ABSTRACT
In spite of these exciting recent advances, a number of results for the H2 molecules led to apparently contradictory conclusions. The Raman band of the H2 molecule in GaAs is split into two components, 8 cm-1 apart, with an intensity ratio of ~ 3:1 [5]. These lines were assigned to ortho and para H2, whose frequencies differ because of ro-vibrational coupling. This interpretation leads naturally to the conclusion that H2, sitting at a Td interstitial site in GaAs, is freely rotating. In contrast to the situation in GaAs, the H2 molecule in Si gives only a single, sharp, H2-vibrational line at 3618.4 cm-1 and no evidence for an ortho-para splitting in its IR absorption spectrum [6,12]. To explain the absence of an ortho-para splitting, it was suggested that there must be a barrier that prevents rotation of the molecule [12]. Uniaxial stress results for the 3618.4 cm-1 line of interstitial H2 in Si were interpreted in terms of an orientationally degenerate defect with low symmetry, reinforcing the suggest ion that the H2 molecule is static [14]. However, several theoretical calculations for H2 at a tetrahedral interstitial site in Si find that <100>, <111>, and <110> orientations have similar energies, making it surprising that the H2 molecule does not rotate [9-12]. Furthermore, recent molecular dynamics calculations indicate that the H2, HD and D2 molecules in Si behave as nearly free rotators, bouncing within the interstitial region [13]. This conclusion is supported by our own model calculations [15] which suggest a rotational barrier of only
~ 0.01 eV, more than an order of magnitude smaller than argued [12] would be needed for consistency with a static model.
