Formation of heterovalent interfaces: A combined photoemission and ab initio DFT study of GaP/Si heterostructures

• Applicant: Professor Dr. Thomas Hannappel
• Subject Area: Physical Chemistry of Solids and Surfaces, Material Characterisation ; Experimental Condensed Matter Physics ; Theoretical Condensed Matter Physics
• Term: from 2018 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 391502515

• Final Report Year: 2022

Final Report Abstract

The double layer stepped Si(100) surface preparation in the arsenic ambient and the investigation of these surfaces in situ by RAS, XPS, LEED and STM have been carried out at TU Ilmenau. In dependence on the type of As source either TBAs or background As4 and on process route, we are able to control the dimer orientation of either majority A-type or B-type domain on the Si surfaces. We have seen that for low offcut Si(100) surface (~0.1°), the surface prepared in background As4 and cooling in low H2 pressure (50 mbar) leads to an energetically driven, majority of A-type, (1×2) reconstructed surface. In contrast, for the surface with high offcut (~4°), the well-ordered, kinetically driven, A-type, (1×2) reconstructed surfaces can be prepared in a controlled manner at around 830°C while TBAs is applied directly from the precursor. The analysis by XPS and the quantified component ratio between the bulk and the close to the surface region confirm that the high amount of As is present on the Si(100) 0.1° surfaces. From the STM scans, we confirm that the As-modified Si(100) surfaces do not entirely consist of symmetric dimers but also exhibit asymmetric dimers as zig-zag chains. We suggest that the symmetric dimers could be Si-Si or As-As dimers and the asymmetric dimers could be As-Si dimers or As-As dimers with an asymmetric distribution of bonded hydrogen atoms. After the Si(100) surface preparation, GaP(100) buffer layers of different thicknesses were grown. Subsequently, depth profiling by SIMS, GCIB-XPS, AR-XPS, photoemission data analysis, and DFT simulations have been carried out at IoP in Prague (Czech Republic). HAXPES measurements were carried out at SPring-8 (4 times), DESY (2 times), and partially at BESSY (1 time). From the HAXPES measurement, the core level peak positions, deconvolution of peak line shape, interface model and band alignment were derived. It has been shown that the interface model of the GaP/Si(100):H sample has a good agreement with the SIMS measurement performed on the GaP samples which shows P dopant concentration is high in the Si interfacial layer. SIMS measurements were ordered at a Berlin company. Moreover, we observe almost no fluctuations of atomic concentration at the interface however small change in atomic compositions was observed in the GaP/Si(100):H sample. In the GaP(As)/Si(100) sample, arsenic core levels chemical shift, valance band offset (VBO) and atomic concentration of As were determined. VBOs between 0.70 - 0.75 eV was obtained for Asrich interfaces independently on Si(100) surfaces. Moreover, the As atomic concentration between 0.5 - 1 at.% was measured for the samples GaP/Si(100):As with A-type and B-type (6 nm thick overlayer, p-Si, 4° miscut) and between n-doped/p-doped Si substrates (8 nm, A-type, 0.1° miscut). The chemical shift of 0.6±0.1 eV in P 2p and of 0.3±0.1 eV in Si 2p peaks were further measured by HAXPES. A ratio of interface/bulk components has changed with photoelectron emission angle: localization of P-Si bonds at interface was confirmed for P-rich interfaces of GaP/Si grown on single-domain Si substrates. For P-rich interfaces, we found dependence of VBO on GaP overlayer thickness. In addition, strong band bending was observed in 8 nm thick GaP/Si(100) single domain (P-rich) heterostructure whereas band bending is significantly less pronounced in GaP/Si(100) surface two domain (Ga-rich) heterostructures. In addition to these measurements, ARPES measurement took place at PSI synchrotron (ADRESS beamline, Switzerland) and Neutron reflectivity (NR) measurements and simulations were at Diamond Light Source (UK). Simulation of NR intensity shifts revealed an interlayer with a lower scattering length at the GaP/Si(100):H interface.

Publications

• “Metalorganic vapor phase epitaxy of III–V-on-silicon: Experiment and theory” Prog. Cryst. Growth Charact. Mater. 64 (2018) 103
  O. Supplie, et.al.
  
• Atomic surface structure of MOVPE-prepared GaP(111)B”, Appl. Surf. Sci. 534 (2020) 147346
  P. Kleinschmidt, et al.
  
• “GaP/Si(0 0 1) interface study by XPS in combination with Ar gas cluster ion beam sputtering”, Appl. Surf. Sci. 514 (2020) 145903
  O. Romanyuk, et al.
  
• “Hard X-ray photoelectron spectroscopy study of core level shifts at buried GaP/Si(001) interfaces”, Surface and Interface Analysis 52 (2020) 933
  O. Romanyuk, et.al.
  
• “A route to obtaining low-defect III-V epilayers on Si(100) utilizing MOCVD”, Crystal Growth & Design 21 10 (2021) 5603
  M. Nandy, et al.
  
• “Band bending at heterovalent interfaces: Hard X-ray photoelectron spectroscopy of GaP/Si(0 0 1) heterostructures”, Appl. Surf. Sci. 565 (2021) 150514
  O. Romanyuk, et al.