Enhancement of the Electron G-Factor in Silicon - Heterostructures.

The effects of a magnetic field tilted relative to the normal to the plane of conduction on resistivity oscillations in 2-dimensional electron systems were measured in silicon MOS and strained SiGe heterostructures. Since the spin splitting should be proportional to the total magnetic field, while the Landau level splitting should be proportional only to the normal component of the field, the ratio of the two splittings should be a decreasing function of tilt angle cosine. Two special regimes were investigated, where the ratios of the two splittings were even and odd multiples of 1/2. For odd half integers, levels are evenly spaced, and the strongest minima will switch from filling factors n = 4,8,... to n = 2,6,... as the ratio of the splittings increases. When the ratio is an even half integer, levels of opposite spin quantum number from adjacent Landau levels are degenerate, and the associated minima become maxima as the gaps between the levels disappear. The spin splitting is assumed to be proportional to the effective g-factor, g*, and this g* is calculated from the angles at which the phase reversals or level crossings occur. The resulting values of g* for the strained SiGe heterostructures fall on the same curve of g* vs. N _{rm S} as data from MOS specimens. Since the accepted theory for g-factor enhancement, which attributes it to an exchange interaction between adjacent levels with differing populations, depends the level broadening and certain material properties of the Si-SiO_2 system, the agreement between the data from the MOS devices and heterostructures is unexpected, and may indicate that further refinement of the theory is necessary. Observations of the behavior of the minimum at n = 3 as the tilt angle was changed also indicated that the valley splitting, like the spin splitting, is a function of total field, rather than normal field, as is the case for the Landau levels. The data are consistent with the theory of Fang et al. for the valley splitting in Si MOS as a function of field, though the earlier work made no distinction between total and perpendicular field.