Report

Electrical Manipulation of Magnetization Reversal in a Ferromagnetic Semiconductor

See allHide authors and affiliations

Science  15 Aug 2003:
Vol. 301, Issue 5635, pp. 943-945
DOI: 10.1126/science.1086608

Abstract

We report electrical manipulation of magnetization processes in a ferromagnetic semiconductor, in which low-density carriers are responsible for the ferromagnetic interaction. The coercive force HC at which magnetization reversal occurs can be manipulated by modifying the carrier density through application of electric fields in a gated structure. Electrically assisted magnetization reversal, as well as electrical demagnetization, has been demonstrated through the effect. This electrical manipulation offers a functionality not previously accessible in magnetic materials and may become useful for reversing magnetization of nanoscale bits for ultrahigh-density information storage.

Magnetization reversal is a fundamental process for writing information, or bits, onto magnetic materials used in data storage, and is generally done by applying magnetic fields locally to the magnetic material. In order to realize higher data density per unit area, the magnetic energy density of the material has to be increased to make the nanometer-scale magnetic bits stable against thermal fluctuations, which at its limit pushes the required magnetic fields for writing too high to generate. Manipulation of magnetization reversal by other means has thus become an important challenge for magnetic information storage (15). We show that electrical manipulation of the magnetization processes is possible in a semiconducting ferromagnetic material and demonstrate electrically assisted magnetization reversal, as well as electrical demagnetization.

For the electric field–effect experiments that we present, the ferromagnetic semiconductor Mn-doped InAs [(In,Mn)As] (6) is incorporated as a channel layer in a metal-insulator-semiconductor field-effect transistor (FET) structure. In such a ferromagnetic FET device, application of an external electric field E to a thin p-type (In,Mn)As channel has been shown to modify the ferromagnetic transition temperature TC of the channel layer through the change of hole concentration p (7); because holes are known to mediate the ferromagnetic interaction among Mn localized spins, negative (positive) bias increases (decreases) p and results in an increase (decrease) of TC. The local magnetization M of the magnetic channel layer is probed by the use of the anomalous Hall effect. The magnetic easy axis direction in (In,Mn)As is perpendicular to the sample plane because of the strain in the channel layer (810).

In our devices, the (In,Mn)As layer thickness of the two FETs is 5 nm and 4 nm for sample A and sample B, respectively. The (In,Mn)As layers were grown by molecular beam epitaxy on an InAs-(Al,Ga)Sb buffer layer structure; the structure consists of, from the surface side, 5 nm InAs, 200 to 300 nm (Al0.8Ga0.2)Sb, and 50 nm AlSb on semi-insulating GaAs (001) substrate. The Mn composition of sample A and B is x = 0.063 and 0.033, respectively. The channel layer of the Hall-bar geometry FETs (Fig. 1A) is covered with a 0.9-μm-thick spun-on SiO2 gate insulator and then by a Cr/Au metal gate electrode (11).

Fig. 1.

(A) Hall bar-shaped field effect transistor having a ferromagnetic semiconductor (In,Mn)As channel. To probe the magnetization M of the channel, Hall resistance RHall = VHall/I proportional to the channel magnetization is measured. (B) Temperature dependence of RHall (∞ M) versus magnetic field μ0H curves with square-shaped hysteresis up to temperatures below 50 K in sample A. Sample A has a ferromagnetic transition temperature of 52 K. No electric field is applied (E = 0). Magnetic field sweep rate is 3.7 mT/min.

Figure 1B shows the magnetic field H dependence of the Hall resistance, RHall, of sample A at zero gate bias from 10 to 50 K; RHall is dominated by the anomalous Hall effect, and thus RHall is proportional to M in the temperature range of interest. The coercive force, μ0HC, at 10 K is 44 mT, where μ0 is the permeability of vacuum, and decreases with an increase of temperature until it finally vanishes at ∼50 K, slightly below TC. The observed decrease in the remanence of RHall as temperature increases from 10 to 40 K can be explained by the combination of the temperature dependence of M and the anomalous Hall coefficient. At zero gate electric field, TC, which can be determined from the Arrott plots through the anomalous Hall effect (7), is 52 K for sample A and 38.5 K for sample B. Application of electric fields E of ±1.5 MV/cm results in a ∓1.5 K change of TC for sample A and a ∓2.0 K change for sample B, in accordance with the previous report (7).

Clearly, application of electric field has a marked effect on HC (Fig. 2) for sample A at 40 K. Here, HC is modified by a factor of 5, from 1 mT at E = –1.5 MV/cm to 0.2 mT at E = +1.5 MV/cm, while keeping the square shape of hysteresis loops virtually unchanged. From the gate capacitance, E = ±1.5 MV/cm is calculated to induce a change in hole concentration of ∓2.7 × 1012 cm2. The small change in remanence can be explained by the change of M itself.

Fig. 2.

Electric-field dependence of the magnetic hysteresis curves measured by RHall of sample A at 40 K. Application of E = ±1.5 MV/cm results in a change of coercive force μ0HC by a factor of 5, from 1 mT at –1.5 MV/cm to 0.2 mT at +1.5 MV/cm, without affecting the square shape of the hysteresis. Magnetic field sweep rate is 0.37 mT/min.

Because the magnitude of HC is a function of an external electric field, the magnetization-reversal process can now be electrically assisted. Using sample B at 32 K (Fig. 3A), we first saturated the magnetization of the (In,Mn)As channel layer under E = –1.5 MV/cm by applying a large enough positive magnetic field, and then reduced the field through 0 mT to a small negative magnetic field μ0H0 of –0.2 mT. This is the initial state indicated by point A in the inset, where two RHall-B curves under E = 0 and –1.5 MV/cm are shown. The change of RHall as a function of time, t, is displayed in Fig. 3A. Because the magnitude of μ0H0 = –0.2 mT is smaller than the coercive force μ0HC = –0.5 mT under E = –1.5 MV/cm, the state remains at point A until we switch E to 0 at t = 25 s. There is an absence of relaxation in this 25-s period, which indicates that the state is stable. In response to the switching, the sign of RHall changes from positive to negative, showing that electrical switching triggers magnetization reversal. Because |HC| < |H0| at E = 0 V/cm, the electrical switching brings the state to point B of the inset. Once the magnetization is reversed, the sign of RHall and thus M remains negative and shows only a small variation when E is switched back and forth between 0 and –1.5 MV/cm, as shown by region C and region D (corresponding to point C and D in the inset, respectively). This electrically assisted magnetization reversal (“assisted” because a small negative magnetic field is needed) without changing applied magnetic fields or temperature demonstrates the possibility of a new scheme for Curie point writing, where magnetization reversal is assisted by making the system closer to or making it go beyond its Curie temperature (12).

Fig. 3.

(A) Time evolution of RHall resulting from a sequence of applied electric fields in sample B at 32 K, showing an electrically assisted magnetization reversal. The initial positive RHall at t = 0 is prepared under E = –1.5 MV/cm by first applying a large enough positive magnetic field to saturate the channel magnetization and then reducing the field to μ0H0 = –0.2 mT. This state corresponds to state A on the hysteresis curve under E = –1.5 MV/cm (inset). The sign change of RHall, i.e., magnetization reversal, occurs in response to switch-off of the electric field (E = 0) at t = 25 s, which makes the magnitude of HC smaller than H0. The state is then at state B on the hysteresis curve under E = 0. RHall (and thus M) shows only a small variation upon switching E back and forth between 0 and –1.5 MV/cm, as the state goes back and forth between state C and state D (inset). (B) Electrical demagnetization of the channel layer by electric field. Here, a hysteresis loop is first measured under E = –1.5 MV/cm (closed triangles). Switching E to +1.5 MV/cm eliminates the hysteresis (closed circles); turning E back to –1.5 MV/cm results in an initial magnetization curve (a virgin curve, open triangles), demonstrating an electrical demagnetization.

At a higher temperature range, one expects to be able to completely demagnetize the material solely by applying an external electric field without resorting to conventional techniques such as increasing temperature or using oscillating magnetic fields. With sample B at 33 K (Fig. 3B), the RHall – μ0H hysteresis loop shown in closed triangles under E = –1.5 MV/cm is measured. Then E is set to +1.5 MV/cm, which makes the RHall – μ0H curve a linear one with no hysteresis (closed circles). Switching back E to –1.5 MV/cm, one can obtain an initial curve (i.e., a virgin curve; open triangles), demonstrating an isothermal demagnetization of a ferromagnetic material. The initial magnetization curve suggests that domain nucleation is responsible for the magnetization reversal in (In,Mn)As, as opposed to domain pinning, because the magnetization of the initial curve reaches saturation at a magnetic field much smaller than the coercive field in the hysteresis loop (13). We expect that well-developed magnetic domains similar to the one in a ferromagnetic semiconductor such as (Ga, Mn)As (1416) are present in (In,Mn)As as a result of the long-range nature of carrier-mediated ferromagnetic interaction, even though (In,Mn)As is a dilute magnetic alloy.

Because electrical manipulation of magnetization processes presented here offers functionalities not previously accessible, ferromagnetic semiconductors are expected to play a key role in magnetoelectronics and spintronics (17, 18) once their TC reaches well beyond room temperature. The highest TC reported for (In,Mn)As and (Ga,Mn)As (19) in its thin-film form is 60 K (x = 0.063) (20) and 160 K (x = 0.074) (21), respectively, which is well accounted for by the Zener model of carrier-induced ferromagnetism (9, 10). Because theoretical prediction by the same model shows that TC increases linearly with Mn concentration and p1/3 with hole density, an increase of Mn concentration by a factor of 5 for (In,Mn)As and a factor of 2 for (Ga,Mn)As, without reducing hole concentration, should allow TC to go beyond room temperature. This has not yet been realized experimentally; such a possibility has been hampered so far by the formation of NiAs-type MnAs as a second phase that reduces the Mn concentration in the host and/or by the formation of Mn interstitials (a double donor) that reduce the hole concentration (2224). Other material systems like GaN and ZnO are predicted to show promise in realizing room-temperature carrier-induced ferromagnetism if the physics demonstrated for the model holds for these materials (9, 10).

Supporting Online Material

www.sciencemag.org/cgi/content/full/1086608/DC1

SOM Text

Fig. S1

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article