T199 The Geology and Environmental Impacts of Lithium Resources in North America and other Globally Significant Deposits (Posters)

Tuesday, 11 October 2022
2:00 PM - 6:00 PM
Exhibit Hall F

PREDICTION OF LI VIA GEOCHEMICAL DATA AND MACHINE LEARNING ALGORITHMS IN SOILS FORMED ON PEGMATITE BEDROCK: A PRELIMINARY STUDY

Paper number: T199-3
Program: Technical Sessions
Time: 2:00 PM – 6:00 PM

Authors

Abstract

With the need for more sustainable, green energy and carbon-free alternative energy, lithium (Li) is in high demand for rechargeable batteries and grid-scale energy storage. Lithium-cesium-tantalum (LCT) pegmatites are important resources of Li and other critical metals such as niobium (Nb), tantalum (Ta), and cesium (Cs). Traditional methods of geological exploration are time-consuming and expensive. Portable X-ray fluorescence (pXRF) spectrometry may help to overcome these issues when properly calibrated. Although pXRF cannot detect Li, it can detect and quantify Li pathfinder elements, such as Cs, Rb, Sn, etc. In this preliminary study, we evaluate pXRF data and a machine learning algorithm to predict Li contents in soil samples based on Li pathfinders.

Soil samples were collected at two LCT pegmatites: Animikie Red Ace (ARA) and Kings-X2 (KX2), Florence County, WI; and a mineralogically-simple pegmatite, at Sturgeon River Quarry (SRQ), Dickinson County, MI. To date, results were collected on 16 soil samples formed directly on pegmatites and 15 samples collected along a transect perpendicular to the ARA LCT dike, a lepidolite-spodumene bearing, highly-fractionated dike, < 3 m thick. The ARA transect expands across the host rocks, a quartzo-feldspatic micaschist. Soils were collected from the surface horizons (H or A), dried, sieved (< 2 mm), pulverized, and analyzed by ICP-OES and pXRF for bulk composition. Statistical models were created based on the geochemical data. Soil samples were randomly separated into modeling (70%) and validation (30%) datasets. A Random Forest approach was employed in the modeling process. The R2 coefficient and root mean squared error (RMSE) were used to compare the predicted and observed values and to validate the best Li predictive models using pXRF data. The low K/Rb ratios in ARA and KX2 are consistent with high degrees of magma fractionation of the pegmatite substrate, as seen in Table 1.

Table 1. Means of K and Rb (mg kg-1) and K/Rb in soils, determined with pXRF

K

Rb

K/Rb

ARA

32782

1629

29

ARA transect

20582

62

449

KX2

13504

705

39

SRQ

25282

89

324

To complete this study, we will integrate the geochemistry of ~130 soil and 18 rock samples from pegmatites and host-rock transect to create a more robust Li predictive model, as a potential fast and inexpensive exploration method.

Geological Society of America Abstracts with Programs. Vol. 54, No. 6

doi: 10.1130/abs/2022AM-382238

© Copyright 2022 The Geological Society of America (GSA), all rights reserved. Permission is hereby granted to the author(s) of this abstract to reproduce and distribute it freely, for noncommercial purposes. Permission is hereby granted to any individual scientist to download a single copy of this electronic file and reproduce up to 20 paper copies for noncommercial purposes advancing science and education, including classroom use, providing all reproductions include the complete content shown here, including the author information. All other forms of reproduction and/or transmittal are prohibited without written permission from GSA Copyright Permissions.