Effect of TiO₂ on the Liquid Zone and Apparent Viscosity of SiO₂-CaO-8wt%MgO-14wt%Al₂O₃ System

Abstract

The effect of TiO₂ on the liquids zone and apparent viscosity of SiO₂-CaO-8wt%MgO-14wt%Al₂O₃ system were studied in the present work. At fixed CaO/SiO₂ between 0.5 and 1.3, higher TiO₂ content decrease the slag viscosity indicating that TiO₂ additions up to 50 wt.% behaved as a viscosity-decrease agent by loosening the silicate network structure. The free running temperature increase at TiO₂ content from 10 wt.% to 30 wt.%. At fixed TiO₂ content of 20, 30 and 40 wt.%, increasing the CaO/SiO₂ resulted in lower viscosity due to the depolymerization of the structure. Four different viscosity models were discussed and two of them were employed to predict the viscosity and found that the Urbain’s Model agrees well with experimental data at high viscosity (4–12 dPa·s) and the KCC’s Model agrees well with experimental data at a lower viscosity (0–4 dPa·s).

1. Introduction

Cost issues and decrease of high grade raw materials has sparked the interest in using of other sources (such as vanadium-titanium magnetizes (VTM)) of iron bearing materials which were regarded as uneconomical before. In the blast furnace process of using VTM, most of the iron and part of vanadium can be reduced into the hot metal. However, almost all of the titanium enters into the slag in phase formation of perovskite and titanaugite, Ti-rich diopside, forming the high titanium slag with the content of TiO₂ varying from 22 to 25%.¹⁾ Therefore, the work for the influence of TiO₂ on the viscosity of slag has been continued.

Figure 1 demonstrates the effect of TiO₂ additions within a broad compositional range on the quaternary and quinary calcium-alumino-silicate slag systems under inert atmosphere. Ohno and Ross,²⁾ Park et al.,³⁾ Liao et al.,⁴⁾ and Sohn et al.⁵⁾ revealed that TiO₂ acts as a basic oxide resulting in the depolymerization of the slags with higher concentrations of TiO₂ at a fixed (wt% CaO)/(wt% SiO₂) ratio and Al₂O₃ content at 1773 K. The viscosity experiments conducted by Saito et al.⁶⁾ indicated that TiO₂ decreases the viscosity in quaternary CaO–SiO₂–Al₂O₃–TiO₂ slag system. Shankar et al.⁷⁾ found that TiO₂ up to 2 wt% lower the viscosity in CaO–SiO₂–MgO–Al₂O₃. Handfield et al.⁸⁾ described industrial high TiO₂ slags are very fluid melts once completely molten. However, Ohno and Ross,²⁾ and Qiu et al.⁹⁾ reported that TiO₂ can increase the viscosity of the CaO–SiO₂–Al₂O₃– TiO₂– (MgO) slag under reducing atmosphere indicating that it could be caused by the formation of solid particles of TiC and also be caused by the polymerization of orthosilicate ions. Xie et al.¹⁰⁾ indicated that slag thickening occurred under prolonged reduction when a considerable amount of both TiO (3–7 wt%) and TiC (1–3 wt%) was formed, particularly in acidic slags.

Fig. 1.

Effect of TiO₂ on the viscosity of CaO–SiO₂–Al₂O₃–MgO based slags at various (wt% CaO)/(wt% SiO₂) ratio at 1773 K (1500°C).

TiO₂ has been approved as a viscosity-decrease agent in the blast furnace slag under inert atmosphere both by experimental and structure calculation. However, the validity of the above conclusion in a broader compositional range of CaO–SiO₂–Al₂O₃–MgO slag system is not approved. Therefore, the object of this work is to approve the validity of the conclusion that TiO₂ decrease the viscosity of the slag. In other word, the upper limit of TiO₂ content in the slag should be given in which the TiO₂ act as a viscosity reducer.

2. Experimental

The rotating cylinder method was used in the present study. The experimental arrangement including the dimensions of the crucibles and spindle have been provided elsewhere.^11,12) A rotating viscometer (Brookfield, model LVDV-II+ Pro, Middleboro, MA) set on the furnace was connected with a computer to record the value of the torque from the rotation of the spindle. The heater of the furnace is made of MoSi₂. The slag sample was loaded in the molybdenum (Mo) crucible which was put into a corundum crucible, with a graphite crucible in the outer layer for removing the left oxygen in the furnace. The rotating bob was made of Mo. The viscometer was regularly calibrated before the experiment using three standard silicone oils of known viscosities (0.96, 4.92 and 9.80 dPa·s) at 298 K (25°C).

The slag for the viscosity measurement was synthetic slag, which were synthetized with the known chemical composition slag from the analysis pure grade regent. the analysis pure grade regent were first claimed at 1273 K (1000°C) in a muffle furnace to remove all the impurities such as carbonate, hydroxide and moisture, and then weighed. The weighted chemicals (CaO, SiO₂, Al₂O₃, MgO and TiO₂) were pressed into cylindrical pellets before experiments, and about 200 g sample in the Mo crucible to make sure the molten bath at least 40 mm deepth.

The liquid region of the CaO-SiO₂-14wt%Al₂O₃-8wt%MgO-TiO₂ slag system at 1723 K (1450°C) was calculated by FactSage to help design the experimental scheme. FACT oxide database was selected to support the result. The chemical compositions of the slags investigated in the present study are presented in Table 1. The slag used for the viscosity measurement was synthetic slag obtained from analytically pure reagents of a known composition. The reagents were first calcined at 1273 K (1000°C) in a muffle furnace to remove all of their impurities, such as carbonate, hydroxide, and moisture. The pretreated powders were combined according to the desired proportions and then mixed homogenously in a mixer for 30 min.

Table 1. Chemical compositions of the studied slags.

*C/S=CaO(wt.%)/SiO2(wt.%)
No.	Al2O3	MgO	TiO2	CaO	SiO2	C/S*
A1	14.00	8.00	0.00	44.09	33.91	1.30
A2	14.00	8.00	10.00	38.43	29.57	1.30
A3	14.00	8.00	20.00	32.78	25.22	1.30
A4	14.00	8.00	30.00	27.13	20.87	1.30
A5	14.00	8.00	40.00	21.48	16.52	1.30
A6	14.00	8.00	50.00	15.83	12.17	1.30
B1	14.00	8.00	0.00	40.86	37.14	1.10
B2	14.00	8.00	10.00	35.62	32.38	1.10
B3	14.00	8.00	20.00	30.38	27.62	1.10
B4	14.00	8.00	30.00	25.14	22.86	1.10
B5	14.00	8.00	40.00	19.90	18.10	1.10
B6	14.00	8.00	50.00	14.67	13.33	1.10
C1	14.00	8.00	0.00	26.00	52.00	0.50
C2	14.00	8.00	10.00	22.67	45.33	0.50
C3	14.00	8.00	20.00	19.33	38.67	0.50
C4	14.00	8.00	30.00	16.00	32.00	0.50
C5	14.00	8.00	40.00	12.67	25.33	0.50
C6	14.00	8.00	50.00	9.33	18.67	0.50

The mixed powders was heated to 1798 K (1525°C) and held for 3 h to homogenize the slag. The bob was lowered and the upper surface of the bob was inserted 5 mm deep into the slag layer. The spindle was carefully adjusted to ensure that it was placed on the central axis line of the crucible because experimental errors can be easily caused by slight deviations of the spindle from the central axis. Argon was blown into the furnace chamber at a constant gas flow rate of 1.5 L/min to avoid oxidation of the molybdenum crucible and spindle. After the slag was held at 1798 K (1525°C) for 3 h, the spindle began to rotate at a fixed speed of 12 rpm. The viscosity was recorded after the reading had stabilized at the desired temperature. Measurements were carried out at every 25 K from 1798 K (1525°C) to 1673 K (1400°C). The equilibration time for viscosity measurements was 30 min, and viscosity measurements are reported as the average value of data collected 2 min after the 30 min thermal equilibrium was achieved. The standard errors of viscosity were found to be less than ±0.02 dPa·s. When the measurement was finished, the slag was reheated to 1773 K (1500°C). Then the viscosity was measured again at 10 seconds intervals with decreasing the temperature continuously at a rate of 2 K (°C)/min, as a result, the viscosity-temperature curves were got. After completing the viscosity measurements, slag samples were reheated to 1773 K (1500°C) and quenched on a water-cooled copper plate. The composition of the samples were analyzed after the experiment by X-Ray Fluorescence spectroscopy to confirm the composition and found that the measured values are very close to the designed chemical compositions. Besides, the quenched samples also used to check the slag structure using X-Ray Diffraction.

3. Results and Discussion

3.1. Effect of TiO₂ and Basicity on Viscosity

Figure 2 represents the influence of TiO₂ on the viscosity with varying basicity. TiO₂ reduced the viscosity, especially for the slag with lower basicity at 0.5, supporting that TiO₂ is a network-modifying oxide. As a network-modifying oxide, TiO₂ is expected to depolymerize the slag network structure. When basicity was increased to 1.3, the viscosity firstly decrease by adding TiO₂, however, it will increase when TiO₂ content reach to 20–30 wt%, the reason for the increase of viscosity is the formation of solid phase near the liquids line as shown in Fig. 3. Hence, the results of calculation does not agree well with the experimental values. At a high basicity of 1.3 with TiO₂ content form 20 wt% to 30 wt%, the liquid phase line of TiO₂ content near 20 wt% should move to the place where the TiO₂ content near 30 wt%. At the basicity of 0.5, the TiO₂ content more than 50 wt% should locate out the liquid phase region. The generation amount of perovskite (CaTiO₃) calculated by FactSage is shown in Fig. 4(a). With increasing of basicity, the amount of perovskite increases and the precipitation temperature of the perovskite becomes higher. The XRD results of as-quenched melts are provided in Fig. 4(b). The XRD studies were carried out to identify the different mineral phases in the sample. At a basicity of 0.5, the mineral phases identified is rutile. At a basicity of 1.3 and TiO₂ content of 20 wt%, the XRD pattern shows no characteristic peaks, suggesting that the sample were amorphous. However, the mineral phases identified is perovskite with increasing content of TiO₂ to 30 wt%, demonstrating again the reason for the increase of viscosity.

Fig. 2.

Effect of TiO₂ on the viscosity with varying basicity.

Fig. 3.

Liquid region of the slag system calculated from FactSage at 1723 K (1450°C) and modification with experimental values.

Fig. 4.

The generation amount of perovskite calculated by FactSage (a) and XRD result of as-quenched CaO–SiO₂–14wt%Al₂O₃–8wt%MgO-20wt%&30wt%TiO₂ system (b).

Viscosity sensitivity with temperature (thermal stability) is used to describe the viscosity variation with temperature. The so-called short slag means its viscosity change occurs in a narrow interval of temperature and bad thermal stability than the long slag does. Figure 5 shows the effect of TiO₂ on the viscosity-temperature curves of the slags with varying basicity. It is could be found that with basicity at 1.1, the viscosity changed dramatically at TiO₂ from 20 wt% at a certain temperature, but the change at 40 wt% with basicity at 0.5. The critical point agrees well with the precipitation temperature of the perovskite. Therefore, the free run temperature which means the critical point of the curves could be got by drawing a tangent line of the slopes is −1 on the curve. It is evident from Fig. 6 that the free running temperature increase with TiO₂ from 10 wt% to 30 wt%, which means that poor viscosity thermal stability with TiO₂ content increasing caused by the precipitation. However, the free running temperature decrease when TiO₂ content at 40 wt%.

Fig. 5.

Effect of TiO₂ on the viscosity-temperature curves of the slags.

Fig. 6.

Effect of TiO₂ on the free running temperature of the slags.

The viscosity data can be described by an Arrhenius equation over the entire temperature region in this study.   

$$\eta ={\eta }_{0}exp\left(\frac{E_{\eta }}{RT}\right)$$(1)

where η₀, E_η, R, and T are a constant, the apparent activation energy of viscous flow, the universal gas constant, and absolute temperature, respectively. Variations of E_η can reveal the change of frictional resistance for viscous flow and suggest a change in the structure of the molten slag or more directly a change in the cohesive flow units forming the slag structure. Generally, Arrhenius equation should be applied for temperature range where fully liquid state (no precipitation) is assumed. So the E_η value was determined from the viscosity data of fully liquid temperature range. The viscous activation energy were got and plot in the Fig. 7. The change of viscosity and viscous energy show the same regularity indicating that the main factor is the same as impact viscosity and viscous activation energy. It should be noted that, the dash line is an assumption. There is no adequate viscosity data of fully liquid to calculate the viscous activation energy when basicity at 1.3 with TiO₂ content form 20 wt.% to 30 wt.%.

Fig. 7.

Effect of TiO₂ on the apparent activation energy of the slag.

The influence of basicity on the viscosity with TiO₂ form 10 wt% to 30 wt% are showing in Fig. 8. It is obvious that the lower the basicity, the higher the viscosity. CaO is a well-known typical basic oxide that can modify the melt structure effectively. The addition of CaO can result in a large-scale de-polymerization with lower basicity. However, the increase of CaO will result in precipitation of perovskite. Therefore, to get appropriate viscosity with high TiO₂ content, the maximum of basicity must be less than 1.3.

Fig. 8.

Effect of basicity on the viscosity with different TiO₂ content.

3.2. Models and Prediction of Viscosity

A large amount of data are available on the viscosity of blast furnace slag. A summary covering part of the published viscosity models suitable for use in metallurgical applications and the specific chemical and metallurgical systems to which they have been applied is given in Table 2. Riboud et al.¹³⁾ classified the slag components into five different categories, depending on their ability to break or form polymeric chains in the molten slag. The Iida’s Model¹⁴⁾ is based on the Arrhenius type of equation and divides all the oxides into acidic oxides, basic oxides and amphoteric oxides, where using the basicity index to consider the network structure of the slag. TiO₂ was assigned into acidic oxides which increase the viscosity in both of the above models and diagree for this study. The Urbain’s Model is one of the most widely used slag viscosity models and assumes Weymann-Franke relation.¹⁵⁾ In the model the slag constituents are classified into three categories: glass former, glass modifiers and amphoterics. KCC’s Model¹⁶⁾ is structural based, and the temperature dependence of viscosity is calculated by Arrhenius equation. In Urbain’s and KCC’s model, TiO₂ act as a network modifying oxide and depolymerize the structure.

Table 2. Classification of slag viscosity models.

T-dpendence	Name	Details of model	Comments/ Validity
Weymann-Frenkel	Riboud13)	AW; BW functions of 5 groups “CaO”+“SiO2”+“Al2O3”+“CaF2”+“Na2O” Where X“CaO” contains XCaO +XMgO +XFeO +XMnOetc. ln A=−19.81+1.73 XCaO+3.58 XCaF2+7.02 XNa2O−35.76 XAl2O3 B=31140−23896 XCaO−46356 XCaF2−39159 XNa2O−68833XAl2O3	Different sets of coefficients for each ternary system, Works well for variety of slags
Arrhenius + basicity index	Iida14)	η(Pas)=Aη0 exp (E/Bi) where η0 is viscosity of hypothetical network-forming melt and Bi=basicity index E=11.11−3.65×10−3 T and a=1.745−1.962×10−3T+7×10−7 T2 and Bi=∑(αi%i)B/∑(αi%)A where A=acid oxides and B=basic oxides or fluorides η=1.8×10−7 (M=i Tm)0.5exp (Hi/ RTm)/V)m0.667exp (Hi/ RTm) where Hi=5.1 Tm and values of α given	Requires use of basicity index
Weymann-Frenkel	Urbain15)	AW; BW functions of 3 groups” Glass formers: XG=XSiO2+XP2O5 Amphoterics XAl=XAl2O3+XB2O3+XFe2O3 Modifiers: XM=XCaO+XMgO+XNa2O+3 XCaF2+XFeO+XMnO+2XTiO2 BW=B0+B1XG+B2XG2+B3XG3 and Bi=αI+biα+ciαi2 ln AW=0.2693BW+11.6725	This model is simple and is applicable to a wide range of slags
structurally based + Arrhenius	KCC16)	ln η=ln A+E/RT; lnA=k(E−572516)−17.47 k=∑(xikJ)/∑xi; (i,j≠SiO2) E=(572516×2)/(nOSi+αAlnOAl+∑αinOi+∑αAl,inOAl,i+ ∑αiSinOiSi+∑αjAl,inOjAl,I+αPnOP+TinOTi+αFenOFe+αCaF2nO CaF2	Used in various industrial processes

Figure 9(a) shows the comparison between the measured viscosity and the estimated ones for CaO–SiO₂–MgO–Al₂O₃–TiO₂ slags, from which it can be seen that KCC’s Model and Urbain’s Model also did a disappointed prediction for the viscosity (0–12 dPa·s)with high average deviations more than 28%. However, the Urbain’s Model agrees well with experimental data at high viscosity (4–12 dPa·s). The significant deviation at lower viscosity (0–4 dPa·s) may be due to the Urbain’s Model is based on silicate slag system and over simplification of the basic oxides. Increasing the basicity and the TiO₂ content will result the decreasing of SiO₂ content, on the one hand, and decrease of the viscosity, on the another hand; and the slag will transform into titania-based. The KCC’s Model agrees well with experimental data at a lower viscosity (0–4 dPa·s). The KCC’s Model is a structurally based and overestimate the TiO₂ as a structure modifier.

Fig. 9.

Comparison between measured viscosity (a) and estimated viscosity and the iso-viscosity diagram at 1723 K (1450°C) (b).

The iso-viscosity distribution diagram can be obtained by these two methods, as shown in Fig. 9(b), the red region with appropriate viscosity (<10 dPa·s) is the requirement of smooth BF operation. The viscosities less than 4.0 dPa·s were calculated by KCC’s Model, and the viscosities more than 4.0 dPa·s were calculated by Urbain’s Model. The red line present liquid area which calculated from thermodynamic software FactSage with phase diagram module, what’s more each curve describes the distribution of the viscosity related to the slag composition. The diagram is specified for the slag where the viscosity increase with TiO₂ content and basicity decrease.

4. Conclusions

In present work, viscosity of the quinary system CaO–SiO₂–MgO–Al₂O₃–TiO₂ were experimentally measured. The increase of TiO₂ additions and basicity lower the viscosity in liquid phase under inert atmosphere. The precipitated phase of CaTiO₃ has a significant effect on the apparent viscosity. The Urbain’s Model agrees well with experimental data at high viscosity (4–12 dPa·s) and the KCC’s Model agrees well with experimental data at a lower viscosity (0–4 dPa·s).

5. Future Work

Research regarding the viscosity and structure of high-temperature slag bearing high TiO₂ will continue to be an important topic. Model to predicate the viscosity of slag systems bearing high TiO₂ will continue to evolve. The role of TiO₂ in the molten structure need more evidence to determine whether it break the Si–O–Si bend as CaO or jus modify the structure through Ti⁴⁺ substitute Si⁴⁺ or the TiO₂ octahedron exists in isolation. In order to understand the slag structure deeply, further analyzed techniques (NMR, FTIR, XPS etc.) should be used.

Acknowledgement

This study was supported by the State General Program of National Natural Science of China (Grant No. 51374263).

References

• 1)   Z.  Zuotai,  W.  Guanghua,  T.  Ping and  S.  Seetharaman: ISIJ Int., 48 (2008), 739.
• 2)   A.  Ohno and  H.  Ross: Can. Metall. Q., 2 (1963), 259.
• 3)   H.  Park,  J. Y.  Park,  G. H.  Kim and  I.  Sohn: Steel Res. Int., 83 (2012), 150.
• 4)   J. L.  Liao,  J.  Li,  X. D.  Wang and  Z. T.  Zhang: Ironmaking Steelmaking, 39 (2012), 133.
• 5)   I.  Sohn,  W.  Wang,  H.  Matsuura,  F.  Tsukihashi and  D. J.  Min: ISIJ Int., 52 (2012), 158.
• 6)   N.  Saito,  N.  Hori,  K.  Nakashima and  K.  Mori: Metall. Trans. B, 34 (2003), 509.
• 7)   A.  Shankar,  M.  Görnerup,  A. K.  Lahiri and  S.  Seetharaman: Metall. Trans. B, 38 (2007), 911.
• 8)   G.  Handfield and  G. G.  Charette: Can. Metall. Q., 10 (1971), 235.
• 9)   G.  Qiu,  S.  Ma,  Q.  Deng,  X.  Lv and  H.  Wang: Metal. Int., 17 (2012), 94.
• 10)   D.  Xie,  Y.  Mao and  Y.  Zhu: VII Int. Conf. on Molten Slags Fluxes and Salts, The South African Institute of Mining and Metallurgy, Johannesburg, (2004), 28.
• 11)   Z.  Yan,  X.  Lv,  J.  Zhang,  Y.  Qin and  C.  Bai: Can. Metall. Q., (2015), DOI: 10.1080/00084433.2015.1126903.
• 12)   Z.  Yan,  X.  Lv,  D.  Liang,  J.  Zhang and  C.  Bai: Metall. Trans. B, (2015), DOI: 10.1007/s11663-016-0676-z.
• 13)   P.  Riboud,  Y.  Roux,  L. D.  Lucas and  H.  Gaye: Fachber. Huttenprax. Metallweiterverarb., 19 (1981), 859.
• 14)   T.  Iida,  H.  Sakai,  Y.  Kita and  K.  Shigeno: ISIJ Int., 40 (2000), S110.
• 15)   G.  Urbain: Steel Res. Int., 58 (1987), 111.
• 16)   G.  Zhang,  K.  Chou and  K.  Mills: Metall. Trans. B, 45 (2014), 698.