Repeatability and reproducibility in surface analysis in the semiconductor industry are key for supporting efficient process development and high volume manufacturing (HVM). Long-term repeatability is critically important when comparing to historical data, while reproducibility is required to support technology transfers when HVM of specific devices is to be carried out at multiple sites. This, however, introduces a number of unique challenges for running a characterization facility. In this work, the authors will describe several examples that can result in reproducibility issues. The examples will be taken in the areas of x-ray photoelectron spectroscopy and secondary ion mass spectrometry. The first and foremost origin of repeatability and reproducibility issues is instrumental variation. A second important contributor to reproducibility issues is sample variability. The authors will show that assessing long-term instrumental stability is potentially hindered by long-term variation of sample characteristics. The authors will also show that an understanding of characterization techniques is paramount to understand such issues. In addition to “pure” technical causes of repeatability and reproducibility issues, the human factor needs to be considered as well. This involves, for instance, decision making in data treatment during, for example, fitting procedures, statistical treatments, etc. Practical examples are given to illustrate this. With present day characterization depending more heavily on computational support/commercial software, potential detriments to characterization repeatability arising from software will again be made evident. Finally, the authors will illustrate with round-robin results that by combining all above-mentioned factors, widely varying results can be obtained on the same samples.

中文翻译：

---

获得可再现的数据：半导体技术中的表面分析示例

半导体行业中表面分析的可重复性和可重复性是支持有效的工艺开发和大批量生产（HVM）的关键。与历史数据进行比较时，长期可重复性至关重要，而当要在多个站点执行特定设备的HVM时，需要重现性来支持技术转移。但是，这给运行表征设施带来了许多独特的挑战。在这项工作中，作者将描述几个可能导致重现性问题的示例。这些示例将在X射线光电子能谱和二次离子质谱仪领域进行。可重复性和再现性问题的第一个也是最重要的起源是仪器的变化。导致重复性问题的第二个重要因素是样品变异性。作者将表明，样品特性的长期变化可能会阻碍评估仪器的长期稳定性。作者还将表明，对表征技术的理解对于理解此类问题至关重要。除了重复性和再现性问题的“纯”技术原因外，还需要考虑人为因素。例如，这涉及在拟合过程，统计处理等过程中进行数据处理时的决策。给出了一些实际示例来说明这一点。由于当今的特征分析在很大程度上取决于计算支持/商业软件，由软件引起的表征重复性的潜在危害将再次显现出来。最后，作者将通过循环法说明通过结合所有上述因素，可以在同一样本上获得相差很大的结果。