Heavy mineral analysis by ICP-AES a tool to aid sediment provenancing

Correlation and provenancing of sediments/sedimentary rocks can be achieved by several techniques; a common approach is through the identi ﬁ cation and quanti ﬁ cation of heavy minerals using a petrological microscope. This can be time consuming, the analysis of heavy minerals by inductively coupled plasma atomic emission spec- troscopy o ﬀ ers a faster alternative, by determining key elements associated with speci ﬁ c heavy minerals. Here we outline a method for determining heavy mineral species though ICP-AES using high temperature fusion with a lithium metaborate ﬂ ux to ensure complete dissolution of resistate minerals. The method was tested in a provenance study of desert sands from the United Arab Emirates. The results are compared with those derived from traditional optical microscopy. These show good agreement for minerals with speci ﬁ c geochemical signatures, whilst the overall geochemistry of the heavy mineral concentrate was diagnostic of potential sediment sources. This geochemical approach is capable of processing large numbers of samples rapidly and is advocated as a screening technique. A combination of geochemical and mineralogical data produced by these techniques provides a powerful diagnostic tool for studies of heavy mineral signatures in sediments frequently used in mineral reconnaissance, paleogeographic reconstruction and reservoir characterisation in the petroleum in- dustry.


Introduction
Understanding the origin or provenance of sediments/sedimentary rocks is important in the reconstruction of sediment pathways and paleogeography (Garzanti et al., 2007 andWeltje andEynatten, 2004), as a tool for mineral reconnaissance (Morton and Berge, 1995), or oil and gas reservoir characterisation (Averill, 2001). Heavy mineral provenancing is traditionally carried out through the identification and quantification of heavy mineral concentrates using a petrological microscope (Morton and Hallsworth, 1993). The main advantage of analysing the heavy mineral fraction is to reduce the dilution from minerals such as quartz, feldspars, carbonate and clays that can typically constitute up to 99% of the sediment and are not generally provenance specific (Morton and Hallsworth, 1999).
Optical microscopy provides information on grain morphology, physico-chemical alteration and the accurate identification of individual heavy mineral species. Such detailed assessment of mineral concentrates can aid in assessing whether the heavy mineral component of a sediment has been altered through selective dissolution, diagenesis or preferential sorting during transportation (Andò et al., 2012). The disadvantages, however, include the difficulty in identifying opaque minerals, the potential for human error and the time associated in undertaking a study with numerous samples. An experienced analyst can analyse an average of three samples per day, involving the identification of 300-500 non-opaque heavy minerals per sample (Morton and Hallsworth, 1993). Minerals excluded from the count include opaque minerals (e.g. magnetite, ilmenite), minerals that are potentially unrepresentative (carbonates, authigenic minerals) or minerals whose specific density is affected by inclusions of another mineral of greater density (magnetite, ilmenite, etc.) than that of the host mineral (Hounslow and Morton, 2004).
Optical identification of the heavy minerals is typically achieved using a standard ribbon counting method (Galehouse, 1971). Provenance is determined by focusing on specific pairs of minerals with similar shape, density and resistance to alteration to create a mineral index, e.g. rutile zircon index [RZi] or chrome-spinel zircon [CZi]. One or more mineral indices are chosen to differentiate sediment sources Hallsworth, 1993 andGarzanti et al., 2013). Heavy minerals can be further divided into those that are inherently resistant to weathering, dissolution and transportation (e.g. zircon, monazite, tourmaline and rutile) and those that are less resistant (typically olivine, pyroxene and amphibole) (Garzanti et al., 2013).
Some of the earliest provenance based studies employed mineralchemical associations using atomic absorption spectrometry (AAS) (Luepke, 1980). For this technique, large volumes of sediment were required to obtain sufficient heavy mineral concentrate for analysis and the data interpretation was limited by the relatively poor detection limits of the AAS technique. Recent provenance studies based on heavy minerals still rely on optical microscopy supported by whole-rock chemical data obtained by inductively coupled plasma-atomic emission spectroscopy (ICP-AES), inductively coupled plasma-mass spectroscopy (ICP-MS), qualitative evaluation of materials by scanning electron microscope (QEM*SEM) or X-ray diffraction (XRD) on bulk samples (Pearce et al., 2010 andPe-Piper et al., 2016). Raman spectrometry is also used to aid in the identification of single heavy minerals (Worobiec et al., 2007). Current ICP technology, combined with appropriate dissolution procedure, is capable of producing geochemical data for a large number of heavy mineral separates rapidly due to dual-view plasma optics and better optical resolution. Combined with a better understanding of provenance indicators, the associated geochemistry can be employed as a prescreening technique or as a complimentary method which can be used concurrently with optical microscopy.
A wide selection of sample dissolution procedures have been employed for geochemical analysis by ICP-AES (Miles and Cook, 2007). These usually involve decomposition with mineral acids either with or without hydrofluoric acid; or fusion with a flux such as lithium metaborate or alkali metal carbonates, oxides or hydroxides. In general, fusion techniques are more effective at dissolving resistant minerals, such as zircons, than acid digestions. Poor recoveries of Zr from granites, rhyolites, andesites and low Cr recoveries from harzburgites are typically evident of incomplete digestions (Lapworth et al., 2012).
Analysis of heavy mineral concentrates by ICP-AES following an appropriate digestion offers a novel approach to provenance studies which is inherently faster than traditional microscopy. A throughput of at least 30 samples per day can be achieved by ICP-AES, including sample fusion and subsequent dissolution, representing a ten-fold increase in productivity compared to analysis by optical microscopy. Fandrich et al. (2007), also acknowledged the time consuming nature of ore characterisation using optical microscopy compared to QEM*SEM analysis.
Where the elemental composition of a heavy mineral concentrate is not diagnostic of a particular mineral phase, its geochemical signature can, in certain cases, still be diagnostic of a specific geological terrain. For example, a mafic or ophiolitic terrain would tend to be enriched in Cr, Fe, Ni, Ti, and Mg (Wallace et al., 2015) whilst a granitic source would be dominated by Ca, P, Ti and Zr (Arslan and Aslan, 2006) and a high grade metamorphic terrain would be represented by increased Al, Ca, Fe, and Mg (Sukhorukov, 2007). However, reconstructing sedimentary pathways are rarely this simple; geological terrains are often complex, constructed of multiple lithologies, and/or sedimentary basins could be infilled from several sources, over a period of time. In this instance, the geochemistry may not necessarily be diagnostic of a specific geological terrain but may still yield valuable data.
In this paper we demonstrate the validity of a geochemical approach to heavy mineral provenance studies, by comparing the interpretation of geochemical data with results from the traditional optical microscopy. Data was acquired through a provenance study of the Rub' al Khali desert sands of the United Arab Emirates (UAE) (Fig. 1).

Study location
The Rub' al Khali (Arabic for 'empty quarter') desert of the UAE is one of the largest free-moving bodies of sand on Earth Abdullah, 2011 andGlennie andSinghvi, 2002). It was formed in a Neogene elongate basin at the southern end of the Arabian Peninsula and is characterised by large sand dunes up to 250 m in height Abdullah, 2011 andGlennie andSinghvi, 2002). This enormous body of sand was previously interpreted as being derived from several sources including the Arabian Shield, Hajar Mountains and Zagros Mountains (Garzanti et al., 2013, Garzanti et al., 2003, Crouvi et al., 2010and Evans and Carter, 2002. Geological mapping of the UAE (Farrant et al., 2012) indicate four main sediment facies along the northern upwind margin of the Rub' al Khali desert varying in age from Miocene to Quaternary. These include: 1. Miocene age fluvial sandstones exposed along the Arabian Gulf coast derived from the Arabian shield; 2. Quaternary palaeodune sandstones of the Ghayathi Formation derived from sediments washed into the Arabian Gulf; 3. Quaternary fluvial fanglomerates and sandstones emanating out from the Hajar Mountains (Hili Formation); and 4. Quaternary to recent aeolian dune sediments representing a variable mix of sediment sources, blown inland by the northwesterly Shamal winds Abdullah, 2011 andEdgell, 2006) (Fig. 1). The detrital heavy-mineral assemblages for the studied samples are dominated by ferromagnesian minerals (average 36%); including calcic amphibole (25.8%), clinopyroxene (6.8%) and orthopyroxene (3.4%). The other major constituents are epidote (35.7%), garnet (12.6%). The remaining mineral phases include: apatite (1.1%), Cr-spinel (3.0%), rutile (1.2%), staurolite (0.5%), titanite (1.5%), tourmaline (3.1%) and zircon (4.3%). The remaining 1% of the averaged heavy mineral component includes trace quantities of the minerals andalusite, brookite, glaucophane, kyanite, monazite, sillimanite and blue-green spinel.

Reagents and standards
Samples and reference materials (RMs) were fused using lithium metaborate (LiBO 2 ) [Sigma-Aldrich: 99.9% trace metals basis] and sodium peroxide (Na 2 O 2 ) [Acros organics 96%]. All concentrated acids and hydrogen peroxide (H 2 O 2 ) used in the digestions were Romil-SpA™ super purity reagents. Deionised water (Milli-Q™) with a resistivity of 18.2 MΩ cm at 25°C was used for all dissolutions and dilutions.
A certified RM NIM-D (dunite) from the National Institute for Metallurgy, Republic of South Africa and the three in-house AHMCs were included in each batch of fusions. Aqueous ICP-AES calibration standards in 5% HNO 3 were prepared on the day of analysis from ROMIL PrimAg® single element standards. Three independent quality control (QC) standards in 5% HNO 3 prepared from Fisher Scientific single element ICP solutions were incorporated in each analytical run. Independent analytical QC solutions were prepared from a Perkin Elmer Pure multi-element standard (As, Ba, K, La, Mg, Mn, Ni, Sr and Zn) and SPEX CertiPrep® standards 1 (As, Ba, K, La, Mg, Mn, Ni, Sr and Zn) and 4 (Mo, P, S, Si, Ti and Zr).

Artificial heavy mineral concentrates
Three artificial heavy mineral concentrates (AHMCs) were created to provide a standard concentrate with a known mineral and chemical composition for the development of a robust fusion method and for use as quality control materials during routine analysis. They were created by mixing pure minerals in varying proportions (Table 1) to reflect the average assemblages of naturally occurring heavy mineral concentrates (HMCs), the chosen minerals are representative of the natural concentrates associated with those sampled in the UAE. The major and trace element composition of these AHMCs were determined by wavelength dispersive X-ray fluorescence spectrometry (XRFS). Fused beads (pre-ignited and prepared in a lithium tetraborate/metaborate flux) were analysed on a PANalytical Axios mAX WD-XRF spectrometer using Omnian interpretative software.
• A ramped fusion with Na 2 O 2 at 550°C followed by dissolution in HCl and HF acids (Watts and Chenery, 2010).
• Flash-fusion at 1050°C with LiBO 2 , followed by dissolution in a mixture of HNO 3 and HF acids.
The relative effectiveness of these sample dissolution methods is summarised in Table 2. This shows the percentage recovery for a range of key major and minor elements compared to the total concentration data obtained by XRFS. Each fusion or digestion was performed in triplicate. The very low recoveries for some elements, especially Cr and Zr (1 to 6%), by the HF-hot block method indicate an almost incomplete dissolution of the resistate minerals zircon and chrome-spinel. Fe concentration was greater than the upper calibration limit for the defined method, but due to the incomplete digestion of Cr and Zr phases, repeat at further dilution was not performed. Mean recoveries for Cr and Zr  using a Na 2 O 2 fusion were 74% and 86%, respectively. The efficiency of this fusion decreased with increasing zircon content (AHMC1 = 105%, AHMC2 = 81% and AHMC3 = 73% zircon). An anomalous 206% recovery of P for AHMC1 is due to a single high P result; due to the overall poor Cr and Zr values, this was disregarded. Fusion with LiBO 2 provided the most complete digestion and was identified as the most appropriate method for the dissolution of heavy minerals. During the main study, 37 AHMCs were analysed (see supplementary data for full AHMC data) during the analysis of the UAE heavy mineral concentrates and performance data is illustrated in Fig. 2.
Correlations between the percentage of mineral(s) within the AHMC's and the ICP-AES data can be used to construct a representative calibration curve. Fig. 3 shows four calibration curves, three of these calibrations display good correlations with corresponding R 2 values: Cr, R 2 = 0.99, P, R 2 = 0.96 and Zr, R 2 = 0.99, however, Ti has a poor calibration correlation (R 2 = 0.22). The AHMC's have the following percentages of major Ti bearing minerals (ilmenite, rutile and titanite): AHMC1 15%, AHMC2 21.4% and AHMC3 22.6%, however, AHMC1 has 20% augite (clinopyroxene), which, with a presumed Ti content of 2.03% would still greatly influence the Ti content of AHMC1.

Sample preparation
A total of 118 samples were selected for analysis, constituting of the four aforementioned sediments (1. Miocene age fluvial sandstones, 2. Ghayathi Formation, 3. Hili Formation and 4. Dune sands of Quaternary to recent age). Subsamples ranging from 100 to 200 g were separated, disaggregated with a pestle and mortar if required, and dispersed in deionised water using a reciprocal shaker overnight. The samples were then sieved to produce a 63-125 μm (fine sand) fraction and then dried overnight before separation of the heavy minerals using a lithium heteropolytungstate heavy liquid at a density of 2.85 g/cm 3 (LST-Fastfloat™). The average heavy mineral yield for the 63 to 125 μm fraction was 4%. A representative subsample was mounted on a glass side using Canada Balsm. An average of 600 transparent grains was counted per sample using a standard ribbon counting method (Galehouse, 1971).

Analytical method for ICP-AES determinations
A representative sub-sample of each HMC from the UAE was retained for optical microscopy and the remaining sample was micronised to reduce the particle size to < 20 μm. A 0.1 g portion of the milled concentrate was fused with 0.9 g LiBO 2 flux in a platinum crucible and flash-fused at 1050°C for 15 min in batches of 30, which included one fusion blank, one duplicate, two AHMCs and a CRM (NIM-D). When cool, the crucibles were placed into vessels containing 50 mL deionised water, 5 mL conc. HNO 3 and 1 mL conc. HF, capped and placed on a reciprocal shaker overnight. After 18 h, a further 44 mL deionised water was added to the vessels prior to analysis (see supplementary data for a full suite of elements for all 118 samples from the UAE).
To compensate for any potential enhancement of the analytical signal due to the increased ionisation potential of lithium ions in the plasma, calibration standards, blanks and QC solutions were matrixmatched using 10% 0.18 mol L − 1 LiBO 2 . Samples were diluted by a factor of 10, therefore the analysed molarity for both samples and standards was 0.018 mol L − 1 LiBO 2 . Analyte wavelengths were chosen to minimise any spectral interferences. Major elements were determined using a Perkin Elmer 7300 DV instrument in radial view mode to limit saturation of the segmented-array charge-coupled device detectors, whilst trace elements were measured in axial mode for higher sensitivity and better detection limits. Instrumental parameters specific to the analysis of HMCs by ICP-AES are given in Table 3.

Results
In order to compare the performance characteristics of elements with concentration ranges that varied over several orders of magnitude, the data were first normalised by setting the minimum and maximum concentrations of each element to 0 and 1 (Kumar-Jain and Bhandare, 2011). Mineralogical data were normalised using the same method to provide comparable data.
A direct relationship between the mineralogical and geochemical data is restricted to specific heavy minerals that possess a unique geochemical signature. The three main comparisons that can be constructed are Zr to zircon, Cr to chrome-spinel and Ti to rutile, anatase, brookite and titanite (ilmenite, another major Ti bearing mineral is excluded as opaque heavy minerals could not be identified) as shown in Fig. 4. Direct Cr, Ti and Zr comparisons for all samples can be found in supplementary data. A two-tailed non-parametric test, the Mann-Whitney U test (Wang and Yue, 2002), was used to determine whether there was any statistical difference between the mineralogical and geochemical data. In this test, the sample populations were assigned a Z-score based on the normalised data. If the magnitude of this score is greater than the Zcritical (1.96) then there is a 95% probability that the two populations (i.e. chemistry and mineralogy) are statistically different from one another. The calculated Z-scores were Ti = − 0.96, Cr = 3.48 and Zr = 4.43 (Fig. 4a-c). However, these values are somewhat misleading due to the provenance-specific geochemistry associated with each sample group. Therefore individual Z-scores were calculated for each of four sample groups in each element-mineral comparison (Table 4). Overall, most of the individual Z-scores indicated good agreement between the two datasets. However, the Z-scores for Zr (Miocene and Modern sand), Cr (Miocene) and Ti (Hili Formation) indicated that the chemical and mineralogical datasets were significantly different.
The poor Z-scores for Zr (Miocene: 3.52 and Dune sands: 2.68) are not necessarily reflected in Fig. 4a; the Miocene samples appear to have a greater degree of scatter whilst the Dune sands appear to plot along the 1:1 line. The sample populations associated with the Z-scores for Cr (Miocene: 6.01) and Ti (Hili: 3.02) appear to be discrete but biased towards the x-axis (ICP-AES) (Fig. 4b and c). The Z-scores for Cr (Ghayathi: − 1.90) and Ti (Ghayathi: 0.82) are both below the Z-critical (1.96) yet plot as discrete populations which appear biased towards the Y-axis (Mineral counts) (Fig. 4b and c).
The normalised geochemical and mineralogical data were also plotted on ternary diagrams using two groups of elements that can be diagnostic of provenance: Zr, Cr, Ti and Ca, Fe, Mg. Ternary plots associated with the more resistant heavy minerals ( Fig. 5a and b) were used to compare Zr, Cr and Ti with their associated minerals, i.e. zircon, Cr-spinel and transparent Ti minerals (anatase, brookite, rutile and titanite). In these plots the general population trends between the geochemical and mineralogical data are similar. Thus, the Miocene sediments are comparatively enriched in Zr and Ti, the Ghayathi and Hili formations contain more Cr than the Miocene sediments and the modern dune sands appear to cluster between the other three groups. The obvious difference between the two plots is the degree of scatter. In the geochemical plot (Fig. 5a) the populations are discrete and well defined whereas in Fig. 5b the groups are less well constrained. In contrast, the geochemistry data associated with the less resistant heavy minerals (Fig. 5c) are not as distinctive as their more resistant counterparts, which makes a direct comparison with the mineralogy difficult. Whilst the occurrence of some minerals can indicate particular source rocks (e.g. orthopyroxene is often derived from pyroxenites, peridotites and dunites, whereas the presence of clinopyroxene can indicate mafic igneous rocks) it is more difficult to associate differences in the geochemistry of the HMCs with specific rock types. However, it is still possible to infer particular geological terrains from the HMC  geochemistry.
The Hili Formation sediments are more ferromagnesian (Fig. 5c) and their mineralogy appear to be more dominated by amphibole and orthopyroxene (Fig. 5d). Most of the Miocene sediments trend towards a ferro-calcic signature (Fig. 4c); this is also reflected in their mineralogy, which is dominated by calc-amphibole and clinopyroxene (Fig. 5d).
The Ghayathi Fm geochemistry is distinctive and discrete, with consistent ratios of Mg, Ca and Fe (Fig. 5c); the associated mineralogy appears to be biased towards clinopyroxene, with subordinate calcamphibole (Fig. 5d). Most of the modern dune sands plot within the other sediment populations, with the exception of a small but discrete population of dune sands with an enriched Ca/Mg signature that is relatively depleted in Fe (Fig. 5c). This separate group of modern dune sands is not reflected in the associated mineralogy (Fig. 5d).

Discussion
Several factors are likely to contribute to potential differences between the mineralogical and geochemical methods. One is the shape and volume of the minerals, which is not taken into account during typical mineral counting. Thus a small, slender, prismatic zircon crystal (~49 mm 3 ) would be counted the same as a larger, equant, wellrounded crystal (~1020 mm 3 ) but the difference in volume (mass) would affect the related Zr concentration determined by ICP-AES. In this example there is a 95% difference in volume between a slender prismatic crystal and a larger spherical grain. This phenomenon has previously been noted by Ratcliffe et al. (2007).
Differences in crystal volume can also result from the partial dissolution of specific minerals. This effect is directly related to mineral stability, weathering and diagenesis (Andò et al., 2012) and is commonly observed in less resistant minerals such as pyroxenes and amphiboles. Selective dissolution in these minerals is typically observed as 'hacksaw'-terminated crystals (Fig. 6a-c). It is difficult to assess its impact on the related geochemistry. In this study, most of the sediments are still fairly immature due to the UAE's arid environment and there is little evidence to suggest that there has been a large degree of selective dissolution. Samples with evidence of 'hacksaw' dissolution were restricted to areas in close proximity to the Arabian Gulf.
Another difference that could influence a comparison between mineralogy and geochemistry is the fact that some minerals are excluded from the mineral counts; these include opaque minerals, authigenic minerals and carbonates such as the detrital dolomite shown in Fig. 6d. This dolomitic component is described further by Lacinska et al. (2014) and also accounts for the discrete population (enriched in Ca/Mg) of dune sands described in Fig. 5c. These minerals are included in the HMC geochemistry data and therefore care must be taken when interpreting differences between the datasets.
A final difference between the two methods is related to the stoichiometry of some minerals, which can differ depending on their mode of formation (Aqrawi andEvans, 1994 andAl-Juboury et al., 2012). This particularly applies to chrome-spinel, which would affect the Cr concentration determined (Thayer, 1956) and thus the Chrome:Crspinel ratio. In addition, the colour and optical-transparency of Crspinels can potentially bias results. Cr-spinels can be identified optically as reddish-brown, dusky brown or dusky red to opaque (optically isotropic) (Thayer, 1956 andBaumgartner, 1985), the redness of Cr-spinel has been attributed to its relative chrome content (Thayer, 1956). Whilst the variations of colour and relative transparency can be accounted for optically, a suite of opaque (optically-isotropic) Cr-spinels Fig. 4. Comparison of normalised mineral counts and element concentrations from UAE HMC. a, Zr:zircon. b, Cr:chrome-spinel. c, Ti:anatase, brookite, rutile, titanite. could not and would therefore be excluded from the mineral count. And as previously mentioned (Fig. 3), a minor chemical component of a heavy mineral (e.g. Ti in Augite) may preferentially increase the overall Ti signature of the sample if a high concentration of the mineral exists.
These potential biases can assist the interpretation of data from the two techniques. For example, the greater spread of Zr concentrations in the Miocene sediments (Fig. 4a) could infer a difference in zircon grain morphology. The Cr relationship in Fig. 4b suggests that the Cr content  of the Ghayathi Formation spinels is generally lower than those associated with either the Miocene sediments or the Hili Formation. This does not, however, preclude the possibility that some Cr-spinels may have been missed during the mineral counting. The Ti comparison (Fig. 4c) appears to be the most scattered but this cannot be associated with any single factor such as grain volume, the inclusion of Ti bearing opaque minerals or variable stoichiometries, as these could all contribute to the overall deviation of data from an exact correlation.
As previously mentioned, a heavy mineral index can be used to aid the determination of sediment provenance. In this case study, the chrome-spinel:zircon index (CZi) and its corresponding geochemical index (CrZri) are the most appropriate. A direct comparison of both indices (Fig. 7) shows a positive correlation with an R 2 is of 0.63. The Mann Whitney U test was used to further demonstrate that the two datasets are comparable within a 95% confidence limit. This test produced a Z-score of 0.32, well below the Z-critical value of 1.96, showing that there is no statistical difference between the CZi and CrZri indices. Once again the Miocene sediments display the greatest degree of scatter, which is attributed to the disparity between the different volumes of zircons.
Indices based on mineralogical and geochemical data for each sediment type were calculated (100 × Cr / [Cr + Zr])and subdivided into three categories, low (< 33), medium (33-66) and high (> 66)a high index is indicative of a Cr-rich, Zr-poor signature. Fig. 8a (CZi) and Fig. 8b (CrZri) display these categories with their associated locations within the UAE. The mineralogical and geochemical maps indicate similar trends, with higher CZi and CrZri indices for the Ghayathi and Hili formations and modern dune sands in the east of the UAE.

Conclusion
This study demonstrates the validity of a novel geochemical approach to the heavy mineral provenancing of sediments. A procedure specifically tailored to the analysis of heavy mineral concentrates by ICP-AES was developed and applied to desert sands from the United Arab Emirates. Some of the many variables associated with either analytical method were considered including grain size, mineral dissolution, the occurrence of opaque or non-detrital heavy minerals and an assumed mineral stoichiometry. In spite of this complexity, an index derived from the geochemistry of heavy mineral concentrates showed remarkably good agreement an index based purely on the optical identification of these minerals. Further interpretation and manipulation of this data is will be presented in a separate paper (Farrant et al., 2017).
The advantages of this geochemical approach include: • Faster analysis time for large numbers of samples. This method represents a ten-fold increase in productivity compared to analysis by optical microscopy without compromising the final interpretation.
• The quantification of REE. Whilst not used in this particular case study, REE data can still be useful for particular studies.
• Geochemistry of the HMC does not discriminate between transparent and opaque heavy minerals. In particular chrome-spinel, which can be particularly difficult to identify optically.
Optical identification of HMC is still required for definitive identification of heavy minerals which lack a unique geochemical signature and for the observation of physical and chemical surface alteration. Although this geochemical method may not eliminate the need for the identification of heavy minerals by optical microscopy entirely, it can be employed as a powerful screening tool to facilitate larger scale provenance studies used in applications such as mineral reconnaissance and petroleum exploration. the identification of heavy minerals through optical microscopy and their application to sediment provenance studies. Tom Bide assisted with the collection of samples. This paper is published with the approval of the Executive Director, British Geological Survey (NERC). I would also like to thank the reviewers of this work, who helped to improve the final manuscript: Sergio Andò and anonymous.

Appendix A. Supplementary data
Science impact statement 1. Analysis of heavy minerals by ICP-AES can determine key elements associated with specific heavy minerals. 2. Capable of processing large numbers of samples rapidly as a screening technique. 3. Used in defining heavy mineral signatures in key sediments associated with conventional and unconventional energy, mineral reconnaissance and paleogeographic reconstruction.