Mechanical and Microstructural Behavior of Mine Gold Tailings Stabilized with Non-Conventional Binders

Abstract

Recent tailing dam failures in Brazil have been attributed to liquefaction. Chemical stabilization offers a promising solution to enhance the strength and stiffness of tailings and mitigate liquefaction potential. This study investigated the mechanical and microstructural behavior of gold mine tailings (GMTs) stabilized using (i) an alkali-activated binder composed of sugar cane bagasse ash (SCBA), hydrated eggshell lime (HEL), and sodium hydroxide (NaOH) and (ii) Portland cement (PC). Drained and undrained triaxial shear tests and scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDS) analyses were performed. Specimens stabilized with Portland cement exhibited a strong strain-softening behavior and the highest strength, with 5.3 MPa under 200 kPa confining pressure compared to 2.3 MPa for alkali-activated samples and 740 kPa for untreated GMTs. The addition of either binder also increased both the peak effective friction angle and the critical state stress ratio, confirming an enhanced shear strength. SEM-EDS analyses confirmed the formation of cementitious reaction products, explaining these improvements. This research validates both binders as viable solutions for tailing stabilization, with the novel alkali-activated binder offering a sustainable alternative for large-scale applications.

1. Introduction

Mine tailings are by-products of mineral processing, consisting of anthropogenic granular geomaterial, water, and significant concentrations of heavy metals. These materials are typically deposited in open-air tailing ponds without treatment [1]. In Brazil, the 2019 Brumadinho tailing dam failure released several million cubic meters of mine tailings, resulting in the loss of over 270 human lives and causing severe environmental degradation, including widespread contamination of soil, surface water, and ecosystems in the affected region [2,3]. A technical investigation attributed the Brumadinho failure to continuous internal deformations due to the creep and reduced shear strength within an unsaturated zone following intense local rainfall. These factors collectively triggered flow liquefaction, ultimately causing the dam to fail [4].

Mine tailings disposed of conventionally often exhibit low in situ density and shear strength, making them susceptible to instability and failure. To prevent future dam collapses, it is essential to enhance the understanding of the geotechnical behavior of these materials. One promising approach is the stabilization of tailings, which can be implemented during the stacking of compacted dry tailings or prior to dam decommissioning to improve in situ properties [5,6]. In this context, previous studies have investigated the behavior of tailings subjected to conventional cement stabilization [7,8]. Chemical stabilization has demonstrated potential to significantly enhance the strength and stiffness of tailings, thereby reducing the likelihood of liquefaction.

Consoli et al. [5] investigated the performance of compacted iron ore tailing–Portland cement mixtures subjected to varying degrees of compaction. Drained triaxial tests (CID) revealed a peak strength followed by strain-softening behavior across all specimens. Increasing the compaction degree from 89% to 99% (Standard Proctor effort) did not affect the peak effective friction angle (ϕ′_peak) but nearly doubled the effective cohesion intercept (c′). The critical state friction angle (ϕ′_cs) was determined to be 36.3°. Further, Consoli et al. [6] examined the mechanical behavior of reconstituted gold tailings artificially stabilized with varying amounts of Portland cement, all compacted to achieve a similar high void ratio. Saturated undrained triaxial (CIU) tests were conducted under low confining pressures. Incorporating 1.0% Portland cement by dry mass was sufficient to prevent liquefaction, as evidenced by the generation of negative excess pore pressures, while maintaining a common critical state line (CSL) for both untreated and lightly cemented specimens, with a critical state friction angle (ϕ′_cs) of 34.8°. Higher cement contents (6.25% and 12.5% by dry mass) altered the ultimate mechanical behavior of the tailings, increasing the ϕ′_cs to 57.9°.

Da Silva et al. [9] investigated the mechanical behavior of cement-stabilized iron ore tailings (IOTs) through unconfined compressive strength (UCS) tests and triaxial testing. The UCS tests evaluated various cement–tailing blend ratios under compacted conditions. Additionally, nine drained triaxial tests were conducted: six specimens were cured under atmospheric pressure and three under a confining pressure of 300 kPa. The tests involved drained axial loading under constant mean effective stress (p′) and lateral unloading stress paths, with shearing performed at mean effective stresses of 300 and 3000 kPa. The results indicated a stress dependency in the effective shear strength parameters. The p′–q stress paths for samples cured under atmospheric and confining pressures showed no significant differences, with both converging toward the same failure envelope. However, specimens cured under confining pressure exhibited a reduced radial strain rate, which contributes to enhanced global stability of the structure by mitigating the risk of liquefaction-induced instability.

Alternative binders have been explored for tailing stabilization due to their potential to reduce environmental impacts compared to traditional Portland cement. Among these, alkali-activated materials, produced through the alkaline activation of aluminosilicate precursors to generate cementitious binding agents, have emerged as a viable and sustainable option.

Kiventerä et al. [10] investigated using calcium sulfoaluminate–belite (CSAB) cement to stabilize gold mine tailings characterized by elevated heavy metal and sulfate concentrations. The study evaluated both mechanical strength and heavy metal leaching behavior. After just 7 days of curing, all hazardous elements were effectively immobilized, with immobilization performance remaining stable over time. Additionally, high mechanical strength and substantial sulfate immobilization were achieved with mine tailings comprising up to 50% of the total binder content. Servi et al. [11] evaluated the mechanical behavior of iron ore tailings stabilized with Portland cement and an alkali-activated binder composed of rice husk ash, carbide lime, sodium hydroxide, and silicate. Results showed an enhanced mechanical response of tailings for both studied binders, based on tests of unconfined compressive strength, split tensile strength, and durability by cycles of wetting, drying, and brushing. Farenzena et al. [12] investigated the mechanical behavior of iron ore tailings (IOTs) stabilized with an alkali-activated cement synthesized from two by-products from the IOT beneficiation process, metakaolin (MK) and sodium silicate (SS), tested under plane strain conditions. For uncemented samples, the peak friction angle (ϕ′_peak) was 31.8º and 35.4º for dry unit weights of 17 kN/m³ and 19 kN/m³, respectively. For 5% cement addition, ϕ′_peak was 33.7º and 36.9º, and c′_peak was 34 kPa and 44 kPa for 17 kN/m³ and 19 kN/m³, respectively. Moreover, even a 1% binder content showed a noticeable improvement in the initial tangent shear modulus, indicating enhanced stiffness.

In a more recent study, metakaolin-based geopolymers effectively stabilized iron mine tailings, with the binder percentage and water content being critical factors. A mixture of 30% binder and 15% water in the tailings achieved a compressive strength of 7.72 MPa after 7 days, exceeding the ideal target of 7 MPa. Increasing the mixing time and reducing water content further improved the strength, demonstrating a more complete geopolymerization reaction [13]. Diosdado-Aragón et al. [14] studied an alternative to ordinary Portland cement for mine backfilling. An industrial residue containing magnesium carbonate and magnesium oxide was used as an alkaline activator with various mine tailings, including those from massive sulfide, stockwork, IOCG (iron oxide–copper–gold), and carbonatite deposits. This combination consistently improved the environmental behavior of the tailings by reducing their acid generation potential and improving leachate water quality. The alkali-activated pastes could be reclassified as inert wastes, offering a more sustainable and environmentally friendly solution for mine waste management.

In addition, a study on stabilizing iron ore tailings for dry stacking found that a combination of a polymeric solution and polypropylene fibers is highly effective. The experimental phase included direct shear tests and splitting tensile tests. The polymeric solution increased the iron ore tailings’ strength parameters by promoting particle cementation. Conversely, adding 0.5% polypropylene fibers enhanced the material’s ductility, which is crucial for preventing brittle failure in dry stacking structures. The optimal and most economical polymeric dosage was found to be 1:4 (polymer solution to water ratio), with the full performance gains developing after 28 days of curing [15].

Thus, the current study was undertaken to advance the understanding of mine tailing behavior following chemical stabilization with non-conventional cementing agents. The experimental program addresses several research gaps. First, past studies generally do not compare alkali-activated binders and Portland cement used for tailing stabilization in terms of strength and LCA, among other factors. This is important because the use of conventional cements is already well-established, and non-conventional binders need to show advantages to be used in practice. Second, the majority of research does not use triaxial compression tests to address the mechanical behavior of the stabilized tailings. This is fundamental for practical application, as geotechnical parameters for stability analysis, for example, are usually taken from those tests. Finally, for the first time, an alkali-activated binder comprising sugarcane bagasse ash (SCBA), hydrated eggshell lime (HEL), and sodium hydroxide (NaOH) was studied for tailing stabilization.

To achieve the aforementioned objectives, the experimental program involved conducting drained and undrained triaxial tests (CID and CIU) and SEM-EDS analysis on gold mine tailings (GMTs). The tailings were stabilized with two distinct binders: (i) an alkali-activated binder comprising sugarcane bagasse ash (SCBA), hydrated eggshell lime (HEL), and sodium hydroxide (NaOH) and (ii) conventional Portland cement (PC). This study was specifically designed to directly compare the influence of these two binder systems on the geomechanical behavior of the stabilized mixtures.

2. Materials and Methods

2.1. Materials

The materials evaluated in this study included (i) gold mine tailings (GMTs), (ii) sugarcane bagasse ash (SCBA), (iii) hydrated eggshell lime (HEL), (iv) sodium hydroxide (NaOH), and (v) Portland cement (PC). The GMTs, sourced from a mine in northeastern Brazil, represented the underflow of a tailings slurry with a solids content exceeding 70% by weight. SCBA, a by-product of sugarcane bagasse combustion at 800 °C, was obtained from a sugar industry in southern Brazil. HEL was laboratory-synthesized from eggshells collected in Porto Alegre, with production procedures detailed by Tonini de Araújo et al. [16]. The alkali activator used was NaOH micro pearls with 98% purity. For comparative purposes, high-early-strength Portland cement (Type III, ASTM C150/C150M, ASTM [17]) with a specific gravity of 3.15 g/cm³ was used to evaluate performance relative to alkali-activated binders.

2.2. Materials Characterization

Materials characterization included determination of unit weight of solids, Atterberg limits, grain size distribution, specific surface area via the BET method, and pozzolanic activity index (

Table 1. Physical properties of the materials used in this study.

Physical Properties	Materials			Standards
	GMT	SCBA	HEL
Unit weight of solids (g.cm−3)	2.71	2.08	2.24	ASTM D854 [19] NBR 16605 [20]
Liquid limit, LL (%)	-	-	-	ASTM D4318 [21]
Plastic limit, PL (%)	-	-	-
Plasticity index, PI (%)	non-plastic	non-plastic	non-plastic
Minimum void ratio	0.69	-	-	ASTM D698 [22]
Maximum void ratio	1.17	-	-	ASTM D4254 [23]
Uniformity coefficient	6.47	-	-	NBR 7181 [24], NBR 6502 [25]
Coefficient of curvature	1.39	-	-
% of coarse sand (0.6 < diameter < 2.0 mm)	64	-	-
% of medium sand (0.2 < diameter < 0.6 mm)	0	-	-
% of fine sand (0.06 < diameter < 0.2 mm)	0	-	-
% of silt (0.002 < diameter < 0.06 mm)	35	-	-
% of clay (diameter < 0.002 mm)	1	-	-
Mean particle diameter, D50 (µm)	91	31.01	7.43	-
90% finer particle diameter, D90 (µm)	-	55.85	23.37	-
10% finer particle diameter, D10 (µm)	15	9.18	2.53	-
Specific surface area (m2·g−1)	-	125.15	4.18	-
Pozzolanic activity index (mg Ca(OH)2/g of pozzolan)	-	817.6	-	NBR 15895 [26]
USCS classification	SM	-	-	NBR D2487 [27]

). Chemical and mineralogical compositions were assessed using X-ray fluorescence (XRF) and X-ray diffraction (XRD). The gold mine tailings primarily consist of 93.89% silica (SiO₂) and 1.20% alumina (Al₂O₃), with quartz identified as the dominant mineral phase. SCBA contains 60.6% silica, 13.8% iron oxide (Fe₂O₃), and 5.8% alumina, with mineralogical components including quartz (SiO₂), hematite (Fe₂O₃), magnetite (Fe₃O₄), and amorphous phases. HEL is predominantly composed of calcium oxide (72.9%) and contains portlandite [Ca(OH)₂], calcite (CaCO₃), and magnesium oxide (MgO), as reported by Consoli et al. [18].

Particle size distribution was determined using a laser diffraction particle size analyzer (CILAS, model 1064, Orléans, France). Specific surface area was measured through the Brunauer–Emmett–Teller (BET) method using QuantaChrome equipment (model NOVA 1781200e, Boynton Beach, FL, USA). The XRF technique was performed in a Malvern Panalytical X-ray fluorescence spectrometer (model Zetium, Malvern, UK), using a pressed sample. XRD analysis was carried out in a Rigaku X-ray diffractometer (model Miniflex 300, Tokyo, Japan; 30 kV, 10 mA, and COD database).

2.3. Optimal Composition and Preparation of Alkali-Activated Binder

Prior to the tailing stabilization procedure, alkali-activated pastes composed of SCBA, HEL, and NaOH were prepared and tested to determine the optimal alkali-activated binder combination. This optimal composition was subsequently used for all tailing–binder mixtures. Unconfined compressive strength (UCS) tests were conducted on the alkali-activated pastes molded in cylindrical PVC (polyvinyl chloride) molds with a diameter of 37 mm and a height of 74 mm.

To identify the ideal binder composition, a full factorial design with three factors—the SCBA/HEL ratio, the water/binder (W/B) ratio, and NaOH molarity—was employed, totaling 200 experiments. An analysis of variance (ANOVA) was performed to determine the influence of these factors on the response variable, UCS. The results from the ANOVA indicated that all three factors and their interactions had a statistically significant influence (p < 0.05) on the UCS, with the highest magnitude of effect observed for NaOH molarity, followed by the W/B ratio. These main effects are visually presented in Figure 1.

While Figure 1 showed that a 1 M molarity and a W/B ratio of 0.8 yielded a higher UCS, the effects of the SCBA/HEL ratio at values of 2.33 and 4 were very similar. To determine the precise optimal combination, a Tukey’s multiple comparison test was conducted, as the interaction among the factors was significant. The comparison was performed by holding one factor level constant and comparing the others. For a fixed W/B ratio of 0.8, the treatments with SCBA/HEL ratios of 4 and 9 were statistically equivalent, but the former showed a higher mean strength. Likewise, for a fixed molarity of 1 M, the treatments with SCBA/HEL ratios of 4 and 2.33 were also statistically equal, yet the former again demonstrated superior strength and lower calcium consumption. Therefore, the optimal factor combination was determined to be an SCBA/HEL ratio of 4 (80% SCBA and 20% HE), a W/B ratio of 0.8, and a molarity of 1M, which corresponded to a UCS of 2.39 MPa. This composition corresponded to a Na₂O-to-binder mass ratio of 2.61% and was identified as the optimal alkali-activated binder formulation for subsequent tailing stabilization.

The molding procedure consisted of four steps:

• SCBA and HEL were weighed, and sodium hydroxide pellets were dissolved in distilled water to prepare alkaline solutions at concentrations of 0.5, 1, 2, and 3 mol·L⁻¹. The required volume of alkaline solution for each mixture was determined based on selected water-to-binder (w/b = 0.7, 0.8, 0.9, 1.0, 1.2) and SCBA-to-HEL (90/10, 80/20, 70/30, 60/40 by mass) ratios. The total binder mass was computed as the sum of SCBA and HEL, subtracting the solution mass.
• The dry components were homogenized, and the alkaline solution was gradually added while mixing until a uniform paste was obtained.
• The mixture was placed into molds and compacted by vibration.
• Specimens were cured for 7 days before demolding to ensure sufficient strength. All samples were stored in sealed plastic bags inside a humid chamber maintained at 23 °C and 95% relative humidity to prevent moisture loss.

The selection of a 7-day curing period, rather than a longer duration, was based on the rapid strength and development observed in the material. As previously mentioned, the results demonstrated that the alkali-activated pastes achieved a compressive strength of 2.39 MPa within this timeframe, already exceeding the minimum requirement of 2.1 MPa for soil–cement bases as stipulated by DNIT (Brazilian National Department of Transport Infrastructure) [28]. This finding is crucial because it validates the material’s viability for practical engineering applications in a significantly shorter period.

2.4. Stabilized GMT Sample Preparation

First, the gold mine tailings were oven-dried and sieved by ABNT [29] standards. A target void ratio of 0.77 was selected to simplify the moist tamping procedure, representing an intermediate value between the minimum (0.69) and maximum (1.17) void ratios determined for the material (Table 1). This selected void ratio also falls within the range commonly reported in the literature for gold mine tailings disposed of in dams, which varies from 0.66 to 1.5 [30,31].

For the triaxial tests (CIU and CID), cylindrical specimens with a diameter of 38 mm and a height of 76 mm were prepared. The target void ratio of 0.77 and a binder content of 15% (as recommended by Ferrazzo et al. [32,33]) were used to calculate the required dry mass for each constituent material: (i) gold mine tailings alone for the reconstituted (GMT–water) specimens; (ii) a mixture of tailings, SCBA, and HEL for the alkali-activated binder specimens; and (iii) a mixture of tailings and Portland cement for the Portland cement-stabilized specimens.

After the materials were weighed, the following mixing procedures were adopted: For the GMT–water mixtures, gold mine tailings and water were combined to achieve a target moisture content of 17%, which facilitated manual compaction without exudation. For the tailing–alkali-activated binder specimens, tailings, SCBA, and HEL were dry-mixed, followed by the gradual addition of NaOH solution (corresponding to 2.61% alkali content) and distilled water until a uniform paste consistency was achieved. Additional water was added as needed to attain the 17% moisture content. For the tailing–Portland cement specimens, tailings and cement were mixed first, followed by the addition of distilled water to form a homogeneous paste.

The GMT–water specimens were manually compacted using a split mold. A membrane was first inserted into the mold, and a vacuum was applied to ensure proper adhesion. The tailing–water mixture was then compacted in three layers using the moist tamping method described by Ladd [34], with each layer slightly under-compacted and the surface of the first and second layers scarified to promote bonding. Static compaction was performed in three layers within the same split mold for the binder-stabilized specimens (tailing–alkali-activated binder and tailing–Portland cement). Upon completion of the molding process, the GMT–water specimens were weighed within the mold (with the mold weight subtracted later), and their dimensions were recorded after the mold was removed and the specimens were placed in the triaxial apparatus. Binder-stabilized specimens were demolded, measured, and weighed. Only specimens within the dimensional tolerances, height within ±1.0 mm and diameter within ±0.5 mm of the target values, were selected for testing.

The GMT–water specimens were tested immediately after preparation, without any curing period. However, after the mold was removed, binder-stabilized specimens were sealed in a plastic bag to avoid moisture loss and cured in a humid chamber (23 ± 2 °C and 95% moisture) for 7 days.

2.5. Triaxial Tests

Isotropically undrained (CIU) and drained (CID) triaxial compression tests were conducted in the gold mining tailings (both in stabilized and non-stabilized samples) to evaluate the material’s behavior. Procedures for these tests’ execution followed the standards ASTM D4767 [35] and ASTM D7181 [36], respectively, and Viana da Fonseca et al. [37]. Triaxial compression tests were performed on a Humboldt HM-5020 triaxial loader (Humboldt Manufacturing Company, Elgin, IL, USA).

After the specimens were placed in the triaxial apparatus, the chamber was filled with distilled water. Saturation was initiated through a two-step percolation process: (i) carbon dioxide (CO₂) was flushed through the specimens for one hour to facilitate saturation, as CO₂ is more soluble in water than air; (ii) distilled water was then percolated for one hour. During both gas and water percolation, an effective stress of 15 kPa was maintained by applying a confining pressure of 25 kPa and a backpressure of 10 kPa.

Following percolation, backpressure saturation was performed in six incremental stages, increasing the backpressure by 40 kPa every 30 min until reaching a final confining pressure of 265 kPa and a backpressure of 250 kPa. Saturation was assessed using Skempton’s B-value [38], which indicates the degree of saturation based on the response of pore pressure to applied stress.

The average B-value for GMT samples was 0.99, which is greater than the minimum required B-value of 0.95 (ASTM D4767 [35] and ASTM D7181 [36]). However, the GMT samples stabilized with the alkali-activated binder and Portland cement presented average B-values of 0.83 and 0.74, respectively. Although [38] indicates that fully saturated soils typically present B-values close to 1.0, cemented or artificially stabilized soils frequently exhibit reduced B parameters. The lowering of B as stiffness increased reflects the lower compressibility of the mineral skeleton as the consolidation pressure and curing time are increased [39]. One example of low B-values associated with cement-stabilized clayey silt has been reported by [40]: the measured B-values for higher consolidation pressures ranged from 0.55 to 0.90 with an average value of 0.76.

Once fully saturated, specimens were isotropically consolidated to the target effective confining pressures before shearing. Shearing was conducted under both drained (CID) and undrained (CIU) conditions at a constant axial displacement rate of 2 mm/h. This shear rate was established after determining the loading rate that will allow pore pressure to dissipate. The t₅₀ and t₉₀, corresponding to 50% and 90% primary consolidation, respectively, were calculated during the consolidation phase to determine the shear rate, as preconized by ASTM D4767 [35] and ASTM D7181 [36].

Area and membrane corrections were applied following the procedure described by La Rochelle et al. [41]. Specimens were sheared to axial strains between 30% and 35%, while pore pressure, axial load, and displacement were continuously recorded throughout the tests. The CIU and CID test program is summarized in

Table 2. Triaxial CIU and CID tests conducted on non-stabilized and stabilized gold mine tailings (GMTs).

Material	Sample	P0′ (kPa)
Non-stabilized gold mine tailings	CIU-50-T	50
	CID-50-T
	CIU-100-T	100
	CID-100-T
	CIU-200-T	200
	CID-200-T
Gold mine tailings + alkali-activated binder	CIU-50-AA	50
	CID-50-AA
	CIU-100-AA	100
	CID-100-AA
	CIU-200-AA	200
	CID-200-AA
Gold mine tailings + Portland cement	CIU-50-PC	50
	CID-50-PC
	CIU-100-PC	100
	CID-100-PC
	CIU-200-PC	200
	CID-200-PC

. Low effective confining pressures of 50, 100, and 200 kPa were selected to simulate field-representative conditions of tailing storage facilities, which are often saturated and subjected to low confining stresses—critical conditions for assessing material stability [6].

2.6. SEM/EDS for Microstructure Characterization

Scanning electron microscopy (SEM) and energy-dispersive spectroscopy (EDS) were employed to analyze the microstructure and chemical composition of the stabilized samples, respectively. SEM imaging was conducted using a JEOL JSM-IT500HR (JEOL, Tokyo, Japan) microscope under the following conditions: backscattered electron (BSE) mode, 500× magnification, 20 kV accelerating voltage, and samples sputter-coated with a Pt/Pd alloy. EDS analysis was conducted using an Oxford Instruments system (Oxford Instruments, Abingdon, UK) with an Ultim Max detector and processed with Aztec software (version 6.1).

3. Results and Discussion

3.1. Triaxial Test Results

Figure 2, Figure 3, Figure 4 and Figure 5 present the stress–strain curves, volumetric strain curves, pore pressure responses, and p′–q stress paths, respectively, for non-stabilized and stabilized gold mine tailings subjected to CID and CIU triaxial tests under effective confining pressures of 50, 100, and 200 kPa.

Figure 2 shows that shear strength increases with higher effective confining pressures in CID and CIU tests for stabilized and non-stabilized GMT specimens. As illustrated in Figure 3 and Figure 4, the tendency for dilation decreases with increasing confining pressure across all tested samples. Most specimens exhibited an initial increase in pore pressure and volumetric strain within the axial strain range of approximately 2.5% to 10%, followed by highly negative pore pressures at larger strains. Including binders enhanced the generation of positive pore pressures while reducing the magnitude of negative pore pressures. The initial compressive behavior is attributed to particle rearrangement, followed by a dilative response associated with grain interlocking—a characteristic behavior of medium-density granular materials recently emphasized by Fourie et al. [42] for mine tailings.

As shown in Figure 2a, CIU tests on non-stabilized GMT specimens exhibited a strain-hardening response, consistent with the generation of negative pore pressures observed in Figure 4a. Moreover, CID tests displayed mild strain-softening behavior. Due to the development of negative pore pressures, the CIU tests resulted in higher deviatoric stress than the CID tests.

For GMTs stabilized with the alkali-activated binder (Figure 2b), both CID and CIU tests exhibited pronounced strain-softening responses. In the CIU tests, this behavior is attributed to a reduction in effective stress toward the end of shearing, likely caused by cementation degradation. Again, the CIU specimens developed higher deviatoric stresses due to negative pore pressure generation. Lastly, GMT specimens stabilized with Portland cement (Figure 2c) displayed strong strain-softening behavior and the highest overall strength, reaching 5.3 MPa at 200 kPa confining pressure, compared to 2.3 MPa for the GMT–alkali-activated specimens and 740 kPa for the non-stabilized GMTs at the same confining pressure.

The peak strength (q_peak) represents the maximum deviatoric stress for the tested sample, while the critical state (q_cs) represents the last measured deviatoric stress in the test.

Table 3. Comparative metrics between Portland cement and alkali-activated specimens across all confining pressures.

Type of Binder	Name	qpeak (kPa)	qcs (kPa)	Difference Between Peak and Critical State (qpeak − qcs)/qpeak	Average (qpeak − qcs)/qpeak
GMT	CID-50-T	167.03	133.36	20%	21%
	CID-100-T	326.13	248.82	24%
	CID-200-T	606.11	497.66	18%
	CIU-50-T	777.70	744.73	4%	2%
	CIU-100-T	646.05	635.15	2%
	CIU-200-T	739.88	733.28	1%
GMT stabilized with the alkali-activated binder	CID-50-AA	919.12	364.45	60%	59%
	CID-100-AA	1473.68	572.10	61%
	CID-200-AA	2304.95	1037.80	55%
	CIU-50-AA	813.58	594.19	27%	30%
	CIU-100-AA	1656.04	1211.24	27%
	CIU-200-AA	2319.33	1472.86	36%
GMT stabilized with Portland cement	CID-50-PC	5084.54	859.41	83%	78%
	CID-100-PC	5022.77	1072.13	79%
	CID-200-PC	5372.07	1542.08	71%
	CIU-50-PC	4186.61	1549.67	63%	55%
	CIU-100-PC	3782.79	2031.38	46%
	CIU-200-PC	5022.44	2266.80	55%

provides comparative metrics between Portland cement and alkali-activated specimens across all confining pressures. Regarding the CID tests performed on GMTs, the peak strength reduced by an average of 21% compared to the residual strength. As for the CIU tests, the peak strength dropped by an average of 2%. For the CID triaxial tests performed in GMTs stabilized with the alkali-activated binder, the peak strength reduced by an average of 59% compared to the residual strength, while for CIU tests, the peak strength dropped by an average of 30%. Finally, the CID test performed in GMTs stabilized with Portland cement showed that the peak strength reduced by an average of 78% compared to the residual strength, while for CIU tests, the peak strength dropped by an average of 55%. This agrees with the aforementioned observations.

Figure 5 presents the p′–q stress paths for non-stabilized and stabilized gold mine tailings. The peak effective friction angle and cohesion intercept were higher for the GMT–Portland cement samples (ϕ′_peak = 67.3º, c′ = 162.2 kPa), followed by GMT–alkali-activated samples (ϕ′_peak = 61.6º, c′ = 29.7 kPa) and non-stabilized GMT samples (ϕ′_peak = 35.7º, c′ = 0 kPa). Also, the stress ratio, q/p′, at critical state M_tc as well as the critical state frictional angle ϕ′_cs increased with binder addition: M_tc = 1.39 and ϕ′_cs = 34.3º for non-stabilized GMT samples, M_tc = 1.85 and ϕ′_cs = 44.9º for GMT–alkali-activated samples, and M_tc = 2.14 and ϕ′_cs = 52.00º for GMT–Portland cement samples.

According to Carvalho et al. [43], forming interparticle bonds induced by binder addition enhances the peak friction angle and cohesion intercept, leading to a cohesive–frictional strength response. This bonding effect accounts for the higher peak strength parameters observed in the GMT–Portland cement and GMT–alkali-activated specimens compared to the non-stabilized samples. Consequently, the incorporation of binders has the potential to improve the stability and safety of tailing storage facilities significantly.

As noted, the behavior of cemented tailings is not purely frictional; however, stress–dilatancy analysis (q/p′ vs. δε_v/δε_s) remains a valuable tool for interpreting their shear behavior. Before bond breakage, the response is predominantly elastic and governed entirely by the cementation bonds, resulting in an almost vertical stress path referred to as the “cemented trend” or “cohesive trend”. Once bond degradation initiates, the material exhibits pronounced dilation, progressing toward the peak stress ratio—characterized as the “cohesive + frictional trend”. Following peak strength, strain softening occurs, and the response transitions to a “frictional trend”. Ultimately, the material approaches a dilatancy rate of zero, corresponding to the critical state condition [44,45].

Figure 6 presents the stress–dilatancy relationships during shearing for non-stabilized and stabilized gold mine tailings. Data for the highest confining pressure (200 kPa) were excluded to enhance the clarity of the plotted results.

As expected, the non-stabilized GMT specimens did not exhibit a “cemented trend”. Instead, the material initially compressed and then dilated, consistent with the behavior of a typical frictional granular material. In contrast, the GMT–alkali–activated and GMT–Portland cement specimens displayed a distinct “cemented trend”, which persisted up to q/p′ values of approximately 2.1 and 2.6, respectively. The dilation rates for these stabilized specimens were higher than those of the non-stabilized samples, as the initial bonding in cemented tailings suppresses early dilation, leading to a more abrupt dilative response once bond breakage initiates. The enhanced shear resistance of the cemented specimens also contributes to the extended and elevated “cemented trend”. After reaching the peak stress ratio, all specimens transitioned to a purely frictional behavior, characterized by a marked reduction in dilatancy and stress ratios.

According to Robertson [46], based on case histories, most flow liquefaction failures occur in young, uncemented, non-plastic, or low-plasticity soils. Given that the studied gold mine tailings (GMTs) are non-plastic, they are susceptible to liquefaction if they exhibit contractive behavior under saturated field conditions and undergo strain-softening when subjected to static loading. In this context, adding binders promotes interparticle cementation, significantly reducing the potential for flow liquefaction by enhancing the material’s strength and stiffness.

Finally, Tonini de Araújo et al. [47] conducted a comparative assessment of the social, environmental, and economic impacts of two soil stabilization strategies: (i) the alkali-activated binder used in this study—comprising SCBA and HEL activated by NaOH—and (ii) conventional Portland cement. For the life cycle analysis, the authors considered a functional unit of 1m³ of stabilized soil that should present comparable strength values. Thus, an unconfined compressive strength value of 2.1 MPa was defined as the target (7 days of curing at 23 °C). Based on this, two dosages were defined within each cement type: high-density/low-binder content (HDLB) and low-density/high-binder content (LDHB).

The study used a proportional scale, assigning a value of 1 to the highest impact dosage in each category to calculate the relative impact of the others. Low-density, high-binder content mixtures (AA-LDHB and PC-LDHB) exhibited a larger environmental footprint than high-density, low-binder content mixtures (AA-HDLB and PC-HDLB), primarily due to the higher quantity of raw materials needed for the former. The PC-LDHB mixture had the most significant environmental impact in six out of ten categories, including global warming and human toxicity, while the AA-LDHB mixture had the highest impact in the remaining four, such as acidification and ozone layer depletion. The PC-HDLB and AA-HDLB dosages, in that order, were identified as the least impactful. These results are presented in

Table 4. Environmental impacts for every dosage option.

Dosage	Abiotic Depletion	Global Warming	Ozone Layer	Human Toxicity	Fresh Water	Marine Aquatic	Terrestrial Ecotoxicity	Photochemical Oxidation	Acidification	Eutrophication
AA-LDHB	0.951	0.625	1	0.937	0.704	0.673	0.293	1	1	1
AA-HDLB	0.745	0.478	0.762	0.716	0.546	0.527	0.224	0.761	0.746	0.748
PC-LDHB	1	1	0.25	1	1	1	1	0.627	0.556	0.587
PC-HDLB	0.737	0.737	0.186	0.738	0.737	0.737	0.736	0.463	0.411	0.434

.

Thus, Tonini de Araújo et al.’s [47] analysis concluded that the alkali-activated binder represents a more sustainable alternative. Therefore, stabilizing gold mine tailings (GMTs) with this binder enhances the structural safety of tailing dams and offers a greener, more sustainable substitute for Portland cement.

While the aforementioned study focused on improving the mechanical properties of mine tailings, it is important to note that such materials may contain leachable contaminants, including heavy metals, at potentially hazardous levels. The applied stabilization technique also promises to reduce contaminant mobility and leachability, rendering the tailings non-hazardous.

In this context, Ferrazzo et al. [33] investigated how a mixture of an alkali-activated binder (AAB) and traditional Portland cement (PC) affected the encapsulation of metals in waste foundry sand (WFS). The AAB was the same as that studied by Tonini de Araújo et al. [47], i.e., composed of SCBA/HEL at a ratio of 80/20 and NaOH solution (2.61% Na₂O), with a water/binder rate equal to 0.8. The authors evaluated the metal leaching behavior of both the WFS-AAB and WFS-PC mixtures through batch and column tests, following the NBR 10005 [48] and ASTM D4874 [49] standards, respectively. Both mixtures successfully eliminated metal toxicity, but the WFS-AAB matrices were particularly effective at encapsulating heavy metals like cadmium (Cd), chromium (Cr), and lead (Pb) from both the WFS and the sugar cane bagasse ash.

3.2. Microstructural Changes

Figure 7, Figure 8 and Figure 9 present the SEM-EDS analyses for non-stabilized GMTs, GMTs stabilized with the alkali-activated binder, and GMTs stabilized with Portland cement, respectively, all captured at 1000× magnification.

SEM imaging reveals that the morphological structure of the tailings consists of minerals with irregularly shaped grains. These grains typically coexist in two typical forms: smooth-surfaced particles (point A), primarily composed of SiO₂ and Al₂O₃, and rough-surfaced particles (point B), dominated by Fe₂O₃ [50]. The EDS analysis produced a chemical map (Figure 7c) and elemental distribution profile (Figure 7d), indicating a significant presence of silicon (Si), iron (Fe), and aluminum (Al). Among these, silicon exhibited the highest concentration, followed by iron and aluminum, with a relatively uniform distribution throughout the sample.

For the GMT–alkali-activated sample (Figure 8), the SEM image reveals tailing particles embedded within a cementitious matrix composed of reaction products exhibiting notable compositional heterogeneity and an amorphous morphology. The EDS-generated chemical map and elemental distribution (Figure 8c,d) highlight the predominant presence of silicon (Si), calcium (Ca), aluminum (Al), and sodium (Na). These elements are indicative of the coexistence of C–S–H and (C,N)–A–S–H gels, with sodium originating from the alkaline activator. Additionally, remnants of sugarcane bagasse (point A) are visible, suggesting that some organic material remained unreacted during the alkali activation process.

For the GMT–Portland cement sample (Figure 9), the SEM image reveals a smoother and more homogeneous microstructure compared to the alkali-activated sample. The tailings are embedded within a cementitious matrix characterized by the presence of needle-like and reticular C–S–H phases, along with calcium hydroxide crystals (point A) [51]. The EDS chemical map and elemental distribution (Figure 9c,d) indicate the predominant presence of silicon (Si), calcium (Ca), aluminum (Al), and sodium (Na). These elements suggest the formation of C–S–H gel, with possible traces of (C,N)–A–S–H gel, similar to that observed in the alkali-activated samples.

The cementitious reaction products observed in Figure 8 and Figure 9 account for the enhanced shear strength of the stabilized samples relative to the non-stabilized GMTs, as evidenced by the triaxial test results.

4. Conclusions

Triaxial CID and CIU tests, along with SEM-EDS analyses, were conducted to evaluate the mechanical behavior of gold mine tailings (GMTs) stabilized with an alkali-activated binder and Portland cement, in comparison to non-stabilized GMTs. The following conclusions can be drawn:

• Initially, all specimens exhibited positive pore pressure and volumetric strain, followed by the development of highly negative pore pressures at larger strains. The addition of binder increased the generation of positive pore pressure and reduced the extent of negative pore pressure development.
• CIU tests on non-stabilized GMTs exhibited a strain-hardening response, consistent with the generation of negative pore pressures. In contrast, CID tests showed mild strain-softening behavior. Specimens stabilized with the alkali-activated binder and Portland cement demonstrated pronounced strain-softening responses in both CID and CIU tests.
• The GMT–Portland cement specimens exhibited pronounced strain-softening behavior and the highest strength, reaching 5.3 MPa at 200 kPa confining pressure, compared to 2.3 MPa for GMT–alkali-activated specimens and 740 kPa for non-stabilized GMT specimens under the same confining pressure.
• The peak angle of shearing resistance at effective stresses was higher for the GMT–Portland cement samples (ϕ′_peak = 67.3º), followed by GMT–alkali-activated samples (ϕ′_peak = 61.6º) and non-stabilized GMT samples (ϕ′_peak = 35.7º).
• The critical state line (CSL) stress ratio (M_tc), critical state friction angle (ϕ′_cs), and cohesion intercept (c′) all increased with binder addition: for non-stabilized GMT, M_tc = 1.39, ϕ′_cs = 34.3°, and c′ = 0 kPa; for GMT–alkali-activated specimens, M_tc = 1.78, ϕ′_cs = 43.4°, and c′ = 19.7 kPa; and for GMT–Portland cement specimens, M_tc = 1.80, ϕ′_cs = 43.8°, and c′ = 145.1 kPa.
• The GMT–alkali-activated and GMT–Portland cement specimens exhibited higher dilation rates than the non-stabilized tailings, as interparticle bonding in cemented tailings initially suppresses dilatancy. Additionally, the greater shear resistance of the cemented specimens contributes to a more pronounced “cemented trend”.
• SEM imaging reveals that the tailings’ morphological structure consists of minerals with irregularly shaped grains, which occur in two typical forms: smooth-surfaced grains primarily composed of SiO₂ and Al₂O₃, and rough-surfaced grains dominated by Fe₂O₃.
• For the GMT–alkali-activated sample, SEM imaging shows tailings particles embedded within a cementitious matrix formed by reaction products exhibiting significant compositional heterogeneity and an amorphous structure. The chemical map indicates the coexistence of C–S–H and (C,N)–A–S–H gels, with the latter incorporating sodium cations from the alkaline activator.
• The SEM image of the GMT–Portland cement sample reveals a smoother and more homogeneous surface compared to the alkali-activated sample. The tailings are embedded within a cementitious matrix composed of needle-like and reticular C–S–H phases, along with calcium hydroxide crystals.
• The higher peak angle of the GMT–alkali-activated and GMT–Portland cement samples compared to non-stabilized samples shows that binder addition increases the safety of stackings. The cementitious reaction products seen in the SEM-EDS analysis explain the higher shear resistance of the stabilized samples.
• The selection of a 7-day curing period was based on the material’s observed rapid strength development, which meets the minimum requirements for practical applications. The results demonstrate that the stabilized tailings achieve a compressive strength of 2.39 MPa within this timeframe, exceeding the 2.1 MPa minimum for soil–cement bases as stipulated by DNIT [28]. This finding validates the material’s viability for field-level applications in a shorter period. However, the study’s limitations, such as the short curing period and the absence of cyclic loading tests, are acknowledged. These factors are critical for evaluating the material’s long-term durability and resistance to liquefaction under dynamic conditions. Therefore, future work should focus on assessing these aspects and conducting field-scale validation to further translate these laboratory findings into practical engineering solutions.