Aluminum nitride thin film development using statistical methods

Extensive studies on reactively magnetron sputtered aluminum nitride (AlN) thin films and the evaluation of the material properties influenced by the deposition parameters were performed utilizing statistical methods. The use of the inverse piezoelectric effect of poly-crystalline AlN thin films in actively deformable micro mirrors are of prior interest for this work. To achieve piezoelectric material properties but also to respect technological conditions in MOEMS manufacturing processes the textural quality, the grain size, the intrinsic material stress, the deposition rate and therefore the non-uniformity in layer thickness are investigated. Wide, randomized series of experiments on process pressure, nitrogen / argon gas flow ratio, plasma rf power and target to substrate separation of the AlN sputter deposition process on amorphous titanium aluminid thin films on silicon substrates were performed. Polynomial based models of the thin films properties influenced by the deposition parameters are presented. The qualities of these models are evaluated by statistical methods. With the use of these models advantageous set points of the deposition process are presented. This set points enables highly textured polycrystalline AlN films, low or zero stressed films, big grain size and low non-uniformity in layer thicknesses.