<<

Granular Films W. D. Doyle, Chairman

Magnetic properties of metallic Co- and Fe-based granular alloys Gang Xiao and Jian-Qing Wang Department of Physics, Brown Univcrsi~, Providence, Rhode Island 029.12 We have studied the magnetic properties of Co-Ag and Fe-Ag granular alloys made using vapor-quenching techniques and thermal annealing. Magnetic coercivity (H,.j and remanence can be controlled over a large range by varying annealing temperature and volume fraction. A large H, on the order of 2 kG has been obtained in the Co-Ag system. We have investigated magnetic anisotropy, the effect of , and coale.scence in these nanostructured materials.

Magnetic granular have received considerable at- state properties, most of our measurements were carried out tention in the past few years.l-lu In these composite materi- at low temperatures. The saturated magnetization of the mag- als, ultratine magnetic of a few nanometers in size netic component (Fe or Coj is approximately equal to that of are embedded in a metallic or insulating matrix by certain the bulk. In addition to magnetic measurements, we have synthesis processes. Because of their microstructure and the also performed magnetoresistance and measure- tunability in materials and geometrical parameters, these ma- ments. Both Fe&g and Co-Ag show GMR effect and ex- terials possess different and sometimes enhanced properties traordinary Hall effect.s-‘a The Hall resistivity can be de- when compared with their bulk counterparts. In particular, a scribed by pXY= p.!! + R,M. In this relation, p-z, is due to the giant magnetic coercivity (H,) has been achieved in Fe-SiQ, ordinary Hall effect and is linear in magnetic field. The sec- granular films,’ with H, enhanced by as much as three orders ond term, due to the extraordinary Hall effect,13 results from of magnitude over the bulk value. The large H, and magne- the left-right asymmetry in electron scattering in magnetic tization of the material make it suitable for application in systems. Because of its linear relationship with M, the field magnetic recording. It has also been discovered that metallic dependence of pxY after removing p,“,, shows a one-to-one granular alloys, such as Co-Cu, Co-Ag, and Fe-Ag, exhibit correspondence with the magnetic curve. There- giant magnetoresistance effects (GMRj,h-r” with a magni- fore, when the sample plane is perpendicular to the external tude rivaling the best multilayers with GMR. These develop- magnetic field, we can consistently obtain remanent magne- ments in granular solids call for more systematic studie.s of tization and coercivity from either Hall effect or magnetiza- their magnetic properties. The understanding of the novel tion measurements. It is noted that it is more efficient and GMR effect also requires the exploration of the underlying economical to measure the Hall effect than the magnetization magnetic state which strongly correlates with the behavior of using a SQUID magnetometer. GMR. In this work, we present a study on two transition Figure 1 shows the magnetization curves at T=5 K for metal granular solids, Co-Ag and Fe-Ag, both of which have two representative as-prepared samples, Co2&gs0 and been shown to have GMR effect.*-‘” The main focus here is Fez&gsr,, with H perpendicular and parallel to the sample how the thermal treatment and volume fraction of the mag- plane (both samples are specifie.d by volume fractions of netic particles affect the magnetic properties. We have fabricated Co-Ag and Fe-Ag granular films by their components). The initial susceptibility measurement in- taking advantage of the immiscibility between Co (or Fe) dicates that the superparamagnetic transition temperature is and Ag in alloy formation. High vacuum sputtering from a about 20-30 K. Therefore, at 5 K, the magnetic moment cold-pressed composite target yields a -separated film vectors of the particles are fixed in random directions. As with magnetic particles precipitating from the Ag matrix. shown in Fig. 1, the Co&gyO sample has a much larger H, The particle size of an as-sputtered sample is very small, on than the Fe,dgsO sample. Most interestingly, the easy axis the order of 1-2 nm. We have also fabricated samples using tends to be out of plane for Co-Ag, whereas it is in plane for codeposition technique and obtained similar physical proper- Fe-Ag. ties. Thermal annealing is an effective means to enlarge the The anisotropy in Co-Ag is most likely caused by the particle size. We have prepared a series of samples with dif- crystalline and shape anisotropies of the Co particles. The ferent thermal annealing as well as varying volume fraction. maximum H, which could result from crystalline or strain Phase separation has been confirmed by using a combination anisotropy is about ~‘300 (fee) or 600 G respectively.‘” The of analysis, i.e., transmission electron microscopy (TEM), large perpendictdar H, value of about 2 kG seen in Fig. 1 x-ray diffraction, and magnetic susceptibility measurement. can be accounted for by magnetocrystalline anisotropy. The Detailed results will be presented elsewhere.” Analysis of perpendicular H,: in Co-Ag is larger than the parallel H, . To phase separation for the Co-,4g system can also be found in account for the observed perpendicular H, in Co-Ag, it is Ref. 12. reasonable to conjecture. that the Co particles are slightly We used a superconducting quantum interference device enlongated along the growth direction. Another possible (SQUID) magnetometer to measure the magnetic hysteresis source of anisotropy is from the interface between the mag- curve of our samples. Since we are interested in the ground netic particles and the surrounding Ag matrix, Because of the

6604 J. Appl. Phys. 75 (lo), 15 May 1994 0021-8979/94/75(iO)/660413/$8.00 0 1994 American Institute of Physics

Downloaded 12 Jul 2001 to 128.148.60.168. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp 1 ! I I 4 CozoAgso1 ~~~~‘-7 0 (3.5 6 YJ m -‘\ 1000 0.0 \

-0.5 ( i,l 0 I I cc 0 100 200 300 400 500 600 i-z ‘I_ ==l,i? ‘I’c5K P Annealing Temperature (“Cj E k& 1 .o Fe,,&,, .A=== 400 I I L I I 1 cl.5 ~~ Fe20A4i0i 0.40 ! :// “ x 300 F / oii "0 0,f/ -o.35 c m 200 d4 - 0.30 3 =i-!,5 _&!’ ? . 100 .o-D-o-o <--::-:.-.--+ ’ -0.25 - 1 .o T=5K 0 I I 1 ---L---L -1 0.20 i-. ..---.- L ---.I_.~~~ 0 100 200 300 400 500 _ 1.R -0 5 0.0 0.5 1.0 H(T) Annealing Temperature (OC)

FIG 1. hlagnetic hysteresis curves in perpendicular and parallel field FIG. 2. Coercivity H, vs annealing temperature T* for Co&gsu (upper cirmfigmations for CO&&~ (upper panel) and Fe&g,, (lower panel) panel) and Fe&g*,, (lower panel). Also shown in the lower panel is the at T-5 R. remanence M,IM, for Fe&g8,, . T=5 K. small size of the particles, interfacial effect could play an and particle morphology. Thermal treatment is an effective important role. Nevertheless, without the proposed growth means to induce phase separation, to enlarge particle size, texture, it is difficult to imagine how the interface anisotropy and to reduce crystalline disorder and stress in granular ma- could cause the observed perpendicular anisotropy under the terials. In Fig. 2, we present the results of annealing on H, influence of the large demagnetization field of the thin film. for Co2,figs0 and Fe&gss. Also included in Fig. 2 is an The H, for the Fe-Ag system is of the magnitude of 300 annealing temperature (T,4j dependence of M,IM, for G. While it is considerably larger than that of a polycrystal- Fea&gss. Thermal annealing was done in high vacuum line Fe film (a few tens of Gj, it is much below the limit of (1X 10e7 Torr) at a chose.nTA for 15 min, followed by natu- crystalline (540 G) and strain (600 G) anisotropy for single- ral cooling. From the TEM micrograph, it was found that the domain particles.‘” The shape anisotropy in Fe can provide a Co particle size increases from about 20 to 130 A as T, maximum H, of 10.7 kc,‘” although experimentally such a reaches 605 “C. X-ray diffraction revealed that the Co par- large N, in single-domain Fe particles has never been dis- ticles have fee structure and are [ 1111 textured.s covered. In Fe-SiO, granular films, a maximum H, value of As shown in Fig. 2, H, for Co&Igsa drops by a large about 3 kG has hcen observed,! where the Fe particle size is amount once TA exceeds 250 “C. This variation of H, has about 50 A. We have obtained the particle sizes in our been verified more than once. Such a drop in Ii, is desirable FezaAgsosamples using both TEM and analysis of superpara- because it tends to reduce the saturation field and hysteresis magnetic behavior.” The FeZsAgs sample used for Fig. 1 in GMR effect, which was indeed observed.” There are a few has an average particle size of 29 A. Thermal annealing at possible causes for this drop. First of all, it is not due to the 300 “C enlarges the particle size to 51 A, but the H, was enlarged particle size which may induce multidomain forma- found to decrease to about 200 G. The comparison between tion and reduce H, . Within our Te4range, the Co particle size Fe-Ag and Fe-SQ granular systems shows convincingly is always below the critical single-domain particle size.‘” that particle-matrix interface could be one of the important Most likely, as the particle size grows, the original growth factors in the magnetic anisotropy of Fe-based systems. texture. diminishes, which reduces the particle shape anisot- Among the magnetic parameters of a ferromagnetic ropy. Another possible cause is that annealing makes phase , the ground state magnetization is basically an intrinsic separation more complete and tends to reduce crystalline dis- parameter, whereas Hi and remanence M,/iGI, are primarily order and stress. Both effects will reduce pinning force for extrinsic parameters. They are sensitive to disorder, stress, magnetization, and, therefore, lower H, .

J. Appl. Phys., Vol. 75, No. IO, 15 May 1994 G. Xiao and J.-Q. Wang 6605

Downloaded 12 Jul 2001 to 128.148.60.168. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp for the Fe~uAglOo-xas-sputtered system. In the low x region (it4and Co-based3y” 3J. R. Chiidress and C. L. Chicn, J. Appl. Phys. 70, 5885 tt1991). granular solids. This is in sharp contrast with the behaviors “?i. S. Zhang. S. H. Lieu, R. J. DeAngelis, K. W. Lee, C. R Reed, and A. of our Fe-Ag and Co-Ag systems, where moderate annealing Nazareth, J. Appl. Phys. 69, 5273 jlYYlj. “S. H. Lieu, S. MaIhotra, Z. S. Shan, D. J. Sellmyer, S. Nafis, J. A. Wool- depressesH, . The difference is due to the fact that the ear- lam, C. P Reed, R. J. DeAngelis, and G. M. Chow, J. Appl. Phys. 70,5882 lier studies were performed at room temperature, whereas (1991). ours are at low temperature. At T=300 K, thermal relaxation ‘A. E. Berkowitz, J. R. Mitchell, hf. J. Carey, A. P. Young, S. Bang, F. E. reduces H, severely for moderately annealed samples where Spada, F. T, Parker. A. Hutten, and G. Thomas, Phys. Rev. Let. 68, 3745 (199.2). the magnetic particles are small in size. In fact, any super- ‘3. Q. Xiaq J. S. Jiang, and C. L. Chien, Phys. Rev. L&t. 68, 3749 (1992). paramagnetic sample has zero hysteresis. For samples an- ‘P. Xiong, F. Ciao, J. Q. Wang, J. Q. Xiao, J. S. Jiang, and C. L. Chien, nealed at high temperatures, and therefore, having large par- Phys. Rev. L&t. 22, 3220 (1992). ticle size, thermal agitation is less important, and H, appears ‘G. Xiao, J. Q. Wang, and P. xiong, Appl. Phys. Lett. 62,420 (1993). “‘J. Q. Wang, P ?iiong, and G. Xiao, Phys. Rev. B 47, 8341 (1993). higher. In order to eliminate the effect of thermal relaxation, “J. Q. Wang and 0. Xiao, Phys. Rev. B 49, 3982 i1994). measurements need to be done at low temperatures. “5. Q. Xiao, J. S. Jiang, and C. L. Chicn, Phys. Rev. B 46, 9266 (1992). Next we turn our attention to the. effect of particle vol- r3L. Berger and G. Bergmann, in The Hull @$x-l nnd Its &~Zications, ume fraction (ru) on magnetic properties. Here, x was deter- edited by C. L. Chien and C. R Westgate (‘Plenum, New York, 1979), p. 554. mined from the composition and the of each “The Physica Principh o~n-iug~ze?ism,by A. H. Morrish [Kriegler, Mala- component. Figure 3 shows H, and IVJM, as functions of x bar, 1983), p, 332.

6606 J. Appl. Phys., Vol. 75, No. IO, 15 May 1994 G. Xiao and J.-Q. Wang

Downloaded 12 Jul 2001 to 128.148.60.168. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp