The Utilization of mXRF to Predict Carbon and Nitrogen Isotope Trends in the Cretaceous Mancos C Formation, New Mexico
Randazzo N., Gabriel J., Barros Lourenco R., Reinhardt E., Kim S.-T., Bhattacharya J.P., and Genovese C.
McMaster University
Stable isotope analyses on siliciclastic core samples are often used to divide chemostratigraphy and reconstruct paleoenvironments. However, conventional stable isotope sampling techniques are only limited to the centimeter scale while the sampling scale of Energy Dispersive-micro X-Ray Fluorescence (ED-mXRF) using an ITRAX Core Scanner can reach millimeter to micrometer. We scanned a Cretaceous-age sediment core from the Mancos C Formation in the San Juan Basin, New Mexico using ED-XRF for several redox sensitive elements at a 500 nm resolution. The samples taken for stable isotope measurements were also analyzed using an Elemental Analyzer attached to a Continuous Flow-Isotope Ratio Mass Spectrometer (CF-IRMS) to obtain δ13Corganic and d15Nsediments values and Total Organic Carbon (TOC). Afterwards, the XRF scanning results were then compared with the stable isotope measurements of the same core at a 22 cm resolution.
Assessment of the correlation between the ED-XRF data and the stable isotope data were based on multiple statistical models which were built to compare our elemental data with δ13Corganic and d15Nsediments values and TOC. These models include Ordinary Least Squares Regression (OLS) and Weighted Least Squares (WLS) Regression, as well as Machine Learning-based regressions: Ridge, Lasso, and Elastic Net regressions. A strong correlation between these two chemostratigraphic techniques would provide a means to use ED-XRF data as a non-destructive way to estimate stable isotope trends on the millimeter and micrometer scale and inform sampling strategies for stable isotope analysis of the siliciclastic core samples.
All five regression models discussed above were trained using 70% of our δ13Corganic and d15Nsediments values and TOC data and were able to successfully predict the values of the remaining 30% of data with absolute R2 values of ~0.99. Ni counts were found to be the best predictor of δ13Corganic values, Ba and U trended strongly with d15Nsediments values, and Cr counts were able to predict the TOC.
Our study also compared the elemental and stable isotope data with high resolution sequence stratigraphy to explore the link between environmental change and anoxic events. Our results found the elemental and stable isotope data reflected the depositional environments identified in the sequence stratigraphy.
Finally, this study briefly compared two different carbonate removal techniques (“acid-washing” and “fumigation”) and commented on which technique worked best for the conditions of this experiment. The results of our tests were then compared with the data from the bulk sediment analysis, which included a mixture of both the organic material and carbonates. Preliminary worked suggests that acid-washing was more successful at removing carbonate materials, with some fumigated samples having low δ13Corganic values which were similar to the bulk samples.
