Characterization of micaceous minerals occurring in phosphate ores in Brazil: emphasis on the interstratified phyllosilicates

Micaceous minerals were characterized in three phosphatic ores from mines in Western Minas Gerais State, Brazil, all having complex chemical and mineralogical compositions, with emphasis on interstratified phyllosilicates. Knowledge of the relationship between the physical and chemical characteristics of the phyllosilicates contributed to a better understanding of the physicochemical processes in the mineral processing stages. Characterization of the minerals was carried out mainly by X-ray diffractometry (XRD), which allowed identifying vermiculite and three interstratified phyllosilicate phases. These minerals presented regular stratification of the unit cell, with different percentages of vermiculite and mica. Complementary studies by scanning electron microscopy (SEM) with EDS microanalysis, thermogravimetry and chemical analysis confirmed the identifications by XRD.


Introduction
Phosphate ores are vital, non-renewable, non-recyclable and essential resources in the phosphorus-based chemical and fertilizer industry (Liu et al., 2017). The main Brazilian phosphate deposits are located in the Alto do Paranaíba Igneous Province, in the states of Goiás (Catalão) and Minas Gerais (Araxá and Tapira) and have low P 2 O 5 content (Albuquerque et al., 2012). In these deposits, after the action of weathering, the micaceous apatitic ores can be encountered in the lower part of the isalteritic saprolite horizon, along with large amounts of goethite, anatase, quartz, phlogopite, vermiculite and interstratified phyllosilicates (Azzone and Rubert, 2010).
Phyllosilicates are an important group of minerals that are part of the main components of the earth's crust. Minerals that have tetrahedral and octahedral layers or lamellae are phyllosilicates and are differentiated by the way the layers are stacked in the unit cell, such as: mica, chlorite, serpentine, clay minerals and vermiculite (Escamilla- Roa et al., 2013;Bian and Kawi, 2020).
In currently mined phosphatic ores, apatite is the main phosphorus-bearing mineral and, therefore, the mineral of interest; while carbonate and silicate minerals, including phyllosilicates, quartz, clays, iron and titanium minerals, feldspars, pyroxenes, calcite and dolomite, are the main gangue minerals to be removed (Liu et al., 2017;Sis and Chander, 2003;Mohammadkhani et al., 2011).
The phyllosilicates are important gangue minerals and can pose serious problems in the selectivity in the froth flotation of these apatite ores. The weathered species, vermiculites and the The samples were dry comminuted in an agate mortar and sieved, obtaining 100% of particles passing the screen size of 38µm, which were used in the analyses for characterization.
Quantitative chemical analyzes were performed by wavelength dispersion X-ray fluorescence (WDS), using 2g of each sample prepared by the fusion technique with Li 2 B 4 O 7 flux in a Claisse fusion machine. The equip-ment used was a PW 2404 spectrometer (Philips-PANalytical), with a rhodium anode tube.
Particles from the three samples were examined and the images obtained using a scanning electron microscope (SEM), FEI brand, model Quanta 200 FEG (field-emission gun). The chemical analyzes of microregions were performed with an energy dispersion Xray spectrometer (EDS) Bruker brand, Esprit software. The mineral phases present in the samples were verified using the X-ray diffractometry (XRD) technique, powder method. The PW1710 X-ray diffractometer (Philips-PANalytical), with point detector, graphite crystal monochromator, copper tube (Z=29), average λ Kα = 1.54184Å and λ Kα1 = 1.54056Å was used. The instrument conditions were: power supply of 50kV and 35mA, scanning range from 3 to 90° in 2θ, step size of 0.06° (2θ) and counting time of 3 seconds.
When performing the XRD analyses, it was observed that misalignment may occur in the equipment's goniometer, thus displacing the peaks. This misalignment can even occur depending on how the sample is placed and adjusted to be analyzed in the equipment. Due to this potential problem, the following methodology was used: a natural sample of a chlorite schist, composed of chlorite, quartz and phlogopite called "linde" was used as standard to observe if there was displacement of the diffraction peaks during the analysis and make the necessary corrections.
This standard sample was submitted to quantitative analysis by X-ray diffractometry (XRD), using the procedure based in the Rietveld method.
The following procedure has always been followed: the standard was run in the diffractometer; then, the samples were analyzed by XRD and the standard sample was submitted to XRD as a second, final check. The differences between the ideal principal lines (100 planes) and the actual observed lines were calculated. The calculations were made for the main lines of chlorite and phlogopite in the "linde" standard as follows: for chlorite the ICDD 83-1381 standard number was used, which shows the main peak at 14.2497Å, and for phlogopite, the 85-2275 card number, which has the main line at 10.3800Å. The difference between the main line of the ICDD form and the respective value found in each analyzed pattern was calculated. These standard cards were from the PDF-2 Release 2010 database of the ICDD -International Center for Diffraction Data; the software used to read and compare the data was X´PertHighScore version 2011, from Panalytical.
In Table 1, these differences are presented. Next, the mean value found between the patterns analyzed before and after the analysis of the samples was calculated and thus the mean value of the displacements for the two important lines of the diffractogram was found. With the mean value found, the correct position of the peaks of the three samples studied was calculated.
To identify the interstratified phyllosilicates, data from Reynolds (1980) and Moore & Reynolds (1997) were used. It was based on the comparison between observed total diffractograms and model diffractograms for mixed or interstratified layers.

Characterization
interstratified varieties, are particularly harmful (Assis et al., 2010). The problem is even more serious for the interstratified silicates because they are poorly studied, mainly in the apatite ores. This is the main motivation of this study, to characterize these minerals and to understand their influence in the process.
This article's main objective is to identify the minerals present in three samples of micaceous apatite ores, with emphasis on interstratified phyllosilicates. The knowledge of the relationships between the physical and chemical char-acteristics of these phyllosilicates can contribute to a better understanding of the physical-chemical processes of the mineral processing of such ores. This study is justified in view of a near future scenario that points to the need to use these ores with more complex mineralogy.
Thermogravimetric analyzes were carried out up to 1000°C with a heating rate of 10°C/min, N 2 flow of 20 mL/min in an alumina crucible; the equipment used was the Shimadzu DTG-60H thermobalance.
The results confirmed the presence of the most interesting minerals in the study: interstratified phyllosilicates, vermiculite and apatite, as shown in the following diffractograms.
In the Rocha Alterada sample, X-ray diffraction identified the following minerals (

X-ray diffraction and quantitative chemical analysis
All samples have interstratified phyllosilicates, identified by the presence of the peaks between 11.59Å and 12.63Å in the X-ray diffractograms and in Table 3. This identification was based on the comparison between the observed complete diffractograms and the model diffractograms for mixed layers or interstratified layer phyllosilicates, according to data from Reynolds (1980) and Moore and Reynolds (1997). Table 4 shows the data from the Reynolds model (1980) and the X-ray diffraction data from the samples. According to the values presented by Reyn-olds (1980), it was observed that the interstratified phyllosilicates present in the samples correspond to mixed structures of the mica/vermiculite type, the layers being ordered or regular. However, the proportions of mica and vermiculite were different in the three cases, as shown in Table 4.
The mineral compositions of the three samples are consistent with their chemical analyzes in Table 2 and with results from previous research (Chula and Brandão, 2004;Neumann et al., 2011).

Juliana Angélica Evangelista de Carvalho et al.
All phyllosilicates observed in SEM images have lamellar habit and exfoliation in very thin plates (Figures 4, 5 and 6). In all EDS microanalysis, these phyllosilicates showed variable compositions, but within short ranges, of the SiO 2 , Al 2 O 3 , MgO, CaO and FeO elements (reported as oxides), in Tables 5, 6 and 7. Therefore, the K 2 O content is the only microchemical parameter to distinguish the mineral species: a) when the % of K 2 O is very low or tends to zero, the mineral is vermiculite (Gruner, 1934); b) when this content lies between 1.0 and 8.5%, this characterizes the mixed layer or interstratified phyllosilicate; c) the original or even slightly altered mica would have K 2 O contents ranging from 9.2 to 11.8% (WebMinerals, 2017). EDS microanalysis (Table 5) was performed on the big particle in Figure  4, which provided high levels of silicon oxide, iron and aluminum, average content of magnesium oxide and low content of calcium and titanium. The potassium oxide content (5.75%), together with the main particle morphology characterizes an interstratified phyllosilicate. Figure 5 shows a lamellar particle. By microanalysis (Table 6), carried out in the particle center, an interstratified phyllosilicate is identified, with a higher content of the mica component, due to the relatively high K 2 O content, but still below 9.2%.     Figure 4. Table 6 -EDS microanalysis of the Rocha Alterada sample in Figure 5.

Thermogravimetric analysis
The thermogravimetry (TG) results of the Mosaic sample are shown in Figure  7 and Table 8. The temperature and mass loss data are consistent with the composition of minerals that undergo pyrolysis in the studied temperature range (Speyer, 1993;Földvári, 2011). In this sample, these minerals were vermiculite, interstratified phyllosilicate and dolomite.   In Figure 6 there is a lamellar particle in the central part of the image. Micro-analysis (Table 7) was performed at the particle center. Thus, this is practically a biotite mica, due to the percentage of K 2 O already very close to the 9.2 limit.    Figure 6.
The thermogravimetric data of the Mica Base 1 sample, shown in Figure   9 and Table 10, demonstrate that, also here, interstratified phyllosilicate and vermiculite are the ones that lose mass, like H 2 O. Figure 8 and Table 9 show the thermogravimetric analysis of the Ro-cha Alterada sample. The minerals that undergo pyrolysis are interstratified phyllosilicate and vermiculite.

Acknowledgments References
Quantitative chemical, XRD, TG and SEM/EDS analyzes indicated the mineralogical compositions for the studied samples. Detailed analysis by XRD allowed to determine the specific unity cell stratification, that is, the stacking sequence of the mica and vermiculite layers, in the 3 types of mixed layer phyllosilicates identified in the samples, which were always ordered (regular), according to the model by Reynolds (1980 Potassium content, reported as K 2 O, is the only microchemical parameter to distinguish the phyllosilicate in EDS microanalyses. a) when the % of K 2 O is very low or tends to zero, it is vermiculite; b) when this content is between 1.0 and 8.5%, this characterizes the mixed layer or interstratified phyllosilicate; c) the original or even slightly altered mica would have K 2 O contents ranging from 9.2 to 11.8%.
The improved knowledge of the mineralogical composition of these complex ores can contribute to better explain problems encountered in their concentration. The micaceous minerals which are a great part of this problem and are now better known, including the fact that fresh mica (biotite) is very rare. The weathered products are the dominant phyllosilicates: vermiculite and interstratified species.
The authors would like to thank everyone who contributed to this work. They also thank the following institu-tions for their academic and financial support: Escola de Engenharia, UFMG, PPGEM and CAPES / PROEX. Also thanks to CNPq for a scholarship and research grant.
The thermogravimetry results are consistent with the data collected in the other analyses. In the case of the Mica Base 1 sample, which has the highest interstratified phyllosilicate content, thermogravimetry showed the highest relative mass loss, with 4.1%, compared to the vermiculite loss, which was 3.1%.
In two samples, two ranges were observed at the highest temperatures, where there was mass loss. Both are attributed to the interstratified phyllosilicate, as follows: in the lower range, the pyrolysis would be of the H 2 O from the vermiculite layers, and in the higher range, the loss would be due to the hydroxyls of the mica layers.