ELECTROLYTE COAGULATION AND COEFFICIENT OF ELECTROLYTE ACTIVITY' WOLFGAKG OSTWALD University of Leipzig, Leipzig, Germany Received July 1 , 1988 I. INTRODUCTION
I believe that a colloid chemist, if asked today to explain the coagulation of a lyophobic hydrosol by electrolytes, will make a rather unhappy face. Most presumably, when explaining simple and well-investigated cases, as, for example, the flocculation of arsenic trisulfide sols with neutral salts, a conscientious colloid chemist will even voice a warning to the effect that this matter is not as simple as it looks. We can summarize the most important steps in the theory of electrolyte coagulation of feebly solvated sols in abbreviated form by the following headings : 1. Neutralization of charge and equivalent ion exchange. 2. Critical potential. 3. Adsorption as the determining variable of coagulation. 4. The electrokinetic, and not the electrochemical, potential is determinative. 5 . Density of charge instead of magnitude of charge. 6. Compression of the double layer. The sequence of these headings corresponds approximately to the historical development of the theory. We all know that a t least the first four assumptions mentioned are wrong. The electrokinetic potential or its reduction is also not determinative for coagulation, as we are finding more and more cases where the rate of electrophoretic migration increases just prior t o coagulation (Tuorila, Lagemann, Kruyt and collaborators, Bull and Gortner, Mukherjee, and others). I have nothing to say against the last two assumptions. They may be right; however, they are insufficient. They, too, have so far been unable to give a fairly comprehensive quantitative theory of coagulation. The newest development of the theory, for example, that by the Dutch colleagues (Verwey, Hamaker, etc.), becomes continually more complicated and speculative. Presented a t the Fifteenth Colloid Symposium, held a t Cambridge, Massachusetts, June 9-11, 1938. 981
982
WOLFGANG OSTWALD
If a professor is obliged t o discuss this unsatisfactory condition of the theory of coagulation for thirty or more years, in every term of the academic year, then it may easily happen that he becomes more and more impatient. Either he becomes resigned or he commences to curse. The latter course is in general more fruitful. Such mental discontent leads one to the experiment of disregarding for the time being the different approaches of the existing theory, or even to forget them entirely, and to consider the possibility of coming closer to the problem in a way entirely diferenl from any so far used. Such radical doubts as to the suitability of the theoretical assumptions existing lead, for example, to the following considerations : A sol is composed not only of a disperse part but also of its dispersion medium. So far, the properties of the disperse part, i.e., of the micelles, have always stood in the foreground of the theory. Their composition, magnitude of charge, their potential, their double layer and its changes, etc., were the center of the theories advanced. The role of the dispersion medium was of decidedly less importance. The dispersion medium was considered in the theory primarily as the carrier of the micelles and of the stabilizing and coagulating ions. Possibly the different evaluation of the d e s played by the disperse part and the dispersion medium is incorrect. Possibly we are going a step further when, contrary to prior methods, we put the properties of the dispersion medium in the foreground. The dispersion medium of a hydrosol is without exception an electrolyte solution both in the condition of stability and in the condition of coagulation. Our concepts of the inner structure of an electrolyte solution are today somewhat a t variance withfhose of the days of van’t Hoff and Arrhenius. We assume that such a solution is a priori highly dissociated and that thc ions in the solution medium are present in a statistic and kinetic type of ionic lattice. In a salt solution there exists, therefore, a certain segregation of the ions. The factor, i, in the concentration equation of van’t Hoff and Arrhenius has today been substituted, following the procedure of G. N. Lewis, by the factor j , the coefficient of activity. Instead of referring to the ion concentration ic, we talk about the activity, fc = a, wherefrepresents the coefficient of activity. We can either determine f thermodynamically or evaluate the coefficient theoretically, in accordance with the physical theory of Milner-DebyeHuckel. Just as the numerical factor of van’t Hoff received the physical significance of the degree of dissociation, we can explain the coefficient of activity from a physical point of view, although it also was first introduced as a numerical factor only. It is a measure of the interionic attraction and therefore of the inner stability of the lattice. These ionic forces will be weaker, the more the value of f in extreme dilutions approaches unity. From this standpoint a sol exhibits a certain resemblance to a mixed
COAGULATION OF SOLS BY ELECTROLYTES
983
crystal, but of course only a statistical one. The comparatively giant micolles are built into the highly disperse ionic lattice. If we have a stable sol, the interionic forces will carry these micelles. If the sol is coagulated by dialysis, the interionic forces become too weak t o carry the particles any longer. The mutual lattice is torn apart. If the sol coagulates by an increase in electrolyte concentration, a kind of “auto-cleansing” of the statistical ionic lattice takes place. The interionic forces become so large that they drive the micelles together and expel them. It is not the micelles which coagulate owing t o mutual attraction; the dispersion medium coagulates the micelles by aggregation and expulsion from the ionic lattice. This, unquestionably, is a different picture from the usual one. It is now my purpose to demonstrate to you that one can go a step further on this basis than has been done so far, and especially can one actually obtain quantitative related functions for coagulation. 11. THE COEFFICIENT OF ACTIVITY
I n the following we shall consider only coagulation by typical neutral salts and discard, for example, flocculation by H+ and OH- ions or heavymetal salts. Furthermore, we want to keep in mind that the flocculation values so far have been rather inaccurate data, owing to the difficulty of expressing a time reaction with a single figure. Finally, we shall restrict our discussion to truly lyophobic sols, in which the so-called “ion spreading” between ions of equal valency is experimentally negligible. We shall therefore select a specific type, for example, the negative arsenic trisulfide hydrosol. According t o the briefly sketched idea, it is the coefficient of activity itself and not the activity, the product o f f and c, which is determinative. The coefficient of activity referring to one ion type, can, according to the physical theory for water a t H O C . , be defined as -1ogf’
= 0.52:
dc
where z represents the valency and u the ionic strength. u = 1/2(m+.z5 m-.z?), where m+ = molality of the cation, z+ the valency of the cation, m- the molality of the anion, and z- the valency of the anion. Both ions are therefore taken into consideration. The value off+ varies greatly with the value of both ions and with concentration, as can be seen in table 1. I n the first column the type of salt is given, in the second the value for the molality of the salt solution, which corresponds to a constant value for f+ = 0.70. We find that, depending on the structure and the valence of the ions, the coefficient of activity of very different concentrations (molalities) corresponds to the same value for f+, i.e., the same intensity of
+
984
WOLFGANG OSTWALD
interionic forces. The relative figures in the last column will remind a colloid chemist vividly of the relative coagulation values for salts of different valencies. I n these cases too, as is known, one needs decidedly smaller concentrations of polyvalent cations for coagulation, and here, too, the
TYPE
i
or SALT'
____
+ -
TABLE 1 MOLALITY (VI) A T P
p
0 70
I I
1 - 1 1:- 2 Is- 3 14- 4
I
096 068 050 042
,
2 - 2
0 0022 0 0017
I
3 - 1s 31 - 2s 3 - 3
0 00022 0 00017 0 000145
2
0 0 0 0
- 19
100,000 70, 800 52,000
13,800
I
2,290 1,770 229 177 151
~
I 4
-
6
-
0 000040 0 000004
1, la
42 4
* The subscripts represent the number of ions present in a molecule. TABLE 2 Arsenic trisulfide sols (H. Schulze: 6 . prakt. Chem. 26, 431 (1882)) TYPE OF 8ALT
1
NUXBER OF BALTS
1 - 1
11
2 13- 3
5
4
1 5 6 3
1 2 -
14-
2 - 12 2 - 2 3 - 13 32 - 73
1
1
1
f+
mh ( V E A Y VALUE)
0 109 0 0578 0 0375 0 0405 0 00172 n 00246(1) 0 000176 0 000112
0.69 0.71 0.73 0.69 0.72
1
om*
1
0.69 0 51 0 39 0 23 0 85 0 63 0 89
0.72 1 0.74 0 81 1 ______ _______ Mean value off' (with *) = 0.71 3~ 0 03 (about 4 per cent). Slean value o f f (wlthout *) = 0 70 0 55 (about 4 per cent).
-
+
counter ions of t h r same valence have a certain, although decidedly smaller, influence on the coagulation value. I have checked many hundreds of coagulation values t o see whether there exists a quantitative relation between coefficient of activity and coagulation d u e . Tables 2, 3, 4, 5 and 6 give a few examples thereof.
985
COAGULATION O F SOLS BY ELECTROLYTES
I n these tables mk represents the coagulating molality. To eliminate the influence of the cation spreading, only the mean values for mi; of one valency class of ions were used in the calculation. Where rnore than one
,
TYPE OF SALT
TABLE 3 Arsenic trisulfide sols NUMBER OF SALT8
i
ntk(MEmNYALUE)
1
f+
(S. E. Linder and H. Picton: J. Chem. SOC.67, 63 (1895)) 1 - 1 11- 2 2 - 12 2 - 2 3 - la
0.0954 0.0316 0,00128 0.00186( !) 0.000099 0.0001 12
13 4 21 7 2 6
32 - 2a
0.70 0.78
0.76 0.67* 0.78
0.74
(H. Freundlich: 2. physik. Chem. 44, 135 (1903)) 1 - 1 12-
5 1 1
2
2 - 12 2 - 2
0.0502 0 0164 0 000672 0 000810
7
I
0.77 0.83 0.82
0.77
TABLE 4 Antimony trisulfide sols
(H. Schulze: J. prakt. Chem. 27, 328 (1883)) 1 - 1 12- 2 2 - 12 2 - 2 3 - la
0.129 0.0732 0.00216 0.00402(!) 0.000247
0.66 0.68 0.69 0.63(!) 0.67
(A.Iwanitekaja-Orlowa: Kolloid-Beihefte 18, 1 (1923)) 2
-
1 - 1 lz(ch1orides)
2 - 12(nitrates)
3
-
13
3 4 ~
4
3
1
1
i
0.0293 o 000362 0 000475 0 000023
I
i
~
0.82 0.86 0.84 0.88
ion is present in the molecule, as, for example, in potassium sulfate, thc molar coagulation concentration only refers as usual to one cation (= 1/2 KzSO4).
986
WOLFGANG OGTWALD
The tables, which could be increased in number by dozens, demonstrate that the coefficient of activity actually represents approximately the measure for the flocculation power of neutral salts in hydrophobic sols. Ehctrolyte-containing dispersion media zuill flocculate at approximately the same value for the coeflcient of activity of the dominating ions. This is the first quantitative consequence of this new method of consideration. TABLE 5 InfEuence of counter i o n s o n arsenic trisulfide sols (Wo. Pauli and E. Valk6) I CONCENITRATION OF SOL I N
AUTHOR
l QR.41?dd,P,R -1
FORMULA YEIQHT I > EQUIVALENT ILLIMOLE, 'ER LITER
-_-I
I
Freundlich
SAL1
~
1 8
1
KC1 KzSOr 1 K4Fe(CN)e
49.5 65.5
l+
mk ~-
--
0.0195 0.0328
0.774 0.774
0.085 0.050 0.046
0.72
0 0332 0.0218 0 0178
0.811 0.812 0.791
1 Ghosh and Dhar
85 100 185
1
Weiser and Nicholas
6 0
I
I
1 KC1 K8Oa I KAFe(CN)B
33.2 43.5 71.2
0.73 0.68
__.__-
TABLE 6 Ferric oxide sols
1 - 1 11- 2
2 - 2
5
,
0 01175 0 000296 0 00021
~
I
2
1
0.88
1 -
0.87 0.88
(H. Freundlich, Joachimsohn, and Ettisch: Z physik Chem. 141, 249 (1929)) 1 - 1
1
2
1
12-
3 14- 4
13-
I
I
1 1
, '
1
0.013 0.00039 0 000031 0 000015
' 1
0 .88 0.86 0.87 0.80
111. QCANTITATIVE RELATION B E T W E E N COAGULATION VALUE AND VALENCE
If this relation stands, then automatically, or better, mathematically, a new quantitative formulation of the Schulze-Hardy valence rule results. The former mathematical formulations of this rule, as, for example, those of Whetham, Robertson, and also myself hare already been found in-
987
COAGULATION OF SOLS BY ELECTROLYTES
sufficient, for the reason that they did not consider the influence of the equally charged counter ions. From the relation = constant and the equation defining the coefficient of activity, we now obtain for the molecular flocculation power (l/mk) the relation :
fl
TABLE 7 Arsenic trisulfide sols l l m . n+ (RELATIYE)
TYPE OF SALT
1 1.6
1 - 1 12- 2 13- 3
a .o
1,- 4
2.6
1
1.65
2.6 2.8
48 64
51 +47
3 - la 32 - 23
486
573
608
838
2,660 27,216
1,720 14,050
-
14
I
6 - la
4
EXPERIMENTAL)
2 - 12 2 - 2
4
0
llm.n+ (MEAN VALUE;
__
10 20 IO
VALLNCV
FIG. 1 FIG.2 FIG. 1. Relatian between coagulation value and valence FIG. 2. Effect of concentration of sol on the coagulation value
Here we only have one vonstant which is characteristic for the sol, whereas the numerical factor 2 = 1/0.50, the characteristic constant of the coefficient of activity equation for water at 18"C., is independent of the
988
WOLFGANG OSTWALD
sol. We find that in this equation the valence of t'he counter ions is also considered. Table 7 shows 1 1 0 ~ -closely the new formulation of the Schulze--Hardy rule corresponds to experiment. Mean values taken from the above tables on measurements of arsenic trisulfide sols have been used. Recently, Verwey and Hamaker have also evaluated quantitative relations between coagulation value and valence, but in rather speculative ways. These results are graphically demonstrated in figure 1. The valence is given on the abscissa aiid on the ordinate the logarithm of l/mk. The circles represent the experimental mean values of table 7 taken from work on arsenic trisulfide sols. One realizes that the relations as postulated by T'erwey and Hamaker are principally in error, notwithsta,nding the many assumptions which had to be made ad hoc. The relation between the logarithm of the coagulating power aiid the valence cannot be represented by a straight line, as was done by Yerwey and Hamaker, but only by a curved one, as it results directly, for example, from the theory of t'he coefficient of activity. IV. THE EFFECT O F SOL CONCENTRATIOK ON THE COAGULATION VALUE
It is known that the concentrat'ion of the sol (Ca0,jinfluences the coagulation vzlue, and, according t o the valence of the dominant ion, frequently in a different way. The most import,aiit case, the basis for the so-called Burtor's rule, can be demonstrated by figure 2. When dealing wit'h n nionovalent, cation, the coagulation molality decreases with increasing sol concentration; in the case of trivalent or higher ralent cations, it increases. Weiser aiid others have demonstrated that exceptions t o this course are known, as, for example, with chromic hydroxide and ferric hydroxide sols. I do not intend t o discuss in this paper the theory of these relations and should like to refer t o a paper published e1sewhere.l From an experiniental point oi view it should be pointed out that, accordilig t o recent investjigations in my laboratory undertaken at extreme dilution3 of the sols, all curves bend upward. In the C R S C of infinitely dilute sols, one would therefore need infinitely concentrated salt solutions to obtain coagulation, an uiiquestioiiably obvious conclusion. How 1.7-ill these curves look if we plot the coefficient of activity af tlie coagulation point against the sol concentration, instead of rlie coagulntiou ure of this lien- method of grapliical can place all curves in one coijrdinate system, and all curves approach each other with increasing sol conccn?ration. Figures 4 and 5 give fiirtlier examples. We find that with increasing sol concentration the curves approach a Kolloid-Z. 80,301 (1037).
989
COAGULATION OF SOLS BY ELECTROLYTES
common point, T h e rule that the coeflcient of activity i s a constant in the case of coagulation will be more exact the more concentrated the sol. As a result of very recent investigations, which have been undertaken with a
6oo
CI.1
CSOI
FIG.3
FIG.4
FIG. 3. Plot of coefficient of activity a t the coagulation point against sol concentration.
FIG. 4. Plot of coe5cient of activity a t the coagulation point against sol concentration.
O9
t
IC W
SOL CO’IIENTR4TION
GI1
FIG. 6
FIG. 6 FIG. 5. Plot of coefficient of activity a t the coagulation point against sol concentration. FIG.6. Plot of coefficient of activity a t the coagulation point against SO1 concentration for arsenic trisulfide 8018. Appearance of “over optimal” sol concentration.
special technique for determining coagulation points, there seemed to exist ‘‘over optimal’’ sol concentrations, as is demonstrated in figure 6 , using arsenic trisulfide sols. Admittedly, so far only one single case of this type
990
WOLFGANG OSTWALD
has been found, so that it is premature t o decide whether such an intersecting of the curves is a general rule. 1’. PHENOMENON O F DOUBLE COAGULATIOK
Every colloid chemist is familiar with the phenomenon of double coagulation, or the so-called “irregular series” of coagulation. Table 8 represents a common example of a mastic sol. If one successively increases the concentration of a salt with polyvalent cations (negative sols), one obtains two zones of coagulation and three coagulation values: fkl, f1.2, and fk3. If one considers that every hydrosol needs a minimum ion concentration for TABLE 8 Coagulation values obtained in treating a gum mastic sol, concentralion 0.688 g . per liter, with a l u m i n u m chloride (Boutaric) m
0.0000010 0.0000067 0.0000075 0,0000100 0.0000125 0.0000250 0.0000300 0.0000400
3
RESULTS
COAQULATION VALUES
___
Stable Stable Coagulated in 3 min. Coagulated instantly Coagulated in 1 min. Coagulated in 5 min. Coagulated in 2 hr. Stable
fi, = 0.936 0.933
i
0.923
fk;
=
0.995 0.994
991
COAGULATION OF SOLS BY ELECTROLYTES
LOG m
%I
FIQ. 7
FIG. 8
FIG. 7. Plot of log of molality against f* and f FIG. 8. Coagulation by salt mixtures TABLE 9 Coagulation of a gum mastic sol (Buxton, Teague, and Shaffer) m
I
RESULTB
I
FeCla 4
0.0000208
Xo coagulation
0.0000557 0.0000833 0.000163
S o coagulation Coagulation No coagulation
:f
= 0.79;fri= 0.985
0.0333
Coagulation
&
= 0.74
0 0000557 0 0000833 0 000166
Coagulation Coagulation S o coagulation
0 00833
Coagulation
0.0000833 0.000166 0.000333 0.000666
S o coagulation Coagulation Coagulation Y o coagulation
0.0333 0.0833
,
I
JX = 0.83
fL
Coagulation
O 985
-
j & = 0.74
fL
S o coagulation - . ~
=
f,G = 0.86
0.971 (0.74)
f;;
0.63
992
TI’OLFGALI\’G OSTWALD
two zones of stability and three zones of instability. The t s o S-curvcs demonstrate the changes of the two coefficients of activity with the logarithm of concentration. Let us consider the points where the S-curveb intersect with the dability zones. \-e find that always two coagulation points helong together:fko andjkz on one hand, andfkl and f k 3 on the other hand. ‘ h o of these points lie a t the same height. That means nothing else but that at double coagulation the four coagulation values are quanTABLE 10 Forma! gold, undialyred, treated with a l u m i n u m chloride
__(Fuchs and Pauli: m
0.00002
0.00006 0.00007 0.00008
Kolloid-Beihefte 21, 412 (1925)) I
RESULT AFTEII 24 HR.
!
S o noticeable change
I
jFl
Violet, not completely coagulated Coagulated
0.0003 to 0,002 0.003
Red color Red
0.018
Blue, coagulated
= 0.823 = 0.809 = 0.797
Red-violet color
to
mk3
0.804
jL3
-log f + = 4.52/F..; -log j - = 0 . 5 6
TABLE 11 Formol gold sol (Boutaric and DuDin)
.41C1,
ThCis FeC13
I
-----
’
0 00000319 0.966 0 00000758 0.966 0 00000146
0.911
1
0 00026 0 977 0 00095 0.961 0 000441 0 962 0 000823 0.949 0 000243 0 976 0 001625 0.937 ~
~
-
titatircly related to each other. The ualrit of the posztzaa coe&zent of actzvzty at the coagulatzon poznt of a negatire sol equals the value of the negatix c coefficient of actil ity at the point of coagulation of the chargereversed positive sol. etc The last two figure\ have deinonqtrated that this coefficient of actk ity rule is supported by facts. The examples in tables 9, 10, and 11 might further substantiate this To be able t o test the equality of f ~ andfiz o more accurate data in regard
COAGULATION O F SOLS BY ELECTROLYTES
993
to the electrolyte content at dialytic flocculation are needed. From the available experimental values for f k 2 one can deduce that these dialytic coagulation values must be, in full accord with experience, extremely small. VI. COAGULATION BY SALT MIXTURGS
A group of the most peculiar phenomena are found in the coagulation by salt mixtures, We place the coagulation value of a monovalent salt a t 100 and also the coagulation value of a single trivalent salt. If we now examine the coagulating power of mixtures of ‘these two salts we very rarely obtain additivity. As figure 8 demonstrates, we find either seiisitination, super-additivity, or antagonism. This latter effect is particularly striking. The coagulation molality of the mixtures can be many hundreds above the 100 per cent for the individual values. The reason for this anomaly is not yet clear. Might it be possible that the proposed theory would also give us some new ideas as to these phenomena? Naturally one can also evaluate the coefficient of activity of salt mixtures. I am refraining from reproducing here the somewhat cumbersome formulas and prefer simply to give somp actual examples. If we look at a mixture of equimolal solutions, we have, for example:
0.001 m KCl; ljT = 0.964 0.001 m AlCl,; 0.001 m KCl
= 0.449
+ 0.001 m AlCI3;I
= 0.709
+
The coefficient of activity of the mixture therefore lies between the values for the individual coefficients. If we now look at a mixture of both salts, of unequal molality but of eqital coeficient of actzvztg, we obtain the following figures: 0.0612
m KC1;
~ f += 0.750
0.000126 W L AlClS;rIf+ = 0.750 0.0612 m KC1
+ 0.000126 m AICla;I
+IIf+ =
0.914
In such a mixture we obtain a material increase of the joint coefficient of activity. According to the first-mentioned coefficient of activity rule, this means that when using such mixtures a much higher molality is necessary for coagulation than would correspond t o additiyity. The appearance of ion antagonism in such coagulation experiments is also contained in the new theory. Figure 9 gives only one example from a large number of theoretical calculations available, namely, the coinbination of a monovalent salt with a polyvalent one. We can see that the antagonism becomes more pronounced the larger the valence difference between the two salts. I n a
994
WOLFGANG OSTWALD
mixture of two salts of the same valence no antagonism exists. These, as well as further theoretical postulates, are in full accord with experiment. Figures 10 and 11 give two examples of the attempt to evaluate experimental cases quantitatively by the new theory. That we have not obtained better agreement is due to the fact that the sols were too dilute, so that the first coefficient of activity rule could not be exactly fulfilled. Compare, for example, the differences in the f i values in figure 10. Furthermore, the location as well as the height of the maximum are extremely sensitive towards minute yariations of the proportions in the mixture.
x
L.SOl
FIG. 10 FIG. 9 FIG. 0. Coagulation by a monoislent salt combined with a polyvalenl one. FIG.10. An cxample of the attempt to evaluate experimental cases quantitatiLely by the new theory.
Experimental error\ in the fifth decimal point of the molality are clearly noticeable in the curws. VII. THE EFFECT O F CHANCES IK BOLVEST AND TEMPERATURE
The equation defining the coefficient of activity as given above holds only for water at 18°C. If we change the nature of the solvent, for esample, by the addition of alcohol or, if we coagulate a t other temperatures, we must apply the extended equation for f+ which is given by:
995
COAGULATION OF SOLS BY ELECTROLYTES
where e = elementary electric charge, D = dielectric constant, T = absolute temperature, k = Boltzmann's constant, and the root content = ionic strength. As can be seen, the coefficient of activity is a multifunction of the dielectric constant and temperature. It may not be overlooked that D in itself already is a function of 5". If the proposed theory is correct, the coagulation molalities in the cane? discussed must decrease with increasing temperature as well as with increasing addition of a solvent with lower D, for example, alcohol. Figurc 12 reveals the decrease of mk for the case ill which j+, the coefficient ot activity at the point of coagulation, equals 0.75. This means that by the addition of ethanol the coagulation value can be reduced much more eniily and much more markedly than by an increase in temperature. Tkw curvature of the two curves is opposite.
WATCR
LlHANOL
FIG. 12 FIG.11. Another example of the attempt t o evaluate experimental cases qunntitatively by the new theory. FIG. 12. Effect of ethanol and of temperature upon the coagulation of arsenic trisulfide sols by salts.
If this deduction is tested experimentally, one actually obtains (with normal and neutral salts) these two typcs of curves, as shown in figure9 13 and 14, resulting from m y own measurements on arsenic trisulfide sols. Naturally one can also start with a n ethanol or propanol sol, as, for example, Weiser and JIack (8) did, and change D by the addition of water, RS can be 3een on the right-hand sideof figure 14. One obtains the same curves. The theory also permits one to predict the magnitude of such a variation for the coagulation value when varying the temperature and the D of the solvent. For this purpose it is only necessary t o introduce into the above
996
WOLFGAKG OSTWALD
formula, which contains D and T,the new values of D and T . Instead of the numerical factor 0.505 we will obtain other numerical factors. If the theory is correct the j : values must remain constant, notwithstanding variations in temperature and the nature of the solvent.
m,
0
50
tn
#KO‘
0
50 l1
100.
FIG. 13. Effect of temperature on the Coagulation of arsenic trisulfide sols by salts
%
-
%Q 9
1T’lANOL
FIG.14. Effect of ethanol on the coagulation of arsenic trisulfide sols by salts
Table 12 shows some figures taken from measurements on arsenic trisulfide sol. The measurements are not simple t o perform, because, especially at higher temperatures, the sols theniselres hydrolyze and it is therefore necessary to work yery fast. However. the constancy of f: is, on the arerage, to be considered as Yery good
997
COAGCLATIOK OF SOLS BY ELECTROLYTES
KCI KCl
'
0.766 I1 0 772 0 772 0 770 (0 770)1 0 775 0 774 I I11 0.785 0.7851 l(0 767)1(0 755) (0.776) I
KzSO, K2S0d
1
1 0 783' 0 75-11 0 78'3
0 783
I1 '0798~
'
I
1
1 0.781 0 7801 0.7761 (0 767) (0.755) (0.773)
'
BaClz I I 0 761, 0 7621 0 763 0 764; 0 76j, 0 7711 0 7761 0 7701(0779) 0 773 0 7771 BaCh BaCIa ::1 10 785~0 7801 0 780
I
~
TABLE 13 Experiments on ferric oxide sol (C.H. Sorum)
I
I KIIa; 2.0
g. Fez03
per liter
ETHANOL
1
D
percent
0 10.2 20.4 43.1 54.9
80.4 74.5 68.7 54 0 47.4
0 10 20 45
80 74 68 56 53
0 10 20 40 45
1 1 I
(JD
0 500 0.561 0 632 0 909 1 102
I
I, I
0.763 0.773 0.780 1 0.781 ~
ntk~o~NaCl
0.022 0.016 0.012 0.008 0.007
0.84 0.86 0.86 0.83 0.81
0.014 0.011 0.010 0.008 0.007
0.87
I
IIIb; 4.0 g. FFe9Oa per liter
40
IIIc; 8.0 g. FezOs per liter
1
1 0
0 0 0 0 0
500 562 623 857 935
80.4 74 6 68 9 56 1 53 0
0 0 0 0 0
500 562 623 857 935
4 6 9
1
'
1
1
1
0 0 0 0
014 008 007 007
0 006
0.87 0.87 0.84 0.836
0 .a7 0 .88 0 .89 0.86 0.86
alcohol. Table 14 shows propanol sols of mercuric sulfide, studied by Weiser and Mack (8), t o which water had been added. Table 15 refers t o my own measurements on arsenic trisulfide hydrosols. The constancy of
TABLE 14 Mercuric sulfide-propanol sols (a.B. Weiser and G. L. Mack)
pE&tFD ~
I 0 5 10 15 20 25 30 35 40 45
22 0 23 25 24 5 27 1 29 8 32 4 35 5 38 5 41 5 450
,
~
0 5 15 22 25
22.0 23.25 27.1 30.3 32.4
22 23 21 27 29 32 35 38
0 25 5 1 8 4 5 5 11 5 45 0
0
20 25 30 35 40 45
I,.
NaCl
0 10 20
__-___-
'
1 1
50
~
Df:
CaC12
0.0033 0.0037 0,0043 0,0064 0.0092
'
0.63 0.63 0.68 0.68 0.66
__-.
I' ~
I
24.3 22.5 20.6 17.75 15 3 13 6 11 85 10 9 9 35 8 28
MgCh
~
~
1 1
1
O.Oal4
0.89 0 .87
0.0870 0.041 I 9 0.0619 0 0,31 0 0452 0 0182 0 0313 o 0~235 0 03335
____
0 .86 0 .84 0.82 0.EO 0.78 0.76 0.74 0.71
LaClr
1
-~ -
1 1
30
40 46.6 47 6
3.50 3.25 2.56 2.11 1.96
____ 10 15
per cent
mii
LiCl
5
ETEANOL
I
cD.z;
~
0.016 0.038 0.032 0 0254 0.0192 0.017
0.779 0.791 0.779 0.780 0.783
0 0375 0 0350
0.802 0.820
0.836 03 3 7 0 338 0 .E26 (0,801)
0.777 (0 766)
0.798 So1 E: 2.53 g. per liter
~_______-.
0 10
20 30 10
50 56 57 38
I
0 0608 0 0517
0 0419 0 0321 0 0263 0.0220 0 0203
0.786 0.786 0.786 0.786 0.786 0.787
0.761 0.748 0.761 0.766 0.761 0.736
0.784 0.786 0.766 (0 763) (0.755) (0.733)
0.738 0.786 (0,737) 998
COAGULATION OF SOLS BY ELECTROLYTES
999
fi is unexpectedly good.
Only the trivalent Isnthanum chloride does not conform to the rule a t higher alcohol concentrations. Table 16 demonstrates the influence of water on an ethanol sol. If one compares the figures in table 16 in horizontal order one again obtains a n exampie for the correctness of the basic coefficient of activity rule (Ak - rule): fk is approximately constant even if the valence of the coagulating salt is varied. TABLE 16 Influence of water on coagulation of arsenic trisulfide-ethanat sols Sol contains 1.7 g. arsenic trisulfide per liter NaCl WATER
mx
per cent
96 86.4 76.8 67.2 57.6 48.0
0 0045 0 0025 )0.0045(
0.796 0.803 0.796
0.0425 0.04375 0.0475 0.08125 0.03163 0.03250
0.813 0.806 0.814 0.806
(0.0r14)
(0.0412) O.OQ 0.0417
(0.610) (0.720) 0.797 0 .ma 0.809
a -808
VIII. SUMMARY
Summing up, the tables and curves as given seem t o demonstrate that the ncw idea proposed herein, namely, to place the properties of the dispersion medium in the focal point of the theory, is rather fruitful. Seemingly, the quantitativc characteristics of the electrolytic dispersion medium aiid the coefficient of activity offer a tool which permits a quantitative control of a number of coagulation phenomena. Ixt u’i rcturii to our original question: What actually happens during the coagulation of a hydrophobic sol by neutral salts? We now might formulate the following reply: The coagulating forces seem t o be the same as the interionic attraction forces between ions of the dispersion medium. This results from the fact that the coefficient of activity, the measure €or these forces, also plays a predominant rale in coagulation. Nevertheless, it iniglit ))e wire to add that the entire mechanism even today is not as simple as the procedurc actually looks. The new theory is by no means contrary to the many attempts t o evaluate the rOle played by the double layer between the disperse part and the dislmsion medium. The only difference is that now the doubIe layer is looked upon as an interposed apparatus of the dispersion medium, and not as the coagulating motor itself. The dispersion medium brings about coagulation and not the double layer. It is the arm that hits and not the cane.
1000
WOLFGANG OSTTVALD
REFERESCES (1) R U Z ~ G H A., 1 7 . : Kolloid-2. 76, 2 (1936); 79, 156 (1937). (2) H.~MAKER,H. C.: Hydrophobic Colloids (Symposium held a.t Utrecht, November, 1937), p. 16. D. R. Centen’s Uitg. hIij. X . C., Amsterdam (1938). (3) LEDEREH,E. L.: Kolloid-Z. 76, 54 (1936). (4) OJTWALD, Wo., AND COLLABORATORS (H. KOXOROS, K . HOFFMASN,H . A. WANNOTV, ANI W. W. STUART): Kolloid-Z. 73, 301 (1935); 76, 297 (1936); 76, 51, 159 (1936); 78, 324 (1937); 79, 49, 287 (1937); 80, 186, 304 (1937); 81, 48 (1937). (5) PoRultr, C. 11.: Kolloid-Z. 68, 311 (1931). (6) VERWEY,E. J . W.: Chem. Rev. 16, 363 (1935). Hydrophobic Colloids (Syminm held a t Utrecht, Sovember, 1037), p. 58. D. 13. Centen’s Uitg. , ri. C., Amsterdam (1938). H, A,, AND HOFFYANN, IC,: Kolloid-Z. 77, 46 (1936); 80, 294 (1937). (7) >b , E., A N D ll.icx, G. L . : J . Phys. Chem. 28, 1254 (1924). (8) 7,