The work was focused on three major aspects: (a) Development of a robust protocol for generating an oil-water interface decorated with nanoparticles, and basic characterization of particle motion; (b) Construction of a high-speed total internal reflection scattering microscope; and (c) High-speed tracking and motion analysis of nanoparticles at fluid interfaces.
These tasks defined five milestones of the project, which were successfully achieved:
Milestone 1: The protocol for oil-water interfaces with nanoparticles was generated. Special emphasis was given to: preventing NP aggregation, the stability of the interface for a few hours (to allow sufficient time to probe it), cleanliness of the interface, and choice of oil which is refractive index-matched to glass as well as having the appropriate viscosity.
Milestone 2: Characterization of NP motion at low speed. This included the acquisition of movies, particle positioning and tracking, as well as basic motion analysis such as MSD and diffusion coefficient extraction.
Milestone 3: The TIR-scattering microscope was constructed. Originally we thought interferometric scattering microscopy (iSCAT) will be used to track the nanoparticles, however, TIR-scattering turned out to be more suitable for our experiments because of its intrinsic ability to image large fields of view.
Two approaches were taken, both of which were based on objective type-TIR; (i) two micromirrors (MM) that guide the illumination light into the objective and collect the reflected light from the interface, where the scattering light from the NPs is collected via a 45 mirror underneath the two MM, and (ii) central 45 rod mirror (RM) with diameter of 80% of the back focal plane of the objective, where the illumination and reflected light paths pass the sides of the RM. The two approaches were constructed and tested for sensitivity as well as precision and both showed excellent performance, with nm precision at exposure times shorter than 10 microseconds and large field-of-view of ~50x50 micrometres squared. We combined the two set-ups with a high-performance fast CMOS camera that we hired for the purpose of the experiments.
Milestone 4: Performing a series of experiments of nanoparticle diffusion at oil-water interfaces, varying the size and material of nanoparticles (20 and 30 nm in diameter, and gold and silver). In addition, we acquired movies with simultaneous fast (10-100 kHz) and slow (25-75 Hz) imaging speed so we could compare the dynamics at the two-time scales. In addition, we acquired movies capturing the landing (absorption) events of a nanoparticle to the oil-water interface.
Milestone 5: Advanced motion analysis of all experiments. This included characterization of displacement correlation as a function of length-scale at all the different conditions (fast vs low time-scales, different particles sizes and materials). In addition, an analysis of the single particle motion was performed to extract the effects of meta-stable states of NPs bound to the interface, as well as the dynamics when a particle is absorbed.
Our experiments revealed that fluid interfaces have a unique inertial response to perturbation at short-timescales, which is different than the inertial response of bulk fluids. This is surprising as the interface itself is a thin boundary layer between two bulk fluids. We showed that the response, which decays logarithmically and is anti-correlated at large distances, is a result of propagation of thermally-excited capillary waves at the interface. This unique behaviour disappears as the capillary waves decay, and therefore could not be captured at slow imaging.
Our experiments also revealed signatures of different stability states of NPs at fluid-interfaces. The dynamics at the single particle level showed that although the motion can be characterized as normal diffusion (or Brownian), it is not Gaussian, i.e. its Van Hove distribution exhibits features of different processes, related for example, to different meta-stable states of the NP at the interface. This is not the case at long time-scales, where the motion is characterized by a Gaussian distribution, since the lifetime these states are short, and usually averaged out at video-rate imaging. Interestingly, the absorption dynamics itself, i.e. when a NP lands at the interface and reaches its equilibrium state, is extremely fast (not resolvable by our imaging tools), in contrast to micro-particles, which slowly absorb to their equilibrium position.
