《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料 Processing and characterisation of model optomechanical composites in the system sapphire fibre/borosilicate glass matrix

Materials P rocessing Technology ELSEVIER of Materials Processing Technology 16 Processing and characterisation of model optomechanical composites in the system sapphire fibre/borosilicate glass matrix A.R. Boccaccini a,*. D. Acevedo a,A F. Dericioglu. C. Jana b National Institute for Materials Science, 1-2-1 Sengen, Tsukuba-shi, Ibaraki 305-0047, Japan eSchott-Jenaer Glas GmbH. Otto Schott StraBe 13. D-07745 Jena germany Received 8 July 2004; accepted 11 March 2005 Abstract Optomechanical composites based on the system sapphire fibre/borosilicate glass matrix were fabricated and characterised. Different techniques of fabrication were used: composites with randomly orientated chopped sapphire fibres were produced by powder technology and pressureless sintering, whilst unidirectionally oriented fibre composites were fabricated by hot-pressing as well as by sandwiching glass slides and arrays of parallel fibres followed by heat-treatment. Pressureless sintered samples were poorly densified and were opaque. Hot-pressed and"sandwich structure"composites were dense and exhibited strong interfaces between fibres and matrix. Only the "sandwich structure composites were transparent and showed significant light transmittance in the visible wavelength range, only 20% lower than that of the unreinforced matrix. Due to the strong matrix/fibre interface limited fibre pull-out during composite fracture was observed. The fabricate B nsparent composites represent an improved version of the traditional material wired glass. They are candidate materials for applications in igh performance fire and impact resistant windows requiring high impact strength and avoidance of fragmentation upon fracture O 2005 Elsevier B. V. All rights reserved Keywords: Composite materials; Glass matrix; Hot-pressing; Sintering: Ceramic fibres; Mechanical properties; Optical properties 1. Introduction been carried out to understand the toughening mechanisms induced by the presence of the reinforcements and to inves- Silicate glasses have extremely low fracture toughness val- tigate the parameters that lead to satisfactory mechanical ues,which leads to the well known low reliability of these behaviour of glass and glass-ceramic matrix composites materials in load-bearing applications. Hence, strengthening [5-l1l and toughening of silicate glasses are required if these mate Limited previous research has been carried out focusing on rials are to find wider use in structural applications [1] improving simultaneously mechanical and functional prop- Forming composites by incorporation of reinforcing ele- erties of glass matrix composites. These include composites ments in glass and glass-ceramic matrices is a very effective for electric and electronic applications(e.g. induction heat approach to improve the mechanical properties of glasses, ing equipment, microwave components, electronic package including fracture strength, fracture toughness as well as substrates, connectors), high-temperature applications(e.g thermal shock and impact resistance Reinforcements thermal insulators, jet engine thermocouples), dimensional most commonly used are in the form of ceramic whiskers, stability applications(e.g. mirror supports for telescopes )and platelets, particulates or fibres [2-5]. Numerous studies have optical applications(e. g. impact resistant windows, struc ures for micro fluidics)[4, 12-20]. In particular, compos ding author. Tel 075946731;fax:+442075843194 ites exhibiting favourable optical and mechanical properties E-mail address: a boccaccini @imperial ac uk(A R. Boccaccini) are called [ 18], which are the ss: GEMPPM de lyon,69621 Villeurbanne, France. focus of the present work 0924-0136/S- see front matter O 2005 Elsevier B V. All rights reserved doi: 10. 1016/j- imatprotec. 2005.03.011

Journal of Materials Processing Technology 169 (2005) 270–280 Processing and characterisation of model optomechanical composites in the system sapphire fibre/borosilicate glass matrix A.R. Boccaccini a,∗, D. Acevedo a,1, A.F. Dericioglu b, C. Jana c a Department of Materials, Imperial College London, Prince Consort Road, London SW7 2BP, UK b National Institute for Materials Science, 1-2-1 Sengen, Tsukuba-shi, Ibaraki 305-0047, Japan c Schott-Jenaer Glas GmbH, Otto Schott Straße 13, D-07745 Jena, Germany Received 8 July 2004; accepted 11 March 2005 Abstract Optomechanical composites based on the system sapphire fibre/borosilicate glass matrix were fabricated and characterised. Different techniques of fabrication were used: composites with randomly orientated chopped sapphire fibres were produced by powder technology and pressureless sintering, whilst unidirectionally oriented fibre composites were fabricated by hot-pressing as well as by sandwiching glass slides and arrays of parallel fibres followed by heat-treatment. Pressureless sintered samples were poorly densified and were opaque. Hot-pressed and “sandwich structure” composites were dense and exhibited strong interfaces between fibres and matrix. Only the “sandwich structure” composites were transparent and showed significant light transmittance in the visible wavelength range, only 20% lower than that of the unreinforced matrix. Due to the strong matrix/fibre interface limited fibre pull-out during composite fracture was observed. The fabricated transparent composites represent an improved version of the traditional material wired glass. They are candidate materials for applications in high performance fire and impact resistant windows requiring high impact strength and avoidance of fragmentation upon fracture. © 2005 Elsevier B.V. All rights reserved. Keywords: Composite materials; Glass matrix; Hot-pressing; Sintering; Ceramic fibres; Mechanical properties; Optical properties 1. Introduction Silicate glasses have extremely low fracture toughness val￾ues, which leads to the well known low reliability of these materials in load-bearing applications. Hence, strengthening and toughening of silicate glasses are required if these mate￾rials are to find wider use in structural applications [1]. Forming composites by incorporation of reinforcing ele￾ments in glass and glass-ceramic matrices is a very effective approach to improve the mechanical properties of glasses, including fracture strength, fracture toughness as well as thermal shock and impact resistance [2–5]. Reinforcements most commonly used are in the form of ceramic whiskers, platelets, particulates or fibres [2–5]. Numerous studies have ∗ Corresponding author. Tel.: +44 207 594 6731; fax: +44 207 584 3194. E-mail address: a.boccaccini@imperial.ac.uk (A.R. Boccaccini). 1 Present address: GEMPPM, INSA de Lyon, 69621 Villeurbanne, France. been carried out to understand the toughening mechanisms induced by the presence of the reinforcements and to inves￾tigate the parameters that lead to satisfactory mechanical behaviour of glass and glass-ceramic matrix composites [5–11]. Limited previous research has been carried out focusing on improving simultaneously mechanical and functional prop￾erties of glass matrix composites. These include composites for electric and electronic applications (e.g. induction heat￾ing equipment, microwave components, electronic package substrates, connectors), high-temperature applications (e.g. thermal insulators, jet engine thermocouples), dimensional stability applications (e.g. mirror supports for telescopes) and optical applications (e.g. impact resistant windows, struc￾tures for micro fluidics) [4,12–20]. In particular, compos￾ites exhibiting favourable optical and mechanical properties are called «optomechanical composites» [18], which are the focus of the present work. 0924-0136/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2005.03.011

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 The main difficulty in the development of optomechanical its high thermal capability as well as considerable corrosion composites is the requirement of being able to improve the and thermal shock resistance [25]. In fact, borosilicate glass mechanical and optical properties simultaneously [18-24]. has been widely used in the past as the matrix for SiC and Indeed, most glass matrix composite materials developed to carbon fibre reinforced composites for structural application date are not optically transparent, or even translucent, because [4, 26]. Moreover, borosilicate glass matrices reinforced by of the type of reinforcements used(e.g. SiC or carbon fibres) a-Al2O3 in the form of particles, platelets and fibres have [2-5], and therefore, they cannot be considered to be suitable been the matter of numerous previous investigations due to materials for optomechanical applications. Hence, means of the favourable thermal expansion mismatch between alu- improving the mechanical properties of glasses without sig- mina and borosilicate glass composition [4, 6, 27]. Additional nificantly degrading their optical transparency need to be advantages of borosilicate glass are its optical propertie further investigated and relatively low dielectric constant [25, 28]. Sapphire The selection of appropriate fibres and matrices for fibres were selected because they exhibit outstanding high ptomechanical composites is a complex matter because temperature stability, high chemical durability and excellent numerous factors have to be considered. The main require- mechanical properties[29, 30]. Single crystal sapphire fibres ment for the fibres is that they should have a higher ther- have been used in previous studies to reinforce ceramic and mal stability than the glass matrix because of the usually glass matrices for high-temperature applications [30-331 high temperatures needed for matrix densification [18, 22]. In those studies however no special care was placed on Furthermore, matching the fibre and matrix thermal expan- the optical property(transparency)of composites, except sion coefficients is necessary in order to avoid large residual for some model systems fabricated for academic purposes stresses upon cooling from the fabrication temperature. How- [21]. Thus, to the authors'knowledge, this is the first work ever, the development of compressive residual stresses in the on the system sapphire fibre/borosilicate glass matrix with matrix by having fibres with thermal expansion coefficient the specific aim of producing transparent composites for higher than that of the matrix may be also favourable [ 18, 22]. optomechanical applications Finally, the strength of bonding at the interface between fibres and matrix is an important parameter since it has a large influ- ence on the mechanical behaviour of the final composites, as 2. Materials and experimental procedure it is the case in all brittle matrix composites [2-4]. In addi- tion to these requirements, matching of the fibre and matrix 2.1. Materials refractive indices is also necessary to avoid (or minimise) light scattering, and thus to obtain a transparent or at least Borosilicate glass was selected as the matrix material and 4]. Another possible option to it was used in two different forms: i) powder of mean particle obtain transparent composites is to include optical windows size <40 um(Duran", Schott Glas, Mainz, Germany)and ii) by a relatively large spacing of the reinforcing fibres in the glass plates of thickness 1. 1 mm(Borofloat33, Schott Jenaer matrix[18], in a similar way as in the traditional material Glas, Jena, Germany ) The properties of the glass are summa- wired glass. Recently, we have fabricated oxide-fibre rein- rized in Table 1[34]. The chemical composition of Duran forced glass matrix composites with high light transmittance glass is(in wt%)[34]: 81SiO2, 13B203, 4(Na20+K2O) (only 30% lower than that of the matrix)using the"optical 2Al2O3, which can be considered to be identical to that of windowconcept, which is based on the presence of relatively Borofloat33 large transparent matrix regions surrounded by the reinforc The reinforcement chosen was sapphire fibre of optical ing fibres [22]. Moreover, Dericiogluet al [18, 24] fabricated quality with nominal diameter 150 um(Saphikon",Laser minicomposite reinforced borosilicate glass matrix optome- Components UK, Ltd ) The fibres were received in length of chanical composites with low volume fraction of reinforce- I m and were cut manually to appropriate lengths for compos. ment exhibiting light transmittance higher than 80% of the ites fabrication by using metallic scissors. For all composites transmittance of the matrix, showing that even if the fibres fibres were used in the as-received condition. Sapphire fibres incorporated are opaque, it is possible to achieve considerable were selected because they exhibit outstanding thermome- optical transparency chanical properties [29-33]. This fibre is a monocrystal of A few experimental investigations aiming at producing a-Al2O3 of very high quality exhibiting high strength and optomechanical composites with technical applications have hardness. Additionally, because absence of grain boundaries been carried out, specially in Japan [17-20, 24, in Germany [15,16, 23] and in the UK [22], yet research in this field remains still rather limited. which has thus motivated the Properties of the borosilicate glass durAN[25, 34] present experimental study Density(gcm-3) The objective of this work is to explore and optimise dif- Tensile strength(MPa) 60 ferent methods to produce optically transparent borosilicate Elastic modulus(GPa) 64 glass matrix composites reinforced by single crystal Al2O Coefficient of thermal expansion(C) Refractive index 1.473 (sapphire)fibres. Borosilicate glass was chosen because of

A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 271 The main difficulty in the development of optomechanical composites is the requirement of being able to improve the mechanical and optical properties simultaneously [18–24]. Indeed, most glass matrix composite materials developed to date are not optically transparent, or even translucent, because of the type of reinforcements used (e.g. SiC or carbon fibres) [2–5], and therefore, they cannot be considered to be suitable materials for optomechanical applications. Hence, means of improving the mechanical properties of glasses without sig￾nificantly degrading their optical transparency need to be further investigated. The selection of appropriate fibres and matrices for optomechanical composites is a complex matter because numerous factors have to be considered. The main require￾ment for the fibres is that they should have a higher ther￾mal stability than the glass matrix because of the usually high temperatures needed for matrix densification [18,22]. Furthermore, matching the fibre and matrix thermal expan￾sion coefficients is necessary in order to avoid large residual stresses upon cooling from the fabrication temperature. How￾ever, the development of compressive residual stresses in the matrix by having fibres with thermal expansion coefficient higher than that of the matrix may be also favourable [18,22]. Finally, the strength of bonding at the interface between fibres and matrix is an important parameter since it has a large influ￾ence on the mechanical behaviour of the final composites, as it is the case in all brittle matrix composites [2–4]. In addi￾tion to these requirements, matching of the fibre and matrix refractive indices is also necessary to avoid (or minimise) light scattering, and thus to obtain a transparent or at least a translucent material [15–24]. Another possible option to obtain transparent composites is to include optical windows by a relatively large spacing of the reinforcing fibres in the matrix [18], in a similar way as in the traditional material wired glass. Recently, we have fabricated oxide-fibre rein￾forced glass matrix composites with high light transmittance (only 30% lower than that of the matrix) using the “optical window” concept, which is based on the presence of relatively large transparent matrix regions surrounded by the reinforc￾ing fibres [22]. Moreover, Dericioglu et al. [18,24] fabricated minicomposite reinforced borosilicate glass matrix optome￾chanical composites with low volume fraction of reinforce￾ment exhibiting light transmittance higher than 80% of the transmittance of the matrix, showing that even if the fibres incorporated are opaque, it is possible to achieve considerable optical transparency. A few experimental investigations aiming at producing optomechanical composites with technical applications have been carried out, specially in Japan [17–20,24], in Germany [15,16,23] and in the UK [22], yet research in this field remains still rather limited, which has thus motivated the present experimental study. The objective of this work is to explore and optimise dif￾ferent methods to produce optically transparent borosilicate glass matrix composites reinforced by single crystal Al2O3 (sapphire) fibres. Borosilicate glass was chosen because of its high thermal capability as well as considerable corrosion and thermal shock resistance [25]. In fact, borosilicate glass has been widely used in the past as the matrix for SiC and carbon fibre reinforced composites for structural applications [4,26]. Moreover, borosilicate glass matrices reinforced by
-Al2O3 in the form of particles, platelets and fibres have been the matter of numerous previous investigations due to the favourable thermal expansion mismatch between alu￾mina and borosilicate glass composition [4,6,27]. Additional advantages of borosilicate glass are its optical properties and relatively low dielectric constant [25,28]. Sapphire fibres were selected because they exhibit outstanding high temperature stability, high chemical durability and excellent mechanical properties [29,30]. Single crystal sapphire fibres have been used in previous studies to reinforce ceramic and glass matrices for high-temperature applications [30–33]. In those studies however no special care was placed on the optical property (transparency) of composites, except for some model systems fabricated for academic purposes [21]. Thus, to the authors’ knowledge, this is the first work on the system sapphire fibre/borosilicate glass matrix with the specific aim of producing transparent composites for optomechanical applications. 2. Materials and experimental procedure 2.1. Materials Borosilicate glass was selected as the matrix material and it was used in two different forms: i) powder of mean particle size <40m (Duran®, Schott Glas, Mainz, Germany) and ii) glass plates of thickness 1.1 mm (Borofloat® 33, Schott Jenaer Glas, Jena, Germany). The properties of the glass are summa￾rized in Table 1 [34]. The chemical composition of Duran® glass is (in wt%) [34]: 81SiO2, 13B2O3, 4(Na2O+K2O), 2Al2O3, which can be considered to be identical to that of Borofloat® 33. The reinforcement chosen was sapphire fibre of optical quality with nominal diameter 150 m (Saphikon®, Laser Components UK, Ltd.). The fibres were received in length of 1 m and were cut manually to appropriate lengths for compos￾ites fabrication by using metallic scissors. For all composites, fibres were used in the as-received condition. Sapphire fibres were selected because they exhibit outstanding thermome￾chanical properties [29–33]. This fibre is a monocrystal of
-Al2O3 of very high quality exhibiting high strength and hardness. Additionally, because absence of grain boundaries, Table 1 Properties of the borosilicate glass DURAN® [25,34] Density (g cm−3) 2.23 Tensile strength (MPa) 60 Elastic modulus (GPa) 64 Coefficient of thermal expansion (◦C−1) 3.3 × 10−6 Refractive index 1.473

272 A.R. Boccaccini er al. Joumal of Materials Processing Technology 169(2005)270-280 Table 2 erties of the Saphikon"fibres 35 Diameter(um) ~10mm Tensile strength(GPa) 2.1-3.4 Elastic modulus(GPa) 386-435 Melting point(°C) Coefficient of thermal expansion( K-) 79-8.8×10-6 Uniaxial refractive index 1.760-1.768 there is no light scattering within the sapphire fibres. Table 2 summarizes the main properties of the Saphikon"fibres used 351, and Fig. I shows a scanning electron microscopy (SEM) image of the fibre. This micrograph confirms that the fibres ive a circular cross-section. moreover. it shows that Fig. 2. Arrangement of Saphikon fibres in(a) hot-pressed and(b)"sand- fibres have a very smooth surface, which will influence the wich structure"composites, in a volume fraction of approximately 5% possible toughening mechanisms acting in the composites a smooth interface should lead to greater average pull-out platelet reinforced glass matrix composites where the same lengths, and thus to higher fracture toughness provided there borosilicate glass matrix was used [37] is optimal bonding strength between the fibre and the matrix [36] 2.2.2. Hot-pressing A custom-made vacuum hot-press described in previous 2.2. Preparation of the composites works [38 was used. Samples made of pure glass matrix and fibre reinforced composites were fabricated. The average 2.2 Pressureless sintering fibre length was about 10 mm. The fibres were arranged par A mixture of borosilicate glass powder and 5% in volume allel to each other on a slightly pressed thin layer of powder of chopped sapphire fibres with a length of - l mm was pre- in a carbon die. They were separated an average distance of pared. Fibres were cut using scissors to the required length, I mm, as shown in Fig. 2(a). Subsequently, a second layer and matrix powder and fibres were mixed in dry conditions of powder was added above the fibres and the composites in a rotary mixer for Ih were hot-pressed The mixture was pressed into cylindrical samples of 8mm The volume fraction of fibres in the rectangular samples diameter in a die at room temperature by application of a which were cut out of the hot-pressed disc, as shown in ompacting pressure of about 100 MPa for 2 min No binder Fig. 2(a), was approximately 5%. The parameters used for was used in this operation. The pellets were then sintered in the fabrication of the samples were heating rate 100.Ch-I an electric furnace at 750"C for 2h in normal atmosphere. holding temperature 750 C, holding time I h, applied pres- to cool down in the furnace. The sintering temperature was chosen on the basis of previous investigations on alumina 2.2.3. "Sandwich structure"method This is a simple pressureless method for composite fab- rication introduced recently [22]. The method consists of sandwiching the reinforcing fibres between two glass plates, as shown in Fig. 2(b), and then subjecting the"sandwich structure to a heat treatment to consolidate the composite by exploiting viscous flow of the glass. The as-received plates of borosilicate glass( Borofloat 33)were cut by means ofa dia- mond tip to the desired dimensions(about 2.5 cm x 1. 5 cm) The average length of the fibres was 8 mm. The same dispo- sition as in the hot-pressed samples was used: fibres wer arranged parallel to each other, and the average distance between two fibres was about I mm(see Fig. 2(b)) he heat-treatment consisted of two main steps. In the ⊙AQ1991sEI first step, the glass plates were heated under a high vacuum to clean the surface by degassing. Subsequently, the sandwich structure was subjected to a second heat-treatment. The heat- Fig. 1. Scanning electron microscopy (SEM) image of a Saphikon" fibre ing rate was 100oCh-, the holding temperature was varied used in the present work. between 750 and 775 C the holding time was between 2.5

272 A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 Table 2 Properties of the Saphikon® fibres [35] Density (g cm−3) 3.99 Diameter (m) 150 Tensile strength (GPa) 2.1–3.4 Elastic modulus (GPa) 386–435 Melting point (◦C) 2053 Coefficient of thermal expansion (K−1) 7.9–8.8 × 10−6 Uniaxial refractive index 1.760–1.768 there is no light scattering within the sapphire fibres. Table 2 summarizes the main properties of the Saphikon® fibres used [35], and Fig. 1 shows a scanning electron microscopy (SEM) image of the fibre. This micrograph confirms that the fibres have a circular cross-section. Moreover, it shows that the fibres have a very smooth surface, which will influence the possible toughening mechanisms acting in the composites: a smooth interface should lead to greater average pull-out lengths, and thus to higher fracture toughness provided there is optimal bonding strength between the fibre and the matrix [36]. 2.2. Preparation of the composites 2.2.1. Pressureless sintering A mixture of borosilicate glass powder and 5% in volume of chopped sapphire fibres with a length of ∼1 mm was pre￾pared. Fibres were cut using scissors to the required length, and matrix powder and fibres were mixed in dry conditions in a rotary mixer for 1 h. The mixture was pressed into cylindrical samples of 8 mm diameter in a die at room temperature by application of a compacting pressure of about 100 MPa for 2 min. No binder was used in this operation. The pellets were then sintered in an electric furnace at 750 ◦C for 2 h in normal atmosphere. The heating rate used was 5 ◦C min−1. Samples were left to cool down in the furnace. The sintering temperature was chosen on the basis of previous investigations on alumina Fig. 1. Scanning electron microscopy (SEM) image of a Saphikon® fibre used in the present work. Fig. 2. Arrangement of Saphikon® fibres in (a) hot-pressed and (b) “sand￾wich structure” composites, in a volume fraction of approximately 5%. platelet reinforced glass matrix composites where the same borosilicate glass matrix was used [37]. 2.2.2. Hot-pressing A custom-made vacuum hot-press described in previous works [38] was used. Samples made of pure glass matrix and fibre reinforced composites were fabricated. The average fibre length was about 10 mm. The fibres were arranged par￾allel to each other on a slightly pressed thin layer of powder in a carbon die. They were separated an average distance of ∼1 mm, as shown in Fig. 2(a). Subsequently, a second layer of powder was added above the fibres and the composites were hot-pressed. The volume fraction of fibres in the rectangular samples, which were cut out of the hot-pressed disc, as shown in Fig. 2(a), was approximately 5%. The parameters used for the fabrication of the samples were: heating rate 100 ◦C h−1, holding temperature 750 ◦C, holding time 1 h, applied pres￾sure 10 MPa and cooling rate 100 ◦C min−1. 2.2.3. “Sandwich structure” method This is a simple pressureless method for composite fab￾rication introduced recently [22]. The method consists of sandwiching the reinforcing fibres between two glass plates, as shown in Fig. 2(b), and then subjecting the “sandwich structure” to a heat treatment to consolidate the composite by exploiting viscous flow of the glass. The as-received plates of borosilicate glass (Borofloat® 33) were cut by means of a dia￾mond tip to the desired dimensions (about 2.5 cm × 1.5 cm). The average length of the fibres was 8 mm. The same dispo￾sition as in the hot-pressed samples was used: fibres were arranged parallel to each other, and the average distance between two fibres was about 1 mm (see Fig. 2(b)). The heat-treatment consisted of two main steps. In the first step, the glass plates were heated under a high vacuum to clean the surface by degassing. Subsequently, the sandwich structure was subjected to a second heat-treatment. The heat￾ing rate was 100 ◦C h−1, the holding temperature was varied between 750 and 775 ◦C, the holding time was between 2.5

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 273 and 5h and the cooling rate was varied between 100 and 3. Results and discussion 260Ch-. The degassing step included two different treat ments. Firstly, the sample was under a normal atmosphere and 3.1. Microstructural characterisation after completion of half of the holding time, a high vacuum vas produced in order to clean all the impurities that might 3.1.1. Sintered composit have burnt. All samples were produced using the same heat Successful fabrication of fibre reinforced glass matrix ing and cooling rates(100Ch-l)in both steps, the holding composites relies on knowing the relationship between tem- temperature and holding time for the first step were 500oc perature and viscosity of the glass matrix. As it is well-known and 4 h, respectively, and they were 750C and 5 h for the for Duran-type borosilicate glass [34,371, the range of suit second step, respectively able temperatures for sintering glass powder is very narrow between720and780°C). A difference of±20° C in the sin- 2.3. Characterization techniques tering temperature can have a large effect on the densification The density of sintered and hot-pressed samples was deter- mined by the Archimedes'method. The relative density was gation were not translucent and their densification was not calculated by considering the theoretical density of the com- completely achieved; the sintered relative densities were in posites based on the density data for matrix and fibres, given the range 80-83% of theoretical density. Fig. 3(a-c)show in Tables 1 and 2, respectively, and the volume fraction of SEM micrographs of polished cross-sections of a chopped fibres. For microstructural characterisation, samples were fibre reinforced sintered composite at different magnifica cut,then mounted in resin and polished to 1 um finish with tions. Fig. 3(a)shows that the matrix close to the fibres is diamond paste to obtain flat cross sections for SEM. The porous and that larger defects are situated around the fibre, microstructures of selected sintered, hot-pressed and sand- confirming that it has not been possible to obtain a good wich structure samples were examined using conventional densification of the matrix in this region by pressureless sin- SEM(JEOL LV 5610)working in both secondary electron tering. The presence of randomly orientated fibres impedes and backscattered electron modes the perfect flow of the viscous glass during sintering and Sintered samples were analysed using X-ray diffraction causes a high porosity of the matrix in the region close to the (XRD) to detect any crystallisation of the matrix. Sam- fibres. On the contrary, the matrix far away from the fibres ples were crushed to a fine powder, and then analysed with (Fig 3(b ))exhibits high densification with few isolated pores a Philips PW1700 series automated powder diffractometer Fig. 3(c)is a high magnification image of the fibre/matrix using Cu Ka radiation at 40 k V-40 mA with a seconda interface showing that the lack of densification impedes the ono To gain preliminary understanding of the fracture interface behaviour of the composites and qualitative information The poor densification of the pressureless sintered sam about the interaction(bonding) between fibres and matrix ples achieved in the present work is in broad agreement with during the fracture process, fractures surfaces were also previous studies that showed that the presence of rigid inclu- observed by SEM. ions makes the densification of a glass matrix composite a A preliminary assessment of the optical quality of the sam- difficult task and that the density of the composites decreases ples was obtained by placing selected samples at different with increasing volume fraction of rigid inclusions [39-411 heights over a written text on a white surface and assessing Different explanations for this phenomenon have been pro- the text legibility. This qualitatively demonstrated the light posed. An inhomogeneous distribution of inclusions in the transmitting characteristics of the different composites. a powder can lead to a poor matrix particle packing and for digital camera was used to document this behaviour. Fab- mation of agglomerates, and thus the porosity will increase ricated"sandwich structure" composites were also observed around the fibres [39]. Moreover, studies conducted using the by means of a conventional optical microscope same materials but different mixing conditions have shown Light transmittance of the hot-pressed and"sandwich that optimised wet-mixing techniques lead to higher densities ructure"samples(with and without fibres) was measured at as they allow for more homogeneous mixtures[42]. Since the room temperature by a UV-visible spectrophotometer (UV- present glass powder/chopped fibres mixture was prepared in 1601 Shimadzu, Japan ), in the direction perpendicular dry conditions, the last explanation may be applicable to our the fibres axis. Before light transmittance measurements, results samples were cut and polished to obtain the size required Another factor affecting the densification of composites r the measurements. The thickness of the samples was containing rigid inclusions is the development of residual 2.3+0.1 mm, and the light transmittance was measured for stresses as a consequence of different sintering rates of matrix wavelengths between 350 and 800 nm. The light transmit- and inclusions. These stresses may cause sintering damage, ance of the samples was reported as a percentage of the leading to crack-like voids or isolated pores and conse- transmittance of a reference monolithic borosilicate gla quently to poor mechanical properties of the sintered samples slide(Borofloat 33) [39-42]

A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 273 and 5 h and the cooling rate was varied between 100 and 260 ◦C h−1. The degassing step included two different treat￾ments. Firstly, the sample was under a normal atmosphere and after completion of half of the holding time, a high vacuum was produced in order to clean all the impurities that might have burnt. All samples were produced using the same heat￾ing and cooling rates (100 ◦C h−1) in both steps, the holding temperature and holding time for the first step were 500 ◦C and 4 h, respectively, and they were 750 ◦C and 5 h for the second step, respectively. 2.3. Characterization techniques The density of sintered and hot-pressed samples was deter￾mined by the Archimedes’ method. The relative density was calculated by considering the theoretical density of the com￾posites based on the density data for matrix and fibres, given in Tables 1 and 2, respectively, and the volume fraction of fibres. For microstructural characterisation, samples were cut, then mounted in resin and polished to 1 m finish with diamond paste to obtain flat cross sections for SEM. The microstructures of selected sintered, hot-pressed and sand￾wich structure samples were examined using conventional SEM (JEOL LV 5610) working in both secondary electron and backscattered electron modes. Sintered samples were analysed using X-ray diffraction (XRD) to detect any crystallisation of the matrix. Sam￾ples were crushed to a fine powder, and then analysed with a Philips PW1700 series automated powder diffractometer using Cu K
radiation at 40 kV–40 mA with a secondary crystal monochromator. To gain preliminary understanding of the fracture behaviour of the composites and qualitative information about the interaction (bonding) between fibres and matrix during the fracture process, fractures surfaces were also observed by SEM. A preliminary assessment of the optical quality of the sam￾ples was obtained by placing selected samples at different heights over a written text on a white surface and assessing the text legibility. This qualitatively demonstrated the light transmitting characteristics of the different composites. A digital camera was used to document this behaviour. Fab￾ricated “sandwich structure” composites were also observed by means of a conventional optical microscope. Light transmittance of the hot-pressed and “sandwich structure” samples (with and without fibres) was measured at room temperature by a UV–visible spectrophotometer (UV- 1601 Shimadzu, Japan), in the direction perpendicular to the fibres axis. Before light transmittance measurements, samples were cut and polished to obtain the size required for the measurements. The thickness of the samples was 2.3 ± 0.1 mm, and the light transmittance was measured for wavelengths between 350 and 800 nm. The light transmit￾tance of the samples was reported as a percentage of the transmittance of a reference monolithic borosilicate glass slide (Borofloat® 33). 3. Results and discussion 3.1. Microstructural characterisation 3.1.1. Sintered composites Successful fabrication of fibre reinforced glass matrix composites relies on knowing the relationship between tem￾perature and viscosity of the glass matrix. As it is well-known for Duran®-type borosilicate glass [34,37], the range of suit￾able temperatures for sintering glass powder is very narrow (between 720 and 780 ◦C). A difference of ±20 ◦C in the sin￾tering temperature can have a large effect on the densification of the composites. Pressureless sintered samples fabricated in this investi￾gation were not translucent and their densification was not completely achieved; the sintered relative densities were in the range 80–83% of theoretical density. Fig. 3(a–c) show SEM micrographs of polished cross-sections of a chopped fibre reinforced sintered composite at different magnifica￾tions. Fig. 3(a) shows that the matrix close to the fibres is porous and that larger defects are situated around the fibre, confirming that it has not been possible to obtain a good densification of the matrix in this region by pressureless sin￾tering. The presence of randomly orientated fibres impedes the perfect flow of the viscous glass during sintering and causes a high porosity of the matrix in the region close to the fibres. On the contrary, the matrix far away from the fibres (Fig. 3(b)) exhibits high densification with few isolated pores. Fig. 3(c) is a high magnification image of the fibre/matrix interface showing that the lack of densification impedes the complete bonding of fibre and matrix leading to an imperfect interface. The poor densification of the pressureless sintered sam￾ples achieved in the present work is in broad agreement with previous studies that showed that the presence of rigid inclu￾sions makes the densification of a glass matrix composite a difficult task and that the density of the composites decreases with increasing volume fraction of rigid inclusions [39–41]. Different explanations for this phenomenon have been pro￾posed. An inhomogeneous distribution of inclusions in the powder can lead to a poor matrix particle packing and for￾mation of agglomerates, and thus the porosity will increase around the fibres[39]. Moreover, studies conducted using the same materials but different mixing conditions have shown that optimised wet-mixing techniques lead to higher densities as they allow for more homogeneous mixtures[42]. Since the present glass powder/chopped fibres mixture was prepared in dry conditions, the last explanation may be applicable to our results. Another factor affecting the densification of composites containing rigid inclusions is the development of residual stresses as a consequence of different sintering rates of matrix and inclusions. These stresses may cause sintering damage, leading to crack-like voids or isolated pores and conse￾quently to poor mechanical properties of the sintered samples [39–42].

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 matrix Fig 3 SEM micrographs of Sapphire fibre rced pressureless sintered composites at different magnifications:(a) area around a fibre showing porosity and sintering defects, (b) matrix far away from the fibres exhibiting high densification and (c) high magnification image of the fibre/matrix interface showing mperfect bonding The existence of different"kinds" of porosity in the green lar result was obtained for hot-pressed samples. Indeed it has compact has been also proposed to explain the retardation of been proved in previous studies [27, 28, 43] that cristobalite densification in composites containing rigid inclusions [39]. crystallisation could happen in Al2O3/borosilicate glass com Pores can be embedded in the matrix material only, or they can posites with low volume fractions of Al2O3 (<10%)and even be located at the interface between matrix and inclusions. This when sintering temperature is low (in the range 700-800oC) is confirmed in the present composites by the micrographs analysed above( Fig 3(a and c)). It is observed that pores are situated both in the matrix and at the interface between matrix Counts/s and fibres. These different pore types in the initial compact will have different free surface energies, and thus, they will lead to an overall lower driving force for sintering in the composite than in the inclusion-free compact [39] and thus to a less densified composite. However, Fig. 3(b)shows that far away from the fibres, the matrix was very well densified confirming that the parameters used for sintering(time and temperature)were appropriate for this glass Fig. 4 shows the XRD pattern for a Saphikonfibr reinforced sintered composite. The only crystalline phase detected was corundum, which corresponds to the crystalline of the Saphikon fibres used. No cristobalite has crystallised in the matrix, which is a favourable result from the Fig. 4. XRD patte Saphikon" fibre reinforced sintered composite showing corundum as the only cristalline phase, which corresponds to the point of view of the composite mechanical strength. A simi- structure of the Saphikonfibres

274 A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 Fig. 3. SEM micrographs of Sapphire® fibre reinforced pressureless sintered composites at different magnifications: (a) area around a fibre showing porosity and sintering defects, (b) matrix far away from the fibres exhibiting high densification and (c) high magnification image of the fibre/matrix interface showing imperfect bonding. The existence of different “kinds” of porosity in the green compact has been also proposed to explain the retardation of densification in composites containing rigid inclusions [39]. Pores can be embedded in the matrix material only, or they can be located at the interface between matrix and inclusions. This is confirmed in the present composites by the micrographs analysed above (Fig. 3(a and c)). It is observed that pores are situated both in the matrix and at the interface between matrix and fibres. These different pore types in the initial compact will have different free surface energies, and thus, they will lead to an overall lower driving force for sintering in the composite than in the inclusion-free compact [39] and thus to a less densified composite. However, Fig. 3(b) shows that far away from the fibres, the matrix was very well densified, confirming that the parameters used for sintering (time and temperature) were appropriate for this glass. Fig. 4 shows the XRD pattern for a Saphikon® fibre reinforced sintered composite. The only crystalline phase detected was corundum, which corresponds to the crystalline structure of the Saphikon® fibres used. No cristobalite has crystallised in the matrix, which is a favourable result from the point of view of the composite mechanical strength. A simi￾lar result was obtained for hot-pressed samples. Indeed it has been proved in previous studies [27,28,43] that cristobalite crystallisation could happen in Al2O3/borosilicate glass com￾posites with low volume fractions of Al2O3 (<10%) and even when sintering temperature is low (in the range 700–800 ◦C). Fig. 4. XRD pattern of a Saphikon® fibre reinforced sintered composite showing corundum as the only cristalline phase, which corresponds to the structure of the Saphikon® fibres

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 20kV X2591gg从m 20kU 18 2 SE I Fig. 5. SEM image of the polished section of a hot-pressed sample contain- ing Saphikon" fibres unidirectionally aligned, showing high densification but the presence crack-like defects This should have a negative effect on the mechanical prop- erties of the composites because the different thermal expan- sion coefficients of cristobalite and borosilicate glass matrix may lead to microcracking, therefore decreasing the tensile gth of the composites. The lack of cristobalite forma- tion in the present composites confirms the high resistance to crystallisation of the Duran borosilicate composition and supports the choice of this particular glass as the matrix for 1936 BES 3.1.2. Hot-pressed composites Hot-pressed samples without fibres were translucent, while composites containing fibres were opaque, even though Fig. 6. SEM micrographs of polished surfaces of"sandwich structure"com- posites showing(a)some defects(voids), marked by arrows, localised close the densities achieved were very high in both cases, 99.5 and to the fibres after too short processing time and(b) perfect bonded sample 98% of theoretical density, respectively. Cristobalite crys fabricated employing optimised parameters tallisation was not detected in hot-pressed composites. Fig. 5 shows a polished section of a hot-pressed sample containing optimised(775C,3.5h), the two plates of glass were seen 5 vol% unidirectionally aligned Saphikonfibres. Although to be well bonded together and the interface between them the matrix seems to be well densified without the presence of was"invisible" under SEM(Fig. 6(b)) large cavities or pores(compare with Fig. 3(a), pressureless Sandwich structure" samples with and without fibres sintered sample, there are still some defects and crack-like were transparent. There was some contamination on the sur- voids remaining, mainly close to the fibres, which should be faces after heat-treatment, but this disappeared after polish responsible for high light scattering and the loss of trans- ing. The two slides of glass were in all cases very well bonded parency of these samples. Thus, despite the relative high Fig. 7 shows the matrix region between two fibres ("optical density achieved, the parameters used for hot-pressing were window )in a sandwich structure composite observed with not optimal for producing optically transparent or translucent an optical microscope; fibres are also visible Heat-treatment composites at 775C for 3.5 h was found to be the best thermal process for the fabrication of sandwich structure c 3.1.3. Sandwich structure"composites SEM micrographs of polished surfaces of""sandwich 3. 2. Fracture behaviour structure"composites are shown in Fig. 6(a and b). In some regions the interface between fibres and matrix could only Previous research has proved that a-Al2O3/borosilicate be detected by using SEM in backscattered electrons mode glass systems containing a-Al2O3 platelets as reinforce (Fig. 6(b)), indicating an intimate perfect bond. Some defects ment lead to composites with higher fracture strength (voids) were observed in some samples, localised just cl Young's modulus. hardness and fracture toughness than the to the fibres(Fig. 6(a)). These were probably the result of a matrix without reinforcement [6,371. In the present inves- too short processing time. When the parameters used were tigation, fracture behaviour of the composites was anal

A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 275 Fig. 5. SEM image of the polished section of a hot-pressed sample contain￾ing Saphikon® fibres unidirectionally aligned, showing high densification but the presence crack-like defects. This should have a negative effect on the mechanical prop￾erties of the composites because the different thermal expan￾sion coefficients of cristobalite and borosilicate glass matrix may lead to microcracking, therefore decreasing the tensile strength of the composites. The lack of cristobalite forma￾tion in the present composites confirms the high resistance to crystallisation of the Duran® borosilicate composition and supports the choice of this particular glass as the matrix for composites. 3.1.2. Hot-pressed composites Hot-pressed samples without fibres were translucent, while composites containing fibres were opaque, even though the densities achieved were very high in both cases, 99.5 and 98% of theoretical density, respectively. Cristobalite crys￾tallisation was not detected in hot-pressed composites. Fig. 5 shows a polished section of a hot-pressed sample containing 5 vol% unidirectionally aligned Saphikon® fibres. Although the matrix seems to be well densified without the presence of large cavities or pores (compare with Fig. 3(a), pressureless sintered sample), there are still some defects and crack-like voids remaining, mainly close to the fibres, which should be responsible for high light scattering and the loss of trans￾parency of these samples. Thus, despite the relative high density achieved, the parameters used for hot-pressing were not optimal for producing optically transparent or translucent composites. 3.1.3. “Sandwich structure” composites SEM micrographs of polished surfaces of “sandwich structure” composites are shown in Fig. 6(a and b). In some regions the interface between fibres and matrix could only be detected by using SEM in backscattered electrons mode (Fig. 6(b)), indicating an intimate perfect bond. Some defects (voids) were observed in some samples, localised just close to the fibres (Fig. 6(a)). These were probably the result of a too short processing time. When the parameters used were Fig. 6. SEM micrographs of polished surfaces of “sandwich structure” com￾posites showing (a) some defects (voids), marked by arrows, localised close to the fibres after too short processing time and (b) perfect bonded sample fabricated employing optimised parameters. optimised (775 ◦C, 3.5 h), the two plates of glass were seen to be well bonded together and the interface between them was “invisible” under SEM (Fig. 6(b)). “Sandwich structure” samples with and without fibres were transparent. There was some contamination on the sur￾faces after heat-treatment, but this disappeared after polish￾ing. The two slides of glass were in all cases very well bonded. Fig. 7 shows the matrix region between two fibres (“optical window”) in a sandwich structure composite observed with an optical microscope; fibres are also visible. Heat-treatment at 775 ◦C for 3.5 h was found to be the best thermal process for the fabrication of “sandwich structure” composites. 3.2. Fracture behaviour Previous research has proved that
-Al2O3/borosilicate glass systems containing
-Al2O3 platelets as reinforce￾ment lead to composites with higher fracture strength, Young’s modulus, hardness and fracture toughness than the matrix without reinforcement [6,37]. In the present inves￾tigation, fracture behaviour of the composites was anal-

276 A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 N 1 mm Fig. 7. Optical microscopy microgra ng the matrix region of about Imm width between two sapphire fibres ("optical window)in a polished)transparent sandwich structure composite ysed qualitatively by examining fracture surfaces of different 3.2. 1. Sintered composites Fig. 8(a and b)are SEM images of fracture surfaces of sintered samples. Fig 8(a) shows that there is pull-out of the 19sE1 fibres during the fracture of the composite. Since there are many defects at the interface between fibres and matrix(see Fig. 8. SEM images of fracture surfaces of pressureless sintered samples also Fig. 3(c)), the bonding at the interface is weak and it is showing(a)evidence of pull-out of the Saphikonfibres during fracture of easy to have fibre/matrix debonding and fibre pull out. These the composite and(b) defect-free matrix-fibre interface thus they will contribute to increase the toughness of the adhered to the surface of the fibre (a) indicate rest of glass from the matrix mechanisms(debonding and pull-out )dissipate energy, and matrix-fibre bonding. The arrow composite. It is also seen that there are cracks in the matrix in regions around the fibres and at the interface, which are Typical fracture surfaces showing evidence of fibre debond- most probably a consequence of pores and crack-like flaws ing and pull-out are presented in Fig 9(a and b). Due to the es compared to press in regions where the glass matrix is in close contact with sintered materials, they are expected to exhibit higher frac- the fibres, bonding between matrix and fibres can be free ture strength. Moreover, due to the unidirectional alignment of defects. In this case a strong fibre/matrix bonding may will exhibit anisotropic mechanical behaviour with highest the pulled-out fibres(arrows in Fig. 8(a)confirms that the strength and stiffness in the fibre direction, exploiting a load bonding between borosilicate glass and sapphire fibres w transfer mechanism [71 very strong after sintering at 750C if there was close con- tact between them. Thus, it may be anticipated that fracture 3.2.3. "Sandwich structure"composites ughness of the samples will be only partially improved by sandwich method" led to defect-free composites debonding and pull-out mechanisms. Moreover, the samples without These composites therefore should have will show low fracture strength because of the relatively high higher fracture strength than the sintered samples. SEM micrographs of fracture surfaces of"sandwich structure composites, shown in Fig. 10(a-c), confirm that when the 3. 2.2. Hot-pressed composites matrix is in close contact with the fibres a strong bonding may very similar, in qualitative terms, to that of pressureless was develop between them. Fig. 10(a)shows an impression left The fracture behaviour of the hot-pressed composite in the matrix by the fibre when the sample was fractured. The tered samples, as inferred from fracture surface analyses matrix was broken and some glass matrix stayed adhered to

276 A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 Fig. 7. Optical microscopy micrograph showing the matrix region of about 1mm width between two sapphire fibres (“optical window”) in a (non￾polished) transparent sandwich structure composite. ysed qualitatively by examining fracture surfaces of different samples. 3.2.1. Sintered composites Fig. 8(a and b) are SEM images of fracture surfaces of sintered samples. Fig. 8(a) shows that there is pull-out of the fibres during the fracture of the composite. Since there are many defects at the interface between fibres and matrix (see also Fig. 3(c)), the bonding at the interface is weak and it is easy to have fibre/matrix debonding and fibre pull out. These mechanisms (debonding and pull-out) dissipate energy, and thus they will contribute to increase the toughness of the composite. It is also seen that there are cracks in the matrix in regions around the fibres and at the interface, which are most probably a consequence of pores and crack-like flaws produced during sintering. However, Fig. 8(b) indicates that in regions where the glass matrix is in close contact with the fibres, bonding between matrix and fibres can be free of defects. In this case a strong fibre/matrix bonding may be assumed. Furthermore, the presence of layers of glass on the pulled-out fibres (arrows in Fig. 8(a)) confirms that the bonding between borosilicate glass and sapphire fibres was very strong after sintering at 750 ◦C if there was close con￾tact between them. Thus, it may be anticipated that fracture toughness of the samples will be only partially improved by debonding and pull-out mechanisms. Moreover, the samples will show low fracture strength because of the relatively high porosity. 3.2.2. Hot-pressed composites The fracture behaviour of the hot-pressed composites was very similar, in qualitative terms, to that of pressureless sin￾tered samples, as inferred from fracture surface analyses. Fig. 8. SEM images of fracture surfaces of pressureless sintered samples showing (a) evidence of pull-out of the Saphikon® fibres during fracture of the composite and (b) defect-free matrix-fibre interface in an area of good matrix-fibre bonding. The arrows in (a) indicate rest of glass from the matrix adhered to the surface of the fibre. Typical fracture surfaces showing evidence of fibre debond￾ing and pull-out are presented in Fig. 9(a and b). Due to the lower porosity of these composites compared to pressureless sintered materials, they are expected to exhibit higher frac￾ture strength. Moreover, due to the unidirectional alignment of the Saphikon® fibres in these composites, the materials will exhibit anisotropic mechanical behaviour with highest strength and stiffness in the fibre direction, exploiting a load￾transfer mechanism [7]. 3.2.3. “Sandwich structure” composites The “sandwich method” led to defect-free composites without porosity. These composites therefore should have higher fracture strength than the sintered samples. SEM micrographs of fracture surfaces of “sandwich structure” composites, shown in Fig. 10(a–c), confirm that when the matrix is in close contact with the fibres a strong bonding may develop between them. Fig. 10(a) shows an impression left in the matrix by the fibre when the sample was fractured. The matrix was broken and some glass matrix stayed adhered to

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 277 strength and elastic modulus than the matrix. moreover. the higher thermal expansion coefficient of the fibre than that of the matrix implies that, upon cooling from the fabrication temperature, the matrix will be in tangential compression [7]. This should increase the tensile fracture strength of the composite. It has been also shown that compressive residual thermal stresses in the matrix caused by thermal expansion coefficient mismatch have important effects on the toughen- ing mechanisms acting in brittle matrix composites [ 5,8], and that a compressive residual stress field in the matrix should increase the matrix microcracking stress [71 3.3. Optical 5eum92122 The matching of refractive index of fibre and matrix in the present composites is not perfect; meaning that only low volume fraction of fibres may be used to achieve a transpar similar to the pressureless sintered composites, only hot-pressed and sandwich structure composites are discussed in this section The macroscopic appearance of polished"sandwich struc es is shown in Fig. 11(a and b), which qual- itatively demonstrates the transparency of the samples. The images show that it is possible to see through the composites and thus light scattering effects are minimised. The under laying text remains clearly legible even if the composite is not directly above the text (Fig. 1l(b)). This behaviour was expected be the dispe of the fibres in the ma leaving optical windows( Fig. 7), as in the conventional mate- rial wired glass. The matrix regions between the fibres should have the same light transmittance as the monolithic borosil Fig9. Typical fracture surfaces of a hot-pressed sample showing evidence cate glass matrix of Saphikon" fibre(a) debonding and (b) pull-out. Measurements of the light transmittance of the compos- ites in the UV and visible wavelength ranges were carried out the fibre surface after fracture. This result confirms that there to quantify the transparency of the samples. The results are exists a strong bonding between a-Al2O3 and borosilicate shown in Fig. 12. The figure shows the light transmittance of glass. As proved in previous investigations [31, 36, 371, chem- three samples: a hot-pressed sample without fibres, a"sand ical compatibility between Al2O3 inclusions and borosilicate wich structure" sample without fibres(reference matrix)and glass produces a very strong chemical bonding and leads to a"sandwich structure" sample with fibres, before and after strong, but brittle composites. This was confirmed by the polishing. It is seen that the hot-pressed sample has lost very short"pull-out "length in these composites, as shown in almost all of its optical transparency; it has a transmittance of Fig 10(b), and the relatively sharp interface between sapphire 18% relative to the borosilicate glass plate. It is thus difficult fibre and borosilicate glass matrix, as shown in Fig. 10(c) to envisage that it will be feasible to obtain a transparent fibre The results indicate the need to coat the fibres with a suit- reinforced composite by this technique of fabrication, even if able material to produce the weak interface required for ful high densities(as high as 98% of theoretical density) could exploiting toughening mechanisms such as fibre pull-out. a be achieved, as discussed in Section 3. 1.2 typical material suggested for composites with silicate matri- The"sandwich structure" sample without fibres has a ces and alumina-containing fibres is SnO [31, 36]. Chawla 100% transmittance in the wavelength range 350-800 nm et al. have studied the role of SnO coating, which has no This means that there was no loss of transparency by joining diffusion in alumina and very little diffusion in silicate glass two plates of glass together in a sandwich structure. This also 31]. Other works have been conducted using TiO2 coatings confirms that the processing parameters used for fabrication on aluminosilicate fibres, which were embedded in silicate were optimised for this system glass matrices [22] The transmittance of a"sandwich structure"composite In principle, sapphire fibres have suitable properties to before and after polishing(to 3 um using a diamond grid) einforce borosilicate glass matrices, i.e. much higher tensile was also measured. As expected, Fig. 12 shows that the pol

A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 277 Fig. 9. Typical fracture surfaces of a hot-pressed sample showing evidence of Saphikon® fibre (a) debonding and (b) pull-out. the fibre surface after fracture. This result confirms that there exists a strong bonding between
-Al2O3 and borosilicate glass. As proved in previous investigations[31,36,37], chem￾ical compatibility between Al2O3 inclusions and borosilicate glass produces a very strong chemical bonding and leads to strong, but brittle composites. This was confirmed by the very short “pull-out” length in these composites, as shown in Fig. 10(b), and the relatively sharp interface between sapphire fibre and borosilicate glass matrix, as shown in Fig. 10(c). The results indicate the need to coat the fibres with a suit￾able material to produce the weak interface required for fully exploiting toughening mechanisms such as fibre pull-out. A typical material suggested for composites with silicate matri￾ces and alumina-containing fibres is SnO2 [31,36]. Chawla et al. have studied the role of SnO2 coating, which has no diffusion in alumina and very little diffusion in silicate glass [31]. Other works have been conducted using TiO2 coatings on aluminosilicate fibres, which were embedded in silicate glass matrices [22]. In principle, sapphire fibres have suitable properties to reinforce borosilicate glass matrices, i.e. much higher tensile strength and elastic modulus than the matrix. Moreover, the higher thermal expansion coefficient of the fibre than that of the matrix implies that, upon cooling from the fabrication temperature, the matrix will be in tangential compression [7]. This should increase the tensile fracture strength of the composite. It has been also shown that compressive residual thermal stresses in the matrix caused by thermal expansion coefficient mismatch have important effects on the toughen￾ing mechanisms acting in brittle matrix composites[5,8], and that a compressive residual stress field in the matrix should increase the matrix microcracking stress [7]. 3.3. Optical properties The matching of refractive index of fibre and matrix in the present composites is not perfect; meaning that only low volume fraction of fibres may be used to achieve a transpar￾ent material, exploiting the optical window concept [18,24] similar to wired glass. Due to the lack of densification of the pressureless sintered composites, only hot-pressed and sandwich structure composites are discussed in this section. The macroscopic appearance of polished “sandwich struc￾ture” composites is shown in Fig. 11(a and b), which qual￾itatively demonstrates the transparency of the samples. The images show that it is possible to see through the composites, and thus light scattering effects are minimised. The under￾laying text remains clearly legible even if the composite is not directly above the text (Fig. 11(b)). This behaviour was expected because of the disposition of the fibres in the matrix leaving optical windows (Fig. 7), as in the conventional mate￾rial wired glass. The matrix regions between the fibres should have the same light transmittance as the monolithic borosili￾cate glass matrix. Measurements of the light transmittance of the compos￾ites in the UV and visible wavelength ranges were carried out to quantify the transparency of the samples. The results are shown in Fig. 12. The figure shows the light transmittance of three samples: a hot- pressed sample without fibres, a “sand￾wich structure” sample without fibres (reference matrix) and a “sandwich structure” sample with fibres, before and after polishing. It is seen that the hot-pressed sample has lost almost all of its optical transparency; it has a transmittance of 18% relative to the borosilicate glass plate. It is thus difficult to envisage that it will be feasible to obtain a transparent fibre reinforced composite by this technique of fabrication, even if high densities (as high as 98% of theoretical density) could be achieved, as discussed in Section 3.1.2. The “sandwich structure” sample without fibres has a 100% transmittance in the wavelength range 350–800 nm. This means that there was no loss of transparency by joining two plates of glass together in a sandwich structure. This also confirms that the processing parameters used for fabrication were optimised for this system. The transmittance of a “sandwich structure” composite before and after polishing (to 3 m using a diamond grid) was also measured. As expected, Fig. 12 shows that the pol-

A.R. Boccaccini et al. Joumal of Materials Processing Technology 169(2005)270-280 15KV 9 10 SEI 13 Z6 SE I Matrix 28kU X3, 008 Fig. 10. SEM micrographs of fracture surfaces of "sandwich structure" composites showing(a) impression left in the matrix by a fibre debonded during fracture,(b)very short"pull-out"length in these composites and (c)the relatively sharp interface between sapphire fibre and borosilicate glass, indicating strong fibre/matrix bonding. ished sample has better transmittance properties than the non- Fig. 12, can be analysed quantitatively by the following rela- polished sample; polishing the sample increases its optical tion, which was developed for optomechanical composites transmittance by 20%.The"sandwich structure"composites with regularly aligned fibres in a transparent matrix [24] exhibit an almost constant transmittance of about 60% before are in agreement with the images in Fig. 11, showing the Te=Tml_D polishing and of about 80% after polishing. These results (1) high transparency of the composites, and they confirm that e"optical window"concept is a convenient way to fabri- where Te and Tm are the light transmittance of the compos- cate optomechanical composites, as proposed in the literature ite and the monolithic matrix, respectively, and Dr and Lr [18, 24]. The present results are better in terms of transparency are the fibre diameter and fibre to fibre spacing, respectively than those obtained in our previous work [22], where only Incorporating the diameter of the sapphire fibre( 150 um) 60% transparency in Nextel fibre reinforced soda-lime glass and the fibre to fibre spacing(l mm)in Eq. (1), the light composites fabricated by the"sandwich structure"method transmittance of the polished"sandwich structure"compos- was achieved. The results are similar to those obtained ite relative to that of the monolithic matrix(Tc/Tm)is found recently by Dericioglu and Kagawa [44], who achieved 80% to be%. This value is in good agreement with the expe relative transparency in SiC fibre reinforced MgAl2O4 matrix imentally determined relative light transmittance value given optomechanical composites, in which the fibre diameter and in Fig. 12 for the polished composite that reads-80%.This fibre to fibre spacing were identical to those of the present result demonstrates the effectiveness of the optical window sandwich structure"composites. The light transmittance of concept in the present sapphire fibre/borosilicate glass com- the polished "sandwich structure"composite, which is given posites and its efficiency in providing optical transparency to relative to that of the monolithic borosilicate glass matrix in he resulting optomechanical composite

278 A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 Fig. 10. SEM micrographs of fracture surfaces of “sandwich structure” composites showing (a) impression left in the matrix by a fibre debonded during fracture, (b) very short “pull-out” length in these composites and (c) the relatively sharp interface between sapphire fibre and borosilicate glass, indicating strong fibre/matrix bonding. ished sample has better transmittance properties than the non￾polished sample; polishing the sample increases its optical transmittance by 20%. The “sandwich structure” composites exhibit an almost constant transmittance of about 60% before polishing and of about 80% after polishing. These results are in agreement with the images in Fig. 11, showing the high transparency of the composites, and they confirm that the “optical window” concept is a convenient way to fabri￾cate optomechanical composites, as proposed in the literature [18,24]. The present results are better in terms of transparency than those obtained in our previous work [22], where only 60% transparency in Nextel® fibre reinforced soda-lime glass composites fabricated by the “sandwich structure” method was achieved. The results are similar to those obtained recently by Dericioglu and Kagawa [44], who achieved 80% relative transparency in SiC fibre reinforced MgAl2O4 matrix optomechanical composites, in which the fibre diameter and fibre to fibre spacing were identical to those of the present “sandwich structure” composites. The light transmittance of the polished “sandwich structure” composite, which is given relative to that of the monolithic borosilicate glass matrix in Fig. 12, can be analysed quantitatively by the following rela￾tion, which was developed for optomechanical composites with regularly aligned fibres in a transparent matrix [24]: Tc = Tm
1 − Df Lf (1) where Tc and Tm are the light transmittance of the compos￾ite and the monolithic matrix, respectively, and Df and Lf are the fibre diameter and fibre to fibre spacing, respectively. Incorporating the diameter of the sapphire fibre (∼150m) and the fibre to fibre spacing (∼1 mm) in Eq. (1), the light transmittance of the polished “sandwich structure” compos￾ite relative to that of the monolithic matrix (Tc/Tm) is found to be ∼85%. This value is in good agreement with the exper￾imentally determined relative light transmittance value given in Fig. 12 for the polished composite that reads ∼80%. This result demonstrates the effectiveness of the optical window concept in the present sapphire fibre/borosilicate glass com￾posites and its efficiency in providing optical transparency to the resulting optomechanical composite.

A.R. Boccaccini et al. /Journal of Materials Processing Technology 169(2005)270-280 4. Conclusions The system sapphire fibre reinforced borosilicate glass matrix composite was studied aiming at developing"optome- chanical composites". Different techniques of fabrication were used: randomly orientated chopped fibre reinforced Imperial College composites were fabricated by pressureless sintering, unidirectionally oriented fibre reinforced composites were fabricated by hot-pressing and by sandwiching two slides London of glass and an array of parallel fibres(sandwich structure composites) Pressureless sintered samples were porous, specially near the fibre/matrix interfaces and there was poor contact between the fibres and the matrix Hot-pressed and"sandwich struc ture"composites were dense and showed a strong interface between fibres and matrix. Interface engineering should be introduced by coating the fibres with a suitable material(e.g SnO2)in order to obtain a weaker interface and improve toughness by inducing significant fibre pull-out effect. Pressureless sintered and hot-pressed samples were due to residual ever, the hot-pressed unreinforced matrix was translucent. On the other hand," sandwich structure"composites were transparent and showed significant light transmittance in the mper visible wavelength range, only 20% lower than that of the n unreinforced matrix(borosilicate glass slides). These results indicate that this technique of fabrication is viable for pro- duction of"optomechanical composites"with borosilicate 0 mm glass matrix. The samples fabricated, which exhibit strong fibre/matrix interface bonding, represent an improved(but less cost-effective) version of the traditional fire and impact es qualitatively demonstrating the transparency of the samples. The images resistant material wired glass. The composites should there how that it is possible to see through the composites. The under-laying text fore be interesting materials for high performance fire resis- remains clearly legible if the sample is placed (a)in direct contact with th tant windows, requiring high impact strength and avoidance text and(b )even if the composite is not directly above the tex of fragmentation upon fracture, obviously in cases where stringent requirements may justify the higher costs of the Acknowledgments Experimental assistance of Mr. Norbert Galy is appre- d-b ciated. ARB acknowledges financial support of the Royal Society, London, UK(Grant nr. 574006G503 References 350400450 600650700750800 [I.w. Donald, Review. Methods for improving the mechanical prop erties of oxide glasses, J. Mater. Sci. 24(1989)4177-4208 [2]KM. Prewo, J.J. Brennan, G K. Layden, Fiber reinforced glasse 12. Results of the measurements of light transmittance of selected san and glass-ceramics for high performance applications, Ceram. Bull the UV and visible wavelength ranges: hot-pressed sample without 65(1986)305-322 fibres(a), "sandwich structure" sample without fibres(b) and"sandwich 3R. D. Rawlings, Glass-ceramic matrix composites, Composites 25 structuresample with fibres, before and after polishing(c and d, respec 994)372-379 tively). For the composites, light transmittance was measured perpendicu- 14 A.R. Boccaccini, Glass and glass-ceramic matrix composite materi- larly to the fibre axes. als: a review, J. Ceram Soc. Jpn. 109(7)(2001)S99-0S10

A.R. Boccaccini et al. / Journal of Materials Processing Technology 169 (2005) 270–280 279 Fig. 11. Macroscopic appearance of polished “sandwich structure” compos￾ites qualitatively demonstrating the transparency of the samples. The images show that it is possible to see through the composites. The under-laying text remains clearly legible if the sample is placed (a) in direct contact with the text and (b) even if the composite is not directly above the text. Fig. 12. Results of the measurements of light transmittance of selected sam￾ples in the UV and visible wavelength ranges: hot-pressed sample without fibres (a), “sandwich structure” sample without fibres (b) and “sandwich structure” sample with fibres, before and after polishing (c and d, respec￾tively). For the composites, light transmittance was measured perpendicu￾larly to the fibre axes. 4. Conclusions The system sapphire fibre reinforced borosilicate glass matrix composite was studied aiming at developing “optome￾chanical composites”. Different techniques of fabrication were used: randomly orientated chopped fibre reinforced composites were fabricated by pressureless sintering, unidirectionally oriented fibre reinforced composites were fabricated by hot-pressing and by sandwiching two slides of glass and an array of parallel fibres (sandwich structure composites). Pressureless sintered samples were porous, specially near the fibre/matrix interfaces and there was poor contact between the fibres and the matrix. Hot-pressed and “sandwich struc￾ture” composites were dense and showed a strong interface between fibres and matrix. Interface engineering should be introduced by coating the fibres with a suitable material (e.g. SnO2) in order to obtain a weaker interface and improve toughness by inducing significant fibre pull-out effect. Pressureless sintered and hot-pressed samples were opaque due to residual porosity and sintering defects. How￾ever, the hot-pressed unreinforced matrix was translucent. On the other hand, “sandwich structure” composites were transparent and showed significant light transmittance in the visible wavelength range, only 20% lower than that of the unreinforced matrix (borosilicate glass slides). These results indicate that this technique of fabrication is viable for pro￾duction of “optomechanical composites” with borosilicate glass matrix. The samples fabricated, which exhibit strong fibre/matrix interface bonding, represent an improved (but less cost-effective) version of the traditional fire and impact resistant material wired glass. The composites should there￾fore be interesting materials for high performance fire resis￾tant windows, requiring high impact strength and avoidance of fragmentation upon fracture, obviously in cases where stringent requirements may justify the higher costs of the material. Acknowledgments Experimental assistance of Mr. Norbert Galy is appre￾ciated. ARB acknowledges financial support of the Royal Society, London, UK (Grant nr. 574006.G503). References [1] I.W. Donald, Review. Methods for improving the mechanical prop￾erties of oxide glasses, J. Mater. Sci. 24 (1989) 4177–4208. [2] K.M. Prewo, J.J. Brennan, G.K. Layden, Fiber reinforced glasses and glass-ceramics for high performance applications, Ceram. Bull. 65 (1986) 305–322. [3] R.D. Rawlings, Glass-ceramic matrix composites, Composites 25 (1994) 372–379. [4] A.R. Boccaccini, Glass and glass-ceramic matrix composite materi￾als: a review, J. Ceram. Soc.Jpn. 109 (7) (2001) S99–0S109