1.1 Spintronics: Historical Background

The electron was discovered way back in 1897 by J. J. Thomson, and the electron charge was measured with perfection by R. Millikan. The presence of a magnetic moment for electrons was discovered in the year of 1922 [23], and later on it was also established that the electron spin is quantized. Wolfgang Pauli formalized the theory of spin in 1927 by using the basic foundations of quantum mechanics. He pioneered the description of the spin state by introducing the so-called Pauli matrices. A study on spin polarized tunneling on ferromagnetic/insulator/superconducting aluminum junctions was made [24]. This showed the conservation of the spin in electron tunneling, and gave rise to the possibility of spin sensitive tunneling between two ferromagnetic films. Later Mikhail D’yakonov and Vladimir Perel’ predicted the spin Hall effect in 1971 - a spin flow perpendicular to the current flow direction [25]. Subsequently, the first experimental confirmation of the prediction was made [26]. Julliere in 1975 discovered an increase in resistance ( 10 % at 4.2 K), when the magnetic layers in a Fe/Ge/Co stack were switched from the parallel to the anti parallel configuration (the first experiment on tunnel magneto-resistance TMR) [27]. The proposal for the idea that the spin-polarized current can be injected into a semiconductor when a current is passed through a ferromagnet/semiconductor junction was made in 1976 [28]. However, it took until the 1980’s for the process technology to be able to fabricate multi-layered devices with each layer thickness in the range of nanometers. The Physics Nobel Prize in 2007 was given to Peter Grünberg and Albert Fert for the discovery of the giant magnetoresistance (GMR) in ferromagnetic thin film multi-layers [29], which showed the general agreement on the importance of spintronics.

Currently, the advancements in TMR technology [30] led to a switch towards TMR-based read heads. Exploiting the TMR effect for the magnetic field sensing is advantageous due to the much higher resistance of the layer stack (KΩ instead of Ω) and at least one order of magnitude bigger resistance modulation (~300% in comparison to ~5%) [31]. The magnetic random access memory (MRAM) devices are a further important and practical spintronic device class. The initial MRAM devices were based upon the GMR effect, but due to the advances in TMR stacks the MRAM devices transitioned to TMR based structures [32], as produced by Freescale Semiconductor and IBM. Recently, another technique known as spin transfer torque (STT) magnetization switching has received great attention [33343536]. STT-MRAM facilitates the control of the magnetization by entirely eletrical means, thus featuring better scaling capabilities, and requires less switcing energy. Another goal of spintronics has been to envision spin-based logic devices to replace the charge-based logic devices. The STT-MTJ technology has been reported to be attractive for building logic configurations, which combines non-volatile memories and the logic circuits (logic-in-memory architecture [37]) in order to overcome the scaling obstacles of CMOS logics [38391140].

The idea of a spin field-effect transistor, or SpinFET, is to control a spin signal analog to the charge-based transistor via applied voltage instead of a magnetic field. They are attractive candidates as basic for spin-driven integrated circuits. The first spin field effect transistor (SpinFET) was proposed by Datta and Das [41] in 1990, which is composed of two ferromagnetic terminals separated by a non-magnetic material. A good elaboration of the device structure of a spin transistor, operating principle, and performance can be found in [42]. The spin injection in semiconductors by different means is a topic under intense research, and will be highlighted later. However, if one is able to resolve the problem of spin injection into a semiconductor with sufficient accuracy, the next challenge will be the manipulation and control of the spin transport through the conducting channel. This is usually achieved by applying an external magnetic field to rotate the spin, although in principle the presence of spin-orbit coupling (SOC) allows to control spin electronically. Indeed, the SOC in the semiconductor heterostructures can be tailored by voltage gates on the top of the heterostructures, hence allowing to control the spin by voltage [41]. The practical realization of such structures is still a research topic, and hence the proper understanding of the spin-orbit coupling in semiconductors must be further developed.