Reduction Roasting of Tin-Bearing Iron Concentrates Using Pyrite

Abstract

A novel method of reduction roasting of tin-bearing iron concentrates using pyrite (FeS₂) is proposed, in which hematite in tin-bearing iron concentrates is reduced to magnetite (Fe₃O₄) and cassiterite (SnO₂) is transformed to volatile tin sulphide (SnS). In a certain range, tin removal and iron reduction rates both increase with roasting temperature, pyrite amounts and residence time. When pyrite amounts exceed 5.00 mass%, some of the produced magnetite will be sulfurated to FeS and the evaporation of SnS (g) will also be obstructed. Tin content of the concentrate is decreased to 0.07 mass% and hematite is reduced to magnetite completely under the conditions of N₂ flow rate of 5.00×10⁻⁶ m³/s, roasting temperature of 1473.0 K, residence time of 60 min, pyrite addition amounts of 5.00 mass% and concentrates particle size of 0.075 mm. The roasting products can meet the standard of BF ironmaking, which requires tin content in iron ores less than 0.08 mass%, and the tin removed can be recycled through smoke dust collection.

1. Introduction

In recent years, the Chinese iron and steel industry develops rapidly, and the corresponding iron ores external dependence degree has kept high for a long time, going against the industry further development. Besides, complex iron ores containing tin, vanadium, titanium and other nonferrous metals have taken an overwhelmingly dominant position in China nowadays.^1,2,3,4) Realizing sustainable development of the iron and steel industry, China must use domestic resources to produce a steady source of high-quality iron concentrates.

There is a tin-bearing iron ore which is mainly found in the north region of China.^5,6) Tin impurity reduces quality of the corresponding iron concentrates, and these concentrates cannot be used directly as blast furnace iron-making materials. Many approaches were reported to utilize these iron ores and the following findings were obtained. The process of iron ores treated by a joint physical separation process of grinding, reselection, magnetic separation and flotation could not obtain qualified iron concentrates, because tin phases (SnO₂) were embedded in iron phases and it was difficult for being separated through milling method.³⁾ In the methods of selective sulfurization or chlorination, tin could be effectively separated from iron ores, but the disadvantages which included equipment corrosion, environmental pollution, atmosphere and iron phases in the products hard to control were also apparent.^{4,5,6,7,8,9,10,11,12)} Using reduction method to treat tin-bearing iron concentrates, tin contents of the products can be decreased lower than 0.01 mass%, but the process is difficult to be extended in industry due to its uneasy controlment.^{5,6,7,8,9,10,11)}

Pyrite (iron-pyrite, FeS₂) is the most widely used sulfidizer in the tin-containing materials treatment. The ferrous ion Fe²⁺, element sulfur S, and the disulfide S₂²⁻ can be formed in the decomposition of pyrite,¹³⁾ and these products make pyrite possess reductive and sulfide property. Thinking of this, a novel method to reduce hematite in tin-bearing iron concentrates parallel to removing tin using pyrite has been proposed in this study. The tin removed can be recycled through smoke dust collection. Different with the earlier research, a special reductant, such as C, CO or H₂,^{5,6,7,8,9,10,11,14,15,16,17)} was not added to the reaction system. The treatment conditions, such as roasting temperature and time, pyrite addition amounts and particle size of the tin-bearing iron concentrates have been investigated.

2. Experimental

2.1. Sample Characterization

The tin-bearing iron concentrate used in this study was collected from a company in Yunnan province, China. Chemical composition (mass%) of it is depicted in Table 1. Table 1 shows that the major element present is iron (64.85 mass%) and tin content is of 0.27 mass%, both being well worth to be reclaimed. “Others” in Table 1 is mainly composed of “O” (existing as Fe₂O₃ and Fe₃O₄ etc.). Main mineral constituents of the samples, established by X-ray diffraction and EPMA, are reported in Figs. 1 and 2. Figure 1 shows that main iron phases of the tin-bearing iron concentrates are Fe₂O₃ and Fe₃O₄, and mass ratio of them is close to 1:1. Figure 2 also shows that the concentrates mainly contain hematite and magnetite, and minor amounts of cassiterite are embedded in iron phases at every size fraction. Pyrite was used as reducing agent and sulfidizer in the roasting process, the FeS₂ content of which is over 98.00 mass%.

Table 1. Chemical composition of the tin-bearing iron concentrates (mass%).

Components	Fe	CaO	SiO2	Al2O3	Sn	Sb	Pb	Zn	Cu	Others
Contents	64.85	1.55	0.91	0.87	0.27	0.34	0.06	0.04	0.01	30.07

Fig. 1.

XRD pattern of the tin-bearing iron concentrates.

Fig. 2.

Mineral phase’s distribution obtained by EPMA.

2.2. Roasting Experiments

100 g of tin-bearing iron concentrates samples were used in every experiment, and addition amount of pyrite was calculated based on the quality ratio of it to tin-bearing iron concentrates, being range from 0.00 mass% to 100.00 mass%. After mixed carefully, the mixtures of tin-bearing iron concentrates and pyrite were roasted in a tube furnace in the range of 1073.0–1473.0 K under N₂ atmosphere. After a proper time held, the reactor was pulled out for cooling down and analysis.

2.3. Characterization

The thermo gravimetric analysis (TG) and differential scanning calorimetry (DSC) of the samples were performed on a thermal analyzer (NETZSCH, STA 449 F3). The typical measurements were performed at a heating rate of 0.17 K/s under Ar atmosphere. Chemical composition and mineralogy of the samples were characterized by chemical analysis, scanning electron microscope (SEM), point counting, energy dispersive spectrometer (EDS), and electron probe microanalysis (EPMA). All of the products phase compositions were detected by an X-ray diffraction (Rigaku D/max-3B) with a Cu-Kα radiation. Spectra were registered between 10° and 90° with a step of 8° and per step of 1 minute. The thermodynamic data of species were given by FactSage thermochemical software. Mathematical expression of the tin removal rate was defined as:   

$$\mathrm{R}=\left(1-\frac{{\mathrm{M}}_{\mathrm{T}}\times {\mathrm{W}}_0}{{\mathrm{M}}_{\mathrm{C}}\times {\mathrm{W}}_{\mathrm{C}}}\right)\times 100\%$$(1)

Where, M_C stands for the mass of tin-bearing iron concentrates used, W_C for the tin content in the origin tin-bearing iron concentrates, M_T for the total mass of the roasted residue, W₀ for the tin content in the roasted residue.

3. Thermo-chemical Behavior of the Mixture of Pyrite and Tin-bearing Iron Concentrates

It has been confirmed by many studies that the decomposition of pyrite proceeds through a multi-step process in the following sequence:^18,19,20)   

$$\mathrm{p}\mathrm{y}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{e}\left(\mathrm{F}\mathrm{e}{\mathrm{S}}_2\right)\to \mathrm{p}\mathrm{y}\mathrm{r}\mathrm{r}\mathrm{h}\mathrm{o}\mathrm{t}\mathrm{i}\mathrm{t}\mathrm{e}\left(\mathrm{F}{\mathrm{e}}_{1-\mathrm{x}}\mathrm{S}\right)+{\mathrm{S}}_{\mathrm{x}}\to \mathrm{t}\mathrm{r}\mathrm{o}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{e}\left(\mathrm{F}\mathrm{e}\mathrm{S}\right)+\mathrm{S}$$

The above results demonstrate that pyrite can be thermo-decomposed into pyrrhotite, troilite, and element sulfur during heat-treatment, all of which have the reduction ability to react with hematite.

Figure 3 shows TG-DSC curves of the mixture of tin-bearing iron concentrates and pyrite (addition amounts of 50.00 mass%) under Ar atmosphere. There are six endothermic peaks in the range of 323.3–1558.3 K in addition to the dehydration stage approximately at 394.2 K. Each of these peaks corresponds to one weight-loss stage. At 703.8–864.0 K, the pyrite breaks down to sulphur vapour and pyrrhotite,^18,19,20,21) corresponding to weight-loss of 5.72 mass%, and there is another weak endothermal peak as a shoulder appearing near 880.1 K, which is related to the reduction of hematite by the decomposed products of pyrite.¹⁹⁾ At 1232.5–1320.0 K, the weight-loss (−2.00 mass%) occurs and the corresponding endothermic peak is attributed to the further decomposition of the primary pyrrhotite before pyrite is completely decomposed.^18,19,20,21)

Fig. 3.

TG-DSC curves of the mixture of tin-bearing iron concentrates and pyrite (addition amounts of 50.00 mass%).

Reactions that may occur in temperature range of 950.0–1558.3 K are listed as follows:^19,22)   

$$\mathrm{F}\mathrm{e}{\mathrm{S}}_2=\mathrm{F}\mathrm{e}{\mathrm{S}}_{\mathrm{x}}+\left(2-\mathrm{x}\right)\mathrm{S}\left(\mathrm{x}=1-2\right)$$(2)

  

$$\mathrm{S}+6\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3=4\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4+\mathrm{S}{\mathrm{O}}_2$$(3)

  

$$\mathrm{F}\mathrm{e}{\mathrm{S}}_2+4\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3=3\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4+2\mathrm{S}$$(4)

  

$$\mathrm{S}\mathrm{n}{\mathrm{O}}_2+{\mathrm{S}}_2=\mathrm{S}\mathrm{n}\mathrm{S}+\mathrm{S}{\mathrm{O}}_2$$(5)

  

$$3\mathrm{S}\mathrm{n}{\mathrm{O}}_2+4\mathrm{F}\mathrm{e}\mathrm{S}=4\mathrm{F}\mathrm{e}\mathrm{O}+3\mathrm{S}\mathrm{n}\mathrm{S}+\mathrm{S}{\mathrm{O}}_2$$(6)

Figure 4 shows standard Gibbs free energy changes of these reactions in temperature range of 323.3–1558.3 K. It is known that the reduction of Fe₂O₃ by FeS₂ involves indirect (Eqs. (2) and (3)) and direct (Eq. (4)) ways.¹⁹⁾ It is seen from Fig. 4 that the Fe₂O₃ is reduced to Fe₃O₄ indirectly by FeS₂ above 703.4 K (Eq. (3)), and directly above 961.9 K. Besides, Fig. 4 shows that the initial temperature for sulfurization of SnO₂ to SnS (Eq. (5)) is about 956.8 K, and the reaction between FeS and SnO₂ (Eq. (6)) occurs hardly since FeS is stable below 1558.3 K.

Fig. 4.

The standard Gibbs free energy changes of reactions of (3)–(6) with temperature.

Enthalpy changes of the related reactions at temperatures of interest are plotted in Fig. 5, and some transitions during the reaction process (see Fig. 3) are found to be driven by enthalpy. The FeS undergoes a phase transformation from FeS(s) into FeS(l) theoretically starting at 1463.0 K (see Fig. 5), as a result of which there is almost no change in weight but a strong exothermic peak at 1423.2 K in Fig. 3. Based on the SnS theoretical phase transitions at 1549.2 K in Fig. 5, SnS(l)→SnS(g), the weight-loss of 3.17 mass% occurring in the range of 1463.0–1533.0 K shown in Fig. 3, may be attributed to the evaporation of SnS(g).

Fig. 5.

Enthalpy changes of reaction of (3)–(6) with temperature.

4. Results and Discussion

In order to reduce hematite and achieve the purpose of removing tin, it is necessary to optimize the operating parameters of tin-bearing iron concentrates roasted with pyrite. For this purpose, the effects of roasting temperature, time, pyrite addition amounts and particle size of the tin-bearing iron concentrates were investigated.

4.1. Effects of Roasting Temperature and Time

Under N₂ flow rate of 5.00×10⁻⁶ m³/s, residence time of 60 min, pyrite addition amounts of 50.00 mass% and concentrates particle size of 0.075 mm, five roasting temperatures of 1073.0, 1123.0, 1273.0, 1373.0 and 1473.0 K were chosen for studying the effects on the iron phases reduction and tin removal rate.

The pyrite decomposes into Fe_1−xS and S_x starting at 703.8 K (see Fig. 3), and the iron phase’s reduction, shown as Eqs. (3) and (4), can be carried out at the roasting temperature (1073.0–1473.0 K) according to Fig. 4. XRD patterns of the roasted products at different temperatures are presented in Fig. 6 (‘Unroasted’ is tin-bearing iron concentrates without pyrite addition). It is obvious that temperature is an important factor for the reduction of hematite to magnetite. Temperature increase improves the reaction rate between hematite and decomposed products of pyrite. The peak intensity of magnetite (reduction products) gradually increases and that of hematite decreases with temperature (see Fig. 6), and the transformation is basically completed when temperature reaches 1473.0 K.

Fig. 6.

XRD patterns of the roasted products at different temperatures.

The pyrite cannot be further decomposed when temperature is relatively low,¹⁸⁾ and the formation of S is little, causing the tin removal reaction (Eq. (5)) also happens little at a low temperature. Figure 7 shows that tin removal rate increases slightly with temperature raised from 1073.0 to 1273.0 K, but it increases obviously to 64.29 mass% with temperature raised further to 1473.0 K. At a higher temperature, the formation of tin sulphide (SnS) is promoted, and some SnS (l) can be transferred to SnS (g) which is easy to volatilize, both being favorable to increasing tin removal rate. In order to reduce hematite and remove tin as much as possible, roasting temperature is fixed at 1473.0 K.

Fig. 7.

Effects of roasting temperature on the tin removal rate.

Figure 8 shows the hematite can partially be reduced into magnetite after being roasted for 40 min, and this transformation is completed with residence time prolonged further to 60 min. Figure 9 shows variations of tin removal rate as a function of roasting time. The tin removal rate increases gradually in the primary 60 min, from 1.79 mass% to 64.29 mass%, and then remains nearly constant. To decrease the process energy consumption and improve tin removal rate, residence time is fixed at 60 min.

Fig. 8.

XRD patterns of the products roasted for different time.

Fig. 9.

Effect of residence time on the tin removal rate.

4.2. Effects of Pyrite Amount

The FactSage 6.4 was used to calculate the equilibrium of roasted products in Gibbs free energy minimization under isothermal, isobaric and fixed mass conditions. Required data for computation were taken from FactPS database of the programme.

In Equilib program of FactSage, Fe₂O₃(s) (32.425 g), Fe₃O₄(s) (32.425 g), CaO(s) (1.55 g), SiO₂(s) (0.91 g), Al₂O₃(s) (0.87 g), SnO₂(s) (0.71 g), FeS₂(s) and N₂ (g) were selected as reactant precursors, where N₂ (g) was an inert gas to strict control the retained content of O₂ in the atmosphere. The calculations were performed in 100 g tin-bearing iron concentrates with variable pyrite amounts added under temperature of 1473.0 K and atmosphere pressure of 1 atm. The phases of iron sulfide (FeS), magnetite (Fe₃O₄), hematite (Fe₂O₃), cassiterite (SnO₂), tin sulphide (SnS), sulfur dioxide (SO₂) and element sulfur (S₂) were assumed to be present in the roasted products and results of phase’s equilibrium are present in Fig. 10. It shows that increase of pyrite amounts improves formation of Fe₃O₄ and SnS under pyrite amounts less than 5.00 mass%. When pyrite amounts increase over 5.00 mass%, the Fe₃O₄ amounts decrease and FeS amounts increase obviously.

Fig. 10.

Equilibrium amounts of species in roasted products as a function of pyrite amounts.

The mixtures of tin-bearing iron concentrates and pyrite at different quality ratios were roasted at 1473.0 K for 60 min. Figure 11 shows that almost all diffraction peaks are assigned to magnetite when pyrite addition is 5.00 mass% and the FeS phase appears and increases with pyrite amounts from 5.00 mass% to 100.00 mass%. These results are in agreement with those obtained from Fig. 10. When pyrite amounts exceed 5.00 mass%, some of the produced magnetite will be sulfurated to FeS, as a result of which magnetite amounts decrease and FeS amounts increase.

Fig. 11.

XRD patterns of the products with different pyrite addition amounts.

Figure 12 shows that tin removal rate increases sharply with pyrite addition from 0.00 mass% to 5.00 mass% and remains nearly constant as pyrite addition increases further. SnO₂(s) can be transformed into SnS (g) completely based on the equilibrium-analysis (see Fig. 10), and there will be no tin residue observed in the roasted products if SnS(g) volatilizes completely. However, the max tin removal rate is only about 72.86 mass% in the roasting experiments (see Fig. 12).

Fig. 12.

Effect of pyrite amounts on the tin removal rate.

With pyrite addition amounts of 50.00 mass%, the mixture was roasted at 1473.0 K for 60 min, and a scanning electron microscope provided with a detector of back scattered electrons (SEM-BSE) was used to characterize the products morphology. Some pictures were made (Figs. 13(a) and 13(b)), and EDS analyses (Fig. 14) for the selected areas were performed. Some holes can be observed in the products morphology (see Fig. 13(a)), and the Fe–O, Fe–O–Sn–S, and Fe–S–O phases are found to be formed at the surface of these holes deduced from Figs. 13(b) and 14, appearing to be spherical particles. Taking into consideration of equilibrium-analysis and effects of pyrite amounts on the tin removal rate, it could be stated that the formation of these holes is caused by the aggregation of gaseous SnS, SO₂ and S₂ within the mixture of iron sulfide (FeS) liquid and magnetite particles.^23,24)

Fig. 13.

SEM photos of the roasted products with pyrite addition amounts of 50.00 mass%.

Fig. 14.

EDS analyses of 1, 2 and 3 microareas of the SEM photos.

In order to decrease the formation amounts of FeS which hinders the evaporation of SnS(g), pyrite addition amounts are fixed at 5.00%.

4.3. Effects of Particle Size

Surface area of reaction particles increases as the particle size decrease, and it is beneficial to increasing iron reduction and tin sulfidation rates. Figure 15 shows phase composition of the products with 5.00 mass% pyrite addition and roasted for 60 min at 1473.0 K, the results show that peak intensity of the magnetite increases obviously while tin-bearing iron concentrates average particle size decreases from 0.425 mm to 0.075 mm. Obviously, the effect of its particle size on the tin removal rate is very small (shown in Fig. 16) due to its low content, although it has been changed during the reduction roasting process.

Fig. 15.

XRD patterns of the roasted products at different particle sizes of tin-bearing iron concentrates.

Fig. 16.

Effect of tin-bearing iron concentrates particle size on the tin removal rate.

As discussed above, the optimum parameters are established as follows: roasting temperature, 1473.0 K; roasting time, 60 min; pyrite amounts, 5.00 mass%; particle size, 0.075 mm. After roasted, the hematite in tin-bearing iron concentrates can be reduced to magnetite completely, meanwhile, the tin content is decreased from 0.27 mass% to 0.07 mass% (tin removal rate of 72.86 mass%). The roasted concentrates can meet the standard of BF ironmaking, which requires Sn content in the iron ores less than 0.08 mass%.⁶⁾ Besides, the tin removed can be recycled through smoke dust collection. It indicates the decomposed products of the pyrite can pre-reduce hematite into magnetite and achieve the purpose of Sn effective removal from tin-bearing iron concentrates.

5. Conclusions

The reduction of hematite to magnetite and effective removal of Sn from the tin-bearing iron concentrates through co-roasting with pyrite are feasible. The factors including pyrite amounts, roasting temperature, residence time and particle size of tin-bearing iron concentrates have important influences. The hematite in tin-bearing iron concentrates can be reduced to magnetite directly by FeS₂ above 961.9 K and indirectly above 703.4 K. The initial temperature for sulfurization of SnO₂ to SnS is about 956.8 K using FeS₂, and the reaction between FeS and SnO₂ occurs hardly since FeS is stable below 1558.3 K.

In a certain range, the tin removal and iron phases reduction rates both increase with roasting temperature, residence time and pyrite amounts. When pyrite amounts exceed 5.00 mass%, some of the produced magnetite will be sulfurated to FeS and the evaporation of SnS (g) will be obstructed. As a result, magnetite formation and tin removal rate decrease. The optimum conditions are established as follows: roasting temperature, 1473.0 K; roasting time, 60 min; pyrite amount, 5.00 mass%; particle size, 0.075 mm. After roasted under these conditions, the hematite in tin-bearing iron concentrates can be reduced into magnetite completely, meanwhile, the tin content is decreased from 0.27 mass% (unroasted sample) to 0.07 mass%. It can meet the standard of BF ironmaking, which requires Sn content in the iron ores less than 0.08 mass%, and the roasted concentrates can be used directly as blast furnace iron-making materials. In addition, the tin removed can be recycled through smoke dust collection.

Acknowledgments

The authors wish to express thanks to National Science Fund for Distinguished Young Scholars (51304092) for financial support of this research.

References

• 1)   C. X.  Xie,  H. M.  Li,  R. J.  Wang,  K. Y.  Xiao,  L.  Sun and  Y. L.  Liu: Geol. Bull. China, 28 (2009), 80 (in Chinese).
• 2)   J.  Liu and  S. Y.  Jin: China Min. Mag., 18 (2009), 1 (in Chinese).
• 3)   C.  Xie: Hunan Nonferr. Met., 11 (1996), 13 (in Chinese).
• 4)   T.  Jiang,  Y. B.  Zhang,  Z. C.  Huang and  G. H.  Li: Trans. Nonferr. Met. Soc. China, 15 (2005), 902 (in Chinese).
• 5)   Y. B.  Zhang,  T.  Jiang,  G. H.  Li,  Z. C.  Huang and  Y. F.  Guo: Ironmaking Steelmaking, 38 (2011), 617.
• 6)   Y. B.  Zhang,  Z. J.  Su,  Y. L.  Zhou,  G. H.  Li and  T.  Jiang: Int. J. Miner. Process., 124 (2013), 15.
• 7)   W. S.  Huang: Tin, Metallurgical Industry Publishing Company. Beijing, (2001), 291.
• 8)   G. Z.  Qiu,  T.  Jiang,  J. C.  Xu and  Y. Z.  Cai: Cold-bonded Pellet Direct Reduction, Central South University Press. Changsha, (2001), 184.
• 9)   D.  Huang,  Y. B.  Zhang,  G. H.  Han,  G. H.  Li and  T.  Jiang: 39th TMS Annu. Meeting and Expo., Vol. 1, TMS, Warrendale, PA, (2010), 393.
• 10)   T.  Jiang,  Y. B.  Zhang,  G. Q.  He,  G. H.  Li and  Z. C.  Huang: Proc. 24th Int. Mineral Processing Cong. (IMPC), IMPC, Beijing, (2008), 3952.
• 11)   Y. B.  Zhang,  G. H.  Li,  T.  Jiang,  Y. F.  Guo and  Z. C.  Huang: Int. J. Miner. Process., 110–111 (2012), 109.
• 12)   A. W.  Fletcher,  D. V.  Jackson and  A. G.  Valentine: Inst. Min. Metall. Trans, 76 (1967), 145.
• 13)   M. C.  Mayoral,  M. T.  Izquierdo,  J. M.  Andres and  B.  Rubio: Thermochim. Acta, 390 (2002), 103.
• 14)   P.  Baguley,  D. H.  John and  P. C.  Hayes: Metall. Trans. B, 14 (1983), 513.
• 15)   J. P.  Gaviría,  A.  Bohé,  A.  Pasquevich and  D. M.  Pasquevich: Phys. B, 389 (2007), 198.
• 16)   A.  Pineau,  N.  Kanari and  I.  Gaballah: Thermochim. Acta, 447 (2006), 89.
• 17)   P. K.  Strangway and  H. U.  Ross: Can. Metall. Q, 5 (1966), 221.
• 18)   G.  Hu,  K.  Dam-Johansen,  S.  Wedel and  J. P.  Hansen: Prog. Energy Combust. Sci., 32 (2006), 295.
• 19)   Y. Y.  Liu,  B. C.  Zhao,  Y.  Tang,  P. Y.  Wan,  Y. G.  Chen and  Z. J.  Lv: Thermochim. Acta, 588 (2014), 11.
• 20)   S.  Gornostayev,  J.  Härkki and  O.  Kerkkonen: Fuel, 88 (2009), 2032.
• 21)   P.  Walder and  A. D.  Pelton: J. Phase Equilib. Diffus., 26 (2005), No. 1, 23.
• 22)   L.  Li,  Z. J.  Qiu,  H.  Wang,  Y. G.  Wei,  B.  Liao and  X. L.  Sang: China J Nonferrous Met., 24 (2014), 519 (in Chinese).
• 23)   S.  Giraud,  T.  Duffar,  E.  Pihan and  A.  Fave: J. Cryst. Growth, 432 (2015), 83.
• 24)   H. F.  Gu,  Q.  Chen,  H. J.  Wang and  H. Q.  Zhang: Appl. Therm. Eng., 91 (2015), 938.