Metal penetration and dopant redistribution beneath alloyed Ohmic contacts to n-GaAs

Abstract

In order to estimate the depth of penetration of Ohmic metals into n-GaAs during the alloying process, marker layer samples fabricated by molecular-beam epitaxy were utilized. The samples consisted of n-GaAs substrates onto which were grown a 500-nm undoped GaAs buffer layer, followed by 20 layers, each 50 nm thick, alternately doped with Si at levels of 1×1016 and 1×1018 atoms/cm3. Ohmic metals consisting of Ni/Ge/Au/Ni/Au or Ni/Ge/Au/TiB2/Au were evaporated onto the marker layer samples and alloyed on a strip heater. Secondary ion mass spectroscopy (SIMS) depth profiling showed that metal penetration extended as much as 0.5 μm into the GaAs. In addition, significant redistribution of the marker layer Si was observed, and is assumed to result from a vacancy enhanced diffusion process. In a SIMS depth profile performed from the back side of the wafer no dispersion of the metal depth penetration was observed, suggesting that differences in penetration depth of the metals observed in front side profiles are the result of ion beam mixing, surface roughness, and/or inhomogeneities within the alloyed region.

On the annealing mechanism of AuGe/Ni/Au ohmic contacts to a two-dimensional electron gas in GaAs/AlxGa1−xAs heterostructures

“…Our results suggests that for solving this problem the focus should be at maintaining sufficient contact area between Ge-rich Ni grains and the Ge-doped Al x Ga 1−x As layer at long annealing times. This can possibly be engineered by changing the layer thickness, order and composition of the initial AuGe/Ni/Au metallization 12,14,15,16,19,21,26 , or by including a Pt, Nb or Ag layer below the top Au layer that suppresses the intermixing of this Au with layers at the wafer surface 7,15,19,22,24 (uniform Ni/Ge/As layers been reported 21 ). Alternatively, one can reduce the depth of the 2DEG by etching before deposition of AuGe/Ni/Au (up to the point where this results in depletion of the 2DEG).…”

“…This probably enhances the in-diffusion of Ge. It was for example also observed that the creation of such vacancies near the surface, enhances the diffusion of Si dopants from the doping layer (much deeper into the material) into neighboring layers [15].…”

“…At the same time, there is diffusion of atomic Ge, Ni and Au (at similar concentrations) into the heterostructure, which penetrates deeper [8,19,20,15,27]. In particular, Ge diffuses out of the Ni-rich grains into the Al x Ga 1−x As layers, and a good ohmic contact is formed when the Al x Ga 1−x As layers are sufficiently doped with Ge all the way up to the 2DEG.…”

“…We can only explain this result if we assume that the 2DEG square resistance underneath the contact is significantly increased (to values comparable to the total observed contact resistance) for under annealed contacts. This probably results from the in-diffusing Ge (and atomic Au and Ni 15,27 ), which already introduces strain and scatter centers in the 2DEG layer before optimal contact conditions are reached. For optimal annealed contacts (here with total resistance of typically 7 Ω, independent of circumference), the square resistance underneath a contact must have returned to a low value of order 10 Ω.…”

Characterization of low-resistance ohmic contacts to a two-dimensional electron gas in a GaAs/AlGaAs heterostructure

“…The AuGe/Ni/Au material was first used by [9] to make an ohmic contact to n-GaAs in 1967. Subsequent studies aimed at improving such contacts and understanding the annealing mechanisms [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]. Later on, with the increasing importance of the 2DEG in a GaAs/Al x Ga 1−x As heterostructure, research focussed on making ohmic contacts to the buried 2DEG [25][26][27][28][29][30][31][32][33].…”