berryh@udel.edu (John Berryhill) (01/12/90)
I'm having some trouble with contacts on n-AlGaAs (x=.1 or so). I'm depositing about 5000 Angstroms of Au-Ge alloy (12% Ge) and following that with 1000A of nickel. Some of the time, it works just fine, but too often the contacts tend to ball up when I'm alloying them to the AlGaAs layer. As I understand it, the purpose of the Ni layer is to keep that from happening. Is there something else I should be doing in addition to the Ni to prevent it from balling up? -- John Berryhill 143 King William, Newark DE 19711
brooks@sierra.Stanford.EDU (Michael B. Brooks) (01/12/90)
I never expected to find this on the net John! I`m happy to reply, since this is my thesis research project here at SU. Try the Ni first, in this recipe, as a starter. If possible, give Rapid Thermal Annealing a whirl: E beam evaporate: 50-100A Ni ~750A Au/Ge Eutectic 250A Ni 1000A Au (for reduced resistivity, the overlayer of Au is essential) RTA times 30-45 sec or so, 450C or less furnace anneal, 450C, under 5 min., forming gas Note that the amount of Au/Ge is variable, and can be evaporated as separate layers of the same overall composition. One must have a thin layer of Ni first to prevent the balling effects that you noticed (it`s more complicated than this, a black art really) See Gallium Arsenide, by MJ Howes and Dv.Morgan (John Wiley & Sons, 1985), or Gallium Arsenide Processing Techniques by RE Williams, (Artech House, 1984) for starters. Mike Brooks/Stanford Electronics Labs (solid state)/SU
berryh@udel.edu (John Berryhill) (01/12/90)
In article <454@sierra.stanford.edu> brooks@sierra.UUCP (Michael B. Brooks) writes: >E beam evaporate: >50-100A Ni >~750A Au/Ge Eutectic >250A Ni >1000A Au (for reduced resistivity, the overlayer of Au is essential) Do you have a reference for this by any chance? Also, have you tried the recipe by Shih and Blum, "Contact Resistances of Au-Ge-Ni, Au-Zn and Al to II-V Compounds" from Solid State Electronics, V15,n11-A, pp. 1177-1180? Apparently they thermally evaporated an alloy of the Au-Ge eutectic plus 5%wt. Ni in one shot. Our E beam machine is a little flakey and I'd like to stick with thermal evaporation if I could. Of course, one of the problems with thermal evaporation of alloys is that the species will often come out of the boat separately. -- John Berryhill 143 King William, Newark DE 19711
brooks@sierra.Stanford.EDU (Michael B. Brooks) (01/14/90)
In reply to the reference request about the ohmic contacts recipe (n-type GaAs) there is a body of literature based on M. Murakami`s group at IBM (Yorktown Heigths). The recipe is based on J.Vac.Sci.Technol. B 4(4), Jul/Aug 1986 pp.903-911, "Microstructure studies of AuNiGe Ohmic contacts to n-type GaAs", by Murakami, Childs, Baker & Callegari. Thickness of the Au-Ge layer varies somewhat (our work). This is not specifically for contacts to AlGaAs (it`s for GaAs) but it will probably work since other workers use such recipes. A similar recipe is from Rai & Ezis et.al., J.Appl.Phys. 63 (9), 1 May 1988, p.4723, and is for GaAs-AlGaAs MODFETs: This is 50A Ni, 170A Ge, 330A Au, 150A Ni, and 2000A Au, which was put down via e-beam , followed by "transient annealing" at 525C for a short time. What I have cited is hardly representative of multitudes of contact schemes available. These are similar to what I use, and would be considered "traditional" in some sense, generally known to work. They all require horrendous amounts of optimization based on your particular set of equipment. If I was using a thermal system for metal evap I would be extremely careful to limit the heat transported to GaAs. MEtallization reactions happen at low temps (<250C) in short times, and can mess things up if not carefully controlled. Beyond this thermal evap ought to be OK. I haven`t tried the contact scheme you have mentioned John, though I`ll try to check out the reference. Mike Brooks/Stanford Electronics Labs (solid state)/SU