MATERIALS CIENCE EIEERIG ELSEVIE Materials Science and Engineering A 417(2006)341-347 www.elsevier.com/locate/msea Mechanical properties of oxide-fiber reinforced glass matrix composites with Bn or SnO2 interfaces D Raab,K. Pfeifer D. Hulsenberg, A R bo occaccin o,* a Technische Universitat ILmenau, Institut fiir Werkstoftechnik, Gustav-Kirchhof-Str 6, D-98693 iLmenau, germany Imperial College London, Department of Materials, Prince Consort Road, London SW7 2BP U Received in revised form 28 October 2005; accepted 3 November 2005 Abstract With the aim of developing optomechanical inorganic materials, boron nitride- coated Nextel M 440 fibers and tin oxide-coated Zen TronM glass fibers were used for manufacturing continuous and short fiber reinforced borosilicate glass matrix composites. No evident loss in tensile strength occurred for the BN-coated Nextel M fibers whereas coating the Zen TronM fiber with tin oxide caused a significant strength decrease Composites with <12 vol% of short or continuous fibers were fabricated by a standard slurry infiltration and hot-pressing process. The mechanical properties of the composites were investigated by uni-and biaxial flexural strength tests. Nextel M 440 short fiber reinforced composites with boron nitride interfaces demonstratedquasi-ductile fracture behaviour under biaxial stress loading with biaxial fracture strength values of up to 88 MPa. The boron nitride layers caused fiber pull-out during fracture of long fiber reinforced composites, leading to a fracture toughness(Ki) value of3. 3 MPam 4. In contrast to this, the Zen Tron M glass fiber reinforced glass matrix composites with SnO2 interface exhibited brittle fracture o 2005 Elsevier B.V. All rights reserved Keywords: Glass matrix composites; Oxide fibres; Optomechanical composites; Interfaces; Fracture toughness 1. Introduction composites containing oxide fibers exhibiting suitable optical and thermomechanical properties, the so-calledoptomechani- The reinforcement of glass with ceramic fibers or particles is cal composites[7-13] a common approach to improve mechanical, thermo-mechanical The successful development of glass matrix composites and functional properties of this material [1-3]. Usually carbon with effective reinforcement and improved mechanical prop- or SiC fibers are used as reinforcing elements [1]. The resulting erties strongly depends on achieving optimal properties of glass matrix composites are useful lightweight materials, e.g. fiber-matrix interfaces. The responsible mechanisms for rein- for the manufacturing of specialized machine parts, such as hot forcement and enhancement of fracture toughness in these brittle glass and metals handling tools [4], or for other applications in matrix composites, such as crack bridging, crack deflection, fiber the aerospace machinery and automotive sectors [1-3, 5, 6] debonding or fiber pull-out are well understood [1, 2]. In the nfortunately, in the course of the manufacturing process the case of oxide fiber reinforced glass matrix composites, special outstanding glass property itself-optical transparency -is lost attention should be placed ( besides the optical compatibility of because of the black carbon or SiC fibers used or due to the fibers and glass matrix)on engineering the microstructure of different optical properties of the composite constituents. The the interfaces, due to the possibility of extensive chemical reac- formation of carbon-rich interphases due to reactions between tions between the fibers and the silicate matrix during processing fibers and matrix is another reason leading to opaque com- [14-16]. Many different interfaces have been studied to deter- posites [3-5]. Therefore, current research efforts are focused mine their suitability in glass matrix composites reinforced by on the development of transparent or translucent glass matrix oxide fibers [ 12-16]. The main role of the interfaces is to prevent a strong bonding of fiber and matrix during the technological steps of composite manufacturing, and thus to enable a sliding ng author.Tel:+44207594673l;fax:+442075946757 process between fiber and matrix during fracture From the opti- E-mail address: a boccaccini@ imperial ac uk(A R. Boccaccini) cal point of view, besides meeting the mechanical and thermal 0921-5093/S-see front matter Elsevier B V. All rights reserved doi:10.1016 J.msea.2005.11001
Materials Science and Engineering A 417 (2006) 341–347 Mechanical properties of oxide-fiber reinforced glass matrix composites with BN or SnO2 interfaces D. Raab a, K. Pfeifer a, D. Hulsenberg ¨ a, A.R. Boccaccini b,∗ a Technische Universit ¨at Ilmenau, Institut f ¨ur Werkstofftechnik, Gustav-Kirchhoff-Str. 6, D-98693 Ilmenau, Germany b Imperial College London, Department of Materials, Prince Consort Road, London SW7 2BP, UK Received in revised form 28 October 2005; accepted 3 November 2005 Abstract With the aim of developing optomechanical inorganic materials, boron nitride-coated NextelTM 440 fibers and tin oxide-coated ZenTronTM glass fibers were used for manufacturing continuous and short fiber reinforced borosilicate glass matrix composites. No evident loss in tensile strength occurred for the BN-coated NextelTM fibers whereas coating the ZenTronTM fiber with tin oxide caused a significant strength decrease. Composites with ≤12 vol% of short or continuous fibers were fabricated by a standard slurry infiltration and hot-pressing process. The mechanical properties of the composites were investigated by uni- and biaxial flexural strength tests. NextelTM 440 short fiber reinforced composites with boron nitride interfaces demonstrated “quasi”-ductile fracture behaviour under biaxial stress loading with biaxial fracture strength values of up to 88 MPa. The boron nitride layers caused fiber pull-out during fracture of long fiber reinforced composites, leading to a fracture toughness (Kic) value of 3.3 MPa m1/2. In contrast to this, the ZenTronTM glass fiber reinforced glass matrix composites with SnO2 interface exhibited brittle fracture behaviour. © 2005 Elsevier B.V. All rights reserved. Keywords: Glass matrix composites; Oxide fibres; Optomechanical composites; Interfaces; Fracture toughness 1. Introduction The reinforcement of glass with ceramic fibers or particles is a common approach to improve mechanical, thermo-mechanical and functional properties of this material [1–3]. Usually carbon or SiC fibers are used as reinforcing elements [1]. The resulting glass matrix composites are useful lightweight materials, e.g. for the manufacturing of specialized machine parts, such as hot glass and metals handling tools [4], or for other applications in the aerospace machinery and automotive sectors [1–3,5,6]. Unfortunately, in the course of the manufacturing process the outstanding glass property itself – optical transparency – is lost because of the black carbon or SiC fibers used or due to the different optical properties of the composite constituents. The formation of carbon-rich interphases due to reactions between fibers and matrix is another reason leading to opaque composites [3–5]. Therefore, current research efforts are focused on the development of transparent or translucent glass matrix ∗ Corresponding author. Tel.: +44 207 5946731; fax: +44 207 5946757. E-mail address: a.boccaccini@imperial.ac.uk (A.R. Boccaccini). composites containing oxide fibers exhibiting suitable optical and thermomechanical properties, the so-called ‘optomechanical composites’ [7–13]. The successful development of glass matrix composites with effective reinforcement and improved mechanical properties strongly depends on achieving optimal properties of fiber–matrix interfaces. The responsible mechanisms for reinforcement and enhancement of fracture toughness in these brittle matrix composites, such as crack bridging, crack deflection, fiber debonding or fiber pull-out are well understood [1,2]. In the case of oxide fiber reinforced glass matrix composites, special attention should be placed (besides the optical compatibility of fibers and glass matrix) on engineering the microstructure of the interfaces, due to the possibility of extensive chemical reactions between the fibers and the silicate matrix during processing [14–16]. Many different interfaces have been studied to determine their suitability in glass matrix composites reinforced by oxide fibers[12–16]. The main role of the interfaces is to prevent a strong bonding of fiber and matrix during the technological steps of composite manufacturing, and thus to enable a sliding process between fiber and matrix during fracture. From the optical point of view, besides meeting the mechanical and thermal 0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2005.11.001
D Raab et al. Materials Science and Engineering A 417(2006)341-347 demands, the interfaces should not influence optical wave guid- Temperatur→ Pressure ing in the visible wavelength, which is essential to achieve a transparent, or at least translucent, material [9-13] In this study, the role of two different interfaces in oxide fiber reinforced borosilicate glass matrix composites was exam- g ined and discussed. Zen Tron M glass fibers were coated with tin dioxide and Nextel TM 440 fibers with boron nitride via chem- 8400 ical vapour deposition(CVD), as described elsewhere [17, 18] E300 The fibers were characterized by tensile strength testing and A200 987654321 atomic force microscopy(AFM). a borosilicate glass matrix (type 756f) was reinforced with desized or coated fibers, both in chopped and continuous form. The fiber volume content was kept low(<12 vol%)to allow for comparison with ear- Time [min] lier related investigations on similar glass matrix composites Fig. 1. Hot-pressing conditionsemployed for optimal composite manufacturing tin-tetramethyl in the presence of oxygen at 450C. Details of 2. Experimental procedures the Cvd coating procedure are given elsewhere [ 17,1 Fiber tensile strength tests were carried out using an 2. Materials INSTRON 4467 facility with a 10N load cell according to DIN Norm ( din V ENV 1007-6, 2002). At least 50 desized and The properties of the glass matrix and of the fibers used as coated fibers were measured for each condition reinforcements are shown in Table 1. because of the refrac. The surface characteristics of desized and coated fibers were ive index mismatch between fibers and matrix, translucent but determined by atomic force microscopy (TMX 2000 Explorer not transparent composites were expected. Moreover, due to the SPM, TopoMetrix). The measurements were carried out in the thermal expansion mismatch between fiber and matrix in com- contact mode under atmospheric conditions posites reinforced with Zen Tronglass fiber, a residual micro- compressive stress field in radial direction should develop in the 2.3. Composite fabrication and characterization atrix. while in the case of Nextel TM fibers. radial microtensile stresses are expected The fiber rovings were infiltrated using a standard slurry infil- tration process, which has been described in the literature [5] 2.2. Fiber coating and characterization Glass particles of size doo <70 um were used. For short fiber composites, the infiltrated rovings were cut to a mean length Coating of the Next! TM fiber with boron nitride films of of 0.3 mm. Composite manufacturing was carried out by hot thickness 40 or 150 nm and of the Zen Tron M glass fiber with pressing at 750 C applying a pressure of 5 MPa for 15 min a tin dioxide film of 70 nm thickness was carried out by con- Fig. I shows the temperature-pressure-time schedule of the hot tinuous CVD under atmospheric pressure in a vertical hot-wall pressing cycle. A short holding time at maximum temperature reactor Before coating, the tiber roving was thermally desized. and the application of the uniaxial pressure during cooling main- The deposition of boron nitride was carried out by reaction tained until the beginning of the glass transformation range were of boric acid-trimethy late and ammonia at 1100C. analo- found to be advantageous for achieving composites with very gously, tin dioxide was obtained by thermal decomposition of low porosity (1%) Table 2 presents an overview of the different composites ricated. The volume fraction of fibers of all composites Table I Properties of the glass matrix and fibers used lass matrix. Fi 三 thermal expansion coefficients of the manufactured composites of fiber a20,,0°c(ppm/K) Fiber concentration silicate) Continuous Short Continuous Short fiber Manufactured by TELUX AGY Refractive index, nD 1.523 Nextel m 440 desized 4.8 9 Thermal expansion coefficient Nextel M 440. 40 nm 5.15 NA (20,,300°)(10-6K-) BN coatin mation temperature(C) 500 Nexte//M,s 440,150mm4.97 s modulus( GPa) BN coating ng strength(MPa) Zen Tron desized 4.36 strength(GPa) 52- 207 Fiber diameter(um) 10-12 SnO coating
342 D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 demands, the interfaces should not influence optical wave guiding in the visible wavelength, which is essential to achieve a transparent, or at least translucent, material [9–13]. In this study, the role of two different interfaces in oxide fiber reinforced borosilicate glass matrix composites was examined and discussed. ZenTronTM glass fibers were coated with tin dioxide and NextelTM 440 fibers with boron nitride via chemical vapour deposition (CVD), as described elsewhere [17,18]. The fibers were characterized by tensile strength testing and atomic force microscopy (AFM). A borosilicate glass matrix (type 756f) was reinforced with desized or coated fibers, both in chopped and continuous form. The fiber volume content was kept low (<12 vol%) to allow for comparison with earlier related investigations on similar glass matrix composites [7]. 2. Experimental procedures 2.1. Materials The properties of the glass matrix and of the fibers used as reinforcements are shown in Table 1. Because of the refractive index mismatch between fibers and matrix, translucent but not transparent composites were expected. Moreover, due to the thermal expansion mismatch between fiber and matrix, in composites reinforced with ZenTronTM glass fiber, a residual microcompressive stress field in radial direction should develop in the matrix, while in the case of NextelTM fibers, radial microtensile stresses are expected. 2.2. Fiber coating and characterization Coating of the NextelTM fiber with boron nitride films of thickness 40 or 150 nm and of the ZenTronTM glass fiber with a tin dioxide film of 70 nm thickness was carried out by continuous CVD under atmospheric pressure in a vertical hot-wall reactor. Before coating, the fiber roving was thermally desized. The deposition of boron nitride was carried out by reaction of boric acid–trimethylate and ammonia at 1100 ◦C. Analogously, tin dioxide was obtained by thermal decomposition of Table 1 Properties of the glass matrix and fibers used Glass matrix, 756f Fiber, ZenTronTM (S-glass) Fiber, NextelTM 440 (alumino silicate) Manufactured by TELUX AGY 3M Refractive index, nD 1.49 1.523 1.616 Thermal expansion coefficient (20, ..., 300 ◦C) (10−6 K−1) 4.8 2.8 5.3 Transformation temperature (◦C) 500 816 – Young’s modulus (GPa) 45 87 190 Bending strength (MPa) 52 – – Tensile strength (GPa) – 4.58 2.07 Fiber diameter (m) – 10 10–12 Fig. 1. Hot-pressing conditions employed for optimal composite manufacturing. tin-tetramethyl in the presence of oxygen at 450 ◦C. Details of the CVD coating procedure are given elsewhere [17,18]. Fiber tensile strength tests were carried out using an INSTRON 4467 facility with a 10 N load cell according to DIN Norm (DIN V ENV 1007-6, 2002). At least 50 desized and coated fibers were measured for each condition. The surface characteristics of desized and coated fibers were determined by atomic force microscopy (TMX 2000 Explorer SPM, TopoMetrix). The measurements were carried out in the contact mode under atmospheric conditions. 2.3. Composite fabrication and characterization The fiber rovings were infiltrated using a standard slurry infiltration process, which has been described in the literature [5]. Glass particles of size d90 < 70m were used. For short fiber composites, the infiltrated rovings were cut to a mean length of 0.3 mm. Composite manufacturing was carried out by hot pressing at 750 ◦C applying a pressure of 5 MPa for 15 min. Fig. 1 shows the temperature–pressure–time schedule of the hotpressing cycle. A short holding time at maximum temperature and the application of the uniaxial pressure during cooling maintained until the beginning of the glass transformation range were found to be advantageous for achieving composites with very low porosity (1%). Table 2 presents an overview of the different composites fabricated. The volume fraction of fibers of all composites was Table 2 Measured thermal expansion coefficients of the manufactured composites Condition of fiber surfaces α20, ..., 300 ◦C (ppm/K) Fiber concentration (vol%) Continuous fiber Short fiber Continuous fiber Short fiber NextelTM 440 desized 4.81 4.74 9 9 NextelTM 440, 40 nm BN coating 5.15 N/A 12 N/A NextelTM 440, 150 nm BN coating 4.97 4.50 12 6.5 ZenTronTM, desized 4.36 4.53 9 9 ZenTronTM, 70 nm SnO2 coating 4.30 4.18 5 5
D Raab et al. /Materials Science and Engineering A 417 (2006)341-347 23 0 nm 2,3μm0pm 150 nm B desized s-glass fiber 70 nm tin dioxide Zen Tron 1g. 4. AFM images of fiber surfaces: desized and tin oxide-coated Zen tron Fig. 2. Comparison of single fiber tensile strength data for desized and coated glass fibers NextelM 440 fibers and for desized and coated Zen TronM fibers(gauge length measured Fig. 2 shows the results of these tests for Nextel M and in the range of 5-12%. Discs of 50 mm in diameter and -3 mm zen tron TM fibers for desized NextelTm 440 fibers as well as thickness were fabricated. Microstructural examination was car- for the 40 and 150 nm BN-coated NextelTM fibers. no significant ried out by scanning electron microscopy (SEM)(CamScan 44, differences in fiber tensile strength values were found. The data Cambridge Instruments). From hot-pressed discs, rectangular scattering is also comparable. However, a pronounced decrease test bars of dimensions 40 mm x 6 mm x 3 mm were cut and pol- in fiber tensile strength was found for the tin dioxide-coated ished. Three-point bending strength tests were performed using Zen Tron TM glass fiber in comparison to the desized fiber.Fiber an INSTRON 4467 facility with a 2N load cell according to degradation due to the thermal treatment during the CVDcoating DIN Norm DIN ENV 658-3. At least five samples were tested process is excluded as a cause of strength loss because of the for each condition. For biaxial flexural strength testing discs relatively high strength of the desized fiber treated at nearly the were thinned to I mm thickness and polished with SiC-paper same temperature as that used for the CVD coating process. A ( um). The apparatus used for measuring the biaxial strength diffusion of tin ions into the glass fiber during the coating process test has been described elsewhere [19]. At least 10 samples were cannot be ruled out however and it should be considered further used for each composition and the results were averaged (see Section 3.2 below). Chemical reactions at the Sno/glass The thermal expansion coefficient of composites was mea- interface and locally stressed regions on the fiber surface as sured using a dilatometer(NETZSCH TMA 402). Test bars of a result of the thermal treatment during CVD might be other dimensions 3 mm x 3 mm x 25 mm were used for the measure measure- contributing reasons leading to the decrease of tensile strength of coated Zen TronM fibers Preliminary assessment of fracture toughness(KI Figs. 3 and 4 show AFM images of desized and coated obtained by the indentation technique, using loads of IoN. The fibers surfaces for both NextelTM and ZenTronTM fibers, respec Heckel-equation[20] was used to calculate Kl from the length tively. Fig. 3 shows that the fine-grained structure of the desized of the cracks emanating from the corners of Vickers impres- Nextel TM 440 fiber becomes smudged by the CVD coating pro- cess. Grain boundaries are still visible in the Bn coating and 3. Results and discussion there is no significant difference between the structure of coat- ings of 40 and 150 nm thickness. earlier investigations have also 3.. Fiber characterization shown no significant changes in surface roughness of BN films prepared by a similar CVD coating process [18]. In some cases, e To evaluate the influence of the CVD coating process on the larger particles were found on the coated fiber surface, as shown chanical properties of fibers, the fiber tensile strength was for example on the sample image for the 40 nm coating(Fig 3) 2,3pm 0 desized Nextel 440 fiber 50 nm boron nitride 150 nm boron nitride Fig 3. AFM images of fiber surfaces: desized and boron nitride-coated Nextel M fibers
D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 343 Fig. 2. Comparison of single fiber tensile strength data for desized and coated NextelTM 440 fibers and for desized and coated ZenTronTM fibers (gauge length 26 mm). in the range of 5–12%. Discs of 50 mm in diameter and ∼3 mm thickness were fabricated. Microstructural examination was carried out by scanning electron microscopy (SEM) (CamScan 44, Cambridge Instruments). From hot-pressed discs, rectangular test bars of dimensions 40 mm × 6 mm × 3 mm were cut and polished. Three-point bending strength tests were performed using an INSTRON 4467 facility with a 2 N load cell according to DIN Norm DIN ENV 658-3. At least five samples were tested for each condition. For biaxial flexural strength testing discs were thinned to 1 mm thickness and polished with SiC-paper (3m). The apparatus used for measuring the biaxial strength test has been described elsewhere [19]. At least 10 samples were used for each composition and the results were averaged. The thermal expansion coefficient of composites was measured using a dilatometer (NETZSCH TMA 402). Test bars of dimensions 3 mm × 3 mm × 25 mm were used for the measurements. Preliminary assessment of fracture toughness (KIc) was obtained by the indentation technique, using loads of 10 N. The Heckel-equation [20] was used to calculate KIc from the length of the cracks emanating from the corners of Vickers’ impressions. 3. Results and discussion 3.1. Fiber characterization To evaluate the influence of the CVD coating process on the mechanical properties of fibers, the fiber tensile strength was Fig. 4. AFM images of fiber surfaces: desized and tin oxide-coated ZenTronTM glass fibers. measured. Fig. 2 shows the results of these tests for NextelTM and ZenTronTM fibers. For desized NextelTM 440 fibers as well as for the 40 and 150 nm BN-coated NextelTM fibers, no significant differences in fiber tensile strength values were found. The data scattering is also comparable. However, a pronounced decrease in fiber tensile strength was found for the tin dioxide-coated ZenTronTM glass fiber in comparison to the desized fiber. Fiber degradation due to the thermal treatment during the CVD coating process is excluded as a cause of strength loss because of the relatively high strength of the desized fiber treated at nearly the same temperature as that used for the CVD coating process. A diffusion of tin ions into the glass fiber during the coating process cannot be ruled out, however, and it should be considered further (see Section 3.2 below). Chemical reactions at the SnO2/glass interface and locally stressed regions on the fiber surface as a result of the thermal treatment during CVD might be other contributing reasons leading to the decrease of tensile strength of coated ZenTronTM fibers. Figs. 3 and 4 show AFM images of desized and coated fibers surfaces for both NextelTM and ZenTronTM fibers, respectively. Fig. 3 shows that the fine-grained structure of the desized NextelTM 440 fiber becomes smudged by the CVD coating process. Grain boundaries are still visible in the BN coating and there is no significant difference between the structure of coatings of 40 and 150 nm thickness. Earlier investigations have also shown no significant changes in surface roughness of BN films prepared by a similar CVD coating process [18]. In some cases, larger particles were found on the coated fiber surface, as shown for example on the sample image for the 40 nm coating (Fig. 3). Fig. 3. AFM images of fiber surfaces: desized and boron nitride-coated NextelTM fibers
D Raab et al. Materials Science and Engineering A 417(2006)341-347 Table 3 Mechanical properties of the manufactured composites(mean values) Three-point bending test Biaxial strength test Bending strength(MPa) Continuous fiber Short fiber Continuous fiber hort fiber Continuous fiber Short fiber 440 uncoated extel M 440/40nm BN coating NextelTM 440/150 nm BN coating 98 4M44 The relative error of bend ngth and biaxial strength data is 10%. The relative error of Kie data is 10%(NA, not assessed The occasional defects are thought to be caused by an uncon- 3.3. Nextel M 440 fiber reinforcement trolled coating process. Nevertheless, a turbostratic hexagonal structure was demonstrated for the boron nitride phase coat- Because of the thermal expansion coefficient mismatch ings by TEM investigations, as shown in previous investigations (a>M)in these composites, radial tensile stresses surround- ing the Nextel fibers develop. There is, however, a residual The desized Zen Tron glass fiber in Fig 4 shows a relatively hoop compressive stress field in the matrix, which is beneficial smooth surface as expected for an amorphous material. On the for the strength of the composite material contrary, the tin dioxide CVD coating exhibits a relatively rough As expected, unidirectional and short NextelM440 fiber crystalline structure in AFM imaging. Raman spectrometry and reinforced composites showed increased strength in three-point X-ray diffraction(XRD) have shown that the tin oxide coating bending tests for both 40 and 150 nm BN fiber coatings, as consists mainly of cassiterite phase [211(results not shown here). compared to uncoated fibers In the case of continuous fiber rein- Besides the changes in fiber chemistry due to interactions with forcement, the fracture toughness increased up to 3. 3 MPam /2 tin ions and the possible development of surface microstresses, for composites with 40 nm thick boron nitride interfaces. For the a mechanical aspect of fiber damaging should not be ruled out, short fiber reinforced composites, no enhancement of the frac caused by the formation of relatively large cassiterite crystals. ture toughness was observed. This is a consequence of the low This crystal growth can lead to surface stresses, which in turn fiber content in these composites(see Table 2)and due to the fact might prompt the formation of microflaws on glass fiber sur- that the basic toughening mechanisms(mainly crack deflection and crack bridging)are different from those active in contin- uous fiber composites, e.g. fiber debonding and pull-out [5,6 3. 2. Composite characterization Moreover it can be assumed that the edges of the short fibers are not coated, and therefore diffusion reactions may occur between The thermal expansion coefficients(a)determined for differ- fibers and matrix which should cause strong interface bonding ent composites containing short or continuous fiber reinforce- leading to brittle fracture In biaxial flexural strength tests, no ment are listed in Table 2. The a-values for the unidirectional significant differences between the strength values of the com- continuous fiber reinforced composites were measured in the posites with uncoated and BN-coated fibers(150 nm thick) were fiber direction. The experimental results are in broad agreement found for both continuous and short fiber reinforcement. Only with values calculated by the rule of mixtures using the a-values composites with short fiber reinforcement with a 40 nm thick BN for matrix and fibers given in Table 1 The results of the mechanical strength tests are summa- rized in Table 3. The fracture behaviour of the composites was investigated by both three-point bending and biaxial flexural tial applications of the composites as structural components, e.g. 280 rength tests. The latter test is relevant considering the poten- in architecture, where flat composite panels will be enclosed in a metal or plastic frame, thus possibly exposed to biaxial stresses due to thermal expansion mismatch of the materials involved [22]. Moreover, the biaxial test method is useful to obtain rel- ive data to compare the behaviour of conti and short fiber reinforced composites. The test allows for com- Displacement [mm] parison of the influence of the fiber orientation in continuous fiber composites and the quasi-isotropic behaviour of short fiber Fig. 5. Load-displacement curves composites regarding both mechanical strength and work of reinforced, interface: 150nm BN,(2) ngth test: (1) short fibre bre reinforced, interface 50 nm BN, (3)short fibre reinforced, no coating
344 D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 Table 3 Mechanical properties of the manufactured composites (mean values) Fibers Three-point bending test Biaxial strength test Bending strength (MPa) Fracture toughness (MPa m1/2) Bending strength (MPa) Continuous fiber Short fiber Continuous fiber Short fiber Continuous fiber Short fiber NextelTM 440 uncoated 61 78 1.0 1.43 61 59 NextelTM 440/40 nm BN coating 91 NA 3.3 NA NA 76 NextelTM 440/150 nm BN coating 98 88 2.3 1.45 59 63 ZenTronTM uncoated 51 61 1.1 1.45 53 38 ZenTronTM/70 nm SnO2 coating 84 63 0.9 1.01 54 64 The relative error of bending strength and biaxial strength data is 10%. The relative error of Kic data is 10% (NA, not assessed). The occasional defects are thought to be caused by an uncontrolled coating process. Nevertheless, a turbostratic hexagonal structure was demonstrated for the boron nitride phase coatings by TEM investigations, as shown in previous investigations [21]. The desized ZenTronTM glass fiber in Fig. 4 shows a relatively smooth surface as expected for an amorphous material. On the contrary, the tin dioxide CVD coating exhibits a relatively rough crystalline structure in AFM imaging. Raman spectrometry and X-ray diffraction (XRD) have shown that the tin oxide coating consists mainly of cassiterite phase [21](results not shown here). Besides the changes in fiber chemistry due to interactions with tin ions and the possible development of surface microstresses, a mechanical aspect of fiber damaging should not be ruled out, caused by the formation of relatively large cassiterite crystals. This crystal growth can lead to surface stresses, which in turn might prompt the formation of microflaws on glass fiber surfaces. 3.2. Composite characterization The thermal expansion coefficients (α) determined for different composites containing short or continuous fiber reinforcement are listed in Table 2. The α-values for the unidirectional continuous fiber reinforced composites were measured in the fiber direction. The experimental results are in broad agreement with values calculated by the rule of mixtures using the α-values for matrix and fibers given in Table 1. The results of the mechanical strength tests are summarized in Table 3. The fracture behaviour of the composites was investigated by both three-point bending and biaxial flexural strength tests. The latter test is relevant considering the potential applications of the composites as structural components, e.g. in architecture, where flat composite panels will be enclosed in a metal or plastic frame, thus possibly exposed to biaxial stresses due to thermal expansion mismatch of the materials involved [22]. Moreover, the biaxial test method is useful to obtain relevant quantitative data to compare the behaviour of continuous and short fiber reinforced composites. The test allows for comparison of the influence of the fiber orientation in continuous fiber composites and the quasi-isotropic behaviour of short fiber composites regarding both mechanical strength and work of fracture. 3.3. NextelTM 440 fiber reinforcement Because of the thermal expansion coefficient mismatch (αF > αM) in these composites, radial tensile stresses surrounding the NextelTM fibers develop. There is, however, a residual hoop compressive stress field in the matrix, which is beneficial for the strength of the composite material. As expected, unidirectional and short NextelTM 440 fiber reinforced composites showed increased strength in three-point bending tests for both 40 and 150 nm BN fiber coatings, as compared to uncoated fibers. In the case of continuous fiber reinforcement, the fracture toughness increased up to 3.3 MPa m1/2 for composites with 40 nm thick boron nitride interfaces. For the short fiber reinforced composites, no enhancement of the fracture toughness was observed. This is a consequence of the low fiber content in these composites (see Table 2) and due to the fact that the basic toughening mechanisms (mainly crack deflection and crack bridging) are different from those active in continuous fiber composites, e.g. fiber debonding and pull-out [5,6]. Moreover, it can be assumed that the edges of the short fibers are not coated, and therefore diffusion reactions may occur between fibers and matrix which should cause strong interface bonding leading to brittle fracture. In biaxial flexural strength tests, no significant differences between the strength values of the composites with uncoated and BN-coated fibers (150 nm thick) were found for both continuous and short fiber reinforcement. Only composites with short fiber reinforcement with a 40 nm thick BN Fig. 5. Load–displacement curves of selected samples (borosilicate glass/NextelTM 440 fiber) for biaxial flexural strength test: (1) short fibre reinforced, interface: 150 nm BN, (2) continuous fibre reinforced, interface: 150 nm BN, (3) short fibre reinforced, no coating
D Raab et al. /Materials Science and Engineering A 417 (2006)341-347 Fig. 6. SEM micrographs of fracture surfaces of samples fractured in biaxial flexure test: Nextel M 440 short fiber reinforced composites with uncoated fibers(left) and with 150 nm BN-coated fibers(right). interface exhibited increased fracture strength. Fig. 5 shows typ- these brittle materials caused by the complex stress field. The ical load-displacement curves for different composites obtained fracture is of statistical nature and no correlation between the in biaxial flexural strength tests Comparing load-displacement number of fracture fragments and the measured strength values curves, non-brittle fracture behaviour was found for compos- could be determined ites with boron nitride-coated fibers(40 and 150 nm BN coating thickness). In contrast to this, continuous fiber reinforced com- 3.4. ZenTronm glass fiber reinforcement posites without fiber coating break in brittle manner(not shown in Fig. 5), whereas the boron nitride coating leads to a non-brittle In these composites, because of aF <aM(see Table 1) fracture behaviour, shown in Fig. 5 radial compression stresses develop in the matrix. Regardless As expected, the fracture surface analysis of the short fiber of any possible diffusion processes at the interface leading to reinforced composites by SEM imaging did not indicate crack strong fiber-matrix bonding, the presence of radial compres- deflection at the fiber-matrix interface for composites containing sion stresses should contribute to clamp the fiber, and thus fiber uncoated fibers. The crack runs from the matrix into the fiber pull-out should be prevented in these composites. The thermal (Fig. 6, left). Indication of crack deflection and fiber debond- expansion mismatch is thus unfavourable in these composites to was observed, however, on similar composites containing promote toughening by fiber pull-out 150 nm BN-coated fibers(Fig. 6, right) The desized ZenTron glass fiber reinforced composite co. The macroscopic crack patterns were optically analysed to showed comparable strength values for continuous fiber rein- mpare crack initiation, crack deflection and bifurcations on forcement in both three-point bending tests and biaxial flexural different samples Selected sample images are shown in Fig. 7. strength tests For the short fiber reinforcement, the biaxial flex In all cases, the crack started in the central zone of the disk. This ure strength value is significantly lower than the value measured is consistent with the equibiaxial stress distribution in biaxial for the uniaxial fiber reinforced composite. In all cases, brit- flexure strength tests[19]. In the case of the short fiber reinforced tle fracture was observed. In contrast to this, the three-point samples, more than one bifurcation centre were found. In these bending strength test and biaxial strength test data for the tin composites, crack deflection occurs and the crack propagation is oxide-coated fibers were similar for the short fiber reinforced non-linear. Besides the radial crack pattern, samples containing composites, whereas the continuous fiber reinforced composites uncoated fibers showed also a circumferential crack branching. as expected, showed a higher strength in three-point bending Contrary to these findings, the sample with uncoated continu- The fracture toughness did not essentially improve regardless of ous fiber reinforcement showed straight crack propagation from fiber coating and fiber architecture(short or continuous fibers) the centre of the disk(right image in Fig. 7). In most cases, Earlier investigations have shown the ability of tin oxide inter crack propagation was orthogonal to the unidirectional fibers faces to deflect cracks in glass matrix/alumina fiber composites and bifurcation started at fiber-matrix interfaces. This fracture [15]. However, this effect is more probable in composites with pattern occurs because of the great variety of failure modes in high fiber content and low coating thickness [16] Fig. 7. Crack 150 nm BN-coated short fibers(middle)and uncoated continuous fibers(right)
D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 345 Fig. 6. SEM micrographs of fracture surfaces of samples fractured in biaxial flexure test: NextelTM 440 short fiber reinforced composites with uncoated fibers (left) and with 150 nm BN-coated fibers (right). interface exhibited increased fracture strength. Fig. 5 shows typical load–displacement curves for different composites obtained in biaxial flexural strength tests. Comparing load–displacement curves, non-brittle fracture behaviour was found for composites with boron nitride-coated fibers (40 and 150 nm BN coating thickness). In contrast to this, continuous fiber reinforced composites without fiber coating break in brittle manner (not shown in Fig. 5), whereas the boron nitride coating leads to a non-brittle fracture behaviour, shown in Fig. 5. As expected, the fracture surface analysis of the short fiber reinforced composites by SEM imaging did not indicate crack deflection at the fiber–matrix interface for composites containing uncoated fibers. The crack runs from the matrix into the fiber (Fig. 6, left). Indication of crack deflection and fiber debonding was observed, however, on similar composites containing 150 nm BN-coated fibers (Fig. 6, right). The macroscopic crack patterns were optically analysed to compare crack initiation, crack deflection and bifurcations on different samples. Selected sample images are shown in Fig. 7. In all cases, the crack started in the central zone of the disk. This is consistent with the equibiaxial stress distribution in biaxial flexure strength tests[19]. In the case of the short fiber reinforced samples, more than one bifurcation centre were found. In these composites, crack deflection occurs and the crack propagation is non-linear. Besides the radial crack pattern, samples containing uncoated fibers showed also a circumferential crack branching. Contrary to these findings, the sample with uncoated continuous fiber reinforcement showed straight crack propagation from the centre of the disk (right image in Fig. 7). In most cases, crack propagation was orthogonal to the unidirectional fibers and bifurcation started at fiber–matrix interfaces. This fracture pattern occurs because of the great variety of failure modes in these brittle materials caused by the complex stress field. The fracture is of statistical nature and no correlation between the number of fracture fragments and the measured strength values could be determined. 3.4. ZenTronTM glass fiber reinforcement In these composites, because of αF < αM (see Table 1) radial compression stresses develop in the matrix. Regardless of any possible diffusion processes at the interface leading to strong fiber–matrix bonding, the presence of radial compression stresses should contribute to clamp the fiber, and thus fiber pull-out should be prevented in these composites. The thermal expansion mismatch is thus unfavourable in these composites to promote toughening by fiber pull-out. The desized ZenTronTM glass fiber reinforced composite showed comparable strength values for continuous fiber reinforcement in both three-point bending tests and biaxial flexural strength tests. For the short fiber reinforcement, the biaxial flexure strength value is significantly lower than the value measured for the uniaxial fiber reinforced composite. In all cases, brittle fracture was observed. In contrast to this, the three-point bending strength test and biaxial strength test data for the tin oxide-coated fibers were similar for the short fiber reinforced composites, whereas the continuous fiber reinforced composites, as expected, showed a higher strength in three-point bending. The fracture toughness did not essentially improve regardless of fiber coating and fiber architecture (short or continuous fibers). Earlier investigations have shown the ability of tin oxide interfaces to deflect cracks in glass matrix/alumina fiber composites [15]. However, this effect is more probable in composites with high fiber content and low coating thickness [16]. Fig. 7. Crack patterns of borosilicate glass/NextelTM 440 fiber composites, reassembled after biaxial flexure test. Composites with: uncoated short fibers (left), 150 nm BN-coated short fibers (middle) and uncoated continuous fibers (right)
D. Raab er al./ Materials Science and Engineering A 417(2006)341-347 Fig8. SEM micrographs of fracture surfaces in biaxial flexure test: borosilicate glass/Zen Tron M continuous fiber composites containing desized fibers(left)and 70 nm tin oxide-coated fibers(night) increase of fracture toughness or flexure strength in both contint ous and short fiber reinforced composites. No crack deflection at the matrix-tin dioxide-fi ber interfaces. neither indication ofother toughening mechanisms, such as fiber debonding or pull-out were found. The decrease of fiber strength during CVD coating and the possible diffusion of Sn-ions during hot pressing could 30μm be responsible for the relatively poor mechanical performance of these composite Fig9. Laser scanning micrograph of a Vickers'microindentation on a borosil- Acknowledgements ate glass/ZenTron M continuous fiber composite with 70 nm tin oxide-coated The authors thank Mrs. J.A. Roether(Imperial College Lon- don) for revising the manuscript. The work was supported by Investigating fracture surfaces by SEM, no feature could the German Research Foundation(FG). The authors are grate- be discerned which could be linked with a mechanism acting ful to TELUX WeiBwasser(Germany) for providing the matrix to enhance the fracture toughness in the present composites glass. Thanks are also due to AGY Europe for the Zen TronM (Fig 8). Besides the clearly visible rings around fibers represent- glass fiber samples ing the tin oxide coating in Fig. 8 (right ), there are no apparent differences between the two fracture surfaces. Moreover, neither References significant crack deflection was detected at fiber-matrix inter faces, nor indication of fiber debonding or pull-out could be [1]AR. Boccaccini, R D. Rawlings, Glass Technol. 43C(2002)191-201 observed 2]K M. Prewo, J.J. Brennan, G K. Layden, Am. Ceram. Soc. Bull. 6: These sEM observations were confirmed by Vickers (1986)305-313. microindentation tests. As shown in Fig9 cracks starting from [4]W. Beier, Faserverstarkte Glaser, 44th Intemational Scientific Collo- the corners of the Vickers'impression were not deflected at matrix-tin dioxide-fiber interfaces. It was found that cracks [5].R. Boccaccini, in: N.P. Bansal(Ed ) Handbook of Ceramic Compos- propagate unimpeded through matrix, interfaces and fibers ites, Kluwer Academic Publishers, New York, 2005, p. 461(Chapter demonstrating that the composites would fail in a brittle manner, [6]IW. Donald, Key Eng Mater. 108-110(1995)123-144 with little or no crack deflection or fiber pull-out [ D. Hulsenberg, P. Fehling, H. Kem, D. Raab, T. Mache, G. Marx, K. Weise, A.R. Boccaccini, Mater. Congr. (2002) 4. Conclusions []D. Hulsenberg, P. Fehling, T. Leutbecher, D Raab, Machine Dyn. Prob- Turbostratic boron nitride CVD-coated Nextel TM 440 fibers are suitable reinforcements for borosilicate glass matrix compos- [10] H. Iba, T. Chang, Y. Kagawa, H. Minakuchi, K. Kanamaru, J. Am ites. Thermal degradation of the fiber and changing of surfac roughness due to the C VD coating process were found to be neg. [II]B. Fankanel, E. Muller, K. Weise, G. Marx, Key Eng. Mater. 206-213 (2002)11091112. igible. The composites exhibited enhanced fracture toughness [12] A.R. Boccaccini,S Atiq, G. Helsch, Comp. Sci. Technol. 63(2003) in comparison to the unreinforced matrix, especially in the case 779-783. of continuous fiber reinforcement Enhanced fracture strength [13] H. Iba, T. Naganuma, K. Matsumura, Y. Kagawa, J. Mater. Sci for continuous and short fiber reinforcement was confirmed (1999)5701-5705 in three-point bending strength tests. However, no significant [14R.J Kerans, R.S. Hay, T.A. Parthasarathy, M. Cinibulk, J. Am. Cerar Soe.85(1l)(2002)2599-2632. increase of fracture strength in both types of composites was [15]AMaheshwari, K KChawla, TA.Michalske, Mater. Sci.Eng.A observed in biaxial flexure strength tests (1989)269276
346 D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 Fig. 8. SEM micrographs of fracture surfaces in biaxial flexure test: borosilicate glass/ZenTronTM continuous fiber composites containing desized fibers (left) and 70 nm tin oxide-coated fibers (right). Fig. 9. Laser scanning micrograph of a Vickers’ microindentation on a borosilicate glass/ZenTronTM continuous fiber composite with 70 nm tin oxide-coated fibers. Investigating fracture surfaces by SEM, no feature could be discerned which could be linked with a mechanism acting to enhance the fracture toughness in the present composites (Fig. 8). Besides the clearly visible rings around fibers representing the tin oxide coating in Fig. 8 (right), there are no apparent differences between the two fracture surfaces. Moreover, neither significant crack deflection was detected at fiber–matrix interfaces, nor indication of fiber debonding or pull-out could be observed. These SEM observations were confirmed by Vickers’ microindentation tests. As shown in Fig. 9 cracks starting from the corners of the Vickers’ impression were not deflected at matrix-tin dioxide–fiber interfaces. It was found that cracks propagate unimpeded through matrix, interfaces and fibers, demonstrating that the composites would fail in a brittle manner, with little or no crack deflection or fiber pull-out. 4. Conclusions Turbostratic boron nitride CVD-coated NextelTM 440 fibers are suitable reinforcements for borosilicate glass matrix composites. Thermal degradation of the fiber and changing of surface roughness due to the CVD coating process were found to be negligible. The composites exhibited enhanced fracture toughness in comparison to the unreinforced matrix, especially in the case of continuous fiber reinforcement. Enhanced fracture strength for continuous and short fiber reinforcement was confirmed in three-point bending strength tests. However, no significant increase of fracture strength in both types of composites was observed in biaxial flexure strength tests. Tin oxide coating on ZenTronTM glass fibers does not bring an increase of fracture toughness or flexure strength in both continuous and short fiber reinforced composites. No crack deflection at the matrix-tin dioxide-fiber interfaces, neither indication of other toughening mechanisms, such as fiber debonding or pull-out were found. The decrease of fiber strength during CVD coating and the possible diffusion of Sn-ions during hot pressing could be responsible for the relatively poor mechanical performance of these composites. Acknowledgements The authors thank Mrs. J.A. Roether (Imperial College London) for revising the manuscript. The work was supported by the German Research Foundation (DFG). The authors are grateful to TELUX Weißwasser (Germany) for providing the matrix glass. Thanks are also due to AGY Europe for the ZenTronTM glass fiber samples. References [1] A.R. Boccaccini, R.D. Rawlings, Glass Technol. 43C (2002) 191–201. [2] K.M. Prewo, J.J. Brennan, G.K. Layden, Am. Ceram. Soc. Bull. 65 (1986) 305–313. [3] T. Leutbecher, D. Hulsenberg, Adv. Eng. Mater. 2 (2000) 93. ¨ [4] W. Beier, Faserverstarkte Gl ¨ aser, 44th International Scientific Collo- ¨ quium, Technische Universitat Ilmenau, 1999. ¨ [5] A.R. Boccaccini, in: N.P. Bansal (Ed.), Handbook of Ceramic Composites, Kluwer Academic Publishers, New York, 2005, p. 461 (Chapter 19). [6] I.W. Donald, Key Eng. Mater. 108–110 (1995) 123–144. [7] D. Hulsenberg, P. Fehling, H. Kern, D. Raab, T. Mache, G. Marx, K. ¨ Weise, A.R. Boccaccini, Mater. Congr. (2002). [8] D. Hulsenberg, P. Fehling, T. Leutbecher, D. Raab, Machine Dyn. Prob- ¨ lems 28 (2004) 131–137. [9] A.F. Dericioglu, S. Zhu, Y. Kagawa, Cer. Eng. Sci. Proc. 23 (2002) 485–492. [10] H. Iba, T. Chang, Y. Kagawa, H. Minakuchi, K. Kanamaru, J. Am. Ceram. Soc. 79 (1999) 881–884. [11] B. Fankanel, E. M ¨ uller, K. Weise, G. Marx, Key Eng. Mater. 206–213 ¨ (2002) 1109–1112. [12] A.R. Boccaccini, S. Atiq, G. Helsch, Comp. Sci. Technol. 63 (2003) 779–783. [13] H. Iba, T. Naganuma, K. Matsumura, Y. Kagawa, J. Mater. Sci. 34 (1999) 5701–5705. [14] R.J. Kerans, R.S. Hay, T.A. Parthasarathy, M. Cinibulk, J. Am. Ceram. Soc. 85 (11) (2002) 2599–2632. [15] A. Maheshwari, K.K. Chawla, T.A. Michalske, Mater. Sci. Eng. A 107 (1989) 269–276
D. Raab et al. Materials Science and Engineering A 417(2006)341-347 [16]SN. Patankar, C. Suryanarayana, F.H. Froes, Scripta Metall. Mater. [20]K. Heckel, Einfuhrung in die technische Anwendung der Bruch- (1991)1787-1792. mechanik. 3. Aufl. Hanser Verlag. 1991 [17 P. Fehling, D. Hulsenberg, T. Mache, K. Weise, G. Marx, S. Stockel, [21K. Weise, S Stockel, M. Winkler, K. Nestler, G. Marx, in: B. wielage, Annual Meeting Exposition of the Americar nic Society, vo G. Leonhard(Eds ) Verbundwerkstoffe und Werkstofiverbunde, Wiley- 104. St Louis. MO. USA. 2002 VCH, Weinheim, Germany, 2001, p. 359 D. Dietrich, S Stockel, G. Marx, Fresenius J. Anal. Chem. 361(1998) [22]SJ. Bennison, A. Jagota, C.A. Smith, J. Am. Ceram. Soc. 82(1999) 653-655 176l-1770. A. Atkinson, A Selcuk, Acta Mater. 47(1999)867-874
D. Raab et al. / Materials Science and Engineering A 417 (2006) 341–347 347 [16] S.N. Patankar, C. Suryanarayana, F.H. Froes, Scripta Metall. Mater. 25 (1991) 1787–1792. [17] P. Fehling, D. Hulsenberg, T. Mache, K. Weise, G. Marx, S. St ¨ ockel, ¨ Annual Meeting & Exposition of the American Ceramic Society, vol. 104, St. Louis, MO, USA, 2002. [18] D. Dietrich, S. Stockel, G. Marx, Fresenius J. Anal. Chem. 361 (1998) ¨ 653–655. [19] A. Atkinson, A. Selcuk, Acta Mater. 47 (1999) 867–874. [20] K. Heckel, Einfuhrung in die technische Anwendung der Bruch- ¨ mechanik, 3, Aufl., Hanser Verlag, 1991. [21] K. Weise, S. Stockel, M. Winkler, K. Nestler, G. Marx, in: B. Wielage, ¨ G. Leonhard (Eds.), Verbundwerkstoffe und Werkstoffverbunde, WileyVCH, Weinheim, Germany, 2001, p. 359. [22] S.J. Bennison, A. Jagota, C.A. Smith, J. Am. Ceram. Soc. 82 (1999) 1761–1770
