Solid State Communications, Vol. 49, No. 1, pp. 99-101, 1984. Printed in Great Britain.
0038-1098/84 $3.00 + .00 Pergamon Press Ltd.
METALLURGICAL BEHAVIOUR OF Ni/Au-Ge OHMIC CONTACTS TO GaAs A. Iliadis and K.E. Singer The University of Manchester, Institute of Science & Technology, P.O. Box 88, Manchester M60 1QD, England
(Received 15 September 1983 by S.A. Amelinckx) Auger depth proffding and contact resistance measurements have been used to study sintered Ni/Au/Au-Ge and Ni/Au-Ge contacts to GaAs. The results indicate that Ge incorporation is by a process of solid state diffusion rather than by the eutectic melting-recrystallization mechanism found in Au-Ge contacts. In spite of the improvement in the homogeneity of the contact brought about by using Ni, the thermal stability of these contacts is found to be worse than the simple Au-Ge ones. Auger depth profiling and contact resistance measurements.
1. INTRODUCTION Au-Ge AND Ni/Au-Ge ARE THE MOST extensively used ohmic contacts to n-GaAs and have been the subject of many studies [ 1-7] and a recent review . Low resistance ohmic behaviour follows from the incorporation of Ge onto Ga sites giving a heavily doped surface region and field emission through the resulting thin barrier. In Au-Ge contacts, Ge doping occurs through a eutectic melting-recrystallization mechanism with Ga being preferentially gettered by the overlying metal film giving rise to the necessary group III vacancies. It is well established that the resultant contact is metallurgically inhomogeneous with current confined to recrystallized areas. In a recent paper  we have presented evidence to show that below the sintering temperature, Ge is unevenly distributed at the interface, and that for normal sintering conditions (450°C for 2 - 3 min) the reaction is initiated via the Au-Ge eutectic system at sites where the Ge content is at the eutectic level. The inhomogeneity in the final contact results from the initial uneven distribution of Ge. The use of an overlying Ni film was first suggested by Braslau et al.  who showed that it much reduced the inhomogeneity of the Au-Ge contact. Braslau suggested that the Ni did not melt during sintering and thus acted like a uniform blanket to hold the liquid A u - G e in contact with GaAs. However, in a later study Robinson  used Auger depth prof'ding to show that below the Au-Ge eutectic temperature (360°C) the overlying Ni moved through the Au-Ge Fdm to accumulate at the GaAs interface. This led to the view of the Ni as being a "wetting agent'. Yoder  pointed out that Ni is a fast diffuser in GaAs and its presence greatly enhances the diffusivity of Ge. This work describes the results of further measurements on Ni containing Au-Ge contacts using
2. EXPERIMENTAL (1 0 0) n-type epitaxial GaAs on semi-insulating substrate was used, with doping ranging from 9 x 10 is cm -3 to 2 X 1017 cm -a. H2804: H202:H20 (10:1 : 1) and HC1 were used for etching and the films were deposited in a conventional evaporation system as described previously . Three sets of films were deposited simultaneously on each wafer by consecutive resistive evaporations. The contacts were Ni/Au/Au-Ge, Ni/Au-Ge and Au/Au-Ge and consisted of 350 h of Au-Ge from an evaporant charge of eutectic composition (12 wt.% Ge), followed by 1000 A of Au and/or by 200 A of Ni. Sintering was usually at 450°C for 2.5 min. The Auger depth profdes from as-evaporated and sintered Ni/Au/Au-Ge and Ni/Au-Ge contacts are shown in Figs. 1 and 2 respectively. Unlike the data from Robinson  there is no significant separation of Au and Ge in the as-deposited films, but the profiles from the sintered contacts exhibit a similar movement of Ni to the GaAs interface as he observed. The sintered contacts also show clearly developed plateaux in the Ga and As lines at the interface (corresponding to the peak in the Ni and Ge distributions more clearly seen in Fig. 2b) indicating homogeneous compound formation at the interface. The Auger signal in these profiles was calibrated against elemental standards and relative sensitivity factors were used. No matrix effects were taken into account. Scanning electron microscope examination of the sintered contacts showed only slight evidence for microsegregation. The maximum development of interfacial compounds corresponded to the minimum in the measured specific contact resistance, r e . Increasing either the 99
METALLURGICAL BEHAVIOUR OF Ni/Au-Ge OHMIC CONTACTS TO GaAs
NilAWAu-6s' f-t~s As evaporated
Vol. 49, No. 1
Ni/Au-6e/GoAs As-evaporat ed
80 100 120 160 160 1B0 Sputter time(rains)
6~ o/ ~ area
8=0 1~ 120 1~,0 160 Sputter time (rains)
Fig. 1. (a) Auger depth profile of as-deposited Ni/Au/ A u - G e contact. (b) Auger profffle after sintering at 450°C for 2.5 rain. sintering time or temperature led to considerably more diffuse depth profiles, greater penetration of Ni into the GaAs, and an increase in contact resistance re. For example, with an epilayer of free carrier concentration of 5 x 1016 cm -3 the minimum specific contact resistance was 3 x 10 -6 ~-cm 2 and occurred after 2.5 min sintering at 450°C. Increasing the sintering time to 5 min gave 5 x 10 -6 I2-cm2, whilst increasing temperature to 550°C for 2.5 min gave r e = 5 x 10 -5 ~2-cm~. A similar effect was observed during thermal life testing. Samples sintered to give the minimum specific contact resistance were heated at 330 ° C for periods of up to 100hr. Over the first 50hr increases in re of factors of 2 to 3 were observed; thereafter the resistance remained constant up to the maximum testing time for the Ni/Au/Au-Ge contacts. Au/Au-Ge contacts
Sputter time (rains) Fig. 2. (a) Auger depth profile of as.deposited N i / A u Ge contact. (b) Auger profile after sintering at 450°C for 2.5 min. annealed under identical conditions showed no significant change in contact resistance. 3. DISCUSSION The Auger profiles from many samples sintered to give minimum rc show a number of common features. Firstly, the Ni is always observed to move from the front surface to the interface with the GaAs, and the Ge distribution closely follows that of the Ni. It appears that the Ni has effectively gettered the Ge away from the Au (in some samples the Ge is below the level of detectability in the region of the Au). This result is close to the observations of Robinson . Robinson suggested that the role of Ni was to improve the wetting of the molten A u - G e to GaAs. However, the results of the present work suggest that owing to the manner in which Ge is
Vol. 49, No. 1
METALLURGICAL BEHAVIOUR OF Ni/Au-Ge OHMIC CONTACTS TO GaAs
leached away from the Au, the Au-Ge eutectic cannot play a part in contact formation. A more likely mechanism is that Ge is incorporated into the GaAs by a mechanism of solid state diffusion, and it is for this reason that the contacts have a homogeneous structure. The fact that the Ge is associated with the Ni at the interface supports the suggestion by Yoder  that the normally slow diffusion of Ge into GaAs is in some way enhanced by the much faster diffusing Ni. The second common feature in the profiles is the formation of interfacial compounds. This observation supports the identification of NiGe and NiAs at the reacted interface by Rackham et al.  using convergent beam electron microscopy, and that of Ogawa [ 11 ] who observed a number of interracial compounds although no contact resistance data were given in order to correlate the effect of the compounds on the contacts. Excess sintering (time and/or temperature) beyond that required to yield minimum re, always corresponded to the break up of these well defined interfacial layers in this work, with a resultant deeper Ni penetration into GaAs. Again, Yoder  has pointed out that such further Ni diffusion could lead to a reduction in the Ge donor concentration under the contact and a consequent increase in re. The results presented here are in accord with this model. 4. CONCLUSION Evidence has been presented which suggests that the action of Ni in Ni containing Au-Ge contacts to GaAs is to change the Ge incorporation mechanism from being eutectuc melting.recrystallization (as in the simple
Au-Ge contacts) to one relying on solid state diffusion. The structural homogeneity of the sintered contact is a result of the absence of melting-recrystallization processes. For application to practical devices it has been considered that the inhomogeneous nature of Au-Ge contacts could lead to unreliability. However, the results presented here show that although the Ni containing contacts have a more uniform structure, their stability to thermal stress is in fact worse that the simpler Au-Ge contacts. Acknowledgements - We wish to thank Mullards Hazel Grove for supplying the GaAs wafers and the SRC for the Auger system support grant. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
J.S. Harris, Y. Nanichi, G.L. Pearson & G.F. Day, J. AppL Phys. 40, 4575 (1969). C.J. Madams, D.V. Morgan & H.J. Howes, Electron. Lett. 11,574 (1975). J. Gyulai, J.W. Mayer, V. Rodriguez, A.Y.C. Yu & H.I. Copen, J. Appl. Phys. 42, 3578 (1971). G.Y. Robinson, Solid State Electron. 18, 331 (1975). K. Ohata, 12th Ann. Proc. IEEE Reliability Phys. Syrup., Las Vegas, 278, (May 1974). A. Christou, Solid State Electron. 22, 141 (1979). A. Iliadis & K.E. Singer, Solid State Electron. 26, 7 (1983). M.N. Yoder, Solid State Electron. 23,117(1980). N. Braslau, J.B. Gunn & J.L. Staples, Solid State Electron. 10,381 (1967). G.M. Rackham, J.W. Steeds & D.N. Merton-Lyn, CVD Proj. Report No. RU41-4 (AP) University of Bristol (1978). M. Ogawa, J. Appl. Phys. 51,406 (1980).