Comparative study of magnetic, microwave properties and giant magnetoimpedance of FeNi-based multilayers with different structure

Journal of Alloys and Compounds 615 (2014) S296–S299

Contents lists available at ScienceDirect

Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom

Comparative study of magnetic, microwave properties and giant magnetoimpedance of FeNi-based multilayers with different structure G.V. Kurlyandskaya a,b,⇑, E. Fernández a, A. García-Arribas a,c, V.N. Lepalovsij b, S.O. Volchkov b a

University of the Basque Country UPV-EHU, Bilbao, Spain Ural Federal University, Ekaterinburg, Russia c BCMaterials, Bilbao, Spain b

a r t i c l e

i n f o

Article history: Available online 3 February 2014 Keywords: Magnetic multilayers Magnetic properties Magnetoimpedance Ferromagnetic resonance

a b s t r a c t FeNi(100 nm)/Cu(500 nm)/FeNi(100 nm) and [FeNi(100 nm)/Cu(3 nm)]4/FeNi/Cu(500 nm)/[FeNi(100 nm) /Cu(3 nm)]4/FeNi(100 nm) multilayers in the shape of elongated stripes were prepared by rf-sputtering. Magnetization curves were measured by vibrating sample magnetometry. Magnetic domains were studied by the Bitter technique. Magnetoimpedance (MI) was studied as a function of the external magnetic field (H) for a frequency range of 1–300 MHz. Ferromagnetic resonance was measured as a function of the applied field for a 8.85 GHz frequency. Maximum MI sensitivities for total impedance at 75 MHz of 1.1 X/Oe (at 3.0 Oe) for [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi and 0.8 X/Oe (at 6 Oe) for FeNi/Cu/FeNi were obtained. Despite the increase of the magnetic inhomogeneities related to small contribution of mixed interfaces in the [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi structure, its MI response is more adequate for applications. Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction Magnetic, microwave properties and giant magnetoimpedance (MI) of magnetic multilayers are hot topics of basic research and technological applications [1]. High frequency characterization like ferromagnetic resonance (FMR) and MI play a more and more important role when complete characterization is requested [2,3]. It is also important to mention that precise MI measurements of thin films and multilayered structures request a rather complex procedure. At the same time, rapid characterization of MI materials would be a great advantage. As both FMR and MI are magnetic phenomena in which a dynamic magnetic permeability is involved, a number of attempts were made to use conventional FMR measurements for MI materials characterization [3–4]. Despite years of experimental and theoretical efforts [5,6] for MI multilayers, the maximum theoretical value of MI is still much higher than the obtained experimental results [7]. In recent years visible progress was made in the enhancement of both MI value and MI sensitivity of thin film based structures. The strategy of employing multilayers instead of thin films was successful [8,9]. There were many comparative studies of MI structures with open and closed magnetic flux ⇑ Corresponding author at: University of the Basque Country UPV-EHU, Bilbao, Spain. E-mail addresses: [email protected] (G.V. Kurlyandskaya), eduardo.fernandez @ehu.es (E. Fernández), [email protected] (A. García-Arribas), vladimir.lepalovskiy@ usu.ru (V.N. Lepalovsij), [email protected] (S.O. Volchkov). http://dx.doi.org/10.1016/j.jallcom.2014.01.185 0925-8388/Ó 2014 Elsevier B.V. All rights reserved.

[6,10]. In the case of the MI sandwich with opened magnetic flux central conductor has the same width as the width of the magnetic layers. Keeping in mind that magnetic properties of permalloy films depend on the film thickness due to the existence of a transition into a ‘‘transcritical’’ state [9], magnetically soft FeNi-based multilayers were designed with the total thickness of magnetic layers being much higher than the critical thickness of the transition into ‘‘transcritical’’ state [8,10]. At the same time the direct comparison of the magnetic properties and MI for ferromagnet/conductor/ferromagnet (F/C/F) structures in the case when the ferromagnetic layers are either a thin film or a multilayer is still absent. In this work magnetic, microwave properties, magnetic domain structure, MI and FMR were comparatively analyzed for F/C/F MI sandwiches of the same geometry with opened magnetic flux (where F is either a single FeNi layer, or a [FeNi/Cu]4/FeNi multilayer, both prepared in the same conditions.

2. Experimental F/C/F MI sandwiches were deposited onto glass substrates by ion-plasma sputtering in an Ar atmosphere at a working pressure of 103 Torr, providing FeNi films of the highest quality [10]. The preliminary vacuum was 2106 Torr. Ferromagnetic layers were either FeNi(130 nm) film (S-I) or [FeNi(100 nm)/Cu(3 nm)]4/ FeNi(100 nm) multilayer (S-II). The thickness of permalloy for FeNi/Cu/FeNi sandwich was on purpose slightly higher than the FeNi layer thickness of 100 nm in [FeNi/Cu]4/FeNi multilayer, but both of them were below the critical thickness of the transition into ‘‘transcritical’’ state of 200 nm (Fig. 1). The total thickness of the magnetic layers in [FeNi/Cu]4/FeNi multilayers was as high as 500 nm. The

S297

G.V. Kurlyandskaya et al. / Journal of Alloys and Compounds 615 (2014) S296–S299

S-I

S-II

Table 1 Selected properties of FeNi-based multilayers: Hc – coercivity, Ha – anisotropy field; Hk – FMR resonance field for external in plane field; H\ – FMR resonance field for external plane applied perpendicular to the plane of the sample; DH – FMR resonance line width.

FeNi

FeNi Cu

Cu

M (G)

400

0

-400

S-I S-II

-800 -12

-8

-4

0

4

8

H (Oe) Fig. 1. VSM hysteresis loops and description of the MI structures. The field during measurements was applied parallel to the stripe length.

thickness of the Cu lead was 500 nm providing high MI for reasonably low frequencies in accordance with previous studies [9]. To obtain a transverse magnetic anisotropy, the depositions were done in in-plane magnetic field of 100 Oe applied along the short side of the element. FeNi and Cu deposition rates were 0.11 and 0.14 nm/s, respectively. The samples had 9  0.5 (mm) shape. The magnetization curves were measured by a vibrating sample magnetometer (VSM) using low field measuring system with KEPCO power supply MOD.NO.BOP 100-10MG. Low field measuring configuration in combination with precise Gaussmeter allowed us to make measurements with the accuracy up to 0.1 Oe. Magnetic domain structures were studied by the Bitter technique [11]. The FMR was measured by a cavity perturbation technique with conventional homodyne detection at room temperature for resonance frequency, f, of 8.85 GHz. The whole MI stripe fit into the resonance cavity and therefore FMR and MI characterizations were done for essentially identical samples. The external magnetic field, H, was rotated in plane perpendicular to the plane of the sample for measurements of the angular dependences of FMR field. For ‘‘in plane’’ configuration, the resonance field Hres = H|| and for ‘‘out of plane’’ configuration, Hres = H\ [2,10]. A half power width, DH, was determined from the microwave absorption measurement for each sample. The values of the effective magnetization 4pMs were calculated as follows:

x c

¼ Hjj ðHjj þ 4pM s Þ

Ha (Oe)

Hk (kOe)

H\ (kOe)

4pMeff (kOe)

DH (Oe)

0.4 0.2

8.2 2.8

0.9 0.8

12.4 12.9

9.4 10.0

90 170

Fig. 2 shows field dependences of total impedance of S-I and S-II samples for selected frequencies. The shape of the curves is typical for longitudinal MI of the film with transverse magnetic anisotropy: the position of the MI curve peaks in both cases corresponds roughly to the values of Ha. Fig. 3 shows clear difference in the frequency dependences of the maximum values of total impedance (DZ/Z)max and real part (DR/R)max ratios. (DZ/Z)max values show non-linear frequency dependence with the maximum corresponding to f  45 MHz for S-I structure and f = 25 MHz for S-II structure. For the frequencies above 180 MHz S-I structure becomes more efficient. The simplest explanation can be given on the basis of the estimation of skin penetration depth (d) responsible for the impedance variation in the FeNi-based multilayers deposited in similar conditions [12]: in the case of S-I structure d becomes comparable to the thickness of the FeNi layer of 130 nm for the frequencies above the 180 MHz. (DR/R)max frequency dependence is quite different. Very wide maximum corresponds to f  100 MHz for S-II structure and (DR/R)max is a monotonously increasing function for S-I structure. The value of the MI sensitivity with respect to an external field is a principally important parameter for sensor applications. Fig. 4

7

50 MHz 75 MHz 100 MHz

5 4 3

ð2Þ

where x = 2pf, c is a gyromagnetic ratio and g = 2.12. Eqs. (1) and (2) can be used to determine effective magnetization, 4pMeff, and thereby the angular dependence of the resonance field. The longitudinal MI (alternating current flows parallel to the external magnetic field applied along the long side of the stripe) was measured in a magnetic field created by a pair of Helmholtz coils [9]. The impedance measurements were performed using radio frequency (RF) techniques by placing the sample as a part of the microstrip line with 50 O of characteristic impedance [4,10]. Total impedance (Z), its real (R) and imaginary (X) parts were measured by Agilent Network Analyzer as a function of the magnetic field for a frequency, f, range of 1–700 MHz. MI was measured both in an increasing and a decreasing magnetic field. The analysis of the experimental data was limited by the frequencies corresponding to the quasi-static MI range [4]. MI sensitivities with respect to an applied field for total impedance s(Z) and its real part s(R) were calculated by differentiating Z(H) or R(H) curves. MI ratios were calculated as follows: DZZ ¼ 100  ZðHÞZðH¼0Þ ; DRR ¼ 100  RðHÞRðH¼0Þ . ZðH¼0Þ RðH¼0Þ

25 MHz

(a)

6

ð1Þ

 

x ¼ H?  4pM s c

Hc (Oe)

S-I S-II

Z (Ω )

 2

Sample

2 1

S-I

0

7

(b)

25 MHz 50 MHz 75 MHz

6

100 MHz

5

Z (Ω )

800

4 3

3. Results and discussion Fig. 1 shows the VSM hysteresis loops for both multilayered structures. The shape of the loops confirms the formation of the in-plane transverse magnetic anisotropy induced during the deposition. The values of the coercive field, Hc, and the anisotropy field, Ha, are collected in Table 1. One can see that the coercivity is much lower for the S-II sample and the anisotropy field is much higher for the S-I case.

2 1

S-II

0 -50

0

50

H (Oe) Fig. 2. MI of S-I (a) and S-II structures (b).

S298

G.V. Kurlyandskaya et al. / Journal of Alloys and Compounds 615 (2014) S296–S299

(a)

S-I:

(ΔR/R)max;

(ΔZ/Z)max

S-II:

(ΔR/R)max;

(ΔZ/Z)max

1

(a) o

α= 0

S-I

300

S-II

P (arb. un.)

200

ΔH

ΔH

100

0

0 0

100

200

0.0

0.5

f (MHz)

(b)

3

1

(b) o

50 MHz 75 MHz 100 MHz

FeNi/Cu/FeNi 0

2

S-II Lorentz fit

P (arb. un.)

R (Ω)

25 MHz

1

1 α= 0

α= 0

S-I

2

1.0

1.5

H (kOe)

P (arb. un.)

(ΔR/R)max, (ΔZ/Z)max (%)

400

0 0.0

o

S-I Lorentz fit

0.5

1.0

1.5

H (kOe)

0 4

6

8

H (Oe)

0.0

0.5

1.0

1.5

2.0

H (kOe)

Fig. 3. Frequency dependences of the maximum values of MI ratios of FeNi/Cu/FeNi (S-I) and [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi (S-II) MI structures (a); field dependences of real component of MI: selection of the interval of maximum sensitivity.

Fig. 4. FMR lines of S-I and S-II structures for external magnetic field oriented inplane of the samples (a); Lorentz fit for FMR curve of S-II inset shows Lorentz fit for FMR curve of S-I (b).

shows an example of s(R) evaluation. As the first step we select the R(H) close to linear dependence range possibly in a low field, which is considered to be a work interval for MI magnetic field sensitive element: the green rectangular indicates 5.5 Oe < H < 6.5 Oe range. Table 2 lists MI sensitivities for selected frequencies. The most important observation is that the tendencies observed for the maxima of MI are different from the tendencies observed for the maxima of sensitivities. The work interval for the external field 2.5 Oe < H < 3.5 Oe (case S-II) is much lower and more appropriate for technological applications. S-II structures show higher s(Z) and s(R) sensitivities for all frequencies comparing with S-I. The highest sensitivity value was observed for s(Z), f = 100 MHz case of S-II structure. At the same time, for S-II structure, if we select a reasonably small frequency of 25 MHz, the sensitivity results s(Z) = 0.8 X/ Oe, which is a very competitive value. One of the main reasons of high sensitivity of a S-II structure is the total thickness of the magnetic multilayers. A large variation of the current distribution is allowed in the case of thick magnetic layers: d  (pflr)1/2 and the skin effect is expected to play a relevant role when d < D/2 (d – is a classic skin depth., l – is a transverse magnetic dynamic permeability and r – is a conductivity). Another reason of the observed difference in MI sensitivity of S-I and S-II structures is the difference in magnetic permeability of S-I and S-II samples which can be roughly estimated using the data of Fig. 1(b): for S-II it is about twice higher than the permeability corresponding to S-I. In order to further understand the S-I, S-II MI behaviour the FMR of whole MI sensitive elements was measured. Fig. 4(a) shows

the FMR results of S-I and S-II multilayers for a = 0°, i.e. for the case of magnetic field applied in plane of the multilayered structure (a is an angle between the external magnetic field and a plane of the sample). In all cases, only one resonance line was observed. The half power width was determined from the microwave absorption measurement for each sample (Table 1). The value of effective magnetization 4pMeff = 9.4 kOe is slightly lower for S-I and practically the same 4pMeff = 10.0 kOe as the value of the effective magnetization calculated from the VSM measurements, 4pMs = 10.1 kOe. This indicates that the FeNi layers of the multilayered structure contain less stresses and imperfections, i.e. Cu sub-layer introduction can play a positive role leading to mechanical stresses relaxation. Surprisingly, at the same time, the line width is higher for S-II. The FRM lines show a reasonable Lorentzian fit (Fig. 4(b)) but the regression is much better for S-I case (R2 = 0.995) than in S-II case (R2 = 0.986). In both cases only one resonance line was observed (Fig. 4). Even for the S-I with the highest Hc value, a good agreement between the calculated (as in [2]) and experimental angular dependences was obtained. What, in this case, causes the deviation of the shape of the resonance line from a Lorentzian and increase of the line widths? Recently, we have shown that in the case of FeNi-based multilayers, interesting structural features can be detected by X-ray diffraction probably due to the existence of interdiffusion between layers [13]. We suppose that the line widths increase and deviations from the Lorenzian shape in the S-II case is due to mixed interfaces at the surfaces of the 3 nm Cu layers.

S299

G.V. Kurlyandskaya et al. / Journal of Alloys and Compounds 615 (2014) S296–S299 Table 2 Selected properties of [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi and FeNi/Cu/FeNi MI multilayers: RDC – dc resistamce; RH – work interval of MI sensitive element. Sample

RDC (X)

RH (Oe)

S-I S-I S-II

1.32

5.5–6.5 2.5–3.5 2.5–3.5

1.63

f = 25 (MHz)

f = 50 (MHz)

f = 75 (MHz)

f = 100 (MHz)

s(Z) (X/Oe)

s(R) (X/Oe)

s(Z) (X/Oe)

s(R) (X/Oe)

s(Z) (X/Oe)

s(R) (X/Oe)

s(Z) (X/Oe)

s(R) (X/Oe)

0.40 0.07 0.80

0.20 0.02 0.50

0.70 0.12 1.00

0.4 0.02 1.10

0.80 0.20 1.10

0.60 0.02 1.10

0.80 0.18 1.00

0.70 0.07 1.20

magnetization parallel to the short side of the MI element. The average width of the bar domains was of order of 200 lm. There was no indication of deviation of magnetization inside the domains from in-plane position or presence of the charged areas near the borders. We therefore suppose that the FeNi-based MI multilayer provides better conditions for the formation of closed magnetic flux near the borders. 4. Conclusions Magnetic, microwave properties, magnetic domain structure and MI were comparatively analyzed for FeNi/Cu/FeNi and [FeNi/ Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi multilayers of the same geometry. FeNi/Cu/FeNi multilayer has higher coercivity and anisotropy field. Maximum MI sensitivities for total impedance at 75 MHz of 1.1 X/ Oe for [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi and 0.8 X/Oe for FeNi/ Cu/FeNi were obtained. Despite the increase of the magnetic inhomogeneities related to small contribution of mixed interfaces in the [FeNi/Cu]4/FeNi/Cu/[FeNi/Cu]4/FeNi structure, its MI response is more adequate for applications. The presence of Cu-sublayers allows to increase the total thickness of the multilayer to be much higher then the critical thickness of the transition into ‘‘transcritical’’ state. They also insure better conditions for stress relaxation and magnetic flux closing. Acknowledgments This work was performed under MAT2011-27573-C04, project no. 215 ‘‘Magnetodynamics of High-Permeability Nanostructured Media’’ at the Ural Federal University and ETORTEK-ACTIMAT grants. Collaboration with Prof. S.M. Bhagat (University of Maryland) is highly appreciated. Selected measurements were made by I. Orue at UPV-EHU SGIker service. Fig. 5. Magnetic domains revealed by Bitter technique: SI (a and b); S-II (c). The arrows indicate the orientation of magnetization inside domains (schematic drawing is shown on the left of part (c).

Magnetic domain observation gives a further insight (Fig. 5). In the case of S-I multilayer there is accumulation of the ferrofluid typical for ‘‘charged’’ surfaces (especially near the borders) and the domain structure is not exactly uniform over the length of the MI element. One can see the difference in the average length of the closure domains for selected typical areas of S-I sample (Fig 5(a and b)). Even so, the domain structure indicates that the effective magnetic anisotropy is transverse with the easy magnetization axis parallel to the short side of the element, closure domains are quite large, having a typical length of the order of 10% of the width of the MI element. It is rather difficult to make a reasonable drawing of the magnetic moment distribution because the domain walls can be mixed Neel–Bloch walls. Micromagnetic calculations are necessary for the development of an appropriate model. In the S-II case (Fig. 5(c)) very well defined bar domains without closure domains were observed with the orientation of the

References [1] M. Coisson, G. Barrera, F. Celegato, F. Vinai, J. Phys. Conf. Ser. 365 (2012) 012003. [2] G.V. Kurlyandskaya, S.M. Bhagat, A.V. Svalov, E. Fernandez, A. Garcia-Arribas, J.M. Barandiaran, Solid State Phenom. 168–169 (2011) 257–260. [3] S.E. Lofland, S.M. Bhagat, M. Dominguez, J.M. Garcia-Beneytez, F. Guerrero, M. Vasquez, J. Appl. Phys. 85 (1999) 4442–4444. [4] J.M. Barandiarán, A. García-Arribas, D. de Cos, J. Appl. Phys. 99 (2006). 1039043. [5] A. Antonov, S. Gadetsky, A. Granovsky, A. Diachkov, M. Sedova, N. Perov, T. Furmanova, A. Lagarkov, Physica A 241 (1997) 414–419. [6] L.V. Panina, K. Mohri, Sens. Actuators A Phys. 81 (2000) 71–77. [7] L. Kraus, J. Magn. Magn. Mater. 354 (1999) 167–196. [8] M.A. Correa, F. Bohn, C. Chesman, R.B. da Silva, A.D.C. Viegas, R.L. Sommer, J. Phys. D: Appl. Phys. 43 (2010) 295004–295007. [9] Y. Sugita, H. Fujiwara, T. Sato, Appl. Phys. Lett. 10 (1967) 29–231. [10] G.V. Kurlyandskaya, A. García-Arribas, E. Fernández, A.V. Svalov, IEEE Trans. Magn. 48 (2012) 1375–1380. [11] A. Hubert, R. Schäfer, Magnetic Domains, Springer, Berlin, Germany, 1998. [12] E. Fernández, A.V. Svalov, G.V. Kurlyandskaya, A. García-Arribas, IEEE Trans. Magn. 49 (1) (2013) 18–21. [13] N. Villar Alzola, G.V. Kurlyandskaya, A. Larrañaga, A.V. Svalov, IEEE Trans. Magn. 48 (4) (2012) 1605–1608.