Unsupervised machine learning, and in particular data clustering, is a powerful approach for the analysis of datasets and identification of characteristic features occurring throughout a dataset. It is gaining popularity across scientific disciplines and is particularly useful for applications without a priori knowledge of the data structure. Here, we introduce an approach for unsupervised data classification of any dataset consisting of a series of univariate measurements. It is therefore ideally suited for a wide range of measurement types. We apply it to the field of nanoelectronics and spectroscopy to identify meaningful structures in data sets. We also provide guidelines for the estimation of the optimum number of clusters. In addition, we have performed an extensive benchmark of novel and existing machine learning approaches and observe significant performance differences. Careful selection of the feature space construction method and clustering algorithms for a specific measurement type can therefore greatly improve classification accuracies.

Single-molecule break-junction measurements are intrinsically stochastic in nature, requiring the acquisition of large datasets of “breaking traces” to gain insight into the generic electronic properties of the molecule under study. For example, the most probable conductance value of the molecule is often extracted from the conductance histogram built from these traces. In this letter, we present an unsupervised and reference-free machine learning tool to improve the determination of the conductance of oligo(phenylene ethynylene)dithiol from mechanically controlled break-junction (MCBJ) measurements. Our method allows for the classification of individual breaking traces based on an image recognition technique. Moreover, applying this technique to multiple merged datasets makes it possible to identify common breaking behaviors present across different samples, and therefore to recognize global trends. In particular, we find that the variation in the extracted molecular conductance can be significantly reduced resulting in a more reliable estimation of molecular conductance values from MCBJ datasets. Finally, our approach can be more widely applied to different measurement types which can be converted to two-dimensional images.

We aim to enhance the carbon fibre (CF)-matrix interface by synthesizing carbon nanotubes (CNTs) on the surface of the CF, creating a hierarchical composite. A 12 nm thick aluminium oxide film applied by atomic layer deposition (ALD) provides protection of the CF from deterioration during CNT growth in a chemical vapour deposition (CVD) process. However, the adhesion of alumina to CF, grown in classical water/trimethylaluminium ALD is severely diminishing during CNT growth, as detected by interface shear strength (IFSS) measurements. In our approach to improve the CF-alumina adhesion, we employed a pre-treatment of the CF with ozone and entirely replaced water with ozone in the ALD process, to promote the covalent bonding of the alumina to the CF surface. The current results show a new perspective in achieving the CNT synthesis on the CF while successfully mitigating its detrimental effects on the fibre mechanical properties.

We analyse the electrical response of narrow graphene nanogaps in search for transport signatures stemming from spin-polarized edge states. We find that the electrical transport across graphene nanogaps having perfectly defined zigzag edges does not carry any spin-related signature. We also analyse the magnetic and electrical properties of nanogaps whose electrodes have wedges that possibly occur in the currently fabricated nanogaps. These wedges can host spin polarized wedge low-energy states due to the bipartite nature of the graphene lattice. We find that these spin-polarized low-energy modes give rise to low-voltage signatures in the differential conductance and to distinctive features in the stability diagrams. These are caused by fully spin-polarized currents.