Abstract
This thesis focuses on the advanced devices based on the metal-insulator-metal (MIM) structure, the first one is MIM capacitor, and another one is resistive random access memory (RRAM). The first part is MIM capacitor, our works revolve around how to suppress the quadratic voltage coefficient of capacitance, so called, VCC-α. However, VCC-α will become larger as the dielectric constant of oxide increases that brings out our main goal, how to increase the capacitance density but suppress VCC-α at the same time. In this part, we adopted amorphous TiO2 and Y2O3 stack as dielectric of MIM capacitor, the main reason is that yttrium has larger atom radius would let TiO2 maintain in amorphous phase. Hence, after annealed TiO2/Y2O3 shows higher capacitance density but TiO2 stays in amorphous state at the same time. Due to capacitance density and VCC-α has an inverse relationship, so we need to utilize another way called “canceling effect” of VCC-α. Canceling effect is achieved by two different oxide films, one has positive VCC-α and another one has negative VCC-α. Compare to adopt only one oxide film, once stack these two oxide films the effective VCC-α will become smaller due to positive VCC-α and negative VCC-α compensates each other. Furthermore, by utilizing the crystalline TiO2 to get higher capacitance density, but the VCC-α will become larger rapidly at the same time. To reduce the VCC-α, nitrogen plasma treatment on the surface of crystalline TiO2 was adopted; moreover, to further suppress the VCC-α, SiO2 is stacked on the crystalline TiO2 because of SiO2 has negative VCC-α that can compensate the positive VCC-α introduced by crystalline TiO2. In this work, MIM capacitor shows 30 ppm/V2 of VCC-α, 11.2 fF/μm2 of capacitance density, these results fit the requirements of MIM capacitor in 2018 by ITRS. But compare to novel metal oxide, SiO2 has much lower dielectric constant (κ) value, so the next part is replacing SiO2 to ZrTiOx. It is worth to mention that ZrTiOx demonstrates negative VCC-α characteristic like SiO2, but has much higher κ value, 22.5. Thus, the results of this work we got a much higher capacitance density of 14.38 fF/μm2 and 68 ppm/V2 of VCC-α. In the RRAM topic, we spent lots of effort to discuss how the property of ZrTiOx affects the electrical characteristics of RRAM. First, we explored the influence of crystalline ZrTiOx compared to amorphous one. Second, we examined how the position of IL effects switching mechanism, based on this, we planned three experiments each has different IL position, one has no IL, another has an IL on the bottom electrode and the last one has IL on the both top and bottom electrode. In this series experiments, we found that amorphous ZrTiOx demonstrates better uniformity, lower switching voltage, faster switching speed and larger sense margin; on the other hand, the position of IL is a critical factor that affects switching voltage, thus, IL could be modulated to achieve higher switching speed and better power consumption. In summary, one-sided IL shows better electrical characteristics. In addition, if we want to achieve higher storage capacity, 3D stack structure is an inevitable solution; however, to avoid the misread situation due to unexpected current flow, for one RRAM device we need another diode to form a 1D1R structure. In this work, we combined a simple metal-semiconductor diode and a RRAM to a TaN/ZrTiOx/Ni/n+-Si structure, the advantages are following, first, this structure is compatible with the incumbent ULSI technology; second, reduced the cost due to discard the expensive and hardly processing metal, such as W and Pt. Additionally, this structure is combined the one-sided IL formation that will further improve the electrical characteristics of RRAM.