Single-step generation of 1D FeCo nanostructures

Magnetic one-dimensional structures are attractive nanomaterials due to the variety of potential applications they can provide. The fabrication of bimetallic 1D structures further expands the capabilities of such structures by tailoring the magnetic properties. Here, a single-step template-free method is presented for the fabrication of 1D FeCo alloy nanochains. In this approach, charged single-crystalline FeCo nanoparticles are first generated by the co-ablation of pure Fe and Co electrodes under a carrier gas at ambient pressures and attracted to a substrate using an electric field. When reaching the surface, the particles are self-assembled into parallel nanochains along the direction of an applied magnetic field. The approach allows for monitoring the self-assembly particle by particle as they are arranged into linear 1D chains with an average length controlled by the deposited particle concentration. Magnetometry measurements revealed that arranging nanoparticles into nanochains results in a 100% increase in the remanent magnetization, indicating significant shape anisotropy. Furthermore, by combining x-ray microscopy and micromagnetic simulations, we have studied the local magnetization configuration along the nanochains. Our findings show that variations in magnetocrystalline anisotropy along the structure play a crucial role in the formation of magnetic domains.

Most self-assembly methods utilize colloidal NPs and organize them into 1D structures with the help of attached ligand molecules, templates, or external fields [2,3].The latter is frequently employed to create magnetic NCs by applying a magnetic field during the synthesis or while drop-casting the particle suspension onto substrates [17][18][19][20][21][22][23].Although this approach is simple, cost-effective, and scalable, it typically lacks precise control over the spatial distribution of the 1D structures on the substrate due to chain aggregation during solvent evaporation [2,3].Recent progress in the precise positioning of the NPs into separated NCs along pre-defined directions onto substrates relies on top-down lithography to spatially confine the suspension [24,25], thereby significantly increasing the complexity and decreasing the cost-effectiveness and scalability of the methods.
An alternative approach is to use a magnetic field to attract and self-assemble gas-phase-generated NPs onto substrates [9,10,26,27].These techniques are typically limited to out-of-plane structures as the magnetic field both attracts the NPs and guides the assembly.Furthermore, the method results in large bundles and nanostructures with limited control of the self-assembly and final geometry.Electrically charged gas-phasegenerated NPs offer additional control as an electric field can be used to attract them to the substrate.Recently, new techniques have emerged combining charged aerosol NPs and top-down processing of dielectric materials to enable electric-field assisted aerosol lithography [28] and 3D printing of micro and nanostructures [29,30].
Previous studies of self-assembled NCs have mainly been limited to elemental ferromagnetic systems and Fe-oxides.To increase the scope of future applications, it is of interest to expand the possible building blocks to nanoalloys as they provide a wider range of materials, often with enhanced properties compared to their monometallic counterparts [31][32][33].Herein, we present a bottom-up approach for generating metallic FeCo alloy NPs and self-assembling them into parallel NCs with controlled average length directly along a substrate surface.The approach combines the simplicity of magnetic-field-directed self-assembly with the added control of using electric fields to attract charged aerosol NPs to the surface [14,34].The method employs an aerosol technique based on spark ablation where NPs are generated by a discharge between two electrodes and transported away with a carrier gas at ambient pressure [35].The approach is capable of producing not only elemental NPs but also bi-magnetic systems [36,37], and is ideal for generating alloy systems, as the transitionmetal ratio of the resulting bimetallic NPs can be precisely controlled by the stoichiometry of the alloy electrodes [38,39].In the presented work, we demonstrate that single-crystalline FeCo alloy NPs with a Fe:Co ratio close to 50:50 can be generated by the co-ablation of single-element Fe and Co electrodes.The resulting NPs are attracted to a substrate using an electric field, where they are self-assembled along the direction of an applied magnetic field.Compared to other magnetic-field-directed techniques, this has the distinct advantage that the magnetic field can be applied independently of the electric field capturing the NPs, thereby allowing the self-assembly to be controlled by varying the direction of the applied magnetic field to generate NCs oriented vertically or along the substrate surface [34].Furthermore, the approach opens up for controlling and studying the self-assembly from the first particle, exemplified here by following the growth of individual NCs, particle by particle, and demonstrating how the average chain length can be controlled by adjusting the deposited particle concentration.Moreover, the technique provides a facile fabrication of free-standing 1D structures that allow for direct integration into devices, as recently demonstrated for Co NCs [14].Finally, by combining magnetometry, x-ray microscopy, and micromagnetic simulations, we demonstrate that arranging the NPs into NCs leads to a significant shape anisotropy and that local variations in the magnetocrystalline anisotropy along the chains play a key role in the domain formation.

Experimental and simulations
The generation of alloy FeCo NPs starts from the co-evaporation of materials by repetitive sparks between two high-purity metallic Fe and Co rods (GoodFellow, 5.0 mm diameter) under the flow of carrier gas [35,39].As the vapor cools down, condensation of sub-10nm primary particles takes place, which then form larger irregular agglomerates by collision.To avoid oxidation of particles, a carrier gas composed of 95% N 2 and 5% H 2 (1.68 lpm) is used, and the pressure of the gas is kept constant at around 1015 mbar [14,38,40].The agglomerates are then passed to a Ni 63 bipolar diffusion charger to give an even charge distribution to the materials.After that, the agglomerates are passed through a tube furnace where they are compacted into faceted NPs at 1473 K and then size selected (D = 40 nm) using a differential mobility analyzer based on their electrical mobility.The deposition of particles is performed in an electrostatic precipitator [41], and for the deposition under a magnetic field, substrates are placed on a permanent magnet with an in-plane field of 300 mT.To be able to deposit the desired amount of particles on substrates, the particle concentration in the gas is monitored online with a TSI Electrometer Model 3068B, and the deposition time for desired coverage is calculated based on this reference [42].X-ray diffraction patterns of the NPs are obtained using Stoe Stadi MP, Mythen 1k detector, Cu-K alpha radiation, λ = 1.54178Å. Electron microscopy is performed using a Jeol JEM-3000F transmission electron microscope equipped with an Energy-dispersive x-ray spectroscopy (XEDS) detector in transmission (TEM) and scanning (STEM) modes, and also Hitachi-SU8010 Cold Field Emission scanning electron microscope (SEM).To investigate the magnetic behavior, the structures are produced on single crystalline SiO 2 stripes, and the measurements are done using a superconducting quantum interference device vibrating sample magnetometer (SQUID-VSM, MPMS 3, Quantum Design).The scanning transmission x-ray microscopy (STXM) measurements were performed at the SoftiMAX beamline [43,44] of the MAX IV laboratory.

Nanoparticle generation and directed self-assembly
Charged NPs suspended in a carrier gas are generated using an aerosol technique based on spark ablation, and the setup is schematically shown in figure 1.The NPs are produced from material evaporated in the repetitive ablation between two electrodes through high-power electric discharge pulses under a flow of inert carrier gas at atmospheric pressures [35,47].In this work, single-element Fe and Co electrodes are employed as anode and cathode, respectively, and as will be shown, equiatomic FeCo alloy NPs are produced through the ablation process, demonstrating the feasibility and flexibility of the method for generating bi-metallic building blocks directly from single-element materials.The as-produced irregular agglomerate particles are given a known charge before being compacted into single-crystalline faceted NPs as they are transported through a furnace.The resulting NPs are then size-selected based on their electric mobility and attracted to the substrate using an electric field.The substrate is placed on a NdFeB magnet, and as the NPs arrive at the surface, they self-assemble via magnetic dipole-dipole interactions to form 1D NCs along the direction of the applied field [14,34], as illustrated in figure 1(a).The NPs in this work are not monodispersed.However, the size distribution of the NPs can be significantly decreased by sorting the agglomerates based on their electric mobility before compaction in the tube furnace [39,48].
The self-assembly of FeCo NPs under an in-plane magnetic field is investigated by consecutive depositions followed by identical-location SEM imaging at each stage.This approach allows for observing the same site after each deposition, thus providing insight into NC growth as more particles are added to the substrate.Figures 1(b)-(e) show representative SEM images for 10, 15, 20, and 25 NPs/μm 2 particle concentrations, together with zoomed-in images of three selected areas, figures 1(f)-(h).The results indicate three main stages in the formation of NCs: (1) nucleation, (2) growth, and (3) bundle formation.At low coverages, single NPs are randomly distributed along the surface.With an increasing number of particles reaching the substrate, magnetic dipole-dipole interactions between incoming and deposited particles result in the formation of dimers and trimers.These structures exhibit an increased magnetization and magnetic field and act as nucleation centers, initiating the formation of NCs by attracting additional particles arriving in the vicinity, see figures 1(f), (g).The effect self-amplifies as the structures grow, resulting in NPs being attracted from larger distances and the formation of increasingly longer parallel chains with few single particles and smaller structures situated between them.At higher coverages, arriving NPs are attracted on top of the NCs, indicated by arrows in figures 1(h), leading to the formation of bundles composed of intertwined chains.Although the growth rate of individual NCs varies, the average chain length has a linear dependence on the deposited particle concentration.Figure 2 displays the average length of chains plotted per particle concentration.The average length is obtained by analyzing a large area of the sample (≈550 μm 2 ) by measuring the areas of individual chains and dividing them by the average 40 nm diameter.The results show a linear increase from 404 to 822 nm, respectively, for 10 to 25 NPs/μm 2 particle concentrations, demonstrating that average length can be controlled by the number of deposited NPs.
This technique not only enables self-assembly on solid surfaces but also facilitates the fabrication of freestanding NCs that can extend up to several microns, as shown in figure 3(b).Here, the self-assembly tends to start at the edges, likely due to local enhancements of the electric field that attract the first particles and initiate the selfassembly as additional particles arrive in the vicinity.As previously demonstrated for Co NCs, this approach provides a facile integration into devices by directly self-assembling chains across the source and drain [14].

Structural characterization
The crystal structure of the produced NPs is investigated by XRD measurements on high-coverage samples prepared without size selection to maximize the particle yield and, thus, the diffracted intensity.Figure 3(a) displays the resulting diffractogram with intensities at 2θ angles corresponding to an equiatomic FeCo bcc structure with an average lattice parameter of 2.851(0.001)Å [49,50].TEM microscopy is used to further characterize the crystallinity and composition of NPs.For this, chains of particles are generated on a TEM grid with lacey carbon using the magnetic field-assisted self-assembly technique, figure 3   spacing of 2.022 and 1.415 Å corresponding to FeCo {110} and {200} planes, respectively.This leads to an average lattice parameter of 2.845(0.022)Å consistent with the value obtained from the XRD diffractogram.Fast Fourier transform (FFT) of the particles displayed in figure 3(c) demonstrates well-defined spots corresponding to {220} or {200} planes for the two particles suggesting a single-crystalline structure.Moreover, the images suggest that the particles are in the form of truncated octahedrons assembled with random crystal orientations with regard to each other, similar to what was previously observed for Co NCs [14].XEDS-mapping of NCs presented in figure 3

Magnetic properties 3.3.1. Magnetization curves
Figure 4 displays SEM images of two samples denoted S1 and S2, the first prepared by attracting the NPs using only the electric field (S1) and the second by applying an additional in-plane magnetic field (S2).The particle coverage is chosen to obtain a high density of 1D structures to maximize the signal for the magnetometry measurements and is kept the same for both samples.The average particle size is below the single-domain critical diameter for FeCo (D c ≈ 51 nm), and the particles can spontaneously magnetize in the carrier gas [51].Since the particle magnetization has no preferential direction, the NPs in S1 self-assemble into randomly oriented interconnected chains and bundles, figure 4(a), in contrast to the parallel NCs in S2 obtained by applying an in-plane magnetic field (figure 4(b)).Magnetic hysteresis loops of the samples are presented in figure 4(c) with the magnetic field applied parallel to the long axis of the NCs in S2.The magnetic parameters extracted from the loops are summarized in table 1, revealing significant differences between the two systems.As shown in figure 4(c), S2 exhibits a smaller saturating field and larger remanent magnetization than S1 (≈100% increase).The observed differences can be explained by the high aspect ratio of the NCs in S2, leading to a significant shape anisotropy along the chain axis.For S2, the parallel NCs have a shape anisotropy collinear with the applied magnetic field.In contrast, S1 is composed of 3D networks of randomly oriented and interconnected chains.At sufficient high external fields, the magnetization saturates as the individual particle moments in S1 align along the applied field direction.However, as the field is lowered, the shape anisotropy starts dominating and rearranging the particle magnetization along the randomly distributed chains, resulting in a reduced M r /M s .Interestingly, the coercivity of S2 has decreased by ≈17%.One possible explanation for this could be the extent of demagnetizing interactions in the samples.Note that the magnetic response of the structures to an externally applied field is highly influenced by the magnetic interactions between the constituent elements.In S2, the structures are arranged in a more compact configuration on the surface, thereby increasing the interactions between the structures.Consequently, when a NC switches its magnetization, it exerts a demagnetizing field on its neighboring structures, leading to a reduction in the total coercive field of the sample [52].Another potential explanation could involve the presence of fanning magnetization reversal mode, which can be present in particle  chains.In such structures, magnetization reversal can occur in coherent or fanning mode.In the former, the magnetization reversal of particles happens collectively, whereas, in the latter, the magnetization of individual particles rotates in alternate directions during the reversal process to minimize the total magnetic energy of the system by decreasing the stray field.In this case, as predicted by theory and observed in previous studies [53][54][55], randomly oriented structures show a slightly higher coercivity than that of parallel structures.

Scanning transmission x-ray microscopy
The local magnetization configuration along individual NCs is investigated by combining STXM and x-ray magnetic circular dichroism (XMCD).The sample is first pre-characterized using SEM, and then STXM is performed at selected representing areas.STXM images are obtained by raster scanning the sample across a 25 nm x-ray beam while detecting the transmitted intensity.Figure 5(a) shows the x-ray absorption spectrum (XAS) obtained from the NCs, measured by recording images at different photon energies across the Co L 3 -edge using right (I + ) and left (I − ) circularly polarized x-rays.The difference in absorption between the two helicities provides an XMCD spectrum, which is proportional to the projected 3d magnetization onto the direction of the x-ray beam.Mounting the sample at a 30-degree angle to the x-ray beam and tuning the photon energy to the maximum Co-L 3 XMCD signal thus provides spatially resolved XMCD images with a dark and bright contrast corresponding to the magnetization direction along the structures.Figure 5(b) displays an SEM image of representing FeCo NCs, self-assembled directly along a TEM grid, and figure 5(c) shows the x-ray absorption image of the corresponding region.The XMCD contrast corresponding to the remanent magnetic state of the NCs after applying a magnetizing field of 2 T along and perpendicular to the chain axis are displayed in figures 5(d) and (e), respectively.After magnetizing parallel to the long axis, the NCs in figure 5(d) show a predominantly bright remanent contrast, demonstrating a uniform longitudinal remanent magnetization along the chains.This can be understood by the high shape anisotropy of the NC, keeping the magnetization along the chain axis.In contrast, dark and bright XMCD contrast is observed after applying the field perpendicular to the chain axis, as shown in figure 5(e), demonstrating the formation of domains with the magnetization in different directions along the chain axis.These findings are similar to the previous study of Co NCs, where it was demonstrated that the emergence of magnetic domains is due to variations in the orientation of the magnetocrystalline anisotropy (MCA) of the NPs along the chains [14].
Micromagnetic simulations are used to further study the effect of cubic MCA on the remanent states of FeCo NCs.A NC with a length of ≈1μm made up of NPs with diameters randomly selected in the 30-50 nm range is considered to represent the size distribution observed along the NCs.The MCA directions are randomly introduced to each NP along the chain to account for the expected variations of the interparticle crystallographic orientation observed from the TEM images in figure 3(d).Similar to the XMCD contrast in figure 5(d), the simulated state exhibits a uniform magnetization with the magnetic moments predominantly oriented along the chain axis.The effect of local variations in the crystallographic directions along the chains is studied by considering three NCs with identical geometry but different randomly chosen orientations of the MCA for each particle.The relaxed equilibrium magnetic configurations of the three NCs after applying a transverse magnetic field are shown in figure 5(g).All three chains exhibit magnetic domains, consistent with the experimental results in figure 5(e).Furthermore, despite sharing the same geometry, the three random sequences of cubic MCA result in three distinct distributions of the magnetic domains along the chains.In contrast, using the same NC geometry but now aligning the MCA of each particle with the long axis of the chain and the shape anisotropy results in a uniform magnetization without any domains, as seen in figure 5(h).It should be noted that the simulated structures are not meant to mimic the imaged NCs in figure 5 but to demonstrate that the emergence and distribution of magnetic domains after applying a transverse magnetic field are highly influenced by local variation in the MCA along the chains.

Conclusions
This study presents a single-step template-free approach for the fabrication of magnetic nano-alloys directly from single-element bulk materials and self-assembling them into 1D structures with a controllable average length.Magnetic measurements reveal a significant shape anisotropy when self-assembling the particles into elongated 1D structures.Furthermore, domain formation in individual NCs is investigated by STXM-XMCD imaging and micromagnetic simulations.The results show that the formation and distribution of the domains are highly dependent on the orientation of the MCA of the individual particles along the chains.

Figure 1 .
Figure 1.(a) Schematic of FeCo NP formation and deposition from single-element electrodes including (1) formation of irregular agglomerates from Fe and Co atomic cloud, (2) acquiring a known charge distribution, (3) reshaping the materials into singlecrystalline particles by heating at 1473 K, (4) size selection based on the electrical mobility, (5) online particle detection and analysis, and (6) deposition under electric and magnetic field.(b)-(e) Identical-location SEM images corresponding to deposited particle coverages of 10, 15, 20, and 25 NPs/μm 2 .(f), (h) Magnified images corresponding to regions I-III in (b)-(e) after each deposition step.(h) Linking of two single chains and forming a bundle when added particles settle on top of the chains, indicated by arrows (scale bars = 200 nm).
(b).HRTEM image of two NPs of a freestanding NC is shown in figure3(c).The image displays faceted NPs with an average interplanar

Figure 2 .
Figure 2. NC length per deposited particle concentration obtained by SEM image analysis.

Figure 3 .
Figure 3. (a) XRD diffractogram of NPs collected without magnetic field showing single-phase FeCo bcc structure.(b) SEM image of NCs self-assembled on a TEM grid with lacey carbon film and (c) High-resolution TEM image of the NPs with the FFT shown as inserts.(d) STEM electron image and XEDS maps of a free-standing FeCo NC demonstrating a homogeneous mixture of Fe and Co with Fe:Co atomic ratio of ≈50: 50.
(d) suggests the homogeneous distribution of the alloy elements as the Fe and Co signals match nicely throughout the structure.The results show a 49.3:50.7 Fe:Co atomic ratio in agreement with the XRD lattice parameter.

Figure 4 .
Figure 4. SEM images of (a) S1 composed of randomly oriented interconnected chains, and (b) S2 made up of parallel NCs.(c) Magnetization curves from S1 and S2 recorded at 300 K with the field applied along the chain axis in (b).

Figure 5 .
Figure 5. (a) XAS and XMCD spectra recorded from the NCs at the Co L 3 -edge using right (I + ) and left (I − ) circular polarized x-rays.(b) SEM image of NCs self-assembled directly along the surface of a TEM grid, and (c) the corresponding STXM image recorded from the total absorption I + + I − at the Co L 3 absorption edge.(d) Remanent XMCD contrast (I + − I − ) obtained after ex-situ magnetizing along the chain axis.(e) STXM-XMCD image after ex-situ magnetizing perpendicular to the chain axis.(f)-(g) Simulated remanent magnetic configurations along a NC made up of NPs of various sizes and random interparticle MCA orientation after magnetizing (f) along the chain axis and (g) perpendicular to three NCs made up of identical NPs but with different sets of random MCA orientations.(h)Simulated remanent magnetic configurations after applying a magnetizing field perpendicular to a NC with the same sequence of NPs as in (f)-(g), but now aligning the MCA of each particle with the long axis of the chain.

Table 1 .
Summarized values of the magnetization curves at 300K.The saturating field, H s (mT), is defined as a 3% deviation from saturation magnetization.