Adjustment of Magnetic Characteristics for [Co/Pt] Multilayer Thin Films

1. Introduction

The study of thin ferromagnetic/heavy metal multilayer films is believed to be an urgent task of nanospintronics. The interest in such films fabrication and investigation is due to a number of structural and magnetic features that provide an unique set of magnetic properties. The first to mention is a pronounced anisotropy of magnetization with an easy axis directed towards normal the growth plane even in the case of very small film thicknesses [1]. The atomic ordering of the films provides specific magnetic ordering within Dzyaloshinskii-Moriya interphase exchange interaction mechanism [2]. The attributive elements of such interaction are the set of nano-size vortex magnetic domains (skyrmions) which are highly interesting objects for magnetic memory applications [3,4]. The magnetic properties of the films are retained at temperatures significantly exceeding room temperature. Finally, since noble metals (usually Pt) are used as heavy metal part of the multilayer structure such films possess improved corrosion resistance.

One prospective representative of ferromagnet/heavy metal system is a Co/Pt multilayer film [5,6,7,8]. These films with varied Co vs Pt thicknesses are widely investigated which is due to the highest technological and structural flexibility of the layers: minor changes in the deposition technology lead to significant modifications of the phase composition of the films. A wide variation in the composition, in turn, allows a wide variation in the magnetic properties and micromagnetic structure [6,7,8]. In this paper, we have studied Co/Pt films fabricated by electron-beam alternating evaporation of Co and Pt targets in a high vacuum. The films consisted of 10 layers of cobalt and 10 layers of platinum sequentially deposited on the surface of a Si substrate. The selection of evaporation modes ensures the formation of either non-homogeneous layers of a Co_xPt_1-x solid solution with a periodic composition modulation (varying x), or a multilayer structure of the Co/Pt type with blurred interfaces. We report on investigating the effect of such a phase composition modification on the magnetic properties and micromagnetic structure of the films. The conditions that ensure the formation of isolated micromagnetic domains (skyrmions) in a pre-magnetized film are discussed.

2. Experiment

The Co/Pt ferromagnetic films were formed in a vacuum chamber on Si substrates by electron beam evaporation at a temperature of 200ºC. Prior to the Co/Pt growth, a 2 nm thick Al₂O₃ layer was deposited on the surface of the samples to prevent Co diffusion into the wafer [9]. The Co/Pt growth was performed by alternately evaporating separate Co and Pt targets with an electron beam in a high vacuum of 3·10^-6 Torr. The layer thicknesses in the Co/Pt multilayer structure were controlled by a piezoelectric weighing sensor. In the present experiment the base thickness values were set to 4 Å for Co and 5 Å for Pt with a total number of repetitions of 10. The exact thickness value was adjusted using a tuning factor (tooling), which varied from t = 0.85 to t = 1.5. The value of t is a dimensionless tuning coefficient, which is the one of operating parameters in the technological equipment used. The tooling is responsible for the simultaneous change in the thickness of the Co and Pt layers. Since the tooling remained unchanged throughout the growth experiment, a structure of the type [Co(4×t)/Pt(5×t)]₁₀ was formed, and the layer thicknesses of Co and Pt in each of 10 layers were the same: 4×t Å for Co and 5×t Å for Pt. The entire composition of Co and Pt thus remained constant (although the total film thickness varied). The impact of the t coefficient at fixed compositions was considered in [6]. In this work, three values of the coefficient were considered: t = 0.85 – bilayer thickness is 7.5 Å (structure A), t = 1 – bilayer thickness is 9 Å (structure B) and t = 1.5 – bilayer thickness is 13.5 Å (structure C).

LIBRA 200MC transmission electron microscope (TEM, Carl Zeiss NTS GmbH, Germany) which operated at 200 kV was used to study the structural properties of the samples. Cross section samples were prepared using Fischione XTEM toolkit. At final stage of preparation samples were polished by 1 kV Ar⁺ ions at small angles according to the samples plane. Thus, we got extended wedge in cross-sections. This allowed us to find area of the same thickness along electron beam for every sample to be sure experimental conditions for amplitude and phase contrast were the same in every case.

The element distribution profiles within the deposited films were measured using X-ray photoelectron spectroscopy (XPS) method. Before measuring the CoPt films, the surface was cleaned by ion beam. During the cleaning procedure, the film surface was bombarded with Ar⁺ ions with energy of 1 keV and ion current of density 0.7 μA/cm² for 3 minutes. The ion beam axis was directed at an angle of 45° to the sample normal. The ion beam diameter was 20 mm. The necessity for the procedure was dictated by the presence of various adsorbates on the surface. The profiles were measured by recording the angular dependence of the Co 2p_3/2 and Pt 4f photoelectron lines with a step of 1°. At the next stage, the concentrations of chemical elements were estimated for each angle using the relative sensitivity coefficient method [10]. Rotating the sample changes the penetration depth of X-rays, so information is collected from a different depth of the sample at each angle. Thus, the angular dependence of the concentration can be recalculated as a function of depth using the mathematical apparatus of the angle-resolved X-ray photoelectron spectroscopy method [10,11,12]. The maximum penetration depth for the studied samples can be estimated as ~ 4 nm, which corresponds to at least 3-5 nominal periods of the multilayer structure.

The magnetic properties were estimated based on the results the anomalous Hall effect measurements at room temperature. The magnetic field was directed normal to a film plane. The Hall resistance R_H(H) is a nonlinear function of the external magnetic field and has both a component proportional to the magnetic field strength (B₀) and a component proportional to the structure magnetization R_s(M) [13,14]. For the studied structures, R₀ << R_s in the magnetization saturation region. Therefore, R₀ was considered to be zero, and the R_H(H) dependence is thus similar to the magnetic field dependence of magnetization. Comparison of the R_H(H) curves for different structures allows drawing conclusions about the modulation of magnetization when varying the process parameters. The full similarity between R_H(H) and M(H) dependences was confirmed in our earlier work [6].

The micromagnetic structure of the deposited Co/Pt films was studied by magnetic force microscopy (MFM) using a Smart SPM microscope (AIST-NT) according to the tow-pass technique involving a probe with a low magnetic moment [15]. Changes in the domain structure were made by applying a magnetic field of a PPP-LM-MFMR probe (Nanosensors) during MFM scanning. The magnetic moment of Nanosensors probe is sufficient to partly magnetize the scanned area of the film and to form skyrmions using the method described in [16,17]. To manipulate the magnetic domain structure, the samples were “magnetized”, which consisted in bringing the MFM probe to a minimum distance from the surface and moving it over a certain area (in a mode similar to MFM scanning). As a result of the impact of the probe’s magnetic field on the surface, a magnetization reversal of some region of the films with dimensions from 3x3 to 5x5 µm² was performed.

2. Results and Discussion

Figure 1 shows the bright-field TEM micrographs of samples A and C with the lowest and highest tooling correspondingly.

The layers of Co/Pt (dark contrast), Al₂O₃ (bright contrast) and Si can be visualized at the images. We note that the technologically set Co/Pt layer thickness is in a good agreement with the values obtained by TEM. The multilayer structure of the Co/Pt can be visualized in case of high bilayer thickness, however due to a polycrystalline structure of the film the multilayer contrast is quite nonuniform. In Sample A with the lowest t value the multilayer contrast was not visualized indicating that Co and Pt in sample A are highly intermixed.

Further investigation of multilayer structure was carried out via the element distribution profiles measurements performed by XPS technique. Figure 2 shows the profiles for samples A, B, and C, which differ in bilayer thickness. Since the upper layer of the sample was subjected to ion cleaning and its structure was changed by the ion beam, the data obtained for a distance of ≤ 1 nm from the surface were excluded from consideration. The TEM signal intensity profiles were plotted at the same graph for samples A and C. The results indicate that the Co and Pt layers in the studied structures are strongly mixed with one another. At the same time, nearly uniform mixing (with respect to the experimental error) is typical only for sample A with the smallest bilayer thickness (t = 0.85). No difference between Co and Pt content with respect to experimental error was revealed which is in the good agreement with vague bright-field contrast from obtained from TEM microgrpaph (Figure 2a). For two other samples, at an analysis depth above 1 nm, periodic oscillations of the Co×Pt composition are visualized. The oscillation period is approximately equal to the period of the multilayer structure (9 Å for sample B and 13.5 Å for sample C). We note that the oscillation period for sample C is in the very good agreement with the period of bright-field contrast oscillation revealed from the TEM micrograph (Figure 2c), evidencing that indeed the Co vs Pt composition is an oscillating function in this case.

We believe that strong mixing between the pure Co and Pt, which could have taken place during the film deposition, has led to formation of Co_xPt_1-x solid solution with a periodically modulated composition. In such case the modulation period is equal to the bilayer period of the multilayer structure. In addition to the period, the modulation depth (the average difference between the maximum of one component and the minimum of the other) also changes. For sample C with the greatest bilayer thickness (13.5 Å for t = 1.5), this parameter was 10 %, for sample B (t = 1), the composition modulation depth does not exceed ≈ 7 %, for sample A the modulation depth is below the experimental error of 2.5 %.

Figure 3 shows the magnetic field dependences of the Hall resistance, reflecting changes in magnetization. For structures A and B, the R_H(H) dependence is a closed hysteresis loop. Note that for sample B, the R_H(H) dependence exhibits a “wasp waist” feature (visual “narrowing” of the loop in the region of zero magnetization). A similar feature was reported for two-phase magnetic films [6]. The R_H(H) dependence for structure C differs significantly from the previous ones: a hysteresis loop is not observed, and the magnetic field of magnetization saturation H_S increases significantly compared to structures A and B (from ~ 1000 to ~ 4000 Oe). Figure 4 shows the MFM images of the studied structures. In the initial unmagnetized state, a labyrinthine domain structure with close domain sizes is formed in films A and B. For structure C with the greatest bilayer thickness, it differs significantly from the data for structures A and B: a scan of a smaller area allows one to reveal a domain structure, but the domain size is significantly smaller than that of structures A and B. The shape of the domains does not correspond to the labyrinthine domain structure. When part of the sample is magnetized by the magnetic field of the magnetic force microscope tip, dark areas with ~ 0.1 µm diameter appear on the MFM image (Figure 4d). These features were previously [6] interpreted as skyrmions based on Lorentz transmission electron microscopy data. The skyrmion density (N_Sk), defined as the number of skyrmions per unit area, is ~ 6 μm^-2 for structure B and 3 μm^-2 for structure A. For structure C, it is impossible to detect magnetic skyrmions. The research results are systematized in Table 1.

Let us proceed to a discussion of the obtained results. First of all, we note the similarity of the magnetic properties and micromagnetic structure for samples A and B (Figure 3, curves A and B, Figure 4a,b respectively). At the same time, the layer structure revealed for these two samples differs significantly: in the first case, a close to uniformly mixed Co_xPt_1-x solid solution is revealed, whereas structure B corresponds to a solid solution with a periodically modulated composition (Figure 1a,b). In the case of structure C with the greatest bilayer thickness (t = 1.5), the magnetic loop parameters and micromagnetic structure differ fundamentally from the results obtained for the samples A and B (Figure 3 and Figure 4), although the layer structure for samples B and C is similar (Figure 2b,c respectively). In other words, in the batch under study, samples A and B are close to one another in magnetic properties, whereas samples B and C are close in the structure of layers. We associate this discrepancy with the degree of mixing between the Co and Pt layers. Presumably, the resolution of the angle-scanning XPS method is approximately equal to the period of structure A (0.85 nm). For this reason, the modulation of the composition in Figure 1a is not resolved. In the TEM micrographs (Figure 1a) the modulation of the composition was not resolved neither which is believed to be due to low contrast resolution in polycrystalline films.

The calculated value of the modulation depth for structure C (10 %) is affected by the resolution capabilities of the methods used. The significant difference in the properties of structure C indicates that the modulation depth in this structure is significantly higher, and the film structure is closer to a multilayer [Co/Pt]₁₀ with highly blurred heterointerfaces. In the case of a decreased layer thickness for structures A and B (t=1 and t=0.85 respectively), the mixing between Co and Pt increases, and film structure is closer to a Co_xPt_1-x solid solution with a periodically modulated composition. The proximity of magnetic properties of samples A and B is determined by the proximity of their structure. In particular, in both cases the presence of skyrmions in magnetized Co-Pt films is observed, which is caused by the Dzyaloshinsii-Moriya interaction (DMI), leading to a special type of magnetic ordering [18,19,20]. A change in the degree of blurring of heterointerfaces significantly changes the value of the DMI constant, as well as the magnetic anisotropy constant [21]. This explains the change in the skyrmion density from structure to structure.

4. Conclusions

Thus, in the present paper, multilayer magnetic [Co(4×t)/Pt(5×t)]₁₀ films were grown and investigated. The adjusted growth parameter was the thickness of the bilayers (t) whereas the Co vs Pt composition remained unchanged. It was found that during the growth process, the Co and Pt layers were significantly intermixed, and the value of t makes it possible to control intermixing degree and the phase composition of the layers. Selecting relatively small values of t, a Co_xPt_1-x solid solution can be formed, whereas larger t values lead to a growth of a multilayer structure with blurred heterointerfaces. The variation of a structure type changes drastically the magnetic properties. In particular, skyrmions were observed in the films representing solid solutions. The density of skyrmions depends on the composition modulation depth, which was revealed earlier [6]. The bilayer thickness coefficient (t) has proven to be a very sensitive adjusting parameter, providing control over the magnetic properties and micromagnetic structure of CoPt layers.