Effect of Fe/SiO₂ Ratio and Fe₂O₃ on the Viscosity and Slag Structure of Copper-Smelting Slags

Abstract

Secondary copper smelting is an effective means of treating waste resources. During the smelting process, the viscous behavior of the smelting slags is essential for smooth operation. Therefore, the effects of Fe/SiO₂ ratio and Fe₂O₃ contents on the viscous behavior of the FeO−Fe₂O₃−SiO₂−8 wt%CaO−3 wt%MgO−3 wt%Al₂O₃ slag system were investigated. The slag viscosity and activation energy for viscous flow decrease with increasing Fe/SiO₂ from 0.8 to 1.2, and increase as the Fe₂O₃ content increases from 4 wt% to 16 wt% at Fe/SiO₂ ratio of 1.2. However, under the conditions of Fe/SiO₂ of 0.8 and 1.0, the viscosity and activation energy for viscous flow show a minimum value at Fe₂O₃ content of 12 wt%. Fe₂O₃ exhibits amphoteric properties. In addition, the increase in Fe₂O₃ content raises the breaking temperature of the slag, while the Fe/SiO₂ ratio has the opposite effect. Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy show that increases in Fe/SiO₂ ratio lead to simplification of the silicate network structure, while increases in Fe₂O₃ content improves the formability of the network. This study provides theoretical support for the related research and application of secondary copper smelting.

1. Introduction

The mixed feed containing primary and secondary raw materials is and has long been a standard operating procedure in the copper-smelting industry. In particular, in the production practice of oxygen-enriched bottom-blow smelting, the separation of copper matte from slag is achieved by mixing gold concentrate, copper concentrate, flux, and secondary resources for smelting. As the copper matte is an excellent solvent for gold and silver, most of the precious metals are dissolved into the copper matte, thus realizing the recovery of gold and silver [1,2]. Secondary copper resources generally include industrial residues, sludge, and domestic waste rich in copper or precious metals. Secondary resources can be reduced and rendered harmless and simultaneously multimetal (copper, gold, silver, etc.) components can be recovered together by dispensing secondary resources into the smelter.

As a by-product of the secondary copper smelting process, iron–silicate–based slag (usually this slag contains significant amounts of Fe₂O₃ and traces of CaO and Al₂O₃) is formed in the smelter that must be tapped out at the end of the process. It has been reported that the fayalite phase precipitates formed in iron–silicate slag at approximately 1473 K increases slag viscosity [3]. At a comparatively low slag temperature of approximately 1370 K, silicate phases containing iron and calcium precipitate that increase slag viscosity even further [4]. When the iron content in the slag is high, the magnetite phase also precipitates in the slag, existing as a fine and diffuse suspended solid in the high-temperature melt [4]. The viscosity of slag is critical to the smooth operation of the process because high viscosity causes several issues, including high copper losses and reduced metallurgical reaction rates, which are common in the secondary copper smelting, particularly for slags with high Fe₂O₃ content. Therefore, an in-depth understanding of the physical properties of slags containing Fe₂O₃ and the role of iron oxides with different valences, in particular, is critical for optimizing the smelting process.

At present, studies on the role of iron oxides in the melt mainly focus on typical calcium–silicate−based slag systems, such as CaO−SiO₂−Al₂O₃−Fe_xO [5], CaO−SiO₂−Al₂O₃−MgO−FeO [6,7,8], CaO−SiO₂−Al₂O₃−MgO−Fe_TO−P₂O₅ [9], and CaO−SiO₂−FeO−ZnO−Al₂O₃ [10], which are characterized by a high calcium–silica ratio. In comparison, less research has been performed on iron–silicate–based slags to date. The viscosity of iron–silicate slags is primarily measured under iron-saturated conditions [3,11,12,13,14]. As a result, almost all iron oxides present in the slag are in the form of divalent iron, and the role of trivalent iron in the slag is rarely reported. According to the literature [6,7,10], FeO is a network modifier with regular octahedral coordination, which decomposes to produce free oxygen. This depolymerizes the silicate network structure and reduces the slag viscosity. The role of Fe₂O₃ in the silicate structure, on the other hand, is still being debated [15]. In slag, Fe³⁺, similar to Al³⁺ and Ti⁴⁺, is an amphoteric ion capable of forming tetrahedral and octahedral structures [16,17,18]. Saito et al. [19] and Wright et al. [20] investigated the effect of Fe₂O₃ on the viscosity of 40 wt%CaO−40 wt%SiO₂−20 wt%Al₂O₃ slag and iron–silicate slag, respectively, and found that the slag viscosity decreased with Fe₂O₃ addition. Mysen et al. [21] also reached the same conclusion in CaO−SiO₂−Fe−O system; they observed that the viscosity decreased distinctly as the Fe₂O₃ content reached 5 wt% compared to iron-free melts. In addition, Sasaki et al. [22] discovered that the role of Fe³⁺ ions in sodium silicate melts (Na₂O–SiO₂–FeO–Fe₂O₃), which increased viscosity. Similarly, Viswanathan et al. [23] found that when the Fe₂O₃ content exceeds 7% in the 68%FexO−32%SiO₂ slag system, the viscosity increases sharply. Furthermore, the slag’s Fe/SiO₂ ratio is a critical indicator in determining slag viscosity. In the decreasing order of importance, the homogeneous liquid slag viscosity of iron–silicate metallurgical slag is affected by various factors: Fe/SiO₂ ratio > temperature > alternative oxides, such as Al₂O₃, CaO, MgO, and Cu₂O > Fe³⁺/Fe²⁺ ratio. Park et al. [24] explored the viscous behavior of copper-smelting slag and found that an increase in Fe/SiO₂ ratio decreases the slag viscosity. Ducret et al. [12] observed that the FeO−Fe₂O₃−SiO₂−CaO−MgO slag viscosity decreases significantly as the Fe/SiO₂ ratio increases from 1.30 to 2.00. However, the extent to which factors affect the viscosity varies across slag systems and slag compositions. Therefore, it is necessary to conduct a systematic investigation on the viscous behaviors of the FeO−Fe₂O₃−SiO₂−CaO−MgO−Al₂O₃ slags.

In this paper, the viscosity of the FeO−Fe₂O₃−SiO₂−CaO−MgO−Al₂O₃ slag system was measured using the rotating spindle method. The breaking temperature, viscosity and activation energy for viscous flow was analyzed under high-temperature conditions with different Fe/SiO₂ ratios (0.8−1.2) and Fe₂O₃ contents (4−16 wt%). In addition, the relationship between slag structure and viscosity was studied via FTIR and Raman spectroscopy. This paper will be of significance in the smelting treatment of secondary resources in the processes of copper industry.

2. Experimental

2.1. Sample Preparation

All chemicals, including SiO₂, Fe₂O₃, MgO, Al₂O₃, CaO, and Fe powder, were of analytical reagent grade purity. Except the Fe powder, all reagents were preroasted at 1273 K for 2 h to remove any volatile impurities. The reagents were then precisely weighed and uniformly mixed according to the predetermined composition listed in

Table 1. Chemical compositions of samples in viscosity determination experiments (wt%).

No.	Designed Composition (wt%)							Analyzed Composition (wt%)
	Fe/SiO2	Fe2O3	SiO2	FeO	MgO	Al2O3	CaO	Fe/SiO2	Fe2O3	SiO2	FeO	MgO	Al2O3	CaO
1	0.8	4	42.18	39.82	3	3	8	0.82	4.6	41.4	39.5	2.8	3.2	8.0
2	0.8	8	41.99	36.01	3	3	8	0.82	8.5	41.0	35.6	3.1	3.3	8.0
3	0.8	12	41.79	32.21	3	3	8	0.82	12.6	40.9	31.8	3.0	3.2	8.0
4	0.8	16	41.59	28.41	3	3	8	0.82	16.6	40.5	27.9	3.1	3.1	7.8
5	1.0	4	37.44	44.56	3	3	8	1.02	4.6	36.9	44.1	2.9	3.1	8.0
6	1.0	8	37.26	40.74	3	3	8	1.02	8.5	36.3	40.2	3.0	3.0	8.0
7	1.0	12	37.09	36.91	3	3	8	1.00	12.5	37.0	36.4	2.9	3.2	7.8
8	1.0	16	36.91	33.09	3	3	8	1.02	16.5	36.0	33.0	2.9	3.1	8.1
9	1.2	4	33.65	48.35	3	3	8	1.22	4.6	33.0	47.8	3.0	3.2	8.1
10	1.2	8	33.49	44.51	3	3	8	1.24	8.9	32.7	43.9	3.1	3.2	7.7
11	1.2	12	33.33	40.67	3	3	8	1.24	12.8	32.1	39.9	3.1	2.9	7.9
12	1.2	16	33.18	36.82	3	3	8	1.24	16.7	32.0	35.8	3.0	3.1	8.2

. The composition of “FeO” was achieved by mixing Fe and Fe₂O₃ powder, where the molar ratio of Fe and Fe₂O₃ was set to 1.0. On this basis, additional Fe₂O₃ was added to investigate the effect of Fe₂O₃ on the viscous behavior. The Fe/SiO₂ ratio in sample slags is defined as the ratio of total Fe (sum of divalent and trivalent iron) to SiO₂ content. The finely mixed samples were held in a molybdenum crucible and premelted at 1673 K for approximately 2 h under sufficient purging with 0.4 L/min Ar atmosphere. Subsequently, the liquid melt was poured into water for rapid quenching, and the bulk slag was then dried, crushed, and ground into powder for the next experiments.

2.2. Experimental Apparatus

In this paper, we used the rotating column method to determine viscosity. Figure 1a,b show the experimental equipment and the sizes of the molybdenum crucible and spindle, respectively. The molybdenum spindle can be rotated using a corundum shaft attached to a digital viscometer (DV2TRV, Brookfield Engineering Labs., Inc., Middleborough, MA, USA), which was mounted on a lift, allowing the spindle to be submerged and removed from the slag. To prevent slag’s oxidation at high temperatures, a high purity graphite sleeve was placed between the corundum tube and molybdenum crucible. The slags were melted in a vertical resistance furnace with six U-shaped MoSi₂ heating rods. A type B thermocouple (Pt−10 wt% Rh/Pt−13 wt% Rh) enclosed in a corundum sheath was placed directly below the crucible’s bottom and maintained within ±3 K of the target temperature in real-time using a PID controller. Before measurements, the viscometer was calibrated at room temperature with standard castor oils of varying concentrations.

2.3. Process for Viscosity Determination

For viscosity experiments, 170 g premelted slag was compacted and charged into a molybdenum crucible, which was placed on a graphite base in the uniform temperature zone of the furnace. The furnace was procedurally heated to 1673 K and held there for approximately 60 min. During the holding period, a fine molybdenum wire was inserted into the slag and stirred to accelerate the homogenization of the slag. Simultaneously, the molybdenum wire was further lowered to the bottom of the crucible and then immediately removed to confirm if the melt pool depth is approximately 40 mm. If not, an additional sample was added to ensure that the depth is appropriate. After the holding time, the spindle was gently dipped into the melt pool, eventually resting 10 mm above the crucible’s base. Then, the spindle is rotated to further mix and homogenize the slag. Throughout the experiment, 0.4 L/min of Ar was introduced into the furnace to prevent the slag’s oxidation.

In general, viscosity measurements were performed using a fixed-point test, i.e., holding the slag for a time at each temperature to obtain a uniform and stable slag [17,25,26]. In order to reduce the oxidation of Fe²⁺, shorten measurement time and obtain the relationship between viscosity and temperature over a wide range of intervals, viscosity measurements were conducted under continuous cooling cycles. Some other literature has also used continuous cooling methods for viscosity testing [10,27,28]. In the viscosity determination experiments, the resistance furnace was ramped up to 1673 K and then cooled with a rate of 3 K·min⁻¹. The viscosity was measured at a maximum temperature of 1673 K with different spindle speeds before the cooling procedure. The findings demonstrate that the viscosity varies by 2% with different rotational speeds, confirming that slag at high temperatures exhibits Newtonian viscous behavior. Thus, a constant rotational speed of 12 r min⁻¹ was chosen to measure the viscosity. The spindle rotation was stopped when the viscosity suddenly increased or reached the limit value of the instrument. In order to verify the reproducibility of the experiments, we selected samples numbered 1, 7, and 11 for repeat testing, and the results showed that the experiment was reproducible with an error of no more than 2%. After completing the viscosity test, the furnace was reheated to the initial maximum temperature of 1673 K. At this time, the spindle was raised from the melt and the slag was poured into water for quenching.

Figure 2 shows the X-ray diffraction (XRD; D8 ADVANCE, Bruker, Rheinstetten, Germany) profile of the quenched slag, which confirms the amorphous state of the sample and represents the structure of the slag at high temperatures. According to the results presented in Table 1, the chemical composition of the quenched samples reveled by X-ray fluorescence spectroscopy (XRF, ZSX Primus II, Rigaku Corporation, Tokyo, Japan) exhibits no significant changes before and after the experiment. It should be noted that the almost negligible MoO₃ content shown by XRF confirms that there is no significant chemical reaction between molybdenum and iron. Therefore, MoO₃ can be ignored during the viscosity measurements. In our experiments, the maximum measurement error of XRF did not exceed 0.5%. The classical K₂Cr₂O₇ titration method was used to determine the presence of Fe²⁺ in quenched slag. The Fe³⁺ content of the slag is calculated by subtracting the Fe²⁺ content from the total Fe content. In addition, the microstructure of the quenched slag was characterized through FTIR (VERTEX70, Bruker, Rheinstetten, Germany) and Raman (HR800, Jobin Yvon, Palaiseau, France) spectroscopy.

3. Results and Discussion

3.1. Effect of Fe/SiO₂ on the Breaking Temperature

The viscosity of the FeO−Fe₂O₃−SiO₂−CaO−MgO−Al₂O₃ fluxes versus temperature at fixed Fe₂O₃ contents for different Fe/SiO₂ ratios is plotted in Figure 3. The viscosity increases slowly with decreasing temperature, as predicted. When the temperature falls below a certain threshold, the viscosity rapidly increases, and an obvious inflection point appears on the viscosity-temperature curve, namely the breaking temperature (T_Bk) [29,30], which has been marked in the curves of Figure 3. It is noticed that at fixed Fe₂O₃ contents, the T_Bk value gradually decreases as the Fe/SiO₂ ratio increases from 0.8 to 1.2, and this trend is more pronounced at higher Fe₂O₃ content. This is due to the increase of Fe/SiO₂ makes the absolute amount of FeO in the slag increase. Slags containing high melting point substances are said to have higher breaking temperatures as well as greater crystallization ability [25]. FeO has a lower melting temperature (1642 K) than other oxides, allowing it to expand the area of solid–liquid coexistence. In addition, FeO can combine with SiO₂ to form the fayalite phase, which has a much lower melting point (1478 K). Specifically, Figure 4a–c give the results of cooling precipitation calculations for spinel, fayalite and clinopyroxene phases present in the slag at 1373–1473 K for an Fe₂O₃ content of 4 wt%, respectively. It can be seen that spinel, fayalite and clinopyroxene phases are present in the slag for different Fe/SiO₂ ratios (0.8, 1.0 and 1.2). With the increase of Fe/SiO₂ ratio, although the spinel phase slightly increases at this time, the fayalite phase with a lower melting point in the slag is the majority and gradually increases, whereas the content of clinopyroxene phase with a higher melting point gradually decreases, thus leading to the decrease of T_Bk.

The calculated cooling phase equilibrium of the slag at 1673–973 K (Fe/SiO₂ = 1.0, Fe₂O₃ = 8 wt%) is given in Figure 5. According to Figure 5, with cooling, the spinel phase began to precipitate at 1575 K, accounting for 0.04%. The iron olivine phase began to precipitate at 1465 K, accounting for about 0.41%, at which time the spinel precipitation had reached 5.89%. With the continuation of cooling, the pyroxene phase began to precipitate at 1403 K, accounting for 0.57%, at which time the spinel accounted for 9.35% and the iron olivine phase accounted for 22.92%. It can be seen that with the continuous cooling, the type and quantity of solids precipitated from the slag are increasing, thus causing the slag viscosity to rise continuously.

3.2. Effect of Fe/SiO₂ on the Slag Viscosity

The viscosity of the FeO−Fe₂O₃−SiO₂−8-wt% CaO−3-wt% MgO−3-wt% Al₂O₃ melt concerning Fe/SiO₂ ratio at 1523 and 1473 K is shown in Figure 6a. The viscosity of all slags decreases with increasing Fe/SiO₂ ratio at a given temperature, and this trend is more pronounced at low temperatures and Fe/SiO₂ ratios, whereas at high temperatures and Fe/SiO₂ ratios, this effect becomes less significant. For example, at a low temperature of 1473 K and Fe₂O₃ content of 16 wt%, the viscosity decreases remarkably by 0.75 Pa·s with increase in Fe/SiO₂ ratio from 0.8 to 1.0. However, when Fe/SiO₂ ratio continues to increase to 1.2, viscosity decreased by only a small amount of 0.31 Pa·s. At a higher temperature of 1523 K, the viscosity decreases by 0.42 Pa·s when Fe/SiO₂ increases from 0.8 to 1.0, which is weaker than that at 1473 K. Continue to increase Fe/SiO₂ to 1.2, the viscosity change is 0.27 Pa·s, which also became relatively weaker.

The iron–silicate slag contains several complex silicate network structures consisting of a large number of [SiO₄]⁴⁻ tetrahedral units connected by bridge oxygen. As Fe/SiO₂ ratio increases, the absolute amount of FeO in the slag increases. The decomposition of FeO generated a large amount of free oxygen (O²⁻), which will react with the bridge oxygen to produce nonbridge oxygen, causing the complex network structure to depolymerize. The disappearance of the network structure causes a reduction in slag viscosity. Conversely, as Fe/SiO₂ ratio increases, the absolute content of SiO₂ in the melt decreases, and the majority of the complex network structure simplify. As a result, a subsequent increase of the Fe/SiO₂ exhibits a weakened effect in reducing the viscosity. In addition, high temperatures can destroy the slag network structure because excess heat can supply sufficient energy to destroy the remaining complex network structure and thus, reduce the viscosity [31]. Therefore, at a high temperature of 1523 K, the increase in Fe/SiO₂ has no obvious effect on the depolymerization because, more silica–oxygen tetrahedral network structures are depolymerized relative to 1473 K.

Figure 6b shows the effect of Fe/SiO₂ ratio on the viscosity of the studied melt system, as calculated by Factsage 7.3. The predicted results are consistent with the variation trend of the measured results. As Fe/SiO₂ ratio increases, the viscosity decreases. However, the measured viscosity value is slightly higher than the predicted value, which may be due to the difference in oxygen potential, where some Fe²⁺ oxidizes to Fe³⁺ during the cooling process. According to the literature [18,19], Fe³⁺ is an amphoteric ion that can form both tetrahedral and octahedral structures; the tetrahedral structure is predominant and exists as a network polymer. When Fe²⁺ is oxidized to Fe³⁺, their coordination structure in the slag shifts from octahedral to tetrahedral, and the slag’s local cohesion increases, increasing the viscosity [32].

3.3. Effect of Fe/SiO₂ on the Slag Structure

In this section, FTIR and Raman spectra was analyzed to characterize the quenched slag structure. Figure 7a,b show the effects of Fe/SiO₂ ratio on the FTIR spectrum of the studied slag system with a fixed Fe₂O₃ content of 12 and 16 wt%, respectively. The FTIR spectrum of silicate slag typically falls between 1200 and 400 cm⁻¹. The [SiO₄]⁴⁻ tetrahedron has a symmetrical stretching vibrational band in the 1200–750 cm⁻¹ range [33]. The asymmetric stretching vibration band of the [AlO₄]⁵⁻ tetrahedral has been identified at around 620 cm⁻¹ [10]. The characteristic vibrational band at approximately 500 cm⁻¹ belongs to the Si−O symmetric bending vibration [24,31].

As visible in Figure 7a, the lower limit of the characteristic symmetric stretching vibration band of [SiO₄]⁴⁻ tetrahedra shifts from 778 to 755 cm⁻¹ with increasing Fe/SiO₂ ratio from 0.8 to 1.2 at 12 wt% of Fe₂O₃. In particular, the convoluted band’s width widens and intensity decreases, implying an increase in the spacing of Si and O in the [SiO₄]⁴⁻ structure and depolymerization of the slag structure. Furthermore, as Fe/SiO₂ increases, the groove depth of the Si−O vibration band at approximately 500 cm⁻¹ becomes shallower, revealing the disappearance of the complex silicate structure.

It is noteworthy that as Fe₂O₃ content increases, the position of the characteristic symmetric stretching vibration band of [SiO₄]⁴⁻ tetrahedra in Figure 7b shifts left (toward large wave number) relative to Figure 7a, i.e., the variation range of the lower limit of the band shifts from 778−755 cm⁻¹ at 12 wt% Fe₂O₃ to 790−764 cm⁻¹ at 16 wt% Fe₂O₃, indicating that Fe₂O₃ behaves as an acidic oxide and Fe³⁺ exists as a network former with tetrahedral coordination, making the silicate network structure more complex. Furthermore, there are no observable changes in the vibration bands of the [AlO₄]⁵⁻ tetrahedra at approximately 620 cm⁻¹ or the vibration band of the [FeO₄]⁵⁻ tetrahedra at approximately 690 cm⁻¹.

Raman spectra were obtained to further investigate the complex anionic silicate structures in slags. The deconvoluted peak plots of the Raman spectra are shown in Figure 8, and the results are listed in

Table 2. Deconvolution results of the characteristic peaks of Raman spectra of samples with varying Fe/SiO₂ ratios at Fe₂O₃ content of 8 wt%.

No.	Fe/SiO2	Q0	Q1	Q2	Q3	NBO/Si
1	0.8	0.02	0.19	0.43	0.36	1.87
2	1.0	0.04	0.26	0.43	0.27	2.07
3	1.2	0.12	0.32	0.32	0.24	2.32

. The silicate structural vibration peak can be seen as a large hump between 800 and 1200 cm⁻¹. The high content of Fe³⁺ in the slag allows the appearance of [FeO₄]⁵⁻ tetrahedral stretching vibration band at approximately 690 cm⁻¹ [34]. With reference to the research by Mysen et al. [35,36], the original Raman spectrum was processed using the Gaussian-deconvolution method with the minimum correlation coefficient r² ≥ 0.99. In this study, the concept of Qⁿ is adopted to represent the structural unit of the symmetric stretching band of [SiO₄]⁴⁻ tetrahedra, where n represents the number of bridging oxygen present in each [SiO₄]⁴⁻ tetrahedron. Qⁿ is generally available in silicate structures as Q⁰, Q¹, Q², and Q³ forms, which exists in the shape of monomers ([SiO₄]⁴⁻), dimers ([Si₂O₇]⁶⁻), chains ([SiO₃]²⁻), and sheets ([Si₂O₅]²⁻), respectively [37,38]. The integrated area of the Qⁿ can be used to calculate the number of structural units remaining in the melt. Changes in the average number of nonbridging oxygen (NBO/Si) can indicate variations in the silicate structure, and NBO/Si can be calculated roughly by multiplying the area fraction of Qⁿ by the amount of nonbridging oxygen present in Qⁿ, as shown in Formulae (1) and (2) [25,38,39]:

$$X_{Q^i}=A_{Q^i}/{{\displaystyle \sum }}_{i=0}^4{A}_{Q^i}$$

(1)

$$NBO/Si=4X_{Q^0}+3X_{Q^1}+2X_{Q^2}+X_{Q^3}$$

(2)

where $X_{Q^i}$ is the molar fraction of Qⁱ, $A_{Q^i}$ is the integrated area of Gaussian peak corresponding to Qⁱ, and NBO/Si is the average number of non-bridging oxygen per silicon.

It is speculated from Figure 8 and Table 2 that as Fe/SiO₂ ratios increases from 0.8 to 1.2 in the slag, the fractions of Q⁰ and Q¹ structural units increase and Q² and Q³ decrease. Similarly, the [FeO₄]⁵⁻ stretching vibrational band located at approximately 690 cm⁻¹ decreases with increasing Fe/SiO₂ ratio. Additionally, the number of NBO/Si increases with increasing Fe/SiO₂ ratio as shown in Table 2. These results show that as Fe/SiO₂ ratio increases, the degree of polymerization of the silicate network structure becomes more simplified, which is consistent with the changes in viscosity and FTIR results.

3.4. Effect of Fe₂O₃ on the Breaking Temperature

The variation of viscosity with temperature in the FeO−Fe₂O₃−SiO₂−8-wt% CaO−3-wt% MgO−3-wt% Al₂O₃ slag system at different Fe₂O₃ contents is given in Figure 9. The findings demonstrate that viscosity changes at different Fe₂O₃ contents follow a similar pattern and viscosity increases smoothly as temperature decreases. As previously stated, the slag’s viscosity varies significantly at low Fe/SiO₂ ratios, as shown in Figure 7. The previously reported viscosity data [12,24,40,41] for iron-silicate based slags are also given in Figure 9d. It can be seen that the data obtained in this study are in good agreement with those in the literatures. Since this study is a six-membered system of FeO−Fe₂O₃−SiO₂−8 wt% CaO−3 wt% MgO−3 wt% Al₂O₃, it is bound to differ from the results in the literature. In addition, the viscosity measurement is also related to parameters such as the material shape of the spindle and crucible. Therefore, the data in this study can be considered as reasonably reliable. Meanwhile, the T_Bk of the slag gradually increased with increasing Fe₂O₃ content at fixed Fe/SiO₂ ratios. The phase diagram of the FeO−Fe₂O₃−SiO₂−8 wt% CaO−3 wt% MgO−3 wt% Al₂O₃ slag system calculated by Factsage 7.3 is given in Figure 10, along with isotherms between 1473 K and 1673 K. According to the phase diagram, the studied slag components are in the crystalline region of the fayalite and spinel phase. With increasing Fe₂O₃ content, the initial crystalline phase of the slag gradually shifts from the low-temperature fayalite phase to the high-temperature spinel phase. Therefore, the T_Bk of the slag gradually increases with the addition of Fe₂O₃ content.

3.5. Effect of Fe₂O₃ on the Slag Viscosity

Figure 11a depicts the effect of Fe₂O₃ content on slag viscosity at various temperatures and fixed Fe/SiO₂ ratios. It can be seen that for Fe/SiO₂ ratios of 0.8 and 1.0, the slag viscosity shows a minimum value with the increase of Fe₂O₃ content. For example, at a Fe/SiO₂ ratio of 0.8, the viscosity decreases from 0.73 to 0.45 Pa·s as the Fe₂O₃ content increases from 4 to 12 wt%, and increases sharply to 1.04 Pa·s as the Fe₂O₃ content continues to increase to 16 wt%. Moreover, as Fe/SiO₂ ratio rises to 1.2, the slag viscosity follows a different trend, i.e., it increases with the addition of Fe₂O₃. This can be attributed to the amphoteric behavior of Fe₂O₃.When the Fe/SiO₂ ratio is relatively low, Fe₂O₃ behaves similarly to other basic oxides, which can decompose network-modifying O₂⁻ at high temperatures and react with the bridge oxygen in the slag; thus, depolymerizing the slag and lowering the slag viscosity. When Fe/SiO₂ ratio is high, Fe₂O₃ exhibits the properties of acidic oxides, such as SiO₂, and the [FeO₄]⁵⁻ tetrahedra forms a more complex network structure with the [SiO₄]⁴⁻ tetrahedra, making the slag more aggregated with increased slag viscosity.

In addition, for low Fe/SiO₂ ratio, the absolute amount of SiO₂ in the slag is high and the slag structure is relatively complicated. At this time, Ca²⁺ acts as a charge compensator particularly when Ca²⁺ is near [FeO₄]⁵⁻ tetrahedra which are not fully supplied for slags at low basicity due to the scarcity of Ca²⁺. Consequently, more [FeO₆]⁹⁻ octahedra would be formed to replace Ca²⁺ as the network modifier, depolymerizing the silicate structure and lowering slag viscosity. Moreover, the bond energy of the Fe−O bond is lower than that of the Si−O bond. Therefore, even if Fe³⁺ exists in the form of tetrahedra, the bond energy of the entire system will decrease with the increase of Fe₂O₃. At the low Fe/SiO₂ ratio, when the Fe₂O₃ content exceeds 12 wt%, the slag viscosity increases. This is due to the relatively high melting point of Fe₂O₃, which improves the crystallization ability of the slag and increases the slag viscosity. In addition, analogous to the action of Fe/SiO₂ ratio, the effect of Fe₂O₃ on viscosity becomes weak at high temperatures.

Figure 11b shows the slag viscosity predicted using Factsage 7.3, which is consistent with the experiments at high Fe/SiO₂ ratio. However, no amphoteric behavior of Fe₂O₃ was observed at low Fe/SiO₂ ratios according to the software calculations, which may be due to the deviation of the viscosity model used by the software from the actual measurement results.

3.6. Effect of Fe₂O₃ on the Slag Structure

Figure 12a,b show the FTIR spectra at different Fe₂O₃ contents for fixed Fe/SiO₂ ratios of 0.8 and 1.2, respectively. It illustrates that the vibrational band of the silicate tetrahedra becomes wider firstly and then narrower with the increase of Fe₂O₃ content at an Fe/SiO₂ ratio of 0.8, and the depth of the corresponding convolution band becomes shallower and then deeper, and this result is consistent with the change of viscosity. Similarly, the same trend is observed for the height of the Si−O bending vibration band. For the slag with Fe/SiO₂ ratio of 1.2, the width of the vibrational band of [SiO₄]⁴⁻ tetrahedra gradually narrows and the height of the groove increases with increasing Fe₂O₃ content, and the symmetric Si–O bending vibration band also becomes pronounced with Fe₂O₃ addition, which is consistent with the viscosity measurement results. Moreover, similar to the effect of Fe/SiO₂ ratio, the variation of [AlO₄]⁵⁻ tetrahedral vibration band and [FeO₄]⁵⁻ with the Fe₂O₃ content is not significant.

Figure 13 depicts the deconvolution results of Raman spectra with characteristic structural unit vibrations in an iron–silicate slag system at various Fe₂O₃ contents and a fixed Fe/SiO₂ ratio of 1.2. It is suggested that with the addition of Fe₂O₃, the Q⁰ and Q¹ units decrease and the Q² and Q³ units increase, as indicated by the variation of the peak heights in the Raman spectra. Similarly, as Fe₂O₃ content increases, the stretching vibration band of [FeO₄]⁵⁻ at 690 cm⁻¹ increases, suggesting that with the addition of Fe₂O₃, a gradual polymerization occurred within the silicate network, which is in consonance with the FTIR and viscosity measurements.

3.7. Activation Energy for Viscous Flow

The temperature dependence of the slag viscosity can be expressed by the Arrhenius equation as shown in Formula (3):

$$\eta =A\exp \left(\frac{E_{\eta }}{RT}\right)$$

(3)

where η is the viscosity, Pa·s; A is the pre-exponential constant; E_η is the activation energy for viscous flow, J·mol⁻¹; R is the universal gas constant, 8.314 J·(mol·K)⁻¹, and T is the absolute temperature, K. The changes in E_η can provide a reference for variation in friction of viscous flow and can express the slag structure changes [42].

Figure 14 depicts plots of the natural logarithm of viscosity (lnη) versus the inverse of temperature (1/T) of all slags above the breaking temperature with varying Fe₂O₃ contents and Fe/SiO₂ ratios. It is evident that for all studied compositions, lnη and 1/T are linearly correlated with a linear correlation coefficient r² > 0.99, indicating that molten slags follow the Arrhenius behaviors. Thus, the E_η values are derived from the slopes of the lines using the Arrhenius formula and are shown in

Table 3. E_η values at different Fe/SiO₂ ratios and Fe₂O₃ contents.

Fe/SiO2 Ratio	Fe2O3 Content/wt%	Activation Energy/(kJ·mol−1)
0.8	4	142.16
	8	136.77
	12	131.79
	16	141.51
1.0	4	138.14
	8	129.56
	12	127.97
	16	136.25
1.2	4	109.47
	8	112.39
	12	112.63
	16	132.18

. It should be noted that the viscous activation energies derived in the experiments were obtained at a rate of cooling of 3 K/min. As expected, the activation energy decreases with increasing Fe/SiO₂ ratio from 0.8 to 1.2 at a fixed Fe₂O₃ content and increases with increasing Fe₂O₃ contents at the Fe/SiO₂ ratio of 1.2. Moreover, the changes in E_η with Fe₂O₃ additions to the slag system at fixed Fe/SiO₂ ratios of 0.8 and 1.0 revealed a distinctive V-shaped pattern. The variation tendency of the E_η is consistent with the measured viscosity values and is supported by changes in slag structure, i.e., the smaller the E_η, the simpler the silicate structure of the slag. The values of the activation energy are similar to the results reported by Park et al. [24] who investigated the FeO_t−SiO₂−Al₂O₃ copper-smelting slags, where the E_η is 118 and 129 kJ/mol with the addition of Al₂O₃ is 0 and 5 mass pct, respectively, at a fixed Fe/SiO₂ ratio of 1.27.

4. Conclusions

The viscous behaviors of FeO−Fe₂O₃−SiO₂−8 wt%CaO−3 wt%MgO−3 wt%Al₂O₃ slag system were investigated by the rotating spindle method. The slag structure with different Fe/SiO₂ ratio and Fe₂O₃ content was explored by FTIR and Raman analysis. The main conclusions obtained are shown below:

(1) The slag viscosity and activation energy for viscous flow decreased with increasing Fe/SiO₂ ratios (0.8–1.2) and increased with increasing Fe₂O₃ contents (4–16 wt%) at the Fe/SiO₂ ratio of 1.2. Slag viscosity at low Fe/SiO₂ ratio (0.8 and 1.0) exhibited a minimum value by increasing Fe₂O₃ content in slag. Fe₂O₃ shows the amphoteric behaviors in slag system.

(2) The T_Bk of the slag decreased with increasing Fe/SiO₂ ratios. With the addition of Fe₂O₃ content, the primary crystalline phase of the slag gradually changes from the low-temperature fayalite phase to the high-temperature spinel phase, resulting in an increase in the T_Bk of the slag.

(3) The degree of polymerization of present slags decreases with increasing Fe/SiO₂ ratio and increases with the addition of Fe₂O₃ content at the Fe/SiO₂ ratio of 1.2. A decrease in Fe/SiO₂ ratio or increase in Fe₂O₃ content promoted the formation of [FeO₄]⁵⁻ tetrahedra, which further complicates the silicate network structure.