5:15 PM - 6:30 PM
[MTT42-P02] Factors limiting the accuracy and precision of peak positions and intensity ratios in Raman spectra: Optimisation of optics and measurement environment for carbon isotope ratio measurements of CO2
Keywords:Raman spectroscopy, Carbon isotope ratio, Quantification, Spectral resolution, Noise
Raman spectroscopy has been established as one of the most versatile quantitative analytical methods in materials science research due to its non-destructive analysis and high spatial resolution. Experimental calibration is performed to relate Raman spectral characteristics to the state of materials to quantify density, pressure, temperature, concentration, strain, etc. In recent years, Raman spectroscopy has been used to improve and develop non-destructive local isotope analysis methods for H2O, CO2, CH4, HCO3-, CO32-, N2, humic acid, calcite, etc. However, the accuracy of these methods is still far from being applied to the measurement of isotope ratios in natural samples. For example, Arakawa et al. (2007)1 and Yokokura et al. (2020)2 reported 1σ = 20‰ and 8.7‰, respectively, for the accuracy of carbon isotope ratio measurements of CO2. In Yokokura et al. (2020)2, the reason for the improvement in measurement accuracy is that the focal length of the monochromator is longer, the pixel resolution is higher, and the measurement time is longer, but it is not clear which factors contribute most to the improvement. Other parameters such as excitation wavelength, grating engraving number and size, slit width, detector noise characteristics, reciprocal line dispersion, and pixel size can also have a significant effect on measurement accuracy, but the relationship between these parameters and measurement accuracy is not well understood. Once we know which optics can be upgraded most efficiently to improve the accuracy of our measurements, we will be able to overcome the 1σ = 8.7‰ barrier reported by Yokokura et al. (2020)2 while reducing cost and time. In this study, we use simulated CO2 Raman spectra to evaluate how the above parameters affect the accuracy of intensity ratio measurements.
The factors limiting the accuracy of Raman spectroscopy can be divided into three main categories: (1) noise, (2) spectral resolution, and (3) thermal and mechanical stability of the instrument.
To obtain the shot noise, the efficiency of Raman scattering (count/sec) is used as an input parameter, and the average number of photons (e-) arriving at the detector in a certain time interval is estimated from the quantum efficiency of CCD (%) and A/D count (e-/count). The square root of this value was then randomly generated as shot noise according to Poisson statistics. Dark noise and readout noise were randomly generated as Poisson and Gaussian distributed noise, respectively, using the measured values of 1.00×10-5 (e-/pix/sec) and 3.36 (e-/times0.5). The effect of the instrumental spectral resolution on the peak bandwidth was introduced following the model of Liu et al. The projection of the slit on the CCD and the spectrum of the sample were assumed to be Gaussian distributed, and the resulting bandwidth was calculated to be the square root of (spectral resolution)2+(natural bandwidth)2.
Based on the above model, it is possible to generate spectra taking into account the effects of noise and spectral resolution by assuming the analysis time (sec), the number of vertical pixels (pix) (i.e. binning width), the shape of the Raman spectrum, the peak position (cm-1), the natural bandwidth (cm-1) and the amplitude (count/sec). It is possible to generate spectra taking into account the effects of noise and spectral resolution. In this study, the Raman spectra of CO2 generated by the above method with varying the parameters were fitted with the spectrum analysis software (GRAMS/AI) to investigate which parameters affect the measurement accuracy such as peak position and intensity ratio. Finally, we estimate under what conditions the carbon isotope ratio of CO2 can be measured with an accuracy of ‰.
The factors limiting the accuracy of Raman spectroscopy can be divided into three main categories: (1) noise, (2) spectral resolution, and (3) thermal and mechanical stability of the instrument.
To obtain the shot noise, the efficiency of Raman scattering (count/sec) is used as an input parameter, and the average number of photons (e-) arriving at the detector in a certain time interval is estimated from the quantum efficiency of CCD (%) and A/D count (e-/count). The square root of this value was then randomly generated as shot noise according to Poisson statistics. Dark noise and readout noise were randomly generated as Poisson and Gaussian distributed noise, respectively, using the measured values of 1.00×10-5 (e-/pix/sec) and 3.36 (e-/times0.5). The effect of the instrumental spectral resolution on the peak bandwidth was introduced following the model of Liu et al. The projection of the slit on the CCD and the spectrum of the sample were assumed to be Gaussian distributed, and the resulting bandwidth was calculated to be the square root of (spectral resolution)2+(natural bandwidth)2.
Based on the above model, it is possible to generate spectra taking into account the effects of noise and spectral resolution by assuming the analysis time (sec), the number of vertical pixels (pix) (i.e. binning width), the shape of the Raman spectrum, the peak position (cm-1), the natural bandwidth (cm-1) and the amplitude (count/sec). It is possible to generate spectra taking into account the effects of noise and spectral resolution. In this study, the Raman spectra of CO2 generated by the above method with varying the parameters were fitted with the spectrum analysis software (GRAMS/AI) to investigate which parameters affect the measurement accuracy such as peak position and intensity ratio. Finally, we estimate under what conditions the carbon isotope ratio of CO2 can be measured with an accuracy of ‰.