Sorting of carbon nanotubes by their chirality is the current bottleneck in the way to their broad employment based on their exceptional electronic and optical properties. Despite the extensive effort, there is no known method, which would result in really pure chirality ensembles. Previously reported sorting protocols result in enrichment rather than in sorting, alter electronic structure, and suffer from low yield. This is mostly due to the statistical approach, where the nanotubes with mixed chiralities are treated as a set. In this thesis, we propose a new sorting technique based on nanotube-by-nanotube compartmelization, characterization, and sorting in a continuously running droplet-based microfluidic device. A new microfluidic platform for droplet-based experiments in femtoliter scale has been fully developed in this work. We report manufacture of full-glass spectroscopy-friendly microfluidic chips with the characteristic length of channels below 3 μm. A novel procedure for immersed planar metallic electrodes suitable for the harsh glass processing is introduced. Selective treatment of the channels with a hydrophobic self-assembled monolayer is implemented for stable water-in-oil droplet microfluidics. Several unique properties of the system: the size, the unorthodox capacity of low droplet formation frequency, effective trap system, and tailored fluorescence detection system, were developed with the aim of lowering the detection limit down to single nanotube level. The automated processing of fluorescence spectra triggers the dielectrophoretic sorting valve deflecting the nanotube carrying droplets to either the reservoir or to the waste. Besides the primary goal, this microfluidic platform represents a powerful experimental tool to be employed in various fields of research. A method of individualization of carbon nanotubes in an aqueous dispersion based on sonication and centrifugation is systematically addressed in this thesis. The purity, level of individualization, quality of individualized nanotubes, and long-term stability are found to be critically dependent on the sonication parameters, mainly the sonication power. For the purpose of water-in-oil droplet microfluidics, a unique protocol for depletion of the surfactant at unaltered level of individualization is reported. The development described in this thesis brought the project to the very doorstep of automated carbon nanotubes sorting, one nanotube at the time. We believe that successful realization of the sorting would allow a major breakthrough in small-scale applications of single nanotube devices with precise chirality requirements for achieving the desired behavior.