G E. Youngblood et al. /Composites Science and Technology 62(2002)1127-1139 the direct and gaseous conduction components might be 烈小点 comparable. The product components depend on the magnitudes of fa and fg, and should be very dependent upon the surface roughness, gap dimensions, tempera- ture and environmental conditions. Separately deter- mInin ng fa and fg would be dificult because these quantities likely vary with changing environmental con- ditions in a non-reproducible manner. However, hg 0 改 nder vacuum conditions. Therefore, the product terms fahd and ghg, can at least be estimated by performing experiments on the same sample in both vacuum and a known atmosphere. To a good approximation, the interfacial conductance measured in vacuum should represent only fahd term. Then to determine the component ghg, the total interfacial conductance h is Fig. 5. A transverse view by SEM of a polished surface for a uniaxial measured in the gaseous phase of interest, and fahd is Hi-Nicalon M/PIP-SiC (amorphous Ceraset M) composite after two subtracted from it crystallization temperature for Ceraset (a1600C) 4. The effective transverse thermal conductivity of However, the hTT conditions (30 min at 1100oC in example SiC/SiC composites argon) were sufficient to form a fairly stable, Si-C-O matrix structure and to expel most of the gaseous com 4.1. Uniaxial Hi-Nicalon/ PIP-SiC composite ponents. The infiltration and HTT cycle was repeated one time to increase the composite density. After the To experimentally assess the analytical solutions for second HTT, a rod was core-drilled from the center of the H-J model, the transverse Kefr for a particular uni- the plug, and several disks for thermal diffusivity mea- axial SiCe/ Sic composite was measured in three diffe surements were cut from the rod with fiber alignment ent atmospheres at a fixed temperature. Using Keff, Kf, either parallel or perpendicular to the flat disk surfaces Km, a and Ve values determined experimentally, the The individual disks were 62-mm diameter by 2-mm overall interfacial conductance, h, was calculated from thick. Normally, for laser flash thermal diffusivity mea- estimated by the method described above and compared however later tests requiring packaging in small y Eq.() for each atmosphere. Then fghg-values were surements one prefers a diameter-to-thickness ratio >5, to calculated values based on known values of gas ther- tion capsules limited the size of our disks to a relatively mal conductivity and a gap thickness estimated from small diameter microstructural analysis Representative fiber diameter and packing fraction The particular composite selected for this analysis was values were determined by SEM examination of made with Hi-Nicalon(Hi-Nicalon is a trademark used polished disc surface with fiber ends normal to the sur- by the Nippon Carbon Co., Yokohana 221, Japan) Sic face. Five randomly selected areas were examined using fiber contained within a polymer impregnated and pyr- a commercial image analysis routine(Prism View olyzed(PIP)-SiC matrix. The procedure used to fabri- Prism View is a trademark used by Analytical Vision, cate uniaxial Hi-Nicalon/PIP-SiC composites for this Inc, Rayleigh, NC, USA). The average diameter of the study follows. First, several Hi-Nicalon fiber tows(2- Hi-Nicalon fiber was 13.8+1.5 um, and a representative cm lengths) were aligned inside a piece of shrink-fit fiber packing fraction for the uniaxial composite was tubing that was then filled with a liquid pre-ceramic 0.566+0.040 polymer( CerasetTM, Ceraset is a trademark used by An SEM micrograph showing a transverse view of the DuPont Lanxide, Newark, DE, USA). The polymer was typical packing and alignment of the Hi-Nicalon fibers cured at 190C in a vacuum for 30 min so that the is given in Fig. 5. The space between the individual tubing simultaneously shrank and compressed the fibers fibers appears uniformly infiltrated by the PIP-SiC into a relatively high packing fraction (>0.6). After matrix. However, the matrix contains numerous curing, the now-rigid plug (8-mm diameter) was shrinkage cracks running both parallel and perpend removed from the tubing and wrapped in graphfoil cular to the fiber lengths. The parallel cracks obviously prior to a high-temperature treatment (HTT). The PIP- would interrupt the transverse heat conduction paths matrix component was purposely kept amorphous so between most of the filaments. Also, numerous direct that fiber contributions would dominate the overall fiber-to-fiber contacts are observed in Fig. 5, although composite thermal conductivity. For this reason, the the actual contact area between touching cylindrical HTT temperature(1100C)was selected below the Sic fibers should be limitedthe direct and gaseous conduction components might be comparable. The product components depend on the magnitudes of fd and fg,and should be very dependent upon the surface roughness,gap dimensions,tempera￾ture and environmental conditions. Separately deter￾mining fd and fg would be difficult because these quantities likely vary with changing environmental con￾ditions in a non-reproducible manner. However, hg 0 under vacuum conditions. Therefore,the product terms fdhd and fghg,can at least be estimated by performing experiments on the same sample in both vacuum and a known atmosphere. To a good approximation,the interfacial conductance measured in vacuum should represent only the fdhd term. Then to determine the component fghg,the total interfacial conductance h is measured in the gaseous phase of interest,and fdhd is subtracted from it. 4. The effective transverse thermal conductivity of example SiCf/SiC composites 4.1. Uniaxial Hi-NicalonTM/PIP-SiC composite To experimentally assess the analytical solutions for the H–J model,the transverse Keff for a particular uni￾axial SiCf/SiC composite was measured in three differ￾ent atmospheres at a fixed temperature. Using Keff, Kf, Km, a and Vf values determined experimentally,the overall interfacial conductance, h,was calculated from Eq. (4) for each atmosphere. Then fghg-values were estimated by the method described above and compared to calculated values based on known values of gas ther￾mal conductivity and a gap thickness estimated from microstructural analysis. The particular composite selected for this analysis was made with Hi-Nicalon (Hi-Nicalon is a trademark used by the Nippon Carbon Co.,Yokohana 221,Japan) SiC fiber contained within a polymer impregnated and pyr￾olyzed (PIP)-SiC matrix. The procedure used to fabri￾cate uniaxial Hi-Nicalon/PIP-SiC composites for this study follows. First,several Hi-Nicalon fiber tows ( 2- cm lengths) were aligned inside a piece of shrink-fit tubing that was then filled with a liquid pre-ceramic polymer (CerasetTM; Ceraset is a trademark used by DuPont Lanxide,Newark,DE,USA). The polymer was cured at 190
C in a vacuum for 30 min so that the tubing simultaneously shrank and compressed the fibers into a relatively high packing fraction ( f50.6). After curing,the now-rigid plug ( 8-mm diameter) was removed from the tubing and wrapped in graphfoil prior to a high-temperature treatment (HTT). The PIP￾matrix component was purposely kept amorphous so that fiber contributions would dominate the overall composite thermal conductivity. For this reason,the HTT temperature (1100
C) was selected below the SiC crystallization temperature for Ceraset ( 1600
C). However,the HTT conditions (30 min at 1100
C in argon) were sufficient to form a fairly stable,Si–C–O matrix structure and to expel most of the gaseous com￾ponents. The infiltration and HTT cycle was repeated one time to increase the composite density. After the second HTT,a rod was core-drilled from the center of the plug,and several disks for thermal diffusivity mea￾surements were cut from the rod with fiber alignment either parallel or perpendicular to the flat disk surfaces. The individual disks were 6.2-mm diameter by 2-mm thick. Normally,for laser flash thermal diffusivity mea￾surements one prefers a diameter-to-thickness ratio 55, however later tests requiring packaging in small radia￾tion capsules limited the size of our disks to a relatively small diameter. Representative fiber diameter and packing fraction values were determined by SEM examination of a polished disc surface with fiber ends normal to the sur￾face. Five randomly selected areas were examined using a commercial image analysis routine (Prism ViewTM; Prism View is a trademark used by Analytical Vision, Inc.,Rayleigh,NC,USA). The average diameter of the Hi-Nicalon fiber was 13.81.5 mm,and a representative fiber packing fraction for the uniaxial composite was 0.5660.040. An SEM micrograph showing a transverse view of the typical packing and alignment of the Hi-Nicalon fibers is given in Fig. 5. The space between the individual fibers appears uniformly infiltrated by the PIP-SiC matrix. However,the matrix contains numerous shrinkage cracks running both parallel and perpendi￾cular to the fiber lengths. The parallel cracks obviously would interrupt the transverse heat conduction paths between most of the filaments. Also,numerous direct fiber-to-fiber contacts are observed in Fig. 5,although the actual contact area between touching cylindrical fibers should be limited. Fig. 5. A transverse view by SEM of a polished surface for a uniaxial Hi-NicalonTM/PIP-SiC (amorphous CerasetTM) composite after two infiltration and HTT-1100
C cycles with f=0.566. G.E. Youngbloodet al. / Composites Science andTechnology 62 (2002) 1127–1139 1133