M. F. BECHTOLD
532
Vol. 59
POLYMERIZATION AND PROPERTIES OF DILUTE AQUEOUS SILICIC ACID FROM CATION EXCHANGE BYM. F. BECHTOLD Contribution No. 8662 from the Chemical Department, Experimenlal Slation, E. I . du Pont de Nemours & Company, Wilmington, Delaware Received December 88,1964
.
Dilute aqueous silicic acid prepared by cation exchange of sodium silicate has been studied with regard to its polymerization kinetics, gelation and titration characteristics, copolymerization tendencies, and precipitation with organic compounds. When first formed, the silicic acid is of low molecular weight and in good solution in water. The kinetics of its polymerization as determined by light scattering and freezing point experiments can be accounted for by an ideal bimolecular condensation of a “monomer” with a n apparent functionality of slightly over two (2.0004-2.05). Although some divergence from the ideal, such a8 cyclization or unequal functional group activity, is indicated and may in part account for this unexpectedly low value, the data indicate that the actual functionality cannot be as high as 3. The silicic acid solutions are also characterized by an optimum pH of about 5.05 for fastest gelation rtnd by a sharp titration inflection point at a pH of about 4.65. Polymerization shifts these values to the vicinity of p H 5.85 and of pH 5.30, respectively, and this apparent decrease in hydrogen ion concentration during polymerization is independent of the titratable strong acid factor. The cooperative action of silicic acid in producing non-gelling or slowly gelling cation exchange effluents with certain other gelable metal acids and hydroxides is explained by copolymerization. “Nalcite” MX, the bottom of the tube being fitted with a fine mesh screen over the outlet tube, which projected into a receiving flask. The top of the column was fitted with a separatory funnel t o contain the influent solution, which was drawn through the column a t the rate of 1-2 liters per hour by reducin the pressure in the receiving vessel, which was a vacuum fiftering flask. The following procedure was adhered to throughout: (a) The column was back-washed with 2 liters of HeO, (b) down-washed with 960 cc. of 2% (by weight) H2S04, (c) down-washed with 2 liters of H2O. The resin was then let stand covered with HzO until used. (d) Diluted sodium silicate solution or other influent containing less base than the “break-through” quantity, equivalent t o 6.7 g. Na20, was passed each time, displacing the water covering the resin into the receiving vessel. Commonly used effluents were prepared from the following influent compositions: for standard silicic acid effluent50.6 cc. Grasselli 20-WW silicate 930 cc. HzO, for standard metasilicic acid effluent-28.4 g. NazSiOs.SH20 983.8 g. H20. B. p H Determinations and Titrations.-The Beckman Model G instrument was used with the standard glass elec“BOIS.” trode-calomel couple with temperature compensation and calibration with buffer solutions. Materials C. Analytical Methods.-In analyses for SiOz, the sample The samples of commercial sodium.silicate solutions used was heated with concentrated HCl, filtered, ignited, in this study werc obtained froni the Grasselli Chemicals weighed, then checked by volatilization as SiF,. Sodium Department of the du Pont Company, and designated as oxide was determined gravimetrically by precipitation with Grasselli 20-WW; the followin analysis was typical: den- uranyl zinc acetate. In some instances, which are indisity a t 25’ 1.388 g./cc.; Si%z, 28.13%; Na20, 8.48%; cated, water was determined by titration with the Karl A1203,0.094%; Ti02, 0.006%; Fe, 0.005%; and H2O (by Fischer Reagent. Separations of alumina and silica were difference), 63.28%. performed by digesting with HC1 to dissolve alumina folSodium metasilicate nonahydrate ( Na2SiOa.9H20) J. T. lowed by filtration of silica, which was analyzed as usual. Baker and Company C.P. rade was used. At 0.1 molal D. Refractive Index.-A Zeiss-Pulfrich refractometer in water, the freezing point fowering (0.54’) was equivalent and a Bausch and Lomb dipping refractometer were used to about 2.9 moles of particles per formula weight. to determine the refractive indices ( n ) of various solution8 “Nalcite” MX resin (National Aluminate Corp.), re- containing silica a t weight fraction (w). Stepwise dilution ported to be a condensation product of formaldehyde with of an aqueous silica system (concentrated from standard 0-, m- and p-phenolsulfonic acids, was used. On a dry silicic acid effluent by boiling off water) with succevsive rebasis, this resin contained 0.77% N and 65% was between fractive index measurements led to a value a t 23” and 35 ai 54G0.7 lngstroms of dnldw as w + 0 of 0.0722 f 0.0015 TI with the Zeiss-Pulfrich and 0.070 =I= 0.003 with the dipping 0.00065% Si02 -and- 0 refractometer. Since dn/dw 4 dn/dC in dilute aqueous wlutiori, a value of 0.072 was used for dn/dC, where C is without, purification for usc as &c rcgcnerant for the catioua g . solute/cm.a of solution in all calculations, except in one exchange columns. instance in which a direct measurement on a polymerizing system showed a value of 0.065. Although the choice of dn/ Method d C within the range 0.065 to 0.072 influences the absolute A. Cation Exchange.-Vertical “Pyrex” brand glass values of molecular weight calculated later, there is little tubes (22” high X ca. 1”dia.) were packed lightlywith 140 g. influence on the value of functionality calculated from the kinetic data. (1) P. G. Bird, U. S. 2,244,325(June 3, 1941). E. Light Scattering and aw.-Right-angle light scatter(2) B. H. Zimm, R . S. Stein and P . Doty, Polymer Bull., 1, 90 ing measurements were made with a Zeiss-Pulfrich mechani(1945). cal photometer equipped with a turbidity head, which was (3) W.D.Treadwell, Trans. Faraday SOC.,81, 297 (1935). similar to the atmaratus of ref. 8 with the following exi 4 ) R. W.Harman, Trire JOURNAL, SO, 359 (IS?O). cept,ions: (1)the collimating lens of the apparat,us was-close ( - 5 ) C. B. Hurd and P. L. hlers, J . A m . Chern. Soc., 68, G1 (1946).
Introduction
The advent of cation exchange of alkaline silicate solutions as a convenient source of substantially salt-free silicic acid‘ and the development of light scattering2 for the estimation of weight average molecular weights of dissolved organic polymers suggested an investigation of the nature of silica in water-soluble or aqueous dispersion form. In general, other in~estigators3-~have explained the chemical and physical properties of silica “sols” by ‘considering them as composed of crystalloid a d colloid phases and by separating the ordinary chemical reactions of the crystalloid phase from reactions tending to build up colloidal particles. It seemed desirable to test the concept that a study of the polymerization reaction could account for most of the physical and chemical properties of silica
R. C.hlerrill, J . Chem. Ed., 24,202 (1947). (7) E. .4.Hauser, ‘THIS JOURNAL, 62, 1105 (1948). (6)
+
+
(8) P. Doty, B. If. Zilnni and H. Mark, J . Chem. Pirys., l a , 44 (1941).
June, 1965
P O L Y M E R I Z A T I O N AND P R O P E R T I E S OF SILICIC
to the light source with only a window a t the cell chamber entrance, ( 2 ) no diaphragm was used a t the cell window entrance, instead (3) filters were used a t this position instead of a t L in the reference apparatus. Light scattering measurements were made on silicic acid solutions at SiOz concentrations of 2 4 . 0 5 % . Each measurement was preceded by settling out or flotation (as judged by constant turbidity readings) of any large foreign particles acquired during cell filling or during dilution of the sample ih the cell. When turbidity ( 7 ) changes with age of silicic acid solutions were followed, fresh samples were withdrawn from the stock solutions for each-measurement, and, if weight average molecular weight ( M w )measurements were desired, the samples were diluted appropriately in the turbidity cell. The C/T data were calculated from the measurements on the diluted soluJtions and the initial Si02 analysis. Then, the values of M,, B (the osmotic coefficient) and p were calculated therefrom by the least Lquares method using the 2 B C / R T and p = selations from ref. 8, H C / r = l / M w (0.5 - Bd22Ml/RTdl), where H i s a function of dn/dC, M I = molecular weight of the solvent, dl = density of solvent and dz is the density of solute; a value of dz = 2.3 g./cm.a was selected from measurements of the partial specific volume of Si02 in the silicic acid/HzO systems. The symbol Mw* is us_ed with the reported data t o indicate the "apparent" Mw. Many of the following Lncertainties, which may result :i only fair correlation of Mw* with the absolute value of M,, are later shown to be circumvented or of second order for the arguments presented: uncertainty in dn/dC for calculation of H , uncertainty in absolute measurement of turbidity, no correction for interference due to large particles9 (dissymmetry of scattering), no correction for any possible depolarization, uncertainty of application of fluctuation theory to charged particles in a polar liquid,1° presence of foreign light scattering particles, the polydispersity of Si02 phase, and changes during polymerization in nature of particles and size distributions due to stirring, catalysis, dilution, or concomitant effects, such as lowering of pH. F. Freezing Point Lowering and ii?,.-The freezing points of the silicic acid solutions were determined as follows: to a one-pint Dewar flask packed in Filter-Cel and equipped with a Beckmann thermometer, inlet funnel, outlet pipet, and mechanically driven glass uplift screw stirrer through the cork was added 140 g. of shaved ice, then 200 cc. of icewater. If the apparatus was previously near 0", equilibrium was established within two minutes; if at room temperature, 5-10 minutes were required. After a reading of the control point, the water was removed by pipetting and was replaced with 200 cc. of the sample near 0'. The difference between the minimum temperature, which was attained within 1-10 minutes and which increased very slowly, and the minimum for water was recorded as Af.p. A sample, usually 50 cc., was then withdrawn for analysis to determine Si02 concentration. The control point in this apparatus was slightly dependent on room temperature, increasing by ca. 0.002' per I", increase in room temperature in the range of 20-30". The calculat,ions of number average molecular weight were made from the formula A?, 1000(1.86)W/Wo(Af.p.) where WJVOis the weight ratio of solute to solvent. The symbol Mi*is used with the data reported to indicate "apparent" M,. A value for the minimum possible M , in a given solution was calculated by assuming all Af.p. _due to Si02 in the sample. A maximum possible value of M , was calculated in some cases by assuming all change in Af.p. with time aa due t o polymerization of silicic acid. G. Gelation.-The incidence of non-pourability of a sample in a bottle about two square inches in cross-section was taken as the gel point.
+
(an)
Theoretical Polarization Kinetics.-The following derived relationships are applied to "silicic acid," assuming it to polymerize by condensation. These have been (9) J. Waser. R. M. Badger and V. Sohornaker, J . Chem. Ph2/8., 14, 43 (1946). (10) P. Doty and R. S. Steiner, ibid., 17, 743 (1949).
ACIDFROM
C A T I O X EXCH.4NGlC
533
published separately," and are repeated liere for convenience (cj. ref. 12)
li?, =
[l
1 + 2fCoKt + (2f - f W
1
(1)
Co is the initial concentration of monomer of molecular weight M o with f functional groups conden,ing bimolecularly with rate constant K , and Ad, and JZn are the weight average and numlxx average molecular weights at polymerization time 1. The calculations of f and K values-giving curves fitting the experimental curves of M , us. i can be performed by the trial and error method1' or, as was done in this work, by selection of two points on the experimental curve, say the mid-point and the highest point, then, using these two sets of coordinates in equation 1 to eliminate K and solve for f. After f is obtained, this value is reapplied to one of the points to find K . Other points on the curve are then calculated from this set of values for f and K . These equations are restricted to those cases in which the activity of the functional groups is not affected by the size of the molecule to which they are attached, functional groups on the same molecule do not react with each other (no cyclization), and the reverse reaction and the weight loss in condensation are negligible. The equations are valid only up to gelation, which OCCUJS only when f > 2. Since at gelation time, tgel, M , = a , the denominator of (1) must then be equal to 0, whence tgel
=
1
- CKO(2f - j - 2 )
(3)
an
Plots of equations 1 and 2 relating 11.1, and with t for various values off, letting C0K = 0.1, are shown in Fig. l.'3,14It is also obvious from equation 3 that violations of the restrictions on equations l and 2 that would perturb the value of K(2f - f2) upon dilution, such as cyclization, could be detected in gelling systems without knowledge of the actual values off or K . In this case, a plot of tgel vs. l/Co would not yield a straight line, unless there are compensations among the factors present. Results
A. Silicic Acid from Sodium Silicates.
1. Characterization of Influents.-Preliminary light scattering measurements on the standard influent (without filtration) gave a turbidity of 1.5 X em. -l a t an age of one-day, and a plot of C/Tvs. C upon dilution leads to M,r of 33,00015~*6 which is interpreted to be the maximum possible for the silica in this solution, since other light scattering particles were probably present. 2. Properties of EfRuents. a. ATw*, p , B.Preliminary light ssattering results on standard effluents led to an M,. of 44,500 a t an age of one day and, in another instance, 21,400 a t an age of two days. Thus, it appeared that ion exchange (11) M. F. Bechtold, J . Polymer Sa.,4 , 219 (1944). (12) G. Oster, J . CoEloid Sei., 2, 291 (1947). (13) G. S. Hattiagandi, ibid., 3, 207 (1948). (14) M.Prasad and K . V. D . Doss. ibad., 4, 349 (1949). (15) P . B. Ganguly, T H I SJOURNAL, SO, 70R (1926). (16) P. Debye and R. V. Nauinan, J . Chem. Phus., 17, 1364 (1949).
M. F. BECHTOLD
534
VOl. 59
the more turbid effluent in Fig. 2 were diluted with results shown in Figs. 3 and 4. The upward curvature of the line relating C / r us. C for high values of C in the two oldest solutions may be due to breakdown upon dilution with stirring of very large crosslinked molecules, which increases their scattering power due to decrease in interference. It might also be related to the polydispersity. Calculation from the data of the points of Figs. 3 and 4 led t o 150
100-
I 5
0
I
I
10
15
TIME.
Fig. 1.-Molecular weight vs. time of polymerization (after equations 1 and 2, letting COK = 0.1).
was accomplished with little or no change affecting ATw in the region of 30,000. A determination of turbidity us. time is shown in Fig. 2 for one effluent up to 30 days and for a low sodium effluent up to 21 days, at which time gelation was incipient. For the less turbid effluent, gelation was incipient a t 14.5 days and, at complete gelation (24 days), an inflection point in the curve was found.
0.005
O.OI0
0.015
0 20
C(q/crn.SI.
Fig. 3.-C/r vs. C in progressive dilutions of samples of 1.78% silicic acid withdrawn at various polymerization times.
150-
t'
9
-
-x
-
I
E
2 100-
* t o m
-
LL L 3
-
50
20
0
10
20
30
T I M E (DAYS).
Fig. 2.-Turbidity in aqueous silicic acid us. polymerieation time a t room temperature: .,1.83% SiOz; 0, 1.78% SiOz, SiOz/NazO = 4450, pH ca. 4.5.
a,+
I n order toTdetermine the course of with time a t room temperature, samples removed from
Fig. 4.--C/r vs. C in progressive dilutions of samples of 1.78% silicic acid withdrawn a t various polymerization times.
the values plotted in Fig. 5. The decreasing slope (dotted line) past 10 days is probably due t o interference, since itfwis actually equal t o co (gelation) at less than about 21 days. The dashed line
POLYMERIZATION AND PROPERTIES OF SILICIC ACIDFROM CATIONEXCHANGE 535
June, 19.55
aw
isldrawn to show the more probable course of vs. b.. T o determine it carefully an appropriate cormction could be made for interference in light scatSered from the large molecules present.
I
I
0.5XI8t
/A
I I
10-4
I
I
I
3
5
7
TIME (DAYS).
I
Fig. 6.--Bw* us. polymerization time calculated from as, M D = 47, 100, and (ii); 0, MD= 60. sumptions (i): . Curves calculated from equation 1 using values off and I< indicated, zero weight loss and dn/dC = 0.072.
0I
/
= 2 +*x
I
1
0
I
10 T I M E (DAYS).
I
I
20
Fig. 5.--aW*us. polymerization time calculated from data of Figs. 3 and 4: 0,calculated values; - - - - - , probable course, if corrected for interference.
and a value of dn/dC actually measured on this system was used. These assumptions (iia) lead to a value of f that is again just slightly over 2. The straight line representing f = 2 and the curve €or f = 2.0448, for example, both of which fit the endpoints of the data are obviously far off from the calculated experimental points, which are on the curve determined by f = 2.00041, I< = 6500. An approximate correction within the scope of assumption (iii) was made by extrapolation of the data of Fig. 5 to M n = 60 and shift of the time scale. The
I n addition t o the simple assumption (i) obtained by extrapolation in Fig. 5 that 1 M O of the Si02 species is about 47,100 two other general assumptions, both of which are later shown to be more rational, are possible: namely, (ii) all initial solute turbidity in excess of that for an aqueous solution of the siliceous species of degree of polymerization one is due t o foreign particles, or, (iii), this excess turbidity is due t o polymerization of the siliceous species prior to or during the ion exchange. Attempts were made to analyze the data on the basis of these extreme assumptions in order to establish limits on the nature of the early stages of polymerization. Approximations within the scope of (ii) were made by subtracting the turbidity a t zero time from later turbidity values. The calculated points are indicated with the best fitting curve calculated for equation 1, with appropriate values of f and K . In Fig. 8, a comparison is made with the early part of Fig. 5 in which interference does not play an important part in the results. The results show assumption (ii) to lead to much more linear reaction than assumption (i), and no interpretation of the results shows an apparTIME (DAYS). ent value off of as much as 3. I n Fig. 7 are shown Fig. 7.-fiw* us. polymerization time calculated on the the results of a further approximate correction of the data (for weight loss during polymerization) in basis that the average polymer unit is which the ATw*was calculated on the basis that the average polymer unit is -0i-(MD = 7 8 )
p"
OH -0-
i-
I
QH
AH (Mo-78)
and dn/dC = 0.065 (for SiOz). Curves calculated from equation 1 with values of f and K indicated, Only solid line fits all experimental points,
M. F. BECHTOLD
536
Vol.
data are, of course, inadequate for an accurate extrapolation of this kind, but the one made, which shifts the time scale t o two days prior to the observed to, is probably not too abrupt, since it was made by a straight line extrapolation of the last two points. Even this shift is adequate to bring down the value off for best fit to a value less than 0.001 above that for a linear reaction, ie., 2.
iio
lysts of constant total activity, the relative positions of the curves would be shifted from that of the uncatalyzed reaction. However, the value of K would be altered to accommodate this possibility, and it is not expected that the general shape or the value off for best fit would be appreciably changed. If the upturn from an initial straight line of the curves of .Mw* us. t is caused by an increasing amount of positive catalyst (such as increasing concentration of OH-) or by a decreasing amount of 0.060 negative catalyst during polymerization, the actual value o f f is, of course, nearer to 2 than to the apparent values. $0.040 The values of f, K , tgel and .dTn*gel calculated v from light scattering data are shown in Table 111 along with the assumed value of M oused. b. Mn*.-Figure 8 shows Af.p. for typical a 0.020 standard effluents us. time. The f.p. lowering decreases to about 20% of its initial value within 5 days and changes little thereafter. This beI I I havior can be explained by assuming that the 0 5 10 15 20 silicic acid polymerizes and _that, within a short time, the re-sidual change in Mn involves such large Time (davsl. . Fig. %-Freezing point lowering vs. polymerization time: values of Mn that no further change in f.p. is ob0,1.89% SiOz, SiOz/NazO = 760; 0 , 1.845% SiOn, SiOz/ servable by the method used. The residual Af.p. a t NazO 3690. advanced age amounts to about 0.012', which must A combination of assumptions under (ii) and be assumed to be the sum of the Af.p. due to all (iii) probably prevails over that of (i). Most impor- silicic acids present plus the Af.p. due to foreign tantly, all of these reduce the apparent functional- particles and any systematic error. This residual ity to a slight but definite value above 2, namely, Af.p. is about 10 times that expected from 'the Na+, OH-, 2.0004 to 2.0009, with the corresponding value of K concentrations of foreign particles (Hf, SOe-, etc.) known to be present plus a reasonable in the range of 6500 to 3200. The values of ATw* calculated from the experimental data under vari- allowance for systematic error. In view of this and conditions discussed later, an appreciable back ous assumptions are given in Table I. reaction (hydrolysis) is suspected. TABLE I If the extreme assumptions are made that; (iv) all residual Af.p. is due to siliceous materials or (v) VALUESOF aw* CALCULATED FROM DATA OF Fraq. 3 A N D all residual Af.p. is due t o foreign materials plus UNDER VARIOUSASSUMPTIONS systematic error, the results-shown in Table IV are obtained by calculating Mn* from Af.p. xalues 0 from the smoothed curves of Fig. 8 (up to Mn* of (47,100) (60) (78) (GO) 0.111 47,800 1 140 10,000) and, for f and K , fitting them to equation 2. 1. 125 63,000 11,700 This assumes, of course, that all restrictions of 2.111 47,800 equation 2 apply. The time scale was shifted by 3,125 63,000 extrapolation to MO = 60 for 1 = 0. This gave a 3.135 147,000 110)000 210,000 to of 0.2-0.5 day prior t o the apparent preparation 5.135 147,000 time. 7.125 1,356,000 1,290,000 2,758,000 It was also shown that use of an M o of 78 to cor9.125 1,35G,OOO respond more nearly to a linear reaction or the use 18.125 4,093,000 4,080,000 of M o as 60 or 600 without shift in the time scale does 21.125 5,174,000 not change appreciably the general trend of the reAssumptions (iv) and (v) both lead to valThe values of B and p calculated from light scat- sults. of 11 2 for the number average degree of polytering data under the various assumptions are ues merization of silicic acid in the freshly prepared efshown in Table 11. These values of p are in the range of 0.40 or less for fluents and (v) leads to a value off slightly over 2 BW* values of about 100,000 or less, which is char- and t o a reasonable gel time of 26 days for the acteristic of linear polymers dissolved in good sol- 1.8991, solutions. Assumption (iv) is absurd, if all vents. This supports the contention that the poly- restrictions on equation (2) prevail, because the mer is nearly linear, i.e., not highly cross-linked. effluent did gel (30-36 days) and this requires the The tendency of p t_oapproach 0.50 asymptotically actual gnvs. 1 curve to be concave upwards. This at high values of M,"* indicates approach of sub- situation, considered with the large residual Af .p. stantial insolubility. This behavior is expected at observed, supports the contention of appreciable very high molecular weights for spherical polymer hydrolysifl of the polysilicic acid. Assumption (iv) is also of interest because it molecules. I n the event the polymerization has occurred in permits calculation of a minimum value of ATn* a t the presence of unknown positive or negative cata- advanced polymerization time, which for the data
+
'CI
I
I
"
I
=i
*
.Jiinc, 1955
POLYMERIZ.4TION .4ND PROPERTIES O F ~ I T J I C I C,i\CID FROM
CATION
537
EXCHANGE
TABLE I1 \'ALUEb O F
B
A N D f l C(ALCULATED P'ROM DATA OF E'IGS.
3
dNU
4
UNDER LrARIOUb -4bbUMPTION6
TinLe (days)
(iii)
0 0.111 1.125 2.111 3,125 3,135 5.135 i .I25 9 . 12Fi 1 3 . 125 21.125
53 7.5 0.13
3.85 2.68
- 1 ,570 0.210 0,4950
0.350 0.395
3.55 2.68 0.190
0.081
0.0545
0.350 0.395 0.4922
0 4968
0.4070
0 . -4968
0.4981;
0,4993
0.4022
0.199 0,0824
0.0191
0.0xc,
0,082.1 0 . ox50 0.00845
0,49c,8
0.4981; 0.4!)97
TABLE I11 VALUES OF f,K , (i)
Mo used
f
&I
AND
&?**gel
CALCULATED UNDER VARIOUS (ii) (iia)
47,100 2.88
60 2 8 X lo-*
78 2 4.1
201,300
4400 8.0 150,100
6500 8.11 380,000
+
+
ASSUMPTIONS
x
(iii)
10-4
60 2 8.8 X
+
3206 9.04 136,600
TABLE IV ESTIMATES OF f,K ANI) .@ne FROM Af.p. Assumptions Effluents
1.89% SiOz 1.845% Si02
concave downward
concave upward
of Fig. 8-is about 2700. Since, from equations 1, 2 and 3, Mnge1= A10[(2 - 2f)/(2 - f)], a maximum value off can be calculated as follows: (2 - 2f)/ (2 - f ) > 2700/60, whence f < 2.047. A minimum value of f cannot be calculated from the present data in this manner but, since gelation occurs, it must be over 2. Consequently the values of 2.0008 and 2.00041 obtained by -various assumptions in the independent study of M," seem more plausible than the value of 2.88 attained by the extreme assumption of M , = 47,100. It is of interest to calculate by equation 2 the course of M,* vs. t using values off and K obtained under the study of The results show much steeper rise in with time than is found under assumption (v). This emphasizes the need for greater precision in Af.p., and determination of the exact contribution of siliceous and foreign materials to residual Af.p. c. Gelation.-A closer examination was made of the phenomena of gelation of silicic acid effluents in light of equations 1, 2 and especially 3. In general, the freshly prepared effluents were adjusted in pH or in polymerization catalyst content, divided into fractions, which were diluted as desired, and then gelation time was observed. Some of the results are summarized in Figs. 9 and 10. These show gel times of a few minutes to more than three months. Freezing points determined on some of the solution showed corresponding abrupt and slow changes. There is no exception to the upward curvature of all plots of tgel us. 1/C. Straight line plots passing through the origin and tgel for the most con-
a;*.
an*
550 550
650 750
1.997, 98.0
2.022, 45.2
centrated solution are expected, provided the restrictions on equations 1, 2 and 3 prevail and the polymerization environment is constant for all dilutions. The use of materials found to be polymerization catalysts, such as C u f f and F- and buffers, did not alter the upward curvature obtained. One possible explanation of this result is thati cyclization is occurring and constitutes a greater percentage of the condensation in dilute solution, thus giving an apparent functionality closer to 2 than the real value of f. Another explanation would be that polymerization is essentially linear ( f = 2) and either an impurity or a small fraction of any remaining functional groups is responsible for the crosslinking reaction leading to gelation. An experiment showing no net change of pH during polymerization in the presence of CuClz is shown in Fig. 10. However, the most dilute solution had a pH lower by 0.31 than the concentrated solution. Better pH control was obtained with an effluent buffered with acetic acid/sodium acetate (Fig. lo), in which the pH ranged from 5.10 to 5.17 and did not change perceptibly during polymerization. The dashed line shows the theoretical course of tgel, if restrictions on equation 3 had prevailed, Further experiments on fast-gelling, buffered solutions were made for a more direct correlation of observed gel times and gel times calculated from equations 1 and 3. The error caused by the time interval between dilution of a sample from the stock solution and the measurement of its turbidity was minimized and in addition the usual practice
M. F. BECHTOLD
538
VOl. 59
(3.12
O‘OJ
/ / /
/ I /
/ 1
/.// /
I
/
7,
Fig. 10.-Gel time vs. ~ / C for O buffered and catalyzed silicic acid solutions (initial, final pH in paren.): 0 1.80% SiOz, acetic aid/sodium acetate buffer; 0, 1.95% SiOt, 0.051% CuC12.2H20.
Fig. 9.-Gel time us. l/Co for various silicic acid solutions (initial pH in paren.): 0 , 1.74% SiO2, 0.0105%. Cu++ (ez CuC12); 8, 1.79% SiOz, adjusted to pH 5.12 with sodium silicate; 8 , 1.80% Si02; 0, 1.83% Si02, 0.0057% F- ( e x NaF).
was followed of allowing coarse particles and bubbles in the turbidity cell to settle out or float off. Changes of turbidity due t o this cause in some solutions resulted in a very slight initial decrease in turbidity with time, followed by a steady rise in turbidit’y due to the progress of polymerization in the solution. The turbidity of the sample a t the instant of dilution was estimated by extrapolation of the turbidity vs. time curve (the part just after settling and flotation effects have ceased). - Values off, K and tgel were calcula’ted from the iM,* us. t curves obtained. The results obtained in this manner with four solutions are summarized in Table V. TABLEV POLYMERIZATION A N D GELATIOXIN FAST-GELLING, BUFFERED SILICICACIDSOLUTIONS Stock soh., % Si02 1.88 1.88 1.88 1.76 pH (initial) 3.60 3.60 3.60 3.35 Vol., 1. 0.6 0.6 0.6 1 1M NaOAo, cc. 5.4 12.6 16.2 21 1 M HOAc, cc. 12.6 5.4 1.8 9 PH 4.32 4.93 5.50 5.01 (f - 2) x 106 23.5 0.9 ... 4 K x 0.525 49.4 ... 0.594 t r e i (calcd.) 0.225 d. 0,0625 d. .,, 0.117d. t g e i (obsd.) >0.33, 0.135d. 0.042d. 0.194d.
