urna Am,Cem.Sox,8615830-37(2003) Thermal Response and Oxidation of TyrannotM-Fiber-Reinforced Si-Ti-C-O Matrix Composites for a Thermal Protection System in High-Enthalpy Dissociated Air Toshio Ogasawara, Takashi Ishikawa, and Takashi Matsuzaki National Aerospace Laboratory of Japan, Mitaka, Tokyo, 181-0015, Japan The thermal response and oxidation of Tyranno no Lox-M Il. Experimental Procedure fiber-reinforced Si-Ti-C-O matrix composites in high-enthalpy dissociated air was investigated in an are jet facility (an are ()Materials wind tunnel). The maximum surface temperature reached The composites under investigation contained TyrannoTM 1310-1670C. Catalytic recombination of oxygen and nitrogen Lox-M fibers woven into an orthogonal 3-D fabric (Z-PlusTM on the composite surface under dissociated air was not signif- SCL, Shikibo, Ltd, Shiga, Japan), or a stitched 2-D plain weave icant. Surface recession was insignificant below 1600.C sur fabric(Z-PlusTM Netshape). The composite constituents are face temperatures and above 5 kPa of oxygen partial pressure summarized in Table I. The optical micrographs in Figs. 1(a) at the stagnation point. Passive-to-active oxidation transition nd(b)illustrate the fiber architecture of the present compos- of the composite agreed with Balat's theory for monolithic ites. Polytitanocarbosilane was used as the matrix precursor silicon carbide. a glass sealant prevented active oxidation of with eight impregnation and pyrolysis cycles being required to the composite for short-time exposures. chieve satisfactory densification. Glass-sealed and unsealed composites were prepared to investigate the effect of glass sealant. The glass-sealed composite specimens were subject to I. Introduction an antioxidation procedure in which a SiO2-Na, O-based glas was repeatedly impregnated into the composite open porosity C ONTINUOUS-CERAMIC- FIBER-REINFORCED ceramIc-matnix com posites exhibit excellent damage tolerance required for aero- chosen because of low viscosity above 1000C. Thermal space applications. The National Aerospace Laboratory of Japan properties are listed in Table IL. Thermal diffusivity and pecific heat were measured by a laser flash method and a DSC te, for use as a structural material in a future reentry space method, respectively. Emissivity of sealed and unsealed com- vehicle, The composite is referred to as NUSK-CMC from the posites was measured using a blackbody furnace(Emissiom- initials of the collaborators. The composite has been developed eter, ThermogageTM, Vatell Corp, VA)at 900 and 1100C in over several years and the present incarnation contains Tyrannoonp N2 gas flow. For direct comparison, commercial graphite Lox-M fibers that have been surface-modified in a carbon mon and a fiber -be oxide atmosphere to produce a 10 nm surface layer rich in Sio matrix composite (Tyrannohex M, Ube Industries, Ltd, Ube, surrounding an inner 40 nm layer rich in carbon. The rationale apan) were tested under similar conditions. behind the surface modification is to promote crack deflection and fiber/matrix debonding at the carbon-rich layer within the fiber (2) Arc Jet Testing ites, orthogonal 3-D woven fabric or stitched two-dimensional This test series was conducted in a 750 kW arc wind tunnel (2-D)fabric preform was used for reinforcement. Matrix densifi- (AWT: Sumitomo Heavy Industries, Ltd, Tokyo, Japan). I cation was achieved through the repeated impregnation and pyrol. Figure 2(a) shows a schematic drawing of a plasma generator of sis of a polymer precursor. The composites exhibit excellent the AWT. In the plasma generator, air is heated by electrical discharge in a water-cooled constrictor arc column. Then, the To apply the composites to the thermal protection system(TPS) gas (air) is supersonically expanded through a converging- of a future reusable space vehicle, thermal response and oxidation under atmospheric reentry conditions should be evaluated in addition to the mechanical properties. An are jet facility(are win Table L. Composite Constituents of N MC tunnel) is often used to simulate the aerodynamic heating condi- tions experienced by space vehicles during reentry. In a high- Tyranno M Lox-M(Si-Ti-C-O) enthalpy convective environment, the catalytic recombination (Ube Industries Ltd )(Si, 54%: Ti, 2%: C reaction of dissociated oxygen and nitrogen on the material 32%;0.12%(mass%) surfaces affects the thermal response of the materials. b-In this Fabric Orthogonal 3-D woven fabric (Z-plusTM SCL study, the thermal response and oxidation behavior of various Shikibo Ltd ) 20 vol%, 20 vol%, and 2 vol% NUSK-CMCs were investigated using an are jet facility. in the x, y, and z directions Stitched 2-D plain fabric (Z-plusTM Netshape. Shikibo Ltd. ) 22 vol%, 22 vol%, and 1 vol% in the x, y, and z directions S6 Treatment(10 nm SiO, -rich layer surrounding R. J. Kerans--contributing editor an inner 40 nm carbon-rich layer at the fiber surface, Ube Industries Ltd Matrix Si-Ti-C-O derived from polytitanocarbosilane 8 PIP cycles(Ube Industries Ltd) eoManuscript No. 187186 Received January 28, 2002: approved December 16. Glass sealant SiO -Na, O-based glass sealant(Kawasaki Heavy stne fember, American Ceramic Society. 830
May 2003 Thermal Response and Oxidation of Tyrannoan-Fiber-Reinforced Si-Ti-C-O Matrix Composites Z fibers Stitching fibers Z ∠mm 2mm Fig 1. Optical microphotographs of Tyrannot Lox-M fiber reinforced Si-Ti-C-O matrix composites(NUSK-CMC): (a)orthogonal 3-D woven fabric (Z-Plust SCL),(b)stitched 2-D plain woven fabric (Z-Plus t Netshape) diverging nozzle into an evacuated test chamber. In the test interlacing loops shown in Fig. I were not present in the final chamber, a sample assembly was mounted onto a mechanical amples. The sample was mounted in a reusable Sic-coated swing arm, and it was positioned 100 mm from a nozzle exit as aphite holder by using three alumina pins with Ni alloy shown in Fig. 2(b). A sample assembly was exposed to the prings. Figure 4 shows a typical test in progress; a shock wave dissociated gas stream, and taken out immediately after the around the sample can be seen. Cold-wall heat flux 4w was desired exposure time measured using a Gardon-type calorimeter (Thermogageo, 3. Samples were 20 mm in diameter and 5 mm thick with a 30 two-color optical pyrometer(IR-CQS21C, Chino, Japan)was taper angle. The tested geometry was that appropriate for the used to determine the surface temperature of the samples. Type leading edge of a blunt body during reentry. The sample R thermocouples were used to measure the internal thermal surfaces were also ground to a flat finish such that the response of the samples during arc jet testing. It is known that Table ll. Thermal Properties of NUSK-CMC, Graphite(IG-110), and Tyrannohex NUSK-CMC G-110 Tyranno Fabric Orthogonal 3-D Orthogonal 3-D Plain weave 2-D Satin weave 2-D Glass sealant Unsealed Sealed Bulk density(g/cm 252.21 1.772.4 Specific heat(/g K) 0.70 0.71 0.320.65 Thermal conductivity (W/mk) 1.74 1062.74 3.03 193 Emissivity 9000.90 0.95 1100 0.89 0.94 0.91 0.900.88
832 Journal of the American Ceramic Sociery-Ogasawara et al Vo.86.No.5 Converger Modular Packs of Diverger Section Anode Cathode Supply (750kW) Plasma generator Test chamber (a) Fig. 2. Schematic drawing of an arc jet facility (arc wind tunnel)used for the experiment (NASDA/NAL 750 kW AWT):(a) segmented constrictor-type plasma generator, (b) plasma generator and test chamber configuration some oxides selectively emit at wavelengths used for optical Ill. Results and Discussion pyrometry at high temperatures. Therefore, the surface temper ature of the composite placed in a blackbody furnace wa (1) Thermal Response easured using both a pyrometer and thermocouples. As a The maximum surface temperatures during arc jet testing are result, it was confirmed that the experimental error of the plotted in Fig. 5 as a function of cold-wall heat flux measured with pyrometer was insignificant up to 1500C a calorimeter, The cold-wall heat fluxes reflected a finite catalytic The electrical power input and the air mass flow rate were surface, because the surface of the calorimeter was made of changed for each arc jet test condition. Typical arc jet test constantan( Cu-Ni alloy). Therefore, it should be noticed that the old-wall heat fluxes are not actual values for the sample surfaces conditions are summarized in Table Ill. Cold-wall heat fluxes because of differences in the catalytic efficiency between the measured with the calorimeter ranged form 0.9 to 2. 1 MW/m and stagnation pressures varied from 13. 1 to 30.7 kPa. The the composites reached 1310-1670C, which was much lower gas enthalpy was estimated by energy balance of the are jet than that of graphite. The difference in observed surface temper- facility, which ranged form 13. 4 to 20.3 MJ/kg. The test atures between glass-sealed and unsealed composites was not chamber pressure before testing was-180-240 Pa Exposure significant. time was 300 s for each sample, which was chosen because Numerical analyses based on the one-dimensional(1-D) finite this is a severe heat flux time for a typical reentry space difference method(FDM) were conducted to evaluate the surface hicle temperature during arc jet testing. In general. the heat transfer After arc jet testing, the mass loss and the surface recession of equation is as follows: each sample were measured. The surface and the cross section of each sample were observed using a scanning electron microscope (SEM, S-4700, Hitachi, Japan). X-ray diffraction(XRD) analysis 0()-a: was conducted using a CuKa source with an X-ray diffractometer (RINT2500, Rigaku, Japan) to investigate the chemical composi where, p, Cn, and K are the density, the specific heat, and the tion of the composite surface. thermal conductivity, respectively, Based on the control volume Ceramic Insulator 5.0 44 号 Graphite holder umina pin Ni alloy Spring Fig. 3. Geometry and dimensions of sample and sample holder assembly used for are jet testing
May 2003 Thermal Response and Oxidation of Tyrannoa-Fiber-Reinforced Si-Ti-C-O Matrix Composites Sample(中20×t5 Radiation equilibrium 2 2000 FCalc.(NUSK-CMC) 1500 ·· 100 aled 3-D sealed 2-D Nozzle Sample assembly 0.5 1.0 2.0 Fig. 4. Typical test in progress. The shock wave around the sample can Measured cold wall heat flux(MW/m tween calculated and experimental results during are jet testing as a function of cold-wall heat chnique the following finite difference equations are obtained: 2 Gardon-type calorimeter. The nominal exposure time s Predicted curves were calculated using the 1-D finite difference aT=ar-T-+ ak+Tk-1+b (2a) (2b) potentials across the boundary layer for the hot and cold walls as a+1-(6x)k ch-hewi pC (2d) where h, hew, and hhw are the gas enthalpies at the boundary layer dge, at a cold wall, and at the surface temperature of the test b=apT?+Q (2e) model, respect/vely. If the gas composition at a hot wall is mainly air, the gas enthalpy at the wall can be obtained by integrating the at=ak-1+ ar+ta specific heat of air from a reference temperature to the hot-wall temperature, where k is the number of divided element air is the distance from an adjacent element, and Q is the heat generation. The boundary h CadT condition at the front surface of the sample can be illustrated by writing a general surface energy balance equation 9=ghw+OET+ r,(h,-hs where cmo (/(g K ))= Co DoT, with Co = 0.979 J/(g K) and Do-15X 10 J/(g K), is the specific heat of air, and Tw and Tref are the hot-wall temperature and the reference temperature where ghw is the hot-wall heat flux at the stagnation point, or is the (300 K), respective me 2-D effect on the heat transfer, a 2-D Stefan-Boltzmann constant, E is the emissivity, Tw is the surface mperature, i, is the mass flux from the sample, hw is the total enthalpy of the vapor species, and hws is the enthalpy of the numerical analysis was also performed using a commercial finite material lement method (FEM) code, ABAQUS 5.8(Hibbitt, Karlsson, For the numerical analyses, the radiation boundary condition and Sorensen, Inc, Pawtucket, RI), Figure 6(a) shows the temper- was given for both front and backside surfaces. Mass flux from the ature distribution of a sample and a sample holder under the cold-wall heat flux of 1 MW/m", which was calculated by 2-D temperature difference on the heat transfer (namely, the"hot-wall analysis results is shown in Fig. 6(b). The 1-D FDM numerical ffect")from the cold-wall heat flux measured with the calorim- results are in good agreement with the 2-D FEM results eter. This correction factor is a function of the ratio of the enthal Numerical results based on 1-D FDM are superimposed on Fig 5: the estimated curve for graphite agrees with the experimental results, On the other hand. the estimated curve for NUSK-CMC is much higher than the experimental results. This is due to the Table Ill. Typical Are Jet Test Condition composite surfaces. -9 From the numerical results, actual heat Air mass flow Gas enthalpy Cold-wall heat Impact pressure Current(A) fluxes for the composites were estimated as shown in Fig. 7. It was predicted that the actual heat fluxes would be about 52% of the 13.4 cold-wall heat fluxes measured with the calorimeter. It was 12.9 1.0 15 reported that SiO, surfaces produced very low catalytic efficien- 16.9 cies compared with metallic materials such as Ni, Pt, Cr, and Cr: 4.5 the heat transfer rate to SiO, was reduced to a minimum value of 18 1.8 only one-third of the value obtained on relatively active NI surfaces,which agrees with the experimental results in this study When the surface temperature exceeded 1650C for the un- sealed composite, the surface temperature increased rapidly above
Journal of the American Ceramic Society--Ogasawara et aL. VoL. 86. No. 5 Surface Heat flux (IMW/m2) 2500 Back side Surface temperature 2000 Te emperature 1500 h11 +1,700+03 @@自@@@ 51000 +2.500e+03 1.300+0 8 O 2-D FEM(ABAQUS) +:00+0 Holder E500● ●2-DHM( ABAQUS) I-D FDM +5.000e+0 200 CL Time(sec) Numerical analysis results based on the 1-D finite difference method(FDM) and the 2-D finite element method(2-D FEM) under a hot-wall heat MW/m2: (a) temperature distribution calculated by 2-D FEM program, ABAQUS 5.8:(b) comparison of temperature profiles between 1-D FDM FEM at the center of the sample surface. 1900 C as shown in Fig. 8. In these samples, significant mass loss Furthermore, the hot air through these pores might hay and surface recession were also observed, which was due to active the internal temperature of the samples. On the other oxidation. However, this was not observed in the glass-sealed pposed that the effect of z-fibers on the thermal through thickness was insignificant because of the low volume The effect of fiber configuration on the thermal response was fraction of z-fibers(2 vol% for the 3-D composite also investigated under the same are jet testing conditions. Figures 9(a) and (b)show the temperature profile of a 3-D woven composite and a stitched 2-D plain weave composite. The surface 2) Mass Loss and Surface Recession ind internal temperatures in the 3-D composite were higher than A summary of the mean mass loss rate of the materials tested in those in the 2-D composite, which was caused by differences in the the arc jet is shown Fig. I l as a function of cold-wall heat fluxes urface roughness. The 3-D composite had large pores(pocket The mean mass loss rate is defined by gion) derived from the orthogonal 3-D woven architecture as thmean=(mo-m)Tr shown in Fi The aerodynamic heating conditions were severe around the pores because of turbulent flow; therefore, the surface temperature was higher compared with a stitched 2-D composite 1900C (over range 2000 1650cT 0.8 1500 0.6 Q52 1000 urface temperature 500 Internal temperature (3.5mm) 0.0 0 50 1.52.0 100 Measured heat flux(MW/m") Time(sec) ig. 8. Temperature profiles of the unsealed orthogonal 3-D c d cold-wall heat fluxes versus measured heat fluxes for under high heat flux condition (measured cold-wall heat flux 2.06 the glass composite(stitched 2-D plain fabric version). The estimate MW/m). The surface temperature exceeded the upper limited values heat fluxes are calculated by I-D FDM, and they are normalized by (1900 C)of the pyrometer, and the thermocouples embedded in the sample