Seven kaolins from Georgia (southeastern U.S.A.), ranging from high to low commercial grade, were characterized by X-ray powder diffraction and chemical techniques to establish that the variation in quality was caused by impurities. The Ca and Cs cation-exchange capacities (CEC) varied from 2.67 to 8.17 and from 3.29 to 8.77 meq/100g, respectively. Selective dissolution and correlation analyses strongly indicated that expandable 2:1 minerals, particularly smectite (1.2-5.9%), were responsible for most of the observed variations in Ca CEC (r = 0.85*). The external surface CEC of kaolinite ranged from 0 to 1 meq/ 100 g. The positive significant correlation (r = 0.90**) between the Ca CEC and the K-mica content (03.9%) suggested that Ca CEC may be related to the degree of mica weathering through an expandable mineral stage.

The Cs-retention capacity (0.19–1.14 meq/100 g) was closely related to Cs-measured vermiculite content (r = 0.80*), and this content plus specific surface (R = 0.93**) or mica content (R = 0.86*). The Cs retention appeared to be primarily related to the presence of interlayer wedges at mica/vermiculite XY interfaces.

The layer charge of five smectites, one vermiculitic material, and five reduced-charge clays was determined by the alkylammonium and structural formula methods. The two sets of results were found to be linearly correlated (r =.961); however, the values that were determined by the alkylammonium method were 20 to 30% lower than those determined by the structural formula method, and the regression slope for their linear relationship was 1.67. The fact that the structural formula method includes the effects of cations on the lateral edges of the clay particles probably accounted for some of the differences in the magnitude of the results but should not have caused the regression slope to deviate substantially from 1.00. Therefore, inaccurate estimates of the packing density of alkylammonium cations in the interlayer space of 2:1 phyllosilicates were deemed responsible for the systematic divergence of the results of the two methods. To satisfy the need for a relationship between the two methods of determining layer charge, an empirical means of adjusting the alkylammonium values has been proposed and shown to yield values of layer charge that are comparable to those determined by the structural formula method.

Authigenic kaolinite and illite are important diagenetic minerals in the Magnus Sandstone, a giant oil reservoir in the northern North Sea. These clay minerals, separated from three wells, show considerable ranges in their oxygen isotopic composition (δ8OSMOW = +9 to + 16%) and hydrogen isotopic composition (δDSMOW = - 55 to - 105%). The variations in δ18O and δD are positively linearly correlated with a high degree of statistical significance for both kaolinite and illite:

$\begin{array}{c}\mathrm{K}\mathrm{a}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{e}:\mathrm{n}=12;\delta \mathrm{D}=6.1{\delta }^{18}\mathrm{O}-169;\mathrm{r}=0.66(>95\mathrm{\%})\\ \mathrm{I}\mathrm{l}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{e}:\mathrm{n}=11;\delta \mathrm{D}=5.9{\delta }^{18}\mathrm{O}-159;\mathrm{r}=0.78(>99\mathrm{\%}).\end{array}$

Formation of the clays in a pore fluid of uniform isotopic composition over a range of temperatures appears unlikely. It is suggested that the observed relationships between clay mineral δ18O and δD are perhaps best explained by a model of precipitation at more or less constant temperature from pore fluids which varied isotopically across the oilfield. The isotopic composition of the formation waters would then lie along the line: δDw = 6.2 δl8Ow - 50. This is most plausibly interpreted as a mixing line with suggested minimal endmembers at (δ18O, δD) values of (+4, -24) and (-4, -76). The first of these represents reasonable isotopic values for Magnus Sandstone formation waters. Although δ18O of the second is compatible with an evolved Cretaceous meteoric water, its δD value is difficult to understand in the context of the model.

Laboratory-hydrolyzed and commercial OH-Al solutions were characterized using kinetics of Al-ferron color development, kinetics of structural OH neutralization with H+, 27Al NMR spectroscopy, and sulfate precipitation. The results showed that the Al13 complexes having the Keggin structure were dominant only in fresh, laboratory-hydrolyzed OH-Al solutions of OH/Al molar ratio = 1.8 and above. These species gradually converted to other polynuclear forms that reacted with ferron slowly, were not detectable by 27Al NMR spectroscopy, and yielded different basic Al sulfates following Na2SO4 addition. These more stable complexes can best be interpreted to have a Al(OH)3-fragment structure. In the three commercial aluminum chlorohydrate (ACH) solutions studied, Al13 complexes accounted for a small portion of the total Al present. More than 80% of the Al was present as species that were not detectable with NMR spectroscopy and resembled the slow-reacting complexes in aged, laboratory-hydrolyzed OH-Al solutions. Small portions of the slow-reacting complexes appeared to be submicron particulates that acted as nuclei for gibbsite formation or aggregates of Al13 complexes that dispersed to Al13 upon dilution. Polyaluminum chloride (PA) solution resembled the moderately aged laboratory-hydrolyzed OH-Al solutions.

The hydrated Mg-vermiculite lattice contains double sheets of water molecules interleaved with silicate layers. In the fully hydrated condition, each water sheet is arranged in a regular hexagonal pattern and the interlayer cations are located midway between water sheets in octahedral coordination.

Release of water molecules from interlayer positions leads at first to a gradual contraction of the basal spacing from 14.81 A, during which the water network becomes increasingly distorted. The contraction ceases temporarily at 14.36 A while further water is withdrawn. When the probability that a cation has a complete octahedral “shell” of water molecules at a given instant falls below a certain value, an abrupt contraction to 13.82 A takes place during which the cations are displaced to sites near the silicate layer surfaces.

Release of further water causes a contraction to a lattice (11.59 A) in which single sheets of water molecules are interleaved with the silicate layers. The 11.59 A phase is stable until only a small proportion of the interlayer water remains, when an approximately regular interstratification of 11.59 A and 9.02 A (dehydrated) layers develops. The final traces of interlayer water are retained in the interior of the crystal with great tenacity owing to the sealing of the crystal near the edges.

Surface area determinations were made on a montmorillonite with various cations emplaced on the exchangeable sites, utilizing nitrogen and carbon dioxide as adsorbates at 77°K and 195°K, respectively, in a dynamic system. From the fraction of a Mississippi montmorillonite less than about 1 μ in size, samples were prepared by replacing the original exchangeable cations with Li+, Na+, K+, Rb+, Cs+, Mg++, Ca++, Ba++, and NH4+, forming a series of homoionic montmorillonite species.

Surface areas from 3-point B.E.T. plots (half-hour adsorption points), with nitrogen as the adsorbate, ranged from 61 m2/g for Li-montmorillonite to 138 m2/g for Cs-montmorillonite, thus reflecting a certain degree of nitrogen penetration between layers. Complete penetration should theoretically result in a surface area of over 300 m2/g for this clay with a nitrogen monolayer between each pair of platelets. The experimental data indicate that the extent of penetration is time-dependent and is also a function of the interlayer forces as governed by the size and charge of the replaceable cation. This finding negates the generally accepted concept that nitrogen at 77°K does not penetrate the layers and provides a measure only of the external surface of expandable clay minerals.

A further measure of the variation of interlayer forces is provided by the adsorption of carbon dioxide at 195°K. Surface area values ranged from 99 m2/g for Li-montmorillonite to 315 m2/g for Cs-montmorillonite. Although the carbon dioxide molecule is larger than the nitrogen molecule, its greater penetration apparently is a result of its being kinetically more energetic (with a larger diffusion coefficient) at its higher adsorption temperature. Similar differences have been found with both adsorbates in the study of microporous substances, such as coal, where activated diffusion is of considerable significance.

Diffusion of K during analytical electron microscopy (AEM) results in anomalously low count rates for this element. As the analysis area and specimen thickness decrease, count rates become disproportionally lower. Adularia and muscovite show different diffusion profiles during AEM; for muscovite a strong dependence of diffusion on crystallographic orientation has been observed. Conditions giving rise to reliable chemical data by AEM are the use of a wide scanning area (>800 × 800 Å) and/or large beam size to reduce the effect of diffusion of alkali elements, a specimen thickness greater than about 1000 Å, constant instrument operating conditions, and the use of a homogeneous, well-characterized standard sample. The optimum thickness range was obtained by determining the element intensity ratio vs. thickness curve for given operating conditions. The standard and unknown should have a similar crystal structure and, especially for strongly anisotropic minerals such as phyllosilicates, a similar crystallographic orientation with respect to the electron beam.

Secondary minerals formed during simulated weathering of nuclear waste glasses have been identified by analytical electron microscopy. A complete description of the reacted glass, from the outermost surface in direct contact with the leachant solution to the reacting front that migrates into the bulk glass, was obtained. Manganese and iron oxyhydroxide phases and saponite were found to have precipitated onto the residual glass surface from the leachant solution. Iron-bearing smectite, serpentine, and manganese and uranium-titanium oxyhydroxides formed in situ in the glass in several distinct bands at different depths beneath the original surface. This sequential development of secondary phases displays a clear trend toward more order and crystallinity in the phases farthest from the reaction front and indicates that complete restructuring of the glass into crystalline phases did not occur at the interface with fresh glass. Additionally, the formation of a discrete uranium-bearing phase, as opposed to uranium uptake by precipitated phases, suggests that stable actinide phase formation rather than ion exchange may be a possible mechanism for retarding radionuclide release to the environment.

Illite diagenesis in Tertiary and Mesozoic shales in the Bergen High area, northern North Sea, was studied using mineralogic, isotopic, and computerized thermal modeling techniques. The Tertiary shales are dominated by smectite, with lesser amounts of illite, kaolinite, and chlorite. At present burial temperatures of >70°C smectite is absent, and the shales contain abundant lath-shaped illite which yields a mixed-layer illite/smectite (I/S) X-ray powder diffraction (XRD) pattern. Transmission electron microscopy (TEM) indicates that the illite laths increase in abundance and thickness with increasing depth; XRD patterns indicate a progressive increase in the illite component of the I/S. The deepest samples were found to contain long-range ordered (R=3) I/S, which showed platy particle morphology with the TEM. K-Ar ages of most of the <0.1-μm-size illite separates imply that illitization was a relatively brief event affecting a thick sequence of sediments during late Cretaceous to early Paleocene time (65–87 Ma); however, measured ages were affected by trace levels of detrital Ar contamination and do not represent the true age of diagenesis.

Several methods of quantifying Ar contamination were used to correct measured ages to obtain a reasonable estimate of the true age of diagenesis. The corrected ages are imprecise due to uncertainties in quantifying the levels of sample contamination, but generally suggest a Paleogene (38–66 Ma) period of illitization. In contrast, simple kinetic models of smectite-illitization suggest much younger ages of diagenesis (0–40 Ma at the Veslefrikk field; 0–60 Ma at the Huldra field). The timing of the diagenesis and the morphologic aspects of the authigenic illite suggest that illite precipitated before late Tertiary compaction and resulted in a decrease in fluid permeability. Low trapping efficiency of early Tertiary sediments, vertical escape of warm fluid from the Brent sandstone, and high heat flow may have promoted illite diagenesis in the shales prior to deep late-Tertiary burial.

Halloysite and kaolinite occur in gneissic weathering profiles developed under a temperate climate in the Leucogia area of NE Greece. X-ray diffraction analyses of samples from three profiles of partially weathered gneisses have revealed that kaolinite becomes more abundant than halloysite with increasing intensity of the decomposition processes. To resolve the genetic relationship between halloysite and kaolinite, microtextures were examined by scanning electron microscopy using polished thin-sections and freshly exposed fracture surfaces, and their chemical compositions were determined using an analytical scanning electron microscope fitted with energy dispersive and wavelength dispersive spectrometers. Different morphological forms of halloysite and kaolinite were detected within the same rock mass. In the earliest stage of weathering, spheroidal aggregates consisting of microcrystalline halloysite are formed on the plagioclase surface. With progressive weathering, spheroidal halloysite converts to tubular halloysite. As weathering advances, tubular halloysite converts to platy halloysite, which in turn converts to kaolinite. Halloysite and kaolinite may coexist in the upper parts of the profiles. Electron microprobe analyses of spheroidal, tubular and platy halloysite and kaolinite show that the chemical composition of the various forms indicate a progressive Fe enrichment and Al depletion with advancing kaolinization, from VIFe0.08VIAl3.70 a.p.f.u. in spheroidal halloysite through VIFe0.29VIAl3.50 in platy halloysite to VIFe0.53VIAl3.32 in newly formed kaolinite. Final-stage, book-type kaolinite approaches its ideal chemical composition (VIFe0.04VIAl3.95). The mineralogy and composition of the kaolin minerals and the development of the various morphological forms of halloysite and kaolinite in different parts of the profiles are attributed to the chemistry of the ambient solutions. However, other physical and microenvironmental conditions (e.g. time and space available) also seem to have strong influence on the precipitation processes producing the morphological variations.

The large contents of octahedral vacancies in published formulae of chlorite from hydrothermal systems and clastic sequences are shown to be largely caused by inclusion of other minerals. Verification is provided by analytical electron microscope (AEM) analyses of chlorite in pelitic rocks from the Gaspé Peninsula, Quebec and the Gulf Coast, Texas. The Gaspé chlorite occurs as discrete crystals locally coexisting with corrensite, and the Gulf Coast chlorite is free of mixed layers other than local serpentine-like 7-Å layers. Unlike most electron microprobe analyses (EMPA), but like other AEM analyses, the reported chlorite formulae do not have significant octahedral vacancies, are not Si-rich and (Fe + Mg)-poor relative to classic metamorphic chlorite, and have nearly equal amounts of tetrahedral and octahedral Al. The studied chlorites and those in metabasites and clastic rocks that could be positively identified as containing no or minimal mixed layering or submicroscopic intergrowths have little or no Ca or alkalis. In contrast, EMPA of chlorite reported for other clastic sequences show variable amounts of Na + K + 2Ca that exhibit a poorly defined positive correlation with the proportion of octahedral vacancies. The EMPA of chlorite from the Salton Sea and Los Azufres geothermal fields that were suggested to contain temperature-dependent amounts of tetrahedral Al (and thus used as “chlorite geothermometers”) show compositional characteristics similar to those reported for several saponite-chlorite transition series in metabasites.

Continuous increases in octahedral occupancy and tetrahedral Al with increasing metamorphic grade are attributed to decreases in abundance of mixed layers or fine-grained intergrown minerals that commonly occur as a result of increasing crystal size and homogeneity in prograde sequences. Use of “chlorite geothermometry” based on the proportion of apparent octahedral vacancies or tetrahedral Al is therefore unwarranted and leads to inaccurate temperature estimates.

Sediments from depths to 670 m in the Barbados accretionary complex and transecting the décollement zone have been studied by transmission and analytical electron microscopy (TEM/AEM). The sediments consist of claystone and mudstone intercalated with layers of volcanic ash. Smectite comprises the bulk of the noncalcareous sediments and forms a continuous matrix enveloping sparse, irregular, large grains of illite, chlorite, kaolinite and mixed-layer illite/chlorite of detrital origin at all depths. The detrital origin of illite is implied by illite-smectite textural relations, well-ordered 2M polytypism, and a muscovite-like composition. K is the dominant interlayer cation in smectite at all depths, in contrast to the Na and Ca that are normally present in similar rocks.

Deeper samples associated with the décollement zone contain small (up to 100 Å thick) illite packets included within still-dominant subparallel layers of contiguous smectite. AEM analyses of these packets imply illite-like compositions. Selected area electron diffraction (SAED) patterns show that this illite is the 1Md polytype. Packets display step-like terminations like those seen in illite of hydrothermal origin. The data collectively demonstrate that smectite transforms progressively to illite via a dissolution-recrystallization process within a depleting matrix of smectite, and not by a mechanism of layer replacement. This illite seems to form at depths as shallow as 500 m and temperatures of 20°-30°C, which is in marked contrast to the much higher temperature conditions normally assumed for this transformation. This implies that the high water/rock ratios associated with the décollement zone are significant in promoting reaction.

The aim of this study was to compare the redox properties of oxides and smectite structural Fe (FeStr). Structural Fe in smectite often accounts for about half of the Fe pool in soils and sediments, and is naturally reduced by indigenous bacteria. This reaction is, however, seldom taken into account in environmental studies. The equation for the oxidation-reduction of FeStr and its application to the calculation of the standard electrode potential of FeStr, E0, are discussed on the basis of literature review. The resulting redox properties were used in a new analysis of previously reported field monitoring data. This agreed well with experimental observations that remained unexplained, and showed that if equilibrium conditions prevail, the FeStr pool in soils will reduce prior to the Fe in oxides. The FeStr pool is, therefore, likely to play a major role in controlling the bio-geochemistry of temporarily water-logged soils, and this reaction should be taken into account in further studies.

The assignment of the 29Si CP/MAS-NMR spectrum of naturally-occurring sepiolite clay was re-examined using 29Si COSY and 1H-29Si HETCOR pulse sequences. Each of the three main resonances at −92.1, −94.6 and −98.4 ppm has been attributed to one of the three pairs of equivalent Si nuclei in the basal plane, and the resonance at −85 ppm to Q2(Si-OH) Si nuclei. On the basis of the COSY experiment, the resonance at −92.1 ppm is unambiguously assigned to the intermediate, near-edge Si sites. The HETCOR experiment revealed that the resonance at −94.6 ppm cross-polarizes almost entirely from the Mg-OH protons, and therefore is assigned to the central Si position. The remaining resonance at −98.4 ppm correlates strongly to the protons of the structural water molecules and therefore is assigned to the edge Si sites. Nearly complete rehydration was achieved at room temperature by exposing sepiolite samples that had been partially dehydrated at 120°C to water vapor or to D2O vapor. The rehydration results support the 29Si NMR peak assignments that were made on the basis of the COSY and HETCOR experiments.

The 29Si CP/MAS-NMR spectrum corresponding to the folded sepiolite structure in which approximately one half of the structural water has been removed by heating to 350°C is reported for the first time. The chemical shift values and relative intensities are significantly different compared to the resonances that are observed in the corresponding spectrum of the true sepiolite anhydride. These observations support the earlier claim that sepiolite heated to ∼350°C exists as a distinct phase to be differentiated from that of the completely dehydrated state.

Adsorption studies indicate that paraquat, diquat, and thionine are bound on bentonite by amounts greater than the measured cation-exchange capacity (CEC) of the clay. Methylene blue, new methylene blue, and malachite green are bound by amounts equal to the CEC. The unipositive organocations form aggregates on the clay surface. Aggregation increases with ionic strength and increases the apparent adsorption capacity by 25%. The aggregates are removed by washing with distilled water. Desorption studies show that the dyes are irreversibly bound, whereas the dipositive organocations are reversibly bound. Ionic strength variation reduces adsorption by 15 and 36% in the monovalent and divalent organocation-clay systems, respectively. In the clay-divalent organocation systems adsorption is greater on Na-saturated clay than on K-saturated clay. Adsorption is unchanged over the pH range 4.5–8.5 and decreases steadily below pH 4.0. Changes in adsorption due to changes in temperature are small. The study indicates that ionic strength is the most important variable in clay-organocation interactions.

Stable, three-dimensionally ordered complexes were formed from synthetically hydrated, highly ordered kaolinite (d(001) = 8.4 Å) and several organic compounds. Removal of the intercalated organic compound by drying or by water washing recovered the 8.4-Å hydrate with its ordered layer stacking essentially unchanged. Some of the complexes were stable for less than a day, whereas others appeared to be stable indefinitely. The compounds that formed ordered complexes were dimethylsulfoxide, formamide, hydrazine-hydrate, 1,1-dimethylhydrazine, ethylene glycol, glycerol, and pyridine. Clay-organic complexes that were prepared from methanol, ethanol, 1- and 2-propanol, acetone, acetic acid, propionic acid, acetaldehyde, N-methylformamide, methylethyl ketone, tetrahydrofuran, and K-acetate were stable only if they were immersed in the intercalating medium and had little or no stacking order.

Many of the organic compounds intercalated by the 8.4-Å hydrate are not known to be intercalated by non-hydrated kaolinite either directly or indirectly. Isolated water molecules appear to be keyed into the ditrigonal holes formed by the basal oxygen of the silicate tetrahedra of the 8.4-Å hydrate. These water molecules, referred to as “hole water,” and fluorine ions that had replaced ~20% of the inner-surface hydroxyls of the 8.4-Å phase sufficiently altered the interlayer bonding to allow an expansion of the inner-layer spaces by a variety of guest molecules. The presence of these guest molecules between the clay layers not only changed the basal spacing and perturbed the infrared (IR) bands arising from the inner-surface hydroxyls, it also shifted the position of the IR band arising from the inner hydroxyl.

Junction probability diagrams show variation in both composition and layer arrangement in mixed-layer clay minerals. These diagrams can represent short-range and long-range ordered, random, and segregated interstratifications. Mineralogical analyses of illite/smectite from shale cuttings, bentonites, and hydrothermally altered tuffs define characteristic reaction pathways through these diagrams. Shale and bentonite analyses fall along pathways joining smectite and illite on diagrams showing nearest-neighbor (R1) layer arrangements. Transition from random to R1-ordered interstratifications occurs in shale samples containing 60–70% illite layers, and in bentonites containing 55–67% illite layers. Analyses of alteration products, however, fall near a line connecting rectorite and illite, which represents the maximum degree of R1 layer ordering. No mineralogical evidence is available to suggest that these alteration samples formed from a smectite precursor. All samples develop next-nearest (R2) and thrice-removed (R3) neighbor ordering along similar pathways. Transition to R2 ordering occurs gradually in samples composed of 65–80% illite layers, and samples containing more than 85% illite layers may show strong R3 ordering.

Muscovite and K-feldspar (adularia) were dry-ground to — 200/inch particles, which were suspended in water. These suspensions were “titrated” with KCl, and the pH was recorded as a function of KCl concentration and temperature. The results indicate that muscovite and adularia react with water to produce a surface film in which H+ has displaced K+. The “titration” curves show some characteristics attributable to exchange reactions, others apparently related to equilibria among solids of fixed composition. Maximum release of K+ from adularia by reaction with water is much greater than that from mica. The interpretation is made that the first result of reaction of mica and adularia with water is a surface layer that grades from an outer portion that is structurally disrupted to an inner portion that retains the original silicate structure but with H+ substituted for K+. Addition of K+ as KCl to the suspending solution displaces H+ from the disrupted zone, but all H+ originally taken up by the solids was not returned to the solution by concentrations of KCl up to 1.0 M. Experiments were of a few hours duration; work by others has shown that the disrupted zone releases appreciable concentrations of silica and alumina to solution over longer time intervals.

These hydrolysis experiments indicate that at 25°C an H-feldspar or H-mica structure is favored over a K-feldspar or K-mica structure except in solutions in which the ratio of aK+ aH+ exceeds 109–10 or 107–8 respectively. These ratios decrease in the temperature range 25-65°C by a factor of about 100.7. These results, where considered in relation to the observed behavior of feldspar and mica under weathering conditions, indicate that the major energy change for the reactions, $2\,mica + 5{H_2}O = \,3\,kaolin + \,2{K^ + }\, + \,2O{H^ - }$$3K - feldspar + 2{H_2}O = \,K - mica + 6quartz + \,2O{H^ - }$ can be considered to result from H+-K+ exchange, and that the energy contribution from other changes is small.

Solid-state nuclear magnetic resonance (NMR) spectroscopy, thermal analysis, and X-ray powder diffraction data on the tubular, hydrous aluminosilicate imogolite were found to be fully consistent with a previously proposed crystal structure consisting of a rolled-up, 6-coordinate Al-O(OH) sheet, bonded to isolated orthosilicate groups. The calculated 29Si chemical shift of this structure agreed with the observed shift within 3 ppm. Thermal dehydroxylation of the Al-O(OH) sheet produced predominantly NMR-transparent 5-coordinate Al, but a few 4- and 6-coordinate sites and some residual hydroxyl groups may also have formed, as shown by NMR spectroscopy. Changes in the 29Si NMR spectrum on dehydroxylation suggest a condensation of the orthosilicate groups, but steric considerations rule out bonding between adjacent silicons. To account for these observations, an alternative mechanism to orthosilicate condensation has been proposed, involving the fracture and unrolling of the tubes, followed by the condensation of fragments to form a layer structure. The layer structure has a calculated 29Si chemical shift of -95.6 ppm, in good agreement with the observed value of -93 ppm.

High-resolution transmission electron microscopy of noncrystalline Fe-Si-Al-oxyhydroxide gels shows a common structure of hollow packed spheres having external diameters ranging from 50 to 1000 Å. Some sphere walls display a concentric structure, particularly if the gel composition is close to that of a crystalline clay mineral (e.g., smectite, kaolin). The spheres probably formed by expansion of void space (bubbles) as the surrounding gel contracted 5–10% because of partial ordering of the Fe-Si-Al-oxygen network. Much of the water contained in such noncrystalline minerals is incorporated within the bubbles.