Effect of Membrane Pore Size on the Measurement of WaterTension in Bentonite Suspensions1

AMOS BANIN, BRIAN G. DAVEY, AND PHILIP F. Low2

ABSTRACT

A pressure-transducer tensiometer and a dynamic osmometerwere used to measure water tensions and osmotic pressures, re-spectively, in clay-water systems. The two methods of measure-ment produced the same results. Different membranes, havingpore sizes ranging from 0.004^ to O.S/* were used with thetensiometer. It was found that the measured water tensionof a given system depended on the pore size of the membrane;as the pore size increased, the tension decreased. However,experience indicates that little or no turbidity appears in asolution when it is separated from a clay suspension by amembrane having pores as small as those of the present experi-ments. Therefore, the relationship between pore size and ten-sion was not due to differences in membrane permeability to

the clay particles. Nor was it due to differences in membranepermeability to dissolved solutes. Tests showed that no lastingtension developed when a NaCI solution or an extract of theclay were utilized in place of the clay suspensions. Two pos-sible explanations were proposed, namely: ( i ) entrance ofparticles into the pores was all that was required to lower thereflection coefficient, and (ii) the gelation of the clay suspen-sions within the pores produced gels having particle densitiesand arrangements and, hence, permeabilities to dissolved sol-utes, that were dependent on the dimensions of these pores. Itwas pointed out that, regardless of the explanation, the exist-ence of such a relationship between membrane pore size andmeasured water tension is noteworthy.

Additional Key Words for Indexing: tensiometers, osmom-eters.

WHEN a clay-water system of any consistency, here-after referred to as a suspension, is separated from

pure water by a porous membrane, the water flows throughthe membrane into the suspension. To prevent flow, pres-sure can be increased on the suspension or decreased onthe pure water. The required increase of pressure is calledthe osmotic pressure; the required decrease is called the

BANIN ET AL.: MEASUREMENT OF WATER TENSION IN BENTONITE 307

1500

1000.

500zO

Z OLJO,COZ>-500CO

LJO

200XO

100-

OLJ

2.I2Q cloy/IOOg water285 hours

0.50 0.75

2.l2g Clay/IOOg water18 hours

8.37g clay/IOOgwater

285 hours

1.00 2.00\ 3.00

\ .

\8.37g clay/IOOg water17.5 hours

1.00 2.00 °3.00°\3.0

CARBOWAX 6000 CONCENTRATION(g/100 ml solution)

Fig. 1—Illustrations of the dynamic method for determiningthe concentration of Carbowax 6000 solution in equilibriumwith a Na-clay suspension.

400

O

300£oLJrroCO23200rrQ-

ooCO 100O

Data from: • Present Studyo Alexandrowicz (1959)

O 1 2 3 4 5 6CARBOWAX 6000 CONCENTRATION

(g/ 100ml solution)Fig. 2—Osmotic pressure of a Carbowax 6000 solution as a

function of concentration.

water tension. Both quantities have the same numericalvalue if the partial molar volume of the water is the samein the clay suspension as it is in pure water.

Recently, osmotic pressures of clay suspensions weremeasured with a dynamic osmometer in which the mem-brane was a section of Visking dialysis tubing. The dynamicosmometer was similar to that described by Alexandrowicz(1959). When the results were compared with those ofLow (1955), who used another kind of dynamic osmometerbut the same membrane, agreement was found. However,agreement was not found when the results were comparedwith those of Leonard and Low (1963). The latter inves-tigators used a tensiometer involving a sintered glass filteras membrane. Hence, it appeared that the measured osmoticpressures of clay suspensions may depend on the kind ofmeasurement, i.e., dynamic or static, or on the kind ofmembrane employed. To further investigate these possibili-ties, the following study was conducted.

EXPERIMENTAL METHODSPreparation of the Clay—Crude Wyoming (Volclay no. 200)

bentonite was mixed with deionized water to form a suspensionof 1.8% and the fraction < 2/j, was separated by decantation.To prepare the Na-Clay, the suspension of < 2/j. particles waspassed through a column of Dowex 50 cation resin saturatedwith Na+. The resulting Na-clay was sedimented by centrifuga-tion in a Sharpies super-centrifuge and redispersed in deionizedwater. This process was repeated until the electrical conduc-tivity of the supernatant solution was < 50 /^mho/cm at 25C.Then the clay sediment was freeze-dried and ground in a ballmill for 15 min. Suspensions at the desired clay concentrationswere prepared from this stock of dry Na-clay.

Tension Measurements—A tensiometer was constructed inwhich the tension developed in pure water, separated from aclay suspension by a thin membrane, was measured by a sensi-tive differential pressure transducer (Dynisco model DPT-85

range ^2.5 lb/in2) and recorded continuously by a L. & N.recorder. Three different membranes were used in the tensio-meter: a Visking dialysis membrane, a Millipore (type-VF)membrane and a Metricel (type GA-4) membrane. The porediameters of these membranes were about .004/i, .01/i, and .8^,respectively. The tensions in several Na-clay suspensions ofdifferent clay concentrations were measured with each mem-brane.

Osmotic Pressure Measurements—Osmotic pressures of claysuspensions were measured by concentration osmometers thatwere similar in principle to those used by Alexandrowicz(1959). These osmometers were prepared by mounting circlesof Visking dialysis membrane on the ends of short glass tubes(inside diam. = 1.70 cm), filling these tubes with 5-ml aliquotsof a given clay suspension, and hanging them vertically in100-ml test tubes containing Carbowax 6000 solutions of dif-ferent concentration. The heights of the tubes containing theclay suspension were adjusted so that the level of the suspen-sion was the same as that of the solution. Periodically thesetubes were removed, blotted and weighed. Then the change inweight of the different tubes was plotted against the respectiveCarbowax concentrations as illustrated in Fig. 1. The inter-section of the resulting line with the abscissa gave the concen-tration of the Carbowax solution with which the suspensionwould be in equilibrium. The corresponding osmotic pressurewas determined from the calibration curve shown in Fig. 2.The lower part of this curve (up to 3 g Carbowax/100 mlsolution) was obtained by using the tensiometer described inthe previous section. For this purpose the Visking dialysis mem-brane was employed. The upper part of the curve was obtainedfrom the data of Alexandrowicz (1959).

RESULTS

The change of the water tension with time, measured bythe pressure-transducer tensiometer, is shown in Fig. 3 forseveral Na-clay systems. The change of the osmotic pres-sure with time, measured by the concentration osmometer,is shown for similar clay systems in Fig. 4.

In Fig. 5 the maximum values of water tension and

308 SOIL SCI. SOC. AMER. PROC., VOL. 32, 1968

= 150

OCLAY CONCENTRATIOfKg/IOOgof water)

I 4.55gH 7.l7g

IH9.73g33tl3.55g

40 6O 80 IOO 120TIME (hours)

140 160

Fig. 3—Change of water tensions with time in Na-clay suspen-sions (measured by pressure transducer tensiometer withMetricel GA-4 membrane).

osmotic pressure, reported as water tension, are plottedversus the clay concentration for the three membranes. Itis evident that the membrane type, and not the measuringdevice, influenced the "equilibrium" tension, and that thelarger the pore size, the smaller the tension at a given clayconcentration. These results are not unlike those of Kemperand Evans (1963) who found, using Carbowax as solute,that the contribution of the osmotic pressure to flow througha membrane decreased as the size of the pores increased.

DISCUSSIONIt is known that, for the accurate measurement of osmotic

pressure, a truly semipermeable membrane is necessary.Staverman (1951) has shown that, if the membrane is"leaky," i.e., if it does not completely exclude the solute,the measured osmotic pressure, nm, will be lower than thetheoretical osmotic pressure, n(, and will be given by:

nm = [1]

where <j is the reflection coefficient. If the membrane ex-cludes the solute entirely, a = 1. If it does not exclude itat all, tr = 0. And if it partially excludes the solute, a willhave a value between O and 1. According to Durbin (1960)and Kemper and Evans (1963), the exact value of adepends on the size of the solute relative to the size ofthe pores.

Seemingly, the foregoing ideas can be invoked to explainthe differences in osmotic pressure observed with the dif-ferent membranes. The clay particles may be regarded asosmotically active solutes because they are surrounded byionic atmospheres and their surfaces attract water. [Asshown by Leonard and Low (1963), the particles per se,acting ideally as individual kinetic units, would make aninsignificant contribution to the osmotic pressure.] Andsome of them are smaller than the pores in the membranesemployed so that, theoretically, o- would be expected tohave a value that is less than unity and that is dependent onthe size of the pores. But a difficulty is encountered. Expe-rience indicates that when a bentonite suspension in the solstate is separated from a solution by a Millipore or similar

120

n5 IOO

Eo~ 80

in 60

40

20

CLAY CONCENTRATION (g/IOOg of woter)

I 1.13H 6.45TU 8.3 7

IOO 200TIME(hours)

300 400

Fig. 4—Change of osmotic pressures with time in Na-clay sus-pensions (measured by concentration osmometer with Viskingdialysis membrane).

membrane and stirred, little or no turbidity appears in thesolution. This indicates that very few particles pass entirelythrough the membrane. Further, most of the suspensionsthat we used (those containing > 4% clay) were in the gelstate in which there is no Brownian motion of the particlesand, hence, no transport mechanism. Consequently, ifequation [1] is applicable and the clay particles were theonly solutes, a- must depend on pore size in a way that doesnot require passage of the particles through the membrane.

It is possible that the clay particles were not the onlysolutes and that, in particular, they were not the ones that"leaked" through the membrane. As one reviewer sug-gested, the different membranes may have caused differentmeasured tensions because they were differentially perme-able to free electrolytes that were present in the suspensions.Presumably, these electrolytes resulted from incompletewashing of the clay or from dissolution of the clay itself.Although our past experience indicated otherwise, we werenot sufficiently confident to discount this possibility entirely.Therefore, it was tested by determining the water tensionsof two NaCl solutions and of an extract from the clay (Thehelp of Mr. Bev Kay in making the necessary tests is grate-fully acknowledged.). The NaCl solutions had normalitiesof 0.002 and 0.02, respectively. The extract was obtained bycentrifuging a Na-bentonite suspension (3 g clay/100 mlwater) which had been allowed to stand for 6 weeks. Forthe tension measurements the pressure-transducer tensiome-ter was used with the Visking membrane. This is the mem-brane that would be expected to show the greatest saltexclusion. In each case the water tension increased at firstand then decreased to zero. The maximum tensions achievedwere 26, 14, and 7 cm of water for the 0.027V NaCl solu-tion, the 0.002/V NaCl solution, and the clay extract, re-spectively. The respective times at which the tensionsbecame zero were 35, 26, and 17 hours. When these timesare compared with those allowed for equilibration (Fig. 3and 4) it is evident that differences in the measured tensions(osmotic pressures) cannot be ascribed to differences inthe permeabilities of the membranes per se to the dissolvedelectrolytes. However, it is possible that clay particlesentered the pores of the membranes and gelled therein to

BANIN ET AL.: MEASUREMENT OF WATER TENSION IN BENTONITE 309

4 6 8 10 12 14CLAY CONCENTRATION (g/IOOg water)

Fig. 5—The effect of membrane pore size on the measuredwater tensions in Na-clay suspensions at different clay con-centrations.

produce, in effect, smaller pores with a greater surfacecharge density. These pores would have a greater ability toexclude electrolytes. In other words, it is possible that claygels within the pores acted as membranes that were moreselective than the original ones. Further, it is possible thatthe density and arrangement of particles within these gelsand, hence, their selectivities, were governed by the dimen-sions of the pores in the supporting membranes. Thus theobserved differences in water tension could have occurred.

Following the reasoning described above, one is obligedto reconsider the studies of Leonard and Low (1963), ofKolaian and Low (1960), and of Ripple and Day (1965).The first-named investigators, using a membrane with largepores, i.e., a fritted-glass membrane, found that no watertension could be measured in suspensions of Na-Aberdeenand Na-Cheto bentonite until the clay concentration washigh enough for the suspension to gel. The latter investi-gators, using similar membranes, found that the measuredwater tension in a clay gel was reduced significantly whenthe gel was disturbed by stirring or shearing. The possibilityexists that, in these studies, the value of a depended on thestate of the system. When the clay concentration was toolow for gelation to occur, or when a gel was disturbed,some of the particles entered the pores without forming aselectively-permeable membrane. The result was a relativelylow value of <r- However, when the clay concentration washigh enough for gelation to occur and there was no disturb-ance, the particles, being implicated in the gel structure,did not enter the pores. Or, if they did, they gelled thereinand formed a selectively-permeable membrane. The result

was a relatively high value of a. In this regard note fromFig. 5 that, when the membrane with large pores (i.e., theMetricel membrane) was used, the water tension was zerountil a clay concentration of 4-5 g/100 g of water wasreached; thereafter, it had finite values. The Na-bentonitegelled at this concentration.

Now, by calling attention to the possible effect of thestate of the system on CT, we do not mean to imply that thestate of the system has no effect on nt, the theoretical ortrue water tension. In fact, evidence cited by Leonard andLow (1963) indicates that this and other properties of thewater change as the suspension gels. And additional evi-dence, obtained recently in this laboratory, indicates thesame. But we do want to recognize that there may be morethan one factor contributing to changes in the measuredwater tension with gelation.

In view of the uncertainty with respect to the basic causeof the differences in measured water tension with differentmembranes, it would be unwise to try to account for thefact that the water tension tended to increase with timewhen it was measured by the pressure-transducer tensiome-ter (Fig. 3) but tended to decrease with time when it wasmeasured by the dynamic osmometer (Fig. 4). However,it should be noted that the increase with time* could havebeen due to either a change in the state of the suspension(Kolaian and Low, 1960; Ripple and Day, 1965) or to theslowness of water transport during equilibration (Klute andGardner, 1962). The decrease with time could have beendue to a slight permeation of the membrane by the Carbo-wax. The results of Kemper and Evans (1963) suggestthat this does occur.

Regardless of the explanation for the effect of pore sizein a membrane on the measured water tension, the factthat such an effect exists is noteworthy. Tensiometers andosmometers are used frequently with different membranesin soil science. One now has reason to wonder about theaccuracy of the results.