APPLICATION OF TOTAL-REFLECTION X-RAY FLUORESCENCE SPECTROMETRY (TXRF) TO GEOLOGICAL OBJECTS: EXPERIENCE OF THE TXRF LABORATORY, CENTRE FOR GEODYNAMICS AND GEOCHRONOLOGY

. Unlike conventional X-ray fluorescence spectrometry, the total-reflection X-ray fluorescence spectrometry is not a widespread and routine method for analyzing solid samples with mineral matrix, but it has a great potential for geochemical, geological, and archaeological studies. Rapid multi-elemental analysis of very small sample amounts can be performed by the internal standard method which does not require the matrix-matched reference materials. This is an undoubted advantage of the TXRF method over the conventional X-ray fluorescence method, especially if there is a limited available sample amount and a lack of well-characterized reference materials. This paper presents our experience with the application of TXRF spectrometry in the elemental analysis of apatite, ceramics, sediments, ores, and nodules. Special attention has been paid to the sample preparation procedure because it is one of the main sources of errors in the analysis. Preparing thin homogeneous specimen from the solid sample with a complex mineral matrix is not easy. Sample preparation strategy should be chosen considering the features of an analytical object, the content of the elements to be determined, and the accuracy required for a reliable interpretation. Consideration is being given to the examples of the preparation of a suspension for rapid analysis of ores and sediments, and to the original techniques of chemical decomposition for apatite and ceramics.


INTRODUCTION
Total-reflection X-ray spectrometry (TXRF) is currently one of the few analytical methods providing a rapid and cost-efficient multi-elemental analysis of various liquid and solid samples. TXRF is a variation of the energy dispersive X-ray fluorescence (XRF). It was first described in 1971 by Yoneda and Horiuchi [Yoneda, Horiuchi, 1971], who used the total reflection of X-rays discovered by Compton in 1923[Compton, 1923 to analyze small amounts of sample on optically flat surfaces. The main parts of an energy dispersive spectrometer are an X-ray tube, a monochromator, and a detector. Fig. 1 presents the principle scheme of TXRF. The essential difference between TXRF and conventional XRF is in the special geometry of the X-ray tube and detector. Unlike XRF, where the primary X-ray beam is directed at 45°, the X-ray beam in the TXRF spectrometer hits the sample on the carrier at a very small angle and is almost totally reflected therefrom. X-ray fluorescence radiation is detected by a silicon-lithium or silicon drift detector. The detector is placed at a very small distance directly above the sample to record the characteristic radiation. Only a small amount of sample is allowed to avoid limiting the total reflection and blocking the detector [Klockenkämper, von Bohlen, 2015].
-A variety of sample preparation procedures (minimal sample pretreatment by direct analysis, decomposition of matrix, preconcentration or extraction); -Quantification by the internal standard method (no external calibration is required); -Fast screening (semi-quantitative and quantitative multi-elemental analysis within 200-1000 s); -Low operational cost (no cooling media and gas consumption).
Currently, different commercial TXRF spectrometers are available on the market and there are many examples of SDD Sample on the carrier X-ray tube 0.1°9 0°F luorescence applying this method to different objects, especially liquids and biological materials. However, unlike conventional XRF or atomic absorption or emission spectral methods, TXRF is not a popular method for the analysis of geological materials, especially in Russia. Due to the above-mentioned advantages, TXRF has a great potential in geochemical, geological, and archaeological studies but special sample preparation procedures should be performed.

APPLICATION OF TXRF TO GEOLOGICAL OBJECTS
General procedures for preparing solid samples are presented in Fig. 2. Fast sample preparation techniques are called "direct" analysis including the semi-quantitative technique of "deposition of particles" and suspension preparation: the latter is the most popular among TXRF users. Mineralization techniques are more time-and reagent-consuming but remain more reliable in terms of the accuracy of measurement. The disadvantage of the mineralization technique is the probability of inaccurate determination of the volatile elements S, Cl, Br, As, due to full or partial evaporation losses that's why in this case the direct analysis is preferable. Extraction procedures are used when the sensitivity of TXRF is restricted for some of the elements (rare-earth elements REE, for example) or there are spectral interference issues caused by a high intensity of matrix elements. The TXRF methods developed by the researchers of the TXRF laboratory (Centre for Geodynamics and Geochronology) for different geological samples are presented in Table 1.
Two TXRF methods for the multi-element analysis of apatite were developed and validated, one is for large crystals (size of 1-2 mm) and another for microcrystals (size of 50-100 µm) [Maltsev et al., 2020[Maltsev et al., , 2021b. The acid digestion strategy was preferable for both of methods with lower detection limits and better precision values as compared to suspension preparation. Apatite elemental content was determined by TXRF in the provenance study of Slyudyanka apatite and for a fission-track study. As is known, some apatite samples are rich in REE and As, Sr, Th, U, etc., and it was shown that TXRF is a well-suitable method for the determination of these elements. Moreover, the advantage of TXRF compared to the routine laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) is a reliable quantitative analysis. The methods were validated through combined procedures: analysis of Durango and Otter Lake well-known apatite samples; using inductively coupled plasma mass spectrometry (ICP-MS) and wavelength-dispersive X-ray fluorescence spectrometry (WDXRF) as reference analytical methods; assessment of the measurement uncertainty.
TXRF can be applied for direct analysis of iron, manganese, ferromanganese, nickel-copper sulfide ores, oceanic and lacustrine nodules Akhmetzhanov et al., 2021;Chubarov et al., 2021]. The suspension preparation (5 mL of 1 % Triton X-100 solution and 20 mg of fine powder with a size <10 μm and internal standard of Ga) gives a possibility for fast reliable determination of major and minor elements by the internal standard method with the recovery values 75-125 %. The determination of trace elements in an ore matrix without preconcentration procedures can only be considered as semi-quantitative. It was shown that the reproducibility of direct TXRF analysis strongly depends on the particle size in the analyzed powder. The poor reproducibility values (>15 %) were observed for samples with an average particle size of more than 10 μm, so that a thorough additional grinding of powder before the suspension preparation is required. Dry mechanical grinding allows obtaining the powder with   [Pashkova et al., 2018a] Note. CRM -certified reference material, LOD -limit of detection, ICP-MS -inductively coupled plasma mass spectrometry, WDXRF -wavelength-dispersive X-ray fluorescence spectrometry, nd -not determined (the concentrations of the analytes considered were significantly higher than LOD). Примечание. CRM -аттестованный эталонный материал, LOD -предел обнаружения, ICP-MS -масс-спектрометрия с индуктивносвязанной плазмой, WDXRF -рентгенофлуоресцентная спектрометрия с дисперсией по длине волны, nd -не определено (концентрации рассматривае мых аналитов были значительно выше LOD).
an average particle size of less than 10 μm, but the distribution of particles is not unimodal, with outliers observed among the measurement results. We have recently applied wet grinding to prepare suspension from sulfide nickelcopper ores. With ZrO 2 grinding balls of a minimum diameter of 1 mm, adding water and the internal standard to the grinding jar of the mixer mill is a good way to achieve a homogeneous suspension with minimal particle size and good distribution of the internal standard. In this case, the reproducibility of the suspension preparation is 3-5 % for major elements in ores and nodules. One of the problems of TXRF analysis is strong peaks overlapping due to poor spectral resolution of the energy-dispersive spectrum. For instance, the differences between the TXRF results and the certified values in CRMs may be more than 30 % for As, Pb, V, Cr, Ba in the nodules due to PbLα/AsKα, BaLα/TiKα, CrKα/VKβ, VKα/TiKβ lines overlapping Chubarov et al., 2021]. Chemometric approaches may be used to solve this problem [Akhmetzhanov et al., 2021]. It was shown that the calibration models based on partial least squares and principal component regressions provide a quantitative determination of Ce, La, and Nd content in ores and nodules using the TXRF method. TXRF was also applied to study the composition of archaeological ceramics for the characterization of raw materials and manufacturing processes [Maltsev et al., 2021c;Pashkova et al., 2021]. TXRF may be considered a promising method in archaeology due to its capability to analyze a very small amount of a sample (<20 mg), which is very important for the preservation of archaeological material. Two sample preparation techniques for the TXRF analysis of ceramics were offered: suspensions for a bulk composition of ceramics and acid leaching for the elemental determination of the ceramics' clay component. The method was validated by the analysis of certified reference materials (CRM), comparison of results of TXRF with ICP-MS and WDXRF results, and interlaboratory studies. Application of TXRF method to the archaeological ceramics from Popovsky Lug, Eastern Siberia (Russia) allowed collecting the first data on the elemental composition of these samples. Additionally, the elemental composition of ceramics may be used for the next chemometric processing, which helps to identify the groups of different origins.
TXRF method was developed to analyze the sediments from Lake Baikal and Lake Khara-Nur (Russia) for environmental and paleoclimate studies [Pashkova et al., 2016[Pashkova et al., , 2018a. Samples for multielement analysis were obtained by cutting a sediment core with a certain step. Since the samples are quite limited in volume, the TXRF method is well-suitable for this task. The simple sample preparation based on mixing of 20 mg of milled sediment sample and 2 mL of an aqueous 1 % Triton X-100 solution and consequent quantification by the internal standard of Ga provides fast determination of Ca, Ti, Mn, Fe, Ni, Cu, and Sr. The external calibrations and correction to the Compton peak signal make it possible to compensate some systematic errors and expand the elements set with Al, K, V, Cr, Zn, Rb, and Ba. TXRF method was also applied to the Br determination in lake sediments with only one difference in quantification procedure: using internal standard of Ge instead of Ga [Pashkova et al., 2016]. The detection limit of Br was 0.4 mg/kg, which is almost two times better than that for the WDXRF method. The TXRF results were validated by the analysis of CRMs of sediments, and the comparison of results to the certified concentrations showed good recovery values. Interpretation of the measured concentration values is possible when the random error of analysis is 2-3 times less than the variations in the contents of the elements throughout the core depth. For most of the elements, the measurement and sample-preparation errors were less than 10-15 %, so the direct TXRF analysis can give information about changes in the element concentration with the core depth.

CONCLUSIONS
The brief examples of the application of TXRF demonstrate the potential of this method for geological, geochemical, and archaeological research. Careful preparation of a small volume of solid samples is the main condition for successful analytical measurements using TXRF spectrometry. We believe some new possibilities of TXRF will be found based on our experience.

CONFLICT OF INTERESTS / КОНФЛИКТ
ИНТЕРЕСОВ The authors have no conflicts of interest to declare. The authors have read and agreed to the published version of the manuscript.