APPLIED PHYSICS LETTERS
VOLUME 79, NUMBER 9
27 AUGUST 2001
Enhanced As–Sb intermixing of GaSb monolayer superlattices in lowtemperature grown GaAs V. V. Chaldyshev, N. A. Bert, Yu. G. Musikhin, and A. A. Suvorova Ioffe Physical-Technical Institute, St. Petersburg 194021 Russia
V. V. Preobrazhenskii, M. A. Putyato, and B. R. Semyagin Institute of Semiconductor Physics, Novosibirsk 630090 Russia
P. Wernera) and U. Go¨sele Max-Plank-Institut fu¨r Mikrostrukturphysik, Halle/Saale D-05120 Germany
共Received 2 March 2001; accepted for publication 18 June 2001兲 As–Sb compositional intermixing was studied by transmission electron microscopy 共TEM兲 in GaAs films grown by molecular-beam epitaxy at low temperature 共LT兲 and ␦ doped with antimony. The TEM technique was calibrated by imaging the as-grown films with ␦ layers containing various amounts of Sb. The calibration allowed us to deduce the effective As–Sb interdiffusion coefficient from apparent thickness of the Sb ␦ layers in the films subjected to isochronal anneals at 400– 600 °C. The As–Sb intermixing in LT GaAs was found to be much enhanced when compared to conventional material. Its temperature dependence yields a diffusion coefficient of D As–Sb⫽2 ⫻10⫺14 exp(⫺0.62⫾0.15 eV/kt) cm2 s⫺1. Since the kick-out mechanism operating under equilibrium conditions is valid for As–Sb interdiffusion in GaAs, the enhanced intermixing was attributed to an oversaturation of arsenic self-interstitials in the LT GaAs films. The effective activation energy for As–Sb interdiffusion in LT GaAs seems to be reasonably close to the migration enthalpy of As interstitials, whereas their concentration was roughly estimated as 1018 cm⫺3. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1394166兴
Gallium arsenide grown by molecular-beam epitaxy 共MBE兲 at low temperature 共LT GaAs兲 has attracted much attention during the last decade due to excellent semiinsulating properties1 and due to the ultrashort carrier lifetime.2 This material has found a number of device applications.3 The main feature of LT GaAs is related to a high arsenic excess incorporated during low-temperature MBE.4,5 The off-stoichiometric arsenic primarily forms antisite defects, AsGa, of which the concentration is as high as 1020 cm⫺3 and can be precisely evaluated using x-ray diffraction and near-infrared optical absorption.6,7 In spite of the huge number of point defects, perfect InAs/GaAs and AlAs/ GaAs heterostructures with abrupt and flat interfaces can be grown by low-temperature MBE.8 –10 The abrupt interfaces are blurred upon anneals due to compositional intermixing.9,11 This process proceeding on the cation sublattice is mediated by gallium vacancies,12 V Ga . The gallium vacancy concentration in LT GaAs films was found using positron annihilation technique13 to be as high as 1017 – 1018 cm⫺3, i.e., much higher than in a conventional stoichiometric material. The large number of vacancies enhances intermixing in InAs/GaAs11 and AlAs/GaAs14 –16 heterostructures grown at low temperature. No data have been published so far on the intermixing on the anion sublattice of LT GaAs. The main point defects associated with LT GaAs 共AsGa and V Ga兲 cannot contribute to the interdiffusion on the arsenic sublattice. The process should be, however, sensitive to the concentration of arsenic self-interstitials, Asi, because a substitutional-interstitial a兲
Electronic mail:
[email protected]
mechanism governs diffusion on the group V sublattice.17 In contrast to AsGa and V Ga , the Asi concentration in LT GaAs is not documented reliably by any experimental technique and is often assumed to be low.18 In this letter we show that the compositional intermixing on the anion sublattice of LT GaAs ␦ doped with Sb is strongly enhanced when compared to that in conventional stoichiometric material. This phenomenon evidences a high concentration of Asi. The LT GaAs films ␦ doped with Sb were grown by MBE at 200 °C in a dual-chamber ‘‘Katun’’ MBE system on semi-insulating 2 in. GaAs 共001兲 substrates. The growth procedure was described in more detail elsewhere.19 The Sb content in the ␦ layers was varied in a wide range with a maximum corresponding to a nominal layer thickness of 1.5 ML. The distance between Sb ␦ layers was varied from 20 to 80 nm. All the samples were cleaved into four parts, of which one was kept as grown, the others were annealed at 400, 500, and 600 °C for 15 min under As overpressure. Conventional 共TEM兲 and high resolution transmission electron 共HRTEM兲 microscopy were employed to study the microstructure and Sb–As intermixing in the LT ␦-GaSb/GaAs heterostructures. Philips EM420 and JEOL JEM 4000EX instruments operating at 100 and 400 kV, respectively, were exploited. The cross-sectional TEM specimens were prepared either by mechanical dimpling followed by 4 keV Ar ion-beam milling or by cleaving technique. The TEM study of the as-grown LT GaAs films ␦ doped with Sb showed a perfect microstructure with no extended defects or clusters. This is demonstrated by the crosssectional dark-field TEM micrograph presented in Fig. 1. The GaSb ␦ layers are marked by arrows, the numbers
0003-6951/2001/79(9)/1294/3/$18.00 1294 © 2001 American Institute of Physics Downloaded 09 Aug 2004 to 128.165.156.80. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
Appl. Phys. Lett., Vol. 79, No. 9, 27 August 2001
FIG. 1. Cross-sectional dark-field TEM micrograph taken in 002 reflection from as-grown LT GaAs sample with various Sb content in delta layers. Arrows at the right side show the positions of the Sb delta layers, and the numbers nearby give the nominal Sb deposit in ML.
nearby indicate the nominal Sb deposit in units of monolayers. The ␦ layers are seen as thin and straight bright lines indicating planar and abrupt interfaces. They are separated by 50-nm-thick LT GaAs spacers. The ␦-GaSb/GaAs interfaces produced by low-temperature MBE are perfect up to a Sb deposition as high as 1.5 ML. The lowest Sb content, which can be confidently distinguished by the TEM technique used, is 0.02 ML. The apparent thickness of the ␦ layers was found to be 1.1⫾0.1 nm 共i.e., 4 ML兲. This value was determined by the HRTEM study 共appropriate HRTEM images were presented in Ref. 19. Since the apparent thickness of Sb ␦ layers is independent of Sb content and coincides with the apparent thickness of In ␦ layers in LT GaAs,11 it can be attributed to the nanoscale roughness of the low-temperature growth surface decorated by the isovalent impurities. Figure 2 shows dark-field TEM images of Sb ␦ layers after an annealing at 400 °C 共a兲 and 500 °C 共b兲. In addition to the thick bright lines associated with the Sb ␦ layers and marked by arrows, dark spots attached to the ␦ layers are
Chaldyshev et al.
1295
FIG. 3. Arrhenius plot of the effective As–Sb interdiffusion coefficient. The data for low-temperature grown GaAs are compared with those known for conventional stoichiometric material 共Ref. 17兲.
seen in cross-sectional micrographs of the annealed films. This dark spot contrast originates from clusters formed due to precipitation of excess arsenic at the Sb ␦ layers. The clusters are surrounded by local strain fields. Their sizes, density, and microstructure depend on the annealing conditions. The precipitation of excess arsenic in LT GaAs ␦ doped with Sb was discussed in more detail elsewhere.19 The Sb ␦ layers marked by arrows in Fig. 2 are apparently as thick as 8⫾1 ML after annealing at 400 °C and 13⫾1 ML after annealing at 500 °C for 15 min. An increase of the annealing temperature to 600 °C results in further broadening of the Sb-containing layers to 21⫾2 ML. These results indicate a pronounced Sb–As. In order to deduce the effective interdiffusion coefficient from the TEM data, the conventional diffusion equation was solved using the TEM detectivity calibration 共Fig. 1兲 and apparent ␦-layer thickness measurements 共see, for instance, Fig. 2兲 as fitting parameters. The details of the D As-Sb evaluation procedure are described in Ref. 11. The effective As–Sb interdiffusion coefficients calculated from TEM data are shown in Fig. 3 as dark points. The least square fit yields the temperature dependence of D As-Sb as D As-Sb⫽2⫻10⫺14 exp共 ⫺0.62⫾0.15 eV/kT 兲 cm2 s⫺1 .
共1兲
This dependence is shown in Fig. 3 by a solid line and extrapolated to higher temperatures where data on compositional intermixing are available for conventional stoichiometric GaSb/GaAs superlattices.17 One can see that the As–Sb intermixing in the LT GaAs is enhanced by many orders of magnitude when compared to the stoichiometric case. Due to a lower activation energy the phenomenon is especially strong in the range of 400– 600 °C. It diminishes with elevation of the annealing temperature and should disappear at 900–1000 °C. It was recently shown17 that As–Sb and As–P interdiffusion in GaAs are governed by a substitutional-interstitial mechanism, whereas As-vacancy-assisted diffusion had been proposed prior to that. The arsenic vacancy concentration must be very low in the LT GaAs extremely enriched by
FIG. 2. Cross-sectional TEM images of Sb delta layers in the LT GaAs samples annealed at 400 °C 共a兲 and 500 °C 共b兲 共dark-field, 002 reflection兲. Arrows show the thickness of the Sb delta layers. Downloaded 09 Aug 2004 to 128.165.156.80. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
1296
Chaldyshev et al.
Appl. Phys. Lett., Vol. 79, No. 9, 27 August 2001
arsenic. Therefore, the enhanced interdiffusion is obviously inconsistent with the vacancy mechanism. In contrast, a very high concentration of arsenic interstitials can be provided by arsenic excess in LT GaAs. Therefore, the kick-out mechanism, which involves As self-interstitials, should account for the enhanced As–Sb interdiffusion observed. So, the same diffusion mechanism seems to be valid in both conventional and low-temperature-grown GaAs. For this mechanism the effective As–Sb interdiffusion coefficient depends on temperature and concentration of arsenic self-interstitials. The effective interdiffusion coefficients should become equal in conventional and LT GaAs at such temperature, when the Asi concentrations coincide in these materials. This is expected at 900–1000 °C when the extrapolation of D As-Sb for LT GaAs crosses the line representing the diffusivity for the conventional stoichiometric material 共Fig. 3兲, which has much stronger temperature dependence. At this temperature and under As-rich conditions Hurle’s thermodynamic analysis20 gives the equilibrium concentration of arsenic interstitials, which is as high as (3 – 6)⫻1018 cm⫺3. This value can be considered as a rough estimate of the Asi concentration in LT GaAs films. The activation energy of As–Sb interdiffusion in LT GaAs is as low as 0.6 eV, whereas a value as high as 4 eV is characteristic of As–Sb intermixing and As self-diffusion in conventional GaAs over a wide temperature range 共Fig. 3兲. Being governed by the kick-out mechanism under equilibrium conditions, the activation of the diffusion on the anion sublattice of the conventional GaAs requires formation of arsenic interstitials and their further migration. Hence, the activation energy, E a (conv), is the sum of the migration enthalpy, H m , and formation enthalpy, H f . However, the formation enthalpy cannot be applied to describe the arsenic interstitial concentration in LT GaAs, because this material is oversaturated by all the excess-arsenic-related point defects due to specific growth conditions. In this case we should consider E a (LT)⫽H m ⫹H g , where H g is an effective energy originated from the balance of Asi with As clusters and other types of point defects. For instance, the reaction AsGa⫽Asi ⫹VGa should be taken into account. Applying balance consideration, the H g can be neglected, and the experimental value of 0.6⫾0.15 eV may be considered as an estimate of the migration enthalpy for arsenic selfinterstitials. Although some transient effects could be expected, similar to those for Al–Ga intermixing in LT GaAs,16 the estimated value of H m looks rather realistic. It is reasonably close to the value proposed from annealing kinetics.21 Finally, our diffusion results show that not only arsenic antisites and gallium vacancies, but also arsenic interstitials play an important role in atomic processes related to the arsenic excess in LT GaAs. While AsGa defects seem to be the main reservoir of excess arsenic, V Ga and Asi control and enhance atomic transport on the cation and anion sublattices, respectively. The oversaturation by V Ga and Asi remains within a wide temperature range up to 900 °C. In spite of quite different electronic properties, mobilities, and formation energies, ‘‘thermalization’’ of these defects appears to happen at approximately the same temperature.22 This indicates a quasiequilibrium in the whole system of native point defects and nanoscale As clusters. In conclusion, enhanced As–Sb intermixing has been re-
vealed by TEM studies of low-temperature-grown GaAs films ␦ doped with Sb. The effective interdiffusion coefficient was evaluated in a temperature range of 400– 600 °C. Its activation energy was found to be much lower than the value characteristic of conventional stoichiometric GaAs. The phenomenon was attributed to the oversaturation of antisite defects in the LT GaAs films by arsenic interstitials, of which the concentration and migration enthalpy were estimated as (3 – 6)⫻1018 cm⫺3 and 0.6⫾0.15 eV, respectively. This work has been supported by Grant Nos. RFBR 9802-17617 and INTAS 97-30930. The research was carried out under the programs: ‘‘Physics of Solid State Nanostructures’’ and ‘‘Fullerenes and Atomic Clusters’’ by the Ministry of Sciences, Russia. The authors are thankful to A. E. Kunitsyn for the optical measurements and R. V. Zolotareva for the careful TEM specimen preparation.
1
F. W. Smith, A. R. Calawa, C. L. Chen, M. J. Mantra, and L. J. Mahoney, IEEE Electron Device Lett. 9, 77 共1988兲. 2 Z. Liliental-Weber, H. J. Cheng, S. Gupta, J. Whitaker, K. Nichols, and F. W. Smith, J. Electron. Mater. 22, 1465 共1993兲. 3 D. D. Nolte, J. Appl. Phys. 85, 6259 共1999兲. 4 M. Kaminska, Z. Liliental-Weber, E. R. Weber, T. George, J. B. Kortright, F. W. Smith, B. Y. Tsaur, and A. R. Calawa, Appl. Phys. Lett. 54, 1881 共1989兲. 5 N. A. Bert, A. I. Veinger, M. D. Vilisova, S. I. Goloshchapov, I. V. Ivonin, S. V. Kozyrev, A. E. Kunitsyn, L. G. Lavrentieva, D. I. Lubyshev, V. V. Preobrazhenskii, B. R. Semyagin, V. V. Tretyakov, V. V. Chaldyshev, and M. P. Yakubenya, Phys. Solid State 35, 1289 共1993兲. 6 X. Liu, A. Prasad, J. Nishio, E. R. Weber, Z. Liliental-Weber, and W. Walukiewicz, Appl. Phys. Lett. 67, 279 共1995兲. 7 N. A. Bert, V. V. Chaldyshev, A. E. Kunitsyn, Yu. G. Musikhin, N. N. Faleev, V. V. Tretyakov, V. V. Preobrazhenskii, M. A. Putyato, and B. R. Semyagin, Appl. Phys. Lett. 70, 3146 共1997兲. 8 K. Mahalingam, N. Otsuka, M. R. Melloch, and J. M. Woodall, Appl. Phys. Lett. 60, 3253 共1992兲. 9 N. N. Faleev, V. V. Chaldyshev, A. E. Kunitsyn, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, and V. V. Tretyakov, Semiconductors 32, 19 共1998兲. 10 V. V. Chaldyshev, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, N. A. Bert, A. E. Kunitsyn, Y. G. Musikhin, V. V. Tret’yakov, and P. Werner, Semiconductors 32, 1036 共1998兲. 11 N. A. Bert, V. V. Chaldyshev, Y. G. Musikhin, A. A. Suvorova, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, and P. Werner, Appl. Phys. Lett. 74, 1442 共1999兲. 12 T. Tan, U. Goesele, and S. Yu, Crit. Rev. Solid State Mater. Sci. 17, 47 共1991兲. 13 J. Soermer, W. Triftshuser, N. Hozhabri, and K. Alavi, Appl. Phys. Lett. 69, 1867 共1996兲. 14 I. Lahiri, D. D. Nolte, J. C. P. Chang, J. M. Woodal, and M. R. Melloch, Appl. Phys. Lett. 67, 1244 共1995兲. 15 J. S. Tsang, C. P. Lee, S. H. Lee, K. L. Tsai, and H. R. Chen, J. Appl. Phys. 77, 4302 共1995兲. 16 R. Geursen, I. Lahiri, M. Dinu, M. R. Melloch, and D. D. Nolte, Phys. Rev. B 60, 10926 共1999兲. 17 M. Schultz, U. Egger, R. Scholz, O. Breitenstein, U. Goesele, and T. Tan, J. Appl. Phys. 83, 5295 共1998兲. 18 M. Luysberg, H. Sohn, A. Prasad, P. Specht, Z. Liliental-Weber, E. R. Weber, J. Gebauer, and R. Krause-Rehberg, J. Appl. Phys. 83, 561 共1998兲. 19 N. A. Bert, V. V. Chaldyshev, A. A. Suvorova, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, and P. Werner, Appl. Phys. Lett. 74, 1588 共1999兲. 20 D. T. J. Hurle, J. Appl. Phys. 85, 6957 共1999兲. 21 J. C. Bourgoin, K. Khirouni, and M. Stellmacher, Appl. Phys. Lett. 72, 442 共1998兲. 22 V. V. Chaldyshev, N. A. Bert, Yu. G. Musikhin, A. A. Suvorova, V. V. Preobrazhenskii, M. A. Putyato, B. R. Semyagin, and P. Werner, Semiconducting and Insulating Materials Conference, SIMC-2000, Canberra, Australia, 2000.
Downloaded 09 Aug 2004 to 128.165.156.80. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
