Journal of Magnetism and Magnetic Materials 304 (2006) e50–e52
The effect of Os interlayers on the thermal stability of magnetic
CoFe/OsMn films
Tai-Yen Peng
a,, C.K. Lo
b,c, San-Yuan Chen
a, Y.D. Yao
d,aaDepartment of Materials Science and Engineering, National Chiao Tung University, Hsinchu 30010, Taiwan
bLaboratory for Spintronics, Electronics and Opto Electronics Research Laboratories, Industrial Technology Research Institute, Hsinchu 31040, Taiwan c
Nano Technology Research Center, Industrial Technology Research Institute, Hsinchu 31040, Taiwan
d
Institute of Physics, Academia Sinica, Taipei 11529, Taiwan Available online 3 March 2006
Abstract
The thermal stability of a multilayer structure of protection layer/Co90Fe10/Os (d nm)/Os20Mn80has been studied as functions of
annealing temperature (Tan) and thickness of Osmium (Os) layer. The insertion of a thin Os layer between the Co90Fe10/Os20Mn80
interface shows better thermal stability. No diffusion evidence was found for samples with d^0:3 nm as examined by Auger electron spectroscopy depth profile at different annealing temperatures up to 400 1C. These samples with Os layer showed the same magnetic behavior and the hysteresis loop with squareness (S) larger than 0.9 were observed before and after annealing.
r2006 Elsevier B.V. All rights reserved.
Keywords: Thermal stability; Os interlayer; Mn diffusion; CoFe/Os/OsMn; CoFe/Os/IrMn
1. Introduction
The development of the high-performance magnetic devices has draw attention in recent years. One of the important factors for this is the increment of thermal stability, especially the magnetic element used in magnetic random access memory (MRAM), magnetic pickup head, sensor, etc. The magnetic behavior is very sensitive to chemical composition, interface and structure, and there-fore, interdiffusion due to heat treatment may cause problems. Many Mn-metal alloys used as antiferromag-netic layer in magantiferromag-netic device have been extensively studied [1–3]; however, the Mn atom causes interdiffusion problem
to degrade the overall performance [4,5]. It was reported
that the doping of Osmium (Os) may block the Mn
diffusion channel up to 400 1C [6]. These motivated us to
study the thermal stability of magnetic multilayer with insertion of Os layer and how does the Os layer play the role on preventing Mn atoms from diffusing. Results of the study may suggest a better way to enhance the thermal stability in magnetic devices.
2. Experiment
The magnetic multilayer of protection layer/Co90Fe10/Os
(d)/Os20Mn80 were RF-magnetron sputtered on SiO2/Si
(1 0 0) substrate with an in-plane magnetic field of 200 Oe during the growth. The thickness of the CoFe and OsMn were fixed at 10 and 20 nm, respectively. The thickness of the Os layer, d, was varied from 0 to 2 nm. It is also important to prevent the specimen from oxidation during annealing by protection layer. After the growth, these samples were ex-situ vacuum annealed for 30 min at
different temperatures (Tan) with a stronger applied field
of 1 kOe along the easy axis. The structure of the samples was examined by X-ray diffraction (XRD), while the magnetic hysteresis properties were measured by magnetic optical kerr effect (MOKE) and vibrating sample magnet-ometer (VSM). Auger electron spectroscopy (AES) depth profile was used to detect composition distribution along the surface normal.
3. Results and discussion
The AES depth profile results of Co90Fe10/Os(0, 1 nm)/
Os20Mn80of the as-grown and Tan¼300 1C are shown in
ARTICLE IN PRESS
www.elsevier.com/locate/jmmm
0304-8853/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jmmm.2006.01.173
Corresponding author. Tel.: +886 3 5912878; fax: +886 3 5912936. E-mail address: TerencePeng@itri.org.tw (T.-Y. Peng).
Fig. 1. Clearly all films were stable at room temperature (Fig. 1(a) and (c)). Without the Os interlayer, Mn migrates
into the top layer at Tan¼300 1C, and also a small amount
of Co moves downward as shown in Fig. 1 (b).
Fortunately, 1 nm layer of Os is thick enough to stop the
diffusion of Mn and Co up to 300 1C (as seen fromFig. 1(c)
and (d)), and this is also confirmed by the MOKE and VSM measurements that no difference in the magnetic hysteresis loops before and after annealing. Once Mn is
mixed up with Co90Fe10layer (likeFig. 1(b)), the coercivity
(HC) and squareness (S) are found to be changed abruptly
as shown in the insertion picture inFig. 2. The insertion of
2 and 1 nm Os layer did not cause so much difference in the Auger depth profile signal; however, there are little
increases in HC and little decrease in S for the 2 nm Os
inserted samples as shown inFig. 2. The diffusion of Mn
into the ferromagnetic layer also causes the reduction of S. All as-grown samples have rather square hysteresis loop.
The S of the non-Os sample at Tan¼400 1C is reduced to
0.25. Even no interdiffusion evidence were found for
sample with either 1 nm or 2 nm Os layers; however, S is
slightly reduced from 0.96 at Tan¼400 1C to 0.75 at
Tan¼440 1C. This could be due to our Auger system that
the chemical information is too small to detect.
It was also told from the XRD analysis that our samples
with Os20Mn80 layer does not show g-phase which
exhibited FM/AFM exchange coupling as reported by
Ref. [6], however, our sample’s properties do agree with
Ref[7,8].
The Os interlayer thickness dependences of S and
normalized HCfor 400 1C annealed samples are shown in
Fig. 3. Samples with Os interlayer retained their S40:9
even though the Os thickness is as thin as 0.3 nm. The HC
increased after annealing until Os being added also indicated the improvement on thermal stability. The
normalized HC, which was defined as HC ðTanÞ=HC
(as-grown), was used to ignore the area difference between
samples. The normalized HC of annealed sample without
Os became near 4 times larger than that of as-deposited
state, while the normalized HC for the annealed sample
ARTICLE IN PRESS
0 200 400 600 800 0 20 40 60 80 100peak to peak atomic percent ( % )
etching time( sec )
Os NO1 O KL1 Co LM2 Mn LM3 0 200 400 600 800 1000 5 10 15 20 25 30 35 40 45 50 55 60
peak to peak atomic percent ( % )
etching time ( sec )
Os NO1 O KL1 Co LM2 Mn LM3 (a) (b) 0 200 400 600 800 0 10 20 30 40 50 60 70 80 90 100
peak to peak atomic percent ( % )
etching time( sec )
Os NO1 O KL1 Co LM2 Mn LM3 0 200 400 600 800 1000 0 10 20 30 40 50 60 70 80 90
peak to peak atomic percent ( % )
etching time ( sec )
Os NO1 O KL1 Co LM2 Mn LM3
(c) (d)
Fig. 1. AES-depth profile for the Co90Fe10/Os (d nm)/Os20Mn80multilayer before (a), (c) and after (b), (d) annealing at 300 1C ((a), (b) d ¼ 0; (c), (d)
d ¼ 1).
with Os only slightly increased. Because the inserting metal layer in the FM/AFM interface decreased the FM/AFM
exchange coupling [9], the optimal Os barrier thickness,
which could not only retain the magnetic properties of the magnetic layer after annealing but also slightly decrease FM/AFM exchange coupling was an important factor to determine the barrier thickness.
According to our experimental data, the insertion of 1 nm Os layer can block the Mn diffusion channel and
retain the magnetic behavior up to Tan¼400 1C. The
exchange field (Hex) for Co90Fe10/Ir20Mn80with 0.3 nm Os
of as-deposited and 350 1C annealed state was 100 and 190 Oe, respectively. However, sample without Os barrier
showed a Hexof 55 Oe after 350 1C annealing while that of
as-deposited state was 105 Oe. Detailed description of the
Os layer on exchange bias effect can be found in Ref.[10].
It could be also found that the sample after annealing with inserted Os layer made the whole magnetic behavior almost the same with the as-deposited state.
The better thermal stability of Co90Fe10/Os20Mn80
structures could be achieved by inserting a thin Os layer. As the annealing temperature up to 400 1C, no diffusion evidence was found for samples with d^0:3 nm as
examined by the AES depth profile, HC and S
measure-ment. After 400 1C annealing, the sample with Os layer
with hysteresis loop showed little larger HC and S40:9
could be obtained even though the thickness of Os was as thin as 0.3 nm.
Acknowledgements
This work was financial supported by ROC MOEA and NSC under the grant Nos. of A331XS3710 and NSC-94-2120-M-001-008, respectively.
References
[1] S. Cardoso, R. Ferreira, P.P. Freitas, P. Wei, J.C. Soares, Appl. Phys. Lett. 76 (2000) 3792.
[2] T. Ochiai, N. Tezuka, K. Inomata, S. Sugimoto, Y. Saito, IEEE Trans. Magn. 39 (2003) 2797.
[3] Y. Fukumoto, K. Shimura, A. Kamijo, S. Tahara, H. Yoda, Appl. Phys. Lett. 84 (2004) 233.
[4] S. Cardoso, P.P. Freitas, C. De Jesus, P. Wei, J.C. Soares, Appl. Phys. Lett. 76 (2000) 610.
[5] M. Takiguchi, S. Ishii, E. Makino, A. Okabe, J. Appl. Phys. 87 (2000) 2469.
[6] S. S. P. Parkin, M. G. Samant, US Patent No. 6, 326, 637, 4 December 2001.
[7] M. Miyakawa, R.Y. Umetsu, K. Fukamichi, J. Phys.: Condens. Matter 13 (2001) 3809.
[8] M. Miyakawa, R.Y. Umetsu, K. Fukamichi, H. Yoshida, E. Matsubara, J. Phys.: Condens. Matter 15 (2003) 4817.
[9] Luc Thomas, A.J. Kellock, S.S.P. Parkin, J. Appl. Phys. 87 (2000) 5061.
[10] T.Y. Peng, C.K. Lo, T.C. Tien, S.Y. Chen, Y.D. Yao, Intermag ASIA 2005 Proceedings Paper BT-06, Nagoya, April 4–8, 2005.
ARTICLE IN PRESS
0 100 200 300 400 500 5 10 15 20 25 30 35 40 45 Annealing temperature (°C ) Coercivity Field ( Oe ) 0.0 0.2 0.4 0.6 0.8 1.0 -200 -100 0 100 200 -1.0 -0.5 0.0 0.5 1.0 d = 0 d = 2Normalized MOKE signal
H (Oe)
d = 0
d = 2 Squareness, S
Fig. 2. The temperature dependence of HC(’ and &) and S (K andJ)
in the Co90Fe10/Os (d nm)/Os20Mn80multilayer, which indicated by d ¼ 0
(dark symbol) and d ¼ 2 (open symbol). The annealing conditions are 30 min at 1 kOe external field. The inserted picture shows the hysteresis loop of samples with d ¼ 0 and 2 nm after 400 1C annealing.
0.0 0.5 1.0 1.5 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
400°C annealing at 1kOe for 30 min. Ta/CoFe/Os(d)/OsMn Os thickness, dOs (nm) Squareness, S 0 1 2 3 4 5 6 Normalized H C
Fig. 3. The squareness and normalized HC varies as a function of the
thickness of Os interlayer.
T.-Y. Peng et al. / Journal of Magnetism and Magnetic Materials 304 (2006) e50–e52 e52