| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Soil Science and Plant Nutrition, Univ. of Western Australia, Nedlands, Western Australia 6907
b Dep. of Soil Science, Waite Agricultural Research Inst., The Univ. of Adelaide, Private Mail Bag 1, Glen Osmond, South Australia 5064
soilsci{at}cyllene.uwa.edu.au
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
, using 0.108 nm2 as the molecular area of adsorbed water. Accordingly, we recommend that the amount of water adsorbed at
(a saturated solution of CaBr2 at 20°C) be used as a single point determination for the routine examination of large numbers of soil samples as an indication of surface extent. We provide reasons why it is inappropriate to convert the water adsorption value to a surface area. We briefly discuss the significance of surface area measurements in studies of the chemical and physical aspects of soil and clay behavior.
Abbreviations: BET, Brunauer–Emmett–Teller • EG, ethylene glycol • EGME, ethylene glycol monoethyl ether
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
, could be used to measure the hydratable surface area of expanding lattice minerals; this was based on the assumption that a monolayer of the adsorbate formed on the internal and external surfaces of montmorillonite. The authors suggested this method had the potential to be extended to measure the surface area of other clay minerals for which retention was restricted to external surfaces only. The technique based on the same general concept has evolved over time, with the introduction by Bower and Goertzen (1959) of an EG–CaCl2 solvate to provide "a fixed vapor pressure" and the proposal of Carter et al. (1965) to use an EGME–CaCl2 solvate in order to attain equilibrium more rapidly as EGME has a much larger vapor pressure than EG. However, Carter et al. (1986), having noted the wide acceptance (USDA, 1982) of the EGME retention method, expressed reservations about the assumptions and limitations involved in the use of EGME retention by soils and clays as a measure of surface area. These authors discussed the assumption that an EGME monolayer covers all interlamellar and external surfaces and commented that the amount of EGME required to form a monolayer is "not absolutely correct" because coverage in the interlamellar spaces may not be complete. Furthermore, they observed that the role of cation solvation needs to be addressed. As a result of an extensive study of EGME retention by soils and clays, Tiller and Smith (1990) concluded that "the widespread use of the EGME procedures for measurement of the total surface area of soils does not appear to be justified". They specifically noted that the EGME technique could result in serious errors for smectitic soils. This work was substantiated and extended by that of Churchman et al. (1991).
It seems timely and appropriate to appraise the EGME method because information is now available on the vapor pressure of EGME and its saturated solutions, notably of CaCl2 (Kellomäki, 1985). In addition, complete adsorption isotherms for EGME on montmorillonites have only recently become available (Kellomäki et al. 1987; Chiou et al. 1993; Chiou and Rutherford, 1997).
Since the retention of EGME by montmorillonite is the basis for the establishment of the technique, we have specifically addressed the retention of EGME by smectites. In order to place recent work in context it is necessary to trace the evolution of the technique culminating in the use of an EGME–CaCl2 solvate to measure the hydratable surface area of soils and clays.
As a background to the following discussion the physical properties of EGME and water are compared in Table 1 .
|
|
Development of the Ethylene Glycol and Ethylene Glycol Monoethyl Ether Techniques |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
Martin (1955) prevented the steady loss of EG over time from EG-saturated clay samples by the inclusion of a free EG surface in the evacuated vessel as well as the dried CaCl2 in order to maintain a "fixed vapor pressure". For Ca-montmorillonite (Wyoming) he measured a retention of 294 mg g-1 after 12 h, and this value remained steady with time. In the absence of the free liquid surface, equilibrium was not attained in the same time and the retention was <220 mg g-1. It was acknowledged by Martin (1955) and later by Sor and Kemper (1959) that the inclusion of a free liquid surface would produce larger relative vapor pressures than those which might be appropriate for monolayer formation.
Bower and Goertzen (1959) introduced the use of a CaCl2–EG solvate. They reported that the retention of EG by Ca-Mississippi montmorillonite (Panther Creek) varied between 250 and 260 mg g-1 for solvates of EG–CaCl2 with molecular ratios (EG/CaCl2) ranging from 0.19 to 0.94; they also reported that a solvate with a molecular ratio of 1.0 appeared relatively stable at 110°C. They concluded from the above that this solvate would form and would exert a fixed vapor pressure over all compositions up to a molecular ratio of 1.0 provided that some EG was present in the system. As a result, they proposed that a standard solvate containing 20 g of EG per 100 g of anhydrous CaCl2
be used in the measurement of "surface areas". Since the EG in excess of a monolayer, which was originally present in a soil or clay slurry, would be transferred to the standard solvate, the final molecular ratio of the solvate would be >0.36. McNeal (1963) found that Ca-Mississippi montmorillonite (API 21, Polkville) retained 249 mg g-1 and the Na-clay 193 mg g-1 when in equilibrium with the standard EG–CaCl2 solvate with a molecular ratio of 0.36.
Eltantawy and Arnold (1974) measured the retention of EG by homoionic Ca-Wyoming montmorillonite by desorption and adsorption procedures in the presence of solvates with a range of molecular ratios. Under desorption conditions and for ratios of 0.036, 0.045, and 0.060, the amounts of EG retained by Ca-montmorillonite (Wyoming) were 152, 193, and 227 mg g-1, respectively. At a ratio of 0.090, the amount of EG retained was 260 mg g-1, and this remained unchanged at ratios of 0.36 and 0.54. However, at molecular ratios of 0.72 and 0.98 (adsorption and desorption) an equilibrium retention of 350 to 360 mg g-1 was established. Eltantawy and Arnold (1974) also reported, in the presence of EG liquid, a retention of 360 mg g-1, which is appreciably greater than the equilibrium value of 294 mg g-1 obtained by Martin (1955) for the same clay and conditions. These results contrast with those of Bower and Goertzen (1959), who reported an EG retention of 250 and 254 mg g-1, respectively, at molecular ratios of 0.69 and 0.94.
In order to attain equilibrium more rapidly, Carter et al. (1965) proposed the use of EGME, which has a much higher vapor pressure than EG; the saturated vapor pressures at 20°C are 545 Pa for EGME and 7 Pa for EG (Kellomäki, 1985; Hales et al., 1981). The standard solvate composition used by Carter et al. (1965, 1986) in their experiments (i.e., 20 g of adsorbate to 100 g of anhydrous CaCl2) has a molecular ratio of 0.25 for EGME. The authors reported that a solvate with a molecular ratio of 1.5 was stable at 70°C. No relative vapor pressures have been given for this solvate or the one with a molecular ratio of 0.25, but the implication was that they are the same. It is assumed that the EGME solvate has the same relative vapor pressure for all compositions, as was demonstrated by Bower and Goertzen (1959) for EG.
One check of the consistency of the monolayer capacity is provided by examining the occupancy of the interlamellar space of montmorillonite by EGME. This can be considered by comparing the density of the interlamellar liquid with the density of the bulk liquid. This is derived from the mass of EGME retained divided by the interlamellar volume, which is calculated from the surface area and the interlamellar distance; this distance is calculated by subtracting the lamellar thickness from the measured d(001) value for the complex. MacEwan (1948) considered that the unit thickness of the aluminosilicate layer was the O-O distance in the clay lattice (0.66 nm) plus twice the van der Waals radius of oxygen (2 x 0.14 nm), that is 0.94 nm. Commonly accepted values are 0.95 and 0.96 nm, and we have used 0.95 nm in our calculations. The d(001) value for the EGME–Ca-montmorillonite complex is 1.707 nm and with a unit layer thickness of 0.95 nm and a surface area of 750 m2 g-1 (see Footnote 1) the interlamellar volume would be 0.284 cm3 g-1. When the amount of EGME retained is 249 mg g-1, the density of the interlamellar liquid would be 0.88 g cm-3; this is 5% below the bulk liquid density of 0.925 g cm-3, thus indicating a close approximation to the monolayer as defined by Dyal and Hendricks (1950), with very little unoccupied space between the solvated Ca ions. For the interlamellar density of EGME to be equal to that of the bulk liquid, the amount adsorbed would need to be 263 mg g-1. Using both the anhydrous CaCl2 and standard solvate (i.e., molecular ratio = 0.25), Eltantawy and Arnold (1973) found that Ca-montmorillonite (Wyoming) retained considerably less (200 mg g-1) EGME than found by other workers; this led to their incorrect conclusion that there was substantial unoccupied space between hydrated Ca2+ ions. Their claim that a retention of 300 mg g-1 of EGME, in the presence of free liquid, represents filling of the interlamellar spaces is obviously not correct since it is significantly greater than the 263 mg g-1 required for this purpose.
From the above it is clear that the application of EG and EGME as adsorbates to the task of surface area measurement has consistently employed procedures in which equilibrium and the vapor pressure of the adsorbate are very poorly defined. Indeed, there are few cases where equilibrium between a defined vapor pressure and the adsorptive surface have been unequivocally demonstrated. In some cases there is doubt about whether results were obtained during adsorption, desorption or somewhere within the hysteresis envelope. For EG, and to a lesser but significant extent for EGME, the final approach to equilibrium must necessarily be very slow because it is driven by such minute vapor pressure differences. We believe that this slowness could easily be mistaken for equilibrium itself. For both adsorbates there has been very little attention given to the confounding effects of capillary condensation. This has been necessitated by the dearth of information about the relative vapor pressure of the adsorbate during experiments.
Estimation of Surface Area from Monolayer Capacity
From the total surface area and the apparent EG monolayer capacity of H-saturated Wyoming montmorillonite (250 mg g-1), Dyal and Hendricks (1950) calculated that the area occupied by each adsorbed EG molecule was
. The total surface area (810 m2 g-1) was obtained using a (0.525 nm) and b (0.920 nm) axis parameters for the unit cell of montmorillonite and the ideal composition of a H-montmorillonite of [(Si8)IV(Al3.34 Mg0.66)VIO20(OH)4–H+0.66].1
In a subsequent paper, Dyal and Hendricks (1952) reported that Ca-montmorillonite (Clay Spur, WY) had an EG monolayer capacity of 280 mg g-1 leading the authors to suggest that their previous value might be too low, by as much as 10%. They also reported substantial variation (90–260 mg g-1) in EG retention with exchangeable cation.
The work of Carter et al. (1965) was principally directed toward establishing a correspondence between the surface areas measured using EG and EGME solvates. Accordingly, they investigated a montmorillonite (API 21-Polkville), an illite (API 36), and a kaolinite (API 5) and reported reasonably good agreement between the two methods for the illite and kaolinite. The amount of EGME retained by the montmorillonite was 232 mg g-1 from which, using a crystallographic reference area of 810 m2 g-1, they established an area for the adsorbed EGME molecule of
. Using this work as a basis, Heilman et al. (1965) established satisfactory correlation between EGME and EG surface areas for a large number of soil samples and presented a recommended procedure for using EGME, which was adopted by the USDA (1982). However, at this point, apart from the use of ethane adsorption by Dyal and Hendricks (1950) on fixed lattice clays (external surface area only), these surface area estimates had not been compared with those obtained by other methods. In this respect, it is significant that Martin (1955) specifically noted, "since the glycol value is a definite quantity, unique for a given clay, it seems superfluous to introduce the uncertainties of assumed molecular packing etc., that are required for the arbitrary conversion of glycol retention to surface area."
|
Comparison of Ethylene Glycol–Ethylene Glycol Monoethyl Ether with Water |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
. He found a close relationship between the two methods, with the ratio of the mass of water adsorbed to EG retained being 
0.83. He also found a good correlation between each of these quantities and the exchange capacity; this is not unexpected since the process involved is the solvation or hydration of the exchangeable Ca ions. A relative vapor pressure of
was chosen because Ca-montmorillonite and vermiculite have a d(001) spacing of 1.5 nm across the range
and also because the water content of 221 mg g-1 corresponds with a monolayer on both internal and external surfaces. This is essentially similar to the basis of the EG method of Dyal and Hendricks (1950) and also implies that a monolayer forms on the external surfaces of nonexpanding clay minerals. However, this last assumption is not supported by other evidence.
In a series of papers, Orchiston (1953, 1954, 1955a, 1955b, 1959a, 1959b, 1959c) applied Brunauer–Emmett–Teller (BET) theory to water vapor adsorption on a range of clays and New Zealand soils with different genetic origins. In the range
, the amount of water vapor adsorbed on the soils was generally well described by the BET equation and allowed the calculation of monolayer capacities. The values of the BET parameter, c, for this wide range of materials were concentrated in the range 4 to 28, with an overall average of 15; this corresponds with completion of the monolayer at
. Quirk (1955) has indicated that the adsorbed water is clustered around the exchangeable cations and the degree to which this corresponds with a monolayer will depend on surface density of charge.
Using the vapor adsorption results of Aylmore (1960), we calculate that at
the water adsorbed by Rocky Gully kaolinite and Willalooka illite corresponds with respective surface areas of 43 and 187 m2 g-1. This calculation assumes the area occupied by a water molecule is
, which corresponds with the area occupied by each water molecule in a hexagonally close-packed monolayer (Foster, 1948). The N surface areas of these soil clays are 36 and 153 m2 g-1, respectively. However, for water vapor adsorption at
, the calculated surface areas are 79 and 397 m2 g-1, respectively, indicating that the amount of water vapor adsorbed at
is about twice that expected for a monolayer.
Comparison of water adsorption at
with surface areas measured with N suggests, for these external surfaces, that
for the kaolinite and 0.088 nm2 for the illite so that the amount of water adsorbed at
exceeds that expected for an hexagonally close-packed monolayer by 20% in each case. This apparent increase in packing density may arise from capillary condensation and from differences in surface accessibility for these two adsorbates. Capillary condensation is expected in regions of crystal overlap, particularly in wedge-shaped pores with a small dihedral angle (de Boer, 1958); this is particularly likely at higher vapor pressures and may explain the artificially large surface areas obtained at
. Furthermore, in regions of crystal overlap, some of the surface is accessible to water but not to N2; we consider that this area is a small fraction (<5%) of the total surface area.
Over time, a significant amount of information has accumulated on water vapor adsorption by montmorillonites (e.g., Mooney et al., 1952; Tardy and Touret, 1987; Cases et al., 1997; Dios Cancela et al., 1997); however, there is not a comparable body of basic information for EGME retention. Chiou and Rutherford (1997) obtained adsorption isotherms for EGME and water vapor on two Ca-montmorillonites up to
. Initially, the EGME isotherms rise more sharply than those for water. For Arizona montmorillonite (Apache County), the amount of EGME adsorbed increased from
; for water vapor the corresponding amounts are 110 and 310 mg g-1. In an earlier paper, Chiou et al. (1993) found, from a BET plot of EGME results, that the monolayer capacity is 215 mg g-1; for Arizona montmorillonite this monolayer is complete at
. The different shapes of these isotherms is cause for some concern.
The shape of the EGME isotherms for Arizona montmorillonite is virtually of Type I (Sing et al., 1985). Isotherms of this shape are commonly associated with either highly localized adsorption or with adsorption by microporous adsorbates and the characteristic plateau behavior corresponds more with void filling rather than with the attainment of a monolayer. The apparent completion of a monolayer at
corresponds with a value of the c parameter in the BET equation of 2400. Gregg and Sing (1982) have warned that such high values of this c parameter denote highly localized adsorption or micropore filling. This implies that the adsorption of EGME is determined more by cation-exchange capacity and micropore volume than by actual surface area. Less stringent limitations may apply to water. Quirk (1955) has noted that, for water, monolayer completion occurs at about
; this corresponds with a value of the c parameter of 18. Although this is a little lower than the range of c values (20–150) in which use of the BET equation is recommended (Gregg and Sing, 1982; Sing et al., 1985), it provides evidence that the adsorption of water on clay surfaces appears less localized than it does for EGME. Additionally, the shapes of the isotherms reported by Chiou and Rutherford (1997) for water tend to be of Type II and therefore more amenable to interpretation. Similar adsorption isotherms were obtained for Wyoming montmorillonite (Chiou and Rutherford, 1997), but these were displaced appreciably downwards towards the abscissa. The use of Chiou and Rutherford's (1997) EGME isotherm for Ca-montmorillonite (Wyoming) to obtain the relative vapor pressure corresponding with Tiller and Smith's (1990) retention value of 249 mg g-1 is precluded, since the exchange capacity of 0.78 mol(+) kg-1 given by Chiou and Rutherford is appreciably less than the commonly reported value of >0.90 mol(+) kg-1 (Dyal and Hendricks, 1952; Eltantawy and Arnold, 1973; Low, 1980).
The marked differences between EGME and H2O adsorption isotherms clearly indicate that EGME molecules have a higher affinity for the clay surface and hence their use as a measure of hydratable surface area introduces another ambiguity.
Capillary Condensation
Using a slit-shaped pore model and zero contact angle, the Kelvin equation predicts that, for different liquids during desorption, the width of the largest pores that remain filled at a given relative pressure depends on the product (
Vm) of the surface tension () and molar volume (Vm) of the adsorbate. It is worth noting that for EGME this product is more than double the value for water, so that capillary condensation of EGME is likely to be even more of a problem than it is with water. As an example, at
during desorption, the width of the largest pores that remain filled with EGME is at least 2.5 nm (this may be larger if adsorbed film thickness is included); for water this value is 1.5 nm. It is significant that pores of 2.5-nm width abound in smectitic soils and have been found to account for the majority of the external surface area (Murray et al., 1985). Such pores may contribute to the retention of EGME in the presence of a free liquid surface.
|
Ethylene Glycol Monoethyl Ether Retention by a Range of Smectites |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
In connection with studies of the swelling of Na-montmorillonite in relation to applied hydrostatic suctions, Low (1980) reported EGME surface areas, chemical composition, exchange capacity, and b-axis dimensions of a number of Ca-saturated smectites. These results are used here to calculate the total surface area of the materials obtained from the crystallographic composition and parameters; these surface area values are then used as a reference to examine the reported EGME surface area measured by Low (1980) using the procedure of Carter et al. (1965). Since the unit cells have virtual hexagonal symmetry,
, where a and b are the axis dimensions. Thus the area per unit cell is 2b2/
3. As there are Avogadro's number of unit cells in the unit cell molecular weight, we can readily arrive at the specific surface area for each smectite from its unit cell composition.
To obtain the EGME surface area for each smectite, Low (1980) used the amount of EGME retained by Ca-montmorillonite (Wyoming) as a reference and assumed that the EGME monolayer was formed on a total surface area of 800 m2 g-1. We have shown here that a crystallographic reference area of 750 m2 g-1 (see Footnote 1) is more appropriate. Accordingly, in order to effect the comparison of the EGME surface areas reported by Low we have reduced his surface area values by the factor 750/800.
In Fig. 1 , we plot the ratio of these corrected area values to the total surface area obtained from the unit cell composition and b-axis dimension against the proportion of the total unit cell charge arising from substitution in the tetrahedral sheet. This is done because the sorption behavior of these minerals is strongly influenced by the origin of the charge. We refer to the ratio of the surface areas as R; for Ca-Wyoming montmorillonite R is obviously one. The identity of the smectites for which R values were calculated, together with their lattice charges (in parentheses), are shown in the legend of Fig. 1.
|
. This test together with the total chemical analysis indicates that 20 of the 25 clay samples, including the reference Wyoming sample, can be classified as smectites. Of these, 18 are montmorillonites and two are beidellites, in which the proportion of charge arising from isomorphous replacement in the tetrahedral sheet exceeds 0.5 (Güven, 1988). It can be seen from Fig. 1 that, apart from Wyoming montmorillonite itself, 15 of these samples have an R value >0.83; the mean value is 0.93. The proportion of isomorphous replacement in the tetrahedral sheet of these 15 clays ranges from 0 to 0.48, and the exchange capacity varies from 0.58 to 1.14 mol(+) kg-1. However, a plot of R values against exchange capacity does not reveal any clear relationship. It is noteworthy that there is no R value greater than that for Wyoming montmorillonite. The average R value of 0.93 for these montmorillonites provides substantial support for the original finding of Dyal and Hendricks (1950) that a monolayer forms on the internal surfaces of montmorillonites. The Otay clay's (Sample 17) R value of 0.78 seems small, as the Italian sample (Sample 12), with a slightly greater lattice charge, has an R value of 0.87. Sample 22 (Texas) has an R value of 0.71, which is less than the 0.92 for Smithfield clay (Sample 21), a sample of almost identical properties. A reason for this spread of R values for the montmorillonites may be the presence of small quantities of impurities as indicated in Table 1 of Low's paper (1980). Those samples that do not contain quartz, mica, or kaolinite are indicated by an asterisk in the legend of Fig. 1. These discordant values may also be the result of some interstratification in the samples, which could be revealed by considering the charge ratio, that is the ratio of the exchange capacity to the lattice charge [both expressed in the same units, e.g., mol(+) kg-1]. Values of the charge ratio significantly less than one would indicate some degree of interstratification. For the Otay sample, the charge ratio is 0.99, and for the Italian clay, it is 0.90. For the Texas clay, the charge ratio is 0.74, suggesting that interstratification is present. Fourteen of the other montmorillonites have charge ratios between 0.84 and 0.93. These values are not regarded as a sufficient deviation from one for the reason that the exchange capacity is based on oven-dried mass and the lattice charge on ignited mass. It is also relevant that the analyses for four of the clays is taken from Schultz (1969), who used the same basis (ignited mass) for both the exchange capacity and lattice charge; these samples are Belle Fourche, Otay, Polkville, and Wyoming, for which the charge ratios were 0.94, 0.99, 0.99, and 0.90, respectively.
India (Sample 11) and New Zealand (Sample 16), with R values of 0.63 and 0.58, can be appropriately described as beidellites since the proportion of the charge accounted for by isomorphous replacement in the tetrahedral sheet is >0.5 (Güven, 1988), being respectively 0.80 and 0.89 of their unit cell charges of 0.76 and 0.75 electrons per unit cell. The charge ratios for these two beidellites are 0.70 and 0.76, respectively, so that the smaller R values are not necessarily a characteristic of beidellites but may be the result of interstratification. The proportions of the lattice charge balanced by K in the India and New Zealand samples are 0.22 and 0.11, respectively. Coupled with the fact that the lattice charge arises principally from tetrahedral replacement, this may mean that the K is not readily displaced by Ca ions when the samples are Ca-saturated prior to the determination of EGME retention. The interstratification which would arise in this way could account for the smaller values of R obtained for these samples as compared with montmorillonites. In view of the occurrence of beidellites in many smectitic soils (Reid et al., 1996), more detailed consideration of this mineral species is warranted.
There is clear evidence that beidellites behave differently from montmorillonite in a number of ways. The amounts of water retained by Mg-montmorillonite (Wyoming) and Mg-beidellite at
and 130 mg g-1, respectively; these have respective lattice charges of 0.77 and 1.1 electrons per unit cell. (Tardy and Touret, 1987). Slade et al. (1991) have shown that the crystalline swelling of montmorillonites and beidellites in aqueous solutions is significantly different. Sodium-saturated Wyoming and other montmorillonites exhibit extensive crystalline swelling, while Na-beidellites do not swell beyond
even in water; it is necessary to saturate beidellites with Li+ ions in order to obtain extensive crystalline swelling. They observed that Na-Nibost with a surface charge density of 1.14 electrons per unit cell, with 0.74 arising in tetrahedral sites, did not undergo extensive crystalline swelling even in distilled water. They also investigated Drayton clay subsoil (a subplastic clay investigated by Norrish and Tiller, 1976), with a surface charge density of 1.06 electrons per unit cell with 1.02 arising from tetrahedral substitution; this clay behaved similarly to the Nibost clay in that it did not exhibit extensive swelling or dispersion in dilute solutions when saturated with Na+ ions (Slade et al., 1991). Slade and Quirk (1991) also studied the expansion of Ca-smectites in concentrated CaCl2 solutions and found that the transition from a d(001) value of 1.55 to 1.9 nm for montmorillonite was incomplete for beidellites.
Harward and Brindley (1966) sought conditions of solvation which might differentiate beidellites from montmorillonites and from vermiculites. Mg-beidellite dried at 105°C and exposed to glycerol vapor had a spacing of 1.45 nm and Mg montmorillonite had a spacing of 1.76 to 1.78 nm. But when exposed to the liquid, both gave spacings of 1.78–1.79 nm. Thus, beidellites behave like montmorillonites toward liquid glycerol and behave like vermiculite toward glycerol vapor where
, resulting in a contraction of the d(001) spacing. Therefore, it seems that the lower retention of EGME by the beidellites could result from the collapse or partial collapse of the d(001) spacings at the relative vapor pressure maintained above the standard solvate or the anhydrous CaCl2 containing EGME.
It might be expected that beidellite would exhibit intermediate behavior in relation to montmorillonite and vermiculite. This idea is supported by the work of McNeal (1963), who reported an EG retention of 267 mg g-1 for Ca-montmorillonite (Wyoming) and 105 mg g-1 for Ca-vermiculite. It seems reasonable to suggest that the Dyal and Hendricks (1950) hypothesis does not apply to beidellite and perhaps some other members of the smectite group of minerals.
Three of the samples represented in Fig. 1 are described as having a weak or very weak 1.77-nm peak and therefore cannot be regarded as smectites; these are Czechoslovakian 650 (Sample 7), Danish (Sample 8), and Guam (Sample 9). Monte Amiata (Sample 14) and Cameron (Sample 5) have 0.34 and 0.51 atoms of K, respectively, per unit cell, corresponding with 1.7 and 2.6% K, which may be compared with illites, which have 6 to 7% K. Schultz (1969), as noted by Low (1980), observed that Cameron (Sample 5) has considerable interlayer illite. The ratio of tetrahedral sheet to lattice charge for clay Samples 7, 8, and 9 exceeds one. For a set of 17 vermiculites studied by Norrish (1973), the ratio of the tetrahedral to total charge varies from 1.2 to 1.9, and the number of Al atoms in the tetrahedral sheet of the unit cell varied between 2.08 and 2.62. This situation contrasts with that of micas, for which the lattice charge of two electrons per unit cell arises entirely from Al substitution for Si in the tetrahedral sheet, and as a result, the ratio of tetrahedral to total charge is one. Thus, for montmorillonites we would expect this ratio to be between 0 and 0.5, for beidellite between 0.5 and 1, and for vermiculites between 1.2 and 1.9.
|
Ethylene Glycol Monoethyl Ether Retention by Fixed Lattice Clay Minerals |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
as the molecular area (Carter et al., 1965). The illites chosen by Tiller and Smith (1990), when examined by x-ray diffraction, showed no evidence of interstratification; the Tumut illite is a fine-grained muscovite. From the information in Table 2 , it can be seen that the EGME area exceeds the N2 area in all cases when Method 1 is used; this is especially so for the kaolinite samples.
|
2 for the illites and 3 for the kaolinites, suggesting that there may be a greater affinity of EGME for the hydroxyl surface of the kaolinite or a small degree of intercalation at the edge of the kaolinite crystals. The difference in the behavior of illite and kaolinites as revealed by both Method 1 and 2 is somewhat surprising, since for polar molecules the solvation of exchangeable cations would be expected to be a dominant feature (Orchiston, 1955a; Quirk, 1955) and the illites have a higher surface density of charge than kaolinites (Greenland and Quirk, 1962, 1963).
Dyal and Hendricks (1950) measured EG retention (24 h) in the presence of anhydrous CaCl2 and also the ethane surface areas (BET) for two hydrous micas, one an illite (supplied by R.E. Grim) and the other from Evansville, TN. The respective ratios for the EG surface area, using
, to ethane surface area were 2.1 and 3.3, which are very much greater than those of 1.2 to 1.5, for EGME surface area to N surface area, given for illites in Table 2. It is relevant that Carter et al. (1965) have established a general correspondence between EG and EGME surface area. The illites used by Dyal and Hendricks (1950) must have had a significant degree of interstratification to which the polar EG molecule had access.
It can be seen from the values of the ratio of EGME surface areas to N surface area (Method 1) in Table 2 that the assumption that EGME forms a monolayer on the surface of clay minerals with a fixed c-axis spacing is not supported.
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
, which is based on a retention value of 232 mg g-1 and a surface area of 810 m2 g-1 for the Polkville montmorillonite (Carter et al., 1965). Tiller and Smith (1990) reported a retention value of 249 ± 5 mg g-1 for Wyoming montmorillonite and if the value of 750 m2 g-1 (Footnote 1) is used, the molecular area at the montmorillonite surface is
. Although this latter value for the molecular area of EGME will reduce the EGME areas in Table 2 by 14%, the values of the EGME areas for the nonexpanding clay minerals are still considerably larger than the N2 surface areas. A further uncertainty is that within the smectite group of minerals the behavior of beidellite has not been defined. The existing evidence, particularly water vapor adsorption (Tardy and Touret, 1987), suggests that beidellites have less affinity for polar molecules than montmorillonites. Other factors that militate against the use of EGME for measuring the hydratable surface area are that its physical and solvent properties are quite different from those of water (Table 1), and we again emphasize that the relative vapor pressure of the standard EGME solvate (molecular ratio EGME/CaCl2 of 1.5) or the other solvates formed by EGME when it reacts with anhydrous CaCl2 is unknown. Given these circumstances it does not seem appropriate to use EGME to measure the hydratable surface area.
Rodevald (quoted by Baver, 1940) introduced the concept of hygroscopicity at the start of this century. Hygroscopicity was originally defined as the water content of a soil at 94.3% relative humidity (aqueous H2SO4, 10%, w/w) and was considered to be related to the specific surface. Anderson and Mattson (1926) showed that, even though soil colloids may adsorb almost equal amounts of water at high vapor pressures, the quantities adsorbed at lower vapor pressures vary considerably. They used
to measure the adsorptive properties of soil colloids.
Earlier, Orchiston (1953, 1954) had shown that the BET equation described water vapor adsorption by soils and clays reasonably well. Because monolayer capacity was attained in the vicinity of
, Quirk (1955) suggested that a single point determination of water vapor adsorption by soil or clay could be made by equilibrating soil samples, in a vacuum desiccator, with a saturated solution of CaBr2, which yields
at 20°C (Hedlin and Trofimenkoff, 1965). Newman (1983) proposed the use of a saturated Ca(NO3)2 solution, which yields
. The basis for this suggestion was that at
a monolayer forms on the internal surfaces of montmorillonite and that capillary condensation of water would not make a significant contribution to the amount of water adsorbed. However, we have shown that the amounts of water retained by an illite and kaolinite at
suggest, when
is used, surface areas that correspond reasonably well with those obtained by N adsorption and that water adsorption at
suggests artificially large surface areas of about twice those obtained by N adsorption. As an example, for Ca-Willalooka illite (see above), the amount of water adsorbed under these conditions is equivalent to 2.6 monolayers (using the surface area measured by N adsorption as a reference). Moreover, the shape of the N adsorption isotherm for this illite is virtually identical to an isotherm interpreted by de Boer (1958)(Fig. 22) in terms of slits and closed wedge-shaped pores. As stated earlier, wedge-shaped pores are sites of early capillary condensation which contribute to enhanced adsorption.
The situation with respect to montmorillonite is somewhat different in that the total amount of water adsorbed is made up of two components: that which is in the interlamellar spaces and that adsorbed on the external surfaces of quasi-crystals (Aylmore and Quirk, 1971), as indicated in Table 3 for two montmorillonites.
|
, then we should use the value of
, suggested by comparison of water and N2 adsorption for kaolinite and illite, rather than 0.108 nm2, the value arising from an hexagonally close-packed monolayer. We can now estimate the amount of water incorporated in interlamellar spaces. This, together with x-ray diffraction data and the assumption that the density of this water is 1.0 g cm-3, allows the assessment of the packing efficiency of water in interlamellar spaces. In Table 3, these calculations are shown for two montmorillonites and disclose, in each case, that this efficiency is <50%. This calculation highlights the difficulties of extracting surface area from water vapor adsorption by smectites.
By contrast, the significance of surface density of charge for interlamellar adsorption of water is brought into sharp focus when the adsorption of water by Mg-Llano vermiculite is considered. The material as used by van Olphen (1969) has a total surface area of 741 m2 g-1, obtained in the same manner as for montmorillonite (Footnote 1), and an external area of only 3.1 m2 g-1; the exchange capacity was 2 mol(+) kg-1. At
the amount of water adsorbed is 175 mg g-1 and the interlamellar volume
is 0.178 cm3 g-1 so that if the density of the water is 1.0 g cm-3, the interlamellar space is fully occupied. The number of water molecules per Mg ion is 9.7, which is similar to the 9.4 molecules per Ca ion estimated to be in the interlamellar space of montmorillonite at
.
From a single crystal x-ray analysis of Ca-Llano vermiculite, Slade and Radoslovich (1985) concluded that the Ca ion has two hydration states, one with eight water molecules of hydration sitting between two di-trigonal cavities in opposing layers and another with six molecules between oxygen triads associated with an isomorphous replacement site of opposing surfaces. An estimate is that there are approximately equal numbers of the two hydration states (P.G. Slade, 1998, personal communication). Thus, on average there would be about seven water molecules for each Ca ion. Between the hydrated cations there is a water network consisting of two sheets arranged in a distorted hexagonal pattern. The locations of water molecules are determined by the surface configuration of the silicate layers, each water molecule being linked by a H bond to a single oxygen in the silicate layer surface and weak H bonding exists within the individual water sheets (Mathieson and Walker, 1954).
Thus at a clay surface, water molecules are retained by two mechanisms: the hydration of the exchangeable cation (primary cation hydration shell) and H bonding to the clay surface (surface hydration). At
, Ca-montmorillonite (Wyoming) has 9.4 water molecules per Ca ion in the interlamellar space and less than half the interlamellar region is filled (Table 3); that is, the interlamellar adsorption at this stage is predominantly involved in cation hydration. However, at
the interlamellar volume of 0.185 cm3 g-1 (see above) is virtually full as the amount of water adsorbed by Ca-montmorillonite is 180 mg g-1 (Aylmore, 1960). The formation of the water network is delayed until p/p0 exceeds 0.19. This contrasts markedly with the behavior of external surfaces.
The similar molecular areas of water adsorbed on Ca-kaolinite and illite at
(am = 0.090 and 0.088 nm2, respectively) suggests that the adsorption of water is almost independent of the surface density of charge on external surfaces. Although the amount of water bound by cation hydration on external surfaces should increase with surface density of charge, this effect is masked by the presence of the water which is H bonded to the surface. The exchange capacities of kaolinite and illite are 0.062 and 0.414 mol(+) kg-1, respectively, and by reference to the N2 surface areas, given above, the surface charge densities are 0.166 and 0.261 C m-2. Expressed alternatively, a Ca ion is associated with 1.92 nm2 of kaolinite surface and with 1.22 nm2 of illite surface; the equivalent value for the external surfaces of Ca-montmorillonite is 2.50 nm2 (0.128 C m-2). The relative amounts of water associated with cation hydration and with surface hydration or network water (Mathieson and Walker, 1954) can be assessed by dividing the number of adsorbed water molecules at
by the number of exchangeable Ca ions. For the kaolinite this value is 21.5 and for the illite 14.0. Slade and Radoslovich (1985) have reported cation hydration numbers for Ca-vermiculite of six to eight, so it can be seen from the above values that the surface associated or network water decreases with surface density of charge since the amount of hydration water must increase with increasing surface density of charge in circumstances when a little more than a monolayer is formed on the surface. At
when there is somewhat more than two layers on an external surface and, in an expanding lattice mineral, there are two layers in the interlamellar space, the amount of surface hydration water will be less in the interlamellar space since there will be twice the number of cations in this space as there is on external surfaces.
While it is acknowledged that, in the routine examination of a large number of soil samples, some general assessment of the surface extent as well as other properties such as particle-size distribution, mineral composition, and exchange capacity are clearly of value, we argue that it is logical that water should be the preferred liquid when the hydratable surface extent of a soil or clay is being considered. The use of water adsorption for comparative purposes presents no real difficulties since large numbers of samples can be placed in a vacuum desiccator in the presence of a saturated solution of a salt. Arai et al. (1976) have provided relevant information about the vapor pressures for several saturated salt solutions. It is noteworthy that Dios Cancela (1997) has constructed water vapor isotherms by using a series of saturated salt solutions. We consider that for a one point determination, adsorption at
is an appropriate indicator of the surface extent since it arises from the use of the BET equation, even though the fundamental assumption for this equation is not strictly met by water molecules clustering around the exchangeable Ca ions rather than forming a hexagonally close-packed monolayer as occurs in the van der Waals adsorption of liquid N2. However, it is clear that such nonspecific adsorption cannot be expected from any adsorbate that is also capable of gaining access to the interlamellar regions of expanding lattice clay minerals; in effect, these requirements are mutually exclusive.
Martin (1955) recommended that it was inappropriate to convert an EG retention value to a surface area in view of the uncertainty concerning the molecular area. The information presented here shows that this reservation is equally valid for water. Such conversions are further confounded by the presence of oxides (Tiller and Smith, 1990) and organic matter as typified by the adsorption of water vapor on peat (Chiou and Rutherford, 1993). These will influence the values obtained for adsorbed water at
. Accordingly, we propose that for comparative purposes the amount of water vapor adsorbed at
be used and that it be designated as hygroscopicity.
Because the concept of surface area within a soil is a complex one, it is necessary to emphasize that the use of surface area to interpret chemical and physical behavior of soils and clays requires special care as the connection between surface area and a soil property is not direct. For example, the swelling of a Ca-illite is the result of the interplay of interparticle forces within the porous matrix established by the organization of the clay particles themselves, as the primary particles (crystals) are arranged in compound particles designated as clay domains (Aylmore and Quirk, 1960; Quirk and Aylmore, 1971; Quirk, 1994). The size of the pores within the clay domains is determined by the distribution of thickness of the primary particles, which in turn is related to surface area (Aylmore and Quirk, 1967; Sills et al., 1973). These pores will be slit- or wedge-shaped with a small dihedral angle, and within these pores repulsive forces operate. These forces result from the perturbation of the water structure near the clay–solution interface and their operation is restrained by ion–ion correlation forces in regions of particle overlap in which the surface separation is <1.0 nm (Kjellander et al., 1988; Quirk, 1994). Furthermore, the clay domain structures are in turbulent or random array so that there are discontinuities within a clay mass which follow the surfaces of these domain structures (Murray and Quirk, 1990a, 1990b). The extent to which the enlargement of these discontinuities contribute to the swelling process, as a result of relaxation in the overall structure with decreasing suction, has not been established. With increasing quantities of exchangeable Na, diffuse double layers develop in the domain pores, and some of the particles are removed from within the potential wells that exist where the primary particles overlap.
There has been undue emphasis and concern about the total area of montmorillonite. For Ca-montmorillonite, the operative particles are the quasi-crystals (Quirk and Aylmore, 1971), which are organized into clay domains. Since the crystalline swelling to
is complete at a suction of 8.8 MPa (Slade and Quirk, 1991; Quirk, 1994), the swelling that occurs at smaller suctions, of more immediate concern for agriculture, results from the interaction between quasi-crystals. Hence the external surface area of these entities is involved. There is evidence that shows that the external surface area of these quasi-crystals varies during the swelling process (Ben Rhaiem et al., 1987; see also Table VIII in Quirk, 1994). By contrast, the whole crystallographic surface area of Li- and Na-montmorillonites is involved in extensive crystalline swelling. However, for Na-beidellites, extensive crystalline swelling does not occur (Slade et al., 1991).
Most of the reported surface area determinations for smectites have been for mineral samples. Murray et al. (1985) have measured the N surface area and pore-size distribution of soil aggregates, as sampled in the field, for twelve smectitic soils from the Emerald Irrigation Area in Queensland. The N surface areas of these soils, expressed in terms of the clay fraction, ranged from 70 to 200 m2 g-1. These surfaces do not exist in isolation, but constitute the walls of slit-shaped pores that are <10 nm across and are predominantly narrower than 3 nm. These pores are within quasi-crystals and domain structures within the soil. The amount of the total soil porosity accounted for by such pore sizes ranges from 20 to 70%. Thus in terms of the physical behavior of these soils, the N or external surface area of the clay is significant since the interlamellar swelling is complete at a suction of <10 MPa, the approximate limit for normal shrinkage.
In relation to the adsorption of ions at inorganic interfaces, the use of surface area measurements may be informative. For a number of goethite samples, Atkinson et al. (1972) found that, when the maximum amount of orthophosphate adsorbed was expressed in terms of the N2 surface area, each phosphate ion was associated with 0.66 nm2 of surface. They noted that Fe atoms, in single coordination with OH, in contiguous unit cells at the (100) face were at an appropriate distance to form a Fe-O-P-O-Fe bridging ligand and that as each unit cell had an area of 0.301 nm2, a surface coverage of 0.60 nm2 would be expected for the dominant (100) crystal face. Such a complex would be expected to be relatively inert and the authors confirmed this interpretation by studying the kinetics of isotope exchange. Such an agreement between crystallographic characteristics would not be expected for kaolinites in which the dominant surface is the basal plane. In fact, Kafkafi et al. (1967) proposed the formation of a bridging ligand for orthophosphate ions between Al atoms at the edge surface of the crystals, which contribute only 10 to 15% of the total surface measured by N2 adsorption.
Finally, we indicate that Mering (1946) has reported that at 350 to 550°C, the interlamellar surface of montmorillonite no longer hydrates so the extent of the external surface could be assessed by water vapor adsorption. Also, from the point of view of discriminating between montmorillonite and beidellite, the use of EGME in the presence of anhydrous CaCl2 merits attention.
|
Conclusion |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
is not significantly affected by surface density of charge when external surfaces are involved, but the amount of water adsorbed in interlamellar regions is clearly dependent on surface charge density, as shown when a comparison is made between Ca-Wyoming montmorillonite and Llano vermiculite. There are enough difficulties in using a polar molecule such as water without compounding these difficulties by using a polar liquid with quite different physical properties (indicated in Table 1).
In view of the information presented here, we recommend that for the routine determination of the surface extent of soils the amount of water adsorbed at
be used and that this be designated as water adsorbed at
or perhaps hygroscopicity (0.19). The interpretation of such values would be facilitated by exchange capacity and mineralogical measurements.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
; if we accept these more recent values together with the John C. Lane Tract analysis for Wyoming montmorillonite (P.G. Slade, 1997, personal communication) of [(Si7.928Al0.072)IV(Al3.057Mg0.434Fe3+0.414TiII0.012MnII0.001)VI O20(OH)4–Ca0.358Na0.014K0.004)], we arrive at an area of 748.6 m2 g-1 for the Ca-montmorillonite. The charge deficiency from chemical analysis of the lattice is 0.740 elementary charges per unit cell (compare the sum of the charge balancing ions of 0.734); this leads to an expected exchange capacity of 0.99 mol(+) kg-1. In this study, we have used 750 m2 g-1 as the reference area for Wyoming montmorillonite. 
Received for publication October 21, 1997.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONDevelopment of the Ethylene...NOTESComparison of Ethylene Glycol...Ethylene Glycol Monoethyl Ether...Ethylene Glycol Monoethyl Ether...DiscussionConclusionREFERENCES |
|---|
