Chlorination of ZnFe₂O₄ in Molten MgCl₂–KCl

Abstract

Recovery of zinc from electric arc furnace dust (EAF dust) has been an important issue for the steelmaking industry, yet a sustainable technology is absent. As a fundamental study to develop a new process of recovering metals from EAF dust by using molten salt, this work clarifies the reaction behavior of solid ZnFe₂O₄ in molten MgCl₂–KCl at temperatures from 773 K to 973 K. MgCl₂ is the chlorinating agent and KCl is an additive to make a low melting point molten salt. The experimental results indicate the efficacy of converting ZnFe₂O₄ to ZnCl₂ and FeCl_x (x = 2 or 3) by MgCl₂. Zn is chlorinated prior to Fe in ZnFe₂O₄ under all conditions, implying the possibility of separating Zn from Fe. Improving the mass transfer in the melt accelerates the reaction. Lower temperatures and larger O₂ partial pressure favor the selective chlorination of Zn, yet the reaction is more stagnant. This work has thus demonstrated the feasibility of treating EAF dust by using MgCl₂-based molten salt.

1. Introduction

Steel production in electrical arc furnace (EAF) has increased globally. A large amount of zinc-bearing dust is generated as a byproduct when using galvanized steel scrap as the major feedstock. Recovery of Zn from EAF dust has been an important issue for the sustainable development of the steelmaking industry.¹⁾

The major components in EAF dust are ZnFe₂O₄, ZnO and Fe₂O₃. Recovery of Zn from EAF dust is not easy, mostly because of the high chemical stability of ZnFe₂O₄. The Waelz process is the dominant technology to recover Zn from EAF dust worldwide. In this technology, Zn-bearing oxides in the dust are reduced by coke at around 1200°C, followed by re-oxidation of Zn vapors. Crude ZnO powder rather than Zn metal is the final product. As environment and energy concerns are becoming more important, the technical defects of the Waelz process such as the low efficiency, high energy consumption and intense CO₂ emission have gradually become clear.^2,3) Recently, Nagasaka et al. proposed a “CaO addition process” which has the potential to replace the Waelz process.^4,5,6) In this technology, CaO is reacted with EAF dust to convert ZnFe₂O₄ into ZnO, and metallic Zn can be extracted from ZnO by hydrometallurgical approaches. Being independent of carbon is the most significant advantage of this technology, yet high reaction temperatures above 1000°C may still lead to a considerable energy consumption.

Chlorination by chlorine gas is one of the promising approaches to recover Zn from EAF dust.⁷⁾ The concept is converting Zn-bearing oxides into ZnCl₂, followed by electrolysis to obtain high-purity metallic Zn. Carbon-based reducing agents are not necessary in the chlorination process and thus a clean production is expected. Matsuura^8,9,10) and Sun¹¹⁾ had proved that Zn in EAF dust could be preferentially chlorinated by controlling the thermodynamic conditions and in this manner separated from Fe. However, because Cl₂ is highly corrosive and toxic, corrosion of apparatus and safety issues remain a challenge. In order to make the process easier and safer in operation, it is helpful to find a Cl₂-free agent to perform the role of chlorination.

Metal chlorides could take the position of Cl₂ as the chlorinating agent and magnesium chloride (MgCl₂) might be the most promising one.¹²⁾ Features of MgCl₂-based chlorinating reactions and applications have been reported in literature. Cooper et al. presented several examples of converting metal oxides into their chlorides by reacting with molten MgCl₂ at elevated temperatures.¹³⁾ Kang et al. reacted titanium ore with molten MgCl₂ to remove iron oxides by selective chlorination.¹⁴⁾ Therefore, it is considered that MgCl₂ can be used to treat EAF dust on the purpose of metal recovery.

The capability of MgCl₂ to chlorinate Zn in EAF dust can be explained by thermodynamic analysis. When reacting EAF dust with MgCl₂, the following reactions may occur.   

$$\begin{array}{l}\mathrm{Z}\mathrm{n}\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_4\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \\ \mathrm{Z}\mathrm{n}\mathrm{C}{\mathrm{l}}_2\left(l\right)+\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)\end{array}$$(1)

  

$$\begin{array}{l}1/3\mathrm{Z}\mathrm{n}\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_4\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \\ 1/3\mathrm{Z}\mathrm{n}\mathrm{O}\left(s\right)+2/3\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_3\left(g\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)\end{array}$$(2)

  

$$\begin{array}{l}1/4\mathrm{Z}\mathrm{n}\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_4\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \\ 1/4\mathrm{Z}\mathrm{n}\mathrm{C}{\mathrm{l}}_2\left(l\right)+1/2\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_3\left(g\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)\end{array}$$(3)

  

$$\mathrm{Z}\mathrm{n}\mathrm{O}\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \mathrm{Z}\mathrm{n}\mathrm{C}{\mathrm{l}}_2\left(l\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)$$(4)

  

$$\begin{array}{l}1/3\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \\ 2/3\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_3\left(g\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)\end{array}$$(5)

  

$$\begin{array}{l}1/3\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(s\right)+\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(s\right)\rightleftarrows \\ 2/3\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_2\left(g\right)+1/3\mathrm{C}{\mathrm{l}}_2\left(g\right)+\mathrm{M}\mathrm{g}\mathrm{O}\left(s\right)\end{array}$$(6)

By using the thermochemical data,¹⁵⁾ the standard Gibbs energy (Δ_rG°) of these reactions can be calculated. As shown in Fig. 1, only reactions (1) and (4) are spontaneous (Δ_rG°<0) at standard conditions, which indicates the possibility of separating Zn from Fe in ZnFe₂O₄ by reacting with MgCl₂. Moreover, the decrease of Gibbs energy of chlorination of Fe [reaction (2)] is larger than that of Zn [reaction (1)] with increasing temperature, indicating the beneficial effect of a low temperature on the selective chlorination of Zn.

Fig. 1.

Variation of standard Gibbs energy of reactions at different temperatures. (Online version in color.)

If using MgCl₂-based molten salt to treat EAF dust, the formed ZnCl₂ would dissolve into the molten salt. According to those properties, the authors have thus proposed a “molten salt process” for Zn recovery from EAF dust by integrating chlorination and electrolysis into the same reactor. Figure 2 shows the concept. EAF dust is added into the bath of MgCl₂-based molten salt. Zn²⁺ would transfer from the solid precursor into the molten salt phase due to selective chlorination and dissolution. Electrolyzing the melt generates liquid Zn on the cathode at the bottom. Solid residue expected to be a mixture of MgO and Fe₂O₃ can be recycled as the raw material in steelmaking plants. The anodic reaction would be evolution of Cl₂, which can also be recycled and reused in chemical plants. Continuous operation of dust addition, product recovery and solid residues removal would enable a high productivity. The molten salt process can simplify the procedure and lower the temperature compared with other technologies.

Fig. 2.

The concept for recovering Zn from EAF dust by selective chlorination and electrolysis in MgCl₂-based molten salt. (Online version in color.)

Aiming at developing such a technology to recover Zn from EAF dust, this work investigates the reaction behavior of solid ZnFe₂O₄ in molten MgCl₂–KCl. KCl with little chlorinating effect was used as an additive to make a low melting point molten salt. The reaction mechanism was clarified by analyzing the products. Feasibility of separating Zn from Fe by selective chlorination in the molten salt was discussed.

2. Experimental

2.1. Materials

Reagent grade Fe₂O₃ (99.9%, Kojundo Chemical Lab. Co. Ltd.), ZnO (99.9%, Kojundo Chemical Lab. Co. Ltd.), MgCl₂ and KCl (99%, Fujifilm Wako Pure Chemical Corp.) were used in this work. Equimolar mixture of powdery Fe₂O₃ and ZnO were ground in an agate mortar, and sintered at 1573 K in air for 3 h to obtain ZnFe₂O₄. The as-heated sample was again ground in an agate mortar to make the powder size smaller than 0.2 mm. Mixture of MgCl₂–38mass%KCl was the chlorinating agent. Prior to melting, the chlorides were dried under vacuum at 180°C for 72 h to remove moisture.

2.2. Procedure

A dense MgO crucible (99%, o.d. 40 mm, i.d. 30 mm, height 120 mm, Nikkato Corp.) containing 50.0 g of MgCl₂–38mass%KCl was set inside a mullite tube (o.d. 135 mm, i.d. 127 mm, height 400 mm) in the furnace. The experimental temperatures were 773 K, 873 K and 973 K. After melting the chlorides, ZnFe₂O₄ was added in different manners to investigate the reaction under static or dynamic conditions, as shown in Fig. 3. In one case [Fig. 3(a)], reaction under static conditions was studied. A bottom-sealed quartz tube (o.d. 8 mm, i.d. 6 mm) containing 0.5 g of ZnFe₂O₄ was immersed into the molten bath in Ar atmosphere. The melt would flow into the tube through a hole (approximately 30 mm²) on the wall. The melt was sampled by using an Al₂O₃ rod (dia. 6 mm) at a specified time interval. In another case [Fig. 3(b)], reaction with Ar or air bubbling was studied. After adding 0.5 g of ZnFe₂O₄ into the bath from the top, Ar or air was flowed (150 ml min⁻¹) through a mullite tube (o.d. 6 mm, i.d. 4 mm) into the melt to stir the molten bath containing suspended ZnFe₂O₄ powders. The distance from the tip of the mullite tube to the crucible bottom was approximately 5 mm. The melt was sampled and analyzed after reaction for a certain time.

Fig. 3.

Experimental apparatus: (a) reaction under static condition (Ar atmosphere); (b) reaction with Ar or air bubbling. (Online version in color.)

2.3. Analysis and Characterization

Concentrations of Zn and Fe in the sampled melt were measured by an inductively coupled plasma optical emission spectrometer (ICP-OES). The MgO crucibles after all sampling were rinsed with distilled water to recover the solid residue, which was characterized by X-ray diffraction (XRD, Smart Lab, Rigaku Corp.) and observed by scanning electron microscope (SEM, VE-8800, Keyence Corp.). The composition was determined by energy-dispersive X-ray spectroscopy (EDX, DAX Genesis APEX2, AMETEK Co., Ltd.).

3. Results and Discussion

3.1. Reaction under Static Conditions

Since the solid ZnFe₂O₄ powders were immersed in molten MgCl₂–KCl of high excess, it is considered that newly formed chlorides (ZnCl₂ or FeCl_x) would dissolve immediately into the melt. Table 1 lists the mass concentration of Zn and Fe in the sampled melt during the reaction under static conditions. At all temperatures, Zn and Fe contents in the melt increase with reaction time, which indicates the proceeding of chlorination and formation of soluble ions. Chlorination of Zn agrees well with the estimation by thermodynamic analysis. Although chlorination of Fe is thermodynamically nonspontaneous under standard conditions, deviation from standard state of FeCl₂ or FeCl₃ (activity less than unity) in the melt makes the reaction happen.

Table 1. Mass concentration of Zn and Fe in the melt during the reaction between ZnFe₂O₄ and molten MgCl₂–KCl in Ar atmosphere under static conditions.

Exp. No.	Temp./K	Sampling time/h	Mass concentration/mass%
			Zn	Fe
SA-01	773	2	0.0007	0.0003
SA-02		18	0.0017	0.0004
SA-03		22	0.0023	0.0003
SA-04		26	0.0031	0.0002
SA-05		42	0.0039	0.0003
SA-06		46	0.0056	0.0006
SA-07		50	0.0048	0.0010
SA-08		66	0.0063	0.0007
SA-09		70	0.0069	0.0014
SA-10		74	0.0082	0.0013
SA-11		90	0.0113	0.0020
SB-01	873	3	0.0039	0.0050
SB-02		18	0.0109	0.0053
SB-03		23	0.0145	0.0069
SB-04		26	0.0181	0.0080
SB-05		42	0.0246	0.0088
SB-06		46	0.0274	0.0094
SB-07		50	0.0208	0.0083
SB-08		66	0.0325	0.0117
SB-09		70	0.0320	0.0159
SB-10		74	0.0336	0.0177
SB-11		90	0.0388	0.0231
SB-12		98	0.0400	0.0242
SC-01	973	2	0.0044	0.0074
SC-02		18	0.0155	0.0098
SC-03		22	0.0307	0.0113
SC-04		26	0.0317	0.0100
SC-05		42	0.0421	0.0147
SC-06		46	0.0457	0.0214
SC-07		50	0.0574	0.0297
SC-08		66	0.0590	0.0462
SC-09		70	0.0643	0.0570
SC-10		74	0.0664	0.0709
SC-11		90	0.0719	0.0860
SC-12		98	0.0778	0.1030

* Maximum content of Zn and Fe in the system: 0.2683 mass% and 0.4584 mass%, given all ZnFe₂O₄ is chlorinated.

Since this work utilized excessive MgCl₂ (>30.0 g) and the amount of each sampled melt is not large (<0.2 g), the total weight of molten salt is assumed to be constant (W_melt = 50.0 g). Neither the evaporation of molten salt is considered. Accordingly, chlorination fractions of Zn and Fe (η) can be calculated from their mass concentrations (m) in the sampled melt by Eqs. (7) and (8).   

$$\begin{array}{l}{\eta }_{\text{Zn}}={\displaystyle \frac{\text{Weight}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{Zn}\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{the}\mathrm{\hspace{0.17em} }\text{melt}}{\text{Total}\mathrm{\hspace{0.17em} }\text{weight}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{Zn}}}\\ \hspace{1em}\hspace{0.5em}={\displaystyle \frac{W_{\text{melt}}\times m_{\text{Zn}}}{W_{{\text{ZnFe}}_2{\text{O}}_4}\times {\displaystyle \frac{M_{\text{Zn}}}{M_{{\text{ZnFe}}_2{\text{O}}_4}}}}}\times 100\%\end{array}$$(7)

  

$$\begin{array}{l}{\eta }_{\text{Fe}}={\displaystyle \frac{\text{Weight}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{Fe}\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{the}\mathrm{\hspace{0.17em} }\text{melt}}{\text{Total}\mathrm{\hspace{0.17em} }\text{weight}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{Fe}}}\\ \hspace{1em}\hspace{0.5em}={\displaystyle \frac{W_{\text{melt}}\times m_{\text{Fe}}}{W_{{\text{ZnFe}}_2{\text{O}}_4}\times {\displaystyle \frac{M_{\text{Fe}}}{M_{{\text{ZnFe}}_2{\text{O}}_4}}}}}\times 100\%\end{array}$$(8)

where W_ZnFe2O4 is the weight of the initial ZnFe₂O₄ (W_ZnFe2O4 = 0.5 g) and M_i is the molar weight of the substance i. As shown in Fig. 4, chlorination fractions of Zn and Fe increase with reaction time, yet the increasing patterns are different. In the case of Zn, the curves are steep at the start and gradually flatten out along with reaction time, which indicates a decline in the reaction rate. It is considered that ZnCl₂ is mostly formed via reaction (1). Accumulation of ZnCl₂ in the melt near the interface (e.g. crevices in ZnFe₂O₄ particles) is the reason for the decrease in chlorination rate of Zn. On the contrary, chlorination of Fe proceeds very slowly at the initial stage. Although chlorination of Fe is not avoidable, chlorination of Zn with a larger Gibbs energy is preferred in the system. Since the total amount of MgCl₂ at the interface is limited, decline in chlorination of Zn at the later stage spares more MgCl₂ available to chlorinate Fe. Therefore, chlorination of Fe becomes faster.

Fig. 4.

Change of chlorination fraction of Zn and Fe in ZnFe₂O₄ with time by reacting with molten MgCl₂–KCl in Ar atmosphere under static conditions. The inset shows the case of Fe at 773 K.

Figure 5 shows the change of molar ratio between dissolved Zn and Fe (x_Zn/x_Fe, x_i means the molar fraction of component i) in the melt with reaction time. The value of x_Zn/x_Fe is 0.5 when Zn and Fe in ZnFe₂O₄ are equally chlorinated [proceeding of reaction (3)]. Most of the data in Fig. 5 locate in the region of x_Zn/x_Fe > 0.5, indicating that chlorination of Zn is dominant over that of Fe. Roughly the value of x_Zn/x_Fe shows an increasing-decreasing trend with time in all temperatures and decreases with rising temperature. According to the thermodynamic analysis in Fig. 1, lower temperatures favor the selective chlorination of Zn from ZnFe₂O₄, although the reaction will be slower.

Fig. 5.

Change of molar ratio between dissolved Zn and Fe (x_Zn/x_Fe) in the melt by reacting ZnFe₂O₄ with molten MgCl₂–KCl in Ar atmosphere under static conditions. (Online version in color.)

3.2. Reaction with Gas Bubbling

Previous results have verified the capability of MgCl₂ to chlorinate ZnFe₂O₄, yet the reaction is extremely slow compared with the reactions of using Cl₂ as the chlorinating agent.⁹⁾ It is considered that improving the mass transfer in the melt may accelerate the reaction. Therefore, another series of experiments with gas bubbling were also conducted. Table 2 lists the concentration of Zn and Fe in the melt after reaction between ZnFe₂O₄ and molten MgCl₂–KCl with Ar or air bubbling at 773 K. Data points after reaction for 30 h with Ar bubbling were obtained in three separated experiments under the same conditions. A large variation may due to experimental faults.

Table 2. Concentration of Zn and Fe in the melt after reaction between ZnFe₂O₄ and molten MgCl₂–KCl with Ar or air bubbling (150 ml min⁻¹) at 773 K.

Exp. No.	Bubbling gas type	Reaction time/h	Concentration/mass%
			Zn	Fe
MAr-01	Ar	5	0.0550	0.0059
MAr-02		10	0.1000	0.0260
MAr-03		30	0.2100	0.0430
MAr-04		30	0.1000	0.0200
MAr-05		30	0.1200	0.0310
MAr-06		50	0.1900	0.0730
MAir-01	Air	5	0.0250	0.0019
MAir-02		10	0.0380	0.0029
MAir-03		30	0.0840	0.0046

Chlorination fractions of Zn and Fe were calculated according to the measured data in Table 2, as shown in Fig. 6 compared with the result under static condition. The results indicate a dramatic improvement in reaction kinetics by introducing gas bubbling. Improved mass transfer due to gas bubbling and enlarged contact area between ZnFe₂O₄ powder and molten salt are the driving factors.

Fig. 6.

Change of chlorination fraction of Zn and Fe in ZnFe₂O₄ with time by reacting with molten MgCl₂–KCl with Ar or air bubbling (150 ml min⁻¹) at 773 K. (Online version in color.)

It can also be noticed in Fig. 6 that bubbling by Ar is more effective in accelerating the reaction compared with that of using air. It is considered that the large partial pressure of O₂ in air is an adverse condition for the fast chlorination of ZnFe₂O₄.

Figure 7 shows the change of molar ratio between dissolved Zn and Fe (x_Zn/x_Fe) in the melt with reaction time under gas bubbling conditions. Larger values are obtained in the case of air bubbling. This observation indicates that air bubbling with a larger O₂ partial pressure benefits the selective chlorination of Zn. This tendency agrees well with the finding by using Cl₂ as the chlorination agent.⁶⁾ In the case of Ar bubbling, the ratio (x_Zn/x_Fe) is smaller and a declining trend with time is shown.

Fig. 7.

Change of molar ratio between dissolved Zn and Fe (x_Zn/x_Fe) in the melt by reacting ZnFe₂O₄ with molten MgCl₂–KCl with Ar or air bubbling (150 ml min⁻¹) at 773 K. (Online version in color.)

The nature of science of chlorination by using chlorides is the same as that by using Cl₂ due to the reaction in the system   

$$\mathrm{M}\mathrm{g}\mathrm{C}{\mathrm{l}}_2\left(in\mathrm{\hspace{0.17em} }melt\right)+1/2{\mathrm{O}}_2(g)\to \mathrm{M}\mathrm{g}\mathrm{O}(s)+\mathrm{C}{\mathrm{l}}_2(g)$$(9)

Therefore, although Cl₂ is not directly used in this work, conversion of oxides into chlorides in the system can be written as:   

$$\mathrm{Z}\mathrm{n}\mathrm{O}\left(s\right)+\mathrm{C}{\mathrm{l}}_2\left(g\right)\to \mathrm{Z}\mathrm{n}\mathrm{C}{\mathrm{l}}_2\left(in\mathrm{\hspace{0.17em} }melt\right)+1/2{\mathrm{O}}_2\left(g\right)$$(10)

  

$$\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(s\right)+3\mathrm{C}{\mathrm{l}}_2\left(g\right)\to 2\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_3(in\mathrm{\hspace{0.17em} }melt)+3/2{\mathrm{O}}_2\left(g\right)$$(11)

  

$$\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(s\right)+2\mathrm{C}{\mathrm{l}}_2\left(g\right)\to 2\mathrm{F}\mathrm{e}\mathrm{C}{\mathrm{l}}_2(in\mathrm{\hspace{0.17em} }melt)+3/2{\mathrm{O}}_2\left(g\right)$$(12)

Here, ZnFe₂O₄ is deemed as a mixture of ZnO and Fe₂O₃. Direction and rate of reactions (10) to (12) are mainly determined by partial pressures of Cl₂ and O₂, which are determined by reaction (9) and external conditions (temperature and atmosphere). Dependence of chlorination rate on O₂ partial pressure can be explained by kinetic analysis. Except for reaction (9), reactions (10) to (12) are far from equilibrium state at the initial stage of chlorination. For these reactions, increase of O₂ partial pressure would inhibit the forward reaction. Therefore, chlorination with Air bubbling is not as fast as that of using Ar.

Dependence of preference in chlorination on O₂ partial pressure can be explained by thermodynamic analysis. Given that all the reactions are at equilibrium, the equilibrium constants (K) for reactions (9) to (12) are   

$$K_9=\frac{a_{\text{MgO}}{P}_{{\text{Cl}}_2}}{a_{{\text{MgCl}}_2}{P}_{{\text{O}}_2}^{1/2}}=\frac{P_{{\text{Cl}}_2}}{a_{{\text{MgCl}}_2}{P}_{{\text{O}}_2}^{1/2}}$$(13)

  

$$K_{10}=\frac{a_{{\text{ZnCl}}_2}{P}_{{\text{O}}_2}^{1/2}}{a_{\text{ZnO}}{P}_{{\text{Cl}}_2}}=\frac{\left({\gamma }_{{\text{ZnCl}}_2}{x}_{{\text{ZnCl}}_2}\right)P_{{\text{O}}_2}^{1/2}}{P_{{\text{Cl}}_2}}$$(14)

  

$$K_{11}=\frac{a_{{\text{FeCl}}_3}^2{P}_{{\text{O}}_2}^{3/2}}{a_{{\text{Fe}}_2{\text{O}}_3}{P}_{{\text{Cl}}_2}^3}=\frac{{({\gamma }_{{\text{FeCl}}_3}{x}_{{\text{FeCl}}_3})}^2{P}_{{\text{O}}_2}^{3/2}}{P_{{\text{Cl}}_2}^3}$$(15)

  

$$K_{12}=\frac{a_{{\text{FeCl}}_2}^2{P}_{{\text{O}}_2}^{3/2}}{a_{{\text{Fe}}_2{\text{O}}_3}{P}_{{\text{Cl}}_2}^2}=\frac{{({\gamma }_{{\text{FeCl}}_2}{x}_{{\text{FeCl}}_2})}^2{P}_{{\text{O}}_2}^{3/2}}{P_{{\text{Cl}}_2}^2}$$(16)

where a_i and γ_i are the activity and activity coefficient of component i relative to pure substance, respectively; P_j is partial pressure of gas component j. Activities of MgO and Fe₂O₃ are unity since MgO and Fe₂O₃ are assumed to exist as pure solid state. According to Eqs. (14), (15), (16), the molar ratio between Zn and Fe in the melt at equilibrium can be calculated by   

$${\displaystyle \frac{x_{\text{Zn}}}{x_{\text{Fe}}}}={\displaystyle \frac{x_{{\text{ZnCl}}_2}}{x_{{\text{FeCl}}_3}+x_{{\text{FeCl}}_2}}}={\displaystyle \frac{{\displaystyle \frac{K_{10}}{{\gamma }_{{\text{ZnCl}}_2}}}}{{\displaystyle \frac{K_{11}^{1/2}{P}_{{\text{Cl}}_2}^{1/2}}{{\gamma }_{{\text{FeCl}}_3}^{}{P}_{{\text{O}}_2}^{1/4}}}+{\displaystyle \frac{K_{12}^{1/2}}{{\gamma }_{{\text{FeCl}}_2}^{}{P}_{{\text{O}}_2}^{1/4}}}}}$$(17)

Combination of Eqs. (13) and (17) leads to   

$${\displaystyle \frac{x_{\text{Zn}}}{x_{\text{Fe}}}}={\displaystyle \frac{{\displaystyle \frac{K_{10}}{{\gamma }_{{\text{ZnCl}}_2}}}}{{\displaystyle \frac{K_9^{1/2}{K}_{11}^{1/2}{a}_{{\text{MgCl}}_2}^{1/2}}{{\gamma }_{{\text{FeCl}}_3}^{}}}+{\displaystyle \frac{K_{12}^{1/2}}{{\gamma }_{{\text{FeCl}}_2}^{}{P}_{{\text{O}}_2}^{1/4}}}}}$$(18)

Assuming that the activity of MgCl₂ and the activity coefficients of ZnCl₂, FeCl₂ and FeCl₃ are constant within a range of composition at a certain temperature, the molar ratio between Zn and Fe in the melt is an increasing function of $P_{{\text{O}}_2}^{1/4}$. Therefore, the molar ratio of Zn to Fe in the melt at equilibrium increases with the increase of partial pressure of O₂, which means that a larger partial pressure of O₂ thermodynamically favors the selective chlorination of Zn from ZnFe₂O₄.

By using thermochemical data,¹⁵⁾ activities of ZnCl₂, FeCl₃ and FeCl₂ at equilibrium under different conditions can be calculated from Eqs. (13), (14), (15), (16). In this work, experiments were conducted in Ar or Air atmosphere. Partial pressures of O₂ are considered to be 1.00 × 10⁻⁵ atm in Ar (given 10 ppm O₂ is contained as impurity) and 2.10 × 10⁻¹ atm in Air. Assuming molten MgCl₂–KCl is an ideal solution, activity of MgCl₂ equals its molar fraction (a_MgCl2 = x_MgCl2 = 0.56). Partial pressure of Cl₂ is calculated by using Eq. (13) and activities of ZnCl₂, FeCl₃ and FeCl₂ are obtained. Table 3 shows the calculated result. Activities of ZnCl₂ at equilibrium are far greater than unity, indicating that chlorination of Zn is highly spontaneous. On the contrary, activities of FeCl₃ or FeCl₂ are much smaller than unity in all cases. The increasing tendency with temperature indicates that chlorination of Fe is favored at higher temperatures. FeCl₂ is the only species in the system of which the activity at equilibrium declines with the increase in O₂ partial pressure.

Table 3. Calculated partial pressure of Cl₂ and activities of ZnCl₂, FeCl₃ and FeCl₂ in the melt after reaching equilibrium for the reaction between solid ZnFe₂O₄ and molten MgCl₂–KCl in Ar or air at different temperatures.

Temperature	773 K		873 K		973 K
Atmosphere	Ar	Air	Ar	Air	Ar	Air
PCl2 (atm)	1.47 × 10−3	2.12 × 10−1	2.85 × 10−3	4.13 × 10−1	4.82 × 10−3	6.99 × 10−1
aZnCl2	53.1	53.1	45.2	45.2	39.8	39.8
aFeCl3	1.00 × 10−5	1.00 × 10−5	3.00 × 10−4	3.00 × 10−4	5.65 × 10−3	5.65 × 10−3
aFeCl2	3.50 × 10−4	3.00 × 10−5	2.46 × 10−3	2.00 × 10−4	1.14 × 10−2	9.50 × 10−4

* O₂ concentration in Ar is assumed to be 10 ppm (P_O2 =1.00 × 10⁻⁵ atm).

* Molten MgCl₂–KCl is assumed to be an ideal solution and activity of MgCl₂ equals its molar fraction (a_MgCl2 = x_MgCl2 = 0.56).

Figure 8(a) shows XRD pattern of the solid residue recovered from the water-rinsed melt in the crucible after reacting ZnFe₂O₄ with molten MgCl₂–KCl for 50 h with Ar bubbling (150 ml min⁻¹) at 773 K. Only the peaks of MgO are detected in the residue, indicating that the ZnFe₂O₄ are mostly consumed in the reaction. SEM observation [Fig. 8(b)] indicates that MgO cubes are surrounded by some fine particles, which are identified to be Fe₂O₃ by EDX analysis [Fig. 8(c)]. It is considered that MgO and Fe₂O₃ are formed by reaction (1).

Fig. 8.

Characterization of the solid residue recovered from the water-rinsed melt in the crucible after reacting ZnFe₂O₄ with molten MgCl₂–KCl for 50 h with Ar bubbling (150 ml min⁻¹) at 773 K: (a) XRD pattern; (b) SEM image; (c) EDX spectrum and composition analysis of the designated area in (b). (Online version in color.)

3.3. Reaction Mechanism

According to the experimental results, reaction mechanism of solid ZnFe₂O₄ in molten MgCl₂–KCl can be proposed, as illustrated in Fig. 9. Initiating at the solid/liquid interface [Fig. 9(a)], the structure of ZnFe₂O₄ is likely to be broken by MgCl₂ in the melt to form new substances. It is possible to achieve a selective chlorination of Zn by controlling the thermodynamic conditions in the system (e.g. a large O₂ partial pressure). In this case, only Zn in ZnFe₂O₄ is chlorinated to form ZnCl₂, generating solid MgO and Fe₂O₃ as byproducts [Fig. 9(b)]. The newly formed ZnCl₂ immediately dissolves in the melt and diffuse toward bulk side driven by concentration gradient. At the end of the reaction [Fig. 9(c)], the solid reactant turns to a mixture of Fe₂O₃ and MgO, while the melt becomes MgCl₂–KCl–ZnCl₂. Dissolution of MgO in the melt can be ignored because the solubility is extremely small. In the case of complete chlorination, both Zn and Fe in ZnFe₂O₄ are chlorinated to form ZnCl₂ and FeCl_x (x = 2 or 3), generating MgO as byproducts [Fig. 9(d)]. At the end [Fig. 9(e)], the system contains solid MgO and molten MgCl₂–KCl–ZnCl₂–FeCl_x. In either case, overall chlorination rate is jointly determined by mass transfer in the melt and chemical reaction at the solid/liquid interface.

Fig. 9.

Reaction mechanism of solid ZnFe₂O₄ in molten MgCl₂–KCl. (Online version in color.)

4. Conclusions

This work investigates the reaction behavior of solid ZnFe₂O₄ in molten MgCl₂–KCl at temperatures from 773 K to 973 K. The results have proven the efficacy of converting ZnFe₂O₄ to ZnCl₂ and FeCl_x (x = 2 or 3) by MgCl₂. Zn is chlorinated prior to Fe in ZnFe₂O₄ under all conditions, indicating the possibility of separating Zn from Fe. Improving the mass transfer in the melt accelerates the reaction. Lower temperatures and larger O₂ partial pressure favor the selective chlorination of Zn, yet the reaction is more stagnant. This work has thus demonstrated the feasibility of treating EAF dust by using MgCl₂-based molten salt. Improvement in reaction kinetics will be the challenge to be tackled in the future study.

Acknowledgements

This work is partially supported by the “ISIJ Research Promotion Grant”. The authors would like to thank Prof. Il Sohn (Yonsei University) and Prof. Youngjo Kang (Dong-A University) for their valuable advices.

References

• 1)   K.  Nakajima,  K.  Matsubae-Yokoyama,  S.  Nakamura,  S.  Itoh and  T.  Nagasaka: ISIJ Int., 48 (2008), 1478. https://doi.org/10.2355/isijinternational.48.1478
• 2)   H.  Kubo,  K.  Matsubae and  T.  Nagasaka: J. Surf. Finish. Soc. Jpn., 66 (2015), 86. https://doi.org/10.4139/sfj.66.86
• 3)   X.  Lin,  Z.  Peng,  J.  Yan,  Z.  Li,  J.-Y.  Hwang,  Y.  Zhang,  G.  Li and  T.  Jiang: J. Clean. Prod., 149 (2017), 1079. https://doi.org/10.1016/j.jclepro.2017.02.128
• 4)   T.  Nagasaka and  K.  Matsubae-Yokoyama: Method for Recovering Zinc from Electric Furnace Steelmaking Dust and Device for Recovering Zinc from Electric Furnace Steelmaking Dust, U.S. Patent US20160177416A1, (2016).
• 5)   R.  Chairaksa-Fujimoto,  Y.  Inoue,  N.  Umeda,  S.  Itoh and  T.  Nagasaka: Int. J. Miner. Metall. Mater., 22 (2015), 788. https://doi.org/10.1007/s12613-015-1135-6
• 6)   T.  Miki,  R.  Chairaksa-Fujimoto,  K.  Maruyama and  T.  Nagasaka: J. Hazard. Mater., 302 (2016), 90. https://doi.org/10.1016/j.jhazmat.2015.09.020
• 7)   C.  Pickles: J. Hazard. Mater., 166 (2009), 1030. https://doi.org/10.1016/j.jhazmat.2008.11.110
• 8)   H.  Matsuura,  T.  Hamano and  F.  Tsukihashi: ISIJ Int., 46 (2006), 1113. https://doi.org/10.2355/isijinternational.46.1113
• 9)   H.  Matsuura,  T.  Hamano and  F.  Tsukihashi: Mater. Trans., 47 (2006), 2524. https://doi.org/10.2320/matertrans.47.2524
• 10)   H.  Matsuura and  F.  Tsukihashi: Miner. Process. Extr. Metall., 117 (2008), 123. https://doi.org/10.1179/174328508X290920
• 11)   G.  Sun,  H.  Matsuura and  F.  Tsukihashi: Mater. Trans., 58 (2017), 59. https://doi.org/10.2320/matertrans.M-M2016835
• 12)   H.  Sato and  K.  Okumura: Tetsu-to-Hagané, 106 (2020), 858 (in Japanese). https://doi.org/10.2355/tetsutohagane.TETSU-2020-032
• 13)   E. I.  Cooper and  D. H.  Kohn: Ceram. Int., 9 (1983), 68.
• 14)   J.  Kang and  T. H.  Okabe: Mater. Trans., 54 (2013), 1444. https://doi.org/10.2320/matertrans.M-M2013810
• 15)   O.  Knacke,  O.  Kubaschewski and  K.  Hesselmann: Thermochemical Properties of Inorganic Substances, Springer-Verlag, Berlin, (1991), 464, 677, 690, 1123, 1136, 2340, 2348, 2364.