Fabrication and Characterisation of a Carbon Nanotube Field Effect Transistor Aptasensor Platform

<p><strong>Carbon nanotube field effect transistors (CNT FETs) are known to be viable platforms for aptasensors, a type of biosensor. Little work has been done in determining the relationship between the CNT network and aptasensor sensitivity. This is partly due to the unavailability of fabrication methods to control CNT network semiconducting purity, CNT bundling, and CNT coverage. CNT FETs will have electrical properties which are related to their semiconducting purity and their network morphologies, specifically the amount of CNT bundling and CNT network coverage. It is expected that for unbundled CNT networks that at certain semiconducting purities and network densities that metallic-semiconducting (m-s) junctions will dominate conduction. These m-s junctions are known to be important to aptasensor sensitivity.</strong></p><p>In this thesis a new automated atomic force microscopy (AFM) analysis method to characterize CNT bundling and network coverage has been developed. The automated AFM analysis method uses a custom written MATLAB program which fits the height histograms of the images to Gaussian peaks. The positions and areas of the Gaussian peaks can be used to calculate average CNT bundle diameter and coverage. The automated AFM analysis method has been demonstrated to be able to the determine average CNT bundle diameters in CNT networks of various morphologies, which can be used to determine the presence of any CNT bundling in the networks. The average bundle diameters characterized by the automated method were found to be lower than the true average bundle diameters determined from manual tracing. Although automated analysis does not determine true average bundle diameter the average bundle diameters determined form automated analysis were found to be related to the true average bundle diameter and therefore can be used to compare across networks. The automated AFM analysis method has also been demonstrated to be able to estimate CNT network coverage. The automated method does not provide the true coverage of the CNTs as it includes broadening of the CNTs due to the AFM tip but does provide a value which will depend on network coverage allowing for comparison between networks of different CNT densities.</p><p>Various rounds of CNT network deposition have been performed with the intent of depositing unbundled CNT networks of varying CNT coverage and semiconducting purity. The fabrication rounds have been characterized using the new automated AFM analysis method. It has been found that unbundled CNTs can be deposited using aqueous surfactant CNT dispersions purchased from NanoIntegris, IsoNanotubes, deposited onto SiO2 functionalized with poly-L-lysine (PLL) as an adhesive for the CNTs. This method of deposition was found to be unable to deposit CNTs at a uniform and controlled coverage due to the coffee-ring effect, an effect that causes the dispersed CNTs to travel to the edge of a drying drop. It was also found that this deposition method is unable to deposit metallic CNTs from metallic IsoNanotubes solution, possibly due to the different surfactants used to disperse the metallic CNTs. It was found that a variation on the deposition method which involves doing deposition in a humid environment has allowed for better coverage control, and for metallic CNTs to be deposited at greater densities. Deposition of a network of 90% semiconducting CNTs followed by a network of 99% semiconducting CNTs has been demonstrated which may be able to be used to vary semiconducting purity.</p><p>Raman spectroscopy has been performed on the CNT networks made from various semiconducting purities: 99.9% semiconducting, 90% semiconducting, and 99% semiconducting with 90% metallic CNT spiking. Raman spectroscopy has been able to verify the CNT diameter ranges given by the supplier, 1.2 nm to 1.7 nm. Raman spectroscopy has also been able to verify network uniformity in terms of coverage and amount of bundling that had been characterized by AFM. Peaks associated with metallic CNTs have been identified in the G-bands with two laser energies, 2.41 eV and 1.96 eV, which are resonant with mostly semiconducting and mostly metallic CNTs respectively. These peaks have been able to confirm the addition of metallic CNTs for the 99% semiconducting networks which have been spiked with 90% metallic CNTs but more work is needed to correlate the peaks to actual metallic network purity.</p><p>Finally, CNT FETs fabricated from the CNT networks have been characterized electrically and used for sensing of adenosine and phenylalanine. It was found that the CNT FET characteristics of the 99.9% and the 90% networks did not vary significantly. Both had low subthreshold swings when gated using an ionic liquid gate, below 100 mV/dec. It was found that in some cases subthreshold swings were higher due to the presence of bundling in the networks. Both 99.9% and 90% semiconducting CNT FETs have been using for the sensing of adenosine, and both showed a maximum response at an adenosine concentration of 100 pM. Current responses at the 100 pM adenosine additions ranged from 1 nA to 30 nA with no differences in current response seen based on the CNT purity. Networks made with 90% semiconducting CNTs have also been demonstrated for the detection of phenylalanine.</p><p>The maximum response for phenylalanine sensing was at a phenylalanine concentration of 8.5 μM, and in this case current responses at the addition were between 1 nA and 6 nA. The differences in the sensitivity of the adenosine and phenylalanine sensing are likely due to differences in the aptamer structures.</p>