Microanalytical techniques

Element quantifi cation in stage-one and stage-two has been through bulk analytical procedures (e.g., FP-XRF and XRF). However, in-situ examination of sulphide trace element distribution and sulphide alteration products would allow for a better understanding of each sulphide-bearing mesotextural group. Th rough understanding these micro-scale processes, better waste rock management/ rehabilitation strategies can be developed, as implied in Jamieson (2011). In-situ element analysis can be undertaken

through several microanalytical techniques which are commonly used in ARD studies (Table 2.15). However, integration of these techniques into ARD predictive protocols is not presented in the literature. Th is is likely due to the absence of clear guidelines as to when and how to apply these techniques in addition to their relatively high analytical costs. Stage-three addresses this, with the following section selecting the most appropriate techniques when considering time and cost, and provides an overview of standard methodologies.

Table 2.16 Microanalytical techniques used to determine mineral composition in mine waste material, dissolution and other weathering textures, and the residence sites of trace elements (compiled from Diehl et al., 2007 and Jamieson, 2011).

Analytical technique Application to mine waste

mineralogy

Example studies GMT approach

application?

Scanning electron microscopy (SEM)

Mineral species, mineral textures, particle size, cleavage, grain bounda- ries, surface weathering, deformation structures, microfaults, veins, fractures, semi-quantitative EDS data

Hudson-Edwards et al. (1999), Walker et al. (2005), Weisner and Weber (2010)

Yes

Electron probe microanalysis (EPMA)

Exact residence of minor and trace elements, spatial distribution of minor and trace metals, semi quantitative and quantitative data

Hudson-Edwards and Edwards (2005), Moncur et al. (2009), Walker et al. (2009), Corriveau et al. (2011)

Yes

Transmission electron micros- copy (TEM), scanning trans- mission electron microscopy (STEM)

High-resolution imaging, may include chemical information and electron diff raction

Petrunic et al. (2009) No

Micro-XRF diff raction using conventional or synchrotron sources (XAS)

Grain scale mineral identifi cation based on crystal structure; application to poorly crystalline materials

Walker et al. (2009), DeSisto et al. (2011)

No

Synchrotron-based X-Ray fl uorescence (micro-XRF)

Mineral characterisation based on oxi- dation state and short-range structure

Foster et al. (1998), Walker et al. (2009), Pérez López et al. (2011), Carbone et al. (2011)

No

Conventional micro-XRF Minor and trace element semi-quanti- tative data

Bernaus et al. (2006) Yes

Laser ablation ICPMS Minor and trace element quantitative data

Al et al. (2007), Ohlander et al. (2007), Savage et al. (2008)

Yes

Raman spectroscopy Mineral identifi cation Das and Hendry (2011) Yes

Micro- particle induced X- Ray emission (micro-PIXE)

Minor and trace element quantitative data

Cabri et al. (1993), Cabri and Campbell (1998), Jamieson et al., (2005)

No

Techniques selected for review in the GMT approach include LA-ICP-MS and μXRF for element mapping; EMPA and LA-ICPMS for element quantifi cation; and MLA-SEM for examining microtexture. Synchrotron based technologies were avoided by the GMT approach as these specialist techniques cannot be routinely used, and essentially will not provide a signifi cant amount of new data which will impact upon sample classifi cation or indeed the management of a particular mesotextural group.

2.6.2.1 SEM-EDS

Scanning electron microscopy (SEM) is the fi rst microanalytical technique recommended. Previously, application of SEM-EDS was suggested exclusively for samples with sulphide composition abnormalities (Mills, 2011). However, application should not be limited. For example, sulphide minerals often contain micro-inclusions which are not readily identifi ed in optical microscopy studies (e.g., Maslenikov et al. 2009; Th omas et al., 2011). Presence of inclusions causes strain to the crystal structure, diminishing the sulphide’s resistance to oxidation (Jambor, 1994; Kwong, 1995; Plumlee, 1999). However, through SEM studies, micro-inclusions are better identifi ed and the relative rate of weathering can be assessed. A broad number of samples from EAF and AF groups should be analysed, particularly those representative of textural variations (i.e., diff erent mineral-associations, sulphide morphologies, sizes, and fracturing). Observations following those listed in Table 2.16 should be made. In this research, a FEI Quanta 600 environmental scanning electron microscope (ESEM; CSL, UTAS) was used, with polished thin sections and laser mounts carbon coated prior to analysis.

2.6.2.2 EPMA

EPMA is the next recommended technique, and should be used to obtain compositional data from distinct oxidation products identifi ed in SEM and optical microscopy studies (i.e., on fewer samples than were analysed by SEM). Identifi cation of these products will enhance the understanding of trace element mobility, as often these act as temporary stores (Lin and Herbert Jr. 1997; Hudson-Edwards and Edwards, 2005; Lottermoser, 2010). Th ese products often form rims and fractures on primary sulphides (e.g., Blowes and Jambor, 1990; Jambor, 2003; Moncur et al., 2009), therefore a small beam size (micron-scale) is required. Quantitative analysis can be hindered as a result of the porous nature of these secondary products (e.g., brown iron oxyhydroxide/ferrihydrite) and admixed impurities, thus results obtained may only be semi-quantitative (Haff ert et al., 2010). Analysis should be undertaken on samples representative of each primary sulphide, as these may be required for use an internal standard as part of LA-ICP-MS analysis (see next section). In this research, a Cameca SX100 electron microprobe was used (CSL, UTAS), with polished laser mounts carbon coated prior to analysis.

2.6.2.3 LA-ICP-MS

Following SEM-EDS and EPMA, LA-ICP-MS analysis is recommended for a well chosen sub-set of samples representative of each sulphide mineral (and textural type) identifi ed per acid forming mesotextural group. Th e main objectives are to determine trace element distribution in primary sulphides in order to understand potential eff ects on oxidation, and if possible, to deduce secondary mineral trace element contents as lower detection limits are achieved by LA-ICP-MS than EPMA (Koenig, 2008). Applications of spot and line analyses have been demonstrated in previous ARD studies (Ohlander et al., 2007; Savage et al., 2008). Element mapping can be undertaken using EPMA (e.g., Lin and Herbert Jr., 1997; Hudson-Edwards and Edwards, 2005; Diehl et al., 2007), however, this is highly time consuming, and better resolution maps can now be collected using LA-ICP-MS. Examples of LA-ICP-MS element mapping in ARD studies is yet to be published, therefore, the methods used here followed those outlined in Large et al. (2009), Th omas et al. (2011) and Danyushevsky et al. (2011). In this research, a New Wave 213nm solid state laser microprobe coupled either to an Agilent 4500 or an Agilent 7700 quadrupole

ICP-MS was used. Quantifi cation of LA-ICP-MS analyses requires an internal standard as defi ned by Longerich et al. (1996). Danyushevsky et al. (2011) recommended that for routine cost-eff ective analysis, the concentration of the internal standard should be easily measurable by other routine micro-analytical techniques such as EPMA, or estimated from stoichiometry. Iron is often chosen as the internal standard for many sulphides (Danyushevsky et al., 2011). Galena and sphalerite were analysed in this research, so stoichiometric Pb and Zn measured by EPMA were used, respectively. During analyses, the calibration standard STDGL2b2 (Danyushevsky et al., 2011) was used, which contains a wide range of chalcophile, siderophile and lithophile elements.

LA-ICP-MS analyses are performed in an atmosphere of pure He. Sulphide element mapping is undertaking by ablating a set of parallel lines arranged in a grid over the sample so the space between lines matched that of the beam (Th omas et al., 2011). In this research, 15, 25 or 30 μm beam sizes were used depending on the size of the target grain. Th e beam is rastered over the lines at a speed matching the beam size with a nominal 10 Hz repetition rate, thus, every position in the sample is ablated 10 times contributing to fi ve consecutive pixels in the fi nal image, and ablated to a depth of c.5 μm (Th omas et al., 2011). Background levels and drift are measured on the STDGL2b2 standard before and after every image, and if analysis is calculated to exceed 1 hour, then additional standard analyses are performed in the middle of the run. Collection of standard data allows for element map quantifi cation. Image processing involves drift correction, application of a median fi lter to remove artefacts generated during processing, subtraction of background from fi ltered counts, and replacement of fi ltered counts less than background with standard deviation values for that element (Th omas et al., 2011). Images are fi nally produced for each element, with concentration indicated by a logarithmic colour scale (Th omas et al., 2011). Whilst examples of LA-ICP-MS mapping analyses are presented in Large et al. (2009) and Th omas et al. (2011) for pyrite and pyrrhotite, few examples exist for other sulphide minerals such as arsenopyrite, galena and sphalerite.

2.6.2.4 Micro-XRF

A potential limitation of LA-ICP-MS is the cost per analysis (>$100 per hour; CODES, UTAS); therefore, a lower cost element mapping technique was sought for inclusion in the GMT approach. Conventional μXRF was selected for use as it is a non-destructive, highly accurate technique. Adams et al. (1998) provide a concise overview of μXRF, and attributed its development to the production of relatively simple and cost eff ective devices for obtaining a small dimension X-Ray beam, i.e., capillary optics. Recent examples of μXRF application are given in Croudace et al. (2006); Rothwell et al. (2006), Katsuta et al. (2007); Coralay and Kadioglu (2008) and Genna et al. (2011), with few studies demonstrated applications in mine waste characterisation (e.g., Bernaus et al., 2006; Hayes et al., 2009). In this research, a Horiba XGT 7000 μXRF was used (CSL, UTAS). Th is machine uses X-Ray guide tubes (either 100 μm or 10 μm) from which a 10μm high intensity X-Ray beam is irradiated onto the sample. Measurement of elements Na (z=11) to U (z=92) is possible. Th e analysis probe is set in vacuum, making it possible to analyse the sample at normal atmospheric pressure. A sample on the XY scanning stage can be visually observed via a CCD camera from the same axis as the X-Ray beam. High speed measurements are possible up to 50 times faster than conventional equipment (i.e., EPMA), providing

much greater usability. Fluorescence and transmission X-Rays from the sample are measured during analysis, with this data used to create element maps. Samples require minimal preparation prior to use (i.e., no coatings as with SEM and EPMA are required), and the sample stage can accommodate for a 10cm x 10cm maximum area of analysis. Th is implies that intact drill-core samples can be analysed at microscale resolution. As no published methods for analysis using this instrument were available, a period of method development was undertaken as part of this research (Appendix 2.2). Several types of sample and various preparation methods were trialled (including optical thin sections, whole and half drill core samples and resin mounted samples). Examples of μXRF analyses performed on laser mount samples only are presented for comparison with LA-ICP-MS data in this research (Chapter 3).

2.6.3 MLA

MLA represents a unique automated method of combining BSE image analysis, X-ray mineral identifi cation and advanced imaging and pattern recognition analysis (Gu, 2003; Fandrich et al., 2007). Applications of MLA in other mining-related disciplines (i.e., applied mineralogy, metallurgical processing) are well established, with examples presented in Bruckard et al. (2010); Chapman et al. (2011); Hunt et al. (2011) and Rizmanoski (2011). Currently, there are no published examples of MLA application in predictive ARD studies. Th erefore, MLA is recommended as a stage-three technique, and should be performed on at least one representative sample from each EAF and AF group. Polished laser mounts (as prepared for SEM, EPMA and LA-ICP-MS) or MLA tiles should be analysed. Th e MLA system used in this research was the FEI Quanta 600 SEM equipped with 2 EDAX ultra thin window Si(Li) energy dispersive X-ray (EDS) detectors (CSL, UTAS).

Th ere are eight basic MLA measurement modes, which vary from a purely BSE-based technique to an almost exclusively X-Ray analysis point counting technique (XMOD; Gu, 2003; Fandrich et al., 2007). Th e GMT approach recommends the use of sparse phase liberation_Lite (SPL_Lite), extended back scattered electron (XBSE), grain-based X-Ray mapping (GXMAP) and X-ray modal (XMOD) analyses. Prior to analysis, a mineral standard library must be collated for each individual sample suite (i.e., per site) to allow for accurate mineral identifi cation. High quality X-Ray spectra are collected for each mineral identifi ed in the sample suite (Fandrich et al., 2007). During the analysis, spectra for unknown minerals are also collected if specifi ed. After analyses these particles can be selected and identifi ed using EDS, and retrospectively incorporated into the mineral library. SPL_Lite analysis should be performed as standard on every selected EAF and AF sample. Th is measurement mode works by searching images for particles with BSE grey levels greater than that of the standard, and subsequently maps them (Fandrich et al., 2007). In this research a nickel standard was used. Applications of XBSE and GXMAP analyses to produce mineral maps for whole samples were explored. XBSE operates similar to SPL_Lite, and GXMAP uses X-Ray mapping on phases that cannot be segmented by BSE grey levels alone (Fandrich et al., 2007). Th e GXMAP technique was of particular use to diff erentiate between pyrite and magnetite for Ernest Henry samples (Chapter 5). XMOD is based on a point counting method whereby mineral identifi cation is determined by one X-Ray analysis at each counting point (as defi ned

by the operator; Fandrich et al., 2007). X-Ray spectra are saved for off -line classifi cation, with modal mineralogy information (i.e., percentages) calculated. Th ese data can be used for comparison with QXRD values gathered at stage-one, or to supplement modal mineralogy data per mesotextural group. Off -line data processing is typically performed using MLA Image View software (Gu, 2005), with particles and images classifi ed based on mineralogy. Classifi ed particles and images can be analysed further using Texture Viewer (Nguyen, 2009).