NEW METHOD FOR ENHANCED SAMPLING STRATEGY OF BASALT: LASER-INDUCED BREAKDOWN SPECTROSCOPY (LIBS) ANALYSIS OF THE CARRIZOZO BASALT FLOW, NEW MEXICO

Significant chemical variation exists among samples that are superficially very similar, such as basalts. This limits representative sampling because it is impossible to recognize chemically interesting samples in the field. Laser-Induced Breakdown Spectroscopy (LIBS) is a rapid, field-portable analytical technique that allows one to distinguish between samples of similar composition. In LIBS analysis, a high-power laser is focused on the sample surface, ablating a small amount of material and forming a high-temperature plasma. As excited atoms in the plasma cool, electrons decay to lower-energy orbitals, emitting light in the form of photons. The photons are collected by fiber optic or telescope, diffracted, and recorded on a CCD camera. The resulting spectrum records concentration information from nearly the entire periodic table, as well as isotopic information, providing a rich chemical fingerprint of the material.

The 5000 ka Carrizozo basalt flow in central New Mexico is an ideal test case for the sensitivity of LIBS for distinguishing between basalts with similar compositions. The flow erupted in two major episodes, in a total of five pulses. Each pulse has been sampled in two or three locations in vertical transects; these samples have been analyzed by XRF. The range in composition of 25 samples is as follows: SiO₂ = 48.8-52.4; TiO₂ = 1.6-1.8; Al₂O₃ = 15.2-16.4; Fe₂O₃^* = 10.8-11.6; MgO = 5.9-7.7; CaO = 7.9-9.1; Na₂O = 3.5-4.2; K₂O = 1.3-1.7; P₂O₅ = 0.3-0.4 wt%.

Gravel-sized particles of the basalt samples were analyzed with LIBS, collecting 100 spectra per sample and averaging those spectra into a single whole-rock spectrum. Spectra will be analyzed with the multivariate analysis techniques Principal Component Analysis (PCA) and Projection to Latent Structures Regression (PLSR) to determine the minimum compositional difference that can be detected by LIBS. Multivariate analysis takes advantage of the large amount of useful information in each spectrum; not only can samples with different compositions be recognized, but the elements causing those differences can be identified.