Back to Table of Contents

44228-G5
Studies of Interface Structure and Properties of Thin Films

Theodosia Gougousi, University of Maryland (Baltimore County)

The atomic layer deposition of HfO₂ thin films is studied on GaAs(100) surfaces. GaAs substrates exhibit higher carrier mobility than Si substrates and as a result are considered as possible materials for advanced electronic devices. Two different chemistries that utilize Hf-amide precursors were investigated: i) tetrakis(dimethyl)amino hafnium (TDMAHf) and H₂O at a deposition temperature of 275°C and ii) tetrakis(ethylmethyl)amino hafnium (TEMAHf) and H₂O at a deposition temperature of 250°C.  Several GaAs starting surfaces were investigated, including native oxide and both HF and NH₄OH-treated substrates.  Wet chemical etching in either HF or NH₄OH solutions are shown to remove most of the Ga and As native oxides (Figure 1).  

i)                    TDMAHf + H₂O: For the TDMAHf process spectroscopic ellipsometry (SE) confirms linear growth rates of ~1.1 Å/cycle for all surfaces.  Rutherford backscattering spectrometry (RBS) shows that deposition on the surface native oxide results in very smooth growth of ~ 2.9×10¹⁴ Hf/cm²/cycle. For the HF and NH₄OH-etched GaAs surfaces steady state is reached after 10 ALD cycles with comparable growth rates (Figure 2). The interface of HfO₂ films deposited on GaAs surfaces is probed by X-ray photoelectron spectroscopy.  Both the HF and NH₄OH treatments passivate the surface and prevent the oxidation of the interface during the deposition of coalesced HfO₂ films (> 15 ALD cycles) (Figure 3).  Deposition of HfO₂ films on the native oxide GaAs surfaces show gradual consumption of the native oxides during the process, indicating the presence of an “interfacial cleaning” mechanism comparable to that observed for other ALD processes.[1]^,[2],[3],[4]  The As-oxide and most of the Ga-oxide is completely removed after 20 ALD cycles.  The presence of As oxides is not detected for films as thick as ~100Å (100 cycles) deposited on native oxide substrates (Figure 4).

ii)                   TEMAHf + H₂O:  A similar to the previously observed interface cleaning reaction is observed for deposition on GaAs native oxides. The oxide consumption reaction appears to be proceeding somewhat slower as it requires ~25 cycles for complete removal of the interface oxides compared to ~20 cycles for the TDMAHf process (Figure 5). A possible explanation if this slower reaction is confirmed is through a difference in the ALD mechanism for the two precursors. In the case of the TDMAHf precursor the ALD reaction can only proceed through reduction of the ligands but for TEMAHf an additional pathway may be possible through beta hydride elimination.[5] If this secondary path is accessible it is almost certain it will produce different products than the reductive path. These "different" products may not react as efficiently with the Ga and As native oxides. We are currently investigating this effect in more detail including RBS surface coverage measurements.

Figure 1: XPS scan of the As 2p, As 3d and Ga 2p regions for different GaAs surfaces: (a) native oxide, (b) cleaned in JTB-100, (c) etched in 30% NH₄OH aqueous solution and (d) etched in HF solution.

Figure 2: Hf atom surface coverage per ALD cycle as measured by RBS for the TDMAHf + H₂O chemistry. The three starting surfaces are surfaces (a), (c) and (d) from Figure 1.

Figure 3. Deposition of ~15 Å of HfO₂ using the TDMAHf + H₂O chemistry on HF and NH₄OH etched GaAs surfaces does not cause regrowth of the native oxides indicating surface passivation.

Figure 4. Gradual consumption of the surface native oxides is observed during the deposition of HfO₂ films from TDMAHf and H₂O on cleaned GaAs native oxide surfaces.

Figure 5. Gradual consumption of the surface native oxides is observed during the deposition of HfO₂ films from TEMAHf and H₂O on cleaned GaAs native oxide surfaces. The film thickness is based on growth rates measured on native oxide Si surfaces. RBS surface coverage measurements will be performed to allow a direct comparison between the data on Figures 4 and 5.

Terms of Use Security Privacy Contact Copyright © 2009 American Chemical Society