Carbon nanotube field-effect biosensors (CNT-bioFETs) are ultraminiaturized devices that can be used to measure single-molecule kinetics of biomolecules on time scales going from a few microseconds to several minutes, as demonstrated for nucleic acid hybridization [1] and folding [2] as well as for enzyme function [3]. Experiments indicate that the sensitivity of CNT-bioFETs originates from the interplay between the nanotube’s conductance, which is monitored by the device, and the electrostatic potential generated by the biomolecule under investigation, which is localized on the nanotube [4,5]. The measured conductance exhibits characteristic transitions between two levels (or more) as a function of time, as the biomolecule folds or performs its function. Yet, the origin of this electrostatic gating of the carbon nanotube by a single biomolecule is not well understood at the molecular scale.

To bridge this gap, we employ molecular dynamics (MD) and Hamiltonian replica exchange (HREX) simulations to unveil: (1) the interactions between the biomolecule and the nanotube to which it is attached in the device and (2) the electrostatic potential on the nanotube as the state of the biomolecule changes. We address these questions by considering three prototypical cases: the function of the Lysozyme protein, the hybridization of 10-nt DNA sequence and the folding of a DNA G-quadruplex, which were previously characterized using CNT-bioFETs [1-5].

Our simulations show that the lysozyme, the 10-nt DNA sequence and the DNA G-quadruplex interact differently with the nanotube to which they are attached. Consequently, the electrostatic potential (ESP) that they generate on the nanotube is very sensitive to the type and state of the biomolecule. When compared to experiment, the ESP distribution for the with-ligand and without-ligand states of the Lysozyme protein are in line with the measured two-level conductance by CNT-bioFETs. For the DNAs, however, the ESP distribution for their different states does not agree with the measured two-level conductance. Experiments imply that the DNA strand is not interacting with the nanotube, which is not what our simulations suggest. The reason for this apparent conflict could arise from the impact of the external electric field imposed by the gate electrode in CNT-bioFETs on highly charged systems such as DNAs, as supported by our recent simulations.

The significance of this work is twofold. First, it contributes to a better understanding of the inner working of carbon nanotube field-effect biosensors, which is crucially needed to support the development of these promising devices in the lab. Second, it provides the structural ensemble of the biomolecules and their interactions with the nanotube in these devices, which can serve as a starting point for a finer characterization of their effect on the carbon nanotube’s conductance at the ab initio level.

[1] S. Sorgenfrei et al. Nat Nanotechnol, 2011, 6, 126-132.

[2] D. Bouilly et al. Nano Lett, 2016, 16, 4679-4685.

[3] Y. Choi et al. Science, 2012, 335, 319-324.

[4] S. Sorgenfrei et al. Nano Lett, 2011, 11, 3739-3743.

[5] Y. Choi et al. Nano Lett, 2013, 13, 625-631.