Scalable Manufacturing of AgCu_{40(wt %)}–WC Nanocomposite Microwires

Metal matrix nanocomposites (MMNC) have been broadly investigated during the recent decade for their wide range of applications. The particle strengthening effect becomes remarkable when the particle size scales down to nanometers, because the interaction between nanoparticles and dislocation stands out [1], especially for thermally stable ceramic nanoparticles like silicon carbide [2,3], tungsten carbide [4], titanium carbide [5], titanium diboride [6], and aluminum oxide [7]. One of the primary issues for mass production of MMNCs is the poor wettability between metal matrix and ceramic nanoparticles, which could induce the nanoparticle agglomeration and consequently poor dispersion. Thus, most research has focused on identifying suitable metal matrix–ceramic nanoparticle combinations and renovating efficient nanoparticle incorporation processes to achieve a homogeneous nanoparticle dispersion in MMNC, in which the most applicable techniques have been the ultrasonic cavitation-based process [8–10] and stir casting [11].

Most MMNCs obtain extraordinary mechanical strength, which is promising for applications in the fields of automobile [12], aerospace [13], and biomedical [14] industries. However, bulk MMNC as functional materials had not been thoroughly developed, characterized, and applied in the industry due to the manufacturing difficulties and the limitation for high-volume nanoparticle incorporation and dispersion. Especially for metal microwires, efficient nanoparticle incorporation could enhance their mechanical properties superior to traditional composites. Furthermore, a comprehensive analysis of the mechanical and electrical properties of MMNC microwires is urgently needed to broaden their practical applications in multiple fields. For example, MMNC microwires could be fabricated to biomedical intracellular sensors and bioprobes for living cell bioelectricity signal detection, which demands micro/nanowires of both excellent mechanical properties for tissue penetration and high electrical conductivity for signal quality [15] to substitute the semiconductor microwires [16]. MMNC microwires could also be applied as filler materials in brazing, especially between metals and ceramics, since nanoparticles could enhance the filler performance and the additional ceramic composition could enhance the wettability between fillers and ceramic base materials [17,18].

This study integrated the production of MMNC using a salt assisted stir casting with thermal fiber drawing of AgCu₄₀ microwires for the first time. Solidification processing, such as stir casting assisted by molten salt, was carried out to produce the AgCu₄₀–WC nanocomposite efficiently. By taking advantages of good wettability between Cu and WC at high temperature [4,19] and a nanoparticle self-stabilization mechanism [5], this research demonstrated efficient incorporation and uniform self-dispersion of WC nanoparticles into AgCu₄₀ alloys. After that, thermal fiber drawing was utilized to fabricate microwires from the bulk nanocomposite as a scalable, efficient, and economical production method. The thermal fiber drawing method, also known as Taylor wire process, has been developed as a manufacturing method to produce metallic microwires with a high production rate. It has realized that the fabrication of microwires with high aspect ratios for various metals like Sn, Cu, Zn, Au [16,20,21], for multiple functions, such as flexible in vivo fiber probes for brain signal detection [22] and metal core assisted optical fiber [23]. AgCu₄₀ alloy was used as the matrix because of three reasons: first, Ag and Cu wet well with WC; second, the alloy of this composition has a liquidus temperature of 840 °C, which matches well with the borosilicate glass as the cladding [21,24]; finally, the addition of Cu could tune the molten metal core viscosity when the Ag–Cu core is in a semisolid zone during the thermal fiber drawing process.

The schematic of the experimental setup for stir casting is shown in Fig. 1(a). Pure Ag and Cu bulk samples were weighed and melted together in a graphite crucible within a furnace at 1100 °C under the protection of argon. After both metals were melted, the furnace temperature was set to 950 °C to maintain the alloy in the liquid state. Then, a three-blade steel stirrer was placed right above the metal melt surface. WC nanoparticles (US Research Nanomaterial, Houston, TX) were premixed with sodium chloride (NaCl) and potassium aluminum fluoride (KAlF₄) powders at a volume ratio of 2:9:9. After the powder mixture was loaded with the metal melt, the mechanical stirrer was turned on at 300 rpm for 30 min. The furnace was maintained at 950 °C for another 30 min to evaporate most of the molten salts. The metal melt was then taken out of the crucible and cooled down to room temperature. Finally, the remaining salts, which appear in white, were removed from the top surface of the bulk metal matrix nanocomposite ingot.

Stir casting with the assistance of molten salt has already been demonstrated as a solid method in nanocomposite incorporation [5,6]. Such method created a vortex in both molten metal and molten salts, accelerating the liquid flow to expedite the nanoparticle self-incorporation. Molten salts contributed to three aspects in the incorporation process: first, they embodied the nanoparticles and covered the metal melts to prevent oxidation; second, since WC nanoparticles have a better wettability with the molten metal than with the molten salts, nanoparticle preferred to stay in molten metals, thus, to initiated a self-incorporation; third, the fluoride salt dissolved the inevitable surface oxide on the metal melt, facilitating the engulfment of nanoparticles into molten metal. The stirring blade was placed at the interface to create turbulence for more efficient nanoparticle incorporation.

The thermal fiber drawing process was used for manufacturing of ultralong nanocomposite microwires. When a macroscale glass/polymer preform rod with a metal core is heated above the melting temperature of the metal core while the glass/polymer cladding becomes viscous, the rod with the metal core is stretched and scaled down into microwires through a constant feeding and pulling. AgCu₄₀–WC nanocomposite ingot was melted in an alumina crucible under argon protection and then solidified to rods of 2.0 mm in diameter after the nanocomposite melt was sucked into fused silica tubes by a vacuum pump. The nanocomposite rods were then taken out by mechanically breaking the fused silica tubes before they were inserted into borosilicate glass tubes (Corning Pyrex, National Scientific Company, Quakertown, PA), which have an inner diameter of 2.0 mm and an outer diameter of 6.5 mm. The glass tube was sealed under vacuum using an oxygen/propylene torch, to serve as the preform for thermal drawing. After the preform was drawn in a furnace, multiple drawn-out fibers were inserted into another borosilicate tube for following drawings. The diameter of the nanocomposite core reduced from 2.0 mm to 200 μm, 60 μm, and 6 μm subsequently. The schematic of the experimental setup is shown in Figs. 1(b) and 1(c). The furnace temperature during the drawing process was set to be 850 °C, where nanocomposite was melted, and the borosilicate glasses maintained a relatively high viscosity around 10⁷ Pa·s, which was suitable for thermal drawing.

For characterization, the borosilicate glass cladding for the nanocomposite microwires was dissolved by a combination of hydrofluoric acid and buffered solution wet etching. The microwires with glass cladding were first submerged into a 49% hydrofluoric acid solution for 10 min, and then were transferred into a 49% buffered solution for 2 h, followed by de-ionized water rinsing. The exposed microwires were placed on a silicon wafer for baking on a hot plate at 100 °C under a delicate nitrogen flow. The mechanical and electrical performances of the microwires were also characterized.

AgCu₄₀ with 10 vol % WC and 22 vol % WC nanocomposite samples were prepared by stir casting process assisted by molten salt. The scanning electron microscopy (SEM) in backscattering mode was used to study the nanocomposite samples. The nanoparticle dispersion of AgCu₄₀–10 vol % WC indicated a certain degree of pseudo-dispersion, as shown in Fig. 2(a). In the area of highly concentrated areas, the nanoparticles are still dispersed, as shown in the magnified image Fig. 2(b). The corresponding light intensity analysis images of Fig. 2(a) were shown in Fig. 2(c). Similar images of AgCu₄₀–22 vol % WC were shown in Figs. 2(d)–2(f). The relatively bright phase represented an area of high WC nanoparticle concentration and the dark phase represents an area of much lower WC concentration. ImageJ was used to analyze image intensity that could quantitatively indicate the comparison of nanoparticle dispersion between these two samples. Higher intensity variation, denoted by the standard deviation, represented a higher possibility of the appearance of “microclusters,” implying AgCu₄₀–10 vol % WC obtained a pseudo-dispersion. Furthermore, nanoparticle size distribution was studied by analyzing the SEM images using ImageJ, shown in Figs. 2(g) and 2(h). Because the original nanoparticles were 150–200 nm in diameter and the noises during the image processing could cause inaccurate counts of nanoparticles that are smaller than 100 nm, the nanoparticle size distribution was reasonable, indicating that no severe agglomeration and sintering happened during the incorporation.

Grain structures were analyzed on AgCu₄₀–22 vol % samples and AgCu₄₀ sample after applying etchants (ammonium hydroxide (NH₄OH) and hydrogen peroxide (H₂O₂)) on the polished surface. The grain structure of the pure AgCu₄₀ sample was shown in Fig. 2(i). Given that Ag and Cu have limited miscibility with each other, AgCu₄₀ obtained a majority of bilayer eutectic phase and Cu-rich phase. During the solidification process, the Cu-rich phase nucleated and grew at the first stage when the alloy cooled down and the eutectic phase grew along the Cu-rich phase. However, in nanocomposites solidification process, the Cu-rich phase growth was restricted by the nanoparticle. Therefore, the grain size of the copper-rich phase reduced from 16.3 μm to 5.2 μm, shown in Fig. 2(j). Afterward, the nanoparticles were pushed to the solidification front and the nanoparticles affected the solidification of eutectic phase by blocking the directional growth. Thus, SEM images revealed that the eutectic phase embodied most of the nanoparticles and the grain structure altered from organized bilayer dendrite to randomly and disorderly oriented. The microstructure transformation and grain refinement could contribute to the changes in the mechanical strength and the electrical conductivity. The characterization of the eutectic grain could not be entirely analyzed because nanoparticles interfered with the observation of the grain boundaries.

The success of nanocomposite fabrication has established the system of AgCu₄₀–WC, which applied the self-dispersion mechanism to overcome the nanoparticle agglomeration during the incorporation process. The self-dispersion mechanism in this study referred to the force balancing between nanoparticle–nanoparticle van der Waals attraction and nanoparticle-molten metal surface tension. It was also known as nanoparticle self-stabilization mechanism, where a high energy barrier prevents nanoparticles from aggregation because of a reasonable wettability between WC nanoparticles and AgCu molten metals [2]. However, such theories also implied that pseudo-dispersion would happen when the attraction force is too large, which is due to the large nanoparticle size, i.e., 150 nm.

Vickers microhardness test was performed on AgCu₄₀ nanocomposite samples that were obtained from stir casting, under the condition of 200 gf for 10 s. Figure 3(a) showed the microhardness of the nanocomposite with respect to the nanoparticle concentration, compared with the pure Ag and pure Cu samples manufactured in the same casting method. AgCu₄₀–22 vol % WC raised the microhardness from 90 HV to 147 HV, and the result also showed that the Vickers microhardness linearly increases with the nanoparticle concentration. The microhardness curve indicated that WC nanoparticles contributed to high hardness (more than 50% enhancement), refined grain structure, and nanoscale precipitates.

Furthermore, the electric conductivity test was performed by CDE ResMap 178 4-point probe (Creative Design Engineering, Cupertino, CA) on nanocomposite thin sheets, as shown in Fig. 3(b). The electrical conductivities of AgCu₄₀–10 vol % WC and AgCu₄₀–22 vol % WC are 19.7±1.0% and 20.1±2.5% IACS, respectively. The result suggested that the conductivity of the nanocomposites decayed in a near logarithmic trend with the nanoparticle concentration. This is mainly due to the interfacial energy difference at the AgCu₄₀ matrix–WC nanoparticle boundaries, associated with the different free electron energy levels and different electronic band structures [25–27]. Because nanoparticles restricted the grain growth, there were more grain boundaries in the nanocomposite, thus increasing the possibility of electron scattering at the grain boundary. As a result, the electrical conductivity decreased.

For nanoparticle redistribution analysis, nanocomposite fibers from three consequent thermal fiber drawings were polished from the side surface. AgCu₄₀–10 vol % WC was used as a sample to produce the microwires, while the viscosity of AgCu₄₀–22 vol % WC was too high for thermal drawing. Images obtained from SEM (Fig. 4) show nanoparticle pseudo-dispersion and redistribution during consequent thermal drawing processes. The pseudo-dispersion zones of high WC nanoparticle loading were well distributed after the first drawing and divided into smaller pseudo-dispersion zones after the second drawing. It seems that the diameter reduction did not significantly influence nanoparticle dispersion inside the microclusters. However, when nanocomposite wire was reduced to 6 μm, smaller than the average size of microclusters, the microclusters were aligned in the microwire and constrained by the glass cladding, forming a bamboo shape with nonuniform diameters. These microwires, thus, could not further be characterized by tensile testing.

The bamboo-like structure in continuous microwires was observed after thermal fiber drawing. The primary reason for the node structure was the high viscosity of the pseudo-dispersion zones when the microwires were reduced to 5 μm. The viscosity could be even larger than that of the glass cladding at the drawing temperature. Therefore, the nanocomposite core could not shrink along with the glass cladding. Further drawing to submicron size would not guarantee the continuity of the metal wire due to the pseudo-dispersion zones. These problems could be potentially solved through further researching using nanoparticles of smaller size and better dispersion.

Tensile testing by DMA Q800 (TA Instrument, New Castle, DE) was performed to measure the mechanical properties of the AgCu₄₀–10 vol % WC nanocomposite microwires (60 μm in diameter) obtained from the second drawings, compared with AgCu₄₀ microwires produced in the same method, shown in Fig. 5. The mechanical properties were also illustrated in Table 1. The strain rate was set to $4×10−4s−1$ in the tensile test. AgCu₄₀–10 vol % WC microwires offered an ultimate tensile strength of 354$±$35 MPa, while only 203 MPa for the reference AgCu₄₀ microwire sample. Furthermore, the nanocomposite microwires maintained an elongation of 5.2%$±$1.1%, similar to that of the pure alloy samples. From the results of tensile testing, the nanocomposite microwire obtained an improved ultimate tensile stress without losing its ductility. The strength improvement could be contributed to the Orowan strengthening, the grain structure alternation, and grain refinement. However, defects on the microwire surface, owing to the mismatch of thermal expansion coefficient between metal and glass, and inhomogeneous nanoparticle dispersion could affect the results and weaken the strengthening effect, making the strength of the nanocomposite microwires smaller than the actual value.

This result implied that nanocomposite microwires are promising to be utilized as bioprobes and biosensors, owing to its mechanical strength and conductivity. Compared to the current application using tin (Sn) with polymer cladding as the bioprobes [22], where Sn only obtains an electrical conductivity of 15% IACS and a tensile strength of 220 MPa, AgCu₄₀–10 vol % WC has better mechanical strength and higher electrical conductivity. Furthermore, bare microwires could be used without structural supporting to further enhance the quality of received signals.

Here, we successfully manufactured the AgCu₄₀ nanocomposite with highly concentrated WC nanoparticles (up to 22 vol %) via a stir casting process assisted by molten salt, and thermal drawing method was applied to fabricate continuous nanocomposite microwires. Based on the thermal drawing method, this microwire production method can be scaled up easily for mass production. We have produced bulk nanocomposite alloy with Vickers microhardness of 147 HV and electrical conductivity of 20.1% IACS. Furthermore, we were able to produce AgCu₄₀–10 vol % WC nanocomposite microwires with an ultimate tensile strength of 354$±$35 MPa and an elongation of 5.2$±$1.1%. The self-dispersed WC nanoparticles remarkably improve the performance of AgCu₄₀ alloy and provide opportunities for functional devices. However, there is still space for improvement in the process, and further research is needed.

We thank J. Zhao at University of California, Los Angeles for his discussion in thermal fiber drawing. We also thank C. Cao and G. Yao at University of California, Los Angeles for their input for nanoparticle incorporation. We appreciate the help form C. Linsley at University of California, Los Angeles for assistance on tensile testing.

• The National Science Foundation (NSF) under the Grant No. CMMI #1449395.