Andrew Pollard1,Piers Turner1,Keith R Paton1,Elizabeth Legge1,Konstantinos Despotelis1
National Physical Laboratory (NPL)1
Andrew Pollard1,Piers Turner1,Keith R Paton1,Elizabeth Legge1,Konstantinos Despotelis1
National Physical Laboratory (NPL)1
As a key component of energy storage devices, electrodes containing graphitic particles must be accurately characterised to determine their material properties, properties which will affect the final performance of the device itself. Raman spectroscopy has long been a ‘go-to’ technique for measuring carbon materials, as the resulting Raman spectra reveal many of the material properties, such as the level of disorder, strain, doping, particle lateral size and thickness. However, although this technique has been used for many decades, there are still glaring issues with reproducibility, as well as the quantification of the Raman peak metrics in relation to the material properties. As a non-destructive and rapid technique, these issues become a crucial barrier, when trying to use this technique for both academic research and as an industrial quality control tool.<br/>Towards this effort, firstly we report the results of an international interlaboratory study (ILC) conducted under VAMAS. This study compared Raman spectroscopy data for an idealised graphitic material, that is a graphene layer grown using chemical vapour deposition, gathered from 17 participants across academia, industry (including instrument manufacturers) and National laboratories and will be used as the basis for the international standard ISO/TS 21356-2. By comparing reported Raman metrics with the measurements of the same regions made by the lead participant, variations in the reported peak intensity ratios and peak fits could be explored for both instrumentation and data analysis. Due to a lack of relative intensity calibration, the relative difference reported in the 2D- and G peak intensity ratios (<i>Ι</i><sub>2D</sub>/<i>Ι</i><sub>G</sub>) was up to 200%. It was also shown that the standard deviation for Γ<sub>2D</sub> values reported by different software packages, was 15× larger for Lorentzian fit functions than for pseudo-Voigt functions. This study has shown that by adopting a relative intensity calibration and consistent peak fitting and data analysis methodologies, these large, and previously unquantified, variations can be significantly reduced, allowing more reproducible and comparable measurements of graphitic materials using Raman spectroscopy.<br/>Secondly, we assess how Raman metrics measured for graphitic particles in powder form can be used to determine impurities and the limits of detection, namely for larger and undesired graphite particles that, in this case, have been deliberately added to a more homogeneous powder of smaller particles. When a sufficiently large number of spectra are acquired, it is found that by processing and classifying individual spectra, rather than the averaged spectrum, a reasonable estimate of the fraction of unexfoliated material can be obtained. These results highlight the care that must be taken when interpreting results of Raman spectroscopy measurements.