Tuning static and dynamic properties of FeGa/NiFe heterostructures Free

Multiferroic materials have the ability to control magnetism via an electric field as opposed to using current, which is inefficient and causes resistive heating.¹ The search for the ideal single-phase multiferroic system is ongoing, but strain-coupled composite systems offer a promising alternative.² By interfacing a magnetostrictive material with a piezoelectric material, the magnetic state can be switched with low energy cost. These composite systems have a wide variety of potential applications, due in part to the ability to tailor their properties via the choice of magnetostrictive and piezoelectric materials.

Iron-gallium, Fe_100-xGa_x, where 12 < x < 25, is a promising magnetostrictive material for use in composite multiferroics due to its large magnetostriction of up to 275 ppm, high magnetization of up to 18 kG, and a saturation field below 100 Oe for epitaxial films.³ Additionally, FeGa has a high piezomagnetic coefficient of up to 2 ppm·Oe⁻¹ and a compliance (inverse of stiffness) of 1.33 × 10⁻¹¹ Pa⁻¹.⁴ Both of these contribute positively to the magnetomechanical coupling factor, k², which physically describes the ability of a material to convert magnetic energy into mechanical energy. FeGa has a magnetomechanical coupling factor of 0.4, the highest value for a magnetic material not containing a rare earth element. This, combined with its ductility, makes it a promising material for use in strain-coupled multiferroic systems. When boron is added to FeGa (forming Fe₇₀Ga₁₈B₁₂), the high-frequency properties are enhanced due to the reduction in magnetocrystalline anisotropy; this reduction in anisotropy is caused by the interstitial boron doping disrupting the crystal structure of FeGa.⁵ Doping has been shown to be a viable option for enhancing the high-frequency properties of FeGa but at the cost of the saturation magnetostriction and magnetization, which is undesirable for strain-coupled multiferroics.

Permalloy, Ni_100-yFe_y, where 17 < y < 24, is a magnetic material traditionally used in transformers and is characterized by very low loss when converting magnetic energy into electrical energy. This is because the magnetocrystalline anisotropy can be controlled easily via growth conditions as well as heat treatments.⁶ At the composition of 81% Fe and 19% Ni, NiFe has almost zero magnetostriction, which is generally considered a form of loss for a magnetic material in an alternating magnetic field.⁷ Thin film NiFe has been grown by various techniques and has been shown to possess uniaxial anisotropy and almost zero coercivity.⁸ NiFe is used as electromagnetic shielding for electronics due to its high permeability and lossless magnetic cores because of its low magnetostriction.

Much of the previous work interfacing hard and soft ferromagnets in thin film form has involved the idea of exchange spring magnetism. The thicknesses of these magnetic layers are on the order of the exchange length of the constituent materials (several nanometers) and are therefore strongly exchange-coupled.⁹ The coupling between the hard ferromagnetic (FM) layer and the soft FM layer reduces the energy required to flip the magnetization of the composite. One study has shown that interfacing FeGa and NiFe in the form of nanowires can affect the static magnetic properties of the bi-phase composite, specifically the coercivity and anisotropy field.¹⁰ In this study, the layer thickness was fixed at 20 nm for the NiFe and 60 nm for the FeGa, both far larger than the exchange length of either material (5.7 nm for NiFe and 3.3 nm for FeGa).^11,12 Both the number and thickness of the individual layers were found to affect the coercivity and magnetic anisotropy of wires, which is attributable to change in shape anisotropy. The coercivity of the heterostructures was found to be similar to that for pure FeGa nanowires; however, the magnetoelastic behavior was not reported for this system.

The impact of the interfacing different magnetic materials on the magnetostriction is critical for device integration. While there have been no studies on the magnetoelastic and high frequency properties of FeGa/NiFe multilayers, several other studies have explored such properties in similar material systems. For a system of (TbFe₂/Fe₃Ga)_n grown via DC magnetron sputtering, it was found that the magnetic properties were dependent on both the number of layers and their thickness.¹³ The composite films demonstrated a magnetostriction value of 550 με and a coercivity of 400 Oe.¹⁴ The hybrid material exhibited the complementary properties of the phases: the high saturation magnetostriction of TbFe₂ and the low coercivity and high magnetization of FeGa. The dynamic properties of magnetic multilayers were assessed for a trilayer of NiFe (5 nm)/FeCoN (100 nm)/NiFe (5 nm), which exhibited a significantly reduced coercivity of 7.8 Oe (from 18 Oe) along the easy axis.¹⁵ The addition of NiFe doubled the permeability compared to single-phase FeCoN, from 600 to 1200. Since the NiFe layers were not exchange-coupled to one another (because they were separated by 100 nm), they were likely coupled through the magnetostriction of the FeCoN layer. The reduction in coercivity and enhanced permeability was attributed to a decrease in the magnetic anisotropy dispersion (MAD); however, the authors looked at only single heterostructures. Another study showed that reducing the layer thickness in a Fe/NiFe material system enhanced the magnetic softness of the multilayers.¹⁶ Holding the total thickness constant, they divided the structure into more layers, up to 39 layers total. They indeed found that in the sample with the thinnest layers, the MAD was significantly narrower compared to that of samples using thick layers.

In this work, FeGa/NiFe multilayers were grown in a thin film form to take advantage of their complementary magnetic properties. The goal is to capture the high magnetization and magnetostriction of FeGa and the low coercivity, narrow linewidth, and high permeability of NiFe by interfacing the materials at the nanoscale. In this study, both the individual layer thickness and total number of layers were varied to observe the tunability of the static and dynamic magnetic properties.

Thin films of FeGa and NiFe were sputter-deposited at room temperature onto Si (100) and PMN-PT (011) substrates using a DC magnetron system with a base pressure of 2 × 10⁻⁷ Torr. Two 4-in. targets are used in the study: Fe₈₅Ga₁₅ (99.99% purity) and NiFe (Ni/Fe 81/19, 99.99% purity). Argon gas pressure and sputtering power to the target were optimized to be 0.6 mTorr (10 sccm) and 100 W to minimize residual stresses to −0.6 MPa and 86.0 MPa for FeGa and NiFe, respectively. This optimization was more critical for FeGa, as it can negatively impact saturation magnetostriction.¹⁷ With each target independently powered, growth rates of 3.5 nm/min and 3.3 nm/min were determined for pure FeGa and NiFe films, respectively. There was no change in the growth rates observed for FeGa on NiFe or vice-versa. Multilayers of FeGa/NiFe were grown with individual layer thicknesses ranging from 6 to 50 nm. In order to account for thickness-dependent demagnetization energy, samples were grown with a constant total thickness of 100 nm, and the total volume ratio between FeGa and NiFe was held 1:1. All multilayers are sandwiched between two FeGa capping layers, while the individual layer thickness was varied. The number of NiFe layers in the multilayer is used to denote the number of bilayers (BLs): 7 bilayers corresponding to 7 FeGa/NiFe bilayers on a FeGa bottom layer. The number of interfaces is two times the number of NiFe layers.

The composition of the multilayers was determined via X-ray photoelectron spectroscopy (XPS) with a monochromated Al Kα source, and its variation as a function of depth was further determined by XPS profiling with Ar⁺ ion sputtering. The structure of the films was determined using X-ray diffraction with a Cu Kα source. Room temperature magnetic hysteresis (M-H) curves of the multilayers were measured via superconducting quantum interference device (SQUID) magnetometry, able to detect a moment as low as 10⁻⁷ emu. A step size of 2 Oe was used to accurately determine the coercive field of each sample. Electron paramagnetic resonance (EPR) spectroscopy at the X band (9.6 GHz) was used to probe the ferromagnetic resonance (FMR) linewidth. Permeability studies were performed with a broadband stripline measurement system, similar to one described elsewhere.¹⁸ Frequencies of 1–8 GHz were swept while applying different magnetic biases of up to 800 Oe to the cavity. The permeability was extracted from the S₁₁ parameters using conformal mapping. To determine the magnetostriction, the multilayers were grown on PMN-PT (011) substrates that are able to provide an anisotropic strain of 750 με with applied voltage (TRS Ceramics^®). The samples were then electrically poled in situ at up to 400 V in the SQUID magnetometer. The saturation magnetostriction was extracted from the SQUID measurements by way of calculating the change in stored magnetoelastic energy after poling.

Single phases of FeGa and NiFe were deposited separately on Si (100) substrates to confirm their respective composition. XPS analysis of the individual phases indicated the compositions of Fe₈₅Ga₁₅ and Ni₈₁Fe₁₉. XPS depth-profiling of a multilayer structure confirmed that there is no measurable intermixing between the FeGa and NiFe in the multilayer structure. XRD measurements of the bilayer samples showed that all samples have peaks corresponding to the FeGa (110) peak and NiFe (111) peaks. The grain size of the films was found to decrease once individual layer thickness decreased below 10 nm.¹⁹ 

Figure 1 shows the static and dynamic magnetic properties of FeGa and NiFe as well as a 5 BL structure to illustrate their complementary magnetic properties. Single-phase FeGa was shown to have a saturation magnetization of 1150 emu/cc and a coercivity of 60 Oe, and NiFe was shown to have a saturation magnetization of 780 emu/cc and a coercivity of less than 1 Oe, both of which agree well with the values reported in the literature for sputtered films.²⁰ Additionally, the NiFe shows strong in-plane uniaxial anisotropy. Using EPR, the linewidth of single-phase FeGa was found to be 260 Oe, while that of single-phase NiFe was 30 Oe. For all of the bilayer samples, the saturation magnetization was found to be around 1050 emu/cc, higher than that predicted by the rule of mixtures based on the 1:1 ratio of FeGa:NiFe, indicating that the layers are strongly exchange-coupled. This effect is similar to that seen in exchange spring systems, where the resulting composite maintains the high magnetization of the soft phase and the high coercivity of the hard phase.²¹ However, studies of systems with more similar magnetic softness are limited.

The coercivity and FMR linewidth at 9.6 GHz are shown in Fig. 2 as a function of the number of bilayers, while maintaining a total thickness of 100 nm and a 1:1 volume ratio of FeGa:NiFe. The coercivity decreased with decreasing layer thickness, in an almost linear fashion, which can be attributed to the decrease in magnetic anisotropy dispersion. As the layers decrease in thickness, the distribution of the possible orientations of the magnetic domains narrows, with the majority lying in plane. This reduces the energy required to rotate the magnetization in plane. It also reduces the average crystallite size, which will reduce the overall magnetocrystalline anisotropy, enhancing the soft magnetic properties of these films.¹⁶ The FMR linewidth shows similar behavior, decreasing linearly until 7 bilayers (7 BLs), where the FMR linewidth increases dramatically. The broadening of the FMR linewidth in the 7 BL sample (6 nm FeGa/7 nm NiFe) can be attributed to the enhanced coupling of the FeGa layers due to the layer thickness being much closer to the exchange length. The 5 BL sample shows a very narrow FMR linewidth of around 35 Oe, close to the value for single-phase NiFe, which is 30 Oe. Both the coercivity and FMR linewidth show a similar trend, indicating that the 5 BL sample, corresponding to a layer thickness around 10 nm, is the optimal configuration in this study.

Broadband FMR measurements were performed on a 5 BL sample, and the biasing magnetic field was varied from 50 to 400 Oe. The real and imaginary components of the permeability were extracted via conformal mapping, the results of which are shown in Fig. 3.¹⁸ The sample shows prominent real and imaginary components of the permeability, indicating that the conductive losses of the film do not predominate the signal.

A small magnetic field of 50 Oe shows the highest permeability value of 700. Additionally, the curves shift to higher frequency with the applied magnetic field, from a f_FMR of 1.9 GHz at 50 Oe bias up to 6.6 GHz at 400 Oe, which is in agreement with Kittel's equation: $fFMR=γ(Hk+Hdc+ΔHeff)(4πMs+Hk+Hdc+ΔHeff)$⁠. FMR studies are critical to understanding the impact of layer thickness and number on the coercivity and anisotropy and are underway for these bilayer samples.

This material demonstrated strong sensitivity to magnetic fields as well low loss for a conducting film, but in order for it to be integrated into a strain-coupled multiferroic device, the magnetoelastic properties needed to be assessed. Magnetostriction was extracted from the M-H loops by integrating the area of the magnetic hysteresis loops with and without applied strain. The difference in the area represents the stored magnetoelastic energy (B₁₁), and from this, the magnetostriction can be extracted.²² 

Using this technique, the magnetostriction of single-phase FeGa was determined to be 60 ppm, which served as a point of comparison for the composites. Figure 4 shows the saturation magnetostriction and the saturation field as a function of the number of bilayers. The 7 BL sample, corresponding to the FeGa layer thickness of 6 nm, has the highest saturation magnetostriction of all samples measured, up to 40 ppm, which is 67% that of single-phase FeGa. The 3 BL sample shows the lowest strain response of the samples measured, corresponding to a saturation magnetostriction of 15 ppm. Values of saturation magnetostriction for 1–7 BL samples ranged from 15–40 ppm. The grain size for the 4 BL sample estimated from the XRD is 9.5 nm, where the FeGa layer thickness is 10 nm.¹⁹ This leads to an enhanced shape anisotropy, where the grains shift more coherently with applied in-plane strain compared to the samples with thicker layers, resulting in a larger saturation magnetostriction as well as a higher saturation field. The enhancement at 7 BLs is likely caused by the thinner layers being able to couple more effectively because they are approaching the exchange length of FeGa. This would also explain the larger saturation field, as the FeGa layers with higher coercivity can couple together more effectively.

While the saturation magnetostriction is an important figure of merit for strain-coupled composites, the saturation field must also be considered, especially for applications where high sensitivity of the strain response is required. The sensitivity of the strain response can be approximated by determining the ratio of saturation magnetostriction to saturation field. The ratio for the 5 BL sample is 2, the highest of all samples measured, indicating high strain sensitivity to applied magnetic fields.

This work demonstrated that FeGa/NiFe heterostructures can allow for the tuning of magnetic properties via the number of layers and their thickness. The 5 BL configuration, corresponding to a structure of (8.5 nm FeGa/10 nm NiFe)₅/8.5 nm FeGa/Si, showed the optimal values of FMR linewidth and coercivity in addition to a high permeability. Hybrid magnetic materials are a promising alternative to single-phase ferromagnetic materials as well as doped material systems for resonator or sensor applications. The low coercivity, high permeability, and high strain sensitivity of these samples make them extremely promising candidates for high frequency, strain-coupled multiferroic systems.

This work was supported by the NSF Nanosystems Engineering Research Center for Translational Applications of Nanoscale Multiferroic Systems (TANMS) under Cooperative Agreement Award No. EEC-1160504. This work made use of the UCLA Integrated Systems Nanofabrication Cleanroom (ISNC) and the UCLA Molecular Instrumentation Center (MIC).