Optical manipulation at the micro- nano-scale is a fascinating topic due to its inherent non-invasive properties and multifaceted applications in various fields such as biology, sensing, micro-fluidics, and micro- nano-robotics. This thesis involves intensive efforts dedicated to realizing the rotational motion and trapping. Despite recent developments, selecting the most efficient and powerful nanomotor is still not obvious. Therefore, this thesis provides a comprehensive understanding of the current state of affairs for optically-driven motors and solutions for nanomotor design, numerical analysis, and nanofabrication.

I demonstrate that a machine learning algorithm as a global solution can produce nanorotor structures generating improved optical torques for a given incident light. The combination of two different fields, machine learning, and nanomotors, via nanofabrication requires substantial optimization processes in deep learning algorithms, fabrication parameters, and optical measurements, which have been described in great detail in this thesis. A torque predictor made of a convolutional neural network (CNN) is connected to a deep convolution generative adversarial network (DCGAN) nanorotor generator. An electromagnetic surface integral equation (SIE) scheme provides a numerical method to obtain the scattering cross-section (SCS), multipole expansion, optical torque, and electric field enhancement distribution for the generated nanorotors. The nanorotor structures obtained from the algorithms are realized by nanofabrication techniques such as electron beam lithography (EBL), thin film deposition, and various etching processes, and are implemented in a SiO2 embodiment to build nanomotors. The nanorotors suggested by the artificial intelligence exhibit improved rotation speeds compared to conventional designs based on V-shape, rods, or gammadion structures.

The realization of sophisticated structures at the nanoscale is critical for linking an ideal design with real devices. The importance of ion beam etching (IBE) is growing due to its ability to fabricate tiny gaps and sharp features for plasmonic responses. However, as the required features go down to a few tens of nanometers, inevitable fences, which occur due to physical bombardment and redeposition, cause the expansion of the structure. This gives rise to unexpected behaviors for the fabricated nanostructures. To quantify and solve these issues, a series of structures have been fabricated using different etch mask heights and, subsequently, analyzed by investigating their cross-section.

Optical trapping has attracted a lot of attention due to its promising applications. In addition to the rotating motion described above, we introduce in this thesis an experiment aimed at visualizing trapping events in plasmonic apertures in this thesis. The idea relies on a combination of optical trapping and Foster resonance energy transfer (FRET) between trapped fluorescent particles (FRET donor) and the plasmonic apertures that have been functionalized with FRET acceptor molecules. A dyed polystyrene (PS) beads solution is injected on top of the plasmonic nanostructures in a confined chamber. For the FRET measurement, a laser is focused onto the nanoaperture using an objective. The dyed PS beads are drawn to the nanoaperture by optical forces, and FRET occurs once PS beads are close enough to the functionalized plasmonic nanoaperture.