Detection and classification of host–guest interactions using β-cyclodextrin-decorated carbon nanotube-based chemiresistors

• Electronic detection of chemicals using β-CD-decorated CNT chemiresistors.
• We studied the effect of CNT surface modification on the sensor performance.
• We used correlation functions with ANN to demonstrate selectivity.
• PTCA linked β-CD-decorated CNT chemiresistors outperformed devices without PTCA.
• The distinct correlation patterns highlight the selectivity of the sensor.

Carbon nanotubes (CNTs) in a chemiresistor setup have been widely explored in bio/chemical sensing. Detection of certain molecules with environmental and health related importance such as 9-anthracenecarboxylic acid, diclofenac sodium, and curcumin using electrochemical methods/unfunctionalized CNTs suffer from lack of response, high limit of detection (LOD) and poor selectivity. The key to overcome these issues is to decorate CNTs with host (receptor) molecules like β-cyclodextrin (β-CD) that interact with guest (target) molecules by host–guest complex formation. To improve guest recognition, and consequently, the sensor performance, effective immobilization of β-CD on the CNT surface using a non-covalent bridging molecule such as 3, 4, 9, 10-perylene tetracarboxylic acid (PTCA) is required. Furthermore, the selectivity can be assessed using the conductance correlation patterns of different host–guest systems in conjunction with a pattern classification tool. Our results indicate that PTCA linked β-CD-decorated CNT chemiresistors showed a good linear detection range (∼100 pM–100 nM), sensitivity (∼3 × 10−3–9 × 10−2 nM−1) and LOD (∼62 pM–101 nM), compared to devices without PTCA, in the detection of the guest molecules. The distinction in correlation patterns of different host–guest systems was corroborated by pattern classification yielding a classification accuracy, sensitivity, and specificity of ∼91.83%, ∼90.13%, and ∼85.39%, respectively.