Tungsten steel is one of the most important high speed steels. Addition of tungsten in steel will bring great hardness and good stability at high temperature. Usually, ferro-tungsten alloy was added into steel during refining process of steel. The ferro-tungsten alloy is manufactured by reduction of scheelite (CaWO4) or Wolframite with Si-Fe alloy or aluminum alloy . Since high melting point of tungsten, the manufacturing of ferro-tungsten requires high temperature and leads to large energy consumption, serious environmental pollution and other problems.
There have been many attempts to direct use of oxides instead of ferroalloys as alloying agents in steelmaking [2–6]. Chychko [2–4] performed plant trial (3 ton scale) using three different alloying mixtures (MoO3+C; MoO3+C+FeOx; and MoO3+C+CaO) as Mo-alloying agents in steelmaking. Mo yield up to 93% had been achieved in plant trial. Alternative method using scheelite directly for productions of tungsten steel has also been proposed [7, 8]. According to this method, scheelite ore or scheelite ore mixed with reductants (Si-Fe alloy, Aluminum alloy) was directly added into molten steel, and reduced by Si, Al in steel or reductants. Successful plant trail had been carried out in a 20 ton Electrical Arc Furnace to produce high W die steel using scheelite . The overall yield of W could be up to 92%.
Investigation on reduction kinetic of scheelite provides important fundamental knowledge to control and optimization of process of ferrotungsten manufacture or direct reduction of scheelite in steelmaking. However, regarding reduction of scheelite, only carbothermic reaction was studied aiming at producing WC from scheelite [9, 10]. There are very few reports on reduction kinetics of CaWO4 by Si. In this work, the isothermal reduction kinetics of CaWO4 by Si from 1,423 K to 1,523 K, where all reactants and products are in solid state, was investigated by using X-ray diffraction quantitative analysis. Scanning electronic microscopy (SEM) attached with energy dispersive spectral (EDS) was employed to investigate the microstructure of reaction interface.
CaWO4 for reduction experiments was synthesized through solid-state reaction between CaCO3 and WO3. CaCO3 (purity>99%) and WO3(purity>99%) powder was taken as raw materials. Raw powders were weighed and then mixed thoroughly in an agate mortar. Powder mixtures were pressed into a pellet with diameter of 8 mm and then placed in a corundum crucible. The pellets were heated at 1,343 K for 24 h, then crushed and ground into powder. As shown in Figure 1, XRD analysis on synthesized powder has verified that CaWO4 with high purity has been synthesized.
Analytical grade Si powder (200 seize) was adopted as reductant. Synthesized CaWO4 and Si powder with molar ratio of 3:2 were mixed in an agate mortar with ethanol as mixing media. The mixed powder was pressed into pellet with mass of about 0.55 g and diameter of 8 mm under pressure of 2.6 MPa.
The reduction of CaWO4 by silicon was carried out in a vertical corundum tube furnace with silicon carbide as heating elements. An S-type (Pt-Pt/10 Rh) thermocouple was mounted underneath the even temperature zone of furnace to control and measure temperature inside furnace. Argon with high purity (>99.999%) was introduced from bottom of corundum tube and flow rate of argon was maintained at 200 ml/min (STP) using a rotameter. After temperature in the furnace reached right temperature, pellets were put in a graphite crucible (ID: 9 mm; OD: 12 mm; H: 35 mm), and then crucible was placed in the even temperature zone of furnace. The crucible were quickly pulled out from furnace and quenched in air when heating was completed. To investigate the effect of different atmosphere, reduction experiments were also carried out in air or in argon with corundum crucible.
The samples were ground into fine powder prior to XRD measurements. XRD examinations were performed using a 18 KW X-ray diffractometer (RIGAKU TTR III) in air with Cu Ka radiation (λ=1.5418 Å, target voltage: 40 kV, target current: 300 mA). Patterns were collected by scanning from 2θ=10° to 90° with step size of 0.01° and a scanning rate of 10°/min. Before XRD quantitative analysis, the background and Kα2 radiation contribution of acquired XRD patterns were removed by employing Crystallographica Search-March (Oxford Cryosystems) software.
Before SEM examination, samples were embedded with resin, ground, polished and coated with gold. SEM examinations were performed using SEM (FEI MLA 250). The working voltage was 25 kV.
Results and discussion
Isothermal reduction kinetics of CaWO4 by Si from 1,423 K to 1,523 K
The reduction of CaWO4 by Si could be expressed as following equation: (1)Figures 2–4 showed XRD patterns for samples treated for various periods at 1,423 K, 1,473 K and 1,523 K respectively. It could be seen that diffraction peaks of CaWO4, Si and W could be distinguished in the patterns. However, there is not any product phase except W which could be distinguished in term of XRD patterns. This could be due to the formation of some non-crystalline phases. Nevertheless, the fraction conversion could be determined by only employing diffraction peaks of CaWO4 and W.
The fractional conversions were determined using an X-ray quantitative analysis method based on RIR (Relative Intensity Ratio) values . The method of X-ray quantitative analysis adopted in the present work was described briefly as follows.
The ratio of mass percentage of CaWO4 to W could be calculated from intensity of the strongest peaks for CaWO4 (peak(1,1,2)) and W(peak (1,1,0)) according to following equation. (2)where ω(CaWO4) and ω(W) are mass percentages of CaWO4 and W respectively. RIR(W) and RIR(CaWO4) are relative intensity ratio values for W and CaWO4 respectively. I(CaWO4) and I(W) are intensities of the strongest peaks for CaWO4 (peak(1,1,2)) and W(peak (1,1,0)) respectively.
The fractional conversion could be further calculated from ω(CaWO4)/ω(W) according to the following equations. (3)
RIR(W)/RIR(CaWO4) could be calculated from relative intensity ratios of W and CaWO4 obtained from ICDD cards. Alternatively, it could be also determined from XRD measurements on mixture of W and CaWO4 with ω(CaWO4)/ω(W) = 1. Figure 5 showed XRD patterns for CaWO4+W mixture (ω(CaWO4)/ω(W) = 1). The XRD measurements were repeated twice. Mean value for RIR(W)/RIR(CaWO4) is 1.467.
The fractional conversions for various periods at 1,423 K, 1,473 K and 1,523 K were determined from XRD patterns in Figures 2–4 and shown in Figure 6. According to phase equilibrium data for the present system , all reactants and products are in solid state at 1,423–1,523 K. Reaction between solid Si and solid CaWO4 particles takes places during reduction process. After contact between Si and CaWO4, a product layers could be generated around reactant. With progress of reaction, this layer could be gradually accumulated. This product layer isolates the reactant from each other, therefore hinders the overall reaction. Continuous reaction could only persist through (1) diffusion of reactant (in the form of atoms or ions) through the product layer to the reaction interface; (2) chemical reactions among reactants at the reaction interface. In most cases, the diffusion of reactant through the product layer is slower than interfacial chemical reaction and often becomes the rate-controlling step of the overall reaction [13, 14]. Jander model  and Ginstling–Brounshtein model  are most frequently used to describe kinetics of three-dimensional diffusion controlling reaction. The expressions of Jander model and Ginstling–Brounshtein model are as follows: (4) (5)
where α is fractional conversions. t is time, k is rate constant and follows Arrhenius equation where E and A designate the apparent activation energy and pre-exponential factor respectively.
In some cases, the interfacial chemical reaction is slower than diffusion through the product layer and should be considered as rate-controlling step. The equation for three-dimensional chemical reaction could be as follow. (6)Experimental kinetics data were fitted using Jander, Ginstling–Brounshtein and 3D chemical reaction model. Fitting results are shown in Figures 7–9. It could be seen from Figures 7 and 8 that both Jander and Ginstling–Brounshtein model fit the kinetic data with fairly accuracy. The correlation coefficients attained by the least square fitting are as large as 0.9945 and 0.9925 for Jander and Ginstling–Brounshtein model respectively. The good agreements between fitted and experimental kinetic data using Jander and Ginstling–Brounstein indicate that 3D diffusion could be rate-determining step of overall reaction. In comparison, as shown in Figure 9, there are large differences between fitted kinetic data using 3D interfacial reaction model and experimental kinetic data, and the correlation coefficient is only. The bad fitting with 3D interfacial reaction model indicates that interfacial chemical reaction could not be rate-determining step of overall reaction. Accordingly, it could be inferred from above fittings that the reduction of CaWO4 by Si from 1,423 K to 1,523 K could be controlled by three-dimensional diffusion through the product layer.
The slopes of straight line in Figures 7 and 8 directly provided rate constants k at different temperatures. As shown in Figures 10 and 11, the natural logarithms of rate constants k were plotted versus reciprocal of absolute temperature to yield the apparent activation energy and pre-exponential values. The temperature dependences of rate constant obtained from the two fitting lines are expressed as follows: (7) (8)Only few reports could be found in literature on kinetics of silicothermic reduction. These studies mainly focused on kinetics of silicothermic reduction of calcined dolomite with regard to Pidgeon process [17, 18]. These studies all suggested that the reduction of dolomite by silicon was controlled by solid-state diffusion of reactants, which is similar to the reaction mechanism revealed in the present work. Wulandari et al.  reported an activation energy value of 299~322 kJ/mol for silicothermic reduction of calcined dolomite in flowing argon, which is close to our activation energy value.
Microstructure of samples
To further elucidate the mechanism for reaction, microstructures for samples with various reaction time and temperature were investigated using SEM-EDS. Figure 12 showed some examples of SEM micrographs for samples reduced for various periods at various temperatures. The light grey phase and dark grey phase were distinguished as CaWO4 and Si phase respectively. The white small particles found between light grey and dark grey phases were determined as W. The composition of black phase was determined as mixture of CaO, SiO2 and WO3, which could be formed by reaction between product (CaO, SiO2) and CaWO4. This phase (denoted by CaxSiyWzOt, x, y, z, t could be variable) was not found in XRD analysis (see Figures 2–4), which may be due to its amorphous nature.
It could be seen from the SEM micrographs that produced tungsten particle has a mean particle size less than 1 µm. The nature of tungsten particle determined that tungsten particle should not be seen as an obstacle for diffusion of reactant to interface. Si particle was separated from CaWO4 by CaxSiyWzOt phase. The continuous reduction requires transfer of Si or CaWO4 across CaxSiyWzOt phase to contact the other reactant phase. As seen in Figure 12(a-d), CaWO4 particle was surrounded by small tungsten particles. It is postulated that Si diffuses across CaxSiyWzOt phase to react with CaWO4 at interface between CaWO4 and CaxSiyWzOt phase. The main resistance for overall reduction could come from the diffusion of Si in CaxSiyWzOt phase.
Effects of different atmosphere on reduction of CaWO4 by Si
The kinetics of CaWO4 reduction by Si was investigated in argon with graphite crucible. To investigate the effect of different atmosphere on reduction, reduction experiments were also conducted in two different atmospheres: in air and in argon with corundum crucible. Since reaction between graphite and trace oxygen in argon would deplete the oxygen in argon, the oxygen pressure around graphite crucible is relatively lower than that around corundum crucible. Therefore, oxygen pressure in graphite crucible is lowest. The XRD patterns for samples in different atmosphere with reaction time of 90 min at 1,473 K were shown in Figure 13. It could be seen from the figure that the amount of produced W increases with decrease of oxygen pressure. The retard of CaWO4 reduction at high oxygen pressure could be attributed to the surface oxidation of Si. Si powder tends to be oxidized at high oxygen pressure and a thin oxidation layer could form on the surface of powder. The oxides “coating” is an obstacle for diffusion of Si to reaction interface, therefore overall reaction slowed down. It is proposed that weak reducing atmosphere would be beneficial to the reduction of Si.
The kinetics of reduction of CaWO4 by Si was investigated by using quantitative XRD analyses at 1,423 K, 1,473 K and 1,523 K. Following conclusions could be reached:
Both Jander model and Ginstling–Brounshtein model could describe well the kinetics of reduction of CaWO4, whereas 3D interfacial reaction model could not describe the kinetics of reduction of CaWO4.
The rate-determining step for reduction of CaWO4 could be diffusion across product layer.
The values of activation energies obtained by fitting kinetic data using Jander and Ginstling–Brounshtein model are as great as 379.93 kJ/mol and 387.16 kJ/mol, respectively
The oxygen partial pressure during reduction has impact on the kinetics of reduction. The reduction was retarded by increase of oxygen partial pressure.
Financial supports from NSFC (grants no. 51174018), Beijing Higher Education Young Elite Teacher Project (No. YETP0346) and State key laboratory of advanced metallurgy are gratefully acknowledged.
Song SQ, Xue ZL, Zhang B, Wang WX, Liu RN, Wang GL. Smelting vanadium steel by direct alloying with self-reduction briquette of V2O5. Ironmaking Steelmaking 2014;41:7–11. Web of ScienceCrossrefGoogle Scholar
Chen ZX, Li JR. Investigation of scheelite instead of ferrotungsten in steelmaking. Iron Steel 1992;27:15–18 (in Chinese). Google Scholar
Li ZB, Guo PM, Zhang HS. Thermodynamic analysis of smelting alloy steels with scheelite, molybdenum oxide and vanadium slag. J Iron Steel Res 1999;11:15 (in Chinese). Google Scholar
Chung BF. Quantitative interpretation of X-ray diffraction patterns of mixtures. I. matrix-flushing method for quantitative multicomponent analysis. J Appl Crystallogr 1974;7:519. CrossrefGoogle Scholar
Yan BJ, Liu JH, Zhang JY. Kinetic studies of LaCoO3-δ preparation from solid state reactions. High Temp Mater Processes 2004;23:163–76. Google Scholar
Shu QF, Liu JH, Yan BJ, Zhang JY. Phase formation mechanism and kinetics in solid-state synthesis of undoped and calcium-doped lanthanum manganite. Mater Res Bull 2009;44:649–53. CrossrefGoogle Scholar
Ginstling M, Brounshtein BI. Concerning the diffusion kinetics of reactions in spherical particles. J Appl Chem USSR 1950;23:1327. Google Scholar
About the article
Published Online: 2015-01-10
Published in Print: 2015-12-01