Fabrication of Tantalum and Hafnium Carbide Fibers via Forcespinning for Ultrahigh-Temperature Applications

In this work, a novel method for producing ultrafine tantalum and hafnium carbide fibers using the Forcespinning technique via a nonhalide-based sol-gel process was investigated. An optimal solution viscosity range was systematically determined via rheological studies of neat PAN/DMF as a function of fiber formation. Subsequently, Forcespinning parameters were also systemically studied to determine the optimal rotational velocity and spinneret-to-collecting rod distance required for ideal fiber formation. TaC and HfC fibers were synthesized via Forcespinning utilizing a mixture of PAN and refractory transition metal alkoxides (i.e., tantalum (V) ethoxide and hafnium (IV) tert-butoxide) in DMF solution based on optimal conditions determined from the neat PAN/DMF. In all instances after calcination, powder X-ray diffraction (PXRD) and energy dispersive spectroscopy (EDS) indicated that TaC and HfC fibers were produced. TGA/DSC confirmed the chemical stability of the resulting fibers.


Introduction
While refractory transition metal-carbides (RTM-Cs) have been studied as far back as the 1800s, interest in these materials increased in the late 1950s as the space race intensified [1]. Research of RTM-Cs gained even more interest during the late 1980s for aerospace applications [1]. As hypersonic technology pushes speeds beyond Mach 5, surface temperatures can reach upwards to 2000°C in the presence of highly reactive gas species [1]. erefore, materials that can survive/mitigate these extreme environments are critical for the future application of hypersonic technologies. Currently, carbon/carbon (C/C) composites are the material of choice due to their mechanical, thermal, and shock properties. Unfortunately, carbon is susceptible to oxidation at high temperature in various environments and has poor stability in strong erosion environments [2]. Silicon carbide (SiC) fibers have also been investigated as a replacement for carbon fibers. However, SiC fibers can only retain their mechanical properties up to 1900°C. SiC generates gaseous reaction products through native oxidation from a combination of >1100°C, high pressure, and external gas flow [3]. Due to the thermal limitation of carbon and silicon carbide fibers handicapping ultrahigh-temperature performance, RTM-C materials for use in high-temperature environments have been identified as viable replacements [4]. RTM-C materials possess strong covalent bonding and degrees of metallic bonding that lead to high melting points (>3700°C (4000 K)), high hardness, good high-temperature strength, high resistance to oxidation, excellent electrical conductivity, and exceptional resistance to harsh chemical and thermal environments [2,[5][6][7].
is combination of properties allows RTM-Cs to survive in more extreme environments than that of existing structural materials. erefore, utilizing these materials as fiber reinforcements in composites would significantly enhance the thermal resistance as well as the damage tolerance in ultrahigh-temperature composites [2][3][4][5][6][7].
To use these RTM-C materials as fiber reinforcements in composites, a readily scalable route to produce high quality materials in fiber form is critical. Methods to produce hafnium and tantalum carbide (HfC and TaC) materials from literature have reported several routes that have led to the successful production of these materials in fiber and powder form [8]. However, most of these methods utilize harsh reagents and halide precursors of the refractory transition metal, which are known to lead to impurities upon degradation (i.e., Cl-contamination). e sol-gel process has demonstrated the feasibility to utilize metal alkoxides (groups IV, V, and VI) to synthesize particles and fibers, with little to no impurities [9][10][11][12].
Furthermore, these methods are not all conducive to produce quality fibers through traditional techniques. Currently, the most common ways to fabricate fibers include melt-spinning, self-assembly, phase separation, and electrospinning methods [13]. Due to the simplicity and lowcost operation of electrospinning, it is the most commonly used method to fabricate nanofibers. Unfortunately, this technique has material/solvent limitations due to electrostatic fields and intrinsically low throughput via traditional lab scale electrospinning fabrication which has shown only to produce around 0.1 g/h of fibers [14,15]. On the other hand, Oregon et al. produced polycaprolactone (PCL) fibers at a rate of 0.32 g in two minutes (scaled up to 9.6 g/h) via the Forcespinning TM technique which corresponds to a 3000% increase in output [14]. Forcespinning TM is a process developed by Sarker et al. that is similar to electrospinning. However, it uses centrifugal forces instead of electrostatic forces to generate fibers from a solution ejected from a spinning metal head [15]. is leads to the rapid production of fibrous materials ranging from nanoscale to micron scale [16]. By using centrifugal forces, the viable material options for Forcespinning TM are not limited by their dielectric properties allowing a wide spectrum of material combinations [17][18][19][20]. By combining Forcespinning TM with the solgel process, ultrahigh-temperature ceramic fibers such as silicon carbide (SiC) and RTM-C fibers such as titanium carbide (TiC) and zirconium carbide (ZrC) have been previously synthesized successfully. While Firbas et al. were the first to produce TiC and ZrC fibers via the Forcespinning TM technique, higher melting point TaC and HfC have yet to be synthesized via this method [17].
In this work, a method for producing stable TaC and HfC fibers using the Forcespinning TM technique and nonhalidebased sol-gel metal precursors is introduced. First, an ideal solution viscosity was systematically determined via rheological studies of neat PAN/DMF. Next, Forcespinning TM parameters were systemically studied to determine the optimal angular velocity and spinneret-to-collecting rod distance required for ideal fiber formation. Lastly, sol-gel preparation, fiber processing parameters, and characterization data were presented for TaC and HfC fibers. To our knowledge, this is the first time TaC and HfC fibers have been synthesized via Forcespinning TM . With these results, the high-throughput production of RTM-C fibers for reinforcements in composites is now viable through the Forcespinning TM technique.

Materials and Methods
2.1. Materials. All syntheses were performed inside an argon-filled glovebox, and all chemicals were used as received unless otherwise noted. Anhydrous dimethylformamide (DMF), anhydrous acetic acid, polyacrylonitrile (PAN, MW 150,000), and tantalum (V) ethoxide were purchased from Sigma Aldrich. Hafnium (IV) tert-butoxide was purchased from Alfa Aesar.

PAN Solution Preparation
. PAN solutions were prepared at various weight percentages ranging from 10 to 20% w/w. To prepare solutions, desired amounts of PAN and DMF were added to a chemical jar with a magnetic stir bar. e solution was then sealed and allowed to stir until PAN was completely dissolved. A 3 mL aliquot of the PAN solution was drawn for the spinning process, and 1 mL aliquot was drawn for rheological studies to compare the various PAN solution weight percentages.

RTM-C Precursor Solution Preparation.
Fiber precursor solution consisted of a PAN solution and tantalum (V) ethoxide solution combined. e PAN solution (13% w/w) recipe consisted of 3.897 g of PAN and 30.00 g of DMF. e reactants were added to a chemical jar with a magnetic stir bar; the solution was sealed and allowed to stir until PAN was completely dissolved. For the tantalum (V) ethoxide solution, 4.500 g of tantalum (V) ethoxide and 9.000 g of acetic acid was added to a chemical jar with a magnetic stir bar; the solution was sealed and allowed to stir until dissolved. e "fiber solution" was tantalum (V) ethoxide/acetic acid solution mixed with the PAN/DMF solution, which was stirred overnight. Lastly, a 3 mL aliquot of the fiber solution was drawn for the spinning process, and 1 mL aliquot was drawn for rheological studies to compare to the optimal neat PAN solution. e same process was used with hafnium (IV) tert-butoxide, instead of tantalum (V) ethoxide to generate HfC-fibers.

RTM-C Fiber Preparation.
e fibers were generated using the FibeRio L-1000 M Cyclone Forcespinning TM System. e Forcespinning TM technique uses centrifugal force to extrude the solution out of metal heads onto vertical metal rods at chosen radial separation distances. Spin speed and collecting rod distance were systematically studied, and results are seen in the manuscript and supporting information. Systematic studies led to the ideal program being a multistep program consisting of spinning at 1000 rpm for the five seconds and then ramping up to 3500 rpm for 1200 seconds. e metal rods were placed 17.78 cm away from the spinneret. e resulting fibers were then collected and placed in a box furnace at 220°C with a ramp rate of 5°C per minute and then were held at 220°C for two hours. After cooling to room temperature, the fibers were then placed in a tube furnace at 1600°C with 5°C per minute ramp rate under the flow of argon and dwelled for two hours. e resulting fibers were characterized by PXRD, SEM/EDS, and TGA.

Fiber
Characterization. Powder X-ray diffraction (PXRD) patterns were collected on a PANalytical X'Pert Pro diffractometer (5-70°two theta) employing CuKα radiation (1.5406Å) and a RTMS X'Celerator detector at a scan rate of 0.02°/sec using a zero-background holder. e PXRD patterns were analysed using JADE 9.6.0 software (Materials Data Inc., Livermore, CA) and indexed using the Powder Diffraction File PDF-4+. ermogravimetric analyses (TGA)/differential scanning calorimetry (DSC) analyses were taken under flowing air and nitrogen atmosphere from room temperature to 1000°C at a ramp rate of 10°C/min using a Mettler Toledo STARe System. Scanning electron microscopy was conducted using a Hitachi S-5200 FE-SEM with an Oxford Instruments X-MAX EDAX accessory operating with an accelerating voltage of 20 kV.

Rheological Testing.
e production of fiber materials is nontrivial and relies on numerous synthetic and instrumental variables. Of the many parameters, the viscosity of the sol-gel is the most readily and easily tailorable to produce fiber materials in any instrument. erefore, rheological studies were done on PAN/DMF solutions to determine the optimal solution viscosity needed to produce fibers via Forcespinning TM utilizing our instrumentation.
Rheological testing of the neat PAN/DMF solutions of various weight percentages from 10 to 20 wt% was measured as a function of shear rate. Forcespinning TM polymer solutions are generally higher in polymer content than electrospinning solutions, due to the variation in the jet generation and extension behavior of the two methods. e neat solution served as a surrogate to identify the optimal conditions for RTM-C solutions, and their viscosity profile results are shown in Figure 1(a). At 10 wt % (lowest loading), the PAN/DMF solution displays nearly Newtonian behavior, up to a shear rate of ∼200 1/s, with a viscosity of 570 mPa s.
As PAN content is increased, the polymer solutions show increasing viscosity levels, and the transition to shear thinning behavior occurs at progressively lower shear rates, as the system shows further chain entanglement. All these solutions are expected to have chain entangled character, based on prior literature of PAN/DMF solutions [20]. e stable formation of a fiber requires a polymer concentration sufficient to obtain chain entanglement. Wang et al. presented rheological characterization of PAN/DMF solutions that determined a minimum concentration (C min ) for the entry into the chain entanglement regime at approximately 5 wt%. For polymer solutions above C min , Wang et al. found that the specific viscosity follows the power law η sp ∼ ϕ 4.78 [21]. Figure 1(b) plots the specific viscosity vs. volume % PAN for estimation of the power law characteristic for these solutions.
e higher concentration solutions made here differ in molecular weight and have a power law exponent of 6.25, indicating rapid development of elastic character with drying.
e production of fibers involves the surface tension and viscosity of the polymer solution, chain entanglement and relaxation dynamics of the polymer solution, evaporation rate of the solvent, and extension forces on the fiber in transit from the orifice tip to the collector system [22]. Fiber production by Forcespinning TM was first tested on pure PAN/DMF solutions to correlate solutions viscosity with fiber formation, as viscoelastic instabilities can lead to formation of beads when viscosity is low. As can be seen from Figure 2, the average viscosity obtained from a shear rate ranging from 0 to 1000 1/s of the different solutions increased as the PAN wt % increased. e 10-13 wt % solution yielded poor nanofiber morphology with droplets and beads due to low viscosity as seen in Figure S2. e 17-20 wt % produced short nanofibers of low yield indicating that the viscosity of the solution is too high for the given centrifugal force as seen in Figure S2. However, the 15 wt % yielded uniform fibers with no droplets. e measured viscosity in Figure 1(a) is shear thinning, and therefore, values at the highest shear rate of 1000 1/s were chosen to characterize suitable fiber production. is gives 2.80 Pa s as the viscosity target for the RTM-C precursor solution.
Control over fiber properties including diameter, regularity, and shape requires tailoring the processing parameters of the Forcespinning TM process. Operationally, there are three instrument parameters which can be varied to control fiber formation: angular velocity, Ω, collector distance, and the polymer concentration. Forcespinning TM has a five-step process of fiber formation. e steps include (1) jet exit, (2) orbital trajectory, (3) aerodynamic fiber vibration, (4) orbital expansion, and (5) fiber collection. ere is a critical angular velocity to produce a jet, but there is a second critical angular velocity in which the jet diameter is reduced well below the tip orifice diameter [8]. us, the fiber size is relatively insensitive to the orifice diameter, which is important for the capillary backpressure needed to induce flow to the tip. ere is a minimum rotational speed related to the fiber extension phase as well to prevent the fibers from being pulled back to the spinneret shaft, and a maximum speed related to the break-up of the polymer solution jet. erefore, the 15 wt % PAN solution was investigated at different speeds and collector distances to determine the optimal conditions. As seen in Figure S3, fiber formation is drastically affected by varying spin speeds and collecting distances. Low rpms (≤2000 rpm) lead to droplet formation due to not exceeding the second critical angular velocity needed to reduce the fiber diameter below the orifice diameter. Too high (≥5000 rpm) of spin speed leads to very small fibers that lack enough strength to form continuous fibers. e collecting distance also needs to be optimized in order to determine at which distance from the spinneret the fibers are fully formed. erefore, we investigated distances ranging from around 15 cm-18 cm for the collector. Systematic investigation of these variables based on the fiber formation and yield led to 3500 rpm and a collecting distance of 17.78 cm to be the optimal speed and distance required for the best fiber formation.
It was anticipated that the addition of the tantalum (V) ethoxide/acetic acid solution to the 15% wt PAN/DMF solution would increase the viscosity, ultimately leading to a change of solution performance under the previously determined conditions. From previous studies with pure PAN/ DMF solution, the optimal solution viscosity required for our instrumentation was 2.80 Pa s. To account for the additional attraction between Ta-ethoxide and PAN, we Advances in Materials Science and Engineering reduced the PAN/DMF concentration to 13 wt % resulting in a total of 33 wt % tantalum (V) ethoxide/acetic acid solution to generate the "fiber solution." e rheology of the fiber solution was investigated, in comparison with the neat 15 wt % PAN/DMF solution, and their behavior was comparable as seen in Figure 1(c). At low shear rate, the viscosity for the TaC precursor fiber solution is higher than the 15 % PAN/DMF solution, indicating that the Taethoxide generates additional interaction between the PAN chains. e 1000 1/s shear rate viscosity of the TaC precursor was 3.07 Pa s, which is comparable to the 15% PAN/DMF solution determined to be optimal. is procedure was replicated for the Hf tert-butoxide/acetic acid solution. us, these solutions were used throughout the rest of this work. After the RTM-C fiber solution was made, the fiber materials were generated at 3500 RPM with the collecting rods at 17.78 cm away from the spinneret.

PXRD.
After green body curing at 220°C and carbonization at 1600°C of the reagents, PXRD data were collected on the isolated products. As seen in Figure 3(a), the product from the Ta "fiber solution" yielded diffraction peaks at 35.5°( 111), 40.7°(200), 58.9°(220), and 70.0°(311) which was indexed to TaC (PDF #00-035-0801) [6,[21][22][23][24][25][26]. ere was some presence of tantalum oxide (Ta 2 O 5 , PDF #00-025-0922) as this is typical when using metal alkoxides as a metal precursor [11]. Using semiquantitative (S-Q) phase analysis in Bruker's Diffrac Eva V5.2 software, it was estimated that Ta-fibers were 91.8% TaC (00-035-0801) and 8.2% Ta 2 O 5 (00-025-0922). Figure 3(b) presents the PXRD spectra of the product obtained from the Hf solution. e spectra exhibited diffraction peaks at 33.4°(111), 38.8°(200), and 56.1°(220) which was indexed to HfC (PDF #00-039-1491). Remnants of hafnium oxide (HfO 2 , PDF #00-053-0560) are present as Advances in Materials Science and Engineering seen in the spectra as this is typical when using metal alkoxides as a metal precursor, but the major product is HfC [11]. Using semiquantitative (S-Q) phase analysis in Bruker's Diffrac Eva V5.2, it was estimated that Hf-fibers were 96.8% HfC (00-039-1491) and 3.2% HfO 2 (00-053-0560). Figure 4 shows the SEM images of the assynthesized TaC fibers. e TaC fibers exhibited diameters with an average size of 475 nm. Figure 5 presents the SEM image along with EDS images of tantalum and carbon collected on the TaC fibers. e EDS analysis confirms the presence of tantalum and carbon over the selected area as seen in the bottom image. Figure 6 shows the SEM images of the as-synthesized HfC fibers. e HfC fibers exhibited diameters with an average size of 340 nm. Figure 7 presents the SEM image along with EDS images of hafnium and carbon collected on the HfC fibers. e EDS analysis confirms the presence of hafnium and carbon over the selected area as seen in the bottom image. e TGA curve in Figure 8(a) shows that the TaC fibers were thermally stable below around 450°C while in air. After 450°C, an increase in weight was observed which is attributed to the oxidation of TaC to Ta 2 O 5 which is further verified by PXRD as seen in S4 [6,22]. Note that the magnitude of this weight increase indicates that oxidation may be limited to the surface of TaC via the formation of a Ta 2 O 5 passivation layer being formed. e oxidation mechanism of TaC is as follows [27] [29]:

SEM/EDS.
After around 560°C, the weight loss was due to the oxidation of any free carbon present and loss of CO 2 from the system [20]. Also, as seen in Figure 8(b), the fibers  Advances in Materials Science and Engineering exhibited no change in mass (decomposition) under N 2 . e TGA curve in Figure 8 shows that the synthesized HfC fibers exhibited thermal stability below around 500°C while in air.
After 500°C, an increase in mass was observed due to the oxidation of HfC to HfO 2 and verified by PXRD (S4). e oxidation of HfC is as follows [21,22]: After around 650°C, the weight loss was due to the oxidation of any free carbon present and loss of CO 2 from the system [20]. Also, as seen in Figure 8(b), the fibers exhibited no change in mass (decomposition) under N 2 . Note that the weight increases at around 500°C is larger, and the weight decreases at around 650°C which is smaller than that of the TaC fibers. is may be due to the possibility that TaC fibers had more trapped free carbon than the HfC fibers. Raman spectra are presented in Figure S4.

Conclusions
TaC and HfC fibers have been successfully synthesized via the Forcespinning TM method. e rheology of the varying wt % of PAN and DMF was determined to understand the entanglement and optimize the precursor solution for fiber formation. is surrogate sample solution led to the optimal fiber solution parameters to produce RTM-C fibers. e RTM-C fibers were characterized in order to confirm their formation, crystal structure, and stability. Upon successful heat treatment of the fiber, PXRD confirmed the formation of crystalline TaC and HfC fibers. TGA analysis showed that the RTM-C fibers maintain oxidation resistance and thermal stability in air up to 450°C. Microscopy analysis determined the average diameter of the fibers was 475 nm and 340 nm for TaC and HfC fibers, respectively. Also, microscopy showed that Hf and Ta were distributed evenly throughout the fiber. is method allows for simple and rapid synthesis of these fibers with the ability to make grams of fibers quickly. Ongoing studies will examine the mechanical behavior, decreasing oxide content, and fiber incorporation in the composite.
Data Availability e rheological, powder X-ray diffraction, scanning electron microscopy/energy dispersive spectroscopy, and thermogravimetric analysis data used to support the findings of this study are included within the article.  Advances in Materials Science and Engineering