LATTICE SYMMETRY OF SILICON AT LOW TEMPERATURES. K. Munakata, Z. Lu, A. Kohno, Y. Soejima, A. Okazaki, Department of Physics, Kyushu University, Fukuoka 812-81, Japan

By means of X-ray diffraction according to the Bond method coupled with a quadruple-crystal monochromator, the lattice spacing of perfect crystals of silicon has been measured. It is found that the lattice symmetry at low temperatures is not cubic. In general, the lattice symmetry slightly distorted from cubic can be identified from the splitting of {hhh} and {h00} . If the symmetry is trigonal, monoclinic or triclinic, the data of {hhh} will be enough to specify the symmetry. It turned out that this is the case for silicon at low temperatures. Measurements were made in a temperature range 20-300 K, and repeated over three or more thermal cycles, on cooling and heating. In the previous experiment, while the temperature dependence of d(440), the lattice spacing of (440), was normal, d(444) showed a variety of temperature dependences depending on specimen and its thermal history. One specimen showed a normal temperature dependence of d(440) and anomalous of d(444), indicating a break of the cubic symmetry. In the present experiment, a specimen showed normal behaviour of d(444) above 120 K. However, below this temperature, i.e. in the region of negative thermal expansion, it showed slightly anomalous behaviour; in this temperature region, a set of four {444} spacings that are equivalent in the cubic lattice were examined simultaneously. While in the first thermal cycle the four values of d{444} were all different, in the third cycle the values were separated into two: one for 444 and three for the other three, namely 444, 444 and 444. The separation of the two sets is marginally significant in terms of the precision of the data. Only a trigonal lattice symmetry is consistent with this. The results suggested an effect of a kind of aging that was also observed in the Pendellösung experiment of silicon at low temperatures.