Magnetoresistance and magnetic properties of Co/Ir multilayers on MgO(110) substrates H. Yanagihara Institute of Applied Physics, University of Tsukuba, Tsukuba 305, Japan K. Pettit IBM Almaden Research Center, San Jose, California 95120-6099 M. B. Salamon Department of Physics, University of Illinois, Urbana, Illinois 61801 Eiji Kita Institute of Applied Physics, University of Tsukuba, Tsukuba 305, Japan S. S. P. Parkin IBM Almaden Research Center, San Jose, California 95120-6099 A Co 12 Å /Ir 16 Å multilayer deposited on MgO 110 is found to have a strong uniaxial anisotropy, comparable to bulk hcp-Co. Experimental results suggests the crystallographic relation of 100 001 Co/ 110 001 Ir/ 110 001 MgO. A remarkable difference is seen between easy and hard axis magnetization processes, due to competition between the in-plane anisotropy and the exchange coupling. Considering the characteristics of the MH loops and magnetoresistance curves, we simulate the magnetization and magnetoresistance processes numerically. The results of our simulation indicate the existence of biquadratic coupling in Co/Ir 110 . © 1997 American Institute of Physics. S0021-8979 97 71208-8 I. INTRODUCTION oxidation. The ML structure was confirmed by medium- Magnetic multilayers MLs have been attractive be- angle x-ray diffraction with Co K radiation 1.79 Å . cause of their enormous potential for applications. Various Magnetization curves MH loops were measured with a magnetic characteristics have been reported in magnetic ML SQUID magnetometer. MR measurements were carried out systems having both magnetic anisotropy and magnetic cou- using a conventional dc four-contact method. The sample for pling between layers.1,2 Of particular significance is the ap- MR measurements was cut into a strip shape along pearance of giant magnetoresistance GMR in ML systems, MgO 001 . Magnetic fields up to 15 kOe were applied along which is associated with the antiparallel configuration of MgO 001 and MgO 11¯0 directions. Both MH measure- magnetization between adjacent layers. Such configurations ments and MR measurements were performed at 10 K and are most readily realized by an antiferromagnetic coupling room temperature. In-plane magnetic torque measurements between adjacent layers. In recent years, the existence of /2 were carried out with an automatic torque magnetometer at biquadratic interlayer coupling has been reported both room temperature in a magnetic field of 14 kOe. The torque- experimentally3 and theoretically.4 The presence of strong measurement sample was cut into a rectangle 2.9 2.2 mm2 biquadratic interactions could lead to a greater variety of in order to reduce demagnetization effects; the same sample magnetization configurations, and consequently, to a large was used for MH measurements. variety of magnetoresistance MR curves. In this article, we report strong in-plane magnetic III. RESULTS AND DISCUSSION uniaxial anisotropy, comparable to that of bulk hcp-Co An x-ray diffraction pattern of the Co/Ir ML was ob- Ku 4.6 106 erg/cc , in Co/Ir MLs grown on MgO 110 . tained in reflection geometry with the scattering vector nor- Using a model taking account of exchange energy up to a mal to the surface. Peaks due to an artificial structure were biquadratic term, the model can reproduce the magnetic-field observed around the hcp-Co 100 and the fcc-Ir 220 Bragg dependence of the magnetization and MR, and hence sup- peaks; however, there was no peak around the angle corre- ports the existence of a biquadratic term in the Co/Ir 110 sponding to the Bragg peak of the fcc-Co 111 . The result ML. supports the crystallographic relation of 100 hcp-Co/ 110 Ir/ 110 MgO. II. EXPERIMENT Figure 1 shows magnetization curves at RT and at 10 K after subtraction of the substrate diamagnetic background. A Co/Ir ML was deposited with a magnetron sputtering With the magnetic field applied along MgO 001 , a stairca- system onto Fe 5 Å /Pt 50 Å /Ir 50 Å as a buffer layer de- selike open hysteresis loop was obtained. On the other hand, posited on a MgO 110 single-crystal substrate. We chose a when the field was applied along the MgO 11¯0 direction, Co layer thickness of 12 Å and Ir layer thickness of 16 Å there was little hysteresis and the magnetization changed corresponding to the second maximum of antiferromagnetic monotonically with a saturation field of 19 kOe 22 kOe coupling.5 The ML is composed of 100 repeats of Co 12 at 10 K . The in-plane torque profile suggests that the anisot- Å /Ir 16 Å , with a 34 Å Ir cap layer deposited to prevent ropy of this sample is uniaxial in the film plane; the easy and J. Appl. Phys. 81 (8), 15 April 1997 0021-8979/97/81(8)/5197/3/$10.00 © 1997 American Institute of Physics 5197 Downloaded¬06¬Mar¬2001¬to¬148.6.169.65.¬Redistribution¬subject¬to¬AIP¬copyright,¬see¬http://ojps.aip.org/japo/japcpyrts.html FIG. 2. Magnetoresistance vs applied field along MgO 001 circles and MgO 11¯0 diamonds . a and b are measured at RT and 10 K, respec- FIG. 1. Magnetization vs applied field along the in-plane two directions: tively. The field is applied up to 15 kOe. MgO 001 circles and MgO 11¯0 diamonds . a RT, b 10 K. The inset shows MH curves up to 55 kOe at each temperature. model. Assuming a coherent rotation model of the magneti- hard axes correspond to MgO 001 and 11¯0 directions, re- zation process, without domain motion or pinning effect, we spectively. The torque curves were dependent on the external can find the in-plane configuration of magnetic moments by fields, but saturated above an external field of 13 kOe. We minimizing the reduced energy 2E/M ,6 estimated the uniaxial anisotropy (Ku) as 5 106 erg/cc from the amplitude of torque. The value is close to that of H1 hcp-Co K , Hk sin2 cos2 cos2 sin2 u 4.6 106 erg/cc and we presume that the hcp- 2 2 2 Co 001 lies along the MgO 001 . Supposing that only Co atoms are responsible for the magnetic moments of the ML, H cos 2 cos , 1 we find the magnetization (M) to be 940 emu/cc at RT, and 2 cos2 2H cos 2 1110 emu/cc at 10 K. This distinctly lower value of M com- pared with bulk hcp-Co 1440 emu/cc suggests mixing or where is an angle between the magnetic moments of Co alloying at interfaces between Co and Ir layers. adjacent layers, and are the angles of the net magnetiza- The MR curves at RT and 10 K are shown in Fig. 2. The tion and applied magnetic field measured from the easy axis, MR ratio [ R/R(Hs)] is only 0.3% at room temperature and and t is the Co layer thickness. The energy terms are ex- 1.2% even at 10 K. Parabolic MR curves with little hyster- pressed as fields; that is, Hk 2Ku/M, H1 4J1/tM, esis are obtained when the field is applied along hard axis. H2 4J2/tM. Because the above configurations cannot be Along the easy axis, however, the MR curves exhibit squared solved analytically, in general, we use an iterative plateaus and large hysteresis of several kOe. The MR pro- calculation.9 The net magnetization parallel to the field is cesses are quite consistent with the MH curves at every tem- proportional to cos /2 cos , and in the simplest MR perature, indicating that the resistance mechanism is strongly picture, R(H) is similarly proportional to cos . We re- dependent on the magnetization. When H 15 kOe at RT, gard 100 bilayers as infinite repeats of bilayers,7,10 and ne- there is little difference between hard- and easy-axis MR, glect coupling-energy differences that depend on layer num- and this implies that each magnetic configuration is equiva- bers. lent and saturated to a nearly parallel configuration. The Some special magnetic characteristics, such as saturation saturated torque curves will be given by the same mecha- fields and remanence, can be expressed analytically using the nism. transformed fields above. The saturation fields along easy Here, in order to investigate the origin of these peculiar and hard axes are Hs,e H1 2H2 Hk , and Hs,h curves of MH and MR, we consider the free energy (E) H1 2H2 Hk , respectively. The condition for finite re- using a model6,7 containing in-plane uniaxial anisotropy manence along the easy axis is 2H2 Hk H1 0. When the (Ku), first-order bilinear exchange J1 , second-order bi- easy axis MH curve possesses a finite remanence, the con- quadratic exchange4 J2 , and magnetostatic energy. The figuration is given by cos 1[(Hk H1)/2H2], 0. Simi- Co/Ir MLs have out-of-plane anisotropy as strong as the larly, the condition for no remanence along the hard axis is uniaxial anisotropy of hcp-Co due to the interface anisotropy;8 however, we ignore this in order to simplify the 2H2 Hk H1 . 2 5198 J. Appl. Phys., Vol. 81, No. 8, 15 April 1997 Yanagihara et al. Downloaded¬06¬Mar¬2001¬to¬148.6.169.65.¬Redistribution¬subject¬to¬AIP¬copyright,¬see¬http://ojps.aip.org/japo/japcpyrts.html In the hard axis magnetization process, there is no hys- teresis and the relation /2 is kept in any field. With the applied field along the easy axis, the MR curve see the inset of Fig. 3 has a wide plateau with large staircaselike hyster- esis similar to the experimental curves. Although there is disagreement between the simulated MR curve and the ex- perimental data around zero field along the hard axis, most characteristics of the experimental results could be repro- duced. Decreasing the easy axis field from saturation, the spins fan out from saturation 0, 0 to 0, 0 and then make a discontinuous spin-flop transition at H 0 to FIG. 3. MH curves obtained by iterative simulation for two cases. The solid zero net moment , /2 . The experimental MH line and dotted line are MH curves for easy and hard axes, respectively, loops, however, show a finite remanence. This AF configu- obtained for case II: H1 0.99, H2 0.79, and Hk 0.83. The circles and ration changes to , /2 in weak negative fields and diamonds are obtained for case I: H1 3.26, H2 0, and Hk 0.14. These then switches to negative M 0, at the switching values are determined from the relations, that is, Hs,h/Hsw 5 and Eq. 2 . The inset shows MR curves for case II. field. For H Hsw , the magnetization process is reversible. From the above discussion, the H2 term is seen to be neces- sary to reproduce the ratio of Hs,h to Hsw , supporting the The switching field is defined as the magnetic field along the existence of the strong biquadratic coupling in this ML. easy axis at which the and /2 configuration be- comes unstable in an increasing field, and is given by IV. CONCLUSION H We have found that Co 12 Å /Ir 16 Å MLs deposited sw Hk Hk H1 2H2 for Hk H1 2H2 . 3 on MgO 110 show strong anisotropy, comparable to bulk The above equations are necessary conditions because the hcp-Co. Experimental results suggests the crystallographic changes of the submagnetization configurations are strongly relation of 100 001 Co/ 110 001 Ir/ 110 001 MgO. dependent on the initial state. Within the coherent rotation model, strong biquadratic cou- Considering the hard axis MH curve, we conclude that pling in Co/Ir 110 is required to reproduce the experimental the zero-field configuration must have H 0 , because MH and MR results. there is no remanence, and that is always /2. The maxi- mum value of the easy axis MR is close to the zero-field MR ACKNOWLEDGMENTS value obtained along the hard direction so that the magnetic configuration of the plateaus on the easy axis MR curve must We are grateful to Professor A. Tasaki of University of also have H 0 . The plateau in the MR curves and the Tsukuba. This research was partly supported by DOE Grant ratio of H No. DEFG02-91-ER45439 at the University of Illinois. s,h to Hsw 5 at 10 K give us a key to determine appropriate values of the parameters. Figure 3 shows simu- 1 lation results of MH for the two cases which satisfy the W. Folkers, J. Magn. Magn. Mater. 94, 302 1991 . 2 B. Dieny and J. P. Gavigan, J. Phys., Condens. Matter 2, 187 1990 . above relation Hs,h/Hsw 5 and Eq. 2 . In case I we set 3 M. Ru¨hrig, R. Scha¨fer, A. Hubert, R. Molser, J. A. Wolf, S. Demokritov, H2 0 while in case II we include a biquadratic term. We and P. Gru¨nberg, Phys. Status Solidi A 125, 635 1991 . chose the fields so as to make H 4 J. C. Slonczewski, Phys. Rev. Lett. 67, 3172 1991 . s,h 3.4 in both cases. It is 5 clear that we could not reproduce the experimental MH S. S. P. Parkin unpublished . 6 H. Fujiwara and M. R. Parker, J. Magn. Magn. Mater. 135, 23 1994 . curves in case I H 7 1 3.26 and Hk 0.14 ; however, quite N. S. Almeida and D. L. Mills, Phys. Rev. B 52, 13504 1995 . similar MH curves to the experimental data are obtained in 8 F. J. A. den Broeder, W. Hoving, and P. J. H. Bloemen, J. Magn. Magn. case II H Mater. 93, 562 1994 . 1 0.99, H2 0.79, and Hk 0.83 . Each parameter 9 should be optimized to obtain better agreement between ex- K. Pettit, M. B. Salamon, S. S. P. Parkin, and S. Gider to be published . 10 B. Dieny, J. P. Gavigan, and J. P. Rebouillat, J. Phys., Condens. Matter 2, perimental and simulation results. 159 1990 . J. Appl. Phys., Vol. 81, No. 8, 15 April 1997 Yanagihara et al. 5199 Downloaded¬06¬Mar¬2001¬to¬148.6.169.65.¬Redistribution¬subject¬to¬AIP¬copyright,¬see¬http://ojps.aip.org/japo/japcpyrts.html