A novel image analysis method integrated with SEM-EDS is proposed in this study for assessing the dispersion state of nanosilica (NS) particles in the hardened cement matrix. The method entailed capturing a statistically significant number of SEM micrographs of a cement-NS paste and their EDS elemental maps, the latter of which revealed that at 3-days, distinct C-S-H gel with a consistently low Ca/Si ratio (<0.9) had formed through pozzolanic reaction. Thereafter, pixels with Ca/Si ratio of 0.6–0.9 were extracted from a micrograph using a ternary plot of the atomic percentages of Ca, Si, and Al. Following this procedure, binary images were obtained, based on which, for the first time, the agglomeration and dispersion of NS particles in hardened cement paste were characterized using various quantitative descriptors, such as fractal dimension and Delaunay triangulation. A relatively small mean absolute error (1.16 vol%) validated the accuracy of the proposed methodology.

本研究提出了一种与 SEM-EDS 相结合的新型图像分析方法，用于评估纳米二氧化硅 (NS) 颗粒在硬化水泥基体中的分散状态。该方法需要捕获大量具有统计学意义的水泥-NS 糊状物的 SEM 显微照片及其 EDS 元素图，后者显示在 3 天时，具有持续低 Ca/Si 比 (<0.9) 的独特 CSH 凝胶具有通过火山灰反应形成。此后，使用 Ca、Si 和 Al 原子百分比的三元图从显微照片中提取 Ca/Si 比率为 0.6-0.9 的像素。按照此程序，获得二值图像，在此基础上，首次使用各种定量描述子表征硬化水泥浆体中 NS 颗粒的团聚和分散，例如分形维数和 Delaunay 三角剖分。相对较小的平均绝对误差 (1.16 vol%) 验证了所提出方法的准确性。