Mineral exploration is increasingly taking advantage of real time techniques that dramatically reduce the costs and time taken to obtain results compared to traditional analytical methods. Portable X-ray fluorescence (pXRF) is now a well-established technique that is used to acquire lithogeochemical data. To date, however, benchtop scanning electron microscopes, equipped with energy dispersive systems (bSEM-EDS) have received little attention as a possible mineral exploration tool.
This study examines the utility of combining pXRF and bSEM-EDS to characterize the igneous stratigraphy and its relationship to Cu-Pd mineralization in a drill hole at the Four Dams occurrence, located within the Eastern Gabbro assemblage of the Coldwell Alkaline Complex, Canada.
The first part of this study compares field portable and laboratory techniques. Seventy-two powdered samples analysed by pXRF are compared with traditional major elements analysed by inductively coupled atomic emission spectroscopy (ICP-AES) and trace elements by inductively coupled plasma spectrometry (ICP-MS), and the compositions of 128 olivine, clinopyroxene and plagioclase grains analysed by bSEM-EDS are compared with traditional electron microprobe data. Our results show that each portable technique yields results similar to their lab-based counterparts within the analytical capabilities and precisions of the respective instruments.
The second part presents a case study for the application of pXRF and bSEM-EDS to resolve questions related to igneous stratigraphy as an aid to mineral exploration in a complicated geological setting. A major problem for Cu-Pd exploration in the Coldwell Complex of NW Ontario is that the oxide-rich units that host Cu-Pd mineralization in the Marathon Series are petrographically similar to the barren oxide-rich units in the Layered Series. However, the mineralized units are geochemically distinctive. Our results show that the mineralized Marathon Series can be distinguished from the barren Layered Series, including oxide-rich units of both, by combinations of P2O5, Ba, Zr and V/Ti values, determined by pXRF, combined with plagioclase, olivine or clinopyroxene compositions measured by bSEM-EDS. The combination of pXRF and bSEM-EDS thus shows considerable promise as an exploration technique.
Supplementary material: Comparisons of pXRF analyses with ICP-AES, ICP-MS and IR analysis are available at http://doi.org/10.6084/m9.figshare.c.3291518
Diamond drilling is arguably one of the most important techniques in mineral exploration, and the geochemical analysis of drill core is an essential component of any diamond drilling program. Conventional laboratory geochemical analyses have high accuracy and precision. However, time delays in obtaining results mean that lab-based measurements are not well suited for making rapid decisions in the field. A number of field-portable analytical techniques are designed to collect data with minimal time delay and at low cost. Thus, the mineral exploration industry is increasingly using field portable instruments in modern mineral-exploration campaigns. Portable X-ray fluorescence analysers (pXRF) and benchtop scanning electron microscopes with energy dispersive spectrometers (bSEM-EDS) are two real-time acquisition instruments that can rapidly collect lithogeochemical and mineral chemical data in the field.
The first portable XRF instrument was produced in 1991, following the introduction of cooled Si-PIN diodes and improvements to various components (Nicola et al. 2011). The first media analysed were simple, with few interferences, e.g. archaeological ceramics and alloys. Since 1991, significant improvements such as the use of silicon drift detectors (SDD) and miniaturized X-ray tubes that can run at higher power, have made the pXRF technique increasingly applicable to the analysis of geological materials. The advantages of pXRF over lab XRF include the smaller size (portability), lower cost, and faster data collection, making it suitable for on-site lithogeochemical analysis. Applications of pXRF initially focused on environmental studies, with few related to mineral exploration. The pioneering work of Potts et al. (1995) provided a preliminary assessment of pXRF in the analysis of silicate rocks. Since then, there have been numerous studies that utilize pXRF in exploration geochemistry, including evaluations of pXRF performance in analyzing geological materials (e.g. Makinen et al. 2006; Haffert & Craw 2009; Morris 2009; Hall et al. 2012, 2014) and applications of pXRF to volcanogenic massive sulphide (Ross et al. 2014) and REE deposits (Simandl et al. 2014). Interactive applications of pXRF include using ‘fit for purpose’ data for pathfinder elements in gold exploration (Arne et al. 2014) and the development of a workflow in the mining setting that highlights the importance of QA/QC control (Fisher et al. 2014). There are many other notable publications on the applications of pXRF to mineral exploration; however, it is beyond the scope of this paper to provide a complete review of the subject.
Unlike pXRF lithogeochemical analyses, there are few applications of field-based real-time mineral-chemical analysis. However, the development of benchtop scanning electron microscopes, equipped with energy dispersive spectrometers (bSEM-EDS), offers the possibility of collecting mineral composition data directly in the field, in contrast with conventional mineral-chemical analysis from lab-based analytical instruments, e.g. scanning electron microscope with energy dispersive spectrometers and electron microprobe. The bSEM-EDS instrument can easily sit on a table in a field setting and samples do not require any preparation, such as polishing or carbon coating, prior to insertion into the instrument. This enables real-time imaging and EDS analysis at the sub-micron scale. Benchtop SEM instruments have been available for several years but, to our knowledge, this technique has not been applied to mineral exploration. The current study combines pXRF with bSEM-EDS analysis as tools to improve decision-making in the field. In this study, these real-time analytical techniques are applied to Cu-Pd exploration at the Four Dams occurrence in the Eastern Gabbro of the Coldwell Alkaline Complex, northwestern Ontario, and the results are compared to those obtained by conventional laboratory analyses to establish the reliability of these techniques.
The Coldwell Alkaline Complex is the largest alkaline intrusive complex in North America (25 km diameter, 580 km2 area) and is part of the Midcontinent Rift (Walker et al. 1993). It intruded the Archean Schreiber-White River greenstone belt at 1108±1 Ma (Heaman & Machado 1992) during the early stages of the Midcontinent Rift (Miller & Nicholson 2013). The Coldwell Alkaline Complex consists of a margin of ultramafic, gabbroic and meta-volcanic rocks that make up the Eastern Gabbro Suite (Good et al. 2015) and three intrusive centres composed predominantly of syenite and nepheline syenite (Walker et al. 1993) (Fig.1).
The Eastern Gabbro suite occurs along the outer margin of the Coldwell Alkaline Complex (Fig. 1). Good et al. (2015) identified nine major lithologies that were further grouped into three major distinctive magmatic series that were defined, from oldest to youngest, as meta-basalt, the Layered Series and the Marathon Series.
Significance of Igneous stratigraphy to Cu-Pd mineralization in Eastern Gabbro
Meta-basalt (fine-grained gabbro in Good et al. 2015) occurs at or near the base of the Eastern Gabbro and comprises approximately one fifth to one third of the volume of the Eastern Gabbro. It consists of pyroxene-hornfels-grade rocks that were intruded by the Layered and Marathon Series. The Layered Series is the next youngest unit and comprises the bulk of the Eastern Gabbro. It consists of massive to modally layered olivine gabbro with lesser amounts of weakly layered oxide augite melatroctolite, and less commonly gabbroic anorthosite. The Marathon Series is the youngest unit. It consists of numerous small intrusions composed predominantly of ophitic gabbro, apatite-bearing olivine clinopyroxenite, apatitic clinopyroxenite, and oxide melatroctolite (Good et al. 2015). The relative ages for these units were confirmed in the field by cross cutting relationships, particularly within the numerous units of igneous breccia.
Most of the copper and all of the known PGE mineralization within the Eastern Gabbro suite are hosted by the Marathon Series. In particular, at the Marathon deposit, the most important sub unit of the Marathon Series is the Two Duck Lake gabbro, an ophitic gabbro and pegmatitic unit. Because of the significance of the Marathon Series as hosts for Cu-Pd mineralization, it is critical that the entire suite of Marathon Series intrusions can be distinguished from other Eastern Gabbro rocks for mineral exploration, particularly if it is possible to differentiate these units in the field. Note that although some disseminated chalcopyrite and pyrrhotite mineralization does occur in the Layered Series, it is associated with albite-actinolite alteration and contains only trace PGE (<0.005 ppm).
The Eastern Gabbro in the Four Dams area is located c. 3 km NW of the Marathon Cu-Pd deposit (Fig. 1) and contains units of both the Layered Series and Marathon Series. The Four Dams area is thus an ideal setting to establish whether pXRF and bSEM-EDS can be combined to establish an igneous stratigraphy that could be used to help guide future exploration.
Drill Hole FD-13–34
Drill hole FD-13–34 (Fig. 2) was selected because it intersects a complete section of igneous stratigraphy at the Four Dams occurrence, including units of both Layered Series and Marathon Series. The hole was drilled at an azimuth of 32.9° and dip of 69.6° for 375 m. The Layered Series at Four Dams consists of layered olivine gabbro and oxide augite melatroctolite and the Marathon Series consists of apatite-bearing clinopyroxenite, oxide melatroctolite, and a minor amount of ophitic gabbro, similar to the Two Duck Lake gabbro at the Marathon Deposit.
A 70 m thick sequence of oxide-rich units consisting of, from top to bottom, oxide augite melatroctolite and oxide melatroctolite, which separates the Layered Series from the Marathon Series. The oxide melatroctolite exhibits a sharp contact with the overlying Layered Series oxide augite melatroctolite. This contact is important because it forms a 350 m long stratigraphic marker horizon that facilitates interpretation of the local geology. Both oxide units are petrographically similar, but essentially belong to two different igneous series, i.e. the oxide augite melatroctolite is within the PGE barren Layered Series whereas the oxide melatroctolite is in the Marathon Series and potentially hosts PGE mineralisation. Being able to differentiate both oxide units is important, not only because the oxide melatroctolite is potentially mineralized, but it can help define the contact between the barren Layered Series and the PGE mineralized Marathon Series.
Two types of igneous breccia occur within drill hole FD-13–34 and are representative of breccia units found throughout the Four Dams area. The first type occurs between 47 and 60 m down the hole and consists of xenoliths of meta basalt rocks cut by Marathon Series intrusions. The second type occurs between 297 and 316 m down the hole and consists of Layered Series units cut by Marathon Series intrusions.
Two types of Cu-Pd mineralization occur in drill hole FD-13–34. The first type occurs within the Layered Series between 13 and 65 m down the hole and contains up to 0.41% Cu and trace (<0.005 ppm) Pd. The second type occurs primarily within the apatitic clinopyroxenite unit between 270 – 278 m and 290 – 336 m down the hole and contains up to 0.72% Cu and 0.56 ppm Pd.
The olivine gabbro of the layered series at Four Dams is medium- to coarse-grained (1 – 3 mm) with modal layering defined by a gradational variation in the abundance of plagioclase. The gabbro has an intergranular texture (Fig. 3A) and is mainly composed of, in decreasing order of abundance, euhedral plagioclase, subhedral clinopyroxene, subhedral olivine, magnetite, and less than 6 modal % fine-grained apatite that is typically enclosed in, or interstitial to, olivine and plagioclase.
The oxide augite melatroctolite is medium- to coarse-grained and has similar textures to the layered olivine gabbro. It is distinguished by the presence of abundant (10 – 30 modal %) disseminated magnetite (Fig. 3B). The mineral composition of this unit comprises, in approximate decreasing order of abundance, olivine, plagioclase, clinopyroxene, magnetite, and apatite. Olivine and clinopyroxene are euhedral, and occur between subhedral to anhedral plagioclase clusters. The oxide augite melatroctolite contains disseminated sulphide minerals (2 – 3 modal %) that are dominated by fine-grained pyrrhotite with trace chalcopyrite.
Where the Marathon Series crosscuts Layered Series rocks in drill hole FD-13–34, intrusive contacts are sharp and lack chilled margins. The oxide melatroctolite and the apatitic (olivine) clinopyroxenite (see below) are interpreted to be part of the Marathon Series because they also commonly occur as thin lenses within the main body of the Two Duck Lake gabbro at the Marathon deposit and typically contain disseminated chalcopyrite and pyrrhotite with elevated PGE concentrations (Good et al. 2015).
The oxide melatroctolite is medium-grained and equigranular and consists of subhedral magnetite (40 – 60 modal %), olivine and clinopyroxene, interstitial plagioclase, and from 2 to 30 modal % euhedral apatite. Between 247.8 and 263.4 m, the oxide melatroctolite becomes coarser grained (2 – 3 mm) and is characterized by coarse-grained plagioclase laths that vary from 20 to 50 modal % (Fig. 3C).
The apatitic (olivine) clinopyroxenite consists mainly of medium- to coarse- grained subhedral olivine, clinopyroxene, and magnetite with interstitial plagioclase and euhedral apatite (Fig. 3D), and commonly contains disseminated pyrrhotite and chalcopyrite. Some chalcopyrite occurs intergrown with plagioclase that replaces early plagioclase (Fig. 3E and F). Chemical (EDS) analyses (discussed later) shows that the late plagioclase is more calcic than the early plagioclase.
The ophitic gabbro is coarse-grained (2 – 4 mm) and is distinguished by an ophitic to subophitic texture. It is composed of, in approximately decreasing order of abundance, subhedral plagioclase, euhedral to subhedral olivine, anhedral interstitial clinopyroxene, magnetite and apatite. Although this unit constitutes a major part of the Marathon Series and is the main host to the Cu-PGE mineralization at the Marathon deposit (Two Duck Lake gabbro) it occurs as thin discontinuous intrusions in the Four Dams area.
Seventy-two powdered samples were collected by grinding 1 m-long channels along the drill core c. 2 mm wide and 1 cm deep using a Thermo Scientific portable grinder. Sample locations for powdered samples are given in Figure 2. A typical sample required 2 – 5 min to collect and produced c. 10 – 200 µm of powder. A few pXRF analyses were made on powders that were re-ground, but this had little effect on the results. Samples were taken at intervals of c. 4.5 m along the length of the drill hole. Cross-contamination was minimized by washing the collection vial, cleaning off the grinder blade, pre-contaminating the vial by grinding some of the sample, shaking the vial to coat the vial with the sample dust and dumping out the vial, and then collecting the powdered sample to be analysed. All powders were then loaded into PREM-4331 XRF sample cups capped with 4-μm PREM-F2540 XRF sample cup films for the pXRF analysis.
The potential for contamination of Fe, Cr and Ni by the grinder blade was tested by sampling a piece of pure quartz. The pXRF analyses are presented in Table 1 and show that grinder blade-related contamination for most elements except Ni is insignificant, with less than 0.1 wt. % for most major elements and lower than the detection limit for trace elements. Approximately 100 ppm of Ni is added to each sample during grinding.
Twenty-three polished thin section blocks that were representative of the different gabbro units and styles of mineralization (e.g. chalcopyrite-rich, bornite-rich and pyrrhotite-rich) were cut from drill core FD-13–34 (Fig. 2) for petrographic and bSEM-EDS analysis.
A Niton XL3t+ GOLDD+ pXRF analyser equipped with a silicon drift detector (SDD), and a high energy, 50 kV X-ray Ag anode tube was used in this study. All pXRF measurements were carried out on powders after calibrating the pXRF analyser using a suite of matrix-matched standards, including two U.S. Geological Survey reference materials, BHVO-2 basalt and W-2a diabase, and seven Marathon gabbro internal standards. This suite of standards covers a wide range of concentrations for elements measured in the current work: SiO2, Al2O3, CaO, Fe2O3 (as total Fe), TiO2, K2O, P2O5, MgO, Ba, V, Zr, Sr, Cu and S. Major elements were determined using the instrument's ‘mining mode’, with two beams, whereas minor and trace elements were determined using the ‘soil mode’, with three beams; beam time for both modes was 60 s, as suggested by Hall et al. (2014) and Fisher et al. (2014). The QA/QC control throughout all analyses was carried out by analyzing two standards after each batch of unknown samples (20 samples typically). A blank standard (98.2 wt. % SiO2) was analysed at the beginning and the end of each batch measurement to monitor cross-contamination from the test stand and check for machine drift. Typically, a new calibration was conducted after every two batches of unknown samples. The intent was to acquire the highest quality of results and less-frequent calibration may have yielded a similar result.
Conventional whole-rock analysis
Whole-rock chemical analyses were carried out by the ALS Mineral Division in Vancouver on powders that were initially analysed by pXRF. After further pulverization, major and trace elements were determined by ICP-AES and ICP-MS, after lithium borate fusion on a minimum 2 g of pulp samples. Total S was determined using a Leco combustion furnace on 0.01 – 0.1 g of sample, in which the sample is heated to roughly 1350°C in an induction furnace. SO2 is produced through a reaction with oxygen and is measured by an infra-red detector. Ferrous Fe was determined by H2SO4-HF acid digestion and titration on 1 g of sample. Quality-control was achieved by analysing a wide array of standards, blanks, and duplicates after each batch of samples, in accordance with ALS geochemical quality-control procedures. Some additional data from Stillwater Canada Inc. is also used to compare S and Cu values. These were also analysed by ALS; S by Leco, as above, and Cu by aqua regia dissolution and ICP-AES analysis.
Benchtop SEM-EDS analysis
The major constituents of plagioclase, clinopyroxene and olivine were determined using a JEOL JMC-600 NeoScope SEM equipped with a JEOL JED-2300 energy dispersive X-ray analyser at Western University. Analyses were conducted using 15 kV accelerating voltage, a high probe current, a working distance of 19 mm, and 30 – 40 μm beam size with standardless ZAF corrections. Unpolished thin section blocks were analysed under low vacuum mode and carbon-coated polished thin sections were analysed under high vacuum mode.
Electron microprobe analysis
Electron microprobe analyses were carried out using a JEOL JXA-8530F field emission probe at Western University. Microprobe analyses were conducted on the same carbon-coated polished thin sections that were analysed by bSEM-EDS. Analyses were conducted using a 15 kV, 20 nA beam. The beam was focused to a 1 μm spot for analysing olivine and clinopyroxene, and to a 5 μm spot for analysing plagioclase. The peak counting time was 30 s for all elements and background counts were a total of 30 s. A variety of synthetic and natural standards were used to calibrate different elements during probe analyses.
Comparison of Field Portable Results to Lab-Based Analayses
All major and trace element concentrations determined by pXRF and by ICP-AES, ICP-MS and IR are provided in Supplementary Material. The down-hole profile of pXRF data relative to lab-based data (Fig. 4) shows that the pXRF data broadly replicate the lithogeochemical patterns defined by the conventional data. The correlation coefficients have R2 values of >0.9 for all elements, except for MgO (Table 2). Mg is the lightest element and thus should have the poorest precision and accuracy by pXRF. However, the MgO R2 value is 0.77, which is still a reasonable correlation between these two methods. Therefore, pXRF is able to provide analyses comparable to lab-based analyses for the elements included in the current study.
To further illustrate the accuracy of the pXRF analyses, the differences of pXRF results relative to lab-based results (RD) are shown on box-and-whisker diagrams (Fig. 5). pXRF results for major elements, excluding Al2O3, generally have an accuracy in the range of ±20% RD (in the 25 – 75 percentile range); accuracy for pXRF Al2O3 is between ±10 and ±35% RD (in the 25 – 75 percentile range), which is inferior to that for other major elements. For minor and trace elements, Ba, Sr and Zr have better accuracy than other elements, with the majority of RD values within ±20% (in the 25 – 75 percentile range). By contrast, the accuracy for pXRF S, V, and Cu is inferior, and most of their RD values are outside the ±20% range (in the 25 – 75 percentile range). Since there are excellent correlations between pXRF and ICP-AES/ICP-MS/IR Al2O3, V, Cu, and S analyses, recalibration can be applied to improve data accuracy. Figure 6 shows the improvement of accuracy for pXRF Cu and S analyses after such recalibration. The pXRF data accuracy and precision reported here are similar to those evaluated by previous studies of Hall et al. (2014), Piercey & Devine (2014) and Ross et al. (2014).
It is not possible to carry out a full calibration and ZAF correction with matrix-matched materials using the bSEM-EDS. However, mineral ratios rather than absolute concentrations are commonly used to examine mineral chemical variations in suites of magmatic rocks. In this study the Mg# (molar MgO/(MgO+FeO)×100) of olivine and clinopyroxene, and the An% (molar Ca/(Ca+Na)×100) of plagioclase were determined, because these are useful indices for assessing magma evolution and mechanisms for forming magmatic PGE mineralization (Barnes 2004; Mungall & Naldrett 2008). The highest quality bSEM-EDS data will be obtained from carbon-coated polished thin sections, but unpolished cut blocks were also analyzed to establish field portability. Consequently, Mg# of olivine, Mg# of clinopyroxene, and An% of plagioclase determined by bSEM-EDS for carbon-coated polished thin sections and unpolished blocks are compared to electron microprobe data (Table 3 and Fig. 7).
The precision of electron microprobe analyses was determined by analyzing the same spot 10 times: results are ±0.07 for Mg# of olivine, ±0.30 for Mg# of clinopyroxene, and ±0.15 for plagioclase An% (1σ standard deviation). These standard deviations (machine error) are well below the sample variability that were determined through wavelength-dispersive spectrometer (WDS) analyses conducted on 3 or 4 grains of each mineral type in each sample (Table 3). Machine error for bSEM-EDS analyses of carbon coated polished thin sections are ±0.51 for Mg# of olivine, ±1.0 for Mg# of clinopyroxene, and ±0.63 for An% of plagioclase (1σ standard deviation). These values are higher than those for the electron microprobe. The sum of the standard deviation of natural variation of all three minerals (determined from 3 or 4 grains per sample) and the machine error for the electron microprobe analyses is < ±1 to ±3, compared to ±1 to ±4 for bSEM-EDS analysis for carbon coated polished thin sections (Table 3 and Fig. 7).
For the bSEM-EDS analysis of unpolished blocks it is necessary to use the low vacuum mode to reduce charge build-up from electrons accumulated on non-carbon coated surfaces, but nevertheless it is possible to measure a mineral directly in the field. Considering that rough surfaces will diffract X-rays from the detector, which results in artificially low measured concentrations (Morris 2009; Fisher et al. 2014), it is difficult to make accurate measurements on unpolished blocks. Whereas data accuracy may be lower for unpolished blocks, the data may still be of sufficient quality to aid decision making in the field. To evaluate this, we first determined the machine error of Mg# for olivine and clinopyroxene and An% for plagioclase by repeat analysis of single grains in the unpolished blocks. These errors are ±1.9 for Mg# of olivine, ±1.7 for Mg# for clinopyroxene and ±2.0 for An% (1σ standard deviation). As anticipated, the errors are higher than those from carbon coated polished thin sections, but they are still of high enough precision to be useful for mapping igneous stratigraphy. Next, the bSEM-EDS analyses of 11 unpolished blocks are compared to their probe counterparts conducted on corresponding carbon-coated polished thin sections (Table 3 and Fig. 7). The correlation between bSEM-EDS and probe analyses is excellent (R2>0.9) for Mg# of olivine and clinopyroxene, and reasonable (R2=0.69) for An% of plagioclase, indicating bSEM-EDS analyses of unpolished blocks are comparable to their probe counterparts. However, where unpolished blocks were analysed, repeated measurements of different mineral grains in each sample show higher variability compared to analyses of carbon-coated polished thin sections. Standard deviation values of the above ratios for unpolished blocks (combination of machine error and natural variability) are typically between 3 and 8. However, the analysis of unpolished blocks is of sufficient precision to aid exploration geologists with decision-making in the field and it should be possible to improve analyses from cut blocks by polishing them in the field.
We have also compared bSEM-EDS analyses of carbon-coated polished thin sections to microprobe analyses, with both sets of analyses conducted on the same grains (Table 3 and Fig. 7). The correlations between bSEM-EDS and electron microprobe analyses are excellent for Mg# of olivine (R2=0.90) and An content of plagioclase (R2=0.85), and reasonable for Mg# of clinopyroxene (R2=0.73).
In addition, sample precision for bSEM-EDS analyses of carbon-coated polished thin sections are improved substantially when compared to bSEM-EDS analyses of unpolished blocks (Table 3 and Fig. 7). Figure 8 compares down-hole variations in Mg# of olivine and clinopyroxene, and An% of plagioclase between probe analyses and bSEM-EDS of unpolished blocks and carbon-coated polished thin sections. The carbon-coated polished thin sections (bSEM-EDS) and probe analyses methods display broadly similar down-hole variations throughout the igneous stratigraphy, including both the Layered Series and the Marathon Series, indicating interpretations regarding the igneous stratigraphy based on bSEM-EDS analyses will be comparable to their probe counterparts. Fewer data points were collected from unpolished blocks, but the downhole variation is similar to bSEM-EDS analyses from carbon-coated polished thin sections. However, it must be kept in mind that bSEM-EDS analysis in the current work is semi-quantitative, without calibration using standards, and, in addition, peak overlaps (e.g. overlaps for Cr, Ti, V) are common issues for EDS analysis. The data quality of bSEM-EDS analysis will therefore inevitably be less accurate and precise than those determined from WDS analysis on an electron microprobe, particularly with respect to data accuracy.
Geochemistry of Igneous Statigraphy
As shown above, the variation of the molar Mg# for olivine and clinopyroxene is important. However, it is difficult to estimate this ratio for whole-rock compositions because of the presence of magnetite. However, most of the whole-rock TiO2 is contained in magnetite. The laboratory analyses include the determination of FeO by titration from which the whole-rock Fe2O3 was calculated by difference from total Fe2O3. Figure 9 shows a linear correlation between laboratory-based Fe2O3 and TiO2. The data was fit by a least square to a linear equation: Fe2O3 (wt. %) =1.3966TiO2 (wt. %) +1.2732. This equation was then applied to pXRF TiO2 data to deduce pXRF Fe2O3 concentrations, after which, the pXRF FeO data was calculated by subtracting pXRF Fe2O3 from pXRF total Fe2O3. This allowed pXRF whole-rock Mg numbers to be calculated: Mg#WR is molar MgO /(MgO+FeO)×100 (Table 4).
The key pXRF lithogeochemical parameters for defining the different igneous units are Mg#WR, CaO, P2O5, Ba, and Zr and the V/Ti ratio (Supplementary Material and Table 4). Figure 10 shows box-and-whisker plots of these parameters and their down-hole variations are plotted in Figure 11. In Figure 10, Layered Series rocks have lower CaO, P2O5 and V/Ti ratios, and higher Ba and Zr contents than Marathon Series rocks. However, the Mg#WR values of the Layered Series rocks are similar to those of the Marathon Series rocks. Good et al. (2015) also observed a lower V/Ti ratio in Layered Series rocks compared to Marathon Series rocks at the Marathon deposit. Thus, the V/Ti ratio seems to be a good discriminant between these two Series throughout the Eastern Gabbro. The higher Ba contents in the Layered Series gabbros is also consistent with the data of Shaw & Penczak (1996) who recognized Ba-rich biotite in this series. In addition, there is a positive correlation between pXRF K2O and Ba for samples in this study (Fig. 12), indicating that Ba is controlled by biotite (negligible K-feldspar is present in the samples examined in this study, although K-feldspar is observed elsewhere in the Eastern gabbro).
Down-hole pXRF whole-rock chemistry can also be used to differentiate between the Layered and Marathon Series rocks. P2O5 contents in the Layered Series are nearly constant, whereas the Marathon Series rocks have variable P2O5 contents (representing apatite-rich cumulate layers rather than interstitial apatite) that overall increase with depth (Fig. 4). Down-hole pXRF variations of Zr contents in the Layered Series are variable but generally increase with depth. By contrast, the Marathon Series rocks have decreasing Zr contents with depth (Fig. 4). The down-hole variation in pXRF V/Ti ratio of the Layered Series gabbros is nearly constant at around 0.015, in contrast to that of the Marathon Series rocks where it generally increases with depth from 0.015 to 0.1 (Fig. 11). The Mg#WR, although similar in the Layered and Marathon Series, does show subtle variation. The Mg#WR decreases with depth in the Layered Series gabbros (reverse differentiation) v. the Marathon Series rocks, which have an increase in Mg#WR with depth (normal differentiation) (Fig. 11).
The concentrations of Hf, Nb, Ta, and Zr in the different rock units from the ICP-MS analyses are given in Table 5 and their down-hole variations are shown in Figure 13. The Layered Series gabbros contain Hf, Nb, Ta, and Zr concentrations in the range 0.9 – 4.0 ppm, 16 – 73 ppm, 0.7 – 3.0 ppm, and 37 – 168 ppm, respectively, and all of these elements generally increase downward in the Layered Series gabbros (reverse differentiation). The Marathon Series, by contrast, has variable Hf, Nb, Ta, and Zr contents that overall continually decrease with depth (normal differentiation). The down-hole variation in pXRF Zr is included in Figure 13 for comparisons, which is broadly similar to down-hole variations of ICP-MS Hf, Nb, Ta, and Zr.
bSEM-EDS mineral chemistry
Down-hole variations in the Mg# of olivine, Mg# of clinopyroxene, and the An content of plagioclase are shown in Figure 14. The Layered Series gabbros show an overall downward decreasing trend in Mg# of olivine, Mg# of clinopyroxene, and An% of plagioclase. By contrast, the Mg# of olivine and clinopyroxene, and An% of plagioclase gradually increase downward in both the oxide melatroctolite and the apatitic clinopyroxenite in the Marathon Series.
The application of pXRF and bSEM-EDS to Cu-Pd mineral exploration
As discussed above, applications of pXRF to mineral exploration has been demonstrated by many studies, whereas, to our knowledge, this is the first application of bSEM-EDS to mineral exploration. In this section, we examine the combination of pXRF and bSEM-EDS to increase efficiency in mineral exploration.
The down-hole variations in Cu and S values from pXRF are compared to exploration assay data from drill hole FD-13–34 (Fig. 15 and Table 6). There is a strong correlation between the two patterns, despite the different sampling intervals (assay data were determined on continuous 2 m sample intervals, whereas samples for this study were collected over 1 m intervals at 3 m spacing). These results show that pXRF is able to quantify Cu concentrations on powders down to at least 100 ppm. In addition, two zones of mineralization (a zone characterized by low Cu/S ratios at the top of the Layered Series and a zone characterized by high Cu/S ratios at the bottom of the Marathon Series (Fig. 15)) are identified in drill hole FD-13–34 by both exploration assay and pXRF data. The Pd abundance in the upper zone with low Cu/S is at or below detection limits of ICP-MS analysis, whereas mineralization in the Marathon Series with high Cu/S is PGE enriched (unpublished Stillwater Canada data), suggesting high Cu/S ratios seem to be a proxy for PGE mineralization at this locality. Figure 16 further illustrates how the pXRF data can be plotted to identify two different Cu/S ratios corresponding to the two mineralized zones described above. Whereas pXRF is unable to detect PGE, bSEM-EDS can be used as a complimentary technique to verify Pd mineralization.
Several platinum group minerals (PGM) and Au-Ag alloys in Marathon Series gabbros within the high Cu/S ratio zone were identified using bSEM-EDS (Figs 17 and 18). The PGM in Figure 17 are from carbon-coated polished thin sections, but the Au-Ag alloy grain shown in Figure 18 is from an unpolished block. This demonstrates that the bSEM-EDS can be used to identify mineralization in saw-cut drill core in the field, without having to wait for assays. In another example, down-hole variations in whole-rock MgO or Mg#WR from pXRF (Figs 4 and 11) are difficult to interpret without knowledge of mineral chemistry. However, bSEM-EDS analyses show a ‘saw-tooth’ pattern for Mg# of olivine between depths of 270 and 290 m (within the high Cu/S zone) (Figs 8 and 14). Similar patterns have been interpreted by Good et al. (2015) as magma recharge conduits that are favourable for PGE mineralization. Thus, the combination of pXRF and bSEM can be used in the field to identify possible recharge zones. The last example is the replacement of early plagioclase by later more-calcic plagioclase, An50 by An60 by bSEM-EDS analysis (Fig. 19). This texture is notable because chalcopyrite is intergrown with the later plagioclase (Fig. 3E and F) and precious metals are associated with the chalcopyrite (Fig. 19). Similar textures were described in the Marathon Main zone by Good & Crocket (1994), thus another application of bSEM-EDS is to identify favourable textures in the field.
The application of pXRF and bSEM-EDS to mapping igneous stratigraphy
This study shows that the combination of pXRF and bSEM-EDS can be used in the field to map igneous stratigraphy in a manner similar to laboratory-based studies that combine whole-rock lithogeochemistry and mineral chemistry (Li et al. 2000; Egorova & Latypov 2013). The interpretation of results from only one method is more difficult. In this study the Layered Series gabbros including the oxide augite melatroctolite (barren) and the Marathon Series gabbros including the oxide melatroctolite (potentially mineralized) can be distinguished from variations in the major-element and trace-element abundances, e.g. Layered Series rocks contain lower CaO, P2O5 and V/Ti ratio and higher Ba and Zr contents than Marathon Series rocks. We have also shown that Zr, determined by pXRF can serve as a proxy for other high field strength elements and the down-hole variations of these highly incompatible element abundances is useful to help differentiate between the PGE-barren Layered Series and the PGE-enriched Marathon Series (normal differentiation for the Marathon Series gabbros v. reverse differentiation for the Layered Series gabbros). Shaw (1997) described normal and reverse trends for mineral compositions in the Layered Series. In this study we have shown that these trends can also be recognized using bSEM-EDS, in particular for the Mg# of olivine, Mg# of clinopyroxene, and An% of plagioclase. It is the combination of whole-rock lithogeochemical and mineral chemical data that allow for a more robust interpretation of the igneous stratigraphy. Unravelling the mechanisms that caused the reverse and normal trends for the Layered Series and the Marathon Series, respectively, is beyond the scope of this paper. The significance of this work is that variations and trends in the various igneous rocks in the Coldwell Alkaline Complex have been characterized using a combination of field portable techniques. This methodology can be used to guide sampling in an iterative fashion in the field, which increases the efficiency of both mineral exploration and academic research.
The reliability of results from the field portable pXRF and bSEM-EDS are demonstrated in the current study. Both portable instruments are able to provide analyses comparable to their lab-based counterparts. In addition, pXRF is capable of estimating Cu grade down to c. 100 ppm and reliably identifies different Cu/S values that can assist in Cu-Pd mineral exploration because, for mineralization in the Coldwell Alkaline Complex, high Cu/S ratios serve as a proxy for potential PGE mineralization. The technique of bSEM-EDS can also be used to identify PGM grains directly in the unpolished blocks. The ability to rapidly identify PGE mineralization in the field can improve the decision making process in exploration. This study was also successful in differentiating between petrographically similar gabbros that belong to two different magmatic series using pXRF and bSEM-EDS: the Marathon Series, which hosts PGE mineralization, and the Layered Series, which is barren of PGE mineralization. Overall, the combination of pXRF and bSEM-EDS can greatly improve sampling and hypothesis testing in field-based research.
Acknowledgements and Funding
We gratefully acknowledge NSERC, Stillwater Canada Inc. and the China Scholarship for funding this project. John McBride, Katrina McLean and Ryan Ruthart of Stillwater Canada are particularly thanked for their help with field work. Additional support for pXRF analyses was provided by Elemental Controls Ltd. and for bSEM analyses by JEOL. The manuscript was improved by constructive reviews by S. Piercey and G. Hall.
- Received October 1, 2015.
- Revision received May 11, 2016.
- Accepted May 27, 2016.
- © 2016 The Author(s)