Availableonlineatwww.sciencedirect.com °" Science Direct c000e5 Part B: engineering ELSEVIER Composites: Part B 39(2008)362-373 www.elsevier.com/locate/compositesb Damage tolerant, translucent oxide fiber/glass matrix composites Dagmar Hulsenberg", Peer Fehling, Thomas Leutbecher TU Ilmenau, Department of Glass and Ceramics Technology, P.O. Box 100565, 98684 Ilmenau, German Received 21 July 2006: accepted 16 January 2007 Available online 26 January 2007 Abstract Starting out from damage-tolerant oxidic composites with carbon (C)-coated fibers, the main part of our investigations was concentrated on Nextel 440-fibers/glass matrix composites. The matrix glasses were chosen in consideration of suitable thermal expan- sion coefficients and optical refractive indices. The fiber pull-out is enabled by a CVd coating with boron nitride(BN). In comparison with a single BN layer, a BN /TiO -double layer yields better mechanical and optical properties. All composites are made by means of a traditional lab method including hot pressing. Varying differences of the thermal expansion coefficients of fibers and matrix glasses(minus, zero, plus) allow the preparation of composites presenting various mechanical properties and also different damage tolerances. A composite, which is both damage-tolerant and translucent at one and the same time requires an identical thermal expansion coefficient and, in addition, an identical optical efractive index of the fibers and the matrix. Bubbles must be avoided in any case. e 2007 Elsevier Ltd. All rights reserved Keywords: A. Glass fibres; B. Damage tolerance; B. Optical properties/techniques: E. Heat treatment; Glass-matrix composites 1. Introduction the load transported by the fiber will be. By applying a suit ible intermediate layer between the fibers and the glass lass is characterized by some very positive properties matrix (i.e an appropriate coating of the fibers), the crack such as its transparency in the visible wavelength range, propagation as well as the fracture toughness can advanta- its high electrical resistance and adaptable thermal expan- geously be influenced. As soon as a crack develops in the sion coefficient. However, its application is limited in par- matrix, the coated fiber will either divert or bridge it. From ticular because of its low crack resistance, the sudden a certain limit load on, the glass matrix will present multi- crack propagation and the resulting catastrophic fracture. crack growth. If the load increases further, the fibers will be Different attempts were made to overcome these problems. extended even more, thus entailing a relative movement The mechanical strength and the fracture toughness of between the fibers and the matrix. This, in turn, will cause act glass products can be improved by its reinforcing the fiber debonding and a fiber pull-out. When the loading rs. The material will not become ductile, but que has reached the ultimate strength of the fibers, the compos- ite will definitely br If a fiber-reinforced glass matrix is exposed to stress, the If the bond between the fibers and the matrix is so load will be transferred to the fibers so that the brittle glass strong that the fibers cannot be pulled out, the composite matrix is relieved. The degree of load transfer depends on will remain brittle. This depends- among other factors both E-moduluses and on the fiber volume content of the also on the properties of the fiber coating, whether chemi composite. The higher the ratio Efiber/Ematrix is, the higher cal reactions take place between the constituents, whether surface roughness brought about by nanocrystals prevents Corresponding adua or. TeL: +493677692801: fax: +49 3677691436. the fibers from sliding in the glass matrix, or whether the de(D.Holsenberg). desired pulling-out of the fibers really takes place. The 1359-8368/S.see front matter e 2007 Elsevier Ltd. All rights reserved doi:10.1016/j.compositesb.2007.01.00
Damage tolerant, translucent oxide fiber/glass matrix composites Dagmar Hu¨lsenberg *, Peer Fehling, Thomas Leutbecher TU Ilmenau, Department of Glass and Ceramics Technology, P.O. Box 100565, 98684 Ilmenau, Germany Received 21 July 2006; accepted 16 January 2007 Available online 26 January 2007 Abstract Starting out from damage-tolerant oxidic composites with carbon (C)-coated fibers, the main part of our investigations was concentrated on Nextel 440-fibers/glass matrix composites. The matrix glasses were chosen in consideration of suitable thermal expansion coefficients and optical refractive indices. The fiber pull-out is enabled by a CVD coating with boron nitride (BN). In comparison with a single BN layer, a BN/TiO2-double layer yields better mechanical and optical properties. All composites are made by means of a traditional lab method including hot pressing. Varying differences of the thermal expansion coefficients of fibers and matrix glasses (minus, zero, plus) allow the preparation of composites presenting various mechanical properties and also different damage tolerances. A composite, which is both damage-tolerant and translucent at one and the same time requires an identical thermal expansion coefficient and, in addition, an identical optical refractive index of the fibers and the matrix. Bubbles must be avoided in any case. 2007 Elsevier Ltd. All rights reserved. Keywords: A. Glass fibres; B. Damage tolerance; B. Optical properties/techniques; E. Heat treatment; Glass-matrix composites 1. Introduction Glass is characterized by some very positive properties such as its transparency in the visible wavelength range, its high electrical resistance and adaptable thermal expansion coefficient. However, its application is limited in particular because of its low crack resistance, the sudden crack propagation and the resulting catastrophic fracture. Different attempts were made to overcome these problems. The mechanical strength and the fracture toughness of compact glass products can be improved by its reinforcing by fibers. The material will not become ductile, but quasiductile [1–3]. If a fiber-reinforced glass matrix is exposed to stress, the load will be transferred to the fibers so that the brittle glass matrix is relieved. The degree of load transfer depends on both E-moduluses and on the fiber volume content of the composite. The higher the ratio Efiber/Ematrix is, the higher the load transported by the fiber will be. By applying a suitable intermediate layer between the fibers and the glass matrix (i.e., an appropriate coating of the fibers), the crack propagation as well as the fracture toughness can advantageously be influenced. As soon as a crack develops in the matrix, the coated fiber will either divert or bridge it. From a certain limit load on, the glass matrix will present multicrack growth. If the load increases further, the fibers will be extended even more, thus entailing a relative movement between the fibers and the matrix. This, in turn, will cause the fiber debonding and a fiber pull-out. When the loading has reached the ultimate strength of the fibers, the composite will definitely break. If the bond between the fibers and the matrix is so strong that the fibers cannot be pulled out, the composite will remain brittle. This depends – among other factors – also on the properties of the fiber coating, whether chemical reactions take place between the constituents, whether surface roughness brought about by nanocrystals prevents the fibers from sliding in the glass matrix, or whether the desired pulling-out of the fibers really takes place. The 1359-8368/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesb.2007.01.002 * Corresponding author. Tel.: +49 3677 692801; fax: +49 3677 691436. E-mail address: dagmar.huelsenberg@tu-ilmenau.de (D. Hu¨lsenberg). www.elsevier.com/locate/compositesb Available online at www.sciencedirect.com Composites: Part B 39 (2008) 362–373
D. Hiilsenberg et al/ Composites: Part B 39(2008)362-373 latter will lead to a controlled, slow breaking, i.e., damage between the fibers and the matrix. All sorts of chemical tolerance. Consequently, it is necessary to evaluate the interaction, force- or form-fit connection must be properties of the interface, in particular the shear strength prevented so as to be able to optimize the bond between the fibers and In contrast to this, the optical transparency requires to the matrix have as few disturbing interfaces as possible, so no inter- Earlier studies and publications concentrated only on nal cracks. flaws or bubbles this aspect of damage tolerance [4-6. Carbon or Sic-fibers with C-coating were used. The coating similar to graphite In order to achieve these contradicting aims, the follow allowed the fibers to slide in the glass matrix. The compos- ing prerequisites must be provided ites were characterized by excellent mechanical properties at room temperature and a good thermal stability. How Both the fibers and the matrix must be oxidic and opti- ever, they were black. The transparency of the matrix glass cally transpar or the nanocrystalline glass ceramic matrix was not The covering layers must also be optically transparent adopted by the composite Both the fibers and the matrix must present absolutely Japanese authors investigated the possibility of produc- identical optical refractive index. ing transparent glass matrix composites [7, 8]. The trans-- The thermal expansion coefficients of both the fibers and mission of the inspected oxynitride fibers/glass matrix the matrix must be adapted composites amounted to 40% in the visible wavelength The fibers shall soften at considerably higher tempera range, with the fiber volume portion being only 7%. It tures than the matrix glass was not possible to explain the negative influence exerted The E-modulus of the fibers must be clearly larger than that of the ma chemical bond between the fibers and the matrix. the com osites showed catastrophic fracture. No comment was The fibers may be amorphous(like S-glass) or nanocrys- nade by the authors concerning the interface between the talline(without influence on the transparency like Nextel fibers and the matrix Other authors [9] investigated the influence exerted by The fiber coating should fulfill the following functions different transparent fiber coatings on the mechanical prop- erties of oxidic composites. They tested uncoated, SnOx-or 1 It should make sure that no chemical interactions BN-coated Nextel 480-fibers. Compared with non-rein between the fibers and the matrix due to diffusion pro- forced matrix glass, only the BN-coated Nextel 480-fiber cesses or other reactions may take place was able to increase the strength of the material. Neither 2. It should form a smooth layer without any disturbing in the composite with the uncoated fibers nor in the one crystallizations so as to allow the fibers to slide with the SnOz-coated fibers, the fibers pulled out. The 3. The transparency in the visible wavelength range should pull-out was only observed in the case of BN coating. be guaranteed The SnO2 layer caused a strong bond with the matrix. 4. It should be thermally resistant and resistant to environ- Although SnO2 does not react with the Al_O3 of the Nextel mental influences even in an oxidizing atmosphere 480-fibers, it reacts very intensely with the Sio of the glass matrix instead. The fact that the BN-coated fibers slide As the thickness of the covering layers is far below the within the glass matrix may be due to the plate-like nano- wavelength of the visible light, the optical refractive index crystalline structure of the BN layers. Surprisingly, the does not play a crucial role cited authors did not test other coatings like, for example, TiO2 or ZrOz. One could possibly expect that they react 3 Experiments less intensely with the glass matrix, compared with Sno The aim of our own research work was to investigate com- 3. 1. Overview of the constituents tested posites presenting three properties: they should be damage tolerant, stronger than compact glass, and translucent [1 Two different types of fibers were chosen, the above mentioned S-glass fiber and the Nextel 440 fiber. Their 2. Preconditions composition and properties are shown in Table 1 In order to study some certain combinations of fibers The following opposing parameters have to be and matrix glasses with adapted and non-adapted thermal optimize expansion coefficients and refractive indices, four matrix glasses were used. Table 2 represents their composition The improvement of fracture toughness by crack deflec- and properties. Glass type 8650 contains a big quantity tion, the debonding and pulling out of the fibers from of lead oxide, therefore requiring oxidizing conditions dur the matrix can only be achieved through a soft interface ing the preparation of the composite
latter will lead to a controlled, slow breaking, i.e., damage tolerance. Consequently, it is necessary to evaluate the properties of the interface, in particular the shear strength, so as to be able to optimize the bond between the fibers and the matrix. Earlier studies and publications concentrated only on this aspect of damage tolerance [4–6]. Carbon or SiC-fibers with C-coating were used. The coating similar to graphite allowed the fibers to slide in the glass matrix. The composites were characterized by excellent mechanical properties at room temperature and a good thermal stability. However, they were black. The transparency of the matrix glass or the nanocrystalline glass ceramic matrix was not adopted by the composite. Japanese authors investigated the possibility of producing transparent glass matrix composites [7,8]. The transmission of the inspected oxynitride fibers/glass matrix composites amounted to 40% in the visible wavelength range, with the fiber volume portion being only 7%. It was not possible to explain the negative influence exerted by the constituents used on the transparency of the composites although the thermal expansion coefficients and the refractive indices were adapted. Due to the strong chemical bond between the fibers and the matrix, the composites showed catastrophic fracture. No comment was made by the authors concerning the interface between the fibers and the matrix. Other authors [9] investigated the influence exerted by different transparent fiber coatings on the mechanical properties of oxidic composites. They tested uncoated, SnO2- or BN-coated Nextel 480-fibers. Compared with non-reinforced matrix glass, only the BN-coated Nextel 480-fiber was able to increase the strength of the material. Neither in the composite with the uncoated fibers nor in the one with the SnO2-coated fibers, the fibers pulled out. The pull-out was only observed in the case of BN coating. The SnO2 layer caused a strong bond with the matrix. Although SnO2 does not react with the Al2O3 of the Nextel 480-fibers, it reacts very intensely with the SiO2 of the glass matrix instead. The fact that the BN-coated fibers slide within the glass matrix may be due to the plate-like nanocrystalline structure of the BN layers. Surprisingly, the cited authors did not test other coatings like, for example, TiO2 or ZrO2. One could possibly expect that they react less intensely with the glass matrix, compared with SnO2. The aim of our own research work was to investigate composites presenting three properties: they should be damagetolerant, stronger than compact glass, and translucent [10]. 2. Preconditions The following opposing parameters have to be optimized: – The improvement of fracture toughness by crack deflection, the debonding and pulling out of the fibers from the matrix can only be achieved through a soft interface between the fibers and the matrix. All sorts of chemical interaction, force- or form-fit connection must be prevented. – In contrast to this, the optical transparency requires to have as few disturbing interfaces as possible, so no internal cracks, flaws or bubbles. In order to achieve these contradicting aims, the following prerequisites must be provided: – Both the fibers and the matrix must be oxidic and optically transparent. – The covering layers must also be optically transparent. – Both the fibers and the matrix must present absolutely identical optical refractive index. – The thermal expansion coefficients of both the fibers and the matrix must be adapted. – The fibers shall soften at considerably higher temperatures than the matrix glass. – The E-modulus of the fibers must be clearly larger than that of the matrix. – While joining the fibers and the matrix, no bubbles, grain boundaries or cracks may be left. The fibers may be amorphous (like S-glass) or nanocrystalline (without influence on the transparency like Nextel 440). The fiber coating should fulfill the following functions: 1. It should make sure that no chemical interactions between the fibers and the matrix due to diffusion processes or other reactions may take place. 2. It should form a smooth layer without any disturbing crystallizations so as to allow the fibers to slide. 3. The transparency in the visible wavelength range should be guaranteed. 4. It should be thermally resistant and resistant to environmental influences even in an oxidizing atmosphere. As the thickness of the covering layers is far below the wavelength of the visible light, the optical refractive index does not play a crucial role. 3. Experiments 3.1. Overview of the constituents tested Two different types of fibers were chosen, the abovementioned S-glass fiber and the Nextel 440 fiber. Their composition and properties are shown in Table 1. In order to study some certain combinations of fibers and matrix glasses with adapted and non-adapted thermal expansion coefficients and refractive indices, four matrix glasses were used. Table 2 represents their composition and properties. Glass type 8650 contains a big quantity of lead oxide, therefore requiring oxidizing conditions during the preparation of the composite. D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373 363
D. Hilsenberg et al/Composites: Part B 39(2008)362-373 Table I could possibly be avoided by applying a microwave Composition and properties of used fibers(specification by the producers) assisted CVD at reaction temperatures of around 500oC Dimension Nextel 440 This could be a task to be solved in the future The Sno, layers prepared presented relatively big cassit erite crystals which -together with certain reactions taking place with the glass matrix prevented the coated fibers %%K from sliding along within the matrix when stress was exerted. The composites turned brittle and broke, as other 2.8 authors had already stated earlier [9]. Therefore, research work with SnO -coated fibers was stopped Refractive index n 1.523 Transformation temperature, Tg 3.2. Single fiber tensile strength Tensile strength MPa 4580 2070 E-modulus The tensile strength of the single fibers (as received Fiber diameter desized and coated) was measured at room temperature Before the measurement was made. some of the fibers were exposed to different temperatures in order to simulate the Table 2 and properties of the used matrix glasses(specifications by conditions prevailing during CVD coating and hot pressing of the composites Fig. I shows the results of the delivered S-glass and Ne Dimension Duran 756 N-SK 4 tel 440 fibers after a 5 h pre-treatment at different temper- Producer Schott Telux SchottSchott atures in air. Although the tensile strength of the S-glass 63.035.2 35.4 fibers in their original state is very high, it decreases so mass% 3.1 8.0 4.6 3.8 strongly, already after a pre-treatment at 500C, that it 134 falls below that of the Nextel 440 fibers. In the tensile strength of the Nextel 440 fibers remains nearly unchanged in the interesting temperature range of the hot g of the composites(up to 850C)and the previous K,O CVD coating(at 900C). Therefore, this type of fiber is 43.8 well suitable for the experiments Cs,O Fig. 2 shows an overview of the tensile strengths of the Thermal 10-K differently coated Nextel 440 fibers measured. The thin Refractive index n 143 1491.613 1.6 black lines illustrate how the measuring values vary around 500658 475 the averages. From the results, the conclusion can be drawn that the coating process alone does not remarkably influ- MPa 805257 ence the tensile strength(cf. Table 1). Only after exposing the composite to high temperatures for another 5 h, the ten- sile strength will decrease. However, some differences can be seen here as well. A statistically solid fact is that heating to 6 The intermediate layers investigated were pyrolytic car- 750 C over 5 h reduces the tensile strength only in the case on(for comparison purposes), SnO,(cassiterite crystals), of the TiO2-and BN-single layers Double BN/TiO2 coated TiO2(anatase nanocrystals), BN(turbostratic) and BN/ fibers will not lose the tensile strength considerably TiO2 double layers. Fig. 3 shows the surface of a coated Nextel 440 fiber The carbon coating was effected either by the conversion without any heating at all and after reheating. Here, atten- of novolak into amorphous carbon or by the chemical tion has to be drawn to the fact that Nextel 440 is a nano- vapour deposition(CVD). All other layers were prepared crystalline fiber [13]. The thin(30 nm BN+ 30 nm TiO2 by CVD. The authors wish to thank Prof. Marx and his fel- surface layer(Fig 3, left-hand side)allows the crystals of low workers from the TU Chemnitz, Institute of Physical the fibers to be imaged almost directly. They have a spher- Chemistry, for preparing the CVD layers [11, 12]. Before ical shape, thus generating a smooth surface, which allows coating, the fibers were thermally desized(over 3 h at a the fiber to slide. During the thermal sourcing out( Fig. 3, temperature of 500C). As will be shown later, the strength right-hand side), the crystals grow slightly. Nevertheless of the S-glass fibers is strongly decreasing with increasing their topographical structure will not be destroyed temperature. However, coating with BN requires tempera- tures of about 900C. As such temperatures are too high 3.3. Preparation of the test samples for S-glass fibers, the following investigations are mainly concentrated on composites containing Nextel 440-fibers. The method applied was taken from [15]and is sch The problem of the temperature stability of S-glass fibers ically illustrated in Fig. 4. The description begins in the top
The intermediate layers investigated were pyrolytic carbon (for comparison purposes), SnO2 (cassiterite crystals), TiO2 (anatase nanocrystals), BN (turbostratic) and BN/ TiO2 double layers. The carbon coating was effected either by the conversion of novolak into amorphous carbon or by the chemical vapour deposition (CVD). All other layers were prepared by CVD. The authors wish to thank Prof. Marx and his fellow workers from the TU Chemnitz, Institute of Physical Chemistry, for preparing the CVD layers [11,12]. Before coating, the fibers were thermally desized (over 3 h at a temperature of 500 C). As will be shown later, the strength of the S-glass fibers is strongly decreasing with increasing temperature. However, coating with BN requires temperatures of about 900 C. As such temperatures are too high for S-glass fibers, the following investigations are mainly concentrated on composites containing Nextel 440-fibers. The problem of the temperature stability of S-glass fibers could possibly be avoided by applying a microwave assisted CVD at reaction temperatures of around 500 C. This could be a task to be solved in the future. The SnO2 layers prepared presented relatively big cassiterite crystals which – together with certain reactions taking place with the glass matrix – prevented the coated fibers from sliding along within the matrix when stress was exerted. The composites turned brittle and broke, as other authors had already stated earlier [9]. Therefore, research work with SnO2-coated fibers was stopped. 3.2. Single fiber tensile strength The tensile strength of the single fibers (as received, desized and coated) was measured at room temperature. Before the measurement was made, some of the fibers were exposed to different temperatures in order to simulate the conditions prevailing during CVD coating and hot pressing of the composites. Fig. 1 shows the results of the delivered S-glass and Nextel 440 fibers after a 5 h pre-treatment at different temperatures in air. Although the tensile strength of the S-glass fibers in their original state is very high, it decreases so strongly, already after a pre-treatment at 500 C, that it falls below that of the Nextel 440 fibers. In comparison, the tensile strength of the Nextel 440 fibers remains nearly unchanged in the interesting temperature range of the hot pressing of the composites (up to 850 C) and the previous CVD coating (at 900 C). Therefore, this type of fiber is very well suitable for the experiments. Fig. 2 shows an overview of the tensile strengths of the differently coated Nextel 440 fibers measured. The thin black lines illustrate how the measuring values vary around the averages. From the results, the conclusion can be drawn that the coating process alone does not remarkably influence the tensile strength (cf. Table 1). Only after exposing the composite to high temperatures for another 5 h, the tensile strength will decrease. However, some differences can be seen here as well. A statistically solid fact is that heating to 750 C over 5 h reduces the tensile strength only in the case of the TiO2- and BN-single layers. Double BN/TiO2 coated fibers will not lose the tensile strength considerably. Fig. 3 shows the surface of a coated Nextel 440 fiber without any heating at all and after reheating. Here, attention has to be drawn to the fact that Nextel 440 is a nanocrystalline fiber [13]. The thin (30 nm BN + 30 nm TiO2) surface layer (Fig. 3, left-hand side) allows the crystals of the fibers to be imaged almost directly. They have a spherical shape, thus generating a smooth surface, which allows the fiber to slide. During the thermal sourcing out (Fig. 3, right-hand side), the crystals grow slightly. Nevertheless, their topographical structure will not be destroyed. 3.3. Preparation of the test samples The method applied was taken from [15] and is schematically illustrated in Fig. 4. The description begins in the top Table 1 Composition and properties of used fibers (specification by the producers) Dimension S-glass Nextel 440 Producer – Owens-Corning 3W SiO2 mass% 65 28 Al2O3 mass% 25 70 B2O3 mass% – 2 MgO mass% 10 – Thermal expansion coefficient, a 106 K1 2.8 5.3 Refractive index, n – 1.523 1.616 Transformation temperature, Tg C 816 – Tensile strength MPa 4580 2070 E-modulus GPa 87 186 Fiber diameter lm 10 11.6 Table 2 Compositions and properties of the used matrix glasses (specifications by the producers) Dimension Duran 756 N-SK 4 8650 Producer – Schott Telux Schott Schott SiO2 mass% 79.7 63.0 35.2 35.4 Al2O3 mass% 3.1 8.0 4.6 3.8 B2O3 mass% 10.3 20.0 12.2 13.4 MgO mass% 0.9 – – – CaO mass% 0.8 – – – BaO mass% 3.7 2.5 46.9 – Na2O mass% 5.2 3.5 – – K2O mass% – 3.0 – – PbO mass% – – – 43.8 Cs2O mass% – – – 3.55 Thermal expansion coefficient, a 106 K1 3.3 4.8 7.4 5.1 Refractive index, n – 1.473 1.49 1.613 1.61 Transformation temperature, Tg C 530 500 658 475 Bending strength MPa 80 52 57 50 E-modulus GPa 63 45 84 58 364 D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373
4000 Nextel 440 fiber 0100200300400500600700800900100011001200130014001500 Fig. 1. Single fiber tensile strength at room temperature of differently pre-treated(5 h in air, different temperatures)S-glass-and Nextel 440-fibers. 40nm BN+110nmTIo2 40nm BN+40nmTIO2 0onm BN+40 nmT:o2 desized 40nm BN+110nmT:O2 desized 1500 2000 2500 Fig. 2. Single fiber tensile strength of differently coated Nextel 440-fibers, not pre-treated and 5 h heated up to 750C[14] left-hand corner and is continued clockwise. The starting blowing air and then pulled by rolls through a slurry bath. materials(at the top, left-hand side) are, on the one hand, The slurry consists only of the glass powder and distilled differently prepared rovings(bundles of differently coated water. During this transport process, the powder starts fibers) wound on coils. On the other hand, pulverized sticking to the surface of the fibers. In this way, the rovings matrix glass is employed. Then, the rovings are infiltrated transport the matrix glass powder out of the slurry by by the glass powder(at the top, in the middle). For this, adhesion. Then, they are reeled up in parallel on a capstan the fibers of the rovings are separated from each other by The resulting layers are now cut and removed from the
left-hand corner and is continued clockwise. The starting materials (at the top, left-hand side) are, on the one hand, differently prepared rovings (bundles of differently coated fibers) wound on coils. On the other hand, pulverized matrix glass is employed. Then, the rovings are infiltrated by the glass powder (at the top, in the middle). For this, the fibers of the rovings are separated from each other by blowing air and then pulled by rolls through a slurry bath. The slurry consists only of the glass powder and distilled water. During this transport process, the powder starts sticking to the surface of the fibers. In this way, the rovings transport the matrix glass powder out of the slurry by adhesion. Then, they are reeled up in parallel on a capstan. The resulting layers are now cut and removed from the Fig. 1. Single fiber tensile strength at room temperature of differently pre-treated (5 h in air, different temperatures) S-glass- and Nextel 440-fibers. Fig. 2. Single fiber tensile strength of differently coated Nextel 440-fibers, not pre-treated and 5 h heated up to 750 C [14]. D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373 365
Composites: Part B 39(2008)362-373 170nm 2,3Hm 207nm 2,3μm Coated fiber After thermal annealing. Fig 3. Topography of the Nextel 440-fiber, coated with 30 nm BN and 30 nm TiO2 [14(AFM picture). Fig. 4. Principle of composites preparation surface of the capstan. Next, the disks are stacked and Table 3 middle)shows a vacuum furnace to provide reducing or Fiber dirction bonlcar' es preparation dried(Fig. 4, on the right). Hot pressing is carried out in an INSTRON installation. Fig. 4(at the bottom, in the doo of the matrix glass <70um 8-35% also neutral atmosphere during hot pressing. For the Slurry Deionized water without additives matrix glass 8650, which requires oxidizing atmospherePressure during hot pressing, an additional furnace(not shown here) Temperature is available which can be inserted in the hot press. In Fig4 N-SK4-matrix 850°C (down on the left), several composite disks are shown: In Duran-matrix 850°C the middle, a black one(1) with C-coated fibers, then two white ones(2) whose optical and thermal properties of 8650-matrix the fibers and the matrix glass are not adapted, and four Atmosphere translucent disks(3)with differently -in any case better N-SK4-matrix roperties Table 3 shows an overview of the manufacturing param 8650matrix eters employed. The hot pressing temperature is mainly influenced by the transformation temperature Tg of results presented in this paper were all obtained in tests car matrix glass. The pressure varied from 5 to 15 MPa ried a pressure of 5 MPa
surface of the capstan. Next, the disks are stacked and dried (Fig. 4, on the right). Hot pressing is carried out in an INSTRON installation. Fig. 4 (at the bottom, in the middle) shows a vacuum furnace to provide reducing or also neutral atmosphere during hot pressing. For the matrix glass 8650, which requires oxidizing atmosphere during hot pressing, an additional furnace (not shown here) is available which can be inserted in the hot press. In Fig. 4 (down on the left), several composite disks are shown: In the middle, a black one (1) with C-coated fibers, then two white ones (2) whose optical and thermal properties of the fibers and the matrix glass are not adapted, and four translucent disks (3) with differently – in any case better adapted – properties. Table 3 shows an overview of the manufacturing parameters employed. The hot pressing temperature is mainly influenced by the transformation temperature Tg of the matrix glass. The pressure varied from 5 to 15 MPa. The results presented in this paper were all obtained in tests carried out at a pressure of 5 MPa. Fig. 3. Topography of the Nextel 440-fiber, coated with 30 nm BN and 30 nm TiO2 [14] (AFM picture). Fig. 4. Principle of composites preparation. Table 3 Parameters of the composites preparation d90 of the matrix glass <70 lm Fiber volume content 8–35% Fiber direction Unidirectional Slurry Deionized water without additives Pressure 5 Mpa Temperature N-SK 4-matrix 850 C Duran-matrix 850 C 756-matrix 700 C 8650-matrix 600 C Atmosphere N-SK 4-matrix Vacuum Duran-matrix Vacuum 756-matrix Vacuum 8650-matrix Oxidizing 366 D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373
esults CVD-Carbon(45nm, t=25MPa) 4.1. Experiments with C-coated fibers The tests and results obtained are described in greater detail in the doctoral thesis submitted by the co-author Leutbecher [13]. The composites are, of course, black and opaque. The aim of these experiments was to gain experi 55nm, t=34MPa ence with the generation of fiber pull-out and with the damage-tolerant behavior. Here, two composites are --- CVD-Carbon (25nm,= S-glass fibers/C-layer/756 glass matrix. Nextel 440 fibers/C-layer/ Duran glass matrix. Fig. 6. Stress-strain-curves of differently carbon coated Nextel 440-fibers in a Duran glass matrix In the case of the black composite, the refractive indices and their adaptation do not play any role at all. In both also show the brittle fracture behavior of pure(non-rein- cases. th Terence of the thermal expansion coefficients forced )hot-pressed matrix glass powder. One can clearly is A%Xp-M#0. One could expect that during cooling after recognize the typical damage-tolerant behavior when C- hot pressing, the 756 glass clamps the S-glass fibers, thus coated fibers are employed. The composites contain a preventing them from pulling out. The expansion difference portion of fibers ranging from 30 to 35 vol%.The more between the Nextel 440 fibers and Duran glass is just the homogeneous and dense the coatings on the fibers are, opposite, which could support the fiber pull-out. Surpris- the better their mechanical properties will be. Therefore, abi, we found fiber pull-out in both cases, which is prob- CVD-carbon layers bring about better results of the com- due to the high elasticity and the good sliding posite compared with C-layers made from Novolak. The behavior of the nanocrystalline graphite layers thickness of the carbon layer influences the stress-strain The mechanical properties of the composites were tested behavior in the way as can be seen from Fig. 6. Despite this through three-point bending in our laboratory. For this, fact. however, the absolute bending strength mainly ( this machine had been used as hot press before). The test ence of the thermal expansion coefficients and particularly The lowering speed of the punch was I mm/min, the force matrix) seem to cause this effect was measured by means of a 100 N-dial gauge. The sam Fig. 7 illustrates the fiber pull-out for the combination ples presented the following dimensions: length: 50 mm, of S-glass fiber/C-layer/756 glass matrix, whereas it width: 6mm, thickness: 2 mm, distance between the shown for the combination of Nextel 440 fiber/C-layer supports: 40 mm. For one measuring value, a total of 20 duran glass matrix in Fig 8 The different thermal expansion coefficients of the fibers were determined and matrix glasses exert an influence on the shear stress Fig. 5 shows the stress-strain curves for the S-glass/756 glass matrix composites, Fig. 6 illustrates this curve for the Nextel 440/Duran glass matrix composites. Both figures CVD-Carbon (25nm, t=78MPa) hot-pressed strain [% Fig. 5. Stress-strain-curves of differently carbon coated Sglass-fibers in a Fig. 7. SEM-picture of fiber pull-out, composite S-glass-fiber/CVD
4. Results 4.1. Experiments with C-coated fibers The tests and results obtained are described in greater detail in the doctoral thesis submitted by the co-author Leutbecher [13]. The composites are, of course, black and opaque. The aim of these experiments was to gain experience with the generation of fiber pull-out and with the damage-tolerant behavior. Here, two composites are discussed: – S-glass fibers/C-layer/756 glass matrix. – Nextel 440 fibers/C-layer/Duran glass matrix. In the case of the black composite, the refractive indices and their adaptation do not play any role at all. In both cases, the difference of the thermal expansion coefficients is DaF–M 6¼ 0. One could expect that during cooling after hot pressing, the 756 glass clamps the S-glass fibers, thus preventing them from pulling out. The expansion difference between the Nextel 440 fibers and Duran glass is just the opposite, which could support the fiber pull-out. Surprisingly, we found fiber pull-out in both cases, which is probably due to the high elasticity and the good sliding behavior of the nanocrystalline graphite layers. The mechanical properties of the composites were tested through three-point bending in our laboratory. For this, the testing machine type INSTRON 4467 was used again (this machine had been used as hot press before). The test procedure is based on the DIN ENV 658-3 standard [16]. The lowering speed of the punch was 1 mm/min, the force was measured by means of a 100 N-dial gauge. The samples presented the following dimensions: length: 50 mm, width: 6 mm, thickness: 2 mm, distance between the supports: 40 mm. For one measuring value, a total of 20 samples were tested. Then, the average and the variability were determined. Fig. 5 shows the stress–strain curves for the S-glass/756 glass matrix composites, Fig. 6 illustrates this curve for the Nextel 440/Duran glass matrix composites. Both figures also show the brittle fracture behavior of pure (non-reinforced) hot-pressed matrix glass powder. One can clearly recognize the typical damage-tolerant behavior when Ccoated fibers are employed. The composites contain a portion of fibers ranging from 30 to 35 vol%. The more homogeneous and dense the coatings on the fibers are, the better their mechanical properties will be. Therefore, CVD-carbon layers bring about better results of the composite compared with C-layers made from Novolak. The thickness of the carbon layer influences the stress–strain behavior in the way as can be seen from Fig. 6. Despite this fact, however, the absolute bending strength mainly depends on the fibers and matrix glasses used. The difference of the thermal expansion coefficients and particularly their signs (clamping of the S-glass fiber by the 756 glass matrix) seem to cause this effect. Fig. 7 illustrates the fiber pull-out for the combination of S-glass fiber/C-layer/756 glass matrix, whereas it is shown for the combination of Nextel 440 fiber/C-layer/ Duran glass matrix in Fig. 8. The different thermal expansion coefficients of the fibers and matrix glasses exert an influence on the shear stress Fig. 5. Stress–strain-curves of differently carbon coated S-glass-fibers in a 756 glass matrix. Fig. 6. Stress–strain-curves of differently carbon coated Nextel 440-fibers in a Duran glass matrix. Fig. 7. SEM-picture of fiber pull-out, composite S-glass-fiber/CVDCarbon/756 glass matrix. D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373 367
368 D. Hiilsenberg et al Composites: Part B 39(2008) 4.2. Mechanical properties of the Nextel 440 fibers/756 glass matrix composites If Nextel 440 fibers and 756 glass as matrix are used, the sign of the difference of the thermal expansion coefficients becomes inverted compared with the S-glass fibers. Theo- retically, a fiber pull-out can already be expected here Now, no graphite-like carbon is being used any more as sliding layer, but boron nitride(BN) instead. Will it react with the fiber and/or the matrix? We did not find any reac- tions between Nextel 440 fibers and the Bn layers. How- ever,any chemical reactions do take place between the Bn layers and the matrix glasses. Tiny bubbles of a size of a few nanometers were found in the interface between BN and the matrix(see Section 4.5, Fig. 18). They cause an internal light scattering. In combination with the Fig8. SEM-picture of fiber pull-out, composite Nextel 440-fiber/CVD. absorption of light by atomic boron(which resulted from Carbon/Duran glass matrix(35% fiber volume content) the chemical reactions), both effects will cause a reduction which can be transmitted between the fiber and the matrix the bN layer and t r er to prevent the reactions between when being under load conditions. These shear stresses with a second fiber coating, in our case TiO2, to protect it. may either support(when they are very low) or prevent Although not all transmission problems can be solved in (when they are high) the fiber pull-out. The method of this way, this measure will not only bring about a consid- measuring them in the interface between fiber and matrix erable improvement in the optical properties, but will also is described in detail by Kuntz [17]. The measurements influence the mechanical properties positively. Therefore, were made by Leutbecher together with Kuntz at the Uni- most of the composites were prepared using double-coated versity of Bremen. The schematic layout for the"push-in"(BN/TiO2) Nextel 440 fibers test can be seen from Fig 9. The shear stresses t measured Fig. 10 shows the stress-strain curves of composites between the matrix and the fibers are also represented in made of differently coated Nextel 440 fibers and 756 glass Figs. 5 and 6. As expected, t is generally higher for the matrix. In principle, these curves are similar to those repre composite S-glass fibers/756 glass matrix compared with sented by Fig. 5. However, instead of the C-coating, there the composites Nextel 440 fibers/Duran glass matrix. is a BN/TiO2 intermediate layer, which leads to a white Therefore, the fiber pull-out is more distinct in the latter composite(see also Fig. 4, down on the left(2). In com- parison with Fig. 5, the ultimate fracture strain is much higher,which is due to AF_M=+0.5x 10-6K-.The more the fibers contract during cooling after hot pressing compared with the matrix glass, the more clearly a gap will develop between them, which facilitates the sliding of the fibers in the matrix. The maximum stress sustained will not differ that greatly if C-coated S-glass fibers(Fig. 5)or BN/TiO2-coated Nextel 440 fibers are used( Fig. 10) Indenter As expected, the non-reinforced hot-pressed 756 matrix class(Fig. 10, curve 1)presents again-also in this new test-nearly the same brittle behavior as in Fig. 5. Its rein- forcement with desized Nextel 440 fibers(curve 2)results in a higher E-modulus and only a slight improvement strength. Due to the fact that there is no sliding lay Matrix and that a direct reaction takes place between the fib and the matrix, also this composite will become brittle If a fiber pull-out takes place, the exact position of the Fiber photograph of a broken composite. The fiber pull-out can clearly be recognized. Fig. 12 illustrates the transition between the fiber and the matrix. The fiber bN, TIO and also the matrix can clearly be distinguished, with ther flaws nor other defects being detected. The EDX anal- Fig.9. Principle of the push-in test [17] ysis carried out in well-chosen points on the surface of the
which can be transmitted between the fiber and the matrix when being under load conditions. These shear stresses may either support (when they are very low) or prevent (when they are high) the fiber pull-out. The method of measuring them in the interface between fiber and matrix is described in detail by Kuntz [17]. The measurements were made by Leutbecher together with Kuntz at the University of Bremen. The schematic layout for the ‘‘push-in’’ test can be seen from Fig. 9. The shear stresses s measured between the matrix and the fibers are also represented in Figs. 5 and 6. As expected, s is generally higher for the composite S-glass fibers/756 glass matrix compared with the composites Nextel 440 fibers/Duran glass matrix. Therefore, the fiber pull-out is more distinct in the latter composite. 4.2. Mechanical properties of the Nextel 440 fibers/756 glass matrix composites If Nextel 440 fibers and 756 glass as matrix are used, the sign of the difference of the thermal expansion coefficients becomes inverted compared with the S-glass fibers. Theoretically, a fiber pull-out can already be expected here. Now, no graphite-like carbon is being used any more as sliding layer, but boron nitride (BN) instead. Will it react with the fiber and/or the matrix? We did not find any reactions between Nextel 440 fibers and the BN layers. However, any chemical reactions do take place between the BN layers and the matrix glasses. Tiny bubbles of a size of a few nanometers were found in the interface between BN and the matrix (see Section 4.5, Fig. 18). They cause an internal light scattering. In combination with the absorption of light by atomic boron (which resulted from the chemical reactions), both effects will cause a reduction in transmission. In order to prevent the reactions between the BN layer and the matrix glasses, the BN was provided with a second fiber coating, in our case TiO2, to protect it. Although not all transmission problems can be solved in this way, this measure will not only bring about a considerable improvement in the optical properties, but will also influence the mechanical properties positively. Therefore, most of the composites were prepared using double-coated (BN/TiO2) Nextel 440 fibers. Fig. 10 shows the stress–strain curves of composites made of differently coated Nextel 440 fibers and 756 glass matrix. In principle, these curves are similar to those represented by Fig. 5. However, instead of the C-coating, there is a BN/TiO2 intermediate layer, which leads to a white composite (see also Fig. 4, down on the left (2)). In comparison with Fig. 5, the ultimate fracture strain is much higher, which is due to DaF–M = +0.5 · 106 K1 . The more the fibers contract during cooling after hot pressing compared with the matrix glass, the more clearly a gap will develop between them, which facilitates the sliding of the fibers in the matrix. The maximum stress sustained will not differ that greatly if C-coated S-glass fibers (Fig. 5) or BN/TiO2-coated Nextel 440 fibers are used (Fig. 10). As expected, the non-reinforced hot-pressed 756 matrix glass (Fig. 10, curve 1) presents again – also in this new test – nearly the same brittle behavior as in Fig. 5. Its reinforcement with desized Nextel 440 fibers (curve 2) results in a higher E-modulus and only a slight improvement in strength. Due to the fact that there is no sliding layer and that a direct reaction takes place between the fiber and the matrix, also this composite will become brittle. If a fiber pull-out takes place, the exact position of the sliding plane will be important. Fig. 11 shows the REM photograph of a broken composite. The fiber pull-out can clearly be recognized. Fig. 12 illustrates the transition between the fiber and the matrix. The fiber, BN, TiO2 and also the matrix can clearly be distinguished, with neither flaws nor other defects being detected. The EDX analysis carried out in well-chosen points on the surface of the Fig. 8. SEM-picture of fiber pull-out, composite Nextel 440-fiber/CVDCarbon/Duran glass matrix (35% fiber volume content). Fig. 9. Principle of the push-in test [17]. 368 D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373
D. Hiilsenberg et al/ Composites: Part B 39(2008)362-373 369 TO FVC% Inm BN-40nm TO, FVC 12% 5-40nm BN-11Onm TO,, FVC 10% △a(FM)=0.pmK Fig10.Three-point-bending-test curves of differently coated Nextel 440-fibers and 756 glass matrix: AzF-M=0.5x10-K- Matrix BN Fig. Il. SEM-picture of a fracture surface of a composite Nextel 440) 40 nm BN, 40 nm TiO/756 glass Fiber broken composite shows where the Ti+ is located, cf. Fig. 13. On the left, the Ti portion on the surface of 30 nn double-coated fiber is given(arbitrary units). The figure Fig 12. TEM-picture of the interphases in a composite Nextel 440/40 nm in the middle shows the situation on the surface of a for- BN, 20 nm Tio) /756 glass (picture made at the TU Chemnitz) merly double-coated fiber after fiber pull-out. The EDX analysis revealed that the Ti had almost completely left the fiber. The TiOz layer, which had seemingly got lost, is glass is considerably higher than that of the Nextel 440 found in the valley of the matrix glass(channel where the fiber. From this fact, it may be concluded that the matrix iber was originally running) as can be seen from Fig 13, will clamp the fibers when the hot-pressed composite is on the right. Hence, one can draw the conclusion that the cooled, thus preventing the fiber pull-out even if they are sliding or fiber pull-out takes place between the BN- and coated with BN. In this case, the composite is not dam the TiOz-layer. The TiOz-layer reacts with the matrix age-tolerant. Nevertheless, a composite was prepared and (probably through diffusion), or the comparatively rough tested as the refractive indices of the fiber and the matrix anatase surface prevents each kind of relative movement correspond well with each other and N-SK 4 glass does between itself and the matri not contain any lead oxide. The tensile strength of the dif- ferently coated fibers is already illustrated in Fig. 2. Only 4.3. Mechanical properties of Nextel 440-fibers/N-SK 4 glass the fibers coated with TiO2 or BN-single layers present a Matrix coMposite significantly lower tensile strength, which could have an influence on the properties of the composite. The tensile Comparing Tables I and 2 with each other, one can strength of the double-coated fibers, whose layers present immediately see that the expansion coefficient of N-sk 4 different thicknesses, differs only slightly from that of the
broken composite shows where the Ti4+ is located, cf. Fig. 13. On the left, the Ti4+ portion on the surface of a double-coated fiber is given (arbitrary units). The figure in the middle shows the situation on the surface of a formerly double-coated fiber after fiber pull-out. The EDX analysis revealed that the Ti4+ had almost completely left the fiber. The TiO2 layer, which had seemingly got lost, is found in the valley of the matrix glass (channel where the fiber was originally running) as can be seen from Fig. 13, on the right. Hence, one can draw the conclusion that the sliding or fiber pull-out takes place between the BN- and the TiO2-layer. The TiO2-layer reacts with the matrix (probably through diffusion), or the comparatively rough anatase surface prevents each kind of relative movement between itself and the matrix. 4.3. Mechanical properties of Nextel 440-fibers/N-SK 4 glass matrix composites Comparing Tables 1 and 2 with each other, one can immediately see that the expansion coefficient of N-SK 4 glass is considerably higher than that of the Nextel 440 fiber. From this fact, it may be concluded that the matrix will clamp the fibers when the hot-pressed composite is cooled, thus preventing the fiber pull-out even if they are coated with BN. In this case, the composite is not damage-tolerant. Nevertheless, a composite was prepared and tested as the refractive indices of the fiber and the matrix correspond well with each other and N-SK 4 glass does not contain any lead oxide. The tensile strength of the differently coated fibers is already illustrated in Fig. 2. Only the fibers coated with TiO2- or BN-single layers present a significantly lower tensile strength, which could have an influence on the properties of the composite. The tensile strength of the double-coated fibers, whose layers present different thicknesses, differs only slightly from that of the Fig. 10. Three-point-bending-test curves of differently coated Nextel 440-fibers and 756 glass matrix; DaF–M = 0.5 · 106 K1 . Fig. 11. SEM-picture of a fracture surface of a composite Nextel 440/ 40 nm BN, 40 nm TiO2/756 glass. Fig. 12. TEM-picture of the interphases in a composite Nextel 440/40 nm BN, 20 nm TiO2/756 glass (picture made at the TU Chemnitz). D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373 369
D. Hiilsenberg et al Composites: Part B 39(2008)362-373 led-out nber surface remained fber channel Fig 13. Ti-presence, measured by EDX, in the debonded areas of the composite Nextel 440/BN, TiO2/756 glass desized fibers. The different thicknesses of the layers do not ing from the differences in thermal expansion. Both curves seem to exert a particularly strong influence indicate a completely brittle-elastic behavior. The fiber 2 The composites are white-translucent, not transparent. being coated only with TiO2, the mechanical properties his results despite adapted refractive indices- from of the composite will not improve either. This curve will the thermally caused internal micro stresses cover the two curves described above and is therefore not The stress-strain curves of different Nextel 440 fibers/N- drawn. As the TiO2 reacts with the N-sK 4 glass, no differ sk 4 glass matrix composites can be seen in Fig 14 ence to the behavior of the desized fiber can be stated It becomes immediately evident that none of the samples ter a bn coating has been carried out, the com- breaks damage-tolerant. Furthermore, all samples present posite properties will change greatly. Here, however, the nearly the same curve ramp in the coordinate origin, i.e., effect exerted by a single BN layer(curve 2)is not as strong nearly the same E-modulus. It is determined by the as the effect exerted by a BN/TiO2 double layer(curves 3, 5 mechanical properties of the matrix glass rather than by and 6). The reasons for this are explained in the preceding the relatively low fiber volume portion. This is the bad chapter. The definitive fracture of all composites, however, point about the results obtained is brittle. If the bending stress exceeds a value of about The very good points, however, are the high tensile 180 MPa and increases further, the sample will stretch strength of the composites, the high fracture strain, and more than below this value. Depending on the type of coat the large area below the stress-strain curve ing, the fracture strain is 8-17 times higher than in the case The pure, non-reinforced hot-pressed matrix glass of non-reinforced glass, and can even reach the unexpect- (curve 1)presents a tensile strength whose value is the same edly high value of 3.5%. The explanation of this result as specified by the manufacturer for the compact glass has not completely been convincing so far. It could be, so material (cf. Table 2). The composite with desized fibers to say, a competition taking place between internal micro (curve 4) presents even worse values, which is due to chem- stresses caused by Ao or clampings, and the fiber pull-out ical reactions taking place between the fibers and the caused by the turbostratic BN layers. Over the whole per matrix during hot pressing, and to internal stresses result- iod when loading is exerted, the radial compressive stress ed fiber FVc 8% 2.40nm BN. FVC 16% 5-35nm BN-30nm TiO. FVc 4001 3-40nm BN-110nm TiO2, FVC 16% 6-100nm BN-40nm TiO2, FVC 3 50 Fig. 14. Stress-strain-curves of Nextel 440-fibers/N-SK 4-glass matrix composites. The fiber-coating and the fiber-volume-content vary
desized fibers. The different thicknesses of the layers do not seem to exert a particularly strong influence. The composites are white-translucent, not transparent. This results – despite adapted refractive indices – from the thermally caused internal micro stresses. The stress–strain curves of different Nextel 440 fibers/NSK 4 glass matrix composites can be seen in Fig. 14. It becomes immediately evident that none of the samples breaks damage-tolerant. Furthermore, all samples present nearly the same curve ramp in the coordinate origin, i.e., nearly the same E-modulus. It is determined by the mechanical properties of the matrix glass rather than by the relatively low fiber volume portion. This is the bad point about the results obtained. The very good points, however, are the high tensile strength of the composites, the high fracture strain, and the large area below the stress–strain curve. The pure, non-reinforced hot-pressed matrix glass (curve 1) presents a tensile strength whose value is the same as specified by the manufacturer for the compact glass material (cf. Table 2). The composite with desized fibers (curve 4) presents even worse values, which is due to chemical reactions taking place between the fibers and the matrix during hot pressing, and to internal stresses resulting from the differences in thermal expansion. Both curves indicate a completely brittle-elastic behavior. The fiber being coated only with TiO2, the mechanical properties of the composite will not improve either. This curve will cover the two curves described above and is therefore not drawn. As the TiO2 reacts with the N-SK 4 glass, no difference to the behavior of the desized fiber can be stated. Only after a BN coating has been carried out, the composite properties will change greatly. Here, however, the effect exerted by a single BN layer (curve 2) is not as strong as the effect exerted by a BN/TiO2 double layer (curves 3, 5 and 6). The reasons for this are explained in the preceding chapter. The definitive fracture of all composites, however, is brittle. If the bending stress exceeds a value of about 180 MPa and increases further, the sample will stretch more than below this value. Depending on the type of coating, the fracture strain is 8–17 times higher than in the case of non-reinforced glass, and can even reach the unexpectedly high value of 3.5%. The explanation of this result has not completely been convincing so far. It could be, so to say, a competition taking place between internal micro stresses caused by Da or clampings, and the fiber pull-out caused by the turbostratic BN layers. Over the whole period when loading is exerted, the radial compressive stress Fig. 13. Ti4+-presence, measured by EDX, in the debonded areas of the composite Nextel 440/BN, TiO2/756 glass. Fig. 14. Stress–strain-curves of Nextel 440-fibers/N-SK 4-glass matrix composites. The fiber-coating and the fiber-volume-content vary. 370 D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373
D. Hiilsenberg et al/ Composites: Part B 39(2008)362-373 developing during cooling is acting on the fibers. These are well distributed micro stresses which. at first. will not exceed the strength of the composite at no point. As the portion of BN-coated fibers in the composites amounts to 16-17 vol% only, the E-modulus of the matrix glass will p mechanical behavior will only be linear(Hooke 's law.If, 9 101 be predominant. Up to a bending stress of 180 MPa, the however, the bending stress exceeds this value, then the the fibers to slide despite the radial compressive stress. At005 what stress sliding sets in will depend on the thickness of fiber coating Fig. 16. Stress-strain- curves of two samples of a Nextel 440-fiber/25 nm The fiber pull-out is illustrated in Fig. 15 and does not BN-layer/8650 glass matrix composite with 15 vol% fiber content. seem to differ from the damage-tolerant behavior of other composites. The fracture aspect of the matrix, however, shows another interesting effect. The very small lumps of Due to the fact that the financial support of the research glass look like broken, hardened single-sheet safety glass. made from double-coated Nextel 440 fibers and an 8650 Therefore, the following preliminary conclusion concern- glass matrix will be have to be investigated further in the ing the failure mechanism of these special composites can future. They allow better mechanical properties to be be drawn: The fibers as well as the micro hardened glass matrix will break at the same time when stress is high xpected and strain very strong 4.5. Optical properties of the composites 4.4. Mechanical properties of Nextel 440-fibers/8650 matrix glass composites As described above, composites with Because of identical values of x and n for the fibers and C-coated fibers are black le matrix, these composites should be both damage-toler--BN-coated Nextel 440 fibers and 756 glass matrix are ant and transparent in the case of a bn coating of the white, i.e., because of the non-adapted optical refractive fibers. Fig. 16 shows the stress-strain curves for two test index they are not transparent, and samples of one and the same composite. The two curves- BN-coated Nextel 440 fibers and N-SK 4 glass matrix prove the variability of the mechanical properties(extre- are translucent white although their refractive indices mely deviating curves were chosen). The fibers seem to be are identical; the internal micro stresses(A), however clamped slightly, which increases strength in comparison cause light scattering with Fig. 10. The fiber pull-out occurs only in a very restricted stress range. From this, it follows that a sligl The composites with an 8650 glass matrix should now be difference in a, 1.e. F >M, will support damage tolerance transparent as both a and n for the fibers and the matrix better than completely identical values of a of both the fiber glass are identical. Indeed, they really are transparent, and the matrix but transparency is not as good as expected. As already pointed out above, the investigations were only carried out with fibers with a single bn-coating. The results of the wavelength-dependent transmission obtained for com- posites with an N-sK 4 glass matrix as well as an 8650 glass matrix are shown in Fig. 17. In the case of the N-sK glass matrix, transmission will increase in the visible wavelength range by up to 20%if the fiber is coated with TiO2, or only by up to 15% in the case of a BN/TiO2 double-layer coating. If 8650 glass is used as matrix, the transmission of the composite will be considerably higher already in the case of a fiber with a sin gle BN coating, increasing up to 45%. The composite wil be colorless transparent. However, colorless actually stand in contradiction to the behavior of the curve in Fig. 17 This fact can only be explained on the basis of the effects exerted by the Mie or the rayleigh scattering Mie scatter Fig. 15. Fiber pull-out and matrix fracture in a Nextel 440/BN, TiO,/ ing will occur if heterogeneities of the order of magnitude of the wavelength used are found in the glass, i.e., ranging
developing during cooling is acting on the fibers. These are well distributed micro stresses which, at first, will not exceed the strength of the composite at no point. As the portion of BN-coated fibers in the composites amounts to 16–17 vol% only, the E-modulus of the matrix glass will be predominant. Up to a bending stress of 180 MPa, the mechanical behavior will only be linear (Hooke’s law). If, however, the bending stress exceeds this value, then the BN interface between the fibers and the matrix will enable the fibers to slide despite the radial compressive stress. At what stress sliding sets in will depend on the thickness of fiber coating. The fiber pull-out is illustrated in Fig. 15 and does not seem to differ from the damage-tolerant behavior of other composites. The fracture aspect of the matrix, however, shows another interesting effect. The very small lumps of glass look like broken, hardened single-sheet safety glass. Therefore, the following preliminary conclusion concerning the failure mechanism of these special composites can be drawn: The fibers as well as the micro hardened glass matrix will break at the same time when stress is high and strain very strong. 4.4. Mechanical properties of Nextel 440-fibers/8650 matrix glass composites Because of identical values of a and n for the fibers and the matrix, these composites should be both damage-tolerant and transparent in the case of a BN coating of the fibers. Fig. 16 shows the stress–strain curves for two test samples of one and the same composite. The two curves prove the variability of the mechanical properties (extremely deviating curves were chosen). The fibers seem to be clamped slightly, which increases strength in comparison with Fig. 10. The fiber pull-out occurs only in a very restricted stress range. From this, it follows that a slight difference in a, i.e. aF > aM, will support damage tolerance better than completely identical values of a of both the fiber and the matrix. Due to the fact that the financial support of the research work carried out so far has come to a stop, composites made from double-coated Nextel 440 fibers and an 8650 glass matrix will be have to be investigated further in the future. They allow better mechanical properties to be expected. 4.5. Optical properties of the composites As described above, composites with – C-coated fibers are black, – BN-coated Nextel 440 fibers and 756 glass matrix are white, i.e., because of the non-adapted optical refractive index they are not transparent, and – BN-coated Nextel 440 fibers and N-SK 4 glass matrix are translucent white although their refractive indices are identical; the internal micro stresses (Da), however, cause light scattering. The composites with an 8650 glass matrix should now be transparent as both a and n for the fibers and the matrix glass are identical. Indeed, they really are transparent, but transparency is not as good as expected. As already pointed out above, the investigations were only carried out with fibers with a single BN-coating. The results of the wavelength-dependent transmission obtained for composites with an N-SK 4 glass matrix as well as an 8650 glass matrix are shown in Fig. 17. In the case of the N-SK glass matrix, transmission will increase in the visible wavelength range by up to 20% if the fiber is coated with TiO2, or only by up to 15% in the case of a BN/TiO2 double-layer coating. If 8650 glass is used as matrix, the transmission of the composite will be considerably higher already in the case of a fiber with a single BN coating, increasing up to 45%. The composite will be colorless transparent. However, colorless actually stands in contradiction to the behavior of the curve in Fig. 17. This fact can only be explained on the basis of the effects exerted by the Mie or the Rayleigh scattering. Mie scattering will occur if heterogeneities of the order of magnitude of the wavelength used are found in the glass, i.e., ranging Fig. 15. Fiber pull-out and matrix fracture in a Nextel 440/BN, TiO2/ N-SK 4 glass composite. Fig. 16. Stress–strain-curves of two samples of a Nextel 440-fiber/25 nm BN-layer/8650 glass matrix composite with 15 vol% fiber content. D. Hu¨lsenberg et al. / Composites: Part B 39 (2008) 362–373 371