Sandia National Laboratories, Albuquerque, NM 87185, USA.
Microsc Microanal. 2010 Jun;16(3):259-72. doi: 10.1017/S1431927610000267. Epub 2010 Apr 8.
In this article, we use simulated and experimental data to explore how three operator-controllable parameters--(1) signal level, (2) detector resolution, and (3) number of factors chosen for analysis--affect quantitative analyses of scanning transmission electron microscopy-energy dispersive X-ray spectroscopy spectrum images processed by principal component analysis (PCA). We find that improvements in both signal level and detector resolution improve the precision of quantitative analyses, but that signal level is the most important. We also find that if the rank of the PCA solution is not chosen properly, it may be possible to improperly fit the underlying data and degrade the accuracy of results. Additionally, precision is degraded in the case when too many factors are included in the model.
在本文中，我们使用模拟和实验数据来探讨三个操作员可控制的参数——（1）信号水平，（2）探测器分辨率，以及（3）用于分析的因子数量——如何影响通过主成分分析（PCA）处理的扫描透射电子显微镜-能量色散 X 射线光谱图像的定量分析。我们发现，信号水平和探测器分辨率的提高都可以提高定量分析的精度，但信号水平是最重要的。我们还发现，如果 PCA 解的秩选择不当，则可能无法正确拟合基础数据并降低结果的准确性。此外，当模型中包含太多因子时，精度会降低。
