CFA Films in Amorphous Substrate: Structural Phase Induction and Magnetization Dynamics

We report a systematic study of the structural and quasi-static magnetic properties, as well as of the dynamic magnetic response through MI e®ect, in Co2FeAl and MgO//Co2FeAl single layers and a MgO//Co2FeAl/Ag/Co2FeAl trilayered ̄lm, all grown onto an amorphous substrate. We present a new route to induce the crystalline structure in the Co2FeAl alloy and verify that changes in the structural phase of this material leads to remarkable modi ̄cations of the magnetic anisotropy and, consequently, dynamic magnetic behavior. Considering the electrical and magnetic properties of the Co2FeAl, our results open new possibilities for technological applications of this full-Heusler alloy in rigid and °exible spintronic devices.


Introduction
The broad range of phenomena in the¯elds of spintronics and magnonics with potential for technological applications requires the development and optimization of ferromagnetic materials with spe-ci¯c static and dynamic magnetic properties. While systems with high damping parameter are explored in spin pumping and inverse spin Hall e®ect applications, 1-3 systems with low damping parameter present a remarkable importance for magnonics devices [4][5][6] and ultrafast sensors. 7,8 Regarding low damping materials, the Full-Heusler materials 9 composed by Co 2 FeAl (CFA) corresponds to an interesting alternative for injection and detection of spin-polarized currents, 10 even though a completely disordered structure is observed. 11 In particular, this alloy, besides the low damping parameter, can reach up to 100% spin polarized current at the Fermi level, justifying the strong interest on this material in the last years. At the same time, the static magnetic characteristics of CFA can be employed in distinct devices based on/or exploring the induced uniaxial anisotropy and/or magnetoelastic properties. 12 With respect to the magnetic properties of CFA full-Heusler alloy, a complex magnetic behavior has been previously reported. 13 Thus, the exploration of its applicability requires investigations of the magnetization dynamics in saturated and nonsaturated magnetic states of the samples, as well as, in resonant and nonresonant regimes. To this end, the Magnetoimpedance (MI) e®ect presents itself as an interesting tool. The MI e®ect corresponds to the changes in the electrical impedance of a ferromagnetic material when submitted to an external magnetic¯eld. The MI response is strongly dependent on the magnetic anisotropy and its orientation with the applied¯elds in the experiment, [14][15][16][17][18] bringing insights on the relations among structural and magnetic properties and magnetization dynamics.
The production of samples with controlled structural, electrical and magnetic properties is a hard task. The full-Heusler alloys can be described by X 2 YZ, where X and Y are transition metals, as Co, Fe and Ni, and Z is an element of the sp group, e.g., Al and Si. The well-known crystallographic structures L2 1 , B2 and A2, associated to the type of occupation of the three available lattice sites, are strongly dependent on the growth parameters and subsequent annealing procedures. 10 The obtainment of the aforementioned crystallographic structural phases are usually performed by growing the layers onto MgO or Si oriented substrates. 7,19 Although these substrates are rigid, this obtainment is an important step to the development of spintronic devices in°exible organic substrates, in which the employed substrate usually presents amorphous structure.
Motivated by these features, in this paper we present a new route to induce crystalline structure in the CFA alloy grown onto amorphous substrate. We explore the structural and quasi-static magnetic properties, as well as the dynamic magnetic response through MI e®ect, in CFA single layers and a tri-layered¯lm. We show that the structural phase of the CFA alloy onto amorphous substrate evolutes from an amorphous state to an A2 crystalline phase by the insertion of a MgO bu®er layer. This change in the structural phase leads to remarkable mod-i¯cations of the magnetic anisotropy and, consequently, dynamic magnetic behavior. Since the main features of the¯lms in glass amorphous substrates can be mirrored in nanostructures grown onto°exible substrates, 20,21 our results correspond to an important step to the employment of the CFA alloy in°exible spintronic devices.

Experimental Procedure
For this study, we produce¯lms with three di®erent geometries. The¯rst one is a Co 2 FeAl(500 nm) single layer. For the other¯lms, we consider a 5 nmthick MgO bu®er layer and grow a MgO//Co 2 FeAl (500 nm) single layer and a MgO//Co 2 FeAl (250 nm)/Ag(100 nm)/Co 2 FeAl(250 nm) trilayered lm. Figures 1(a)-1(c) presents a schematic representation of the produced¯lms. The samples are grown onto amorphous glass substrates with dimensions of 6 Â 6 mm 2 . They are deposited by the magnetron sputtering technique using the following parameters: base pressure of 6 Â 10 À7 Torr, Ar pressure during disposition of 10 À3 Torr, 90 W set in the RF source for the deposition of the MgO layer, and 100 W and 10 W in a DC source for the CFA and Ag layers, respectively. Using these parameters, the deposition rates are 0.02 nm/s, 0.56 nm/s and 0.18 nm/s for MgO, CFA and Ag, respectively. To induce magnetic anisotropy, an external magnetic eld of 100 Oe is applied in the plane of the substrate during the deposition process. In particular, for the production, the substrate is previously annealed at 450 C for 1 h. The MgO and Ag layers are deposited at room temperature, while the CFA layers are grown with the substrate at 150 C. The employed temperature is chosen based on a systematic study of the structural and magnetic properties for epitaxial CFA¯lms performed by Qiao et al. 22 The samples are characterized from the structural and magnetic point of views. The structural characterization is obtained through X-ray di®raction (XRD), experiments performed using a Rigaku MiniFlex system. The quasi-static magnetic properties are analyzed via magnetization curves, acquired with a LakeShore Model 7407 vibrating sample magnetometer, with maximum external inplane magnetic¯eld of AE 350 Oe.
Finally, the magnetization dynamics is investigated through MI measurements. The MI experiments are performed using an impedance analyser Agilent model E4991, with E4991A test head connected to a stripline, i.e., a microstrip in which the sample is the central conductor, as shown in Fig. 1(d). The real R and imaginary X components of the impedance Z are measured over a wide range of frequencies, from 0:3 GHz up to 3:0 GHz, in a linear regime with 1 mW (0 dBm) constant power, and magnetic¯eld varying between AE 350 Oe. The measurements are acquired with the¯eld in the plane of the¯lms in two con¯gurations: (i) Longitudinal MI, in which the probe current and external magnetic¯eld are applied along the easy magnetization axis of the sample and (ii) transverse MI, where both are applied transverse to the easy axis. In particular, these MI con¯gurations di®erentiate by rotating the sample, i.e., modifying the easy axis with respect to the H and I ac direction.

Results and Discussions
Here, we present and discuss the results related to the structural and magnetic characterization, as well as to the MI response of the¯lms. Figure 2 shows the XRD results for CFA¯lms. For the CFA single layer deposited onto glass substrate without the MgO bu®er layer, an amorphous structural character is evidenced. On the other hand, the¯lms for grown onto the MgO bu®er layer present the CFA [022] texture, assigned by the well-de¯ned and high intensity peak at 44:73 . Moreover, the trilayered¯lm, besides the Ag [111] texture, also shows the CFA [004] for one, identi¯ed by the presence of a small peak located at 65.12 .
Considering the CFA structural phases, the absence of [111] and [002] textures allows us to infer that, except for the sample without the MgO layer, the CFA alloy presents the A2 structure, since the [022] and [422] textures are present, associated with the complete chemical disorder among the Co, Fe and Al sites. 10,13,23 Due the thicknesses of the CFA and Ag layers, no information on the MgO layer is obtained. However, the insertion of this MgO bu®er is responsible for CFA structural orientation, given that the sample produced without the MgO layer presents amorphous structural character.
The magnetic properties of the CFA alloy are strongly dependent on the growth process. Thus, similar to the structural character, the magnetic anisotropy presents a dependence with sample geometry. Figures 3(a)-3(c) show normalized magnetization curves obtained for the CFA¯lms. For the CFA single layer, despite of a magnetic¯eld is applied during the deposition, isotropic in-plane magnetic properties are ver¯ed. This behavior can be associated with the stress stored during the deposition and with the low structural quality of this sample. Similar features have been previously reported for a CFA/Au/CFA trilayered¯lm grown onto amorphous glass substrate. 11 For the CFA single layer, a small coercive¯eld of H c % 2:0 Oe is found for both measurements, while the¯eld saturation H s is around 150 Oe, not shown here.
On the other hand, with the insertion of the MgO bu®er layer, the quasi-static magnetic behavior presents noticeable changes. The°atness of the MgO layer over the substrate enables the growth of samples with structural phase induction. Thus, the magnetic¯eld applied during the deposition, combined with the structural order and quality in the interface, gives rise to the magnetic anisotropy in the CFA layers.
For the MgO//CFA single layer, the angular dependence of the curves indicates an uniaxial in-plane magnetic anisotropy, induced by the¯eld applied during the deposition. The easy magnetization axis is identi¯ed in the measurement at 0 , for which presents a squared curve with coercive¯eld of H c % 5:0 Oe. For the measurement at 90 , a slightly tilted loop with smaller remanence appears once the magnetic eld is perpendicular to the easy axis. In this case the saturation¯eld is H s % 25 Oe.
For the MgO//CFA/Ag/CFA trilayered¯lm, a biphase magnetic behavior is found. The two-stage magnetization process is characterized by the magnetization reversal of one CFA layer at low¯elds, followed by the reversal of the other CFA layer at higher¯elds. In principle, the biphase magnetic behavior suggests that the CFA layers are uncoupled. The easy magnetization axis remains along the direction of the¯eld applied in the substrate during the deposition, as expected. Thus, this magnetic behavior can be associated with distinct magnetic properties of CFA layers. The CFA layer grown onto MgO bu®er layer keeps the soft magnetic properties, with switching¯eld of around 5:0 Oe for the measurement at 0 , while the other one grown onto the Ag layer presents harder magnetic properties, with switching¯eld of % 22 Oe. This fact re°ects the striking in°uence of the seed layer in which the CFA is deposited. The plateau located at around zero magnetization indicate an antiparallel alignment between the magnetization of the CFA layers. This is a¯ngerprint of the same magnetic contribution of both layers, and similar induced magnetic anisotropy direction of the CFA layer, despite they have distinct anisotropy constants.
For both samples produced using the MgO bu®er layer, uniaxial anisotropy is con¯rmed by the behavior of the M r =M s as a function of the applied ¯eld direction, as shown Fig. 3(d). The maxima in the trend curves indicate the easy magnetization direction, while the minima indicate the hard axis one. From a general point of view, the trilayered¯lm presents higher anisotropy induction, when compared with CFA¯lm grown onto MgO bu®er layer. It is related with the stress stored during deposition and its dependence with the CFA layer thicknesses. It is well-known that quasi-static magnetic properties play a fundamental role in the dynamic magnetic response and MI behavior. 24,25 The shape and amplitude of the MI curves are dependent on the orientation of the applied¯eld and ac current with respect to the magnetic anisotropies, magnitude of the external¯eld and probe current frequency as well as are directly related to the mechanisms responsible for the transverse magnetic permeability changes: skin and ferromagnetic resonance (FMR) e®ects. In particular, at moderate frequency range (MHz), the skin depth is smaller than t=2, where t is the thickness of the sample. This mechanism changes the transverse permeability of the sample and, consequently, the MI. At high frequency range (GHz), beside the strong skin e®ect, the FMR e®ect is present in the MI measurement, changing strongly the transverse permeability. 26,27 Based on that, experiments of longitudinal and transverse MI are investigated. Given that the CFA single layer has isotropic magnetic anisotropy, here just the results of the anisotropic MgO//CFA single layer and MgO//CFA/Ag/CFA trilayered¯lm are explored.
Using the microstrip con¯guration, where the sample is the central conductor, both electrical and magnetic contributions are obtained during the MI measurement. To make easier a direct comparison between the results obtained through the longitudinal and transverse MI measurements, the normalized Z is variation is considered. To this end, ÁR/R and ÁX/X are calculated through and where R(H) and X(H) are the real and imaginary components of the impedance at a given external magnetic¯eld (H), and R(H max ) and X(H max ) are the respective components at H ¼ 350 Oe, the maximum value of the external magnetic¯eld, where the samples are in the magnetic saturation state. This procedure is performed for each one of the frequencies employed in the experiment. Figure 4 presents the ÁR=R and ÁX=X normalized components of the impedance as a function of the external magnetic¯eld for the MgO//CFA single layer for the frequency range of 0:5 GHz up to 3:0 GHz, for the MgO//CFA single layer. Although the curves are acquired over a complete magnetization loop and present hysteretic behavior, here just part of the curve, when the¯eld goes from the maximum positive value to maximum negative one, is presented. For the transverse MI experiment [ Figs. 4(a) and 4(b)], where the external magnetic eld and electrical ac current are applied perpendicularly to the easy magnetization axis [see Fig. 1(d)], the curves present a double peak structure for the whole frequency range, as expected, 13,16 since the ac¯eld generating by the I ac current is parallel to the easy magnetization axis. At the intermediate frequency range, the R peaks position is close to the H s values veri¯ed through the magnetization curves. On the other hand, as the frequency is raised, the R and X peaks displace toward higher elds, a behavior veri¯ed above $ 2:0 GHz which is related to the increase of the FMR e®ect contribution to the MI variations. 18 For the longitudinal MI experiment [Figs. 4(c) and 4(d)], a clear change of the peaks structure with frequency is found, as expected. In this con¯guration, the easy magnetization axis is along the external magnetic¯eld and electrical current. Consequently, the R component of the impedance presents a single peak structure at the intermediate frequency range. As the frequency is raised, the FMR e®ect becomes mainly responsible for the MI variations, splitting the single peak in a double peak structure at $ 2:2 GHz [see horizontal dashed line in Fig. 4(c)]. Above this value, as the frequency is increased, there is the wellknown displacement of the peaks toward higher¯elds as the frequency is increased.
These results reveal the dependence of the dynamic magnetic response with the probe current frequency and magnetic¯eld. In this sense, it is possible to explore another representation of the same data by using the plot of the imaginary component as a function of the real one. 3,28 Figure 5 shows the ÁX/X as a function of ÁR/R for selected frequencies, for the MgO//CFA single layer. At this frequency range, below 1.4 GHz, we have identi¯ed that the strong skin e®ect changes the transverse magnetic permeability and, consequently, the MI variations. However, this analysis allows us to verify the frequency limit where the FMR e®ect starts acting in the sample and contributing for the MI e®ect. For both MI experiments, the FMR contribution is veri¯ed even for the lowest employed frequencies, $ 0:4 GHz. It can be found from the fact that the experimental points follow a circular trajectory. 28,29 This behavior is more evident in transverse MI experiment [ Fig. 5(a)], since the alternating magnetic¯eld H ac is along the easy magnetization axis, triggering the excitation of local magnetic anisotropies.
Similar analyses are performed for the MgO// CFA/Ag/CFA trilayered¯lm. They are respectively presented in Figs. 6 and 7. The insertion of the metallic Ag layer leads to a reduction of the electrical resistance of the whole sample. This feature improves the MI performance even the magnetic volume of the sample does not change.
Although the behavior related to the peaks structure and mechanisms governing the magnetization dynamics are similar, for the trilayered¯lm noticeable characteristics are revealed. As expected, the quasi-static magnetic properties are re°ected in the magnetization dynamics. At the intermediate frequency range, the peaks are located at $ 45 Oe in the transverse MI experiment [ Fig. 6(a)]. At the same time, the frequency limits where the FMR e®ect becomes the main mechanism responsible for the MI variations decrease considerably, when compared with the value found for the sample without the Ag layer. In this case, for the transverse MI experiment, the frequency is $ 1:5 GHz, while for the longitudinal MI one it is $ 1:8 GHz.
The limits between the frequency ranges in which distinct mechanisms act in the MI variations are related with the magnetic anisotropy of the¯lms. These magnetic features can be observed in the plot of the ÁX/X as a function of ÁR/R (Fig. 7). In this sample, the FMR e®ect contributes, irrespective of MI experiment (transverse or longitudinal), with the MI variation from 0:4 GHz. However, for both, the circular trajectories are not completely well de¯ned, a fact related to the imperfect alignment between alternating magnetic¯eld and magnetic anisotropy, due to anisotropy dispersion, di®erent than that observed in Fig. 5(a).
Finally, Fig. 8 shows the ÁR=R versus H at 0:4 GHz for the MgO//CFA single layer and MgO// CFA/Ag/CFA trilayered¯lm, together with magnetization curves, re°ecting the relation between the quasi-static magnetic properties and dynamic magnetic response. For both, the results represent the classical MI behavior for an anisotropic system when the electrical current and the external magnetic¯eld are applied transverse to the easy axis direction. In particular, the peaks are located close to the H s values found from the magnetization curves. Besides, for the trilayered¯lm, a plateau near zero¯eld is observed (see the red rectangle in the Fig. 8(b)), re°ecting the quasi-static magnetic properties for this sample. Despite the biphase magnetic behavior, no evidence of asymmetric MI e®ect is veri¯ed. This is a striking signature of the uncoupling of the ferromagnetic CFA layers in the trilayered¯lm. 25 From a general point of view, the features found for the CFA¯lms may be of interest for a wide sort of future applications. The manipulation of thicknesses, spacers, bu®er and seed layers in these geometries enables us to taylor the dynamic magnetic response and to tune the frequemcy limits in which the FMR e®ect emerges. At the same time, the insertion of the MgO bu®er layer allows us to obtain crystalline CFA¯lms with very interesting magnetic properties and rich magnetization dynamics. The integration of the present study with future works, in which the properties of the CFA alloy are explored in°exible amorphous substrates, is an utmost question for the employment of the CFA alloys in°exible spintronics devices.

Conclusions
In conclusion, we have investigated the structural features, quasi-static magnetic properties, and dynamic magnetic response through magnetoimpedance e®ect of Co 2 FeAl and MgO//Co 2 FeAl single layers and of a MgO//Co 2 FeAl/Ag/Co 2 FeAl tri-layered¯lm, all grown onto amorphous substrate. By considering a MgO bu®er layer, we have induced changes in the structural phase of the alloy, thus showing a new route to induce crystalline structure in the Co 2 FeAl alloy. The amorphous Co 2 FeAl single layer has isotropic in-plane magnetic properties. However, the MgO//Co 2 FeAl single layer has presented the well-known magnetic behavior veri¯ed in anisotropic systems with uniaxial magnetic anisotropy. Moreover, the MgO//Co 2 FeAl/Ag/Co 2 FeAl trilayered¯lm has revealed a biphase magnetic behavior, a feature previously veri¯ed in systems as NiFe/spacer/Co nanostructures. Thus, the evolution from an amorphous to an A2 crystalline structure has led to remarkable modi¯cations of the magnetic anisotropy and, consequently, in the magnetic dynamic response. In this sense, we have played with the structural phase, magnetic anisotropy and dynamic magnetic response of CFA¯lms. The fact that the main features of¯lms in glass amorphous substrates can be mirrored in¯lms grown onto°exible substrates places our results as an important step to the employment of the CFA alloy in°exible spintronic devices. We hope that our investigations motivate experimentalists to the development of this scienti¯c story.