions each of which is common to three octahedra which are thus linked together into an infinite layer or sheet which is electrically neutral.In the case of zinc the individual layers of both hydroxide and halide are similar but in the hydroxide the layers are held together by hydroxyl bonds instead of by van der Waals forces as in the halide. Hydroxyl bonds are not present in the hydroxides of the other bivalent metals with the exception of beryllium so that in those cases the structures of hydroxide and halides differ much less. The basic magnesium chlorides 36 seem to be well defined and inter- esting Mg(0H)Cl has the MgCL structure with statistical distribution of the OH and C1 in the octahedral arrangement of ions around each magnesium ion.There are also compounds MgC1,,3Mg(OH)2 and MgCl2,5Mg(OH) which exist in the water-free form and also as 3 5 and 7 hydrates. The water-free forms have the Mg(OH) layer structure but in the hydrates wit,h 5 and 7H20 these layers have been broken up into bands the crystals being built up of these bands separated by bands or chains of H20 molecules. In the case of zinc and cadmium some rather ill-defined hydroxyfluorides are said to exist 3 7 The hydroxy-chlorides -bromides and -iodides of zinc are well crystallised and of two types ZnX,,Zn(OH) and ZnX,,4Zn(OH)2.38 There is also a basic nitrate Zn(N03)2,4Zn(OH)2.38 All these compounds have layer-lattice structures very similar to that of a-Zn( OH) [Zn( OH) is dimorphic]. The structure of ZnX,,Zn(OH) [Zn(OH)Cl] differs most from that of a-Zn(OH),.The chief difference in the other cases is due to an expansion along the c-axis as OH is replaced by the larger halogen ions. Although much work has been done on the corresponding cadmium compounds only two of these seem to be really well established namely Cd(OH)Cl 39 and CdC12,4Cd(OH),.40 There are probably a number of others 41 but the cadmium compounds appear to be less soluble and to crystallise less readily than the zinc compounds. This and the small differences in the X-ray diagrams make results difficult to interpret especially as even when definite compounds clearly exist as in the case of zinc there are indications that intermediate solid solutions (possibly metastable) may appear. This is perhaps due to some random replacement of OH by halogen in the hydroxide layers and is most likely to occur when the basic salt is rapidly prepared as by precipitation.The structure of Cd(0H)Cl differs in an interesting way from that of Mg(0H)Cl. The crystals are made up of layers having the empirical 36 JV. Fcitknorht and F. Hold Helv. Chiin. Actcr 1944 27 1480. 37 W. Foitknecht and 11. Burhor ibid. 1943 26 2177 2196. W. Foitkneclit ibid. 1930 13 23 ; W. Feitknecht and H. Weidmann. ibid. 39 J. L. Hoard and J. D. Grcnko 2. Krist. 1934 87 110; W. Feitknecht and 40 Idem H c l v . Chim. Actrr. 1937 20 1344. 4 1 I d e m ioc. c i t . ; JV. Feitknec.lit h'xperientia 1945 1 230 ; Helv. Chim. Ada 1043 26 1560 1564. W. Gcrbor ibitl. 1937 98 168. 1045 28 1444. BASSETT BASIC SALTS 259 composition Cd(0H)Cl in which every Cd ion is surrounded octahedrally by three OH and three C1 ions in such a manner that t.he OH ions lie all to one side of the plane of Cd ions and the C1 ions t o the other side.All the OH and C1 ions of an octahedron are shared by two other octahedra and in consequence every C1 has three Cd ions to one side of it and three OH to the other in an octahedral arrangement while the OH ions are similarly surrounded by three Cd and three C1 ions (see Fig. 3). The layers in the crystals are held together merely by van der Waals forces for there is no possibility of hydroxyl bonds since the hydroxyl ions of one layer face chlorine ions of the nest layer. Now Mg( OH) and Cd( OH) both crystallise with the same lattice and so do MgCl and CdC1 and the general arrangement of ions in the lattices of Mg(0H)Cl and Cd(0H)Cl is fundamentally similar.Why is the distribution of OH and C1 random in the case of the magnesium compound but fixed in a very definite way in the cadmium compound ? W BIG. 3 Portion of one layer of Cd(0H)Cl. The OH groups (shaded) lie above and the C1 atoms below the plane of the metal atoms. (Reproduced by permission from Wells's " Siructural Inorganic Chemistry ".) It may mean that the linkages in Cd( 0H)Cl have sufficient covalent character to fix the OH and C1 in some favoured position while those in the magnesium structure cannot do this because of their rather more ionic character. Some basic sulphates form layer-lattice structures CoSO,,3Co( OH) is one of these.42 It forms microscopic hexagonal leaflets built up of alternate layers of hydroxide and normal sulphate ; about 4 mols.of zeoliiic water are normally present. This blue basic salt is converted by concentrated aqueous CoSO into a violet salt 2CoS04,3Co(OH),,5H,0 in which the water is also held zeolitically. It is stated7*3 on the other hand that CdSO4,3.5Cd(OH) forms hexagonal plates and that it has a simple layer-lattice structure formed from that of Cd(OH) by replacing every eighth OH by SO,. The interesting mineral 4 2 W. Feitknecht and G. Fischer ibid. 1935 18 40. 4 3 W. Feitknecht and W. Gerber ibid. 1945 28 1454. 260 QUARTERLY REVIEWS hemimrphite 44.ig one of the very small number of silicates containing the Si,O;- ion. Its behaviour on dehydration and its crystal structure show it to be Zn4(0H),(Si,0,),H,0. It is evident that a very large number of basic salts have a layer-lattice type of structure in which layers of hydroxide alternate with layers of normal salt.This appears to be due to the fact that many normal salts and many It has a layer lattice. X (b) F I G . 4 Plans of the structures of (a) Mg,SiO and (b) Mg(OH),. shaded circles Mg and open circles 0 atoms. distinguish between SiO tetrahedra at diflerent heights. Repeat unit Small black circles represent Si I n (a) light and heavy lines are used to T o the left in (b) the Mg-OH bonds are shown and to the right a n octahedral co-ordination group is outlined. x x y norbergite Mg,Si04,Mg(OH),. x y x x y chondrodite 2Mg,SiO,,Mg(OH),. x y x y x x y hurnite 3Mg,Si04,Mg(OH),. x y x y x y x x y .* clinohuinite 4Mg,Si04,Mg(OH),. (Reproduced by permission from Wells's " Structurnl Inorganic Chmistry ".) hydroxides have layer-lattice structures and that the lrations in both salt and hydroxide are surrounded octahecirally by oxygen hydroxyl or halogens.The structures of the layers of many salts and hydroxides are so similar for this reason that they are often able to key together and form very strong composite layers the composite structure being a basic salt. It is evident that the number of layers of normal salt and of metal hydroxide so bound 44 T. Ito and J. West 2. Krist. 1932 82 1. BASSEW BASIC SALTS 26 1 may be expected to depend upon various factors and to vary considerably although there may well be certain proportions of hydroxide and normal salt which are more stable than others. In the case of zinc rn we have seen the basic salts tend to have 4 mols.of Zn(OH) combined with one of zinc halide (or nitrate) but in the case of numerous cupric salts one finds 3Cu(OH) for each mol. of normal salt [atmamite CuC1,,3Cu( OH) ; brochantite CuS0,,3Cu(OH) ; Cu(NO3),,3Cu(0H),}. In all these cases several layers of hydroxide separate each layer of normal salt. The opposite type of structure in which one hydroxide layer is divided from the next by several layers of the normal salt also occurs and is found in the well-known chondrodite series of basic orthosilicate minerals. At one end of this series is the mineral brucite Mg(OH), and at the other olivine Mg,SiO,. The intermediate stages are norbergite Mg,SiO,,Mg(OH) which has single layers of Mg,SiO and Mg(OH) alternating chondrodite 2Mg,Si04,Mg(OH) where two layers of Mg,SiO alternate with one of Mg( OH), humite 3Mg,Si04,Mg(OH), and clinohumite 4Mg,Si04,Mg(OH) (see Fig.4). Some of the hydroxyl groups are generally replaced by fluorine as is so often the case with naturally occurring basic salts. This is an indication of the ionic character of the hydroxyls. Staurolite has a similar type of structure which can be regarded as composed of layers of cyanite interleaved with ferrous hydroxide layers. It can be written 2AI,SiO,,Fe(OH),. V Basic Salts containing Complex Silicate and A1 timinosilicate Anions.- The majority of the minerals belonging to the more complex types of silicate structures appear to be somewhat basic. This does not apply to the pyroxenes which contain simple chain anions (SiO,)2,-. The members of this group such as enstatife MgSiO and diopside CaMg(SiO,), appear to be non-basic.The amphiboles with the more complex %and ions (Si4011)i- all seem to be basic and usually to the extent indicated by the formula (OH),Ca,Mg,(Si,O,,)~- of tremolite. The (Si4011)i- band anions are bound together by the metal ions. There can be wide variation in the nature of these kations the number of which depends upon the extent to which any of the silicon atoms in the Si4Oll have been replaced by aluminium for every aluminium atom taking part in such replacement necessitates one more positive valency among the kations. I n hornblende the best-known of the amphiboles the anion has the composition [Si,Al,0,,]14-. The more such replacement of Si by A1 has occurred the more tightly the anions are held together as the kations are then more numerous.I n other cases the links are relatively weak and the crystals tend to break into long fibres ; asbestos is an amphibole. Hydroxyl bonds have little if anything to do with the behaviour of the amphiboles. This is not the case with chrysotile asbestos which is very closely related to the amphiboles. Its formula is (OH),Mg6(Si,011)~-,H,0. The magnesium ions lie between the oxygen atoms of the Si40, anions and OH anions so that it is chiefly the bonds between the OH groups of adjacent bands which hold the latter together. These are much weaker than the more ionic bonds of the less basic true R 262 QUARTERLY REVIEWS amphiboles. The crystals of this mineral split into fine fibres particularly If the band ions of the amphiboles are linked together endless sheet anions result. These usually contain hexagonal rings linked together &B in a honeycomb formed from six SiO tetrahedra but rings formed from 4 or 8 SiO groups are also found in a few minerals.Among the minerals containing these sheets of hexagonal rings [representable as (S&06),J are the clay minerals (kaolinite dickite w r i t e etc.) the micas and the chlorites. All of these are basic salts and they are so important technically in connection with agriculture and all the industries in which clay or clay products are used that a special " Clay Minerals Group " of the Mineralogical Society has been-formed recently for their study. Similar groups exist in fiance and other countries. It is possibly of some importance in connection with the structures of the clay minerals vermicuZites chlorites etc.that in some cases a t least they have arisen by degradation of mica and other silicates formed at high temperatures. Silicic acid sheet structures may well have survived from the mica structures or have been formed from the three-dimensional struc- tures of the felspar. The clay minerals and related compounds could have been built up from such silicic acid sheets a t ordinary low temperatures. This mode of formation would help to explain the fact that though crystal- line the crystallites of those clays which have been formed from felspars are extremely minute as a rule whereas crystals of the vermiculites which have been formed from micas are often quite large. They are pseudomorphs of the original mica but not in the usual sense of being polycrystalline aggregates. The silica skeleton of the original mica seems to have been retained almost intact.All the minerals of this group itlix. clays vermiculites chlorites etc. owe their formation essentially to the fact that in the silica sheet of the mica type all the vertices of the SiO tetrahedra point towards the same side of the silica sheet and that the distance apart of the 0 atoms a t these vertices in the hexagonal rings is very closely the same as the distance between the OH ions of the hexagonal rings in the layer structures of hydrargiEZite Al(OH) and brucite Mg(OH),. This enables the silica sheets and the hy- droxides to bond together in layer structures. J. W. Gruner 4 5 is especially responsible for the suggested structures of the resulting minerals and these have stood the test of time very well.Halloysite A1,0,,2Si0,,4H20 was considered by Gruner to have the simplest structure and to consist of sheets of silicic acid and aluminium hydroxide held together merely by hydroxyl bonds [ . . . (OH),AI,(OH) . . . Although the structure of this mineral is still under dispute the indications appear to be that it is essentially similar to that of kaolinite but with dis- ordered stacking of the layers. In the typical clay minerals (kaolinite dickite nacrite) interaction has occurred between the OH of the silica and 4 5 Z . Krist. 1934 88 412; Amer. Min. 1934 19 557; 1935 20 475. ectsily. 1 . . . O,Si,(OH) . . . BASSETT BASIC SALTS 263 alumina layers so that there is actual chemical bonding with formation of [Al,(OH),Si,O,],. These double layers are neutral and only held together by hydroxyl bonds in the crystals.These links are easily broken which accounts for the swelling and other physical properties of the clays. By coupling a second silica sheet on to the other side of the aluminium hydroxide layer there results the triple-layer compound pyrophyllite [A12(0H)2Si,0,,] which is also a neutral structure. I n this case the hydroxyl groups are right in the middle of the triple layer and quite close together (cf. the remarks on p. 256) so that in the crystals the triple layers are only held together by weak van der Waals forces. Talc is the corre- sponding triple-layered magnesium compound [Mg,( OH),Si,O,,] no magnesium compound corresponding to kaolinite is known. M~ntmorillonite,~~ the chief constituent of fuller’s earth and bentonite is derived from pyrophyllite by the intercalation of an indefinite amount of water between the triple layers of the latter but there has always been some replacement of A1 by Mg or of Si by A1 (or both) together with the inter- calation between the layers of exchangeable zeolitic kations needed by such replacement (see the section on micas).The vermiculites 47 have a structure which seems to be related to that of both the bentonites and chlorites. Owing to the macro-crystalline nature of the vermiculites their general physical character seems very different from that of the bentonites with their submicroscopic crystals but the inter- calated layer of water molecules (probably with a regular orderly arrange- ment) is present all the same as is shown by the enormous swelling and exfoliation which occurs on heating.The c-spacing in the crystals is quite characteristic and differs from that of the bentonites. The true micas 47 differ from pyrophyllite or talc in having one in four of the silicon atoms replaced by aluminium. This necessitates the presence of an ion having one positive charge for every silicon atom so replaced. In both muscovite K~[Al,(OH)2(Si,A101,)]~ (the mica derived from pyro- phyllite) and phlogopite Ki[Mg,( OH)2(Si3A1010)]~ (mica derived from talc) these positive charges are provided by potassium ions. These are located between the triple-layer anionic structures which are thereby held much more tightly together than in pyrophyllite or talc. The electrostatic forces are still moderate however so that cleavage remains excellent. In the so- called brittle micas half the Si atoms have been replaced by Al so that twice as many positive charges are required as in muscovite.Margarite is Ca;[Al,( OH),(Si,Al,Olo)]~ and the sheet anions are now held together so strongly that cleavage is much less good and the mineral is much harder. The chlorifes 48 provide the last variant of these structures containing siliceous sheets. They are derived from talc by replacing some of the silicon atoms by Al and the triple anionic sheets range in composition between [Mg,(OH),( AlSi3010)]; and [Mg,Al( OH)2(A12Si2010)]; the greater replace- 4 6 J. W. Gruner Zoc. cit. ; s. B. Hendricks and M. E. Jefferson Amer. Min. 1938 47 L. Pauling Proc. Nut. Acad. Sci. 1930 16 453. 48 L. Pauling ibid. p. 578. 23 863; F. A. Bannister Ann. Reports 1938 35 190. 264 QUARTERLY REVIEWS ment of silicon in the second case being made up for by some replacement of Mg by Al.The negatively charged infinite sheets arising in this way are held together in the crystals not by simple positive ions as in the micas but by other positively charged infinite sheets derived from layers of the brucite Mg(OH) structure by replacement of one-third of the Mg by Al. Chlorite may be written as [Mg,Al( OH),]; [Mg,(OH),(AlSi,OIo)];. Silicates with three-dimensional network anions may also be basic; thus the important mineral epidote may be written as Ca2( OH)[(AI,Fe),Si,O,,]. Although the complex negative structures of the silicates have been called anions and considered to be held together by covalencies in the above dis- cussion yet it has to be remembered that the Si-0 link has some ionic character.This may be considerable in respect of some of the links especially those involving Mg and A1 in octahedral co-ordination in the brucite and hydrargillite portions of the complex sheet structures of the pyrophillites micas chlorites etc. VI. Basic Salts with Discrete Molecules.-The only basic salts which come into this category appear to be the remarkable basic beryllium acetate Be,O(CO,CH,) and the similar compounds Be,O(CO,R) (where R = ethyl propyl isopropyl or tert.-butyl). The beryllium atoms in the acetate are arranged tetrahedrally around the central oxygen and the six acetate groups straddle the six edges of the tetrahedron. The structure is truly molecular and the links essentially c0valent.4~ The basic acetate crystallises in the cubic system but a completely symmetrical arrangement of the acidic groups along the tetrahedral edges is only possible in the case of the acetate (and formate).A great decrease in symmetry is found in the basic pro- pionate which is monoclinic. Apart from this difference in symmetry the general arrangement in the molecules of all the above compounds isi no doubt very much the same. I n conclusion the writer wishes to acknowledge the great help he has received in preparing this review from A. F. Wells’s “ Structural Inorganic Chemistry ”. He also thanks Dr. Wells and the Clarendon Press for per- mission to reproduce the four figures and Dr. Max Hey for helpful advice. W. H. Bragg and G. T. Morgan Proc. Roy. SOC. 1923 A 104 437 ; L. Pauling and J. Sherman Proc. Nat. Acad. Sci. 1934 20 340; G. D. Preston and J. Trotter Nature 1943 151 166 ; C. A. Beevers ibid. 1943 152 447.

ISSN:0009-2681    

DOI:10.1039/QR9470100246

出版商:RSC    

年代:1947

数据来源: RSC

摘要:

SOME THERMODYNAMIC PROPERTIES OF HIGH POLYMERS AND THEIR MOLECULAR INTERPRETATION By GEOFFREY GEE Sc.D. (DIRECTOR OF RESEARCH THE BRITISH RUBBER PRODUCERS’ RESEARCH ASSOCIATION) 1. Introductory THE past decade has witnessed a rapid growth in the understanding of the thermodynamic properties of polymers and their solutions. This has been achieved largely by the development of statistical theories of rubber-like elasticity and of the free energy of mixing of polymers with liquids. A number of reviews are available dealing with portions of this field of work but for the most part they are addressed to specialist workers. The object of this Review is to survey in broad outline a number of related topics without entering into detailed discussion of statistical theories. The emphasis will therefore be laid on the experimental thermodynamic data which form the basis for these theories and on their physical significance.An attempt will be made to develop the argument so far as possible in physical terms although it must be realised that this is essentially a field in which we are concerned with quantitative measurements and their mathematical inter-relations. One fundamental experimental difficulty which is common to nearly all the problems to be discussed below is that of ensuring that the system under investigation has reached a state of equilibrium. The significance of the concept of equilibrium in measurements on polymers has recently been very clearly discussed by A. R. Ubbe1ohde.l The difficulty arises from the fact that many processes in a polymer take place so slowly that they may to a good approximation be said not to occur a t all during the time involved in an experiment.When this is the case it is clear that the system cannot be assumed to reach a state of equilibrium with respect to this particular process. It is indeed quite common to find that certain properties of a polymer are very dependent on the previous history of the specimen ; examples will be given later. Even when this is the case the system may still be in equilibrium with respect to other possible changes and it is therefore permissible to apply the thermodynamic criteria of equilibrium.* Care is needed in relating the experimental results to theories in order to make sure that the theory is not based on the assumption of equilibrium with respect to changes which are in fact so slow as to be virtually negligible.The usefulness of considering partial equilibria in this 1 Faraday Society General Discussion on “ Swelling and Shrinking ” Sept. 1946. * This behaviour is of course not confined to polymeric systems although it is more frequent there than in systems containing only components of low molecular weight. Familiar examples are furnished by the existence of supercooled liquids in equilibrim with vapour and the stability of mixtures of oxygen and hydrogen. 265 266 QUARTERLY REVIEWS way depends on the possibility of choosing a time scale for the experi- mental work long enough to permit the rapid processes to be complete and at the same time short enough to exclude other slower processes. 2. Crystallisation of Polymers The most comprehensive study of the crystallisation of a polymer is the work on natural rubber ; this has been summarised by L.A. Wood.2 The phenomena encountered are very much more complicated than in the freezing of a liquid. Apart from the possible occurrence of supercooling the latter is a sharp phase change occurring at a perfectly definite tem- ?" Crystalisation temperature. Fra. 1 Melting range of natural rubber a s a function of the crystallisation temperature. perature. Crystallisation is accompanied by the evolution of the latent heat L and a volume change -A V . The freezing point T can be altered by the application of a hydrostatic pressure P in accordance with the familiar equation dT,/dP = TdV/L . * ( 1 ) In contrast with this simple behaviour a polymer cannot be said t.0 have any single freezing point.The crystallisation of an amorphous polymer by cooling is very dependent on the rate of cooling. If a polymer is allowed to crystallise and then heated melting takes place over a temperature range and this range depends on the temperature a t which crystallisation was " Advances in Colloid Science " Interscience Publishers 1946 pp. 57-93. E. A. Guggenheim " Modern Thermodynamics " Methuen 1933 p. 55. GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 267 carried out. Figs. 1 and 2 illustrate some of the resulting complexities in the melting of crystalline natural rubber. Fig. 1 shows that after crystallisation at a given temperature it is necessary to raise the tem- perature by some 5" in order to cause melting to start and by as much as 20-30" to complete it.It must be emphasised that a time factor is involved in the data thus represented. It is clear from Fig. 1 that crystal- Iisation is possible within the temperature range in which melting occurs and it is only possible to obtain such data because melting is a much faster process than crystallisation. If crystallisation is carried out at - 30" and the temperature then raised to 0" the crystals will melt but further crystallisation will then slowly occur producing crystals which do not melt completely until the temperature is raised to 15". Another possibility FIG. 2 the direction of temperature changes. Double melting range in the melting of the sample of '' stark " rubber. Arrows indicate V,y is the specifi volume. implicit in Fig. 1 is the co-existence of two sets of crystals having different melting ranges.This possibility has been verified experimentally as shown in Fig. 2. A sample of " stark " rubber which had crystallised slowly at room temperature had a melting range of 32-39". A sample waa allowed to crystallise further a t 2" ; on warming the crystals were found to melt in two discrete ranges between 5" and 15" and then between 32" and 39". The phenomena observed in the melting of polythene are much simpler and provided that measurements be made slowly a reversible volume- temperature relation is found,6 as shown in Fig. 3. The three curves refer From the data of N. Bekkedahl and L. A. Wood J . Chem. Phymk 1941 9 193. From the data of L. A. Wood op. cit. ref. 2 p. 70. E. Hunter and W. B. Oakes Trans. Faraday Soc. 1945 41 49. 268 QUARTERLY REVIEWS to three different samples of polythene and it will be noted that although there is a sharp break at the point where crystals first appear on cooling the amorphous material the volume does not change discontinuously at this point.The interpretation placed on the region in which an abnormally large coefficient of expansion is observed is that it is a range in which the degree of crystallisation varies continuously with the temperature. Similar behaviour is shown by the heat content,' and R. B. Richards 8 has estimated from these data the fraction 8 in the amorphous state as shown in Fig. 4. It will be noted from Fig. 4 that crystallisation is never complete an observation which is readily understood in terms of the molecular picture MO G FIG. 3 volume-temperature curve9 for three samph of polythene.(Reproduced by permisewn from Transactions of the Faraday Society 1945 Q1 '51.) of the crystallisation of a polymer. The individual crystallife is shown by X-ray evidence to be small of the order of 200-500 A. side in the case of stretched natural rubber.g It follows that the units from which the crystallites are built are almost certainly not whole molecules but por- tions of molecules and that a single molecule may pass through several crystallites as indicated in Fig. 5.1° It is clear that if crystallisation com- mences from a number of points simultaneously growth from these centres must leave amorphous regions which cannot be incorporated into crystals without a very large-scale reorganisation involving the temporary melting of many crystallites. Such a process would be so slow as t o be quite negligible.7 H. C. Raine R. B. Richards and H. Ryder {bid. p. 57. * Ibid. p. 127. J. Hengstenberg and H. Mark 2. Km'at. 1928 69 271 ; S. D. Gehman and J. E. Field I d . Eng. Chem. 1940 32 1401. lo L R. G. Treloar Rep. Prog. Phyeics 1943 9 118. GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 269 In considering how to apply thermodynamic methods to the crystallisa- tion of polymers two different problems are involved. The first is that of the equilibrium of an individual crystallite in its own environment. There * rl5 loo 8'0 60 do ,/ 20 5 Gmp." C . / / 0'20 / * I 4 / I 4 1 / 0.30 l9 040 FIG. 4 Crystallinity of polythne as a function of temperature. (Reproduced by pemziesion from Transactions of the Faraday Society 1945,4l 132.) is no difficulty in writing down equations analogous to (l) for the effect of the various stresses which may be imposed on the crystallite by its (a) tb) FIG.5 Molecular structure of a crystalline polynter (a) unatretched (b) stretched. (Reproduced permiasion fm Reports Of Progress in Physice 1942-43,9,118.) environment and it is evident that when the material irJ in a steady state at a fixed temperature each individual crystallite must be in equilibrium with the stresses on it A qualitative development of this idea provides 270 QUARTERLY REVIEWS a ready explanation of the phenomena which have been described above. If a polymer is cooled quickly t o a temperature at which crystallisation is possible a number of crystallites will start to grow at points where chains are suitably oriented.Growth will then take place by the incorporation of neighbouring chains but these can only be brought into suitable posi- tions by the slow process of diffusion. The high viscosity of the polymer and the mechanical entanglement of the chains retard growth not only by the slowness of diffusion but also by setting up stresses which are only slowly relieved. At any stage of the growth therefore the crystallite is nearly in equilibrium with its environment to any rapid change of tempera- ture. Hence if a t any stage in the crystallisation the temperature is raised a few degrees (rapidly) the crystallite will start to melt under the influence of the existing stresses. Furthermore since melting involves no bulk transport of matter it is a rapid process and a considerable rise of temperature assisted by the stresses in the environment may therefore cause complete melting even a t a temperature a t which crystallisation is still possible.This is believed to provide an acceptable explanation of the phenomena represented in Figs. 1 and 2 and of the very striking fact,ll that the melting range shown in Fig. 1 is independent of the extent of crystallisation. The fact that such phenomena have not been reported for polythene is to be ascribed to the higher temperatures involved and the lower viscosity of the material; the whole time scale is therefore greatly contracted and the stresses in the individual crystallites due to the sur- rounding amorphous region dechy in times comparable with the period of observation. I n these circumstances the individual crystallites will grow until they begin to interfere with one another's growth.Thermodynamically this is represented by a system of stresses between the crystallites ; incor- poration of more amorphous material into the crystallites involves an increase in these stresses and lowers the melting point.12 An approximate quantitative treatment of this problem has been attempted by E. M. Frith and R. F. Tuckett l3 and by R. B. Richards.* The basis of the treatment is the assumption that on the average each chain is partly in an amorphous region and partly in a crystalline region. Growth of the crystallites restricts the freedom of the amorphous portions and thus diminishes the configura- tional entropy of the system.* It is found possible in this way t o account semi-quantitatively for the melting range of polythene shown in Fig.4. It is clear from the foregoing argument that this theory should not be applied to the melting of natural rubber where the equilibrium assumed in the theory is not attained experimentally. The second thermodynamic problem involved in the crystallisation of polymers is the inter-relation of such quantities as can be measured experi- mentally. I n this case we are not concerned with individual crystallites or stresses within the material but simply with observations made on the l1 Op. cit. ref. 2 p. 66. l2 T. Alfrey and H. Mark Rubber Chem. Tech. 1941 14 625. l3 Trans. Faraday $oc. 1944 40 251. * Compare Q 5 , GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 271 is plotted as a function of temperature there may or may not be a region in which occurs a phase change of the type described in the previous sect'ion.Quite generally a point is found a t which the slopes of these curves change sharply as indicated in Fig. 6. This discontinuity is described as a transition of . .!? $ $ polymer in bulk. Although only a limited equilibrium condition is achieved measurements of the changes of volume and heat capacity accompanying crystallisation under specified conditions do refer to definite thermodynamic processes even though it may not be possible physically to describe pre- cisely what these are. As an illustration we may consider the application of equation (1) to the crystallisation of natural rubber. The melting point * of a sample of stark rubber was found l4 to be increased by pressure according to the equation where p is the pressure in bars.Putting p = 1 we find T = 311 and dT/dP = 0.037 deg./bar. The latent heat of fusion L of a different sample of stark rubber having a melting point of 284" K. was estimated 15 to be 4.0 cals./g. = 167 bars c.c./g. Equation (1) then gives for the expansion on melting This is in accord with the values found experimentally for similar samples of rubber.lO Thus the overall process of crystallisation can be treated thermodynamically as if it were a simple phase change ignoring the physical complexities of the process. 10glo(p + 1300) = 5.9428 - 875/T LIP' = 167 x 0.037/284 = 0.022 c . c . / ~ . -0.98 $1 -0.36 -8p' -69' -48 -0.94 l4 L. A. Wood N. Bekkedahl and R. E. Gibson J . Chem. Physics 1945 13 475. l5 N. Bekkedahl and H. Matheson J . Ree. Nut. Bur. Stand. 1934 13 411 ; 1935 l6 Op.cit. ref. 2 pp. 1-55. * The temperature at which melting waa complete. 15 503. 272 QUARTERLY REVIEWS TABLE I Second-order transition points for Some hydrocarbon polymers Material. 1 Temp. Material. 1 Temp. /I I II I Polyisobutylene . . . Polystyrene . . . . Natural rubber . . . Polyindene. . . . . GR-S rubber . . . . Recently doubt has been thrown on the validity of considering this transition as a thermodynamic singularity. It has been shown 1' that the transition temperature of polystyrene can be very greatly lowered by allowing more time for the attainment of equilibrium in studying the effect of temperature on volume. When the temperature is lowered say from 80" to 75" a rapid contraction occurs apparently complete within the time required for thermal equilibrium giving a thermal expansivity of ca.2.7 x If the temperature is maintained at 75" for several hours a further slow contraction occurs the final volume change giving a thermal expansivity of ca. 4.5 x deg.-l equal to the value found above the transition point. In other words by reducing the rate of cool- ing the transition point can be lowered from 81" to below 75". By pro- ceeding sufficiently slowly R. s. Spencer and R. F. Boyer l7 were able to reach a temperature of 20" without encountering a transition point. Similar observations have been recorded for a number of other polymers and these authors have therefore concluded that the transition points are not true thermodynamic singularities a t all. The Reviewer considers this to be a misleading conclusion. It is obvious that the transition is very dependent on the rate of measurement but as we have seen the existence of pro- cesses slow compared with the time of an experiment does not preclude the application of thermodynamic reasoning to the limited equilibrium attained under these conditions.Exactly the same considerations apply here; a t a specified rate of heating there is clearly an abrupt change in the mechanism of expansion at a certain temperature or a t any rate within a small temperature range. This temperature can quite properly be re- garded as a thermodynamic singularity in spite of its dependence on rate. As an illustration of the usefulness of this method of approach consider the effect of pressure on the transition temperature. Equation (1) is not directly applicable to this problem since dV and L are both zero.In order to evaluate the ratio A V / L suppose the transition from one state to the other to occur at a temperature AT above the equilibrium transition point. A V would then be given by V . Aa .d T where V is the molar volume of the material and da the difference in expansivities above and below the transition. Similarly the latent heat of the transition would no longer be zero but dC,.dT where AC, is the difference in the specific heats abme and below the transition. deg.-1. Combining these we have d V / L = Vda/dCp . (2) . l7 J. Appl. Physics 1946 17 398. GEE THERMODYNAMIC PROPERTIES OF HIQH POLYMERS 273 Since this is independent of AT it will also be true at the transition point Hence from (1) This equation has been used by W. H. Keesom in discussing the second- order transition between helium I and helium I1 ; it does not appear to have been applied t o polymers.A. H. Scott lS found that the transition point of a sulphur vulcanisate of natural rubber containing 19.5% of combined sulphur was increased from 36" to 45" by applying a pressure of 800 bars. Thus for this material aT/aP = 0.011 deg./bar. Values of doc and ACp for rubber hydrocarbon can be obtained from t'he data of N. Bekkedahl and of Bekkedahl aiid H. Matheson.15 aT/aP = T V . A ~ / A C . ' (3) These give V . 3 r = 0.0004 c.c./g. deg. ; dC = 5 bars c.c./g. deg. Hence aT,/aP = 200 x 0.0004/5 = 0.016 deg./bar. order as the value found by Scott for the vulcanisate. This is of the same 4. Ricbber-like Elusticity It is a simple matter to extend the usual discussions of the thermo- dynamic properties of a material to take account of the work done on the body when a force is applied to it.20 In general experiments on the elastic extension of materials are carried out isothermally and a t atmospheric pressure ; in these circumstances the force f required to produce a simple est,ensioii of the length 1 is related to the Gibhs free energy of the material as follows - (4) This increase in the Gibbs free energy which occurs on stretching can be divided in the usual way into changes in heat content ( H ) and entropy (8).To do this it is necessary to study the effect of temperature on the force required to maintain a fixed length 1. We then have To a very good approximation for materials in a condensed phase changes of heat content during processes a t normal pressures may be equated to the corresponding changes of internal energy ( E ) ; this approximation will be made throughout the following discussion.Equations (4)) ( 5 ) and (6) are perfectly general and apply t o any elastic material. Very different values are however found when these equations are applied on the one hand to metals and on the other to rubber-like polymers. dynamics " 1940 p. 331. l8 Leiden Communications Supplement 80b ; cf. J. K. Roberts " Heat and Thermo- l9 J . Res. Nut. Bur. Stand. 1935 14 99. 2o G. Gee Trans. Faraday SOC. 1946 42 585. 274 QUARTERLY REVIEWS Some typical data are plotted in Fig. 7 21 for steel and in Figs. 8 20 * and 9 22 for pure gum vulcanisates of natural rubber using as variable not the length 1 but the extension ratio a equal to Z/Zo where 2 is the unstretched length.Comparing the two figures a number of striking differences are apparent (1) The great difference of extensibility viz. ca. 1% for steel up to 1 0 0 0 ~ o for rubber. (2) The very much larger modulus of elasticity of steel (ca. lo5 times larger than for a typical rubber). (3) The Therniodynun ics of the elastic extension of steel. relntivc impor'tance of thc eiitro1)y and energy contributions. The exten- sioii of steel is accompanied by a consitiernble increase of entropy the rate of iriercasc being yracticallx (*onstant and the force may be said to be due in the main to the increase of internal energy. Sufficiently small exten- 21 Data from " Physical and Chemical Constants " (G. W. C. Kay0 and T. H. Laby) 2 2 L. A. Wood mid F.I,. Itotli J . rlppl. Physics 1944 15 781. * A numerical slip in tlie original lias been corrected. Longmnns 1935 pp. 29 and 30. GEE TRERMODYNAMICI PROPERTlES OF HIGH POLYMERS 275 eiom of natural rubber also involve increases of both entropy and internal energy but at larger extensions the contribution of internal-energy changes to the force becomes relatively unimportant. Over most of the extension range the entropy of extension is negative and makes the major contri- bution to the observed force. (4) The extension of steel is accompanied by a considerable bulk expansion ( A V),* reflected in its low Poisson's ratio. Small elongations of natural rubber take place with only very small volume L 0.10 0.005- Extension r a t i o OL. s n S FIG. 8 Thermodynamics of the elastic extension of rubber.changes (increase) but a t large extensions there is a considerable bulk contraction. Taking these observations in this order we shall now consider what qualitative conclusions can be drawn as to the molecular mechanisms involved. It is clear in the first place from the very high extensibility and low modulus of elasticity that the processes involved in the deformation differentials. * Note that d V is the total expansion ; the other quantities in Figs. 7-9 are all 276 QUARTERLY REVIEWS of rubber-like polymers differ fundamentally from those operative in metals. In the latter extension involves essentially the separation of atoms without change of their relative positions ; the possible movement is small and the opposing forces large. The large deformations possible in rubber-like materials must necessarily involve considerable relative movements of molecules and the forces applied are too small to produce appreciable changes of interatomic or intermolecular distances.The most striking thermodynamic feature of the extension of rubber is the large decrease of entropy which occurs over a wide range of deforma- tion. This has been interpreted in terms of the decreased freedom of the I 2 3 4 5 €xiension ratio a. FIG. 9 Thermodynamics of the elastic extension of rubber. bouring chains to which they are individual molecules. Rubber- like polymers are essentially linear in structure and their molecules possess a high meapure of flexibility. In general most or all of the links in the chain are single C-C 'bonds about which rotation can occur more or less freely.As a result the chain takes up a series of ever- changing configurations. Of all these configurations only one will give the chain its maximum outstretched length whereas in most configurations the chain will be extensively kinked with its ends much closer together. The overall shape and length (i.e. end-to-end distance) of any individual molecule will tend to change randomly but it is evi- dent that a molecule must have some statistically most probable length. In a piece of rubber the molecules are not altogether free their motions being re- stricted by the presence of neigh- held by van der Waals forces. Vulcanised rubbers possess in addition a number of cross links where molecules are joined chemically. Qualitatively these restrictions do not affect the essential features of the picture.In the undeformed rubber the molecules will tend to be highly kinked ; extension of the rubber will involve a net straightening of molecules which are thus constrained to take up less probable configurations. It is this process which accounts for the observed decrease in entropy on extension. The expansion of a material when stretched is a normal feature of the elastic behaviour of all isotropic materials. Its origin may be understood GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 277 physically by considering the applied force as made up of two shears and a hydrostatic tension. So long as the material is isotropic the effect of a tensile force is the same as that of two shear stresses which change the shape of the material without affecting its volume and a hydrostatic tension which causes the material to e~pand.2~ The hydrostatic component is equal in magnitude to one-third of the tensile stress.Hence if Y is the Young’s modulus of the material [= (af/aZ)p,T] and K its compressibility [= - (av/aP)T/v] the relative expansion d V/v produced by a small extension (in the ratio a ) is given by A V / V = i Y K ( a - 1) . * (7) In terms of the Poisson’s ratio a we have the familiar expression 2 0 = 1 - & Y K . - ( 8 ) The very small expansions accompanying the extension of rubber (am0.4999) are thus directly related to its low Young’s modulus. The contraction observed at larger elongations is evidence that the material is no longer isotropic. The large contraction observed in natural rubber a t elongations greater than about 300 yo arises from crystallisation and may indeed be employed as a relative measure of the extent of crystallisation.lo In these volume changes the packing of the molecules is changed and work is done by or against the intermolecular forces. Consequently changes occur in the entropy and internal energy of the material an expansion being accompanied by increases of both quantities. The effects of these changes on the free energy nearly cancel thus leaving the force required to stretch the rubber almost the same as it would have been had the volume been maintained constant by applying a hydrostatic preasure.20 To a first approximation the increase of internal energy on stretching produced in this way will be equal to the increase resulting from the application to the unstretched material of a hydrostatic tension sufficient to produce the same expansion.This will be strictly true for small elongations of any isotropic material. Now this quantity can be calculated thermodynamically and it can thence be shown that ( Z ) P T f = O (”> al P,T f = O - lYPT . (9) where is the coefficient of cubical expansion. This is a perfectly general relationship holding e.g. both for steel (Fig. 7 ) and for natural rubber (Fig. 8). Two consequences may be noted (a) Experimental data on an isotropic material which do not agree with equation (9) must necessarily be wrong. ( b ) The observed increase of energy and entropy for smaEZ exten- sions of rubber are completely unrelated to the mechanism of deformation. K. H. Meyer and A. J. van der Wyk 24 have recently pointed out that measurements of the shear modulus of rubber are free from the complica- tions arising from these volume changes since a shear stress has no hydro- 23 A.E. H. Love ‘‘ The Mathematical Theory of Elasticity ” Cambridge 1934 p. 83. Helv. Chim. Acta 1946 29 1842. s 278 QUARTERLY KEVIEWY static component. shear then for any isotropic body It is indeed easy to show that if m is the amount of ( a H / a f 4 P T f - O = 0 * (10) 5. The Elasticity of an Ideal Network The main conclusion from the thermodynamic data of Figs. 8 and 9 is that the restoring force in a stretched rubber arises from the decreased entropy associated with the straightening out of the randomly- kinkcd molecules. Several attempts have been made to calculate the force by applying the methods of statistical mechanics to an idealised model of the system.25 To do this it is necessary to compute the number of configura- tions go of the system before stretching and the reduced number g after stretching.The deformation being assumed to take place without change of volume or internal energy the entropy increase AS is then given by the application of Boltzmann’s equation by Afid = kin (Sl/SO) * - ( 1 1 ) The model employed in all the treatments published hitherto represents the molecule by a series of equal links joined end to end and oriented at random. No account is taken of the volume of the links and impossible configurations in which two links occupy the same position in space are not excluded from the computation. These chains are linked together at a number of points so as to constitute a three-dimensional network in which the length of a segment i.e.the distance measured along a chain from any junction point to the next is constant. For not too large ex- tensions the force per unit area of the unstrained rubber a t an extension ratio a is given by an expression of the form f = B(a - 1/a2) . * (12) More generally the work W done in deforming a unit cube to a block of sides A, A2 A (where &A2& = 1) is found to be w = *B(A? + + 2; - 3) . * (13) The quantity B has been variously estimated we shall adopt here the value given by F. T. Wall,25 which may be put in the alternative forms B = NkT = pRT/Mc . - (14) where N is the number of chain segments M is the “ molecular weight ” of a segment and p is the density of the rubber. Equations (12) and (13) are valid.only for small deformations ; if any of the chains approach their fully stretched-out length (12) must be replaced by a more complex expression due to Guth and James,25 which may be written 26 A systematic treatment with brief reference to the contributions made by other workers is given by E. Guth H. M. James and H. Mark op. cit. ref. 2 pp. 253-298; cf. also L. R. G . Troloar Trans. Faraday SOC. 1945 41 83. GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 279 where a. is the limiting extensibility of the network and L-1 denotes the inverse Langevin function defined by 1 L(z) = coth x - ; = y . * (16) L-'(y) = x . - (17) It will be noted that equations (13) and (14) define the entire elastic behaviour of a rubber a t not too large deformation in terms of it single molecular parameter which measures the extent to which the rubber is cross-linked.To take account of large deformations it is necessary t o introduce as a second parameter the ultimate extensibility. L. R. G. Treloar 25 has related this also to the degree of cross-linking thus leaving only one adjustable parameter available to fit the whole stress- strain curve. These theor e tic a1 s tress-str ain curves may be compared with experiment in a variety of ways. L. R. G. Treloar 26 has examined the validity of the form of equation (13) by measuring on a single specimen of rubber stress-strain curves in (1) elongation (2) com- pression (3) shear and (4) a comtination of (1) and (3). Some of his results are shown in Fig. 10 ; tlic moht striking feature of them is the way in \vhich a single parameter dwcrihes the elastic hehaviour in four different types of measurement.The agreement between theory and experiment is not quantitatively accurate but the Reviewer has shown 2o that a considerably closer fit is obtained if experiments are carried out on rubber highly swollen with a liqiiid. The elastic behaviour can FIG. 10 Stress-strain data for natural rubber vulcan- isates in (a) compression (b) simple elongation ( c ) shear. Broken curves calculated. be more accurately described by the tii'o-constant equation deduced by H. Mooney 27 on more general mathe- mat ical grounds not involving any molecular model. Equation (15) leads to a stress-strain curve in qualitative agreement with Fig. 9 f becoming infinite a t a = a,,. By treating a. as an adjustable parameter the stress-strain curve of a pure gum natural rubber has been reproduced almost exactly up to 400% elongation.28 The value of a.2 G Ibid. 1943 39 59. 2i J. -4ppl. Physics 1940 11 582 ; R. Rivlin ibid. 1947 18,444 ; L. R. C . Treloar 28 H. 31. James and E. Guth J . Chem. Physics 1943 11 455. Proc. Physical SOC. in the press. T 280 QUARTERLY REVIEWS required is however considerably less than that calculated by L. R. G . Treloar from the degree of cross-linking.Z5 As we shall see below the form of the curve a t high elongations is in any case highly dependent on crystal- lisation ; equation (15) would not be applicable in such circumstances. The final test to be applied concerns the significance of the degree of cross-linking. K. H. Meyer and A. J. Van der Wyk 29 have recently argued that the elastic behaviour of rubber is not determined by cross-linking as equation (14) requires but their data obtained by applying very small deformations for short times hardly seem to constitute an appropriate test.P. J. Flory 30 has shown that the values of M deduced from the elastic behaviour of a series of butyl rubber vulcanisates are in satisfactory agreement with values deduced from their solubility. A similar conclusion has been reached by the Reviewer 31 by comparing M with the amount of sulphur combined in a series of sulphur vulcanisates of natural rubber. The statistical theory of an ideal network thus appears to give a very satisfactory semi-quantitative account of the equilibrium elastic behaviour of real rubbers relating the elastic moduli to a definite molecular property of the rubber.6. Crptallisation of Stretched Polymers The crystallisation of polymers on stretching presents two thermo- dynamic problems (a) the effect of a tension on the melting point of a polymer and ( b ) the effect of crystallisation on its elastic behaviour. The first is closely analogous to the effect of a hydrostatic pressure already discussed in 5 2. It may be shown 32 that the effect of a tension f is given by where A2 is the increase in length of a specimen (held under constant ten- sion temperature and pressure) consequent upon the occurrence of a small amount of crystallisation and Lr is the latent heat of fusion of this same amount of crystalline polymer. There do not appear to be any data avail- able for comparison with this equation but a very simple argument shows that dl must be positive.If we consider an amorphous polymer held at constant length crystallisation will only occur if the Gibbs free energy is thereby reduced Since the tension f is equal to aG/aZ it follows that f must be reduced by any spontaneous process. The reduction of tension on slow crystallisation is in fact a well-known phenomenon; under cer- tain conditions f may actually fall below zero as shown by a tendency for the material to become bowed.33 With the experiment in mind it is easily seen that probable values of A2 and Lr may well make aT/af of the order of 1 deg./kg. for a specimen of 1 cm.2 cross-section. This is con- sistent with the observation that natural rubber can be made to crystallise by stretching even a t temperatures nearly 100" above its normal melting J .Polymer Science 1946 1 49. 31 J . Polymer Science in the press. 3 3 W. L. Holt and A. T. McPherson Rubber Chem. Tech. 1937 10 412. 301nd. Eng. Chem. 1946 38 417. 3 2 G. Gee unpublished work. GEE THERMODYNAMIU PROPERTIES OF HIQH POLYMERS 281 p0irlt.~4 The molecular explanation of this rise of melting point on stretching is of course to be found in the improved alignment thus produced. The effect of uystallisation on the elastic behaviour of a polymer is complex. We have already noted that crystallisation reduces the tension in a sample held at constant length. On the other hand once crystalliw are formed they increase the modulus of elasticity because they act effec- tively as cross links surviving unchanged during further elongation of the rubber although by their presence they prevent the rubber from taking up its most stable state.This behaviour is due of course to the extreme improbability of the co-operative process needed to permit the crystallites to melt and then re-form in more favourable configurations. The extension 14 FIG. 11 Stress-atrain curves for a natural rubber vulcankate at three temperature%. of a polymer when crystallisation is occurring is thus not an equilibrium pro- cess and the statistical equations noted in § 5 should not be applied to it. The considerations sct out in the last paragraph have an important bearing on the experimental problem of studying the thermoelastic behaviour of crystallisable polymers. If a series of complete stress-strain curves are plotted at different temperatures the elongation a t which crystallisation commences increases progressively with rise of temperature with the result that at high temperatures the polymer tends to be more highly extensible and its stress-strain curve is generally below that found at lower temperrt- tures ; some typical results for natural rubber 35 are given in Fig.11. At 3 4 J. E. Field J . Appl. Physics 1941 12 23. 35 G . Gee and T. A. Sharpley unpublished. T* 282 QUARTERLY REVIEWS the same time if the length of the specimen is fixed while the temperature is vaned af/aT is found (at nearly any elongation) to be positive. The explanation of this discrepancy is that in comparing complete stress-strain curves measured a t different temperatures we are in effect comparing two different materials ; the apparent temperature coefficient obtained from mch a comparison has no thermodynamic significance.7. The Absorption of Vapours by Polymers Vapour-pressure curves of swollen polymers fall broadly into two types represented by the data for benzene in rubber 36 and agar-agar in water,37 illustrated in Fig. 12. The former are everywhere convex towards the vapour-pressure axis while the latter are sigmoid with the opposite curva- ture at low vapour pressures. Thermo- .L dynamically the vapour pressure is related to the Gibbs free energy of dilution denoted by the symbol AGO. This is defined as the increase in the Gibbs free energy of the whole system when one mole of liquid is transferred from a reservoir of pure liquid to a large bulk of the swollen polymer. Alternatively we may say that AGO is equal to the difference of chemical potential of the liquid in the swollen polymer and in the pure liquid.It may be shown 36 that for any polymer liquid system FIG. 12 AGO = R T I n ( g / & ) . . (19) Vapour presmre curves of (a) benzene in rubber (b) agar-agar in water. where pp," and p are the vapour pres- sures of the swollen polymer and of the pure liquid. The only assumption involved in the derivation of equation (19) is that the vapour behaves as a perfect gas. This is usually a fairly good approximation which can be corrected if necessary by replacing the vapour pressures by fugacities. No assumptions are made about the nature or structure of the polymer and the equation is in fact valid for solutions generally. The free energy of dilution can be separated into the heat AH and entropy AS, of dilution if vapour-pressure data are available a t different temperatures the necessary relationships being G.Gee and L. R. G. Treloar Trans. Furaday SOC. 1942 38 147 ; G. Gee and R. Friche and J. Ltike 2. Ekktrochem. 1930 86 309. W. J. C . Om ibid. 1946 42 607. 283 The temperature coefficients niust be measured at constant weight com- position of the swollen polymer and it is important to note that they are frequently sufficiently small for vapour pressures to require correction for departure from perfect-gas behaviour. When this type of analysis is applied to the data of Fig. 12 very dif- ferent results are obtained as shown in Fig. 13. The heat and entropy of dilution of rubber by benzene are both positive and this is generally found to be the case for non-polar polymers.38 Highly exothermic swelling is characteristic of the absorption of water by polymers in which it is appreciably soluble.These invariably contain polar groups t o which water molecules could attach themselves by hydrogen bonding and there GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS is little doubt that this is the reason for the large heat evolution observed. In attempting a quantitative statis- tical treatment of the absorption of vapours these two types of system require separate consideration. The positive heat and entropy of dilution found for nearly all non- . aqueous systems affords strong evi- 2 dence that absorption results simply F 5 from the tendency of the solute mole- 1 cules to diffuse into the polymer by 2 virtue of their thermal energy. The IP observed entropy of dilution is thus taken to represent the increase of con- figurational entropy consequent upon mixing the small molecules with the large ones.Another way of looking a t this is to say that the most prob- able way in which the available space can be occupied by the polymer and the liquid * is a random distribution I 1 I -2.0 I FIG. 13 of the two components. This method of approach lends itself to quanti- tative treatment since it permits the use of the Boltzniaiui relation between the entropy and probability ( W ) of a state S = k h W . (21) If W and W are the probabilities of the random mixture and a state of complete separation the increase of entropy AS on mixing is thus given by A S = k In (WJW,) . (22) The problem of evaluating W J W has been approached by considering the 38 G. Gee J .in the press ; op. cit. ref. 2 pp. 145-195. * The vapour molecules when dissolved in the polymer are of course to be con- sidered as in the liquid state since they are closely surrounded by neighbouring molecules but yet able t o diffuse. Heuts and entropies of dilution ( a ) benzene in rubber ; (b) agur-agar in water. 284 QUARTERLY REVIEWS solution to be arranged on a regular lattice so defined that each molecule of liquid occupies one or a small number of lattice points while the polymer occupies a large number of points forming a continuous succession of nearest neighbour sites. The method has been outlined recently 39 and only the simplest form of the final result will be quoted here. If the liquid molecule is sufficiently large or flexible t o require several lattice points the entropy of mixing calculated i s found to be nearly independent of the particular lattice chosen.If z is the ratio between the molecular volumes of the polymer and liquid and v the volume fraction of the polymer then AS 1 - R(In(1 - v,) + v,(l - l/z)} . * (23) Except in dilute solutions the term l/x is generally negligible and the equation then takes a very simple form completely free from any molecular parameters. I n other words according to this theory the entropy of dilution of any polymer by any liquid should be the same. The molecular origin of the heat of mixing lies in the change of inter- molecular contacts. Mixing involves the separation of polymer chains to make room for a molecule of liquid which has to break loose from its neighbours before it is free to take up the vacant site.Once there it is strongly held and in general the energy of the polymer-liquid contacts does not differ widely from that of the contacts they replace. It is this which makes the problem of calculating the heat of mixing so M c u l t . Indeed the existence of different kinds of intermolecular force would seem to preclude the possibility of any simple theory of general applicability. Fortunately for nearly all but hydroxylic compounds much the most important forces are the dispersion forces and by neglecting all others a very simple treatment can be given. This method was widely and 3uc- cessfully applied by J. H. Hildebrand *O to liquid mixtures and may also be used for polymer-liquid ~ y s t e i n s . ~ ~ This treatment gives the heat of dilution in terms o€ the cohesive energy densities (C.E.D.) of the com- ponents.The C.E.D. is the energy rieeded to separate all the molecules in 1 C.C. of a substance; for a liquid it is equal to the latent heat of evaporation a t constant volume (per c.c.). Denoting the C.E.D.s of liquid and polymer by coo e, the calculated heat of dilution is A H = Vo(deo- der)%; . - (24) where V is the molar volume of the liquid. The C.E.D. of the polymer is not so readily obtained. An approximate estimate can be made from the chemical structure of the polymer but this would scarcely be sufficiently precise to use in equation (24). A method of deducing it from the swelling of a non-linked polymer has been described ; 41 this is based on the assump- tion that equation (24) is valid or at least that AH, = 0 when e = e,.I n general equation (24) is not quantitatively accurate for liquid mixtures and in most discussions of polymer-liquid mixtures it has simply been assumed that A H o is proportional to $. By combining a term of this form 39 G. Gee Trans. Paraday SOC. in the press. 4O L L Solubility ” Reinhold 1936. 41 G. Gee I.R.I. Trans. 19.13 18 2tif.i. GEE THERMODYNAMIC PROPERTIES OF HIQH POLYMERS 286; with equation (23) M. L. Huggins 42 obtained for the free energy of dilution Except for dilute solutions V J X may be neglected and combination of (19) and (25) gives the vapour-pressure isotherm This equation has proved astonishingly successful in representing vapour- pressure data over a wide concentration range with a single value of p. Recently however it has been pointed out 39 that systems of limited miscibility show a systematic deviation represented by a decrease in p as the amount of vapour absorbed approaches the saturation value.The importance of this in the determination of two-phase equilibria will be discussed below. It should also be stressed that at least part of the suc- cess of the isotherm arises from a compensation of errors. The form of equation (24) is based on a model which requires each segment of polymer molecule in an infinitely dilute solution to be completely surrounded by solvent molecules. For polymers with flexible chains this will clearly not be true random linking of the chain will result in many intramolecular contacts. The effect of this will be to reduce A H / v ~ in dilute solutions; this has been demonstrated experimentally for rubber in benzene,36 and is believed to be a general phenomenon; ,LL remains relatively unchanged because there is a compensating reduction in AS,.The above treatment is based on an assumed randomness of the solu- tion but the existence of a finite heat of mixing tends to favour the states of lower energy. It has been shown 43 that this effect is negligible for the heats of mixing normally encountered in relatively non-polar systems. The sorption of water by a highly polar polymer represents the other extreme of behaviour. It is now generally recognised that the first amount of water to be taken up is exothermally attached to specsc polar groups,** the order thus imposed on the system being reflected in the negative entropy of dilution. There is still much discussion as to the mechanism of absorp- tion of the remainder,*5 and it would be out of place in such a review aa this to attempt to hold the balance between the different points of View.The various isotherms which have been deduced generally contain three adjustable parameters and not unnaturally they can be made to fit a simple curve such as that shown in Fig. 12 very well. Some of the possible complications of the systems studied will be referred to below. 4 2 Ann. N.Y. A d . Sci. 1942 43 1. W. J. C. Om Trans. Faradug SOC. 1944 40 320 ; E. A. Guggenheim Proc. Roy. SOC. 1944 A 183 203 213. 44 F. T. Pierce J. Textile In&. 1929 20 T 133. 4 6 A. B. D. Cassie Trans. Faraday SOC. 1945 41 450 458 ; “ Fibrous Proteine ” Society of Dyers and Colourists 1946 p. 90; P. H. Hermans ibid. p. 92; G. A. Gilbert &id.p. 96 ; A. J. Hailwood and S. Horrabin Faraday Sou. Discussion (see ref. 1). 286 QUARTERLY REVIEWS 8. The Osmotic and Swelling Pressures of Polymer Solutions and Gels This problem is very intimately related to the preceding one the osmotic or swelling pressure 17 being given by where To is the partial molar volume of the liquid in the solution. In general osmotic pressures are only measured in solutions sufficiently dilute to justify replacing In (1 - tir) in equation (25) by - v - Qv;. Combina- tion with (27) gives for the osmotic pressure A G O = -nV0 . * (27) This equation has been widely used in interpreting osmotic data for polymer solutions,42 but is open t.0 criticism on the grounds that dilute solutions are not accurately described by the present theory.Attempts 46 to formulate a more precise statistical theory of dilute solutions have not yet made very much progress. For the purpose of determining the molecular weight of a polymer from osmotic data the precise form of equation (28) is immaterial the important conclusion being that at infinite dilution the only remaining RT v term is the first -.< equivalent to the familiar van't Hoff form RTc/M v o where c is the concentration and M the molecular weight of the solute. It can be shown 4 7 by a very general argument that this must necessarily be the case for solute particles of any size or shape. Hence the molecular weight can always be determined by extrapolating osmotic data to infinite dilution this being most conveniently done from the approximately linear plot of I7/c against c.9. The Solubility of Polymers in Liquids The conditions for the co-existence of two condensed phases in equi- librium in a binary system of polymer + liquid may be put in the form where p0 p are the chemical potentials of liquid and polymer and the indices 0 r refer to the phases which are respectively dilute and concen- trated with respect to polymer. These conditions are of course identical with those for two phases in any binary system but the polymer-liquid system is peculiar in that in nearly all circumstances one phase is experi- mentally indistinguishable from pure liquid ; the second equilibrium con- dition is then meaningless and ,u is the chemical potential of the pure liquid. In these circumstances the two equations (29) may be replaced by AGO = O * (30) where dQ is the Gibbs free energy of dilution of the concentrated phase.dB P. J. Flory J. Chein. Physics 1945 13 453 ; W. J. C. Orr Trans. Faraduy Soc. 47 G. Gee ibid. 1944 40 261. 1917 43 12. GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 287 Thus a polymer will be completely miscible with a liquid if AGO is negative a t all concentrations. This is equivalent to considering the equilibrium as an osmotic equilibrium (with = 0) in which the polymer is unable to enter the dilute phase. If the polymer is cross-linked this method is precise ; for linear polymers it is adequate for most purposes. The observn- tion which justifies this simplification means that in general a liquid is either miscible with a polymer in all proportions or does not disperse it to any measurable extent ; 48 this behaviour is readily predictable from equation (25).According to the theory of non-polar systems outlined in tj 7 the factor tending to make AGO positive and thus leading to separation into two phases is a large positive AH,. Equation (24) suggests that this will be found in the case of a polymer and liquid differing appreciably in C.E.D. Although this is not an invariable rule it has proved very useful.38 In selecting a solvent for a given polymer the most likely choice would usually be one having a C.E.D. close to that of the polymer. This analysis is readily extended to the solubility of polymers in mixed Liquids.4g If there are two liquids there are three equilibrium conditions analogous to (29) but as in the case of it single liquid one of these may usually be omitted. The remaining two are more complicated this time since in general the liquid absorbed by the polymer differs in composition from the liquid in equilibrium with it and express the fact that there must be no change of Gibbs free energy on a small transfer of either liquid component from one phase to the other.There are six types of intermolecular contact to consider in a ternary mixture and the heat of mixing of the two liquids influences the solvent power of the mixtures very considerably. It may be shown 49 that if this heat of mixing is large the solvent power of the mixture will exceed that of the components separately. This behaviour seems very natural on a crude physical picture the large heat of mixing of the components tending to drive them away from one another which is only possible if they both mix with the polymer.The conditions for enhancement of solvent power on mixing are readily formulated in terms of the C.E.D.s of the liquid and polymer ; it is necessary that the C.E.D. of the polymer be intermediate between those of the liquids. This prediction has been verified by using it to select pairs of liquids which though individually non-solvents for a given polymer would dissolve it when mixed ; Table I1 gives some typical results. The solubility of crystalline polymers presents some new features of interest and practical importance. The two-phase equilibrium in a saturated Aolution of a completely crystalline material is of a different type from that previously considered since the solvent is absent from the solid phase. There is therefore only one equilibrium condition to be satisfied ; this may be put in the form dC&=O.' (31) where AG is the molar increase of Gibbs free energy accompanying the 48 J. N. Brransted 2. phyoikal. Chem. 1931 Bodenstein Festband p. 257. 49 G. Gee Trans. Paraday SOC. 1944 40 463 468. 288 Solvent mixtures. Non-solvents. QUARTEBLY REVIEWS Non-solvent mixtures. TABLE I1 Solubility of GR-S in mixed l i p i d s (a) n-Hexane . . . . 55 (c) isoButyl acetate . . 69 ( b ) Ethyl ether . . . 58 Solvents. ( d ) n-Pentme 50 ( d ) + (e) ( e ) Ethyl (4 4- (f) (el + (f) (f) Methyl acetate 80 acetate 89 I I I I I I dissolution of crystalline solute in the solution. Since the dissolved material is essentially liquid the process of solution may be considered to take place in two stages involving first melting of the crystals and then mixing of the liquid with the solvent.If the free-energy changes in these two steps are AGf and AG, the former must be positive since the crystals are stable with respect to their melt. This positive contribution to AG has to be balanced by an equal negative term AG ; if AGf is large the saturated solution will therefore be dilute. For a substance of low molecular weight AGf is simply related to the melting point Tr and molar latent heat of fusion LI Combination of this expression with the value of AG appropriate for ideal solutions gives an equation for the solubility in terms of Lf and Tr; this has been found very successful for relatively non-polar systems involving solutes of low molecular weight (cf. J. H. Hildebrand 40). When the solute is a polymer the problem becomes much more com- plex as will be evident from the discussion of crystallisation in $ 2.The polymer is never wholly crystalline and the crystalline and amorphous regions are so intimately related structurally that practically no inolecules will be free to disperse in the liquid until almost the whole of the crystal- lites have melted. At this point the polymer becomes amorphous and its solubility is determined by the factors already discussed. The essential new feature to be investigated therefore is the effect of a liquid on the melting point of the polymer. R. B. Richards * has included this problem in the statistical mechanical treatment of crystallisation. The effect of the liquid on the crystalline-amorphous equilibrium is calculated by intro- ducing the entropy of mixing of the liquid with the amorphous regions using an equation equivalent to (23).The final expression for the melting point is somewhat complex and contains three more or less adjustable constants but it is very successful in reproducing the general effect of additions of low-molecular substances to polythene the melting range is lower and melting becomes still more gradual. It is well known that fractional solution or precipitation may be employed to separate heterogeneous polymers into fractions more or less homogeneous AGj = Lj(1 - T/TJ) . * (32) GEE THERMODYNAMIU PROPERTIES OF HIGH POLYMERS 289 in molecular weight. The whole subject of polymer fractionation has been recently reviewed,60 and only a few points need to be noted here. In general fractionation is carried out by using critical mixtures of solvent and non- solvent the molecular weight which just dissolves having been found experimentally l / M = A + Bp .* (33) An equation of this form can be derived,62 with certain approximations fiom the statistical theory outlined in 3 7. The thermodynamic formula- tion of the two-phase equilibrium existing when a part of the polymer has been precipitated from solution is complex. Even if the distribution of molecular sizes is replaced by two species there are four equilibrium con- ditions to be satisfied one for each component of the system. In this case since we are concerned with solutions of finite concentration none of these is negligible. In order to treat the problem statistically the theory described in 8 7 must be extended to the mixing of two polymers.This has been carried out by E. A. G~ggenheirn,~~ whose solution gives for the entropy of mixing of n, nj . . . moles of polymer in the volume ratio vi vj . . . AS = - kZn,Invi ’ (34) The heat of mixing of homologous polymers may be assumed to be zero and their heats of solution in a given liquid proportional to their molecular weights. It is thus R straightforward matter to express the thermo- dynamic equilibrium conditions in terms of molecular quantities but the resulting set of simultaneous equations has not yet been solved in any complete way. In the Reviewer’s opinion it is doubtful whether the present status of the statistical theory justifies the labour involved in numerical solutions. A rather simpler problem is presented by the mutual solubility of polymers.A. Dobry b3 has shown that in general if solutions of two different polymers in the same solvent are mixed two phases are formed. The higher the concentration of the solutions the more completely do the two polymers separate in the two phases. Of 35 pairs of polymers examined only four formed single-phase systems over the whole concentration range studied. Dobry has sought to explain this behaviour in terms of complex formation but in the Reviewer’s opinion a phenomenon of such generality must almost certainly have a simple general explanation. I n fact such an explanation is implicit in the order of magnitude of the entropy of mixing given by equation (34). Let us consider various binary mixtures of two liquids each of molar volume 100 C.C. and two polymers each of molar volume 100,000 C.C.The t o be simply related to the fraction p of precipitant i 60 I,. H. Cragg and H. Hammerschlag Ghem. Reviews 1946 39 79. 61 G. V. Schulz 2. physikal. Chem. 1937 A 179 321 ; R. A. Blease and 6 2 Q. Gee Ann. Reports 1942 39 7. 63 J. Chim. physique in the press ; read at Strasbourg Polymer Conference Nov. R. F. Tuckett Trans. Faruduy SOC. 1941 37 571. 1946. 290 QUARTERLY REVIEWS increases of entropy (XT) on mixing 1 C.C. of each of two components at room temperature are approximately ( a ) Two liquids . . 8.3 cals. ( c ) Two polymers . . 0.0083 cal. This entropy of mixing of two polymers is so small that even a very small positive heat of mixing would result in almost complete immiscibility. Since for most pairs of polymers A H is likely to be positive and of the same order of magnitude as for liquid mixtures,* it appears certain that most pairs of polymers will be immiscible.The high viscosity of polymers makes this very difficult t o test directly with dry polymers but Dobry's work affords convincing evidence of its truth. The presence of a common solvent for the two polymers will of course tend towards the formation of a single phase. The three equilibrium conditions defining the two-phase systems described by Dobry are easily formulated in terms of the statistical theory of polymer solutions but do not appear to be explicitly soluble and have not yet been examined numerically. It can be said with confidence that they are in qualitative agreement with the experimental observations. 10. The Swelling of Polymers in Liquids ( b ) Liquid + polymer .. 4.15 cals. This problem is of course only the limiting case of the absorption of vapours by polymers when the vapour is saturated. Owing to its practical importance and experimental simplicity it has been the subject of a good deal of work and therefore merits special reference here. I n this section attention is confined to polymers which are totally insoluble in the liquid concerned so that the equilibria to be discussed are osmotic equilibria defined by the single thermodynamic condition AGO = 0 . * (30) For a linear polymer Combination of this condition with the statistical expression (25) for A Go gives (v,/x being neglected) It is convenient for the purpose of this section t o re-write this in terms of the volume of liquid Q imbibed by unit volume of polymer at saturation - (36) This has a positive solution if ,u > 0.5 so that a polymer-liquid combination having ,u > 0-5 should not be completely miscible.Doubt has recently been cast on this conclusion by some data on polystyrene in diethyl ketone and n-biityl acetate. E. C. Baughan 54 has reported that the vapour pressures of these systems give ,LL values of 0.67 and 0-82 respectively yet bot'h liquids are solvents for polystyrene. For these systems then p is not independent of concentration and it will be shown below that there is evidence of similar discrepancies in the swelling of natural vulcanisates in poor swelling agents. Despite this apparent limitation the mean value of ,,Y remains a most In ( 1 - v,) + 01. + pv; = 0 . * (36) In (1 + I/&,) = (Qm + I)-' + ~ ( & m + 64 Faraday SOC.Discussion (ref. 1). * Cf. the argument leading up to equation (21). GEE TRERMODYNAMIC PROPERTIES OF HIGH POLYMERS 291 useful parameter by which to characterise a polymer-liquid combination and its evaluation by means of equation (36) from the observed maximum swelling offers the simplest method by which it may be estimated. We shall show below how the method may be extended to values of ,u < 0.5 when cross-linked polymers are available. It is convenient to anticipate at this stage in order to describe swelling data which throw light on the factors determining p. According to equation (24) p would be equal to Yo( l / e - de,)2/RT and although this seldom holds quantitatively there r n I I -'0 FIG. 14 Swelling of natural rubber vulcanisatea in a range of aliphatic liquids.(Hydrocarbons 0 ; ketones A ; estera V ; aldehydes x ; ethers 0 ; n i t r i b +.) is strong evidence that this quantity exercises the controlling effect in the relative swelling of a given polymer in a range of liquids provided the syst'ems be chosen SO as to exclude obvious specific interactions and especially hydrogen bonding. This evidence has been reviewed recently,% but we may repeat here one of its most strikingly successful application^.^^ Fig. 14 shows the equilibrium swelling of five different natural rubber vulcanisates in a range of aliphatic liquids as a function of d Y o ( l / e - l / e ) . The ex- perimental points refer to the top curve and show how different chemical types all lie nearly on a single curve ; similar agreement was found for the 292 QUARTERLY REVIEWS other four vulcanisates.necessary to write empirically where ,uo is a constant but Ic is still slightly dependent on 1/ V,( deo - de,). A range of aromatic liquids gave similar results,but with different k (more nearly 1). The estimation of ,u values from cohesive energy densities is therefore only qualitatively successful but within this limitation is of great use in correlating the swelling powers of a range of liquids. The nest problem to be considered is the effect of mechanical restraints on the swelling of a given polymer in a given liquid. Restraints may be either external or internal i.e. due to the structural rigidity of the material and in general oiily the former can be discussed by thermodynamic methods. A linear polymer which is amorphous and not brittle behaves essentially as a liquid to cont,inuous stresses and cannot therefore support permanently any type of stress other than hydrostatic.A hydrostatic pressure may be applied either to the polymer only some form of porous membrane being used to effect equilibrium with the liquid or to both components. The former case is simply the familiar osmotic or swelling pressure and the effect of a pressure P on AGO is given by To fit one of these curves quantitatively it is p = po + kVo( l / e o - l / e r ) 2 (37) where V the partial molar volume of the liquid in the compressed polymer is usually very near to the molar volume V of the free liquid a t least for fairly small pressures. Thus application of an excess pressure P raises the free energy of dilution by V,P. To calculate the resulting decrease of swelling use may be made of the mathematical identity whence Here Q is the degree of swelling of the polymer (= l/v - l) not neces- sarily at saturation and the left-hand side represents the change of swelling with pressure if the vapour pressure is kept constant.The derivation of this equation has involved no assumptions about the nature of the polymer ; it is indeed equally applicable to liquids. The chemical nature and structure of the polymer will of course control the magnitude of aAC?,/w and hence also of (aQ/aP),@. If the pressure is applied to both phases (38) has to be replaced by - (adQo/aP)T,Q = 7 0 - v * * (41) I n view of the approximate quality of 7 and V, the effect of a hydroatatic pressure applied both to liquid and to polymer may be in either s e w but will usually be unimportant.In aqueous systems however 7 is generally GEE THERMODYNAMIC PROPERTIES OF HIQH POLYMERS 293 less than V, and for these a hydrostatic pressure will increase the maximum swelling. A precisely analogous method may be employed to calculate the effect of any other applied stress on the swelling of any body capable of sustaining the stress. An example in which we shall be interested is that of a polymer constrained (by unidirectional tension or compression) 60 a length I in a specified direction I f f is the force required and I the unstrained length and cross-section in the dry state it is found 39 that at constant T and P In this case we cannot at once evaluate (i3f/i3Q)2 but it is evident physically that swelling a stretched polymer will generally have the effect of diminishing the tension.If this is so extending a piece of swollen polymer will according to (42) diminish its vapour pressure and increase its swelling capacity. In order to carry out experiments on the effect of applied stresses it is necessary to work with a polymer possessing structural rigidity even when swollen. This internal structure also imposes restraints on the swelling of the polymer and may thus affect both the entropy and the heat of swelling. If the structure is such as to leave the polymer macroscopically homogeneous there is no thermodynamic method of evaluating its effect. In certain cases however the structure is sufficiently coarse to enable its constituents to be considered separately. An obvious example of this type is found in a reinforced plastic where the reinforcing layers (which may themselves be swollen) restrain the swelling of the plastic inter layer^.^^ An extremely interesting and thorough analysis of the swelling of wood has been made by W.W. Barkas along similar lines.56 A typical wood cell has the form of a hollow cylinder with a thin elastic sheath. Swelling occurs almost ex- clusively perpendicular to the axis and is restrained by the elastic reaction of the sheath which is comparatively little swollen. The very detailed discussion of this problem given by Barkas could probably be applied to other materials . Attempts which have been made to calculate the effect of internal struc- ture on the swelling of wool and keratin 57 rest on a much more questionable basis. It is assumed that the work done by the water in swelling the structure is the same as would have to be expended in order to produce the same dimensional changes by applying external forces.This is certainly not true in general and would require strong justification before it could be accepted in any particular case. In the Reviewer’s opinion the restoring forces of the structure are very greatly over-estimated by this procedure. The effect of cross-linking on the swelling of rubber-like polymers ia another example of a restraint imposed by an internal structure. This has been analysed on the basis of the statistical mechanics of an ideal network (cf. 9 5) the problem being to calculate the elastic reaction of the network ssF. T. Barwell and K. W. Pepper ibid. s8 Forest Products Research Special Report No.6 H.M.S.O. 1945. 67 A. B. D. Cassie loc. cit. ref. 45. 294 QUARTERLY REVIEWS when swollen by a liquid. It is clear from the model that this restraint takes the form of a negative contribution to the entropy of swelling. This contri- bution has been estimated by James and Guth and by P. J. Flory 25 for the cases of isotropic swelling and of swelling with the polymer restrained in one direction. After introduction of these terms into equation (25) the free energy of dilution becomes (a) Isotropic swelling AGO = RT{ln (1 + $) - ( 1 + Q)-l + p(1 + &)-a + + &)-'/a} . (43) 3 roy.hrc. FIG. 15 Effect of vulcanisation on swelling of natural rubber compounds 1 light petroleum (b.p. 40-60") ; 2 n-propyl acetate ; 3 ethyl acetate ; 4 methyl ethyl ketone ; 5 acetone.(Curves theoretical.) (Reproduced by permission from Transactions of the Faraday Society 1946,42 B 36.) ( b ) Swelling a t length I * (44) where p7. and Mc are the density and molecular weights between junction points of the polymer V is the molar volume of the liquid I is the length dry and unstrained I is the length swollen and strained. These equations permit p to be calculated from the swelling of polymers of known degrees of cross-linking and also give values for aAG,,/aQ which may be substituted into equations (40) and (42). An interesting feature of (44) is that as I + co the negative entropy contribution vanishes leaving AGO = RT(ln(1 + &) - (1 + Q1-l + p ( l + + Pr z.~} v o 1 . 296 AGO equal to the value for an uncross-linked polymer. It can thus be seen that cross-linking should reduce the swelling capacity while extension should increase it towards its original value.These predictions have been tested by using a series of natural rubber corn- GEE THERMODYNAMIC PROPERTIES OF HIGH POLYMEM - some of the results being shown in Figs. 15-17. Values of M were estimated from the elastic moduli of the rubbers when swollen and are therefore not really inde- pendent quantities. Two points are to be observed ( a ) the very good quantitative account of the effects of network structure on swelling in good swelling agents which can be given by this theory and ( b ) the systematic departure found for poor swelling agents. Analysis of the latter by means of equation (42) shows that it is accounted for almost wholly by the term i3AGo/aQ. The con- clusion drawn is that polymer- liquid systems with a relatively small mixing tendency cannot be adequately characterised by a single value of p but that as saturation is approached p falls considerably.This phenomenon which appears to be quite general is illustrated by the vapour- pressure data for natural rubber hydrocarbon in ethyl shown in Fig 18. It represents a defect in the current statistical theory of polymer-liquid systems which has not yet been satisfac- torily explained. I 2.5 z/z* . FIG. 16 Effect of extension o n swelling (1) chloro- form ( 2 ) benzene ( 3 ) toluene (4) light petro- leum (b.p. 40-60'). (Curves theoretical.) (Reprodud bg pcnniS&n from Transactions of the Faraday Society 1940,4!& B 37.) 11. The 8olubility of Gmes in Polymer8 This problem differs from that of the absorption of liquids only in that the reference state of the solute is the gas (usually at 1 atm.pressure) instead of the liquid. Just as the dissolution of crystalline materials may be divided into (a) fusion and ( b ) mixing so the solution of gases may be supposed to proceed through a preliminary liquefaction. This i p not in all cases a physically possible route since we may be concerned with gases which are above their critical temperatures. This objection is more apparent than real since even in the caae of unsaturated vapours the process of liquefaction 296 QUARTERLY REVIEWS lq71(;. I 7 Eflect of extension O?L suxllitLy ( 1 ) l l - p r o ~ ~ ! j l cicrtutr (cidd 0.5 to ordijiutes) ( 2 ) ethyl acetate ( 3 ) nlethyl rtlryl kotutic (4) c t h y l Jor)ticite.(13roke)i curve8 theoreticul.) (Reproduced by per’ntixsion j n j m Trawnctions of the E’:rraday YocIety 1946 42 B 38.) FIG. 18 Free eneryy of di1ritio)i of nutirrul riibbrr by acetone. [Broken curue cnlculrited f r o m e y w t i o n ( 2 5 ) with p = 0.713.1 BEE THERMODYNAMIC PROPERTIES OF HIGH POLYMERS 297 is an unnatural one involving an increase of Gibbs free energy. The only additional difficulty in the case of permanent gases is concerned with the extrapolation required in estimating the free energy of condensation. Writing the Gibbs free energy of solution dG of gas a t 1 atm. pressure in the solution as the sum of AG and AGO the Gibbs free energies of condensa- tion and dilution we have for equilibrium AG,+AGo = O . * (45) The problem is now seen to be very much simpler than that of the solution of crystalline polymers since A@ is connected only with the properties of the gas.If Tb is the condensation temperature and Lb the molar latent heat of condensation assumed independent of T AGO is given by equation (25) but in the case of gas solutions v is invariably close to 1 so that to a good approximation AGO N R T ( l n ( 1 - v,) + 1 + p } . 0 (47) The solubility is usually expressed as the volume of gas (c.c. a t N.T.P.) dissolving in 1 C.C. of polymer under a pressure of 1 atm. Denoting this by 0 its relation to vr is u = 22,400(1 - v~)/VO . * (48) where V is the molar volume of t,he liquefied gas a t the temperature of the experiment ; this can only be estimated by extrapolation. Combination of equations (45) to (48) ghes finally dG = &(T/Tb - 1 ) .' (46) Lb/Tb is the entropy of evaporation which may be taken as 20 cals./mole/" c . while a typical value of In (22,400/v0) is about 6-5. Hence (49) reduces to - ln u 2 4.5 + p - lOTb/T . * (50) This argument has been set out in some detail as it has not been published elsewhere and the final result is particularly simple and striking. It is t o be emphasised that the low solubility of permanent gases is due on this view to their very large positive free energy of condensation a t ordinary temperature. Since this is determined largely by the difference between their condensation temperatures and the experimental temperatures it is natural to find a term Tb/T in equation (50). The value of p is relatively unimportant for permanent gases; this is in accord with the observation that the solubilities of say air in a range of polymers do not differ widely.Without knowing p it is not possible to compare equation (50) with experi- ment exactly but it is found that even if we set ,u = 0 the calculated solu- bilities of hydrogen nitrogen oxygen and methane agree within a factor of 3 with the observed solubilities in natural rubber.58 A further test of the validity of this method of approach is to use it to calculate the heat of solution which is obviously given by AH = /ART - 2OTb . * ( 5 1 ) ~~ 68 G. J. van Amerongen J . Appl. Physics 1946 17 972. 298 QUARTERLY REVIEWS Eliminating ,u between (50) and (51) we have AH E! - RT(4.5 + In a) . ( 5 2 ) Table I11 gives the values of p needed to fit some experimental solubilities to equation (50) and a comparison of the observed heats of solution with those calculated from equation (52) ; the agreement is only approximate but is as good as would be expected from such a crude analysis. TABLE I11 Solubilities of gases in natural rubber at 25" c . Gas. Tb I(. H . . . . . N . . . . . 0,. . . . . coz . . . . so . . . I CH . . . . XH3 . . . . - In a AH^ (cals./mole). Exptl. 900 650 150 - 1900 - 2600 - 5400 - 7000 Eqn. (52). - 9.10 - 1300 - 1850 - 2640 - 3500 - 4500

ISSN:0009-2681    

DOI:10.1039/QR9470100265

出版商:RSC    

年代:1947

数据来源: RSC