Position Sensing using an Asymmetric Carbon Nanotube Dimer and a Tree-Based Classification Approach

This paper presents a nano-particle sensing system composed of an asymmetric carbon nanotube (CNT) dimer along with a tree-based classification approach. Electromagnetic coupling in asymmetric CNT dimers split the CNT plasmonic resonances into two distinct resonances known as the bonding and anti-bonding modes. The proximity of an external nano-particle (NP) to the CNT dimer perturbs the dimer's near-field distribution and causes different shifts in the bonding and anti-bonding resonances depending on the NP location. We have studied one such case of the NP sensing system, where the NP is lying on the dimer plane and moving perpendicular to the dimer axis. The NP movement is divided into six regions around the dimer and a dataset is created for 150 different locations of the NP by mapping them to the relative shifts in bonding and anti-bonding resonances. Finally, a tree-based machine learning algorithm is applied to fit the training data and predict the NP location for a given random pair of relative shifts in bonding and anti-bonding resonances. This new sensing modality predicts the NP location correctly in 90% of the test cases.