Epitaxial growth of Bi on GaAs(100) surfaces

Epitaxial growth of Bi on GaAs(1 00) surfaces s. Horng and A. Kahn Department ofElectrical Engineering, Princeton University, Princeton, New Jersey 08544 (Received 8 February 1989; accepted 31 March 1989) Bi layers deposited on (100) GaAs are investigated with low-energy electron diffraction, Auger electron spectroscopy, and electron energy-loss spectroscopy. The (100) substrates are prepared by simultaneous Ar+ sputtering and annealing. The deposition ofBi on room-temperature (100) GaAs eliminates the reconstruction spots at 0.5 monolayer. No long-range order is obtained in the room temperature grown Bi layer. Post-growth annealing at 250 ·C restores long-range order and produces a sixfold symmetric hexagonal pattern characteristic of the rhombohedral structure of bulk Bi. The attenuation of the Auger peak intensities as a function ofBi thickness indicates that the film grows in a quasi-laminar fashion. The deposition ofBi on high-temperature (100) GaAs (250 ·C) produces a two-dimensional ordered growth. The GaAs (4 X 6) structure is replaced by a (2 Xl) structure after deposition of one-half monolayer Bi. These (2 Xl) structure remains visible up to coverages of at least 25 monolayers. Auger measurements indicate island formation, while the sharpness of the electron diffraction spots indicates good crystallinity at the surface of the Bi crystal. I. INTRODUCTION The growth of Bi thin films has been the subject of many investigations over the years. Considerable interest in the properties of such films was generated by the prediction of quantum size effects. I- 3 Quantum size effects arise from a quantization of the electron or hole wavevector perpendicular to the film imposed by the boundary conditions at the surface. Bi, being a semimetal, is an ideal candidate for such manifestation because the electron effective mass is very small (0.015-D.018 me in the trigonal direction). Thus the deBroglie wavelength is long. If the Bi films can be made thin enough (-100 A), size quantization of the electron and hole bands should drive the energy levels high enough that the band overlap present in the bulk Bi solid is transformed into a band gap, and thus lead to semiconducting behavior. Structures that make use of such an effect involving bandinverted semiconductor Bi/Bi l _ xSbx heterojunctions have been proposed. 4 A better understanding of the epitaxial growth and properties ofBi thin film is therefore a first step in such investigations. Bi films have been grown on various materials such as mica, BaF2 , PbTe, Cu, and NaC1. 5- 8 Most of the Bi films grown are oriented with the trigonal axis (001) perpendicular to the substrate, but are, however, polycrystalline on the growth plane. Recently, Joyce et al. 9 have investigated the deposition ofBi on the GaAs( 110) surface, and claimed that the surface structure of Bi retains the symmetry of the substrate up to a coverage of 10 ML. The bondlength in Bi is 4.54 A. On the GaAs ( 110) surface, the unit cell is rectangular with 3.99 A and 5.65 A sides respectively, while on GaAs( 1(0) it is square with 3.99-A sides. The (100) is the preferred orientation for the growth of GaAs by molecularbeam epitaxy, and the realization of thin epitaxial metal films on the (100) surface is therefore highly desirable. In this paper, we investigate the growth of Bi overlayers on GaAs( 1(0) surfaces at various temperatures and find that Bi overlayers grown at high temperature (100 ·C) has long931 J. Vac. Sci. Technol. B 7 (4), Jul/Aug 1989 range order with a structure which appears to be imposed by the substrate, although the deposited Bi also tends to form islands. II. EXPERIMENTAL TECHNIQUES We used horizontal-bridge grown GaAs, doped with Si to 1.5 X 10 18 cm- 3 , and cut and polished along the (100) plane. The samples were either In-mounted on a molybdenum block or craddled in a molybdenum foil for resistive heating, and placed in a UHV chamber (base pressure = 10- 10 Torr). The (100) surface was cleaned by ion bombardment with a 1.0-1.5 keY Ar+ beam at an incidence angle of 70· and simultaneously annealed at - 500·C. Low-energy electron diffraction (LEEO) was used to monitor the surface structure after annealing. The clean surfaces exhibited the ( 4 Xl) or ( 4 X 6) structure. We believe that the ( 4 Xl) surface is actually an uncompletely reconstructed ( 4 X 6) structure. The conditions that lead to such distinction were somewhat uncontrollable in our experiment. Bi was evaporated from a resistively heated W filament. The deposition was monitored with a quartz thickness monitor from Sloan Technology at a typical deposition rate of -4 A/min. One monolayer (1 ML) of Bi on GaAs(100) has a nominal thickness of 2.23 A. Bi was deposited on samples held at room temperature, 100, or 250 ·C. The sample temperature was calibrated using a thermocouple attached to the molybdenum block or foil. Auger electron spectroscopy (AES) and electron energy-loss spectroscopy (EELS) data were recorded after cooling to room temperature. Low- and highenergy AES spectra were recorded with a single-pass cylindrical mirror analyzer (CMA) to monitor surface contamination and relative change in the Ga and As signals. A 2-V peak-to-peak modulation was used for the high-energy peaks, while 1 V was used for the low-energy peaks. LEED measurements were performed with a standard four-grid optics, with the samples placed in position of normal incidence. EELS data were obtained in the second derivative mode with 0734-211X/89/040931-Q5$01.00 © 1989 American Vacuum Society 931 932 S. Horng and A. Kahn: Epitaxial growth of Bi on GaAs(100) surfaces 932 the single-pass CMA using a 150-eV primary electron beam. The modulation used for EELS was 0.5 V peak-to-peak. III. RESULTS AND DISCUSSION After simultaneous Ar + cleaning and annealing, the ( 1(0) substrates exhibited the 4 X I or 4 X 6 LEED patterns characteristic of a Ga-rich surface [see Figs. l(a),I(b»). The room-temperature deposition of 0.5 monolayers (ML) of Bi was sufficient to eliminate the fractional-order spots in the LEED patterns and, hence, the surface reconstruction. The LEED pattern was entirely eliminated with - 3-4 ML, indicating a lack oflong-range order in the Bi layer grown at room temperature. A post-growth annealing at 250 ·C for 50 min restored the long-range order and produced a sixfold symmetric hexagonal LEED pattern. This pattern was similar to that expected for the rhombohedral structure of Bi in the trigonal direction. The attenuation of the low- and highenergy AES Ga and As peak intensities as a function of Bi deposition is shown in Fig. 2(a). The exponential attenuation of the normalized intensities clearly indicates a quasilaminar growth at room temperature (R T). It was fitted with the empirical relation I = e - d / A where I is the normalized intensity, d is the thickness of overlayer, and A is the electron inelastic mean free path at the relevant energy. Inelastic mean free paths of - 5-6 A for the low-energy peaks [Ga( 55 eV), AsOI eV) ) and - 15-20 A for the high-energy peaks [Ga( 1070 eV)], [As( 1228 eV) 1 were found, in good agreement with expected values within the experimental accuracy. The initial drop at low coverage corresponds to an inelastic mean free path of - 2 A; this exceedingly small value may come from some chemical effects such as the broadening of Auger peaks. Further experiments are needed to identify tile cause of this initial drop. The high-temperature (250 ·C) deposition of Bi led to a different overlayer morphology. First, the initial (4 X I ) or (4 X 6) GaAs structure was changed to a (2 X I) structure upon deposition of - I-ML Bi [Fig. I (c»). This LEED pattern remained visible after further deposition up to coverages of at least 25 ML. The attenuation of the Auger peaks l Fig. 2 (b)], however, indicated that the 250°C growth is not laminar. The attenuation rate of the low- and high-energy peaks must be fitted with unrealistically large inelastic mean free paths, and suggests either islanding or extensive interdiffusion ofGa and As through the interface. The retention of the (2 X I) pattern argues in favor of the former. The 250·C growth appears therefore to follow a Stranski-Krastanov growth model. The EELS spectra is shown in Fig. 3. The 3.4- and 5.6-eV peaks correspond to valence-band to conduction-band transitions. The 10.4 and 16.2 eV are the surface and bulk plasmons, respectively. The 19.5-21 eV (42 eV) peaks correspond to Ga-3d (As-3d) to conduction-band transitions respectively. \0 The attenuation of bulk plasmon peak is slow as expected from the island nature of the growth. The surface plasmon changes at - I ML, corresponding to the transition from (4 X I) or (4 X 6) to (2 X I) in LEED observations. The elimination of the 3.4-eV peak at -0.5 Ais also surprising, because, as a bulk transition, it should not disappear at such low coverage. This quick elimination might result from J. Vac. Sci. Technol. B, Vol. 7, No.4, Jul/Aug 1989 (a) (el FIG. 1 LEEDpaltern.,observcd: (aJ (4 I) and (b) (4x6) GaAs(100) after simultaneous ion bombardment and (c) annealing (2 Xl) pattern after 2 A of Bi depo,itioll at 250 T. the appearance of a 4-e V peak from Bi layer, since the EELS spectrum is taken in second derivative mode which is quite sensitive to change in lineshape. Finally, the EELS spectrum of the sample after deposition of214 ABi at 250°C shows the expected feature for Bi: the plasmon at IS eV and the peak at 10 eV which we tentatively attributed to Bi surface plasmon. The origin of the 4-eV peak remains unknown. Given that Bi deposited at room temperature is not atomically ordered but is quasi-laminar, while Bi deposited at 250°C tends to form islands, an intermediate substrate temperature of 100 °C was investigated. LEED spots were ob- 933 S. Horng and A. Kahn: Epitaxial growth of Bi on GaAs(100) surfaces at 250°C [Fig. 2 (c) J. Since the Bi deposited at high temperature forms islands, the origin of the LEED patterns is somewhat questionable. To resolve this issue, a 230-A-thick Bi layer was deposited on the sample at 100 ·C; AES spectra were taken. The low energy Ga and As peaks (escape depth ~ 5 A) were eliminated. The high-energy peaks (escape tained for all coverages investigated (1/4-100 ML). The LEED pattern showed again a transition from the ( 4 X 6) to the (2 Xl) at a coverage of 2 ML, then turned to 1 X 1 at a coverage of 4 ML. The attenuation of the AES peak intensities at l00·C show that the deposited Bi tends to form islands, though the degree of islanding is slightly smaller than . ~ !"" ., .,c (/)~ w ~ II wol': esca[.e deptll ~ . ol~ c (/) -16A. o "'0 .,~ = N .. II -2A escape dep: II II .... 0 -".... .1 , 0 30 20 Bi coverage (A) 0 0c> " CO -.... 40 !"" ~ . -2A escape depth ol': -6A escape depth - ". !:! .. 0 c> oc> "'0 ,~ - ~ Q. II Q. ..,... Il II It> It> ii" " 2 4 6 81 coverage (A) ~ 10 c: (/) II We ol"'O~ as iiiDN -E " • - ~ 0> c II III IsA ",cape depth • II C\I II !:! C l"C iii DE ~ 0> C II - escape depth CD N N .... II 40 20 " 60 =-. 80 20 -1: BI coverage (A) 40 60 80 BI cov.rage (A) . ~ ;;; (/) C W II C W! IJ) c:] _ !~ C\I escape depth • ol': "'0 II .. ~ • !:! .. E - . 0- c:> sA II Q. • ~ 0- escape depth c:> II ..,.... If) " (b) 8 4 6 81 coverage (A) ~ w-" c:': II 2 0 ;;; (/) ~ , -.;- . ol 8 iii c: ~ 40 30 20 '0 w! -1~ (a) 0 (/) c II CE ~ .1 ol 81 coverage (A) OJ W -2A escape depth II CD N N (/) c ~: escape depth eQ. iiiE> "i~ 933 .1 0 2 4 6 Bi coverage (A) 8 10 ;;c: .1 0 2 4 8 6 BI coverage (Al FIG. 2. Normalized Auger peak intensities as a function of Bi deposition (a 1 at ro{)m temperature, (b) at 250°e. (c) at 100 0e. J. Vac. Sci. Techno/. B, Vol. 7, No.4, JullAug 1989 10 934 S. Horng and A. Kahn: Epitaxial growth of Bi on GaAs(100) surfaces . ~ ?: -;; .. IIlC c C II enw c w- C- C li f-~: - escape depth • E> escape depth N N II 60 40 20 BI coverage (A) ~ C 80 . .1~- 0 20 40 60 BI coverage (A) ?: -.; w! l.w Co!: II • l~ -- II'" It II SA 0-- c>II ~o. • ..... 0- escape depth c> III III ii" .1 80 escape depth III C CE ~ ii Cal ~ en C w" - • 0. o .. o .... o ~ • ~ 1~ • =0. ~ 934 M 0 4 2 Icl 6 8 10 -;;- .1 C 0 2 BI cove,..ge (A) 4 6 8 10 BI coverage (A) Figure 2 continued. depth -15-20 A) were found to be extremely small. These AES data indicated therefore that all of the GaAs surface was indeed covered with Bi and that the mean Bi thickness between large islands was considerably larger than the escape depth of the high-energy AES peaks. Thus the LEEO pattern observed at that coverage could only come from the Bi overlayer. Although the spots were sharp, this LEEO pattern was very dim and the symmetry difficult to determine. The weakness of the pattern can be explained by the fact that the Oebye temperature of Bi is very low (119 K). Moreover, though the symmetry of the LEED pattern at the coverage of 230-A Bi deposited at 100 °C is difficult to determine, it is quite different from the sixfold symmetry observed from the sample with post-growth annealing after RT deposition. It is therefore interesting to note that the substrate did affect the structure of the overlayer. For a more careful examination of these patterns, low-temperature experiments will be necessary. IV. SUMMARY .0 30 20 10 ENERGY lOSS 'IV) FIG. 3. EELS spectra taken for deposition of Bi at 250 'c. J. Vac. Sci. Technol. B, Vol. 7, No.4, JullAug 1989 o Bi layers were grown on GaAs( 1(0) surfaces at various temperatures, and were investigated with LEEO, AES, and EELS. The layers grown at room temperature did not exhibit long-range order as revealed by LEEO, and showed a sixfold hexagonal symmetry after a post-growth annealing at 250°C. The attenuation of AES low- and high-energy peak amplitude indicated a quasi-laminar growth. For the growth at high temperature (> 100 °C), the presence of a LEEO pattern demonstrated long-range order with a symmetry different from the hexagonal symmetry. The attenuation of the AES peak intensity indicated island formation. 935 S. Horng and A. Kahn: Epitaxial growth of Bi on GaAs(100) surfaces ACKNOWLEDGMENTS This work was supported by a grant of the National Science Foundation (DMR-870953l). Partial support was also provided by General Electric and General Motors Corporation through the Presidential Young Investigator Award program. IB. A. Tavger and V. Ya. Demikhouski, Sov. Phys. Solid State 5, 469 (1963). 2V. B. Sandmirskii, SOy. Phys. JETP 25, tol (1967). J. Vac. Sci. Technol. B, Vol. 7, No.4, Jul/Aug 1989 935 'B. Y.Jin, H. K. Wong,G. K. Wong,andJ. B. Ketterson, Thin Solid Films 110,29 (1983). 4D. Agassi and T. K. Chu, Appl. Phys. Lett. 51 (26),2227 (1987). 'N. Garcia, Y. H. Kao, and M. Strongin, Phys. Rev. B 5, 2029 (1972). "S. C. Shin, J. B. Ketterson, and J. E. Hilliard, Phys. Rev. B 30, 4099 (1984); S. C. Shin,J. E. Hilliard, and J. B. Ketterson,J. Vac. Sci. Technol. A 2,296 (1984); S. C. Shin, J. E. Hilliard, andJ. B. Ketterson, Thin Solid Films 111, 323 (1984). 7M. Jalochowski, Phys. Status Solidi (a) 95, KS (1986). "F. De1amareand G. E. Rhead, Surf. Sci. 35, 172 (1973); F. De1amare and G. E. Rhead, Surf. Sci. 35,185 (1973). oJ. J. Joyce, J. Anderson, C. Yu, and G. J. Lapeyre, in Proceedings of the 35th AVS National Symposium (Atlanta, 1988). lOR. Ludeke and A. Koma, CRC Crit. Rev. in Sol. State Sci. 5, 259 (1975).