ULTRAFILTRATION THROUGH CELLOPHANE OF POROSITY ADJUSTED BETWEEN COLLOIDAL AND MOLECULAR DIMENSIONS1 JAMES W. McBAIN AND R. F. STUEWER Department of Chemistry, Stanford University, California Received June 11, 1956
Some years ago we found that the pores of commercial Cellophane swollen in water happened to be of such size that ordinary molecules, such as sucrose or anthracene, passed through freely, while all known colloids were partially or wholly held back (5). Now we find that Cellophane, Sylphrap, and Viscacelle, as a t present manufactured, possess distinctly finer pores, so that a portion even of sucrose or potassium chloride is held back on ultrafiltration. Such cellulose membranes possess the great advantage that they are chemically indifferent and that the same membrane can be used for a series of different solutions or solvents without noticeable change in porosity, provided that the transition from one solvent to another is made by washing through with mutually miscible liquids. Such Cellophane serves as a most useful test to prove the presence or absence of colloidal matter, or as a means of measuring the amount of crystalloid present in the colloidal solution. It was very convenient that unlimited supplies of such uniform material were commercially available, requiring no troublesome laboratory preparation. Naturally, care had to be taken to obtain Cellophane, etc., which had not been waterproofed in any way. Formerly we were interested in the problem of adjusting the size of pores down to finer and finer molecular dimensions, so as to separate the ordinary large molecules, such as sucrose or potassium chloride, from small molecules, such as methyl alcohol and water. This waa readily achieved by deposition of collodion or viscose in the pores by filtering the necessary amount of their solutions through the Cellophane, thus producing molecular sieves of any desired fineness. Cellophane which has not been swollen in water is practically air-tight. Now that commercial membranes are denser, we are faced with the opposite problem of finding some simple method of restoring them to the invaluable position they formerly occupied, intermediate between ordinary Presented a t the Thirteenth Colloid Symposium, held a t St. Louis, Missouri, June 11-13,1936.
1157
1158
JAMES W. MCBAIN AND R. F. STUEWER
recognized molecules and the finer colloidal particles. In a paper on dyeing Morton (6) reported that the permeability of Viscacelle is greatly increased by a treatment with sodium hydroxide. In the present investigation, we have found that solutions of zinc chloride of particular concentration are most satisfactory for achieving our purpose. We are greatly indebted to the manufacturers of Cellophane, Sylphrap, and Viscacelle for samples and even for a specimen of Cellophane that never had been allowed to dry. The subject of ultrafiltration is not quite simple. This mas well brought out in a valuable and careful study by Ershler (l),who studied the effect of stirring upon the apparent permeability of a collodion membrane.2 R e have been led to investigate systematically the effects of pressure and of stirring upon the simplest case, that of a non-electrolyte such as sucrose. Other observations refer to the effect of molecular size and of the electrical effects with salt solutions or charged colloidal soh. K e have also taken occasion to measure the hydration of pectin. I. CONTROL OR ADJUSTMEKT O F THE POROSITY OB CELLULOSE MEMBRANES
Not only is modern sheet ccllulose (Cellophane, etc.) denser than fornierly, but, as was pointed out in previous communications, Cellophane swells less after aging, or a t higher temperature (3). Freshly manufactured Cellophane may swell as much as threefold in thickness (not in length, nor more than a few per cent in width), but after keeping for several years it may swell to less than double its thickness. The change in thickness goes parallel with change in porosity. Dry No. 300 Cellophane and Sylphrap are 0,0009 in. or 0.023 mm. thick; this was the material chiefly used in the present investigation. No. 600 Cellophane, previously used, filters more slowly. It also retains distinctly (at least three-halves) more solute.
Swelling in water With a new and an older sample of Cellophane No. 600, the thickness in inches upon swelling in water for fifteen minutes and for twenty-four hours a t three different temperatures was as shown in table 1. With No. 300 Cellophane swollen in water the dimensions were the same at fifteen minutes as after three hours, as shown in table 2. Thus the Cellophane swollen a t room temperature very slowly shrinks again at 100OC. The porosity and the rate of ultrafiltration also depend upon the tem2 We are in agreement with his main conclusions except for a few sentences in the first two paragraphs of his paper in which, since his membranes were of unsuitable porosity and were, of course, not suitable for distinguishing between molecules and particles, he stated the lzon sepuztur that Kistler’s coarse ones likeaise were not.
1159
ULTRAFILTRATION THROUGH CELLOPHANE
perature of swelling in water. Thus, a t 200 pounds pressure, when 3 per cent sucrose was filtered through 24 sq. cm. of filter: CELLOPHANE SWOLLEN AT
'C.
25
100
~
RATE OF ULTRAFILTRATION
1
j
[
BUCROBE RETAINEDa
grams per minute
pe7 cent
0.38 0.22
13 22.0
Swelling in sodium hydroxide solution The porosity of membranes swollen in sodium hydroxide is dependent upon both temperature and concentration. In general, an optimum concentration exists for each temperature, the optimum being more proTABLE 1 Swelliny of N o . 600 Cellophane in water OLD SAMPLE
N E W BAMPLE (DRY 0.0016 I N . )
TREATMENT OF CELLOPHANE
(DRY
0.0015 IN.)
15 min.
24 hrs.
15 min.
inches
inches
inches
inches
0.0035 0.0035 0.0030
0.0033 0.0032 0.0028
0.0034 0.0034 0,0029
24 hrs.
_ _ _ _ _ ~ _ _ _ Water a t 0°C.. . . . . . . . . . . . . . . . . . . . . . . . 0,0035 Water a t room temperature. . . . . . . . . . . . 0.0035 0.0030 Water a t 100°C.. . . . . . . . . . . . . . . . . . . . . . .
TABLE 2 Swelling of No. SO0 Cellophane in water DIMENIIONS
TREATMENT OF'CELLOPHANE
Original . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water a t room temperature (25OC.). . . . . . . . . . . . . . . . . . . . . Water a t 100°C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water a t lOO"C., then 2 days a t 25%. . , . , . , , , . , . , , . , . , . . Room temperature, then 3 hours a t 100°C... . . . . . . . . . . . . .
40 x 46.5 x 45 x 45 x 45 x
76 x 76 x 76 x 76 x 76 x
0.023 mm. 0.051 mm. 0.038 mm. 0.038 mm. 0.045 mm.
nounced and the swelling greater a t low than a t high temperatures. Membranes swollen a t low temperatures are more permeable than those swollen first a t high temperatures and then brought to lower temperatures, although lowering the temperature after the initial swelling does cause an increase. 3 For the method of calculation and definition of per cent sucrose retained, 1 - 4 , see part 11, below. All ultrafiltrations were carried out with the apparatus previously described, similar to that supplied by Yereinigung Gottinger Werke, Gottingen, with electrical stirrer, which was always used except when otherwise mentioned. The interior was silver-plated, and such parts as the stirrer and filter bed were replaced by pure silver. When electrolytes were present, the plated parts were coated with paraffin. Pressures were obtained from a cylinder of nitrogen.
1160
JAMES W. MCBAIN AND R. F. STUEWER
Conversely, swelling initially a t low temperatures and then heating causes a decrease, but not of sufficient magnitude to give the same result as is obtained with the reverse procedure. Membranes swollen in sodium hydroxide have a tendency to become very brittle, especially when the swelling is large, and so are not satisfactory over a wide range of porosity. Neale (7) has discussed the swelling of viscose cellulose sheets and presents a plausible theory as to the mechanism. The hysteresis effects just pointed out cannot, however, be explained on the basis of osmotic and Donnan effects. Swellzng in zznc chloride solution Zinc chloride is a remarkable swelling agent in that its effectiveness is almost confined to the range 60 to 65 per cent. Kahlbaum’s best grade of anhydrous zinc chloride was employed, and analysis confirmed its purity and freedom from water. Detail of procedure ,for swelling in zznc chloride. The membranes were swollen by first pouring enough of the solution into a flat dish to cover the bottom, then putting in the Cellophane, and finally pouring on the remainder of the solution. Sufficient solution was used to cover the membrane completely, and care was exercised to prevent its adhering to the bottom of the dish. This was accomplished by simultaneously tilting and rotating the dish for the first ten minutes. A coaLing of paraffin on the bottom of the dish is desirable, but not necessary. At the completion of swelling, Le., after fifteen to twenty minutes, the excess liquid was carefully pipetted off so as to leave the membrane lying flat. Wrinkling of the surface may lead to cracking of the membrane upon removal of the swelling agent. After removal of the excess liquid, water was poured in and the dish was rotated to speed up diffusion. After a few minutes the membrane wa9 removed and washed free from zinc chloride by means of dilute hydrochloric acid. The acid was then washed out with distilled water. Determination of the ash left upon burning the treated membrane indicated that satisfactory removal of zinc had been effected. Effect o j zznc chloride. As criteria of the extent and effectiveness of swelling we have used the thickness, the rate of filtration, and the per cent of sucrose retained when a 3 per cent sucroSe solution is filtered through the membrane. Thickness is significant except a t the higher concentrations of zinc chloride, where the surface of the membrane tends to slough off in layers and the softening likewise renders the result uncertain. If insufficient time is allowed for swelling, the sloughing may be uneven. Swelling for long periods instead of fifteen minutes increases sloughing. Sylphrap swells more than Cellophane, and its surface remains smooth. Yiscacelle swells less and sloughs more than Cellophane. The swelling
ULTRAFILTRATION THROUGH CELLOPHANE
1161
increases steadily with lowering of temperature from 40" down to lO"C., but 3°C. is too low for appreciable swelling in fifteen minutes. Table 3 shows the degree of reproducibility of swelling in particular concentrations of zinc chloride as judged by the rate of flow of water in grams per minute. TABLE 3 Eflect of swelling N o . 300 Cellophane i n zinc chloride solutions upon the rate offiltration of water at various pressures (pounds per square inch) CONCENTRAll ON OF ZnCln
RATE OF FILTRATION AT PRESSURES OF
40 lbs
80 lbs.
120 lbs.
200 Ibs.
0.375 0.397 1.408 1.403 2.850 1.782 5.97 4.24
0,558 0.580
0.467 0.450 1.011 0.698 2.125 1.56
0,267 0.285 1.037 1.009 1.968 1.298 4.31 3.13
per cent
60 60 63.0 63.0 63.8 63.8 64.0 64.0
TABLE 4 Increase of thickness on swelling CONCENTRATION OF ZnCh
per cent
Original dry 60 61 62 63 64 65
TEICHNESB OF
Cellophane
Sylphrap
inches
inches
0.0009 0.0030 0.0040 (after use 0.0035) 0.0046 0.0060* 0.0060* (after use 0.0050) >0.0060
0.0009 0.0035 0,0043 0.0054 0,0082** 0.0080**
Visoaoelle inches
0 0009 0.0030 0.0037 0.0045 (0.0055)
* Increase in all dimensions; surface slightly roughened. **Surface still smooth after 15 minutes. However, after 17 hours the surface is sloughed off. The thickness increases on swelling as shown in table 4. With the niembranes swollen in 63 per cent and 64 per cent solution, filtration of a 3 per cent sucrose until two-thirds or three-quarters had passed through left a residue not more concentrated than 3.01, 3.06, and 3.02 per cent, respectively. Likewise, 3 per cent potassium chloride and 3 per cent potassium iodate passed through unaltered, within the error of analysis. The Cellophane swelled distinctly, both in length and breadth. The rate of filtration for Sylphrap was about three-fifths, and for Viscacelle about two-
1162
J A M E S W. MCBAIN AND R. F. STUEWER
fifths that for Cellophane, but the pressure relations were similar to those in table 3. The pressure of 200 pounds appears to compact the menibranes, so that the rates are no longer proportional to the pressure, either with water or with solutions. The effect of pressure upon the retention of sucrose as shown in table 5 will be discussed in part 11. Figure 1 shows the pressure dependence of the rates for Cellophane taken from table 5. As is ordinarily the case with gels, the swollen membranes shrink upon drying and do not regain their original porosity. Experiments upon cellophane which had never been dried in the course of manufacture were run to determine to what extent the structure might be allowed to co1lap.e without this collapse beconling permanent. Accordingly, strips of the Cellophane were weighed, soaked in glycerol soh tions of various conccntrations, and then dried and reweighed. The glycerol was then washed TABLE 5 Effect o j swelling KO.500 Cellophane in z i n c chlorzde solution u p o n the rate of filtration ( R in grams per m i n u t e ) and per cent of sucrose retained Jrowz 9 per cent sucrose soli~tion CONCEITRATION OF
ZnClz
1 PHICKNESS
PER CENT O P SUCROSE RETAIAED A U D R A T E OF FILTRATION AT P R E S S r R 4 8 OF
40 lbs
I
per cent
0 61
120 Ibs
SO lbs
200 Ibs.
I
0 0009
64 65
out and the strips were weighed, dried, and neighed foi a fouith time. The percentages of free space occupied by the glycerol and the percentage of space regained upon reswelling were calculated as follows : Per cent of space kept filled while dry Per cent of space regained on swelling
0 0 9 0 20 3 30 3 41 4 51 5 36 9 53 3 84 6 94 4 96 0 98 2
It is apparent that essentially complete recovery of the original form results if 30 per cent or more of the free space is filled while dry. This affords a means of preserving membranes of this type, if it is necessary t o dry them. They may be kept in glycerol solution for some time TTithout change also. In water decomposition takes place after a few days. When membranes were soaked in 50 per cent glycerol and then dried, after seven aeeks the glycerol was washed out and the membranes w r e found t o have retained a t least nine-tenths of the previous rate of filtration and to have recovered from previous higher prewures.
ULTRAFILTRATION THROUGH CELLOPHANE
1163
The experiment also shows that the swelling process with zinc chloride probably does not consist merely in a breaking down of the old structure, but in a setting up of a new structure. That is, the gel particles have not been torn apart, owing to strains within the gel, but rather have been allowed to orientate in such a manner as to relieve the stress. Upon washing out the zinc chloride the ability to orientate is lost, and the new orientation becomes the preferred one. Thereafter the tendency is to return to this form after compression. The advantages of the method of preparing membranes may be briefly stated. Assuming that no new pores are opened, one may conclude that
40
80 120 160 200 PRESSURE-IBSRRSQ IN
FIG.1. Effect of zinc chloride on porosity of Cellophane. o , water; A, 61 per cent zinc chloride; 0 , 6 4 per cent zinc chloride; 0,65 per cent zinc chloride.
in a series of membranes the number of pores and the relative pore sizes may be readily calculated from areas before and after swelling, and from rates of flow of water. The convenience of the method makes it especially applicable when only a few membranes of varying pore sizes are desired. 11. T H E DEPENDENCE O F RETENTION O F ELECTROLYTES AND O F NONELECTROLYTES UPON MOLECULAR SIZE, CONCENTRATION, STIRRING, PRESSURE, AND OTHER FACTORS
RATE O F
The retention or sieving constant (1- 4 ) is d e h e d by Manegold (2) as the constant proportion of solute held back throughout a given filtration,
1164
JAMES TV. MCBAIN AND R. F. STUEWER
$being set equal to the ratio between the concentration of original solution and residue for a small amount of filtration. If x1 and 2 2 are the concentrations of successive samples of residue of solution remaining above the filter whose volumes are u1 and V Z , it is readily shown that xl/x, = (ul/u*)$, where r#I is identical with that deduced in Manegold’s more complicated expression. Retention will therefore be expressed as lOO(l-r#I) per cent. In all cases approximately 50 g. was placed in the filter, which had been previously rinsed with several portions of the same solution. After standing a small sample was removed for analysis immediately before filtration began. The residue after filtration was also analyzed. Eflect of size of molecule Table 6 is of interest in showing how ordinary Cellophane swollen in water discriminates between molecules of progressively larger size. At 350 pounds pressure the rate for the first three was 0.47 g. per minute per 24 ern.,; for raffinose a t 200 pounds, 0.35 g. TABLE 6 Percentage of different molecules held back b y ordinary Cellophane SOLUTE
I
INITIAL CONCENTRATION
1
FINAL CONCENTRATION
1
PER CENT RETAINED
per cent
Glycerol (C,) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dextrose (C,) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sucrose (&), . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raffinose (C18).............................
2.96 2.86 2.96 2.96
3.02
7.0 3.58
18.1
Eflect of concentration, and the contrast between electrolytes and non-electrolytes Table 7 shows how sucrose compares with a number of electrolytes, and also shows the enormous effect of dilution upon electrolytes first discovered by Ershler for collodion membranes of slightly greater porosity. All these solutions were filtered through the same actual piece of Cellophane a t 200 pounds per square inch, except for the viscous 48 per cent sucrose, where 340 pounds were used. The calculation of $ is necessarily inaccurate for the dilute potassium iodate, because of the change in $ with concentration. Figure 2 contrasts the relative constancy of retention of sucrose with the great effect of dilution upon an electrolyte such as potassium iodate. Two hundred pounds pressure and the same rate of stirring were used throughout. The most dilute solution of sucrose was 0.05 per cent, where the accuracy of analysis is impaired. The high retention of very dilute electrolytes is probably due to the charges upon the walls of the pores of the membrane, in accordance with
1165
ULTRAFILTRATION THROUGH CELLOPHANE
the principle of Donnan equilibrium: the effect being suppressed by sufficient concentration of any electrolyte, as is shown a t the end of table 7. Per cent retained, iOO(1
c
TABLE 7 +),for a series of solutions filtered through the same piece of Cellophane
DATB
RATHI
SOLUTION
December 7, 1935 December 7, 1935 December 14, 1935 December 14, 1935 December 24, 1935 December 24, 1935 December 24, 1935 December 26, 1935 December 27, 1935 January 7, 1936 January 18, 1936 January 22, 1936 January 23, 1936
0.39 0.38 0.33 0.37 0.35 0.35 0.29 0.31
2.94% KCl 3.01% KIOa 3.0170 sucrose 6.97% KIOa 0.083% KIOa 0.92% CdIz 24.3 7 0 CdIz 5.30% NaIOa 48.23% sucrose 2.99% HI08 3.01’% sucrose 2.94% KCl 2.97% KIOB
0.043
0.27 0.27 0.27 0.29
+ 0.74oJ, NaCl + 10% NaCl
Another specimen.. .. ......... Another specimen.. .. ......... Another specimen.. . . . . . . . . . . . . .
0.087% KIOa 3.0% N a I 0 8 17.7% KCl
I
PHIR CENT RETAINED
1.92 1.78 14.50 3.66 48. 3.45 7.55 4.68 8.59 5.54 14.16 3.57 4.5
1
7.1 6.1 0.9
d
1
2
3
4
5
6
1
RRCCNT COtKB4lPATION
FIQ.2 FIG. 3 FIG.2. Dependence of retention on solute concentration. 0 , sucrose; 0 , potassium iodate. FIG.3. Dependence of retention on porosity; retention of sucrose and of potassium iodate of two concentrations by membranes of different porosities. 0 , 3 per cent sucrose; 0,0.08 per cent potassium iodate; X, 3 per cent potassium iodate.
The effect of the charge in keeping dilute electrolytes away and thus simulating a reduced size of pore may be compared with the similar effect in 4 Other influences of Donnan distribution in ultrafiltration were pointed out and discussed by J. W. McBain and W. McCletchie (J. Am. Chem. Soc. 66, 1315 (1933)). For its suppression see reference 3.
1166
JAMES W. MCBAIN AND R . F. STUEWER
dyeing. If so, the rBle of added salt in dyeing is not that of a dispersing agent, as has often been suggested, but rather that the effective pore size is restored. Effect of varying porosity o n retention Figure 3 exhibits for sucrose and potassium iodate the effect of steadily increasing porosity, as measured by rate of flow a t 200 pounds and 140 R.P.M., in permitting more and more complete passage of the solute. I t is evident that long after sucrose is passing completely, there is still a marked effect upon the very dilute potassium iodate. For higher ionicconcentration the potassium iodate curve lies wholly below that of sucrose. In' other words, the exaggerated effect in the dilute electrolyte is a Donnan effect. Gse of membranes for detecting presence of colloid The data in this paper clearly show that membranes must be suitably chosen and employed if they are to detect and measure the presence of colloid with certainty. It must be shown that the particular membrane is sufficiently porous to pass quantitatively all ordinary molecules, including electrolytes in which the ionic concentration is sufficiently high to suppress Donnan effects. In a previous communication (4) it was concluded that nearly saturated solutions of sodium and potassium iodates contained a distinct proportion of colloid. However, reexamination of the original records shows that the experiments assembled in that communication were spread over a series of years, and that although the original Cellophane was indeed tested for its complete permeability to electrolytes and non-electrolytes, the much later experiments on the iodates had been carried out with a later and evidently much denser sample of Cellophane. There therefore remains no definite evidence for the presence of appreciable colloid in these solutions, although it may be significant that the retention of cadmium iodide in concentrated solution is twice that in more dilute solution. Effect of pressure and rate o j stirring Ershler has convincingly pointed out that complete absence of stirring must soon suppress all minor differences between the composition of filtrate and original solution, because of the automatic building up of the compensatingly high concentration in the layer of liquid resting upon the upper surface of the filter. We extend this to point out in the data here deduced that it is always a race between the rate at which the solution is being bodily thrust through and the rate a t which any molecule or particle which cannot freely pass through all portions of the membrane is escaping from its neighborhood,
ULTRAFILTR ATlON THROUGH CELLOPHANE
1167
through electrical influences (Donnan effect), through diffusion, and through convection. Hence, the higher the pressure and the higher the viscosity, and the higher the molecular weight, the greater must be the rate of stirring in order to allow the membrane to exhibit its selective action. Even for a non-electrolyte such as sucrose, these effects are intriguing, as is shown in figure 4. It is quite clear that unless the pores are large enough to pass all similar molecules freely (1 - 4 = 0), great care must be taken in interpreting quantitatively the retentions observed under the actual conditions of the filtration. In figure 4 it is evident that maximum retention occurs when the pressure is 200 pounds per square inch (rate of filtration is 0.44g. per minute) at the fastest practicable rate of stirring, a rate faster than that for which
UUW\L
SSURE-309.
Y
I
103
I
203
I
RPM
303
I
403
FIQ.4. Dependence of retention on pressure and rate of stirring; percentage of sucrose retained in filtration through ordinary Cellophane a t various pressures and rates of stirring. Full lines refer t o 3 per cent sucrose; the dashed line to 30 per cent sucrose a t 200 pounds pressure. the instrument is commercially supplied. The position of the curve for 30 per cent sucrose is predictable from the ratio of viscosities or of rates of filtration when compared with 3 per cent sucrose, except at the lowest rate of stirring. Some of the effects are similar to those discussed in the theory of dependence of heterogeneous reaction upon rate of stirring. SUMMARY
1. Commercial sheet cellulose is much less porous than formerly. It is now necessary to swell in zinc chloride solutions exceeding 63 per cent zinc chloride (37 per cent water), in order to make these membranes freely permeable to all ordinary molecules. Further, the ionic concentration must be sufficient to suppress Donnan effects in the case of very dilute electro-
1168
JAMES W. MCBAIN AND R. F. STUEWER
lytes, including colloids. Contrary to a previous intimation, owing to this change in a commercial membrane there remains no certain evidence for colloid in iodate solutions, although some is indicated in very concentrated aqueous cadmium iodide. 2. With incompletely permeable membranes such as ordinary cellulose, interesting influences of size of molecule, concentration, rate of stirring, pressure, viscosity, and diffusion are observed. REFERENCES (1) ERSHLER, B.: Kolloid-Z. 68,289 (1934). (2) MANEGOLD, E., AND HOFMANN, R.: Kolloid-Z. 61,221 (1930). (3) MCBAIN,KAWAKAMI, AND LUCAS:J. Am. Chem. SOC.66,2768 (1933). (4) MCBAIN,J. W., AND KISTLER,S. S.:J. Phys. Chem. 36,130 (1931). ( 5 ) MCBAINAND KISTLER:J. Gen. Physiol. 12, 187-200 (1928); J. Phys. Chem. 33, 1806-12 (1929); 36, 1306 (1931); Trans. Faraday SOC.No. 107, 26, Part 4 (1930). MCBAIN,KAWAKAMI, AND LUCAS:J. Am. Chem. SOC. 66,2762 (1933). (6) MORTON, T. H.: Trans. Faraday SOC.31,262 (1935). (7) NEALE,S.M.: J. Textile Inst. 20,373T (1928).