1 Introduction

Mining activity in Tunisia started with Carthaginians and Romans (Sainfeld 1952). About 600 mineral occurrences are known in the Northern part of the country; about 50 deposits were exploited, mainly from the end of the nineteenth century. The main deposits are hosted within upper cretaceous to tertiary sedimentary rocks, and consist of veins, impregnations, and replacement masses. The element association includes Pb–Zn–Fe–Ba–F, with subordinate Cd, As, Sb, and others; gangue minerals include calcite and clays. Total production adds up to 55 Mt Fe ore, 2.3 Mt Pb ore, and 2 Mt Zn ore (DGM 2005).

After mine closure, several million tonnes of flotation tailings remain confined within dams. They still contain environmentally significant amounts of potentially toxic elements (PTE), and represent a potential hazard to human health, being located close to human settlements and amidst farming lands. Several studies were devoted to the environmental aspects of these tailings (Souissi et al. 2008, 2013, 2015; Chakroun et al. 2013; Othmani et al. 2013, 2014, 2015; Daldoul et al. 2015), but comparatively little attention was paid to semimetals, such as As and Sb. These highly toxic elements might be overlooked in mining areas, where they were not the main object of exploitation, but quite often they are of relevance for the environmental impact (e.g., Benzaazoua 1996; Casiot et al. 2007; Ashley et al. 2007; Razo et al. 2004; De Gregori et al. 2003; Hiller et al. 2012; Souissi et al. 2008, 2013).

This study deals with the former Pb–Zn mining sites of Jalta (Ja, 1903–1951), Jebel Hallouf-Sidi Bouaouane (JH-SBA, 1892–1986), and Touiref (TO, 1891–1959), located in Northern Tunisia (Fig. 1). The area is characterized by a temperate, sub humid climate [annual rainfall ranging from 333 to 1011 mm for the decade 2001–2010, with minimum in July (2–8 mm) and maximum in January (76–117 mm); average monthly temperatures from 8 to 11 °C in December through February, to 27–28 °C in July–August; INM 2011].

Fig. 1
figure 1

Location map of the study area [adapted from Sainfeld (1952); Perthuisot (1978), by Souissi et al. (2013)]

Full size image

Ore deposits were dominated by Pb–Zn–Ba–F concentrations, hosted mainly in carbonate (limestone and dolostone) rocks, which may locally be accompanied by lesser amounts of silica, clay, gypsum, and others. Ores were extracted in open pits or underground workings. Ore processing was first based on gravimetric separation and then on flotation techniques. The latter gave rise to most of the mine tailings as they are known today, because large amounts of the former gravimetry tailings were reprocessed by flotation.

This paper reports the results of a mineralogical and geochemical study of flotation tailings at these sites. The mobility of heavy metals (Zn, Pb, and Cd) and semi-metals (As and Sb) is estimated by a static leaching test in deionized water, simulating the interaction of wastes with rain and/or surface water. The results contribute to a full assessment of environmental hazards associated with abandoned mine wastes in carbonate settings of Northern Tunisia. We emphasize the attention given to the semi-metals As and Sb.

2 Materials and methods

2.1 Sampling

As specified in Table 1, several samples of 2 kg each were collected at regular intervals from a total of nine sampling points in tailing dams of the three localities.

Table 1 Characteristics of the studied samples
Full size table

Most samples were taken at 30 cm depth; however, at Touiref, it was possible to take samples at different depths corresponding to visual changes in the appearance of the material, indicating the transition from a yellowish oxidized zone (ZO) to an intermediate yellowish-grey zone (ZT) to a dark grey sulfide-bearing zone (ZS). At Jebel Hallouf, in addition to the surficial (30 cm) samples, it was possible to drill a core down to a depth of 5.75 m. Because of macroscopic homogeneity in texture and colour, five subsamples were collected from this core at ~1 m intervals, and combined into a single sample. In the laboratory, samples were dried first in open air and then in oven (45 °C for 24 h). Samples of each sampling point were then carefully mixed, homogenized, and quartered, to obtain one representative sample for each sampling point. This sampling strategy, dictated by practical limitations of the following analytical steps, does not allow recognition of subtle field scale variations, but gives for each sampling site an estimate of the overall characteristics and potential impact.

Grain-size analysis, performed at INRAP (Sidi Thabet, Tunisia) by means of a Mastersizer 2000 laser instrument, pointed out the fine-grained nature of all materials (see below); all subsequent analyses were performed on aliquots of the bulk homogenized samples, with no size distinction.

2.2 Analytical methods

Mineralogical analyses (X-ray power diffraction) were performed at the University of Cagliari by means of a Panalytical X’pert Pro instrument, and complemented by scanning electron microscopy (SEM) observation with semi-quantitative EDS microanalysis at MEMA (University of Firenze, Italy; ZEISS EVO instrument with Oxford Linkam microanalysis). Chemical analyses for major elements were performed by X-ray fluorescence (Rigaku Primus II spectrometer, CRIST, University of Firenze; raw data correction by an in-house standardless ab initio procedure). The procedure is optimized for ordinary rocks, and it is not well suited for samples with unusual composition, such as mine tailings. Moreover, while most major elements are conventionally reported as oxides, for others, there is indetermination on the chemical state (e.g., sulfur occurs both as sulfide and as sulfate). For these reasons, major element analysis is affected by an uncertainty that was estimated in the order of 10–20%.

The loss on ignition (LOI) was based on differential thermal analysis in a muffle furnace. Tailing samples begin to lose volatiles at about 110 °C; the devolatilization process was complete at about 950 °C. Calculation of LOI is based on the difference in weight of sample before and after ignition.

For an accurate determination of selected trace elements (Zn, Pb, Cd, As, and Sb) at the University of Cagliari, samples were first reacted in a microwave oven (Milestone ETHOS 1) with an HNO3–H2O2–HF mixture. The resulting solutions were analyzed by inductively coupled plasma optical emission spectroscopy (ICP-OES; ARL Fisons 3520B instrument); the reproducibility of analysis was checked by running several duplicates and the accuracy by analysis of the certified standard SRM 2710. Both reproducibility and accuracy are within 5%.

In the absence of a specific Tunisian regulation, leaching tests were performed according to Italian environmental law (DL 152/2006). The test is based on the UNI 10802 (2013; www.uni.com) procedure, and consists in the static treatment for 24 h by deionized water of the material (solid/liquid ratio 1:10) under agitation at 150 RPM. The leachates were analyzed by ICP-OES for Zn, Pb, and Cd, and by ICP mass spectrometry (ICP-MS; Perkin Elmer ELAN DRC-e instrument) for As and Sb.

3 Results and discussion

3.1 Grain-size distribution

As previously mentioned, all collected materials are fine-grained (100% <2 mm; >85% <0.5 mm). Specifically, the mean diameters (d 50) are comprised between 19.5 and 169.7 μm (Table 1); the dominant fraction can be classified as either fine sand or very fine sand, except two samples from Jebel Hallouf–Sidi Bouaouane, where the main fraction can be described as clayey silt. The fine-grained nature of these tailings testifies to a protracted grinding procedure to achieve the best liberation of the economic phases. On the other hand, this fine-grained size enhances the reactivity of the tailings, favoring the mobility of PTE (see Brill and Floc’h 2001).

3.2 Mineralogy

Table 2 shows the mineralogical composition of the nine samples as a result of the XRD patterns and SEM–EDS observations. In all samples, calcite is the most abundant mineral, in agreement with the calcareous nature of the host rocks. Primary sulfides, especially arsenopyrite and sphalerite, are always present, whereas galena and pyrite are comparatively less widespread. The widespread occurrence of jordanite, Pb14(As, Sb)6S23, is remarkable. The mineral was identified both by the presence of its strongest reflection (d = 2.24 Å) in the XRD patterns, and by semiquantitative SEM–EDS microanalyses. Jordanite had previously been reported in these deposits by Rouvier (1971); Jemmali et al. (2011, 2013, 2014), and Souissi et al. (2013).

Table 2 Results of the XRD semi-quantitative analysis
Full size table

Secondary minerals are represented mainly by carbonates (cerussite and smithsonite) and gypsum, but we point out the presence of mimetite, Pb5(AsO4)3Cl, and Sb oxides evidenced by SEM images (Fig. 2e). In the Touiref samples collected at different depths, described as “oxidized”, “intermediate”, and “sulfide”, there are no significant changes in the overall mineralogical composition, except for the presence of galena limited to the “sulfide” zone. Figure 2 reports some SEM photomicrographs portraying the occurrence and textural relationships of some of the observed minerals.

Fig. 2
figure 2

SEM photomicrographs of flotation tailings from the study areas. Sample TO a smithsonite (Sm) grains in a gangue of calcite (Ca); b sphalerite (Sp); and c associated grains of galena (G), sphalerite (Sp), and jordanite (J). Sample JHD1 d jordanite; notice the dark rim of (presumably) Sb-oxides e associated grains of galena (G), gypsum (Gy), and pyrite (P); f associated mimetite (M) and iron oxyhydroxides (FeO). Sample JA g galena (G) and cerussite (Ce); h grain with associated galena (G) and barite (B)

Full size image

3.3 Tailing composition

The major element contents of the studied samples are reported in Table 3. The analyses, showing the dominance of Ca (38–52%), an important loss on ignition (LOI; 25–40%), and a subordinate presence of Si, Al, Fe, and Mg are in agreement with mineralogical studies (abundance of calcite, minor quartz, Fe-oxyhydroxides, dolomite, and traces of silicates). Table 3 also includes the contents in heavy metals (Pb, Zn, and Cd) and semimetals (As and Sb) in the tailings. The contents of these elements are consistently of the same orders of magnitude in all samples, with just moderate differences among different localities: values for Pb, As, and Sb are higher at Jebel Hallouf and Sidi Bouaouane; this last deposit also shows, along with Touiref, the highest contents of Zn and Cd.

Table 3 Major and trace element contents of the representative tailing samples
Full size table

The abundance of Zn (6.9–13.7 g kg−1) and Pb (2.9–9.6 g kg−1) corresponds to the presence of sphalerite, smithsonite, galena, and cerussite; some Pb also occurs in jordanite and mimetite. These last two minerals are probably also the main carriers of arsenic (139–837 mg kg−1), as suggested by the strong positive correlation between As and Pb (R 2 = 0.82, p = 0.0008); on the other hand, As is also hosted by arsenopyrite and realgar, and probably also occurs in association with Fe oxyhydroxides; correlation between As and Fe is indeed significant (R 2 = 0.69, p = 0.006). Jordanite is likely the main phase hosting Sb (up to 102 mg kg−1), as indicated by the strong correlation of Sb with As (R 2 = 0.84, p = 0.0005) and Pb (R 2 = 0.93, p < 0.0001), whereas Cd (31–106 mg kg−1) most probably substitutes for Zn in its minerals (as shown, for this district, by Souissi 1987, 2007; Ghorbel et al. 2010). This idea is reinforced by the high degree of positive correlation between Zn and Cd (R 2 = 0.75, p = 0.003).

These contents of potentially toxic elements (PTE) are obviously several orders of magnitude higher than average crustal abundances 6, and also exceed the limits established by many countries for soils (Table 3 reports the Italian limits for industrial soils as defined by DL152/2006).

3.4 Leaching tests

The composition of the solutions obtained from the leaching tests is reported in Table 4.

Table 4 Composition of the solutions obtained from the leaching tests
Full size table

As expected from the calcite-dominated mineralogy, the pH of the leachates is subalkaline (7.6–8.0). The quality of leachates is otherwise rather poor if compared with international standards for waters. All relevant chemical parameters exceed the limits recommended for drinking water by the World Health Organization (WHO 2011), and most values also exceed the limits established by many countries for wastewater discharge (Table 3 reports the Italian limits according to DL152/2006).

In general, there is no correlation at a significant level (p > 0.05) between total contents of the elements in the solid wastes and corresponding contents in the leachates, i.e., the mobile fraction is different from one sample to another. The only exception is for Sb. For the five samples, where Sb concentrations in the leachates could be measured, there is a significant correlation (R 2 = 0.95, p = 0.005) between Sb contents in the leachates and in the solid wastes, in agreement with the dominance in all samples of a single Sb-hosting phase (jordanite).

The considered elements show different degrees of mobility in the conditions of the test. This difference can be quantified by introducing a parameter M defined as the ratio between the concentration of the element (in mg kg−1) in the solid and that (in g L−1) in the corresponding leachate; the lower the value of M, the more mobile the element (for the specific test conditions).

From Table 5, it can be seen that mobility of Zn and Sb is higher than that of As, Cd, and Pb. The difference between the two groups (Zn–Sb vs. As–Cd–Pb) is statistically significant (Mann–Whitney U test, p < 0.01), whereas intra-group (i.e., Zn vs. Sb, and As vs. Cd and Pb) differences are not significant (p > 0.05). A high mobility of Zn compared to Pb, and a comparatively high mobility of semimetals, that at subalkaline pH dominantly occurs in anionic form, is a typical feature of net alkaline mine drainages (NAMD). On the other hand, Cd is less mobile than Zn, in spite of the similar geochemical behavior of the two elements, and of a strong correlation (R 2 = 0.86, p = 0.0005) between their contents in the leachates. This mobility difference was observed in other systems (e.g., Frau et al. 2015), and was ascribed to the occurrence of highly soluble secondary Zn sulfates, whereas Cd preferentially occurs in comparatively less soluble minerals, such as carbonates (e.g., Lattanzi et al. 1998). Indeed, the correlation between Zn and sulfate in the leachates is significant (R 2 = 0.79, p = 0.002); the occurrence of comparatively soluble sulfates (including gypsum) in tailings is demonstrated by the high amount of sulfates in the leachates.

Table 5 Mobility ratio (M: see text for definition) for the analyzed PTE
Full size table

In our samples, arsenic is apparently less mobile than Sb. In mine environments, As may be less or more mobile than Sb depending on the specific conditions (e.g., Ashley et al. 2003; Casiot et al. 2007), even if a higher mobility of As is more commonly reported (e.g., Flynn et al. 2003; Ettler et al. 2010; Majzlan et al. 2011). Both As and Sb can be, at least temporarily, adsorbed onto secondary Fe-oxyhydroxides, but this mechanism seems more effective for Sb than for As (Majzlan et al. 2011). In their study of materials from the Hillgrove mine, New South Wales, Ashley et al. (2003) found that in stream sediments, mobility of Sb is higher than that of As, whereas leaching of stibnite ore concentrate released Sb and As in ratios consistent with those in the starting solid. The authors suggested that Sb mobility is essentially controlled by the rapid formation of secondary Sb oxides. In our case, we can speculate that As mobility is comparatively low, because much of this element is either still contained in primary arsenopyrite, or else it occurs in an insoluble phase, such as mimetite. This low mobility of a toxic element, such as As, is obviously good news, but, while mimetite can be considered a long-term storage of this element (e.g., Flis et al. 2011), oxidation of arsenopyrite may eventually release As into the environment, and should be monitored.

Considering the tailings of Touiref, it is possible to follow the behavior of PTE in depth according to the different zones (oxidized zone, intermediate zone, and sulfide or fresh zone). Results show (Table 5) that As and Zn have the same geochemical behavior: their mobility increases gradually with depth to reach the mobility ratio M = 5.25 and M = 0.36, respectively, in the fresh zone, while the mobility of Cd decreases with depth. The variation of Pb mobility is irregular; the highest mobility is recorded within the fresh zone.

4 Conclusions

Fine-grained waste materials in the former Pb–Zn mining districts of Northern Tunisia contain environmentally significant amounts of potentially toxic elements (PTE), such as Zn, Pb, Cd, As, and Sb. Mineralogical studies identified as PTE carriers’ primary sulfides (galena, sphalerite, pyrite, chalcopyrite, and arsenopyrite) and sulfosalts (jordanite), and secondary oxidized minerals (cerussite, smithsonite, and Sb-oxides), accompanied by calcite, sulfates (gypsum and barite), and clay minerals. Because of the fine-grained nature of wastes, wind deflation and/or runoff may represent non-negligible pathways for PTE dispersion.

Leaching tests with deionized water indicate a near-neutral reaction, accompanied by a comparatively high release of PTE. These values are well above the recommended WHO limits for drinking waters, and suggest that interaction with rain and/or surface waters will produce contaminated drainage downstream mine. A specific risk is the mobilization, at the beginning of the rainy season, of comparatively soluble salts accumulated during the dry season (see Canovas et al. 2010). Moreover, the mobility of As, which is currently comparatively low, might increase in the long term because of oxidation of residual arsenopyrite.

Therefore, these wastes represent a potential threat for the environment and for the nearby human settlements. An adequate monitoring plan is advisable to exactly quantify the actual impact on environment and health. The results of monitoring could indicate the appropriate remediation measures to be undertaken.