Silicide and Germanide Contacts to Silicon and Germanium Nanowires

Open Access
Author:
Dellas, Nicholas S
Graduate Program:
Materials Science and Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
March 18, 2011
Committee Members:
  • Suzanne E Mohney, Dissertation Advisor
  • Suzanne E Mohney, Committee Chair
  • Theresa Stellwag Mayer, Committee Member
  • Zi Kui Liu, Committee Member
  • Joan Marie Redwing, Committee Member
Keywords:
  • silicide
  • Schottky barrier
  • nanowire
  • kinetics
  • nickel
Abstract:
Semiconductor nanowires have shown promise and garnered considerable interest for use in field-effect transistors, thin film transistors, and chemical and biological sensors. Electrical contacts to nanowire based devices have been identified as a source of performance limitations of these devices. For the successful application of nanowires into high-performance devices, a fundamental understanding of contact formation and contact properties is necessary. Furthermore, the formation of metallic silicides in the case of Si nanowires (SiNWs) or germanides in the case of GeNWs have proven to be an effective way to reduce the access and/or series resistance of nanowire devices. Here, the formation of silicides (Ti, V, Pt, Pd, and Ni) and germanides (Ni) in contacts to nanowires have been studied. Additionally, a method for accurate extraction of the Schottky barrier height in axial contacts to nanowires is demonstrated for Ni silicide contacts to n-type SiNWs. We have found that as a basic criteria for forming axial metal silicide contacts to SiNWs from metal contact pads, the metal should be the dominant diffusing species (DDS) in the solid state reaction between the metal and Si for the phase formed. In the case of Ti and V, Si is actually the dominant diffusing species in the first phase formed, namely C49 or C54 TiSi2 and VSi2, and in this case the silicide is formed underneath the contact pad as opposed to along the SiNW itself. For Pt and Pd, the metal is the DDS in the first phase formed, Pt2Si or Pd2Si, and axial silicide segments are formed. However, in the case of Pt silicide formation, more severe annealing conditions cause a transformation to PtSi. In PtSi, Si is the DDS and Kirkendall voids form as a result of the unequal fluxes of Pt and Si causing a break to form at the interface between the PtSi and SiNW. In the case of Pd2Si formation, Pd2Si is the only phase that forms and the segment continues growth with increased annealing time and temperature; however, during silicidation of oxidized SiNWs Pd2Si reacts through the SiO2 shell surrounding the SiNW. This situation would be problematic for forming structures in which the silicide contact is underneath a gate that must provide electrical isolation. A short would form between the gate and silicide contact if a transistor were made. Fortunately, Ni silicides also form axial metal contacts and in the Ni-Si system for every Ni silicide phase that forms Ni is the DDS. We have also identified for the Ni-SiNW system that the orientation of the SiNW can determine the Ni silicide phase that forms. In the case of annealing Ni contact pads to [112] SiNWs, the high-temperature metastable theta-Ni2Si phase forms and is thermally stable until annealing conditions of 700°C and higher. At 700°C branches form as a compressive stress release mechanism and could result in electrical shorts to neighboring devices in high device density applications. When annealing Ni contact pads on SiNWs with [111] growth directions, NiSi2 is the first phase to form and remains stable until temperatures in excess of 600°C where a transformation to NiSi occurs. In addition to identifying differences in the Ni silicide phase formed for different growth direction SiNWs, we have also identified differences in the kinetics of Ni silicide formation for [112] and [111] SiNWs. For [112] SiNWs, the formation of theta-Ni2Si is diffusion-limited with an activation energy of 1.45 ± 0.07 eV/atom. This activation energy is considerably lower than literature values of bulk lattice diffusion through Ni silicide compounds, and thus the diffusion mechanism is attributed to Ni diffusion along the Si/SiO2 interface. For [111] SiNWs the formation of NiSi2 is interfacial reaction limited with an activation energy of 0.76 ± 0.10 eV/atom. Furthermore, for the formation of theta-Ni2Si axial contacts to oxidized SiNWs, it was found the Ni reacted with the core of the SiNW, leaving the SiO2 shell intact surrounding the theta-Ni2Si/SiNW interface. The ability of the theta-Ni2Si contact to react only with the Si core is a requirement for implementation into wrap-around gate Schottky diodes. After identification of a suitable candidate, namely -Ni2Si contacts to SiNWs, was identified these contacts were integrated into wrap-around gate Schottky diode structures. Simulations by Karthik Sarpatwari showed that gating of the metal/semiconductor interface in full wrap-around gate Schottky contacts was an effective approach for extracting the true Schottky barrier height at the metal/semiconductor interface. For our theta-Ni2Si contacts to n-type SiNWs we were able to identify the same linear relationship between the ideality factor (n) and effective Schottky barrier height (φBeff) measured under different gate bias conditions. By extrapolation of the linear φBeff-n plot to n=1 the actual barrier height at the theta-Ni2Si/n-SiNW interface is identified, in this case 0.57 eV. We measured Schottky barrier heights for SiNWs ranging in diameter from 60–100 nm and found no significant (± 0.02 eV) deviation or trend with SiNW diameter. A new fabrication procedure for producing smaller diameter (30 nm and less) is also discussed and issues with integration of smaller diameter SiNWs into these structures due to the reduction of SiO2 by the Al gate are mentioned. Lastly, the formation of axial Ni germanide contacts to GeNWs was examined. We found that axial Ni germanide segments begin forming after annealing at 300°C for 2 min and continue growing with increased time and temperature. The Ni germanide phase is identified by matching of electron diffraction patterns to the Ni2In prototype structure. A stoichiometry of Ni3Ge2 is assigned due to the lack of vacancy ordering observed in the electron diffraction patterns. All other Ni germanide phases with the Ni2In prototype structure (B8 region of the phase diagram) have been reported previously to have some vacancy ordering with the exception of Ni3Ge2. After annealing at temperatures in excess of 400°C, a break is formed in the Ni germanide segment near the Ni germanide/GeNW interface. Plausible reasons for the break formation are discussed. We find the break formation problem can be worked around and that longer Ni germanide segments with average lengths of 1.5 μm can be formed after annealing at 400°C for 5 min.