Effect of TiO2 on the Viscosity and Structure of Mineral Wool Melt from Blast Furnace Slag

The effect of TiO2 on the viscous behaviours of mineral wool melts containing CaO −SiO2−Al2O3−MgO-TiO2 were studied at acidity coefficient Mk in the range of 1.0–1.6 and TiO2 content in the range of 0–6wt%. Based on the experimental data, the critical temperature and activation energy were calculated to analyse the changes in viscosity flow. The morphology and phase of slag were clarified by scanning electric microscopy with energy dispersive spectroscopy and X-Ray diffraction. Fourier transformation infrared and Raman spectra were used to obtain information about the variations in melt structure. The results showed that the viscosity decreased with increasing TiO2 content. The critical temperature (Tcr) was 1537–1645K, and it was relatively stable when the acidity coefficient Mk is 1.3. The activation energy decreased from 182.27, 178.48, 162.75 kJ/mol to 146.67, 139.28, 127.64 kJ/mol with increasing acidity coefficient Mk, respectively. With increasing TiO2 content, the fraction of main crystal phases (CaSiO3) decrease and form perovskite (CaTiO3). In addition, relationships are found to exist between the viscosity and structure of melt. The formation of Ti-O or O-Ti-O bonds weakens the stability of molten slag, and the number of bridging oxygen decreases from 1.85 to 0.65. frequencies represent the variations in bridging oxygen (BO) bond contents of m(Si, Al, Ti)-O-n(Si, Al, Ti) and the content of the Si-O-Si tetrahedra structural unit ( ). The characteristic peak of the original Raman spectra in the high-frequency region (800–1200cm -1 ) shifts from 983cm -1 to 885cm -1 with the increasing TiO 2 content. This indicates that the simple silicate structure units gradually

It is well known that viscosity is affected by both temperature and material composition, and the temperature during the production process is relatively stable, but the raw material composition is not identical. Of the raw materials of mineral wool (metallurgical slag, basalt, and others) [5,6], blast furnace (BF) slag is widely used as its chemical composition (primarily composed of SiO 2 , CaO, Al 2 O 3 and MgO) is similar to the materials in mineral wool. However, a certain amount of TiO 2 is contained in blast furnace slag due to the addition of titanium ores and titanium-containing pellets during BF protection operations, which influences the viscosity of the mineral wool melts in the production process and further affects the quality of the mineral wool fibres. Therefore, it is critical to study the effect of TiO 2 on the viscosity of mineral wool melt.
Several studies have reported the effects of TiO 2 on the physicochemical properties of metallurgical slag. Yan et al. [7] reported that the viscosity of the quinary system CaO-SiO 2 -8 wt%MgO-14 wt%Al 2 O 3 -TiO 2 with basicity (C/S = CaO/SiO 2 ) in the range of 0.5-1.3 and found TiO 2 decreased the viscosity of a quinary slag system. Saito et al. [8] revealed that the apparent activation energy of viscosity flow decreased and the structural units became smaller with increasing TiO 2 content in CaO-SiO 2 -Al 2 O 3 slag system. Gao et al. [9] proposed that the viscosity increased with the addition of TiO 2 in the range of 23.46 wt%-26.45 wt%, which is the result of the formation of high melting point perovskite. The viscosity experiments conducted by Park et al. [10] indicated that TiO 2 acted as a basic oxide and depolymerized complex silicate sheets into simpler structures in CaO-SiO 2 -10 wt%MgO-17 wt%Al 2 O 3 slag. Work by Jiao et al. [11] found TiO 2 lowered the viscosity of CaO-SiO 2 -11.26 wt%Al 2 O 3 -9.13 wt%MgO-TiO 2 -FeO slag and weaken the strength of the silicate network. Chang et al. [12] studied the effect of TiO 2

Apparatus used for viscosity measurements
The slag viscosity measurement was conducted using the rotating cylinder method. The experimental apparatus is shown in detail in Fig. 2. The U-shaped MoSi2 heating element in the furnace was used to heat the apparatus. The furnace tube was composed of corundum and the experimental temperatures were controlled by a Pt-6%Rh and Pt-30%Rh thermocouple inserted into the furnace.
The measurement error was less than ±2 K. When the rotating shaft was working, the torsion wire generated a twist angle due to the viscous action of the slag. The computer then converted the twist angle into a time difference signal and the slag viscosity values were calculated and recorded.

Experimental procedure
The viscosities of the slags were measured as follows. The pre-melted slag (110g) was held in a Mo crucible. The Mo crucible was placed in a constant temperature zone of the reaction chamber in the resistance furnace. The slag was heated to 1773K at the rate of 7K/min in Ar atmosphere and kept for 120 min to homogenize the composition of the slag before the spindle was immersed into the fluid slag. The viscosity measurements were performed during the cooling cycle with an equilibration time of 10 min at each temperature to ensure sufficient thermal equilibration. The measurements were also conducted above the critical temperature of the melt. The critical temperature for the CaO-SiO 2 -12wt%Al 2 O 3 -8wt%MgO-2.2wt% TiO 2 slag is shown in Fig. 3. When the temperature was lower than 1593K, the value of the natural logarithm of the viscosity (lnη) increased significantly, indicating that the large solid phase began to precipitate, and the corresponding temperature was the critical temperature (T cr ). The viscometer was calibrated using castor oil of a known viscosity before each viscosity measurement. Following the viscosity measurements, the samples were heated to the critical temperature and kept for 90 min. The samples were then quenched on a water-cooled copper plate to confirm the morphology and element distribution using SEM-EDS (VEGA 3 XMU/XMH, Tescan, Czech Republic) and XRD, respectively.

Effect of TiO 2 on viscosity of slag
The viscosities of the mineral wool melts based on CaO-SiO 2 -Al 2 O 3 -MgO-TiO 2 at varying temperatures with different acidity coefficients Mk are shown in Fig. 4. The viscosity of the melt decreased with the addition of TiO 2 at a fixed acidity coefficient Mk of 1.0, as shown in Fig. 4(a). However, the viscosity increases with decreasing temperature and the reduced effect of TiO 2 on slag viscosity at a higher temperature. This was because the amount of excess thermal energy sufficiently modified the intricate network structure of the slag, and the effect of TiO 2 was relatively weaker than that of high temperature. The results were in good agreement with some previous studies. Sohn et al. [13] and Xu et al. [14] concluded that the complex network structure is depolymerised to simpler structure with  [15][16][17]. Fig. 4(b) shows that the increase in TiO 2 content lowers the viscosity of melt, which is consistent with the various trends in the acidity coefficient Mk at 1.0. The viscosity of slag with a greater acidity coefficient Mk is relatively high at the same temperature. Although the high temperature weakens the influence of TiO 2 on melt, the effect of TiO 2 on decreasing viscosity is still more obvious than for a lower acidity coefficient Mk. This indicates that the temperature of 1733 K is insufficient to modify the network structure completely. Fig. 4(c) shows clearly that the influences of TiO 2 and temperature on the viscosity of slag are basically constant. However, with a further increase in temperature to 1773K, the reductive effect of TiO 2 on the viscosity is still obvious. This may be because the mole fraction of complex silicate and aluminate unites increases with an increase in the acidity coefficient, and the slag structure tends to polymerise; however, the high temperature cannot completely destroy the network structure. TiO 2 still has a strong effect on the decrease in viscosity.

Effect of TiO 2 on critical temperature
The critical temperature (T cr ) is defined as the temperature at which the viscosity increases sharply or the slag behaves like non-Newtonian fluids in the cooling cycle [18][19][20]. In addition, slag with high acidity coefficient has higher T cr , and it is relatively stable when the acidity coefficient Mk is 1.3. Therefore, the effect of TiO 2 on the T cr of melt is weak at an acidity coefficient Mk of 1.3, and TiO 2 has little effect on the process of mineral wool production.

Activation energy for viscous flow
All the viscosities of the slag were measured above the critical temperature. This means that all the molten slags were in the fluid region, and the melt behaved as Newtonian fluids. Thus, the Arrheniustype equation was applied to express the temperature dependence of the viscosity η: Eq. (1) can be taken logarithmically and written as the following.
where η, A, E η , R, and T are the viscosity, pre-exponential factor, apparent activation energy of the slag, gas constant (8.314J/mol·K), and absolute temperature, respectively.
The apparent activation energy represents the viscous flow barrier, which can be calculated from the relationship of 1/T and ln(η), and the variation reflects the change in slag structure.

Area scanning analysis and phase compositions
To understand the slag microstructure at the critical temperature, back-scattered electron images of

Effect of TiO 2 on slag structure using FTIR spectroscopy
The structure of slag affects viscosity, and the formation of complex network structure leads to an increase in viscosity. Fig. 10 represents the FTIR absorption spectra of all the quenched slags.

Effect of TiO 2 on the slag structure using Raman spectroscopy
FTIR spectra are used in the qualitative analysis of slag structures, whereas a more definitive quantitative analysis is necessary to provide clear information on the variations in the changes of the structural units. Thus, in this study, Raman spectroscopy was used to quantifiably analyse the changes in these units. In previous studies [28,[31][32][33][34], the band in the region of 800-1200 cm -1 is deconvoluted into four typical peaks that label with the presence of tetrahedrally coordinated silicon (Q 0 , Q 1 , Q 2 , Q 3 ) located at 850-880cm -1 , 900-920cm -1 , ~1000cm -1 , and ~1050cm -1 , respectively. In addition, the bands centred at 650-670cm -1 and near 725cm -1 appear in the low-frequency region, as shown in Figs.   12(b) and (c). The corresponding Raman bands for the structural units are provided in Table 2.  From the work of Frantz and Mysen [36], the relationship between the mole fractions of the structural units and the relative areas for the deconvoluted bands are expressed by the following relation: where X i is the mole fraction of species, θ i is the Raman scattering coefficient, and A i is the band area of the species.  Table 3. It can be seen clearly that the content of Q 0 gradually increases, whereas Q 2 decreases with the increase in TiO 2 content. The fractions of Q 1 and Q 3 increase first with increasing of TiO 2 content from 0wt% to 2wt% and then decreases with the further addition of TiO 2 from 2wt% to 6wt%. This assumes that the Q 2 depolymerises to Q 1 or Q 3 , which can be expressed by the following equation.
It should be noted that the percentage of the complex network structure (Q 2 and Q 3 ) gradually decreases and the simple structure (Q 0 and Q 1 ) increases with TiO 2 content from 2wt% to 6wt%, which further proves that TiO 2 weakens the strength of the silicate structure and decreases the DOP of mineral wool melt.

Conclusion
The viscosity of mineral wool melt was measured under Ar atmosphere in fully liquid region, and the slag structure was verified using FTIR and Raman spectra. Good agreement was obtained among viscosity, critical temperature, activation energy, and changes in slag structure. The present study can be summarised as follows: 1. TiO 2 as a network modifier plays a role of depolymerisation in mineral wool melt, and the viscosity of slags decreases with increasing TiO 2 content from 0wt% to 6wt%.