Impact of nanoparticle magnetization on the 3D formation of dual-phase Ni/NiO nanoparticle-based nanotrusses

Magnetic nanoparticles with average size 30 nm were utilized to build three-dimensional framework structures—nanotrusses. In dual-phase Ni/NiO nanoparticles, there is a strong correlation between the amount of magnetic Ni and the final size and shape of the nanotruss. As it decreases, the length of the individual nanowires within the trusses also decreases, caused by a higher degree of branching of the wires. The position and orientation of the non-magnetic material within the truss structure was also investigated for the different phase compositions. For lower concentrations of NiO phase, the electrically conducting Ni-wire framework is maintained through the preferential bonding between the Ni crystals. For larger concentrations of NiO phase, the Ni-wire framework is interrupted by the NiO. The ability to use nanoparticles that are only partly oxidized in the growth of nanotruss structures is of great importance. It opens the possibility for using not only magnetic metals such as pure Ni, Fe, and Co, but also to use dual-phase nanoparticles that can strongly increase the efficiency of e.g. catalytic electrodes and fuel cells.


Introduction
The ability to design structures with a huge surface area to volume ratio is vital for many energy-related applications of nanomaterials (Zhou et al. 2011), such as different electrodes for capacitors, batteries, fuel cells, and electrochemical catalysis (Ning et al. 2016;Royston et al. 2008;Li et al. 2012;Debe 2012). Assemblies of nanoparticles can have such interesting properties. However, they are mostly effective on 2D surfaces as the surface area exposed to reactants is limited in thicker layers and among agglomerated particles. One way to circumvent this is to use a 3D support matrix (Somorjai and Park 2008) or assembling the nanoparticles into more open structures. The latter can be achieved, by using magnetic 1 nanoparticles combined with an applied magnetic field which allows the nanoparticles to self-assemble into nanowires or frameworks of nanowires-so-called nanotrusses. This can be achieved using magnetic particles in a liquid suspension (Kralj and Makovec 2015) or by direct generation and assembly of nanoparticles using plasma-based methods (Ekeroth et al. 2018). The latter has the advantage of avoiding impurities, such as surface oxidization. In a previous work, we have shown that Fe nanoparticles can be generated and assembled, onto a variety of substrates, into long (> 100 μm) nanotrusses through a plasma where the particles are generated followed by magnetic collection (Ekeroth et al. 2018). The low packing density of these trusses makes the material suitable for electrocatalysts since reactants can reach a high fraction of the exposed surface. Other benefits include no extra process steps associated with support structure, the entire structure is composed of active material, and there are electrical pathways to the substrate.
In the present work, nanowires and nanotrusses are formed through direct self-assembly of dual-phase magnetic Ni/NiO nanoparticles. The magnetic property of the nanoparticles is needed for the self-assembly of the nanotrusses in the magnetic field applied. We here investigate the amount of magnetic material needed in the individual nanoparticles for the assembly to occur. This knowledge is essential when designing nanostructures containing materials that are non-magnetic for different applications, e.g., catalytic electrodes. Some interesting materials include CuNi (Kimi et al. 2018), FePt (Adamski et al. 2018), and NiPt (Pham et al. 2017). The growth of individual dual-phase nanoparticles of Ni and NiO has been demonstrated previously, where various oxygen flow rates, 0 to 0.08 sccm, produced particles from pure Ni to (almost) pure NiO particles (Ekeroth et al. 2019). Ni nanoparticles have ferromagnetic properties (at size > 20 nm) (Ishizaki et al. 2016) and NiO (with a size > 4 nm) is antiferromagnetic (Rinaldi-Montes et al. 2014). This allows us to investigate magnetic collection into nanowires and nanotrusses from particles with varying amount of magnetic material. When the NiO fraction in the nanoparticles increases, their ability to form individual nanowires decreases while, at the same time, the occurrence of intrawire branching and inter-wire interlinking increases.

Experimental setup
Nanoparticles are created by generating a high-density flux of Ni ions from a hollow cathode sputtering source using high-power pulses. The Ni cathode (99.98% purity; inner diameter 5 mm and length 54 mm) is placed in a high vacuum chamber with a base pressure of 3 × 10 −7 Torr (4 × 10 −5 Pa) where nanoparticles nucleate and grow. Through the hollow cathode, an Ar sputter gas (99.9997% purity) is flown into the vacuum chamber at a rate of 30 sccm. Diluted O 2 (99.5% Ar) is added to the process through a separate gas inlet in the vacuum chamber wall, located approximately 30 cm downstream of the cathode. The O 2 gas flow is the only parameter that is changed for the different growth conditions. Four growth conditions are used, with effective O 2 flows of 0, 0.02, 0.04, and 0.06 sccm. In Table 1, each condition is given a roman numeral for easier description throughout the paper. The experimental setup has been described earlier (Ekeroth et al. 2019), with the only difference being the position of the substrates. In the present work, the substrate is placed on a permanent magnet with a magnetic flux density of 0.36 T and dimensions of 9 mm in diameter and a height of 7 mm. The magnet is placed inside the mesh at the same height as the anode, see Fig. 1. This allows the substrate to be close to the nanoparticle generation zone, but away from the most intense plasma region. In control experiments, nanoparticles are also collected at the same position, but without a magnet. Substrates are Ti-coated (200 nm) Si wafers.
Scanning electron microscopy (SEM) images are taken using a LEO 1550 instrument equipped with electron high tension (EHT) and in-lens detectors. To investigate the structure of the nanotrusses as well as the particle-particle bonding, suitable samples for transmission electron microscopy (TEM) analysis are prepared. First by placing the deposited sample in isopropyl alcohol in an ultrasonic bath for 5 min to create a suspension of the nanotrusses. Then a small drop (approx. 50 μl) of the suspension was dried onto a TEM grid covered with an amorphous carbon film. The TEM analysis is obtained with a FEI Tecnai G2 operated at 200 kV using high angle angular dark field (HAADF) and energy dispersive X-ray spectroscopy (EDS) combined with scanning TEM (STEM) analysis. The HAADF images are collected with the annular detector spanning an angular range from 80 to 260 mrad.
Grazing incidence X-ray diffraction (GIXRD) measurements are performed using an Empyrean diffractometer in a parallel beam configuration with a line- focused copper anode source (Cu Kα, l = 0.154 nm), operating at 45 kV and 40 mA. The primary beam is conditioned using a parallel beam mirror and a 1/4°d ivergence slit and in the secondary beam path a 0.27°p arallel plate collimator is used. A PIXcel-3D detector is used as an open detector for the data acquisition. The grazing incidence X-ray diffraction scans are performed at an incidence angle of 1°in a 2θ range of 30-80°using a step size of 0.015°and a data collection time of 0.88 s/ step.
Calculations of the magnetic orientation within nanoparticles is done with the "Magnetic Fields, No Currents" module within COMSOL Multiphysics version 5.4 (COMSOL Multiphysics 2019). A cylindrical calculation domain with a diameter of 300 nm and a length of 323 nm is used with the nanoparticle in the center, aligned with the cylinder axis. Magnetic energy is calculated as W m ¼ ∫ τ 1 2 H : Bdτ where the integration is performed over the entire calculation domain τ.

Results and discussion
A series of nanoparticles were grown, and collected on substrates, using conditions with different levels of oxygen added, Table 1. Their phase composition is estimated using semi-quantitative reference intensity ratios of the two phases observed using the gracing incident XRD measurements, Fig. 2.
The O 2 gas flow is the only process parameter altered. The XRD acquired phase composition is calculated from the peaks shown in Fig. 2.
An overview of the structures obtained is presented in Fig. 3, displaying top-view SEM images of samples grown using the four different growth conditions investigated. It shows that nanotrusses are formed from particles grown at the lower oxygen flow rates (conditions i-iii), but not for the highest flow rate investigated (condition iv).
In order to elucidate the 3D structure, SEM and TEM investigations of samples grown using conditions i through iii are shown in Fig. 4. A matrix system is used to describe the specific image, Fig. 4 (x, y), where x determines the row and y the column. Each column represents a specific growth condition (in accordance with Table 1) and the rows display increasing magnification of SEM and TEM images of the three different samples.
Comparing the present results for pure Ni (condition i) with previously published results using Fe nanoparticles (Ekeroth et al. 2018), similarities are observed with trusses highly oriented along the growth direction, i.e.,  Table 1 the magnetic field orientation of the external magnet. The nanotrusses, with lengths of up to 100 μm, are made up of nanowires that interlink. Differences between the Fe and Ni are seen in the size of the individual nanoparticles within the trusses. Fe nanoparticles with sizes of less than 10 nm have been seen to participate in nanotruss formation (Ekeroth et al. 2018). However, no individual Ni nanoparticles smaller than 20 nm are found within the trusses. Fe has a significantly higher saturation magnetization as compared to Ni (217.6 and 55.1 Am 2 /kg, respectively) (Crangle and Goodman 1971). This means that Ni particles need to grow bigger as compared to Fe particles for the attractive force of the magnet at the substrate to overcome the diffusive motion of the nanoparticle in the plasma. Please, see Ekeroth et al. (2018) for details on magnetic capturing of nanoparticles.
As oxygen gas is added to the process, Ni/NiOcomposite nanoparticles are expected to be formed (Ekeroth et al. 2019), which should affect how the nanoparticles are collected and nanostructures formed. For the lowest oxygen gas flow investigated, condition ii, the SEM images ( Fig. 4a (ii) and b (ii)) reveal only a small difference compared to growth without oxygen (Fig. 4a (i) and b (i)). The individual nanowires are no longer easily distinguishable and the bundles seems to be denser, but the overall appearance is still fibrous. From the TEM image ( Fig. 4c (ii)), we can see a higher degree of interlinking between the nanowires compared to the pure Ni nanoparticles (Fig. 4c (i)). This can be one explanation to why the individual wires are not clearly seen in Fig. 4b (ii). At higher magnification, a clear trend is observed within the nanoparticles themselves. With no oxygen, the particles are monodispersed with an average diameter of approximately 35 nm and show a hexagonal projection in the TEM image ( Fig. 4d (i)). Adding the oxygen flow, the particles themselves become increasingly polydisperse both in terms of their size and shape (Fig. 4d (ii)). This is due to the formation of mixed-phase nanoparticles consisting of both metallic/magnetic Ni and insulating/non-magnetic NiO. The different phases are easily distinguished in the STEM-HAADF images of Fig. 5a, b, d where the higher average atomic mass of Ni gives it a higher contrast compared to NiO. EDS mapping confirms this, and in Fig. 5c NiO (blue) is clearly visible between larger and smaller Ni (green) particles. The NiO domains are typically 20 nm by 10 nm in size, as seen in the 2D projections of the TEM. They are cones-shaped crystals formed on a larger (30-40 nm) Ni particle. These domains are then capped by a smaller (8-10 nm) Ni crystal. The generation of these multiphase nanoparticles is more carefully explored and discussed in our previous work (Ekeroth et al. 2019). The nanowires formed by these multiphase nanoparticles produced under growth condition ii mainly constitute a main wire-structure of the larger Ni particles connected to each other, as seen in Fig. 5a. The NiO cone (with the Ni-cap) points out from the wire. This should increase the probability of interlinks forming between wires, as is seen in Fig. 5d. Here the Ni-cap of a multiphase particle seemingly connected to the larger Ni particle of another multiphase. These extra branching sites give rise to the formation of the distinctive nanotruss structures seen in Fig. 4b (ii). How the dual-phase nanoparticles assemble into wires, with the caped NiO to the and b depict the magnetic flux density in a cross section along the symmetry line for two different magnetizations of the particle. The two magnetizations were along and perpendicular to the symmetry axis (dashed line), with opposing magnetic orientation in the two Ni domains for the perpendicular case, see Fig. 6. These two configurations were chosen since other configurations are likely to have higher magnetic energy due to lower symmetry. A magnetization of 5.1 × 10 5 A/m was used in Ni while NiO was treated as air magnetically in the calculations. The calculations show that the configuration in Fig. 6a is favorable as the magnetic energy, W m , is lower than for the configuration in Fig. 6b, 4.79 versus 5.03 eV. The results from the calculations explain the observed formation, with the larger Ni particle taking part in the main structure and the NiO cone with the Ni-cap sticking out. Now focus is turned to the growth condition with more oxygen, condition iii. Here, the general direction of the main structures (micrometer scale) is along the external magnetic field direction, as is the case for growth conditions i and ii. This can be seen in Fig. 3. However, the general appearance of these structures grown under condition iii is fluffier than the corresponding structures of grown under condition i or ii, which display more fibrous structures. See Fig. 4 (row a) for this comparison. The reason for this appearance can be found when comparing the structures at a higher resolution, Fig. 4 (row b). The ends of the nanotrusses at growth condition iii have a high tendency to deviate from direction of the external magnetic field (Fig. 4b (iii)). This results in trusses and wires of extending up to a few hundred nanometers in all directions from the main structure. For growth condition ii, there is a similar tendency. However, here the trusses and wires extend only short distances (less than 100 nm, Fig. 4b (ii)). With no added oxygen, condition i, such features are rare (Fig. 4b (i)). The TEM of structures grown under condition iii (Fig. 4c (iii)) shows a high degree of branching of the nanowires,creating relatively large voids between the wires. As can be seen in the TEM micrograph with higher resolution (Fig. 4d (iii)), the typical shape of the nanoparticles has, for growth condition iii, shifted from spheres to cube-like structures, with small (1-2 nm) features covering the surfaces. From the STEM-HAADF images (Fig. 5e, f, h), it can be seen that the NiO (darker contrast) is both cubic and incorporated in the main wire-structure. EDSstudies (Fig.5g) confirmthatthedifferentphasesare Ni (green) and NiO (blue). A previous study has shown that these structures appear when NiO is dominating with cubically shaped nanoparticles, having multiple small Ni satellites attached to it (Ekeroth et al. 2019). This change in the particlecompositionisaprobableexplanationforthechange observed in the microstructure, with highly branched structures as seen in Fig. 4a-d (iii). However, even if a large fraction of the nanoparticles is of the non-magnetic NiO phase,thefactthattrussesarestillformedisanindicationthat the amount of metallic Ni is high enough to magnetically collect the particles.
For even higher flow rates of oxygen (0.06 sccm, condition iv), it is no longer possible to form truss structures. There are still some nanoparticles collected at the substrate; however, they do not assemble into the characteristic nanotruss-structure. To validate that it is the magnetic properties of the nanoparticles that determines the formation of nanotrusses, control experiments, where the permanent magnet was removed, were also performed and compared to samples collected with a magnetic field present. The results are presented in Fig. 7 with growth condition i in the top row and condition iv in the bottom row. The influence of using a magnet is very clear for condition i but no effect is observed for condition iv. For the latter condition, particle distribution is similar to earlier experiments done with electrically collected metallic nanoparticles (Gunnarsson et al. 2015;Gunnarsson et al. 2016;Pilch et al. 2013a, p. 102;Pilch et al. 2013b, p. 103).
Higher magnification images of samples grown at condition iv, with or without a magnet present (see Fig. 8 for an example), reveal that the individual particles have changed both in shape and size, compared to the particles in growth conditions i-iii with or without a magnet. Here the particles consist of large assemblies with many crystals of cubical shape, similar to what was seen for higher oxygen flow rates in earlier work (Ekeroth et al. 2019). These assemblies appear to be formed in the gas phase, without the aid from a magnetic field, before arriving at the substrate surface.

Conclusion
Thetechniqueofusingmagneticfieldsfordirectassembleof vapor phase formed ferromagnetic nanoparticles into nanotrusses can be applied also to complex multiphase nanoparticles. Through different levels of oxidation, Ni/ NiOdual-phasenanoparticlesaregeneratedwhichassemble into nanostructures (nanotrusses) unless the ferromagnetic Ni-fraction becomes too low. The NiO content of the nanoparticles has a significant effect on the overall nanostructures. With an increasing fraction of NiO, more nanowire branching occurs within the nanotruss. At the highest NiO content studied, only very small domains of Ni remain, and the nanoparticles do not assemble into extensive nanotrusses. The possibility to include non-magnetic phases in the otherwise ferromagnetic nanoparticles opens possibilities to generate nanostructures with improved or altered properties that might find applications in different fields, e.g., as active electrodes in electrocatalysis.
Funding information Open access funding provided by Linköping University. This work was made possible by financial support from the Knut and Alice Wallenberg foundation (KAW 2014.0276) and support from the Swedish Government Strategic Research Area in Materials Science on Functional Materials at Linköping University (Faculty Grant SFO Mat LiU No 2009 00971). SI and TS like to acknowledge support from Tokyo Metropolitan University in the framework of Co-tutorial Program with Linköping University.