Lithium niobate is a key material for photonics with applications in optical modulators and light frequency converters. Present lithium niobate optical devices are realized in bulk lithium niobate single crystals while interests in higher integration and also better performance derives efforts towards high quality epitaxial lithium niobate thin films. In the past, epitaxial lithium niobate thin films obtained using chemical vapor deposition (CVD), pulsed-laser deposition and crystal-ion slicing. Among these methods CVD is the one that can provide large area growth. Optimizing films properties stays an issue in CVD because in CVD of multi-element oxides using multiple precursors the stoichiometry of the film does not reflect the precursor ratios and additionally it varies with precursors total flux values. In this thesis optimization and discovery of multi-element oxide thin films were performed in a high-vacuum CVD reactor that is capable of performing combinatorial experiments. Using this reactor epitaxial lithium niobate thin films were obtained on sapphire {001g} and lithium tantalate {001} substrates with rocking curve full-width at half-maximum (FWHM) values of 0.03° and 0.024° respectively. The epitaxial growth of lithium niobate films was confirmed from X-ray diffraction (XRD) measurements and transmission-electron microscope (TEM) observations. Niobium tetra-ethoxy-di-methy-amino-ethoxide (Nb(OEt)4(dmae)) and lithium tert-butoxide (Li(OBut)) were used as precursors. Optically induced inhomogeneities, also known as optical damage, in lithium niobate limit its usage for high laser intensity applications. Lithium niobate single crystals doped with hafnium showed improved optical damage thresholds. Therefore combinatorial HVCVD experiments were conducted and textured {001} hafnium-doped lithium niobate films, with about 6 [mol%] hafnium, were obtained on sapphire f001g. Although the crystalline quality of the film is not high (rocking curve FWHM 0.8°) its Raman spectrum is still quite close to lithium niobate single crystal and epitaxial lithium niobate films on sapphire. Hafnium tert-butoxide (Hf(OBut)4) was used as precursor for hafnium. In the HVCVD reactor, a precursor transport system was applied such that spatial control of the precursor impinging rate on the substrate can be tailored. The precursor flux distribution that was mostly used in this thesis was a linear gradient with ratio of three across a 150 mm diameter circular area. In fact combinatorial experiments were performed by producing a linear gradient for individual precursors over the substrate with a certain angle between the gradient directions. The linear gradient feature was used to perform efficiently the systematic study of single precursors providing the benefit that one deposition at certain substrate temperature gives information about deposition kinetics for a vast range of precursor flux values. A systematic study of the Nb(OEt)4(dmae) precursor was performed using this feature and chemical-reaction-limited, mass-transport-limited, and desorption-limited regimes of CVD process were identified. The possibility of screening a large range of precursor flux values led us to discovering a CVD regime in which deposition rate is decreased by raising the precursor flux. A model was proposed for interaction of gas with surfaces that can explain all the different CVD regimes observed. The precursor molecules in HVCVD reactor reach the substrate with nearly no gasphase collision. Based on this fact a lift-off process was proposed for large-area in-situ patterning of multi-element oxide thin films.