Reassessment of the Raman CO2 densimeter

Abstract Raman spectroscopy has proven to be an effective tool to confirm the presence and abundance of CO 2 in fluid and melt inclusions. The Raman method for quantifying CO 2 abundance is based on the observation that the distance between the two Raman bands comprising the Fermi diad varies systematically with CO 2 density. In recent years, several Raman densimeters have been developed by different research groups to determine the density of CO 2 in fluid and melt inclusions. The different densimeters that have been proposed predict different densities for the same Fermi diad splitting, leading to large differences in estimated CO 2 contents for inclusions, depending on which densimeter is used to interpret the Raman data. In this study, we examine potential causes for variations in the various densimeters and show that these differences are mainly the result of using different Raman instruments and settings, different collection parameters, and different analytical methods. Twelve experiments were conducted to test the variability associated with changing instrumental and analytical conditions, as well as to understand the differences between the various densimeters, using three different Raman instruments, with different laser sources and dispersion gratings. In all of the experiments, the splitting of the Fermi diad of CO 2 and CO 2 density at pressures from the liquid-vapor curve (6.0 MPa to 0.06 MPa) at ambient temperature (~ 22 °C) was calibrated using a high-pressure optical cell. The results show a consistent behavior whereby all analytical configurations show parallel trends in terms of the variation in Fermi diad splitting as a function of CO 2 density. The slopes of the lines representing the variation in Fermi diad splitting as a function of CO 2 density, as well as low density (pressure) data from other densimeters ( Kawakami et al., 2003 , Yamamoto and Kagi, 2006 , Song et al., 2009 , Fall et al., 2011 , Wang et al., 2011 ) are remarkably similar, with a variation of about ~ 10% and a standard deviation of 3%. The differences observed in all densimeters, including previously published densimeters and the 12 experiments from this study, are most likely a function of variations in instrumentation, laser excitation wavelength, gratings, and analytical protocols used during the experimental calibration of the splitting of the Fermi diad. Based on results of this study, we recommend against using any published densimeter to interpret Raman data collected using an instrument other than that on which the calibration is based, and suggest that researchers develop a calibration that is applicable and specific to their instrument and data collection protocol.