International Journal of Mineral Processing 138 (2015) 20–29
Contents lists available at ScienceDirect
International Journal of Mineral Processing journal homepage: www.elsevier.com/locate/ijminpro
Mechanism of vanadium slag roasting with calcium oxide Juhua Zhang a,⁎, Wei Zhang b, Li Zhang b, Songqing Gu c a b c
Key Laboratory for Ferrous Metallurgy and Resources Utilization of Ministry of Education, Wuhan University of Science and Technology, Wuhan 430081, PR China School of Materials and Metallurgy, Northeastern University, Shenyang 110819, PR China Zhenzhou Research Institute of China Aluminum Corporation, Zhenzhou 540001, PR China
a r t i c l e
i n f o
Article history: Received 9 March 2014 Received in revised form 24 January 2015 Accepted 16 March 2015 Available online 18 March 2015 Keywords: Vanadium slag Calcium roasting Oxidation Thermodynamics Kinetics
a b s t r a c t Thermodynamic analyses, XRD characterization and non-isothermal oxidation kinetic analyses were conducted to find the mechanism of vanadate formation when vanadium slag was roasted with calcium oxide. The effects of heating rate, added amount of CaO, holding temperature and holding time on oxidation efficiency were investigated. Thermodynamic calculations show that the fayalite is more likely to be oxidized than vanadium spinel, formation of calcium vanadates is easier than that of Mn(VO3)2 and Mg(VO3)2 and the oxidation of Fe2+ in augite is hindered by the presence of diopside. TGA results show that lowering the heating rate can improve the oxidation efficiency of vanadium. The maximum vanadium recovery of 93.3% was achieved when the vanadium slag with a ratio of m(CaO)/m(V2O5) of 0.42 was roasted at 850 °C for 2.5 h. Dynamic heating experiments indicate that oxidations of vanadium spinel and augite overlap within 608–959 °C with a heating rate of 3 °C·min−1, while only oxidation of spinel occurs within 657–914 °C at 5 °C min−1. The oxidation was controlled by a 3/2 reaction and a third order reaction, with corresponding overall apparent activation energy values of 140.3 and 247.8 kJ·mol−1 for the heating rates of 3 °C·min−1 and 5 °C·min−1, respectively. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Vanadium, as a very important transition metal, is widely used in various fields (Liao and Bo, 1985; Moskalyk and Alfantazi, 2003). At present, the primary vanadium resources are vanadium slag, stone coal, steel slag and spent catalyst. Vanadium slag which is a by-product of vanadium titanomagnetite accounted for more than 38% of the world's overall vanadium production in 2009 (Polyak, 2011). Processes comprising roasting, leaching, purification of aqueous solution, and precipitation steps are commonly employed in vanadium extraction from vanadium slag to produce vanadium pentoxide, of which, roasting plays a more important role in the whole process chain of vanadium. Since problems like difficulties in sodium salt recovery, utilization of tailings, and environmental pollution, exist in the conventional roasting process with sodium salts (i.e. Na2CO3, NaCl and Na2SO4), this process is not favored; hence, it is essential to develop some new processes for the clean production of vanadium oxide. Sub-molten salt (Liu et al., 2013), carbonate salt (Li et al., 2011) and compound salt roasting for vanadium extraction can decrease the amount of poisonous gas discharge; however, they cannot eliminate the hazard of sodium or potassium in the tailings and waste water. Calcium salt roasting was firstly proposed by the
⁎ Corresponding author. E-mail address:
[email protected] (J. Zhang).
http://dx.doi.org/10.1016/j.minpro.2015.03.007 0301-7516/© 2015 Elsevier B.V. All rights reserved.
Russia Tula factory in 1970s and it was not applied commercially due to lower recovery of vanadium compared with sodium salt roasting (Liao and Bo, 1985). However, it does have benefits of lower cost of additive, no emission of pollutional gas, and no sodium or potassium contained in tailings and waste water, so this process is of increasing interest. Vanadium slag roasting with calcium oxide is an oxidation process. Vanadium embedded in the slag presents as V3 +, and with roasting, V3+ is oxidized to V4+ and V5+ and converted to vanadate which can be dissolved in a subsequent leaching step. Many investigations (Cao, 2012; Li et al., 2012; Li, 2011) have focused on the effects of roasting parameters, including the ratio of m(V2O5)/m(CaO), roasting temperature and roasting time, on the vanadium recovery. Van Vuuren and Stander (1995) studied the oxidation kinetics of synthetic FeV2O4 over the temperature range of 200–580 °C and the oxidation of synthetic FeV2O4 in a sodium carbonate mixture (Van Vuuren and Stander, 2001). Voglauer et al. (2004) investigated the reaction kinetics of vanadium roasting process in steel slag. In this paper, the thermodynamics of a converter vanadium slag roasting process with calcium oxide was analyzed. Effects of heating rate, added amount of CaO, holding temperature and holding time on oxidation and recovery of vanadium, and experimentally the relationships between the heating rate and the optimum holding temperature were studied. The roasted samples prepared under different roasting conditions were characterized by X-ray diffraction (XRD). The oxidation kinetic equations and apparent activation energy values for a vanadium slag roasting process in the presence of calcium
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
21
oxide were obtained using differential scanning calorimetry and thermal gravimetric (DSC–TG) methods. 2. Experimental 2.1. Materials The chemicals used in this study (CaO, H2SO4, (NH4)2Fe(SO4)2· 6H2O, C13H11NO2, H2NCONH2, NaNO2 and KMnO4) were all of analytical grade (Sinopharm Chemical Reagent Co., Ltd, purity N 98%). CaO was used as the additive for vanadium slag roasting and was dried in an oven at 120 °C for 24 h before use. The other chemicals were used to determine vanadium content in the slag with ammonium ferrous sulfate titration method. Water used in experiments was deionized one time. Vanadium slag was supplied by Sichuan Weiyuan Iron & Steel Co., Ltd. China (after magnetic separation). The mass fraction of the raw slag with particle size larger than 250 μm was 16.3%, and after screening, the slag with particle size within 48–75 μm was chosen for roasting experiments in this study. The mineralogical and chemical analyses of this slag were conducted using X-ray diffraction (XRD, X'Pert Pro MPD, PAnalytical B.V. of Netherlands), SEM-EDS (SSX-550, Shimadzu Corporation), X-ray fluorescence spectroscopy (S-MAX, Rigaku) and ICP-OES (Optima 4300DV, American PerkinElmer Company). Chemical compositions of the slag were listed in Table 1 and the X-ray diffraction pattern was shown in Fig. 1. The backscatter image and EDS analysis results of the slag were given in Fig. 2 and Table 2, respectively. Table 1 shows that Fe is the richest element in the slag and then decreasing in concentration are V, Si, Ca, Mn, and Ti successively. The main mineral phases in the converter slag are vanadium spinel, fayalite and augite as shown in Fig. 1. According to the EDS analysis results (Table 2) it is known that the white area in Fig. 2 is the vanadium bearing phase-vanadium spinel [(Mn, Fe) (V, Cr)2O4]; the gray area and the black area are the matrix phases–fayalite (Fe2SiO4) and augite [Ca (Fe, Mg)Si2O6], respectively–dispersed around the spinel phase. The silicon content in augite is higher than in fayalite; the dark gray areas (D and E) are Ca3P2O8 and Ca2P2O7. Since the content of P in slag is only 0.22%, calcium phosphate phases occupy a small fraction of the sample surface area. The volume fractions of vanadium spinel, fayalite and augite are 32.74%, 49.52% and 17.74% respectively.
Fig. 1. X-ray diffraction pattern for vanadium slag.
the corundum work tube were directly open to atmosphere to ensure that the pellets were thoroughly exposed to air. The pellets were heated according to a preset heating system. When the holding time was completed, the roasted pellets were quenched to room temperature. The roasted pellets were ground to fine powder and characterized by XRD. Roasting efficiency was expressed by vanadium recovery upon leaching, and the leaching conditions were liquid to solid ratio of 4:1, temperature of 65 °C, time of 1 h, pH of 2.5, and stirring speed of 500 rpm. The vanadium recovery was calculated by the following equation:
η; % ¼
m0 w0 −m1 w1 100% m0 w0
where η represents the vanadium recovery, m0 is the mass of roasted slag, w0 is the mass fraction of vanadium in the roasted slag, m1 is the mass of the tailing after leaching and w1 is the mass fraction of vanadium in the tailing. The mass fraction of vanadium in the roasted slag and in the tailing was determined by ammonium ferrous sulfate titration. 3. Thermodynamic analysis for roasting process
2.2. Experimental apparatus Roasting tests were carried out in a vertical tube furnace controlled by a Shinaden SR-53 temperature programmed instrument with molybdenum disilicide heating elements as shown in Fig. 3. The isothermal section was about 90 mm long and the temperature variation was within ±1 °C. Leaching experiments were conducted in glass vessels centered in a thermostatic water bath (HH-4, China Changzhou Sino Instrument Co., LTD) and the pH values were measured by a pH-meter (OHAUS Starter 3C). The dynamic oxidation experiments were carried by TG–DSC (STA409CD, NETZSCH German).
Vanadium spinel surrounded by matrix phases—fayalite and augite, is the main vanadium-bearing phase in converter vanadium slag. In order to obtain a high oxidation efficiency of vanadium, the structure of the binding phases should be destroyed firstly to liberate the vanadium spinel and then it is easier for oxygen to contact with vanadium spinel directly. The trivalent vanadium is oxidized into pentavalent vanadium and then converted to dilute acid soluble vanadate.
2.3. Experimental procedure The vanadium slag (48–75 μm) was mixed with additive of CaO and prepared into Ф 8–10 mm pellets. In the roasting experiments, only a few pellets were put in the corundum crucible and the two ends of Table 1 Chemical composition of vanadium slag (mass fraction, %). MFe
SiO2
CaO
MnO TiO2 Cr2O3 MgO
14.3 24.84 Al2O3 P2O5 2.1 6.4 ×
V2O5
FeO
2.29 SO3 1.9 ×
14.3 Na2O 1.0 ×
9.9 K2O 3.9 ×
8.5
10−1
10−1
10−1
10−2
7.4
4.4
3.7
Fig. 2. Backscattered micrograph of the vanadium slag.
22
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
Table 2 EDS analysis of the vanadium slag. Point in Fig. 2
Elements (wt.%)
A B C D E
O
V
Cr
Fe
Mn
Ti
Mg
Si
20.99 24.17 37.01 34.70 37.90
24.40
8.31
22.14 18.96 8.86
15.37 23.96 7.24
5.60
1.58 2.08
3.09
2.63
0.41 17.48 27.01 3.29 8.31
Reactions possibly occurring in the roasting process in the presence of calcium oxide are given as follows. The corresponding thermodynamic data are from Barin (1995). (1) Oxidation of metal iron and fayalite
Ca
Al
P
1.22 13.33 8.29 40.04 29.94
11.60 3.10
21.97 17.02
(3) Formation of vanadate V2 O5 þ CaO ¼ CaV2 O6
ð14Þ
V2 O5 þ 2CaO ¼ Ca2 V2 O7
ð15Þ
2Fe þ O2 ¼ 2 FeO
ð1Þ
V2 O5 þ 3CaO ¼ Ca3 V2 O8
ð16Þ
3=2 Fe þ O2 ¼ 1=2Fe3 O4
ð2Þ
4 4 2 4 FeV2 O4 þ O2 þ CaO ¼ Fe2 O3 þ CaV2 O6 5 5 5 5
ð17Þ
3Fe2 SiO4 þ O2 ¼ 2 Fe3 O4 þ 3SiO2
ð3Þ
V2 O5 þ MgO ¼ MgV2 O6
ð18Þ
2Fe2 SiO4 þ O2 ¼ 2 Fe2 O3 þ 2SiO2
ð4Þ
MnO þ V2 O5 ¼ MnV2 O6 :
ð19Þ
4Fe3 O4 þ O2 ¼ 6Fe2 O3 :
ð5Þ
(2) Oxidation of vanadium spinel
(4) Oxidation and decomposition of augite 4CaFeðSiO3 Þ2 þ O2 ¼ 2Fe2 O3 þ 4CaSiO3 þ 4SiO2
ð20Þ
4FeV2 O4 þ O2 ¼ 2 Fe2 O3 þ 4V2 O3
ð6Þ
CaMgSi2 O6 ¼ MgSiO3 þ CaSiO3
ð21Þ
4MnO þ O2 ¼ 2Mn2 O3
ð7Þ
CaO þ SiO2 ¼ CaSiO3
ð22Þ
2Mn2 O3 þ O2 ¼ 4MnO2
ð8Þ
2CaO þ SiO2 ¼ Ca2 SiO4
ð23Þ
2=3Cr2 O3 þ O2 ¼ 4=3CrO3
ð9Þ
3CaO þ SiO2 ¼ Ca3 SiO5
ð24Þ
CaO þ Al2 O3 þ SiO2 ¼ CaAl2 SiO6 :
ð25Þ
4=3CaO þ 2=3Cr2 O3 þ O2 ¼ 4=3CaCrO4
ð10Þ
2V2 O3 þ O2 ¼ 2V2 O4
ð11Þ
2V2 O4 þ O2 ¼ 2V2 O5
ð12Þ
Fe2 O3 þ TiO2 ¼ Fe2 TiO5 :
ð13Þ
1 2 3
8 11 9
4
10 5 6
7
Fig. 3. Schematic diagram of roasting reactor (1—iron wire; 2—heating element; 3—furnace tube; 4—crucible; 5—pellets; 6—corundum tube; 7—furnace base; 8—furnace shell; 9—thermocouple; 10—temperature control instrument; 11—iron chrome alloy silk).
In pure oxygen atmosphere, when P O2 is 100 kPa, the thermodynamic calculations are illustrated in Figs. 4 and 5. Fig. 4 shows that except for the oxidation of Cr3+ (reactions 9 and 10), decomposition of diopside (CaMgSi2O6), formation of Fe2TiO5, oxidation of Mn2O3 to MnO2 and formation of manganese vanadate (Mn(VO3)2), the standard gibbs free energy changes ΔrGo for other reactions occurring in the process of vanadium slag roasting are negative over the temperature range of 100–1000 °C, which indicates that the oxidation of fayalite and vanadium spinel, and the formation of calcium vanadates are thermodynamically feasible. In Fig. 4(a), ΔrGo for oxidation of metal iron is the most negative indicating that this reaction is the easiest to happen. ΔrGo values for reactions (3) and (4) are more negative than those for reaction (6) and this means that oxidation of fayalite is easier than that of vanadium spinel. Divalent manganese is possible to be oxidized to trivalent manganese and reacts with V2O5 to generate Mn(VO3)2 at temperatures lower than 600 °C. Trivalent manganese only can be oxidized to Mn4 + at temperatures lower than 500 °C. Reaction between Cr3 + in spinel and oxygen within 100– 1000 °C is not possible, but Cr3 + may be converted to CaCrO4 in the presence of CaO at temperatures lower than 800 °C The standard gibbs free energy for reaction (13) becomes negative when the roasting temperature is higher than 600 °C, however the absolute value is smaller than 3 kJ mol− 1, which indicates that the limitation of reaction is small. Fig. 4(b) illustrates that ΔrGo values for formations of calcium vanadates are more negative than for manganese vanadate (Mn(VO3)2) and magnesium vanadate (MgV2O6) and the stabilities
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
23
Fig. 4. ΔrGo-T diagrams for reactions in oxidation process at PO2 of 100 kPa (a) reactions 1—13 and (b) reactions 14–25.
for those three calcium vanadates follow an order of CaV 2 O6 b Ca3V2O8 b Ca2V2O7 at temperature above 500 °C when enough CaO is present. The ΔrGo value for the oxidation and calcification reaction (reaction 17) of vanadium spinel is more negative than these values of formations of vanadates (reactions 14–16), and then it is deduced that oxidation and calcification of vanadium may occur simultaneously in roasting process. Formation of Mn(VO3)2 is possible to proceed at temperature lower than 600 °C. Hedenbergite (CaFeSi2O6) and diopside (CaMgSi2 O 6) are completely isomorphous and form into a mixed crystal-augite (Ca(Fe,Mg)Si2O6). Though the ΔrGo value for CaFeSi2 O6 oxidation is negative, this reaction may be hindered since CaMgSi2 O 6 cannot decompose in the temperature range of 100–1000 °C. Silicon dioxide produced from the decomposition of fayalite can react with Al2 O3 and with CaO to form silicate. Fig. 5 shows that oxidation and formation of vanadates are exothermic and increasing roasting temperature will decrease the barriers to these reactions. When the roasting process is conducted in air and P O2 equals 0.21 × 105 Pa, the calculated gibbs free energy for reactions where oxygen is involved is shown in Fig. 6. The values of ΔrG for all the oxidation reactions become less negative with a decrease in oxygen partial pressure. Apart from reactions 8, 9 and 10, ΔrG values for other reactions are still negative in temperature ranging from 100 to 1000 °C. Compared with Fig. 4(a), the temperature points at which ΔrG values of reactions 8 and 10 change from the negative to the positive are decreased from 517 to 482 °C and from 801 to 766 °C, respectively.
4. Experimental results and discussions 4.1. Effects of heating rate and holding temperature It was found previously that the cooling rate of roasted slag had little effect on vanadium recovery (Cao, 2012). This paper presents the effect of heating rate on vanadium recovery and the relationships between heating rates and the obtained optimum roasting temperatures under the conditions where 6% CaO was added (assuming the vanadium slag weight as 100%), particle size was within 48–75 μm and holding time was 2 h. The results are shown in Fig. 7. Reducing the heating rate from 4 to 2 °C per minute is beneficial to improve the vanadium leaching rate and the maximum vanadium recovery increased by around 7%. The optimum roasting temperature changed from 800 to 850 °C. This can be attributed to the reactions that take place within lower temperature range proceeding more thoroughly by decreasing the heating rate, and it means that the matrix phases are oxidized more extensively and then vanadium spinel becomes exposed and easier to contact with oxygen, which gives rise to a high oxidation efficiency of vanadium. Roasting temperature is a key factor determining the oxidation degree of vanadium. When the roasting temperature is lower, oxidation and calcification of vanadium are insufficient. Conversely, a higher temperature may lead to the burden sintering. As illustrated in Fig. 7, recovery of vanadium increases slowly from 400 to 600 °C — that is to say, the oxidation rate of vanadium is low. At temperatures over 600 °C, oxidation rate increases and vanadium recovery rises considerably. The
21
0
(a)
0
21
(b) 18
18
25
25
-100
22
-100
23
22
23 24
24
-200 -1
14
19
-300
15
20
-500
14
19
-300
15
-400
o
-400
Δ rH ,KJ·mol
o -1 Δ rH ,KJ·mol
-200
17
20
-500
17
-600
-600 16
-700 0
200
400
600
Temperature,°C
800
16
-700 1000
0
200
400
600
800
Temperature,°C
Fig. 5. ΔrHo-T diagrams for reactions in oxidation process at PO2 of 100 kPa (a) reactions 1–13 and (b) reactions 14–25.
1000
24
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
Fig. 6. ΔrG-T diagram for reactions where oxygen is involved at O2 partial pressure of 0.21 × 105 Pa.
maximum of 91.5% occurs at 850 °C. When the temperature further increases, the experimental sample sinters and densifies. This is not conducive to the diffusion of oxygen to the unreacted solid; meanwhile, an insoluble eutectic silicate forming in this higher temperature range may coat the vanadate and result in a low vanadium recovery upon leaching (He et al., 2007). Fig. 8 shows the XRD patterns of vanadium slag roasted at different temperatures for 2 h at heating rates of 4 and 2 °C·min−1 respectively. After roasting at 450 °C for 2 h at a heating rate of 4 °C·min−1, fayalite is not completely oxidized and decomposed, and only partially oxidized into magnetite (Fe2.933O4) and hematite (Fe2O3). When the holding temperature increases to 550 °C, it is completely oxidized to Fe2O3 and Fe2.938O4. Similarly, Fig. 8(b) shows that the temperature for complete fayalite oxidation is higher than 500 °C when a heating rate of 2 °C·min−1 is used. Oxidation of vanadium spinel and formation of vanadate take place at temperatures above 600 °C, which agrees well with the results in Fig. 7. CaV2O5 and V5O9 still exist at 800 °C with a heating rate of 4 °C·min−1 while low valent vanadium (V3+ and V4+) cannot be found at 800 °C with a heating rate of 2 °C·min−1. Fe2+ in the spinel is oxidized to Fe2O3. Oxidation and calcification of vanadium in spinel proceed simultaneously because low valent vanadium oxides and low valent vanadate coexist in the roasted samples. Meanwhile, partial Mn2+ reacts with V2O5 to generate manganese orthovanadate (Mn3(VO4)2). During roasting a large amount of SiO2 converts to aluminosilicate and calcium silicate. In Fig. 8(c), the peak intensity of iron oxide becomes stronger gradually with increasing holding temperature due to the gradual oxidations of vanadium spinel and augite. Diopside (CaMg(SiO3)2) is chemically stable in temperature range from 400 to 1000 °C in accordance with the
100 -1
2 °C min -1 4 °C min
Vanadium leaching rate,%
•
80
•
60 40 20
calculation in Fig. 4(b). Fig. 8(a) shows that augite still exists at 1000 °C. Vanadate cannot be found in the XRD pattern of Fig. 8(c) at 1000 °C since the SEM backscatter images show that they are coated by a large amount of silicate with low melting point which results from the oxidation of augite. 4.2. Effect of the amount of CaO on vanadium recovery The effect of the added amount of CaO on vanadium recovery at different holding temperatures in the roasting–acid leaching process is shown in Fig. 9, where the content of CaO in raw converter vanadium slag is excluded. Fig. 10 gives the XRD patterns of the samples roasted at 850 °C for 2 h with different ratios of m(CaO)/m(V2O5) (m(CaO) referring to the mass of the added CaO). Fig. 9 shows that when the mass ratio of CaO to V2O5 increases from 0.07 to 0.84 the vanadium recovery increases slowly followed by a gradual decrease. The maximum value of 91.46% with a heating rate of 2 °C·min−1 occurs at the ratio of 0.42 when the roasting temperature is 850 °C. The optimum holding temperature changes to 800 °C with a heating rate of 4 °C·min−1 and the same trend is observed, the vanadium recovery increases from 85.23% to 85.78% and then decreases to 82.68% with an increase in mass ratio of CaO to V2O5 from 0.04 to 0.42 and then to 0.84. When the sample is roasted at 900 °C for 2 h using a heating rate of 4 °C·min−1, the vanadium recovery rises from 80.19% to around 81% as the ratio increases from 0.07 to 0.21, levels off within the mass ratio range of 0.21 to 0.63, and then decreases rapidly when the ratio increases further. The vanadium recovery was higher at 850 °C with a heating rate of 2 °C·min− 1 than that obtained when roasting at 800 °C and 900 °C with a heating rate of 4 °C·min−1 over whole mass ratio range of CaO to V2O5 examined in this study, which further confirms that lowering heating rate can improve the vanadium recovery. When only slag is roasted with no addition of CaO, the main vanadate Mn2V2O7 (Fig. 10) that can be dissolved in dilute acidic solution forms. As CaO is added to a value of 0.42, Ca2V2O7 and Mg2V2O7 are generated, and calcium silicate occurs as well. When more CaO is added up to a mass ratio of 0.84, a composite vanadate is produced and more low melting calcium silicate which coasts the vanadate during cooling to room temperature, leading to a decrease in vanadium recovery on leaching. 4.3. Effect of holding time on vanadium recovery
0 400
500
600
700
800
900
1000
T, °C Fig. 7. Effects of heating rate and holding temperature on vanadium leaching rate.
Under the conditions of m(CaO)/m(V2O5) of 0.42, heating rate of 2 °C·min−1, the effect of holding time at different temperatures on vanadium recovery is shown in Fig. 11 and the XRD patterns of products roasted at 850 °C for different reaction time are shown in Fig. 12. In the roasting process, vanadium spinel is firstly oxidized into tetravalent oxide or pentavalent oxide and then converts to calcium
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
25
100 2˚C/min,850˚C 4˚C/min,900˚C 4˚C/min,800˚C
95
Vanadium recovery, %
90 85 80 75 70 65 60 0.0
0.2
0.4
0.6
0.8
1.0
m(CaO)/m(V2O5) Fig. 9. Effect of the added amount of CaO on vanadium recovery with different heating rates.
value of 90% at 850 °C in 15 min and then the vanadium recovery declines due to the successive formations of different vanadates, and then increases again due to nearly equilibrium after 100 min. The reaction products (Kowalski et al., 1995) for CaO and V2O5 at temperature lower than 995 °C exist in three phases, CaV2O6, Ca2V2O7 and Ca3V2O8, and the solubility for these three products is different at the experimental leaching pH and temperature. Apart from the calcium vanadates, some other vanadates or composite vanadates with Mn, Fe, Mg and Ca are also generated (Chen, 1995). At 800 °C, the recovery of vanadium rises slowly from high value of 82% in 90 min and then presents the same trend as observed at 850 °C with increasing holding time. Over all, vanadium recovery at 850 °C only increases by 2.76% with increasing the time from 15 to 150 min. This phenomenon is very different from that obtained in other investigations (Li et al., 2012; Li, 2011; Yin et al., 2012) where vanadium recovery increased dramatically with time in 120 min, 60 min and 150 min respectively at the optimum roasting temperatures. Because in the latter cases, samples were not put into the roasting reactor until the optimum temperature reached and the sample did not experience the lower temperature roasting. This further verifies that lengthening the heating time from 550 to 850 °C can accelerate the oxidation kinetics of vanadium spinel, and this is of interest for energy saving in the whole roasting process in practical production. 4.4. Oxidation kinetics of vanadium slag roasting process
Fig. 8. XRD patterns of slag after roasted at stated temperatures for 2 h ((a)—at heating rate of 4 °C·min−1; (b) and (c)—at heating rate of 2 °C·min−1).
vanadate. Roasting time has a great effect on the oxidation efficiency, and if the holding time is short, low valent vanadium (V3+ and V4+) may not be oxidized to V5+ completely. Fig. 11 shows that the recovery of vanadium at 850 °C is much higher than that at 700 °C, which indicates that the oxidation rate of vanadium is much larger at 850 °C. The recovery of vanadium at 700 °C increases gradually from 60% to 73% with extending holding time from 15 to 180 min, while the oxidation of vanadium reaches a
Dynamic heating experiments (TG–DSC) were conducted under the conditions of 10–12 mg vanadium slag with particle size of 48–75 μm, m(CaO)/m(V2O5) of 0.42 and air flow rate of 50 ml·min−1. The obtained TG–DSC curves are given in Fig. 13. From the curves of TG–DSC at a heating rate of 3 °C·min− 1 (Fig. 13(a)), there are three stages in oxidation process. There is an endothermic peak and the sample weight decreases in the temperature range of 337–431 °C, and then the weight increases by 0.2% over the temperature range of 431–560 °C. When the temperature is higher than 600 °C, a dramatic weight gain of 4.8% and a broad exothermic peak occur simultaneously. According to the thermodynamic analyses, oxidations of fayalite and vanadium spinel are both exothermic and lead to the increase in weight, so it is deduced that crystal water was removed in the first stage. Fayalite was oxidized in the second stage of 431–560 °C. According to the chemical composition analysis of the slag, the theoretical weight gain percentages for Fe2+ and V3 + were 3.52% and 2.36% respectively, and for vanadium spinel (FeV2O4) was 3.0%. Ferrous ions are distributed in fayalite, vanadium spinel and augite phases, and so vanadium spinel oxidation and augite oxidation overlap over the temperature range of 608–959 °C with a weight gain of about
26
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
m
m
m
m
m
m
Fig. 10. XRD patterns of slag roasted at 850 °C for 2 h with different amounts of additive.
4.8%. Compared with the curves for heating rate of 3 °C·min−1, the TG– DSC curve for 5 °C·min−1 presents the same shape but the starting and ending temperatures for the first two stages shift to higher temperature ranges of 399–432 °C and 433–565 °C respectively due to the thermal hysteresis with increasing heating rate. The third stage proceeds at temperature ranging from 657 to 914 °C with weight gain of 3% which is in good agreement with the theoretical weight gain of vanadium spinel oxidation. The oxidation of vanadium slag is considered to be a non-isothermal heterogeneous gas–solid reaction. In a gas–solid reaction aA(s) + bB(g) = cC(s), when the reaction rate is controlled by substance A, it can be given by: dα n ¼ k f ðα Þ ¼ kð1−α Þ dt
where α is the fraction of reacted solid reactant at time t, n the order of reaction and k the reaction rate constant expressed by the Arrhenius equation: −Ea =RT
k ¼ Ae
where T represents the Kelvin temperature, Ea the apparent activation energy, R the gas constant, and A the pre-exponential factor. For a linear heating, the heating rate φ is usually a function of time: φ¼
dT : dt
ð28Þ
ð26Þ
95
Vanadium leaching rate,%
90 85
850˚C 800˚C 700˚C
80 75 70 65 60 0
50
100
150
200
250
t, min Fig. 11. Effect of holding time at different temperatures on vanadium recovery.
ð27Þ
Fig. 12. XRD patterns of slag roasted at 850 °C for different time.
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29 105
0.3 0.0
102
-0.3 101 -0.6
DSC
Exo. 1.5
1.0
103
102
0.5
101
DSC
100
0.0
DSC,mw·mg-1
103
(b) 104
TG,%
TG,%
0.6
105
Exo.
DSC,mw· mg -1
0.9
(a) 104
27
100
-0.9
TG
TG
99 100
200
300
400
500
600
700
800
900
-0.5
99
-1.2 1000
100
200
300
Temperature, ˚C
400
500
600
700
800
900
1000
Temperature, ˚C
Fig. 13. TG–DSC curves for oxidation of vanadium slag at different heating rates (a)—3 °C·min−1; (b)— 5 °C·min−1.
Combining Eqs. (27) and (28) with Eq. (26), we get: dα A −Ea=RT n ¼ e ð1−α Þ dT φ α¼
ð29Þ
mt −m0 m∞ −m0
ð30Þ
where m0, mt and m∞ represent the sample weight at the starting, time t and completion of a reaction, respectively. Separation of variables and integration of Eq. (29) yield: Zα α0
dα A ¼ f ðα Þ β
ZT expðEa =RT ÞdT ¼ g ðT Þ:
ð31Þ
T0
The right-hand expression of Eq. (31) has no exact integral, and many researchers have approached the analytical solutions with some mathematical simplifications. The Coats–Redfern (Coats and Redfern, 1964) equation is: 1−ð1−α Þ1−n ART 2 2RT −Ea=RT ¼ 1− : e Ea 1−n αEa
Fitting the TG data for heating rate of 3 °C·min−1 and 5 °C·min−1 using kinetic equations with different n, the results are displayed in Table 3 and the optimum results of linearization are drawn in Fig. 14. Table 3 shows that when the heating rate was 3 °C·min−1, the overall oxidation reaction of vanadium spinel and augite occurring within 608–959 °C was controlled by a 3/2 order reaction and the overall apparent activation energy Ea and the frequency factor A were 140.2 kJ·mol− 1 and 3.0 × 105 min− 1 calculated on the basis of the slope and intercept of the fitting curve (1) in Fig. 14. When the heating rate increased to 5 °C·min− 1, only oxidation of vanadium spinel happened in the temperature range of 657–914 °C and was controlled by a third order reaction. The corresponding overall apparent activation energy Ea and the frequency factor A were 247.8 kJ·mol−1 and 1.0 × 1012 min−1 respectively. The left-hand side of Eq. (31) can be integrated and the right-hand side can be solved by recursive method with a computer program. Supposing the activation energy, reaction order and the pre-exponential factor are given, the numerical solution of Eq. (31) is obtained. According to the Taylor's equation, 0
ð32Þ
g ðT 0 þ ΔT Þ≈g ðT 0 Þ þ g ðT 0 ÞΔT þ
1 ″ 2 g ðT 0 ÞΔT 2!
ð35Þ
Where: Taking natural logarithm yields: 0
″
" # 1−ð1−α Þ1−n AR 2RT E 1− − a: when n≠1; ln ¼ ln φEa Ea RT T 2 ð1−nÞ
g ðT Þ ¼ ð34Þ
AEa expð−Ea =RT Þ : βRT 2
ð37Þ
2
(2)
Heating rate, °C·min−1
Temperature range, °C
The order of reaction, n 1/3
3/4
1
3/2
2
3
3 5
608–959 657–914
0.988 0.969
0.995 0.982
0.998 0.987
1.000 0.994
0.999 0.998
0.994 1.000
The correlation coefficient (R2) values of linear regression marked in bold (Table 3) show the best results of linearization procedure for TG data at 3 °C•min−1 and 5 °C•min−1 respectively.
2
ln{2[(1-a)
Table 3 Values of R2 for TG data fitting with different kinetic equations (0.25 ≤ α ≤ 0.75).
-12
-12
-13
-13
(1)
-1/2
and intercept of ln(AR/φEa), and then the activation energy and preexponential factor can be calculated.
ð36Þ
With such a numerical solution method, the calculated values of oxidation ratio α against t (or T) can be plotted.
-1] / T }
For most reactions, E is much larger than RT in common reaction temperature range, therefore, the first part in the right side of equation h i Þ is approximately regarded as a constant value. A plot of ln − lnTð1−α or 2 h i 1−n Þ ln 1−Tð21−α against T1 should result in a straight line with slope of –Ea/R ð1−nÞ
A expðEa =RT Þ β
-2
ð33Þ
g ðT Þ ¼
ln{[(1-a) -1] / 2T }
− ln ð1−α Þ AR 2RT E when n ¼ 1; ln 1− ¼ ln − a φEa Ea RT T2
-14
-14
-15
-15 0.00090
0.00095
0.00100
0.00105
-1
1/T, K
Fig. 14. Optimum fitting of linearization for TG data ((1)—3 °C·min−1; (2)—5 °C·min−1).
28
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
100
(a)
Experimental data
α, %
80
60
40
Calculated data
20
0 600
700
800
900
1000
Temperature, ˚C 100 90
(b)
80
α, %
70 60 50 40
Experimental data
30 20
Calculated data
10 0 700
750
800
850
900
Temperature, ˚C Fig. 15. Comparisons between the experimental data and the data calculated with the obtained kinetic parameters (a)—3 °C·min−1 and (b)—5 °C·min−1.
According to the above theoretical basis, the chosen kinetic equations in Fig. 14 can be further verified by comparing the experimental data and calculated ones on the basis of the obtained kinetic parameters, and the results are illustrated in Fig. 15. Fig. 15 shows that the calculated data are in good agreement with the experimental ones, which indicates that the obtained kinetic equations can best describe the vanadium roasting process at heating rates of 3 °C·min−1 and 5 °C·min−1 in corresponding temperature ranges. 5. Evaluation of the results Using calcium roasting process to extract vanadium from converter slag was proposed in the 1970s, however, an understanding of the roasting mechanism of converter slag with CaO is still not clear and needs to be developed. In this paper, from the thermodynamic analyses, it is found that metal iron and fayalite should be oxidized thermodynamically more easily than vanadium spinel. Some Mn2 + in the slag can be oxidized to Mn2O3, while other Mn2+ can react with V2O5 to form Mn(VO3)2 at temperatures lower than 600 °C, so Mn2+ will transfer into the solution if acid leaching is used to extract vanadium from the roasted slag. Cr3+ in spinel is impossible to be oxidized to Cr6+ within the temperature range of 100–1000 °C by oxygen, but it can react and oxidize with CaO to form CaCrO4 in the presence of CaO at temperatures lower than 800 °C in oxygen. When the oxygen partial pressure decreases to 0.21 × 105 Pa, the temperature for the formation of CaCrO4 shifts to a lower temperature range, i.e., lower than 766 °C. It is more difficult for Cr3 + to transform to Cr6 + in a calcium roasting process compared
with sodium roasting, where the gibbs free energy for formation of Na2CrO4 is more negative and the Kp of this reaction is larger. In comparison with manganese vanadate and magnesium vanadate, the formation of calcium vanadates is more favorable thermodynamically, and if enough CaO is available, Ca2V2O7 can form at 500 °C above, which is beneficial to extracting vanadium from the slag with sulfuric acid solution of pH 2.5. Lowering heating rate can promote the oxidation of augite (Ca(Fe,Mg)Si2O6) as shown in Figs. 8 and 13 and make the optimum holding temperature shift to high temperature region, meanwhile, the corresponding maximum vanadium recovery is increased by about 7% when the heating rate is decreased from 4 °C·min− 1 to 2 °C·min− 1. This result is further confirmed by the experimental results in Sections 4.2 and 4.3. Over the ratio range of m(CaO)/ m(V2O5) stocked from 0.07 to 0.84, the vanadium leaching rate at 850 °C with heating rate of 2 °C·min − 1 was higher than those at 800 and 900 °C with a heating rate of 4 °C·min− 1. In addition, a vanadium recovery of over 90.5% was obtained in just 15 min at 850 °C when a heating rate of 2 °C·min−1 was used. This finding is of significance for practical production where roasting is always conducted in multi-heart roaster or rotary kiln. Extending the reaction time in lower temperature range (especially from 550 °C to the optimum holding temperature) not only increases the vanadium recovery but also shortens the roasting time at higher temperature, and overall, the whole roasting time and energy consumption are reduced in the roasting stage. In roasting process, oxidation and calcification of vanadium occur simultaneously, and low valent vanadium oxides and low valent calcium vanadate co-exist in the roasted slag. When the addition of CaO is not enough, V2O5 will react with MnO and MgO to form manganese vanadate and magnesium vanadate. With an increase in the addition of CaO, Mn and Mg incorporated in vanadates will be replaced by Ca and more calcium vanadates generate, especially Ca2V2O7, which is consistent with the thermodynamic calculation. 6. Conclusions (1) The roasting process of vanadium slag with lime was investigated. Thermodynamic analysis predicts that oxidation of fayalite is more favorable than that of vanadium spinel; formation of calcium vanadates is energetically easier than that of manganese vanadate (MnV2O6) and magnesium vanadate (MgV2O6); oxidation of Fe2+ to Fe3+ in augite is hindered by the chemically stable diopside. Oxidation of fayalite, spinel and augite and formation of vanadates are all exothermic reactions. (2) Lowering heating rate from 4 °C·min−1 to 2 °C·min−1 can improve the vanadium recovery, to a value of 91.46% and change the optimum roasting temperature from 800 to 850 °C. A satisfactory vanadium recovery was obtained in 15 min at 850 °C with a heating rate of 2 °C·min−1 and the maximum recovery occurred when the reaction time was extended to 150 min. (3) Kinetic analyses indicate that when the heating rate is 3 °C·min−1, oxidations of vanadium spinel and augite overlap in the temperature range of 608–959 °C while raising heating rate to 5 °C·min−1 oxidation of spinel occurs in a temperature range of 657–914 °C. They are controlled by the rate of a 3/2 reaction and the rate of a third order reaction respectively. The corresponding overall apparent activation energy Ea and the frequency factor A change from 140.2 kJ·mol−1 to 247.8 kJ·mol−1 and 3.0 × 105 min−1 to 1.0 × 1012 min−1 respectively with increasing heating rate from 3 to 5 °C·min−1.
Acknowledgments This project (FMRU2007K10) was supported by the Open Research Fund of Key Laboratory for Ferrous Metallurgy and Resources
J. Zhang et al. / International Journal of Mineral Processing 138 (2015) 20–29
Utilization of Ministry of Education, Wuhan University of Science and Technology. References Barin, I., 1995. Thermochemical Data of Pure Substances. 3rd ed. Wiley-VCH Verlag GmbH, Weinheim. Cao, P., 2012. Research on vanadium slag roasting with calcium salt. Iron Steel Vanadium Titan. 33 (1), 30–34. Chen, H.S., 1995. Study on extracting vanadium pentoxide from roasted vanadium slag with lime. Iron Steel Vanadium Titan. 13 (6), 1–9. Coats, A.W., Redfern, J.P., 1964. Kinetic parameters from thermogravimetric data. Nature 201 (1), 68–69. He, D.S., Feng, Q.M., Zhang, G.F., Qu, L.M., Lu, Y.P., 2007. An environmentally-friendly technology of vanadium extraction from stone coal. Miner. Eng. 20 (12), 1184–1186. Kowalski, M., Spencer, P.J., Neushutz, D., 1995. Slag Atlas. 2nd ed. Verlag Stahleisen GmbH, Düsseldorf. Li, X.S., 2011. Mechanism Research on Oxidation Roasting and Leaching Process of High Calcium Low-Grade Vanadium Slag. Chongqing University, Chongqing. Li, X.S., Xie, B., Wang, G.E., Li, X.J., 2011. Oxidation process of low-grade vanadium slag in presence of Na2CO3. Transit. Nonferrous Met. Soc. China 21 (8), 1860–1867.
29
Li, X.J., Zhang, R.P., Xie, B., 2012. Effects of calcium roasting conditions of vanadiumcontaining slag on leaching rate of vanadium. Chin. J. Process. Eng. 12 (1), 54–58. Liao, S.M., Bo, T.L., 1985. Foreign Vanadium Metallurgy. Metallurgical Industry Press, Beijing. Liu, B., Du, H., Wang, S.N., Zhang, Y., Zheng, S.L., Li, L.J., Chen, D.H., 2013. A novel method to extract vanadium and chromium from vanadium slag using molten NaOH–NaNO3 binary system. AICHE J. 59 (2), 541–552. Moskalyk, R.R., Alfantazi, A.M., 2003. Processing of vanadium: a review. Miner. Eng. 16 (9), 793–805. Polyak, D.E., 2011. U.S. geological survey 2009 minerals yearbook—vanadium. U.S.: Geological Survey, Virginia: 80.1–80.10. Van Vuuren, C.P.J., Stander, P.P., 1995. The oxidation kinetics of FeV2O4 in the range 200– 580 °C. Thermochim. Acta 254, 227–233. Van Vuuren, C.P.J., Stander, P.P., 2001. The oxidation of FeV2O4 by oxygen in a sodium carbonate mixture. Miner. Eng. 14 (7), 803–808. Voglauer, B., Grausam, A., J rgl, H.P., 2004. Reaction-kinetics of the vanadium roast process using steel slag as a secondary raw material. Miner. Eng. 17 (2), 317–321. Yin, D.F., Peng, Y., Sun, Z.H., He, W.Y., 2012. Influencing factors of calcified roasting and thermal analysis for the process of vanadium slag produced from Pangang. Met. Mine 4, 91–94.