The Depression Effect of Micromolecular Depressant Containing Amino and Phosphonic Acid Group on Serpentine in the Flotation of Low-Grade Nickel Sulphide Ore

Abstract

Selective depression of serpentine remains a major challenge in the flotation of low-grade nickel sulphide ores because serpentine slimes impair concentrate grade and recovery. In this study, four structurally related micromolecular depressants bearing amino and phosphonic functionalities were designed, synthesized and systematically evaluated. Micro-flotation screening (depressant range: 0–20 mg·L⁻¹) and bench-scale tests identified an operational optimum near pH 9 and a reagent dosage of ≈18 mg·L⁻¹; potassium butyl xanthate (PBX) was used as a collector and methyl isobutyl carbinol (MIBC) as a frother. Phosphonate-containing molecules (PMIDA and GLY) delivered the largest gains in pentlandite recovery and concentrate selectivity compared with carboxylate analogues and a benchmark depressant. Mechanistic studies (zeta potential, adsorption isotherms, FT-IR, and XPS) indicated that selective adsorption of amino and phosphonate groups on serpentine occurs via coordination with surface Mg sites and by altering the electrical double layer. The DLVO modelling showed that these reagents generate an increased repulsive barrier, mitigating slime coating and entrainment. Contact-angle measurements confirmed selective hydrophilization of serpentine while pentlandite remained hydrophobic. These findings demonstrate that incorporating targeted phosphonate chelation into small-molecule depressants is an effective strategy to control serpentine interference and to enhance flotation performance.

1. Introduction

As the rapid consumption of traditional high-grade nickel sulphide ore and the technological problem of utilization of laterite ore, the interest in the processing and utilization of low-grade nickel sulphide ore has increased in recent years [1,2]. Compared with the difficulties caused by the low concentrations of valuable metals such as nickel and copper, the problems arising from a high proportion of magnesium-silicate gangue are often more problematic for the flotation of low-grade nickel sulphide ores [3,4]. A typical example of magnesium-silicate gangue is serpentine, a relatively soft phyllosilicate (Mohs hardness = 2.5–4) that is readily comminuted during grinding and pulp stirring into fines and slimes (typically <10 μm), thereby complicating flotation [5,6]. These serpentine slimes and fines interfere with the flotation enrichment of pentlandite (the main nickel-bearing mineral in nickel sulphide ore) through two paths: heterogeneous condensation (deposition of fine gangue (serpentine) onto sulfide grains, forming slime coatings that block collector adsorption and reduce bubble-particle attachment) and foam entrainment. These two paths lead to two consequences: 1. hydrophobic surface of pentlandite is coated by serpentine slimes and transformed into hydrophilic surface [7,8]; 2. the Ni grade of flotation concentrate is diluted by a lot of entrained serpentine fines [9,10].

Depressants play an important role in eliminating the adverse effect of serpentine on the flotation enrichment of pentlandite. Experientially, the development and improvement of depressants for serpentine focuses on the following two points: the optimization of molecular size and functional groups [11,12]. For the molecular size, most of prior depressants for serpentine belong to high-molecular polymers, such as starch, dextrin, guar gum, N-carboxymethyl chitosan, carboxymethylcellulose, and other polysaccharides derivatives [13,14,15]. However, high intensity agitation is always necessary to remove serpentine slime from the surface of pentlandite. In this respect, the use of high-molecular polymers is disadvantageous [16]. Therefore, diverse micromolecular depressants for serpentine have been developed in recent years [11,13,17,18,19]. On the other hand, regarding functional groups, carboxyl is the most common functional group contained in known depressants. However, the depression selectivity of monotypic depressants toward magnesium active sites on the serpentine surface is limited, which could also exert a certain degree of depression on pentlandite. Therefore, in recent years, potent chelating groups have been proposed to be introduced into the molecular structure of micromolecular depressants.

For the enhanced depression effect profited from chelating groups, tetrasodium iminodisuccinate is a persuasive case. Compared with oxalic acid, a traditional micromolecular carboxylic acid depressant, the additionally introduced amino group could enhance the depression effect dramatically. At the same depression efficiency (90%), the dosage by using tetrasodium iminodisuccinate was significantly decreased to 10% (Figure 1) [11,13,17,18,19]. Besides the amino group, it is worth noting that the phosphonic acid group is also a highly selective group. Several inorganic and polymeric reagents have been reported as selective serpentine depressants, for example sodium phytate, sodium hexametaphosphate, and phosphorylated starch [16]. Therefore, introducing amino and phosphonic acid groups into micromolecular carboxylic acid depressants is expected to improve selectivity. In this work, four different micromolecular carboxylic acid depressants including amino groups or phosphonic acid groups were selected as candidate depressants. The aim of the present work is to screen an efficient micromolecular depressant and to reveal the depression mechanism of the optimized depressant on serpentine in the flotation of nickel sulphide ore. Comparative studies using oxalic acid are also performed to evaluate the superiority of the optimized depressant.

2. Materials and Methods

2.1. Samples and Reagents

The serpentine and pentlandite pure mineral samples and the batch sulfide ore used in this study were all sourced from Jinchuan, Gansu Province, China. Serpentine was finely ground in an agate ball mill and characterized by laser particle size analysis to obtain the fine fractions required for surface studies. Pentlandite was crushed with a hammer, then dry-ground in a ceramic ball mill and dry-sieved to produce size fractions appropriate for micro-flotation and surface analyses. Representative X-ray diffraction (XRD) patterns confirming the mineral phases are presented in Figure 2. The actual sulfide ore sampled from Jinchuan that was used in the batch flotation tests is summarized in

Table 1. Chemical composition analysis of mineral samples.

Sample	SiO2	MgO	CaO	Al2O3	Fe	Ni	S
Serpentine	41.77	36.06	0.23	0.88	5.41	-	-
Pentlandite	-	-	-	-	27.15	29.28	30.59

.

Micro-flotation experiments employed the 38–74 μm size fraction obtained by sieving, while fractions below 10 μm were reserved for Fourier-transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), zeta potential, and adsorption measurements. Potassium butyl xanthate (PBX) was used as the sulfide ore collector in flotation tests. Four structurally related micromolecular depressants bearing amino and phosphonic acid functionalities were synthesized in the laboratory; full synthetic procedures and characterization are provided in Section 2.3. Methyl isobutyl carbinol (MIBC) served as the frother. Analytical-grade hydrochloric acid (HCl) and sodium hydroxide (NaOH) were used to adjust pH in flotation and adsorption experiments. Potassium nitrate (KNO₃) was used as the background electrolyte for zeta-potential measurements, and potassium bromide (KBr) was used for FTIR pellet preparation. Milli-Q deionized water with a resistivity of 18.2 MΩ·cm was used throughout all experiments to ensure consistency. All chemical reagents were of analytical grade.

2.2. Synthesis and Characterization of the Dioxime Collector

2.2.1. Synthesis Procedure

The synthesis of the four depressants was carried out as follows:

• Iminodiacetic acid (IDA)

Iminodiacetic acid was obtained by hydrolysis of iminodiacetonitrile (IDAN). In a typical run, IDAN was dissolved in a water/ethanol mixture and the solution was heated to 70–90 °C. Aqueous sodium hydroxide (3–4 equiv) was added and the mixture was stirred until reaction completion as monitored by TLC or NMR (several hours). The mixture was cooled and acidified with dilute HCl to pH ≈ 2 to precipitate the free acid. Inorganic salts were removed by filtration, the aqueous phase was concentrated under reduced pressure, and the crude product was recrystallized from ethanol/water to afford IDA as white crystals.

• Nitrilotriacetic acid (NTA)

NTA was prepared by successive alkylation of ammonia with chloroacetic acid (or its halo-derivative) followed by acidification and purification. Briefly, an aqueous solution of ammonia (excess) was cooled and sodium chloroacetate (3.0 equiv) was added portionwise under stirring at 0–5 °C to control exotherm. The reaction was allowed to warm to ambient temperature and then heated to 60–80 °C for several hours to complete alkylation. After reaction, the mixture was neutralized and inorganic salts were removed by filtration. The filtrate was concentrated and acidified to precipitate NTA, which was washed and recrystallized from ethanol/water. Typical isolated yields were 60%–80%. The structure was confirmed by NMR and elemental analysis.

• N-(phosphonomethyl)iminodiacetic acid (PMIDA)

PMIDA was synthesized from IDA via a phosphonomethylation (Mannich-type) sequence. The alkali salt of IDA was dissolved in water and cooled to 0–5 °C. Phosphorous acid (H₃PO₃) or phosphorus trichloride (PCl₃) was added dropwise under stirring, followed by the controlled addition of an aqueous formaldehyde source (paraformaldehyde or formalin) to introduce the methylene linker. The reaction mixture was kept at 0–5 °C during addition to control exotherm and then warmed to 50–70 °C for 2–6 h to complete conversion. After cooling, volatiles were removed, and the pH was adjusted to precipitate PMIDA or to convert it into a desired salt form. The crude product was purified by recrystallization from water/ethanol.

• N-(phosphonomethyl)glycine (GLY)

Glyphosate was prepared from PMIDA by oxidative/hydrolytic conversion following literature-type procedures. In a representative laboratory procedure, PMIDA was dissolved in dilute aqueous acid and subjected to oxidative or hydrolytic conditions (for example controlled H₂O₂ treatment or acid-mediated cleavage under reflux) to effect rearrangement/hydrolysis to yield the glycine derivative. After reaction completion, the mixture was neutralized and the product was isolated by crystallization as the free acid or as a crystalline salt.

2.2.2. Characterization Methods

The purity and molecular structure of the synthesized depressant were characterized using ¹H NMR and ¹³C NMR, as shown in Figure 3.

Iminodiacetic acid (IDA): ¹H NMR (400 MHz, Deuterium Oxide) δ = 2.25 (d, J = 2.5 Hz, 4H). ¹³C NMR (126 MHz, Deuterium Oxide): δ = 179.60, 51.81.

Nitrilotriacetic acid (NTA): ¹H NMR (400 MHz, Deuterium Oxide): δ = 2.18 (d, J = 4.0 Hz, 6H). ¹³C NMR (126 MHz, Deuterium Oxide): δ = 179.16, 58.92.

N-(phosphonomethyl)iminodiacetic acid (PMIDA): ¹H NMR (400 MHz, Deuterium Oxide): δ = 2.25 (s, 4H), 1.50 (d, J = 11.5 Hz, 2H). ¹³C NMR (101 MHz, Deuterium Oxide): δ = 179.71, 60.48, 54.10, 52.68.

N-(phosphonomethyl)glycine (GLY): ¹H NMR (400 MHz, Deuterium Oxide): δ = 2.29 (s, 1H), 1.58 (d, J = 12.8 Hz, 1H). ¹³C NMR (101 MHz, Deuterium Oxide): δ = 179.84, 54.09, 53.93, 48.10, 46.72.

The molecular structures of the four candidate depressants were confirmed by ¹H and ¹³C NMR spectroscopy in deuterium oxide, and the observed chemical shifts and splitting patterns are consistent with the proposed formulas. For iminodiacetic acid (IDA), the ¹H NMR spectrum (400 MHz, D₂O) shows a single resonance at δ 2.25 with apparent doublet character (J = 2.5 Hz) integrating to 4H, which is assigned to the two equivalent methylene groups (-CH₂-) connected to the central nitrogen [20]. The ¹³C spectrum (126 MHz, D₂O) exhibits a deshielded signal at δ 179.60, attributable to the two carboxylate/carbonyl carbons, and a methylene/carbon resonance adjacent to nitrogen at δ 51.81, in agreement with the IDA skeleton [21].

For nitrilotriacetic acid (NTA), the ¹H NMR (400 MHz, D₂O) displays a resonance at δ 2.18 (d, J = 4.0 Hz, 6H) corresponding to the three methylene groups linking the tertiary nitrogen to the three acetate units [22]. The ¹³C NMR (126 MHz, D₂O) shows the carboxylate carbon at δ 179.16 and a higher-field methylene carbon resonance at δ 58.92, consistent with N-substituted acetate carbons [23].

For N-(phosphonomethyl) glycine (glyphosate, GLY), the ¹H NMR (400 MHz, D₂O) contains a singlet at δ 2.29 (1H) and a doublet at δ 1.58 (J = 12.8 Hz, 1H). The large doublet coupling is characteristic of scalar coupling between methylene protons and the adjacent phosphorus nucleus, and therefore, the δ 1.58 signal is assigned to a proton on the phosphonomethyl fragment that experiences 2JH-P coupling, while the singlet at δ 2.29 is assigned to the glycine-derived proton environment [24]. The ¹³C NMR (101 MHz, D₂O) presents the expected carboxylate carbon at δ 179.84 and several methine/methylene carbons between δ 54.09 and 46.72 that match the multiple carbon environments in the phosphonomethylated glycine structure [25].

For N-(phosphonomethyl) iminodiacetic acid (PMIDA), the ¹H NMR (400 MHz, D₂O) shows signals at δ 2.25 (s, 4H) attributed to the pair of equivalent methylene groups adjacent to the nitrogen and at δ 1.50 (d, J = 11.5 Hz, 2H) which is assigned to the phosphonomethyl protons displaying coupling to phosphorus [26]. The ¹³C NMR (101 MHz, D₂O) again shows the carboxylate carbon near δ 179.71 and methylene carbons at δ 60.48, 54.10, and 52.68 that accord with the PMIDA framework and with substitution by a phosphonomethyl unit [27].

In all four cases, the ¹³C resonances near 179 ppm confirm the presence of carboxylate carbonyl groups, and the methylene/methine carbon chemical shifts in the 46–61 ppm range are consistent with carbons bound to nitrogen or phosphorus. The observed proton−phosphorus coupling patterns for the phosphonate-containing compounds further support correct placement of the phosphonomethyl groups. No extraneous resonances attributable to common organic impurities were detected, indicating that the synthesized compounds are of high purity and that the NMR data support the assigned structures.

2.3. Micro-Flotation Experiments

Micro-scale flotation trials were carried out in a 40 mL transparent cell mounted on an XFG-1600 flotation apparatus (Tankuang Ltd., Jilin, China) operated at 1800 rpm. Prior to each run, the cell was cleaned ultrasonically for 3 min. Two grams of pentlandite (38–74 μm) was placed into the cell with 40 mL of deionized water, and the predetermined mass of serpentine (4–10 μm) was introduced when required by the experimental design. Reagents were added in the following sequence: depressant (one of the four candidate molecules), collector (PBX), and frother (MIBC). After each addition, the slurry was conditioned for a fixed interval (see figure captions and results for individual conditioning times); pH adjustments were effected using analytical-grade HCl or NaOH and maintained at the target value throughout conditioning. Air was then supplied, and froth recovery proceeded for 3 min. The floated and non-floated fractions were collected, filtered, dried and weighed to determine recovery and grade. Every experimental condition was replicated at least three times to compute averages and deviations, which were then used to construct error bars.

2.4. Batch Flotation Experiments

Laboratory batch flotation tests were performed to evaluate reagent performance on bulk ore. The flowsheet of the laboratory-scale batch flotation is shown in Figure 4. One kilogram of run-of-mine material (−4 mm) was wet milled in an XMQ-240 × 90 ball mill (Tankuang Ltd., Jilin, China) to a nominal product with 70% passing 74 μm at a pulp density of 65% solids. Rougher and cleaner stages were simulated in an XFD-63 flotation machine (Tankuang Ltd., Jilin, China) fitted with a 3.0 L rougher cell and a 1.0 L cleaner cell. The impeller speed was set to 3000 rpm. After preparing the slurry, reagents were added and conditioned for 4 min prior to aeration. Air was introduced at 1.6 L/h and frothing was allowed for desired time; water was added as necessary to maintain a constant liquid level. Products from each stage (froth and tailings) were collected on filter paper, oven-dried, weighed and sub-sampled for chemical analysis. Mass balances, grades, and recoveries were calculated, and each batch test was repeated to verify reproducibility.

2.5. Zeta-Potential Measurements and Dlvo Calculation

Electrokinetic measurements were performed on a Zetasizer Nano ZS90 (Malvern Instruments, Nottingham, UK). A background electrolyte of 1.0 × 10⁻³ mol·L⁻¹ KNO₃ was used to control ionic strength. For each measurement, 30 mg of mineral sample (<5 μm) was dispersed in 40 mL of the KNO₃ solution and magnetically stirred for 10 min to ensure even suspension. The pH was adjusted to the same values used in flotation tests by adding HCl or NaOH. After pH stabilization, the selected depressant or collector was introduced and the suspension was conditioned for 5 min. The dispersion was allowed to stand for 5 min to permit coarse particles to settle; the supernatant was then withdrawn for zeta-potential determination. Every test was repeated at least three times to compute averages and deviations, which were then used to construct error bars.

To interpret the influence of surface charging and adsorbed reagents on interparticle interactions, the classical DLVO theory was applied to estimate the total interaction energy (V_T) as the sum of the electrostatic double-layer repulsion (V_E) and van der Waals attraction (V_W). Parameters used in the calculations included particle radii, surface potentials derived from measured zeta potentials, and literature Hamaker constants for the mineral−water−mineral systems. The inverse Debye length was computed from the background electrolyte concentration. As previously outlined in reports [28,29], the values of V_T, V_W, and V_E can be calculated using Equations (1)–(3) as follows:

V_T = V_W + V_E

(1)

$$V_W=-{\displaystyle \frac{A_{132}}{6H}} · {\displaystyle \frac{R_{1 }{R}_2}{R_1+R_2}}$$

(2)

$$V_E={\displaystyle \frac{\pi {\epsilon }_a{R}_1{R}_2}{R_1+R_2}} ·\left({\mathsf{\Psi }}_1^2+{\mathsf{\Psi }}_2^2\right)\left[{\displaystyle \frac{2{\mathsf{\Psi }}_1{\mathsf{\Psi }}_2}{{\mathsf{\Psi }}_1^2+{\mathsf{\Psi }}_2^2}}\ln \left({\displaystyle \frac{1+\exp \left(-{\lambda }_{D}H\right)}{1-\exp \left(-{\lambda }_{D}H\right)}}\right)\right]+\ln \left(1-\exp \left(-2{\lambda }_{D}H\right)\right)$$

(3)

In the DLVO framework, H denotes the separation distance between pentlandite and serpentine particles, for which the particle radii of R₁ = 35.4 μm (pentlandite) and R₂ = 3.1 μm (serpentine) were adopted based on measured size distributions. The parameter A₁₃₂ represents the Hamaker constant for the pentlandite−serpentine pair in pure water and was calculated according to Equation (3). Surface potentials Ψ₁ and Ψ₂, used in the electrostatic term, correspond to the pentlandite and serpentine potentials obtained from zeta-potential measurements. The permittivity of the medium is given by ε_a = ε₀ × ε_r, where ε₀ and ε_r are the vacuum permittivity and the relative permittivity of water, respectively (8.85 × 10⁻¹² and 7.85 × 10⁻¹² C²·J⁻¹·m⁻¹ for pure water) [30]. The inverse Debye length (λ_D⁻¹) was taken as 0.180 nm⁻¹ following previously reported values [31,32].

$$A_{132}=\left(\sqrt{A_{11}}-\sqrt{A_{33}}\right)\left(\sqrt{A_{22}}-\sqrt{A_{33}} \right)$$

(4)

Within Equation (4), A₁₁, A₂₂, and A₃₃ symbolize the Hamaker constants of pentlandite, serpentine, and pure water in a vacuum, respectively. As reported by K.E. Bremmell [7], A₁₁, A₂₂, and A₃₃ are 3.3 × 10⁻²⁰ J, 9.7 × 10⁻²⁰ J, and 3.7 × 10⁻²⁰ J, respectively.

2.6. Reagent Adsorption Measurements

Adsorption experiments were designed to quantify collector reagent (PBX) uptake on mineral surfaces under conditions equivalent to those used in flotation [33]. In a typical test, 2 g of mineral (38–74 μm for flotation-scale tests; <10 μm for surface studies when specified) was added to 40 mL of deionized water in a 100 mL beaker; the pH was adjusted to the target value and the suspension was stirred for 10 min. A known concentration of the collector was then added and the mixture was agitated for 10 min to approach adsorption equilibrium. After conditioning, the slurry was centrifuged (6000 rpm, 5 min) and the supernatant was collected for analysis. Residual collector reagent concentrations were determined by an ultraviolet−visible spectrophotometer (UV2600, Shimadzu Corporation, Kyoto, Japan). Adsorbed amounts were calculated from the difference between initial and final concentrations, accounting for solution volume and sample mass.

2.7. XPS Analysis

Surface chemical states before and after reagent exposure were examined by X-ray photoelectron spectroscopy using a Thermo Scientific K-Alpha instrument (Thermo Fisher Scientific, Waltham, MA, USA). Mineral powders (<2 μm) were conditioned with the reagent at the chosen pH for 20 min, centrifuged, rinsed three times with deionized water and dried at 40 °C under vacuum. Dried powders were mounted on conductive adhesive tape and introduced into the XPS chamber. Survey and high-resolution spectra were acquired under standard operating conditions. Binding energies were referenced to the C 1s peak (284.8 eV) [34]; shifts in binding energy and the appearance of diagnostic peaks were used to infer chemical interactions between reagents and mineral surface sites.

2.8. Contact Angle Measurements

The wettability of mineral surfaces was evaluated by static contact angle determination using a JC-2000A contact angle analyzer (Dingsheng Ltd., Chengde, China). Intact block specimens were used for contact angle measurements; each original rock block was sectioned and then lapped and polished into dedicated ore wafers to produce smooth, flat surfaces suitable for wettability testing. On each pellet, a droplet of deionized water (2 μL) was deposited using a microsyringe and the contact angle was recorded immediately.

3. Results and Discussion

3.1. Micro-Flotation Experiments

To examine the influence of serpentine on pentlandite flotation, a series of mixed-mineral micro-flotation tests were performed by varying the serpentine-to-pentlandite ratio; the results are displayed in Figure 5. In addition, the effects of four candidate depressants (IDA, NTA, PMIDA, and GLY) on the mixed pentlandite−serpentine system were investigated across different slurry pH values and reagent dosages; those outcomes are presented in Figure 6 and Figure 7.

As shown in Figure 5, increasing the amount of serpentine produced a pronounced decrease in pentlandite recovery. The introduction of only 1 g/L serpentine reduced pentlandite recovery from above 90% to below 75%. When the serpentine loading was raised to 5 g/L, pentlandite recovery fell further to near 40%. This behavior demonstrates a strong adverse effect of serpentine contamination on pentlandite flotation. Moreover, a saturation trend was apparent: beyond approximately 5 g/L, the rate of recovery loss slowed and approached a plateau, implying that additional serpentine no longer produced a proportional decrease in pentlandite floatability. Accordingly, a serpentine addition of 5 g/L (corresponding to the test condition of 2 g pentlandite + 0.2 g serpentine in the micro-flotation protocol) was adopted for the subsequent pH and depressant dosage studies.

The flotation performance of pentlandite was found to be highly pH-dependent, and an optimum was observed near pH 9 under test conditions. In the absence of serpentine, the pentlandite recovery remained high and relatively stable from pH 2 to 9 but deteriorated markedly in the strongly alkaline region (pH 9–12) where recovery fell toward 60%. When serpentine was present (5 g/L), the negative effect became more severe and the lowest recoveries occurred under strongly alkaline conditions. The addition of 10 mg/L depressant substantially improved recovery at pH 9 (for example, recoveries at pH 9 rose to 66.2% with IDA, 45.1% with NTA, 75.3% with PMIDA, and 87.6% with GLY), indicating that near-neutral to mildly alkaline conditions favor selective suppression of serpentine while retaining collector activity on pentlandite. Mechanistically, the loss of pentlandite floatability at higher pH can be explained by surface hydrolysis of Fe and Ni active sites to form hydrophilic metal−OH species and by increased competition from OH⁻ for surface coordination sites, both of which impede xanthate chemisorption or promote xanthate desorption. In contrast, around pH 9, the balance between collector activity and depressant−serpentine interaction is most favorable: the depressants can effectively complex surface Mg sites on serpentine, driving a negative shift in serpentine surface charge, while pentlandite retains sufficient active metal sites for xanthate adsorption. For these reasons and to maintain consistency across subsequent tests, pH 9 was adopted as the working condition in our comparative reagent studies.

Figure 7 presents the dependence of pentlandite recovery on depressant dosage at pH 9 with 5 g/L serpentine. Recoveries for all four reagents rose markedly as dosage increased from 0 to 20 mg/L, achieving maxima of approximately 73.6% (IDA), 65.3% (NTA), 82.8% (PMIDA), and 79.5% (GLY). These increases reflect successful suppression of serpentine’s adhesive coverage on pentlandite at low to moderate depressant concentrations. However, further increasing depressant dosage beyond the optimum led to a rapid decline in pentlandite recovery. At 20 mg/L, recoveries dropped to about 3% for the most affected cases. This decline can be ascribed to excessive depressant adsorbing onto pentlandite itself, increasing its hydrophilicity and impairing collector attachment; additionally, high reagent loadings raise pulp viscosity and interfere with bubble formation and transport. Therefore, selecting an intermediate depressant dosage is critical to balance serpentine suppression with preservation of pentlandite floatability.

When benchmarked against the conventional depressant sodium hexametaphosphate (SHMP) under identical conditions, PMIDA and GLY achieved comparable or superior restoration of pentlandite recovery at the optimal dosage, whereas IDA and NTA were less effective than SHMP. The superior performance of PMIDA/GLY can be rationalized by stronger Phosphonate-Mg chelation on serpentine and a higher affinity for surface metal sites, which allows effective serpentine passivation at lower dosages and reduces the risk of pentlandite self-depression. In practice, therefore, an intermediate reagent concentration (18 mg/L) balances effective serpentine suppression with minimal adverse impact on pentlandite floatability.

3.2. Batch Flotation Experiments

Following the existing plant flowsheet, bench-scale closed-circuit flotation experiments were carried out on the Jinchuan copper−nickel ore. The experimental flowsheet is presented in Figure 4, and the closed-circuit test results are compiled in

Table 2. The results of the bench-scale closed-flotation experiment.

Depressants	Products	Yields (%)	Grades (%)		Recovery (%)
			Ni	Cu	Ni	Cu
No Depressants	Concentrate	18.07	3.75	2.15	49.91	34.26
	Tailing	81.93	0.83	0.91	50.09	65.74
	Feed	100	1.36	1.13	100	100
IDA	Concentrate	11.39	7.18	4.11	59.43	44.84
	Tailing	88.61	0.63	0.65	40.57	55.16
	Feed	100	1.38	1.04	100	100
NTA	Concentrate	11.33	7.22	4.19	59.81	45.94
	Tailing	88.67	0.62	0.63	40.19	54.06
	Feed	100	1.37	1.03	100	100
GLY	Concentrate	11.19	7.39	4.32	60.42	46.75
	Tailing	88.81	0.61	0.62	39.58	53.25
	Feed	100	1.37	1.03	100	100
PMIDA	Concentrate	11.13	7.45	4.39	61.26	47.82
	Tailing	88.87	0.59	0.6	38.74	52.18
	Feed	100	1.35	1.02	100	100

.

As shown in Table 2, all four small-molecule depressants produced measurable improvements in the beneficiation indices of the Jinchuan copper−nickel ore, with phosphonate-bearing reagents (GLY and PMIDA) giving the most pronounced enhancements. Using IDA as a reference, NTA produced only marginal gains in both grade and recovery. By contrast, GLY increased the Ni grade in the total concentrate from 7.18% to 7.39% and the Cu grade from 4.11% to 4.32%, while Ni and Cu recoveries rose from 59.43% and 44.84% to 60.42% and 46.75%, respectively. PMIDA delivered the best overall performance: the Ni and Cu grades in the concentrate reached 7.45% and 4.39%, representing increases of 0.27 and 0.28 percentage points relative to IDA, and the Ni and Cu recoveries improved to 61.26% and 47.82%, i.e., gains of 1.83 and 2.98 percentage points, respectively. Concentrate yields were comparable across reagents (11.1%–11.4%), indicating that the observed grade and recovery improvements reflect genuine selectivity gains rather than simple mass pull changes. These results demonstrate that incorporation of phosphonate functionality (in PMIDA and GLY) markedly enhances the separation of serpentine gangue from nickel−copper sulfides under the tested conditions, making PMIDA and GLY promising candidates for practical application in the flotation of this ore.

3.3. Zeta-Potential Measurements and Dlvo Calculation

Because electrostatic interactions are a dominant factor governing the heteroaggregation of pentlandite and serpentine, zeta-potential measurements were performed to elucidate how the four depressants influence surface charge and thereby affect flotation behavior. The zeta-potential data are presented in Figure 8.

As shown in Figure 8, untreated serpentine carried a net positive potential over much of the acidic to near-neutral pH range, whereas pentlandite remained negatively charged between pH 3 and 11. This contrast in surface charge fosters a strong electrostatic attraction that promotes serpentine adsorption onto pentlandite and consequently reduces pentlandite floatability. Upon addition of 10 mg/L of the tested reagents, both mineral surfaces exhibited negative shifts in zeta potential, but the magnitude of the shift was markedly larger for serpentine than for pentlandite. At pH 9, the zeta potential of pentlandite changed only slightly from −30.2 mV (untreated) to −29.1, −29.4, −28.5, and −29.7 mV after dosing with IDA, NTA, PMIDA, and GLY, respectively. By contrast, serpentine shifted from +11.0 mV (untreated) to −1.8, −9.6, −22.4, and −25.9 mV with the same reagent additions. The stronger negative shift induced by PMIDA and GLY indicates a more effective neutralization and inversion of serpentine surface charge, which we attribute to coordination of phosphonate groups with surface Mg²⁺ sites and to enhanced anionic adsorption that screens positive surface sites [35]. The comparatively small change on pentlandite suggests limited direct adsorption of these small-molecule depressants onto sulfide metal sites under the tested conditions.

Given that modifications of surface charge directly alter interparticle electrostatic energy, the classical DLVO theory was applied to quantify total interaction energy profiles between pentlandite and serpentine as a function of separation distance [36]. The calculated results are shown in Figure 9. In the absence of depressants, the total interaction energy remained negative across the examined separation range, indicating a net attractive interaction and a propensity for heteroaggregation and surface coating. After treatment with 18 mg/L of each reagent, the DLVO profiles changed substantially. For IDA and NTA, the total interaction energy became less negative but did not reach positive values until very short separations, implying only partial dispersion. In contrast, PMIDA and GLY produced a pronounced increase in the total interaction energy: the interaction curve crossed from negative to positive at separation distances near 1.3 nm for PMIDA and near 0.8–1.0 nm for GLY, signifying that these reagents can generate a repulsive barrier that prevents close approach and adhesion of serpentine onto pentlandite. The greater dispersing effect of the phosphonate-containing reagents is consistent with their larger negative shifts of serpentine zeta potential and with the improved pentlandite recoveries observed in flotation tests. Therefore, the DLVO calculations corroborate the zeta-potential measurements and demonstrate that PMIDA and GLY are more effective than IDA and NTA at converting attractive pentlandite−serpentine interactions into repulsive ones, thereby mitigating slime coating and enhancing selective flotation.

3.4. Reagent Adsorption Measurements

To quantify how serpentine and the four candidate depressants influence collector uptake, PBX adsorption onto pentlandite and serpentine was measured as a function of PBX concentration at pH 9; the results are presented in Figure 10. In the absence of serpentine, PBX uptake on pentlandite increased with dosing and approached a plateau near 0.30 mg/g, with the adsorption rate slowing at PBX concentrations in the 25–50 mg/L range, indicative of surface site saturation. By contrast, pure serpentine exhibited negligible affinity for PBX across the examined range, with adsorbed amounts remaining below 0.02 mg/g, confirming that serpentine is intrinsically non-collectible by xanthate under these conditions.

Introduction of serpentine (5 g/L) into the system caused a pronounced suppression of PBX uptake on pentlandite: the measured adsorption fell to well below 0.08 mg/g, consistent with surface masking or competitive blocking by fine serpentine particles. When 10 mg/L of the novel depressants was applied, PBX adsorption on pentlandite recovered to different extents. Phosphonate-bearing reagents (PMIDA and GLY) restored PBX uptake most effectively, recovering pentlandite adsorption to approximately 0.25–0.30 mg/g. The carboxylate-type reagents (IDA and NTA) produced only partial recovery (0.10–0.15 mg/g), demonstrating weaker mitigation of serpentine interference.

These adsorption trends corroborate the electrokinetic and flotation data: PMIDA and GLY preferentially adsorb to serpentine, likely via strong phosphonate-Mg chelation, neutralizing positive surface sites and preventing serpentine from coating pentlandite. As a result, collector molecules can re-adsorb onto exposed pentlandite active sites. The adsorption curves also reveal saturation behavior: once serpentine surface sites are occupied, further depressant addition offers little extra benefit, and excessive depressant can begin to compete with PBX for pentlandite sites, which explains the recovery decline observed at high depressant dosages.

3.5. XPS Analysis

X-ray photoelectron spectroscopy was used to probe surface chemical changes on serpentine induced by the four candidate depressants [37]; representative high-resolution spectra are shown in Figure 11, Figure 12 and Figure 13. The comparison of the Mg-related region before and after reagent treatment reveals a reproducible positive shift of the Mg core-level peak accompanied by a modest decrease in peak area for all reagents, implying partial coordination of surface Mg sites and a reduction in the number of free Mg surface species. The magnitude of the Mg shift is noticeably larger for PMIDA and GLY than for IDA and NTA, suggesting stronger interaction between phosphonate-bearing molecules and surface Mg.

Figure 11 presents high-resolution Mg 1s and N 1s spectra for serpentine before and after treatment with IDA and NTA, and the data provide clear evidence of reagent adsorption and modification of the serpentine surface. In the Mg 1s region, the pristine serpentine shows a single symmetric feature centered at 1303.44 eV [38]; following exposure to IDA or NTA, this Mg 1s maximum shifts to higher binding energy (to 1304.22 eV for IDA and 1304.19 eV for NTA) and the peak area is slightly attenuated. The positive shift and modest loss of Mg signal intensity are consistent with a change in the local chemical environment of surface Mg atoms caused by interaction with adsorbed organic ligands [39], for example coordination of carboxylate and/or amine functionalities to Mg²⁺ or formation of Mg-O-C/Mg-N type surface species rather than simple physisorption.

Complementary evidence appears in the N 1s window: untreated serpentine shows no discernible N 1s feature, whereas both IDA- and NTA-treated surfaces exhibit a new N 1s envelope centered near 399.8–399.9 eV [40]. The emergence of this nitrogen signal confirms the presence of the amino-bearing molecules on the mineral surface and implies that the nitrogen atom participates in the adsorption environment (either directly via coordination or indirectly through neighboring functional groups). No P 2p signal is detected in these spectra, as expected for reagents that lack phosphorus.

Taken together, the Mg 1s and N 1s changes indicate that both IDA and NTA chemisorb to serpentine, modifying surface coordination and electronic structure of Mg sites. These surface chemical alterations are fully consistent with the zeta-potential shifts and the reduced propensity of serpentine to heteroaggregate with pentlandite observed in flotation and DLVO analyses.

Figure 12 displays high-resolution Mg 1s, N 1s, and P 2p spectra for serpentine before and after treatment with PMIDA. The Mg 1s main feature of pristine serpentine is centered at 1303.44 eV; following PMIDA exposure, this peak shifts to 1304.20 eV and its integrated intensity is modestly reduced. Such a positive binding-energy shift, together with a small attenuation in Mg signal, is indicative of a changed chemical environment for surface Mg atoms, consistent with coordination or partial complexation rather than weak physisorption. Concomitantly, a new N 1s envelope appears at 399.90 eV where none was detectable on the untreated surface, confirming adsorption of the amino-containing PMIDA molecule and implying participation of nitrogen in the interfacial binding environment (either through direct coordination or via adjacent functional-group interactions). Most diagnostically, a clear P 2p doublet emerges at 131.76 eV after treatment [41]; this binding energy is characteristic of phosphonate/phosphate species bound to metal centers. The simultaneous appearance of P 2p and N 1s signals, together with the Mg 1s shift, strongly supports formation of surface Mg-O-P type linkages (phosphonate chelation) and a changed Mg coordination sphere on serpentine. These XPS observations therefore provide direct chemical evidence that PMIDA chemisorbs to serpentine, primarily through phosphonate–Mg interactions with auxiliary involvement of the amino functionality, which explains the pronounced negative zeta-potential shifts and the superior serpentine passivation and dispersion behavior observed in DLVO and flotation tests.

Figure 13 presents high-resolution Mg 1s, N 1s, and P 2p spectra for serpentine before and after treatment with GLY. The Mg 1s main line of the untreated sample is centered at 1303.44 eV; after GLY adsorption, the Mg 1s feature shifts positively to 1304.17 eV with a slight decrease in integrated intensity. Such a binding-energy upshift and modest attenuation of the Mg signal indicate a perturbation of the Mg coordination environment, consistent with formation of new surface bonds rather than mere physisorption. Concomitantly, a previously absent N 1s envelope appears at 399.75 eV, confirming the presence of the amino-containing GLY molecule on the serpentine surface and implying that the nitrogen atom contributes to the adsorption environment (either by direct coordination or as part of a hydrogen-bonded network). Most notably, a P 2p doublet emerges at 131.88 eV after treatment; this chemical shift falls within the range expected for phosphonate/phosphate species coordinated to metal centers. The combined observation of Mg 1s shifting, together with new N 1s and P 2p signals, provides strong evidence that GLY chemisorbs onto serpentine via Phosphonate-Mg interactions with auxiliary involvement of the amino functionality. These XPS results are fully consistent with the pronounced negative zeta-potential shifts and the enhanced dispersion behavior observed for GLY in the DLVO and flotation experiments, and they explain GLY’s superior ability to passivate serpentine compared with purely carboxylate reagents.

3.6. Contact Angle Measurements

Contact angle determination provides a direct metric of surface wettability and thus a convenient probe of the net hydrophobicity changes induced by reagent adsorption [42]. In this work, static water contact angles were measured on polished pentlandite wafers and on serpentine wafers before and after treatment with the candidate depressants; the results are summarized in Figure 14 and Figure 15.

As shown in Figure 14 and Figure 15, the untreated pentlandite surface exhibited a relatively high contact angle (75.53°), consistent with an intrinsically hydrophobic sulfide surface under the present experimental conditions. After conditioning with the collector and in the presence of the depressants, pentlandite contact angles decreased only modestly (to 70–72° for most treatments), indicating that pentlandite retained substantial hydrophobic character and that the tested depressants exerted limited direct hydrophilization of the sulfide surface. By contrast, untreated serpentine was markedly hydrophilic (contact angle: 36–40°). Following exposure to the four depressants, the serpentine contact angles fell substantially, with the largest reductions observed for the phosphonate-containing reagents: PMIDA and GLY reduced the serpentine contact angle to roughly 24–27°, whereas the carboxylate-type reagents (IDA and NTA) produced smaller decreases to about 29–30°.

These wettability changes mirror the adsorption and electrokinetic data and explain the observed flotation behavior. The pronounced hydrophilization of serpentine by PMIDA and GLY is attributed to strong Phosphonate-Mg interactions and formation of surface Mg-O-P linkages, which introduce polar, water-binding moieties and increase surface hydration. In contrast, pentlandite offers fewer sites for such chemisorption by small depressant molecules and thus preserves its collector-derived hydrophobicity. Overall, the selective increase in hydrophilicity of serpentine, most effectively achieved by PMIDA and GLY, enhances hydrophobic contrast between pentlandite and serpentine and underpins the improved pentlandite selectivity observed in flotation tests.

4. Conclusions

This study systematically evaluated four small-molecule depressants (IDA, NTA, PMIDA, and GLY) for selective suppression of serpentine in the flotation of low-grade nickel sulphide ore. Micro-flotation identified an operational optimum near pH 9 and a reagent dosage of 18 mg/L, at which PMIDA and GLY (phosphonate-bearing molecules) delivered the largest recovery gains and selectivity improvements compared with carboxylate-type reagents. Electrokinetic and DLVO analyses showed that PMIDA and GLY induce much larger negative shifts in serpentine zeta potential and generate a repulsive energy barrier that discourages heteroaggregation with pentlandite. Adsorption tests and XPS confirmed preferential chemisorption of PMIDA and GLY on serpentine via Phosphonate-Mg interactions (Mg 1s shifts and emergent P 2p signals), while PBX uptake on pentlandite was restored when serpentine was passivated. Contact angle data corroborated selective hydrophilization of serpentine without substantially altering pentlandite hydrophobicity. Collectively, these results indicate that introducing targeted phosphonate functionality into micromolecular depressants is an effective strategy to mitigate slime coating and entrainment, thereby improving concentrate grade and recovery. Under laboratory conditions, PMIDA and GLY performed comparably or better than benchmark phosphate depressants. Overall, PMIDA and GLY emerge as highly promising, high-performance serpentine depressants with strong potential to enhance selectivity and concentrate quality in the flotation of low-grade nickel sulphide ores.