3.1. The Hydrolysis Behavior and Occurrence form of Fe/V in MTA
The main impurity elements of TiOSO
4 solution obtained from TBFS were Mg, Al, Fe, V, and Mn, where both Fe and V possibly existed in two states (Fe
2+/Fe
3+ and V
3+/VO
2+).
Table 3 shows the theoretical pH at the beginning of the hydrolysis of these impurity ions. The precipitation pH was calculated by the ion concentrations and
Ksp of the corresponding hydroxides. Al
3+, Fe
3+, and V
3+ had low precipitation pH, which meant they were easy to hydrolysis.
Figure 1a summaries the two processes for the synthesis of MTA from the impurity bearing TiOSO
4 solution. Firstly, in order to ensure the hydrolysis ratio would be >95%, CaO slurry was added to neutralize the H
2SO
4 with a decrease in acidity from
F = 3.02 to
F = 1.50 (
F is the acidity factor defined in Equation (1)). The reduction of the solution acidity promoted the hydrolysis of Fe/V. If the solution was directly hydrolyzed at 110 °C, Fe-MTA was obtained consistent with the hydrolysis of Fe
3+. In the sulfate process, Ti
3+ was used to reduce the Fe
3+ to Fe
2+ to inhibit the hydrolysis of Fe and the adsorption of Fe
3+ [
12]. In order to generate Ti
3+, we applied Al reduction (
). The added Al did not significantly affect the concentration of Al in the solution. However, if the solution was reduced, V-MTA was formed. The composition of Fe-MTA and V-MTA is shown in
Table 4; all the impurity contents are in terms of TiO
2 wt.%. Though Al was hydrolyzed, it was colorless and therefore did not affect the pigment quality.
Figure 1b shows the Fe/V hydrolysis behavior in the unreduced solution. The concentration of V was unchanged, while the concentration of Fe was decreasing exponentially. This means the only Fe was hydrolyzed and precipitated with MTA. With the hydrolysis of TiOSO
4, the concentration of H
2SO
4 was increased (
), so finally the hydrolysis of Fe was inhibited and the concentration of Fe was unchanged.
Figure 1c shows the Fe/V hydrolysis behavior in the Al reduced solution. Though Fe was unhydrolyzed due to Al reduction with an unchanged concentration, V was hydrolyzed and the concentration was decreasing exponentially like Fe. The reason for this was the formation of V
3+, as shown in Equations (2)–(4), where
E0 is the standard electrode potential of the reactions at 25 °C.
When the solution was unreduced, no Ti
3+ was formed, so the solution was in the oxidized state. Fe existed as Fe
3+, and V existed as VO
2+. Fe
3+ was easy to hydrolyze and formed Fe-MTA. When the solution was reduced by Al, Ti
3+ was generated, Fe
3+ converted to Fe
2+, and VO
2+ converted to V
3+. V
3+ was easy to hydrolyze and formed V-MTA. According to the Nernst equation, the concentration of Ti
3+ affects the redox potential of the solution and finally determines the concentration of Fe
2+/Fe
3+ and V
3+/VO
2+. Therefore, the aim was to find a suitable Ti
3+ concentration in the solution for minimal Fe/V hydrolysis, as shown in
Figure 1d. The results show that Fe or V in MTA was always >500 ppm at different Ti
3+ concentrations, so the Fe/V could not be removed simultaneously by Al reduction.
Figure 2a shows the XRD patterns of Fe/V-MTA and blank MTA. The blank MTA was synthesized from the Mg/Al-bearing TiOSO
4 solution at the same conditions with Fe-MTA. The solution was prepared from reagents (TiOSO
4, MgSO
4, Al
2(SO
4)
3) without Fe/V impurities. The crystalline product in these samples was anatase (PDF 21-1272).
Table 5 shows the analysis of the (101) peak. Fe/V-MTA had lower diffraction intensity and smaller crystal size than blank MTA, which means the Fe/V in the solution affected the hydrolysis of TiO
2+ and inhibited the growth of the anatase nanocrystals in MTA. The changes of 2θ for Fe/V-MTA indicate that Fe/V may have entered the anatase nanocrystals and changed the lattice spacing.
Figure 2b shows the XPS Ti 2p spectra of Fe/V-MTA and blank MTA. The peaks at ~465.1 and 459.4 eV of blank MTA were assigned to Ti 2p
1/2 and Ti 2p
3/2 of the Ti
4+ state [
26]. For Fe/V-MTA, the negative shift of binding energy ~1.8 eV indicated that some Ti
4+ ions were replaced by Fe/V, and the Ti-O-Ti in anatase changed to Ti-O-Fe/V.
Figure 2c shows the XPS Fe 2p spectra of Fe-MTA. The peaks centered at ~723.9 and 712.6 eV were assigned to Fe 2p
1/2 and Fe 2p
3/2 of the Fe (III) species, and the spin–orbit splitting characteristic of Fe
3+ corresponded to typical features of Fe(OH)
3 [
27].
Figure 2d shows the XPS V 2p spectra of V-MTA. The peaks centered at ~521.8 and 514.4 eV were assigned to V 2p
1/2 and V 2p
3/2 of V
3+, which was close to V(OH)
3 [
28,
29]. Therefore, the results indicate that some of the Fe
3+/V
3+ hydrolyzed with TiO
2+ at high temperature and entered the lattice of anatase nanocrystals so these impurities could not be removed by dilute H
2SO
4 or water.
3.2. Thermodynamic Equilibrium Calculation of the Hydrolysis Process
It has been proposed that Fe/V enters the MTA lattice by hydrolysis with TiO
2+, so the key to controlling the impurity content is to inhibit the hydrolysis of Fe
3+/V
3+. It is known that if a reaction cannot proceed thermodynamically, then kinetically the reaction cannot proceed. Therefore, assessing if the Fe
3+/V
3+ hydrolysis reaction will happen thermodynamically at different reaction conditions will help to identify the Fe
3+/V
3+ non-hydrolysis condition. Therefore, a thermodynamic calculation for the hydrolysis process of impurity bearing TiOSO
4 solution was carried out. The equilibria of the impurities-bearing TiOSO
4 solution for the thermodynamic calculation are given in Equations (5)–(16).
The standard equilibrium constant or relevant electrode potential was obtained from various handbooks and articles [
16,
17,
30,
31,
32]. By using Van’t Hoff, Kirchhoff’s, and Nernst equations, the equilibrium constants for Equations (5)–(16) at 110–150 °C were obtained, as shown in
Figure 3a. By using the Newton–Raphson method, based on the original composition of the solution in
Table 2 and equilibrium constants, the equilibrium composition was calculated, as shown in
Figure 3b,c. Here, α was referred to as the hydrolysis ratio, which was determined by Equation (17).
The temperature (110 °C),
F-value (1.50), and [Ti
3+] (0.01 mol·L
−1) used in the calculation were the experimental conditions. With increasing hydrolysis ratio, [SO
42−], [TiOSO
4], [TiO
2+], and [OH
−] decreased, while [HSO
4−], [H
+], and [H
2SO
4] increased, which was because H
2SO
4 was formed during the hydrolysis of TiOSO
4 (
). The concentration of impurity ions changed a little, but it is worth noting that the concentration of H
+ was relatively low (0.001–0.01 mol·L
−1). As shown in
Table 6, the lower TiOSO
4 concentration compared to that for the sulfate process contributed to the lower free H
2SO
4, corresponding to the lower concentration of H
+. Even at higher
F-value, the free H
2SO
4 of the impurities-bearing TiOSO
4 solution was much lower than for the industrial sulfate process in the industry. Furthermore, a large quantity of SO
42- introduced by the high concentration of impurity ions from TBFS promoted a positive proceeding of Equation (5), so lots of H
+ was consumed.
Based on the equilibrium composition, the concentration quotient (
Qc) for Fe(OH)
3 or V(OH)
3 can be calculated, and whether Fe/V will be precipitated can be judged by comparing it with the corresponding solubility product (
Ksp). If
Qc >
Ksp, precipitation is expected, and vice versa. To validate the results of the equilibrium composition, a comparison was performed for the condition that is used to synthesize Fe-MTA and V-MTA, as shown in
Figure 4. For the Fe-MTA condition (
Figure 4a,b), at any hydrolysis ratio, Fe(OH)
3 formation was expected as
Qc >
Ksp and V(OH)
3 formation was expected as
Qc <
Ksp. For the V-MTA condition (
Figure 4c,d), V hydrolyzed while Fe did not at a low hydrolysis ratio, and neither Fe nor V was hydrolyzed at a high hydrolysis ratio. The calculations were consistent with the experimental results. Furthermore, the
Qc values for Fe(OH)
3/V(OH)
3 significantly decreased with the increasing of α, so increasing the concentration of free H
2SO
4 (
F-value) was expected to inhibit the hydrolysis of Fe/V.
To identify the conditions for Fe/V co-removal, a simulation for different hydrolysis conditions (temperature,
F-value, and Ti
3+ concentration) was performed and the results are shown in
Figure 5. As discussed before, increasing the concentration of free H
2SO
4 close to that of the industry standard is an effective way to inhibit the hydrolysis of Fe/V, so a suitable
F-value was 2.0–3.5. However, increasing the
F-value lowered the hydrolysis ratio, so a higher temperature was necessary (T = 130–150 °C). There were two ways for the co-removal: (1) inhibit Fe hydrolysis under oxidizing conditions (No Ti
3+) or (2) inhibit V hydrolysis under reducing conditions (0.01 M Ti
3+). The results show that Fe
3+ was always hydrolyzed at a low hydrolysis ratio (
Figure 5a,b), while V
3+ had unhydrolyzed conditions (
Figure 5c,d). Therefore, method (2) was optimal for Fe/V co-removal. The possible conditions were (130 °C,
F = 3.0) and (150 °C,
F = 3.5). However, for the condition (130 °C,
F = 3.0), the Ti hydrolysis ratio was only 0.638 after 4 h; likewise, for the condition (150 °C,
F = 3.5), the hydrolysis ratio was 0.842 after 4 h. Therefore, though the two possible conditions may ensure the Fe/V co-removal, they cannot ensure the Ti hydrolysis ratio >0.95 within a reasonable reaction time (4–6 h), which will lower the production efficiency.
3.3. Two-Step Hydrothermal Hydrolysis for Fe and V Co-Removal
To ensure a high hydrolysis ratio (experimental α > 0.95) within a reasonable reaction time (4–6 h), the hydrolysis properties of impurities-bearing TiOSO
4 solution were investigated, as shown in
Figure 6a. For the solution formed by dissolution of TBFS, the
F-value was 3.0. If the
F-value was higher (
F = 3.5), additional concentrated H
2SO
4 was required, and the temperature or reaction time for hydrolysis needed to be higher or longer. If the
F-value was lower (
F = 2.5), V
3+ would hydrolyze during the initial stages of Ti hydrolysis as shown in
Figure 5c,d. Thus,
F = 3.0 was selected as optimal. For
F = 3.0, the conditions whether V
3+ hydrolyzed at different temperatures and hydrolysis ratios were investigated, so that several hydrolysis processes were provided as shown in
Figure 6a,b. If the solution was hydrolyzed at 150 °C for 4 h, V
3+ may have hydrolyzed at the initial stages of Ti hydrolysis (
Qc >
Ksp when α = 0), as shown in
Figure 6b. If the solution was hydrolyzed at 130 °C for 4 h, though the V
3+ was not hydrolyzed, the hydrolysis ratio was very low (0.32, 2 h), as shown in
Figure 6a. Therefore, we provided a new two-step hydrothermal hydrolysis method, as shown in
Figure 6c. For Step 1, the impurities-bearing TiOSO
4 solution was hydrolyzed at low temperature to inhibit the V hydrolysis. According to
Figure 6b, the suitable condition for hydrolysis was at 110–130 °C for 2–4 h to allow Ti to partially hydrolyze. As hydrolysis proceeded, [H
2SO
4] was increased, and the Ti hydrolysis ratio > 0.25, V
3+ was not hydrolyzed at higher temperatures. Therefore, for Step 2, the solution was hydrolyzed at a higher temperature to ensure α > 0.95 within 6 h. According to
Figure 6c, to reduce the total hydrolysis time, the hydrolysis at 150 °C was conducted for 2 h.
Figure 7 shows the experimental validation of the two-step hydrothermal hydrolysis method, where only the conditions of Step 1 were varied.
Figure 7a shows the hydrolysis ratio was increased by the temperature and reaction time of Step 1; α > 0.95 could be achieved only if the reaction time was longer than 4 h or the temperature was 130 °C.
Figure 7b shows the Fe content in MTA was only 21 ppm for the reaction time of 2 h, which met the requirement of TiO
2 pigment (<30 ppm) and was much lower than Fe-MTA (3854 ppm), but it increased rapidly with hydrolysis time. This may have been because the oxygen in the autoclave gradually oxidized Ti
3+ to TiO
2+ as time went on, while Fe
2+ was converted to Fe
3+, resulting in an increase of the Fe content in MTA.
Figure 7c shows that the V content in MTA at the different conditions was almost the same, which was in the range of 140–160 ppm. Despite not meeting the requirement of TiO
2 pigment (>7 ppm), it was much lower than V-MTA (1205 ppm). Considering both the extent of hydrolysis and the Fe/V content, the optimal condition for Step 1 was hydrolysis at 130 °C for 2 h. The sample obtained with high purity is shown in
Figure 7d. Compared with the traditional method (
Figure 1a), the process was simpler. More importantly, Fe and V were both excluded by the two-step method (Fe = 21 ppm, V = 145 ppm). Compared with the optimal Fe/V content in
Figure 1d, the impurities content of the two-step method was notably decreased and the color was white. Compared with previous relevant works [
13,
14,
20], Fe was decreased from ~10,000 ppm to 21 ppm.
Though the two-step method reduced the Fe/V content of MTA, the V impurity was still high. Thus, we studied the possible reason for how V
3+ entered the MTA from which the results are shown in
Figure 8. As shown in
Figure 8a, the reduction in V concentration up to 4 h was significantly less than before (
Figure 1c). Furthermore,
Figure 8b shows that the two-step MTA and blank MTA had the same XRD (101) peak position of anatase, which means the V
3+ did not enter the lattice of MTA. This indicates that V
3+ did not enter the MTA by hydrolysis but was present due to other mechanisms.
Figure 8c compares the Ti 2p XPS spectra of the two-step MTA and blank MTA. The peaks at ~465.1 and 459.4 eV of blank MTA were assigned to Ti 2p
1/2 and Ti 2p
3/2 of the Ti
4+ state. For the two-step MTA, the negative shift of binding energy ~0.2 eV indicated that some V
3+ was located on the MTA surface and affected the electronic state of Ti
4+, as previous studies have shown [
33]. Therefore, V impurities in MTA may have been due to the adsorption of V
3+. Furthermore, the V cannot be washed off, so the adsorption was not a simple physical adsorption. According to the formation and structure of MTA, we provided a possible mechanism for how V
3+ enters the MTA in two-step hydrolysis, as shown in
Figure 8d. The SEM image shows that the two-step MTA formed from impurity bearing TiOSO
4 solution had hierarchical structures [
34]; the aggregation of 30–100 nm primary anatase TiO
2 particles to form final 1–3 μm MTA aggregates. Previous studies show that SO
42− (4–6 wt.% SO
3 of TiO
2) is strongly adsorbed at the surface of the primary particles [
34,
35]. Furthermore, V
3+ can readily complex with SO
42−, so the V
3+-SO
42− complex may be formed on the primary particles surface of MTA, like the interaction between Al
3+ and the hydrated TiO
2 clusters [
36,
37]. As hydrolysis proceeds, primary particles agglomerate and therefore V
3+ is trapped in MTA, which is difficult to wash off.
For future studies of Fe/V co-reduced, the key to reduce the V content in MTA is to inhibit the formation of the V3+-SO42− complex. Specifically, there are two ways: (1) selective complexation of V3+ with complexing agents, and (2) converting V3+ to VO2+ with oxidizing agents to facilitate the dissolution of V in solution.