Correlation between the Effective Amounts of Elements in TbFeCo Thin Films and Their Magnetic Properties

Abstract

In this study, TbFeCo thin films were prepared using DC-magnetron sputtering technique utilizing a conventional batch-type sputtering machine. The properties such as magnetic hysteresis loops and magneto-optical properties of the obtained films were measured. Furthermore, we precisely evaluated the “effective” amounts of elements, i.e., the amounts of constituents released by oxidation, in TbFeCo thin films. We also attempted to control the amounts of effective elements by changing the pre-sputtering condition before film preparation. This is because the pre-sputtering process greatly affected the effective amounts of elements in the prepared TbFeCo thin films, such that the magnetic properties were drastically different from sample to sample by altering the pre-sputtering condition. Insufficient pre-sputtering led to a relatively high oxygen content in the films in comparison with those prepared with sufficient pre-sputtering. Outstanding perpendicular magnetic anisotropy in the out-of-plane direction was observed in a film prepared after 60 min of pre-sputtering, which exhibited a coercivity (H_c⊥) value of 6.4 kOe from magnetic hysteresis measurements and 8.2 kOe from magneto-optical polar Kerr hysteresis measurements at an incident light of wavelength 700 nm. The saturation value of the polar Kerr rotation angle (θ_K) of this film was approximately 0.3°, which is comparable to the theoretically optimized value. Therefore, we demonstrated that high-quality TbFeCo films can be obtained with high reproducibility by using a simple batch-type sputtering machine and that there is a strong correlation between effective amounts of elements in TbFeCo thin films and their magnetic properties.

 

This Paper was Originally Published in Japanese in J. Japan Inst. Met. Mater. 82 (2018) 140–146.

Fig. 8 Comparisons of the magneto-optical properties of TbFeCo films after 3 and 60 min of pre-sputtering. Polar Kerr hysteresis loops measured at (a) the film surface and (b) the substrate side at an incident light wavelength of 700 nm. Changes in θ_r values as a function of incident light wavelength for measurements at (c) the film surface and (d) the substrate side.

1. Introduction

TbFeCo thin film is an amorphous material exhibiting superior perpendicular magnetic anisotropy with large coercivity at room temperature. In comparison with other magnetic materials, TbFeCo thin film shows larger Kerr rotation angle (θ_K);¹⁾ hence, it has been extensively studied and had been applied as a magneto-optical recording material.²⁾ The magnetic moment structure of TbFeCo is composed of moments originated from Tb (4f element) and Fe and Co (3d elements). The magnetic moments from Tb and Fe/Co arrange antiparallel with each other, exhibiting perpendicular magnetic anisotropy in the vicinity of the compensation composition.³⁾ However, off-compensation composition shows deteriorated perpendicular magnetic anisotropy, as well as low coercivity (H_c)^3,4) in the TbFeCo system. A precise composition control of TbFeCo is necessary for the perpendicular magnetic anisotropy at the compensation composition. Thus, various techniques have been attempted to address this issue.^1,4) Recently, TbFeCo, such as ferromagnetic nanowires for spintronic devices^5,6) and perpendicular magnetic tunnel junctions to develop perpendicular magnetic random access memory,⁷⁾ has been investigated. We have been studying a composite material, that consists of TbFeCo thin film and noble-metal particles aiming for modification, control, and enhancement of the magnetic properties of TbFeCo thin film with the assistance of localized-surface plasmon resonance generated by noble-metal particles.

Tb is quite easy to be oxidized among the constituents of TbFeCo, such that it is hard to optimize the conditions of film deposition for the high-quality TbFeCo thin film. In other words, the problems to be addressed in the TbFeCo thin film are suppression of H_c⊥, θ_K, and other magnetic properties owing to the oxidation of the film. To overcome these drawbacks, addition of elements^8,9) and protection layer¹⁰⁾ has been attempted in the TbFeCo and/or TbFe thin film systems to stabilize the perpendicular magnetic anisotropy. Another problem of this material is its high process cost because a load-lock system is necessary to achieve the ultra-high vacuum environments to avoid the oxidation of the film. Moreover, the origin of perpendicular magnetic anisotropy is complex in rare-earth–transition metal amorphous alloy thin film systems, including TbFeCo. These factors result in the difficulty to control precisely the magnetic characteristics of TbFeCo thin film. Each research group attempts to establish the most appropriate deposition conditions based on the empirical tries and errors. Thus, there have been various results owing to the different experimental environments of each research group,^11,12) which is another problem in this material system.

In the present study, we focus on the TbFeCo monolayer system. We strictly controlled and evaluated the “effective” content of constituents in TbFeCo as a function of pre-sputtering conditions. Pre-sputtering is a cleaning process of sputtering targets, as well as Tb chips, to control the film composition prior to the thin film preparation. Herein, we define the non-oxidized elements as elements without any bonds with O, which exist in pure simple substances, as “effective” elements. We attempted to control the effective content of elements by controlling only the pre-sputtering conditions, leading to the suppression of the TbFeCo thin film oxidation. Unlike additional elements and formation of protection layers (i.e., extrinsic approaches), our research approach is an intrinsic one. In this study, we clarify the correlation between the effective amount of elements and magnetic properties based on experiments, and detailed discussion will be made.

So far, the oxidized elements and elements freed from the oxidation had not been distinguished, and they were simply regarded as the “constituent” in the TbFeCo thin film in previous studies. In other words, previous studies simply discussed the magnetic properties in relation to the “constituent”, consisting of oxidized and non-oxidized elements. We noticed that magnetic properties, specifically intrinsic magneto-optical properties of TbFeCo, must be affected only by effective elements. Therefore, we prepared TbFeCo thin films by altering only the pre-sputtering conditions and discussed the correlation between magnetic properties and effective amounts of elements from the perspectives of the magnetic and magneto-optical properties. We also comprehensively quantified the effective amounts of elements. To the best of our knowledge, this study is the first research that is based on this perspective.

This study also aims to explore the cost-effective manner in rare-earth–transition metal amorphous alloy systems. Thus, we attempted to establish the preparation conditions of TbFeCo thin films with superior magnetic characteristics by using a simple, cost-effective, high-versatility batch-type magnetron sputtering machine with a single sputtering target. The key point of the material design that aims at the innovative application of TbFeCo thin film would be the control of the “effective” amount of elements, the clarification of the relationship between the “effective” amount of elements and magnetic properties, and the establishment of film preparation conditions with high reproducibility for high-quality films by using a simple technique.

2. Experiment

2.1 Film preparations

We prepared the TbFeCo thin film on the glass substrate (Corining Inc.; EAGLE XG, 33–35 mm × 25 mm × 0.5 mm) by using a batch-type magnetron sputtering machine (SBR-1104, a customized machine of ULVAC Inc.) with a single sputtering target. Prior to film deposition, the glass substrate was subjected into isopropyl alcohol (once) followed by ultra-pure water (twice) and cleaned under sonication for 10 min for each process, thereby it took 30 min for the cleaning procedures.

We evacuated the sputtering chamber to achieve the base pressure of 1.33–2.67 × 10⁻⁴ Pa. The pre-sputtering was performed for the cleaning of the sputtering target and Tb chips as well prior to film preparation. In this study, the pre-sputtering time was varied. Afterward, the TbFeCo thin film was deposited. We maintained the Ar gas pressure of 4.00 × 10⁻¹ Pa during pre-sputtering and film preparation processes. Moreover, the sputtering power density was set to 0.12 W/cm² for the entire experimental procedures, and the pre-sputtering time was changed from 3 to 120 min. Then, the TbFeCo thin film with the thickness of 65 nm was prepared using the composite-target technique after the pre-sputtering.

In the present study, we used the Tb chips (10 mm × 10 mm × 1 mm; Japan Yttrium Ltd.) set on the TbFeCo alloy target (Tb:Fe:Co = 22:66:12 at%; Toshima Manufacturing Co., Ltd.) as the composite-target. Prior to the experiments, the Tb chips were cleaned under ultrasonication in acetone (once) and, then, in isopropyl alcohol (once) for 10 min. The TbFeCo thin film was deposited using a DC sputtering technique (200–250 V, 0.04–0.05 A, 10 W) at the Ar gas pressure of 4.00 × 10⁻¹ Pa. With this sputtering condition, we deposited the film at the rate of 0.02–0.04 nm/s. The sputtering power density during the film deposition was the same as that of the pre-sputtering (i.e., 0.12 W/cm²). In the present study, all the TbFeCo thin films were prepared under the equivalent experimental conditions, except the pre-sputtering time.

2.2 Film evaluations

The magnetic properties of the film were measured through the vibrating-sample magnetometer (VSM, CA-10A; Toei Industry Co., Ltd.) at room temperature immediately after film deposition. For the magnetic property measurement, the maximum magnetic field was applied (i.e., ±10 kOe) to the film, and high-purified Ni plate (Tokei Industry Co., Ltd.) was utilized for the calibration of magnetic measurements. The thin film prepared was cut to 8 mm × 8 mm, with the same shape and size to that of standard high-purified Ni plate, and then the magnetic properties were measured. The X-ray fluorescence (XRF) method (SEA1200VX; EpoLead Inc.) was used to determine the Tb content and other constituents in the obtained thin film. Through atomic force microscopy (VN-8010; KEYENCE Corp.), we evaluated the thickness of the films. Moreover, X-ray photoelectron spectroscopy (XPS, PHI5000 Versa Probe; ULVAC-PHI Inc.) was used to reveal the chemical-bonding states of the constituents in the thin film as the film depth function at the acceleration voltage of 1 kV. In the XPS measurements, the film was sputtered/etched for 0.66 nm in each cycle, and then all the constituents in the film were analyzed. This measurement was repeated in 21 cycles. We also analyzed the areas of the XPS spectra by using a specific software, MultiPak V9 (ULVAC-PHI Inc.). The magneto-optical properties, θ_K, of the film were measured through ultraviolet spectroscopy for the magnetic-property evaluation system (NEOARK Corp.; BH-M800UV-HD-10) with the application of external magnetic field of ±15 kOe in the range of incident light wavelengths of 250–900 nm.

3. Results and Discussion

3.1 Crystallinity and magnetic properties

Figure 1 shows the XRD diffraction pattern of a TbFeCo thin film with the thickness of 65 nm, which was deposited after 60 min of pre-sputtering. TbFeCo is generally known to be amorphous; however, it has the possibility to be crystallized depending on the deposition conditions. We concluded that the present thin film is amorphous because the diffraction pattern only showed a halo pattern without any traces of peaks originating from the crystal phase(s).

Fig. 1

X-ray diffraction pattern of a TbFeCo film with a thickness of 65 nm.

Figure 2 shows the magnetic properties of these thin films in varied pre-sputtering times prior to film preparation. The results show the changes in the magnetization (M) when the external magnetic field is applied normal to the film surface. Note that M represents the normalized value. Noticeably, the magnetic properties varied among specimens when the pre-sputtering time was within 3–120 min before the film deposition although the preparation condition of TbFeCo thin films was the same. The films after the pre-sputtering times of 3 min (Fig. 2(a)) and 20 min (Fig. 2(b)) both showed the H_c⊥ value of approximately 3.7 kOe. However, the films deposited after 60 and 120 min of pre-sputtering obtained the H_c⊥ values of approximately 6.4 kOe (Fig. 2(c)) and 5.7 kOe (Fig. 2(d)), respectively. We found the tendency of superior perpendicular magnetic anisotropy with sufficient H_c⊥, as well as an increase in the pre-sputtering time. In the magnetic hysteresis curves, slight discontinuities were observed for the films after 20 and 120 min of pre-sputtering. According to a previous study, the magnetic property of the TbFeCo thin film showed a discontinuity when the film was prepared at low sputtering power.¹³⁾ We formed the present films at 0.12 W/cm², which is a remarkably low sputtering power density in this study. Thus, we consider that the film exhibited “flips” in the hysteresis curves.

Fig. 2

Out-of-plane magnetic hysteresis loops of TbFeCo films measured at room temperature. The pre-sputtering time before film preparation was (a) 3, (b) 20, (c) 60, and (d) 120 min. After the pre-sputtering, TbFeCo films were prepared under the same experimental conditions for all samples; namely, a sputtering power density of 0.12 W/cm² and an Ar gas pressure of 4.00 × 10⁻¹ Pa. Note that the y-axis represents the normalized values.

Generally, the magnetic properties of TbFeCo are drastically different as a function of the Tb content in the films;³⁾ thus, we evaluated the Tb content (●) in the films by using the XRF technique. The correlation between the Tb content and H_c⊥ value (△) obtained from Fig. 2 is shown in Fig. 3(a). The Tb content of the films was approximately 26 at% for all the specimens, whereas the H_c⊥ value increased for the films deposited with the pre-sputtering time longer than 60 min. Thus, the pre-sputtering time of more than 60 min is effective for the perpendicular magnetic anisotropy of the TbFeCo thin film. Meanwhile, the saturation magnetization value (M_s) decreased with increasing the pre-sputtering time for both directions, normal (△) and parallel (●) to the film surface (Fig. 3(b)). In the compensation composition of TbFeCo, the magnetic moments arranging anti-parallel configurations are equal, leading to the net magnetic moment, M, being zero. Moreover, the infinite H_c⊥ value would be achieved at this composition.^3,4) By considering the relatively large H_c⊥ values (5.7–6.4 kOe, Fig. 3(a)), in combination with relatively small M_⊥ and M_// values (25–50 emu/cm³, Fig. 3(b)), the films prepared with pre-sputtering time longer than 60 min would be in the vicinity of the compensation composition.

Fig. 3

(a) Dependence of Tb contents (●) in the films and out-of-plane coercivity (H_c⊥) values (△) on the pre-sputtering time. (b) Changes in the saturation magnetization values as a function of pre-sputtering time. △: Out-of-plane, and ●: in-plane magnetizations.

These results would be caused by the suppression of the oxidization of Tb in the film, rather than the film deposition conditions, through the pre-sputtering process for the sputtering target and Tb chips. We consider that the specimens prepared pre-sputtering over 60 min possess large effective amounts of constituents (e.g., Tb) because of the sufficient purification of Tb chips and the sputtering target. The sufficient pre-sputtering probably realizes the vicinity of compensation compositions of the film. However, the Tb content in the film based on Fig. 3(a) (26 at%) is much more than that of the compensation composition of Tb, which is reported to be approximately 22 at%.³⁾ Therefore, the detailed effective amount of elements in the thin film was determined through experiments.

3.2 Effective amount of elements in the films

We evaluated the chemical bonding state of each element as the film depth function by sputtering/etching repetition performed at approximately 0.66 nm per each cycle, corresponding approximately to 0.59 nm/min. The sum of XPS peak areas that correspond to Tb 4d, Fe 2p_3/2, and Co 2p_3/2 (△) and the XPS peak area of O 1s (●) are shown in Figs. 4(a) and 4(b), respectively. Figure 4(a) presents the results of the thin film deposited after a 3 min pre-sputtering, whereas the results in Fig. 4(b) show the thin film formation after the pre-sputtering time of 60 min. We compared the areas in both figures based on the normalized values, rather than the absolute values. There is a trend that the metal amounts (Tb, Fe, and Co) increased as the thin film deepened, although the O content decreased in relation to the film depth. This finding clarified that the effective amounts of elements increased at the deep part of the thin films from the film surface. Figure 4(c) shows the typical example for the calculation of O 1s peak area by using MultiPak V9, which is the specific software for the XPS area analyses. In this study, we subtracted the backgrounds of the XPS profiles and then calculated the XPS peak areas by using the software. Figure 5 shows the comparison of Fe 2p_3/2 (●) and Co 2p_3/2 (△) peak areas for the film after 60 min of pre-sputtering. Both peak areas increased when deepening the film thickness, suggesting that effective Fe and Co contents without bonding with O increased in the deep part of the thin film. This result is consistent with that shown in Fig. 4. Furthermore, the Fe 2p_3/2 and Co 2p_3/2 areas turned to be almost constant after the XPS-sputterings/etchings were repeated (i.e., 10 cycles for Fe 2p_3/2 and 9 cycles for Co 2p_3/2, respectively). Therefore, Fe and Co without O bonding exist in the deep part of the film than in tenth and ninth XPS-sputtering/etching cycles, respectively. Based on these considerations, pure Fe and Co (i.e., Fe and Co without O bonding) exist in deep part of the film at approximately 6.6 and 5.9 nm from the film surface, respectively. In other words, Fe and Co are partially oxidized at approximately 6.6 and 5.9 nm from the surface of the thin films.

Fig. 4

Changes in the chemical bonding states of TbFeCo films measured by X-ray photoelectron spectroscopy as a function of the sputtering cycle, i.e., film depth. ●: Normalized peak areas of O 1s, △: normalized peak areas of the sum of Tb 4d, Fe 2p_3/2, and Co 2p_3/2. The pre-sputtering time was (a) 3 and (b) 60 min. (c) Example for the determination of the XPS peak area of O 1s using the “MultiPak V9” software.

Fig. 5

Peak areas of Fe 2p_3/2 (●) and Co 2p_3/2 (△) as a function of film depth obtained by XPS measurements.

Among the film constituents, Tb is the easiest element to be oxidized.¹⁴⁾ Hence, we analyzed in detail the bonding states of O and Tb. The XPS spectra of Tb 4d in the films prepared after pre-sputtering for 3 and 60 min are compared in Figs. 6(a) and 6(b). The arrows in both figures show the peak positions of Tb 4d (i.e., pure Tb without O bonding). We determined the effective Tb content freed from the oxidation in the films by calculating the peak areas indicated by the arrows by using MultiPak V9. The dashed lines in the figures show the theoretical positions of Tb 4d peak reported in the XPS database.¹⁵⁾ The positions of Tb 4d peak are shifted to lower energy in both films, as well as all the film depths, than the reported Tb 4d positions. Not only Tb 4d but also Fe 2p_3/2 and Co 2p_3/2 spectra shifted to lower energy than those of the theoretical positions. Thus, we deduce that the present film must be charged during the XPS measurements. As Figs. 6(a) and 6(b) show, Tb 4d peak shapes for both films are not evident at the 4th XPS-sputtering/etching layer, which is close to the film surface. The reason is probably due to the atmospheric exposure of the films that form the natural oxidization layer of the film surfaces after the film deposition. The apparent Tb 4d peaks were observed deeper than the eight XPS-sputtering/etching cycles (i.e., deeper than approximately 5.3 nm from the film surface) for both specimens. The shapes of Tb 4d XPS spectra are similar to that of the theoretical shape when the XPS-sputtering/etching cycles were repeated more than 12 times, corresponding to ∼7.9 nm from the film surface. Therefore, we consider that pure Tb elements without O bonding exist deeper than the XPS-sputtering/etching cycle of 12 times. Based on these studies, the partial Tb elements are oxidized at approximately 7.9 nm from the film surface.

Fig. 6

XPS spectra of Tb 4d for various depths of TbFeCo films. The pre-sputtering time was (a) 3 min and (b) 60 min.

Afterward, the Tb 4d and O 1s areas are compared (Fig. 7). Regarding the areas of Tb 4d for the films prepared after 3 (●) and 60 min (△) of pre-sputtering, the effective Tb contents were increased when increasing the XPS-sputtering/etching cycles, as shown in Fig. 7(a). Consequently, the effective Tb amount increases as the film thickness is deepened. The inserted figure in Fig. 7(a) shows the change in the areas of the Tb 4d peak based on the XPS-sputtering/etching cycles of eight times (∼5.3 nm from the film surface) up to 21 times (∼13.9 nm from the surface). The approximated function of the film with the pre-sputtering time of 3 min, which is calculated from the inserted figure indicated by the dashed line, is y = 1.76x + 3.28. The correlation coefficient value of this function, |r|, is 0.756. In the case of the film deposited after the pre-sputtering time of 60 min, the function is approximated to be y = 2.33x + 1.20 with the |r| value of 0.926. By comparing with the inserted figure, the areas of Tb 4d are larger for the film with 60 min pre-sputtering than that of 3 min. This is noticeable from the approximated functions for both specimens. Therefore, the effective Tb content is larger in the case of a specimen with the pre-sputtering time of 60 min than that of 3 min. Based on the XPS experiments, we consider that the 3 min-pre-sputtering is insufficient for the surface purification of the sputtering target and Tb chips. Hence, Tb₂O₃ would be in the TbFeCo thin film. This result can be the reason for the drastic difference of magnetic properties, as shown in Figs. 2(a) and 2(c), for the films after 3 and 60 min of pre-sputtering treatments.

Fig. 7

XPS peak areas of (a) Tb 4d and (b) O 1s as a function of film depth. TbFeCo films were prepared after 3 (●) and 60 min (△) of pre-sputtering in both figures. The insets show the partial results for sputtering cycles from 8 times up to 21 times. The broken lines show the mathematical approximations mentioned in the text.

We also compared the areas of XPS-O 1s spectra for the specimens deposited after the pre-sputtering time of 3 (●) and 60 (△) min, as shown in Fig. 7(b). Similar to Fig. 7(a), the O 1s peak areas obtained from the XPS-sputtering/etching cycles at 8–21 times are shown as the insertion of Fig. 7(b). A remarkable decrease in the O content as the XPS-sputtering/etching cycles increase was concluded for both cases. The approximation function (denoted with dashed line) was y = 7856.4e^−0.13x with the value of |r| = 0.895 for the film deposited after the 3 min pre-sputtering. In the case of the film prepared after the 60 min pre-sputtering, the approximation function was y = 6966.7e^−0.14x with |r| = 0.807. As Fig. 7(b) and the approximated functions show, the film with the 3 min pre-sputtering time clearly shows the relatively large peak areas than that with 60 min. Thus, we consider that the O content in the thin film is larger for the sample with 3 min pre-sputtering than with 60 min. The results shown in Fig. 7 (i.e., decrease in the O content at the deep part of the thin film) agree with the results in Fig. 4. Note that the width of peaks and peak intensities of Tb 4d are much smaller than that of O 1s because the absolute values of Tb 4d areas are even smaller than in the case of O 1s shown in Fig. 7.

The systematic XPS studies revealed that the bonding states of Tb–O, Fe–O, and Co–O are different between the thin films with 3 and 60 min pre-sputtering times. Regarding Tb, which is quite easy to be oxidized, the sufficient purification of the sputtering target and Tb chips enables us to increase the effective Tb content (Figs. 6 and 7). Under the present film preparation conditions, the film deposited was removed from the sputtering chamber to the ambient air/environment, and then the sputtering chamber was evacuated again for the subsequent film preparations. This process leads to the oxidization of the target and Tb chips. In addition, the short pre-sputtering time is not sufficient for the removal of oxidation layers of the target and Tb chips. However, it would be sufficient after the pre-sputtering treatment over 60 min. Consequently, the effective amount of Tb in the film must be different as the function of the pre-sputtering time.

3.3 Magneto-optical properties

We measured the polar Kerr rotation angles, θ_K, which is derived from the magnetic Kerr effect, of the films formed after 3 and 60 min of pre-sputtering. The polar Kerr effect of the TbFeCo thin film is generally useful for the applications at the wavelength of ∼700 nm.^10,13,16) Figures 8(a) and 8(b) show the changes in θ_K with regard to the external magnetic field at the incident-light wavelength of 700 nm. Figure 8(a) shows the results when the light is injected from the film surface, while it is from the backside of the glass substrate in the case of Fig. 8(b). As Fig. 8(a) shows, we obtained the saturated θ_K values of ∼0.05° and ∼0.1° for the film prepared after the pre-sputtering for 3 and 60 min, respectively. The values are smaller than the reported-saturated θ_K value (0.3°)^10,13) in the general TbFeCo thin film. The decrease in the saturated θ_K value shown in Fig. 8(a) is caused by the partial oxidization of constituents in the film surfaces. Specifically, θ_K is mainly dominated by the effective Co when the incident-light wavelength is ∼700 nm, a long wavelength of the incident light.¹⁷⁾ This result implies that Co is partially oxidized in the thin film, leading to smaller saturated θ_K values than the theoretical value. These results are coincident with the results obtained by the XPS measurements, as shown in Fig. 5. However, as presented in Fig. 8(b), the saturated θ_K value is approximately 0.04° for the film deposited after the 3 min pre-sputtering, which is the result measured from the substrate side. Hence, it is not plausible that O diffuses through the entire film with the thickness of 65 nm (i.e., from film surface to the vicinity of the substrate). Therefore, explaining that the film preparation through insufficient purification of the target and Tb chips resulted in the deterioration of the θ_K value is possible because of the O existence in the thin film. Different magneto-optical properties owing to the pre-sputtering time should have a slightly different O amount in the films (Fig. 7). As Fig. 8(b) shows, the saturated-θ_K value is approximately 0.3° for the film prepared after the 60 min pre-sputtering, exhibiting the H_c⊥ value of 8.2 kOe from the Kerr hysteresis curve. Furthermore, thin film near the substrate side is less influenced by the oxidation after the film exposure to the ambient atmosphere because the film is composed of effective constituents freed from the bonding with O. The reason is the sufficient purification of the target and Tb chips was achieved through the pre-sputtering time of 60 min. These results and considerations agree with the present XPS studies (Figs. 5 and 7), thereby suggesting the existence of pure metal elements without O bonding in the deep part of the thin film.

Fig. 8

Comparisons of the magneto-optical properties of TbFeCo films after 3 and 60 min of pre-sputtering. Polar Kerr hysteresis loops measured at (a) the film surface and (b) the substrate side at an incident light wavelength of 700 nm. Changes in θ_r values as a function of incident light wavelength for measurements at (c) the film surface and (d) the substrate side.

Herein, we define the θ_K value at zero external magnetic field as θ_r, and we plotted the θ_r values in relation to the wavelengths of the incident light in the range of 250–900 nm in Figs. 8(c) and 8(d). In the surface of the specimens prepared after the pre-sputtering time of 3 (●) and 60 min (△) (Fig. 8(c)), the polarities of θ_r are positive in the range of short-wavelength region, whereas it became negative at the wavelength of 450 nm, probably due to the interference effect of light, which is generated by the oxidation layer forming on the film surface to flip the polarity of θ_r. For the substrate side (Fig. 8(d)), the same tendency was observed in Fig. 8(c) for the film deposited after the 3 min pre-sputtering. The reversal of θ_r was observed in the wavelength range of 350–500 nm. This result is the same with the case of the film surface, suggesting that the interference effect of light at the backside of the film is caused by the partial oxidation of the constituents in the film. On the other hand, the polarity of θ_r for the specimen deposited after the 60-min pre-sputtering is constantly negative, and the absolute value of θ_r is 0.3° for the wavelengths longer than 400 nm. Therefore, we consider that the O content in this sample is quite few in the vicinity of the substrate side. The θ_r value of 0.3°, which is the theoretical value of the TbFeCo thin film, would be achieved because the film constituents are mostly freed from the bonding with O.

3.4 Correlation between the effective amounts of constituents and magnetic moments

As Fig. 2(c) shows, H_c⊥ was determined to be 6.4 kOe in the film obtained after the pre-sputtering time of 60 min, which is not consistent with the H_c⊥ value obtained from the magneto-optical measurement (8.2 kOe). The different H_c⊥ values from magnetic and magneto-optical measurements would be probably caused by the difference in the measurement methods used (i.e., evaluation of “net” magnetic moments of the film through VSM and an evaluation of the reflected light from the “film surface” through Kerr measurements). Moreover, we observed the reversal of Kerr hysteresis loops regarding the magnetic hysteresis loops. According to previous reports, the reversal of the Kerr hysteresis loop is caused by several factors. Regarding the constituents of the TbFeCo thin film, (i) the amounts of 3d elements (FeCo) are dominant¹⁰⁾ or (ii) 4f element (Tb) is dominant.¹³⁾ Another plausible reason was (iii) the anomalous hysteresis loop induced by the inhomogeneity of the film quality.¹⁸⁾ Various studies mentioned many considerations regarding this issue. We found that the reversal/non-reversal of Kerr hysteresis polarities in the TbFeCo thin film is depended on the underlayer structures of the film even when the films were simultaneously prepared using the same sputtering target/sputtering chamber under the identical experimental conditions, although these experimental facts are not presented in the present paper. Thus, we emphasize that the identification of dominant magnetic moments in the film (i.e., 4f element or 3d elements), based only on the reversal/non-reversal of the Kerr hysteresis polarity from the magneto-optical properties, is not accurate. As Fig. 3(a) presents, we deduced that the magnetic moments generated by the 4f element must be more dominant than those of 3d elements in the present specimens. However, we discuss here the correlation between the effective Tb contents and overall magnetic moments in the film from all the considerations mentioned in this paper.

Figures 9(a) and 9(b) show the relationship between H_c⊥ and M_s⊥ in connection to the effective Tb content for the films prepared after the pre-sputtering time of 3 and 60 min. Figure 9(a) exhibits the case when the magnetic moments generated by the effective 4f element (Tb) are dominant. Figure 9(b) shows the case if the magnetic moments of the effective 3d elements (Fe and Co) are dominant in the present specimens. Both figures were depicted based on the literature.³⁾ If we assume that the effective 4f element is dominant in the current specimens, then H_c⊥ and M_s⊥ should be similar to Fig. 9(a) for the films deposited after the 3 (●) and 60 (△) min pre-sputtering. Hence, the effective Tb content must be larger for the film obtained when the pre-sputtering time is 3 min than that of 60 min. Notably, the effective Tb content is less for the film formed after the 3 min pre-sputtering than that of 60 min according to the aforementioned results and discussions. Thus, determining the relationship represented in Fig. 9(a) is difficult. Conversely, when we assume that the magnetic moments induced by the effective 3d elements are dominant, the relationship can be depicted in Fig. 9(b). In this case, the film prepared after the 60 min pre-sputtering is closer to the compensation composition, resulting in large effective Tb content, compared with that obtained after the 3 min pre-sputtering. This result is consistent with our present results. Thus, our thin films satisfy the relationship represented in Fig. 9(b) (i.e., 3d magnetic moments are dominant compared with that from the 4f element).

Fig. 9

Correlations between the out-of-plane coercivity (H_c⊥), saturation magnetization (M_s⊥), and the effective Tb content (The figures are based on Ref. 3). The dash-dot lines indicate the behavior of H_c⊥ and the broken lines show M_s⊥. The compensation composition of TbFeCo is indicated by arrow. (a) represents the case when the effective 4f (Tb) magnetic moments are dominant in the film rather than the effective 3d (FeCo) magnetic moments. On the other hand, (b) represents the case when the effective 3d magnetic moments are dominant in the film. ●: 3 and △: 60 min of pre-sputtering in both figures.

We consider that Fig. 3(a) shows the “overall” Tb content in the thin films, including Tb₂O₃ and effective Tb, leading to the sum of the Tb content being 26 at%. We expected that our films have 4f-dominated magnetic moments because the Tb content is 26 at%, which is a higher value than the compensation composition of Tb. However, by considering all the results obtained from magnetic, magneto-optical properties and XPS results, the effective Tb content in the film formed after the pre-sputtering time of 60 min would be approximately 22 at%, thereby supporting the relationship shown in Fig. 9(b).

4. Conclusions

In this study, we revealed that the pre-sputtering remarkably affects the oxidation state of each constituent of the TbFeCo thin film when we prepared the film by using the batch-type sputtering machine with a single target. Not only the oxidation state but also the magnetic properties are sensitively influenced by the oxidation state of the constituents. The detailed discussion on the correlation between effective elements and magnetic properties from the perspectives of magnetic, magneto-optical properties and XPS analyses were made in the current study.

In addition, the pre-sputtering performed prior to film preparation is essential to purify the surface of the sputtering target and Tb chips for the batch-type machine with a single target. We successfully established the sputtering condition of the TbFeCo thin film in the vicinity of the compensation composition with superior perpendicular magnetic anisotropy, which exhibits the H_c⊥ value of 6.4 kOe from M–H, as well as the H_c⊥ value of 8.2 kOe from θ_K–H hysteresis curves, respectively. The most appropriate pre-sputtering time under the present conditions is determined to be more than 60 min. The insufficient purification of the sputtering target and Tb chips induces the partial oxidation of Tb, leading to the formation of Tb₂O₃, which is one of the factors for the deterioration of perpendicular magnetic anisotropy of the film because of the decrease in the effective Tb content in the film.

We demonstrated that the pre-sputtering condition enables us to control the oxidation state of the constituents, i.e., the effective amount of elements, through the simple pre-sputtering process even when we use the non-load-lock sputtering system. In other words, controlling the magnetic properties of the materials is possible through the appropriate pre-sputtering conditions by using the batch-type magnetron sputtering machine with a single target. Hence, we obtained the high-quality TbFeCo thin film that exhibits in the vicinity of the compensation composition by controlling of the film oxidation, which avoids the decrease in the effective amount of film constituents. If sufficiently controlled, then, the TbFeCo film is available by proper pre-sputtering conditions by using a commonly utilized simple sputtering machine with high-reproducibility, not by extrinsic ways, such as an addition of elements. Finally, we could innovate in novel material applications that have never been realized in the TbFeCo system.

Acknowledgment

This work was supported by JSPS KAKENHI Grant Number JP16K06273 and Tanaka Kikinzoku Memorial Foundation. A part of this work was conducted at Advanced Characterization Nanotechnology Platform of the University of Tokyo, supported by “Nanotechnology Platform” of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

REFERENCES

• 1)   M.Y.  Teng,  C.Y.  Hsu,  K.S.  Liu and  I.N.  Lin: J. Magn. Magn. Mater. 239 (2002) 338–342.
• 2)   T.  Chen,  D.  Cheng and  G.B.  Charlan: IEEE Trans. Magn. 16 (1980) 1194–1196.
• 3)   M.  Mochida and  T.  Suzuki: IEEE Trans. Magn. 38 (2002) 2096–2098.
• 4)   M.T.  Rahman,  X.  Liu,  M.  Matsumoto and  A.  Morisako: IEEE Trans. Magn. 41 (2005) 2568–2570.
• 5)   D.  Bang and  H.  Awano: J. Appl. Phys. 115 (2014) 17D512.
• 6)   S.  Li,  H.  Nakamura,  T.  Kanazawa,  X.  Liu and  A.  Morisako: IEEE Trans. Magn. 46 (2010) 1695–1698.
• 7)   M.  Nakayama,  T.  Kai,  N.  Shimomura,  M.  Amano,  E.  Kitagawa,  T.  Nagase,  M.  Yoshikawa,  T.  Kishi,  S.  Ikegawa and  H.  Yoda: J. Appl. Phys. 103 (2008) 07A710.
• 8)   T.  Iijima and  I.  Hatakeyama: IEEE Trans. Magn. 23 (1987) 2626–2628.
• 9)   T.  Niihara,  F.  Kirino,  N.  Ogihara,  E.  Koyama,  R.  Sudou,  K.  Shigematsu and  N.  Ohta: IEEE Trans. Magn. 24 (1988) 2437–2442.
• 10)   T.  Kanaizuka,  T.  Ohwada,  Y.  Morihara,  T.  Katayama and  T.  Suzuki: J. Magn. Soc. Jpn. 11(Suppl. S1) (1987) 333–335.
• 11)   T.K.  Hatwar,  A.C.  Palumbo and  D.G.  Stinson: IEEE Trans. Magn. 24 (1988) 2775–2777.
• 12)   E.E.  Marinero,  D.C.  Miller,  A.E.  Bell,  A.  Gupta,  R.N.  Payne and  H.  Notarys: IEEE Trans. Magn. 23 (1987) 2629–2631.
• 13)   K.  Wang,  Y.  Huang,  Z.  Xu,  S.  Dong and  R.  Chen: J. Magn. Magn. Mater. 424 (2017) 89–92.
• 14)   F.  Kirino,  N.  Ohta and  F.  Kugiya: J. Japan Inst. Metals 58 (1994) 359–367.
• 15)   D.D.  Sarma and  C.N.R.  Rao: J. Electron Spectrosc. Relat. Phenom. 20 (1980) 25–45.
• 16)   C.  Xu,  Z.  Chen,  X.  Yang,  S.  Li,  D.  Chen,  S.  Zhou and  T.  Lai: J. Magn. Magn. Mater. 324 (2012) 2034–2038.
• 17)   H.  Tsujimoto,  M.  Shouji and  Y.  Sakurai: IEEE Trans. Magn. 19 (1983) 1757–1759.
• 18)   T.  Chen and  R.  Malmhäll: IEEE Trans. Magn. 20 (1984) 1025–1026.