Kinetic study of ferrocolumbite dissolution in hydrofluoric acid medium

Hydrometallurgy 85 (2007) 87 – 94 www.elsevier.com/locate/hydromet

Kinetic study of ferrocolumbite dissolution in hydrofluoric acid medium M. Rodriguez a , O. Quiroga b , M. del C. Ruiz a,⁎ a

Instituto de Investigaciones en Tecnología Química (INTEQUI) Universidad Nacional de San Luis-CONICET, Argentina b Instituto de Investigaciones para la Industria Química (INIQUI) Universidad Nacional de Salta, Argentina Received 3 January 2006; received in revised form 5 July 2006; accepted 9 July 2006 Available online 2 October 2006

Abstract The recovery of Nb, Ta, Fe and Mn from ferrocolumbite with hydrofluoric acid solutions has been investigated. The experimental tests were carried out in a pressure reactor. The influence of several parameters has been studied in order to deduce the kinetics of the dissolution of ferrocolumbite. The particle size of the mineral was − 45 μm and the solid–liquid ratio was 1.82% w/v; the temperature range was between 348 to 493 K. The reagents and products were characterised by means of X-ray fluorescence (XFR), X-ray diffraction (XRD), inductively coupled plasma optical emission spectrometry (ICP-OES), scanning electron microscopy (SEM), energy-dispersive X-ray spectrometry (EDS), and BET specific surface area measurement. The leaching rate of ferrocolumbite increases with the concentration of hydrofluoric acid and the reaction temperature but is not affected very much by stirring speed. The diffractograms of the residues did not show any structural change with respect to the mineral. SEM and EDS analysis of some residues in which the conversion was approximately 50%, indicated a selective attack on certain particles of the mineral and on certain zones of a particle. This attack was an irregular located-type. The treatment of the experimental data was carried out using the MODELADO software. The model that best represents the dissolution of ferrocolumbite and each one of the oxides in the mineral is:
 b2 t lnð1−X Þ ¼ −b1 lnð1 þ b2 tÞ − 1 þ b2 t This is based on the “nucleation and growth of nuclei” theory and physically represents the attack observed by SEM and EDS on the residues. © 2006 Elsevier B.V. All rights reserved. Keywords: Reaction kinetics; Leaching; Hydrometallurgy; Extractive metallurgy; Industrial minerals; Oxides ores

1. Introduction Niobium and tantalum occur together in nature, mainly in minerals such as pyrochlore, microlite and the columbo-tantalite series (Habashi, 1997; Gupta and Suri, ⁎ Corresponding author. Tel./fax: +54 2652 426711. E-mail address: [email protected] (M.C. Ruiz). 0304-386X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.hydromet.2006.07.005

1994). Both elements have applications in the electronic, chemical, nuclear, and aerospace industries, among others. (Habashi, 1997; Gupta and Suri, 1994; Tolley, 1992). Moreover, tantalum is used in medicine due to its lack of toxicity and great compatibility with tissues (Habashi, 1997). The extraction of Nb and Ta from their minerals or the materials that contain them is carried out by

88

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

pyrometallurgical methods and hydrometallurgical methods. The production of Nb2O5 of technical grade from pyrochlore has been investigated by Habashi and Toromanoff (1983) using an autoclave. Krasilshckik (1991) have used leaching under pressure to dissolve the low-solubility materials, such as the niobium and tantalum oxides. A number of studies have been performed with the aim of getting information about the kinetics of leaching Nb and Ta-containing materials. Majima et al. (1988) determined that values of the apparent activation energies for the leaching of Nb and Ta from columbite and tantalite, in HF–HCl mixture, are betwen 42.7 and 65.5 kJ/mol, and suggested that the temperature increase is effective in enhancing the dissolution reaction, which indicates that the reaction rate is controlled by the chemical reaction. In previous kinetic studies on the dissolution of tantalite in HF–H2SO4 medium, at temperatures between 313 to 353 K, Baram (1965) reported a value of activation energy of 7.2 kJ/mol. Later, this author investigated the dissolution kinetics of the Nb and Ta pentoxides in HF medium, in a temperature range between 298 and 348 K, and found activation energies in the order of 49.79 kJ/mol and 23.85 kJ/mol for Nb2O5 and Ta2O5 dissolution, respectively (Baram, 1972). On the other hand, Ruiz et al. (1999) have studied the variables that affect the extraction of Nb and Ta from a columbo-tantalite material in a pressure reactor, using HF as leaching agent, and observed that the leaching was affected by the solid–liquid ratio, temperature, reaction time and concentration of the fluid phase. Similar results were found by El-Hussaini and Mahdy (2002) for a H2SO4–HNO3 medium. Hoberg and Götte (1985) and Welham (2001) investigated the effect of mechanical activation on the leaching of columbite, finding that it substantially affects the rate and the degree of the dissolution in HF and NaF–HCl media. In this work the pressure dissolution of a ferrocolumbite of San Luis, Argentina, in HF medium has been studied. The aim is to find the kinetic model that best represents the experimental data. 2. Experimental 2.1. Materials The mineral was obtained from the Las Cuevas mine, in San Luis, Argentina. The mineral was ground in a disc grinder and separated to − 45 μm particle size. The specific surface is 1.2 m2/g, determined by BET me-

thod, by N2 adsorption at 77 K on a Micromeritic Acusorb 2100 E. The ferrocolumbite global composition was determined by XRF and ICP-OES, and the results are given in Table 1. Characterization performed by X-ray fluorescence (XRF) with a Philips PW 1400 equipment (Ruiz et al., 1993) indicated that the mineral stoichiometry is: ðMn0:46 Fe0:54 ÞðNb0:65 Ta0:35 Þ2 O6

ð1Þ

The mineral diffractogram obtained with a Rigaku DMax III C equipment (operated at 35 kV and 30 mA, Cu Kα radiaton, Ni filter, λ = 0.1541 Å) is shown in Fig. 1, which indicates the presence of ferrocolumbite, feldspar and quartz (JCPDS, 1993, card N° 33659). Observation of the solid by scanning electron microscopy (SEM) was performed with LEO 1450VP and Philips 515 equipments, equipped with both energy dispersive X-ray spectrometers (EDS). As observed in Fig. 2a and b, the particles have irregular shape with flat sides. All of the other chemical reagents used were of analytical quality. 2.2. Leaching equipment and procedure The leaching tests were performed in a 450-ml Parr reactor, built in Monel. The leaching agent used was a 9% HF (v/v) aqueous solution. For each test 5 g of the mineral were placed into the reactor and 275 mL of the leaching solution was added. N2 was bubbled to remove the air and thus minimize the corroding effects of O2 in the hot HF medium. The mixture was subsequently heated with constant stirring speed at 330 rpm and a heating programme of either 5 or 10 K/min, according to the work temperature (thus all the preheated periods were similar for all tests). Once the experiment was finished, the reactor was cooled down for 25 min, without stirring, and the solid was Table 1 Global composition of the mineral Compound/ element

% in weight

Compound/ element

% in weight

Nb2O5 Ta2O5 TiO2 MnO FeO SiO2 Al2O3 CaO

41.20 36.80 1.16 6.83 9.58 1.48 1.13 0.38

K2O U Zn Sn V Pb Ce La

0.143 0.220 0.060 0.020 b0.020 0.015 0.005 0.003

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

89

formed according to the methodology developed by Ruiz et al. (2004). In each experiment, 100 mg of the residue was used for the preparation of the pellets for the analyses, and the mass of each oxide in the residues was obtained. With these data and the initial mass values of each oxide present in the mineral, the quantity of dissolved ferrocolumbite was calculated, in terms of global conversion, X, defined as: X ¼

mi − mf mi

ð2Þ

where mi and mf are the addition of the masses of the Nb, Ta, Fe and Mn oxides, initial and final, respectively. Fig. 1. XRD diffractogram of the mineral.

filtered. The residue obtained was dried, calcined at 1173 K for 4 h, cooled down and weighed. The quantitative analytical determinations for each studied oxide (Ta2O5, Nb2O5, FeO and MnO) was per-

3. Results and discussion The dissolution of ferrocolumbite with HF can be represented by the following reaction: ðMn0:46 Fe0:54 ÞðNb0:65 Ta0:35 Þ2 O6 þ 16HF →1:3H2 NbF7 þ 0:7H2 TaF7 þ 0:46MnF2 þ 0:54FeF2 þ 6H2 O

ð3Þ

3.1. Effect of temperature and the reaction time The effect of the temperature and the reaction time on the ferrocolumbite dissolution was studied from 348 to 493 K and for 16 to 220 min, respectively. The results show that the rate of mineral dissolution increases with temperature and the time of reaction, reaching a mineral

Table 2 Effect of reaction time and temperature on the dissolution of ferrocolumbite Temperature (K)

Time (min)

Global conversion (X )

348

16 96 132 196 28 73 108 178 208 40 85 110 120 130 190 220

0.12 0.42 0.48 0.55 0.52 0.70 0.75 0.79 0.83 0.76 0.81 0.83 0.83 0.84 0.84 0.84

396

493

Fig. 2. SEM micrographs of ferrocolumbite. a) General view, b) morphological and surface details of the particles.

90

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

Table 3 Effect of the stirring speed on the dissolution of the mineral

3.3. Characterization of the leaching residues

Stirring speed (rpm)

Global conversion (X )

110 220 330 440 550

0.81 0.83 0.83 0.83 0.84

The BET method was used to determine the specific surface area of some residues with different degrees of conversion. No important changes were observed on the specific surface area of the residues. The quantitative determination of Nb, Ta, Fe and Mn in the residues was performed by XRF, as mentioned earlier. The results of these analyses showed that the Nb, Ta, Fe and Mn quantities in the residues correspond to the stoichiometry of the mineral, which indicates that the composition of the particles in the residue remains approximately the same as that of the mineral. The diffractograms of the leaching residues, in different working conditions, do not show any structural change in the residue remained, which is in agreement with what was observed by XRF. This is shown in Fig. 3. In order to understand the type of chemical attack, residues from tests where the conversion was approximately 50% were analysed by SEM and EDS. From the results shown in Fig. 4a, b and c and Table 4, it can be observed that there was a selective attack on certain

dissolution of 81% in 85 min. The results are shown in Table 2. 3.2. Effect of the stirring speed The effect of the stirring speed on the dissolution rate of the oxides present in the mineral was studied at 493 K. The results are shown in Table 3, which indicate that the stirring speed does not have a remarkable effect on the dissolution rate of the mineral in the working conditions. This allows us to infer, according to Habashi (1980), that the rate of mass transfer in the solid–liquid interface is not involved in the global reaction rate of the process.

Fig. 3. XRD diffractogram of the leach residue.

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

91

3.4. Kinetic model The experimental data of the ferrocolumbite dissolution in HF medium were treated with the MODELADO software developed by Quiroga (2002). The most probable model obtained was:
lnð1−X Þ ¼ −b1 lnð1 þ b2 tÞ −

b2 t 1 þ b2 t

 ð4Þ

where b1 and b2 coefficients are defined as: b1 ¼

rG NS0 X0 bMB rS b2 qdp0

ð5Þ

b2 ¼ kN2 NS0 CA2

ð6Þ

Eq. (4) is based on the “nucleation and growth of nuclei” theory (Avrami, 1939), which was originally developed for the modelling of transformation reactions and solid decomposition. Delmon (1969) adapted this theory to fluid–solid reactions, changing the concept of “nucleation” and “growth of nuclei” for “activation in active sites” and “growth of holes”, respectively. When the activation occurs in a sequential mode, the conversion rate of the particle is expressed as: dX X0 mG rN ¼ ð1−X Þ dt V0

ð7Þ

where: Ω0 is the initial particle surface and V 0 its initial volume; rN is the activation rate of the active sites per surface unit; vG is the volume of the hole, which, for a time t is calculated as:

mG ¼ rG V Fig. 4. SEM microphotographs of residues from leaching with HF. a) General view, b) and c) selectively attacked particles.

particles of the mineral. Moreover, Fig. 4b and c show that the attack was selective on certain zones of the same particle. EDS analysis indicated that the most attacked particles have a higher concentration of U. This can be explained taking into account the metamict characteristics of this type of minerals, since the presence of U in certain zones of a particle makes them more vulnerable to the attack.

0

b M B rS t qdp0

!p ð8Þ

where: σG is a coefficient of shape of the nuclei; b is a stoichiometric coefficient; MB and ρ are the molecular weight and the solid density, respectively; dP0 the initial Table 4 Elemental compositions of partially attacked particles, in atom % Zone

% Nb

% Ta

% Fe

% Mn

%U

1 2 3 4

47.18 48.30 40.21 43.72

23.30 18.7 31.05 27.60

18.50 9.65 8.71 9.23

12.40 13.25 14.01 15.60

0.26 1.72 0.10 1.50

92

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

particle diameter; rS the rate of fluid-solid reaction; p is the growth factor, whose value indicates the direction of the nuclei growth. In the specific case of Eq. (4), p = 1 (in one direction) and the activation rate, rN, is defined as: 2

rN ¼

kN2 NS0 NE0 2

m m

ð1 þ kN2 NS0 NE0 tÞ2

ð9Þ

with: NS0 representing initial number of the sites that can be activated per unit of initial surface of the particle; kN2 the kinetic constant of activation of the sites that can be activated; m the reaction order with respect to the fluid reagent; and NEo the number of moles per unit surface of a chemical species E that participates in the process, and whose identity is obtained from the analysis of different kinetic mechanisms for the formation of reaction interfaces. Eq. (9) corresponds to the particular case of sequential activation of spontaneous sites, NS, due to a process of adsorption–reaction–desorption, and its change rate is given by the following equation:

S2* = O* with superscript (*) indicating an activated species. According to this mechanism, Eq. (10) corresponds to the particular case of an activation rate of second order. From Eqs. (11) and (12) it is established that the initial rate of the process is controlled by the elimination rate of Me2Ox (x = 2 or 5, depending on whether the oxides are Fe and Mn or Nb and Ta, respectively). In agreement with this mechanism, Eq. (10) is written as: dNS ¼ kN2 c2A ðNS0 −NS Þ2 dt

where: NS0 represents the number of the initial moles of Me2Ox per unit of initial surface of the particle; and NS, the number of moles of Me2Ox that have been eliminated since the solid and fluid reagents were in contact; and cA the concentration of HF evaluated on the interface of reaction. The integration of Eq. (13) set into = 0 and t = 0, gives: 2

NS ¼ m dNS ¼ kN2 NE0 ðNS0 −NS Þ2 dt

ð10Þ

1) Slow Stage H H Sb S * þ H2 OðaqÞ F H 2HFðaqÞ þ2SY Y Y *1 H F S2 þ F2 MeðaqÞ Sb Sb F F

ð14Þ

2

dNS kN2 c2A NS0 ¼ dt ð1 þ kN2 c2A NS0 tÞ2

ð11Þ

ð12Þ

where: S = Me2Ox (x = 2 or 5), is a metallic oxide present in the mineral; S1* = Me* (if x = 2) and MeO2* (if x = 5);

ð15Þ

Combining Eqs. (7) and (8), for p = 1, and Eq. (15) for rN, and integrating the resulting equation, Eq. (4) is again obtained, with the b1 and b2 coefficients defined by Eqs. (5) and (6). In Eqs. (5) and (6) rS, k and kN are defined as: rS ¼ kcnA

2) Fast Stage F F S1* b S1* b S1* H F Y YH2 OðaqÞ 2HFðaqÞ þ * Y F H * * S2 S2 b S2 b H H þF2 MeðaqÞ

kN2 c2A NS0 t 1 þ kN2 c2A NS0 t

from which it results:

where NS is the number of the active sites per surface unit. To identify the chemical species E of Eq. (9), it is assumed that the global rate of the dissolution process of the mineral is controlled by the stages of adsorption of the fluid reagent and chemical reaction, according to the following mechanism that takes places on the surface of the solid reagent:

Sb

ð13Þ

k ¼ A1 e−RT Ea

EN

kN2 ¼ A2 e− RT

ð16Þ

where k and kN, the kinetic coefficients of the reaction rate and the formation rate of the sites, respectively; A, the frequency factor; Ea and EN, the apparent activation energy and energy of the nucleation, respectively; R, the gas constant and T, the temperature. The results of the correlation of the experimental data, using Eq. (4), for different levels of the factors, are shown in Fig. 5. Eq. (4) represents a model where the solid reagent experiences a topochemical type attack. The results obtained by SEM and EDS analyses, performed on the residues of the ferrocolumbite treated in HF medium,

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

93

Fig. 5. Correlation between the model predictions and experimental data of the ferrocolumbite dissolution in HF medium.

allow us to confirm the validity of the model. These studies showed that the dissolution of the mineral is produced by an irregularly-located type attack, preferably in the U-containing zones. The causes of the kinetic curves flattening in Fig. 5 are likely due to the formation of insoluble layers of fluorides or fluorosilicates of calcium, iron, aluminum or uranium. However, whatever the reason, the “nucleation and growth of nuclei” theory adapted by Delmon justifies the formation of the holes visible in Fig. 4 at the level of the reaction interface, or the formation of layers of fluorides or fluorosilicates of calcium, iron, aluminum or uranium as if they were ash layers. The value of the kinetic coefficient of the reaction rate was calculated using Eqs. (5) and (6). The value of the activation energy of the dissolution reaction was obtained from the Arrhenius plot, and the result is shown in Fig. 6. The slope of Fig. 6 allows us to estimate the apparent activation energy for the ferrocolumbite dissolution, being

Ea = 45.7 kJ/mol. From the Ea value it can be suggested that the temperature increase affects the dissolution reaction, which indicates that after the reaction interface has been formed, the controlling stage of the mechanism is the chemical reaction, in agreeent with what has been postulated by other researchers (Habashi, 1980; Majima et al., 1988; Quiroga et al., 1999). 4. Conclusions The maximum dissolution of the mineral, 83%, is reached in 90 min when using 9% HF acid v/v, at 493 K. The rate of the ferrocolumbite dissolution increases with the temperature and the reaction time. The stirring speed of the pulp affects slightly the lixiviation rate of the mineral in the working conditions. The solid undergoes an irregular-located type attack; that is, the hydrofluoric acid selectively attacks those particles containing uranium.

Fig. 6. Arrhenius plot for the ferrocolumbite dissolution in HF medium.

94

M. Rodriguez et al. / Hydrometallurgy 85 (2007) 87–94

The model, based on the “nucleation and nuclei growth” concept, that best fits the experimental results of the ferrocolumbite dissolution in HF medium is represented by Eq. (4). The characterization of the residue physically supports the results of the kinetic model. The value of the apparent activation energy suggests that the controlling stage after the reaction interface has been formed, of the mechanism is the chemical reaction. Acknowledgments The financial support from Universidad Nacional de San Luis, Agencia Nacional de Promoción Científica y Tecnológica and Consejo Nacional de Investigaciones Científicas y Técnicas is gratefully acknowledged. Appendix A Symbols b Stoichiometric coefficient b1 and b2 Coefficients defined by Eqs. (4) and (5), respectively cA Concentration of HF evaluated on the interface of reaction, mol/l dp0 Initial particle diameter, mm E Activation energy, kJ/mol k Kinetic coefficient of the reaction rate, m/s. kN2 Kinetic coefficient of the formation of the sites, m2/s. MB Molecular weight of the solid reactant, g/mol. mi and mf Addition of the masses of the Nb, Ta, Fe and Mn oxides, initial and final, respectively, mg NE0 Number of moles per surface unit of a chemical specie E NS0 Initial number of the sites than can be activated per unit of initial surface of the particle NS Number of the active sites per surface unit p Growth factor R Gas constant, kJ/kmol K rN Activation sites rate, mol cm− 2 min− 1 rS Reaction solid–fluid rate, mol cm− 2 min− 1 T Temperature, K t Time, s vG Volume of the hole, m3 0 V Initial particle volume, m3 X Solid conversion Greek Symbols ρ Solid density, kg/m3 0 Ω Initial particle surface, m2 σG Coefficient of shape of the hole

References Avrami, M., 1939. Kinetic of phase change. I. General theory. Journal of Chemical Physics 7, 1103–1112. Baram, I.I., 1965. Cinética de la disolución de los pentóxidos de niobio y tantalio en ácido fluorhídrico (traducido al Español). Zhurnal Prikladnoar 38 (10), 2181–2188. Baram, I.I., 1972. Cinética de la disolución de tantalita en una mezcla de ácidos fluorhídrico y sulfúrico (traducido al Español). Tsvettnye Metally (The Soviet Journal of Non-Ferrous Metals) 15, 97–99. Card of JCPDS, Number 33659, 1993. Powder Difraction Files. Delmon, B., 1969. Introduction a la Cinétique Hétérogène. Editions Technip, Paris. El-Hussaini, O.M., Mahdy, M.A., 2002. Sulfuric acid leaching of Kab Amiri niobium–tantalum bearing minerals, central eastern desert, Egypt. Hydrometallurgy 64, 219–229. Gupta, C.K., Suri, A.K., 1994. Extractive Metallurgy of Niobium. CRC Press Inc, Florida. Habashi, F., 1980. Principles of Extractive Metallurgy, vol. I. Gordon and Breach, New York. Habashi, F., 1997. Handbook of Extractive Metallurgy, vol. III. Wiley VCH, Germany. Habashi, F., Toromanoff, I., 1983. Hydrometallurgical production of technical niobium oxide from pyrochlore concentrates. Journal of Less-Common Metals 91, 371–382. Hoberg, H., Götte, J., 1985. The influence of mechanical activation on the kinetics of the leaching process of columbite. International Journal Mineral Processing 15, 57–64. Krasilshckik, V.Z., 1991. Autoclave vapor-phase descomposition of some difficulty soluble substances. Zhurnal Analitical Khimii 41 (4), 586–590. Majima, H., Awakura, Y., Mishima, M., Hirato, T., 1988. Dissolution of columbite–tantalite in acid flouride media. Metallurgical Transactions. B, Process Metallurgy 19 B, 355–363. Quiroga, O.D., Avanza, J.R., Fusco, A.J., 1999. Modelado Cinético de las Transformaciones Fluido–Sólido Reactivo. Editorial Universitaria de la Universidad del Nordeste (EUDENE). Corrientes, Argentina. Quiroga, O.D., 2002. Modelado. Software para el Tratamiento Cinético de Transformaciones Fluido Sólido–Reactivo. INIQUI (UNSa-CONICET). Salta, Argentina. Ruiz, M. del C., González, J., Olsina, R., 1993. Analysis of niobium, tantalum and titanium extracted from tantalite and columbite chlorination. Journal of Chemical Technology and Biotechnology 57 (4), 375–378. Ruiz, M. del C., Rodriguez, M., González, J., Rivarola, J., 1999. Pressure leaching of niobium and tantalum from columbo-tantalite. In: Mishra, B. (Ed.), Extraction and Processing Division Congress. TMS, California, pp. 25–533. Ruiz, M. del C., Rodriguez, M., Perino, E., Olsina, R., 2004. X-ray fluorescence analytical methodology for the determination of Nb, Ta, Fe and Mn extracted in hydrometallurgic processes. Latin American Applied Research 34, 23–27. Tolley, R.J., 1992. Tantalum–Niobium International Study Center, vol. 69, p. 3. Welham, N.J., 2001. Efect of extend grinding on the dissolution of a Ta/Nb concentrate. Canadian Metallurgical Quaterly 40 (9), 143–154.