**3. Analysis of results**

As show in **Figure 1**, the information found, indicates that the evaluated technologies still need to mature, as no evidence of their use at higher TRLs were found. The evaluated studies only included laboratory or prototype applications. However, there is indication that reverse osmosis and ion exchange resins have higher TRLs, as water treatment and purification companies already offer the products at a commercial scale [35].

**Table 2** shows the big diversity in the information retrieved from the research papers, with important variations in the experimental conditions, especially for maximum concentrations and pH ranges. However, variations in efficiencies, and temperature, were lower. Considering commercial or water treatment plant applications, the removal efficiencies presented are adequate; however, some pHs and temperatures are not at environmental conditions, requiring additional expenses.

The range of maximum initial sulfate concentrations in the analyzed samples vary widely ranging from 25,000 mg L−1 in a case using reverse osmosis, 8,000 mg L−1 with CESR, and 5,000 mg L−1 for nanofiltration. On the other end, there were also cases with concentrations below the limits established by international standards (250 mg L−1). Regarding this last point, the studies carried out by adsorption, which are the majority, work with concentrations as low as 20 mg L−1 reporting a removal above 90%. The problem of working with such low concentrations is that they already are below the limit of permitted concentrations. In addition, when up scaling the process to higher concentrations and full-scale applications, the margin of error would increase considerably and probably the system will be saturated easily. Studies should consider sulfate concentrations similar to those found in natural effluents, for example, 750 mg L−1 obtained in the city of Puebla, Mexico [36] or in the rivers in western Canada with records up to 3040 mg L−1, or the sea water that contains around 2700 mg L−1 [7, 11].

Most treatments report removal rates of more than 72%, except for Hong and collaborators [27], where the removal efficiency was 43% using activated carbon (polypyrrole-tailored). Nanofiltration and reverse osmosis had the highest removal rates, with values above 97%. However, it is important to consider some advantages and disadvantages of these systems. For example, for nanofiltration, which removes up to 97% of the sulfates present in water, it is necessary to include pretreatments, and increasing the pressure to values between 40 and 80 psi depending on the studies conditions. This pumping process increases the cost, since it implies higher energy consumption, and at the same time reduces the half-life of the membrane [11]. Similarly, reverse osmosis removes up to 98% (with samples that exceed 6,000 mg L−1); however, it needs pretreatment such as chlorination, softening and/or ion exchange, to avoid flakes formation by reactions on the walls of the membranes [24].

Although not all the reports indicate a required pretreatment, adsorption might also require the modification of the pH to increase the removal efficiency, depending on the pollutant to be removed. According to the methodology and reagents used, adsorption has been used in acidic or alkaline conditions, giving better results with acidic pH. Similarly, temperature is a parameter that directly affects the efficiency of removal by adsorption. In both cases, when pH and temperature modifications are required, removal efficiency costs increase [26].

For biological processes, although removal percentages are above 72%, response time and environmental conditions must be considered. For example, O´Sullivan and collaborators reported a retention time of more than 30 days [22]. However, these processes are among the most economical, as they show, not require temperature changes, work naturally, and typically operate at a neutral pH in the case of constructed wetlands. Regarding the bioprocess, constructed wetlands differs from bioreactors, because the former requires a large land and therefore a high initial investment.

In **Table 2**, the pH and temperature are reported whenever they are present in the study evaluated, as well as the concentration of the samples studied and the percentages of sulfate removal. In addition, it is also important to identify the presence of other components in the water samples, such as competitive ions, that might affect the removal efficiency of sulfates from water. Although Bowel and collaborators mention an extra advantage of their method, because nitrate can be removed together with sulfates during the ettringite precipitation, unfortunately they do not show any results about it [13].

Before introducing a particular method to remove sulphates from water, it is necessary to have all the information previously mentioned for the correct characterization of the samples to be treated, as well as defining their origin. It is important to differentiate the analysis of water from natural sources than the studies using synthetic water, because the natural water contains different factors, which cannot be controlled as with synthetic water. For example, the presence of other anions will compete with sulfates during the adsorption process; also, when using membranes, the presence of calcium sulfate directly affects the generation of flakes, clogging the pores of the membrane [11].

Finally, there are notable differences in the adsorption capacity of evaluated adsorption studies. For example, Rumjit and collaborators [30], with Biochar, reported a maximum adsorption capacity of 153.85 mg g−1, while Rahmati and collaborators [26] reported a value of 9 mg g−1 with activated carbon (see **Figure 2**). Rumjit and collaborators [30], however, reported a pH of 9.8 at room temperature, but no record of the surface area of the biochar tested. On the other hand, Rahmati et al, reported the surface area of their material of 51 m<sup>2</sup> g−1 working at pH 6 and temperature of 45°C but has less removal capacity [26]. Furthermore, Rahmati, with a lower adsorption capacity, reports a higher surface area in the material tested than Ao and collaborators [31] (21.7 m2 g−1), which has the second highest removal capacity from the studies reviewed with 35.2 mg g−1 [31].

Only one of the three biochar studies reports its surface area, which is much lower than the one reported for the activated carbon, where surface areas of up to 1100 m<sup>2</sup> g−1 are reached in at least two research articles. Of the nine articles reported with adsorption, 55.6% do not report surface area data, 22% do not record the data of maximum adsorption capacity, and those that do, present values at different conditions.
