SOLID SURFACES and the Gas-Solid Interface


SOLID SURFACES and the Gas-Solid Interfacepubs.acs.org/doi/pdf/10.1021/ba-1961-0033.ch035Similarby CT DEEDS - ‎Cited b...

0 downloads 114 Views 992KB Size

Density Studies in Clay-Liquid Systems I.

The Density of Water Adsorbed by Expanding Clays

C. T. DEEDS and H. VAN OLPHEN

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

Shell Development Co., Houston 1, Tex.

Apparent pycnometric densities of expanding clays in water and in various organic liquids are compared with crystallographic densities. Dif­ ferences between the various density values are obtained. These differences are interpreted in terms of both abnormal fluid densities in the neighborhood of the clay surfaces and void spaces within particle aggregates which are inaccessible to certain liquids. We concluded that aggregate void space occurs when the clays are immersed in fluids which do not penetrate be­ tween the unit layers; the density of the first two monolayers of water or of polar organic fluids which are adsorbed between the unit layers is slightly below normal in most expanding clays; and the density of water adsorbed beyond the first two monolayers of water is normal within the experimental error.

The properties of a liquid are likely to be modified in the neighborhood of a solid surface, partly because of the molecular geometry of the boundary, and partly because of the effect of adsorption forces on the molecules of the liquid. How far from the surface the adsorption forces significantly affect the properties of the liquid is still a matter of controversy. A case in point is the density of water adsorbed by clays. Although water vapor adsorption isotherms indicate that the adsorption energy of successive adsorbed monolayers' decays rapidly over a dis­ tance of about 10 Α., corresponding with the thickness of four monolayers of water, density anomalies in the water phase at distances of at least 60 A. are suggested by the results of the very carefully conducted experiments by Anderson and Low ( 1 ). These authors compare the volume increments of water squeezed out of a clay paste with the corresponding volume increments of mercury pressed into the paste. In a typical experiment, they find that the water density gradually decreases from the normal value to a value which is about 3% below normal at water contents of the pastes ranging from 5 to 1 gram per gram of clay. This 332

Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

DEEDS AND VAN OLPHEN

Density of Water Adsorbed by Clays

333

range corresponds with a minimum distance range of about 60 to 10 A. from the clay surface. Pycnometric data reported in the literature give undifferentiated values for the average density of the water phase in clay-water systems (2, 3, 4, 6, 7, 8). Deviations of the water density from normality are generally reported, but in spite of the simplicity of the experiments, there is little agreement about the magnitude or even the sign of the deviations. The reason for the scatter of the results is that in some of the work, certain precautions were not taken which are important when working with clay systems. We have repeated some of the pycnometer work, and have extended our investigations to systems with a known amount of preadsorbed water and to different displacement fluids.

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

Experimental Procedure Vacuum-jacketed pycnometers with capillary stoppers and ground-glass caps were used. They were calibrated with water. All weighings were reduced to vacuum. The measurements were carried out at 2 6 . 0 0 ° ± 0.01 °C. The organic fluids were Baker Analyzed or Spectro grade samples. They were thoroughly dried with Linde 4A Molecular Sieve. Water was triple-distilled and was used freshly boiled. The clays were occasionally used in their native form, but they were usually converted to a certain ion form by resin exchange and washed and centrifuged to remove coarse nonclay matter or to separate a particle size fraction. The clays were used in powder form, as flakes, or as a freeze-dried voluminous powder. They were dried to a constant weight at 1 0 5 ° C . at 0.1-mm. pressure and transferred to the pycnometers in a dry box. The dry condition of the clay was checked by measurement of the basal spacing by x-ray diffraction. The expanding clays admit water and certain organic compounds with polar groups between the unit layers of 10-A. thickness; therefore, a large contact area between the liquid and a crystallographically flat surface is created. Since the contact area is of the order of 800 sq. meters per gram, relatively large bulk effects can be expected if density anomalies near the surface occur. Unfortunately, however, the workable clay concentration is limited to a few per cent, since gelation prohibits adequate vacuum deaeration at higher concentrations. In hydrocarbon liquids, in which the clays do not show interlayer swelling, the contact area is considerably smaller, but larger clay concentrations can be tolerated. Results and Discussion The results of the pycnometric determinations are presented in terms of "apparent densities" of the clays in the various displacement fluids. The apparent density of a clay is that density calculated from the data by the assumption that the density of the displacement fluid is normal throughout the liquid phase. When dissimilarities are observed for the apparent densities using different displacement fluids and the apparent densities deviate from the actual density, the possibility of density abnormalities of the liquid in the proximity of the clay surface should be considered. Moreover, consideration must also be given to another possible cause of certain deviations—i.e., the existence of a void space within the clay system which is inaccessible to the liquid. Clays in Hydrocarbon Liquids ("Nonpenetrating Liquids"). When the dry clays are immersed in dry hydrocarbon liquids, the basal spacing of the clay is Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

334

ADVANCES I N CHEMISTRY SERIES

unchanged; therefore, these liquids do not penetrate between the unit layers of the clay. The apparent density of both native and sodium Wyoming bentonite in a number of nonpenetrating liquids is 2.694 grams* per cc. (11 determinations, standard deviation 0.006 gram per cc.). The actual density of the clays in these systems can be evaluated from the unit cell dimensions determined by x-ray diffraction, and the unit cell weight can be computed from chemical analysis data with the application of the RossHendricks method of calculation. The crystallographic density of the sodium Wyoming bentonite was 2.804 grams per c c , and this value is estimated to be accurate within ± 0.012 gram per cc. Therefore, the apparent density of the clay in hydrocarbon liquids is appreciably smaller than the actual crystallographic density of the clay. Since both the contact area and the adsorption forces between clay and hydrocarbon liquids are relatively small, we propose that the discrepancy between apparent and crystallographic density is entirely due to the presence cf void space within the clay phase which is inaccessible for hydrocarbons. The magnitude of this assumed void space is found from the difference between the apparent and the crystallographic specific volumes. Taking the specific data for the sodium clay, we find that the inaccessible void space amounts to 0.3720 - 0.3566 = 0.0154 cc./g. ( ± 0 . 0 0 1 5 cc./g.), or 4.1% of the specific volume. We further propose that this void space occurs within the clay aggregates between partially overlapping unit layers, as sketched in Figure l,a (left). Such a void space is accessible only via the channels between touching unit layers, and since hydrocarbon liquids are unable to penetrate between the unit layers, they would indeed be unable to pass through these channels. Clays in Water (Penetrating Liquid). The apparent densities of the expanding clays in water are higher than those of the same clays in hydrocarbon liquids. For example, the apparent density of sodium Wyoming bentonite in water is 2.793 grams per cc. (four determinations, standard deviation 0.008 gram per cc.). We have also determined the apparent density of a sodium Wyoming bentonite clay with a small amount of preadsorbed water (approximately 300 mg. per gram of clay), using n-decane as the pycnometer fluid. Assuming normal densities for both n-decane and the preadsorbed water, we found that the apparent density of the clay in this experiment is identical with the apparent density of the clay when completely immersed in water (2.786 grams per c c ) . This particular experiment will be discussed in more detail later. Within the experimental error, these values are identical with the crystallographic density of the dry clay. It is tempting to interpret these results simply as follows: The void space within aggregates which is inaccessible to hydrocarbons is accessible to water, and the density of the water phase is normal throughout the system within the experimental error. However, in order to show precisely which assumptions are made in arriving at any conclusions regarding the density of water in the system, we must give the results a somewhat more sophisticated analysis. When comparing the density data obtained in hydrocarbons with those obtained in water, we can formally attribute the difference between the corresponding specific volumes to an "apparent" void space inaccessible to hydrocarbons but accessible to water and, at the same time, define that this space is completely filled if the water is assigned a normal density. This apparent void Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

DEEDS AND VAN OLPHEN (a) Distribution fluid. is

of

Apparent a

behavior

of

unit

layer

hydrocarbon

Aggregate

empty

Density of Water Adsorbed by Clays

void

("inaccessible

aggregates.

Apparent

space voids").

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

Dry

clay

Apparent

immersed

hydrocarbon

behavjor

of

water

in a

in

distribution

of

normal

is

f i l l e d ("selectively

si b l e

(b)

density.

void

interlayer

of

water

Void

space acces-

space").

space.

Hydrated

clay,

fluid.

BENTONITE

VACANT

VACANT

INTERLAYER

SPACE

INTERLAYER

SPACE

unit

Figure 1.

335

layer

Schematic representation of apparent hydration of expanding lattice clays

space, which will be called the "selectively accessible void space," appears to be identical with the previously defined inaccessible void space. Therefore, we shall assume that both relate to the same real void space within aggregates. It is indeed likely that this void space which is created between partially overlapping unit layers would be accessible via the interlayer channels for penetrating liquids such as water (Figure l,a, right). Furthermore, we assume that there is no other aggregate void space which is inaccessible for water. This assumption is reasonable, since such a void space must be accessible for hydrocarbon fluids. When these assumptions are accepted, we see that the identity of the apparent density of the clay in water and the crystallographic density of the dry clay means that the system behaves as if the dry clay expands in water by admitting a sheet Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

336

ADVANCES IN CHEMISTRY SERIES

of water of normal density between the unit layers. However, this observation does not imply that the actual density of the water adsorbed between the unit layers is normal, for the following reasons. Contrary to the situation in the hydrocarbon systems, the crystallographic density of the dry clay is not necessarily the true reference value for the hydrous systems in which the clay occurs in the expanded condition. The c spacing of the dry clay as it exists in the hydrocarbon system is 9.6 A. Since the thickness of the unit layers proper in van der Waals contact is smaller than 9.6 Α., the unit layers in the bentonite clay are not in contact (see Figure l,b). If the c spacing of the uncharged prototype mineral pyrophyllite (9.14 A.) is arbitrarily taken as the van der Waals thickness of the unit layers in the clay, the unit layers in the clay will be 0.46 A. apart. The vacant space between the unit layers in the dry clay is occupied only by the sodium exchange ions. Since these ions are accounted for in the computation of the crystallographic density of the dry clay as it exists in the hydrocarbon systems, a true reference value could be established for these systems. In the expanded hydrous system, on the other hand, water molecules may occupy this originally vacant space in the domain of the oxygen surfaces of the unit layers, which also includes holes within hexagonal rings of oxygen atoms into which the water molecules may sink to some extent. At the same time, the sodium ions which were originally embedded in these holes may diffuse into the water phase. Without knowing exactly how the layers of adsorbed water are embedded in the unit layer surfaces, or where the cations are located in the hydrated clay, we cannot establish an exact crystallographic reference value for the hydrous systems. For these reasons, we stated that the bulk density measurements only indi­ cate that the clay behaves as if a sheet of water of normal density is admitted be­ tween the unit layers (see Figure l,fo). In reality, however, the interlayer water may expand in such a fashion that the originally vacant space becomes occupied. Since the bulk density measurements would not be affected by the expansion, no information on this matter can be obtained from such measurements. However, if the expansion occurs, we can make a rough estimate of the average density of the interlayer water, in so far as we can speak about the bulk property of density on the molecular scale of the interlayer space. In the experiment in which the apparent density of the clay with 300 mg. of preadsorbed water per gram of clay was determined in n-decane, the c spacing of the clay was 15.4 A. If the water between the unit layers should expand to touch the unit layers, its volume would be increased roughly in the ratio of (15.4 — 9.6) to (15.4 —9.14), and the "density" of the interlayer water would be 7.3% below normal. Possibly, a better value for the van der Waals thickness of the unit layers in the clay is 9.3 Α., which we derived by adding twice the radius of an oxygen atom (2 X 1.35 A.) to the 0 — 0 center distance in the unit layers (6.6 Α . ) . Taking this value and applying the same reasoning, we find that the "density" of the interlayer water would be 5% lower than normal if it expands to touch the unit layers. In these estimates, the effects of a partial sinking of the water molecules into the hexagonal holes, as well as of a diffusion of the cations into the water phase, are neglected, but these corrections would be comparatively small, and they would, moreover, partially cancel. The above interpretation is probably valid for most expanding montmorillonite clays, of which only sodium hectorite was studied in addition to sodium Wyoming bentonite. An exceptional behavior is shown by sodium vermiculite clay, an expanding clay differing essentially from the other expanding clays in the charge density, which is about twice as high. Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

DEEDS AND VAN OLPHEN

Density of Water Adsorbed by Clays

337

Unlike the situation in the bentonite and hectorite clays, the apparent density of the sodium vermiculite clay in water is appreciably higher than the crystallo­ graphic density of the clay. The apparent density in water is 3.045 grams per cc. (three determinations, standard deviation 0.012 gram per c c ) . The crystallo­ graphic density is 2.827 grams per cc. for the dry clay which has a c spacing of 9.82 A. Only if a c spacing value of 9.12 A. is used in computing the crystallo­ graphic density is there an equality of the crystallographic density and the appar­ ent density of the clay in water; therefore, the system behaves as if water of normal density touches the oxygen surface of the unit layers (see Figure l,fo). However, the uncharged prototype mineral for this clay is talc, which has a basal spacing of 9.26 A. If we take this spacing as the van der Waals thickness of the unit layers in the vermiculite clay, we find that the interlayer water would actually be somewhat compressed. Since the c spacing of the hydrated clay is 14.82 Α., the interlayer water would be compressed in the ratio (14.82 — 9.12) to (14.82 — 9.26), and the "density" of the interlayer water would be 2.5% above normal, when the corrections for sinking in the holes and redistribution of sodium ions are neglected. The difference between the density behavior of vermiculite and other ex­ panding clays is paralleled by the difference in swelling behavior. In bentonite, the two-layer hydrate has a c spacing of 15.4 Α., which is 5.8 A. or the thickness of two monolayers of water more than the c spacing of the dry clay. The sub­ traction of the same thickness from the 14.82-A. spacing of the hydrated vermicu­ lite gives a reference spacing of 9.02 Α., which is in reasonable agreement with the conclusions drawn from the density determinations. The compression of the interlayer water between the unit layers of the vermiculite clay is probably a consequence of the large attractive force between the negatively charged unit layers and a densely populated layer of sodium ions midway between the unit layers. This large attractive force also keeps the unit layers from swelling beyond a two-layer complex. In conclusion, definite information on the density of adsorbed interlayer water cannot be derived from the bulk density measurements. However, such measure­ ments supplement adsorption isotherm data, and both will be helpful in the inter­ pretation of x-ray structure analysis of the configuration of water molecules and exchange cations in the interlayer space. DRYNESS OF C L A Y . The question of the complete dryness of the clay after heating at 1 0 5 ° C . in vacuum remains an uncertain element. When the clays are fired at about 9 0 0 ° C , dehydroxylation takes place. The weight loss due to dehydroxylation can be computed from the chemical formula of the clay. Generally, the actual weight loss, with reference to the weight of the clay dried at 1 0 5 ° C , is approximately 2% higher than the theoretical weight loss. Therefore, there may be some strongly adsorbed water in the clay dried at 1 0 5 ° C. that is lost in the same temperature range in which dehydroxylation occurs. Such strongly ad­ sorbed water may be retained in the holes in the unit layer surface; alternatively, the excess weight loss may be due to strongly held oxygen atoms in these holes. Another interpretation, proposed by McConnell (5), is the assumption that part of the silicon-oxygen tetrahedrons in the clay are replaced by ( O H ) tetrahedrons to such an extent that the excess weight loss corresponds with the dehydroxylation in the tetrahedral part of the unit layers. A re-evaluation of the analysis of the density data on the basis of either model does not significantly change the estimates for the interlayer water density. 4

Clays in Polar Organic Liquids (Penetrating Liquids). In general, the ap­ parent densities of clays in organic penetrating liquids are the same as those obCopeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

338

ADVANCES IN CHEMISTRY SERIES

tained in water. For example, the apparent density of sodium Wyoming bentonite in ethylene glycol is 2.818 grams per cc. as compared with 2.793 ± 0.008 grams per cc. in water; the apparent density of native Wyoming bentonite in dioxane is 2.843 grams per cc. and in water is 2.835 grams per cc. Therefore, the same packing considerations apply for these liquids as for water. In some systems, however, the apparent density of the clay in the organic liquids is appreciably higher than in water; this indicates that the particular organic is more closely packed in the interlayer space. An example is the pyridine-bentonite system, which will be discussed in another paper. Sometimes, abnormal values can be attributed to chemical changes in the organic liquid which may occur when it is contacted by a clay. For example, the erratic apparent density values' obtained for Wyoming bentonite in acetone, which drifted as a function of time, were determined to be due to the partial conversion of acetone by the dry clay to diacetone alcohol. Density of Water at Larger Distances from the Clay Surface. Since the previously quoted work by Anderson and Low ( 1 ) covers density anomalies in the water phase at larger distances from the clay surface, we attempted to check their results pycnometrically by comparing systems of clays' with preadsorbed water, using n-decane as the displacement fluid and systems of totally immersed clays. As previously mentioned, the apparent densities of the clays in both systems were identical. This result seems to contradict the data of Anderson and Low. If their data were applicable, the apparent density of the clay immersed in water would have been about 20% lower than that of the clay with preadsorbed water immersed in n-decane. According to the limits of accuracy of the pycnometer experiments, and with the assumption that no density anomalies occur at the n-decane-water interface, the average density of the water in the range of waterclay ratios in which density anomalies were observed by Anderson and Low would not deviate more than 0.02 to 0.03% from the normal value, instead of the approximate 1.5% low average value derived by them. E X P E R I M E N T A L D E T A I L S . A fair comparison between the apparent densities of clays immersed in water and of clays with a certain number of preadsorbed monolayers immersed in n-decane requires that each preadsorbed monolayer of water between the unit layers is completed, so that no vacant space within a monolayer exists. The clay should be in the same state of hydration in the entire system. The selectively accessible void space should be completely filled, as well as capillaries in the clay aggregates. The homogeneous distribution of the adsorption water was achieved by slowly equilibrating thin flakes of clay with almost saturated water vapor. After about one month of equilibration, the uniform state of hydration of the clay was shown by the sharpness and order of the x-ray diffraction pattern. The completion of the monolayers was judged from the amount of water taken up by the clay, with the knowledge that about 100 mg. of water is needed per gram of clay for the formation of a monolayer. A very suitable system for this type of experiment is the sodium vermiculitewater system, in which the clay is hydrated with two monolayers of water. Since the clay does not hydrate beyond this stage, there is no danger of creating an incompletely filled third monolayer between the unit layers. The water uptake was 270 mg. per gram, which leaves an excess of about 70 mg. of water per gram to fill the void space of about 0.013 cc. per gram and to fill the capillaries by capillary condensation. The apparent density of the clay in this condition in ndecane was 3.049 grams per cc, which is identical with the apparent density of the clay immersed in water (3.045 ± 0.012 grams per cc.). Two additional experiments were carried out with very thin flakes of sodium Wyoming bentonite. According to x-ray analysis, two-layer complexes were formed, and the water uptake was 275 and 304 mg. per gram, respectively, and the apparent densities were 2.768 and 2.786, respectively. The higher value Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.

DEEDS AND VAN OLPHEN

Density of Water Adsorbed by Clays

339

Downloaded by FUDAN UNIV on February 26, 2017 | http://pubs.acs.org Publication Date: June 1, 1961 | doi: 10.1021/ba-1961-0033.ch035

is identical with the apparent density of the clay in water (2.793 ± 0.008 grams per cc. ) within the experimental error. The other value is somewhat low; this can be attributed to an insufficient excess of water to fill the capillaries completely. In their analysis, Anderson and Low assume that the clay dissociates fully into single unit layers which are at the same distance throughout the system. On this basis, the distance of a certain increment of volume of water from the clay surface is derived from the water contents. Although this assumption may be correct for sodium Wyoming bentonite, a complete dissociation of unit layers is not expected to occur in potassium Wyoming bentonite, for which analogous results were obtained. Against the conclusions of our work on sodium Wyoming bentonite, the objection might be raised that the system of the clay with the preadsorbed water in which the unit layers are two monolayers apart is not directly comparable with that in which the clay is immersed in water, if, in the latter system, the unit layers are much farther apart. Since this objection would not apply to the sodium vermiculite system, it would be interesting to study this system with the technique of Anderson and Low. A possible flaw in the method used by Anderson and Low might be that the development of grain pressure in the clay matrix in the range of high clay concentrations causes local high pressures. Such local pressures may cause small local volume changes of the container, or they may cause some compression of the interlayer water, which according to our analysis, has a low density of packing. It can be shown that the latter factor alone could be responsible for, at most, about one third of the total effect. This compression factor would not exist for the sodium vermiculite system. Literature Cited (1) Anderson, D. M., Low, P. F., Proc. Soil Sci. Am. 22, 99-103 (1958). (2) Bradley, W. F., Nature 183, 1614-15 (1959). (3) de Wit, C. P., Arens, P. L., Trans. IVth Inter. Congr. Soil Sci. 2, 59-62 (1950). Moisture Content and Density of Some Clay Minerals. (4) Hauser, E., Le Beau, D. S., J. Phys. Chem. 42, 1031-50 (1938). (5) McConnell, D., Am. Mineralogist 35, 166-72 (1950). (6) Mackenzie, R. C., Nature 181, 334 (1958). (7) Nitzsche, W., Kolloid Z., 93, 110-15 (1940). (8) Tscapek, W., Z. Pflanzenernähr., Düng. u. Bodenk. 34, 265-71 (1934). RECEIVED June 2, 1 9 6 1 .

Copeland et al.; SOLID SURFACES Advances in Chemistry; American Chemical Society: Washington, DC, 1961.