Gross R., Sidorenko A., Tagirov L. Nanoscale Devices

Подождите немного. Документ загружается.

Molecular Detection with Magnetic Labels and Magnetoresistive Sensors 43

GMR type biosensors are a good starting point, but they do not represent

the most sensitive magnetoresistive device possible. Higher signals can be

obtained by tunneling magnetoresistance (TMR) type sensors [12], which

show a much larger resistance variation over a smaller field range.

Therefore, we are also developing a biosensor based on tunneling

magnetoresistance. Just like our GMR biosensors, it is designed to be

compatible with DNA microarray type applications. Thus, the size of

individual sensor elements is chosen to be comparable to typical DNA spot

diameters (around 100 µm), so that each sequence is detected by one sensor

element.

A SEM image of a sensor element of our TMR type biosensor is shown

as an inset to Fig. 6. One of its ferromagnetic electrodes (3 nm of Co

)

is magnetically hard due to direct exchange interaction with an underlying

15 nm thick antiferromagnetic Mn

film, while the magnetization of the

electrode on the opposite side of the 1.6 nm thick Al

tunneling barrier is

free to rotate (8 nm of Ni

). Due to the unidirectional exchange

interaction, the in-plane sensor characteristic depends upon the direction of

the applied field, which is apparent from the measurements displayed in

Fig. 6. Only when the field is applied parallel to the pinning direction, an

antiparallel magnetization configuration is achieved and the full TMR

amplitude is reached.

Fig. 6. Layout and magnetoresistance response of a TMR-type sensor element.

TMR Type Biosensors

44 J. Schotter, M. Panhorst, M. Brzeska, et al.

Fig. 7. Signal comparison of GMR- and TMR-type biosensors for similar marker

coverage.

Since the stray fields of the magnetic markers are radially symmetric in

our standard measurement setup, the actual sensor response to magnetic

labels can be regarded as the average of its characteristics parallel and

perpendicular to the pinning direction.

The TMR-type biosensor is passivated by the same protection layer as

described above. Afterwards, magnetic labels are directly spotted on top of

individual sensor elements in varying concentrations, and, similarly to the

method described for GMR-type biosensors, their signals are taken in

dependence of a perpendicular magnetizing field. For maximum sensitivity,

an additional in-plane bias field is applied in order to set the operational

point of the sensor close to the switching field of the free magnetic layer.

Figure 7 shows a direct comparison of the resulting signals both for

GMR- and TMR-type biosensors. In each case, the respective sensor

elements are covered to about 6% by Bangs magnetic microspheres with a

mean diameter of 0.86 µm. The reference line displays the signals obtained

for a pair of uncovered sensor elements, clearly indicating the absence of

any response to the out-of-plane magnetizing field. The measurements for

label-covered sensor elements, however, clearly reveal the increasing effect

of the marker’s stray fields onto the sensors’ magnetization configurations

with rising magnetizing field. In principle, the data of both sensor types

share the same features, but the response of the TMR sensor is larger by a

Molecular Detection with Magnetic Labels and Magnetoresistive Sensors 45

factor of 3.5 in this case, which reflects its improved sensitivity to in-plane

fields.

In addition to being more sensitive, TMR type biosensors could also be

employed for single molecule detection, which is an integral part of any

science investigating the physics and applications of such entities. In this

case, the detection of single molecules translates into the detection of single

magnetic labels, which is quite possible provided the size of the

magnetoresistive sensor is comparable to the magnetic label [13]. Because

the sensor elements are easily scaleable to the required dimensions, TMR-

based magnetoresistive biosensors are especially fit for that purpose as they

also offer the unique possibility of on-chip manipulation of biomolecules

by magnetic gradient fields applied to their labels. Currently, we are

working on the integration of on-chip manipulation and TMR-based

detection of single molecules [8].

Conclusion

We have demonstrated the feasibility of GMR-based biosensors for

detecting specific complex DNA sequences down to a concentration of 24

pM. At these low concentrations, the GMR-type biosensor has proven to be

more sensitive than a comparable standard fluorescent DNA-detection

method. Due to the direct availability of an electronic signal and the small

size of the required instrumentation, magnetoresistive biosensors are a

promising choice for the detection unit of future integrated lab-on-a-chip

systems.

Furthermore, TMR-based biosensors with superior signal amplitudes

have been fabricated for the detection of magnetic labels. As these systems

are easily scalable, they could be employed in the regime of single

molecule detection. This is especially true when adding on-chip

manipulation of single molecules by magnetic gradient fields applied to

their labels.

Acknowledgments

This work was supported by the German ministry of education and

research (BMBF) under grant number 13N7859 and by the

Sonderforschungsbereich 613.

46 J. Schotter, M. Panhorst, M. Brzeska, et al.

References

1. Kricka LJ (2001). Clinica Chimica Acta 307:219

2. Liu RH, Yang J, Lenigk R, Bonanno J, Grodzinski P (2004). Anal Chem

76:1824

3. Thorsen T, Maerkl SJ, Quake SR (2002). Science 298:580

4. Heller MJ (2002). Annu Rev Biomed Eng 4:129

5. Umek RM, Lin SW, Vielmetter J, Terbrueggen RH, Irvine B, Yu CJ, Kayyem

JF, Yowanto H, Blackburn GF, Farkas DH, Chen YP (2001). Journal of

Molecular Diagnostics 3:74-84

6. Baselt DR, Lee GU, Natesan M, Metzger SW, Sheehan PE, Colton RJ (1998).

Biosensors and Bioelectronics 13:731

7. Rife JC, Miller MM, Sheehan PE, Tamanaha CR, Tondra M, Whitman LJ

(2003). Sensors and Actuators A 107:209

8. Brzeska M, Panhorst M, Kamp PB, Schotter J, Reiss G, Puehler A, Becker A,

Brueckl H (2004). Journal of Biotechnology 112:25

9. Panhorst M, Kamp PB, Reiss G, Brueckl H (2005). Biosensors and

Bioelectronics 20:1685

10. Graham DL, Ferreira HA, Freitas PP (2004). Trends in Biotechnology 22:455

11. Schotter J, Kamp PB, Becker A, Puehler A, Reiss G, Brueckl H (2004).

Biosensors and Bioelectronics 19:1149

12. Gregg JF, Petej I, Jouguelet E, Dennis C (2002). J Phys D: Appl Phys 35 :R121

13. Tondra M, Porter M, Lipert RJ (2000). J Vac Sci Technol A 18:1125

WEAK MAGNETIC FIELDS DETECTION

TECHNIQUES AND DEVICES

Magnetic Tunnel Junctions Based on Half-Metallic

Oxides

Rudolf Gross

Walther-Meißner-Institut, Bayerische Akademie der Wissenschaften

and

Physik-Department, Technische Universität München

Walther-Meißner Str. 8, D-85748 Garching, Germany

Abstract: Magnetic tunnel junctions (MTJs) are key elements of spintronic

devices with widespread applications in magnetic data storage and

magnetic sensors. They are based on magnetic multilayer structures

and their optimization for applications requires the nano-engineering

of the interfaces in these multilayers. The key figure of merit of MTJs

is the magnitude of the achievable tunneling magneto-resistance

(TMR). A promising way for improving TMR values is the use of

half-metallic ferromagnets, that is, ferromagnets in which only states

of one spin direction are present at the Fermi level. We review the

history and present understanding of spin-polarized tunneling in

MTJs and their improvement by using half-metallic ferromagnets.

Doing so, we give a classification of various types of half-metallic

materials. Furthermore, we address the various existing definitions of

the quantity spin polarization and the experimental methods for

measuring it. Promising half-metallic materials are ferromagnetic

oxides. We review the physical properties and present understanding

of the most prominent representatives of this materials class and give

an overview on recent attempts to fabricate MTJs with high TMR

values from these materials. Here, we particularly focus on problems

related to the nano-engineering of interfaces.

Keywords: magnetic tunnel junctions, spin polarized tunneling, spintronics, half-

metallic oxides, spin polarization, nano-engineering of interfaces,

tunneling magnetoresistance

Electronic mail: Rudolf.Gross@wmi.badw.de

R. Gross et al. (eds.), Nanoscale Devices - Fundamentals and Applications, 49–110.

50 Rudolf Gross

Introduction

The field of magneto- or spin-electronics has attracted increasing interest

over the last decade. This was stimulated both by the interesting

fundamental physics issues associated with the transport, manipulation and

detection of spins in magnetic and non-magnetic materials and the

increasing number of commercial applications of spintronic devices. For

example, the high storage capacity of today's hard-disk drives would have

been impossible without the development of read heads based on magneto-

electronic concepts. Moreover, at present the feasibility of the Magnetic

Random Access Memory (MRAM) is investigated vigorously by several

companies. For future applications more complicated three-terminal devices

such as spin transistors are envisioned. The common idea behind all these

activities is to exploit both the charge and spin degree of freedom of

electrons to arrive at spintronic devices with improved or novel properties

that may be able to satisfy the continuously growing demands to devices

used in our communication and information technology. Most spintronic

devices are based on multi-component materials systems consisting of

magnetic and non-magnetic materials. Their operation usually strongly

depends on the properties of the interfaces in these structures. Therefore,

improving the quality and functionality of spintronic devices always

requires not only a solid understanding of the underlying physics but also of

the involved materials and even more important a controlled nano-

engineering of interfaces. Here, we are discussing the relevant physics,

materials and nano-engineering issues related to magnetic tunnel junctions

with particular emphasis on the use of half-metallic ferromagnets.

Spintronic devices already have and will most likely have their largest

application potential in magnetic storage and sensors (for recent reviews see

[1-11]. For example, for our today's computer industry there is great interest

in the possibility of fabricating a non-volatile random access memory

which retains its information even after removing power from the device –

an ideal memory. The new concept of a nonvolatile Magnetic Random

Access Memory (MRAM) has been proposed and will possibly

revolutionize semiconductor memory and other spintronic devices such as

programmable logic elements in the near future [2, 12-14]. The basic

elements for MRAMs [15, 16] as well as many other spintronic devices are

micron-sized magnetic tunnel junctions (MTJs), which consist of two

ferromagnetic (FM) electrodes sandwiching a thin insulating (I) barrier.

Since the discovery of a large tunneling magnetoresistance (TMR) at room

temperature in MTJ devices in the early 90's, [17-21] this research field has

Magnetic Tunnel Junctions Based on Half-Metallic Oxides 51

become very active. Moreover, the various physical phenomena which

govern the operation of these magnetoresistive devices and the need for

suitable materials made the field of magnetic tunnel junctions very

attractive both from the basic physics and materials point of view. This

stimulated a tremendous research activity in experimental and theoretical

physics as well as materials science aiming at the thorough understanding

of the electronic, magnetic and magnetotransport properties of MTJs.

We emphasize that one of the keys to a successful application of

spintronic devices is the ability to control the magnetization direction in

ferromagnetic materials. Evidently, this can be realized by applying small

magnetic fields. However, this is not an ideal solution for submicron-sized

devices due to the large currents and space required for the generation of

the control fields. An interesting solution has been proposed by Berger [22]

and Slonczewski [23] called spin-transfer torque. This phenomenon, where

the flow of a spin-polarized current is transferring angular momentum to a

ferromagnet and changes the orientation of its magnetization has been

studied both theoretically and experimentally [24-30]. In ferromagnetic

semiconductors additional control is possible both by optical means [31-33]

and by applying a gate voltage [34-35]. With the development of so-called

multiferroic materials it even may be possible to switch the magnetization

direction by applying electrical fields.

Tunneling always has played an important role in understanding spin

effects in electrical transport. First experiments on spin-dependent transport

have been made using normal metal/ferromagnet/normal metal (N/FM/N)

type junctions based on the ferromagnetic semiconductor EuO [36, 37].

When an unpolarized current passes the ferromagnetic semiconductor the

current was found to become spin-polarized [38, 39]. In the early 1970s,

spin polarized tunneling was studied by Tedrow and Meservey in a series of

experiments on ferromagnet/insulator/ superconductor (FM/I/S) type

junctions [40-44] followed by the first tunneling experiments on

ferromagnet/insulator/ferromagnet (FM/I/FM) type junctions by Jullière in

1975 [45].

A magnetic tunnel junction (MTJ) consists of two ferromagnetic

electrodes separated by a thin insulating barrier allowing for spin polarized

tunneling between the junction electrodes. The key feature of a MTJ is the

fact that the tunneling resistance depends on the relative orientation of the

magnetization in the junction electrodes, which can be changed by an

external magnetic field. That is, we observe different resistance values R

and R

(or conductance values G

and G

) for a parallel and anti-parallel

History and foundations of magnetic tunnel junctions

52 Rudolf Gross

magnetization orientation. This phenomenon is called tunneling

magnetoresistance (TMR) or sometimes junction magnetoresistance (JMR)

with the TMR or JMR effect usually defined as

app

pap

app

TMR

−−−

−≡ ==

(1)

.==

app

pap

app

JMR

−

−≡

(2)

Of course, TMR in MTJs is just a manifestation of a magnetoresistance that

yields a change in electrical resistance in the presence of an applied

magnetic field. Historically, the first magnetoresistance effect (beyond the

usual positive magnetoresistance observed for every normal metal) was the

anisotropic magnetoresistance in bulk ferromagnets dating back to

experiments of Lord Kelvin in the 19th century [46].

Although TMR is known since the early experiments of Jullière [45] and

Maekawa et al. [47] about 30 years ago, the interest in TMR initially was

modest due to the fact that the TMR values obtained in the first experiments

were small (only a few percent) and/or could be observed only at low

temperatures [48-52]. Certainly, this was related to the difficulties related to

the controllable and reproducible fabrication of MTJs. However, with the

strong improvement of magnetic thin film heterostructures triggered by the

discovery of the giant magnetoresistance effect [53-55], Miyazaki and

Tezuka [18] as well as Moodera et al. [17] developed reproducible

fabrication processes for MTJs employing smooth and pinhole-free Al

tunneling barriers. For such MTJs for the first time reproducible TMR

values above 10% could be observed at room temperature. This stimulated

an enormous research activity resulting in a steady improvement of TMR

values. Today, for MTJs based on various transition metal alloys TMR

values above 50% have been achieved at room temperature [56-60]. Very

recently, for Fe/MgO/Fe MTJs with epitaxial or highly oriented MgO

barriers TMR values of even above 200% at room temperature have been

obtained [61-65] in agreement with theoretical predictions [66, 67].

The physics behind the TMR is spin-dependent tunneling. The tunneling

probability and hence the tunneling resistance depends on the relative

orientation of the magnetization of the two magnetic layers. The reason for

that is simple. Whereas in non-magnetic metallic materials electrons at the

Fermi level with opposite spin direction (we denote them spin-up and spin-

down in the following) have the same Fermi wave vector and subsequently

Magnetic Tunnel Junctions Based on Half-Metallic Oxides 53

the same tunneling probability for spin-up and spin-down electrons, this is

no longer the case for ferromagnetic materials. Here, due to the exchange

splitting, we usually have different Fermi wavevectors for spin-up and spin-

down electrons and in turn different tunneling probabilities. Moreover, due

to the exchange splitting also the density of states (DOS) giving the density

of occupied and open states is usually different for the spin-up and spin-

down electrons. Since the tunneling current is proportional to the product of

the occupied states in one electrode and the open states in the second times

the tunneling probability, it is evident that the tunneling current is spin

dependent. The spin dependent tunneling of electrons has been discovered

by Tedrow and Meservey [40-43]. Using tunnel junctions with a

ferromagnetic and a superconducting Al counter-electrode separated by an

tunneling barrier, they could use the superconducting electrode to

detect the difference of the tunneling current of the spin-up and spin-down

electrons (for a review see [44]).

Model of Jullière

Already in 1975, Jullière developed a very simple model relating the TMR

value of MTJs to the spin polarization P

and P

of the two ferromagnetic

junction electrodes. This model, which is illustrated in Fig. 1, is based on

two simplifying assumptions. First, it is assumed that the spin of the

electrons is conserved during tunneling. Then, the tunneling of spin-up and

spin-down electrons could be analyzed within a two conduction channel

model, where electrons originating from one spin state in the one

ferromagnetic junction electrode can tunnel only into empty states of the

same spin direction in the other junction electrode. This two-spin channel

model goes back to the pioneering work of Mott [68, 69] and was extended

later by Campbell [70] and Fert [71, 72]. Secondly, it was assumed that the

conductance for a particular spin orientation only depends on the product of

the effective DOS in the junction electrodes. That is, the tunneling

probability was assumed to be the same for both spin directions. With these

assumption the TMR or JMR could be expressed in terms of the spin

polarization in the junction electrodes as

Theoretical models

Gross R., Sidorenko A., Tagirov L. Nanoscale Devices - Fundamentals and Applications

Подождите немного. Документ загружается.