(204e) Surface Reaction Mechanism during the Atomic Layer Deposition of Titanium Dioxide from Titanium Tetraisopropoxide and Ozone | AIChE

(204e) Surface Reaction Mechanism during the Atomic Layer Deposition of Titanium Dioxide from Titanium Tetraisopropoxide and Ozone


Agarwal, S. - Presenter, Colorado School of Mines
Rai, V. R. - Presenter, Colorado School of Mines

In this presentation, the authors will discuss the surface reaction mechanism during the atomic layer deposition (ALD) of TiO2 using titanium tetraisopropoxide (TTIP) as the metal precursor, and O3 and H2O as the oxidizers. The film growth mechanism was investigated using a combination of in situ, real-time attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy, quartz crystal microbalance (QCM), and quadrupole mass spectrometry (QMS). The ATR-FTIR spectroscopy setup is ideal for studying ligand-exchange surface reactions during ALD due to its sub-monolayer sensitivity for surface adsorbates. In our experiments, deposition was done on ZnSe internal reflection crystals. The refractive indices of ZnSe and amorphous TiO2 are closely matched (~2.2) and, therefore, internal reflection on the top surface occurs at the vacuum-film interface. In our geometry we have 25 reflections on each surface and, hence, the IR beam passes through the film 50 times, greatly enhancing the signal-to-noise ratio. Figure 1 shows IR spectra recorded during exposure of the surface to TTIP, using an O3-exposed surface as the reference. Sub-monolayer surface absorbates were easily detected in real time and were identified in these spectra. To quantify the surface coverage, and to determine the growth per cycle (GPC), the mass uptake during each half-cycle was determined using the QCM and the surface reaction products were detected with a differentially pumped QMS. Finally, the total thickness and the refractive index of the films were characterized using ex situ spectroscopic ellipsometry. Using O3 as the oxidizer, relatively contaminant-free TiO2 films were deposited at 150 ºC: this is nearly a 100 ºC lower than the widely-accepted minimum temperature for the TTIP-H2O ALD window. Our data shows that the reactive sites for the adsorption of TTIP after the O3 cycle are surface carbonate species, which have symmetric and anti-symmetric absorptions in the 1400 ? 1700 cm-1 region, shown in Fig. 1. These surface carbonates can be present as mono-, bi-, or poly-dentates. Since all these forms of surface carbonates are known to be stable on a TiO2 surface at the growth temperature, and their absorption bands overlap, it was not possible to determine their relative surface concentration. Presence of surface carbonates was further confirmed by detecting CO2 as a reaction product during both the TTIP and O3 cycles. During O3 exposure, CO2 was formed due to the combustion of isopropoxy ligands. The surface carbonates were then formed due to the back reaction of some of this CO2 with a hydroxyl-terminated TiO2 surface. When the carbonate terminated surface is exposed to TTIP, CO2 is eliminated from the surface. This is in contrast to the commonly-studied metal alkoxide-H2O ALD, where the alkoxides react directly with surface hydroxyl groups that are formed after H2O exposure. During our experiments, we also observe non-ideal ALD behavior. First, we observe that the surface coverage, measured through the integrated absorbance change due to the CHx stretching modes in the ~3000 cm-1 region, is greater for a higher TTIP flux for an otherwise identical precursor dose. Second, the surface carbonates that are formed during the following O3 cycle have different absorption bands, and a lower integrated absorbance for a lower flux. We explain this behavior using a simple surface site balance, and show that each TTIP molecule is able to exchange more ligands with the surface under low flux conditions, resulting in a low apparent coverage in the IR. Finally, under high flux conditions the GPC was measured over the temperature range of 150-275 °C using a QCM and spectroscopic ellipsometry. The GPC was almost constant over this temperature range at an average value of 0.52 Å/cycle, in good agreement with the literature.