THE SILICON/GERMANIUM (111) INTERFACE : THE ONSET OF EPITAXY J. Woicik, R. List, B. Pate, P. Pianetta To cite this version: J. Woicik, R. List, B. Pate, P. Pianetta. THE SILICON/GERMANIUM (111) INTERFACE : THE ONSET OF EPITAXY. Journal de Physique Colloques, 1986, 47 (C8), pp.c8-497-c8-501. <10.1051/jphyscol:1986893>. <jpa-00226224> HAL Id: jpa-00226224 https://hal.archives-ouvertes.fr/jpa-00226224 Submitted on 1 Jan 1986 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
JOURNAL DE PHYSIQUE Colloque C8, supplbment au no 12, Tome 47, dbcembre 1986 THE SILICON/GERMANIUM (111) INTERFACE : THE ONSET OF EPITAXY J.C. WOICIK, R.S. LIST, B.B. PATE and P. PIANETTA Stanford Synchrotron Radiation Laboratory, Stanford, CA 94305, U.S.A. X-ray absorption fine structure, Auger electron spectroscopy, and low energy electron diffraction have been used to study the evolution of the lattice constant and the degree of intermixing of thin (=5 A) silicon films grown on the germanium(ll1) surface by solid phase epitaxy. As the system is annealed from 370 to 800 degrees centigrade, substantial intermixing between the silicon and germanium is observed. An additional peak in the Fourier transform of the EXAFS grows with annealing temperature while the initial single silicon peak of 2.35 A decreases in amplitude and shifts towards larger radii by about 0.1 A. This shift is consistent with strained Si-Si bonding with a bond length of 2.44 A, i.e., that of the germanium substrate. Computer simulations show that the additional peak is due to strained Si-Ge bonding with a bond length of 2.44 A as well. As the system is annealed, the near edge structure becomes similar to that of crystalline germanium while a new feature develops at the silicon Is-threshold which we attribute to excitonic effects as the alloy becomes germanium rich. After the 800 degree C anneal, a faint 1x1 LEED pattem is visible while initially it is not. The above observations indicate that by solid phase epitaxy we have grown a pseudomorphic Si-Ge alloy on a germanium(ll1) substrate with the germanium bulk lattice constant The technological importance of semiconductor superlattices and heterojunctions has grown significantly in the past few years. The silicon/germanium system has become increasingly popular for a variety of reasons. First, both are group IV tetrahedrally coordinated semiconductors which form uniform alloys of any composition range while possessing a significant 4% lattice mismatch. Recent work has shown that strained-layer OeSiISi superlattices can be grown on silicon without misfit dislocations if the layer thickness is not too great [I]. Further, the possibility of producing tunable band gap semiconductors also exists by varying the relative concentration of Si to Ge [2]. Likewise, the SiIGe system serves as a prototypical system for model calculations and for the study of growth dynamics for the more complicated strained-layer superlattices based on ID-V compounds. Our interest lies in studying the evolution of the lattice constant and electronic structure as the two materials intermix. In this way we may gain an understanding of the behavior of the silicon and germanium lattices during the initial stages of epitaxy. Using x-ray absorption fine structure, Auger electron spectroscopy, and low energy electron diffraction, we have studied the evolution of the lattice constant and the degree of intennixing of thin (=5 A) silicon films grown on the germanium(ll1) surface by solid phase epitaxy. The samples were prepared by cleaving germanium single crystals in an ultra-high vacuum chamber @11x10'~~ tom) and depositing 5 A of silicon by electron beam evaporation and subsequently annealing the samples for 15 minutes at 370, 620, and 800 degrees centigrade. The silicon K-edge absorption spectra were taken by monitoring the silicon KLL Auger electrons as a function of photon energy and normalizing to the incident flux from the Jumbo beamline at SSRL after the room temperature growth and after each anneal. The spectra were taken 30 degrees from normal incidence. After the absorption data was collected, a new sample was prepared and the Auger electron experiment was performed under identical conditions. The experiment was divided into two portions in order to avoid electron beam damage to the sample during the collection of the absorption data. After an 800 degree C anneal, a faint 1x1 LEED pattern is typically visible while for the room temperature growth it is not. Figure 1 shows the silicon K-edge absorption spectra as a function of annealing temperature. Figure 2 shows the corresponding fine structure, X(k). In obtaining X(k), two spline knots were used, E, was taken as 1845 ev, and the spectra were normalized to the edge jump. It is evident from Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1986893
C8-498 JOURNAL DE PHYSIQUE figure 1 that as the system is annealed the amplitude of the edge jump decreases by a factor of about three fifths while the near edge structure sharpens and becomes similar to that of crystalline germanium [3]. It is evident from figure 2 that the spectrum corresponding to the room temperature growth exhibits sinusoidal single bond length modulations which decrease in magnitude and become more characteristic of a multi-component signal. Fig. 1. Silicon K-absorption spectra as a function of annealing temperature. Fig. 2. The corresponding fine structure as a function of annealing temperature. Shown in figure 3 is a plot of the absolute peak to peak intensities of the germanium MVV (52 ev) and the silicon LVV (92 ev) surface sensitive Auger electron data as a function of annealing temperature. In obtaining the Auger data, a primary electron beam energy of 3 kev was used. The ratio of the silicon to germanium cross sections is about 10 to 1 and the escape depth for both is approximately 5 A. Originally, the germanium signal is reduced by an escape depth while the silicon signal is nearly that of the bulk material. As the system is annealed, we note a smooth increase in germanium intensity, while in contrast, the silicon intensity decreases smoothly until the 800 degree C anneal where the decrease then becomes quite dramatic. These results suggest a smooth intermixjng process upon annealing until the 800 degree C anneal where the intermixing has become complete enough to allow the germanium to surface segregate. Since the edge jump decreases by a factor of three fifths and since the KLL Auger electrons of 1610 ev used to measure the absorption have an escape depth of 25 A, we estimate the extent of intermixing to be 15 A with a monolayer of germanium on the surface giving rise to the fmallx1 LEED pattern after the alloy has crystallized. In order to determine the sbucture about the silicon atoms as a function of annealing temperature, the k-squared weighted gourier transforms of the fine structure of figure 2 were computed and are shown in figure 4. In obtaining the Fourier transforms, a ten percent Hanning window was used and the data transformed from k=2 to k=9 A-l. The k-range was extended down to k=2 A'l in order to improve our radial resolution. For the room temperature growth, the Fourier transform appears as a singly peaked function corresponding to the Si-Si bond length of 2.35 A. As the system is annealed,
Fig. 3. A plot of the Si and Ge surface sensitive Auger peak to peak intensities versus annealing temperature. -......... 0 100 200 300 400 500 600 700 800 DEGREE C ANNEAL Fig. 4. The k-squared weighted Fourier transforms of the experimentally determined fine structure as a function of annealing temperature. Fig. 5. The Fourier transforms of the simulated data. RADIUS h an additional peak in the Fourier transform grows with annealing temperature while the initial single silicon peak narrows and decreases in amplitude as it continuously shifts towards larger radii by about 0.1 A. Higher Fourier components also become evident indicating that the alloy has crystallized. In order to understand the nature of the second peak, EXAFS data was simulated for Si-Si bonding of 2.35 A, Si-Ge bonding of 2.44 A, and a Si-Ge alloy with 50% Si-Si (2.35 A) and 50% Si-Ge (2.44 A) bonding. The phase shifts and backscattering amplitudes were taken from Teo and Lee 141 and the Debye-Wallet factor from reference 5. The k-squared Fourier transforms of the simulated data were computed in the same fashion as the transforms of the experimental data and are shown in figure 5. We conclude that the second peak in our data is due to Si-Ge bonding and is well resolved since the total phase difference between the Si-Si and Si-Ge bonds passes through pi in our k-range. As the calculation suggests, when two such sine waves of opposite phase are added and then Fourier transformed, the magnitude of the Fourier transform exhibits a node close to their
JOURNAL DE PHYSIQUE average distance. In contrast to the simulated data, as noted above, the peak which corresponds to Si-Si bonding shifts 0.1 A towards larger bond length. This shift is consistent with strained Si-Si bonding with a bond length of 2.44 A, i.e., that of the germanium substrate. We feel that this shift is significant since it corresponds to the bulk crystalline germanium spacing as the silicon is incorporated into the germanium lattice. It is also noted that the near edge structure becomes similar to that of crystalline germanium. Combined with our Auger results, the fact that the Si-Si peak narrows and continuously shifts towards larger bond length as its amplitude decreases suggests a process where the intermixing occurs across the interface with increasing annealing temperature. Our analysis also suggests a variable Si-Si bond length and a variable alloy concentration across the interface for the lower temperature growths. The Si-Ge bond maintains a constant length, namely 2.44 A, while the Si-Si bond length is gradually strained from its initial value of 2.35 A to that of the germanium substrate, 2.44 A. As the silicon and germanium intermix the bond lengths at the interface become strained until the concentration of germanium becomes great enough to accomodate a totally strained system. As mentioned previously, as the system is annealed, the absorption edge evolves into a sharp doubly peaked structure. Shown in figure 6 is the evolution of the absorption edge as a function of annealing temperature on a scale where the growth of the double edge is more apparent. The sha?, double structure closely resembles the K-absorption edge of diamond [a. In the diamond study, the spike at threshold was well modeled by the Elliot form [7] of a lscore exciton in the effective mass approximation. The difference in appearance between the Si-Ge spectra and that of diamond near threshold is attibuted to poorer monochromator resolution and greater core hole broadening at the silicon K-edge. Recent work shows that the excitonic binding energy in semiconductors increases with compositional disorder about the silicon atom An excitonic binding energy at the Si b,m edge of 0.12 ev for pure silicon compared to 1.5 ev for a Si-C alloy is reported [8]. This result agrees with our findings since the exciton becomes clearly resolved from the continuum as the concentration of germanium is increased in the silicon film. The original exciton in the amorphous silicon overlayer apparently lies too close to the continuum and can not be resolved. 8 7 5 - Fig. 6. The evo'ution of the Si K-absorption edge as a function of annealing temperature. 1837.5 1848.B 181125 1845.8 IEY7.5 18588 1852.5 1855.8 Ewrp~ In conclusion, we have studied the solid phase epitaxial growth of thin silicon films on the germanium(ll1) substrate. Our results show that the silicon intermixes with the germanium at elevated temperatures to form a pseudomorphically strained alloy. The initial Si-Si bond length (2.35 i() is gradual1 strained to that of the germanium substrate (2.44 A) in contrast to the Si-Ge bond length (2.44 d) which remains constant and increases in amplitude as the silicon and germanium intennix. Our reported bond lengths are to be compared to the Ge-Ge (2.45 A) and the Si-Ge (2.37 A) bond lengths found in the amorphous alloys of reference 5. Note that for the amorphous alloy, the Ge-Ge bond length is preserved while the Si-Ge bond length is closer to the average of the two. Therefore we conclude that the presence of the germanium substrate serves to strain the Si-Si and Si-Ge bonds. After the 800 degree C anneal, a new structure develops at the 1s-threshold which we attribute to excitonic effects as the alloy becomes germanium rich. 1) R. Hull, J.C. Bean, F. Cerdeira, A.T. Fiory, and J.M. Gibson, Appl. Phys. Lett. 48 (I), (1986). 2) D. J. StukeI, Phys. Rev. B 3 (lo), 3347 (1971).
3) M.G. Proietti, L. Incoccia, S. Mobilio, A. Gargano, and F. Evangelisti, in EXAFS and Near Edpe Structure 111, K.O. Hodgson, B. Hedman, and J.E. Penner-Hahn, eds., (Springer-Verlag, Berlin, 1984), p. 26. 4) Boon-Keng Teo, P.A. Lee, A.L. Simons, P. Eisenberger, and B.M. Kincaid, J. Am. Chem. SOC., 99 (ll), 3854 (1977). 5) L. Incoccia, S. Mobilio, M.G. Proietti, P. Fiorini, and F. Evangelisti, in EXAFS and Near Edge Structure 111, K.O. Hodgson, B. Hedman, and J.E. Penner-Hahn, eds., (Springer-Verlag, Berlin, 1984), p. 284. 6) J.F. Morar, F.J. Himpsel, G. Hollinger, G. Hughes, and J.L. Jordan, Phys. Rev. Lett. 54 (17), 1960 (1985). 7) R.J. Elliott, Phys. Rev. 108 (6), 1384 (1957). 8) F. Evangelisti, F. Patella, M.K. Kelly, R.A. Riedel, G. Margaritondo, P. Fiorini, P.Perfetti, and C. Quaresima, in.proceedings of the 17th International Conference on the Phvsics of Semiconductots, J.D. Chadi and W.A. Hamson, eds., (Springer-Verlag, New York, fl85), p. 1235.