Samenvatting

A new gravitation wave observatory is being built in Japan. This facility is a big Michelson interferometer where the mirrors and sensors will be made cryogenic down to 4 K. The problem is that: the heat cannot be transported out of the mirror by its surroundings because the whole system is in an ultra-high vacuum. Also a fraction of the LASER power, in the order of 2 W, will be stored in the mirror. This heat can only be transported using the suspension of the mirror. To transport this heat, the wires of the mirror can be made real thick. But now the suspension will be to stiff which decreases the vibration noise properties. It will also still take up to one year before the mirrors reach a temperature of 4 K. For a good transportation of the heat, the idea is that: thick sapphire wires will be used to hang the mirror and for good vibration properties silicon cantilevers will hang
the mirror with a low vibration frequency. The bending performance of silicon cantilevers depends highly on the surface conditions because silicon is a brittle material.
The goal of this research is finding the best cantilever blade condition which insures good bending performance. The bending performance depends on several surface conditions. The surface needs to be smooth and without cracks. These cracks cause weak spots in the cantilever. To minimize the cracks several cutting methods are performed, namely: cutting by hand, machine-cutting and High Reactive ion etching. Chemical etching of the edges is used as an after-cut-treatment in attempt to minimize the cracks.
To measure the stress at a certain point of the blade an experimental setup has been made. This setup can compress a blade till it buckles and eventually breaks. The compression distance _x, the compression force Fx, the tip angle _ and the height of the blade center h can be measured. Using this quantities, the stress _ at the surface of the blade can be calculated. The compression distance will be made larger in small steps till the blade breaks. The measurement right before the blade breaks will be used to calculate the stress.
A numerical analysis has been made to support the measurements. This analysis delivers the expected blade shape at any state of bending. Note that this does not predict the breaking of the blade. The forces and stresses in the blade can also be predicted if the E-modulus is known.
The first measurements have been performed with a diamond knife-cut edge. This is done by hand using a protractor and a diamond knife. The average maximum feasible stress is 280 MPa. This blade does have a large breaking bandwidth with a standard deviation is 130 MPa. Big cracks at the edge of the blade ensures weak spots. This weak spots are unwanted, therefore a different solution had to be found.
The second set of measurements are also performed with a diamond knife only now, a cutting