**3. Results**

#### *3.1. Grain Mount Density Gradients*

Grains on each epoxy grain mount were classified into three false colors according to their density; red (>4.25 SG) for the heaviest 1/3 of minerals in the mineral reference library, and green (3.24–4.25 SG) and black (<3.24 SG) representing intermediate- and light-density minerals, respectively (Figure 4). This classification was done to determine if any settling had affected the grains during preparation of the epoxy mounts. Heavy minerals (red) were more abundant in the bottom of the vertical slabs 1 and 3 from the primary mounts. The number of grains was also highest in the bottom 1/3 of slabs 1 and 3. In contrast, the basal surface of the primary mount (slab 2) contained a higher relative abundance of heavy minerals to lighter minerals, as well as a higher overall abundance of grains when compared to the vertical slab surfaces. This contrast between slab 1 and 3 versus slab 2 was observed in all secondary mounts for all till samples, for all size fractions examined. The differences between slabs 1 and 3 versus slab 2 were interpreted to be a settling gradient from top to bottom that formed after the grains were stirred into the epoxy during preparation of the primary mount.

**Figure 4.** False-color grain maps of: (**A**) a portion of the slab 2 basal surface of a HMC epoxy grain mount, and (**B**) the entire slab 1 vertical cross-sectional surface of a HMC epoxy grain mount. Particle abundance per unit area and the relative abundance of heavy (red), intermediate (green) and light (grey) are depicted in the pie charts above for each red-shaded area. Three sections were analyzed on the vertical slabs and they display variations in particle abundance with depth through the slab.

#### *3.2. Mineral Liberation Analysis (MLA) Error Estimation*

Modal mineralogy tables were generated for a portion of basal slab 2 from two di fferent secondary mounts (one of the 185–250 μm fraction and one of the <64 μm fraction) for the duplicate MLA runs (before and after calibration). The absolute di fference between the values generated by each scanning routine (in area percentage) was calculated for each mineral identified. The absolute di fference was divided by the mean of the two area percentage values to prevent bias between minerals found in very low or very high abundances, and then divided by 2 to give an approximation of the percentage error above or below that mean value.

Values were calculated for successive scans of the same area, within the same scanning routine ("run"). This produced four "in-run" error comparison values, one for each of the two size fractions examined in the run before and the run following gun alignment calibration. In addition to the four in-run values, 8 "out-run" values were computed by comparing all possible permutations of before- and after-calibration scans for each of the two size fractions examined. Table 2 details the resulting in-run error values (before and after calibration) and the out-run error values for several indicator minerals.


**Table 2.** Calculated error between two identical MLA sample runs, before and after routine gun alignment calibration, for two size fractions of polished HMC grain mount. Out-run error is calculated using all runs, regardless of pre- or post-calibration.

The identification of minerals using EDS requires the identification of peaks in EDS spectra, with the x-axis location of a peak representing one of the energy emissions of the element of interest and the area of the peak (PB-ZAF corrected) reflecting the relative abundance of that element. The method is therefore semi-quantitative, and a fully quantitative technique like wavelength-dispersive spectroscopy (WDS) is required to obtain precise measures of elemental abundance in a mineral. The semi-quantitative nature of automated mineralogy (EDS) analysis makes it di fficult to discern between minerals with similar composition or highly variable stoichiometry. Conversely, automated mineralogy systems will excel at identifying minerals containing rare elements like U or Zr, or minerals with distinct, simple spectra like many sulfide minerals [45]. Certain minerals were found consistently to represent outlier values. These minerals include garnets, pyroxenes, and other complex silicates, including axinite.

 0.0002

 1.0276

## *3.3. Modal Mineralogy*

Staurolite

 1.0273

The basal slab 2 surface had the highest number of grains per unit area and highest proportion of the heaviest minerals for each sample. We deemed the basal slab 2 the most appropriate for use in the rest of the study and, therefore, mineralogy data are reported only for the slab 2 portion of each of the 16 secondary mounts.

Several reporting metrics can be generated by automated mineralogical software packages, including volume percentage (vol. %), weight percentage (wt. %), grain count, and area percentage. The calculation of 3D vol. % values from a 2D surface requires the generalization of grain volume as a sphere or ellipse using the length of the long axis present on the polished grain surface [43]. This is a powerful tool for many applications of automated mineralogy as it allows a user to establish a "grade" for a sample, when mineral chemistry data is combined with the resultant volume from MLA.

Simandl et al. [46] found a high coe fficient of determination ( *R*<sup>2</sup> =0.90) between the wt. % estimation of Nb-bearing minerals using QEMSCAN ™ and Nb content determined by X-ray fluorescence (XRF), and a similarly high R<sup>2</sup> value for the abundance of monazite with ( Σ La, Ce, Pr, Nd) values obtained from borate fusion/nitric acid digestion/ICP-MS. However, the elements observed are not commonly present in a wide range of minerals. The method of determining wt. % within automated mineralogy software packages assumes either perfect mineral stoichiometry or requires mineral chemistry data (e.g., by electron microprobe) and, therefore, wt. % values are not appropriate for reporting the abundance of most till minerals, which have variable sources and stoichiometry, or where the elements of interest are commonly found in multiple mineral phases. Likewise, volume percentages do not enrich the interpretation of mineral abundance used in indicator mineral exploration and, therefore, area percentage (which both volume percentage and weight percentage are derived from) is the most appropriate reporting metric for till indicator mineral studies.

Although grain counts are an intuitive metric for understanding indicator mineral abundance and are more easily related to traditional visual grain counts for the >250 μm HMC, there is still a small amount of uncertainty introduced by the MLA software when it defines and separates grain boundaries. Area percentage is calculated by dividing the number of pixels assigned to a particular mineral value by the total number of pixels measured. This calculation supplies a metric that, although less intuitive than grain counts, represents a more precise quantitative measure.

Area percentage can be used in tandem with the number of grains to discern information about the occurrence of indicator minerals that would not be observed from either metric alone. Low grain counts with correspondingly high relative area percentage values sugges<sup>t</sup> that a mineral is present as discrete whole grains (Figure 5). High grain counts combined with low area percentage values sugges<sup>t</sup> that a mineral is present as high numbers of smaller grains or disseminated inclusions in other mineral grains. These conclusions must be confirmed by observing the mineral in a false-color MLA grain map. When using grain counts it is important to normalize values to allow for comparison between samples containing varying numbers of grains. Hulkki et al. [47] used two methods for normalizing grain count data, with their disaggregated bedrock samples being normalized using the equation:

(Number of a specific indicator mineral grain in a sample/ Total number of minerals grains in a sample) × 1000

This ratio accommodates the varying number of grains analyzed between subsamples. The stream sediment grain counts of Hulkki et al. [47] were normalized to a 1 kg processed sample weight, although the authors noted that given the extremely low abundance of the Cu minerals of interest, the normalized values were very close to the numbers of observed mineral grains.

Values in this study were normalized using the same method as the bedrock samples described above. Normalizing to total processed sample weight does not consider the varying numbers of grains analyzed between the di fferent size fractions studied, which can range between tens of thousands of grains on 185–250 μm sample mounts to hundreds of thousands on <64 μm sample mounts. Normalizing to indicator grains/1000 grains allows for comparison of abundance values between samples, and between di fferent size fractions of the same sample.

**Figure 5.** A,B conceptual view of the basal surface of epoxy grain mounts: (**A**) three large heavy (black) mineral grains; (**B**) six small heavy (black) mineral grains. The area (%) values for heavy grains on mounts A and B are identical even though mount A contains half as many grains as mount B. C,D conceptual cross section view through epoxy grain mounts: (**C**) grain mount containing a greater total number of grains (30) than mount D, with four black grains out of a total of seven grains exposed on the basal surface; (**D**) grain mount containing a smaller total number of grains (16) than mount C with three black grains out of a total of six grains exposed on the basal surface The actual abundance of heavy (black) grains per total grains mounted in epoxy is lower in mount C (5 out of 30 grains) than in mount D, but due to the preferential settling of heavier minerals through the curing epoxy, the displayed abundance on the polished basal surface is higher in mount C (71%) than D (50%).

The indicator mineral abundances for each indicator mineral for each slab 2 are reported in two different ways (Tables 3 and 4): (1) number of indicator grains normalized to 1000 grains, and (2) area percentage.

Minerals in the four size fractions of the four till samples are dominated by calcic amphibole, ilmenite, and almandine. Minerals present in lesser abundance include epidote, chlorite, biotite, titanite, rutile, and assorted Fe-oxide minerals (Figure 6). Quartz and albite, both less dense than the 3.2 SG heavy liquid used to process the original till samples, were found in high abundances in all samples, primarily as parts of composite particles containing other, denser minerals. Abundances increased with decreasing particle size for the following minerals: ilmenite, almandine, epidote, rutile, titanite, apatite, zircon, and monazite. Other minerals decreased in abundance with decreasing particle size: quartz, chlorite, and biotite.


**Table 3.** Abundance of ore indicator minerals in four till sample up- and down ice of the Izok Lake volcanogenic massive sulfide (VMS) deposit. Mineral abundance data are reported in two ways: (1) number of indicator mineral grains per total grains, normalized to 1000 grains; (2) minerals by 2-dimensional percentage area they take up on the polished mount surface. Values for 0.25–0.50 μm material are from the heavy mineral fraction of the same sample obtained by visual grain picking


**Table 4.** Abundance of alteration indicator minerals in four till sample up- and down ice of the Izok Lake VMS deposit. Mineral abundance data are reported in two different ways: (1) number of indicator mineral grains per 1000 grains; (2) minerals by 2-dimensional percentage area they take up on the polished mount surface. Grey-shaded values are grain counts from the coarse (250 to 500 μm) heavy mineral fraction of the same sample obtained by visual grain picking that have been

#### *Minerals* **2020**, *10*, 310

**Figure 6.** Mineral abundance (normalized to 1000 grains) for the most abundant minerals comprising 50% of <250 μm till heavy mineral concentrates, for four size fractions of four till samples. Red line denotes the relative position of massive sulfide mineralization, and the black arrow denotes the direction of ice flow.

#### *3.4. Alteration Minerals and Metamorphic Equivalents*

Gahnite was present in all four till samples, including the sample up ice, with values ranging from 0.03 to 2.30 grains per 1000 grains. Sample 09-MPB-058, immediately down ice of mineralization, contained the most gahnite of the four samples (0.90 to 2.30). Gahnite was three times more abundant in the coarsest (185–250 μm) fraction of this sample as compared to the finer three fractions. Up ice, the <64 μm fraction had the highest abundance of gahnite (0.08), but significantly lower than all down ice samples.

Corundum was identified by Averill [3] as an indicator mineral of metamorphosed massive sulfide deposits and is found intergrown with gahnite grains in till samples down ice of the Izok Lake deposit. Corundum is present in all till samples, with values ranging from 0.14 to 3.59 per 1000 grains. The highest abundances by far (0.87 to 3.59) were found in the four fractions of sample 09-MPB-058, immediately down ice of mineralization. Of these four fractions for this sample, corundum is most abundant in the coarsest fraction. The fewest grains of corundum (0.14) were detected in the coarsest fraction of sample 09-MPB-060, 1 km up ice of the deposit. Corundum grains are intergrown with other minerals, or present as discrete grains.

Corundum grains intergrown with gahnite were identified in all samples down ice of mineralization. These intergrown grains were found in all four size fractions of sample 09-MPB-058, immediately down ice of mineralization, in the 64–125 μm fraction of intermediate distance sample 09-MPB-075 (3 km down ice), and in the 125–185 and <64 μm fractions of distal sample 12-MPB-902 (8 km down ice). Corundum was intergrown with other minerals, including hercynite (Fe-spinel), chlorite, staurolite, Fe-oxide minerals, hornblende and almandine. Corundum, both discrete grains and grains intergrown with gahnite, from proximal down ice sample 09-MPB-058, contained inclusions of the sulfide ore minerals galena and chalcopyrite.

Staurolite abundance ranged from 5.18 to 16.21 grains per 1000 grains. It was present in all four till samples, in all four size fractions. The highest abundance of staurolite (16.21) was detected in sample 09-MPB-058, just down ice of mineralization. The finest fraction of all four till samples consistently had the highest values of staurolite. In general, content in till decreased as distance down ice increased.

Axinite abundance ranged from 0.03 to 0.33 grains per 1000 grains and was identified in all four fractions of all four till samples. The 64–125 μm fraction of sample 09-MPB-075, 3 km down ice of mineralization, contained the most (0.33) axinite. The least amount of axinite was detected in the 125–185 μm fraction of sample 12-MPB-902, 8 km down ice. There were no consistent patterns of abundance related to size fractions for the four till samples.

Epidote abundance ranged from 14.4 to 61.71 grains per 1000 grains and was identified in all four size fractions of all four till samples. The <64 μm fraction of till sample 09-MPB-058, 0.5 km down ice of mineralization, contained the most (61.71) epidote. The least amount was detected in the coarsest (185–250 μm) fraction of distal down ice till sample 12-MPB-902 (8 km down ice). The greatest abundance of epidote grains is in the 64–125 or <64 μm fraction of all till samples.

Fe-oxide minerals were identified in all size fractions of all till samples from Izok Lake, with a notable increase in abundance in the coarsest (185–250 μm) fraction in the till samples down ice of mineralization. Fe-oxide mineral abundance ranged from 15.65 to 72.16 grains per 1000 grains and was identified in all four size fractions of all four till samples. The 185–250 μm fraction of till sample 12-MPB-902, 8 km down ice of mineralization, contained the most (72.16) Fe-oxide. The least amount was detected in the 125–185 μm fraction of till sample 09-MPB-075, located 3 km down ice of mineralization.

#### *3.5. Ore Minerals*

Chalcopyrite abundance ranges from 0.02 to 0.40 grains per 1000 grains and was identified in all four till samples. It is most abundant (0.40) in sample 09-MPB-058, just down ice of mineralization, in which it is present in all four size fractions. In this sample, it is most abundant in the coarsest fraction and least abundant in the finest fraction. In fact, chalcopyrite was found in all four fractions (except for the 64–125 μm of one sample) of the three till samples down ice of mineralization. In sample 09-MPB-060, 1 km up ice of the deposit, chalcopyrite was only identified in very low abundance (0.02) in the finest fraction (<64 μm).

Sphalerite is the least abundant of the four ore minerals reported here. It was present only in two till samples, with values ranging from 0.03 to 8.09 grains per 1000 grains. It was found in all four size fractions of sample 09-MPB-058 and in two size fractions (125–185 μm and <64 μm) of sample 12-MPB-902, 8 km down ice. In sample 09-MPB-058, it is most abundant in the coarsest fraction and least abundant in the finest fraction. No sphalerite was detected in the till sample up ice.

The abundance of galena in till samples varies from 0.03 to 0.40 grains per 1000 grains. It is most abundant in sample 09-MPB-058, with the highest abundance in the coarsest (185–250 μm) fraction. For each till sample, it is most abundant in the coarsest size fraction. It was also detected in minor amounts in the coarsest fraction of up ice sample 09-MPB-060.

Pyrrhotite abundance ranges from 0.07 to 1.00 grains per 1000 grains. It is most abundant in sample 09-MPB-058, with the highest abundance in the 125–185 μm fraction. It is identified in all size fractions of all till samples, with the highest abundances in the coarsest size fractions observed (185–250 and 125–185 μm).

Pyrite was found in all four size fractions of all four samples. Values vary between 0.06 and 2.50 grains per 1000 grains. Abundance was highest in sample 09-MPB-058, with pyrite most abundant (2.50) in the coarsest size fraction and least abundant (0.25) in the finest size fraction. Pyrite abundance in till samples 09-MPB-075 and 12-MPB-902, 3 and 8 km down ice of mineralization respectively, is lower than in till sample 09-MPB-060, 1 km up ice of mineralization.

Sulfide ore minerals are present in the sample as both discrete, liberated grains and as components in other composite mineral particles. The proportion (area%) of each ore mineral contained as liberated grains or as part of composite particles is contained in Table 5.


**Table 5.** The proportion (area %) of each ore mineral present as either liberated grains or as a component in a composite particle.
