2826
IEEE TRANSACTIONS ON MAGNETICS, VOL. 38, NO. 5, SEPTEMBER 2002
Strain Sensors Based on Magnetostrictive GMR/TMR Structures M. Löhndorf, T. A. Duenas, Alfred Ludwig, M. Rührig, J. Wecker, D. Bürgler, P. Grünberg, and Eckhard Quandt
Abstract—This paper investigates magnetic layer structures suitable for devices measuring mechanical responses such as stress, strain and pressure. Results for giant magnetoresistiance (GMR) multilayers and magnetic tunneling junctions (MTJs) intentionally prepared either with magnetostrictive Fe50 Co50 materials or with amorphous Fe-based alloys serving as sensing (or “free”) layers is discussed in view of possible applications. For MTJs prepared ex-situ with an amorphous FeCoSiB, free layer MR ratios of 33% are obtained leading to gauge factors on the order of 300.
GF = (1
)1
Index Terms—Giant magnetoresistance (GMR), magnetic tunneling junctions (MTJ), magnetostriction, sensors.
I. INTRODUCTION
W
HILE THERE has been considerable research devoted to the use of giant magnetoresistance (GMR) [1] and tunnel magnetoresistance (TMR) layered structures for data storage and magnetic field sensing applications, only a few studies have been dedicated to the use of these devices in broader industrial applications such as stress, strain or pressure sensing [2], [3]. Whereas in the former applications nearly zero magnetostrictive materials are chosen as free layers in order to avoid interaction caused by stress or strain, in this study we have intentionally used highly magnetostrictive alloys and amorphous materials as free layers to promote and govern strain sensitivity without a large reduction in the magnetoresistive (MR) ratio. By the inverse magnetostrictive effect, mechanical straining of coupled magnetostrictive–magnetoresistive structures leads to a change of magnetization direction of the free layer and in turn leads to a change in magnetoresistance. Two different kinds of magnetoresistive devices, i.e., trilayer current-in-plane (CIP) GMR structures and magnetic tunneling junctions (MTJs), have been used in order to develop novel strain sensors. Both the trilayer GMR structures as well as the MTJs consist of two ferromagnetic layers. One of the layers is magnetically pinned (“hard” layer) while the second layer is free to rotate (hence, the term “free”) and serves as the sensing layer. The two ferromagnetic layers are separated either by an insulating tunneling barrier as
Manuscript received February 12, 2002; revised May 10, 2002. This work was supported by the German Ministry of Education and Research under Grant 13N7943. M. Löhndorf, T. A. Duenas, A. Ludwig, and E. Quandt are with the Center of Advanced European Studies and Research (CAESAR), 53111 Bonn, Germany (e-mail:
[email protected]). M. Rührig and J. Wecker are with Siemens AG, 91052 Erlangen, Germany (e-mail:
[email protected]). D. Bürgler and P. Grünberg are with IFF, 52452 Jülich, Germany (e-mail:
[email protected]). Digital Object Identifier 10.1109/TMAG.2002.802466.
for the MTJs or in case of the GMR sensor by a nonmagnetic metallic layer. II. EXPERIMENT The GMR structures are prepared by rf- or dc-magnetron sputtering. Silicon (100) wafers 625 m thick are used as substrates and sectioned before deposition to beams of 3 mm 30 mm. The background pressure is below 10 Pa. All layers are sputtered in the absence of an applied external magnetic field. The layer thickness for the GMR sample in this study are nominally Fe Co (4 nm)/Cu (3 nm)/Fe Co (2 nm)/ Fe Mn (15 nm). For more information on sputtering process parameters, see [4]. The MTJ structures are prepared by magnetron sputtering. The hard layer in all samples consists of 8 nm Ir Mn anti-ferromagnetic (AF) layer and a 2.5 nm thick Co–Fe layer. As tunneling barrier a 1.5 nm thick aluminum layer is deposited and oxidized by plasma oxidation. Magnetostrictive Fe Co films and amorphous (Fe Co ) Si B alloys nominally 6-nm thick are used as free layers. The sputtering process for the MTJ with the magnetostrictive Fe Co free layer as well as the MTJ with the (Fe Co ) Si B free layer was executed with a vacuum break after the oxidation of the aluminum layer. The subsequent sputtering of the free layer was performed four days later in a different sputtering system. The magnetic properties of the magnetostrictive thin films have been investigated by vibrating specimen magnetometer (VSM) measurements of films prepared with the same process parameters. In order to study the influence of applied mechanical stress to MTJs and GMR-based strain sensors, a bending apparatus was constructed which allows for magnetic field versus resistance measurements up to 1.8 T and simultaneous delivery of homogenous strain in parts-per-thousand (‰) range. Homogenous straining of the MTJs and CIP-GMR structures is obtained by using the four point bending method [5]. The strain is introduced by the displacement of a so-called pusher block, which contains two ceramic rods (3-mm diameter) spaced 6 mm apart centered between two fixed support rods (3-mm diameter; 18 mm apart). The induced strain is tensile and for small displacements it is considered to remain within the plane (i.e., plain stress condition) of the sample. III. RESULTS AND DISCUSSION The major loop measurement shown in Fig. 1(a) is performed with the direction of the applied tensile stress parallel to the applied magnetic field and to the easy axis of the GMR stack.
0018-9464/02$17.00 © 2002 IEEE
LÖHNDORF et al.: STRAIN SENSORS BASED ON MAGNETOSTRICTIVE GMR/TMR STRUCTURES
(a)
(b) Fig. 1. (a) Resistance versus magnetic field applied in the parallel configuration for the Fe Co /Cu/Fe Co /Fe Mn sensor structure as a function of applied strain. (b) Resistance versus magnetic field applied in the perpendicular configuration for the Fe Co /Cu/Fe Co /Fe Mn sensor structure as a function of applied strain.
The observed MR ratio for the measurement without any applied mechanical stress or strain (dotted black line) is 3.4%. However, the MR ratio almost remains constant, only a slight increase from 3.4% to 3.5% is observed. Furthermore, in the slope and in the coercivity field value of the pinned hard layer only minor changes are obtained, which are attributed to small magnetization direction changes within the hard layer system with increasing applied stress. While the AF Fe–Mn layer leads to an increase in the coercive field of the Fe–Co hard layer, a field offset due to exchange bias at the AF/FM interface is absent since field annealing is not performed during deposition. The coercivity field value of the free layer is on the order of 2 kA/m (25 Oe) for the unstrained case, whereas the coercivity field value increases with increasing level of tensile stress. For a strain of 1‰, the coercive field is increased by a factor of 2 to 3.9 kA/m (49 Oe) (light gray line). The additional energy contribution from the applied stress to the total free energy of the sensing layer lead to a change of the effective anisotropy of magnetostrictive free layer and therefore stabilizes the initial easy axis. For positive magnetostrictive materials such as Fe Co , the magnetization direction will rotate into the direction of the applied stress due to the additional stress-induced anisotropy in the films. Magfor sputtered Fe Co films are in netostriction constants
2827
the range of 70–90 10 [6]. This stress induced anisotropy causes a steeper slope of the resistance (higher squareness with higher remanence) but also an increase in coercivity of the magnetostrictive free layer. Fig. 1(b) shows a measurement of the same GMR structure in the perpendicular configuration (applied stress is perpendicular to the easy axis of the GMR stack and to the applied magnetic field). The observed MR ratio for the measurement without any applied mechanical stress (dotted black line) is 3.4%. However, a large decrease from 3.4% to 2.7% of the MR ratio is observed for an increase in strain of 1‰. Furthermore, a decrease in the slope of the free layer as well as the hard layer is obtained. The additional energy contribution from the applied stress to the total free energy of the sensing layer lead to a change in the effective anisotropy of the magnetostrictive free layer. Since the applied field is perpendicular to the applied stress direction we expect to see a decrease in the slope of the resistance versus applied field measurement. The decrease in the MR ratio is due to an earlier start of magnetization changes within the hard sub-system at lower aPPLIED field values. As a result, we will not see the full antiparallel orientation of free and hard layer. For strain sensor applications, the gauge factor which is deis an important parameter. For MR fined as GF on the order of 1‰ gauge ratios on the order of 3–4% and factors in the range of 30–40 are expected in the parallel configuration [Fig. 1(a)] or 15–20 in the perpendicular configuration [Fig. 1(b)]. However, MTJ devices with MR ratios higher than 40% have been prepared [3]; by using these MTJ as strain sensors, the potential gauge factor increases by a factor of ten. Fig. 2(a) shows the influence of an applied tensile stress upon 20 m sized MTJ prepared with a magnetostrica 20 m tive Co Fe (6 nm) free layer. After plasma oxidation of the tunnel barrier, the vacuum was broken and the subsequent deposition of the magnetostrictive layers was performed in a different sputtering system. The minor loop measurement shown in Fig. 2(a) is performed in the parallel configuration (applied tensile stress parallel to the easy axis of the MTJ and to the applied magnetic field). The dotted black minor loop (resistance vs. applied magnetic field) represents the unstrained state of the MTJ, while the dashed and the solid loops represent 0.33‰ and 0.66‰ of strain, respectively. The TMR ratio is on the order of 20% and almost constant for all three loops. The switching behavior of the magnetostrictive Fe Co free layer shows an offset of 1.5 kA/m (19 Oe) from the zero-magnetic bias field due to Néel-type coupling between the ferromagnetic layers [7]. The behavior is similar to those found in the GMR experiment: a steeper slope; an increase of the coercive field; and a slight increase in TMR ratio obtained for a change of strain from zero to 0.66‰. Fig. 2(b) represents the measurement of a similar MTJ in the parallel configuration, but this time under compressive stress. A decrease in the slope and a narrower hysteresis of the magnetostrictive free layer is observed. These changes are also attributed to the change of the effective anisotropy of the free layer due to the applied compressive stress. For the maximum magnetization rotation of the free layer by 90 due to the applied stress we expect to see a 50% change from the initial TMR ratio. By analyzing the data, we found this 50% change in TMR ratio of 1.1‰ at a bias field of 1.5 kA/m. (17% to 8%) for a
2828
IEEE TRANSACTIONS ON MAGNETICS, VOL. 38, NO. 5, SEPTEMBER 2002
2
(a)
Fig. 3. The influence of an applied tensile stress upon a 20 m 20 m sized MTJ prepared with an amorphous magnetostrictive (Fe Co ) Si B (6 nm) free layer in the perpendicular configuration.
1.2 kA/m (15 Oe) for the Fe Co free layer. By analyzing the data we found a 50% change of the initial TMR ratio (from of 0.55‰. The obtained gauge factor is 30% to 15%) for a on the order of 300 for the amorphous (Fe Co ) Si B MTJ. IV. CONCLUSION
(b)
2
Fig. 2. Influence of an applied stress upon a 20 m 20 m sized magnetic tunneling junction (MTJ) with a magnetostrictive Co Fe (6 nm) free layer in the parallel configuration (applied stress is parallel to the magnetic field and to the easy axis of the MTJ). (a) With applied tensile stress. (b) Change in resistance of the MTJ as a function of applied compressive stress.
Amorphous magnetostrictive Fe-based alloys are possible candidates for strain gauge applications due to their sensitive magnetoelastic response to small strains and stresses [8]. As a result, (Fe Co ) Si B has been chosen as a candidate for this highly sensitive TMR strain sensor development. The typical magnetostriction constant for those class of materials is on the order of 30 10 [9]. A typical measurement of a 20 m 20 m sized MTJ with an amorphous magnetostrictive (Fe Co ) Si B free layer is shown in Fig. 3. The dotted black minor loop (resistance versus applied magnetic field) represents the unstrained state of the MTJ, whereas the dashed and the solid loops are measured at 0.33‰ and 0.55‰ strains respectively. The TMR ratio is on the order of 33% and almost constant for all three loops. As for the measurement of the Fe Co MTJ the direction of the applied tensile stress is perpendicular to the magnetic field and the easy axis of the free layer (perpendicular configuration). For the free layer thickness of 6 nm, an offset of 1.7 kA/m (21 Oe) from zero magnetic bias field is observed as for the previous measurement of the MTJ with a 6 nm Fe Co free layer. The coercive field values as determined from VSM measurement are 0.2 kA/m (2.5 Oe) for the amorphous (Fe Co ) Si B MTJ and
For the development of strain sensors based on the combination of magnetostriction with the magnetoresistance effect, the use of magnetic tunneling junctions is favorable since the MR ratio of trilayer CIP-GMR structures is on only the order of 3%–4% whereas MR ratios of MTJs are 10%–40%. In addition, MTJ with lateral sizes in the nanometer range have been prepared [3] which in turn could lead to a high spatial resolution combined with high strain sensitivity. Future work focuses on the enhancement of the gauge factor as well as on the preparation of MTJ on flexible substrates such as polymers. REFERENCES [1] A. Fert, P. Grünberg, A. Barthélémy, F. Petroff, and W. Zinn, “Layered magnetic structures: Interlayer exchange coupling and giant magnetoresistance,” J. Magn. Magn. Mater., vol. 140–144, pp. 1–8, Feb. 1995. [2] H. J. Mamin, B. A. Gurney, D. R. Wilhoit, and V. S. Speriosu, “High sensitivity spin-valve strain sensors,” Appl. Phys. Lett., vol. 72, pp. 3220–3222, June 1998. [3] S. S. Parkin et al., “Exchange-biased magnetic tunnel junctions and application to nonvolatile magnetic random access memory,” J. Appl. Phys., vol. 85, pp. 5828–5833, Apr. 1999. [4] T. Duenas, A. Sehrbrock, M. Löhndorf, A. Ludwig, J. Wecker, P. Grünberg, and E. Quandt, “Micro-sensor coupling magnetostriction and magnetoresistive phenomena,” J. Magn. Magn. Mater., vol. 242–245, pp. 1132–1135, 2002. [5] L. Baril, B. A. Gurney, D. R. Wilhoit, and V. Speriosu, “Magnetostriction in spin valves,” J. Appl. Phys., vol. 85, pp. 5139–5141, Apr. 1999. [6] M. D. Cooke et al., “The effect of thermal treatment, composition and substrate on the texture and magnetic properties of FeCo thin films,” J. Phys. D: Appl. Phys., vol. 33, pp. 1450–1459, 2000. [7] B. D. Schrag et al., “Néel ‘orange peel’ coupling in magnetic tunneling junction devices,” Appl. Phys. Lett., vol. 77, pp. 2373–2375, Oct. 2000. [8] K.-H. Shin, M. Inoue, and K.-I. Arai, “Strain sensitivity of highly magnetostrictive amorphous films,” J. Appl. Phys., vol. 85, no. 8, pp. 5465–5467, Apr. 1999. [9] M. Wun-Fogle, H. T. Savage, and M. L. Spano, “Enhancement of magnetostrictive effects for sensor applications,” J. Mater. Eng., vol. 11, no. 1, pp. 103–107, 1989.
