2014 Volume 54 Issue 4 Pages 721-727
During the production of titanium stabilized stainless steel, as titanium in steel has a tendency to reacting with SiO2 in mould fluxes to generate TiO2 into mould fluxes and mould powder can inevitably pick up Ti-bearing inclusions floating up from steel, TiO2 content in the molten mould fluxes gradually increases so that physiochemical properties of the fluxes change. To evaluate the effect of TiO2 increase in mould fluxes on the structure of the mould flux, the glassy slag system CaO–SiO2–CaF2–TiO2 for stainless steel casting fluxes was studied by combining Raman spectroscopy with 29Si and 19F Magic Angular spinning Nuclear Magnetic resonance (MAS-NMR) to obtain the structure information. Both Raman and 29Si MAS-NMR investigation results have shown that Q2 is predominant silicate species in structures of all samples. Ti4+ mainly exists in the form of [TiO4] in slag, and forms TiO2-like clusters with Ti4+ in tetrahedral coordination, which cannot change the degree of polymerization of the silicate network. A small amount of Ti enters into the silicate network as the role of network formation, which slightly enhances the degree of polymerization of the silicate network. According to 19F MAS-NMR spectra, most of the fluorine is exclusively coordinated by Ca2+ corresponding to F–Ca(n) site and only a few Si–F bonds were observed in samples. Increase of TiO2 content has no significant effects on the F- bonds.
Mould fluxes are usually used during continuous casting of steel and play an important role in controlling quality of steel.1,2,3) During the casting of titanium-stabilized stainless steel, titanium in the steel have a tendency to reacting with silica in mould fluxes to generate TiO2 into mould fluxes, and mould powder can inevitably pick up Ti-bearing compounds inclusions floating up from steel, which leads to increase of the concentration of TiO2 in mould flux. This phenomenon can be illustrated by following equation:4)
Increase of TiO2 concentration due to above reaction could have strong influence on physicochemical properties of mould slag such as viscosity and crystallization ability, and performance of heat transfer and lubricity, which would bring instability and even problems during operation of casting.4,5)
Physicochemical properties of melts strongly depend on the microstructure and investigation of structure characteristics can be employed to explain the changes of physiochemical properties.6,7,8) For example, there are correlations between polymerization degree of molten slag and viscosity.9,10) Similarly, it has been shown that the density and electrical conductivity of melts are simple functions of polymerization degree which reflects silicate network structure.10,11,12) From viewpoint of application, an established relationship between structure and properties of mould fluxes would help to better control of mould fluxes during continuous casting. Despite the importance of structural information of mould fluxes, structural studies on mould fluxes are still scarce.1) The investigation on the structure of TiO2-bearing mould flux system of CaO–SiO2–CaF2–TiO2 in the casting of Ti-stabilized stainless steel isn’t reported. On the other hands, although there are some previous reports about TiO2 or CaF2-bearing glass or melts, there is still some controversy on structural role of TiO2 and CaF2 in the system. Raman spectral study on silicate glass performed by Mysen et al. indicated that Ti4+ is in four-fold coordination and forms discrete units in the structure of glasses.14) Another Raman spectral study conducted by Mysen et al.15) showed coexistence of six-fold and four-fold coordinated titanium, and four-fold titanium increases with increase of TiO2 content. Raman and infrared spectral study by Bihuniak et al.16) showed that titanium is present in four-fold coordination in titanium-containing vitreous silicas, while X-ray Absorption Fine Spectra(XAFS)17) and neutron diffraction18) study showed dominant five-fold titanium existence in some glasses and melts. Regarding fluorine in structure of glass, 19F nuclear magnetic resonance with magic angular spinning (MAS-NMR) study show that most fluoride is in sites that are structurally similar to the corresponding simple metal fluoride and a few present of Si–F bonds can be founded in the sodium silicates. In CaO–Al2O3–CaF2–SiO2 system, the principal fluorine species is found to be F–Ca(n) with no Si–F or Al–F by 19F MAS-NMR.1) (6) In CaO–SiO2–CaF2 glasses, fluoride ions are found to be coordinated with four or less than four calcium ions by employing solid-state 19F NMR.20) Therefore, CaO–SiO2–CaF2–TiO2 glass system has been investigated in this study to obtain the structure of mould fluxes in casting of Ti-stabilized stainless steel for developing the more optimal mould fluxes.
Many techniques including IR spectrum analysis, X-ray photoelectron spectroscopy (XPS), X-ray absorption near edge structure analysis (XANES), high temperature X-ray diffraction and neutron diffraction analysis, Raman spectroscopy and NMR, etc. have been employed by many researcher to investigate structure of glass and melts. Among these techniques, Raman spectroscopy and NMR are powerful methods for resolving the structure of local arrangements in glasses. They have obvious advantages: microanalysis, high precision and accuracy, high analysis speed and no destruction, and so on.23,24)
Currently, structures of glassy mould fluxes in CaO–SiO2–CaF2–TiO2 system were investigated by combining Raman spectroscopy with 29Si and 19F Magic Angular spinning Nuclear Magnetic resonance (MAS-NMR) in order to determine the structure of possible Ti-bearing complexes and to assess the interaction between such complexes and silicate.
Some researchers have compared the infared and Raman spectra of various silicate glasses and their melts, which found the glass and melt spectra be similar.23,25) Since structures of quenched glasses could be seen as first approximation of parent melts, the present structural results could be used to explain properties of liquid mould fluxes.
The innovative aspect of this work is to demonstrate the effect of TiO2 on the structure when CaF2 and TiO2 present in silicates and employs Raman spectroscopy and MAS-NMR to analyze at the same time. Since there is few study of glass or melts bearing both CaF2 and TiO2 found in literature, it is also a novel point for the present study that structures of glassy mould fluxes in CaO–SiO2–CaF2–TiO2 system which is the basic mould flux of the Titanium-stabilized stainless steel were studied using multi-spectroscopic technique.
Reagent grade powders of CaCO3 (>99.5 mass%), CaF2 (>99 mass%), TiO2 (>99.5 mass%) and high purity SiO2 (>99.99 mass%) were used as raw materials. CaCO3 were calcined at 1323 K to obtain CaO in a muffle furnace for 10 hours. TiO2, CaF2 and SiO2 powders were also calcined at 773 K to remove moisture. After carefully weighed, the powder mixture was ground in an agate mortar with ethanol as mixing media. Powder mixtures were placed in a platinum crucible and melted in a molybdenum silicide vertical tube furnace at approximately 1723 K to 1773 K in air for 2 hours. High temperature melts were quenched into water to form the glasses. After quenching, the samples were subject to X-ray fluorescence (XRF) to determine the composition. Nominal Compositions of samples and composition of after the melting which is analyzed by XRF are listed in Table 1. All quenched samples were optically transparent without phase-separation and micro-crystallization. According to XPS26) and XANES27) analyses, titanium in glass prepared in air exists in Ti4+ redox state. Due to present glassy samples were melted and quenched in air, Ti4+ was considered to be sole titanium ion in the present glasses.
Powder X-ray diffraction (XRD) was used to verify amorphous state of samples (Fig. 1). Powder X-ray diffraction measurements were carried out on a M21X-SRA X-ray diffractometer (MACScience) equipped with graphite crystal monochromator in air.
XRD pattern of the quenched samples.
Solid-state 29Si and 19F MAS-NMR measurements of the powder glasses were recorded on a 400M FT-NMR spectrometer (BrukerAvance III 400M (9.4T), Germany) using a MAS probe with 4 mm ZrO2 rotor and two pairs of Dupont Vespel caps. The spinning rates were 5 KHz and 25 KHz for 29Si and 19F spectra respectively. All 29Si and 19F MAS-NMR spectra were referenced to the materials of tetramethylsilane (TMS) and PVDF (polyvinylidene fluoride), respectively. The spectra were fitted by assuming Gaussian line shapes for peaks of different structural units. Abundances of structural units were calculated in terms of area fractions of peaks. Relative peak area, including spinning sidebands, was estimated by Gaussian line shapes fits.
Glassy slags were used for nonpolarized Raman spectroscopy measurements. Raman spectra were recorded at room temperature in the frequency range of 100–2000 cm−1 using a laser confocal micro-Raman spectrometer, JY-HR800 (JobinY’von, France). The experiments were performed in room temperature using excitation wavelength of 532 nm and the light source was a semiconductor laser with power of 1 mw. The spectra were fitted by assuming Gaussian line shapes for peaks of different structural units. Abundances of structural units were calculated in terms of area fractions of peaks.
Figure 2 shows the 19F MAS-NMR spectrums for glassy samples with different contents of TiO2 and show signals between 0 and –200 ppm. Spinning side bands were also marked in the figure. The 19F MAS spectrum has a strong and broad peak centered at about –95 ppm, which is in close proximity to the chemical shift of the non-planar F-Ca(n)site.17,18,19,20,28,29,30) At about –150 ppm, there is a small shoulder appearance in all samples, and shoulder become sharp in spectra of samples with 10% TiO2 content. It can be seen from Fig. 2 that introduction of TiO2 in glassy slag brings no significant changes of 19F MAS-NMR spectra, which indicates that TiO2 have no critical effects on the F- bonds.
19F MAS-NMR spectra for samples with different contents of TiO2 (No. 1, 2 and 3).
The 29Si MAS-NMR spectra are presented in Fig. 3 and show signals between 0 and –150 ppm which correspond to the Si–O environments. All spectrums has a maximum centered peak near –78 ppm, which has been marked in the figure. The peak chemical shifts only slightly change from –77.06 to –78.10 with TiO2 addition. According to Stebbins32,33) et al., Si chemical shift becomes more negative with the number of bridging oxygen increasing from 0 to 4 (Q0 to Q4). This peak chemical shifts change suggests that the addition of TiO2 could slightly increase degree of polymerization of the silicate network structure.
29Si MAS-NMR spectra for samples with different contents of TiO2 (No. 1, 2 and 3).
In this study, all Raman spectra were subtracted their background, then deconvolved by Gaussian line shapes which were assigned to individual vibrations of structural units. Before spectra are fitted, it is necessary to correct the data for frequency dependant scattering. In these experiments, Raman spectra were corrected according to the follow equation:23)
Where is intensity of original spectra, is intensity of corrected spectra, ν and ν0 are the frequencies of the exciting line and Raman shift. All original spectrums were shown in Figs. 4(a)–4(b). Figure 4 indicates that, even minor TiO2 (5 mass% in sample 2) are added to the glass sample, there have appeared critical changes in Raman spectra. With the introduction of TiO2, there appears new strong Raman band at about 840 cm–1, while the bands at about 600–720 cm–1 of the Raman spectra for the glass samples gradually become weaker. It was also observed that the relative intensity of the Raman signals at about 860 cm–1 become weaker, and Raman bands near 910 and 970 cm–1 become slightly stronger.
Original Raman spectra for samples with different contents of TiO2 (No. 1,2 and 3) at room temperature.
As shown in Fig. 2, the peak centered at about –95 ppm is predominant for all glassy samples. This dominant peak could correspond to F–Ca(n) (n represents the number of Ca coordinated with fluorine, and fluorine has an unknown number of exclusively Ca2+ nearest neighbors, with n typically between 3 and 4) site.18,19,20,21,22,28,29,30,31) Therefore, most of the fluorine is exclusively coordinate by Ca2+ in the structure of present glassy slags. According to Zeng and Stebbins,19) high-frequency shoulder at about –150 ppm may reflect Si–F site.31) Since peak at about –150 ppm is much weaker than peak at –95 ppm, only very small amounts of Si–F bonds exists in structures. Through calculation of the peak area, the maximum percentage of Si–F bonds is only 3.75%. Based on F environment acquired by NMR spectrums, we could conclude that most of the fluorine in glass samples is exclusively coordinate by Ca2+ and only a few Si–F bonding could be observed.
NMR data for each type of solid silicates permitted us to make the ranges of 29Si chemical shift for Q0, Q1, Q2 and Q3 structure units (superscript 0, 1, 2, 3 is the number of bridging oxygen per SiO4 tetrahedra).34,35) Different Qn species could coexist according to equilibria among Qn species which could be described as follows:36)
In the present work, the 29Si MAS-NMR spectrum of all glass samples is deconvolved using Gauss-Deconvolution method by assuming contributions from structural units of Q3, Q2, Q1 and Q0 with the minimum correlation coefficient r2 ≥ 0.99.6,36,37,38) The best-fit simulations are shown in Figs. 5(a)–5(c). The summary of deconvolution results is listed in Table 3. It is clear that a major peak near –78 ppm is dominated by Q2 and smaller peaks near –70 ppm, –74 ppm and –88 ppm are dominated by Q0, Q1 and Q3, respectively. From the deconvolution results (Table 3), Q2 is a predominant structural form in all samples, Q0 and Q2 has a slight decrease and Q1 and Q3 has a slight increase as increase of TiO2.
Deconvolved results of 29Si MAS-NMR for samples with different TiO2 contents.
The molar fractions of different structure units are related to the peak areas. In this work, the average number of bridging oxygen of each sample is used to explain the change of the silicates structure network, which could be estimated by the area ratio of each structural units (Q0, Q1, Q2 and Q3) multiplied by the number of its bridging oxygen. The average bridging oxygen number of each sample was estimated using 29Si NMR spectral and the effect of TiO2 can be observed from Fig. 6. It could be seen that all average bridging oxygen numbers are near 1.8. For the CaF2–CaO–SiO2 ternary glass, if only Ca–F bond exists, theoretical average bridging oxygen number should be 1.83 when CaO/SiO2 (mass ratio) is 1. The consistence between theoretical and calculated bridging number validates our deconvolution results. As can be seen in Fig. 6, the average number of bridging oxygen also has a slight increase (about 5%) as the content of TiO2 increases from 0 to 10% in mass. That means the majority of Ti4+ form TiO2-like clusters with Ti4+ in tetrahedral coordination and don’t change the degree of polymerization of silicates. At the same time, the average number of bridging oxygen slightly increases, which can be explained as there might be a few TiO2 playing the role of network formation to increase the degree of polymerization.
Effects of TiO2 contents on the number of the average bridging oxygen.
The deconvoluted results of corrected Raman spectra with different content of TiO2 are showed in Figs. 7(a)–7(c) through the Gaussian-Deconvolution method similar to method by Mysen et al.39) with the minimum correlation coefficient r2 ≥ 0.998. All spectrums were successfully fitted in its high-frequency parts from 600–1200 cm–1. Detailed assignments of Raman bands are summarized in Table 2. From the deconvolution results analysis (Table 3), there were an intense band at about 970 cm–1 and a relatively weak band near 870 cm–1, 910 cm–1 and 1050 cm–1 in the TiO2-free sample. According to Mysen et al.14) and McMillan,40) the band centered at about 970 cm–1 is due to stretching with NBO/Si=2 (NBO is the non-bridging oxygen) and is referred to as Q2 species in a chain structure.39,40,41,42) Bands centered at about 870 and 910 cm–1 is assigned to Q0 in monomer structure and Q1 in dimmer structure respectively.7,39,40) There is another band at about 1050 cm–1, which could be assigned to stretching with NBO/Si=1 and referred to as Q3 species in a sheet structure. according to McMillan40,41) and Parkinson et al.42) While Mysen et al.43,44,45) suggested that this band could be also due to Si–O0 stretch vibration in structure units (but not necessarily residing in fully polymerized structural units) or alternatively to a vibration in structure units associated with the metal cation.46) It can be seen that there is overlap between Si–O0 and Q3 for their Raman signals. In the present work, Raman spectral at about 1050 cm–1 has been well deconvolved by assuming a Gaussian peak, and combined with the NMR spectroscopic analysis, this Raman band has been assigned as Q3 species in a sheet structure. With the introduction of TiO2, a new band appear in the spectra at about 840 cm–1. With the increasing of TiO2 content, the band at 840 cm–1 gradually increases in intensity and becomes the dominant peak. Based on Mysen et al.,14) Duan et al.47) and Wang et al.48) the band at about 840 cm–1 in Ti-bearing samples are due to the vibration of [TiO4] monomer structure unit. It can be found that intensity of [TiO4] bands gradually increase with the addition of TiO2 from the Fig. 7. The bands at about 600–750 cm–1 of the Raman spectra for the glasses are due to O–Si–O and O–Ti–O deformation vibration.14)
Deconvolved results of Raman spectra for samples with different TiO2 contents.
According to Frantz and Mysen,44,45) the mole fractions of different structure units is related to the band areas in Raman spectra. The average number of bridging oxygen of each sample is used to explain the change of the silicates structure network, which could be estimated by the area ratio of each structural units (Q0, Q1, Q2 and Q3) in Raman spectra multiplied by the number of its bridging oxygen. The effects of TiO2 on the estimated average number bridging oxygen can be observed in Fig. 6. It can be seen that the average number of bridging oxygen just increases slightly as the content of TiO2 increases from 0 to 10% in mass, indicating degree of polymerization of silicates changes very little. This result is in good agreement with NMR spectra results.
The exact nature of the structural behavior of Ti4+ cannot be obtained with certainty from the Raman spectra, but reasonable structural interpretation may be considered. There are three possible model for structural role of titanium, already suggested from various reports on structure of Ti-bearing glasses, will be considered.15) (1) Ti4+ substitutes for Si4+ in tetrahedral coordination in the structural units in the glasses.49) (2) Ti4+ forms TiO2-like clusters with Ti4+ in tetrahedral coordination.15,16,49,50) (3) Ti4+ as a network modifier, possibly occur in five-fold or six-fold coordination.51,52) In the present work, it can be found that the second model may be employed to interpret deconvolution results. That means Ti4+ forms TiO2-like clusters with Ti4+ in tetrahedral coordination. If it is only situation of the first model, Ti4+ substitutes for Si4+ in tetrahedral coordination in the structural units in the glasses, the number of average bridging oxygen will be significantly larger (for CaO–SiO2 (CaO/SiO2=1)–TiO2 system, if only consider model 1, the bridging oxygen number for sample with 10 mass% TiO2 addition would increase from 1.82 to 2.23) and the degree of polymerization of silicates should be significantly enhanced. As the third model stated, Ti4+ as a network modifier, possibly occur in fivefold or sixfold coordination, it will significantly decrease the degree of polymerization of silicates. It could be seen from Fig. 6 that the number of average bridging oxygen only slightly increases with the addition of TiO2. That means only a few TiO2 enter into the silicate network slightly enhancing the degree of polymerization. Accordingly, the first and the third model could not predict consistent the degree of polymerization with the present results. Therefore, the second model is most appropriate to describe the role of titanium in structure.
Correlation between structural information and physiochemical properties of mould slags could be naturally expected. Many researchers4,5) have reported that viscosity of mould fluxes decreases with increased pickup of TiO2 in fluxes during casting. According to the present structural study, degree of polymerization is slightly enhanced with TiO2 increase. Therefore, the decrease of viscosity could not be attributed to decrease of polymerization degree. Relative low Ti–O strength could be the main reason for viscosity decrease. As field strength (charge/radius, Z/r) of Ti4+ is lower than that of Si4+. The Ti–O bonds are expected to be weaker than Si–O bonds. It was also reported5) that crystallization ability of mould fluxes also decreases with increase of TiO2. The slightly enhanced degree of polymerization could be one reason for decreased crystallization. The other reason could be increased amounts of discrete [TiO4] structural units, which could hamper the crystallization of cuspidine as main crystallization product for mould fluxes.
Structure of glassy CaO–SiO2–CaF2–TiO2 system with varied TiO2 content was investigated by Raman spectroscopy and 29Si, 19F MAS-NMR techniques. The interpretations of spectra lead to the following conclusions.
(1) Both Raman and 29Si MAS-NMR investigation results have shown that Q2 is predominant silicate species in structures of all samples.
(2) Ti4+ mainly exists in the form of [TiO4] in silicates, and forms TiO2-like clusters with Ti4+ in tetrahedral coordination, which cannot change the degree of polymerization of the silicate network. A small amount of Ti enters into the silicate network as the role of network formation, which enhances the degree of polymerization of the silicate network.
(3) 19F MAS-NMR spectra provide F-bonds information in the silicate framework. Most of the fluorine is exclusively coordinate by Ca2+ corresponding to F–Ca(n) site. Only negligible Si–F bonds were observed.
Financial supports from NSFC (grants no. 51174018), the Fundamental Research Funds for the Central Universities (no. FRF-TP-12-019A) and State key laboratory of advanced metallurgy are gratefully acknowledged.