CARBON PERGAMON Carbon38(2000)1323-1337 Mechanical, surface and interfacial characterisation of pitch and PAN-based carbon fibres M.C. Paiva. C.A. Bernardo,*M.Nardin Departamento de Engenharia de polimeros, Universidade do Minho, Campus de A=urem, 4800 Guimaraes, Portuga Institut de Chimie des surfaces et Interfaces, CNRS-UPR 9069, 15 rue Jean Starcky, B.P. 2488, F-68057 Mulhouse Cedex, Fr Received 4 August 1999, accepted 18 November 1999 Abstract The mechanical and surface characteristics of pitch and PAN-based carbon fibres were studied by tensile testing, XPS SEM analysis and wetting measurements. The pitch-based fibres had two different geometries, with circular and ellipsoidal (ribbon-shaped) cross sections. Plasma oxidation was used to treat the surface of the fibres. The interfacial characteristics of untreated and treated fibres were measured by fragmentation tests of single filament composites. The effect of the surface treatment on the mechanical, surface and interfacial properties of the fibres was determined and correlated. It was shown that a relationship exists between the ability of the surface to transfer loads and its oxygen content. Finally, the influence of the on-axisymmetry on the interfacial parameters obtained in the fragmentation tests was assessed. o 2000 Elsevier Science td. All rights reserved Keywords: A. Carbon fibres; B. Surface treatment; D. Interfacial properties, Mechanical properties, Surface properties 1. Introduction 2. Theoretical basis of the work uring the last two decades there has been considerable 2. 1. Mechanical properties of the fibres progress in the production of high mechanical, thermal and electrical property carbon fibres from mesophase pitch The micro-mechanical analysis of an interface requires a [1-3]. This led to the development of fibres from different complete knowledge of the mechanical properties of the precursor compositions and improved production condi- materials involved. In the case of polymer-carbon fibre tions. Different fibre shapes were also prepared [4]. composites, the matrix material can be characterised Usually, their final application on composite materials mechanically with considerable accuracy. This is more requires surface treatment, in order to enhance the surface difficult to achieve for the carbon fibres energy and thus increase the ability to establish strong The tensile strength of carbon fibres is usually assessed interactions with the matrix by single filament tensile tests [5]. The experimental data The present work is a contribution to the characteris- generated by these tests has high scatter, mainly due to the tion of some carbon fibre-polymer interfaces. It can be presence of flaws along the fibres. Thus, the interpretation divided in two main parts; one concems the study of the of the data must be done statistically. Several statistical mechanical and surface characteristics of the fibres, un- distributions have been used to describe tensile strength treated and surface treated, the other deals ith the data, the more flexible so far being the weibull distribution alysis of fibre-matrix interfaces in model composites. [ 6, defined by the two-parameter cumulative distribution The effect of using non-circular fibres is also analysed function described by Eq. (1) F( *Corresponding author. Tel :+35-1-510-101; fax: +35-1- where oo and m represent the Weibull scale and shape 51400 E-mail address. bernardo @eng minho. pt (C.A. Bernardo) The Weibull distribution is typically used to describe life 0008-6223/00/S-see front matter 2000 Elsevier Science Ltd. All rights reserved PII:S0008-6223(99)00266-3
PERGAMON Carbon 38 (2000) 1323–1337 Mechanical, surface and interfacial characterisation of pitch and PAN-based carbon fibres a a, b M.C. Paiva , C.A. Bernardo , M. Nardin * a Departamento de Engenharia de Polımeros ´ ´ , Universidade do Minho, Campus de Azurem, 4800 Guimaraes ˜ , Portugal b Institut de Chimie des Surfaces et Interfaces, CNRS-UPR 9069, 15 rue Jean Starcky, B.P. 2488, F-68057 Mulhouse Cedex, France Received 4 August 1999; accepted 18 November 1999 Abstract The mechanical and surface characteristics of pitch and PAN-based carbon fibres were studied by tensile testing, XPS, SEM analysis and wetting measurements. The pitch-based fibres had two different geometries, with circular and ellipsoidal (ribbon-shaped) cross sections. Plasma oxidation was used to treat the surface of the fibres. The interfacial characteristics of untreated and treated fibres were measured by fragmentation tests of single filament composites. The effect of the surface treatment on the mechanical, surface and interfacial properties of the fibres was determined and correlated. It was shown that a relationship exists between the ability of the surface to transfer loads and its oxygen content. Finally, the influence of the non-axisymmetry on the interfacial parameters obtained in the fragmentation tests was assessed. 2000 Elsevier Science Ltd. All rights reserved. Keywords: A. Carbon fibres; B. Surface treatment; D. Interfacial properties, Mechanical properties, Surface properties 1. Introduction 2. Theoretical basis of the work During the last two decades there has been considerable 2.1. Mechanical properties of the fibres progress in the production of high mechanical, thermal and electrical property carbon fibres from mesophase pitch The micro-mechanical analysis of an interface requires a [1–3]. This led to the development of fibres from different complete knowledge of the mechanical properties of the precursor compositions and improved production condi- materials involved. In the case of polymer–carbon fibre tions. Different fibre shapes were also prepared [4]. composites, the matrix material can be characterised Usually, their final application on composite materials mechanically with considerable accuracy. This is more requires surface treatment, in order to enhance the surface difficult to achieve for the carbon fibres. energy and thus increase the ability to establish strong The tensile strength of carbon fibres is usually assessed interactions with the matrix. by single filament tensile tests [5]. The experimental data The present work is a contribution to the characterisa- generated by these tests has high scatter, mainly due to the tion of some carbon fibre–polymer interfaces. It can be presence of flaws along the fibres. Thus, the interpretation divided in two main parts; one concerns the study of the of the data must be done statistically. Several statistical mechanical and surface characteristics of the fibres, un- distributions have been used to describe tensile strength treated and surface treated, the other deals with the data, the more flexible so far being the Weibull distribution analysis of fibre–matrix interfaces in model composites. [6], defined by the two-parameter cumulative distribution The effect of using non-circular fibres is also analysed. function described by Eq. (1): s m Fs d s; s , m 5 1 2 expS S DD 2 ] (1) 0 s0 *Corresponding author. Tel.: 135-1-510-101; fax: 135-1- where s and m represent the Weibull scale and shape 0 51400. parameters, respectively. E-mail address: cbernardo@eng.uminho.pt (C.A. Bernardo). The Weibull distribution is typically used to describe life 0008-6223/00/$ – see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S0008-6223(99)00266-3
1324 M C. Paira et al. Carbon 38(2000)1323-1337 pmm时tmha22 (4) fibre failure leads to a length dependence of its strength: the longer the fibre length, the larger the number where Buy is the contact angle of the liquid on the solid of flaws that are present and the higher the probability of in the presence of its vapour, x is the surface energy of occurrence of a severe faw. The Weibull distribution has liquid and yi is the dispersive component to be adapted to account for that dependence. a simple that energy. The use of the tensiometric method allows to way to do so is through the weakest link approximation. equate cos 0 y to the force F exerted on the fibre as it It assumes the fibre to be formed by L independent links of penetrates the wetting liquid arbitrary unit length, each link failing or surviving at a given stress level, independently of its neighbours. It also F assumes that the strength distribution of each independent COsTs/ p link is described by a simple Weibull distribution, char- acterised by identical parameters. This leads to a fibre where p is The non survival probability equal to the product of th survIvE sive component, ys, was determined by probability of each link. On the basis of these considera- the two-liquid phase method [ll], using water and a hydrocarbon as wetting liquids. In this method, the non- ons, the Weibull cumulative distribution function F(o or dispersive interactions between the fibre and water can be m) and the corresponding mean strength(o), adapted account for the gauge length dependence of the fibres (L), may be described by Eqs. (2)and (3) wnu.= F(o o, m) where wsw is the non-dispersive interaction between the =aL-n1+1/m) (3) water and the solid, w and y represent the surface energies of water and hydrocarbon, yw is the dispersive here I represents the gamma function component of the surface energy of water, w is the The parameter estimate for the two-parameter Weibull interfacial energy between water/hydrocarbon, distribution was performed for the strength data obtained at contact angle of water on the solid in the presence all gauge lengths simultaneously. In this work, the maxi- the hydrocarbon mum likelihood theory was used to determine the Weibull parameters. The method used was that described in Ston- er's thesis [8], as well as the calculation programs therein 2.3. Interfacial properties In this way, for each type of fibre, a single set of parameters is obtained that fit all the gauge lengths tested bre/ matrix interface was evaluated by fragmenta- [9]. The estimate of the tensile strength at any gauge ts, performed on single filament composites. The length, needed for the interfacial shear strength calcula- consists basically on the tensile test of a single tions, is performed simply by substituting the calculated bedded on a matrix material. as the strain parameters in Eq.(3), for the specified fibre lengt s, the fibre breaks at the points where the ultimate fibre strength is reached. The matrix must have a strain-to- failure at least three times higher than that of the fibre, so 2. 2. Surface composition and energetics that the fragmentation process can reach saturation before matrix yielding. The process will stop at the point where The surface chemistry of the fibres was studied by X-ray the fibre fragments are so small that the shear stress photoelectron spectroscopy (XPS), and the surface func- transferred through the interface is not enough to produce tionality determined and quantified by spectral deconvolu- further fibre breakage. The interpretation of the test data was based on the Kelly-Tyson approach [12]. This Surface energy measurements were performed using a approach considers that there is a maximum shear strength tensiometric method(Wilhelmy technique)[10). The total that can be sustained by the interface, T, characteristic of surface energy of the fibres, y, was estimated as the sum each particular interface. Above T there is decohesion of a dispersive component, ys, corresponding to London(debonding) at the fibre/matrix interface(or yielding of the interactions only, and a non-dispersive (or polar)com matrix). To each T value corresponds a critical fibre onent, ys, that can be associated to other types of length, le, or more accurately, a fibre length distribution interaction(Debye, Keesom, acid-base,.). ys was ranging from 1/2 to le, since any fibre with length just determined by the one-liquid phase method, using a non- above le will still break. Considering an approximately polar liquid and considering the non-dispersive interactions symmetrical distribution of fibre fragments, their average negligible. ys was estimated from Eq(4) length, at T, should then be equal to
1324 M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 data, especially when there is little information about the ]d ]d ] œg L population under study [7]. The flaw-induced nature of cosu 5 2 g ] 2 1 (4) SL/V œ S gL fibre failure leads to a length dependence of its tensile strength: the longer the fibre length, the larger the number where uSL/V is the contact angle of the liquid on the solid of flaws that are present and the higher the probability of in the presence of its vapour, g is the surface energy of L occurrence of a severe flaw. The Weibull distribution has d the wetting liquid and g is the dispersive component of L to be adapted to account for that dependence. A simple that energy. The use of the tensiometric method allows to way to do so is through the ‘weakest link’ approximation. equate cos u to the force F exerted on the fibre as it SL/V It assumes the fibre to be formed by L independent links of penetrates the wetting liquid: arbitrary unit length, each link failing or surviving at a F given stress level, independently of its neighbours. It also cosu 5 ] (5) SL/V assumes that the strength distribution of each independent pgL link is described by a simple Weibull distribution, char- where p is the perimeter of the fibre. acterised by identical parameters. This leads to a fibre nd The non-dispersive component, g S , was determined by survival probability equal to the product of the survival the two-liquid phase method [11], using water and a probability of each link. On the basis of these considera- hydrocarbon as wetting liquids. In this method, the non- tions, the Weibull cumulative distribution function F(s; s , 0 dispersive interactions between the fibre and water can be m) and the corresponding mean strength (s¯), adapted to determined by: account for the gauge length dependence of the fibres (L), may be described by Eqs. (2) and (3): nd d d ] ] ] W 5g 2g 1g cosu 2 2 g s g 2 g d SW W H HW SW/H œ S œ W H œ s m (6) Fs d s; s , m 5 1 2 expS S DD 2 L ] (2) 0 s0 nd where W is the non-dispersive interaction between the SW 21 /m s¯ 5 s0L Gs d 1 1 1/m (3) water and the solid, gW H and g represent the surface d energies of water and hydrocarbon, g W is the dispersive where G represents the gamma function. component of the surface energy of water, g is the HW The parameter estimate for the two-parameter Weibull interfacial energy between water/hydrocarbon, and uSW/H distribution was performed for the strength data obtained at is the contact angle of water on the solid in the presence of all gauge lengths simultaneously. In this work, the ‘maxi- the hydrocarbon. mum likelihood’ theory was used to determine the Weibull parameters. The method used was that described in Ston- 2.3. Interfacial properties er’s thesis [8], as well as the calculation programs therein. In this way, for each type of fibre, a single set of The fibre/matrix interface was evaluated by fragmenta- parameters is obtained that fit all the gauge lengths tested tion tests, performed on single filament composites. The [9]. The estimate of the tensile strength at any gauge method consists basically on the tensile test of a single length, needed for the interfacial shear strength calcula- fibre embedded on a matrix material. As the strain tions, is performed simply by substituting the calculated increases, the fibre breaks at the points where the ultimate parameters in Eq. (3), for the specified fibre length. fibre strength is reached. The matrix must have a strain-tofailure at least three times higher than that of the fibre, so 2.2. Surface composition and energetics that the fragmentation process can reach saturation before matrix yielding. The process will stop at the point where The surface chemistry of the fibres was studied by X-ray the fibre fragments are so small that the shear stress photoelectron spectroscopy (XPS), and the surface func- transferred through the interface is not enough to produce tionality determined and quantified by spectral deconvolu- further fibre breakage. The interpretation of the test data tion. was based on the Kelly–Tyson approach [12]. This Surface energy measurements were performed using a approach considers that there is a maximum shear strength tensiometric method (Wilhelmy technique) [10]. The total that can be sustained by the interface, t , characteristic of i surface energy of the fibres, gS, was estimated as the sum each particular interface. Above ti there is decohesion d of a dispersive component, g , corresponding to London (debonding) at the fibre/matrix interface (or yielding of the S interactions only, and a non-dispersive (or polar) com- matrix). To each t value corresponds a critical fibre i nd ponent, g S , that can be associated to other types of length, lc, or more accurately, a fibre length distribution d interaction (Debye, Keesom, acid–base, . . . ). g was ranging from l /2 to l , since any fibre with length just S cc determined by the one-liquid phase method, using a non- above lc will still break. Considering an approximately polar liquid and considering the non-dispersive interactions symmetrical distribution of fibre fragments, their average d negligible. g was estimated from Eq. (4): length, at t , should then be equal to: S i
M C. Paira et al. Carbon 38(2000)1323-1337 2805, a polycarbonate from Bayer. It is a standard grade (7) for injection moulding, with a weight average molecular weight (Mw) of 22 880 g mol and a polydispersivity Applying the equilibrium condition between the tensile index, measured as the ratio M /M,, equal to 1.86.The rce acting on a fibre of diameter d, and the shear forces glass transition temperature, Tg, was equal to 145+1C, as transferred through the interface, a simple expression, measured by differential scanning calorimetry(DSC),on a proposed by Fraser and Di Benedetto [13], can be obtained Perkin-Elmer DSc7 instrument. The coefficient of thermal for the average interfacial shear strength,t: was measured on a Perkin-Elmer dynamic- mechanical analyser, DMA7, working on thermo-mechani- o() (8) cal mode in expansion, in the temperature range of 90 to For a given fibre/matrix system, the smaller the fibre min,apc was determined to be 97x10C. The mechanical characterisation of Makrolon 2805. in tension. fragments obtained in the fragmentation test, the higher the was performed in an Instron 4505 testing machine. interfacial shear strength equipped with a clip strain gauge extensometer. The resul obtained for modulus, yield strength and strain were 2.17±0.06GPa,659±0.5 MPa and5.3±0.3%, respective 3.E 3.1. Materials 3. 2. Surface characterisation The carbon fibres studied in this work had different 3. 2.I. XPs characteristics and were subject to plasma treatment in Two distinct analyses were performed. a global de- section geometries, and the hoe ate ere pitch-based, two termination of the surface atomic composition,in terms of several conditions. Three of them w unsized a ith different cross- verall carbon, oxygen and nitrogen content, was done ras a commercial fibre, with an ESCALAB 200A-VG SCIENTIFIC spectrometer, treated and sized. For comparison, a PAN-based fibre sing a mg /al double X-ray source with a power of 300 unsized and untreated, was also studied. In terms of W. Then, a thorough spectral analysis was run for the mechanical classification, the pitch-based fibres can be pitch-based P120J and ribbon fibres, untreated and plasma considered as ultra-high modulus (UHM) and high treated for 75 W/3 min. This analysis was performed on a modulus(HM), and the PAN-based fibres as high tensile Leybold LHS 12 spectrometer. The spectral deconvolution strength(HT). The general description of the fibres used, was conducted using a DS 100 Leybold system as well as their code names, is presented in Table 1 The P120J fibres were plasma treated in a range of 3.2.2. Wettability measurements plasma power conditions(75 w,3 and 10 min, 100 W,3 The surface energy measurements were performed on min and 150 W,3 and 10 min). The ribbon and C320 fibres the carbon fibres using a CAHn Dynamic Contact Angle were treated at 75 W for 3 min. The plasma reactor was a Analyser, DCA- 322. The dispersive component of the Technics Plasma 200-G model, with a microwave power surface energy was measured using a-bromonaphthalene of 2.45 GHz. In all cases an oxygen pressure of as the non-polar liquid For the non-dispersive component, 00 Pa was used the octane/water system was adopted. The dispersi The polymer utilised in the present work was Makrolon component of the surface tension of both liquids, octane General description of the carbon fibres used pe Precursor material Cross-section geometry UHM Unsized. uncreate Proprietary treatment Ribbon, Clemson UHM Unsized untreated O, plasma/INCAR C320, Sigri Great Circular Unsized untreated Lakes O, plasma/INCAR
M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 1325 2805, a polycarbonate from Bayer. It is a standard grade 3 l ¯5 ]l . (7) for injection moulding, with a weight average molecular c 4 ¯ 21 weight (Mw) of 22 880 g mol and a polydispersivity Applying the equilibrium condition between the tensile index, measured as the ratio M¯ ¯ /M , equal to 1.86. The w n glass transition temperature, T , was equal to 145618C, as force acting on a fibre of diameter d, and the shear forces g measured by differential scanning calorimetry (DSC), on a transferred through the interface, a simple expression, proposed by Fraser and Di Benedetto [13], can be obtained Perkin-Elmer DSC7 instrument. The coefficient of thermal ¯ expansion, a , was measured on a Perkin-Elmer dynamic- for the average interfacial shear strength, t : PC mechanical analyser, DMA7, working on thermo-mechanid cal mode in expansion, in the temperature range of 90 to t ¯5 ]ss d l . (8) c 2.lc 1308C. For samples prepared with cooling rates of ¯68C/ 26 21 min, a was determined to be 97310 8C . The PC For a given fibre/matrix system, the smaller the fibre mechanical characterisation of Makrolon 2805, in tension, fragments obtained in the fragmentation test, the higher the was performed in an Instron 4505 testing machine, interfacial shear strength. equipped with a clip strain gauge extensometer. The results obtained for modulus, yield strength and strain were 2.1760.06 GPa, 65.960.5 MPa and 5.360.3%, respective- 3. Experimental ly. 3.1. Materials 3.2. Surface characterisation The carbon fibres studied in this work had different 3.2.1. XPS characteristics and were subject to plasma treatment in Two distinct analyses were performed. A global deseveral conditions. Three of them were pitch-based, two termination of the surface atomic composition, in terms of were obtained unsized and untreated, with different cross- overall carbon, oxygen and nitrogen content, was done section geometries, and the third was a commercial fibre, with an ESCALAB 200A-VG SCIENTIFIC spectrometer, treated and sized. For comparison, a PAN-based fibre, using a Mg/Al double X-ray source with a power of 300 unsized and untreated, was also studied. In terms of W. Then, a thorough spectral analysis was run for the mechanical classification, the pitch-based fibres can be pitch-based P120J and ribbon fibres, untreated and plasma considered as ultra-high modulus (UHM) and high treated for 75 W/3 min. This analysis was performed on a modulus (HM), and the PAN-based fibres as high tensile Leybold LHS 12 spectrometer. The spectral deconvolution strength (HT). The general description of the fibres used, was conducted using a DS100 Leybold system. as well as their code names, is presented in Table 1. The P120J fibres were plasma treated in a range of 3.2.2. Wettability measurements plasma power conditions (75 W, 3 and 10 min, 100 W, 3 The surface energy measurements were performed on min and 150 W, 3 and 10 min). The ribbon and C320 fibres the carbon fibres using a CAHN Dynamic Contact Angle were treated at 75 W for 3 min. The plasma reactor was a Analyser, DCA-322. The dispersive component of the Technics Plasma 200-G model, with a microwave power surface energy was measured using a-bromonaphthalene generator of 2.45 GHz. In all cases an oxygen pressure of as the non-polar liquid. For the non-dispersive component, 100 Pa was used. the octane/water system was adopted. The dispersive The polymer utilised in the present work was Makrolon component of the surface tension of both liquids, octane Table 1 General description of the carbon fibres used Designation and Type Precursor material Cross-section Treatment/origin producer geometry P120J, Amoco UHM Pitch Circular Unsized, untreated a O plasma/INCAR 2 P75S, Amoco HM Pitch Circular Proprietary treatment and sizing Ribbon, Clemson UHM Aromatic Ellipsoidal Unsized, untreated a University mesophase pitch O plasma/INCAR 2 C320, Sigri Great HT Polyacrylonitrile Circular Unsized, untreated a Lakes O plasma/INCAR 2 a Instituto Nacional del Carbon, Oviedo, Spain
1326 M C. Paira et al. Carbon 38(2000)1323-1337 and water, can be considered identical ( w=y), so that containing a single fibre aligned in the centre, with a the non-dispersive interactions may be estimated by a dog-bone shape, according to the Din 53504(S3A) simplified form of Eq. (6). Using the two-liquid phase standard tensiometric method, cos osw is calculated as the ratio Fragmentation tests, consisting of the stretching of the between the forces FHy and Fuw [ll], that represent the single filament composites, were made on an Instron 4505 force exerted on the fibre as it enters the first liquid at a speed of 0.5 mm/ min, until the matrix started yielding (hydrocarbon), and the second liquid (water), respectively: At this point, it was considered that the fragmentation of the matri W--YCos esw/H (9) was much higher than the tensile strain of the fibre in air (more than 10 The non-d contribution to the surface energy of measured using optical transmission microscopy. For the fibre P75S and P120J (untreated) fibres, a few tests were performed with a Minimat testing device(Polymer Labora- tory)coupled to a polarised light microscope. In this way, it was possible to monitor the fragmentation process by optical microscopy and simultaneous video recording The surface free energy values for the liquids used in the calculations are presented in Table 2. 4. Results and discussion 33. Micromechanical tests 4. 1. Surface characterisation 3.3. 1. Single filament tensile test 4. 1. Scanning electron microscopy The single filament tensile test is a technique widely Scanning electron microscopy (SEM) of the carbon sed to obtain tensile strength data on fibre form materials fibres was performed with a LEICA-$360 apparatus using The test method used here is described in the literature [8 a microanalysis system LINKexlll. Observation of the and was adapted from the AsTM standard surface revealed some morphology alterations induced by The diameter of the circular fibres was determined by a plasma with the type of laser diffraction technique, with a 10 mw He-Ne laser beam [14. The cross-sectional area of the non-circular SEM micrographs of the P120J fibres are presented in fibres was obtained after embedding their extremity in a Fig. 1. Although the P120J fibres were untreated and block of resin and polishing [8]. The measurements were unsized, a considerable amount of thin flakes was observed done using optical microscopy and image analysis. on the surface that may result from contamination during The tensile tests were performed in an Instron 1122 the production process. The flakes were not significantly equipped with a load beam of 5 N, at a crosshead speed of affected by plasma treatment at 75 w, but their con- centration decreased when the power increased. Only traces were present after the 100-w treatment and they 3.3. 2. Fragmentation tes were almost absent at 150 W. For the 75 W/3 min plasma The samples were processed by compression moulding treatment no significant alteration of the surface morpholo- the polymer in a hot press [9]. Two plates were prepared in gy, relative to the untreated fibres, was observed with the controlled conditions, and kept in a steel frame used as a amplification used Higher treatment levels produced etch- mould. Five to six carbon fibres were positioned straight ing lines along the fibre axis direction( deeper etching for across one of the polymer windows and glued to the metal stronger treatments). a decrease in fibre diameter was also frame. in both extremities. The two frames were assem- observed bled, and the set was compression moulded and cooled In The ribbon fibres behave differently this way, five to six carbon fibres aligned and positioned ment, at 75 W/3 and 10 min. The y ser aces ma ng mother, as can be observed in Fig. Is no net btained. Composite bars were cut from the plaque decrease in the fibre's dimension Surface free energy characteristics of the liquids Xw(mJ m a-Bromonaphthalene 44.6 Octane 21.6
1326 M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 d and water, can be considered identical (g W H ¯g ), so that containing a single fibre aligned in the centre, with a the non-dispersive interactions may be estimated by a ‘dog-bone’ shape, according to the DIN 53504 (S3A) simplified form of Eq. (6). Using the two-liquid phase standard. tensiometric method, cos u is calculated as the ratio Fragmentation tests, consisting of the stretching of the SW/H between the forces FHV HW and F [11], that represent the single filament composites, were made on an Instron 4505 force exerted on the fibre as it enters the first liquid at a speed of 0.5 mm/min, until the matrix started yielding. (hydrocarbon), and the second liquid (water), respectively: At this point, it was considered that the fragmentation process reached saturation, as the yield strain of the matrix FHW HW g ] was much higher than the tensile strain of the fibre in air ] 5 ]cos u . (9) SW/H FHV HV g (more than 10 times). The fragment lengths obtained were measured using optical transmission microscopy. For the The non-dispersive contribution to the surface energy of nd P75S and P120J (untreated) fibres, a few tests were the fibres, g S , was estimated using the geometric mean performed with a Minimat testing device (Polymer Labora- type relation: tory) coupled to a polarised light microscope. In this way, nd nd nd ]] it was possible to monitor the fragmentation process by W 5 2 g g . (10) SW œ S W optical microscopy and simultaneous video recording. The surface free energy values for the liquids used in the calculations are presented in Table 2. 4. Results and discussion 3.3. Micromechanical tests 4.1. Surface characterisation 3.3.1. Single filament tensile test 4.1.1. Scanning electron microscopy The single filament tensile test is a technique widely Scanning electron microscopy (SEM) of the carbon used to obtain tensile strength data on fibre form materials. fibres was performed with a LEICA-S360 apparatus using The test method used here is described in the literature [8] a microanalysis system LINKexLII. Observation of the and was adapted from the ASTM standard. surface revealed some morphology alterations induced by The diameter of the circular fibres was determined by a plasma treatment. These changes vary with the type of laser diffraction technique, with a 10 mW He–Ne laser fibre analysed. beam [14]. The cross-sectional area of the non-circular SEM micrographs of the P120J fibres are presented in fibres was obtained after embedding their extremity in a Fig. 1. Although the P120J fibres were untreated and block of resin and polishing [8]. The measurements were unsized, a considerable amount of thin flakes was observed done using optical microscopy and image analysis. on the surface that may result from contamination during The tensile tests were performed in an Instron 1122 the production process. The flakes were not significantly equipped with a load beam of 5 N, at a crosshead speed of affected by plasma treatment at 75 W, but their con- 0.5 mm/min. centration decreased when the power increased. Only traces were present after the 100-W treatment and they 3.3.2. Fragmentation test were almost absent at 150 W. For the 75 W/3 min plasma The samples were processed by compression moulding treatment no significant alteration of the surface morpholothe polymer in a hot press [9]. Two plates were prepared in gy, relative to the untreated fibres, was observed with the controlled conditions, and kept in a steel frame used as a amplification used. Higher treatment levels produced etchmould. Five to six carbon fibres were positioned straight ing lines along the fibre axis direction (deeper etching for across one of the polymer windows and glued to the metal stronger treatments). A decrease in fibre diameter was also frame, in both extremities. The two frames were assem- observed. bled, and the set was compression moulded and cooled. In The ribbon fibres behave differently after plasma treatthis way, five to six carbon fibres aligned and positioned ment, at 75 W/3 and 10 min. The surface becomes along the plaque length at middle thickness can be smoother, as can be observed in Fig. 2. There is no net obtained. Composite bars were cut from the plaques, decrease in the fibre’s dimensions. Table 2 Surface free energy characteristics of the liquids 22 d 22 22 Liquid g (mJ m ) g (mJ m ) g (mJ m ) L L LW a-Bromonaphthalene 44.6 44.6 – Octane 21.3 21.3 51.0 Water 72.6 21.6 –
M C. Paira et al. Carbon 38(2000)1323-1337 1327 Fig. 1. SEM micrographs of the P120J fibres: (a)untreated;(b)75 W/3 min;(c)75 W/10 min;(d)100 w/3 min;(e)150 w/3 min; (f)150 W/10 min No change was observed for the PAN-based fibres after studied are presented in Table 3. Carbon and oxygen are plasma treatment, at the level of amplification used in the the most abundant elements at the fibres'surface. Other observations(Fig. 3) elements may be present, but only at atomic concentrations lower than 0. 1%, except for the PAN-based fibres, that 4.1.2.XPS contain a considerable amount of nitrogen. The global surface atomic compositions for all fibres Plasma treatment significantly increased the net surface
M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 1327 Fig. 1. SEM micrographs of the P120J fibres: (a) untreated; (b) 75 W/3 min; (c) 75 W/10 min; (d) 100 W/3 min; (e) 150 W/3 min; (f) 150 W/10 min. No change was observed for the PAN-based fibres after studied are presented in Table 3. Carbon and oxygen are plasma treatment, at the level of amplification used in the the most abundant elements at the fibres’ surface. Other observations (Fig. 3). elements may be present, but only at atomic concentrations lower than 0.1%, except for the PAN-based fibres, that 4.1.2. XPS contain a considerable amount of nitrogen. The global surface atomic compositions for all fibres Plasma treatment significantly increased the net surface
M C. Paira et al. Carbon 38(2000)1323-1337 Fi.2. icrographs of the ribbon fibres:(a) untreated; (b)75 W/3 min;(c)75 W/10 min oxygen concentration. The fibres are extensively oxidised 4.1.3. Wettability fter 3 min, even at the lower plasma power The results obtained for the dispersive and non-disper Detailed analysis of the P120J and ribbon fibres, before sive components of the surface energy are presented and after plasma treatment at 75 w for 3 min, revealed the Table 5 type of functionalities present on the fibres surface. The The dispersive component remains considerably con- bands obtained by spectral deconvolution of the Cls peak stant for all the fibres studied, showing no appreciable re shown in Fig. 4, and the functionalities attributed to variation after plasma treatment. In the limit, it could be each are presented in Table 4. The same functionalities considered that a slight tendency to an increase in ys is were detected for both types of fibres, namely hydroxyl observed from the untreated to the plasma treated fibres ether groups, quinone type groups and carboxylic acid, in The magnitude of the results is comparable to that order of decreasing concentration. The relative composi- in the literature for other types of fibres [15]. The non- tion, expressed as the ratio of each peak area to the total dispersive contribution to the surface energy is very small )) is represented in Fig. 5 A crease is. however observed from untreated The hydroxyl or ether groups are present in a higher to treated fibres. The ribbon fibres also show the same concentration in the untreated fibres. Plasma treatment behaviour. Higher values were obtained for the P75s, but introduces more quinone and carboxylic acid groups, this probably reflects interactions with the fibre sizing. The relative to the hydroxyl groups, but the concentration of results found in the literature for other pitch-based fibres. the latter is still dominant. The oxygen uptake is, in although few and obtained with different wetting tech- general, higher for the ribbon fibres for all contributing niques are comparable with those determined in this work. functionalitie No data were obtained for the pan-based fibres due to
1328 M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 Fig. 2. SEM micrographs of the ribbon fibres: (a) untreated; (b) 75 W/3 min; (c) 75 W/10 min. oxygen concentration. The fibres are extensively oxidised 4.1.3. Wettability after 3 min, even at the lower plasma power. The results obtained for the dispersive and non-disperDetailed analysis of the P120J and ribbon fibres, before sive components of the surface energy are presented in and after plasma treatment at 75 W for 3 min, revealed the Table 5. type of functionalities present on the fibres’ surface. The The dispersive component remains considerably conbands obtained by spectral deconvolution of the C1s peak stant for all the fibres studied, showing no appreciable are shown in Fig. 4, and the functionalities attributed to variation after plasma treatment. In the limit, it could be d each are presented in Table 4. The same functionalities considered that a slight tendency to an increase in g is S were detected for both types of fibres, namely hydroxyl or observed from the untreated to the plasma treated fibres. ether groups, quinone type groups and carboxylic acid, in The magnitude of the results is comparable to that reported order of decreasing concentration. The relative composi- in the literature for other types of fibres [15]. The nontion, expressed as the ratio of each peak area to the total dispersive contribution to the surface energy is very small. peak area (A /A ), is represented in Fig. 5. A significant increase is, however, observed from untreated C1s(i) C1s(total) The hydroxyl or ether groups are present in a higher to treated fibres. The ribbon fibres also show the same concentration in the untreated fibres. Plasma treatment behaviour. Higher values were obtained for the P75S, but introduces more quinone and carboxylic acid groups, this probably reflects interactions with the fibre sizing. The relative to the hydroxyl groups, but the concentration of results found in the literature for other pitch-based fibres, the latter is still dominant. The oxygen uptake is, in although few and obtained with different wetting techgeneral, higher for the ribbon fibres for all contributing niques are comparable with those determined in this work. functionalities. No data were obtained for the PAN-based fibres, due to
M C. Paira et al. Carbon 38(2000)1323-1337 Fig. 3. SEM micrographs of the PAN-based fibres: (a) untreated; (b)75 W/3 min,(c)75 w/10 min. the literature for similar systems indicate ys and y values around 50 and 7 mJ respectively [16] Table 3 mined by XPS(at % 4.2. Micromechanical effects P120J Untreated 97.302.70 75W3min90.85915-10.1 4.2. 1. Single filament tensile testing 75W/0min884411.56 The mechanical data for the circular and the ribbon 73 fibres are presented in Tables 6 and 7, respectively. The 150W3min91968.04 differences in cross sections and dimensions are shown in 150W/10min914l8.59 Fig. 6. The Youngs modulus of the ultra-high modulus Ribbon Untreated 96.823.18 fibres could not be determined accurately. In fact, as the 75W/3min89.171083-12.2 compliance of the measuring system is of the same order 75W/10min899010.10 of magnitude of that of the fibres, a high degree of C320 92635.921.4464 uncertainty is introduced in the calculations. For this 75W/3min91.047291678.0 eason, only a lower limit can be presented. This limit is 5S 78.218.82.9824.0 730 GPa for the untreated P120J. 780 GPa for the W/3-min-treated P120J, higher than 800 GPa for the
M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 1329 Fig. 3. SEM micrographs of the PAN-based fibres: (a) untreated; (b) 75 W/3 min; (c) 75 W/10 min. experimental difficulties. However, the results described in d nd the literature for similar systems indicate g S S and g 22 values around 50 and 7 mJ m , respectively [16]. Table 3 Global surface composition as determined by XPS (at.%) 4.2. Micromechanical effects Sample C O N O/C (%) P120J Untreated 97.30 2.70 – 2.8 75 W/3 min 90.85 9.15 – 10.1 4.2.1. Single filament tensile testing 75 W/10 min 88.44 11.56 – 13.1 The mechanical data for the circular and the ribbon 100 W/3 min 91.19 8.81 – 9.7 fibres are presented in Tables 6 and 7, respectively. The 150 W/3 min 91.96 8.04 – 8.7 differences in cross sections and dimensions are shown in 150 W/10 min 91.41 8.59 – 9.4 Fig. 6. The Young’s modulus of the ultra-high modulus Ribbon Untreated 96.82 3.18 – 3.3 fibres could not be determined accurately. In fact, as the 75 W/3 min 89.17 10.83 – 12.2 compliance of the measuring system is of the same order 75 W/10 min 89.90 10.10 – 11.2 of magnitude of that of the fibres, a high degree of uncertainty is introduced in the calculations. For this C320 Untreated 92.63 5.92 1.44 6.4 75 W/3 min 91.04 7.29 1.67 8.0 reason, only a lower limit can be presented. This limit is 730 GPa for the untreated P120J, 780 GPa for the 75 P75S 78.2 18.8 2.98 24.0 W/3-min-treated P120J, higher than 800 GPa for the
M C. Paira et al. Carbon 38(2000)1323-1337 2200 0000 Pl Sat2 Sat1 5 970 160 4000 2000 PI Sat2 Sat1 966 sG日 tev Fig 4. CIs peak of the P120J fibres with curve deconvolution and window for Ols peak;(a)untreated;(b)75 w,3 min treated fibres
1330 M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 Fig. 4. C1s peak of the P120J fibres with curve deconvolution and window for O1s peak; (a) untreated; (b) 75 W, 3 min treated fibres
M C. Paira et al. Carbon 38(2000)1323-1337 1331 Table 4 CIs peak components and functionality attributions BE (ev) BE shift(ev) Attributions CIs(1) 284.10±0.10 Csp and Csp"in the carbon CIs(2 285.20±0.10 1.10±0.10 Aliphatic C, no functionality contaminants adsorbed) CIs (3) 1.90±0.10 C from C-oh and C-O-C CIs(4 ±0.10 2.90± L= CIs (5) 4.30± 289.50±0.20 satellite(→ 290.40+0.20 6.30±0.20 satellite(丌→丌) 7.00±0.30 园Cls(3) ■Cl(4) 15 日Cls(5 0.5 0 P120J P 120JOx Ribbon RibbonOx Fig. 5. Relative osition of the fibres surface, in terms of hydroxyl or ether groups(CIs(3 ) quinone type groups(Cls(4))and carboxylic acid(CIs(5), expressed as the ratio of each peak area to the total peak area ribbon fibres, treated and untreated, 420 GPa for the P75s tal results of strength at any gauge length are presented in and 230 GPa for the pAN-based C320 Table 8. These parameters allow the estimation of fibre The estimated Weibull parameters that fit the experimen- strength at small lengths, for each fibre type, using Eq. (3) Values of the dispersive and non-dispersive components of the surface energy P120J 10±0.2 75 W/3 min 31±2 2.4±0 75 W/10 min 32±2 33±1 54±0.3 00 W/3 min 34±2 50 W/3 min 33± 23.5±3 50 W/10 min 2.7+0 5S 41.6 39±10 7.5±4 Ribbon 7+1 02±0.1 75 W/3 min 30.4±0.2 2.5±0.4 75 W/10 min 2.4±0.4 represents the standard deviation
M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 1331 Table 4 C1s peak components and functionality attributions Component BE (eV) BE shift (eV) Attributions 2 3 C1s (1) 284.1060.10 – Csp and Csp in the carbon fibre structure C1s (2) 285.2060.10 1.1060.10 Aliphatic C, no functionality (contaminants adsorbed) C1s (3) 286.0060.10 1.9060.10 C from C–OH and C–O–C C1s (4) 287.0060.10 2.9060.10 aC=O, quinones C1s (5) 288.4060.10 4.3060.10 –COOH Sat.1 289.5060.20 5.4060.20 Shake-up satellite (p→p*) Sat.2 290.4060.20 6.3060.20 Shake-up satellite (p→p*) Pl. 291.1060.30 7.0060.30 Plasmon Fig. 5. Relative composition of the fibres’ surface, in terms of hydroxyl or ether groups (C1s(3)), quinone type groups (C1s(4)) and carboxylic acid (C1s(5)), expressed as the ratio of each peak area to the total peak area. ribbon fibres, treated and untreated, 420 GPa for the P75S tal results of strength at any gauge length are presented in and 230 GPa for the PAN-based C320. Table 8. These parameters allow the estimation of fibre The estimated Weibull parameters that fit the experimen- strength at small lengths, for each fibre type, using Eq. (3). Table 5 a Values of the dispersive and non-dispersive components of the surface energy d 22 nd 22 nd 22 Sample g (mJ m ) W (mJ m ) g (mJm ) S SW S P120J Untreated 2863 1462 1.060.2 75 W/3 min 3162 2362 2.460.5 75 W/10 min 3262 3361 5.460.3 100 W/3 min 3462 2762 3.660.4 150 W/3 min 3361 23.563 2.860.5 150 W/10 min 3364 2361 2.760.3 P75S Treated and sized 41.6 39610 7.564 Ribbon Untreated 3364 761 0.260.1 75 W/3 min 30.460.2 2362 2.560.4 75 W/10 min 3661 2262 2.460.4 a 6 represents the standard deviation
1332 M C. Paira et al. Carbon 38(2000)1323-1337 Single filament tensile test data for the circular fibres No. fibres Diameter(um) Tensile strength(GPa) ±S.D 95% confidence interval 2.01±0.18 9.6±1.3 1.96±0.14 P120J(MDS) P120J 9.2±1.2 02 P120J 75 W/10 min 8.9±1.1 P120J 1.97±0.26 100 W/3 min 9.3±1.3 P120J 18 1.93±0.30 150 W/3 min 8.8±1.4 1.66±0.23 P120J 1.98±0.26 150w/10m 8.2±1.2 2.04±0.19 00±0.39 P75S 18 1.94±0.16 11.0±1.0 1.63±0.09 1.53±0.13 1.38±0.16 2.1 C320 32 2.67±0.22 2.86±0.18 C320(MDS) 8±0.5 2.91±0.20 2.90±0.22 2.59±0.16 Measured on impregnated strand tensile test data for the ribbon fibres No. fibres Cross section Tensile strength(GPa) area(um) 95% confidence interval Ribbon 2.73±0.23 2.26±0. 2.09±0.25 2.44±0.1 Ribbon 2.39±0.22 75 W/3 min 34±36 2.21±0.21 2.41±0.2
1332 M.C. Paiva et al. / Carbon 38 (2000) 1323 –1337 Table 6 Single filament tensile test data for the circular fibres Sample No. fibres Diameter (mm) Gauge Tensile strength (GPa) tested 6S.D. length (mm) 695% confidence interval P120J 38 6 2.0160.18 61 9.661.3 10 1.9660.14 22 16 1.8660.22 43 21 1.9660.20 a b P120J (MDS) – 10 – 2.37 P120J 25 10 1.9760.29 75 W/3 min 30 9.261.2 15 1.9160.22 28 20 1.6260.21 P120J 23 10 2.2060.23 75 W/10 min 24 8.961.1 15 1.9760.22 23 20 1.8760.23 P120J 24 10 1.9760.26 100 W/3 min 24 9.361.3 15 1.8660.23 23 30 1.5360.23 P120J 18 10 1.9360.30 150 W/3 min 21 8.861.4 15 1.7060.19 25 21 1.6660.23 22 40 1.4960.19 P120J 24 10 1.9860.26 150 W/10 min 28 8.261.2 20 2.0460.19 21 41 2.0060.39 P75S 18 10 1.9460.16 23 11.061.0 21 1.6360.09 20 41 1.5360.13 19 80 1.3860.16 b P75S (MDS) – 10 – 2.1 C320 31 20 2.8860.23 32 7.860.5 30 2.6760.22 31 40 2.8660.18 C320 (MDS) – 7 – 3 C320 19 15 2.9460.18 75 W/3 min 19 7.860.5 20 2.9160.20 20 30 2.9060.22 21 41 2.5960.16 a MDS, Manufacturer’s Data Sheet. b Measured on impregnated strand. Table 7 Single filament tensile test data for the ribbon fibres Sample No. fibres Cross section Gauge length Tensile strength (GPa) 2 tested area (mm ) (mm) 695% confidence interval Ribbon 40 10 2.7360.23 45 320642 25 2.2660.22 41 35 2.0960.25 44 45 2.4460.18 Ribbon 26 10 2.3960.22 75 W/3 min 27 334636 15 2.2160.21 30 30 2.4160.23