Influences of Al₂O₃ and TiO₂Content on Viscosity and Structure of CaO–8%MgO–Al₂O₃–SiO₂–TiO₂–5%FeO Blast Furnace Primary Slag

Abstract

In view of the fact that Ti–bearing blast furnace primary slag has been explored limitedly and its viscosity–structural property is not fully understood, the phase compositions, viscosity and structure of CaO–8%MgO–Al₂O₃–SiO₂–TiO₂–5%FeO slag are investigated by X-ray diffractometer, rotating cylinder method, Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy respectively, considering the effect of Al₂O₃ and TiO₂. The critical temperature that is defined as the temperature below which the viscosity of slag increases quickly, could be explained by the relative amount of perovskite to melilite from phase compositions analysis. The slag viscosity first increases with increasing Al₂O₃ content from 10 to 15 mass%, and then decreases with the further increase of Al₂O₃ to 18 mass%. Increasing TiO₂ content continuously lowers the viscosity. FTIR and Raman spectra results show that increasing Al₂O₃ or decreasing TiO₂ content leads to complex Si–O and Ti–O networks structure, corresponding to the slag viscosity variation. The effect of weak linkages of Si–O–Al is more dominant when Al₂O₃exceeds 15 mass%, which results in the decrease of viscosity.

1. Introduction

The blast furnace (BF) smelting of Ti–bearing iron ore occupies an appreciable proportion in hot metal production because of abundant titanomagnetite resources in China [1,2]. In addition, ironmakers use the burden containing TiO₂ to protect the hearth refractory due to the formation of titanium carbide and nitride for prolonging the furnace campaign life [3,4,5]. The softening–melting of Ti–bearing materials in BF generates the flowable primary slag whose formation could be also regarded as a process of interaction and reorganization of various phase components [6]. The viscosity of Ti–bearing primary slag could influence many aspects, such as permeability of cohesive zone, mass and heat transfer through metallurgical kinetics and metal–slag separation. The Ti–bearing primary slag assigned to CaO–MgO–Al₂O₃–SiO₂–TiO₂–FeO system with high basicity and a significant amount of FeO has different viscous behaviour to the slag in hearth. Recently, the usage of low-cost and high-alumina iron ore is gradually increased because of continuous consumption of the ores rich in iron [7,8]. Al₂O₃ content in the slag may reach a certain level [9] and the relative amount of TiO₂ changes. Therefore, it is important and necessary to understand the effects of Al₂O₃ and TiO₂ on phase compositions and viscosities of Ti–bearing primary slag fundamentally.

The viscosity of CaO–MgO–Al₂O₃–SiO₂–TiO₂ system has been investigated by many researchers [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24].There has been a consensus among the studies [11,12,13,14,15,16,17,18] that TiO₂was a reducing viscosity agent in the quinary slag with fixed MgO, Al₂O₃ and basicity. For the effect of binary basicity, most experts reported that viscosity decreased with increasing basicity [17,18,19,20,21,22]. There was also a different view that higher basicity led to higher slag viscosity in the high alumina and medium titania BF slag [23]. Considering the influences of MgO and Al₂O₃, it was believed that the addition of MgO content can decrease viscosity [13,20,21]. The viscosity increased with increasing Al₂O₃ content in the studies [21,22,23]. Zhen et al concluded that viscosity of Ti–bearing slag with constant CaO, SiO₂ and MgO content increased as Al₂O₃/TiO₂ ratio increased [24]. In the case of CaO–MgO–Al₂O₃–SiO₂–TiO₂–FeO slag, work done by Jiao et al showed that viscosity decreased with increasing TiO₂ or FeO content [25,26]. Hence, the major studies on Ti–bearing slag are relevant to final slag in BF hearth. Although the practical BF process relates to Ti–bearing primary slag, there is comparatively little data available on its viscosities, especially for the influences of Al₂O₃ and TiO₂.Therefore, the investigation of primary slag is required to enrich and improve the theory of Ti–bearing slag smelting.

In this work, CaO–MgO–Al₂O₃–SiO₂–TiO₂–FeO system slag with 8 mass% MgO, 5 mass% FeO and CaO/SiO₂ of 1.3, which is related to Ti–bearing primary slag, is synthesized by melting chemical reagents in a Mo crucible at 1823 K with more than 3 h under Ar flow using the viscosity measurement furnace. The effects of Al₂O₃ and TiO₂on viscosity and phase compositions are investigated via rotating method and X-ray diffractometer (XRD), respectively. Furthermore, the slag structure is explored by Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy for better understanding viscous behaviour.

2. Material and Methods

2.1. Preparation of SlagSamples

Several analytically pure chemical reagents, which include TiO₂, MgO, Al₂O₃, SiO₂, CaCO₃, and FeC₂O₄·2H₂O, are used for preparing the primary slag sample. CaCO₃ and FeC₂O₄·2H₂O are consumed for producing CaO and FeO, respectively. All of the materials are dried and weighed according to the designed compositions shown in

Table 1. Chemical compositions of designed Ti–bearing primary slags (mass%).

No.	Designed						Analyzed
	FeO	CaO	MgO	Al2O3	SiO2	TiO2	FeO
01	5.00	37.87	8.00	10.00	29.13	10.00	4.62
02	5.00	36.74	8.00	12.00	28.26	10.00	4.43
03	5.00	35.04	8.00	15.00	26.96	10.00	4.71
04	5.00	33.35	8.00	18.00	25.65	10.00	4.57
05	5.00	41.26	8.00	12.00	31.74	2.00	4.28
06	5.00	39.00	8.00	12.00	30.00	6.00	4.46
07	5.00	34.48	8.00	12.00	26.52	14.00	4.59

. The mass ratio of CaO to SiO₂ is fixed at 1.30. In each component of No.1~No.4, contents of FeO, MgO and TiO₂ keep constant while Al₂O₃ content increases from 12 to 18 mass%. The other variable is TiO₂ content, changing from 2 to 14 mass%. Apart from FeC₂O₄·2H₂O, other oxides are well mixed in a zirconia ball mill, then pelletized for later use.

2.2. Viscosity Measurement

The viscosity measurement is conducted by a melt physical property comprehensive testing equipment (RTW–10, Northeastern University, Shenyang, China) shown in Figure 1. Mo spindle and crucible are selected touching the slag. Mo crucible is installed in a high–purity graphite crucible for better protecting the slag. The thermocouple near the crucible is used for monitoring the slag sample temperature which should follow the furnace temperature. The programmable viscometer is designed for viscosity measurement at the given shear rates and it is calibrated with castor oil of known viscosities at different temperatures before experiments. The known viscosities of castor oil are 0.986, 0.651 and 0.451 Pa·s, corresponding to 20, 25 and 30 °C, respectively. During the calibration, the castor oil is filled in the Mo crucible with the same sizes for slag viscosity measurement until the depth is 40 mm.

FeC₂O₄·2H₂O is first put into the Mo crucible (40 mm inner diameter and 80 mm depth). As temperature rises, FeC₂O₄·2H₂O graduallydecomposes, softens and melts. Meanwhile, the pelletized ball mentioned in Section 2.1 is carefully added. The furnace temperature is heated to 1823 K, held for more than 3 hours, and then drops to 1773 K. At the moment, the rotating spindle is soaked in molten slag and stirs the slag for 30 min. About 140 g slag (~40 mm depth) is eventually obtained for viscosity measurement on cooling style from 1773 K at every 25 K interval. The equilibration time is 25 min at each temperature. After viscosity measurement, the primary slag is reheated to 1773 K and kept for 60 min, and then poured into water, dried, and crushed by a disk mill for chemical compositions analysis and next phases and structure investigation. In the whole process, a constant high purity Ar gas flow (0.5 L/min) is introduced for protection. The FeO content in every slag sample analyzed by the titration method is given in Table 1, showing little change in FeO content and good protection during experiment.

2.3. Phase Compositions and Structure Investigation

Two parts (each about 5 g) of the post-experimental slag powder are re-melted at 1773 K and kept for 60 min under Ar gas flow using small Mo crucibles formed by punching (29 mm inner diameter, 4.5 mm depth and 0.1 mm thickness). One is slowly cooled to the ambient temperature with 10 K/min in the furnace and examined by XRD (SmartLab SE, Rigaku, Tokyo, Japan) for phases investigation. The other is rapidly quenched in water for structure analysis using XRD, FTIR (Thermo Scientific Nicolet IS5, Nicolet, Madison, WI, USA) and Raman spectroscopy (XploRA PLUS, Horiba Scientific, Edison, NJ, USA). Potassium bromide (KBr) tablet method is adopted to attain FTIR transmitting spectra. The ratio of sample to KBr is 1:150. The spectrum for each slag sample is an average of 32 scans, which deducts the spectrum of blank KBr tablet. Raman spectroscopy is performed on samples by a laser confocal micro-Raman spectrometer, of which the excitation wavelength is 532 nm. The recorded spectral range is 100–2000 cm⁻¹.

3. Results and Discussion

3.1. The Critical Temperatures

Figure 2 shows the viscosity variation of slag containing different Al₂O₃ and TiO₂ content against temperature. As expected, viscosity decreases with increasing temperature. It is noteworthy that the additions of Al₂O₃ and TiO₂ result in various critical temperatures of slags. The critical temperature (CT) is defined as the temperature below which the slag viscosity increases quickly and marked as the dotted circle. A low critical temperature represents a wide thermostable operation region, which is advantage to actual production. The highest CT with changing Al₂O₃ content is observed at 15 mass% Al₂O₃, after which the CT drops dramatically with further rise of Al₂O₃. On the other hand, with increasingTiO₂ content, the CT first decreases, and then increases.

XRD analysis of the slowly cooled, slag samples with varying Al₂O₃ and TiO₂ content are shown in Figure 3. The phases in the Ti–bearing primary slags are melilite and perovskite. The diffraction peak intensities of melilite and perovskite change with adding Al₂O₃ and TiO₂. For the XRD pattern of each slag, the backgrounds is determined and subtracted through Highscore Plus Software (Version 3.0e, PANalytical, Almelo, The Netherlands). Subsequently, the strongest peaks of perovskite and melilite are found and their intensities are recorded as I_P and I_M, respectively. I_P/I_M can be calculated and listed in Figure 3. I_P/I_M is defined as the ratio of the strongest diffraction intensity of perovskite and melilite, indicating the relative amount of perovskite to melilite. From Figure 3a, as Al₂O₃ content increases, I_P/I_M first increases to the maximum value at 15 mass% Al₂O₃, and then decreases on the whole. Figure 3b shows that when TiO₂ content increases, I_P/I_M first decreases and then increases, exhibiting a minimum value at 6 mass% TiO₂. The melting temperature of perovskite is 1970 °C. Comparatively, the melting temperature of melilite is much lower. It has been reported that the slag with more content of high–melting–point substance has the higher melting temperature and stronger crystallization capacity, which would result the higher CT [13,19,27,28,29,30]. Therefore, the variation of CT could be explained by the investigation of phase compositions.

3.2. Effects of Al₂O₃ and TiO₂ on Viscosity

Figure 4 shows the effects of Al₂O₃ and TiO₂ content on viscosity at various temperatures. As depicted in Figure 4a, the slag viscosity initially increases, reaches the maximum at 15 mass% Al₂O₃ and again drops afterwards with increasing Al₂O₃ content. Obviously, the effect of Al₂O₃ is more slight at higher temperature. This is because the relative wide spacing and weak interionic attraction between cations and anions. In such a slag system with much basic oxides, there are sufficient cations for charge compensation [27], which is beneficial to the existence of [AlO₄]⁵⁻–tetrahedrons. [AlO₄]⁵⁻ is believed to be a network former increasing the complexity of networks structure. Therefore, the increase of viscosity could be attributed to the acid characteristics of Al₂O₃, similar to the work by Feng et al that showed the Ti–bearing slag viscosity increased with increasing the Al₂O₃ content from 11 to 15% [22]. However, the Al₂O₃ content more than 15 mass% results to a decrease in the viscosity. Al₂O₃ incorporates into the silicate slag and forms Si–O–Al linkages. The Si–O–Al linkages are weaker than Si–O–Si bonds, which means that the migration of structural units will be easy if excessiveAl₂O₃ is added. This effect may be more dominant to some degree in the Al₂O₃ content exceeding 15 mass% within the slags, resulting the viscosity to be decreased. It should be further confirmed by spectra analysis.

From Figure 4b, when the temperature is lower than 1698 K, TiO₂ content exceeding 6 mass% will increase the viscosity. This is because of the formation of solid perovskite speculated from phase compositions. Above 1698 K, the viscosity decreases by adding TiO₂, which supports that TiO₂ is a reducing viscosity agent. It is proposed that TiO₂ may be a basic oxide and could break the slag networks structure in some reports [11,13,15,26]. The decrease of viscosity with increasing TiO₂ could be attributed to the simplification of networks structure, which will be certified by structure analysis.

The viscosity variation with compositions at high temperature could be interpreted via the viscosity activation energy (E_η) simply, which represents the energy barrier for viscous flow running and indirectly suggests a change in the structure. According to the well-know Arrhenius law, the natural logarithm of viscosity against reciprocal of temperature is described in Figure 5. E_η is calculated from the linear slope, listed in

Table 2. Viscosity activation energy of slags.

No.	Factors	Al2O3 (mass%)	TiO2 (mass%)	Eη (kJ/mol)
01	Al2O3	10	10	92.9
02		12	10	85.3
03		15	10	140.4
04		18	10	116.4
05	TiO2	12	2	87.2
06		12	6	80.7
02		12	10	85.3
07		12	14	70.6

. It is observed that the variation of E_η is an increasing trend but the maximum value of E_η is achieved at 15 mass% Al₂O₃. This is basically in agreement with the above viscosity trend. For increasing TiO₂ content, E_η exhibits a decreasing trend overall. This suggests more simple structural units with low flowing energy barrier generate in the slags, contributing the decreased viscosity.

3.3. Structure Analysis by FTIR and Raman

The XRD results of rapidly quenched slags at 1773 K are shown in Figure 6, indicating that there is no characteristic peaks of minerals. The diffraction peak due to Mo is caused by incomplete separation between rapidly quenched slag and Mo crucible. It is proved that the samples are amorphous and homogeneous and maintain the high temperature state.

The rapidly quenched slags are performed on FTIR spectroscopy to analyze the structure changes with different Al₂O₃ and TiO₂ content qualitatively. As shown in Figure 7, the bands of 1200–750 cm⁻¹ belongs to the [SiO₄]⁴⁻–tetrahedral symmetric stretching which includes four kinds of Si–O networks structural units. The bands at ~850, ~940, ~980 and ~1030 cm⁻¹ are assigned to Q⁰, Q¹, Q² and Q³ (corresponding to non-bridging oxygen per silicon NBO = 4, 3, 2, 1), respectively [31]. In addition, the peaks at 730–630 cm⁻¹, 570–520 cm⁻¹ and ~500 cm⁻¹ are assigned to the asymmetric stretching vibration of [AlO₄]⁵⁻–tetrahedral (network former), [AlO₆]⁹⁻–octahedra (network modifier) and the Si–O–Al rocking, respectively [26,32].

In Figure 7a, by increasing Al₂O₃ content, the center of the [SiO₄]⁴⁻–tetrahedral bands shifts to higher wavenumbers from about 945 to 962 cm⁻¹, resulted from the decrease of Q⁰ and Q¹ and the increase of Q². This indicates that higher Al₂O₃ contentis likely to polymerize Si–O networks. For Al–O networks, the trough of [AlO₄]⁵⁻–tetrahedral has little change. No peaks of [AlO₆]⁹⁻–octahedral are found. This means that Al₂O₃ is certified only to be a network former in the slags, playing an active role in increasing viscosity. The depth of the Si–O–Al rocking does not change with the addition of Al₂O₃ content up to 15 mass%, whereas it decreases with higher Al₂O₃ content. The decrease in the trough of Si–O–Al rocking suggests that the linkage between [SiO₄]⁴⁻ and [AlO₄]⁵⁻–tetrahedrals becomes weaker, which can be the reason for decreasing viscosity in Figure 4a. As can be noted from Figure 7b, with increasing TiO₂ content, the bands of the [SiO₄]⁴⁻–tetrahedral symmetric stretching vibration becomes shallower and its center shifts to lower wavenumbers slightly. No significant variation of the trough of [AlO₄]⁵⁻–tetrahedral is found. However, the depth of the Si–O–Al rocking trough decreases continuously. It is confirmed that TiO₂ behaves as a basic oxide providing free oxygen ions in these slags. As TiO₂ content increases, more O²⁻ ions react with bridged oxygen and depolymerize the silicates, which results in the negative center shift of Si–O stretching bands and the weaker Si–O–Al linkages. This leads to decreasing viscosity as shown in Figure 4b. The FTIR analysis also seems to suggest that TiO₂ prefers to modify the Si–O networks rather than Al–O networks, which is similar to the work done by Park [15]. Ti–O networks structure is difficult to be identified in FTIR spectra and it will be clarified by Raman analysis.

The slag structural units with changing Al₂O₃ and TiO₂ content are further studied by Raman spectroscopy. The original Raman spectra of the rapidly quenched slags at 1773 K are shown in Figure 8. It is obviously discovered that the strong bands locate the frequency region of 600–1100 cm⁻¹ in every spectra curve and change with compositions. The bands of 710~810 cm⁻¹ become more intense relative to the bands at about 860 cm⁻¹. This tendency with changing TiO₂ is much more pronounced, which presents as the position of the highest peak shifting to lower wavenumbers. The bands at 790–830 cm⁻¹ were associated with Ti–O stretching vibrations in TiO₄⁴⁻ monomer which is a simple and small structural unit [19,33]. Additionally, the bands corresponding to O–(Ti, Si)–O deformation vibrations were proposed at 700–750 cm⁻¹ [33,34], which are suggested to be complex structural units [19,24]. The Raman peaks of Al–O stretching vibrations were reported at 530–610 cm⁻¹ [35,36]. It is found that the bands of 530–610 cm⁻¹ change little with varying both Al₂O₃ and TiO₂ content, which is consistent with the FTIR results. Hence, it is believed that the bands between 710 and 810 cm⁻¹ relate to Ti–O networks, such as O–(Ti, Si)–O and TiO₄⁴⁻. The Si–O networks have been successfully studied on the basis of the Qⁿ concept. As mentioned in FTIR analysis, Qⁿ includes Q⁰, Q¹, Q² and Q³, which are corresponding to silicate forms of [SiO₄]⁴⁻, [Si₂O₇]⁶⁻, [SiO₃]²⁻ and [Si₂O₅]²⁻ with the structure of monomers, dimers, chains and sheets, respectively. According to the previous works [32,37,38,39], these silicate units pertain to the Raman peaks at 840~860 cm⁻¹, 900~920 cm⁻¹, 960~1000 cm⁻¹ and 1050~1100 cm⁻¹, respectively. Considering the characteristics of Raman curves of the slags, there are little Q³ units in the present slags.

It is necessary to deconvolute the Raman spectra ranged from 600–1100 cm⁻¹ for obtaining detailed information. First, the backgrounds of Raman spectra is determined and subtracted. Then, based on the proposed Raman peaks of each units, the Raman spectra is deconvoluted by the Gaussian–Deconvolution method with correction coefficient more than 0.99. In current study, ~730 cm⁻¹, ~800 cm⁻¹, ~860 cm⁻¹, ~910 cm⁻¹ and ~1000 cm⁻¹ are assigned to O–(Ti, Si)–O deformation, stretching vibrations of TiO₄⁴⁻, Q⁰, Q¹ and Q², respectively. The typical deconvolution of the Raman spectra is shown in Figure 9. The area of Ti–O networks structural units and Qⁿ can be obtained. The amount of Qⁿ is calculated by Formula 1. Raman scattering coefficient (θ_n) has been summarized in several studies [40,41]. The average number of non-bridging oxygen per silicon (NBO/Si) is employed to represent the polymerization degree of Si–O networks and calculated by the mole fraction and the number of non-bridge oxygen of Qⁿ. The calculation of NBO is expressed as Formula 2. Lower value of NBO/Si, higher polymerization degree of Si–O networks.

$${\mathrm{X}}_{\mathrm{n}}={\mathrm{A}}_{\mathrm{n}}{\mathsf{\theta }}_{\mathrm{n}}/{\displaystyle \sum }_{\mathrm{n}=0}^3{\mathrm{A}}_{\mathrm{n}}{\mathsf{\theta }}_{\mathrm{n}}$$

(1)

where X_n, A_n and θ_n is the mole fraction, Raman band area and Raman scattering coefficient of Qⁿ (n = 0~3).

$$\mathrm{NBO}/\mathrm{Si}={\displaystyle \sum }_{\mathrm{n}=0}^3\left(4-\mathrm{n}\right){\mathrm{X}}_{\mathrm{n}}$$

(2)

The abundance variations of structural units with increasing Al₂O₃ and TiO₂ content are shown in Figure 10. It can be seen from Figure 10athat as Al₂O₃ content increases, the percentage of O–(Ti, Si)–O deformation vibrations increases while that of TiO₄⁴⁻ stretch vibrations slightly decreases, which is an indication of more complex Ti–O networks. NBO is decreased continuously, suggesting polymerization of Si–O networks. In Figure 10b, with the addition of TiO₂ content, both O–(Ti, Si)–O and TiO₄⁴⁻ in the slags increase. Notably, the increase of TiO₄⁴⁻ is much greater than O–(Ti, Si)–O. Meanwhile, NBO of the slag with more content of TiO₂ is higher. The facts indicate that TiO₂ acts as a basic oxide and simplifies Si–O and Ti–O networks. There are some consistency between FTIR and Raman analysis, correlating well with the results of viscosity.

4. Conclusions

The evolutions of viscosity and structure of CaO–MgO–Al₂O₃–SiO₂–TiO₂–FeO system slag with C/S = 1.3, 8 mass% MgO and 5 mass% FeO, which is relevant Ti–bearing primary slag in blast furnace, are studied by a series of methods. Major findings are concluded as follows:

(1).

The change of critical temperatures can be attributed to the variation trend of melting temperature and crystallization capacity, which is able to be explained by the relative amounts of basic phases including perovskite and melilite.

(2).

The viscosity of the slag containing 10 mass% TiO₂ first increases and then decreases with increasing Al₂O₃ content from 10 to 18 mass%, exhibiting the maximum value at 15% Al₂O₃, while an increase in TiO₂ content from 2 to 14 mass% causes a decrease in the viscosity of the slag at 12 mass % Al₂O₃.

(3).

FTIR and Raman analysis confirm that Si–O and Ti–O networks are more complex with increasing Al₂O₃ or decreasing TiO₂ content, causing higher viscosity. The dampening of Si–O–Al trough observed in FTIR indicates a decrease in the linkagebetween [SiO₄]⁴⁻ and [AlO₄]⁵⁻ tetrahedrals, which may be the reason for the lower viscosity with excess Al₂O₃ content.