THE MECHANISM O F THE COAGULATION OF SOLS BY ELECTROLYTES 111. Exchange Adsorption during the Coagulation of Hydrous Oxide Sols BY HARRY B. WEISER AND GEORGE R. GRAY
‘Y3uperequivalent” Displacement of Chloride In recent communications’ a method of procedure has been given which enables one to follow the displacement of chloride ion from the micelles of hydrous oxide sols during the stepwise addition of coagulating electrolyte. The method of titration previously described in detail consists essentially in mixing separate portions of sol, in which is suspended a small amount of calomel, with gradually increasing amounts of electrolyte and determining the chloride concentration potentiometrically after the mixture has stood in a thermostat for at least 48 hours to allow equilibrium conditions to be established. The significance of these observations and of the accompanying change in hydrogen ion concentration, for the precipitation process, have received detailed consideration in the earlier papers and will not be reviewed here. Rabinowitsch2 followed a much simpler method of titration: A definite portion of sol containing suspended calomel was placed in a beaker containing mercury, thus making one-half of a calomel concentration cell. The other half was a standard calomel electrode. The difference in potential was measured at the outset and shortly after each addition of a small amount of coagulating electrolyte. This procedure, which has the advantage of being more rapid than the one we adopted, leads to entirely different results as indicated by the two curves shown diagrammatically in Fig. I . The lower curve W representing the displacement of chloride by sulfate was obtained with a hydrous Fez03sol by the method used in this laboratoryS while the upper curve R was obtained on a similar sol by Rabinowitsch and K a r g a ~ ~ It will be noted that the former curve follows a smooth course well below the straight line showing the amount of sulfate added, while the initial portion of the Rabinowitsch curve lies above the sulfate line indicating that with low concentrations of coagulating agent, appreciably more chloride is displaced than corresponds to the sulfate added. Similar observations were made by Wassiliev and Rabinowitsch with A1103 sol but the phenomenon appeared less frequently than with Fe2O3 sol. Weiser: 3. Phys. Chem., 35, I , 1368 (1931).
* Rahinowitsch and Kargan: Z. physik. Chem.,
witsch: Kolloid-Z., 56, 306 (1931). Weiser. J. Phys. Chem., 35, 16 (1931). ’ Z phgsik. Chem., 133, 203 (1928).
133, 203 (1928); Wassiliev and Rahino-
MECHANISM OF COAGULATION BY ELECTROLYTES
2179
The so-called “superequivalent” displacement of chloride ion during ,the titration process was at first believed by us to be a delusion resulting from the slowness with which the calomel electrode comes to equilibrium. Rabinowitsch, on the other hand, is of the opinion that this factor is relatively unimportant and defends his experimental method. After stating that he has confirmed the experimental results reported by us, he writes:’
CC K,SO,
Added
FIQ.I Diagrammatic Representation of Chloride Displacement Curves obtained on “Titrating” Fe203Sols with K 8 0 p by the Method of Weiser (W) and Rabinowitsch (R).
“There is only one point of discrepancy between your views and those of my laboratory. It regards the question of legitimacy of the procedures used in our respective laboratories. I think that both are equally legitimate. Of course yours presents a more close approximation to the equilibrium in each point of the curve. But ours, which has the advantage of rapidity and simplicity, is no less satisfactory. It only gives the picture of the changes which take place in the first minutes and hours after the addition of each portion of electrolyte, whereas yours tends to find the final state of the system. “The irregular trend of the potentiometric titration curve in our experiments which you ascribe to the calomel electrode not attaining the state of equilibrium, we understand as the picture of the stormy processes of ionic interchange following immediately the addition of concentrated electrolyte solutions to the colloid.” 1 Private communication to Harry B. Weiser. A similar point of view is expressed in an article which appeared after this paper was written. [Rabinowitsch and Fodiman: Z. physik. 4 (1932)l. Chem. 1 5 ~ 403
2180
HARRY B. WEISER ASD GEORGE R . GRAY
This is not convincing for it leaves altogether unanswered the nature of the “stormy process of ionic interchange” which brings about the alleged superequivalent displacement of chloride ion. Adsorption of Precipitating Ions In the investigations above referred to, attention has been given especially to the stepwise displacement of chloride resulting from the adsorption of the precipitating ion. The converse of this procedure consists in determining the actual amount of precipitating ion adsorbed during the stepwise addition of electrolyte both above and below the precipitation value. This has been done by Peterson and Storks’ for the adsorption of chromate by hydrous alumina. A series of samples of the A1*03sol was prepared by the addition of varying amounts of K2Cr04and the adsorption determined by the change in concentration. Below the precipitation value, the sol was ultrafiltered and the ultrafiltrate analyzed, while above the precipitation value, an aliquot part of the supernatant solution was withdrawn for analysis. A curve showing three distinct parts was obtained. The first part, below the precipitation value, was characterized by decreasing adsorption with increasing concentration of electrolyte while above the precipitation value the curve had a form similar to that of the usual adsorption isotherm but with a distinct point of inflection. Peterson and Storks believe that the first portion of the curve represents a neutralization adsorption, the last portion a surface adsorption non-electrical in type, and the intermediate portion a combination of the two types. Since the adsorption of the precipitating ion by hydrous iiI2O3 prepared from the chloride is an exchange adsorption process throughout most, if not all, of the concentration range investigated, the observations of Peterson and Storks may be open to question. Indeed both the choice of precipitating reagent and the experimental method appear to be somewhat unfortunate. I n the first place, it is known that the following equilibrium is set up in chromate solutions: 2CrOq-- 4-2” HzO Cr2O7--. In such solutions we are dealing with two ions and when they are mixed with the sol there will be a shift in the equilibrium depending on the change in pH value.2 In the second place, nothing can be told about the true adsorption equilibrium after ultrafiltration. The process causes a gradual change in the amount of intermicellar solution in contact with the micelles and results in more or less coagulation and agglomeration of the micelles, especially if it is carried very far. A potentiometric method which does not involve ultrafiltration may give a fairly accurate measure of the concentration of ions in equilibrium with those adsorbed in the micelles; but the true equilibrium will be shifted by ultrafiltration. In this investigation attention will be given to the two questions raised in the above survey: ( I ) the alleged superequivalent displacement of chloride
+
J. Phys. Chem., 35, 649 (1931). Cf. U’eiser and Middleton: 3. Phys. Chem., 24, 648 (1920); Ishizaka: Z.physik. Chem., 83, 97 (1913);cf., also, Weiser: J. Phys. Chem., 35, 1368 (1931). 1 2
MECHANISM OF COAGULATION BY ELECTROLYTES
2181
ion from a hydrous oxide sol by a precipitating electrolyte; and (2) the interrelation between adsorption of a precipitating ion and displacement of chloride during the stepwise addition of electrolyte to a hydrous oxide sol.
I. “Superequivalent” Displacement of Chloride Ion The hydrous oxide sols of aluminum, chromium and iron were “titrated” with K2S04by means of the Rabinowitsch method. The observations disclose the causes of the observed superequivalent displacement of chloride and indicate the unsatisfactory nature of the rapid Rabinowitsch procedure.
General Method of Procedure Preparation of Sols. A1203 Sol I was obtained by adding 1.5 N NH40H from a dropping funnel to I liter of a solution containing IOO g AlC13 . 6 H z 0 stirred by a mechanical stirrer, until the precipitate which first formed just failed to redissolve. The mixture was diluted to 3 liters, boiled for 30 minutes and dialyzed in a Neidlel dialyzer a t 75’ until a satisfactory ratio of Alto3 : C1 was obtained, after which it was evaporated to the desired concentration and allowed to age. A1208 Sol I1 was formed by the method of Wassiliev and and Rabinowitsch:2 75% of the theoretical amount of 7.5 N N H 4 0 H was added with constant stirring to I liter of solution containing IOO g of AlC13.6Hz0. The mixture was diluted to 1750 cc and dialyzed as above described. The Crz03 sol was prepared by adding to I liter of solution containing 5 5 g CrC13, small portions at a time of N H 4 0 H containing 2 5 g of concentrated base diluted to 800 cc. The mixture was diluted to 3.5 liters, boiled for 30 minutes, and dialyzed in a Neidle dialyzer at 75’. The Fe203sol was made by adding dilute N H 4 0 H in small portions to 500 cc of a solution containing IOO g FeCl3’6H2O,until 75% of the amount necessary for complete precipitation was used. The sol was diluted to 3.5 liters, dialyzed at room temperature to remove the excess FeC13 and finally at 75’ until the desired ratio of Fez03 to C1 was obtained.
TABLEI Age and Composition of Hydrous Oxide Sols Sol
Age before use
A1203I 3 months at room temperature A1203I1 3 months a t room temperature Cr203 z months a t room temperature I month a t Fe203 2 weeks at 80’ room temperature
+
1
J. Am. Chem. SOC., 38, 1270 (1916). Kolloid-Z., 56, 306 (1931).
Composition MzOs c1 g/l equivalents/l
5.08 3.53
0.0206 0.0268
6.02
0.0140
5.91
o .0096
2182
HARRY B . WEISER AND GEORGE R. QRAY
The oxide content of the several sols was determined by precipitating with a slight excess of “*OH, filtering, washing, and igniting in the usual way. The chloride content was gotten by adding a slight excess of AgN08 and then a large excess of HN03 followed by digestion in the dark for 24 hours at 80’. This dissolved the hydrous oxide and after dilution, the AgCl was filtered into a Gooch crucible, washed, dried and weighed. The age before use and the composition of the several sols are summarized in Table I. Method of Titration. The change in chloride ion concentration during the stepwise addition of 0.5 iV K2SO4was determined by the method of Rabinowitsch and Karganl using the same potentiometer set-up and electrode vessels previously described.* I n all cases, 15 cc of sol were used and 0.5 N KzS04 was added from a micro burette graduated in 0.01 cc. The potential was measured against a 0. I JVcalomel electrode and the chloride activity as KC1 was calculated by means of the Nernst equation. The molar concentration was obtained from activity by means of a graph prepared from data given by Lewis and RandalL3 The chloride concentration was corrected for dilution by the added electrolyte. Preliminary experiments showed that electrodes prepared with sols not previously saturated with calomel required 12 to 15 hours before constant potentials were obtained, while sols which had stood in contact with calomel overnight usually gave constant potentials within 3 hours. Accordingly the electrodes were prepared a t least 1 2 hours before the titration was begun. The titration proceeded as follows: The electrolyte was added from the burette to the sol in the electrode vessel and the mixture stirred thoroughly with a stirring rod. The potential against the 0.1 Y electrode was read at Io-minute intervals for 30 minutes and then a t 15-minute intervals until the potential was constant for three consecutive readings. In general, the alumina sols gave a constant potential after 30 minutes to one hour, while a somewhat longer time was required for the chromic oxide sol and a still longer time for the ferric oxide sol. Above the coagulation point the potential became constant after I O to 30 minutes. Experiments with Alumina Sol Preliminary observations with alumina sol disclosed that there was no superequivalent displacement of chloride if the sol was thoroughly mixed with electrolyte and a reasonable time allowed for the electrode to come to equilibrium conditions. Rapid mixing of sol and electrolyte in an all-glass mixing vessel gave mixtures which showed no evidence of superequivalent displacement of chloride ion at any concentration. I t was observed that the dropwise addition of the relatively strong electrolyte to the sol causes localized coagulation. If stirred sufficiently vigorously, the precipitated hydrous oxide is peptized completely; but if the ‘Z. physik. Chem., 133, 203 (1928). 2Weiser: J. Phys. Chem., 35, 8 (1931). 3 “Thermodynamics,” 344, 362 (1923).
2183
MECHANIBM OF COAGULATION BY ELECTROLYTES
TABLEI1 Titration of
A1203
Sol I with 0.5 N KzSOl [Cl] X 108
corrected
PI x 108
displaced
8.46
9.05
9.05
0.00
14.14 9.18
15.62 11.54 9.92
15.67 11.57 9.95
6.62 2.52 0.90
I
0.05
44.0 51.4 54.5
0.10
41.3
0.10
51.2
0.10
53.4
15.72 10.66 9.78
17.48 11.68 10.62
17.62 11.76 10.69
8.57 2.71 1.64
3.33 3.33 3.33
A B C
0.20
16.03 11.81
0.20
40.8 48.6 49.5
17.86 12.96 12.50
18.09 13.13 12.66
9.04 4.08 3.61
6.67 6.67 6.67
A B C
0.30 0.30 0.30
42.6 45.0 46.1
14.94 13.60 13.02
16.86 15.30 14.64
7.81 6.25 5.59
10.00
A B C
0.40
45.8 44.6 44.9
13.18 13.81 13.65
14.48 15.25
14.86 15.64 15.41
5.81 6.59 6.36
'3.33 13.33 13.33
13.44 14.14 15.00
14.85 15.62 16.62
15.30 16.18
0.50
45.3 44.0 42.5
17.12
6.25 7.13 8.07
16.67 16.67 16.67
A B &C
0.60 0.60
41.4 41.4
15.66 15.66
18.09 18.09
9.04 9.04
20.00 20.00
A B &C
0.70 0.70
40.2 40.9
16.41 15.97
19. I 4 18.52
10.09 9.47
23.33 23.33
A B &C
0.80 0.80
40.0 40.7
16.54 16.10
'9.39 18.85
10.34 9.80
26.67 26.67
39.9 40.3
16.61 16.35
19.58 19.26
10.53
30.00 30.00
&SO4
A B C A B C
0.5 N cc
millivolts
(aC1)
0.00
57.1
0.05 0.05
0.20
0.40
0.40
A
0 .50
B
0.50
c
A B &C
0.90
0.90
X
103
10.58
11.40
15.02
18.30 17.78
18.50 18.I 8
P I x 103
10.21
.67 .67 I .67 I
10.00 10.00
2184
HARRY B. WEISER AND GEORGE R. GRAY
mixture is not well stirred or if sufficient time is not allowed before making the chloride ion measurement, a value appreciably above the normal equilibrium value would be expected. That such is the case is indicated by the following experiments. E f e c t of the Nature of the Stirring following the Dropwise Addition of Electrolyte. The 0.5 N K1S04 was allowed to drop into 15 cc of sol from a micro burette and the following observations made a t 23’: A. The mixture was stirred gently for a few seconds after which it was allowed to stand I O minutes and the potential measured against an N / r o calomel electrode.
n
0 X
&
0
0.4
0.2
0.S
0.6
Cc N/Z K,SO,
1.0
Added
FIG.2 “Chloride Displacement” Curves with A1203 Sol I (A) after and (C) at equilibrium.
IO
minutes, (B) after 2 0 minutes
B. The mixture was then stirred vigorously for 30 seconds, allowed t’o stand for I O minutes and the potential read once more. C. The mixture was again stirred thoroughly and potential readings taken at I O minute intervals until three consecutive readings were the same. The observations on A1203 Sol I are given in Table I1 and shown graphically in Fig. 2 . Line A shows the values obtained on the first readi,ng, line B on the second, and line C the final value. Observations with A1203 Sol I1 are given in Fig. 3. The results are similar to those of Sol I in all essential respects. I t is apparent that the chloride concentrations calculated from the first potential readings show very erratic variations from the equilibrium conditions. The observed superequivalence is the result of two errors inherent in the method, which render it of questionable value. First,, the partial coagulation accompanying the addition of strong electrolyte results in a proportionately greater displacement of chloride than if no coagulation takes place. This relatively greater displacement of chloride in the zone of coagulation accounts for the marked change in direction of the equilibrium chloride displacement curve in this region.’ Second, a few minutes is insufficient time Cf. Weiser: J. Phys. Chem., 35,
IO,
1374 (1931).
MECHANISM O F COAGULATION BY ELECTROLYTES
2185
for the sol-calomel-electrolyte mixture to come to equilibrium. Accordingly the curve obtained by adding the relatively strong solution to the sol, stirring gently, and measuring after I O minutes, is meaningless. The second curve obtained after thorough stirring and waiting for a n additional I O minutes is more regular and the observed superequivalence is much less. Here also the superequivalence results chiefly from partial electrolyte coagulation which is not completely reversed and from delay in setting up equilibrium at the electrode. The final equilibrium curve is similar in form to those obtained by Weiser’s method but the true amount of chloride dis-
0.L
0.t
Cc N/2
06
as
1.0
K,SO., Added
FIQ.3 “Chloride Displacement’’ Curves with Al*Os Sol I1 (A) after IO minutes, (B) after minutes and (C) a t equilibrium.
20
placed is uniformljr higher. The reason is obvious. I n the Rabinowitsch method, partial coagulation a!ways takes place below the precipitation value, while this is avoided in Weiser’s procedure. The excess chloride displaced by the partial coagulation is never taken up completely since the coagulation is not completely reversible. From these observations it is clear that the Rabinowitsch method of procedure has little to commend it. Initial coagulation is distinctly objectionable and the titration curves are meaningless unless an approximate state of equilibrium is attained. If time is allowed for this, the procedure is much more time consuming than the Weiser method. Thus, starting with the sol saturated with calomel, approximately I 5 hours elapsed from the beginning of the titration until the last point in the C curves of Figs. z and 3 were obtained. In the Weiser method about two hours are required to prepare the series of electrodes containing the varying amounts of electrolytes and after these have stood for z days to attain equilibrium, the potentiometric measurements are made in a few minutes. Moreover, the method of mixing avoids the complication of localized coagulation below the precipitation value. Effect of Concentration of Sol. Since the so-called superequivalent displacement of chloride ion results in part from an inherent error in the experi-
2186
HARRY B. WEISER AND GEORGE R. GRAY
Cc N/Z K,SO,
Added
FIG.4 “Chloride Displacement” Curves with AltOi Sols of Varying Concentration (8) after minutes, (B) after 20 minutes and (C) at equilibrium.
IO
mental method, namely, partial electrolyte coagulation, it follows that the effect would be least marked the greater the dilution of sol. This is borne out by the following experiments: A1203 Sol I was evaporated in a vacuum desiccator over HzS04 until its concentratior, was approximately doubled. Titration experiments similar to those described in the preceding section were then made on (I) the concentrated sol ( 2 ) the concentrated sol diluted with one part of water and (3) the concentrated sol diluted with three parts of
MECHANISM OF COAGULATION BY ELECTROLYTES
2187
water. I n order to conserve space the lengthy tables of data are omitted but the results are shown graphically in Fig. 4. To avoid confusion the A , B and C curves for three different sols are plotted separately. The results speak for themselves. There is no “superequivalent” displacement with the most dilute sol if the sol electrolyte mixture is thoroughly stirred before making the potentiometric reading. This merely confirms the earlier observations in this laboratory that there is no indication of superequivalent displacement when the sol is mixed rapidly with the highly diluted precipitating electrolyte.
‘ /
I
s 0
0 X
3
Y
4
0
az
O b
Cc
06
0.8
CIS
N/2 K,504 Added
FIQ.5 “Chloride Displacement” Curves with Crz08Sol (A) after and (C) a t equilibrium.
IO
minutes, (B) after 20 minutes
I n this connection attention should be called to the observation of Rabinowitsch and Karganl that the excess of chloride ion displaced, falls off with dilution of sol and disappears entirely with five-fold dilution of a Fez03sol whose original concentration was 5.76 grams per liter. This is as it should be but the reason for this behavior was apparently overlooked. Experiments with Chromic Oxide Sol Observations on the Cr203 sol were made by the same procedure used with alumina and the results are shown graphically in Fig. 5 . The similarity of curves obtained under similar conditions with the two different sols is a t once apparent. Experiments with Ferric Oxide Sol The observations with Fez03sol are so similar to those with A1203 and CrzOs that no comment is necessary. The data are shown graphically in Fig. 6. Since the chloride content of the sol is less than in the other two sols the chloride displacement is correspondingly less. For this reason the scale in Fig. 6 is made twice as large as in the other figures. 1 Z.
physik. Chem., 133, 224 (1928).
HARRY B. R’EISER AND GEORGE R. GRAY
2188
0
0.2
01
cc
N/z K,SO,
03
0.4
05
Added
FIG 6 “Chloride Displacement” Curves with FezO8 Sol (A) after minutes and (C) at equilibrium
10
minutes, (B) after
20
Conclusion The conclusion from these observations is that the so-called superequivalent displacement of chloride ion during the “rapid” titration of hydrous oxide sols by the dropwise addition of a relatively strong precipitating electrolyte according to the method of Rabinowitsch, is the result of an unsatisfactory experimental procedure which produces localized coagulation of a portion of the sol and does not allow time for equilibrium conditions t o be approached. I t is not typical of any one sol and the experimental conditions cannot be developed uniformly in any case. For this reason the chief value of the rapid procedure is in emphasizing the experimental conditions which should be avoided in studying the change in the adsorption equilibria which follow the stepwise addition of precipitating electrolytes to sols.
11. Adsorption of Precipitating Ions It will be recalled that Peterson and Storks obtained a curve showing three distinct parts for the adsorption of chromate added in gradually increasing amounts to alumina sol. Since the equilibrium curves showing the displacement of chloride from hydrous oxide micelles by adsorption of sulfate and other ions including chromate, show no breaks,’ there was no reason for believing that breaks would occur in the adsorption curve of the Precipitating ion. The following experiments indicate that there are no breaks in the adsorption curves for sulfate added stepwise to the hydrous oxide sols of aluminum, chromium and iron. General Procedure. It was found by preliminary experiments that all of the precipitation concentration of sulfate was adsorbed by the several oxides. Since none of the precipitating electrolyte was left over a t the precipitation point, it is apparent that none would remain unadsorbed below the precipitation value. Accordingly, it was necessary t o make the adsorption measure-
’ Weiser: J. Phys. Chem., 35, I , 1368 (1931).
2189
MECHANISM O F COAGULATION BY ELECTROLYTES
ments above the precipitation value only. This was fortunate since, as we have seen, it is not permissible to ultrafilter the sol and since the lead sulfate electrode was found to be unsatisfactory for the potentiometric estimation of low concentration of sulfate. A definite amount of 0.5 N K2S04 was added from a micro burette graduated in 0.01cc to I 5 cc of sol contained in a test tube. The tube was stoppered and the mixture shaken thoroughly. After standing 24 hours the precipitated hydrous oxide was thrown down by centrifuging for 30 minutes at 3000 r.p.m. in an International Equipment Company Centrifuge. The supernatant liquid was poured off and centrifuged 15 minutes more, after which an aliquot part was analyzed for sulfate by the nephelometric method which was found by preliminary experiments to be quite satisfactory. To a mixture of 5 cc of 1% BaCL solution, 5 cc of 0.1 N HC1 and I cc of 5% sulfate-free gelatin,' was added an amount of liquid (not exceeding 5 cc) which contained not over 0.5 mg of SO4 ion as determined by preliminary experiment, and the final volume was made up to 16 cc with water. This sample was compared with a standard prepared a t the same time using 5 cc of K2S04 solution containing 0.24 mg Sod. The comparison was made after the sample had stood I S minutes, using a Dubosque-leitz nephelometer. I n each case the mean value of IO readings was taken. Adsorption by Alumina. The measurements of adsorption by alumina was accomplished without difficulty except at concentrations just above the precipitation value. I n this region traces of alumina that are shaken up during the centrifuging process remain in suspension persistently. Accordingly, the determination of unadsorbed sulfate after the addition of 0.3 and 0.35 cc of KzSOa was made by the following modification of the procedure described above: A 2 cc sample of the supernatant liquid was added to a mixture of 5 cc of 0.1 N HC1, 5 cc HzO, and 0.5 cc 1% BaCl2 solution. Another 2 cc portion was added to a similar mixture in which water was substituted for the BaClz
TABLE 111 Adsorption of Sulfate by A1203 Sol I (15.0cc of sol with 0.5 N K z S O ~ ) KzSO4
0.5 N cc
0.30
7.20
0.35 0.40
8.40 9.60
0.50
12.00
0.60
14.40 16.80 19.20 21.60
0.70
0.80 0.90 I
so,
added mg
so1
in solution mg
0.61 1.23 1.48 1.74 2.44 3.30 4.31 5.80
so4
&[s04]
$[so41
adsorbed mg
adsorbed X 103
corrected
6.59
8.98 9.73
9.16 9.95 11.29 14.21 16.60 18.76 20.67 21.94
7.17 8.12 10.26 11.96 13.50 14.89 15.80
I 1 .oo
13.80 1 5 .97 17.92 19.63 20.70
Denis and Reed: J. Biol. Cbem., 71, 193 (1926).
X IO^
$[sod
added X 103
10.00
11.67 13.33 16.67 20.00
23.33
26.67 30.00
HARRY B. WEISER AND GEORGE R. GRAY
2190
solution. These samples were compared with a standard prepared a t the same time containing 0.12 mg SO4. The samples were allowed to stand 15 minutes before making the comparison. A correction for the turbidity due to dispersed A1203 was applied to the calculated so4 concentration. The observations are recorded in Table I11 and shown graphically in Fig. 7 , together with the chloride displacement curve from Fig. 2 .
0
14
02
Cc
N/Z K,SO,
0.6
as
1.0
Added
FIQ.7 Curves for the Simultaneous Dis lacement of Chloride and Adsorption of Sulfate by 8olloidal AL03 Sol. I.
The sulfate concentration was corrected to the same basis as the chloride displaced, L e . , to 1 5 cc volume. The adsorption curve follows an unbroken course through the entire concentration range, but there appears to be a slight point of inflection just above the precipitation value corresponding to a similar bend in the chloride displacement curve. The sulfate adsorption is appreciably greater than the amount of chloride displaced. The reason for this is that a part of the sulfate which is taken up corresponds to chloride measurable potentiometrically in the intermicellar solution.' Adsorptzon by Chromzc Oxzde. No difficulty was encountered in the adsorption measurements with Cr203 sol since the supernatant solution above the precipitation value was entirely free from suspended matter. The adsorption data are given in Table IV and the adsorption is shown in Fig. 8 together with the chloride displacement curve from Fig. 5 . ICE. Weiser: J. Phys. Chem., 35, 23, I392 (1931).
MECHANISM OF COAGULATION BY ELECTROLYTES
0
02
04
66
0,I
2191
1.0
Cc N/Z &SOI Added FIQ.8 Curves for the Simultaneous Displacement of Chloride and Adsorption of Sulfate by Colloidal Cr20s Sol.
a
0.1
ai
0.3
Cc N/2 K,SO,
04
1s
Added
FIG.g Curves for the Simultaneous Displacement of Chloride and Adsorption of Sulfate by Colloidal FelOs Sol.
2192
HARRY B. WEISER AND GEORGE R. GRAY
TABLE IV Adsorption of Sulfate by Crz03Sol (15.0 cc of sol with 0.5 N K ~ S O J ) KzSOa o.jN cc
so,
so4
so4
added in solution adsorbed mg mg mg
0.30
7.20
0.10
7.10
0.35 0.40 0.50 0.60
8.40 9.60
0.34 0.88 2.31 3.95 4.88 6.00 7.90
8.06 8.72 9.69 10.45 11.92 13.20 13.70
0.70
0.80
0.90
12.00
14.40 16.80 19.20 21.60
tlSO4l
adsorbed X
103
9.67 10.94 11.80 13.03 13.96 15 82 17.41 17.95
tlS04l correcteS X io3
+[SO41 added
9.86 11.19 12.11 13.42 14.52 16.56 18,33 19.02
10.00
x
103
11.67 13.33 16.67 20.00
23.33 26.67 30 .oo
Adsorption by Ferric Oxide. The adsorption data with ferric oxide sol are given in Table V and Fig. 9.
TABLE V Adsorption of Sulfate by FeSO3 Sol (15.0 cc of Sol with 0.5 N KsSOa) &SO4 0.5 N c,c
0.13
0.15 0.20
0.30 0.40 0 . 50
so4
added mg
3.12 3,60 4.80 7.20 9.60 12.00
so*
so,
in solution adsorbed rn 5 mg
0.24 0.68 1.49 3.03 4.23
5.77
2.88 2.92 3.31 4.17 5.37 6.23
lISO4l
adsorbed
x
103
3.97 4.02 4.54 5.68 7.27 8.38
tIS0ll
corrected
x
108
4.00 4.06 4.60 5.79 7.46 8.63
+IS041 added X
103
4.33 5 .oo 6.67 10.00
13.33 16.67
Conclusion At all concentrations both above and below the precipitation value, sulfate ion enters into exchange adsorption with chloride ion present in the diffuse outer layer of the hydrous oxide particles. The sulfate adsorption curve follows a smooth course above that of the chloride displacement curve through the entire concentration range. summary
The following is a summary of the results of this investigation. I . A comparison is made of the methods employed by Weiser and by Rabinowitsch for following the stepwise displacement of chloride ion from hydrous oxide sols during the dropwise addition of electrolyte to the sol.
MECHANISM OF COAGULATION BY ELECTROLYTES
2193
2, The so-called superequivalent displacement of chloride observed by Rabinowitsch by his method of “rapid titration” is the result of an unsatisfactory experimental procedure which (a) produces localized coagulation of a portion of the sol below the precipitation value and (b) does not allow time for equilibrium to be approached. 3 . The final equilibrium concentration of chloride following localized coagulation of the sol is higher than in the absence of coagulation since the coagulation process is not completely reversible in a reasonable time. 4. At all concentrations both above and below the precipitation value, sulfate ion enters into exchange adsorption with chloride ion present in the diffuse outer layer of the hydrous oxide particles. 5 . Contrary to observations of Peterson and Storks on the adsorption of chromate by hydrous alumina both above and below the precipitation value, the sulfate adsorption curve follows a smooth course above that of the chloride displacement curve throughout the entire concentration range. The most marked break in the adsorption curve obtained by Peterson and Storks is probably due to a displacement of the adsorption equilibrium by ultrafiltration.
The Rice Institute, Houston. Texas.
