Semiconductor magnetoelectronics and prospects for a spin transistor

Author: Monzon, Franklin Gregory

Year: 1999

Degree: Dissertation (Ph.D.)

Advisor: Roukes, Michael Lee

Committee Member: Unknown, Unknown

Option: Applied Physics

DOI: 10.7907/gccy-3j80

Abstract

The first demonstration of spin-coupled electronic transport, in all-metallic devices, occurred over ten years ago. Although the development of similar ferromagnet/semiconductor structures poses unique difficulties, these devices are exciting because they afford the possibility of constructing active circuit elements based upon the manipulation of spin rather than charge.

In this thesis we clearly delineate the requirements that must be met in order to successfully implement a semiconductor spin transistor. We present extensive 4.2 K measurements of NiFe/InAs quantum well spin devices fabricated both by photolithography and electron beam lithography, both wet-etched and dry-etched. These measurements often exhibit strong magnetoelectronic phenomena, not based on spin transport, that complicate the demonstration of spin injection/detection. We use local Hall voltages to carefully characterize submicron ferromagnetic thin films in order to improve the switching behavior of our ferromagnetic contacts and to minimize stray field effects. We describe theoretical calculations for ballistic spin-coupled transport, along with careful on-chip electrical characterization, that reveal our InAs conduction channels to be adequate spin-preserving wires. We conclude that inefficient spin transfer across the NiFe/InAs interface severely degrades the spin polarization of the injected current. Though we therefore do not obtain a conclusive demonstration of spin transport even in our smallest devices, with injector/detector spacings of less than 1 mm, we make firm suggestions for future spin devices, including expectations for what consitutes a definitive demonstration of spin transport in a high mobility two-dimensional electron gas (2DEG).

From a physics standpoint, such a demonstration is very interesting because control of a semiconductor spin-polarized current offers possibilities both for experiments in spin dynamics without the use of high magnetic fields or optical-polarization schemes, and for the investigation of spin-scattering mechanisms in tunable-density (gated) devices. From an engineering standpoint, these same transistor-like devices are intriguing as novel memories or switches. Despite this, the development of semiconductor spin devices has lagged behind the advance of metallic multilayer (GMR) spin devices. This thesis addresses the difficulties inherent in implementing semiconductor spin devices, with the aim of enabling the successful manufacture of a spin transistor in the near future.

Files