The identification of minerals is fundamental to the use and interpretation of earth and planetary materials. Minerals are defined by their chemistry and crystalline structure. A common way to identify minerals involves using instruments such as an Electron Probe Micro-Analyzer (EPMA) to measure the chemistry of a grain or crystal and compare element ratios to known minerals, i.e. stoichiometry, but this requires user expertise and often some prior knowledge of expected minerals. Here, we present MIST (Mineral Identification by SToichiometry), a mineral-stoichiometry-based model to identify geochemical observations with elemental ratios that match natural mineral compositions. MIST uses normalized oxide weight percentages and stoichiometric ratios between elements in a detailed hierarchical rules-based classification scheme based on validated mineral formulas and compositions to identify mineral phases. The model includes tolerances allowing the vacancies and elemental substitutions common in natural mineral analyses. MIST is focused on rock-forming mineral species containing oxygen and is tested against a standard dataset of validated mineral analyses. The current version of MIST, 3.0, can identify 246 mineral species or stoichiometrically indistinguishable sets of species, with the capability to expand the number of species recognized in future versions. MIST outputs precise mineral formulas, relevant mineral endmembers, and values used in intermediate calculations. As with other mineral identification methods, stoichiometric mineral identifications should be compared to other datasets, including oxide totals, textural, or structural information. We used MIST to filter over a million mineral chemistry analyses in the GEOROC database, resulting in over 875,000 natural mineral analyses with standardized labels, formulas, and mineral descriptors that can be used for machine learning models. MIST provides a rapid, accurate, standardized way to recognize minerals in high-resolution chemical datasets while minimizing required mineralogical expertise.

The Perseverance rover traversed across the highly anticipated Margin Unit (MU), Bright Angel (BA), and Crater Rim (CR) terrains of Jezero crater between mission sols ~900 and present. These areas have been sought after for their potential habitable past [1,2]. Here, we report mineralogy derived from chemical analyses in these areas while considering complicating Martian conditions and instrument uncertainties. These results help to continually quantify chemical variations in Jezero and provide mineral data for understanding the complex history of this area.

Identifying minerals is a prerequisite to interpreting geologic history from samples. Various techniques for mineral identification are available, but one method commonly used for individual grains or crystals in a sample is to obtain high-resolution geochemical information of the grain and compare element ratios to known minerals, i.e., stoichiometry.

We developed MIST (Mineral Identification by Stoichiometry) as a first-principles-based computational algorithm to identify geochemical observations with stoichiometric elemental ratios that match real-world compositions. Identifying minerals by stoichiometry works when the geochemical measurement spot size is smaller than the size of the mineral. Beyond this requirement, stoichiometric rules are agnostic to the source of geochemical data. MIST uses normalized oxide weight percentages and stoichiometric ratios between elements in a detailed decision tree approach to identify mineral phases using recognized mineral group hierarchies. The approach incorporates tolerances tailored to the imperfections and elemental substitutions common in real-world minerals. When an observation matches a mineral species, we output the name of the mineral according to IMA rules and a detailed stoichiometric mineral formula. The algorithm has been tested on well over 2000 mineral compositions and currently identifies 250 mineral species with >95% accuracy. MIST is free to use via an online API at mist.rice.edu.

Data quality testing and standardization is a significant area of development in geochemical databases and compilations of geochemical results. MIST can rapidly standardize mineral chemistry datasets by recognizing when a geochemical measurement stoichiometrically matches a mineral. For example, we tested 222,543 geochemical compositions classified as olivine in the GEOROC chemical database. Accuracy is a known issue with databases that collect decades of data, but if each individual user must filter the datasets themselves, they likely select compositions based on the individual’s perception of the tolerance of real mineral variations; MIST can standardize this filtering process, preparing datasets for better use in research projects and as training datasets for machine learning models.

The Mars 2020 Perseverance rover has traversed from the crater floor, through the western fan, and into the Margin Unit(s) of Jezero crater, methodically investigating outcrop exposures with proximity instruments prior to collecting samples for potential future return to Earth. Quantifying chemical differences in the variable mineral assemblages found along the traverse informs the conditions that once existed in Jezero, documents relationships between complementary samples and sample suites, and provides context to maximize return sample science.

The Mars 2020 Perseverance rover has traversed from the crater floor, through the western fan, and into the Margin Unit(s) of Jezero crater, methodically investigating outcrop exposures with proximity instruments prior to collecting samples for potential future return to Earth. Quantifying chemical differences in the variable mineral assemblages found along the traverse informs the conditions that once existed in Jezero, documents relationships between complementary samples and sample suites, and provides context to maximize return sample science.

The Mars2020 Perseverance Rover is exploring Jezero crater on Mars to investigate a potential past habitable environment. The rocks in Jezero crater hold key clues to understanding the past habitability of Mars because this crater was once home to a lake environment. To decipher the conditions of this past environment, it is crucial to identify the primary and secondary minerals of the rocks remaining in Jezero crater.

Thus far, the crater floor has been identified as an igneous olivine cumulate from analysis of the primary minerals. Furthermore, presence of secondary minerals preserves information about the water-rock interactions that occurred in the crater. Identifying primary and secondary minerals is an important first step for compiling the history of Jezero crater.

Identifying the detailed stoichiometry of mineral assemblages can provide valuable insight into the specific geochemistry of the conditions that formed the rocks. While valuable, it is more challenging to extract the exact crystal chemistry of minerals. However, by utilizing geochemical data from the PIXL instrument onboard the Perseverance rover and a developed mineral identification algorithm, we identify and provide the exact stoichiometry of primary and secondary mineralogy of rock targets. This is a meaningful endeavor for constraining the history of Jezero crater and the geochemical interactions that occurred to create the facies.

Identifying minerals is a prerequisite to interpreting geologic history from samples. A variety of techniques for mineral identification are available, but one method commonly used for individual grains or crystals in a sample is to obtain high-resolution geochemical information of the grain and compare element ratios to known minerals, i.e., stoichiometry. High-resolution geochemical data can be obtained using a variety of analysis tools, including EPMA, XRF (e.g., PIXL instrument on Perseverance), EDS, ICP-MS, and LIBS (e.g., ChemCam and SuperCam instruments). However, the process of identifying minerals by stoichiometry historically requires some understanding of expected minerals and a series of binary and ternary diagrams to identify geochemical endmembers to check their stoichiometry. MIST simplifies this step.

MIST 2.0 is a first-principles-based computational algorithm to filter geochemical analyses and identify observations with stoichiometric elemental ratios that match real mineral compositions. MIST is a decision-tree model that filters the geochemical analyses based on elemental ratios and normalized oxide percentages following mineral classification rules based on real mineral structures and stoichiometries, including typical elemental substitutions. Mineral species are identified and reported in five steps, first as observations that fit in mineral classes, then mineral groups, sub-groups, sub-sub-groups and species. Mixtures of phases or chemical ratios that would not fit a real mineral structure are identified as mixtures. The model includes a “tolerance” value for instrument uncertainty or unexpected mineral substitutions, this will be adjustable to either minimize false detections or identify a broader range of possible species. Results should be compared to other information on the sample.

MIST currently identifies 200+ mineral species and has successfully identified minerals in thousands of EPMA analyses and PIXL analyses (see Moreland et al., this meeting). The model outputs the identified mineral, the mineral formula calculated from the stoichiometry of the analysis, and relevant variables for particular mineral groups, such as Ab-An-Or for plagioclase and Wo-En-Fs for pyroxene.

We invite you to use the model on your geochemical datasets at mist.rice.edu.

Mineral identification is critical for interpretation of geological samples. However, geochemical data is often easier to collect on in-situ geological surfaces on Earth and other planets, using techniques including EPMA, XRF (e.g., PIXL instrument on Perseverance), EDS, ICP-S, and LIBS (e.g., ChemCam and SuperCam instruments), among others. When the spot size of the geochemical measurement is smaller than the mineral crystal or grain size, stoichiometry, or ratios between elements, can be used to identify mineral constituents or a suite of mineral polymorphs to improve geological interpretations.

We have developed a computational algorithm, MIST 2.0, to filter geochemical analyses and identify observations with stoichiometric elemental ratios that match real mineral compositions. The algorithm uses normalized oxide percentages and stoichiometric ratios between elements in a detailed decision-tree approach to identify mineral phases. Mineral species are recognized in five steps, first as observations that fit in mineral classes, then mineral groups, sub-groups, sub- sub-groups and species. Mixtures of phases or chemical ratios that would not fit a real mineral structure are identified as mixtures. When an observation matches a single mineral we output the name of mineral according to IMA rules and a stoichiometric mineral formula. The algorithm has been tested on well over 2000 EPMA mineral compositions and currently identifies 150 mineral species with >95% accuracy. We are generating a public website to enable users to use this tool to identify minerals in their geochemical datasets.

There are many instruments on Earth and even deployed on the surface of Mars that measure the chemistry of sand or rock surfaces at high resolution, but in order to interpret the chemistry, we need to know how the elements are bound together into minerals. For example, an increase in magnesium has very different implications if the magnesium is in an igneous mineral that reveals information about its volcanic source, or if it is in magnesium sulfate, a salt produced in the final stages of evaporation or freezing.

Here, we present a computational model we created to identify minerals in high resolution chemical maps. It works by checking the ratios between anions and cations and comparing them to known mineral formulas. This only works when the mineral happens to be larger than the observation spot size, so it does not always return results- but knowing at least one mineral can really help the geological interpretation.

We have shown that the model works well to distinguish mineral types in high resolution chemistry maps from an electron microprobe on Earth. Now we are applying the model to the PIXL instrument on the Mars 2020 rover mission to identify minerals on Mars.

Compositional data from the surface of Mars have been used for decades to describe the character of Martian surfaces at planetary scales (e.g., Gamma-Ray Spectrometer data) and at rover scales (e.g., Pathfinder, MER, and MSL APXS data, MSL ChemCam data). Geochemical data provide critical constraints on the bulk composition of the surface of the planet or specific rocks, but they must be interpreted to constrain geological processes and habitability. For example, an increase in MgO has very different implications if the MgO is contained in olivine, where it may be a signature of the igneous protolith and an indication of minimal water-rock interaction, vs in magnesium sulfate, which is produced in the final stages of evaporation or freezing. Interpretations of the co-varying elements in compositional maps have frequently been used to infer minerals and their abundances, or the elemental maps have been correlated with results from instruments designed to measure mineralogy to infer chemistry of amorphous phases. Here, we describe our approach to an algorithm that will be used to automatically identify putative minerals in high-resolution geochemical data, with application to the Planetary Instrument for X-ray Lithochemistry (PIXL) on the Perseverance rover.