J Mater Sci(2007)42:763-771 DOI10.1007/s10853-006-1443-3 Effect of heat treatment in air on the thermal properties of siC fibre-reinforced composite. Part 1: a barium osumilite (BMas) matrix glass ceramic composite R. Yilmaz·R. Taylor October 2003/ Accepted: 13 April 2006/Published online: 12 January 2007 Springer Science+Business Media, LLC 2007 Abstract The thermal properties have been studied Introduction on a glass ceramic composite comprised of a barium umilite(BMAS)matrix reinforced with SiC (Tyran- SiC fibres have been developed and used in a variety of no) fibres which has been subjected to a heat treatment reinforced glass ceramic composites over the last two in air in the range of 700-1, 200C. Microstructural decades. They are intended to provide reinforcement studies were carried out especially on of the interface and to improve the performance of structural ceramics between fibre and matrix. The presence of a carbon for high temperature applications [1] thin layer in the interface is a typical observation in Sic Ceramics are brittle, have low fracture toughness fibre-reinforced glass ceramic matrix composite sys- and they fail in a catastrophic manner. The brittleness tems. The microstructural evaluation and thermal of the material results from sudden propagation of a properties showed a degradation of interfacial layer crack under the applied stress. Thus, for the future ccurred at low heat treatment temperatures, (700- application of ceramics at high temperatures it is 800C)this was attributed to the fact that, at those necessary to develop ceramic fibres as reinforcement in heat treatment temperatures the carbon rich layer suitable matrices. It is also necessary to understand the formed during processing was oxidised away leaving fracture behaviour of these composites and behaviour voids between fibre and matrix, which were linked by of the interface for design matters in order to optimise isolated silicon-rich bridges. After heat treatment at their mechanical properties higher temperatures of 1,000-1, 200C, the thermal One of the most promising glass ceramic matrix properties were retained or even enhanced by leaving a systems is barium osumilite(BMAS), first developed thick interfacial layer. by Brennan et al. [2]. Thermal properties of BMAS glass ceramic matrix composites have been studied by Johnson et al. 3]. Although the use of such materials ill be governed by the development of the suitable mechanical properties, the accurate thermo-physical properties are also needed such as thermal expansion, thermal diffusivity and conductivity During manufacturing stage, due to fibr R. Yilmaz(凶) reaction, a carbon layer is formed [4, 5] in SiC fibre- Technical Education Faculty. Metal Education Division Sakarya University, Esentepe Campus, 54187 Sakarya reinforced glass ceramic composites. This carbon rich Turkey layer has been observed in MAS, calcium aluminium e-mail: ryilmaz@sakarya. edu.tr silicate (CAS), barium aluminium silicate (BAS) barium magnesium aluminium silicate(BMAs)glass R. Pavlo Manchester Materials Science Centre, University of ceramic matrix composites 6-12 Manchester Institute of Science and Technology. Grosvenor The in-service performance of these composites Street, Manchester M1 7HS. England depends on the environmental conditions. The thermal
Effect of heat treatment in air on the thermal properties of SiC fibre-reinforced composite. Part 1: a barium osumilite (BMAS) matrix glass ceramic composite R. Yilmaz Æ R. Taylor Received: 13 October 2003 / Accepted: 13 April 2006 / Published online: 12 January 2007 Springer Science+Business Media, LLC 2007 Abstract The thermal properties have been studied on a glass ceramic composite comprised of a barium osumilite (BMAS) matrix reinforced with SiC (Tyranno) fibres which has been subjected to a heat treatment in air in the range of 700–1,200 C. Microstructural studies were carried out especially on of the interface between fibre and matrix. The presence of a carbon thin layer in the interface is a typical observation in SiC fibre-reinforced glass ceramic matrix composite systems. The microstructural evaluation and thermal properties showed a degradation of interfacial layer occurred at low heat treatment temperatures, (700– 800 C) this was attributed to the fact that, at those heat treatment temperatures the carbon rich layer formed during processing was oxidised away leaving voids between fibre and matrix, which were linked by isolated silicon-rich bridges. After heat treatment at higher temperatures of 1,000–1,200 C, the thermal properties were retained or even enhanced by leaving a thick interfacial layer. Introduction SiC fibres have been developed and used in a variety of reinforced glass ceramic composites over the last two decades. They are intended to provide reinforcement and to improve the performance of structural ceramics for high temperature applications [1]. Ceramics are brittle, have low fracture toughness and they fail in a catastrophic manner. The brittleness of the material results from sudden propagation of a crack under the applied stress. Thus, for the future application of ceramics at high temperatures it is necessary to develop ceramic fibres as reinforcement in suitable matrices. It is also necessary to understand the fracture behaviour of these composites and behaviour of the interface for design matters in order to optimise their mechanical properties. One of the most promising glass ceramic matrix systems is barium osumilite (BMAS), first developed by Brennan et al. [2]. Thermal properties of BMAS glass ceramic matrix composites have been studied by Johnson et al. [3]. Although the use of such materials will be governed by the development of the suitable mechanical properties, the accurate thermo-physical properties are also needed such as thermal expansion, thermal diffusivity and conductivity. During manufacturing stage, due to fibre and matrix reaction, a carbon layer is formed [4, 5] in SiC fibrereinforced glass ceramic composites. This carbon rich layer has been observed in MAS, calcium aluminium silicate (CAS), barium aluminium silicate (BAS), barium magnesium aluminium silicate (BMAS) glass ceramic matrix composites [6–12]. The in-service performance of these composites depends on the environmental conditions. The thermal, R. Yilmaz (&) Technical Education Faculty, Metal Education Division, Sakarya University, Esentepe Campus, 54187 Sakarya, Turkey e-mail: ryilmaz@sakarya.edu.tr R. Taylor Manchester Materials Science Centre, University of Manchester Institute of Science and Technology, Grosvenor Street, Manchester M1 7HS, England J Mater Sci (2007) 42:763–771 DOI 10.1007/s10853-006-1443-3 123
J Mater sci(2007)42:763-771 microstructural and mechanical properties of the brought about by heat treatment. Changes in the composites can be affected by the environment with microstructure of the specimen were monitored using time. Some studies carried out on glass ceramic matrix standard microstructural characterisation techniques composites [13-16 show that the interface between and physical properties such as thermal diffusivity of fibre and matrix in the composites can be affected by the specimen were measured before and after heat the temperature of the environment. Oxidation occurs treatment at the interface during the heat treatment with temper ature and this results in a degradation of the mechanical properties of the composites. This was confirmed using Experimental mechanical tests such as tensile, three- or four-point bending, fibre pullout, creep, etc. These studies suggest Materials that this behaviour can also be studied using thermal property tests for composites exposed to a heat treat- A 0/90 laminated SiC/BMAS composite was supplied ment at varous temperatures n UK. The by the National Physical Laborator S $ g Limited amount of work has been carried out on the composite was manufactured by Harwell Technology thermal properties of such composites. Hasselmann in England. The preparation route was for the Tyranno and co-workers have carried out the most comprehen- fibre tow to be desized in a furnace, taken through a sive series of thermal diffusivity measurement on a slurry of glass frit, removed and wound on a wheel unit, wide range of composite systems 3, 17-24. Thermal allowed to dry for 20 min, cut and laid up manually in diffusivity or conductivity can be affected by the layers for hot pressing in a graphite die at-1, 200C for relative volume fraction of the constituents(fibre/ 10 min. It was then crystallised via a proprietary heat matrix and porosity), the orientation of the fibres, the treatment, which involved in heating to a temperature particular processing route chosen, and the structure of not exceeding 1, 300C [AEA Harwell Technology Heating to high temperatures can affect the thermal private communication the fibre/matrix interface Thermal diffusivity and thermal expansion measure diffusivity of SiC fibre-reinforced composites [19, ments were carried on the as-received material and primarily because of the change in fibre/matrix inter- after heat treatments in air at temperatures of 700C, face in thermal exposure. a number of studies have800°C,900°C,1,000°C,1,100°C,1,200° C for times of indicated that thermal conductivity and diffusivity of ranging from 1 to 30 h composites can be affected by a thermal barrier resistance of the interface [18-21, 24]. The direction Microstructural examination of heat flow also plays an important role in determining le effective diffusivity of composites in which there is X-ray diffraction analysis fibre/matrix interface resistance. The greatest effe will be observed when heat flow is perpendicular to the X-ray diffraction studies were carried out to identi fibre/matrix interface the phases present in the composites. These were Oxidation resulting in the removal of carbon at performed using a PHILIPS E'XPERT diffractometer interface behaved as a thermal barrier. When the Pw 3710 by using nickel-filtered copper K radiation carbon layer is oxidised, the thermal conductivity at with a graphite secondary monochromator Scans at a the fibre/matrix interface occurs by gaseous conduction step width of 0.005 for 20 values from 20 to 700 were and resulted in lower thermal diffusivity in composites used on samples that were solid bulk plates 10 mm [18. However, there have been relatively few reported square by 2 mm thick. The diffraction traces obtained observations of the effect of thermal exposure on the were compared against standard Xrd patterns for a thermal properties. Certainly there is not any system- range of materials atic investigation undertaken on this. It is possible that the measurement of the thermal diffusivity can be used Optical and scanning electron microscopy as a qualitative non-destructive tool to determine the integrity of the fibre/matrix interfaces and to monitor Sample preparations were taken in three stages; the microstructural changes occurring in the fibres or specimens were ground on a Buehler DATAMET microprocessor grinding/polishing system. METLAP 4 In this work, a detailed microstructural character- wheel with 9 um METaDI diamond slurry, with the ization of the BMAS/SiC system has been presented wheel contra-rotating at 25 r.P. m, then on a Beuhler along with subsequent changes in microstructure Metlap 2 wheel with 6 um diamond slurry, the wheel 2 Springer
microstructural and mechanical properties of the composites can be affected by the environment with time. Some studies carried out on glass ceramic matrix composites [13–16] show that the interface between fibre and matrix in the composites can be affected by the temperature of the environment. Oxidation occurs at the interface during the heat treatment with temperature and this results in a degradation of the mechanical properties of the composites. This was confirmed using mechanical tests such as tensile, three- or four-point bending, fibre pullout, creep, etc. These studies suggest that this behaviour can also be studied using thermal property tests for composites exposed to a heat treatment at various temperatures. Limited amount of work has been carried out on the thermal properties of such composites. Hasselmann and co-workers have carried out the most comprehensive series of thermal diffusivity measurement on a wide range of composite systems [3, 17–24]. Thermal diffusivity or conductivity can be affected by the relative volume fraction of the constituents (fibre/ matrix and porosity), the orientation of the fibres, the particular processing route chosen, and the structure of the fibre/matrix interface. Heating to high temperatures can affect the thermal diffusivity of SiC fibre-reinforced composites [19], primarily because of the change in fibre/matrix interface in thermal exposure. A number of studies have indicated that thermal conductivity and diffusivity of composites can be affected by a thermal barrier resistance of the interface [18–21, 24]. The direction of heat flow also plays an important role in determining the effective diffusivity of composites in which there is fibre/matrix interface resistance. The greatest effect will be observed when heat flow is perpendicular to the fibre/matrix interface. Oxidation resulting in the removal of carbon at interface behaved as a thermal barrier. When the carbon layer is oxidised, the thermal conductivity at the fibre/matrix interface occurs by gaseous conduction and resulted in lower thermal diffusivity in composites [18]. However, there have been relatively few reported observations of the effect of thermal exposure on the thermal properties. Certainly there is not any systematic investigation undertaken on this. It is possible that the measurement of the thermal diffusivity can be used as a qualitative non-destructive tool to determine the integrity of the fibre/matrix interfaces and to monitor microstructural changes occurring in the fibres or matrix. In this work, a detailed microstructural characterization of the BMAS/SiC system has been presented along with subsequent changes in microstructure brought about by heat treatment. Changes in the microstructure of the specimen were monitored using standard microstructural characterisation techniques and physical properties such as thermal diffusivity of the specimen were measured before and after heat treatment. Experimental Materials A 0/90 laminated SiC/BMAS composite was supplied by the National Physical Laboratory in UK. The composite was manufactured by Harwell Technology in England. The preparation route was for the Tyranno fibre tow to be desized in a furnace, taken through a slurry of glass frit, removed and wound on a wheel unit, allowed to dry for 20 min, cut and laid up manually in layers for hot pressing in a graphite die at ~1,200 C for 10 min. It was then crystallised via a proprietary heat treatment, which involved in heating to a temperature not exceeding 1,300 C [AEA Harwell Technology, private communication]. Thermal diffusivity and thermal expansion measurements were carried on the as-received material and after heat treatments in air at temperatures of 700 C, 800 C, 900 C, 1,000 C, 1,100C, 1,200C for times of ranging from 1 to 30 h. Microstructural examination X-ray diffraction analysis X-ray diffraction studies were carried out to identify the phases present in the composites. These were performed using a PHILIPS E’XPERT diffractometer PW 3710 by using nickel-filtered copper K radiation with a graphite secondary monochromator. Scans at a step width of 0.005 for 2h values from 20 to 70 were used on samples that were solid bulk plates 10 mm square by 2 mm thick. The diffraction traces obtained were compared against standard XRD patterns for a range of materials. Optical and scanning electron microscopy Sample preparations were taken in three stages; the specimens were ground on a Buehler DATAMETmicroprocessor grinding/polishing system. METLAP 4 wheel with 9 lm METADI diamond slurry, with the wheel contra-rotating at 25 r.p.m, then on a Beuhler Metlap 2 wheel with 6 lm diamond slurry, the wheel 123 764 J Mater Sci (2007) 42:763–771
J Mater Sci(2007)42:763-771 contra-rotating at 120 r.P. m. After that polishing was were considered but the most suitable condition was done using a TEXMET platen with a wheel using 1 um found to be as follows: 5 kV ion beam energy 0.4A diamond slurry. Final polishing was carried out with current at 30, 25 and 16 impingement angle for 30 h. colloidal silica. Each section takes 10 min and the After milling the foils were then taken from the io polishing pressure was set to 1n per sample and beam thinner and placed directly into the electron maintained at that level during preparation. The microscope for examination. The analytical electron samples were then finished by washing with water for microscopy was carried out using a Philips EM 400 and 1 min and dried. After preparation of the specimens, CM 20 operated at 120 kV and 200 kv, respectively they were mounted on to an aluminium stub and Both were equipped with an Energy dispersive spec- initially coated with carbon or gold in order to prevent troscopy(EDS)system, and the investigation was charging in the microscope. An Edwards coating conducted using bright field, lattice imaging diffraction system E 306A was used for coating. A conducting and micro diffraction techniques silver paste was used with the carbon-coated samples painted on the edge of the sample connecting it with Thermal diffusivity measurement the stub to improve electrical contact. The surface of the heat-treated samples were examined using Philips Thermal diffusivity measurements were carried out 525 scanning electron microscopes(SEM) operating at using the laser flash method originally described by 20 kV with scanning facility operating with a computer Parker et al. [25]. The thermal diffusivity equipment programme-connected microscope used at UMIST has been previously described by Taylor [ 26] Transmission electron microscopy The specimens used in the measurements were in the form of 10 mm- plates with a thickness of appro Discs of the same diameter as the electron microscope mately 2 mm. The specimens were cut from the specimen holder (3 mm in diameter and 2 mm thick) composite plate by using a slow speed diamond saw. were cut from the heat-treated specimens. The spec- In order to ensure optimum absorption of the laser flash imens were prepared for transmission electron micros- at the sample front surface and maximum emissivity for copy(tEm) by a combination of mechanical polishing monitoring the transient temperature at the opposite and ion beam thinning techniques. The specimens were face of the sample, the both faces of the sample were initially ground on 1200 grit wet Sic abrasive until the coated with a colloidal graphite film. Measurements thickness of the specimen was reduced to -150 m. were performed from 100C up to 1,000C with the Further grinding was carried out until the specimen 100C interval. At least three measurements were thickness was 70 um. The foils were then transferred to taken at each measured temperature and averaged a Gatan ion beam thinner. Several milling conditions value was taken for plotting of the graphs Fig. 1 X-ray diffraction spectrum of as-received terial of bMaS/SiC CN Co: Cordierite B: Barium osumilite S: SiC fibre 8 400 300 8 1001B o"T"Jo""4o""5o""6("6
contra-rotating at 120 r.p.m. After that polishing was done using a TEXMET platen with a wheel using 1 lm diamond slurry. Final polishing was carried out with colloidal silica. Each section takes 10 min and the polishing pressure was set to 1 N per sample and maintained at that level during preparation. The samples were then finished by washing with water for 1 min and dried. After preparation of the specimens, they were mounted on to an aluminium stub and initially coated with carbon or gold in order to prevent charging in the microscope. An Edwards coating system E 306A was used for coating. A conducting silver paste was used with the carbon-coated samples painted on the edge of the sample connecting it with the stub to improve electrical contact. The surface of the heat-treated samples were examined using Philips 525 scanning electron microscopes (SEM) operating at 20 kV with scanning facility operating with a computer programme-connected microscope. Transmission electron microscopy Discs of the same diameter as the electron microscope specimen holder (3 mm in diameter and 2 mm thick) were cut from the heat-treated specimens. The specimens were prepared for transmission electron microscopy (TEM) by a combination of mechanical polishing and ion beam thinning techniques. The specimens were initially ground on 1200 grit wet SiC abrasive until the thickness of the specimen was reduced to ~150 m. Further grinding was carried out until the specimen thickness was 70 lm. The foils were then transferred to a Gatan ion beam thinner. Several milling conditions were considered but the most suitable condition was found to be as follows; 5 kV ion beam energy 0.4 A current at 30, 25 and 16 impingement angle for 30 h. After milling, the foils were then taken from the ion beam thinner and placed directly into the electron microscope for examination. The analytical electron microscopy was carried out using a Philips EM 400 and CM 20 operated at 120 kV and 200 kV, respectively. Both were equipped with an Energy dispersive spectroscopy (EDS) system, and the investigation was conducted using bright field, lattice imaging diffraction and micro diffraction techniques. Thermal diffusivity measurement Thermal diffusivity measurements were carried out using the laser flash method originally described by Parker et al. [25]. The thermal diffusivity equipment used at UMIST has been previously described by Taylor [26]. The specimens used in the measurements were in the form of 10 mm2 plates with a thickness of approximately 2 mm. The specimens were cut from the composite plate by using a slow speed diamond saw. In order to ensure optimum absorption of the laser flash at the sample front surface and maximum emissivity for monitoring the transient temperature at the opposite face of the sample, the both faces of the sample were coated with a colloidal graphite film. Measurements were performed from 100 C up to 1,000 C with the 100 C interval. At least three measurements were taken at each measured temperature and averaged value was taken for plotting of the graphs. Fig. 1 X-ray diffraction spectrum of as-received material of BMAS/SiC CMC 123 J Mater Sci (2007) 42:763–771 765
J Mater sci(2007)42:763-771 Results between heating and cooling measurement runs. Te ssess material variability take Characterisation of as-received material on four samples cut from different regions of the as-received plate. The results are plotted in Fig. 2 and To identify the phases present in the as-received show a scatter of *3%. The median value obtained material, the X-ray diffraction studies were carried from this curve will be included, for comparison out on the samples. Figure 1 shows a typical scan over purposes in all future diffusivity/temperature plots the range 10< 20 <70 from which three crystalline Changes in thermal diffusivity were noted for the phases have been identified: Barium osumilite(BaMg2 samples heat treated for times as short as 1 h at Al3(SigAl3O30)), hexacelsian(BaAl3Si2Os)and cor- temperatures of 700 and 900C. However for higher dierite(Mg2AlgSis O1s) temperature anneals at 1,000-1, 200C, a negligible change was noted. The thermal diffusivity data after 1 h Thermal properties heat treatment in the range 700-1, 200C are presented in Fig 3. It can be seen that the greatest degradation Thermal diffusivity measurements thermal diffusivity is noted for the sample heat treated at700° C with the value of44×10-3cm2s-lwas Thermal diffusivity was measured over the tempera- measured at 100C and fair ture range 100-1,000C. There was no change noted 3.44 x 10 cm-s of being recorded above 350C. Fig. 2 Measurement of thermal diffusivity of BMAS/ a AR2 SiC as-received materials AR: Samples cut from different regions of the as- 0.007 0.006 0.005 010020030040050060070080090010001100 erc] Fig 3 Measurement of 0.008 thermal diffusivity of BMAS/ Sic after heat treatment for 900°c I h in air 0.007 0.005 0.003 010020030040050060070080090010001100 Temperature rcl
Results Characterisation of as-received material To identify the phases present in the as-received material, the X-ray diffraction studies were carried out on the samples. Figure 1 shows a typical scan over the range 10 < 2h < 70 from which three crystalline phases have been identified: Barium osumilite (BaMg2 Al3 (Si9Al3O30)), hexacelsian (BaAl3Si2O8) and cordierite (Mg2Al4Si5O18). Thermal properties Thermal diffusivity measurements Thermal diffusivity was measured over the temperature range 100–1,000 C. There was no change noted between heating and cooling measurement runs. To assess material variability, measurements were taken on four samples cut from different regions of the as-received plate. The results are plotted in Fig. 2 and show a scatter of ±3%. The median value obtained from this curve will be included, for comparison purposes in all future diffusivity/temperature plots. Changes in thermal diffusivity were noted for the samples heat treated for times as short as 1 h at temperatures of 700 and 900 C. However for higher temperature anneals at 1,000–1,200 C, a negligible change was noted. The thermal diffusivity data after 1 h heat treatment in the range 700–1,200 C are presented in Fig. 3. It can be seen that the greatest degradation in thermal diffusivity is noted for the sample heat treated at 700 C with the value of 4.4 · 10–3 cm2 s –1 was measured at 100 C and fairly constant value 3.44 · 10–3 cm2 s –1 of being recorded above 350 C. Fig. 2 Measurement of thermal diffusivity of BMAS/ SiC as-received materials. AR: Samples cut from different regions of the asreceived composite Fig. 3 Measurement of thermal diffusivity of BMAS/ SiC after heat treatment for 1 h in air 123 766 J Mater Sci (2007) 42:763–771
J Mater Sci(2007)42:763-771 This value is 39% lower than the value measured for the thermal diffusivity values after 700-900C heat treat as-received material ments all show lower values than the as-received A more detailed set of thermal diffusivity results materials with the ranking order of greatest change after 10 h heat treatments are shown in Fig. 4. The shown after annealing at 700C, 800C and 900oC 1, 200C treatment shows a slight enhancement of thermal diffusivity, as does the 1,100C, whereas heat Microstructural studies treatment at 1000oC shows similar values to the values of the as-received material. all the other heat SEm studies treatments at temperatures lower than 1,000C show lower thermal diffusivities than the as-received mate- Limited SEM studies were carried out on samples heat rial. However, the 700C heat-treated sample again treated at temperatures between 700C and 1, 200C. shows the lowest thermal diffusivity values The most noteworthy observation was that a gap These trends are also maintained for the 30-h heat between fibre and matrix was noted at low tempera treatments for which thermal diffusivity results are ture. This is illustrated in Fig. 6 for a sample heat shown in Fig. 5. Again the heat treatment at 1, 200 treated at 700oC for 30 h. The shows higher thermal diffusivity values than those for residual stresses exist in the samples heated at lowe the as-received material, whereas data for samples temperatures. Residual glassy phases would tend to heated at 1,000C and 1, 100C are very close to that fow through any gaps in the matrix but the extent to of the as-received composite. On the other hand, the which this occurs depends on temperature of the heat Fig. 4 Measurement of 000 thermal diffusivity of BMAS/ Sic CMC after heat treatment in air for 10 h 0008 900°C 0D07 0006 日As recelved △△ 0004 0003 01002003040050060070080090010001100 Temperature rCl Fig 5 Measurement of 0.009 thermal diffusivity of BMAS/ 700°C 末 Sic CMC after heat treatment in air for 30 h 0.008 1100°c 0,006 x1200℃C 0.005 0.003 10020030040050060070080090010001100 Temperature rc
This value is 39% lower than the value measured for the as-received material. A more detailed set of thermal diffusivity results after 10 h heat treatments are shown in Fig. 4. The 1,200 C treatment shows a slight enhancement of thermal diffusivity, as does the 1,100 C, whereas heat treatment at 1,000 C shows similar values to the values of the as-received material. All the other heat treatments at temperatures lower than 1,000 C show lower thermal diffusivities than the as-received material. However, the 700 C heat-treated sample again shows the lowest thermal diffusivity values. These trends are also maintained for the 30-h heat treatments for which thermal diffusivity results are shown in Fig. 5. Again the heat treatment at 1,200 C shows higher thermal diffusivity values than those for the as-received material, whereas data for samples heated at 1,000 C and 1,100 C are very close to that of the as-received composite. On the other hand, the thermal diffusivity values after 700–900 C heat treatments all show lower values than the as-received materials with the ranking order of greatest change shown after annealing at 700 C, 800 C and 900 C. Microstructural studies SEM studies Limited SEM studies were carried out on samples heat treated at temperatures between 700 C and 1,200 C. The most noteworthy observation was that a gap between fibre and matrix was noted at low temperature. This is illustrated in Fig. 6 for a sample heat treated at 700 C for 30 h. These may indicate that residual stresses exist in the samples heated at lower temperatures. Residual glassy phases would tend to flow through any gaps in the matrix but the extent to which this occurs depends on temperature of the heat Fig. 4 Measurement of thermal diffusivity of BMAS/ SiC CMC after heat treatment in air for 10 h Fig. 5 Measurement of thermal diffusivity of BMAS/ SiC CMC after heat treatment in air for 30 h 123 J Mater Sci (2007) 42:763–771 767
J Mater sci(2007)42:763-771 in significant diffusivity changes at the lower heat treatment temperatures (700-800C). There wa marked change in the microstructures of samples heat treated in this temperature range where voids and gap can be seen at the interfaces. In Fig. 8, there is a clearly identifiable gap of -8 nm at the interface between fibre and matrix but also linkages between fibre and matrix were also seen as noted by Plucknett et al. [14] After heat treatment to higher temperature no evidence of voids was detected. Examples of these are the micrographs in Fig 8(a) and(b), which show interfaces in samples heated at 1, 100C and 1, 200C. It is noticeable that the interface has now thickened to Fig6 Back-scattered electron SEM images showing by arrows -20-25 nm. For an even higher heat treatment tem- glassy phases around the fibre after heat treatment at 700C for perature the interface is even thicker(40-50 nm), as evident with sample heat treated at 1, 200C(Fig. 9b) The change on the interface structure as a result of reatment. However, it was not easy for all the glass to thermal treatment can be summarised as a transition flow. Therefore, local concentrations of the glassy from a carbon-rich interphase in the as-fabricated phase occurred. After a higher temperature anneal at composite to a carbon-free interphase after heat treat higher than 900C( Fig. 7)and reduced glassy phase ments in the intermediate temperature range content was seen in the matrix. This reduction sugge (700-800C)and finally to retention of a carbon-rich that some recrystallisation of the residual glassy pha interface at higher ageing temperatures(900-1, 200C) had occurred The interphase formed at the higher heat treatment temperature is also much thicker (45 nm)than that TEM studies formed in the as-received composite(20 nm)[12] Figure 10 collates information from the EDs traces TEM studies were carried out on the selected samples. of the various regions of the interface and compare Whereas, SEM studies were intended primarily to the information with that obtained for as-received identify the changes in phase structure and glassy material after heat treatments at 700C and 1, 200C phase content in the TEM studies. A special emphasis Although EDS analysis does not give reliable results was paid to studying the fibre/matrix interface where due to its limitations for low atomic weight elements microstructural changes had been previously reported such as o and C and the results were affected by [13-15]. Because of difficulties in preparing of suitable the back ground noise, this can provide a useful TEM samples, it was decided that the heat-treated samples to be studied were those heat treated for 30 h nce heat treatment for this period had resulted fbre Fig8 TEM bright field images of the sample heated at 700C Fig. 7 SEM micrograph after heat treatment at 1,000C for 30 h showing gap and voids at the interfac 2 Springer
treatment. However, it was not easy for all the glass to flow. Therefore, local concentrations of the glassy phase occurred. After a higher temperature anneal at higher than 900 C (Fig. 7) and reduced glassy phase content was seen in the matrix. This reduction suggests that some recrystallisation of the residual glassy phases had occurred. TEM studies TEM studies were carried out on the selected samples. Whereas, SEM studies were intended primarily to identify the changes in phase structure and glassy phase content in the TEM studies. A special emphasis was paid to studying the fibre/matrix interface where microstructural changes had been previously reported [13–15]. Because of difficulties in preparing of suitable TEM samples, it was decided that the heat-treated samples to be studied were those heat treated for 30 h in air since heat treatment for this period had resulted in significant diffusivity changes at the lower heat treatment temperatures (700–800 C). There was a marked change in the microstructures of samples heat treated in this temperature range where voids and gaps can be seen at the interfaces. In Fig. 8, there is a clearly identifiable gap of ~8 nm at the interface between fibre and matrix but also linkages between fibre and matrix were also seen as noted by Plucknett et al. [14]. After heat treatment to higher temperature no evidence of voids was detected. Examples of these are the micrographs in Fig. 8(a) and (b), which show interfaces in samples heated at 1,100 C and 1,200 C. It is noticeable that the interface has now thickened to ~20–25 nm. For an even higher heat treatment temperature the interface is even thicker (~40–50 nm), as evident with sample heat treated at 1,200 C (Fig. 9b). The change on the interface structure as a result of thermal treatment can be summarised as a transition from a carbon-rich interphase in the as-fabricated composite to a carbon-free interphase after heat treatments in the intermediate temperature range (700–800 C) and finally to retention of a carbon-rich interface at higher ageing temperatures (900–1,200 C). The interphase formed at the higher heat treatment temperature is also much thicker (45 nm) than that formed in the as-received composite (20 nm) [12]. Figure 10 collates information from the EDS traces of the various regions of the interface and compares the information with that obtained for as-received material after heat treatments at 700 C and 1,200 C. Although EDS analysis does not give reliable results due to its limitations for low atomic weight elements such as O and C and the results were affected by the back ground noise, this can provide a useful Fig. 7 SEM micrograph after heat treatment at 1,000 C Fig. 8 TEM bright field images of the sample heated at 700 C for 30 h showing gap and voids at the interface Fig. 6 Back-scattered electron SEM images showing by arrows glassy phases around the fibre after heat treatment at 700 C for 30 h 123 768 J Mater Sci (2007) 42:763–771
J Mater Sci(2007)42:763-771 a) fibre interface fibre Fig. 10 EDS analysis showing the composition of SiC/BMAS omposite interface after heat treatments at 700C and 1, 200C (NF: EDS analysis from near fibre, MID: Middle of interface. Fig9 TEM image of the sample heated (a)at 1, 100C and(b) AF: Away from fibre, U: Unidentified phase) semi-quantitative analysis. These results may give a oted by other workers [3, 20-23 and estimated general idea about the composition at the interface. theoretically [24]. Temperature changes during service Those clearly show that after the 700C heat treat- of a component can result in structural changes such as ment the carbon present at the interface is removed development of porosity or the formation of other leaving a silicon-rich layer whereas after the 1,200oC phases at the interface. The diffusivity sample thick heat treatment the carbon layer is not only still present ness was 2 mm so since the ply thickness was -0.2 mm but is even enhanced. Near the fibre and near the some interfaces were always normal to heat flow so any matrix are two regions which are carbon rich. Inter- gap at the interface would be expected to affect the estingly Ti, which can only have originated in the fibre thermal diffusivity of the composite. Significant effects is present in all the EDS spectra even that remote fre were observed particularly in the temperature range the fibre interface. The presence of Si, O, Mg, Al 700-900C. The diffusivity measurements were sup- Ti is attributed to diffusion of the matrix elements to ported by SEM and TEM studies. During the process- the interface. This has good agreement with previous ing of Sic fibre-reinforced composites in a glass findings [13, 14]. ceramic matrix a carbon layer is formed at the fibre/ matrix interface either by the reaction first mentioned orDer and Chyung [4(Eq. 1) Discussion SiC+O2→SiO2+C Clearly the fibre/matrix interface plays an important or that also suggested by Cooper and Chyung [4] role in the thermal properties of the composites. A Bemson et al. 5] and Le Strat et al. [ 27(Eq 2) discontinuity between fibres and matrix can result in decrease in thermal diffusivity, a feature that has been SiC 200- SiO2+ 3C
semi-quantitative analysis. These results may give a general idea about the composition at the interface. Those clearly show that after the 700 C heat treatment the carbon present at the interface is removed leaving a silicon-rich layer whereas after the 1,200 C heat treatment the carbon layer is not only still present but is even enhanced. Near the fibre and near the matrix are two regions which are carbon rich. Interestingly Ti, which can only have originated in the fibre is present in all the EDS spectra even that remote from the fibre interface. The presence of Si, O, Mg, Al and Ti is attributed to diffusion of the matrix elements to the interface. This has good agreement with previous findings [13, 14]. Discussion Clearly the fibre/matrix interface plays an important role in the thermal properties of the composites. A discontinuity between fibres and matrix can result in a decrease in thermal diffusivity, a feature that has been noted by other workers [3, 20–23] and estimated theoretically [24]. Temperature changes during service of a component can result in structural changes such as development of porosity or the formation of other phases at the interface. The diffusivity sample thickness was 2 mm so since the ply thickness was ~0.2 mm some interfaces were always normal to heat flow so any gap at the interface would be expected to affect the thermal diffusivity of the composite. Significant effects were observed particularly in the temperature range 700–900 C. The diffusivity measurements were supported by SEM and TEM studies. During the processing of SiC fibre-reinforced composites in a glass ceramic matrix a carbon layer is formed at the fibre/ matrix interface either by the reaction first mentioned by Cooper and Chyung [4] (Eq. 1) SiC þ O2 ! SiO2 þ C ð1Þ or that also suggested by Cooper and Chyung [4], Bemson et al. [5] and Le Strat et al. [27] (Eq. 2) SiC þ 2CO ! SiO2 þ 3C ð2Þ Fig. 10 EDS analysis showing the composition of SiC/BMAS composite interface after heat treatments at 700 C and 1,200 C (NF: EDS analysis from near fibre, MID: Middle of interface, Fig. 9 TEM image of the sample heated ( AF: Away from fibre, U: Unidentified phase) a) at 1,100 C and (b) at 1,200 C 123 J Mater Sci (2007) 42:763–771 769
J Mater sci(2007)42:763-771 The BMAS/SiC (Tyranno) fibre composite as rapid at low temperatures could be due to slower received had a carbon-rich interface typically 20 nm kinetics of oxidation. However as the temperature is thick, a typical thickness also noted by Plucknett et al. increased, two effects could contribute to reducing the [14. The thickness of the carbon-rich layer does effect however depend on hot pressing temperature and time [28, 29]and thicknesses as high as 250 nm have been (a) Any residual glassy phases in the matrix could noted 10]. The carbon exists as an amorphous phase at begin to soften. The extent to which these may lower manufacturing temperatures but if manufactur- flow to fill gaps depends on the viscosity. As ing temperatures are higher than 1, 250C graphitic temperature is increased, viscosity is reduced and carbon formation has been seen [10, 11. The EDS viscous fow can occur analyses of the interfacial layer shows it to consist of C (D) As the temperature is increased further oxidation (42%),O(13%)and Si (23%)which suggests a of the silicon carbide could occur to produce more mixture of C and Sioz. Interestingly for the BMAs SiOz and further carbon(pipe line diffusion) composite the interfacial layer also contained some Ti The results examined for the anneals at the higher which can only have come from the Tyranno fibre temperatures (>1,000C) revealed that the thermal suggesting that the formation of the interfacial layer diffusivity is changed by very little as a result of these had occurred by oxidation of the fibre via one of the heat treatments. The thickness of the interface how two reactions proposed by Eq (1)or(2) ever is much thicker than that of the as-received The most dramatic results during the heat treat- composites, amounting to some -45 nm after the 30 h ments are for the heat treatments in the temperature heat treatment at 1, 200C. Examination by EDs of range 700-900C where significant decreases in diffu- various regions of the interface after a 30 h heat sivity and thermal expansion coefficient are noted. treatment reveals it to contain Si, O and C possibly as TEM micrographs clearly show the development of a result of reaction. The interfacial layer contained Ti all gap at the interface after heat treatments in this the way through thus supporting the view that the temperature range. This is clearly due to the oxidation reaction products were the result of oxidation of the of the carbon to leave a silicon oxide-rich region SiC fibre. Interestingly the presence of Ba, Al,Mg was forming the bridges between matrix and fibre. Pluck- also noted near the matrix interface suggesting some nett et al. [14] reported similar behaviour and stated diffusion from the matrix also. The effect of these heat that the bridges they observed were silica. The degra- treatments on the diffusivity is difficult to quantify dation of the C layer occurs at a temperature lower except that for the higher temperature heat treatments than that at which the fibre oxidises. Two reactions are there is a slight increase in thermal diffusivity after the possible heat treatment at 1. 200oC of some 4% after 30 h. may be postulated for this C+O2→CO (a) Interfacial effects or more probably (b) Crystallisation of residual glassy phases, and (c) Changes in phase structure of the phases in the 2C+O2→2CO matrix Changes in the interface will affect thermal diffu 3 Irrespective of whether the oxidation of C occurs via sIvIty if any contact resistance is reduced. Although the ne formation of CO or CO2 the mechanism whereby differences in thickness of the interface were noted, the reaction occurs is worth investigation. Oxygen this is difficult to quantify. There is a reduction in the diffusion from the BMAS matrix is unlikely at these glassy phase content which could increase the diffusive low temperatures. The interfaces themselves may act ity. According to Winter et al. [30] crystallisation starts as pipe oxidation channels providing a route for gas at 900-950C with the formation of a magnesium transport from the exposed ends of the fibres. aluminosilicate of high-quartz type structure. Osum The reduction in oxygen content of the interfacial lite starts to crystallise at around 970C. The other layer suggests that this mechanism may be a contrib- phases in the matrix begin to crystallise at around utory factor although not necessarily the only one. The 1,000C. This recrystallisation in the residual glasses is largest change in diffusivity occurs after the 700C heat possible at around those temperatures. However, treatment. It is questioned why degradation is most crystallisation of those phases is a time-dependent rapid at 700C. The rate of oxidation of C increases process and with increasing heat treatment, the with temperature and the fact that degradation is less percentage of the crystallised phases is increased 2 Springer
The BMAS/SiC (Tyranno) fibre composite as received had a carbon-rich interface typically 20 nm thick, a typical thickness also noted by Plucknett et al. [14]. The thickness of the carbon-rich layer does however depend on hot pressing temperature and time [28, 29] and thicknesses as high as 250 nm have been noted [10]. The carbon exists as an amorphous phase at lower manufacturing temperatures but if manufacturing temperatures are higher than 1,250 C graphitic carbon formation has been seen [10, 11]. The EDS analyses of the interfacial layer shows it to consist of C (~42%), O (~13%) and Si (23%) which suggests a mixture of C and SiO2. Interestingly for the BMAS composite the interfacial layer also contained some Ti which can only have come from the Tyranno fibre suggesting that the formation of the interfacial layer had occurred by oxidation of the fibre via one of the two reactions proposed by Eq. (1) or (2). The most dramatic results during the heat treatments are for the heat treatments in the temperature range 700–900 C where significant decreases in diffusivity and thermal expansion coefficient are noted. TEM micrographs clearly show the development of a gap at the interface after heat treatments in this temperature range. This is clearly due to the oxidation of the carbon to leave a silicon oxide-rich region forming the bridges between matrix and fibre. Plucknett et al. [14] reported similar behaviour and stated that the bridges they observed were silica. The degradation of the C layer occurs at a temperature lower than that at which the fibre oxidises. Two reactions are possible C þ O2 ! CO2 ð3Þ or more probably 2C þ O2 ! 2CO ð4Þ Irrespective of whether the oxidation of C occurs via the formation of CO or CO2 the mechanism whereby the reaction occurs is worth investigation. Oxygen diffusion from the BMAS matrix is unlikely at these low temperatures. The interfaces themselves may act as pipe oxidation channels providing a route for gas transport from the exposed ends of the fibres. The reduction in oxygen content of the interfacial layer suggests that this mechanism may be a contributory factor although not necessarily the only one. The largest change in diffusivity occurs after the 700 C heat treatment. It is questioned why degradation is most rapid at 700 C. The rate of oxidation of C increases with temperature and the fact that degradation is less rapid at low temperatures could be due to slower kinetics of oxidation. However as the temperature is increased, two effects could contribute to reducing the effect; (a) Any residual glassy phases in the matrix could begin to soften. The extent to which these may flow to fill gaps depends on the viscosity. As temperature is increased, viscosity is reduced and viscous flow can occur. (b) As the temperature is increased further oxidation of the silicon carbide could occur to produce more SiO2 and further carbon (pipe line diffusion). The results examined for the anneals at the higher temperatures (>1,000 C) revealed that the thermal diffusivity is changed by very little as a result of these heat treatments. The thickness of the interface however is much thicker than that of the as-received composites, amounting to some ~45 nm after the 30 h heat treatment at 1,200 C. Examination by EDS of various regions of the interface after a 30 h heat treatment reveals it to contain Si, O and C possibly as a result of reaction. The interfacial layer contained Ti all the way through thus supporting the view that the reaction products were the result of oxidation of the SiC fibre. Interestingly the presence of Ba, Al, Mg was also noted near the matrix interface suggesting some diffusion from the matrix also. The effect of these heat treatments on the diffusivity is difficult to quantify except that for the higher temperature heat treatments there is a slight increase in thermal diffusivity after the heat treatment at 1,200 C of some 4% after 30 h. Various reasons may be postulated for this, these are (a) Interfacial effects, (b) Crystallisation of residual glassy phases, and (c) Changes in phase structure of the phases in the matrix. Changes in the interface will affect thermal diffusivity if any contact resistance is reduced. Although the differences in thickness of the interface were noted, this is difficult to quantify. There is a reduction in the glassy phase content which could increase the diffusivity. According to Winter et al. [30] crystallisation starts at 900–950 C with the formation of a magnesium aluminosilicate of high-quartz type structure. Osumilite starts to crystallise at around 970 C. The other phases in the matrix begin to crystallise at around 1,000 C. This recrystallisation in the residual glasses is possible at around those temperatures. However, crystallisation of those phases is a time-dependent process and with increasing heat treatment, the percentage of the crystallised phases is increased. 123 770 J Mater Sci (2007) 42:763–771
J Mater Sci(2007)42:763-771 Barium osumilite and celsian both start to crystallise at Acknowledgements The authors would like to thank to NPI around 970C, whereas the cordierite start to crystal (National Physical Laboratory) lise at about 1,020C and the highest percentage of with SEM and TEm studies. r yilmaz would also like to thank to crystallisation occurs at 1, 100C. Sakarya University for financial support. Conclusions References The following conclusions can be drawn from this 1. Prewo PM, Brennan JJ, Layden GK(1986)Am Ceram Soc 2. Brennan JJ, Chyung K, Taylor MP(1986) USA Patent Ne 1. The interface is one of the key factors which affect 4589900,May20 thermal properties such as diffusivity of the 3. Johnson LF, Hasselman DPH, Chung KJ(1987)JAm Ceram Soc 70: C135 Doper RF, Chy yung K(1987)J Mater Sci 22: 3148 2. Thermal properties were determined before and 5. Bemson PM, Spear KE, Pontano CG(1988)Ceram Eng Sci after heat treatment at various temperatures and Proc 9: 63 6. Chaim R, Heuer AH (1991)J Am Ceram Soc 74: 1663 mes in air. It has been found that heat treatment 7 Murty VSR, Li J, Lewis MH (1989)Ceram Er at the lower temperatures causes a considerable 10:938 degradation in the thermal diffusivity and the 8. Bonney LA, Cooper RF(1990)JAm Ceram Soc 73: 2916 thermal expansion with the greatest affect being 9. Murthy VSR, Phoraoh MW, Lewis MH (1990) Inst Phys shown after 700 oC heat treatment. However Conf Ser No 111. New Mater Appl 185 10. Lewis MH, Murthy VSR(1991)Compos Sci Technol 42: 221 temperatures higher than 900C heat treatments 11. Lewis MH, Daniel AM, Chamberlian A, Pharaoh MV,Cain resulted in retention in the thermal property values MG(1993)J Microsc 169: 109 and sometimes even higher thermal diffusivity 12. Yilmaz R(1998)PhD thesis. UMIST-UK values were obtained 13. Bleay SM, Scott VD(1992)J Mater Sci 27: 825 14. Plucknett KP, Sutherland S, Daniel AM, Cain RL, Taplin 3. TEM analysis showed interfacial structure degra DMR, Lewis MH (1995)J Microsc 177: 251 lation after heat treatments that were carried out at 15. Kumar A, Knowles KM (1996)J Am Ceram Soc 79: 2369 lower temperature(700oC)on glass ceramic matrix 16. Pharaoh MW, Daniel AM, Lewis MH(1993)J Mater Sci composites caused by the removal of carbon and the 17. Hasselman DPT(1988)Therm Conduct 19:383 occurrence of gaps between fibre and matrix and 18. Bhatt H, Donaldson KY, Hasselman DPH, Bhatt RT (1992) isolated silicon rich bridges linking the two J Mater Sci 27: 6653 4.TEM analysis after higher heat treatment temper- 19. Bhatt H, Donaldson KY, Hasselman DPH, Bhatt RT(1990) atures such as 1, 200C showed that the interfacial 20. Hasselman DPH. Venkatesawaron A, Yu M(1991)J Mater reaction layer was much thicker which sometimes Sci lett 17: 1037 resulted in higher values in the thermal diffusivity. 21. Hasselman DPH. Venkatesawaron A, Tawil H(1991)JAm 5. seM studies show that at lower heat treatment temperature(700C)residual glass in the matrix 22. Tawil H, Bersen LD, Baskaran J, Hasselman DPH (1985)J Mater Sci 20: 3201 migrated to the voids in particular to interfaces 23. Bhatt H Donaldson KY, Hasselman DPH, Bhatt RT(1992) after degradation of carbon layer at interfaces. J Am Ceram Soc 75: 334 6. SEM studies indicated that crystallisation in the 24 Hasselmann DPH, Johnson T(1987)J Compos Mater 21:508 residual glass, which resulted in an increase in 25. Parker WJ, Jenkins RJ, Butler CP, Abbot GL (1960)J Appl Phys32:926 thermal properties of the composites, occurred 26. Taylor R(1980)J Phys E: Sci Instrum 13: 1193 fter heat treatment at temperature(>1, 100C) 27. Le Strat E, Lancin M, Fourches-Coulon M, Marhic c(1998) Philos Mag A 78: 189 7. As a general conclusion it may be said that the 28. Brennan J. Prewo KM(1982)J Mater Sci 17: 2371 thermal diffusivity can be used as a qualitative non- 29. Qui G, Spear KE, Pontano CG(1993)Mater Sci Eng A destructive technique to determine the integrity of 16245 te fibre/matrix interface and to monitor micro- 30. Winter W, bogdonaw C, Muller G, Panshorst W(199 structural changes occurring in the fibres, matrix Glass Technol Ber 66: 109 and interface during manufacturing or service
Barium osumilite and celsian both start to crystallise at around 970 C, whereas the cordierite start to crystallise at about 1,020 C and the highest percentage of crystallisation occurs at 1,100 C. Conclusions The following conclusions can be drawn from this study: 1. The interface is one of the key factors which affect thermal properties such as diffusivity of the composites. 2. Thermal properties were determined before and after heat treatment at various temperatures and times in air. It has been found that heat treatment at the lower temperatures causes a considerable degradation in the thermal diffusivity and the thermal expansion with the greatest affect being shown after 700 C heat treatment. However temperatures higher than 900 C heat treatments resulted in retention in the thermal property values and sometimes even higher thermal diffusivity values were obtained. 3. TEM analysis showed interfacial structure degradation after heat treatments that were carried out at lower temperature (700 C) on glass ceramic matrix composites caused by the removal of carbon and the occurrence of gaps between fibre and matrix and isolated silicon rich bridges linking the two. 4. TEM analysis after higher heat treatment temperatures such as 1,200 C showed that the interfacial reaction layer was much thicker which sometimes resulted in higher values in the thermal diffusivity. 5. SEM studies show that at lower heat treatment temperature (700 C) residual glass in the matrix migrated to the voids in particular to interfaces after degradation of carbon layer at interfaces. 6. SEM studies indicated that crystallisation in the residual glass, which resulted in an increase in thermal properties of the composites, occurred after heat treatment at temperature (>1,100 C) 7. As a general conclusion it may be said that the thermal diffusivity can be used as a qualitative nondestructive technique to determine the integrity of the fibre/matrix interface and to monitor microstructural changes occurring in the fibres, matrix and interface during manufacturing or service. Acknowledgements The authors would like to thank to NPL (National Physical Laboratory) provision of samples of composites and Mr. I. Brough and Mr. P. Kenway for assistance with SEM and TEM studies. R. Yilmaz would also like to thank to Sakarya University for financial support. References 1. Prewo PM, Brennan JJ, Layden GK (1986) Am Ceram Soc Bull 65:305 2. Brennan JJ, Chyung K, Taylor MP (1986) USA Patent No 4589900, May 20 3. Johnson LF, Hasselman DPH, Chung KJ (1987) J Am Ceram Soc 70:C135 4. Cooper RF, Chyung K (1987) J Mater Sci 22:3148 5. Bemson PM, Spear KE, Pontano CG (1988) Ceram Eng Sci Proc 9:63 6. Chaim R, Heuer AH (1991) J Am Ceram Soc 74:1663 7. Murty VSR, Li J, Lewis MH (1989) Ceram Eng Sci Proc 10:938 8. Bonney LA, Cooper RF (1990) J Am Ceram Soc 73:2916 9. Murthy VSR, Phoraoh MW, Lewis MH (1990) Inst Phys Conf Ser No 111. New Mater Appl 185 10. Lewis MH, Murthy VSR (1991) Compos Sci Technol 42:221 11. Lewis MH, Daniel AM, Chamberlian A, Pharaoh MV, Cain MG (1993) J Microsc 169:109 12. Yilmaz R (1998) PhD thesis. UMIST-UK 13. Bleay SM, Scott VD (1992) J Mater Sci 27:825 14. Plucknett KP, Sutherland S, Daniel AM, Cain RL, Taplin DMR, Lewis MH (1995) J Microsc 177:251 15. Kumar A, Knowles KM (1996) J Am Ceram Soc 79:2369 16. Pharaoh MW, Daniel AM, Lewis MH (1993) J Mater Sci Lett 12:998 17. Hasselman DPT (1988) Therm Conduct 19:383 18. Bhatt H, Donaldson KY, Hasselman DPH, Bhatt RT (1992) J Mater Sci 27:6653 19. Bhatt H, Donaldson KY, Hasselman DPH, Bhatt RT (1990) J Am Ceram Soc 73:312 20. Hasselman DPH, Venkatesawaron A, Yu M (1991) J Mater Sci Lett 17:1037 21. Hasselman DPH, Venkatesawaron A, Tawil H (1991) J Am Ceram Soc 74:1631 22. Tawil H, Bersen LD, Baskaran J, Hasselman DPH (1985) J Mater Sci 20:3201 23. Bhatt H, Donaldson KY, Hasselman DPH, Bhatt RT (1992) J Am Ceram Soc 75:334 24. Hasselmann DPH, Johnson T (1987) J Compos Mater 21:508 25. Parker WJ, Jenkins RJ, Butler CP, Abbot GL (1960) J Appl Phys 32:926 26. Taylor R (1980) J Phys E: Sci Instrum 13:1193 27. Le Strat E, Lancin M, Fourches-Coulon M, Marhic C (1998) Philos Mag A 78:189 28. Brennan JJ, Prewo KM (1982) J Mater Sci 17:2371 29. Qui G, Spear KE, Pontano CG (1993) Mater Sci Eng A 162:45 30. Winter W, Bogdonaw C, Muller G, Panshorst W (1993) Glass Technol Ber 66:109 123 J Mater Sci (2007) 42:763–771 771
Copyright of Journal of Materials Science is the property of Springer Science Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holders express written permission. However, users may print, download or email articles for individual use