Times are displayed in (UTC-05:00) Central Time (US & Canada) Change

Iron Redox State and Oxygen Fugacity of Silicate Minerals using Raman Spectroscopy

Session: Advancing Mineralogy and Spectroscopy Across the Solar System in Honor of MSA Roebling Medalist M. Darby Dyar

Presenting Author:

Dr. Laura Brucia Breitenfeld

Authors:

Breitenfeld, Laura B.¹, Dyar, M. Darby², McCanta, Molly C.³

(1) Geosciences, Stony Brook University, Stony Brook, New York, USA; Planetary Science Institute, Tucson, Arizona, USA, (2) Planetary Science Institute, Tucson, Arizona, USA, (3) Earth and Planetary Sciences, University of Tennessee, Knoxville, Tennessee, USA,

Abstract:

Evaluating redox conditions and oxygen fugacity of silicate minerals is essential for understanding mineral formation and the history of water on Earth and across our Solar System. Here, we use Raman spectroscopy, an inelastic scattering technique, to characterize suites of silicates by mineral group. More specifically, we investigate whether Raman spectroscopy can be utilized as a tool to predict iron redox state and oxygen fugacity.

Thus far, we have studied a suite of 77 well-charactered amphibole minerals, and we are currently evaluating a suite of 44 feldspar minerals in experimentally equilibrated charges. Studies of other mineral groups including garnet and pyroxene are planned. All samples were analyzed on a Bruker Senterra Raman spectrometer using the 532 nm laser with a microscope 50× objective attachment. A laser power of 25 mW was employed for 400 co-additions using an integration time of 0.2 seconds. Raman spectral measurements extend from 50 to 4,250 cm^-1 at a 1.5 cm^-1/channel spectral resolution. Each spectrum is associated with the corresponding compositional metadata of the sample.

Partial Least Squares (PLS) multivariate analysis modeling is used to predict iron redox state (amphiboles) and oxygen fugacity (feldspars) using the Raman spectra. PLS regresses one response variable against multiple explanatory variables (intensity at each channel of the spectra). PLS predictions utilize every channel of the spectral range, assigning coefficients to every channel, rather than focusing on specific peaks.

Raman spectra from amphibole minerals are used to estimate %Fe³⁺ (relative to total Fe) using PLS. The accuracy of our model for prediction of %Fe³⁺ is ±8.1% (absolute) expressed as root-mean-square error (RMSE) of the entire data set, covering the range from 0 to 100% with an R² value of 0.85. A strong PLS model coefficients is located at 585 cm^-1, in part due to Fe³⁺– O stretching within this spectral region. One of the advantages of the PLS method is that it produces interpretable coefficients for each channel of the spectra, linking spectral features to prediction variables. Our preliminary feldspar measurements show large model coefficients at 471 cm^-1 (Group I region associated with ring-breathing modes), when predicting log f_O2 relative to NNO. Studies of this type underscore the potential of Raman spectroscopy to extract compositional and environmental information about rock-forming minerals.