Thin Solid Films, 149 (1987) 251-260 GENERAL FILM BEHAVIOUR
F U N D A M E N T A L AND PRACTICAL ASPECTS OF A L L O Y I N G ENCAPSULATED GOLD-BASED CONTACTS TO GaAs A. J. BARCZ, E. KAME~SKA AND A. PIOTROWSKA Institute of Electronic Technology, AI. Lotnik6w 46, 02-668 Warsaw (Poland) (Received September 23, 1986; accepted November 4, 1986)
Au/GaAs and Au(Zn)/GaAs protected with either S i O 2 o r A1203 overlayers were alloyed at temperatures of 400-500 °C. Rutherford backscattering spectrometry, secondary ion mass spectrometry, scanning electron microscopy and X-ray diffraction show that the extent of interaction expressed in terms of the amount of gallium in the metal or the total volume of decomposed GaAs is reduced by a factor of 20 or more compared with unprotected contacts. A lower specific resistance of (SiO2)Au/Zn/Au ohmic contacts to p-GaAs is obtained over a wider range of processing temperatures. Also, the morphology of both the surface and the interface of the contact is significantly improved. The observed effects of encapsulation made it possible to formulate a revised model of interaction between gold and GaAs when annealed in the conventional open-system configuration. It is postulated that gold does not react directly with GaAs. The primary process responsible for decomposition of the semiconductor is the enhanced evaporation of arsenic through the metallic layer. Rapid reaction of the released gallium with gold is a secondary process which can be substantially limited by preventing the loss of arsenic.
Gold and gold-based alloys have been used to form ohmic contacts to GaAs devices for nearly 20 years 1. In spite of considerable work in this field 2 14, both the understanding of the basic processes which determine the electrical behaviour of the contact and the actual performance of the contacts are still far from satisfactory. Nevertheless, ohmic contact is formed by heating a suitably prepared gold-dopant mixture for a short time at temperatures in the range of 400-500 °C. Substantial migration of gallium atoms leads to A u - G a alloys of compositions which, according to the A u - G a phase diagram 15, may liquefy at these processing temperatures. It is postulated that, on cooling, a thin layer of the underlying GaAs receives high doping so that tunnelling may occur through the modified Schottky barrier ~0. Incorporation of large quantities of gallium has been detected by a number of techniques such as Rutherford backscattering spectrometry (RBS)12A3'16'17, Auger electron spectroscopy (AES) 5'7"18, energy-dispersive analysis of X-rays 0040-6090/87/$3.50
© Elsevier Sequoia/Printed in The Netherlands
A. J. BARCZ, E. KAMINSKA, A. PIOTROWSKA
(EDAX) 12, secondary ion mass spectrometry (SIMS)9,13 or electron microprobe analysis (EMPA)17. X-ray and electron diffraction studies have enabled identification of almost the entire spectrum of phases for the A u - G a system: ~ gold-rich solid solution 19, [3hexagonal (AuTGa2) 10,13,16,19 21, orthorhombic Au2Ga 16,19,22 and orthorhombic AuGa 19. In contrast with gallium, arsenic was found to evaporate rapidly through the metallic layer 23-26 at a rate higher than that from uncovered GaAs 27. The alloyed metallization appears highly inhomogeneous (in thin contacts even discontinuous) both laterally and vertically with respect to the initial semiconductor surface. A typical picture of the resultant interface consists of metallic inclusions 2°, rectangular on GaAs(100) or triangular on GaAs(111), which may be deep enough to make the alloyed contact hardly compatible with the fabrication geometry of submicron planar devices 28. The whole alloying process is extremely complex and not fully controlled; hence there is a large scatter in the data obtained in different laboratories. The aim of this work is primarily to determine the origin of the semiconductor dissolution, i.e. whether it is due to a direct chemical reaction with the contact constituents or to the thermal decomposition of GaAs itself. Futhermore, we shall show that substantial dissolution of GaAs can be considered as a byproduct of the reaction rather than a necessary condition for creating an ohmic contact. This was achieved by studying the metallurgical and electrical consequences of furnace annealing of gold or gold-based contacts encapsulated under different dielectric layers. An improved contact resistivity has already been reported 29 when an SiO2 overlayer was used in place of nickel in a standard Au/Ge/Ni metallization to n-GaAs. We shall demonstrate that appropriate protection from the loss of arsenic can be advantageous in all aspects of the technology of ohmic contacts. 2.
The substrates used in this study were (100)-oriented GaAs single crystals doped with zinc to p = 7 ×10-Z4m 3. The polished wafers were cleaned and chemically etched in a 3HeSO4:lH2OE:lHzO solution followed by hydrofluoric acid and 8 H E S O 4 : l H 2 0 2 : l H 2 0 . Either pure gold 100-300nm thick or, for electrical evaluation, Au(20 nm)/Zn(40 nm)/Au(270 nm) sandwich layers were deposited onto GaAs by evaporation from a resistance-heated boat in a conventional oil-diffusion-pumped system equipped with a liquid nitrogen trap. The pressure during evaporation was kept below 10 - 6 Torr. Prior to alloying, some of the samples were covered with different protective films of thickness about 200 nm: SiO2 chemically vapour deposited at 300 °C, AIEO3 r.f. sputtered at 150 °C or SiO2 magnetron sputtered at room temperature. Annealings were carried out in an opentube furnace with an H z flow at temperatures of 350-550 °C for 3 min. Most of the experiments described here were conducted with pure gold on GaAs(100) heated in hydrogen at 420 °C for 3 min. These annealing parameters had been shown to yield the minimum Au/Zn contact resistance to p-GaAs 13. For further analysis the dielectric films were removed chemically. The structure and composition of the contact layers were determined by complementary use of 2 MeV He + RBS, SIMS, X-ray diffraction (XRD) and
ALLOYING AU-BASED CONTACTS TO GaAs
scanning electron microscopy (SEM). RBS was applied for quantitative analysis of the in-depth distribution of the metal and to calibrate the SIMS signals. SIMS profiles were taken using a 4 keV Ar ÷ primary beam 150 ~tm in diameter scanned over an area of 2.5 mm x 2 mm. Secondary ions originating from the central part of the rastered area were separated in a quadrupole analyser and monitored using a pre-programmed unit. Because the high intensity of the gallium signal saturated the electron-multiplier under optimum conditions, it was necessary to design an electronic switch that moved the energy window to higher energies in the intervals when Ga ÷ counts were accumulated. Glancing angle XRD provided phase identification within the contact region. SEM revealed the morphology of the surface and, after appropriate chemical etching, also that of the interfaces. The etchant 1 0 ~ o C H 4 N 2 S: HC1:15~/oH20 (20:1:1) was found to be inert to GaAs. The specific resistance r c of the Au/Zn/Au/GaAs contacts was measured in a four-point configuration 3°. The test structure consisted of dots 200 lam in diameter arranged on a straight line at intervals of 500 lxm. 3. RESULTS A first observation of the effect of alloying an encapsulated Au/GaAs contact can be made with the unaided eye. After removing the protective oxide the underlying metallic surface remains golden coloured, indistinguishable from that of a non-alloyed sample. Similar non-protected contacts change their colour to silverish of an intensity roughly proportional to the temperature. Figure 1 shows scanning electron micrographs of the surface and interface of contacts alloyed with or without a chemically vapour-deposited SiO2 overlayer 200nm thick. It is immediately seen that the protected metallic surface (b) has a much better morphology than the unprotected surface (a) does. Micrographs (c) and (d) were performed on GaAs with the metal removed. For the conventionally annealed sample (c), an array of rectangular depressions aligned along the (110) direction cover about 30~ of the GaAs surface. The use of a cap (d) reduces both the area of pits and their average depth. By taking scanning electron micrographs at different angles it was possible to estimate the total reduction in the pit volume (which corresponds to the amount of decomposed GaAs) to yield a factor of 30 or more. Experiments carried out with sputtered SiO2 or A120 3 gave equivalent results except that in the latter case the resulting pits were less uniform with rounded corners. The spectra of backscattered He ÷ ions incident at a random direction on the Au/GaAs samples before and after annealing at 420 °C for 3 min in either opened or capped configuration are presented in Fig. 2. The full curve represents the spectrum from an as-deposited sample. The triangles correspond to a conventionally alloyed contact. A decrease in height of the uniform portion of the gold spectrum suggests that the gold contains certain component(s) of the GaAs. It will be shown below that this species is mainly gallium. From the tabulated values of the stopping crosssections for gold and gallium 31 a ratio of gallium to gold of 10~o is calculated. For the uncapped sample the signals from gold and GaAs partially overlap; the gold tail
A. J. BARCZ, E. KAMI/~SKA, A. PIOTROWSKA
Fig. 1. Scanning electron micrographs of a gold contact 300 n m thick on GaAs(100) annealed at 420 °C for 3 rain in H 2 (a),(c) without or (b),(d) with a chemically vapour-deposited SiO 2 overlayer 200 n m thick: (a),(b) contact surface (SiO 2 removed); (c),(d) GaAs surface (both SiO 2 and metallization removed).
reflects a distribution of the gold atoms into the substrate. As a consequence the spectrum originating from GaAs shifts towards higher energies and its intensity decreases. An interesting feature of the gold signal distribution is a fairly well-defined inflection point, indicated with a horizontal arrow. This is interpreted as a boundary between a uniform Au(Ga) layer and metallic protrusions. It should be noted that the height of this point relative to the maximum gold signal gives a direct measure of the pit coverage, here 55~o. A similar reasoning adopted for a cap-alloyed sample (circles) leads to the conclusion that no gallium is detected within the bulk gold layer and that much less gold is incorporated into GaAs (pit coverage, 8~). Because of its high sensitivity and adequate mass resolutions, SIMS appeared to be especially useful for studying the atomic transport in the contact volume. Figure 3 shows in-depth profiles of As + and Ga + secondary ion intensities monitored as a function of the sputter erosion time. For clarity, the signals are plotted on a logarithmic scale and gold profiles are not shown. In the non-processed sample there is already a detectable amount of gallium present in the gold film (the
ALLOYING AU-BASED CONTACTS TO G a A s
250 nm Au
1.5 ENERGY, MeV
~ I 1000
SPUTTER TIME [s]
Fig. 2. 2 MeV He + backscattering spectra of a gold contact 220 nm thick on GaAs: - - , as deposited; A, annealed at 420 °C for 3 min; O, same annealing but with a chemically vapour-deposited SiO2 cap (removed for analysis). Horizontal arrows represent the boundary between the uniform contact layer and the metallic inclusions. Fig. 3. 4keVAr ÷ SIMS profiles ofGa ÷ (filled symbols)and As ÷ (open symbols)in an Au(250nm)/GaAs contact: O, 0, non processed; A, A, annealed at 420°C for 3 min; [3, I , same annealing but with a chemically vapour-deposited SiO2 cap (removed for analysis). gallium b a c k g r o u n d in the apparatus was less than 50 counts s - 1), confirming the ability of G a A s to decompose even at r o o m temperature when in contact with gold 32. Annealing at 420 °C for 3 min results in a uniform distribution of gallium in gold, to a count rate of 1.9 x 104 counts s - 1. If we take as a reference the gallium signal from the bulk G a A s we m a y attempt to derive the gallium content from a simple proportion: 50~o x 1.9 × 104/1.2 × 105 = 8~o G a in the A u G a alloy, a figure which falls very close to that obtained with RBS. In view of the well-known matrix effects in S I M S this result is rather fortuitous but enables us to measure quantitatively the contact stoichiometry. In contrast, no arsenic is found in the alloyed sample. Annealing carried out with a chemically vapour-deposited SiO2 cap (squares) results in a certain in-diffusion of arsenic into the gold layer and in a m u c h smaller incorporation of gallium (0.4~). X R D phase identifications of (SiO2)Au/GaAs and (SiO2)Au/Zn/Au/GaAs heated to different temperatures are summarized in Table I. Because of the small lattice expansion for the A u G a ct solid solution its presence has been deduced from the gallium profiles and percentages as measured by S I M S or RBS. The lack of certain lines in the 13(AuTGa2) diffraction patterns suggests a preferred orientation of 13crystallites. N o evidence for the 13phase was found in contacts containing zinc. This m a y be due to the saturation of the whole population of gold atoms by ct solution and the c o m p o u n d Au3Zn.
A. J. BARCZ, E. KAMINSKA, A. PIOTROWSKA
TABLE I PHASECOMPOSITIONFOR GOLD (300 nm) AND Au(20 nm)/Zn(40 nm)/Au(270 nm) CONTACTSON GaAs AS A FUNCTION OF ANNEALINGCONDITIONS Contact
Phases observedfor the following annealing temperatures
Au/GaAs Au/Zn/Au/GaAs SiO2/Au/GaAs SiO2/Au/Zn/Au/GaAs
:t-Au-10 at.% Ga AuTGa2 ct-Au-10 at.% Ga Au3Zn[H] ct-Au4).3 at.% Ga
ct-Au-12 at.% Ga AuTGa 2 ct-Au 12 at.% Ga AuaZn[H ] ~t-Au4).6 at.% Ga
No Au or ct-AuGa New unidentified phases
a-Au4).3 at.%Ga Au3Zn[H]
ct-Auq3.6at.% Ga AuaZn[H ]
ct-Au-1 at.% Ga AuvGa 2 (traces) =-Au-1 at.~o Ga Au3Zn[H] AUTGa2 (traces)
The specific contact resistance is plotted v s . annealing temperature in Fig. 4 for Au/Zn/Au and SiO2/Au/Zn/Au/GaAs. It should be noted that, apart from the slightly lower absolute value of re, encapsulation allows heating over a wider range of temperatures without noticeable degradation of the contact. 20 20 4O
15 E o
z _ to
ANNEALING TEMPERATURE [°C]
Fig. 4. Specific resistance of an Au/Zn/Au contact to p-GaAs vs. processing temperature with ( without (- - -) surface protection by SiO2 (annealing time, 3 min).
4. DISCUSSION The aim of this experiment was to demonstrate that reaction between GaAs and a gold-based contact can be substantially restrained when annealed under a
ALLOYING AU-BASED CONTACTS TO
protective encapsulant of S i O 2 o r A1203. The technological advantages of this technique are self-explanatory: better contact morphology for easier bonding, lower specific resistance and, possibly the most important of all, suppressed penetration of the gold alloy into the semiconductor which otherwise unavoidably downgrades the performance of planar devices. Regularly shaped metallic protrusions are believed to be the product of solidification of the liquid portion of the contact. Our observation of lower resistivity achieved for a contact with substantially reduced population of pits seems to contradict a model according to which conduction through the array of metallic indentations is responsible for ohmic behaviour 2s. The pit density presented in Fig. l(d) is an average taken from the inspection of a large number of samples. Higher pit coverages were usually correlated with imperfections visible on the protective layer and having the form of cracks or pinholes. In contrast, there certainly exists a lower limit to which the decomposition of GaAs can be reduced by applying a surface barrier to prevent the escape of arsenic. This limit is determined by the finite amount of both arsenic and gallium accommodated within the metallic layer itself (see Fig. 3). Arsenic must then probably be located at grain boundaries because its solid solubility in gold is negligible a3. The results obtained with cap-alloyed contacts make it possible to formulate a model for the interaction between gold and GaAs on conventional open-system heating. The commonly adopted mechanism for the formation of an ohmic contact requires high doping of the underlying semiconductor by the initially deposited donor or acceptor elements 1°. The only evidence that such a layer exists was provided when fabricating an n +-p + tunnel junction by alloying an AuGe contact to p +-GaAs 34.35. Our attempt to identify a zinc-rich p + layer in GaAs (after removal of the alloyed AuZn contact) by capacitance-voltage measurements or SIMS analysis gave no sign of a highly-doped volume of the semiconductor. Nevertheless, zinc must somehow play a doping role simply because A u - Z n does form ohmic contacts to moderately doped p-GaAs and pure gold does not. The problem is only how such a layer might be produced. For the annealing temperatures and times employed the classical diffusion of dopants into single-crystal GaAs may safely be neglected. A mechanism which is favoured in the literature assumes partial dissolution of GaAs in the molten metal with subsequent regrowth and formation of an n ÷ (p+) film 1°. However, congruent dissolution of a compound with a melting point of 1240°C would necessitate A u - G a - A s ternaries or A u - Z n - G a - A s ( A u - G e - G a - A s ) quaternaries with an arsenic-to-gallium ratio of unity and with eutectic point(s) below 500 °C. This has not been found experimentally for the first system 36 and remains very unlikely for the other two. Limited solubility of GaAs in an AuGe eutectic melt was claimed above 500 °C 37 but the amount of arsenic that may (and certainly does) escape before solidification was not taken into account. At this point we should like to stress that, energetically, GaAs is a stable stoichiometric compound with a standard free enthalpy 3s of - 100 kJ m o l - x and its only weakness is to lose arsenic by vaporization at relatively low temperatures. In a series of mass spectrometric studies 23-26 the evaporation rate of arsenic from beneath gold-based contacts was found to be higher than that from pure GaAs. This means that gold can be considered to be a catalyst that decreases the activation energy of arsenic evaporation in a similar fashion to the behaviour reported for antimony on GaAs 39.
A . J . BARCZ, E. KAMINSKA, A. PIOTROWSKA
From this it follows that it is the blocking of the arsenic loss that practically inhibits the decomposition of GaAs and that arsenic evaporation is the primary process that initiates the interaction between the gold and the semiconductor. In view of this it is worth while to recall that arsenic is not present in the contact alloyed in the open system (Fig. 3). It therefore becomes clear that GaAs cannot form the hypothetical p÷ (n ÷) film by epitaxial regrowth from the melt because only one constituent (gallium) is available in the molten contact. In conclusion, we postulate that ohmic character is determined by a suitable arrangment of bonds involving dopant atoms at a very intimate metal-semiconductor interface without, however, creating a finite layer of appreciable thickness. Additional information on the Au-GaAs interaction can be gained by analysing the thermodynamic properties of the compounds involved. The standard enthalpy of formation for GaAs is 38 - 8 2 kJ mol-~. No experimental data have been found for the heats of formation of the A u - G a compounds. We made use of computed values based on a semi-empirical procedure developed by Miedema 4° and its further refinements 41. The data have been shown to be accurate for binary systems containing at least one transition or noble metal element. If we now take the 1:1 AuGa compound of highest negative enthalpy ( - 34 kJ moi - 1) and consider the reaction of gold with it according to the scheme Au + G a A s ~ AuQa + As
we realize that this reaction is associated with a high positive change in enthalpy AH = ( - 3 4 ) - (-- 82) = + 48 kJ m o l - 1, i.e. strongly prohibited. For other A u - G a compositions the situation is even worse. The above estimate is another strong indication that gold does not react directly with GaAs. The only model that seems to be consistent with the results of this work is a two-step interaction, as follows. First, the GaAs decomposes by evaporation of arsenic, with a catalytic action of gold: Au
2GaAs ~ 2Ga + As2 T
Secondly, the gold reacts with the released gallium: mAu + nGa ~ AumGa,
Reaction (3) is expected to be very rapid because of the negative enthalpies for all A u - G a compositions including solid solutions 4 ~ and the high mobility of gallium in gold at low (down to room) temperatures 42. Reactions (2) and (3) will proceed until the gold becomes saturated with gallium. Fast removal of the excess gallium appears to be a necessary condition for arsenic to evaporate and, conversely, excessive accumulation of gallium is a factor terminating the evaporation of arsenic as well as the whole reaction. It has been shown earlier that the addition of gallium to the gold contact significantly lowers the arsenic evaporation rate 43. If it is thick enough the "saturated" contact takes the form of a uniform layer consisting of ~-AuGa solution with AuTGa2 (Au2Ga) crystallites within or near the GaAs surface. By sealing the Au/GaAs structure with a suitable protective film the overpressure of arsenic exerted on the GaAs surface prevents excessive decomposition of
ALLOYING AU-BASED CONTACTS TO GaAs
the s e m i c o n d u c t o r . T h e o u t c o m e o f a r e c e n t e x p e r i m e n t w i t h A r ÷ b e a m a l l o y i n g o f the A u / G a A s s y s t e m 44 a p p e a r e d to be v e r y s i m i l a r to t h a t of t h e r m a l h e a t i n g : the a r s e n i c was f o u n d to be p r e f e r e n t i a l l y s p u t t e r e d off t h r o u g h the g o l d l a y e r a n d , m o r e o v e r , the e x t e n t o f i o n m i x i n g was r e d u c e d w h e n b o m b a r d m e n t was p e r f o r m e d t h r o u g h a t h i n A1203 surface barrier. ACKNOWLEDGMENTS F r u i t f u l d i s c u s s i o n s w i t h P r o f e s s o r M.-A. N i c o l e t ( C a l i f o r n i a I n s t i t u t e o f T e c h n o l o g y ) are g r e a t l y a c k n o w l e d g e d . T h e a u t h o r s are also i n d e b t e d to Dr. A. R. M i e d e m a of Philips, E i n d h o v e n , for g e n e r o u s l y p r o v i d i n g the c o m p u t a t i o n s o f t h e r m o d y n a m i c d a t a c o n c e r n i n g b i n a r y systems. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
N. Braslau, J. B. Gunn and J. L. Staples, Solid-State Electron., 10 (1967) 381. J. Gyulai, J. W. Mayer, V. Rodrigues, A. Y. C. Yu and H. J. Gopen, J. Appl. Phys., 42 (1971) 3678. A.K. Shinha and J. M. Poate, Appl. Phys. Lett., 23 (1973) 666. V.L. Rideout, Solid-State Electron., 18 (1975) 541. G.Y. Robinson, Solid-State Electron., 18 (1975) 331. M. Wittmer, R. Pretorius, J. M. Mayer and M.-A. Nicolet, Solid-State Electron., 20 (1977) 433. A. Christou, Solid-State Electron., 22(1977) 141. D.C. Miller, J. Electrochem. Soc., 127 (1980) 467. M. Heiblum, M. I. Nathan and C. A. Chang, Solid-State Electron., 25 (1982) 185. A. Piotrowska, A. Guivarc'h and G. Pelous, Solid-State Electron., 26 (1982) 179. C.R.M. Grovenor, Thin Solid Films, 104(1983)409. A.K. Rai, R. S. Bhattacharya and Y. S. Park, Thin Solid Films, 114 (1984) 379. E. Kaminska, A. Piotrowska, A. Barcz, J. Adamczewska and A. Turos, Solid-State Electron., 29 (1986) 279. C.J. Palmstrom and D. V. Morgan, in M. J. Howes and D. V. Morgan (eds.), Gallium Arsenide, Wiley, New York, 1985. C.J. Cooke and W. Hume-Rothery, J. Less-Common Met., 10 (1986) 42. J.M. Vandenberg and E. Kinsbron, Thin Solid Films, 65 (1980) 259. B. Pecz, E. Jaroli, Gy. Radnoczi, R. Veresegyhazy and I. Mojzes, Phys. Status Solidi A, 94 (1986) 507. Bor-Long Twu, Solid-State Electron., 22 (1979) 501. Xian-Fu Zeng and D. D. L. Chung, Thin Solid Films, 93 (1982) 207. K. Kumar, Jpn. J. Appl. Phys., 18 (1979) 713. T. Yoshiie and C. L. Bauer, J. Vac. Sci. Technol. A, 1 (1985) 554. T. Yoshiie, C. L. Bauer and A. G. Milnes, Thin Solid Films, 111 (1983) 149. T. Sebestyen, M. Menyhard and D. Szighethy, Electron. Left., 22 (1976) 96. I. Mojzes, T. Sebestyen and D. Szighethy, Solid-State Electron., 25 (1982) 449. E. Kinsbron, P. K. Gallagher and A. T. English, Solid-State Electron., 22 (1979) 517. S. Leung, L. K. Wong, D. D. L. Chung and A. G. Milnes, J. Electrochem. Soc., 130 (1983) 462. I.R. Arthur, Surf Sci., 43 (1974) 449. N. Braslau, J. Vac. Sci. Technol., 19 (1981) 803. F. Vidimari, Electron. Lett., 15 (1979) 675. E. Kuphal, Solid-State Electron., 24 (1980) 69. W.K. Chu, J. W. Mayer and M.-A. Nicolet, Backscattering Spectrometry, Academic Press, New York, 1978. A. Hiraki, K. Shuto, S. Kim, W. Kammura and M. Iwami, Appl. Phys. Lett., 31 (1971) 611. M. Hansen, K. Anderko, Constitution of Binary Alloys, McGraw-Hill, New York, 2nd edn., 1958.
A. J. BARCZ, E. KAMII'~SKA, A. PIOTROWSKA
34 N. Honolyak, D. L. Keune, R. D. Burnham and C. B. Duke, Phys. Rev. Lett., 2,~ (1970) 589. 35 A.M. Andrews and N. Honolyak, Solid-State Electron., 15 (1972) 601. 36 M.B. Panish, J. Electrochem. Soe., 114 (1967) 516. 37 M. Otsubo, H. Kumabe and H. Miki, Solid-State Electron., 20 (1977) 617. 38 J. Barin, O. Knacke and O. Kubashewski, Thermochemical Properties of Inorganic Substances, Suppl. Vol., Springer, Berlin, 1977. 39 C.Y. Lou and G. A. Samorjai, J. Chem. Phys., 55 (1971) 4554. 40 A.R. Miedema, Philips Technical Rev., 36 (1976) 217. 41 A . K . Niessen, F. R. de Boer, R. Boom, P. F. Chatel, W. C. M. Matterns and A. R. Miedema, Calphad, 7 (1983) 51. 42 V. Simic and Z. Marinkovic, Thin Solid Films, 34 (1976) 179. 43 T. Sebestyen, I. Mojzes and D. Szighethy, Electron. Lett., 16 (1980) 505. 44 A.J. Barcz, M. Domanski, E. Kaminska and J. Jagielski, Proc. Conf. on Ion Beam Modification o f Materials, Catania, 1986, in Nucl. Instr. Methods. Sect. B, to be published.