MATERIALS ELSEVIER Materials Characterization 58(2007)922-927 Non-destructive testing of satellite nozzles made of carbon fibre ceramic matrix composite, C/SiC J Rebelo Kornmeier a,.M. Hofmann .S. Schmidt Technische Universitat Miinchen, ZWE, FRM-lL, Forschungsneutronenquelle Heinz Maier-Leibnitz, Lichtenbergstr: I D-85747 Garching, germany EADS-Space-Transportation, Willy-Messerschmi-StrD-85521 Minchen, Germany Received 25 January 2006: received in revised form 14 September 2006: accepted 18 September 2006 bstract Carbon fibre ceramic matrix composite materials, C/SiC, are excellent candidates as lightweight structural material performance hot structures such as in aerospace applications. Satellite nozzles are manufactured from C/SiC, using, for instance, the Liquid Polymer Infiltration(LPD) process. In this article the applicability of different non-destructive analysis methods for the characterisation of C/Sic components will be discussed. By using synchrotron and neutron tomography it is possible to characterise the C/SiC material in each desired location or orientation. Synchrotron radiation using tomography on small samples with a resolution of 1. 4 um, i.e. the fibre scale, was used to characterise three dimensionally fibre orientation and integrity, matrix homogeneity and dimensions and distributions of micro pores. Neutron radiation tomography with a resolution of about 300 um was used to analyse the over-all C/SiC satellite nozzle component with respect to the fibre content. The special solder connection of a C/SiC satellite nozzle to a metallic ring was also successfully analysed by neutron tomography. In addition, the residual stress state of a temperature tested satellite nozzle was analysed non-destructively in depth by neutron diffraction. The results revealed almost zero stress for the principal directions radial, axial and tangential, which can be considered to be the principal directions. c2006 Elsevier Inc. All rights reserved. Keywords: C/SiC; Neutron; Synchrotron; Tomography 1. Introduction aggressive environments. SiC has an excellent high L.I. Why use C/SiC fibre ceramic matrix composite density, good oxidation resistance and high hardness material in the aerospace industry? however, it is also notch-sensitive and low in toughness. Composites consisting of carbon fibres embedded in a Silicon carbide(Sic) offers great potential for SiC matrix combine the elevated mechanical properties tructural applications in the aerospace industry which of carbon fibres with the high oxidation resistance of the requires structural materials for high temperatures and SiC matrix. In addition to high strength due to load transfer from the SiC matrix to the fibres and higl fracture toughness energy can be absorbed by fibre pull il address: joana. kommeier(@fnma tum. de out from the SiC matrix, causing crack deflection or 044-5803/S-see front matter o 2006 Elsevier Inc. All rights reserved. doi:10.06 matcha.200609.010
Non-destructive testing of satellite nozzles made of carbon fibre ceramic matrix composite, C/SiC J. Rebelo Kornmeier a,⁎, M. Hofmann a , S. Schmidt b a Technische Universität München, ZWE, FRM-II, Forschungsneutronenquelle Heinz Maier-Leibnitz, Lichtenbergstr. 1, D-85747 Garching, Germany b EADS-Space-Transportation, Willy-Messerschmitt-Str.D-85521 München, Germany Received 25 January 2006; received in revised form 14 September 2006; accepted 18 September 2006 Abstract Carbon fibre ceramic matrix composite materials, C/SiC, are excellent candidates as lightweight structural materials for high performance hot structures such as in aerospace applications. Satellite nozzles are manufactured from C/SiC, using, for instance, the Liquid Polymer Infiltration (LPI) process. In this article the applicability of different non-destructive analysis methods for the characterisation of C/SiC components will be discussed. By using synchrotron and neutron tomography it is possible to characterise the C/SiC material in each desired location or orientation. Synchrotron radiation using tomography on small samples with a resolution of 1.4 μm, i.e. the fibre scale, was used to characterise three dimensionally fibre orientation and integrity, matrix homogeneity and dimensions and distributions of micro pores. Neutron radiation tomography with a resolution of about 300 μm was used to analyse the over-all C/SiC satellite nozzle component with respect to the fibre content. The special solder connection of a C/SiC satellite nozzle to a metallic ring was also successfully analysed by neutron tomography. In addition, the residual stress state of a temperature tested satellite nozzle was analysed non-destructively in depth by neutron diffraction. The results revealed almost zero stress for the principal directions, radial, axial and tangential, which can be considered to be the principal directions. © 2006 Elsevier Inc. All rights reserved. Keywords: C/SiC; Neutron; Synchrotron; Tomography 1. Introduction 1.1. Why use C/SiC fibre ceramic matrix composite material in the aerospace industry? Silicon carbide (SiC) offers great potential for structural applications in the aerospace industry which requires structural materials for high temperatures and aggressive environments. SiC has an excellent high temperature strength and elasticity modulus, low density, good oxidation resistance and high hardness, however, it is also notch-sensitive and low in toughness. Composites consisting of carbon fibres embedded in a SiC matrix combine the elevated mechanical properties of carbon fibres with the high oxidation resistance of the SiC matrix. In addition to high strength due to load transfer from the SiC matrix to the fibres and high fracture toughness energy can be absorbed by fibre pullout from the SiC matrix, causing crack deflection or blunting. Materials Characterization 58 (2007) 922–927 ⁎ Corresponding author. E-mail address: joana.kornmeier@frm2.tum.de (J. Rebelo Kornmeier). 1044-5803/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.matchar.2006.09.010
J Rebelo Kornmeier et al. Materials Characterization 58(2007)922-92 1. 2. Description of the C/SiC nozzle fabrication great importance to determine the residual stress state of the C/SiC composite nozzles Several processing methods have been developed for fabricating continuous fibre ceramic matrix composites. 1.4. Connection between C/SiC nozzle and metallic The C/SiC composites are usually manufactured by an component infiltration processes. In the following the Liquid Polymer Infiltration(LPD) process, which was used for In order to connect the C/SiC composite nozzle to a the samples analysed in this study, will be briefly metallic ring a special brazing method was developed 3]. Prior to sold Here, the C/SiC composite is made via the so called composite is perforated with a solid state small impulse polymer route. A carbon fibre bundle coated with a laser NdYAG, see Fig. 1(a). The perforation method was polymer is impregnated with a powder-filled polymer, optimised with respect to the optimal shape, dimensions the precursor, and laminated to form prepregs. Subse- and distributions of the perforations and the surface area quently, the wound fibre cloth structure is laminated, to be perforated. The cavities were then filled with compacted in an autoclave and cross linked. Afterward solder in a high vacuum, Fig. 1(b)3 this green composite is pyrolysed without pressure and without moulding tools at temperatures around 1300- 2. Experimental 1900 K in an inert gas atmosphere. Such processes are relatively flexible since the composition of the precursor 2.1. Tomography an be tailored. A shrinking of the matrix occurs during the pyrolysis step owing to the generation of gaseous Tomography in general is a method which provides ecies. As a consequence, several pyrolysis sequences cross sectional images of an object from transmission and re-impregnations have to be applied in order to data, obtained by irradiating it with specific radiation achieve a low enough residual porosity [1] from many different directions. From these projections a tomographic image is then mathematically recon- 1.3. Residual stress of the fibre ceramic matrix structed. Here the projection at a given angle represents composite material (a) Generally, temperature gradients occurring within the omponent during the fabrication processes can lead to residual macros-stresses on the scale of the component. Moreover, when the material has more than one phase phase-specific residual stresses arise during cooling as a consequence mainly of the difference of the thermal expansion coefficient of the phases, i.e. between fibre and the matrix in the case of composite materials. The phase-specific residual stresses of one phase is the sum of residual macro stresses plus load stresses plus micro residual stresses of this phase [2]. The material resistance and consequently the lifetime of the compos- (b) ite component under service conditions depends on the residual stress state. The fracture of ceramic matrix composites is usually governed by matrix cracking followed by interactions of the newly-formed cracks with fibres and interfaces such as fibre pull-out and debonding. These fracture mechanisms are activated by tensile stresses acting parallel to the fibre axis. Furthermore, it is obvious that the stresses acting norma to the fibre axis have an important influence on the load transfer from the fibre to the matrix as well as on the debonding process. Therefore, in order to optimise the aterial behaviour for certain external loads. it is of Fig. 1.(a) Laser perforation; (b) cavities filled with solder
1.2. Description of the C/SiC nozzle fabrication Several processing methods have been developed for fabricating continuous fibre ceramic matrix composites. The C/SiC composites are usually manufactured by an infiltration processes. In the following the Liquid Polymer Infiltration (LPI) process, which was used for the samples analysed in this study, will be briefly described [1]. Here, the C/SiC composite is made via the so called polymer route. A carbon fibre bundle coated with a polymer is impregnated with a powder-filled polymer, the precursor, and laminated to form prepregs. Subsequently, the wound fibre cloth structure is laminated, compacted in an autoclave and cross linked. Afterwards this green composite is pyrolysed without pressure and without moulding tools at temperatures around 1300– 1900 K in an inert gas atmosphere. Such processes are relatively flexible since the composition of the precursor can be tailored. A shrinking of the matrix occurs during the pyrolysis step owing to the generation of gaseous species. As a consequence, several pyrolysis sequences and re-impregnations have to be applied in order to achieve a low enough residual porosity [1]. 1.3. Residual stress of the fibre ceramic matrix composite material Generally, temperature gradients occurring within the component during the fabrication processes can lead to residual macros-stresses on the scale of the component. Moreover, when the material has more than one phase, phase-specific residual stresses arise during cooling as a consequence mainly of the difference of the thermal expansion coefficient of the phases, i.e. between fibre and the matrix in the case of composite materials. The phase-specific residual stresses of one phase is the sum of residual macro stresses plus load stresses plus micro residual stresses of this phase [2]. The material resistance and consequently the lifetime of the composite component under service conditions depends on the residual stress state. The fracture of ceramic matrix composites is usually governed by matrix cracking followed by interactions of the newly-formed cracks with fibres and interfaces such as fibre pull-out and debonding. These fracture mechanisms are activated by tensile stresses acting parallel to the fibre axis. Furthermore, it is obvious that the stresses acting normal to the fibre axis have an important influence on the load transfer from the fibre to the matrix as well as on the debonding process. Therefore, in order to optimise the material behaviour for certain external loads, it is of great importance to determine the residual stress state of the C/SiC composite nozzles. 1.4. Connection between C/SiC nozzle and metallic component In order to connect the C/SiC composite nozzle to a metallic ring a special brazing method was developed [3]. Prior to soldering, the surface of the C/SiC composite is perforated with a solid state small impulse laser NdYAG, see Fig. 1(a). The perforation method was optimised with respect to the optimal shape, dimensions and distributions of the perforations and the surface area to be perforated. The cavities were then filled with solder in a high vacuum, Fig. 1(b) [3]. 2. Experimental 2.1. Tomography Tomography in general is a method which provides cross sectional images of an object from transmission data, obtained by irradiating it with specific radiation from many different directions. From these projections a tomographic image is then mathematically reconstructed. Here the projection at a given angle represents Fig. 1. (a) Laser perforation; (b) cavities filled with solder. J. Rebelo Kornmeier et al. / Materials Characterization 58 (2007) 922–927 923
J. Rebelo Kornmeier et al. Materials Characterization 58(2007)922-927 84 mm 94 mm from top 20 mm from top (a) Fig. 2. Fibre ceramic matrix composite nozzle produced by LPl. The dark grey colour represents the material which has the highest absorption the integral of the image in the direction specified by The main difference between tomography with X- that angle [4, 5]. As a result cross sectional views of the rays and neutrons may be explained as follows: X-rays object can be non-destructively visualised in any desired interact with the atomic shell, i.e. they are scattered or ocation and orientation absorbed by electrons. The more electrons an element (b) Niob ring (a) C/SiC Fig 3. Connection between the C/SiC nozzle produced by LPI and the metal ring. a) Cross section of the C/SiC nozzle connected with metal rin b)and c) Three-dimensional representations of the isolated solder. d)C/SiC nozzle and Niob ring components
the integral of the image in the direction specified by that angle [4,5]. As a result cross sectional views of the object can be non-destructively visualised in any desired location and orientation. The main difference between tomography with Xrays and neutrons may be explained as follows: X-rays interact with the atomic shell, i.e. they are scattered or absorbed by electrons. The more electrons an element Fig. 2. Fibre ceramic matrix composite nozzle produced by LPI. The dark grey colour represents the material which has the highest absorption. Fig. 3. Connection between the C/SiC nozzle produced by LPI and the metal ring. a) Cross section of the C/SiC nozzle connected with metal ring. b) and c) Three-dimensional representations of the isolated solder. d) C/SiC nozzle and Niob ring components. 924 J. Rebelo Kornmeier et al. / Materials Characterization 58 (2007) 922–927
J Rebelo Kornmeier et al. Materials Characterization 58(2007)922-92 neutrons that pass through the sample are recorded by a nitrogen cooled CCD-camera system. The basic princi- ple of the detector is the combination of a CCD-camera SiC matrix with a neutron-sensitive scintillator screen ( Li6 or gd as neutron absorber). The light from the screen is reflected to the camera by a mirror and focused on the CCD-chip by a special lens [6] The nozzle was incrementally rotated between 0- 180C using 600 steps whilst irradiating the nozzle for Fibre 30 s for each angle. The resolution obtained was about 300 In Fig. 2 intensity distributions are shown for cross sections parallel (a)and perpendicular, (b) and(c), to the nozzle axis In Fig. 2 the dark grey colour represents the Fig. 4. Three-dimensional view of an exemplary sample of ceramic material which has the highest absorption, i. e. where the composite(C/SiC) fibre concentration is higher. It can be clearly seen where the nozzle was reinforced. the fibre concentration has, the more it attenuates X-rays. Neutrons, on the being higher in the interior side of the throat region. other hand. interact with the atomic nuclei but show no obvious regularity across the periodic table of elements. 2. 3. Neutron tomography of the solder connection Interaction strongly depends on the inner structure of the atomic nuclei, meaning that even isotopes of the same Neutron tomography was also applied to the nozzle element may often provide very different levels of shown in Fig. 3d)in order to verify the solder distribution in contrast in the projection. The high degree of neutron the joint section between the nozzle and the metal scattering caused by hydrogen and the penetration component using thermal neutrons. The different absorp- capacity of neutrons for most metals are of particular tions coefficients of the materials enable their identification industrial significance [5]. and, hence, a representation of their distribution over the components volume. The results are presented in Fig. 3 2. 2. Neutron tomography of the nozzle showing in a)a cross section of the C/Sic nozzle connected with the metal ring, the dark grey colour representing the The neutron tomographies were carried out at the solder material. Such cross sectional views can be obtained instrument NEUTRA at the Paul Scherrer Institut(PSD) at any axial position and angle. Three-dimensional in Switzerland using thermal neutrons with an energy representations of isolated solder are also possible, see range between 2 meV and 100 meV. The thermal Fig 3b)and c). It can be seen that almost all cavities were Fig. 5. (a)Temperature tested nozzle;(b)residual stress measurement directions at throat region.(b) Residual stress measurement apparatus at HMI
has, the more it attenuates X-rays. Neutrons, on the other hand, interact with the atomic nuclei, but show no obvious regularity across the periodic table of elements. Interaction strongly depends on the inner structure of the atomic nuclei, meaning that even isotopes of the same element may often provide very different levels of contrast in the projection. The high degree of neutron scattering caused by hydrogen and the penetration capacity of neutrons for most metals are of particular industrial significance [5]. 2.2. Neutron tomography of the nozzle The neutron tomographies were carried out at the instrument NEUTRA at the Paul Scherrer Institut (PSI) in Switzerland using thermal neutrons with an energy range between 2 meV and 100 meV. The thermal neutrons that pass through the sample are recorded by a nitrogen cooled CCD-camera system. The basic principle of the detector is the combination of a CCD-camera with a neutron-sensitive scintillator screen (Li6 or Gd as neutron absorber). The light from the screen is reflected to the camera by a mirror and focused on the CCD-chip by a special lens [6]. The nozzle was incrementally rotated between 0– 180 °C using 600 steps whilst irradiating the nozzle for 30 s for each angle. The resolution obtained was about 300 μm. In Fig. 2 intensity distributions are shown for cross sections parallel (a) and perpendicular, (b) and (c), to the nozzle axis. In Fig. 2 the dark grey colour represents the material which has the highest absorption, i.e. where the fibre concentration is higher. It can be clearly seen where the nozzle was reinforced, the fibre concentration being higher in the interior side of the throat region. 2.3. Neutron tomography of the solder connection Neutron tomography was also applied to the nozzle shown in Fig. 3d) in order to verify the solder distribution in the joint section between the nozzle and the metal component using thermal neutrons. The different absorptions coefficients of the materials enable their identification and, hence, a representation of their distribution over the component's volume. The results are presented in Fig. 3 showing in a) a cross section of the C/SiC nozzle connected with the metal ring, the dark grey colour representing the solder material. Such cross sectional views can be obtained at any axial position and angle. Three-dimensional representations of isolated solder are also possible, see Fig. 3b) and c). It can be seen that almost all cavities were Fig. 5. (a) Temperature tested nozzle; (b) residual stress measurement directions at throat region. (b) Residual stress measurement apparatus at HMI, Berlin. Fig. 4. Three-dimensional view of an exemplary sample of ceramic composite (C/SiC). J. Rebelo Kornmeier et al. / Materials Characterization 58 (2007) 922–927 925
J. Rebelo Kornmeier et al. Materials Characterization 58(2007)922-927 filled with solder, thus providing the desired homogeneity figure the fibre orientation and integrity, matrix homoge of the connection. As an advantage with respect to neity, dimensions and distributions of micro pores can measurements with X-rays, the high density of the Niob clearly be seen. Using image analysis it is also possible to flange is not a problem for neutron radiation obtain quantitative characterisation of the C/SiC material 2. 4. Synchrotron tomography of a nozzle sample 2.5. Neutron residual stress Characterisation of the composite material on the fibre The non-destructive analysis of phase-specific residual scale is not possible by neutron tomography as the fibre stresses is only possible by means of diffraction methods diameter is only 6 to 7 um. However, synchrotron While conventional X-ray diffraction stress analysis only tomography is able to reveal three dimensionally the yields information from a small surface layer, between 5- orientation of carbon fibres and their integrity [7]. Such a 100 um, the large penetration depth of neutrons offers the characterisation tool can identify if the silicon has reacted que possibility for non-destructive residual stre with the fibre. In the Sic matrix unreacted silicon can be analysis within bulk samples and components. detected and the dimensions and distributions of pores in Residual stress measurements in the sic matrix of the any desired location or orientation can be quantified. C/SiC nozzles were carried out at the neutron diffraction Respective exemplary measurements were carried out at facilities of Hahn Meitner Institut(HMi) Ber the X-ray microtomographic device at the Materials thermal neutrons. The measurements were performed at Science Beamline MS of the Swiss Light Source(SLS) the throat region of a temperature tested nozzle in the three using an energy of 15 ke V and a pixel resolution of 1. 4 um principal directions, axial, tangential and radial, as ( the data being binned). Using a 2048 x 2048 pixel CCD- indicated in Fig. 5(b). During the prior temperature test, camera [7] this creates a 1. 4 mm field of view. Therefore the nozzle had been subjected to a temperature of 1900C the high resolution requires samples smaller than 1. 4 mm for two and an half hours, see Fig. 5(a). The residual stress to be measured. In our case samples of I mm cross measurements were carried out with the centre of the section were examined, see Fig. 4. Five hundred two- gauge volume at 3 different points in depth, the first one dimensional projections of the sample in equidistant being located 2 mm underneath the surface. The othertwo ngular intervals from 0o to 180 were acquired. The points were chosen in increments of2 mm, at 4 and 6 mm exposure time was 5 s for each projection resulting in a depth, respectively. The gauge volume element was total scan time of approximately 40 min. Flat and dark defined with a size of 4 x 4x4 mm by slits in the primary field corrected data have been reconstructed with a and reflected beam. In order to maintain an identical gauge standard filtered-backprojection algorithm [5]. The result- volume at different sample orientations, the diffraction ng three-dimensional image is shown in Fig 4. In this angle 20 had to be near 90o. Therefore, for a wavelength Point Imn : Fig. 6. Residual stress values for different positions in depth at the throat of the temperature tested nozzle
filled with solder, thus providing the desired homogeneity of the connection. As an advantage with respect to measurements with X-rays, the high density of the Niob flange is not a problem for neutron radiation. 2.4. Synchrotron tomography of a nozzle sample Characterisation of the composite material on the fibre scale is not possible by neutron tomography as the fibre diameter is only 6 to 7 μm. However, synchrotron tomography is able to reveal three dimensionally the orientation of carbon fibres and their integrity [7]. Such a characterisation tool can identify if the silicon has reacted with the fibre. In the SiC matrix unreacted silicon can be detected and the dimensions and distributions of pores in any desired location or orientation can be quantified. Respective exemplary measurements were carried out at the X-ray microtomographic device at the Materials Science Beamline MS of the Swiss Light Source (SLS) using an energy of 15 keVand a pixel resolution of 1.4 μm (the data being binned). Using a 2048× 2048 pixel CCDcamera [7] this creates a 1.4 mm field of view. Therefore the high resolution requires samples smaller than 1.4 mm to be measured. In our case samples of 1 mm2 cross section were examined, see Fig. 4. Five hundred twodimensional projections of the sample in equidistant angular intervals from 0° to 180° were acquired. The exposure time was 5 s for each projection resulting in a total scan time of approximately 40 min. Flat and dark field corrected data have been reconstructed with a standard filtered-backprojection algorithm [5]. The resulting three-dimensional image is shown in Fig. 4. In this figure the fibre orientation and integrity, matrix homogeneity, dimensions and distributions of micro pores can clearly be seen. Using image analysis it is also possible to obtain quantitative characterisation of the C/SiC material. 2.5. Neutron residual stress The non-destructive analysis of phase-specific residual stresses is only possible by means of diffraction methods. While conventional X-ray diffraction stress analysis only yields information from a small surface layer, between 5– 100 μm, the large penetration depth of neutrons offers the unique possibility for non-destructive residual stress analysis within bulk samples and components. Residual stress measurements in the SiC matrix of the C/SiC nozzles were carried out at the neutron diffraction facilities of Hahn Meitner Institut (HMI) Berlin using thermal neutrons. The measurements were performed at the throat region of a temperature tested nozzle in the three principal directions, axial, tangential and radial, as indicated in Fig. 5(b). During the prior temperature test, the nozzle had been subjected to a temperature of 1900 °C for two and an half hours, see Fig. 5(a). The residual stress measurements were carried out with the centre of the gauge volume at 3 different points in depth, the first one being located 2 mm underneath the surface. The other two points were chosen in increments of 2 mm, at 4 and 6 mm depth, respectively. The gauge volume element was defined with a size of 4×4×4 mm3 by slits in the primary and reflected beam. In order to maintain an identical gauge volume at different sample orientations, the diffraction angle 2θ had to be near 90°. Therefore, for a wavelength Fig. 6. Residual stress values for different positions in depth at the throat of the temperature tested nozzle. 926 J. Rebelo Kornmeier et al. / Materials Characterization 58 (2007) 922–927