CARBON PERGAMON Carbon40(2002)55l-555 Texture of PAN-and pitch-based carbon fibers Oskar Paris, Dieter Loidl.Herwig Peterlik Erich Schmid Institute of Materials Science, Austrian Academy of Sciences, and Metal Physics Institute, University of Leoben Jahnstrasse 12.4-8700 Leoben. Austria Institute of Materials Physics, University of vienna, Boltmanngasse 5, A-1090 vienna, austria Received 25 November 2000; accepted 26 May 2001 Abstract o The new technique of scanning microbeam X-ray diffraction is used to get information about the axial as well as the SS-sectional crystallographic texture PAN- and mesophase-pitch-based carbon fibers. The resulting preferred orientation in the axial direction is higher than the values obtained from conventional X-ray diffraction measurements on fiber bundles. a change in azimuthal width of the 002 reflection was observed across some of the fibers. which can be attributed to a radial folded cross-sectional texture for pitch-based fibers, and to a different preferred orientation of skin and core layers for PAN-based fibers. 2002 Elsevier Science Ltd. All rights reserved Keyords: A. Carbon fibers; C. X-ray diffraction; D. Textures 1. Introduction tion distribution of the crystallites. Therefore, a more detailed analysis of the overall fiber texture has to be Carbon fibers combine high tensile strength and high carried out on single fibers, e.g. by transmission electron tensile modulus with low weight. They are the ideal microscopy (TEM)[1, 4-7, 12]. The cross-sectional texture reinforcing material for light weight structures, e.g. in of pitch-based fibers was also vely investigated by aerospace applications. Carbon fibers consist of stacked (polarized light) optical microscopy (OM) on polished hexagonal carbon layers, forming small coherent units fiber cross sections [9, 13, and by scanning electron (crystallites) of only a few nanometres size in the stacking microscopy (SEM) on cross sections of fractured fibers direction [1]. The stacking direction of the layers is [7,9, 13-17. Many different cross-sectional structures such preferentially perpendicular to the fiber axis(axial pre- as radial, radial-folded, onion-skin and flat-layer have been ferred orientation), and determines primarily the modulus identified for pitch-based fibers by SEM(see Ref [15] for of the fibers [2]. The fiber strength is a function not only of an overview). However, sEM provides structure infor- flaws, but also of the axial as well as the cross-sectional mation on a much larger scale than the typical size of the orientation distribution of the stacked carbon layers [1, 3- crystallites, and the topological information gained from 9 fracture surfaces is not necessarily directly related to the The axial preferred orientation is usually determined on crystallographic orientation distribution of the crystallites fiber bundles by X-ray diffraction(XRD)in'fiber-geome To determine the crystallographic texture, diffraction tech- try, i.e. the X-ray beam is perpendicular to the long axis niques have to be used. To overcome time consuming fiber of the fibers [2, 10, 11. By investigating fiber bundles, sectioning for electron diffraction in the TEM, it is however, inevitable tilts of the fibers themselves in the proposed to scan single fibers across a high brilliance bundle cannot be separated from the tilt of the layers X-ray microbeam. a beam diameter much smaller than the within the fibers. Moreover, XRD on fiber bundles does diameter of the fibers can be provided by third generation not provide information about the cross-sectional orienta- synchrotron radiation sources [18]. In recent years it was shown for several types of single fibers that one can indeed Corresponding author. Tel: +43-3842-455-1148 fax: +43. get local cross-sectional texture information by using 3842-455-1116. canning microbeam XRD [18-22) and quantitative re- E-mail address: paris @unileoben ac at(O. Paris) sults could be obtained also for carbon fibers [23, 24. The 0008-6223/02/S-see front matter 2002 Elsevier Science Ltd. All rights reserved PII:S0008-6223(01)00139-7
Carbon 40 (2002) 551–555 Texture of PAN- and pitch-based carbon fibers a, b b Oskar Paris , Dieter Loidl , Herwig Peterlik * a Erich Schmid Institute of Materials Science, Austrian Academy of Sciences, and Metal Physics Institute, University of Leoben, Jahnstrasse 12, A-8700 Leoben, Austria b Institute of Materials Physics, University of Vienna, Boltzmanngasse 5, A-1090 Vienna, Austria Received 25 November 2000; accepted 26 May 2001 Abstract The new technique of scanning microbeam X-ray diffraction is used to get information about the axial as well as the cross-sectional crystallographic texture of single PAN- and mesophase-pitch-based carbon fibers. The resulting preferred orientation in the axial direction is considerably higher than the values obtained from conventional X-ray diffraction measurements on fiber bundles. A change in azimuthal width of the 002 reflection was observed across some of the fibers, which can be attributed to a radial folded cross-sectional texture for pitch-based fibers, and to a different preferred orientation of skin and core layers for PAN-based fibers. 2002 Elsevier Science Ltd. All rights reserved. Keywords: A. Carbon fibers; C. X-ray diffraction; D. Textures 1. Introduction tion distribution of the crystallites. Therefore, a more detailed analysis of the overall fiber texture has to be Carbon fibers combine high tensile strength and high carried out on single fibers, e.g. by transmission electron tensile modulus with low weight. They are the ideal microscopy (TEM) [1,4–7,12]. The cross-sectional texture reinforcing material for light weight structures, e.g. in of pitch-based fibers was also extensively investigated by aerospace applications. Carbon fibers consist of stacked (polarized light) optical microscopy (OM) on polished hexagonal carbon layers, forming small coherent units fiber cross sections [9,13], and by scanning electron (crystallites) of only a few nanometres size in the stacking microscopy (SEM) on cross sections of fractured fibers direction [1]. The stacking direction of the layers is [7,9,13–17]. Many different cross-sectional structures such preferentially perpendicular to the fiber axis (axial pre- as radial, radial-folded, onion-skin and flat-layer have been ferred orientation), and determines primarily the modulus identified for pitch-based fibers by SEM (see Ref. [15] for of the fibers [2]. The fiber strength is a function not only of an overview). However, SEM provides structure infor- flaws, but also of the axial as well as the cross-sectional mation on a much larger scale than the typical size of the orientation distribution of the stacked carbon layers [1,3– crystallites, and the topological information gained from 9]. fracture surfaces is not necessarily directly related to the The axial preferred orientation is usually determined on crystallographic orientation distribution of the crystallites. fiber bundles by X-ray diffraction (XRD) in ‘fiber-geome- To determine the crystallographic texture, diffraction techtry’, i.e. the X-ray beam is perpendicular to the long axis niques have to be used. To overcome time consuming fiber of the fibers [2,10,11]. By investigating fiber bundles, sectioning for electron diffraction in the TEM, it is however, inevitable tilts of the fibers themselves in the proposed to scan single fibers across a high brilliance bundle cannot be separated from the tilt of the layers X-ray microbeam. A beam diameter much smaller than the within the fibers. Moreover, XRD on fiber bundles does diameter of the fibers can be provided by third generation not provide information about the cross-sectional orienta- synchrotron radiation sources [18]. In recent years it was shown for several types of single fibers that one can indeed *Corresponding author. Tel.: 143-3842-455-1148; fax: 143- get local cross-sectional texture information by using 3842-455-1116. scanning microbeam XRD [18–22], and quantitative reE-mail address: paris@unileoben.ac.at (O. Paris). sults could be obtained also for carbon fibers [23,24]. The 0008-6223/02/$ – see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S0008-6223(01)00139-7
O. Paris et al./Carbon 40(2002)551-555 Table I Some parameters for the fibers, together with the results of the present investigatio Fiber name and CRE FWHM FWHM (g/cm) (° S E35(DuPont ±27 21.5 FT500(Tonen Pitch 4.83 HTA7-AR (Tenax) 3.50 2673 35.4 HTA7-18(Tenax) HTA7-21(Tenax) 3.50 1.78 10000 21.3 HTA7-24(Tenax) 3.35 1.91 The four PAN-based fibers differ by their final hTT (as-received, 1800, 2100 and 2400"C as indicated by the name). The mean fiber rad as determined by OM and SEM on the very same fibers as used for the diffraction experiments. The modulus was obtained from tensi tests and the density data were supplied by the data sheets from the companies. crr is the amount of crystallites building up the radial folded structure and o, their average tilt angle with respect to the main radial orientation [24]. The full-width at half maximum(FWHM) of the azimuthal intensity distribution of the 002 reflections was determined for fiber bundles(FB)and single fibers(SF, mean value weighted by fiber thickness present work shows that simultaneously the and co-workers [25, 26, and 2D diffraction patterns from cross-sectional texture of carbon fibers are obtained b single carbon fibers can be found in our earlier work the interpretation of the results for PAN-based and pitch- [23, 24]. In a fiber diffraction pattern, the 002 reflection is based fibers is different and requires careful treatment spread around the equator in azimuthal direction, from which the distribution of tilt angles of the crystallites with respect to the fiber axis is obtained. For none of the fibers 2. Experimenta investigated we could observe general hkl reflections indicating an essentially two-dimensional crystal structure Two types of pitch-based fibers and one type of PAN- (turbostratic structure)[1, 2, 6]. The 10 band occurs as a based fiber were investigated. The fibers were received ring of increasing intensity from the equator towards the from commercial sources, and therefore no details about meridian due to the random orientation of the [101 the precursor materials and the production routes could be directions within the 2D layer planes around the stacking figured out. The pitch-based fibers were chosen so that direction of the crystallites they exhibited a considerable difference in Youngs modulus. The PAN-based fibers were exposed to different 3. I. Cross-sectional texture of single fi heat treatment temperatures(HTT) to obtain the same effect. Four PAN-based fibers were measured: as-received n a recent paper [24 we have shown that the cross and HTT of 1800, 2100 and 2400.C. Some properties of sectional texture of carbon fibers can be quantitatively the fibers are shown in table 1. the microbeam diffraction modeled from data measured with microbeam xrd in measurements were carried out at the microfocus beamline fiber geometry. The model is based on the fact that for a at the European Synchrotron Radiation Facility(ESRF) in non-random cross-sectional texture (e.g. radial, radi Grenoble(France), providing a microbeam with a diameter folded or onion skin) the total equatorial intensity from of =2 um. Each fiber was linearly scanned through the 002 reflection and from the 10 band exhibit different beam in fiber geometry with a step width of 0.5 um, and a dependencies on position when scanning the fibers across a two-dimensional (2D) diffraction pattern was collected for microbeam. For a radially folded structure, two parameters every scanning step using an area detector. Details of the could be derived from fitting of the diffraction data: (i) the experimental set-up and data treatment can be found amount(CRe) of crystallites which form the radial folded elsewhere [23]. For comparison, fiber bundles were also structure, the rest(I-cre) being randomly oriented, and investigated with the fibers slightly pre-stretched to keep (ii) the average tilt angle of the crystallites with respect parallel. Diffraction data were collected in the same to the main radial orientation(radial fold) geometry as for the single fiber experiments using a The analysis of the present data using this concept yield laboratory X-ray instrument equipped with an area detec- a random cross-sectional texture for all four PAN-based fibers and a radial folded structure for the two pitch-based fibers. The result of the fits to the diffraction data for the pitch-based fibers are summarized in Table 1. A range of 3. Results and discussion ±5% in the fraction cor and of±s° in the tilt angle should be taken as the uncertainty of the fits [24]. The radial the main features of diffraction from carbon fibers were folded structure for the two pitch-based fibers could be ready described in detail in the literature, e.g. by Ruland also qualitatively deduced from SEM micrographs of
552 O. Paris et al. / Carbon 40 (2002) 551 –555 Table 1 Some parameters for the fibers, together with the results of the present investigation Fiber name and Fiber Radius Modulus Density cRF v w FWHM FWHM 3 manufacturer type (mm) (GPa) (g/cm ) (%) (8) SF (8) FB (8) E35 (DuPont) Pitch 5.33 241 2.10 8 627 21.5 26.3 FT500 (Tonen) Pitch 4.83 542 2.11 21 646 11.9 15.8 HTA7-AR (Tenax) PAN 3.50 206 1.77 0 – 35.4 39.5 HTA7-18 (Tenax) PAN 3.55 257 1.77 0 – 28.4 32.3 HTA7-21 (Tenax) PAN 3.50 273 1.78 0 – 21.3 24.6 HTA7-24 (Tenax) PAN 3.35 315 1.91 0 – 17.7 20.9 The four PAN-based fibers differ by their final HTT (as-received, 1800, 2100 and 24008C as indicated by the name). The mean fiber radius was determined by OM and SEM on the very same fibers as used for the diffraction experiments. The modulus was obtained from tensile tests and the density data were supplied by the data sheets from the companies. cRF is the amount of crystallites building up the radial folded structure and w their average tilt angle with respect to the main radial orientation [24]. The full-width at half maximum (FWHM) of the v azimuthal intensity distribution of the 002 reflections was determined for fiber bundles (FB) and single fibers (SF, mean value weighted by fiber thickness). present work shows that simultaneously the axial and the and co-workers [25,26], and 2D diffraction patterns from cross-sectional texture of carbon fibers are obtained, but single carbon fibers can be found in our earlier work the interpretation of the results for PAN-based and pitch- [23,24]. In a fiber diffraction pattern, the 002 reflection is based fibers is different and requires careful treatment. spread around the equator in azimuthal direction, from which the distribution of tilt angles of the crystallites with respect to the fiber axis is obtained. For none of the fibers 2. Experimental investigated we could observe general hkl reflections, indicating an essentially two-dimensional crystal structure Two types of pitch-based fibers and one type of PAN- (turbostratic structure) [1,2,6]. The 10 band occurs as a based fiber were investigated. The fibers were received ring of increasing intensity from the equator towards the from commercial sources, and therefore no details about meridian due to the random orientation of the [10] the precursor materials and the production routes could be directions within the 2D layer planes around the stacking figured out. The pitch-based fibers were chosen so that direction of the crystallites. they exhibited a considerable difference in Young’s modulus. The PAN-based fibers were exposed to different 3.1. Cross-sectional texture of single fibers heat treatment temperatures (HTT) to obtain the same effect. Four PAN-based fibers were measured: as-received, In a recent paper [24] we have shown that the cross and HTT of 1800, 2100 and 24008C. Some properties of sectional texture of carbon fibers can be quantitatively the fibers are shown in Table 1. The microbeam diffraction modeled from data measured with microbeam XRD in measurements were carried out at the microfocus beamline fiber geometry. The model is based on the fact that for a at the European Synchrotron Radiation Facility (ESRF) in non-random cross-sectional texture (e.g. radial, radial Grenoble (France), providing a microbeam with a diameter folded or onion skin) the total equatorial intensity from the of ¯2 mm. Each fiber was linearly scanned through the 002 reflection and from the 10 band exhibit different beam in fiber geometry with a step width of 0.5 mm, and a dependencies on position when scanning the fibers across a two-dimensional (2D) diffraction pattern was collected for microbeam. For a radially folded structure, two parameters every scanning step using an area detector. Details of the could be derived from fitting of the diffraction data: (i) the experimental set-up and data treatment can be found amount (c ) of crystallites which form the radial folded RF elsewhere [23]. For comparison, fiber bundles were also structure, the rest (12cRF) being randomly oriented, and investigated with the fibers slightly pre-stretched to keep (ii) the average tilt angle w of the crystallites with respect v parallel. Diffraction data were collected in the same to the main radial orientation (radial fold). geometry as for the single fiber experiments using a The analysis of the present data using this concept yields laboratory X-ray instrument equipped with an area detec- a random cross-sectional texture for all four PAN-based tor. fibers and a radial folded structure for the two pitch-based fibers. The result of the fits to the diffraction data for the pitch-based fibers are summarized in Table 1. A range of 3. Results and discussion 65% in the fraction c and of 658 in the tilt angle should RF be taken as the uncertainty of the fits [24]. The radially The main features of diffraction from carbon fibers were folded structure for the two pitch-based fibers could be already described in detail in the literature, e.g. by Ruland also qualitatively deduced from SEM micrographs of
O. Paris et al./Carbon 40(2002)551-555 553 fracture surfaces of fibers from the same bundle, and from tion) can be attributed to a tilt of the fibers within the optical micrographs of the polished cross-sections of the bundles. This is an important finding, since the preferred very same fibers as investigated with microbeam XRd orientation is one of the major parameters controlling the Young's modulus of the fibers From the experimental fact that only a small amount of There is not a unique relationship between the FWHM rystallites contribute to the radial folded cross-sectional and the Youngs modulus(given in Table 1) for all type exture(21% for FT500 and 8% for E35), we conclude of carbon fibers investigated, e.g. E35 has a lower modulus that the cross-sectional structure of the two pitch-based than HTA7-18, but the FWHM is narrower. This is fibers is built up by two 'phases, a randomly oriented probably the consequence of differences in size, shape and phase and a radial folded phase. Differences between the perfection of the crystallites, generally observed for PAN two fibers concern the relative amount of the phases as and pitch-based fibers [4, 12, 27]. when considering PAN- well as a difference in the radial fold, i.e. the larger angle and pitch-based fibers separately, the finding that the p for FT500 means a stronger undulation around the main degree of preferred orientation determines the modulus of radial direction(Table 1). The possible existence of two the fibers is consistent with the literature [2, 4, 9, 10, 15, 271 distinct phases in pitch-based fibers was already reported in the literature [4, 17 3.3. Correlation between axial preferred orientation and 3.2. Axial preferred orientation of single fibers and fiber cross-sectional texture bundles In Fig. 1, there is no significant dependence of the The width(FWHM) of the azimuthal intensity dis- FWHM on position for the PAN-based fibers with low tribution of the 002 reflection was used as a ure for HTT (HTA7-AR and HTA7-18). A slight, but systematic the preferred orientation, i.e. the higher the FWHM, the decrease from the center to the border of the fibers can be lower the degree of preferred orientation. Fig. la shows the seen for HTA7-21 and HTA7-24, the effect being more FWHM (in degrees) as a function of position across the pronounced the higher the HTT. This is an indication for a fibers, and Fig. Ib an enlargement of Fig. la for the two skin-core structure with higher preferred orientation at the fibers with the smallest FWHM. In Table 1. the mea skin of the fibers, developing with increasing values of the FWHM from Fig. I are compared to those accordance with the literature [ 12]. Unfortunately, our data obtained from fiber-bundles. Remarkably, the values from are strongly smeared due to the relatively large size of the the fiber-bundle measurements are consistently about 4-5 X-ray beam(2 um, compared to a fiber diameter of 7 degrees higher than the corresponding values for the single um). New developments of significantly smaller mi- fibers. This difference(actually it would be a deconvolu- crobeams down to 0.1 um [18, 22] could enable the determination of many more details about such skin layers in the future mpared to the position dependence of the FwHM for the PAN-based fibers, the opposite trend is seen for the two pitch-based fibers in Fig. 1. E35 exhibits only a slight increase. whereas FT500 shows a considerable increase of FWHM towards the border of the fiber. A straightforward interpretation would suggest a higher degree of axial preferred orientation in the fiber core, even though the ccurrence of such an "inverse skin-core structure would f 13 be very surprising. The degree of preferred orientation was al ways found higher at the fiber border, not only in carbon fibers [12], but also for example in viscose fibers [21] However, the above interpretation is not necessarily cor- rect if the cross-sectional fiber structure is not random which is indeed the case for the radially folded structures of FT500 and E35(Section 3.1 Fig. 1.(a)Full width at half maximum( FWHm) of the azimuthal To understand this. one has to consider the whole 002 intensity distributions of the 002 refiection as a function of the Ix, rather than only the width(FWHM). A fit of I fiber) n symbols, PAN-bas HTA7-AR (squares) with a single Gaussian gives indeed rather poor fit results HTA7-18(circles), HTA7-21(triangles), HTA7-24(diamonds). In Fig 2a, two /(rr-curves are plotted, one from the center losed symbols, pitch-based fibers: E35(black triangles) and one from the border of fr500. Each of the two curves FT500(black circles).(b) Larger scale representation of (a)for was fitted with the sum of two Gaussians(solid lines in HTA7-24 and FT500(symbols are the same as in(a)) Ig
O. Paris et al. / Carbon 40 (2002) 551 –555 553 fracture surfaces of fibers from the same bundle, and from tion) can be attributed to a tilt of the fibers within the optical micrographs of the polished cross-sections of the bundles. This is an important finding, since the preferred very same fibers as investigated with microbeam XRD orientation is one of the major parameters controlling the [24]. Young’s modulus of the fibers. From the experimental fact that only a small amount of There is not a unique relationship between the FWHM crystallites contribute to the radial folded cross-sectional and the Young’s modulus (given in Table 1) for all types texture (21% for FT500 and 8% for E35), we conclude of carbon fibers investigated, e.g. E35 has a lower modulus that the cross-sectional structure of the two pitch-based than HTA7-18, but the FWHM is narrower. This is fibers is built up by two ‘phases’, a randomly oriented probably the consequence of differences in size, shape and phase and a radial folded phase. Differences between the perfection of the crystallites, generally observed for PAN two fibers concern the relative amount of the phases as and pitch-based fibers [4,12,27]. When considering PANwell as a difference in the radial fold, i.e. the larger angle and pitch-based fibers separately, the finding that the w for FT500 means a stronger undulation around the main degree of preferred orientation determines the modulus of v radial direction (Table 1). The possible existence of two the fibers is consistent with the literature [2,4,9,10,15,27]. distinct phases in pitch-based fibers was already reported in the literature [4,17]. 3.3. Correlation between axial preferred orientation and 3.2. Axial preferred orientation of single fibers and fiber cross-sectional texture bundles In Fig. 1, there is no significant dependence of the The width (FWHM) of the azimuthal intensity dis- FWHM on position for the PAN-based fibers with low tribution of the 002 reflection was used as a measure for HTT (HTA7-AR and HTA7-18). A slight, but systematic the preferred orientation, i.e. the higher the FWHM, the decrease from the center to the border of the fibers can be lower the degree of preferred orientation. Fig. 1a shows the seen for HTA7-21 and HTA7-24, the effect being more FWHM (in degrees) as a function of position across the pronounced the higher the HTT. This is an indication for a fibers, and Fig. 1b an enlargement of Fig. 1a for the two skin-core structure with higher preferred orientation at the fibers with the smallest FWHM. In Table 1, the mean skin of the fibers, developing with increasing HTT in values of the FWHM from Fig. 1 are compared to those accordance with the literature [12]. Unfortunately, our data obtained from fiber-bundles. Remarkably, the values from are strongly smeared due to the relatively large size of the the fiber-bundle measurements are consistently about 4–5 X-ray beam (¯2 mm, compared to a fiber diameter of ¯7 degrees higher than the corresponding values for the single mm). New developments of significantly smaller mi- fibers. This difference (actually it would be a deconvolu- crobeams down to 0.1 mm [18,22] could enable the determination of many more details about such skin layers in the future. Compared to the position dependence of the FWHM for the PAN-based fibers, the opposite trend is seen for the two pitch-based fibers in Fig. 1. E35 exhibits only a slight increase, whereas FT500 shows a considerable increase of FWHM towards the border of the fiber. A straightforward interpretation would suggest a higher degree of axial preferred orientation in the fiber core, even though the occurrence of such an ‘inverse’ skin-core structure would be very surprising. The degree of preferred orientation was always found higher at the fiber border, not only in carbon fibers [12], but also for example in viscose fibers [21]. However, the above interpretation is not necessarily correct if the cross-sectional fiber structure is not random, which is indeed the case for the radially folded structures of FT500 and E35 (Section 3.1). To understand this, one has to consider the whole 002 Fig. 1. (a) Full width at half maximum (FWHM) of the azimuthal intensity distribution as a function of azimuthal angle x, intensity distributions of the 002 reflection as a function of the I(x), rather than only the width (FWHM). A fit of I(x) position across the fiber diameter (position50 is the center of the with a single Gaussian gives indeed rather poor fit results. fiber). Open symbols, PAN-based fibers: HTA7-AR (squares), In Fig. 2a, two I(x)-curves are plotted, one from the center HTA7-18 (circles), HTA7-21 (triangles), HTA7-24 (diamonds). Closed symbols, pitch-based fibers: E35 (black triangles), and and one from the border of FT500. Each of the two curves FT500 (black circles). (b) Larger scale representation of (a) for was fitted with the sum of two Gaussians (solid lines in HTA7-24 and FT500 (symbols are the same as in (a)). Fig. 2a)
O. Paris et al./Carbon 40(2002)551-555 the cross-section ( Section 3. 1), and different degrees of preferred orientation in axial direction, respectively, could uccessfully describe all observed results. One has just to attribute the two Gaussian components used for fitting of the l(xr-curves(Fig. 2a)to the two phases, i.e. the random phase to the broad component and the radial folded phase to the narrow component(Fig 3). According to the model developed in Ref [24], the(volume normalized) contribu- tion of a radially folded phase to the 002 diffraction signa will decrease towards the fiber edge. while the contributio of a random phase will remain constant. This explains the increase of the area ratio in Fig. 2b. In the fiber center, the area of the two gaussia ponents will be roughly azimuthal angle x Degrees) proportional to the relative amount of the two phases. An se with smaller FWHM (radially folded phase) is estimated for FT500 from Fig. 2b. This is in good agreement with the value of 21+5% b reflections and the 10 bands (Table 1), supporting our interpretation further. In this respect it is also not surpris- ing that E35 shows not such a distinct position dependence FWHM(degrees of the axial preferred orientation (Fig. la), since the amount of radially oriented phase is considerably smaller this fiber Fig. 3 tries to visualize the interpretation of our data in a ,·。●·。。·●· very simplified sketch. The drawing should however not be over-interpreted, since we cannot deduce any details about how the two phases are contained within the fibers. In articular, the disc-shaped units in Fig. 3 were just used for position y (um) convenience to illustrate the correlation found between the Fig. 2.(a) Azimuthal intensity distributions loo(x) for the fiber FT500 for two different positions: close to the fiber center(open triangles) and close to the fiber edge(black circles). The full lines are fits to the data using Eq. (1) with parameters a,=7.40 0,=4.02(fixed for both experimental curves), a=0.48 for the fiber center and a=0.68 for the fiber border. (b) The area ratio A, /,=ao, /(1-a)o, as a function of position for FT500 shown by the black circles. The insert shows that there is a linear 十 nip between the area ratio and the FwHM(from Fig. 1) lx=aex here the widths have been determined for one curve and were then kept fixed for all other curves. All measured distributions /(x)across the fiber could be perfectly fitted Fig. 3. Sketch of the correlation between the cross-sectional by Eq. (1)with fixed widths(,=7.40, 02=4.02), the texture and the axial preferred orientation. Single hexagonal amplitude a of the broad component being the only carbon layers(rather than crystallites) are schematically visualized by circular disks. These disks are oriented preferentially parallel to arying parameter. Fig. 2b shows the ratio of the areas the fiber axis with an average tilt angle o(standard deviation of a lA, =ao,/(1-a)o, as a function of position. Remark bly, the area ratio in Fig. 2b shows exactly the same Gaussian, see Eq. (1))with respect to the fiber axis. For the random phase these disks are randomly oriented position dependence as the FWhm in Fig. 1b. To under- within the fiber cre whereas they form a radially folded line this fact, the linear relationship between the area ratio structure for the se se(right side, radial folded traces are Two phases, with random and radial folded structure in random phase (on)than for the radial folded phase(%/ or the and the FWhM is shown in the insert in Fig. 2b indicated by the ) The tilt angle o is larger
554 O. Paris et al. / Carbon 40 (2002) 551 –555 the cross-section (Section 3.1), and different degrees of preferred orientation in axial direction, respectively, could successfully describe all observed results. One has just to attribute the two Gaussian components used for fitting of the I(x)-curves (Fig. 2a) to the two phases, i.e. the random phase to the broad component and the radial folded phase to the narrow component (Fig. 3). According to the model developed in Ref. [24], the (volume normalized) contribution of a radially folded phase to the 002 diffraction signal will decrease towards the fiber edge, while the contribution of a random phase will remain constant. This explains the increase of the area ratio in Fig. 2b. In the fiber center, the area of the two Gaussian components will be roughly proportional to the relative amount of the two phases. An amount of ¯25% for the phase with smaller FWHM (radially folded phase) is estimated for FT500 from Fig. 2b. This is in good agreement with the value of 2165% from the modeling of the total intensity from the 002 reflections and the 10 bands (Table 1), supporting our interpretation further. In this respect it is also not surprising that E35 shows not such a distinct position dependence of the axial preferred orientation (Fig. 1a), since the amount of radially oriented phase is considerably smaller for this fiber. Fig. 3 tries to visualize the interpretation of our data in a very simplified sketch. The drawing should however not be over-interpreted, since we cannot deduce any details about how the two phases are contained within the fibers. In particular, the disc-shaped units in Fig. 3 were just used for convenience to illustrate the correlation found between the Fig. 2. (a) Azimuthal intensity distributions I (x) for the fiber 002 FT500 for two different positions: close to the fiber center (open triangles) and close to the fiber edge (black circles). The full lines are fits to the data using Eq. (1) with parameters s 5 7.408, 1 s 5 4.028 (fixed for both experimental curves), a50.48 for the 2 fiber center and a50.68 for the fiber border. (b) The area ratio A /A 5 as /(1 2 a)s as a function of position for FT500 is 12 1 2 shown by the black circles. The insert shows that there is a linear relationship between the area ratio and the FWHM (from Fig. 1). 2 2 x x I(x) 5 a expSD SD 2 1 ] ] (1 2 a) exp 2 (1) 2 2 2s 1 2 2s where the widths have been determined for one curve and were then kept fixed for all other curves. All measured Fig. 3. Sketch of the correlation between the cross-sectional distributions I(x) across the fiber could be perfectly fitted texture and the axial preferred orientation. Single hexagonal by Eq. (1) with fixed widths (s 5 7.408, s 5 4.028), the 1 2 carbon layers (rather than crystallites) are schematically visualized amplitude a of the broad component being the only by circular disks. These disks are oriented preferentially parallel to varying parameter. Fig. 2b shows the ratio of the areas the fiber axis with an average tilt angle s (standard deviation of a A /A 5 as /(1 2 a)s as a function of position. Remark- 12 1 2 Gaussian, see Eq. (1)) with respect to the fiber axis. For the ably, the area ratio in Fig. 2b shows exactly the same ‘random phase’ (left side) these disks are randomly oriented position dependence as the FWHM in Fig. 1b. To under- within the fiber cross section, whereas they form a radially folded line this fact, the linear relationship between the area ratio structure for the second ‘phase’ (right side, radial folded traces are and the FWHM is shown in the insert in Fig. 2b. indicated by the thick lines). The tilt angle s is larger for the Two phases, with random and radial folded structure in random phase (s ) than for the radial folded phase (s ). 1 2
O. Paris et al./Carbon 40(2002)551-555 axial preferred orientation and the cross-sectional texture, microbeam XRD experiments. Special thanks also to E and have nothing to do with any real shape of the Haberz and G. Moser for sample preparation and optical crystallites or of possible fibrils containing several crys- microscopy, and to P. Fratzl for valuable discussions allies. Moreover, the question whether the pitch-based Financial support from the Austrian Science Foundation fibers do really consist of two distinct phases(e.g fibrils of under project no. P-14294-PhY is gratefully acknowl- a highly oriented phase in a less oriented matrix), or edged whether there is a whole continuum of orientational order of the basic layers, cannot be ultimately concluded from our investigations. Such information could only be pro- References vided by high resolution electron microscopy(HRTEM) Investigations referring to this are currently in preparation [1 Oberlin A Carbon 1984 22: 521-41 2]Ruland w. Appl Polym Symp 1969: 9: 293-301 3] Reynolds WN, Sharp JV Carbon 1974: 12: 103-10 4. Conclusions 14] Diefendorf RJ, Tokarsky E. Polym Eng Sci 1975: 15: 150-9 5] Bennett SC, Johnson DJ, Johnson w. J Mater Sci Microbeam X-ray diffraction can provide detaled quan- [6)Guigon M, Oberlin A, Desarmot G. Fibre Sci Technol 1984;20:55-72. and the cross-sectional macro-texture in single carbon [7 Endo M. J Mater Sci 1988: 23: 598-605 fibers. In particular, the following have been shown [8 Kumar s, Anderson DP, Crasto AS. J Mater Sci 1993;28:423-39 1. The axial preferred orientation can be derived without [9] Huang Y, Young RJ. J Mater Sci 1994, 29: 4027-36 e influence of fiber tilt included in fiber bundle [10] Perret R, Ruland WJ. J Appl Cryst 1970; 3: 525-32 relating the preferred orientation to the fiber modulus. [12] Bennett SC, Johnson DJ Carbon 1979: 17: 25-l9'o measurements. This knowledge ortant whe 2. The cross sectional texture of the four PAN-based fibers [13] Fortin F, Yoon SH, Korai I, Mochida I. J Mater Sci is random. The axial preferred orientation increases 1995;30:4567-83 with increasing HTT, and a higher degree of preferred [14] Vezie DL, Adams ww. J Mater Sci Lett 1990, 9: 883-7 orientation develops with increasing HTT in a thin skin [15 Edie DD. Carbon 1998: 36: 345-62 [16] Barnes AB, Dauche FM, Gallego NC, Fain CC, Thies MC layer, compared to the fiber core. The cross-sectional Carbon1998;36:855-60 texture of the two pitch-based fibers was found to [17] Hong SH, Korai Y, Mochida I Carbon 1999: 37: 917-30 consist of a mixture of a radial folded phase and a [18] Riekel C Rep Prog Phys 3233-62 andom phase. Quantitative values for the amount of the [19] Riekel C, Cedola A, Heidelbach F, Wagner K. Macro- phases and the magnitudes of the undulations of the molecules 1997- 30: 1033-7 radial folds were derived. Such information about the [20] Miller M, Czihak C, Vogl G, Schober H, Riekel CMacro- detailed cross-sectional texture of carbon fibers could molecules 1998: 31: 3953 be relevant in relating fiber structure to mechanical [21] Muller M, Riekel C, Vuong R, Chanzy H. Polymer 200041:2627-32. properties, such as fiber strength [22 Muller M, Burghammer M, Flot D, Riekel C, Morawe C 3. The axial preferred orientation and the cross-sectional Murphy B et al. J Appl Cryst 2000: 33: 1231-40 texture of the two pitch-based fibers are correlated. The [23] Paris O, Loidl D, Peterlik H, Muller M, Lichtenegger H, andomly oriented phase within the cross-section ex Fratzl P. J Appl Cryst 2000: 33: 695-9 a lower preferred orientation than the radial [24] Paris O, Loidl D, Muller M, Lichtenegger H, Peterlik H. J Cryst 2001:, in press [25] Ruland W, Tompa H. Acta Crystallogr 1968: A24: 93-9 226] Fourdeux A, Perret R, Ruland wJ Appl Cryst 1968: 1: 252 Acknowledgements [27] Huang Y, Young RJ. Carbon 1995: 33: 97-107 We are grateful to the esre staff, in particular to M Burghammer and C. Riekel for help with the scanning
O. Paris et al. / Carbon 40 (2002) 551 –555 555 axial preferred orientation and the cross-sectional texture, microbeam XRD experiments. Special thanks also to E. and have nothing to do with any real shape of the Haberz and G. Moser for sample preparation and optical crystallites or of possible fibrils containing several crys- microscopy, and to P. Fratzl for valuable discussions. tallites. Moreover, the question whether the pitch-based Financial support from the Austrian Science Foundation fibers do really consist of two distinct phases (e.g. fibrils of under project no. P-14294-PHY is gratefully acknowla highly oriented phase in a less oriented matrix), or edged. whether there is a whole continuum of orientational order of the basic layers, cannot be ultimately concluded from our investigations. Such information could only be pro- References vided by high resolution electron microscopy (HRTEM). Investigations referring to this are currently in preparation. [1] Oberlin A. Carbon 1984;22:521–41. [2] Ruland W. Appl Polym Symp 1969;9:293–301. [3] Reynolds WN, Sharp JV. Carbon 1974;12:103–10. [4] Diefendorf RJ, Tokarsky E. Polym Eng Sci 1975;15:150–9. 4. Conclusions [5] Bennett SC, Johnson DJ, Johnson W. J Mater Sci 1983;18:3337–47. Microbeam X-ray diffraction can provide detailed quan- [6] Guigon M, Oberlin A, Desarmot G. Fibre Sci Technol titative information about the axial preferred orientation 1984;20:55–72. and the cross-sectional macro-texture in single carbon [7] Endo M. J Mater Sci 1988;23:598–605. fibers. In particular, the following have been shown. [8] Kumar S, Anderson DP, Crasto AS. J Mater Sci 1993;28:423–39. 1. The axial preferred orientation can be derived without [9] Huang Y, Young RJ. J Mater Sci 1994;29:4027–36. [10] Perret R, Ruland WJ. J Appl Cryst 1970;3:525–32. the influence of fiber tilt included in fiber bundle measurements. This knowledge is important when [11] Takaku A, Shioya M. J Mater Sci 1990;25:4873–9. [12] Bennett SC, Johnson DJ. Carbon 1979;17:25–39. relating the preferred orientation to the fiber modulus. [13] Fortin F, Yoon SH, Korai I, Mochida I. J Mater Sci 2. The cross sectional texture of the four PAN-based fibers 1995;30:4567–83. is random. The axial preferred orientation increases [14] Vezie DL, Adams WW. J Mater Sci Lett 1990;9:883–7. with increasing HTT, and a higher degree of preferred [15] Edie DD. Carbon 1998;36:345–62. orientation develops with increasing HTT in a thin skin [16] Barnes AB, Dauche FM, Gallego NC, Fain CC, Thies MC. layer, compared to the fiber core. The cross-sectional Carbon 1998;36:855–60. texture of the two pitch-based fibers was found to [17] Hong SH, Korai Y, Mochida I. Carbon 1999;37:917–30. consist of a mixture of a radial folded phase and a [18] Riekel C. Rep Prog Phys 2000;63:233–62. [19] Riekel C, Cedola A, Heidelbach F, Wagner K. Macro- random phase. Quantitative values for the amount of the molecules 1997;30:1033–7. phases and the magnitudes of the undulations of the [20] Muller M, Czihak C, Vogl G, Schober H, Riekel C. Macro- ¨ radial folds were derived. Such information about the molecules 1998;31:3953–7. detailed cross-sectional texture of carbon fibers could [21] Muller M, Riekel C, Vuong R, Chanzy H. Polymer ¨ be relevant in relating fiber structure to mechanical 2000;41:2627–32. properties, such as fiber strength. [22] Muller M, Burghammer M, Flot D, Riekel C, Morawe C, ¨ 3. The axial preferred orientation and the cross-sectional Murphy B et al. J Appl Cryst 2000;33:1231–40. texture of the two pitch-based fibers are correlated. The [23] Paris O, Loidl D, Peterlik H, Muller M, Lichtenegger H, ¨ randomly oriented phase within the cross-section ex- Fratzl P. J Appl Cryst 2000;33:695–9. hibits a lower preferred orientation than the radial [24] Paris O, Loidl D, Muller M, Lichtenegger H, Peterlik H. J ¨ Appl Cryst 2001;, in press. folded phase. [25] Ruland W, Tompa H. Acta Crystallogr 1968;A24:93–9. [26] Fourdeux A, Perret R, Ruland W. J Appl Cryst 1968;1:252– 4. Acknowledgements [27] Huang Y, Young RJ. Carbon 1995;33:97–107. We are grateful to the ESRF staff, in particular to M. Burghammer and C. Riekel for help with the scanning