Fast alloying technique for improved ohmic contacts to n-GaAs

Fast alloying technique for improved ohmic contacts to n-GaAs

Solrd-Sture Elecrronrc.~ Vbl. 27. No. 10, pp. 925-926, Printed in the U.S.A 1984 003R-1101/84 $3 00 + .Ou Pergamon Press Ltd. LETTER TO THE EDITOR ...

175KB Sizes 2 Downloads 34 Views

Solrd-Sture Elecrronrc.~ Vbl. 27. No. 10, pp. 925-926, Printed in the U.S.A

1984

003R-1101/84 $3 00 + .Ou Pergamon Press Ltd.

LETTER TO THE EDITOR Fast alloying technique for improved ohmic contacts to n-GaAs

(Received 30 May 1983; in revised form 7 November 1983)

vestigated. A difference of IO-20°C between the optimum Alloying and sintering techniques, which are the basic alloying temperature and the temperature optimum from methods of contact fabrication in device technology, are the -aspect of reliability was found[6], in good agreement admittedly still a technical art. with results described in Ref.171. This last value was found Recently a new “face down” alloying method was proposed[l]. According to this method when cooling is about 510°C. The high reliadihty of the contact obtained, when annealed at 51O”C, is explained by the metallurgical forced by a copper block placed on top of a wafer the heater stability of the grain structure[7J Different types of the current is turned off. By application of the copper block a AuGe, AuGeNi metalhzation system, such as higher cooling rate is achieved while the substrate side is at AuGeNi + Au and AuGeNi + Ga + Au were investigated lower temperature. An analysis of the possible explanations in our laboratory, and various surface morphologies were for the lowering of the specific contact resistivity was carried observed by SEM. In good agreement with Miller’s out by the authors of Ref. [l]. According to this analysis the obscrvation[8] the electrical characteristics of AuGe conhypothesis of the formation of a thin n+ contact layer and the mechanism of conduction suggested by Popovic[Z] seems tacts was remarkably insensitive to the morphology of the metalhzation. It should be noted that from this point of view to be the best possible mechanism for the improved contact only relatively thick (about SOOnm) contacts were inproperties. However, it is not clear whether the higher vestigated. Rectangular regions indicating epitaxial growth cooling rate or the “face down” technique are the main during cooling[9] were also observed using a galliumreason of this improvement. containing contact system[lO]. Our experimental results have shown that the fast alloying A possible explanation for the contact formation at high process is the most effective method to obtain reliable ohmic temperatures with rapid heating and cooling rates is the contacts of low specific resistivity. For our experiments following. Three phenomenological processes determine the n-n+ epitaxial structures were used. The polished wafers of contact formation in a given structure. These are: the known thickness and with concentrations between 1.5 and 2.2 x 10’Scn-3 were degreased and etched. The contact solution, characterized by the solubility at a given temperature, the thermal decomposition of the compound semimaterial was an eutectic alloy of AuGe with about 5% Ni conductor material in the presence of the contact added. The thickness of the eutectic film was 400 nm, and material[lO] and the regrowth of the compound semiconit was deposited by vacuum evaporation of the eutectic ductor from a solid[9], from a hquid[l l] or from thin[l2] alloy. Using our technique GaAs samples after metallization phase during the heating and mainly during the cooling. are placed on a gold plated copper stud surrounded by a During the heat cycle compound formation and interheater of low thermal inertia. A short heat pulse or a metallic formation, volatile component loss and diffusion sequence of heat pulses are used for the heat treatment. A relatively high flowing rate (300 m&c) hydrogen stream is take place. As it was shown in Ref. [13] the main difference between poor and good ohmic contacts is the presence or directed upon the heater and the stud. The formation of absence of gold and germahium in solid solution at the ohmic contacts is controlled by checking the resistance of interface. Contacts heated at lower temperature give a poor the sample during the alloying process[3,4]. After the ohmic character, showing similar intermetallic precipitates formation of the specific contact resistance is measured[5]. at the interface, but no significant Au or Ge concentration Various technological parameters such as the value of the at the GaAs surface[l3]. This fact and the structural insenpeak temperature of the heat treatment, the heating and sitivity seem to indicate that the critical factor influencing cooling rates, the atmosphere during the heat cycle and the surface preparation before the metallization were inthe quality of the contacts is the dopant distribution in GaAs[l3] and not the chemical processes and structures vestigated in the course of our work. Summarizing our results it can be stated, that an optimal temperature of about formed during the heat cycle. This dopant distribution is strongly influenced by the temperature and by the cycle 5OO“Cwas found for the alloying of contacts. Alloying at different heating rates in the range of 150. .24O”C/sec was time. These two parameters are the most important ones influencing the dopant distribution and incidentally the also investigated. It was found that the specific contact resistance has an optimum at a heating rate of 2OO”C/sec value of the specific contact resistance and the character of the contact. Other factors such as the presence of an element and has a value of only (0.9. . .2) x IO- 5 Q/cm2 for a typical X-band Gunn structure, without an n+ contact coexisting in the contact system can also drastically change the diffusion of an individual component[l4]. The solubility layer[6]. of GaAs in AuGe alloys is relatively high at 5OOT, which Forming gas and nitrogen are the most commonly used is the beginning of the saturation range. Volatile component gases during heat treatment. We investigated various gases loss becomes higher at about 5OOC[lO]. In this temperature and hydrogen seems to be the optimal one. We can speculate range the solubility is sufficiently high and the volatile on at least two possible mechanisms. One is a chemical component loss, which strongly influences the specific conreaction during the heat pulses, the other one is the faster tact resistivity [lo], is not so high. Using a short heat cycle a cooling caused by the high specific heat of hydrogen. In the course of our study NH,OH and H,SO, based thin n+ layer forms and a conduction mechanism proposed solvents were used to investigate the effect of the surface by Popovic [2] takes place leading to a lower specific preparation[6]. The best results were obtained using an contact resistivity. NH,OH-based etchant, a H,SO,-based method resulted in poor contacts. It should be notedthat the difference between the lowest and highest values of the specific contact re- Research Institute for Technical Physics I. Morz~s sistance is only one order of magnitude. of the Htmgarian Academy of Sciences, Reliability aspects of the technology were also in- Budapest, Ujpest 1. P.O.B. 76, Hungary. 925

926

Letter REFERENCES

1. M. I. Nathan and M. Heiblum, Solid-St. Elecrron. 25, 1063 (1982). 2. R. 8. Popovic, Solid-St. Electron. 21, 1133 (1978). 3. I. Mojzes, Phys. Stat. Sol. (a) 47, K183 (1978). 4. 1. Mojzes, Acta Phys. Acad. Scient. Hung. 48, 131 (1980).. 5. L. Gutai and I. Mojzes, Appl. Phys. Lett. 26,325 (1975). 6. I. Mojzes, A. Bama and B. P. Bama, Proc. of MICROCOLL’78 Budapest, Aug. 1978. Vol. II. V-2/10.1 (1978). I. M. Ogawa, J. Appl. Phys. 51, 406 (1980).

to the Editor 8. D. C. Miller, J. Electrochem. Sot. 127, 467 (1980). 9. A. Christou, Solid-St. Electron. 22, 141 (1979). 10. I. Moizes. T. Sebestven and D. Szieethv. _ ,, Solid-St. Elect&. 25, 449 (198i). 11. M. Otsubo, H. Kumabe and H. Miki, Solid-Sr. Electron. 20, 141 (1977). 12. T. Sebestyen, H. L. Hartnagel and L. H. Herron, IEEE Trans. on Electron Deu. ED-22, 1073 (1975). J3. C. R. M. Grovenor, Solid-St. Electron. 24, 792 (1981). 14. M. Heiblum, M. I. Nathan and C. A. Chang, Solid-S/. Electron. 25, 185 (1982).