Distribution of P₂O₅ between P-rich Phase and Matrix Phase in P-bearing Steelmaking Slag

Introduction

With the rapid development of steel industry in China, the cumulative amount of slag has increased to more than 200 million tons, which affects the environment. Owing to the dephosphorization in steel-making process and the past use of phosphorus-rich ore in recent years, slags rich in phosphorus are being produced. If P₂O₅ in the slag can be enriched and separated, the separated P₂O₅ phase may be used as phosphoric fertilizer or phosphoric fertilizer additives, while the rest of the ingredients may be recycled in the iron- and steel-making processes, e. g., sintering, hot metal desiliconization, and hot metal dephosphorization process. If steel slag recycling is realized, the utilization of converter slag in agriculture will minimize environmental pollution and create financial benefit.

The fertilizer efficiency of slag phosphate fertilizer depends on %(P₂O₅) and P₂O₅ solubility. The %(P₂O₅) in slag phosphate fertilizer is related to the phosphorus enrichment and separation from the slag, while the degree of phosphorus enrichment is dependent on the phosphorus distribution among the different phases in the slag [1, 2, 3]. Therefore, to efficiently recover phosphorus resources from slag to use as fertilizer, it is critical to study the phosphorus distribution between the P-rich phase and matrix phase in slag. To date, phosphorus existence form in slag and phosphorus distribution between steel and slag have been extensively studied [3, 4,, 5,, 6, 7]. Inoue and Suito [8] found that the phosphorous transfer rate from slag to 20 to 50 μm long dicalcium silicate (2CaO·SiO₂) particles is considerably fast and the 2CaO·SiO₂ particle changes to the particle with the composition of n2CaO·SiO₂-3CaO·P₂O₅ (for short nC₂S-C₃P) solid solution within 5 s. Shimauchi et al. [9] studied the distribution of P₂O₅ between solid dicalcium silicate and liquid phases in the CaO–SiO₂–Fe₂O₃ system, and found that high P₂O₅ content and reasonable T.Fe content were controlled to show the high distribution ratio, and the addition of MgO and MnO has no effect on the distribution ratio of P₂O₅. Panlevani et al. [10] studied the distribution of P₂O₅ between the nC₂S-C₃P solid solution and liquid phase, and investigated the effect of MgO, MnO, and Al₂O₃ on the distribution of P₂O₅ in slags with variable iron oxide (FeO or Fe₂O₃) content.

There is, however, lack of studies regarding the effect of CaF₂ and Na₂O on the distribution of P₂O₅ between P-rich phase and matrix phase in the slag. In this paper, the effect of MnO, MgO, Na₂O and CaF₂ on the phosphorus existence form and the distribution ratio of P₂O₅ between P-rich phase and matrix phase in slag was studied systematically, γP2O5 and C_P in P-rich slag phase and matrix phase were analyzed, additionally, the effect of adding different oxides on distribution ratio of P₂O₅ between P-rich slag phase and matrix phase in slag was discussed, which provide the theoretical foundation for phosphorus enrichment and separation in slag, and provide the necessary research basis for the resource utilization of P-bearing steelmaking slag using as slag phosphate fertilizer.

Research method

Reagent grades CaO, SiO₂, Fe₂O₃, P₂O₅, MgO, MnCO₃, CaF₂ and Na₂CO₃ were mixed in various ratios to produce the CaO-SiO₂-Fe₂O₃-P₂O₅ slag system. To investigating the effect of other elements on the distribution of P₂O₅ between the P-rich phase and matrix phase, Na₂O, CaF₂, MgO and MnO were also added. The mixing conditions are summarized in Table 1. Figure 1 shows the compositional range of slags in the CaO-SiO₂-Fe₂O₃ ternary phase diagram calculated by using the FactSage 6.3 for samples X. As shown in Figure 1, all experimental slags were in the dicalcium silicate (C₂S) primary zone, C₂S precipitated first during the cooling process, and then nC₂S-C₃P solid solution (mainly Ca₁₅(PO₄)₂(SiO₄)₆ and Ca₅(PO₄)₂SiO₄) formed. Most of the iron oxide would be in the form of FeO in steelmaking condition, but Fe₂O₃ was used as iron oxide in this experiment and the heating was carried out under the air atmosphere, the reasons are as follow: (1）Ito et al. [8] disclosed that the distribution behavior of P₂O₅ between solid solution and liquid phases was not different, when the iron oxide changed from FeO to Fe₂O₃. (2) Shimauchi et al. [9] found that FeO was solved into the solid solution but the content of Fe₂O₃ in solid solution was negligible small. (3) According to the preliminary experiments, it was found that phosphorus enrichment behavior between solid solution and liquid phases was not affected, when the iron oxide changed from FeO to Fe₂O₃.

Figure 1:

Observed slag composition in the ternary phase diagram of the CaO-SiO₂-Fe₂O₃ system.

The mixed slags (200 g) were placed in magnesia crucible, 60 mm in diameter and 100 mm long that were subsequently placed inside graphite crucibles and heated in a MoSi₂ electric resistance furnace up to 1773 K. The temperature was maintained for 30 min to ensure the slag fully melting. Then, the slags were cooled to 1623 K at 3 K/min and maintained at this temperature for 1 h to fully promote the precipitation of 3CaO·P₂O₅. Then, they were cooled to 1423 K at 3 K/min, and the furnace was subsequently turned off and the slag samples were left to cool in the furnace (Figure 2). The phase morphology of the prepared slag samples was observed by an S-360 scan electron microscope, and each phase composition was analyzed by a Tracor Northern spectrometer. After each experiment, the slags were ground to less than 300 mesh (48 μm), and mineralogical phases were determined by XRD analysis. Diffraction patterns were measured in a 2θ range of 10~90° using copper K_α radiation of 40 kV and 30 mA, and the scan speed was 5°/min.

Figure 2:

Experimental conditions for precipitation of C₂S-C₃P solid solution.

Experiment results

Figure 3 shows a SEM-BSE image (Scanning Electron Microscope- Back scattered Electron Imaging) of a sample after cooling. The EDS date for the sample is given in Table 2. Figure 4 shows the X-ray diffraction results of synthetic slag samples.

Figure 3:

SEM-BSE image of experimental samples (1- P-rich phase, 2- RO phase, 3- Matrix phase).

Figure 4:

XRD results of different slag composition modification.

The results suggest that the P-bearing slag mainly consists of P-rich phase, matrix phase, and RO phase. The %(P₂O₅) in matrix phase is low, the matrix phase mainly comprise of nCaO·SiO₂ and merwinite (Ca₃MgSi₂O₈), and the RO phase mainly consists of iron oxide or iron and magnesium oxide.

%(P₂O₅) in the P-rich phase in A slag is about 20%, the phosphorus in P-rich phase is mainly in the existence of 6C₂S-C₃P solid solution. For adding MgO or MnO into the CaO-SiO₂-P₂O₅-Fe₂O₃ slag, MnO and MgO mainly enters into RO phase to form MnFe₂O₄ and MgFe₂O₄, and has little change in the %(P₂O₅) in P-rich phase (also about 20%), so which has little effect on the degree of phosphorus enrichment and phosphorus occurrence form.

In the A(3%Na₂O) slag, the added Na₂O mainly replaces calcium ions(Ca²⁺) in the P-rich phase (6C₂S-C₃P solid solution) to form C₂S-C₃P and Na₂Ca₄(PO₄)₂SiO₄, finally the P-rich phase is mainly 6C₂S-C₃P, C₂S-C₃P exists along with Na₂Ca₄(PO₄)₂SiO₄, and generally contains about 23–26% P₂O₅, and its particle size increased to more than 100 μm. In the A(6%Na₂O) slag, %(P₂O₅) in the P-rich phase further increased to 32%, and owning to the binding capacity of sodium ions(Na⁺) combine with phosphate ions((PO₄)³⁻) are stronger than calcium ions(Ca²⁺), calcium ions(Ca²⁺) in phosphorus-rich phase (6C₂S-C₃P or Na₂Ca₄(PO₄)₂SiO₄) are replaced by sodium ions(Na⁺), then Na₃PO₄ are formed, so the phosphorus existence form is changed.

In the A(3%Ca₂F) slag, %(P₂O₅) in the P-rich phase is also obviously higher than in the A slag. The added fluorine mainly enters the P-rich phase to form fluorapatite (Ca₅(PO₄)₃F), and %(P₂O₅) in the P-rich phase is generally about 34%, the particle size further increased to more than 100 μm. In the A(6%Ca₂F) slag, %(P₂O₅) in theP-rich phase is increased to 37%, whereas the particle size remained the same as that in the A(3%Ca₂F) slag, and the content and crystallization rate fluorapatite (Ca₅(PO₄)₃F) in the P-rich phase increased with increasing of CaF₂ content. The addition of CaF₂ significantly increases the %(P₂O₅) in the P-rich phase. Fluorapatite cannot be dissolved in 2% citric acid solution, therefore, the P₂O₅ solubility of slag decreases [11, 12].To further verify existence form of phosphorus and fluorine in P-bearing slag, the phase diagram at 1273 K was calculated in CaO-P₂O₅-CaF₂ system using the FactSage 6.3 software. As shown in Figure 5, P and Q in the diagram correspond to the area of A(3%Ca₂F) slag and A(6%Ca₂F) slag respectively. It shows that the phosphorus and fluorine mainly precipitated as fluorapatite crystal during the slag cooling process, which is consistent with the results of XRD.

Figure 5:

The CaO-P₂O₅-CaF₂ ternary phase diagram at 1273 K that calculated by FactSage 6.3.

Discussion

Based on SEM and EDS date in Figure 3 and Table 2, the P-bearing slag consists of P-rich phase, matrix phase, and RO phase, and the phosphorus is mainly in the P-rich phase and matrix phase. To improve dephosphorization and to increase the applications of P-bearing slag, the distribution of P₂O₅ between the P-rich phase and matrix phase is discussed in this paper.

The effect of slag composition on L_p’ between P-rich phase and matrix phase

The phosphorus in slag is transferred from the matrix phases to the 2CaO·SiO₂ particles to form nC₂S-C₃P solid solution. The distribution ratio of P₂O₅ between P-rich phase and matrix phase (L_p’=%(P₂O₅)_SS/%(P₂O₅)_M) is used to represent the degree of phosphorus enrichment in the slag in this study, %(P₂O₅)_SS stands for the P₂O₅ mass fraction in the P-rich phase, and %(P₂O₅)_M stands for the P₂O₅ mass fraction in the matrix phase. The bigger the L_p’ is, the more phosphorus is transferred from the matrix to the P-rich phase.

The relation between L_p’ and %(T.Fe) in the matrix phase of different slag systems is shown in Figure 6. L_p’ increases with increasing of %(T.Fe) in the matrix phase. The addition of MgO and MnO into the CaO-SiO₂-Fe₂O₃-P₂O₅ slag system affects the L_p’ negligibly. %(P₂O₅)_SS increases with increasing Na₂O in the slag, however, since %(P₂O₅)_M also increases, L_p’ does not change. However, L_p’ increases evidently with increasing CaF₂ in the slag. It is mainly due to that the melting point of slag is reduced effectively, the fluidity of the slag is increased, the phosphorus transfer from matrix phase to P-rich phase in the slag is promoted, and a new phase fluorapatite is formed with the addition CaF₂ into the slag, so the %(P₂O₅)_SS increases significantly.

Figure 6:

The effect of %(T.Fe) in the matrix phase on L_p’ in different slag systems.

The basicity of matrix phase has an important effect on L_p’, the %(CaO) of the matrix phase is used to show the basicity, as shown in Figure 7. In the CaO-SiO₂-Fe₂O₃-P₂O₅-X (X=MgO, MnO, Na₂O and CaF₂) slag system, L_p’ decreases with increasing of %(CaO) in the matrix phase.

Figure 7:

The effect of %(CaO) in the matrix phase on L_p’ in different slag systems.

The effect of slag composition on the activity coefficient of P₂O₅

As the P-rich phase and the matrix phase of the final slag is in equilibrium, the activities of P₂O₅ in both phases are the same, thus, L_p’ is proportional to the ratio of the activity coefficient of P₂O₅ between the matrix phase and P-rich phase, as shown in eq. (1), where a is the activity; γ is the activity coefficient, X is the mol fraction, and kis the coefficient for converting the mass percentage to mol fraction. Subscripts P₂O₅ (M) and P₂O₅ (SS) denote the activity and activity coefficient of P₂O₅ in the matrix phase and P-rich phase, respectively. The activity coefficient of P₂O₅ in the matrix phase is calculated by using a regular solution model [7], while the activity coefficient of P₂O₅ in the P-rich phase is calculated by using eq. (1).

(1)LP=(%P2O5)SS(%P2O5)M=kaP2O5(SS)×γP2O5(M)aP2O5(M)×γP2O5(SS)=kγP2O5(M)γP2O5(SS)

Activity coefficient of P₂O₅ in the P-rich phase

Figures 8 and 9 show the relation between %(T.Fe), %(CaO) in matrix phase and the γP2O5(SS) in different slag systems. The γP2O5(SS) decreases with the addition of MgO, MnO, Na₂O, and CaF₂ in the slag, whereas the γP2O5(SS) decreases minimally for the addition MgO, MnO, and CaF₂ in the slag, and the γP2O5(SS) decreases significantly after adding Na₂O. Meanwhile, γP2O5(SS) increases with decreasing the %(CaO) in the matrix phase in different slag systems. In the regular model, the interaction energy between other cations and Ca²⁺ is negative, and the γP2O5(SS) and Ca²⁺ content in the slag is a negative correlation, which results in γP2O5(SS) increases with decreasing Ca²⁺ when the %(CaO) in the slag decreases.

Figure 8:

The effect of %(T.Fe) in the matrix phase on the γP2O5(SS) in different slag systems.

Figure 9:

The effect of slag compositions on the γP2O5(SS) at different %(CaO) in the matrix phase.

Figure 10 shows the relation between %(P₂O₅)_SS and the γP2O5(SS) in different slag systems. %(P₂O₅)_SS minimally changes with increasing %(MgO) and %(MnO) in the slag, whereas %(P₂O₅)_SS increases with increasing %(Na₂O) and %(CaF₂) in the slag. In the CaO-SiO₂-Fe₂O₃-P₂O₅-X(X=MgO, MnO, Na₂O and CaF₂) slag system, γP2O5(SS) increases with increasing %(MnO), %(MgO) and %(CaF₂), whereas γP2O5(SS) decreases with increasing %(Na₂O).

Figure 10:

The effect of %(P₂O₅)_SS on the γP2O5(SS) in different slag systems.

Activity coefficient of P₂O₅ in the matrix phase

Figures 11 and 12 show respectively the relation between %(T.Fe), %(CaO) in the matrix phase and the γP2O5(M) in different slag systems. The γP2O5(M) minimally changes with increasing %(MgO), %(MnO) and %(CaF₂) in the slag, whereas γP2O5(M) decreases with increasing %(Na₂O) in the slag. Meanwhile, γP2O5(M) increases with decreasing %(CaO) in the matrix phase. As the P-rich phase and the matrix phase of the final slag are in equilibrium, the activities of P₂O₅ in both phases are the same (see eq. (1)), when the %(P₂O₅)_SS and the matrix phase are constant, γP2O5(SS) also increases with decreasing %(CaO), which is consistent with the results shown in Figure 9.

Figure 11:

The effect of %(CaO) in the matrix phase on the γP2O5(M) in different slag systems.

Figure 12:

The effect of %(T.Fe) in the matrix phase on the γP2O5(M) in different slag systems.

The effect of slag composition on the phosphorus capacity

Phosphorus capacity (C_P) is used to judge the capacity of a slag to contain phosphorus. The P-rich phase and the matrix phase in the slag are regarded as two different slag systems and the C_P in the P-rich phase (C_P(SS)) and the matrix phase (C_P(M)) are studied to analyze the phosphorus distribution between P-rich phase and matrix phases. The C_P was calculated by using eq. (2) [13].

(2)logCP=0.51×(23NCaO+17NMgO+8NFetO+33NNa2O+20NCaF2+13NMnO−26NP2O5)+29920T−19.280+log(%P)/NP2O51/2

Phosphorus capacity of the P-rich phase

Figures 13 and 14 show respectively the relation between %(T.Fe), %(CaO) in matrix phase and C_P(SS) for in slag systems. C_P(SS) increases with increasing MgO, MnO, Na₂O and CaF₂ in the slag. C_P(SS) increases with increasing %(CaO) in the matrix phase, as shown in Figure 13. Figure 14 shows that when the %(T.Fe) in matrix phase is less than 2%, C_P(SS) decreases with increasing %(T.Fe) in the matrix phase, whereas when %(T.Fe) in the matrix phase is more than 2%, C_P(SS) increases with increasing %(T.Fe) in the matrix phase.

Figure 13:

The effect of %(CaO) in the matrix phase on C_P(SS) in different slag systems.

Figure 14:

The effect of %(T.Fe) in the matrix phase on C_P(SS) in different slag systems.

It can be seen from Figure 15 that C_P(SS) linearly decreases with increasing %(P₂O₅)_SS. C_P(SS) is lowest in the slag with CaF₂, whereas C_P(SS) is highest in the slag with Na₂O, which is consistent with the results of Suito [14]. The decreasing C_P means that the capacity of containing phosphorus decreases as well.

Figure 15:

The effect of %(P₂O₅)_SS on C_P(SS) in different slag systems.

Phosphorus capacity of the matrix phase

Figures 16 and 17 show respectively the relation between %(T.Fe), %(CaO) in matrix phase and C_P(M) in different slag systems. C_P(M) increases with the addition of MgO, MnO, Na₂O and CaF₂ into the slag, and C_P(M) increases with increasing %(CaO) in the matrix phase. From Figure 18, it can be seen that when %(P₂O₅)_M is less than 3%, C_P(M) linearly increases with increasing %(P₂O₅)_M. When %(P₂O₅)_M is more than 3%, C_P(M) minimally changes with increasing %(P₂O₅)_M. C_P(M) is highest for the slag with Na₂O, and lowest in the slag with CaF₂.

Figure 16:

The effect of %(CaO) in the matrix phase on C_P(M) in different slag systems.

Figure 17:

The effect of %(T.Fe) in the matrix phase on C_P(M) in different slag systems.

Figure 18:

The effect of %(P₂O₅)_M on C_P(M) in different slag systems.

Conclusions

The effect of slag composition on phosphorus existence form in P-bearing slag was well researched, and the distribution behavior of P₂O₅ between P-rich phase and matrix phase was analyzed. The key findings are as follows:

1. The slag is composed of P-rich phase, matrix phase, and RO phase. Phosphorus in the slag is mainly in the form of nC₂S-C₃P solid solution in the P-rich phase. For CaO-SiO₂-Fe₂O₃-P₂O₅-X (X=MgO, MnO, Na₂O, and CaF₂ respectively) slag system, MgO and MnO are in RO phase and have no effect on the phosphorus existence form in the slag and the %(P₂O₅)_SS. Na₂O and CaF₂ in the slag change the phosphorus existence form in the slag, and significantly increase the %(P₂O₅)_SS.

2. With increasing %(T.Fe) and decreasing %(CaO) in the matrix phase, L_p’ increases. MnO and MgO have little effect on L_p’, whereas Na₂O and CaF₂ strongly affect L_p’. Furthermore, L_p’ clearly increases with increasing CaF₂ in the slag.

3. %(CaO) in the slag strongly affect γP2O5(SS) and γP2O5(M). Both γP2O5(SS) and γP2O5(M) decrease with increasing %(CaO) and γP2O5(SS) increases with increasing %(P₂O₅) in the slag. MgO, MnO and CaF₂ in the slag minimally affects γP2O5(SS) and γP2O5(M), whereas Na₂O in the slag causes γP2O5(SS) and γP2O5(M) to decrease obviously.

4. With increasing %(P₂O₅)_SS, C_P(SS) decreases and C_P(M) increases. When %(T.Fe) in the matrix phase is less than 2%, C_P(SS) decreases with increasing %(T.Fe) in the matrix phase. When %(T.Fe) in the matrix phase is more than 2%, C_P(SS) increases with increasing %(T.Fe) in the matrix phase. For %(P₂O₅)_M is less than 3%, C_P(M) linearly increases with increasing %(P₂O₅)_M. For %(P₂O₅)_M is more than 3%, C_P(M) minimally changes with the increasing %(P₂O₅)_M. C_P(SS) and C_P(M) increase with increasing %(CaO) in the matrix phase. In the CaO-SiO₂-Fe₂O₃-P₂O₅-X (X=MgO, MnO, Na₂O and CaF₂) slag system, C_P(SS) and C_P(M) are highest when Na₂O is added and lowest when CaF₂ is added.