EXCHANGE EXCHANGE Studies in Adsorption A THESIS SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY EARL PETTI JOHN June, 1918 Studies in Adsorption A THESIS SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY EARL PETTIJOHN June, 1918 t. Studies in Adsorption PART I. AN ATTEMPT TO DETERMINE APPROXIMATE GRAIN SIZE AND THE MEASUREMENT OF THE MAXIMUM THICKNESS OF SURFACE FILMS. Introduction. The determination of mean grain size is of importance in a number of problems. Among these are problems involving the flow of water through soil and adsorption problems. In the first case grain size determines the number, size and kind of pores through which the water flows, in the second it determines the amount of surface available for adsorption. In the former case King 1 has done some work on esti- mating pore space and relating the rate of flow to the diameter of the grain; in the latter, mass is universally substituted for surface, a procedure which is apt to introduce a variable error. The first part of this paper deals with a new method of estimating the diameter of small grains and gives results for materials of known surface. The thickness of water films formed on glass and sand has been investi- gated quite fully. The second part of the paper describes a simple method of determining the maximum film which can form on materials of this kind. Materials. Sand. Ordinary river sand was used. It was treated with con- centrated hydrochloric acid until no test for iron was shown. The acid was then washed out with distilled water and the sample dried in the air. Four samples were obtained by sifting. The first sample contained all of the sand which passed the ten-mesh screen but was retained by the twenty-mesh screen. The second, third and fourth samples consisted of the fractions from the original lot retained by the forty-, sixty- and eighty-mesh screens, respectively. These are called ten, twenty-, forty- and sixty-mesh sands in this paper. The grains in this lot of sand were far from spherical, no two diameters being the same. An approximation of the surface was obtained by weighing a counted number of grains (4000 to 5000), to get the average weight per grain, and determining the specific gravity. On the assumption that the grains were spherical the diameter and surface of a single grain could be calculated. It was realized when these values were obtained that they were at best only approximations. Ottawa Sand. A single sample of sand called in this paper "Ottawa Sand" consisted of well rounded grains. This sample gave values by the above mentioned method which were very close to the true value for the diameter and surface. It was considered to be of known surface. 1 Nineteenth Annual Report Geological Survey, 1897-98, pages 67-294. 444249 Glass Pearls. The glass pearls used were solid, round and of various sizes, as indicated in the table below. A few, which were poorly formed were removed from the lot by rolling them down an inclined board. Those which were not round could be easily picked out in this way. The pearls were from two different sources, and apparently of different kinds of glass. They differed considerably in specific gravity. The first lot was purchased retail. The material was sold under the name of "glistening dew" and was used to decorate fancy cards. Two samples were obtained from this lot by "elutriation." A quantity of the pearls were placed in a tube and delivered from it at a slow rate into a rising column of water. Under these conditions by properly regulating the current, the lighter ones were carried up and the heavier ones sank to the bottom. These samples are No. 9 and No. 10, in the tables. The second lot consisted of five samples, Nos. i, 3, 5, 7 and 8. The individual pearls in each sample were of the same diameter except for No. 7 which contained pearls of two sizes. These samples were obtained from Germany and when received were coated with dye. All of the samples were cleaned by boiling in concentrated nitric acid, washing free from acid and air drying. The diameter surface and volume of the pearls in each lot was determined by the method used for the sand. Since the pearls were very nearly spherical in form they were considered to be of known surface. Precipitated Silica. Precipitated silica was only used in the preliminary work in this paper. No attempt was made to determine the surface or diameter of the particles of the powder accurately. The microscope showed it to be very fine, but far from uniform. The sample used was of German origin. It was necessary to wash free from iron before using it. The following table gives the weights and the specific gravities of the materials used. TABLE I. Data on Materials Used. Samples. Mean weight of Sand. single grain in gram. Specific gravity. io-mesh 0.000168 2.643 2O-mesh o.ooono 2.645 40-mesh o . 000030 2 . 650 6o-mesh o . 000007 2 . 666 Ottawa o . 000686 2 . 656 Pearls No. i 0.003988 3-101 No. 3 0.002625 3-090 No. 5 0.000853 3-079 No. 7 0.000261 3-125 No. 8 0.00012 1 3-069 No. 9 0.000182 2.505 No. 10 0.000122 2.496 Part I. General Considerations. There are at present three methods of determining the surface of small grains. They are: i . Count-weight Method. 2. Average Diameter Method. 3. King's Method. The count-weight method has already been described, it being the method used in the calculation of the diameter and the surface of the samples used in this piece of work. It is accurate only if the particles are spherical and of the same diameter. The average diameter method consists in measuring the diameter of a large number of grains and using the average obtained for calculating the surface and volume. It will also give accurate results if the grains are spherical and of very nearly the same diameter. This method as well as the former one may give results far from the actual ones for grains that are not spherical. King's method consists in determining the time taken for a given volume of air or water to pass through a certain packed volume of the material to be tested. A formula, is given for the amount of air flowing through the apparatus in a given time. The quantity for unit time varies with the square of the diameter. The method is a first attempt to determine diameter and surface independent of the individual particle. The above equation is derived from a mathe- matical study of the factors involved in the passage of air through such a medium. The equation and the experimental results reported by King check with a fair degree of accuracy. The method described in this paper is in some respects similar to that of King, the results being derived from the rate at which water is removed from the surface instead of the rate at which air passes through a mass of packed grains. Experimental Work. The first work done was of a preliminary nature and was carried out for the purpose of determining the magnitude of the changes that could be expected with the materials used. Air dry samples of Ottawa sand, ignited and unignited silica, and twenty-, forty- and sixty-mesh sands were placed over phosphorus pentoxide to dry, being weighed at intervals. Seventy-five gram samples were used. The samples were placed in crystallizing dishes all of them being placed in the same desiccator to insure their drying under uniform conditions. The results obtained are shown in Table II, and on Plate I. A second series of determinations was then made by placing weighed samples of precipitated silica (air-dry), over different concentrations of sulfuric acid. The desiccators in which the samples were subjected to the vapors of the sulfuric acid solutions were themselves placed in a large oven, electrically heated. Under these conditions the effect of tempera- ture and vapor pressure on the film could be studied simultaneously. TABLE II. Loss in Weight of Air-dry Material Placed Over Phosphorus Pentoxide. Time in Ottawa 20-mesh 40-mesh 60-mesh Ignited Un ignited hours. sand. sand. sand. sand. silica. silica. I 5 /6 0.2 O.l8 0.12 0.09 0.05 0.05 3 6 /i2 0.34 0.32 0.23 0.15 0.07 0.06 4 n /i2 0.48 0.45 0.34 0.24 o.io 0.08 6 l /4 0.65 0.62 0.48 0.33 0.14 o.i i 7 3 A 0.85 0.80 0.63 0.43 0.19 0.14 io l /z .... .... 0.63 0.28 0.18 nVa 0-72 0.31 0.20 Results expressed in milligrams. TABLE III. Variation in Weight of Air-dry Silica when Temperature and Vapor Pressure are Varied* Series i. 12% Sulfuric Acid. Temperature, C. Vapor pressure. Increase per gram. 50 88.0 0.00408 43 61.3 0.00540 33 34-5 0.00729 19 13.6 Series 2. 44% Sulphuric Acid. 50 48.3 0.00242 43 33-7 0.00257 33 18.8 0.00267 19 8.0 0.00304 Series 3. 52% Sulfuric Acid. 50 31.5 0.00040 43 22. o 0.00080 33 13-3 0.00090 19 5-5 0.00150 Series 4. 70% Sulfuric Acid. 50 5.9 0.00170 43 4.3 0.00180 33 3.0 0.00180 23 1.6 0.00170 On removing a sample from the oven for weighing it was allowed to cool to room temperature in the air. Since the sample was at a higher tem- perature than that of the room no accumulation of moisture could take place on it from the air, and any loss of moisture by it to the air would of necessity be considered as condensed moisture, and not as a part of the surface film. Reheating was continued at the same temperature until a constant weight was obtained. If the weight of the sample so treated is greater than the weight of the air dry material some moisture has been taken up which is not lost during cooling. This moisture may be consid- ered as a part of the semi -permanent film on the surface of the grain. 4. Q 8 Time in hours e . & i 1 \ \ /- 1 PLATE! . Relation between vapor ^ v pressure ana increase ^k f in ireiaht / / ^ ^ v /s 7~ : V. m 4/ 5 i -2 / // &^f^-7 M| " 40 60 to IOC Vapor Pressure If on the other hand the material on exposure to the air reverts back to the weight of the air-dry sample no thicker film than that present on the air-dry sample can be formed at that temperature. Vapor pressures greater and less than those of the air under ordinary conditions were used so that a tendency to revert to the air-dry weight could be checked from either side. The values obtained are shown in Table III, and on Plate II. Discussion of Results. When a pile of small particles like any one of these samples is exposed to conditions favoring evaporation of the surface film of moisture, the greater part of the moisture must evaporate by way of the air spaces between the particles composing the pile. If the particles are of the same shape and if the same arrangement of particles holds in the different piles, the size of the air spaces will be determined by the diameter of the particles. It might therefore be assumed that there would be a direct relation between the size of particles and the rate of evapora- tion. The results in Table II and the curves on Plate I show that the relationship holds for the materials used, the loss during a given interval decreasing as the grain size decreases. The relationship is not a simple one, however, since the capacity possessed by a surface for holding a liquid film, as well as the arrangement of particles in the piles may differ. That the relationship is not simple for the three samples of meshed sand is probably due to the irregularity of the grains and the resulting irregular arrangement of them in the pile. The results from the second series of determinations are shown graphic- ally on Plate II. Solid lines connect points at the same temperature, while "broken ones connect those for the same vapor pressure. No attempt was made to definitely determine either the shape of the curve or the actual amount of liquid held, since the purpose of this part of the work was only to get the magnitude of the change caused by variation of temperature and vapor pressure. Samples placed over 70% acid showed a decrease in weight, all of the others an increase, over that of the air-dry sample. In taking the warm sample out of the oven and cooling it in the air, there is no tendency for the moisture in excess of that held by the air dry sample to evaporate. If the air-dry sample has the smaller amount there is no tendency for the sample to take up moisture from the air. From the curves it may be noted that the change in weight of the film formed over any one concentration of acid is very slight as the vapor pres- sure is increased, and that the slight change is in the direction of a de- crease in weight with increase in vapor pressure. The amount of the de- crease for a certain increase in temperature also varies with the acid strength, being greater with the lower concentrations. This is equivalent to saying that the decrease for a certain temperature change is greater as the vapor pressure is greater. Heating the sulfuric acid container would have two effects. It would decrease the capacity of the solid for holding moisture, and it woud in- crease the amount of vapor present in the air and available for the forma- tion of a film. The effect upon the solid would undoubtedly result in less liquid being held. Increasing the vapor present might cause either an increase or a decrease in the amount of liquid held, depending on the intensity of the force which holds the film to the solid. As the vapor pres- sure increases water will evaporate from the sulfuric acid solution and may evaporate from the film already present on the surface of the pearl also. In the experiments above, the resultant of the two effects brought about a slight decrease in weight as the temperature increaseed. The results indicate that the thickest film that can form is formed when the temperature is low and the vapor pressure high ; also that the amount held gradually decrease as the vapor pressure increases with rise of temperature. While the results shown do not make this a certainty, they do make it probable, since the only other possibility lies in the curve for one strength of acid (broken line curve of Plate II), showing a maxi- mum at some temperature below 19 and decreasing again for vapor pres- sures below this maximum. The curves indicate quite clearly that the film is not chemically com- bined, since the amount of liquid held at constant temperature, decreases with the vapor pressure of the sulfuric acid solution. This is not a normal behavior for chemical compounds, which lose their water, if the reaction is a reversible one stepwise, rather than at a gradual rate. These results show roughly the magnitude of the change when a part of the film ordinarily present is removed by drying, and also the effects that may be expected if temperature or vapor pressure changes. The results as obtained were so small that it seemed impossible to get measurable results on pearl samples of known surface. A few trials showed that the weight of pearls used would have to be at least 150 grams, and that this weight would considerbly increase the time required to obtain equi- librium conditions. The fact that the loss in weight varied with the size of the grain was used as a basis for a new method of determining surface. It will be seen from the work already described that the rate of loss of moisture from a mass of sand grains varies with the size of the grain, and that the larger the grain the greater the loss for any interval of time. It seemed probable that the loss was dependent only indirectly on the grain size, the direct factors being the amount and the shape of the un- occupied space. King 1 in his investigation of the flow of water through soils determined the amount of this space for different soils, calling it "pore space." The following formula was used in determining it: Vd W looVd in w r hich, V = volume of sand and pore space, d = density of sand, W = weight of sand used, P = pore space. For rounded grains it was found to vary with the size of the grain, be- tween 32% and 40% and for ordinary soils between 32% and 47%. The force which holds the liquid in the spaces existing in a mass of small grains is physical in its nature. The nature of the surface of the grain and the size, shape and arrangement of the pores in the mass, de- termines the magnitude of the force. For material of the same kind the nature of the surface need not be considered. If spherical grains of uniform size are used, and if they are arranged as compactly as possible, both the pore space and the liquid held in it are constant in amount. The rate of loss of this liquid should depend on the 1 Loc. cit. 10 size, shape and arrangement of the pores. If each sample is made up of spherical grains of one definite size it seems probable that the shape of the pores would be the same in every sample and that the size of the pores would vary with the size of the grain composing the sample. It also seems probable that the arrangement of the pores would be the same. Whether this is true or not there is a relationship between the size of the grain, the pore space and the rate of loss of liquid from the pores ; and spherical grains which pack so as to give a constant pore space should give results which would show this relationship most definitely. If it is possible to determine how the rate of loss varies with the diameter of the pearl and with the pore space, the pore space can be calculated and the diameter, that is the effective diameter, of the grain can be de- termined from the rate of loss of liquid. For spherical grains the actual diameter would be determined, for other shapes the effective mean di- ameter from which to calculate the surface. The glass pearls and Ottawa sand already described provided just the kind of material needed for determining this relationship. The centrifuge at once suggested itself as a medium by which the liquid could be removed and a method involving its use was devised. TABLE IV. Losses per Gram of Sample on Centrifuging. Ottawa Sand. Time in mins. Sample No. 1. No. 2. No. 3. No. 4. IO O.OoSl 0.0087 0.0083 0.0084 10-20 0.0070 0.0088 0.0093 0.0073 20-30 0.0082 0.0090 0.0085 O.OOSO Pearls No. 10. 10 O.OO8O 0.0065 0.0064 0.0083 10-20 0.0064 0.0078 0.0062 0.0070 20-30 0.0057 0.0061 o . 0050 0.0059 30-40 0.0056 0.0059 0.0050 0.0055 Pearls No. 8. IO 0.0040 0.0039 0.0046 0.0045 IO-2O 0.0043 0.0054 0.0042 0.0052 20-30 0.0047 0.0050 0.0047 0.0049 30-40 0.0047 0.0051 0.0049 .... Pearls No. 7. 10 0.0046 0.0042 0.0056 0.0058 IO-2O 0.0048 0.0049 0.0053 0.0052 20-30 0.0042 0.0046 .... 0.0045 30-40 0.0045 Method Used. Porcelain Gooch crucibles were used to hold the samples while they were being centrifuged, those being chosen which had per- forations too small for any of the sizes of pearls used to pass through. They were as nearly the same size and shape as could be obtained. A II weighed sample of pearls was transferred to a Gooch crucible and distilled water was added to cover the sample. The sample was kept covered with water while 100 cc. of water was passed through the pearls. This was done in order to wet the pearls uniformly and to remove bubbles of air. The greater part of the free liquid was then removed by suction, the Gooch crucible was transferred to the centrifuge and the sample was rotated for ten minutes at the rate of 1000 revolutions per minute. The crucible was then removed and weighed and alternately centrifuged for a ten-minute period and reweighed until the original weight was obtained. The weights of duplicate samples as well as those of the different sizes were varied somewhat to see whether an appreciable effect on the loss in weight would result. Ottawa sand and pearls Nos. 10, 7 and 8, were used, the loss in liquid being determined for ten minute intervals with four samples of each. The results are given in Table IV. Discussion of Results. In considering this method of determining the diameter experimentally there were five factors which, it was thought, might cause results to vary. 1 . Size and Shape of Crucibles. It was possible to select crucibles which were of practically the same diameter and height. This source of error was, therefore, of but very little importance. There is no doubt that a difference in the diameter of the crucible would cause the weight loss to vary, since the centrifugal force acts over the section of the crucible. 2. Number and Size of Perforations in the Crucibles. No effort was made to measure accurately the size and number of perforations in the base of the Gooch crucibles. It seems probable that they would have an influence on the rate of loss of liquid unless they were numerous enough to readily take care of all of the water driven from the pearls during a ten-minute period. Apparently such differences as existed had no effect on the relative rate of loss of liquid. 3. Amount of Water Present when Centrifuging Began. A rough at- tempt was made to adjust the amount of liquid when centrifuging began, through the application of suction for a definite time, following a definite pre- liminary treatment. The method used did not do this with any accuracy since the amount of liquid held by any one sample of pearls was not the same in any of the four determinations made. It was expected that this might lead to larger losses particularly during the first period of rotation. That it had no such regular effect can be seen from the results. This is probably due to the fact that on centrifuging a constant pressure is applied to the pearls from the surface layers down, and this pressure effects the removal of a certain quantity of water at the bottom, independ- ent of the am ount of water actually present in the crucible. It was found 12 that after the first centrifuging the top fayers of pearls were practically dry, and as centrifuging was continued the dry layer deepened until all of the water was removed. 4. System of Packing the Pearls. There are innumerable possibilities in arrangement when a mass of small pearls are poured into a crucible, and the pore space may vary greatly with the method of packing em- ployed. It seems probable that the greater part of the variations found in the above results are due to this factor. The centrifugal force applied for drying purposes was the means used in these experiments to control the method of packing. The results, especially during the last two periods of rotation show that the system of packing in duplicate samples must have been closely the same. 5. Variation in Weight of Sample Centrifuged. The weights of samples taken were purposely varied slightly to see whether any change in the rate of drying would result. The rate of loss on drying was found to be fairly constant for the same sample and such variations as were found did not correspond to the variations in the weight of the sample. Losses per Gram (see Table IV). The loss per gram during a ten- minute period of centrifuging while showing some variation are fairly close. Occasional high and low results occur particularly during the first two ten-minute periods. It was considered that drying conditions would be more uniform for the third ten-minute period when a layer of dry pearls covered the wet ones and when any moisture on the inside of the crucibles and above the top layer of pearls would be removed. For this reason the values obtained in this interval were used in the calculations. It is readily seen that the loss on centrifuging any one sample is pro- portional to the time of centrifuging, as would be expected. TABLE V. Data for Centrifugal Samples. Av. Loss Vol. per gm. % pore Pore space Sample. Diameter. per gram. (in cc.) space. (in cc.) K. Ottawa Sand 0.079 0.0084 -59 36.26 0.22 0.136 Pearls 7 0.054 0.0044 -45 34-oo 0.15 0.128 Pearls i o 0.046 0.0057 0.60 33.40 0.20 0.136 Pearls 8 0.041 0.0048 0.49 34.20 0.17 0.140 A relation was found to exist between the diameter of the pearls, the rate of loss and the pore space. Diameters were obtained by the count- weight method, the rate of loss experimentally by centrifuging, and the pore space by calculation from King's formula. A tabulation of the data is given in Table V. The rate of loss was found to vary as the square root of the diameter and as the pore space. Expressing this in the form of an equation we have 13 L = KPD H , in which, L = loss per gram, P = pore space, and D = the diameter. Values of "K" are shown in Table V. The variations in "K" indicate a variation of about 0.002 cm. in the diameters obtained. No application of this method was made to samples of irregular shapes, since there is no method for getting a standard for the effective diameter. No. 7 was composed of pearls of two sizes which accounted in part for its smaller amount of pore space. The sample also shows a low value for "K." While the results obtained show the relation anticipated they do not give the desired degree of accuracy. A part of this failure is no doubt due to the method of calculating pore space. It must also be noted that the value of "K" as given is limited to the crucibles used, since the loss per gram could be decreased by decreasing the number of perforations in the bottom of the crucible. Neither pore space nor diameter would be effected by this, so that "K" would have to vary if the relationship held. The method as developed gives a general relationship but does not per- mit of determining the individual factors in the equation with an accuracy sufficient to warrant its application in determining diameters more closely than 0.002 cm. This variation is too large for the diameters considered. As a result of this no additional work was done along this line. MEASUREMENT OF THE THICKNESS OF FILM FORMED ON GLASS AND SAND. Introduction. A large amount of work has been done up to the present time on the formation of a film on the surface of glass or silica, in which water has been used as the liquid to produce the film. Thus, Ihmori, 1 Parks, 2 Briggs, 8 Katz. 4 and Langmuir, 6 give values for the thickness of the film formed on glass, silicate or quartz surfaces. There is a considerable difference in the values obtained as might be expected, since the materials used differed considerably in chemical com- 1 Wied. Ann., 31, 1006 (1887). *Phil. Mag., [6] 5, 517 (1903). 3 J. Phys. Chem., 9, 617 (1905). 4 Proc. Acad. Wetenschappen, 1915, p. 445. 8 /. Am. Chem. Soc., 38, 2221 (1916). 14 position and nature of surface. A part of the material was in the form of small grains (sand and quartz), some being relatively coarse and some extremely fine powder. The glass used was in the form of thin sheets, in some cases curved (spherical), and in others plane. Each of these factors would have an influence on the thickness of film obtained, as would also the temperature and vapor pressure at which the film was formed. Some typical results obtained for the film thickness are given in the fol- lowing table : TABLE I. FILM THICKNESS VALUES. No. Nature of material. Film thickness. Investigator. 1. Glass globes o .0000033 Ihmori 2. Cotton silicate (glass wool) o .0000133 Parks 3. Sand (microscopic powder) 0.00000045 Briggs 4. Quartz (very fine powder) o .0000013 Katz Anorthite (as above) o .0000062 Katz 5. Glass (incandescent lamp globes) o .00000166 Langmuir There are two theories regarding the formation of a film on a solid. According to the first the force acting is physical in its nature, and the intensity of its effect varies inversely as some power of the distance be- tween the two molecules concerned. The force is similar to the force of gravitation but acts through the distance between molecules. According to this theory successive layers of molecules may be built up on the surface of a solid to a thickness such that the attractive force of the solid just equals the tendency of the outer layer of the film to evaporate. The second theory assumes that a chemical reaction takes place and that the water taken up becomes a part of a more or less stable chemical compound. According to this theory a variable amount of water could be taken up by solids depending only on the capacity of the solid to form a loose compound with it. The second theory has usually been assumed to hold for the film of water forming on glass surfaces, free or loosely combined alkali present in the glass, being the substance with which the water reacts. Ihmori, 1 found that keeping the glass in boiling water for some time decreased the amount of water which it would take up. He believed that alkali was removed during this boiling, and that the decrease in the amount of moisture taken up was due to this fact. It seemed worth while to check the values obtained using glass and sand, having known surfaces, if possible. Since no work had been done to determine the thickest film which can form without free liquid ap- pearing, a method of doing this was worked out. This film was com- pared with the film formed with other liquids to see whether there was any basis for the theory that a chemical compound formed with water. 1 Loc. cit. 15 Materials. Solids: Sand. Ordinary river sand was used. It was treated with cone, hydrochloric acid until no test for iron was shown. The acid was then washed out with distilled water and the sample dried in the ah". Four samples were obtained by sifting. The first sample contained all of the sand which passed the lo-mesh screen but was retained by the 20- mesh screen. The second, third and fourth samples consisted of the fractions from the original lot retained by the 40-, 60- and 8o-mesh screens, respectively. These are called 10-, 20-, 40- and 6o-mesh sands in this paper. The grains in this lot of sand were far from spherical, no two diameters being the same. An approximation of the surface was obtained by weigh- ing a counted number of grains (4000 to 5000), to get the average weight per grain, and determining the specific gravity. On the assumption that the grains were spherical the diameter and surface of a single grain could be calculated. It was realized when these values were obtained that they were at best only approximations. Ottawa Sand. A single sample of sand called in this paper "Ottawa Sand," consisted of well-rounded grains. This sample gave values by the above mentioned method which were very close to the true value for the diameter and surface. It was considered to be of known surface. Glass Pearls. The glass pearls used were solid, round and of various sizes, as indicated in the table below. A few, which were poorly formed, were removed from the lot by rolling them down an inclined board. Those which were not round could be easily picked out in this way. The pearls were from two different sources, and apparently of different kinds of glass. They differed considerably in specific gravity. The first lot was purchased at retail. The material was sold under the name of "Glistening Dew" and was used to decorate fancy cards. Two samples were obtained from this lot by "elutriation." A quantity of the pearls were placed in a tube and delivered from it at a slow rate into a rising column of water. Under these conditions by properly regulating the current, the lighter ones were carried up and the heavier ones sank to the bottom. These samples are No. 9 and No. 10, in the tables. The second lot consisted of 5 samples, Nos. i, 3, 5, 7, and 8. The individual pearls in each sample were of the same diameter except for No. 7 which contained pearls of two sizes. These samples were obtained from Germany and when received were coated with dye. All of the samples were cleaned by boiling in cone, nitric acid, washing free from acid and air drying. The diameter, surface and volume of the pearls in each lot were determined by the method used for the sand. Liquids. Distilled water and a series of organic liquids were used to form the films. Specific Gravity of Solids. The specific gravity of the sand and the pearl samples was determined by displacement of water. A specific gravity bottle was weighed, empty, full of water, and then with a known weight of sample substituted for a part of the water. To avoid air bubbles, the weighed sample was run into water in a fine stream. The bottle was then placed in a partial vacuum and let stand for several hours before the final filling and weighing was done. Method of Determining Film Thickness. In most of the previous work done on determining film thickness, the film has been formed by subjecting the sample to the vapors of water and establishing an equilib- rium at the contact surface. Usually the water vapor was at or near its saturation point. As a check on the results obtained in this way, the method used in this paper was evolved, which consists in getting an equilibrium of the film, by the use of liquid water rather than vapor, and getting it with the air saturated. This would give the maximum film which could form, and at the same time, would, by the magnitude of the results obtained, indicate whether there was an essential difference between a film formed from the vapor and one formed from the liquid. Considering the sand and pearl samples already de- scribed, the method involves the addition of small amounts of liquid to them, thus gradually building up on them a film of water. As successive layers of molecules are added to this film a thickness is finally reached at which the surface molecules act as normal molecules. That is they evaporate, flow, exert surface tension, etc. Any liquid beyond this amount would remain in the liquid condition. It was only necessary to get a definite test for the point at which these new properties exhibit themselves. Apparatus. The first and simplest arrangement used for this purpose consisted of a buret and an Erlen- meyer flask. The weighed sample of sand was placed in the flask and liquid added from the buret a drop at a time, with thorough shaking between, until a final drop caused the grains to stick to the flask. When this occurred water was present as free liquid. This "sticking point" was taken as the end -point of the titration. Fig. i. An ordinary buret soon proved unsatisfactory for delivering the liquid, especially so in cases where the liquid was volatile. Delivering the 17 liquid into an open flask also introduced errors with these liquids. To avoid these losses due to volatility of the liquids, and to limit definitely the volume of air saturated during a titration, a weight buret, Fig. i, was substituted for the ordinary buret and the liquid was delivered into a closed flask. Procedure. In carrying out a single determination the following procedure was used: 200 g. of the air- dry sample was weighed and transferred to the clean, dry, Brlenmeyer flask. The flask was then closed by means of the stopper carrying the weight buret. The liquid was run in a drop at a time, the sand being thoroughly shaken after the addition of each drop. Toward the end of the titration only frac- tions of a drop were added, these being removed by tipping the flask to bring the pearls in contact with the tip of the buret. A final addition of liquid caused a large number of the pearls to. stick to the walls of the flask. The weight of the liquid used gave the amount of liquid taken up when a film of maximum thickness formed. Correc- tions were made in the case of volatile liquids for the amount of liquid necessary to saturate the air in the flask under the working conditions. After a determination in which sand was used the sand was air-dried and then heated to strong redness in a large platinum dish. After partial cooling it was transferred to a desiccator over phosphorus pentoxide, and kept for future determinations. The pearl samples were not ignited. They were boiled with strong nitric acid, to which some hydrochloric was added and were air-dried after being washed free from acid. This procedure was followed for the purpose of determining whether a chemical reaction was involved in the holding of the liquid. If the pearls were air-dried, there would be much less tendency for an unstable chemical compound to be broken down, than if they were dried in vacua. The intention was to have the chemical compound, if it formed at all, present at the time of 'titration, and not formed during it. It seemed improbable that any chemical compound formed by the method used, would decom- pose on exposure to ordinary conditions of temperature and pressure. Experimental. In order to obtain the relationship between surface and amount of liquid to produce sticking, a series of determinations was conducted using the glass pearls, water being used as the titrating liquid. Under these conditions the only variables were those of the solid, in- cluding the nature of the surface, the size and the specific gravity of the pearls. For samples from the same source no difference in the nature of the surface was to be expected. During the whole of this work an attempt was made to find other material suitable for titration and of known surface. Results with this material would permit conclusions to be drawn regarding the capacity i8 of different surfaces to hold liquid films and would thus show the effect of the other variants. No other material was found that could be used in this way. Reproducibility of Results. The apparatus as used was subject to some error due to the fact that the quantity of liquid added could only be controlled by opening the lower stopcock of the weight buret. To give an idea of the accuracy obtainable with this apparatus, a series of results obtained with each of two liquids is included. The remaining liquids gave results correspondingly accurate. Liquid. Water. Nitrobenzene. Sample 200 g. pearls No. 8 200 g. pearls No. 3 Weight of liquid, grams o . 1 19 o .040 0.119 o 041 0.117 0.039 o.i 16 0.042 O.I2O O.042 O.I2I 0.038 O Il6 0.118 Average 0.118 o .040 Greatest variation from average o .003 =2.5% = 5 % Variation between highest and lowest values 4.2% 10% The per cent, error introduced depended principally on the amount of surface titrated or on the amount of liquid added, the greatest variation amounting to from 0.004 to 0.006 g. of liquid. The results obtained and the calculations of film thickness are given in the accompanying table: TABLE II. TITRATION VALUES USING WATER. Sample. Diameter Surface Titration Film Pearls. in cm. sq. cm./g. Sp. gr. Weight. liquid per g. thickness. No i O I "^67 14 6? 7 IOI o .003988 o . ooo 1 90 o .00001 29 No. 3.. \j . * 3*j / O . 1 1 80 **f ^ / 17 .09 O * ^ Jx 1 OQO o .002626 O OOO2 I 8 o .0000128 No. 5 0.0808 24.03 O **ar^* 3.079 o 000853 0.000303 0.0000126 No. 7 , o .0542 35 -46 3 I2 5 o .000261 o . 000402 o .00001 13 No. 8 0.0410 46.40 3-069 O.OOOI2I 0.000595 0.0000128 No o o .0540 44 Q-J 2 SOS o .000206 O OOO207 o . 0000066 No. 10 0.0460 *1^T 7O 53 *4 ^ o w o 2 .496 O . OOO 122 \j . wt_rv^^*y i o .000375 o .0000070 SANDS. ^y Ottawa 0.0790 28.58 2.656 0.000686 0.000374 o 0000130 lo-mesh o . 0494 46.00 2.643 0.000168 O.OOI3IO 0.0000285 2O-mesh 0.0430 52.28 2.646 O.OOOI IO O.OOI I2O 0.0000214 4O-mesh 0.0280 81 .90 2.650 0.000030 O.OOI IOO 0.0000135 6o-mesh 0.0170 129 57 2.666 0.000007 0.001480 0.0000114 Discussion of Results. The liquid required for a titration may be used to form a uniform film of liquid over the surface of the pearls up to 19 the thickness at which flow would occur. If this is the case a negligible amount of liquid would be required actually to support the grains, this amount being added after the uniform film had been added, and "stick- ing" would result from a concentration of this added amount at the con- tact surface of flask and pearl through the action of capillary forces. On the other hand, the whole amount of liquid required may be neces- sary to support the pearls through the action of surface tension. In this case no film would form but all of the liquid added would concentrate at the contact surface, and "sticking" would occur as soon as the surface tension was sufficient to support the pearl. In order to determine which of these two hypotheses held or whether the amount used in titrating was the resultant of both effects, some calculations were made of the amount of liquid necessary to support a single grain. Fig. 2 will explain the letters used and the method of calculation followed. Consider a pearl weighing o.oooi g., held to the surface of the flask by surface tension. The liquid holding the pearl may be considered as occupying a volume represented in section by (OBCED), the lowest level of this volume being the circumference of a circle whose radius is a. Surface tension may be considered as acting along this circumference. If the surface tension and the weight supported by it are known, the length of Fig. 2. the circumference required to support the pearl is given o.oooi /sur. ten. Substituting actual values and placing the quotient equal to the circum- ference of a circle enables one to calculate the value of a in the same units as are used for expressing the surface tension (cm.). Having the value of a, the value of h (thickness of liquid acting) may be calculated, since by geometry, 2O and all of the terms except h are known. Knowing both a and h, the volume of liquid holding the pearl can be estimated. It was assumed that the volume of liquid necessary to sup- port the pearl would be that required to half fill the volume represented on the figure by (OB ED). This is believed to be in excess of the actual amount needed. Calculating this value for the smallest pearl used, one weighing 0.00012 g. gave 0.0537 cc - P 61 " g- f pearls. Calculating the same value for pearls No. i, the heaviest pearls used, gave 0.058 cc. per g. of pearls. These amounts are negligible when the amounts required for a titration are considered. This shows clearly that although the end- point is marked by the appearance of "sticking," which is a surface-tension effect, surface tension itself cannot account for the liquid required for a titration. The amount of liquid required is directly proportional to the surface, and the film thickness is uniform for the same kind of glass. As to the actual thickness of film found, it is of the same order as that found by earlier investigators who worked with the vapor phase of water in forming the film. The results are higher than those of all except Parks. It does not seem probable, in view of the results obtained, that there is any difference in the nature of the film itself, whether water in the vapor phase or water in the liquid phase is used to form the film. It also seems probable that what actually takes place on the surface of the grain is a condensation of water vapor. In the titrations in which water was used to form the film it was found that the titrated sample on air drying, wou 1 d again take up the same amount of liquid. It is difficult to see how this could take place repeatedly if a chemical reaction was involved in holding the liquid. The results obtained with samples No. 9 and No. 10, are only about half as large as those obtained with the rest of the samples. This can only be due to a different surface capacity for holding liquid. The results are close to those obtained by Katz with ornithite, and by Ihmori with glass globes. Ottawa sand gives the same value for film thickness that the larger series of pearls does. It seems probable that this is a chance agreement, since the surface of the sand differed considerably from that of the pearls, both in hardness and in texture. As a whole the results indicate that there are two factors which in- fluence the amount of liquid necessary to form the maximum thickness of film. The first factor is the amount of surface, the actual area that the film must cover. The second is the nature of the surface itself, its capacity to hold a film. 21 Not a great deal is known regarding the variants which determine the capacity factor of a surface. It is probably related to the free energy present in the atoms of the surface layers. The work up to this point indicated that the film thickness should be independent of the liquid used, providing the liquid is not too viscous to spread readily. It also indicated the desirability of applying the titration method to the determination of the surface of irregular particles like sand grains. For the purpose of obtaining surface values for the meshed sands, complete titrations for each of these samples with water and with each of the organic liquids was carried out. An attempt was made also to titrate finer sands, loo-mesh, i5o-mesh, 2oo-mesh. These however, would not permit of an even distribution of the liquid over the surface, and no satisfactory titrations were obtainable. To determine whether the same thickness of film would be found with a different liquid, titrations were carried out first with the pearls and then with the sands making use of the organic liquids. The determinations with the pearls were not completed when they were found to check closely for the first liquids used but those on the sands were completed for all of the liquids. The organic liquids used were chosen so that the specific gravity, volatility, surface tension, etc., varied. The results obtained from these two series of titrations are given in Table III, and will be discussed together. TABLE III. TITRATION VALUES FOR SANDS AND PEARLS WITH ORGANIC LIQUIDS. Liquids. Nitrobenzene K o Sands. )-mesh. 20-mesh. .00136 O.OOII7 .00133 O.OOIO8 00122 0.00107 .00131 0.00108 .OOI26 O.OOII2 .00134 O.OOII6 .00131 O.OOII6 40-mesh. 0.00109 O.OOII2 O.OOIO9 0.00107 O.OOII2 O.OOIIO O.OOIOS O.OOIO9 No. 8. 0.00056 0.00059 0.00055 0.00059 0.00059 0.00053 60-mesh. 0.00148 O.OOI5I 0.00148 0.00155 0.00149 0.00146 0.00142 No. 9. 0.00027 0.00030 O.OOO23 0.00029 0.00030 0.00031 Ottawa. 0.00039 0.00037 0.00039 0.00039 0.00039 0.00038 0.00039 No. 10. 0.00040 0.00037 0.00039 0.00045 0.00039 0.00038 Water o Aniline o Dimethylaniline o Phenyliodide o Toluol o Turpentine o Pyridine Liquids. N i trobenzene No. 1. 0.00018 0.00019 Pearls. No. 3. No. 5. 0.00020 0.00030 O.OOO22 O.OOO3O 0.00022 0.00032 0.00021 Water Aniline Dimethylaniline . . . Phenyliodide Toluol. . 0.00017 0.00018 22 Discussion of Results. The results show that the thickness of the film is independent of the kind of liquid used for titrating, and that the sand titrations can be checked with as good an accuracy as titrations of pearls. Occasional results vary, but the uniformity for the whole series is pronounced. This proves definitely that the surface tension of the liquid has no effect on the amount of liquid required for a titration. The surface tension of water is much greater than that of the other liquids but the volume required per gram is the same. This could not be true if the surface tension influenced the amount of liquid required to produce "sticking." It also proves that there is no chemical reaction in the ordinary sense of the term, when a film of water forms on glass. While it might be possible to imagine such an effect between water and glass, it is obviously impossible to do so with the rest of the liquids of the series. In addition to this the calculated film thickness for different sizes of pearls is found to be the same, showing that the volume for titration varies with the surface. The definite conclusion can be drawn that these films are not due to the formation of a chemical compound, but that they are held by the free surface energy of the solid. It seems certain that the same force holds a thinner film. While these films are formed by the addition of liquid to the solid, the inference is that the same conclusion may be drawn for a film formed from the vapor phase. This inference is supported by the fact that the values obtained for the film thickness when formed from the vapor phase are only very slightly lower than those formed by the addition of liquid. It seems probable that a liquid film forms in both cases, but that with the unsaturated vapor phase it never becomes thick enough to show as a normal liquid on the surface of the solid, while when liquid is used the formation of free liquid marks the end of the titration and indicates the thickest film that can form without free liquid being present. In titrating sands a simple relationship such as was found for the pearl samples does not exist between titrated amount and calculated surface. This is partly due to error in calculating the surface, on the assumption that the grains are spherical, and partly to the fact that extra liquid is required to fill the etchings in the surface. However, if relative effective surfaces are sought they may be expected to be proportional to the titra- tion values since there is no application of the surface which would not involve the etchings and so produce results which would be proportional to those obtained by the titration method. 23 Summary. This paper describes a new method of obtaining the thickness of the maximum film which can form on a surface without free liquid being present. Evidence is presented to show that the liquid forming the film does not combine chemically with the solid. The method has been applied to sand and to glass, and films have been formed with water and with several organic liquids. The film thickness is found to be independent of the liquid used and of the size of the solid particle. The method gives accurate values for the effective surface of sand particles, providing that surfaces of the same kind are compared. PART II. THE ADSORPTION OF COPPER SULPHATE BY GLASS AND SAND. Introduction. Adsorption has been applied as a general term to in- clude any one or a group of effects taking place at the contact surface between two different phases. In a specific case it may be a capillary effect, or an adsorption, or a chemical change. Or it may be a combina- tion of two or more of these. The term "Adsorption" as at present used, indicates simply that the action taking place, which is always a change in concentration, is limited roughly to the surface, and is relatively small in amount. In view of the fact that the term is used to include so many effects that may differ entirely in their nature, such as the formation of a film of liquid or gas on any solid, the accumulation of any dissolved substance on the surface of any solid, liquid, or gas in contact with its solution, etc., it is not surprising that the equation used to express the relation between ad- sorption and the factors which influence it, must be a general one. On the contrary, it is surprising that any equation, no matter how general, will apply to so many seemingly different processes. The equation which is generally used 1 for the adsorption isotherm, and which has been found to hold for a good many individual cases, is as follows : X/M = kC n . in which, X is the amount adsorbed, M is the mass of the adsorbing substance, C is the concentration of the adsorbed substance, "k" and ' V are constants depending on the materials used. "n" may be positive or negative, whole or fractional. 1 Zeit. Phys, Chemie, 57-425, 1906. 2 4 It will be noted that the equation as given does not include the surface factor at all, in spite of the fact that adsorption is defined as a "change in concentration of the adsorbed substance at the surface of contact." Its omission is brought about chiefly by two factors, the necessity of using very large surfaces in order to get a measurable effect, and the practical difficulties involved in the subjection of a large, known surface to adsorp- tion. The solid materials ordinarily used for an adsorbing surface are in a finely divided condition in order to increase the surface as much as possible, without at the same time increasing the bulk of the material used. With material of this kind it is very probable that the relation between surface and weight is contant. That is, two grams of finely ground charcoal, clay, or silica, have a surface double that of one gram of the same sample of material. In order for this to hold exactly, it is necessary to assume that the particles of the portions of the adsorbing substance used are all of the same mean size, but even without this assumption, the results ob- tained by substituting mass for surface would be more accurate than those based on the values of the surface derived in any other way known at the present time. In effect then, the term "M," in the equation given above is a relative measure of the surface involved. If we view adsorption as a purely physical effect, a change in concen- tration without chemical reaction occurring, and produce it upon a known surface, and with a known concentration of a solute, the adsorption per unit surface should be a constant quantity. Since the same materials are used, "k" and "n" should have constant values, and if "C" is also kept constant, X/M or X/S should give a constant value (S, represents the surface). Much attention has been given to the solubility of glass in acid, alkali, and salt solutions, in order to determine to what extent the error introduced from this source, influences analytical results. No attempt has been made, so far as our study of the literature reveals, to determine whether glass has a tendency to concentrate certain metallic ions or compounds on its surface, either by a process of physical adsorp- tion, or by a double decomposition, resulting in a solution of the glass and the precipitation of the metal on the surface It is obvious, however, that if a concentration of the solute or of one of its ions does take place on the surface of the glass, it only takes place to a very slight extent, since the exactness of quantitative procedure would reveal even small variations from this cause. Purpose of the Work. Since glass is so universally used in analytical work as a container for sojutions of all kinds, it seemed worth while to attempt to measure the increase in concentration at the surface of glass, 25 in a specific case. The determination of the increase in concentration, or the adsorption, with a given solution, and the variation in amount adsorbed with the concentration of the solution, were both of importance. If increase in concentration took place, it might result from causes which were purely physical, and it might result from a chemical reaction occurring at the contact surface. With pure silica a chemical reaction would not be expected. Comparative values using silica might help to decide whether, when glass is used, the adsorption is physical or chemical It was hoped that a critical examination of all of the results obtained would lead to a definite conclusion concerning the nature of the process. In all of the work on adsorption that has been done up to the present time, only one size of particle has been used, and the total surface exposed has been varied by increasing or decreasing the weight of this sample. The pearl samples described in the former paper afforded an adsorbing medium of known surface. The surface of the meshed sand samples was also known approximately. Both materials had shown the surface relationship when water was adsorbed. If the adsorption was purely physical, results similar to those obtained with water, in the former paper could be expected. The use of these samples also permitted us to vary the surface exposed without varying the weight of sample exposed. If the adsorption depended pri- marily on the surface exposed, its amount would vary as the surface varied. If other factors were involved, such as the mass of the individual particles, this surface relationship would not be found. The only objection to this material was that the size of the grains necessarily limited the surface which could be exposed for adsorption, which would result in very small adsorption values. Copper sulfate was chosen as the substance to be adsorbed, principally because of the ease and accuracy with which the copper present could be determined. No precipitation or filtration was required, a point of very great importance when very dilute solutions are used. The adsorption of copper sulfate by glass might be either physical or chemical in its nature. Since we desired to study an adsorption which might result from either in order to distinguish between them, it fulfilled the requirements in this regard also. Materials. Pearl samples and meshed sands, as in the previous paper. Copper sulfate solutions made up from carefully recrystallized copper sulfate. Method. The method used consisted in placing one hundred grams of the pearls in a clean dry Erlenmeyer flask. Over this was poured one hundred cc. of the copper solution. After shaking and letting stand for a definite time, ten cc. of the liquid was pipetted off and the copper present in it determined volumetrically. This was repeated at the intervals noted in the tables. 26 In order to be sure of the end point used, and also to check any varia- tion in value of the titrating solution, a blank was run before and after each series of determinations. The blank consisted in the titration of a ten cc. portion of the solution being used, with no pearls present. A lower titration value for the solution taken from the pearls, than that ob- tained by running a blank, indicated that an adsorption had taken place. The iodide titration method was used to determine the amount of copper present in the solution. Ammonia was added to a portion which had been pipetted off, until an excess was present, as indicated by the appearance of a deep blue color. Acetic acid was next added to acid reaction, fol- lowed by about a gram of potassium iodide. The iodine liberated was titrated with sodium thiosulfate solution, and the copper present calculated from the amount of thiosulfate required. In titrating these dilute solutions it was found that, after the disap- pearance of the blue color, a light reddish violet color persisted, a few ad- ditional drops being required to cause its disappearance. The titration was continued to the disappearance of the reddish violet color, as a more definite color change occurred at that time. Checks run using the two end- points indicated that either could be used without materially effecting the results obtained. The thiosulfate solution used in titrating was standardized by means of a copper sulfate solution, prepared from pure copper foil, and diluted to a strength corresponding to that of the thiosul- fate solution. Jena glass flasks were used to hold the pearls during adsorption and were also used for the titrations. Any variations due to the use of glass flasks should be present to the same extent in the blank determinations. Experimental. A few preliminary experiments indicated that it was necessary to use the greatest care in cleaning and handling the pearls. If the adsorption was physical and reversible, placing the pearls in a current of running water for some time should remove the adsorbed material. On trying this out with precipitated silica and with the pearl samples, it was found that all of these samples retained copper. Iron was also present in all. Boiling with aqua regia and then washing with distilled water, and drying, removed both to of these metals. In addition adsorption values were greatly reduced after boiling with aqua regia, showing that a great part of the effect obtained with the material as first used, was chem- ical in its nature. As a result, samples subjected to adsorption were boiled in aqua regia, washed, and air-dried before being used with a second solution. Check determinations, using the same concentration of copper sulfate solution as had been used previous to this treatment showed that this method of cleaning the pearls had no appreciable effect either on the individual ad- sorption values obtained, or on the shape of the adsorption curve. 27 The results obtained are shown in the following tables (I and II). On the accompanying plates, curves are plotted based on these results. Most of the values represent an average of two separate determinations, though some of the later ones have not been so checked. TABUS I. Adsorption Results. i. Copper sulfate solution contains 0.000042 gram Cu per cc. Values in terms of grams Cu adsorbed from 10 cc. by 100 grams of solid. Amount adsorbed in Samples. 20 min. 4H hrs. 29 hrs. 51 hrs. Pearls i O.OOOII 0.00052 o . 00062 Pearls 3 0.00010 o . 00034 0.00052 o . 00065 Pearls 5 O.OOO26 o . 00035 0.00043 O.OOO59 Pearls 7 0.00033 o . 00095 0.00155 0.00153 Pearls 8 0.00071 o . 00073 0.00138 O.OOI5I Sand 4o-mesh O.OOO25 0.00019 O.OOO47 o . 00005 Sand 6o-mesh 0.00004 O.OOOI6 0.00012 Sand 8o-mesh 0.00001 0.00005 0.00004 Sand Ottawa O.OOOO5 0.00019 o . 00032 o . 00030 Prec. Silica O.OOII4 0.00191 O.OOI9I 0.00210 2. Copper sulfate solution contains 0.000113 gram Cu per CC. Pearls i 0.00028 0.00061 O.OOIOO O.OOII5 Pearls 3 0.00083 0.00104 0.00125 O.OOI3I Pearls 5 o . 00073 0.00072 0.00070 0.00069 Pearls 7 0.00143 0.00226 0.00274 o . 00326 Pearls 8 O.OOI22 0.00176 0.00214 O.OO222 Sand 40-mesh 0.00028 o . 00034 0.00031 0.00055 Sand 6o-mesh o . 00024 0.00032 0.00023 Sand 8o-mesh o . 00007 0.00015 O.OOOIO O.OOOI7 Sand Ottawa o . 00004 O.OOO2O O.OOO22 Prec. Silica 0.00311 0.00512 0.00514 0.00482 3. Copper sulfate solution contains 0.000153 gram Cu per CC. Pearls i O.OOIIO O.OOI2O Pearls 3 . . . 0.00150 O.OO2OO Pearls 5 . . . o . 00080 O.OOIOO Pearls 7 . . . 0.00290 o . 00300 Pearls 8 . . . 0.00240 o . 00300 Pearls 9 . . . 0.00040 o . 00070 Pearls 10 o . 00080 o . 00090 4. Copper sulfate solution contains 0.000352 gram Cu per cc. Pearls i O.OOOIO o . 00030 O.OOI2O 0.00140 Pearls 3 . 00020 o . 00030 0.00270 o . 00300 Pearls 5 0.00050 0.00090 0.00140 0.00160 Pearls 7 O.OOOIO O.OOO2O 0.00370 o . 00400 Pearls 8 O.OOO2O 0.00080 o . 00300 o . 00330 Pearls 9 o . 00070 . . . 0.00070 O.OOIOO Pearls 10.. . . . 28 TABLE I (Continued}. 5. Copper sulfate solution contains 0.000655 gram Cu per cc. Values in terms of grams Cu adsorbed from 100 cc. by 100 grams of solid. Amount adsorbed in Samples. Pearls i . . . Pearls 3 10 min. 0.00070 O OO2IO 30 min. 2^ hrs. 24 hrs. O.OOO8O O.OOIOO O.OOI2O 0.00260 0.00290 0.00370 0.00025 0.00040 0.00180 O.OOI3O O.OO3OO O.OO4IO O.OO25O O.OO28O O.OO29O' 0.00030 0.00040 0.00100 O.OOI7O O.OO28O O.OO3IO ution contains 0.00150 gram Cu per 30 min. 3J hrs. 24 hrs. 0.00025 0.00028 0.00031 O.OOOlS O.OOO25 O.OOO32 0.00025 0.00032 0.00036 0.00032 0.00055 0.00057 0.00032 0.00032 O.OOO53 O.OOO27 0.00040 0.00047 0.00031 0.00032 0.0004.2 49 hrs. 0.00140 o . 00470 O.OO2IO o . 00470 o . 00300 O.OOI4O O.OO29O CC. 49 hrs. 0.00032 0.00032 O.OOO32 0.00083 0.00061 o . 00047 o . oood.8 Pearls 5 ... o 00030 Pearls 7 . . . . o 00070 Pearls 8 ... Pearls 9 ... Pearls 10. . o . 00070 O . OOO2O o 00040 6. Samples. Pearls i . . . Copper sulfate sol 13 min. O.OOO29 Pearls 3 ... o 00018 Pearls 5 O OOOI I Pearls 7 ... 0.00018 Pearls 8 ... o 00025 Pearls 9 ... Pearls 10. . 0.00016 . .0.00021 TABLE II. Adsorption Values for All Concentrations of Copper Sulfate Used. Amount adsorbed from concentration of 0.00150 000655 0.000352 0.000153 0.000113 0.000042 Samples. gm./cc. gm./cc. gm./cc. gm./cc. gm./cc. gm./cc. 29 hrs. Pearls i . . . . o . 0003 O.OOI2 O.OOI2 O.OOII O.OOIO 0.0005 Pearls 3 , . , . 0.0003 0.0037 0.0027 0.0015 O.OOI2 0.0005 Pearls 5 . . . . O.OOO4 O.OOlS 0.0014 O.OOO8 o . 0007 0.0004 Pearls 7 o . 0006 O.OO4I 0.0037 0.0029 0.0027 0.0015 Pearls 8 0.0005 0.0029 0.0030 O.OO24 O . OO2 I 0.0014 Pearls 9 . . . . o . 0005 O.OOIO 0.0007 O.OOO4 .... Pearls 10 O.OOO4 0.0031 0.0008 .... 50 hrs. Pearls i , ... o . 0003 0.0014 0.0014 O.OOI2 O.OOII 0.0006 Pearls 3 ... o . 0003 o . 0047 0.0030 . 0020 0.0013 o . 0006 Pearls 5 ... o . 0003 O . OO2 I 0.0016 O.OOIO 0.0007 O.OOO6 Pearls 7 ... 0.0008 o . 0047 o . 0040 o . 0030 0.0015 Pearls 8 ... o . 0006 o . 0030 0.0033 o . 0030 0.0022 0.0015 Pearls 9 0.0005 0.0014 O.OOIO O.OOO7 Pearls 10 ... o . 0005 0.0029 O.OOO9 30 min. Pearls i . . . 0.0002 0.0008 0.0003 .... 0.0003 0.0001 Pearls 3 . , . O.OOO2 o . 0026 o . 0003 O.OOO8 O.OOOI Pearls 5 . . O.OOO2 O . OOO2 0.0009 O.OOO7 o . 0003 Pearls 7 . . . o . 0003 0.0013 0.0002 .... 0.0014 0.0003 Pearls 8 . . . 0.0003 0.0025 0.0008 .... 0.0012 0.0007 Pearls 9 . . . o . 0003 0.0003 .... Pearls 10 o . 0003 O.OOI7 2 9 Discussion of Results. The results in the above tables show that a defi- nite and positive concentration of copper takes place at the surface of the glass pearls. The adsorption is not large for any of the samples used, the maximum value being 0.00514 gram of copper adsorbed by 30 grams of precipitated silica. For glass, the maximum is 0.00470 gram of copper, adsorbed by 100 grams of pearls. One hundred grams of pearls represents 1500 sq. cms. to 5300 sq. cms. of surface depending on the sample used. Results using the meshed sand samples are included for some of the first concentrations used. The adsorption values are small and they are very irregular. Since the surface was not definitely known, and since the results obtained were too small and too irregular to permit definite conclusions to be drawn, these samples were omitted in the later work. The original values obtained with precipitated silica were very large. Careful washing and drying reduced the values obtained to one-third of the former value. Only thirty grams of silica could be used with 200 cc. of copper sulfate solution. The values given in the tables are for these amounts. The largest result obtained in the series was that with pre- cipitated silica when the concentration of copper sulfate was 0.000113 gram per cc. While the result is larger than any of those obtained using pearls it is also true that the surface exposed is enormously larger. This result, therefore, really agrees with those obtained with the meshed sands, all of them indicating that the adsorption of copper from copper sulfate solution, by silica, is smaller than the adsorption of the same substance by glass. It is impossible, from the results obtained to give a definite value for the adsorption of copper by glass. The amount of adsorption varies with the concentration of the copper sulfate solution, and also with the size of pearls used. The variation is not regular for either concentration or pearl size. Results were obtained using six different concentrations of copper sul- fate. For the most part the results show good agreement. The ad- sorption-time curves, as well as the adsorption-concentration curves, ap- pear as smooth curves. There is no agreement between curves represent- ing different sizes of pearls, however. Less regularity might be expected for the thirty-minute interval, since this represents a period of rapid change in amount adsorbed. On this account curves have not be plotted for the thirty-minute period. In the work done on adsorption up to the present time, only a single sized grain (usually a fine powder or a colloid) has been used. l With this 1 Z. phys. chem., 57, 425 (1906); Trans. Lon. Chem. Soc., 91-92, 1666 (1907); Biochem. Z., 23, 27-42 (1910); Kolloid Z., 15, 10-18; Z. anorg. Chem., 60, 306-8 (1908); Compt. rend., 151, 72-5 (1911); Proc. Acad. Wettenschappen, 15, 445-54 (1913). 30 Milligram f Adfortreet o, 5 - / OH Z 60- Jl 01-U 13 60-i a wo Son mtsh Sa ltvo San nesh San ef- Co/=t nd-Copp j-Copp d- Copfr * PLATE I Adso rp f/'o n - Tim e Cur ^ er Cone.' 000042 a. per ce. er Cone. = . oooi / 3 tj. per- cc. er Cone. =. ooo//3 cj. per cc. er Cone. = . 000042 ej. per cc. r es ^ ==r.= =.-r= i=^- ~ :jl ==-^=: L-r^: ^4^-: fn(i- .a Time in hours PLATE E Atfsorpf ''on-Time Curves Sand Samples Free. Silica - Cu. Cone. : oew/3 a. per cc. Prtc. Si lie a Cu. Cone. * . oeoo4z q. per cc. 4O-meshSai9a Cu. Cone.*. 000113 a.pir cc. SO- mesh Sand Cu. Cone. - .000113 ci. per cc. 4o~mesh Sandcu. Cone. *. 000042 2r. per cc. ' 8O mesh Sand Cu.Conc.*.oooo4Zq. per cc. IS 20 25 30 35 Time in hours I Copper Cone. = .oooess a. per cc. 2T Copper Cone. =.OOS352 q. p fr cc. HI Copper Cone. =.000153 q. per cc. IS Copper Cone. =.ooo//3