Nanoscale structural evaluation of 0-3 magnetic nanocomposites fabricated by electro-infiltration

Magnetic nanocomposites with 0-3 connectivity, whereby a 0D magnetic nanoparticle phase is embedded into a 3D magnetic metal matrix phase, have gained increased interest for use in applications ranging from integrated power inductor cores to exchange-spring magnets. The electro-infiltration process, in which a metal phase is electroplated through a nanoparticle film phase, is an inexpensive approach compatible with semiconductor fabrication methods for the formation of these nanocomposites. Past demonstrations of electro-infiltrated nanocomposites have relied on scanning electron microscopy and energy dispersive x-ray spectroscopy to evaluate the 0-3 composite structure. However, a detailed investigation of the boundary between the particle and metal matrix phases cannot be performed with these tools, and it is unknown whether the particle/matrix interfaces are dense and void-free. This detail is critical, as the presence of even nanoscale voids would affect any potential magnetic exchange coupling and hence the overall electromagnetic properties of the material. This work seeks to explore the phase boundary of 0-3 magnetic nanocomposite fabricated by electro-infiltration by using scanning transmission electron microscopy and energy-dispersive x-ray spectroscopy to analyze the nanostructure of two different composites—a nickel/iron-oxide composite and a permalloy/iron-oxide composite. High-resolution imaging indicates that the interface between the particle phase and matrix phase is dense and void-free. These results will help guide future studies on the design and implementation of these magnetic nanocomposites for end applications.Magnetic nanocomposites with 0-3 connectivity, whereby a 0D magnetic nanoparticle phase is embedded into a 3D magnetic metal matrix phase, have gained increased interest for use in applications ranging from integrated power inductor cores to exchange-spring magnets. The electro-infiltration process, in which a metal phase is electroplated through a nanoparticle film phase, is an inexpensive approach compatible with semiconductor fabrication methods for the formation of these nanocomposites. Past demonstrations of electro-infiltrated nanocomposites have relied on scanning electron microscopy and energy dispersive x-ray spectroscopy to evaluate the 0-3 composite structure. However, a detailed investigation of the boundary between the particle and metal matrix phases cannot be performed with these tools, and it is unknown whether the particle/matrix interfaces are dense and void-free. This detail is critical, as the presence of even nanoscale voids would affect any potential magnetic exchange coupling and he...


I. INTRODUCTION
The fabrication and application of magnetic nanocomposites, defined by having one or more of their phases possess a characteristic length scale of 1-100 nm, has gained interest due to the desire for new and better hard and soft magnetic materials. These new materials could decrease reliance on rare-earth magnets and improve the performance and efficiency of devices in areas such as power electronics (Sullivan, 2009;Balamurugan et al., 2012;Coey, 2012;Lewis and Jimenez-Villacorta, 2013;Sullivan et al., 2013;and Silveyra et al., 2018). Figure 1a shows the structure of a 0-3 magnetic nanocomposite, which is made from magnetic nanoparticles surrounded by a magnetic metal matrix. These composites have been made using different methods, such as cluster deposition, which requires the use of expensive and/or modified equipment (Balasubramanian et al., 2011 andLiu et al., 2011), and magnetic composite electroplating, which results in low fill ratios on the order of <5% (Guan and Nelson, 2005;2006). Over the last decade the electro-infiltration method, developed by Hayashi et al. and Wen et al., has been suggested as a relatively cheap and semiconductor fabrication compatible method for making 0-3 magnetic particle/matrix composites (Hayashi et al., 2012;Wen et al., 2014;2016;2017;and Smith et al., 2019).
The magnetic and material properties of 0-3 electro-infiltrated composites have been previously documented for various applications (Wen et al., 2016;2017;and Smith et al., 2019). These works relied on scanning electron microscopy (SEM) and energy dispersive x-ray spectroscopy (EDS) to demonstrate successful electro-infiltration, but these methods do not afford the spatial resolution to analyze the boundary between the particle and matrix phases. The phase boundary of these composites is of great importance, because the presence of voids between the particles and the metal matrix can significantly affect the properties of the material as a whole. Specifically, voids between the particle and matrix phases is of known importance for applications such as exchange-spring magnets, where voids would negatively effect the exchange-coupling interactions of the two phases (Skomski and Coey, 1993). The phase boundary is also of importance in inductor core materials, where voids would degrade the overall permeability of the core (Silveyra et al., 2018). This work explores the phase boundary between the particles and metal matrix of 0-3 nanocomposites made using electroinfiltration. To do this, high-resolution S/TEM combined with EDS, is used to probe two different composite samples-a nickel/ironoxide composite and a permalloy/iron-oxide composite. Results show that the iron-oxide nanoparticles (IONs) are surrounded by metal to form a dense, void-free boundary.

II. MATERIALS AND METHODS
Iron-oxide nanoparticles with an average diameter of 31 nm were synthesized using a two-step semi-batch thermal decomposition method adapted from Vreeland et al. and put into suspension (Vreeland et al., 2015). The electroinfiltration process flow can be seen in Fig. 1b. First, a photoresist layer is patterned on top of a silicon substrate that has a 20 nm titanium (Ti) adhesion layer and a 100 nm gold (Au) seed layer sputtered upon it, and the ironoxide nanoparticles are drop cast into the resulting molds. These molds are left to dry into a porous particle film for at least 3 hours in a fume hood. In the next step, the particle layer is placed into an electroplating bath of either Ni or NiFe, where these metals are electroplated through the film at a current density of 10 mA/cm 2 at 54 ○ C (for Ni) or 25 ○ C (for NiFe) with no agitation. A NdFeB retaining magnet, placed ∼1 cm from the backside of the substrate, is used to hold the magnetic particles in place during plating. Finally, the film is removed from the plating bath, rinsed, and the photoresist layer is stripped to leave behind a composite material. After fabrication, a FEI Helios NanoLab 600 Dual-Beam FIB/SEM was used to cut lamellae of the composite materials, and S/TEM images were taken using a FEI Tecnai F20 S/TEM instrument operating at 200 keV with an EDS detector.

III. RESULTS AND DISCUSSION
A. Identification of layers within the sample cross-sections EDS line scans taken along the thickness of each of the samples are used to reveal and identify the presence different layers formed during fabrication. Fig. 2a shows a line scan plot for the Ni/ION composite with each layer labeled, and Fig. 2b shows a line scan plot for the NiFe/ION composite with each layer labeled. In Fig. 2a, the presence of silicon and silicon oxide come from the substrate. Following these should be a 20 nm titanium adhesion layer, but this does not appear due to the low resolution of the line scan. Instead, the 100 nm gold seed layer is observed, present directly above the substrate. After this, the 2 μm Ni/ION composite layer appears, as supported by the combined presence of nickel and iron in the line scan. After the composite layer, a transition is made to a pure nickel layer. Due to the presence of an un-infiltrated layer of particles that is ∼1 μm thick above the pure nickel layer, it is suggested that stress from the electroplating step caused the nickel to stop infiltrating the particle layer and, instead, heave it upward as pure nickel continued to plate below it. The exact cause of this phenomenon requires further exploration.
In Fig. 2b, after the expected silicon, silicon oxide, and gold layers, the NiFe/ION composite layer comes into view. Unlike the Ni/ION composite, this layer does not transition into a pure NiFe layer at any point, and instead transitions to an un-infiltrated particle layer. This transition suggests the plating step had been stopped prematurely before the particle layer could be filled, or before stress caused the particle layer to be heaved up, as seen in Fig. 2a.

B. Inspection of composite layers via S/TEM
Next, inspection of the different layers of the cross-section is necessary to examine how the particles are distributed throughout the material and better understand the connectivity of the composite. Fig. 3 shows S/TEM images taken of the Ni/ION and NiFe/ION composite samples. In Fig. 3a,  layer, which appears to be densely filled with nanoparticles. The presence of different shades of black and gray from one nanoparticle to the next seen in this image is a consequence of the lack of alignment of the different particles and their crystal orientations. In Fig. 3b, an image of the pure nickel layer can be seen, with a lack of the many nanoparticles that were densely packed in the Ni/ION layer. A few single IONs can be seen within this layer, but the make-up is primarily nickel metal. In Fig. 3c, the un-infiltrated layer of particles can be seen, with the various uninfiltrated nanoparticles visible and surrounded by empty space. Due to Z-contrast, elements with greater atomic mass are shown to be brighter. The presence of nanoparticles in the pure nickel layer, as seen in Fig. 3b, was previously unknown, and supports the idea that the pure nickel layer heaves up the un-infiltrated particle layer after some point during the plating step, as the particles seen in the pure nickel layer may have fell off during this lifting process and were surrounded by nickel. Fig. 3d shows the cross-section of the NiFe/ION composite layer, with nanoparticles present throughout. Fig. 3e shows the interface between the NiFe/ION composite layer and the un-infiltrated particles layer, which indicates the position where the electroplating of the permalloy was stopped. Finally, in Fig. 3f, the un-infiltrated nanoparticle layer in the sample is shown, with particles shown to be densely packed together and surrounded by empty space. These images support that there has been a successful infiltration of permalloy through the particle layer, as particles embedded in the metal matrix are clearly visible.

C. EDS confirmation of particle/matrix interface and composition
The final part of this investigation confirms that the electroplated metals do surround the particles in the composite and form a dense and void-free boundary, as evaluated with 2D EDS mapping. Fig. 4 shows two high-resolution EDS maps of an area in each of the two composites, with Fig. 4a showing the Ni/ION composite, and Fig. 4b showing the NiFe/ION composite. The color green represents the presence of nickel, and the color red represents the presence of iron. In Fig. 4a, there is a clear separation between the nickel matrix (green) and the iron-oxide nanoparticles (red) without any visible voids, indicating a dense, void-free boundary between the two phases. Likewise, Fig. 4b shows a similar behavior for the NiFe/ION composite layer, where the Ni-Fe matrix (red and green) are separate from the iron-oxide nanoparticles (red), with no visible voids present at their interfaces. These images support the conclusion that these composite materials successfully lead to a lack of voids between the particle and matrix phases.

IV. CONCLUSION
This work offers a definite confirmation that the boundary between the particle and metal matrix phase of electro-infiltrated Ni/ION and NiFe/ION nanocomposites are dense, and void-free. Previous studies of these composites confirmed the successful infiltration by examining the properties of the composites and/or through the use of SEM and EDS, but investigation of the phase boundary was limited by resolution. Through the use of S/TEM and EDS, it is now known that metals electroplated through particle film in the electro-infiltration process do fill in the spaces between particles which leads to a dense and void-free boundary. This dense, void-free interface between the particle and matrix phases in the composites is necessary for applications such as exchange-spring magnets and improved inductor cores, and further suggests the electro-infiltration fabrication process as an attractive pathway towards realizing such materials. This work provides support for past studies on electro-infiltrated 0-3 nanocomposites and encourages further investigations on the applications of these composite materials.