Concurrent magnesium and boron extraction from natural lithium brine and its optimization by response surface methodology

Highlights

• A new system for simultaneous extraction of impurities from natural lithium brine was disclosed.

• A dialkylphosphate ion liquid dissolved in 2-ethylhexanol/kerosene was used as a dual extraction system.

• The extraction capacity of the concurrent system was greater than 99% for both Mg and B.

• The dual extraction process was optimized and validated using an experimental design known as response surface methodology.

Abstract

The high demand of lithium for electromobility and energy storage requires an efficient and sustainable optimization of the current extraction from natural lithium brine process. In this context, this study develops a simultaneous Mg²⁺ and B³⁺ extraction system, which could be a contribution to the elimination of precipitation stages and the generation of high purity lithium brine. This dual system composed of a dialkylphosphate ionic liquid and a 2-ethylhexanol/kerosene mixture as magnesium and boron extractors, respectively, was optimized by a mathematical and statistical method known as response surface methodology (RSM). The RSM method identified, through a small number of experiments, the interdependencies of the parameters involved in the extraction process, and calculated their optimal values to achieve the maximum extraction efficiency of Mg²⁺ and B³⁺. The optimization was performed from two approaches. The first, maximizing the extraction of Mg²⁺ and B³⁺, resulted in experimental efficiencies of 99.17% of Mg²⁺, 99.36% of B³⁺ and 23.54% of Li⁺; whereas the second, in which lithium co-extraction was minimized, yielded efficiencies of 83.56% Mg²⁺, 99.22% B³⁺ and 3.36% Li⁺, results that demonstrate the high extraction capacity of the system, as well as proving the validity of the statistical model.

Introduction

Non-Conventional Renewable Energy and electromobility have emerged as substitutes for fossil fuels worldwide. In this regard, lithium-ion batteries (LIBs) have taken on a fundamental role, which through increased load capacity and cost decline, accomplished annual growth rates of 24% from 2015 to 2018, driven primarily by transport applications [1]. The projection of the global demand for lithium for electromobility and other industrial applications predict that the demand for lithium metal will increase 8% annually from 2017 to 2025 and lithium demand in LIBs is expected to range from 300 to 600 thousand metric tons of lithium per year, comprising virtually 66% of the current Li production worldwide [2], [3].

Chile has the largest lithium reserves in the world, about 52% of the total, and this resource is extracted from continental brines [4]. The brine from the Salar de Atacama in Chile has the highest average lithium concentration known; however, the current extraction process is not sufficiently efficient.

The extraction begins by pumping brine well from the Salar into a series of evaporation ponds where solar evaporation takes place for approximately 18 months [5], [6], [7], generating a concentrated lithium brine (5–6% Li⁺, Fig. 1 (in blue)). Subsequently, to remove ions that do not precipitate in the evaporation stage, the concentrated brine is pumped to a treatment plant. At this point, due to the acidity of the brine, the boron present is removed as boric acid by liquid-liquid extraction using ideally isooctyl alcohol or 2-ethylhexanol [8], [9]. The boron-free brine is purified of magnesium and calcium in a second step by precipitation and liquid solids separation using lime and sodium carbonate. Finally, to the resulting brine, which has a lithium content of around 0.6% at the moment, is added sodium carbonate to precipitate lithium carbonate (Fig. 1 (in black)) [10], [11]. Unfortunately, these last two stages involve a loss of lithium, occluded in the precipitation, economic cost in the use of lime and carbonate salts, a significant amount of magnesium salts wastes, and an elevated increase in sodium on the final product.

Consequently, a variety of methods for lithium enrichment in brines have been widely studied in the literature [11], [12], [13], [14], [15], including liquid-liquid extraction, which has been extensively researched due to its efficient and economical application [16], [17], [18], [19]. On the other hand, a large number of articles have used ionic liquids (ILs). Fluorinated ionic fluids based on imidazolium combined with triisobutyl phosphate (TIBP) or TBP [20], [21], [22], [23], phosphonium ionic liquid [24], and dialkyl phosphates ionic liquids [25], [26], demonstrated an effective extraction capacity of lithium from brines (over 80%).

In terms of magnesium extraction, commercial extractants such as Versatic 10 (neodecanoic acid), Cyanex 272 (bis-(2,4,4-trimethylpentyl) phosphinic acid) and DEHPA (di-(2-(2-ethylhexyl) phosphoric acid)) [27], [28], [29], have shown a high extraction capacity from synthetic solutions. In this context, it is remarkable the work performed by Li et al. [30], where they demonstrated that the mixture of Cyanex 293 (mixture of trihexyl and octylphosphine oxides) with a β-diketones (HPMBP - 4-benzoyl-3-methyl-1-phenyl-2-pyrazolin-5-one) efficiently and selectively extracted magnesium from synthetic brines, and in addition, demonstrated the existence of a strong synergism in the combined system.

Boron extraction from lithium brine has been neglected in several studies; however, it continues being a crucial step in the extraction process. The presence of boron makes the separation of lithium difficult [31], and quantities of this contaminant in battery-grade lithium carbonate could cause serious disadvantages in degree of purity [32] and crystallization [33], [34].

It is proven that depending upon the aqueous media pH, boron can exist as borate ion [B(OH)₄]⁻ or boric acid [B(OH)₃] [35]. In this respect, boron removal has been studied through several processes such as electrodialysis [36], [37], absorption and filtration membranes [38], [39], reverse osmosis [40], [41], chemical precipitation [42], among others [43], [44]. Nevertheless, solvent extraction is considered one of the most effective techniques for boron extraction due to the low cost, high selectivity, large extraction capacity, and simple execution.

Selected alcohols classified as monohydric alcohol [45], [46], [47], [48], di-hydroxy alcohols (diols) [49], [50], [51], [52], [53], [54], and mixed alcohols [55], [56], [57], have prominent properties for extracting boric acid from the brine under acid conditions. Prime among them was the study realized by Zhang et al. [58] who used 2-ethylhexanol to extract boron from salt lake brine. The results exhibited a boron extraction efficiency of 99.5% in a 50% v/v kerosene system and high-purity products by employing water as the stripping agent.

In view of all the studies focused on brine purification, an extraction medium capable of simultaneously removing Mg²⁺ and B would be auspicious for the extraction process, since: (1) it would eliminate successive precipitation stages, reducing the process time, (2) it would decrease the expense of precipitation salts, (3) it would eliminate the precipitated magnesium waste, accumulated in the salar during years, (4) it would avoid the loss of occluded lithium in precipitated magnesium salts (5) it would provide a high purity lithium brine, and (6) it would allow the flexible manufacture of lithium products. For example, intervening the current process in the solvent extraction step by adding a dual extraction system could provide a new optimized process, as shown in Fig. 1 (in red).

Besides, optimization of parameters in extraction is one of the most important stages in the development of an efficient and economic process. The traditional “one-factor-at-a-time approach” is time-consuming, and the interactions between independent variables are not considered [59]. Hence, the univariate method of optimization of variables commonly used [18], [19], [20], [21], [22], [23], [24] is inadequate for subsequent scaling studies because it does not show the combined effect of all the variables of the process and requires a greater number of experiments that ultimately lack reliability.

Experimental designs are commonly used to evaluate and optimize an extraction process obtaining the maximum amount of useful information in a small number of experiments, minimizing costs, and maximizing the desired responses [60], [61]. One of the most commonly used experimental designs for optimization is the response surface methodology (RSM), a mathematical and statistical technique used to develop empirical models of a process [62], [63], which is established on the fit of a polynomial equation to the experimental data.

According to the preliminary range of extraction variables determined by a single-factor experiment, a three-level-three-factor Box-Behnken design (BBD) can be applied to determine the best combination of extraction variables for the purification of the brine. The BBD is a rotating, spherical response surface, including a center point and midpoints between the corners, circumscribed on a sphere. This design can be applied for the optimization of various chemical and physical processes [64], [65], [66], [67], where the number of experiments is determined according to the requirements of the process.

Intending to make a major contribution to research on the purification of lithium brines, this work discloses the feasibility of concurrent extraction of impurities using a combined system of dialkylphosphate ionic liquid (Fig. 2), (1-butyl)triethylammonium bis(2-ethylhexyl)-phosphate ([N₂₂₂₄][DEHP]) and a 2-ethylhexanol/kerosene mixture, which enable removal of both elements in the same stage. While [N₂₂₂₄][DEHP] removes Mg²⁺ a pH value ≤ 2, where boric acid prevails and the co-extraction of lithium is minimal, 2-ethylhexanol:kerosene acts as solvent and extractant of boron at once.

While most studies in the field have only been demonstrated in synthetic brines, this method was applied in natural brine from the Salar de Atacama, and considering the complexity of a dual system, variables involved such as the amount of extractant (EQ), the number of contacts (NC), and the ratio between the volume of organic and the volume of aqueous (Vo/Va) were analyzed and optimized by RSM, demonstrating the efficiency of the system.