Oxidation of TaSi2-Containing ZrB2-SiC Ultra-High Temperature Materials

Hot pressed coupons of composition ZrB2 -20 v% SiC -5 v% TaSi2 and ZrB2 -20 v% SiC -20 v% TaSi2 were oxidized in stagnant air at temperatures of 1627 and 1927°C for one, five and ten 10-minute cycles. The oxidation reactions were characterized by weight change kinetics, x-ray diffraction, and SEM/EDS. Detailed WDS/microprobe quantitative analyses of the oxidation products were conducted for the ZrB2 -20 v% SiC -20 v% TaSi2 sample oxidized for five 10-minute cycles at 1927°C. Oxidation kinetics and product formation were compared to ZrB2 -20 v% SiC with no TaSi2 additions. It was found that the 20 v% TaSi2 composition exhibited improved oxidation resistance relative to the material with no TaSi2 additions at 1627°C. However, for exposures at 1927°C less oxidation resistance and extensive liquid phase formation were observed compared to the material with no TaSi2 additions. Attempts to limit the liquid phase formation by reducing the TaSi2 content to 5 v% were unsuccessful. In addition, the enhanced oxidation resistance at 1627°C due to 20 v% TaSi2 additions was not achieved at the 5 v% addition level. The observed oxidation product evolution is discussed in terms of thermodynamics and phase equilibria for the TaSi2-containing ZrB2-SiC material system. TaSi2-additions to ZrB2-SiC at any level are not recommended for ultra-high temperature (>1900°C) applications due to excessive liquid phase formation.


INTRODUCTION
ZrB2 and related materials are considered for ultra-high temperature applications such as leading edges of hypersonic vehicles due to their high melting points. However, application of these materials is limited due to their poor oxidation resistance, low toughness, and low thermal shock resistance [1]. This paper addresses methods to improve the oxidation resistance of ZrB2. ZrB2 oxidizes to form B2O3 and ZrO 2 . At temperatures between 450 and about 1100C, liquid phase boria is formed [2]. This liquid phase provides a somewhat protective liquid layer that slows oxygen transport to the underlying substrate. At temperatures greater than 1100C, the B2O3 boils off allowing rapid transport of oxygen through the zirconia scale to the underlying ZrB 2 substrate. Additions of 20 v% SiC to ZrB2 [3] provided improved oxidation resistance since the formation of a silica scale provides protection to much higher temperatures dependent on the amount of additive and oxidation condition [4]. However, at temperatures above the melting point of silica (T m = 1723C [5]), the liquid film can be lost to volatility and shear forces in high velocity applications [6]. Many studies have been conducted to add additional constituents to the ZrB2-SiC system in attempts to further increase the oxidation resistance. TaSi2 is one such constituent that has been considered to increase the oxidation resistance.
It has been demonstrated that Ta additions to ZrB 2-based materials showed improved oxidation resistance relative to the baseline materials at moderately high temperatures (T<1627C). Talmy et al. showed that two to twenty mol% TaB2 additions to ZrB2-SiC resulted in improved oxidation resistance at temperatures between 1200 and 1400C [7]. Similarly, Ta5Si3 and TaSi2 additions in the amounts of 8-31 v% showed improved oxidation resistance relative to ZrB 2 at temperatures between 1100 and 1400C in air for times of two to three hours [8,9]. In all cases, Talmy showed evidence for immiscible glass formation in the oxidation products and attributed the improved oxidation resistance to the steeply rising liquidus temperatures, increased viscosity, and decreased oxygen diffusivity [7] of the phase-separated glasses. Opila et al. [10] showed that ZrB 2 -20 v% SiC -20 v% TaSi2 resulted in a ten-fold reduction in oxide thickness relative to ZrB 2 -20 v% SiC after oxidation at 1627C for 100 minutes in stagnant air. This improvement was attributed to the presence of Ta rather than the increased Sicontent. The mechanism for improved oxidation behavior was not determined, however, it was proposed that this result was either due to formation of immiscible glass phases or Ta-doping of the ZrO 2 scale.
TaSi2 has a melting point of approximately 2200C [11,12] so the material itself has high temperature capability. The expected oxidation products, SiO 2 and Ta2O5 have melting points of 1723C [5] and 1872C (a-Ta2O5) [13] respectively. Upon oxidation at temperatures above 1872C, TaSi2 is therefore expected to form a liquid phase. For mixed ZrB2-SiC-TaSi2, the oxidation products are expected to be more complex. The formation of Ta2O5-xZrO2 phase V with x reportedly varying between 5 and 8 [14][15][16] is possible. The melting temperature of this phase is unknown, but has been reported to be >1870C [14]. Thus, this oxidation product is also likely to limit the temperature capability of TaSi2-containing ZrB2-SiC compositions. It was proposed [10] that additions of small amounts (<20v%) of TaSi2 to ZrB2 might result in oxidation to form Ta2O5 in solution with ZrO 2 , thus preventing formation of these low melting oxides. While the extent of Ta2O5 solubility is unknown, a phase diagram for the related Nb 2 O 5 -ZrO 2 system shows 5 to 10 mol% solubility of Nb2O5 in ZrO 2 [16].
The objectives of this work are to: 1) Extend the previous work with ZrB2 -20 v% SiC -20 v% TaSi2 [10] to the lower TaSi2 level of 5 v% for developing enhanced oxidation resistance at both 1627 and 1927C, and 2) Interpret the results of a detailed characterization of ZrB 2 -20 v% SiC -20 v% TaSi2 before and after oxidation using microprobe and quantitative WDS to understand liquid phase formation arising from TaSi 2 additions, the whereabouts of the Ta in the oxide scale, and possible steps to mitigate liquid-phase formation.

MATERIALS AND METHODOLOGY
ZrB2-SiC-TaSi2 coupons were prepared from powders described in Table 1. Processing history is summarized in Table 2. Several batches of the same composition were made. The powders were mixed, and then milled using Si 3N4 media in hexane in a Si3N4 mill for 24 h. The powders were vacuum hot pressed using a graphite die. Further details can be found in reference [17]. Sample coupons of 2.5 cm x 1.2 cm x 0.3 cm were machined from hot pressed plates. The coupons were ultrasonically cleaned in detergent (Micro-90, International Products Corporation), de-ionized water, acetone, and alcohol prior to oxidation. X-ray diffraction and Energy Dispersive Spectroscopy of as-machined and cleaned coupons indicated ZS20TS contained ZrB2, SiC, TaSi2 and a minor amount of a (Zr,Ta) boride solution. Initial surface areas and sample weights were recorded. The samples were oxidized at 1627C in stagnant air in a bottom loading CM Inc. MoSi2 element box furnace. The samples were oxidized at 1927C in stagnant air in a bottom loading DelTech, Inc. zirconia element furnace. Three coupons were loaded into a slotted calcia-stabilized zirconia firebrick setter (98.7% purity). Two lines of contact existed between the setter and sample. The coupons were oxidized for ten minute cycles. On heat-up to 1627C, the furnace temperature reached approximately 100 degrees less than the set temperature in 1 minute. Heating to the set temperature was complete 5 minutes into the heating cycle. The DelTech furnace temperature fell by approximately 75 degrees during the loading process. Samples were removed from the furnace and cooled in ambient conditions for 10 minutes between cycles. A coupon was removed after one cycle, five cycles, and ten cycles. A maximum oxidation time of 100 minutes was achieved. Post-oxidation weight changes were recorded where possible. Some coupons stuck to the zirconia setter due to extensive glass formation.
Optical macrographs were taken of all coupons after oxidation. The sample surfaces were analyzed by X-Ray Diffraction (XRD), Field Emission Scanning Electron Microscopy (FESEM) Hitachi S-4700, and Energy Dispersive Spectroscopy (EDS) to determine phases present before and after oxidation as well as oxide microstructures. Several materials were also examined by FESEM and EDS in cross-section prepared by non-aqueous cutting and polishing procedures to preserve boria-containing phases. The ZSTS20 coupon previously oxidized for five cycles (50 minutes total exposure) at 1927C was also characterized extensively by Wavelength Dispersive Spectroscopy (WDS) microprobe (JEOL 8200). Each reported phase composition determined by microprobe is the average of five measurements.

Oxidation Kinetics
The weight change results for ZS, ZSTS5, and ZSTS20 obtained at 1627C in stagnant air are shown in Fig. ( 1). Macrophotos of the materials after oxidation at 1627C are shown in Fig. (2). Surface SEM images are shown in Fig. (3) for all three compositions while cross sections of the oxide scale are shown in Fig. ( 4). XRD results for all the oxidized materials were similar and showed only ZrO 2 (monoclinic) with minor amounts of ZrO 2 (cubic). No additional phases were observed. Macrographs of ZS, ZSTS5, and ZSTS20 after oxidation at 1927C are shown in Fig. ( 5). The results for ZS and ZSTS20 were previously published [10] but are shown here for comparison. The weight change results and macrophotos for the sample exposed at 1627C show that ZrB 2 -20 v% SiC -20 v% TaSi2 shows the best oxidation behavior. The addition of 5 v% TaSi2 to ZrB2 -20 v% SiC is insufficient to lead to phase separation in the glass, whereas 20 v% TaSi2 additions result in immiscible glass formation as shown in Fig. ( 3). The oxidation thickness is reduced approximately ten times for 20 v% TaSi2 relative to the material with no additives as shown in Fig. (4) [10]. The oxide cross-section for the sample with 5 v% TaSi2 is not shown since the scale was bubbled and varied irregularly in thickness between 4 and 100 µm for a sampling of 20 locations. At 1927C, the material with no TaSi2 exhibits the best oxidation behavior since liquid phase formation is limited. Extensive liquid formation is observed for both the 5 and 20 v% TaSi 2 addition levels.

Microstructural/Compositional Characterization of ZSTS20 Before and After 1927C Exposure in Stagnant Air for 50 Minutes
In previous work [10], EDS analysis of the ZSTS oxidation products was complicated by the fact that the Ta Ma peak (1.709 keV) has almost complete overlap with the Si Ka peak (1.739 keV). Ta identification was made primarily by observing the low intensity Ta La peak at 8.145 keV. In addition, phases with high Si content had much darker contrast than those with high Ta content. Nevertheless, it was uncertain where Ta was incorporated into the oxidation products. In addition, the composition of the low melting phase was unknown, preventing an understanding of possible ways to improve the high temperature oxidation resistance of ZSTS materials. In this paper, microprobe wavelength dispersive spectroscopy (WDS) was used to more completely characterize the oxidation products. The following microprobe results are presented beginning with the substrate and proceeding outward through the layers of the oxide products formed to identify the key reactions occurring during the oxidation exposure at 1927C. Results are summarized in Table 3.

Substrate Before Oxidation
After hot pressing, in addition to the starting phases ZrB 2 , SiC, and TaSi2, an additional phase zirconium rich boride phase (Zr,Ta)B x was observed as shown in Fig. ( 6). This solution phase has been previously observed [8,10] and is reasonable since both ZrB2 and TaB2 have a hexagonal crystal structure. XRD results showed that the (100), (101), and (111) peaks for ZrB2 all exhibited shoulders displaced by 0.1 to 0.5 degrees 2theta higher indicating reduction of the lattice parameters of the hexagonal ZrB2 structure with the introduction of TaB2

Substrate After Oxidation
After oxidation, the TaSi2 is no longer found in the substrate throughout the entire thickness as shown in Fig.  (6). Fig. (6b) was obtained near the center of the coupon. The Ta is now present as a carbide (Ta,Zr)C x . The amount of ZrB2 and SiC have both decreased relative to the starting material whereas the amount of (Zr,Ta)B x phase has increased. Porosity is found throughout the substrate.  Figs. (7, 8) also show that Ta and B are present at low levels throughout much of the oxide scale. Fig. (2). Macrophotos of coupons after oxidation at 1627C in stagnant air. , and ZSTS20 (bottom) after oxidation at 1627C for 100 min in stagnant air. Details of area in white box on left are shown in higher magnification on the right. ZrO 2 and glass (g) phases are labeled in the right hand micrographs. Note that only the ZSTS20 shows evidence of immiscible glass formation: g1 is silica rich while g2 contains Al and Mg impurities. Fig. (4). Secondary electron images showing cross-sections of the ZS and ZSTS20 after oxidation at 1627C for 100 min in stagnant air [10]. White arrows show the oxide thickness.

Oxidation of TaSi2-Containing ZrB2-SiC Ultra -High Temperature Materials
The Open Aerospace Engineering Journal, 2010, Volume 3 45 Fig. (5). Macrographs of ZS, ZSTS5, and ZSTS20 after oxidation at 1927C for 10 minute cycles in stagnant air.  Fig. (6). Backscattered electron image of ZSTS20 substrate before and after oxidation for 50 min at 1927C in stagnant air. Four phases are present in the hot-pressed material (from brightest to darkest contrast): TaSi 2 , (Zr,Ta)B 1 .5 , ZrB2, and SiC. The exposed sample contains (from brightest to darkest contrast) (Zr,Ta)C x , (Zr,Ta)B x , ZrB 2 , SiC and porosity. Fig. (7). Backscattered electron image and corresponding WDS dot maps for the substrate/oxide interface of ZSTS20 after oxidation at 1927C for five 10-minute cycles in stagnant air. The red line corresponds to the approximate location of the ZrO 2 /ZrB 2 interface as discussed in the text.

Oxidation of TaSi2-Containing ZrB2-SiC Ultra -High Temperature Materials
The Open Aerospace Engineering Journal, 2010, Volume 3 47 Fig. (8). Backscattered electron image and corresponding WDS dot maps for the substrate/oxide interface of ZSTS20 after oxidation at 1927C for five 10-minute cycles in stagnant air. The red oval focuses on an area containing ZrO2, SiO2, and micron-sized, Ta-rich oxyboride phases.

Middle Portion of Attached Scale
reaction with the surrounding amorphous phase, as shown in The phases present in this portion of the scale are large Fig. ( 10).
columnar ZrO 2 grains in SiO 2 , as shown in Fig. (9). In Outer Surface, Detached Scale addition to the finely dispersed Ta oxyborides, a band of The phases observed on the outer detached scale and larger grains of Ta(C,B) is observed amongst the other oxide shown in Fig. ( 11) include ZrO 2 found both as large grains phases. Finally, the ZrO 2 grains appear to show surface and as dendrites, SiO x with small amounts of dissolved Ta and Zr and x=3, and a (Zr,Ta) oxide with a cation/anion stoichiometry of 2:5 as would be expected for Ta 2 O 5 , however, two thirds of the cations are Zr. Fig. (10). Backscattered electron image of the middle of the oxide scale formed on ZSTS20 after oxidation at 1927C for five 10minute cycles in stagnant air. Same area as Fig. ( 9). The red arrows highlight the bright contrast surface layer found on zirconia suggesting Ta-enrichment. Fig. (11). Backscattered electron image of zirconia dendrites observed on the outer detached oxide scale formed on ZSTS20 after oxidation at 1927C for five 10-minute cycles in stagnant air

Optimization of TaSi2 Content
Twenty vol % addition of TaSi 2 to ZS promoted improved oxidation behavior at 1627C. Both the weight change (Fig. 1) and the oxide thickness (Fig. 4) confirm this result. The appearance of the oxide surface indicates glass immiscibility, as shown in Fig. (3). Talmy et al [7][8][9] attribute similar observations of glass immiscibility to Ta additions to the system. EDS analysis [10] of the immiscible phases in Fig. (3) were inconclusive. The darkest phase showed EDS consistent with SiO 2 . The medium contrast phase showed results consistent with Al and Mg impurities concentrated in a SiO 2 -rich phase. In addition, a trace peak attributed to Ta and or Zr was also observed [10]. The bright phase had EDS results consistent with ZrO 2 and a slight trace of Ta or Si. The trace of Ta or Si could be oxide doping or a sampling volume effect. The sample was unavailable for further microprobe analysis to better characterize the composition of the surface oxide phases since it had been cross-sectioned, mounted and polished for oxide thickness measurements. Thus, while glass immiscibility has been observed, the mechanism for improved oxidation resistance due to Ta additions is still not certain.
The oxidation weight gain of ZSTS5 was similar to ZS (no TaSi2 additions) at 1627C. After only ten minutes of oxidation the ZSTS5 scale showed significant amounts of the pink scale observed previously for the ZSTS20 samples (Fig.  2). However, this pink oxide was not able to establish a protective scale at longer times and the glassy appearance of the ZSTS5 after ten cycles was similar in appearance to the ZS material. SEM images of the surface oxide (Fig. 3) show more uniform dispersion of the ZrO 2 in a continuous silicarich phase relative to ZS, but glass immiscibility of the oxidized ZSTS20 was not observed.
At 1927C, it was hoped that the reduction in TaSi2 from 20 to 5 vol % would eliminate the excess liquid formation attributed to Ta-containing low melting phases, possibly by incorporation of the Ta into the ZrO 2 . While five vol % TaSi2 additions to ZS did reduce the amount of liquid phase relative to ZSTS20, there was still formation of significant amounts of liquid phase indicated by sample distortion, as shown in Fig. (5).
Not only did the reduction in TaSi2 to 5 vol % fail to provide additional oxidation protection to ZS at 1627C, the reduction was not enough to eliminate liquid phase formation at 1927C. Thus, optimization of TaSi 2 content is not possible to allow improved oxidation behavior at both 1627 and 1927C.

Substrate Phases: As Hot-Pressed
The composition of the hot-pressed material is summarized in Table 3 while images of the material can be seen in Fig. (6). The formation of (Zr,Ta)B 1 . 5 indicates that TaSi2 is unstable with respect to ZrB 2 . The stoichiometry of the new phase (Zr,Ta)B 1 . 5 suggests that Ta is added into the ZrB2 structure without additional boron, resulting in a material with a lower relative boron content. XRD results indicated that the lattice parameters of the (Zr,Ta)B 1 . 5 phase were slightly smaller. These results are consistent with substitution of the smaller Ta atom on the Zr site and/or incorporation of B vacancies in the (Zr,Ta)-diboride phase. Table 4 shows the relative thermodynamic stability of a series of compounds important for this material system [5]. It can be seen that silicides are the least stable with Ta compounds less stable than the corresponding zirconiumbased compounds. These trends are consistent with the observed conversion of TaSi2 to (Zr,Ta)B 1 . 5 during hotpressing.

Substrate Phases: After Oxidation at 1927C
After oxidation, no TaSi2 was observed in the matrix. In addition to ZrB2, SiC, and (Zr,Ta)B 1 . 5 , a new phase (Ta,Zr)C 1 . 4 is formed. The amount of the (Zr,Ta)B 1 . 4 has increased, the tantalum content in this boride phase has increased and the resulting B stoichiometry has also decreased slightly consistent with more Ta addition to the structure. This suggests continued reaction between TaSi 2 and ZrB2. A possible explanation for additional loss of TaSi 2 and the appearance of both voids and the tantalum-rich carbide could be a reaction of the following form: TaSi2 + SiC + 1.5 O2(g) = TaC + SiO(g), ^G rxn = -79kJ/mol (1) with the free energy data obtained from the FACT database [5]. This reaction is a type of active oxidation in which both TaSi2 and SiC are consumed forming TaC as the silicon leaves the system as a vapor species. TaC is more stable than SiC (AGf 1927C= -140kJ/mol vs -41kJ/mol) [5] making this a likely reaction route.

Oxide Phases Formed: ZrO 2
ZrO2 is observed as large grains tending toward a columnar structure throughout the scale. The amount of Ta dissolved in the ZrO2 structure increased toward the outer surface of the oxide, but remained low, 1.3 at% or less. This suggests that incorporation of Ta into the ZrO 2 is limited for in situ thermally grown oxides. Higher contents of Ta2O5 have been incorporated in zirconia by traditional powder or sol gel ceramic processing techniques [18][19][20]. There are two consequences of this limited solubility of Ta in thermally grown zirconia. First, oxygen ion transport through the zirconia will not be reduced substantially due to oxygen vacancy concentration reduction associated with pentavalent Ta occupying quadravalent zirconium sites. Ionic conductivity and ion diffusivities are related through the Nernst Einstein equation [21]. Thus, oxygen ionic conductivity measurements as a function of dopant concentration can give information about the changes in oxygen diffusivity. While data for the dependence of ionic conductivity in Ta2O5-doped ZrO2 are not available in the literature, data obtained for ZrO2 co-doped with Y2O3 and Ta2O5 show less than an order of magnitude decrease in ionic conductivity for 2mol% Ta2O5 additions [22,23]. Thus, a Ta2O5 content of about 1% in ZrO2 is not likely to substantially reduce the oxidation rates of ZS materials. Second, since solubility of Ta in thermally grown ZrO 2 is limited, any excess Ta is likely to form low melting oxide phases which are detrimental to oxidation resistance at temperatures exceeding their melting point. The formation of zirconia dendrites shown in Fig. ( 11) is evidence of zirconia incorporation in and precipitation from these low melting phases.
There is additional evidence of zirconia reaction with Ta in the bright contrasting phases formed around each zirconia grain as seen in Fig. ( 10). Microprobe was not conducted on these surface reaction phases, so the composition is unknown. One possibility is that this phase is the V-phase, Ta2O5-xZrO2 (x = 5 to 8). An additional zirconium tantalate phase was observed phase separated in the silica in the outer detached layer of the oxide scale, though this phase has the 2:5 stoichiometry of Ta2O5.

Oxide Phases Formed: SiO2
SiO2 is observed as the continuous phase that surrounds the columnar zirconia grains as well as any Ta-containing phases. Microprobe results indicate the expected SiO 2 oxygen stoichiometry near the oxide/substrate interface. The oxygen stoichiometry deviates moving outward through the oxide scale while correspondingly more Ta is incorporated in the silica. A maximum of 2.5 at% Ta is observed in the silica. At the outer surface, small spheroids of phaseseparated zirconium tantalate are observed in the silica as previously mentioned and shown in Fig. ( 11). The stoichiometric and compositional trends indicate that "pure" silica is formed at the oxide/matrix interface and more complex silicate melts are formed in the outer portion of the oxide as the scale thickens.

Oxide Phases Formed: Ta-Oxides
Ta2O5, the thermodynamically stable oxide of Ta is not observed anywhere in the oxide scale. Near the substrate/oxide interface, small submicron precipitates of Taoxyboride are dispersed in both the ZrO 2 and the SiO2 (Fig.  8). The observation that these particles are dispersed in the zirconia supports the limited solubility of Ta in the ZrO 2 .
Proceeding outward through the scale, larger 10 µm sized particles of Ta(C,B) grains are observed in a layered structure (bright phase in Fig. 9). This layer corresponds to the highest concentration of Ta observed in the oxide scale as shown in Fig. (7). The existence of these large non-oxide grains is surprising since they are surrounded by oxide. However, based on the Ellingham diagram calculated using the Fact database [5] shown in Fig. ( 12), tantalum oxides are the least thermodynamically stable, so that for a limited amount of oxygen, zirconia is expected to form preferentially. A similar phenomenon was observed in which BN remained unoxidized when surrounded by SiO2 due to the relative thermodynamic stabilities of the boria and silica [24].
As discussed above, two zirconium tantalate phases are observed in the outer portions of the scale while more Ta is observed dissolved in both the zirconia and the silica. These more complex tantalates are found in the outer portions of the scale suggesting further Ta reactions occur at longer exposure times. Fig. (12). Ellingham diagram for oxides in the ZSTS system calculated with FACT53 data and FactSage [5].

CONCLUSIONS
Previous results [10] have shown that additions of 20 v% TaSi2 to ZrB2-20 v% SiC resulted in markedly improved oxidation resistance at 1627C in air for 100 minutes. This improved oxidation resistance is likely attributed to Tainduced glass phase immiscibility and resulting lower transport rates of oxygen through the more viscous glasses. However, it has now been demonstrated that addition of the smaller amount of 5 v% TaSi 2 to ZrB2-20 v% SiC is insufficient to result in glass immiscibility and improved oxidation resistance. TaSi2 additions to ZrB2-20 v% SiC in both 5 and 20 v% amounts resulted in extensive liquid phase formation when exposed for 50 minutes in stagnant air at 1927C. Extensive microprobe characterization of the 20 v% TaSi2-containing sample, 1927C exposure, has shown that TaSi2 additions are not suitable for applications at 1927C for a number of reasons. TaSi 2 is unstable in the matrix with respect to ZrB2. During oxidation, TaSi2 in the presence of SiC appears to undergo active oxidation forming TaC and SiO(g) leaving voids in the substrate. During oxidation, Ta is incorporated in the zirconia in amounts of 1.3 at% or less, suggesting that in-situ doping of the zirconia to decrease oxygen transport rates through the zirconia is limited. During oxidation, Ta also forms oxyboride, silicate, and zirconate phases which result in extensive amounts of liquid phase formation and poor oxidation resistance. Ta-additions to ZrB2-20 v% SiC are not recommended for ultra-high temperature (>1900C) applications.