Flater Erin E, Zacharakis-Jutz George E, Dumba Braulio G, White Isaac A, Clifford Charles A
Department of Physics, Luther College, 700 College Drive, Decorah, Iowa 52101, United States.
Department of Physics, Luther College, 700 College Drive, Decorah, Iowa 52101, United States.
Ultramicroscopy. 2014 Nov;146:130-43. doi: 10.1016/j.ultramic.2013.06.022. Epub 2013 Jul 9.
Quantitative determination of the geometry of an atomic force microscope (AFM) probe tip is critical for robust measurements of the nanoscale properties of surfaces, including accurate measurement of sample features and quantification of tribological characteristics. Blind tip reconstruction, which determines tip shape from an AFM image scan without knowledge of tip or sample shape, was established most notably by Villarrubia [J. Res. Natl. Inst. Stand. Tech. 102 (1997)] and has been further developed since that time. Nevertheless, the implementation of blind tip reconstruction for the general user to produce reliable and consistent estimates of tip shape has been hindered due to ambiguity about how to choose the key input parameters, such as tip matrix size and threshold value, which strongly impact the results of the tip reconstruction. These key parameters are investigated here via Villarrubia's blind tip reconstruction algorithms in which we have added the capability for users to systematically vary the key tip reconstruction parameters, evaluate the set of possible tip reconstructions, and determine the optimal tip reconstruction for a given sample. We demonstrate the capabilities of these algorithms through analysis of a set of simulated AFM images and provide practical guidelines for users of the blind tip reconstruction method. We present a reliable method to choose the threshold parameter corresponding to an optimal reconstructed tip shape for a given image. Specifically, we show that the trend in how the reconstructed tip shape varies with threshold number is so regular that the optimal, or Goldilocks, threshold value corresponds with the peak in the derivative of the RMS difference with respect to the zero threshold curve vs. threshold number.
对原子力显微镜（AFM）探针尖端的几何形状进行定量测定，对于可靠测量表面的纳米级特性至关重要，包括对样品特征的精确测量以及摩擦学特性的量化。盲尖端重建是指在不知道尖端或样品形状的情况下，通过AFM图像扫描来确定尖端形状，这一方法最著名的是由Villarrubia [《美国国家标准与技术研究院研究报告》102 (1997)] 确立的，自那时起一直在不断发展。然而，由于对于如何选择关键输入参数（如尖端矩阵大小和阈值）存在模糊性，而这些参数会强烈影响尖端重建的结果，这阻碍了普通用户实施盲尖端重建以产生可靠且一致的尖端形状估计。本文通过Villarrubia的盲尖端重建算法对这些关键参数进行了研究，在该算法中，我们增加了让用户能够系统地改变关键尖端重建参数、评估一系列可能的尖端重建结果，并为给定样品确定最佳尖端重建的功能。我们通过分析一组模拟AFM图像展示了这些算法的能力，并为盲尖端重建方法的用户提供了实用指南。我们提出了一种可靠的方法，用于为给定图像选择与最佳重建尖端形状相对应的阈值参数。具体而言，我们表明重建尖端形状随阈值数量变化的趋势非常规律，以至于最佳（即“恰到好处”）的阈值与均方根差相对于零阈值曲线与阈值数量的导数峰值相对应。
