Mechanism on the Separation of Vanadium and Titanium from Vanadium Slag by Roasting with Ammonium Sulfate

Abstract

The technology of simultaneously recovering V and Ti from vanadium slag via ammonium salt roasting has proven to be an efficient route. However, due to the phase stability and complex chemical composition of vanadium slag, intermediate materials containing Fe, V, Ti and Mn are difficult to be characterized critically. This work aims to investigate the decomposition and transformation of vanadium slag during ammonium salt roasting, using a combination of FT-IR, XRD, XPS and SEM techniques. It was found that the lattice structure of Fe-contained spinel would be transformed from FeV₂O₄ to Fe²⁺V_nFe_2−nO₄ (0 < n < 2) during directly roasting in the air. However, there is no obvious change for Ti-contained and Mn-contained spinel. Using NH₄HSO₄ (ABS) as an additive and roasting the slag in the N₂ atmosphere, those spinels would be decomposed into various sulfate salts. Meanwhile, when the slag was roasted with NH₄HSO₄ in the air, a part of Fe(II) in (NH₄)₂Fe(SO₄)₂ would be transferred into Fe(III), but V(III), Ti(IV) and Mn(II) from those salts would remain the same valance state. Ultimately, about 88% V and 81% Ti were recovered, when vanadium slag was roasted at 663.15 K with a 1:5 ratio of slag-to-NH₄HSO₄ and followed by 8 vol.% H₂SO₄ leaching.

1. Introduction

Vanadium slag is a strategic vanadium extracting source, which is produced from the hot-metal treatment prior to the basic oxygen steelmaking (BOS) or during the converter steelmaking process by using vanadium–titanium magnetite (VTM) ore [1,2]. This VTM ore has been extensively found in different countries, e.g., South Africa, Russia and China, etc., and serves as a vanadium production raw material [3]. Especially in China, the vanadium slag, a by-product of VTM ore from the iron and steelmaking industry contributes to an over 85% V₂O₅ production [4,5]. Referring to the physical and chemical property of the vanadium slag, the primary stable phases are the spinel containing Fe, V, Ti, Cr, and Mn, while the veinstone phase is quartz [6,7]. Since the V-O bond in the spinel crystal structure is quite stable, the only way to extract V from the slag is to break the bond as well as decompose the spinel structure, which is quite difficult. The conventional process for V extraction from the vanadium slag is sodium or calcium roasting followed by an acid leaching technology. For this method, the slag was roasted at 1123.15 K using Na₂CO₃, NaCl, CaO or CaCO₃ as an additive agent [8,9,10]. The primary mechanism of V-contained phase transition during the roasting of the two methods can be expressed as Equations (1) and (2) [11,12].

FeV₂O/4O₂(g) + 2NaCl = 2NaVO₃ + 1/2Fe₂O₃ + Cl₂(g)

(1)

FeV₂O₄ + CaCO₃ + 5/4O₂(g) = 1/2Fe₂O₃ + CaV₂O₆ + CO₂(g)

(2)

Although thermodynamic calculation indicates that the Gibbs free energy change of reactions (1) and (2) are −329.07 and −1063.39 KJ/mol at 298.15 K, i.e., these two reactions are sufficient to proceed at room temperature in the thermodynamic perspective. The actual case in the industry is to roast the vanadium slag at 1123.15 K to achieve suitable reaction kinetics. This fact leads to inefficient energy consumption [13,14]. Besides, those sodium and calcification roasting methods discharge amounts of Na⁺ containing waste or Ca²⁺ containing gypsum, which is quite detrimental to the environment. In addition, these methods abandon titanium resources in the leaching tailings. Recently, several researchers have paid attention to investigating the decomposition behavior of spinel and tend to develop a new efficient and environmental-friendly vanadium extraction method. For instance, Zhang et al. researched the thermodynamic calculation and phase transition of vanadium slag in the pressure acid leaching and a 98% vanadium extraction ratio at 413.15 K can be obtained [15]. Du et al. investigated the vanadium decomposition mechanism in the concentrated NaOH solution. By using this method, a 95% of vanadium recovery ratio at 393.15 K has been reported [16]. Lv et al. researched the transformation behavior of FeV₂O₄ by the mechanical activation method and they obtained a sufficient vanadium recovery ratio of over 90% [17]. Liu et al. applied an electrolytic device with a 0.4 A and 2.8~3.0 V condition to remove the vanadium from the slag and obtain over 70% recovery ratio at 348.15 K [18]. Among these works, the method reported by [16,17,18] was complicated to apply in the actual industry. Moreover, the application of the pressure acid leaching method on an industrial scale requires to build up a large facility, e.g., autoclaves [19]. These technologies are not easy to upgrade to the existing vanadium industry since a huge amount of investment to build a new plant is needed. Thus, if there were an efficient technical route that could use the roasting equipment from conventional industrial processes, it would have better applicability and promotion for the clean production of vanadium.

In a previous work by the authors, we proposed a route to simultaneously recover V and Ti from the vanadium slag by the efficient (NH₄)₂SO₄ (AS) roasting [5]. A relatively high vanadium extraction ratio of over 90% can be obtained, and a mild leaching condition using 6% H₂SO₄ at 333.15 K has been selected. This method provides a potential chance to optimize the vanadium extraction utilizing the traditional roasting-leaching equipment referring to the vanadium extraction industrial application. However, due to the complex microstructure and phase composition of vanadium slag, the intermediate materials containing Fe, V, Ti and Mn are difficult to be characterized, and several scientific issues have not been clearly revealed yet. According to this point, the current work aims to investigate the chemical decomposition and phase transformation of vanadium slag during ammonium salt roasting. Specifically, the thermodynamic calculations of chemical reactions during the NH₄HSO₄ (ABS) roasting process were first performed. Subsequently, the designed roasting experiment was carried out to characterize the phase transformation of Fe, V, Ti and Mn compounds, using Fourier-transform infrared spectroscopy (FT-IR), X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and scanning electron microscope (SEM). Finally, a sustainable separation route of V, Ti and Fe after ammonium salt roasting and acid leaching was suggested.

2. Thermodynamic Analysis

The thermodynamic calculation uses Fe, FeV₂O₄, Fe₂TiO₄, MnO, Fe₂SiO₄, CaFe(SiO₃)₂ and CaMgSiO₆ to represent the mainly mineral composition in the vanadium slag. This selection is made based on the consideration of both the actual phase in the vanadium slag as well as the capability of the thermodynamic database in the HSC Chemistry software [13,20]. Possible chemical reactions of the roasting process in the presence of ammonium sulphate are shown in

Table 1. Reactions occurred in (NH₄)₂SO₄ roasting process.

NO.	Reaction Equation	NO.	Reaction Equation
1	(NH4)2SO4 = NH4HSO4 + NH3(g)	11	NH4HSO4 + FeO = FeSO4 + NH3(g) + H2O(g)
2	2Fe + O2(g) = 2FeO	12	NH4HSO4 + 1/3Fe2O3 = 1/3Fe2(SO4)3 + NH3(g) + H2O(g)
3	3/2Fe + O2(g) = 1/2Fe3O4	13	NH4HSO4 + 1/3V2O3 = 1/3V2(SO4)3 + NH3(g) + H2O(g)
4	3Fe2SiO4 + O2(g) = 2Fe3O4 + 3SiO2	14	NH4HSO4 + 1/2V2O4 = VOSO4 + NH3(g) + H2O(g)
5	2Fe2SiO4 + O2(g) = 2Fe2O3 + 3SiO2	15	NH4HSO4 + MnO = MnSO4 + NH3(g) + H2O(g)
6	4CaFe(SiO3)2 + O2(g) = 2Fe2O3 + 4CaSiO3 + 4SiO2	16	NH4HSO4 + 1/3Fe2TiO4 = 1/3TiOSO4 + 2/3FeSO4 + NH3(g) + H2O(g)
7	CaMgSiO6 = MgSiO3 + CaSiO3	17	NH4HSO4 + 1/4FeV2O4 = 1/4V2(SO4)3 + 1/4FeSO4 + NH3(g) + H2O(g)
8	4FeV2O4 + O2(g) = 2Fe2O3 + 4V2O3	18	NH4HSO4 + 1/2Fe2SiO4 = FeSO4 + NH3(g) + 1/2SiO2(g) + H2O(g)
9	2V2O3 + O2(g) = 2V2O4	19	NH4HSO4 + CaFe(SiO3)2 = FeSO4 + NH3(g) +CaSiO3 + H2O(g) + SiO2
10	2V2O4 + O2(g) = 2V2O5	20	NH4HSO4 + 2FeSO4 + 1/2O2(g) = Fe2(SO4)3 + NH3(g) + H2O(g)

. Using HSC Chemistry software (Metso Outotec, 6.00, Helsinki, Finland), the $\Delta G_T^o$, $\Delta H_T^o$ of 1~20 reactions are analyzed and presented in Figure 1 and Figure 2. The No. 1 reaction shows that (NH₄)₂SO₄ (AS) will be decomposed into ABS at above 533.15 K. Therefore, to simplify this research, the critical discussion in this work was based on ABS as the additive.

According to the reactions listed in Table 1, it can be divided into three parts, such as reaction 1 for the additive agent decomposition, reactions 2~10 for oxidation of different components in the vanadium slag, and the reactions 11~20 decomposition reactions in the presence of ABS. To simplify the thermodynamic calculation process, it is assumed that the gas partial pressure is 1 standard atmosphere. Figure 1 shows that the enthalpy change, $\Delta$_rH, of reactions 1, 7 and 11~16 was higher than 0 KJ/mol in the range of 273.15~773.15 K. This indicates that these reactions were all endothermic. When the temperature was in the range of 623.15 to 673.15 K, the value of $\Delta$_rH in reactions 2~6, 8~10 and 13 were below 0 KJ/mol. This fact indicates that these reactions were all exothermic type.

As shown in Figure 2, the thermodynamic calculation of the reaction (1) indicates that the $\Delta$_rG was positive under 538.15K, but it turns out to be negative at the elevated temperature. The $\Delta$_rG of other chemical reactions, except for the reactions (7) and (18) (decomposition of CaMgSiO₆ and Fe₂SiO₄), are all below 0 KJ/mol over 573.15 K. The results of the $\Delta {\mathrm{G}}_{\mathrm{T}}^{}$ analysis indicate that the main phase of vanadium slag may be oxidized in the air or decomposed with the addition of ABS during roasting when the temperature was above 573.15 K.

Zhang et al. have pointed out that the decomposition of the vanadium slag in the sodium or calcium roasting process begins with the oxidation of FeO in the spinel at 1073.15 to 1123.15 K, which is the key reaction for the vanadium extraction [13]. This result was consistent with the thermodynamic calculations of reactions 2 to 6 and 8 to 10. However, according to the thermodynamic analysis of reactions 11 to 20, the phase of FeV₂O₄, Fe₂TiO₄, Fe₂SiO₄, and CaFe(SiO₃)₂ can also be decomposed directly by ABS at high temperature. Our previous study also concluded that the optimized temperature for vanadium extraction via ammonium sulfate roasting was 643.15 K. Using this condition, about 91% of the vanadium in the slag can be recovered [5]. However, the thermodynamic calculation shows both oxidation roasting and ABS roasting can contribute to the decomposition of the iron spinel. A detailed roasting experiment should be systematically investigated.

3. Experiment

3.1. Raw Materials

The vanadium slag used in this work was obtained from the hot-metal treatment procedure of an iron and steelmaking company. The slag was ground by a planetary ball mill and screened to be a granularity of 0.075 mm in diameter. All the slag was dried over 383.15 K for 9 h before roasting. The ABS with an analytical grade was provided by Aladdin Chemical Reagent Co., Ltd. (Fengxian District, Shanghai, China). The distilled water used in this work was self-produced in the laboratory. The chemical composition of the slag used here is presented in

Table 2. Chemical composition of the vanadium slag used in this work.

Composition	Fe2O3	SiO2	V2O5	TiO2	MnO	CaO	MgO	Cr2O3	Al2O3
wt.%	38.50	27.75	18.88	12.75	10.40	2.27	2.33	1.90	2.19

.

3.2. Characterization Methods

An FTIR spectrometer (IRPrestige 21, Shimadzu, Kyoto, Japan) was used to measure vibrational spectra in the range of 450 to 4000 cm⁻¹ with a 0.5 cm⁻¹ resolution. The XRD analysis (DX2007, Shandong FY factory, Shandong province, China) was carried out with the condition of 1°/min from 10 to 75°, Cu Kα radiation (λ = 0.154056 nm), to determine the existence of different phases in the samples. The ICP-OES (iCAP 6000, T-Fisher Scientific, Austin, TX, USA) was used to analyze the chemical composition of the leaching solution. The SEM (JSM7500F, JEOL, Kagawa, Japan) was used to observe the morphology of the slag before and after roasting. The XPS (XSAM800, Stretford, UK) was used to analyze the valence state of the different surface elements (V, Ti, Fe, Mn) of the slag.

3.3. Experimental Procedure

Three series of roasting experiments were designed to study the phase transformation of the spinel compounds. Detailed experimental plans are shown in

Table 3. Detailed roast experimental plan in this work.

Exp. No.	Amount of Slag	Flow Rate of Atmosphere
1	3 g vanadium slag	250 mL/min air
2	1 g vanadium slag + 3 g ABS	250 mL/min N2 gas
3	1 g vanadium slag + 3 g ABS	250 mL/min air

.

In each roasting test, the raw vanadium slag was placed in the ceramic crucible and roasted in the tube furnace (GSL-1500X, Anhui KJ Material Sci. & Tech. Co., Hefei, China). The initial temperature of the device was 293.15 K, and the endpoint temperature was 653.15 K. The heating rate was 10 K/min. When the temperature heated up to 653.15 K, the sample was placed in the hot zone of the furnace. The furnace chamber was cleaned by using a vacuum pump followed by purging a nitrogen gas with a purity of 99.9%. The roasting was operated independently at 10 and 60 min. After roasting, the slag samples were collected to characterize the phase evolution. Subsequently, a leaching procedure was conducted using H₂SO₄ at 65 °C in the flask assembled in the water bath. The formulation to calculate V and Ti is shown as:

$$E=\frac{C_i\times V}{m\times w_i}\times 100\%$$

where V and C_i represent the volume of the leaching solution and concentration of the element in the solution, m represents the mass of the slag and w_i represents the percentage of elements in the slag.

4. Results and Discussions

4.1. Transformation of Functional Groups in the Spinel

To detect the functional groups of the roasted slag, the FT-IR analysis were conducted, and the results are shown in Figure 3.

In Figure 3a–d, the positions of 3451, 3438, 3318, 1639 and 1637 cm⁻¹ in Figure 3 were all caused by the hydroxyl (-OH) from the water molecule. The absorption bands that detected at 1070, 1065, 1063, 1053, 1051, 1012 and 1011 cm⁻¹ were all attributed to the stretching vibration of Si-O-Si in the samples. While, the stretching vibration of [SiO₄] were shown in 960, 951, 949 and 916 cm⁻¹. This result indicates that there are at least two types of silicon-containing phases in the samples. The positions of 888, 887, 874, 875 and 852 cm⁻¹ belong to V-O absorption band, and, the absorption band of Fe-O occurs at 614, 593, 583, 580, 579, 578, 478, 477, 468 and 456 cm⁻¹ [21].

Compared with Figure 3a, some of the bands were shifted obviously due to the structural change in Figure 3b–d. However, it is hard to distinguish those phases using FT-IR individually. Furthermore, the absorption band of O=S=O occurred at 1469, 1420 and 1402 cm⁻¹ in Figure 3c,d. This is because the band belongs to the dual characteristic covalent peak of S=O in sulfate species. Either TiOSO₄ or VOSO₄ may exist in the samples, and this fact needs to be examined further. The HSO₄⁻ functional group of NH₄HSO₄ shows its characteristic bonds were at the position of 1284, 1268, 1228 and 1182 cm⁻¹. However, the characteristic bond of NH₄⁺ group from ammonium species cannot be detected due to its overlap with O=S=O bond at the position of 1402 cm⁻¹ [22].

4.2. Structures Evolution of Different Phases in the Vanadium Slag

Figure 4a–c shows the XRD pattern of the phase structure evolution from the raw slag to the roasted slags. In Figure 4a, it was found that FeV₂O₄, Fe₂TiO₄, MnV₂O₄, Fe₂SiO₄, and Ca(Fe, Mg)(SiO₃)₂ were the stable phase in the raw slag. After 10 min roasting in the air, the mineral phase in the slag is presented to be Fe²⁺V_nFe_2−nO₄ (0 < n < 2), Fe₂TiO₄, Mn₂VO₄, Ca(Fe, Mg)(SiO₃)₂ and Fe₂SiO₄. When the roasting time was extended to be 60 min, the characteristic peaks of Fe²⁺V_nFe_2−nO₄ (0 < n < 2) at diffraction angles of 18.232^o, 29.989°, 35.322°, 53.081° and 62.323° were still existing. However, two new characteristic peaks at 42.252° and 42.941° were observed. A further analysis indicates that these two phases were CaSiO₃ and Fe₂O₃. Therefore, it can be inferred that the olivine phase in vanadium slag was also decomposed at this time.

In Figure 4b, after 10 min roasting with ABS in N₂ gas, a part of FeV₂O₄, Fe₂TiO₄, Fe₂SiO₄, and CaFe(SiO₃)₂ in the slag was decomposed, but there was no oxidization process. After 60 min roasting, the production was (NH₄)₂Fe(SO₄)₂, NH₄V(SO₄)₂·2H₂O, NH₄V(SO₄)₂, MnSO₄, SiO₂ and TiOSO₄ phases. Compared with the raw vanadium slag, all the raw phases were decomposed completely. However, there was no oxidization process for both V(III) and Fe(II). When the raw slag was roasted with ABS in the air, a part of the iron element was oxidized into Fe(III), which is shown in Figure 4c. In addition, the characteristic covalent peak of S=O detected by FT-IR analysis was confirmed from TiOSO₄.

In terms of the Fe²⁺V_nFe_2−nO₄ (0 < n < 2) phase, it was a complex vanadium iron compound. FeV₂O₄ and Fe₂VO₄ are two representative phases of Fe²⁺V_nFe_2−nO₄ (0 < n < 2), of which the standard pattern is shown in Figure 4a. Compared with the XRD standard pattern of FeV₂O₄, there was only a little change in characteristic peaks of Fe₂VO₄. Fe(II) was in the FeV₂O₄, but a Fe(II) and a Fe(III) in the Fe₂VO₄ [23]. It was difficult to give an accurate value of n in the phase of Fe²⁺V_nFe_2−nO₄ (0 < n < 2). However, it was suggested that a lattice structure transformation of vanadium–iron spinel occurred.

4.3. Valence State of Key Metal Elements

A further XPS analysis was conducted to confirm the valence state of Fe, V, Ti and Mn in the slag, and the results are shown in Figure 5a–d. Figure 5a shows that the Fe 2p, Fe 3p, V 2p, Mn 2p and Ti 2p elements were detected by XPS in the slag. In terms of the curve (I) from raw slag, it presents Fe 2p_3/2 in 710.5 eV, Fe 2p_3/2 in 711.05 eV, Fe 2p_1/2 in 724.3 eV and Fe 2p_1/2 in 725 eV, are all correspond to Fe²⁺-O bond [24,25,26]. As known, it is a complex process for vanadium slag production, and a very few iron phases may also be wrapped in the silicate phase theoretically. However, this part of iron cannot be detected by instruments or calculated accurately by chemical composition analysis in our research. We can see that the peaks in the position of 515.8 and 516.8 eV binding energy were both V 2p_3/2, which contribute to 40.09% and 59.91% proportion in V³⁺-O bond, respectively [27,28]. It is Ti 2p_3/2 in 458 eV and Ti 2p_1/2 in 464.19 eV that corresponds to Ti⁴⁺-O bond [29,30], while Mn 2p_3/2 in 641.2 eV, Mn 2p_1/2 in 652.6 and 653.4 eV, correspond to Mn²⁺-O bond [31,32]. This fact indicates that Fe²⁺, V³⁺, Ti⁴⁺ and Mn²⁺ were the main valence state of each element in the raw slag.

The curves (II) to (IV) expressed in Figure 5b give the valence state of Fe in the roasted slag. In curve (II), the results show that the strong peaks in the position of 710.5 and 725.0 eV binding energy were Fe 2p_3/2 and Fe 2p_1/2, which corresponds to the Fe²⁺-O bond [25]. Meanwhile, the strong peaks in the position of 711.4 and 724.0eV binding energy were Fe 2p_3/2 and Fe 2p_1/2, which corresponds to Fe³⁺-O bond [25,26]. The XPS analysis shows that both Fe (II) and Fe (III) were in the curve (II), which is consistent with the XRD analysis. Although there was no obvious valance and energy level change in V³⁺-O bond, the proportion of V 2p3/2 in position of 515.8 and 516.8 eV binding energy was changed to 27.14% and 72.86%, which also indicates the lattice transition of FeV₂O₄ [27,28]. In the curve (III), the XPS spectrum indicates that the binding energy of 711.6 and 713.6 eV presents Fe 2p_3/2, which corresponds to the Fe²⁺-O bond [33,34]. The binding energy of 725.0 eV presents Fe 2p_1/2, which also corresponds to the Fe²⁺-O bond. In the curve of (IV), the binding energy of 712.3 eV was Fe 2p_3/2 and 725.0 eV was Fe 2p_1/2, which correspond to Fe³⁺-O and Fe²⁺-O bonds, respectively [25,35].

In terms of V, Mn and Ti spectrum from Figure 5c,d, there was no valance state change in V³⁺-O, Mn²⁺-O and Ti⁴⁺-O bond. These results indicate that the vanadium slag can be decomposed by the process of air oxidization, which is consistent with the thermodynamic calculations, FT-IR and XRD analysis. Overall, the phase transformation of Fe, V, Ti and Mn compounds in vanadium slag via roasting with (NH₄)₂SO₄ can be shown in

Table 4. Phase transformation of Fe, V, Ti and Mn compounds in vanadium slag via roasting.

Materials	Roasting Atmosphere	Main Phase	Main Valance Sate
raw slag	without roasting	FeV2O4	Fe2+, V3+, Ti4+, Mn2+
		Fe2TiO4
		MnV2O4
raw slag	air	Fe2+VnFe2−nO4 (0 < n < 2)	Fe2+, Fe3+, V3+, Ti4+, Mn2+
		Fe2TiO4
		MnV2O4
raw slag + ABS	N2	(NH4)2Fe(SO4)2	Fe2+, V3+, Ti4+, Mn2+
		NH4V(SO4)2·2H2O/NH4V(SO4)2
		TiOSO4
		MnSO4
raw slag + ABS	air	NH4Fe(SO4)2	Fe2+, Fe3+, V3+, Ti4+, Mn2+
		NH4V(SO4)2·2H2O/NH4V(SO4)2
		MnSO4
		TiOSO4

. Thus, when the slag was directly roasted in the air, the lattice structure of the Fe-contained spinel would be transformed from FeV₂O₄ to Fe²⁺V_nFe_2-nO₄ (0 < n < 2), but there is no obvious change for Ti-contained spinel and Mn-contained spinel. Using NH₄HSO₄ (ABS) as an additive and roasting the slag in the N₂ atmosphere, those spinels would be decomposed into various salts, such as (NH₄)₂Fe(SO₄)₂, NH₄V(SO₄)₂·2H₂O, NH₄V(SO₄)₂, MnSO₄ and TiOSO₄. Meanwhile, when the slag was roasted with ABS in the air, a part of Fe(II) in(NH₄)₂Fe(SO₄)₂ would be transferred into Fe(III), but V(III), Ti(IV) and Mn(II) from those salts would remain the same valance state.

4.4. Morphology of the Roasted Slag

Figure 6a–d shows the morphology of the raw and the roasted slags observed by SEM. It is seen that the particle of the raw slag was quite dense, appearing in the form of blocks or rods. Even though the slag was roasted in the air for 60 min, there was no obvious change in its morphology. When the slag was roasted with ABS, it presented to be porous with a flaky crystal of ABS. However, it seems no obvious change in apparent morphology for using air or N₂ gas.

4.5. Effect of the ABS Amount on the Extraction Ratio of V and Ti from the Slag

The effect of ABS amount on the extraction ratio of V and Ti was investigated at 663 K, with a roasting time of 60 min in the air. Thereafter, the slag was leached at 338.15 K for 90 min using an 8 vol.% H₂SO₄ solution.

Figure 7 shows the extraction ratio of V was approximately 50% at a mass ratio of 1:1, and about 88% V and 81% Ti would be extracted at 1:5. When the mass ratio of V-slag to ABS was increased from 1:1 to 1:11, the extraction rate of V increased first from 49.83% to 88.59% and then decreased to 73.2%, and the Ti extraction rate increased first from 41.38% to 81.18% and then decreased to 66.96%. Compared with no ABS added, an appropriate ABS amount will promote the V and Ti extraction. However, excessive ABS will decrease the extraction rate because it discharges more NH₃ gas and leads to the stratification of raw material [5].

Based on the phase analysis result and extraction data, a sustainable technological route for recovering V and Ti from vanadium slag could be suggested, which is shown in Figure 8. The whole process consists of oxidation roasting, leaching, Ti-hydrolysis, V-precipitation, Fe-precipitation and AS circulation. Besides precipitation, possible separation routes also contain solvent extraction, ion exchange, and ionic liquid separation [36,37,38,39]. The energy conservation and emission reduction of extraction V and Ti from V-slag by ammonium salt can refer to the reference [40].

5. Conclusions

This work focused on investigating the decomposition and transformation mechanism of spinel from vanadium slag using ammonium salt as an additive agent. Thermodynamic calculation indicates that the main phase of vanadium slag, such as FeV₂O₄, Fe₂TiO₄, MnV₂O₄, Fe₂SiO₄ and Ca(Fe, Mg)(SiO₃)₂, can be oxidized in the air or be decomposed with ABS addition when the temperature was above 573.15 K. A further experimental investigation indicates that when the slag was roasted in the air at 653.15 K, a part of Fe (II) was oxidized into Fe(III), causing the lattice transformation from Fe₂VO₄ to Fe²⁺V_nFe_2−nO₄ (0 < n < 2). When the vanadium slag was roasted with ABS in N₂ gas, there was no obvious oxidization process for both V(III) and Fe(II), however, the slag will be decomposed into (NH₄)₂Fe(SO₄)₂, NH₄V(SO₄)₂·2H₂O, NH₄V(SO₄)₂, MnSO₄, SiO₂ and TiOSO₄. Meanwhile, if the vanadium slag was roasted with ABS in the presence of air, a part of the iron element will be oxidized into Fe(III). Under this condition, the main phases of the roasted samples were NH₄Fe(SO₄)₂, (NH₄)₂Fe(SO₄)₂, NH₄V(SO₄)₂·2H₂O, NH₄V(SO₄)₂, MnSO₄, SiO₂ and TiOSO₄. When the slag was roasted with ABS at a 1:5 mass ratio and 663.15 K for 60 min, 88% V and 81% Ti could be recovered using an 8 vol.% H₂SO₄ solution. This work aims to give an applicable reference for the application of the ammonium salt roasting process in the sustainable recovery of V, Ti and Fe from vanadium slag in the industry.