Otimização do processo de obtenção de RE ₂ O ₃ a partir da xenotima usando planejamento estatístico

ce Cerâmica Cerâmica 0366-6913 1678-4553 Associação Brasileira de Cerâmica Resumo O estudo da produção da mistura de óxido de terra rara (RE2O3) de xenotima via hidrometalurgia foi realizado para avaliar o significado dos efeitos de três fatores: temperatura de fusão (A), proporção de NaOH/xenotima (B) e tempo (C) para a primeira fase de estudo de fusão alcalina usando um planejamento fatorial completo 2³. Além disso, foram estudados sete fatores, incluindo a proporção líquido/sólido (LS), inerte (I1), excesso de HNO3 (EA), temperatura (TL), inerte (I2), tempo (HL) e inerte (I3) para a segunda e terceira etapas, que foram lixiviação do ácido oxálico e precipitação, usando o planejamento experimental Plackett-Burman (PBD). Fatores de otimização para os efeitos de significância foram realizados utilizando parcelas de interações, análise de variância (ANOVA), gráficos de Pareto, gráficos de superfície, análise do teste t de Student e o teste F. Um modelo de regressão foi sugerido e satisfatoriamente ajustado para os dados experimentais do processo de fusão alcalina, revelando um coeficiente elevado de determinação (R2=0,92) a um valor F calculado bem acima do valor F tabulado, a um nível de confiança de 95%. O RE2O3 foi caracterizado por análise química, difração de raios X, microscopia eletrônica de varredura e área de superfície específica. Este processo resultou em um material com propriedades físicas e químicas idênticas aos compostos isolados de terras raras e pode ser usado como um aditivo de sinterização alternativo em cerâmicas avançadas e indicou que o custo de produção final de RE2O3 pode ser menor em comparação com Y2O3 comercial. INTRODUCTION Rare earth (RE) has its use in glass, metal, nuclear technology, ceramics, and are widely used as high-performance catalysts, as well as for high efficient production of permanent magnets and phosphors. The electronic structure of the elements of RE confer special properties such as magnetism, luminescence and laser applications, among others1), (2), (3. The rare earth elements are ranked, usually into two subgroups: a light fraction or subset of cerium (ceric lands), comprising the elements of atomic numbers 57-63 (La to Eu) and a heavy fraction or subgroup yttrium (yttric lands) which contain the elements of atomic numbers 64 to 71 (Gd to Lu), besides yttrium itself. Despite its low atomic weight, yttrium is categorized with the heavy rare earths because of its mode of occurrence, ionic radius and other properties which are closer to the elements of the heavy fraction4), (5. The main sources of rare earths are the classic ores/minerals such as monazite, xenotime and bastnasite. Xenotime, a phosphate, and some RE-carrying clays are the main sources of supply of rare earth elements from the heavy fraction6), (7. Xenotime is a phosphate mineral with rare earth crystal structure (space group I41/amd, Z=4), which typically consists of 25% Y2O3 and other heavy elements, but may have as much as 60% Y2O3 and 40% other rare earth oxides8), (9. Brazil has identified RE resources of the order of 30 million tons, with the largest reserves and the highest concentration of rare earths being found in lateritic and silexitos soils associated with carbonatite complexes. Here, the term identified resources is used with the same meaning assigned to identified resources by the U.S. Geological Survey: resources whose characteristics, location, content and quantity are known or estimated from specific geological evidence, including economic, marginally economic and subeconomic components3. The literature reports that with phosphate matrix, the minerals xenotime and monazite have rare earth extracted by acid or basic lixiviation/leaching10), (11. Experimental studies using alkaline leaching and alkaline fusion to recover rare earth oxide indicated a yield of about 95% of rare earths extracted from xenotime10. Using an acid route, monazite and xenotime are digested in sulfuric acid and finally leached in hydrochloric acid producing a mixture of rare earth elements2. The effects of various parameters involving the hydrometallurgical process has the influence of many factors that can be studied quantitatively, where the importance of each variable must be determined4), (12), (13), (14. In this study, the optimization of the production process of the mixed rare earth oxide (RE2O3) from xenotime via hydrometallurgy was divided into three stages; the first was the process of alkaline fusion based on full 2³ factorial design. The second step was the process of leaching, and the third stage was the process of oxalic precipitation, where both phases were based on Plackett-Burman design (PBD). The full 2³ factorial design is an important and simple statistical tool. The observation of the effects of the variables of the melting temperature (A), the ratio of NaOH/xenotime (B), time (C), and their interactions is of utmost importance to understand the process of acid leaching15. The PBD allows estimating all k=N-1 main effects, where N represents the number of experiments with minimum variance, proving to be efficient in the evaluation of a large number of variables and identifying the most significant16), (17. In PBD design, seven factors were evaluated: the liquid/solid ratio (LS), inert (I1), excess HNO3 (EA), temperature (TL), inert (I2), time (HL), and inert (I3) for the second and third steps, which were the oxalic acid leaching and precipitation, respectively. The advantage of this method is to reduce the number of experiments that investigate interactions between factors chosen from the RSM yields of reactions for each experiment18. In both plans response surface methodology (RSM) was used, an optimization technique that has been used with great success in the modeling of various industrial products15), (18. MATERIALS AND METHODS For the optimization, the methodology for the production of RE2O3 was performed in three steps: the first step of the alkaline fusion process, based on the full 2³ factorial design; the second and third stages of oxalic acid leaching and precipitation, respectively, were performed using the Plackett-Burman design (PBD). Alkaline fusion: xenotime from Pitinga Mine, located in Presidente Figueiredo (AM-Brazil), and 98% NaOH in flakes (Sinc) were stoichiometrically weighed according to the experimental variables and levels used in the construction of the full 23 factorial design, which observed the effects of varying the fusion temperature (A), the ratio of NaOH/xenotime (B), and time (C). The materials were mixed in a double cone mixer for 1 h and the mixtures obtained were heated in an electric furnace for times and temperatures pre-determined in the full 2³ factorial design, with the objective of transforming xenotime, which is an insoluble rare earth phosphate - insoluble in nitric acid. The product of the alkaline fusion was fragmented and crushed in a ball mill. Then, an aqueous leaching was performed to separate the insoluble product Na3(RO3) from sodium phosphate (Na3PO4), soluble in water. This step was performed in a vacuum filtration system. The product was dried at 100 °C and sieved with/until a particle size of 40 mesh. Acid leaching: the product from the previous step, consisting of a salt of rare earth sodium was reacted in a glass reactor by reflux with nitric acid (HNO3 66 Be). The system showed a mixture consisting of a liquid phase (liquor) containing dissolved rare earth and a solid phase containing insoluble impurities. In the acid leaching step, the experimental design used was the statistical Plackett-Burman approach (PBD) that investigated the interactions with the seven factors of liquid/solid ratio (LS), inert (I1), excess HNO3 (EA), temperature (TL), inert (I2), time (HL), and inert (I3). Oxalic precipitation: in this third step, we used the liquor/solution diluted with water to 5 g/L at pH=1, obtained by acid leaching conditions previously laid down by the same Plackett-Burman design (PBD) used in the second stage. A solution of oxalic acid (C=80 g/L) was added to the liquor in constant agitation at 200 rpm at room temperature to avoid loss of rare earth. Then, the oxalate formed was separated from the reaction medium by sedimentation followed by filtration. After this step, the product was subjected to washing with water to remove residual soluble salts. The rare earth oxalate was calcined at 800 °C for 4 h. Experimental schedules: the full 2³ factorial design was used to evaluate the effects of three factors (k) of the alkaline fusion step. The total number of experiments required for this experimental design method is given by 2k=23=8. The observation of the effects of varying the fusion temperature (A), the ratio of NaOH/xenotime (B), processing time (C), and their respective values are shown in Table I. The purpose of using the Plackett-Burman design (PBD) is to evaluate the effects of selected factors and identify the most influential in the process of the second and third steps, which is the oxalic acid leaching and precipitation, respectively. PBD investigated interactions with seven factors which are liquid/solid ratio (LS), inert (I1), excess HNO3 (EA), temperature (TL), inert (I2), length (HL) and inert (I3), presented in Table II which shows variables (real and inert) and their respective higher (+) and lower (-) levels studied. Table I - Factors and levels analyzed during the stage of alkaline fusion. Tabela I - Fatores e níveis analisados durante o estágio de fusão alcalina. Factor Notation Level (-) Level (+) Fusion temperature A 600 °C 700 °C Ratio NaOH:xenotime B 1.0 w/w 1.3 w/w Time C 180 min 240 min Table II - Variables in the leaching acid and their respective levels. Tabela II - Variáveis da lixiviação ácida e seus respectivos níveis. Variable Notation Level (+) Level (-) Liquid/solid ratio LS 10/1 kg 5/1 Inert I1 - - Excess HNO3 EA 20% 10% Temperature TL 100 °C 90 °C Inert I2 - - Time HL 20 min 10 min Inert I3 - - Response methodology (RSM): for this study the statistical schedules were performed using software (Statistic Ver 7.0). The response values used included the yield of the reactions in the process steps, and the reactions were performed according to the values shown in Tables I and II. Upon statistical analysis, it was possible to obtain a mathematical model that represents the process for the studied levels. For a 2³ factorial design, the mathematical equation that describes this system can be represented by the linear polynomial equation19; this model (Eq. A) is a simpler model with fewer factors for a well-established, systematic errors are absent, and normalized residuals resulted from experimental errors which exhibit a normal distribution according to a widely accepted statistical convention: y ^ X 1 , X 2 , X 3 = b 0 + b 1 X 1 + b 2 X 2 + b 12 X 1 X 2 + b 13 X 1 X 3 + b 23 X 1 X 2 + b 123 X 1 X 2 X 3 + ε X 1 , X 2 , X 3 (A) Characterization: the chemical composition of RE2O3 was determined by optical emission spectrometry with inductively coupled plasma (ICP-OES Varian Vista MPX). The crystal structure was characterized by X-ray diffraction (XRD) using Rich Seiferst equipment (Debeyeflex 1001) with CuKα radiation, operating at 25 mA/30 kV. The specific surface area was determined by N2 adsorption (BET method), Quantachrome, Nova 1000 equipment. Morphological characterization was performed by LEO scanning electron microscope (SEM, 1450-VP). RESULTS AND DISCUSSION For the evaluation of the effects studied in the first stage of alkaline fusion, Table III presents a matrix of 23 factorial design with the factors fusion temperature (A), ratio of NaOH/xenotime (B), time (C), and responses obtained in the yield of the reactions. The yield of alkaline fusion is the transformation of xenotime to a rare earth compound Na3RO3 insoluble in water and Na3PO4, according to: R P O 4 S + 6 N a O H S → N a 3 R O 3 S + N a 3 P O 4 S + 3 H 2 O V (B) Table III - Response values, R1 and R2, obtained in alkaline fusion, using full 2³ factorial design, conducted in replicated. Tabela III - Valores de resposta, R1 e R2, obtidos na fusão alcalina, usando planejamento fatorial completo 23. Run number A B C R1 (%) R2 (%) 1 - - - 67.87 67.27 2 + - - 69.73 74.47 3 - + - 69.80 70.20 4 + + - 88.93 89.80 5 - - + 67.47 71.13 6 + - + 77.13 80.20 7 - + + 83.73 84.53 8 + + + 92.93 96.20 The results obtained in planning indicate an increase in the average yield of 67.27% to 96.20%, when all variables are simultaneously level -1 to +1. Therefore, within this range of operation, the alkaline fusion gave a better yield, or an efficient transformation of xenotime in Na3RE3 and Na3PO4. Table IV presents the estimated effects, standard errors, and Student t-test for yield response of alkaline fusion, according to the 23 factorial design. Through statistical analysis, it was found that the main effects of factors A, B and C, and the effect of interaction between A and B were statistically significant at a 95% confidence level. The positive effect of the interaction A*B means that the yield of fusion is favored with a concomitant increase of temperature and relative NaOH:xenotime ratio, i.e. in this case a synergistic effect of interaction occurs between the variables B and A. After examination of the significant regression on the yield of the reaction factors, a linear model was fitted to experimental data: Y i e l d % = 78 . 21 + 5 . 46 A + 6 . 30 B + 3 . 45 C + 1 . 99 A B (C) Table IV - Estimates of the effects, standard errors and Student t-test for yield response of alkaline fusion, using the full factorial design 23. Tabela IV - Erros padrão e teste t de Student para a resposta do rendimento da fusão alcalina, utilizando o planejamento fatorial completo 23. Factor Estimate (%) Standard error tcalc Average 78.21* ± 0.73 107.43 Temperature (A) 10.92* ± 1.46 7.50 NaOH:xenotime (B) 12.61* ± 1.46 8.66 Time (C) 6.91* ± 1.46 4.74 AB 3.98* ± 1.46 2.73 AC -1.02 ± 1.46 0.70 BC 2.76 ± 1.46 1.89 * - statistically significant values at 95% confidence. where A, B and C are the coded values for temperature, NaOH:xenotime relation, and fusion time, respectively. The statistical significance of the empirical model of the first order was evaluated by F test (ANOVA), which revealed a high coefficient of determination (R2=0.92) and also a Fcalculated value well above the Ftabulated, to a 95% confidence level (Table V). For the regression to be statistically significant, the value of Fcalculated, or the MQR/MQr ratio should be at least 4 times the value of Ftabuleted16), (21. In this case, Fcalculated is approximately 10 times greater than the Ftabulated, which shows that the model described by Eq. C is highly significant. Fig. 1 shows a graph of the values calculated by the model on the basis of experimental values obtained. It can be seen that the predicted values are well adjusted to the experimental data, confirming again that the model is highly significant. Figure 1: Graph of the values calculated by the model on the basis of experimental values obtained. Figura 1: Gráfico dos valores calculados pelo modelo com base nos valores obtidos experimentalmente. Table V - Analysis of variance for the adjustment of the first order to yield data on the alkaline fusion step model. Tabela V - Análise de variância para o ajuste a fim de se obter dados sobre o modelo de fusão alcalina. Source variation SQ Gl MQ Fcalculated Ftabulated Regression 1367.01 4 341.75 33.87 3.36* Waste 110.95 11 10.09 Total 1477.96 15 Notes: F set values at 95% confidence; % explained variation (R 2 ): SQ R /SQ T =0.92. The RSMs determined by the linear model equation, which describes the values of yield of alkaline fusion depending on the manipulated variables in the process are shown in Fig. 2, where we have each variable in their relationships with their statistical significance for this study. The maximum yield predicted by the model is obtained at higher levels of temperature and the relation NaOH:xenotime (Fig. 2a). All main effects of the independent variables are positive, presenting positive correlation with the variable yield response so that as the manipulated variables increased, the yield of the reaction also increased (Figs. 2b and 2c). Figure 2: Response surfaces and contour plots showing the effects of the independent variables on positive correlation with the variable yield response: (a) NaOH:xenotime and temperature; (b) time and NaOH:xenotime; and (c) temperature and time. Figura 2: Superfícies de resposta e efeitos mostrados através das variáveis independentes com relação positiva à variável de resposta: (a) NaOH:xenotima e temperatura; (b) tempo e NaOH:xenotima; e (c) temperatura e tempo. The alkaline material from the fusion process at optimized conditions (all variables in the upper level), after washing with water, was reacted with nitric acid under different experimental conditions, according to: N a R O 3 S + 6 N a O H 3 a q → 3 N a N O 3 a q + R N O 3 a q + 3 H 2 O 1 + r e s í d u o s (D) For the acid leaching stage, the PBD results are expressed in terms of output, i.e. the maximum conversion of yttrium and rare earth salts in solution, given in Table VI. This is Plackett-Burman design for the study of 4 variables with 8 trials during the stage of acid leaching, in case I1 and I2. Fig. 3 shows the significance of the effects after statistical analysis of the design data by PBD. Through analysis of the Pareto chart, Fig. 3a showed that, among the estimated impact of the 7 factors assessed, the excess HNO3 (EA) had the greatest importance in the process, followed by temperature (TL). Fig. 3b confirms through desirability analysis that the excess HNO3 (EA) and temperature (TL) were the two factors that had significant positive correlations with the yield of the acid leaching step. Figure 3: Estimated effects of studied variables (a), and results of desirability analysis (b) showing that the excess HNO3 (EA) and temperature (TL) had significant positive effects with the yield of the acid leaching step. Figura 3: Efeitos estimados das variáveis estudadas (a) e resultados da análise de conveniência (b) mostrando que o excesso de HNO3 (EA) e temperatura (TL) tiveram efeitos positivos e significativos com o rendimento da lixiviação ácida. Table VI - Plackett-Burman design for the study of 4 variables with 8 trials during the stage of acid leaching. Tabela VI - Plackett-Burman para o estudo de 4 variáveis com 8 ensaios durante a fase de lixiviação ácida. Run nº LS I1 EA TL I2 HL I3 Yield (%) 1 + + + - + - - 74.72 2 + + - + - - + 86.45 3 + - + - - + + 82.89 4 - + - - + + + 84.38 5 + - - + + + - 86.56 6 - - + + + - + 83.81 7 - + + + - + - 86.55 8 - - - - - - - 87.24 The statistical significance of each effect can be determined by Student’s t-test. For the design of PBD, the number of dummy variables is the number of degrees of freedom for the entry in the table of t, for a given confidence interval15), (20. In this work, a confidence level of 95% with the tabulated value of t equal to 3.18 was used. Comparing the values of t given in Table VII, it can be verified that at this level all effects showed a tcalc value lower than ttabul. This indicated a lower significance of effects on the yield of the acid leaching. Table VII - Estimates of the effects, Student’s t test and p-value for the factors used in the leaching step (PBD). Tabela VII - Estimativas dos efeitos, teste t de Student e valor de p para os fatores utilizados na etapa de lixiviação (PBD). Variable Notation Estimate (%) tcalc p-value Liquid/solid ratio LS -2.840 -1.21 0.31 Excess HNO3 EA -4.165 1.78 0.17 Temperature TL 3.535 1.51 0.23 Time HL 2.040 0.87 0.45 Lastly, the oxalic precipitation was achieved in solution produced in the leaching step (experiment 8), employing lower levels of the studied factors. The concentration of precipitant solution used for conversion of the oxalates to oxides of rare earths was 80 g/mL, thus obtaining a yield of 98.25%. Oxalic acid has great selectivity, allowing for the attainment of rare earth oxalates which are insoluble in the reaction medium used21), (22), (23. According to the statistical design, we obtained oxalate by calcination from XRD pattern in Fig. 4a for the mixed rare earth oxide, in the precipitation step at 800 °C for 2 h. The decomposition of the oxalate group occurred between 340-410 °C as reported in the literature24. Fig 4b shows the XRD pattern of a physical mixture of the main constituent oxides observed in the chemical analysis performed on RE2O3 by ICP-OES (Table VIII). It can be seen in the mechanical mixture spectrum that the diffraction peaks of the individual oxides are identified. The behavior of the mixed rare earth oxide observed in the XRD pattern suggested that the constituents of the same metals, as oxides, form a solid solution of yttrium matrix due to the similarity of the rare earths (atomic radius, ionic radius, electronegativity and valence). The peaks were slightly displaced with respect to the rare earth oxides with the highest concentration found in xenotime22), (23), (24), (25), (26. Figure 4: X-ray diffraction patterns of: (a) mixed rare earth oxide after calcination at 800 °C for 2 h; and (b) physical mixture of the main constituent oxides. Figura 4: Difratogramas de raios X de: (a) mistura de óxidos de terra rara calcinada a 800 °C por 2 h; e (b) mistura dos pós de óxidos constituintes. Chemical analysis by atomic emission spectroscopy by induced plasma (Table VIII) done in mixed rare earth oxide showed the highest percentages of oxides: Y2O3 (44.1 wt%) and Yb2O3 (19.2 wt%). The great variability of the composition of natural xenotime is verified by literature8), (22), (23), (24), (25), (26), (27), (28, showing that there are different compositions of xenotime with heavy rare earths, like Y, U and Th. Specifically, xenotime grains usually obtained from granite around the world are characterized by compositions comprising 70 to 80 mol% Y (PO4) + 16 to 25 mol% heavy in rare earth phosphates8. Table VIII - Chemical analysis performed on RE2O3 by ICP-OES. Tabela VIII - A análise química realizada em RE2O3 por ICP-OES. Oxide % mass Oxide % mass Y2O3 44.1 Sm2O3 0.45 Yb2O3 19.2 ZrO2 0.16 Er2O3 13.6 Nd2O3 0.15 Dy2O3 10.5 CeO2 0.09 Ho2O3 3.16 CaO 0.06 Tm2O3 2.88 La2O3 0.04 Lu2O3 2.55 Eu2O3 0.02 Gd2O3 1.26 TiO2 0.01 Tb4O7 0.88 The result presented for the analysis of specific surface area via BET was 6.48 g/cm3. This value showed that the powder obtained had characteristics suitable for direct application without the need for grinding, however for synthesis of advanced ceramics, it needs to be mixed and ground with other parts, thereby losing its initial characteristics. The bimodal particle size distribution of RE2O3, with a wide range of distribution, varied from 0.40 to 80 μm. This wide distribution facilitates the compaction of the powder, which could make mixtures of ceramics to be sintered via liquid phase. Fig. 5 shows SEM micrographs in which it was observed that the particles form evenly-rounded clusters. The irregular particles provide more contact points, which facilitates the compaction of the material and thereafter the same in sintering. Figure 5: SEM micrographs of RE2O3 particles showing evenly-rounded clusters. Figura 5: Micrografias obtidas por microscopia eletrônica de varredura das partículas de RE2O3 mostrando aglomerados uniformemente arredondados. CONCLUSIONS The study of optimization of the mixed rare earth oxide (RE2O3) by the process of hydrometallurgy from the use of the full 23 factorial and Plackett-Burman (PBD) designs, and also the study of the effects of these factors on alkaline fusion steps, oxalic acid leaching and precipitation, had the following results in the step of alkaline fusion using full 2³ design, average yield of 67.27% increased to 96.20% when all variables simultaneously passed the level -1 to level +1. Analyzing Student t-test, the main effects of factors A (melting temperature), B (ratio of NaOH/xenotime), C (time), and the effect of interaction between A and B were statistically significant at a 95% level of confidence. The well-adjusted results in the graph of the predicted values versus experimental values gave an empirical model of first order with coefficient of determination R2=0.92, and with the analysis of the F test a Fcalculated value well above the Ftabulated, for a 95% level of confidence, was obtained. The RSM was obtained satisfactorily and correlated positively with the yield response variable so that, insofar as the manipulated variables increased, the reaction yield also increased. In the oxalic acid leaching and precipitation using the PBD, the analysis of the Pareto chart and desirability function showed that among the seven factors evaluated, the excess HNO3 (EA) showed the greatest importance in the process followed by temperature (TL) with significant positive correlations with the yield of the acid leaching step. Considering the attainment of oxalic precipitation solution from the acid leaching step with four of the seven most significant factors according to the PBD design, the concentration of precipitant solution used for conversion of oxalates in rare earth oxides was 80 g/mL, thus obtaining a satisfactory yield of 98.25%. This resulted in a material with adequate characteristics determined by chemical analysis, X-ray diffraction, scanning electronic microscopy and specific surface area, and indicated that the final cost production of RE2O3 can correspond to 80% reduction compared with commercial Y2O3. ACKNOWLEDGMENTS The authors thank the financial support of the Brazilian research funding institutions: CAPES, CNPq and FAPESP-00/08832-5. REFERENCES 1 1 K.M. Franken, Sci. Technol. 30, 7 (1995) 1941. Franken K.M. Sci. Technol. 30 7 1995 1941 1941 2 2 Y. Jiang, A. Shibayama, K. Liu, T. Fujita, Hydrometallurgy 76 (2005) 1. Jiang Y. Shibayama A. Liu K. Fujita T. Hydrometallurgy 76 2005 1 1 3 3 J.P. Barbosa, F.E. Lapido Loureiro, in: Rare earth minerals: chemistry, origin and ore deposits, Natural History Museum, London (1993) 7. Barbosa J.P. Lapido Loureiro F.E. Rare earth minerals: chemistry, origin and ore deposits Natural History Museum London 1993 7 7 4 4 R. Dehghan, M. Noaparast, M. Kolahdoozan, S.M. Mousavi, Int. J. Miner. Process. 89 (2008) 9. Dehghan R. Noaparast M. Kolahdoozan M. Mousavi S.M. Int. J. Miner. Process. 89 2008 9 9 5 5 C.M. Moore, In: “Mineral facts and problems”, U.S. Bureau of Mines, USA (1981) 739. Moore C.M. Mineral facts and problems U.S. Bureau of Mines USA 1981 739 739 6 6 C.K. Gupta, N. Krishnamurthy, Extractive metallurgy of rare earths, CRC Press, Boca Raton (2004). Gupta C.K. Krishnamurthy N. Extractive metallurgy of rare earths CRC Press Boca Raton 2004 7 7 S. Radhika, B.N. Kumar, M.L. Kantam, B.R. Reddy, Hydrometallurgy 110, 1-4 (2011) 50. Radhika S. Kumar B.N. Kantam M.L. Reddy B.R. Hydrometallurgy 110 1-4 2011 50 50 8 8 L.A. Boatner, Rev. Mineral Geochem. 48, 1 (2002) 87. Boatner L.A. Rev. Mineral Geochem. 48 1 2002 87 87 9 9 G.A. Moldoveanu, V.G. Papangelakis, Hydrometallurgy 117-118 (2012) 71. Moldoveanu G.A. Papangelakis V.G. Hydrometallurgy 117-118 2012 71 71 10 10 P. Alex, A.K. Suri, C.K. Gupta, Hydrometallurgy 50, 3 (1998) 331. Alex P. Suri A.K. Gupta C.K. Hydrometallurgy 50 3 1998 331 331 11 11 R.C. Vickery, The chemistry of yttrium and scandium, Pergamon Press, New York (1960). Vickery R.C. The chemistry of yttrium and scandium Pergamon Press New York 1960 12 12 P. Kikot, M. Viera, C. Mignone, E. Donati, Hydrometallurgy 104 (2010) 494. Kikot P. Viera M. Mignone C. Donati E. Hydrometallurgy 104 2010 494 494 13 13 K. Wang, J. Li, R.G. McDonald, R.E. Browner, Hydrometallurgy 109 (2011) 40. Wang K. Li J. McDonald R.G. Browner R.E. Hydrometallurgy 109 2011 40 40 14 14 A. Shemi, S. Ndlovu, V. Sibanda, L.D. Van Dyk, Int. J. Miner. Process. 127 (2014) 10. Shemi A. Ndlovu S. Sibanda V. Dyk L.D. Van Int. J. Miner. Process. 127 2014 10 10 15 15 D.C. Montgomery, Design and analysis of experiments, Wiley, New York (2001). Montgomery D.C. Design and analysis of experiments Wiley New York 2001 16 16 P. Wang, Z. Wang, Z. Wu, Chem. Eng. J. 193-194 (2012) 50. Wang P. Wang Z. Wu Z. Chem. Eng. J. 193-194 2012 50 50 17 17 A. Tuncuk, S. Ciftlik, A. Akcil, Hydrometallurgy 134-135 (2013) 80. Tuncuk A. Ciftlik S. Akcil A. Hydrometallurgy 134-135 2013 80 80 18 18 A. Vatanara, A.N. Rouholamini, K. Gilani, R. Asgharian, M. Darabi, M. Rafiee-Tehrani, J. Supercrit. Fluid 40 (2007) 111. Vatanara A. Rouholamini A.N. Gilani K. Asgharian R. Darabi M. Rafiee-Tehrani M. J. Supercrit. Fluid 40 2007 111 111 19 19 Z. Ekinci, E. Şayan, A.V. Beşe, O.N. Ata, Int. J. Miner. Process. 82 (2007) 187. Ekinci Z. Şayan E. Beşe A.V. Ata O.N. Int. J. Miner. Process. 82 2007 187 187 20 20 R.L. Mason, R.F. Gunst, J.L. Hess, Statistical design and analysis of experiments, with aplications to engineering and science, John Wiley & Sons, New Jersey (2003). Mason R.L. Gunst R.F. Hess J.L. Statistical design and analysis of experiments, with aplications to engineering and science John Wiley & Sons New Jersey 2003 21 21 R. Chi, Z. Xu, Metall. Mater. Trans. B 30 (1999) 189. Chi R. Xu Z. Metall. Mater. Trans. B 30 1999 189 189 22 22 J. Baltazar-Rodrigues, P. Rodrigues Jr, G.K. da Cruz, M.H. Lente, J.A. Eiras, Mater. Res. 17, 1 (2014) 1. Baltazar-Rodrigues J. Rodrigues P. Jr Cruz G.K. da Lente M.H. Eiras J.A. Mater. Res. 17 1 2014 1 1 23 23 X. Li, T. Liu, Q. Lin, R. Cao, Cryst. Growth Des. 10, 2 (2010) 608. Li X. Liu T. Lin Q. Cao R. Cryst. Growth Des. 10 2 2010 608 608 24 24 L. Muresan, E.-J. Popovici, R. Grecu, L.B. Tudoran, J. Alloy Compd. 471, 1-2 (2009) 421. Muresan L. Popovici E.-J. Grecu R. Tudoran L.B. J. Alloy Compd. 471 1-2 2009 421 421 25 25 H.-J. Förster, Am. Mineral. 83 (1998) 1302. Förster H.-J. Am. Mineral. 83 1998 1302 1302 26 26 R. Chi, Z. Xu, Metall. Mater. Trans. B 30 (1999) 189. Chi R. Xu Z. Metall. Mater. Trans. B 30 1999 189 189 27 27 M.S. Safarzadeh, M.S. Bafghi, D. Moradkhani, M.O. Ilkhchi, Miner. Eng. 20 (2007) 211. Safarzadeh M.S. Bafghi M.S. Moradkhani D. Ilkhchi M.O. Miner. Eng. 20 2007 211 211 28 28 S.C. Chehreh, M. Rudolph, T. Leistner, J. Gutzmer, A. Urs Peuker, Int. J. Mining Sci. Technol. 25 (2015) 877. Chehreh S.C. Rudolph M. Leistner T. Gutzmer J. Urs Peuker A. Int. J. Mining Sci. Technol. 25 2015 877 877