GeoScienceWorld
Volume

Quantitative Mineral Analysis of Clays

By C. S. Calvert, B. L. Davis, J. E. Dodd, F. T. Dulong, J. W. Hosterman, L. R. Johnson, R. C. Jones, D. A. Palkowsky, D. R. Pevear, L. C. Poppe, R. C. Reynolds Jr. and Maynard Slaughter
Edited by D. R. Pevear and F. A. Mumpton

Abstract

Quantitative determination of the weight percentages of mineral phases in rocks and soils challenges theorist and analyst alike. Although precision – analytical reproducibility – is often described in great detail, few, if any, commercial laboratories or published methods say much about accuracy – the approach of the determined value to the "real" value. Results are often reported with four significant figures, but for most analyses, only one or two figures have real significance. Technologists and non-technologists, unfamiliar with the above problems, typically assume that quantitative phase analysis (QPA) is similar to elemental chemical analysis, not realizing that the variabililty of minerals is much greater than that of atoms. Accuracy may be of far more that academic interest. In court, the amount of kaolinite in soil adhering to a boot may be critical piece of forensic evidence in a murder case, and in an equity hearing, the effect of clay in a petroleum reservoir on predicted production may result in each 0.1% of smectite being worth millions of dollars. Many geoscientists have a rather casual attitude towards quantitative mineral analysis, perhaps because they realize how semiquantitative such analyses can be. X-ray powder diffraction (XRD) peaks are commonly normalized to give "quick and dirty" results, the accuracy of which is unknown. That these results fail to correlate with other data may be solely due to bad mineral analyses. This volume describes XRD methods including a complete treatment for QPA of oriented clay aggregates. Both theory and practice are detailed, and the section on preferred orientation is especially enlightening.

  1. Page 1
    Abstract

    “… but can you tell me how much is there?”

    Quantitative determination of the weight percentages of mineral phases in rocks and soils challenges theorist and analyst alike. Anyone who has sent carefully prepared mineral mixtures to several commercial laboratories understands the situation. For samples sent out by my company, “highly accurate” results (i.e., within 3% absolute or 10% relative) were reported for only 46–74% of the analyses; generally, clay minerals gave the worst results.

    Although precision— —analytical reproducibility— —is often described in great detail, few, if any, commercial laboratories or published methods say much about accuracy— —the approach of the determined value to the “real” value. Results are often reported with four significant figures, but for most analyses, only one or two figures have real significance. Technologists and non-technologists, unfamiliar with the above problems, typically assume that quantitative phase analysis (QPA) is similar to elemental chemical analysis, not realizing that the variabililty of min-erals is much greater than that of atoms. Accuracy may be of far more that academic interest. In court, the amount of kaolinite in soil adhering to a boot may be critical piece of forensic evidence in a murder case, and in an equity hearing, the effect of clay in a petroleum reservoir on predicted production may result in each 0.1% of smectite being worth millions of dollars.

    Many geoscientists have a rather casual attitude towards quantitative mineral analysis, perhaps because they realize how semiquantitative such analyses can be. X-ray powder diffraction (XRD) peaks are commonly

  2. Page 3
    Abstract

    Quantitative analysis by X-ray powder diffraction (XRD) pro-edures is difficult even for simple, well-characterized solid phases. The intrinsic chemical variability of minerals adds a measure of uncertainty to results, and sample-to-sample variations in preferred orientation, commonly encountered with micas and clays, represents still another source of error that is difficult to control or evaluate. Analysis of clays by XRD methods should probably be considered excellent if the results are accurate to about 10% of the amounts present, and, perhaps, 20% if the concentrations are less than 10%. These numbers are based on the assumption that optimum techniques have been used. The analyst often cannot use the best techniques because of small sample size, poor crystal morphology (which destroys the preferred orientation so necessary for the detection of some of the best reflections to use), interferences on key analytical peaks, and samples that do not remain aggregated during drying. No invariant methodology is possible for quantitative XRD analyses of clays; instead, the analysts must select the best techniques consistent with the characteristics presented by the sample.

    Acceptable results are obtained by attention to many details, and the purpose of this chapter is (1) to enumerate and discuss such details, (2) to evaluate the kinds of errors that occur under given circumstances, and (3) to point out the larger errors involved if these details are ignored. No final method is proposed, and no attempt is made to review and evaluate the many XRD methods that have been published, although individual aspects of

  3. Page 37
    Abstract

    A computer program for semiquantitative mineral analysis used by the U.S. Geological Survey was designed in 1972 by J. C. Hathaway and translated into Fortran language for the IBM-360–65 computer by R. W. Bowen, both of the Geological Survey. Since then, the program has been upgraded and adapted for other computer systems. The program is currently operational on an IBM-PC-XT or IBM-PC-At microcomputer or on any other equivalent microcomputer, using digital data from an automated Diano 8500 series X-ray powder diffraction (XRD) system. It can also be used on a Prime minicomputer, using digital data from an automated Philips 3501 XRD system.

    The central algorithm operates on the digitized data of an XRD pattern. The generalized calculations of the program: (1) smooth the digitized data with a running-weighted 5-count average, (2) determine and subtract the background, (3) determine the position of the peaks and their integrated intensities, (4) qualitatively identify the peaks on the basis of a library of standard minerals supplied by the user, and (5) estimate the relative weight percentages of as many as 12 minerals utilizing an over-determined system of linear equations, which are solved by means of the least squares technique.

    To create a library of standard minerals, the user must prepare and analyze samples of these minerals under the same conditions as those to be used for the unknown sample. The library should consist of XRD data for each mineral standard, the run parameters of the XRD unit, and two sets of test factors. The

  4. Page 51
    Abstract

    Until only about a decade ago, researchers working with soils, sediments, and geological specimens used X-ray powder diffraction (XRD) principally for qualitative determinations of the phases present in a sample. Only rough semi-quantitative estimates of mineral concentrations could be made from the strip charts produced by early diffractometers. With the availability of microcomputers, however, most new diffractometers use programmed step-scanning as a means of collecting intensity data over the 2θ range of interest; many older diffractometers have been converted to step scanners as well. Mineral identification is now routinely accomplished by a variety of search-match procedures utilizing digitized XRD data.

    Aside from mineral identification, the most powerful use of digitized XRD data is peak-profile refinement. Commonly, the clay mineralogist determines mineral content by visually inspecting the XRD pattern. Most of the minerals found in soils and sediments are encountered with sufficient frequency that a computerized search-match procedure is often unnecessary. What is not always clear from visually inspecting the XRD pattern is that subtle differences in the physical properties between samples, such as diffracting domain size, degree of crystallinity, degree of hydration, and crystal structure substitutions occur. The real advantage of curve fitting/peak decomposition, therefore, is the ability to determine quantitatively subtle peak parameter differences between samples collected across a transect, down a vertical section or a test well, or after specific chemical or physical treatments. Furthermore, for most materials, the determination of specific mineral properties is more important in understanding the chemistry and physics of a sample than simply

  5. Page 103
    Abstract

    The theoretical basis and experimental verification of quantitative multiphase X-ray powder diffraction (XRD) analysis using reference intensity ratios have been well established by Chung (1974), Hubbard et al. (1976), Davis (1980, 1981, 1984, 1988), and Davis and Johnson (1982). Because of recent advances in sample preparation (Calvert et al., 1983; Davis, 1984), preferred orientation can no longer be used to explain the inability to obtain quantitative mineral analyses by XRD. Quantitative phase analyses by XRD, however, are sometimes limited because the basic reference intensity constants themselves are not always known with sufficient accuracy and because adequate reference intensities are lacking for mineral components that have a range of elemental compositions.

    The quantitative multicomponent analysis procedure known as the reference-intensity method (RIM) is defined to include X-ray powder transmission (XRT), X-ray powder diffraction, and improved sampling techniques for the analysis of both crystalline and noncrystalline materials. The sample preparation procedure of RIM is critical and consists of (1) particle size reduction, (2) suspension of the sample as an aerosol, and (3) collection of sample particles onto a substrate consisting of a porous glass-fiber filter. Studies with highly inequant crystallites of MoO3 (Davis, 1984) have demonstrated that random orientation of particles during suspension can be achieved even with heavy mass loadings.

    Mass absorption coefficients measured by XRT allow for the correction of integrated intensities for thin-layer transparency and matrix effects and also permit the determination of noncrystalline components by comparison of the observed with the calculated mass absorption coefficients obtained from the

  6. Page 119
    Abstract

    The quantification of minerals and their compositions can help determine geochemical and geological processes in all types of rocks. Several methods of quantitative mineral analysis have been tried. Microscopic methods succeed on some coarse-grained rocks, but fail on fine-grained rocks. X-ray powder diffraction (XRD) and infrared (IR) methods can determine a few minerals with some accuracy, but these techniques generally fail for bulk rock analysis, especially clay mineral analysis. Serious spectral overlap, preferred orientation, and variable mineral compositions render most of these methods qualitative or, at best, semiquantitative, and even the newest automated X-ray diffractometers cannot produce quantitative mineralogical analyses for most rocks. Elaborate XRD methods, such as spectral deconvolution (see, among others, Slaughter, 1981; Jones, 1989), have promise in limited applications, but most of these methods are tedious and expensive. Special sample preparation techniques, such as the deposition of dust on filter membranes for XRD and IR, while useful for special purposes, do not generally work for whole-rock analysis.

    Clay minerals are especially difficult to quantify, because they commonly contain structural disorder and have a range of compositions. (This particular difficulty of analyzing clay minerals is illustrated by the work of van der Marel, 1966). In addition, individual members of clay mineral groups, such as those of the smectite group, cannot generally be differentiated qualitatively by XRD methods if they are present together in a rock.

    Although current methods individually do not provide sufficient information to quantify the mineral content of rocks, a combination of methods may provide enough

  7. Page 153
    Abstract

    Quantitative mineral analysis is necessary if highly accurate estimates of mineral types and proportions in a bulk rock are required to evaluate or predict the properties or behavior of that rock. In this chapter, “highly accurate” is defined as within 3% absolute or 10% relative (±10% of the amount of a given phase present) to the actual mineral proportions. This range of accuracy is not arbitrarily chosen, but represents what in the authors’ experience is “about the best that can be done” for typical natural rock samples containing many phases. This definition may not sound very good to those used to standard chemical analyses; indeed, it is really a semiquantitative analysis or, more precisely, an “accurate estimate”.

    Although identification of minerals can be straightforward, the estimation of mineral proportions is seldom simple because of natural variability in the properties of mineral phases. A mineral is by definition a “naturally formed chemical element or compound having a definite range of chemical composition and usually a characteristic crystal form (structure)” (Anonymous, 1976). Due to the variable composition and crystallinity of minerals, particularly clay minerals, no single property can accurately define the abundance of most of the common minerals in rocks.

    An accurate quantitative mineral analysis requires: (1) proper interpetation of both whole-rock and clay-fraction X-ray powder diffraction (XRD) data to identify all phases present, and (2) application of several measured sample properties (e.g., chemistry) to define mineral proportions. Although XRD by itself may provide rapid estimates of mineral proportions, variations in particle

  8. Page 167
    Abstract

    Many researchers still generate X-ray powder diffraction (XRD) data using strip chart recorders. Either templates or conversion tables are then used to convert these patterns from O2θ to interplanar spacings, i.e., d-values, thereby allowing the Joint Committee on Powder Diffraction Standards (JCPDS) XRD diffraction files to be used to determine which phases are present. The merits of templates over conversion tables are the time and labor they save. Although computerized X-ray diffraction systems that automatically convert O2θ to d-values are now commercially available, the cost of these systems ($25,000-$150,000) prevents many researchers from using them.

    A computer program has therefore been written to produce plotter-generated templates to convert O2θ to interplanar spacings (Figure 1). This program prompts the user for all necessary information and contains options to select the radiation wavelength, English or metric scales, various combinations of strip chart and goniometer speeds, and 29 range.

    The program utilizes the Bragg equation and was written in the C programming language. It operates under Unix system 5 or MS-DOS version 2.x. Although the program was written to operate with the CALCOMP 970 and 936 plotters, the plotting routines were written into the program in a modular fashion, thereby allowing the user to substitute routines for other X-Y plotters. This feature makes the program more versatile and, therefore, functional for a large number of users. The location of these routines within the software is given in the program documentation.

    Scales for O2θ and d-values on the conversion template are labeled to allow

Purchase Chapters

Recommended Reading