AQUEOUS SOLUTIONS OF SODIUM SILICATES. I PREPARATION AND ELECTRICAL CONDUCTIVITY BY R. W. HARMAN.
Introduction No satisfactory proof of the possible number and constitution of sodium silicates, and no explanation of their behaviour in solution and the behaviour, in general, of solutions of varying ratios NazO: SiOz has, so far, been advanced. At the time this work was undertaken our knowledge of the alkali silicates was meagre, vague, and in large part contradictory. Till quite recent times it could be summed up by saying they formed solutions which are largely hydrolysed and in which most of the silica exists in the colloidal state; that several sodium silicates had a more or less hypothetical existence, but the only one whose composition was definitely established was NazSiOa, or sodium metasilicate, crystallising in a number of various hydrates about which much conflicting data had appeared. Moreyl then proved that NazSiO, and NazSipOj separated as crystalline salts at temperatures between 400°-~0000Cfrom solution, and from fused melts. Later Bogue2showed that the degree of hydrolysis as determined by E. M. F. measurements was unexpectedly low and not in accord with earlier work. No attempt had been made systematically to investigate the whole problem or to formulate a theory not in conflict with all the existing data. In the present investigation the system NazO-SiO2 HzO a t 2 5 O C has been studied from the point of view of: (I) conductivity; (2) transport numbers; (3) hydrolysis; (4) sodium ion activity; ( 5 ) silicate ion; (6) lowering of vapour pressure and of freezing points; ( 7 ) heterogeneous equilibria. Confusion has arisen through an indiscriminate use of the terms “sodium silicate’’ and “water-glass”, such terms being useless unless an analysis of the substance is quoted, but this, unfortunately, has not always been done. Throughout the whole of this work, the different silicates and mixtures will be designated by the ratio N a 2 0 : SiOz in equivalent proportions, this being the simplest and most convenient system of nomenclature and one already finding general and serviceable use in industry. Thus a ratio of 1:2 contains one equivalent of NazO in grams to two equivalents of S O z . All concentrations, except where otherwise stated, are expressed in weight normality (N,) with regard to their sodium content, i. e. in pram-equivalents of sodium per 1000 grams of water. Thus, a I N, solution of ratio I :4 contains ~ ( N a z 0 . 4 S i O zexpressed ) in grams, in 1000grams of water. Preparation of Materials. Preparation of silicates by fusion and subsequent solution of the melt presents many difficulties, especially if an exact ratio and a pure product J. Am. Chem. SOC.,36, 215 (1914); J. Phys. Chem. 28, 1167 (1924). Chem. SOC., 42, 2575 (1920).
* J. Am.
1156
R. W. HARMAN
are required. It was, therefore, decided to prepare the crystalline nonahydrate of sodium metasilicate, and to use this as the starting point for the preparation of the other ratios.
Preparation of NazSi03-9Hz0. Jordis' made many attempts to prepare the different sodium silicates and the other alkali silicates in a crystalline form, but he was successful only in obtaining crystals of sodium metasilicate with I O , 9, 7, 6, 5, and 2 molecules of water of crystallisation. It is doubtful whether all these hydrates exist, but this point receives full consideration later in a paper on heterogeneous equilibria of the system Na20-SiOz-Hz0 to be published shortly. Vesterberg2 undoubtedly obtained the pure salt, NazSi03 by crystallisation from an alcoholic solution of a strongly alkaline ratio and recrystallisation from a dilute alcoholic solution. Brunner, Mond and Co., kindly supplied several of their commercial solutions of ratios varying from I :2 to I :4 but all of these, on analysis, were found to contain from 0 . 2 7 ~to 0.57~ impurity, consisting chiefly of Al, with traces of Fe, Ca, Mg, and SO4. The sodium metasilicate was prepared according to Vesterberg from the purest of these commercial solutions. The first crystallisation took over a week, but subsequently by means of inoculation, crystallisation was brought about in a few minutes. The crystals were washed with 50 cc. water and 50 cc. alcohol on a Buchner funnel, drained thoroughly and finally dried between filter paper in a carbon dioxide-free atmosphere. This is essential as the metasilicate is quite rapidly decomposed by COz. Analysis showed that these crystals contained a small excess, 2.3Y0 of NazO over that required by by the formula Na2SiO3.gH2O,and 0.2% A1203 as impurity, evidently due to the formation of a sodium alumino-silicate. These crystals were then recrystallised from a saturated solution to which 2 0 cc. NaOH of sp. gr. 1.26 and 60 cc. alcohol were added to every 3 litres of warm water required to dissolve them. On seeding, the crystals came down in a few minutes, and were washed, drained, and dried as before. Analysis showed this recrystallised silicate to contain 0.9% excess NazO and 0.1% Alz03. A second such recrystallisation yielded crystals corresponding to NazSiO3 .9H20 within the limits of experimental error. These rhombic crystals had a melting point of 47°C (Vesterberg, m. p. 48'; Erdenbrecher, 47'). On dissolving the impure crystals in water dilute solutions were quite clear, but from solutions twice normal and over a residue separated which was filtered off, analysed, and found to contain a very high proportion of AI, while the filtrate was a very pure sodium silicate solution, but its molecular ratio Na20: Si02 had increased slightly. Evidently, the residue was a Z. anorg. Chem. 56, 296; 58,98 (1908); Chem. Zt. 38,922 (1914). 2Eighth Int. Congress Appl. Chem. 2, 235 (1912).
AQUEOCS SOLUTIONS O F SODIUM SILICATES
1157
sodium alumino-silicate, containing comparatively a small proportion of sodium, or it may possibly have been an alumino-silicate. Electrolytic Preparation of Solution of Varying Ratios NazO:SiOz. The electrolytic removal of alkali from salts has been performed by Spencer and Proud’ and by Lottermoser2 who thereby prepared paraitungstates and other complex tungstates; by Kroger3 who obtained silicic acid from a dilute solution of a sodium silicate by means of a Hildebrand cell; and by Codd4. The same type of cell as used by Kroger and by Codd is described in Smith’s “Electro-analysis” p. 303. The method consists in the use of a rotating platinum anode and a mercury cathode. The difficulty lies in efficient stirring and mixing of the mercury used as cathode, otherwise the sodium amalgam formed redissolves in the silicate solution with little or no diminution in the alkalinity of the solution. Of course, some of the amalgam finds its way into the outer dish and is there decomposed by the water, but some remains in contact with the silicate solution and redissolves therein, especially in the more concentrated solutions of low ratio, e. g. in a 2X, I : I solution. This necessarily makes the method slow; the fact, with a current density of 0.04 amps per sq.cm. in the above solution, it required 24 hours to remove ~7~ of the alkali. This difficulty may be overcome in two ways, either by removing, as it is formed, the sodium amalgam from contact with the silicate solution, or by increasing the C.D. so that, the rate of formation of amalgam is greater than its rate of solution in the silicate solution, The first way is very difficult of satisfactory solution, as the amalgam being so light in comparison with the mercury, floats on its surface, and the mechanical devices tried whereby the mercury in contact with the solution is completely changed, often failed to remove it. Practically the only satisfactory method found was to siphon off the silicate solution, charge the cell with fresh mercury, and recommence electrolysis. This was effective but naturally cumbersome, and would be impracticable for use on a large scale. The second method of increasing the C.D. considerably hastens the removal of the alkali and gave good results; but it has its limitations in the fact that, above a certain limit, increase of C.D. causes separation of solid silica on the platimun anode. This limiting C. D., above which silica separates on the anode, varies not only with the dilution but also markedly with the ratio. The more concentrated the silicate solution and the greater the proportion of silica in the ratio, the lower must be the limiting C.D. With a 2N, solution of ratio I:I or 1:2 a C.D. of 0.044 amps per sq. em. scarcely diminishes the alkalinity of the solution; a C.D. of 0.15 a,mps Kolloid-Z. 31, 36 (1922). Kolloid-Z. 30, 346 (1922). 3 Kolloid-Z. 30, 16 (1922). hppl. Chem. Abs. 1924, 56.
I158
R. W. HARMAN
per sq.cm diminishes the alkalinity quite rapidly but yet does cause separation of solid silica. With ratio I :4 a 3KWsolution gave a very thick deposit of silica with a C.D. of 0.11 amps per sq.cm. A zK, solution with a C.D. of 0.11 also gave a thick deposit but with a C.D. of 0.044 amps per sq.cm., although silica was deposited on the anode, a t the end of 4 hours the solution on analysis was found to be o.SN, and its ratio was 1:5.2. This solution was very opalescent and after two days set to a gel and later on exhibited synaeresis. A 0.5 N, solution gf ratio I :4 with a C.D. of 0.13 amps per sq.cm. also deposited silica, but at the end of I O hours, during which time the C.D. gradually fell, the resulting solution was found to be 0.08 N , with a ratio of 1:40. This dilute solution showed no signs of gel formation. Thus with ratio I : I a z N ,solution may be electrolysed with a C.D. of 0.15 amps per sq.cm. but with ratio I : 4 a IN, solution gives a deposit with as low a C.D. as 0.044. It has been found possible by this means to prepare 2 pu’, solutions of ratios I : 2 , I K ~of , I :3, and I : 4. Higher ratios than these set to a gel, viz., 1:s above o.rN,, but in very dilute solutions the removal of alkali can proceed until the solution is practically one of pure silicic acid. The alkalinity of these solutions was determined by pipetting off 5 cc. of the silicate solution, diluting with 150 cc. water and then titrating with 0.1 N HCL using methyl orange as an indicator. The end-point in such a titration is quite sharp and since no silica separates in such dilute solution, the method is quite accurate for comparison purposes as required here. Before it was used in an investigation, the final solution was always analysed in duplicate and its concentration and ratio thus accurately found.
Conductivity Kohlrauschl investigated a t I 8°C the electrical conductivity a t concentrations ranging from 6N to o.ooo~Nof solutions of ratios I:I and 1:3.4) and also of a dilute solution of caustic soda to which were added increasing amounts of silicic acid until the ratio was I :3. He found that in dilute solution, the metasilicate was an excellent conductor, but in concentrated solution its conductivity was very low. The more acid silicate was also a good conductor in dilute solution but a very poor one in concentrated solution. Kohlrausch ascribed this high conductivity to complete hydrolytic decomposition of the salt into acid and base, a view which is no longer tenable on account of the results of hydrolysis experiments. He also concluded that combination of base and acid continued until they present in the proportion of I :2 . Kahlenberg and Lincoln2 determined the conductivity of solutions of ratios I :I I : z and I :5, in dilute solution only, as their method of preparation, mixing NaOH with a solution of silicic acid, made it impossible to prepare more concentrated solutions than 0.25 gm. equiv. per litre owing to gel formation. 2
Wied. Ann. 47, 736 (1892); Z. physik. Chem. 12, 773 (1893). J. Phys. Chem. 2, 77 (1892;.
AQUEOUS SOLUTIONS O F SODIUM SILICATES
1159
Their results are in fair agreement with Kohlrausch’s and they too attributed the high conductivity to hydrolytic dissociation, concluding that there was 100% hydrolysis of NazSi03at o.osN, and that the increase of conductivity with further dilution was due to a decrease of the retarding influence of the colloidal silica on the ions.
Experimental A modification of Kohlrausch’s method was employed1, consisting of the use of a double rotating commutator whereby a reflecting galvanometer could be used to indicate the balance point, which was thus found very accurately. The bridge wire, about j metres long, wound on a rotating drum, had 1000 divisions but the sensitivity of the galvanometer allowed a reading of I in IO,OOO. The bridge wire was calibrated several times throughout the course of the measurements. The resistances in ohms of any one solution for about 10-12 different positions of the contract on the bridge wire agreed at at least within I in 100,but generally the agreement for I O readings was within I in 1000. The conductivity cell, of the ordinary type with a tightly-fitting lid, was made of hard glass and was thoroughly steamed and cleaned before use. Since the object of this investigation was not extreme accuracy in dilute solution, the water used throughout the whole of this investigation was not specially prepared conductivity water, but freshly distilled from the laborastill. Its conductivity varied between 1.6 to 2.1 x IO-^ mhos. The cell constant and the conductivity water were determined quite frequently, on an average before every 2 or 3 silicate solutions were determined. The cell was immersed in a water thermostat electrically heated and reulated to 2 5 O C + 0.01, the temperature being registered by a standardised thermometer. The stock solution, i. e. the most concentrated of any series, was kept in a silver flask, and the more dilute solutions were prepared therefrom by adding a weighed quantity of water to a certain weight of the stock solution, These solutions were allowed to stand a t least 12 hours before they were used in a determination, as the conductivity changes a little with their age for the first 1-5 hours after dilution. All weights, thermometers, measuring vessels etc. used throughout this work on silicate solutions were standardised before use. The densities of the various solutions were determined by means of a pyknometer. About 4 density determinations were made for each ratio and the remaining values were obtained by interpolation and extrapolation from the curve so obtained. The solutions of any one ratio could be obtained in three ways,( I) by dissolving crystalline NazSi03.gH20 for ratio I : I and by electrolysing such a solution as previously descirbed for the higher ratios. This way gave the purest solutions. See Fitzpatrick: Proc. Phys. SOC.London, 15, 13 (1896); Whetham: Phil. Trans. 194A,321 (190).
I 160
(2)
(3)
R. W. HARMAN
by adding caustic soda to ratio I :4 to get the lower ratios. by using the commercial solutions supplied by Brunner, Mond and
co. Solutions of different ratios at various Concentrations obtained by any of these three ways gave identical results, both in conductivity and in the other measurements to be described.
Calculation of Equivalent Conductivity Since the concentrations are expressed in weight normality i. e. in gram equivalents per 1000 gms. water, we get for the equivalent conductivity A the expression,X ( 2 0 0 0 x) A ___ Nw. p where X = specific conductivity x = no. gms. solid. in 1000 gms. water Nw = weight normality p = density of the solution.
+
Conductivity Results
T
TABLEI Gm. equiv. S a per 1000 gms. water.
Gms. solid per 1000 gms. €120.
Spec. Cond.
Density
Ratio
X.IO-~
5.868 3.057
145 I ,068 1,034 I ,013 I . 006 I . 003
0.020
I .222
I .OOI
0.010
0.611 0.305
I
1 . I33 0.516 0.204 0.096, 0.050
0.005
1-48.90
69.31 31.54 12.50
I
.o
0.796 0.398 0 . I59 0.0796 0.0398 0,0159 0.00796
92.15 46.07 36.67 18.34 7,335 3.667 1.834 0,7335 0.0366
108.3 85.0 48.8 23 . o 12.6 7.182
I.
3,054
.ooo
1' 5 5 2
I . 000
0.792
Ratio 2.0
1.097 I . 050
1.043 1.021
I
,009
I . 005
2
:I
115.2 85.89 71.96 47.97 22.08 13.83
I .002
7.151
I .OOI
3.052 1.543
1.000
2j°C
I:I
P
2.435
=
Equiv. Cond.
A =
X(IOOO+
Nw. P
44.6 75.9 95 . o 112.7
730.8 143.6 152 ' 7 155.2 15%. 4
x)
1161
AQUEOUS SOLUTIONS O F SODIUM SILICATES
Ratio X 2
.o
I.o
0.5 0.2
0.I
0.05 0.02 0.01 0.005
X. IO-^
P
1.110
0.500
0.204 0.IO0
0.05 0.02
0.01 0.005
X
(ZOO0
233.68 100.43 45.65 18.425 9.130 4.565 I.826 0.913 0.450
I . I80
50.80
21.50
1.085 I .042
37.16 24.44
I ,017
12 '
34.26 49.05 62.59
242.90 121.4j 60.77 24.29 12.145 6,077 2.429 1.214 0.607
I. 190
I .070 I .03j
1.017 1.008 I .004 I .002 I .OOI I .ooo
32.09 50.23
66.76 86.20 99.20 107.04 114.20 118. IO
120.14
1:2
I ,009 I. 004 I .002 I.0 0 1 I .ooo
75
7.27 3.90 1.68 0.947 0.466
84.00 89.50 93.20
39.60 31.83 22.48 11.44 6.634 3 785 I.636 0.8516 0.4490
20.46 31.42 45.41 57.33 66.48 75.63 81.75 85.16 89.80
31.36 23.75 16.08 9.563 5.757 3.284
16.17 23.24 33.I4 48.25 j7.80 6j.80 75.06 81.j4 86.04
72.70
78.00
Ratio I :3 2.0
1.0
0.5 0.2 0.I
0.05 0.02
0.01 0.005
I. 136 I .050 1.022
I.0 1 0 I ,007 I .0 0 1
I.000
Ratio 2.0
I.o 0.j 0.2 0.I 0.05
300,184 I jo.092 75.046 30.018 Ij.009 7.504
'
I .003
I
:4
I . 260 1.175
1.043 1.021
1.011
I .005
,
3.002 I. jor
1.002
I. 500
0.01
1.001
0.00;
0.7504
I.000
0.81j4 0.4302
0.02
+ x)
Nw. P
I , 140
'
62.05 49.94 33.38 17.24 9.920 5.352 2.284 I.I81 0.6007
A =
152.45 76.22 38.11 15.24 7.622 3.811 1 524 0.762 0.381
Ratio 2,450
I :I. j
1162
R. W. HAHMAN
These results are shown graphically in Figs. I and 2 . In Fig. I the equivalent conductivity has been plotted against the concentration, while Fig. 2 shows the equivalent conductivity plotted against the logarithm of the concentration. I n Table I1 the results are collected together, and those of NaOH at 2 5 O C are also given, for purposes of comparison. A few of these results have been found by interpolation from the curves in Fig. I so that the conductances for all the ratios could be compared at the same concentration.
TABLE I1 EQUIVALENT CONDUCTIVITY
liT
NaOH
2 :I
1.0
142.0 172.5
0.5
200.0
0.2 0.1
209.0 214.5
157.32 85.57 107.80 136.90 157.5
0.05
220.0
175.5
0.02
225.5
0.01
227.5
0.005
228.0
190.1 193.0 194.2
2.0
0.0
I :I
57.25 81.25 96.80 112.70
130.80 143.80
152.7 155.0 158.0 160.0
I:I
57.50 81.20 96.5 113.0 130.0 142.6 151.8 156.0 159.0
1:I.j
I:2
32.09
25.80 36.10 49.05 62.59
50.23
66.76 86.20 99.20. 7 2 . 7 0 107.04 78.00 114.20 84.00 118.10 8 9 . 5 0 120.14 93.20 121.00 95.00
T :3 20.46 31.42 45.41 57.33 66.48 75.63 81.75 85.'16 89.90 91.00 I
=
25°C I :4 16.17 23.24 33.14 48.25 57.80 65.80 75.06 81.50 86.04 88.00
Discussion of Results From the curves, Fig. I , of equivalent conductivity against concentration, it is seen that,( I ) the values obtained for ratio I :I, are practically identical with those of Kohlrausch a t all concentrations. (2) ratio I : I , i. e. sodium metasilicate Na2SiO3,has a very high conductivity in dilute solution; (3) ratio 2 : I gives practically the same values as I :I at concentrations 1-2 N,. This is very remarkable. (4) all the other ratios are quite fair conductors in dilute solution, but show an abnormally low conductivity in concentrated solution, especially the higher ratios I :3 and I :4. This high conductivity of Ir'azSiOs is undoubtedly due to the presence of the very mobile hydroxyl ion formed. by hydrolysis, but no exact quantitative conclusions can be accurately drawn from these conductivit,y data as to the extent of hydrolysis. In ratios 2 : 1 , I : I and I : I . ~the low conductivity in concentrated solutions is mainly due to a decrease in hydrolytic dissociation, as a decrease in the concentration of the hydroxyl ion would cause a large decrease in the conductivity. In ratios I :2, I :3 and I :4 where the hydroxyl ion concentration is very low, even in dilute solution, the fair conductivity of dilute solutions points
AQUEO’CTS SOLUTIONS O F SODIUM SILICATES
1163
to a high degree of ionisation and a fairly mobile silicate ion, while in concentrated solution there may be either very little ionisation or there may be complex or colloid formation. In the latter case the osmotic activity would be abnormally low. Equivalent Conductivity at Infinite Dilution The value of the equivalent conductivity at infinite dilution is most difficult to determine. Several extrapolation methods are discussed by Kraus
in his recent book on “Electrically Conducting Systsms” by means of which A m for binary salts may be determined with some accuracy, but no such methods have been worked out for ternary salts, and even then it would be hypothetical to assert that some or all of these ratios correspond to ternary salts. NazSjOs undoubtedly is, but owing to partial hydrolysis the problem becomes complicated. In such cases, only an approximate value may be deduced for A, by extrapolation from the curve where the equivalent conductivity is plotted against the logarithm or the cube root of the concentration. Fig. 2 shows the plot of equivalent conductivity against the log of the concentration for ratios I : I , I : I . ~ 1, : 2 , 1 : 3 , and 1:4. It can be seen that the curves conform to the type for which the extrapolation is considered valid, i. e. they eventually become asymptotic to a line parallel to the concentration axis. KO accurate measurements have been made at very low concentrations but they have been made at sufficiently low concentration to warrant this extrapolation, and the values of A m so obtained should be accurate to within 2 or 3 units.
1164
R. W. HARMAN
The values of HOhave also been obtained from the expression1,-
I/A
=
I/AE
+ K (CA)
by plotting values of I/L against those of (CA)”-’, then taking a value of n such as would make the graph nearly a straigdt line, and also two other neighbouring values of n on opposite sides of this one. These three graphs were produced so as to intercept the I/A axis, and the most probable value of the intercept chosen to repiesent the value of I/ACX.
Fro. 2
The values of A, as found by extrapolation from the log curves in Fig. 2 and by Noyes’ method agreed closely. The final values assigned to ACX are shown below,Ratio A, I: I
I6 0
I: 1.5
I21
1:2
1:3 1:4
95 91 88
Mobility of the Silicate Ion
If we know the concentration of the various ions in the solution and their mobilities at this concentration then the specific conductivity at that concentration may be calculated from the expression,-
See A. A. Xoyes: Carnegie Inst. Pub. 1908, 63.
I 165
AQUEOUS SOLUTIONS O F SODIUM SILICATES
where C denotes the concentration and U or V the mobilities of the ions. The concentration of the various ions, has been found by means of hydroyxl ion concent,ration, sodium ion concentration, and freezing point measurements to be published shortly; and, although, in these measurements of the concentration of the ions, the laws strictly applicable only t o infinitely dilute or ideal solutions have been applied to more concentrated solutions, still the values are sufficiently accurate for the purposes of this calculation. Assigning values of 4j and 180 to the sodium and hydroxyl ions respectively, we can thus calculate the mobility of the silicate ion for each ratio, with the following result,Ratio Mobility silicate ion (dilute solution) 1:1 60 1:2
35
I :3
43
1:4
41
It seems justifiable to assume the presence of only three kinds of ions viz., sodium, hydroxyl, and silicate ions formed by hydrolytic and ionic dissociation, and knowing their mobilities and the extent of hydrolysis, Am for each ratio may be approximately found. In this way are obtained the values below, those found by extrapolation from the log curves being also given for camparison. Ratio
Aa (mobility)
Aa (extrapolation)
I : I
165
I60
1:2
95 90
95 91
88
88
1:3 1:4
In t,he higher ratios I : z , I :3 and I :4where the hydroxyl ion concentration is practically negligible the agreement is good. In the metasilicate, as would be expected, owing to the distrubing influence of the hydroxyl ion making the application of the mobility method doubtful, the values are not so close, but are in sufficiently close agreement to warrant the conclusion that the mobilties, as given above, are fairly accurate. The most interesting conclusions are obtained, however, when the equivalent conductivity is plotted against the ratio. This is done in Fig. 3, where the values of NaOH are given under ratio I :0 . Considering dilute solutions first, it is seen that, :(I) the equivalent conductivity of a caustic soda solution to which silica is added falls linearly and rapidly until the ratio I : 2 is reached, there is a sharp change of direction at I : 2 . (2) (3) there is no change of direction at I : I , i. e. NazSiOa. (4) after I : 2 the conductivity again falls regularly and linearly but not nearly so rapidly.
1166
R. W. HARMAN
The very sharp change of direction at ratio I : 2 would seem to indicate the existence in solution of the definite salt Na20.zSiOz.There is also other conclusive evidence, such as the separation from solution of the crystalline hydrate NanSi03.gH20,that a definite salt NazO.SiO2 also exists in solution. The fact that there is no change of direction in the curve a t I:I indicates that the salt NazSiOais largely hydrolysed and that it is the gradual disappearance of the mobile hydroxyl ion that causes the conductivity to decrease linearly as ratio I:Z is approached.
FIG. 3 Equivalent Conductivity against Ratio
The fact that beyond ratio 1 : 2 further addition of silica causes the conductivity to decrease regularly, but only slightly, indicates, either : (I) that the repression of the degree of hydrolysis is only very slight, or, ( 2 ) if the hydroxyl ion practically disappears a t these ratios, then the silicate ion in ratio I :4 either, is more mobile than that at I :3, or its mobility is the same, but it is there in greater concentration. Considering the concentrated solutions, we notice :, of direction begins to appear a t 2 : I , which be( I ) a t ~ . I aN change comes more pronounced the more concentrated the solution. at 0 . jN, a change of direction appears at I : I which becomes very (2) pronounced at higher concentrations.
AQUEOUS SOLUTIONS O F SODIUM SILICATES
I
167
(3) at 1.ONV-2.0N, the conductances of ratios 2: I and I : I are practically the same. This is very remarkable. (4) the change of direction at I : Z is still quite definite except for the highest concentration, but it is not so sharp as in dilute solution. ( 5 ) beyond I : 2 the slope of the curve is the same as in the more dilute solutions. If a change of direction of the curve indicates the existence of definite salts at that point, then there are salts of the compositions 2NaZ0.SiOz,KaZO SiOz, and NazO. 2Si02. The existence of the last one appears very probable and there is good evidence that NazSiOs also exists. However, one would not be justified in concluding from these curves alone that these three salts exist here in solution; especially the salt 2NazO.SiO2. Ratio 2 :I may be a mixture of NaOH and NazSi03; both of these salts in solution give rise to hydroxyl ion and when mixed the total hydroxyl ion concentration will not be equal to, but less than, the sum of the separate hydroxyl ion concentrations, and so one would expect diminished conductivity on this account. This point is further discussed in a later paper on hydrolysis. It is worth noting that the fall in conductivity from NaOH to ratio I : 2 is practically constant for all concentrations, being I 20-130 units. In concentrated solutions above ratio I : 2 we note the same behaviour as in dilute solutions, but as here the conductivity is abnormally low, and as the mobility of the silicate ion is moderately high, we can conclude that the osmotic activity of these concentrated solutions must be very low. We should, therefore, expect not only little ionisation but probably the formation of complexes or aggregates, and that these solutions would exhibit colloidal properties.
Summary ( I ) Crystalline NazSiOs has been prepared, and from a solution of this salt, various other ratios NanO:Si02 have been prepared by removing the alkali electrolytically. (2)
I :z, I
Conductivity measurements have been made of ratios 2 :I, I :I,
I :I.5 ,
:3. and I :4 a t concentrations ranging from 2N, to o.oo5Nw.
(3) Solutions of 2 : 1 and I:I are excellent conductors; I :2, I : 3 , 1:4 are good conductors in dilute solution but abnormally low in concentrated solution. (4) Hydrolysis into NaOH and colloidal silicic acid is not sufficient to account for this conductivity and calculations have been made of the mobility of the silicate ion.
( 5 ) On plotting equivalent conductivity against ratio, there is only one change of direction and that a very sharp one at I:Z in dilute solution; but in concentrated solution changes of direction also occur at 2 :I and I :I.
I 168
R. W. HARMAN
(6) It appears likely that salts corresponding to I :I and I : 2 exist in solution; and that the other ratios are mixtures of these with NaOH or “hydrated” silica as the case may be. ( 7 ) In concentrated solutions either the extent of ionisation and hydrolysis is very low, or aggregate or colloidal formation takes place. I wish to thank the Commissioners of the 18j~ Exhibition for a Scholarship which has enabled me to carry out this investigation, and to express my gratitude to Professor Donnan, at whose suggestion this work was undertaken, for his constant kindly interest and advice. T h e Ramsay Laboratories of Physical and Inorganic Chemistry University College, London. J u n e $2, 1925.