*4.2.1 Sample preparation and analysis*

Representative sediment samples (from impacted and non-impacted sediments in 2015), polluted soils and tailings were transported in hermetically sealed plastic containers and dried at 40°C for 48 h, ground, sieved (mesh #10, < 2 mm) and homogenized by quartering [140] Portions of 100 g were re-milled (Fritsch ball grinder), sieved (mesh #200 < 74 μm), and dried at 96°C for chemical analysis. The samples were preserved at room temperature (20°C) in hermetic containers [141]. All analyses were conducted by LABQA-UNAM (Accredited Laboratory No. R-0593-031/14 by *Entidad Mexicana de Acreditación*). Analytical reproducibility was inspected following the laboratory's QA/QC analytical procedure, using spike samples and certified international reference material, and preparing blanks. All analyses were duplicates, all reagents were analytical grade or high purity, and the water was ultrapure deionized (Nanopure).

Total concentration was measured through X-ray fluorescence (XRF) with a portable model DP-6000-CC Thermo Scientific XRF Olympus analyzer, used following the 6200-method [142]. Sequential Extraction was performed using a modified Tessier et al. [6] method. The procedure consisted of five successive extractions: Fraction I (F1): Exchangeable (1 M MgCl2, pH 7, shaken 250 rpm, 1 h at room temperature (19–23°C)); Fraction II (F2): Carbonates (1 M CH3COO�/CH3COOH buffer, pH 5, shaken 250 rpm, 5 h at room temperature); Fraction III (F3): Fe/Mn Oxides (0.3 M, Na2S2O4 + 0.025 M, citric acid+0.175 M, sodium citrate, shaken at

250 rpm, 5 h at 96 3°C); Fraction IV (F4) Organic matter/sulfides: (3 mL [HNO3] + 5 mL 30% H2O2, pH 2, shaken 250 rpm, 5 h, at 85 5°C); and Fraction V (F5): Residual phase, total concentration minus the sum of fractions I-IV. Oral gastric bio-accessibility was determined with method NOM-147 [143], analogous to SBET, RBALP, and SBRC [144]. Solid samples were mixed with glycine (C2H5NO2) at pH 1.50 0.05, reached with HCl 1 M, in a ratio (1 g:100 mL); shaken 1 h at 30 rpm, controlling the temperature with an immersion recirculation heater at 37 2°C. Concentrations were measured in accordance with ICP-OES EPA 6010 [145].

### *4.2.2 Statistical analysis*

The Mann–Whitney U test was applied to test the null hypothesis in this work, that is, there will be no statistically significant differences in metal(loid) concentrations in groups of sediments, sediments and polluted soils, or sediments and tailings. This function takes two data samples as parameters, uses the median as a measure of central tendency, and then sends the results with a p-value showing the statistical significance. All analyses use a significance level of p = 0.05. If the p-value ≤0.05, the conclusion is to reject the null hypothesis and to accept a difference between the ranks of the two groups (sediments, soil, tailings).

### **4.3 Results and discussion**

**Tables 3–5** present the results of the sequential extraction and total concentrations of three representative metal(loids) from the acid spill: As, Cu, and Fe. No anionic sequential extraction for As was made, since the results are similar to those obtained with cationic sequential extraction [134]. The recovered fractions were: Exchangeable (I); Carbonates (II); Fe and Mn-oxide/hydroxides (III); Organic Matter/Sulfides (IV); and Residual (V). **Figures 4**–**7** show the recovery percentages for the three representative elements in each fraction, and **Figure 4** presents the recovery fractions in wastes and polluted soils. There are no statistically significant differences in total concentrations of As, Cu, or Fe, between impacted and baseline sediments (p values = 0.19, 0.21, and 0.07 respectively). Indeed, total As and Fe concentrations means in impacted sediments were slightly lower than those of the baseline sediments (**Tables 3**–**5**). On the other hand, sequential extraction does provide valuable information: As is the only element for which recovery in F3-fraction was very significant, Cu was recovered in different fractions, including F1, and Fe was retrieved from impacted sediments, predominately in fraction F5 followed by F3 (**Tables 3**–**5**). The dominant As fractions were F3 (linked to Mn/Fe-oxides) and F5 (residual). F3 distribution was variable along the river, with higher values in the backwaters where sediments have more easily precipitated (**Figure 4**). The differences between impacted and baseline sediments were only statistically significant for Cu in F1-fraction (p = 1.6 <sup>10</sup><sup>2</sup> ), and for Fe in F3 (p = 3.5 <sup>10</sup><sup>3</sup> ) and F5 (p = 1.4 <sup>10</sup><sup>3</sup> ). Therefore, Fe can also be used as a tracer of the impact of the acid solution spill, although the differences between impacted and baseline sediments are most evident in the case of As. In samples of non-impacted sediments, fraction F5 equaled 98.95%, while in impacted sediments an important percentage of As concentration belonged to F3-fraction. Significant differences between baseline and impacted sediments were statistically proved (F3, p = 3.4 <sup>10</sup><sup>5</sup> and F5, p = 3.2 <sup>10</sup><sup>5</sup> ) (**Figure 4**). Significant differences were also found when statistically comparing fraction F3 of impacted sediments with the same fraction of tailings and polluted soil not affected by the spill


**Table 3.** *As Sequential extraction and total concentration.*


**Table 4.** *Cu*

 *Sequential extraction and total concentration.*



**Figure 4.** *Sequential extraction for As.*

**Figure 5.** *Sequential extraction for Cu.*

**Figure 6.** *Sequential extraction for Fe.*

**Figure 7.** *Sequential extraction for As in polluted soils and tailings.*

(p = 2.5 � <sup>10</sup>�<sup>4</sup> and p = 1.6 � <sup>10</sup>�<sup>3</sup> , respectively). This behavior indicates that they mainly contain arsenopyrite in relatively high concentrations, which is the most reported As-mineral in the area [146] and recovered in the F5-residual fraction [147]. After sediment 24, in samples 25 and 26, practically all of the As was recovered in F5 fraction. This behavior was also observed in the following samples (data not shown), indicating no acid solution impact downstream at those sites. The high As recovery in fraction F3 of impacted sediments is mainly a consequence of the chemical changes that took place between the waste rock and the dam (the leaching process), and likely in the river as well. The acid solution spilled from the dam was a lixiviate formed during the pretreatment of rock waste deposits. Those low-grade Cu minerals were doused with a weak sulfuric acid solution, to destroy the basic minerals occurring naturally in waste rock and that partially neutralize the added sulfuric acid. This process reduces future Cu leaching. In the waste rock, most of the As was in the form of arsenopyrite (FeAsS) and was possibly present in lower concentrations as scorodite [147]. Additionally, traces of As2O3 have been reported in the waste rock deposit. Impure sulfuric acid is added to waste rock to boost the microbiological oxidation of Sminerals. Sulfur oxidation increases Cu recuperation from Chalcocite and Chalcopyrite. Under acidic conditions, AsIII can be partially oxidized to arsenates by the MnO2 from waste rock [148]. Arsenates over pH 2 lose H<sup>+</sup> -ions, forming negatively charged species which could be retained on jarosite by sulfate substitution and/or forming inner and outer sphere complexes [149]. They can also be sorbed by schwertmannite, amorphous ferrihydrite, maghemite and goethite [150]. These As retention processes could likely happen at the dam and or in the river when basic materials added to the water lead to the formation of amorphous Fe compounds with high sorption capacity. The arsenates sorbed onto Fe compounds were recovered in F3-fraction due to FeIII-reduction. Non-oxidized AsIII (arsenopyrite) was recovered on the F5-fraction, as Ankan and Schreiber [147] also observed. From sample sites 23 and up, As was recovered in F5. Although only sediments 24 and 26 are shown in **Figure 4**, the same behavior was observed for the rest of the analyzed sediments (data not presented). Thus, the segment of the river impacted by the acid solution spill was no greater than 30 km, an area significantly smaller than what was initially considered before remedial action was taken (190 km), indicating that control actions were effective. Although total concentration could not be used as a guide for the impacted area, the results show that sequential extraction allowed the distinction of the impact zone from other anthropogenically polluted materials containing natural minerals. The As in fraction F3 was the best tracer.
