<l 
 
 ar V 
 10448 
 
 i 
 
 THE 
 
 FORMATION OF COLLOIDS 
 
 SVEDBERG 
 
(ftoritpll HniOTrHitg ffiibrarjj 
 
 Sfifuca, Netn fork 
 
 BOUGHT WITH THE INCOME OF THE 
 
 SAGE ENIX)WMENT FUND 
 
 THE GIFT OF 
 
 HENRY W. SAGE 
 
 1891 
 
ArV1044 A"*" ""'"""^ '■'""^ 
 The formation of colloids. 
 
 3 1924 014 777 696 
 
B Cornell University 
 
 ^ Library 
 
 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/cu31924014777696 
 
MONOGRAPHS ON THE PHYSICS AND CHEMISTRY OF COLLOIDS 
 
 THE FORMATION 
 OF COLLOIDS 
 
 BY 
 
 THE SVEDBERG 
 
 PROFESSOR OF PHYSICAL CHEMISTRY IN THE 
 UNIVERSITY OF UPSALA 
 
 With 22 Illustrations. 
 
 NEW YORK 
 
 D. VAN NOSTRAND COMPANY 
 
 25, PARK PLACE 
 

 |\S\^-5^*] 
 
 TrinteJ In Qreal BrUain. 
 
PREFACE. 
 
 Owing to the rapid development of the science of 
 colloids, it has become a rather difficult task to write 
 a complete and modern text-book on this subject. 
 The literature containing the primary material is 
 verj' great, and grows greater every day. It seemed, 
 therefore, worth while to try to give a survey of the 
 chemistry and physics of colloids in the form of a 
 series of small monographs, treating the main parts 
 of the science separately. In this way it is possible 
 to move more freely and select the problems to be 
 reviewed according to their importance for the 
 subject. Moreover, a series of monographs can 
 easily be kept up to date by bringing out new 
 editions of the books dealing with those parts of the 
 subject that develop with special rapidity. 
 
 In the first monograph I mean to give a survey of 
 the processes which cause the formation of colloids — 
 or of heterogeneous systems with a relatively large 
 boundary surface on the whole (disperse systems 
 according to Wo. Ostwald)— especially with regard 
 to the conditions that determine the degree of 
 subdivision of the systems formed. As the forma- 
 tion of gels is a change of state within the disperse 
 system, it wUl not be considered in this monograph, 
 which deals with the formation of coUoids from non- 
 colloid material, but will be treated in one of the 
 following monographs. In order to bring forth 
 the leading ideas as clearly as possible, I have 
 refrained from giving detailed directions for the 
 preparation of the various colloids. Such informa- 
 
VI PREFACE. 
 
 tion _ may be found in my book, Herstellung 
 kolloider Losungen (Dresden, 1909), or in the original 
 papers referred to here. 
 
 My sincere thanks are due to Mr. Emil Hatschek 
 for his kindness in reading the manuscript before it 
 was sent to press. 
 
 THE SVEDBERG. 
 
 CJPSALA. 
 
CONTENTS. 
 
 PAGE 
 
 Introduction ........ 9 
 
 Formation of Disperse Systems in vacuo . . .15 
 Ultra- microscopic films produced by condensation of 
 metal vapours. Structure and stability. 
 
 Formation of Disperse Systems in Gases — the Gaseous 
 Dispersion Medium in some cases subse- 
 quently EXCHANGED FOR A LlQUID^ONE . . 19 
 Microscopic and ultra-microscopic films produced by 
 condensation of metal vapours in various gases. Pure 
 thermal vaporization. Vaporization by means of canal 
 rays. Degree of dispersion and experimental con- 
 ditions. Injection of superheated mercury vapour 
 into water. Dipping of incandescent metal wires in 
 water. Synthesis of colloids by means of the electric 
 arc. Free direct current arc in a gas. Condensation 
 of metal vapour on a solid and on a fluid surface. 
 Submerged arc. Condensation and dispersion product. 
 Colloids prepared by means of the free continuous 
 current arc. Enclosed continuous current arc. Inter- 
 mittent current arc. Damped oscillatory current arc. 
 Dispersity of dispersion product. Alloy electrodes. 
 Amount of dispersion and condensation product. 
 Purity of sols. Degree of dispersion of condensation 
 product. Colloids prepared by means of the damped 
 oscillatory arc. Undamped oscillatory current arc. 
 Vaporization of metal foil or wire by means of a heavy 
 current. Experiments of early investigators. 
 
 Condensation of supersaturated vapour by adiabatic 
 expansion. Variation of degree of dispersion with 
 degree of supersaturation. Formation of disperse, 
 systems in gases by chemical reactions. Smoke. 
 
 Dispersion processes in gases. Spray. Dust. 
 
 Formation of Disperse Systems in Liquids and in 
 
 Solids ........ 54 
 
 Condensation Processes. 
 
 Cooling. Lowering of solubility. Productionjof the 
 material of the disperse phase in supersaturated 
 
viii CONTENTS. 
 
 PAGE 
 
 solution. Chemical reactions. Reduction, oxidation, 
 dissociation, double decomposition. Preservation of 
 primary degree of dispersion. Purification. Organic 
 colloids. 
 
 Reduction. Gold sols. Mechanism of reduction and 
 condensation process. Gold nuclei. Particles of 
 other material acting as nuclei in gold reduction. 
 Degree of dispersion of gold sols. Action of light in 
 gold reduction. Effect of dilution. Formation of gold 
 ruby glass. Experiments of early investigators on 
 gold colloids. Silver sols. Silver mirrors. The latent 
 photographic image. Electrolysis of silver solutions. 
 Formation of silver sols by illuminating metallic silver 
 immersed in alcohol or water. Boiling of silver in 
 water. Classification of reduction processes. Pre- 
 paration of colloids of other metals by reduction. 
 Experiments of early investigators. 
 Oxidation. Sulphur sols. Experimental conditions. 
 Fractional coagulation. Purification. Electrolytic 
 pulverization. Experiments of early investigators. 
 
 Dissociation. Sulphur sols. Nickel sols. Organic 
 compounds. Dissociation of salts. Pyrosols. Blue 
 rock salt. 
 
 Double decomposition. Rapid reactions. Experi- 
 mental conditions and degree of dispersion. Concen- 
 tration of reacting ions. Temperature. Nature of 
 reacting ions. Slow reactions. Formation of As^Sj- 
 sols. Sols of other sulphides. Formation of sols by 
 hydrolysis. Fe^Oj-sols. Experiments of early investi- 
 gators. Decomposition of salts by acids and by 
 alkalies. Silver halides. Ferrocyanides. Benzosols- 
 of sodium halides. 
 
 Dispersion Processes. 
 
 Partial dissolution, grinding, emulsification. Oxidation 
 of tellurium sol. Peptization. Grinding with partial 
 dissolution. Sols of zirconium and tungsten, etc. Sols 
 of boron ; experiments of early investigators. Forma- 
 tion of sols by pure grinding. Grinding with solid 
 diluent. Two different emulsification processes. Mer- 
 cury sols. Homogenization of milk. Emulsification of 
 fats and hydrocarbons. 
 
 Name Index ........ 121 
 
 Subject-Matter Index ...... 123 
 
THE FORMATION OF 
 COLLOIDS. 
 
 INTRODUCTION. 
 
 The question of the formation of colloids will always 
 be connected with the name of Thomas Graham. 
 He was the first to point out the colloids as a special 
 class of chemical systems, and he was the first of 
 all investigators to discover a series of methods for 
 preparing colloid solutions (i). The study of the 
 works of earlier chemists has taught us since that 
 among them there were some who dealt with colloid 
 solutions and also, incidentally, expressed fairly 
 correct opinions as to the constitution of colloids, 
 as, for instance, Berzelius (2), Selmi (3), Faraday (4), 
 but none of them drew the attention of the chemical 
 world to the specific behaviour of colloids by means 
 of systematic series of experimental investigations 
 as Graham did. 
 
 Graham's opinion may be generally expressed 
 thus : " Crystalloids and colloids represent two 
 forms of existence ; they appear like different worlds 
 of matter." A radical distinction is assumed " to 
 exist between colloids and crystalloids in their 
 intimate molecular constitution." A crystalloid and 
 a colloid solution of the same substance must, 
 according to this view, be considered to contain two 
 allotropic modifications of the substance. Researches 
 during the last two decades have considerably 
 modified this point of view. That the colloid 
 solutions are heterogeneous systems has, chiefly by 
 the work of Linder and Picton (5), and especially by 
 
10 THE FORMATION OF COLLOIDS. 
 
 that of Zsigmondy (6), been proved beyond all 
 doubt ; and it is evident that, under suitable 
 conditions, every kind of matter can be brought into 
 colloid form without any change in its intra-molecular 
 constitution. Present-day research considers the 
 colloids and the crystalloids as two worlds, not from 
 an intra-molecular point of view, but from an extra- 
 molecular. The colloids are heterogeneous systems, 
 but, owing to their fine structure and large relative 
 or specific surface, they deviate considerably from 
 such systems as were formerly looked upon as 
 heterogeneous. In fact, they form a region between 
 these and the crystalloid solutions. Wo. Ostwald 
 has called them " the world of neglected dimensions," 
 because of the lack of interest which until very 
 lately caused them to be a disregarded field of 
 study (7). In a system containing two or more 
 phases, one of which is continuous (corresponding 
 to the solvent of a common crystalloid solution), 
 the continuous phase is called dispersion medium, 
 and the non-continuous phases, one, or more in 
 number (corresponding to the dissolved substance 
 or solute of a crystalloid solution), are called disperse 
 phases (8). Graham introduced the term "sol" 
 for a colloid solution. When the dispersion medium 
 is water, we have a hydrosol ; when it is alcohol, an 
 alcosol, etc. 
 
 Colloids or disperse systems in general may be 
 formed in two ways, differing in principle: by 
 condensation or by dispersion (9). In one case 
 matter is brought together within a smaller boundary 
 surface than before ; in the other case the matter is 
 dispersed so that the specific surface is increased. 
 In many cases diverse systems are formed by a 
 combined process of condensation and dispersion. 
 Often, on superficial observation, the process may 
 be considered, for instance, as a dispersion, while 
 an intimate analysis of the phenomenon shows that 
 
GENERAL PRINCIPLES. ii 
 
 a condensation process is inserted so .as to make 
 such a process the direct cause of the formation of 
 the disperse system (lo). Examples of such " false 
 dispersion processes " are met with in most forms of 
 electric coUoidation of metals. Thus, in general, the 
 pulverization of metals by means of the electric arc 
 must be a vaporization of metal with subsequent 
 condensation to a disperse system. In the case of 
 electrolytic pulverization the metal is often dissolved 
 as a chemical compound, then reduced, and at the 
 same time condensed, to a disperse system. If the 
 state of the system in the period that directly 
 precedes the formation of the disperse system be 
 taken into consideration, the classification of the 
 processes of formation in condensation and dispersion 
 wiU evidently be clearly defined. 
 
 The condensation processes are far more numerous 
 and common than the dispersion processes, and, in 
 fact, it is quite obvious that this must be so. As 
 every decrease of surface is accompanied by a libera- 
 tion of energy and vice versd, no negative surface ten- 
 sion being known, it is obvious that condensation is a 
 spontaneous, dispersion a forced, process. Of course, 
 the former case occurs more easily than the latter. 
 
 The primary condition for condensation is the 
 existence of a system of higher dispersity than the one 
 desired. If such a system, most often a moleciilar 
 one, is to be produced from condensed material, 
 this generally means the establishment of a forced 
 process {e.g., vaporization of compact metal in the 
 electric arc). It is only then that the spontaneous 
 process can begin to act (condensation of the metal 
 gas to metal powder). The secondary condition 
 for obtaining by condensation a disperse system 
 with more than a merely ephemeral existence is the 
 breaking off or the retardation of the condensation 
 process after having reduced the surface of the 
 system to a suitable extent. 
 
12 THE FORMATION OF COLLOIDS. 
 
 The primary condition for dispersion is the 
 occurrence of a condensed system of a lower degree of 
 dispersion than the one desired. This low degree of 
 dispersion can always be attained by meaiis of a 
 spontaneous process. The secondary condition for 
 obtaining a somewhat stable disperse system is the 
 prevention or the retardation of the spontaneous 
 retrogression of the dispersion process. 
 
 One of the most important tasks when investigat- 
 ing the formation of disperse systems is to state the 
 relations between the conditions of experiment and 
 the degree of dispersion of the system formed. 
 
 A distinction must be made between the primary 
 degree of dispersion (ii), which exists immediately 
 after the carrying out of the colloid-forming process, 
 and the secondary degree of dispersion, which is most 
 frequently brought about rather quickly in conse- 
 quence of the aggregation of the primary particles, 
 i.e., by coagulation. Further, one has to consider the 
 tertiary degree of dispersion, which is the result of a 
 process very slow compared with the two former 
 processes, i.e., recrystallization. In case of con- 
 densation the primary mean size of the particles is 
 determined by the number of condensation centres 
 per unit mass of the material of the future disperse 
 phase, and the primary distribution of size of the par- 
 ticles is regulated by the supply of material to the 
 various condensation centres. If, for instance, 
 centres are generated or inserted into the system 
 during the condensation itself, those centres that 
 appear later will receive less material than the 
 earlier ones, and thus they will become smaller than 
 these (12). 
 
 The centres of condensation may : 
 
 I. Be present in the system before the condensation 
 process sets in or be introduced independently of the 
 processes that lead to condensation [e.g., condensation 
 of vapour on ions in the adiabatic expansion experi- 
 
CONDENSATION CENTRES. 13 
 
 ments of C. T. R. Wilson (13), precipitation of gold on 
 small gold particles according to Zsigmondy (14) ). 
 
 2. Or they may he formed by the condensation f re- 
 cess or in intimate connection with it {e.g., ordinary 
 reduction of gold to gold colloid). 
 
 Case I is of course by far the more simple and 
 the more easily controlled. Besides the number of 
 centres and the transport of mass to them, i.e., time 
 of condensation for every single centre, one has to 
 take into account the nature of the centres or nuclei. 
 It has been shown that some nuclei are able to cause 
 condensation at high degrees of supersaturation but 
 not at low ones, while other kinds of nuclei extend 
 their action to lower degrees of supersaturation. 
 . In such a case the degree of dispersion rises with the 
 supersaturation. 
 
 In case 2, where the nuclei are formed exclusively 
 or to a great extent by the condensation process, it 
 is necessary to know how they are formed in order to 
 estimate the relation between the experimental con- 
 ditions and the degree of dispersion. In this respect 
 our knowledge is still very incomplete. It seems that 
 the primary degree of dispersion always rises with the 
 supersaturation, and consequently in general with 
 concentration (15) . If, however, the process by which 
 the material for the condensation is produced does 
 not take place fast enough, it may happen that the 
 supersaturation does not rise but falls with the 
 ■concentration (16), In many cases where a fall of the 
 degree of dispersion occurs with increasing concen- 
 tration it may be due to the fact that it is not the 
 primary but the secondary, or even the tertiary, 
 degree of dispersion that is observed. 
 
 Bibliography. 
 
 1. Graham, Phil. Trans., 151, 183 (1861). 
 
 2. Berzelius, Lehrb. d. Chem., 3 Aufl., 3, 65 (1834) ; 
 5 Aufl., 2, 269 (1844). 
 
14 THE FORMATION OF COLLOIDS. 
 
 3. Selmi, Nuovi Ann. di Scienze Natur., Bologna 
 (2), 8, 404 (1847). Cf. Koll.-Zeitschr., 8, 113 (iQii)- 
 
 4. Faxaday, Phil. Trans., 147, 145 (1857). 
 
 5. Linder and Picton, Journ. Chem. Soc, 61, 114, 
 137 and 148 (1892) ; 67, 63 (1895). 
 
 6. Zsigmondy, Zur Erkenntnis der Kolloide, Jena 
 
 (1905)- 
 
 7. Wo. Ostwald, Die Welt der vernachlassigten 
 Dimensionen, Dresden (1915). 
 
 8. Wo. Ostwald, Koll.-Zeitschr., i, 291 (1907). 
 
 9. Svedberg, Nova acta, Upsala (4), 2, No. i, p. 3 
 (1907). 
 
 10. Svedberg, Herstellung kolloider Losungen, Dres- 
 den, p. I (1909). Kohlschiitter, Die Erscheinungs- 
 formen der Materie, Leipzig, p. 180 (1917) . 
 
 11. Cf. Mecklenburg, Zeitschr. anorg. Chem., 74, 
 262 (1912). Oden, Arkiv for kemi., Stockholm, 7, 
 No. 26, p. 4 (1920). 
 
 12. Zsigmondy, Kolloidchemie, 2te Aufl., Leipzig, 
 p. 144 (1918). 
 
 13. C. T. R. Wilson, Phil. Trans., A189, 265 (1897) ; 
 A192, 403 (1899). 
 
 14. Zsigmondy, Zeitschr. phys. Chem., 56, 65 (1906). 
 
 15. V. Weimarn, Grundziige der Dispersotdchemie, 
 Dresden, p. 54 (1911) ; Zur Lehre von den Zustdnden 
 der Materie, Dresden, p. 165 (1914). Oden, Arkiv fSr 
 kemi., Stockholm, 7, No. 26, p. 33 (1920). 
 
 16. Picton, Journ. Chem. Soc, 61, 137 (1892). 
 Biltz, Nachr. Ges. Wiss., Gottingen, Math.-Phys. 
 Klasse, 1906, p. 141. Svedberg, Zeitschr. phys. 
 Chem., 65, 624 (1909). 
 
FORMATION OF DISPERSE SYSTEMS 
 IN VACUO. 
 
 The simplest form of condensation is the one that 
 taJies place in vacuo (i). Here we have no ordinary 
 dispersion mediimi, and the disperse system formed 
 can exist only through the support of a sohd or fluid 
 surface, which^acts as a dispersion medium. The 
 origin and nature of such condensates are of con- 
 siderable interest in judging of the primary quahties 
 of condensation. The researches of M. Knudsen (2) 
 and L. Hamburger (3) have shown that the con- 
 densation in vacuo is determined to a great extent 
 by the temperature of the walls and of the gas just 
 before condensation takes place. Accorc&ng to 
 Knudsen, it seems that a metal molecule which 
 encounters a surface of the same substance always 
 adheres to it. If the surface consists of another 
 material there is a critical temperature dividing the 
 domain of temperature into two parts. Below this 
 point the molecule adheres to the surface, above it 
 the molecule is reflected. For a glass waU and 
 mercury molecules Knudsen found the critical 
 temperatm-e to be about — 135", for glass and sUver 
 it is above + 575°. Thus mercury vapour adheres 
 without reflection to a glass surface colder than 
 — 135°, and silver when it is colder than + 575°. The 
 consequence of the absence of reflection is that if an 
 object is placed between the source of the metal gas 
 and the wall a shadow free from condensate will be 
 seen on it. The metal molecules are (in vacuo) sent 
 out in straight lines from the source [e.g., from a fine 
 incandescent wire), and in the case of non-reflection 
 
i6 THE FORMATION Op COLLOIDS. 
 
 the part of the wall that lies in shadow wiU get no 
 condensate on it. Above the critical point there 
 is no shadow, for metal is brought over to the 
 shaded part of the wall by diffuse reflection. On 
 account of the complete reflection of the metal 
 vapour by a glass wall of a temperature above the 
 critical point, very high degrees of supersaturation 
 may be obtained. R. Wood (4) and L. Hamburger 
 have observed this phenomenon in cadmium. Thus 
 Hamburger found that a cadmium wire heated in 
 vacuo by means of an electric current gave off gas 
 that did not condense on a glass waU at ordinary 
 temperatures. If the waU was touched with a cotton 
 plug dipped in liquid air, a rapid condensation took 
 place at that very point and lasted even when the 
 cooling had been stopped. Magnesium and zinc 
 showed the same phenomenon, but it was never 
 observed with silver, gold, platinum, tungsten, 
 molybdenum, nickel, iron and copper. These metals 
 are distinguished by the shadow phenomenon. 
 According to Knudsen, the critical condensation 
 temperature lies below — 78° for cadmium, mag- 
 nesium and zinc. It seems that condensates formed 
 on a wall cooled below the critical point have a much 
 higher degree of dispersion than those formed above 
 it. It also appears that even within the region above 
 it a lower temperature of the wall gives to the con- 
 densate a higher degree of dispersion. The tempera- 
 ture of the metal gas is also a very important factor. 
 Thus Hamburger has shown that elements with a 
 very high melting point, such as tungsten, carbon, 
 molybdenum, platinum, nickel, iron, give con- 
 densates of a very high degree of dispersion, often 
 optically quite unresolvable and without any 
 Tyndall light. In his experiments these elements, 
 owing to their high melting points, could be brought 
 to very high temperatures. Such elements as silver, 
 gold, copper, which could not be heated to those 
 
METAL FILMS. 17 
 
 very high temperatures, gave condensates of a little 
 lower degree of dispersion, and the easily melted 
 metals, magnesium, zinc, cadmium, gave the coarsest 
 condensates. Thus it is obvious that the degree of 
 dispersion rises with the diference of temperature 
 between the metal gas and the wall, viz., with the degree 
 of super saturation. 
 
 In these condensation experiments we are dealing 
 with the highest differences of temperature that can 
 possibly be obtained. Thus, for instance, the differ- 
 ence in temperature between tungsten gas from an 
 incandescent tungsten wire and a cold wall is about 
 3,000°. Hamburger's ultra-microscopic investiga- 
 tions of the condensates and his measurements of 
 their electric conductivity, carried out in collaboration 
 with E. Osterhuis, prove that some of these con- 
 densates, especially those derived from very high 
 melting -metals and condensed at low temperatures, 
 are to be looked upon as real films of amorphous 
 metal . These films disintegrate in time spontaneously 
 int discrete particles, probably with crystallization. 
 Heating accelerates the disintegration, and it seems 
 that the higher the vapour pressure of the metal, the 
 easier the disintegration of the film. Tungsten, which 
 of all substances has the lowest vapour pressure, 
 gives, the most stable films (thickness down to 0-5 mm) '> 
 silver, with a considerably higher vapour pressure, 
 gives very unstable films ; while cadmium, zinc and 
 magnesium, with their high vapour pressure, give no 
 films at all, but only granular condensates, the degree 
 of dispersity of which is lowered rather rapidly in 
 consequence of the growth of the larger grains at the 
 expense of the smaller ones. This disintegration of 
 the continuous metal films into particles is to be 
 considered as a continued condensation, for the sur- 
 face of the new system is no doubt smaller than that 
 of the film. 
 
 In transferring the condensates thus formed into 
 
 F.C. 3 
 
i8 THE FORMATION OF COLLOIDS. 
 
 the liquid state a considerable decrease in the degree 
 of dispersion takes place. Thus Knudsen observed 
 that continuous films of mercury, brown in trans- 
 mitted light, which, had been formed at a low tem- 
 perature ( — 183°), disintegrated into small drops when 
 the temperature was allowed to rise. 
 
 If the glass wall on which condensation is to take 
 place is covered beforehand with a thin film (for 
 example, 10 ix/jl thick) of some salt, no coherent metal 
 film is formed. Probably the metal molecules are 
 separated by the salt molecules. Such films represent 
 in a fresh condition, and at low metal content, a sort 
 of solid solution of metal in salt, and, when old and 
 •with higher metal content, probably a very fine- 
 grained metal colloid, the salt acting as a dispersion 
 medium. They have a considerably lower light 
 absorption than the coherent metal films or the fine- 
 grained pure condensates of metal particles. In 
 order to prevent the darkening of electric incan- 
 descent lamps, the makers have recently begun to 
 coyer the inside of the glass globes with very thin 
 films of certain salts, and the result has been very 
 good indeed. These observations are of great value 
 for our opinion as to the formation of disperse systems 
 in the solid state, which will be explained in detail 
 below. 
 
 No dispersion processes taking place in vacuo have 
 as yet been studied. 
 
 BiBLIOGEAPHY. 
 
 1. M. L. Houllevigue, Ann. chim. et phys. (8), 20, 
 138 (1910). 
 
 2. M. Knudsen, Annul. Phys. (4), 50, 472 (1916). 
 
 3. L. Hamburger, Koll.-Zeitschr., 23, 177 (1918). 
 
 4. R, Wood, Phil. Mag. (6), 32, 364 (1916). 
 
FORMATION OF DISPERSE SYSTEMS IN 
 GASES— THE GASEOUS DISPERSION ME- 
 DIUM IN SOME CASES SUBSEQUENTLY 
 EXCHANGED FOR A LIQUID ONE. 
 
 Next to condensation in vacuo, the simplest case 
 we have to consider is condensation in an indifferent 
 gas. It appears that the nature of the gas has a con- 
 siderable influence on the degree of dispersion of the 
 condensate. Two investigations of V. Kohlschiitter, 
 one dealing with the vaporization of metals by heat 
 in a glass or quartz tube and condensation on a glass 
 wall (i), and the other vaporization of metals by 
 means of canal rays and condensation on a glass wall 
 (= cathodic disintegration in vacuo) (2), are of great 
 interest. In both cases the degree of dispersion of the 
 condensate rises with the molecular weight of the 
 indifferent gas (Fig. i). Thus cadmium and zinc 
 gave more finely grained condensates in nitrogen 
 than in hydrogen, arsenic and selenium more finely 
 grained ones' in carbon dioxide than in hydrogen. 
 When silver is vaporized by canal rays the condensate 
 is most fine-grained in argon, less so in nitrogen, and 
 coarsest in hydrogen. This relation between the 
 molecular weight and the degree of dispersion of the 
 system formed vanishes with increasing rarefaction 
 of the gas. It is, however, perceptible at rather low 
 pressures, and is stiU marked at 50 mm. Kohlschiittei 
 found that the degree of dispersion of the condensate 
 rises with the pressure of the gas (Fig. i). This appears 
 to be opposed to Hamburger's observation of very 
 highly disperse metal condensates in vacuo. When 
 
 2—2 
 
20 THE FORMATION OF COLLOIDS. 
 
 Nitrogen. Hydrogen. 
 
 
 
 -«w- '. 
 
 / ' 
 
 /• . '. ■ 
 
 ♦ *• 
 
 • : *. • • •. • 
 
 >* • ? * • , 
 
 V • • •• " 
 
 - * , » • 
 
 1 o 
 
 / • 
 
 • v^ 
 
 
 
 ^■^:.;:ii^' 
 
 
 ,._ _..- . ^. - -^ 
 
 Fig. 1. — C^a^sation of zinc vapours. (According 
 to Kohlschiitter.) 
 
 collecting the condensate on a wall in the presence 
 of a gas, the condensation probably takes place to 
 some extent in the gas itself, i.^., near the wall, but 
 not on the wall. In vacuo the condensation must, of 
 
KIMURA'S EXPERIMENTS. 21 
 
 < 
 
 course, take place on the wall. This may account for 
 the above-mentioned discrepancy between the obser- 
 vations of Kohlschiitter and Hamburger. 
 
 By the injection of superheated mercury vapour 
 into water I. Nordlund prepared disperse systems of 
 mercury (3). The degree of dispersion was rather 
 high in part of the system. Here we are obviously 
 concerned with the condensation of mercury vapour 
 in an atmosphere of water vapour. The condensa- 
 tion may partly take place more freely in this 
 atmosphere, partly close to the cool water-surface. 
 The small mercury droplets seem to be formed in the 
 latter, the large ones in the former way. 
 
 Some very peculiar observations made by 
 Kimura (4) would have come under this heading 
 had they laeen correct. Kimura heated a metal wire 
 (in a flame or by means of an electric current) and 
 dipped it quickly in water. An ultra-microscopic 
 examination proved the presence of some small 
 particles. In the case of the easily oxidizable metals, 
 aluminium, zinc, iron, nickel, antimony, tin, bis- 
 muth, Kimura's observations are to be explained 
 by the action of an oxidation process. In the case 
 of platinum and gold the explanation might have 
 been as follows : The incandescent wire is to some 
 extent surrounded by a layer of metal vapour, and 
 by immersing the wire in water this vapour is con- 
 densed in an atmosphere of water vapour to ultra- 
 microscopic particles, which are then taken up by the 
 water. A careful examination of Kimura's experi- 
 ments by H. Nordenson (5) has proved, however, that 
 in the case of platinum and gold no formation of 
 metallic particles takes place. The particles observed 
 by Kimura are to be explained as dust or some other 
 contamination brought into the water during the 
 process. 
 
 The most effective method for the vaporization of 
 the metals is to form an electric arc between elec- 
 
22 THE FORMATION OF COLLOIDS. 
 
 trodes of the metal in question. Even the most 
 refractory metals are in this way vaporized rather 
 abundantly. The amount of vaporization and the 
 degree of dispersion of the colloid formed depend 
 not only on the electric properties of the arc {i.e., 
 potential difference, current intensity, and period of 
 current), but also on the nature of the surrounding 
 gases and the solid or liquid surfaces near to it. It 
 has been customary to distinguish between arc 
 dispersion in gases and in liquids. In both cases, 
 however, the arc is burning in gas ; the difference 
 lies only in the manner of condensation of the metal 
 vapour. Even in this respect there are transition 
 forms, and thus it seems most rational to treat all 
 these phenomena together and to consider them all 
 under the heading " Formation of Disperse Systems 
 in Gases." 
 
 With regard to the electrical conditions, there are 
 two extreme cases : direct current arc and high- 
 frequency current arc. Transition forms between 
 these are, intermittent current arc and low-frequency 
 alternating current arc. From a thermal point 
 of view there are two forms to be considered, 
 free arc and enclosed arc. With regard to conden- 
 sation, we have condensation in gas, on a liquid' 
 surface, and on a solid surface. 
 
 If a direct current arc, for instance, between silver 
 electrodes, is maintained in an indifferent gas, 
 e.g., nitrogen, the condensation takes place in the 
 nitrogen, and a rather coarse condensate is produced, 
 settling gradually on the solid walls of the vessel. 
 But, as F. Ehrenhaft and his students have shown, 
 the smallest particles remain suspended in" the gas 
 for hours (6). According to some incidental obser- 
 vations made by these investigators the degree of 
 dispersion rises as the strength of the current 
 diminishes (7). 
 
 If the arc is permitted to burn close to a solid 
 
ARC PULVERIZATION. 23 
 
 surface — for instance, of glass — in such a way that 
 the hot metal gas strikes the surface, part of the 
 metal gas is condensed on it. Owing to the rapid 
 cooling, and probably, also, to the breaking off of 
 the coagulation process, the degree of dispersion of 
 this condensate is higher than when it is formed 
 freely in the gas. If the arc is allowed to burn close 
 above the surface of a liquid, the condensation takes 
 place partly in the surface, and this condensate is 
 taken up by the liquid, forming a coUoid solution (8). 
 The degree of dispersion is greater than when the 
 condensation takes place in a gas. 
 
 The condensation on a liquid surface can be 
 brought to completion by surrounding the arc on aU 
 sides with liquid. Thus we have the submerged arc 
 or the so-caUed arc pulverization process in a liquid. 
 It seems, however, that, even in this case, the main 
 part of the metal gas is condensed in the vapour from 
 the liquid that surrounds the arc, and not on the 
 surface between the vapour and the liquid. The 
 metal vaporization in the arc is much greater if the 
 arc is surrounded by a liquid than by a gas. Thus 
 the loss in weight of the electrodes in the case of the 
 cadmium arc of i-o ampere is, in nitrogen, about 
 0-0004 g™- psr minute ; in ethyl ether, o-ooi6 gm. 
 per minute, or four times more in ether than in 
 nitrogen (8). The explanation of this increased 
 vaporization of metal may be as f oUows : Owing 
 to the effective condensation the equilibrium in th? 
 arc is rapidly disturbed, and this causes abundant 
 vaporization of metaJ in order to re-establish the 
 equilibrium. 
 
 Besides the metal gas and the gaseous products 
 from the surrounding liquid, the arc also contains 
 metal globules formed from the melted metal on the 
 electrode surfaces by the action of electro-mechanical 
 forces. Some of these melting globules leave the 
 arc and are taken up by the surrounding liquid. 
 
24 THE FORMATION OF COLLOIDS. 
 
 forming a rather coarse sediment on the bottom of 
 the vessel (9). 
 
 The portion of the coUoid formed by condensation 
 of metal gas has a much higher degree of dispersion 
 than the portion formed by dispersion of melted 
 metal from the electrodes. The loss in weight of the 
 electrodes rises proportionally to the square of the 
 current, and is but little influenced by the length of 
 the arc. The decomposition of the medium by the 
 
 60 volt 
 
 ^. 
 
 
 1 
 
 A 
 
 ^m^ 
 
 
 R 
 
 Fig. 2. — Arrangement of electric pulverization. 
 (According to Bredig.) 
 
 arc increases very rapidly as the length of the arc 
 increases (10). 
 
 The preparation of coUoid solutions by means of 
 the arc was discovered by Bredig (i8g8) (11), and 
 has since been studied specially by the writer and his 
 collaborators (1907-1920) (12). Bredig used a con- 
 tinuous current of 5 to 10 amperes and 30 to no 
 volts, and the arc was allowed to burn freely in the 
 hquid. As a dispersion medium he used pure water 
 that was cooled by immersing in ice and water the 
 vessel in which the arc was burning. Fig. 2 gives 
 
BREDIG'S SOLS. 25 
 
 the arrangement for electric pulverization according 
 to Bredig. A is an ammeter, P the pulverization 
 vessel, and R an adjustable resistance. 
 
 He obtained hydrosols of the noble metals, viz., 
 platinum, palladium, iridium, gold, and silver. 
 Using very pure, cooled water that had been freed 
 from air, he could even get the hydrosol of cad- 
 mium. More positive metals than cadmium give 
 sols of their oxides. Degen (1902) (13) and Burton 
 (1906) (14) tried to disperse metals in organic liquids, 
 such as ethyl alcohol and methyl alcohol, by means 
 of the arc, but the considerable decomposition of the 
 medium prevented the formation of pure and stable 
 sols.' 
 
 The mean temperature of the continuous current 
 arc, and, consequently, the vaporization of metal in 
 it, can be considerably increased by using a suitable 
 thermal protection. Thus we get what the writer 
 has called an enclosed arc in contradistinction to the 
 free arc described above. It has been shown that 
 of the two electrodes the anode is most sensitive to 
 thermal protection. Of course, this phenomenon is 
 the more marked the greater the cooling power of 
 the surrounding medium. Thus it is more marked 
 when the arc is immersed in a liquid than in a gas. 
 If an arc between silver wires of about i mm. 
 diameter is enclosed in a quartz tube provided with 
 a little hole right in front of the arc, and if a current 
 of nitrogen is let in from both ends of the tube, as 
 shown in Fig. 3, we get an effective arrangement for 
 producing silver vapour of high temperature and 
 great concentration. A magnetic field may be 
 appUed in order to facilitate the emergence of the 
 silver gas through the hole. 
 
 If there is a great distance between the hole of 
 the quartz tube from which the nitrogen gas loaded 
 with sUver vapour emerges and any solid or fluid 
 wall, the result is a condensate, the degree of disper- 
 
26 THE FORMATION OF COLLOIDS. 
 
 sion of which is equal to the condensate obtained 
 from the free arc in a gas. If, however, the aforesaid 
 distance is small enough to let the silver vapour, while 
 
 silver 
 
 nitrogen 
 
 N 
 
 1 
 
 glass 
 
 rubber 
 
 quartz 
 nitrogen 
 
 S 
 
 nitrogen 
 
 Fig. 3- — ^Arrangement for the production of silver vapour. 
 
 still hot, encounter a solid or fluid wall, a condensate 
 with a much higher degree of dispersion is formed. 
 Thus we get a highly disperse, yellowish-red conden- 
 sate on a glass wall placed some miUimetres in front 
 
ENCLOSED ARC. 
 
 27 
 
 of the hole, while the free arc only produces a black, 
 coarse condensate. If the arc vapours are blown 
 against an ethyl alcohol surface, the silver vapour is 
 condensed to very small particles, which form a 
 reddish-yellow alcosol. The free silver arc burning 
 close to an alcohol surface also produces a silver 
 alcosol, though a rather dilute one, and the particles 
 are much greater, the sol being greenish-black in 
 
 r" 
 
 Fig. 4. — Spectrum of the free silver arc in ethyl alcohol (a) ; 
 the free silver arc in nitrogen (6) ; and the enclosed silver 
 arc in ethyl alcohol (c). 
 
 colour. Thus the higher the speed of the conden- 
 sation process and the shorter the period during 
 which the particles are kept in the gaseous medium 
 (where coagulation proceeds very rapidly), the 
 higher the degree of dispersion of the product 
 formed. 
 
 If the hole in the quartz tube opens under the 
 surface of a liquid, the condensation and, conse- 
 quently, also the formation of colloid, are much more 
 complete. In some experiments with the quartz 
 tube immersed in ethyl alcohol the writer found a 
 
28 ■ THE FORMATION OF COLLOIDS. 
 
 formation of colloid several times greater than with 
 the free arc immersed in the same liquid. When 
 
 hot metal vapour con- 
 denses on a liquid surface 
 there is the risk that 
 some chemical decompo- 
 sition of the liquid may 
 take place. This process 
 is still more marked if the 
 liquid is allowed to touch 
 the hot electrodes, and if 
 vapours of the liquid get 
 into the arc. Spectro- 
 scopical investigations 
 have shown that in the 
 case of the free arc, e.g., 
 sUver in ethyl alcohol, the 
 arc is to some extent- 
 built up of gas from the 
 liquid. This gas is partly 
 decomposed, the spec- 
 trum of the arc showing, 
 in addition to >the silver 
 lines, the band spectrimi 
 of carbon (Fig. 4, a) . The 
 spectrum of the free silver 
 arc in nitrogen is given in 
 Fig. 4, b. 
 
 In the case of the 
 enclosed arc (arrangement 
 shown by Fig. 5) , however, 
 the hot nitrogen and silver 
 gases coming out from 
 the hole in the quartz 
 tube and encountering the alcohol only cause very 
 slight decomposition, the spectrum of the arc showing 
 only the silver lines and some continuous light 
 probably emitted by the hot quartz tube (Fig. 4, c) . 
 
 Fig. 5a. — Arrangement 
 electric pulverization 
 metals with enclosed arc. 
 
APPARATUS FOR ENCLOSED ARC. 29 
 
 It is obvious that the glowing parts of the quartz 
 tube are emitting vapours of sihcic acid, and, as a 
 
 Fig. 5b. — Apparatus for electric pulverization of metals 
 with enclosed arc. 
 
 matter of fact, the colloids formed contain consider- 
 able amounts of colloid silicic acid. In a silver 
 alcosol prepared by means of this arrangement 
 
30 THE FORMATION OF COLLOIDS. 
 
 Borjeson found the disperse phase to contain 40 
 
 per cent, of sihcic acid. 
 
 As stated above, a considerable 
 change takes place in the produc- 
 tion of colloid, even when only 
 the anode, and not the cathode, 
 is thermally protected against the 
 cooling power of the liquid. Thus 
 with the arrangement indicated 
 by Fig. 6 we get a silver coUoid 
 of a much higher degree of dis- 
 persion than with the free silver 
 arc in alcohol. 
 
 The nature of the indifferent 
 gas used in the experiments with 
 the enclosed arc influences the 
 degree of dispersion of the system 
 formed. Thus in the case of the 
 only partly protected arc (Fig. 6) 
 we get a coarser colloid when 
 using hydrogen instead of nitro- 
 gen as the indifferent gas. In the 
 case of the completely enclosed 
 arc (Fig. 5) the difference is 
 only slight. It is of interest to 
 notice that Kohlschiitter observed 
 just the same difference between 
 nitrogen and hydrogen with regard 
 to the condensation on a glass wall 
 of metal vapour produced in a 
 
 Fig. 6.— Arrange- purely thermal way. The agree- 
 ment for electric ment may, perhaps, be only acci- 
 pulverization o f dental ; as far as the arc is con- 
 metals with partly j ii_ -u -u 
 
 protected arc cerned, the phenomenon can be 
 
 explained by the greater thermal 
 conductivity, i.e., the greater cooling power of the 
 hydrogen, while in Kohlschiitter's experiment the 
 explanation can hardly be the one suggested. 
 
INTERMITTENT CURRENT ARC. 
 
 31 
 
 The intermittent current arc, such as we get it, 
 for instance, by inserting a mercury turbine inter- 
 rupter (interruption number about 150 per second) 
 in the current circuit of a common direct current arc, 
 differs from the ordinary continuous arc in the 
 following respect (15) : the portion of the system 
 formed that consists of rather large particles is much 
 increased. This is just what might be expected, for 
 
 
 
 
 
 
 
 
 ©1 
 
 a 
 
 . 
 
 L H I 
 
 3 
 
 
 
 
 
 {} M 
 
 Fig. 7. — Arrangement for electric pulverization of metals with 
 damped oscillating arc. 
 
 in the case of the intermittent arc the metal gas as 
 well as the vapours from the liquid are condensed 
 to a great extent, or even completely, at every 
 interruption, thus causing the Hquid to fill up the 
 space formerly occupied by the arc and scattering 
 the melting globules of arc into the surrounding 
 liquid. The low-frequency alternating current arc 
 (frequency about 50 per second) produces nearly the 
 same kind of disperse systems as the intermittent arc. 
 
32 THE FORMATION OF COLLOIDS. 
 
 The arc formed by an alternating current of very- 
 high frequency — a so-called oscillatory current — 
 (period lo-' to lo-* seconds) shows a rather different 
 behaviour. Two forms of such arcs have been 
 studied, viz., the damped oscillatory arc produced 
 by a Leyden jar circuit (i6) and the undamped 
 oscillatory arc (wave current) (17) as produced by a 
 sort of Poulsen circuit invented by the writer (18). 
 Figs. 7 and 8 show the arrangement of apparatus. 
 L is a Leyden jar, H a hot-wire ammeter, P the 
 pulverization apparatus, I an induction coil, C a 
 
 '+'*0 volt 
 
 L H P A 
 
 Fig. 8. — Arrangement for electric pulverization of metals with 
 undamped oscillating arc. 
 
 condenser, M a mercury interrupter, R an adjustable 
 resistance, A an ordinary ammeter. 
 
 We will first consider the oscillatory arc with' 
 damping. If the arc is surrounded by a gas, the loss 
 in weight of the electrodes, i.e., the production of 
 colloid, is very small, only about ^g of the value 
 observed for the free direct current arc, the effective 
 intensity of the current being the same. If, however, 
 the arc is immersed in a liquid, the loss of weight is 
 of the same order of magnitude as for the free con- 
 tinuous current arc in a liquid. Just as in the case 
 of the low-ffequency alternating current arc and the 
 
PULVERIZATION OF ALLOYS. 33 
 
 intermittent current arc two different processes take 
 place, viz., vaporization of the metal and condensa- 
 tion of the metal gas to small particles and a real 
 dispersion of melted metal (19). The dispersion pro- 
 duct formed is separated by a rather large interval 
 from the part of the disperse phase formed by the 
 condensation process, and it can easily be removed 
 from the condensate by means of sedimentation. In 
 the case of cadmium in ethyl alcohol the radius of 
 the globules formed by pure dispersion lies between 
 the limits 0-5 to 25 ju,, while the mean value of the. 
 radius of the condensation product is about 5 fj-fx. 
 For tin the average maximum value of the radius 
 wasTound to be about ro /j, and for bismuth 100 ;u,. 
 These figures, of course, only indicate the order of 
 magnitude of the radii. 
 
 When pulverizing electrodes of a binary alloy in 
 ethyl alcohol Borjeson found that the two metals 
 were distributed between the condensation and the 
 dispersion products in the following way. If the two 
 metals had nearly the same boiling point the sedi- 
 ment and the disperse phase of the colloid solution 
 were found to have about the same composition. If 
 the metals had different boiling points the sediment 
 was richer in the metal of high boiling point (Table i) . 
 This statement is strong evidence in favour of the 
 view that the sol is formed by condensation of metal 
 gas and the sediment by dispersion of melted metal. 
 As to the composition of the various parts of the 
 sediment, it was found that the fine-grained fractions 
 contained less of the volatile metal than did the 
 coarser fractions. This fact shows that the melting 
 globules must, have been formed when the arc was 
 still burning and that they must have remained in 
 the arc for some time, permitting the volatile metal 
 to evaporate to some extent from the surface of the 
 globules. Owing to their larger specific surface the 
 evaporation per unit mass must, of course, be greater - 
 
34 THE FORMATION OF COLLOIDS. 
 
 in the case of the smaller globules than in the case 
 of the larger ones. The hypothesis suggested by 
 Benedicks (20) that the melting globules are formed 
 by the streaming of the hquid against the molten 
 parts of the electrode surfaces when the arc breaks 
 down at the interruptions of the current cannot be 
 supported. If such a process occurs it must be of 
 very little importance. 
 
 Table i. 
 
 
 Ratio of components. 
 
 Alloy and boiling point of 
 components. 
 
 In 
 electrodes. 
 
 In sediment. 
 
 
 Coarse 
 fraction. 
 
 Fine- 
 grained 
 fraction. 
 
 In sol. 
 
 Au (2,500°)— Cd (780°) , 
 Au (2,500°)— Sn (2,270°) 
 Bi (1,420°)— Cd (780°) 
 
 0-84 
 0-74 
 10 
 
 1-7. 
 075 
 1-4 
 
 30 
 
 o-8i 
 
 2-0 
 
 052 
 0-71 
 
 0-77 
 
 In liquids the high-frequency arc always produces 
 more finely grained and purer colloids than the free 
 continuous or low-frequency arc. The amount of 
 condensate and of dispersion product and their 
 degrees of dispersion as well as the" decomposition 
 of the hquid (the dispersion medium) depend on the 
 electrical conditions of the oscillation circuit, viz., 
 intensity of current, capacity, self-induction, resist- 
 ance, length of the arc, etc. They are also dependent 
 on the nature of the electrode metal and the nature 
 • and temperature of the dispersion medium. 
 
 With a capacity of about 0-0004 ™-f-. low self- 
 induction and low resistance in the oscillation circuit, 
 the fraction of the disperse phase produced by pure 
 dispersion is about 20 per cent, for cadmium in ethyl 
 
DISPERSION PRODUCT. 
 
 35 
 
 alcohol. With increasing capacity this fraction 
 increases, amounting to about 40 per cent, at a 
 capacity of 0-003 ^nf- ^.nd to about 55 per cent, at 
 0-09 m.f. In the case of metals with a high melting 
 point the fraction produced by pure dispersion is 
 smaller than in the case of a metal of low melting 
 point (Table 2) (21) . 
 
 Table 2. 
 
 Medium = ethyl alcohol. Capacity = 0-003 ^i- 
 Current intensity = 1-4 to 1-7 amperes. 
 
 Metal. 
 
 Dispersion product in 
 percentage of total loss 
 in weight of electrodes. 
 
 ' Melting point. 
 
 Pt 
 
 Au 
 Zn 
 Cd 
 Sn 
 Bi 
 
 20-2 
 
 25-8 
 41-0 
 41-6 
 56-8 
 59-3 
 
 1,700° 
 1,064° 
 
 419° 
 320° 
 232" 
 268° 
 
 The loss in weight of the electrodes is proportional 
 to the square of the virtual current and is equal for 
 both electrodes. In Fig. 9 the curves representing 
 the relation between current and total loss in weight 
 of electrodes in ethyl ether are shown. 
 
 The current intensity being equal, the pulveriza- 
 tion decreases a little with increasing capacity, 
 increases a little with increasing self-induction, 
 decreases rapidly with increased resistance and with 
 increased arc length. The decomposition of the 
 medium decreases rapidly with increased capacity, 
 and increases rapidly with increased self-induction, 
 increased resistance, and a little with increased arc 
 length. Low temperature depresses the decomposi- 
 tion of the medium considerably, but has no per- 
 ceptible influence on the j^ate of pulverization of the 
 
 3-2 
 
36 THE FORMATION OF COLLOIDS. 
 
 1^* O/iv^fJUtt 
 
 Fig. 9. — Variation of total loss in weight of electrodes per 
 minute with current in the case of electric pulverization 
 in ethyl ether by means of , the damped oscillatory arc. 
 
SIZE OF PARTICLES. 37 
 
 electrodes. Hence the ratio between decomposition 
 of medium and pulverization of electrodes is smallest 
 when the capacity is great, when the self-induction, 
 the resistance and the arc length are small, and when 
 the temperature is low. Under -these circumstances 
 we also get the purest sols. 
 
 The question of the relation between the degree of 
 dispersion of the condensate and the experimental 
 conditions is a rather complicated one. Coagulation 
 sets in immediately after the formation of the par- 
 ticles, and thus alters the degree of dispersion in 
 a manner very difficult to control. Borjeson by 
 measuring the decrease of the degree of dispersion 
 with time was able to calculate the approximate size 
 of the particles at the time zero, and found the initial 
 degree of dispersion to be almost independent of 
 capacity, self-induction,' and current intensity. The- 
 initial size of particles was found to be larger in fluids 
 of higher viscosity and accordingly somewhat larger 
 at low temperatures. The particles from various 
 nietals do not generally differ much in size. About 
 one minute after the formation of the sol Borjeson 
 found the following figures (Table 3) : — 
 
 Table 3. 
 
 O-Scillatory arc. Medium : ethyl alcohol. 
 Temperature = — 75° 
 
 
 Metal. 
 
 Radius of particles in fj-ij.. 
 
 Au 
 
 
 2-8 
 
 Zn 
 
 
 2-9 
 
 Pt 
 
 
 3-8 
 
 Cd 
 
 
 50 
 
 Sn 
 
 
 12-6 
 
 Bi 
 
 
 17 
 
 The large values of Sn and Bi may be due to coagulation. 
 
38 THE FORMATION OF COLLOIDS. 
 
 The decrease in the degree of dispersion in most 
 cases reached a Umiting value after some time, and 
 this secondary degree of dispersion was found to 
 
 depend on many 
 factors. Low tem- 
 perature is one of 
 the principal means 
 for attaining sols of 
 high secondary 
 degree of dispersion. 
 Further, the rate of 
 production of colloid 
 must be slow, i.e., 
 the capacity and the 
 intensity of current 
 must not be too high 
 and the arc npt too 
 short. 
 
 By means of this 
 form of electric pul- 
 verization the sols of 
 alniost aU metals 
 have been obtained 
 (22). The following 
 dispersion media 
 have been used : 
 water; methyl, 
 ethyl, propyl, 
 butyl, and amyl 
 alcohol ; ethyl ether ; 
 acetone ; ethyl and 
 amyl acetate ; chlo- 
 roform, etc. As pul- 
 
 -Spark 
 electric pulverization of metals. 
 
 verization apparatus the writer used a kind of spark- 
 micrometer that could be immersed into a beaker 
 filled with dispersion medium (Fig. 10) . In preparing 
 the sols of the alkali metals it is necessary to work in 
 an iridifferent atmosphere and to use an indifferent 
 
SODIUM SOLS. 
 
 39 
 
 dispersion medium, e.g., ethyl ether, of a high degree 
 of purity. Low tem- 
 perature faciUtates the 
 process. In Fig. ii a 
 suitable apparatus for 
 the preparation of 
 sodium and potassium 
 sols is represented. 
 Some fragments of the 
 metal to be pulverized 
 are introduced into the 
 tube T, which is pro- 
 vided with two holes at 
 the bottom and two 
 platinum wire electrodes 
 Pi and Pg. The flask F 
 contains a supply of 
 dispersion medium and 
 some purifying sub- 
 .stance, e.g., sodium wire 
 in the case of ethyl 
 ether. Hydrogen can 
 be conducted through T 
 and F by means of the 
 rubber tubes Rj and Rj. 
 Sparking and pulveriza- 
 tion go on almost auto- 
 matically when the cur- 
 rent is closed, and can 
 be facilitated by slightly 
 tapping the tube T. 
 
 With regard to the 
 production of colloids by 
 means of the high-fre- 
 quency arc without 
 damping (" wave cur- 
 rent arc "), the process on the whole very much 
 resembles that with the damped high-frequency arc. 
 
 Fig. II. — Apparatus for_ the 
 preparation of sodium and 
 potassium sols. 
 
40 THE FORMATION OF COLLOIDS. 
 
 It differs from it in the following respect : The 
 production of colloid is much greater in the case, of 
 the wave current arc. For i ampere it is about three 
 times as great. The decomposition for i ampere 
 is, however, only increased to about double, so that 
 the specific decomposition, i.e., the ratio between 
 decompo-sition and loss in weight of electrodes, is 
 smaller in the case of the wave current arc. The 
 order of the various metals, arranged according to 
 the total loss in weight of electrodes per minute, is 
 not quite the same in the case of the wave current 
 as in that of the damped oscillatory current arc. 
 In Table 4 some figures demonstrating this fact are 
 given. 
 
 Table 4. 
 
 Current intensity =1-5 ampere. 
 
 Oscillatory current arc; capacity 
 = 0-003 m-f- 
 
 Wave current arc ; capacity 
 = o'oi.^ m.f. 
 
 Metal. 
 
 Total loss in 
 
 weight of electrodes 
 
 in milligrammes 
 
 per minute. 
 
 Metal. 
 
 Total loss in weight 
 
 of electrodes in 
 
 milligrammes per 
 
 minute. 
 
 Pb . 
 
 Bi . 
 Sb 
 
 Cd 
 
 Zn . 
 Au . 
 Pt . 
 Al 
 
 Ag . 
 Cu . 
 
 
 45 
 33 
 
 25 
 21 
 
 8 
 6 
 
 3 
 2 
 
 1-3 
 
 I-O 
 
 Sb . 
 Bi 
 
 Pb . 
 Cd 
 
 Zn . 
 Au . 
 Pt . 
 
 ii ■ : 
 
 Al . 
 
 130 
 
 75 
 64 
 48 
 
 25, 
 10 
 
 4 
 
 2-6 
 2-0 
 O-I 
 
 The loss in weight of electrodes is not equally 
 divided between cathode and anode, but the anode 
 loses more than the cathode. With increasing length 
 
PULVERIZATION OF WIRES. 41 
 
 of arc this difference becomes less marked. The 
 ratio between dispersion ancj condensation product is 
 a little less than in the case of the damped oscillatory 
 arc. The secondary degree of dispersion of the 
 colloid formed is about the same as in the case of the 
 damped oscillatory arc, but macroscopic coagulation 
 often takes place during or immediately after the 
 formation. 
 
 Another rnethod of preparing disperse systems 
 with metal as disperse phase, which is clearly related 
 to the process just described, consists in passing an 
 electric current of high intensity through a foil or 
 wire of the metal in question. The metal is partly 
 volatilized and partly melted, and gives rise to a 
 disperse system. If the process takes place in gas, 
 the particles settle on the walls of the vessel, forming 
 coloured films. Such films were already obtained 
 by van Marum (1787) (23), and very ingeniously 
 studied by Faraday (1857) (24) ■ If the process takes 
 place near a liquid surface* some of the particles can 
 be collected in the liquid, forming a sol with it (25). 
 If the metal is immersed in a liquid while the current 
 is sent through it, a layer ot vapour from the liquid 
 around it will instantly be produced, so that the 
 condensation of the metal vapour will take place 
 partly in gas and partly on the liquid surface bound- 
 ing the liquid vapour region. The condensate is 
 collected in the liquid (25) . Part of the melted metal 
 is probably disintegrated and dispersed by the sudden 
 movement of the liquid and the vapours from it. 
 The volatilization may partly be caused by the Joule 
 heat and partly by small electric arcs formed between 
 parts of the wire at the moment of its melting. If the 
 volatilization of the wire is carried out in a transverse 
 magnetic field, the degree of dispersion of the colloid 
 obtained seems to be a little higher than without 
 such a field. Probably this is due to the very short 
 duration of the arc in a transverse magnetic field. 
 
42 THE FORMATION OF COLLOIDS. 
 
 Hence the metal vapour will be readily spread and 
 condensed before it has time to form large particles. 
 Hitherto we have dealt with condensation 
 processes occurring when vapours come into contact 
 with gaseous, liquid or solid material at a lower 
 temperature. In doing so, moreover, we have 
 confined ourselves almost exclusively to the con- 
 densation of metal vapour. The experiences gained 
 there may be applied to other cases, viz., the forma- 
 tion of colloid by condensation of- non-metaUic 
 vapours. Such processes are, however, only very 
 little studied as yet, and are, therefore, at present of 
 little interest for our description of the formation of 
 colloids. (Ehrenhaft prepared disperse systems in 
 gases by condensation of sulphur vapour, and 
 Westgren obtained a sulphur hydrosol by injecting 
 sulphur vapour into water.) 
 
 Another very important case of formation of 
 disperse systems by condensation in a gas is the 
 condensation of supersaturated vapour by adiabatie 
 expansion. In this case the whole volume of the 
 vapour is cooled all through and at once, in contra- 
 distinction to the cases treated above, where the 
 vapour was cooled along the surfaces of contact between 
 the hot vapour and the cold material that it en- 
 countered. The latter class of processes may, 
 therefore, suitably be called surface condensation 
 processes, the former volume condensation processes. 
 
 In the cases of adiabatie condensation studied so 
 far, one is dealing almost exclusively with conden- 
 sation of vapours to liquid drops. The only case 
 where the adiabatie expansion led to a disperse 
 system with solid disperse phase is E. Meyer's 
 experiment with regard to the condensation of water 
 vapour in air under o° (26). Disperse systems 
 formed by adiabatie expansions are rather unstable, 
 as is generally the case with disperse systems in 
 
ADIABATIC EXPANSIONS, 43 
 
 gases. These investigations have, however, given 
 very important and detailed information as to the 
 mechanism of the condensation process itself , and are, 
 accordingly, of great interest for the chemistry of 
 colloids. 
 
 We shall first consider the phenomena to be 
 observed in connection with the condensation of 
 water vapour in air (27). If a volume v^, saturated 
 with water vapour at, e.g., + 20°, is exposed to a 
 series of adiabatic expansions, the cooling will 
 
 increase as the ratio of volumes — increases, Wg being 
 
 the volume of the vapour at the end of the expansion. 
 Hence the supersaturation of the water vapour 
 immediately after the expansion will increase with 
 
 2 
 
 If condensation takes place it will go on until so 
 
 much water is condensed that the heat set free by 
 this process makes the remaining water vapour just 
 saturated at the temperature attained. The diagram 
 (Fig. 12) shows the maximal cooling ^2. the equi- 
 librium temperature tj^, and the supersaturation S 
 
 plotted against the ratio — as abscissa. Here S = ~ 
 
 Vi P2 
 
 when pi deiiotes grammes of water per cubic centimetre 
 
 immediately after expansion and pj the equilibrium 
 
 concentration of water at the temperature ij- The 
 
 initial pressures are to be chosen so that after every 
 
 expansion there shall be normal pressure (760 mm. 
 
 mercury) in the gas. In Fig. 12 are also drawn the 
 
 curve for the mass of water condensed per cubic 
 
 centimetre {q-,^ and the curve for the percentage .of 
 
 water condensed (= — . 100). 
 Pi 
 We will now consider more closely the conden- 
 sation at various degrees of supersaturation. In 
 a pure mixture of air and water vapour there is 
 
44 THE FORMATION OF COLLOIDS. 
 
 no condensation at supersaturations under about 
 
 4-2 f^ < i-asV If dust particles or products from 
 
 chemical processes be present (e.?., smoke particles 
 or particles formed by the photochemical action of 
 
 -30-- 
 
 
 -25 
 
 
 -20°- 
 
 
 
 -10.10 
 
 
 / / 
 
 -20 
 
 
 
 tz/ / 
 
 
 
 -10°- 
 
 
 
 
 
 
 .15 
 
 
 
 / S'^'^^^/a^^^^ 
 
 
 
 0°- 
 
 
 -10 
 
 -6 
 
 •5.10 
 
 
 //yC^\/ 
 
 
 . 
 
 + 10°- 
 
 
 
 
 
 
 -5 
 
 . 
 
 h^ 
 
 
 s 
 
 q 
 
 r T 
 
 t20° 
 
 ^ '" ^1 1 1 1 1 
 
 J 
 
 "r 
 
 1.1 
 
 1.2 
 
 1.3 
 
 ).<. 
 
 100 p.c. 
 
 50 p.c. 
 
 T 
 
 Fig. 12. — ^Variation of maximal cooling temperature t^, equi- 
 librium temperature t-^, supersaturation 5, ma?s of water 
 condensed per cubic centimetre q and percentage of water 
 condensed q\fi with ratio of volumes 1/2/^1 ii the case of 
 condensation of water vapour by adiabatic expansion. 
 
 a strong illumination), water is condensed on these 
 nuclei even at low supersaturation. Hence the 
 degree of dispersion in this case depends entirely on 
 the number of such nuclei present in the gas. In 
 the pure air and water vapour mixture condensation 
 begins at 5 = 4'2, and, according to Wilson's investi- 
 gations, it is the negative ions in the gas that serve as 
 
WILSON'S EXPERIMENTS. 45 
 
 condensation centres. This form of condensation ap- 
 plies to the interval S = 4'2to 5 ( - = 1-25 to 1-28 j. 
 
 Within the interval S = 5-8 to 6-8 (^ = 1-31 to 1-34) 
 
 the positive ions act as condensation centres. As 
 the " natural " ionization in air is very small, being 
 caused in the main by the radioactivity of the earth 
 and atmosphere, there are, under normal conditions, 
 only a few centres in pure air that are able to act 
 as condensation nuclei at supersaturations between 
 4-2 and 6-8. Hence the degree .of dispersion of the 
 system formed is very small. Accordingly, the 
 condensate of that interval is called rain-like. The 
 size of the drops varies from about 0-2 mm. in radius 
 at 5 = 4*2 to about 0-02 mm. at S = 6-8, but the 
 condensate always contains drops of rather different 
 size. If the ionization is increased, for instance, by 
 exposing the gas to X-rays or radium rays, the 
 number of centres and, consequently, the degree of 
 dispersion of the condensate are increased. In this 
 way it is possible to produce within the interval 
 5 = 4-2 to 6-8 dense fogs with small drops of rather 
 uniform size. C. T. R. Wilson very ingeniously 
 employed this process to make the paths of single 
 a- and /3-particles visible (28). Passing on to still 
 higher degrees of supersaturation, we find a very 
 rapid increase in the degree of dispersion of the 
 condensate at S = 8. The condensate is now fog- 
 like. When the supersaturation 12 is reached, 
 the number of drops formed per cubic centimetre 
 does not increase any longer, but remains constant. 
 As the mass of water condensed still increases, the 
 degree of dispersion begins to decrease slowly from 
 the value 5 = 12. The radius of the drops at S = 12 
 is about 0-6 ju, and is fairly uniform through the whole 
 system. The diagram (Fig. 13) shows the mass of 
 condensate per cubic centimetre (5') as a function of 
 
46 THE FORMATION OF COLLOIDS. 
 
 the supersaturation (S). In the diagram are also 
 plott-ed a series of measurements of the number of 
 drops per cubic centimetre (A'^) at various degrees of 
 supersaturation (29) . This latter curve illustrates 
 very well the change in the degree of dispersion of the 
 condensate with increasing supersaturation. The 
 values of iV certainly refer to condensates of some- 
 
 10.10- r o -100.0D0 
 
 50.00D 
 
 Fig. 13. — Variation of mass of condensate q and number of 
 drops N per cubic centimetre with supersaturation S in 
 the case of condensation of water vapour by adiabatic 
 expansion. 
 
 what different total masses, but, as the q curve shows, 
 this liiass only varies very little within the interval 
 of observation (5 = 4-2 to 19-3). As to the nature 
 of these very numerous nuclei which begin to act at 
 S = 8, and of which the number for water vapour in 
 air is about 100,000 per cubic centimetre, it has been 
 stated that they are electrically neutral and are very 
 probably derived from the water vapour and ^not 
 from the air. Experiments with other gases, viz., 
 
NATURE OF NUCLEI. 47 
 
 hydrogen and carbon dioxide, instead of air have 
 given quite the same values for N within the interval 
 S = 8 to 15. Within the region 5 = 4-2 to 6-8, ^V is 
 to some degree dependent on the nature of the gas. 
 This phenomenon is explained by the fact that the 
 natural ionization varies in different cases. Accord- 
 ing to P. Lenard the condensation centres in the 
 interval 5 = 8 to 15 are complex water molecules (30) . 
 
 The condensation by adiabatic expansion of 
 vapours other than water has been studied especially 
 by Andren, and it has been stated that on the whole 
 the process takes place in the same way (31). The 
 constants of the process are, however, different and 
 characteristic for each substance. Thus the values 
 of 5 where highly disperse condensates just begin to 
 be formed for methyl alcohol, ethyl alcohol, propyl 
 alcohol and benzene are 4-0, 2-5, 3-5 and ii-o respec- 
 tively. The maximal number of drops per cubic 
 centimetre is 160,000, 340,000, 360,000 and 190,000 
 respectively. 
 
 It is obvious that these results are of great impor- 
 tance for our views as to other kinds of condensation 
 processes in gases. The existence of certain critical 
 degrees of supersaturation characteristic of the 
 various substances, above which the degree of dis- 
 persion of the formed system Itecomes high, is pro- 
 bably not confined to condensation caused by 
 adiabatic expansion. It is very probably a phe- 
 nomenon common to most forms of condensation in 
 gases. 
 
 The supersaturation necessary for the forijiation of 
 a disperse system in a gas may also be produced by 
 the formation of a new suustance by chemical reac- 
 tion. The new substance must, of course, be present 
 at the moment of formation in a molecular state, 
 i.e., as gas in a gas, and, provided the concentration 
 be great enough, this represents a supersaturation. 
 
48 THE FORMATION OF COLLOIDS. 
 
 Two cases must be taken into consideration. Either 
 the condensation takes place simultaneously in the 
 whole volume of the gas (volume condensation), or 
 it takes place along certain surfaces within the gas 
 (surface condensation). We have examples of the 
 former case in the formation of disperse systems by 
 photochemical reactions, of the latter in the pro- 
 duction of smoke by the mixing together of certain 
 gases, e.g., HCl + NH^. When gases which react on 
 one another to form a solid or liquid phase are 
 mixed, it may of course be possible for the con- 
 densation process to take place in a homogeneous 
 system and, consequently, simultaneously throughout 
 the whole gas volume. But such a case assumes that 
 the mixing process is carried out very quickly and 
 that the supersaturation lasts long enough for the 
 system to become quite homogeneous — a condition 
 very rarely realized in practice. 
 
 The formation of disperse systems in gases by such 
 processes has so far been studied only very slightly. 
 Tyndall made some interesting experiments on the 
 formation of highly disperse fogs by photochemical 
 reactions (32) . A glass tube about | metre in length 
 and 6 to 8 cm. in diameter was fiUed with a photo- 
 chemically sensitive gas mixture that could give rise 
 to some non-volatile reaction product. Air with a 
 slight trace of amyl nitrite gas and a little hydro- 
 chloric gas proved to be very suitable. An intense 
 beam of light from an arc lamp was then thrown 
 into the tube along its axis. In the glass tube, at 
 first optically empty, there arose after a minute or so 
 a blue fog, made visible by the illuminating beam of 
 light, and this fog gradually became more dense and 
 at the same time more and more white in colour. At 
 last the tube contained a thick white fog. The Ught 
 diffused by the fog was at first totally polarized in a 
 plane perpendicular to the axis of the tube, but as 
 the fog became more and more white the non- 
 
DISPERSION PROCESSES. 49 
 
 polarized part became more prominent. Obviously 
 we are dealing here with a disperse system that 
 passes from a high degree of dispersion to a low 
 one. 
 
 In general, the degree of dispersion seems to be 
 higher in volume condensation than in surface con- 
 densation. Thus the Tyndall fogs are relatively 
 highly disperse, while ammonium chloride smoke 
 formed by rnixing ammonia gas and hydrochloric gas 
 is less disperse. The latter is thought to be a surface 
 condensation, the former a volume condensation 
 process. 
 
 The smoke foriiied by the incomplete combustion 
 of bituminous coals, i.e., hydrocarbons of high carbon 
 percentage, may be regarded as the production of a 
 disperse system in a gaseous medium by a condensa- 
 tion process preceded by a chemical dissociation. The 
 particles act as nuclei for the condensation of water 
 vapour even at low degrees of supersaturation 
 (c/. p. 44) , and thus give rise to those dirty fogs from 
 which great towns in damp climates, e.g., London, 
 are suffering. 
 
 Real dispersion processes leading to the formation 
 of disperse systems in gases are very incompletely 
 known. The degree of dispersion is on the whole 
 rather low. In this category we have the mechanical 
 pulverization of a liquid by means of a gas jet 
 (sprayer, Schoop's metallization process (33) ) , and 
 by means of a hard surface against which the liquid 
 is thrown (rain beating against a rock), grinding 
 of solid bodies (starch dust in mills, ore dust in 
 bucking-mills). Of much greater interest is the 
 formation and disintegration of that peculiar form 
 of disperse systems which is represented by bubbles 
 and foam. If a gas is set free in a liquid to a concen- 
 tration above saturation, or if it is injected into the 
 liquid from outside, it will of course rise to the surface 
 
 F.C 4 
 
50 THE FORMATION OF COLLOIDS. 
 
 of the liquid in the form of gas bubbles. As a 
 bubble rises there will ultimately be formed a very 
 thin layer of liquid between the bubble and the 
 atmosphere above the liquid surface. The specific 
 surface of this thin liquid lamella is very great, i.e., 
 the degree of dispersion is very high. Hence the 
 lamella is to be regarded as a dispersed system. 
 When, in consequence of the continuous rising of the 
 bubble, the thickness of the lamella has passed a 
 certain limit, differing in different liquids and gases, 
 the lamella bursts. The bursting is accompanied by 
 a reduction of the specific surface and the fragments 
 are thrown explosively into the atmosphere, forming 
 a fine spray. Thus a new disperse system with a 
 smaller specific surface, i.e., a smaller degree of dis- 
 persion than the lamella, is formed hy condensation. 
 Here we have the analogy to the condensation pro- 
 cess in vacuo mentioned above, when a continuous 
 metal film formed on a glass wall disintegrates into 
 ultra-microscopic particles. Provided the liquid is 
 not too volatile, its surface tension low, and its 
 viscosity great, as is the case in soap solutions, 
 rather stable lamellae can be produced (soap bubbles 
 and soap foam) . If the liquid in the lamellte solidifies 
 we get a solid foam of great stability (pumice-stone, 
 certain forms of blast furnace slag) . 
 
 Thus the pulverization of liquids by evolution of 
 gas in the interior of the liquid is a combined disper- 
 sion and condensation process. In practice it has 
 been applied to produce coloured flames. Little 
 gas bubbles are formed in a salt solution, for 
 instance by means of electrolysis. They rise to 
 the surface and then burst, forming small drops 
 of salt solution, which are carried away to the 
 flame by the gas current. The very inconvenient 
 production of sulphuric acid spray that occurs when 
 lead accumulators are charged is due to a similar 
 process. 
 
DISPERSION OF SOLIDS. 51 
 
 Gas production in a solid body may also give rise 
 to a disperse system (34). Thus oxalates and car- 
 bonates, when heated, are often disintegrated into a 
 fine dust, which is thrown out into the surrounding 
 atmosphere by the carbon dioxide set free within 
 them. In the same way dry potassium perman- 
 ganate, when heated, sends out dust of manganese 
 oxides dispersed by the production of oxygen. 
 Volcanic ashes and fine volcanic dust are formed by 
 a similar process (35). Here it is water which, 
 having been dissolved in the silicate mineral under 
 high pressure, is suddenly volatilized when the stones 
 are ejected from the volcano into air at atmospheric 
 pressure, and explosively pulverizes the material (36). 
 This volcanic dust is of importance, partly because 
 of the catastrophes it has caused (the destruction of 
 Herculaneum and Pompeii, the burning of St. Pierre), 
 and partly because of the splendid sunsets and the 
 luminous night clouds to which it gives rise. With 
 regard to the mechanism of these disintegration pro- 
 cesses, it has not yet been ascertained whether thin 
 lamellae are formed as an intermediate product or 
 whether the material is broken up uniformly. 
 
 Bibliography. 
 
 1. Kohlschiitter and Ehlers, Zeitschr. Elektrochem., 
 
 18, 373 (1912). 
 
 2. Kohlschiitter and Noll, Zeitschr. Elektrochem., 
 18, 419 (1912). Kohlschiitter, Jahrb. Radioakt., 9, 
 
 355 (1912). 
 
 3. Nordlund, Quecksilberhydrosole, Diss., Upsala, 
 p. 22 (1918) ; Koll. -Zeitschr., 26, 121 (1920). 
 
 4. Kimura, Mem. Coll. Science and Engin., Kyoto, 
 5, 211 (1913). 
 
 5. Nordenson, Kollchem. Beih., 7, 106 (19.15). 
 
 6. Ehrenhaft, Sitzungsber. Akad. Wiss. Mathem.- 
 nat. Kl. Wien, 119, 8.30.(1910). 
 
52 THE FORMATION OF COLLOIDS. 
 
 7. Konstantinowsky, Sitzungsber. Akad. Wiss. 
 Math.-nat. Kl. Wien, 123, 1707 (1914). 
 
 8. Svedberg, Meddel. NobelinsL, Stockholm, 5, 
 No. 10, p. 3 (1919). 
 
 9. Borjeson, Researches on the Electric Synthesis 
 of Colloids, Diss., Upsala (1921). 
 
 10. Svedberg, Nova acta, Upsala, loc. cit., p. 45. 
 
 11. Bredig, Zeitschr. angew. Chem., 1898, p. 951 ; 
 Anorganische Fermente, Leipzig (1901) ; Zeitschr. 
 phys. Chem., 32, 127 (1900). Cf. Tichomiroff and 
 Lidow, Beihl. Annal. Phys., 8, 232 (1884). Threlfall, 
 Phil. Mag. (5), 38, 450 (1894). 
 
 12. Svedberg, Ber. Deutsch. chem. Ges., 39, 1705 
 (1906) ; Nova acta, Upsala, loc. cit., p. 14 ; Herstellung 
 kolloider Losungen, Dresden, p. 423 (1909) ; Meddel. 
 Nobelinst., loc. cit. Borjeson and Svedberg, Koll.- 
 Zeitschr., 25, 154 (1919). Nordlund, loc. cit. 
 Borjeson, loc. cit. 
 
 13. Degen, Beitr. Kenntnis koll. Metalle, Diss., 
 Greifswald (1903). 
 
 14. Burton, Phil. Mag. (6), 11, 425 (1906). 
 
 15. Borjeson, loc. cit. 
 
 16. Svedberg, Nova acta, Upsala, loc. cit., p. 29. 
 
 17. Borjeson and Svedberg, loc. cit. 
 
 18. Svedberg, Physikal. Zeitschr., 15, 361 (1914). 
 
 19. Borjeson, loc. cit. 
 
 20. Benedicks, Kolloidchem. Beih., 4, 234 (1913). 
 
 21. Borjeson, loc. cit. 
 
 22. Svedberg, Nova acta, loc. cit., p. 68. 
 
 23. van Marum, Verhandel. Teyler's tweede 
 Genootsch., Haarlem, 4 (1787). 
 
 24. Faraday, Phil. Trans., 147, 152 (1857). 
 
 25. Svedberg, Koll. -Zeitschr., 24, i (1919). 
 
 26. E. Meyer, Mitteil. phys. Ges., Ziirich, No. 18 
 (1916). 
 
 27. C. T. R. Wilson, Phil. Trans., loc. cit. 
 
 28. C. T. R. Wilson, Proc. Roy. Soc, A8s, 285 
 (1911) ; A87, 277 (1912). 
 
BIBLIOGRAPHY. 53 
 
 29. Andren, Zdhlung und Messung der komplexen 
 Molekiile einiger Ddmpfe nach einer neuen Konden- 
 sationsmethode, Diss., Upsala, p. 28 (1918). 
 
 30. Lenard, Sitzungsber. Akad., Heidelberg, 1914, 
 No. 29. 
 
 31. Andren, loc. cit., pp. 53, 59, 62, 67, 91. 
 
 32. Tyndall, Phil. Mag. (4), 37, 384 (1869). 
 
 33. Kasperowicz and Schoop, Das Elektro-Metall- 
 spritzverfahren, Halle (1920). 
 
 34. Kohlschutter, Erscheinungsformen der Materie. 
 
 P- 143- 
 
 35. Zsigmondy, Kolloidchemie, p. 33 (1918). 
 
 36. Barus, Amer. Journ. Science (4), 9, 161 (1900). 
 Lacroix, La montagne Pelee et ses eruptions, Paris 
 (1904). 
 
FORMATION OF DISPERSE SYSTEMS IN 
 LIQUIDS AND IN SOLIDS. 
 
 Condensation Processes. 
 
 The supersaturation necessary for condensation 
 in a liquid medium may be caused by cooling, by 
 adding a reagent that lowers the solubility of some 
 substance dissolved in the medium, or by producing 
 the material of the disperse phase {e.g., by a chemical 
 reaction) in a concentration high in relation to its 
 solubility. 
 
 The formation of disperse systems according to the 
 first principle has not yet been subjected to exhaus- 
 tive investigation. The cooling cannot be brought 
 about, as in the case of the adiabatic expansion 
 of a gas, simultaneously in the whole volume of the 
 liquid, but must be conducted through the liquid. 
 The condensation therefore becomes to some extent 
 a surface condensation, but in practice it may very 
 nearly approach volume condensation, provided the 
 cooling be conducted rapidly enough through the 
 liquid. Some examples may be mentioned. The 
 cooling with ice of a solution of benzene in water 
 saturated at boiling heat gives a benzene colloid, i.e., 
 a benzene-hydrosol. By placing an alcoholic solution 
 of sulphur in contact with liquid air we get a sulphur- 
 alcosol (i) . A solution of water in pentane under the 
 same conditions gives an ice-pentanosol (2), etc. 
 ' When such a "cryosol" is slowly warmed up, the 
 condensation process at first proceeds by lowering the 
 degree of dispersion of the system, but as the solu- 
 bility of the disperse phase rises, a point is reached 
 
CONDENSATION BY COOLING. 55 
 
 where a kind of dispersion process sets in, the size 
 of the particles, i.e., the degree of dispersion of the 
 system, being reduced by re-solution of the particles. 
 P. P. von Weimarn has observed the following phe- 
 nomenon. If a test tube containing some cubic 
 centimetres of an alcoholic solution of sulphur is 
 dipped in liquid air, the solution freezes to a clear 
 glass, which no doubt contains the sulphur partly 
 in a supersaturated molecular form and partly in a 
 highly disperse colloid form. If the test tube is then 
 taken out of the liquid air and slowly warmed, there 
 appears at first a slight blue opalescence, which soon 
 becomes stronger and more whitish. As the tem- 
 perature rises still more the opalescence diminishes, 
 turns blue and pale as at the beginning, and at last 
 quite vanishes. The slight bluish opalescence at the 
 .beginning corresponds to the small particles formed 
 by condensation, the white colour is due to the 
 particles growing greater and greater, and the 
 returning of the blue opalescence means the diminish- 
 ing of their size by re-solution (dispersion), which 
 ultimately leads to their returning entirely to the 
 molecular state. A quantitative investigation of 
 the various degrees of dispersion obtained bj^ cooling 
 such systems to various temperatures would be of 
 great interest. If the experiment is arranged so as 
 to cause the coohng to take place very rapidly and 
 uniformly throughout the whole system, an analogy 
 would thus be estabHshed with the investigations in 
 gases of the variation of the degree of dispersion 
 with the degree of supersaturation. 
 
 Condensation by means of a substance that lowers 
 the solubility of the material of the future disperse 
 phase is also a simple form of condensation that might 
 be worth studying more closely. If a solution of 
 palmitic acid is injected into water the solubility of the 
 acid is reduced, and consequently a supersaturation 
 is created that leads to precipitation of high disperse 
 
56 THE FORMATION OF COLLOIDS. 
 
 palmitic acid (3). The degree of dispersion depends 
 on the magnitude of the supersaturation and on the 
 speed with which it is effected, i.e., on the rate of 
 supersaturation. P. P. von Weimarn in an analogous 
 way prepared disperse systems of sulphur, phos- 
 phorus and selenium (4). 
 
 It is probable that we must regard the electrolytic 
 pulverizing of metals, studied especially by Haber (5), 
 as a peculiar case of lowering of solubility. If an 
 alkaline solution is electrolysed with a lead cathode, 
 there occurs a sort of pulverizing of the cathode, 
 brown clouds of metallic lead emanating from it. In 
 an acid solution a similar but less marked pheno- 
 menon is to be observed. In the former case Haber 
 has shown that an alloy is formed between the 
 cathode metal and the alkali metal set free by the 
 current, and this alloy is then decomposed by the 
 water. The solvent of the lead solution, i.e., the 
 alkali metal, is thus removed and the lead is left 
 behind in the water, in which it is almost insoluble, 
 as a disperse system. Probably the phenomenon 
 consists in a condensation of lead molecules to colloid 
 particles and lead lamellEe, the latter disintegrating 
 into particles as the condensation goes on. In the 
 case of electrolysis in acid solution the hydrogen very 
 likely plays the part of the alkali metal. We may 
 perhaps equally well assume that at first a hydride 
 is formed and that this compound is decomposed by 
 the water. If such be the case this phenomenon 
 should come under the heading " Dissociation " (see 
 p; 90). 
 
 A process quite analogous to the cathode-pul- 
 verization in alkaline solutions takes place, as Haber 
 has proved, when water is allowed to act upon certain 
 alloys of alkali metals and various other metals. The 
 phenomenon can easily be observed if fragments of a 
 sodium-lead alloy are thrown into water. 
 
 The most important and as yet most exhaustively 
 
CHEMICAL REACTIONS. 57 
 
 studied processes giving rise to disperse systems are 
 those in which the material of the disperse phase is 
 produced in high concentration. Most commonly 
 this is effected by means of a chemical reaction, but 
 other forms, too, are known. Thus colloids of radio- 
 active substances are often formed by the element 
 itself being produced within the system. 
 
 Here also it may be useful to bear in mind the two 
 cases of condensation, viz., volume condensation and 
 surface condensation. An example of the former case 
 in its pure form is the formation of colloid sulphur in 
 an alcoholic solution by means of ultra-violet illumina- 
 tion (transformation into another allotropic modifica- 
 tion). All condensation by photo-reaction is more or 
 less a volume condensation. When solutions that 
 react with each other are mixed, in most cases sur- 
 face condensation at the contact surface between the 
 solutions takes place. If the condensation or the 
 reaction proceed slowly enough, volume condensa- 
 tion may occur even in this case. Pure volume 
 condensation most probably gives rise to systems of 
 a more homogeneous degree of dispersion than 
 mixed surface and volume condensation. The pure 
 surface condensation very likely produces systems 
 with rather unequal particles. The degree of super- 
 saturation at different points in the limit layer 
 between the solutions will depend on the diffusion, 
 i.e., on the time and on the distance from the first 
 contact surface. 
 
 As regards the classification of the various forms 
 of condensation processes arising from chemical 
 reactions, we may, according to the kind of reaction, 
 put them under one of the following headings : — 
 
 Reduction. 
 
 Oxidation. 
 
 Dissociation. 
 
 Double decomposition. 
 Perhaps it would be more rational to choose for 
 
58 THE FORMATION OF COLLOIDS. 
 
 the purpose of classification cliaracteristic features 
 from the condensation process itself and not from the 
 chemical reaction that precedes the condensation. 
 Such a form of classification we have in the two 
 groups : condensation with spontaneous formation 
 of niiclei ; condensation with addition of nuclei. The 
 degree of supersaturation might also be taken as a 
 basis of classification. 
 
 Now in most cases we know only very little of the 
 mechanism of the condensation process. Thus it 
 appears that for the present it will be best to adhere 
 to the classification according to the chemical 
 reaction, especially as a close study of the proper 
 reactions and the condensation processes accom- 
 panying them ma}' very likely be the only way to 
 arrive at the general laws of condensation. 
 
 As stated above, we must distinguish between the 
 primary, the secondary and the tertiary degree of 
 dispersion. The primary degree of dispersion being 
 commonly the highest, it is obvious that the methods 
 of preparation in most cases aim at suppressing those 
 processes which lead to secondary or tertiary degree 
 of dispersion. The chief methods of attaining this 
 are : — 
 
 1. Very low solubility of the disperse phase in the 
 
 dispersion medium. 
 
 2. Suitable (not too high) ion concentrations. 
 
 3. Low concentration of the disperse phase. 
 
 4. Low temperature. 
 
 5. The presence of a protective colloid. 
 Conditions such as those just mentioned may, 
 
 however, in some cases influence the condensation 
 process in such a way as to form a less high primary 
 degree of dispersion. We then get the optimum 
 conditions in the form of a compromise. 
 
 In cases where these conditions are not fulfilled 
 and coagulation therefore takes place, the primary 
 structure may often be regained by — 
 
PURIFICATION. 59 
 
 1. Lowering the concentration of the ions present 
 
 (washing) . 
 
 2. Replacing unsuitable ions by suitable ones 
 
 (washing and peptization). 
 
 The two latter processes were previously considered 
 as dispersion processes, but recent investigations 
 have shown that they are in general only reversible 
 coagulations. The degree of dispersion is not altered, 
 only the distribution of the particles in the dispersion 
 medium is changed. 
 
 The removal of electrolytes from colloid solutions 
 may be carried out in different ways. There is the 
 simple washing of the coagulum on a filter or by 
 sedimentation and decanting, the filtration through 
 collodion or gelatin membranes, which permit the 
 crystalloids to pass freely but retain the colloids 
 (ultra-filtration) (6), Graham's classical dialysis (7), 
 and lastly electrolysis (8). The methods of ultra- 
 filtration and dialysis have of late been considerably 
 improved and now constitute a very important 
 procedure in the preparation of colloids (g). 
 
 There is a large and very important class of disperse 
 systems, most probably formed by condensation pro- 
 cesses, concerning the mechanism of formation of 
 which we are still almost ignorant, viz., the organic 
 . colloids, especially those found in plants and animals. 
 Owing to the investigations of E. Fischer (10), we 
 know, for instance, that the proteins are built up of 
 amino-acids, but the conditions under which they give 
 rise to those highly disperse colloids that play such 
 a prominent part in the organic world have not yet 
 been elucidated. Some of them may be completely 
 dried and yet redissolve to hj'drosols when brought 
 in contact with water. Many salts of organic sub- 
 stances form colloid solutions in water but crystalline 
 solutions in organic solvents, e.g., alcohol, benzene. 
 In some cases the process may be regarded as a 
 hydrolysis. The formation of hydrosols of many dye- 
 
6o THE FORMATION OF COLLOIDS. 
 
 stuffs comes under this heading (loa). On the other 
 hand, there are dyestuffs that form hydrosols when 
 dissolved in water which are certainly not hydrolysed 
 to any noticeable degree. The alkali salts of the 
 higher fatty acids seem to be almost completely 
 hydrolysed in concentrated aqueous solutions, while 
 they are not so much hydrolysed in dilute solutions, 
 quite contrary to the experience with inorganic salts. 
 It cannot be a hydrolysis equilibrium of the ordinary 
 kind, as such solutions show no raising of the. boiling 
 point, notwithstanding their content of " free " 
 alkali (ii). 
 
 The swelling and redissolution of, e.g., gelatin, 
 and the colloid dissolution of dyestufis might be 
 considered as a kind of dispersion process. The 
 structure or degree of dispersion brought about 
 were, however, probably latent in the substance 
 before it came in contact with the solvent, and the 
 process is therefore probably to be looked upon as a 
 change of state in the system and not as a formation 
 of a new disperse system. 
 
 Reduction. 
 The best known reduction process is the reduction 
 of chlorauric acid (HAUCI4) to gold. Almost every 
 conceivable reducing agent has been studied, viz., 
 hydrogen, hydrogen peroxide, hydrogen sulphide, 
 carbon monoxide, carbon disulphide, nitric oxide, 
 phosphorus, phosphorus tetroxide, hypophosphoric 
 acid, sulphur dioxide, sodium thiosulphate, sodium 
 bisulphite, ferrous sulphate, tin, stannous chloride, 
 acetylene, terpenes, alcohols, glycerine, aldehydes, 
 acrolein, oxalic acid and oxalates, tartaric acid, 
 sugars, starches, phenols, hydroxide acids, hydro- 
 chinones, hydrazines, hydroxylamines, protalbic 
 acid, electric sparks (formation of nitric oxide), 
 a-, ^- and y-rays, etc., etc. (12). In spite of these 
 extensive investigations we know but very little of 
 
GOLD REDUCTION. 6i 
 
 the mechanism of the reducing process itself. It 
 still remains to be settled in which stages it takes 
 place, what are the intermediate products, whether it 
 is independent of the condensation that follows, etc. 
 
 As to the reducing process itself it seems to proceed 
 differently in a pure (or acid) and in an alkaline solu- 
 tion. The simplest and most easily studied case is 
 reduction with hydrogen peroxide in a pure solution 
 of HAUCI4 : 
 
 2HAUCI4 + 3H2O2 = 2Au + 8HC1 + 3O2 (13). 
 
 Here neither the reducing agent itself nor the pro- 
 ducts formed, except the hydrochloric acid, are 
 electrol5rtes. Hence we may easily follow the course 
 of the reduction by measuring the electric con- 
 ductivity of the solution. This quantity will rise 
 during the reduction process from the value for 
 HAuClj at the dilution in question to the value for 
 4HCI at the same dilution. Measurements carried 
 out by H. Nordenson (14) and by the writer have 
 brought out the following facts. 
 
 First case : No nuclei are present at the beginning 
 of the reduction. When the reducing agent is added 
 there is a sudden increase in conductivity of about 
 30 per cent. (A — B, Fig. 14). At the same time the 
 yellow colour of the solution decreases considerably. 
 Then the conductivity rises for some time very slowly 
 or remains almost constant, and so does the colour 
 too (B — C in Fig. 14) . After this period there follows 
 one of rapid rise in the conductivity (C — D in Fig. 14), 
 accompanied by a sudden red coloration of the liquid. 
 When examined in the ultra-microscope the system 
 at this period is seen to contain a great number of 
 particles, none of which were present during the 
 stages A — B and B — C. When the intensity of 
 colour no longer increases the conductivity has 
 likewise attained its.limiting value for 4HCI. 
 
 Second case : Nuclei are present at the beginning 
 of the reduction. From the addition of the reducing 
 
62 THE FORMATION OF COLLOIDS. 
 
 agent to the end of the reduction the conductivity as 
 well as the colour intensity now increases almost pro- 
 portional to time (Fig. 14, A — D'). If nuclei are 
 added to the reduction mixture during the stage 
 B — C we get an intermediate case represented by 
 the curve A— B— C"— D". 
 
 As to the explanation of these phenomena the 
 following hypothesis suggests itself. The part A — B 
 of the curve denotes reduction of about J of the 
 
 conductivity of 4 H CI 
 
 3 
 o 
 
 > time 
 
 Fig. 14. — Variation of conductivity with time when reducing 
 chlorauric acid with hydrogen peroxide. 
 
 HAUCI4 to Au. Hereby a highly supersaturated 
 solution of gold is formed and the progress of the 
 reduction is stopped according to the law of mass- 
 action. The solubilitj? of gold in water being of 
 course very low, viz., less than io~^, the degree of 
 supersaturation will prove to be quite enormous, 
 vi?;., 10' at a concentration of lo"* normal. Most 
 probably this enormous supersaturation gives rise 
 to a disperse system of an exceedingly high degree 
 of dispersion. As the condensation proceeds, the 
 
ACTION OF NUCLEI. 63 
 
 small primary particles being aggregated or coagu- 
 lated to secondary particles, the reduction goes on 
 slowly along the part B — C of the curve. When the 
 size of the greater number of particles has reached a 
 certain value they begin to act as centres of condensa- 
 tion for molecular gold (at C in Fig. 14) , and then the 
 process is finished along the straight line C — D. The 
 addition of nuclei at B or at C" will only have the 
 effect of displacing the part C — D to B — D' or to 
 C" — D". In support of this hypothesis we may 
 mention the observation made by Zsigmondy and by 
 others that gold particles do not act as condensation 
 nuclei unless their diameter has reached a certain 
 value, viz., about 2 /x/a (15). 
 
 The reduction of HAUCI4 by hydrogen peroxide 
 with the addition of nuclei has in recent years become 
 of great value to the physics and chemistry of 
 coUoids (16). Two important applications are to be 
 noted. Firstly, it is possible to prepare series of gold 
 colloids with various degrees of dispersion by adding 
 various amounts of gold nuclei to the reduction 
 mixture. Secondly, the degree of dispersion of gold 
 colloids with very small (amicroscopic) particles may 
 be determined by immersing the particles in the 
 reduction mixture and allowing them to grow large 
 enough for ultra-microscopic investigation. Recently 
 Borjeson (17) has shown that a great many other 
 kinds of small particles act as nuclei in the reduction 
 of gold by hydrogen peroxide, viz., metals such as 
 platinum, silver, mercury, tin, copper, bismuth, 
 cadmium, zinc, iron, aluminium, sulphides of 
 antimony and arsenic, sulphur. Oxides of tin, iron 
 and silicon as well as dyes and organic colloids, e.g., 
 gelatin and gum arable, do not act as nuclei. WTien 
 deaUng with these " nuclei methods," whether it is a 
 question of preparing a gold colloid of a certain 
 degree of dispersion or of determining the degree of 
 dispersion of a coUoid — the particles of which are too 
 
64 THE FORMATION OF COLLOIDS. 
 
 small to be measured with the ultra-microscope — the 
 fundamental condition is that there should be no 
 spontaneous production of nuclei in the reduction 
 mixture. According to Westgren the concentration 
 of the HAuClj should not be less than lo""' normal 
 and the number of nuclei added not less than 5-10"^ 
 per cubic centimetre, otherwise spontaneous produc- 
 tion of gold nuclei will take place before the reduc- 
 tion is completed. 
 
 That under suitable conditions no spontaneous 
 nuclei appear is demonstrated by the following 
 table (5) given by Westgren : — 
 
 Table 5. 
 
 Cubic centimetres gold 
 nuclei fluid (= K'. 
 
 Number of particles formed 
 per 10"^ cc. (= «). 
 
 n/K. 
 
 8 
 12 
 16 
 20 
 
 121 
 169 
 231 
 289 
 
 I5-I 
 14-1 
 14-4 
 14-5 
 
 Even in the case of nuclei of material other than 
 gold the process may be carried out in such a manner 
 that all the gold from the solution is deposited on 
 the nuclei. A table given by Borjeson shows this. 
 
 Table 6. 
 
 Nuclei : Cadmium, prepared by the damped high-fre- 
 quency alternating current arc in ethyl alcohol at — 70° C. 
 
 Volume Cd— sol in the reduction 
 
 experiment per 50 cc. reduction 
 
 mixture. 
 
 Radius of the nuclei as determined 
 
 by the reduction experiment and 
 
 the measurement of the degree of 
 
 dispersion of the gold sol formed. 
 
 0-2 cc. 
 O-I C.C. 
 0-05 C.C. 
 0025 C.C. 
 
 ii-4/ifi 
 10-7 iMfi 
 
 II -0 fifl 
 
 10-9 /i/i 
 
PURIFICATION. 65 
 
 The method has not yet been sufficiently worked 
 out and we may still expect to make much progress 
 by means of a careful examination. 
 
 In order to get rid of the main part of the HCl 
 formed by the reduction dialysis or filtration through 
 collodion membranes may be used. In consequence 
 of the great sensitiveness of gold hydrosols to small 
 electrolytic impurities partial coagulation will often 
 occur when they are dialysed. Gold hydrosols of 
 not too high a degree of dispersion may be purified 
 by repeated sedimentation and by replacing the dis- 
 persion medium by pure water. In this manner 
 Westgren was able to prepare gold hydrosols almost 
 free from HCl. 
 
 In those cases where alkali is added to the gold 
 solution before reduction gold hydroxide is. probably 
 formed first and then an alkali auric salt (18). The 
 reduction process wiU thus consist in precipitating 
 the gold from the solution of this salt. The reaction 
 between the chlorauric acid and the alkali may be 
 written in the following manner. 
 
 HAUCI4 + 4KOH = Au(0H)3 -h 4HCI + H2O 
 Au(0H)3 + KOH = KAuOa + 2H2O 
 or 
 
 HAuCli + 2K2CO3 + 2H2O = Au(0H)3 + 2CO2 + 
 
 4KCI 
 2Au(OH)3 + K2CO3 = 2KAUO2 + 3H2O -(- CO2. 
 
 If formaldehyde be added we have 
 
 2KAUO2 + 3HCHO + KOH = 2Au + 3HCOOK + 
 
 2H2O 
 or 
 
 2KAuO, + 3HCHO + K2CO3 = 2Au + 3HCOOK+ 
 KHCO3 + H2O. 
 
 This agrees with the statement of Vanino that 
 2 mol. Au .requires 11 mol. NaOH for reduction (19). 
 Naumoff has prepared KAuOg and has shown that 
 
 r.c. 5 
 
65 THE FORMATION OF COLLOIDS. 
 
 the solutions of this salt are reduced in the same way 
 as mixtures of HAUCI4 and alkali. 
 
 Even when we have to deal with " pure " solu- 
 tions of HAUCI4 there will always be present small 
 quantities of Au(0H)3, formed from the HAUCI4 by 
 hydrolysis. This decomposition of the gold solution 
 is accelerated by light (20) . At high degrees of dilu- 
 tion it may in time lead to a complete liberation of 
 all the hydrochloric acid. The Au(0H)3 being only 
 sparingly soluble in water, a colloid solution of 
 this gold compound is formed. The properties 
 of the gold colloid prepared by reduction of this 
 Au(OH)3-sol will of course depend on the state of the 
 latter. 
 
 The degree of dispersion of gold colloids formed by 
 reduction of HAUCI4 without the addition of con- 
 densation nuclei is, according to Zsigmondy (21), 
 dependent on two factors, viz, : 
 
 1. The spontaneous production of nuclei (s.p.n.). 
 
 2. The velocity of growth of the particles (v.g.p.). 
 This means applying to the formation of colloids 
 
 the view held by Tammann concerning crystallization 
 in fused substances, and it seems, at least in the 
 present state of research, to be of considerable value 
 for the classification of the phenomena observed in 
 the process of the formation of gold colloids by 
 reduction. 
 
 The greater the number of gold nuclei produced 
 the higher the degree of dispersion, and the sooner 
 the nuclei are produced in the course of the reduction 
 process the more uniform will be the size of the par- 
 ticles. Great velocity of growth will generally 
 depress the degree of dispersion of the system, as 
 there is not sufficient time for the production of 
 nuclei before the gold supply of the solution is 
 already exhausted. Of course this view holds good 
 only for the primary degree of dispersion. The 
 secondary degree of dispersion will, moreover, depend 
 
INFLUENCE ON NUCLEI. 67 
 
 on the presence of coagulating and protective sub- 
 stances in the solution. 
 
 Zsigmondy's school has devoted much time and 
 energy to finding out the influence of various sub- 
 stances added to the gold reduction mixture. It is 
 impossible, however, to formulate any general laws, 
 and the conclusions drawn by the various investi- 
 gators do not hold when the experimental conditions 
 are only slightly altered. 
 
 Hiege studied the reduction by formaldehyde in 
 alkaline solution and obtained the following 
 results (22). In pure water, freed with great care 
 from contaminations, the reduction took place so 
 rapidly that the formaldehyde could not even be 
 completely mixed with the gold solution before the 
 liquid was already coloured. The first particles 
 formed before the mixing is completed will act as con- 
 densation nuclei and so will those formed a little 
 later, etc. Hence the degree of dispersion becomes 
 very heterogeneous. The process is a " surface con- 
 densation." With regard to the influence of sub- 
 stances added to the gold solution before reduction 
 the following classes may be distinguished. 
 
 1. V.g.p. reduced. 
 
 Under this heading fall various colloids and high 
 molecular substances, e.g., gelatin, salts of protalbic 
 acid, soaps, oils and fats ; among inorganic sub- 
 stances KBr, KJ have the same action but less 
 pronounced. 
 
 The reduction is retarded whether gold particles 
 (nuclei) be added or not. The degree of dispersion 
 is in general increased by the substance added. 
 
 2. S.p.n. reduced. 
 
 Potassium ferro- and ferricyanide, ammonia and 
 its salts have this effect. 
 
 Reduction is retarded if nuclei are not added, but 
 is not retarded if nuclei are added. 
 
 3. S.p.n. accelerated. 
 
 5—2 
 
68 THE FORMATION OF COLLOIDS. 
 
 Potassium thiocyanate, potassium oxalate, sodium 
 citrate, congo red, benzopurpurine, and most 
 inorganic salts in very low concentrations. 
 
 The reduction of HAUCI4 in alkaline solution by 
 hydrazine chlorhydrate (23) was studied by the 
 
 > concentpalion of gelatin (per cenl) . 
 
 Fig. 15. — Variation of time required for the red colour to 
 appear when reducing chlorauric acid .in the presence 
 of gelatin of various concentration. 
 
 writer at an early date (24) . He found the appear- 
 ance of the red colour retarded by gelatin, the retar- 
 dation being almost proportional to the gelatin 
 concentration (Fig. 15). When gelatin was present 
 electrolytes were found to accelerate the appearance 
 of the red colour, MgCla being more effective than 
 KCl and BaClg more effective than MgClg. Non- 
 
ACTION OF LIGHT. 
 
 69 
 
 electrolytes, e.g., urea, alcohol, glycerine, milk-sugar, 
 pyridine, had no effect. ;: 
 
 The observations of Hiege and of the writer may be 
 interpreted as follows. The spontaneous production 
 of gold nuclei is a kind of coagulation and is therefore 
 accelerated by coagulating electrolytes in small 
 quantities. Higher concentrations drive the aggre- 
 gation of the particles too far. Weak reducing agents, 
 e.g., potassium thiocyanate, potassium citrate, Congo 
 red, etc., added before the reduction give rise to 
 small gold particles which act as nuclei. Protective 
 electrolytes, e.g., NH3, K4Fe(CN)e, K3Fe(CN)e, 
 retard the production of nuclei. Protective colloids 
 do not retard the s.p.n., their particles being too 
 coarse for these very small gold particles. The 
 growth of the particles is also a kind of coagulation. 
 It is retarded by protective colloids and accelerated 
 by coagulating electrol5rtes. 
 
 As to the nature of the protective action, it most 
 probably consists in an absorption of the gold 
 particles by the particles of the protective colloid. 
 
 Table 7. 
 
 Exposure (seconds). 
 
 Time for coJour to appear 
 (minutes). 
 
 Diameter (jllju). 
 
 
 
 10 
 
 90 
 
 I 
 
 5 
 
 70 
 
 2 
 
 5 
 
 50 
 
 3 
 
 4 
 
 40 
 
 5 
 
 J 
 
 30 
 
 10 
 
 ^ 
 
 20 
 
 20 
 
 Very short. 
 
 10—15 
 
 30 
 
 Very short. 
 
 10 
 
 60 
 
 Very short. 
 
 10 
 
 e.g., gelatin, albumin, gum arable, etc. This process 
 takes a considerable time (order of magnitude : 
 
70 THE FORMATION OF COLLOIDS. 
 
 some minutes) and thus it happens that the con- 
 densation is «rather far advanced before the pro- 
 tective colloid has had time to come into full action. 
 H. Nordenson has. shown that the reduction of 
 
 by means of HgOj is 
 If a dilute solution of 
 
 HAUCI4 to gold colloids 
 influenced by light (25). 
 
 100- 
 
 
 * lime of exposure (sec.) 
 
 Fig. 16. — ^Variation of degree of dispersion with time of expo- 
 sure in the case of reducing chlorauric acid with hydrogen 
 peroxide and illuminating the reduction mixture with 
 ultra-violet light. 
 
 HAUCI4 is mixed with some drops of HgOg and 
 immediately exposed to the ultra-violet rays of a 
 strong mercury vapour lamp the reduction is greatly 
 accelerated and the degree of dispersion of the 
 system formed is considerably higher than without 
 light. The size of the particles at first increases very 
 rapidly with the duration of exposure and then 
 becomes constant (Table 7 and Fig. 16). A glass 
 
FARADAY'S GOLD SOLS. 71 
 
 plate inserted between the lamp and the solution 
 reduces the action of the light quite markedly, the 
 diameter of the particles being in this case about 
 45 jjLfjL, instead of about 10 /x/^ without the glass plate. 
 Thus the light action is mainly due to the ultra- 
 violet rays.- 
 
 A satisfactory explanation of this very remarkable 
 phenomenon has not yet been found. According to 
 Nordenson's experiments pure solutions of HAUCI4 
 are only very slowly reduced by light and the dis- 
 perse system formed has a low degree of dispersion, 
 i.e., the particles are rather large. Hence the 
 phenomenon must be connected with condensation 
 only and is very probably due to rapid formation of 
 condensation nuclei by the hght. 
 
 A reduction method that gives gold colloids of a 
 very high degree of dispersion is Faraday's (26) 
 classical process, improved by Zsigmondy (27). 
 Here HAUCI4 in slightly alkaline solution is reduced 
 by phosphorus. In practice HAuCl^ of about i-io"* 
 normal with about 5'io~"* KgCOg is mixed with some 
 drops of an etheric solution of phosphorus. The 
 reduction takes place very slowly, a day often being 
 necessary for it to be completed. The diameter of 
 the particles is about 5 /i/x and can be reduced still 
 more by the addition of gelatin to the reduction 
 mixture. The mechanism of the process is rather 
 complicated. Some low oxide of phosphorus is 
 formed first, and this oxide acts as a reducing 
 agent. The cause of the very high degree of dis- 
 persion is not yet known. Perhaps it may be 
 attributed to the very rapid and abundant formation 
 of ion nuclei accompanying the oxidation of phos- 
 phorus. 
 
 As a rule the greater the dilution of the HAuClj 
 the higher the degree of dispersion of the gold colloid 
 (Fig. 17) (28). This law holds good for all sorts of 
 gold reduction processes. The fact seems to be 
 
72 THE FORMATION OF COLLOIDS. 
 
 opposed to the general rule stated above, viz., the 
 greater the supersaturation the higher the degree of 
 dispersion. The explanation might be that in high 
 dilutions the growth of the particles is much depressed. 
 Hence more nuclei can be formed before the supply of 
 gold is exhausted. Oden has suggested to the writer 
 
 400- 
 
 
 ■a 
 
 as 
 P. 
 
 300- 
 
 200- 
 
 100- 
 
 r 
 
 -3 
 
 -1 -2 -3 -h -5 
 
 — > log of H Au Clu Concentration (mol/liter) ■ 
 
 Fig. 17. — ^Variation of degree of dispersion with concen- 
 tration in the case of formation of gold hydrosol 
 by reduction of chlorauric acid. 
 
 that the velocity of reduction may be considerably 
 diminished by the dilution, while the formation of 
 nuclei, being very likely a coagulation process, may 
 be diminished to a lesser extent. 
 
 The reduction of a gold salt to metallic gold may 
 of course take place in some solvent other than Water. 
 A case that is of great interest both theoretically and 
 
GOLD RUBY GLASS. ^2, 
 
 practically is the formation of small gold particles 
 in glass (29). It is not stated whether this reaction 
 is a true reduction of gold salt by some component 
 of the gla^s or a mere decomposition of it by heat. 
 In any case we will deal with the case here as it shows 
 a close analogy to the formation of gold hydrosols 
 by reduction. 
 
 If a small quantity of a goM compound, e.g., gold 
 chloride, is added to a suitable glass at red heat (lead 
 or barium glass) and the glass is allowed to cool 
 slowly, a colourless glass is obtained, which becomes 
 ruby red when heated again quickly (but not to 
 melting). This red glass — gold ruby glass — as 
 Zsigmondy and Siedentopf first showed, contains a 
 great many ultra-microscopic gold particles (30) . If 
 a strip of the colourless gold glass is heated at one 
 end there will be a certain fall of temperature along 
 the glass strip in consequence of the thermal conduc- 
 tion. The result will be a glass with gold particles, 
 the size of which decreases as the distance from the 
 heat source increases. On the other hand, the number 
 of particles per unit volume is constant through the 
 whole strip. If the heating is continued further the 
 glass again becomes colourless. Zsigmondy suggested 
 the following explanation of these phenomena. At 
 first the gold chloride is reduced to molecular metallic 
 gold, which is kept in solution in the fused glass. By 
 cooling the glass a considerable supersaturation is 
 created and a condensation to very fine particles 
 takes place, some of which reach the nucleus limit, 
 i.e., become large enough to act as nuclei. But, owing 
 to the enormous viscosity of the glass at lower tem- 
 peratures, these condensation centres do not grow. 
 If the glass is quickly heated again, the system will 
 enter into a region where the velocity of growth is great 
 (the viscosity small) . Hence the small invisible gold 
 nuclei will grow to particles of ultra-microscopic size. 
 The experiment with the glass strip heated at one 
 
74 THE FORMATION OF COLLOIDS. 
 
 end shows that the velocity of growth of the particles 
 increases as the temperature rises and that the num- 
 ber of nuclei is not altered during the short period 
 of reheating. At a somewhat higher temperature, 
 however, the gold particles should be redissolved in 
 the glass. Zsigmondy has tried to connect all these 
 phenomena in the formation of gold ruby glass — as 
 well as those in the formation of colloids by conden- 
 sation in general — with some experiences of Tammann 
 concerning crystallization iji fused substances (31). 
 Tammann found the production of crystaUization 
 centres to increase as the degree of supersaturation 
 rises, then to reach a maximum, and finally to decrease 
 with increasing supersaturation. The maximum 
 velocity of growth does not in ge'neral coincide with 
 the maximum production of nuclei. Generally it 
 will be necessary to heat the melted matter to a 
 temperature above the one at which the production 
 of nuclei has its maximum in order to get a sufficient 
 velocity of growth. It may, however, be doubted 
 whether the crystallization and condensation in pure 
 melted salt and in solutions really take place in so 
 completely analogous a way that conclusions drawn 
 from the one phenomenon may be applied to the 
 other. At least, this must be done with great care 
 and caution. Thus it ought to be mentioned that 
 the number of nuclei produced in fused salt is only 
 some hundred per cubic millimetre, while in gold ruby 
 glass and in colloid solutions the number amounts to 
 thousands of millions per cubic millimetre. 
 
 It is of some interest to notice that colloid gold 
 formed by reduction had been known long before 
 the idea of a colloid was originated (32) . The 
 " aurum potabile " of the alchemists was very 
 probably a disperse system with gold as a disperse 
 phase. Macquer, in his Dictionnaire de Chymie, 
 second edition, Paris, 1774, tome 3, expresses a 
 rather modern view on this subject, viz. : " However, 
 
EARLY INVESTIGATIONS. 75 
 
 all these gold tinctures are nothing but gold which is 
 made extremely finely divided and floating in an 
 oily fluid. They are therefore, properly speaking, no 
 tinctures and, as stated by M. Le Baron in his Edition 
 of Lemery's Chemistry, they 'may be called drinkable 
 gold only so far as we connect with this name no 
 other idea than the one that the gold is swimming in 
 a fluid and is made into particles so fine that it may 
 itself be regarded as potable in the form of a fluid." 
 
 Johann Juncher's Conspectus Chemice (" ins 
 Teutsche iibersetzt," Halle (1749) ) contains many 
 statements which point to the formation of gold 
 colloids by reduction processes. In Table XXXIII. 
 we read : " Even in the case when not more than a 
 single drop of gold solution is distributed in some 
 ounces of ordinary water so as to make it imper- 
 ceptible both to sight and to smell the water will be 
 coloured red everywhere as soon as a drop of tin- 
 solution is added, which is to be regarded as a very 
 clear statement of the extreme smallness of the 
 particles." 
 
 In an essay published in London, 1794, Mrs. 
 Fulhame made a very clear statement regarding the 
 coloration of pieces of silk dipped in gold solution and 
 treated with phosphorus or hydrogen (33) . In the first 
 part of the nineteenth century several investigators" 
 obtained red or purple liquids by the reduction of gold 
 salts. Oberkampf (34) (1811), e.g., treated dilute gold 
 chloride with hydrogen and obtained liquids " d'une 
 belle couleur rouge pourpre, semhlahle a celle du vin 
 . . . sans qu'il se format aucun precipite." 
 
 Some very interesting remarks concerning the state 
 of the gold in gold glass were made by J. B. Richter 
 as early as 1802, and have recently been brought to 
 light by Wilh. Ostwald. In the middle of the nine- 
 teenth century Faraday undertook his famous 
 studies " On the experimental relations of gold to 
 light," published in the Trans, of the Roy. Soc. (35). 
 
76 THE FORMATION OF COLLOIDS. 
 
 Here the preparation of stable highly disperse gold 
 hydrosols was described and the heterogeneous 
 character of the red fluids clearly demonstrated. 
 According to Faraday, " A quick and ready mode of 
 producing the ruby fluid is to put a quart of the 
 weak solution of gold (containing about o-6 of a grain 
 of metal) into a clean bottle, to add a little solution 
 of phosphorus in ether, and then to shake it well for 
 a few moments : a beautiful ruby or amethystine 
 fluid is immediately produced, which will increase 
 in depth of tint by a little time. Generally, however, 
 the preparations made with phosphorus dissolved 
 in sulphide of carbon are more ruby thaif those 
 where ether is the phosphorous solvent. The process 
 of reduction appears to consist in a transfer of the 
 chlorine from the gold to the phosphorus and the 
 formation of phosphorous acids and hydrochloric 
 acid, by the further action of the water." The varia- 
 tion of the degree of dispersion with the dilution of 
 the gold salt at the reduction, the formation of secon- 
 dary particles (aggregates), and the necessity of using 
 very clean vessels for the preparation of stable gold 
 sols were also recognized by him. " Fluids thus pre- 
 pared may differ much in appearance. Those from 
 the basins, or from the stronger solution of gold, are 
 often evidently turbid, looking brown or violet in 
 different lights. Those prepared with weaker solu- 
 tions and in bottles, are frequently riiore amethystine 
 or ruby in colour and apparently clear. The latter, 
 when in their finest state, often remain unchanged 
 for many months and have all the appearance of 
 solutions. But they never are such, containing in 
 fact no dissolved but only diffused gold. The par- 
 ticles are easily rendered evident, by gathering the 
 rays of the sun (or a lamp) into a cone by a lens, and 
 sending the part of the cone near the focus into the 
 fluid. The cone becomes visible, and though the 
 illuminated particles cannot be distinguished because 
 
SILVER REDUCTION. ^^ 
 
 of their minuteness, yet the light they reflect is golden 
 in character, and seen to be abundant in proportion 
 to the quantity of solid gold present." 
 
 " A gradual change goes on amongst the particles 
 diffused through these fluids, especially in the case 
 where the gold is comparatively abundant. It 
 appears to consist of an aggregation. Fluids, at 
 first clear or almost clear to ordinary observation, 
 become turbid ; being left to stand for a few days, a 
 deposit falls." 
 
 " All the vessels used in these operations must be 
 very clean ; though of glass they should not be sup- 
 posed in proper condition after wiping, but shoiild 
 be soaked in water, and after that rinsed with distilled 
 water. A glass supposed to be clean, and even a 
 new bottle, is quite able to change the character of a 
 given gold fluid." 
 
 Next to gold reduction, the reduction of silver 
 attracts the attention. Silver compounds, especially 
 AgOH and some organic silver compounds, are easily 
 reduced. 
 
 The reduction of AgOH in aqueous solution by 
 hydrogen is of especial interest for the theory of 
 formation of colloids. 
 
 2AgOH + Ha ^ 2Ag + aHgO 
 
 This process was studied by Kohlschiitter (36). 
 He found that silver is deposited only at the surface 
 of contact between the solution and the wall of the 
 vessel containing it. The nature of this wall also has 
 a considerable influence upon the degree of dispersion 
 and other qualities of the coUoid formed. Thus 
 more fine-grained sols were obtained when the 
 reduction was carried out in a- vessel of quartz or 
 ordinary glass than in one of Jena glass. When 
 reduced in a platinum vessel the silver is deposited 
 in form of crystals on the wall. Thus the degree of 
 
78 THE FORMATION OF COLLOIDS. 
 
 dispersion is the lowest for a platinum surface. 
 Kohlschiitter made use of this fact in order to purify 
 his silver sols from AgOH. If a silver sol containing 
 AgOH is poured into a platinized platinum basin 
 and a current of hydrogen passed through the liquid 
 for some hours, almost all the Ag of the AgOH is 
 deposited on the wall of the vessel as silver crystals 
 while the silver colloid is left unchanged. 
 
 The formation of silver mirrors by reduction of 
 silver salts in solution is of interest in this connection. 
 This process has also been studied by Kohlschiit- 
 ter (37) . He found that the conditions for the depo- 
 sition of the silver in the form of a mirror were slow 
 velocity of reaction and the presence of certain high 
 molecular and colloid substances, probably acting 
 as protective colloids. The former condition may 
 be fulfilled by using complex silver compounds. 
 The mirror silver is probably built up of very fine 
 particles deposited close to each other. 
 
 If the dilution of the silver salt be very great the 
 formation of condensation nuclei is so diminished 
 that no silver deposit takes place. This is the case 
 when ammoniacal silver compounds are reduced by 
 aldehyde in great dilution. Here Zsigmondy and 
 Lottermoser found that gold particles added to the 
 reduction mixture act as condensation centres (38). 
 A silver colloid is formed with particles containing a 
 small gold nucleus. Silver particles also act as con- 
 densation centres in most mixtures of silver com- 
 pounds and reducing agents. The formation of the 
 photographic image is probably due to such a 
 process. Liippo-Cramer obtained series of beautifully 
 coloured silver sols by reducing silver nitrate (0-4 per 
 cent.) at 25° C. with hydroquinone in the presence 
 of gelatin (i per cent.) and various quantities of 
 silver nuclei (39). As the number of silver nuclei 
 added increased, the particles in the system formed 
 increased in number, and at the same time the colour 
 
CAREY LEA'S SILVER SOLS. 79 
 
 changed : blue gray — > blue -^ blue violet -^ ruby 
 red — > brownish yellow. 
 
 The great influence of small quantities of colloids 
 on the formation of silver mirrors is also observed 
 in case of hydrosol formation as a very high protec- 
 tive action. Such substances also raise the degree 
 of dispersion of the system, as was the case with the 
 gold hydrosols. Carey Lea's famous method of pre- 
 paring colloid silver depends on this (40). Silver 
 nitrate is reduced by ferrous citrate, and the deposit 
 formed is dissolved in water and reprecipitated with 
 ammonium nitrate. 
 
 By precipitating the sols thus obtained with 
 alcohol it is possible to get a substance containing 
 99 per cent, of silver, which forms a silver sol when 
 brought in contact with water. The remainder is 
 some ferrous colloid that acts as a protective colloid. 
 
 The silver hydrosols prepared according to Carey 
 Lea's method contain particles of various sizes. 
 Oden has shown that such a sol may be divided into a 
 series of sols with particles of nearly equal size by 
 means of fractional coagulation (41). As a coagula- 
 tor Oden used ammonium nitrate in a concentration 
 varying from o-o8 normal to 0-25 normal. The 
 smallest particles require the highest concentration 
 for coagulation. 
 
 Another form of reduction that has been studied 
 by Kohlschiitter in the case of silver is the electro- 
 lytic metal deposition (42) . At the cathode there is 
 at first formed a supersaturated solution of metallic 
 silver, and the higher the silver concentration of the 
 electrolyte and the current density, the higher the 
 supersaturation will be. Thus we get black spongy 
 deposits at high concentration and current density, 
 and crystalline, ghttering, compact deposits at low 
 concentration and current density. The degree of 
 
8o THE FORMATION OF COLLOIDS. 
 
 dispersion observed is in most cases, however, a 
 secondary one, and consequently it is very much 
 influenced by coagulating factors. The coagulating 
 power of the electrolytes present being reduced by 
 dilution, by using dilute solutions we here obtain 
 as the end-product a more highly disperse system. 
 Thus, J. Billiter prepared a silver hydrosol (together 
 with silver crystals) by electrolyzing 0-003 normal 
 AgNOj solution (43). According to observations 
 made by the writer, the formation of colloid is greater 
 if the cathode surface is made rather small in order 
 to raise the current density. Billiter was able to 
 produce hydrosols of some other metals, e.g., mercury, 
 lead, by the same method. 
 
 Guided by some observations made by J. Blake 
 concerning the pulverization of silver in the electric 
 arc (44) , the writer was able to prepare silver alcosols 
 by the electrolysis of alcohol with silver electrodes. 
 It has been proved by H. Nordenson that this process 
 fakes place in such a way that silver is first dissolved 
 at the anode (in the form of some chemical com- 
 pound) , and is then reduced at the cathode (45) . The 
 .silver cathode may in fact be replaced by a platinum 
 cathode without affecting the production of silver 
 colloid. This process may be given as an example of 
 a " false " dispersion process, the colloid-producing 
 process being in itself a condensation. Whether the 
 reduction be purely electrochemical or caused by the 
 hydrogen set free at the cathode is not yet decided. 
 
 There are some other phenomena connected with 
 those just described which at first sight also seem to 
 be dispersion processes, but which on closer inspection 
 prove to be condensation (reduction) processes. The 
 author observed that a silver plate submerged in 
 water or alcohol gave rise to a silver colloid when 
 illuminated by ultra-violet light or X-rays (46), and 
 Margerita Traube-Mengarini found that even the 
 boiling of silver in water produces some silver 
 
ACTION OF LIGHT. 8i 
 
 colloid (47). She also thought she had observed 
 the production of hydrosols by boiling platinum and 
 gold in water, but these results have not been con- 
 firmed by later investigators. In the case of silver 
 the author found that even alcohol which had been 
 left for some time in contact with a silver plate in 
 absolute darkness gave some colloid when illuminated 
 with ultra-violet light. Nordenson, who undertook 
 a very careful examination of all these phenomena (48) , 
 has shown that silver is oxidized both by water and 
 alcohol, and is thus dissolved (as AgOH or some 
 organic compound). This silver solution may then 
 be reduced by illumination (Svedberg) or by traces 
 of reducing agents (Traube-Mengarini) . The dis- 
 solution itself (oxidation process) is greatly accele- 
 rated by light, especially ultra-violet, and this must 
 be considered as the cause of the rather rapid pro- 
 duction of silver sol. As stated by the author, some 
 other metals, e.g., lead and copper, also give sols 
 when illuminated in water and alcohol ; but accord- 
 ing to Nordenson these sols contain the metal in an 
 oxidized state. It seems that in this case, too, the 
 oxidation process is very much increased by light, 
 no reduction, however, taking place afterwards. Sols 
 are therefore produced only when the oxides or 
 hydroxides are but very slightly soluble in the dis- 
 persion medium used. 
 
 No direct dispersion of metal through Hght is as 
 yet known. 
 
 By means of the reduction method we are also able 
 to get sols of the metals of the platinum group and of 
 mercury, bismuth, copper, tellurium and selenium. 
 In these cases the condensation process has not been 
 subjected to any detailed study, all the work being 
 as yet of a preparatory nature. Some of the most 
 interesting methods of this kind will be mentioned 
 below. 
 
82 THE FORMATION OF COLLOIDS. 
 
 As has been pointed out by the writer some years 
 ago, all reduction methods may be placed in one of 
 the following classes, according to the manner in 
 which the electric charges of the ions to be reduced 
 are removed (49) : — ■ 
 
 I. The electric charges may be transferred to 
 hydrogen either by using elementary gaseous 
 hydrogen or by using some suitable organic com- 
 pound in the presence of water, hydrogen ions being 
 thereby produced. The following examples illus- 
 trate this reaction : 
 
 (2Au - + 6C1') -f 3H2 = 2Au + (6H ■ -f 6C1') 
 (Fulhame (50) (1794), Oberkampf (51) (1811), 
 Vanino (52), (1905).) 
 f (2Ag ■ + 2OH') + H^ = 2Ag + (2OH' + 2H-) 
 1 2OH' + 2H • = 2H2O 
 
 (Kohlschiitter (53) (1908). ) 
 (2Au ■•• + 6C1') + 3H2O + 3HCHO = sAu + (6H • 
 + 6C1') + 3HCOOH 
 
 (Zsigmondy (54) (1898).) 
 
 2.. The charges may be transferred to neutral 
 metal masses, which thereby send an equivalent 
 quantity of metal ions into the solution, e.g., 
 
 (4AU - 4- 12CI') +3Sn = 4Ati + (3Sn-- + 1201') 
 
 (Fischer (55) (1827).) 
 
 3. The charges may be transferred to such ions of 
 low oxidation degree as have a tendency to pass into 
 a higher oxidation degree, e.g., 
 
 (2Hg ■ -f 2NO3') + (Sn •• + 2NO3) = 2Hg + (Sn •- 
 + 4NO3') 
 
 (Lottermoser (56) (1898).) 
 
 4. The charges may be neutralized by introducing 
 negative electrons in the system (electrolysis), e.g., 
 
 r(4Hg • -f 4NO3') + 20 = 4Hg + 4NO3' 
 1 4NO3' + <0 + 2H2O = (4H • -f 4NO3') + O, 
 
 (Billiter (57) ^1902).) 
 
VARIOUS REDUCTIONS. 83 
 
 The first and the third groups include those reduc- 
 tion processes that have hitherto been of the greatest 
 importance in practice. 
 
 In the third group we have Carey Lea's method, 
 mentioned above, for preparing silver hydrosols, and, 
 further, Lottermoser's method. By the aid of the 
 latter process, which uses stannous salts as a reducing 
 agent, the hydrosols of mercury, bismuth and copper 
 may be prepared. 
 
 In the first group are included almost all other 
 reduction methods of importance. The reduction 
 of silver by hydrogen (Kohlschiitter), and gold by 
 hydrogen peroxide (Stoeckl and Vanino, Doerinckel), 
 by phosphorus (Faraday, Zsigmondy), by formalde- 
 hyde (Zsigmondy), and by hydrazine chlorhydrate 
 (Gutbier) have already been described. Hydrazine 
 hydrate, hydrazine chlorhydrate, phenylhydrazine 
 chlorhydrate, and hydroxylamin chlorhydrate have, 
 according to Gutbier, proved to be very good re- 
 ducing agents. The hydrosols of the metals of the 
 platinum group, as well as those of selenium and 
 tellurium, may be prepared by their aid (58) . Paal 
 and his students have worked out some very efficient 
 reduction methods (59). The alkaline salts of pro- 
 talbic and lysalbic acid are used as protective 
 colloids. In the case of gold and silver no other 
 reducing agent is necessary. The preparation of the 
 hydrosols of platinum, osmium, palladium, iridium, 
 copper, tellurium and selenium requires, however, 
 the addition of a special reducing agent such as 
 hydrazine hydrate, sodium amalgam, etc. Most of 
 Paal's hydrosols can be evaporated to dryness and 
 dissolved again in the form of sols, the protective 
 colloids present preventing the metal particles from 
 getting too close together. 
 
 The reduction of metals other than gold has not 
 been known very long. Some experiments of H, 
 Rose (1828) may be regarded as an indication of the 
 
 6—2 
 
84 THE FORMATION OF COLLOIDS. 
 
 formation of colloid silver (60), and Dobereiner 
 (1832), perhaps, observed an unstable sol of plati- 
 num (61) . The silver sub-citrate of jWohler (62) (1839) 
 was recognized by Newbury (1886) as a silver hydro- 
 sol (63). He remarks that the solution of Wohler's 
 substance, though a clear red liquid in transmittent 
 light, appears "cloudy" in reflected light. " It seems 
 to me highly probable," he writes, " that this red 
 colour is caused by finely divided metallic silver." 
 Muthmann at the same time (1887) demonstrated 
 that the red substance, when dialyzed, did not diffuse 
 through the parchment membrane (64). 
 
 Oxidation. 
 
 There are only two oxidation processes that are 
 of importance for the formation of colloids, viz., the 
 oxidation of hydrogen sulphide to sulphur (Selmi 
 (1844) (65) ) , and of hydrogen selenide to selenium : 
 
 2H2S + O2 = 25 -f 2H2O 
 2H2Se -I- O2 = 2Se + iU^O 
 or 
 
 2H2S + SO2 = 3S + 2H2O 
 
 In the latter case thio-acids are also formed : 
 
 5H2S + 10H2SO3 = sHaSsOe + 12H2O. 
 
 The decomposition of thiosulphates by acids as a 
 process furnishing colloid sulphur (Engel (1891) (66), 
 Raffo (1908) (67) ), may very well come under this 
 heading, for the main part (|) of the sulphur pre- 
 cipitated is formed by oxidation of H2S by SO2 : 
 
 3Na2S203 + 3H2SO, = 3H2S2O3 -f 3Na2S04 
 
 H2S203 = S-F-,S02 + H20 
 
 2H2S2O3 + 2H2O = 2H2S + 2HaS04 
 
 2H2S + SO2 = 3S + 2H2O 
 
 sNagS^Os + 3H2SO4 = 4S + 3Na2SO, + 2H2SO4 
 + 3H2O. 
 
OXIDATION. 
 
 85 
 
 Oden has carefully investigated the formation of 
 sulphur sols by the reaction bet-ween HgS and SO 2 and 
 by the decomposition of thiosulphates (68). The 
 results he obtained are summed up below. 
 
 Wlien HgS is conducted into a solution of SO 2 in 
 -water, the amount of colloid sulphur formed and its 
 degree of dispersion depend upon the concentration 
 of SO2, as is sho-wn in the folio-wing Table. All 
 masses are taken in grammes and refer to 100 c.c, 
 1-8 normal SOg. 
 
 Table 8. 
 
 Conceiitration 
 of SO2. 
 
 Colloid 
 sulphur. 
 
 "Insoluble" 
 
 (non-colloid) 
 
 sulphur. 
 
 Amicro- 
 
 scopic 
 
 particles. 
 
 0-910 
 0-160 
 0-052 
 
 Traces. 
 
 Nearly 
 
 visible 
 
 particles. 
 
 Submicro- 
 
 scopic 
 particles. 
 
 1-8 normal. 
 
 1-44 " 
 0-9 
 
 8-333 
 
 9-891 
 
 13-022 
 
 1-936 
 Traces. 
 
 0-098 
 0-288 
 0-402 
 
 14-908 
 16-98 
 
 4-218 
 2-099 
 0-206 
 
 3-205 
 
 7-632 
 
 12-764 
 
 0-45 .. 
 0-225 >■ 
 
 1-936 
 
 Traces. 
 
 The amount of colloid sulphur at first increases as 
 the concentration of SOg decreases, reaches a 
 maximum, and then very rapidly decreases. The 
 total mass of sulphur increases -with the dilution of 
 SO 2, indicating that the reaction does not follo-w the 
 simple scheme 
 
 2H2S + SO2 = JS + 2H2O, 
 
 but that, according to the statements of earUer 
 investigators (Wackenroder (69), Selmi (70), Debus 
 (71), Spring (72), Lewes (73), Shaw (74), Hertlein (75) ), 
 it might give rise to various kinds of thio-acids and 
 also to some siilphuric acid. At great dilutions 
 (0-125 — 1'8 n.) the reaction will almost com- 
 pletely follow the simple scheme, the amount of 
 sulphur found being 16-98 and the theoretical value 
 
86 
 
 THE FORMATION OF COLLOIDS. 
 
 17-30. The table further shows that the degree of 
 dispersion of the siilfhur decreases as the concentration 
 decreases. 
 
 As to the decomposition of sodium thiosulphate 
 by sulphuric acid, Oden made the following observa- 
 tions : The formation of colloid sulphur decreases as 
 the concentrations of the reacting components 
 decrease, and decreases also, but only a little, with 
 lowering of temperature. 
 
 Table 9. 
 30 c.c. 3-normal NaaSaO., ; lo c.c. concentrated H2SO4. 
 
 Water. Cubic 
 centimetres. 
 
 Colloid sulphur. 
 Grammes. 
 
 Insoluble 
 sulphur. 
 Grammes. 
 
 Small 
 particles. 
 Grammes. 
 
 Large 
 particles. 
 Grammes. 
 
 
 
 1-691 
 
 I 031 
 
 0-910 
 
 0-781 
 
 5 
 
 1-353 
 
 1-365 
 
 0-022 
 
 I-33I 
 
 10 
 
 1-273 
 
 1-282 
 
 Traces. 
 
 1-273 
 
 20 
 
 1-070 
 
 1-463 • 
 
 Traces. 
 
 1-070 
 
 30 
 
 0-384 
 
 2-123 
 
 — 
 
 0-384 
 
 50 
 
 0-059 
 
 2-439 
 
 
 0-059 
 
 The experiments tabulated above (Table 9) were 
 carried out in the following manner : 30 c.c. of 
 3-normal NagSgOg solution were mixed with various 
 quantities of water, and then 10 c.c. concentrated 
 sulphuric acid Were added. As the concentration of 
 NaaOgSg diminishes, not only does the qtiantity of 
 colloid sulphur decrease, but also the degree of dis- 
 persion of the system. Some other experiments 
 showed that a dilution of the sulphuric acid had an 
 analogous effect on the formation of colloid sulphur. 
 
 In practice it is advisable not to choose too high a 
 temperature, for the higher the .temperature the 
 stronger is the irreversible coagulating power of the 
 cations present, viz., the sodium ions. It has also 
 proved advantageous to use rather small quantities 
 
ODIN'S SULPHUR SOLS. 87 
 
 of solutions, 100 c c. or so. After the formation of 
 the colloid sulphur the reaction mixture should be 
 immediately coagulated with pure sodium chloride 
 solution and the precipitate separated from the 
 liquid by centrifuging. The bulk of the deposit dis- 
 solves in water, forming a sulphur hydrosol. The 
 residue, consisting of " insoluble " sulphur, is removed 
 by centrifuging or by sedimentation. By once more 
 adding sodium chloride solution and centrifuging, the 
 sulphur may be purified. The colloid sulphur con- 
 tained in the reacting mixtures from 2H2S + SOg 
 may be freed from impurities in the same way. 
 
 Sulphur hydrosols obtained by this or by the other 
 methods are built up of particles of various size from 
 a few ^ju, to some hundreds of /^/.t. Oden has worked 
 out a very ingenious method of dividing the original 
 polydisperse sol into a series of sols with particles 
 of almost equal size, i.e., into a series of monodisperse 
 sols. When dealing with the polydisperse sulphur 
 sols he observed that the small particles were less 
 sensitive to coagulating agents than the larger ones, 
 i.e., by slowly adding a salt solution the larger 
 particles were first precipitated. Thus it should be 
 possible to fractionate the sulphur sol by means of a 
 reversibly coagulating salt. Oden used sodium 
 chloride. In Table 10 such a series of sols obtained by 
 Oden is reviewed. 
 
 The concentration necessary for coagulating a 
 sulphur sol, i.e., the precipitation value, varies with 
 the temperature. The higher the temperature, the 
 higher is the precipitation value. At a given salt 
 concentration the temperature at which coagulation 
 sets in is higher for the larger particles. Thus on 
 slowly cooling a polydisperse sulphur sol, the greater 
 particles will be first precipitated, then the particles 
 that are a little smaller, and so on. According to 
 Oden, a method for separating the particles of various 
 size may be founded on this phenomenon. 
 
THE FORMATION OF COLLOIDS. 
 
 The removal of the electrolyte used for the purify- 
 ing and fractionating of the sulphur colloid is carried 
 out in the following manner. The main part of the 
 electrolyte is removed by cooling the concentrated 
 sol to about — 10°, when a fairly salt-free soluble 
 coagulum is deposited. A very efficient purification 
 may then, according to Oden, be attained by electro- 
 
 Tablk id. 
 
 The sulphur sol 
 
 
 is not coagu- 
 lated 
 
 is coagulated 
 
 Ultra-microscopic characteristics. 
 
 by NaCl of the following 
 
 
 concentration 
 
 in normality : 
 
 
 0-25 
 
 00 
 
 Faint amicroscopic light-cone, no 
 submicrons. 
 
 0-20 
 
 0-25 
 
 Clearly visible amicroscopic light- 
 cone, no submicrons. 
 
 0'l6 
 
 0-20 
 
 Strong amicroscopic light-cone, no 
 submicrons. 
 
 13 
 
 016 
 
 Particles just visible (about 25 fi/j, in 
 diameter) . 
 
 O-IO 
 
 0-I3 
 
 No amicroscopic light-cone, diameter 
 of particles, 90 /x/x. 
 
 007 
 
 O-IO 
 
 No amicroscopic light-cone, diarneter 
 of particles, 140 mi. 
 
 
 
 0-07 
 
 No amicroscopic light-cone, diameter 
 of particles, 210 fj.fi. 
 
 lysing the sol in an apparatus provided with suitable 
 diaphragms. 
 
 The oxidation of hydrogen selenide to selenium 
 hydrosol is of less importance and has not yet been 
 sufficiently studied. 
 
 A curious case of electrolytic pulverization studied 
 by E. Miiller and .his students, and hitherto looked 
 upon as a dispersion process, may most probably be 
 regarded as an oxidation process (76) . If a platinum 
 
EARLY INVESTIGATIONS. 89 
 
 plate is covered to some extent with sulphur and 
 used as a cathode in pure water, colloid sulphur is 
 produced at the contact surface between the sulphur 
 and the platinum. Selenium and tellurium show 
 analogous phenomena. The explanation is probably 
 that hydrogen sulphide is first formed by a reaction 
 between the electrolytically produced hydrogen and 
 the sulphur, and that this HgS is then oxidized by 
 the oxygen dissolved in the water. Another explana- 
 tion would be that a very unstable hydrogen . per- 
 sulphide (H2S2) is formed at the cathodg and that it 
 dissociates into H2S and colloid stdphur. In this 
 case the process should be included in the following 
 group (dissociation processes) . 
 
 The formation of sulphur hydrosols by oxidation 
 of hydrogen sulphide water may probably be one of 
 the colloid formation processes first observed. The 
 gradual clouding of the natural sulphur waters has 
 long been known. In fact, our knowledge of it is 
 perhaps of the same date as our acquaintance with 
 those waters. Scheele (77) (about 1770), Berg- 
 man (78) (1782), Le Veillard (79) (1789), Berthol- 
 let (80) (1798), Berzelius (81) (1808) and Dobe- 
 reiner (82) (1813) mention such reactions in their 
 treatises. Berzelius writes as follows : "In water it 
 (HoS) dissolves abundantly ... if the water con- 
 tain atmospheric air some part of the air is destroyed, 
 the hydrogen being oxydized to water and the sul- 
 phur set free, on jvhich the water gets a milky 
 appearance." Wackenroder (83) (1846) studied the 
 turbid Hquid obtained by passing HgS through an 
 aqueous solution of SO2, and Selmi (84) (1844-5) 
 made this liquid the subject of a close examination 
 that brought him to a very clear and modern view of 
 the form of existence of the sulphur in it and its ; 
 colloid chemical reactions. He recognized the con- 
 nections between such products as the disperse 
 systems of Prussian blue, AsjSg, casein, albumin, 
 
90 THE FORMATION OF COLLOIDS. 
 
 Wackenroder's liquid, etc., and called them pseudo- 
 solutions : "II resulte, en outre, que le soufre emulsion- 
 nahle presente des phenomenes analogues a ceux qui 
 s'observent dans heaucoup d'autres corps qui jouissent 
 de la propriete 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. . . ." 
 Some years later (1888) Debus (85) expresses a 
 similar opinion : " If we look back on the properties 
 of the sulphur as it is contained in solution in Wacken- 
 roder's liqui^', and can be obtained from it by partial 
 evaporation, we find that it possesses all the pro- 
 perties which Graham describes as characteristic of 
 the colloids." 
 
 " The sulphur dissolved in Wackenroder's solution 
 does not diffuse through porous clay or parchment. 
 It is held in solution by very feeble forces. Slow and 
 gradual separation takes place when its solutions are 
 kept for some time, or complete precipitation if 
 apparently inert substances, such as sodium chloride, 
 charcoal powder, or basic sulphate, are added. The 
 unstable condition of its molecules, their slow change 
 into other modifications, and finally its gummy, 
 sticky condition remind one of the colloids." 
 
 Dissociation. 
 
 The formation of disperse systems by dissociation 
 is as yet but very imperfectly known. Some pro- 
 cesses will be treated here which are probably to be 
 referred to this class. 
 
 As already mentioned above, some (|) of the 
 colloid sulphur formed by the action of acids upon 
 thiosulphates is derived from a dissociation of thioic 
 acid : 
 
 H2S2O3 = 5 + SO2 + H2O. 
 
 Some polysulphides also produce colloid sulphur on 
 dissociating. 
 
DISSOCIATION. 91 
 
 Wa. Ostwald observed that a solution of nickel 
 carbonyl in benzene when boiled gives rise to a nickel 
 benzosol (86) . The process is nt) doubt to be regarded 
 as a dissociation, viz. : 
 
 Ni(CO)4^m' + 4CO. 
 
 All organic compounds dissociate at high tempera- 
 tures, and may under suitable conditions give rise to 
 disperse systems of carbon. The Acheson graphite 
 used in lubrication is manufactured by treating 
 carbon dust formed in the carborundum furnaces 
 with gallotannic acid and water (87) . 
 
 Much more important is the dissociation of salts, 
 especially metal halides, by which process the metal 
 under suitable conditions may be deposited in highly 
 disperse form. Among this group of phenomena 
 maj' be included the darkening of the silver halides 
 in the light. According to R. Lorenz, it is possible to 
 follow the formation and growth of the silver particles 
 in pure silver bromide by means of the ultra-micro- 
 scope (88). In the optically empty mass there first 
 appears an amicroscopic hght-cone, which gradually 
 disintegrates into submicroscopic particles. As has 
 been shown by R. Lorenz and his students, many 
 metal halides dissociate spontaneously, even in the 
 dark, into metal and hahde, e.g., PbClg, TlCl, 
 TlBr, AgCl, AgBr, and this process takes place both 
 in the fused salts and in sohd crystals. Of course 
 the degree of dissociation is rather small. Optically 
 empty systems of these salts exist only in the presence 
 of some excess of hahde. As an example, the prepa- 
 ration of optically empty PbClg will be described. 
 About 100 gm. of commercial lead chloride is melted 
 in a tube of hard glass placed within an electric resist- 
 ance furnace and treated for one to two hours with 
 a mixture of dry hydrochloric and chlorine gas. 
 Then the glass tube is taken out of the furnace and 
 cooled. The congealed mass consists of pure opti- 
 
92 THE FORMATION OF COLLOIDS. 
 
 cally empty crystals of lead chloride. If a piece of 
 metallic lead is introduced into a pure fused specimen 
 of PbClg of this sort there occurs a kind of pulveriza- 
 tion of the metal, a brownish black lead fog emanating 
 from it. This phenomenon may perhaps be explained 
 in the following way. By the presence of the metallic 
 lead the equilibrium in the melted salt is disturbed 
 and the dissociation of the halide into metal and 
 chlorine is increased. If such a black fused mixture 
 is cooled and then ptit to an ultra-microscopic test, it 
 will prove to contain numerous metal particles, some 
 of which are amicroscopic and others submicroscopic. 
 The system is therefore to be regarded as a colloid 
 solution of metal in salt. R. Lorenz has adopted 
 the name " pyrosol " for such systems. 
 
 This " metal pulverization " in fused salts was 
 first observed in the electrolysis of such systems and 
 regarded as a kind of electric cathode pulverization 
 (Faraday, Bunsen). The earlier investigators did 
 not, however, pay attention to the fact that all salts, 
 even " dry " ones, always contain a little water that 
 cannot be removed by heating only. For a halide 
 it is necessary to heat the salt in a current of HCl at 
 a slowly rising temperature in order to get rid of the 
 water and avoid hydrolysis. If this precaution is 
 not taken, oxides and oxychlorides will be formed 
 during the electrolysis, and the metal fog phenomenon 
 cannot be clearly observed, but appears as part of the 
 general clouding of the electrolyte. 
 
 Another very interesting phenomenon, observed 
 by H. Siedentopf, may perhaps be of an analogous 
 nature (89). He found that dry crystals of rock 
 salt heated with sodium vapour in vacuo assumed a 
 blue coloration and became quite like the natural 
 blue rock salt sometimes found in salt mines. Wlien 
 examined in the ultra-microscope the blue salt, the 
 artificial as well as the natural variety, proved to 
 contain numerous metallic particles, arranged in the 
 
DOUBLE DECOMPOSITION. 93 
 
 directions of cleavage of the crystal. The explana- 
 tion may be that the sodium vapour penetrating into 
 the fine clefts of the crystal disturbs the equilibrium 
 of the hahde and causes some dissociation of it and 
 formation of sodium colloid by a condensation pro- 
 cess. Obviously there is also another explanation — 
 adopted by Siedentopf — viz., that the sodium colloid 
 is produced directly from the sodium vapour by 
 condensation in the fine clefts of the crystal. Now 
 blue rock salt may also be prepared in another 
 manner, viz., by exposing ordinary rock salt to 
 cathode or radium rays. In this case the mechanism 
 of the process must of course be a dissociation of 
 NaCl into Na -\- CI. The identity of the products 
 obtained by this method and by the treating of rock 
 salt with sodium vapour may be interpreted in 
 favour of the opinion held by the writer. 
 
 Many other forms of coloration caused by cathode 
 and radium rays in various substances should, 
 according to Doelter, be regarded as colloid colours 
 (90) . The chemical reactions, dissociations or double 
 decompositions to which the substances may be 
 subjected are as yet almost completely unknown. 
 
 Double Decomposition. 
 
 A great many double decomposition reactions 
 give rise to disperse systems. 
 The cases of most interest are : 
 
 1. Formation of oxide sols by 
 
 (a) Hydrolysis. 
 
 (6) Precipitation with alkaUs. 
 
 (c) Decomposition of salts with acids. 
 
 2. Formation of sulphide sols by 
 
 (a) Precipitation with hydrogen sulphide. 
 
 [b) Decomposition of sulpho-salts. 
 
94 THE FORMATION OF COLLOIDS. 
 
 3. Formation of salt sols by precipitation with 
 easily soluble 
 {a) Halides. 
 (6) Sulphates. 
 
 [c) Chromates. 
 
 [d) Phosphates. 
 
 [e) Ferro- and ferri-cyanides. 
 
 As to the question of the relation between the 
 experimental conditions and the degree of dispersion 
 of the system formed, a great many experiments 
 have been carried out with various double decompo- 
 sition reactions in order to settle this matter. P. P. 
 V. Weimarn, who has devoted a great deal of work 
 to this subject (91), pronounces the opinion that as 
 the concentration of the reacting substances decreases, 
 the degree of dispersion — measured a quarter to 
 half an hour after the reaction — first decreases and 
 then increases again. When measured about a 
 month after the reaction the degree of dispersion is 
 supposed always to rise with concentration. This 
 view V. Weimarn has tried to maintain, especially by 
 his investigations on the formation of BaSO^ from 
 MnS0i + Ba(SCN)2. 
 
 Recent investigations by Oden have shown that the 
 degree of dispersion always diminishes as the con- 
 centration decreases (92). By means of a special 
 apparatus he was able to measure the distribution of 
 the various sizes of particles in BaS04 — suspensions 
 produced when, under various conditions, he mixed 
 a solution of barium ions, e.g., Ba(SCN)2, with a 
 solution of sulphate ions, e.g., (NH4)2S04. The 
 formation of secondary particles, i.e., aggregates of 
 primary particles, easily misleads the investigator. 
 Oden found that secondary particles are very often 
 formed when the size of the particle is small. Above 
 100 nfx, scarcely any are formed, but below 20 /x/n one 
 nearly always gets secondary particles if no special 
 
BaSOi SOLS. 
 
 95 
 
 precautions are taken, e.g., the addition of citrate, in 
 order to prevent aggregation and to cause such 
 a^ggregates as may have been formed owing to local 
 
 f^« 
 
 w 
 
 Sa-SOf, 
 
 '^oe..^ I 
 
 i \ 
 
 ! \ 
 \ 
 
 \ 
 
 
 Fig. i8. — ^Variation of degree of dispersion, as represented by 
 the distribution curve, with concentration of reacting ions. 
 (According to Oden.) 
 
 high concentration of electrolytes to redissolve into 
 primary particles. The chemical reaction, being a 
 pure ionic reaction, is supposed to take place exceed- 
 ingly fast, and the supersaturation immediately before 
 
96 THE FORMATION OF COLLOIDS. 
 
 condensation should therefore be equal to the ratio 
 between the total concentration of BaS04 and the 
 solubility of BavS04. Thus it is in agreement with 
 the result obtained for condensation in gases that 
 
 iso 
 
 200^ 
 
 ISO 
 
 /oc 
 
 o't- 
 
 BaSO,, 
 
 i 
 
 I :I5° \\ I 
 
 \ : \ V 
 
 % 
 
 /I /\ 
 
 / 
 
 /»0^ 
 
 \ 
 
 \ 
 
 •X 
 
 N, 
 
 \"-1---->N 
 
 I I I 
 
 -1o 
 
 I I 
 
 -as 
 
 Ol/L 
 
 I 
 
 Fig. 19. — ^Variation of degree of dispersion with temperature 
 during the reaction. (According to Oden.) ^,j 
 
 the degree of" dispersion increases as the concentra- 
 tion rises. The production of nuclei seems to be 
 • almost exclusively spontaneous. Oden found that 
 particles of BaS04 added to one of the reaction com- 
 ponents before the reaction have little or no influence 
 on the degree of dispersion of the system formed. In 
 
DISTRIBUTION OF PARTICLES. 
 
 97 
 
 Table ii. 
 
 Experiments on the relation between the degree of tiisper. 
 sion, as represented by the distribution curve, and the 
 concentration of the reacting ions, i.e., the degree of 
 supersaturation . 
 
 
 F{r). 
 
 r. 
 
 Molar concentration of BaS04-ion at the moment of reaction. 
 
 
 O'l 
 
 O'OI, ' O'OOI. 
 
 20 
 
 _ 
 
 
 
 1-35 
 
 1-5 
 
 — 
 
 — 
 
 9-9 
 
 1-2 
 
 — 
 
 
 
 38-9 
 
 I-O 
 
 — 
 
 4-6 
 
 105-0 
 
 0-9 
 
 — 
 
 
 
 178 
 
 0-8 
 
 — 
 
 239 
 
 485 
 
 0-75 
 
 
 
 72-5 • 
 
 275 
 
 070 
 
 2-7 
 
 118 
 
 
 o-6o 
 
 i6-o 
 
 187 
 
 
 0-55 
 
 — 
 
 218 
 
 
 050 
 
 42-0 
 
 244 
 
 
 046 
 
 • — 
 
 224 
 
 
 0-40 
 
 102 
 
 198 
 
 
 0-37 
 
 — 
 
 — 
 
 
 0-35 
 
 137 
 
 151 
 
 
 0-33 
 
 161 
 
 — 
 
 
 0-31 
 
 194 
 
 — 
 
 
 0-30 
 
 219 
 
 47-0 
 
 
 0-27 
 
 255 
 
 
 
 0-25 
 
 273 
 
 
 
 0-23 
 
 268 - 
 
 
 
 0-20 
 
 •258 
 
 
 
 p-i8 
 
 233 
 
 
 
 Coarse 
 material. 
 
 — 
 
 I'2 
 
 /F{r)dr= 5-4 
 J per cent. 
 
 — 
 
 Fine- 
 grained 
 residue. 
 
 o-i8 
 
 fF{r)dr= 72-4 
 ■4 per cent. 
 
 0-30 
 
 fF{r)dr=iy7 
 J per cent. 
 
 o'75 
 
 lF(r)dr=i7 
 ,■65 per cent. 
 
 F.C. 
 
98 THE FORMATION OF COLLOIDS. 
 
 Table ii and Fig. i8 a few of Oden's most charac- 
 teristic experimental series are presented. 
 
 Here r denotes the radius of the particles in \x. (or, 
 properly, the radius of a sphere of the same 
 material sinking with the same velocity in the 
 same fluid) . 
 
 F\j) the functional relation between the percentage 
 mass of particles and the radius, and 
 
 \ 
 
 -ZJi -t.5 
 I I I 
 
 A 
 
 cuaot 
 
 J\ 
 
 \r 
 
 m-', 
 
 V, 
 
 c=g2_, 
 
 £^ 2J ao • 3J 
 
 III III I 
 
 Fig. 
 
 20. — ^Variation of degree of dispersion with solubility of 
 the disperse phase formed. (According to Od^n.) 
 
 F(r)dr the percentage mass of particles in, the 
 radius interval dr. 
 
 In the figures the Inr are plotted as abscissae against 
 the rF{r) as ordinates in 'order to have a suitable 
 form of graphic representation. 
 
 The temperature of the reacting mixture influences 
 the degree of dispersion of the system produced. 
 Oden found the following values (Table 12) (Fig. 19) 
 for the reaction products of 0-2 molar solutions of 
 Ba(SCN)2 and (NH^I^SO^ at 0°, 15° and 100°. 
 
INFLUENCE OF TEMPERATURE. 
 
 99 
 
 Thus the degree of dispersion becomes lower as the 
 temperature rises, a fact that is very probably due to 
 the increased solubility of the disperse phase at 
 higher temperatures. 
 
 Table 12. 
 
 7. 
 
 F(r). 
 
 
 0° 
 
 15° 
 
 100° 
 
 I-O 
 
 
 
 
 
 
 0-8 
 
 
 
 — 
 
 140 
 
 0-7 
 
 60 
 
 — 
 
 28-7 
 
 0-6 
 
 II-6 
 
 
 
 61 -2 
 
 0-55 
 
 — 
 
 — ^ 
 
 
 
 0-50 
 
 21-6 
 
 — 
 
 164 
 
 0-45 
 
 — 
 
 21-3 
 
 224 
 
 040 
 
 740 
 
 88-3 
 
 296 
 
 0-35 
 
 150 ' 
 
 146 
 
 338 
 
 0-33 
 
 — 
 
 200 
 
 329 
 
 032 
 
 204 
 
 — 
 
 
 
 030 
 
 314 
 
 353 
 
 256 
 
 0-27 
 
 419 
 
 437 
 
 135 
 
 0-25 
 
 972 
 
 444 
 
 96-5 
 
 0-24 
 
 934 
 
 — 
 
 
 
 023 
 
 540 
 
 378 
 
 91-3 
 
 0-22 
 
 22Q 
 
 — 
 
 — 
 
 0-20 
 
 174 
 
 241 
 
 84-5 
 
 018 
 
 165 
 
 
 
 017 
 
 
 
 
 Fine- 
 
 o-i8 
 
 0*20 
 
 0-20 
 
 grained 
 residue. 
 
 •J per cent. 
 
 JF{y)dr = 40-4 
 J per cent. 
 
 fF{r)dr = 12 
 J per cent. 
 
 Oden compared the degree of dispersion of systems 
 of BaSOi, SrSOi and CaSOj produced by mixing 0-2 
 molar solutions of the nitrates with o'2 molar solu- 
 tions of ammonium sulphate (Table 13) (Fig. 20), 
 
 7-2 
 
100 THE FORMATION OF COLLOIDS. 
 Table 13. 
 
 BaSOj. 
 
 SrS04. 
 
 CaSOj. 
 
 r. 
 
 F{t). 
 
 r. 
 
 FM. 
 
 r. 
 
 F{r). 
 
 I-O 
 
 1-3 
 
 10 
 
 2-74 
 
 40 
 
 O-IO 
 
 07 
 
 4-6 
 
 9 
 
 6-67 
 
 30 
 
 0-27 
 
 0-5 
 
 16-4 
 
 8 
 
 IO-6 
 
 25 
 
 062 
 
 0-45 
 
 36-0 
 
 7 
 
 i8-6 
 
 20 
 
 3-57 
 
 0-40 
 
 87-2 
 
 6 
 
 270 
 
 18 
 
 4-16 
 
 0-37 
 
 154 
 
 5-5 
 
 26-0 
 
 17 
 
 4-95 
 
 0-35 
 
 215 
 
 5-0 
 
 21-4 
 
 16 
 
 23-1 
 
 0-32 
 
 287 
 
 4-5 
 
 14-4 
 
 15 
 
 i8-6 
 
 0-30 
 
 355 
 
 40 
 
 no 
 
 14 
 
 II-4 
 
 0-27 
 
 465 
 
 3-5 
 
 8-8o 
 
 . 13 
 
 661 
 
 026 
 
 494 
 
 3-2 
 
 '6-44 
 
 12 
 
 3-59 
 
 0-25 
 
 548 
 
 — 
 
 ■ — 
 
 10 
 
 1-37 
 
 0-24 
 
 579 
 
 — 
 
 ■ — ■ 
 
 8 
 
 0-84 
 
 0-23 
 
 578 
 
 
 
 - 
 
 
 0-22 
 
 545 
 
 
 
 
 
 0-2I 
 
 ; 488 
 
 
 
 
 
 0-20 
 
 429 
 
 
 
 
 
 019 
 
 ■ 379 
 
 
 
 
 
 018 
 
 327 
 
 
 
 
 
 0-17 
 
 280 
 
 
 
 
 
 o-i6 
 
 209 
 
 
 
 
 
 0-15 
 
 167 
 
 
 
 
 
 0-13 
 
 102 
 
 
 
 
 
 0-I2 
 
 83-5 
 
 
 
 
 
 
 
 CO 
 
 
 CO 
 
 
 
 Coarse 
 
 fF(r)dr = 
 
 Coarse 
 
 fF{r)dr=2-5 
 
 
 
 material. 
 
 material. 
 
 Fine- 
 
 0-12 
 
 Fine- 
 
 3'2 
 
 Fine- 
 
 8 
 
 grained 
 
 fF{,)dr=i-g 
 
 grained 
 
 fF{r)dr=i 
 
 grained 
 
 fF{r)dr=2-S 
 
 residue. 
 
 ^ percent. 
 
 residue. 
 
 ^ per cent. 
 
 residue. 
 
 ^ percent. 
 
 With regard to the secondary and tertiary degree 
 of dispersion studied by Oden in. the case of BaSOj, 
 reference will be made to these investigations in a 
 
SULPHIDE SOLS. loi 
 
 subsequent monograph dealing with the changes of 
 state in colloids. 
 
 In cases when the chemical reaction proceeds 
 slowly, e.g., 'in the formation of colloid AsjSj from 
 As^Gg-solution by the action of HgS or during 
 hydrolysis' of FeClg, the degree of dispersion .decreases 
 as the concentration increases, as was the case 
 with the reduction of HAUCI4. It is not yet known, 
 however, to what extent the formation of secondary 
 particles, i.e., aggregates, may be responsible for the 
 effect observed. 
 
 Berzelius had already observed that the reaction 
 
 AsaOg + 3HaS = As^S^ + 3H2O 
 
 gave rise to a yellow solution of AS2S3, and he remarks 
 that " this solution probably ought to be regarded 
 more as a suspension of transparent particles " (93). 
 Schulze found that by means of the said method 
 systems of various properties can be obtained accord- 
 ing to the concentrations of AsgOg at which the reac- 
 tion takes place. He writes as follows : " It is very 
 remarkable that the fluids which are obtained by 
 diluting the more concentrated ones do not in all 
 respects resemble the directly made solutions of the 
 same degree of concentration. In transparency the 
 former are very inferior to the latter, and they show 
 in reflected light a more or less pure yellow colour in 
 contradistinction to the rich yellow colour of the 
 solutions directly made " (94). 
 
 A little later Picton and Linder (95) prepared 
 AsaSj-sols by various double decomposition methods, 
 and they notice, a point that Schulze had overlooked, 
 that the properties of these " pseudo-solutions " 
 depend on the size of the particles of which they are 
 built up. The AsjSg-sol that contained the greatest 
 particles (just visible in a microscope), the " a-solu- 
 tion " of Linder and Picton, was prepared by mixing 
 
102 THE FORMATION OF COLLOIDS. 
 
 arsenical tartrate solution with hydrogen sulphide 
 water. The ' ' j8-solution, ' ' obtained by mixing sodium 
 arsenite and hydrogen sulphide water, could not be 
 resolved with the microscope, and " presented the 
 appearance of a perfectly homogeneous ftquid." Still 
 it showed no trace of diffusion. The " y-solution," 
 prepared by mixing concentrated AsjOj-solution, 
 about 9-5 per cent., with hydrogen sulphide water, 
 resembled the jS-solutipn, but showed a slight diffu- 
 sion. Yet it was not filtrable. Finally the " 8-solu- 
 tion," obtained by the same reaction as the y-solu- 
 tion, but from a more dilute (about 2 per cent.) AsgOg- 
 solution, was diffusible as well as filtrable. It still 
 gave, however, the Tyndall light-cone. Linder and 
 Picton remark that the 8-sol must be looked upon as 
 containing particles of smaller ^ize than the y-sol. 
 Some investigations of the writer concerriing the 
 light absorption in AsgSg-sols show that the degree of 
 dispersion rises more and more as the dilution of 
 AsjOj is increased (96). He studied dilutions up to 
 i-io~' normal. This result has been confirmed by 
 recent experiments of G. Borjeson, who succeeded in 
 deterniining the absolute size of the particles in a 
 series of AsgSj-sols obtained from AsgOg solutions of 
 various concentrations (97). In Table 14 some of his 
 measurements are given : 
 
 Table 1:4. 
 
 Concentration of AS293. ^^^'"^ "S"''" °^ 
 
 I- 10-^ normal. i 39 /i(» 
 
 5-I0-* ,, I i6 fijii 
 
 i-io--* ,, I II nfi 
 
 I 
 
 Some other processes by which sulphide sols may 
 be obtained will also be mentioned. Reactions of 
 
HYDROLYSIS. 103 
 
 the type where only shghtly electrolytic dissociated 
 substances are formed have, besides the case already 
 mentioned, been used also in the following cases. 
 Copper and zinc sulphide hydrosols were prepared by 
 Linder and Picton by treating the hydrates suspended 
 in water with hydrogen sulphide (98) . Lottermoser 
 treated mercury cyanide, glycocol copper, and copper 
 acetoacetic ester with hydrogen sulphide and 
 obtained sulphide sols of those metals (99). He 
 even succeeded in carrying out the reaction in 
 organic solvents, t.g., alcohol and ether, and thus 
 obtained sulphide organosols. 
 
 Reactions which give rise to strongly dissociated 
 products may be used for the preparation of sulphide 
 sols in any of the following ways, viz. : i. By 
 operating in highly diluted solution. 2. By adding a 
 protective colloid. 3. By removing the electrolytes 
 or replacing the ions present by more suitable, i.e., 
 less coagulating, ions. Process i was first observed 
 by Berzehus (100) and by Schulze (loi), and later on 
 practised by Winssinger (102). Process 2 has been 
 used by many investigators. Paal dissolved pro- 
 talbic or lysalbic silver in ammonium sulphide, and 
 thus obtained colloid silver sulphide protected by 
 the ammonium salt of the protalbic or lysalbic 
 acid (103) . After dialyzing, the sol was precipitated 
 with alcohol and the precipitate dried. This pro- 
 tected colloid silver sulphide dissolves easily in 
 water, forming a hydrosol again. Process 3 was also 
 observed by Berzelius (104). Spring used it for the 
 preparation of copper sulphide sol (105) . He washed 
 copper sulphide coagulum with HgS-water. 
 
 Hydrolysis, i.e., the decomposition of a compound 
 by water, is a reaction of great interest for colloid 
 chemistry. It may be regarded as a double 
 decomposition reaction, viz., 
 
 MeAc + H2O :^ MeOH + HAc, 
 
104 THE FORMATION OF COLLOIDS. 
 
 where Me stands for a metal atom, and Ac for an acid 
 group. 
 
 In many cases it takes considerable time to reach 
 equilibrium. Goodwin (io6), who studied the hydro- 
 lysis of ferric chloride, has suggested that, in such 
 cases, first a rapid reaction of the kind 
 
 Fe-- + H20 = FeOH- + H • 
 
 takes place, and after that a slow formation of 
 FcgOg. This product then condenses to a FcgOg 
 —or Fe(0H)3 — sol. The hydrolytic equilibrium 
 varies with the dilution. At 25°'C., Goodwin found 
 the following values (Table 15) : — 
 
 Table 
 
 15. 
 
 
 Concentration 
 
 Degree of hydrolysis 
 
 mol/Ut. 
 
 in 
 
 percentage. 
 
 0-45 
 
 
 2 
 
 0-225 
 
 
 II 
 
 o-ogo 
 
 
 37 
 
 0'045 
 
 
 53 
 
 0-0225 
 
 
 67 
 
 0-0090 
 
 
 84 
 
 0-0045 
 
 
 91 
 
 Thus there is practically no hydrolysis above the 
 concentration 5-10""^ mol/lit, and below 3'io~* there 
 is almost total- hydrolysis. The velocity of the 
 hydrolysis differs considerably for various degrees of 
 temperature and dilution. It increases as the 
 temperature and the dilution increase, and is likewise 
 increased by insolation. At 25° C, Goodwin found 
 that for the concentration i-io~* equilibriiun was 
 practically attained in three hours, whereas a week 
 was needed when the concentration was 6*i0"~* The 
 writer (107) carried out a series of measurements on 
 the light absorption in hydrosols of FegOg prepared 
 
EARLY INVESTIGATIONS. 105 
 
 by hydrolysis of FeClg at various dilutions, from 
 i-iQ-* to i-iQ-''. He found that the absorption- per 
 unit mass of FegOg first increases, and, after reaching 
 a maximum, decreases again, just as in the case 
 of other series of sols with an increasing degree of 
 dispersion. Some ultra-microscopic observations by 
 W. Biltz support the view that, when a disperse 
 system is formed by hydrolysis, the degree of 
 dispersion increases as the dilution increases (108) . 
 
 Gay-Lussac (109) (1810) and Berzelius (no) (1833) 
 were the first investigators to mention the formation 
 of peculiar " soluble " forms of substances by 
 hydrolysis. Berzelius describes the preparation of 
 " 6-silicic acid," i.e., the hydrosol of SiOg, thus : 
 " In its purest- state it is formed by oxidation of 
 silicon sulphide at the expense of water ; hydrogen 
 sulphide is set free, and the &-silicic acid dissolves in 
 the water. In concentrated form the solution soon 
 coagulates to a jeUy-like mass." The formation of 
 " pseudo-solutions " by hydrolysis of, acetates was 
 studied by Crum (in) (1853), Pean de Saint- 
 Gilles (112) (1855), and also by Graham (113) (1861) 
 in his fundamental series of colloid investigations. 
 Graham carried out the hydrolysis in a dialyzer, and 
 thus accelerated the reaction by removing the free 
 acid. Later investigators have followed him in that 
 path. In the second half of the nineteenth century 
 a great many experiments concerning the preparation 
 of hydrosols by hydrolysis have been carried out on 
 various salts, and on other compounds too. A 
 method employed by Grimaux is especially interest- 
 ing on account of the absence of inorganic products 
 except the colloids (114). He prepared FejOg 
 hydrosol by mixing ferriethylate with water, and 
 SiOg hydrosol by heating an aqueous solution of 
 the methylester of silicic • acid. The methylalcohol 
 contained in the sihcic acid hydrosol thus obtained 
 may be driven off by boihng. 
 
io6 THE FORMATION OF COLLOIDS. 
 
 A great many other double decomposition reactions 
 have been used for the preparation of disperse 
 systems. Among these are some of the classical 
 procedures of Graham, first described in his famous 
 paper, " Liquid Diffusion applied to Analysis," of 
 1861. He prepared " soluble silicic acid " by 
 decomposing silicate of soda with hydrochloric acid 
 and dialyzing for 24 hours. Many oxides were 
 precipitated from their salts as coagula, and then 
 peptized with some suitable salt and dialyzed, 
 e.g., AI2O3 with AICI3, FcgOg with FeClg, CrgOg with- 
 CrClg, SnOg with SnCl4, etc. 
 
 Lottermoser studied the formation of silver halides 
 in dilute solution (115). If n/ioAgNOg is slowly 
 'poured into n/ioKJ, an AgJ-hydrosol is formed and 
 remains stable until no excess of KJ is present, when 
 coagulation takes place. Inversely, if n/ioKJ is 
 poured into n/ioAgNOj, an AgJ-hydrosol is likewise 
 formed, and remains stable until all excess of AgNOg 
 is removed. The two forms of AgJ-hydrosols thus 
 obtained carry electric charges of contrary signs, 
 the former being negative and the latter positive. 
 
 The hydrosols of the ferrocyanides can be prepared, 
 according to Zsigmondy (116), in an analogous way, 
 viz., by pouring the dilute metal salt solution into the 
 dilute solution of potassium ferrocyanide, or in- 
 versely, taking care that there should be excess of 
 the solution reacted upon when the operation is 
 finished. 
 
 Many substances not to be obtained as hydrosols 
 because of their great solubility in water may be 
 prepared as organosols. Some rather interesting 
 methods of this kind have been worked out by 
 Paal (117). Michael had observed (118) that chlor- 
 acetic ester, when brought into reaction with sodium 
 malonester, acetoacetic ester and their simple alkyl 
 substitution products dissolved in benzene gave rise 
 to a yellowish, slightly opalescent liquid. As no 
 
DISPERSION PROCESSES. 107 
 
 sodium chloride was deposited, Michael believed that 
 some addition-prodiict, e.g., CiiHjgOgClNa, had been 
 formed. Paal now showed that the sodium chloride 
 existed free in the liquid in the form of a colloid. It 
 could be precipitated with petroleum ether as a gel 
 and redissolved in benzene as an almost pure 
 NaCl-sol, protected by a trace of some high mole- 
 cular organic substance formed as by-product in the 
 reaction. Paal and his students also obtained the 
 organosols of NaEr and NaJ by analogous methods. 
 
 Dispersion Processes. 
 
 As our knowledge of the mechanism of the forma- 
 tion of disperse systems has increased, an increasing 
 number of those processes formerly regarded as dis- 
 persion processes have proved to be in realitj' con- 
 densation phenomena, e.g., the formation of colloids 
 by means of the electric arc, by electrolysis, by ultra- 
 violet light, by boiling, etc. As pure dispersion 
 phenomena there only remain a few processes. Three 
 groups may be distinguished, viz. : 
 
 1. Partial dissolution. 
 
 2. Grinding. 
 
 3. Emulsification. 
 
 The formation of disperse systems by partial 
 dissolution or the raising of the degree of dispersion 
 of a low disperse system by the same means has been 
 studied by P. P. v. Weimarn (119). As has already 
 been mentioned (p. 55), a highly disperse sulphur 
 alcosol prepared by the rapid cooling of a solution of 
 sulphur in alcohol, when warmed, has its degree of 
 dispersion first lowered in consequence of the growth 
 of the larger particles at the cost of the smaller ones, 
 and then raised in consequence of the resolution of 
 the particles in the dispersion medium. If tellurium 
 is dissolved in potassium hydroxide solution, and the 
 solution is then diluted {e.g., 5 c.c. per 1,000 c.c. HgO), 
 
io8 THE FORMATION OF COLLOIDS. 
 
 a coarse suspension of tellurium is obtained. If air 
 or oxygen is passed through the liquid the tellurium 
 particles are oxidized and partly dissolved, most 
 probably because of the formation of potassium 
 tellurate. Hence the degree of dispersion increases 
 (P. P. V. Weimarn). Starting with a pure solution 
 of potassium tellurate, we may, inversely, as was 
 already shown by Berzelius (120) (1834), precipitate 
 tfie tellurium in the form of fine particles by passing 
 oxygen through it. When further treated with oxy- 
 gen these particles will be partially oxidized and 
 redissolved, i.e., the degree of clispersion is diminished. 
 Thus in this case wp get by the same means, viz., by 
 treating with oxygen, first a condensation process — 
 formation of colloid Te by oxidation of KgTe : 
 
 KaTe + O + H2O = Te + 2KOH, 
 
 and then a dispersion process, i.e., partial dissolution 
 of the Te-particles according to tMreactidn : 
 
 Te + 2KOH +.O2 = KgTeOg + H^O. ,. 
 
 Those processes have not yet become of any prac- 
 tical importance for the preparation of colloid 
 solutions. 
 
 As stated above, the so-called peptization pro- 
 cesses are in general to be considered as reversible 
 coagulations. .They are not real dispersion processes, 
 the degree of dispersion not being altered by the 
 peptization,, but. only, the mutual distances of the 
 particles. There may, however, exist peptizations 
 where a partial dissolution of the particles is of a 
 certain importance; thiis, for instance, according 
 to P. P.v. Weimarn (121), the peptization of AljOg 
 by alkalis (formation of aluminate) and of SiOj by 
 alkalis (formation of silicate), etc.. 
 
 Anintermediate position is occupied by the coHoida- 
 tion methods of Wedekind (122) and Kuzel (123). 
 The mechanism of these processes is not yet quite 
 
KUZELS SOLS. 109 
 
 clear, but it may most probably consist in a sort of 
 combination of a real dispersion process and an 
 ordinary peptization. According to Wedekind, metal- 
 lic zirconium is prepared by the reduction of some 
 suitable zirconium compound, e.g., zirconium dioxide 
 with magnesium or zirconium potassium fluoride 
 with potassium, and the product obtained is treated 
 with hydrochloric acid. If the mass is then washed, 
 with water, part of the zirconium goes into solution 
 as a colloid. The reduction product may again be 
 treated ■with acid and it -vtrill, when washed, give oS a 
 new portion of zirconium colloid. Kuzel's method 
 is carried out in the following way. The metal to be 
 brought into the colloid state; e.g., tungsten, moly- 
 bdeniim, silicon, zirconium, titanium, is first pul- 
 verized very finely by grinding, is then treated alter- 
 nately with acid and alkali, and washed with water 
 between these treatments. No doubt the coUoida- 
 tion depends here to som.e extent on a real dissolu- 
 tion of the particles by the acid and alkali, but in the 
 opinion of the author it may also to a considerable 
 extent be the result of an ordinary peptization, i.e., 
 a separa;tion of the aggregated particles. In the ease 
 of pure mechanical grinding it is generally possible, 
 as wiU be shown later on, to get colloids if the par- 
 ticles are only preveiited from aggregating too 
 closely together; This- fact is obviously in favour of 
 the view just put fortb, that the chemical reagents 
 in Wedekind's and Kuzel's methods only serve to a 
 great extent to bring about the loosening of the 
 particles from one another. Kuzel's coUoids were 
 used in the manufacture of filaments for electric 
 incandescent lamps. The sol was coagulated with 
 some suitable electrolyte, and the plastic mass 
 obtained pressed through fine holes. The wire thus 
 formed was then heated so as to get the particles to 
 sinter together to a solid wire. 
 
no THE FORMATION OF COLLOIDS. 
 
 It may be of interest to note that Wedekind's 
 method goes back to some old observations made by 
 Davy in 1808 (124) and by BerzeUus in I824 (125). 
 Davy found that boron, prepared by reduction of 
 boric acid with potassium, is redissolved when washed 
 with water : " The solutions obtained, when passing 
 through a filter, had a faint olive tint, and contained 
 sub-borate of potash and potash. In cases when, 
 instead of water, a weak solution of muriatic acid 
 was used for separating the saline matter from the 
 inflammable matter, the fluid came through the 
 filter colourless." 
 
 The preparation of disperse systems by means of 
 purely mechanical grinding is, as yet, only very little 
 studied. It seems, however, that the view held by 
 earlier investigators, that only rather coarse disperse 
 systems cotdd be obtained in this manner, cannot 
 be maintained. On the contrary, recent work 
 points to the conclusion that, under suitable con- 
 ditions, it will generally be possible to produce 
 rather highly disperse systems by purely mechanical 
 subdivision. On grinding a substance there are 
 most probably formed, besides the coarse particles, 
 very fine ones as well ; but as the grinding goes on 
 these coalesce and form aggregates, which appear as 
 coarse particles. As in the case of many chemical 
 coUoidation processes, a secondary degree of 
 dispersion is taken for the primary one. If the 
 aggregation is prevented, we get the primary 
 structure, i.e., highly disperse systems. 
 
 P. P. V. Weimarn suggested the use of some 
 indifferent solid dilution matter, to be mixed during 
 the grinding with the substance to be sub- 
 divided (126). If the dilution matter and the 
 substance be chosen so as to allow the former to be 
 dissolved while leaving the particles of the substance 
 unaffected when put into some liquid dispersion 
 medium, it will be possible to obtain a colloid solution 
 
GRINDING. Ill 
 
 of the latter. He pointed out further that, by 
 means of successive dilution and grinding with the 
 solid dilution matter, it should be possible to prepare 
 a series of coUoids with successively increasing degrees 
 of dispersion. N. Pihlblad has shown this to be 
 experimentally realizable (127). He ground sulphur 
 with urea in an agate mortar ; first o*i gm. S + i gm. 
 urea, then o-i gm. of this mixture + i g™- urea, and 
 so on. On dissolving the ground products in water, 
 he obtained a 'series of sulphur hydrosols with 
 increasing degree of dispersion. Similar experiences 
 were made when grinding in the same way the 
 dye-stuff " anihnblau 2 B " (from A.G.F.A., Berhn), 
 which is insoluble in water. A. L. Stein, a student 
 of V. Weimarn (128), is said to have carried out 
 many such experiments, but as yet we are without 
 any detailed information regarding the nature of the 
 dilution matter, etc. 
 
 G. Wegelin (129) has succeeded in preparing rather 
 highly disperse systems by simply grinding the pure 
 substances with or without water, e.g., silicon, 
 antimony, tungstic acid, titanic acid, molybdic 
 acid, melted vanadic acid, vanadic trioxide. On the 
 other hand, he had no success with copper, bismuth, 
 tellurium, selenium, sulphur, graphite, CuO, PbOg, 
 ZnO, CoO, FegOj, CaFg, violet chromic chloride. 
 We may suspect, however, that, especially in the 
 case of the very easily dispersed vanadic acid, some 
 chemical reaction plays a part in the process. The 
 question deserves further examination (130). 
 
 The formation of disperse systems by emulsiiica- 
 tion is important, both from a theoretical and from 
 a practical point of view. The investigations by 
 I. Nordlund (131) have shown that in the case of 
 emulsification-we have to distinguish between two 
 different processes, viz. : — 
 
 1. Crushing. of drops (dispersion process) . 
 
 2. Formation of lamellae (dispersion process) and 
 
112 THE FORMATION OF COLLOIDS. 
 
 bursting of the lamellae, small drops being thereby- 
 produced (condensation process). 
 
 When a disperse system is prepared by the shaking 
 up of two non-miscible hquids whose interfacial 
 tension is high, it is mainly process 2 that occurs. 
 
 o-o 
 
 ^ 
 
 Fig. 21. 
 
 A B 
 
 -Two diflferent methods of dispersing mercury. 
 (According to Nordlund.) 
 
 Nordlund demonstrated this fact by the following 
 experiments (Fig. 21) : — 
 
 In experiment " A," water (mixed with potassium 
 citrate to a concentration of 2*5 . lo"^ normal, in 
 order to preserve the primary degree of dispersion 
 and prevent coagulation) in the form of bubbles is 
 
MERCURY LAMELLA. 113 
 
 forced to rise through mercury, producing very thin 
 mercury lamellae when reaching the surface of the 
 mercury ; in experiment " B," mercury is ejected 
 through a fine glass tube against . a glass wall 
 under water (mixed with potassium citrate as 
 above). In the former case a comparatively rich 
 and very highly disperse mercury hydrosol is 
 obtained ; in the latter, only a very dilute and 
 coarsely disperse hydrosol. Thus, by the bursting 
 of the mercury lamellae, a highly disperse system 
 is obtained ; on the other hand, by breaking up 
 mercury drops against a wall, only a slightly disperse 
 system results. The ultra-microscopic test showed 
 that the particles were amicroscopic (with arc-light 
 illumination and with a " slit ultra-microscope ") 
 and smaller than the particles in any mercury 
 sol yet known. The author has carried out Nord- 
 lund's experiments with the system oil-water, and 
 found that in this case no clear difference appears 
 between " A " and " B," very likely because of 
 the low surface tension of the oil compared with 
 mercury. 
 
 In the case of oils and fats, and when dealing with 
 substances with low surface tension in general, 
 rather highly disperse systems are obtained by the 
 disruption of drops. Thus, if a coarse fat emulsion 
 is ejected against a hard wall, the fat droplets are 
 disintegrated and, accordingly, the degree of disper- 
 sion is increased. The so-caUed homogenization of 
 milk is carried out in this manner (132). Hot milk 
 is forced between iron plates at a pressure of 
 about 200 atmospheres ; a reduction of the size 
 of the fat drops to about one-tenth is attained. 
 When fats or hydrocarbons are emulsified with 
 water, a Httle alkali (in the case of glycerides) or 
 soap is usually added, in order to lower the surface 
 tension (or interfacial tension) of the dispersion 
 medium, thus facilitating the disintegration of the 
 
114 THE FORMATION OF COLLOIDS. 
 
 drops and preventing the small droplets formed 
 
 from uniting again (133). 
 
 Hatschek, who has carried out a 
 series of valuable investigations on 
 emulsions (134), has given many 
 good hints concerning the prepara- 
 tion of emulsions by mechanical 
 dispersion. An automatic emul- 
 sifying apparatus devised by 
 Hatschek is shown in Fig. 22 
 (135). The cylinder C is filled 
 to about one-third with a suitable 
 soap solution, e.g., 1 per cent, 
 sodium oleate, the oil to be emul- 
 sified is allowed to flow very 
 slowly down from the pipette P 
 through the funnel F and tube 
 T, while the ball- tube B is con- 
 nected to a filter pump. Pro- 
 vided the flow of oil be slow 
 enough, the oil will form but a 
 thin film on the inner wall of the 
 tube T, and the air passing out 
 at the lower edge of the tube 
 will break it up into fine droplets. 
 The air current at the same 
 time causes a continuous thorough 
 stirring, thus permitting the emul- 
 sification to go on automatically 
 
 throughout the whole liquid. The percentage of 
 
 disperse phase may rise to 70 or over. 
 
 Fig. 22. — Apparatus 
 for the emulsifying 
 of oils and hydro- 
 carbons. (Accord- 
 ing to Hatschek.) 
 
 Bibliography. 
 
 1. v. Weimarn, Dispefsoidchemie, Dresden, p. 70 
 (1911). 
 
 2. Wo. Ostwald and v. Weimarn, Koll. Zeitschr, 
 6, 181 (1910). 
 
BIBLIOGRAPHY. 115 
 
 3. Svedberg, English Patent No. 12108 (1908). 
 
 4. V. Weimarn, Dispersoidchemie, p. 67. 
 
 5. Reed, Journ. Franklin Inst., 139, 283 (1895). 
 Haber, Zeitschr. anorg. Chem., 16, 438 (1898). Bredig 
 and Haber, Ber. deutsch. chem. Ges., 31, 2741 (1898). 
 Haber, Zeitschr. Elektrochem., 6, 40 (1900) ; 8, 245 
 (1902); II, 660 and 828 (1905). 
 
 6. Malfitano, Comptes rend., 139, 1221 (1904). 
 Bechhold, Zeitschr. phys. Chem., 60, 257 (1907) ; 64, 
 .328 (1908). 
 
 7. Graham, Phil. Trans., 151, 199 (1861). 
 
 8. See, e.g., Oden, Nova acta, Upsala (4), 3, No. 4, 
 p. 66 (1913). 
 
 9. Jordis, Zeitschr. Elektrochem., 8, 677 (1902L 
 Zsigmondy and Heyer, Zeitschr. anorg. Chem., 68, 
 169 (1910). Glenny and Walpole, Biochem. Journ,., 
 9, 284 and 298 (1915). Wegelin, Koll. Zeitschr., 18, 
 225 (1916). Hatschek, Laboratory Manual, p. 16. 
 
 10. E. Fischer, Ber. Deutsch. chem. Ges., 39, 530 
 (1906). 
 
 loa. See Zsigmondy, Kolloidchemie, p. 315 (1918). 
 
 11. See Zsigmondy, Kolloidchemie, p. 307 (1918). 
 
 12. Cf. Svedberg, Herstellung kolloider Losungen, 
 p. 222. 
 
 13. Stoeckl and Vanino, Zeitschr. phys. Chem., 30, 
 98 (1899). 
 
 14. Nordenson, Ueher die Bedeutung des Lichtes 
 fur die Bildung und Stabilitdt kolloider Losungen, 
 Diss., Upsala, p. 107 (1914). 
 
 15. Reitstotter, Kolloidchem. Beih., 9, 248 (1917). 
 Zsigmondy, Kolloidchemie, p. 148 (1918). 
 
 16. Doerinckel, Zeitschr. anorg. Chem., 63, 344 
 (1909). Westgren, Die Brownsche Bewegung, Diss., 
 Upsala, p. 68 (1915). 
 
 17. Borjeson, Koll. Zeitschr., 27, 18 (1920). 
 
 18. Naumoft, Zeitschr. anorg. Chem., 88, 43 (1914)- 
 
 19. Vanino and Hartl, Koll. Zeitschr., i, 272 (1907). 
 
 20. Nordenson, Diss., loc. cit., p. 114. 
 
 8—2 
 
ii6 THE FORMATION OF COLLOIDS. 
 
 21. Zsigmondy, Kolloidchemie, p. 143 (1918). 
 
 22. Hiege, Zeitschr. anorg. Chem., 91, 145 (1915). 
 
 23. Gutbier, Zeitschr. anorg. Chem., 31, 448 (1902). 
 
 24. Svedberg, Koll. Zeitschr., 5, 318 (1909). 
 
 25. Nordenson, Diss., loc. cit., p. 122. 
 
 26. Faraday, Phil. Trans., i/^'j, 159 (1857). 
 
 27. Zsigmondy, Zur Erkenntnis der Kolloide, p. 100 
 (1905). 
 
 28. Zsigmondy, Zur Erkenntnis der Kolloide, pp. 
 136 and 173 (1905). Svedberg, Koll. Zeitschr., 4, 
 168 (1909) ; Die Existens der Molekiile, p. 10 (1912)! 
 Biltz, loc. cit. . 
 
 29. Zsigmondy, Zur Erkenntnis der Kolloide, p. 128 
 
 (1905)- . 
 
 30. Siedentopf and Zsigmondy, Annul. Phys. (4), 
 10, I (1903). 
 
 31. Tammann, Zeitschr. phys. Chem., 25, 441 (1898) ; 
 Zeitschr. Elektroch., 10, 532 (1904). 
 
 32. See, e.g., Vanino, Zeitschr. prakt. Chem. (2), 
 73> 575 (1906). Svedberg, Herstellung kolloider 
 Losungen, p. 14 (1909). Wi. Ostwald, Koll. 
 Zeitschr., 4, 5 (1909). 
 
 33. See abstract in Annal. de Chim., 36, 58 (1798). 
 
 34. Oberkampf, Annal. de Chim., 80, 140 (1811). 
 
 35. Faraday, Phil. Trans., 147, 145 (1857). 
 
 36. Kohlschiitter, Zeitschr. Elektroch., 14, 49 
 (1908)'; Koll. Zeitschr., 12, 285 (1913). 
 
 37. Kohlschiitter and Fischmann, Liebigs Annal., 
 387. 86 (1911). 
 
 38. Zsigmondy and Lottermoser, Zeitschr. phys. 
 Chem., 56, 77 (1906). 
 
 39. Liippo-Cramer, Koll. -Zeitschr., 7, 99 (1910). 
 
 40. Carey Lea, Amer. Journ. Science (3), 37, 476 
 (1889). 
 
 41. Od6n, Zeitschr. phys. Chem., 78, 682 (1912). 
 
 42. Kohlschiitter, Koll. Zeitschr., 12, 289>(i9i3).' 
 
 43. Billiter, Ber. Deutsch. chem. Ges., 35, 1929 
 (1902). 
 
BIBLIOGRAPHY. 117 
 
 44. Blake, Amer. Journ. Science (4), 16, 431 (1903). 
 
 45. Nordenson, Kolloidchem. Beih., 7, 130 (1915). 
 
 46. Svedberg, KoU. Zeifschr., 6, 129 (1910). 
 
 47. Traube-Mengarini and Scala, Koll. Zeitschr., 6, 
 65 and 240 (1910) ; 10, 113 (1912). 
 
 48. Nordenson, Kolloidchem. Beih., 7, 91 (1915). 
 
 49. Svedberg, Herstellung koUoider Losungen, p. 13 
 {1909). 
 
 50. Fulhame, lac. cit. 
 
 51. Oberkampf, loc. cit. 
 
 • 52. Vanino, Ber. Deutsch. chem. Ges., 38, 463 
 
 (1905)- 
 
 53. Kohlsdhiitter, Zeitschr. Elektrochem., 14, 49 
 (1908). 
 
 54. Zsigmondy, Liebigs Annal., 301, 29 (1898). 
 
 55. Fischer, Annal. Phys. (2), 9, 255 (1827). 
 
 56. Lottermoser, J own. prakt. Chem. (2), 57, 484 
 (1898). ■ 
 
 57. Billiter, loc. cit. 
 
 58. Gutbier, Zeitschr. anorg. Chem., 32, 51, 91, 106 
 and 347 (1902) ; 42, 177 (1904). Gutbier and 
 Hofmeier, Journ. prakt. Chem. (2), 71, 358 and 452 
 
 (1905)- 
 
 59. Paal, Ber. Deutsch. chem. Ges., 35, 2224 and 
 
 2236 (1902) ; 37, 124 (1904) ; 38, 326 and 534 and 
 1398 (1905) ; 39, 1550 (1906) ; 40, 1392 (1907) ; 50, 
 722 (1917). 
 
 60. Rose, Annal. Phys. u. Chem. (2), 14, 183 (1828). 
 
 61. Dobereliner, Ann. Pharm., 2, 4 (1832). 
 
 62. Woliler, Annal. Phys. (2), 46, 629 (1839). 
 
 63. Newbury, Chem. News, 54, 57 (1886). 
 
 64. Muthmann, Ber. Deutsch. chem. Ges., 20, 983 
 (1887). ■ ■ 
 
 65. Sebni, Annali del Majocchi, 15, 88, 212 and 235 
 (1844) ; Ann. Chim. et phys. (3), 28, 210 (1850). 
 
 66. 'Engel,' Compt. rend., 112, 866 (1891). 
 
 67. Raffo, Koll. Zeitschr., 2, 358 (1908). 
 
 68. Oden, Nova acta, Upsala (4), 3, No. 4 (1913)- 
 
ii8 THE FORMATION OF COLLOIDS. 
 
 69. Wackenroder, Arch. d. Pharm., 47, 272 (1846) ; 
 48, 140 (1846). 
 
 70. Selmi, loc. cit. 
 
 71. Debus, Journ. Chem. Soc, 53, 278 (1888). 
 
 72. Spring, Rec. des trav. chim. des Pays-Bas, 25, 
 253 (1906). 
 
 73. Lewes, Journ. Chem. Soc, 41, 300 (1882). 
 
 74. Shaw, Journ. Chem. Soc, 43, 351 (1883). 
 
 75. Hertlein, Zeitschr. -phys. chem., 19, 287 (1896). 
 
 76. Miiller, Zeitschr. Elektrochem., 11, 521 and 931 
 (1905) ; Ber. Deutsch. chem. Ges., 38, 3779 (1905). . 
 
 77. Scheele, Efterlemnade href och anteckningar utg. 
 af Nordenskiold, Stockholm, pp. 416 and 448 (1892). 
 
 78. Bergman, Kleine Physische und Chemische 
 Werke, Frankfurt am Mayn, i, 342 (1782). 
 
 79. Le Veillard, Crell's Chem. Annul., i, 440 (1789). 
 
 80. Berthollet, Annul. Chim., 25, 233 (1798). 
 
 81. Berzelius, Lurohok i kemien, Stockholm, i, 120 
 (1808). 
 
 82. Dobereiner, Schweiggers Journ., 8, 400 (1813). 
 
 83. Wackenroder, loc cit. 
 
 84. Selmi, loc. cit. 
 
 85. Debus, loc cit. 
 
 86. Wa. Ostwald, Koll. Zeitschr., 15, 204 (1914). 
 
 87. See Acheson Journ. Frunklin Inst., 164, 375 
 (1907). Thome, Journ. Chem. Soc, 109, 202 (1916). 
 
 88. R. Lorenz, Zeitschr. unorg. Chem., 91, 46, 57 
 and 61 (1915) ; Koll.. Zeitschr., 18, 177 (1916). 
 
 89. Siedentopf, Phys. Zeitschr., 6, 855 (1905). 
 
 90. Doelter, Rudium und die Farhen, Dresden 
 (1910). 
 
 91. V. Weimarn, Zur Lehre von den Zustdnden der 
 Muterie, p. 164 (1914). 
 
 92. Oden, Arkiv for kemi, Stockholm, 7, No. 26 
 (1920). 
 
 93. Berzelius, Lehrb. der Chem., 5 Aufl., 2, 269 
 (1844). 
 
 94. Schulze, Journ. frukt. Chem. (2), 25, 431 (1882). 
 
BIBLIOGRAPHY. 
 
 119 
 
 95. Picton and Linder, Journ. Chem. Soc, 61, 137 
 (1892) ; 67, 63 (1895). 
 
 96. Svedberg, Die Existenz der Molekule, p. 27 
 (1912). 
 
 97. Borjeson, Koll. Zeitschr., 27, 18 (1920). 
 
 98. Linder and Picton, Journ. Chem. Soc, 61, 114 
 (1892). 
 
 99. Lottermoser, Journ. prakt. Chem. (2)^ 75, 293 
 (1907). 
 
 100. Berzelius, Lehrbuch der Chem., 3 Aufl., 3, 209 
 
 (1834). 
 loi. Schiilze, Journ. prakt. Chem. (2), 27, 320 
 
 (1883). 
 
 102. Winssinger, Bull. Acad. Belg. (3), 15, 390 
 (1888). 
 
 103. Paal and Voss, Ber. Deutsch. chem. Ges., 37, 
 3862 (1904). 
 
 104. Berzelius, Lehrb. der Chem., 3 Aufl., 3, 126, 
 2og, 222 and 439 (1834). 
 
 105. Spring, Chem. News, 48, loi (1883). 
 
 106. Goodwin, Zeitschr. phys. Chem., 21, i (1896). 
 
 107. Svedberg, Die Existenz der Molekiile, p. 24. 
 
 108. Biltz, Nachr. Ges. Wiss., Gottingen, Math.- 
 Phys. Kl., 1906, p. 141. 
 
 109. Gay-Lussac, Annal. de Chim. 74, 193 (1810). 
 no. Berzelius, Lehrb. der Chemie, 3 Aufl., 2, 122 
 
 (1833). 
 
 111. Crum, Journ. Chem. Soc, 6, 217 (1853). 
 
 112. Pean de Saint-Gilles, Compt. rend., 40, 568 
 
 (1855). 
 
 113. Graham, Phil. Trans., 151, 183 (1861). 
 
 114. Grimaux, Compt. rend., 98, 105 and 1434 
 (1884). 
 
 115. Lottermoser, Journ. prakt. Chem. (2), 72, 39 
 
 (1905); 73,374(1906). 
 
 116. Zsigmondy, Kollotdchemte, p. 300 (1918). 
 
 117. Paal and Kiihn, Ber. Deutsch. chem. Ges., 39, 
 2859, 2863 (1906) ; 41, 51. 58 (1908). 
 
120 THE FORMATION OF COLLOIDS. 
 
 iiS. Michael, Ber. Deuisch. chem. Ges., 38, 3217 
 
 (1905)- 
 
 119. V. Weimarn, Dispersoidchemie, pp. 69 and yy 
 (1911). 
 
 120. Berzelius, Annul. Phys. (2), 32, i (1834). 
 
 121. V. Weimarn, Dispersoidchemie, p. 79. 
 
 122. Wedekind, Zeitschr. Elektrochem., 9, 630 
 (1903). . ^ 
 
 123. Kuzel, Osterreich. Patent, Kl. 12&, No. A 2573 
 (1906). Lottermoser, Koll. Zeitschr., 2, 347 (1908). 
 
 124. Davy, Phil. Trans., 1809, p. 78. 
 
 125. Berzelius, Vetenskapsacad. Handl., Stock- 
 holm, 1824, p. 89. 
 
 126. V. Weimarn, Dispersoidchemie, p. 82. 
 
 127; Pihlblad, Lichtabsorption und Teilchengrosse, 
 Diss., Upsala, p. 46 (1918). 
 
 128. V. Weimarn, Zur Lehre von den Zustdnden der 
 Materie, p. 183 (1914). 
 
 129. Wegelin, Koll. Zeitschr., 14, 65 (1914). 
 
 130. See Bancroft, Brit. Assoc. Report, Sect. B, p. 6 
 (1918). 
 
 131. Nordlund, Quecksilberhydrosole, Diss., Upsala, 
 p. 18 (1918) ; Koll. Zeitschr., 26, 121 (1920). 
 
 132. G. Buglia, Koll. Zeitschr., 2, 353 (1908). 
 B. Spear, in Zsigmondy's Chemistry of Colloids, 
 p. 265. 
 
 133. Donnan, Zeitschr. phys. Chem., 31, 42 (1899). 
 Pickering, Koll. Zeitschr., 7, 11 (1910). Schlaepfer, 
 Journ. Chem. Soc, 113, 522 (1918). 
 
 134. See, e.g., Brit. Assoc. Report (1918). 
 
 135. Hatschek, Laboratory Manual, p. 63 (1920). 
 
NAME INDEX. 
 
 Acheson, 91, 118 
 Andr^n, 47, 53 
 
 Bancroft, 120 
 
 Barns, 53 
 
 Bechhold, 115 
 
 Benedicks, 34, 52 
 
 Bergman, 89, 118 
 
 BerthoUet, 89, ii8 
 
 Berzelius, 9, 13, 89, loi, 103, 
 
 105, 108, 110, 118, 119, 120 
 Billiter, 80, 82, 116, 117 
 Biltz, 14, 105, 116, 119 
 Blake, 80, 117 
 Borjeson, 30, 33, 37, 52,'63,'64, 
 
 102, 115, 119 
 Bredig, 24, 25, 52, 115 
 Buglia, 120 
 Bunsen, 92 
 Burton, 25, 52 
 
 Carey Lea, 79, 83, 116 
 Crum, 105, 119 
 
 Davy, no, 120 
 Debus, 85, 90, 118 
 Degen, 25, 52 
 
 Dobereiner, 84, 89, 117, 118 
 Doelter, 93, 118 
 Doerinckel, 83, 115 
 Donnan, 120 
 
 Ehlers, 51 
 
 Ehrenhaft, 22, 42, 51 
 Engel, 84, 117 
 
 Faraday, 9, 14, 41, 
 
 76. 83, 92, 116 
 Fischer, 82, 117 
 Fischer, E., 59, 115 
 
 52, 71. 75. 
 
 Fischmann, 116 
 Fulhame, 75, 82, 117 
 
 Gay-Lussac, 105, 119 
 Glenny, 115 
 Goodwin, 104, 119 
 Graham, 9, 10, 13, 59, 90, 105, 
 
 106, 115, H9 
 Grimaux, 105, 119 
 Gutbier, 83, 116, 117 
 
 Haber, 56, 115 
 
 Hamburger, 15, 16, 18, 19, 21 
 
 Hartl, 115 
 
 Hatschek, 114, 115, 120 
 
 Hertlein, 85, 118 
 
 Heyer, 115 
 
 Hiege, 67, 69, 116 
 
 Hofmeier, 117 
 
 HouUevigue, 18 
 
 Jordis, 115 
 Juncher, 75 
 
 Kasperowicz, 53 
 Kimura, 21, 51 
 Knudsen, 15, 18 
 Kohlschutter, 14, 19, 20, 21, 
 
 30. 51. 53. 77. 78. 79. 82, 83, 
 
 116, 117 
 Konstantinowsky, 52 
 Kiihn, 119 
 Kuzel, 108, 109, 120 
 
 Lacroix, 53 
 Lemery, 75 
 Lenard, 47, 53 
 Le Veillard, 89, 118 
 Lewes, 85, 118 
 Lidow, 52 
 
122 
 
 NAME INDEX. 
 
 Linder, 9, 14, loi, 102, 163, 
 
 119 
 Lorenz, 91, 92, 118 
 Lottermoser, 78, 82, 83, 103, 
 
 106, n6, 117, 119, 120 
 Liippo-Cramer, 78, 116 
 
 Macquer, 74 
 Malfitano, 115 
 van Marum, 41, 52 
 Mecklenburg, 14 
 Meyer, 42, 52 
 Michael, io6, 107, 120 
 Miiller, 88, 118 
 Muthmann, 84, 117 
 
 Naumoff, 65, 115 
 
 Newbury, 84, 117 
 
 Noll, 51 
 
 Nordenson, 21, 51, 61, 70, 71, 
 
 80, 81, 115, 116, 117 
 Nordlund, 21, 51, 52, iii, 112, 
 
 113, 120 
 
 Oberkampf, 75, 82, 116, 117 
 Od^n, 14, 72, 79, 85, 86, 87, 88, 
 
 94. 95. 96, 98, 99, 100, 115, 
 
 116, 117, 118 
 Osterhuis, 17 
 Ostwald, Wa, 91, 118 
 Ostwald, Wi., 75, 116 
 Ostwald, Wo., 10, 14, 114 
 
 Paal, 83, 103, 106, 107, 117, 
 
 119 
 Pean de Saint-Gilles, 105, 119 
 Pickering, 120 
 
 Pictou, 9, 14, loi, 102, 103, 119 
 Pihlblad, iii, 120 
 
 Raffo, 84, 117 
 Reed, 115 
 Reitstotter, 115 
 Richter, 75 
 Rose, H., 83, 117 
 
 Scala, 117 
 
 Scheele, 89, ii8 
 
 Schlaepfer, 120 
 
 Schoop, 49, 53 
 
 Schulze, 101, 103, 118, 119 
 
 Selmi, 9, 14, 84, 85, 89, 117, 
 
 118 
 Shaw, 85, 118 
 
 Siedentopf, 73, 92, 93, 116, 118 
 Spear, 120 
 Spring, 85, 118, 119 
 Stein, III 
 Stoeckl, 83, 115 
 Svedberg, 14, 52, 81, 115, 116, 
 
 117, 119 
 
 Tammann, 66, 74, 116 
 Thome, 118 
 Threlfall, 52 
 Tichomiroff, 52 
 Traube-Mengarini, 80, 81, 117 
 Tyndali, 49, 53, 102 
 
 Vanino, 65, 82, 83, 115, 116, 
 
 117 
 Voss, 119 
 
 Wackenroder, 85, 89, 90, 118 
 
 Walpole, 115 
 
 Wedekind, 108, 109, no, 120 
 
 Wegelin, in, 115, 120 
 
 V. Weimarn, 14, 55, 56, 94, 107, 
 
 108, no. III, 114, 115, 118, 
 
 120 
 Westgren, 42, 64, 65, 115 
 Wilson, C. T. R., 13, 14, 44, 
 
 45. 52 
 Winssinger, 103, 119 
 Wohler, 84, 117 
 Wood, 16, 18 
 
 Zsigmondy, 10, 13, 14, 53. 63, 
 66, 67, 71, 73, 74, 78, 82, 83, 
 106, 115, 116, 117, 119, 120 
 
SUBJECT-MATTER INDEX. 
 
 Acheson graphite, 91 
 Adiabatic expansion, 12, 42 
 Alcosols, 10, 25, 27, 38, 54, 103, 
 
 107 
 Alkali metals, electric sols of, 
 38, 39 
 
 salts of fatty acids, sols of, 60 
 Alloys, electric pulverization 
 
 of. 33. 34 
 Aluminium, electric pulveriza- 
 tion of, 36, 40 
 oxide hydrosols, 106, 108 
 Antimony, electric pulveriza- 
 tion of, 36, 40 
 hydrosol, iii 
 Arsenic, condensation in carbon 
 dioxide, 19 
 hydrogen, 19 
 Arsenious sulphide sols, loi 
 
 degree of dispersion of, 102 
 Aurum potabile, 74 
 
 Barium sulphate sols, 94-100 
 Benzene hydrosol, 54 
 
 vapour, c ndensation of, 47 
 Bismuth, electric pulveriza- 
 tion of, 35, 36, 37, 40 
 
 reduction of, 81, 83 
 Blue rock salt, formation of, 
 
 92-93 
 Boron hydrosol, 110 
 
 Cadmium, condensation in hy- 
 drogen, 19 
 nitrogen, 19 
 vacuo, 16, 17 
 electric pulverization in alco- 
 hol, 33, 35, 37. 40 
 in ether, 23, 36 
 in nitrogen, 23 
 in water, 25 
 Calcium, electric pulverization 
 of, 36 
 
 Canal rays, vaporization of 
 
 metals by, 19 
 Carbon, condensation in vacuo, 
 16 
 disperse systems of, 16, 91 
 Cathode rays, action on rock 
 salt, 93 
 coloration of various sub- 
 stances by, 93 
 Chromic oxide hydrosols, 106 
 Cobalt, electric pulverization 
 
 of, 36 
 Condensation, by adiabatic 
 expansion, 12, 42 
 by cooling, in liquids, 54, 55 
 by formation of a new 
 
 element, 57 
 by lowering of solubility, 54, 
 
 55 
 by production of material of 
 
 disperse phase, 54 
 in gases, chemical reactions, 
 
 47. 48 
 influence of gas on degree 
 
 of dispersion, 19 
 photochemical reactions, 
 48 
 in liquids and in solids, 54 
 by chemical reaction, 57 
 in vacuo, action of salts, 18 
 critical temperature of, 15 
 degree of dispersion ■ of 
 
 condensate, 16, 17 
 shadow phenomenon, 16 
 primary conditions for, 1 1 
 with addition of nuclei, 58, 
 61 
 formation of nuclei, 58, 61 
 Copper, condensation in vacuo, 
 16 
 electric pulverization of, 36, 
 
 40 
 reduction of, 81, 83 
 
124 
 
 SUBJECT-MATTER INDEX. 
 
 Copper sulphide sols, 103 
 Crushing of mercury drops, 1 1 1 
 Cryosols, 54 
 
 Degree of dispersion, condi- 
 tions for maintaining the 
 primary, 58 
 methods for regaining the 
 
 primary, 59 
 primary, 12, 58 
 secondary, 12, 58 
 tertiary, 12, 58 
 Dialysis, 59, 105, 106 
 Disperse phase, 10 
 systems, 10 
 
 formation in gases, 19 
 liquids and in solids, 54 
 vacuo, 15 
 Dispersion, 11 
 
 primary conditions for, 12 
 medium, 10 
 processes in gases, 49 
 liquids, 107 
 Dispersity, 11 
 Dissociation, 57, 90 
 Double decomposition, 57, 93 
 degree of dispersion and ex- 
 perimental conditions, 94 
 
 Electric arc, anode thermally 
 
 protected, 30 
 condensation product, 24 
 direct current, 22 
 dispersion product, 24 
 electrical conditions, 22 
 enclosed, 22, 25 
 free, 22 
 
 high-frequency current, 22 
 influence of gas in enclosed 
 
 arc, 30 
 intermittent current, 22 
 low-frequency current, 22 
 products of, 23 
 
 damped oscillatory, 32 
 
 undamped oscillatory, 40, 
 
 41 
 pulverization, arrangement 
 for continuous current, 
 
 24 
 arrangement for damped 
 oscillatory current, 32 
 
 Electric arp, pulverization, 
 arrangement for un- 
 damped oscillatory cur- 
 rent, 31 
 arrangement with enclosed 
 
 arc, 28, 29 
 electrical conditions, 24, 
 
 34. 35 
 in gases, 22 
 in liquids, 22 
 of alloys, 33 
 thermal protection of, 25 
 vaporization of metals by, 21 
 Emulsification, 107 
 apparatus for. 114 
 of mercury, 1 1 1 
 of oils and fats, 113 
 Etherosols, 38, 39, 103 
 Ethyl alcohol vapour, con- 
 densation of, 47 
 
 Fats, emulsification of, 113 
 Ferric oxide hydrosols, 104 
 Ferrocyanide hydrosols, 106 
 Ferrous citrate, reducing agent 
 
 in silver reduction, 79 
 Formaldehyde, reducing agent 
 
 in gold reduction, 65, 67 
 
 Gelatin, - swelling and redisso- 
 
 lution, 60 
 Gold, condensation in vacuo, 16 
 
 electric pulverization of, 25, 
 
 35. 36, 40 
 hydrosols, degree of disper- 
 sion of, 66-70, 71-72 
 historical notes, 74-77 
 purification of, 65 
 reduction of, 13,' 60-77, 83 
 action of light, 70-7 1 
 influence of various 
 substances added, 67- 
 69 
 mechanism of, 61-62, 
 
 65-66 
 oil nuclei, 63 
 ruby glass, 73-74, 75 
 Grinding, 107, no, in 
 and peptization, 109 
 with solid dilution matter, 
 no 
 
SUBJECT-MATTER INDEX. 
 
 125 
 
 Halides, dissociation of, 91 
 Hydrazine chlorhydrate as re- 
 ducing agent, 83 
 hydrate as reducing agent, 
 83 
 Hydrogen, reducing agent in 
 silver reduction, 77 
 peroxide, reducing agent in 
 
 gold reduction, 61-64 
 selenide, oxidation of, 84 
 sulphide, oxidation of, 84 
 Hydrolysis, 59, 93, 103 
 and degree of dispersion, 105 
 historical notes, 105 
 Hydrosol, 10 
 
 Hydroxylamin chlorhydrate as 
 reducing agent, 83 
 
 Ice-pentanosol, 54 
 Iridium, electric sols of, 25 
 Iron, condensation in vacuo, 16 
 electric pulverization of, 36 
 
 Lead, electric arc pulverization 
 of, 36, 40 
 electrolytic pulverization of, 
 56 
 Liquid lamellae, 50 
 Lysalbic acid as protective 
 colloid, 83, 103 
 
 Magnesium, condensation in 
 vacuo, 16, 17 
 electric pulverization of, 36 
 Magnetic field, application in 
 electric pulverization, 25, 26, 
 28, 29, 41, 42 
 Manganese, electric pulveriza- 
 tion of, 36 
 Melting globules, 23 
 Mercury, condensation in 
 vacuo, 15, 18 
 emulsification of, 111-113 
 injection of vapour into 
 
 water, 21 
 lamellae of, in 
 reduction of, 8r, 83 
 sulphide sols, 103 
 Metal wires, heated and dipped 
 into water, 21 
 heating in vacuo, 16-18 
 
 Metal wires, volatilization in 
 liquids, 41 
 
 Metals, determination of elec- 
 tric pulverization of, 36, 40 
 
 Methyl alcohol vapour, con- 
 densation of, 47 
 
 Molybdenum, condensation in 
 vacuo, 16 
 electric pulverization of, 36 
 hydrosols, 109 
 
 Molybdic acid hydrosols, in 
 
 Nickel benzosol, 91 
 
 carbonyl, dissociation of, 91 
 condensation in vacuo, 16 
 electric pulverization of, 36 
 
 Nuclei, 13 
 
 action of in gold reduction, 
 63-64 
 
 Oils, emulsification of, 113 
 Organic colloids, formation of, 
 59 
 
 compounds, dissociation of, 
 91 
 Organosols of halides, 106 
 
 ice, 54 
 
 metals, 25, 27, 38, 39 
 
 sulphides, 103 
 
 sulphur, 54, 107 
 Oxidation, 57, 84 
 Oxide sols, formation of, 93 
 
 Palladium, electric sols of, 25 
 Partial dissolution, 107 
 Peptization, 59, 106, 108 
 Phenylhydrazine chlorhydrate 
 
 as reducing agent, 83 
 Phosphorus, reducing agent in 
 gold reduction, 71, 76-77 
 
 sols, 56 
 Platinum, condensation in 
 vacuo, 16 
 
 electric pulverization of, 25, 
 
 35. 36, 37. 40 
 metals, reduction of, 81, 83 
 Potassium, electric sols of, 39 
 Propyl alcohol vapour, conden- 
 sation of, 47 
 Protalbic acid as protective 
 colloid, 83, 103 
 
126 
 
 SUBJECT-MATTER INDEX. 
 
 Protective colloids in gold 
 reduction, 67-70 
 silver reduction, 79 
 sulphide formation, 103 
 Proteins, formation of, 59 
 Pulverization of alloys by 
 water, 56 
 of metals, by boiling, 81 
 electrolytic, in fused salts, 
 92 
 solutions, 56 
 of sulphur, selenium and 
 
 tellurium, electrolytic, 88 
 mechanical, 49 
 
 by gas production in solids, 
 
 51 
 Purification of sols by electro- 
 lysis, 59 
 by filtration, 59 
 by sedimentation, 59 
 Pyrosols, 91-92 
 
 Radium rays, action on rock 
 salt, 93 
 coloration of various sub- 
 stances by, 93 • 
 reducing agent in gold reduc- 
 tion, 60 
 Reducing agents in gold reduc- 
 tion, 60 
 Reduction, 57, 60 
 
 measurement by means of 
 
 conductivity, 61 
 methods, classification of, 
 82-83 
 historical notes, 74-77, 
 83-84 
 Removal of electrolytes, 59 
 Reversible hydrosols, 83, 103 
 Rock salt, coloration of, 92-93 
 dissociation of, 93 
 
 Salt sols, 94 
 Sediment, 24 
 
 from damped oscillatory cur- 
 rent arc, 33, 34, 35 
 intermittent current arc, 
 
 31 
 
 low-frequency current arc, 
 
 31 
 
 Sediment from undamped oscil- 
 latory current arc, 41 
 Selenium, condensation in car- 
 bon dioxide, 19 
 hydrogen, 19 
 electrolytic pulverization of, 
 
 89 
 hydrosols, 56, 81, 83, 84, 89 
 reduction of, 81, 83 
 Silicic acid hydrosols, 105, 108 
 Silicon hydrosols, 109, m 
 Silk, coloration of by gold 
 
 solution, 75 
 Silver, condensation in vacuo, 
 15. 16 
 electric pulverization of, 25- 
 
 30, 36, 40 
 halide hydrosols, 106 
 hydrosols, degree of disper- 
 sion, 78-80 
 fractional coagulation, 79 
 reversible, 79 
 mirrors by reduction, 78 
 reduction of, 77-81, 83 
 by electrolysis, 79-780 
 by illumination, 80-81 
 influence of vessels, 77-78 
 on gold nuclei, 78 
 on silver nuclei, 78 
 spectrum of arc, 27, 28 
 vaporization by canal rays 
 in argon, 19 
 hydrogen, 19 
 nitrogen, 19 
 vapour, arrangement for pro- 
 duction of, 25, 26 
 condensation of, 26, 27 
 Size of particles, distribution 
 of primary, 12 
 primary, 12 
 Sodium amalgam as reducing 
 agent, 83 
 electric sols of, 39 
 Sols by damped oscillatory cur- 
 rent arc, 32, 34, 38, 39 
 enclosed arc, 27, 28, 29 
 free continuous current arc, 
 
 24. 25 
 intermittent current arc, 31 
 low-frequency current arc, 31 
 partly protected arc, 30 
 
SUBJECT-MATTER INDEX. 
 
 127 
 
 Sols by undamped oscillatory 
 current arc, 40 
 degree of dispersion of 
 electric sols, 37 
 Stannic oxide hydrosols, 106 
 Stannous salts as reducing 
 
 agent, 83 
 Submerged arc, 23 
 Sulphide sols, formation of, 93 
 Sulphur alcosols, 54 
 
 formation by partial reso- 
 lution, 107 
 electrolytic pulverization of, 
 
 88-89 
 hydrosols, 84-90 
 
 degree of dispersion and 
 ■ experimental conditions, 
 85-86 
 formation of, 56, 84-90 
 formation by grinding, i n 
 by ultra-violet illumi- 
 nation, 57 
 fractional coagulation of, 
 
 87 
 historical notes, 89-90 
 purification of, 88 
 vapour, condensation of, 42 
 Supersaturation and degree of 
 dispersion, 46 
 critical degrees of, 47 
 Surface condensation, 42, 48, 
 49, 54. 57 
 
 Tellurium, electrolytic pul- 
 verization of, 89 
 hydrosol, formation of by 
 partial resolution, 108 
 reduction of, 81, 83 
 ThaUium, electric pulverization 
 of, 36 
 
 Thioic acid, dissociation of, 90 
 Thiosulphates, decomposition 
 
 of, 84 
 Tin, electric pulverization of, 
 
 .35, 36, 37 
 Titanic acid hydrosols, 1 1 1 
 Titanium hydrosols, 109 
 Tungsten, condensation in 
 vacuo, 1 6, 17 
 
 hydrosols, 109 
 Tungstic acid hydrosols, iii 
 
 Ultrafiltration, 59 
 Ultra-violet light, reducing 
 agent in gold reduction, 70 
 silver reduction, 80 
 
 Vanadic acid hydrosol, iii 
 
 trioxide, iii 
 Volume condensation, 42, 48, 
 
 49, 54, 57 
 
 Washing by filtration, 59 
 
 sedimentation, 59 
 Water vapour, condensation 
 
 of, 42-47 
 Wave current, 32, 39 
 
 X-rays, reducing agent in 
 silver reduction, 80 
 
 Zinc, condensation in hydro- 
 gen, 19 
 nitrogen, 19 
 vacuo, 16, 17 
 electric pulverization of, 35, 
 
 36, 37, 40 
 sulphide sols, 103 
 Zirconium hydrosols, 109 
 
 THE WHITEFRIARS PRESS, LTD., LONDON AND TONBRIDGE, 
 
NEW BOOKS ON CHEMISTRY 
 
 THE FUNDAMENTAL PROCESSES OF DYE CHEMISTRY. By 
 H. E. Fierz-David, Professor of Chemistry at Federal ±lign 
 School, Zurich. , Translated by F. A. Mason, Ph.D , 
 Research Chemist, British Dyestuffs Corporation, Ltd. 
 With 45 Illustrations, including 19 Plates. Price 21s. net.- 
 
 A TEXT-BOOK OF PRACTICAL CHEMISTRY. By G. F. Hood, 
 M.A., B.Sc, Chemistry Master, Nottingham High School, 
 and J. A. Carpenter, M.A., late Science Master, St. Paul's 
 School. With 162 Illustrations. Price 21s. net. 
 
 TREATISE ON APPLIED ANALYTICAL CHEMISTRY. Edited 
 by Prof. V. Villavecchia. Translated by T. H. Pope, 
 B.Sc. Vol. I. With 58 Illustrations. 21s. net. Vol. II. 
 With 105 Illustrations. 25s. net. 
 
 TREATISE ON GENERAL AND INDUSTRIAL CHEMISTRY. 
 By Dr. Ettore Molinari. Second English Edition. 
 Translated by T. H. Pope, B.Sc, F.I.C. Vol. I.— Inorganic. 
 With 328 Illustrations and 2 Plates. 42s. net. Vol. II. — 
 Organic. Pt. I. 254 Illustrations. 305. Pt. II. In 
 preparation. ^ 
 
 INTRODUCTION TO QUALITATIVE CHEMICAL ANALYSIS. 
 By C. R. Fresenius. Seventeenth Edition, translated by 
 C. A. Mitchell, M.A., F.I.C, Editor of "The Analyst." 
 With 57 Illustrations. 36s. net. 
 
 EXPLOSIVES. Their Manufacture, Properties, Tests, and History. By 
 Arthur Marshall, A.C.G.I., F.I.C, F.C.S. Second 
 Edition. 2 vols. With 158 Illustrations. £$ 3s. net. A 
 Short Account of Explosives. 7s. 6d. net. A Dictionary of 
 Explosives. 15s. net. 
 
 A HANDBOOK OF COLLOID CHEMISTRY. By Dr. W. 
 
 OsTWTALD. Second Edition. Trans, by Dr. Martin 
 Fischer. 93 Illustrations. i8s. net. 
 
 AN INTRODUCTION TO THE PHYSICS AND CHEMISTRY 
 OF COLLOIDS. By Emil Hatschek. Third Edition. 
 With 17 Illustrations. 5s, net. 
 
 LABORATORY MANUAL OF ELEMENTARY COLLOID 
 CHEMISTRY. By E. Hatschek. With 20 Illustrations. 
 7s. 6d. net. 
 
 J. 6 A. CHURC:hILL