From Agro-Industrial Waste to Gold Lixiviant: Evaluating Cassava Wastewater Applications in Artisanal Mining

Abstract

Artisanal and Small-Scale Gold Mining (ASGM) is a primary source of global mercury pollution, creating an urgent need for sustainable, low-cost alternatives to amalgamation. This study investigates the use of cassava wastewater (manipueira), a cyanogenic agricultural byproduct, as a lixiviant for a gold concentrate (14.30–15.87 ppm Au) from an artisanal mine. Two approaches were evaluated: direct leaching with manipueira in natura (250 ppm CN⁻) in single and double 8 h and 12 h cycles, and leaching with a cyanide solution concentrated from dilute manipueira (100 ppm CN⁻) via a simplified air-stripping system. Results were benchmarked against the mine’s amalgamation (44.7% recovery) and 30-day heap leach (75.8% recovery) processes. The most effective method observed was a two-cycle, 8 h leach with manipueira in natura, which achieved a mean gold recovery of $76.75\pm 4.71\%$. This result is comparable to the efficiency of the site’s lengthy heap leach process and suggests a promising, faster, route to eliminating mercury use. Longer (12 h) leaching cycles yielded lower recoveries, suggesting process limitations such as preg-robbing. The cyanide concentration method proved inefficient, recovering a maximum of 12.40% of the available cyanide and resulting in a weaker lixiviant. The findings demonstrate that while direct leaching is a viable alternative to mercury, the inherent instability of manipueira necessitates a focus on developing efficient, controlled systems to extract and concentrate its cyanide content, thereby creating a standardized “green” reagent from a large-volume agricultural waste stream.

1. Introduction

Artisanal and Small-Scale Gold Mining (ASGM) involves an estimated 10 to 19 million people in more than 80 countries and represents 37.7% of anthropogenic mercury emissions [1,2,3,4,5,6]. Artisanal miners produce between 380 and 870 tonnes of gold annually, using an estimated 640 to 1000 tonnes of mercury in the process [7,8,9,10,11]. Mercury amalgamation is a common extraction method due to its low cost, availability, and simple application, which does not require formal training or specialized equipment [12,13,14,15]. The historical significance and environmental legacy of ASGM are exemplified in Brazil, where official production peaked in 1983, accounting for over 88% of the national total. Although this official share declined to approximately 17% by 2017, the sector remains a key component of the national mining industry [16,17,18]. This history of intensive activity has had lasting environmental consequences; for example, mining in the Madeira River basin between 1975 and 2002 is associated with the release of an estimated 3000 tonnes of mercury into the Amazon environment [19,20,21,22,23]. This case illustrates the dual role of ASGM in the region over recent decades as both an economically important activity and a primary source of environmental contamination.

The use of cyanide in ASGM as a secondary step after amalgamation is increasing [24,25,26,27]. Miners use processing centers for initial amalgamation and leave behind mercury-contaminated tailings. These tailings, which can contain over 70% of the original gold, are then processed with cyanide to extract the remaining metal [28,29,30,31]. This practice can lead to the formation of mercury–cyanide complexes that are not recovered and are instead discharged with tailings into local water systems [32,33,34]. Although cyanide can be an effective reagent, its application in many regions has spread through heap and vat leaching, often without adherence to safety or environmental protocols. This results in high cyanide consumption, sometimes up to 1 kg of NaCN per tonne of ore, and frequently without proper treatment of the residual cyanide [35,36,37].

In the broader mining industry, cyanide salts and solutions have been common reagents for gold extraction since the early 20th century [38,39,40]. However, in the ASGM context where mercury is often present, the combined use of these reagents can lead to the formation of stable and mobile mercury–cyanide complexes, exacerbating environmental hazards [32,41]. When applied correctly in mercury-free systems, cyanidation is a viable alternative capable of achieving over 90% gold recovery [42,43]. As an available and relatively inexpensive industrial chemical, cyanide is considered a suitable reagent to replace mercury for processing nearly 90% of the world’s gold [35,44,45]. Its industrial use and environmental risks can be managed safely with appropriate investment and specialized labor; however, these costs are often neglected in artisanal settings [38]. The primary danger of cyanide is the formation of highly toxic hydrogen cyanide (HCN) gas, which can occur if the pH of the solution drops below 10.5. Inhalation of HCN can be lethal, and chronic exposure poses serious health risks. Safe implementation in ASGM requires strict pH control, adequate ventilation, the use of personal protective equipment (PPE), and the implementation of detoxification protocols for tailings, such as natural degradation or chemical treatment, to neutralize residual cyanide before discharge. These measures, while essential, represent a significant operational and financial challenge in resource-limited artisanal contexts [46,47,48,49]. Natural processes can degrade cyanide into less toxic chemicals, and with proper management, it does not typically persist or cause chronic health problems at low concentrations. Furthermore, the potential to recover and recycle cyanide from plant tailings offers both economic benefits and a means to comply with strict environmental regulations, reinforcing its suitability as a managed alternative to mercury [50,51].

A potential biosourced alternative to industrial cyanide derives from the natural process of cyanogenesis, a chemical defense mechanism present in over 2500 plant species [52,53,54]. These plants synthesize and store cyanogenic glycosides, which are non-toxic compounds that can be hydrolyzed by enzymes to release hydrogen cyanide (HCN) [55,56]. Cassava (Manihot esculenta) is a plant known for this characteristic, producing the glycosides linamarin and lotaustralin [57,58,59,60,61,62,63]. This natural phenomenon provides the scientific basis for using plant-derived liquids as a source of cyanide for metallurgical applications. Pioneering research applying this principle to gold extraction was conducted by Torkaman et al. [64,65]. Their work utilized manipueira, the liquid effluent from cassava processing, as a bioleaching agent. In laboratory experiments on a Colombian ore sample, they demonstrated that this method could extract over 80% of the available gold in a 24 h period, providing a compelling proof-of-concept for a low-cost and accessible alternative to both mercury and industrial cyanide [65].

This article is a follow-up study built directly upon Torkaman’s foundational work, investigating the gold leaching potential of manipueira both in natura (untreated, freshly collected) or as a concentrated cyanide solution derived from it, using ore samples from an artisanal gold mine in Brazil, located relatively near a large cassava starch industry. The study is based on a comparative methodological approach, evaluating direct leaching against a concentrated solution, and its applied context, benchmarking the laboratory results against the performance of an active industrial amalgamation and heap leach operation. The main goals of the study were to assess the feasibility of these different process routes using a locally sourced agricultural by-product and to generate the data needed to inform the design of a potential future pilot-scale application.

2. Study Area

The materials for this study were sourced from two distinct sites, located approximately 180 km from each other, in the state of Pernambuco, Brazil (Figure 1).

The bioleaching agent, manipueira, was obtained from “Maná Agroindustry”, a large-scale cassava starch production factory located in the municipality of Araripina. This facility processes up to 400 tonnes of cassava tubers per day, generating a significant effluent stream of up to 128,000 L of manipueira daily. The starch extraction process at the factory, as depicted in Figure 2, involves washing and peeling the cassava tubers, grating them into a pulp, and then dewatering the pulp to separate the starch. The liquid wastewater from this dewatering stage is the manipueira, which is rich in cyanogenic compounds. The solid by-products from peeling and screening are repurposed as animal feed, while the final starch product is dried for market.

The ore used for the experiments was a sluice box concentrate with unknown prior concentration of gold, sourced from an Artisanal Gold Processing Center (AGPC), “Simas Mining”. The operational practices at this center are rudimentary, with no systematic control over feed grades, no prior mineralogical characterization, and minimal control over the reagents used. Consequently, while the material is considered representative of the typical concentrate processed at the site, significant batch-to-batch heterogeneity is expected due to variations in orebody geology and inconsistent upstream processing. The operational workflow at this center (Figure 3) begins with ore extraction at the open-pit mine located in the municipality of Salgueiro, followed by hauling the material to the processing center, situated 30 km away, in the town of Serrita, both in Pernambuco State, Brazil. Upon arrival at the primary processing site, the ore undergoes a multi-stage comminution circuit that includes manual breaking of large rocks, primary and secondary reduction in jaw crushers, and fine grinding through a series of hammer mills. The resulting finely ground ore is then subjected to whole-ore amalgamation through five concrete sluice boxes, each lined with mercury-coated copper plates, to capture the gold.

After the initial gold recovery via amalgamation, the tailings are collected in a designated pit. Once the pit is filled, typically over a period of six to seven months, these mercury-bearing tailings are transported to a secondary processing site 10 km away. There, the material is stockpiled and subjected to a long-term heap leaching process using a sodium cyanide solution for a minimum of 30 days. The gold-rich leachate is passed through activated carbon to adsorb the gold. Finally, the loaded carbon is transported to a separate, non-disclosed facility in the state of Bahia for elution and final gold recovery, completing a complex processing chain that relies on both mercury and cyanide.

A previous diagnostic campaign conducted at this site reported a gold recovery of $44.7\pm 11.7\%$ using whole-ore amalgamation alone, $75.8\pm 4.32\%$ through heap leaching, and an overall recovery of $86.6\pm 2.89\%$. These values served as key comparative parameters in the present study.

3. Materials and Methods

The experimental methodology, adapted from Torkaman et al. [64,65], was divided into two distinct approaches to evaluate the use of manipueira (cassava wastewater) as a lixiviant for gold ore. The first method, carried out in June 2023, involved direct leaching using manipueira in natura under various time and cycle conditions. The second method, conducted in September 2024, focused on the preparation of a concentrated cyanide solution, derived by stripping and capturing cyanide from a large volume of manipueira, which was then used in a comparative leaching test.

3.1. Leaching with Manipueira in Natura

The leaching experiments were conducted in duplicate using in natura (untreated, freshly collected) cassava wastewater as the primary lixiviant. For each batch test, a 50 g sample of gold ore was added to 200 mL of manipueira. The initial free cyanide (CN⁻) concentration in the manipueira was determined using a Chemetrics K-3015 Colorimetric Test Kit (Chemetrics by AquaPhoenix Scientific, 4295 Catlett Rd, Midland, VA, USA) before the start of each experiment and was found to be 250 ppm (mg/L). This kit operates by reacting the free cyanide in the sample with reagents, including barbituric acid and isonicotinic acid, to produce a blue color of varying intensity, which is then visually compared against a calibrated color scale ranging from 0 to 1 mg/L. For concentrations exceeding this range, samples must be diluted. Although visual colorimetric methods are generally considered less precise than instrumental techniques, previous research by Torkaman et al. [64] demonstrated that results from this method are consistent and comparable to those obtained by titration and ion-specific electrodes for analyzing cyanide in manipueira. Therefore, this method was selected for its reliability, speed, and suitability for rapid, on-site measurements in a field setting, providing data sufficient for process monitoring.

The pH of the resulting slurry was immediately measured (around 4.5) and adjusted to approximately 10.5 by gradually adding a 50% (m/v) NaOH solution. This adjustment was aimed to ensure the chemical stability of the cyanide ions and prevent the formation of volatile hydrogen cyanide (HCN). The experiments were carried out under constant agitation using a magnetic stirrer (between 100 and 150 rpm) and were structured into four distinct conditions: a single 8 h leaching cycle, two consecutive 8 h leaching cycles, a single 12 h leaching cycle, and two consecutive 12 h leaching cycles.

For the two-cycle experiments, the solid residue from the first leaching cycle was separated from the pregnant solution by filtration and was then subjected to a second, identical leaching cycle using a fresh 200 mL volume of pH-adjusted manipueira. This two-stage approach was designed to assess the extent of gold recovery under repeated exposure to the lixiviant.

To counteract the consumption of cyanide during the process, the CN⁻ concentration was monitored at the midpoint of each cycle (i.e., at 4 h for the 8 h tests and 6 h for the 12 h tests). Immediately following this measurement, an additional 100 mL of fresh manipueira was introduced into the reactor to replenish the cyanide concentration, increasing the total slurry volume to 300 mL.

3.2. Concentrating Cyanide from Manipueira

To evaluate the leaching potential of a concentrated cyanide solution derived solely from manipueira, a simplified cyanide stripping and concentration procedure was implemented, as depicted in Figure 4. A volume of 60 L of fresh manipueira was collected from the cassava starch production facility and placed in a 200 L sealed plastic barrel. The initial cyanide concentration was measured immediately upon collection and was found to be 100 ppm. The pH of the in natura solution was around 4.5. The barrel was equipped with an air sparging system, wherein an air compressor bubbled air through the manipueira solution. The air flow was controlled manually based on the visual observation of the bubbling intensity inside the scrubbing flasks. The stripping was conducted in duplicate batches of 2 h and 4 h durations. This process was aimed to strip the cyanide from the acidic solution by volatilizing it as HCN gas. The gas-laden air exiting the barrel was directed through a series of three consecutive scrubbing flasks, each containing 300 mL of a 1 M sodium hydroxide (NaOH) solution, aimed at trapping the HCN gas and converting it into a stable sodium cyanide (NaCN) solution. The final combined solution from the traps had its CN⁻ content measured and was diluted as needed to achieve a target concentration of 400 ppm (mg/L). This lixiviant solution was split into duplicates and then used in a single, 12 h batch leaching test with 50 g of ore to serve as a comparative benchmark for the manipueira in natura experiments.

3.3. Laboratory Analysis and Metallurgical Balance of Au

At the conclusion of the final leaching cycle for each condition, the residual solid ore was collected, filtered, washed, and dried. A sample of the original, unleached ore, and the dried solids were then sent to SGS Geosol Laboratories, in Vespasiano, Minas Gerais State, Brazil, for gold content analysis via the FAA303 technique. This method consists of a standard fire assay procedure where the ore sample is pulverized to below 200 mesh and then fused with a fluxing agent and lead oxide, collecting the precious metals in a lead button. The lead is subsequently removed through cupellation, leaving a precious metal bead (dore) that is dissolved in aqua regia. The final gold concentration in the resulting solution is determined by Atomic Absorption (AA) spectrometry.

Metallurgical balance for gold was performed based on procedures adapted from Anene et al. [66]. The % recovery for gold (Au) after the leaching process can be calculated based on the grades of the feed and tailings streams using Equation (1):

$$\%\mathrm{Recovery}=\left[1-\left({\displaystyle \frac{A{u}_{\mathrm{TL}}}{A{u}_{\mathrm{FD}}}}\right)\right]\times 100$$

(1)

where $A{u}_{\mathrm{T}\mathrm{L}}$ stands for the gold grade of the tailings and $A{u}_{\mathrm{F}\mathrm{D}}$ is the Au grade of the feed.

3.4. Data Analysis and Experimental Limitations

All experiments were conducted in duplicate (n = 2); however, given the small sample size, the data analysis was limited to descriptive statistics, primarily the calculation of mean values and the presentation of the interval defined by the mean ± two standard deviations. While calculating a standard deviation from only two samples has limitations, it provides a standardized and conservative estimate of variability. Formal statistical hypothesis testing was not performed, as it would not be appropriate for the limited dataset. The interpretation of the results focuses on observed trends and magnitudes of difference rather than statistical significance.

It is important to acknowledge the limitations inherent in this study, which was conducted under field conditions constrained by the resources available in an artisanal mining context. The use of improvised and unsophisticated equipment, particularly for the cyanide stripping and concentration tests, significantly influenced the outcomes. Furthermore, the natural variability of the raw materials, both the gold content in the ore concentrate and the cyanide concentration in the manipueira, introduces a level of uncertainty that is characteristic of real-world ASGM operations. The results should therefore be interpreted as a practical demonstration of potential rather than a precisely controlled laboratory study.

4. Results and Discussion

4.1. Lixiviant and Ore Characterization

As previously stated, the initial free cyanide (CN⁻) concentration of the manipueira in natura used in the June 2023 leaching experiments was determined to be 250 ppm. The manipueira collected for the concentration experiments in September 2024 had an initial concentration of 100 ppm.

A representative sample of the concentrate used for the first experimental campaign (June 2023) was analyzed by Fire Assay (FAA303) and determined to have a gold grade of 15.87 ppm. A different batch of concentrate used for the second campaign (September 2024) had a gold grade of 14.30 ppm.

4.2. Gold Leaching Performance

The gold recovery results from the duplicate tests, illustrated in Figure 5, highlight the potential of using manipueira in natura as a replacement for whole-ore amalgamation. The leaching performance is first discussed, followed by an interpretation of the underlying mechanisms and process dynamics. The most effective condition tested was the two-cycle, 8 h leaching process (8H2C), which achieved a mean gold recovery of $76.75\pm 4.71\%$. This outcome is particularly significant when compared to the current operational performance at the study site. The mean recovery is similar to the $75.8\pm 4.32\%$ reported for the site’s 30-day cyanide heap leach process and represents a considerable improvement over the $44.7\pm 11.7\%$ reported for their whole-ore amalgamation. This demonstrates that a rapid, 16 h leaching process can yield gold recoveries comparable to those of a month-long conventional chemical process, while completely eliminating the use of mercury.

The benefit of a second leaching cycle is evident in the 8 h tests. Mean recovery increased from $51.78\pm 27.53\%$ in a single cycle to $76.75\pm 4.71\%$ in two cycles, with lower variability between the duplicate results. The observed increase suggests that replenishing the lixiviant is critical for achieving higher and more consistent extraction, a common practice in cyanidation circuits.

Conversely, extending the leaching time to 12 h yielded lower mean recoveries and greater variability. The single 12 h cycle (12H1C) produced a mean recovery of only $31.09\pm 25.33\%$, and the two-cycle test (12H2C) reached a mean of $45.89\pm 33.91\%$.

The experiment using the concentrated lixiviant (12H1C-CON) produced a mean recovery of $59.40\pm 5.89\%$. While this is higher than the mean recoveries from the single-cycle in natura tests, it falls short of the 82.4% recovery reported by Torkaman and Veiga [65] in a study which used a different ore and a naturally high-concentration manipueira (600 ppm CN⁻), underscoring that direct comparisons must account for differences in experimental conditions.

4.3. Interpretation of Leaching Behavior

The counterintuitive decrease in performance with longer leaching times (12 h vs. 8 h) may be attributed to preg-robbing, a phenomenon where dissolved gold–cyanide complexes are re-adsorbed by carbonaceous material, such as starches and fine solids, present in the manipueira. This effect is well documented in the works by Miller et al. [67], Garcia Rosales et al. [68] and, in the context of cassava-based leaching, the possibility was first suggested by Torkaman et al. [64] as a potential limiting factor for extended leaching with this organic-rich lixiviant. While a detailed analysis of the organic phase was beyond the scope of this field-based study, the observed trend suggests that shorter, more intense leaching cycles are preferable to prolonged exposure. The absence of an 8 h, two-cycle test with the concentrated lixiviant is a limitation of this study; such a test would be valuable for future work to determine if higher cyanide concentration can mitigate these effects.

4.4. Cyanide Consumption and pH Control

The experimental log provided critical insights into the process dynamics, specifically regarding cyanide consumption and pH stability (

Table 1. Summary of process parameters from experimental log.

Cyanide (ppm)			pH Range
Initial	Mid-Cycle	Final	Initial	Final
250	<10	20	4.1	9.9–10.4
250	30	20	4.1–4.4	10.3–10.8
250	30	20	5.9	10.4–10.5
250	30	20	4.4–5.9	10.3–10.8

). Across all test conditions, a rapid and significant decrease in free cyanide concentration was observed. For instance, in most tests, the initial concentration of 250 ppm dropped to approximately 30 ppm by the midpoint. By the end of each cycle, the residual cyanide was consistently low, around 20 ppm. This high rate of consumption highlighted the need to replenish the cyanide content in the solution at mid-cycle to ensure sufficient reagent was available for gold dissolution. Furthermore, the log confirms the importance of pH management. The initial acidic pH of the manipueira (pH 4.1–5.9) required immediate correction. Even after adjustment, the pH tended to drift downwards during the leaching process. This behavior highlights the practical challenges in an artisanal setting and reinforces the safety concerns discussed previously, as a drop below pH 10.5 increases the risk of toxic HCN gas formation. The data demonstrate that maintaining both sufficient lixiviant concentration and a safe, stable alkaline environment requires active and ongoing process control.

4.5. Cyanide Stripping and Concentration Efficiency

The results of the cyanide concentration experiments, presented in Figure 6, reveal that the stripping device itself was the primary cause of the low recovery efficiency. Based on a mass balance calculation, the overall cyanide recovery from the 60 L of dilute manipueira (containing an initial 6 g of CN⁻) was poor. After four hours of sparging, a maximum of only 0.74 g of CN⁻ was captured in the scrubbing solutions, corresponding to a recovery of just 12.40%. This inefficiency is attributed to the rudimentary, open-system design of the stripping apparatus, which failed to create adequate conditions for HCN volatilization and capture. CN⁻ measurements from the individual scrubbing flasks confirm this; in a representative 4 h test, the first flask captured 10.00% of the initial cyanide, while the second and third flasks captured only 2.00% and 0.40%, respectively. This steep decline indicates that while the first scrubber was somewhat effective, a significant amount of uncaptured HCN was likely lost to the atmosphere.

Consequently, the lixiviant prepared for the 12H1C-CON test was insufficiently concentrated, which directly explains its modest gold recovery (59.40%). The inefficiency of this improvised method stands in stark contrast to industrial processes like AVR (Acidification, Volatilization, and Reneutralization) or vacuum distillation, which can achieve over 95% cyanide recovery but require significant capital investment and process control [51,69]. The results highlight the critical need for a more controlled and engineered system for ASGM. Future iterations should incorporate a closed-loop design to recycle the exhaust gas, allowing residual HCN to be passed through the scrubber solution multiple times, thereby maximizing capture efficiency and minimizing environmental losses.

While the use of manipueira in natura shows promise, its direct application in ASGM presents significant challenges due to the inherent instability of its cyanide content. The concentration of cyanogenic glycosides varies considerably depending on the cassava species, as well as the environmental and climatic conditions under which the plant was cultivated. This variability is compounded by the purchasing practices of the cassava starch and flour producers, who typically buy cassava based on its starch content rather than by specific species, resulting in a mixed and inconsistent feedstock. Furthermore, once processed into manipueira, the solution is prone to rapid fermentation, which, without immediate pH adjustment, leads to the swift volatilization of HCN gas. Such variability makes direct, large-scale application unreliable for a consistent metallurgical process. Harnessing the enormous effluent stream from large-scale producers like Maná Agroindustry, which generates up to 128,000 L of manipueira daily, represents a significant opportunity. Therefore, for this biolixiviant to be viably adopted in ASGM, it is crucial to develop methods to first stabilize the cyanide in the solution and then to implement an efficient process for extracting and capturing it. Such a system would not only mitigate the issue of inconsistent potency but could also establish a sustainable, green stream of cyanide production, transforming a variable agricultural byproduct into a standardized and effective tool for mercury-free gold extraction.

5. Conclusions

The study demonstrated that manipueira, a widely available agricultural byproduct, can serve as an effective lixiviant for gold extraction in ASGM, offering a viable alternative to mercury amalgamation. The best condition identified, a two-cycle, 8 h leaching process, achieved a mean gold recovery of $76.75\pm 4.71\%$, a rate similar to but much faster than the site’s current 30-day heap leach process and vastly superior to its amalgamation practice. This finding confirms that a rapid, mercury-free process using a local, renewable resource is technically feasible. However, the research also highlighted two critical challenges for practical implementation. First, the leaching performance of manipueira in natura appears to be constrained by leaching duration, possibly due to preg-robbing from its high organic content, making direct industrial application potentially unreliable without optimization. Second, the simplified, low-cost method for concentrating cyanide from dilute manipueira proved to be highly inefficient, with a maximum cyanide recovery of only 12.40%. Such inefficiency resulted in a weak lixiviant and consequently lower gold recovery in a single leaching batch, underscoring that the apparatus was not adequate for the task.

The most promising pathway for the adoption of this technology in ASGM is not the direct use of raw manipueira, but rather the development of an optimized and controlled engineering system. Future research and development should focus on designing an efficient, closed-loop process to reliably strip, capture, and concentrate the cyanide from the vast effluent streams of large-scale cassava producers. Such a system could transform a problematic agricultural waste into a standardized and sustainable “green” cyanide reagent, providing the ASGM sector with a powerful tool to eliminate mercury use and improve its environmental performance.