NiMn bilayers with different seed and cap layers

NiMn bilayers with different seed and cap layers

Journal of Magnetism and Magnetic Materials 257 (2003) 190–194 Structural and magnetic properties of NiFe/NiMn bilayers with different seed and cap l...

342KB Sizes 1 Downloads 14 Views

Journal of Magnetism and Magnetic Materials 257 (2003) 190–194

Structural and magnetic properties of NiFe/NiMn bilayers with different seed and cap layers B. Dai*, J.W. Cai, W.Y. Lai State key Laboratory of Magnetism, Institute of Physics and Center for Condensed Matter Physics, Chinese Academy of Sciences, Group CM 04, P.O. Box 603, Beijing 100080, China Received 18 July 2002

Abstract The structural and magnetic properties of Ni0.81Fe0.19/Ni0.42Mn0.58 bilayers with Ta or (Ni0.81Fe0.19)0.58Cr0.42 as the seed and cap layers were investigated. It was found that Mn diffused into Ni0.81Fe0.19 layer during the transformation of Ni0.42Mn0.58 layer from nonmagnetic to antiferromagnetic phase through annealing, which causes the decrease of magnetic moment in the ferromagnetic layer. Since (Ni0.81Fe0.19)0.58Cr0.42 cap layer can accommodate Mn atoms but Ta layer cannot, the reduction of the magnetization for the films with (Ni0.81Fe0.19)0.58Cr0.42 seed and cap layers was less than that with Ta seed and cap layers. On the other hand, the grain size of the bilayers with (Ni0.81Fe0.19)0.58Cr0.42 seed layer was much larger than that of the films with Ta seed layer, which leads to a better thermal stability for the former. The present results indicate that (Ni0.81Fe0.19)0.58Cr0.42 can be promising seed and cap layers in spin valves based on Mn-alloyed antiferromagnets. r 2002 Elsevier Science B.V. All rights reserved. PACS: 75.70.i; 75.70.Cn; 75.70.Ak Keywords: Exchange bias; Interdiffusion; Thermal stability; Bilayers

1. Introduction A spin valve based on the giant magnetoresistance (GMR) was first introduced in 1991 [1]. Applications of the GMR spin valve, including magnetic field sensors, read heads for hard drives, galvanic isolators and magnetoresistive random access memory, have been expanded thereafter. The prerequisite for spin valve devices is a good thermal stability and a sufficient exchange bias field (Hex ) resulting from the exchange coupling *Corresponding author.

between an antiferromagnetic (AF) pinning layer and a ferromagnetic (FM) pinned layer. Typically, there are two kinds of AF materials exhibiting exchange bias effect. One is the AF oxide, such as CoO [2], NiO [3] and Co1xNixO [3]. Spin valves based on these AF materials exhibit poor stability and relatively small exchange bias field. The other kinds of AF materials is Mn-based antiferromagnets, including FeMn, IrMn, PdMn, PtMn and NiMn [4,5], which exhibit large Hex and relatively good thermal stability [6]. However, a universal disadvantage for these Mn-based antiferromagnets is the interdiffusion, especially Mn

0304-8853/03/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 0 2 ) 0 1 1 7 0 - 8

B. Dai et al. / Journal of Magnetism and Magnetic Materials 257 (2003) 190–194

migration, when the spin valves are heated [7,8]. So thermal stability is still a key problem and various methods have been exploited for this purpose. Selecting a suitable seed layer is an efficient way to control the microstructure and furthermore alter the magnetic properties. The Ta thin film is a classical seed layer to induce a strong (1 1 1) texture of NiFe, CoFe layer [9,10]. More recently, Lee et al. have reported that a thin Ni0.81Fe0.19 film grown on a thin (Ni0.81Fe0.19)1xCrx (0:38oxo0:45) seed layer exhibits a large enhancement in AMR value, which is attributed to large (1 1 1) textured grains of the NiFe film [11]. In this article, we have studied the structural and magnetic properties of Ni0.81Fe0.19/ Ni0.42Mn0.58 bilayers with Ta or (Ni0.81Fe0.19)0.58Cr0.42 as the seed and cap layers, and compared the thermal stability and the interdiffusion in NiFe/NiMn bilayers with these two kinds of seed and cap layers.

191

annealed at 2401C for 5 h and cooled down to room temperature in a vacuum oven (base pressure 1  105 Pa) with a static magnetic field of 800 Oe parallel to easy axis of the films. After this procedure, we again heated these annealed films at preset temperatures (Tp ¼ 10022401C) for 1 h in the vacuum but the magnetic field direction reversed to study the variety of exchange paths [12] and the distribution of blocking temperatures Tb : The hysteresis loop and exchange bias field were measured by a room-temperature vibrating sample magnetometer. The surface morphology and the crystalline textures of the films were determined by the atomic force microscopy (AFM) and X-ray diffraction (XRD), respectively. The concentration profiles of each element in the layers were obtained through Auger electron spectroscopy (AES) analysis.

3. Results and discussion 2. Experimental details Samples were deposited by a four-target DC magnetron sputtering apparatus onto corning glass substrates. The base pressure in this system was less than 5  105 Pa and Ar pressure was 0.5 Pa. Ta and Ni0.81Fe0.19 alloy targets were used to deposit Ta and NiFe films. An Ni target with some Mn slices symmetrically attached was used to deposit NiMn films, and similarly an Ni0.81Fe0.19 target with some Cr slices attached was used to deposit NiFeCr films. By adjusting the number of Mn and Cr slices, Ni0.42Mn0.58 and (Ni0.81Fe0.19)0.58Cr0.42 films (denoted as NiMn and NiFeCr, respectively) were readily fabricated. The composition of the films was determined by ICP-AES. Two kinds ( ( of samples, Glass//Ta(15 A)/NiFe(150 A)/NiMn( ( and Glass//NiFeCr(15 A)/NiFe( (330 A)/Ta(50 A) ( ( ( were sequen(150 A)/NiMn(330 A)/NiFeCr(50 A) ( tially deposited at sputtering rates of 0.7–1.2 A/s for different materials in the presence of a 400 Oe in-plane magnetic field provided by permanent magnets to induce an easy axis beforehand. In order to transform face-centered-cubic (fcc) nonmagnetic NiMn to face-centered-tetragonal (fct) antiferomagnetic phase, the samples were

Fig. 1 shows the M2H curves for NiFe/NiMn bilayers with NiFeCr and Ta as seed and cap layers in the as-deposited state and after annealing at 2401C for 5 h, respectively, where the magnetic moment M is normalized by the saturated moment Ms0 of the corresponding as-deposited sample. One can note that both samples exhibit no exchange bias in the as-deposited state, and both have nearly the same pinning field about 250 Oe with reduced moment after annealing. The most interesting fact here is that, after annealing, the NiFe/NiMn bilayer with NiFeCr as seed and cap layers has much less reduction of Ms than that with Ta. The nonmagnetic NiMn is transformed into AF materials through annealing, and the exchange bias is thus set up, which is consistent with the results previously reported [4,5]. The magnetization reduction for the samples after annealing seems to correlate with Mn diffusion into the NiFe layer, which destroys partially the magnetic moments of NiFe layer. In order to clarify this conjecture, the Auger depth profile was carried out for the annealed samples, and the results are shown in Fig. 2. As can be seen, the signal from Mn element is visible in NiFe layer for both

B. Dai et al. / Journal of Magnetism and Magnetic Materials 257 (2003) 190–194

M / MS0

192

M / MS0

(a)

1.5

0.9

1.0

0.6

as-depo

0.5 0.0

0.0

-0.5

-0.3

-1.0

-0.6

-1.5 -150

-0.9 -100

-50

0

50

100

150

(b)

1.5

0.9

1.0

0.6

as-depo

0.5

0.0

-0.5

-0.3

-1.0

-0.6

(c)

-100

-50

0

50

100

-500

-400

-300

-200

0.3

0.0

-1.5 -150

annealed

0.3

150

Hex (Oe)

-0.9

(d)

-100

0

annealed

-500

-400

-300

-200

-100

0

100

Hex (Oe)

Fig. 1. Magnetic hysteresis loops for NiFe/NiMn films with NiFeCr or Ta as seed and cap layers in the as-deposited state and after annealing at 2401C for 5 h: (a) NiFeCr as-deposited; (b) NiFeCr after annealing; (c) Ta as-deposited; and (d) Ta after annealing.

samples. However, Mn element in NiFe layer is less for a sample with NiFeCr as the seed and cap layer than that with Ta, which is in agreement with the different moment reduction for the samples with different seed and cap layers. From Fig. 2, one can also note that there is a substantial amount of Mn diffused into the NiFeCr cap layer, but little Mn diffused into the Ta cap layer. The crystalline NiFeCr layer can accommodate an Mn element, but amorphous Ta layer cannot. As for the NiMn layer, there is no much difference of Mn content for these two samples. So it is natural that the quantity of Mn intruded into NiFe layer is less for the sample with NiFeCr seed and cap layer. Finally, we would like to point out that some Fe diffused into NiMn layer, which was previously observed in the systems of IrMn [13] and FeMn [14]. Fig. 3 shows the XRD patterns of the asdeposited samples. Obviously, NiFe layer has a higher (1 1 1) texture for the sample with NiFeCr

seed and cap layers, which is consistent with the results reported by Lee et al. [11]. However, the NiMn (1 1 1) texture is hardly altered by the NiFeCr seed layer further. To investigate the surface topography and lateral dimension of NiMn films, we fabricated two samples, with the struc( ( ( tures of NiFeCr(15 A)/NiFe(150 A)/NiMn(330 A) ( ( ( and perand Ta(15 A)/NiFe(150 A)/NiMn(330 A), formed AFM observations on them. The threedimensional AFM micrographies, which are (  5000 A ( scans, are shown in Fig. 4. For 5000 A the film with an NiFeCr seed layer, the NiMn ( in diameter on an grains are as big as 240 A average. However, the film with Ta seed layer has ( in much smaller NiMn grains, about 180 A diameter. So the grains of NiMn for the sample with NiFeCr seed and cap layer are much larger than those with Ta. It has been reported that small AF grains are susceptible to the thermal fluctuation, while large

B. Dai et al. / Journal of Magnetism and Magnetic Materials 257 (2003) 190–194

193

Atomic concentration (%)

70 Ni Mn Fe

60 50 40 30 20 10 0

NiFeCr

-0.5

0.0

NiMn

0.5

1.0

NiFe

1.5

2.0

2.5

3.0

NiFeCr

3.5

4.0

Sputter time (min)

(a) 70

Ni Mn Fe

Atomic concentration (%)

60 50 40 30 20 10 0

Ta

-0.5 0.0

NiMn

0.5

1.0

(b)

1.5

NiFe

2.0

2.5

3.0

Ta

3.5

4.0

Sputter time (min)

Fig. 2. AES depth profiles for NiFe/NiMn films with (a) NiFeCr and (b) Ta as seed and cap layers after annealing at 2401C for 5 h, respectively.

500 NiMn(111)

NiFeCr

400

Fig. 4. The three-dimensional AFM topographies of as-depos( ( ited samples with structures: (a) NiFeCr(15 A)/NiFe(150 A)/ ( and (b) (15 A)/NiFe(150 ( ( ( NiMn (330 A) A)/NiMn (330 A), respectively.

Ta

Intensity (cps)

NiFe(111)

300 200 100 0 40

42

44

46

48

50

2θ (deg)

Fig. 3. XRD profiles for the as-deposited NiFe/NiMn films with NiFeCr or Ta as seed and cap layers.

AF grains with large potential barrier prevent them from thermal agitation and contributes to an insensitive temperature dependence of exchange bias [5,15]. So we can expect that the sample with NiFeCr seed and cap layer is thermally more stable than the sample with Ta seed and cap layers. To verify this hypothesis, heat treatment of the samples with magnetic field direction reversed was carried out at various temperatures. Fig. 5 shows the resulting exchange bias field versus annealing temperature T; where Hex is rescaled by the factor

B. Dai et al. / Journal of Magnetism and Magnetic Materials 257 (2003) 190–194

194

reduction of the magnetization for the films with NiFeCr seed and cap layer was less than that with Ta seed and cap layers because NiFeCr layer accommodates Mn atoms. Besides, the sample with NiFeCr seed layer has larger NiMn grains, a narrower distribution of exchange paths, and is thermally more stable. Conclusively, NiFeCr can be a substitute for Ta in the spin valves based on Mn-alloyed AF materials.

250 200

Hex (Oe)

150

NiFeCr Ta

100 50 0 -50 0

50

100

150

200

250

Acknowledgements

o

Temperature ( C)

Fig. 5. Exchange bias field versus annealing temperature with the magnetic field direction reversed.

Ms (annealed)/Ms (as-deposited) because of the Ms reduction during the annealing processes. It can be seen that Hex in the sample with Ta seed and cap layers starts to decrease when the annealing temperature is above 1101C, but it keeps the high value unchanged until 1301C for the sample with NiFeCr seed and cap layers. Moreover, the Hex decreases more steeply and the direction of exchange bias reverses at lower temperature for the former. The curve of Hex 2T is determined by the distribution of exchange paths [12, 15] and, vice versa, the distribution of Tb can also be deduced from the Hex  T curve. From the present experiment, both samples having almost 50% of the exchange paths, have a Tb higher than 2201C, and most importantly, the film with NiFeCr seed and cap layers has a better thermal stability, and a narrower distribution of exchange paths. Based on the above experiments, we would like to point out that NiFeCr has more merits, and could be a promising seed and cap layer for spin valves in the future.

4. Conclusions Diffusion of Mn into NiFe layer leads to the reduction of the magnetization for NiFe/NiMn films during the annealing process. However, the

This work was supported by the Major Program of the National Natural Sciences Foundation of China (Grant No. 19890310).

References [1] B. Diney, V.S. Seriosu, S.S.P. Parkin, B.A. Gurney, D.R. Wihoit, D. Mauri, Phys. Rev. B 43 (1991) 1297. [2] W.H. Meiklejohn, C.P. Bean, Phys. Rev. 102 (1956) 1413; T. Ambrose, C.L. Chien, Appl. Phys. Lett. 65 (1994) 1967. [3] M.J. Carey, A.E. Berkowitz, Appl. Phys. Lett. 60 (1992) 3060; H. Yamane, M. Kobayashi, J. Appl. Phys. 83 (1998) 4862. [4] T. Lin, D. Mauri, N. Staud, C. Hwang, J.K. Howard, G.L. Gorman, Appl. Phys. Lett. 65 (1994) 1183. [5] S. Mao, N. Amin, E. Murdock, J. Appl. Phys. 83 (1998) 6807. [6] J.P. Nozi"eres, S. Jaren, Y.B. Zhang, A. Zeltser, K. Pentek, V.S. Speriosu, J. Appl. Phys. 87 (2000) 3920. [7] M.F. Toney, C. Tsang, J.K. Howard, J. Appl. Phys. 70 (1991) 6227. [8] J.H. Lee, H.D. Jeong, C.S. Yoon, C.K. Kim, B.G. Park, T.D. Lee, J. Appl. Phys. 91 (2002) 1431. [9] H. Fujiwara, K. Nishioka, C. Hou, M.R. Parker, S. Gangopadhyay, R. Metzger, J. Appl. Phys. 79 (1996) 6286. [10] G. Choe, S. Gupta, Appl. Phys. Lett. 70 (1997) 1766. [11] W.Y. Lee, M.F. Toney, P. Tameerug, E. Allen, D. Mauri, J. Appl. Phys. 87 (2000) 6992. [12] S. Soeya, T. Imagawa, K. Mitsuoka, S. Narishige, J. Appl. Phys. 76 (1994) 5356. [13] J.H. Lee, H.D. Jeong, C.S. Yoon, C.K. Kim, B.G. Park, T.D. Lee, J. Appl. Phys. 91 (2002) 1431. [14] M.F. Toney, C. Tsang, J.K. Howard, J. Appl. Phys. 70 (1991) 6227. [15] S. Araki, M. Sano, M. Ohta, Y. Tsuchiya, K. Noguchi, H. Moriya, M. Matsuzaki, IEEE Trans. Magn. 34 (1998) 1426.