This MSc research project is a feasibility study to explore the possibilities of using ferrofluid bearings in high precision, low load applications. Ferrofluids are fluids containing suspended ferromagnetic particles, which are influenced by magnetic fields due to their ferromagnetic properties. In similarity to for example iron powder, a ferrofluid moves towards the location with the highest gradient of a magnetic field. In the case of permanent magnets, the highest gradients occur on the edges. When the ferrofluid is applied on a magnet, pressure builds up within the ferrofluid, providing the ability to carry a load. In this case the ferrofluid acts as a hydrostatic bearing. The absence of stick-slip in these types of bearings makes them suitable for high-precision positioning systems. Ferrofluid bearings might typically be applied in precision XY-stages, as are used in biological microscopic research, where it is desired to image a specimen quickly with high resolution. This is commonly achieved by stitching multiple images of roughly 0.1×0.1 mm together to one full image. For this application, actuation in the planar degrees-of-freedom (DoF) is required. Since the ferrofluid bearings do not restrict rotations, all three (instead of two) planar DoF should be actuated. During planar motion, a ferrofluid trail is left behind, thereby lowering the stage with approximately 2 ?m/mm. When the vertical DoF are also actuated, the stage can bring (and keep) the microscopic sample into focus of the microscope. In order to apply the ferrofluid bearing to the mentioned microscopic application, a 6 DoF motion stage has been built with a planar range of 10×10 mm and a vertical range of 0.2 mm, see Figure 1. The moving stage has a mass of 0.15 kg, resulting in a steady state ferrofluid-film thickness of approximately 80 ?m. The stage is actuated by six Lorentz actuators. The magnetic field that is required for the Lorentz actuators is delivered by the same permanent magnets that are used for the ferrofluid bearings. This synergy between bearing and actuator makes the system compact and lightweight. Each magnet delivers the magnetic field for two coils; one for vertical- and one for planar actuation. The six coils of the Lorentz actuators are etched in a 4-layer printed circuit board (PCB). This PCB simultaneously acts as the surface on which the ferrofluid bearings can move. The non-linear position dependency of the actuator-forces is modelled in Matlab and it is verified that this model describes their behaviour with 95% accuracy. This model is also implemented in the control scheme. Possible microscopic samples can be placed on top of the moving stage. The bottom of this stage (see Figure 2) consists of three magnets, three interferometer mirrors, a PMMA support structure and an iron top-plate. This iron top-plate is used as a target for the capacitive sensors and a flux-path for the magnets, thereby increasing their efficiency. The Lorentz coils are an inductive load, so feedback current amplifiers are required to drive these loads with a flat (input to force) frequency response. Custom-made amplifiers were designed and manufactured of which the gain and current-limit can be altered easily. Their performance was tested up to 30 kHz, resulting in a flat frequency response and a phase lag of only 5 degrees while the cumulative noise-level up to 1 kHz remains lower than 0.025 mA. For validation of the system, a capacitive/interferometric measurement system was used to provide position feedback. These high resolution sensors are by far the most expensive components of the setup. In future applications these sensors could be replaced with for example position sensitive detectors (PSDs) and integrated in the same PCB. The 6 DoF demonstrator stage is a multiple-input-multiple-output (MIMO) system, but decoupling matrices decompose it into six single-input-single-output (SISO) systems, each having their own controller. The control scheme is built in dSPACE, which executes the high-level programming language of Matlab/Simulink. Open-loop measurements show that up to 500 Hz the planar motion can be treated as a pure mass, so that a simple PID controls are sufficient to control the planar motion up to this frequency. Due to the highly damped vertical motion, a PI-controller is sufficient for out-of-plane control. The behaviour of the demonstrator stage is slightly dependent on its vertical position: when the vertical position is increased, the surface area of the ferrofluid is decreased, causing a lower damping. Planar steps of 0.1 mm settle within 10 nm in 0.03 seconds. The vertical motion has a control bandwidth of 100 Hz, 250 nm steps in this direction have a settling time (position within ±1% of reference) of 0.02 seconds. These specifications allow the system not only to be used in digital microscopy but also in other applications which require higher specifications, such as white light interferometry, where typical vertical steps of half the light’s wavelength (250 nm) are required. The settling times are mainly limited by the maximum force of the actuators. When faster settling times are required in future applications, the system can easily be altered to meet those requirement by adding more layers to the PCB and/or increasing the number windings at each layer. It can be concluded that a system with ferrofluid bearings can be used in low load applications were fast, high precision positioning is required. The ferrofluid bearings do not seem to limit the precision that is achieved by the demonstrator stage.