Analysis of Dephosphorization Thermodynamics Based on the Melt Structure of CaO–SiO₂–FeO–MgO System

Abstract

To reduce the phosphorus content in steel, the CaO–SiO₂–Fe_xO–MgO-based dephosphorization slag was designed to clarify the microscopic reaction behavior of phosphorus. In the present study, Raman spectra of final slag were measured and deconvoluted. The spectral analysis showed that P⁵⁺ ions removed from liquid iron existed in the form of Q⁰(P) and Q¹(P) groups. According to the result, phosphate capacity and phosphorus distribution ratio (L_P^*) were redefined. The results showed that lgL_P^* was proportional to molar ratio of Q⁰(P)/Q¹(P), indicating that Q⁰(P) exerted a significant influence on L_P^* compared with Q¹(P). As the increases of CaO/SiO₂ from 0.15 to 1.25 and FeO from 7.2 to 21.9 mass% in the final slag, lgL_P^* gradually increased, whereas lgL_P^* showed a downward trend when FeO increased to 29.2 mass%. This was because the increase of O²⁻ ions generated by the dissociations of CaO, MgO and FeO continuously destroyed the network structure and formed more Q⁰(P) units, which were compensated by the increasing Ca²⁺, Mg²⁺, and Fe²⁺ cations to form stable groups. Meanwhile, the Q⁰(Si) units formed by slag depolymerization further played a role in fixing Q⁰(P). However, due to the stronger polarization of Fe²⁺ than Ca²⁺ and Mg²⁺, Q⁰(P) and Q¹(P) units were easily deformed and decomposed by Fe²⁺. The excessive Fe²⁺ diluted the proportion of Ca²⁺ and Mg²⁺, and made Q⁰(P) and Q¹(P) lose stability. The P⁵⁺ ions in the Q⁰(P) and Q¹(P) units were reduced to liquid iron, and the rephosphorization phenomenon occurred, resulting in a decrease of lgL_P^*.

1. Introduction

With the increasingly fierce competition in the steel market and the increasing demand for high-quality steel, the phosphorus content has become one of the important factors to measure the steel product quality. Phosphorus is an element of representative impurities that tends to segregate at grain boundaries, leading to the reduction of brittle characteristic at low temperature, crack resistance, weldability and mechanical properties of steel.^1,2) Low phosphorus content is essential for steel applications where high ductility is required, such as thin sheets, deep drawn structures, pipelines, and automobile exteriors.³⁾ The ability of phosphorus to strengthen and embrittle ferrite imposes restrictions on the maximum phosphorus content for the aforementioned applications. Therefore, phosphorus content should be kept as low as possible in conventional steel grades.

The modern refining process of liquid steel has almost no dephosphorization function, so basic oxygen smelting has become the key link to control the phosphorus content. Many scholars have carried out a lot of research on enhancing the dephosphorization efficiency of converter.^4,5,6,7) Studies have reported that low temperature, high oxygen partial pressure, and high CaO/SiO₂ ratio are advantageous for dephosphorization from liquid steel.^8,9) For example, Li et al.¹⁰⁾ studied the phosphorus distribution ratio (L_P) of CaO–FeO–SiO₂–Al₂O₃/Na₂O/TiO₂ slag and carbon-saturated iron under different temperatures and compositions. The results showed that L_P decreased with the increase of temperature and increased with the increase of basicity and FeO content. Meanwhile, it was found that the effect of temperature on L_P was weaker than that of basicity, but stronger than that of FeO content. Im et al.¹¹⁾ also proposed that L_P depended more on the CaO content than on the Fe_xO content of slag. Furthermore, Basu et al.¹²⁾ experimentally determined the influences of FeO concentration and basicity on the equilibrium phosphorus partition ratios of CaO–SiO₂–Fe_xO–P₂O₅–MgO slags at temperatures of 1600 and 1650°C. The results showed that the L_P initially increased with basicity but attained a constant value beyond basicity of 2.5. An increase in FeO concentration up to approximately 13 to 14 mass% was beneficial for phosphorus partition.

In addition, the mineral structure and the composition of phosphorus-rich phase in dephosphorization slag have been studied.^{13,14,15,16,17)} It was reported that the phosphorus was mainly enriched with the form of nCa₂SiO₄∙Ca₃(PO₄)₂ (hereafter denoted as nC₂S∙C₃P) in the phosphorus-rich phase whether the dephosphorization slag was cooled in the atmosphere of air or argon.^18,19,20,21) Besides, the measurements of the equilibrium distribution ratio of P₂O₅ between the solid C₂S·C₃P and the liquid phase indicated that P₂O₅ was concentrated in the solid C₂S·C₃P with a high distribution ratio.²²⁾ Hence, phosphorus in the liquid steel is usually considered to be oxidized to P₂O₅ and enter into the slag, and then is fixed by CaO to form C₃P, which forms a solid solution with C₂S in the slag. However, the above-mentioned solid solution does not really exist in the molten slag. For phosphorus-containing slag, phosphorus exists in the form of a tetrahedron containing one P=O double bond and three P–O or P–O–P bonds, and is divided into four structural units, Q⁰(P), Q¹(P), Q²(P), and Q³(P), according to the bridging oxygen number from low to high.^23,24) Similarly, silicon also acts as a network former to form five silicon-oxygen tetrahedrons, Q⁰(Si), Q¹(Si), Q²(Si), Q³(Si), and Q⁴(Si).^25,26) In order to truly evaluate the dephosphorization ability of molten slag, it is necessary to analyze the dephosphorization mechanism based on the real structure of molten slag.

In the present work, a typical dephosphorization slag of CaO–SiO₂–FeO–MgO system was designed to conduct the equilibrium experiments of dephosphorization at 1600°C. The melt structures of final slags were determined via Raman spectroscopy, and the Raman spectra were deconvoluted. According to the structural information, the ion theory of dephosphorization was expounded, and the phosphate capacity and phosphorus distribution ratio were redefined. Then, the relationship between the melt structure and phosphorus distribution ratio was analyzed and the dephosphorization thermodynamics was elucidated from the perspective of microstructure. The results would have great significance in providing guidance for enhancing dephosphorization efficiency and improving the steel quality.

2. Experimental Procedures

2.1. Materials Preparation

Referring to the variation of molten slag in the actual steelmaking process²⁷⁾ the lower melting point region was selected to determine the composition of the initial slag of dephosphorization, as shown in Fig. 1 and Table 1. The initial slags 1–4 and 5–8 were designed to investigate the effect of FeO and CaO/SiO₂ on the dephosphorization efficiency, respectively. The initial slags were prepared with AR-grade CaO, SiO₂, FeC₂O₄∙2H₂O, and MgO. FeO was added in the form of FeC₂O₄∙2H₂O. To remove moisture and impurities, each reagent except FeC₂O₄∙2H₂O was pretreated with high-temperature calcination in a muffle furnace.

Fig. 1.

Isothermal liquidus of the CaO–SiO₂–FeO-10%MgO slag system. (Online version in color.)

Table 1. Compositions of initial slags and the matrix phase in final slags (mass%).

Samples	CaO/SiO2	Initial composition/Composition of matrix phase in final slags
		CaO	SiO2	FeO	Fe2O3	MgO	P2O5
1	1.0/0.88	40/34.025	40/38.661	10/7.203	0/0.022	10/19.320	0/0.769
2	1.0/0.83	35/28.963	35/34.853	20/15.221	0/0.017	10/17.737	0/3.209
3	1.0/0.80	30/26.560	30/33.271	30/21.865	0/0.029	10/14.641	0/3.634
4	1.0/0.85	25/24.875	25/29.171	40/29.200	0/0.140	10/13.113	0/3.501
5	0.2/0.15	10/6.356	50/42.836	30/19.344	0/0.026	10/31.361	0/0.077
6	0.5/0.43	20/16.542	40/38.464	30/18.445	0/0.044	10/25.660	0/0.845
7	0.7/0.67	25/22.572	35/33.723	30/20.407	0/0.031	10/19.914	0/3.353
8	1.4/1.25	35/37.295	25/29.886	30/20.020	0/0.032	10/8.602	0/4.165

The iron block containing phosphorus was prepared by melting a pure iron of 280 g and a ferrophosphorus alloy of 5 g in the MoSi₂ furnace under an argon atmosphere. The final phosphorus content of iron block was confirmed by using ICP-OES with a resolution of 0.006 nm to be 0.47 mass%P.

2.2. Dephosphorization Experiment

Eight sets of laboratory-scale dephosphorization experiments were carried out in a high-temperature quenching furnace. The mass ratio of molten slag and liquid iron used in the dephosphorization experiment was 1:4. A Fe-0.47 mass%P of 20±0.5 g and a mixed initial slag of 5 g were placed in the MgO crucible (31 mm ID) hung with Mo wire in the hot zone of furnace. High-purity argon gas (99.999%) was introduced at a fixed flow rate of 1 L∙min⁻¹ to maintain the oxygen partial pressure at about 10⁻⁴ atm. Then, the temperature was raised to 1600°C and kept for 3 h. The preliminary experiment confirmed that the phosphorus content in molten slag and liquid iron remained constant after 15 minutes, which can be regarded as a quasi-equilibrium state. Therefore, a constant temperature of 3 hours was considered to be sufficient to reach the equilibrium of the dephosphorization reaction. Finally, Mo wire was loosened, and molten slag and liquid iron fell into the bucket directly below with MgO crucible to complete the quenching.

2.3. Analysis Method

The chemical analysis of the iron samples after the dephosphorization experiment was determined by ICP-OES, as listed in Table 2. The morphology and chemical composition of final slags were analyzed by SEM-EDS and XRD, as shown in Figs. 2 and 3. A small amount of precipitated phase was found in the final slag. Combining with XRD analysis, the precipitated phase was (Fe, Mg)O, which is generally considered to be caused by the fact that the quenching rate is insufficient.²⁸⁾

Table 2. Compositions of the iron samples after dephosphorization reaction (mass%).

Samples	O	P	Mn	Si	C	Al
1	0.120	0.180	0.012	0.108	0.020	0.024
2	0.092	0.160	0.010	0.122	0.026	0.022
3	0.149	0.057	0.008	0.134	0.016	0.019
4	0.199	0.072	0.013	0.06	0.012	0.026
5	0.171	0.370	0.014	0.128	0.014	0.025
6	0.075	0.170	0.009	0.050	0.032	0.023
7	0.096	0.120	0.011	0.129	0.025	0.020
8	0.104	0.032	0.014	0.116	0.023	0.017

Fig. 2.

Typical SEM images of final slags after dephosphorization. (Online version in color.)

Fig. 3.

Typical XRD pattern of final slag after dephosphorization.

The composition analysis of the matrix phase and precipitated phase of final slag was performed by using SEM-EDS to obtain the mass fractions of CaO, SiO₂, MgO, P₂O₅, and total iron (TFe₁) in the matrix phase. The mass fractions of total iron (TFe₂) and FeO of final slags were determined by XRF. Then Fe₂O₃ content in the matrix phase could be obtained by subtracting FeO content from TFe₂ content. Since Fe₂O₃ was not contained in the precipitated phase, Fe₂O₃ was considered to originate entirely from the matrix phase. The mass fraction of FeO in the matrix phase was calculated by the content of TFe₁ and Fe₂O₃. The composition of matrix phase is shown in Table 1.

The melt structures of final slags were measured by LabRAM HR800 Raman spectrometer equipped with a CCD detector. The specific measurement conditions are as follows: An Ar⁺ laser with an excitation wavelength of 633 nm was used as the light source to record the Raman spectrum. Before the test, a single crystal silicon standard sample with a known characteristic peak (520 cm⁻¹) was used to calibrate the wavenumber. The scanning range of the band was 200–2000 cm⁻¹, and the spectral resolution was 0.65 cm⁻¹. Although a small amount of FeO was precipitated in the final slag, the reason why we still used Raman spectroscopy to measure its melt structure was that the precipitation FeO as a network modification did not affect the formation of the main vibrational peaks in the Raman spectra. The subsequent result is discussed in terms of the composition of matrix phase.

3. Results and Discussion

3.1. Dephosphorization Rate and Equilibrium Distribution Ratio of Phosphorus

Figure 4 shows the P content in the iron samples and P₂O₅ content in the final slags under the condition of different initial slags. As the FeO concentration in final slag increases from 7.2 to 29.2 mass%, the P content in the iron samples first decreases from 0.18 to 0.05 mass% and then increases to 0.072 mass%, whereas the P₂O₅ content in final slags increases from 0.746 to 3.634 mass% and then decreases to 3.501 mass%. FeO exhibits the dual roles of promoting and hindering dephosphorization. As CaO/SiO₂ ratio increases from 0.15 to 1.25, the P content in the iron samples gradually decreases from 0.37 to 0.032 mass% and P₂O₅ content gradually increases from 0.077 to 4.165 mass%, indicating that high CaO/SiO₂ ratio is beneficial to dephosphorization. However, the P content in the iron samples and the P₂O₅ content in final slag cannot truly represent the reaction of molten slag and liquid iron due to the continuous changes in the total amount of iron and slag. Therefore, according to Eqs. (1) and (2), the dephosphorization rate (η_P) and the equilibrium distribution ratio of phosphorus (L_P) are calculated to evaluate the dephosphorization capacity of molten slag.   

$${\eta }_{\mathrm{P}}=\frac{{(m_{\mathrm{P}})}_{\mathrm{f}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} } \mathrm{s}\mathrm{l}\mathrm{a}\mathrm{g}}-{(m_{\mathrm{P}})}_{\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} } \mathrm{s}\mathrm{l}\mathrm{a}\mathrm{g}}}{{[m_{\mathrm{P}}]}_{\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} } \mathrm{i}\mathrm{r}\mathrm{o}\mathrm{n}}}\times 100\%$$(1)

  

$$L_{\mathrm{P}}=\frac{w{({\mathrm{P}}_2{\mathrm{O}}_5)}_{\mathrm{f}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} } \mathrm{s}\mathrm{l}\mathrm{a}\mathrm{g}}}{w{[\mathrm{P}]}_{\mathrm{f}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} } \mathrm{i}\mathrm{r}\mathrm{o}\mathrm{n}}}$$(2)

where (m_P)_{final slag} represents the mass of (P) in final slag; (m_P)_{initial slag} represents the mass of (P) in initial slag; [m_P]_{initial iron} represents the mass of [P] in the initial iron samples; w(P₂O₅) is the mass fraction of P₂O₅ in final slag; w[P] is the mass fraction of P in the iron sample after dephosphorization.

Fig. 4.

P content in the iron samples and P₂O₅ content in final slags under the condition of different initial slags. (Online version in color.)

According to the result of dephosphorization experiment, the calculated values of η_P and L_P are shown in Fig. 5. Through comparison, it is found that when FeO content is less than 21.9 mass% and CaO/SiO₂ ratio is less than 1.25, the values of η_P and L_P increase with increasing FeO and CaO/SiO₂. This phenomenon shows the same trend as the changes of P content in the iron samples and P₂O₅ content in final slag, and accords with the thermodynamic conditions of enhancing dephosphorization. Meanwhile, it can be seen from the increasing magnitude in the values of η_P and L_P that the increase of CaO/SiO₂ ratio has a more significant influence on η_P and L_P than the increase of FeO, which is consistent with the result reported by Basu et al.⁸⁾ As FeO content exceeds 21.9 mass%, η_P and L_P tend to decrease with the increase of FeO, where FeO play a role in hindering dephosphorization.

Fig. 5.

Dephosphorization rate and equilibrium distribution ratio of phosphorus under the condition of different initial slags. (Online version in color.)

3.2. Structural Analysis of Final Slag

Figure 6 shows the Raman spectra of final slags. Referring to the existing literature,^28,29,30,31) the vibration peaks in the wavenumber range of 400–1200 cm⁻¹ are deconvoluted. In the process of deconvolution, each characteristic peak is continuously fitted until the minimum correlation coefficient between the original curve and the fitted curve is greater than 0.99. The deconvolution result is shown in Fig. 7.

Fig. 6.

Raman spectra of final slags. (Online version in color.)

Fig. 7.

Deconvolution results of Raman spectra of final slags: (a) sample No. 1; (b) sample No. 2; (c) sample No. 3; (d) sample No. 4; (e) sample No. 5; (f) sample No. 6; (g) sample No. 7; (h) sample No. 8. (Online version in color.)

According to the information of the structural units in Fig. 7, the relative area fraction of each characteristic peak is counted. Actually, the area fraction fails to accurately reflect the change in the proportion of structural units caused by the variation of slag composition. Hence, its molar fraction can be calculated with the aid of Raman scattering coefficients according to Eq. (3), and the calculation results are shown in Fig. 8.   

$$X_n=\frac{A_n/S_n}{\sum _{n=0}^4{A}_n/S_n},n=0,1,2,3,4$$(3)

Where X_n represents the molar fractions of Qⁿ(Si) and Qⁿ(P); A_n represents the relative area fractions of Raman characteristic peaks; S_n represents the Raman scattering coefficients of Qⁿ(Si) and Qⁿ(P); S₀, S₁, S₂, S₃, and S₄ for Qⁿ(Si) is 1.000, 0.514, 0.242, 0.090,³⁴⁾ and 0.015; S₀ and S₁ for Qⁿ(P) is 1.000 and 0.576.²³⁾

Fig. 8.

Molar fractions of structural units in final slags. (Online version in color.)

To verify the accuracy of the deconvolution results, the experimental and theoretical values of the non-bridging oxygen number (NBO/T_exp and NBO/T_theor) were calculated using Eqs. (4) and (5), as shown in Fig. 9. NBO/T_exp in Fig. 9 is close to NBO/T_theor, which indicates that the molar fractions of structural units obtained by the present experiment have higher accuracy.   

$$\begin{array}{l}\mathrm{N}\mathrm{B}\mathrm{O}/{\mathrm{T}}_{exp}=\frac{x_{\mathrm{S}\mathrm{i}}}{x_{\mathrm{S}\mathrm{i}}+x_{\mathrm{P}}}\cdot \sum _{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{\hspace{0.17em} } {\mathrm{Q}}^n(\mathrm{S}\mathrm{i})}{X}_n\cdot \left(4-n\right)\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+\frac{x_{\mathrm{P}}}{x_{\mathrm{S}\mathrm{i}}+x_{\mathrm{P}}}\cdot \sum _{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{\hspace{0.17em} } {\mathrm{Q}}^n(\mathrm{P})}{X}_n\cdot \left(3-n\right)\end{array}$$(4)

  

$$\mathrm{N}\mathrm{B}\mathrm{O}/{\mathrm{T}}_{\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{o}\mathrm{r}}=\frac{2x_{\mathrm{O}}-4x_{\mathrm{S}\mathrm{i}}-5x_{\mathrm{P}}}{x_{\mathrm{S}\mathrm{i}}+x_{\mathrm{P}}}$$(5)

Where x represents the molar fraction of elements in final slag.

Fig. 9.

Experimental and theoretical non-bridging oxygen numbers of final slags. (Online version in color.)

As can be seen from Fig. 8, three silicon-oxygen tetrahedrons (Q⁰(Si), Q¹(Si), and Q²(Si)) and two phosphorus-oxygen tetrahedrons (Q⁰(P) and Q¹(P)) exist in final slag after dephosphorization. Due to less Fe₂O₃ content, the Fe³⁺-related structural units were not detected in the slag. With the increase in the CaO/SiO₂ ratio and FeO concentration, the molar fractions of simple silicon-oxygen tetrahedron and phosphorus-oxygen tetrahedron such as Q⁰(Si) and Q⁰(P) gradually increase, indicating gradual depolymerization of molten slag.

3.3. Redefinition of Phosphorus Distribution Ratio and Phosphate Capacity

The transfer process of phosphorus between molten slag and liquid iron can be described as follows: The P dissolved in liquid iron loses electrons at the interface and is oxidized to P⁵⁺ and diffuses into the molten slag; the O dissolved in liquid iron gains electrons at the interface and is reduced to O²⁻. Due to the strong electrostatic force between P⁵⁺ and O^2– ions, P⁵⁺ exists in the molten slag in the form of PO₄³⁻ unit. This process can be represented by Eq. (6). Based on the ion theory of dephosphorization, the phosphate capacity of molten slag (C_P) was defined.²⁷⁾ The formula is shown in Eq. (7).   

$$2[\mathrm{P}]+5[\mathrm{O}]+3({\mathrm{O}}^{2-})=2(\mathrm{P}{\mathrm{O}}_4^{3-})$$(6)

  

$$C_{\mathrm{P}}=K_{\mathrm{P}}\cdot \frac{a_{({\mathrm{O}}^{2-})}^3}{f_{(\mathrm{P}{\mathrm{O}}_4^{3-})}^2}=L_{\mathrm{P}}{}^2\cdot \frac{1}{f_{[\mathrm{P}]}^2\cdot a_{[\mathrm{O}]}^5}$$(7)

Where $a_{({\mathrm{O}}^{2-})}$ and a_[O] represent the activities of O²⁻ in molten slag and the dissolved O in liquid iron, respectively; f_[P] and $f_{(\mathrm{P}{\mathrm{O}}_4^{3-})}$ represent the activity coefficients of P dissolved in liquid iron and $\mathrm{P}{\mathrm{O}}_4^{3-}$ in molten slag, respectively; K_P represents the equilibrium constant of dephosphorization reaction (Eq. (6)).

However, PO₄³⁻ unit is, in fact, not the only form of P⁵⁺. According to previous structure analysis²³⁾ and the abovementioned Raman deconvolution results, P⁵⁺ in the CaO–SiO₂–FeO–MgO–P₂O₅ slag mainly exists as PO₄³⁻ (Q⁰(P)) and P₂O₇⁴⁻ (Q¹(P)) units. Hence, the ion equation and equilibrium constant (K_P^*) of the dephosphorization reaction at the interface of molten slag and liquid iron can be expressed as follows:   

$$4[\mathrm{P}]+10[\mathrm{O}]+5({\mathrm{O}}^{2-})=2(\mathrm{P}{\mathrm{O}}_4^{3-})+({\mathrm{P}}_2{\mathrm{O}}_7^{4-})$$(8)

  

$$K_{\mathrm{P}}{}^{*}=\frac{{\gamma }_{(\mathrm{P}{\mathrm{O}}_4^{3-})}^2\cdot w{(\mathrm{P}{\mathrm{O}}_4^{3-})}^2\cdot {\gamma }_{({\mathrm{P}}_2{\mathrm{O}}_7^{4-})}\cdot w({\mathrm{P}}_2{\mathrm{O}}_7^{4-})}{f_{[\mathrm{P}]}^4\cdot w{[\mathrm{P}]}^4\cdot a_{[\mathrm{O}]}^{10}\cdot a_{({\mathrm{O}}^{2-})}^5}$$(9)

Where ${\gamma }_{(\mathrm{P}{\mathrm{O}}_4^{3-})}$ and ${\gamma }_{({\mathrm{P}}_2{\mathrm{O}}_7^{4-})}$ represent the activity coefficients of $\mathrm{P}{\mathrm{O}}_4^{3-}$ and ${\mathrm{P}}_2{\mathrm{O}}_7^{4-}$ dissolved in molten slag, respectively.

According to Eqs. (8) and (9), the equilibrium distribution ratio of phosphorus (L_P^*) and phosphate capacity (C_P^*) are redefined as Eqs. (10) and (11), respectively. In Eqs. (10) and (11), the mass fractions of phosphorus-oxygen groups and free oxygen ions (FO) can be obtained from the analysis results of melt structure and the calculation formulae are shown in Eqs. (4), (12), and (13). The calculation results of L_P^* and C_P^* are shown in Fig. 10. The newly defined L_P^* shows the same change trend as the result of previous L_P mentioned in Section 3.1 with the increases of FeO content and CaO/SiO₂. The difference is that the newly defined L_P^* and C_P^* amplify the trend of change due to the high exponential form in the new formulae and show the sensitivity to the changes in structure and composition. The newly defined phosphorus distribution ratio realizes the analysis of the dephosphorization efficiency from the ionic point of view, which can lay a good foundation for the thermodynamic calculation of the ionic reaction of dephosphorization.   

$$L_{\mathrm{P}}{}^{*}=\frac{w{(\mathrm{P}{\mathrm{O}}_4^{3-})}^2\cdot w({\mathrm{P}}_2{\mathrm{O}}_7^{4-})}{w{[\mathrm{P}]}^4}$$(10)

  

$$C_{\mathrm{P}}{}^{*}=K_{\mathrm{P}}{}^{*}\cdot \frac{a_{({\mathrm{O}}^{2-})}^5}{{\gamma }_{(\mathrm{P}{\mathrm{O}}_4^{3-})}^2\cdot {\gamma }_{({\mathrm{P}}_2{\mathrm{O}}_7^{4-})}}=L_{\mathrm{P}}^{*}\cdot \frac{1}{f_{[\mathrm{P}]}^4\cdot a_{[\mathrm{O}]}^{10}}$$(11)

  

$$x_{{\mathrm{O}}^{2-}}=x_{\mathrm{F}\mathrm{O}}=x_{\mathrm{O}}-x_{\mathrm{B}\mathrm{O}}-x_{\mathrm{N}\mathrm{B}\mathrm{O}}$$(12)

  

$$x_{\mathrm{B}\mathrm{O}}=\frac{1}{2}\cdot \frac{x_{\mathrm{S}\mathrm{i}}}{x_{\mathrm{S}\mathrm{i}}+x_{\mathrm{P}}}\cdot \sum _{\mathrm{f}\mathrm{o}\mathrm{r}{\mathrm{Q}}^n(\mathrm{S}\mathrm{i})}{X}_n\cdot n+\frac{1}{2}\frac{x_{\mathrm{P}}}{x_{\mathrm{S}\mathrm{i}}+x_{\mathrm{P}}}\cdot \sum _{\mathrm{f}\mathrm{o}\mathrm{r}{\mathrm{Q}}^n(\mathrm{P})}{X}_n\cdot n$$(13)

Where BO represents the bridging oxygen in the [SiO₄] and [PO₄]-tetrahedrons.

Fig. 10.

Phosphate capacity and the equilibrium distribution ratio of phosphorus under the condition of initial slags: (a) different Fe_xO content; (b) different CaO/SiO₂ ratios. (Online version in color.)

Figure 11 shows the comparisons of lgL_P^* defined based on the ion theory with lgL_P calculated by the molecular theory and the empirical formula.^8,20,22) The values of lgL_P^* in the present study are positively correlated with those obtained by other methods, but the calculated value is slightly larger due to the difference in the definition formula.

Fig. 11.

Comparison of phosphorus distribution ratio. (Online version in color.)

3.4. Relationship of the Equilibrium Distribution Ratio of Phosphorus and Melt Structure

Figure 12 shows the relationships of the equilibrium distribution ratio of phosphorus, the molar ratios of structural units, NBO/T and the moles fraction of FO under the condition of different FeO content and CaO/SiO₂. In figure, lgL_P^* is positively correlated with the molar ratio of Q⁰(P)/Q¹(P), indicating that the high concentration of Q⁰(P) is beneficial to dephosphorization. In other words, phosphorus is more stable in the form of Q⁰(P) in the molten slag after being removed from the liquid iron. To prove this point, we try to explain in conjunction with the molecular theory of dephosphorization. The dephosphorization reaction occurs at the interface of molten slag and liquid steel and is generally considered to produce C₃P. The ionic formula of Q⁰(P) is PO₄³⁻, and three P–O non-bridging oxygen bonds in the ionic groups are connected to 3/2 Ca²⁺ to form 3/2Ca∙PO₄, as shown in Fig. 13. When this ionic group is enlarged by 2 times, it is actually C₃P. The conclusion further illustrates that the stable phosphate ion existing in the molten slag is Q⁰(P). In fact, Q¹(P) is also a group formed after fixing P⁵⁺ ion. In Eq. (10), L_P^* is proportional to the square of the concentration of Q⁰(P) and the first power of the concentration of Q¹(P). Through the comparison the powers of the concentration of structural units, it can be found that the effect of Q⁰(P) content on L_P^* is more significant than that of Q¹(P), which also confirms the above results.

Fig. 12.

Comparisons of the equilibrium distribution ratio of phosphorus, the molar ratios of structural units, NBO/T and the molar fraction of FO. (Online version in color.)

Fig. 13.

Schematic diagram of the relationship between Q⁰(P) and 3CaO·P₂O₅. (Online version in color.)

It is noted that lgL_P^* is also in proportion to Q⁰(Si)/(Q¹(Si)+Q²(Si)), NBO/T, and FO with increasing FeO content and CaO/SiO₂ when FeO content is less than 21.9 mass%. According to the above discussion, it is not difficult to understand the reason why lgL_P^* increases with the increase of NBO/T and FO. An increase in NBO/T and FO means that the polymerization degree of slag is lower and more O²⁻ ions exist in the network structure, which is conducive to the formation of Q⁰(P), so lgL_P^* increases. Meanwhile, it can be seen from Eq. (6) that when the concentration of FO in the molten slag increases, the activity of O²⁻ increases, which promotes the dephosphorization reaction and is beneficial to the improvement of the phosphorus distribution ratio.

The increase of NBO/T also indicates the increase of simple Q⁰(Si) units. The observations on the mineralogical phase of the dephosphorization slag suggested that C₂S precipitation accompanied with phosphorous enrichment in the slag was key to the increase of L_P^*.³²⁾ In the melt structure, C₂S is actually the combination of Q⁰(Si) (SiO₄⁴⁻) and two charge compensators (Ca²⁺), as shown in Fig. 14. Hence, the increase of Q⁰(Si) also plays a role in fixing Q⁰(P) unit. However, the inverse proportion of NBO/T, Q⁰(Si)/(Q¹(Si)+Q²(Si)), FO and lgL_P^* when FeO content exceeds 21.9% is speculated to be related to the cations in the molten slag, which will be explained in the following discussion.

Fig. 14.

Schematic diagram of the relationship between Q⁰(Si) and 2CaO∙SiO₂. (Online version in color.)

In Fig. 12, lgL_P^* gradually increases with increasing CaO/SiO₂ from 0.15 to 1.25. The increase of CaO introduces more free O²⁻ ions that can fix P⁵⁺, resulting in the formation of more Q⁰(P) groups in the slag. Meanwhile, the introduced Ca²⁺ compensates the charge of anionic group, so that Q⁰(P) can stably exist in the network structure, which is the dephosphorization product C₃P considered by molecular theory. The process of P⁵⁺ being fixed by CaO can be represented in Fig. 15(a).

Fig. 15.

Schematic diagram of the analysis of dephosphorization based on melt structure: (a) the effect of CaO on dephosphorization; (b) the effect of FeO on dephosphorization. (Online version in color.)

A special phenomenon, different from the increasing CaO/SiO₂, is found that lgL_P^* increases initially, followed by a decrease with the gradual increase of FeO content. By comparing the relationship between lgL_P^* and the molar ratio of cations (as shown in Fig. 16), it was found that this phenomenon is considered to be closely related to the molar ratio of Ca²⁺ and Mg²⁺ to Fe²⁺. In Fig. 16, the molar ratio of (Ca²⁺+Mg²⁺)/Fe²⁺ has a decreasing trend with the increase of FeO content and CaO/SiO₂. When (Ca²⁺+Mg²⁺)/Fe²⁺ is lower than 2.905, lgL_P^* shows a sudden drop. It is known that both Ca²⁺, Mg²⁺, and Fe²⁺ act as network modifiers to compensate for the charge of anionic groups. However, due to the stronger polarization of Fe²⁺ than Ca²⁺ and Mg²⁺,^33,34) Fe²⁺ can cause an obvious deformation of anionic group, resulting in the instability of Q⁰(P) and Q¹(P) groups. The excessive increase of Fe²⁺ dilutes the molar concentrations of Ca²⁺ and Mg²⁺, which decompose Q⁰(P) and Q¹(P). The P⁵⁺ ions in the decomposed Q⁰(P) and Q¹(P) groups diffuse to the slag-metal interface and are reduced to the dissolved phosphorus in the liquid iron. For the increase of FeO content, on the one hand, the introduced O²⁻ can combine with P⁵⁺ to form Q⁰(P), and on the other hand, the introduced Fe²⁺ cannot stabilize the Q⁰(P) and Q¹(P) units. In the present study, hence, when FeO content is less than 21.9 mass%, Ca²⁺ and Mg²⁺ plays a major role in fixing P⁵⁺ and causes an increase in lgL_P^*; when FeO content is greater than 21.9 mass%, the increasing Fe²⁺ weakens the effect of Ca²⁺ and Mg²⁺, resulting in a decrease in lgL_P^*. The Steps 2 and 3-1 in Fig. 15(b) demonstrate the fixation process of P⁵⁺ and the Step 3-2 shows the decomposition process of Q⁰(P) and Q¹(P) by FeO.

Fig. 16.

Comparisons of the equilibrium distribution ratio of phosphorus and molar ratio of cations. (Online version in color.)

3.5. Analysis of the Thermodynamic Conditions of Dephosphorization

It can be known from the thermodynamic equilibrium condition that the oxidative atmosphere and high basicity (CaO/SiO₂) are favorable for dephosphorization. It can be understood from the point of view of ion theory that in the case of high oxygen potential, the more oxygen can be dissolved into the liquid iron, and continuously diffuse to the slag-metal interface. The process can be depicted in the Step 1 of Fig. 15(b). At the interface, the reduction reaction of oxygen leads to the corresponding increase of O²⁻ in the slag, so Q⁰(P) increases accordingly, L_P^* gradually increases, and the dephosphorization rate is improved significantly. However, at a higher oxygen potential, the excessive iron is oxidized into the molten slag, resulting in a sharp increase in Fe²⁺ in the slag. Fe²⁺ has a strong polarization force, which tends to be around Q⁰(P) and Q¹(P), and thereby Q⁰(P) and Q¹(P) are polarized and deformed. The polarized Q⁰(P) and Q¹(P) units are difficult to exist stably, dissociate into P⁵⁺ and enter the liquid iron after being reduced.

When the basicity of molten slag is higher, both O²⁻ and Ca²⁺ generated by the decomposition of CaO increase, which play the major role in binding P⁵⁺ ion and stabilizing Q⁰(P) and Q¹(P) units, respectively, resulting in an increase in lgL_P^* and a higher dephosphorization rate. However, excessive CaO content can make the slag viscous, which is not conducive to dephosphorization from a kinetic point of view. Meanwhile, considering the dual roles of FeO on dephosphorization, the effect of Fe²⁺ on the deformation of phosphorus-oxygen group can be eliminated by replacing part of Fe²⁺ with Ca²⁺.Therefore, the CaO/FeO ratio in the slag should be at an appropriate value to obtain a higher phosphorus distribution ratio. According to the above results, metallurgists can appropriately control the basicity and FeO content of steelmaking slag to improve the dephosphorization efficiency, reduce the phosphorus content in the steel, and further improve the quality of the steel.

4. Conclusions

In the present study, the effect of FeO and CaO/SiO₂ on the equilibrium distribution ratio of phosphorus was studied through the dephosphorization equilibrium experiment of molten slag and liquid iron at 1600°C. Then, the melt structure of final slag was determined by Raman spectroscopy. According to the analysis results of Raman spectra, it was found that there were only two forms of phosphorus ions in the slag, Q⁰(P) and Q¹(P). Hence, the ionic equation of dephosphorization reaction was established, the formulas for phosphate capacity and the equilibrium distribution ratio of phosphorus were redefined and the thermodynamic conditions for enhanced dephosphorization were analyzed in detail. The results showed that the presence of P⁵⁺ in the slag in the form of Q⁰(P) could significantly improve L_P^*, indicating that Q⁰(P) was more stable than Q¹(P). With increasing CaO/SiO₂ from 0.15 to 1.25, the increasing free O²⁻ ions could combine with P⁵⁺ to form a stable Q⁰(P) unit, which was fixed by Ca²⁺ and Mg²⁺; hence, the dephosphorization rate, equilibrium distribution ratio of phosphorus and phosphate capacity both showed an increasing trend. As the molar concentration of FeO gradually increased, the O²⁻ ion dissociated by FeO also played the role in forming Q⁰(P). However, when the FeO content increases to 29.2 mass%, the excessive Fe²⁺ ion decreased the proportion of Ca²⁺ and Mg²⁺ ion, thereby reducing the stability of Q⁰(P) and Q¹(P), so that P⁵⁺ at the interface was reduced to the dissolved phosphorus in the liquid iron.

Acknowledgment

The authors sincerely thank the National Natural Science Foundation of China (Grant Nos. 51974075), the Open Funds of State Key Laboratory of Metal Material for Marine Equipment and Application (Grant No. SKLMEA-K202001).

References

• 1)   Z. H.  Tian,  B. H.  Li,  X. M.  Zhang and  Z. H.  Jiang: J. Iron Steel Res. Int., 16 (2009), 6. https://doi.org/10.1016/S1006-706X(09)60036-4
• 2)   H.  Li,  S.  Barui,  S.  Mukherjee and  K.  Chattopadhyay: Metals, 12 (2022), 268. https://doi.org/10.3390/met12020268
• 3)   T. A.  Bloom,  D. R.  Fosnacht and  D. M.  Haezebrouck: Iron Steelmaker, 17 (1990), 35.
• 4)   K.  Ito,  M.  Yanagisawa and  N.  Sano: Tetsu-to-Hagané, 68 (1982), 342 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.68.2_342
• 5)   O.  Kovtun,  M.  Karbayev,  I.  Korobeinikov,  C.  Srishilan,  A. K.  Shukla and  O.  Volkova: Steel Res. Int., 92 (2021), 2000607. https://doi.org/10.1002/srin.202000607
• 6)   A. T.  Morales and  R. J.  Fruehan: Metall. Mater. Trans. B, 28 (1997), 1111. https://doi.org/10.1007/s11663-997-0067-6
• 7)   C. M.  Lee and  R. J.  Fruehan: Ironmaking Steelmaking, 32 (2005), 503. https://doi.org/10.1179/174328105X48142
• 8)   S.  Basu,  A. K.  Lahiri and  S.  Seetharaman: Metall. Mater. Trans. B, 38 (2007), 357. https://doi.org/10.1007/s11663-007-9057-y
• 9)   G.  Li,  T.  Hamano and  F.  Tsukihashi: ISIJ Int., 45 (2005), 12. https://doi.org/10.2355/isijinternational.45.12
• 10)   F.  Li,  X.  Li,  S.  Yang and  Y.  Zhang: Metall. Mater. Trans. B, 48 (2017), 2367. https://doi.org/10.1007/s11663-017-1023-8
• 11)   J.  Im,  K.  Morita and  N.  Sano: ISIJ Int., 36 (1996), 517. https://doi.org/10.2355/isijinternational.36.517
• 12)   S.  Basu,  A. K.  Lahiri and  S.  Seetharaman: Metall. Mater. Trans. B, 38 (2007), 623. https://doi.org/10.1007/s11663-007-9063-0
• 13)   H.  Suito,  Y.  Hayashida and  Y.  Takahashi: Tetsu-to-Hagané, 63 (1977), 1252 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.63.8_1252
• 14)   H.  Suito and  R.  Inoue: Trans. Iron Steel Inst. Jpn., 24 (1984), 40. https://doi.org/10.2355/isijinternational1966.24.40
• 15)   H.  Yu,  X.  Lu,  T.  Miki,  K.  Matsubae,  Y.  Sasaki and  T.  Nagasaka: Resour. Conserv. Recycl., 180 (2022), 106203. https://doi.org/10.1016/j.resconrec.2022.106203
• 16)   C. M.  Yoon and  D. J.  Min: Ceram. Int., 48 (2022), 15017. https://doi.org/10.1016/j.ceramint.2022.02.030
• 17)   R.  Saito,  H.  Matsuura,  K.  Nakase,  X.  Yang and  F.  Tsukihashi: Tetsu-to-Hagané, 95 (2009), 258 (in Japanese). https://doi.org/10.2355/tetsutohagane.95.258
• 18)   W.  Yang,  J.  Yang,  Y.  Shi,  Z.  Yang,  F.  Gao,  R.  Zhang and  G.  Ye: Steel Res. Int., 92 (2021), 2000438. https://doi.org/10.1002/srin.202000438
• 19)   X.  Yang,  H.  Matsuura and  F.  Tsukihashi: ISIJ Int., 49 (2009), 1298. https://doi.org/10.2355/isijinternational.49.1298
• 20)   H.  Suito and  R.  Inoue: Tetsu-to-Hagané, 70 (1984), 366 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.70.3_366
• 21)   W.  Fix,  H.  Heymann and  R.  Heinke: J. Am. Ceram. Soc., 52 (1969), 346. https://doi.org/10.1111/j.1151-2916.1969.tb11948.x
• 22)   W.  Yang,  J.  Yang,  Y.  Shi,  Z.  Yang,  F.  Gao,  R.  Zhang and  G.  Ye: Ironmaking Steelmaking, 48 (2021), 69. https://doi.org/10.1080/03019233.2020.1731256
• 23)   C.  Liu,  R.  Zhang,  X.  Zhao,  J.  Jia and  Y.  Min: J. Non-Cryst. Solids, 557 (2021), 120579. https://doi.org/10.1016/j.jnoncrysol.2020.120579
• 24)   R. K.  Brow: J. Non-Cryst. Solids, 263–264 (2000), 1. https://doi.org/10.1016/S0022-3093(99)00620-1
• 25)   R.  Zhang,  Y.  Wang,  X.  Zhao,  J.  Jia,  C.  Liu and  Y.  Min: Metall. Mater. Trans. B, 51 (2020), 2021. https://doi.org/10.1007/s11663-020-01888-8
• 26)   P.  McMillan: Am. Mineral., 69 (1984), 622.
• 27)   M. Y.  Zhu: Modern Metallurgical Technology, Metallurgical Industry Press, Beijing, (2011), 207.
• 28)   W. L.  Dong,  H. W.  Pan,  L.  Luo,  C. X.  Ji,  H. B.  Li and  Z. H.  Tian: Iron Steel, 55 (2020), 75. https://doi.org/10.13228/j.boyuan.issn0449-749x.20190107
• 29)   B. O.  Mysen,  D.  Virgo,  W. J.  Harrison and  C. M.  Scarfe: Am. Mineral., 65 (1980), 900.
• 30)   B. O.  Mysen,  D.  Virgo and  F. A.  Seifert: Rev. Geophys., 20 (1982), 353.
• 31)   R.  Zhang,  Y.  Min,  Y.  Wang,  X.  Zhao and  C.  Liu: ISIJ Int., 60 (2020), 212. https://doi.org/10.2355/isijinternational.ISIJINT-2019-413
• 32)   F.  Liu,  G.  Wang,  Y.  Zhao,  J.  Tan,  C.  Zhao and  Q.  Wang: Ironmaking Steelmaking, 46 (2019), 392. https://doi.org/10.1080/03019233.2017.1403134
• 33)   J.  Ma,  J.  Zhang,  Z.  Wang and  X.  Xing: 7th Int. Symp. on High-Temperature Metallurgical Processing, Hoboken, NJ, (2016), 699. https://doi.org/10.1002/9781119274643.ch86
• 34)   F. M.  Shen: Physical Chemistry of Metallurgy, Higher Education Press, Beijing, (2017), 160.