Dedicated, Comprehensive and Fast Mineral Analysis with AZtecGeo

Electron microscopists and geoscientists have worked together to refine workflows for efficient whole-sample mineralogical characterisation. This collaboration led to the adaptation of automated workflows, in which geological samples were characterised by EDS on a pixel-by-pixel basis, generally referred to using the umbrella term ‘automated mineralogy’. 

Outside the geosciences, Feature/particle analysis approaches typically involve segmentation performed on BSE images (their inherent contrast mechanism can be treated as a proxy for compositional variations) by specifying grayscale thresholds. Particles that meet specific greyscale thresholds are then analysed for their composition using EDS. This approach has several advantages in comparison with pixel-by-pixel approaches:

• Data Quality: the composition of a mineral can be derived from one single EDS spectrum collected across its full area. Thus, 10s of thousands of counts per spectrum can be collected in a fraction of a second. Compared to pixel-by-pixel approaches, many individual spectra with only a couple of thousand counts per spectrum (at most – often far fewer) would be collected across the same area. With its improved counting statistics, a BSE guided approach can achieve improved quantification accuracy, repeatability, and lower detection limits.
• Spectrum Processing: With a relatively high-count spectrum, AZtecLive’s spectrum processing engine (TruQ^® IQ) is applied to greater effect on every single particle. Therefore, it is not necessary to pre-define elements to be analysed – instead allowing them to be identified and quantified automatically – critical when key phases and elements could be found at any time within > 100,000 different particles.
• Speed; High throughput with up to 4 large area Ultim Max ∞ EDS detectors, achieving acquisition rates of up to 120,000 features per hour – even with a single detector.

Figure 1- A typical example of the mineral abundance and composition data particle analysis can produce in seconds. (A) represents isolated particle, its EDS composition and subsequent classification.

The applicability of feature analysis for geology is vast. With rock texture and mineralogy being fundamental characteristics for many geological interpretations or industrial processes, this analytical technique holds a high value. Though originally developed for the field of mineral processing, automated mineral analysis has been adapted for a variety of functions including: petroleum reservoir assessment, CO2 sequestration, mineral exploration, petrology and accessory mineral study.

Note: we will refrain from using the term ‘automated mineralogy’ due to the strong connections with mineral processing applications and liberation/association statistics. For automated mineralogy software with such functionality, see AZtecMineral.

Introducing AZtecGeo- Quantify the Chemistry and Morphology of Up to 120,000 Particles Per Hour

AZtecGeo is a specialized user profile within our AZtecFeature particle analysis platform which is specifically designed to support geoscientists. It includes a range of distinct features tailored to streamline analytical setup and enable rapid mineral classification, making complex analyses more efficient and accessible.

Built-in mineral classification scheme – Gain access to a comprehensive classification library featuring over 200 common rock-forming minerals. This built-in resource greatly reduces the knowledge and time required for fast and accurate classification of EDS data.

Preset image and EDS acquisition settings - Optimize data collection with pre-configured settings designed for speed and quality. This includes BSE imaging and EDS acquisition parameters such as; image resolution, dwell time, spectrum counts, and process time.

Guidance and support – Access detailed step notes at each stage of the workflow for clear explanation and guidance.

Beyond the advantages of the dedicated user profile, AZtecGeo leverages the power of AZtecFeature to deliver fast, accurate, and intelligent particle analysis. This proven software automates particle detection, morphology measurement, compositional analysis, and classification. Designed for efficiency, it provides reliable results while achieving analysis speeds of up to 120,000 particles per hour, all within an intuitive and highly customisable workflow.

AZtecGeo in Practice

To demonstrate the power of combining AZtecGeo with AZtecFeature, three analyses were performed reflecting common academic and commercial requirements for automated mineral analysis. All data was collected using a field emission SEM equipped with a single Ultim Max ∞ 100 mm² EDS detector. All analyses were performed within the AZtec 6.2 software platform enabled with access to AZtecFeature and AZtecGeo.

Quantifying Mineral Abundance

As previously discussed, the mining industry has been a major driver behind the development of geological applications of automated particle analysis. In this example, a particulated Zn-ore sample was embedded in epoxy resin, polished to expose individual particles, and coated with a 30 nm carbon layer to ensure electrical conductivity during analysis. 

Application Objective: The goal of this analysis is to analyse and identify all of the mineral grains; quantifying their composition and relative abundances. This information is critical for assessing the distribution of economically valuable and/or deleterious phases within an orebody, which will have a crucial control on its economic viability. 

Challenges: Ore deposits exhibit significant geospatial variability, often containing a wide range of mineral phases. In this sample, thousands of individual grains are present, ranging from single-phase minerals to complex multi-phase aggregates. A robust mineral classification scheme is essential—not only for assessing the content of valuable commodities (e.g., Zn) but also for identifying minerals that contain penalty elements or recoverable by-products. Additionally, effective image segmentation is crucial to accurately distinguish rock fragments composed of multiple mineral phases. 

Our Solution: AZtecFeature provides an optimal platform for segmenting BSE images with precision. Its intuitive threshold selection tools enable users to exclude epoxy, accurately delineate composite grains into discrete features, and focus EDS analysis where it is needed.

Figure 2 - Flow diagram displaying the different stages of particle analysis completed using AZtecGeo. Rapid mineral abundance quantification can be determined with ease.

A minimum of 20,000 counts per feature were acquired and processed for over 6000 mineral grains within 30 minutes. Classification results and mineral abundance data from AZtecGeo were instantly extracted and visualized using simple pie charts in Excel. 

The analysis confirmed a significant presence of sphalerite (Zn ore mineral), with minor galena (Pb ore mineral) as a by-product. The primary gangue minerals—carbonates, quartz, feldspar, and pyrite—account for approximately 85% of the sample. When applied across multiple samples from an orebody, this approach enables real-time monitoring of ore grade fluctuations and gangue mineralogy, providing valuable geometallurgical insights.

Locating Sparse Mineral Phases

From a geological perspective, not all minerals hold equal significance. Geoscientists often focus on a select group of compositionally variable, well-characterised mineral phases that provide insights into specific geological processes or physiochemical conditions. Zircon, for instance, is widely used for U-Pb geochronology and O¹⁸ oxybarometry, offering a reliable method for dating rocks and understanding the activity of oxygen during their formation. 

In this example, AZtecGeo was employed to identify key accessory phases—zircon, apatite, and monazite—within a heterogeneous till sample. The till fragments were mounted in conductive resin, polished, and coated with a 30 nm carbon layer to ensure electrical conductivity during analysis.

Application Objective: Till is a glacially derived material that forms as glaciers erode, transport, and deposit rock debris over vast regions. As a result, till contains a diverse assemblage of mineral fragments sourced from underlying bedrock. Crucially, the presence of specific accessory minerals within till can serve as indicators of concealed mineral resources. Additionally, these minerals provide insight into sediment provenance and the movement of glaciers during past glaciations.

Challenges: By definition, accessory minerals are rare, can be exceptionally small, and inconsistently distributed within a sample, making their identification particularly challenging. This is especially true in till, where an unusually high mineralogical diversity can obscure their presence. Once located, it is essential to pinpoint multiple suitable targets for follow-up correlative analyses, such as BEX, EBSD, and Raman, to further characterize their properties.

Our Solution: We undertook a survey scan of a till sample to detect the presence of geologically insightful accessory minerals. All mineral phases were segmented using BSE greyscale thresholds, again removing analysis of epoxy and breaking down composite particles into mineral constituents (Fig. 3).

Figure 3 - Flow diagram displaying the different stages of particle analysis completed using AZtecGeo. Rapid accessory mineral searches can be completed with ease, streamlining the way to targeted compositional analysis.

A minimum of 20,000 counts per feature could be acquired and processed for over 11,000 mineral grains within the space of 50 minutes. Each of these was classified into a mineral phase by the AZtecGeo classification scheme immediately.  The dataset could then be easily filtered to isolate the important accessory phases - monazite (n = 64), apatite (n = 178) and zircon (n = 108) within the sample. As the stage coordinates are remembered for each particle, relocation and correlative analysis of targeted phases with BEX, EDS mapping, Raman or EBSD could be carried out with ease. Otherwise, a large area map (Fig. 3) could be exported to guide analyses on a separate system.

High-Throughput Amphibole Analysis for Petrological Studies

Petrologists frequently rely on the composition of abundant rock-forming minerals to reconstruct the geological conditions in which they formed. Magmatic amphibole, for example, serves as a crucial indicator of temperature and pressure during crystallisation. By analysing a large population of crystals, researchers can uncover compositional variability, sub-populations, and cryptic magmatic processes. 

Application Objective: Volcanic systems often exhibit significant mineralogical diversity within their amphibole populations. These variations can provide key insights into geological processes such as magma mixing, crystallization histories, and storage conditions beneath active volcanoes.

Challenges: Manually analysing amphibole populations using traditional ‘point and ID’ methods is labour-intensive and time-consuming. Additionally, cognitive bias in selecting crystals for analysis can lead to an incomplete or skewed representation of chemical variability. To obtain statistically meaningful results, many analyses must be conducted efficiently whilst trying to avoid cognitive bias. Aside from user associated problems, the accurate detection and quantification of minor elements requires 10s of thousands of counts. Meaning high-speed data collection is a necessity for accurate petrological insight.    

Our Solution: To streamline amphibole analysis, we implemented a strict threshold filter designed to exclude non-target materials such as glass and other minerals (see image). Using this automated approach, a total of 811 amphibole crystals were successfully analysed within just 20 minutes over 86 individual fields covering an area of 50 mm². Montaging these fields with the ability to reconstruct border particles produces a robust dataset that captures the amphibole diversity within the sample.

Figure 4 - Flow diagram displaying the different stages of targeted amphibole analysis completed using AZtecGeo. Rapid, population-wide analysis can be completed with ease, allowing unbiased and comprehensive population composition to be assessed.

To ensure the detection of minor elemental constituents (<2 wt.%), a minimum threshold of 30,000 counts per feature was applied. Using AZtecGeo’s automated classification system, the amphibole population was predominantly identified as hastingsite and edenite, with minor occurrences of hornblende. This level of classification, performed in real time, provides immediate insights into the mineralogical composition without the need for manual sorting.

Across the analysed crystal population, a bimodal distribution of Si, Al, and Ti was observed. These trends were instantly visualized within AZtec and further explored using exported compositional data in Excel. The clustering of amphibole compositions suggests the presence of two distinct populations, potentially representing crystallization from separate magma batches. This finding highlights the power of automated mineral classification in deciphering complex magmatic histories.

Conclusions

These examples demonstrate how AZtecGeo enhances the powerful particle analysis capabilities of AZtecFeature for mineral quantification. With an extensive mineral classification library, optimized analytical settings, and expert-guided workflow support, this approach enables the rapid, accurate, and accessible acquisition of compositional and textural data across large sample areas. Unlike traditional automated mineralogy solutions, AZtecGeo offers a flexible and cost-effective alternative, making it an invaluable tool for both commercial and academic geological research.