Underlayer effect on the soft magnetic, high frequency, and magnetostrictive properties of FeGa thin films Free

Recent research has shown great potential for the voltage control of magnetism at the nanoscale in magnetoelectric (ME) or multiferroic materials and heterostructures.^1,2 This is motivated by the promise of next-generation electrical and electronic devices with a lower energy cost. Many applications that have received significant attention in recent years, exploiting ME coupling, include random access memory, spintronics, mechanically actuated antennas, and RF or microwave devices.^3–12 While there are many materials with either large magnetostriction or soft magnetic properties, the current bottleneck in achieving high efficiency and ME coupling in these devices is synthesizing ferromagnetic materials that exhibit simultaneously a large magnetostriction and soft magnetic properties in order to achieve a high magnetoelastic coupling.¹³ Furthermore, to achieve fast switching and low loss operations in ME spintronic and microwave applications, it is necessary to have a low Gilbert damping coefficient.^13–16 This functional property often requires materials engineering to realize as the large spin-lattice coupling that is typically responsible for high magnetostriction also results in high magnetic hysteresis and large Gilbert damping coefficients.¹⁷ 

Fe_xGa_1−x alloys have been of interest due to the high magnetostriction observed for bulk and polycrystalline alloys which makes them promising for integration in strain-mediated ME devices.^18–22 However, one of the barriers for high frequency applications of FeGa thin films has been their large ferromagnetic resonance (FMR) linewidths (∼620–700 Oe at X-band).^23,24 For sputtered FeGa thin films, it has been well documented that the structure, magnetic softness, and magnetostrictive properties can be heavily influenced by their deposition parameters.^25–27 Indeed, more recent works have shown that the fabrication of high quality epitaxial films can be used to achieve greatly reduced linewidths (∼80–220 at X-band).^28,29 Similarly, the addition of small interstitial atoms (e.g., C, B, and N) to FeGa thin films have been explored and found to promote excellent soft magnetic properties by reducing their grain size and diminishing their magnetocrystalline anisotropy.^20,30,31

One approach that has been previously explored to enhance the soft magnetic properties of FeCo alloys has been to use an underlayer material between the sputtered film and substrate.^32–41 The resulting enhancement in the soft magnetic properties was attributed to the refined grain size and impact of the stress at the interface on the magnetoelastic energy.^32,39,42,43 Similarly, for FeGa thin films, a recent study showed that a non-magnetic underlayer (Cu) can be used to improve its soft magnetic and high frequency characteristics.⁴⁴ In this work, the underlayer effect in FeGa with Ta (a non-magnetic material with a lattice constant different than that of Cu) and NiFe (a magnetic material with a lattice constant similar to that of Cu) was explored and compared to that of Cu. The choice of these materials is primarily to assess the effect of the lattice match at the interface with the FeGa film.

The films in this study were grown via DC magnetron sputter deposition using a ULVAC JSP 8000 sputter system with a base pressure of 2 × 10⁻⁷ Torr at room temperature with 4-in. targets. Si (100) substrates were used for all of the depositions without the removal of the native oxide. The FeGa films were sputtered using a target with 80/20 Fe/Ga composition at 200 W DC bias power and an Ar pressure of 0.5 mTorr; the Ta, Cu, and Ni₈₀Fe₂₀ (NiFe) underlayers were sputtered at 100 W DC bias power and an Ar pressure of 0.5 mTorr in the same chamber. SEM imaging was used to confirm the thickness of the films. The resulting composition (81% Fe and 19% Ga) of the sputtered FeGa films was determined via x-ray photoelectron spectroscopy (XPS) with a monochromated Al Kα source. The structural characterization of the films was determined via x-ray diffraction (XRD) using a Panalytical X’Pert Pro x-ray powder diffractometer with a Cu Kα source and fit with the Fityk software package.⁴⁵ Atomic Force Microscopy (AFM) imaging of the surface microstructure was performed using a Bruker Dimension Icon Scanning Probe Microscope with a Bruker RTESPA-300 AFM tip with an 8-nm nominal tip radius.

The room temperature magnetic hysteresis curves of the multilayers were measured via superconducting quantum interference device (SQUID) magnetometry using a Quantum Design MPMS3. The high frequency magnetic linewidth was measured using a short-circuited strip line connected to a vector network analyzer (VNA) with details described elsewhere.⁴⁶ For these measurements, the samples were placed facing the strip line and a large saturating magnetic field was first applied parallel to the strip line to establish a baseline for the measurement. The reflection coefficient (S₁₁) was then measured as a function of bias magnetic field (0–600 Oe) and frequency (100 MHz to 6 GHz). The magnetostrictive properties were performed by depositing FeGa, with and without an underlayer, on thin Si cantilever substrates (100 μm thickness) and utilizing an MTI-2000 fiber-optic sensor to detect the deflection of the cantilever tip due to changes in the stress of the FeGa thin films. Details are described elsewhere.³⁰ 

In this study, 100-nm FeGa thin films were deposited either directly onto Si substrates or with a thin 2.5-nm Ta, Cu, or NiFe underlayer. Figure 1 shows the in-plane magnetic hysteresis loops for the 100-nm FeGa films deposited on different underlayer materials normalized to the saturation magnetization. All of the films exhibited strong in-plane magnetic anisotropy. The FeGa film deposited directly onto a Si substrate, without an underlayer, showed a coercivity of 84 Oe. The coercivity of FeGa was reduced to 54 Oe when deposited onto a 2.5-nm Ta underlayer and further decreased to 17 and 15 Oe when deposited on 2.5-nm Cu and NiFe underlayers, respectively. These results follow a similar trend to that previously observed for Fe₆₅Co₃₅ films where a Ta underlayer resulted in a modest decrease in easy-axis coercivity while Cu and NiFe underlayers promoted a larger decrease.³² In addition, the FeGa films deposited with both Cu and NiFe underlayers displayed an enhanced uniaxial anisotropy, as observed from the increase in remnant magnetization (M_r) as summarized in Table I.

The high frequency characteristics of FeGa films deposited with different underlayers were studied using broadband FMR spectroscopy. Figure 2 shows the S₁₁ absorption as a function of the magnetic bias field (0–600 Oe) at a fixed frequency of 6 GHz. These are the cross sections of the entire FMR spectra collected for the frequency range of 100 MHz–6 GHz (see Fig. S1 in the supplemental material). For a 100-nm FeGa film deposited without an underlayer, the FMR spectra are characterized by a very low peak absorption (∼0.3%) and very broad FMR linewidth (>600 Oe at 6 GHz) that extends beyond the maximum magnetic field applied. For an FeGa film deposited with a Ta underlayer, a small enhancement in the FMR linewidth (∼465 Oe at 6 GHz) can be observed. In contrast, the FeGa films deposited on a Cu or a NiFe underlayer were characterized by a dramatic enhancement in the FMR response with linewidths decreasing to as low as ∼178 Oe and ∼160 Oe at 6 GHz, respectively.

The effective Gilbert damping coefficient, α_eff, is calculated and reported in Table I by fitting the FMR linewidth of the absorption as a function of frequency for the entire FMR spectra in Fig. S1 in the supplemental material to the following equation: $ΔH=2αeffωγ+ΔH0$⁠, where ω is the frequency, γ is the gyromagnetic ratio (≈2.8 MHz/Oe), and ΔH₀ is the frequency-independent linewidth broadening. The FeGa films deposited with Cu and NiFe underlayers show a significant decrease (∼75%–78%) in their effective Gilbert damping coefficient compared to an FeGa film without an underlayer.

The enhanced soft magnetic properties of the FeGa films grown on the Cu and NiFe underlayers must originate from the impact of the underlayer on its microstructure. The structural characterization of the FeGa films grown on different underlayers was first investigated with XRD. All of the FeGa films showed primarily a bcc (110) diffraction as the strongest diffraction line. Figure 3 shows the spectra highlighting the bcc (110) diffraction for a 100-nm FeGa film without an underlayer compared to those sputtered on Ta, Cu, and NiFe underlayers. The films deposited onto Cu and NiFe underlayers, which show the largest enhancement in their soft magnetic properties also displayed the largest shift of the (110) diffraction line position which is caused by a relative change in strain compared to FeGa deposited directly onto a Si substrate. Compared to FeGa deposited directly on a Si substrate, this shift in peak position represents a relative increase of 0.28% and 0.21% compressive film strain for the FeGa films on Cu and NiFe underlayers, respectively. This relative change in strain was calculated from the XRD data using Braggs law, $d=λ/2dsinθ$⁠, where a change in the relative strain between the two samples causes a shift, Δd, in the lattice constant: $Δε=Δd/d1=(d2−d1)/d1=sinθ2/sinθ1−1$⁠.

The FeGa films deposited on both the Cu and NiFe underlayers showed an increase (∼30%) in the intensity of their (110) diffraction peak compared to Ta or no underlayer, indicating an increased (110) polycrystalline texture. This is consistent with previous studies where a Cu buffer layer encourages a (110) crystalline texture along the growth direction for FeGa films.⁴⁷ This enhancement can be attributed to the close lattice match of the FeGa (110) (d = 2.06 Å) film texture to the underlying Cu (111) (d = 2.09 Å) and NiFe (111) (d = 2.05 Å) film texture which is highlighted in Fig. S2 in the supplemental material. In contrast to Cu and NiFe, Ta exhibits a preferential β-(002) diffraction at 33.7° (d = 2.66) that has a large lattice mismatch with FeGa.

AFM imaging (Fig. 3) was used to probe the differences in the microstructure of the FeGa films that can appear in their surface morphology when grown on the different underlayers. The surface roughness remained in the range of 1.1–1.4 nm for all of the FeGa films. More interestingly, the magnetically softer FeGa films deposited on Cu or NiFe underlayers exhibited a smaller and more uniform grain width distribution (29 ± 7 nm and 29 ± 6 nm, respectively) than the magnetically harder FeGa samples deposited directly on Si or with a Ta underlayer (46 ± 23 nm and 39 ± 14 nm, respectively).

The properties discussed thus far for the FeGa films deposited on the different underlayers are summarized in Table I. This serves to highlight the correlation and change in microstructure with the static and dynamic magnetic properties. Note that the calculated change in the bcc (110) peak intensity (ΔI₁₁₀) and the change in film strain (Δε) are reported relative to the FeGa films without an underlayer.

In order to study the impact of the thickness of FeGa on its soft magnetic properties with an underlayer, varying thicknesses of FeGa films were deposited using Cu as the underlayer. The rationale for Cu as the underlayer material is to decouple the effect of two magnetic phases present if NiFe were used (e.g., exchange effects across the interface which becomes a greater fraction of the magnetic volume at smaller FeGa thickness). For the same reason, all of the samples were also capped with a 2.5-nm Cu layer to reduce the oxidation of the FeGa layer that becomes a more significant fraction of the total volume at smaller thicknesses.

The normalized in-plane magnetic hysteresis for these samples is shown in Fig. 4. The saturation magnetization before normalization (not pictured) decreases linearly with the film thickness. A clear dependence of the coercivity on the FeGa thickness can be observed, where the coercivity decreases from ∼17 Oe for a 100-nm film down to ∼12 Oe for a 10-nm film. In addition, from the corresponding XRD spectra it can be seen that there is an increase in the linewidth of the bcc (110) FeGa diffraction peak as the thickness decreases (0.55° for a 100-nm film to ∼1.3° for a 10-nm film). This is indicative of a trend toward smaller grain size as the film thickness decreases.

The value of α_eff for the FeGa films on a 2.5-nm Cu underlayer as a function of thickness was determined based on the FMR spectra in Fig. S3 in the supplemental material. It was found that α_eff decreases from 0.053 to 0.004 for a 100-nm film compared to a 10-nm film. This trend, along with the decrease in coercivity, is summarized in Fig. 5. These trends are consistent with the previous studies on FeGa films where coercivity and α_eff increase with film thickness due to an increase in film roughness and inhomogeneity.^21,48

In order to obtain the magnetostriction measurements for the FeGa films, a perpendicular AC magnetic field is applied along the short axis of the silicon cantilever, while initially a constant 100 Oe bias field is applied in the long axis in order to saturate the magnetization and assess the full magnetostriction during the measurement. The magnetic field induced stress, b, is calculated from the deflection at the cantilever tip using the following relation:⁴⁹ $b=−dts2Es/3tfl2(1+vs)$⁠, where d is the deflection, t_s and t_f are the substrate and film thicknesses (100 μm and 100 nm, respectively), l is the distance between the clamping edge and the probe location (27 mm), and E_s and ν_s are Young's modulus and Poisson ratio of the Si substrate [169 GPa and 0.069, respectively, along the [110] in-plane direction for a Si (100) substrate⁵⁰].

For thin films, the magnetostrictive stress is the more relevant parameter to describe the magnetostrictive effects because the in-plane strain is prevented by the substrate clamping and thus one can measure only the stress; this also avoids the need to measure the elastic properties of thin films which can be difficult and cannot necessarily be assumed to be the same as the bulk. However, for comparison with other literature on magnetostrictive thin films, magnetostriction in terms of strain can be calculated from the relation: $λ=−23(1+νfEf)×b$⁠, where E_f and v_f are Young's modulus and Poisson ratio of the FeGa film which are approximated from the relation that $(Ef1+νf)=50GPa$⁠.⁵¹ 

From the data in Fig. 6, the FeGa film deposited without an underlayer reached a saturation magnetostriction of 99 ppm. The FeGa film grown on the Cu underlayer largely maintained the same magnetostriction, displaying a saturation magnetostriction of 95 ppm. Interestingly, the FeGa film grown on the NiFe underlayer showed a 27% increase in the saturation stress, reaching 125 ppm. While the literature values of magnetostriction reported for FeGa thin films can vary significantly across an order of magnitude, which may be due to differences in deposition parameters and measurement techniques,^31,52,53 the importance of the results here is to highlight that the enhancement in the soft magnetic properties of the FeGa films can be achieved without a trade-off in magnetostriction.

In summary, the effect of 2.5-nm Ta, Cu, and NiFe underlayers on the soft magnetic and microstructural properties of FeGa thin films was compared. It was found that up to an 82% decrease in coercivity and ∼78% decrease in effective Gilbert damping coefficient can be achieved with the optimal NiFe underlayer material. Both Cu and NiFe, which have a good lattice match to the FeGa films, influence the microstructure of the FeGa films by promoting an increased (110) polycrystalline texture, smaller grain size, and an increase in compressive film strain. Additionally, the films were able to maintain their high magnetostriction with an underlayer, making it an excellent material for application in both microwave and strain-mediated magnetoelectric devices.

See the supplementary material for the full FMR spectra of the FeGa films with different underlayers and the complete XRD spectra of the FeGa, Ta, Cu, and NiFe films.

We acknowledge the use of the fabrication facility at the Integrated Systems Nanofabrication Cleanroom (ISNC), the Nano and Pico Characterization Lab, and the Molecular Instrumentation Center (MIC) at the California NanoSystems Institute (CNSI) at UCLA. This work was also supported by the NSF Nanosystems Engineering Research Center for Translational Applications of Nanoscale Multiferroic Systems (TANMS) under the Cooperative Agreement Award (No. EEC-1160504).

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