Pollution sources and metallic elements mobility recorded by heavy minerals in soils affected by Cu-smelting (Legnica, SW Poland)

The observed heavy mineral assemblage was diverse in both magnetic and non-magnetic fractions. The quantitative modal composition was not analyzed for this material, but clearly, different types of Fe-rich phases dominate in magnetic fraction, which was also several times more abundant by weight compared to non-magnetic one. Modal based on simple grain-by-grain calculations is given in Table 2. The Fe-rich phases were divided into two groups of Fe oxide (SEM-EDX analyses showed only Fe and O spectra), and Fe oxide-mix (SEM-EDX spectra were dominated by Fe and O, but also showed other elements). The detected assemblages were generally similar between L4 and L7 profiles for magnetic fraction, but the L7 sample contained a higher proportion of Fe oxide and Fe oxide-mix phases, as compared to the Ti-Fe phase (ilmenite and titanomagnetite) that dominated in L4 sample. Also, 12% of particles showed Cu, Pb, Zn, As or Co signals in SEM-EDX spectra in L7, whereas only 7% of such phases were found in L4. The main difference for non-magnetic fraction was abundant apatite in agricultural soil L4, which was absent in the forest soil L7 sample. The phase most probably represented diverse fertilizers.

Some particles were monomineral (garnet, ilmenite, zircon, rutile, dolomite), but many of them were complex, and composed of several phases. The complex particles could be classified as natural or anthropogenic. The first group included rock fragments (mainly amphibolite), whereas the second group included slag fragments. Sulfides were rare and always showed signs of weathering. Dolomite was probably also related to mining and represented gangue rock or could represent a flux. This study was particularly focused on the particles that may have contained elevated contents of potentially toxic elements and they were considered also, in the context of their stability (i.e. is alteration visible or not). Such particles included Fe oxide and Fe oxide-mix, sulfides, slag fragments and Pb-rich silicate glass. Particles containing anthropogenic zircon were also observed.

Based on SEM analyses the Fe oxide-mix group can be further divided into three subgroups depending on the additional elements, which the particles contained: (1) approximately 50% of Fe-mix-oxides contained only elements such as Si, Al, Mg, Ca, (2) in 25% of particles, in addition to these elements sulfur was detected, and minor elements including potentially toxic ones may or may not have occurred, and (3) in 25% of particles, variable elements were detected depending on the phase (e.g. Cu, Pb, Zn, Co, V, Mn, Ni), but no sulfur.

This group is strongly diversified in terms of grain size, morphology, and composition (Fig. 2) with Fe always being the dominating component. In this group, we included phases with Fe [wt%] content exceeding 40% and divided them into two sub-groups based on chemical composition as analysed by electron microprobe: Fe oxides with high totals (mostly exceeding 90%) and Fe oxides with low totals (mostly within the range 65%–90%, Fig. 3A). The lower totals were most probably a combination of several factors such as the presence of hydroxyl group, Fe³⁺, or porous structure. This division did not reflect the real variability in Fe oxide phases that may include Fe oxides sensu stricto such as magnetite (Fe³⁺₂Fe²⁺O₄) and hematite (Fe³⁺₂O₃), as well as ferrihydrite (Fe³⁺₂O₃·0.5(H₂O)) and Fe oxyhydroxides (Fe³⁺O(OH) polymorphs such as goethite, lepidocrocite, feroxyhyte, and akaganeite). It is difficult to assign analyses of particular Fe oxides to a mineral name, firstly because they represent a mixture of different Fe phases, but also because admixtures of other elements (Si, Al) contaminate the analyses. The other elements could indicate the presence of unweathered minuscule silicates or silica and Al oxyhydroxides precipitates from secondary fluids. Alternatively, they could be components sorbed or formed within the Fe oxides after their deposition in soils. Therefore, a possible assumption is that the first group is mostly represented by magnetite and hematite, whereas the second group is dominated by various Fe oxyhydroxides and/or ferrihydrite. The first group contains spherical particles (Figs. 2A and 2F), angular monomineralic fragments (Figs. 2B, 2C, 2F), larger porous particles (Figs. 2B, 2C, 2G), and grains within complex slag fragments (Fig. 2F). The second group is mostly represented by anhedral, strongly porous grains (Figs. 2D, 2F, 2G). Uncommon occurrences include a Fe-rich glass-like material containing quartz grains (Fig. 2E). The two groups differ in chemical composition, the low-total group contains higher concentrations of other elements, most notably Si (Fig. 3B) compared to the high-total group, which is probably due to the presence of admixtures with sizes below the resolution of the microprobe beam. Also, metallic elements such as Pb and Cu occur usually above the detection limits of electron microprobe in the low-total group and are below detection in high-total Fe oxides (Figs. 3C and 3D). Based on electron microprobe analyses, sulfur was detected in 24 out of 41 analyzed low-total Fe oxide phases.

Figure 2.

Figure 3.

This group included angular fragments composed of several phases and it was strongly diversified in terms of phase composition and texture. Figure 4 shows examples of six slag fragments and each of them contains different phases with variable metallic elements contents. Spinels and silicates are common phases, but they differ in composition between the slag fragments; also, glassy material is strongly compositionally diverse (Figs. 4A and 4B and Supplementary Material). For example, Pb content in slag glass is either below the detection limit (white symbols in Figs. 4A–4C), reaches 10 wt% (red star, Fig. 4B), or is as high as 40 wt% (Fig. 4D). Some fragments do not contain Cu, Zn, and Pb, but show elevated contents of other elements such as V (Fig. 4A) or Ba (Fig. 4E). Also alteration features range from extensive (rims in Figs. 4A and 4E) to absent (Figs. 4B–4D).

Figure 4.

Pb-rich silicate glass occurs in both profiles and consists of silica (25–50 wt%), alumina (5–10 wt%), and Pb (41–68 wt% as PbO). It also contains Fe, K, Ca, and Na and traces of Mg, Ti, Zn, and Cu. The fragments show signs of alteration that form discontinuous rims with low Pb contents around the fragments and a similar reaction propagates along the fractures (Fig. 5A) or irregular tunnels (Fig. 5B). The Pb-poor rims reach the thickness of 25 μm in L7 compared to 15 μm in L4. Generally, the composition of the glass can be divided into three groups related to the intensity of the alteration. Group 1 is represented by the primary glass and is characterized by high Pb contents (>50 wt% PbO) and inverse correlation of Pb and silica. Closer to the rims, patches characterized by the uneven appearance and darker color in the Back Scattered Electron (BSE) image occur (Figs. 5A and 5B). They contain less Pb and more Si, Al, and K than the primary glass. The outermost rims are Pb-poor and have scattered Si and other element contents (Figs. 5C and 5D); some of the analyses also have low totals, suggesting a hydrated composition. The scattering is expected as evident from SEM compositional mapping (Figs. 5E–5G). Clearly, low-Pb rims are composed of several inter-mixed phases.

Figure 5.

Other particles containing metallic elements are sulfides (mostly chalcopyrite, bornite plus rare galena) and metals (predominately Fe, but alloys of other metals were also found e.g. Sn or tiny Pb droplets). As seen in Table 2 these phases are very rare (metal alloys were not detected by particle-by-particle counting), but they may contain high concentrations of metallic elements other than the main constituents. Chalcopyrite contains Ag, whereas bornite Pb, Zn, Co, and Ni. Iron alloys contain over 0.1 wt% of Ag and As (Table S2a and S2b in Supplementary Material). Sulfides are commonly surrounded by altered rims composed mainly of Fe oxides and silica (probably amorphous silica precipitate), but also containing other metallic elements such as Mn and Pb. The chalcopyrite grain shown in Fig. 6A is altered along rims and fractures, some Fe oxide patches are observed also within the grain, and precipitation of Fe-Mn oxides is observed along the fractures (Fig. 6A). Within the more porous weathered fragments other elements such as Pb, Co, As, P and Si were also detected (only EDX analyses exist for this particular grain). In contrast, the chalcopyrite shown in Fig. 6B is characterized by a thin rim composed mostly of Fe oxide, but also containing some W. As we used sodium polytungstate to separate heavy phases the liquid might have reacted with sulfide particles, this was probably the case for the grain in Fig. 6B. However, as other grains show extensive alteration with no W addition, we conclude that the majority of sulfides are altered by natural processes.

Figure 6.

Spherical particles and breccia-like fragments composed of fragmented zircon and Zr-rich glass (Fig. 6C) constitute approximately 1%–5% of the zircon particles found in the studied soils. The composition of this anthropogenic zircon is indistinguishable from the natural one analyzed in this study. The glass in breccia-like fragments is composed of SiO₂ (73 wt%), Na₂O, and Al₂O₃ (7–8 wt% each) and contains up to 5 wt% of ZrO₂, interestingly Ce contents above the detection limit were also measured (approx. 1500 ppm, Table S2b in Supplementary Material). The breccia-like fragments can be remnants of abrasive materials used in the smelter (e.g. Cho et al., 2008).

The main difference between agricultural and forest soil is the presence of abundant apatite grains in the former. The grains most probably represent a phosphate fertilizer and provide insight into its reactivity. Several types of apatites were noted with the dominance of oval, porous grains (Fig. 6D) or large angular fragments with no porosity. The oval grains were analyzed in this study showing Pb and Cu contents slightly above the detection for a few grains (Table S2b in Supplementary Material). Interestingly, high metal contents are limited to minuscule grains, usually observed within pores. It is unclear if these are newly precipitated or mechanically included.

As was noted by studies dealing with soils polluted by mining and smelting activities (see Table 1 and references within) the heavy minerals in soils may originate at different stages of mining and smelting procedure. Ettler et al. (2016) grouped the grains based on their size, phase composition, and morphology into five categories, which were related to diverse sources and similar sources may be envisaged for particles analyzed in this study: (1)

Particles related to mining operations or derived from mine tailings. They are angular, 5–150 μm in size, and composed of ore and gangue rock minerals (Ettler et al., 2016). In this study, sulfides and dolomite may be associated with this group.

Among the studied phases Fe oxides are the most common and not straightforward to classify. They could be derived from several smelting-related sources or could be formed within the soil as secondary minerals or precipitates. Some of them have a spherical morphology and low porosity and can be regarded as particles emitted from the smelter formed during rapid cooling of the liquid oxide (Fig. 2A). Similar particles were observed in soils surrounding the Zn-Pb smelters near Olkusz (Cabala & Teper, 2007). Some Fe oxides may occur in slag fragments (Fig. 4), which may be rich in primary magnetite, a common phase formed during rapid cooling and transformation processes within smelter and converter facilities, (Lanteigne et al., 2012). However, the majority of the particles have diverse sizes and irregular shapes (Figs. 4C and 4E). These Fe oxides are characterized by porous structures that may suggest that they represent slag fragments or sulfides that were oxidized to diverse Fe oxides when they were deposited in soils. This is consistent with experiments that simulated the reaction between slag and acidic water leading to the formation of Fe oxide and silica gel (Tyszka et al., 2021). We suggest that similar products of intense weathering of smelter-derived particles are common in the studied soils. However, particles of Fe oxide with low totals may also have been formed naturally in soils by precipitation and it is difficult to distinguish them from the products of slag and sulfide weathering.

The analyzed assemblage may be compared to phases emitted recently from the Legnica smelter that were deposited on spider webs (Trzyna et al., 2022). Anthropogenic silicates were observed both in our study and that of Trzyna et al. (2022). The authors noted also a high proportion of sulfides in fractions <2.5 μm and the general absence of sulfides in fractions >10 μm. The sulfides were rare in our study, however, we analyzed mostly fractions >10 μm. The sulfur signal was detected in at least 50% of Fe oxides with low totals, which may indicate that the sulfides were commonly present in the soil, but were weathered and replaced by Fe oxides. However, the adsorption of sulfates present in soils cannot be excluded. Also, Fe oxides were rare in the study of Trzyna et al. (2022) and attributed mostly to erosion of local constructions (rail tracks, industrial metal structures). The abundance of such particles in the soils and their general absence in recent air-deposited material may favor their precipitation in the soils. However, the differences between phases from recent air deposition and those in soils may be also due to the change in the phase composition of emitted material with the soil containing particles emitted over 70 years of the smelter activity. Summarizing, the origin of Fe oxides is complex and possible origins include anthropogenic particles, secondary products after anthropogenic particles weathering and soil precipitates. Establishing more exact proportions between the particles of diverse origin requires more detailed in-situ analyses of phases coexisting within a single Fe oxide particle.

So far, heavy minerals in polluted soils have been used to detect possible weathering reactions and estimate the stability and mobility of metallic elements in soils affected by mining and smelting (Table 1). A similar approach can be applied to particles analysed in this study. For example, we observed that Pb-rich silicate glass is readily weathered potentially releasing a considerable amount of Pb. On the other hand, sulfides are characterized by more complex reactions. Some of them are slightly weathered (Fig. 6B) and some show more extensive weathering (Fig. 6A). The experiments performed on Cu-slags showed preferential weathering of glass compared to sulfides in alkaline conditions (pH >7, Potysz et al., 2018). However, the soils around Legnica are slightly acidic (pH 5.6–6.5) to near-neutral (6.6–7.2) and not alkaline (Hołtra & Zamorska-Wojdyła, 2020), which should not promote extensive glass weathering and also stabilize sulfides. Therefore, the range of observed weathering features probably reflects the complex nature of the soils. As anthropogenic material has been released from the smelter for over 70 years, it probably weathered at different rates and conditions at different times.

As noted above Fe oxides may be a product of extensive weathering and they may also contain metallic elements such as Pb, Cu, and Zn. Tyszka et al. (2021) showed that secondary Fe oxides may rapidly form in slag material in experimental conditions under acidic pH (simulated acid rain conditions). Extensive weathering is not expected in near-neutral soil, but the particles may have been already weathered and transported from slag disposal sites or be affected by fluctuating pH e.g. after heavy rains with more acidic pH. Legnica lies in the area, where pH in the period 2002–2019 is generally low, in the range of 4.8–5.2, compared to other regions in Poland (https://powietrze.gios.gov.pl/pjp/maps/chemistry/precipation). The longer timespan of pH measurements can be accessed for the Śnieżka mountain from the European Monitoring and Evaluation Programme (Tørseth et al., 2012). It is evident that the precipitation was more acidic with pH values often below 4.0 in the early 90’s and it is possible that even lower values were noted before, in the period commonly affected by acid rains in Poland. Such a situation may be compared to the Świętochłowice slag dump, which contained high amounts of fine-grained weathered material composed mostly of Fe oxides formed after long-term deposition of slag affected by recurring acid rains (Tyszka et al., 2014). The elevated contents of elements such as Cu, Pb, and Zn are probably due to the presence of secondary phases that trap these elements or minuscule remnants of primary phases (sulfides). Unfortunately, the resolution of both SEM-EDX and electron microprobe used in this study is too low to recognize individual phases in complex Fe oxides. Further research by field emission SEM may give better characteristics of the phases and their potential mobility. However, despite the difficulties of assigning elevated contents of metallic elements to a particular phase, the complex Fe oxides must be considered as the main carrier of smelter-derived pollution currently residing in the studied soils.

This study and others shown in Table 1 confirm the usefulness of heavy minerals to characterize pollution sources and the fate of the particles derived from mining and smelting. However, the vital information on the general pollution in the area should assign metallic elements to particular phases in the context of the whole soil metallic elements budget. This can be done if the weight proportion of heavy mineral fraction in the soil sample is estimated. A good example of quantitative studies done in a similar context of anthropogenic particles accumulated in soils and sediments is research on glass microspheres. It was shown that the number of microspheres differs between their depositional sites that delineated the routes of their preferential transport (Gałuszka & Migaszewski, 2018; Migaszewski et al., 2022). These conclusions were only possible because the total number of microspheres was calculated per kilogram of the sample. Such an approach was not applied in this study, but rough estimates can be done based on approximate modal proportions and the number of analyzed grains. For example, assuming that Pb-rich glass was fully separated within a non-magnetic fraction and most of the fraction was mounted, we expect 10–15 grains of Pb-rich glass per 500 g of material used for heavy mineral separation. Further assuming a spherical shape and 100 μm radius of the particles as well as Pb content of 50 wt% and density around 5 g/cm³ (estimated after Kacem et al., 2017), it can be calculated that <0.5% of the total Pb budget is contained in these particles. On the other hand, Fe oxide particles are much more abundant (estimated at several thousand particles per 500 g soil) and they may contain >1% of the total Pb. Taking into account that these are probably minimal estimates the calculations show that heavy minerals may contain an important part of the metallic elements budget, but time-consuming methods are required to estimate it properly.