Time-effective, unsupervised clustering techniques are exploited to discriminate nanometric metal disks patterned on a dielectric substrate. The discrimination relies on cluster analysis applied to time-resolved optical traces obtained from thermo-acoustic microscopy based on asynchronous optical sampling. The analysis aims to recognize similarities among nanopatterned disks and to cluster them accordingly. Each cluster is characterized by a fingerprint time-resolved trace, synthesizing the common features of the thermo-acoustics response of the composing elements. The protocol is robust and widely applicable, not relying on any specific knowledge of the physical mechanisms involved. The present route constitutes an alternative diagnostic tool for on-chip non-destructive testing of individual nano-objects.

利用时间有效的，无监督的聚类技术来区分在介电基板上构图的纳米级金属盘。区分依赖于聚类分析，聚类分析应用于基于异步光学采样的从热声显微镜获得的时间分辨的光学迹线。该分析旨在识别纳米图案磁盘之间的相似性，并相应地对其进行聚类。每个簇以指纹时间分辨迹线为特征，综合了组成元素的热声响应的共同特征。该协议健壮且可广泛应用，不依赖于所涉及物理机制的任何特定知识。本路线构成了对单个纳米物体进行片上无损检测的另一种诊断工具。