APPENDIX A: Quantitative Mineralogy: A Comparison of Mineral Liberation, X-‐ray Diffraction, Rietveld Refinement, Quantitative Evaluation of Minerals by Scanning Electron Microscopy and Whole Rock Analyses This appendix consists of a report summarizing the comparison of mineral quantification through mineral liberation analysis (MLA), X-‐ray diffraction (XRD), Rietveld refinement (RR), quantitative evaluation of minerals by scanning electron microscopy (QEMSCAN) and whole rock analysis, calculated using the MINSQ worksheet, as well as all of the results from the aforementioned quantification methods as follows: A.I XRD A.II RR A.III MLA results A.IV QEMSCAN A.V MINSQ 122 A.1 Abstract There are many methods for quantifying mineralogy, but which is most appropriate for a given application? Using petrography results as a basis for comparison, this question is considered by quantifying ten samples of homogeneous, quartz diorite, taken over the Relincho Cu-‐Mo porphyry deposit in Atacama, Chile. A comparison of the strengths, weaknesses and relative accuracy of X-‐Ray Diffraction (XRD), Rietveld Refinement (RR), Mineral Liberation Analysis (MLA), Quantitative Evaluation of Minerals by Scanning Electron Microscopy (QEMSCAN), and calculation from whole rock data using the MINSQ spreadsheet, identify QEMSCAN, RR and MINSQ as the most effective methods of quantification for these samples. QEMSCAN and MLA use a carbon coated, polished thin sections to produce a coloured mineral map and can report modal mineralogy as well as physical properties if requested. XRD, RR and MINSQ use pulp material and only determine modal mineralogy. Though QEMSCAN and MLA are powerful mineral quantification tools, result accuracy is dependent upon substantial geological knowledge of the sample material and an accurate spectral library. As spectral libraries tend to improve with increasing analyses, QEMSCAN and MLA are ideal for large-‐scale, long-‐term quantification programs, though ill suited for small-‐scale projects with no pre-‐existing spectral library. Methods XRD, RR and MINSQ quantify mineralogy relatively quickly, easily and inexpensively. Attractive for small-‐scale programs, XRD, RR and MINSQ are less practical for large-‐scale projects as interpretation is completed sample-‐by-‐sample with little automation of the interpretation. 123 A.2 Introduction A variety of methods exist for mineral quantification, all of which have particular applications for which they are best suited. Petrographic quantification is well applied to small-‐scale programs, but is laborious and can be fraught with human error, including misidentification of minerals and over or under estimation of composition. Other quantification methods, such as Quantitative Evaluation of Minerals by Scanning Electron Microscopy (QEMSCAN), and Mineral Liberation Analysis (MLA), require an accurate spectral library for quantification, and can produce a wealth of information including visually intuitive coloured maps, physical properties of mineral grains and elemental distribution matrices. Quantification by X-‐ray diffraction (XRD), Rietveld refinement (RR) and calculations from whole rock and multi-‐element results using the MINSQ spreadsheet (Herrmann & Berry, 2002) are simple by comparison to MLA and QEMSCAN, producing only modal mineralogy, but also requiring less geological input. Each of these six methods has a suitable application, though none are suitable for every situation. A comparison of quantification results of ten homogeneous, quartz diorite samples from the Relincho Cu-‐Mo porphyry deposit in Atacama, Chile by QEMSCAN, MLA, XRD, RR and MINSQ examines the strengths and weaknesses of each methodology. A description of each methodology, along with a comparison of strengths, weaknesses and relative accuracy, allows for an assessment of each method. Scanning Electron Microscope (SEM) and Electron Microprobe (EMP) data in combination with petrographic observations offer a basis for comparison and accuracy for quantification results. Based on these results, the reader can make an informed decision as to which methodology is most appropriate for their application. A.3 Project Background The Relincho deposit is located at the southern extent of a belt of Paleocene porphyry deposits running from southern Peru to central Chile (Figure A.1). A Paleocene, homogeneous quartz diorite hosts mineralized Paleocene porphyry units of the same source: the Los Morteros batholith. Average composition of the quartz diorite host is 60-‐75% plagioclase (An32 from EMP results), 8-‐15% quartz, 8-‐15% hornblende (magnesio-‐hornblende from EMP results), 4-‐8% biotite and 0-‐2% magnetite. Alteration assemblages 124 are potassic, propylitic and phyllic defined by secondary biotite -‐ K-‐feldspar ± glassy limonite (from chalcopyrite), epidote -‐ chlorite -‐ hematite ± albite ± calcite ± pyrite, and chlorite -‐ quartz -‐ sericite ± calcite, respectively. Temporally the potassic alteration occurred first, followed by propylitic, then by phyllic. Propylitic alteration extends most distally, and potassic is strongly associated with mineralization (Figure A.2). A.4 Methods Mineral quantification methodologies determine modal mineral composition and speciation by analyzing either pulp materials or polished thin sections (PTS). The Vancouver Petrographics made PTS were described, then carbon coated for QEMSCAN, MLA, SEM and EMP work. Approximately 6 kg of sample material was submitted to Acme Laboratories in Vancouver, Canada for preparation, XRD, whole rock and ultratrace multi-‐element analyses. Whole rock and multi-‐element results were processed using the MINSQ spreadsheet to quantify the mineralogy (Herrmann & Berry, 2002). The RR work was performed on pulp made from the remaining material leftover from the PTS. The PTS based methodologies (MLA and QEMSCAN) are both non-‐destructive, produce a coloured mineralogical map and have the added capacity to report inter-‐mineral relationships, elemental distribution and physical properties including grain size, shape and density. Pulp-‐based methodologies (XRD, RR and quantification from whole rock data) produce a list of minerals and their concentrations; for the XRD based methodologies (RR and XRD) a diffraction pattern is also reported. Figure A.3 provides a side-‐by-‐side comparison of products. A.4.1 Sample Selection Samples were chosen to represent the major alteration assemblages at different intensities. Ten quartz-‐diorite surface samples taken across the Relincho deposit were allotted alteration intensity values from 1 to 3 representing insipient (less than 30% of the original mineral replaced by the alteration mineral), partial (30-‐50%) and complete (over 50%) based on hand sample and thin section observations (Table A.1). 125 Figure A.1: Map of the global regional porphyry belts and the porphyry belts of Chile including the location of the Relincho deposit Figure A.2: Alteration footprint of the Relincho deposit, based on hand sample and thin section observations 126 Figure A.3: Products from each quantification method for sample AA-‐03A: A. XRD diffusion pattern with the excerpt from the report stating modal abundances; B. RR with coloured lines indicative of minerals, and a grey line below zero indicating the difference between the sample diffraction pattern and the fit of the identified minerals; C. MLA results with mineral list and modal abundances. An outline shows the area run for QEMSCAN, note pixelation due to instrument issues; D. QEMSCAN results with the respective list of minerals and modal compositions; E. MINSQ results spreadsheet, minerals with * used electron microprobe calculated compositions for quantification. Quartz…….. Moderate Abundance (> 25 %) Albite…….. Moderate Abundance (> 25 %) Orthoclase…….. Minor to Moderate Abundance (10-25%) Clinochlore…….. Minor Abundance (5-10 %) Quartz 23.61 % Clinochlore II 3.14 % Actinolite 3.56 % Albite low, calcian 56.56 % Orthoclase 10.36 % Magnetite 1.22 % Titanite 1.55% 127 Alteration in Mafic Sites Alteration in Felsic Sites Sample Alt Ass Biotite Chlorite Epidote Musc Albite K- Feldspar AA-03A Pr 3 1 2 1 AC-01 Pr 3 2 AH-03 F C-04A K-Ph 2 2 1 F-03 K-Pr 1 1 1 1 G-08 K-Pr 1 2 2 2 K-08 K 3 1 2 K-18 K-Ph 3 2 1 Q-08B Pr-Ph 3 1 2 S-06A K-Pr 2 2 2 3 Table A.1: Selected samples and their respective alteration. Assemblages are: potassic (K, dark grey highlight), propylitic (Pr, medium and light grey), phyllic (Ph, light grey and white) and fresh (F), or combinations written in chronological order. Alteration intensity indicated by 1 (<30% replacement of primary mineral with alteration mineral), 2 (30%<replacement<60%), or 3 (>60% replacement). 128 A.4.2 Mineral Quantification Methods A brief description of each method is necessary for understanding the results, strengths and shortcomings of each. All require the operator to have a basic geological understanding of the sample material, without which erroneous results may occur. It is important to note that QEMSCAN and MLA analyses require an accurate, precursory, spectral library that matches, as closely as possible, the spectra of minerals in the samples. One mineral could be represented by several spectra based on slight crystallographic or compositional changes. Creating these libraries is laborious, but essential for accuracy and completeness. Whenever possible users are encouraged to create the library themselves and be involved with data processing. All samples have undergone petrographic quantification, using SEM results to ensure correct mineral identification. Each sample has been photographed in polarized and cross-‐polarized light prior to quantitative work. A.4.2.1 X-‐ray Diffraction Samples at Acme were prepared by jaw crushing to 80% passing 2 mm, riffle splitting a 500 g sub-‐sample, and pulverizing by ring and puck mill to 85% passing 0.85 mm. The pulp material was then mounted and exposed to a beam of X-‐rays, which are diffracted by crystal faces. Sample rotation exposes different crystal faces, which diffract the X-‐ray beam; a detector receives these diffractions and produces a diffraction pattern. Crystal structure of a mineral dictates intensity and position of diffraction peaks, creating characteristic patterns for each mineral. Diffraction patterns are compared to a mineralogical database, the closest fit mineral is assigned. As the XRD uses crystallography for identification, it cannot identify anything amorphous. Analyses for XRD were completed by Acme on a Siemens D500 diffractometer, acquiring data with DataScan (version 4) software and analyzing with Jade (version 8.01) software, both of which were created by MDI software of Livermore, California. Quantification was made on the basis of maximum peak height ratios using mineral patterns extracted from the International Powder Diffraction Database (IPDD). There are no regular calibrations of this XRD. Cost of XRD analysis at Acme Laboratories, Vancouver is $ 110 CAN per sample, plus preparation. 129 A.4.2.2 Rietveld Refinement Rietveld refinement is performed on XRD results. A second set of XRD analyses was completed at the University of British Columbia for the purpose of applying RR. Specific preparation is required to reduce preferential orientation; originally put forth by Raudsepp & Pani, 2003, is summarized and slightly modified here. Samples were pulverized, then ground under ethanol in a vibratory micronizing mill for 7 minutes to reduce the grain size to <10 μm. The ethanol-‐sample slurry was spread on to glassware to allow the ethanol to evaporate. Dried sample material was then broken up and scraped off of the glassware using a razor, and gently ground with a mortar and pestle. The ground sample was packed into a holder, ensuring that: particles are randomly oriented, the area of the X-‐ray beam is confined to the sample at all angles of diffraction, and the sample is thick enough that the X-‐ray beam cannot pass through it. Data are collected using step-‐scan X-‐ray powder-‐diffraction, collected over a range of 3 to 80o 2θ in increments of 0.03o 2θ with a counting time of 0.7s/step using CoKα radiation. Diffraction patterns are calculated using known crystal data for each phase; these are then fitted to the sample’s diffraction pattern using least-‐squares refinement. Bruker software EVA was used to identify phases; Topas software was used for RR. EVA can suggest a list of minerals to fit the sample pattern if requested, though the user is strongly advised to use this list with caution. Manually fitting and refining mineral spectra was performed in Topas to minimize the discrepancy between the sample pattern and the mineral spectra. Topas interpolates mineral quantities based on peak intensities and spectra fit, normalizing the crystalline phases to 100%. Over-‐estimation of quantities occur when the number of mineral specimens present is underestimated, therefore it is important to note the discrepancy between the mineral spectra and sample pattern to ensure that all mineral species have been accounted for. Though a pre-‐set mineral list is not required for RR, mineralogical knowledge will significantly improve the quality of results. Because of its XRD platform, RR is incapable of identifying anything amorphous. Analyses were performed on a Bruker D8 Focus XRD with a LynxEye detector in the Earth, Ocean and Atmospheric Sciences department of the University of British Columbia. This XRD is checked weekly by running a known standard of corundum to ensure the 130 correct peak positions and intensities. The cost of having an experienced technician prepare, analyze and interpret results is $ 200 CAN per sample at UBC. A.4.2.3 Mineral Liberation Analysis MLA uses a PTS to create a low noise, high-‐resolution, grey-‐scale, Back Scatter Electron (BSE) image by dividing the PTS into a grid of evenly sized “frames”, imaged one at a time from left to right, bottom to top, across the surface area defined by the user. These frames are stitched together into a single BSE image with mineral phases identified by their boundaries based on greyscale changes (Fandrich, et al., 2007). Each distinct greyscale phase is given a false colour in what is called the phase segmentation process. All distinct mineral phases are defined by an Average Atomic Number (AAN), which is generally unique to each mineral and proportional to BSE brightness. MLA software groups phase areas with like greyscales as potentially identical minerals and allots each area at least one X-‐ray measurement, delivered by an Energy Dispersive Spectrometer (EDS). Each phase area’s X-‐ray spectrum is processed using the MLA Image Processing Tool and Mineral Reference Editor software, which attempts to match the EDS spectrum to the X-‐ray spectral library defined by the user. The user prescribes a ‘looseness of fit’ criterion for the spectrum matching: if the spectrum does not meet the criterion when attempting a match to the X-‐ray spectra in the library database, the program reports the phase as “unknown”. If the fraction of “unknowns” is unacceptable, the spectral library must be revised and the data reprocessed. The spectral library must be carefully compiled to be as representative as possible of pertinent mineral phases; incomplete, or incorrect spectra result in a large “unknown” population, or misidentified minerals. Quantities reported by MLA are based on surface area and normalized to 100%. Spectra can be grouped together to be representative of a mineral. For instance the mineral phase biotite may include a group of five X-‐ray spectra, each representative of a slightly different composition or crystallography of biotite. In this way it is possible to group and ungroup the spectra as the user deems appropriate, for instance breaking down and re-‐grouping the biotite spectra to distinguish Fe-‐rich biotite from Mg-‐rich biotite. Analyses were completed at the Teck Metals Ltd. Applied Research and Technology facility in Trail, BC on an FEI Quanta 600 SEM with a tungsten filament, using dual Bruker 131 XFlash 4010 silicon drift detectors. Image focus, BSE grey level drift and position of the automatic X-‐ray measurements are checked for before the analysis of each sample. The EDS is calibrated against the Cu spectral energy before each sample was analyzed to ensure that all elements are reported correctly. The cost of MLA on a PTS at ALS Chemex is $ 600 CAN per sample, not including PTS preparation or interpretation. A.4.2.4 Quantitative Evaluation of Minerals by Scanning Electron Microscopy QEMSCAN uses a SEM with four light-‐element EDS to produce mineral maps in a method similar to that of the MLA. As in MLA, QEMSCAN requires a detailed, accurate spectral library to ensure quality results. An electron beam steps across the sample at a pixel size set by the user; BSE signals and EDS spectra are collected for each pixel, from which minerals are identified. A false-‐colour mineral map, as in MLA, is processed using the iDiscover software, producing a coloured mineral map, categorized as indicated by the user. If a spectrum does not match anything in the spectral library it is reported as “unknown”; if there is a substantial unknown concentration the spectral library must be reviewed. It should be noted that QEMSCAN is incapable of differentiating between magnetite and hematite; this must be done through optical mineralogy or in hand sample. As with MLA, mineral spectra can be grouped or ungrouped as the user needs. QEMSCAN work was completed at the Advanced Mineralogy Research Center at the Colorado School of Mines. Their system features four Bruker X275HS silicon drift X-‐ray detectors, a Carl Zeiss EVO50 platform, a four-‐quadrant semiconductor diode BSE detector and a secondary electron detector. Calibration was completed each time a sample was run using pure Au, Cu and quartz standards. Analytical information is processed using iMeasure® and iDiscover® software. The cost of QEMSCAN analysis at the Colorado School of Mines is $ 200 USD an hour for commercial use, or about $ 600 USD for a 26 mm x 46 mm PTS, though analytical time varies depending on the amount of data requested (cost does not include PTS preparation). QEMSCAN and MLA methodologies are similar, but have some distinct differences. 1-‐ QEMSCAN uses four detectors, MLA uses two 2-‐ QEMSCAN uses a 25 mm working distance, MLA uses 10 mm 132 3-‐ QEMSCAN takes readings at regular intervals preset by the user covering the entire sample. MLA takes readings based on identified phase boundaries: if no boundary is detected, no reading will be taken. A.4.2.5 Whole Rock Analysis Using the same pulp as the Acme XRD analysis, major oxides were determined by a lithium metaborate/tetraborate fusion, dilute nitric acid digestion with ICP-‐ES finish on a 0.2 g sample. Trace elements were determined by the same fusion and digestion, with an ICP-‐MS analysis, while total carbon and sulphur were determined by Leco analysis. Multi-‐element ultratrace results are from aqua regia digestion of a 30g sample split, with an ICP-‐MS finish. Using the concentrations of major oxides and other elements the MINSQ spreadsheet (Herrmann & Berry, 2002), modal mineralogy was calculated using the least squares method and the Solver analysis tool in Microsoft Excel. User created mineral lists and their compositions were used for quantification. To improve accuracy, EMP calculated mineral compositions of actinolite, hornblende, Mg-‐biotite, K-‐feldspar, plagioclase and albite were used. Compositions were calculated using methodologies presented by Schumacher, 1997 for amphiboles, Deer et al., 2004 for biotite, and Deer et al., 2001 for feldspars. Mineral lists were tweaked from sample to sample to accommodate compositional change. These ten samples were analyzed along with the other 280 samples from the survey, with blanks, standards (matrix matched and non-‐matrix matched), field duplicates and pulp duplicates inserted at a rate of 1 in 20 each. Whole rock analysis at Acme Laboratories, Vancouver starts at about $ 30 CAN per sample, plus preparation. The use of the MINSQ2 spreadsheet is free of charge and available online (http://www.codes.utas.edu.au/5_NewsAndMedia/wally.htm). A.5 Results Table A.2 provides a comparative summary of modal mineralogy for each of the above-‐described methodologies. Not all methodologies can distinguish between different Fe-‐oxides; therefore they have all been grouped in this table as “Fe-‐Oxides”. 133 A.5.1 X-‐Ray Diffraction Acme XRD mineral quantifications were reported in terms of very minor (1-‐5%), minor (5-‐10%), minor to moderate (10-‐25%), moderate (>25%) and moderate to significant (≤50%). Results have been expressed here numerically and adjusted to sum to 100%. XRD identified quartz, albite, biotite, clinochlore, chlorite, hornblende, orthoclase, muscovite and phlogopite. In general these results have higher quartz and albite, but lower biotite and hornblende concentrations than the other methodologies. No plagioclase or epidote was reported. A.5.2 Rietveld Refinement The RR results identified albite (14.7-‐64.7 %), actinolite (1.7-‐5.8 %), biotite (2.2-‐3.8 %), calcite (0.5-‐3.7 %), clinochlore (0.8-‐7.1 %), chlorite minerals (3.6 %), epidote (0.4-‐8.7 %), hematite (0.3 %), K-‐feldspar (0.8-‐59.5 %), magnetite (0.6-‐1.9 %), muscovite (1.8-‐4.1 %), plagioclase (56.6-‐68.2 %), quartz (12.3-‐28.3 %), sphene (0.7-‐1.4 %), and stilbite (1.5%). Modal mineralogy is comparable to MLA and QEMSCAN results, though albite and quartz are generally lower, and plagioclase higher than in other methodologies. Actinolite was identified more frequently and in higher concentrations by RR than by any other methodology. Only RR identified stilbite. A bias was potentially introduced during preparation: biotite would not pulverize as finely as the other minerals and tended to float. As a result, the biotite would not pass through the sieve and would have to be re-‐pulverized. It is possible that biotite passed through disproportionately. 134 Table A.2: Summary of modal mineralogy results: petrographic observations (PTS), X-‐Ray Diffraction (XRD), Rietveld Refinement (RR), Mineral Liberation Analysis (MLA), Quantitative Evaluation of Minerals by Scanning Electron Microscopy (QS) and MINSQ (WR). Protolith (P) and alteration (A) minerals are indicated. Elements written in italics have EMP results used in the MINSQ calculations. Mineral P A PTS XRD RR MLA QS WR PTS XRD RR MLA QS WR PTS XRD RR MLA QS WRQuartz 26.00 40.00 23.61 45.21 24.10 22.78 26.00 45.00 26.78 48.33 30.80 25.51 14.00 23.00 14.10 18.92 12.80 11.65Plagioclase 46.00 56.56 40.30 47.50 15.98 46.00 59.30 34.76 44.90 52.60 61.00 68.21 67.12 65.10 54.57 PotassicHornblende 2.00 0.12 1.60 1.05 1.00 0.20 10.00 8.00 4.10 2.61 PropyliticMg-Biotite 1.48 4.00 0.27 0.90 1.37 Phyllic + PropyliticActinolite 3.57 1.62 0.42 5.82 5.24 0.83 PhyllicAlbite 5.00 41.00 1.66 7.40 37.38 10.00 45.00 7.12 14.00 10.31 3.00 46.00 1.97 6.00 12.97 Protolith MineralFe-Biotite 3.00 0.03 0.10 0.01 4.00 23.00 2.23 2.81 1.50Calcite 0.01 0.17 0.16Clinochlore
UBC Theses and Dissertations
Surface lithogeochemistry of the Relincho porphyry copper-molybdenum deposit, Atacama region, Chile :… Greenlaw, Lauren 2014
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
- 33426-ubc_2014_november_greenlaw_lauren_appendices.pdf [ 437.48MB ]
- JSON: 33426-1.0107673.json
- JSON-LD: 33426-1.0107673-ld.json
- RDF/XML (Pretty): 33426-1.0107673-rdf.xml
- RDF/JSON: 33426-1.0107673-rdf.json
- Turtle: 33426-1.0107673-turtle.txt
- N-Triples: 33426-1.0107673-rdf-ntriples.txt
- Original Record: 33426-1.0107673-source.json
- Full Text