Polymer-derived HfSiCO monolithic ceramics and fibers using Starfire’s SHP-199

Abstract

The introduction of the first commercially available polycarbohafnium precursor opened new pathways in the polymer-to-ceramic route towards ultra-high temperature ceramics. We present a straightforward chemical crosslinking method to produce HfSiCO ceramics by reacting SHP-199 with tetraethoxysilane (TEOS), catalyzed by the intrinsic acidity of SHP-199. This sol-gel–like process produces robust HfSiCO monoliths and fibers with a 92% ceramic yield, excellent thermal stability, and structural integrity up to 1600 °C. While pyrolyzing SHP-199 alone results in nearly pure HfC, the SHP-199/TEOS mixture investigated here forms a multiphase ceramic containing HfO₂, SiO₂, and HfSiO₄, but no SiC or HfC. Strong Raman signals of carbon at 1600 °C suggest kinetically hindered reduction of the oxide. This work demonstrates the development of a synergistic Hf-Si network and provides an accessible pathway to high-temperature HfSiCO ceramics and fibers.

1. Introduction

Ultra-high temperature ceramics (UHTCs) include borides, carbides, and nitrides of group 4 and 5 transition metals. They are known for their exceptionally high melting points, low thermal expansion coefficients, high thermal shock resistance, and high hardness at elevated temperatures, enabling them to operate in extreme environments.^1–6 UHTCs are especially valuable for hypersonic flight and space re-entry vehicles.^2,7,8 HfC has the highest melting point among UHTCs, and HfB₂ has been identified as one of the most promising options for high-temperature applications due to its high thermal conductivity and oxidation resistance, which can be further improved by adding silicon carbide. ⁹

Traditionally, Hf-based UHTCs are produced by starting from HfO₂ or Hf metal through powder metallurgical processing involving high-temperature solid-state synthesis. The polymer-derived ceramic (PDC) route to UHTCs, as opposed to the solid-state process, is less resource-intensive, allows for chemical modifications that can influence the composition and properties of the final ceramic, and enables the fabrication of various ceramic forms.^6,10 Polymeric precursor mixtures can be melt-spun into fibers, spray, spin, or dip-coated onto substrates, cast into microstructural components, crosslinked with sol-gel processing, supercritically dried into aerogels, or shaped via additive manufacturing. Green bodies are then pyrolyzed to produce UHTCs. ¹¹

Hitherto, the PDC route towards Hf-based UHTCs involved the modification of polycarbosilanes and polycarbosilazanes using amines and alkoxides of Hf.^6,10 The advent of SHP-199 – the first commercially available polycarbohafnium precursor delivered by Starfire Systems – changed the perspective, and new studies were initiated to investigate and modify the hafnium precursor.^12–16 SHP-199 has been designed to produce crystalline HfC upon pyrolysis to 1600 °C. ¹² The polymer is delivered in a brown aqueous solution that contains 30-40% poly(hydroxyhafniumdioxobutyne) by mass. According to the patent literature, the polymer is synthesized by the reaction of tetrachlorohafnium and 2-butyne-1,4-diol in water (Scheme 1). ¹²

Scheme 1.

Reaction scheme to produce SHP-199 preceramic polymer. ¹² The ratio x:y is approximately 1:1.

Previous work with this polymer demonstrated that near-pure HfC powder is indeed produced after pyrolysis at temperatures of 1700 °C and higher, using rapid laser heating. ¹⁶ Traditional furnace pyrolysis techniques yielded impure HfC-containing ceramic powder at 1600 °C. ¹³ Co-pyrolysis of SHP-199 with a polysilazane or a siloxane showed improved oxidation resistance as compared to SHP-199 pyrolyzed on its own.^14,15

In this work, we demonstrate that chemical crosslinking involving SHP-199 is readily achievable, enabling the fabrication of monoliths and fibers. Hence, the approach unlocks the full potential of processing and shaping techniques for this polymer that the PDC route enables.

2. Methods

2.1. Materials

The SHP-199 solution (Starfire Systems, Inc., Glenville, NY) and 98% tetraethoxysilane (TEOS) (Sigma-Aldrich, St. Louis, MO) were used as-received. Ethanol 200 Proof (Decon Labs, King of Prussia, PA) was dried over 3Å molecular sieves (Alfa Aesar, Heysham, England). Water was filtered through an ARIES-1102D High Purity D.I. Loop (Aries Filterworks, West Berlin, NJ). Synthesized products were thermally treated in alumina crucibles placed into either the STF1200-50X600 quartz tube furnace (Across International, Sparks, NV) under flowing nitrogen, for samples pyrolyzed up to 1000 °C, or the TF1700 alumina tube furnace (Across International, Livingston, NJ) under flowing argon, for samples pyrolyzed up to 1600 °C. The N₂ and Ar were obtained from Airgas (Grand Prairie, TX).

2.2. Synthesis

For preparing HfC materials, SHP-199 was pipetted directly from the bottle and used as-is (S0). Excess water in SHP-199 can be removed by heating the solution to 40 °C under vacuum. The dry product (SDry) was a beige powder with a texture similar to wet sand, which could be easily redissolved in deionized water.

To prepare HfSiCO materials, SHP-199 and TEOS were added to a vial at a 1:2 mass ratio. Based on an estimated 70 wt% of water in SHP-199 (see subsection 2.4), this results in approximately a 1:1 molar ratio of TEOS-bound ethoxy groups to water molecules in the mixture. Note that the elemental ratio Si:Hf in the mixture is about 10:1. The combined SHP-199/TEOS precursor was gently shaken by hand for 3 minutes.

The condensation and gelation of the SHP-199/TEOS precursor mixture could be followed by probing/testing its viscosity. After 10-15 minutes, it became possible to manually draw preceramic fibers from the mixture using tweezers. The fibers were placed on a Teflon sheet and covered with an overturned glass dish to air-dry overnight (F0). If the mixture is instead allowed to solidify in the vial overnight, the SHP-199/TEOS precursor formed an amber monolithic gel. This monolith was rinsed once with water and once with ethanol before being air-dried in a fume hood overnight (M0).

2.3. Processing and pyrolysis

Samples S0, M0, and F0 were pyrolyzed under 300 mL/min flowing inert gas. One series of samples (S10, M10, and F10) was produced by heating under N₂ at a rate of 10 °C/min to 1000 °C, then holding the temperature for 6 hours before cooling. A second series (S16, M16, and F16) was pyrolyzed under Ar by first heating at 10 °C/min to 1000 °C, then increasing at 5 °C/min to 1600 °C and holding this temperature for 6 hours. In all furnace experiments, sacrificial graphite sheets placed upstream of the alumina crucible containing the sample prevented contamination from oxygen impurities in the gas stream. During pyrolysis, samples prepared directly from SHP-199 (S samples) broke into small grains and fine powders. In contrast, specimens prepared with the addition of TEOS (M and F samples) remained intact.

2.4. Characterization

Powder X-ray diffraction (XRD) analysis using a Bruker D8 Advance diffractometer (Madison, WI) with Cu Kα radiation, operating at 40 mA and 40 kV characterized the crystallinity of SDry and pyrolyzed products. The scan parameters were 2 °/min across the 10-80 ° 2θ range, with an increment of 0.03 °. The MDI JADE software (International Centre for Diffraction Data, Newtown Square, PA) analyzed crystallinity, phase contents, and calculated the crystallite sizes of the resulting ceramics.

Thermogravimetric analysis (TGA) using a TA Instruments Discovery SDT650 (New Castle, DE) was first used to estimate the water content in the aqueous SHP-199 solution. The flow of N₂ gas was set to 300 mL/min, and the system was purged for 30 minutes at room temperature. Thereafter, the furnace heated at a rate of 10 °C/min to 150 °C and help that temperature for 1 hour to drive off excess water. To this point, the system has lost approximately 70 wt%. The mass loss is due to both water and HCl formed by the decomposition of hafnium oxychloride (see subsection 3.1). ¹⁷ Subsequent thermal analysis of samples S0 and M0, ground with a mortar and pestle, followed the same heating protocol, then continued with heating the system at 10 °C/min to 1400 °C.

Scanning electron microscopy (SEM) was conducted using a Hitachi S-3000N microscope (Tokyo, Japan) with an accelerating voltage of 20kV. A Hummer VI sputter coater (Anatech Ltd., Alexandria, VA) with a Pd/Au target coated the samples prior to investigation.

Raman spectroscopy was performed using a Thermo Scientific DXR3 Raman Microscope spectrometer (Madison, WI) with a 532 nm laser set to 5 mW, covering a spectral range of 100-3500 cm^-1. The baseline was corrected using Omnic software.

3. Results

3.1. X-ray diffraction of SDry

Figure 1 presents the X-ray diffractogram of SDry, the solid obtained from drying the SHP-199 solution under vacuum at moderate temperatures. The diffractogram shows prominent peaks that are in good agreement with the pattern of HfOCl₂•8 H₂O. ¹⁸ The hafnium oxychloride is a product of dissolving HfCl₄ in water (see Scheme 1) and is generated during the polymer synthesis.^12,17

Figure 1.

X-ray diffractogram of SDry, the solid obtained by drying the SHP-199 solution. The lines indicate 2θ-values of major peaks reported for hafnium oxychloride, HfOCl₂ • 8 H₂O. ¹⁸

3.2. Sol-gel process

Since the SHP-199 solution is highly acidic (pH ≈ 1-2), it catalyzes the hydrolysis of TEOS with the water present in the solution (Scheme 2a). The acidity likely originates from the HfOCl₂, which generates HCl in aqueous solution. ¹⁹ However, Hf atoms in the polymer provide additional Lewis acidic metal centers. Hydrolyzed TEOS will further condense with itself (Scheme 2b) and via hydroxyl groups on the polycarbohafnium (Scheme 2c). In analogy with the sol-gel synthesis of HfSiO₄, the formation of Si-O-Hf bonds can be inferred. ²⁰ Confirmation may be possible using ¹⁷O nuclear magnetic resonance (NMR) spectroscopy or X-ray photoelectron spectroscopy.^21–24 These characterization methods are, however, outside the scope of this study.

Scheme 2.

Reaction scheme for (a) the acid-catalyzed hydrolysis and (b) condensation crosslinking of TEOS with itself and (c) with hafnium-bound hydroxyl groups.

3.3. Monolithic samples (M0, M10, M16)

The SHP-199/TEOS monoliths are shown at their wet, dried (M0), and pyrolyzed (M16) stages in Figure 2. The wet gel, after washing but before air-drying, has a brittle texture and a translucent light amber color (Figure 2a). The wet gel is kept in the same vial through mixing, crosslinking, and drying. Removing the wet sample from the vial is difficult and results in breakage because the sample is fragile at this stage. After air-drying overnight, the diameter shrinkage observed for sample M0 is 32%, measured with calipers. The sample turned a darker amber color, but it is still translucent and fully intact (Figure 2b). The pyrolysis process, which produces M16, caused the sample to crack into several large pieces (Figure 2c). This is due in part to an estimated 10% of additional shrinkage. The bulk of the sample turned opaque black, indicative of free carbon formation (see subsection 3.6).

Figure 2.

Digital photographs of monolithic crosslinked SHP-199/TEOS (a) before drying, (b) after air drying (M0), and (c) after pyrolysis to 1600 °C (M16).

3.4. Thermogravimetric analysis

The TGA analyses of the as-received polymer (S0) and the SHP-199/TEOS monolithic gel (M0) are shown in Figure 3. To enable a direct comparison, the mass remaining after the isothermal hold at 150 °C was normalized to 100 wt% for both samples. Clearly, S0 loses mass throughout the entire heating process. The most significant loss, about 20%, occurs up to 500 °C (heating rate 10 °C/min). Heating the sample further to 1100 °C results in an additional 5 wt% loss. Beyond 1100 °C, the mass loss increases significantly due to the onset of carbothermal reduction, where excess carbon reacts with available oxygen to produce CO, and hafnia, HfO₂, is transformed into hafnium carbide, HfC (see subsection 3.5).

Figure 3.

TGA of S0, M0, and reference TEOS microspheres. ²⁵

The TGA of M0 sharply contrasts with that of the pure polymer (S0). M0 only loses 7 wt% during heating to 600 °C. Afterward, it stays stable without significant additional mass loss up to 1200 °C. Even at 1400 °C, the system’s ceramic yield remains 92%, much higher than the 65% for the S0 sample at the same temperature. Crosslinked TEOS microspheres following a similar solvent-free synthesis with a 1:1 ratio of TEOS-bound ethoxy groups to water were prepared for comparison. ²⁵ These were subjected to the same TGA program, except for a 16-hour hold at 150 °C to account for the drying step in the literature. At 1400 °C, the TEOS microspheres retained 93% of their mass, losing only 7% by weight, consistent with previous studies. Using the TGA data of dried SHP-199 and TEOS microspheres and calculating a weighted average of ceramic yields—assuming 0.3 g of dried Hf-polymer after 150 °C for every 2 g of TEOS—an expected mass retention of 89% at 1200 °C is estimated for the combined system. Obviously, the yield of M0 at 1200 °C is 3 percentage points higher, and the advantage is further enhanced at 1400 °C. Overall, this “synergistic effect” suggests that the Hf-polymer reacted with TEOS and is well integrated within the TEOS/SiO₂ matrix, preventing excessive mass loss of organic groups during pyrolysis.

3.5. X-ray diffraction of pyrolyzed samples

The XRD patterns of SHP-199 annealed at 1000 and 1600 °C, S10 and S16, respectively, are shown in Figure 4. The pattern of S10 displays a hump at low 2θ-values (15-40 ° 2θ), indicative of a significant amorphous phase in the sample. On top of this, the XRD shows major Bragg peaks that can be used to identify two phases of crystalline hafnia, monoclinic m-HfO₂ and tetragonal t-HfO₂, in approximately equal amounts.^26,27 Based on the peak widths, the MDI JADE software approximated crystallite sizes of 13 nm for m-HfO₂ and 9 nm for t-HfO₂. Disordered carbon is expected to produce diffuse Bragg peaks near 25 ° and 42 ° 2θ. ²⁸ However, the high amount of m-HfO₂ makes these signatures difficult to resolve. Note that even ordered graphene sheets are not discernible in XRD spectra against hafnia. ²⁹ The XRD pattern of the sample annealed at 1600 °C, S16, exhibits intense diffraction peaks of HfC, with a calculated crystallite size of 43 nm and lattice parameter of 4.636 Å.^30,31 A close inspection of the lattice parameters derived from the XRD pattern suggests the presence of dissolved oxygen, resulting in an estimated composition of HfC_0.96.^32–34 The sample is nearly phase pure, with a small amount of HfO₂ indicated by low-intensity peaks between 28 ° and 32 ° 2θ, which correspond to the highest peaks of the m-HfO₂ and t-HfO₂ spectra. Thus, although it does not show up in the XRD, the amount of carbon in the S10 sample must be substantial, since it requires a C/HfO₂ molar ratio ≈ 3.1 after pyrolysis and before the onset of carbothermal reduction to attain HfC without remaining HfO₂. ³⁵

Figure 4.

X-ray diffractograms of pyrolyzed polymeric precursors: SHP-199 to 1000 °C (S10), SHP-199 to 1600 °C (S16), SHP-199/TEOS to 1000 °C (M10), and SHP-199/TEOS to 1600 °C (M16). Crystalline phases are identified by a selection of major peaks.

Previously, Mujib et al. ¹³ pyrolyzed SHP-199 following a similar procedure. Their XRD pattern of SHP-199 shows peaks of m-HfO₂ even at 1600 °C, when HfC dominates the pattern. The differences between the present products and those of Mujib et al. are attributed to the inclusion of sacrificial carbon sheets upstream of the sample crucible, which react with oxygen impurities in the gas stream. This reduces unwanted oxidation and allows a more complete conversion of the existing HfO₂ phases into HfC, which is otherwise difficult to obtain in solution-derived HfC.^35,36

XRD patterns of the SHP-199/TEOS monolithic gel annealed at 1000 and 1600 °C, labeled as M10 and M16, respectively, are shown in Figure 4. The diffraction pattern of M10 reveals a strong amorphous background, similar to that of S10. The main crystalline phase, accounting for about 90% of the crystalline content, is m-HfO₂, with a crystallite size of 25 nm. The remaining 10% of the crystalline phase is t-HfO₂, with a crystallite size of 14 nm.

After annealing at 1600 °C, the XRD pattern of M16 shows four crystalline phases: m-HfO₂, t-HfO₂, α-cristobalite-SiO₂, and HfSiO₄, along with a residual amorphous hump. The latter is partly due to grinding the sample in a mortar and pestle before analysis. About 60% of the crystalline portion is α-cristobalite-SiO₂ with a crystallite size of 60 nm. ³⁷ Hafnon, HfSiO₄, with a crystallite size of 50 nm, makes up roughly 20% of the crystalline content of M16. ³⁸ The remaining crystalline portion consists of m-HfO₂, with 40 nm crystallites, while t-HfO₂ accounts for less than 1%. Interestingly, no carbide phase, either HfC or SiC, is detected in the XRD pattern of M16.

In previous work, t-HfO₂ has been identified as the most prominent form of HfO₂ at lower temperatures, transitioning to m-HfO₂ at moderate temperatures, and then completely converting to HfC at the highest temperatures. ³⁵ A similar trend is observed for sol-gel derived hafnon, and the conversion can be hastened by the addition of a mineralizer. ³⁹ These observations are in excellent agreement with the phase changes observed in Figure 4.

3.6. Raman spectroscopy

The Raman spectra for S10, S16, M10, and M16 are shown in Figure 5. They are dominated by the so-called D- and G-bands related to carbon segregations in the material. ⁴⁰ Although the XRD patterns do not show prominent peaks associated with crystalline or turbostratic carbon, the Raman spectra indicate the presence of graphite-like carbon features. This holds even for the S16 sample, where the flat background of the XRD otherwise indicates the absence of ordered carbon in the sample (see subsection 3.5). Going into more detail, S10 exhibits the maxima of the D- and G-bands at 1348 and 1558 cm^-1, respectively, whereas for S16, they occur at 1368 and 1550 cm^-1. M10 shows maxima at 1330 and 1609 cm^-1, and M16 at 1338 and 1598 cm^-1. M samples exhibit more defined spectra, with sharper D- and G-bands, whereas the S samples display much broader lines with an apparent overlap of the bands. Moreover, the G bands in S10 and S16 appear 40-50 cm^-1 lower than in M10 and M16. Overall, this indicates less order within carbon segregations in S samples compared to M samples. ⁴⁰ The higher degree of order of carbon segregations in M samples is corroborated by the appearance of overtones in the Raman spectra. Both M10 and M16 display overtones of D- and G-peaks at 2930 cm^-1 and 2934 cm^-1, respectively, that are attributed to a D+G peak. Additionally, M16 displays peaks at 2675 cm^-1 and 3235 cm^-1 that are assigned as 2D and 2D′ modes, respectively. Interestingly, the Raman spectrum of M16 shows additional peaks at 213 cm^-1, 352 cm^-1, 401 cm^-1, 448 cm^-1, 985 cm^-1, and 1022 cm^-1, which can be attributed to hafnon, HfSiO₄, in the sample. ⁴¹

Figure 5.

Raman spectra of SHP-199 pyrolyzed to 1000 °C in nitrogen (S10) and to 1600 °C in argon (S16), and Raman spectra of SHP-199/TEOS solgel pyrolyzed to 1000 °C in nitrogen (M10) and to 1600 °C in argon (M16). The inset shows a magnification of the M16 spectrum between 200 and 1000 cm^-1 with peaks characteristic of hafnon.

Despite the presence of well-ordered carbon segregations in M16 indicated by Raman spectroscopy, no SiC or HfC was found in the XRD as a product of carbothermal reduction (see subsection 3.5). A kinetically hindered carbothermal decomposition of SiCO has been previously reported, with the process thought to be slowed by the presence of hafnium.^42,43

3.7. Scanning electron microscopy

Figure 6 shows SEM micrographs of fibers drawn from the SHP-199/TEOS mixture, including as-obtained samples and those annealed at 1000 and 1600 °C, labeled as F0, F10, and F16, respectively. The fiber widths, manually drawn with tweezers, range from 10 to 100 μm. In some cases, flat surfaces are visible, likely due to the sample resting on the Teflon sheet or the crucible during pyrolysis. As-obtained fibers and those pyrolyzed to 1000 °C exhibit smooth surface textures, with the latter sometimes showing tendrils after pyrolysis. Conversely, fibers exposed to 1600 °C display rough, porous surfaces, indicating the release of gaseous species and possibly significant material rearrangements. Due to the nature of the synthesis, the fiber samples are thought to be identical to the monolithic samples except for morphology. Thus, this “carbothermal reduction” of the oxide matrix is kinetically dampened and does not produce either SiC or HfC (see subsection 3.6).^42,43

Figure 6.

SHP-199/TEOS pulled fibers (a and b) unpyrolyzed (F0), (c and d) pyrolyzed to 1000 °C (F10), and (e and f) pyrolyzed to 1600 °C (F16).

4. Conclusions

The direct pyrolysis of SHP-199 under a controlled inert atmosphere presents a straightforward method for producing HfC/C ceramic powders. However, this process cannot produce a mechanically sound monolith due to the mass loss resulting from solvent evaporation and the extreme volume shrinkage that occurs during pyrolysis. Thermal crosslinking of SHP-199 yields ceramic particles but not monoliths, which is why chemical crosslinking was of interest.

SHP-199 self-catalyzes its reaction with TEOS, resembling acid-catalyzed sol-gel processing.^19,20 The sol-gel route is well-known for synthesis of HfB₂, HfO₂, HfC, HfSiO₄, and HfN.^20,44–47 The low pH of SHP-199 is attributed to the presence of dissolved HfOCl₂, possibly with additional Lewis acidic metal centers within the polymer. Indeed, SHP-199 reacted quickly with TEOS, forming hard monoliths. Their diameter shrank ∼30% upon drying, but they did not fracture.

The SHP-199/TEOS monoliths demonstrate significant thermal stability, with a near-constant mass above 600 °C and a ceramic yield of 92% at 1400 °C. Notably, the mass retention of the combined system of SHP-199/TEOS is larger than the average of SHP-199 and TEOS separately. This strongly suggests a synergistic effect through the incorporation of Hf into the network structure, as previously inferred.^19,20 Monolithic samples retained their structural integrity after pyrolysis to 1000 °C and to 1600 °C and could be easily ground into powder for XRD analysis. The SHP-199/TEOS mixture could even be drawn into fibers that retained their shape after pyrolysis.

The SHP-199/TEOS ceramic with a Hf:Si ratio of 1:10 develops a multiphase ceramic comprising HfO₂, SiO₂, and HfSiO₄, without the formation of SiC or HfC. Strong Raman signatures of carbon in the product at 1600 °C, together with the absence of crystalline carbide phases, indicate that reduction of the oxides is kinetically hindered, which is a known behavior for ceramics of similar composition.^42,43

The present work demonstrates that processing and shaping techniques enabled by the PDC route apply to the new polycarbohafnium polymer, SHP-199, when chemically crosslinked using a common siloxane precursor. Hence, the approach unlocks the full potential of PDC processing, including additive manufacturing. The process warrants further investigation of the sol-gel process, and particularly a detailed characterization of local bonding and its changes during pyrolysis. Future work will focus on developing routes to alter the relative amounts of Hf, Si, C, and O and to synthesize HfC/SiC compounds.