| Summary: | 碩士 === 國立清華大學 === 電子工程研究所 === 87 === In recent years, some perovskite-type ceramic thin films of high dielectric constant have been proposed to be a promising material for charge storage capacitors in Gbit DRAM cells. Regarding their device applications, the electrical properties, such as low leakage current, fast dielectric response, low dielectric loss, high dielectric breakdown and long lifetime are especially important in device quality and reliability. The targets for ideal 256Mbit and Gbit era DRAM dielectrics include the following :
(1) SiO2 equivalent thickness,<0.5 nm for 256 Mbit,0.2 nm for Gbit.
(2) Leakage current density <1x10-7 A/cm2 at 100 kV/cm.
(3) Lifetime 10-year at 85oC and 1.6V.
(4) Stability 1015 cycles at > 100 MHz.
(5) General compatibility to semiconductor processing.
Some perovskite-type ceramics, such as BaTiO3 is the most widely known ferroelectric oxide, and has been intensively investigated as the most suitable candidate for memory devices because of the high dielectric constant. Fig. 1-2 shows the dielectric constant vs. temperature plots for ferroelectric ceramics. As shown in the figure, there exists ferroelectric to paraelectric phase transition temperature-curie temperature. Although the dielectric constant of ferroelectric ceramics is much higher than traditional dielectric material, it is not stable between DRAM operation temperature since the curie temperature is high. Recently, some perovsikte ceramics have been investigated for DRAM application such as BST (BaxSr1-x)TiO3 and PZT Pb(Zr,Ti)O3 etc. which is stable between DRAM operation temperature because of low curie temperature.
Another issue for applying perovskite ceramics on DRAM application is leakage current. The reduction of dielectric dispersion and leakage current is very important because it decreases the device performance and the electric charge once stored. It is expected that defect states in the bulk film and the interfacial between bottom electrode and bulk film control the mechanism of leakage current and dielectric dispersion. Besides, the defect states deeply influence the ac response at high frequency. Many efforts have been made to improve the film properties such as post-annealing in an oxidizing ambient after film deposition is need to reduce the dielectric dispersion and the leakage current. Pure O2 is the oxidizing ambient usually used in post-annealing.
Ba(Ti1-xSnx)O3, which is a solid solution of BaTiO3 and BaSnO3, exhibited ferroelectric transition at temperatures between 0 and 60℃ when the x is between 10 and 20 [9]. It has been proposed that the cubic phase of bulk Ba(Ti1-xSnx)O3 will be actually fully stabilized at room temperature if the concentration of Sn is above 20%, and the cure temperature is about 20oC [10]. Therefore, Ba(Ti1-xSnx)O3 with the concentration of Sn larger than 20 mole % is expected to exhibit good paraelectric characteristics.
In this thesis, Ba(Ti0.8Sn0.2)O3(BTS) thin films are deposited by magnetron radio-frequency(RF) sputtering method and we discuss the effects post-annealing after deposition BTS thin films. The electrical and dielectric property of BTS films on Pt/TiO2/SiO2/Si(100) substrates with different post-annealing conditions are studied, and the leakage current density is extremely low for DRAM application. For development of VLSI technology, we propose the equivalent circuits of our BTS thin film capacitors at different frequency range, and the defect in BTS films is also analyzed by complex plane analysis. Besides, the electronic noise due to defect states is also discussed.
Here, we summarize the major contributions of this work.
1. The Ba(Ti0.8Sn0.2)O3 thin film deposited by magnetron RF sputtering can be applied to Gbit DRAM because of high dielectric constant and low leakage current. Our BTS thin films are paraelectric phase and are stable during DRAM operation temperature.
2. The equivalent circuit of BTS thin film capacitors in different frequency range was proposed. From the equivalent circuit, the resonant frequency was observed and the dielectric relaxation can be explained. By analyzing the resonant frequency, we conclude that dielectric relaxation is due to grain boundary defect and interface defect, instead of shallow trap term.
3. The electrical noise vs. bias was observed. It is shown that higher electrical field induce higher noise. By analyzing noise at different temperature, the noise is dominated by defect in the dielectric film, instead of thermal contribution.
4. N2O shows more efficient than O2 to repair defect in BTS thin film during RTA process. This is quite agree with the results of our experiments and proposed literature.
In this thesis, the AC and DC dielectric properties of BTS thin film have been discussed. The experiments in this work is very useful to characterize dielectric material and is very useful to VLSI technology.
|
|---|
