Introduction

In 2020, total production of steel in the world was 1.878 billion tons of crude steel, 26.3% of which was obtained in electric arc furnaces [1]. Electric arc furnace dust (EAFD) is a by-product of electric steelmaking with high contents of zinc, lead, and iron. Electric arc smelting can produce about 15–25 kg of dust per ton of crude steel [2]. Currently, recycling proportion of generated dust is about 50% that leads to obtaining approximately 6% of all zinc produced in the world [3].

EAFD contains zinc both in the forms of hardly soluble ferrite (ZnFe2O4) and free highly soluble oxide (ZnO) [4]. There are also significant contents of chlorides [5] and sometimes fluorides [6] that hinder direct EAFD processing by hydrometallurgical zinc plants. The presence of zinc ferrite is undesirable due to a necessity for the application of additional stages of strong sulfuric acid leaching and iron precipitation after conventional weak sulfuric acid leaching [7]. Considerable contents of halogens require additional stages for their removal because they pass into the leaching solution and impede the further electrolysis process. Chlorine ions in the sulfate solutions cause corrosion of electrodes, increase energy consumption, and can also induce the release of gaseous chlorine, which deteriorates working conditions in the electrolysis plant [8]. Fluoride ions hamper stripping of electrolytic zinc from aluminum cathode [9,10,11]. Thus, to involve EAFD as a raw material for zinc production, its pretreatment is necessary to remove halogens and to decompose zinc ferrite.

In order to recycle EAFD with comprehensive extraction of valuable components, many authors have proposed various technologies using both pyrometallurgical [12] and hydrometallurgical methods [13]. The main EAFD recycling method is the Waelz process, which is pyrometallurgical processing of the dust in a rotary kiln with carbonaceous reducing agent [14]. At the present time, the Waelz process treats more than 85% of the EAFD involved in recycling [15]. In the Waelz process, carbothermic reduction at 1250–1300 °C decomposes ZnFe2O4, so zinc evaporates and passes into zinc-rich Waelz oxide. The Waelz oxide also accumulates halogens, so the next treatment stage is used to remove them either by calcination at temperatures over 900 °C [16, 17] or by leaching with aqueous solutions of sodium carbonate [18, 19]. Despite the widespread prevalence of the Waelz process, its disadvantage is impossibility to avoid the second-stage treatment of the Waelz oxide. Moreover, considerable consumption of coke breeze in the range of 180–350 kg per ton of dust should be noted [20] that leads to significant CO2 emission [21]. Currently, the global upcoming trends in industry is reduction of CO2 emissions [22, 23] and development of carbon–neutral economy [24], so in the foreseeable future, the Waelz process probably can be substituted for other methods for EAFD recycling with low greenhouse gas emissions. Among many different approaches, EAFD calcination with lime [25,26,27,28,29,30] is noteworthy due to the possibility of rotary kiln application, like the Waelz process, but probably lower CO2 emission compared with the one owing to a feasibility of the processing without coke breeze in the charge. It is important to note that decomposition of zinc ferrite and removal of halogens during the calcination of EAFD with lime occur within a single pyrometallurgical stage.

This research focuses on the in-depth investigation of zinc, lead, and iron behavior during EAFD calcination with lime. The experiments were carried out to study the effect of temperature, calcination time, and lime amount on the decomposition of zinc ferrite, as well as on the behavior of other components of EAFD.

Materials and Methods

Thermodynamic simulation was performed by HSC Chemistry 9.9 (Finland) software [31] using Equilibrium Composition module. The calculation was carried out per 100 kg of EAFD at temperature range of 600–1100 °C using gas phase containing 20% O2 + 80% N2 and model composition of the dust based on the results of chemical and phase analyses. It was assumed that all phases are pure substances with activity of 1.

The EAFD sample containing 29.95% Zn and 0.77% Pb, as well as reagent grade CaO, was used in the experiments. CaO was preliminary calcined at 1000 °C for 2 h to remove residual calcium hydroxide and carbonate. EAFD and CaO were ground in an agate mortar up to the fraction of 0.2 mm and mixed in certain ratios by V-shell blender for 24 h. The mixtures were pressed into tablets with a pressure of 170 MPa; each tablet was 1 g in weight and 17 mm in diameter. 3 g of the mixture in corundum crucible was placed in a preheated muffle furnace at the temperature in the range of 800–1100 °C. Calcination time was in the range of 1–120 min. Selective leaching of zinc oxide from ferrite in calcined samples was carried out by aqueous solution of NH4Cl + NH4OH [32]. 2 g of the sample was placed in a flask containing 200 ml of the mixture prepared by 22 g NH4Cl dissolution in 80 ml of NH4OH with a density of 0.9 g/cm3 and 120 ml of water. The solution was stirred at 50–60 °C for 2 h and filtered out. The residue was flushed out 4–6 times by 1% NH4Cl solution, then heated up to 50 °C, and flushed out 2–3 times by hot water. The obtained calcined samples and the ammonia-chloride solutions were analyzed by atomic emission spectroscopy with inductively coupled plasma (ICP-OES) by Agilent 725 (USA) device.

The chemical composition of the dust was analyzed by the atomic emission spectrometer Agilent 725 (USA) (Zn, Pb, Fe contents), LECO CS-400 (USA) device (carbon content), and X-ray fluorescence spectrometer ARL QUANT'X (Switzerland) (the rest elements). Chlorine in the calcined samples was determined by argentometric titration and turbidimetric methods [33]. Fluorine in the EAFD sample was analyzed by potentiometric titration using the state standard GOST 14048.7– 80. The XRD patterns were obtained by X-ray diffractometer ARL X'TRA (Switzerland) using Cu-Kα radiation.

Phase composition of the samples was determined using original software. The software algorithm was based on matching the experimental XRD pattern and the entries from the ICCD PDF [34] and COD [35] databases by varying intensity, width and shape of the peaks. The ICCD and COD data were used "as is": without changing the number, position, and relative intensity of peaks. The criteria for the satisfactory matching the experimental XRD pattern were the minimum standard deviation of the model from the experimental data, the correspondence of the composition and number of components for the chemical model, and the feasibility of the obtained set of variable parameters for the physical model. The semi-quantitative analysis of the components was assessed by the Reference Intensity Ratio (RIR) method using their I/Icor from the databases.

Mössbauer absorption spectra were obtained by Mössbauer spectrometer MS1104EM (Russian Federation). The γ-radiation source was 13 mCi 57Co in a matrix of metallic Rh. The spectra were collected at room temperature (296 ± 3 K) and at liquid nitrogen temperature (77.5 ± 0.5 K) until the noise/signal ratio reaches 1 or 2%. The experimental data processing was carried out for Mössbauer spectra of high resolution (1024 points) using the SpectRelax 2.8 software (Russian Federation). In the case of the processing using a model, the spectra were fitted by combinations of symmetrical doublets and sextets with controlled ratios of the intensity and width of the resonance lines. Values of chemical shift are given relative to α-Fe.

Particle size distribution in the samples was analyzed by laser light diffraction method using Fritsch Analysette 22 Next Nano (Germany) device.

To specify the zinc and lead phase distribution, methodology of chemical phase analysis was applied that described in detail in [32]. Figure 1a and b demonstrates two particular flowsheets for the selective dissolution of different zinc and lead phases, respectively.

Fig. 1
figure 1

The flowsheets of chemical phase analysis of Zn (a) and Pb (b) phases in EAFD

Full size image

A polished section was prepared to examine the calcined EAFD sample by the methods of optical microscopy (OM) and scanning electron microscopy (SEM). A reflected-light optical microscope METAM LV-34 (Russia) and a camera RisingCam RT5109 (China) were used to snap photomicrographs of the polished section. The sample was also analyzed using scanning electron microscope Tescan VEGA 3SB (Czech Republic) and energy-dispersive X-ray analyzer (EDX) INCA SDD X-MAX (United Kingdom).

Results

EAFD Characterization

Table 1 shows chemical composition of the EAFD sample.

Table 1 Chemical composition of the EAFD sample, wt%
Full size table

Figure 2 illustrates the XRD pattern of the EAFD sample; Table 10 in Appendix lists the quantitative composition of the sample using RIR method with the marked entry numbers of the identified phases in the XRD databases. The results indicate that the main minerals in the EAFD sample are ZnFe2O4 and ZnO. Furthermore, considerable contents of NaCl and KCl are present, which are likely the major halogen compounds of the sample. A certain percentage of charge materials of electric steelmaking such as CaCO3 and C are detected. The other identified phases are oxides and silicates such as Mn3O4, Fe3O4, FeMnO3, and Ca3Al2(SiO4)3. The presence of amorphous phases is clear due to an amorphous ring existence at 28–40° 2θ.

Fig. 2
figure 2

XRD pattern of EAFD, where (1) ZnO, (2) Fe3O4, (3) ZnFe2O4, (4) FeMnO3, (5) Cu0.4Zn0.6Fe2O4, (6) NaCl, (7) MgO, (8) KCl, (9) C, (10) Ca3Al2Si3O12, (11) CaCO3, (12) Mn3O4 are identified (entry numbers in the legend are given from the ICCD and COD databases)

Full size image

Table 2 shows the results of chemical phase analysis of zinc and lead minerals.

Table 2 Distribution of zinc and lead minerals in the EAFD sample
Full size table

The obtained data demonstrate that zinc content both as chlorides and as sulfides is insignificant; the main zinc-containing phases are ferrite and oxide in approximately equal proportion. As in the case of zinc, the lead content in the chloride and sulfate form is negligible, but major phases are distributed differently. Most of the lead is in silicate form, but some amounts of sulfates, oxides, metal, and ferrites are also present.

Figure 3 depicts the Mössbauer spectra of EAFD collected at the both temperatures; Table 3 lists values of its parameters. They involve an intense and wide doublet, as well as a poorly resolved combination of broad and low-intensity resonance lines of a distorted sextet. The spectrum obtained at 78 K has a relatively high intensity and symmetry, while the spectrum obtained at 298 K is asymmetric, not so high intensity, and is clearly a superposition of several subspectra. A strong temperature dependence of both the intensity and the profile indicates that components in the superparamagnetic state are present in the EAFD sample. The change in the symmetry of the ferromagnetic part of the spectra is a characteristic behavior of magnetite.

Fig. 3
figure 3

Mössbauer spectra of the EAFD sample obtained at 296 and 78 K, as well as models for their description

Full size image
Table 3 Parameters of the Mössbauer spectra of the EAFD sample
Full size table

Most of the ferromagnetic part of the spectra can be described using the many-state superparamagnetic relaxation model [36] by two sextets with a single set of relaxation parameters (Table 3, ##2–3). The Mössbauer parameters of these sextets correspond to Fe3+ atoms in tetrahedral voids and Fe3+/Fe2+ atoms in octahedral voids of the spinel-type crystal lattice [37]. In particular, the hyperfine parameters are hybrid between the isostructural magnetite and maghemite [38], which suggest that the subspectra are related to partially oxidized magnetite Fe3-δO4 [39, 40]. It should be noted that magnetite and maghemite, as well as their solid solutions, are isostructural and almost indistinguishable in XRD patterns. If we assume that the sample particles have a spherical shape and the value of the magnetic anisotropy constant at room temperature is 105 J/m3, which is equivalent to the average value for nanosized maghemite and magnetite [41, 42], the size of the corresponding magnetic domains is estimated at the level of about 5.5 nm. It is consistent with a noticeable broadening of the diffraction peaks for the corresponding phase in the XRD pattern (Fig. 2).

The rest of ferromagnetic part of the spectra is the sextet (Table 3, #1) which is probably a substance related to copper ferrite [43], namely, the substituted copper ferrite Cu1-xZnxFe2O4 [44, 45]. The subspectrum area significantly increases with decreasing the collecting temperature, which can be associated with ferrimagnetic interactions between ferrimagnetic nanoparticles of inhomogeneous composition distributed over the diamagnetic matrix of the composite. Apparently, the mentioned area change of the subspectrum #1 with the temperature shift occurs simultaneously with the outer most intense and wide doublet #4 (Table 3), which is one of the two describing the paramagnetic part of the Mössbauer spectra. Therefore, one can assume that this doublet partially describes copper ferrites with a high substitution degree of copper by zinc and Tc (Curie point) in the studied temperature range [46]. On the other hand, it should be noted that this subspectrum is likely a compound doublet. It can include another doublet with similar parameters close to FeMnO3, which has Tc below the nitrogen boiling point [47,48,49]. A substantial width of the corresponding resonance lines even at 78 K indicates a low crystallinity of this phase, which agrees well with the XRD pattern (Fig. 2), where the corresponding peaks are also significantly broadened.

The second doublet, which has the significantly smaller values of the quadrupole splitting and the line width, corresponds to zinc ferrite ZnFe2O4 [50, 51]. The relatively narrow resonance lines for this doublet are consistent with a low width of the XRD peaks that indicate a high crystallinity of the corresponding phase.

Thermodynamic Simulation

Based on the results of chemical and phase analysis without taking into account components of low contents, initial EAFD composition for the simulation was accepted as follows: wt%: 58.69 ZnFe2O4; 17.47 ZnO; 2.34 Ca3Al2Si3O12; 1.18 MgO; 0.38 CuFe2O4; 4.20 CaO; 4.29 CaCO3; 2.61 KCl; 2.16 NaCl; 1.49 C; 0.30 Fe3O4; 1.55 SiO2; 1.65 Mn3O4; 0.61 Mn2O3; 0.16 Cr2O3; 0.71 Pb2SiO4; 0.10 PbO; 0.06 PbSO4; 0.05 Pb. Figure 4 shows the results of thermodynamic simulation of equilibrium states.

Fig. 4
figure 4

Effect of CaO addition on equilibrium amounts of major zinc, iron, and calcium compounds in the dust at 1000 °C (a) and temperature dependence of equilibrium amounts of zinc and calcium compounds with addition of 30% CaO to 100% EAFD (b)

Full size image

The thermodynamic simulation has evidenced that the main reaction of zinc ferrite decomposition is interaction with the formation of dicalcium ferrite:

$$ {\text{ZnFe}}_{{2}} {\text{O}}_{{4}} + {\text{ 2CaO}} \to {\text{Ca}}_{{2}} {\text{Fe}}_{{2}} {\text{O}}_{{5}} + {\text{ ZnO}} $$
(1)

Figure 4a shows that in order to convert zinc into the highly soluble form completely, at least 26 wt% CaO should be added. Accordingly, in the preliminary experiments, a small excess of 30 wt% CaO addition was used.

Chlorine Behavior

To define more accurately the required calcination temperature, a preliminary experimental study of chlorine behavior was carried out. To prevent the negative consequences of halogen influence, chlorine and fluorine should be removed from the calcine to the required level of < 0.06% Cl and < 0.01% F [52]. The EAFD sample has already complied with the requirements for fluorine content (Table 1), so we studied only the behavior of chlorine. Table 4 demonstrates the residual chlorine content in the calcined EAFD samples obtained during the experiments.

Table 4 Remaining chlorine content in the samples obtained by the calcination of EAFD + 30% CaO at various temperatures and duration, wt%
Full size table

The results have indicated that to conform with the requirements for the chlorine content, the EAFD calcination is required at least at 1000 °C for 120 min. Based on these data, we chose duration of 1–120 min for further experiments.

Zinc Behavior

Figure 5 shows the influence of CaO amount and calcination time on zinc ferrite conversion into the oxide form.

Fig. 5
figure 5

Effect of CaO addition on the relative zinc percentage in the oxide form at 1000 °C after 60 min of the calcination (a) and calcination duration at 950–1050 °C with the addition of 30% CaO to 100% EAFD (b)

Full size image

It can be seen that the EAFD calcination without lime and with insignificant lime addition led to decreasing of zinc percentage in the highly soluble ZnO form (Fig. 5a) that is probably due to partial ferritization by the interaction of free zinc and iron oxides contained in the dust. At 20 wt% CaO addition to EAFD, 87.7% of zinc was in ZnO form, while 30 wt% CaO addition resulted in 90.2% of zinc in ZnO form. Further increasing of CaO addition influenced the remaining zinc ferrite decomposition insignificantly.

Figure 4b demonstrates that the share of zinc in the highly soluble ZnO form at all chosen temperatures was in the range of 80–90% already after 10 min of calcination. It is noteworthy that the curves at all the temperatures are almost the same that is consistent with the other study [25]. Evidently, the best result was 90% of zinc in highly soluble ZnO form that was obtained at 1000 °C after 45 min of the calcination. Thus, to convert 90% of zinc into ZnO form, the EAFD calcination with 30 wt% CaO at 1000 °C for 45 min is quite sufficient. However, to remove also chlorine to the required level, it is necessary to increase the calcination time to 120 min.

Lead Behavior

Figure 6 shows the kinetic curves of Pb removal in the range of 900–1100 °C.

Fig. 6
figure 6

Effect of calcination duration on lead vaporization from EAFD with 30% CaO at 900–1100 °C

Full size image

As can be seen, the Pb removal rate after the 120 min calcination is less than 30% at 900 °C and above 80% at 1000–1100 °C.

A rate-controlling step of Pb removal process in the range from 1 to 120 min was assessed by conventional model fitting of the experimental data with models based on different kinetic equations. Table 5 compares adequacy of the equations for the experimental data at different temperatures.

Table 5 Comparison of determination coefficient of several kinetic models for lead removal during calcination of the EAFD sample with CaO in the mass ratio of 100:30
Full size table

As follows from the listed data, the Jander equation gives the best fit to the Pb removal experimental curves that corresponds to the diffusion-controlled process. Figure 7a clearly demonstrates the adequacy of the Jander model at 900–1000 °C. Nevertheless, the Jander equation as the other models is in poor agreement with the experimental curve of 1100 °C.

Fig. 7
figure 7

Plot of Jander equation F(x) = [1 − (1 − x)1/3]2 vs. time at 900–1100 °C for Pb removal during the calcination of EAFD with 30% CaO (a) and Arrhenius plot for the process (b)

Full size image

The rate constants at different temperatures were evaluated from the appropriate slopes in Fig. 7a. Apparent activation energy was calculated using the logarithmic form of the Arrhenius equation:

$$\mathrm{ln}k=\mathrm{ln}A+ \frac{E}{RT}$$
(2)

In the equation, E is the apparent activation energy (J/mol), k is the rate constant (min−1), A is the preexponential factor (min−1), R is the universal gas constant (8.314 J/(mol∙K)), T is the temperature (K), respectively. Figure 7b shows the results of determination of the apparent activation energy by the graphical method using the Eq. (2). Only the curves fitted to the Jander equation were taken into account, so the curve of 1100 °C was excluded from the calculation.

Correspondence of the Pb removal process with the Jander model agrees well with previously reported data [53], but the values of apparent activation energy differ by more than 2 times: 365 kJ/mol in our study vs. 175 kJ/mol in [53]. Taking into account lead mineralogy in the EAFD sample (Table 2), to elucidate the behavior of lead during the calcination, thermodynamic calculations were carried out using HSC Chemistry software. Figure 8a demonstrates the equilibrium amounts of Pb compounds at different temperatures during the calcination of EAFD.

Fig. 8
figure 8

Temperature dependence of equilibrium amounts of lead compounds with addition of 30% CaO to 100% EAFD (a) and temperature dependence of the Gibbs free energy change of the reactions (3–7) (b)

Full size image

As follows from the equilibrium calculation, there is only an amount of lead as PbO in the condensed phase; condensed PbO decreases and PbO(g) increases with a temperature rise over 850 °C. Moreover, above 1000 °C, the gas phase contains small amounts of lead chloride and metallic lead vapors.

Probably, lead components, which were detected by chemical phase analysis (Table 2), during the EAFD calcination behaves as follows. A small amount of lead chloride contained in the dust vaporizes already during the heating up to the required temperature [54]. Lead sulfide interacts with atmospheric oxygen to form oxysulfates and sulfates [55] and then the oxysulfates and the sulfate already contained in the dust in the range of 800–1000 °C decompose stepwise to oxide according to the scheme [56, 57]: PbSO4 → PbO∙PbSO4 → 2PbO∙PbSO4 → 4PbO∙PbSO4 → PbO.

It is reported that the removal of lead oxide from EAFD can occur through interaction with chlorides by the following reactions [53]:

$$ {\text{2NaCl }} + {\text{ PbO }} + {\text{ 2SiO}}_{{2}} + {\text{ Al}}_{{2}} {\text{O}}_{{3}} \to {\text{ PbCl}}_{{{2}({\text{g}})}} + {\text{ 2NaAlSiO}}_{{4}} $$
(3)
$$ {\text{2KCl }} + {\text{ PbO }} + {\text{ 2SiO}}_{{2}} + {\text{ Al}}_{{2}} {\text{O}}_{{3}} \to {\text{ PbCl}}_{{{2}({\text{g}})}} + {\text{ 2KAlSiO}}_{{4}} $$
(4)

Similar reactions are likely to be possible for silicates, which are the main lead phases in EAFD in this study:

$$ {\text{4NaCl }} + {\text{ Pb}}_{{2}} {\text{SiO}}_{{4}} + {\text{ 3SiO}}_{{2}} + {\text{ 2Al}}_{{2}} {\text{O}}_{{3}} \to {\text{ 2PbCl}}_{{{2}({\text{g}})}} + {\text{ 4NaAlSiO}}_{{4}} $$
(5)
$$ {\text{4KCl }} + {\text{ Pb}}_{{2}} {\text{SiO}}_{{4}} + {\text{ 3SiO}}_{{2}} + {\text{ 2Al}}_{{2}} {\text{O}}_{{3}} \to {\text{ 2PbCl}}_{{{2}({\text{g}})}} + {\text{ 4KAlSiO}}_{{4}} $$
(6)

Lead silicate can react with lime to form PbO:

$$ {\text{Pb}}_{{2}} {\text{SiO}}_{{4}} + {\text{ 2CaO }} \to {\text{ PbO }} + {\text{ Ca}}_{{2}} {\text{SiO}}_{{4}} $$
(7)

Figure 8b shows free Gibbs energy change of the reactions (3–7) at varied temperature. The presented data indicate that the reactions (3–7) are possible from 600 to 750 °C and above; the reaction (5) is the most thermodynamically probable at temperatures over 800 °C among the chlorination reactions. However, the reaction (7) is more probable than the reactions (3–6), so PbO is likely formed from lead silicate that is confirmed by the thermodynamic simulation (Fig. 8a).

Thus, based on the previous calculations, one might assume that the main part of lead during the EAFD calcination transforms into PbO and vaporizes as PbO(g). It is supported by the fact that lead oxide has a significant vapor pressure at temperatures above 950 °C [58, 59] that is consistent with our experimental data (Fig. 6). In addition, a different evaporation condition compared with [53] is likely explained by the twofold value of apparent activation energy (Fig. 7b).

However, it is noteworthy that from 18 to 24 rel.% Pb has been already vaporized at 900–1000 °C during the heating and within the first minute of the holding, while at 1100 °C, this value is 62% (Fig. 6). Consequently, the chlorination of a certain part of lead by the reactions (3–6) and its evaporation in the chloride form during the calcination of EAFD is also possible. This assumption is confirmed by authors [60, 61], who has found out a high rate of Pb removal as lead chlorides at the temperature range of our study, as well as a possibility of lead vaporization at the temperatures even below 850 °C. After the main part of chlorides contained in the EAFD has been vaporized, a lead evaporation mechanism is also changed. This also explains a discrepancy between the experimental curve at 1100 °C and the kinetic models because a major part of lead was removed during the heating by the chlorination mechanism; then the rest was evaporated during holding as PbO(g). The mechanism change is also confirmed by the more negative value of the Gibbs free energy change of the chlorination reaction (5) than the reaction (7) above 1000 °C. This observation is of practical importance because the addition of chlorides to EAFD can increase the rate and reduce the temperature of the Pb removal.

It can be inferred from the obtained data that a temperature above 1000 °C is favorable for Pb removal from EAFD. Temperature of 1000 °C and the calcination duration of 120 min are the experimental conditions of best results for 87% Pb removal. It creates an opportunity to obtain during the calcination an intermediate product that can be used in lead production.

Calcined EAFD Characterization

Table 6 shows chemical composition of the calcined EAFD sample at the optimal calcination conditions.

Table 6 Chemical composition of the calcined EAFD sample with the addition of 30% CaO at 1000 °C for 120 min, wt%
Full size table

Figure 9 illustrates the XRD pattern of the calcine obtained at optimal conditions. Table 11 lists the quantitative composition of the sample using RIR method with the marked entry numbers of the identified phases in the XRD databases.

Fig. 9
figure 9

XRD pattern of calcined EAFD with 30% CaO at 1000 °C for 120 min, where (1) ZnO, (2) Ca2Fe1.7Zn0.15Si0.15O5, (3) Ca2Fe2O5, (4) Ca2Cr0.5Fe1.5O5, (5) CuCr1.7Mn0.3O4, (6) CaO, (7) Ca(Fe,Mg)(CO3)2 (8) (Mn2.58Fe0.34Ca0.08)(Al1.99Fe0.01)Si3O12, (9) Ca2SiO4, (10) K2CaSiO4 are identified (entry numbers in the legend are given from the ICCD and COD databases)

Full size image

XRD analysis has confirmed that the main part of zinc transforms into ZnO during the calcination according to reaction (1). The peaks are quite narrow due to high temperature treatment. The main iron-containing phase is dicalcium ferrite with a substitution of iron by different minor elements. An inconsiderable part of CaO remains unreacted that led to formation of complex carbonates probably during an air storage between cooling of the calcined EAFD sample and XRD analysis. There is a small amorphous ring on the XRD pattern; therefore, amorphous phases are present in the calcined EAFD.

It is notable that particle size of the calcined EAFD sample was increased compared to the initial EAFD sample. If the EAFD sample contains 31.7% particles < 1 μm and 82.7% particles < 10 μm, these values for the calcined EAFD sample grinded up to less 200 μm are 3.5% and 47.8%, respectively.

Figures 10 and 11 depict the Mössbauer spectra of the calcined EAFD with 30 wt% CaO and without CaO addition, respectively; Tables 7 and 8 list values of their parameters.

Fig. 10
figure 10

Mössbauer spectra of the calcined EAFD sample with 30% CaO at 1000 °C for 120 min, models for their description, and the corresponding hyperfine magnetic field probability distribution functions

Full size image
Fig. 11
figure 11

Mössbauer spectra of the calcined EAFD sample at 1000 °C for 120 min without CaO addition and models for their description

Full size image
Table 7 Parameters of the Mössbauer spectra of the calcined EAFD sample with 30% CaO at 1000 °C for 120 min
Full size table
Table 8 Parameters of the Mössbauer spectra at 296 K of the calcined EAFD sample at 1000 °C for 120 min without CaO addition
Full size table

The Mössbauer spectrum of the sample calcined at 1000 °C without the addition of CaO is a doublet that is a superposition of two subspectra for high-spin Fe3+ in an octahedral oxygen environment (Fig. 11, Table 8). Hence, this spectrum has confirmed the occurrence of additional ferritization of various elements including Zn during calcination without CaO (Fig. 5a). Therefore, one can assume that all changes in the Mössbauer spectra after the EAFD calcination with CaO are associated only with the formation of calcium-iron compounds. As follows from the spectra, they consist of more than ten resonance lines of different width and intensity. Moreover, the resonance lines in the outer part of the spectrum obtained at room temperature have a significant width and a distortion toward to the inner part of the spectrum. However, the resonance lines for the spectrum obtained at 78 K have a quite symmetry shape, a reduced line width, and a doubled intensity. Such an effect of the temperature indicates the presence of iron compounds with a complex poorly ordered magnetic structure in the sample. Accordingly, the main part of spectra is two sextets, which described by similar hyperfine magnetic field probability distribution functions with the same area of the subspectra. It can be stated clearly that the corresponding subspectra describe the belonging of iron atoms to the same phases, moreover, in a ratio of 1:1. According to the values of parameters (Table 6), the sextets are related to dicalcium ferrite derivatives [62,63,64,65] with the brownmillerite structure [66, 67]. The distinction of iron compounds in this structural type is the placement of Fe3+ atoms in two different crystallographic positions with octahedral and tetrahedral oxygen environments, respectively [68]. Another distinguishing feature is high values of the quadrupole splitting of both the subspectra that require taking into account the quadrupole interaction constant to the second order of perturbation theory expressed by the parameters a+ and a- [69,70,71]. For this sample, the a+/a- ratio is -1, which corresponds to the value of the polar angle between Vzz and the direction of the internal magnetic field Hhf θ = 90° as in the paper [72]. Appendix (Table 12, Fig. 13) contains more rigorous description of the spectra in terms of full diagonalization of hyperfine Hamiltonian without any change in the given conclusion.

Hyperfine magnetic field probability distribution function illustrated in Fig. 10 for the both temperatures are trimodal at least (the leftmost mode observed in the spectra at room temperature can obviously be ignored due to the border effect). Evidently, the sample contains a number of solid solutions based on dicalcium ferrite Ca2MxFe2-xO5, which are mainly different in a decrease of hyperfine magnetic field value with an increase in the proportion (x) of impurity cations (M3+) substituting Fe3+ atoms in both sublattices [73, 74]. Probably, different modes in the probability distribution functions are related to different impurity compositions of dicalcium ferrite. These impurities must obey the rules of isomorphic substitution [75]; according to the data on the chemical composition of the sample (Table 6), they are likely Al3+, Cr3+, and Mn3+ [76,77,78,79]. The rest of the spectra is a doublet with a high values of quadrupole splitting and isomeric shift, which corresponds to Fe3+ atoms in a tetrahedral oxygen environment [80]. Due to the absence of corresponding iron-containing compound in the XRD pattern, the doublet is likely to be related to an amorphous phase. The values of the parameters suggest that it is a composite based on potassium-silicon glass phase with a low glass point with a high content of potassium or calcium–aluminum glass with a low content of calcium or a combination of the both ones [81, 82], where Fe3+ atoms substitute Si4+ and Al3+ in tetrahedral positions, respectively, to form a solid solution.

Figure 12 shows photomicrographs of the calcined EAFD with 30 wt% CaO obtained using optical and scanning electron microscopy, respectively. Table 9 lists the composition of some points of SEM image, which are indicated in Fig. 12b.

Fig. 12
figure 12

OM (a) and SEM (b) images of the calcined EAFD with 30% CaO at 1000 °C for 120 min. The image (a) consists of dicalcium ferrite area 1) (dark gray) and zinc oxide area 2) (light gray). The elemental composition of marked points in the SEM image is in Table 9

Full size image
Table 9 The elemental composition of points indicated in Fig. 12(b) obtained by EDX analyzer
Full size table
Table 10 Characteristics of two approaches based on the Waelz process and the calcination with lime in a rotary kiln for the EAFD treatment
Full size table

The microscopy methods have pointed out that the calcined EAFD consists of a dicalcium ferrite base with zinc oxide and dicalcium silicate impregnation. The EDX analysis of points of dicalcium ferrite area in the Fig. 12 detected Zn, Mn, Si, and Al besides Ca and Fe. It cannot be ruled out that a part of the impurities in the sample points 1A, 1B, and 1C is in other separate components, which may be accidentally detected due to essential spatial and depth resolution of EDX analysis [83]. However, taking into account quite stable qualitative composition of these and other dicalcium ferrite points, XRD pattern (Fig. 9), which involves 3 various dicalcium ferrite entries, as well as Mössbauer spectra parameters (Fig. 10; Table 7), we assume that dicalcium ferrite include Zn, Mn, Si, and Al as substitutional impurities.

Among the impurities of dicalcium ferrite, zinc deserves particular attention. According to Fig. 5, 90.2% Zn is in the form of ZnO. The Mössbauer analysis (Fig. 10; Table 7) suggests that there is no significant amount of iron in the zinc ferrite form in the calcined EAFD sample; the additional estimation resulted that less than 0.8% of iron can be potentially included in zinc ferrite. Hence, the cumulative evidence suggests that the rest of zinc is mainly a substitutional impurity of dicalcium ferrite that predetermines zinc losses in a residue during following leaching of the calcine.

Discussion

The study has shown that the EAFD calcination with 30 wt% CaO at 1000 °C for 120 min leads to the conversion of 90.2% Zn into highly soluble form and to the removal of 87% Pb, as well as of 95% Cl that is sufficient for the use of the calcined dust during further electrolysis. The only one required pyrometallurgical stage at relatively low temperatures with reduced fuel consumption is an undoubted advantage of this technology compared to the Waelz process. It is noteworthy that other papers [27, 28, 30], which studied the process on a laboratory scale, also achieved an incomplete zinc conversion into the oxide form at the level of about 90%. However, real dynamic calcination conditions in a rotary kiln, which are the better conditions for the interaction of solid phases compared to the static experimental ones, can increase the completeness of zinc transformation. In support of this assumption, the paper [84] has confirmed comparable to the Waelz process total zinc recovery of 95%, where the EAFD calcination technology with lime followed by alkaline leaching was tested in semi-industrial conditions for the production of zinc powder.

Touching upon greenhouse gas footprint of EAFD processing by the Waelz process and the calcination with lime, it is important to emphasize some points. The main sources for the CO2 emissions for the approach based on the Waelz process are coke breeze combustion, flux (limestone or dolomite) decomposition, and gas fuel combustion at the Waelz oxide treatment stage to remove halogens (if applicable). It is notable that coke breeze in the Waelz process is indispensable because it is not only fuel and reducing agent, but also prevents charge melting and incrustation in a rotary kiln [14]. As for the approach based on the calcination with lime, CO2 emissions not only from gas fuel combustion during the processing should be taken into account, but also from carbonate decomposition and gas fuel combustion during the production of lime from limestone. The conventional gaseous fuel used in the mentioned processes is natural gas, which is combusted with the lowest CO2 emissions of all fossil fuels at about 15 kg C/GJ compared to coke breeze with about 29 kg C/GJ [85]. Evidently, because of a similarity of the processing conditions for the EAFD calcination with lime and the lime production from limestone, to reduce energy costs and CO2 emissions, these two processes should be combined, i.e., EAFD can be calcined with limestone at once; such method invites further investigations.

Table 10 clearly demonstrates the main distinctive features of the EAFD processing using the Waelz process and the calcination with lime. Besides the above-mentioned advantages in costs due to reduced temperature and avoidance of a halogen removal stage of the EAFD calcination with lime, it generates 6–8 times less sublimate than the Waelz process that allows to use a simpler gas cleaning system. On the other hand, an approximately threefold growth of the calcine amount compared to the Waelz oxide is expected that requires a corresponding increase in productivity of the next leaching stage. Additionally, it should be appreciated that the Waelz process is highly flexible for the charge feedstock of various zinc speciation including oxide, ferrite, sulfide, and sulfate, while the performance of the calcination with lime has still proved only for the EAFD containing zinc in ferrite and oxide form.

One more point to be mentioned for the EAFD calcination with lime is unclear prospect for the residue utilization after the calcination and zinc oxide leaching, regardless of solvents used. Until recent times, no specific study has been carried out to utilize such the residue. On the other hand, it is known that the Waelz slag from EAFD processing, which considered as a hazardous waste [86], is partially recycled in construction industry. However, its recycling is limited due to significant residual contents of Zn and other heavy metals that often disallow to meet the standards according to leaching tests for construction materials [87].

Taking into account the given advantages and disadvantages, we assume that the method of EAFD processing based on its calcination with lime should be considered for the construction or reconstruction of plants, which recycle EAFD, on the same level with the Waelz process or even as a substitution of Waelz process.

Conclusions

The paper investigated in detail the chemical and mineralogical composition of EAFD and determined its phase distribution of zinc, lead, and iron. The thermodynamic calculation, as well as laboratory experiments to investigate zinc transformation from ZnFe2O4 into ZnO, Pb removal during the EAFD calcination with lime, and Mössbauer analysis of the calcined EAFD sample have revealed the behavior of valuable dust components. The calcination has led to the conversion of the main part of iron from ferrite and magnetite form to Ca2Fe2O5. It was found that the major part of zinc transforms into ZnO during the calcination at 900–1100 °C, while its minor part remains as a substitutional impurity of dicalcium ferrite. At the same time, the main part of lead vaporizes both as chlorides and PbO(g); the primary mechanism of lead evaporation depends on the calcination temperature and EAFD composition.

The optimal calcination conditions, which led to obtaining 90.2% Zn in free oxide form, as well as volatilization of 87% Pb and 95% Cl, are as follows: EAFD:CaO mass ratio is 1: 0.3, the temperature is 1000 °C, and the duration is 120 min.