The conformal growth of Pb-based ferroelectric films on oxide electrodes at low temperatres by direct liquid injection metalorganic chemical vapor deposition (DLI-MOCVD) is significant to realize the three-dimensional (3D) capacitor structures for high-density(≥128Mb) ferroelectric random access memories (FRAMs). This dissertation has focused on the low temperature DLI-MOCVD process of Pb-based ferroelectric films such as $PbTiO_3$ and $Pb(Zr,Ti)O_3$ [PZT].
Firstly, I have optimized the metalorganic source delivery which is the first step toward DLI-MOCVD process. Among various solid metalorganic precursors, the three precursors of $Pb(TMHD)_2$, $Ti(TMHD)_2(O-i-Pr)_2$, and $Zr(TMHD)(O-i-Pr)_3$ were selected because of their similar vaporization temperatures. It was found that the selection of suitable solvent was also significant in the DLI system. In terms of selective vaporization and sufficient solubility (≥0.1M), the solvent of n-butylacetate was chosen. But the Pb-solution exhibited the lack of long-term stability. To rule out this aging problem, the Pb-solutions were used within 24h after synthesis. The vaporization temperature was determined at 230℃ at which the amount of non-volatile residues were minimal.
I have addressed a new issue on the thermochemical stability of $IrO_2$ bottom electrodes during the growth of PZT films on them using DLI-MOCVD. This issue has not been addressed so far since the use of $IrO_2$ as well as DLI-MOCVD method is not prevalent. The $IrO_2$ films have been regarded as a leading candidate for the bottom electrode of ferroelectric capacitors. This study, however, revealed that $IrO_2$ was not compatible with the DLI-MOCVD process of PZT films. The $IrO_2$ electrode was thermally dissociated when annealed in vacuum ambient with a low oxygen pressure. It was also reduced by carbon and hydrogen decomposed from the solvent molecules. The reduction of $IrO_2$ by solvent was more pronounced with decreasing temperature, which is attributed to the longer residence time of the solvent molecules on the $IrO_2$ surface at lower temperatures. The reduction of $IrO_2$ also occurred by the metal constituents of Zr and Ti in PZT films because they have higher chemical affinities with $O_2$ than Ir. The use of 20-nm-thick Pt film as an interlayer assured that the $IrO_2$ electrode maintains its stability during the growth of PZT at 420℃.
The growth behavior of Pb-based ferroelectric films at low temperatures around 400℃ was investigated. The impurity level of carbon was below Auger electron spectroscopy (AES) detection limit even though $PbTiO_3$ was grown at the temperature as low as 360℃. The film growth was strongly dependent on the substrate temperature. It was also highly susceptible to the source input ratio of [Pb]/[Ti] at the fixed temperature. That is, the even smaller variations of substrate temperature and [Pb]/[Ti] brought about the greater variations of film composition. It is difficult to obtain the stoichiometric films having pure perovskite phases at low temperatures.
The effects of the pulsed plasma on the low temperature (380℃) growth of Pb-based ferroelectric films have been systematically investigated. With the appropriately tuned pulsed plasma, the stoichiometric $PbTiO_3$ films having pure perovskite phases could be obtained over wider range of precursor supply ratio [Pb]/[Ti]. In the growth of PZT films, the processing window of [Pb]/([Zr]+[Ti]) for obtaining stoichiometric films was also somewhat widened by the pulsed plasma. The pulsed plasma has facilitated the control of Pb-based ferroelectric film composition without degradation of step coverage characteristic. The adoption of pulsed plasma is expected to improve the reproducibility and stability of a mass-production compatible DLI-MOCVD process.
