Process optimization and characterization of Ni-Ge-Au ohmic contacts to n+ GaAs heterostructures

Abstract

A process for the formation of low resistance Ni-Ge-Au ohmic contacts to n+ GaAs heterostructures was optimized using multivariable screening and response surface experiments [1,2]. Of seven variables screened by a fractional factorial experiment, the strongest effects were: total Ge+Au evaporative thickness, Ge:Au ratio and post-alloy cooling time. A response surface experiment run on these three variables enabled the development of an empirical model of ohmic contact resistance (Rc), with a predicted optimum value of 0.07 ± 0.03 Ωmm. Twenty confirmation runs yielded an average Rc of 0.07± 0.04 mΩm, a reduction of 50% on the standard average process value of 0.14 ±0.1Ω mm. A non-optimized (Rc = 0.17 Ωmm) and an optimized (Rc = 0.04 Ωmm) sample were characterized using SEM-EDS, cross-sectional AEM, and XPS techniques. The non-optimized sample has prominent surface dendrites and a heterogeneous microstructure consisting of blocks of Ni0.5Ge0.4As0.1 in a matrix of Au0.7Ga0.2As0.1 and Ge-rich phases. Some of the Ge-rich crystals extend downward 50 nm into the substrate and are epitaxial to the n+ GaAs. The optimized sample has faint surface dendrites and a homogeneous fine-grained microstructure consisting of islands of Ni0.5Ge0.25As0.25 in a matrix of Au0.8Ga0.15As0.05. XPS depth profiles are in qualitative agreement with these results - the non-optimized sample shows a higher Ge content in the metal layer and less Ge in the GaAs substrate relative to the optimized sample. This work shows that 1) multivariable experimental methods can be used to optimize Ni- Ge-Au ohmic contacts to n+ GaAs, 2) the critical parameters for this process are Ge+Au layer thickness. Ge:Au ratio, and post alloy cooling time, and 3) lower Rc values are realized by effective diffusion of Ge from the metal layer into the GaAs substrate [3].