Magnetoresistive performances in exchange-biased spin valves and their roles in low-field magnetic sensing applications

n g H G ence Cau Giay, Ha Noi 100000, Viet Nam tion (UT a r t i c l e i n f o This is an open access article under the CC BY license ( fabrication techniques of GMR are still useful [5,16e18]. The per- formance of GMR is very sensitive to the fabricating conditions and the specific equipment [19]. This can also be found in various esides, numerous icular equipment, n of GMR metallic 5], the ion beam ion [31], the elec- nges of the GMR etween the mag- ent used. Further- directly determine the architecture and the features of the mag- netic sensors. For example, the exchange bias field (Heb) expresses the strength of the external field making a saturation, where the sensor is inactive with any external magnetic fields. The intercou- pling field (Hin) is induced by the coupling between a pinned layer (PL) and a free layer (FL). The Hin can cause a shift of the operating point of theMR curve. TheHin is mainly caused by the thickness and the roughness of the spacer layer, and by the stray field of the PL as * Corresponding author. E-mail address: sulv@itims.edu.vn (V.S. Luong). Contents lists availab Journal of Science: Advance .e l Journal of Science: Advanced Materials and Devices 3 (2018) 399e405Peer review under responsibility of Vietnam National University, Hanoi.effect have been reported. Nevertheless, extensive studies on the more, the magnetoresistive properties are the factors which1. Introduction Giant magnetoresistance (GMR) devices are widely used in various applications [1e4]. The most important application of GMR is in the data storage area [5e10]. GMR based sensors have also found wide applications in the automotive markets, navigation, aerospace [11e13], and electronic compass devices [14,15]. In the last two decades, numerous comprehensive studies on the GMR articles on the GMR spin valve effect [20e24]. B works have discussed in detail about the part methods and technical conditions for fabricatio multilayers, such as the sputtering method [2 deposition [26e30], the chemical vapor deposit trodeposition [32e35], etc. However, the challe fabrication still remain owing to the interplay b netoresistive properties and the specific equipm© 2018 The Authors. Publishing services by Elsevier B.V. on behalf of Vietnam National University, Hanoi.Article history: Received 18 May 2018 Received in revised form 2 August 2018 Accepted 14 September 2018 Available online 19 September 2018 Keywords: GMR Giant magnetoresistance Magnetic sensors RF sputtering Spin valvehttps://doi.org/10.1016/j.jsamd.2018.09.004 2468-2179/© 2018 The Authors. Publishing services b ( b s t r a c t The magnetoresistive properties of pinned spin valves (SV) and their roles in low-field sensing applications were characterized. The magnetoresistive parameters were extracted, including the exchange bias (Heb) field as a function of the iron content in the CoFe layer and the antiferromagnetic (AFM) thickness, the magnetoresistance (MR) ratio versus the spacer thickness, the coercivity (Hc) as a function of the seed layer, and the composite layer [NiFe/Co] used. These parameters are crucial in determining the features of the magnetic sensors. Eventually, the selected SV film structure of (Si/ SiO2)/Ta(50 Å)/[NiFe(30 Å)/Co(15 Å)]/Cu(24 Å)/Co80Fe20(25 Å)/IrMn(100 Å)/Ta(50 Å) was found sig- nificant, and the SV elements were patterned using the lithographic lift-off method with the active cell dimensions of 2 mm
150 mm. To define a pinning axis, a cool-field anneal was applied at 250 C for 30 min in a magnetic field of 2 kOe. A Wheatstone half bridge was engineered using two SV elements and two external resistors. The operation point of the sensor was well tuned using a tiny permanent magnet. A sensitivity of 5 V/T was observed with a linear range of ±2 mT. To demonstrate the performance of the designed sensor, a measurement of the Earth magnetic field was carried out. The engineered SV sensor finds its usefulness in low-field magnetometer and electronic compass applications.c Hanoi Community College (HCC), Trung Kinh, d Hung Yen University of Technology and Educa EHY), Dan Tien, Khoai Chau, Hung Yen 160000, Viet NamHa Noi 100000, Viet Nam b Institute of Engineering Physics (IEP), Ha Noi University of Science and Technology (HUST), Ha Noi 100000, Viet NamOriginal Article Magnetoresistive performances in excha their roles in low-field magnetic sensin Van Su Luong a, *, Anh Tuan Nguyen a, Quoc Khanh Anh Tue Nguyen b, Tuan Anh Nguyen c, Van Cuong a International Training Institute for Materials Science (ITIMS), Ha Noi University of Sci journal homepage: wwwy Elsevier B.V. on behalf of Vietnamge-biased spin valves and applications oang a, Tuyet Nga Nguyen b, iap d and Technology (HUST), Dai Co Viet, Hai Ba Trung, le at ScienceDirect d Materials and Devices sevier .com/locate/ jsamdNational University, Hanoi. This is an open access article under the CC BY license cedwell [36]. Furthermore, for the low magnetic field measurement, errors should be kept as small as possible. The error arises from the hysteresis (Hc) of the FL. A large Hc of the FL leads to a big error. Regarding the responsivity of the sensor, the high MR ratio, e.g. the large change of the resistance in an external magnetic field pro- vides a high sensitivity and a large signal-to-noise ratio. The MR ratio depends onmany factors, such as the texture of the seed layer, the composition of the ferromagnetic alloy layer, and the critical thickness of the spacer or each of the functional layers. Eventually, the cool-field anneal process also strongly affects the MR perfor- mance owing to the inter-diffusion. In fact, the technical factors interact complicatedly together on the optimization of the MR performance so that during the fabrication of the spin valve (SV) films for sensor developments, technicians must decide the trade- off between these factors to find out the appropriate parameters for the SV fabrication. This case study, therefore, is emphasized on the technical factors that affect the magnetoresistive properties of the SVs fabricated by the RF sputtering deposition towards low magnetic field sensing applications, including Heb, Hc, and MR performance, etc. For an ideal comprehensive investigation, the variables should be changed one by one, and the interplay between all the above parameters should be considered. This, however, would take a lot of time and, therefore not be feasible. We, thus, refer to the knowledge reported on the SV performances to apply in our works. Here, we focused on some important factors, such as Heb, Hc, and the MR ratio. We changed the fabrication variables in the range based on our previous reports to optimize them separately one by one [37]. For these investigations, the SV structure of (Si/SiO2)/Ta(tTa)/[NiFe/Co]/Cu(tCu)/Co100-xFex/IrMn(tIrMn)/Ta was chosen. In the sensor applications, the SV films were patterned into micron scale cells and used to form a half-bridge sensor. The magnetic properties of the patterned SVs also strongly depend on the micro-scale size of the components [38,39]. Finally, the per- formance of the patterned SV was demonstrated via the Earth's magnetic field measurement. The obtained experimental results of the SV sheet films and the prototype sensor are presented and discussed. 2. Experimental Exchange-biased SVs were prepared by the RF sputtering tech- nique. The vacuum in this procedure was 6
105 Pa, and the working pressure (Argon) was 0.7 Pa. The distance from the target to the substrate was about 8 cm. Firstly, a 400 nm layer of SiO2 was sputtered onto a 5 mm
5 mm oxidation silicone substrate. The antiferromagnetic (AFM) layer was then sputtered using an alloy target of Ir25Mn75, while the ferromagnetic (FM)material layer of the CoeFe alloy was formed by co-sputtering from two pure Co and Fe targets. The composition of Co100-xFex was tuned by fixing the sputtering power of the Co target while adjusting the sputtering rate of Fe. The content of the as-prepared CoFe layer was characterized by the X-ray energy dispersive spectroscopy attached to a field emission scanning electron microscop of the model JEOL JSM-7600F. The film thickness was measured by an atomic force microscop. The sput- tering rate was determined by the thickness of each deposited functional layer versus the deposition time. The average deposition rate was about 30÷50 Å/minute. A cool-field annealling procedure was used to define the bias direction as the easy axis using a static magnetic field of 2 kOe. The heating time was 30 min, and the cooling time to room temperature (RT) lasted 60 min. The SV films were magnetically characterized using a vibration sample magne- tometer (VSM). The magnetoresistance (MR) was measured by a conventional four-probemethodwith a bias current passing in-plane V.S. Luong et al. / Journal of Science: Advan400of the SV.In the sensor applications, the lithographic lift-off method was used to pattern the device. The active bar SV was of 2 mm
150 mm in size. The easy axis was patterned into the short dimension of the bar, hereby, the cross shape anisotropy between the FL and the PL contributed to a collapse of the hysteresis of the patterned cells. The line width of the contact pads was 25 mm, and the bonding pad was 150 mm
150 mm, which was designed by a photo-mask for silver (Ag) deposition. The chip dimensionwas cut into 1 mm
2 mm. To construct a sensor, two cut chips were glued to the printed circuit board (PCB) for wire bonding, while the other side of the PCB was soldering connected by two identically passive resistances. To demonstrate the features of the engineered sensor, a sweeping magnetic field generated by a Helmholtz coil with the amplitude of ±10 mT was applied to the sensor. The sensitivity of the sensor was determined by the slope of the sensor's output versus the sweeping field curve. In addition, since the intercoupling field will cause a shift of this curve, so the magnetic bias technique was applied using a tiny permanent magnet. The field strength of the permanent magnet for biasing the operation point was controlled by adjusting the distance between the sensor and the magnet. Finally, an Earth's magnetic fieldmeasurement was carried out for verifying the features of the engineered sensor. A manual rotation frame was set up allowing the rotation to take from 0 to 360. The sensor probe was fixed in the center of the rotation frame and rotated with an interval of 10. Each point of the rotation, the output of the sensor was recorded by a data acquisition (DAQ), which was a multifunctional module of MyDAQ provided by Na- tional Instruments. The software was coded in LabVIEW. We first study the magnetoresistive properties of SV sheet films for finding out the appropriate parameters of the SV film fabrication. The conventional approach of sputtering was used to prepare the films based on our previous work [37] and refering to the recent reports on the SV film fabrication technology [40,41]. 3. Results and discussion 3.1. Magnetoresistive properties of the SV sheet films 3.1.1. Exchange bias (Heb) Exchange-biased spin valves were first introduced by Dieny et al. in 1990 [17,20,42], based on a simple sandwich structure of a GMR layer with an additional AFM layer, which is in contact with one FM layer of the sandwich structure. The result of the AFM-FM contact is an interfacial exchange interaction, which is the so- called exchange biasing effect. This structure is a simple exchange-biased spin valve structure. The other (free) FM layer in the sandwich structure was unpinned and can rotate freely under a weak external magnetic field. This free layer was made of soft magnetic materials. The magnetic properties are demonstrated by the M-H loop, shown in Fig. 1(a). The interesting feature of a spin valve structure is that the M-H loop is asymmetric caused by the exchange biasing effect. As previous reports show that the GMR in multilayers is symmetric, so a strong magnetic field is needed to reverse the magnetization direction and, thus, a static magnetic field bias or modulation technique is required for the low field magnetic sensing and measurements. Suppose that the SV is exposed to a strong magnetic field that larger than the Heb (AFM saturated), and the magnetization of both the PL and FL is parallel, the SV is, thus, insensitive to the external magnetic fields. Therefore, the working range of an SV sensor is only active within the reversed magnetization state of the FL and must be smaller than Heb [17,20,42e44]. In sensor applications, Heb should be as large as possible [45]. In this work, the reference layer wof CoFe was used due to its high magnetic moment and high Materials and Devices 3 (2018) 399e405interfacial coupling with the AFM layer. Among the ferromagnetic of C NiFe ncedmaterials, Co and its alloy with iron play a significant role in the performance of SV [18]. Fig. 1(b) shows the obtained experimental results of Heb and the MR ratio as a function of the Fe content in the CoFe alloy. The Heb increases with the increasing iron content and reaches a maximum at 40%, inducing a close-to-maximum magnetic moment of the CoFe alloy leading to a large exchange coupling energy in the IrMn/ CoFe. The MR is reduced monotonically with the increasing Fe content because the high Fe concentration induces a lower mag- netic moment in the CoeFe alloy leading to the weakening of the interfacial exchange coupling in the IrMn/CoFe [46,47]. The effects of the microstructure including grain size and texture of the AFM (IrMn-111) layer on the Heb has been reported by M. Pakala et al. [48]. The Heb dependence on the Fe content has also been revealed that an Fe content of 30% induces a high exchange anisotropy [49], while other work claimed that a high exchange bias could be reached at 45% of Fe [46]. On the other hand, a series of the pub- lished papers have confirmed that good MR performance is ob- tained with 10% of Iron [50e53]. It is also revealed there that the Heb and the MR performance are very sensitive to the specific fabrication equipment. Because of the trade-off between Heb and MR ratio, in this work, x ¼ 20% was the iron content chosen for the SV film fabrication. Furthermore, the Heb also strongly depends on the AFM layer thickness. Fig. 1(c) shows the effects of the IrMn layer thickness, which was varied from 50 Å to 300 Å on theMR behaviors and the Heb value. A lower Heb in the thicker AFM layer is caused by the suppression of the (111) texture [54], while the MR is reduced significantly in a thicker AFM layer owing to the shunting current [55,56]. When the thickness of the IrMn layer has further increased, the Heb has reached a maximum at 100 Å and it drops as this film Fig. 1. (a) Illustration of a M-H loop of a typical SV, (b) Heb and MR versus Fe contents Ta(50 Å), and (c) MR and Heb as a function of IrMn thickness of SVs Si/SiO2)/Ta(50 Å)/[ V.S. Luong et al. / Journal of Science: Advabecomes thicker (so, at thicknesses >100 Å). As a result, both the Heb strength and theMR ratio reached a maximum at tIrMn ¼ 100 Å, and this was chosen for the further investigations. 3.1.2. Magnetoresistance performance (DR/R0) The change of the resistance in the spin valve depends on the relative magnetization angles between the PL and the FL, as defined by DR/R0 in Fig. 2(a), where R0 is the base resistance of the SV in the zero external field and DR is calculated using the resistances in the parallel and anti-parallel magnetization states of the PL and the FL. In the small field region close to H¼ 0, the magnetizing direction of the FL is reversed, and that is actually theworking region of SV. This is also a crucial advantage of the SV in low magnetic field sensing and measurements as well as in the data storage applications because it can detect an extremely weak magnetic field, e.g., the magnetic field induced by a data bit memory, the biological mag- netic fields, and the Earth's magnetic field etc. The resistive sensi- tivity could be defined by the slope of the DR/R0 versus DH curvewithin the reversemagnetization of the FL. Beside the dependences of theMR ratio on the iron content in the CoFe alloy layer and on the thickness of the AFM layer, as mentioned above, the MR perfor- mance is also sensitive to the initial texture of the buffer layer. Therefore, the result of the cool-field anneal process, and especially the thickness of the spacer layer are the main factors affecting the MR performance of SVs. Fig. 2(b) shows the effect of the seed layer (Ta) on the MR ratio. The maximumMR is observed at tTa ¼ 50 Å. At the thickness above 50 Å, the MR ratio decreases with the further increasing Ta thick- ness owing to the stable bcc-phase of Ta in the thicker layer, where the (111) texture disappears. Currently, a seed or buffer layer is considered a standard part of the SV structures. The benefit of the seed layer is to induce a (111) texture in the SV structure [44,57]. The (111) texture has been reported to boost an enhancement of the MR in SV films [58]. Moreover, with the present of the (111) texture, the Heb is also enhanced [48]. Therefore, the disappearance of the (111) texture due to the lack of a seed layer leads to a decrease inHeb and in the working temperature of SV films as well [54]. In fact, the magnetoresistive properties of SVs are very sensitive to the seed layer used. The effect of the various buffer layers in a typical SV structure with a strong (111) texture has also been reported by Ryoichi et al. [59]. Finally, a Ta layer thickness of 50 Å was used for the seed layer in all spin SVs studies in this work. The crucial role of the non-magnetic spacer layer in the SV is that it provides the coupling mechanism between the two FM layers. The coupling sign of magnetizations of the FM layers can be controlled by tuning the spacer thickness leading to the magneto- resistance oscillating with the varying spacer thickness. Therefore, this investigation aimed to find out a critical spacer thickness that provides the maximum MR ratio with an appropriate Hin. The oFe alloy of SVs (Si/SiO2)/Ta(150 Å)/Co(45 Å)/Cu(30 Å)/Co100-xFex(25 Å)/IrMn(250 Å)/ (30 Å)/Co(15 Å)]/Cu(24 Å)/Co80Fe20(25 Å)/IrMn(tIrMn)/Ta(50 Å). Materials and Devices 3 (2018) 399e405 401dependence of theMR ratio on the non-magnetic layer thickness of the SVs has also been extensively studied [20,38,44]. Fig. 2(c) shows the dependence of the MR on the non-magnetic Cu layer thickness, tCu. At tCu < 24 Å, theMR ratio increases, but the Hin become stronger leading to the shift of the working point. This is mainly caused by the roughness of spacer layers, the pinholes, and the interlayer coupling field of the PL [36]. In the lowmagnetic field sensing applications, the bias point (or operation point) of the SV should be as close and symmetrical as possible around the zero field (H ¼ 0). At the thicknesses above 24 Å, e.g. tCu ¼ 30 Å, the MR decreases with an increasing spacer layer thickness. The suppres- sion of the MR could be explained for two reasons. Firstly, the probability of the bulk scattering is proportional to the thickness of the Cu conductive layer. This scattering is not dominating against the electrons passing the FM layers, so theMR is reduced. Secondly, because the high shunting current of the thicker spacer layer also contributes to reduce the MR ratio [45,60,61], while Hin is improved, it revealed that the coupling between the PL and the FL is f SV 15 Å e20( cedweaker in a thicker spacer layer. Finally, by the trade-off between the MR and Hin, tCu ¼ 24 Å was chosen for SV film fabrications. As we know, a cool-field anneal process is an indispensable step in the fabrication proce