Defect model of melt-grown GaAs

  • PDF / 1,627,978 Bytes
  • 16 Pages / 595.44 x 841.68 pts Page_size
  • 44 Downloads / 275 Views

DOWNLOAD

REPORT


I. INTRODUCTION In spite of intensive investigation for over a decade the atomic nature of the deep level electron trap, labeled EL2 by Martin et al.,1 which plays a crucial role in highpurity semi-insulating (SI) GaAs, remains elusive. It appears that EL2 incorporates the antisite defect As Qa although it is quite likely that additional structure is involved to account for its apparent metastability. For details on these features of EL2 we recommend the excellent 1984 review by Makram-Ebeid et al.2 We also cite the recent update given by von Bardeleben et al? in a paper describing their own extensive investigations and the theoretical overview of EL2 to be found in the work of Wager and Van Vechten4 in their description of their model of this defect. Various suggestions for the atomic structure of EL2 have been put forward over the years. Van Vechten5 in 1975 suggested that antisite defects and their complexes with vacancies would be abundant in binary semiconductors. Recently he and Wager4 have proposed the complex As Ga VGa VAs to account for the many properties of EL2, including metastability. Based on an analysis of epitaxial data Zou 6 had earlier suggested the same complex for EL2; he and co-workers7 have also used that model to account for the properties of EL2 in strained GaAs. As a result of recent experiments Spaeth8 and von Bardeleben et al? have proposed an As interstitial to be involved in EL2: As Ga As,. Other suggestions have included As aggregates by Ikoma et al.9'10 and As Ga VAs by Walukiewicz et al.n It is well established that EL2 is present in GaAs grown from a melt rich in As and that it is not present in J. Mater. Res. 2 (5), Sep/Oct 1987

http://journals.cambridge.org

a melt too deficient in As. Furthermore, with proper doping by acceptors, GaAs grown from an As-rich melt can be made semi-insulating with a room-temperature electron concentration below 108 cm" 3 . Intentional doping with the deep level acceptor Cr was used in the early years to compensate unwanted shallow donors and to partly compensate the deep donor EL2 to produce SI GaAs. In recent years high-purity SI GaAs has been grown by the liquid-encapsulated Czochralski (LEC) method in which there are few shallow donors and EL2 is apparently compensated by residual shallow carbon acceptors C As . Johnson et al.12 demonstrated this compensation mechanism in a theoretical model that also shows an abrupt Sl-p transition as the As content in the melt is reduced below a critical value, in agreement with experiment. Because of the sensitive balance in the compensation among shallow donors, EL2, and acceptors, a reduction in EL2, caused by growth from a melt too deficient in As, can result in markedly /"-type material. At the other extreme, in more As-rich material, too few acceptors, possibly caused by too little carbon, can result in an increase in electron concentration13'14 although the material may still be SI, due to partial compensation of EL2 by an as yet unidentified double acceptor. Some fine comprehensive data on not-intention