Phase and Morphology Transformations in Sulfur-Fixing and Reduction Roasting of Antimony Sulfide

Abstract

Metallurgical extraction of antimony (Sb) currently has the limitations of high energy consumption and adverse environmental impact. In this study, we proposed a cleaning process to extract Sb by metallurgy and beneficiation based on S-fixing and reduction roasting of Sb₂S₃. Metallic Sb can be obtained directly by using zinc oxide (ZnO) and carbon as sulfur-fixing and reducing agents, respectively, at 600–1000 °C, wherein S is fixed in the form of ZnS. The thermodynamic feasibility of the process of roasting and the effects of a range of process parameters on Sb generation were investigated comprehensively. The optimum conditions for metallic Sb generation were determined to be as follows: temperature of 800 °C, C powder size of 100–150 mesh, ZnO content of 1.1 times its stoichiometric requirement (α), and reaction time of 2 h. Under the optimum conditions, the proportion of Sb distributed in the metal phase reached 90.44% and the S-fixing rate reached 94.86%. The phase transformation of Sb progressed as follows: Sb₂S₃→Sb₂O₃→Sb. The Sb particle had mainly spherical and hexahedral morphologies after quenching and furnace cooling, and bonded little with ZnS. This research is potentially beneficial for the further design process of Sb powder and ZnS recovery by mineral separation.

1. Introduction

Antimony is a critical basic material, widely used in the manufacture of flame retardants, alloys, brake pads, and catalysts [1]. In 2014, the European Commission designated Sb as a critical raw material [2]. The ores from which Sb can be economically extracted are typically sulfides such as stibnite (Sb₂S₃) and jamesonite (Pb₄FeSb₆S₁₄), and these are distributed across China, Bolivia, Burma, Russia, and Tajikistan [3]. The pyrometallurgical process, involving oxidative volatilisation (>1200 °C) and reduction smelting (>1000 °C) at high temperatures, carried out in blast furnaces and reverberatory furnaces, respectively, has been the primary method of Sb extraction and accounts for more than 90% of the total Sb output [4]. The significant shortcomings of this method are high energy consumption (>3.6 × 10¹⁰ J/t Sb), large generation of low-concentration SO₂ [5], and risk of occupational disease due to volatilisation of As and Sb [6,7].

To improve the operational environment, researchers have developed several processes of intensified smelting, such as O₂-enriched smelting [8,9], molten salt smelting [10,11], and sulfur-fixing smelting [12]. In O₂-enriched smelting, O₂-enriched air is used to oxidise Sb₂S₃ concentrate to obtain high-concentration SO₂. This practice indicates that Sb₂O₃ coheres on the gas duct and exhibits a low melting point and bondability. We developed a process of reduction-matting-smelting to directly produce Sb at between 1000 °C and 1300 °C, and found that the direct recovery rate of Sb was low for its volatilisation to proceed. Padilla et al. investigated S-fixing roasting of Sb₂S₃ using lime as a desulfurisation agent in the temperature range 700–850 °C [13,14]. This method showed potential for application in Sb extraction at low temperature with SO₂-free emission. However, it is difficult to dispose of CaS, the product of the S-fixing reaction, which is yet another hazardous solid waste.

Industrial hydrometallurgical practices can be divided into acidic and alkaline processes based on the leaching agents employed. In the acidic method, FeCl₃/SbCl₅-HCl mixture is used as the leaching solution and membrane electrowinning is subsequently conducted to extract Sb and regenerate the leaching reagent [15,16]. The main issues of this method are equipment corrosion by Cl, and hyperplasia of Fe²⁺ ions. In the alkaline method, the Na₂S-NaOH system is used as the leaching solution [17,18]. The obvious shortcoming of this method is the low current efficiency of Sb electrode position due to side reactions, as well as the requirement for wastewater disposal [19].

An economical and environment-friendly method is imperative for Sb extraction. In our previous work, a process of molten salt smelting [10] was adapted to smelt Sb₂S₃ directly in the molten NaCl-Na₂CO₃ system; the significant advantages of this process were low temperature and S-fixing, but high consumption of molten salt and difficulties in molten salt separation were its shortcomings. However, it was found that the direct reaction between Sb₂S₃, ZnO, and C was completed in the absence of a molten medium. Hence, it was inferred that the process entailed roasting of Sb₂S₃ with sulfur-fixing and reductant to yield Sb metal and other sulfides at low temperature, following which the solid roasted products could be separated by beneficiation. Meanwhile, Pb and Zn smelters entail the production of large quantities of secondary ZnO ash (ZnO > 70%) [20,21], which are low in value and cannot return to the hydrometallurgical process of Zn because of corrosion compositions.

Therefore, our team proposed a method of combined concentration and smelting process to extract Sb from stibnite. Firstly, stibnite was engaged in S-fixing and reduction roasting using ZnO and C as the S-fixing agent and reductant, and the products are Sb and ZnS. The roasted products were then separated by gravity and flotation to obtain Sb powder and ZnS concentrate. The prominent advantages of this new method are its low operating temperature, and the clean production and synergetic disposal of ZnO ash. This work focused on the phase and morphology transformations of Sb in the processes of S-fixing and reduction roasting. The objective of this work was to obtain a high Sb generation rate and favourable particle characteristics of Sb for the subsequent process of beneficiation, and also to provide a potential method for the extractive metallurgy of sulfide ore.

2. Experimental

2.1. Materials and Instruments

The main reagents used in the experiment were ZnO, Sb₂S₃, and C powder; C powder and ZnO were procured from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), and Sb₂S₃ from Guangdong WengJiang Chemical Reagent Co., Ltd. All reagents used in the experiment were analytically pure. The particle size of Sb₂S₃ was 250–300 mesh. Roasting was carried out in a tubular furnace (SGMT60/12A, Sigma Furnace Industry Co., Ltd., Luoyang, China). The components of the equipment used for roasting are shown in Figure 1.

2.2. Methods

The total amount of Sb₂S₃, ZnO, and C powder used was 15 g, in the molar ratio 1:3:3. The powder was taken in a beaker, and alcohol (75 mL) was added. The beaker was then placed in an ultrasonic machine and the mixture was mechanically stirred for 30 min. The beaker was then placed in a drying oven at 60 °C. The powder was then placed in a mould and crushed into pieces after volatilisation of the alcohol. The raw material, pressed into blocks, was loaded into a crucible and roasted in the tube furnace at a certain temperature in N₂ atmosphere. The N₂ flow-rate during the trials was 2 L/min. The gas generated was successively pumped into KMnO₄ solution, magenta solution, and Na₂CO₃ solution to detect SO₂ generation according to Reaction (1). The pH values of the KMnO₄ solution were recorded every 5 min using a pH meter. The sample was withdrawn from the furnace after the setting time. The phases and morphologies of the products were measured using a Rigaku D/max 2550 VB + 18 kW X-ray powder diffractometer and a scanning electron microscope (SEM, JSM-6490LV, JEOL, Akishima, Tokyo, Japan), respectively. The phase content of Sb in the roasted products was also analysed by chemical titration (oxidation method by cerous sulfate [22])

2KMnO₄ + 5SO₂ + 2H₂O = 2MnSO₄ + K₂SO₄ + 2H₂SO₄

(1)

The roasted products were dressed directly through flotation and gravity separation to obtain ZnS concentrate and Sb powder. The Sb powder can be purified by the current pyrometallurgical refining process with the addition of Na₂CO₃ and O₂ blowing. The process flow chart is shown in Figure 2. In this paper, we focused on the phase and morphology transformations in the process of S-fixing and reduction roasting of Sb₂S₃, and the separation of the roasted products was not studied in detail.

2.3. Thermodynamic Calculations

The reactions possible during the experiment are shown in

Table 1. Reactions possible during the experiment.

Reductions	No.
Sb2S3 + 3ZnO + 1.5C = 2Sb + 3ZnS + 1.5CO2(g)	(2)
Sb2S3 + 3ZnO + 3C = 2Sb + 3ZnS + 3CO(g)	(3)
Sb2S3 + 3ZnO = Sb2O3 + 3ZnS	(4)
Sb2O3 + 1.5C = 2Sb + 1.5CO2(g)	(5)
Sb2O3 + 3C = 2Sb + 3CO(g)	(6)
Sb2S3 + 2Sb2O3 = 6Sb + 3SO2(g)	(7)
ZnO + CO(g) = Zn + CO2(g)	(8)
ZnO + C = Zn + CO(g)	(9)
C + CO2(g) = 2CO(g)	(10)

. To investigate the thermodynamic feasibility of these reactions, the HSC software was used to calculate the relationship between the Gibbs free energy (ΔG^θ) and temperature for each reaction, and the results are shown in Figure 3. The diagram indicates that the ΔG^θ values of reactions (7) and (8) are always greater than zero, while the ΔG^θ values of all the other reactions are less than zero in the temperature range 500–1000 °C. For the direct reducing reaction of ZnO shown in (9), the ΔG^θ value is also positive below 1000 °C, so the reducing reaction of ZnO cannot be carried out before 1000 °C. This result is also in agreement with standard industrial practice. With increase in temperature, the ΔG^θ values of Reactions (2), (3), (5) and (6) become more negative. This suggests that an increase in temperature is favourable for the direct generation of Sb and reduction of Sb₂O₃. However, the increase in temperature is less effective on Reaction (7). This indicates that, in theory, an increase in temperature can promote the reaction and facilitate the formation of Sb.

3. Results and Discussion

3.1. Effect of Temperature on Roasting Results

Sb₂S₃, ZnO and C powder were mixed in a molar ratio of 1:3:3. The particle size of the C powder was 150–200 mesh, and the roasting time was 2 h. The effects of roasting temperature on the Sb phase distribution and SO₂ emission were evaluated in the temperature range 600–1000 °C. The results of Sb phase distribution are shown in Figure 4a, and the changes in the pH of KMnO₄ solution with time are shown in Figure 4b. The pH value of the KMnO₄ solution reflects the generation of SO₂ during roasting.

It can be seen from Figure 4a that the main phase of Sb in the roasted products was Sb₂O₃ at temperatures lower than 800 °C and little metallic Sb was obtained. As temperature increased to 800 °C, the generation rate of metallic Sb increased sharply, and its proportion rose to 89.4%. As temperature increased further i.e., beyond 800 °C, the proportion of Sb changed negligibly. During the entire process of heating, the proportion of Sb₂S₃ decreased, indicating that the conversion reactions of Sb₂S₃ progressed adequately. The XRD patterns of the roasting products of Sb₂S₃-ZnO-C mixture at different temperature were investigated in our previous report [23], and the results proved that Sb₂O₃ and Sb were yielded after 600 °C and 800 °C respectively, which was in very good agreement with the phase distribution results. Figure 4b shows that the pH value of the KMnO₄ solution at all temperatures decreased sharply in the first 5 min and then gradually stabilised. This phenomenon was likely due to the oxidisation of Sb₂S₃ to yield SO₂; the tube was exposed to air when the furnace door was opened for the reactant mixture to be loaded, and hence the pH of the solution dropped dramatically. When O₂ was depleted, the production of SO₂ ceased and the pH of the solution stabilised, thereby resulting in very little SO₂ generation in the subsequent process of roasting. However, the pH value of the KMnO₄ solution increased with temperature, and also achieved equilibration quickly at 1000 °C. As calculated in Figure 3, an increase in temperature can promote the sulfur-fixing and reducing reaction and facilitate the formation of Sb, and sulfur was fixed by ZnO in the form of ZnS rather than SO₂ and S^O [14]. So the pH of KMnO₄ solution did not decrease.

The SEM and energy-dispersive X-ray spectroscopy (EDX) analysis of the products of roasting at six temperature values are presented in Figure 5 and

Table 2. EDS results of the elemental contents of the roasted products at the selected points (atom %).

Points	Sb	Zn	S	O
1	96.00	-	-	4.00
2	20.22	49.16	28.03	2.60
3	90.13	4.08	1.64	4.15
4	41.18	36.67	19.55	2.6
5	100	-	-	-
6	1.06	69.44	30.50	-

, respectively. As shown, large amounts of Sb₂O₃ with perfect rhomboidal dipyramid morphology were obtained at 500 °C and 600 °C. It is established in the literature [1] that Sb₂O₃ has two crystal forms, i.e., cubic and rhomboidal, and the former transforms into the latter at temperatures beyond 550 °C. Sb₂O₃ melts as temperature increases beyond 656 °C. Hence, coarse and regular Sb₂O₃ particles were observed in the temperature range 500–600 °C, whereas the roasted products presented fine particles at 700 °C, entirely devoid of Sb₂O₃ and almost mixed with ZnS. At 800 °C, metallic Sb with spherical particles were obtained as Sb has a low melting point of 630 °C and the particles grew continuously with increase in temperature. Meanwhile, pure ZnS grains with low Sb content were obtained beyond 800 °C, which indicates that the S-fixing and reduction reactions had been carried out thoroughly. Based on these findings, along with the results displayed in Figure 4 and Figure 5, the formation of Sb can be divided into two steps. The first step is the cross-reaction between Sb₂S₃ and ZnO to obtain Sb₂O₃ and ZnS, and the second step is the reduction of Sb₂O₃ by C to Sb. Comprehensive consideration suggests that the best roasting temperature is 800 °C.

3.2. Effect of C Particle Size on Roasting Results

Sb₂S₃, ZnO and C powder were taken in the molar ratio 1:3:3. The roasting temperature and time were 800 °C and 2 h, respectively. The effect of the size of C powder on roasting was investigated. The phase distribution ratio of the roasted products is shown in Figure 6a, and the changes in the pH values of the KMnO₄ solution with time are shown in Figure 6b.

Figure 6a indicates that the C powder size had little effect on the phase distribution of Sb. Sb was mainly in the form of metallic Sb in the roasted products with a proportion of about 89%, followed by Sb₂O₃, which accounted for about 8% of the total Sb content. As the powder was mixed evenly and compacted into blocks, there was relatively sufficient contact among the particles, and hence the C powder size had little effect on the phase distribution of Sb. It can be seen from Figure 6b that the smaller the particle size of C powder, the higher the pH of the KMnO₄ solution is. This indicates that the effect of S-fixing improves with the fineness of C powder. This is because greater fineness of the particle enhances contact between reactants, thereby accelerating the reaction rate of S-fixing, reducing the emission of SO₂ and improving the effect of sulfur-fixing. However, the finer the granularity, the higher the cost. Therefore, the particle size of 100–150 mesh is the most suitable.

3.3. Effect of ZnO Content on Roasting Results

The roasting temperature and time were 800 °C (1073 K) and 2 h, respectively. The C powder particle size was 100–150 mesh. The phase distribution ratio of Sb in the roasted product and the pH value of the KMnO₄ solution at different content of ZnO (0.8–1.2 times the stoichiometric requirement), according to equilibrium Reaction (2), are shown in Figure 7.

Figure 7a indicates that the proportion of Sb gradually increased from 72.0% to 91.3%, while that of Sb₂S₃ decreased from 18.2% to 0.31% as the content of ZnO increased from 0.8 to 1.2 times the stoichiometric requirement. When the content of ZnO reached 1.1 times the stoichiometric requirement (α_ZnO), the proportion of Sb gradually stabilised, reaching 91.3%. The proportion of Sb₂O₃ changed negligibly and accounted for about 8%. As the reaction system can be described in two steps, i.e., ZnO reacts with Sb₂S₃ to generate Sb₂O₃ and ZnS, following which Sb₂O₃ reacts with C powder to generate Sb, a change in the amount of ZnO can affect the amounts of Sb₂S₃ and Sb, but not that of Sb₂O₃.

From the pH values measured, it can be seen that increasing the amount of ZnO increased the final pH of the solution after stabilisation. This shows that increasing the amount of ZnO can improve the effect of S-fixing but no significant enhancement was seen at ZnO contents greater than 1.1 times the stoichiometric requirement. As ZnO is a sulfur-fixing agent, the amount of ZnO directly affects the S-fixing effect. Therefore, the ZnO content of 1.1 times the stoichiometric requirement was selected.

3.4. Effect of Time on Roasting Results

The roasting temperature was 800 °C. The C powder particle size was 100–150 mesh and the content of ZnO was 1.1 times the stoichiometric requirement. The effects of roasting time on the phase distribution of Sb in the roasted products and the pH value of the KMnO₄ solution are shown in Figure 8.

Figure 8a indicates that the proportion of Sb increased gradually and that of Sb₂O₃ decreased with time. At 2 h, the reduction was nearly complete, and the proportion of Sb was 91.3%. However, as Sb is gradually reduced by volatilisation, the proportion of Sb started to decrease, and that of Sb₂O₃ started to increase after 2 h. The roasting time had little effect on the proportion of Sb₂S₃. The pH value of the KMnO₄ solution was mostly stable, and the roasting time had little influence on the effect of S-fixing. Therefore, the optimum roasting time was determined to be 2 h and the proportion of Sb obtained was 91.3%.

3.5. Effect of Cooling Mode on Roasting Results

To investigate the effect of the cooling mode on the crystal forms and structures of the roasted products, programmed cooling was carried out at a rate of 3 °C/min, and natural cooling was performed in the tube furnace. The SEM analysis was used to examine the product morphology. The changes in temperature during the two cooling modes, and the SEM images of the products are presented in Figure 9 and Figure 10, respectively. As shown, in the case of the programmed cooling mode, the temperature declined linearly from 800 °C to 400 °C in 133 min. However, in the case of natural cooling, the decrease in temperature was initially rapid but progressed slowly thereafter at a rate lower than that of programmed cooling. As shown in Figure 10, large amounts of Sb particles with perfect dipyramid morphology were obtained in both the cooling modes. The Sb particles obtained in the natural cooling mode presented aggregation and were larger; recrystallisation was observed on the surface of the Sb particles, which were determined to be composed of large microcrystals. When compared with natural and splat cooling, programmed cooling yielded smooth and single-crystal Sb particles with rhombohedral crystals and homogenous distribution in the roasted mixture. However, in both the cooling modes, the Sb particles obtained were independent and had little bonding with ZnS, thereby facilitating the separation of Sb and ZnS in the subsequent process of beneficiation.

3.6. Energy Consumption Comparision

Through a low-temperature sulfur-fixing and reduction roasting and a gravity-flotation combined beneficiation process, metallic Sb was obtained. This represents a 200–300 °C decrease in temperature compared with the traditional process, which may reduce energy and refractory materials consumption dramatically. The economic feasibility based on energy consumption of the new process was calculated and compared with the current process in

Table 3. The standard energy consumption to produce 1t of antimony by the current pyrometallurgy process and this metallurgy and beneficiation combined process via the present study.

The Conventional Pyrometallurgy Process		Metallurgy and Beneficiation Combined Process
Unit operation	Energy cost GJ·t−1	Unit operation	Energy cost GJ·t−1
Oxidising-roasting	3.05 × 1010	Roasting	3.28 × 1010
Reduction-smelting	1.45 × 1010	Gravity separation	4.69 × 108
		Flotation separation	5.50 × 108
Total costing =	4.5 × 1010	Total costing =	3.38 × 1010
CO2 emission	4.62 t	CO2 emission	3.29 t
SO2 generation	0.81 t	SO2 generation	0.04 t

Calculation standard: where calorific value coke = 3 × 107 J/kg, bituminous coal = 2.7 × 107 J/kg, anthracite coal = carbon = 2.9 × 107 J/kg, S content in all fuel was 1 pct; CO₂ emitted during the consumption of fuel. Energy consumption of the gravity separation and flotation separation were 36.1 kWh/t and 60.0 kWh/t.

. The conventional pyrometallurgy process primary development is the blast furnace. According to the production practice [6,7], the fuel consumption of every ton of Sb in the oxidising-roasting and reduction-smelting process is as follows: coke: 906 kg, bituminous coal: 123 kg and anthracite coal: 507 kg; this resulted in 4.62 t CO₂ greenhouse gas emission. Meanwhile, the sulfur in stibnite in the form of sulfide was oxidised to SO₂, which is low in concentration and consumes large alkaline solutions to absorb. However, the fuel consumption of the new process was estimated as 1131 kg anthracite coal for every ton Sb; 37% of it is used for reduction and other 63% for combustion. The energy cost calculation of new method was based on the practice of gold ore shaking table separation and zinc concentrate flotation [24]. The energy cost of total beneficiation is low, and the total cost is 3.38 × 1010 GJ/t, which is found to be 24.9 pct lower than 4.5 × 1010 GJ/t of energy required in the conventional process. Meanwhile, the CO₂ emission decreased to 3.29 t; in particular, the SO₂ generation decreased to 0.04 t. This represents a large environmental benefit compared with the conventional process.

4. Conclusions

The thermodynamic mechanisms and experimental conditions of S-fixing and reduction roasting in the process of Sb smelting were studied. The thermodynamic analysis shows that the ΔG^θ value of the overall reaction of S-fixing and reduction is less than zero and that Sb₂O₃ can be reduced to Sb under the given conditions in the system. It can be inferred from the experiments that the course of the reaction can be divided into two main steps, i.e., S-fixing and reduction. The optimal reaction conditions are as follows: roasting temperature of 800 °C, carbon powder size of 100–150 mesh, content of ZnO 1.1 times the stoichiometric requirement, and roasting time of 2 h. Under these optimum conditions, the proportion of Sb, as determined from the phase distribution of Sb in the product, is 91.3%. The natural and programmed cooling modes both yielded rhombohedral Sb particles that had little bonding with ZnS, while the programmed cooling mode also yielded smooth and single-crystal Sb particles, and these were convenient for the separation of Sb and ZnS in the subsequent process of beneficiation. The proposed process is greener and has an increased economic attraction to the extraction Sb from stibnite.