Pedogenic smectite from a young (Holocene) tropical soil was reacted in Al-rich solution at 150ºC for a range of reaction times (3 to 120 days) in orderto study mechanisms and rates associated with the transformation of smectite to kaolinite via interstratified kaolinite-smectite (K-S). As has been observed in tropical soils, the overall reaction rate is logarithmic, with rapid initial transformation of smectite to K-S with ~50% smectite layers, followed by progressively slower transformation of intermediate K-S to kaolinite-rich K-S and eventually Fe-kaolinite. Sub-micron hexagonal non-Fe-bearing kaolinite forms in the final stage (after 120 days) as a minor mineral in an assemblage dominated by Fe-kaolinite. The pedogenic smectite used as starting material consisted of two end-members, Fe-beidellite and Al-smectite, enabling comparison of reaction pathways. Fe-beidellite transforms to K-S or Fe-kaolinite within 3 days, whereas Al-smectite transforms much more slowly, appearing to reach a maximum rate in intermediate stages. This difference is probably due to hydrolysis of relatively weak Mg-O and Fe-O bonds (relative to Al-O bonds) in Fe-beidellite octahedral sheets, which drives rapid reaction, whereas the driving force behind transformation of Al-smectite is more likely to be related to stripping of tetrahedral sheets which reaches its maximum rate at intermediate stages. Multiple analytical approaches have indicated that Al is rapidly fixed from solution into smectite interlayers within K-S, and that K-S and Fe-kaolinite inherit octahedral Fe and Mg from precursor smectite; as the reaction progresses, octahedral sheets become progressively more Al-rich and Fe and Mg are lost to solution. These results demonstrate that: (1) early-formed pedogenic smectite in tropical soils is expected to transform to kaolinite via interstratified K-S; (2) K-S has a strong potential to sequester plant-toxic Al in tropical soil; and (3) the presence in tropical soils of Fe-kaolinites with relatively large cation exchange capacities may be related to inheritance of octahedral sheets from precursor smectite and K-S.

The mechanism for the kaolinization of smectite is extremely complex. The purpose of this study was to explore this mechanism by providing more microscopic information about kaolinite-smectite (K-S) intermediate phases. Crystal-chemical changes were investigated and integrated in a model of the transformation mechanism. Eight K-S samples from three localities, derived from volcanic ash beds, were studied using transmission and analytical electron microscopy (TEM, AEM) and high-resolution TEM (HRTEM). The study completes a previous investigation, using several analytical techniques. The samples cover the range of K-S composition available from the previously studied sample set. Analysis by TEM indicated the preservation of particle morphology throughout the process. Most K-S particles had anhedral, smectite-like morphology, and only the most kaolinitic specimen revealed the coexistence of anhedral and euhedral, hexagonal particles. Analytical electron microscopy showed large chemical variations within samples, corresponding to various degrees of smectite kaolinization. Comparison of chemical results (Si/Al) and d060 values (proxy for octahedral composition) with the extent of kaolinization from thermogravimetry (TG) indicates that chemical changes in the octahedral sheet occur mainly when the proportion of kaolinite is 40–70%. The results above are consistent with kaolinization occurring via layer-by-layer transformation through the progressive loss of individual tetrahedral sheets in smectite layers and subsequent chemical changes in the octahedral sheet. Such a mechanism would produce the results observed in this study: (1) most particles preserve their original morphology; (2) significant variation in terms of the extent of transformation of particles within samples, and (3) formation of crystal structures intermediate between those of smectite and kaolinite, with parts of the tetrahedral sheets missing (kaolinite-like patches). Such structures become least stable at kaolinite ∼50%, when the perimeter of the kaolinite-like patches is largest and chemical changes in the octahedral sheet can occur more easily. Kaolinite layers could not be resolved by HRTEM in most cases and showed lattice fringes corresponding to superstructures. A model was established to quantify kaolinite and smectite layers in the HRTEM images with results which matched TG-derived values.

Twenty two samples were studied to investigate the nature and evolution mechanism of mixed-layer kaolinite-smectite (K-S). We examined the <2 µm or <0.2 µm fraction of K-S formed by hydrothermal and hypergenic alteration of volcanic material. The samples are from three localities: 20 specimens from a Tortonian clay deposit in Almería, Spain; one specimen from weathered Eocene volcanic ash from the Yucatan Peninsula, Mexico; and one sample from a weathered Jurassic bentonite from Northamptonshire, England. The samples were studied using chemical analysis, X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR). The XRD patterns of the oriented, glycolated mounts were modeled using NEWMOD and the proportion of smectite and kaolinite layers was determined, ranging between 0 and 80% kaolinite. The analysis of the OH-stretching region of the FTIR spectra at different temperatures (180–550°C) showed the progressive dehydroxylation of kaolinite domains and, perhaps, of smectite domains, but no detailed information could be obtained about the sequential OH loss in different cation environments. The abundance and short-range ordering of the octahedral cations were studied using the OH-bending bands. The chemical and FTIR-estimated octahedral cation abundances were broadly similar. Aluminum showed a tendency to mix with Fe and Mg rather than to form AlAl pairs. Al-for-Mg substitution accompanying kaolinization was evident from the increase in AlAl pairs and decrease in AlMg pairs. Iron is retained in the structure. No other octahedral cation rearrangement was observed. The intensity of the 750 cm−1 band, assigned to translational vibrations of external OH groups in a kaolinitic environment, was quantified and modeled in relation to kaolinite layer proportion. The chemical data show that there are residual interlayer cations in kaolinite domains, which, in accordance with the model mentioned above, disturb external OH-translation vibrations. These results indicate the persistence of certain chemical and structural smectite features in kaolinite domains and thus support a smectite kaolinization process via a solid-state transformation. This confirms previous XRD, thermal, chemical and NMR analyses of the same sample set.

Kaolinite-smectite mixed-layers have been found to have a complex structure with smectite and kaolinite domains within layers. Here we further investigate this structure in samples with 0–80% kaolinite layers, as determined by X-ray diffraction, by means of magic angle spinning nuclear magnetic resonance (MAS NMR) of 29Si and 6Li. The 29Si NMR experiments were carried out on two samples (55 and 80% kaolinite layers), before and after their exchange with Mn2+, a paramagnetic ion that causes NMR signal loss from neighbouring nuclei, in order to investigate the distance between Mn ions and Si atoms in kaolinite sites. The 29Si NMR intensity from such sites (at ~–91 ppm) was reduced upon Mn exchange, indicating that some Mn ions are located near kaolinite Si sites. The position of the 6Li peak changes slightly (–1.3 to –1.8 ppm) but progressively with increasing kaolinite content (0–80% kaolinite layers) of four K-S specimens, suggesting two slightly different chemical environments for interlayer Li, one related to smectite and the other to kaolinite. The two sets of experiments are consistent with a complex structure of kaolinite-smectite, including smectite and kaolinite domains within layers and/or interlayers of varying smectitic and kaolinitic character.