Half Metals: the Absolute Magnets

Half Metals: the Absolute Magnets

A major recent development in spintronics has been the introduction of a special class of metals called half-metals (HM). Half metals are defined by a property called spin polarization (SP). Normal metals, such as copper (Cu) or aluminum (Al) have no net spolarization, meaning that half the electrons are spin up and half are spin down at the Fermi Level (EF). Ferromagnets, such as iron (Fe) and cobalt (Co) have a small imbalance of spin-up and spin-down electrons. Half metals, in contrast, produce 100% spin polarization; that is, all the electrons at the Fermi Level in the half metal are either spin up or spin dow. This characteristic has major implications for the efficiency of spintronic devices and increases their potential magnetoresistance by a huge factor. For example, if one were to substitute a half metal such as CrO2 for the ferromagnetic layers in a spin valve GMR device, the anti-parallel configuration would have a theoretically infinitely high resistance due to strong spin-dependent scattering, while the parallel configuration would maintain a minimal scattering and thus a small resistance. Half metals can thereby produce very high magnetoresistance required for the GMR devices. These high magnetoresistance values provide significantly heightened sensitivity in computing devices, specifically the read head in the computer hard drive.

Physicists have predicted several materials to be half metallic, however, the only material that has been definitively classified as a half-metal at this point is chromium dioxide (CrO2). The experimental results for many other candidates including some metal oxides, Heusler alloys, and zinc blends, remain ambiguous partly because these structures are too complex to be consistently and accurately synthesized.

Chromium dioxide is the only binary oxide that is a ferromagnetic metal, which has a tetragonal rutile structure with a tetragonal unit cell (a=b=4.42Å, c=2.92Å). Although widely used in recording applications for 30 years (a particulate recording medium in magnetic tape), its electronic structure was known only recently.  Band structure calculations show there is a large energy gap (> 1eV) for the minority spin band at the Fermi level. In experiments, point contact Andreev reflection (PCAR) yields a spin polarization as high as 98.4% for this material at low temperature.

The CrO2 is a thermodynamically metastable phase in atmosphere. Its surface will spontaneously decompose to Cr2O3, a much more stable phase. Thus it has been difficult to prepare high-quality thin films of this material using conventional techniques under normal growth conditions. Till now, the most reliable way to deposit CrO2 thin film is by using atmospheric chemical vapor deposition (CVD). In this method, CrO2 thin film can be grown on TiO2 substrate (in either polycrystalline or epitaxial form) and the whole deposition process is highly sensitive to the surface of the substrate.

The CVD deposition reactor consists of a quartz tube placed inside a two-zone furnace. Chromium trioxide (CrO3) powder is used as the precursor and is loaded into a quartz
boat in the source zone, whereas the substrates are placed on a tilted susceptor in the reaction zone. Prior to placement in the reactor, the substrates are ultrasonically cleaned in organic solvents and then dipped in a dilute HF bath for a few minutes. They are subsequently rinsed in distilled water and blown dry. Oxygen is used as a carrier gas for the sublimed CrO3 to be transported to the reaction zone, where it decomposes on the substrate to form CrO2 with evolution of O2. The phase purity and morphology of the films is dependent on the substrate, the source temperature, and the oxygen flow rate. We have obtained single-phase films at substrate temperatures of 390–450 °C, with a source temperature of around 280 °C and oxygen flow rate of ~100 sccm.

Figure 4 shows the temperature dependence of the saturated magnetization for a 4000 Å thick (100)-oriented CrO2 film measured along the [100] easy axis direction. A sharp magnetic transition is observed with a Curie temperature of 393 K. The value of the saturation magnetization at low temperature is ~650emu/cc. Figure 5 shows the magnetic hystersis loops at 5 K for an unstrained film, with the applied field in the plane in the [001] and [010] directions. The magnetic easy axis is clearly along [001] (c axis) with close to ideal 100% remanence, while [010] (b axis) is the magnetic hard axis direction exhibiting reversible behavior. As seen in the bottom inset, the magnetization loop along the [001] direction is a square with a relatively low coercive field. From the magnetization data along the hard axis, the anisotropy field is determined to be ~1350 Oe at 5 K. The magnetocrystalline anistropy constant K1 is then calculated to be 4.4 X 10^5 ergs/cc.

There are many methods currently in use for determining the spin polarization of metals. Prominent among these are spin-resolved photoemission, tunnel junctions, and Andreev reflection.

In spin-resolved photoemission, electrons in the half metal are excited by means of a light beam. For the majority spin, the photoemission spectrum shows a conducting Fermi cut-off, whereas for the minority spin, the spectrum  shows an insulating gap. The spin polarizations of the ejected electrons can be measured at different energies of the incident light, however, the most direct measurements of the spin-polarized density of states are obtained near the Fermi energy. Although this method of measurement has been highly useful to experimentalists, it does have a few shortcomings, namely, poor energy resolution and high sensitivity to the nature of the surface being examined (Venkatesan).

A second method for determining the spin-polarization of materials is the tunnel junction. Magnetic tunnel junctions (MTJs) are usually composed of two ferromagnetic layers with an insulator (such as MgO or Al2O3) in between. The magnitude of the tunneling magnetoresistence is related to the spin-polarization of the individual ferromagnetic layers. When the ferromagnetic layers are oriented such that their magnetic moments run parallel to one another, the spin dependent scattering of the carriers is small, and the junction has a low resistance. However, when the magnetic moments run antiparallel to one another, the exact opposite effect occurs: the spin-dependent scattering of the carriers is much larger and the resistance across the junction is at a high. The difference between these two resistances can provide insight into the spin-polarization of the ferromagnetic material.

Andreev reflection is widely considered the most reliable method for measuring the spin-polarization of half metals. This process occurs when a current passes between a superconductor and a normal metal at a point contact. For an electron to move from a normal metal to a superconductor, it must first form a Cooper pair—a pair of electrons bound together by the very small attraction between electrons, which will achieve an energy lower than the Fermi energy. This is accomplished by simultaneously injecting spin-up and spin-down electrons into the electron hole. In this manner, the observed current is doubled in a normal metal. In a half metal, however, Cooper pairs cannot form because these pairs consist of electrons of opposite spins. Since there are no states available, the current will drop to zero (Venkatesan). Using this method, the spin-polarization of various materials can be deduced. In point contact Andreev reflection, a superconducting point contact tip such as Sn, Nb, or Pb is pressed against the material in question. This method extremely utile and has verified the half-metallicity of CrO2.


Half Metals and Their Properties



Curie Point (K)


Spin Polarization





515 emu/cm3
at 300K


Fig. (a)



390 emu/cm3 at 300 K


Fig. (b)




98.7 emu/cm3
at 300K

99% (photoemission)
58-92% (Andreev)

Fig. (c)

Double Perovskites



390 emu/cm3
at 300K

90% (MTJ)

Fig. (d)

Zinc Blendes



1000 emu/cm3

99% (photoemission)




720 emu/cm3


Fig. (e)




320 emu/cm3 at 300K

(tunnel junction)

Fig. (f)