Controlling and patterning the effective magnetization in Y₃Fe₅O₁₂ thin films using ion irradiation

Ion irradiation has previously been used to alter magnetic characteristics of various garnet materials.^1–4 These studies were mostly conducted on micron-thick films using heavy ions with MeV-GeV energies. With the increased availability of high quality thin films of Y₃Fe₅O₁₂ (YIG), studies on how light ions with low (keV) energies affect them are now possible.⁵ 

Nanoscale devices fabricated out of magnetic heterostructures are central to the emerging field of spintronics. A complicating factor in developing these devices is the interface between dissimilar materials. While depositing different materials is one way to create multicomponent structures, another avenue is to pattern regions using ion irradiation to alter local magnetic properties.^3,6 Magnetic textures formed through irradiation establish interfaces defined by varying magnetic fields rather than abrupt changes in materials, which should result in cleaner interfaces.

A promising material for patterned ion irradiation is YIG since it is an insulator with low magnetic damping, eliminating effects from conduction electrons and allowing for the sensitivity needed in linewidth measurements. YIG also has the benefit of being well characterized and is widely used in experimental as well as commercial applications.⁷ The method in this study offers the freedom to control magnetic properties over a wide range of applications; from affecting 4πM_eff with minimal changes to damping, to changing both significantly.

In this experiment, we used epitaxial 20 nm thin film YIG grown on a (111) oriented Gd₃Ga₅O₁₂ (GGG) substrate by off-axis sputtering as described in Ref. 8. A variety of instruments and techniques were used to characterize our YIG films before and after irradiation. Broadband ferromagnetic resonance (FMR) measurements were employed to determine the effective magnetization through the angular dependence of the resonance field, as well as Gilbert damping determined from the frequency dependence of the linewidth. Magnetic force microscopy (MFM) measurements were used to probe the changes in the saturation magnetization as a result of ion irradiation. Cavity based FMR measurements were conducted on samples with arrays of irradiated patterns requiring higher sensitivity than that provided by a broadband stripline.

Radiation patterning of YIG samples with Ga⁺ was performed with a FEI Helios Nanolab 600 Dual Beam Focused Ion Beam / Scanning Electron Microscope (FIB/SEM). Fluences ranged from 1.32 × 10¹² − 1.26 × 10¹⁴ ions/cm² at an energy of 30 keV. Samples requiring full areal exposure with He⁺ were sent to Leonard Kroko, Inc. for irradiation with fluences of 4 × 10¹⁴ − 2 × 10¹⁶ ions/cm² at an energy of 60 keV. These energies and fluences were chosen based on SRIM⁹ calculations such that most ions passed through the YIG and the density of vacancies created was comparable to samples irradiated with Ga⁺ using our in-house FIB.

Angular and frequency dependent data were taken using a broadband FMR spectroscopy setup equipped with a rotatable magnet capable of supplying DC fields up to 2 T. The sample was mounted on a microwave stripline connected to a Gigatronics 12000A, sourcing frequencies from 2 to 20 GHz. FMR signals were obtained through a field modulation technique where the transmitted microwave power signals were measured by a diode read out by a SR850 lock-in amplifier. Measurements were repeated as a function of polar angle by rotating the magnet relative to the normal of the fixed sample.

Magnetic force microscopy measurements were performed in a home built setup outfitted with a diamond cantilever where a micron-sized magnetic particle has been affixed to the tip. The sample and cantilever are immersed in a large (0.5 T) external field to saturate the magnetization of our film out-of-plane. We use a high coercivity material, SmCo₅, for our particle in order to maintain the freedom to align our probe moment with or against the sample magnetization. The cantilever is driven using a form of active feedback known as self excitation.¹⁰ We recorded the change in resonant frequency as the probe was scanned at a fixed height across a sample containing 2 μm $×$ 65 μm irradiated bars.

Single frequency, high sensitivity FMR measurements were conducted using a Bruker EMX X-Band electron paramagnetic resonance spectrometer at a cavity resonance of 9.664 GHz. Samples were mounted on a Teflon holder in a quartz tube and rotated so that the DC magnetic field was oriented normal to the YIG plane. With the goal of characterizing a micron-sized irradiated structure, we created an array of non-periodic, well-separated dots; having a diameter of 2 μm and a minimum center-to-center distance of 4 μm. This ensures minimal interaction, avoids unintended resonances due to a periodic structure, and gives a large enough signal to measure.

Ferromagnetic resonance offers a spectroscopically precise method of measuring interactions within magnetic materials.¹¹ These interactions in a ferromagnet can be framed in terms of various fields in which ferromagnetically ordered spins precess at a frequency f around an effective field, $H→eff$⁠. The effective field includes the external field, $H→$⁠, as well as internal fields such as anisotropy and the demagnetizing field. The resonant frequency of precession is related to the free energy density, F, by¹² 

where γ is the gyromagnetic ratio and M_s is the saturation magnetization, and is evaluated at the equilibrium orientation of the magnetization. This precession can be driven by a microwave field, $h→rf$⁠, perpendicular to $H→eff$⁠. The equilibrium polar and azimuthal angles of our film magnetization, θ and ϕ, are obtained from minimizing the free energy density¹³ 

where N is the demagnetizing tensor, K_u is the uniaxial anisotropy constant, $M^i$ is the directional cosine of the magnetization vector in the i-th direction, and K₁₍₂₎ is the first (second) order cubic anisotropy constant. The effective magnetization is related to the saturation magnetization and uniaxial anisotropy field, H_u = 2K_u/M_s, by 4πM_eff = 4πM_s − H_u.

A schematic of the measurement geometry used for broadband FMR measurements on 20 nm thick YIG samples where half of the film surface was irradiated with 30 keV Ga⁺ of various fluences is shown in Fig. 1 (inset left). From our SRIM simulations, we find that the peak implantation depth of the Ga⁺ occurs at 15.8 nm with a straggle of 7 nm. We extracted the resonant field, H_res, of the YIG film as the external field direction, θ_H, was swept from -10° to 100° in 10° increments (Fig. 1). Using Matlab^®, we fit our measured H_res(θ_H) to Eqs. 1 and 2, obtaining 4πM_eff for each fluence (Fig. 2a). Similar measurements were performed before and after He⁺ bombardment and are plotted in Fig. 2b. The helium ions travel completely through the 20 nm YIG film, resulting in a more uniform irradiation across the film thickness than the Ga⁺ samples.

The results shown in Fig. 2 demonstrate the ability to control the effective magnetization of our YIG films through irradiation. An increase in 4πM_eff is seen, reaching a peak value of 183 G for Ga⁺ and 424 G for He⁺. A loss of signal was observed at a fluence of 1.26 × 10¹⁴ and 2 × 10¹⁶ ions/cm² for Ga⁺ and He⁺ respectively, though further study is required to find a more precise cutoff fluence⁴ in each case.

An interesting result from this study is how irradiation affects magnetization dynamics, described by the Gilbert damping parameter, α. We performed frequency dependent measurements of the resonance linewidth for the completely irradiated He⁺ samples; which were chosen due to the larger signal, greater increase in 4πM_eff, and more uniform damage throughout the sample thickness. We see two main regimes, indicated by the shaded regions in Fig. 2b. The effective magnetization can be altered significantly with minimal changes to the damping (light shaded region), or the effective magnetization can be held nearly constant with significant changes to the damping (dark shaded region). This is a promising result for patterning magnetic textures for devices due to the nearly independent control of these two parameters.

A change in 4πM_eff indicates a change in 4πM_s and/or H_u. We performed MFM measurements due to their sensitivity to stray magnetic fields, arising purely from 4πM_s when the sample is fully saturated, in order to separate these contributions. Treating our cantilever magnetic particle as a dipole with moment $μ→$⁠, the magnetic force exerted by the sample on the tip is given by $F→=(μ→⋅∇)H→$⁠. In our measurement geometry, indicated in Fig. 3a, this force leads to a shift, Δf, of the cantilever resonant frequency¹⁴ 

In Eq. 3, f₀ is the cantilever resonant frequency in the absence of magnetic forces, k is its effective spring constant, and $Hzstray$ is the stray field due to the sample magnetization. This allows us to measure changes in the film saturation magnetization (Δ4πM_s) through monitoring the shifts in cantilever resonant frequency as we scan across magnetic landscapes.

Bars with dimensions 2 μm $×$ 65 μm were irradiated with 30 keV Ga⁺ at a fluence of 1.36 × 10¹³ ions/cm². A scan across the width of a bar is shown in Fig. 3b, where the magnetic particle is oriented anti-parallel to the external field (and therefore the sample magnetization). The decrease in frequency when scanning across the irradiated bar indicates that the saturation magnetization is reduced. This is because a region whose magnetization has been reduced by ΔM, in an otherwise uniformly magnetized film, creates stray fields equivalent to a lone region of magnetization $ΔM→$⁠. Therefore, its effect will be the same as a bar whose magnetization is parallel to the probe moment and will produce an attractive force that opposes the cantilever restoring force, thus reducing the frequency as the cantilever passes over the bar. To ensure that this effect was magnetic in nature, and not arising from surface charges or changes in topography, we flipped the magnetic particle moment and repeated the MFM scan. The inversion of the frequency change confirms the magnetic nature of the force since only the direction of the moment, $μ→$⁠, was changed.

Combining this result with our FMR measurements, we find that the increase in 4πM_eff is due to an increase in the magnitude of the anisotropy field, H_u. These results agree with the previous studies of similar systems that concluded 4πM_s is reduced through damage from irradiation^1,2,4,15 while giving rise to strain,¹⁶ which increases H_u.

An application of patterning 4πM_eff on a micron scale is shown in Fig. 4 where a non-periodic array of irradiated dots was measured using resonant cavity FMR. The effective field for the films in our geometry, which are saturated out-of-plane, is given by H_eff = H − 4πM_eff. Due to the larger demagnetizing field than the surrounding material, an increase in 4πM_eff results in a magnetic field well. Resonances confined to these wells result in discreet modes of localized precession (localized modes). The observed resonance fields for the localized modes agree well with micromagnetic simulations using code as in Refs. 17 and 18.

We have demonstrated control over the effective magnetization of a YIG thin film through ion irradiation. We find that 4πM_eff can be altered significantly with minimal changes to the damping, or that 4πM_eff can be held nearly constant with significant changes to the damping, depending on the range of fluences. This offers nearly independent control of these two parameters, which may prove useful for device fabrication. Patterns of micron-sized dots were successfully created using irradiation whose results agree well with theoretical expectations, demonstrating the feasibility of application to future micro- and nano-structures. Further studies are required to better understand the mechanisms involved in this change in effective magnetization, namely extracting the anisotropy fields and 4πM_s before and after irradiation, as well as their impact on Gilbert damping.

This work was primarily supported by the US Department of Energy (DOE), Office of Science, Basic Energy Sciences, under Award No. DE-FG02-03ER46054 (experiment design, device fabrication, data acquisition and analysis), and Award No. DE-SC0001304 (sample growth). This work was supported in part by the Center for Emergent Materials, an NSF-funded MRSEC under Award No. DMR-1420451 (sample fabrication and characterization). We also acknowledge technical support and assistance provided by the NanoSystems Laboratory at The Ohio State University.