| AVS 57th International Symposium & Exhibition | |
| Electronic Materials and Processing | Tuesday Sessions |
| Session EM-TuP |
| Session: | Electronic Materials and Processing Poster Session |
| Presenter: | J. Shen, University of California at San Diego |
| Authors: | W. Melitz, University of California at San Diego J. Shen, University of California at San Diego J. Clemens, University of California at San Diego E. Chagarov, University of California at San Diego A.C. Kummel, University of California at San Diego |
| Correspondent: | Click to Email |
For highly scaled atomic layer deposition (ALD) of gate oxides (EOT < 1nm) on III-V semiconductors, the general requirements for an unpinned, high mobility oxide/III-V interface are as follows: (a) The metal precursor cannot disrupt the substrate during deposition; (b) The metal precursor must form a monolayer of nucleation sites in order for aggressive scaling of the equivalent oxide thickness (EOT); (c) The oxide has to be resistant to atom donation to/from the substrate; (d) The oxide needs to bond weakly to the interface or to form nearly covalent bonds to the interface to balance polarity. The nucleation/passivation layer has to enable all four requirements. Half cycle room temperature ALD of trimethylaluminum (TMA) and dimethylaluminium ethoxide (DEAE) have been performed on InAs and InGaAs surface to compare two precursors for the same oxide one of which is oxygen-free and one which contains oxygen. The electronic properties of the clean and deposited surfaces are probed via scanning tunneling spectroscopy and microscopy (STS and STM ), and Kelvin probe force microscopy (KPFM). Previous STS and KPFM studies for both clean InAs and InGaAs (4×2) surfaces determined the Fermi level is pinned 0.3eV above the valance band; DFT studies show that the surface are pinned by homodimers in the trough. STM and STS show that TMA forms an ordered monolayer of absorbates which unpin the Fermi level suggesting that an ordered monolayer layer might be a requirement for unpinning. However, STM of DEAE reacted InGaAs(001)-(4x2) shows a nearly amorphous monolayer layer while STS shows even this amorphous layer unpins the interface. The influence of larger ligands on the DEAE might account for more degeneracy in bonding configurations making order structures less probable. The results are consistent with a multitude of bonding configuration being able to unpin the Fermi level as long as the pining sites are removed and the presence of oxygen in the precursor not being an impediment to passivation as long as there is still attractive interactions between the absorbates which promote monolayer formation.