Distribution Characteristics of Phosphorus in the Metallic Iron during Solid-State Reductive Roasting of Oolitic Hematite Ore

Abstract

The distribution characteristics of phosphorus in the metallic iron, and the reduction behaviors of hematite and fluorapatite during solid-state reductive roasting of a phosphorus-rich (1.3 mass%) oolitic hematite ore were investigated in the temperature range of 900°C–1200°C. Experimental results show that phosphorus in the iron is present in the forms of αFe and Fe₃P, resulted from the simultaneous reduction of fluorapatite and hematite at temperatures up to 1100°C. The phosphorus content in the metallic iron particles decreases from the edge to the center in the pellet cross-section. For the aim of phosphorus removal, the selective reduction of hematite and fluorapatite should be achieved; while for the enrichment of phosphorus as ferrophosphorus, the reduction of fluorapatite, and affinity of P₂ and Fe should be guaranteed. These findings could provide guidance for technique development for the utilization of phosphorus-rich oolitic iron ores via pyro-methods.

1. Introduction

Oolitic iron ores are generated from ooids composed of several concentric layers, in which 0.4–1.1 mass% phosphorus mainly occurs in the form of apatite Ca₅(PO4)₃(F,Cl,OH). Oolitic iron ores have yet to be commercially exploited, due to its poor liberation and fine dissemination of iron minerals.¹⁾

In the modern steel industry, it has generally been considered that phosphorus is deleterious to the mechanical properties of steel because the embrittlement induced. For this reason, the phosphorus content in steels is limited to less than 0.04 mass%.²⁾ Thus, to improve the use of phosphorus-rich iron ores, phosphorus removal is usually necessary prior to their use for ironmaking feedstock.

To recover iron with limited content of phosphorus from the oolitic hematite ore, the process of carbothermic reduction followed by magnetic separation^3,4,5,6,7) or screening^8,9) has been found to be an effective way. However, when the reduction of apatite and iron oxides happens simultaneously at an excessively high temperature, phosphorus will migrate into the metallic iron phase, which is unfavorable for the magnetic separation between iron and phosphorus.^10,11,12) This leads to a dilemma between the reduction of iron oxides and apatite during the reductive roasting. To ensure the desired magnetic separation of metallic iron particles from gangues, high roasting temperatures are required for sufficient reduction of iron species together with the growth of metallic iron particles, but high temperatures facilitate the reduction of apatite to a large extent.^3,13,14)

Alternatively, without prior phosphorus removal, a method for production of ferrophosphorus from oolitic hematite ore where migration of phosphorus to metallic iron occurs during reduction roasting at 1250°C was proposed.¹⁵⁾ High-phosphorus powdered metallic iron was then refined by a duplex steelmaking process, and the refining slag could be used as a phosphate fertilizer.

Excellent works have been conducted on the phosphorus distribution between the hot metal and molten slag,¹⁶⁾ as well as that between the solid iron and molten slag¹⁷⁾ or semi-solid slag.¹⁸⁾ However, the phosphorus distribution between the solid iron and solid oxides has been being unclear. In the viewpoint of that the behavior of phosphorus during solid-state carbothermic reduction is imperative to the Fe–P separation for dephosphorization, and to the affinity of phosphorus and iron for ferrophosphorus production. The aim of the present work is to obtain insights on phosphorus distribution characteristics in metallic iron in the reduced hematite ore. Carbothermic reduction thermodynamics of fluorapatite and hematite, and the reduction behaviors of hematite and fluorapatite during the reductive roasting of phosphorus-rich oolitic hematite ore in the temperature range of 900–1200°C were investigated.

2. Experimental

2.1. Materials

The oolitic hematite ore sample from Hunan province, China was used. As shown in Table 1, the total iron content of the iron ore sample is 51.28%, and other constituents include 9.53% SiO₂, 5.66% Al₂O₃, and 1.30% P. XRD pattern in Fig. 1 shows that the ore is mainly constituted by hematite, quartz, fluorapatite and kaolinite.

Table 1. Main chemical composition of the oolitic hematite ore/mass%.

Fetotal	P	SiO2	Al2O3	CaO	MgO	S	LOI*
51.28	1.30	9.53	5.66	3.57	0.18	0.05	3.06

*LOI  is mass loss on ignition

Fig. 1.

XRD pattern of the oolitic hematite raw ore.

The microstructure and elemental mapping of the ore are presented in Fig. 2. This ore consists of several concentric layers of hematite, kaolinite and fluorapatite in the ooids, and individual quartz grains with a relatively large size.

Fig. 2.

Backscattered electron image (BEI) of the oolitic hematite ore (a); and EDS maps of Fe (b), Si (c), Al (d), P (e) and Ca (f).

2.2. Methods

The ground oolitic hematite ore sample with particle size of 65 mass% passing 0.074 mm was balled into green pellets with a diameter of 12–16 mm, and then dried in a baking box at 105°C for 4 h.

Reductive roasting experiments were performed in a vertical tube resistance furnace with an inner tube diameter of 70 mm as shown in Fig. 3, which was described in detail in the literature.¹⁹⁾ Lignite was crushed and screened to a size range of 1–5 mm used as the external reductant. About 40 g dry pellets were charged into a cylindrical heat-resistant stainless steel pot (inner diameter of 50 mm and height of 200 mm). An excessive amount of coal (150 g) was used to cover the pellets to ensure the sufficient reducing atmosphere. After the furnace was preheated to the target temperature (900°C, 1100°C and 1200°C), the pot was placed in a high temperature zone of the vertical resistance furnace and roasted for 90 min. After roasting, the pot was taken out of the furnace and cooled down to ambient temperature under an anoxic condition. Subsequently, a portion of reduced pellets were polished for SEM-EDS analysis, and the rest were ground for XRD analysis.

Fig. 3.

Schematic diagram of the reduction apparatus.

The raw lump ore (cut into 10 mm cubes) and reduced pellets were mounted by resin, cut and polished, and then the cross-sectional microstructures of them were examined by a scanning electron microscope equipped with an energy diffraction spectrometer (SEM-EDS, JEOL, JSM-6490LV, Japan). The X-ray diffraction (RIGAKU, D/Max 2500) was carried out by using a copper Kα X-ray source, with a scanning angle in the range from 10° to 80° (2θ) and step size of 0.02° (2θ). A voltage of 40 kV and a current of 250 mA were used.

3. Results

After roasting, phosphorus could exist in different states, depending on the reduction ratio of fluorapatite. On the one hand, the non-reduced part of phosphorus presents as phosphate. On the other hand, gaseous phosphorus from the reduced apatite will either emit to the atmosphere or be detained in the metallic iron. Sasabe et al.¹⁰⁾ investigated the relationship between dephosphorization and iron metallization ratio, and found that the majority of P was detained in the reduced iron as the phosphorus emission kept almost constant at 13% when the iron metallization ratio exceeded 30%. Similarly, it was found that the distribution ratio of phosphorus in metallic iron phase increased from 45% to approximately 70% with increasing roasting temperature from 1175°C to 1275°C by Sun et al.¹⁵⁾ Therefore, the repartition characteristic of phosphorus in metallic iron is critical to understanding of the behaviors of phosphorus in the reduced oolitic hematite ore.

According to the Fe–P binary phase diagram (Fig. 4), the retention phosphorus will present as a solid solution of γFe and αFe or phosphides Fe₃P, Fe₂P and FeP, depending on the content of P. Moreover, eutectic mixtures of intermetallic compounds of αFe–Fe₃P (melting point of 1048°C) and Fe₂P–FeP (melting point of 1262°C) will form with the increasing affiliated phosphorus.

Fig. 4.

Fe–P binary phase diagram.²⁰⁾

Thereafter, to study the repartition of phosphorus in the reduced iron ore, it is necessary to obtain concentration profiles of phosphorus and iron in the metallic iron particles. However, bulk analysis is not able to elucidate the heterogeneous distribution of phosphorus in the metal. In this work, the distribution characteristics of phosphorus in the metal phase of oolitic hematite pellets reduced at different temperatures were studied by SEM-EDS analysis, and results are presented in Figs. 5, 6, 7.

Fig. 5.

Backscattered electron images (BEI) of pellets reduced at various temperature and EDS spectra of representative Fe–P particles: 900°C (a, b), 1100°C (c, d) and 1200°C (e, f); light gray part represents metallic iron and dark gray part represents silicates or phosphates depending on the temperature.

Fig. 6.

Concentration profiles of P and Fe in metallic iron particle of the pellet reduced at 1100°C for 90 min.

Fig. 7.

Change of P content in representative metallic iron particles across the pellet /atom%, (a) reduced at 1100°C and (b) reduced at 1200°C.

According to Fig. 5, at 900°C, there is no phosphorus within the metallic iron particles as fluorapatite keeps stable at this temperature. The original structure of oolitic hematite ore was almost inherited, and the size of metallic iron particles was small (Fig. 5(a)). When the temperature was elevated to 1100°C, metallic iron particles were combined with phosphorus because fluorapatite began to be reduced. Line scanning of metallic iron particle was further performed, and the result shows that αFe and Fe₃P were present in the metallic iron particle (Fig. 6). A eutectic mixture of αFe and Fe₃P would form at 1100°C, as indicated by the Fe–P binary diagram. During the solidification, αFe precipitated initially and was enwrapped in the matrix of Fe₃P. For this reason, metallic iron particles gathered together due to the presence of molten iron in the eutectic mixture (Fig. 5(c)). When the temperature further increased to 1200°C, metallic iron particles aggregated more obviously (Fig. 5(e)).

Moreover, it is observed that the phosphorus content in metallic iron particles decreased from the edge to the center of the pellet cross-section, as shown in Fig. 7. P content decreased from 5.31 to 0 atom% at a temperature of 1100°C, and that decreased from 3.83 to 0.16 atom% at 1200°C. The affinity of gaseous P₂ and metallic iron is closely related to the specific texture of oolitic iron ore. The raw ore consists of several concentric layers with a matrix of quartz or fluorapatite in the center, and the hematite layer is adjoined with gangue mineral layers (Fig. 2). Therefore, prior to reduction of fluorapatite, the fresh metallic iron generated during reductive roasting would enwrap the unreduced fluorapatite layer. When the fluorapatite is reduced, the gaseous phosphorus penetrates into the metallic iron readily.

4. Discussion

As phosphorus distribution relies on the reduction of fluorapatite and hematite, the standard Gibbs free energies for the selected reactions were calculated by using HSC-Chemistry 5.0 software, which is feasible to examine their priorities even though performed at standard conditions. Plots of standard Gibbs free energy against temperature are shown in Fig. 8.

Fig. 8.

Plots of ΔG^o vs. temperature of reactions 1–9.

The reduction of hematite proceeds stepwisely (Fe₂O₃→Fe₃O₄→FeO→Fe) when the temperature exceeds 570°C. The controlling step is the reduction of FeO to Fe.²¹⁾ Additionally, fayalite and hercynite are readily generated along with the formation of FeO in the presence of silica and alumina, as shown in Eqs. (2) and (3). The carbothermic reduction of Ca₅(PO₄)₃F requires an extremely high temperature, but it becomes easier in the presence of Al₂O₃ and SiO₂, as demonstrated in Fig. 8 that the ΔG^o values of Eqs. (7), (8), (9) are smaller than that of Eq. (6). Silica and alumina serve to promote the reduction of fluorapatite, providing a thermodynamic driving force.^22,23,24)   

$$\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{C}=\mathrm{F}\mathrm{e}+\mathrm{C}\mathrm{O}$$(1)

  

$$2\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{S}\mathrm{i}{\mathrm{O}}_2=\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4$$(2)

  

$$\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3=\mathrm{F}\mathrm{e}\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_4$$(3)

  

$$1/2\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4+\mathrm{C}=\mathrm{F}\mathrm{e}+1/2\mathrm{S}\mathrm{i}{\mathrm{O}}_2+\mathrm{C}\mathrm{O}$$(4)

  

$$\mathrm{F}\mathrm{e}\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_4+\mathrm{C}=\mathrm{F}\mathrm{e}+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3+\mathrm{C}\mathrm{O}$$(5)

  

$$\begin{array}{l}2/15\mathrm{C}{\mathrm{a}}_5{(\mathrm{P}{\mathrm{O}}_4)}_3\mathrm{F}+\mathrm{C}=3/5\mathrm{C}\mathrm{a}\mathrm{O}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+1/15\mathrm{C}\mathrm{a}{\mathrm{F}}_2+1/5{\mathrm{P}}_2+\mathrm{C}\mathrm{O}\end{array}$$(6)

  

$$\begin{array}{l}2/15\mathrm{C}{\mathrm{a}}_5{(\mathrm{P}{\mathrm{O}}_4)}_3\mathrm{F}+3/5\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3+\mathrm{C}\\ =3/5\mathrm{C}\mathrm{a}\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_4+1/15\mathrm{C}\mathrm{a}{\mathrm{F}}_2+1/5{\mathrm{P}}_2+\mathrm{C}\mathrm{O}\end{array}$$(7)

  

$$\begin{array}{l}2/15\mathrm{C}{\mathrm{a}}_5{(\mathrm{P}{\mathrm{O}}_4)}_3\mathrm{F}+7/10\mathrm{S}\mathrm{i}{\mathrm{O}}_2+\mathrm{C}\\ =2/3\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{i}{\mathrm{O}}_3+1/30\mathrm{S}\mathrm{i}{\mathrm{F}}_4+1/5{\mathrm{P}}_2+\mathrm{C}\mathrm{O}\end{array}$$(8)

  

$$\begin{array}{l}2/15\mathrm{C}{\mathrm{a}}_5{(\mathrm{P}{\mathrm{O}}_4)}_3\mathrm{F}+2/3\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3+41/30\mathrm{S}\mathrm{i}{\mathrm{O}}_2+\mathrm{C}\\ =2/3\mathrm{C}\mathrm{a}\mathrm{A}{\mathrm{l}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_8+1/30\mathrm{S}\mathrm{i}{\mathrm{F}}_4+1/5{\mathrm{P}}_2+\mathrm{C}\mathrm{O}\end{array}$$(9)

To further characterize the solid-state reduction behavior of phosphorus-rich oolitic hematite ore, XRD analyses of the pellets reduced at various temperatures were performed, and the results are shown in Fig. 9. The reduction of fluorapatite was enhanced when reducing temperature was increased from 900 to 1200°C. At 900°C, the diffractions of Ca₅(PO₄)₃F remained unchanged compared to those of the raw oolitic hematite ore (Fig. 1), which suggests that the fluorapatite keeps stable at this temperature. As the temperature was elevated to 1100°C, diffraction peak intensity of Ca₅(PO₄)₃F decreased and the peaks of CaAl₂Si₂O₈ emerged, indicating the reduction of fluorapatite. At 1200°C, diffractions of anorthite became more intensive, indicating that most of fluorapatite had been transferred to anorthite.

Fig. 9.

XRD patterns of oolitic hematite reduced at various temperatures (A- anorthite CaAl₂Si₂O₈; F-fluorapatite Ca₅(PO₄)₃F; He- hercynite FeAl₂O₄; I-iron Fe; P- fayalite Fe₂SiO₄; Q-quartz SiO₂; W-wustite FeO).

As shown from the results of thermodynamic calculation, the carbothermic reduction of Ca₅(PO₄)₃F is facilitated greatly by the presence of Al₂O₃ and SiO₂. However, the previous literatures merely focus on the intensified effect of SiO₂ rather than Al₂O₃.^12,14,25,26) Hence, the effects of SiO₂ and Al₂O₃ on the reduction of hematite and fluorapatite are imperative to be taken into account together, due to the high content of quartz in the raw oolitic hematite ore as well as the formation of amorphous SiO₂ and γ-Al₂O₃ resulted from the thermal decomposition of kaolinite.^27,28)   

$$\begin{array}{l}\mathrm{A}{1}_2{\mathrm{O}}_3·2\mathrm{S}\mathrm{i}{\mathrm{O}}_2·2{\mathrm{H}}_2\mathrm{O}\stackrel{980°\mathrm{C}}{⟶}\\ \gamma \text{-}\mathrm{A}{1}_2{\mathrm{O}}_3+2\mathrm{S}\mathrm{i}{\mathrm{O}}_{2(\mathrm{a}\mathrm{m}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{s})}+2{\mathrm{H}}_2\mathrm{O}\end{array}$$(10)

It can be confirmed by Fig. 9 that, the diffractions of FeAl₂O₄ increase with reducing temperature from 900°C to 1100°C, due to the thermal decomposition of kaolinite. Moreover, the increased intensity of anorthite diffractions at the expense of fluorapatite and quartz reveals the fact that the reduction of fluorapatite was enhanced by the gangue components such as SiO₂ and Al₂O_3, owing to their close association between hematite, fluorapatite, kaolinite and quartz in the raw ore.

Therefore, it can be concluded that, to achieve the selective reduction of hematite over fluorapatite, on the one hand, the roasting temperature should be lower than the temperature at which reduction of fluorapatite initiates. On the other hand, additives such as Na₂SO₄, Na₂CO₃, CaO and Ca(OH)₂ can be used to suppress the reduction of fluorapatite by consuming SiO₂ and Al₂O₃ in advance.^12,13,14,26)

In regard with the treatment of high-phosphorus oolitic hematite ore via carbothermic reduction, measures should be taken in accordance with the target products: 1) for dephosphorization, the selective reduction of hematite and fluorapatite should be achieved; 2) for phosphorus enrichment via ferrophosphorus production, the maximum reduction of fluorapatite and the affinity of gaseous phosphorus and metallic iron should be guaranteed. Moreover, to realize the goals mentioned above, the destruction of the alternating-concentric layers composed of hematite, fluorapatite, quartz and kaolinite in the oolitic hematite ore is essential.

5. Conclusions

The phosphorus distribution characteristics as well as the reduction behaviors of hematite and fluorapatite were investigated, and requirements for the techniques development of high-phosphorus iron ores were discussed. The following conclusions can be drawn:

(1) Reduction of fluorapatite accompanies with that of hematite at temperatures up to 1100°C, and the phosphorus in the metallic iron presents in the forms of αFe and Fe₃P. The eutectic mixture of αFe and Fe₃P with a melting point of 1048°C is favorable for the accumulation of ferrophosphorus particles.

(2) The phosphorus content in metallic iron decreases from the edge to the center of the pellet cross-section. Due to the alternating-concentric layers of hematite and fluorapatite in the raw ore, gaseous P₂ generated by the reduction of fluorapatite readily penetrates into the metallic iron.

(3) The reduction of Ca₅(PO₄)₃F is enhanced by the presence of Al₂O₃ and SiO₂. Besides the effect of SiO₂, that of Al₂O₃ on the reduction of hematite and fluorapatite is imperative to be taken into consideration together, due to the high content of Al₂O₃ commonly contained in the oolitic iron ores. To achieve the selective reduction of hematite and fluorapatite for the aim of dephosphorization, basic compounds could be used to suppress the reduction of fluorapatite by consuming SiO₂ and Al₂O₃ in advance. Future works should include the phosphorus distribution characteristics in the system of FeO–CaO–Al₂O₃–SiO₂.

Acknowledgements

The authors wish to express their thanks to National Natural Science Foundation of China (51174230 and 51234008), and New Century Excellent Talents in University (NCET-11-515) for the partial financial support of this research. This work was also financially supported by Co-Innovation Center for Clean and Efficient Utilization of Strategic Metal Mineral Resources.

References

• 1)   S.  Song,  E. F.  Campos-Toro and  A.  López-Valdivieso: Powder Technol., 243 (2013), 155.
• 2)   H.  Tang,  J.  Wang,  Z.  Guo and  T.  Ou: J. Iron. Steel Res. Int., 20 (2013), 17.
• 3)   G.  Li,  C.  Xie,  Y.  Zhang,  Q.  Li and  T.  Jiang: 1st Int. Symp. on High-Temperature Metallurgical Processing, TMS, Warrendale, PA, (2010), 505.
• 4)   G.  Li,  T.  Lei,  M.  Rao and  T.  Jiang: 3rd Int. Symp. on High-Temperature Metallurgical Processing, TMS, Warrendale, PA, (2012), 483.
• 5)   S.  Weissberger,  Y.  Zimmels and  I.  Lin: Metall. Trans. B, 17 (1986), 433.
• 6)   K.  Ionkov,  S.  Gaydardzhiev,  A.  Correa de Araujo,  D.  Bastin and  M.  Lacoste: Miner. Eng., 46 (2013), 119.
• 7)   Z.  Peng and  J. Y.  Hwang: Int. Mater. Rev., 60 (2015), 30.
• 8)   E.  Matinde and  M.  Hino: ISIJ Int., 51 (2011), 220.
• 9)   E.  Matinde and  M.  Hino: ISIJ Int., 51 (2011), 544.
• 10)   M.  Sasabe,  Y.  Ida and  T.  Yokoo: Tetsu-to-Hagané, 100 (2014), 325.
• 11)   Y.  Li,  T.  Sun,  J.  Kou,  Q.  Guo and  C.  Xu: Miner. Process. Extr. Metall. Rev., 35 (2014), 66.
• 12)   Y.  Sun,  Y.  Han,  P.  Gao,  Z.  Wang and  D.  Ren: Int. J. Miner. Met. Mater., 20 (2013), 411.
• 13)   G.  Li,  S.  Zhang,  M.  Rao,  Y.  Zhang and  T.  Jiang: Int. J. Miner. Process., 124 (2013), 26.
• 14)   W.  Yu,  T.  Sun,  J.  Kou,  Y.  Wei,  C.  Xu and  Z.  Liu: ISIJ Int., 53 (2013), 427.
• 15)   Y.  Sun,  Y.  Han,  P.  Gao and  D.  Ren: Int. J. Miner. Met. Mater., 21 (2014), 331.
• 16)   Y.  Kobayashi and  S.  Kodama: ISIJ Int., 52 (2012), 960.
• 17)   N.  Maruoka,  S.  Ono,  H.  Shibata and  S.-y.  Kitamura: ISIJ Int., 53 (2013), 1709.
• 18)   F.  Pahlevani,  S.-y.  Kitamura,  H.  Shibata and  N.  Maruoka: ISIJ Int., 50 (2010), 822.
• 19)   G.  Li,  T.  Shi,  M.  Rao,  T.  Jiang and  Y.  Zhang: Miner. Eng., 32 (2012), 19.
• 20)   S. K.  Zečević,  J. B.  Zotović,  S. L.  Gojković and  V.  Radmilović: J. Electroanal. Chem., 448 (1998), 245.
• 21)   H. Y.  Sohn and  M. E.  Wadsworth: Rate Processes of Extractive Metallurgy, Plenum Press, New York, (1979).
• 22)   K.  Jacob,  D.  Reynolds and  W.  Hill: Ind. Eng. Chem., 21 (1929), 1126.
• 23)   J.  Mu,  F.  Leder,  W. C.  Park,  R. A.  Hard,  J.  Megy and  H.  Reiss: Metall. Trans. B, 17 (1986), 861.
• 24)   K.  Jacob and  D.  Reynolds: Ind. Eng. Chem., 20 (1928), 1204.
• 25)   J.  Yin,  X.  Lv,  C.  Bai,  G.  Qiu,  S.  Ma and  B.  Xie: ISIJ Int., 52 (2012), 1579.
• 26)   S.  Bai,  S.  Wen,  D.  Liu,  W.  Zhang and  Q.  Cao: ISIJ Int., 52 (2012), 1757.
• 27)   G. Z.  Qiu,  T.  Jiang,  G. H.  Li,  X. H.  Fan and  Z. C.  Huang: Scand. J. Metall., 33 (2004), 121.
• 28)   G.  Li,  T.  Jiang,  N.  Sun,  X.  Fan and  G.  Qiu: Light Metals 2009 - TMS 2009 Annual Meeting and Exhibition, TMS, Warrendale, PA, (2009), 57.