Cornell XDlntvetsit^ OF THE IRcw l^orft State (tollcae of Hariculturc B.|.4i^M...,. s.^L!^.lb.^.. Cornell University Library QD 549.T3 The chemistry of colloids and some techn 3 1924 003 139 270 The original of tliis book is in tine Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924003139270 THE CHEMISTRY OF COLLOIDS THE CHEMISTRY OF COLLOIDS AND SOME TECHNICAL APPLICATIONS W. W. TAYLOR, M.A., D.Sc. LECTURER IN CHEMICAL PHYSIOLOGY AT THE UNIVERSITY OF EDINSURGU SECOND IMPRESSION NEW YORK LONGMANS, GREEN AND CO. LONDON: EDWARD ARNOLD l-RIKTRD BY WILLIAM CLOWES AND SONS, LIMITED, LONDON AND BECCLES, ENGLAND. PEE FACE It is curious that althougli colloid chemistry owes its development in no small degree to British investigators, hitherto there has been not only no English text-book on the subject, but no text-book in English available, the foreign works that have been translated dealing with particular aspects of the subject only, or with its bearings on other sciences. The present book, based on my lectures on Hetero- geneous Systems delivered to advanced students in the University of Edinburgh, will, it is hoped, meet the want of a convenient text-book, and at the same time serve as a reference book for workers in other sciences, who are becoming more and more interested in colloids. To increase its usefulness in the latter direction, a section on the^applications of colloid chemistry has been added. It makes no claim to be complete or exhaustive; indeed, many important applications are not even mentioned, bub it seemed better to deal at some length with a number of selected examples, rather than to give a mere catalogue of the many branches of pure and applied science in which it plays a part. Further, a large amount of experi- mental data has been included throughout the book, with the idea that this may be of use to those interested in the subject. The practical instructions for the preparations will, if carefully adhered to, enable any one to prepare sols vi PREFACE whether for purposes of research or of demonstration. In this connexion I desire to express my indebted- ness to Svedberg's Die Methoden zur Herstellung Kolloider Losungen, as also generally to Freundlich's Kapillarchemie and von Weimarn's Grundzuge der Disjtersoidchemie. My thanks are also due to Messrs. Carl Zeiss, who kindly lent the block for Figure 2 ; to my wife for draw- ing the remaining figures; and to Mr. J. D. M. Eoss, M.A., B.Sc, for valuable assistance in reading the proofs. W. W. T. University of Edinburgh. /an. igis. CONTENTS PAET I GENBEAL PBOPEETIES OP COLLOIDS CHAFTES PAGE I. Introduction 1 II. General Differences between Suspensoids and Emulsoids 18 III. Diffusion and Dialysis 24 IV. Osmotic Pressure and Molar Weight . . 32 V. Optical Properties ...... 38 VI. Brownian Movement 47 VII. The Size of Particles and Ultrafiltration . 54 VIII. Electrical Properties 61 IX. PRECrPITATION 98 X. Protection and Mutual Precipitation . . 128 XI. The Properties of Gels ..... 141 PAET II METHODS OF PEEPABATIQN XII. Introduction , .165 XIII. Crystallisation Methods 176 XIV. Solution Methods ...,,,. 204 XV. Electrical Dispersion Methods » . , . 218 viii CONTENTS PART III ADSOEPTION CHAPTEK PAG« XVI. Sdeface Phenomena 221 XVII. SUKFACB CONCBNTKATION 232 XVIII. Adsorption 243 PAET IV APPLICATIONS OP COLLOID CHEMISTEY XIX. Semi-Colloids 263 XX. Dyeing 269 XXI. Tanning, The Soil, and Purification op Sewage 285 XXII. Applications op Colloid Chemistry to Biology 295 Subject Index , 319 Author Index 326 CHEMISTRY OF COLLOIDS PART I GENERAL PROPERTIES OF COLLOIDS CHAPTER I INTEODUCTION When Graham, as the. result of his epoch-making experi- ments on colloids, was led to make the well-known state- ment that "they (crystalloids and colloids) appear like different worlds of matter, and give occasion to a corre- sponding division of chemical science. The distinction between these kinds of matter is that subsisting bet-st^een material of a mineral and the material of an organised mass," the science of chemistry was still dominated by the idea of the fundamental difference between inorganic and organic substances. It was natural, therefore, that he should extend this already accepted distinction to the new region of chemistry which he had just discovered. But, just as the advance of knowledge has removed the distinction in the one case, so too the tendency of recent work has been to bring the other under suspicion. For many years physical chemists have inclined more and more to the view that crystalloids and colloids are not different hirids of matter (as Graham considered) but, rather, are different states of matter, and that the same B 2 GENERAL PROPERTIES OF COLLOIDS chemical substance may be obtained in the one state or the other by suitable alteration of the conditions under which it is produced. WhUe this exceedingly important change in our funda- mental conception regarding colloids is not due to any one investigation, Paterno's researches, although recently shown by him to be erroneous (p. 285), were the first to indicate that a substance (tannin) may form a sol in one liquid (water), and a true solution in another liquid (acetic acid). Many have contributed to the development of the idea, but it was only recently carried to its logical conclusion by P. P. von Weimarn, who has succeeded in mathematically formulating the main factors which govern the appearance of a substance in either the colloid or the crystalline state ; these con- clusions he has confirmed in an elaborate series of investi- gations. As a result of these investigations we have not only the means afforded of classifying the diverse methods hitherto employed in the preparation of particular colloids, but have indicated in any given case the necessary modi- fications of the conditions for obtaining a substance in the colloid state. These developments are so recent that they can hardly be said to have been applied in a systematic manner, and yet von Weimarn has already succeeded in obtaining more than 400 substances in the colloid state. In another direction there have been in recent time very considerable advances : in the development of the connexion between the special properties of the colloid state and capillarity or surface phenomena. Here the question of surface energy is all-important. As a surface implies heterogeneity, being an interface between two phases, we can at once proceed to classify systems according to the nature of the phases (at least INTRODUCTION 3 two in number) which are necessary for the existence of the interface. The special properties ascribable to surface energy only come into prominence when it amounts to a moderately large fraction of the total energy of the system ; this is connected with the ratio of the surface to the volume (or mass) of the system, which in turn may be expressed in terms of the " specific surface." The specific surface is the surface divided by the volume. The surface of a cube, the side of which is I, is 6Z*, and its volume is l^ ; the specific surface is &jl. Thus a cube of 6 cm. side and the inscribed sphere of 6 cm. diameter have unit specific surface. A large specific surface can only be obtained by the reduction of at least one spatial dimension of one of the phases, hence the name micro- chemistry, suggested by Ostwald, in place of capillary chemistry, the name adopted by Freundlich. Unfortu- nately, the name has already been misapplied to a special branch of chemical analysis — the analysis of microscopic quantities. What is usually known as colloid chemistry is only a part, though the most important practically, and the most fully developed part, of the general science of microchemistry. By a systematic application of the two ideas — of heterogeneity involving at least two phases, and of the increase of the specific surface by reduction of one, two, or three spatial dimensions of one of the phases, we arrive at a complete classification of the possible microchemical systems. The phase which is divided into small separate volumes is known as the disperse phase, the other is the dispersion medi^im ; when the dispersion takes the form of parallel plates, there is no distinction between the two phases. The dispersity is the degree to which the reduc- tion of the dimensions of the disperse phase has been carried ; it is best expressed numerically in terms of 4 GENERAL PROPERTIES OF COLLOIDS specific surface, but more frequently in terms of the linear dimensions of the disperse phase, e.g. the thickness of films, or the diameter of separate particles. Diminution of one dimension gives rise to films and bubbles; of two dimensions to threads; and of three dimensions to fine powders, drops, etc. Leaving out of account the two former cases, we have the following classes : — I. The dispersion medium is a Gas.' The disperse phase is (a) liquid, producing cloud t)r mist; (6) solid, producing dust or smoke. II. The dispersion medium is a Liquid. The disperse phase is (a) gas, giving rise to foam ; (6) liquid, producing emulsions (and gels ?) ; (c) solid, producing suspensions. III. The dispersion medium is a Solid. The disperse phase is (a) gas ; gas bubbles in a solid produce solid foam or scoriae ; Q}) liquid, as in the liquid inclusions in minerals; some gels probably come into this class. (c) solid, producing solid mixtures. Emulsions and suspensions form by far the most im- portant classes at the present time, comprising as they do what were formerly known as colloids. There is some doubt whether all gels ought not to be regarded as emulsions possessed of considerable rigidity. This ques- tion will arise later, and will then be more fully discussed. Solid mixtures have as yet received little theoretical con- sideration, owing to the difficulties involved. They are, ' Since all gases are completely miscible, there is no disperse system consisting of gases alone. INTRODUCTION S however, of great practical importance, as the mere mention of cements and heterogeneous alloys will indicate. Here, too, the question arises whether, in some of these cases, one at least of the phases may not be liquid rather than solid, and should accordingly be placed among the gels. Nomenclature. — When the history of the development of colloid chemistry is recalled, it is hardly surprising that there is a certain amount of confusion in respect of nomen- clature. To Graham we are indebted for the general term colloid in contrast with crystalloid. He also distinguished between the two conditions in which the colloids he first investigated, as gelatine and silicic acid, were obtainable, the term sol being applied to the form in which the system was apparently liquid, while to the solid, jelly-like form he applied the term gel. If one of the components was water, the two forms were a hydrosol and a Tiydrogel; if it was alcohol, they were an alcosol and alcogel, and so on. These terms are still in common use, but are no longer sufficient. The earlier work was largely done on either naturally occurring colloids, or on others, the discovery of which was more or less in the nature of accident, and which were similar in properties to these natural colloids. These be- long to the group II. (6) in the above classification (p. 4), consisting of systems of two liquids, and known as emul- sions. Now, the colloids of the group II. (c), consisting of a solid disperse phase in a liquid dispersion medium, were investigated at a later date, and as they are on the whole of a simpler nature than the emulsion colloids, many of the most important properties of colloids were not found out until comparatively recent 'times. The terms sol and gel are frequently applied to them ; the former is un- objectionable, not so the latter. It is not desirable to 6 GENERAL PROPERTIES OF COLLOIDS apply the same terms to processes or states which may be essentially different, though superficially the same. As already suggested in the above classification, gels usually consist of two liquid phases, but may in some cases consist of a solid phase and a liquid phase. In the case of suspension colloids there is little doubt that the sol contains a solid phase and a liquid phase, and this is probably also true of the precipitated colloid, to which the term gel has been applied. It is not impossible that here the two phases may be liquid, but even if it be so, the phenomena are not comparable with those exhibited by the gels proper, and it is better not to extend the name to them. Starting from the fundamental idea that colloids are heterogeneous systems, one phase of which is liquid, and the other either liquid or solid, we have first of all, when the dispersity is not high, the two classes known as emul- sions and suspensions respectively. As the dispersity in- creases we have the typical emulsion colloids and suspension colloids, which may conveniently be named emulsoids and suspensoids, the corresponding general term being dispersoids. Some, as P. P. von Weimarn, would still further subdivide the region between this dispersity and the apparent homogeneity of solutions, into emulsides suspensides, and solutides, the latter merging into true solutions. But these terms have not as yet found common recognition, partly because the phenomena can be ade- quately described and discussed without them. The classification of sols into suspensoids and emul- soids, based upon the state of the disperse phase, is not altogether satisfactory, for some systems, in which the disperse phase is undoubtedly liquid, exhibit characteristic properties of suspensoids, while others, in which it may be INTRODUCTION 7 solid, are more like the emulsoids. The determining factor seems to be the presence or absence of affinity between the two phases, whether in the form of solubility or chemical combination, the formation of a gel being a sig- nificant manifestation of the difference between the two classes. The term lyophile has been applied to those systems in which there is marked affinity between the phases, and lyophdbe to the others. When water is the dispersion medium the terms hydrophile and hydrophobe are commonly used. Broadly speaking, this classification is the same as into emulsoids and suspensoids. Lyotrope Series. — Another aspect of the affinity between disperse phase and dispersion medium may be referred to here. There are many striking differences between suspensoids and emulsoids, or better, between lyophobe and lyophile sols, and especially in their behaviour towards chemical reagents. With the former, the effect of the added reagent is produced almost exclu- sively on the disperse phase, while with the latter, its effect on the dispersion medium is also of importance, and may even overshadow the other, and so become the pre- dominant feature. It is well to remember at the outset that the ordinary physical properties of a lyophobe sol are very little different from those of the pure dispersion medium, while the converse holds for lyophile sols. Now there is a large number of reactions in solutions which are markedly influenced by the addition of foreign substances, which from their nature seem debarred from taking any chemical part in the reaction. The inference is that their influence on the chemical reaction is an indirect effect, the direct effect being a change in the nature of the solvent, pro- duced by the added substance. This is confirmed by the following facts. 8 GENERAL PROPERTIES OF COLLOIDS If the reagents are arranged in order of their influence on the reactions, the same sequence is obtained in widely differing reactions. The reaction may be promoted by certain reagents, and hindered by others, in comparison with the reaction in the pure solution. This sequence does not appear to follow any recognisable chemical order, as, e.g. valency, atomic weight, etc. In some reactions the order is reversed, but the sequence is still maintained. The order is frequently reversed when the same reaction takes place in acid and alkaline solutions, i.e. the sub- stances which favour the reaction in acid solutions hinder it in alkaline solutions, and conversely, the sequence remaining the same in both cases. Among the reactions in which this lyotrope influence has been recognised are the following: the catalysis of esters, the inversion of cane sugar, the setting of gelatine, and the heat-coagulation of albumin. It is very significant that the same sequence is observed in the effect of sub- stances on many of the physical properties of water as compressibility, viscosity, density, surface tension. This it is which points to the lyotrope character of the effect in the chemical reactions, the essential properties of the water, on which its behaviour as a solvent and as a dis- persion medium depend, being changed in a definite manner to a fixed extent by these substances. Appended are a few examples of reactions and properties, which will illustrate the lyotrope series. The ions are arranged in order of their effect upon the velocity or the property ; thus ions which precede {H^O) reduce the velocity of reaction or the numerical value of the compressibility, etc., of water, the first mentioned causing the greatest reduction, while those which succeed {H^O) increase the velocity, etc., the last mentioned having the greatest effect. INTRODUCTION 9 1. The Hydrolysis of esters by acids. Anions: SO4 (^^^O) CI < Br. Kations : {H,/)) Li < Na < K < Eb < Cs. SO4 retards the action, the other anions and the kations accelerate it. 2. The Hydrolysis of esters by bases. Anions : I > NO3 >Br > 01 > A (^R^O) Sg O3 < SO4 Kations : Cs > Eb > K > Li (H^O). The ions which accelerate the acid hydrolysis retard the basic hydrolysis, and vice versa. 3. The Compressibility of aqueous solutions. Anions : CO3 > SO4 > CI > Br > NO3 > I {H^O). Kations : Na > K > Li > NH^ {H^O). In all cases the solutions are less compressible than water 4. The Surface Tension of aqueous solutions. {E^O) I< NO3 < Cl< SO4 < CO3. The surface tensions of the solutions are greater than that of water. 5. The Viscosity of aqueous solutions. Anions : NOg > CI {H^O) SO4 [Potassium salts]. Kations : Cs > Eb > K (ffgO) Na < Li [chlorides]. The viscosity of the solutions of KNO3, KCl, CsCl, EbCl is less than that of water (so called " negative viscosity "), that of the other salts is greater than that of water. Coagulation, again, is a term which is applied more or less indiscriminately to several phenomena, e.g. to the formation of a precipitate in emulsoid sols, as the proteids, especially by the action of heat, and also to the precipita- tion of suspensoid sols, especially by the addition of electrolytes. As, however, little is known regarding the mechanism of the former type of reaction, this dual use of the term is not of much consequence. But undoubtedly lo GENERAL PROPERTIES OF COLLOIDS the latter set of changes are better designated as pre- cipitations. Sol-gel Transformation. — The ambiguity in the use of the term gel has already been referred to. The reversible transformation of sol ^ gel is a phenomenon, the extreme importance of ■which is likely to increase, as so many other effects are bound up with it. The nomenclature is admittedly inadequate; verbs to express the processes and names of the processes themselves are much needed. It is now difficult to justify such a statement as " the gel dissolves to a sol," for if the gel dissolved, the result would be a solution, not a sol. The term gelatinisation is frequently used for the change from sol to gel. It is felt to be too long, for some have reduced it to gelatination, but as gel is the generalised form of gelatine, gelation is obviously the most appropriate term, and the corresponding verb is surely gelate. The term gelation has been occasionally used. For the reverse change we are not in so happy a, position ; solution and solate are unquestionably the logical antitheses of gelation and gelate, but we already have solute for a thing, and solution for both a thing and a process, and the more recent solvate as an adjective and a noun. Graham introduced the verbs pehtise and peptise, and the corresponding terms pektisation and peptisation to express these changes, and probably considered the similarity of the two words to express two directly opposed processes to be an advantage, instead of as now the very reverse. At any rate they have not been generally adopted, though curiously the adjective pelctous is coming into use, especially among biologists, and peptise is also frequently used. Now the real objection to peptise and peptisation is that they originally were, and still are, strictly applied to INTRODUCTION ii only one of the many methods by which a gel may be transformed into a sol ; and it is most undesirable to stamp a general idea with a name which has been and still is properly enough applied to one small part of the idea. There seems no escape from the adoption of the following terms : — gel, to gelate, gelation, sol, to solate, solation, they are unambiguous and brief, are logically connected, and will save an enormous amount of circumlocution. Degree of Dispersity. — The fundamental idea under- lying the chemistry of colloids, and much more, is that of heterogeneity, but between certain well-defined limits. The upper limit has been fairly definitely established, and lies above the limit of microscopic visibility (which may be placed at 10"^ cm.). At this dispersity the properties of the substance begin to differ appreciably from the properties of the substance as it ordinarily occurs; at this stage the phenomenon known as Brownian movement, just observable at about 10 "^ cm., is well developed, and the surface energy is no longer a negligible fraction of the total energy of the system (p. 222). Particles between these limits (10~^ cm. and 10~^ cm.) are termed microns in Zsigmondy's nomenclature. The lower limit can not be below the dimensions of molecules (about 10~^ cm.). Ultramicroscopic hetero- geneity has been directly observed at 5 X 10~' cm., the particles between this and 10"^ cm. are termed sM6??wcroms, and the existence of particles of approximately lO"' cm. diameter has been demonstrated in an indirect manner by Zsigmondy; these are known as amierons. Below this again must lie the apparent homogeneity of true solutions, which, however, must ultimately be regarded as truly heterogeneous. J2 GENERAL PROPERTIES OF COLLOIDS There is thus unbroken continuity between the coarsest-grained heterogeneity and the apparent homo- geneity of the true solutions, and the molecular state in gases. Historical Note. — Graham is usually regarded as the founder of colloid chemistry, and so in a certain sense he was, but many of the properties on which he based his ideas, and some of the ideas, were already known to his predecessors, as were also many of the ideas which have been put forward since his time. A glance at the section of the book dealing with methods of preparation will show that metal sols, prepared by reduction of salts, were fairly well known, and several investigators expressed the view that these so-called metal solutions were suspensions and not solutions at all. While the metal sols formed by kathode reduction were generally considered to be hydrides, Euhland (1815) and Poggendorf (1848) pronounced them to be metals in a very fine state of division. And Berzelius, in his "Lehrbuch" (1844), having described arsenious sulphide sol as obtained by the action of hydrogen sulphide on arsenious oxide solution, says : " For the present this solution is rather to be regarded as a suspension of trans- parent particles, for arsenious sulphide gradually separates out as a precipitate." ^ The sulphur sol, formed by the reaction between sulphur dioxide and hydrogen sulphide in water, was discovered by Wackenroder (1846). He states that "if the liquid is frozen and thawed, a great part, but not all, of the suspended sulphur separates out. But it is im- mediately and completely separated out in large flocks if ' " Inzwisohen ist diese Losung wobl melir als eine Suspension von durohsiohtigen Teilchen zu betraehten, denn allmiihlioii scheidet sieh das Schwefelarsenik ab und fallt nieder." INTRODUCTION 13 a neutral salt of an alkali, as NaCl, is added to the acid liquid." 1 And Sobrero and Selmi (1850), in an elaborate account of the same sulphur sol, make the following remarks : " If water is added to it, it divides up, forming an emulsion, from which it does not separate out, even on prolonged standing (several months)." ... "If a little aqueous solution of a neutral potassium or sodium salt is added to the emulsion, sulphur is immediately precipitated, but, singularly, if a sodium salt is used, the sulphur does not lose the property of dividing itself up again in water. All that is necessary is to decant the liquid containing the sodium salt, and to wash the precipitate several times with distilled water ; after the second or third washing the sulphur does not settle out, but regenerates the emulsion. If a potassium salt, especially the sulphate has been used, the precipitated sulphur has completely lost the property of emulsifying in water. ... In spite of repeated washing, it always retains a trace of the potassium sulphate employed for the precipitation. . . ." " This enormous quantity of sulphur is, one would be inclined to say, dissolved, for it hardly affects the limpidity of the liquid. . , . Sulphur can thus be modi- fied in an extraordinary manner by the siibstances present at the time of deposition, these adhering obstinately, probably by simple adhesion, and can either acquire the property of emulsifying in water, or assume a state of aggregation which prevents it dividing up in water. It thus appears that sulphur exhibits phenomena analogous ' " Lasst man sie gefrieren und wieder aufthauen, so soheidet sich zwar ein grosser Theil, aber nioht aller suspondierter Sohwefel ab. Augenblioklioh jedooh wird derselbe in grossen Plooken rein abgesohie- deu, wenn man die stark saure Plussigkeit mit ein em neutralen Alkalisalze, vornehmlich mit Koohsalzlosung, vermisoht." 14 GENERAL PROPERTIES OF COLLOIDS to those observed with many other substances, which possess the power of dispersing and dividing themselves in a liqnid, without- completely dissolving in it, as e.g. soap, starch, and prussian blue, on which one of us (Selmi) has previously made observations similar to those now described. These facts are related to a set of phonomena which M. Selmi has classed together under the name of pseudo-solutions. The number of pseudo- soluble substances seems to be pretty large." ^ The first observations on the hydrolysis of salts ^ " Si Ton y ajoute de I'eau 11 s'y divise en formant une Emulsion dont il ue se s^pare plus, mtoe par. un repos tr^s-prolongd (plusieurs mols)," ... " Si, a I'^mulsion de se soufre, on ajoute un peu de solu- tion aqueuse d'un sel neutre de potasse ou de soude, on obtient immS- diatement un preoipit^ de soufre, mals (ohose singulifere) si Ton a employ^ un sel de soude pour la precipitation, le soufre n'a pas perdu la propriety de se deviser dans I'eau. II suffit, pour s'en assurer, de decanter le lic[uide contenant la sel sodique, et laver le pr^cipitS plusieurs fois avec de I'eau distillSe ; au deuxi^me ou au troisi^me lavage, le soufre ne se depose plus : il xhginkte I'fimulsion. Si, au oontraire, on a employe un sel potassique, surtout le sulfate, le soufre pr6cipit6 a perdu oomplfetement la propriety de s'^mulsionner dans I'eau. . . . Malgr6 les lavages r^p6t6s, il retient toujours un peu de sulfafe de potasse employ^ pour la precipitation." . . . " Cette 6norme quantite de soufre est, on dirait, dissoute, car eUe n'alt^re presque pas la limpidity du liquide. . . . Le soufre peut done §tre modifi^, dans sa manidre d'etre, d'une fajon toute par- ticulifere par la presence des corps au milieu desquels il se depose, et qui y adherent avec opini^tret6, probablement par simple adiiesion, et acquferir tantot I'Smulsion ability, tantot un 6tat d'aggregation qui I'empeche de se diviser dans I'eau. II resulte, en outre, que le soufre 6mulsionnable presente des phfinom^nes analogues a ceux qui s'observent dans beaucoup d'autres corps qui jouissent de la propri^t^ de se disperser et se diviser dans un liquide, sans toutefois s'y dissoudre absolument, tels que le savon, I'amidon, et le bleu de Prusse, sur lequel un de nous, M. Selmi, a d^ja fait des observa- tions analogues a celles que nous venons d'exposer. Ces faits se rattachent a un ordre de phtoom^nes que M. Selmi a bien caract^- ris^s, et qu'il a r^unis sous le nom de pseudosolutions. II parait que le nombre des corps pseudosolubles est assez grand." (Ann. Chim, Phys., 1850, [iii.] 28, 210-218.) IIMTKUUUUiiUJN ■ 15 resiilting in the formation of sols were made by Gay Lussac (1810) on aluminium acetate. Berzelius ^ then gave a full description of the preparation of j8 silicic acid by the hydrolysis of silicon sulphide. " j8 silicic acid is obtained in its purest form when silicon sulphide is oxidised by water ; hydrogen sulphide is evolved as gas, and the j8 silicic acid dissolves in the water. In a more concentrated state the solution soon turns to a gelatinous mass." Fremy 2 (1853), on the same subject, added little to Berzelius' account ; he says : " The hydrated silica remains completely dissolved in the water, and this siliceous water is sometimes extremely stable; if dilute, it can be kept for months without depositing any silica, but it becomes gelatinous and gives up the silica when it is concentrated or boiled, or when a soluble alkaline salt is added to it." The peptisation of silicic acid gel by alkalies was described by both Berzelius (1833) and by Kiihn (1853). The latter 3 also states that "by treating silicic acid jelly with water it is easy to obtain 5 per cent, solutions which 1 " In ihrem reinsten Zustand bildet sie sich duroh Ozydation von Schwefelkiesel auf Kosten von Wasser; es entwiokelt sioh Sohwefelwasserstofi-gas, und die $ Kieselsaure lost sioh in Wasser auf. In konzentrierterem Zustande gesteht die Losung bald zu einer gallertartigen Masse." (Berzelius, Lehrbuch, 3 auf. 1833.) 2 " la silica hydratSe qui reste enti^rement en dissolution dans I'eau. Oette eau silioeuse presente quelquefois une grande stability ; on peut la conserver pendant des mois entiers lorsq'elle est ^tendue sans qu'eUe laisse deposer de la silioe ; mais elle devient g^latineuse et abandonne la silioe qu'elle tient en dissolution quand on la concentre lorsqu'on la ports t I'^buUition ou qu'on introduit dans la liqueur des sels aloalins solubles." (Ann. Chim. Phys., 1853, [iii] 38, 312.) ' " Die Hauptbedingung zur Darstellung einer soloher Auflosung besteht darin, dass KieselsauregaUerte, welcbe dazu dienen soU, in mogliobst ausgedehnter BesohafEenbeit bereitet und gewissermassen so lange in statu nascendu erhalten wird, bis sie zur Auflosung hinlanglicb vorbereitet ist." [J. pr. Chem., 1853, 59, 1.) i6 GENERAL PROPERTIES OF COLLOIDS are quite stable. The chief condition in making such solutions is to have the silicic acid in as fine a state as possible, and to keep it in statu nascendi until it is being prepared for dissolving up." The researches of Orum (1853) and of P^an de Saint Gilles (1854) on the hydrolysis of aluminium acetate and ferric acetate come next. In the former case Crum obtained pure aluminium hydroxide sol by heating the dilute solution for ten days in a closed vessel, and boiling off the acetic acid in shallow basins. He failed to do the same with ferric acetate, but Pean de Saint Gilles suc- ceeded in the following year, although he was unable to get rid of all the acetic acid. Both of them carried out precipitation reactions with neutral salts. Then with Graham's comprehensive researches on dialysis and its application to the purification of sols, and on the preparation of the impure sols by peptisa- tion, by washing out, by double decomposition between salts, and by the action of acids on soluble salts, as sili- cates, stannates, etc. (1861-1864), we enter on the period when colloid chemistry was recognised as a special branch of science. The historical development of each special department is briefly outlined in the text. The treatment of the subject falls naturally into three divisions — the general properties of colloids (sols aud gels) ; the methods of preparation, and the factors which govern the production and stability of colloids ; the con- nection between the properties of colloids and surface phenomena, including in particular surface concentration and adsorption in gases and solutions. If justification is needed for dealing with the properties of colloids before detailing the methods of preparation, it may be found in the fact that although obviously colloids were obtained or INTRODUCTION 17 prepared before their properties could be ascertained, the latter were earlier made the object of scientific investiga- tion. Thus, while at first the production of a new colloid was of the nature of an accident, their general behaviour was well made out on these natural or accidentally discovered colloids. Not until these general properties had been recognised did it become possible to treat systematically the methods by which colloids might be produced and preserved. This, indeed, has only been done within the last few years by von Weimarn. The order adopted in the text is therefore — I. The General Properties of Colloids (sols and gels). II. The Methods of Preparation. III. Surface Phenomena and Adsorption. IV. Applications of Colloid Chemistry. CHAPTER II GENERAL DIFFERENCES BETWEEN SUSPENSOIDS AND. EMULSOIDS Introduction. — Suspensoids are usually defined as sols in wHcli the disperse phase is solid, while in emulsoids it is lic^uid, the dispersion medium being in each case liquid. Now this alone will not account for all the divergences between them. It is, indeed, not impossible that in both the disperse phase is liquid. Apart from the theoretical considerations, based on the optical behaviour,^ which are in favour of spherical particles, there are many facts which point to the same conclusion. In many of the methods of preparation, there is every scope for Ostwald's " Eule of the appearance of the least stable form." As has been shown time and again, when a new phase, which is finally solid, makes its appearance suddenly, whether from vapour or solution, it appears first as a liquid ; it may run through many intermediate (labile) forms before reaching its final solid form. Sulphur, for instance, forms globules, which crystallise later. Crystal- lisation as spherolites is well known. Even in the case of some metals, e.g. burnished gold, silver, the surface layer has been found to be of the natiu'e of a liquid film.^ The difference must, therefore, be sought elsewhere. Quincke long ago expressed the opinion that what we > MaxweU Garnett, Phil. Tram., 1904, 203, 385 ; 1906, 205, 237. » Beilby, Proc. Boy. Soc, 1903, 72, 218. SUSPENSOIDS AND EMULSOIDS 19 now call emulsoids were two-phase liquids, each of which contained both components. Gelatine sols contain amicrons only at high temperatures, while sub microns appear at lower temperatures; the same occurs with agar sols. Hardy 1 investigated agar, and found that, by pressing out liquid from the gel, two phases were obtained, which contained agar in different amounts, e.g. Agar gel (2*23 per cent. agar). Concentration of Agar. Temp. Liquid. Solid. 36° 0-47 3-2 13" 0-12 4-7 5° 0-09 3-0 Spiro also regarded the " salting out " of albumin as a separation into two liquid layers, each of which contained all three components, but in different proportions. In suspensoids, on the other hand, we have a disperse phase, solid or liquid, which does not combine with the dispersion medium, and is practically insoluble in it, e.g. the metals, sulphur, metallic sulphides, etc. The hydrox- ides can combine, more or less, with water, and are more like emulsoids in many respects. Emulsoids may then be regarded as intermediate between suspensoids and solutions, and may in some cases be solutions of substances of high molar weight. Some evidence of this will be found below (see e.g. Optical Properties). The behaviour of silicic acid also points in the same direction ; when first made by the action of hydrochloric acid on water glass, it is in true solution, as shown by its optical properties and its diffusion. Graham found that 5 per cent, dialysed away in 24 hours, and that no further loss occurred in the next four days.^ ' Proc. Boy. Soc, 1900, 66, 95. ' Phil. Trans., 1861, 151, 183. 20 GENERAL PROPERTIES OF COLLOIDS The soluble acid changes with time into a sol, in which Zsigmondy found amicrons only. Even this diffuses to some extent (see below) and has a distinct osmotic pres- sure ; its viscosity, too, increases continuously until gel is formed. Many dyes exhibit similar anomalies; methyl violet diffuses rapidly through parchment paper, but has abnormally high molar weight, and, in the ultramicro- scope shows a light cone together with a few sub- microns. Many other instances might be given, in which there are evidences of the existence both of solution and of sol, but one more must suffice. Casein is an acid sub- stance, insoluble in water. It combines with a fixed proportion of alkali, and is then soluble. This solution possesses a good electric conductivity, which changes normally with dilution, but it does not diffuse through parchment paper. Its molar weight, based on its con- ductivity and its combining weight with sodium hydroxide, is estimated ^ at 5000 ; a substance with this molar weight should diffuse fairly well (see below). This fundamental difference between suspensoids and emulsoids finds expression in many of their general pro- perties, and most of all, in those properties which undergo marked changes as the result of solution ; such are density, viscosity, and surface tension. Density. — The density of suspensoids can be calculated by the law of mixtures, as Linder and Picton 2 proved for arsenious sulphide sol. Density AsjSj (per cent.). Obaerved. Calculated. 4.4 1033810 1-033810 2-2 1-016880 1-016905 1-1 1-008435 1-008440 0-01719 1000137 1-000134 ' Sackur, Zeitsch. physikal. Chem., 1902, 41, 672. * Chem. Soc. Journ., 1895, 67, 71. SUSPENSOIDS AND EMULSOIDS 21 The densities of emulsoids do not appear to have been determined accurately, but contraction occurs when gelatine and starch take up water (i.e. the volume of starch + volume of water > volume of sol). When a ferric hydroxide sol is precipitated, there is a slight dilatation,^ whereas the volume of an arsenious sulphide sol is unchanged. According to Gayda,^ the density of albumin is 1-299 ; a volume contraction takes place on solution, and also a slight contraction on dilution of a concentrated solution. Chick and Martin * give the density of dry casein as 1'318, the density of a 7'85 per cent, sol is 1-024, which gives 1-39 for dry casein. Similar results were obtained with egg albumin and serum albumin. Surface Tension. — Linder and Picton found the surface tension of arsenious sulphide sol and ferric hydroxide sol to be the same as that of water, and the same is true for metal sols. The surface tension of emulsoids is, in many cases, much smaller than that of water. Quincke* gives numerous examples; thus, the surface tension of dilute sol of egg-white is 28 per cent., and of gelatine 12 per cent, less than that of water. In accordance with this difference, suspensoids are not adsorbed, as a rule, while emulsoids usually are. (See Adsorption, Chap. XVI.) Viscosity. — As the concentration of most suspensoids is small (Zsigmondy's gold sol 0-12 per cent, arsenious sulphide sol 4"4 per cent.; these are maximum concen- trations. Metal sols are usually below 0*05 per cent. • Chem. Soc. Jown., 1895, 67, 71. ' Bidchem. Zeitsch., 1912, 39, 400. » Koll. Zeitsch., 1913, 12, 69. * Wied. Ann., 1885, [iii] 36, 582. 22 GENERAL PROPERTIES OF COLLOIDS and the others below 1 per cent.), their viscosity is not likely to differ greatly from that of the dispersion medium. On precipitation, the viscosity is usually found to be slightly decreased. The concentration of emulsoids is not so restricted, and their viscosity is often much greater than that of the medium; it also steadUy increases with increase of con- centration, as the following figures ^ show : — Sol Concentration t Viscosity Gelatine . 1 % 20° 0-021 >j • 2 J) 0-037 Egg-white . 1-6 „ 17-5° 0-0115 >» . 2-9 „ » 0-0125 Silicic acid . 0-81 „ 20° 0-012 )) . 0-99 „ jj 0-016 i) . 1-96 „ 3) 0-032 . 3-67 „ „ 0-165 The viscosity of water at 17-5° is 0-0126, and at 20° it is 0-0120. Pauli^ gives the viscosity of 1 per cent, albumin as 1-068 (water = 1). As both these results for albumin were obtained by the capillary tube method, it is not easy to reconcile the differences found for albumin. The swinging disc method applied to emulsoids is known to yield values which are invariably higher than those by the capillary tube method. The difference may amount to 10 per cent. This and other peculiarities of the results of the swinging disc method are explicable on Quincke's assumption, which Garrett also adopts, of two liquid phases, each containing both components. Ferric hydroxide sols also give a difference in viscosity, as determined by these methods. Pauli observed that the viscosity of albumin sols is • Garrett, Dissertation, Heidelberg, 1903, p. 51. « Eolloidchemische Studien am Eiweiss, 190S, p. 9. SUSPENSOIDS AND EMULSOIDS 23 diminished by small amounts of neutral salts (Fig. 1), while it is greatly increased by both acids and alkalies. It is increased 18 per cent, by O'Oln HCl, or by 0"03n. Fia. 1. 001 OOZ 003 ConcentraJiOTL. 004- OOSn, NaOH. The effect of neutral salts is attributed to adsorp- tion, that of acids and alkalies to chemical changes (in the direction of solution). There are, according to Wo. Ostwald,^ several other factors affecting the viscosity of emulsoids, e.g. dispersity, hydration or solvation of the particles, electric charge, previous thermal or mechanical treatment, addition of small amounts of more viscous sols, and time (p. 121). Hatschek ^ has developed a theory of the viscosity of two-phase systems, which leads to the equation V = r,(l + 4-5/) where i>. and x\ are the viscosity coefficients of the dis- persion medium and the sol, and / is the ratio of the volume of the disperse phase to the volume of the sol. The same expression, with the exception of the numerical value of the coefficient of/, has been obtained by Einstein ' and others. ' For these and other references on this subject see Trans. Faraday Soc, 1913, 9, 34-107. CHAPTEE III DIFFUSION AND DIALYSIS In the course of his experiments on liquid diffusion, Graham found that solutions could be divided into two classes, according to their rates of diffusion. Salts and crystalline substances generally {crystalloids) had diffusion coefficients of the same order, while certain non-crystalline organic substances {colloids), as starch,'gum7gelatine, etc., and some inorganic substances, diffused at a very slow rate, or not at all. The division was not very rigid, for the rates i of diffusion in each class varied within wide limits. On extending his investigations to the diffusion of solutions into a gelatine gel, he found the two classes were more sharply differentiated, in that the diffusion of crystalloids in the gel was practically the same as in water, whereas that-of the colloids was reduced to a very small amount, and in most cases was entirely prevented. Thus, the diffusion of a 10 per cent, solution of sodium chloride into water, and into a 2 per cent, gelatine gel were the same, as the following figures show : — Number of Water 2% gelatine layer. 1 days at 9*. 8 days at 10°. 1-2 0-030 0-030 3-4 0-079 0-061 5-6 0-215 0'212 7-8 0-529 0-562 9-10 1-100 1-116 11-12 1-907 2-168 13-14 2-844 2-393 15-16 3-294 3-450 DIFFUSION AND DIALYSIS 25 Caramel, on the other hand, had scarcely begun to diffuse into the gelatine at the end of 8 days, so that no similar table for it is possible. Gelatinised starch, gum, and coagulated albumin, also had very little effect on the rate of diffusion of crystalloids. It wUl be noticed that the two series of figures are by no means identical, or even parallel, although Graham concluded from them, and from other similar results, that diffusion of crystalloids is not hindered by gels.^ \ In the further modification of the diffusion experiment suggested by these results, the solution is separated from the pure solvent by a thin layer of colloid, such as a sheet of parchment paper or animal membraiie. 'The differences \ between crystalloids and colloids were again observed, ! and in a more marked degree. The former readily diffused through such a septum, frequently with imdiminished velocity, while the diffusion of the latter, already small, was stiU further reduced. The connexion between the two phenomena is clearly shown by the figures below, •compiled from Graham's results : — Times of equal Amount dialysed Substance. Difiusion. in equal times. Sodium chloride ... 1 I'O Ammonia 6 0-85 Alcohol 2 0-47 Glucose 3 0-36 Cane sugar .... 3 0'47 Gum arable .... 7 0-008 Tannin 11 0-015 Albumin 21 0003 Caramel 42 0-005 This process, which obviously permits of the con- venient separation of crystalloids from colloids, is known » Graham, Experimental Besearches, p. 598. 26 GENERAL PROPERTIES OF COLLOIDS as dialysis. The apparatus employed by Graham con- sisted of a short, wide glass cylinder, or of a wooden hoop, on which was tied a sheet of bladder, or of parch- ment paper. The dialyser was floated in a larger vessel containing water, which could be renewed from time to time. This form of dialyser is not at all convenient. The interface, on which the rate of dialysis must depend, is relatively small; again, the solutions to be dialysed, frequently of a very unstable nature, are exposed to atmospheric influences for prolonged periods. Various practical forms'STdlalyser are now obtainable. Tubing of parchment paper may be had in any length, and is obtainable from any instrument dealer. It should always be tested for leaks, which can usually be closed up with gelatine, or with white of egg, which is then coagulated by heat. Before use, the tubing should be well purified by soaking it, inside and outside, in many changes of pure water. Inattention to this important detail is accountable for many undesired results, especially in the dialysis of sols which are coagulated by minute traces of certain salts. In the dry state the tubing is liable to crack, and should be well soaked before use. It can then be bent into a U, and placed in a beaker; if the U-tube projects a little above the edge, a slow stream of water can be run through the beaker. With the large surface and the constant renewal of the water, dialysis is rapid. As the ends can be closed with corks, or folded together, the risk of contamination or decompo- sition is much reduced. For small quantities of substance, and for substances like albumin, parchment paper thimbles, as made by Schleicher and SchuU, are very convenient. They may be mounted in the following way : slip the open end over a carefjilly fitted glass tube, fixing it by means of a short DIFFUSION AND DIALYSIS 27 piece of good rubber tubing. This joint is quite water- proof, and the rubber does not come in contact with the sub- stance. The tubes are clamped in a beaker, so that the thimble is completely immersed ; the water is constantly- renewed in the beaker. The whole tube can be sterilised and the glass ends can be plugged with antiseptic cotton wool With these thimbles the author has dialysed crystallised egg albumin until it was electrolyte-free, without fermentation occurring, even in summer. Eecent investigations have shown that parchment paper is not the most efficient septum. It is slower, and may not only retard the diffusion of the crystalloids, but may also allow the colloid to escape. Collodion films, if suitably prepared, are highly efficient, effecting separations more completely in much less time than does parchment paper. They should be made when required, and can be made of any desired size or shape ; the degree of perme- ability can also be varied to suit given requirements, by a simple alteration in the manipulation. Full instruc- tions and bibliography are to be found in a paper by Bigelow and Gemberling ; ^ but the following details will enable anyone to make them. The collodion solution is made as follows : 75 c.c. of ether are poured over 3 g. of commercial pyroxylin in a corked flask; in 10-15 minutes 25 c.c. of ethyl alcohol are added. The pyroxylin dissolves quickly and com- pletely to a clear liquid, which does not require filtra- tion. To make a sheet, pour a few c.c. of the solution on a clean, dry glass plate, and spread it by tilting to and fro. Allow the layer to dry until it is gelatinous, and will bear lightly rubbing without wrinkling. Loosen the edges, and peel off. Plunge into water. Larger sheets of uniform ' Amer. Chem. J., 1907, 29, 1576. 28 GENERAL PROPERTIES OF COLLOIDS thickness can be made by pouring the collodion on to a surface of mercury in a shallow dish. To make a collodion sac proceed as follows : A small orifice, 1-2 mm, in diameter, is blown in the bottom of a tube of the desired diameter. This hole is closed with one or two layers of collodion, none of which should go inside the tube. This can be done by touching the bottom of the tube with a cork wet with collodion, allowing a portion of the solvent to evaporate, and repeating as often as necessary. The tube is then coated with collodion by rotating it on the surface of the collodion, which is in a tilted vessel. The tube is exposed to air until the coating has "set," and does not stick to the finger; it is then plunged into water, and water is also poured into the tube. If it is immersed too soon, the film is white, opaque, and brittle ; if too late, the film cannot be removed without injury. The proper time, between 2-15 minutes, depends mainly on the consistency of the collodion ; it is easily found after a few trials. The sac is removed by blowing into the tube and gently pulling the membrane ; this forces water through the orifice and between the sac and tube. They can easily be made up to 40 cm. by 2-3 cm. A very convenient and effective collodion dialyser is obtained by soaking a dried Soxhlet thimble in the above collodion solution, allowing it to set, and plunging it into water. The thimble is cemented to a glass tube by means of collodion, and is then inserted in a glass tube which just admits the thimble. Highly pure water is run in at the bottom of the jacket at a very slow rate, and escapes by a side tube close to the top. Dialysis is very rapid and complete ; e.g. the electric conductivity of a ferric hydroxide sol, prepared by Kreke's method (p. 195), fell in 2-3 days to a value much lower than any hitherto recorded. DIFFUSION AND DIALYSIS 29 A comparison of the efficiency of collodion, parchment paper, and gold-beater's skin will be found below. The last-named material is the best yet used for dialysis, and parchment paper is the worst, although it still seems to be the one most frequently employed. As a test, a Zsigmondy gold sol was dialysed in three dialysers of identical pattern and size, the septa being gold-beater's skin, col- lodion, and parchment paper. The rate of dialysis was followed by determining the specific conductivity (k) every 24 hours. Time of Dialysis Specific 1 Conductivity r{icxi 0»). in days. Gold-beater's skin. Collodion. Parchment paper, 984 984 984 1 305 416 850 2 81 208 468 4 55 102 256 5 30 85 159 10 23 26 47 12 11 21 43 17 8 11 15 19 11 11 The undoubted superiority of gold-beater's skin is obvious. If a time-conductivity curve is drawn, counting in each case from the conductivity 11, it will be seen that it takes about the same time (6-7 days) in each case to reduce the conductivity from 30 to 11, all the differences occurring in the earlier stages of dialysis. Membrapes made from pure samples of coUodion are more durable, and also more permeable, than those made from commercial samples. In any case, the permeability decreases with age, but, as a rule, a film is useful for one to three months. Dialysis is the basis of one of the general methods of preparing sols. It was also for a long time the generally 30 GENERAL PROPERTIES OF COLLOIDS accepted criterion of a sol, i.e. a substance, which to all appearance was in solution, but yet was unable to diffuse through a dialyser, was classed as a colloid, and the apparent solution as a sol. It should be particularly observed that Graham's results do not show that, as has occasionally been assumed, there is no diffusion and dialysis of sols. Though his classification into crystalloids and colloids is baaed upon the differences between them in this respect, he every- where gives figures for the dialysis and diffusion of colloids ; indeed, in the same paper he states in this very connexion, " in nature there are no abrupt transitions, and distinctions of class are never absolute." It may have been convenient to disregard the tran- sitions, and to regard colloids as non-diffusible, but recent, more accurate investigation has confirmed the correctness of Graham's view, and his experimental results, qualita- tively at least. Thus, Linder and Picton i found that an arsenious sulphide sol diffused to a considerable extent. After 11 days, the outer vessel contained 0"1185 g. of arsenious sulphide in 350 c.c. of water, and the inner cylinder contained 0*8210 g. in 77 c.c. Further experi- ments with carefully purified sols gave similar results. Again, in another direction, Bechhold and Ziegler^ have found that the diffusion of both electrolytes and non- electrolytes is diminished by gels, the effect depending on the concentration of the gel. The addition of various substances to the gel causes further alteration in the rate of diffusion; some, e.g. sodium sulphate, reduce the diffusion still further, others, as urea, increase it. As many other properties of gelatine (and other gels) are changed in opposite directions by these substances, this is probably ' Chem. Soc. Journ., 1892, 61, 187. ' Zeitsch. ■physikal. Chem., 1906, 56, 105. DIFFUSION AND DIALYSIS 31 another illustration of this influence (see Lyotrope series, pp. 7 and 123). Finally, the presence of crystalloids in a sol often has a marked influence on the dialysis of the sol. For instance, a silicic acid sol will diffuse through parch- ment paper if mixed with sodium chloride. The fairly rapid diffusion of some sols may be due to this influence. The disperse phase will adsorb small amounts of the crystalloids, and their presence may cause diffusion. Linder and Picton noticed that an arsenious sulphide sol, to which a soluble tartrate had been added, diffused appreciably into water, although the particles were so large as to be just visible under the microscope. Or, perhaps, this also may be another instance of lyotrope influences. The theory of the Brownian movement leaves no room for doubt that sols diffuse, the difference between colloids and crystalloids being one of degree only. CHAPTER IV OSMOTIC PRESSURE AND MOLAR WEIGHT Closely connected with the question of dialysis is that of osmotic pressure, for, as is well known, direct measure- ment of osmotic pressure depends on the impermeability of the membrane by the solute, while the solvent can freely pass through it. The most important application of the laws of osmotic pressure is to the determination of the molar weight of soluble, non- volatile substances. This was done by van 't Hoff', who proved that the gas laws could be quantitatively applied to dilute solutions, osmotic pressure taking the place of gas pressure, and volume of solution the place of gas volume. Thus the formula for calculating molar weights, which was previously limited to gases and volatile substances, was at once extended to all soluble substances. But before entering into details of the osmotic pressure of sols, and the attempts to deduce the molar weight of colloids from the related methods of depression of freezing point, etc., it is of interest to note that Graham, from purely chemical considerations, for these other methods were not known until twenty years later, had been led to assign high molar weights to colloids. It seems not at all unlikely that this idea, thus early expressed, dominated later research in this direction for many years. From the fact that sols of ferric hydroxide and aluminium hydroxide, prepared by dialysis of the basic OSMOTIC PRESSURE AND MOLAR WEIGHT 33 chlorides, could not be entirely freed from chloride, how- ever prolonged the dialysis, Graham ^ concluded that they were compounds of (condensed) bases of very high molar weight, and deduced the chemical equivalent of the base in the usual manner from the ratio of base to acid, which he had found to be approximately constant. After 19 days' dialysis, there was one equivalent of HCl to 30*3 of re203 ; after 38 days there was one equivalent of HCl to 31 "2 of Cr203; similar ratios were found for other similar sols (see p. 108). Similarly, the gels of silicic acid and stannic acid are dissolved by very small quantities of alkali, eg. 1 part of sodium hydroxide liquefies 200 parts of silicic acid gel, and the acid reaction of 100 parts of the latter is neutralised by 1*85 parts of potassium oxide. These compounds of acids or bases of high chemical equivalence were distinguished by the prefix " colli " or "co." Graham suggested that "the basis of coUoid- ality may really be this composite character of the molecule." * Osmotic Pressure. — Pfeffer, in his classic researches on osmotic pressure, gave measurements of the pressure of some sols. The values are small in comparison with those for other substances. Thus the pressures of 1 per cent, solutions of the following substances ^ are — Potassium sulphate 193 cm, mercury „ nitrate . 178 ,, Cane sugar . . . 47 „ Dextrin .... 16-5 „ Gum arable . . . . 7-2 „ That this is a real pressure, in the case of gum at least, ' Graham, Experimental Researches, pp. 582, 596. ' Osmot. Unterauchimgen, Leipzig, 1877. D 34 GENERAL PROPERTIES OF COLLOIDS is supported by the proportionality between concentration and pressure, as the following figures show : — c P p/e 1 7-2 7-2 18 119-7 6-7 jj 120-4 6-9 Whether this is to be ascribed to the colloid itself, or to impurities not completely removed from it, has long been a debatable question. On the one hand, there should be less difficulty in determining the osmotic pressure of sols than of solutions, for it is clear that semi-permeable membranes are much more impermeable by colloids than by electrolytes or ordinary solutes. This difference would lead to dialysis of the impurities, in consequence of which the osmotic pressure would steadily fall during the experiment. Further, a sol prepared in different ways, and possibly containing different impurities, would be expected to have different pressures, according to the nature of the impurities. This is not at all what is observed. Different prepara- tions of a sol may have different initial osmotic pressures, but they fall to the same final value, which is not zero, but a definite positive value. It is usually small, but is too large to be put down to experimental error. It may be due to a small residue of electrolyte or other active solute which cannot be removed by dialysis, but in this case it must be regarded as something other than an impurity, rather as an essential constituent of the sol (see p. 108). Suspensoids. — Direct measurement of the osmotic pressure of suspensoids gives very variable values, which are always small. The reason for this want of concord- ance is not very apparent. Perrin's experiments * on the » Ctrnpt. Bend., 1908, 146, 967 ; 1908, 147, 330. OSMOTIC PRESSURE AND MOLAR WEIGHT 35 arrangement of the particles of a gamboge sol under the action of gravity and its own osmotic pressure, based upon the kinetic theory, afford an estimate of the osmotic pressure, since all the other factors are known. The relation is logJ^l,„^A(l_i) where Mq ^^^ '"' ^^® ^^^ number of particles in unit volume in the layers at the heights and h, m is the mass and p the density of the particles, g is the gravity constant, and k is the osmotic pressure of a single particle in unit volume. The value of A is 43 X 10" ^^ On the assumption that this is the same as the pressure exerted by the particle as a gas or solute, , ET N being the number of molecules in a mol (6 X lO^^). If R is calculated from these values, it comes to 2'1 cal. (instead of 1-98 cal.). The molar weight of gamboge calculated from these data is 3 X 10^. If use is made of Thovert's 1 relation, M X D^ = con- stant (60 X IQ-i"), the molar weight of arsenious sulphide calculated from the diffusion constant 01 X 10-^ is over 6000. Emulsoids. — The molar weight of albumin calculated in the same way from Graham's value of D = 0'07 X lOrS at 13° is 13,000. This molar weight corresponds to an osmotic pressure of about 7 mm. mercury (= 100 mm, water), which ought to be measurable. With the greater diffusion of emulsoids in general, there is also a greater osmotic pressure, and recently direct measurements have > Compt. Bend., 1901, 133, 1197 ; 1902, 184, 507 ; 1902, 135, 579. 36 GENERAL PROPERTIES OF COLLOIDS been made. The osmometer membrane is usually col- lodion or parchment paper. The following results may be of use : — Concentration. Osmotic pressure. Molar Sol. g. per litre. mm. mercury. weight Egg albumin ' . 12-5 2(K^ 11,000 Gelatine ^ . . . 12-5 6 36,000 Starch iodide ^ . 30 15 34,000 Dextrin 3 . . . 10 165 1,000 Gum arabic ^ . . 10 72 2,400 Probably the best values are obtained from the diffu- sion-constants and the above relation of Thovert. The diffusion constants of several albumins and enzymes have been accurately determined by Herzog.* The results are as follows : Substance. D at 18°. M.W. = ^ Egg albumin . . . 0-059 17,000 Ovomucoid . . . 0*044 30,000 Clupein sulphate . 0-074 11,000 Pepsin 0-070 12,000 Lab 0-066 14,000 Invertin .... 0-033 54,000 Emulsin .... 0-036 46,000 Freezing Point, Boiling Point, Vapoui- Pressure. — It is quite evident from the above estimates of molar weights, that it is useless to apply the ordinary molar weight methods to sols. Taking the freezing point method as the most accurate, and the one which gives the largest number to be determined, an osmotic pressure of 9 mm. ' Lillie, Amer. J. Physiol., 1907, 20, 127 ; see also Moore and Boaf, BiocUm. J., 1906, Z, 34. 2 JBodewald, Zeitsch. physikal. Chem., 190O, 33, 583. , ' Pfefier, Osmot. Untersuchungen, Leipzig, 1S77. ' Zeitsch. Elektrochem., 1907, 13, 533. OSMOTIC PRESSURE AND MOLAR WEIGHT 37 mercury (= 125 mm. water) corresponds to a freezing- point depression of only 0'001°. Owing to the nature of these emulsoid sols, the experimental error is certainly much larger than in ordinary cases, and, obviously, large concentrations would be necessary to give a readable depression. The question then arises how far such con- centrated sols can legitimately be considered to be dilute. The determination of the boiling point or of the vapour pressure of these sols is attended with quite unusual difficulties, the nature of which need not be specified, the more so as no useful estimate of molar weight is afforded by them. Many attempts to estimate the molar weights of col- loids were made in the earlier times, and frequently with most inadequate apparatus and methods. The results are therefore quite valueless, though, strange to say, they have found their way into text-books, probably because they were in accord with preconceived ideas. CHAPTER V OPTICAL PEOPEKTIES Macroscopic Observations. — The heterogeneity of a coarse suspension is usually at once apparent to the unaided eye, because of its opaque appearance in transmitted and in reflected Ught. As the size of the particles decreases, so too do these effects. We thus have all gradations between opacity and absolute transparency, when judged by the naked eye. Frequently, however, the heterogeneity is still observable when a powerful beam of light is sent through the liquid. This was noticed by Faraday ^ to be the case with the gold sols prepared by him. When the sol appears to be quite clear in ordinary light, the appearance, when a concentrated beam of light is sent through it, is similar to fluorescence. It differs from fluorescence in that the light is polarised in its passage through the liqtiid, which is not the case with fluorescence. Tyndall Effect. — This phenomenon is usually known as the Tyndall Effect, as Tyndall 2 applied it first with good results in the case of dust and mist in the atmo- sphere. The light is mainly polarised in a plane normal to the path of the beam, but the amount of polarisation and the angle depend on the size of the particles. The extinction with a crossed Mcol prism is therefore only easily observed when the particles are of approximately • Phil. Mag., 1857, [iv.] 14, 401, 512. » Phil. Mag., 1869, [iv.] 37,384; Proc. Soy. Soc, 1869, 17, 223. OPTICAL PROPERTIES 39 uniform size, and the corresponding angle sought out. If the particles are below* 100 (ifi in diameter, the polarisa- tion is complete, and the angle is 90°. The Tyndall Effect is easily seen if a small rectangular slit is placed in the focus of a projecting lantern (arc by preference), and the beam is passed through a sol in a rectangular glass tank. By rotation of a mounted Nicol, such as is used for demonstration purposes in physics, the polarisation is readily detected in a Zsigmondy gold sol, or in a good arsenious sulphide soL For the reasons given above, the clearer the sol is to the naked eye, the better it is for this purpose. Even if there is not complete extinction, the difference in luminosity as the Nicol is rotated is quite striking. The absence of polarisation in the case of fluorescent solutions (quinine bisulphate, very dilute eosin, or fluorescein) should be shovra at the same time. In the Tyndall Effect we have a powerful means of detecting heterogeneity, where even the best misroscope fails to reveal it. But its delicacy is in actual practice a drawback, for it is a difficult matter to obtain an " optically void " liquid. The ordinary processes of puri- fication : filtration, distillation, are quite useless. The only certain way of freeing a liquid from accidental heterogeneous impurities is to produce a (colloid) pre- cipitate in the liquid in a closed vessel ; the settling out of the precipitate carries down all the other suspended matter.* Zinc hydroxide, ferric hydroxide, aluminium hydroxide, and barium sulphate are among the most efficient precipitates. This action is partly mechanical, _ partly a mutual precipitation of sols of opposite electric charge (p. 132). Obviously this makes the application of the Tyndall Effect to the recognition of the heterogeneity of sols a • Spring, Bull. Acad. Boy. Belg., 1899, [iii.] 37, 174. 40 GENERAL PROPERTIES OF COLLOIDS somewhat doubtful one. Spring's results show that aqueous solutions of salts of the alkalies and alkaline earths are optically clear, but that solutions of salts of aluminium, chromium, iron, copper, mercury, etc., are never so. The explanation is that these salts are hydro- lysed, and the product is not a soluble basic salt, as is frequently assumed, but a hydrosol, probably of the basic hydroxide, as the Tyndall Effect disappears when acid is added. Lobry de Bruyni maintains that the Tyndall Effect is produced whenever the molecules of solvent and solute are of markedly different size, in support of which he cites aqueous solutions of cane sugar and raffinose. Although they are usually regarded as crystalloids, the solutions show the Tyndall Effect. Colour. — The colour of sols is in many cases very striking. Thus Faraday's gold sols varied from blue to a magnificent ruby red ; other gold sols range from green to violet, silver sols yellow to green, platinum sols brown to black, arsenious sulphide yellow to orange, antimonious sulphide orange to red. As a rule, however, the colour of sols is bluish-white, e.g. sulphur and mastic, just as fine dust and mists in gases are bluish, in agreement with the theory developed by Eayleigh.^ Except in the case of dyes, emulsoids also are bluish in colour, e.g. starch sol. The colour of the coloured sols seems to depend to some extent on the dispersity, as well as on the con- centration of the disperse phase. The method of reduc- tion, the amount of reducer added, or the rate at which it is added, produce different colours in both gold sols and silver sols. Similar changes can be produced by minute ' Bee. Trav. chim., 1900, 19, 251. ' PTiil. Mag., 1871, [iv.] 41, 107, 274, 447; 1881, [v.] 12, 81 ; 1899, [v.] 47, 375. OPTICAL PROPERTIES 41 quantities of electrolytes. These changes are usually put down to change in size, and, conversely, conclusions have been drawn as to the dispersity from the colour of the sol. Zsigmondy's investigations do not bear this out, as the following table shows. The sizes were estimated by the ultramicroscopic method (see p, 54). Colour of Gold sol. Average size in /ifi. Rose 6 Deep red 10,15,18,32 Violet-red 2.3,32,75 Violet 35,95 Blue-violet .54 Dirty blue 30 Purple-red 38 It must be understood that all these sols were practi- cally as stable as the origipal sol. They could be filtered, dialysed, and kept for indefinite periods without under- going any appreciable change. Steubing's researches' confirm this result. He suggests that the shape of the particles is different, as the blue colours obtained by adding electrolytes are not the same as the colour of the original blue sols. But Svedberg's observations ^ on the alkali metal sols in ethyl ether, given below, indicate a displacement of the colour towards the red (greater wave-lengths) with increased size of particles. Colour of Ether Sols. Small Large Colour of metal- Metal, particles. particles. vapour. Li . . brown brown — Na. . purple- violet blue purple K . . blue blue-green blue-green Bb. . green-blue green green-blue Cs . . blue-green green-blue — ' Ann. Physik., 1908, [iv.] 26, 329. ' Methoden z. Berstellwng Koll. Los., 1909, p. 481. 42 GENERAL PROPERTIES OF COLLOIDS In pentane, sodium is reddish-purple, and potassium is blue at —70°, green-blue at higher temperatures. The colour is thus displaced in the same direction (towards the red) by increasing atomic weight. microscopic Observations. — As the limit of microscopic visibility is about 10-^ cm,, a sol whose particles are less than about 0'15 /* will not be recognisable even with the best microscope (magnification 2250), Schulze ^ was unable to detect any particles in arsenious sulphide sol, while Linder and Picton ^ observed them in it, but not in antimonious sulphide sol. Zsigmondy ^ and Bredig * were unable to detect particles in their gold sols, even with the highest magnification. Particles which are small enough to exhibit Brownian movement (10~3 cm,) and are yet visible under the micro- scope (10 ~^ cm.) are termed microns; those which fall below this limit are submiorons. The TJltramicroscope. — There is a wide interval between the limit of microscopic visibility (0'15 /n) and molecular dimensions (0"00016 /*, the diameter of a mole- cule of hydrogen), within which fall the submicron sols and the true solutions, if the latter do not coincide with the lower limit. It is possible to explore this region by the aid of the Tyndall Effect, but, as already explained, it is difiicult to apply, owing to its sensitiveness and un- certainty as to the absence of accidental heterogeneity. The ultramicroscope which renders ultramicroscopic particles visible, gets over this difficulty. Its lower limit, which is fixed by the intensity of the available light, is about 5/i/x (5 X 10"' cm.), i.e. about O'Ol of a wave-length " J. pr. Chem., 1882, [ii.] 25, 431. ' Chem. Soc. Joum., 1892, 61, 137. ' Zur Erhenntvfls d. Kolloide, p. 79, * Aiiorg. Fermente, p. 27. OPTICAL PROPERTIES 43 of light. This gives the lower limit of size of submicrons. Smsfller heterogeneous particles, whose existence is inferred on various grounds, are termed amicrons. The principle of the ultramicrosoope, on which the various forms of the instrument are based, is as follows : A very thin layer of the sol is illuminated by a beam of light, and as the number of particles illuminated is small the light scattered by each one is not interfered with by that from the others ; they are thus seen as coloured discs. To render these discs more readily visible, the field is kept dark, and, since the intensity of the scattered light varies as the intensity of the source of illumination, the latter is made as powerful as possible. In the Siedentopf and Zsigmondy ultramicrosoope the light from an arc lamp (or sunlight from a heliostat) is focussed on an adjustable micrometer slit, by means of which the dimensions of the beam can be varied at will to a known extent. The slit is focussed on a microscope objective, from which the beUm passes in a horizontal direction through the observation cell. This was devised by Biltz, and is a rectangular chamber in a glass tube, with quartz glass windows on two adjacent sides, one of which faces the illuminating objective, and the other (upwards) the observing microscope. The illuminated layer is a few thousandths of a millimeter in depth. The light which is not scattered by the particles passes through the ceU, and so a dark ground is secured. The dark field is secured by a simple device in the Cotton and Mouton instrument. In it the beam is not directed horizontally through the cell, but in an oblique upward direction, so that the main beam is totally reflected at the upper cover glass. The drawback to it is that the illuminated layer is not in one plane, and its volume is not known. 44 GENERAL PROPERTIES OF COLLOIDS The latest form, the cardioid ultramicroscope, deserves notice, owing to its greater simplicity in comparison with the original pattern. An extremely intense illumination of the particles, and a very dark ground are secured by the cardioid condenser, the principle of which is easily seen from Fig. 2. The Biltz chamber is replaced by a special microscope slide of quartz glass. A circular depres- sion is ground in it to a uniform known depth of a few /i, with a deeper channel at the rim. A small drop of the Pig. 2. sol is placed in this depression, and the quartz cover placed over it. The excess of liquid flows into the chan- nel, and so the illuminated layer of known uniform thickness is obtained. For general purposes of observing and counting the number of submicrons, this cell is most convenient, but for kataphoretic phenomena (q.v.) the proximity of the cell walls may be a disadvantage. There are several forms of dark-ground microscopes, which dispense with the optical bench and horizontal illumination. The illumination is not very intense, and their utility is limited to comparatively large particles. OPTICAL PROPERTIES 45 Ultramicroscopic Observations. — Moderately concen- trated sols show as a rule only a brilliant cone of light, which, on dilution of the sol, is either resolved into a number of individual bright particles, or reduced in intensity until it disappears altogether. The former is usually associated with the presence of submicrons, the latter with amicrons. But there seem to be reasons for the belief that submicrons and amicrons are, to some extent and under some conditions, reversibly, convertible into each other. The usual method of investigating a sol is to dilute it with an optically void dispersion medium, until the number of submicrons in the field of view can be counted, and to calculate from that the number of particles in the original sol by allowing for the dilution. Some recent work by Coward ^ on this point goes to show that this, in certain cases at least, may not be permissible. Although emulsoids exhibit the Tyndall Effect, some indeed being opalescent in ordinary light, their ultra- microscopic character is not, as a rule, well defined. It may be, as Zsigmondy suggested, that the visibility of the particles depends on the difference between the refractive index of the disperse phase and medium. In the case of metal sols the difference is very great ; it is not large in the case of most emulsoids. It is significant that the basic hydroxide sols are, ultramicroscopically, more like the emulsoids than the suspensoids. Other reasons have already been given for the view that they belong to the emulsoids. The ultramicroscope reveals the presence of a few submicrons with many amicrons. On dilution the light cone is gradually extinguished. Submicrons have been observed in albumin, gelatine, glycogen, and agar sols, Zsigmondy 2 carefully examined > Trans. Faraday Soc. 1913, 9, 142. ' Zur Erhenntnis d. KoUoide, p. 174. 46 GENERAL PROPERTIES OF COLLOIDS a soluble starch sol; a 3 per cent, boiled sol contained amicrons only, but submicrons made their appearance after a time. They disappeared again on heating. The ultramicroscope does not enable us to ascertain the colour of particles, as will be readily understood from the explanation of the principle of the instrument. CHAPTEE VI BEOWNIAN MOVEMENT In 1827 E. Brown,* the botanist, observed under the microscope that pollen graias, when suspended in water, are not at rest but in constant motion. They oscillate around a mean position, and the movements persisted without apparent diminution for an indefinite period. This peculiar and characteristic movement appears to take place without exception, when a disperse phase of fairly high dispersity is suspended in a medium which does not unduly impede it. The upper limit of size is not very large, for particles of about O'Ol mm. diameter just show it. At this dispersity the movements are very slow, and the distances travelled are very small; they increase rapidly with increase of dispersity. At very high dispersity (submicrons) the purely oscillatory motion observed by Brown is partially replaced by a continuous irregular motion through the liquid along a series of zig-zag straight lines. When Zsigmondy first observed this in gold sols by the aid of his newly discovered ultramicroscope, he was so impressed by this difference that he regarded it as something quite different from Brownian movement.^ The earlier investigators established the universality of this phenomenon, when a highly disperse phase is ' Phil. Mag., 1826, 4, 101 ; 1829, 6, 161 ; Edin. New Phil. Jour., 1828,5,358; 1830,8,41. ' Zsigmondy, Zur Erkenntnis d. Kolloide, p. 107. 48 GENERAL PROPERTIES OF COLLOIDS suspended in a medium which permits of the motion, i.e. whenever a solid, liquid, or gas is highly dispersed in a liquid of not too great viscosity, or when a solid or liquid is dispersed in a gas. The viscosity of the medium has a marked influence, as has also the temperature, increase of temperature increasing the path. In spontaneous suspen- sion the Brownian movement overcomes the action of gravity, since a very fine powder covered with water becomes uniformly distributed through the water in course of time. Small quantities of alum, lime, acids caused the motion to cease in opalescent liquids, the particles clump- ing together.^ Many attempts at explanation were made, usually directed to fixing the cause of the motion on external influences, such as local heating, or absorption of light with one-sided heating, thus inducing convection cur- rents, or impurities causing irregular changes in the surface tension between the particles and the liquid, Wiener,^ Gouy,' and others succeeded in proving that the move- ment could not be ascribed to external influences. Its persistence in a preparation for years without apparent diminution is irreconcilable with the assumption of acci- dental or extraneous causes, as is also the reproducibility of the phenomenon, even quantitatively. Wiener supposed it to be a consequence of the kinetic nature of heat, or, in other words, to be caused by the impacts of the smallest particles or molecules of liquids, which are in constant motion, on the microscopic grains. Similar explanations based on the kinetic theory have been put forward by Gouy,^ Eamsay,* and others. ' Sohulze, Pogg. Ann., 1867, [ii.] 129, 366. ' Pogg. Ann., 1863, [ii] 118, 79. ' Physique, 1888, 7, 561 ; Compt. Bend., 1889, 109, 102. « Chem. News, 1892, 66, 90. BROWNIAN MOVEMENT 49 As seen in the ultramicroscope, the motion of sub- microns is oscillatory and translatory. The two character- istics of the Brownian movement are (1) the amplitude A, and (2) the period of oscillation r. The amplitude is the distance between the extreme and mean positions of the particle, and the oscillation period is the time taken by the particle to make a complete oscillation, i.e. to travel from the one extreme position to the other, and back again. The velocity v is then given by ^4A The amplitude depends in the first instance on the size. This is shown by the following figures given by Zsigmondy ^ for various gold sols. Diameter (in /lyn). Amplitude (in n), 6 >10 10 3-4 (also 20) . 35 1-7 The relation between velocity and size of particle had been the subject of earlier inquiry. Exner^ found for a gamboge suspension that, when the diameter varied from 0*4 jLt to 1*3 [I, the velocity varied from 3*8 jn to 2*7 /* per second; with a diameter of 3 /* the motion was barely perceptible, and totally ceased when it was 4 /*.' The velocities observed by Zsigmondy and Svedberg in metal sols are much greater, over 100 n per second lor particles between 10 nfi and 50 ixfi. The only other factor of importance is the viscosity of the medium ; neither the nature of the particles nor the other properties of the liquid appear to have much effect. This is fully established by Svedberg's extensive researches ^ ' Zsigmondy, Zur Erkenntnis d. Kolloide, p. 107. 2 Ann. Physik., 1900, [iv.] 2, 843. ' Zeilsch. Elekirocliem., 1906, 12, 853, 909. K 50 GENERAL PROPERTIES OF COLLOIDS on a series of platinum sols in water and organic liquids, prepared by his method (see p. 217). The sizes were between 40 [i/j, and 50 n/j,. An improved method of obser- vation was employed by running the sol at a known slow rate through the observation cell. The particles are then seen as curves of light, the amplitude (A) and wave-length (A) of which can easily be estimated by means of a micro- meter eyepiece. The oscillation period r is given by Xqt/v, V being the volume of sol passing across the area under observation, q, in the time t. As stated before, the mean velocity is 4A/r. The following table shows the variation in the amplitude with the viscosity tj of the medium. Absolute Amplitude. Medium. Acetone . . . Ethyl acetate . Amyl acetate . Water. . . . w-Propyl alcohol Isobutyl alcohol Isoamyl alcohol Glycerol . . . The approximate constancy of Aij shows that the amplitude for any fixed size of particles is inversely propor- tional to the viscosity of the dispersion medium. The oscillation period, calculated as explained above, increases as the amplitude increases ; thus 4A/r is a con- stant, and is independent of the nature of the dispersion medium. This is proved by the following figures of Svedberg : Amplitude, viscosity. A in ^. 17 X 10'. At, X 10» 3-1 3-2 9-9 2-0 4-6 9-2 1-0 5-9 8-9 1-1 10-2 11 0-7 22-6 16 0-6 39-3 24 very small 43'4 — 830 — Medium. A (in m). T (in sec). 4A/t. Acetone . . . 31 0-032 390 Ethyl acetate . . 2-0 0-028 290 Amyl acetate . . 1-5 0-026 230 Water .... 1-1 0-013 340 m-Propyl alcohol. 0-7 0-009 310 BROWNIAN MOVEMENT 51 Theory of Brownian Movement. — We have already seen that any explanation based on outside sources of energy is untenable. The source of the energy of the movement must be sought within the system : it can only be seated in the particles or in the liquid medium. As far back as 1892, Eamsay ^ explained it by the impacts of the liquid molecules on the particles, i.e. the kinetic energy of the molecules of liquid is the source of the kinetic energy of the particles. The modern theories of Einstein ^ and of Smoluchowski ' are based upon similar considerations. But if we assume, as there are grounds for doing, that there is continuity from coarse heterogeneity, through" suspensions and suspensoids, to solutions (and to mole- cules), and if molecules, whether in a liquid or gas, are possessed of molecular motion, due to their intrinsic energy, it does not seem inconsistent to regard the Brownian movement of the larger particles of sols and suspensions as due to the same cause. That is, the move- ment of these particles is of the same nature and is due to the same causes as the movement of the liquid or gas molecules of the dispersion medium (which are usually supposed to cause it). The application of the kinetic theory, however, has taken the line previously indicated, that the motion is caused by the impacts of the molecules on the particles. This at once gives the qualitative connexion between size and velocity ; it also follows from it that the velocity of the particles is independent of their chemical nature. The objection that the molecular impacts wiU be so numerous, and therefore so evenly distributed that they 1 Chem. News, 1892, 65, 30. ' Einstein, Zeitsch. physikal. Chem., 1907, 59, 451 ; Ann. Physik., 1905, [iv.] 17, 549 ; 1906, 19, 371 ; Zeitsch. Elektrochem., 1907, 13, 41. = Smoluohowski, Ann. Physik., 1906, [iv.] 21, 756. 52 GENERAL PROPERTIES OF COLLOIDS will neutralise eaoli other and no motion result, is not valid, for the probability that at any given instant there will be a given excess of impacts in one direction has been shown to be very great. The objection would apply to the system over a long period of time, but the time of observation cannot be so regarded. It is readily seen, too, that when the particles are very small, as in some of the gold sols, uni-directional impacts will rapidly increase, and the paths will be longer, will indeed become very similar to the paths of the molecules themselves. Thus, as Zsigmondy observed, the oscillatory movements give place to a rapid sequence of zig-zag straight lines. The equation which Smoluchowsti deduced is — 4A2: ^^^ ^T 27 2 SottjN in which A and t are amplitude and oscillation period, tj the viscosity of the medium, r the radius of the particles, and N" the number of molecules in 1 gram molecule, E and T are the gas constant and absolute temperature. Einstein, by different methods, arrived at the same formula, without the numerical factor 64/27. The com- parison of the amplitudes observed by Svedberg with those calculated from the two formulae is given below. Amplitude (in /x). A. A, A, Aa observed calculated Medium Svedberg Smolucbowski Einstein Acetone. . , . 31 11 0-71 2-8 Ethyl acetate . . 2-0 0-68 .0-44 3-0 Amyl acetate . . 1-5 0-58 0-38 2-6 Water . . . . 11 0-31 0-20 3-6 Propyl alcohol , 0-7 017 Oil 41 BROWNIAN MOVEMENT 53 As the last column shows, the observed values are all about three times as large as the calculated ones, i.e. the formula is correct except for a numerical factor. The same, of course, is true of the other formula, as the two differ only by a numerical factor. CHAPTEE VII THE SIZE OF PAETICLES AND ULTRAFILTRATION Size of Particles. — Neither the form nor the size, nor even the colour, of the particles can be directly ascertained by means of the ultramicroscope, for what is observed is a disc of light. As already stated, Bredig ^ fixed the upper limit at 0'14 /j,, which is the limit of visibility with the best microscope (mag. 2250). With this magnification he was unable to detect individual particles in his gold sols. The lower limit is fixed by molecular dimensions, which according to the kinetic theory is 00016 /x (for hy- drogen). Various estimates of the size, based on the effect of the particles on the passage of light through the sol, have been made, but since the advent of the ultramicroscope, they are of secondary interest. The Siedentopf and Zsigmondy method ^ of determining the size is as follows : — Chemical analysis of the sol gives the mass m of the disperse phase in unit volume of the sol ; the number of particles n in unit volume of the sol is obtained by a direct count of the number of particles in the illuminated volume of the sol in the ultramicroscope. This volume is fixed by the depth of the illuminating beam, and by the area of the field (micrometer squares) in ' Anorg. Fermente, p. 20. ' Zsigmondy, Zur Erhenntnis c?, Kolloide, p. 93. THE SIZE OF PARTICLES 55 which the particles are counted. The volume v of the particle is given by — m np p being the density of the particle (the density of the substance in ordinary masses is taken). The linear dimension is calculated from the volume, on the assump- tion that the particle is a cube or a sphere. The dispersity of a sol is usually expressed in terms of the diameter, obtained in this way. It is better to use the specific surface for this purpose (see p. 223). If the Brownian movement is rapid, it is easier to estimate the average distance between the particles than to count their number. The number can easily be calculated, from this distance. Observations on many sols lead to the conclusion that the size is not even approximately constant for a particular substance ; it ranges between 6 /i/x and 250 /x/^j, according to the method of preparation or subsequent treatment. Even larger particles are found along with submicrons, and frequently there are a few submicrons, and a large number of amicrons, which are less than 6 [1(1. That these are still sols, and not solutions, is inferred from the fact that they can be " developed." ^ Thus a mixture of gold salt and reducing agent is more rapidly reduced if an amicron gold sol is added to it. Addition of such a gold sol will even accelerate the reduction of an alkaline silver solution by formalin. By repeated application of this process, in which the amicrons increase in size owing to the further deposition of gold on them, Zsigmondy succeeded in preparing a series of gold sols with particles of any desired size.. ' Zsigmondy, Zeitsch. physikal, Chem. 1906, 56, 63 ; Lottermoser, ibid., 77, j S6 GENERAL PROPERTIES OF COLLOIDS From their ultramicroscopic character, emulsoids seem to consist mainly of amicrons, unless, as is not altogether improbable, the disperse phase is not present in isolated particles, but in a more or less continuous net-work ; or, owing to the smaller differences in refractive index of the disperse phase and the medium, it may be impossible to detect emulsoid particles so easily as suspensoids. On the other hand, emulsoids exhibit more of the general properties of solutions, which is consistent with the smaller size of the particles in emulsoid sols. An interesting resume of the methods of estimating the size of particles in a sol is given by Henri.* The ultramicroscopic method is very uncertain, for frequently the chemical nature of the particles is not known : a certain amount of a metal may be in form of oxide, and may even be in solution, or it may be hydrated, any one of which renders the calculation erroneous. The methods which connect the size of the particle with other properties of the sol are — (1) The determination of the density of dispersion at different heights (Perrin) — , Too , .A.rMO" tiq, n = number of particles in equ9,l volumes at heights and h. p, pi = density of dispersion medium and disperse phase. r = radius of the particles. (2) The relation between Brownian movement and size of particle (Einstein) — I being the displacement in time t. i KoU. Zeitsch., 1913, 12, 240. THE SIZE OF PARTICLES 57 (3) The velocity of sedimentation — 9 V 2 2(/)-yOi)^ V being the velocity. Eor gold hydrosols the equation becomes r^ = 2'5 X 10~^^. It is only applicable when the dispersity is low, for if r = 10 [i/j. the time to fall O'l mm. is 7 hours. (4) From diffusion (Sutherland) — - KT 1 + a" O = • -7; n VTTTjr A is a constant = 0'815, and a the mean path of a molecule of the dispersion medium (a/r vanishes if the radius is great in comparison with the mean path) ; at ordinary temperature the formula is — S = 1-58 X 10-4(cm2, 24 hours). It is only applicable to highly disperse sols, e.g. if 7" = 1 [ifj,, 8 = 0*158 (cm.2, 24 hours), while for urea it is 0-S7 (cm.2, day). (5) Intensity of scattered light (Eayleigh) — c is the intensity of the scattered light. A is a factor depending on the refractive indices of the liquid and the disperse phase, on the intensity of the incident rays, and the angle at which the scattered light is observed. n is the number of particles. A is the wave-length of the scattered light. (6) There is a close connexion between the size of particles and the light absorption by the sol (Garnett, S8 GENERAL PROPERTIES OF COLLOIDS Wood, Svedberg), but so far it is only of use in deter- mining cbanges in size. Ultrafiltration. — Most sols can be filtered through filter paper without appreciable change. Small quantities may be adsorbed to begin with, but beyond that no further loss is experienced. Emulsoids often filter very slowly, owing to their great viscosity. Linder and Picton ^ tried the filtration of arsenious sulphide sols through porous earthenware, and found that the size of particles varied in different samples. They recognised four kinds of sols : a, visible in the microscope ; J3, showed Tyndall Effect ; y, retained by a porous plate ; and S, filtered unchanged through a porous plate. Barus ^ made an estimate of the size of the particles in a silver sol by filtration through porous plates. He tried plates of gradated porosity, and calculated the size of pore of the plate which just permitted filtration to take place. He arrived at a size of 36 fi[i for one silver sol. More recently, Beehhold ^ has elaborated the method, and has applied it in several useful directions. He made a graduated series of filters by impregnating a support of filter paper, fabric, or wire gauze, with gelatinous colloids of varying concentration. Collodion in acetic acid, and gelatine in water, are the most suitable substances ; the filter is subsequently treated chemically : the collodion dipped into water to render it porous ; the gelatine treated with formaldehyde to harden it (see also p. 27). The filters are standardised with a haemoglobin sol. These filters can be used to separate the colloid from the dispersion medium, as in arsenious sulphide sol or > Chem. Soe. Journ., 1892, 61, 148. 2 Sill. Jour., 1896, 48, 51. ' Vortr. 78 Vers. d. Naturf. u. Aemte, 1906 ; Vortr. 14 Vers. Bunsdn- Oes., 1907 ; Zeitsch. physikal Chem., 1907, 60, 257 ; 1908, 64, 328. ULTRAFILTRATION 59 ferric hydroxide sol, or to concentrate albumin and similar emulsoids, the filtrate containing no emulsoid. They can also be employed to free sols from soluble filtrable impurities, the process being similar to purification by dialysis. Another important application of ultrafiltra- tion is in the preparation of graduated series of sols, the sizes being regulated by filtration through a series of graduated filters. The approximate size of particles in sols can also be ascertained by finding which of a series of filters will just allow the particles to pass into the filtrate. The sizes obtained in this way agree on the whole with the ultra- microscope results, Bechhold gives the following series in the order of diminishing size : — Prussian blue, Bredig's platinum, ferric hydroxide, casein, arsenious sulphide, gold (Zsigmondy's 40 /tju,), gelatine, bismuth hydroxide, gold (Zsigmondy's amicrons 1-4 fifi), haemoglobin, Paal's silver, serum albumin, [diph- theria toxin], coUargol, haematin, protalbumoses, silicic acid, deuteroalbumoses, dextrin. It must be remembered that this process of ultra- filtration is a complex one, as ordinary filtration, dialysis, and adsorption can all play a part in it. Even with filter paper, and still more with porous plate, adsorption and precipitation may and do take place. This is especially the case with positive hydrosols. They cannot be filtered even through filter paper without partial retention of the colloid by the filter. The reason is that the paper becomes negatively charged in contact with water (p. 68), and at the entrance to each pore a corresponding quantity of positive colloid is precipitated (p. 132). Bredig's ^ experiments on his negative gold sol may be cited in illustration of what usually takes place. He 1 Anorg, Ferrmnte, p. 27. 6o GENERAL PROPERTIES OF COLLOIDS filtered the sol through a Pukall cell ; the first filtrate was colourless, owing to adsorption or precipitation of the gold by the cell, after which the sol passed through the filter unchanged. The pores of the filter at last became stopped up with the larger particles of gold, and the rest of the filtrate was colourless. For details of the precautions necessary in the preparation and accurate use of ultrafilters, the original papers of Bechhold should be consulted. CHAPTEE VIII ELECTRICAL PROPERTIES In no other branch of the subject is it more necessary to keep in mind the two points of view, viz. the facts, and the theories which have been advanced to explain the facts. As regards the former, there is no longer any serious question, since the main phenomena have been established in a satisfactory manner. When we come to the origin of the electrical charge of colloids we are con- fronted by one of the most difficult problems, a complete solution of which is still awaited, and which, when found, will prove an invaluable aid towards the elucidation of the whole subject. The phenomena will be described first, and then the principal hypotheses which have been advanced will be discussed in detail. The electric conductivity of sols will be taken first, and then migration in an electric field. Electric Conductivity Suspensoids. — The electric conductivity of suspensoids is very small, so little removed from that of the dispersion medium that the question arises whether this small con- ductivity is not due to the small quantities of (adsorbed) electrolytes, which, as has already been pointed out, are impossible to remove, at least by washing or dialysis. 62 GENERAL PROPERTIES OF COLLOIDS The following data * will give an idea of the electric conductivity of several typical sols. Specific conductivity (k) in mhos. Sol. Sol. Medium. Au (Bredig) . . 1-14 X IQ-s to 1-6 „ 0-81x10-6 Au (Zsigmondy) 7"9 „ 1'2 „ Pt (Bredig) . . 2-9 „ 11 Au (Blake) . . 12-7 „ 1-5 Fe(0H)3 . . . 35-7 „ 2 Various attempts have been made to decide by experi- ment how far this conductivity is due to the sol, and how far to the " active " electrolyte. Malfitano ^ determined the electric conductivity of sols of ferric hydroxide from hydrolysed ferric chloride, and of arsenious sulphide, before and after filtration through a collodion membrane. As the latter retained the suspenspid, and allowed all (?) the rest to filter through, the conductivity of the filtrate should fall off, if it is due to the suspensoid, while, in the other event, it should remain unchanged by this process. No diminution of conductivity was observed in any instance, from which he concluded that the conductivity is due to the dissolved impurities. Applying the same method of collodion filtration, Duclaux^ found the con- ductivity of a ferric hydroxide sol to be diminished by filtration, and regarded the difference as the conductivity of the suspensoid. It amounted to about 200 X 10"^ mho. This method is open to the objection that the collodion filtration may change the concentration of the medium in addition to removing the disperse phase, and that the ' Billifcer, SUztmgsber. Kaiserl. Akad. Wiss. Wien., 1902, IH, 1395 ; Bigelow and Gemberliug, J. Armr. Chem. Soc, 1907, 29, 1576 ; Whitney and Blake, ibid., 1904, 26, 1339. "■ Compt. Bend., 190i, 139, 1221. = Ibid., 1905, 140, 1468. ELECTRIC CONDUCTIVITY 63 small observed differences may as well be put down to this as to the removal of the disperse phase. As electrolysis affords perhaps the most certain means of removing electrolysable impurities from a colloid, with- out any other appreciable alteration, the application of this method should solve the problem. By determining the electric conductivity of the original sol, and after sub- jection of the sol to successive electrolyses, Whitney and Blake ^ found that it diminished after each electrolysis, but did not finally approach zero. Instead, it converged to a definite value which they regarded as the true electric conductivity of the pure sol. The following values were obtained : Gold sol. « X 10'. After dialysis . . . . 13-2 „ 1 electrolysis . 7-7 „ 2 „ . 4-2 „ 3 . 2-7 „ 4 „ . 2-1 „ 5 . 1-8 [6 ,. . 1-6] ■7 „ . . 1-4 8 „ . . 1-3; The conductivity of the water was IS X 10"^. In the actual experiment, the sol was contained in a glass tube closed at both ends by gold-beater's skin. The lower end dipped into a glass dish ; the electrodes were applied to the damp membranes, a wider glass tube being fixed by a rubber band to the upper end (Fig. 4, p. 77). The gold migrated to the anode (lower end) and was deposited on the membrane. On treatment with pure water, the gold again diffused into the liquid with regene- ration of the sol. The conductivity of the water in the outer vessel rose considerably. The above figures 'would seem to indicate that the i Loc. cit,, p. 62. 64 GENERAL PROPERTIES OF COLLOIDS pure sol would have the same electric conductivity as the water, since two or three further repetitions of the treat- ment should reduce the conductivity of the sol to that of the water. The figures in square brackets are values extrapolated from the curve plotted from the experimental values in the above table. Kohlschutter's ^ silver sol (see Preparations, p. 183), had an initial conductivity of 129 X 10"^ mho. By puri- fication with hydrogen gas in a platinum vessel the silver hydroxide present (no other electrolyte can be present) was reduced to silver, and the conductivity fell to one-tenth of its initial value, and then remained constant at about three times the specific conductivity of the water, viz. 7 X 10"^ mho. In a few instances it fell as low as 4 to 5 X 10 "6, and in some others could not be reduced below 10 X 10-« mho. In several silver sols prepared according to this method by the author, the conductivity fell as low as 7 X 10~^ in Jena glass flasks, without any treatment in platinum. The initial conductivity of the saturated silver hydroxide solution was about 38 X 10 ~6. Emulsoids. — The figure already given for ferric hydroxide sol may be taken as an indication that emul- soids have often a higher electric conductivity than suspensoids. Although Pauli^ found that albumin sol, carefully purified by dialysis, was practically a non-conductor, Whitney and Blake ^ obtained fairly high values for silicic acid sol, k for a 1"47 per cent, sol being 100 X 10~^ mho, while a 0'2 per cent, gelatine sol gave K = 68 X 10~^ mho. These are of the same order as 1 Zeitsch. EUktrochem., 1908, 14, 49. 2 Beitr. Chem. Phys. Path., 1906, 7, 531. 3 Amer. Chem. Soc, 1904, 26, 1374. ELECTRICAL MIGRATION 6$ the 200 X 10~® mho. observed by Malfitano for ferric hydroxide sol. It would appear from these results that there may be two classes of emulsoids, so far as electrical conductivity is concerned : some, comprising the electrically neutral sols, as albumin, have no electrical conductivity when they are pure, the variable value usually found being due to the impurities; while others, the electrically charged sols, possess a considerable conductivity. If the albumins are amphoteric substances, their vanishingly small electric conductivity is readily understood, for the conductivity of the simple amino-aoids (e.g. glycocoU) is extraordinarily small in comparison with that of the parent acid. From a theoretical point of view, a sol in which the disperse phase is electrically charged, whether positively or negatively, and the greater number of sols come into this category, must exhibit in some degree the phenomena of electric conductivity. Electrical Migration The migration of colloids in an electric field forms a special part of the general electrical phenomena, which occur at a boundary or interface between two phases, one of which at least is a liquid. Quite possibly the same phenomena occur at every interface, whether the phases are gas, liquid, or solid. The experimental difficulties are very great unless one phase is a liquid, and we shall only consider such cases. At all such interfaces an electrical potential difference is established, of the nature of an electrical double layer. Consequently, displacement of one phase against the other takes place when the system is placed in an electric field. We have, therefore, two separate cases. In the one, the 66 GENERAL PROPERTIES OF COLLOIDS solid is -fixed, as in a capillary tube or a porous diaphragm, and all the displacement is thrown on the liquid, which is shown by a difference of level, or by a flow of liquid across the diaphragm. In the other, the solid is suspended in the liquid, and as no permanent difference of level can persist in the liquid, the solid moves through the liquid. The terms " electroendosmosis " and " kataphoresis," frequently applied indiscriminately to both of these phenomena, may conveniently be employed to distinguish between them. Electroendosmosis Electroendosmosis, or the movement of a liquid across a diaphragm or through a capillary tube towards one of the electrodes when a current of electricity is passing, was noticed by Eeuss ^ long ago, and has since been the object of many investigations, theoretical and practical. When a poor electrolyte is electrolysed with a porous cell, the difference of level on the two sides of the cell is very noticeable. Wiedemann ^ was led by this to inquire into the matter. He found that under comparable conditions the difference of level depended on the applied E.M.F. and also on the viscosity of the solution. The latter fact suggested that the two electrically charged layers which suffer displacement are both liquid, the one adhering immovably to the soM, while the other moves with the rest of the liquid. The experiment can also be so arranged that no differ- ence of level is maintained, the liquid which is forced through the diaphragm being run off by a side tube. This ' For full historical references see Wiedemann, EUhtricitUt, 1893, 1, p. 993. - Wiedemann, Fogg. Arm., 1852, [ii.] 87, 321. ELECTROENDOSMOSIS 67 is a convenient form of the experiment, the volume transferred in unit time being a convenient measure of the effect. The formulae connecting the various factors con- cerned are given here (their deduction will be found at the end of the chapter). The equation for the volume of liquid transferred is oeED V ^ 4:TTr]l in which g- = area of diaphragm (or tube). e = potential difference of double layer. E = E.M.F. at the electrodes. D = dielectric constant of the liquid. rj = viscosity of the liquid. I = distance between electrodes. From it we see that the volume transferied is propor- tional to (1) the area of the diaphragm or the cross- section of the capillary, (2) the potential difference between the liquid and solid, (3) the fall of potential between the electrodes, and (4) the dielectric constant ; it is inversely proportional to the viscosity of the liquid. If, instead of allowing the liquid to escape, the pressure is allowed to rise, the equation is p^2eED The height to which the liquid will rise in tubes of the same material, is proportional to the applied E.M.F. and inversely proportional to the square of the radius of the tube. Quincke ^ tested these results by measuring the differ- ence of level on the two sides of a single glass capillary, when a known E.M.F. was applied. The equation shows 1 Pogg. Ann., 1861, [ii.] 113, 513 ; Teresohin, Wied. Ann., 1887, [iii.] 32, 333. 68 GENERAL PROPERTIES OF COLLOIDS that e, the potential difference between glass and water, should be a constant. This is the case. r (cm.). E (volt). h (cm.). c (volt). 0-D449 76 0-0125 0-054 >j 152 00245 0-053 0-0188 154 0-0880 0-055 Other investigators have found e to be constant, and of approximately the same value as Quincke obtained, e.g. — r (cm.). E (volt). h (cm.). e (volt). 0-369 5,120 1-9 0-041 V,220 3-1 0-048 8,860 3-8 0-048 10,200 4-3 0-046 11,500 4-8 0-046 Since most substances become negatively charged in contact with water, the water travels to the hatliode. This has been observed with capillary tubes of glass and shellac, and with diaphragms of porous earthenware, asbestos, carborundum, wool, cotton wool, arsenious sul- phide, and many other substances. A small number have been found to be positively charged in water. They comprise basic oxides and hydroxides, e.g. aluminium oxide and cobalt oxide, etc. ; and some salts, as anhydrous chromic chloride, barium carbonate, etc. Effect of Acids and Alkalies. — The potential difference between the two phases may be changed by the addition of various substances to the water ; and this change may be in either direction. The potential difference may increase or it may diminish, and in the latter case may be reversed in sign, passing through zero on the way. Perrin ^ has investigated this very fully, and some of his results are given below. The effect of acids and alkalies is very marked. The general effect is as follows. » J. cHm.phys., 1904, 2, 601. ELECTROENDOSMOSIS 69 Nlegative diaphragms become more negative in alka- line solutions ; in acid solutions the diaphragm becomes less negative, and with increasing concentration of acid it becomes electrically neutral and finally is positive. Positive diaphragms behave in the corresponding manner, i.e. they become less positive, and finally negative on the addition of alkalies, ■while they become more strongly positive in acid solutions. In Perrin's experiments the variation or reversal of sign of the charge on the diaphragm was indicated by the volumes of liquid transported to the kathode or anode in unit time. The + or — indicates uniformly the sign of the charge on the solid (also migration of the liquid to the anode and kathode respectively). Diaphragm. Solution. V (c.o./min.), Carborundum 0-02 m. HCl + 10 0-008 0-002 -15 water -50 0-0002 m. KOII -60 002 —105 Similar effects are observed with other diaphragms. It is worthy of note that the effect of acid and alkali is greatest at very small concentrations, rapidly falling off as the concentration increases, until practically a constant value is reached. The following figures for a naphthalene diaphragm show this very well. Diaphragm. Solution. V (c.e./min.). Naphthalene 0-02 m. HCl +38 0-01 +39 0-001 +28 0-0002 + 3 0-0002 m. KOH -29 0-001 —60 0-02 -60 70 GENERAL PROPERTIES OF COLLOIDS If the volumes are plotted (with due regard to sign) against the concentrations of acid and alkali (H' and OH'), the two tranches of the curve are symmetrical with respect to the point of electrical neutrality, which is very slightly displaced to the acid side, corresponding to the fact that the substances are negative in pure water. +4-0r +20 S %-zo -M -60- -002n(0H') OOln, Alkali. Concentrattoiv. Fig. 3. OOlrv 00Z7l(H') Acid, With positive diaphragms there would be a similar dis- placement to the alkali side. Effect of Salts. — The influence of neutral salts has also been investigated by Perrin. For this purpose he used solutions which were already acid or alkaline owing to added acid or alkali. The effect of salts is very important in the case of kataphoresis, where, however, it is frequently complicated by other unavoidable effects, which mask or ELECTROENDOSMOSIS ^\ even obliterate the primary effect. Some details will therefore be given in the corresponding case of electro- endosmosis. Negative Diaphragms. Carborundum in Alkali. Salt. V (o.c./min.). -60 01 m. NaBr —14 0-002 m. Ba(]Sr03)2 . . . —26 0002 m. La(N03)3 . . —18 Chromic Chloride in Alkali. -76 0-001 m. MgCJa .... -10 0-001 m. MgSO^ .... — 6 Positive Diaphragms. Chromic Chloride in Acid. +90 0-1 m. KBr +33 0-001 m. MgSO^ .... +23 0-0005 m. K3Fe(CN)6 . . + 3 The concentration effect is similar to that already noted with respect to acids and alkalies : very small concentrations have a very large effect, which is only slightly increased by subsequent increase of concentration This is shown below. Carborundum in Alkali (negative). Concentration of La(N03)3(mols. per litre). V (c.c. per min.). . -60 0-00004 —58 0-0002 —18 0-001 — 1 Further addition of the salt must finally reduce the ^2 GENERAL PROPJERTIES OF COLLOIDS electroendosmosis to zero, or reverse its direction. The latter actually occurs, just as addition of acid to a negative diaphragm, and of alkali to a positive one, reduces the potential difference to zero, and iinally reverses the sign. The reversal is not so pronounced, and the reversed potential difference cannot be made so large as can be produced by acids and salts. The Valency Rule. — The foregoing figures lead to a highly important generalisation regarding the influence of salts on the potential difference between the solid and the liquid. Considering the figures for the negative diafhragm, we find that the valency of the kation is. the predominant factor. The concentration of the bivalent kation Ba" which produces a given decrease in V, is very much less than that of the univalent kation Na", and the concentration of the trivalent La"" is much less again. The figures for magnesium chloride and sulphate show that the valency of the anion does not matter, there being little difference between the univalent CI' and the bivalent SO/'. With the ^positive diaphragm^ on the other hand, the valency of the anion is the principal factor, the effect of the salt on the electrical charge of the diaphragm or on the volume of liquid increasing with increasing valency of the anion. This may be summarised as follows : the electric charge on the diaphragm is reduced by the ion of opposite sign; further, the magnitude of the effect increases with the valency of that ion, and is independent of the valency of the other ion. ' This must be regarded as a broad generalisation, and it will subsequently be shown to hold in the case of kata- phoresis, and also of the precipitation of suspensoids by electrolytes. It must be at once admitted that the matter is not so simple as this ; ions of the same electric sign as the diaphragm do not have the same effect. There are ELECTROENDOSMOSIS 73 differences, again, between ions of the same valency, which cannot be explained by differences in ionisation. H" and OH' in particular differ from other univalent ions. Instances of these will be found under kataphoresis, and more especially under precipitation of sols by electrolytes. That the valency of the ion of opposite electric sign is the main factor in the alteration of potential on the diaphragm is plainly seen from the symmetrical nature of the figures obtained, when the results for + and — diaphragms are reduced to a common standard. This has been done by Freundlich,! who calculated the concentra- tion of various salts which reduced V to one-half of the value when no salt was present. Carborundum (uefative). Salt. Concentration (millimols per litre). Na-Br 50 Ba"(N03)2 2 La-(N03)3 0-1 Chromic Chloride (positive). KBr' 60 MgS04" 1 K3Fe(GN)6"' .... 0-1 Other Liquids. — Quincke examined other liquids besides water; turpentine is negative to glass and carbon disulphide is positive to it. There was no migration of ether or petroleum. Methyl alcohol, ethyl alcohol and acetone all showed marked electroendosmosis. Perrin^ extended the list very considerably. There was marked migration of methyl alcohol, ethyl alcohol, acetone, acetylacetone, and nitrobenzene ; to a less degree of amyl alcohol and acetic acid ; none of chloroform, ether, benzene, carbon disulphide, petroleum, and oil of turpentine. 1 Preundlioh, Kapillarchemie, p. 238. « J. chim.phys., 1904, 2, 601. 74 GENERAL PROPERTIES OF COLLOIDS The same general relations hold for these liquids as for water. The quantitative results ^ for methyl alcohol fully bear this out. Substances, such as chromic chloride, manganese dioxide, which are positive in water, are also positive in alcohol ; the others are negative in alcohol, as in water. The addition of non-electrolytes does not affect the migration,but the addition of acids and alkalies has the same effect on positive and negative diaphragms, as in water. The valency rule applies equally in alcohol and in water. Emulsoids. — Very little seems to be known regarding electroendosmosis of liquids in contact with emulsoid gels. Isolated observations have been made on the behaviour of semi-permeable membranes of aluminium hydroxide, ferric hydroxide and chromic hydroxide, all of which are probably emulsoid gels. Thusj in electrolysis through a iilm of aluminium hydroxide .there was a very marked migration of liquid into the anode compartment ^ ; the water was therefore negative, as was to be expected, since aluminium hydroxide is a positive sol, and wanders to the kathode. In a later research on the same subject, electro- endosmosis took place across a diaphragm of filter paper soaked in gelatine ; the direction was not recorded. Kataphoresis Suspensoids. — From the relation between electroendos- mosis and kataphoresis, we should expect the general conclusions regarding the former, which have been dis- cussed in the preceding section, to hold equally for kata- phoresis. Except in so far as disturbing factors interfere with the observations, this is found to be true. Thus suspended particles of shellac, clay, cotton wool, starch, 1 Baudouin, Compt. Bend., 1904, 138, 898. = Taylor and Inglis, Phil. Mag., 1903, [vi.] 5, 301; Taylor and Millar, Proc. Boy. Soc. Edin., 1906, 26, 447. KATAPHORESIS 75 sulphur, etc., in water migrate to the anode. Most sus- pensoid sols do the same, e.g. the metals, as gold, silver, platinum; the sulphides, as arsenious sulphide, zinc sulphide, etc. ; while the hydroxides of the metals, as aluminium hydroxide and ferric hydroxide, aud some other substances, migrate to the kathode. This is in agreement with the electroendosmosis results recorded in the previous section. This migration of sols in an electric field was first observed by Linder and Picton i in arsenious sulphide sol, which migrated to the anode, and in ferric hydroxide sol, which migrated to the kathode. The behaviour of many sols has been investigated since. The following list, though in no way complete, may be useful : — Positive Sols (to kathode). Ferric hydroxide Titanic acid Aluminium jj Methyl violet Chromic j> Methylene blue Cadmium 3J Magdala red Zinc » Some metals, Pb, Bi, Fe Thorium )> (almost certainly hydr- Zirconium » oxide sols) Cerium JJ Negative Sols (to anode). Metals, e.g. Au, Ag, Pt, Pd, Ir, Cd, etc. Most metallic compounds (except oxides, etc.), e.g. AgCl, Agl, AsgSg, PbS, CdS, Prussian blue, etc. Sulphur, selenium, tellurium. Silicic acid, stannic acid. Vanadium pentoxide. Molybdenum blue, tungsten blue, mastic, gamboge, fuchsin, eosin, indigo, aniline blue, methyl aniline green. 1 Ckem. Soc. Journ., 1892, 61, 148. ;6 GENERAL PROPERTIES OF COLLOIDS The velocity of migration of particles suspended in a liquid, under a fall of potential, is given by the equation 4777; where u = velocity, c = the potential difference of the double layer, JI = the fall of potential (volt/cm.), D = dielectric constant, and ij = the viscosity of the liquid. The derivation of this formula is given at the end of the chapter. Inserting Quincke's value of e for glass and water (0'05 V), and the values of the other symbols, Smoluchowski 1 calculated u for a fall of potential of 1 v/cm. to be 34 X 10"^ cm. per second. 0-05 X 1 X 81 o^ ^ in-fi / " = 300 X 300 X 4,. X 0-0105 = 34 X 10 « cm./sec. The two factors 300 in the denominator are to convert e and H into absolute units. The velocity can be determined macroscopically, microscopically, or ultramicroscopically. In the first method, a vertical tube is iilled with the sol (the ends may be closed with gold-beater's skin), and an electrode applied at each end. The direction of the current is such that the sol migrates to the under electrode, by which a sharp upper boundary is maintained. The distance travelled in a known time is read off on a scale (Fig. 4). In the other methods,^ the actual migration of a single particle is observed under the microscope, and the distance travelled by it in a known time is measured by means of a micrometer eyepiece. The ultramicroscope must be used if the particles are submicrons. Certain precautions 1 Bull. Acad. Scien. Cracow, 1903, 182. 2 Cotton and Mouton, J. cJiim.phys., 1906, 4, 363 ; Coward, Trans. Faraday Soc, 1913, 9, 142. KATAPHORESIS 11 i^~ are necessary if correct results are to be obtained. The particles selected for observation must be near the middle of the liquid, for particles near the cell walls may even travel in the reverse direction. The reason is that electro- endosmosis occurs at the cell wall, and the liquid moving in the contrary direction to the sol (the cell wall, of glass or quartz, usually has the same sign as the sol) drags the particles along with it. Cotton and Mouton have described the diverse nature of the observations when the microscope is focussed on layers at different depths in the cell. Then, the particles must be fairly equidistant from the electrodes, and the observations must be made immediately the current is applied. The disturbances due to neglect of these precautions are twofold. When charged particles reach an electrode of opposite charge, they usually give up their charges to the electrode, and remain precipitated in the vicinity of the electrode. They may, however, take up a charge of the same sign as the electrode, and then set off back to the other electrode. Eeversal of the (Charge may occur in another way. As we have seen in electroendosmosis, very small changes in the concen- tration of electrolytes, and especially of H" or OH', pro- duce large changes in the charge on the solid, and may cause reversal. When a sol is submitted to prolonged kataphoresis, the traces of electrolytes accumulate at the electrodes, at the same time decreasing the H' at the kathode, and increasing it at the anode. If, therefore, the original sol is neutral, the kathode liquid may become FlQ. i. 78 GENERAL PROPERTIES OF COLLOIDS alkaline, and the anode liquid will become acid, with corresponding changes in the charge on the particles. This reversed migration, however caused.must obviously vitiate the results. Still another disturbing process may take place. The particles with reversed charge may come in contact with some of the original particles, and the two electrically neutralise each other, causing mutual precipi- tation. If the experiment is unduly prolonged this effect may cause serious error in the middle of the solution. There is yet another disturbing factor, first indicated by Coward,^ which must not be overlooked, viz. a charged particle may be attracted to an electrode of the same sign as the particle, if the potential difference between them is sufficiently great. For this reason, observations near the electrodes are always liable to error. These sources of error apply also to the macroscopic methods, and their effects were noticed by Linder and Picton. For demonstration purposes, the familiar apparatus of Nernst for demonstrating ionic migration may be used, a sharp boundary between the water and the coloured sol being obtained in the usual manner by loading the sol with a non-electrolyte (sugar or urea). More convenient still is a small form of the Nernst apparatus, which can be mounted in the lantern and projected on the screen. The U tubes are of small thin- walled glass tubing. The filling tube has no tap, but is constricted to a very narrow opening just below its junction with the U tube. Instead of being placed centrally, it is better to bend it to one side, as shown in Fig. 5, but not in the same plane as the U tube. It is easier to work, and is out of the field of the lantern ; the apparatus is clamped by this tube. By means of the 1 Loc. cit., p. 76. KATAPHORESIS 79 rubber tube and small clip the sol can be introduced slowly and to the requisite amount. To use it, the side tube is quite filled with the loaded sol, and the clip is closed. The sol which has escaped into the U tube is washed out with water. Water is then placed in the U tube, and the electrodes (a flat helix of thin platinum wire) inserted. The ap- paratus is then mounted in the lantern. The clip is cautiously- opened, and the sol flows equally into the two limbs with sharp in- terfaces. The clip is closed as soon as the electrodes are covered with water. The amounts of water and sol must be so adjusted that the interfaces are well below the elec- trodes, A suitable voltage is llOV or 220V. With this simple apparatus the direction of migration, the reversal of direction on reversal of the current, and the approximate velocity for a potential gradient of 1 volt/cm. can be easily obtained, as owing to the magni- fication, the migration is visible in a very little time. Some at least of the disturbances caused by prolongation of the experiment, such as the precipitation near the electrode, and the mutual precipitation by the particles of reversed charge, can also be observed. Two such tubes may be shown on the screen at one time, and if a negative sol (gold or arsenious sulphide) is placed in one, and a positive sol (ferric hydroxide) in the other, the migration in opposite directions may be con- vincingly shown, as the electrode connexions can easily be shown on the screen at the same time. 8o GENERAL PROPERTIES OF COLLOIDS Velocity of Migration. — The observed velocity is in good agreement with the value, 34 X 10-^ cm./sec., cal- culated from the formula, as the figures below show.i 11 (om./sec. per Sol. Size of particles. 1 v/om ■)• AS2S3 50 [MfM 22 X 10- 5 (L. P.) Quartz . . 1/^ 30 (W. B.) Au . . . < 100/^/1. 40 (W. B.) „ . . . 3J 21-6 (B.), Pt . . . J> 30 (W. B.) J, ... J) 20-3 (B) Ag . . . Ji 33 (C. M.) if - )t 23-6 (B.) ( 4 different \ '> \ preparations , J) 19-7 » „ ... , J» 19-6 ii ,, ... . )> 19-3 ,, Fe . . . , li 19 J, Fe(0H)3 . . JJ 30 (W. B.) The agreement of these figures and the theoretical value shows that the potential difierence between water and all these different substances is of the same order as that between glass and water. The values of e calculated from the above values of ti, vary from 0"028V. to 0'058V. The metals lead and bismuth give very low values of u and c: they and iron are positive sols, like ferric hy- droxide. The velocity is practically independent of the size and form of the particles, as the formula requires. Most remarkable of all is the fact that the average velocity is practically the same as the migration velocity of the average slow moving ions of electrolytes, Na' being 45 X 10-5 cm./sec, and Li'= 36 X 10-^ cm./sec. for ' Linder and Picton, Chem. Soc. Journ., 1897, 71, 508; Whitney and Blake, J. Amer. Chem. Soc, 1904, 26, 1389 ; Burton, Phil. Mag., 1904, [vi.] 11, 425 ; Cotton and Mouton, loo. cit., p. 76. KATAPHORESIS 8i 1 v/cm. fall of potential. The velocity of migration of suspensoids, whether microns or submicrons, and of ions is thus practically constant. Effect of Electrolytes, — The presence of electrolytes, and especially of acids and alkalies (H* and OH'), has such a marked effect on electroendosmosis that we naturally look for similar effects here. This expectation is justified in the few instances in which it is possible to test it. The difficulty is that another action of electrolytes comes in, which is wanting in electroendosmosis. Suspensoids are by no means so stable as a diaphragm-liquid system ; they are very susceptible to electrolytes, undergoing a series of changes, which usually end in complete precipitation of the sol. As will be shown in the next chapter, these changes are almost certainly connected with the electrical changes, mainly the electrical neutralisation of the charged particles by the ions of the electrolyte, which we are led to infer from the behaviour of diaphragms, but which are difficult to verify by experiment. Burton ^ investigated the influence of the trivalent kation, Al'", on negative gold sol and silver sol. Silver Sol. Concentration of Al2(SOj3 u (cm./sec.) x 10' in millimols per litre. at 18°. — 22-4 0-0052 — 7-2 0-014 + 5-9 0-0284 + 13-8 Whitney and Blake had previously noticed that the addition of H' (nitric acid) or of Ag' (silver nitrate) to silver sol caused the particles to migrate to the kathode instead of to the anode. 1 PUl. Mag., 1903, [vi.] 13, 472. G 82 GENERAL PROPERTIES OF COLLOIDS The Electric Charge on a Particle. — An estimate of the charge on a single particle of gold sols and silver sols has been made by Burton, on the assumption that in the pre- cipitation of these sols by aluminium salts, the amount of Al"" which just precipitates the gold or silver is such that the amount of positive electricity on it is equal to the amount of negative electricity on the particles pre- cipitated. The volume of a particle is 2 X lO-^* c.c, therefore 100 c,c. of a sol (with 6 "5 mg. Ag) contains 3 X 10^" particles. This volume of sol required 3'0 X 10~^ and 2'6 X 10~s A]2(S04)3 for precipitation, from which the charge on a particle is 2'8 X 10-^ electrostatic units, and the charge on 1 gm. equivalent of silver is 4 per cent, of the charge on Ag' (1 gm. equivalent of silver ion). Other liquids. — Quincke observed that most sub- stances were positive in oil of turpentine, sulphur being the only exception. Lycopodium in ether migrates to the kathode ; sulphur, vermilion, antimonious sulphide, to the anode. Billiter ^ found that alcohol reduced the migration in a platinum sol to zero, and then reversed the direction of migration. To Burton 2 we again owe a series of quantitative experiments, which show the general relations quite plainly. The symbols in the table are the same as in the formula on p. 67. 1 Zeitsch. EleUrochem, 1902, 8, 633. ' Phil. Mag., 1904, [vi.] 11, 425. 22 + 0-046 10-2 + 021 4-5 + 0-024 3-6 + 019 2-8 + 0-015 2-3 — 0-054 1-7 — 0-040 KATAPHORESIS 83 Dispersion Medium. Sol. D. ?;. u x 10^ e (volts). Methyl alcohol Pb 33 0-00612 Bi Ethyl alcohol Pb 25-8 0-0123 Sq Zn Ethyl malonate Pt 10-7 0-0223 Ag Au 1-4 — 033 The value of u depends principally on the viscosity of the liquid; e does not differ much from the values in water. The sols of the base metals in the alcohols are positive, as in water. A broad generalisation regarding the sign of the charge on a solid in contact with a liquid, first put forward by Coehn,^ may be stated : the substance with the greater dielectric constant is positive to the other substance. The D.C. of oil of turpentine is 2-23, that of glass is 4-7, according to its composition, and that of water is 81. In agreement with the rule, glass is positive in oil of turpen- tine, and negative in water. Water has a much higher dielectric constant than most other substances, and, as we have seen, most substances are negative in water. Emulsoids. — When an emulsoid is placed in an electric field, migration is observed to take place as in suspen- soids. But the disturbing factors are much more powerful than in the latter. The velocity is also much smaller, u for gelatine being + 25 X 10"^, and for silicic acid — 15 X 10"^ cm./sec. It is partly due to this, that reversal at the electrodes, and mutual precipitation in the body of the liquid are more pronounced. Indeed, in a gelatine sol some of the particles were found to be travel- ling from each of the electrodes at the commencement of the experiment. ' Wicd. Ann., 1898, [iii.] 64, 217. 84 GENERAL PROPERTIES OF COLLOIDS This suggests that the charge on an emulsoid is largely an accidental matter, being determined -by the medium in which it happens to be, and that its normal condition is electrical neutrality. Gelatine, if charged initially, is ready to lose its charge to the electrodes, or, if not, is ready to receive charges from the electrodes or from ions in the vicinity of the electrodes. Confirmation of this view is to be found in the observations on albumin. Coagulated egg-albumin does not migrate in either direction in neutral solutions, but migrates to the anode in alkaline solutions, and to the kathode in acid solutions.^ "We must conclude from this that albumin is electrically neutral in water and in neutral solutions, and that it becomes negative in alkalies (OH'), and positive in acids (H). Pauli's experiments ^ on a specially pure albumin sol still further confirm these conclusions. The albumin was dialysed until no more electrolyte diffused away. It was contained in three vessels connected together, the elec- trodes being in the two outer vessels. The concentration of the sol in the three vessels was determined by Kjeldahl's nitrogen method ; the current was connected for periods up to twenty-four hours. There was no kata- phoresis, from which we must conclude that the albumin is isoelectiiic with water. In solutions with an acid reaction (H') the albumin migrated to the kathode, and was therefore positively charged ; while in alkaline solutions (OH') it migrated to the anode, and was negatively charged. It remained unaffected by neutral salts. It is quite open to explain this behaviour of albumin ' Perrin, Coiv^t. Bend., 1903, 136, 1388; Hardy, J. Physiol, 1904, 24, 288. 2 Beitr. cheni. Physipl. Path., 1906, 7, 531. ELECTRICAL PROPERTIES 85 by regarding it as an amphoteric substance, which as a base forms salts with acids, in which the albumin forms the kation, and as such migrates to the kathode, as all kations do. As an acid, it will form salts with bases, in which it functions as anion, and so travels to the anode. But when we recall Perrin's observations (p. 69) on the effect of H' and OH' oh electroendosmosis with diaphragms of such substances as carborundum and naphthalene, to neither of which can be ascribed acid or basic properties, or amphoteric electrolyte structure, we are forced to the opinion that the possible amphoteric character of albumin is not the essential factor in its kataphoretic behaviour. Eather, that pure albumin, perhaps like most pure emul- soids, is electrically neutral, and takes up + or — charges, especially in solutions containing H' or OH' respectively. (See further, p. 90.) Oil emulsions made by shaking two c.c. of petroleum (density 0-9) with 100 c.c. of water, or by boiling the oil and water together, or by dissolving the oil in alcohol, pouring the solution into water, and finally driving off the alcohol by boiling, were examined by Hatschek.i The dispersity was the same in all three cases, being 4 X 10"^ cm. The oil was negative to the water, and the velocity of kataphoresis was — 43 X 10"^ cm./sec. per 1 v/om. ; from this the value of e is O'OSV, and e = 4'4 X 10-' electrostatic units, values which are practically the same as for the negative metal sols. Theoeetical In this section will be found first the deduction of the equations concerning electroendosmosis and kataphoresis already given in the two preceding sections ; and following ' Koll. Zeitsch., 1911, 9, 159 ; also Lewis, 1909, 4, 211. 86 GENERAL PROPERTIES OF COLLOIDS it, a discussion of the various theories ■which have been suggested to account for the origin of the potential dif- ference at the boundary between solid and liquid. The Volume Equation. — In the steady state resulting when a constant E.M.F. is applied to electrodes in two vessels connected by a capillary tube, no increase of hydrostatic pressure occurring, the E.M.F. must be con- sidered merely to overcome the initial resistance, and thus to impart the constant velocity. The force of friction is proportional to the viscosity, to the area of the moving surfaces, and to the velocity gradient. For the reason already given (p. 66) we may consider both the moving surfaces of the electrical double layer to be liquid, the liquid moving against a film of liquid, which firmly adheres to the solid. The viscosity is thus given by ij, the viscosity of the liquid. The mean ' velocity u is given by the length of the column of liquid which passes out of the tube in unit time, V being the volume of liquid ; then ■nr^u = v (1) in which r is the radius of the capillary. The gradient of velocity -r- can be put = -' (p = the ctp o distance from the solid wall, and 8 the distance of the moving side of the double layer from the wall), since S is a small distance, and the liquid at S moves with the velocity u, and the layer itself has zero velocity. The frictional force for unit surface is thus : J-, or, puttmg in M = ^, ^ The electrical force acting on unit surface of the moving side of the electrical double layer is «H, f- being the charge ELECTRICAL PROPERTIES 87 on this unit surface, and H the potential gradient produced by the external E.M.F. These two forces balance in equilibrium, and we have «H=.^ (^) The double layer may be regarded as a condenser of potential e in a medium whose dielectric constant is D. Then e = D this. with the preceding equation (2), gives cDH 477S from which 1 A _ 4, (3) Substituting j for H, E being the E.M.F. applied to the electrodes, and I their distance apart, and substituting q the cross-section of a diaphragm for m'^ the cross-section of a single capillary tube, we finally have ?€ED ,., ^ = 1^ (^> in which form it is given on p. 67. The Pressure Equation. — If the liquid is not allowed to escape at constant (zero) pressure, the hydrostatic pressure P produced by the migration of a volume v of liquid is given by Poiseuille's law : TrPr* 88 GENERAL PROPERTIES OF COLLOIDS The equilibrium pressure P is obtained by combining this with equation (3) : TrPr* r2eED from which 2r]l 4.7]l ^ 2eED (5) These are the two laws for electroendosmosis. Velocity of Kataphoresis Equation. — The law for the kataphoresis of solid particles in a stationary liquid is readily deduced from the above relations. Imagine the particle to be a small cylinder placed axially in the capillary tube, and to be stationary, while the liquid flows past the cylinder in the annular space between it and the tube. If the annular space be regarded as made up of a series of tubes, the velocity of the liquid is obtained by combining equations (1) and (3) : r2eDH '2^/, :— - ' from which ^=4^ (^> If the cylinder be now regarded as moving and the liquid as stationary, the velocity must still be the same (equality of action and reaction). Theories regarding the Electrical Properties The formulae which have just been deduced, have been shown in the preceding sections to be fully borne out by the experimental results. It is well, however, to state explicitly that they throw no light on the origin of the electrical phenomena described in those sections. The ELECTRICAL PROPERTIES 89 development of the equations is based upon the assump- tion of "an electrical double layer," one layer of which can move, while the other remains stationary. The nature of the double layer, or how it comes into being, or is maintained, are quite immaterial. These are, of course, questions of the greatest im- portance and interest. No final decisive answer can be given at the present time, but several very suggestive indications have been advanced, and are worthy of closer consideration. On the well-established ground that the exceptions or irregularities most often afford the clue, we may first take the positive sols, those basic hydroxides of metals and basic dyes, which, unlike the great majority of hydrosols, migrate to the kathode. This peculiarity has been ex- plained in various ways; e.g. being metallic hydroxides, they can form metallic (positive) ions, by electrolytic dis- sociation. This is true, but they must at the same time have produced an equivalent amount of OH' or of some other negative ion. Further, zinc sulphide (and most other metallic compounds) can also give rise to metallic ions (kations), but they, notwithstanding this, migrate to the anode. Now if aluminium hydroxide, for instance, in pure water is positive, while the water is negative, it would appear that the hydroxide has in some way united with part of the H' of the water, leaving the equivalent excess of OH' to produce the negative charge of the liquid phase. Pure water contains equivalent amounts of H' and OH', though the actual concentrations are very small. From this we would infer that aluminium hydroxide dissolves, or adsorbs, or retains by permeation or diffusion, H' more readily than OH'. There is abundant experimental evidence to support this 90 GENERAL PROPERTIES OF COLLOIDS view. The formation of a semipermeable membrane by reaction between aluminium salts and ammonia is evidence that aluminium hydroxide is impermeable by OH', for otherwise, instead of a thin film, which does not increase in thickness, being formed, the reaction would continue until all the Al"" or OH' was used up. It has also been shown ^ that even a large E.M.F. fails to drive OH' across such a film of aluminium hydroxide. We may thus take it as proved that aluminium hydroxide cannot dissolve OH' to any extent, and that OH' cannot diffuse into it. It has also been demonstrated that a film of aluminium hydroxide is permeable by H', both by simple diffusion and by the application of an E,M. F. It is immaterial whether this per- meability is regarded as a solution effect or as a diffusion. Thus, when aluminium hydroxide is suspended in even pure water, which contains equal concentrations of H" and OH', the H' can dissolve in, or diffuse into it, while the equivalent of OH' will be left behind, and will accumulate in the layer next the particles. We will next consider the case of albumin, as the type of another set of exceptional substances. Highly purified albumin shows no kataphoresis in water or in neutral salt solutions ; it therefore has no electric charge. From the present point of view, this indicates that albumin either cannot dissolve H' or OH', or that both are equally soluble (or diffusible) in it. If the latter view is taken, the concentration of H' and OH' in albumin will vary directly as their concentration in the liquid, i.e. in acids, which contain more H' and less OH' than water, it will take up an excess of H', and will be positive. The positive charge will increase with increase in acidity of the liquid. In alkaline solutions there is more OH' and less H' than in water ; the albumin will become negative, * Taylor and Inglis, Toe. cit., p. 74. ELECTRICAL PROPERTIES 91 and the charge will increase as the alkalinity increases. In each case the effect of increasing acid or alkali will fall off, as a saturation limit may soon be reached. Admittedly, these ideas cannot be carried much further, for practically nothing is known regarding solubility or diffusibility of ions in most substances. Nor does it follow because these' " explanations " iit the facts, that they are correct. There may be many possible explana- tions, only one of which can be right. Indeed, the most probable of the suggestions is that one which substitutes adsorption for diffusion or solution: and this, partly because adsorption is a wider and more universal phenome- non than the others, and partly because it can be observed or assumed to occur in more instances than they. The adsorption theory explains the predominant effect of H' and OH' on the electric charge by stating that these ions are more readily adsorbed than other ions, and that, further, OH' is more readily adsorbed than H" ; from which most sols are negative to water. It is not so clear why OH' is less readily adsorbed than H' by those basic hydroxides which form positive sols, nor why albumin adsorbs both ions equally. Freundlich suggests that the basic substances produce OH' in the adherent layer, and that, owing to its greater diffusion velocity than the other ions present, it diffuses into the moving layer, imparting to it a negative charge, and leaving the particle positively charged ; and similarly with the feebly acid substances and H'. The great influence of acid or alkaline reaction of the medium on the magnitude, and even on the sign of the charge on the sol, is readily seen from this point of view. The fact that the effect of ions is relatively greatest at small concentrations is what would be expected from the relation between adsorption and concentration. It is 92 GENERAL PROPERTIES OF COLLOIDS plain, too, that only adsorption of ions of opposite charge to that on the sol can reduce the charge on the sol, for the adsorption of the other ion would increase the charge. The valency rule also is capable of explanation from the same point of view (see p. 102). Further discussion of these and other theories must be deferred to a later section. CHAPTER IX PRECIPITATION One of the most striking features of certain of the sols prepared by Graham and other of the early investigators of colloids, and one which attracted their attention, is their extraordinary sensitiveness to chemical reagents. Aluminium hydroxide sol, for instance, prepared by Graham's method, gives a dense precipitate when one or two drops of dilute sulphuric acid are added to it ; this precipitate is aluminium hydroxide. The sol is still more sensitive to neutral sulphates ' {e.g. sodium sulphate). Many other sols are equally sensitive. In some cases this process can be reversed, and the sol regenerated by merely washing away the reagent with pure water, but in others this cannot be done, there being apparently no direct method of regaining the sol. Between these extremes lie the majority of sols. In many cases the reversibility is a question of time ; the precipita- tion is reversible at first, but, sooner or later, passes through stages of incomplete reversibility to complete irreversi- bility. The division of precipitations into reversible and irreversible depends on an arbitrary interval of time. Instability, the tendency to undergo change either spontaneously or under the influence of very small amounts of reagents, appeared to be characteristic of colloids, though possessed by different colloids to very different extents. The metal sols, in whatever way 94 GENERAL PROPERTIES OF COLLOIDS prepared, are, as Faraday, Zsigmondy, and Bredig found, probably the most susceptible to changes, which seemed at first sight to be spontaneous, due to inherent instability, but which were later shown to be dependent on the method of preparation, and especially on the nature of other substances formed in the reaction simultaneously with the sol. On the other hand, many sols were extremely stable, and remained unchanged for years, although they were highly sensitive to small amounts of reagents. This susceptibility to reagents is, as a rule, limited to electrolytes: salts, acids, and bases ; indifferent substances, whether inorganic or organic, have usually no appreciable effect, until they are added in large quantity. Sols vary very much in their susceptibility to reagents ; some are readily precipitated while others are not affected by some salts, and require large amounts of other salts to cause precipitation. They have accordingly been classed as hydrophobe and hydrophile sols, or more generally, as the distinction also applies to other dispersion media, lyophobe and lyophile sols. Broadly, this division agrees with that into suspensoids and emulsoids. The propriety of the terms will be evident if the fundamental differences between suspensoids and emulsoids are recalled (p. 18). It is necessary also to bear in mind the present use of the term "gel" (Introduction, p. 5). As the phenomena of precipitation are so varied, they will be dealt with in the following order : — A. The action of electrolytes on suspensoids : (ci) qualitative, (&) quantitative, (c) valency rule, {d) adsorption, {e) stabilising effect of ions. PRECIPITATION 95 B. The " complex " theory of colloids. 0. The action of reagents on emulsoids : (a) albumin, (b) gelatine, (c) silicic acid. D. Protection of susj)ensoids by emulsoids. E. Mutual precipitation of sols. The Action of Eleotkolytes on Suspensoids When successive small quantities of a salt are added to a sol, the sol becomes opalescent, and a flocculent precipitate finally appears, which settles out, and leaves the medium clear and free from the disperse phase. This behaviour, first observed by Selmi, and later by Graham, is well known to every student in the case of arsenious sulphide, which is obtained as a sol by interaction of arsenious acid and hydrogen sulphide solutions; no precipitate is obtained until acid or salts have been added. The changes are readily followed under the microscope ; when opalescence commences, the particles, while visibly increasing in size by coalescence, are in rapid Brownian movement. As the size increases, the movement falls off, and when the particles settle out, they are no longer in motion. The course of these changes can be followed more completely in the ultramicrosoope. Commencing with an amicron sol, the first effect is the production of submicrons, seen as individual discs of light in active Brownian move- ment. On further addition of electrolyte, they decrease in number by uniting with each other to form microns. Along with this the Brownian movement becomes sluggish, and, as the micron stage passes into the macron stage, diminishes to complete cessation; sedimentation then occurs. 96 GENERAL PROPERTIES OF COLLOIDS This cessation has been considered by some to be the cause of precipitation, the primary effect of the electrolyte being on the Brownian movement, slowing it down. It had, indeed, been observed that small concentrations of electrolytes actually do reduce it, and at last cause it to cease. Svedberg showed that the sequence is the reverse of that stated above, i.e. the electrolyte causes the par- ticles to unite, and the Brownian movement becomes slower in consequence of the larger size. There was no change in the amplitude before and after the electrolyte was added, until the size began to increase. The following figures refer to a silver sol, to which increasing amounts of aluminium sulphate were added. It was possible to determine the amplitude before a sensible amount of precipitation had occurred. Even so great an amount of aluminium salt as to reverse the electric charge on the sol had no effect on the amplitude. Kataphoretio velocity Amplitude, (cm./sec. X 10=). in ft. —2-1 1-3 -2-6 1-3 +0-43 1-2 +1-76 1-2 These figures make it quite clear that the alteration in the Brownian movement is due to antecedent changes in the sol. The cause is rather to be sought in the electrical phenomena. Burton noticed in his experiments on the effect of Al"" on the kataphoresis of silver sols, that the latter appeared to be at maximum instability when the kataphoresis was reduced to zero, i.e. at the point of electric neutrality, the isoelectric point. Of the four silver sols in the table on p. 81, the pure sol was stable for indefinite periods, the two following {u = — 7"2 and PRECIPITATION 97 +5"9 X lO"") precipitated in a few hours, and the last (m = -|- 13-8 X 10~5) was more stable, not being completely precipitated in four days. The effect of the electrolyte is first of all to diminish the potential difference between the disperse phase and the medium, and precipitation only occurs when the potential difference has been brought sufficiently near to zero ; smaller amounts of electrolyte have practically no effect on the stability of the sol. Linder and Picton ^ added to a ferric hydroxide sol one- sixth of the amount of NaCl necessary for its complete precipitation ; it remained clear even on boiling. As the sol could no longer be filtered unchanged through a porous plate, its dispersity must have decreased, although not so much as appreciably to diminish its stability. We must conclude, then, that a definite concentration of electrolyte is necessary for precipitation, and that, since smaller concentrations may cause only partial precipita- tion or even none at all, it is not improbable' that larger concentrations may also produce partial or no precipitation. The Precipitation Concentration. — The existence of this limiting concentration was first demonstrated by Bodlander^ in the case of kaolin suspensions. This is well shown by the following table : — Acid concentration, ^^ , . TT a(-> Kaolin. ^^^ X 10-^ g. per litre. 00 7-7 0-12 7-5 0-16 7-7 0-19 7-6 0-25 6-9 0-28 6-1 0-29 4-8 " Chem. Soc. Journ., 1905, 87, 1992. « Giittinger Nachrichten, 1893, 267. H 98 GENERAL PROPERTIES OF COLLOIDS The aame holds for sols ; below this limit of concentration there is no precipitation even after a long interval of time.i An arsenious sulphide sol contained 9*57 milli- mols per litre ; in portions of this sol containing 1"219 and 2"438 millimols of potassium chloride per litre, the concentrations of arsenious sulphide at the end of 340 days were 9'60 and 9'45 millimols respectively. With 3'90 millimols of potassium chloride per litre, the sol was almost completely precipitated in this time. It is a peculiar circumstance that the manner of adding the reagent has a marked effect on precipitation ; when the reagent is added very slowly, or a little at a time, a much larger amount is necessary for complete precipitation. The sol appears to become " acclimatised." For instance, the amount of barium chloride necessary for complete precipitation of an arsenious sulphide sol in two hours was ascertained. When this amount was added in small quantities at a time, precipitation was far from complete in two hours, and a much larger amount was then necessary. The slower the reagent is added, the greater is this excess. This is a general occurrence. The Valency Rule. — -The most striking thing about the precipitation of sols by electrolytes is the enormous differences in the precipitation concentration of different salts. This is equally true of suspensions and of' sols. The following figures, given by Bodlander,^ illustrate this. The concentrations are not precipitation-concentra- tions, but are the concentrations which exactly doubled the rate of sedimentation in the pure suspension. 1 Preundlioh, Zeitsch. physikal. Chem., 1903, 44, 144, ' Jahrb. Mmeral, 1893, 2, 147. PRECIPITATION 99 Kaolin Suspension. Mg. equivalents in Electrolyte. 100 o.o. suspension, HGl 0'0017 HNO3 0-0016 HCOO.CCls 0-0016 H2SO4 0-0020 H2C2O4 0-3700 0-0183 H3PO, NaOH 1-186 Ba(0H)2 0-3700 NaNOg 0-0133 CaClg 0-0029 MgQa 0-0016 Similar results were obtained by both Hardy and Bechhold for mastic sols and a gold sol ' : — Electrolyte. HCl . HNO3 . H2SO4 . H CjOi 2 NaCl . K,S04 . BaClg . MgSO^ KOH . Mastic (Hardy), gm. equiv. per litre. . 0-004 . 0-004 . 0-004 . 0-009 . 0-12 . 24 . 0-022 . 0-028 Mastic (Bechhold). Gold sol (Hardy), gm. equiv. per litre. 0-008 0-008 0-008 0-013 0-026 0-004 0-09 Electrolyte. NaOl . AgNOg . HCl. . H2SO4 . MgSO^ . ZnS04 . mg. equiv. per litre. . 1000 . 125 10 10 . 100 . 100 Electrolyte. BaClj . CaClo . , CdSO^ Al2(S04)3 PeaCSOi)^ A1(N03)3 Fed, . mg. equiv. per litre. . 50 . 50 . 25 . 0-5 . 0-5 . 0-6 . 09 ' Hardy, Zeitsch. physikal. Chem., 1900, 37, 385 ; Bechhold, Zeitsch. physical. Chem., 1904, 48, 385. 100 GENERAL PROPERTIES OF COLLOIDS It is obvious from these results that these electrolytes fall naturally into three groups, the members of each group possessing the same precipitating power. As the same anions are to be found in all three groups, the enormous differences cannot be ascribed to them, but are either due to the unionised molecule or to the kations. The kations in each group are of the same valency ; the univalent kations H', Na', K', etc., in the first group, with the smallest precipitating power ; followed by the bivalent kations Ca", Ba", Mg", Zn", etc., and by the third group, the trivalent kations Al"", Fe'", etc., with the greatest precipitating power. All the sols in the above tables are negative ; it is . clear that the precipitation is mainly effected by the ion of opposite electric charge, and that the precipitation- concentration is a function of the valency of these ions. If this conclusion is correct, the precipitation of posi- tive sols should depend on the valency of the anion. Ex- periments made to test this have fully confirmed the valency rule. Hardy, who first established this rule, foimd the following figures : — Eeeeio Hydeoxide Sol. Concentration gm. equiv. Electrolyte. per litre. HGl 0-5 HNO3 0-5 NaCl 0-5 KOH 0-001 BaCla 0-001 Electrolyte. H2SO4 . HaCaO, . K2SO4 . MgSO^ . H3 citrate. Concentration gm. equiv. per litre. 0-002 0-002 0-0006 0-0005 0-0007 Certain peculiarities in these results require explana- tion, but on the whole they confirm the valency rule, the three groups containing the univalent, bivalent, and trivalent anions respectively. A very extensive series of experiments by Freundlich ^ fully establishes the rule. ' Loc. cit., p. 98. PRECIPITATION lOI Fereic Hydroxide Sol (16 m.mols per litre). Electrolyte. NaCl . , KCl . , BaClg . , Cono. mg. equiv. in 1 litre. . 9-25 . 9-03 . 9-64 KNO, 11-9 KBr . Ba(N03), KI . . 12-5 14-0 16-2 Electrolyte. HCl ; Ba(0H)2 K2SO4 . MgS04 H2SO4 . Cone. mg. equiv. in 1 litre. ca400 . . 0-42 . . 0-204 . . 0-217 . . 0-194 ca 0*5 Platinum Sol (-7 m.atom per litre). Cone, (m.mol Electrolyte. per litre). NaOH 130 NaCl 2-5 KCl 2-2 AgNOj 0-22 Electrolyte. BaCl, . U02(N03)2 Pb(N03)2 . iAl2(SO,)3 Cone, (m.mol per litre). . . 0-058 . . 0-065 . . 0-011 . . 0-013 Arsenious Sulphide (7*54 m.mols per Utre). Electrolyte. Univalent Kations- ^K3 citrate kC2H302 . KCH02 4K2S04 KCl . KNOg . Nad . LiCl . HCl . iH2S04 Guanidine nitrate Strychnine „ Aniline chloride Toluidine sulphate , p. ohloraniline chloride . . , Gone. 240 110 86 65-5 49-5 50-0 51-0 58-5 30-8 30-1 16-4 8-0 2-52 1-17 1-08 Electrolyte. Morphine chloride . Crystal violet . . Neo fuchsin . . . Bivalent Rations — MgSO^ .... MgClg CaCl2 SrCl, BaCfg Ba(N03)2. . . . ZnClg U02(N03)2 . . . Quinine sulphate . Benzidine nitrate , Trivalent Rations — AICI3 A1(N03)3. . . . Cone. 0-425 0-165 0-114 0-810 0-717 0-649 0-635 0-691 0-687 0-685 0-642 0-24 0-087 0-093 0-095 I02 GENERAL PROPERTIES OF COLLOIDS Hardy's rule that the precipitating power of an electro- lyte depends on the valency of the ion, whose electric charge is opposite to that on the sol, is thus fully estab- lished. We thus have the same law applying to the amount and direction of electroendosmosis and kataphoresis on the one hand, and to precipitation on the other. It is then highly probable that the latter is a conseq^uence of the reduction or elimination of the potential difference between the disperse phase and the medium. If the electric charge on the particles is due to the adsorption of kations or anions, according as the sol is -|- or — , it is easily seen that the neutralisation of this charge requires the adsorption of anions or kations. Further, the readily adsorbed H' and OH', which we have seen to have so powerful an effect in electroendosmosis, should also have a greater precipitating power than other univalent ions. The data given above contain illustrations of such an effect. Not only so, but any ion which is readily adsorbed ought to have a lower precipitation- concentration than other ions of the same valency. Among such are the organic ions. We shall return to this later on. Explanations of the Valency Rule. — Various explana- tions have been advanced. Qualitatively, there is little difficulty, if it is remembered that precipitation is a con- sequence of removing a definite electric charge from the disperse phase by means of ions of opposite charge. As the charges on uni-, bi-, and tri-yalent ions are in the ratio 1:2:3, the equivalent amounts will be in the ratio 3:1-5: 1. The quantitative relation is not so easy. Whetham ^ applied the theory of probabilities, and reached the con- clusion that the ratios of the molecular preoipitation- ' Phil. Mag., 1899, [v.] 48, 474. PRECIPITATION 103 concentrations of univalent, bivalent, and trivalent ions are Cj : C2 : Cg = k^ : k^ : k h being a constant. Freundlich assumes that neutral salts, whatever their valency, are equally adsorbed from equimolecular solutions, and that they follow the adsorption law. The charges on uni-, bi-, and tri-valent ions being in the ratio 1:2:3. the amounts to be adsorbed, in order to produce complete log or 3 Log of n l-ogofl ] I -Z -1 . o Log Concentration Fig. 6. precipitation, will be in the ratio 3 : 1'5 : 1. The concen- trations of solutions in which bivalent and trivalent ions will be adsorbed to these extents, compared with the necessary adsorption of univalent ions, can be read from the adsorption curve (p. 251). These should be approxi- mately the precipitation-concentrations of the solutions. The simplest test of this is to plot the logarithms of the concentrations against the logarithms of 3, 1'5, and 1. The curve will be a straight line, if the adsorption law is followed. When this is done for the data for arsenious sulphide and platinum sols the curves are practically straight lines (Fig. 6). 104 GENERAL PROPERTIES OF COLLOIDS Adsorption of the Precipitating Ion. — ^Perhaps the best confirmation of the above ideas as to the mechanism of precipitation is to be found in the fact, that when a sol is precipitated, the precipitate " carries down " with it part of the reagent, or rather, of the ion of opposite charge. This was first noticed by Linder and Picton ^ with arsenious sulphide sol and barium chloride. The supernatant liquid was found to be acid, and a corresponding quantity of barium was found in the precipitate. It was so firmly held by the precipitate that no amount of washing with water availed to remove it. It could be removed by washing it with a solution of some other salt, the kation of which took the place of the barium. These observations have been subsequently confirmed and extended. According to the explanations of the preceding section, only the precipitating ions should be adsorbed, and in electrically equivalent amounts. The amounts of different ions adsorbed by a given amount of precipitate ought then to be chemically equivalent. This is the case.^ Thus 100 c.c. of arsenious sulphide sol adsorbed the following weights of ions : — Weight adsorbed (gram), Ion, observed. calculated, Ba . . . . 0-0076 0-0076 Ca . . . . 0-0020 0-0022 Sr . . . . 0-0039 00049 K . . . . 0-0036 0-0043 The weights equivalent to 0-0076 g. Ba are given in the last column. Preundlich^ examined several organic ions, which also agreed with the rule. The number of milligram equivalents 1 Ohem. Soc. Journ., 1895, 67, 63. « Whitney and Ober, J. Amer. Chem. Soc, 1902, 23, 842. » KoU. Ztitsch., 1907, 1, 321. PRECIPITATION 105 of kations adsorbed by 100 millimols of arsenious sulphide is very nearly 2 (the number varies from 2-5 for Ca to 1-8 for aniline). The impossibility of washing out the adsorbed ion with -water, and the quantitative replacement of it by other ions, correspond closely with the behaviour of adsorbed substances in general. The positive sols afford further confirmation. Organic anions have an unusually low precipitation-concentration for ferric hydroxide sol, e.g. — Sodium chloride .'. . 300 formate acetate benzoate salicylate 56 30 32 20 Now organic acids are as a rule strongly adsorbed, and so the amount of anion necessary for electrical neutralisa- tion and precipitation will be adsorbed from more dilute solutions. The Stabilising Effect of Ions. — Attention has just been directed to the specially low precipitation-concentration of organic anions on a positive sol. A glance at the tables on p. 101 shows that these salts have an abnormally high precipitation-concentration for the negative sols. The same holds for OH', and the reverse for H". Acids and bases occupy an exceptional position in all the tables, whether relating to -|- sols or — sols. Thus for ferric hydroxide the precipitation-concentrations are — BaCla 9-64 HCl 400 Ba(0H)2 0-42 The figure for barium chloride is about the same as that for any other chloride, except hydrochloric acid, and the figure for any univalent anion except OH', is nearly the same as for 01'. io6 GENERAL PROPERTIES OF COLLOIDS It thus appears that H' exerts an influence on a positive sol which is opposed to precipitation, while OH' has the opposite effect. The corresponding effect with negative sols will be that OH' hinders precipitation, while H' favours it, and this is what actually happens. The pre- cipitation-concentration of sodium chloride on a platinum sol was 2*5, of sodium hydroxide 130. The figures for a mastic sol were : sodium chloride = 0'12, and hydrochloric acid = 0'004. Bodlander's figures for kaolin are a good illustration, e.g. — HNO3 0-0016 ISTaNUa 0-0133 NaOH 1-19 CaCla 0-003 Ba(0H)2 0-37 In all these examples we see that H' and OH' exercise opposing influences on both positive sols and negative sols, biit that each of them has a similar effect on sols of the same sign as themselves, i.e. H' stabilises posi- tive sols, and OH' stabilises negative sols. This is what the facts of electroendosmosis, and the theories regarding the origin of the potential difference at the interface of liquid and solid, would lead us to expect. A positive diaphragm and a positive sol will have their positive charge increased by the presence in the liquid of a highly adsorbable kation, such as H' and the organic kations; they will be more readily discharged by highly adsorbable anions, such as OH' and organic anions. The converse will hold for negative diaphragms and sols. The precipitating effect of an electrolyte is due to a balance between two opposing influences : a suspending or stabilising effect due to the ion of the same sign as the sol, and a labilising or precipitating effect due to the ion PRECIPITATION 107 of opposite sign. The valency rule is approximately true because in the majority of cases, i.e. inorganic salts, the one force is much greater than the other. Still, even in the case of neutral salts the differences between univalent ions are far greater than the experimental error. Many instances could be cited in illustration of this view, but two will suffice. Bredig discovered a method of preparing metallic sols by electrical dispersion of solid metals under water. These sols, which are negative, were not very stable, and Bredig recommends the addition of a trace of alkali to pure water, to increase the stability. In Kohlschiitter's method of preparing silver sol (p. 183) there is always some unreduced silver hydroxide left. Most of this can be reduced by hydrogen gas in a platinum basin, but not quite all of it. The less hydroxide remains, the less stable is the sol. Unless very pure conductivity water is employed (electrolyte-free) the sols usually pre- cipitate towards the end of the purification. We must also revert to Burton's experiments on the kataphoresis of silver or gold sol, to which varying amounts of Al"' had been added (p. 81). The trivalent kation, which is so potent in discharging and precipitating the negative sols, reverses the direction of kataphoresis, and also increases the stability of the sol when added in larger quantity. In this connexion, his experiments with quadri- valent ions are interesting. He failed to obtain precipita- tion with stannic chloride or platinum chloride. There are two possible explanations for this failure. Probably in neither case does the solution contain a quadrivalent kation ; stannic chloride is hydrolysed into stannic acid sol, which is emulsoid, and would "protect " the silver sol from precipitation by electrolytes (p. 128) ; if the platinum chloride was PtCl^ (and not HgPtCle), it was probably entirely converted into an oxy-acid (Il2PtOCl4), in which io8 GENERAL PROPERTIES OF COLLOIDS the kation is H* and the platinum is part of the anion. The other explanation is that, in accordance with the valency rule, the precipitation-concentration of a quadrivalent ion is exceedingly small, and a small excess over this amount would, as in the above case of Al'", convert the silver sol , into a stable positive sol. The "Complex" Theoey of Colloids These relations between the stability of sols and the presence of ions or of electrolytes in the disperse phase have naturally led to the question whether a colloid is not essentially a complex made up of the. two. Thus the above silver hydrosol will be either xAg, yAgOH, or more simply xAg.yOTl', and so for other sols. This idea is found quite early in the history of colloids. Graham observed that dialysis, however prolonged, did not remove all the chloride from his ferric hydroxide sol, the limit being 98'5 per cent, of ferric hydroxide to 1"5 per cent, of hydrochloric acid. The amount of chloride was later reduced to less than half this, 99-15 per cent. Fe(0H)3 to 085 per cent. HCl, and recently again by Nicolardot ^ to 99*84 per cent. re(0H)3 to 0-16 per cent. HCl. The same thing occurs in many other instances, e.g., zirconium hydroxide sol,^ prepared by hydrolysis of the chloride and dialysis, contained 0"0048 g. CI' to 0'667 g. Zr02. The sulphide sols prepared by Linder and Picton ^ were found by them to contain hydrogen sulphide, which could not be removed without precipitation of the sol. They determined both the metal and the sulphur ; the excess of sulphur amounted to several per cent. They regarded the hydrogen sulphide as an essential constituent of the sols, ' Becherches stir le sesguioxide de fer, Th^se, Paris, 1905. . 2 Ruer, Zeitsch. anorg. Chem., 1905, 43, 282. = Chem. Soc. Journ., 1892, 61, 114. COMPLEX THEORY OF COLLOIDS 109 and assigned formulae to them, e.g. SAsgSg, HgS ; 22CuS, HgS; 12ZnS, H2S. Wyrouboff 1 and pupils represented the various dialysed ferric hydroxide sols as chlorides of "condensed" ferric hydroxides, a reminiscence of Graham's " co " salts. This idea was further developed by Duclaux,^ and extended to other groups of sols. In particular, copper ferrocyanide was shown to contain K', no matter what excess of CUSO4 had been employed. The formula varied between Cu2Pe(CN)e, iK4re(CN")6 and Cu2Fe(CN)6, ^K4Fe(CN)g. Similar results were found for cadmium sulphide from cadmium sulphate and hydrogen sulphide ; the sol always contained all three substances, whether there was excess of either the one reagent or the other. The stability of the sol depends on the presence of these ions. From this standpoint, precipitation by electrolytes becomes a chemical reaction, a double decomposition in fact, and the valency rule becomes obvious. The equi- valence of the ions necessary for equal precipitation has already been dealt with, but Duclaux's figures for ferric hydroxide sol are so striking that they are given below.2 The sol contained 0-0203 equivalents Fe, and 0-00166 equivalents CI per litre. 10 c.c. of the sol (= 16-6 X 10-^ equiv. CI) required 17 X 10-6 equiv. SO4 16-5 C6H5O7 (^citraie) 15-2 CrOi 19 X 10-6 equiv. PO4 16 OH 13 Fe(C 17 CO3 Similar figures were obtained for the negative copper ferrocyanide sol. 1 Bull. Soc. chim., 1899, 21, 137. 2 Compt. Bend., 1904, 138, 144, 809; 1905, 140, 1468, 1544; 1906, 143, 296, 344 ; J. Chim. Phys., 1907, 5, 29. no GENERAL PROPERTIES OF COLLOIDS In the attempt to prepare pure silicic acid sol by Graham's method, Jordis * found that the more it was purified the more unstable it became, and formed the conclusion that sols are chemical compounds. The amount of salt associated with the colloid varies with the concentration of the salt in the liquid,^ which is contrary to chemical combination, but is in agreement with adsorption. Lottermoser^ has shown that silver halide sols can be made from very dilute solutions of silver nitrate and KX. If silver salt is added to potassium iodide solution, a pilver iodide sol is obtained, which is stable so long as there is an excess of potassium iodide ; but so soon as Ag" and I' are exactly equivalent, the sol is unstable and is precipi- tated. These sols are negative. Starting in the reverse order, the sol is stable so long as there is excess of silver nitrate, and is precipitated when there is none. These sols are positive. "We are thus led to conclude that the stability of sols is connected with the presence in the sol of small quantities of other substances, usually substances related to or derived from the colloid itself. For one class of sols this seems improbable : the metal sols, especially the sols of noble metals, as gold and platinum, and most particularly when prepared by Bredig's method. The gold sols formed by reduction of auric chloride contain CI', for after precipitation 01' is formed in the liquid.* But in Bredig's and Svedberg's electrical dispersion methods (p. 213) the sols are formed at very high temperatures in the presence of air. It is by no means impossible for traces of oxides or hydroxides to be ^ Zeitsch. aiiorg. Cliem., 1903, 38, 16 ; Zeitscli. Elektrochem., 1904, 10, 509. 2 Henri and Mayer, Compt. Bend., 1904, 139, 924. ' J.pr. Chm., 190S, 72, 39; 1906, 78, 874. « Steubing, Ann. Physik., 1908, [iv.] 26, 835. PRECIPITATION in formed under these conditions. The base metals when treated in this way give sols which, in appearance and general behaviour, are like hydroxide sols.^ Even if air is excluded, the dispersion medium is undergoing chemical decomposition, owing to the high temperature, and so the possibility of the formation of compounds is increased. The instability of Kohlschiitter's silver sol, when freed as far as possible from silver hydroxide, has already been referred to (p. 107). The Action of Eleotkolytes on Emulsoids The differences between suspensoids and emulsoids in their behaviour towards electrolytes are very marked, even more so than their electrical behaviour would lead us to anticipate. They are not merely differences between liquid and solid disperse phases, for ferric hydroxide sol, which is probably liquid, does not differ in this respect from other sols which are undoubtedly solid. As has been pointed out (p. 19), emulsoids are not only liquids, but par- take more of the nature of solutions, part at least of the col- loid probably being in solution. Whether this is the reason or not, the behaviour of emulsoids towards electrolytes is so diverse and so complex, that classification and compre- hensive survey are (for the present) out of the question. All that can profitably be attempted is to present the more general features in a few of the most important and best-known cases. Even the classification of precipita- tion as reversible and irreversible is not justifiable, as reversibility is mainly a matter of time. It is rather unfortunate, too, that so much of the work has been done on the natural emulsoids, as white of egg, serum albumin, gelatine, frequently without any assurance as to the ' Burton, Phil. Mag., 1906, [vi.] 11, 425. 112 GENERAL PROPERTIES OF COLLOIDS purity or even uniformity of the material. This circum- stance makes the comparison of results a doubtful matter, and has surely retarded progress. The oil emulsions of Hatschek (p. 85) are not par- ticularly sensitive to electrolytes. Thus, the dispersity of a 1 in 5000 emulsion of light petroleum, whose dispersity is about 4 X 10"^ cm. was not changed by sodium chloride, but 155 m.mol of sodium sulphate per litre produced a clearing up. 19 m.mol of hydrochloric acid per litre had barely any effect on the sol, but a marked effect was pro- duced at a concentration of 23 m.mol per litre. The particles became larger and the original reddish coloured emulsion with a bluish opalescence became white. I. Albumin The Action of Electrolytes on Albumin. — If we leave out of account acids and alkalies, salts may be grouped into three classes, according to their action on albumin sols. The first group produces precipitates, even when in small concentration; it comprises salts of the heavy metals." The other two groups only cause precipitation at really considerable concentrations ; in the one group, the precipitation is reversible, while in the other it is not. The second group comprises the alkali salts : K, Na, NH4, Li, and Mg ; while the third group contains the alkaline earth salts. The difference is one of time, for in the former the precipitation becomes irreversible after a time, and in the latter it is reversible at first, but very rapidly ceases to be so. Much of the work has been done on purified " natural " egg-white ; egg-white is beaten up to a stiff froth, poured into a tall narrow cylinder, and the clear liquid which collects at the bottom is separated and diluted with water. This not only contains a mixture of albumins, but is not PRECIPITATION 113 electrolyte-free ; it is in fact alkaline. From the electrical differences between neutral and acid or alkaline albumin, it is not surprising tliat this preparation does not exhibit the same behaviour as pure albumin. It is necessary to bear in mind the possibility of these differences ; only pure neutral albumin is suitable for experimental work. The three principal constituents of egg-white, egg-albu- min, ovomucoid, and a globulin, differ widely in many respects, and yet are not very readily separated from each other. Pure crystallised albumin is prepared by Hofmeister's method, or by modifications of it. The properties of serum albumin, again, differ considerably from those of egg-albumin. Reversible Precipitation, Salting Out. — When a satu- rated solution of ammonium sulphate is added, little by little, to albumin sol, no precipitate forms until an equal volume of the salt solution has beep added, but precipi- tation is not complete until a further amount is added. To be more exact, precipitation commences when 10 c.c. of the mixture contain 6"2 c.c. of saturated ammonium sulphate solution, and is complete when 10 c.c. of the mixture contain 6'8 c.c. of the saturated solution. These limits are pretty close together, much more so than for many allied substances ; the total amount required also differs from substance to substance. The figures for egg globulin are 2*9 and 4"6 ; this indicates a method of separation which is made use of If water is added, the precipitate dissolves, and will reappear on addition of sufficient ammonium sulphate solution. Hofmeister utilised this in the preparation of pure albumin.^ The usual method is a modification which 1 Hofmeister, Zeitsch. physiol. Chem., 1889, 14, 163; 1891, 16, 187. Hopkins, J. Physiol., 1898, 23, 130; 1900, 25, 806. Sohulz, Zeitsch. ]jhysiol. Chem., 1899, 29, 86. I 114 GENERAL PROPERTIES OF COLLOIDS gives a larger yield, and is altogether more convenient. It is as follows : Mix equal volumes of egg-white and satu- rated solution of ammonium sulphate. Beat to a stiff froth, and let it stand overnight. Filter oif the precipi- tate of globulin and mucoid, and add gradually to the clear filtrate, with gentle stirring, a 10 per cent, solution of acetic acid, until a slight permanent precipitate is formed. This mixture should be just acid to litmus. Then add 1 c.c. of the 10 per cent, acid for each 100 c.c. ; a bulky precipitate is formed, which becomes crystalline in five hours. Allow to stand till next day, to obtain the full yield (60 g. per litre). To purify : filter, wash the pre- cipitate three times with half-saturated ammonium sulphate solution containing O'l per cent, of acetic acid. Dissolve the crystals in the minimum of water, and add saturated solution of ammonium sulphate until a pre- cipitate is formed ; then add 2 c.c. more to each litre. The albumin recrystallises in 24 hours. Repeat this treatment as often as necessary. Schulz and Zsigmondy have shown that 3 to 6 recrystallisations are necessary to remove all the colloid impurities. Finally, the crystals are dissolved in pure water, and dialysed to remove the ammonium salts (p. 26). This method has been given at length, because it may be of use, but mainly because it has been applied by von Weimarn, with suitable modification, to obtain in % crystalline state substances which are usually regarded as amorphous (p. 121). This precipitation differs from that of suspensoids in the large precipitation-concentration, in reversibility, and also in the absence of a valency effect, either of kation or anion. The following figures ^ for purified native albumin show this. ' Hofmeister, Arch, exper. path, pharm., 1888, 24, 247. PRECIPITATION 115 Concentration Concentration Salt. mols per litre. Salt. mols per litre. Sodium citrate . . 0'56 Sulphate, Li . . . 0-78 „ tartrate . » 0-78 K . . . 0-79 „ sulphate . 0-80 Na . . . 0-80 „ acetate . 1-69 NH^ . . 1-00 „ chloride . 3-62 Mg . . . 1-32 „ nitrate . 5-42 „ chlorate . 5-52 Sodium iodide and thiocyanate produce no precipitate. Pauli^ concludes from his extensive results that kation and anion effects are independent, the former favouring precipitation, while the latter hinder it. The order of anions arranged according to their pre- cipitating power is thus — citrate > tartrate > SO4 > C2H3O2 > CI > NO3 > CIO3 > I > CNS and of kations— Li K > Na > NH4 > Mg We shall come across this order of ions again. The order for salts of alkaline earths ^ is — kations : Ba > Ca anions : C2H3O2 < CI < NO3 < Br < I < CNS This is the reverse of the order of anions in the alkali salts. It is noteworthy that addition of barium chloride produces an acid reaction in a neutral albumin sol. If an albumin, to which a trace of acid has been added, is used instead of neutral or natural albumin, the anion ^ Beitr. chem. physiol. path., 1903, 3, 225. ' Pauli, 'Beitr. chem. physiol. path., 1904, 5, 27 ; Posternak, Ann. Institut Pasteur, 1901, 15, 85. ii6 GENERAL PROPERTIES OF COLLOIDS \' . . order is reversed ; and if much acid is added, the precipita- tion is irreversible. With 0'03a HCl, the order is — kations : Na > K > NH4 > Mj? anions: SO4 < NOg < Br < CNS - This difference of acid and alkaline albumin sols is undoubtedly connected with the change in sign of the electric charge on albumin in acid and alkaline liquids. Irreversible Coagulation of Albumin. — The line of demarcation between reversible and irreversible coagula- tion is so indefinite that already some irreversible coagulations have been considered. Coagulation by Heat. — When aqueous albumin sols are heated, coagulation takes place at some definite temperature, which is fairly constant for each kind of albumin. The nature of the change is unknown, but is apparently chemical, for natural albumins become more alkaline, and acid albumins become neutral or less acid. The two chief factors are the reaction of the sol, and the kind and amount of salt present. Complete coagulation only occurs if the sol is faintly acid. If the sol is too acid, or is not acid at all, more or less albumin remains in the sol. Further, a dialysed salt- free egg albumin is not coagulated by heat. This was first noticed by Aronstein,^ and has been frequently con- firmed. It is, nevertheless, not quite accurate, for co- agulation does occur, when the heating is continued for a sufficient time ; it occurs at a constant temperature, too, which is close to the ordinary coagulation temperature. Serum albumin, if free from salts, is completely coagulated by heat ; and the presence of traces of acid or alkali com- pletely prevents coagulation. The effect of salts on heat-coagulation has been closely * Pfluger's Archiv., 3874, 8, 75. PRECIPITATION 117 stvidied by Pauli and others, both with natural albumin, and with pure dialysed albumin. The results with the former are very complicated, and will not be referred to. The following figures show the influence of salts ; they were obtained with pure dialysed serum-albumin, the coagulation temperature of which was 60'3°. The table gives the coagulation temperatures of the albumin for the given concentrations of salts. Coagulation Temperature, Equi iv. concentration. Salt. 0-01 0-02 0-03 0-04 0'05 NaCl . . , . 63-2° 65-7° 66-4° 67-2° 67-9' ¥a2S04 . , . 66-7° 68-0° 68-5° 69-1° 69-7' NaA . . . 66-9° 69-2° 70-6° 71-5° 72-1' NaCNS . . 68-0° 69-7° 70-6° 71-6° 72-5° These figures correspond to the adsorption curve (p. 251) ; in agreement with it, small concentrations pro- duce relatively larger effects than do larger concentrations. In every case the coagulation temperature is raised; it slowly rises up to concentrations of l-2n, in some cases to a constant value at 2-5n, e.g. alkali chlorides, bromides, nitrates. In another group (fluoride, sulphate, acetate, citrate) a maximum temperature is reached, after which it begins to fall, and may even be below that of the pure albumin. With iodide and thiocyanate the temperature rapidly rises after l-2n, and at 2-3n no coagulation occurs even on boiling. Here, again, we find the same lyotrope sequence of anions. There appears to be a connection between this effect of salts on heat coagulation, and the effect on viscosity (p. 23). Acid and Alkaline Albumin. — As amphoteric substances,* albumins unite with acids and bases, and become posi- tive and negative sols. Consequently, they differ very ii8 GENERAL PROPERTIES OF COLLOIDS much from neutral albumin ; some of these differences have already been stated. One further instance will suffice. A trace of acid or of alkali completely prevents the heat-coagulation of serum-albumin, but the acid sol is completely coagulated by a trace of a salt, e.g. sodium chloride, and the alkaline sol is partially coagulated by salts, especially by traces of calcium. Precipitation by Salts of Heavy Metals. — Here the effect is wholly due to the kation, and precipitation takes place at very small concentrations. In all cases the precipita- tion is irreversible, except under certain special circum- stances. The precipitate always contains the precipitating kation. Three types of reaction occur. I. Silver salts precipitate at all concentrations from 0"ln, to 6n ; the precipitate does not dissolve in excess of silver solution, but some remains in solution so long as excess of albumin is still present. II. Copper salt solutions from O'OOln to In give a precipitate; above In no precipitate is formed. A secondary precipitate is formed with a 6nCu" solution. III. Zinc salts from O'OOln to 0'5n give a precipitate ; from 0'5n to 4n no precipitate ; and beyond 4n a pre- cipitate is formed. This is for a 1 : 10 egg-albumin. If the sol contains more albumin, the region of non-precipitation is reduced. The behaviour of zinc sulphate is illustrated in Fig. 7. In all these cases the precipitates contain metal as well as albumin, and have been regarded as chemical compounds. If so, their composition is remarkably variable, e.g. the copper albumin precipitates contain anything from 1'4 to 20 per cent, of CuO; the same applies to the others. A few examples are given, the figures give the percentage of metal in the egg-albumin metal precipitate. PRECIPITATION 119 (1) Ag 2-17, 3-3, 3-9, 4-3, 4-86, 6-26. (2) Cu 0-7-2-2, 1-2, 1-2-1-35, 1-35-2-65, 2-24-2-65, 3-55, 3-95. (3) Za 0-9-3-7. In conclusion, the precipitation of albumin by heavy metals bears a strong resemblance to the mutual precipita- tion of sols, and to the precipitating and stabilising action of multivalent ions (p. 137). Concenfy-atlon of ZnSO^. 1,2,3.= Increasing Concentrations of AlbunvLrh. Fia. 7. Coagulation by other msans. — Many organic substances, e.g. alcohol, phenol, formaldehyde, tannin, coagulate albumin irreversibly. Alcohol does not coagulate pure salt-free albumin, but does so if a trace of salt is added. This follows the same regularities as heat-coagulation. Urea and many other indifferent organic substances hinder the coagulation by alcohol. There are also many colouring matters, either basic or 120 GENERAL PROPERTIES OF COLLOIDS acid, which coagulate albumins. These reactions are of great importance in the staining of tissues; the mech- anism is probably the precipitation of negative albumin sol by the positive sol (or by the kation) of the basic dyes, and conversely of the positive albumin sol by the negative sol (or by the anion) of the acid dyes (see below). Albumins can be separated from the sol by mechanical means : ^ shaking, bubbling air or a gas through it, etc. Eamsden also noticed that all albumins form in time a solid or a very viscous skin at a still surface. This is a general occurrence with substances which greatly lower the surface tension of the liquid, and so can pass into the surface layer (see Adsorption). This phenomenon is well shown by some dyes (fuchsin, methyl violet). A similar coagulation occurs at the interface between aqueous albumin and organic liquids insoluble in water (toluene, chloroform), especially when the interface is enlarged by shaking the liquids together. The action is the same as in air, accumulation of colloid in the surface layer until solidification occurs (p. 258). The skins are insoluble in water (the denaturising process appears to be spontaneous, and therefore irreversible). II. Gelatine While the main interest of the albumins centres in precipitation and coagulation, in gelatine and similar colloids, it is only of secondary interest, the chief problem being that of the reversible change — sol to gel, to which processes the terms solation and gelation are applied (see Introduction, p. 10). This process is entirely different from coagulation, as will appear immediately. 1 Eamsden, Proc. Boy. Soc, 1903,72,156; Zeitsch.physikal. Chem., 1904, 47, 336. PRECIPITATION 121 The properties of gelatine vary much with the presence of impurities. By adopting the principle of Hofmeister's original method of crystallising albumin, von Weimarn 1 has succeeded in crystallising gelatine and agar. A very dilute, quite clear gelatine solution in aqueous alcohol is maintained at 60°-70° in a desiccator. The water is slowly abstracted by, e.g., dry potassium carbonate, which does not take up alcohol vapour. The solubility slowly decreases as the concentration of the alcohol increases. At last the gelatine commences to crystallise out. (See also p. 114.) A further difficulty is that gelatine is not very stable, at least in presence of water, and especially if heated. This is shown by the change in viscosity which a gelatine sol undergoes when kept at constant temperature. ^ iscosity of 2 per cent. Gelatine at 100 Time (hours). V 2-29 2 1-68 4 1-58 6 1-47 10 1-40 11 1-39 After 36 hours' heating, gelation no longer occurs (change to j3 gelatine). Ultramicroscopic investigations are in ac- cord with this. A sol prepared at 60° contains submicrons, but after prolonged heating at 100°, only amicrons. Gelatine is not amphoteric, like albumin, but is essentially acid ; ^ notwithstanding this, it appears to be primarily a positive sol. Beversible Sol-Gel Transformation. — If a moderately 1 Grundziige d. Dispersoid Chemie, 1911, p. 106, " Levites, Koll. Zeitsch., 1908, 2, 210. ' Hofmeister, Ber., 1878, 2, 299. 122 GENERAL PROPERTIES OF COLLOIDS strong sol of gelatine in water is made at high temperatures, on cooling it sets to a jelly, which is reconverted into the sol on warming. This change does not take place at a definite temperature (like a melting-point), for the process is a continuous one, as is shown by the cooling curve, or the temperature-viscosity curve. Fairly comparable results can be obtained by selecting some arbitrary standard of elasticity or of viscosity. Such a device was 0-25 0-5 0-7^' Conceivtratiorh Fig. 8. In that adopted by Pauli,i who observed the temperature at which the thermometer was just held firmly by the gel ; or by Levites,2 who measured the times which elapsed before the viscosity became so great that the sol no longer ran out of the inverted vessel. In the study of gelation, either the temperature or the time of gelation may be determined ; 'the latter is prefer- able. Schroder 3 measured the power of setting by the 1 Arch, gesam. Physiol., 1898, 71, 1. « J. Buss. Phys. Chem. Soc, 1902, 34, 110. » Zeitseh. johysikal. Qhem., 1903, 45, 75. 10% 15% 21° 25-5° 29-6° 29-4° PRECIPITATION 123 change in viscosity of the sol with time (drjjdt). What- ever method is adopted, it is essential that the previous treatment of the sol be uniform, as gelatine exhibits "hysteresis" in a marked degree, more especially after long heating. One cause of this is given above. Pure gelatine sets between 18° and 25°, and when heated melts between 26° and 29°. The temperature varies with the concentration; below 0'25 per cent, or 0'18 per cent.,^ it does not set even at 0° C. This variation is shown by the following 2 : — 5% Sets . . . l7-8° Melts , . 26-1° The temperature is also affected by salts and by organic substances. Some salts raise the gelation temperature, others lower it. The order is as follows : — Eaise . . SO4 > cr> T > A (HgO) Lower . . 01 < CIO3 < NO3 < Br < I The kation effect is very small, the order seems to be — Na > K > NH4 > Mg Glucose and glycerol raise it, while alcohol and urea lower it. The same lyotrope order was found in Levites' investigation on the time of gelation, and by Schroder for the effect of salts on dr)/dt (Fig. 8). The effect of salts increases slowly at first with increasing concentration, and then more rapidly. The curves are somcM'hat irregular. Schroder found that if At]/ At at 25° was > 0*0095, the sol would form a gel within 24 hours, but if it was < 0*0075, it would remain liquid (Aij is the differences in viscosity at 25° at the times ti and (2). > Levites, Koll. Zeitseh., 1908, 2, 162 ; Rohloft, Physikal. Zeit., 1907, 8, 442. ' Pauli and Kona, Beitr, chetn. physiol. path., 1902, 8, 4. 124 GENERAL PROPERTIES OF COLLOIDS As the effect of salts on gelation is additive, many mixtures of salts will leave the gelation temperature unchanged. Salting Out, and Precipitation of Gelatine by Salts. — This process is totally different from gelation. In the first place it is irreversible ; further, it is produced by salts which favour gelation and by those which retard it. Organic substances influence gelation, but do not precipitate (certain exceptional substances, e.g. tannin excluded). It only occurs at high concentrations of salts, in this respect resembling the salting out of albumins, e.g. precipitation is produced by 4nKCl (which lowers the gelation tempera- ture about 13°), and also by 0"75nN'a2SO4 (which raises the gelation temperature about 3°). The order of anions, which alone have any considerable effect, is again lyotrope — SO4 > Citrate > Tartrate > Acetate > CI The action is not additive,^ and the effect of mixtures is very complicated. Effect of Electrolytes on the Osmotic Pressure. — Pure gelatine sols have a fairly high osmotic pressure (for 1"25 per cent, sol it is 6"2 mm. mercury), and this is not altered by non-electrolytes ; but neutral salts lower it markedly .^ The order is again lyotrope — Cl(0-44) > SO4(0-47) > NO3(0-47) > Br(0-53) > 1(0-57) > C]SrS(0-60) The numbers are fractions of the osmotic pressure of the pure sol. Both acids and alkalies greatly increase the osmotic ' Pauli, ArcMv. gesam. Physiol., 1898, 71, 336. ■' Lillie, Amer. J. Physiol., 190Y, 20, 127. PRECIPITATION 125 pressure of gelatine, from 8 mm. to 39 mm. in 0-0024n HCl, and to 25 mm, in 0-0024n KOH. Agar and other similar substances show the same characteristics as gelatine ; in particular the effect of salts on gelation is very similar, the order being lyotrope. III. Silicic Acid This, the earliest artificial emulsoid, was discovered by Kiihn ^ in 1853. Graham prepared it by dialysis, and noticed that its stability increased the longer it was dialysed. The pure sol is slightly acid (100 g. require 1"85 g. KOH for neutralisation), and is not so stable as a neutral sol, but its stability is increased by small quantities of hydrochloric acid. Kiihn's sol and Graham's sol behaved quite differently to reagents ; the former was irreversibly precipitated by sulphuric acid, or by freezing, the latter was not precipitated by acids or neutral salts, but by carbonates. These different results are^characteristic of emulsoids. Four main facts, however, stand out : the sol is un- stable ; gelation is irreversible ; the stability is affected by acids and alkalies (as is also the sign of the charge on the colloid) ; the maximum of stability occurs at the iso- electric point, with a small H* concentration. Gelation is irreversible ; all the investigations agree on this (except Kiihn's method of preparation). The change takes place spontaneously, as the viscosity steadily increases with time until the gel is formed.^ It is accele- rated by rise of temperature, and by electrolytes. The temperature effect is not very large, e.g. boiling does not ' J.pr. Chem., 1853, [i.] 59, 1. ° Garrett, Dissertation, Heidelberg, 1903, p. 51, 126 GENERAL PROPERTIES OF COLLOIDS necessarily cause gelation. The effect of cooling is un- certain, some, as Kuhn and Ljubawin,i gtate that complete gelation occurs on freezing the sol, while others, as Mylius and Groschuff,2 did not find gelation to occur on cooling a sol to -23°. 1-5 0-5n(OH') 0-5n(H') 7 Alkaline „ Acid Concentration. Pig. 9. Eifect of Electrolytes. — The action of acids and alkalies is the most important factor. Billiter ^ found kataphoresis to the anode in alkaline and feebly acid solutions ; in these it is a negative sol, while at greater H" concentra- tions it travels to the kathode, and is thus positive. The sol is therefore at its isoelectric point in feebly acid solu- tion. Now Tlemming,* who made a close study of the ■ J. Buss. Phys. Chem. Soc, 1889, 21, 397. 2 Ber., 1906, 39, 116. ' Zeitsch. physikal.'Cliem., 1905, 51, 150. » Ibid., 1902, 41, 443. PRECIPITATION 127 effect of H' and OH' on the rate of gelation, found the maximum velocity at small OH' concentration, and the minimum at small H" concentration. With greater OH' concentration the velocity decreased, and it increased with greater H" concentration (Fig. 9). We have thus a sol which, unlike the suspensoids, is most stable at its iso- electric point. In this respect it resembles egg-albumin. According to Hardy's results,^ the behaviour of silicic acid sols to neutral salts is very similar to that of negative suspensoids. This is probably true for alkaline sols, which are strongly negative. Thus, for instance, the pure sol (feebly negative) is not sensitive to kations, K", H', ^^4, Ba", etc., but becomes so on addition of ammonia, which increases the negative charge. It is then at once precipi- tated by barium chloride (this gel at once dissolves when excess of hydrochloric acid is added). Pappada,^ on the other hand, found that neutral salts only act at great con- centrations, and accelerate gelation in accordance with the lyotrope series : SO4 > 01 > NO3 : Cs > Eb > K > Na > Li Stannic acid, titanic acid, tungstic acid, molybdic acid, and also starch sols are more or less like silicic acid sols in their general behaviour. 1 Zeitsch. physikal. Ckem., 1900, 33, 391. 2 Gazz. cliim. ital., 1903, 33, [ii.] 272 ; 1905, 35, [i.] 78. CHAPTER X peotection and mutual pkecipitation Protection of Sols Faraday, who prepared gold sols by reduction of auric chloride with phosphorus (p. 179), discovered that they were more stable when "jelly" (no doubt gelatine) was added to them.^ The gold chloride was evaporated to dryness with the jelly ; the resulting red-coloured jelly did not change colour with salt, as the " ruby fluids " did. Purple of Cassius, too, which has long been known, is another example of increased stability of a colloid in the presence of another. Many other instances could be given in which use was made of this " protective " action, without any precise knowledge of its nature. The pro- tection of less stable sols by organic sols (of suspensoids by emulsoids) was first definitely recognised by E. von Meyer and Lottermoser.2 They observed that albumin prevented the precipitation of silver sol by salts. The latter subsequently reached the conclusion ^ that "on addition of very stable colloids, as albumin, gelatine, agar, or gum arable, to a silver sol, no precipitation is caused by electrolytes until this stable colloid is gelatinised. The less stable silver sol is thus 'protected' against the > Phil. Trans., 1857,^ 154. " J.pr. Chem., 1897, [ii.] 56, 241, ^ Anorg. Kolloide, 1901, p. 50. PROTECTION OF SOLS 129 electrolyte by the more stable colloid ; it becomes more like the latter in its behaviour." Zsigmondy then investigated the action quantitatively by means of his gold sol (p. 185), which is itself very stable, and is also very sensitive to salts. .The degree of protection is expressed by the " gold number," ^ which is the weight (in mg.) of colloid which just fails to prevent the change from red to violet in 10 c.c. of a gold sol (0-0053-0-0058 per cent.), when 1 c.c. of 10 per cent, solution of sodium chloride is added to it. The reciprocal of this number would be a more convenient expression of the protecting power. The following are taken from Zsigmondy's table : — Colloid. Gold number. Gelatine . . . . . 0-005-0-01 Isinglass . . . . 0-01-0-02 Casein (in NH3) . 0-01 Egg-albumin . , . 0-15-0-25 Gum arabic . 0-15-0-25 ; 0-5-4 Gum tragacanth . ca. 2 Dextrin . . . , . 6-12; 10-20 Starch, wheat . . ca. 4-6 „ potato . . ca. 25 Sodium stearate . 10 (at 60°); 0-01 (at 100°) „ oleate . . 0-4-1 Stannic acid (old) . 00 Urea .... . CO Cane sugar . . - . 00 The gold number is a useful means of characterising substances, especially in cases which present difficulties to ordinary analytical methods. Thus 1 per cent, of gelatine will reduce the gold number of dextrin from 10 to less 1 Zeitsch. anal. Chem., 1902, 40, 697. K 130 GENERAL PROPERTIES OF COLLOIDS than 1. It has already proved useful in the fractionation of albumin.^ Preparation. Gold number. White of egg (fresh) . . . 0-08 Albumin (Merck) ', , . . 0-1-0-3 Globulin 0-02-0-05 Ovomucoid 0-04-0-08 Albumin (cryst.) .... 2-8 Alkali albumin 0-006-0-04 If alkali is added to any of them, the differences disappear ; all the alkali albumins have the same low gold number. This protection is not confined to metal sols, nor is it exerted by organic colloids alone. The sulphide sols — arsenious sulphide, cadmium sulphide, etc. — are protected in the same manner,^ though to different extents.; further, the order of efficiency of the protecting colloid differs from sol to sol. Gold sol and silver sol have been prepared with silicic acid^ as protective colloid, and zirconium hydroxide* sol exceeds even gelatine in its protection of gold sol. Another well-known example, is that of ferric hydroxide sol ; a basic ferric chloride solution when dialysed gives no visible precipitate with silver .nitrate even when a considerable quantity of chloride is still present, as can easily be shown in other ways. Paal's protalbic acid and lysalbic acid^ (products of the action of alkali on albumin) are extremely active protectors. They have been used in the preparation of a large number of stable sols — comprising metals, metallic oxides and hydroxides, and salts — containing high per- centages of colloid (see Preparations). These sols may be evaporated to dryness, and even after years will dissolve readily in water containing a little alkali. ' Sohulz and Zsigmondy, Beiir. chem. physiol. path., 1902, 3, 137. ' Miiller and Artmann, Osterr. chem. Zeit., 1904, 7, 149. » Kiispert, Ber., 1902, 36, 2815, 4066. * Biltz, ibid., 4431. ' Ibid., 2195. PROTECTION OF SOLS 131 We may thus say that a suspensoid sol, when mixed in suitable proportion with an emulsoid sol, loses most of its characteristic properties, and gains those of the protecting emulsoid. Bechhold ^ suggested that the protection is the result of adsorption. A thin layer of the emulsoid is adsorbed at the interface, and thus confers its stability on the adsorbing particle. Thus only substances which are readily adsorbed {i.e. which lower the surface tension of the dispersion medium) can act in this way. The objection that the amount of the very active colloids is too small to cover the particles is only an apparent one, for the concentration of most of the sols protected is very small ; e.cf. in the Zsigmondy test, the amount of gold is 0*5 mg. and of gelatine required for protection is 0"005-0'01 mg. The difference is still less if account is taken of the difference in density of gold and gelatine. According to Zsigmondy the coating of gelatine is invisible in the ultramicroscope, and there is no apparent diminution of the Brownian movement. If the action is due to adsorption of the protecting colloid, it is clear why their protective power is not parallel with their stability, for adsorption, and possibly other factors too, will come into play. The protection is not merely against electrolytes, for the stability of the mixture is practically the same as that of the protector, e.g. as regards solation and gelation; heat ; evaporation ; imbibition ; solation by digestion with water, acid, alkali, etc. ; ultrafiltration. The relative amounts of the two colloids are of the utmost importance, for precipitation takes place with smaller amounts of the protective colloid ^ (see p. 135). Owing to the small concentration of metal sols it is not 1 Bechhold, Zeitsch. physikal. Chem., 1904, 48, 385. * Priedemann, Archiv. Hygiene, 1906, 15, 376. 132 GENERAL PROPERTIES OF COLLOIDS easy to get this effect with them, and so the protection is very marked. It may be added that the reverse of pro- tection has been effected.' By adding excess of a mastic sol to an albumin sol, the albumin could be readily precipitated by means of electrolytes, etc. (See also next section.) Mutual Pkeoipitation of Sols When Lottermoser mixed various sols with a view to obtaining mixed gels by precipitation of the mixture with salts, some of the sols precipitated each other without the addition of any salt. On making a table of these in- compatible sols, he was surprised to find that the classifi^ca- tion was the same as he had already obtained from their electrical migration.^ In other words, a positive sol will precipitate a negative sol, while sols of the same electric sign will not do so. The precipitate contains both colloids, but, owing to experimental difiBculties, he was unable to determine their amounts. ' Very little ferric hydroxide sol sufficed to precipitate a large amount of silicic acid sol ; with 20 CO. of the latter (0-144 g. SiOg) and 10 c.o. of the former sol (0-040 g. FejOs), the liquid above the precipitate was still brown. The experimental difficulties are three in number : difficulty of filtration without adsorption (or precipitation), spontaneous precipitation of the excess of colloid in the sol, and slowness of the reaction. Occasion- ally spontaneous solation occurred. The mutual precipitation of certain dyes had previously been noticed by Linder and Picton ; 3 they also knew that these dyes were colloids of opposite electric charge. Suspensoids. — A quantitative method was devised by ' Michaelis and Eona, Biochem. Zeitsch., 1906, 2, 2] 9. ' Anorg. KoUoidi, 1901, p. 77. 8 Chem.'Soc. Joum., 1897, 71, 568. MUTUAL PRECIPITATION OF SOLS 133 Biltz; it is a synthetic inethod,i whereas Lottermoser's unsuccessful ones were analytical. Varying known amounts of the one sol are added to a constant amount of the other, and the result noted. Very small amounts cause no visible change, more causes partial precipitation, larger amounts cause complete precipitation, while still larger amounts give less precipitate, and at last no precipitation occurs. The region of complete precipitation is fairly narrow, while, with large excess of either sol, no precipitation occurs at all. The following shows this. 10 c.c. gold sol (= 1"4 mg.) + 5 c.c. thorium hydroxide sol of variable concentration. Appearance, mg. ThO^. Immediate. 0'5 no precipitate 1"0 trace of precipitate 2'0 slow precipitation 2*5 rapid complete precipitation 3*0 slow „ „ 4"0 trace of precipitation 5-0 no „ 30 minutes later, slight opalescence trace of precipitate slow precipitation complete „ no change The optimum amounts of various positive sols for gold sol differ considerably, as also does the optimum amount of a positive sol for various negative sols, but the order is always the same, e.g. : Negative sol. Optimum of positive sol (ia mg.). FC2O3. ThOj. CeOa. ZrOj. Cr.Oj. AI2O3. Au 1-4 mg. . A3.S324 „ . SbjSj 28 „ . 3 13 32 2-5 6 20 4 4 11 1-6 2-0 6-5 0-3 0-5 30 0-1 -0-2 20 2-0 ' Ber., 1904, 37, 1095. 134 GENERAL PROPERTIES OF COLLOIDS There is obviously an amount of one sol which is equivalent to a given amount of the other ; it is not, how- ever, a chemical equivalence, nor is the precipitate, which must contain the two colloids in practically constant pro- portions, to be regarded as a chemical compound, in spite of its constant composition. The equivalence is electrical ; as Billiter first suggested,^ the maximum precipitation occurs when the positive charge on the one sol exactly equals (and neutralises) the negative charge of the other. This is in agreement with what we learnt regarding the precipitation of suspensoids by electrolytes, and the maximum of instability at the isoelectric point (p. 96), The optimum precipitation may not correspond exactly to electrical equivalence, for obviously the number of particles required a,nd their size (i.e. the electric charge and the dispersity) as well as the relative concentrations of the two sols, must affect the precipitation, as also does the rate at which they are mixed (p. 98), In accordance with this, the composition of the precipitate is not quite constant. It always contains both, but so does the remain- ing sol, unless complete precipitation has occurred, i.e. the precipitate on either side of the optimism does not contain the whole of the component which is in defect. This distinguishes it from ordinary chemical reactions. In fact, if the positive and negative sols are standard- ised by means of barium chloride (the change of titre on precipitation, p. 104), complete precipitation is found to occur when the volumes are inversely proportional to the titre, which is directly proportional to the electric charge. Henr?2 had already tried the effect of two sols of the same electric sign upon each other. As would be antici- pated from the protection effect, they not only did not » Sitzungsber. Kaiserl. Ahad. Wiss. Wien., 1904, 113, 1159. ' Compt. rend. Biol, 1904, 55, 1666, MUTUAL PRECIPITATION OF SOLS 135 precipitate each other, but the mixed sol acquired the stability of the more stable component. On either side of the optimum mixture, the sol is negative or positive, according as one or other component is in excess, although the sol still contains both ; and this holds whether partial precipitation has taken place or not. The following observations of Billiter ^ illustrate this. ■'a re(0H)3 sol + AsgSg sol. 10 0.0. of tlie mixture contaiu Observation. Fe,03(mg.). As,S3(mg.). Optical. Kataphoresis. 0-61 20-3 opalescence to anode 6-08 16-6 immediate precipitation „ 9-12 14-5 complete „ none 15-2 10-4 immediate „ to kathode 24-3 4-14 slight opalescence „ 27-4 2-07 no change „ The curve Kg. 10 is of the same form as that for the action of zinc sulphate on albumin (p. 119). Emulsoids. — So far we have considered mainly the action of suspensoids on suspensoids. When we come to emulsoids, their behaviour, as might be anticipated from their electrical properties and behaviour on precipitation, is not so simple. The mutual precipitation of positive and negative gelatine (and other similar emulsoids) during kataphoresis has been referred to already (p. 83). Albumin (electrically neutral) is precipitated by basic emulsoids, as histone, and basic dyes, because in them the albumin becomes negative, while they are positive sols. Similarly, it is precipitated by acid emulsoids, as silicic acid sol, and acid dyes, in which it becomes positive, while the acids are negative. ' Zeitsch. physilcal. Chem., 1905, 51, 142. 136 GENERAL PROPERTIES OF COLLOIDS Similar reactions occur with tannin and gallic acid,' which precipitate both albumin and gelatine. Suspensoids and Emulsoids. — The action of emulsoids on suspensoids has been described in the previous section. As there stated, the protection stage is preceded by precipitation, when the emulsoid is added slowly, or in small amount. The reason is now evident. It is also clear that there may be a balance between the precipita- tion and protection effects. Here, possibly, the acid nature of gelatine and its protective power together more than balance the precipitation effect. On addition of FiQ. 10. ammonia to the gelatine, which makes it more negative, the ferric hydroxide sol is precipitated ; if instead, the ammonia is added to the mixture of sols, no precipitation occurs, but the mixed sol migrates to the anode.^ Similar anomalies have been observed with albumin.^ which is precipitated by both positive and negative sols. The essential feature, then, of the action of a positive 1 Dreaper and Wilson, /. Soc. Chem. Ind., 1906, 25, 515, « Sitzimgsber. KaAserl. Ahad. Wiss. Wien., 1904, 113, 1159, * Priedemann, Zeitseh. physikal. Chem., 1905, 61, 146, MUTUAL PRECIPITATION OF SOLS 13; sol on a negative sol, is that excess of either sol protects, while precipitation only occurs in a middle zone, which is usually narrow (Fig. 10). Action of Multivalent Ions on Sols. — This takes us back to the remarkable phenomenon observed by Burton in the action of Al'" on gold sol and silver sol (p. 107). A very small concentration of Al"" does not precipitate the sol, at slightly larger concentrations precipitation is complete, while with excess of Al"" no precipitation occurs. The original sols are negative ; they pass through the isoelectric point, and then become positive. This is very similar to the protection of sols, which can also precipitate each other. In each case, there is a middle zone of precipita- tion, and on either side of it stable sols, but of opposite electric charge. Other multivalent kations behave like Al"". Now these salts are hydrolysed, and thus contain positive hydroxide sols. Their action on negative sols may therefore be due to protection at the two extremes, with mutual precipitation at intermediate concentrations. As Burton showed, the greater the excess, the greater the (reversed) charge and the stability. A new effect, however, must now come in : the precipitating effect of the anion on the now positive sol. The second region of non- precipitation is accordingly followed by a second zone of precipitation, which is finally complete. The whole series of changes on successively increasing the hydrolysed salt is : (1) first zone of non-precipitation ; (2) first zone of precipitation ; (3) second zone of non-precipitation ; (4) second zone of precipitation (Fig. 11). 138 GENERAL PROPERTIES OF COLLOIDS The following illustrate this * : — Mastic sol + A]2(S04)3 0-008-0033 0-083 0-167 0-33-0-83 1-67-33-3 Precipitation. none slight complete none complete -Complete %• Mastic +Al2(SGf.)^ ■Complete No. ^ Platiruun+Fe CI3 - X X X )( — (3-8) -0-5 +0-5 +7 Log of Concentration. Fig. 11. The accompanying change in kataphorcsis is shown below 2 — Platinum sol (Bredig) + FeClg FeCl, (millimols in 1 litre). Precipitation. Kataphoresis. 0-021-0-056 none to anode 0-083-0-222 complete none 0-333-6'67 none to kathode 16-3-668-7 complete none Fig. 11 illustiates this behaviour. ' Friedemann and Neisser, Milnch. medizin. Wochenschrift, No, 11, 1903. ' Buxton and Teague, Zeitsch. physikal. Chem., 1907, 67, 7G. MUTUAL PRECIPITATION OF SOLS 139 That the second precipitation zone is due to the action of the anion of the salt on the reversed sol is home out by the above tables. In each case, the first precipitation zone begins at 0'083 m.mol Al"* and Fe"', and ends at 0333 m.mol ; in the former the second precipitation, zone commences at 1'67 m.mol Al"", vrhile in the latter it is ten times as great, 16'3 m.mol Fe"*. In the former the anion is SO4", and in the latter 01' ; the difference is in agreement with the valency rule. This phenomenon is easily seen in the case of dyes, for they comprise all three cases : (a) true solutions, as eosin and methylene blue, which act on sols in the same manner as other electrolytes ; (6) most dyes exert pro- tection, i.e. there is a precipitation zone, with a non- precipitation zone on each side of it. Frequently the middle zone is very sharply defined and within very narrow limits ; (c) Nile blue and mastic sol behave in a similar manner to aluminium salts and mastic. As we have seen, the behaviour oi albumin with zinc salts and salts of heavy metals is very similar, the curves being of the same kind (c/. Figs. 7 and 11). It has been suggested that salts of multivalent ions possess greater precipitating power because of the hydro- lysis, i.e. that the precipitation is really produced by the hydroxide sol, and not by the ion, but this is not the case. The charge on a sol is very much less than on an equivalent amount of the corresponding ion (p. 82), and a correspondingly larger amount of the former will be required. Biltz showed that 24 mg. of arsenious sulphide required 2 mg. AI2O3 as hydroxide sol for precipitation, while, as Al"', 0-13 mg. AI2O3 were sufficient. Even complete hydrolysis, which is far from being the case, could not give the required amount of sol. It is not probable that the increased positive charge on the I40 GENERAL PROPERTIES OF COLLOIDS sol will be increased to a sufficient extent by the presence of H", to make up for such large differences in amount. Proof that the relative amounts of positive and negative sols for optimum precipitation vary with the dispersity is afforded by the recent experiments of Galecki and Kastorskij.^ The ferric hydroxide sol contained amicrons only, and was opalescent, its concentration was 10"28 mg. in l.c.c. and its kataphoresis velocity was + 35-7 X 10"^ (cm. sec. volt cm.). Of the two gold sols, the one, Au/, prepared by reduction with formaldehyde, contained sub- microns ; its concentration was 0-07 mg. in 1 c.c. and its velocity was — 39*3 X 10~^. The other, Au^, prepared by reduction with phosphorus, was of the same concen- tration, and had almost the same velocity ( — 36"2 X 10~^), but contained amicrons only. On the average 1 mg. Au/ = 4-98 mg. re203 and 1 mg. Aup = 18-4 mg. Fe^O^ When the amount of gold sol added was insufficient to cause any precipitation, the Au/ sols contained amicrons only, while Aup at iirst cleared up the opalescence of the ferric hydroxide, but submicrons appeared later. The following figures require no further explanation. PejOj. Au. -Velocity (cm. (in mg.) A%. 0-7 — 39-3 sec. volt cm.). . AUj,. -36-2 Optimum for precipitation (as above), 61-68 0-28 + 30-1 92-52 007 + 33-4 102-8 0-0 + 35-7 + 22-77 + 26-20 + 35-7 1 Koll. Zeitsch., 1913, 13, 143. CHAPTEE XI THE PEOPEETIES OF GELS Gels are usually obtained when emulsoid sols are cooled or evaporated down ; when suspensoids are treated similarly, the disperse phase is usually obtained as a loose powder. The former contain both disperse phase and dispersion medium, and form two fairly distinct types : the gels which retain their elasticity and coherence on drying, and those which lose their elasticity and become powdery on drying. The sol-gel reversible transformation has already been dealt with (p. 121) and will not be again referred to ; there still remain the two important subjects of imbibition and hydration and dehydration of gels, and the structure of gels, including the relation between amorphous and crystal- line solids. But before these are discussed, it may be useful to present a summary of the general properties of gels. As emulsoids are generally recognised to be two-phase liquid systems, it seems reasonable to conclude that gels are also two-phase liquid systems with extremely great viscosity, since there is no apparent decrease in the number of phases on cooling or evaporating down emulsoid sols. As the viscosity of emulsoids indicates, there is in emul- soids a gradual and continuous transition from sol to gel (p. 122). This conclusion was reached empirically by Biitschli (1892 onwards) as a result of microscopic obser- ' rations on the structure of gels of starch, gum, gelatine, and 142 GENERAL PROPERTIES OF COLLOIDS other substances. Many of the characteristic properties of gels are connected with this two-phase structure. Density. — There is always contraction when a dry gela- tinous colloid and water unite to form a gel ; but the gel considered alone undergoes considerable expansion when it takes up water. This initial contraction on imbibition has long been known, and was accurately determined by Liidekingi in the case of gelatine. The density of the sample of gelatine he used was 1'412, and the density calculated from the density of gels of 14 per cent, and 35 per cent, gelatine was 1"9. The actual figures are— Concentration d. d. g. in 100 g. of gel. observed. calculated, 13-6 1-069 1-041 32-0 1-135 1-103 58-5 1-242 1-206 This behaviour of gels is in agreement with that of emul- soid sols (p. 21), and is further evidence of the similarity of their nature. Thermal Exfpansion. — The thermal expansion of gels is practically the same as that of the water or other liquid which makes up the greater part of the gel. A gelatine gel becomes warmer when it is stretched, and cooler when it is compressed ; this is also the case with rubber. Compressibility. — The compressibility of gelatine gel is much higher than that of solids,^ j3 = 10 X lO"-^^ instead of about 1 X 10-^ in the case of solids. The compressi- bility rises as the temperature is raised, and when the gel becomes sol, j3 = 48 x 10~^, which is the value for water. It is not improbable that the high value of /3 for the » Wled. Ann., 1888, [iii.] 38, 552. 8 Barus, J. Amer. Cliem. Soc, 1898, 6, 285. PROPERTIES OF GELS 143 gel is due to the phenomenon being the sum of two effects, the compressibility of the dilute liquid phase which forms the larger part of the gel, and the elasticity of the more concentrated liquid phase which forms the containing walls of the gel. Elasticity and Viscosity. — The modulus of elasticity of a cylinder of gel is given by P^ E,=- P = weight attached to the cylinder to cause extension. AJ = the increase in the length I. r = radius of the cylinder. This is the modulus of elasticity for extension, the modulus for shearing strain is E.=. ^' 2(1 + /.) fi, being the ratio of the relative contraction of the diameter to the relative change in length. The value of n for gela- tine is 0*5, hence there is no actual change in volume when a gelatine gel suffers extension, even when the extension is great. The value of Ed in gelatine gels increases as C^, the square of the concentration of gelatine. The relaxation time increases much with increase of concentration, being 10 minutes for a 10 per cent, gel, and 40 minutes for a 20 per cent. gel. Since Ej X relaxation time = viscosity, the viscosity of gels must show enormous increases with increase of con- centration. The addition of other substances to the gel affects the elasticity just as it affects gelation (p. 123). Sulphate and polyhydroxyl organic compounds, as sugar, glycerol, act 144 GENERAL PROPERTIES OF COLLOIDS in the one direction (increase of E), while chloride and nitrate have the opposite eifect. This is as might be expected : the substances which increase the elasticity of the gel favour gelation, the others favour solation. The viscosity must also be affected by the addition of substances to a gel, as they affect both the modulus of elasticity and the relaxation time. Optical Properties. — Double Refraction on Mechanical Deformation. — This is another property which gels have in common with emulsoid sols, and as it was not referred to in describing sols, some detail may be given here. It is well known that glass becomes doubly refracting when subjected to strain, such as unequal pressure. As glass is a super-cooled, highly viscous liquid, it occurred to Kundt ^ that other highly viscous liquids, which yet have the ordinary properties of liquids, ought to show the same effect. The apparatus consists essentially of a massive cylinder rotating coaxially inside a larger hollow cylinder, which is stationary. The liquid is placed in the annular space and is thus subjected to the mechanical deforma- tion. Polarised light is sent through the liquid and the emergent light is analysed by a suitable eyepiece. No double refraction' was observed with cane sugar solution or glycerol, but it was produced in even very dilute sols of gum, gelatine, collodion, etc. The effect therefore cannot be due to ordinary viscosity, as the glycerol and the sugar had much greater viscosities than the sols. It most probably arises from the peculiarity of emulsoids and gels in being two-phase liquids, the inter- face between the phases of which is possessed of elastic properties, to which also the peculiarities of the viscosity of emulsoids is also almost certainly due. The double refraction, D = w — '^o' ^^ ^'^ difference » Wied. Arm., 1881, [iii.] 13, 110. PROPERTIES OF GELS 145 of the refractive indices of the extraordinary ray and the ordinary ray, and is given by a where A is the wave-length of the monochromatic light employed, A the difference of phase, and a the thickness of the plate or layer of gel. D is proportional to Al/l, the relative lengthening, and this ratio may be taken as a measure of the effect. It is approximately proportional to the concentration of the gel, and salts have the same effect on it as on the modulus of elasticity, i.e. chloride and nitrate diminish it, sulphate has practically no effect.^ Gelatine is optically active; chloride and nitrate diminish the rotation, while sulphate does not alter it. Diffusion in Gels. — The early observations on the diffu- sion of solutions in gels have already been referred to at some length (p. 24). It may be recalled that Graham concluded that the rates of diffusion of salts in gelatine gels were the same as in pure water, but that his results did not really bear him out in this, the differences he found being greater than his experimental error. Others, too, arrived at the same conclusion for salts in other gels, e.g. in silicic acid and agar. It was demonstrated later that the differences are not large so long as the gel is dilute, but become much greater in concentrated gels.^ Substances which influence the sol-gel transformation and the elasticity of the gel, also affect the permeability of the gel by other substances. Alcohol and glycerol. 1 Lieok, Arm. Physik., 190i, [iv.] 14, 139. ' Beohhold and Ziegler, Zeitsch. physihal. Chem., 1906, 56, 105. L 146 GENERAL PROPERTIES OF COLLOIDS reduce the diffusion in gelatine or agar, -while urea, chloride, and iodide increase it. Liesegang's Rings. — This complicated phenomenon has given rise to much discussion. A glass plate is coated with gelatine sol containing ammonium bichromate ; when the gelatine is set and partly dry, a drop of silver nitrate solution is placed in the middle of the plate, and the plate kept in a horizontal position in a moist atmosphere, to prevent evaporation. The silver nitrate diffuses out from the drop, and silver chromate is deposited in con- centric rings at ever increasing distances from each other. Ostwald^ explained their formation as an effect of supersaturation : a certain excess of Ag* is required to produce the supersaturation in the metastable region, which will necessitate the formation of nuclei of the precipitate. When this has happened, the whole of the excess of silver chromate will separate out, and so the adjoining region becomes poorer in chromate ion and the silver ion has further to go before the next crystallisa- tion can take place. There are other factors, however, some of which have been indicated by Liesegang, and by Bechhold.2 In the first place, there is the protective action of emulsoids on the precipitation of suspensoids, and this is exceptionally strong in the case of gelatine (p. 129). It is thus probable that silver chromate will remain as a suspensoid sol long after the metastable concentration (if it exists at all) has been passed. Ostwald's explanation does not explain why the centre of the plate, where the drop of silver nitrate is put on, always remains clear, the first deposit being a ring near I Lehfhiich, 2, II. p. 778. » Zeitsch. physikal. Chem., 1907, 59, iii ; 1905, 52, 185, PROPERTIES OF GELS 147 the margin of the drop ; for it cannot be owing to want of Ag'. Nor does the protective action of the gelatine afford an explanation, unless the question of time is introduced. The clear centre may, however, be a result of the protect- ing action of an excess of either ion in the precipitation of sols of insoluble silver salts, e.g. in the titration of silver by chloride or of chloride by silver, the silver chloride remains as sol so long as there is excess of the ion to be precipi- tated, and it is precipitated only when there is no excess of either ion (p. 110). It may well be, then, that the very reverse of Ostwald's explanation takes place, and that precipitation only occurs when the amounts of silver ion and chromate ion are equivalent. There is still another effect which may exert a power- ful influence, viz. the permeability, not merely of the gelatine, but of the ring of precipitated silver chromate by salts. This itself is a subject of great complexity, and the few generalisations which have been made out will be discussed below. Only it may be said here that a gel consisting of a semi-permeable precipitation film is, as a rule, impermeable by the ions which make up the film, otherwise a thin continuous film would not be produced by the reaction. Accordingly, if the silver chromate is first of all precipitated as a gel or a two-phase liquid, it will certainly not be without effect on the diffusion of silver ion from the centre to the region beyond the ring last deposited. It may even cause decomposition of the silver nitrate, the silver ion being retained while the nitrate passes through, thus clearing out the chromate and causing silver ion to diffuse further before the next ring can be formed. It might be inferred from the above, that no silver ion could get past the first ring ; but this is not the case, for the precipitated silver chromate gel will, sooner or later (and probably assisted by the excess of silver ion), 148 GENERAL PROPERTIES OF COLLOIDS break down, either by crystallisation or other diminution in dispersity, and will thus lose its impermeability. This does not exhaust all the possibilities, but enough ' has been said to show that the phenomenon is by no means a simple one, and that a full and satisfactory explanation is not yet possible. Precipitation Films. — The properties of precipitation membranes, which came into prominence during the development of the osmotic pressure theory of solution, have been the subject of many investigations, without any finality having been reached. The structure of the membrane must be an important factor in the behaviour of any particular film, and this must depend to. a large extent on the conditions under which it is precipitated, and these again must largely affect the stability of the form in which the precipitate first separates out. As a rule these membranes are unstable, and liable to change, to deteriorate in impermeability, under influences which are mainly unknown ; hence it is not surprising that the results of investigations on them are frequently not in close agreement. The state in which a precipitation film first separates from solution, and its subsequent changes, are indicated by von Weimarn's Theory (p. 174). The more insoluble the substance, the easier it is to obtain it in the form of a cellular jelly, which will under suitable conditions form a coherent semi-permeable membrane. The more dilute the reagents, the more likely is the precipitate to separate out in a crystalline form, or, if it is very insoluble, as an incoherent powder or open network. The permeability will vary with the structure of the film, and as this is often much altered by the action of salts (ions), the film may change very rapidly when salts are present, although it had previously remained unchanged PROPERTIES OF GELS 149 for long periods. These changes may be either reversible or irreversible; if the former, it may be very difficult to detect the change. The permeability of a film by substances may be connected with the molar weight of the diffusing substance, or with the influence of the substance on the nature of the membrane, increasing or decreasing its elasticity, which, as we have just seen, affects the diffusion of substances in gels. There is also the possibility of irreversible coagulations or other changes brought about by the action of ions on the colloid films, which may act in the direction either of diminishing or increasing the permeability, and may even produce the one effect at low concentrations, and the reverse at higher concentrations. Notwithstanding this, a few general conclusions have been arrived at concerning such semi-permeable mem- branes. No one now maintains that the permeability is connected with the size of the pores in the film, i.e. that the films function as molecular or ionic " sieves." It is much more likely that the principal factor is the lyotrope character of the particular solute. Thus acids and alkalies, chlorides, bromides, iodides, nitrates, and chlorates are more or less able to pass through most precipitation films ; sulphates, phosphates, carbonates, salts of organic acids, salts of bivalent kations as Ca, Mg, Zn, etc., and salts of heavy metals are less so, and often cannot do so at all, whUe.polyhydroxy organic compounds also come into the same class. This grouping seems to point to lyotrope influences, especially the classification of the organic substances, though the effect of salts might also point to a valency influence, as in the precipitation of sols (p. 98). Adsorption must likewise play a part in the phenomena, and along with it the- mutual precipitation of positive ISO GENERAL PROPERTIES OF COLLOIDS and negative sols. The nature of the semi-permeable membrane may be temporarily or permanently changed in consequence of either adsorption or of the precipitation of other substances on the membrane. Hydeation and Dehydration of Gels Non-Elastic Gels The question of the composition and chemical nature of the hydrates of metallic oxides, such as ferric oxide, man- ganese dioxide, and stannic oxide, has always been a difficult one, and there is little doubt that many such " compounds '' still find a place in the dictionaries, in spite of the evidence of their indefinite nature. For instance, J. M. van Bemmelen,' to whom is due a great part of our knowledge of this subject, showed as long ago as 1878 that the amount of " hydrate water " of ferric oxide was variable, an accidental amount, and that the substance continuously lost water in dry air, and gained it in moist air, the amount depending, among other things, on the temperature and the moisture. The following figures show this : — Mols of Water to 1 of Oxide. State of gel. t. SiO^. SnOj. Al^O^. Fe,0„ Just dry 15° 44 3-0 5-6 >6 In air 15° 1-5 2-6 4-4 5^ In dry air 15° 0-5 1-0 2-6 1-6 In air 100° 0-2 1-0-6 2-3 1-2-1 The vapour pressure curves show decisively that the solid is not a definite hydrate, a chemical individual, for according to the phase rule a hydrate, which, on giving up some of its water as vapour, is in contact with a lower » Ber., 1878, 11, 2232 ; 1880, 13, 1466. HYDRATION & DEHYDRATION OF GELS 151 hydrate, or the anhydrous compound, forms a univariant system (two components, three phases), the vapour pressure of which is constant at constant temperature. As this is not the case, there can be only two phases, the gas phase and one solid phase, in which the vapour pressure varies with the temperature and the concentration in the phase, just as in the case of an unsaturated solution. The water in the solid phase (from the phase rule point of view, it is immaterial whether the phase be regarded as solid or liquid) is therefore not in the form of a chemical compound, but is adsorbed or dissolved. Real compounds are formed under certain conditions, e.g, BeO,H20, and Al203,3H20, which separate out in a crystalline state with a definite composition, and behave as true compounds. Hydration and Dehydration of Silicic Acid Gel. — The system Si02,H20 has been systematically investigated by van Bemmelen,^ who determined the isothermal at- 15°. The gels were obtained by the action of dilute hydro- chloric acid on sodium silicate, a little ammonia being added ; the gels were then well washed. Gels of various concentrations were obtained by varying the concentra- tion of the sodium silicate. The results were the same, whether the gel was in powder, in grains or in clumps. The gels were placed in desiccators containing sul- phuric acid of thirty-six different concentrations (from H2SO4,0-25H2O down to H2SO4,90H2O). The amount of water was ascertained by weighing at stated intervals of time. The results are given in Fig. 12 for a gel which originally contained IOOH2O to lSi02. The temperature was 15°. The dehydration follows the curve AOOiOq, which is made up of two similar curved portions, separated by an intermediate linear portion. From A to the loss of » ZeiUch. anorg. Chem., 1896, 13, 233. rS2 GENERAL PROPERTIES OF COLLOIDS water is a continuous process, but is in no sense an equili- brium, for thougb at any point the system can take up water, it does not follow the curve, but strikes out towards the saturation pressure (indicated on the diagram) which it reaches in a very little time. This latter process is more nearly reversible, as subsequent dehydration follows a slightly divergent curve to reach the curve AO near the point at v^hich it left it. At the middle portion of the curve, OOi, is reached ; it is approximately a horizontal straight line, indicating that large quantities of water can be removed without making much difference in the vapour pressure. This part 15 r Vapour Pressure of Water //. 0-5 1 1-5 2 2-5 3 35 * Concentration, of Water in Gel. Pie. 12. of the curve corresponds to a curious change in the appearance of the gel, which up to this point had been clear and transparent. At O it becomes opalescent, "com- mencing in the interior and gradually spreading through the mass, until it is yellow by transmitted light and bluish by reflected light. These colours suggest the appearance of a new highly disperse heterogeneity, and this proves to be the case. The opalescence is caused by bubbles of air or of water vapour appearing where there had previously been liquid water. A similar appearance has been observed ■yvith many other gels and transparent substances with HYDRATION & DEHYDRATION OF GELS 153 very fine porosity, as meerschaum. As dehydration pro- ceeds the gel becomes white as porcelain, and then opaque. At Oj the reverse series of changes commences, the gel becoming white, then opalescent, and finally clear as glass. Between and Oj the reaction is irreversible ; at any point the gel can take up water, but it does so, as before, by a short cut to saturation. As before, this hydration is approximately reversible, the system returning to a point on OOi. From Oi the curve OiOg is a continuation of AO, but differs from it in the reaction being reversible, both hydra- tion and dehydration being represented by the same curve. Hydration of the Dry Gel. — The gel dried over con- centrated sulphuric acid still has about 0'2H2O to SiOg. Starting at Oq the hydration follows the dehydration curve as far as Oj from which point it lies above OiO ; this portion O1O2 is not a straight line, like OjO, but is much flatter than the rest of the hydration curve. The process is not reversible along O1O2, the dehydration follows the paths shown on the diagram. The characteristic opal- escence and opacity also make their appearance in this region, but not to so marked an extent as on dehydration. From Og, which is about 0'2H2O above 0, and the vapour pressure 2-3 mm. higher, the third portion of the curve O2O3 begins; the curve ends in O3 at the vapour pressure of water. Dehydration from O3 does not exactly follow the hydration curve, but comes down to if the gel is far past Og. If it is near Og it comes down to an intermediate point on 00|. All these are indicated on the diagram by the direction arrows. Silicic acid gels are unstable systems of the general formula SiOgjWHaO, the value of n depending on the physical conditions and also on the previous history of 154 GENERAL PROPERTIES OF COLLOIDS the system. The latter, known as hysteresis, is of great importance, not only in this particular property of gels, but in many of the properties of colloids generally. The hydration and dehydration of other non-elastic gels, as ferric hydroxide, stannic acid, etc., show the same charac- teristics as silicic acid gel, although there are many individual differences in each case. Elastic Gels The differences in the hydration phenomena of elastic gels, of which gelatine or agar may be taken as the type, and the non-elastic gels are sutBciently great to warrant their separation, although they have much in common, and the groups merge into each other. Perhaps the most outstanding differences are the absence in the elastic gels of the middle portion of the dehydration curve (OOj, Fig. 12), and the fact that an elastic gel which has taken up all the water it can from water vapour, can still take up large amounts of liquid water, much more indeed than it has already taken up from the vapour. As with non-elastic gels, there is no evidence of chemical combination, no fixed ratio of solid to water which is reproduced under varying external conditions. The amount of water imbibed depends on many con- ditions : the temperature, whether the water is liquid or vapour, the presence of other substances in the water, and the previous history or treatment of the gel. The process of imbibition is accompanied by con- traction, if the total volume of the gel and water is considered, but in many cases the solid experiences a large expansion, and can then give rise to large pressures. This and the other phenomena of imbibition are clearly very important from the biological side, as many of the tissues of plants and animals possess the power in a HYDRATION & DEHYDRATION OF GELS 155 marked degree. Other elastic gels are starch, agar, pro- teins, gelatine. One difficulty of investigation is that the substances examined are not pure, many of them are mixtures, and most of them are easily decomposed; in addition to which equilibrium is often very slowly reached. Hofmeister considered imbibition to be made up of three distinct processes : capillary, endosmotic, and mole- cular imbibition. The last, which is of the nature of adsorption in its not being chemical combination, is what is now ordinarily meant by imbibition. He arrived at the following generalisations. The quantity of liquid which can be taken up reaches a maximum, which depends on the gel, on the liquid, on the temperature, and on the viscosity of the liquid. There is always contraction, and always evolution of heat; from which, according to le Chatelier's theorem, heat will hinder or diminish imbibi- tion, while cold and pressure will favour it. Weight and Volume Change. — A plate of dry gelatine weighing 0'904g.i was placed in an atmosphere saturated with water at ordinary temperature. The gain in weight in eight days was 0"37 g. and no further gain took place in twenty days. When the plate was immersed in water at ordinary temperature, it took up 5 '6 g. of water in one hour, and still more on prolonged soaking. The one maximum is about 50 per cent, of the dry weight, the other is 1000 per cent. This extra water is easily given up to dry air, or even in moist air. The greater part of it is so loosely held that the gel has the same vapour pressure as water, and under constant conditions loses water at the same rate as a similar surface of water ; only a small residue is more firmly held, which cannot be removed at all at 15°, but is slowly given up at 95°. » Schroeder, Zeitsch. physihal. Chem., 1903, 45, 109. IS6 GENERAL PROPERTIES OF COLLOIDS The volume change is easily demonstrated, if a 50 c.o. flask with long narrow neck is filled loosely with shreds of gelatine, water added up to a mark on the neck, and the stopper inserted. The level of the water falls as imbibition proceeds. The temperature must be kept con- stant, as heat is evolved during the reaction. If gelatine is hardened by alum or alcohol or formalde- hyde, it largely loses its power of imbibition, becoming more like the non-elastic gels. Imbibition also depends on the liquid, 6.g. alcohol is not imbibed by gelatine. There is no imbibition by starch in petroleum, or by albumin in alcohol, benzene, ether, or turpentine. This alone is sufficient to mark off imbibition from adsorption, for there would be some adsorption with these liquids, although probably not so large as with water. There must be some connexion between imbibition and the solubility of the liquid in the gel, as the liquid must pass through the walls of the gel to dilute the liquid in the interior. Further examples of imbibition are rubber in ether, chloroform, or other organic liquids, and pyroxylin (collodion) in ether or alcohol. The same behaviour is observed here as with gelatine and water. To make a gelatine sol from dry gelatine and water, it is much quicker to soak the gelatine in cold water, and to apply heat only when imbibition is fairly complete ; hot water and dry gelatine form a sol very slowly. Similarly with collodion (see p. 27): it is soaked in either alcohol or ether, and after imbibition has taken place, the other liquid is added, when solation is rapid. The Pressure of Imbibition. — As the total volume change is a contraction, the pressure due to the expansion of the gel by water is not observed at all unless the gel is separated from the water by a semi-permeable membrane. HYDRATION & DEHYDRATION OF GELS 157 If gelatine is packed in a porous earthenware cylinder, connected with a manometer, and placed in water, the pressure bursts the cylinder.^ Eeinke determined the pressures by means of the oedometer. Circular discs of dry seaweed (Laminaria) were placed in the bottom of a steel cylinder, the piston of which was pierced by a number of very fine channels, which allowed free passage of water to or from the sea- weed (the latter really provided the semi-permeable membrane, the steel piston only acting as a strong support, as when pressure was applied to the piston, the holes would be closed by the outer membrane of the seaweed). Weights were placed on the top of the piston, and the height to which the piston was raised was measured. As , the discs increased only in thickness and not in area, the height gives the volume and hence the amount of water which has been taken up. Ten layers of Laminaria, 0"1 mm. in thickness, and 50 sq. mm. area, gave the following pressures : — W (per cent. H^O in vol. per cent, of air- P (atmospheres). h (mm.). dry substance). 1 3-3 330 3-2 2-05 205 7-2 0-97 97 21 0-35 35 41 0-16 16 These pressures are equilibrium pressures, for similar values were obtained for the pressures at which water was just pressed out from seaweed containing varying amounts of water. The pressures required to remove water when the amount present is very small, become enormous. Similar forces come into play when elastic gels are dried in other ways : e.ff. glass plates coated with gelatine and 1 Schroeder, loc. cit. 158 GENERAL PROPERTIES OF COLLOIDS then dried often become double-refracting, owing to the strain, and if the glass is thin it may, become very curved. Similarly, if an elastic gel in a glass or porcelain dish is dried up in a desiccator over sulphuric acid, or in very dry air, it not infrequently happens that the surface of the basin is torn away with the gel as it contracts. Heat of Imbibition. — When a gel takes up water heat is given out, as the following figures show : — csl. per g. gel. Gelatine 5*7 Starch 6'6 Gum arable 9'0 „ tragacanth lO'S Now the solation of a gel absorbs heat, as heat is evolved when emulsoids are coagulated; hence it is difficult to arrive at correct values of the heat of imbibi- tion. Experiments on starch with known amounts of water show that the greatest amount of h^at is evolved for small amounts of water, as is only to be expected, since only a small fraction of the water is difiBcult to remove, and requires a high temperature to do it (p. 155). Velocity of Imbibition. — Eeinke used the oedometer, by means of which he measured the pressure of imbibition, to determine the velocity of the process. Since the discs of seaweed do not change in area, only in thickness, it is easy to find the velocity by determining the change of h with the time, the pressure on the piston being constant and the amount of gel being known. If Wf is the amount of water taken up in the time t, and w^ the maximum amount which can be taken up under the conditions, we have — dw , , , or k = -log ^ — HYDRATION & DEHYDRATION OF GELS 159 The table contains the results of such an experiment. t (min.). w. fe. t (min.) w. h. 1 40 0-108 16 306 0-095 4 121 092 20 345 0-106 8 201 0090 24 372 0-121 12 258 0090 00 392 Hofmeister ^ determined the velocity of imbibition of water by thin plates of agar and gelatine (0-5 mm. in thickness) by weighing the plates at stated intervals of time. As the weighing took some time, the times are not very exact. The same formula holds for his results, e.g. Gelatine Plate. t (min.). 5 10 15 'W (g-)- 3-08 3-88 4-26 0-090 0-084 0-084 t (min,). 20 25 oa ■u) (g.). 4-58 4-67 4-96 k. 0-064 0-075 Imbibitioa in Solutions. — Hofmeister 2 extended his investigations on imbibition to salts in water, in order to find the distribution of the solute between the gel and the solution. He found, however, that both the velocity of imbibition and its amount varied considerably with the nature of the solute. He used thicker plates of gel, so that maximum values were not obtained. The plates were soaked in solutions of salts and of indifferent substances, the concentrations varied from 0-5 m to 4 m. Remarkable differences were observed, the imbibition in some solutions being as much as five times as great as in other solutions of eq^uivalent concentration. Arranged in order of increasing effectiveness in pro- moting imbibition, the series is — ' Arch. exp. Physiol. Path., 1890, 27, 395. ' Ibid., 1891, 28, 210, 238. i6o GENERAL PROPERTIES OF COLLOIDS sodium — sulphate, tartrate, citrate, acetate; alcohol, glucose, cane sugar ; (water) ; chlorides — potassium, sodium, ammonium ; sodium — chlorate, nitrate, bromide ; to which might be added iodide and thiocyanate, both coming after sodium bromide. The substances before water hinder imbibition, less water being taken up in the same time, and also less water taken up in the maximum, while the salts after water favour imbibition. The order is again the lyotrope series which" we have come across so often. As already seen, the modulus of elasticity is affected by salts in the lyotrope order ; this is the same series in the reverse order. As was to be expected, a substance which renders a gel more easily deformed (reduction of E) will favour imbibition. There is a still closer connexion between the effect of salts on gelation and imbibition, so that the curves representing the effect of any salt on these processes are practically reciprocal, i.e. what favours gelation hinders imbibition, and conversely. When a gel undergoes imbibition in a solution, it takes up both solute and solvent, but in different amounts which depend to different extents on the concentration of the solution. The water taken up increases with the con- _ centration of the solution to a maximum (at about 14 per cent, solution of NaCl), and then decreases slowly as the concentration increases. The amount of salt taken up increases with the concentration, and always approximately proportionally to the concentration. In the presence of salt (NaCl), the amount of water imbibed exceeds that taken up in pure water, the increase being from 0'2 to 18 per cent, Not only is the amount larger, but the velocity is also increased. In a 5 per cent. STRUCTURE OF GELS i6i solution of NaCl, the total amount was greater, and the rate much greater. The action of acids and bases on the imbibition by gelatine is similar to the action of salts. At small con- centrations (0'005-0"01m), imbibition is less than in pure water, reaches a minimum and increases to a maximum at 0*025m, both the rate and the amount being greater ; the latter is 3-4 times as great as in water. The Structure of Gels As emulsoid sols are two-phase liquids, and as gels are obtained from them by cooling or by evaporation, the conclusion is natural that gels also possess the same structure as emulsoids : two solutions, the one, more concentrated, forming the enclosing walls of the cells which contain the other, more dilute solution. The only difference is that the one solution, if not both, is more concentrated than before, and thus confers the greater viscosity and rigidity. The structure then is cellular, the walls being a highly viscous solution enclosing a more dilute solution. This is confirmed by observations in many directions, at least so far as the existence of the cell walls is con- cerned, especially by J. M. van Bemmelen, and above all by Biitschli, who closely examined the microstructure of a series of precipitates and gels. Now all these micro- structures are more or less after the " honeycomb " pattern, and several other observers have confirmed Butschli's observations. This structure is recognisable not only in plant and animal cells and tissues, but also in gelatine precipitated with tannin or alcohol or chromic acid, and in albumin precipitated by heat or by salts, etc. Hardy detected it in albumin coagulated by heat, or by mercuric chloride, bichromates, or thiocyanates ; in silicic acid gels, M i62 GENERAL PROPERTIES OF COLLOIDS and in gelatine precipitated by mercuric chloride, formalde- hyde or chromates. The albumin structures are always open network (the mesh is 0*5-1"5jli) ; gelatine, if from sols of over 7 per cent, concentration, forms closed nets, and open network below 7 per cent. Pauli, and others with him, have expressed doubt as to this structure, maintaining that it is not pre-existent in the original sol or gel, but is a secondary structure of the altered gel. He concludes that only genuine coagulations show this microstructure. According to von Weimarn, these structures arc secondary, but the reasons which lead him to this conclusion are widely different from Pauli's objections. To quote : — " There are, however, precipitates so fine-grained that in the microscope the form of the grain appears to be round, owing to diffraction and other phenomena. Ob- viously the crystalline nature of such precipitates must be proved in other ways, since the microscope fails on account of the fineness of grain." "These microscopically amorphous precipitates exhibit various characteristic structures — network, honeycomb, rayed, dendritic, and the like — under the microscope, un- less the preparation consists of a single layer of detached grains. All these structures are of secondary formation, the only original structure of these precipitates is the flvs-grained crystalline structure." To avoid these secondary deformations special pre- cautions must be taken, both in collecting specimens for observation, and in handling them. Object glasses are placed at the bottom of the vessel before the precipitate is formed, and are removed successively when a very slight trace of the precipitate has formed on them ; they are ex- amined at different magnifications. The glasses are STRUCTURE OF GELS 163 replaced by others. In this way some of them will contain only one layer of detached particles, which at suitable magnification will be found to be crystalline. If the precipitate settles quickly in large quantity, small portions must bQ withdrawn from time to time from different parts of the liquid, and put into a large volume of the pure dispersion medium. The object glasses are placed in position before sedimentation of these diluted samples has comnienced. But the chief cause of these structures is to be found in the treatment of the specimens, i,e. in the cover glass and the resulting compression and superposition of layers. If, for example, a moist highly disperse barium sulphate precipitate, whose crystalline character is recognisable as distinct crystal needles, at very great magnifications by an immersion lens is placed on a slide and pressed down with a cover glass, the crystalline character wiU be recognisable only at places where no needles are super- posed, and if there are several layers of needles, the preparation will appear amorphous with an indistinct grained structure. Such a preparation examined with a dry objective of insufficient magnification will show no crystalline nature, but will, instead, appear as grained flecks, closed or open network, or as a dendritic structure. PART II METHODS OF PREPARATION CHAPTEK XII INTEODUCTION The preparation of a sol, a disperse system the dispersity of which lies between certain limits, must obviously follow one of two lines : either the dispersity is increased, or it is diminished. In the former case, we start with the (massive) solid, in the latter, with the substance in the form of molecules or ions in solution or vapour. If the process of solution of a crystalline solid in a liquid is regarded as a continuous change, the dispersity of the solute continuously increasing, and the crystallisation of a solid from solution regarded as the reverse process, it is evident that every system during the processes of solution and crystallisation passes through the degree of dispersity corresponding to the sol stage. There are then two main groups into which the methods of preparation fall, I. Crystallisation Methods, and II. Solution Methods. All the known methods come under these two heads. The electrical dispersion method of Bredig, and the later modifications of it, may appear to be possible exceptions, i66 PREPARATION OF COLLOIDS but are almost certainly to be regarded as crystallisation methods. It is, however, convenient to treat them as a separate group. This classification is due to von Weimarn, and is similar to that adopted by Svedberg, v7ho classifies the methods as Condensation Methods, and Dispersion Methods. In the former the disperse phase is produced by " condensation " from ions in solution, and the latter includes all those methods which start from a solid phase of low dispersity. It therefore takes in the electrical dispersion methods. In spite of a formal similarity, there is a considerable difference between the two classifications ; the latter is somewhat arbitrary, especially as regards dispersion methods, some of which, as Svedberg states, are more probably condensation methods, von Weimarn, on the contrary, bases his classification on the continuity between coarse heterogeneity and the apparent homo- geneity of solutions, so that every crystallisation begins with the formation of a highly disperse system, a sol, and in every process of solution the solid phase, just before complete solution, forms a highly disperse system, a sol. Thus, to prepare a sol it is merely necessary to arrest the process of crystallisation or of solution at the appropriate stage. Simple as this may appear, in practice it is not always easy to effect, and the art of preparing sols consists in so arranging the conditions that — (1) The disperse system is obtained in a fairly stable condition, and (2) Its concentration is not too small. As a brief historical sketch will amply show, the early preparations were mainly a matter of chance ; some one accidentally discovered the conditions under which a particular sol was obtained in a stable form. As these INTRODUCTION 167 were multiplied, certain general conditions were, of course, recognised, more especially after the general properties of sols had become better known. It thus became possible to classify methods, and to apply known methods which were likely to be successful in any given case. It was not, however, until von Weimarn elaborated his theory regarding the processes of crystallisation and solution that systematic treatment became possible. As his theory has had highly important results, and has already led to the preparation of a very large number of sols, and is capable of almost indefinite application, it will be briefly described. von Weimarn's Theory. — Two stages in the separation of a crystal solid from a solution must be taken into con- sideration. The one is the " initial condensation " of the solid from the solution, which must take place in the form of a highly disperse phase, the size being approximately that of the molecules or ions of the substance in solution. These " amicron " particles increase in size by two distinct processes: aggregation crystallisation, and ordinary re- crystallisation. In the former, when two particles come near enough to each other, they become similarly oriented, and unite to form a larger particle. In the latter, each crystal grows at the expense of the substance in the surrounding solution. The concentration and the stability of the disperse system depend on these two factors : the initial velocity of the condensation, and the velocity of the growth of the crystals. Doubtless other factors are also concerned, which have been as yet unconsidered, and, in consequence of their omission, the results are not quanti- tatively exact. For crystallisation to take place at all, there must be an excess of solute present, above that necessary for saturation under the conditions of experiment. If Q is i68 PREPARATION OF COLLOIDS the total amount of solute in unit volume of the solvent, and L is the solubility of the solute in the solvent, Q — L, the number of mols which must finally appear as a solid of constant solubility L, may be regarded as the " conden- sation pressure," or the pressure under which the molecules begin to condense to crystals. Against this pressure we have the resistance of the system to condensation, which is not the same for different solids or solvents. With a given condensation pressure, the "condensation resist- ance " will vary with the solubility of the disperse phase, i.e. with the molecular cohesion of the dissolved molecules to those of the dispersion medium. Since, at the outset, the disperse phase is practically of the same dimensions as the molecular complexes of the solution (it is invisible or barely visible in the best ultramicroscope), the initial reaction may be treated as homogeneous, and the velocity of initial conden- sation „y. , condensation pressure , Q — L , P W = k J -1 ^^-7— = Jc — = — = A- = iU condensation resistance L L in which A is a constant, Q and L, as above, the amoimt of solute in unit volume of the solution, and the solubility of the disperse phase in the dispersion medium. . P = Q — L is the actual excess of mols of solute in unit volume, and p U = Y-is the specific supersaturation at the initial conden- sation. The second factor, the velocity of the growth of the particles of the disperse phase, is expressed by the Noyes- Nernst formula- — INTRODUCTION 169 in which D is the diffusion coefficient ; 8 the thickness of the adhering layer (of uniform con- centration, equal to the length of the diffusion path) ; S the surface of the disperse phase ; c the concentration of the solution ; I Jhe solubility of the disperse phase of given dis- persity ; c — I the absolute supersaturation. For the crystallisation of a pure liquid the correspond- ing equations are — where T — T^ = the absolute over-cooling ; E = the latent heat of crystallisation ; U = the specific over-cooling. And V = ^ S(i - Ti) where H = the coefficient of thermal conductivity ; t = the melting-point for given dispersity ; and the other symbols have the same significance as above. Let us first consider the velocity of condensation. In the first place, the factor of primary importance is not P, p the absolute excess of solute, but j, the specific excess. An example will make this clear. With a given value of P (say, a few grams per 100 cc), a very soluble sub- stance, as, e.g., sodium chloride, will deposit nothing at first, and finally a few crystals will be formed. With the same value of P, an almost insoluble substance, as silver chloride, will give an immediate curdy precipitate. Owing to the large value of L in the former case, the I70 PREPARATION OF COLLOIDS velocity of condensation is much smaller than in the latter case. The value of P, nevertheless, is not without influence. The results are very different, according as a given value of U is given by a large P or by a small L. In the one case, a large amount of disperse phase must be produced, in the other very little ; if U is large, the former will give a concentrated gelatinous precipitate (a gel), the latter a large number of highly disperse particles (a sol). Thus, by suitable alteration of P or L or of both, we can secure the initial separation of the disperse phase in any desired form. The duration of the disperse phase in this form is governed by the second formula, which expresses the rate of growth of the disperse phase under the conditions pre- vailing. The smaller the absolute supersaturation (c — l) is, the smaller is V and the greater the stability of 'the disperse phase. The same result is attained by decreasing D, the rate of diffusion ; this occurs if the solution is very concentrated, in which case the disperse phase is also very concentrated, and further increases the stability. The conditions for obtaining a stable highly disperse phase are, therefore, small value of P and a corresponding small value of L, in order that U may be sufficiently large to give a large number of points of crystallisation. Suit- able values of P may be obtained by any of the usual means, as, e.g. (a) rapidly cooling a concentrated solution to a sufficient extent ; (6) adding a solution of the sub- stance to a liquid in which the solvent is completely soluble, but in which the solute is practically insoluble ; (c) production of the disperse phase by a chemical reaction between substances which are sufficiently soluble to give the desired value of P. A series of illustrative examples may now be given. INTRODUCTION 171 (1) Sodium chloride is so soluble in water (L large) that, even if it were possible to obtain a sufficiently large value of P, the suspensoid stage would be of short duration, owing to the rapid growth of the crystals in the concen- trated solution. (2) But if sodium chloride is produced by the inter- action of sodium ethylate or sodium thiocyanate and hydrochloric acid in, say, a mixture of ether and amyl alcohol, in which sodium chloride is practically insoluble, a curdy or gelatinous precipitate, similar to silver chloride, is formed. (3) If the concentrations in the above reaction are so chosen that P is not too large, a stable suspensoid sol of sodium chloride will be formed. (4) If an aqueous solution of silver nitrate, saturated at 20°, is cooled in liquid air, a clear glass is obtained, which contains the silver nitrate in the form of a highly disperse phase along with ice. This system is stable at low temperature in spite of the large value of P, because diffusion is reduced to zero. A fuller examination of the case of barium sulphate will be found of interest. The solubility of barium sulphate at 18° is 0*00024 g. in 100 c.c. ; this is not very insoluble, compared with, e.g. aluminium hydroxide or even silver iodide. It is so large that with solutions of barium nitrate or chloride (the ordinary reagents) and the ordinary laboratory solutions of sulphates, the values of P obtainable are not large enough to give large values of U ; and for the same reason the subsequent growth of the crystals is rapid. Under ordinary conditions of pre- cipitation, therefore, barium sulphate is obtained as a more or less immediate, crystalline precipitate, which, of course, is exactly what is wanted for analytical purposes. 172 PREPARATION OF COLLOIDS By making use of more soluble salts, as, e.g. barium thiocyanate (Ba(CNS)2,3H20), and manganous sulphate (MnS04,4H20), it is possible to obtain barium sulphate in any form, from a clear gelatinous mass to a clear solution. The equivalent solutions are mixed in equal volumes; correspondingly large volumes of the very dilute solutions must be used, otherwise there will not be a visible quantity of barium sulphate to separate out. Strictly, the product of the volume and the concentration should be a constant. The following table, given by von Weimarn, indicates the nature of the precipitate : — Equivalent v^ _ — concentration excess of " L of the reagents BaSOi (L = 0-0002 g. (n). (g. in 100 cc). in 100 c.c.) 0-00005 to 0-00014 0-0006 3 0-00014 to 0-0017 0-0017 to 0-75 0-75 to 3 to 7 0-0006 0-0096 0-0096 4-38 4-38 17-51 17-51 40-9 8 48 48 21,900 21,900 87,500 87,500 204,500 Nature of the precipitate. No precipitate in a year — mioro-orystals to be ex- pected in a few years (macro- crystals from large amounts of solution). Slow precipitation at U = 8. Suspensoid stage at U = 25 (momentary). Complete separation in months to hours. Precipitation in a few seconds at U = 48. Beyond this, instantaneous precipitation. Crystal skeletons and needles. At U = 21,900 crystals are barely recog- nisable. Immediate formation of " amorphous " precipitates. Cellular clear jelly. It is clear from this that to get a suspensoid of barium sulphate, it is necessary to diminish L, since to increase P leads, as was to be anticipated, to the formation of a (concentrated) gel. This can be done by the addition of INTRODUCTION 173 alcohol to the water. The suspensoid stage appears to be most stable at about U = 8. Eetaining this value, and reducing L to one-tenth of its value in water by means of alcohol, we must also reduce Q to one-tenth ; the velocity of growth will also be reduced to the same extent. The excess of barium sulphate in 100 c.c. is reduced to O'OOOIS g. and the stability of the disperse phase is greatly increased. For the sol to be stable, the solubility of the disperse phase must not exceed O'OOOl g. in 100 c.c. of the dispersion medium. The differences in the nature of the precipitate, according as the appropriate value of U is attained by change in P or in L, are further illustrated by a com- parison of the barium sulphate gel obtained by the above method with 7 n. solutions in water (P large, L un- changed) with the gel obtained by Kato's method (p. 202), in which solutions of sulphuric acid and of barium acetate in a mixture of alcohol and water (P smaller, L reduced) also produce a gelatinous precipitate of barium sulphate. The two precipitates are very similar in appearance, but their behaviour on dilution with water is strikingly different. The former gives at once a dense white crystalline precipitate, while the latter forms a clear sol, or is at least converted into an opalescent bluish sol, which is very stable. On the addition of an acid, as hydrochloric acid, a white precipitate is at once produced, owing to the increased solubility of the barium sulphate in the acid liquid, and the consequent rapid growth of the particles. One further instance may be given which shows clearly how the necessary conditions for a sol or a, gel may be calculated. The solubility of aluminium hydroxide is so small that it is impossible to determine it. We may assume it to be 10 times less soluble than silver 174 PREPARATION OF COLLOIDS iodide, say, 10 -^ g. in 100 c.c, whicli is 20,000 times less than that of barium sulphate. With the latter gels are obtained at P = 17'5 ; a similar precipitate of aluminium 17'5 hydroxide will be obtained at P = orjooo ~ 0'0009. This is confirmed by experiment. At a concentration of 0-0005 g. A1(0H)3 in 100 c.c. the precipitate is highly disperse, above 0*01 g. it is a gel. When, as here, the amount which separates out is so minute, there is not enough to form a skin on the drops, and a network or a fine-grained gel or gelatinous precipitate is obtained. The third stage of the table is not attainable at ordinary tem- peratures, as the concentrations are too small ; but at the boiling-point a microcrystalline precipitate is formed at P = 0*0005 g. Ammonia increases the solubility to such an extent that large microcrystals are readily obtained. Summarising, we have — 1. Substances which are very soluble. The suspensoid stage is only reached at large values of U, and the result is a concentrated suspensoid sol, a gel. If U is small the suspensoid stage is not recognisable. 2. Substances of small but measurable solubility (about 10"* g. in 100 cc). The suspensoid state is reached at both large and small values of U : the former gives a gel as in 1, the latter a suspensoid sol of small concentration. 3. Substances which are practically insoluble. At small values of U, the suspensoid stage is not recognis- able; with large. U dilute sols are obtained. A gel is obtained only at extremely large values of U. 4. However small the solubility, the precipitate is obtained as crystals with small values of U, if the volume and interval of time are sufficiently large. 5. Every highly disperse system obtained by the crystallisation method strives to diminish its dispersity, INTRODUCTION 175 because the smallest particles are more soluble than the larger. A brief discussion of von Weimarn's theory of the solution method will be found on p. 206. The crystallisation methods may be classified as follows, and will be described in this order : — 1. Cooling a solution or liquid. 2. Eeplacing one liquid by another. 3. Eeduction methods — (a) By hydrogen, carbon monoxide, acetylene, hydrazine, etc. (b) By phosphorus. (c) By organic substances, with or without pro- tection. (d) By metals, or kations. («) By kathode reduction. 4. Oxidation — Sulphur from hydrogen sulphide and air or sul- phur dioxide. 5. Hydrolysis — (a) Of metallic salts. (6) Of other compounds, e.g. silicon sulphide, ferric ethylate, etc. 6. Precipitation methods — (a) No electrolyte produced. (5) Electrolytes produced, with or without pro- tection. CHAPTER XIII CKYSTALLISATION METHODS I. Cooling Solution A FEW c.c. of a 0"02 per cent, solution of sulphur (or phosphorus) in alcohol is cooled in liquid air ; a clear glass is obtained, which is a highly disperse solid sol. This is evident when the tube is allowed to warm up to room temperature. The liquid becomes opalescent blue, chang- ing to white as the temperature rises. This is succeeded by a blue opalescence, and the sol gradually passes into a solution. An ice sol can be prepared in a similar manner. When a very dilute solution of water in ether, chloro- form, or xylene (or other liquid of low melting-point) is rapidly cooled in liquid air or in a mixture of alcohol and solid carbon dioxide, the water crystallises out as highly disperse ice, and a liquid sol is obtained. This method can be extended indefinitely to either low or ordinary temperatures, and to gases, liquids, or solids. II. Eeplacement of Solvent In this method the solvent is replaced by a liquid in which the solute is insoluble, and thus appears as a disperse phase. The preparation of a mastic sol or a gamboge sol is the classical example of this process. CRYSTALLISATION METHODS 177 Gum mastic is insoluble in water, but is readily soluble in alcohol or ether. When a dilute alcoholic solution is added gradually to a quantity of water, with constant stirring, the gum is suddenly precipitated in a highly disperse condition, and, owing to its insolubility, the dispersity does not appreciably diminish. Here we have a small value of P, while L is very small, and U is large, which fulfil the necessary conditions. If the mastic solution is concentrated, or if a large quantity of it is added to the water, a gel is obtained which is also fairly stable. The method is obviously capable of almost indefinite extension. A few instances will sufBce. (1) Sulphur {Selenium, Phosphorus) The solubility of sulphur in alcohol is 0'05 g. in 100 g. at 18°, and 0'4: g. at 78°; of phosphorus in alcohol, 0-3 g. at 18°, and 0-4 g. at 78°. Both are practically insoluble in water. Crystallised selenium is insoluble in alcohol, but dissolves in carbon disulphide, its solubility being 0-02 g. in 100 g. at 46°. To prepare sulphur or phosphorus sols, 5-25 c.c. of cold saturated solution are slowly added to 1000 c.c. of cold water, with vigorous stirring. The sols are colourless, with a bluish opalescence, and are stable for days or weeks. For the selenium sol, 5-25 c.c. of a hot saturated solution are similarly added to 1000 c.c. of cold ether (water cannot be used, as carbon disulphide also is insoluble in it). It is of a beautiful rose colour with a metallic sheen, and is as stable as the others. (2) Silver Iodide Complex salts or double salts, one of whose components is an insoluble salt, and which are decomposed by water N 178 PREPARATION OF COLLOIDS into their components, may yield sols. Thus, if silver iodide is dissolved in potassium iodide, and the solution poured into a large volume of water, a silver iodide sol is obtained. Its stability is increased by the potassium iodide, which at very small concentrations diminishes the solubility of silver iodide. (3) Gelatine (or Agar) If a very dilute solution of gelatine or agar in boiling water is poured into a large volume of boiling alcohol, extremely stable suspensoid sols are obtained. III. Reduction Methods Historical Sketch. — The potable gold of the alchemists was probably a gold sol, since it was a solution of gold salts in ethereal oils ; these slowly reduce gold chloride with production of gold sols. The production of a red gold solution by the action of a tin solution on gold chloride was described in a text-book so long ago as 1749. In 1794 was published an elaborate research on the colours produced on silk dipped in solutions of gold salts, and subsequently treated with hydrogen or phosphorus in ether. Red and purple shades were obtained. Out of many early references, mention may be made of reductions with hydrogen and phosphine (1811) ; of the action of egg-white and of isinglass, which gave reddish- purple sols ; and of starch, which gave blue sols. We have here very early (1821) the action of "protective" colloids in conferring stability. Gold chloride was also reduced by hydrogen and hydrogen sulphide, by gallic acid, and by oxalic acid, etc. (1822). Finally, BerzelLus, in his Lehr- huch (1844), gave many recipes for the preparation of gold sols of different colours. CRYSTALLISATION METHODS 179 Curiously enough, in spite of this wealth of recorded observations on gold sols, they were discovered anew by Faraday 1 in 1856, who obtained them by reduction with phosphorus in ether, and were rediscovered once again by Zsigniondy2 in 1898. The latter also rediscovered Faraday's work, while both apparently knew nothing of all the early work. All the earliest work seems to have been confined to gold, no doubt partly because of its alchemistic importance, partly because of its ready reduction. Wohler* reduced dry silver meUitate or citrate by means of hydrogen gas at 100° ; a brown mass was obtained, which dissolved in water with a yellow- brown colour (1839). Carey Lea's* reduction of silver solutions with strong solutions of ferrous citrate gives gels containing also ferric ion (which probably acts as a "protector"), from which the sol is obtained by washing out the citrate. From the precipitating effect of citrate, the sol would appear to be positive, i.e. the negative silver sol + excess of Fe"* (see p. 137). Carey Lea regarded the gels as " allotropic " forms of silver. The modern history of reduction methods begins with Zsigmondy's rediscovery of gold sols by the reduction of a faintly alkaline solution of auric chloride with form- aldehyde. He sought out the conditions for obtaining, as often as required and with small risk of failure, a uniform highly disperse gold sol, which was at once stable and sensitive to reagents. Further, his method could be applied with but sHght modification to other noble metals. This was done by Lottermoser^ (1901), who • Phil. Trms., 1857, 145. " Annalen, 1898, 301, 29. = Pogg. Ann., 1839, [ii.] 46, 629. ' Amer. J. Science, 1889, [iii.] 37, 476 ; 38, 47. ■" TJeher cmorg. Koll., 1901. i8o PREPARATION OF COLLOIDS prepared platinum, palladium, rhodium, and iridium sols by this method. According to von Weimarn's theory, the preparation of stable highly disperse sols of metals by reduction of their ions in solution should be an easy matter, for the metals are insoluble in the dispersion medium ; thus it is easy to have a large value of U with very small P, and vanishingly small L, and practically no D. Hence there will be a very large number of points of conden- sation, and no subsequent growth in the solution, for all the ion will have been changed into metal, which is insoluble. This is the actual experience, but there is a difficulty of a different origin. The highly disperse sols are precipitated by traces of electrolytes, and these may be produced by the reduction or be present as impurities in the water. On this account many inorganic reducers are inadmissible, especially acids and salts, as their ions are usually powerful precipitants. If they are used, pro- tective colloids must also be present. When stannous ion is used as the reducer, stannic acid is formed along with the metal sol, and so stable sols are obtained. Examples of this are the purple of Cassius, the alchemical preparations of gold, and Lottermoser's preparation of mercury, by means of stannous chloride. Pure suspensoid sols are usually not stable unless very dilute, so that if concentrated sols are wanted, as they may be for medical or technical purposes, recourse must be had to the stabilising action of protective colloids. Paal^ has elaborated a method which makes use of two degradation products of egg-albumin, which he names protalbic acid and lysalbic acid. Their sodium salts are obtained by the action of sodium hydroxide on albumin ; 1 Ber., 1902, 35, 2206. CRYSTALLISATION METHODS i8i they exert a powerful protection, and at the same time are strong reducers. They may be used alone, or in con- junction with other reducers, as hydrazine, sodium amal- gam, etc. Others, too, have employed protective colloids for the same purpose. These sols are not so useful for many scientific purposes, as their properties differ so much from those of the pure sols, but they are extraordinarily stable, and may be dried up to hard, metallic-looking grains, which after years will completely solate when placed in water. It is very doubtful whether their physiological activity is at all comparable to that of the unprotected sols. Kathode reduction may give rise to sol formation, BUliter* obtained a yellow-brown mercury sol at the kathode when a 0*004 n. solution of mercurous nitrate was electrolysed. He employed a U tube with side tubes, the bend being plugged with glass wool. On prolonged electrolysis at 220 volts and 0'2-0*3 ampere, the solution was converted into a turbid gray liquid. The purity of the water is an important factor in the production of stable sols. Bourdillon's method of distilla- tion 2 is simple and most effective. Using this water (k = 0"3 to 1 X 10 -® mho), the author has obtained brilliant stable sols of gold, silver, ferric hydroxide and arsenious sulphide. Naturally it must also be used for the dialysis. It is quite impossible to deal exhaustively with all the reducers which have been employed successfully. Full details are to be found in Svedberg's invaluable book.3 Among the reducers which have been found effective are the following : — 1 Ber., 1902, 35, 1929. » Chem. Soc. Journ., 1913, 103, 791. ' Methoden z. Darsiellung Eolloider LSsungen, 1909, l82 PREPARATION OF COLLOIDS Reducer. Sol. Author. Hj Ag Kohlschiitter ' CO Au, Pd Ag Donau ^ Kohlschiitter ' SO, \ H3POJ Ag v.d. Pfordten^ Se Schulze * Te Gutbier = P Au Faraday = Zsigmondy ' Terpenes, alcobols Au Vanino ' Acetylene Au Blake" Formaldehyde Au Zsigmondy ' Pt, Pd, Rh Ir Lottermoser '<> „ + silicic acid Au, Ag Kiispert" Acrolein Au, Pt, Pd, Os, Eu Castoro '= Carbohydrates Au Vanino " N,H, 1 Ag, Au, \Se, Te Pt, Pd. Ir, ■ Gutbier '« NHjOH Au, Te C,H,N,H, Au, Te n With Protective Sols. Protalbates and lysalbates Protalbate + Hj or NjH^ Lysalbate + NaHg Ag,0 (to 80%) Ag (93%) Au (93%) Pt (78%) Pd (76%) Ir (40%) Cu (red and blue) Cu (brown and blue) Se Ps al» Polyphenols Phenol acids Au, Pt, Ag, Hg Au, Ag, Pt H G enrich" irhowski '* Fe" (citrate) Sn" Ag Hg, Bi, Cu Carey Lea '= Lottermoser " ' Zeitsch. Etektrochem., 1908, 14, 49. 2 Monatsh., 1905, 26, 525 ; 1906, 27, 71. = Ber., 1885, 18, 1407. * /. pr. Ghem., 1885, [ii.] 32, 390. » Zeitsch. anorg. Chem., 1902, 82, 51, 91, 105. « Loc. cit., p. 179. ' Loc. cit, p. 179; also Zeitsch. anal. Ghem., 1901, 40, 697. ' Ber., 1906, 39, 1696; 1905, 38, 463. ' Amer. J. Science, 1903, [iv.] 16, 381. " Loc. cit., p. 179. " Ber., 1902, 36, 2815, 4066. '' Zeitsch. anorg. Ghem., 1904, 41, 126. " Koll. Zeitsch., 1907, 2, 51. " Zeitsch. anorg. Ghem., 1905, 4S, 77; 1902, 31, 448; J. pr. Ghem., 1905, [ii.] 71, 358, 452 ; Zeitsch. a/norg. Ghem., 1902, 82, 51, 91, 106, 347 ; 1904, 39, 112. ' " Ber., 1902, 35, 2206, 2224, 2236; 1904, 37, 124. '" Ber., 1906, 39, 1545, 1550; 1905, 38, 584, 526. " Ber., 1903, 36, 609, 1215. " Ibid., p, 1815, •» Loc. cit., p. 179, CRYSTALLISATION METHODS 183 A detailed account of the following methods is given as being the most generally useful and illustrative : — 1. Kohlschiitter . silver sol with hydrogen. 2. Donau . . . gold sol with carbon monoxide. 3. Zsigmondy . . ,, „ formalin. ,, „ phosphoruiS. 4. Castoro . . . platinum sol with acrolein. 5. Gutbier . . . red and blue gold sols with hydrazine. (1) Silver Sol by reduction with Hydrogen (Kohlschiitter) The method consists in the reduction of silver oxide or hydroxide in presence of water by means of hydrogen gas. As the two products are silver and water, the method is one which yields an electrolyte-free sol — 2AgOH + H2 == 2Ag + 2H2O Hydrogen gas is bubbled through a hot (50°-60°) saturated solution of silver oxide, contained in a round- bottomed flask, which is preferably of Jena glass. Solid silver oxide is present to keep up the concentration. In 8-10 hours about O'5-l litre of intensely coloured sol is obtained. Eeduction is too slow below 50°, while above 60° the sols are very unstable. Analysis, showed the presence of some unchanged silver oxide. As the action appeared to take place mainly at the interface between the liquid and the vessel, Kohlschiitter tried the effect of various surfaces. In a platinum vessel a crystalline deposit was obtained in place of the sol. This suggested a means of purification, as follows. About 80 CO. of the sol is placed in a blackened platinum basin, and protected from air by a bell-jar. Pure hydrogen is passed through the liquid by means of a tube with a platinum tip. In 12 hours the residual silver i84 PEEPARATION OF COLLOIDS oxide was reduced, the silver being deposited as bright crystals which firmly adhered to the basiiii During this treatment the electric conductivity fell to 7-8 X 10~^ mho, a tenth of its former value. The specific con- ductivity of the water was about 2 X 10~^ mho. (For other interesting details of this work, see p. 107.) This preparation can be made as a lecture experiment, if a saturated solution of silver oxide is substituted for the suspension. About 150-200 c.c. of filtered solution are heated to 60° in a Jena flask fitted with Jena glass tubes, and a brisk current of pure hydrogen passed through it. In 20-50 minutes a perfectly clear, intense yellow sol is obtained, which presents a most brilliant appearance in the ultramicroscope. It contains about 0'03 g. Ag in 1000 c.c. (2) Gold Sol by reduction with Carbon Monoxide (Donau) Pure gold, precipitated by oxalic acid, is dissolved in aqua-regia, evaporated on a water-bath, and dissolved in pure water to form a 0'002-0'05 per cent, solution. If ordinary distilled water is used, a protective colloid must be added. The carbonic oxide, prepared from oxalic acid and sulphuric acid, may be used along with the carbon dioxide, or freed from it; the mixed gas gives a better colour. The gas is stored in a gasometer, and, after bubbling through water, is passed into the gold solution by means of a capillary tube. The solution may be cold or warm; in a few minutes a red colour is noticeable. The reduction is complete and the end of it is ascertained by the electric conductivity. Large quantities can be made at a time, if the gas is led in slowly. A litre of 0"002 per cent, sol may be had in about 2 hours. " The sol is of a clear red colour, and is very stable. The carbon CRYSTALLISATION METHODS 185 dioxide can be removed by boiling, and the hydrochloric acid by dialysis. A very stable palladium sol is obtained by treating a palladous chloride solution (0"0005-0"05 per cent.) in the same manner. The reaction is quicker, and is complete. Most of the acid can be removed by dialysis. Kohlschutter made a silver sol by reduction of silver oxide solution with carbon monoxide. It was not very stable. (3) Gold Sol by reduction with Formaldehyde (Zsigmondy) The following are Zsigmondy's later directions, and by adhering to them it is easy to prepare a pure red gold sol. About 120 c.c. of pure water are put into a Jena glass beaker of 300-500 c.c. capacity, and raised to boiling point. During the warming 2*5 c.c. of the gold chloride solution (6 g. of crystallised HAuCl4,3H20 in a litre), and 3-3*5 c.c. of pure potassium carbonate solution (0"18 n.) are added. As soon as it boils, 3-5 c.c. of formaldehyde solution (0"3 c.c. of commercial formalin in 100 c.c. of water) are added moderately quickly, a little at a time, with constant vigorous stirring (a soft glass rod must not be used ; a Jena glass tube closed at the ends will do). The reaction commences in a few seconds, or, at longest, a minute ; the liquid turns a pale red, rapidly changing to an intense bright red, which undergoes no further change. The sol may be purified by dialysis. It appears to be impossible to prepare more than 120- 150 c.c. at a single operation, with larger quantities the colour is not so good or clear. If a large quantity *is required, the individual samples are collected and sub- mitted to purification together. i86 PREPARATION OF COLLOIDS (4) Gold Sol by reduction with Phosphorus (Zsigmondy) Faraday's gold sols, prepared by adding a solution of phosphorus in ether to aqueous solution of gold chloride, varied very much in colour and stability. Zsigmondy, by combining it with his formaldehyde method, has devised a method which gives, without difficulty, gold sols of the highest dispersity. The directions and quantities are the same as in the formaldehyde method (see above), except that the solution is not boiled, as reduction takes place at the ordinary temperature, and that a few drops of a solution of white phosphorus in ether are added instead of the formalin. The action is slow, the liquid turning first pale yellow- brown, then gradually a pure red, which is perfectly clear, without a trace of opalescence by either transmitted or reflected light. The Tyndall effect is scarcely observed in it. The size of the particles is about l-2-l*5 fi/x, and is exceedingly uniform. In the ultramicroscope a few isolated gilded dust particles are observed, and bfeyond that, only a feeble cone of light which gradually fades away on dilution. The sol is very sensitive to electrolytes. (5) Platinum Sol by reduction with Acrolein (Castoro) About 500 c.c. of a solution of HgPtClg (1 g. in a litre) are made slightly alkaline with a few drops of potassium carbonate solution, and raised to boiling point. The flame is removed, and 2-4 c.c. of acrolein solution (Kahlbaum's 33 per cent, solution) are added. The yellow colour disappears in a few seconds, and is succeeded by a brown colour, which finally becomes black. It is filtered, diluted with its own volume of water and dialysed. The sol is quite clear in thin layers, and is very stable. CRYSTALLISATION METHODS 187 For dilute sols it is better to make 50-60 c.e. of the platinum solution up to 200 c.c. with pure water, and to proceed as above, using correspondingly less acrolein. Gold, palladium, osmium, and ruthenium sols may also be prepared by this method. (6) Blue Gold Sol by reduction with Hydrazine (Gutbier's Methods) In these hydrazine, phenylhydrazine, and hydroxyl- amine are employed as reducers. 1 g. gold chloride is dissolved in 1000 c.c. of pure water, and exactly neutralised with very dilute sodium carbonate. A solution of hydrazine hydrate (1 c.c. of Kahlbaum's commercial 50 per cent, solution to 2000 c.c. of water) is added drop by drop. The reaction begins at once with production of a deep blue sol. In all a few c.c. of the hydrazine solution are required for complete reduction. The sol is deep blue by both transmitted and reflected light, is clear, and very stable. If too much reducer is added, the sol is not clear and soon settles out. A peculiarity of the phenylhydrazine reduction of gold is that stable sols of different colours can be obtained by simply adjusting the amount of reducer. The following is a striking experiment : — Dilute 5 c.c. of gold solution (0*1 per cent.) with 300 c.c. of water in a large beaker, and run in from a burette 0"2 to 0*5 c.c. of a freshly prepared solution of phenylhydra- zine hydrochloride (0*4 per cent.). On stirring the liquid becomes deep red. Now add more reducer, drop by drop, when the colour changes to violet ; after addition of 5 c.c. the colour is violet-blue, a further quantity changes the colour to blue, and finally to deep blue when 12 c.c. have been added. Silver, platinum, palladium, iridium, tellurium, and 1 88 PREPARATION OF COLLOIDS selenium sols, with gum arabic as protective colloid, have been made by this method. The colour of the gold sols produced by reduction with hydrazine hydrate depends on the reaction of the solution. If the solution is faintly acid the sol is bright blue, if it is almost neutral the sol is violet or purple, while a brilliant red sol is obtained if the reaction is alkaline (author's unpublished results). Thus — 100 c.c. water, 1 c.c. gold solution (1%), 1-1"5 cc. potassium carbonate solution (O'lS n.), 1-2 c.c. hydrazine hydrate (1 c.c, of 50% solution to 2000 c.c. of water), give a bright blue sol. If 2 c.c. of potassium carbonate are used the sol is violet, and with 2'5 c.c. of potassium carbonate and 3 c.c. of hydrazine hydrate solution the colour is a brilliant clear scarlet. All three sols are very stable, and are easily reproduced. Faal's Method. .Frotalbic Acid and Lysalbic Acid as Protective Colloids This method is important because of the highly con- centrated sols which can be prepared by it (for "Pro- tection," see p. 128). There are two ways in which these substances have been employed ; in the earlier method they were used both as reducers and protectives, while in the later they are simply protectives, other powerful reducers, such as hydrazine, sodium amalgam, etc., being used to effect the reduction. The latter method is the more useful. Two examples will suffice. Silver Oxide Sol and Silver Sol.^1 part of sodium prot- albate is dissolved in 15 parts of water, and some sodium hydroxide is added. Silver nitrate is then added until a CRYSTALLISATION METHODS 189 precipitate forms, which is dissolved by more sodium hydroxide. Silver nitrate and sodium hydroxide are added alternately until the desired concentration is reached (up to 3'2 parts of AgNOs, or 2 parts Ag). If a silver oxide sol is wanted, this solution is dialysed. For silver sol the undialysed silver hydroxide sol is heated on the water- bath until reduction is complete; this is known by the absence of any reaction with one drop of hydrazine solu- tion. The sol is then dialysed and evaporated in a flat basin on the water-bath. It consists of shining grains with a beautiful blue-green metallic sheen, and is com- pletely soluble in water, even after two years if a drop of sodium hydroxide solution is added. Gold, platinum, and other metal sols are obtained in a similar manner, with the aid of hydrazine as reducer. Selenium. — 3 g. of sodium protalbate are dissolved in 45 c.c. of water, and 1"63 g. of SeOg ( = 1 g. Se) dissolved in water is added, along with sodium hydroxide in suffi- cient quantity to redissolve any precipitated protalbic acid. 2 g. of commercial hydrazine hydrate solution are added, and the liquid then rendered just acid in reaction by means of dilute hydrochloric acid. The liquid froths up and becomes blood-red in colour. Alcohol is added to diminish the frothing. The reduction is aided by gentle heat, which causes red flocks of colloid selenium and free protalbic acid to settle out. These are dissolved by addition of sodium hydroxide, and the solution is then dialysed. The sol is quite clear and of the colour of arterial 'blood. It can be evaporated on the water-bath and dried in a vacuum over sulphuric acid. It is a dark red, shining, and enamel-Uke substance, and is completely soluble in water. It contains 33 per cent, of Se, and 5 per cent, of Na. The solubility in water is retained for several years. igo PREPARATION OF COLLOIDS Mercury (Lottermoser), A very dilute solution of mercurous nitrate is poured into a dilute solution of stannous nitrate. An excess of stannous salt is necessary, and neither solution should con- tain more free acid than will just prevent the separation of basic salts. The liquid becomes deep brown in colour. The mercury sol is precipitated by a strong solution of ammonium citrate as a fine black precipitate, and the liquid carefully neutralised with ammonia. The liquid is syphoned off, the rest being sucked out by means of a porous earthenware filter. The precipitate is dried in a vacuum over sulphuric acid. It has a silvery lustre and forms in water a deep brown sol. This preparation owes its stability, no doubt, to the protection exerted by the emulsoid stannic acid, which is produced at the same time by the hydrolysis of the stannic nitrate. , This accounts for the salting out and the subse- quent regeneration of the sol. Bismuth and copper sols have also been prepared by modifications of this method. IV. Oxidation Methods The oxidation of hydrogen sulphide to water and sulphur seems to be the only instance of this method. When a solution of hydrogen sulphide is exposed to air a bluish opalescent sol is first noticed, which subse- quently turns yellowish-white and deposits sulphur. Or a solution of hydrogen sulphide is acted on by Sj solution of sulphur dioxide, when a series of complicated reactions occurs, one product of which is sulphur gel and sol (Wackenroder's solution, 1846).i Wackenroder prepared it by passing hydrogen sulphide into a saturated solution of sulphur dioxide at room temperature ; he observed that • Annalen, 1846, 60, 189. CRYSTALLISATION METHODS 191 the opaque yellowish-white liquid is very stable, and is apparently homogeneous, as it can be filtered unchanged. ' Selmi and Sobrero (1850) ^ passed sulphur dioxide and hydrogen sulphide simultaneously into a flask full of water. The yellow precipitate of sulphur dissolved in water and was reprecipitated by salts. If sodium salts were used, the precipitate redissolved in water, but not if potassium salts had been used. In the latter case the precipitate is elastic and remains so for months. The precipitate contains potassium sulphate which cannot be washed out. The acid liquid contains sulphur sol, which is precipitated by potassium or sodium salts. They con- sidered it to be an emulsion (like soap, starch, prussian blue, in Selmi's previous researches). Debus 2 fully investigated Wackenroder's solution and gave details of the preparation. Eecently Eaffo* prepared sulphur sol from sodium thiosulphate and sulphuric acid as follows : 50 g. of pure Na2S203,5H20 are dissolved in 30 c.c. of water, and added, drop by drop, with constant stirring to 70 g. of sulphuric acid (of density 1'84) contained in a 300 c.c. cylinder, which stands in cold water. Hydrogen sulphide is first formed, then sulphur dioxide, followed by a whitish precipitate, changing to lemon yellow. When the action is over, 30 0.0. of water are added and the whole shaken up and heated to 80° for 10 minutes. The sulphur dioxide is expelled, and the thick turbid mass clears up to a sulphur yellow liquid, which is filtered through glass wool, kept cold for 12 hours, is reheated and then refiltered. This treatment is repeated until all the insoluble sulphur is removed. At this stage the sulphur is a yellow mass which dissolves on warming to a perfectly clear yellow I Atm. Chim. Phys., 1850, [iii.] 28, 210. ' Chem. Soc. Journ., 1888. 53, 278. = Koll. Zeitsch., 1908, 2, 358, 192 PREPARATION OF COLLOIDS sol. It is allowed to settle out, centrifuged, washed with cold water, centrifuged again, dissolved in water, precipitated by sodium carbonate, centrifuged, and so is obtained as a sediment of almost pure soluble sulphur, which solates in distilled water. One such sol contained 4:"6 per cent, of S and 1'5 per cent, of N'a2S04. Its stability is not very great, and is diminished if the sodium sulphate is dialysed away. With sodium salts a gel is precipitated, which solates in pure water, or on heating to 80°. With potassium salts the gelation is irreversible. A sol which had been neutralised by sodium carbonate remained clear when kept in a stoppered vessel. V. Hydrolysis Methods The hydrolysis of salts is really a special case of double decomposition, in which water is one of the re- agents ; this is clearly seen if the reaction takes place in non-aqueous solution. Theoretically, all salts undergo hydrolysis, but not to a recognisable extent unless the acid or the base, or both, of the salt is a weak electrolyte. The equations Fe-'CV + 3H"0H' ^ SH'Cr + re(0H)3 and NaCN' + H-OH'^lsra-OH' + HCN illustrate the reaction. In the latter instance the acid is very weak, but is soluble. In the former the base is probably very weak, but is certainly very insoluble, and thus the equilibrium point is very far on the right side. Hydrolysis is, as indicated in the above equations, reversible, and increases with the dilution, and also with rise of temperature. When, as in the first reaction, one product is an insoluble substance, we have the necessary conditions for obtaining a stable sol. Even in cases CRYSTALLISATION METHODS 193 where the hydrolysis is slight, a sol may be obtained by dialysis, if the salt is not nearly so diffusible as the acid, which is usually the case. Or the acid may be volatile, in which case hydrolysis can be completed by boiling off the acid. The salts which have been most frequently utilised for preparations are those of organic acids (acetate chiefly, because acetic acid is both volatile and a weak acid), and nitrates and chlorides, owing to their solubility and the ready diffusibility of the acids. Another factor of importance is the precipitating effect of anions on positive sols, which all the hydroxide sols formed by hydrolysis of these salts are. Sulphates are more readily hydrolysed than chlorides or nitrates, but ferric sulphate cannot be used to prepare a ferric hydroxide sol, owing to the precipitating effect of the bivalent anion SO4" on the positive sol, and this in spite of the stabilising effect of the H" ; in consequence a precipitate of so-called basic ferric salt is produced instead of a sol. The earliest preparation of a sol by hydrolysis was the hydrolysis of silicon sulphide, which was discovered by Berzelius (1833).^ Crum^ then prepared an aluminium hydroxide sol from aluminium acetate by boiling off the acetic acid. He failed to get ferric hydroxide sol in the same way, but this was done by Pean de St. Gilles a year later.^ Ther^ followed the dialysis researches of Graham, but these processes are really solution processes, as he digested the solution with precipitated hydroxide before dialysis. The hydrolysis of ferric chloride was fully investigated by Debray,^ and more especially by Krecke.^ The latter found that with solutions containing over 4 per cent, of 1 LeJirbuch, 3 Edition, II., 122, ' Atmalen, 1853, 89, 156. ' Compt. rend., 1855, 40, 568. * Compt. rend., 1869, 68, 913. * J. pr. Ohem., 1871, [ii.] 3, 286. 194 PREPARATION OF COLLOIDS ferric chloride the hydrolysis is reversed on cooling, while below 1 per cent, it is quite irreversible. The temperature at which hydrolysis is complete also depends on the con- centration, thus : Oonoentration. g. FeClj in 100 g. of solution, Temperature. 82 130° 8 110° 4 100° 1 83° 0-25 64° 0-06 36° Most of the free acid can be removed by dialysis. Goodwin 1 showed by measurement of the electric con- ductivity that the hydrolysis of ferric chloride is a time reaction. The same holds for zirconium chloride, and for aluminium acetate. As usual, a few typical preparations will be given. (1) Hydrolysis of Acetates Aluminium Hydroxide Sol ^ A solution of aluminium acetate (containing not more than 0'5 per cent, of A1(0H)3) is heated to 100° in a closed vessel for 10 days. The liquid is diluted in a wide flat basin, and is kept boiling vigorously until all the acetic acid is driven off, water being constantly added to replace that boiled off. If the depth of the liquid does not exceed 5 mm., 1 to 2 hours' boiling will be sufficient. The final concentration should be about 1 part of A1(0H)3 in 1000 parts of water. The sol is transparent, tasteless, and neutral. ' Zeitsch. physikal. Chem., 1896, 21, 1. « Orum, Annalen, 1853, 89, 156. CRYSTALLISATION METHODS 195 (2) Hydrolysis of Nitrates (a) Zirconium Hydroxide Sol ^ 30 g. of zirconium nitrate are dissolved in 500 c.c. of water, and evaporated to dryness on a water-bath. The residue is dissolved in 500 c.c. of water and again taken to dryness ; this treatment is repeated several times. The final product is a clear gum, which swells up in water, and forms a very viscous sol. On dilution an opalescent sol is obtained. (&) Ferric Hydroxide Sol ^ A solution of iron in nitric acid is freed as far as possible from acid. It is then dialysed in a parchment- paper tube, the external water being changed three times a day. The dialysis is continued until nitric acid can no longer be detected in the exterior water. This takes about four to six days. The sol is clear brown, with a slight opalescence by reflected light, Sols of the following metallic hydroxides have also been prepared by this method : chromium, aluminium, tin, bismuth, cerium, thorium, zirconium. In every case the sol contained traces of nitrate. (3) Hydrolysis of Chlorides Ferric Hydroxide Sol * A few c.c. of a 32 per cent, solution of ferric chloride are added slowly to 1000 c.c. of boiling water. The liquid becomes dark red-brown, and remains so on cooling ; it ' MiUler, Zeitsch. cmorg. Chem., 1907, 52, 316. ' Biltz, Ber., 1902, 35, ii31. » Kreoke, J. pr. Chem., 1871, [ii.] 3, 286, 196 PREPARATION OF COLLOIDS contains ferric hydroxide sol and hydrochloric acid. It will serve many purposes without purification. Practically all the acid can be removed by dialysis. (4) Hydrolysis of Silicon Sulphide' Berzelius, and later Fremy, obtained silicic acid sol by the action of water on silicon sulphide. Hydrogen sulphide is evolved and j3 silicic acid remains in solution. It is very stable (it is an electrolyte-free preparation, as the gas escapes), but becomes gelatinous if it is concentrated or boiled, or if alkaline salts are added. (5) Hydrolysis of Organic Compounds (a) Ferric Sydroxide Sol from Ferric Ethylate^ Anhydrous ferric chloride is dissolved in absolute alcohol, and treated with sodium ethylate. Sodium chloride is precipitated, the ferric ethylate remaining in solution. This is poured into a large quantity of water, when a clear brown sol is obtained. (&) Silicic Acid Sol from Methyl Silicate 2 A solution of 8 g. of methyl silicate in 200 c.c. of, water is boiled under a reflux condenser. The methyl alcohol is then removed by boiling off one-fourth of the liquid. A 2*25 per cent, sol is obtained, which, as the method indicates, is very stable against heating. ■ ' Berzelius, Lehrbuch III. Auf . 2, 1832, 122 ; Frfimy, Ann. Chim. Phys., 1853, [iii.] 38, 812. » Giimaux, Compt. rend., 1884, 98, 105, 1434. CRYSTALLISATION METHODS 197 VI. Other Crystallisation Methods This section contains all the crystallisation methods which are not reductions or hydrolyses. As already suggested, there is no scientific importance in this sub- grouping. These methods are mainly double decompositions, one product of which is insoluble. They are thus typical cases of von Weimarn's theory, and special attention is again directed to it, and to the examples already de- scribed in detail, which properly come into this section (p. 171). The stability of the sol depends to a great extent on the nature of the other product. If it is an electrolyte, the sol may soon be precipitated by it, unless the solutions are very dilute ; or it may not be obtained at all unless a protective colloid is present. A^e have thus three classes : (a) The other product of the reaction is a non- electrolyte. (6) The other product is an electrolyte, and the con- centrations must be very small. (e) A protective colloid is added to increase the stability of the sol. Historical Sketch. — In the third edition of his Lehrbuch (1833) Berzelius described the preparation of soluble arsenious sulphide from arsenious oxide and hydrogen sulphide, and its precipitation by hydrochloric acid. He also states that osmium sulphide is obtained as a brown solution by the action of hydrogen sulphide on osmic acid, and that it, too, is precipitated by acids. Then Graham (1867) in his classical experiments pre- pared many sols by the action of acids or alkalies on soluble salts, as e^g. hydrochloric acid on silicate, stannate, molybdate, or potassium hydroxide on stannic chloride; also by double decomposition of salts in dilute solution 198 PREPARATION OF COLLOIDS and subsequent dialysis to remove the electrolyte product, e.g. copper ferrocyanide, prussian blue. Schulze (1882) returned to the action of hydrogen sulphide on oxides of arsenic and antimony. In the latter case no sulphide sol was formed, and he investigated the action of hydrogen sulphide on tartar emetic ; other salts are produced in the reaction, and the concentration is an important factor. No sol is obtained if the tartar emetic solution is more than 1'6 per cent., and partial precipita- tion occurs between 1"25 and 0'5 per cent. Winssinger (1888) prepared many sulphide sols by the action of hydrogen sulphide gas on very dilute solutions of salts, e.g. of platinum, palladium, gold, silver, thallium, lead, bismuth, iron, nickel, cobalt ; other sulphides, e.g. of tungsten and molybdenum, by dissolving the sulphide in ammonium sulphide and reprecipitation with acetic acid. The salts were removed by dialysis. Zinc sulphide sol was prepared by passing hydrogen sulphide into a suspen- sion of zinc hydroxide. It is doubtful if this last method is not a solution method ; it may at least become one if excess .of hydrogen sulphide is passed in. Linder and Picton (1892) substituted hydrogen sulphide solution for the gas, at the same time keeping up the concentration by passing in more gas. They employed hydroxides in suspension, and also salts. Lottermoser, acting on mercuric cyanide and copper glycine with hydrogen sulphide, obtained stable sols of mercuric sulphide and copper sulphide, the other products being hydrocyanic acid and glycine (amino-acetic acid), both non-electrolytes. He also prepared organosols from the same substances by substituting organic liquids for water as solvent. He also discovered later that, at sufficiently low con- centrations, silver nitrate gives sols with many salts which CRYSTALLISATION METHODS 199 form insoluble silver salts, provided that one or other of the reagents is present in slight excess (see p. 110). In this way he prepared many silver sols, e.g. AgCl, AgBr, Agl, AgCN, AgCNS, AgOH, AgaCOg, AgjCrO^, AggS, etc. Solution processes may also play a part in some at least of these preparations. Many sols prepared by double decompositions are not very stable, and none are so except at great dilution. If more concentrated sols are required, an emulsoid must be added to confer stability. For instance, Paal's protalbic acid and lysalbic acid methods are equally effective for other insoluble metallic compounds. By the action of the appropriate sodium salt on silver protalbate or lysal- bate, Paal prepared sols of AgOH, Ag2C03, AggPO^, AggS, AgCl, AgBr, Agl ; he also prepared Hg20 sol by the same method. Other emulsoids may be used; HgCl, HgBr, Hgl, Ag2Cr04 sols have been made with albumin as pro- tector ; with the aid of gelatine Lobry de Bruyn (1898) obtained sols of AgOl, AgBr, Agl, AggCrO^, sulphides and hydroxides of many heavy metals, lead iodide, prus- sian blue, copper ferrocyanide, manganese dioxide, etc. According to him concentrated sugar solution also protects. In some instances, a protective colloid may be pro- duced by the reaction. Michael observed that the reaction between chloracetic ester and sodiomalonic ester (and other similar compounds) in benzene solution gave clear solu- tions of a pale yellow colour, with a faint opalescence, and concluded that an addition compound had been formed, as no sodium chloride separated out as usual. Paal (1906) showed it to be a sol of sodium chloride, which is pre- cipitated as a gel by ligroin; this gel is re-solated in benzene. An organic bye-product is supposed to act as protector; sols of sodium bromide and iodide have been 200 PREPARATION OF COLLOIDS obtained by similar reactions in anhydrous benzene or ether. The presence of a protective colloid does not seem necessary for the stability of sodium chloride sol in organic liquids, such as benzene, in which it is practically insoluble (see p. 171). As before, a few typical or important preparations are given in detail. A. Electrolyte-free Sols I. Arsenious Sulphide Sol^ Pure arsenious acid is dissolved iu water, which is kept boiling ; in this way a strong solution is obtained. This solution is allowed to flow into a saturated solution of hydrogen sulphide, through which a current of the gas is continuously passing. The uncombined hydrogen sul- phide is removed by a current of hydrogen (Linder and Picton). Schulze obtained very concentrated sols by adding arsenious oxide to the sol, after the saturated solutions had been treated with hydrogen sulphide, then adding more H2S. After several repetitions the sol contained 37 per cent, of arsenious sulphide. II. Zinc Sulphide Sol^ A current of hydrogen sulphide is passed into water containing in suspension pure zinc hydroxide, which has been precipitated by ammonia and washed with cold water by decantation. The sol is very slightly opal- escent. Copper sulphide sol is prepared in a similar way. ' Linder and Pioton, Chem. Soc. Journ., 1892, 61, 137; Schulze, J. pr. Chem., 1882, [ii.] 25, 431. » Winssinger, Bull. Soc. Chim., 1888, [ii.] 49, 452. CRYSTALLISATION METHODS 2oi III. Mercuric Sulphide Sol * Hydrogen sulphide is passed into a cold saturated solution of mercuric cyanide (12 g. in 100 c.c). The hydrocyanic acid cannot be completely removed by a current of hydrogen sulphide or carbon dioxide, and the sol becomes unstable if it is dialysed. The best way is to distil off the hydrocyanic acid under reduced pressure in an atmosphere of hydrogen sulphide. The sol is deep brown in colour. Copper glycine treated in a similar manner gives copper sulphide sol. Organosols are obtained by using ether or benzene as solvents instead of water. B. Sols containing Electrolytes I. Silicic Acid Sol ^ A solution of sodium silicate (112 g.) is poured into an excess of dilute hydrochloric acid (67*2 g. HCl), the total amount of water being 1000 c.c. The sodium chloride and the excess of acid are dialysed away ; the dialysis is complete in four days (no precipitate with silver nitrate). The sol contained 60*5 g. of silicic acid, 6'7 g. had escaped during dialysis, and the concentration was 4-9 per cent, of silicic acid. II. Antimonious Sulphide Sol ^ 4"3 g. antimonious oxide and 13 g. of tartaric acid are made up to 1000 c.c. of solution; this wiU. give a 0'5 per cent. sol. The solution is saturated with hydrogen ' Lottermoser, J. pr. Chem., 1907, [ii.] 75, 293. " Graham, Phil. Trams., 1861, 181, 183. » Sohulze, J.pr. Chem., 1883, [ii.] 27, 320. 202 PREPARATION OF COLLOIDS sulphide gas. The sol is deep red, and may be purified by dialysis. III. Metallic Sulphide Sols ^ For sols of platinum sulphide or gold sulphide, the solution of the corresponding chloride is made as neutral as possible, and hydrogen sulphide is passed in. In spite of the hydrochloric acid, the dark brown sols are fairly stable, remaining unchanged for many weeks. A gold sol contained 0'55 g. AU2S2 in a litre. A silver sulphide sol from silver nitrate contained 0'6 g. Ag2S in a litre; a slightly more dilute sol remained unchanged for over two months. Dialysis increases the stability. IV. Barium Sulphate in Aqueous Alcohol^ Aim. sulphuric acid solution is diluted with 2 volumes of alcohol, and there is added an equivalent amount of a 1 m. barium acetate solution, which has been diluted with 6 volumes of alcohol. A transparent gel of barium sulphate, which forms a clear sol with a large quantity of water, is obtained. If the gel is dried below 40° C. it will still solate in water. C. A Protective Colloid is present I. Gelatine {or Sugar) ^ A 10 per cent, solution of gelatine, purified from electrolytes, is mixed with an equal volume of the solu- tions (0"l-0'05 n.), which by their interaction will produce au insoluble compound. 1 Winssinger, I.e., p. 200. ' Kato, Mem. Coll. Sci. Eng. Kyoto, 1909, 2, 187. IChem. Soc. Journ., 1910, A. ii., 850.] » Lobry de Bruyn, Eec. trav. cUm., 1900, 19, 236; Ber., 1902, 38, 3079. CRYSTALLISATION METHODS 203 II. Sulphide Sols with Casein or Gum Arabic * If hydrogen sulphide gas is passed through a solution of silver salt or cadmium salt, to which gum arable or casein have been added, stable sols of the sulphides are obtained. III. Paal's ProtalUc Acid and Lysalbic Acid Methods 2 (a) For silver hydroxide sols, see p. 188. (&) Hydroxides of Bi, Co, Ni, Fe, Mn ; AggCOg, AggS, AgCl, etc., may all be prepared by the action of the appropriate sodium salt on the corresponding metallic salt of protalbic acid or lysalbic acid. IV. Sodium Chloride Sol in Benzene 5 g. of malonic ester and 0"7 g. of sodium are added to 30 c.c. of anhydrous benzene, cooled, and 4 g. of chloracetic ester added; and heat is applied till all has dissolved. The orange coloured opalescent liquid is cooled and the gel is precipitated by adding 5-6 volumes of petroleum ether. It solates completely in benzene, but is insoluble if dried in a vacuum. The dried substance contained 58 per cent, of NaCl, and 26 per cent, of Na. There is thus excess of sodium, and a large amount of (adsorbed) organic substance. » Muller, Oesterr. Ckem. Zeit., 1904, 7, 149. 2 Ber., 1906, 39, 1436, 2859, 2863 ; ibid., 1908, 41, 51, 58. CHAPTER XIV SOLUTION METHODS The process generally known as " peptisation " (Graham) is a typical example of the solution method. As the term gelation has been used to express the transformation of sol into gel, solation might appropriately be applied to the reverse process. Lottermoser,! starting from his observations on the formation of suspensoid silver compounds, in which he found that excess of one or other ion was necessary for the production of the sol (p. 110), considers that a peptiser (or sol-former) must contain one, or, in some cases, either of the ions of the disperse phase. Thus, silver chloride, as a fine precipitate, forms a sol on digestion with a solution containing either Ag' or CI'. When aluminium hydroxide is peptised with aluminium chloride solution, it is because of the Al"", or with sodium hydroxide because of the OH'. Miiller ^ showed that ferric chloride, thorium nitrate, chromium nitrate peptise aluminium hydroxide as well as hydrochloric acid does. Equal quantities of aluminium hydroxide (= 1'224 g. of AI2O3) in 250 C.C. of water required for complete peptisation — ' Zeitsch. physikal. Chem., 1908, 62, 359; Koll. Zeitsch., 1908, 3, 31. ' Koll. Zeitsch., 1907, 2, supplement 6-8; Zeitsch. anorg. Chem., 1908, S7, 311. SOLUTION METHODS 205 20 c.c. of 0-05 n. HOI. 1 c.c. of FeCls solution (containing 3 per cent, of FegOj). 4"5 c.c. of Cr(N03)3 solution (containing 1'45 per cent, of CraOg). 10 c.c. of Th(N03)4 solution (containing 2"55 per cent, of ThOs). Calculating back to (H*)Cr or (H')N03', the amounts required for peptisation were — Hydrochloric acid Ferric chloride . Chromic nitrate . Thorium nitrate . 0-001 HCl 0-0012 HCl 0-0026 HNO3 0-0038 HNO3 while the amount of aluminium hydroxide peptised requires 0-036 HCl or HNO3 to form normal salt. These results are not necessarily at variance with Lottermoser's theory, for it is quite probable that the first action between ferric chloride and aluminium hydroxide is the formation of some ferric hydroxide and aluminium chloride. Or ferric chloride, like all the above salts, is hydrolysed into basic hydroxide and free acid, which is thus in a position to act on the aluminium hydroxide with production of Al"'. It is rather curious that hydrochloric acid and ferric chloride are apparently equally effective in peptising aluminium hydroxide. The amount of peptiser required is not proportional to the amount of the precipitate, but depends largely on the previous treatment of the precipitate. Some metallic hydroxides are peptised by alkalies ; thus Hantzsch ^ showed that zinc hydroxide and beryllium hydroxide form sols when treated with alkalies, and do not dissolve in them. Many metallic sulphide precipitates are peptised by treatment with hydrogen sulphide. ' Zeitsch. anorg. CJiem., 1902, 30, 289. 2o6 PREPARATION OF COLLOIDS In all these instances, as von Weimarn indicates, we have, as peptiser, substances which are capable of forming soluble compounds with the substance which is peptised ; e.g. with aluminium hydroxide we have aluminium chloride and the aluminates; with silicic acid, we have silicates formed with alkalies, and with acids, a soluble complex of unknown composition. In the silver salts, soluble complex salts are formed with both silver nitrate and the soluble halides. The solution theory of Noyes and Nernst must there- fore apply to this process, as it does to the crystallisation process. The formula is — V = ^S(L-c) in which L is the solubility, c the concentration of the solution at any given time, and the other symbols have the same significance as on p. 168. [It is the same equa- tion, except that L and c are reversed, in order to apply it to the process of solution (increase of c), instead of to crystallisation (decrease of c)]. If V is large, the sol stage will soon be passed ; V will be small only when D or (L — c) is small. Substances of high molar weight diffuse slowly, and their sol stage is of long duration, as, e.^. gelatine. Even if (L — c) is small, the sol stage will not be stable, unless D is also small, since the larger particles will grow rapidly at the expense of the smaller ones. The Noyes-Nernst formula, however, only applies when the processes are reversible, and when diffusion is slow in comparison with the other processes. If then the solid undergoes a slow change before passing into solution, or if the actual process of solution is much slower than the rate of diffusion, the sol stage may be comparatively stable. Peptisers, as we have seen, are substances which SOLUTION METHODS 207 can, if in sufficient concentration (or if sufficient time is allowed), dissolve the solid, which is insoluble in the dis- persion medium. This being so, the velocity of solution can be slowed down to any desired extent by simply reducing the concentration of the peptiser. "When the sol stage is reached, the peptiser may be removed, wholly or in part, by dialysis. The method is a very general one. As already explained, the crystallisation method indicates how a substance may be obtained as a highly disperse precipi- tate (gel). By means of a suitable peptiser and dispersion medium, this can be converted into a sol. For instance, barium sulphate and carbonate and other insoluble salts of the alkaline earths can be peptised by the correspond- ing acid (which forms soluble acid salts) in a mixture of alcohol and water, in which the disperse phase is sufficiently insoluble to prevent subsequent decrease in dispersity by recrystallisation. Many other examples of a similar kind will be found below. There is another method of preparation which is difficult to place. As the dispersity certainly increases, it has this much in common with the solution process. Svedberg classes it along with peptisation as mechanical- chemical dispersion under the heading " washing-out " method. Many reactions give rise to an insoluble sub- stance which is not obtained as a sol, except with very dilute solutions, because it is precipitated by the other product, which is an electrolyte (see previous section). In some of these instances the insoluble precipitate is con- verted into a sol»when the excess of this electrolyte is removed by washing or dialysis. The phenomenon of a precipitate coming through the filter on thorough washing is well known. There are many other reactions also in which a 2o8 PREPARATION OF COLLOIDS substance is produced in a highly disperse state, and only requires the removal of electrolytes to pass into the sol. Davy.i e.g., reduced boron oxide with potassium, and found that the boron washed through the filter before the potas- sium hydroxide was completely removed. The same happens with silicon and zirconium (Berzelius). The latter also prepared sols of silicic acid and molybdic acid. The silicic acid gel produced by the action of water on silicon fluoride passes into a sol on washing with water. Molyb- denum tetrachloride is precipitated by ammonia as a gelatinous precipitate, for it is insoluble in salt solutions ; when these are washed out, a clear reddish-yellow sol is obtained. The dispersity of the precipitate is of great importance in solution methods. Kiihn ^ noticed that silicic acid gel formed a sol only if it was dilute and had not lost much water. The same difference has been noted between a gel of freshly precipitated antimonic oxide (Sb205,6H20), and one which has been air dried. Berzelius ^ also described the solation of sulphides by washing sulphide precipitates with water. Many other instances could be given, one will suffice. Platinum black, prepared by means of formic acid, often forms a stable sol when thoroughly washed. Berzelius seems to have been the first to observe the solation of a gel in the case of silicic acid. Finely divided silicic acid dissolves in potassium carbonate solu- tion, forming j3 silicic acid, which dissolves in water. The potassium carbonate is not decomposed, for no carbon dioxide is given off ; and when it is neutralised with acid, the silicic acid does not separate out, proving it to ba ^ Schweigg. J., 1811, 2, 48. ' J.pr. Chem., 1853, [i.] 59, 1. ' Fogg. Ann., 1824, [ii.] 2, 139. SOLUTION METHODS 209 dissolved by the water and not by tbe alkali. The sol- ation of stannic acid, titanic acid, and prussian blue were all known to Berzelius. Then come Graham's important researches ^ on peptisa- tion(1861, 1864). He prepared aluminium hydroxide, ferric hydroxide, and chromic hydroxide sols by digestion with the chlorides, titanic acid by means of hydrochloric acid, ferrocyanides of iron and copper by means of oxalic acid or oxalates. The sols were purified by dialysis. Eeference has already been made to A.. Miiller's researches, as also to the peptisation of sulphides by hydrogen sulphide, which was first observed by Spring.^ Kuzel has recently (1907) described an important method of peptisation. The substance, usually a metal, as e.g. chromium, platinum, is very finely powdered by mechani- cal means, and treated with moderately hot dilute solutions of different etching reagents. Usually an acid reagent is followed by an alkaline one. A detailed example is given below, von Weimarn's method for insoluble salts of alkaline earths has already been given (p. 171). A similar method of Neuberg will be found below. I. Boron (silicon, zirconium) ^ An oxide of the element is reduced by potassium at red heat. The fused mass is treated with water, and the precipitate washed with pure water until the filtrate begins to run through coloured. Subsequent washing gives the sol. II. Ferric Hydroxide Sol' A solution of ferric chloride is saturated with freshly > PMl. Trans., 1861, 161, 183. 2 Ber., 1883, 16, 1142. ' Davy, I.e., p. 208. 2IO PREPARATION OF COLLOIDS precipitated ferric hydroxide, or ammonium carbonate is added to it so long as the precipitate redissolves. The red solution is dialysed; after 19 days there were 30'3 equivalents of Fe to 1 of HCl. III. Aluminium Hydroxide^ 50 c.c. of a solution of aluminium chloride (= 2*45 per cent, of AI2O3) are diluted with water and precipitated by ammonia, as in gravimetric analysis. The precipitate is well washed with hot water, and is then transferred to a flask containing 250 c.c. of water. 0'05 n. HCl is added from a burette, the solution being heated to boiling. After each addition water is added to replace that boiled off. The opalescent liquid can be filtered unchanged. IV. Cadmium Sulphide 2 An ammonia solution of cadmium sulphate is com- pletely precipitated with hydrogen sulphide. The pre- cipitate is thoroughly washed by decantation with water, suspended in pure water and a current of hydrogen sul- phide passed in. The precipitate becomes milky and finally disappears. The sol is then boiled until no more hydrogen sulphide is driven off. v. Tungsten 3 10 kg. of tungsten, mechanically powdered as fine as possible, are heated for at least 5 hours, better for 24-48 hours, on a water-bath with 75 kg. of 15 per cent, hydro- chloric acid, with vigorous stirring and frequent renewal of the acid. The acid is run off, and the metal washed by decantation with distilled water until a sol begins to form. • Miillor, I.e., -p. 204. ' Spring, Bull. Acad. Boy. Belg., 1887, [iii.] 14, 312. 8 Kuzel, D. B. P., 186980, 1907. SOLUTION METHODS 211 It is then treated for 5-24 hours with 75 kg. of 1 per cent, potassium cyanide solution. After being washed with water it is agaia heated for 24 hours with 75 kg. of 1 per cent, ferrous sulphate. After complete removal of the iron by washing with water, the metal is again heated with an alkaline liquid, e.g. a 2 per cent, solution of methylamine, or a 0"5 per cent, solution of sodium hydroxide in alcohol. After two or three repetitions of this treat- ment with acid followed by alkali the tungsten forms a perfect sol with water. By this method sols of chromium, manganese, molyb- denum, uranium, tungsten, vanadium, tantalum, titanium boron, silicon, thorium, zirconium, platinum, and other metals can be obtained. Suitable reagents for the method are — acid : organic acids, phenols, salts with acid reaction, as ammonium chloride, aluminium chloride; alkaline: caustic alkalies, alkaline carbonates, potassium cyanide, ammonia, organic bases as pyridine, methylamine. Other liquids may be used instead of water, as methyl alcohol, ethyl alcohol, glycerol. VI. Barium Carbonate in Methyl Alcohol* Carbon dioxide passed into a solution of barium oxide in methyl alcohol gives a thick gel of barium carbonate. If more gas is passed in, the gel dissolves, forming an opalescent liquid. On evaporation in a vacuum it leaves a clear gel, which re-solates on addition of methyl alcohol. TJnclassified Methods I. Conversion of one Sol into Another One example has already been given in Paal's prepara- tion of silver sol from silver hydroxide sol (p. 189). ' Neuberg, Koll. Zeitsch., 1908, 2, 321, 354. 212 PREPARATION OF COLLOIDS Schneider 1 acted on stannic acid sol with hydrogen sulphide, and so obtained the sulphide sol. Meyer and Lottermoser 2 converted silver sol into silver halide sols by means of free halogens, or of salts which readily give up halogen. Lottermoser also obtained cupric oxide sol from copper sol by oxidation in air. As already stated, the sols of base metals (as prepared by the electrical dispersion methods) are really oxide or hydroxide sols. II. Sols formed by the solution of Alloys (1) Some alloys give rise to a highly disperse sol, when one constituent is insoluble in an acid and the others are soluble in it. Schneider^ made a gold-tin-silver alloy, and treated it with nitric acid. A black powder was left which was well washed with water. This residue gave a beautiful ruby sol with ammonia. On dialysis, purple of Cassius sol was obtained. In this case the stannic acid protects the highly dis- perse gold ; perhaps the alloy is to be regarded as a highly disperse gel (cp. pp. 171 and 176). (2) Wedekind^ reduced zirconium oxide with mag- nesium. A part of the black reduction product begins to go through the filter when treated with moderately strong hydrochloric acid and washed with water. The sol is deep blue and opalescent. i > Zeitseh. anorg. Ohem., 1894, 6, 80. ' J.pr. Chem., 1897, [ii.] 56, 247; ibid., 1898, [ii.] 57, 543. » Zeiisch. Ehktrochem., 1903, 9, 630. CHAPTEK XV ELECTRICAL DISPERSION METHODS There are two distinct methods of electrical dispersion, kathode dispersion and dispersion in an electric arc between two metallic conductors in the dispersion medium. The latter, which is by far the more important, was discovered by Bredig so late as 1898. The former goes back to the times of Eitter and Davy. A tellurium kathode in water with a platinum anode sends out a cloud of brown powder. Using a high potential, Davy obtained a purple-red liquid which spread through the water. This has been fully investigated by E. Miiller.* He found that a rod of pure tellurium as kathode, 5 cm. distant from a platinum anode in pure water, gives at 200 volts brown clouds, which form a brown sol. Selenium and sulphur sols were obtained by partially covering a platinum kathode with fused selenium or sulphur. The sol is formed at the junction of the plat- inum and the other element. High voltages are not necessary ; sols are formed at even 4 volts. Bredig,2 who was engaged in an investigation of the decomposition of organic substances by the electric arc, tried the effect of passing an arc between metal wires under water. Dark-coloured liquids resulted ; gold wires gave red or violet liquids, which were very similar to Zsigmondy's gold sols. Some metals, e.g. mercury and 1 Zeitsch. Elektrochem., 1905, 11, S21, 931. * Ibid., 1908, 4, 51; Anorg. Fermenie, Leipzig, 1901, 214 PREPARATION OF COLLOIDS tin, gave suspensions of low dispersity ; platinum, palla- dium, iridium, silver, and cadmium gave metal sols; thallium gave a hydroxide sol. Later, it was found that many metals give rise to hydroxide sols only, as zinc, iron, aluminium. Metals with high solution pressures (> hydrogen) are not likely to yield metal sols in water. When Bredig substituted organic liq[uids as the dispersion medium, another difficulty was encountered, the de- composition of the liquid with production of carbon, of which the sol contained a large amount, e.g. 75 per cent. of the solid in a platinum sol was carbon. Burton dis- persed several metals (copper, lead, bismuth, tin, iron, zinc) in methyl and ethyl alcohol, and platinum, silver, gold in ethyl malonate, but all these sols were very impure. These difficulties .were satisfactorily overcome by Svedberg.i In the first place, in applying Bredig's method to the preparation of organosols, he found that increasing the active surface of the metal was favourable to sol formation. Insertion of a condenser in parallel with the electrodes was a still further improvement. These methods worked with the kathodically softer metals, silver, gold, copper, lead, tin, but failed with the "hard" metals, e.g. aluminium. Very positive metals, as calcium, caused greater decomposition than negative metals; the catalyser metals, cobalt, nickel, platinum, also gave rise to strong decomposition, while bismuth and tin had the least effect in this direction. Svedberg gives a list of 28 metal sols in isobutyl alcohol, with their colour and relative stability. The sols are mostly dark coloured, perhaps because of the » Ber., 1905, 38, 3616 ; iUd., 1906, 89, 1705 ; Koll. Zeitsch., 1907, 1, 229, 257; ibid., 1908, 2, xxix-xliv; Methoden zur Herstellung KoUoider LBswngen, Dresden, 1909. ELECTRICAL DISPERSION METHODS 215 0'2 — 2 per cent, of carbon which they contain. A few examples may be given. Colour. Metal. Transmitted. Eeflected. Stability. Ag . . . green brown black 24 hours Al . . . brown black black 00 Au . . . dark violet black 28 hours Cd . . . brown grey black 00 Cr . . . brown black black 1-2 weeks Cu . . . grey black blue black 20 hours Pt . . . brown black black CO Zn . . . brown red grey black 00 C . . . yellow brown grey brown 00 P . . . colourless flesh — Se . . . vermilion pale red 1-2 days Si . . . brown yellow dark grey 00 Te . . . brown black a few hours Svedberg then used oscillatory discharges instead of a direct current ar», and with care was able to prepare pure metal sols in water and other liquids. By means of it, organosols of the alkali and alkaline earth metals have been obtained. The best conditions for sol formation are — Capacity . . . Self-induction Ohm's resistance Length of spark as large as possible as small „ Related metals are dispersed to a similar extent ; ease of dispersion in any series increases with the atomic weight. The order is — Cu,Ag,Au; Mg, Zn, Cd; AI, Th; Ni, Pt. As already mentioned, the amount of decomposition of the dispersion medium depends on the nature of the metal. 2i6 PREPARATION OF COLLOIDS The following figures for alcohol sols indicate the superiority of Svedberg's method. Comparison of Alcohol Sols prepared ly Bredig's Method {Degen), and ly Svedberg. Percentage of Carbon. Metal. Degen. Svedberg. Bi — 0-25 Cd 15 0-33 Mg 62 ' 1-4 Pt 73 1-2 Al — 1-6 1. Gold Sol and Silver Sol by Bredig's Method An ammeter (A), a rheostat (B), which will afford 4-12 amperes with the voltage (220 v.) and two electrodes + - A Pig. 13, of pure gold wire, 1 mm. in thickness, and 6-8 cm. in lengthj are connected in series with the lighting circuit. The gold wires are pushed thr6ugh narrow glass tubes for convenience in handling. The resistance is adjusted until ELECTRICAL DISPERSION METHODS 217 8-10 amperes pass with a 1 mm. arc between the wires under water. The wires are momentarily touched and drawn apart. A better sol is obtained if a O'OOl n. solution of sodium hydroxide is used instead of pure water, but in either case salts should not be present. So long as the ar^ passes, purple red clouds are sent out from the wire ; if the arc is extinguished, the wires are short-circuited again. The metal comes from the kathode alone ; the anode usually gains slightly in weight. In this manner Bredig prepared sols of gold, silver, platinum, iridium. Silver sol is the easiest to prepare : — "Wires of 1 mm. diameter easily disperse in pure water, without addition of alkali, with a current of 4-8 amp. The dark sol is sometimes red-brown to deep olive-green in colour ; the dilute sol is yellow. They are very stable." Cadmium sol could only be prepared with pure water and in an oxygen-free atmosphere. 2. Billiter's modification of Bredig's Method > A thin layer of the metal is electrolytically deposited on another metal, e.g. lead, copper, nickel, zinc, on iron; iron on zinc ; mercury on zinc or iron ; or aluminium foil is wrapped round zinc or iron wire. These are used as kathodes in Bredig's process, the anode is iron or aluminium wire. All the sols thus obtained are brown in colour ; most of them are oxides or hydroxides. 3. Svedberg's First Method Small pieces of thin metal foil are suspended in the dispersion medium in a cylindrical vessel. The electrodes ' Ber., 1902, 35, 1929. 2l8 PREPARATION OF COLLOIDS are wires of " hard " metal, e.g. iron or aluminium, and are connected with the 220 volt circuit. The foil is kept in rapid movement, and sparks readily pass from one piece to another. The current is read off on an ammeter, 20-50 milliamperes being sufficient. If a condenser of 0-32 microfarad capacity is joined in parallel with the electrodes, the dispersion is quieter and there is also less decomposition of the liquid. By this method sols of gold, silver, copper, lead, tin can conveniently be obtained in water or organic liquids, as alcohol, acetone, ether, chloroform. 4. Svedberg's Second Method Earth Fig. 14. The secondary terminals of a large induction coil (12-15 cm. spark) are connected with the electrodes, and in parallel with a glass condenser of 225 sq. cm. surface. The electrodes dip into the liquid contained in a basin, at the bottom of which is placed the metal in the form of grains or sroall pieces (Tig. 14) ; the electrodes need not be ELECTRICAL DISPERSION METHODS 219 of the same metal. Dispersion takes place rapidly and quietly, and many sols can be prepared in this manner. The contents of the basin may be kept in gentle motion by means of one electrode. 5. Sodium Sol in Ether The apparatus consists of a small flask with a long neck, and a hard glass tube which just fits into the neck. A rubber tube slipped over the junction makes it gas- tight. The lower end of the tube is closed up except for one or two minute holes, which afford com- munication between the tube and the flask. Platinum wires are stretched on the inside of the tube from bottom to top, where they are fused in. Dry ether and small pieces of sodium, the latter to a depth of 0'5-2 cm., are placed in the flask. The top of the tube is connected with a supply of pure dry hydrogen, which is passed through the apparatus for 1-5 hours, after which the ether is sucked by means of a pump into the tube, and the current turned on. For other metals, as potassium, rubidium, caesium, a low temperature is necessary. The flask is placed in a mixture of solid carbon dioxide and acetone, or, better, of liquid air and alcohol, in a Dewar vessel. The ether must be very pure and dry. Pig. 15. 220 PREPARATION OF COLLOIDS Other liquids may be used, e.g. pentane, methane, chloroform. Isobutyl alcohol is the most generally useful alcohol, as lower ones may give alcoholates, and higher ones are decomposed. 6. Electrical Dispersion of Ifon-inetals (Svedberg) Any solid element or compound which has a moderate electric conductivity can be dispersed by this method, e.g. carbon, silicon, selenium, tel- lurium; and minerals, as magnetite, copper sulphide. The following is a con- venient form of apparatus for poor conductors : a glass fun- nel with a ground edge is fitted with a conical vessel of aluminium, in which is placed the powder and the liquid. The funnel is covered with a perforated glass plate, through which an aluminium wire is led. The glass funnel acts as a convenient insulator ; the conical vessel and the wire are connected with the induction coil, and the wire may also be earthed. The powder is' dispersed as sparks pass between the electrodes. Sols of sulphur, phosphorus, mercuric oxide, copper oxide, and prussian blue in isobutyl alcohol were prepared with this apparatus. The sols are very dilute, but can be concentrated by evaporation. Fig. 16. PART III ADSORPTION CHAPTER XVI SUBFACE PHENOMENA Some justification may seem necessary for the departure, in this section, from the principle adopted in the rest of the book, in which the phenomena are described first, and then the various theories and explanations which are deemed of sufficient importance are discussed. Here what is to some extent the reverse procedure has been preferred, for the reason that it seemed of the utmost importance to have a guide through the mass of facts and observations, and because there is, in this case, one central idea round Avhich these ascertained facts can be ranged, and which therefore effects the necessary simplification. This essential fact is that there is always a difference in concentration at the boundary between two hetero- geneous phases, whether the phases be gas, liquid or solid. This " surface concentration " is the principal factor in adsorption, from which all the generalisations may be deduced; the deviations from these laws are due to secondary disturbing factors, which vary from case to case; or, in certain instances, the phenomena are not really adsorptions, but bear a more or less close general resemblance to them. 222 ADSORPTION This being the position, we will in the first place ascertain the facts regarding surface concentrations, and to do this, we must take into consideration surface energy and surface tension. From them we will deduce the empirical generalisations regarding adsorption, and finally, will consider the other phenomena which are liable to be classed as adsorptions. Surface Energy and Surface Tension. — The surface of a liquid, i.e. the interface between it and the gas phase, whether its own vapour or any other gas, is the seat of a form of energy, the surface energy. This energy may be regarded as characteristic of the liquid state, just as volume energy is of the gas state. Its two factors are surface tension and the area of the surface, the former being the " intensity " factor, and the latter the " capacity " factor; the corresponding factors of volume energy are pressure and volume. Surface energy is therefore the product of the area and the tension — surface energy = surface X surface tension. It may be well to note here that, contrary to the case of gases, where the energy decreases as the volume increases, the surface energy of a liquid increases with increase of surface, and that, therefore, the surface of a liquid tends to diminish spontaneously, while to extend the surface energy must be supplied to the liquid. This, as we shall see later, is of fundamental importance in the consideration of surface concentration and adsorption. Owing to the small numerical value of the intensity factor, surface energy forms an insignificant fraction of the total energy of a liquid, unless the development of the surface is extraordinarily great. Specific Surface. — The relation of the surface of a system to its volume can be conveniently expressed by the specific SURFACE PHENOMENA 223 surface, a conception introduced by Wo, Ostwald. It is obvious that the actual surfaces of equal volumes of a substance may' vary enormously, from the minimum surface of a single sphere to indefinitely high values in the; case of thin sheets, rods or very small particles, whether drops or powder. The specific surface is the surface divided by the volume. For a cube with length of side I, the surface is &l^, and the volume is l^, and the specific surface is &jl, from which it follows that a 6 cm. cube has unit specific surface. Surface energy is negligibly small, unless the specific surface is at least 10,000. Surface Tension of Pure liquids. — ^For a detailed account of surface tension and the methods of measure- ment, the larger textbooks of Physics or of Physical Chemistry should be consulted. All that can be done here is to present those aspects which are of immediate importance from our present point of view. The methods of measurement fall into two classes, the static and the dynamic methods. The former include the rise in a capillary tube, the weight of the hanging drop, the adhesion of a disc to the surface, and direct measurement of the curvature of the surface. The latter include the oscillating jet, the oscillating drop, and the capillary waves methods. As a rule all the methods give concordant values in the case of pure liquids, but not in the case of solutions, for which the static methods do not agree with the dynamic methods, although each class gives practically concordant results. The numerical values of the surface tension of liquids are usually small ; for water, which has a greater surface tension than most liquids, it is 77 ergs at 0° C. ; i.e. 77 ergs are necessary for the production of 1 sq. cm. of water surface. 224 ADSORPTION Below are given the surface tensions of a few typical liquids. Surface Tension. I.— Liquids at 20° C. Water . . . . . 73-0 Formic acid 37-1 Methyl alcohol . 23-0 Acetic „ 23-5 Ethyl . 22-0 Propionic „ 26-2 Propyl . 28-6 n-Butyric „ 26-3 Isobutyl „ . 22-5 Aniline . . 43-8 Isoamyl . 23-5 Pyridine 38-0 Ether . . . 16-5 Nitrobenzene 41-8 Acetone . . , . . 23-0 Benzene 28-8 Toluene . . 28-2 II.— Metals. Mercury. . 15° C. 436 Cadmium . 365° C. 810 Sodium . . 90° 520 Iron . . . 1200° 1000 Tin . . . 277° 610 ■ III.— Fused Salts. KI . . . 700° C. 86 KNO3 . . 338° C. 110 KCl . . . 790° 100 NaNOg . . 339° 106 IV. — Liquefied Gases. H2 . . . -252° C. 2 CI2 . . - 72° C. 33-6 N2 . . . -196° 8 CO . . —190° 11 O2 . . . — 1 33° 13 CO2 . +16-2° 2 The Temperature Coffiecient. — The variation of surface tension with the temperature is given by the formula — (Ti = oqO- — at) from which there must be a temperature at which ct = 0. This is most probably the critical point (Frankenheim), at which the difference between liquid and gas disappears. As a matter of fact, the critical temperature can be SURFACE PHENOMENA 225 calculated fairly well by determining the surface tension at two temperatures, and extrapolating to (t= 0. It also follows that liquids which are near their critical points, e.g. condensed gases, will have small surface tensions, while liquids far removed from their critical points, as molten metals and fused salts, will have large surface tensions. The value of a also depends to some extent upon the value of a; for ordinary liquids a = 003, while it is much larger for substances with small a, as gases, for which a = 0"015, and much smaller for those with large a, as metals, for which a = 0003. The Molar Surface Energy. — Eeference may be made to this in passing, as it is of interest and importance. The molar surface energy is a coUigative property of liquids, and is as general as the gas constant E is for gases. The molar surface energy is the product of the molar surface and the surface tension. If V is the molar volume the molar surface is proportional to V^, since the volume of a sphere is as the cube of the radius, while its surface is as the square ; the molar surface energy is thus VV, omitting the numerical factor v'SBtt. Ebtvos and Kamsay and Shields found that it decreases proportionately with rise of temperature, becoming zero at the critical temperature (see above), and that the temperature coefficient is the same for all liquids. If then Wt is the molar surface energy at the temperature t, and Wq that at 0°, we have — Wt = Wq — Bi where B is a constant independent of the nature of the liquid. This equation becomes identical in form with the gas equation if the temperature is counted from the critical Jemperature as zero, at which a = 0. 226 ADSORPTION Expressing these temperatures by D, and the molar surface by a>, the equation becomes — CTco =BD analogous to pv = ET , The value of B is 2* 12 in absolute units. By means of it we can determine the molar weights of liquids, and if they are associated liquids, the deviation in their value of B affords a measure of the degree of association. Thickness of the Surface Layer. — The thickness of the surface layer can easily be deduced from the fundamental consideration of Stefan, that to bring a liquid particle into the surface requires half the work necessary for its transformation into vapour. From this the total amount of energy which can be imparted to a liquid in the form of surface energy is JW (W being the heat of vaporisa- tion), while the surface tension gives the energy required to form unit surface. In the case of water ^W = 2'5 X lO^" erg for 1 g., and CT = 77 erg, from which 1 g. of water can cover 16 X 10'' sq. cm., and the thickness of the layer is 0-6 X 10-8 cm. This also is the limit at which the properties of the substance perceptibly differ from those of the substance in bulk. This will become more apparent if we consider the variation of vapour pressure of a liquid with the curvature of the surface. Vapour Pressure of Curved Surfaces. — Since the effect of surface tension is to reduce the surface to a minimum, and since the surface of two drops of a liquid is greater than that of the united drop, while the volume and mass remain the same, the tendency is for a system consisting of an assemblage of drops to unite into a single drop. This may be effected either by coalescence of the drops by SURFACE PHENOMENA 227 actual contact, or by the transference of vapour from the smaller to the larger drops. The simplest method of getting at the difference of vapour pressure between a curved surface and a plane surface of a liquid is as follows : — •A capillary tube is placed in the liquid, which is contained in a vessel so large that the surface in it is plane. The height to which the liquid will rise is rd a being the surface tension, r the radius of the tube, and d the density of the liquid. The vapour pressure at the curved surface in the tube is equal to the vapour pressure at the plane surface, less the hydrostatic pressure of a column of the vapour, whose height is h. This pressure is equal to JiD, if D is the absolute density of the vapour. Thus dp=p—pi=hJ)=^J) The meniscus of the liquid is concave, and dp is negative, i.e. the vapour pressure at a concave surface is less than that at a plane surface. Conversely it is greater to the same extent at a convex surface. Thus the smaller the drop the greater its vapour pressure ; consequently small drops will distil isothermally over to the larger ones. The actual differences are very small, e.g. the vapour pressure of a drop of O'OOl mm. radius is about 1 per cent, greater than that of the plane surface. The Pressure in the Surface layer — The surface layer is under a greater pressure than the bulk of the liquid, owing to the contractile force of the surface tension. It can be calculated in the following manner : — 228 ADSORPTION A quantity of the liquid under the pressure p is con- nected with a spherical drop of the same liquid by means of a capillary tube. In equilibrium the pressure p, which tends to increase the size of the drop, is counterbalanced by. the surface tension which diminishes the surface, and consequently the size of the drop, i.e. for a small change in radius of the drop dr, the work done by the pressure pdv is equal to adw, the opposing work done by the surface tension, v being the volume of the drop, and w its surface. We thus have adw — pdv = since increase of volume and increase of surface denote loss and gain of energy respectively. , For the change in radius dr, the increase in volume dv is iTrr^ . dr, and the increase of surface dw is Snr . dr. Thus a . Snr .dr^p. 4t7Tr^ . dr whence » = — In the case of water, with a = 77 erg, p = 154/n Since atmospheric pressure is about 10^ units, a drop of water whose surface pressure is 1 atmosphere will have a radius of 1"5 X 10 ~* cm. or a diameter of 0"003 mm. or 3/1. This must be reflected in the other properties, e.g. the density. The compressibility of water is 0'00005 for 1 atmosphere, hence the density of the above drop will be O'OOOOS times greater than that of water in bulk. The other properties are changed to a proportionate extent. Surface Concentration. — From these and other con- siderations, for instance, of the " internal pressure " of liquids which is produced by the great difference in the molecular forces on the two sides of the interface, and SURFACE PHENOMENA 229 which attains enormous magnitudes, it has been con- cluded that the surface layer is a layer of great com- pression, a conclusion which has many important corollaries. One illustration must suffice. If the surface is a layer under great pressure, the concentration in the surface of a solution must be different from that in the bulk of the liquid, or briefly the " volume concentration." For just as unequal temperatures in a dilute solution cause an unequal distribution of the solute, a fact experimentally demonstrated by Soret, and later theoretically deduced by van't Hoff, who found Soret's results to be in agreement with the gas law as applied to solutions, so from the same law unequal pressures at constant temperature must also produce an unequal distribution. Only here, owing to the peculiarities of hydrostatic pressure, it is not so easy to perform the corresponding experiment. The subject will be referred to later (p. 250), where the converse view is adopted, that the . experimentally proved differences between surface concentration and volume concentration in solutions are taken as proof of the existence of a layer of compression in the surface. It is plain that in a pure liquid too, the surface concentration will be greater, if the surface layer is under great pressure. The Surface Tension of Solutions. — There are thus important differences between the surface tension of pure liquids and of solutions, which until recently had not re- ceived the attention they merit, although they were clearly indicated in the conclusions reached by Willard Gibbs. As we have already seen, every free liquid surface tends to diminish its energy. In the case of a pure liquid this can take place in only one way, viz. by the diminution of the surface (the capacity factor), since the 230 ADSORPTION intensity factor, the surface tension, is a constant. But in a solution, it may be effected in this manner, or by decrease in the intensity factor, the surface tension. And this in general can always occur, for increase in concen- tration of the one component will produce an increase of surface tension, and of the other component a decrease. A solution can therefore always decrease its surface tension, and its surface energy, by the appropriate change in con- centration of the surface layer. In other words, a solution can diminish its surface energy, with a fixed area of surface, by an excess of that component which reduces the surface tension entering the surface layer. Thus the composition of the surface layer of a solution is different from that of the rest of the solution, the "molar fraction" of that component which lowers the surface tension being increased. It is desirable, for the sake of clearness, to adhere to the use of the term "molar fraction" in expressing the composition of a solution, thus avoiding the arbitrary and unjustifiable distinction between the components, which the terms " solvent " and " solute" imply. If a solution contains 100 mols of solvent and solute together, of which a mols are solute, the molar fractions of solute and solvent are a and 100 — a respectively. It is clear from the above that the surface may contain either an excess or defect of solute. There is an important difference between the two cases; the former, surface excess, may be very large even in dilute solution, but surface defect is necessarily small. Static and Dynamic Surface Tensions. — This surface concentration manifests itself in many ways, one of the most important of which is that since time is required for the necessary readjustment of concentrations when a fresh surface is produced, or an existing one extended, the SURFACE PHENOMENA 231 surface tension of solutions is not constant, but varies with the time which has elapsed since the surface was produced. In consequence of this, the various methods of measuring" surface tension usually give different results according as they depend on dynamic or static methods (p. 223). In the former fresh surfaces are always being formed, while in the latter the surface is an old one. The dynamic surface tension is nearer that of the solvent, while the static surface tension is usually smaller, since a solution usually has a smaller surface tension than the solvent. The values of the former will not as a rule be the real dynamic surface tension, but will lie somewhere between it and the static surface tension. The latter presents no special difficulty, except in certain cases, where the difference in concentration is accompanied or followed by slow chemical changes, as with the soaps, where the surface tension may continue to change for hours or days. The following figures illustrate the diff'erences between the dynamic and static surface tensions of some solutions. Water . . . Sodium oleate . , Heptylic acid . (T (Dynamic), a (Static). . . . 75 75 . . . 26 79 . . . 54 68 The static values were obtained by the capillary tube method, and the dynamic values by the oscillating jet method. Sodium Sulphate in Water. Static (capillary tube). t = 10°. % 74-0 4-01 74-8 5-07 751 6-53 75-6 Dynamic (capillary waves). t = 15". % 73-3 5 72'8 10 73-4 15 72-6 CHAPTER XVII SURFACE CONCENTRATION The Surface Concentration Formula. — The equation con- necting the surface concentration and the surface tension of the solution was first developed by Willard Gibbs (1876-78)/ later by J. J. Thomson (1888),2 and in the form given here by Milner (1907).* Consider a solution, the surface and volume of which are capable of independent reversible alteration, the latter by means of a semi-permeable membrane, which separates it from water. If s is the surface, v' the volume, a the surface tension, and;^' t^^ osmotic pressure, the work done on the system by increasing s by ds at constant v is ads, and that done by increasing v by dv at constant s is —pdv. Since the final result is independent of the order in which these reversible changes are performed ads — (p-\- rr-dsjdv = —pdv -\-(a-{- -j-dvjds , . , da dp dv ds i.e. the surface tension will vary with the volume (and therefore with the concentration) only when the osmotic pressure depends on the surface. ' Trans. Connecticut Academy, vol. III., 439. ' Applications of Dynamics to Physics and Chemistry, p. 190. » Phil. Mag., 1907, [vi.] 13, 96. SURFACE CONCENTRATION 233 The surface excess u is derived as follows : u is the number of mols in each sq. cm. of surface withdrawn from the volume, and ineffective on the osmotic pressure. If N is the number of mols in the original solution, the volume concentration, on which alone a and p depend, is — N — us c = V Changing to c as the variable we have — d(7 da dc c da dv~ do dv~ V dc and dp dp dc u dp ds dc ds V dc from which da It dp do ~ c dc If the gas law holds, ^ = ET ; or if not, J = iRT, whence dc c c da The relation between concentration and surface tension is perhaps more easily derived in the following manner.^ The surface w of the solution contains 1 mol of the solute in virtue of surface concentration, and a is the surface tension. A very small quantity of solute enters the surface from the solution and diminishes a hj da; the corresponding energy is wda. Now this energy must exactly balance the energy required for the separation of the same quantity of solute from the solution, i.e. to 1 Ostwald, Outlines of General Chemistry, 3rd English Edition, 1912, p. 499. 234 ADSORPTION overcome the osmotic pressure due to this quantity of solute in the solution. If ■;; is the volume which contains unit weight, and dp is thejdifference between the osmotic pressures of the solution before and after its removal, the energy is vip. We thus have — wdxj + vdp ■=■ If the gas law is assumed to apply to these solutions, V = RT/p, hence wda = df) P da RT or -J- = dp wp and since p is directly proportional to the concentration, the formula gives the desired relation between a and c — da_ _RT dc cw As w is the surface which contains 1 mol excess of solute, u the excess of solute in unit surface is Ijw, which gives finally — c da "~ ~RT Ic It should be noted that the surface tension concerned here is the equilibrium, i.e. the static, surface tension. This equation shows that if a- increases with increase of concentration, u is negative, i.e. the molar fraction of the solute in the surface diminishes, while if a diminishes with increase of concentration, u is positive, and the molar fraction of solute in the surface increases. The same con- clusion has already been reached qualitatively (p. 230). Verification of the formula presents serious difficulties, SURFACE CONCENTRATION 235 and so far only qualitative agreement has been attained, i.e. the molar fraction changes in the direction indicated by the equation. The experimental demonstration of the equation, which was first derived by W. Gibbs in 1876-78, was long in being realised. In 1896 the author, working in Ostwald's laboratory, observed that the electric conductivity of a dilute aqueous solution of sodium oleate changed when the solution was stirred up with the electrodes, so as to produce foam on the surface, and gradually returned to its initial value as the bubbles slowly collapsed. This could be repeated as often as desired. Attempts to separate the foam on a large scale, in order to prove the difference by analytical methods, were unsuccessful. Zawidski ^ (1900) effected the analytical proof in the case of salts and acetic acid by the simple expedient of adding saponin, and analysing the foam carried over by bubbles of air. The differences in concentration were considerable. Miss Benson 2 (1903), working with aqueous amyl alcohol, which, like soap solutions, does not require extraneous substances to produce foam, found the surface concen- tration to be 0'0394 n., the original concentration being 0-0375 n. Milner's ^ results are as follows : — Sodium Chloride. — The increase in a is 1"72 dynes per sq. cm. for increase in c of 1 mol per litre, i = 1'74, E = 8-32 X IC erg per degree, T = 290, whence u =— 4-09 X 10~ii mol per sq. cm., or 0'024 mg. per square metre. Acetic Acid.- i = I'OS, and ^t = +3-3 X 10-1° mols per sq. cm., or 0'20 mg. per square metre. 1 Zeitsch. physikal. CHem., 1900, 35, 77. 2 Amer. J. Phys. Chem., 1903, 7, 532. ' hoc cit., p. 232. 236 ADSORPTION Sodium Oleate. — No finite value for ^- could be ob- dc tained, as the following figures show : — Relative Belative surface surface Conoentration. tension. Concentration. tension. 0-02 . . . . 2-330 0-00225 . . . , 2-305 0-066 . . . . 2-300 0-00206 . . . . 2-240 0-0082 . . . . 2-275 water . . . . 6-15 u was estimated by determining the decrease in electric conductivity produced by bubbling air through the solu- tion and removing the foam. This estimate came to about 1-2 X lO"!" mol per sq. cm., or 0-4 mg. per sc[uare metre. Milner observed that the static and dynamic surface tensions of the soap solution were very different. In spite of this want of quantitative agreement, the generalisation is one of extreme importance in its bearing upon adsorption, which is primarily and in the main a matter of surface extension, and therefore of surface con- centration. Before passing on to adsorption it is accord- ingly necessary to inquire into the surface tension of solutions, with especial reference to the variation of surface tension with concentration. Surface Tension of Solutions ; the a, c curve. — Eefer- ence to the table of surface tensions on p. 224 will show that, of ordinary liquids, water has the largest surface tension, and clearly only a solvent with a comparatively large value of a can show much variation of a, at least in the direction of diminution. And so it is that, apart from other properties of water as a solvent (unless, indeed, these are connected in some way with the large value of a), water differs markedly from many other solvents in that SURFACE CONCENTRATION 237 very large decreases of surface tension do occur, and that most aqueous solutions have a smaller surface tension than water. All the possible varieties of curve between two components with different surface tensions are known, but for the present, at least, these variations of type are of no particular moment. Gibbs pointed out that a small quantity of solute may cause a large decrease of surface tension, but cannot increase it to any great extent (see p. 241). Accordingly, the majority of a, c curves lie below the straight line joining the values of o- of the two components ; further, they either pass through a minimum or exhibit a tendency to a minimum. Naturally there are instances where the a, c curve is the connecting straight line, but this only occurs when the two substances are chemically similar and when their surface tensions are not far apart. Surface Tension of Solutions. — The surface tension of aqueous solutions of salts, and especially of inorganic salts, is greater than that of water, and increases linearly with the concentration. This is in accordance with the general rule, for the surface tension of salts is undoubtedly greater than that of water. Solutions of hydrogen chloride and hydrogen bromide, which as pure liquefied gases have smaller values of a, have slightly smaller surface tensions than water. Surface Tension of Electrolytes in Water. t = 18° ; c = 1-5 mol per litre ; )3 1-76 Methyl alcohol-carbon disulphide 18° 0-82 It is plain that the mutual solubility of the liquids has a marked influence here, as might be predicted, for when two liquids become completely miscible at the critical solution point, the interface and therefore the surface tension disappear. The values are also influenced * Loc. cit., p. 257. ADSORPTION 259 to a marked extent by the addition of acids and alkalies, and stUl more by substances like soap. Lewis 1 investigated the adsorption of sodium glyco- choUate and methyl orange in water-paraffin oil, having shown that these substances greatly lower the surface tension, e.g. — Sodium Olycochollate in Water-petroleum. Concentration. (71,2 + "^2,3, t^® second liquid will com- pletely remove the disperse phase from the original sol." (c) If t7j,2 > (^2,3 + <^i,3, 01" if 11° one a is greater than the other two together, the disperse phase will collect at the interface between the two liquids. These conclusions have been verified by Eeinders, both in the case of suspensions and of sols. With the latter • Cp. Freundlich, Kapillarchemie, 1909, p. 137. ' This case corresponds to the " wetting" of a solid by a liquid. If "■«,» > "1,9 + "s,! the liquid will spread over the surface of the solid until there is no direct contact between the solid and the gas. The interface between solid and gas will disappear in favour of the solid-liquid interface and the liquid-gas interface, since the free energy of the former ia greater than the sum of the other two. ADSORPTION 261 any of the three can happen, provided the dispersity is low, but if the dispersity is high, a sol may exhibit anomalies, one part behaving in one way, and the rest in one of the others. In addition, emulsoid sols may form a separate layer between the others. A few illustrations will suffice. In gold hydrosols isobutyl alcohol produced no effect until the separate layer was formed, when the gold separated out at the interface in a thin layer. The colour by transmitted light was violet-blue to blue-green, and by reflected light a brilliant gold. The layer creeps up between the alcohol and the glass, causing the alcohol to appear coloured, although it is not so. Amyl alcohol, benzene, benzine, ether, carbon disulphide, and carbon tetrachloride behave similarly. A gold sol in amyl alcohol (produced l)y reduction of auric chloride in amyl alcohol with phosphorus), which was yellow-brown in colour and very stable, was shaken up with water. No gold entered the water layer, but violet-red gold separated out at the interface and on the glass. Ether sols were prepared in the same manner, but were not stable ; when shaken up with water, all the gold left the ether, forming a yellow-red sol in the water, and a slight brown precipitate at the interface. A freshly prepared hydrosol (from gold chloride and phosphorus in ether) which was brown-red in colour, was quite indif- ferent to ether, no separation occurring at the interface. The effect of gum arable as a "protecting" colloid was as follows : — Gold sol -f- gum arable. 2Dd liquid. No gum. O'OOI per cent. 0*005 per cent. isobutyl blue at Inter- blue at interface, blue at interface, very alcohol face. slow, some red slow, most of the sol in the water. gold red in the water. ether blue at inter- red, all in the red, all in the water. face. water. 262 ADSORPTION As was to be expected, the protection of the gum arabic, which is hydrophile, hinders the transference of the dis- perse phase (hydrophobe) into the alcohol or ether in which the gum is itself lyophobe. Of other sols which were examined, ferrw hydroxide hydrosol was unaffected by any of the other liquids, and arsenious sulphide hydrosol was completely separated out at the interface with isobutyl alcohol and amyl alcohol, and was not changed by the others. Frequently part of an apparently uniform sol would remain in one liquid (usually the original dispersion medium) while the rest was precipitated at the interface. The separation of the disperse phase at the interface appears to be an exceedingly common occurrence, and forms a good criterion for sols ; for example, alkaK blue, the blue acid of congo red, the acid of erythrosin, all go to the interface when the hydrosols are shaken up with carbon tetrachloride. Analogous phenomena must often occur at interfaces between solids and liquids, e.g. gold sol when shaken up with charcoal, barium sulphate, aluminium hydroxide, fibres, and particularly colloid dyes on fibres. They commonly pass under the name of adsorptions, but a more suitable name would be adhesions. As with increasing dispersity there is a continuous change from sols to solu- tions, there is no abrupt division between adhesions and adsorptions. PART IV APPLICATIONS OF COLLOID CHEMISTRY CHAPTEE XIX SEMI-COLLOIDS There is an extensive series of substances which are intermediate in properties between emulsoid sols and solutions (the emulsides and solutides of von Weimarn's nomenclature). In accordance with this position, their character is indefinite, as they exhibit some of the pro- perties of each, and so far as is known, there is no clue as to which class they belong in respect of any given property. Naturally, the first idea is to look for some connexion between the molar weight of the substance and its general behaviour, but this at once fails, for substances of practically the same molar weight, and not widely different composition, are frequently to be found in different classes, as a glance at the table on p. 270 will show. Among the important classes of substances which come into this category are some of the highest impor- tance in technical science, as casein, the soaps, dyes and colouring matters, tannins, degradation products of albumin, etc. While albumin is a genuine emulsoid, the products of the action of alkalies (NaOH) on it, Paal's protalbic acid 264 APPLICATIONS OF COLLOID CHEMISTRY and lysalbic acid (p. 130) are semi-colloids, e.g. the latter, whicli is soluble, diffuses in water and througli parchment paper ; the molar weight, deduced from the freezing point, is 700-800. But the solutions are opalescent, and syrupy when concentrated. Further, they are strongly " protec- tive," as their use in Paal's method of preparation testifies. Peptones, on the other hand, do not " protect," hut diffuse slowly; their molar weight, determined by the freezing point method, is about 500-1000. They are usually regarded as amphoteric electrolytes.^ The electrical behaviour of semi-colloids is of consider- able interest, for many of them are fairly good electrolytes, even when osmotically they are remarkably inert ; this aspect of the question has been but little studied. Owing to the individual peculiarities of the semi- colloids, it is inipracticable to "give a systematic account of them. A brief description of some technically important ones is all that is attempted. Casein. — Casein is insoluble in water, but is essentially acid, as is shown by the reddening produced by pressing moist casein on to blue litmus paper. It dissolves in alkaline solutions, which might be attributed to solation by peptisation; it appears, however, to combine with a definite amount of alkali irrespective of the concentra- tion (1 g. casein requires 0'88 millimol of NaOH). The resulting solution is opalescent, which is usually put down to hydrolysis, casein being reformed. The addition of acids to the alkaline solution produces turbidity, casein being precipitated, while excess of alkali removes the opalescence. From the electric conductivity, i.e. by the applica- tion of Ostwald's rule for determining the basicity of an acid from the variation of the equivalent conductivity (/u) 1 Neumann, Zeitsch. jahysiol. Chem., 1905, 45, 216. SEMI-COLLOIDS 265 of the sodium salt with the dilution (v), Sackur concludes that casein is a tetrabasic acid, which, together with the above equivalence with sodium hydroxide, gives a molar weight of about 5000.1 It does not diffuse through parchment paper, though a substance with this molar weight should diffuse to a measurable extent (p. 36), and should also exert a considerable osmotic pressure. Soaps. — ^The soaps constitute a highly complicated chapter in chemistry. From the present point of view, their interest centres in the fact that they are representa- tive of a class of substances which (1) Are themselves probably non-colloid, but may undergo changes in certain liquids, giving rise to colloids, and hence have been termed coUoidogens ; (2) Exhibit very different behaviour at low and high concentrations ; (3) Are known as electrolyte colloids, for the reason that, while the aqueous solutions are good electrolytes, their" osmotic properties (boiling point, vapour pressure, etc.) are so far abnormal that not only do they not indicate the ionisation, which the electric conductivity leads us to expect, but they differ so slightly from those of the pure liquid, that the idea of solution is precluded. In other words, while the solutions are fair electrolytes^ their boiling point and vapour pressure are approximately the same as for water. Thus, Krafft 2 found no elevation of boiling point for concentrated solutions of soaps, while there were slight rises for dilute solutions. Smits ^ obtained similar results for sodium palmitate at concentrations from 0*3 to 0'6 ' Zeitsch. physikal. Cliem., 1902, 41, 672 ; Beitr. chem. Physiol. Path., 1903, 3, 193. ' 2 Ber., 1894, 27, 1747 ; 1895, 28, 2566 ; 1896, 29, 1328 ; 1899, 32, 1584. » Zeitsch. physikal. Chem., 1903, 45, 608. 266 APPLICATIONS OF COLLOID CHEMISTRY mols per litre. For higher concentrations he used a tensimeter owing to the experimental difficulties of the boiling point method. There was no diminution of vapour pressure in a 28 per cent. (1 mol per litre) solution, and only 1"3 mm. diminution in a 14 per cent, solution. For the osmotic pressure of soap solutions, see p. 314. Concentrated solutions are clear when hot, and gels are formed on cooling. Dilute solutions are frequently opalescent, and the opalescence is increased by heat. All the solutions have electric conductivity ; the soap is there- fore ionised or hydrolysed, or probably both. Hydrolysis is proved by the extraction of fatty acid from the solution by shaking up with toluene ; the hydrolysis is not into acid and base, but into acid salt and base, and occurs even when excess of base is added. In absolute alcohol the soaps give normal molar weights by the boiling point method. A comparative study of the sodium salts of the fatty acids shows that the soap character begins with the Ci2 acid. Oleic acid is more soluble than the saturated acids, and its salts are less hydrolysed. MacBain and Taylor ^ were unable to get any satisfactory determinations of the boiling point (which the author can corroborate from personal experiences), so they studied the electric conductivity at 90° of solutions of sodium palmitate, using silver vessels, as others were attacked chemically by the soaps at this high temperature. They appear to have obtained the same value for the conductivity, whether the solution was made up from solid soap and water, or from acid and sodium hydroxide. This is a very important observation, for it indicates that a true, i.e. a reversible, equilibrium is established between > Zeitsch. physikal. Chem., 1911, 76, 179. SEMI-COLLOIDS 267 electrolytes in solution and colloids in the sol condition. There is a maximum conductivity (89'47) at 0*5 m., and a minimum (82*37) between 0*2 m. and 0"1 m. Addition of sodium hydroxide to the solutions pro- duces a precipitate, which is a mixture of sodium palmitate and an acid salt, N"aHPa2, in spite of the alkaline reaction of the liquid. Whether this precipitation is the precipita- tion of a sol, a salting out, or a gelation, is not very clear. The precipitation of the acid salt by alkali from alkaline solution seems to be similar to the gelation of silicic acid by small quantities of alkali (p. 127). As the reaction takes place with solutions up to 56 per cent. (2 m.), it is' plain that the hydrolysis of sodium palmitate is — 2NaPa -f HOH ^ NaOH + NaHPag and not NaPa + HOH ^ NaOH -1- HPa The nature of "salting out" of soaps by adding a mixture of sodium chloride, carbonate and hydroxide, seems to be as uncertain as the above action of sodium hydroxide. It has been variously described as salting out proper, analogous to the action of ammonium sulphate on albumin, or of many salts on organic substances not necessarily colloid ; as a precipitation of a sol by electro- lytes ; and as a " common ion effect," i.e. the precipitation of a sparingly soluble salt by adding a large amount of a soluble salt with a common ion (e.g. the decrease in solubility of silver acetate on adding sodium acetate or silver nitrate to the solution). Dried soap swells in water, becoming a gel^ ; this process is accelerated by alkali, and this fact is made use of by the manufacturer. The cuttings of soap to be remade up are treated with dilute alkali instead of with water. ' Goldsoliinidt, Koll. Zeitsch., 1908, 2, 103, 227. 268 APPLICATIONS OF COLLOID CHEMISTRY The surface tension of soap solutions is dealt with on p. 236. As the solutions are so complex, it is not sur- prising that the surface tensions are not well-defined. While sodium oleate lowers the surface tension of water to a great extent, sodium palmitate, according to Bottazzi,^ does not lower it at all, and the addition of a little sodium hydroxide produces no change. He concludes that the lowering of the surface tension is due to the undissociated sodium oleate, while the hydrolytic products are much less active, or are quite inactive. Among semi-colloids other than those mentioned above are substances used in dyeing and tanning. Owing to their importance they will form the subject of separate chapters. ' Atti JJ. accad. Lined, 1912, 365 (abstract in Koll. Zeitsch., 1913, 12, 205). CHAPTER XX DYEING The question whether dyeing is a chemical or a physical process is one which has agitated scientists and practical men alike as long as there has been a science of chemistry, and the main effect of the advent of colloid chemistry has been to render the question stUl more complicated by providing yet another set of explanations to those already existing, without increasing materially the prospect of a final answer. Still, it affords several points of attack, and there is little doubt that many of the phenomena of colloids must be concerned in some of the processes of the dyer's art. The fibres cotton, silk, wool, are more or less comparable with gels, both in structure and general properties ; they exhibit the phenomena of imbibition and dehydration, and possess great development of surface. Many dyes are true colloids or semi-colloids in water, and of the others, many greatly lower the surface tension of water, and are thus readily adsorbed from water, as also are the other two classes of colloids and semi- colloids. Notwithstanding this, it is vain at the present moment to expect that colloid chemistry can provide a single or simple explanation of dyeing, or an answer to the question whether dyeing is. a chemical or a physical process. The reason of this is to be found in the question. As Ostwald 270 APPLICATIONS OF COLLOID CHEMISTRY well expresses it, " We are not justified in speaking of ' a theory of dyeing,' as if any single theory can be applied to all the processes of dyeing. Dyeing is not a scientific, but a technical notion, which denotes uniform results without regard to the means employed. Science is bound to take into consideration the means, and, if they are found to be different, it must put forward correspondingly different theories of the dyeing processes which depend on them." ^ The original rival explanations were physical action, with solid solution as a later extension, and chemical reaction between the fibre and the dye. To these are now added colloid reactions and adsorption at interfaces, the latter naturally taking first place, though the mutual precipitation of colloids of opposite electric charges, and the various types of pseudo-adsorption (p. 252) must also be important factors, especially though not exclusively when mordants are used. But before entering on this, the main subject of interest, it is desirable to outline briefly the colloid properties of dyes. The Colloid Nature of Dyes. — As stated in the previous chapter, many dyes form semi-colloids in water, others are undoubtedly colloids, i.e. form emulsoid or suspensoid sols with water as dispersion medium, while others again form true solutions. The following table shows to which group a number of important dyes belong. Colloid. Semi-colloid. Crystalloid. Alkali blue Capri blue Alizarin red Aniline blue (soluble) Fuchsia Auramine Azoblue Methyl violet Bismarck brown Benzazurin Neutral red Ohrysoidin Benzopurpurin Nile blue Eosin ' OutUnes of General Chemistry, 3rd English Edition, 1912, p. 495. - Colloid. Carmine Congo red „ fast blue „ brown Diamine green „ blue Induline (soluble) Naphthol yellow Mght blue Violet black DYEING Semi-ooUoid. 271 Crystalloid. riuorescein Magdala red Methylene blue Patent blue Picric acid Pyronine Ehodamine Safranine Thionine The colloids show distinct microns and submicrons in the ultramicroscope, and otherwise behave as typical sols ; usually the emulsoid or hydrophile character predominates, but in some (e.g. congo red), the dye is precipitated by small amounts of salts, and, as they are mostly negative sols, the precipitating power of the salt depends on the valency of the kation (p. 98). As a rule the sol is electropositive or electronegative according as the colouring substance forms the kation or anion of the dye. The semi-colloids exhibit the irregularities associated with the class to which they belong (see Chap. XIX.) ; their molar weight in water is 2-3 times the normal, while, like the soaps, the molar weights in alcohol are normal.! 'p^Q aqueous solutions are good electrolytes, and submicrons are visible in the ultramicroscope. Michaelis 2 observed that the addition of salts to sols of semi-colloids, and even to solutions of the soluble dyes, produces submicrons, i.e. they shift the dye from one class to the next. Thus a dilute fuchsin contained no sub- microns, but when sodium chloride was added to the hot solution and the solution rapidly cooled, many submicrons were visible. This change no doubt plays an important 1 Krafft, Bur., 1899, 32, 1610. » Virch. Arch. Anat. Physiol, 1905, 179, 195. 272 APPLICATIONS OF COLLOID CHEMISTRY part in the effect of neutral salts on the taking -up of dyes by fibres from the dye-bath ; neutral salts, as is well known, increase in some cases to a remarkable extent the amount of dye taken up by the fibre, and colloid dyes are among the most readily taken up dyes. The chemical theory has failed to supply any plausible explanation of this action. A similar transference from one class to the next more colloid class can also take place in other ways. The dye salt may form true solutions, but give rise, by hydrolysis or other reaction, to colloid base or acid (or it may be to colloid ion) in a manner analogous to the action of water on soaps (p. 265), or by the production of insoluble salts, e.g. of fatty acids, in the form of sols. Precipitation of Dyes by other Colloids. — The mutual precipitation of colloids is strikingly exemplified in the case of dyes.l Among dyes are to be found all types of colloids — suspensoids and emulsoids, .semi-colloids and colloidogens — and salts which form true solutions. Of the three types of action (p. 135) (1) simple precipitation, and (2) precipitation with protection and consequent non- precipitation when excess of either sol is present, are well known; the third type with two zones of precipitation and two of non-precipitation does not appear to have been observed. (1) Eosin (anion) and methylene blue (kation), being electrolytes and non-colloid, precipitate suspensoids sols — eosin, the positive sols such as aluminium hydroxide and ferric hydroxide, methylene blue, the negative sols such as platinum and mastic — so soon as the critical concentration has been reached. A similar action occurs with emulsoids under the conditions which convert them into positive or negative • Buxton and league, Zeiisch. physikal. Chcm., 1907, 60, 4G9, DYEING 273 colloids, the negative emulsoids are precipitated by basic dyes, the positive by acid dyes. Methylene blue precipi- tates sols of hsemoglobin, serum albumin (native and heated), congo red, etc., at high, but not at low concentra- tions. (2) The colloidogen, semi-colloid and colloid dyes with suspensoids and emulsoids, such as gelatine, tannin, albumin, etc., are examples of the second type : a zone of precipitation witb a zone of non-precipitation on each side of it. The possibility of titrating one dye with another, as night blue with eosin, or with other colloids, as night blue with tannin, depends on this reaction, the precipitation zone being very sharply defined, and the ratio of the two substances remaining the same when the absolute concen- tration is changed. The ratio corresponds in the former case to the chemical equivalents, but that there is not true salt formation is indicated by the fact that ' the ratio varies with the rate at which the precipitant is added. Solid Skins on Solutions of Dyes. — Solutions of many dyes, e.g. fuchsin, form a skin on the surface, wben they are kept for some length of time. The nature of this skin is not always known, but is due to increase of surface con- centration. These substances lower the surface tension of water to a very considerable extent, and so, even in dilute solutions, the concentration of the surface layer may be extremely high ; and as they are mainly emulsoid or hydrophile colloids, they form gels or even solid skins on the surface. They may also be formed at the interface of two im- miscible liquid s in which the colloid is insoluble. Eamsden,^ who has investigated this phenomenon, especially in the ' Zeitsch. physikal. Chem., 1904, 47, 336. T 274 APPLICATIONS OF COLLOID CHEMISTRY case of albumins, has applied it to the separation of colloids from liquids by simply shaking up with air or a suitable liquid (toluene or chloroform in the case of albumins). The method applies to albumins, soaps, and saponins, as well as to dyes. This concentration of the colloid in the surface or interface is certainly followed in many instances by irreversible changes, e.g. albumin and dyes become insoluble, in whole or in part, or become incapable of solation, etc. Two classes of substances which are nearly related remain to be discussed ; they are lakes and mordants. We will take the latter first. Mordants. — As a broad generalisation mordants may ■ be said to be " coUoidogenic " substances, i.e. they are themselves truly soluble, but by various changes give rise to colloids, either sols, or more usually gels. These gels produced on or in the fibre of the fabric to be dyed then react with the colouring matter of the dye-bath, whether by chemical action, by adsorption, by solid solution, or, as is extremely likely in some cases, by mutual precipitation of colloids of opposite electric charge, is immaterial at present. By this means dyes which are not fast when used alone are more firmly fixed. The mere mention of the substances chiefly used as mordants is sufficient evidence of their colloid character: aluminium salts, chromic salts, ferric salts, stannic chloride, tannin, etc. Let us take aluminium mordanting, for instance. A solution of aluminium acetate is partly hydrolysed in solution into aluminium hydroxide sol and acetic acid. This, when brought into contact with, say, cotton, will probably deposit some aluminium hydroxide on the surface of the fibre, for the sol is positive and cotton in contact with water is negative ; we should thus expect precipita- DYEING 27s tion of the sol on the cotton. But whether this takes place or not, some aluminium acetate will soak into the fibre, and will there hydrolyse, depositing aluminium hydroxide gel in the fibre, the acetic acid escaping either by diffusion or by evaporation, if the fabric is steamed or dried. If the former action, mutual precipitation of the aluminium hydroxide on the surface, takes place, the formation of the gel in the fibre could take place as follows. The aluminium acetate which has soaked into the fibre before the precipitate has been deposited on its surface, cannot diffuse out again, for the film is impermeable by Al"', but acetic acid can easily dialyse out, leaving an aluminium hydroxide gel impregnating the fibre. Thus the mordant would be both embedded in the fibre, and also to some extent deposited on its surface. The fact that the basic sulphates are good mordants while the basic chlorides are not so, is probably connected with the precipitating effect of the bivalent anion, SO4", on the positive aluminium hydroxide sol in the former, while in the latter the stabilising effect of the hydrogen ion H", or of the Al'" on it prevents its precipitation by the univalent anion CI' (pp. 137, 193). In the case of the basic acetate, which also is a good mordant, the salt is more fully hydrolysed than the chloride or sulphate, and the acetic acid is so weak that it can have only a feeble stabilising effect on the aluminium hydroxide sol. The use of hichromaie as a mordant appears to depend ultimately on the precipitation of chromic hydroxide gel on or in the fibre. The bichromate is decomposed in dilute solution into chromate and chromic acid ; the latter is adsorbed by the fibre and is probably partially reduced by the material of the fibre, but the reduction is completed e.g. by sulphurous acid. Chromic chloride and chromic fluoride are also used as mordants for wool. 276 APPLICATIONS OF COLLOID CHEMISTRY Stannic chloride by similar reactions can deposit stannic acid gel in the fibre, for in water it is practically completely hydrolysed into hydrochloric acid and stannic acid sol. Tannin as a mordant deserves separate mention. It is an emulsoid sol in water, and is readily adsorbed. Its adsorption is increased when free acid is added to the solution, but this soon reaches a maximum, and further addition of acid reduces the amount adsorbed, as the following figures show : — Adsorption of Tannin hy Cotton. Amount adsorbed. Tannin + no acid 30 per cent. + 1 g. acetic acid per litre 35 „ „ + 2 40 + O „ „ ,, 50 „ ,, "T 10 „ „ „ oo „ „ + 20 „ „ „ 31 „ „ lakes. — Tannin forms lakes with basic dyes ; these may be salts, but are more likely formed by the mutual pre- cipitation of the negative tannin and the positive (basic) dye. According to Witt^ there is no definite molecular ratio between the dye and the tannin. The ratio will obviously depend on the electric charge on the particular gel and sol, and will in any case not necessarily approxi- mate to molecular ratios (p. 134). That this is the nature of the reaction is supported by the fact that the tannin lakes are soluble in excess of tannin, for a sol which pre- cipitates when the approximate amount is added protects the other sol if added in excess (p. 133). The lakes proper are formed from alizarin (a weak dibasic acid, dihydroxy anthraquinone, C14H8O4) and 1 Chem. Zeit., 1885, 12. DYEING 277 metallic oxides, of which aluminium hydroxide is the type. The presence of an oxide of a bivalent metal, e.g. CaO, is also necessary for a proper colour and fast dyeing. The chief lake is Turkey red on cotton, which contains aluminium, and calcium, and also a fatty acid. A pure aluminium mordant on wool also requires an oxide of the MO type to give the colour; here magnesium oxide is found to be the best. Pure aluminium rioinoleate powder, if heated with water and alizarin, begins to attract the colouring matter at 40° C, at which it melts, and the bright red colour is gradually developed. The formation of lakes with dyes and aluminium hydroxide may be compared with the action of the latter on red gold sols, which gives bright red precipitates. Thk Theories of Dyeing The essential fact on which the chemical theory of dyeing is based is that there is a decomposition of the dye salt, especially in the case of basic dyes, the coloured katibn being fixed on the fibre, while the anion remains in the bath. This is represented as a chemical reaction ^ between the dye and the fibre, which is stated to be an amphoteric electrolyte, and so functions as an acid to basic dyes, and a base to acid dyes, an insoluble product being deposited on or in the fibre. The following objections appear to be fatal to this theory. (1) The dye is not insoluble, for it can in many cases be readily removed by . treatment with water or other 1 Kneoht, Ber., 1889, 2X, 1556, 2804; 22, 1120. Suida, Sitsungsber. Jcaizerl. Akad. Wiss. Wien., 1904, 113, 724; Zeitsch. Angew. Chem., 1909, 2131. 278 APPLICATIONS OF COLLOID CHEMISTRY liquid, and there is, in fact, reversible equilibrium between the dye on the fibre and in the bath. (2) That the same decomposition occurs with other substrates in which chemical action (amphoteric or other- wise) must be considered improbable. Such are pure cellulose, charcoal, glass, asbestos, and other silicates.^ (3) According to Losev it is doubtful if a similar decomposition of acid dyes takes place, and it certainly does not occur with the substantive dyes, so that in these cases the chemical reaction is quite lacking. As the facts relating to the first of these objections are of fundamental importance, they may be stated at once. In an investigation on the behaviour of picric acid to various adsorbents. Walker and Appleyard ^ found that the amount of dye taken up from the bath varied continuously with the concentration of the bath, relatively larger amounts being taken from weak baths than from strong ones. The formula expressing the relation was found to be — Cpllk 2-V ^water = k This is the well-known adsorption formula. If the dyed silk was treated with water, the action was reversed, and the same end result was obtained whether the dye was on the silk or in the solution. Further, less picric acid was taken up by silk from alcohol solutions, and none at all from benzene solutions. From these results, it would appear that the action of picric acid on silk is purely an adsorption effect, the amount adsorbed depending not only on the nature of the substance adsorbed, but also on that of the solvent. » Georgievies, Farberieit., 1894-5, 19, 9, 129, 188, 286. Preundlich. and Losev, Zeitsch. physikal. Chem., 1907, 69, 284, ' Chem. Soc. Journ., 1896, 69, 1334. DYEING 279 Adsorption generally is greater from liquids whose surface tension is large (as water) than from those with small surface tensions (as alcohol and benzene), the decisive factor being the effect of the particular solute on the surface tension of the liquid (p. 238). It is very difficult to reconcile some of the well-known facts of dyeing with the chemical theory, e.g. the effect of adding neutral salts to the bath (p. 271), but above all the increased dyeing produced by adding bases to basic dyes, and acids to acid dyes. The stronger the acid and the greater its concentration, the greater is the dyeing action. Adsorption of Dyes. — In the chapter devoted to adsorp- tion (p. 243) will be found several references to dyes, which show that the initial stage at least of the reaction between dye and fibre may be regarded as an adsorption. One only will be repeated here. The ratio of the amounts of two adsorbed substances is practically independent of the nature of the adsorbent (p. 246), and this applies equally to fibres (wool, cotton, silk) as to the substances usually employed as adsorbents (charcoal, aluminium oxide, silica, etc.). Again, as pointed out above, there is apparently equilibrium between fibre and solution, just as there is with the other adsorbents. Lastly, the connexion between the concentration of the dye-bath and the amount taken up is expressed by the adsorption formula, the numerical values of the index beiug similar to those of other ad- sorptions. This, the result of the work of Walker and Appleyard, already referred to, of Schmidt,^ and especially of Freundlich and Losev,^ is the main objection to the theory of solid solution, first suggested by Witt.^ 1 Zeitsch. physikal. Chem., 1894, 15, 56. " Freundlich and Losev, loc. dt., p. 278. » mrbermt., 1890-1, 15, X. 28o APPLICATIONS OF COLLOID CHEMISTRY The Solid Solution Theory. — According to this view, the dye is distributed between the two immiscible solvents — the liquid and the fibre — being in the ordinary solution state in the one, and as a solid solution in the other. Many arguments haye been adduced in favour of this theory, but the above objection led to the general abandon- ment of this theory in favour of the adsorption theory, although many of the properties of dyed fibres were not in agreement with what was to be expected, if the process of dyeing was purely an adsorption. In particular, the microscopic appearance on the whole gave the impression of solution throughout the fibre, and not of a surface adsorption. For example, starch grains are uniformly coloured,! and microscopic sections across dyed fibres have revealed in some cases the existence of irregular deposits on the external surface of the fibre, or mainly in the outer layers, but in many others the dye was uniformly distributed through the fibre.^ The former structure occurs with the insoluble dyes as Turkey red (see Mor- dants, p. 274), and with inorganic pigments; the latter occurred especially in direct dyeing on wool and silk with basic dyes. It also occurs with cotton and mercerised cotton with substantive dyes, and with basic dyes if the cotton is previously mordanted with tannin. As in many other cases there is evidence that adsorp- tion, which is a rapid process, is followed by slow changes, such as diffusion into the interior of the adsorbent. Examples of this are found in hydrogen and platinum or palladium, acids in various adsorbents both fibres and inorganic substances,^ iodine and hydrogen in 1 Fischer, Zeitsch. physikal. Chem., 1908, 63, i80. 2 Suida, loc. cit., p. 277. » Georgievios, Koll. Zeitsch., 1912, 10, 31 ; Zeitsch. physikal. Chem., 1914, 87, 669. DYEING 281 charcoal.^ though in the case of iodine and charcoal later re- search has failed to find any evidence of solid solution, the equilibrium being rapidly attained in benzene ; in carbon disulphide it was very slow, owing to decomposition of the carbon disulphide with liberation of sulphur which com- bined with the iodine.2 On the whole, then, it is not improbable that in dyeing, too, adsorption is followed by similar processes. According to Eeinders^ pure adsorption only occurs with those dyes which form sols, and in the fibre are also colloid (gels ?), among which he classes precipitation of gold sol or arsenious sulphide sol on charcoal or barium sulphate, or of charcoal suspension on paper, and also the dyeing of wool and cotton with true colloid dye sols, such as the blue acid of congo red. These dyeings are not fast, and the dye can be removed by washing* Eeinders and Lely have recently shown that the fact that the distribution of dye between fibre and bath follows the adsorption law instead of the distribution law (Henry's law) cannot be accepted as a criterion in this question, for the distribution of dyes between two immiscible liquids frequently follows the same adsorption law. They in- vestigated the distribution of a number of dyes between water and isobutyl alcohol, both with and without the addition of acid (HCl), and alkali (KOH) ; the dyes were— Basic dyes: methylene blue, fuchsin, crystal violet, new fuchsin. Acid dyes : crystal ponceau, patent blue, erythrosin A, roccellin, quinoline yellow, alkali blue. Substantive dye : congo red. 1 Davis, Chem. Soc. Journ., 1907, 91, 1666. MaoBain, Zeitsoh, physikal. Chem., 1909, 68, 471. ' Dora Walter, Eoll. Zeitsch., 1914, 14, 242. 3 Eoll. Zeitsch., 1913, 13, 96. * Pelet Jplivet, Die Theorie des FoLrbeprozessen, Dresden, 1910, p. 142 282 APPLICATIONS OF COLLOID CHEMISTRY They found (1) that the distribution of the dyes between water and isobutyl alcohol is expressed by the adsorption formula — ^alcohol jr n Cwater in which - is less than 1. The value of - varies from 0*3 n n for erythrosin A to 1 for alkali blue and crystal ponceau. In most cases it is not constant, but rises with increased concentration. (2) The addition of potassium hydroxide to the basic dyes, and of hydrochloric acid to the acid dyes increase the concentration of dye in the alcohol layer, just as they increase the amount of dye taken up by fibres, (3) Wool and silk which are dyed with basic dyes are readily decolorised by the alcohol (compare the non- dyeing of silk by picric acid in benzene, p. 278). (4) A colourless solution of rosaniline in water dyes wool and silk a deep red, i.e, the colour of the salts ; the same thing happens when the solution is shaken up with isobutyl alcohol or amyl alcohol. In this case salt for- mation is excluded. It is probable that there are two modifications of the base, colourless and red, the former more soluble in water, the latter more soluble in alcohols (and in fibres), and so taken up from the aqueous layer.^ Eeinders explains the small value of 11% as partly due to the differences in molar weight of the dyes in the two 1 A similar thing occurs witli picric acid, whioli is readily adsorbed from the yellow aqueous solution by, e.g., platinum black on platinised electrodes, and is not extracted from them by washing with water. It is extracted by alcohol and benzene, but in the form of a colourless solution, which resulted in the extraction being overlooked, until it was noticed that the addition of water to the alcohol produced an intense yellow colour. (Private ppnamunioation from Professor J. Walker.) DYEING 283 layers, for as already stated their molar weights in alcohol are usually normal, while in water they are 2-3 or more times the normal, as the following figures show ^ : — Dye. Molar weight. In waler. In alcohol. Theoretical. Puohsin .... Methyl violet . . . Benzopurpurin . Diamine blue . . 520-620 800-870 3000 3400 320-340 400-420 387 408 724 999 The change in molar weight with concentration may be the cause of the change in the value of - with concentra- tion. Or the explanation may be that the dyes are both hydrolysed and ionised in water, and not in the alcohol ; the ions would not be extracted from the water layer by the alcohol, and thus the distribution would vary with the concentration of the ions ; similarly, the distribution of the hydrolytic products between the two solvents would depend on their solubility in each, and, naturally, to a different extent from that of the undissociated dye. That this must have some effect on the distribution is borne out by the effect of acid on acid dyes, and of bases on basic dyes, the results of which are as would be expected from the above considerations. The addition of acid to basic dyes, or of base to acid dyes, often causes an increase in the concentration of the dye in the alcohol layer. This may be referred to the decreased hydrolysis, and consequent increase in concentration of the undissociated dye salt in the water layer, which, of course, means an increase in the alcohol layer also. The effect is not so great as in the 1 Krafft, Ber., 1899, 32, 1608. 284 APPLICATIONS OF COLLOID CHEMISTRY other case, where acid is added to acid dye and base to basic dye, and the changes in colour point to other effects. The following figures illustrate the effect of acid and alkali on methylene blue : — Equivalents to 1 oi methylene blue Calcobol Cwater HOI. Large excess 8 4 2 1 1'08 0-86 0-50 0-41 0-41 041 KOH. 0-8 2-6 6-2 30 0-41 0-57 1-25 333 Finally, the adsorbability of the three dyes, crystal violet, new fuchsin, and patent blue, follows the above order, independently of the nature of the adsorbent, whether it is charcoal, silk, wool, cotton, or cellulose (p. 246)j and the same order is retained in their distribu- tion between water and isobutyl alcohol. We are thus forced to conclude that the invalidity of the simple distribution law and the validity of the adsorption formula cannot be considered a decisive objection to the solid solution theory in favour of the adsorption theory. On the other hand, the adsorption theory affords a reasonable explanation of the striking differ- ences in the adsorption of dyes in water and in other liquids such as alcohol and benzene. It seems probable that the initial stage in most instances is an adsorption pure and simple, but that, if the dyeing is to be fast, it is succeeded by other stages which may in some cases be chemical changes and physical in others, resulting in a decrease in solubility of the dye in the original solvent. CHAPTER XXI tanning, the soil, and purification of sewage Tanning After so lengthy a discussion of the main processes of dyeing, which, however, makes no claim to be exhaustive or complete, it is unnecessary to devote much space to tanning, since to a great extent the main problem is the same, viz. how far the actions of the chemicals on the skin are to be regarded as chemical reactions, adsorptions, or mutual precipitations of colloids. The colloid nature of the substances concerned in tanning is even more pronounced than in the case of dyeing. The skin contains albumin- and gelatine-like substances in the form of gels, which swell in water, this being influenced to a marked degree by acids, alkalies, and salts. The tanning materials are mainly colloids, and hydrophile rather than hydrophobe. Tannin, e.g., forms undoubtedly an emulsoid sol in water, for it is opalescent, does not dialyse, and does not lower the freezing point of water. It is interesting to note that Patemo,^ who first discovered this, also stated that its molar weight in acetic acid was normal. He has quite recently, however, found this to be erroneous.^ the lowering of the freezing point being due to impurities and moisture in the tanning ; it is as much a colloid in acetic acid as in water. One result of • Zeitsch. physikal. Chem., 1889, 4, 457. ' Paterno and Salimei, Koll. Zeitsch., 1913, 13, 81. 286 APPLICATIONS OF COLLOID CHEMISTRY its colloid nature is that a larger amount of tannin is extracted from tanning material by distilled water than by ordinary water, the presence of small quantities of salts preventing solation, bivalent ions (Ca" or CO3") being more powerful than univalent ions. In the treatment preliminary to tanning proper, the swelling during the treatment with lime, and the persist- ence with which the last traces of lime are retained, necessitating extraction with acid, are both instances of colloid reactions. The process of " pickling " has been very fully studied by Proctor ; ^ it consists in treating the skin before tanning with very dilute sulphuric acid, during which a large amount of water is imbibed, the fibres swelling to a con- siderable extent. The skins are then placed in a strong solution of salt, when dehydration takes place, the skin becoming leather-like in texture. This action of neutral salts only takes place after acid imbibition. Proctor made out the general relations in the simpler case of the imbibition of water by gelatine under the influence of acids and salts. A gelatine which took up about eight times its weight of water, took up about fifty times its weight of water in very dilute hydrochloric acid ; very weak acids scarcely altered the amount imbibed. Salts in neutral solution increased imbibition, but in acid solution they dehydrated the gelatine ; the process of dehydration was very susceptible to traces of acid. This is another general characteristic of colloids ; such processes as gelation and solation, imbibi- tion and dehydration of gels are generally greatly influenced by very small quantities of acids and alkalies (p. 161). The action of the acid on the gelatine, according to Proctor, is to form a salt-like substance. > KoU. Chem. Beihefte, 1911, 2, 243-281. TANNING 287 In tanning proper, adsorption and mutual precipitation of sols in the liquid by the gels in the skin are undoubtedly the primary processes in the case of tannin and the vegetable 'extracts. The nature of the leather depends on the state of the skin (due to differences in the preliminary treatment) and the materials in the tanning liquor ; in the latter the colloids which are most readily adsorbed are deposited in great concentration on the surface, leaving the interior unaffected, while those which are less adsorbable and more diffusible have time to penetrate into the skin. As a general rule what is aimed at is a gradual tanning. This is attained in several ways, by appropriate selection of less adsorbable reagents at the commencement of the tanning, followed by more adsorbable ones. The same is effected by systematic exhaustion of the liquor; fresh extract is used with skins which are almost completely tanned, thus removing the most adsorbable substances and depositing them on the surface, while the fresh skins are treated with the almost spent liquor, which naturally contains only the least adsorbable substances. The adsorption is greatly influenced by the nature of the previous swelling of the skin; if acid has been iised, adsorption is much quicker and much greater than after alkaline treatment. In mineral tanning, in which chromium, aluminium, and iron alone are concerned, there is probably precipita- tion of a hydroxide sol contained in the basic salts employed by the colloids in the skin, although there is also the possibility of diffusion of salt into the skin, followed by decompositions similar to those occurring in mordanting which result in the deposition of metallic hydroxide gel in the fibres. In the two-bath process of chrome tanning, as in bichromate mordanting (p. 275), chromic hydroxide is produced in the fibre by the 288 APPLICATIONS OF COLLOID CHEMISTRY reduction of chromic acid by reducing agents, and if thiosulphate and acid are used as the reducing agent, sulphur sol is also formed (soluble hydrophile colloids from the skin or the chromic hydroxide " protecting " it), and some of it is also adsorbed. Secondary Changes. — Whatever the initial process, there are, in the case of both vegetable and mineral tanning, secondary changes, which are slow and irrever- sible. At first a considerable amount of tannin, or of chromium, aluminium, or iron, can be extracted with water, and the skin has not fully acquired the properties of leather. As the latter increase, the amount of tanning agent extracted by water decreases. In tanning with fats, as with tannin, there is the possibility of oxidation, and in the former case it has been suggested that the oxidation products are the actual tanning agents. Other actions supposed to take place are dehydration and poly- merisation, it being further suggested that these actions are catalytically accelerated by the fibre. There is ap- parently little direct evidence bearing on these suggestions. Iron tanning is not so satisfactory as aluminium, and still less so than chromium. According to Stiasny, ferric salts are too fully hydrolysed, and therefore too indiffusible, and too quickly precipitated, while the inferiority of alu- minium is due to its salts not being hydrolysed enough. If hydrophile colloids, such as soap, blood, albumin, gelatine, are added (as " protectors "), iron tanning is improved. It is quite possible, too, that the iron catalytically oxidises the fibre, and so causes brittleness. The superiority of chromium may be connected with the presence in chromic salts of complex ions, no similar substances being known in the case of ferric and aluminium solutions. Just as Proctor ^ was enabled to throw light on the 1 Koll. Zeitsch., 1907, 2, 85. TANNING 289 action of acids and salts in " pickling," by a study of the behaviour of gelatine (p. 286), so Abegg and Schroeder obtained valuable information from an investigation of the action of tanning agents on gelatine. The melting point of the gelatine was 36°. Formalin acted slowly, raising the melting point ; the piaximum attained was 48°, The time necessary for this to be reached was inversely proportional to the concentration of the formalin. Alum and chrome alum do not harden the gelatine as formalin does, and the melting point is lowered, by 10 per cent, alum to 34°, and by 10 per cent, chrome alum to 32° ; the action is very rapid (a few minutes), and previous treatment of the gelatine with sodium carbonate was of no effect. Iron alum, with or without sodium carbonate, caused the gelatine to become hard and brittle; the melting point was 40°-38° in very dilute solution. Bichromate followed by sodium thiosulphate, picric acid, and tannin were tried, but the melting point remained unchanged. The lowering of the melting point by the alums is due to the action of salts on the melting point of gelatine (p. 123) ; this effect is superposed on the tanning effect. Washing out the tanned gelatine produced a curious effect — the melting point of the mineral-tanned gelatine rose well above the original melting point after brief extraction with water and fell on prolonged washing to the original melting point. The following figures are of interest : — 290 APPLICATIONS OF COLLOID CHEMISTRY Gelatine (m.p. 36°). Melting point. Tanning agent. At onoe. 30 min. washing. Long washing 10% Cr. alum 32° 37° 36° „ Al alum 34° 42° 36° „ Pe alum 48° 45° 41° Tannin , , 34° — Formalin . . — 48°(48hrs.) 46°(55hr£ The Soil The application of colloid chemistry to the study of soils has been attended with marked success. The colloid nature of one of the chief constituents of soil, humus, has long been known, and to its colloid properties its import- ance is mainly due. There are also other colloids present in soil, the chief being (a) Humus, and decayed organisms. (b) Ferric hydroxide and aluminium hydroxide, etc. (c) Amorphous silicates, resulting from the weathering of crystalline silicates. {d) Bacteria. Humus is, perhaps, the most important of all ; it is emulsoid in character, resembling albumin in many respects — in salting out, imbibition with water, adsorp- tion of salts, migration to the anode ; it acts as a sub- strate for bacteria, and as a "protective'' colloid for suspensoid colloids in the soil. The question whether it is an acid or not seems at last to be definitely settled. It has frequently been held that its neutralisa- tion with bases was simply the formation of adsorption compounds, but recent work leaves little room for doubt. Tbe solution has a greater electric conductivity than oleic acid solution; it is said to invert cane sugar, and THE SOIL 291 to react with iron, hydrogen being liberated. Oden pre- pared humus acid suspension, and a colloid-free ammonium humate, and concludes that alkali first peptises humus, and subseq[uently acts upon it chemically, for alkali humate is certainly not a colloid, since no particles are visible in the ultramicroscope ; further, salts do not cause precipitation, the solution diffuses, and there is no change in concentration when it is shaken up with charcoal or barium sulphate. The gelation of humus sols gives rise to membranes which make the soil impervious to air and water. As a negative colloid humus is precipitated by kations and in accordance with the valency rule more readily by Ca"" and Fe" or AV. They alter the permeability, the amount of water taken up by imbibition, and also the way the soil " works." The adsorption of potassium is probably also affected. The adsorption of dyes by humus is made use of in soil analysis, e.g. methyl violet serves to distinguish colloids from finely-divided crystalloids; colloid aluminium silicates and aluminium hydroxide can be distinguished from amorphous silica (silicic acid) by means of alizarin, and the two former from each other by naphthol yellow -f- acetic acid, which colours the aluminium hydroxide alone. AU these reactions depend on the attraction of colloids for basic or acid dyes according to the sign of the electric charge on the colloid. As Eohland has shown, colloid clays adsorb prussian blue, organic dyes, colouring matters from fseces and urine, and also albumins, oils, soaps, etc., from drainage waters. A knowledge of the amount of clay in a soil is thus an important matter. Plasticity tests afford only a rough guide to the amount, and better results are obtained by ascertaining the hygroscopicity, i.e. the adsorption of 292 APPLICATIONS OF COLLOID CHEMISTRY water vapour by the surface. It is, however, difficult to apply, as the extent of surface changes on drying, and humus imbibition also vitiates the results. The adsorp- tion of malachite green by the clay affords a good and reliable estimate of the amount of clay. The adsorption of salts, of calcium and potassium in particular, by soil is highly important; potassium is strongly adsorbed by clays, phosphate by calcium carbonate (nodoubt somechemical actionin this case), also precipitated calcium phosphate precipitates ferric oxide and aluminium oxide. The adsorption of potassium is a very rapid process, as pure adsorptions usually are, while the adsorption of phosphate is much slower, as is to be expected, since, as stated above, the adsorption of phosphates may be partly a chemical action. Another important action of fertilisers is the solation of silicic acid gel (probably also of other colloid gels), usually stated as dissolving silicic acid ; the sol rises by capillary action to the surface, and is there again gelated, and encrusts the surface. The weathering of rocks gives rise to colloids, mainly in the form of gels. The precipitated colloids produced during weathering are not chemical compounds of alumina with silica, i.e. amorphous aluminium silicates, but are mixtures of the single gels, produced by the mutual pre- cipitation of positive and negative sols. PUIIIFICATION OF SeWAGE AND WATEK The oxidisable impurities in sewage are colloids in the form of sols. This was first suggested by Biltz and Krohnke,^ who proved it to be the case by dialysis, and by the migration of the substances to the anode when subjected to a faU of potential. They are thus negative 1 Ber., 1904, 37, 17i5. PURIFICATION OF SEWAGE 293 colloids, and are precipitated by positive sols, sucli as ferric hydroxide sol and aluminium hydroxide sol. They are also rapidly precipitated by ferric salts and aluminium salts ; and this not because of the hydrolysis of these salts into the above positive sols, but because of the powerful precipitating effect of trivalent kations, as Fe'", Al'", etc., on negative sols (p. 139). Biltz applied the adsorption formula, but without much success. The methods of chemical purification of waste water in chemical manufactures are usually based on the above reaction. For example, aluminium hydroxide is used as a clearing agent in the waste waters in starch factories. According to Eohland (p. 291), some kinds of native plastic clay oan be used instead of artificial colloid aluminium hydroxide in these purification processes. The purification of water for town supplies by filtration through beds of sand, charcoal, etc., affords examples of adsorption from solutions at the enormous surface of the filter. The addition of alum to water before mechanical filtration has several effects. In the first place, the acid solution will affect the electric charge on the colloids in the water, especially the emulsoid colloids, and so will increase their precipitation by salts. Then, as mentioned above, aluminium ion has a very powerful precipitating action on negative sols, and if positive sols are present the bivalent sulphate ion will also rapidly precipitate them. Some of the impurities which may occur in water are very strongly adsorbed. For instance, lead may be dis- solved from the lead pipes, and find its way into the water supply. It is completely taken out by a single filtration through a layer of closely-packed cotton wool (a single filter paper will not do it). This is extremely convenient for analysis, as the lead is easily removed from 294 APPLICATIONS OF COLLOID CHEMISTRY the cotton wool by means of acid. The lead is apparently present as a sol of lead carbonate or hydroxide. Eohland states that the yellow colouring matter from the sulphite cellulose process is not adsorbed readily, as it is not a colloid. It must be turned into a colloid by oxidation in air and addition of potassium ferrocyanide before it can be removed by adsorbents CHAPTER XXII APPLICATIONS OF COLLOID CHEMISTRY TO BIOLOGY The Equilibrium between Blood and Oxygen. — The question of the solubility of gases, especially the important gases oxygen and carbon dioxide, in blood has long engaged the attention of scientists. The variation in the amount taken up with the external pressure does not follow Henry's law (that the solubility of a gas in a liquid varies directly as the pressure), very large amounts being taken up at low pressures in comparison with those at high pressures. The amount taken up at any pressure is enormously greater than that dissolved by water under the same conditions,^ as the following figures show : at a pressure of 150 mm. 100 c.c. of water take up 0"7 c.c. of oxygen, while 100 c.c. of dog's blood take vlt^ 24 c.c. ; as the ceutrifuged plasma takes up only 0'65 c.c, the corpuscles are responsible for the difference, and not the salts or other substances in the blood. Since Liebig's time, the accepted explanation of this deviation from Henry's law has been that it is due to the formation of a chemical compound of haemoglobin and oxygen. There are several difficulties in accepting this explanation. Not only does the composition of this compound vary with the pressure, but also with other 1 Bohr, Nagel's Handbuch d. Physiol, 1905, I;, 1, p. 54; Loewy, Physikal. Chem. u. Medizm, 1907, I., p. 231; also Philip, Physical Chemistry : Its bearing on Biology and Medicine, 1913, p. 26. 296 APPLICATIONS OF COLLOID CHEMISTRY varying conditions of preparation ; further the maximum amount that can be taken up is by no means constant. Thus the maximum amount of oxygen that can be taken up should bear some constant relation to the amount of iron (representing haemoglobin), but 1 g. of Fe corre- sponds to — 328-468 c.c. in dog's blood. 3U1-45U » JJ ox j> 280-401 }} ,, pig's »> 379-426 ". JJ horse's )j Again, in any given case, the maximum saturation is not reached even at a pressure of 10 atmospheres ^ : — p cm. Hg. c.c. atmospheres. c.c. 17 7-3 1 20-0 25 10-1 2 20-9 34 10-8 3 21-6 44-0 15-6 5 22-7 56 16-9 7 23-1 10 23-4 The first of these difficulties was apparently solved by Bonders, on the basis of chemical dissociation, analogous to the dissociation of calcium carbonate. But to "explain" all these phenomena, it was necessary to assume more than one kind of haimoglobin, and several dissociation products of haemoglobin, one containing iron (hsemochro- mogen), another one iron-free. The former again dissociates into a part containing oxygen, and another which is oxygen-free. As Wolfgang Ostwald^ pointed out, all the available data for the amounts both of oxygen and of carbon dioxide 1 Bert, La session baromUrigue, etc., Paris, 1878, p. 643. ' Koll. Zeitsch., 1907, 3, 264, 294. BIOLOGY 297 in blood are expressed by the adsorption formula, and give typical adsorption curves ; the following example will suffice 1 : — mm. Hg. X in per cent, of total saturation. X calculated. 2 16 25 45 63 82 32-5 52-0 63-5 74-0 83-0 93-5 26-1 52-9 61-5 75-1 84-2 92-1 The figures in the last column are calculated from the adsorption formula — — = k en m with k = 20-6 and - = 0-34. The curve in Fig. 21, represents p and x in columns 1 and 2. There can be little doubt that the gases are adsorbed by the blood, probably at the surface of the disperse phase. The effect of artificial colloids on the solubility of gases has been the subject of investigation by Geffcken.2 With ferric hydroxide sol there was no difference between the solubility of oxygen in water and in the sol, while nitrous oxide was less soluble in the sol. And more recently, Findlay^ and co-workers have carefully in- vestigated the problem, using carbon dioxide and nitrous 1 Loewy, Engelmcmn's Arch. Physiol., 1904, 231. " Zeitsch. physikal. Chem., 1904, 49. » Chetn. Soe. Journ., 1910, 97, 536; 1912, 101, 1459; 1913, 103,486; 1914, 106, 291. 298 APPLICATIONS OF COLLOID CHEMISTRY oxide, and a large variety of colloid sols, as starch, egg albumin, gelatin, dextrin, ferric hydroxide, silicic acid, and suspensions of silica and charcoal, at pressures from 250 mm. to 1000 mm. of mercury. At low pressures the solubility of the gas may be greater or less than in water, but diminishes with rise of pressure, falling to a constant minimum or passing through a minimum and rising slightly after it. As there is no effect at all comparable 20 4-0 60 Pressure I JTiTn,. Tnercury) Pig. 21. with the case of oxygen in blood, it is evident that some other factor must be operative in the latter case, and this may be chemical in its nature. Enzymes and Inorganic Ferments. — The organic fer- ments which have been isolated are amorphous, albumin- like substances; it is usually impossible to remove them from organic tissues with which they are associated. During reaction they must be either in the form of gel on the surface of the tissue, or in the form of a sol (probably BIOLOGY 299 emulsoid or hydrophile). In either case the reaction is obviously heterogeneous, and adsorption, surface con- centration, and diffusion must therefore take an important part in the changes. If the chemical change is a rapid one, the Nemst-Brunner principle must also apply, and if the enzyme is a sol, Brownian movement will come into play, as ■will also changes in the specific surface under the influence of reagents, especially of acids, alkalies and salts. When Bredig discovered his method of preparing sols of pure platinum and other metals, whose action as catalysts in the massive state was known, he was naturally led to examine their catalytic properties in comparison with the same metals in the solid state on the one hand, and with the organic catalysts, the enzymes, on the other. The results obtained by Bredig and his pupils are embodied in his interesting book, Anorganische Fermente (Leipzig, 1901). Qualitatively the metal sols were found to catalyse the same reactions as platinum black, e.g. the union of hydrogen and oxygen, the oxidation of hydriodic acid by atmospheric oxygen, and the decomposition of hydrogen peroxide, which is also catalytically decomposed by many organic ferments. The activity of the sols is enormous ; thus 1 c.c. of a platinum sol, containing from O'OOOl mg, to "000003 mg. of metal, showed marked catalytic properties. Sols of other substances which act upon hydrogen peroxide were much less active than this, the concentrations corresponding to the above being — 20 million litres for 1 g. atom Ft . , 20 r MnOj . 10 C02O3 2 CuO .' 1 PbOa . 01 300 APPLICATIONS OF COLLOID CHEMISTRY These concentrations are comparable with those at which the activity of ferments, etc., is perceptible. Quantitatively there is a close resemblance between decompositions by metal sols and by enzymes. The decomposition of hydrogen peroxide into water and oxygen is a very convenient reaction to study; it is a reaction of the first order if platinum black is used, and remains so whether the solution is acid, alkaline, or neutral. Here the chemical reaction is extremely rapid, and the velocity actually measured is the diffusion of the hydrogen peroxide across the adherent layer, the tempera- ture coefficient being correspondingly small. The reaction with platinum sol in neutral solution is not strictly one of the first order, although it is approxi- mately so, for if the sol is a very active one, the velocity increases as the amount of hydrogen peroxide falls off. This is not due to the formation of a compound of platinum and oxygen, since the rise is repeated, even if the platinum has been previously treated with the peroxide. In acid solution the reaction is also of the first order. There is a difference in this connexion between the enzymes and the metal sols. O'SuUivan and Tompson,! and Tamman ^ have shown that the inversion of sugar by enzymes approximates to the logarithmic law (reaction of the first order), but is not in exact accordance with it, because of changes in the condition and activity of the ferment, which are functions of time, and in some instances of the amount of substance catalysed. In alkaline solution very peculiar results were obtained. The activity is very much increased, so that the sol must be 3-30 times more dilute than in water, a suitable con- centration being 1 g. atom in 300,000 litres. Dilute 1 Chem. Soc. Journ., 1890, 87, 926. » Zeitsch. physikal. Chem., 1889, 3, 35; 1895, 18, 436. BIOLOGY 301 alkali therefore increases the velocity, which increases at first with the concentration of sodium hydroxide to a maximum, and then diminishes, finally becoming less than in neutral solution, e.g. — 0-05 H2O2 + 0-000003 Pt Concentration of NaOH i *^ '^^ 2^6 Ta? B^ A e (min. lecompi Concentration of ^h 256 Ta? 34 28 24 \ 1 4 i NaOH ( ^^ " i * 1 normal , Jr/ <""^^-) f°.^' 1 255 34 28 24 25 22 50% decomposition ^ , Jr/ (""^^•> ^°^ i 34 34 70 162 520. 50% decomposition ) Fig. 22a is the corresponding curve. Thus in normal sodium hydroxide the velocity is reduced to half the velocity in neutral solution. This is very similar to the effect of alkali on the decomposition of hydrogen peroxide by emulsin. Jacobson ^ had found that the enzymes of malt and of pancreas, and emulsin decomposed hydrogen peroxide much more rapidly in weak alkaline solutions, e.g. — Emulsin + H2O2 Concentration 1 n 1 i 1 1 1 1 ofKOH J " T3 7 -40 t6 is normal Time (min.) for j 33 3 g ^5 ^^ 30 170 c.c. O2 J The curve (Fig. 22b) is very similar to that for platinum. The order of reaction cannot be determined, for in very dilute alkali solutions the velocity is independent of the concentration of the peroxide ; at greater concentrations the order of reaction comes out between 0-1. The cause » Zeitsch. physiol. Chem., 1892, 16, 3i0. 302 APPLICATIONS OF COLLOID CHEMISTRY of this may be the stabilising effect of bydroxyl ion on negative sols (p. 105), changing sub-microns, to amicrons, and causing an increase in the specific surface and Brownian movement. A similar explanation may also hold for the analogous case of enzymes. Confirmation of this view may be found in the observation that ferric hydroxide sol, which has marked catalytic powers, is quite inactive in alkaline solution, according to Bredig's observations. It is a positive sol, and the effect of hydroxyl ion upon it 45t OSn Concentration, Pig. 22a. 002n CmiceniraiioTi, Fig. 22b. 004n. should be the direct opposite of the effect on negative sols. If the concentration of the platinum sol is diminished in the ratio 2 : 1, the velocity constant Tc decreases in the ratio 3 : 1, e.g. — Pt concentration. k. 28-16 X 10-6 0-115 21-12 „ 0-072 14-08 „ 0-040 10-56 „ 0-024 5-28 „ 0-0084 2-64 „ 0-0027 Finally, there is a striking parallelism between the ' poisoning " action of various substances on platinum sol BIOLOGY 303 and on enzymes. Substances such as hydrogen sulphide, mercuric chloride, hydroxylamine, hydrocyanic acid, etc., even at very small concentrations, reduce or entirely stop the catalysis : e.g. — 0-06 H2O2 + 0-00001 Pt Concentration HON 1 (mol. per litre) / 0"° O'OOOOOOOS 00000001 0'0000002 Time in minutes for \ t,.k ^c , 00 n ^r. n 50% decomposition I ^^ 15-4 22'0 40'9 i.e. 0"0014 mg. of hydrocyanic acid in 1 litre reduces the velocity to half. Jacobson finds the same for the catalysis of hydrogen peroxide by emulsin and by pancreatic ferment, and Kobert found a similar sensitiveness of the catalysis of hydrogen peroxide by blood (haemase) to hydrocyanic acid (0'002 mg.). It is interesting to note that the order in which the reagents are mixed makes a great difference to the result; the poisoning of the catalyser, whether platinum or enzyme, is always more severe when the hydrocyanic acid is added to it before the hydrogen peroxide. Both the platinum and the enzymes "recover" their activity after a time. As Freundlich points out, the variation of the " poisoning effect " with the concentration of the poison is in accordance with the adsorption formula, though it is, of course, not suggested that the mere adsorption of the substance on the surface slows down or prevents the action (by offering a resistance to diffusion). Agglutination. — An enormous amount of work has been done during the recent development of colloid chemistry towards the elucidation of some of the perplex- ing and complicated reactions of biology in accordance with the newly acquired ideas, and in no direction have more strenuous efforts been made than in the case of toxins and antitoxins, and the other reactions of a similar nature. 304 APPLICATIONS OF COLLOID CHEMISTRY So extensive is the subject that entire books have been devoted to it ; obviously, then, nothing can be done here beyond indicating how far the outstanding features of these reactions are colloid in nature, whether adsorptions or mutual precipitations of sols and gels. A word of warning may not be out of place here, although it may be equally necessary in other subjects. There is an undeniable tendency not only to apply a new idea or theory in many new directions in a loose and vague manner, but also to regard such applications as "explanations." Mass-action, osmotic pressure, ionisation, hydrolysis, amphoteric electrolytes, and now colloids, adsorption, and surface tension, have been successively thus employed, and often in cases of so complex a nature that it was a priori highly improbable that any one explanation could take in all the relations concerned. Particularly is it necessary to remember this in dealing with adsorption. The adsorption formula is merely an empirical expression, which more or less closely fits a large number of facts, which are by no means necessarily the same in nature, and to show that a particular pheno- menon can be expressed by the adsorption formula certainly does not explain the phenomenon. What are commonly known as emulsions of bacteria are really suspensions, varying in dispersity according to the kind of bacteria, and are precipitated by quite definite amounts of electrolytes, as are suspensions generally. They are, as a matter of fact, not very sensitive, as they are not precipitated by kations of the alkalies or of light metals, but are precipitated by acids (hydfion), by kations of heavy metals, and by aluminium and ferric ion (see table, p. 306). They thus stand between suspensoid and emulsoid sols, and on this account have been represented as suspensoids, protected by an emulsoid BIOLOGY 305 sol. This is supported by the fact that bacteria sols are not " protected " to any considerable extent by emulsoids, such as gelatine or dextrin, which are usually so powerful in this respect. When the immune serum is added to the bacteria sol, the latter becomes much more sensitive to electrolytes. Indeed, no clumping of the bacteria or sedimentation occurs on the addition of the immune serum, unless salts are present ; hence, if the sol and the serum have been dialysed, there is no precipitation. It would thus appear that the agglutinin in the immune serum destroyed the " protecting " part of the bacteria sol, which thus became a suspensoid sol. The method of working is as follows. The bacteria sol (typhus, cholera, etc.) is treated with formalin, which kills the bacteria, the sol is repeatedly centrifuged, changing the water, or it is dialysed. The sol is opalescent, and remains without sedimentation for a long time. To obtain the agglutinin-bacteria, the serum of animals, which have been injected with the same bacteria as the sol, is added to a culture of the bacteria, and the resulting agglutinin-bacteria sol is treated in the same manner as the bacteria sol. The agglutinin-bacteria sol is not precipitated by hydroxyl ion, but is readily precipitated by acids and salts of heavy metals. The valency rule applies, the pre- cipitating power increasing with the valency of the katiOn. There is an " optimum " of precipitation at one definite ratio of bacteria sol to agglutinin, and no pre- cipitation occurs if there is a large excess of either. This is analogous to the mutual precipitation of positive and negative sols (p. 132). The following table illustrates the differences between bacteria sol (typhus) and the corre- sponding agglutinin-bacteria in their precipitation by 3o6 APPLICATIONS OF COLLOID CHEMISTRY electrolytes. For comparison, the corresponding figures ^ for a mastic sol are added : — Electrolyte, N"aCl . NaNOs . 2^8,804 2 • KOH . HCl . . H2SO4 2 11.020300 AgNOg .' MgSO^ . OaOlg . BaClg . OdSO^ . Ni(N03)2 HgOlg . Al2(SO,)3 2 I'e2(S04)3 Concentration (m.mol. per litre) for precipitation in 24 hours. Agglutinin- Bacteria. bacteria. Mastic. 00 25 1000 — 25 — — 55 — 00 00 1 0-5 10 1 0-25 10 1 1 25 1 125 00 1-3 50 00 2-3 25 CO 2-5 — 5 0-5 1-3 12 1-3 0-25 — 0-08 0-08 0-2 0-04 The change from emulsoid to suspensoid is also shown hy the precipitation of agglutinin-bacteria by dyes. Similar changes occur when egg-albumin is coagulated by heat. The mutual precipitation with its zone bounded on both sides by zones of non-precipitation is accompanied by the analogous behaviour towards multivalent ions, e.g. AY" and Te"" (p. 137), and also towards acids (H") ; the latter of which, however, has not been observed with I Bechhold, Zeitsch. physikal. Chem., 1904, 48, 385. BIOLOGY 307 inorganic sols. Its occurrence in this instance is probably connected with the anomalous position of an albumin-like substance with suspensoid properties, the albumin cha- racter rendering the sol very susceptible to hydrion (H") as emulsoid sols of albumin are known to be. Henri' pointed out an important difference between the mutual precipitation of sols and agglutination in that the former only occurs between sols of opposite electric charge, while the latter occurs with both. For instance, red blood corpuscles, which are negative, can be precipi- tated by both positive and negative sols. He assumes that the corpuscles are surrounded by a pellicle which can fix salts (by adsorption ?), especially magnesium sulphate and calcium sulphate. These salts act upon any pre- cipitable sol, producing a coagulum round the corpuscle ; they can be removed by diffusion into an isotonic sugar solution, after which the blood corpuscles are much less susceptible to precipitation by sols. Similarly by soak- ing them in solutions of salts, especially of chlorides and sulphates, they become more readily precipitated by sols, and especially by ferric hydroxide sol. The nature of the combination between the bacteria and the agglutinin is a problem of much the same kind as that of the combination of fibres and dyes. From the constancy of the ratio of the two which is necessary for optimum precipitation, it was at first concluded that the combination was chemical in nature. But this alone is of no value as a criterion, since, to mention only one instance, the mutual precipitation of positive and negative sols also fulfils this condition, even in cases where chemical reaction in the ordinary sense cannot possibly be supposed to take place. Biltz^ then suggested that they were > Compt. rend., 1904, 138, 1461. 2 Zeitsch. physikal. Chem., 1904, 48, 615. 3o8 APPLICATIONS OF COLLOID CHEMISTRY adsorption compounds formed from a highly disperse emulsion and a sol. He showed further that the dis- tribution of agglutinin between the bacteria and the immune serum followed the adsorption law, as the follow- ing figures show : — Typhus Bacillus and Agglutinin, . Agglutinin in Agglutinin in agglutinin-bacilli. solution. Observed. Calculated. 20 180 170 60 340 360 500 1500 1520 3500 6500 5960 9000 11000 11400 The figures in column 3 are calculated from the adsorption formula X L — = acn m in which the constant a = 21'9, and - = 0"637. n The action between bacteria and agglutinin can thus undoubtedly be expressed as an adsorption. But there are difficulties in the way of accepting this as a final explanation, for a most important feature of the whole reaction is the fact that sols of any one kind of bacteria are affected only by the agglutinin produced in serum by the injection into the animal of the same kind of bacteria, and by no other. Gelatine, however, acts as a specific agglutinin to both typhus and cholera bacilli. This, while confirming the colloid nature of the reaction, is difficult to work out on the line of chemical combination, although, as Proctor found in investigating the action of acids and salts ' Biltz, Med. nalurverein. Archw., 1907, 1, 363, calculated from Elsenberg and Volk, Zeitsch. Hygiene, 1902, 40, 155. BIOLOGY 309 on gelatine, gelatine forms a chemical compound of a salt- like nature with acids (p. 286). This specific character of the reaction is generally considered to be strongly in favour of a chemical reaction between the two, a point of view advocated throughout by Arrhenius in his book on Immunochemie, If the combination is chemical in nature, it must be of a peculiar and restricted kind. The probability of deciding the problem in the meantime seems to be somewhat remote. Toxins and Antitoxins. — The nature of the reaction between a toxin and its antitoxin has, like the similar reaction of agglutination, attracted much attention in recent years, and the position concerning it is very similar to that in the case of agglutination. The substances concerned in the toxin-antitoxin reaction are undoubtedly colloid in nature more so on the whole than in the agglutination reaction, some of them, indeed, being so far removed from the dispersity associated with eniulsions that they diffuse fairly rapidly. The diffusion coefficients of some toxins and antitoxins in gelatine have been determined by Arrhenius.* cm Diphtheria toxin . . D = 0-016 X lO'S ^ sec. „ antitoxin . 0'0017 „ Tetanus lysin . . . 0-043 „ antilysin . . 0-0024 The substances are thus colloids, and the antitoxins apparently more so than the corresponding toxins. No conclusions regarding molar weight by the application of Thovert's relation (p. 35) are permissible, as the diffusions were carried out in gelatine and not in water. The first striking similarity between the colloid 1 Immunochemie, p. 17. 310 APPLICATIONS OF COLLOID CHEMISTRY reactions and the toxin-antitoxin reaction is what is known in the latter as the Danysz effect. When a diphtheria toxin is treated with its antitoxin, the reduction in toxicity depends on the manner in which it is added, i.e. an amount of antitoxin which is exactly sufficient to neutralise a given amount of toxin when added all at once, is not nearly sufficient to neutralise the same amount of toxin when added little by little, with moderate intervals between each addition. And the longer the intervals, the greater is the insufficiency. This is a well-known charac- teristic of colloid reactions, e.g., in the precipitation of suspensoid sols by electrolytes,^ and in the salting out of emulsoids, as albumin.^ If the salt is added all at once, more albumin is precipitated than if it is added gradually ; when 5 c.c. of a saturated solution of ammonium sulphate were added at one operation to 5 c.c. of a white of egg sol, the filtrate contained 017 g. of albumin; when the addition was spread over twenty-four hours the filtrate contained 0*214 g., and over forty-eight hours it contained "237 g. Similarly with the mutual precipitation of positive and negative sols, the ratio of the amounts which produce complete precipitation is a function of the time over which the addition is extended. This " acclimatisation " is an important objection to explanations of the phenomena on the basis of chemical combination, or of chemical equi- librium, for even when the latter requires appreciable time for equilibrium to be established, the equi- librium itself is independent of the time. Bredig noticed too that the manner in which hydrocya'nic acid and hydrogen peroxide were added to platinum sol made very great differences in the result (p. 303). He also states that an old gold sol is much less sensitive to ' Preundlich, Zeitsch. physikal. Chem., 1903, 44, 143. ' HOber and Gordon, Beitr. chem. Physiol. Path., 1904, 5, 436. BIOLOGY 311 ammonia than a freshly prepared one, and that the addi- tion of such old insensitive sols to new very sensitive ones confers on the mixture "immunity" against ammonia.^ There are thus many colloid analogies with the Danysz effect. Several investigations of the electrical behaviour of toxins and antitoxins have been made with a view to explaining their action on each other, but with no very definite result. Diphtheria toxin and antitoxin were found by Field and Teague 2 to migrate distinctly. Bechhold also found that diphtheria toxin was slightly weakened at the anode, while the antitoxin migrated to the kathode. In a mixture containing an excess of toxin, the excess migrated to the kathode, provided that the experiment was per- formed immediately after mixing the toxin and antitoxin. ^ The main question here, as in agglutination, is whether the reaction is due to chemical combination or to an adsorption. The chemical reaction which naturally is at once suggested, is the neutralisation of acid and base. Against this is not only the fact that, as stated above, the proportion of the two necessary for neutralisation varies with the manner in which the addition is made, but also the further fact that as the amount of antitoxin to a fixed amount of toxin increases, the amount of free toxin diminishes, but never entirely disappears. The corre- sponding chemical reaction is the equilibrium between a weak acid and a weak base, such as ammonia and boric acid, in which case too there will always be free boric acid (or ammonia), however great an excess of ammonia (or boric acid) is added. It is chemical reactions of this type on which Arrhenius has based the chemical 1 Bredig, Anorganische Fermente, p. 28. ' Journal of Experimental Medicine, 9, p. 86. ' BecJjfepJd, Eoll. Zeitsch., 1907, 2, 184, abstract. 312 APPLICATIONS OF COLLOID CHEMISTRY theory of the reaction. The equilibrium equation for the above case of a weak acid and weak base in aqueous solution is — in which ci is the concentration of free acid, C2 „ „ free base, c ,, „ combined acid, and k is a constant. The following table shows how closely the above equa- tion holds for the reaction between toxin and antitoxin. In it Ci, C2, and c are the amounts of free tetanus lysin, of antilysin, and of bound lysin respectively. The value of k is 0'115 ; n in the first column is the amount of antilysin added to 1 of lysin, the second column contains the amount of free lysin as determined by experiment, and the third column the amount calculated from the above equation : — Tetanus Lysin and Antilysin. n c, (obs.). c, (calo.). 100 100 0-05 82 82 0-1 70 66 0-15 52 52 0-2 36 38 0-3 22 23 0-4 14-2 13-9 0-5 10-1 10-4 0-7 6-1 6-3 1-0 40 4-0 1-3 2-7 2-9 1-6 2-0 2-5 2-0 1-8 1-9 Biltz was the first to suggest that the reaction might be an adsorption, and in support of this view showed that the results are in agreement with the adsorption formula. BIOLOGY 313 The following table, calculated by Biltz 1 from Madsen's results 2 on the reaction between diphtheria toxin and antitoxins shows how closely the experimental results are expressed by the adsorption formula : — Diphtheria Toxiri, and Antitoxin. Free toxin in Toxin bound to the antitoxin. solution. obs. calo. 1-2 197 180 5-6 210 210 11-1 222 230 17-2 237 240 28-0 240 255 32-2 270 260 49-8 251 270 57-6 283 275 72-8 272 280 74.4 (511) 283 The calculated results in the third column are obtained from the adsorption formula — X L — = ac» m in which a = 177, and - = 0-102, n A comparison of the two tables shows that both the adsorption formula and the equation for the equilibrium between a weak acid and a weak base express the experi- mental results very closely, when the uncertainties of the experimental data are borne in mind ; on the whole, the adsorption formula is in closer agreement. There is, how- ever, the same difficulty here as in the case of agglutina- tion, namely, the specific nature of the reaction. The fact is that diphtheria antitoxin is the only one which can 1 Biltz, loc. cit, p. 308. ' Madsen, Arrhenius' Im/munochemie, p. 131. 314 APPLICATIONS OF COLLOID CHEMISTRY neutralise diphtheria toxin, though it has not yet been made clear how this is to be explained by chemical action of the type of neutralisation of acid and base, which is certainly a general and not a specific reaction. Osmotic Pressuee and Semipeemeability, Membeane Hydeolysis The earlier determinations of the osmotic pressure exerted by biologically important colloids gave conflicting results, partly on account of inadequate methods, and partly because of unsuitable treatment of the substances. Eecent work has shown that very considerable osmotic pressures are attained in some cases. Owing to the colloid character of the substances, there is usually no trouble in securing a membrane which is sufficiently jmpermeable by the colloid, while allowing water and crystalloids to pass freely. Parchment paper and collodion films (p. 27) are found most generally useful. As they are readily permeable by salts, any permanent osmotic pressure recorded cannot be ascribed to electrolytes or other crystalloid impurities. The effect of electrolytes on the osmotic pressure of sols has also been investigated ; these, especially in the case of albumin, are of interest from the biological point of view. As a rule, the osmotic pressure of sols is lower if salts are present, but in a few cases the pressure is increased, although the membranes are readily permeable by these salts. The cause of this is found in the change of dispersity of the disperse phase. Moore and Parker ^ determined the osmotic pressure of soap solutions, and obtained fairly large values. They ; » Amer, J. Physiol, 1902, 7, 262. BIOLOGY 315 concluded that soap is not dissolved as single molecules, but as " solution aggregates." (For a general account of the soaps, see p. 265.) Particular interest attaches to Bayliss's ^ recent work on the osmotic pressure of Congo red, which is the sodium salt of a disulphonio acid, the theoretical molar weight of which is 696. The apparatus used was of the Moore and Eoaf 2 type, in which a parchment paper septum is clamped between two metal chambers provided with flanges, the septum being supported by a rigid metal grid. The values obtained were slightly below those for a solution containing single molecules, and this behaviour was confirmed by vapour-pressure determinations, which gave normal values. Notwithstanding this, Congo red is a colloid, for it does not diffuse or dialyse, migrates in an electric field to the anode, and is precipitated by kations in accordance with the valency rule ; at the same time, it is a good electrolyte, and in dilute solution is largely ionised, up to 80 per cent, at a dilution of 500 litres, from which the osmotic pressure should be from 1"5 to 3 times the calculated value. Similar results have been obtained by Biltz,^ the osmotic pressure of Congo red corresponding to a molar weight of 602, when the osmometer was immersed in pure water. When the outer liquid was a solution containing the products of dialysis, concentrated until its electric conductivity was the same as that of the contents of the osmometer, the osmotic pressure fell, and then gave a molar weight of 2333. The conditions obtaining in the dialysis of the salts of the nature of Congo red, say NaE, have been investigated 1 BayUss, Proc. Boy. Soc, 1909, B. 81, 209; 1911, B. 84, 229; Koll. Zeitsch., 1910, 6, 23. ^ Siochem. J., 1906, 2, 34; or Philip, Physical Chemistry: its Bearing on Biology and Medicine, p. 182. » Zeitsch. physikal. Chem., 1909, 68, 357 ; 1910, 73, 481 ; 1911, 77, 91. 3i6 APPLICATIONS OF COLLOID CHEMISTRY by Donnan.i He shows that if a salt NaE, like Congo red, is dissolved in water and separated from pure water by a semipermeable membrane like parchment paper, a large amount of hydrolysis may take place. Qualitatively at least this is readily deduced. The membrane is per- meable by Na" and not by E', nor by the unionised NaR ; it is also permeable by H* and OH', and also by HgO ; consequently it will be permeable by NaOH and not by HE. This being so, the salt which in water may not be hydrolysed at all, can be largely decomposed in water when separated from pure water by the membrane, and especially will this be possible where, as is the case with Congo red, the acid is more colloid than the salt. An aqueous solution of the blue acid of Congo red shows submicrons in the ultramicroscope, and the osmotic pres- sure corresponds to a molar weight about 20 times that for single molecules. The same kind of thing will also occur with substances like aluminium chloride, and with simpler salts in the case of membranes like aluminium hydroxide which are permeable by various ions to very different extents. As experiments on the electric conductivity of salts across precipitation films showed, an aluminium hydroxide membrane is readily permeable by H", K', Na', CI', but not by SO4" ; from which it follows that an aqueous solution of say sodium sulphate, which is not hydrolysed to a measurable extent, can be separated by dialysis through such a film into sodium hydroxide and sulphuric acid. Similarly a weak acid like carbonic acid may be enabled to decompose a salt of a much stronger acid, if the latter is able to dialyse across the membrane much more rapidly than the other substances. ' Zeitsch. EUMrochem., 1911, 71, 572; Chem. Soc. Jottrn., 1911, 99, 1554. BIOLOGY 317 At the same time electrical potential differences may also be established at such a membrane. With Congo red on one side of the film, sodium chloride is distributed unequally on the two sides of the film, being higher on the side opposite the Congo red. According to Donnan, these inequalities of concentration render fallacious any calculations of molar weight from the osmotic pressures observed in such cases. Casein and sodium hydroxide give rise to a similar condition of matters,^ for the sodium hydroxide crosses over to the casein against the osmotic pressure. Until the casein is almost saturated no alkali will remain on the outside of the osmometer. Alkali, therefore, moves against the osmotic pressure to the colloid side in order to unite with the casein. Moore attributes the osmotic pressure to this union, and assigns to the membrane the purely mechanical role of holding together the aggregates; he maintains the impermeability of the membrane to be a fiction, stating that free ions pass through the membrane readily, while ions anchored to the colloid are retained on the colloid side of the membrane and exert pressure. With living cells the different concentrations of sodium, potassium, chloride and phosphate ions inside and outside the cell are re- garded as expressions of the specific af&nities of the definite colloid of each type of cell for these ions, and do not imply that there is a membrane acting as a " closed gate" to these ions. Moore and Eoaf^ are apparently not in favour of selective permeability of membranes, preferring on the whole selective adsorption by proto- plasm. ' Moore, Eoaf and Webster, Biochem. J., 1911, 6, 110. ' See further, Hober, Arch. ges. Phvsiol, 1913, 150, 15; and Moore and Roaf, Koll. Zeitsch., 1913, 13, 133, 3x8 APPLICATIONS OF COLLOID CHEMISTRY The question arises whether there is any real dif- ference between the two. Selectively adsorbing proto- plasm in the cell would probably act as a selectively permeable membrane, if it was spread out as a film between the two phases. SUBJECT INDEX " Acclimatisation," 98, 310 Aoid albumin, 115, 117 " Active " electrolyte, 62 solutes, 238 " Adhesions," 262 Adsorbents, efficiency, 215 Adsorption, 221, 243 equilibrium, 245 formula, 246 isothermal, 246, 248, 256 nature of solvent, 247 negative, 244 and precipitation, 102 at liquid interfaces, 258 of dyes, 254, 279, 291 of gases, 254 of gases by blood, 296 of ions, 91, 104 of'salts by soil, 291 pseudo-, 252, 262 theory of, 250 Agar, 125, 178 concentration of phases, 19 Agglutination, 308 and adsorption, 308 efieot of dyes, 306 salts, 805 Albumin, acid, 115, 117 alkaline, gold number, 130 amphoteric nature, 65, 117 coagulation, 116, 119, 120 crystallised, 113 density, 21 difEusion, 36 conductivity, 64 gold number, 129 kataphoresis, 84 molar weight, 35 osmotic pressure, 35 precipitation, 112, 113, 118, 119, 135 preparation, 113 salting out, 19, 113 solid skins on, 274 surface tension, 21 viscosity, 22 Alizarin lakes, 277 Alkali metals, organosols, 219 Alloys, sols from, 212 Aluminium acetate, hydrolysis, 14, 193, 194 hydroxide, crystalline, 174 hydrosol precipitation, 93 preparation, 194, 210 ion on sols, 96, 188 lakes, 277 mordants, 274 Amiorons, 11 development of, 55 Amplitude of Brownian movement, 49 and precipitation, 96 Antimonious sulphide, hydrosol, 201 Antitoxins, 309 kataphoresis, 311 Arsenious sulphide hydrosol, 12 density, 20 dialysis, 30 difEusion, 30 hydrogen sulphide in, 108 kataphoresis, 80 molar vreighty, 35 precipitation, 98, 101 by ferric hydroxide hydrosol, 135 preparation, 200 and alcohols, 262 Babium carbonate aloosol, 211 sulphate hydrosol, hydrogel, 171, 172, 202 Bichromate mordants, 275 Bismuth hydrosol, 190 Blood, oxygen equilibrium, 295 Blue gold hydrosol, 187, 188 Boiling point of sols, 36 of soap solutions, 266 Boron hydrosol by peptisation, 209 Brownian movement, 11, 47 amplitude, 49 and precipitation, 96 dispersity and, 56 320 SUBJECT INDEX Brownian movement oscillation period, 49 theory, 51 Cadmium hydrosol, 216 sulphide hydrosol, 210 Capillary analysis, 251 Casein, 20, 264 gold number, 129 membrane hydrolysis, 317 protection by, 203 Cellular structure of gels, 161 Charcoal adsorption of gases, 254 adsorption of salts, 253 Clay, estimation by adsorption, 291 adsorption of salts, 292 Clupein, diffusion, 36 Co-silicates, etc., 33 Coagulation, 9 of albumin, 116, 119, 120 Collodion dialysers, 27, 29 imbibition, 156 Golloidogens, 265 Colloids and crystalloids, 1, 5, 24 " complex " theory of, 108 electrolyte, 265 Colour of sols, 40, 41, 215 and dispersity, 41 Compressibility of gels, 142 Condensation methods of prepara- tion, 166 Congo red osmotic pressure and dialysis, 315 Conversion of one sol into another, 211 Copper hydrosol, 190 ferrocyanide hydrogel, composi- tion, 109 oxide hydrosol, 212 salts on albumin, 118 sulphide hydrosol, 200, 201 organosols, 201 Cotton, adsorption of tannin, 2Y6 Critical point and surface tension, 224 Crystallisation methods of prepara- tion, 165, 175 velocity, 168 Curved surfaces,vapour pressure, 22G Danxsz effect, 310 Dehydration of gels, 150, 154 Density of sols, 20 Dextrin, molar weight, 36 Dextrin, osmotic pressure, 33 Dialysers, comparison, 29 Dialysis, 24, 31 methods,, 26 of Congo red, soaps, etc., 315 Diffusion, 24, 30 and dispersity, 57 molar v^eight, 36 into gels, 24, 145 of toxins and antitoxins, 309 Disperse phase, 3 equilibrium with ion, 267 Dispersion electrical, 213 medium, 3 methods of preparation, 166 Dispersity, 3, 11 and colour, 40 and solation, 208 determination of, 54 Dispersoids, 6 Distribution law, 248 and dyeing, 281 of sols between two liquids, 259 Double refraction of gels, 144 Dyeing, 269 and adsorption, 278 and chemical action, 277 and distribution law, 281 and solid solution, 280 effect of acid and alkali, 279, 284 of salts, 272 Dyes, action on agglutin-bacteria,306 on albumin, 119 of salts on, 271 adsorption of, 254, 278, 291 colloid, 270 distribution between two liquids, 262, 281 mutual precipitation, 182, 135, 272 solid skins on, 273 titration of with dyes, tannin, etc., 273 Dynamic surface tension, 223, 230 at liquid interfaces, 258 of solutions, 231, 241 Elastic gels, hydration of, 154 Elasticity of gels, 143 Electric charge of sols, 82 conductivity, 61 albumin, 65 casein, 264 silver hydrosol, 64, 184 Electric conductivity of soaps, 266 SUBJECT INDEX 321 Electrical dispersion, 213 Bredig's method, 214, 216 of non-metals, 220 organosols, 214, 216, 218 Svedbetg's methods, 214, 217- 220 at kathode, 213 and decomposition of liquid, 214, 216 Electrical properties of sols, 65 theory of, 85, 88 Electroendosmosis, 66 acid and alkali on, 68 salts on, 70 valency rule, 72 of organic liquids, 73 Electrolyte colloids, 265 Emnlsides, 263 Emulsin, difiusion and molar weight, 36 decomposition of hydrogen per- oxide, 301 Emulsions, 4, 6 oil, 85 kataphoresis, 85 precipitation, 112 Emulsoids, 6, 18 protection, 129 mutual precipitation, 135 precipitation of suspensoids, 136 viscosity of, 21 Enzymes, catalysis of hydrogen per- oxide, 299 " poisoning " of, 303 and inorganic ferments, 298 Equilibrium between blood and gases, 295 between ions and sol, 267 Pbbmentb, inorganic, 298 Ferric ethylate, hydrolysis, 196 hydroxide hydrosol,chloridein,108 conductivity, 62 kataphoresis, 80 • precipitation, 97, 100 preparation, 195, 196, 209 protection by, 130 hydrogel, adsorption of ASjO,, 249 ion on platinum sol, 138 salts, hydrolysis, 193 Fluorescence and Tyndall Efieot, 39 Freezing point of sols, 36 Fused salts, surface tension of, 224 Gamboge, hydrosol, 176 Gases, adsorption, 254 solubility in sols, 297 Gel, 4, 10 Gelatine, 120 compressibility, 142 conductivity, 64 crystalline, 121 density, 142 gelation and solation, 121 hysteresis, 123 imbibition, 155, 159 kataphoresis, 83 molar weight, 36 mutual precipitation, 136 optical activity, 145 osmotic pressure, 36, 124 protection by, 202 salting out, 124 suspeusoid sol, 178 tanning of, 286, 289 viscosity, 22, 121, 123 Gelation, 10 and elasticity, 144 and imbibition, 160 of gelatine, 121 of silicic acid, 125 of soap, 267 Gels, difiusion in, 24 difiusion of, 25, 145 double refraction, 144 elastic gels, 154 hydration and dehydration, 150, 154 imbibition, 154, 156, 160 salts on, 159 non-elastic, 150 permeability of, 145, 147, 149, 816 structure of, 161 thermal properties, 142 vapour pressure of, 150 viscosity, 143 Gold ethersol, 260 hydrosol, blue, scarlet, violet, 187, 188 conductivity, 62 distribution between alcohols, 261 precipitation by ferric hydroxide hydrosol, 140 preparation, 179, 184-189 " Gold number," 129 potable, 178 sulphide hydrosol, 203 322 SUBJECT INDEX Gum aratio molar ■weight and os- motic pressure, 36 protection by, 203 H" and OH', opposed effect on eleo- troendosmosis, 68 on gelation of silicic acid hydrosol, 127 on precipitation of sols, 106 Heterogeneops systems, classifica- tion, 4 " Honeycomb " structure, 161 Humus, colloid properties, 291 Hydration of gels, 150, 154 Hydrolysis, 14, 192 and mineral tanning, 283 valency rule, 139 membrane, 314 of Congo red, 315 methods of preparation, 192 of ferric salts, 193, 195, 196 methyl silicate, 196 salts, 192 silicon sulphide, 193, 196 soaps, 266 Hydrophile, hydrophobe sols, 7 precipitation of, 94 Hydrosol-hydrogel, 5 Hysteresis of gelatine, 123 silicic acid sol, 19, 154 IcB sol, 175 Imbibition, 154 acids on, 286 maximum, 160 pressure, 156 salts on, 160, 286 velocity, 158 Immunity, 305, 311 Invertin, diffusion, 36 Ions, adsorption of, 91, 104, 118 stabilising effect, 105 Irreversible precipitation, 98, 111, 125 of albumin, 116, 118 Isoelectric point, 96 Kaolin suspensions, precipitation, 99, 106 Kataphoresis, 66, 74 and dielectric constant, 83 and mutual precipitation, IBS, 140 electrolytes on, 81 Kataphoresis in organosols, 82 of albumin, 84 of emulsoids, 83 of oil emulsions', 85 of silver hydrosol + Al'", 81, 96 of suspensoids, 74 velocity of, 76, 80, 88 Kathode reduction, sols by, 181 Lab, diffusion, 26 Lakes, 276 Liesegang's rings, 146 Liquid inclusions, 4 interfaces, adsorption at, 258 precipitation at, 120 surface tensions at, 258 Liquids, surface tension, 224, 229 Lyophile, lyophobe sols, 7, 94 Lyotrope series, 7-9, 30, 149 and gelation of gelatine, 128 of silicic acid, 127 heat coagulation of albumin, 117 imbibition, 160 osmotic pressure of gelatine, 124 precipitation of albumin, 115 Lysalbic acid, 130, 188, 264 Mastic hydrosol, 177 Mechanical filtration of water, 298 precipitation of albilmin, 120 dyes, 120 Membrane hydrolysis, 314 of Congo red, 315 Mercuric sulphide hydrosol, 201 Mercury hydrosol, 190 surface tension, 255 Metal' ethersols, 41 hydroxide hydrosols, 203 sols, 12 Metals, surface tension, 224 Methyl violet dialysis, 20 kataphoresis, 75 Micron, 11, 42 Microscopic observations on sols, 42 visibility, 11 Mineral tanning, 287 " Molar fraction," 230 Molar surface energy, 225 Molar weight and diffusion co- efficieilt, 35 of casein, 265 of colloids, 32 of dyes, 283 of soaps, 265 SUBJECT INDEX 323 Molybdio acid hydrosol, 127 Mordants, 271 tannin, 276 Multivalent ions on sols, 137, 139 Mutual precipitation of sols, 132 and kataphoresis, 135, 140 of dyes, 132, 139, 272 of emulsoids and suspensoids, 136 of ferric hydroxide hydrosol and arsenious sulphide hydro- sol, 136 and gold hydrosol, 140 gold hydrosol and thorium hydroxide hydrosol, 133 sols, 132 optimum concentration, 183, 134, 140 zones of, 133, 137, 139 solubility of liquids, 258 Negative adsorption, 244, 252 diaphragms, 71 sols, 76, 271 Nonylic acid, surface tension, 240 Oedometeb, 157 Oil emulsions, 85 kataphoresis, 85 precipitation, 112 Optical activity of gelatine, 145 properties of sols, 38 gels, 144 " Optical void," 39 Organic liquids, electroendosmose, 73 Organosols, alkali metals, colour, 41 preparation, 219 and water, 261 Brownian movement, 50, 52 kataphoresis, 83 metal, 215 preparation, 177, 178, 202, 203, 211, 214, 218, 219, 220, 261 OscUlation period, 49 Osmotic pressure and biology, 314 of Congo red, 315 of soaps, 314 of sols, 32 electrolytes on, 314 Oxidation methods of preparation, 190 Faai;'s methods of preparation, 180, 188 Pektisation, 10 Peptisation, 10 methods of preparation, 204, 209 of silicic acid, 15 Permeability of aluminium hy- droxide hydrogel, 90 gels, 145, 147, 149, 316 "Pickling "hides, 286 Platinum hydrosol, 186, 189 catalysis, 298 kataphoresis, 80 organosols, 50 Potential difference between glass and water, 68 metals and organic liquids, 83 Precipitation and adsorption, 102 concentration, 97 films, 148 of albumin, 112 dyesby other colloids, 272 emulsoids. 111 night blue by eosin, 273 suspensoids, 95 "salting out," 113, 124 valency rule, 98, 102 Preparations, 165 classification of methods, 165 I. Crystallisation methods, 167, 197 classification, 175 cooling solution, 176 hydrolysis methods, 192 oxidation methods, 190 reduction methods, 178 with protection, 188 replacement of solvent, 197 II. Electrical dispersion methods, 212 Billiter's method, 217 Bredig's method, 216 Svedberg's methods, 217-220 III. Solution methods, 204 peptisation, 204, 209 IV. Preparations — agar, 178 aluminium hydroxide, 194, 210 antimonious sulphide, 201 arsenious sulphide, 200 barium carbonate, 211 sulphate, 171, 202 bismuth, 190 blue gold, 187, 188 boron, 209 cadmium sulphide, 210 324 SUBJECT INDEX IV. Preparations — continued. copper, 190 oxide, 212 ferric hydroxide, 195, 196, 209 gelatine, 178 gold, 184, 185, 186, 187, 188, 189, 216, 218, 261 ice, 176 mastic, 177 mercury, 190 sulphide, 201 metal organosols, 218, 219 non-metal organosols, 220 osmium, 187 palladium, 185, 187 phosphorus, 176, 177 platinum, 186, 187, 189 purple of Cassius, 212 selenium, 177, 188, 189 silicic acid, 196, 201 silicon, 209 silver, 188, 185, 187, 188, 216, 218 iodide, 177 oxide, 188 sulphide, 203 sodium chloride organosols, 203 sodium ethersol, 219 sulphur, 176, 177, 191 tellurium, 187, 220 tungsten, 210 zinc sulphide, 200 zirconimn, 212 hydroxide, 195 Protalbic acid, 130, 264 Protection of sols, 128, 135 in preparations, 180, 180, 188, 202 QoADEiVALBNT ions, action on sols, 107 Eeduction methods of preparation, 178, 182 kathode, 181 Reversible precipitation of albumin 113 sols, 98 Salting out, 113, 267 Semi-colloids, 263 electrical properties, 2G4 Semi-permeable membranes, 147, 148, 314 Sewage purification, 292 Silicic acid gel dehydration, 151 hydration, 154 peptisation, 15 Silicic acid sol, 125 conductivity, 64 gelation, 125 hysteresis, 19, 153 kataphoresis, 83 preparation, 193, 196, 201 viscosity, 22 Silicon sulphide, hydrolysis, 198, 196 Silver hydrosol, 183 conductivity, 64, 184 kataphoresis, Al'" on, 96 salts on albumin, 118 sulphide hydrosol, 203 Size of particles, 54, 59 Soaps, 265 osmotic pressure, 314 surface tension, 231, 236, 268 Sodium chloride organosols, pre- paration, 171, 199, 203 organosols, preparation, 219 Soil, 290 adsorption of dyes, 291 salts, 291 Sol, 5 Solation, 10 Solid skins on albumin and dyes, 120, 273, 274 solution theory of dyeing, 280 Sols, boiling point, freezing point, vapour pressure, 86 Brownian movement, 47 colour, 40 conductivity, 61 density, 20 dialysis and difiusion, 24 dispersity, 54 filtration, 58 kataphoresis, 74 mutual precipitation, 132 optical properties, 38 osmotic pressure and molar Vfeight, 32 surface tension, 21 viscosity, 21 Solubility of gases in sols, 297 Solutides, 263 Specific surface, 8, 222 and adsorption, 245 SUBJECT INDEX 325 Spontaneous suspension, 48 Stabilising effect of ions, 105 Stability of metal sols, 215 Stannic acid hydrosol, 127 sulphide hydrosol, 212 Starch sol, 127 Static surface tension, 223, 230 at liquid interfaces, 258 of solutions, 230, 286 Structure of gels, 161 secondary, 163 Submicrons, 11 ■ reversible changes, 45 Surface concentration, 221, 228, 232 acetic acid, 235 sodium chloride, 235 sodium oleate, 236 defect and excess, 230, 238-4, 241 energy, 222 layer, thickness, 226 pressure, 227, 229 tension, 222 " active " solutes, 238 and surface energy, 222 at liquid interfaces, 258 concentration curves, 236, 240 dynamic, 223, 230, 241, 258 of emulsoids, 21 of fused salts, 224 of liquefied gases, 224 of liquids, 223 of mercury in gases, 254 of metals, 224 of soaps, 268 of solutions, 229, 237-240 lyotrope series, 9 of suspensoids, 21 temperature coefficient, 224 Suspensions, 4, 6 Suspensoids, 6, 18 mutual precipitation, 132 precipitation by emulsoids, 136 viscosity, 21 Tahnin, 276, 285 adsorption by cotton, 276 Tannin lakes, 276 molar weight in vyater and acetic acid, 285 Taiming, 285 gelatine, 289 mineral, 287 "pickling," 286 secondary changes, 288 Thorium hydroxide hydrosol, pro- tection by, 133 Thovert's relation, 35 Titanic acid hydrosol, 127 Toxin-antitoxin, 309 Tungstic acid hydrosol, 127 Turkey red, 277 Tyndall Efiect, 38, 40 Ultbafiltbation, 58 Ultramicroscope, 42, 44 Valency rule, 72, 98, 102 Vapour pressure at curved surfaces, 226 of gels, 150 of sols, 36 Velocity of crystallisation, 168 gelation, 122 kataphoresis, 76, 80 Viscosity of gelatine, 121 gels, 143 sols, 21 solutions, lyotrope series, 9 Wackbneodeb's solution, 190 Water, pure, for preparations, 181 purification, 293 Weathering of rooks, 292 Zinc salts on albumin, 118 sulphide hydrosol, 200 Zirconium hydrosol, 212 hydroxide hydrosol, 195 Zones of precipitation and non-pre- cipitation, 137-140 AUTHOR INDEX Abkgg, 289 Appleyard, 246, 278, 279 Aronstein, 116 Arrlienius, 309, 311 Aitmann, 130 Baeus, 58, 142 Baudouin, 74 Bayliss, 315 Bechhoia, 30, 58, 59, 99, 131, 146, 306, 311 Beilby, 18 van Bemmelen, 150, 151, 161 Benson, 235 Bert, 296 Berzelius, 12, 15, 178, 193, 196, 197, 208, 209 Bigelow, 27, 62 Billiter, 62, 82, 126, 184, 135, 181, 217 / BUtz, 43, 130, 133, 139, 195, 249, 292, 307, 308, 312, 313, 315 Blake, 62, 63, 64, 80, 81, 82, 182 Bodlander, 97, 98, 106 Bohr, 295 Bottazzi, 268 Bredig, 42, 54, 59, 94, 107, 110, 165, 213, 214, 299, 310 Brown, 47 Bruyn, L. de, 40, 199, 202 Burton, 80, 81, 82, 96, 107, 111,137, 214 ButsoMi, 141, 161 Buxton, 138, 272 Oastobo, 182, 183, 186 Chick, 21 Ooehn, 83 Cotton, 77, 80 Coward, 45, 78 Crum, 16, 193, 194 Davis, 281 Davy, 208, 213 Debray, 193 Debus, 191 Dewar, 257, 258 Donau, 183, 184 Donders, 296 Donnan, 316, 317 Dreaper, 136 Duclaux, 62, 109 Einstein, 51, 52, 56 Eisenberg, 308 Botvos, 225 Exner, 49 Paeadat, 38, 94, 128, 179, 182 Field, 311 Findlay, 297 Fischer, 280 Flamming, 126 Frankenheim, 224 FB^my, 15, 196 Freundlich, 3, 73, 91, 98, 100, 101, 103, 104, 245, 278, 279, 303, 310 Friedemann, 131, 136, 138 Galecki, 140 Garbowski, 182 Garnett, 18, 57 Garrett, 22, 125 Gayda, 21 Gay Lussac, 15 GefEoken, 297 Gemberling, 27, 62 Georgievios, 278, 280 Gibbs, 229, 232, 235, 237 Goldsobmidt, 267 Goodvrin, 194 Goppelsroder, 254 Gordon, 310 Gouy, 48 Graham, 1, 5, 10, 16, 19, 24, 25, 30, 33, 35, 98, 95, 108, 109, 125, 145, 193, 197, 201, 204, 209 Grimaux, 196 AUTHOR INDEX 32; GrosohufE, 126 Gutbier, 182, 183, 187 Hantzsch, 205 Hardy, 19, 84, 99, 100, 102, 127, 161 Hatsohek, 85, 112 Henri, 56, 110, 134, 307 Henrioh, 182 Herzog, 36 Hober, 310, 317 van't HofE, 32, 229 Hofmann, 259 Hofmeister, 113, 114, 121, 155, 159 Hopkins, 113 Ihglib, 74 Jacoeson, SOI, 303 Jolivet, 2H1 Jordis, 110 Kastobskij, 140 Kato, 173, 202 Kneoht, 277 Kobert, 303 KohlBchutter, 64, 107, ill, 182, 183, 185 Krafit, 265, 271, 283 Krecke, 193, 195 Krohnke, 292 Kuhn, 15, 125, 126, 20S Kundt, 144 Kiispert, 130, 182 Ku2el, 209, 210 Lagebgeen, 251 Lea, 179, 182 Levites, 121, 122, 123 Lewis, 85, 259 Lieck, 145 Liesegang, 146 Lillie, 36, 124 Linder and Picton, 20, 21, 30, 31, 42, 58, 76, 78, 80, 97, 104, 108, 132, 198, 200 Ljubawin, 126 Loe-wy, 295, 297 Losev, 278, 279 Lottermoser, 110, 128, 132, 133, 179, 180, 182, 190, 198, 201, 204, 212 Liideking, 142 MacBain. 266, 281 Hadsen, 313 Malfitano, 65 Martin, 21 Mayer. 110 Meyer; 128, 212 Michael, 199 Michaelis, 132, 271 MiUar, 74 Milner, 232, 235, 241 Moore, 314, 315, 317 Mouton, 77, 8U MuUer, A., 180, 195, 203, 204, 209, 210 Miiller, E., 213 Mylius, 126 Neisseb, 138 Neuberg, 209, 211 Neumann, 264 Nicolardot, 108 O'Sdllivan, 300 Ober, 104 Oden, 291 Ostwald, W., 3, 146, 233, 235, 245, 252 269 Ostwald, Wo., 223, 296 Paal, 130, 180, 182, 188, 199, 203, 211 Pappada, 127 Parker, 314 Paterno, 2, 285 Pauli, 22, 64, 84, 115, 117, 122, 123, 124, 162 Pean de St. Gilles, 16, 193 Perrin, 34, 56, 68, 69, 70, 73, 84 Pfeffer, 33, 36 V. d. Pfordten, 182 Philip, 295 Pioton. See Linder, Poggendorf, 12 Posternak, 115 Proctor, 286, 288, 308 Quincke, 18, 21, 67, 73, 82 Baffo, 191 Eamsay, 48, 51, 225 Ramsden, 120, 273 Rayleigh, 40, 67 Reinders, 259, 260, 281 Reinke, 157, 168 Reuss, 66 Bitter, 213 328 AUTHOR INDEX Eoaf, 315, 317 Rodewald, 36 Rohland, 291, 293, 294 BohlofE, 123 Bona, 123, 132 Ruer, 108 Rutland, 12 Sackue, 20, 265 Salimei, 285 Schmidt, 245, 279 Schneider, 212 Schroeder, 122, 123, 155, 157, 289 Sohulz, 113, 114, 130 Sohulze, 42, 182, 198, 200, 201 Selmi, 13, 14, 95, 191 Shields, 225 Siedentopf , 54 Smits, 265 Smoluchowski, 51, 52, 76 Sobrero, 13, 191 Soret, 229 Spiro, 19 Spring, 40, 209, 210 Steubing, 41 Stiasny, 288 Suida, 277, 280 Sutherland, 57 Svedberg, 41, 49, 50, 52, 58, 96, 110, 166, 181, 207, 214, 217, 218, 220 Tamman, 300 Taylor (R. L.), 266 Taylor (W. W.), 28, 64, 74, 181, 184, 188, 235 league, 138, 272, 311 Teresohin, 67 Thomson, 232 Thovert, 35, 36 Tompson, 300 Travers, 257 Tyndall, 38 Vanino, 182 Volk, 308 Wackbneodeb, 12, 190 Walker, 246, 278, 279, 282 Walter, 281 Webster, 317 Wedekind, 212 von Weimarn, 2. 6, 17, 114, 121, 162, 166, 167, 172, 206, 209 Whetham, 102 Whitney, 62, 63, 64, 80, 81, 82, 104 Wiedemann, 66 Wiener, 48 Wilson, 136 Winssinger, 198, 200, 202 Witt, 276, 279 Wohler, 179 Wood, 58 Wyroubofi, 109 Zawidski, 235 Ziegler, 30, 146 Zsigmondy, 11, 20, 41, 42, 45, 47, 49, 52, 54, 55, 94, 114, 129, 131, 179, 182, 183, 185, 186 Au-.^M THE END PKINTED Br WILLIAM CLOWES AND SONS, LIMniCD, LONDON AND BECCLES, ENGLAND.