|
|
|
Mechanistic Study of Metalorganic Chemical Vapor Deposition of (Ba,Sr)TiO3 Thin FilmsY. Gao, C. L. Perkins,(a) S. He,(b) T. Tran, S. Thevuthasan, M. A. Henderson, P. Alluri,(c,d) and T. Baum(e) Supported by OBER EMSL Operations, OBER/LTR, and Motorola, Inc. During the last two decades, the storage density of DRAM has increased nearly fourfold every three years (SIA 1997). Each successive DRAM generation has been required to maintain the same bit cell capacitance while the amount of lateral space available in the DRAM bit cell decreased significantly. To maintain adequate capacitance in such small lateral areas, DRAM capacitors have employed increasingly complex three-dimensional cell structures with reduced dielectric film thickness. Currently, the thickness of the dielectric films is limited by high leakage currents while the intricate cell geometry adds complexity to the processing that ultimately impacts yield and manufacturing cost. An alternative approach to maintaining the required amount of charge on the bit cell capacitor is to use new materials, such as (Ba,Sr)TiO3 (BST), which have higher dielectric constants than traditional dielectrics. Although much progress has been made in the preparation of BST thin films using MOCVD, the precursor chemistry and film-formation mechanisms are still not well understood. The absence of such knowledge greatly hinders the development of the MOCVD process for the preparation of device-quality BST thin films. We have studied MOCVD growth of BST thin films using Ba(thd)2, Sr(thd)2 and Ti(O-iPr)2(thd)2 precursors, focusing on understanding the deposition chemistry, the origin of high-step coverage, and the relationship among the reaction chemistry and kinetics, film microstructure and crystallinity, conformal step coverage, and dielectric properties. The mechanistic study of MOCVD growth of BST thin films was carried out using surface science techniques in conjunction with a MOCVD study. Several mechanistic questions have been addressed using isotopic labeling experiments both in a surface science apparatus and under real MOCVD processing conditions. Some of the results have been reported elsewhere (Gao et al. submitted, 1999). The precursor chemistry study was carried out in a UHV chamber equipped with a quadrupole mass spectrometer (QMS), a differentially pumped ion gun, a double pass cylindrical mirror analyzer, a rear-view LEED apparatus, a three-sector HREELS spectrometer and various gas dosing capabilities (Henderson 1996). A series of temperature programmed desorption (TPD) experiments were performed on adsorbed Sr(thd)2 molecules to study gas phase products of the reactions between the substrate and the precursor molecules with and without gas phase oxygen. In an oxygen ambient, TPD shows oxidation of Sr(thd)2 on the SrOx precovered surface to have at least four different reactions which involved the removal of carbon from the precursor ligands (Figure 4.6). In two of these, gas phase oxygen was incorporated into the oxidative products. The other two processes that do not directly involve gas phase oxygen for a carbon removing reaction could also be associated with the ligand oxidation because these peaks are absent in the TPD traces without an oxygen background. The chemistry of Sr(thd)2 on SrOx/Pt(111) without an oxygen background was also explored. It is apparent that there is much less chemistry occurring as witnessed by the disappearance of all TPD peaks. The chemistry of Sr(thd)2 also depends on the substrate material. The absence of the higher temperature peaks for Sr(thd)2 on clean Pt(111) indicates that the oxidation of Sr(thd)2 molecules could complete at lower temperatures on clean Pt than on SrOx covered Pt. BST thin film growth was performed in an oxide MOCVD system equipped with a downstream microwave plasma source, a Fourier transform infrared spectrometer for in-situ gas-phase analysis, and a multi-wavelength ellipsometer for monitoring growth rate and film thickness (Gao 1999). Isotopic labeled 18O2 experiments were carried out to understand the film-formation reactions under MOCVD growth conditions because oxygen in the BST films can originate from both the gas phase oxidants (18O), and the (thd) and (O-iPr) ligands (16O). TOF-SIMS analyses reveal both M18O and M16O (M = Ba, Sr, Ti) in these BST films, indicating that both oxidation and thermal decomposition of the precursor molecules occur under these growth conditions. Figure 4.7 indicates that the (thd) and (O-iPr) ligands are only partially substituted by the 18O2 (i.e., oxidation of the precursor molecules) and are an important source of M-O bonds for the growing film. The retention of 16O in the BST film without incorporation of carbon suggests that the cleavage of the O-C11H19 and O-C3H7 bonds occurs for the (thd) and (O-iPr) ligands, respectively. The cleavage of those O-C bonds was caused by both thermal decomposition as in the case of the growth of BST films in inert atmosphere, and by an intermediate role of the gas phase oxygen to remove carbon and hydrogen from the precursors. The latter was confirmed by the TPD results (Figure 4.7), which reveal that the gas phase oxygen attacks the O-C11H19 bonds to form CO, CO2, and other hydrocarbon species.
The M18O/M16O ratios (Table 4.1) decreased considerably for the BST film prepared using 18O2 plasma enhancement. This suggests that the more reactive oxygen radicals produced by the microwave plasma play a much bigger role in the film formation reactions by breaking the O-C bonds rather than substituting the precursor ligands, compared to molecular oxygen. This result is of interest because use of oxygen plasma in BST film processes typically increases the film surface roughness. It is likely that rapid removal of hydrocarbons from the precursor molecules by oxygen plasma reduces the surface mobility of the reactants, resulting in a rougher surface. In addition, the BST film deposited at 590°C in 18O2 show higher M18O/M16O ratios than those in the BST film prepared at 650° C in 18O2, indicating that more precursor molecules were directly oxidized at the lower temperature. Thus, at low deposition temperatures, the thermal decomposition of the precursors becomes less dominant relative to the oxidation of the precursors for the growing film, resulting in greater incorporation of 18O in the film. Compared to the film grown at 650°C in 18O2, the M18O/M16O ratios for the BST film at 650°C in a 50%18O2-50%N216O mixture are much lower. The reduction of 18O incorporation in the film indicates direct involvement of N2O in the film-formation reactions via either direct oxidation of the precursor molecules by N216O or the intermediate role in the oxide formation as aforementioned for oxygen. In addition, we found that the film growth rate in 50%18O2-50%N216O is slightly lower than that in 18O2, indicating that the reaction rate using N2O is lower than that using O2. It is likely that N2O molecules are thermally decomposed in multiple steps to create atomic oxygen, which then reacts with the precursor molecules.
ReferencesGao, Y., Thin Solid Films, 346, 73 (1999). Gao, Y., T. Tran, and P. Alluri, "An Isotopic Study of Metalorganic Chemical Vapor Deposition of (Ba,Sr)TiO3 Thin Films," Appl. Phys. Lett. 75, 415 (1999). Gao, Y., C. L. Perkins, S. He, P. Alluri, T. Tran, S. Thevuthasan, and M. A. Henderson, "Mechanistic Study of Metalorganic Chemical Vapor Deposition of (Ba,Sr)TiO3 Thin Films," J. Appl. Phys., submitted. Henderson, M. A., Surf. Sci. 355, 151 (1996). Semiconductor Industry Association (SIA), The National Technology Roadmap for Semiconductors (1997).
William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147 |
