Fabrication and broadband ferromagnetic resonance studies of freestanding polycrystalline yttrium iron garnet thin films

There has been great demand for flexible electronic materials and devices in wearable products and implantable systems^1,2 due to various applications, such as flexible display,³ Flexible Radio Frequency Identification (RFID),⁴ and artificial neural networks.⁵ In the last decade, researchers have put much effort into developing flexible electronic materials.^6,7 Meanwhile, it is equally important to develop flexible magnetic materials. In this regard, the ferrimagnetic insulator yttrium iron garnet Y₃Fe₅O₁₂ (YIG) has been widely studied in magnonics⁸ and spintronics^9,10 due to its high Curie temperature, long spin transmission length, and low Gilbert damping constant.^11–16 However, achieving flexibility for magnetic materials, especially YIG magnetic insulators, is rather challenging. Recently, researchers have successfully fabricated low-damping flexible YIG thin films on muscovite mica.¹⁷ Freestanding single-crystalline YIG membranes have also been produced through a lift-off process.¹⁸ Indeed, the demand for freestanding magnetic materials remains high, which provide a viable solution to achieve data storage and magnetic processing.¹⁹ 

In this article, a freestanding polycrystalline YIG thin film was synthesized by pulsed laser deposition (PLD) using a Sr₃Al₂O₆ (SAO) sacrificial layer. Here, water-soluble Sr₃Al₂O₆ (SAO), as a sacrificial layer, has become a powerful solution for freestanding film processing.^20–23 The magnetism of freestanding YIG film was confirmed by a vibrating sample magnetometer (VSM) and broadband ferromagnetic resonance (FMR) spectroscopy. In contrast to the heterostructure, after deposition, the static and dynamic magnetic properties of a freestanding polycrystalline YIG thin film changed significantly. We attributed this change to stress release during transfer to a thermal release tape (TRT). This work provides a new method for preparing freestanding YIG films.

Before growing the heterostructure, GGG substrates were soaked in a mixture of concentrated sulfuric acid and hydrogen peroxide for 10 min to remove the surface oxide surface layer. Then, the substrates were ultrasonicated in acetone and alcohol to clean the substrate surfaces. The deposition temperature in the PLD cavity was maintained at 700 °C and the oxygen pressure was set at 5 × 10⁻⁶ Pa using a coherent KrF excimer laser (248 nm) with pulses of 200 mJ at a pulse rate of 10 Hz to grow the SAO sacrificial layer; the temperature and pressure were set at 750 °C and 1 Pa, respectively, using a laser with 300 mJ pulses at a pulse rate of 5 Hz to grow the GGG and YIG layer. After deposition, the heterostructures were annealed at 800 °C for 4 h in air. We have grown 200 nm SAO, 10 nm GGG, and 300 nm YIG thin films.

The crystalline structure and the surface roughness of the as-grown and transferred YIG thin films were studied by x-ray diffraction (XRD, Rigaku SmartLab x-ray diffractometer) and atomic force microscopy (AFM, Bruker, Dimension Icon), respectively. The static and dynamic magnetic properties were investigated using a vibrating sample magnetometer at room temperature (VSM, Lake Shore 7404) and broadband ferromagnetic resonance (FMR, NanoOsc) spectroscopy at room temperature. For FMR measurements, the microwave frequency is set at 2–16 GHz and the microwave power is set at 10 dBm. The derivative of FMR absorption intensity was extracted using a sweeping magnetic field model.

A flexible polycrystalline YIG thin film was developed and processed as schematically shown in Fig. 1(a); the fabrication process involved film deposition and transfer. First, the SAO sacrifice layer, GGG buffer layer, and YIG film were deposited one after the other on the (111)-oriented GGG substrate, as detailed in Sec. II, forming a GGG/SAO/GGG/YIG heterostructure. Here, the lattice constants of YIG and SAO are 12.38 and 15.844 Å, respectively. There is a large lattice mismatch between YIG and SAO. To ensure good crystal lattice matching for YIG thin film deposition, a GGG thin buffer layer was applied between SAO and YIG films to reduce residual stress. Then, flexible TRT was pasted on the top of the SAO/GGG/YIG heterostructure, acting as a support layer to maintain the integrity of the GGG/YIG thin film after separation. Finally, the GGG/SAO/GGG/YIG/TRT stack was immersed into deionized water for about 30 min to dissolve the SAO sacrificial layer separating GGG/YIG from the GGG substrate and transferring it to the TRT. The resulting structure is very flexible; see the photograph of the bent YIG thin film on the TRT support layer [Fig. 1(b)]. Intermediate and partially enlarged photographs of the TRT/YIG/GGG layer under a high-resolution optical microscope are shown in Figs. 1(c) and 1(d), respectively. We noticed that the GGG/YIG bilayer film was clearly cracking and curling. The main reason for the film cracking and curling is that the bilayer film deposited on GGG/SAO after annealing is polycrystalline, and the growth orientation is inconsistent, which leads to the fact that different domains in the film experience different stresses on the SAO film. When the deionized water completely dissolves the SAO film, the stress on the bilayer film will be completely released. Due to the stress released by different domains not being the same, the stress difference at the grain boundary is maximized and the grain boundary causes irregular fractures after stress is released.

The crystal structure of the GGG/SAO/GGG/YIG heterostructure after deposition and TRT/YIG/GGG after transfer from the GGG substrate are characterized by x-ray diffraction (XRD), as shown in Figs. 2(a) and 2(c). The XRD pattern in Fig. 2(a) shows that the as-deposited YIG thin film is polycrystalline, which is consistent with the Selected Area Electron Diffraction (SAED) pattern shown in the inset of Fig. 2(a). Moreover, after transferring the YIG from the GGG substrate to the TRT support layer, only two reflections from YIG remain, shown in Fig. 2(c), demonstrating clean separation of the YIG film from the GGG substrate with the SAO sacrifice layer completely dissolved. Figures 2(b) and 2(d) show the AFM surface profiles of as-grown and transferred polycrystalline YIG thin films onto TRT substrates, respectively. The root-mean-square (rms) roughness of the obtained polycrystalline YIG thin film was 0.44 nm, demonstrating that the as-grown YIG thin film after growth was atomically flat. However, the rms roughness of the flexible polycrystalline YIG thin film exhibits a large surface roughness (0.89 nm), in accordance with the photograph shown in Fig. 1(c), which may be due to cracking and twisting of the YIG thin film after the release of stress. Since the GGG buffer layer with a paramagnetic property is extremely thin compared to the polycrystalline YIG film, the effect of the GGG buffer layer on it will not be considered in the following discussion of the magnetic properties.

To further analyze the functional properties of the transferred polycrystalline YIG thin film, the VSM measurement was used to measure the static magnetic properties of the as-grown and transferred YIG thin films on the TRT support layer, shown in Fig. 3. The measurements were carried out at room temperature with an applied magnetic field parallel and perpendicular to the film surface. Both as-grown and transferred films exhibit clear magnetic hysteresis loops, indicating that the transfer process does not chemically affect the YIG thin film.¹⁹ The normalized static magnetic hysteresis (M-H) loops of the as-deposited polycrystalline YIG thin film and the measured in-plane (IP) and out-of-plane (OOP) directions at room temperature are shown in Fig. 3(a). We found that an in-plane magnetic hysteresis loop obviously exists in the as-grown polycrystalline magnetic YIG thin film, and the magnetic moment of the thin film essentially has no out-of-plane components, which is consistent with a YIG thin film directly deposited on the GGG substrate.²⁴ In addition, the coercivity field H_c extracted from the in-plane magnetic hysteresis loop is 16.2 Oe, which is almost 20 times higher than that of a single crystal YIG thin film.²⁵ The increase in the coercivity field is mainly due to the polycrystalline nature of the YIG film. In contrast, the normalized magnetic hysteresis loops after lift-off, shown in Fig. 3(b), have obvious differences, which consist of an increase in the out-of-plane magnetic components and a decrease in the coercivity field (H_c = 12.6 Oe). Changes in the static magnetic properties of thin YIG films can be associated with relaxation of the deformation of a freestanding YIG film after the removal of the substrate constrain, which significantly weakens the magnetic anisotropy caused by the clamping effect of the substrate.^19,26 Before the lift-off process, the magnetic components of the polycrystalline YIG thin film are held in the plane by stress and the GGG buffer layer. When transferred to the TRT support layer, the stress on the YIG thin film is completely released, causing part of the in-plane magnetization return to the out-of-plane direction due to the magnetic anisotropy. In addition, it is obvious that the out-of-plane magnetic hysteresis loops change before and after lift-off.

Figure 4 shows the dynamic magnetic properties of the as-grown and transferred polycrystalline YIG thin films on the TRT substrate, measured by the method of broadband ferromagnetic resonance (FMR) spectroscopy. During the FMR measurement, the samples had covered caps on the coplanar waveguide. An external magnetic field H was applied parallel to the film plane and perpendicular to the microwave field. The microwave frequency is used in the range of 2–16 GHz, and the coplanar waveguide signal linewidth is 500 µm. The absorption derivative spectra of polycrystalline YIG thin films before and after lift-off at a frequency of 12 GHz are shown in Figs. 4(a) and 4(b), respectively. It has been observed that the signal is resonant at every frequency. Here, a Lorentz fit was used to obtain the peak-to-peak linewidth (ΔH_pp) of the film from the FMR signals. It was found that the linewidth ΔH_pp of as-grown and transferred polycrystalline YIG thin films on the TRT substrate is larger than that of a single crystal film. In addition, ΔH_pp = 112.99 Oe of the flexible YIG film increases compared to ΔH_pp = 88.40 Oe before lift-off. However, the intensity of FMR absorption actually decreases, which may be associated with an increase in the surface roughness of a flexible polycrystalline YIG thin film and a grain mismatch.²⁷ The general magnetic properties of a polycrystalline YIG thin film before and after lift-off are also obtained from the relationship between the extracted resonance field (H_res) and the microwave frequency, as shown in Fig. 4(c), which is described by the Kittel law,^28,29 for the in-plane FMR,

Here, H_k is the field of magnetocrystalline anisotropy, 4πM_eff is the effective saturation magnetization of the YIG thin film, and $γ$ = 2.8 GHz/kOe is the gyromagnetic ratio. Fitting curves in Fig. 4(c) are in good agreement with experimental data. The measured anisotropy field and effective saturation magnetization of the as-grown and transferred polycrystalline YIG thin films on the TRT layer were −16.92 and 1792.07 Oe and −68.87 and 1964.18 Oe, respectively. After the transfer of a polycrystalline YIG thin film onto the TRT substrate, the anisotropy field and effective saturation magnetization increased significantly. An increase in the anisotropy field, which is consistent with the aforementioned VSM measurement shown in Fig. 3, and the decrease in the resonance field are mainly due to the release of stress.

To determine the damping constant of polycrystalline YIG thin films, we investigated the peak-to-peak FMR linewidth (ΔH_pp) as a function of microwave frequency for a YIG thin film before and after lift-off. The results are shown in Fig. 4(d). The Gilbert damping α is obtained by the formula^30,31

where f is the microwave frequency, which is in the range of 2–16 GHz, and ΔH_pp,0 is the zero-frequency offset arising from long-range magnetic inhomogeneities. The damping constant α of the as-grown and transferred polycrystalline YIG thin films on the TRT layer is 9.45 × 10⁻³ and 5.4 × 10⁻³, respectively. Meanwhile, the zero-frequency offset ΔH_pp,0 of polycrystalline YIG thin films with and without transferring is 89.44 and 40.46 Oe, respectively. We found that the damping constant α and zero-frequency offset ΔH_pp,0 of polycrystalline YIG thin films are at least an order of magnitude larger than those of single crystal YIG thin films, and the damping constant α of the flexible polycrystalline YIG thin film is especially smaller than that of the YIG thin film without transferring. In contrast, The zero-frequency offset ΔH_pp,0 of the freestanding polycrystalline YIG thin film becomes larger, which is attributed to the inhomogeneity of the cracked flexible polycrystalline YIG thin film. We related the decrease in the damping constant α with stress release and an increase in magnetic anisotropy and surface roughness.^17,32

In summary, a GGG/SAO/GGG/YIG heterostructure was fabricated by PLD and a freestanding YIG thin film was obtained by dissolving a sacrificial layer of SAO, with both as-deposited and freestanding YIG films showing polycrystalline properties. The freestanding YIG thin film demonstrates outstanding mechanical flexibilities. The static and dynamic magnetic properties of the flexible polycrystalline YIG thin film were confirmed by VSM and FMR measurements. Compared to the GGG/SAO/GGG/YIG heterostructure, the flexible polycrystalline YIG thin film exhibits greater magnetocrystalline anisotropy and a lower damping constant, which was caused by the release of stress during the transfer of the polycrystalline YIG thin film. Thus, this work offers a viable solution for a flexible YIG thin film that can be used in a number of applications.

This work was supported by the National Key Research and Development Plan (Grant No. 2016YFA0300801); the National Natural Science Foundation of China (Grant Nos. 51702042, 61734002, 61571079, and 51672007); the National Key Scientific Instrument and Equipment (Project No. 51827802); and the Sichuan Science and Technology Support Project (Grant Nos. 2021YFG0347 and 2021YFG0091).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.