A systematic review and meta-analysis of sputter yield data for xenon ions normally incident on graphite at energies below 2000 eV was undertaken to identify systematic errors, determine the best model parameter values to represent yield as a function of energy, quantify uncertainty, and determine if the data support differences in yields for different types of graphite. A critical examination of the 11 published data sets for high density graphite, pyrolytic graphite, and amorphous carbon showed that, in general, they were carefully controlled to minimize errors. The most significant quantifiable systematic errors were those caused by the neglect of doubly charged ions, chemical erosion, and the impact of secondary electron emission on ion flux measurements. The effects of gas uptake and outgassing on mass loss measurements and unrepresentative surface textures may have biased other experiments, but these effects could not be quantified. The semi-empirical Eckstein model for yield as a function of energy was fit to data for the three graphite types using a hierarchical Bayesian statistical model, producing recommended fit parameters and probability distributions representing uncertainty in yields. The results showed that differences in yield for high density graphite and pyrolytic graphite were not statistically significant. Apparent differences in yield for amorphous carbon disappeared when the single data set available for energies below 150 eV was corrected for reasonable values of double ion content. Recommended procedures to avoid systematic errors and additional experiments and modeling to fill in gaps in our understanding are included.
Read full abstract