Abstract
A successful method of mineral exploration in glaciated terrain is the use of indicator minerals recovered from carefully selected glacial sediments, and subsequently traced back to their bedrock source. The successful application of indicator mineral methods relies on efficient and effective recovery as well as the correct identification of a wide variety of indicator minerals. The Geological Survey of Canada (GSC) has developed protocols for ongoing and future research projects to achieve the highest quality for reporting indicator mineral data. Such protocols include the use of field duplicate samples, blank samples, and base material spiked with known numbers, morphologies, species, and sizes of indicator minerals. Field duplicate samples serve to estimate sediment heterogeneity. Spiked samples are used to monitor the accuracy of the sample processing and mineral identification methods for recovering specific minerals. Blank samples serve to detect potential carry-over contamination. In certain instances, a specific sample processing order is essential and should be communicated to the commercial processing laboratory. Ore-rich samples collected near known mineralization are to be processed last, to reduce chances of carry-over contamination. Repeated indicator mineral counts should be carried out on at least 10% of the heavy mineral concentrates to measure reproducibility (precision) of the mineral counts. All indicator mineral data, original laboratory reports, heavy mineral concentrates, unmounted picked grains, and grain mounts are now archived at the GSC, using specific guidelines.
Indicator minerals are defined as mineral grains present in transported detrital sediments or regolith and indicative of a specific type of mineralization, alteration, or bedrock lithology (McClenaghan 2005). Most indicator minerals typically have a high density; they can be physically separated and concentrated from their host sediments (e.g. till, glaciofluvial sediments, stream sediments, beach sediments, alluvium, laterite, etc.) by gravity, size, and magnetic processes, and they are relatively stable in the surface environment (Averill 2001; McClenaghan 2005). Kimberlite indicator minerals (KIMs) recovered from glacial sediments play a key role in diamond exploration both in indicating the presence of kimberlite and, by their major and trace element composition, its diamond fertility (e.g. Fipke et al. 1995; McClenaghan & Kjarsgaard 2001, 2007; McClenaghan et al. 2002; Lehtonen et al. 2005; Cookenboo & Grütter 2010).
Indicator minerals are also used in exploration for base metals (e.g. Williams & Cesbron 1977; Averill 2001, 2007, 2009; Griffin et al. 2006; Plouffe et al. 2006, 2008; Spry & Teale 2009; Bouzari et al. 2010, 2011; Paulen et al. 2011), precious metals (e.g. McClenaghan & Cabri 2011), rare metals (e.g. Lehtonen et al. 2011), and gemstones other than diamonds (e.g. Galbraith et al. 2009). Research continues to identify additional useful indicator minerals associated with a broad range of mineral deposit types (e.g. McClenaghan et al. 2008; Gent et al. 2011). As such, the use of indicator minerals in glacial sediments is rapidly developing as a broadly used exploration method. Most importantly, numerous mineral exploration targets have been identified (Parent et al. 2004; Plouffe et al. 2006; Paulen et al. 2011) and discoveries (see list of examples in Averill 2001) in Canada and elsewhere have been made by tracing indicator minerals in glacial sediments back to their bedrock source, for example: kimberlites (McKinlay et al. 1997; Carlson et al. 1999; Doyle et al. 1999; Graham et al. 1999; Kirkley et al. 2003; Strand et al. 2009); gold (Prest 1911; Averill & Zimmerman 1986; Sopuck et al. 1986; Sauerbrei et al. 1987; Thomson et al. 1987); uranium (Geddes 1982); and skarns (Aumo & Salonen 1986).
The success of indicator mineral methods in glaciated terrain relies in part on the effective processing of carefully selected glacial sediments for the optimal recovery of indicator mineral grains. During processing, indicator mineral grain loss should be minimized (close to zero), minerals must be correctly identified, and samples must not be contaminated from external sources or other samples. These requirements are essential because a single grain of a specific indicator mineral in a 10 kg sample can be significant for exploration in some geological regions (e.g. Strand et al. 2009; McClenaghan 2011).
Published or archived results and metadata for indicator mineral content in glacial sediments should include evidence that quality control measures were applied and that results can be compared with other surveys. However, protocols to ensure quality in the processing, analysis and reporting of indicator mineral data are not as advanced as those currently used for matrix geochemistry protocols for soils and sediments (e.g. Cook & McConnell 2001; McClenaghan et al. in press). Several commercial indicator mineral processing laboratories in Canada have developed their own internal quality assurance (QA) and quality control (QC) measures and their data should be considered when monitoring QA/QC (Averill & Huneault 2003; Quirt & Maki-Scott 2003; Whiteford 2003; Baumgartner 2006; de Souza 2006; Le Couteur & McLeod 2006; Mircea 2006; Hozjan & Averill 2009; Michaud & Averill 2009). In addition to insisting on internal laboratory QA/QC, users of commercial laboratories should independently and consistently follow protocols to measure the quality of the data obtained from those laboratories. Procedures for monitoring quality of indicator mineral data have been applied in some government surveys (e.g. Prior et al. 2005; Plouffe et al. 2006, 2008; Thorleifson et al. 2007; McMartin et al. 2011a, 2011b; Watson 2011) but the approaches used are very different in each case, and QA/QC protocols are lacking in some surveys.
In most countries, there are minimum standards for disclosure of mineral exploration activities and data. In Canada, for example, National Instrument 43–101 (http://www.cim.org/committees/NI_43-101_Dec_30.pdf) requires that technical information reported by exploration or mining companies must include an indication that a QA programme is in place and is being implemented. The collection of large samples (>10 kg), especially in remote regions, for indicator minerals and sample processing represents a significant expense for exploration and survey field programmes. QA protocols can ensure that this investment returns data of acceptable quality.
The main objective of this paper is to provide a description of the protocols being developed and implemented at the Geological Survey of Canada (GSC) for monitoring the quality of indicator mineral data, from field sampling to archiving the information in databases.
QA/QC protocols: From Field to Archive
The QA/QC protocols followed in indicator mineral surveys at the GSC are presented below from beginning to end of the survey: (1) field procedures; (2) selecting a laboratory and preparing samples prior to indicator mineral separation and analysis; (3) requested laboratory procedures; and (4) archiving of data and indicator minerals.
Field procedures
Basic QC measures for the selection of sample sites, the recording of field notes, and the numbering and labelling of samples must be implemented. These protocols are described in detail in Spirito et al. (2011). The design of an indicator mineral survey including sample spacing is dependent on the objectives of the GSC project and style of sediment dispersal which are also described in Spirito et al. (2011). For till samples, after considering site selection and geological context (cf. Spirito et al. 2011), the sample is collected and clasts larger than a few centimetres as well as organic debris are removed by hand. Glaciofluvial sediment samples (e.g. esker) can be sieved on site (c. 1 cm) to remove pebbles (e.g. Little 2004).
Sampling tools
At the onset of a sediment sampling programme, sampling tools should be examined and thoroughly cleaned in order to reduce their role as potential sources of contamination. Paint, varnish and other types of surface coatings may be removed from tools prior to sampling. Given the nature of sediments, hard tools made of various metal alloys are required; plastic tools should not be used. Contamination from these tools may occur in the form of metal shavings of various unnatural forms and should be expected and recognized. Knowing the chemical composition of the metallic portion of the sampling tools (especially drill bits) might be necessary if the heavy mineral concentrates (HMCs) are also submitted for geochemical analyses. To avoid cross-contamination, sampling tools must be cleaned in between sampling sites using a steel or hard-bristle brush. Water can be used to clean the tools if available.
Samples should be collected in new plastic bags (at least 6 ml), plastic pails, or in metal pails lined with large plastic bags. Rice or cloth bags should not be used for till or glaciofluvial sediment sampling as they are slightly porous and allow loss of fine-grained material and/or contamination. Care must be taken when transporting samples in the field to avoid puncturing the bags and contaminating samples.
Sample size
Sample size is largely dependent on glacial sediment texture, but may also be dictated by the range and type of analyses to be performed. As a general guide, a glacial sediment sample must contain an average of 5–10 kg of sand (Clifton et al. 1969; Averill 2001) to obtain an adequate number of indicator mineral grains to be useful for mineral exploration. If gold or platinum group minerals (PGMs) are targeted, a significant silt component is also important because most grains of these minerals are silt-sized (Averill 2001, 2009). Sample size is also in part a function of the signal-to-noise ratio. Therefore, in regions with elevated background indicator mineral content, larger samples may be necessary to account for sediment heterogeneity and sample representativity.
In the field, a consistent sample size is collected based on volume (e.g. full pail, full sample bag, etc.). The weight of the sediment will vary according to moisture content, sediment compaction and composition. Generally, for most GSC projects, a full 22 -L pail equivalent to c. 20–40 kg is collected in regions where the till contains small amounts (ca. <25%) of sand-sized material (e.g. Western Canada Sedimentary Basin or Cochrane Till in northern Ontario). In regions underlain by igneous and metamorphic rocks where the till contains c. >50% sand (e.g. Canadian Shield, Appalachians), a 10 -L pail or a large sample bag of equivalent volume is used (c. 10–20 kg) (Fig. 1). Similarly, 10 L samples of glaciofluvial sediments are typically collected which correspond to c. 12–25 kg, because for the same volume, glaciofluvial sediment is usually heavier than till.
Typical sample size (large bag) collected for indicator mineral analysis in regions underlain by igneous and metamorphic rocks where the till matrix contains c. >50% sand. The smaller sample is processed for geochemical analyses on the matrix portion of the sediment.
Duplicate samples
Duplicate samples can be collected in the field similarly to the protocols followed in the Canadian national geochemical reconnaissance surveys (e.g. Cook & McConnell 2001; Friske & Hornbrook 1991). Duplicate samples are primarily collected to estimate sediment heterogeneity (site variability). Comparing the indicator mineral content of field duplicate samples only estimates sediment heterogeneity because the variability in their indicator mineral content is influenced by sediment heterogeneity and the precision of the laboratory separation and mineral identification procedures.
A field duplicate sample is collected from a second location, ideally up to 10 m from the first site, in the same sedimentary unit and from similar depth (Fig. 2). The field duplicate sample is placed in a separate sample container and given a separate sample number. To optimize sampling time, field duplicates should be collected at accessible sites where sediment is easy to sample and abundant. If no separate duplicate samples are collected in the field (e.g. because of helicopter-supported survey with limited weight capacity), one large sample can be subdivided into two separate containers with different sample numbers prior to sending the samples to the processing laboratory.
Duplicate sample (arrow) collected approximately 10 m from a routine sample. It is collected at the same depth within the same till unit as the routine sample.
Field contamination
Anthropogenic contamination of glacial sediments will affect the indicator mineral component and should be expected in areas proximal to present- and past-producing mines and related infrastructure (e.g. Bajc & Hall 2000; McMartin et al. 2002; Hozjan & Averill 2009; Michaud & Averill 2009). Near-surface till that appears to be undisturbed can actually be highly contaminated in these regions. Areas with a long mineral exploration history should also be approached with caution. Sites affected by waste drilling water and fluids, drill cuttings and other sources of contamination should be identified (Fig. 3). These areas should be sampled with extreme caution. For instance, samples should be collected at a minimum depth of 0.5 m even if unoxidized till is exposed at surface, sampling equipment should be cleaned carefully, and clean sample containers should be closed until the last possible moment before they are filled with sediment sample.
Till sampling sites near potential anthropogenic contamination: (A) till sampling site (ss, arrow) near a diamond drill; note the drilling fluids (df, arrow) in the foreground which could be a source of contamination; (B) sampling till in an open-pit mine; the top part of the section represents waste rock and only in the lower part of the exposure is till exposed; sampling site shown with arrow.
Other sources of contamination should be completely avoided. Coveralls and other types of clothing that have been used around heavy machinery, drilling, or rock-cutting operations must not be worn. The sampler must not be wearing any jewellery, as rings are sources of contamination (Kontas 1991) and could impact geochemical analyses of HMCs.
Selecting a laboratory and preparing samples prior to processing for indicator mineral analysis
Selection of a commercial processing laboratory to carry out the recovery and identification of indicator minerals is critical. The commercial laboratory should be willing to use and disclose internal procedures used to monitor QA/QC. Taking a tour of the laboratory is strongly recommended to help identify where potential contamination or loss of minerals may occur (Doherty 2009). It is recommended that the general geological setting and known mineral deposits from which the samples were collected be communicated to the laboratory prior to sample processing. Laboratory personnel can then pay particular attention to the expected suites of indicator minerals.
Prior to sending samples to a processing laboratory, specific information to monitor data quality must be requested. In addition to the indicator mineral grain identification and counts, the processing laboratory should report:
sample processing flow chart for every sample batch;
starting weight and subsequent weights of each fraction produced at each step in the process;
listing of magnetic separations performed, the equipment used including its setting, and the weights of each ferromagnetic and paramagnetic fraction produced;
the sample processing order, to ensure the laboratory has followed the requested sequence (see later section);
batch number(s) and sample numbers in each batch;
internal QA/QC procedures followed and their results;
the weight of the split of the heavy mineral fraction picked for indicator minerals, if only a split of this fraction was examined for indicator minerals instead of the entire fraction.
Blank samples
A blank sample (i.e. a sample that contains no indicator minerals of interest) should be processed at the beginning of each sample batch. This first blank sample cannot prevent contamination; it is used to detect cross-contamination from previously processed sample batches in the laboratory. Contamination may extend beyond the first sample if laboratory processing equipment is highly contaminated. In addition to an initial blank sample, it is estimated that one additional blank sample should be inserted for at least every 50 samples, as a minimum, in order to assess the potential for cross-contamination within the same sample batch. If fewer than 50 samples are processed in a batch, it is recommended that one blank be inserted at the beginning and a second blank part way through the batch. If the sediment samples are suspected of having large numbers of a specific indicator mineral of interest, it is advisable to insert a blank sample after a suspected rich sample, to monitor cross-contamination between samples. Certain commercial laboratories process their own internal blank sample after processing what they observe to be a highly anomalous (more than a 1000 indicator mineral grains) sample (e.g. Averill & Huneault 2006).
The blank material may be sediment that has been previously tested and is known to contain no indicator minerals of exploration interest. Currently, the blank material used by the GSC is a weathered Silurian-Devonian granite (grus) of the South Nepisiguit River Plutonic Suite (Wilson 2007; McClenaghan et al. 2012) collected approximately 66 km west of Bathurst, New Brunswick (Fig. 4). Following numerous tests, the HMCs from the weathered granite were determined to contain mainly hornblende, with trace amounts of garnet, epidote, limonite (goethite), titanite (leucoxene), pyrite and very rare gold grains.
(A) Sampling site of the unconsolidated weathered Silurian-Devonian granite (grus) used for GSC blank heavy mineral samples; (B) close-up view of the blank sample material.
Spiked samples
A spiked sample is defined as a sample voluntarily contaminated with an established number of known species of indicator mineral grains, referred to here as spiking grains. Spiked samples are used to quantitatively monitor the effectiveness of a processing laboratory at recovering and identifying specific indicator minerals (Michaud & Averill 2009). Spiked samples are viewed as the equivalent of primary standards used to evaluate analytical precision of geochemical analyses. Several commercial laboratories use them to internally monitor QA (Averill & Huneault 2003; Quirt & Maki-Scott 2003; Whiteford 2003; Baumgartner 2006; de Souza 2006; Le Couteur & McLeod 2006; Mircea 2006; Michaud & Averill 2009). Regardless of the internal measures used in commercial processing laboratories, QA is also monitored independently by the GSC.
The base material, that is, the sediment into which the spiking grains are introduced, should be as similar as possible to the routine samples. The composition of the base material should be well characterized following processing and picking of several splits of the base material, to ensure that it does not contain the indicator minerals being sought (Michaud & Averill 2009). The base material should not be low density quartz as this material simplifies heavy mineral separation and embellishes recovery rates (Michaud & Averill 2009). Furthermore, the base material should not contain exceptionally high concentrations of heavy minerals (Michaud & Averill 2009). The spiking grains should be: (1) within the size range of the expected indicator minerals (typically between 0.25 and 2.0 mm); (2) variable in size; (3) derived from natural sediments and, if possible, not from crushed bedrock (see details below); (4) not fractured or well cleaved such that they can easily break into smaller particles during processing (Hozjan & Averill 2009; Michaud & Averill 2009); and, (5) positively identified as a specific mineral species by scanning electron microscope (SEM) methods.
For accurate identification once recovered, spiking grains should be laser-etched (Whiteford 2003) or as a minimum, photographed or imaged using a SEM. Laser-etched grains should be examined prior to spiking to ensure that the etching process has not fractured them and made them susceptible to breakage. If possible, angular grains obtained from crushed bedrock must not be used as spiking grains because they will not behave in the same manner as typical sub-angular to sub-rounded grains derived from sediments during the concentration process (Hozjan & Averill 2009). They may also differ significantly in morphology compared to mineral grains naturally present in the sample such that they are easily recognized by the laboratory and, therefore, potentially bias the assessment of processing and picking recovery rates. Artificial density markers with specific size range and density are commercially available and can be used for spiking (Gent et al. 2011), but the following limitations should be considered: (1) angular shapes do not behave in the same manner as natural grains during sample processing; (2) some markers are magnetic and will be prematurely removed during ferromagnetic separation; (3) solvents, such as acetone, used to dilute heavy liquids and to clean HMCs may alter the surface and the shape of the density markers (Michaud & Averill 2009); and, (4) some markers are made of plastic and are unsuitable if samples are dried at high temperature (Towie & Seet 1995).
An alternative to using spiked samples is the insertion of indicator mineral-rich samples, collected near a known mineralized bedrock source, into sample batches. This procedure is not recommended at the GSC for two reasons: (1) the exact number of indicator minerals in a mineral-rich sample is unknown due to the heterogeneity of glacial sediments; and, (2) indicator mineral-rich samples represent a potential source of carry-over contamination.
It is suggested that as a minimum, one spiked sample should be submitted per group of 50 samples or equivalent to 2% of the sample batch. The introduction of the spiking grains to a base sample should be done in a clean environment devoid of potential external contamination. The spiking grains should be mixed thoroughly into the sample and not sprinkled on top of the sample. If grains are simply placed on top of the sample, they can be lost if some sediment is accidentally spilled. In addition, certain laboratories remove a representative split from each sample prior to processing them for indicator minerals and some of the spiking grains may be removed inadvertently at this stage (Averill & Huneault 2003; Michaud & Averill 2009). As part of the GSC protocol, the laboratory is instructed not to remove an archive split from any of the samples to be processed for indicator minerals. Instead, the archive split is obtained from a separate sample collected for geochemical analysis at the same site and sampling depth as the indicator mineral sample.
A mineral grain library is under development at the GSC and will be used as a source of spiking grains. Various species of indicator minerals from different project areas throughout Canada will be collected and stored for spiking. Spiking grains inserted and recovered during sample processing will be returned to the mineral grain library.
For base material, the GSC is using till recovered from a borrow pit near Almonte, Ontario, (Henderson 1973; Richard 1976) which is texturally typical of till derived from the southern Canadian Shield (Fig. 5). The average composition of the sediment is reported in Table 1 following the processing and picking of eight (13 kg) separate splits at the commercial laboratory of Overburden Drilling Management Limited. Heavy mineral separations were completed using a combination of shaking table and heavy liquid (specific gravity (SG) 3.2; methylene iodide diluted with acetone). Per 10 kg of table feed (<2 mm material), 72 g of non-ferromagnetic heavy minerals were recovered on average from each of the eight samples, including the following: 51 g in the 0.25–0.5 mm-size fraction, 13 g in the 0.5–1.0 mm-size fraction, and 3 g in the 1.0–2.0 mm-size fraction. Ferromagnetic minerals were removed with a hand magnet prior to sieving those size fractions. Heavy minerals were examined in all size fractions using a binocular microscope, and suspected indicator minerals were counted and some verified using SEM imaging.
(A) Sampling site of till base material near Almonte, Ontario; (B) close-up view of the sample material.
Average content of heavy minerals in till sample from Almonte, Ontario (GSC base material for spiking); N = 8 samples processed and picked; 0.25 to 2 mm sized fraction
The spiked and blank samples can be inserted in a sample batch in the field or after the field season depending on logistics. Preferably, the spiked and blank samples should be sent to the laboratory along with the routine survey samples so that the inserted QA/QC samples cannot be identified by the commercial laboratory’s staff.
Labelling system and order of processing
The labelling system for blank and spiked samples should be part of the sequence used for the routine sample set so that the QA/QC samples are not recognized by the laboratory and do not receive additional attention. The order in which all samples are to be sequentially processed should be predetermined and communicated to the laboratory. If known, the samples collected near mineralized zones with potentially high concentrations of sought indicator minerals should be processed at the end of the batch to limit cross-contamination. Samples should be shipped to the processing laboratory in pails or other secure containers to limit damage to sample bags and reduce the risk of accidental spills or contamination.
Laboratory procedures for the recovery of indicator minerals
Towie & Seet (1995), Gent et al. (2011) and McClenaghan (2011) provide an account of several processing methods available for the recovery of indicator minerals from unconsolidated sediments. The purpose of this section is not to repeat those reviews but to summarize the methods adopted by the GSC to recover and identify indicator minerals and to conduct chemical analyses on individual mineral grains. These adopted methods build upon pioneering indicator mineral studies by McLeod (1959), Dreimanis (1958) and Lee (1963, 1965, 1968) and have been utilized in numerous GSC projects for over two decades with satisfactory results. They have been recommended by Spirito et al. (2011) for ongoing and future GSC projects.
At the GSC, samples collected for indicator minerals are processed in a commercial laboratory following the flow chart depicted in Figure 6. Glacial sediment samples are first disaggregated and wet-sieved at 2 mm. The <2 mm fraction is then pre-concentrated on a shaking table to remove the lightest minerals. If a sample batch is targeted primarily for recovery of kimberlite indicator minerals (KIMs), each sample is passed across the shaking table twice to enhance recovery of the larger (>0.5 mm) indicator minerals (e.g. McClenaghan et al. 1996, 2004). The >2 mm fraction is retained for clast lithology studies. The heavy mineral fraction recovered from the shaking table is micro-panned for gold grains which are counted, measured and classified as pristine, modified or reshaped (DiLabio 1990). At this stage, other minerals of economic interest such as PGMs, sulphides and uranium minerals that are recovered during panning are counted and their abundance and size reported. Heavy minerals recovered from the shaking table are further concentrated with heavy liquid (e.g. methylene iodide), most commonly with a SG of 3.2, to include the lowest density indicators in the heavy mineral separates (e.g. Cr-diopside and fosteritic olivine). As indicated in Figure 6, separations at SG 2.9 to 3.3 may be used depending on project objectives. The ferromagnetic fraction of the heavy minerals is removed with a hand magnet and stored, or retained for examination and further analysis. The non-ferromagnetic heavy mineral concentrate (NFM-HMC) is sieved into three size fractions, 1–2 mm, 0.5–1 mm and 0.25–0.5 mm, because indicator mineral identification is facilitated on concentrates of limited grain-size range (Towie & Seet 1995). The 0.25–0.5 mm-size fraction is further separated with an electromagnet, set at different amperage, resulting in non-paramagnetic to strongly paramagnetic fractions which further facilitate mineral identification based on magnetic properties. Amperages used are reported by the commercial laboratory. At this stage, the variability of a mineral's magnetic properties based on its composition has to be considered (Gent et al. 2011; McClenaghan 2011).
Flow chart depicting the processing of glacial sediment samples for indicator minerals adopted at the GSC; modified from McClenaghan et al. (2012). *Samples targeted for recovery of kimberlite indicator minerals are passed twice across the shaking table.
Visual identification of indicator minerals is carried out by experienced personnel in a commercial laboratory using binocular microscopes. Mineral identification is assisted with a SEM and ultraviolet light (Gent et al. 2011; McClenaghan 2011). The recognition of a specific mineral species in a heavy mineral concentrate relies on the expertise and experience of the personnel. Heavy mineral identification on smaller size fractions (e.g. 0.18–0.25 mm) is carried out by certain laboratories upon request. In addition to mineral abundance, commercial laboratories are able to report grain morphology and the presence of specific surface features such as kelyphite rims on kimberlitic garnets (e.g. Mosig 1980; McCandless 1990; McClenaghan & Kjarsgaard 2007). This information will assist in the assessment of relative transport distance from the bedrock source.
The chemical composition of indicator minerals provides key information about the genesis of the mineralization, alteration or bedrock lithology and, in some cases, the mineral potential of the deposit (e.g. diamond grade of a kimberlite). Therefore, individual indicator mineral grain analysis is conducted at the GSC using the following methods: SEM, electron microprobe (EMP), or laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) (Jackson 2009). For EMP and LA-ICP-MS analyses, grains are mounted in polished epoxy disks (Gibbs 2007).
Apart from the manual and visual inspection of HMCs for the detection of indicator minerals, automated methods are now available in commercial laboratories and include mineral liberation analysis (MLA) and quantitative evaluation of material by scanning electron microscopy (QEMSCAN) (e.g. Keulen et al. 2009; de Souza et al. 2011). These methods have been recently used at the GSC for a few projects and complement the visual identification. The automated methods allow the rapid identification of minerals in the usual picking-size fraction (0.25–2.0 mm) as well as the smaller <0.25 mm fraction that is difficult and time-consuming to visually examine. These automated methods are currently more expensive than the conventional visual mineral identification methods and are being tested with different mineral suites to assess their accuracy. Although there are benefits in using these automated methods, there is an advantage to having a trained person examine a HMC. Features such as adhering minerals or host rock attached to indicator minerals, as well as indicator mineral grain morphology, can only be recognized by a trained person and are important to help identify the minerals correctly. In addition, the number of mineral grains that can be examined by automated methods is presently limited. A 10 kg till sample typically yields c. 20 g of 0.25–0.5 mm heavy minerals, or c. 200 000 mineral grains. When manually and visually inspected, every grain is examined whereas at most 2000 grains (or 1% of the HMC) are analysed by automated methods – an insufficient number to quantify or even detect many indicator minerals.
After receiving data from the indicator mineral processing laboratory
A number of steps are required once data are received from the commercial heavy mineral processing laboratory to ensure data quality and that the information is complete before data interpretation and mineral potential assessment.
To verify the reproducibility of visual grain counts, at least 10% of HMC samples should be resubmitted for mineral identification and picking (Doherty 2009; McClenaghan 2011). Indicator mineral-rich and -poor samples should be selected for this test. Samples selected for recounting can also be chosen on the basis of till matrix geochemical results (e.g. McClenaghan et al. 2011). Samples should be renumbered and resubmitted to the same laboratory so that laboratory personnel cannot identify the samples being re-examined. To evaluate the precision of mineral species identification, the concentrates can be sent to a different mineral picking laboratory.
All QA/QC results should be communicated to the indicator mineral processing or analytical laboratory to ensure that the sources of any perceived errors are identified and corrected. Open communication between the client and the processing laboratory will lead to improvement in the laboratory methodology and will further enhance the quality of the reported data.
Chemical analyses on indicator mineral grains
As indicated above, a number of analytical methods are available for mineral grain chemistry. Regardless of the analytical method, certified mineral reference standards and duplicate grain analyses are required to monitor the analytical accuracy and precision, respectively (de Souza 2006; Doherty 2009).
For most indicator minerals, such as the kimberlite indicator mineral (KIM) suite of Cr-pyrope, eclogitic garnet, Mg-ilmenite, Cr-diopside, olivine and chromite, mineral grains can be analysed on automated runs using a SEM and an EMP, which have the advantages of less analytical time and thus lower costs. Each of these indicator minerals, in most cases, only requires a single analysis to characterize its composition and this is usually performed on the centre of the grain. Some indicator minerals require more precise location of the analytical beam because of adhering gangue minerals, mineral inclusions or the polyminerallic nature of the grain. For some mineral species, the EMP operator must be present during analysis so that if the desired mineral species in the grain is not analysed by the beam, the grain can be moved and re-analysed at that time. For example, gahnite grains recovered from till down-ice of metamorphosed volcanogenic base metal deposits often have adhering mica and quartz that can obscure much of the gahnite.
Data and sample archiving
Once indicator mineral data are received from a processing or analytical laboratory (single grain chemistry), the data should be archived in such a way as to preserve the unmodified laboratory reports in digital form. The archived mineral abundance report should include the original mineral classification. The mineral chemistry report should include the confirmed or reclassification based on chemical composition. HMCs, picked indicator minerals, epoxy grain mounts or slides, and other fractions produced during processing, picking and analysis of each sample are recovered from the processing or analytical laboratory and stored at the GSC for future consultation. At the GSC, laboratory reports and associated metadata are being archived in the Canadian Database of Geochemical Surveys, which can be accessed via the Geoscience Data Repository (http://www.gdr.nrcan.gc.ca/geochem). Sample metadata are entered into the GSC sample management system.
Indicator mineral metadata
The indicator mineral data collected from each survey or study are initially released to the public in open file format. Published information should include the following metadata and data listings as a minimum (from Spirito et al. 2011):
Sample medium: till, glaciofluvial sediments, stream sediments, etc.;
Name of processing laboratory;
Name of mineral identification laboratory (if different from the processing laboratory);
Weights of material processed for recovery of indicator minerals (original sample weight, weight of table feed (<2 mm);
Pre-concentration method (e.g. panning, hydro-separator, shaking table, dense media separator, Knelson® concentrator, jig, rotary spinal concentrator, other);
Heavy liquid separation: name of liquid, and density;
Magnetic separation: type of magnet used (e.g. hand magnet, Frantz®, roll magnet, or other methods) and amperages if an electromagnet is used;
List of all size and density fractions prepared and their individual weights;
Weight and size range of fraction(s) examined for indicator minerals and percentage of concentrate examined for each sample if the complete concentrate was not examined;
Mineral identification method: visual scan under the binocular microscope, MLA, QEMSCAN, SEM, or other methods;
Mineral chemistry determination method, machine operating conditions, and laboratory name for: EMP, SEM, LA-ICP-MS, other;
Raw indicator mineral count data as reported by the picking laboratory;
Indicator mineral count data as confirmed by EMP, SEM or other methods;
Indicator mineral count data as values normalized to total sediment weight processed (e.g. number of grains per 10 kg table feed (<2 mm).
Note: the total indicator mineral grain counts are never combined; they are reported separately for each size fraction.
In summary, to successfully apply the protocols described in this paper, specific procedures have to be implemented at various stages of a project. To facilitate the implementation of the protocols, Table 2 presents a checklist of the basic steps followed by GSC staff. These protocols are offered as a reference for any agency or company conducting indicator mineral surveys. Furthermore, although this paper focuses on indicator minerals in glacial sediments, some of the guidelines and principles may easily be adapted for the recovery of indicator minerals from non-glacial sediments (e.g. stream sediments, soils, beach sediments, alluvium).
Check lists for QA/QC for processing of glacial sediments for indicator minerals (IMs)
Examples
The following three examples of indicator mineral surveys demonstrate that contamination and the missed recognition of minerals can occur, and that QC measures and interpretation of data can help to identify these problems.
The presence of potential KIMs in 20 glaciofluvial sediment samples in NE British Columbia, Canada, (Simandl et al. 2005), generated interest because diamondiferous kimberlites were not known in this region (Simandl 2004). A follow-up sampling survey in the same region reported additional potential KIMs in glaciofluvial and stream sediments and the presence of one diamond (Simandl et al. 2006). The recovered diamond was nearly 0.8 mm in size. The HMC from which the diamond was recovered contained abundant fluorite and no additional KIMs. The presence of fluorite along with KIM led to the interpretation that the diamond in the sample was a result of cross-contamination since there is no known relationship between diamond and fluorite in kimberlite-hosted diamond deposits (Simandl et al. 2006). The diamond could have been derived from a sample previously processed in the commercial laboratory and its source is still unknown. This example demonstrates that cross-contamination can occur and, in specific instances, can be identified by determining the mineral assemblages present. In this case, the use of blank samples could have helped to identify cross-contamination.
As part of a regional till sampling survey aimed at determining the diamond exploration potential of NW Alberta, Canada, till samples along with one spiked sample were submitted to a commercial laboratory for indicator mineral processing (Plouffe et al. 2006; 2008). The base material for the spiked samples was till from the Brownvale region in Alberta (Dufresne et al. 1996; Prior et al. 2005). It was spiked with 40 KIMs (10 pyrope, 11 Mg-rich olivine (forsterite), 9 chromite and 10 Cr-diopside grains) recovered from stream sediment and bedrock samples of the Buffalo Head Hills kimberlite field 180 km away (Prior et al. 2005). The initial results provided by the processing laboratory reported recovery of the spiking grains as follows: 7 out of 10 pyrope, 10 out of 10 Cr-diopside, 5 out of 9 chromite, 2 out of 11 olivine. The olivine grains were colourless, typical of the Buffalo Head Hills kimberlites, and thus difficult to visually identify. The QC results were communicated to the processing laboratory and all HMC samples (N=50) were re-examined by another person from the same laboratory. Following this second count, the final picking of the spiked sample was as follows: 100% recovery of pyrope, 100% recovery of Cr-diopside, 89% recovery of chromite, and 73% recovery of olivine spiking grains. Plus, five additional olivine grains were found in one routine till sample and one pyrope grain and one Cr-diopside grain were observed in a second routine till sample. Because of the greater picking experience of the second person, a large number of dark grey to black sphalerite grains were also identified in several samples, which led to the identification of a large sphalerite glacial dispersal train (Plouffe et al. 2006, 2008; Paulen et al. 2007, 2011). This particular example demonstrates that the mineral identification stage is highly operator-dependent. Spiking samples with indicator minerals that have characteristics similar to the expected minerals represents an efficient method of measuring detectability of those minerals by the processing laboratory. Therefore, if a particular indicator mineral is expected in a sample batch, it should be communicated to the personnel at the processing laboratory so that they can confirm their capacity for recognizing the indicator mineral, and that a person who is familiar with the characteristics of that mineral can be assigned to examine the samples.
Till samples were collected over the Thompson Nickel Belt in northern Manitoba, Canada, to characterize the geochemical and mineralogical signatures in till near known magmatic Ni-Cu mineralization (McClenaghan et al. 2011). The silt and clay-sized fraction (<0.063 mm) was geochemically analysed and indicator minerals were recovered from the 0.25–2.0 mm heavy mineral fraction. A strong correlation exists between the highest Ni concentrations and the number of pentlandite grains per 10 kg till and between elevated Cu levels and number of chalcopyrite grains per 10 kg (McClenaghan et al. 2009). Initially, poor correlations between Pt and Pd concentrations in till and the number of visible sperrylite (PtAs2) grains were noted. Three till samples that contained elevated Pt and Pd concentrations but reported to contain no sperrylite grains were re-examined and sperrylite grains were found (McClenaghan et al. 2011). This example demonstrates that geochemical analysis can be an indicator of the presence of specific indicator minerals and can be used to verify indicator mineral results.
Conclusions
The GSC has developed protocols for collecting and processing glacial sediment samples to recover indicator minerals for its own regional surveys and case studies. Specific procedures, summarized in Table 2, are used at all stages of a project, from the field to the archiving of the data and samples. By following these protocols, contamination of the samples will be limited in the field and GSC staff will be able to monitor data quality, reproducibility and accuracy and to detect problems such as cross-contamination which can occur in a heavy mineral processing laboratory. The protocols presented in this article will continue to be improved over time as more samples are processed, and as mineral separation and identification procedures at commercial laboratories evolve and improve.
Acknowledgments
This paper first benefited from the reviews completed by M. Parent and S.A. Averill and secondly from the journal reviews by T.E. McCandless and L.H. Thorleifson. Some key elements and concepts presented in this paper were developed following discussions with A. Bajc, T. Ferbey, T.A. Goodwin, B.A. Kjarsgaard, and L.H. Thorleifson. Comments and feedback on the protocols presented in this paper are welcome and can be communicated to the author or co-authors. ESS Contribution number: 20110278.
- © 2013 AAG/Geological Society of London
References
Processing of glacial sediments for the recovery of indicator minerals: protocols used at the Geological Survey of Canada
