Wrong Bonds at Compound Semiconductor Grain Boundaries

  • PDF / 467,180 Bytes
  • 6 Pages / 420.48 x 639 pts Page_size
  • 40 Downloads / 244 Views

DOWNLOAD

REPORT


WRONG BONDS AT COMPOUND SEMICONDUCTOR GRAIN BOUNDARIES W. R. L. LAMBRECHT, C. H. LEE ', M. METHFESSEL,t M. VAN SCHILFGAARDE,t C. AMADOR* AND B. SEGALL*. * Department of Physics, Case Western Reserve University, Cleveland, OH 44106

t

Fritz-Haber-Institut der MPG, D-1000 Berlin 33, Faradayweg 4-6, Germany. : SRI-International, 333 Ravenswood Av., Menlo Park, CA 94025

ABSTRACT The role of cation-cation and anion-anion "wrong" bonds at compound semiconductor grain boundaries is investigated. Results are presented for the {110} and {001} inversion domain boundaries in SiC and GaAs. The role of electrostatic effects, compensation of over- and undersaturated bonds, "chemical" reconstructions and bond length relaxations are discussed. 1. INTRODUCTION The reconstructions of many grain boundaries in elemental semiconductors involve 5and 7-fold rings. [1] A new feature of grain boundaries in compound semiconductors is that these odd-numbered rings necessarily contain cation-cation and/or anion-anion bonds. Kohyama et al. [21 have shown that these "wrong" bonds may imply a considerable energy cost. They found that the (122) E = 9 boundary in SiC has a reconstruction similar to that of the same boundary in Si but has a four times larger energy per unit area. We have previously studied the {110} inversion domain boundary in SiC [3,41 and found an even higher energy for this boundary which only contains "wrong" bonds. Several questions arise at this point. Is there a definite energy associated with each type of wrong bond? Is the energy of a boundary given by the sum of the energies of the wrong bonds of each type? Is the idea of energetic "wrong"bonds generally applicable to compound semiconductors? Intuitively, one reason for the high energy of wrong bonds is the electrostatic repulsion between like ions placed in nearest neighbor positions. It is well known that SiC is fairly ionic in spite of being a IV-IV compound. The reason for its ionicity is that the C valencelevels lie much deeper than the Si-valence levels and thus attract more electrons. For non-isovalent semiconductors, another effect possibly plays a role. Cation-cation bonds are undersaturated and anion-anion bonds are oversaturated. Thus some bonding states may remain empty and some anti-bonding states may become filled respectively. Both effects would weaken the bonding. On the other hand, they would lead to a metallic character of the boundary. One might question whether these effects would in fact occur, or, whether the wrong bonds of the two types will tend to locally compensate each other by appropriate reconstructions. We have started to investigate these questions for the simplest type of grain boundaries, the inversion domain boundaries (IDB). These are boundaries between domains with inverted cation and anion positions. IDBs often occur as a result of growth on stepped {001) surfaces of an elemental semiconductor. They have been extensively investigated in 3C-SiC by Cheng et al. [5] and in GaAs by Ueda et al. [6]. We present results for the (0011 and {