2014 Volume 54 Issue 4 Pages 985-989
The effect of Al2O3/TiO2 ratio on viscosities of CaO–MgO–Al2O3–SiO2–TiO2 melts was investigated by the rotating cylinder method in this study. In addition, structural characterizations of these quenched vitreous samples were also studied by Raman spectroscopy. It was indicated from the experimental results that viscosity increases as gradually increasing Al2O3/TiO2 ratio while keeping the contents of other components constant. The Raman spectra analyses indicated that TiO2 mainly exists in the form of [TiO4] as a network former in composition range of 3–17 mol%. With increasing the Al2O3/TiO2 ratio, the [TiO4] content decreases and the degree of polymerization of the melt increases resulted from the increase of Al2O3 which behaviors as an acidic oxide and incorporates into the SiO2 network with the charge balance of CaO. Consequently, there will be an increase of viscosity with increasing Al2O3/TiO2 ratio.
It is well known that viscosity is an important physical property for aluminosilicate melt. Considerable attentions have been paid on viscosity for its great importance in understanding the fluid dynamic of molten slags and slag-metal reaction kinetics during the pyrometallurgy process.
CaO–Al2O3–SiO2 system is a very important fundamental system during the iron-making and steelmaking processes, so it is meaningful to study the viscosities of the CaO–Al2O3–SiO2 based slags. Titanium is a minor element in most rock-forming silicate melts. Nevertheless, this element is petrologically significant. For example, liquidus boundaries between olivine and pyroxene shift toward olivine-rich portions of appropriate systems as TiO2 is added to the melt.1) Ohno and Ross2) found TiO2 additions increase the slag viscosity in the CaO–SiO2–Al2O3–TiO2 slags under reducing atmospheres of C/CO equilibrium mainly because of the reducing the TiO2. Shankar et al.3) revealed the effect of TiO2 in the CaO–SiO2–MgO–Al2O3 slag system, where the viscosity decreased with TiO2 up to 2 mass%. Satio et al.4) found TiO2 lowered the viscosity in the CaO–SiO2–MgO–Al2O3 slag system as adding 10 mass% and 20 mass% TiO2. Sohn5) investigated the influence of TiO2 from 0 to 10 mass% on the viscous behavior of CaO–SiO2–17 mass%Al2O3-10 mass%MgO slags. Both Al2O3 and TiO2 are amphoteric oxides, so it is significant to distinguish their different effects on viscosity. However, there is no related study on the influence of Al2O3/TiO2 ratio. Therefore, in this work, the viscosity variation of CaO–MgO–Al2O3–SiO2–TiO2 molten slag with Al2O3/TiO2 ratio will be studied. Furthermore, the Raman spectroscopy analyses will also be used to illustrate the structure change corresponding to the viscosity variation.
Slag samples were prepared using reagent-grade SiO2, Al2O3, MgO, TiO2, and CaCO3 powder, all of which were calcined in a muffle furnace at 1273 K for 10 hours, to decompose any carbonate and hydroxide before use. The detailed information of the chemical reagents was shown in Table 1. Then the prepared CaO and other reagents were precisely weighted according to the compositions shown in Table 2, and then mixed in the agate mortar thoroughly. In each composition of Table 2, contents of CaO, MgO and SiO2 keep constant, but Al2O3/TiO2 ratio gradually increases. The mixtures were packed into a Mo crucible and pre-melted in an induction furnace at 1873 K for 3 hours with the protection of Ar gas. After pre-melting, the slag sample, together with the crucible, was preserved in a desiccator.
The viscosity measurement was carried out using the rotating cylinder method. The schematic diagram of the experiment apparatus can be found in our previous paper.6) During the viscosity measurement, both the crucible and the spindle should be properly aligned along the axis of the viscometer, which is very important because a slight deviation from the axis can cause large experimental errors. Then, the furnace was heated up to about 1823 K and held for 60 min under the protection of Ar gas, before immersing the spindle into the slag and measuring the viscosity. The viscosity measurement was carried out at every 10 K or 25 K interval on cooling. At each experimental temperature before measuring, the melt was kept for 30 min first to ensure the melt uniform. The variations of viscosity due to the different rotating speeds range from 100 to 200 rpm were less than 1.5%, confirming that the molten was Newtonian fluid. The average value was adopted as the viscosity value. All the measured viscosities were given in Table 3. After completing the viscosity measurements, the furnace was reheated up to 1823 K to pull out the spindle, which was cleaned for the next experiment.
Glass samples with different compositions were prepared by conventional melting and quenching method. Raw materials were mixed, put into a platinum crucible and then melted at 1873 K for 4 hours in Ar atmosphere. After that, the melts were quenched by water to obtain the glass samples which were proved to be amorphous by XRD as shown in Fig. 1.
Typical XRD pattern of the quenched glass sample.
Raman spectra were acquired using a laser confocal Raman spectrometer (JY-HR800, Jobin Y’von, France). Its precision of wave number is better than 0.01 cm–1. The experiments were carried out at room temperature by using excitation wavelength of 532 nm. The light source was a semiconductor laser with power of 1 MW. The frequency band measured in this work was ranged from 100 to 2000 cm–1.
For four compositions in Table 2 with the same contents of CaO, MgO and SiO2 but different contents of TiO2 and Al2O3, the temperature dependences of viscosity are shown in Fig. 2, from which it can be seen that the Arrhenius law is obeyed and the viscosity decreases as increasing the temperature. Furthermore, the viscosity increases as increasing the ratio of Al2O3/ TiO2.
Variations of viscosity with the ratio of Al2O3/TiO2 at different temperatures.
Figure 3 presents the room-temperature Raman spectra of the four glass samples, respectively. All the backgrounds of the measured Raman signals have been subtracted. It is found that the peak of spectra of the glasses become broad and shift to higher frequency from composition A1 to A4.
Raman spectra for samples with different molar ratio of Al2O3/TiO2 at room temperature, after background subtraction.
All the measured Raman spectra are deconvolved by Gaussian-Deconvolution method with the minimum correction coefficient ≥ 0.999. The squares of the deviations between the observed and calculated Raman envelopes with Gaussian line shapes,7) χ2, are also given. The deconvoluted results have been shown in Figs. 4, 5, 6, 7. From all the spectra in the figures, obvious bands are found in the frequency of 600–1200 cm–1, so the focus of attention in this work is the middle frequency of the Raman spectra. Assignments of Raman peaks have been listed in Table 4. As shown in Figs. 4, 5, 6, 7, there exist four bands in the middle frequency range as 700–728 cm–1, 820–843 cm–1, 940–966 cm–1, 1010–1040 cm–1, respectively. Area fractions of various bands could be calculated according to the deconvolved spectra data. Figure 8 shows the area fractions of bands as functions of x (Al2O3)/x (TiO2).
Deconvoluted result of Raman spectra for sample A1.
Deconvoluted result of Raman spectra for sample A2.
Deconvoluted result of Raman spectra for sample A3.
Deconvoluted result of Raman spectra for sample A4.
Effects of Al2O3 contents on the structure of silica network.
The structure of TiO2-free silicate has been subjected to many studies1,8,9,10,11,12,13,14,15,16) by Raman spectroscopy, that the peak at about 948 cm–1 is due to SiO32– stretching with NBO/T = 2 (non-bridging oxygen per tetrahedrally coordinated cation) and referred to as Q2 (subscript refers to the number of bridging oxygen) species in a chain structure.13,17,18) The 820–843 cm–1 band probably corresponds to the vibration of Ti–O–Si structural group,17) or Ti–O–Ti vibrations,1) or both. It is indicated that Ti4+ exists in the glasses in the form of [TiO4]. According to Mysen et al.,1) 700–728 cm–1 is probably assigned to deformation of O–Ti–O or O–(Si, Ti)–O in chain or sheet units or both. Based on the reports of You et al.,13) Tsunawaki14) and Mcmillan,15,16) the peak around 1040 cm–1 is due to stretching with NBO/Si = 1 and referred to as Q3 species in a sheet structure. In the present spectral, there are no bands at wavenumbers less than 700 cm–1 that could be attributed to Ti–O stretch vibrations of Ti in six-fold coordination.1) Therefore, it is possible that no or very little [TiO6] exists in the present system and all the TiO2 behaviors as an acidic oxide.
It is known that Al2O3 is amphoteric oxide and its behavior depends on the basicity of melts to which it is added. Al3+ ions are glass network intermediates and can enter as both the network formers and the network modifiers. For all these four compositions, the contents of basic oxides (CaO + MgO) are higher than that of the Al2O3. Therefore, Al2O3 would behave as an acidic oxide and incorporate into the network of SiO2 with the charge compensation of basic oxide, which means Al3+ ion enter as the network formers in the form of [AlO4]. It also can be seen from the Raman spectroscopic results of Figs. 4, 5, 6, 7 that 940–966 cm–1 and 1010–1040 cm–1 bands assigned to the Q2 and Q3, which means Al3+ ion enter as the network formers in the form of [AlO4].
Besides, the abundance changes of bridging oxygen are mainly discussed by comparison of areas of two bands 948 cm–1 and 1040 cm–1. Band 1040 cm–1 (Q3) refers to the structural unit with more bridging oxygen than band 948 cm–1 (Q2), and the area ratios of Q2 to Q3 reflect the abundance changes of bridging oxygen. According to Frantz and Mysen,17,18) the mole fractions of different structure units are related to the band areas Ai according to the equation: Xi = θi·Ai and the ratio of Raman scattering coefficients θi for Q3 and Q2 is θ3/θ2 = 2.92, where Xi represents the molar fraction of different structural unit. Therefore, it can be obtained that
The molar fraction ratio of Q2 to Q3 could be estimated using Eq. (1). The effects of Al2O3 on the abundance of Q2 to Q3 can be observed. It can be seen from Fig. 8 that the estimated ratio of Q2 to Q3 decreases as the content of Al2O3 increases. That means the percentage of Q3 increases compared with that of Q2, which indicates that the percentage of bridging oxygen increases.
(1) In investigation of the phase relations in the SiO2–TiO2 system, Devries et al.19) indeed noticed weight losses that could be attributed to the formation of little Ti2O3, which would have been present in solid solution with TiO2. As for any redox reaction, the most reduced valence is favored by higher temperatures with the result that the abundance of Ti3+ could become significant in air about 2000°C at ambient pressure.20) All experimental work mentioned in this study has been performed at temperatures that were not extremely high. Meanwhile, as a matter of fact, Ti3+ ion gives a glass a purple color that is readily observed if this valence state of titanium is present in significant amounts.20,21,22) However, all glasses of this study are colorless. Therefore, titanium mainly exists in the form of Ti4+ ions.
(2) According to the results of Raman spectroscopic study, it can be found that, [TiO4] and [AlO4] are the predominant forms of Ti4+ and Al3+ in the composition range of present study. The reason increasing viscosity is that the Al3+ ions enter as the glass network formers in the form of [AlO4] and take place of some Ti4+ ions in the network as substituting Al2O3 for TiO2. When Al3+ ions enter as the network formers in the form of [AlO4], to keep the equilibrium of electric charge there should be one M2+ ion in neighbor of two [AlO4] or one non-bridging oxygen ion as well as one [AlO4].17) Therefore, when substituting 1 mol of Al2O3 for 1 mol TiO2, 1 mol [TiO4] disappears but 2 mol [AlO4] adds by consuming 1 mol of CaO which initially behaviors as the network modifier. So, in this study, when the Al3+ ions replace Ti4+ ions in the network, the number of bridging oxygen ions increases which can be seen from the Raman spectroscopic results of Fig. 8, in which the ratio of Q2 to Q3 gradually decreases, thereby, the degree of polymerization increases, so is the viscosity.
(1) In this study, the influence of Al2O3/TiO2 ratio on the viscosity of CaO–MgO–Al2O3–SiO2–TiO2 melts was investigated by using the rotating cylinder method. It was found that viscosity increases as gradually substituting Al2O3 for TiO2.
(2) The Raman spectroscopy technique has been used to provide deep insight into the structure of the CaO–MgO–Al2O3–SiO2–TiO2 melt. Both Ti4+ and Al3+ mainly enter into the glasses as the network formers in the forms of [TiO4] and [AlO4], respectively. When substituting Al2O3 for TiO2 in CaO–MgO–Al2O3–SiO2–TiO2 melts, the number of non-bridging oxygen decreases, which leads to the increases of viscosity.
Thanks are given to the financial supports from China Postdoctoral Science Foundation (2012M510318 and 2013T60061) and the Fundamental Research Funds for the Central Universities (FRF-TP-13-002A) as well as and National Natural Science Foundation of China (51304018).