Nature of Structural Disorder in Natural Kaolinites: a New Model Based on Computer Simulation of Powder Diffraction Data and Electrostatic Energy Calculation

A new model for the description of the structural disorder in natural kaolinite materials is proposed, based on the stacking of two 1:1 layers and their enantiomorphs, and encompassing previously proposed models. The layers, where randomly stacked along the c axis (using probabflistic functions nested in recursive algorithms), correctly describe the observed powder diffraction patterns of natural kaolinites with any density of structural faults. The proposed model was evaluated using electrostatic energy calculations against earlier models of disorder based on layer shift, layer rotation, statistical occupancy of the A1 octahedra, or enantiomorphic layers. The present 4-layer model has a minimum of potential energy with respect to the previous models. As expected, the fully ordered triclinic structure of kaolinite possesses the absolute minimum of potential energy. Key Words--Energy calculation, Kaolinite, Layer disorder, XRPD simulation. I N T R O D U C T I O N At present (a) there is no general model to describe the structure disorder in kaolinite, (b) the widely used empirical disorder indices, such as the Hinckley index, bear little relationship to the structural description of stacking disorder, and (c) proposed models of stacking disorder have not been compared with electrostatic energy calculations. These latter calculations were limited to the understanding of the hydroxyl crystal chemistry and bonding in the fully ordered structure (Giese 1982, Collins and Catlow, 1991, Bleam 1993). We present here the results of computer simulations of X-ray diffraction powder patterns of kaolinites, using a general recursive algorithm applied to a new structural model of the stacking disorder. The simulations are compared to experimental powder diffraction patterns of natural kaolinites, and are shown to describe correctly the diffraction profiles ofkaolinites with widely different fault densities. The technique is based on a rigorous recursion model, and it is of simple and general application. The proposed structure model is supported by the results of energy calculations. Kaolinite, A12Si2Os(OH)4, is a dioctahedral 1:1 layer aluminosilicate (Bailey 1980, Giese 1991), with one sheet composed oftetrahedrally coordinated Si and one sheet of octahedrally coordinated A1, with only 2/3 of the octahedra being occupied in an ordered or a statistical way. The 1:1 layer has a triclinic distortion, due to the atom relaxation around the vacant a luminum site (usually denoted as the B-site). The non-hydrogen atom positions obey lattice C-centering (Adams 1983, Suitch and Young 1983, Thompson and Withers 1987, Copyright 9 1995, The Clay Minerals Society Young and Hewat 1988, Bish and Von Dreele 1989, Collins and Catlow 1991, Bish 1993) with space group C 1 for the unit triclinic layer. Atomic positions from Bish and Von Dreele (1989) were used, yielding a simulated X-ray powder pattern in perfect agreement with the most ordered known natural kaolinite, the sample from Keokuk, Iowa (Bish and Von Dreele 1989, Giese 1991, Bish 1993). EXPERIMENTAL PROCEDURES X-ray powder pattern: experimental profiles Two natural kaolinite samples were selected because of their purity and because they represent well-ordered and poorly-ordered materials. The "ordered" kaolinite (KAOSAR) is from a 6 cm thick vein of hydrothermally recrystallized pure kaolinite from Casa Locchera, Nuraghe Mandras sas Ebbas, Sardinia, Italy (Coulon 1971). It is purer and more "ordered" than Clay Minerals Society Standard KGa1 (Van Olphen and Fripiat 1979). The "poorly-ordered" specimen is the Clay Minerals Society Standard KGa-2 from Warren County, Georgia (Van Olphen and Fripiat 1979). It contains minor amounts of anatase. The two powder specimens were analyzed for grainsize distribution using laser diffraction (Fritsch Particle Size Analysette 22). Both samples have a bimodal distribution, with grain-size peak maxima at 5 and 15 ~m. The two powder samples were then side-loaded in an a luminum holder, and the X-ray powder diffraction data (Figure 1) were collected on a Philips goniometer using graphite monochromatized CuKa radiation, a Bragg-Brentano parafocussing geometry, a step/scan