The photocatalyst beta-TaON is of interest due to promising properties, such as stability, suitable band gap for visible light, and carrier mobility. We implemented a combinatorial, material discovery approach that used pulsed laser deposition (PLD) for thin-film growth, X-ray diffraction (XRD) for phase determination, and machine learning for data reduction. A lateral compositional gradient of TaOxNy was grown across the surface of an alpha-Al2O3 (012) wafer. After annealing, XRD scattering patterns were collected across the lateral gradient. Unsupervised machine learning separated the XRD data into four clusters (phases); one of which turned out to be the desired monoclinic beta-TaON phase. Using high-resolution XRD, we determined that the beta-TaON region of the film was a 260 angstrom thick single-crystal epitaxial with the substrate, having out-of-plane beta-TaON (100)//alpha-Al2O3 (012) and in-plane beta-TaON (010)//alpha-Al2O3 (2 (1) over bar0). X-ray reflectivity (XRR) analysis of the beta-TaON region of the film showed an electron density matching that expected for beta-TaON. X-ray photoelectron spectroscopy (XPS) showed a Ta5+ valence state in the beta-TaON region of the film. This combinatorial approach, which produces a library of phases on a single wafer, proved to be very efficient for the growth of a material's phase of interest.