INTERPRETATION OF X-RAY DIFFRACTION PROFILES OF CLAY MINERALS FROM BOTTOM SEDIMENTS OF LAKE BAIKAL

E.P. Solotchina, T.E.Gorelik, V.M. Gavshin, and G.N. Anoshin

United Institute of Geology, Geophysics and Mineralogy, Siberian Branch RAS, Novosibirsk, Russia

Keywords: X-ray powder diffraction, clay minerals, structural modeling, bottom sediments, Lake Baikal.

Mineralogical and crystal chemical investigations of clay component taken from Lake Baikal Cenozoic sediments by deep drilling are directed at solving major scientific problem as the research on changes of paleoclimate dynamic. As evidenced by the Earth history, physical weathering was dominated and clay facies were precipitated during cooling periods, and relatively warming periods were characterized by enhanced chemical weathering and precipitation of diatomic facies . Detail investigations of a composition and structure of clay minerals making up sediments of both types allow to single out time intervals according to different paleoclimatic episodes.

Polymineral clay fractions are composed of fine-dispersed, poor crystallized illite, chlorite, kaolinite, smectite and mixed-layer illite/smectites, The diffraction pattern is complicated by the presence of non-clay minerals: quartz. feldspar and etc.

The method of full profile analysis [1] is practically the only way permitting to interpret complicated diffraction spectra of multi-component clay mineral samples. However, a search for the optimal solution takes a lot of time because of the strong cross-correlation between parameters of predicted structural models.

There has been developed the algorithm in which: 1) calculation of diffraction pattern of structural models of mixed-layer minerals is performed by well-known R.C.Reynold's program supplemented with the interactive service; 2) diffraction peaks of associated minerals, in particular, quartz, are simulated by Pearson's functions; 3) summing up calculated spectra of proposed phases is made with various weight factor; 4) the final stage is realized by sequential iterative searching parameters till the best fitting of the calculated curve with experimental one.

Possible variants of relations between parameters characterizing the composition, concentration of different types of layers and their stacking sequence are estimated by the analysis of the multidimensional distribution of correlation coefficients calculated for theoretical and experimental diffraction profiles in specific angular intervals. These parameters provide a very complex influence on the kind of a theoretical curve, so the correlation function has of great number of local extrema (especially for polymineral samples), and the direct use of methods of mathematical programming for the search for correlation function maxima is difficult. Even universal optimization algorithms by Nelder-Mead [2] intended for discontinuous functions give the uncertain convergence. By this reason the preliminary analysis of properties of multivariate distribution of the correlation function is required. Using sliding windows with the several different radii for the definition of profile correlation allows to construct the convergent algorithm for searching parameters in multidimensional space for nonunimodal optimization problem.

The performed investigations allow to interpret poorly expressed diffraction spectra of sediments described. It has been established the presence of several mixed-layer minerals with various concentration of illite and smectite layers of the different order degree and dispersity. These parameters along with variable relations of clay minerals throughout a section and their composition may serve as the indicator of paleoclimate changes recorded in bottom sediments of Lake Baikal.

The work is supported by RFFI Foundation grant No 96-05-65943 and No 97-05-96370.

1. Drits V.A., Sakharov B.A. (1976) X-ray Analysis of Mixed-layer Minerals. Nauka, Moscow, 256 pp (in Russian)
2. David M. Himmelblau (1972) Applied Nonlinear Programming. McGraw-Hill Book Company