Bull. Kitakyushu Mus. Nat. Hist. Hum. Hist., Ser. A, :, March, Xray fluorescence analysis of major and trace elements in carbonate rocks using glass bead samples asushi MORI Kitakyushu Museum of Natural History and Human History, Higashida, ahatahigashiku, Kitakyushu, Japan (Received July, ; accepted October, ) ABSTRACT Easy and quick analytical techniques of Xray fluorescence spectrometry for carbonate rocks were investigated. The analyzed elements are SiO, TiO, Al O, total Fe O,,,, Na O, K O, P O,,,,,,,,,,,,,,, and. These elements were determined on a single glass bead made of a mixture of. g of sample powder without prior calcination, g of Li B O flux and about mg of LiI releasing agent. Synthetic standard samples for calibration of a spectrometer were prepared from pure chemical reagents and were fused to glass beads likewise. Although loss on ignition during the fusion resulted in deviation of the flux/sample ratio of these glass beads, reliable calibration lines were obtained by theoretical corrections for matrix effects, flux/sample ratio and spectral lineoverlap based on a fundamental parameter approach. Analytical results of GSJ geochemical reference samples were in good agreement with the recommended values and showed sufficient precision and sensitivity. INTROCTION In Xray fluorescence (XRF) analysis of geological materials, there are several methods of sample preparation such as fused glass beads and pressed powder pellets (e.g., HATTORI, ). The glass bead method is valid to eliminate mineralogical and grain size effects, so it is widely used for silicate rocks. For carbonate rocks, the powder pellets are more popular than the glass beads (KOCMAN, ). This is because highca and CO contents of carbonate rocks result in crystallization and bubbling of glass beads, respectively. GOTO et al. () have succeeded in avoiding these problems by calcination of sample powders and addition of pure SiO prior to fusing the glass beads. However, the calcined sample powders are hard to handle due to moisture absorption, and the complex procedure requires much time for sample preparation. This report attempts XRF analysis of carbonate rocks using glass beads made of sample powders only. By omitting calcination of sample powders and addition of pure SiO, the sample preparation will become considerably easy and quick. The analytical techniques described by MORI & MASHIMA () are improved for carbonate rocks. Correction for flux/sample ratio of the glass beads is newly introduced to eliminate the influence of loss on ignition (LOI). Accuracy, precision and sensitivity of this method will be reported. ANALTICAL TECHNIQUES Standard samples Seventeen synthetic standard samples were prepared from pure chemical reagents for calibration of an XRF spectrometer. The reagents used and their drying conditions, which follow those of MORI & MASHIMA (), are shown in Table. The reagents of Ca, Na and K were carbonates. The flux and releasing agent were Li B O (Merck Spectromelt A) and LiI, respectively. The releasing agent was dissolved into water to make a wt.% LiI solution. First, the reagents of major elements (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, P and ) were mixed. The mixtures of major elements, about g each, were individually homogenized for hours with an agate mortar, and then dried at room temperature for > hours in a desiccator. Next, the reagents of trace elements were diluted with the flux to obtain the ppm mixtures of (),, and, (), and, () and, (), and, and (),, and. The grouping of the trace elements was arranged for which no spectral interference of analysis lines occurs in each mixture. The mixtures of trace elements, about g each, were individually homogenized for hours with an agate mortar, and then dried at room temperature for > hours in a desiccator. Finally, the mixtures of major elements and the mixtures of trace elements were blended at appropriate ratios, which were decided so that the total amount of the reagents might be g in
asushi MORI Reagent Table. Reagents used for standard samples. Assay (%) Major elements SiO TiO. Al O Fe O... CaCO. Na CO. K CO. KH PO Trace elements O. NH O. O.. O. SO CO. O. O O. CO. O. O. O. O. Flux Li B O (Spectromelt A) Releasing agent LiI. Manufacture S M S Abbreviations: = ako Pure Chemical, S = Strem Chemicals, M = Merck, RT = room temperature. M Drying conditions (;) (hours) RT RT RT.. >.. > > (min) (min) each blend. The flux/sample ratio was adjusted to : by adding the flux to the blend. Chemical compositions of the standard samples are shown in Table. The standard samples were fused to glass beads with a Tokyo Kagaku TK automated high frequency bead sampler. After adding about mg of the LiI solution (corresponding to about mg of anhydrous LiI), each standard sample was homogenized with an agate mortar. The mixture was put into an AuPt alloy crucible combined with a mold of glass beads. The mixture was heated to C spending about minutes and was totally melted holding the temperature for the following minutes. The molten mixture was agitated by spinning and tilting for the following minutes under the same temperature and was cast into a glass bead after being cooled in the crucible. The fusion made the glass bead lighter than the mixture of sample and flux. LOI was almost equal to the total amount of CO and H O in each standard sample, suggesting that these volatile components have been almost entirely lost from the glass bead. Dispersion of the mixture did not occur during fusion of all the samples. This is probably because relatively slow heating in the fusion procedures prevented abrupt bubbling of the mixture. In some cases, tiny gas bubbles remained within the glass beads. The gas bubbles were removed by fusing the glass beads again. Sample preparation The samples used in this report were GSJ geochemical reference samples provided by the National Institute of Advanced Industrial ience and Technology: JLs limestone, JDo dolomite, JCp coral and JCt fossil shell. The GSJ samples of JA andesite, JB and JB basalts, JGa and JG granodiorites, JGb gabbro, JH hornblendite and JR rhyolite were mixed with a pure CaCO reagent (ako Pure Chemical, assay >. %) and were used for examination of analytical precision and sensitivity. The mixing ratio was :. First, the sample powders and flux were dried at C for hours in a drying oven. Then,. g of each sample powder and g of the flux were mixed after cooled in a desiccator. About mg of the LiI solution was added to the mixture. The mixture was homogenized with an agate mortar and was put into the AuPt alloy crucible. Finally, the mixture was fused to a glass bead by the same procedure as the standard samples. Instrumental conditions The instrument used was a PANalytical MagiX PRO wavelength dispersive XRF spectrometer at Kitakyushu Museum of Natural History and Human History. The XRF spectrometer was equipped with a Philips k Rh anode Super Sharp Xray Tube (P/) and a Philips P/ RC sample changer. The operating conditions are shown in Table. The calibration and analysis were automatically performed with the help of a SuperQ software package, which was attached to the XRF spectrometer. The instrumental drift was corrected by measuring synthetic monitor samples (see MORI & MASHIMA, ). Calibration lines In order to set up calibration lines, theoretical matrix corrections based on a fundamental parameter approach were applied to all the analyzed elements. The matrix correction model used was that of DE JONGH (). As described above, the flux/sample ratio of the standard glass beads was
XRF analysis of carbonate rocks Table. Calculated chemical compositions of standard samples. (wt.%) SiO TiO AlO FeO NaO KO PO LOI (ppm) LOI = loss on ignition calculated from compositions of reagents. N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N........... N...........
asushi MORI Element SiO TiO Al O Fe O * Na O K O P O Line Si Kα Ti Kα Al Kα Fe Kα Mn Kα Mg Kα Ca Kα Na Kα K Kα P Kα Kα Kα Kα Kα Kα Kα Kα Kα Kα Kα Kα Lα Lα Lα Lα Table. Instrumental operating conditions of XRF spectrometer. Xray tube Angle ( Counting time Collimator Analyzing θ) PHD Detector (sec) (µm) crystal (k) (ma) Peak BG BG Peak BG LL PE LiF PE LiF LiF PX LiF PX LiF Ge LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF LiF FPC FPC FPC FPC FPC FPC FPC FPC FPC......................................................................... : : : : : : : : : : : : : : : : : : : : : : : : : Lβ LiF... : Common settings: anode = Rh, filter = none, chamber = vacuum, sample spinner = on, PR gas flow rate =. l/hr. Abbreviations: BG = background, PHD = pulse height distribution, LL = lower limit, UL = upper limit, FPC = gas flow proportional detector, = scintillation detector, = duplex detector (FPC + Xesealed detectors in tandem). *Total iron as Fe O. UL adjusted to :. However, it decreased due to LOI during fusion. Correction for the flux/sample ratio (KATAOKA et al., ; SHOJI et al., ) was employed to eliminate the influence of LOI. In addition, spectral lineoverlap corrections were introduced into the matrix model. The equation of the calibration line for element i is written as: i = (D i + E i I i )(+Σα ij j + R j )+ΣL ij j R j where i is concentration, I i is fluorescent Xray intensity, α ij s are matrix correction coefficients, R is standard flux/sample ratio (i.e., in this report), R is a deviation from the standard flux/sample ratio calculated from LOI content, L ij s are lineoverlap correction coefficients, and D i and E i are intercept and slope of the calibration line, respectively. The values of α ij s, L ij s, D i and E i were calculated with the SuperQ software package on the basis of fluorescent Xray intensities measured on the standard glass beads. LOI was regarded as a base element in the calculations. RESULTS AND DIUSSION Accuracy of calibration lines The calculated coefficients of the calibration lines are summarized in Tables, and. Accuracies and Kfactors, which indicate reliability of the calibration lines, were calculated by the following equations: Accuracy = Kfactor = Σ(C chem C calc ) n k (C Σ chem C calc ) n k C chem C
XRF analysis of carbonate rocks α i,tio......................... α i,al O.......................... α i,sio SiO TiO Al O Fe O Na O K O P O........................... se element: loss on ignition. SiO TiO Al O Fe O Na O K O P O α i,.......................... α i,.......................... α i,.......................... α i,fe O.......................... α i,.......................... Table. Matrix correction coefficients. α i,.......................... α i,.......................... α i,.......................... α i,.......................... Table. Extended. α i,.......................... α i,.......................... α i,na O.......................... α i,.......................... α i,k O.......................... α i,.......................... α i,p O.......................... α i,.......................... α i,.......................... α i,.......................... α i,.......................... α i,.......................... α i,.......................... α i,..........................
asushi MORI TiO Fe O Na O P O TiO Fe O Na O P O L i,sio.. L i,. L i,tio.. L i,. where n is the number of samples, k is the number of correction coefficients, C chem is an original concentration, C calc is a calculated concentration, and C is a weighting constant. They are listed in Table. The corrections for matrix effects, flux/sample ratios and spectral lineoverlaps were successfully achieved in both the major and trace elements. For example, accuracy of the calibration lines was improved from. wt.% to. wt.% for Fe O and from ppm to ppm for (Fig., Table ). The Table. Line overlap correction coefficients. L i,al O L i, L i,..... Table. Extended. L i,.... L i,... L i,. L i,. L i,.... L i,na O. L i,. L i,. L i,.. calibration line of also came to pass the vicinity of the origin. As shown in Table, the accuracy is. wt.% for,. wt.% for SiO and <. wt.% for other major elements. That is ppm for, ppm for and < ppm for other trace elements. The Kfactors indicating reliability of the calibration lines are below. and. for major and trace elements, respectively (Table ). These findings suggest that the calibration lines have sufficient quality for quantitative analysis.
XRF analysis of carbonate rocks Table. Calculated results of calibration lines. Intercept D i Slope E i Accuracy Kfactor SiO TiO Al O Fe O * Na O K O P O..................................................... (wt.%)......... (ppm).......................... *Total iron as Fe O. Precision and sensitivity Analytical results of the GSJ samples are shown in Table. These values are average compositions of threetimes measurements. The relative standard deviations are σ of the measurements. LOI was calculated from a sum of H O and CO contents of the recommended values in IMAI et al. (,, ) and OKAI et al. (). The analytical results are in good agreement with the recommended values. For the major elements of >. wt.%, the differences between the analytical results and the recommended values are typically below %. The relative standard deviations of the analyzed values are below. %. For the trace elements of > ppm, the differences between the analytical results and the recommended values are typically below %. The content of the JB + CaCO sample is considerably lower than the recommended value. The reason of this difference is not clear., and indicate somewhat greater difference (below %) with respect to some lowconcentration samples. Figure shows relationships between the relative standard deviation and the concentration of the trace elements. As indicated by the regression lines (y=ax b ), the standard Fig.. Calibration lines for Fe O and. The open and solid circles show the data before and after the corrections for matrix effects, flux/sample ratio and spectral lineoverlap, respectively. deviations increase towards zero concentration. They reach % at ppm for,,,, and, at ppm for,, and, at ppm for,,, and, and at ppm for. The precision and sensitivity of this method are comparable to those of MORI & MASHIMA () and are sufficient to petrological studies. SUMMAR Major and trace element compositions of carbonate rocks were analyzed by XRF spectrometry using glass bead samples. The glass beads with flux/sample ratio of : were made of sample powders. In the fusion procedures, abrupt bubbling and dispersion of the glass beads were prevented by relatively slow heating. ystallization of the glass beads did not occur even in the pure CaCO sample. By omitting calcination procedure of sample powders and addition of pure SiO, the sample preparation became easy and quick compared with that in earlier literatures.
asushi MORI (wt.%) SiO TiO AlO FeO* NaO KO PO LOI Total (ppm) R............ Table. Analytical results of GSJ geochemical reference samples. JLs JDo JCp JCt JA + CaCO JB + CaCO XRF RSTD R XRF RSTD R XRF RSTD R XRF RSTD Calc XRF RSTD Calc XRF RSTD.................................................. n.d............................................. R = recommended values after IMAI et al. () and OKAI et al. (), Calc = calculated compositions based on recommended values after IMAI et al. () and IMAI et al. (), XRF = analytical results, RSTD = relative standard deviations (σ in %), = not detected. *Total iron as FeO. Loss on ignition calculated from a sum of HO and CO contents of the recommenden values.....................................................................................................................................................
(wt.%) SiO TiO AlO FeO* NaO KO PO LOI Total (ppm) JB + CaCO JGa + CaCO JG + CaCO JGb + CaCO JH + CaCO JR + CaCO XRF analysis of carbonate rocks Table. Extended. Calc............ XRF............ RSTD....................... Calc............ XRF............ RSTD......................... Calc............ XRF............ RSTD.......................... Calc............ XRF............ RSTD........................ Calc............ XRF............ RSTD......................... Calc............ XRF............ RSTD........................
asushi MORI Fig.. Relationships between relative standard deviations (RSTD) and concentrations of trace elements. The curves and equations indicate the regression lines (y = ax b ).
XRF analysis of carbonate rocks Fig. Extended.
asushi MORI The calibration lines were set up by theoretical corrections for matrix effects, flux/sample ratio and spectral lineoverlap based on a fundamental parameter approach. These corrections were successfully achieved and yielded reliable calibration lines with wide optimal ranges of composition. The optimal ranges are wt.% of, wt.% of, wt.% of total Fe O, ppm of and ppm of (see Table ). Thus the calibration lines cover various carbonate and related rocks such as limestone, dolostone and skarn. This method is, however, inappropriate for carbonatite containing > ppm of, and. Analytical results of GSJ geochemical reference samples were in good agreement with the recommended values and showed sufficient precision and sensitivity for petrological studies. ACKNOLEDGMENTS I am grateful to Drs. S. KAKUBUCHI and T. NISHIAMA for their critical reading of the manuscript. I thank Drs. T. MORISHITA, T. MIAMOTO and T. ANAGI who gave me GSJ geochemical reference samples (JDo, JCp, JCt and JB). This study was partly supported by GrantinAid for oung ientists B (No. ) from the Ministry of Education, ience, Sports and lture, Japanese Government. REFERENCES DE JONGH,. K.,. Xray fluorescence analysis applying theoretical matrix corrections Stainless steel. XRay Spectrometry, :. GOTO, A., T. HORIE, T. OHBA, & H. FUJIMAKI,. XRF analysis of major and trace elements for wide compositional ranges from silicate rocks to carbonate rocks using low dilution glass beads. Japanese Magazine of Mineralogical and Petrological iences, : (in Japanese with English abstract). HATTORI, H.,. Preparation of glass disc sample of rock for light element analysis by Xray spectrometry. Bulletin of the Geological Survey of Japan, : (in Japanese with English abstract). IMAI, N., S., TERASHIMA, S. ITOH & A. ANDO,. compilation of analytical data for minor and trace elements in seventeen GSJ geochemical reference samples, igneous rock series. Geostandards Newsletter, :. IMAI, N., S., TERASHIMA, S. ITOH & A. ANDO,. Compilation of Analytical Data on ne GSJ Geochemical Reference Samples, Sedimentary Rock Series. Geostandards Newsletter, :. IMAI, N., S. TERASHIMA, S. ITOH & A. ANDO,. compilation of analytical data for five GSJ geochemical reference samples: The instrumental analysis series. Geostandards Newsletter, :. KATAOKA,., S. SHOJI & H. KOHNO,. Correction for loss on ignition (LOI), gain on ignition (GOI) and dilution ratio of fusion beads in Xray fluorescence spectrometry (). Advances in X ray Chemical Analysis Japan, : (in Japanese with English abstract). KOCMAN,.,. Xray fluorescence analysis of gypsum, anhydrite and carbonate rocks. In: S. T. AHMEDALI (ed.), Xray fluorescence analysis in the geological sciences, advances in methodology, pp.. Geological Association of Canada. MORI,. & H. MASHIMA,. Xray fluorescence analysis of major and trace elements in silicate rocks using : dilution glass beads. Bulletin of the Kitakyushu Museum of Natural History and Human History, Ser. A, ():. OKAI, T., A. SUZUKI, S. TERASHIMA, M. INOUE, M. NOHARA, H. KAAHATA & N. IMAI,. Collaborative analysis of GSJ/AIST geochemical reference materials JCp (Coral) and JCt (Giant Clam). Chikyukagaku (Geochemistry), : (in Japanese with English abstract). SHOJI, S., K. AMADA, E. FURUSAA, H. KOHNO & M. MURATA,. Correction for loss on ignition (LOI), gain on ignition (GOI) and dilution ratio of fusion beads in Xray fluorescence spectrometry (). Advances in Xray Chemical Analysis Japan, : (in Japanese with English abstract).