Controlling the anisotropy and domain structure with

AIP ADVANCES 4, 027104 (2014)

Controlling the anisotropy and domain structure with oblique deposition and substrate rotation N. Chowdhury and S. Bedantaa

School of Physical Sciences, National Institute of Science Education and Research (NISER), IOP Campus, Bhubaneswar-751005, India (Received 30 October 2013; accepted 28 January 2014; published online 5 February 2014)

Effect of substrate rotation on anisotropy and domain structure for a thin ferromagnetic film has been investigated in this work. For this purpose Co films with 10 nm thickness have been prepared by sputtering with oblique angle of incidence for various substrate rotations. This method of preparation induces a uniaxial anisotropy due to shadow deposition effect. The magnetization reversal is studied by magneto-optic Kerr effect (MOKE) based microscope in the longitudinal geometry. The Co films prepared by rotating the substrate with 10 and 20 rpm weakens the anisotropy but does not completely give isotropic films. But this leads to high dispersion in local grain anisotropy resulting in ripple and labyrinth domains. It is observed that the substrate rotation has moderate effect on uniaxial anisotropy but has significant effect on C 2014 Author(s). All the magnetization reversal process and the domain structure. ⃝ article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/1.4865248]

Understanding and tailoring of anisotropy in magnetic thin films has been interesting due to its application in data storage devices and magnetoresitive devices. For optimizing the performance of these devices a high uniaxial anisotropy is required. In particular, in plane uniaxial anisotropy is required in magnetoresistive random access memory and head assemblies of the hard drive.1 From fundamental point of view anisotropy plays an important role in determining the magnetic properties of thin films. Various anisotropies such as magnetocrystalline, shape, surface and strain anisotropy can contribute to the net anisotropy in magnetic thin films. Anisotropy can be induced in a magnetic film by oblique deposition,2 applying magnetic field during deposition,3 post annealing (annealing in presence of magnetic field or magnetic annealing)4 etc. Effect of oblique deposition on magnetic anisotropy is well established.1, 2, 5–9 It is known that during oblique deposition due to steering10 and shadow effects2 an in plane uniaxial anisotropy is induced. Due to shadow effect columnar structures of grains grow with a tilt towards the deposition angle.8 Hence both the shape and strain2, 11 anisotropy contributes in the effective anisotropy. Also by changing the angle of incidence, easy axis orientation can be changed from perpendicular to parallel direction with respect to the incident flux.6 Lisfi et al. showed that the magnetic domain structure depends strongly on the growth angle on obliquely deposited Co thin film.12 However in some applications e.g. in microwave devices13, 14 and in fundamental studies it is desirable to overcome the anisotropy. One method is to rotate the substrate so that homogenous thickness and isotropic thin film is obtained. Kim et al. showed that coercive field increases with increasing the rotational speed of the substrate holder.15 But the detailed study on the change of micromagnetic domain structure and anisotropy of cobalt thin film with substrate rotation has not been studied systematically. In this paper we report the effect of oblique deposition and rotation of substrates on anisotropy as well as on the domain structure. The uniaxial anisotropy of cobalt thin film decreases when the

a Email: [email protected]

2158-3226/2014/4(2)/027104/6

4, 027104-1

⃝ C Author(s) 2014

027104-2

N. Chowdhury and S. Bedanta

AIP Advances 4, 027104 (2014)

substrate is rotated but it does not become completely isotropic even when the substrate rotation is increased to 20 rpm. Interestingly the domain structure changes from striped domains for 0 rpm to patch like domains superimposed on ripple state for 10 and 20 rpm cobalt thin films. This may arise due to dispersion in local anisotropy. Formation of ripple and labyrinth domains have been studied in details in past by several groups.16–20 Harte proposed theory for ripple domains in a polycrystalline ferromagnetic thin film with inhomogeneity in anisotropy arising due to several factors e.g. inhomogenous strain between crystallites.21 Cohen studied the effect of magnetic anisotropy dispersion systematically by annealing the Ni-Fe films and observed ripple and labyrinth pattern.17 McCord et al. showed that partial rotation of magnetization with the development of ripple and labyrinth domains were observed in exchange bias system of CoFe/IrMn.20 It was attributed to the anisotropy dispersion produced in the ferromagnetic layer caused by the small applied magnetic field during deposition.20 But in this paper we report that just by substrate rotation during the film deposition one can observe ripple domains and labyrinth pattern. In addition the magnetization reversal is partly also governed via coherent rotation of spins. This may be attributed to the dispersion of anisotropy in Cobalt thin films due to high misalignment in the local grain anisotropy resulting from the substrate rotation during deposition. The samples were prepared by sputtering in a UHV chamber manufactured by Mantis Deposition Ltd., UK. The angle between the substrate normal and the incident flux from the target is 30 degrees. The samples have the following layer structure: Al2 O3 (2 nm)/Co(10 nm)/Al2 O3 (2 nm) on Si(100) substrate. Three samples have been prepared which are (i) with 0 rpm [Sample A], (ii) with 10 rpm [Sample B] and (iii) with 20 rpm [Sample C], respectively. Cobalt was deposited using DC power of 56 watts and Al2 O3 was deposited using RF power of 120 watts at room temperature. The surface topography has been imaged by atomic force microscopy (AFM). These three samples show comparable roughness of 0.4 ± 0.1 nm. However it should be noted all the samples have an Al2 O3 capping layer. Therefore the information on the roughness of Co is not possible to obtain from these AFM measurements. But in-situ characterization of Co will give the information how roughness varies as a function of substrate rotation. However it is expected that roughness should gradually become less for samples prepared with higher substrate rotation. Also the grain size should decrease with increase in substrate rotation because of lesser time to grow. To study the anisotropy and micromagnetic domain structures, angle dependent static hysteresis measurement with simultaneous domain imaging was performed using longitudinal MOKE (LMOKE) microscope manufactured by Evico Magnetics Ltd. Germany by varying the angle (θ ) between the easy axis and the external magnetic field. Figure 1(a) shows the plot for coercivity vs. angle θ with respect to easy axis (EA) for all the samples. It can be seen from Figure 1(a) that sample A (without rotation) shows six fold anisotropy.5 In addition to a strong uniaxial magnetic anisotropy due to shadow effect, there exists weak intrinsic

FIG. 1. (a) Coercivity vs. θ for Sample A (red solid circles), Sample B (blue solid triangles) and Sample C (green solid stars). (b) Mr /Ms vs. θ for the samples showing the uniaxial anisotropy behaviour.

027104-3

AIP Advances 4, 027104 (2014)


TABLE I. Comparison of coercive field for easy and hard axis and their difference for sample A (0 rpm), B (10 rpm) and C (20 rpm).

Sample 0 rpm (sample A) 10 rpm (sample B) 20 rpm (sample C)

Hc for easy axis ! E A" HC in mT 9.057 4.767 7.226

Hc for hard axis ! H A" HC in mT 4.7 3.423 6.492

!

HCE A − HCH A in mT

"

4.357 1.344 1.11

FIG. 2. Comparison of hysteresis loops measured along easy and hard axis by LMOKE at room temperature, for sample A, B, and C, respectively.

magnetocrystalline anisotropy. The uniaxial anisotropy behaviour is also present in Sample A and B as seen in Figure 1(a) shown by the green color stars and blue color triangles. The anisotropy comparison can be better visible in Figure 1(b) which shows the plot for angular dependence of squareness Rs , defined as Mr /Ms where Mr = remament magnetization and Ms = saturation magnetization. Here the existence of clear easy and hard axis for sample A can be inferred. It can be seen that the change in squareness (Figure 1(b)) for sample B and C is small. This implies that even though we have rotated the sample to get isotropic film, a very weak uniaxial anisotropy still exists. It should be noted that a peak near hard axis (HA) is observed for all three samples. This peak arises due to the misalignment of local grain anisotropy.22 Idigoras et al. showed experimentally and theoretically that the height and width of the peak increases with the increase in disorder. From Figure 1(b), it can be seen that the height and width of the peak increases abruptly from 0 rpm to 10 rpm but the change is minimal when rotation is increased from 10 to 20 rpm. Quantitatively this can EA HA be from Table I which shows the difference " ! (i.e." HC − HC ) of coercivity along hard axis ! realised HCH A peak from the coercivity along easy axis HCE A . It is clear that lesser the difference more is the height of the peak which is a signature of more disorder. Hence from Table I it is evident that increase in substrate rotation during deposition results in misalignment and dispersion in local grain anisotropy. It should be noted that this dispersion in anisotropy due to rotation has marked effect on domain structure which will be shown later in this paper. Figure 2 shows a direct comparison of hysteresis loops for sample A, B, and C respectively for easy axis and hard axis. Figure 2 provides a direct visualization of squareness i.e Mr /Ms for all three samples. It can be seen from graphs that the squareness of hard axis increases with rotation. This implies that sample C has weaker anisotropy energy. Hence it may be expected that sample C (20 rpm) is more isotropic. It can also be seen that the anisotropy field Hk is comparable to coercive field for sample B and sample C which are characteristics of an inverted films.16 The value of Hk can be deduced from the saturation field of hard axis.23, 24 Domain structure for all the samples were studied by performing longitudinal Kerr microscopy obtained during hysteresis loop measurements. It was observed that reversal for sample A along EA, 30o and 60o (Figure 3) were corroborated by domain wall motion and nucleation. But along HA no domains are observed during magnetization reversal for Sample A. In the later case the magnetization reversal was only due to coherent rotation of spins. For sample B and sample C we observe ripple domains superimposed on patch like domains for easy axis. As we increase the angle

027104-4


AIP Advances 4, 027104 (2014)

FIG. 3. Domain images for sample A, B and C are shown for θ = 0o (EA), 30o , 60o and 90o (HA), respectively. The images are taken near the coercive fields for the respective angles and the samples. The inset in Figure 3(i) shows the zoomed image (60 × 60 µm2 ) for the marked region. The inset in figure (j) and (k) shows high resolution image (57 × 52 µm2 ).

FIG. 4. (a) Labyrinth domain images for Sample B and C taken by Kerr microscope for θ = 30o are shown in (a) and (b) respectively (enhanced). [URL: http://dx.doi.org/10.1063/1.4865248.1] [URL: http://dx.doi.org/10.1063/1.4865248.2]

from easy axis labyrinth domains are observed for sample B and sample C which is characteristic of inverted film. The inset in Figure 3(i) shows the zoomed image for the area marked. Inset in Figure 3(j) and 3(k) shows high resolution images for 30 and 60 degrees respectively. The images clearly show the labyrinth pattern. The origin of such labyrinth domains in these samples (A and B) arises due to local dispersion in anisotropy in cobalt film due to substrate rotation. For better visualization of the formation of labyrinth domains and magnetization reversal see video 1 and 2 for sample B and C, respectively. These videos are movies showing magnetization reversal along one branch of hysteresis loop for these samples measured along 30o w.r.t. EA. The still images from these videos are shown in Figure 4(a) and 4(b), respectively. The effect of local anisotropy dispersion can also be seen for these three samples from the Kerr microscope images taken at remanence of hard axis hysteresis loop as shown in Figure 5. No

027104-5


AIP Advances 4, 027104 (2014)

FIG. 5. Kerr images for hard axis at remanence for Sample A (0 rpm), Sample B (10 rpm), and Sample C (20 rpm) labeled as image (a), (b) and (c) respectively.

FIG. 6. (a) Hysteresis measured by Longitudinal Kerr microscope for sample C for θ = 30o . (b) to (i) are domain images taken at fields marked by point 1 to 8 in (a), respectively.

magnetic domains are visible for sample A as seen in Fig 5(a). The reason for this is that sample A has the least dispersion of anisotropy. But fine domains are visible for Sample B (Fig. 5(b)) and C (Fig. 5(c)), which hints the presence of high dispersion in local anisotropy with rotation. Idigoras et al. showed that for a Co thin film, at hard axis near remanence, fine domains can be observed due to the misalignment of local grain anisotropy.22 Figure 6(a) shows the hysteresis and 6(b)–6(i) shows the images for reversal of 30 degrees wrt EA for sample C. It can be seen that initially the magnetization reverses by partial rotation giving rise to longitudinal magnetization ripple domains (Figure 6 and 6)16 to reduce the stray field energy. On increasing the magnetic field labyrinth pattern was observed. Finally with increase in field, reversal is completed by domain wall motion as higher fields are required to move the domain walls formed. It can be seen from Figure 6(h) that domain walls are present even at higher fields. Similar magnetization reversal was observed for both samples B and C, when magnetic field was applied for any angle other that easy axis. This observation suggests the existence of high dispersion in local anisotropy as observed in inverted films.16, 20 In our case we attribute the variation in local in anisotropy to the misalignment of gains due to substrate rotation. In summary we have studied the effect of substrate rotation on the anisotropy and magnetic domain structure on Co (10 nm) thin film deposited by sputtering. The sample without rotation showed a high uniaxial anisotropy which was induced due to the oblique deposition along with weak magnetocrystalline anisotropy. Rotation of substrate by 10 and 20 rpm weakens the anisotropy but does not completely give isotropic properties. The change in anisotropy from 10 to 20 rpm is minimal. However interestingly for samples with 10 and 20 rpm substrate rotation, magnetization reversal at low fields proceeds via partial rotation where small ripples are formed. On increasing the field complicated labyrinth domain structure was observed and finally reversal is completed by domain wall motion. This is attributed to the inhomogeneous dispersion of anisotropy in Cobalt thin films due to high misalignment in the local grain anisotropy which results from substrate rotation during deposition.

027104-6


AIP Advances 4, 027104 (2014)

Our results show that by rotating the substrate during deposition one can control the anisotropy moderately. However the little change in anisotropy has a severe consequence on the domain structure of the ferromagnet. This approach of controlling the domain microstrucutre by substrate rotation may have applications in future magnetic devices based on domains. ACKNOWLEDGMENTS

We thank National Institute for Science Education and Research (NISER) and the Department of Atomic Energy of the Government of India for the financial support. We would like to acknowledge Mr. Sovakara Singh for his help in Kerr microscopy experiments. We are grateful to Mr. Vanarajsinh J. Solanki and Prof. Shikha Varma for AFM measurement. We are also thankful to Dr. G.S.Babu for discussion. 1 Y.

Fukuma, Z. Lu, H. Fujiwara, G. J. Mankey, W. H. Butler, and S. Matsunuma, J. Appl. Phys. 106, 076101 (2009). O. Smith, M. S. Cohen, and G. P. Weiss, J. Appl. Phys. 31, 10 (1960). 3 S. Bedanta, T. Eimuller, W. Kleemann, J. Rhensius, F. Stromberg, E. Amaladass, S. Cardoso, and P. P. Freitas, Phys. Rev. Lett. 98, 176601 (2007). 4 M. Takahashi and T. Kono, J. Phys. Soc. Jpn. 15, 936 (1960). 5 Ya-Peng Fang, Wei He, Hao-Liang Liu, Qing-Feng Zhan, Hai-Feng Du, Qiong Wu, Hai-Tao Yang, Xiang-Qun Zhang, and Zhao-Hua Cheng, Appl. Phys. Lett. 97, 022507 (2010). 6 M. T. Umlor, Appl. Phys. Lett. 87, 082505 (2005). 7 T. G. Knorr and R. W. Hoffman, Phys. Rev. 113, 1039 (1959). 8 Y. Hoshi, E. Suzuki, and M. Naoe, J. Appl. Phys. 79, 4945 (1996). 9 Paritosh and D. J. Srolovitz, J. Appl. Phys. 91, 1963 (2002). 10 S. V. Dijken, L. C. Jorritsma, and B. Poelsema, Phys. Rev. Lett. 82, 4038 (1999). 11 J. L. Bubendorff, S. Zabrocki, G. Garreau, S. Hajjar, R. Jaafar, D. Berling, A. Mehdaoui, C. Pirri, and G. Gewinner, Europhys. Lett. 75, 119 (2006). 12 A. Lisfi and J. C. Lodder, Phys. Rev. B 63, 174441 (2001). 13 J. B. Youssef, N. Vukadinovic, D. Billet, and M. Labrune, Phys. Rev. B 69, 174402,(2004). 14 H. Kijima, S. Ohnuma, and H. Masumoto, IEEE Transactions on Magnetics 47, 3928 (2011). 15 J. Y. Kim, I. O. Hwang, H. C. Hong, and D. H. Shin, “Proceedings of magneto-optical recording international symposium 96,” Magn. Soc. Jpn. 20, 165, supplement no s1 (1996). 16 S. Methfessel, S. Middelhoek, and H. Thomas, J. Appl. Phys. 32, 1959 (1961). 17 M. S. Cohen, J. Appl. Phys. 34, 1841 (1963). 18 D. O. Smith and K. J. Harte, J. Appl. Phys. 33, 1399 (1962). 19 H. W. Fuller and M. E. Hale, J. Appl. Phys. 31, 238 (1960). 20 J. McCord, R. Sch¨ afer, R. Mattheis, and K.-U. Barholz, J. Appl. Phys. 93, 5491 (2003). 21 K. J. Harte, J. Appl. Phys. 39, 1503 (1968). 22 O. Idigoras, A. K. Suszka, P. Vavassori, P. Landeros, J. M. Porro, and A. Berger, Phys. Rev. B 84, 132403 (2011). 23 M. T. Johnson, M. T. Johnsony, P. J. H. Bloemenzx, F. J. A. den Broedery, and J. J. de Vriesz, Rep. Prog. Phys. 59, 1409 (1996). 24 Chiao-Sung Chi, Bo-Yao Wang, Way-Faung Pong, Tsung-Ying Ho, Cheng-Jui Tsai, Fang-Yuh Lo, Ming-Yau Chern, and Wen-Chin Lin, J. Appl. Phys 111, 123918 (2012). 2 D.