Increasing capacitance density could be achieved by mainly two methods: increasing the capacitor node area and adopting higher dielectric constant (κ) material. The former relates mostly to integration issues, such as deep capacitor hole etching with an aspect ratio over 50 and filling the hole with the conformal electrode material. The latter is mainly a material issue, requiring appropriate material selection and an extremely conformal film deposition process. It also must be compatible with the electrode material, which means no adverse interfacial reaction and phase-pure high-κ film growth.Given the industry-compatible and matured electrode fabrication process of TiN grown by the atomic layer deposition (ALD) process, ZrO 2 has been the leading high-κ material in DRAM fabrication. This is because the crystalline structure of the thin-film ZrO 2 transforms from the medium-κ (≈20) monoclinic phase to the high-κ (>≈40) tetragonal phase due to the surface-energy effect. [6,7] Also, the ALD process of ZrO 2 is well matured to secure mass production. However, undoped ZrO 2 has suffered from a high leakage current problem. The problem could be ascribed to the local current flow through the grain boundaries of the polycrystalline ZrO 2 [8] and the n-type nature (the Fermi level is close to the conduction band edge) of the material by the presence of oxygen vacancies. [9,10] This problem has been overcome by adopting a thin Al 2 O 3 insertion layer (IL) for the relatively thicker ZrO 2 or doping the ZrO 2 with Al for thinner films. [8][9][10][11][12] However, such a strategy has sacrificed capacitance density owing to the inclusion of the lowκ amorphous Al 2 O 3 layer (κ ∼ 6-9) or the degraded crystallinity of the Al-doped ZrO 2 . [10,13,14] Therefore, the DRAM industry has spent the enormous effort to optimize the dielectric stack structure, but it has become evident that next-generation dielectric material is necessary for further scaling.Among the diverse candidates with even higher κ-values, SrTiO 3 showed a severe incompatibility with the TiN electrode. [15,16] Hf-doping into the ZrO 2 could be a viable and immediate option, but it involves a risk of loss of discharging density and slow operation speed. This problem is owing to the possible involvement of the (anti-) ferroelectric effect of the Hf-doped This study examines the influences of the Al 2 O 3 and Y 2 O 3 insertion layers (ILs) on the structural and electrical features of ZrO 2 thin films for their application to dynamic random access memory capacitors. The ultra-thin Al 2 O 3 IL (0.1-0.2 nm) dissolves into the ZrO 2 layers, which causes the top and bottom portions of the ZrO 2 film to merge and have smaller lattice parameters. However, the thicker Al 2 O 3 IL (>≈0.4 nm) forms a continuous layer and separates the top and bottom portions of the ZrO 2 film. Interestingly, the diffusion of Al does not occur in this case. Overall, the dielectric constant (κ) of the ZrO 2 /Al 2 O 3 /ZrO 2 film is lower than that of the undoped ZrO 2 film due to the involveme...