A reference-free clustering method for the analysis of molecular break-junction measurements

Abstract

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.

Unravelling the conductance path through single-porphyrin junctions

“…To further investigate the charge transport pathways, an unsupervised clustering algorithm has been used to subdivide the different datasets in four different classes 20. Classes A–C are associated with the presence of a molecule inside the junction, whereas class D (see Fig.…”

“…For this reason, we synthesized seven different compounds with closely related geometry features, and measured them using an automatized mechanically controlled break-junction technique (MCBJ). By employing an unsupervised clustering algorithm19,20 on a unique set of data consisting of almost 100 000 conductance traces, we identify classes of common behaviours in the breaking traces and their correlation with structural features of the molecules under investigation enables us to link each of them to a different electron pathway across the porphyrins.…”

Connectivity dependent thermopower of bridged biphenyl molecules in single-molecule junctions

“…In this case the so-called junctioncharacteristic plateau can be observed in the IZ curve. Using a non-supervised clustering technique (k-means in Matlab) [28][29][30] through which we can identify and separate individual IZ curves with similar behaviors, we could easily separate curves with a molecular junction from others that presented just a tunneling behavior. With all the resulting curves, those with a signature of a molecular junction being formed, the corresponding conductance was calculated dividing by the bias voltage applied (V bias = 200 mV) and conductance 1D histograms were built, shown in Fig.…”

Conformation-dependent charge transport through short peptides

“…These short plateaus are found for all peptide junctions studied ( Fig To help sorting the plateaus, we employ a reference-free classification algorithm based on clustering of the data with the K-means++ method. 28 The algorithm groups traces that have similar one-and two-dimensional histograms, thus highlighting their most prominent features. Figures 2c,d show two of these classes which clearly display distinct features around a particular conductance value both in the one-and twodimensional histograms.…”