Switching of magnetic easy-axis using crystal orientation for large perpendicular coercivity in CoFe2O4 thin film

Abstract

Perpendicular magnetization and precise control over the magnetic easy axis in magnetic thin film is necessary for a variety of applications, particularly in magnetic recording media. A strong (111) orientation is successfully achieved in the CoFe₂ O₄ (CFO) thin film at relatively low substrate temperature of 100 °C, whereas the (311)-preferred randomly oriented CFO is prepared at room temperature by the DC magnetron sputtering technique. The oxygen-deficient porous CFO film after post-annealing gives rise to compressive strain perpendicular to the film surface, which induces large perpendicular coercivity. We observe the coercivity of 11.3 kOe in the 40-nm CFO thin film, which is the highest perpendicular coercivity ever achieved on an amorphous SiO₂ /Si substrate. The present approach can guide the systematic tuning of the magnetic easy axis and coercivity in the desired direction with respect to crystal orientation in the nanoscale regime. Importantly, this can be achieved on virtually any type of substrate.

Switching of magnetic easy-axis using crystal orientation for large perpendicular coercivity in CoFeO

OPEN

R

P

Sagar E. Shirsath, Xiaoxi Liu, YukikoYasukawa, Sean Li & Akimitsu Morisako

Perpendicular magnetization and precise control over the magnetic easy axis in magnetic thin O

the highest perpendicular coercivity ever achieved on an amorphous SiO

approach can guide the systematic tuning of the magnetic easy axis and coercivity in the desired

The pursuit of a fundamental understanding and the control of nanomagnetism have received signicant interest due to its novel properties and wide ranges of potential applications. Examples of nanomagnetic materials include nanosized room-temperature ferromagnets for memory applications or materials with high magnetization and switchability that lend themselves to thin lm architectures for device development. The ability to control magnetic properties such as saturation magnetization, remanent magnetization, and coercivity is important not only for a fundamental understanding of magnetism but also for the applications of magnetic data storage, MRI contrast-enhancement agents, and magnetic hyperthermia for biomedical therapeutic purposes14. Cobalt ferrite, CoFe2O4 (CFO), a ferrimagnetic inverse spinel, is one of the most commercially signicant members of the magnetic ferrite class. CFO resides in an inverse spinel structure (space group Fd3m; no. 227), which is a cubic crystal system with oxide anions arranged in a cubic close-packed lattice and metal cations tetragonally or octahedrally surrounded by oxygen (Fig.1a). Each unit cell consists of 32oxygen ions, 16 octahedral sites and 8 tetrahedral sites. For CFO, Fe3+ ions occupy the tetrahedral sites and half of the octahedral sites, whereas Co2+ ions are located at the remaining octahedral sites5.

Below the Curie temperature (Tc) of 860 K, CFO presents a long-range collinear ferrimagnetic order with antiferromagnetic intersublattice exchange interactions6. CFO has a magnetic easy axis along the (100) directions7,8 with a high saturation magnetization (400emu/cm3), and it has a positive rst-order magnetocrystalline anisotropy constant (K1 2 106 ergs/cm3), an order of magnitude greater than other spinel structure ferrites, resulting in a high coercivity (Hc). These properties make CFO as a unique material for magnetic spin lters9 and spintronics devices such as charge-/strain-driven multiferroic nanostructures1012. It is a common material of choice used in magnetic recording media, and it serves as a ferromagnetic component of read/write heads13,14.

All these applications depend upon the characteristic values of Hc and saturation magnetization (Ms) which is magnetic easy-axis dependent. Particularly in hard disk drives (HDD), the properties of magnetic recording media depend upon the relative orientation of the magnetic easy axis. A preferential perpendicular magnetization and anisotropy in CFO are necessary for a variety of application. Reports have shown a magnetic easy axis tailored through a lattice mismatch between lm and substrate7,15,16, with very expensive single crystal or sapphire

School of

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

Figure 1. Schematic representation of crystal structure orientation and its eect on magnetic properties.

(a) Cubic spinel crystal structure of CFO, indicating tetrahedral and octahedral sites. (b) A (311)-plane that dominates the random orientation. (c) A (111)-plane that dominates (111) orientation in the cubic spinel CFO. (d) The XRD pattern of randomly oriented CFO thin lm; the inset shows the lm texture of polycrystalline lm with a random orientation. (e) The XRD pattern of (111)-oriented CFO thin lm; the inset shows the lm texture of epitaxial lm with a strong (111) orientation. (f) The hysteresis loops of randomly oriented CFO, and (g) the hysteresis loops of (111)-oriented CFO, where the blue loop shows the magnetization measured in the perpendicular direction, and the green loop shows the magnetization measured along the in-plane direction.

substrates. However, to obtain the perpendicular anisotropy, precise control over the direction of the magnetic easy axis of the CFO thin lm deposited on an inexpensive amorphous substrate remains signicant challenge; therefore, an eective strategic approach needs to be develop.

In this study, we demonstrate a fundamental approach for systematically modulating the maximized perpendicular magnetic anisotropy and magnetic easy axis of CFO thin film based on a consideration of the eects of crystal orientation, compressive strain anisotropy and oxygen vacancies. The (111)-preferred and oxygen-decient (311) randomly oriented CFO thin lms are fabricated by the DC sputtering technique and then systematically examined by various characterizations. We demonstrate that the oxygen-decient lm aer post-annealing gives rise to a compressive strain perpendicular to the lm surface. We extend this concept to develop a CFO thin lm with large perpendicular coercivity in a desired direction with moderate saturation magnetization. Our strategy is given in a scheme where the in-plane and perpendicular coercivity are associated with a crystal orientation (Fig.1). This approach can lead to the systematic tuning of the magnetic easy axis and coercivity, particularly with respect to crystal orientation in a nanoscale regime; importantly, this can be achieved on any types of substrate.

Methods

CoFe2O4 thin lms 40 nm thick were grown on amorphous SiO2/Si substrate by DC sputtering. A commercially available ferrite with a stoichiometric composition of the CoFe2O4 target (Kojyundo Chemical laboratory Co., LTD) was used for the deposition of the CFO thin lms. The chamber was evacuated to a pressure below 3104 Pa, and then, the total working gas pressure of the Ar, 0.4 Pa, was fed into the chamber. The facing target sputtering (FTS) technique17 was used in the present work, which stoichiometrically transfers the composition from the target to the lm, in which the distance between the plasma and the substrate was approximately 6 cm, and the DC power of 30 W was applied for the deposition. The samples were grown for 1h (40nm) with dierent substrate temperatures (Ts), viz. room temperature (RT), 50, 100, 150, 200, 250, 300, 325, 350 and 375C. Aer the deposition of lm, the samples were subjected to post-annealing (Ta) at 400 and 800C.

Characterizations. The prepared CoFe2O4 thin lms were characterized by means of X-ray diraction (XRD; Rigaku SmartLab). Pole gure analysis was conducted using 3D Explore soware, reection high-energy electron diraction (RHEED; JEOL JEM-2010), X-ray photoelectron spectroscopy (XPS; AXIS-ULTRA DLD, Shimadzu), vibrating sample magnetometry (VSM; Tamagawa Factory Co. Ltd., custom-made), atomic/magnetic force microscopy (AFM/MFM; Veeco, Innova) and eld-emission scanning electron microscopy (FE-SEM; Hitachi SU8000).

The X-ray diraction patterns of 40-nm CFO thin lm samples deposited at dierent substrate temperatures (Ts) are shown in Fig.2(a). As seen in Fig.2(a), the sample deposited at room temperature shows that the (311) diract line is the most intense peak, indicating the random orientation of the CFO thin lm. The XRD pattern shows a weak contribution from (111), (222), (400) and (333). It should be noted that no contribution from the (220)

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 2. (a) X-ray diraction 2 scans (background removed) of post-annealed CFO. The XRD pattern of Ts= RT and Ts= 375C, blue in color, shows a randomly oriented CFO, XRD pattern of Ts= 100 to 325C; the green in color shows (111)-oriented CFO and XRD pattern; Ts = 50 shows the (111)-dominated random orientation of CFO thin lm. The vertical dotted lines indicate the positions of the lattice plane reections. (b) Variation of XRD peak intensity in the (111) and (311) planes with varying substrate temperatures.(c) Variation of the out-of-plane (a, c-axis) lattice parameter and the rocking curve with the substrate temperature of the CFO thin lm. The horizontal dotted line in part c indicates the lattice parameter of bulk CFO.

plane is detected in the randomly oriented CFO. As the Ts increased from 50 to 100C, the intensity of the (311) line decreased, and the (111) orientation became dominant with the group lines of (222 and 333). The polycrystalline CFO lms generated XRD patterns with all possible crystal orientations of the material, which was similar to the bulk powders. However, (111) preferred oriented or textured lms demonstrated XRD patterns with certain Bragg reections that were more pronounced than others. It is of great signicance to grow (111)-oriented CFO thin lms on an amorphous SiO2/Si substrate at the low temperature of 100C. Most importantly, we observed that this (111) orientation was true for the as-deposited CFO lm without any need for post-annealing. The further increase of Ts resulted in a similar pattern but with strong diraction intensity. Figure2(a,b) show that the CFO deposited at a Ts = 250 C showed the strongest intensity for the (111) plane. The intense (111), (222) and (333) diraction lines of the CFO spinel structure indicated a strong preferential orientation, with the (111) directions of the crystallites perpendicular to the lm plane. As the Ts increased 350C, the (111) preferential orientation was not observed, and the CFO thin lm again showed a random orientation. It is worth to mention here that, XRD peak at 2 = 57.4 is corresponding to (333) and (511) planes and is observed in all the XRD patterns. The (333) plane is observed if the samples are (111) oriented and the plane (511) is observed when the sample is randomly oriented. The full width at half maximum (FWHM) of a rocking-curve analysis of the (111) plane of CFO thin lm deposited at dierent substrate temperatures is shown in Fig.2(c). CFO thin lm deposited at Ts= 250C shows the lowest FWHM (1.9) of the rocking curve conrmed the high (111) orientation. The substrate temperature-dependent change in the crystal orientation of the investigated CFO thin lm was related to the crystallographic change in the polycrystalline or epitaxial thin lms, which was driven by a reduction in the system total anisotropy energy, i.e., Etotal = [(s+ i)/h] + Mhkl2, where s is surface energy, i is interfacial

energy, and h is lm thickness18. The last term, Mhkl2, is elastic strain energy density, is intrinsic residual strain, and Mhkl is the biaxial modulus, which is closely related to the crystallographic direction. It should be noted that we used an amorphous SiO2/Si substrate, and therefore, i can be neglected.

The s for dierent planes of CFO are anisotropic and are: s(111) (208 ergs/cm2) < s(400) (1486 ergs/cm2) < s(220) (1916 ergs/cm2) < s(311) (2344 ergs/cm2)19. The closed-packed (111) planes of the CFO have the lowest surface energy which led to energetically favorable (111)-oriented CFO spinel surfaces. This was valid for the CFO thin lm deposited at Ts = 100350 C. Apart from the s minimizing eect, Mhkl2 also inuences the total energy and the lm orientation. Mhkl2 depends on the crystallographic orientations (hkl), and the eective Mhkl of the grains. The Mhkl can be calculated as a function of the grain orientation (hkl) and the stiness constants (C11, C12 and C44) as Mhkl = C11 + C12 + K [2(C12 K)2/(C12 + 2K)], where K = (2C44 C11 + C12) (h2k2+k2l2+h2l2)/(h2+k2+l2)2,20.

The CFO thin film deposited at Ts = RT, 50 C and 350 C should have the values of Mhkl2, Mhkl (111)

(3.68 1012 dynes/cm2) > Mhkl (311) (3.09 1012 dynes/cm2). The smaller strain energy density for (311) leads to a random orientation of the CFO thin lm rather than an orientation in the (111) direction because the formation of (111) texture lms possesses larger strain energy density than the formation of other textures. The discussion from here onwards is primarily related to the CFO thin lms deposited at Ts = RT, post-annealed at 800C (hereaer referred to as CFORT(Random)) and Ts= 250C and post-annealed at 400C (hereaer referred to as CFO250(111)) because these lms showed the strongest random and (111) orientations, respectively.

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 3. RHEED pattern of the CoFe2O4 thin lm along the 1010

[ ]direction. (a) CFORT(Random) (post-annealed at 800C), (b) CFO375(Random) (post-annealed at 400C), and (c) CFO250(111) (post-annealed at 400C). At the top (0, 3/2) and (0, 1) streaks have been labeled for clarity. (d) 3D pole gure pole gure of CFO250(111)

(post-annealed at 400C).

The variation of average out-of-plane (OOP) lattice parameter (a; c-axis) with the substrate deposition temperature for all CFO thin lms is depicted in Fig.2(c). It is observed that the a lattice parameter increased from 8.344 to 8.396 with the increase in the substrate temperature from Ts = RT to 250 C. However, the further increment in the Ts (>250 C) again shrank the lattice parameter. It remained almost unchanged for the (111)-oriented CFO thin lms deposited in the range of Ts= 100250 C, revealing that there was no detective lattice distortion, and it fully relaxed toward the bulk value of CFO (a=b=c= 8.391). This likely also occurred because the CFO thin lms deposited in this range were in a thermo-equilibrium state without a signicant intrinsic strain, which is important because CFO is known to be highly magnetostrictive. However, the alattice parameter for the randomly oriented CFO thin lms deposited at Ts= RT, 50C and 350C were signicantly lower than the (111)-oriented and CFO bulk values, indicating the contraction in the OOP direction. Therefore, randomly oriented CFO thin lms were in an out-of-plane compressive residual strain that was Ts dependent. To study the strain eect, we determined the in-plane lattice parameter (a||; a-axis) by employing the XRD from the in-plane direction of the CFO lm (Supplementary Fig. 2). The calculated lattice parameters for the most intense (311) peak for the CFORT(Random) were a=8.342 anda||= 8.427. The c/a (a/a||) ratio was 0.989, which conrmed the face-centered cubic to face-centered tetragonal ( fct) distortion of randomly oriented CFO.

The in-plane and out-of-plane strains () were calculated using the formula: = (a a0)/a0, where a is the in-plane (a||) or out-of-plane (a) lattice parameters and a0 is the bulk CFO unstressed lattice parameter. CFORT(Random) post-annealed at 800C showed the highest out-of-plane compressive strain () of 0.583% and an in-plane tensile strain (||) of 0.425%. This marked dierence between the and || strains was attributed to the fact that the lattice parameters (a||) and (a) diered by 1%. One of the reasons for the observed strain in post-annealed CFORT(Random) lm was attributed to the dierence in the thermal expansion coefficient () between the SiO2/Si substrate (3.5106/K) and CFO (1105/K). This dierence in caused tension along the in-plane direction during the cooling of the CFO thin lm to room temperature aer post-annealing. However, the observed strain and therefore oxygen vacancies may also have contributed to the tension, which is discussed later on in the XPS results. By contrast, (111)-oriented CFO thin lms deposited at Ts = 100325 C showed a small OOP tensile strain, 0.06%, which conrmed the lattice relaxation.

We studied the ex-situ RHEED patterns along the [1010] direction in reciprocal space. Figure3(a,b) shows the RHEED patterns of CFORT(Random) and CFO375(Random), respectively, and conrms the excellent crystallinity. In these images, diraction rings are observed, indicating the random and polycrystalline nature of CFO thin lms. Figure3(c) represents the RHEED pattern of CFO250(111); the streaks observed on the RHEED pattern represent the reciprocal lattice. The observed smooth streaks (0, 1) and (0, 1)-type planes corresponded to the FCC

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 4. XPS spectra of O1s core level of CFO thin lms. (a) CFORT(Random) (as deposited), (b) CFORT(Random)

(post-annealed at 800C), (c) CFO250(111) (as-deposited), (d) CFO250(111) (post-annealed at 400C), (e) CFO375(Random)

(as deposited) and (f) CFO375(Random) (post-annealed at 400C). Schematic (x) shows the illustration of the fcc CFO lattice dimension of the as-deposited CFORT(Random) with oxygen vacancies and oxygen at interstitial sites. (y) isthe schematic showing the eect of the post-annealing temperature on the CFORT(Random) lm. Lattice wall of the bottom layer of CFO thin lm is covalently bonded with the substrate and posses the fcc structure, however asthe layer thickness increased it transformed from the fcc to the fct structure with the removal of oxygen from the interstitial sites.

sublattice. No spotty pattern was observed that indicated that the deposited lm grew in two dimensions. The intermediate (0, 3/2) and (0, 3/2) streaks revealed the characteristic RHEED pattern of the spinel structure. The slight presence of rings around the RHEED streaks suggested the crystalline quality was not excellent, which is related to the post-annealing of the CFO thin lm. To obtain more clarity of the (111) orientation in CFO250(111),

we measured the pole gure of the (111) plane (Fig.3d). It was observed that a pole was located at the center, which showed that the (111) plane of CFO was parallel to the sample surface and further conrmed that the lm had a strong (111) texture.

XPS spectroscopy is the most appropriate method to identify the oxidation state and chemical composition of CoFe2O4. Figure.4 shows the XPS spectra of CoFe2O4 thin lms prepared at dierent synthesis parameters. The

XPS spectra primarily involve the Co2p, Fe2p and O1s peaks, and the shape of these peaks depends greatly on the chemical state of the atoms. In the case of Fe, the Fe2p peak was split into two components, Fe2p1/2 at 723.5eV and

Fe2p3/2 at 710.1eV, due to spin-orbit coupling (supplementary Fig. 3). It is known that a noticeable satellite peak in the intermediate Fe2p1/2 and Fe2p3/2 peak appears due to the presence of only Fe3+ ions. However, in the present case, the intensity of such a satellite peak (718eV) was less evident. We were more interested in the O1s peak as the CFO lm deposition was carried out in Ar gas atmosphere only without introducing the oxygen gas. Oxygen does play a vital role in governing the structural and magnetic properties in an oxygen closed-packed cubic spinel system such as CFO. The O 1s XPS spectra were deconvoluted into three symmetric peaks corresponding to lattice oxide (Fe/Co-O), surface hydroxyl (Fe/Co-OH) and loosely bound oxygen, such as absorbed O2

or adsorbed H2O at Oi = 529.67529.94 (0.05) eV, Oii = 530.89531.17 (0.15) eV and Oiii = 532.01532.91 (0.9) eV, respectively. In the present case, the main peak Oi of O 1 s at 529.91 eV shied to a lower binding energy at 529.68 eV only in the case of CFO deposited at Ts = 375 C. This peak almost remained constant at 529.83529.87 eV for as-deposited and post-annealed CFO250(111); therefore, it was considered not to be oxygen decient. However, more importantly, the Oi peak position suggested that the binding energy of the as-deposited CFORT(Random) was 529.67eV, and aer post-annealing, it shied to a higher binding energy of 529.94eV.

It should be considered that the amorphous inorganic framework of the CFO thin lm deposited at room temperature is porous and may absorb some oxygen from the atmosphere. These absorbed oxygen ions reside at interstitial sites and are diused out aer post-annealing as the lm become dense. Because the lm is covalently bound to the substrate lattice, the wall density can be increased not in an in-plane direction but in the OOP

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

direction. Because of the increased density, the CFO lattice can distort to relax the strain with oxygen vacancies by compressing the c-axis lattice parameter and expanding the a-axis lattice parameter. We schematically demonstrate this mechanism in Fig.4 (x and y) for a better understanding. It has been reported that presence of oxygen at interstitial sites may enhance the c-axis lattice parameter of the perovskite structure21. As reported earlier that oxygen vacancies increase the c-axis lattice parameter of oxide thin lms such as La0.65Sr0.35MnO3/SrTiO322,

HoMnO3/Al2O323, LaMnO3/SrTiO324, LaAlO3, BiFeO325 and (LaSr)CoO/NGO, LSAT26. However, our results for post-annealed cubic CFORT(Random) are contrary to these reports, rejecting the possibility of the presence of oxygen at interstitial sites aer post-annealing. The shi in Oi can be attributed to the increased oxygen vacancies, which decreased the work function and resulted in the Oi (O1s) peak shis towards higher binding energies27.

Variations in the Oii and Oiii peaks further claried the oxygen vacancies in the CFO thin lm. Oii can also be related to oxygen in oxygen-decient regions. The position of the Oii peak generally has a higher binding energy by 1.5 eV than Oi28. However, as per Fig.4 it is higher only by1.25 (0.03) eV, which may be due to the contribution from oxygen-decient regions. The calculated Oii/Oi ratios were approximately 0.49 (0.1), 0.38 (0.2) and 0.18 (0.1) for the post-annealed CFORT(Random), CFO250(111) and CFO375(Random), respectively. A decrease in the ratio of Oii/Oi with the increase in the substrate temperature conrmed the reduction of the oxygen vacancies in CFO thin lms. It should be noted here that the contribution of OH to the Oii region may not be ignored; this become even more signicant for the CFO250(111) lm because no peak shi of Oi was observed. As discussed earlier, the non-evident or very weak Fe3+-state intermediate satellite peak between Fe2p3/2 and Fe2p1/2 at 718eV indicates that Fe2+ ions produced by the reduction from Fe3+ ions which is due to oxygen deciency. Furthermore, to sustain the charge neutrality, oxygen vacancies in the lm were compensated by a transformation of the Fe cation oxidation state, from Fe3+ to Fe2+ ions. Therefore, the spinel Fe3O4 phase may be formed with the reduction of oxygen in the CFO lms during sputtering.

Quantitative analysis of overall chemical composition is found to be Co0.95Fe2.05O4-, Co0.98Fe2.02O4 and

Co0.99Fe2.01O4 for the post-annealed CFORT(Random), CFO250(111) and CFO375 (Random) thin films, respectively. The cation distribution by the XPS data was estimated as (Co0.38Fe0.62)A[Co0.57Fe1.43]BO4- and

(Co0.15Fe0.85)A[Co0.83Fe1.17]BO4 for post-annealed CFORT(Random) and CFO250(111) lms, respectively. As reported earlier, the Fe2+ ions in CFO occupied only B-sites, thus pushing the Co2+ ions to the A-site29.

The surface morphology of CFO thin films was observed by SEM (Fig.5). The film surface was smooth for CFORT(Random) and CFO250(111). The disc-shape grains were of an irregular size and shape, giving rise to the random orientation for the CFORT(Random) (Fig.5a). In contrast to these uniformly isolated spherical-shaped grains, a narrow grain-size distribution was observed for the CFO250(111) (Fig.5b). As the Ts increased to 375C (CFO375(Random)), some region of the CFO thin lm melted, exposing some voids and cracks (Supplementary Fig. 4) gave rise the random orientation for CFO375(Random). The average grain size estimated from these images are 35, 30 and 110nm for Ts= RT, 250C and 375C respectively. The representative AFM images are depicted in Fig.5, exhibiting grains with dierent shapes. Post-annealing CFORT(Random) had a root mean square roughness (RMS)

of 1.856nm, whereas for CFO250(111), it was 0.903nm. The corresponding MFM images show that it consisted of domains with a cluster-like structure for CFORT(Random) lm in which magnetization was conned up and down with dark and bright contrasts, respectively.

To reveal the co-relation of crystal orientation with magnetization, the magnetic behavior of CFO thin lm was analyzed by performing VSM. The hysteresis loops of the representative CFO thin lm samples obtained by the VSM are depicted in Fig.6. Magnetic properties such as saturation magnetization (Ms), the remanent ratio (R = Mr/Ms) and coercivity (Hc) extracted from hysteresis loops of as-deposited and post-annealed (400 and 800C) CFO thin lm and are presented in Fig.7 and supplementary Fig. 5. It is noticed that all the post-annealed randomly oriented CFO lms possess a large perpendicular coercivity compared to in-plane coercivity and exhibit a higher Ms, Mr, and R and a magnetic easy axis in the OOP direction. In contrast, all the (111)-oriented CFO lms exhibiting a magnetic easy axis in the in-plane direction show higher in-plane coercivity than perpendicular coercivity and have a higher Ms, Mr, and R along the in-plane direction. As discussed earlier, the lattice of post-annealed CFORT(Random) lm was under face cubic tetragonal distortion, which suggests that the magnetoelastic energy (Kme) may contribute to the control of the magnetic easy axis, where Kme is directly proportional to (a||a)30. Randomly oriented CFO lm has Kme>0; therefore, Kme is uniaxial, which directs the easy axis in the OOP direction. It is exactly opposite for (111)-oriented lms, where Kme< 0 and has a negative value, leading to a biaxial in-plane anisotropy, which makes an easy axis along the in-plane direction. The analysis of Kme also suggests that all the randomly oriented CFO lms were under OOP compressive strain; this strain forced the magnetic easy axes to lie parallel to the compression direction, which caused the perpendicular magnetic anisotropy. By contrast, (111)-oriented lms showed little OOP tensile strain, which compensated for compression along the in-plane direction and thus resulted an increase in strain anisotropy. Increased strain anisotropy translated into an increased magnetization along the magnetically easy in-plane direction for (111)-oriented lms. Moreover, the IP M-H loop of CFO250(111) has strong contribution to magnetic anisotropy of the CFO lm. It makes strong demagnetization energy and aects signicantly in anisotropy constant. This nding clearly suggested that the crystal orientation and compressive strain drove the direction of the magnetic easy axis in the CFO thin lm.

The present work was designed to achieve high coercivity. Randomly oriented post-annealed CFORT(Random)

lm showed the highest perpendicular coercivity (Hc), 11.27kOe, with an in-plane coercivity (Hc||) of 4.7kOe. By contrast, (111)-oriented post-annealed CFO250(111) lm showed an Hc|| of 4.1 kOe, which was higher than its Hc (3.0 kOe). Considering the deposition of CFO on the SiO2/Si substrate, it is worth emphasizing that the observed high perpendicular coercivity measured at room temperature in post-annealed CFORT(Random) lm was,

to the best of our knowledge, greater than 5.37 kOe (91%), compared to the highest coercivity of 5.9 kOe for CFO thin lm previously reported by Raghunathan et al.31. A comparable coercivity in CFO was reported by Yin et al.32; however, they used single-crystal quartz and (0001) sapphire substrate and employed a high substrate temperature of 550C.

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 5. SEM, AFM and MFM images of CFO thin lm. (a) SEM of CFORT(Random) post-annealed at 800C for 4h; (b) SEM of CFO250(111) post-annealed at 400C for 4h; (c) AFM image and (d) its corresponding MFM image of post-annealed CFORT(Random), whereas (e) AFM image and (f) is its corresponding MFM image of post-annealed CFO250(111) thin lm.

The observed very high coercivity in post-annealed CFORT(Random) lm can be interpreted in terms of the total uniaxial anisotropy energy (Ea) that controls the spin alignment in thin lms. Ea is the sum of several factors, including thickness (t), surface (Ks), shape (Ksh), stress (Ka) and magnetocrystalline (Ku) anisotropies; therefore, it can be written as Ea=Ks/t+Ksh+Ka+Ku, where Ks, Ksh, and Ka and Ku are the extrinsic and intrinsic contributions, respectively. A positive Ea indicates an easy axis is in the perpendicular direction and OOP magnetic anisotropy, whereas a negative value of Ea indicates in-plane direction and in-plane magnetic anisotropy30. All CFO thin lms under investigation are 40-nm thick; therefore, Ks/t can be neglected because it is eective for much thinner lm. Ksh is equivalent to 2Ms2; the calculated values are 0.4106erg/cm3 and 1.05106erg/cm3 for CFORT(Random) and CFO250(111) lms, respectively. Because the Ksh is negative, it contributes only to in-plane magnetization33. Relating Ksh with Kme, Kme has a positive value for CFORT(Random), whereas it is negative for the CFO250(111) lm. This indicates that a positive Kme is sufficient to overcome a negative Ksh in CFORT(Random); therefore, Ksh has less of an eect and can be neglected. CFO is known to have a high Ku because of the spin orbit stabilized ground state and unquenched orbital momentum, caused by a trigonal crystal eld on the Co2+ cations. However, according to the single-ion model, this is true only when CFO possesses a fully inverse spinel structure, where Co2+ ions occupy only the octahedral site. As observed from the XPS results, Co2+ ions occupy

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

Figure 6. Room temperature variation of magnetization (M) with applied magnetic eld (H) of post-annealed CFO thin lms. M-H loops of CFORT(Random), CFO50(Random) and CFO375(Random) are of randomly oriented CFO thin lm where perpendicular coercivity is higher than that of in-plane coercivity. In these cases magnetic easy-axis is along perpendicular (c-axis) direction. M-H loops of CFO100(111), CFO250(111) and

CFO325(111) are of (111) oriented CFO thin lm where in-plane coercivity is higher than that of perpendicular coercivity. In these cases magnetic easy-axis is along in-plane direction (a-axis) direction.

a tetrahedral A site in a greater amount, suggesting that the CFORT(Random) lm possesses a partial inverse spinel structure. Therefore, Ku does not play a signicant role in the observed high Hc in the CFORT(Random) lm.

The stress () was calculated using the relation = Y, where Y is the Youngs modulus for CFO (Y =1.5 1012 dyne/cm2)34. The in-plane (||) and out-of-plane () stresses calculated for the CFORT(Random) lm were found to be 6.37 109 and 8.74 109 dyne/cm2, respectively, whereas it was = 0.89 109 dyne/cm2 for the CFO250(111) lm. Here, negative sign indicates the compression whereas positive sign indicates the tension. Using these stress values, the stress-induced anisotropy constant (Ka) was estimated, Ka=(3/2) , where is the magnetostriction coefficient and is taken as; 100 = 590 106 3436. The estimated value of Ka for post-annealed CFORT(Random) is 7.7 106erg/cm3 in perpendicular direction which is higher than bulk the CFO (2 106 ergs/cm3). Therefore the total uniaxial anisotropy energy (Ea) in post-annealed CFORT(Random) lm is mainly governed by the Ka and thus played a deciding role in the high Hc for CFORT(Random) lm. By contrast, the moderate Hc observed in the CFO250(111) lm correlated with a lower Ka as a result of lattice-strain relaxation. The Ms value for the CFORT(Random) lm was slightly lower than that of bulk CFO and CFO250(111). This may be attributed to strain, a partially inverse spinel structure, and/or antiphase boundaries, which form when islands that nucleate at dierent areas of the substrate merge and are out of phase with each other37. The lower Ms for CFORT(Random) is also attributed to the oxygen deciency, which led to a decrease in the super-exchange interaction of the Co2+ cation and the O2 anion.

Conclusions

40-nm CFO thin lm with a (111) orientation and a (311)-random orientation was prepared by DC magnetron sputtering on an SiO2/Si substrate. CFO was deposited at dierent substrate temperatures, ranging from room temperature to 375 C. The (111) orientation was achieved at a very low substrate temperature, 100 C; however, this orientation was strongest at a substrate temperature of 250 C, as evidenced by the XRD patterns and rocking-curve measurements and the RHEED images. In the case of randomly oriented CFO, the out-of-plane lattice constant was much smaller than that of the in-plane lattice constant, revealing the fact that randomly oriented CFO was under compressive residual strain, possessing an fct structure. By contrast, (111)-oriented CFO showed a relaxed crystal structure and was under slight tensile strain with an average grain size of 30nm. The XPS results showed that randomly oriented post-annealed CFO had some oxygen deciency; this deciency and the increase in lm density produced compressive strain in this lm. It was observed that randomly oriented CFO

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

Figure 7. Variation of coercivity (Hc) and saturation magnetization (Ms) with dierent substrate temperature (Ts) of CoFe2O4 thin lm. The magnetization measurements were carried out in perpendicular

and In-plane directions of the lm surface. Yellow pillar region in the Ms, and Hc graphs denotes the magnetic measurements of randomly oriented CFO thin lm. Ms values are considered to be remain unchanged for both perpendicular and In-plane directions.

showed a magnetic easy axis along the perpendicular direction (c-axis); exactly the opposite, the (111)-oriented lm showed the magnetic easy axis along the in-plane direction (a-axis). Randomly oriented CFO thin lm possessed a large perpendicular coercivity of 11.3 kOe. This large coercivity is associated with the compressive strain anisotropy induced by the oxygen deciencies. The ultra-smooth surface observed from the AFM and the large perpendicular coercivity makes randomly oriented CFO thin lm suitable for magnetic recording media. Importantly, the demonstrated novel approach provides a facile way to control and manipulate the magnetic easy axis in CFO thin lm on virtually any substrate.

SCIENTIFIC REPORTS

9

www.nature.com/scientificreports/

References

1. Noh, S. et al. Nanoscale magnetism control via surface and exchange anisotropy for optimized ferrimagnetic hysteresis. Nano Lett. 12, 37163721 (2012).

2. Sellmyer, D. & Skomski, R. Advanced Magnetic Nanostructures Springer, New York (2006).3. Sun, S., Murray, C. B., Weller, D., Folks, L. & Moster, A. Monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattices. Science 287, 19891992 (2000).

4. Prozorov, T. et al. Cobalt ferrite nanocrystals: out-performing magnetotactic bacteria. ACS Nano 1, 228233 (2007).5. Schmitz-Antoniak, C. et al. Electric in-plane polarization in multiferroic CoFe2O4/BaTiO3 nanocomposite tuned by magnetic elds. Nature Commun. 4, 2051, doi: 10.1038/ncomms3051 (2013).

6. Sun, Qi -C., Birkel, C. S., Cao, J., Tremel, W. & Musfeldt, J. L. Spectroscopic signature of the superparamagnetic transition and surface spin disorder in CoFe2O4 nanoparticles. ACS Nano, 6, 48764883 (2012).

7. Dix, N. R. et al. Selectable spontaneous polarization direction and magnetic anisotropy in BiFeO3 CoFe2O4 epitaxial nanostructures. ACS Nano 4, 49554961 (2010).

8. Bozorth, R. M., Tilden, E. F. & Williams, A. J. Anisotropy and magnetostriction of some ferrites. Phys. Rev. 99, 17881798 (1955).9. Ramos, A. V. et al. Inuence of oxidation on the spin-ltering properties of CoFe2O4 and the resultant spin polarization. Phys. Rev. B 78, 180402(R) (2008).

10. Zheng, H. et al. Multiferroic BaTiO3-CoFe2O4 nanostructures. Science 303, 661663 (2004).11. Zavaliche, F. et al. Electric eld-induced magnetization switching in epitaxial columnar nanostructures. Nano Lett. 5, 17931796 (2005).

12. Vaz, C. A. F., Homan, J., Anh, C. H. & Ramesh, R. Magnetoelectric coupling eects in multiferroic complex oxide composite structures. Adv. Mater. 22, 29002918 (2010).

13. Quickel, T. E., Le, V. H., Brezesinski, T. & Tolbert, S. H. On the correlation between nanoscale structure and magnetic properties in ordered mesoporous cobalt ferrite (CoFe2O4) thin lms. Nano Lett. 10, 29822988 (2010).

14. Dai, Q. et al. Self-assembled ferrimagnet-polymer composites for magnetic recording media, Nano Lett. 10, 32163221 (2010).15. Gao, X. S. et al. Switching of magnetic anisotropy in epitaxial CoFe2O4 thin lms induced by SrRuO3 buer layer. J. Phys. D.: Appl. Phys. 42, 175006 (2009).

16. Coll1, M. et al. Low temperature stabilization of nanoscale epitaxial spinel ferrite thin lms by atomic layer deposition. Adv. Funct. Mater. 24, 53685374 (2014).

17. Naoe, M., Yamanaka, S. & Hoshi, Y. Facing targets type of sputtering method for deposition of magnetic metals at low temperature and high rate. IEEE Tans. Mag. MAG-16, 646648 (1980).

18. Thompson, C. V. Structure evolution during processing of polycrystalline lm. Annu. Rev. Mater. Sci. 30, 159190 (2000).19. Zheng, H. et al. Controlling self-assembled perovskitespinel nanostructures. Nano Lett. 6, 14011407 (2006).20. Thompson, C. V. & Carel, R. Stress and grain growth in thin lms. J. Mech. Phys. Solids 44, 657673 (1996).21. Nakamura, T., Ling, Y. & Amezawa, K. The eect of interstitial oxygen formation on the crystal lattice deformation in layered perovskite oxides for electrochemical devices. J. Mater. Chem. A 3, 1047110479 (2015).

22. Orgiani, P. et al. Evidence of direct correlation between out-of-plane lattice parameter and metal-insulator transition temperature in oxygen-depleted manganite thin lms, Appl. Phys. Lett. 100, 042404 (2012).

23. Lee, D. et al. Giant exoelectric eect in ferroelectric epitaxial thin lms. Phys. Rev. Lett. 107, 057602 (2011).24. Xu, Z. T. et al. Evidence for a crucial role played by oxygen vacancies in LaMnO3 resistive switching memories. Small 8, 12791294 (2012).

25. Rondinelli, J. M. & Spaldin, N. A. Structure and properties of functional oxide thin lms: Insights from electronic-structure calculations. Adv. Mater. 23, 33633381 (2011).

26. Kim, Y. M. Probing oxygen vacancy concentration and homogeneity in solid-oxide fuel-cell cathode materials on the subunit-cell level. Nature Mater. 11, 888894 (2012).

27. Maeng, J. et al. The eect of excimer laser annealing on ZnO nanowires and their eld eect transistors. Nanotechnol. 20, 095203 (2009).

28. Thgersen, A., Rein, M., Monakhov, E., Mayandi, J. & Diplas, S. Elemental distribution and oxygen deciency of magnetron sputtered indium tin oxide lms. J. App. Phys. 109, 113532 (2011).

29. Moyer, J. A., Vaz, C. A. F., Negusse, E., Arena, D. A. & Henrich, V. E. Controlling the electronic structure of Co1xFe2+xO4 thin lms through iron doping. Phys. Rev. B 83, 035121 (2011).

30. Lis, A. et al. Reorientation of magnetic anisotropy in epitaxial cobalt ferrite thin lms. Phys. Rev. B 76, 054405 (2007).31. Raghunathan, A., Nlebedim, I. C., Jiles, D. C. & Snyder, J. E. Growth of crystalline cobalt ferrite thin lms at lower temperatures using pulsed-laser deposition technique. J. Appl. Phys. 107, 09A516 (2010).

32. Yin, J., Ding, J., Liu, B., Miao, X. & Chen, J. Nanocrystalline Co-ferrite lms with high perpendicular coercivity. Appl. Phys. Lett. 88, 162502 (2006).

33. Coey, J. Magnetism and magnetic materials Cambridge University Press, New York (2010).34. Bozorth, R. M., Tilden, E. F. & Williams, A. J. Anisotrophy and magnetostriction of some ferrites. Phys. Rev. 99, 17881798 (1955).35. Suzuki, Y., Hu, G., Van Dover, R. B. & Cava, R. J. Magnetic anisotropy of epitaxial cobalt ferrite thin lms. J. Magn. Magn. Mater. 191, 18 (1999).

36. Fritsch, D. & Ederer, C. Epitaxial strain eects in the spinel ferrites CoFe2O4 and NiFe2O4 from rst principles. Phys. Rev. B 82, 104117 (2010).

37. Margulies, D. T. et al. Origin of the anomalous magnetic behavior in single crystal Fe3O4 lms. Phys. Rev. Lett. 79, 5162 (1997).

Acknowledgements

One of the authors (SES) is very thankful to the Japan Society for the Promotion of Science (JSPS) for providing the postdoctoral fellowship and nancial support (Grant-in-Aid) for scientic research.

S.E.S., X.L. and A.M. designed the project, analyzed the results. S.E.S. prepared the thin lm samples. S.E.S., X.L. and Y.Y. performed the calculations and characterization work. S.E.S. led the writing of the paper with contributions from X.L., S.L. and A.M.

Additional Information

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Shirsath, S. E. et al. Switching of magnetic easy-axis using crystal orientation for large perpendicular coercivity in CoFe2O4 thin lm. Sci. Rep. 6, 30074; doi: 10.1038/srep30074 (2016).

SCIENTIFIC REPORTS

10

www.nature.com/scientificreports/

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

11