The Scanning Electron Microscope (SEM) can be applied in different ways to perform an element mapping of the sample to visualize the distribution of individual elements across the sample. Traditional element mapping is performed with the software provided together with the energy dispersive spectroscopy (EDS) detector on the SEM, where an energy window in the analysed spectrum is selected in the software and screened for X-rays with the relevant energy. Results are typically provided as relative intensities, compared to high values within the sample. This makes it hard to compare between samples, especially if they do not both have the same composition. A second disadvantage of this technique is that several elements may have peaks in the same energy window. The rise of automated quantitative mineralogy (AQM) analysis techniques allowed for element mapping by the matching of spectra against a database (e.g., Gottlieb et al., 2000; Gäbler et al., 2011). There, reference spectra with small variations in composition could be included in the databases, which omitted the problem of peak overlap created by the energy windows. However, there will still be an uncertainty in the precise concentrations of the minor elements in the spectrum. Here, we applied the ZEISS Mineralogic software platform for the AQM, which applies a full spectrum deconvolution for every analytical point by applying the single spectrum mode of the EDS software. Therefore, semi-quantitative results are available in weight (wt)% concentrations for every pixel in the generated mineral map. This presents a hitherto unused potential for precise element mapping on single minerals in a mixed aggregate.

Samples in this study are derived from cores into partially depleted oil fields in Siri Canyon in the Danish North Sea, which are currently under consideration as sites for CO₂ storage. The oil fields are in glauconitic sandstones, a reservoir type, which has not previously been tested for CO₂ storage. Since glauconitic clasts comprise 20–30 vol% of the sandstones (Weibel et al., 2010; Kazerouni et al., 2012), they may potentially have large influence on the CO₂ reservoir stability. Glauconitic clasts occur with substantial variations, both in coloration, morphologically, structurally, and chemically (e.g., Odin and Matter 1981; Amorosi 1997; Kelley and Webb 1999; Meunier and El Albani 2007; Rudmin et al., 2017). Therefore, the AQM analysis has been applied to map compositional variations.

Glauconite is an iron (Fe)-rich, potassium (K)-poor mica that can be interstratified with minerals of the smectite-group. Glauconitic clasts form by glauconitisation of different particles in the shallow marine environment, upper slope and the outer shelf (50–500 m water depth; e.g., Odin and Matter 1981; Van Houten and Purucker 1984; Kelley and Webb 1999; Meunier and El Albani 2007; Rudmin et al., 2017). Various parent minerals have been identified, such as degraded micaceous clay minerals, carbonate clasts, argillaceous/fecal pellets, bioclasts, feldspar, rock fragments (e.g., Odin and Matter 1981; Sánchez-Navas et al., 2008; Banerjee et al., 2016; Tribovillard et al., 2021; Tounekti et al., 2021). Glauconitic clasts are known for their greenish colour and formed as pellets or by weathering of biotite or other 2:1-stratified micaeous minerals in shallow marine environment, near a coast with relatively warm water and at reducing conditions (McRae 1972). Compositional differences correlate with the different ratios of illite and smectite layers in the glauconitic clasts, which can be determined by X-ray diffraction techniques as an average for all glauconitic clasts in the sample (McRae 1972). However, individual glauconitic clasts can show considerable variation within a single sample and individual glauconitic clasts may show a different composition in the rim and the core of the clast (Velde and Odin 1975).

Elemental variations in glauconitic clasts are used in provenance and to explain processes active in the shallow marine environment during their formation (e.g., Meunier and El Albani 2007; Baldermann et al., 2017) but also in diagenetic studies to explain changes in composition with depth (e.g. Kazerouni et al., 2012; Bansal et al., 2017). Both the formation processes and the diagenetic alterations include the same set of elements, which might be problematic for the interpretation of the glauconitic bulk chemical data. Hence, the aim of this contribution is two-fold: Firstly, we present a new method for mineral-specific element mapping. Secondly, we apply this new method in a case study on glauconitic clasts, where we try to resolve different processes acting on these clasts by element mapping with the aim to differentiate between provenance and formation-related processes versus diagenetic changes.

AQM of glauconitic sandstones were performed on the SEM at the Geological Survey of Denmark and Greenland (GEUS), using the ZEISS Mineralogic software platform. Analyses were performed with acceleration voltages of 12–15 kV, a 120 μm² aperture. Analysis time per spectrum is set such that each spectrum contains at least 2,500 counts. For AQM, the backscattered electron contrast (BSE) is applied to display the contrast between different mineral phases in the sample. During AQM, a mosaic of BSE frames of a representative part of the sample is compiled. Each frame in the mosaic is thereafter analysed with the EDS detectors applying a user-defined step-size of 0.5–20 µm between EDS analytical spots. The chemistry in each spot is interpreted as a mineral species, which is forming one pixel in the false-coloured mineral map. Oxygen is quantified as one of the elements in the spectrum and resulting quantities are normalized to 100%. The classification of minerals is operator-based, where obtained mineral compositions are compared to known compositions in Deer et al. (1985) and the online database www.webminerals.com. Further details on the software and applied method can be found in Keulen et al. (2020). The precision of the analyses in this study were tested by repeating analyses on the same samples, where the concentration of the major elements could be reproduced within 1 wt%.

Most thin sections have been investigated several times at different magnifications and step-sizes to obtain both an overview over the bulk mineralogy, grain size distribution and the range of glauconitic compositions at low magnifications, and on the internal structure of glauconitic clasts at higher magnifications applying smaller step sizes. For the high-magnification maps created with a step-size of 0.5 µm, analyses were performed at 12 kV, to ensure a small interaction volume for the incoming electron beam with the analysed thin section and to avoid a mixed EDS signal from neighbouring clasts. Graham and Keulen (2019) showed that circa 50% of the EDS signal comes from an area less than 250 nm in diameter when applying 10 kV. By applying the same technique, we can show that for glauconitic clasts analysed at 12 kV, 80% of the EDS signal comes from a 0.5 µm-wide interaction volume (see Supplementary Figure S1).

The Mineralogic software has been applied to create element maps of the samples for individual minerals in the sample, while the other minerals are blanked. This is possible, because the mineral maps created with ZEISS Mineralogic are based on EDS spectrum deconvolution for every single pixel, including matrix quantification (Phi-Rho-Z or ZAF), omission of escape peaks and double peaks, like for a single spot EDS measurement. Therefore, quantitatively determined element wt% concentrations are available for each pixel in the mineral map (see also Keulen et al., 2020 for further description; Figure 1). A new mineral list was created for the false-colour mineral map, in which all minerals, except for the mineral of interest, are now coloured white, and the target mineral is coloured by concentrations of specific elements in the mineral, e.g., the Fe concentration in glauconite (see Figures 2 and Supplementary Figure S2, Figure 3). In this way, mineral-specific element maps can be created, with absolute element concentrations, and these concentrations can be used to draw diagrams (Figure 4).

We have applied this technique on a set of 22 thin sections of glauconitic sands from the Siri Canyon in the Danish North Sea Basin (see Figure 1). Siri Canyon is characterized by Palaeocene hemipelagic mudstone and marls, intercalated with glauconitic sand deposited by mass flow and turbidites (Hamberg et al., 2005; Schiøler et al., 2007). Remobilisation of the sand after deposition resulted in injectites in both stratigraphically younger and older strata (Svendsen et al., 2010). The samples were collected from 13 cored wells with varying present-day burial depths of 1,694–3019 m depth and represent four different gravity flows, Bor, Tyr, Kolga members and Frigg sand (Figure 1). Note, that Frigg sand is an informal denotation of presumably Kolga Member, which is injected into a stratigraphically higher level, i.e., the Horda Formation. Samples were collected from the center of the glauconitic sand intervals. Only a few gravity flows were present in each well. The glauconitic sands are homogeneous in composition, with circa 50% quartz, around 20–30% glauconitic minerals in the shallow part of the Siri Canyon (Weibel et al., 2010). The remaining minerals are 10% feldspars, 0–7% carbonates (mainly siderite) and very minor muscovite, biotite, other clays, metamorphic minerals, accessory minerals, rock salt and pyrite. The samples are remarkably homogeneous in composition and grain size (see also Hamberg et al., 2005; Friis et al., 2007). The gravity flows originate from the Stavanger platform and are deposited as dipping members. The individual drill cores reached the same members at different burial depths, thus the effect of provenance versus alteration at higher depth/temperature could be investigated within the glauconitic gravity flows.

The variation in glauconitic clast composition has been studied from element maps for iron (Fe), aluminium (Al), magnesium (Mg), potassium (K), and silicon (Si). In Figures 2; Supplementary Figure S2 the element maps have been sorted by the individual members of the Siri Canyon. Most glauconitic clasts, irrespective of their member, drill core or depth have a core and a rim. Many have internal fractures along which a change in element composition has been observed. In most samples, the glauconitic clasts nearly all have a similar composition, within a few wt% margin. However, some samples, especially those at shallow depth, show a very wide range of compositions, for example Vi1_1731.40 (uppermost Tyr Member) in Figures 2; Supplementary Figure S2. Glauconitic grains that still show a mica-cleavage usually are more Al-rich and more depleted in Fe-K than surrounding glauconitic clasts (see e.g. Figures 3A,B and Supplementary Figure S2).

Here, we shortly discuss some trends of the element distribution with depth, between different members and between drill cores.

Different processes seem to play a role in the element exchange in glauconitic clasts and each of the studied elements show their own pattern. Three different processes affect the compositions: 1) variations with depth, 2) core versus rim differences and 3) provenance variations. These will be discussed separately.