Tuning the Metal-Insulator Transition in N-Type Silicon with a Magnetic Field

Abstract

Since the classic calculation of Yafet, Keyes, and Adams1 demonstrating the shrinkage of shallow donor wave functions in large static magnetic fields, it has been widely recognized that a magnetic field could be used to tune a metallic sample through the Metal-Insulator (MI) transition. Several experimental groups2–5 have successfully tuned n-type InSb through the MI transition with modest magnetic fields. For InSb it is easy to bring the magnetic length λ=(hc/eH)1/2 to a value much less than the donor Bohr radius (aD *~600A) with reasonable fields since λ=81A (10/H(T.))1/2. One thereby readily achieves the strong field limit (λ<<aD*) for InSb. For n-type Si and Ge the donor Bohr radii are very much smaller (aD*<20A for Si and aD*<47A for Ge) and one remains in the weak field limit (intermediate regime for Ge) for the largest static laboratory fields (H~30T.) available. As a result there have been no reported successful efforts in tuning the MI transition in Si with a magnetic field to compare with the remarkably successful tuning of the critical density Nc for Si :P by Paalanen et al.6 utilizing uniaxial stress. Below some new experimental evidence is presented, for both barely insulating and barely metallic samples, which can be interpreted in terms of tuning of Nc by a large static magnetic field.