On-site geochemical and indicator mineral analytical techniques in critical raw-metal exploration in the glaciated terrain

1 Introduction

Increasing self-sufficiency in critical minerals is one of the key objectives of the mineral strategies of the European Union (EU) and Finland. The aim is to reduce dependence on imports from outside the EU and ensure the availability of domestically produced raw materials for high-technology industries. Metals defined as critical by the EU include rare earth elements (REEs), antimony, cobalt, lithium, magnesium, nickel, niobium, tungsten, and platinum group elements (PGEs), which are used in the electronics industry, smart technologies, catalytic converters, and batteries. In Finland, as in other Nordic countries, there is significant potential for the discovery of new economically viable mineral deposits (Eilu, 2012; Sarapää et al., 2013). However, mineral exploration for new critical raw material (CRM) deposits is challenging in glaciated terrains, where continuous till cover and only few bedrock outcrops hinder the identification of mineralized bedrock.

In these regions, where the bedrock and its mineral resources are found under sub-cropping conditions, transported overburden, together with surficial geological exploration and research methods, has traditionally been used (Hirvas and Nenonen, 1990; Sarala, 2015). The oldest and effective methods include surface boulder hunting and glacial till sampling, combined with analyses of the geochemical composition and lithology of till (Gleeson et al., 1989). Among these, geochemical methods are most widely used and are nowadays supported by several portable analytical techniques such as handheld X-ray fluorescence (pXRF) (Hall et al., 2015; Sarala et al., 2015a; Lemière and Uvarova, 2019) and portable X-ray diffraction (pXRD) (Sarala and Koskinen, 2018). Furthermore, studies on heavy minerals (HMs) are highly effective for several commodities, including diamond, gold, PGEs, and various base metals (Lehtonen et al., 2005; McClenaghan, 2005; Mackay et al., 2015; Layton-Matthews and McClenaghan, 2022). There are also examples demonstrating that following appropriate modification, heavy mineral concentration and analytical methods can be used to identify indicator minerals for REEs in the glaciated terrain (Lehtonen et al., 2011; 2015; McClenaghan et al., 2023).

HM analysis is based on the concentration and identification of ore minerals (Peuraniemi, 1982) and is suitable for both soil and rock samples. It typically uses a large sample size (10 kg–100 kg), from which the heaviest minerals (specific gravity >3.3 g/cm³) are separated by gravity-based concentration methods. Various technical approaches are available for the concentration process, most of which rely on gravity and water.

Conventionally, mineral identification is carried out visually by examining the shape, color, and surface textures of mineral grains under a microscope (Averill, 2001). This method involves the estimation of abundance of individual minerals and has proven particularly effective for the study of heavy minerals such as sulfide minerals (e.g., arsenic-, copper-, iron-, nickel-, and zinc-bearing sulfides), chromite, magnetite, PGE minerals, rutile, scheelite, and gold (Lehtonen et al., 2005; McClenaghan and Cabri, 2011).

In addition to the essential ore minerals, many other minerals are characteristic of specific types of ore deposits and generally are found in more abundance and are more widely distributed in the bedrock than the ore minerals themselves. These minerals are referred to as indicator minerals because they reflect the presence of ore deposits and the enrichment of specific elements in the bedrock and/or soil (Lehtonen et al., 2015; McClenaghan and Paulen, 2017). Owing to their greater abundance and easier detection in soil samples, indicator minerals can be used effectively in mineral exploration, often providing a more practical exploration target than the essential ore minerals themselves.

Traditional HM analysis methods are not fully effective in the study of critical minerals because it is challenging to visually identify many of the characteristic indicator minerals or distinguish them from conventional HMs. In addition, many of the indicator minerals have lower densities than the minerals typically recovered in conventional HM concentrates; therefore, they are not retained in the traditionally enriched heavy-mineral fraction. To date, no well-established method has been developed for their efficient separation and concentration from sediment samples (Lehtonen et al., 2011). Due to the challenge of visually identifying the wide range of indicator minerals associated with critical metals, HM analysis methods have not been widely adopted for the exploration of all types of critical mineral deposits, particularly those containing REEs.

In the search for critical minerals, considerable potential exists not only in the discovery of new ore deposits but also in the re-evaluation of old mine sites, particularly their tailings and waste rock piles (Burlakovs et al., 2015; Juutinen et al., 2023; Kinnunen et al., 2022). The recovery and reuse of these materials have received increasing attention in recent years, and new approaches to their utilization are being explored within the framework of circular economy. Research on indicator minerals serves as an effective tool for the characterization and exploration of large tailings deposits and waste rock piles, helping assess their potential for hosting recoverable critical minerals. In principle, the same methods could be applied for mapping and evaluating the mineral potential of historical mine-waste deposits at abandoned mining sites.

To develop exploration methods of critical mineral and metals, the project titled Automated identification of indicator minerals in the search for critical minerals (Indika), funded by the European research development funding (ERDF), was carried out between 2016 and 2019. This project utilized the latest analysis and mineral identification technologies, and practical fieldwork was conducted to evaluate the suitability of new automated and field-based, or on-site, indicator mineral exploration methods for the exploration of different types of critical mineral and metal deposits in glaciated terrain in Finland (Lehtonen et al., 2015; Sarala et al., 2015b). The aim of the Indika project was to develop a comprehensive research concept for CRM indicator minerals, including pretreatment of soil samples, on-site method testing, and laboratory analyses. In addition, the project evaluated the suitability of individual methods and/or analyzers for indicator mineral research (Sarala et al., 2019). The research hypotheses were as follows:

• By applying different sample processing and enrichment methods, the separation of critical minerals and indicator minerals from other mineral materials can be optimized. This approach makes it possible to identify the most suitable processing methods for different sample sizes, varying grain size fractions, and the specific density of indicator minerals.
• On-site analysis methods, i.e., the analysis and research of mineral samples directly in the field, can accelerate the mineral exploration procedure and reduce the need for expensive laboratory analyses. Such methods include portable XRD and XRF analyses Their application could result in significant cost savings in mineral exploration.
• Applied research on new automated mineral identification methods can improve the speed and efficiency of the analysis and identification of CRM indicator minerals in soil samples. Suitable techniques include laboratory-based field emission scanning electron microscopy (FE-SEM-EDS) and mineral liberation analyzer (MLA). These methods can also be used to rapidly validate the results of on-site analysis methods.
• Comprehensive testing of field and laboratory methods provides reliable information on the performance of the selected method under northern environmental conditions. The testing also includes an assessment of the usability of different sample preparation procedures and data processing approaches.
• Piloting the developed methods at a new test site enables the evaluation of the overall performance of the proposed methodological framework. At the same time, comprehensive documentation facilitates the transfer of knowledge and provides opportunities for companies and future research projects to apply the developed methods and utilize the results.

The aim of this study is to develop and evaluate improved methodologies for exploration of indicator minerals in glaciated terrains. Specifically, the objectives are to

• Assess the effects of different sample processing and enrichment methods on indicator mineral recovery.
• Compare field-based (on-site) and laboratory-based analytical techniques.
• Evaluate the role of grain size and density fractions in the concentration of indicator minerals.
• Determine the practical applicability of these methods in mineral exploration workflows, including cost- and time-efficiency.

2 Study areas

Studies have been carried out in northern Finland (Figure 1), mostly on the northern side of the Arctic Circle. Owing to repeated glaciations in the Northern Hemisphere during the Quaternary, the bedrock is largely covered by glaciogenic sediments composed mainly of till. In the latest ice divide zone of the last (Weichselian) glaciation, sub-glacial erosion has been weak, resulting in short transport distances for the till (Sarala and Ojala, 2008; Johansson et al., 2011). In this region, geochemical exploration methods based on till, pre-glacial weathered bedrock, stream water, stream sediments, and surface weak leach geochemistry, which is based on the mobile metal ion (MMI) theory (Mann et al., 1998), have proven to be effective. Furthermore, indicator mineral studies complement till geochemistry and provide additional support for mineral exploration (Sarala, 2015).

The four test study areas are Mäkärä, Kyörtesselkä in Sokli, and Rautuvaara and Mustikkakangas in Vuolijoki (Figure 2).

The region containing the Mäkärä Au-REE prospect has been explored for several decades, with the most recent exploration campaigns having begun in 2009 (Sarapää and Sarala, 2012). The target is located within the strongly deformed Tana Belt, which comprises amphibolite, garnet-biotite, and arkose gneisses, thrust together with the Lapland granulite belt onto the Central Lapland belt (CLb) during 1.9 Ga (Sarapää et al., 2013). The Tana Belt in northern Finland is one of the most prospective regions for REEs and gold (Au) exploration in northern Europe (Sarapää et al., 2015). It is characterized by pronounced La and Y anomalies in regional surficial and lithogeochemical datasets. High Y concentrations indicate the enrichment of heavy REEs in the bedrock. This area also contains a substantial amount of Au, as evidenced by the occurrence of Au grains in the till and weathered bedrock, along with the presence of large paleoplacer gold deposits (Puustinen, 1991; Sarapää and Sarala, 2012).

Deep pre-glacial weathering and elevated Au contents in saprolite and glacial till show a strong positive correlation with the positive electromagnetic anomalies, which are associated with sulfide-bearing bedrock. According to Sarapää and Sarala (2012), the highest La and Y concentrations in the till correlate well with the Th maxima identified in airborne radiometric datasets. Exploration studies revealed two REE occurrences within the target area: the Au–hematite–quartz vein containing up to 3 ppm Au and 0.04%–0.1% REEs at Mäkärä and elevated REE concentrations up to 0.4% in the fine fraction of till at Vaulo. The elevated REE content resembles those of ionic-adsorption clay deposits in China. The principal REE-bearing minerals include monazite, rhabdophane, xenotime, and kaolinite. These findings indicate the significant potential of REEs in the Tana Belt because the Mäkärä–Vaulo area covers only a very small part of the belt (Sarapää et al., 2013).

Another study area is located in the eastern part of central Lapland, at Kyörtesselkä, next to the Sokli carbonatite massif. The Kyörtesselkä target is considered a prospective area for exploration of P–Fe–Nb–REE mineralization (Sarapää et al., 2015). The bedrock in the target is composed of a fenitic zone surrounding carbonatite massif and late-stage carbonatite and lamprophyre dykes that crosscut the fenitic zone that originally consisted Archean mafic volcanic rocks and tonalitic gneisses. These lithological units are considered to have the highest potential for REE mineralization in the target area. The Sokli carbonatite massif itself is a funnel-shaped, multistage pluton 5 km in diameter (Vartiainen, 1980). In addition to hosting an unexploited phosphate ore deposit, the Sokli intrusion is enriched in Nb, Ta, Zr, REEs, and U (Korsakova et al., 2012).

The third study area was the tailings area of the former Rautuvaara Fe–Cu mine in western Finnish Lapland. The mine operated during 1962–1988, but the processing plant at the site remained in operation until 1996. The tailing consists of process sands derived from ores mined at several different deposits that were transported to the Rautuvaara for enrichment processing (Räisänen et al., 2015). The main body of the tailings pond consists of ore-processed sand from the Rautuvaara mine, and the uppermost sand materials represent the latest ore materials from Laurinoja, Pahtavuoma, and Saattopora gold-bearing sulfidic ores (Juutinen et al., 2023). According to drilling observations, the thickness of the tailings ranges between 3 m and 9 m. In the water-saturated layers, the tailings remain largely unoxidized and have not undergone chemical weathering. In contrast, the near-surface layers have been oxidized and also chemically weathered, resulting in the release of metals into the pore water. Some of these metals are retained within the tailings deposit, whereas others are transported out of the waste area (Räisänen et al., 2015).

In addition, the methodological testing of the on-site research concept was piloted in the Vuolijoki region in Central Finland (Hugg and Heiskanen, 1983). The area hosts the former Otanmäki Fe–Ti–V mine and the known Pentinpuro Fe–Ti–V deposit in mafic intrusion host oxide ore bodies as lenses within gabbroic to anorthositic wall rocks. The area has also been recognized with prospective potential for REE mineralization. A 1-day sampling campaign was conducted in the Mustikkakangas study area on the SE side of Pentinpuro to trace the potential source for surficial boulders containing elevated concentrations of REEs (Sarala et al., 2019).

3 Methods

In total, four study areas were investigated: Mäkärä, Kyörtesselkä, Rautuvaara, and Vuolijoki. A total of 42 samples were used during the different stages of method development and validation. The study design incorporated the parallel application of field-based and laboratory-based methods. At each sampling site, replicate samples were collected and processed separately to enable comparison between field-based concentration techniques and laboratory-based sample processing and analytical workflows.

3.1 Sampling and sample materials

Till samples were collected from hand-made test pits in the Kyörtesselkä and the Rautuvaara tailings study areas, whereas the samples of the Mäkärä and Vuolijoki study areas were collected from the tractor-excavated test pits. The sample size in the analysis for indicator/heavy minerals was 12 L (c. 20 kg/sample–25 kg/sample). At each sampling site, three parallel samples (triplicates) were collected: (1) for field-based testing, (2) for laboratory analyses, and (3) as a reference/control sample for potential further analyses. In total, 42 samples were collected from the four study areas, including 18 samples from Mäkärä, nine from Kyörtesselkä, six additional samples from Rautuvaara, and nine from the Vuolijoki pilot site. In addition, small till samples (c. 300 g–500 g) were collected for the geochemical analyses and the second small sample set (c. 500 g–1,000 g) for the pXRD and pXRF analyses.

3.2 Laboratory techniques

Initially, two till samples (one each of the Mäkärä and Kyörtesselkä till samples) were used to determine the distribution of heavy minerals within the till samples, especially the enrichment of indicator minerals of critical metals in certain grain size, density, and magnetic fractions. The volume of each test sample was 12 L. The till test samples were first wet-sieved under a 1-mm size fraction, and then the heaviest fraction was separated using a shaking table. The shaking table was selected as the pretreatment method because an experienced operator can influence the density of the resulting concentrate and the method is well suited for the recovery of HMs from fine till material (<100 µm) (Lehtonen et al., 2015). In addition, the shaking table has a gentler approach on fragile mineral grains than the Knelson concentrator. Subsequently, magnetite was removed by weak and strong magnetic separation. Based on the results, sample processing was optimized to concentrate the fractions containing the most indicators using pretreatment, laboratory, and preparation methods. Electron microscopic analyses focus on expanding optical analysis focused on expanding the mineral libraries and developing procedures for reliable identification of indicator minerals.

Based on the first testing results of the Kyörtesselkä and Mäkärä test samples, the sample processing process was confirmed (Figure 4). The process flow presented in Figure 4 represents the laboratory-based sample-processing procedure only. Field-based concentration methods (spiral concentration and Knelson) were applied separately and in parallel to laboratory workflows using subsamples of the same material.

The developed procedure, its stages, and the results are presented in more detail in a separate report by Lukkari and Lehtonen (2019).

There were 11 actual preprocessed and analyzed project samples in total. The volume of their sample sizes were 12 L. There were four till samples from the Mäkärä site, three till samples from Kyörtesselkä, and four from the Rautuvaara tailings pond in Kolari. No samples were used in laboratory analyses from the Vuolijoki pilot site.

3.3 Heavy liquid separation

Heavy liquid separations were performed for the earlier mentioned 11 shaking-table preprocessed till samples with undiluted methylene iodide (3.3 g/cm³) and LST (lithium heteropolytungstate aqueous solution; 2.8 g/cm³). Methylene iodide separation was performed using a separatory funnel for the mineral material with a grain size of more than 0.25 mm. The separation of finer mineral material was enhanced by using a centrifuge. The disadvantage of the LST method was high sample loss and imprecise separation.

3.4 FE-SEM+EDS

The mineralogical composition of the magnetic mineral-free fractions obtained via heavy liquid separation was determined with a field emission scanning electron microscope (FE-SEM) coupled with an energy-dispersive spectrometer (EDS) in the laboratory of the Geological Survey of Finland (GTK) in Espoo. The samples were analyzed using a JEOL JSM 7100F FE-SEM connected to an Oxford Instruments X-Max 80 mm² EDS and operated by INCA software. For analysis, 25-mm diameter surface thin sections were prepared from the sample fractions, cast in epoxy and polished, and coated with graphite to provide electrical conductivity. One mount was prepared from each fraction that was studied.

The analysis parameters of the FE-SEM-EDS were high vacuum, COMPO backscatter signal, 20 kV accelerating voltage, and 1.3 nA current intensity. Mineral grains (10,000–20,000) from each preparation were analyzed in an automated run. The EDS spectrum was generated from the center of the longest grain diameter, with a minimum grain size of 7 µm (ECD).

3.5 Field techniques

In connection with sampling, different methods for the concentration of the till samples in the field were tested. During the summer and autumn of 2017, samples were collected from the Au–REE mineralization (Mäkärä target), the southern fenitic zone of the Sokli carbonatite deposit, and the Rautuvaara tailing sand basin in Kolari. Additional samples were collected during the summer of 2018 from the Mustikkakangas REE study area in Vuolijoki. The aim was to compare the proportions of REE indicator minerals obtained using two concentrating techniques: a spiral concentrator and a Knelson concentrator.

First, each sample was mixed throughout to homogenize the sample material before being divided into two subsamples. As a preprocessing step, both subsamples were wet-sieved in the field to the <2-mm size fraction. The first subsample was then concentrated using a spiral concentrator (also known as a Gold hound concentrator), whereas the second was processed using the Knelson concentrator. The resulting concentrate weights ranged from some tens of grams for the spiral concentrator to hundreds of grams for the Knelson concentrator. Spiral concentrator concentrates are usually suitable for direct microscopic examination, whereas the Knelson concentrates require further post-processing, such as using micro-panning or heavy-liquid separation (McClenaghan et al., 2023). Under field conditions, micro-panning is the most practical and cost-effective method for separating the heaviest mineral fraction in the indicator mineral research. In addition, magnetite is typically removed from the concentrates to reduce the amount that must be examined.

3.6 Portable XRD

An Olympus Terra 542 Portable XRD Analyzer owned by GTK was used for pXRD testing. An introduction to the Olympus Terra analyzer and analytical procedure and data interpretation was provided by Sarala and Koskinen (2018). In addition to pXRD, the samples were analyzed using laboratory XRD and FE-SEM-EDS instruments in CMO. The pXRD analytical results were processed and compared using two software applications: XPowder and PDXL2.

In this study, the final concentrates were separated into two size fractions (<150 μm and >150 μm) to estimate the grain size distribution of indicator minerals and their abundance in different fine material fractions. The 150-μm threshold was selected because it represents the upper grain size limit for pXRD analysis. The coarser material must be pulverized prior to pXRD analysis.

For validation, reference analyses were performed in the GTK’s mineralogical laboratory using a Bruker XRD instrument and diffrac.eva software and in CMO using Rigaku SmartLab XRD and PDXL2 software. For the laboratory XRD, the samples from both <150 µm and >150 µm size fractions were pulverized prior to analysis. In addition, some duplicate XRD samples were analyzed at the ALS laboratory to compare the results with those obtained from the commercial laboratory.

3.7 Portable XRF

In this study, we used an Olympus Innov-X Delta 6000 portable X-Ray Fluorescence analyzer. The analyzer is equipped with a 4-W X-ray tube with an Rh anode, 8 keV–40 keV, 5 μA–200 μA, and a high-resolution, thermo-electrically cooled Si PIN diode detector. The GeoChem mode was used for all measurements. The time for a single measurement was set as 40 s: 10 + 15 +15 s for beams 1–3, respectively. Each sample was measured three times, and the average of the three measurements was used for analysis (cf. Sarala et al., 2015a).

The elements determined in the GeoChem mode are MgO, Al₂O₃, SiO₂, P₂O₅, S, Cl, K₂O, CaO, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Y, Zr, Nb, Mo, Pd, Ag, Cd, Sn, Sb, Ba, La, Ce, Hf, Ta, W, Pt, Au, Hg, Tl, Pb, Th, and U. The detection limits are not constant in the case of pXRF. Instead, when the measured concentration of an element is below the instrument’s detection capability or is considered unreliable, the pXRF reports the result as <LOD (limit of detection), indicating that the element was not detected.

3.8 Summary of analytical techniques

The analytical workflow can be summarized as follows:

• Sampling and subdivision into parallel field and laboratory samples.
• Field-based concentration (spiral and Knelson methods).
• Laboratory pretreatment (sieving, shaking table, and magnetic separation).
• Heavy liquid separation.
• Mineral identification (FE-SEM-EDS).
• Supporting analyses (pXRD and pXRF).

4 Results

4.1 Heavy mineral concentration in the laboratory

The mineralogy results of the heavy mineral concentrates from Mäkärä, Kyörtesselkä, and Rautuvaara indicate that the enrichment process for the indicator minerals was successful. The indicator minerals have been enriched in the analyzed fractions in concentrations of up to several percent or even tens of percent. They are generally found in till and tailings sands at ppm to ppb levels. The analyses identified monazite [(Ce, La, Nd, and Sm) (PO₄)] and pyrochlore [Ca₂Nb₂O₇] as the principal indicator minerals for critical metals.

Table 1 lists the minerals present in the heavy fractions with their chemical formulas and mineral abbreviations. The indicator minerals observations derived from different grain size fractions from Mäkärä, Kyörtesselkä, and Rautuvaara are listed in Table 2.

TABLE 1. Minerals occurring in the project samples, their chemical formulas, and their abbreviations according to Kretz (1983). Mineral abbreviations that are not in Kretz’s classification are marked in italics.

TABLE 2. Indicator mineral observations derived from the heavy fractions (d > 3.3 g/cm³) of the Mäkärä, Kyörtesselkä, and Rautuvaara samples in two grain size classes.

The mineral abbreviations used are shown in Table 1. Samples marked with an asterisk have been run on a Frantz magnetic separator with a current of either 0.3 A or 0.5 A.

The most common heavy minerals in the Mäkärä till samples are epidote, rutile, clinozoisite, Fe-oxide, zircon, almandine, Al-silicate, and hornblende. Among the main minerals, zircon is also classified as an indicator mineral. In addition, other indicators such as monazite, apatite, allanite, xenotime, and baddeleyite were observed. The zircon content of the heavy mineral concentrates is up to 29 vol%, and the next most common indicator mineral is monazite, with a content between 0 and 2.8 vol%.

In the heavy mineral concentrates of the Kyörtesselkä till samples, the main minerals are rutile, zircon, epidote, clinozoisite, Fe oxide, almandine, pyrochlore, baddeleyite, and monazite. Of these, the indicator minerals are zircon, pyrochlore, baddeleyite, and monazite. In addition, the indicator minerals apatite, allanite, xenotime, and zirkelite are also present. The pyrochlore content in the concentrates varies between 0.1 and 10.9 vol%, the monazite content is 0.1 vol%–11.7 vol%, and the baddeleyite content is 0 vol%–8.1 vol%. The zircon content is as high as 20.8 vol%.

The main minerals in the concentrates of the Rautuvaara tailing samples are feldspar, Fe oxide, hornblende, clinopyroxene, pyrite, and almandine. The most common indicator mineral is tungsten-containing rutile (W-rutile), where tungsten content is in the range of 1.5 wt%–5 wt%. W-rutile content is in the range of 0.1 vol%–2.0 vol%. Other indicator minerals, monazite, cobaltite, apatite, allanite, xenotime, zircon, baddeleyite, and thorite, are found in small concentrations. In addition, small amounts of gold grains were also found. The monazite content varies between 0.1 and 1.1 vol%.

In the samples processed with the Frantz magnetic separator, critical minerals present in trace amounts were effectively identified and separated into different magnetic fractions. Zircon, baddeleyite, and pyrochlore are enriched in the 0.5-A fraction, and monazite is enriched in the 0.3-A fraction. Critical minerals mainly occur predominantly in the 63-µm–100-µm grain-size fraction. However, monazite and pyrochlore in the Kyörtesselkä sample 13.2 and xenotime in the Mäkärä sample 8.2 are concentrated in the coarser 100-µm–160-µm fraction.

4.2 pXRD analyses

A total of 13 till samples from the Mäkärä study area, 10 samples from Kyörtesselkä, and nine from Vuolijoki were analyzed in the field using pXRD. The sample consisted of till collected from the C horizon and concentrated as mentioned in Section 3.5. Six Rautuvaara tailing sand samples were taken from two sample points within the tailing area and from three different layers (Juutinen et al., 2023). The samples were enriched using both a spiral concentrator and a Knelson concentrator. Two concentration runs were performed with the Knelson concentrator; the first recovered the heavy mineral fraction, while the second run recovered medium–heavy minerals. Prior to pXRD analysis, most of the concentrated samples were further processed by micro-panning to reduce the amount of lighter minerals. The grain-size fraction below 150 µm was sieved from the micro-panned sample and used as is in pXRD analysis. The 150 µm–2 mm portion was pulverized with a mortar for pXRD analysis. Measurement times with the pXRD analyzer ranged from approximately 15 min to more than an hour depending on the quality of the diffraction pattern obtained.

Table 3 compares the analytical results obtained by pXRD/XPowder and Rigaku laboratory XRD/PDXL2 with those obtained from two FE-SEM analyses of the same sample, RM_POS-2028-7.1. The sample was collected from the Vuolijoki study area and concentrated using a Knelson concentrator, followed by micro-panning. The comparison shows a consistent trend in the XRD analysis results: the proportions of Fe oxides are underestimated, whereas those of plagioclase are overestimated. On the other hand, the proportions also vary between different FE-SEM analyses, particularly with respect to the relative abundances of Fe oxides to ilmenite.

TABLE 3. Mineral proportions in the Vuolijoki sample RM_POS-2018-7.1 to show comparisons of pXRD (pXRD/XPowder) and lab-XRD (Rigaku/PDXL2) analyses vs. FE-SEM analyses performed in Oulu university’s CMO and Espoo GTK’s laboratories.

The pXRD results show the same minerals as in the other analyses, except for chlorite and rutile. Pyrite, aeschynite, and vermiculite are only identified in the pXRD column. Allanite is only identified in the laboratory XRD analysis but not in any FE-SEM analysis. In addition to the main minerals, zircon is identified in all analyses, and monazite is identified in all analyses except the laboratory XRD analysis.

Table 4 (Mäkärä), Table 5 (Kyörtesselkä), and Table 6 (Vuolijoki) present the REE minerals identified by pXRD in different sample fractions. The last column lists the REE minerals identified in the sample by FE-SEM analysis.

TABLE 4. REE minerals identified by pXRD in different sample fractions (fine <150 µm and coarse 150 µm–2 mm) of the Mäkärä samples.

SC, spiral concentrator, Knel = Knelson; Numbers 1 and 2 correspond to the concentrator runs. Explanations of mineral abbreviations can be found in Table 1. In addition, Cer = cerite, Rha = rhabdophane, Anc = ancylite, Eux = euxenite, Aes = aeschynite, and Fer = fergusonite. No results were obtained from samples: marked as a dashed line, and no analyses: empty boxes.

TABLE 5. REE minerals identified by pXRD in different sample fractions (fine <150 µm and coarse 150 µm–2 mm) of the Kyörtesselkä, Sokli samples.

Knel = Knelson; numbers 1 and 2 correspond to the concentrator runs. Explanations of mineral abbreviations can be found in Table 1. In addition, Cer = cerite, Rha = rhabdophane, Anc = ancylite, Eux = euxenite, Aes = aeschynite, and Fer = fergusonite. No results were obtained from samples: marked as a dashed line, and no analyses: empty boxes.

TABLE 6. REE minerals identified by pXRD in different sample fractions (fine <150 µm and coarse 150 µm–2 mm) of the Vuolijoki samples.

SC, spiral concentrator, Knel = Knelson; numbers 1 and 2 correspond to the concentrator runs. Explanations of mineral abbreviations can be found in Table 1. In addition, Cer = cerite, Rha = rhabdophane, Anc = ancylite, Eux = euxenite, Aes = aeschynite, and Fer = fergusonite. No results were obtained from samples: marked as a dashed line, and no analyses: empty boxes.

The indicator mineralogy of the Mäkärä area demonstrates a highly enriched REE-bearing mineral assemblage and support earlier observations and assessments pertaining to the high REE potential in the area (cf. Sarapää and Sarala, 2012; Sarapää et al., 2013). The indicator minerals are derived from a pre-Quaternary weathering crust, from which glaciers eroded REE-bearing minerals and incorporated them into the till material. Monazite, columbite, and xenotime are the dominant REE-bearing minerals found throughout the area. Most of the indicator minerals can be classified as heavy minerals (density >3.3 g/cm³) as they are predominantly concentrated in the fine fraction <150 µm of the first concentration run (both SC1 fine and Knel 1 fine). However, an interesting observation is that few indicator minerals were identified in the FE-SEM + EDS analysis.

In Kyörtesselkä, mineralogy is somewhat different due to carbonate-rich bedrock. Columbite, monazite, pyrochlore, rhabdophane, ancylite, and xenotime are the most common minerals in HM concentrates. Most of these are REE-bearing minerals that support the recognized high CRM potential in Sokli and its surrounding areas. An interesting observation is the pronounced occurrence of xenotime in the fine fraction recovered during the second Knelson concentration run. This enrichment cannot be readily explained by the deviation of the density of xenotime 4.4 g/cm³–5.1 g/cm³ as it is similar to that of other observed REE minerals such as monazite (4.6 g/cm³–5.7 g/cm³), pyrochlore (4.4 g/cm³–4.9 g/cm³), columbite (5.3 g/cm³–7.3 g/cm³), and ancylite (3.95 g/cm³). The enrichment may instead be related to the high abundance of other REE minerals in the concentrate or to differences in the crystal morphology and particle behavior during gravity separation.

The mineral composition differs somewhat from that of the Mäkärä and Kyörtesselkä study areas as allanite, aeschynite, cerite, and pyrochlore are relatively abundant minerals. However, columbite and monazite are also common. A characteristic feature of the area is the high diversity of the REE mineral species present in many samples, indicating a heterogeneous REE-rich bedrock.

4.3 Rautuvaara tailings sands

In the Rautuvaara tailings, the pXRD spectra were generally of poor quality, most likely due to the material matrix and mineralogical composition. The main minerals identified by pXRD were albite, clinochlore, pyroxene, pyrrhotite, ilmenite, and rutile. Other significant minerals, including allanite, arsenopyrite, pyrite, realgar, arsenolite, columbite, monazite, and xenotime, were detected only in the fine fraction (<0.063 mm) of the HM concentrate using XPowder analysis. These minerals were also detected in the FE-SEM analyses; however, their proportions are variable, and particularly, proportions of ilmenite and albite were clearly overestimated compared with the pXRD results.

FE-SEM analyses in the micro-panned material and pXRD analysis of the non-micro-panned material revealed interesting details concerning monazite, rutile, and xenotime. Their occurrence indicates that the tailings sands are potential sources of REEs. Furthermore, pXRD enabled the identification of several As-bearing minerals, including realgar (arsenic sulfide) and arsenolite (arsenic oxide). These minerals were also only recognized in the FE-SEM analysis. This indicates that the identification of certain minerals may depend on the FE-SEM classification criteria used in the laboratories; nevertheless, the pXRD diffractograms appear to contain peaks consistent with the presence of these minerals. In addition, pXRD identified arsenopyrite, which was abundant in all FE-SEM analyses.

4.4 Portable XRD vs. laboratory XRD

Some XRD samples from the Mäkärä area were sent to ALS laboratory to compare the pXRD results with those obtained from laboratory-based XRD analyses. Compared with laboratory XRD spectra, the spectra produced by pXRD are of lower quality due to the faster measurement speed and lower resolution. The noise level is relatively high, making it difficult to distinguish small peaks from the background. One solution is to radically extend the measurement time, which would mean at least an hour of measurement time per sample.

The angular range measured in the portable XRD analyzer is 5°–55°, which is significantly narrower compared to that of laboratory XRD, where the angular range is 2°–120°. The narrow angular range proves challenging in identifying phases where many peaks are located outside the range. If these phases are present in the sample, quantitative analysis will not be successful.

Table 7 compares the XRD results obtained from the ALS laboratory from the Mäkärä till sample RM_JOV$-2017-3.1 with the results obtained with the portable XRD on different sample fractions concentrated using both spiral and Knelson concentrators (spiral concentrator HM fraction <150 μm; Knelson HM fraction run 1 <150 µm and Knelson HM fraction run 2 <150 µm). The samples concentrated with a Knelson concentrator were also analyzed with the Rigaku XRD using PDXL2 program at CMO.

TABLE 7. Comparison of pXRD results with ALS laboratory XRD and Oulu Rigaku XRD results. The results show mineral division in the three different concentrates (<150 µm) of till sample RM_JOV$-2017-3.1 from the Mäkärä study site.

Samples analyzed in the ALS laboratory are marked as ALS1-3; KK, spiral concentrator; Knel1, Knelson run1; Knel2, Knelson run2.

The results demonstrate that pXRD can be used to identify large groups of minerals with a performance comparable to that of laboratory-based XRD techniques. It means that important indicator minerals can be identified directly in the field as part of sample processing and analysis. This provides a substantial benefit for mineral exploration by enabling a more efficient and rapid exploration workflow. However, a large variation is observed in the reported abundances of different minerals not only between pXRD and laboratory XRD analyses but also between laboratories. This makes a true comparison difficult and, in some cases, may result in a significant source error.

4.5 pXRF analyses

Portable XRF analyses were used to support the mineralogical study conducted in the field. Although the element concentrations are relatively low, several meaningful geochemical features can be identified. For example, concentrations of Ca and P, along with Ba, Nb, Zr, and Y, are clearly elevated at Kyörtesselkä. This observation is consistent with the till composition that reflects the underlying bedrock. The low concentrations of battery metals such as Co, Cu, and Ni do not yield particularly encouraging results from an exploration perspective; however, this is understandable given the mineralization types present in the area, which are not typically associated with elevated concentrations of these metals.

Use of portable analyzers in the glaciated terrain has the following restrictions: a) sampling should be conducted at an appropriate depth/stratigraphic layer to get the most representative sample materials, b) the abundance of indicator minerals can be low in till due to glacial erosion and mixing of materials during transportation and deposition of till, and c) for the same reasons, element concentrations may occur at levels that are close to or below the detection limits of the portable analyzers. In addition, the analytical coverage of portable analyzers for certain minerals and elements can be limited, which restricts the usability of the analyzers. The low concentrations of many elements (contents generally low or values not detected), including CRMs and REEs, are seen in the pXRF results of the study areas (Table 8).

TABLE 8. Summary of selected elements’ concentrations from single measurements of bulk soil samples from a) Mäkärä (five samples; 15 single measurements), b) Kyörtesselkä (10 samples; 29 single measurements), and c) Vuolijoki (14 samples; 41 single measurements). Samples were analyzed with Olympus Delta using the GeoChem mode.

5 Discussion

The results show that the use of multiple sampling locations enabled the evaluation of method performance across different geological settings. One objective was to assess the applicability of the new procedures for the exploration of CRM and REEs, and the different ore exploration and mine tailings targets were well suited for this purpose. Heavy minerals and their indicators have traditionally been used in the exploration of gold, diamonds, sulfidic base metals’ ore deposits, and magmatic Ni–Cu–PGE deposits (Peuraniemi, 1982; Peuraniemi and Heinänen, 1985; Sarala et al., 2015b; Averill, 2001; McClenaghan and Kjarsgaard 2007; Sarala and Peuraniemi, 2007; McClenaghan and Cabri, 2011; McClenaghan and Paulen, 2017). The intense development of FE-SEM methods has enabled its efficient use in the study of the fine fractions of till (Lehtonen et al., 2011; 2015).

The aim of the HM research was to develop an efficient and reliable concentration procedure for the recovery of indicator minerals from the till, weathered bedrock, and mine tailings and to support the identification of these indicator minerals using automated/semi-automated scanning electron microscopes with EDS.

The efficient indicator mineral enrichment process is important for method development because indicator minerals generally occur at very low concentrations, particularly in till. However, statistically significant numbers of indicators may be present in the concentrated fractions, which makes it possible to compare sample-specific concentrations and generate useful exploration data. If the enrichment process is insufficiently effective or the indicator mineral observations remain sporadic for other reasons, the results lack a reliable basis for comparison and the applicability to ore exploration remains uncertain.

The usability of the research method is also influenced by its laboriousness. Technically well-functioning sample processing and analytical procedures are of limited value if they highly time-consuming and resource-intensive—thus increasing costs. One of the key objectives for methodological development is to eliminate the sample fractions that are not relevant to the research objectives as early as possible in the processing workflow, thereby enhancing efficiency and reducing resource consumption.

The results of this study indicate that the optimal sample preparation procedure depends on the target minerals and their chemical composition, density, and grain size. Sampling from multiple locations improved the reliability of the study by demonstrating that the optimized processing methods are applicable across variable geological environments. Based on the results of the test samples, it was found that

• The shaking table is well suited for use before concentration.
• Sieving can be used to eliminate less-relevant grain size fractions early in the sample processing phase, thereby reducing the amount of material requiring further processing.
• Weak and strong magnetic separation can further remove unimportant sample fractions and concentrate indicator minerals into specific fractions.
• Of the two heavy-liquid separation methods evaluated, methylene iodide proved to be the most suitable for research on indicator minerals as it effectively separated the most important indicator minerals from the most common bedrock minerals.
• The most critical fractions for research on indicator minerals are indicator mineral concentrates with grain sizes of 63 µm–100 μm and 100 µm–150 µm.

5.1 Sources of error in the laboratory sample preprocessing and concentration

In principle, multistage sample processing involves sources of error. These include the loss of sample material and increased risk of contamination at various stages of the sample processing workflow.

In addition, different stages of the process involve different risk factors. One limitation of use of the shaking table is its poor recovery of leave-/plate-like minerals. For example, flaky or leafy gold is not obtained in the concentrate but may instead be found in the process waste. A challenge associated with use of magnetic separators is that it is difficult to clean them thoroughly, thereby increasing the risk of contamination. During heavy-liquid separation, lighter indicator mineral material may be lost, e.g., as inclusions in the porous minerals such as goethite that can easily be floated during the heavy liquid separation.

A key challenge in sample preparation is ensuring sample representativeness as the prepared sample typically comprises only a small proportion of the test fraction. In the production of thin sections, it is possible that the sample material becomes segregated and that the heaviest mineral phases are lost during grinding. The analytics emphasize the professionalism of the operator and the interpretation of the results. Care was taken to minimize all sources of error through application of careful working practices. Nevertheless, these factors may have had a cumulative effect on the results, although their overall impact was likely minor.

5.1.1 FE-SEM+EDS

The advantages of the identification method based on the automated FE-SEM + EDS analysis include the ability for accurate imaging, analysis, and concentrations of the mineral grains contained in the till samples. At the same time, this method enables other significant observations relevant to mineral exploration, such as the identification of inclusions within the indicator mineral grains and other mineral matter attached to them from their source rock. Correlations between the concentrations of different indicator minerals and associated mineralogy can also reveal mineral assemblages that can be used to assess the type and mineralogical characteristics of the source rock. Variations in indicator minerals’ concentration, grain size, and grain morphology may, in turn, indicate the length of the transport distance (e.g., Lehtonen et al., 2005).

If necessary, the assembly of indicator mineral grains detected by FE-SEM the electronics quantitatively with a microanalyzer (EPMA) on the same sample preparations, thereby confirming the identification of minerals with complex chemical formula. In addition, microanalytical research provides an analysis of trace elements in indicator minerals, the results of which may be significant in characterizing the original source(s).

The technical limitation of automated FE-SEM + EDS analysis in the research on indicator minerals are primarily related to the grain size range of the sample material. In practice, fractions <5 µm and >500 µm cannot be effectively analyzed using this method. Other limitations include the labor-intensive nature of the method and its high cost. The analytical phase is preceded by numerous work steps, including sample collection, pre- and laboratory processing, and preparation of analytical mounts. Sample handling, analysis, and interpretation of the results require skilled personnel and a high-level laboratory infrastructure. In addition, differences in mineral libraries and the software used in the identification of minerals may introduce systematic errors during the interpretation phase. For these reasons, sample processing, analysis, and data interpretation are inherently time-consuming. Thus, this method is not applicable to large sample volumes but is best used as a complementary technology alongside other exploration methods and field-based analytical techniques (till geochemistry, pXRF, and pXRD). Based on other methodological experiences, it is recommended to select only a portion of the samples for FE-SEM analysis.

5.2 Portable XRD

Portable XRD seems to work well in identifying major indicator or ore minerals in concentrated till samples. However, compared with laboratory-based XRD instruments, pXRD produces diffraction patterns that are less accurate and cover a more limited 2θ angle range, which is important for the identification and precise positioning of XRD peaks. Still, they can reliably identify the most common minerals and, under favorable conditions, also less common minerals. The identification of REE-bearing minerals requires a purpose-designed enrichment process and supporting data from other analytical methods. For example, elemental concentrations data obtained via pXRF analyzer can be used to confirm, verify, and support identifications of minerals based on XRD results. Reliable quantitative mineralogical analysis cannot be derived directly from the graph produced by pXRD, at least not without further calibration, study of the analysis software, use of appropriate reference materials, and determination of the coefficients used in calculating the proportions.

Rather than emphasizing the number of samples, greater attention should be paid to sample quality, careful sample handling, and minimizing the sources of error that may distort the measurement. The most reliable pXRD results are obtained when the diffraction pattern is as clear as possible, i.e., the peaks can be well resolved, and the background noise is low. To obtain a high-quality diffraction pattern, the measurement time must be sufficiently long, ranging from approximately 15 min to 2 hours, depending on the sample. The grain size distribution of the material loaded into the sample box must be as even as possible. Within the framework of this study, the optimum grain size was found to be 0.063–0.150 mm, although a narrower grain-size range would likely yield even better results. Care should also be taken during sample loading to ensure that the grains can rotate freely in the vibrating sample holder. Magnetic separation prior to pXRD analysis would likely be beneficial both to obtain a better diffraction-pattern quality and to improve the concentration of REE-bearing minerals.

In the comparison of analysis software, neither of the programs used proved to be clearly better. In both cases, effective use of the software requires familiarization with the functions and analytical tools. The XPowder program uses the AMCSD mineral database and in the PDXL2 ICDD database. The latter is more widely recognized and is considered more reliable. Both the GTK laboratory in Espoo and the CMO of the University of Oulu use the ICDD database. Its advantage is, among other things, a quality assessment of mineral records, based on which the user can see whether the record selected for analysis has been proven reliable.

Comparison of FE-SEM and pXRD results shows that some of the most common minerals, at least biotite, chlorite, and calcite, were almost entirely absent from the pXRD results. Naturally, those minerals with most diagnostic diffraction peaks that lie outside the 2θ range of the portable XRD analyzer are most likely to remain unidentified, for example, calcite and fluorite. Other factors may also contribute to unsuccessful mineral identification. These include the amorphous or metamictic state of mineral or weak reflections characteristic of the mineral that do not produce distinct peaks above the background noise in the XRD diffraction pattern. The reason may also be the behavior of the mineral during the measurement. If mineral grains tend to accumulate along the edges of the sample chamber rather than move freely within the vibrating chamber, they will not be included in the analysis. Some light or platy minerals may be lost during the concentration process, particularly during washing, which can further reduce their likelihood of detection.

Incorrect mineral identifications can be influenced by the overlap of diffraction peaks from several different minerals, as well as from the user’s interpretation of significant peaks and the degree of correspondence between the observed diffraction pattern and the reference patterns in the database. A thorough familiarization with XRD analysis software and the many sources of errors can significantly improve the interpretation of diffraction patterns. The use of reference materials and standard samples and creation of own custom mineral libraries help develop a more streamlined analysis process for a sample set with uniform mineralogy.

Comparison of the FE-SEM + EDS and pXRD results also showed that pXRD may identify minerals that are not detected in FE-SEM analysis due to technical limitations. For example, minerals with identical elemental compositions but different crystal structures produce distinct diffraction patterns and can, therefore, be distinguished by XRD, even though they are indistinguishable from each other in FE-SEM analysis. Consequently, pXRD can provide complementary mineralogical information that is not obtainable from elemental composition data alone.

5.3 Portable XRF

Portable XRF can now be considered a standard on-site analytical method and has become widely used in mineral exploration. Recent advances in instrument performance, average value counting, calibration procedures, and data processing have further increased the reliability of portable XRF measurements for exploration applications. It provides rapid, nondestructive, and cost-effective elemental analyses directly in the field. The pXRF instruments require only small sample amounts and minimal sample preparation–even natural, moist till samples can be used in analysis (Sarala et al., 2015a). The pXRF can be flexibly used in the analysis of rock, soil, drill core, and sediment samples, making it valuable tools at different stages of mineral exploration. The ability to obtain near real-time geochemical data significantly reduces the time required for decision-making during exploration campaigns and enables more efficient targeting of sampling locations (Sarala, 2016; Lemière and Uvarova, 2019; Balaram and Sawant, 2022).

In glaciated terrains, however, the application of pXRF presents specific challenges. Extensive Quaternary deposits, particularly glacial till, commonly conceal the underlying bedrock. As a result, mineral exploration relies heavily on indirect methods that identify geochemical and mineralogical signatures transported by glaciers. The heterogeneous composition of till, variable grain size, and the presence of weathered and transported material can influence pXRF measurements and complicate the interpretation of geochemical anomalies (Sarala, 2015; Lemière, 2018). Furthermore, the dilution of ore-related minerals by abundant background material may reduce the detectability of pathfinder and critical elements. In particular, mineral alteration and weathering can result in the depletion or enrichment of certain element concentrations.

Despite these limitations, pXRF has proven to be an effective screening tool for reconnaissance exploration in glaciated environments. In this study, pXRF demonstrated its ability to analyze both major and trace elements. However, element concentrations for many CRMs (e.g., REEs and PGEs) are relatively low in till, making direct analysis using pXRF difficult. In such cases, pathfinder elements can often be used to compensate limitation.

5.4 Portable analyzers as a part of mineral exploration analysis

The portable, handheld analyzer enables the rapid identification of anomalous concentrations of major and trace elements, as well as a wide range of indicator minerals. Although on-site analytical methods (pXRF and pXRD) have limitations in terms of accuracy and detection limits, they offer clear advantages in rapid decision-making and can significantly reduce exploration time and costs when used in combination with laboratory validation.

These methods allow exploration programs to prioritize samples for more detailed laboratory analyses. When integrated with glacial geological observations, heavy mineral studies, indicator mineral investigations, and laboratory-based analytical methods, they can substantially improve exploration efficiency while reducing analytical costs.

Recent advances in instrument performance, calibration procedures, and data processing have further increased the reliability of portable measurements for mineral exploration applications. In some cases, portable analyzers can be used as a standalone analytical tool, such as pXRF in base metal exploration, where element concentrations are in the hundreds or thousands of ppm. However, the greatest benefit is achieved when portable analyzers are used as part of a multi-method exploration strategy rather than as a standalone analytical technique. In the exploration of critical raw materials, where the rapid assessment of large numbers of samples is required, portable XRF has become an increasingly important component of modern mineral exploration workflows in glaciated terrains.

6 Conclusion

This article describes the activities applied to development of research methods involving heavy and indicator minerals commonly used in the exploration of precious and base metals, as well as new identification techniques and analytical instruments for the exploration of so-called critical minerals and metal ores. The results show that:

• The separation of indicator minerals of critical minerals and metals from other sample material can be optimized by applying different sample handling and enrichment methods. In this way, the most suitable processing methods can be found for samples of different sizes and varying grain size fractions as well as for the specific gravity classes of the indicator minerals. The optimal sample preparation method depends on grain size distribution and mineralogical composition, with the 63–150 µm heavy fraction providing the most reliable indicator mineral.
• On-site analysis methods, i.e., portable, hand-held analyzers speed up the mineral exploration procedure and reduce costly laboratory analyses. Such methods include, for example, portable XRD and portable XRF analyzers. This would lead to significant cost savings in mineral exploration as the samples can be screened beforehand and by that way diminish the amount of samples to be sent to the laboratory. However, on-site analytical methods are most effective when used in combination with laboratory techniques, rather than as standalone tools. Field methods significantly improve efficiency, but their analytical limitations in the case of CRM must be considered in interpretation.
• Applied research has focused on advanced laboratory techniques, such as automated mineral identification methods, which can accelerate and improve the efficiency of the analysis and identification of critical minerals in soil samples. One suitable technique is the field emission scanning electron microscope equipped with energy-dispersive X-ray spectroscopy (FE-SEM-EDS). These methods can be used to validate the results of on-site analytical methods.
• Comprehensive testing of both field and laboratory methods provides reliable information on the functionality of the method in northern environmental conditions. The testing also included evaluating the usability and data processing of different types of sample preparations.
• Through practical pilot testing and the results of methodological development for exploring critical metals and minerals, the functionality of the new expertise and methodological concept was evaluated. Another benefit is that the documented information on methods can be utilized by companies and follow-up projects.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. It was used to check the language of the revised manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Anna Ladenberger, Geological Survey of Sweden, Sweden

Reviewed by: Malin Andersson, Geological Survey of Norway, Norway

Joao Silva, University of São Paulo, Brazil