3.1. Structural and Morphological Investigations of Fe Linear Structures
As described in Refs. [
20,
21,
23], Mo faceted linear structures with [001] crystallographic orientation can be prepared, this direction being favored at Mo/Fe interfaces as the easy axis of magnetization of Fe. Following similar deposition conditions, even some differences may occur due to the use here of a different deposition method [
32,
33,
34,
35], firstly a 5 nm thick Mo film was deposited on an Al
2O
3 and it was thermally treated after deposition, in order to obtain a planar flat surface. Then, a Mo layer having a thickness of 15 nm was re-deposited on the planar Mo surface while varying the deposition temperature, in order to obtain the arrays of linearly trenched structures [
22]. For all samples, we used these structures as a template for the growth of Fe and FeCo thin films. AFM, TEM images, and sketches of representative samples are presented in
Figure 1. The AFM images correspond to planar atomic surfaces of the Al
2O
3 substrate (
Figure 1a) and to the first wetting Mo layer (
Figure 1b). Linear patterns observed in the approximately diagonal direction of the image represent atomic steps corresponding to the Al
2O
3 substrate with an average periodicity of 35 nm, followed by the first thermal treated Mo layer.
Figure 1c shows the profiles of the structures over the lines presented in the two images. Beyond the AFM resolution, trenches were evidenced by TEM. In
Figure 1d the TEM images (inset: chemical contrast) in the profile plane of a Fe sample with a nominal thickness of approximately 3 nm and lateral periodicity of trenches of approximately 15 nm are shown. Fe molds on the structured Mo surface and accumulates deep inside the trenches.
3.2. Magnetic Characterization
Usually, two types of magnetic measurements are performed for the characterization of the nanosystems: (i) hysteresis loops for the investigation of the magnetic reversal mechanisms and (ii) specific measurements of zero-field-cooling—filed-cooling (ZFC-FC) procedures for the investigation of the magnetic relaxation mechanisms. The measuring geometry for the vectorial investigation of the magnetization reversal depends on the orientation of the applied magnetic field and will be defined as follows according to the inset of
Figure 2a: (i) longitudinal, with the magnetic field applied along Oz direction (along the wave), (ii) transversal, with the magnetic field applied along the Ox direction (perpendicular to the wave, but in the film plane) and (iii) perpendicular, with the magnetic field applied along to the Oy direction (perpendicular to the film plane).
According to the present samples, three types of nanostructures should be critically revised: (i) assembly of magnetic nanowires, (ii) assembly of magnetic islands/clusters, and (iii) undulated magnetic films. However, depending on the undulated film characteristics (lateral periodicity, deep and longitudinal uniformity) the last system can cover at the limit the first two cases, as well as the case of a quasi-uniform thin film. For example, undulated templates with deep well-formed trenches of high lateral periodicity and very low thickness of the magnetic film can lead to non-interacting filiform/quasi-cylindrical nanowires. Specific magnetization reversal mechanisms depend on the nanowire diameter [
36,
37]. If the lateral periodicity is decreased, the assembly of nanowires that interact by dipolar magnetic interactions is obtained [
38]. If the longitudinal uniformity of the nanowires is lost, separate segments of nanowires of different aspect ratios (nanocluster-like) are obtained [
39]. Depending on the separation gap and the transversal periodicity, such clusters can interact along the chain (unidimensional magnetic systems) as well as between chains (bi-dimensional magnetic systems). Moreover, depending on the shape and size of the magnetic clusters, they can become superparamagnetic above a certain temperature [
40]. Depending on the wetting conditions and type of magnetic materials, a very broad distribution of magnetic clusters (both in size and aspect ratio) can be formed along the trenches. If the inter-chain magnetic interaction is mediated by the finest clusters which become superparamagnetic above a certain temperature, a unidimensional magnetic system is expected above this temperature and a bi-dimensional one, below. Magnetic relaxation and blocking temperatures of possible magnetic clusters can be observed in this case by typical ZFC-FC measurements.
In order to interpret the magnetic measurements of the prepared systems, firstly theoretical expectations on the magnetization reversal mechanism obtained by micromagnetic simulations will be briefly discussed in the case of some ideal systems derived from undulated magnetic thin films. The magnetization reversal of an undulated Fe thin film with a thickness of 3 nm, deep of the valley of 5 nm, and lateral periodicity of 15 nm is shown in
Figure 2a. The simulations have been performed using the OOMMF (Object Oriented Micromagnetic Framework) public domain software, developed by the Applied and Computational Mathematics Division of National Institute of Standards and Technology (NIST) [
41]. The considered geometry and the discretization space are presented in the inset. Half of the loops (from the highest positive applied magnetic field to the highest negative field) are shown. The three-magnetization reversal effects presented in
Figure 2 correspond to the three above-mentioned directions of the applied magnetic field: (i) field applied along the wave (longitudinal geometry), (ii) field applied perpendicular to the wave but in the film plane (transversal geometry) and (iii) field applied perpendicular to the film plane (perpendicular geometry).
In longitudinal geometry, it is observed the typical jump of the relative magnetization (current magnetization relative to the spontaneous magnetization which can be experimentally approximated by saturation magnetization at low temperatures) from 1 to −1, specific to a magnetization reversal with the field applied along a magnetic easy axis. In this case, the easy axis is along the wave (Oz axis). On the other hand, the magnetization reversal in perpendicular geometry is linearly increasing with the applied field, according to a coherent spin rotation typical to a hard magnetic axis. Finally, in transversal geometry, the magnetization reversal approaches well a coherent spin rotation, except for a narrow interval of low fields (order of magnitude lower than the switching/jumping field in longitudinal geometry) where a fast reorientation of the spins take place. The situation is not similar to a magnetic field applied perpendicular to a narrow angular distribution of easy axes due to the rectangular shape of the loop in longitudinal geometry (field applied along the easy axis). Moreover, due to the lower value of the switching field in longitudinal geometry with respect to the saturation field in perpendicular geometry (that means with the film applied along an easy and hard axis of magnetization), the magnetization reversal mechanism in undulated thin films has to be represented by magnetic walls displacement anywhere a switching field is evidenced (e.g., in both longitudinal and transversal geometry) [
36,
40]. At variance to the film with a deep of the valley of 5 nm where the switching field in longitudinal geometry is about 6000 Oe, in the case of the film with a lateral periodicity of 15 nm and deep of the valley of 3 nm, the switching field in longitudinal geometry is about 4000 Oe whereas in transversal geometry the jump in magnetization is larger and at a switching field closer to the switching field in longitudinal geometry. At the same time, the coherent rotation in perpendicular geometry develops much harder (i.e., lower magnetic susceptibility) than in the former case. To note that in a planar geometry (0 nm deep of the valley), with in plane magnetic anisotropy, a full magnetization reversal from 1 to −1 appears at the same very low field of order of hundred Oe in both longitudinal and transversal geometries whereas the coherent reversal is the hardest. At this point, one may conclude on undulated thin films of Fe (and by extension of any soft ferromagnetic film) that the magnetic reversal mechanism depends not only on the direction of the magnetic field with respect to the structuring wave, but also on the wave characteristics with respect to the film thickness. In the case of a lateral periodicity of about 15 nm as in the present case, for nanometer-sized valley deeps larger than the film thickness, a strong in-plane uniaxial anisotropy (shape related) with the easy axis along the waves is evidenced by micromagnetic calculations. The magnetization reversal in longitudinal geometry takes place at a switching field higher than a few thousand Oe, via a fast domain wall displacement. In transversal geometry (with the field applied along the film but perpendicular to the wave), a coherent rotation takes place initially, followed by a final switching by domain wall displacement in lower negative fields of the order of hundred Oe. The coherent spin rotation which is in work in perpendicular geometry is saturated in fields higher than 10 kOe. By decreasing the valley deep down to the film thickness, the discrepancies between the magnetic reversal loops in longitudinal and transversal geometries become lower and lower, involving however almost similar switching fields of the order of kOe. The saturation of the coherent spin rotation in perpendicular geometry becomes harder and harder, involving applied magnetic fields of many tenths of kOe.
Besides, the magnetization reversal of magnetic clusters depends on both the size and the shape of the clusters. Ellipsoidal magnetic nanoparticles of enough low size and aspect ratio with magnetic monodomain structure can be superparamagnetic above the blocking temperature whereas below the blocking temperature the magnetization reversal respects the Stoner-Wohlfarth mechanism [
41]. Rectangular hysteresis loops are obtained with the field applied along the easy axis, the switching field being dependent on the shape anisotropy energy and taking values of hundreds of Oe in the case of ellipsoidal Fe and FeCo nanoparticles of high aspect ratios. A coherent spin rotation takes place with the magnetic field applied perpendicular to the easy axis with the saturation field equating to the switching field. Given the trenched structure of the template, ellipsoidal magnetic nanostructures may be dominantly formed in case of unappropriated wetting of the deposited material. In fact, this limiting case can be obtained starting also from the unidirectional case of filiform configurations along the trenches. Accordingly, in the peculiar case of undulated templates with trenches of high lateral periodicity, non-interacting filiform nanowires (deposited only in the valley) can be obtained for the deposited effective thickness of the film lower than half of the deep of the valley. The magnetization reversal of the assembly respects the magnetization reversal of a nanowire, which is strongly dependent on its diameter. Coherent rotation behavior (Stoner-Wohlfarth like) is specific to nanowires of soft magnetic materials (Ni, Fe, and even Co) with diameters lower than 4–5 nm whereas for higher diameters other mechanisms are in work (transverse wall mode, curling mode, etc.). However, in all cases a distinction can be made in the reversal loops obtained with the applied field along the easy axis (longitudinal geometry in the present case) and perpendicular to the easy axis (either transversal geometry or perpendicular geometry in the present case). Rectangular loops are specific to longitudinal geometry and linear ones to transversal and perpendicular geometries (with saturation field equating to the switching field in case of only coherent rotation and a higher saturation field than the switching field in case of transverse wall mode). If dipolar interactions take place between the nanowires, deviations from both the rectangular and linear variation of the magnetization versus field may appear. Such much-rounded shapes of the loops can be also due to local discontinuities of the nanowires, leading to the distribution of magnetic segments, easily associated with a distribution of ellipsoidal nanoparticles of different sizes and aspect ratios.
Moreover, before the presentation of the experimental vectorial magnetic reversal curves depending on the direction of the applied magnetic field with respect to sample-related fixed axes, it is worth mentioning some important specificities for its interpretation. There are well-known pitfalls of SQUID measurements in the case of magnetic measurements on non-uniform magnetic configurations of extremely low magnetic moments [
42,
43]. It was proved that due to different filling factors of the SQUID pick-up coils, large variation in the measured signal (e.g., the total magnetic moment at saturation) can be observed depending on the orientation of the sample with respect to the coil geometry (such discrepancies are even more evident in case of nanostructures grown on diamagnetic substrates). Usually, the pick-up coils are concentric to the superconducting coil which generates the applied magnetic field, and hence the measured signal will depend on the orientation of the magnetic structure with respect to the field. Therefore, not the values of the saturation magnetization in different geometries will be critically discussed, but the overall vectorial behavior (the shape) of the hysteresis loops at different orientations of the applied field with respect to the nanostructured samples. It should be noted that the diamagnetic contribution of the substrate was properly subtracted in the case of the hysteresis loops presented in the next sections.
3.2.1. Magnetic Characterization of Fe Undulated Systems (with an Effective Thickness of 3 nm) Deposited on Mo Trenches
Magnetic properties of Fe nanostructures, with a thickness of 3 nm—as derived from the TEM image (
Figure 1d), i.e., the magnetic anisotropy, the magnetic moment, and the easy axis of magnetization are determined from the observed magnetic behavior at various orientation of the applied field versus sample plane and a fixed in-plane direction as well as at different temperatures. Magnetic hysteresis loops acquired at 10 K and at different orientations of the applied field with respect to the structured undulated plane, as well as at different temperatures are presented in
Figure 3a and
Figure 3b respectively.
Neglecting the values of the saturation magnetization (due to the pitfalls of such specific SQUID measurements) two main observations should be underlined from
Figure 3a: (i) rounded hysteresis loops of similar shapes and coercive fields of about 280 Oe are obtained in transversal and longitudinal geometry at 10 K (no dominant easy axis specific to strongly undulated thin films) and (ii) a relative fast saturation within a not opened loop in perpendicular geometry. According to the above discussion, the only situation for such a behavior can be explained by a quasi-discontinuous undulated thin film formed by three-dimensional islands of Fe of nanometer size (about 5 nm), which strongly interact at low temperature. Hence, at 10 K, the magnetization reversal will be specific to a combination between the long-range magnetic structure of an undulated thin film with an equivalent thickness higher to the valley deep (whit negligible difference between magnetization reversal in transversal and longitudinal geometry) and an assembly of quasi-spheroidal nanoparticles which open the low coercive loop specific to the magnetic frozen state below the blocking temperature. As a consequence, due to nanoparticle-related components, the similar reversal loops in longitudinal and transversal geometries will present a relatively low coercive field as compared to a simple undulated thin film whereas. Also, the magnetic reversal in perpendicular geometry will present a higher magnetic susceptibility in low fields, followed by an apparent saturation due to the compensation of the diamagnetic susceptibility of the substrate with the positive susceptibility of the undulated film.
By increasing the temperature, the finest nanoclusters (smallest Fe islands) become progressively superparamagnetic, canceling the interactions between the larger clusters. As a consequence, the long-range magnetic order specific to the undulated thin film structure disappears and the coercive field in the longitudinal magnetization reversal (specific to only an assembly of non-interacting magnetic nanoparticles) decreases progressively with the temperature while more and more nanoparticles become superparamagnetic (
Figure 3b) This behavior is also nicely supported by the ZFC-FC curves presented in
Figure 3c. The ZFC-FC procedure was carried out by measuring the magnetic signal with an applied magnetic field of 200 Oe while increasing temperature, after the previous cooling of the sample in a null magnetic field. As the temperature is increased, a progressive increase of the magnetic moment is observed due to a size distribution of the magnetic entities (and implicitly of the magnetic moment) as a consequence of their deblocking followed by their orientation towards the applied field. The range of temperatures at which the magnetic moment is increasing corresponds to the distribution of sizes of the magnetic entities and thus, of the blocking temperature. An average blocking temperature of about 136 K (in longitudinal geometry) was estimated at the maximum of zero-field cooling (ZFC) curves. From the variation of the coercive fields in
Figure 3b with the
a final blocking temperature of about 150 K was determined. Despite the large error in the determination of this value due to the lack of experimental points, it is in good agreement with the blocking temperature estimated from the ZFC-FC curves. To note the very slow decrease of the curves above the branching point, proving a very broad size distribution of the Fe islands in the film. The magnetic moment measured at room temperature supports also the existence of magnetically frozen nanoparticles even at room temperature.
3.2.2. Magnetic Characterization of FeCo Undulated Systems (with an Effective Thickness of about 3 nm) Deposited on Mo Trenches
For comparison, a FeCo sample was prepared in the same experimental conditions as the Fe undulated system characterized above, keeping the same effective thickness of 3 nm of the FeCo magnetic layer. Magnetic reversal loops measured in different geometries and at different temperatures are presented in
Figure 4a,b. There is not a notable difference between the magnetization reversal loops collected at 10 K with the field applied in the sample plane (longitudinal, transversal, and oblique geometries, the last one meaning the field oriented at 45 degrees with respect to longitudinal and transversal direction), similar to the previous case. The coercive field is in this case about 200 Oe (so, slightly lower than in the case of the Fe film), inferring on one hand an undulated long-range magnetic structure with an effective thickness higher than the valley deep, and on the other hand, rather the formation of oblate FeCo islands/nanoparticles (e.g., with lateral sizes larger than the highness). As a consequence, the magnetic susceptibility in perpendicular geometry is much lower in the case of the FeCo oblate nanoparticles/spheroids than in the case of the quasi-spheroidal Fe ones, and therefore, the overall contribution to the magnetic susceptibility at 10 K coming from the undulated FeCo thin film with long-range magnetic order, the system of FeCo oblate nanoparticles in the magnetic frozen regime and the diamagnetic substrate becomes negligible. This is an additional reason for the very low values of magnetization over the whole range of applied magnetic fields during the magnetic reversal in perpendicular geometry. On the other hand, the evolution of the coercive field with temperature in longitudinal geometry is quite similar to the one of the 3 nm Fe film. Following the same explanation, the finest FeCo oblate nanoparticles become superparamagnetic at higher temperatures (as proven also by the ZFC-FC measurements in
Figure 4c), canceling out the long-range magnetic interaction of the non-homogeneous undulated FeCo thin film. The magnetic behavior is due to only the assembly of non-interacting FeCo nanoparticles which enter progressively the superparamagnetic state at a higher temperature.
The blocking temperature estimated from the ZFC curve measured in longitudinal geometry with an applied field of 200 Oe is about 157 K (
Figure 4c). The insignificant decrease of magnetization above the branching temperature infers also FeCo nanoparticles of large size and shape distribution, many particles remaining in the blocked state also at room temperature. However, the larger blocking temperature and coercive field in longitudinal geometry in the case of the FeCo system as compared to the Fe system, give support for an in-plane anisotropy of the largest nanoparticles, higher in the case of FeCo systems. Nevertheless, this situation becomes possible only for quasi-ellipsoidal nanoparticles (prolate spheroids) oriented along the trenches. Hence, a higher in-plane uniaxial anisotropy of the system can be kept at 300 K (easy axis along the ellipsoidal nanostructures) in the FeCo systems. Given this behavior, two FeCo systems grown on trenches of higher periodicity will be considered, with the aim of diminishing the lateral dipolar interaction between the ellipsoidal nanoparticles which are magnetically blocked at room temperature.
3.2.3. Magnetic Characterization of FeCo Nanostructured System (with a Thickness of 3 nm) Deposited on Mo Trenches Prepared at Higher Temperature Than Those Presented in Section 3.2.2
Considering a further possibility to control the magnetic properties of these nanosystems, we prepared Mo templates using a temperature gradient in the range (500–200 °C), which induces a larger periodicity than that of the first two samples [
22]. These templates were used for the deposition of FeCo films of two different effective thicknesses.
Hysteresis loops were collected at 10 K in different geometries and at different temperatures and the ZFC-FC curves of the FeCo system with the same thickness (3 nm) and deposition conditions similar to the one presented in
Section 3.2.2, are shown in
Figure 5. However, in this case, the facetted template should have a higher periodicity of the trenches than the previous system. If concerning the magnetization reversal in longitudinal and transversal geometries the situation is similar to of the previous FeCo film (with almost the same shape and a coercive field of 200 Oe in both geometries), a significant difference appears in the perpendicular geometry (
Figure 5a). This behavior might be induced by the Co composites and the magnetization reversal mechanism that is strongly dependent on the geometrical configuration of the finest FeCo nano-islands/nanoparticles which are blocked at low temperature and assure the long-range magnetic interaction of the undulated film and develops along a more rounded hysteresis loop, of high coercive field. The only explanation for this behavior, is the formation of oblate islands/spheroids grown on the wave side, in such a way that a field perpendicular to the undulated film will make a small but finite angle with the plane of the quasi-bidimensional island. According to the Stoner-Wohlfarth model, the magnetization reversal develops in this case gradually (with low magnetic susceptibility), but with a sensible coercive field. By increasing the temperature, the finest FeCo nanoparticles become superparamagnetic, cutting out the interaction among the largest prolate islands (nanoparticles assimilated to prolate spheroids with the long axis along trenches. Accordingly, the coercive field of the assembly of non-interacting prolate nanoparticles decreases with temperature, as the effect of more and more nanoparticles entering the superparamagnetic regime (
Figure 5b), similar to the case of the previous FeCo film. However, the ZFC-FC curves, in this case, suggest a much-increased blocking temperature above 300 K (
Figure 5c), meaning longer prolate magnetic nanostructures along the trenches.
3.2.4. Magnetic Characterization of a Thicker FeCo System than the One Presented in Section 3.2.3
In order to further study the influence of the film thickness on the magnetic properties of such nanostructures, a much thicker FeCo system of approximately 15 nm (5 times higher thickness than the systems presented above) was prepared on Mo linear trenches. The grooved Mo templates were deposited at a temperature gradient of (500–300 °C) similarly to the system previously presented. Hysteresis loops collected at 10 K in different geometries and at different temperatures as well as the specific ZFC-FC curves are illustrated in
Figure 6. A noticeable difference can be observed in
Figure 6a as compared to
Figure 5a. That is the higher coercive field (800 Oe as compared to 200 Oe) reflected by the two loops of similar shape collected in longitudinal and transversal geometry. That means the undulated thin film contribution to the in-plane anisotropy of the system becomes dominant over the oblate nanoislands grown on the wave side, as will be easily expected from the deposition of a thicker film. On the contrary, in the perpendicular geometry where the spins of the undulated film respond very hardly to the field, the dominant increase of magnetization is given by the magnetization reversal of the finest oblate nanoislands which are magnetically frozen at low temperature and hence the shape of the loop is similar to the previous case. At increasing temperature, the finest oblate nanoparticles become superparamagnetic and cancel out the long-range magnetic interaction between larger clusters, leading to magnetic configurations specific to undulated thin films. Hence, the coercive field of the assembly of the prolate nanoclusters decreases at increasing temperatures due to the increased number of superparamagnetic nanoparticles/nanoislands (
Figure 6b). The average blocking temperature, well above 300 K as deduced from the ZFC-FC curves presented in
Figure 6c is even higher than in the previous case, proving the formation of prolate nanoislands of higher aspect ratio (or anisotropy constant).
Due to the higher effective thickness of the inhomogeneous FeCo film, the in-plane magnetic properties of the FeCo system were also characterized at room temperature by magneto-optical measurements using a MOKE magnetometer.
Specific vectorial hysteresis loops (at different orientations of the field with respect to the trenches direction), as well as polar representations of the relative remnant magnetization (remnant to saturation magnetization ratio) and of the normalized coercive field (current coercive field reported to the maximal coercive field) obtained from the MOKE hysteresis loops collected at different azimuthal angles, are shown in
Figure 7. The ample variation of the shape of the hysteresis loops with the azimuthal angle, translated also in the large oscillations of both the relative remnant magnetization and normalized coercive field, proves clearly the high uniaxial anisotropy of the system along the trenches achieved at room temperature in such systems.