XRD APPLICATION NOTE X-ray diffraction (XRD) for grade control of iron ores Beneficiation techniques for iron ore are receiving worldwide attention due to the development of new deposits of lower grade ore. This application note describes the benefits of resolving the mineralogical ore composition by use of X-ray diffraction (XRD) in order to optimize concentrator feed and operational efficiency. Benefits Cost-effective and quick return on investment (ROI): wrong classification of 1 grade block with 2, t iron ore represents a loss of 1 k No human factor for grade estimation Short measurement times for quick and easy quality check of concentrator feed Simple sample handling, without involving chemicals Fully automatable analysis also in combination with XRF The Analytical X-ray Company
Increasing operational efficiency in iron ore beneficiation with XRD The mineralogy of ore sample is often not considered or at best inferred from visual geological logging of the sampled material. As such, mineral contaminants that cannot be identified visually and / or by chemical analysis alone are often not quantified. For example the identification of gibbsite via chemical data is typically problematic due to the co-occurrence of kaolinite and aluminous goethite. The correct identification of gibbsite is important as gibbsite can have a significantly adverse impact on sinter efficiency of the iron ore. Gibbsite-bearing ores require longer sintering times and higher sintering temperatures in order to promote melt formation. Information provided by XRD gives the mine geologist the opportunity to create maps of the spatial distribution of minerals such as gibbsite or kaolinite (see Figure 1) and provides the possibility to blend and / or isolate these minerals throughout the defined grade blocks. Common practice in grade control for iron ore mines include the chemical analysis of ore and waste for key elements. Selected grade blocks are designated to high grade, low grade or waste destinations based on the abundance of these elements. In this example the theoretical grade blocks could be redefined on the basis of the XRD data (Figure 2). In comparison with the original high grade blocks domain (Figure 2A) the revised high grade blocks could be separated into three different classes; high grade (HG), high grade gibbsite (HGG) and high grade beneficiation (HGC) (Figure 2B). Both high - and low grade beneficiation (LGC) feed have been defined on the basis of relatively high kaolinite content making these blocks amenable to beneficiation. Low grade other (LGO) comprises relatively low kaolinite material which is not readily amenable to traditional forms of beneficiation. Waste (W) are grade blocks with a very low total iron content (Paine et al. 211). Figure 1. Contour plots showing the distribution of kaolinite and gibbsite for one iron ore deposit, Western Australia (Paine et al. 211), 6 samples measured Figure 2. Original (A) and revised (B) grade blocks defined on the basis of varying mineralogy, see also Figure 6, for one iron ore deposit, Western Australia (Paine et al. 211)
Cluster analysis for grade control Cluster analysis of the raw scans was performed prior to subsequent analyses, such as phase identification and quantification. It can be used to simplify data processing significantly by automatically sorting closely related scans into clusters. The most characteristic scans of each cluster are identified. The amount of data that has to be processed in order to characterize each cluster is drastically reduced this way. After completion of the cluster analysis of all the characteristic ore grades, the results can be used to filter all following similar samples with varying mineralogy before quantitative analysis is applied. In this case, all 6 scans of different iron ore samples were clustered. Figure 3 shows a magnified view of the scans plotted on top of each other, visualizing the analysed phases of the ore minerals. Cluster analysis was applied and a principal component analysis (PCA) plot was calculated, as can be seen in Figure 4. Each circle represents one scan which stands for one iron ore sample, respectively. The size of the circles represents the total iron content. Three different clusters are found with around each cluster displayed semi-transparent confidence spheres. The shape of the PCA plot indicates the presence of a complete range of ore grades. The PCA plot was modified in order to display one additional property of the datasets as can be seen in Figure 4. The total Fe content, determined by XRF analysis, is plotted as circles with different sizes. Big circles indicate high Fe content (blue) and the small circles indicate low Fe content (green). The chemical data underline the 3 different grades determined from XRD. Figure 3. Large amount of data measured in a short time High grade high total Fe Medium grade Low grade low total Fe Figure 4. Principal component analysis (PCA) plot of 6 iron ore samples divided into 3 different clusters (grades) with different mineralogy, displaying semi-transparent confidence spheres around each cluster
Phase identification and quantification Phase identification The most characteristic scans of each cluster are used for phase identification. In addition to the main iron ore minerals hematite and goethite the samples also contain significant amounts of kaolinite, gibbsite, and quartz, Figure 5. Intensity [counts] 15 Hem / Phase identification was performed using PANalytical s HighScore Plus software package. Batch programs within the software are available to enable automation of phase identification for the demanding mining environment, delivering results at the push of a button. 1 5 Kao Gib Hem Qua Kao Kao Hem Figure 5. Measurement and phase identification of one iron ore sample from Western Australia ( = thite, Hem = Hematite, Kao = Kaolinite, Gib = Gibbsite, Qua = Quartz) 15 2 25 3 35 4 45 2θ (Co) 68.9 % 26.6 % Phase quantification Modern methods like the Rietveld refinement in combination with highspeed data collection allow for fast and reliable quantification of iron ores. The full pattern Rietveld method has several advantages compared to classical quantification methods. All crystalline phases can be quantified in just a few seconds without line overlaps, sample height and preferred orientation influencing the results. No standards, monitors or calibration are needed. Counts 15 1 Hematite 26.6 % thite 68.9 % Kaolinite-2M 1.4 % Gibbsite 1.1 % Quartz 2. % 2. % 1.1 % 1.4 % Figure 6 shows the graphical output of a Rietveld refinement of one typical iron ore sample. The difference plot highlights the fit between the measurement and the profile calculated by the software. 5 1 2 3 4 5 6 7 8 2θ (Co) Figure 6. Rietveld refinement of the same iron ore sample as in Figure 3 (red = measurement, blue = calculation; below = difference plot, RProfile = 3.7 (mathematical quality of the fit))
During the Rietveld quantification crystallographic parameters such as unit cell dimensions and occupation factors are calculated. These parameters can be used to identify goethite with different Al contents as shown in the last column of Table 1. Within the crystallographic structure of goethite, the Fe atoms can be substituted with Al atoms up to 3 mol%. The 3 ore grades are: High grade iron ore : High goethite, medium hematite, low gibbsite-kaolinite-quartz Medium grade iron ore: High goethite, low hematite, medium gibbsite-kaolinite-quartz Low grade iron ore: Low goethite, medium hematite, high gibbsite-kaolinite-quartz All 6 samples were analyzed and quantified using the measurement conditions (see Table 1). The concentration ranges for the different grades correlate with the three clusters as demonstrated in Figure 7. The benefit of resolving the mineralogical composition of iron ore is to optimize the concentrator feed. Gibbsite and kaolinite show different properties during the beneficiation process. Gibbsite and aluminous goethite cannot be removed from the ore but can be controlled by blending with low-al containing ore. Kaolinite can be liberated during wet concentration. The three grades specified in this example are distinguished by the goethite content and the concentration of the Al containing phases. The hematite content does not vary significantly between the three grades. Quantitative results are summarized in Table 2 showing the minimal, maximal and average mineral concentrations. % % 1 9 8 7 6 5 4 3 2 1 18 16 14 12 1 8 6 4 2 thite High grade Medium grade Low grade Kaolinite High grade Medium grade Low grade % % 6 5 4 3 2 1 6 5 4 3 2 1 Hematite High grade Medium grade Low grade Gibbsite High grade Medium grade Low grade Figure 7. Phase concentrations goethite, kaolinite and gibbsite versus the cluster number of 6 iron ore samples Cluster 1 High grade 2 Medium grade 3 Low grade Hematite thite Gibbsite Kaolinite Quartz Magnetite Mol % Al in goethite Min 3.1 56.1.... < 5 Max 23.1 94.3 1.1 4.6.8 2.2 < 5 Average 16.9 75.3.6 2.2.1.7 < 5 Min 5.9 5.7.5... < 5 Max 39.1 91.2 4.1 8.8 4.8. < 5 Average 19.5 74.2 1.9 3.5.9. < 5 Min 15.7 37.4 1.9 4.3.. < 5 Max 49.5 76.8 4.9 16.2 3.7. 5 to 12 Average 3.3 55.2 3. 9.9 1.6. 5 to 12 Table 1. Concentration ranges of the phase content of 6 iron ore samples obtained by Rietveld refinement
Precision, accuracy and detection limits To check the accuracy, precision and determination limits of the Rietveld refinements, synthetic mixtures were prepared and measured ten times. Table 2 shows the results of a phase mixture prepared from pure goethite, hematite, magnetite and quartz. Average values indicate an accuracy of less than.5 %. Standard deviations between.2% and.6% prove the high precision of the method. Expected concentration thite Hematite Magnetite Quartz 6% 25% 1% 5% 1 59.4 25. 1.5 5.1 2 59.5 24.6 1.3 5.6 3 59.8 25. 1.4 4.8 4 6.1 24.7 1.5 4.7 5 6.2 25.1 1. 4.7 6 6. 24.7 9.8 5.5 7 59.6 25. 1.3 5.1 8 59.8 24.9 1.3 5. 9 59.7 24.8 1. 5.5 1 59.5 25.2 1.2 5.1 Average 59.8 24.9 1.2 5.1 Absolute RMS.27.19.23.33 Relative RMS %.45.78 2.26 6.49 Table 2. Accuracy and precision determined on a synthetic mixture of pure goethite, hematite, magnetite and quartz The detection limits of minor phases were tested on mixtures of goethite with small amounts of quartz (see Table 3). The determination limit was found to be.5%. Expected concentration thite Quartz thite Quartz thite Quartz 99.5%.5% 99% 1% 98% 2% 1 99.6.4 99.2.8 98. 2. 2 99.5.5 99. 1. 98. 2. 3 99.5.5 98.8 1.2 98. 2. 4 99.6.4 99.2.8 97.8 2.2 5 99.5.5 99. 1. 98.1 1.9 6 99.4.6 99.1.9 97.8 2.2 7 99.3.7 99.2.8 97.9 2.1 8 99.6.4 98.9 1.1 98.2 1.8 9 99.5.5 98.8 1.2 98. 2. 1 99.4.6 98.9 1.1 98.1 1.9 Average 99.5.5 99. 1. 98. 2. Absolute RMS.1.1.16.16.13.13 Relative RMS %.1 19.5.16 16.11.13 6.4 Table 3. Determination limits of quartz in goethite
Experimental setup Instrument setup Instrument CubiX 3 Iron X-ray tube Cobalt LFF Incident beam optics Fixed divergence slits 1/2, antiscatter slit 1, mask 15 mm,.4 rad Soller slits Incident filter Iron Sample stage Spinning sample stage Diffracted beam optics Fixed antiscatter slits 1/2,.4 rad Soller slits Detector X Celerator Scan range 8 to 8 2θ Step size.2 2θ Scan time 5 minutes 56 seconds Table 4. Configuration and measurement conditions Sample preparation For this study sixty iron ore samples from one deposit in Western Australia (WA) were analyzed. The grain size of iron ore is far too large for direct analysis. Samples were, therefore, ground and pressed with a compact sample preparation machine into steel ring sample holders. Sample handling and preparation are critically important for reliable phase quantification. It is recommended that a standardized or automated sample preparation procedure is used. The measurements were made using a CubiX 3 Iron industrial diffractometer, suitable for all types of iron-containing samples. A cobalt tube with incident beta filter (Fe) is especially suited for ironcontaining materials, as it produces high-resolution data without disturbing sample fluorescence. The measurements can be done in less than 6 minutes using a high-speed X Celerator detector.
Conclusions Rapid XRD analysis and associated cluster analysis offer additional criteria for the definition of grade blocks in iron ore mining and have advantages compared with other definitions that are only based on chemistry. The additional information, such as the concentration and spatial distribution of gibbsite, aluminous goethite and kaolinite provide the mine geologist with the possibility to blend and/or isolate these minerals throughout the defined grade blocks. In addition, material with a tendency for higher degrees of beneficiation can be identified and domained. Conversely, material with limited upgrade potential can be identified without submitting it to traditional forms of beneficiation. In sum, the use of XRD with its modern optics, high-speed detectors and software provides rapid and accurate mineral analysis suitable for process control environments and can increase operational efficiency. Summary XRD can be used for fast and accurate cluster analysis, phase identification and quantification of the minerals present in iron ore. The ore contains phases as hematite, goethite and magnetite as well as the gangue minerals gibbsite Al(OH) 3, kaolinite Al 2 Si 2 O 5 (OH) 4 and quartz SiO 2. In addition, different XRD methods can distinguish between goethite with varying Al contents. Theoretical grade blocks can be defined on the basis of the mineralogy determined by XRD, indicating ore grades that can be effectively and economically upgraded in the concentrator. Exploration of iron ore deposits can be done using cluster analysis. This method can distinguish between different ore grades and facilitate multidimensional compositional mapping of ore deposits. Cluster analysis identifies regions of favorable mineral compositions without detailed knowledge about the phase content. Global and near PANalytical B.V. Lelyweg 1, 762 EA Almelo P.O. Box 13, 76 AA Almelo The Netherlands T +31 () 546 534 444 F +31 () 546 534 598 info@panalytical.com www.panalytical.com References PAINE, M., KÖNIG, U. & STAPLES, E. (211): Application of rapid X-ray diffraction (XRD) and cluster analysis to grade control of iron ores. In: Broekmans, M.A.T.M. (ed.): Proceedings 1th International Congress for Applied Mineralogy (ICAM), Trondheim, 495-51. ISBN-13: 978-82- 7385-139- Further reading KÖNIG, U., GOBBO, L. & MACCHIAROLA K. (211): Using X-ray diffraction for grade control and minimizing environmental impact in iron and steel industries. Proceedings Iron Ore 211 conference, 11-13 July, Perth, WA, 49-56. KÖNIG, U., GOBBO, L. & REISS, C. (211): Quantitative XRD for ore, sinter and slag characterization in the steel industry. In: Broekmans, M.A.T.M. (ed.): Proceedings 1th International Congress for Applied Mineralogy (ICAM), Trondheim, 385-393. ISBN-13: 978-82- 7385-13 Regional sales offices Americas T +1 58 647 11 F +1 58 647 1115 Europe, Middle East, Africa T +31 () 546 834 444 F +31 () 546 834 499 Asia Pacific T +65 6741 2868 F +65 6741 2166 Although diligent care has been used to ensure that the information herein is accurate, nothing contained herein can be construed to imply any representation or warranty as to the accuracy, currency or completeness of this information. The content hereof is subject to change without further notice. Please contact us for the latest version of this document or further information. PANalytical B.V. 29. Printed in The Netherlands on 5% recycled, chlorine-free paper. 9498 72 19511 PN899