Building clay media with variable anisotropy in particles organization
Clay particles (particularly ones from in this work) have generally platelet shape with a low aspect ratio (i.e., thickness /diameter of the basal surface of < 0.1 (Fig. I.8)), as calculated from Ar adsorption measurements for kaolinite, vermiculite, and illite (Hassan et al., 2005; Reinholdt et al., 2013); Chapter III in this work, respectively). Such morphology dictates that particles tend to settle on their basal surfaces in an aqueous environment, a behavior known as the preferred orientation of clay particles. Therefore, two preparation methods were used to produce compressed cylindrical porous media (i.e., compressed samples), either isotropic or anisotropic, regarding particles’ preferred orientation. Isotropic samples were prepared by uniaxial compression of dried clay powder. For that, the clay powder was directly placed inside (poly)methyl methacrylate (PMMA) tubes (diameter of 0.95 cm and a height of 7 cm) designed to be mounted directly to the TD setup. The samples were compressed with the aid of two metal pistons (one on each side) using a hydraulic press, as illustrated in Fig.II.1. For the Anisotropic samples, thanks to a protocol proposed for kaolinite by Dabat et al. (2020), we developed a methodology that allows to; (i) obtaining samples from either swelling or none-swelling clays having the same total porosity (ε) as the isotropic ones. Still, at a considerably higher degree of anisotropy in particles organization. (ii) to directly perform TD experiments. This protocol consists of the following steps, as illustrated in Fig.II.2:
1. Prepare a clay dispersion from the desired clay material at a known concentration. For that, the mass of the clay was decided based on the aimed porosity and final thickness of the sample (more details ahead). The volume of the dispersion corresponds to the required centrifugation runs (i.e., ten runs). For instance, samples for TD experiments are prepared in PMMA tubes with a diameter of 0.95 cm and a height of 7 cm. By subtracting the volume occupied by the (PTFE) cap (Fig. II.2) and the final sample volume at the end of the preparation. Hence, the remaining volume of the tube is ~ 3cm3 (Fig. II.2d). This volume can be filled with ~3 mL of dispersion. Therefore, for ten centrifugation runs, we need ~30 mL of dispersion. For the solid /solution ratio in this work, we used clay dispersions at a concentration ranging from 20 to 40 g L-1.
2. The second step starts with transferring 3 mL of the clay dispersion into the tube sealed from the bottom with a PTFE cap (II.2a). Then, the tube is placed into the centrifuging swinging bucket rotor (II.2b). Note that we used a special adapter to prevent tube rotation inside the containers (II.2a).
3. In the third step, horizontal centrifugation runs at ~23 850 g for 25 minutes were performed (Centrifuge Avanti J 301, rotor JS-24.38 from Beckman Coulter®; Fig. II.2b). After each run, the clear supernatant was siphoned, and a new aliquot was added ((Fig. II.2c)). These steps were repeated ten times to obtain a sufficient mass in the tube (between 1-1.5 g) (II.2.d).
4. The fourth step, drying the sample: after centrifugation, the sample must be dried to determine the dry clay mass necessary for ε calculation (more details ahead). However, an unsuccessful attempt to dry one vermiculite sample (Fig.II.3a) at 60˚C proved to introduce significant shrinkage and distortion of its structure; a similar observation was reported in Loeber, (1992) and Durrieu et al., (1997). Therefore, we developed the protocol previously used for clay cakes preparation (Loeber, 1992; Pret, 2003) to dry the sample without compromising its geometry and particle organization. For this, the sample was first flash-frozen by immersion in slush nitrogen (SN2) for 2-5 minutes (Fig.II.2e) and then dried by lyophilization at – 40°C for 2-3 hours (Fig.II.2f). Complete water removal was verified when the sample maintained a stable mass during the last hour of lyophilization. Finally, the sample was mounted to the TD cell and uniaxially compressed to the desired height (Fig.II.2g-i).
It is necessary to mention that SN2 was used instead of liquid nitrogen (LN2) because the latter introduced deformation in the sample, as shown in Fig.II.3b. Such deformation means that the internal organization of the particles is compromised, and the non-homogenous freezing of the sample causes it. Indeed, when the sample is immersed in LN2, it raises its temperature to the boiling point (i.e., ‒196 °C). Hence, a layer of nitrogen vapor is formed around the sample that impedes heat transfer with the LN2. This phenomenon is known as « gas film » (Goldstein et al., 1992; Pret, 2003).
To avoid the formation of the gas film, LN2 is placed under a high vacuum for few minutes until its complete solidification. And by suddenly breaking the vacuum, a « metastable paste » of solid and liquid nitrogen is formed with an average temperature of approximately ‒210°C (i.e., SN2) (Goldstein et al., 1992; Pret, 2003). Thus, the sample temperature is no longer sufficient to raise nitrogen temperature to its boiling point at – 196°C.
Fig. II.1. Illustration of the method used to prepare clay samples at a low degree of particles preferred orientation (i.e., isotropic). (a) clay dried powder is placed in ploy methyl methacrylate (PMMA) tubes sealed from bottom with metal piston. (b) clay powder is compressed to a certain height by hydraulic press after installing an upper metal piston (c) the two pistons are removed after compression.
Total porosity ε of all samples used in this work was calculated as the difference between the sample volume (considering the diameter of the tube and the sample’s thickness after compression, L) and the volume occupied by the solid only (i.e., clay particles). The solid volume can be determined based on the sample dry mass and the clay grain density. The latter was calculated to be 2.62 g cm-3 for Na-kaolinite based on the structural formula and the crystal structure parameters refined by Sakharov et al., (2016) for KGa-2 kaolinite. For the 0.1-0.2 µm size fraction of Na-IDP used in diffusion experiments, the grain density was calculated to 2.83 g cm-3, based on the structural formula, the (a, b parameters) and the c* parameter obtained during this work (Chapter III). Finally, for the 0.1-0.2 µm size fraction of Na-vermiculite, a grain density of 2.73 g cm-3 was used based on the (a, b) cell parameters and the chemical composition of the 2:1 layer reported by Arguelles et al., (2010) and the c* parameter value for dehydrated layers at 10 Å (Gieseking, 1975). However, Na-vermiculite displays a mono-hydrated state (1W) at room humidity and a bi-hydrated state (2W) in water-saturated conditions, respectively (Faurel, 2012). Thus, two hydrated densities of 2.49 and 2.24 g cm 3 for 1W and 2W Na-vermiculite, respectively, were determined based on water content and c* parameters derived by Faurel. (2012) for 1W (4.13 H2O/O20(OH)4, c* = 11.97 Å) and 2W layers (9.9 H2O/O20(OH)4, c* = 14.86 Å). The value of hydrated density at 1W is essential to consider when preparing the sample from vermiculite powder (i.e., isotropic sample), it allows for considering the quantity of adsorbed water at room humidity when measuring the powder mass.
Fig. II.2. Illustration of the protocol used to prepare clay samples at high degree of particles preferred orientation (i.e., anisotropic). (a) Clay dispersion is placed in ploy methyl methacrylate (PMMA) tubes sealed from bottom with poly tetrafluoroethylene (PTFE) caps placed in turn into swinging buckets. (b) PMMA tubes are centrifuged at ~23850 g for 25min in a horizontal swinging bucket rotor. (c) Supernatant is removed, and new clay dispersion is added. (d) 10 centrifugation cycles are performed to obtain enough mass in the tube. (e-f) The tube is immersed in slush nitrogen (SN2) for few minutes to remove adhesion between the PTFE cap and the sample and immediately placed into a Lyophilizer to remove water by sublimation. (g) The sample is compressed to the desired height corresponds to the designed total porosity value. (h) Sample is mounted to the diffusion cell and the final height is maintained by threaded rods and nuts. (i) Diffusion cell is mounted into the Through-diffusion (TD) setup.
Note that for Na-kaolinite and Na-IDP, only the interparticle porosity (εinterp.) is accessible by the water tracer, resulting in ε = ε interp.. By contrast, Na-vermiculite exhibits both ε interp. and interlayer porosity (ε interl.). By considering the difference in the c* parameter values between dehydrated layers at 10 Å (Gieseking, 1975) and fully water-saturated layers at 14.86 Å (Faurel, 2012), the interlayer volume corresponds approximately to 1/3 of the particle thickness in water-saturated conditions (i.e., during diffusion experiments). Consequently, the proportion of the ε interp. vs. ε interl. porosities for Na-vermiculite can be calculated for using the following relation:
Finally, for each sample used in this work, a duplicate was prepared for particle organization measurements (more on this in the next section).
Fig. II.3. Images of 0.1-0.2µm Na-vermiculite samples prepared by centrifugation to obtain high anisotropy in the particles preferred orientation. (a) Deformation in the geometry of the sample dried at 60˚C. (b) Deformation in the sample that was flash-frozen by plunging into liquid nitrogen (LN2).
(C) None-deformed sample that was flash-frozen by plunging into slush nitrogen (SN2). (d) longitudinal section in one sample to show the successive beds of clay particles, each two beds (dark and light colored) correspond to one centrifugation cycles.
Induration and slicing of the clay porous media
Induration of the duplicate samples was necessary to produce thin sections (i.e., lamellas) from which the two-dimensional X-Ray Scattering (2D-XRS) measurements were obtained (details are reported in the next section). The induration method relied on the thermal impregnation with methyl methacrylate (MMA, C5H8O2). This technique was developed by Sammaljärvi et al., (2012) and successfully applied to Na-kaolinite samples in Dabat et al., (2020). The duplicates samples were prepared in polyetheretherketone (PEEK) tubes (diameter of 0.64 cm and height of 7.5 cm) instead of PMMA because the latter can dissolve when in contact with MMA. Perforated PEEK caps and cellulose membranes with a pore size of 0.1 μm were set at each side to maintain the total porosity and particle organization during the induration process (Fig.II.4a). First, the tubes (containing the samples) were placed under a vacuum for 20 minutes in a hermetic cell to remove the residual water that may prevent MMA vapor from penetrating the sample pores. Next, liquid MMA mixed with thermal initiator benzoyl peroxide (BPO, at 0.5% of MMA mass) was introduced into the tubes without breaking the vacuum (Fig.II.4b). Sustaining the vacuum is necessary to maintain the pores in air-free condition, thus facilitating the access of MMA gas to the smallest pores. The tubes were kept in contact with liquid MMA for a few days up to two weeks (depending on the sample permeability) to promote the capillary saturation of the pores by the MMA+BPO mixture as much as possible. After saturation, the tubes were sealed with PTFE caps and placed in a water bath at 55°C for at least 24 hours (Fig.II.4c) for a complete polymerization of MMA into PMMA.
After induration, the samples were sliced with a circular saw longitudinally and transversely with respect to the tube axis (Fig.II.4d). The resulting slices were reduced on a polishing table to lamellas with a thickness of ~500 µm. This thickness is necessary to allow sufficient transmission of the X-ray beam during the 2D-XRS measurements (Dabat et al., 2020).
Analyzing the 〈preferred〉 orientation of clay particles
This work uses the P2 order parameter to describe the degree of anisotropy in a particle’s preferred orientation in the samples. Hence, we discuss in this section the theory behind the degree of anisotropy in particle’s preferred orientation and how to obtain the Orientation〈〉 Distribution Function (ODF) from the 2D-XRS measurements, from which the P2 order parameter is calculated (Dabat et al., 2019, and references therein).
Fig. II.4. Illustration of the method used to endure the clay samples. (a) cross-section in the sample tube. (b) hermetic cell used to put the sample under vacuum and to introduce the mixture of methyl methacrylate (MMA, C5H8O2) and the thermal initiator benzoyl peroxide (BPO). (c) the sample is placed in a thermal bath at 55˚C. (d) After induration, the sample is cut into lamellas: (i) the sample is cut into two vertical halves, (ii) cutting the longitudinal and transversal lamellas (the cutting planes are shown in light color cuboid). The lamellas are reduced to a thickness of 500 µm by polishing. (e) longitudinal lamellas of Na-vermiculite.
The Orientation Distribution Function (ODF)
The orientation of a platelet-shaped clay particle can be defined in an orthogonal framework as a function of the vector perpendicular to its basal surface zʹ (Fig. II.5), the latter being defined in spherical coordinates as a function of two angles: θ, which is the azimuthal angle between zʹ and the z-axis, and Φ, the polar angle in the (xy) plane (Fig. II.5). For clay particles with uniaxial symmetry whose z-axis is the only symmetry axis (Cebula, 1979; Wenk et al., 2008; Hubert et al., 2013, and references therein), the probability for a particle orientation is equal over all possible Φ angle values. Therefore, the ODF can be described as a function of only its θ angle, i.e., ƒ(θ), with the following ( )≥ constraints:0 Eqs. (II.2) and (II.3) indicate that the ODF is always positive, with the same probability of the particles pointing upward or downward. The term ƒ(θ)sin(θ) in Eq. (II.4) corresponds to the normalization of all the particles in the sample which have an angle θ with the z-axis.
Two-dimensional X-ray Scattering (2D-XRS) measurements
As discussed earlier, ODF can be extracted from 2D-XRS diffractograms. In this work, these measurements were performed at the Laboratoire de Physique des Solides (LPS) – Orsay, in collaboration with Erwan Paineau.
The experimental setup for 2D-XRS is illustrated in Fig.II.6. The X-ray beam is generated via a copper rotating anode generator (RU H3R, Rigaku Corporation, Japan) and equipped with a multilayer W/Si mirror (Osmic), which generates a monochromatic beam (λCuKα = 1.5418 Å) with a spot size of 1 mm2 that passes through the slits to cover a 600 x 600 µm² surface of the sample. The X-ray beam is directed perpendicular to the studied sample by mounting it on a goniometer head (Fig.II.6). The scattered signals (the diffraction cones) from the particles in the sample are collected on a 2D-detector (MAR345, marXperts GmbH®, Germany, 150-µm pixel size). The X-ray beam is blocked after the sample by a beam stop to avoid the detector’s saturation. Increasing the sample-to-detector distance will increase the diffraction cones spreading and bring in focus the reflection rings at smaller values at the expense of the ones at larger values. For samples from this work, this distance was set to 250 mm; such configuration makes it possible to reach a scattering vector modulus down to Qmin = 0.2 Å-1 (Q = 2π/d = 4π/λ sin( ), where λ is the incident wavelength, and 2 is the scattering angle) corresponding to diffraction peaks range between 2 to 28 Å. The XRS patterns were acquired with a typical acquisition time of 900 s.
Fig. II.6. How particles organization is measured from thin lamella of the clayey sample. Featuring an illustration of the X-ray scattering with a 2D-detector (2D-XRS) instrumental setup and the measurement of the particle orientation by azimuthal integration (along ) of the scattered intensity.
Treatment of 2D-XRS patterns
2D-XRS patterns are recorded from the longitudinal and transversal lamellas of the duplicated samples (see section II.3.1). The longitudinal lamellas were cut perpendicular to the sedimentation plane to allow the anisotropy in the particle’s preferred orientation to appear in the ODF. By contrast, the transversal lamellas were cut parallel to the sedimentation plane to verify the transverse isotropy of the sample (i.e., adequately prepared transversal lamella results in isotropic ODF).
Fig. II.7. shows 2D-XRS patterns recorded from longitudinal and transverse lamellas of one anisotropic vermiculite sample. The measured Bragg peaks scattered intensities (I) are displayed in color scale from white to black with increasing intensity. The black rings correspond to the hkl Bragg peaks of the clay in the sample. These patterns are processed by the software IMAGE developed at LPS. This software allows for full integration over the whole diffractogram to produce a 1D pattern of the Bragg peaks, as shown in Fig.II.8a.
Table of contents :
Chapter I Clay minerals and clay porous clay media
I.1. Clay structure and classification
I.1.1. The definition of clay minerals
I.1.2. The unit cell of clay minerals
I.1.3. The structure of the clay sheet
I.1.4. Linkage and deformation of the tetrahedral and octahedral sheets
I.1.5. The classification of the clay minerals
I.1.6. The chemical formula of the clay minerals used in this work
I.1.7. External surface properties and structural charge in clay minerals
I.1.8. The Electrical Double Layer (EDL)
I.2. Clay porous media
I.2.1. Clay particles, aggregates, and porosity types
I.2.2. Definition and main properties of a clayey medium
I.2.3 Diffusion of water and solutes in a clayey medium
Chapter II Material and Methods
II.1.1. Santa Olalla vermiculite
II.1.2. Illite du Puy
II.1.3. KGa-2 kaolinite
II.2. Conditioning and preparation of the different size fractions
II.3. Building the clay porous media
II.3.1 Building clay media with variable anisotropy in particles organization
II.3.2. Induration and slicing of the clay porous media
II.4. Analyzing the preferred orientation of clay particles
II.4.1. The Orientation Distribution Function (ODF)
II.4.2. Two-dimensional X-ray Scattering (2D-XRS) measurements
II.4.3. Treatment of 2D-XRS patterns
II.4.4. Calculating the 〈P2〉 order parameter
II.5. Experimental techniques used to measure diffusion in the clay porous media
II.5.1. Through-Diffusion (TD) experiment
II.5.2. 1H NMR pulsed Gradient Spin Echo experiments
Chapter III Mineralogical and morphological characterization of different size fractions of Illite du Puy
III.1. Materials and Methods
III.1.1. Illite du Puy materials
III.1.2. Bulk mineral quantification using XRD refinement on randomly oriented powders
III.1.3. Quantification of clay mineralogy using profile modelling of XRD 00ℓ reflections
III.1.4. Microchemical analysis by electron microscopy
III.1.5. Low-pressure nitrogen and argon adsorption at 77 K
III.2. Results and discussion
III.2.1. Mineralogy comparison between V-IDP and E-IDP raw materials
III.2.2. Crystal-chemistry and geometrical characterization of sub-fractions from E-IDP sample
III.3. Concluding remarks and perspectives
III.6. Supplementary Data (S.D.)
Chapter IV Water diffusion in Na-vermiculite, illite and kaolinite
IV.1. Water diffusion in Na-saturated illite du Puy (Na-IDP)
IV.1.1. Particle organization measurements in IDP
IV.1.2. Experimental challenges and modification of the diffusion cell
IV.1.3. Results and discussion
IV.2. Water diffusion in Na-Santa Olalla (SO) vermiculite
IV.2.1. Article; Role of interlayer porosity and particle organization in the diffusion of water in swelling clays
IV.2.1. Supplementary data for “Role of interlayer porosity and particle organization in the diffusion of water in swelling clays”
IV.3. A comparison of HDO diffusion in Na-IDP, Na-vermiculite, and Na-kaolinite
IV.4. Summary of Chapter IV
Chapter V: Na+ and Cl- diffusion in Na-vermiculite and Na-illite
V.2 Summary of the TD experiments
V.3. Na+ and Cl- diffusion in Na-saturated Illite du Puy (Na-IDP)
V.4. Na+ and Cl- diffusion in Na-Santa Olalla (SO) vermiculite
V.5. Summary and conclusions
General conclusions and perspectives