CARBON PERGAMON Carbon4l(2003)1193-1203 Effect of carbon fiber surface functional groups on the mechanical properties of carbon-carbon composites with HTT SR Dhakate.OP Bahl Carbon Technology Unit, Engineering Materials Division, National Physical Laboratory, Dr. K.S. Krishnan Marg, New Delhi 110012, Received 18 March 2001; accepted 25 January 2003 Abstract The present investigation describes the itative measurement of surface functional groups present on commerciall available different pan based carbon fibers, their effect on the development of interface with resol-type phenol stages of heat treatment. An ESCa study of the carbon fibers has revealed that high strength(ST-3 )carbon fibers posse formaldehyde resin matrix and its effect on the physico-mechanical properties of carbon-carbon composites at variou almost 10% reactive functional groups as compared to 5.5 and 4.5% in case of intermediate modulus(IM-500)and high modulus(HM-45)carbon fibers, respectively. As a result, ST-3 carbon fibers are in a position to make strong interactions with phenolic resin matrix and HM-45 carbon fibers make weak interactions, while IM-500 carbon fibers make intermediate interactions. This observation is also confirmed from the pyrolysis data(volume shrinkage)of the composites. Bulk density and kerosene density more or less increase in all the composites with heat treatment up to 2600C. It is further observed that bulk density is minimum and kerosene density is maximum upon heat treatment at 2600C in case of sT-3 based composites compared to HM-45 and IM-500 composites. It has been found for the first time that the defection temperature(temperature at which the properties of the material start to decrease or increase)of fiexural strength as well as interlaminar shear strength is different for the three composites(A, b and C)and is determined by the severity of interactions established at the polymer stage. Above this temperature, flexural strength and interlaminar shear strength increase in all the composites up to 2600C he maximum value of fiexural strength at 2600C is obtained for HM-45 composites and that of Ilss for ST-3 composites o 2003 Published by Elsevier Science Ltd Keywords: A. Carbon fibers; C. Electron spectroscopy; D. Functional groups, Mechanical properties 1. Introduction in carbon-carbon composites between fiber and matrix, within fiber bundles between the different microstructures Carbon-carbon composites are used in a number of which may exist within the matrix. It is well known that demanding applications such as space, defense, turbine the fiber-matrix interaction depends upon the surface blades, etc. [1, 2]. Their performance is known to depen functional groups of the carbon fibers and the matrix on the type of carbon fibers, matrix precursors, nature of precursor 3, 4]. In particular, the mechanical properties of bonding between fiber and matrix(fiber-matrix interface) carbon-carbon composites are very sensitive to the bond- and processing conditions [3, 4]. The nature of the interface ing between fibers and the matrix and its stress transfer and its influence on physico-mechanical properties is capability. In this respect, continuous research and de- extremely complex. Different types of interface may exist velopment work is going on from the last 3-4 decades, and especially regarding the role of interface on controlling the overall performance of carbon-carbon composites [5-10] Corresponding author. Tel:+91-11-574-6290 fax:+91-11- A qualitative correlation between the amount of 572-6952 functional groups, the nature of the interface and E-mail address. dhakate @csnpl ren nic, in(SR. Dhakate) site properties has been reported by Fitzer et al. [7 0008-6223/03/S-see front matter 2003 Published by Elsevier Science Ltd doi:10.1016/S0008-6223(03)00051-4
Carbon 41 (2003) 1193–1203 E ffect of carbon fiber surface functional groups on the mechanical properties of carbon–carbon composites with HTT S.R. Dhakate , O.P. Bahl * Carbon Technology Unit, Engineering Materials Division, National Physical Laboratory, Dr. K.S. Krishnan Marg, New Delhi 110012, India Received 18 March 2001; accepted 25 January 2003 Abstract The present investigation describes the quantitative measurement of surface functional groups present on commercially available different PAN based carbon fibers, their effect on the development of interface with resol-type phenol formaldehyde resin matrix and its effect on the physico–mechanical properties of carbon–carbon composites at various stages of heat treatment. An ESCA study of the carbon fibers has revealed that high strength (ST-3) carbon fibers possess almost 10% reactive functional groups as compared to 5.5 and 4.5% in case of intermediate modulus (IM-500) and high modulus (HM-45) carbon fibers, respectively. As a result, ST-3 carbon fibers are in a position to make strong interactions with phenolic resin matrix and HM-45 carbon fibers make weak interactions, while IM-500 carbon fibers make intermediate interactions. This observation is also confirmed from the pyrolysis data (volume shrinkage) of the composites. Bulk density and kerosene density more or less increase in all the composites with heat treatment up to 2600 8C. It is further observed that bulk density is minimum and kerosene density is maximum upon heat treatment at 2600 8C in case of ST-3 based composites compared to HM-45 and IM-500 composites. It has been found for the first time that the deflection temperature (temperature at which the properties of the material start to decrease or increase) of flexural strength as well as interlaminar shear strength is different for the three composites (A, B and C) and is determined by the severity of interactions established at the polymer stage. Above this temperature, flexural strength and interlaminar shear strength increase in all the composites up to 2600 8C. The maximum value of flexural strength at 2600 8C is obtained for HM-45 composites and that of ILSS for ST-3 composites. 2003 Published by Elsevier Science Ltd. Keywords: A. Carbon fibers; C. Electron spectroscopy; D. Functional groups, Mechanical properties 1. Introduction in carbon–carbon composites between fiber and matrix, within fiber bundles, between the different microstructures Carbon–carbon composites are used in a number of which may exist within the matrix. It is well known that demanding applications such as space, defense, turbine the fiber–matrix interaction depends upon the surface blades, etc. [1,2]. Their performance is known to depend functional groups of the carbon fibers and the matrix on the type of carbon fibers, matrix precursors, nature of precursor [3,4]. In particular, the mechanical properties of bonding between fiber and matrix (fiber–matrix interface) carbon–carbon composites are very sensitive to the bondand processing conditions [3,4]. The nature of the interface ing between fibers and the matrix and its stress transfer and its influence on physico–mechanical properties is capability. In this respect, continuous research and deextremely complex. Different types of interface may exist velopment work is going on from the last 3–4 decades, and especially regarding the role of interface on controlling the overall performance of carbon–carbon composites [5–10]. *Corresponding author. Tel.: 191-11-574-6290; fax: 191-11- A qualitative correlation between the amount of surface 572-6952. functional groups, the nature of the interface and compoE-mail address: dhakate@csnpl.ren.nic.in (S.R. Dhakate). site properties has been reported by Fitzer et al. [7]. Also, 0008-6223/03/$ – see front matter 2003 Published by Elsevier Science Ltd. doi:10.1016/S0008-6223(03)00051-4
1194 S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 Table I Properties of carbon fibers used (from manufacturer's data sheet) Diameter Density Tensile strength Tensile modulus Strain to failure (GPa) ST-3 6.0 IM-500 5 HM-45 191 2200 0.5 there are plenty of data available on the analysis of surface The composites were coded as follows. functional groups of carbon fibers and their influence on the development of interface and mechanical properties of lymer matrix composites. In addition to this, some data (A)ST-3 carbon fiber composites are available in the literature describing the effect of heat (B)IM-500 carbon fiber composites. treatment temperature(HTT)on the mechanical properties (C)HM-45 carbon fiber composites lyarylacetylene(PAA)and furfuryl alcohol based arbon-carbon composites [11-13]. Some studies are also available on the influence of fiber surface functional The polymer composites were heat-treated to 400- 2600C under an inert atmosphere. The heat-treated groups and various types of surface treated carbon fibers composites were characterized for volume shrinkage on the development of interface with phenolic resin matrix density and mechanical properties during each stage of [7,14,15 heat treatment. Flexural strength was measured by the In the present investigation, a systematic approach was three-point bending technique on a universal Instron adopted to understand the influence of carbon fiber surface testing machine(Model 4411, ASTM standard D-790-80) functional groups on interface development by measuring with a span length to depth ratio of 30: 1. The interlaminar the surface functional groups quantitatively and their shear strength(ILSS)was measured using ASTM standard D-2344-74 with span length to depth ratio of 8: 1. The kerosene density was measured by the kerosene pickup method using the Archimedes principle. The transverse 2. Composite preparation and characterization coefficients of thermal expansion were measured using a thermo-mechanical analyzer(TMA)attached to a Mettler Three types of commercially available PAN-based car hermal system TA-3000, in the range 50-900C under an bon fibers, manufactured by Toho Beslon Inc, and ert atmosphere. The optical micrographs of composites orayca Industries Inc, Japan, were used as reinforcement. heat treated at 2600C were observed using polarized light (a) ST-3(high strength, likely HTT-1200-1500C) (b)IM-500(intermediate modulus, likely HTT 2200°℃C) 3. Results and discussion Since these fibers are heat-treated to different tempera. 3. 1. ESCA studies of carbon fibers tures, they exhibit different physical and mechanical properties (see Table 1)as well as different types of The surface composition of carbon fibers obtained by surface functional groups. The surface functional groups present were measured by ESCA, using an SSI 301 Ko radiation(spot diameter 300 um, 80 W; radiation able energy 14866 ev)under a residual pressure of 5x10-8 Surface composition bon fibers Torr (I Torr=133.322 Pa) Fiber type nidirectional polymer composite samples(150 mmx element N(1s) 4.0 mmX4.5 mm) were prepared using the wet winding and match mold die technique [16] with 45+2% fiber T-3 volume. The resol type phenol formaldchyde resin was HM.-45 9223 used as the matrix precursor for composite preparation. HM-4
1194 S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 Table 1 Properties of carbon fibers used (from manufacturer’s data sheet) Fiber Diameter Density Tensile strength Tensile modulus Strain to failure 3 type (mm) (g/cm ) (MPa) (GPa) (%) ST-3 6.0 1.75 4200 250 1.68 IM-500 5.5 1.77 4700 300 1.57 HM-45 6.0 1.91 2200 440 0.5 there are plenty of data available on the analysis of surface The composites were coded as follows. functional groups of carbon fibers and their influence on the development of interface and mechanical properties of (A) ST-3 carbon fiber composites. polymer matrix composites. In addition to this, some data (B) IM-500 carbon fiber composites. are available in the literature describing the effect of heat (C) HM-45 carbon fiber composites. treatment temperature (HTT) on the mechanical properties of polyarylacetylene (PAA) and furfuryl alcohol based carbon–carbon composites [11–13]. Some studies are also The polymer composites were heat-treated to 400– available on the influence of fiber surface functional 2600 8C under an inert atmosphere. The heat-treated groups and various types of surface treated carbon fibers composites were characterized for volume shrinkage, on the development of interface with phenolic resin matrix density and mechanical properties during each stage of [7,14,15]. heat treatment. Flexural strength was measured by the In the present investigation, a systematic approach was three-point bending technique on a universal Instron adopted to understand the influence of carbon fiber surface testing machine (Model 4411, ASTM standard D-790-80) functional groups on interface development by measuring with a span length to depth ratio of 30:1. The interlaminar the surface functional groups quantitatively and their shear strength (ILSS) was measured using ASTM standard influence on composite properties as a function of HTT. D-2344-74 with span length to depth ratio of 8:1. The kerosene density was measured by the kerosene pickup method using the Archimedes principle. The transverse 2. Composite preparation and characterization coefficients of thermal expansion were measured using a thermo-mechanical analyzer (TMA) attached to a Mettler Three types of commercially available PAN-based car- thermal system TA-3000, in the range 50–900 8C under an bon fibers, manufactured by Toho Beslon Inc., and inert atmosphere. The optical micrographs of composites Torayca Industries Inc., Japan, were used as reinforcement. heat treated at 2600 8C were observed using polarized light microscopy. (a) ST-3 (high strength, likely HTT |1200–1500 8C). (b) IM-500 (intermediate modulus, likely HTT |1600– 1800 8C). (c) HM-45 (high modulus, likely HTT .2200 8C) 3. Results and discussion Since these fibers are heat-treated to different tempera- 3 .1. ESCA studies of carbon fibers tures, they exhibit different physical and mechanical properties (see Table 1) as well as different types of The surface composition of carbon fibers obtained by surface functional groups. The surface functional groups present were measured by ESCA, using an SSI 301 spectrometer employing monochromatic and focused Al K Table 2 a radiation (spot diameter 300 mm, 80 W; radiation28 Surface composition of carbon fibers energy 1486.6 eV) under a residual pressure of 5310 Torr (1 Torr5133.322 Pa). Fiber type Atomic % Unidirectional polymer composite samples (150 mm3 \element C (1s) O (1s) N (1s) Si (2p) 4.0 mm34.5 mm) were prepared using the wet winding ST-3 86.63 8.19 2.35 1.86 and match mold die technique [16] with 4562% fiber IM-500 92.28 7.23 0.00 0.49 volume. The resol type phenol formaldehyde resin was HM-45 92.21 6.18 0.00 1.60 used as the matrix precursor for composite preparation
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 220 294 2892286.82844282 2916 2844 B80 660 2868 BINDING ENERGY A Fig 1.(a) ESCA spectra of (a)ST-3, (b) IM-500 and (c) HM-45 carbon fibers (b) Deconvolution spectra of o Is for (a)ST-3, (b)IM-500 and(c)HM-45 carbon fibers. Relative percentage of functional groups on the carbon fibers obtained by ESCA Functional group (28418) (284.05,28483) Phenolic or hydroxyl (B. E, ev) Carboxylic B. E. eV (290.48) (290.92) (B. E, ev) (39802,400.43)
S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 1195 Fig. 1. (a) ESCA spectra of (a) ST-3, (b) IM-500 and (c) HM-45 carbon fibers. (b) Deconvolution spectra of O 1s for (a) ST-3, (b) IM-500 and (c) HM-45 carbon fibers. Table 3 Relative percentage of functional groups on the carbon fibers obtained by ESCA Functional group Relative atomic percentages of functional groups ST-3 IM-500 HM-45 Graphitic carbon 69.9 61.9 90.88 (B. E., eV) (284.18) (284.06) (284.05, 284.83) Phenolic or hydroxyl 18.76 28.6 – (B. E., eV) (285.42) (285.08) Carbonyl 7.01 5.44 4.3 (B. E., eV) (287.34) (287.90) (288.49) Carboxylic 4.31 4.05 4.8 (B. E., eV) (290.13) (290.48) (290.92) Nitrogen containing 2.35 – – (B. E., eV) (398.02, 400.43)
1196 S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 the ESCa studies is presented in Table 2. As expected 3.2. Interactions of fibers with matrix [17 the oxygen content on the surface decreases from ST-3 to HM-45 and the C/o ratio increases from St-3 The specific interactions are postulated to be Lewis HM-45, confirming the well known fact that the oxygen- acid-base type interactions or electron acceptor-donor ontaining functional groups, and consequently the reac- interactions [30, 31]. The polymer used here is a thermoset- tivity of carbon fiber surface, decrease with increasing ting phenolic resin which has acidic functional groups HTT. The nitrogen-containing functional groups are pres- whereas fibers possess both acidic and basic functional ent only in ST-3 fibers, from the residual nitrogen present groups. It is well known that carboxylic groups are in the precursor fibers responsible for strong interactions with a polymer matrix Fig. la shows the ESCA spectra of the three carbon having basic functional groups [ 18, 32]. Therefore, in the fibers. The three major peaks observed between binding present case these groups are not likely to play an energies 200 to 600 ev of carbon(C Is), oxygen(o Is) important role. It is found that ST-3 fibers possess maxi and nitrogen(N Is)correspond to graphite-like carbon and mum(9.46 relative percentage) functional groups which various functional groups. The peak around 101 ev are basic in character and would make strong interactions corresponds to silicon Si(2p). The C Is peak at 284. 24 ev whereas HM-45 fibers possess minimum reactive func- assigned to the carbon element only and between 284.05 tional groups (4.3 relative percentage) and would thus and 284.83 ev to graphitic carbon, 285.08-285.42 eV to make weak interaction with the phenolic resin matrix On hydroxyl (C-OH), 287.34-288.49 ev to carbonyl the other hand, the IM-500 fibers possess a maximum (C=O)and 290.13-290.92 ev to carboxylic(COoH) amount of total functional groups(38.09 relative per- functional groups of three different carbon fibers is differ- relative percentage (of carbonyl), which can make interac- ent(Fig. 1b, Table 3), which may be due to the structural ions with phenolic resin matrix intermediate between ST-3 difference of the carbon fibers surfaces. The fibers used in and HM-45 carbon fibers [ 33] the investigation are heat-treated at increasing temperature ST-3 to HM-45) as described in Section 2. As a conse- quence, orientation of carbon layers preferentially im- 3.3. Volume shrinkage vs HTT of composites proves parallel to fiber axis leading to different degrees of graphitization and therefore different electrical conduc- During heat treatment, carbon fiber reinforced polymer tivities of the fibers. The ESCA spectrum for the ST-3 opposites show changes in all the three directions of carbon fibers shows a very weak n Is peak as compared to composites, i. e, length, width and thickness, which are due the c Is and o Is peaks, which arises from the residual to the thermal degradation and shrinkage of polymer nitrogen present in the fibers. Nitrogen-containing func matrix [7, 34]. In unidirectional composites, the changes tional groups appear in ST-3 fibers at a binding energy of taking place in the direction parallel to the fiber axis are 398.02 ev, corresponding to the aromatic amines and controlled by the longitudinal thermal expansion of the piperidine structure, at 400. 43 ev to aliphatic amines, carbon fibers themselves. The fiber thermal expansion nitrile and amides [21, 22], while in the cases of IM-500(positive or negative) is very small [35]. As a consequence. and HM-45 carbon fibers these groups are not detected Deconvolution of these peaks gives relative percentages of functional groups present and these are compiled in Table The high strength carbon fibers (ST-3) possess the highest number of surface functional groups and as a result the surface of these fibers may be more disordered possessing comparatively high active surface area. The ST-3 fibers possess carboxylic, phenolic and hydroxyl 25 functional groups which are acidic in nature while car- 20 bonyl and some of nitrogen containing groups are basic in character [23-29]. The high modulus HM-45 carbon fiber, on the other hand, possesses a minimum number of surface functional groups and a better ordered graphite-like carbon fiber surface. Therefore such fibers should have the lowest active surface area. The intermediate modulus(IM-500) carbon fibers possess the greatest number of hydroxyl 04008001200 200024002800 functional groups. The contribution of carboxylic groups is HTT(C) almost the same in all the three fibers while carbony olume shrinkage observed with heat treatment tempera- groups decrease from ST-3 to HM-45 fibers ture of composite
1196 S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 the ESCA studies is presented in Table 2. As expected 3 .2. Interactions of fibers with matrix [17], the oxygen content on the surface decreases from ST-3 to HM-45 and the C/O ratio increases from ST-3 to The specific interactions are postulated to be Lewis HM-45, confirming the well known fact that the oxygen- acid–base type interactions or electron acceptor–donor containing functional groups, and consequently the reac- interactions [30,31]. The polymer used here is a thermosettivity of carbon fiber surface, decrease with increasing ting phenolic resin which has acidic functional groups HTT. The nitrogen-containing functional groups are pres- whereas fibers possess both acidic and basic functional ent only in ST-3 fibers, from the residual nitrogen present groups. It is well known that carboxylic groups are in the precursor fibers. responsible for strong interactions with a polymer matrix Fig. 1a shows the ESCA spectra of the three carbon having basic functional groups [18,32]. Therefore, in the fibers. The three major peaks observed between binding present case these groups are not likely to play an energies 200 to 600 eV of carbon (C 1s), oxygen (O 1s) important role. It is found that ST-3 fibers possess maxiand nitrogen (N 1s) correspond to graphite-like carbon and mum (9.46 relative percentage) functional groups which various functional groups. The peak around 101 eV are basic in character and would make strong interactions, corresponds to silicon Si (2p). The C 1s peak at 284.24 eV whereas HM-45 fibers possess minimum reactive funcis assigned to the carbon element only and between 284.05 tional groups (4.3 relative percentage) and would thus and 284.83 eV to graphitic carbon, 285.08–285.42 eV to make weak interaction with the phenolic resin matrix. On hydroxyl (.C–OH), 287.34–288.49 eV to carbonyl the other hand, the IM-500 fibers possess a maximum (.C=O) and 290.13–290.92 eV to carboxylic (–COOH) amount of total functional groups (38.09 relative perfunctional groups [18–20]. The chemical shift for all the centage) but the reactive functional groups are only 5.44 functional groups of three different carbon fibers is differ- relative percentage (of carbonyl), which can make interacent (Fig. 1b, Table 3), which may be due to the structural tions with phenolic resin matrix intermediate between ST-3 difference of the carbon fibers surfaces. The fibers used in and HM-45 carbon fibers [33]. the investigation are heat-treated at increasing temperature (ST-3 to HM-45) as described in Section 2. As a consequence, orientation of carbon layers preferentially im- 3 .3. Volume shrinkage vs. HTT of composites proves parallel to fiber axis leading to different degrees of graphitization and therefore different electrical conduc- During heat treatment, carbon fiber reinforced polymer tivities of the fibers. The ESCA spectrum for the ST-3 composites show changes in all the three directions of carbon fibers shows a very weak N 1s peak as compared to composites, i.e., length, width and thickness, which are due the C 1s and O 1s peaks, which arises from the residual to the thermal degradation and shrinkage of polymer nitrogen present in the fibers. Nitrogen-containing func- matrix [7,34]. In unidirectional composites, the changes tional groups appear in ST-3 fibers at a binding energy of taking place in the direction parallel to the fiber axis are 398.02 eV, corresponding to the aromatic amines and controlled by the longitudinal thermal expansion of the piperidine structure, at 400.43 eV to aliphatic amines, carbon fibers themselves. The fiber thermal expansion nitrile and amides [21,22], while in the cases of IM-500 (positive or negative) is very small [35]. As a consequence, and HM-45 carbon fibers these groups are not detected. Deconvolution of these peaks gives relative percentages of functional groups present and these are compiled in Table 3. The high strength carbon fibers (ST-3) possess the highest number of surface functional groups and as a result the surface of these fibers may be more disordered possessing comparatively high active surface area. The ST-3 fibers possess carboxylic, phenolic and hydroxyl functional groups which are acidic in nature while carbonyl and some of nitrogen containing groups are basic in character [23–29]. The high modulus HM-45 carbon fiber, on the other hand, possesses a minimum number of surface functional groups and a better ordered graphite-like carbon fiber surface. Therefore such fibers should have the lowest active surface area. The intermediate modulus (IM-500) carbon fibers possess the greatest number of hydroxyl functional groups. The contribution of carboxylic groups is almost the same in all the three fibers while carbonyl Fig. 2. Volume shrinkage observed with heat treatment temperagroups decrease from ST-3 to HM-45 fibers. ture of composites
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 these composites exhibit very little change parallel to shrinks onto the fibers when the interaction is maximum carbon fibers. However, appreciable changes in the width (as A). Such composites should exhibit nd thickness direction of shrinkage, as is observed Fig 2 shows the volume shrinkage of polymer compos ix interactions are weak(see Section 3.2) ites with heat treatment temperature. The amount of and as a result there is minimum shrinkage shrinkage depends upon the type of polymer matrix used in the fabrication of composites and fiber-matrix interactions ]. During heat treatment, chemical bonding between the 3.4. Bulk density vs HTT of composites fiber surface and the free functional groups of the resin undergoes rearrangement. Simultaneously, the molecular The bulk or apparent density is the density of compos- chains in the polymer matrix undergo pyrolysis and their ites containing voids and porosity. Table 4 shows the rearrangement results in matrix shrinkage. Up to 400C changes in density observed with HTT of the composites the shrinkage pattern of all the composites does not show These changes in density may be due to two factors much difference, but with increasing HTT shrinkage is maximum during the whole range of temperature and ()Dimensional changes due to shrinkage during (ii) Weight loss due to evolution of volatile product composites B and C it is 11-12%. The higher shrinkage in olysIs the case of composite A is attributed to strong fiber-matrix interactions. With increase in heat treatment temperature shrinkage increases continuously in all the composites and The initial density of the polymer composites depend it is -25% in composite A, 22% in composite B and 19% upon the density of( fibers used. its volume fraction in composite C at 1800C. Above 1800C, the conversion (which is kept the same in all three composites)and of non-graphitic carbon into a graphite-like carbon struc- compactness of the composites. The bulk density of ture(in the matrix), i.e., a reorientation of graphitic planes, lymer composites varied between 1.42 and 1.52 g/cm' starts taking place thus resulting in further composite pon heat treatment to 600C, a gradual density decreas shrinkage [11]. Further, with a continuous increase in heat is observed due to evolution treatment temperature to 2600C, the reorientation of formation of pores which results in volume expansion [36] graphitic planes takes place more extensively and the Above 600C, product evolution decreases to a large extent of shrinkage observed depends on the extent of extent as pyrolysis of the resin matrix is almost complete reorientation. The incremental shrinkage between 2000 and and there is an increase in density up to 1400C which is 2600C is as high as 6-7% in composite A and only 4% due to structural changes. The small density decreas in composite C. It is important to mention here that the between 1400 and 1800"C in composites A and B is due initial fiber volume(45+ 2%)is kept the same in all three to the formation of closed microporosity [37, 38]because cases. This brings out very clearly the effect of reactive of evolution of nitrogen-containing reaction products urface functional groups on the the fiber-matrix interac- tions. Consequently, during heat treatment the matrix m gto gs 4001800220026003000 HTT(C) Fig. 3. Change in kerosene density of composites with stages of
S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 1197 these composites exhibit very little change parallel to shrinks onto the fibers when the interaction is maximum carbon fibers. However, appreciable changes in the width (as in composite A). Such composites should exhibit a and thickness directions are observed. maximum amount of shrinkage, as is observed. In compoFig. 2 shows the volume shrinkage of polymer compos- site C, fiber–matrix interactions are weak (see Section 3.2) ites with heat treatment temperature. The amount of and as a result there is minimum shrinkage. shrinkage depends upon the type of polymer matrix used in the fabrication of composites and fiber–matrix interactions [9]. During heat treatment, chemical bonding between the 3 .4. Bulk density vs. HTT of composites fiber surface and the free functional groups of the resin undergoes rearrangement. Simultaneously, the molecular The bulk or apparent density is the density of composchains in the polymer matrix undergo pyrolysis and their ites containing voids and porosity. Table 4 shows the rearrangement results in matrix shrinkage. Up to 400 8C changes in density observed with HTT of the composites. the shrinkage pattern of all the composites does not show These changes in density may be due to two factors: much difference, but with increasing HTT shrinkage increases linearly in all the composites. For composite A it is maximum during the whole range of temperature and (i) Dimensional changes due to shrinkage during minimum in case of composite C. Up to 800 8C, 15% pyrolysis. shrinkage is noticed in the case of composite A, whereas in (ii) Weight loss due to evolution of volatile products composites B and C it is 11–12%. The higher shrinkage in during pyrolysis. the case of composite A is attributed to strong fiber–matrix interactions. With increase in heat treatment temperature, shrinkage increases continuously in all the composites and The initial density of the polymer composites depends it is |25% in composite A, 22% in composite B and 19% upon the density of carbon fibers used, its volume fraction in composite C at 1800 8C. Above 1800 8C, the conversion (which is kept the same in all three composites) and of non-graphitic carbon into a graphite-like carbon struc- compactness of the composites. The bulk density of3 ture (in the matrix), i.e., a reorientation of graphitic planes, polymer composites varied between 1.42 and 1.52 g/cm . starts taking place thus resulting in further composite Upon heat treatment to 600 8C, a gradual density decrease shrinkage [11]. Further, with a continuous increase in heat is observed due to evolution of reaction products and treatment temperature to 2600 8C, the reorientation of formation of pores which results in volume expansion [36]. graphitic planes takes place more extensively and the Above 600 8C, product evolution decreases to a large extent of shrinkage observed depends on the extent of extent as pyrolysis of the resin matrix is almost complete reorientation. The incremental shrinkage between 2000 and and there is an increase in density up to 1400 8C which is 2600 8C is as high as 6–7% in composite A and only 4% due to structural changes. The small density decrease in composite C. It is important to mention here that the between 1400 and 1800 8C in composites A and B is due initial fiber volume (456 2%) is kept the same in all three to the formation of closed microporosity [37,38] because cases. This brings out very clearly the effect of reactive of evolution of nitrogen-containing reaction products. surface functional groups on the the fiber–matrix interactions. Consequently, during heat treatment the matrix Table 4 Variation in bulk density of composites 3 HTT Bulk density (g/cm ) (8C) A BC 150 1.42 1.46 1.53 400 1.37 1.42 1.45 600 1.36 1.40 1.44 800 1.41 1.45 1.50 1000 1.45 1.50 1.53 1400 1.51 1.55 1.55 1800 1.50 1.53 1.57 2200 1.53 1.55 1.58 Fig. 3. Change in kerosene density of composites with stages of 2600 1.55 1.57 1.61 heat treatment
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 graphitization(and hence high density of the matrix as Porosity in composites heat-treated at 2600C* well as the carbon fiber surface Porosity Table 5 gives information about the total, open as well as closed porosity in the three composites heat treated at 2600C. Even though the same matrix has been used, oper 3l.4 porosity is maximum(13.9%) in composite A and mini 17.5 mum(8%)in composite C. It is well known that when soft 13.9 8 carbon is heat treated to temperatures of the order of Determined from true(2.26 ) and bulk densities 2200C and above, it becomes graphitized and, as a consequence of structural reorganization, a lot of porosity gets opened up [39]. However, in the present case,even though the maximum HTT (2600C)is the same in all the Composites A and B were made from carbon fibers which composites, the open porosity values are entirely different are not heat treated to more than 1400-1800C and In case of composite A, fiber-matrix interactions are therefore, contain some amount of nitrogen containing strong this generates the maximum amount of stress, which functional groups on their surface. Composite C, on the is thought to induce stress graphitization in the carbon other hand, does not show any decrease in the density matrix(at the interface). As a result, a large amount of since HM-45 carbon fibers (likely heat treatment porosity(previously closed) gets opened up. Likewise, in temperature>2200C)are nitrogen free (Table 2). Further, composite C interactions are weak and hence minimum on heat treatment to 2200 and 2600C, the bulk density degree of stress graphitization is induced. As a conse- gradually increases in all the composites, because of quence there is minimum open porosity in composite C breakdown of microporosity due to structural rearrange- By comparing the data of bulk and kerosene density it ment of carbon atoms. The final bulk density of compos- observed that the bulk density and kerosene density ites varies between 1.55 and 1.61 g/cm. The bulk density generally increase in all the composites with heat treatment of all the composites increases more or less with HTT In to 2600C. It is further observed that bulk density is the case of glass like carbon it decreases with htt due to minimum and kerosene density is maximum upon heat the continuous volume expansion of pores [38]. In the treatment at 2600C in composite A, as compared composites the volume expansion is thought to be con- composites B and C. This is again thought to be due to the trolled by fiber-matrix interactions act that, when carbon composites are heat treated above 2200C they become graphitized and, as a consequence of structural reorganization significant opening porosity is 3.5. Kerosene density vs. HIT of composites created. The extent of graphitization will be greater in ST-3 fiber-containing composite (e.g, due to frozen stres In order to get an idea about the total open porosity in s). As a result, kerosene density of composite A is the the composites, experiments were conducted to measure highest even through its bulk density is minimum kerosene density since kerosene is known to wet al carbons is capable of entering pores of up to 50 nm in size The kerosene density and bulk density give information about the open porosity present in the composites the kerosene density of composites A, B and C with hTT. Initially at 1000C the density of all the composites is in the range 1.60-1.62 omposite A while it increases from 1.65 to 1.75 g/cn for composite B. The sudden density increase in the case of composite A may be due to preferential graphitization (in the immediate vicinity of carbon fibers)of isotropi carbon derived from phenolic resin as well as the graphiti- zation of carbon fibers [12(see Fig. 7, composite A) However, in case of composites B and C kerosene density 0001400180022002600 registers a continuous increase with the ultimate value of HTT (C) 1.75 g/cm only at 2600C. This observation clearly Fig. 4. Variation in CTE in transverse direction of composites A brings out the role of fiber-matrix interaction in affecting B and C with stages of heat treatment
1198 S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 Table 5 graphitization (and hence high density) of the matrix as Porosity in composites heat-treated at 2600 8C* well as the carbon fiber surface. Porosity Composite type Table 5 gives information about the total, open as well (%) as closed porosity in the three composites heat treated at ABC 2600 8C. Even though the same matrix has been used, open Total 31.4 30.5 28.8 porosity is maximum (13.9%) in composite A and miniClosed 17.5 20.2 20.8 mum (8%) in composite C. It is well known that when soft Open 13.9 10.3 8 carbon is heat treated to temperatures of the order of 3 * Determined from true (2.26 g/cm ) and bulk densities. 2200 8C and above, it becomes graphitized and, as a consequence of structural reorganization, a lot of porosity gets opened up [39]. However, in the present case, even though the maximum HTT (2600 8C) is the same in all the Composites A and B were made from carbon fibers which composites, the open porosity values are entirely different. are not heat treated to more than 1400–1800 8C and In case of composite A, fiber–matrix interactions are therefore, contain some amount of nitrogen containing strong this generates the maximum amount of stress, which functional groups on their surface. Composite C, on the is thought to induce stress graphitization in the carbon other hand, does not show any decrease in the density matrix (at the interface). As a result, a large amount of since HM-45 carbon fibers (likely heat treatment porosity (previously closed) gets opened up. Likewise, in temperature.2200 8C) are nitrogen free (Table 2). Further, composite C interactions are weak and hence minimum on heat treatment to 2200 and 2600 8C, the bulk density degree of stress graphitization is induced. As a consegradually increases in all the composites, because of quence there is minimum open porosity in composite C. breakdown of microporosity due to structural rearrange- By comparing the data of bulk and kerosene density it is ment of carbon atoms. The final bulk density of compos- observed that the bulk density and kerosene density 3 ites varies between 1.55 and 1.61 g/cm . The bulk density generally increase in all the composites with heat treatment of all the composites increases more or less with HTT. In to 2600 8C. It is further observed that bulk density is the case of glass like carbon it decreases with HTT due to minimum and kerosene density is maximum upon heat the continuous volume expansion of pores [38]. In the treatment at 2600 8C in composite A, as compared to composites the volume expansion is thought to be con- composites B and C. This is again thought to be due to the trolled by fiber–matrix interactions. fact that, when carbon composites are heat treated above 2200 8C they become graphitized and, as a consequence of structural reorganization significant opening porosity is 3 .5. Kerosene density vs. HTT of composites created. The extent of graphitization will be greater in ST-3 fiber-containing composite (e.g., due to frozen stresIn order to get an idea about the total open porosity in ses). As a result, kerosene density of composite A is the the composites, experiments were conducted to measure highest even through its bulk density is minimum. kerosene density since kerosene is known to wet all carbons is capable of entering pores of up to 50 nm in size. The kerosene density and bulk density give information about the open porosity present in the composites. Fig. 3 shows changes in the kerosene density of composites A, B and C with HTT. Initially at 1000 8C the density of all the composites is in the range 1.60–1.62 3 g/cm . In the case of composites A and B, the density increases moderately between 1000 and 2000 8C, and then 3 it registers a sharp increase from 1.64 to 1.80 g/cm for3 composite A while it increases from 1.65 to 1.75 g/cm for composite B. The sudden density increase in the case of composite A may be due to preferential graphitization (in the immediate vicinity of carbon fibers) of isotropic carbon derived from phenolic resin as well as the graphitization of carbon fibers [12] (see Fig. 7, composite A). However, in case of composites B and C kerosene density registers a continuous increase with the ultimate value of 3 1.75 g/cm only at 2600 8C. This observation clearly Fig. 4. Variation in CTE in transverse direction of composites A, brings out the role of fiber–matrix interaction in affecting B and C with stages of heat treatment
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 1400 曰A 田B 工HOzuXH0 m c 400 15040060080010001400180022002600 HTT (C) Fig. 5. Change in flexural strength of composites with stages of heat treatment. 3.6. Coefficient of thermal expansion(CTE)vs. HTT of are expected to be maximum in composite A and minimum in composite C. This would explain why composite A shows a maximum rise and composite C a minimum CTE The thermal expansion measurement gives an idea of rise between 2000 and 2600C [11. This sugges dimensional stability and the preferred orientation and strong intercrystalline bonds in composite C prevent the graphitization of unidirectional carbon-carbon composites. full strain free c-axis due to minimum graphitization. The Cte parallel to the fiber axis is dominated by fiber thermal expansion [35]. In the transverse direction, how- ever, the CTE contribution from the matrix is significant as 3.7 Flexural strength vs. HTT of composites well, as it would depend on the type of fibers and thei surface characteristics. Therefore in the present study Cte Variations in flexural strength with HTT in the range measurements were carried out in the transverse direction 150 to 2600C are shown in Fig. 5. The strength of polymer composites made with high strength carbon fibers Fig. 4 shows the CTE variation in the transverse (composite A) is higher than that of composites made with direction. At 1000C the CtE of composite C is maximum intermediate and high modulus carbon fibers(composites B lat of composite A is minimum. The minimum value of and C). The ST-3 fibers possess maximum reactive CTE in composite A is attributed to the disordered functional groups as compared to IM-500 and HM-45 structure of the carbon fibers. (In composite C the carbon carbon fibers(Section 3. 1). As a result, ST-3 fibers are fibers possess more ordered structure). The lower thermal expected to have strongest interactions with the polymer expansion is a clear manifestation of the presence of a matrix and thus possess a high stress transfer capability strong boundary restraint and anisotropic contraction o With heat treatment, all the composites show a decrease in fibers and matrix. With increasing hTT, the Cte in all flexural strength up to a particular temperature and above it cases registers a gradual increase to almost 2000C, which an upward trend is observed. In the case of composite A is attributed to marginal improvements in the matrix flexural strength decreases suddenly from polymer stage to structure. Above 2000C there is a sharp increase for 600C, it increases thereafter to a maximum HTT of omposite A and moderate increases for composites B and 2600C On the other hand, fiexural strength of composi C. The former is thought to be due to the onset of stress B and c decreases from polymer stage to 1400 and graphitization [11]. Stress graphitization clearly depends 1000C, respectively, in the same manner as observed for on the degree of frozen stresses in the composites which composite minima in flexural strength in compos
S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 1199 Fig. 5. Change in flexural strength of composites with stages of heat treatment. 3 .6. Coefficient of thermal expansion (CTE) vs. HTT of are expected to be maximum in composite A and minimum composites in composite C. This would explain why composite A shows a maximum rise and composite C a minimum CTE The thermal expansion measurement gives an idea of rise between 2000 and 2600 8C [11]. This suggests that dimensional stability and the preferred orientation and strong intercrystalline bonds in composite C prevent the graphitization of unidirectional carbon–carbon composites. full strain free c-axis due to minimum graphitization. The CTE parallel to the fiber axis is dominated by fiber thermal expansion [35]. In the transverse direction, however, the CTE contribution from the matrix is significant as 3 .7. Flexural strength vs. HTT of composites well, as it would depend on the type of fibers and their surface characteristics. Therefore in the present study CTE Variations in flexural strength with HTT in the range measurements were carried out in the transverse direction 150 to 2600 8C are shown in Fig. 5. The strength of only. polymer composites made with high strength carbon fibers Fig. 4 shows the CTE variation in the transverse (composite A) is higher than that of composites made with direction. At 1000 8C the CTE of composite C is maximum intermediate and high modulus carbon fibers (composites B that of composite A is minimum. The minimum value of and C). The ST-3 fibers possess maximum reactive CTE in composite A is attributed to the disordered functional groups as compared to IM-500 and HM-45 structure of the carbon fibers. (In composite C the carbon carbon fibers (Section 3.1). As a result, ST-3 fibers are fibers possess more ordered structure). The lower thermal expected to have strongest interactions with the polymer expansion is a clear manifestation of the presence of a matrix and thus possess a high stress transfer capability. strong boundary restraint and anisotropic contraction of With heat treatment, all the composites show a decrease in fibers and matrix. With increasing HTT, the CTE in all flexural strength up to a particular temperature and above it cases registers a gradual increase to almost 2000 8C, which an upward trend is observed. In the case of composite A, is attributed to marginal improvements in the matrix flexural strength decreases suddenly from polymer stage to structure. Above 2000 8C there is a sharp increase for 600 8C, it increases thereafter to a maximum HTT of composite A and moderate increases for composites B and 2600 8C. On the other hand, flexural strength of composites C. The former is thought to be due to the onset of stress B and C decreases from polymer stage to 1400 and graphitization [11]. Stress graphitization clearly depends 1000 8C, respectively, in the same manner as observed for on the degree of frozen stresses in the composites which composite A. The minima in flexural strength in compos-
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 ites A, B and c are observed at 1800, 1400 and 1000C, In carbon fiber reinforced polymer composites, the flexural strength depends upon fiber-matrix bonding and mechanical properties of the fibers and the matrix pre- cursor. The strength of phenolic resin decreases upon heat treatment to 600C [40]. Therefore, all the composites exhibit a steady decrease in strength upon heat treatment. Above 800C, the matrix becomes strong and composite failure is controlled by crack propagation via fiber-matrix interactions. In composites A and b, the fracture would initiate at the tip of prestressed microcracks perpendicular to the fiber surface [7. However, in such composites the energy at the tip of the notches or cracks is expected to be smaller than the bonding energy between the fibers and the matrix. The crack, would therefore not stop at the fiber B surface but pass straight through the composite causing it to fail at low loading and low fexural strength [40). In posite C, the fiber-matrix bonding is quite weak and crack branching is expected to occur easily at the interface The matrix crack propagates to the fiber surface, stops there, is eliminated by absorbing fracture energy ol propagates further along the fiber-matrix interface. Hence, composite fracture by this mechanism requires a higher fracture energy and therefore composite C exhibits a higl ultimate breaking strength [41]. On the other hand,the strain to failure of high modulus carbon fibers and carbon- 6]. Ther 10m composite C would fail at the strain level at which fibers or matrix reach the maximum obtainable stress From Fig. 5 it is observed that the flexural strength, after eaching a minimum value, in each case registers an upward trend. The mechanical properties of brittle materi- als are highly notch sensitive. As described in our earlier 一B Fig. 7. Optical micrographs of composites A, B and C heat- study [42, the thermal stresses keep increasing upon heat treatment of composites and, as a result, the d spacing in all three composites increases, but only up to 2200C 5040060080010001400180022002600 falling thereafter because of stress relaxation [42]. How HTT(C) ever, the deflection temperature is different for each Fig.6.Change in ILSS of composites with stages of heat composite. Due to the generation of thermal stresses, many treatment microcracks are expected in the matrix, which will be
1200 S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 ites A, B and C are observed at 1800, 1400 and 1000 8C, respectively. In carbon fiber reinforced polymer composites, the flexural strength depends upon fiber–matrix bonding and mechanical properties of the fibers and the matrix precursor. The strength of phenolic resin decreases upon heat treatment to 600 8C [40]. Therefore, all the composites exhibit a steady decrease in strength upon heat treatment. Above 800 8C, the matrix becomes strong and composite failure is controlled by crack propagation via fiber–matrix interactions. In composites A and B, the fracture would initiate at the tip of prestressed microcracks perpendicular to the fiber surface [7]. However, in such composites the energy at the tip of the notches or cracks is expected to be smaller than the bonding energy between the fibers and the matrix. The crack, would therefore not stop at the fiber surface but pass straight through the composite causing it to fail at low loading and low flexural strength [40]. In composite C, the fiber–matrix bonding is quite weak and crack branching is expected to occur easily at the interface. The matrix crack propagates to the fiber surface, stops there, is eliminated by absorbing fracture energy or propagates further along the fiber–matrix interface. Hence, composite fracture by this mechanism requires a higher fracture energy and therefore composite C exhibits a high ultimate breaking strength [41]. On the other hand, the strain to failure of high modulus carbon fibers and carbonized matrix are nearly in the same range [6]. Therefore, composite C would fail at the strain level at which fibers or matrix reach the maximum obtainable stress. From Fig. 5 it is observed that the flexural strength, after reaching a minimum value, in each case registers an upward trend. The mechanical properties of brittle materials are highly notch sensitive. As described in our earlier Fig. 7. Optical micrographs of composites A, B and C heattreated at 2600 8C. study [42], the thermal stresses keep increasing upon heat treatment of composites and, as a result, the d spacing in all three composites increases, but only up to 2200 8C, falling thereafter because of stress relaxation [42]. However, the deflection temperature is different for each Fig. 6. Change in ILSS of composites with stages of heat composite. Due to the generation of thermal stresses, many treatment. microcracks are expected in the matrix, which will be
S.R. Dhakate, O.P. Bahl/ Carbon 41(2003)1193-1203 120l maximum in composite A, composites B and defined columnar type texture which is due to the strong C. As argued above, these ix interactions [12], while in composite B a crack initiation and explains lamellar type texture with less extinction lines is observed eratures. For composites A, B and C these are -1800, In composite C a predominantly lamellar type texture is 1400 and 1000C, respectively, and they indicate that the observed which is known to be due to the weak fiber- stress relaxation depends on the severity of fiber-matrix matrix interactions [46] interactions established at the polymer stage. As the degree of interactions increases(strong fiber-matrix bonding), the higher should be the deflection temperature. Above the deflection temperature, an improvement in the flexural 4. Conclusions strength is attributed to modification of the interface exhibiting improved crack tip deflection [43, 44. This is It is observed in the present investigation that the type of in composite C which sh owS 50% higher fiber-matrix interactions (i.e, strong or weak) depends on ngth(at 2600C)than composite A, even though the the relative percentage of reactive functional groups pres- strength of carbon fibers in composite C is only 50% of the ent on the carbon fiber surface and not the total amount of rength of carbon fibers in composite A unctional groups present. In the case of hi (ST-3)carbon fibers, almost 10% of the functional groups are reactive as compared to 5.5% and 4.5% in the case of 3. 8. ILSS ws. HTT of composites intermediate(IM-500) and high modulus(HM-45) carbo fibers. Consequently, ST-3 carbon fibers are in a position Fig 6 shows the changes in interlaminar shear strength to establish strongest interactions and HM-45 the weakest of composites A, B and C with HTT. At polymer stage nteractions with the matrix (150C)composite A has the maximum ILSS followed by The volume shrinkage occurring during heat treatment composites B and C. The maximum ILSS is a direct ( pyrolysis) is maximum in the case of ST-3 and minimum evidence of strong fiber-matrix interaction in composite in the case of HM-45 composites A, while the minimum ILSS in composite C is evidence of The flexural strength and interlaminar shear strength in weak interactions(see Section 3.2). The fibe all the composites is maximum at the polymer stage and site a possess maximum reactive functional groups and the decreases with HTT up to a characteristic temperature fibers in composite C possess minimum reactive functional (deflection temperature). To the best of our knowledge, this groups. is the first report that the deflection temperature depends Upon heat treatment, due to the evolution of volatile upon the severity of fiber-matrix interaction established at products and reorientation of matrix structure, the interfa. the polymer stage. Above the deflection temperature, cial strength between fibers and matrix degrades, and as a flexural strength to a modification of the esult, ILSS decreases in all the composites. The extent of nterface to favor ack tip deflection. this decrease is higher in composite A and lower in The coefficient expansion is confirmed to composite C. This is directly related to the volume depend strongly upon fiber-matrix interactions and matrix akage observed during HTT(see Fig. 2). The higher the cross sectional shrinkage (composite A) the more the interfacial flaws, thus explaining the lower ILSS value in of composite A at 1000C. In all the composites, in the temperature region 1000-1800C, the value of ilss is Acknowledgements nearly the same but above 1800C it increases and is maximum in case of composite A. The maximum ILSS value at 2600'C for all the composites is directly related The authors are grateful to Dr. K. Lal, Director, National the severity of interactions established at the polymer stage Physical Laboratory, New Delhi for his kind permission to which persist at higher heat treatment temperature. Lower publish these results. The authors also thank Dr. Anil K. ILSS values for composites B and C are also directly Gupta, Head of Engineering Materials Division for the related to weak interactions established at the polymer encouragement through out this investigation. stage which persist with HTT. These observations are in greement with those of Dhami et al. [45]. It is also observed in the optical micrographs(Fig. 7) that, in the case of composite A heat treated at 2600C, the fiber and References the matrix are in close proximity due to the strong fiber matrix interaction even at 2600C. Composite A shows a [1] Buckley JD l-carbon, an overview. Am Ceramic Soc lamellar type texture with strong extinction lines or a well Bull1988:67:364-8
S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 1201 maximum in composite A, followed by composites B and defined columnar type texture which is due to the strong C. As argued above, these microcracks act as centers of fiber–matrix interactions [12], while in composite B a crack initiation and explains the varying deflection tem- lamellar type texture with less extinction lines is observed. peratures. For composites A, B and C these are |1800, In composite C a predominantly lamellar type texture is 1400 and 1000 8C, respectively, and they indicate that the observed which is known to be due to the weak fiber– stress relaxation depends on the severity of fiber–matrix matrix interactions [46]. interactions established at the polymer stage. As the degree of interactions increases (strong fiber–matrix bonding), the higher should be the deflection temperature. Above the deflection temperature, an improvement in the flexural 4. Conclusions strength is attributed to modification of the interface It is observed in the present investigation that the type of exhibiting improved crack tip deflection [43,44]. This is more glaring in composite C which shows 50% higher fiber–matrix interactions (i.e., strong or weak) depends on the relative percentage of reactive functional groups pres- strength (at 2600 8C) than composite A, even though the ent on the carbon fiber surface and not the total amount of strength of carbon fibers in composite C is only 50% of the functional groups present. In the case of high strength strength of carbon fibers in composite A. (ST-3) carbon fibers, almost 10% of the functional groups are reactive as compared to 5.5% and 4.5% in the case of intermediate (IM-500) and high modulus (HM-45) carbon 3 .8. ILSS vs. HTT of composites fibers. Consequently, ST-3 carbon fibers are in a position to establish strongest interactions and HM-45 the weakest Fig. 6 shows the changes in interlaminar shear strength interactions with the matrix. of composites A, B and C with HTT. At polymer stage The volume shrinkage occurring during heat treatment (150 8C) composite A has the maximum ILSS followed by composites B and C. The maximum ILSS is a direct (pyrolysis) is maximum in the case of ST-3 and minimum in the case of HM-45 composites. evidence of strong fiber–matrix interaction in composite A, while the minimum ILSS in composite C is evidence of The flexural strength and interlaminar shear strength in weak interactions (see Section 3.2). The fibers in compo- all the composites is maximum at the polymer stage and decreases with HTT up to a characteristic temperature site A possess maximum reactive functional groups and the (deflection temperature). To the best of our knowledge, this fibers in composite C possess minimum reactive functional is the first report that the deflection temperature depends groups. Upon heat treatment, due to the evolution of volatile upon the severity of fiber–matrix interaction established at the polymer stage. Above the deflection temperature, products and reorientation of matrix structure, the interfa- flexural strength increases due to a modification of the cial strength between fibers and matrix degrades, and as a interface to favor improved crack tip deflection. result, ILSS decreases in all the composites. The extent of The coefficient of thermal expansion is confirmed to this decrease is higher in composite A and lower in depend strongly upon fiber–matrix interactions and matrix composite C. This is directly related to the volume structure. shrinkage observed during HTT (see Fig. 2). The higher the cross sectional shrinkage (composite A) the more the interfacial flaws, thus explaining the lower ILSS value in of composite A at 1000 8C. In all the composites, in the Acknowledgements temperature region 1000–1800 8C, the value of ILSS is nearly the same but above 1800 8C it increases and is maximum in case of composite A. The maximum ILSS The authors are grateful to Dr. K. Lal, Director, National value at 2600 8C for all the composites is directly related to Physical Laboratory, New Delhi for his kind permission to the severity of interactions established at the polymer stage publish these results. The authors also thank Dr. Anil K. which persist at higher heat treatment temperature. Lower Gupta, Head of Engineering Materials Division for the ILSS values for composites B and C are also directly encouragement through out this investigation. related to weak interactions established at the polymer stage which persist with HTT. These observations are in agreement with those of Dhami et al. [45]. It is also observed in the optical micrographs (Fig. 7) that, in the References case of composite A heat treated at 2600 8C, the fiber and the matrix are in close proximity due to the strong fiber– matrix interaction even at 2600 8C. Composite A shows a [1] Buckley JD. Carbon–carbon, an overview. Am Ceramic Soc lamellar type texture with strong extinction lines or a well Bull 1988;67:364–8
1202 S.R. Dhakate, O.P. Bahl /Carbon 41(2003)1193-1203 [2] Fitzer E. The future of carbon-carbon composites Carbon [23] Puri BR. In: Walker PL, Chemistry and physics of carbon, vol 6, New York: Marcel Dekker, 1971, P. 191 3] Weisshaus H, Kenig S, Siegmann A Effect of materials and [24] Bansal RC, Donnet JB, Stoeckli HF In: Active carbon, New rocessing on the mechanical properties of c-c composites York: Marcel Dekker, 1988, p. 119 Carbon 1991;29:1203-2 225] Boehm HP, Voll M. Basische oberflachenoxide auf kohler 4 Dillon F, Thomas KM, Marsh H. The influence of matrix stoff-. Adsorption von sauren. Carbon 1970: 8: 227-40 microstructure on mechanical properties of CFRC compos- 26] Contescu A, Vass M, Contescu C, Putyera K, Schwarz JA tes. Carbon1993;31:1337-48 Acid buffering capacity of basic carbons revealed by their 5] Fitzer E, Geigle KH, Weiss R Chemical interactions between continuous pK distribution. Carbon 1998: 36: 247-58 carbon fiber surface and epoxy resins. Carbon 1980: 18: 389-[27 Boehm HP. Some aspects of the surface chemistry of carbon blacks and other carbons. Carbon 1994: 32: 759-69 [6] Bradshaw WG, Vidoz AE. Fiber-matrix interactions in [28] Papirer E, Li S, Donnet JB. Contribution to the study of nidirectional carbon-carbon composites. Ceram Bull basic groups on carbon. Carbon 1987, 25: 243-8 1978;57:193-8 29 Papirer E, Dentzer J, Li S, Donnet JB. Surface group or I Fitzer E, Geigle KH, Huettner W. Infl of carbon fiber nitric acid oxidized carbon black sample determined by surface treatment on the mechanical properties of carbon- chemical thermodesorption. Carbon 1991; 29: 69-72 carbon composites. Carbon 1980: 18: 265-70 30] Drago RS, Voel GC, Needham TE. A four-parameter equa- [8 Manocha LM, Yasuda E, Tanabe Y, Kimura S. Effect of tion for predicting enthalpies of adduct formation. JAm carbon fiber surface treatment on mechanical properties of Chem Soc1971;,93:6014-26. C-c composites. Carbon 1988, 26: 333-7 31 Drago RS, Parr LB, Chamberlain GS Solent effects and their 9 Manocha LM, Bahl OP, Singh YK. Mechanical behaviour of elationship to the E and C equation. J Am Chem Soc carbon-carbon composites made with surface treated carbon 1977;9:3203 fibers. Carbon 1989- 27: 381-7. 32] Krekel G, Huttinger K, Hoffman WP. The relevance of [10] Bahl OP, Dhami TL. In: Proceedings of the International surface chemistry of carbon fiber in their adhesion to high Conference"Carbon and Carbonaceous Composite Materi- temperature thermoplastic. Part Il. Surface chemistry. J Mater Sci 1994- 29: 3461-8. observation on ess B3 Dhakate SR, Bahl OP, Sahare PD. Oxidation behaviour of Pan based carbon fiber reinforced phenolic resin matrix 1991:29:1155-63 composites. J Mater Sci Lett 2000: 19(21): 1959-61 [12] Hishiyama Y, Inagaki M, Kimura S, Yamada S. Graphitiza- [34] Jenkins GM, Kawamura K. Polymeric carbons. Cambridge on of carbon fiber/glassy carbon composites. Carbon 35] Morgan wC. Thermal expansion of graphite crystals. Carbon [13] Manocha LM. The effect of heat treatment temperature on 1972:10:73-80 the properties of polyfurfuryl alcohol based carbon-carbon [36]Choe CR, Lee KH. Effect of processing pal the omposites Carbon 1994: 32: 213-23 mechanical properties of carbonized phenolic [14] Chlopek J, Blzewicz S. Effect on processing variables on the Carbon1992;30:247-9 properties of carbon-carbon composites. Carbon [37 Kipling JJ, Sherwood JN, Shorter PV, Thompson NR. The 991;29:127-31 pore structure and surface area of high temperature polymer [15] Takano T, Uruno T, Kinjo T, Tlomak P, Ju CP. Structure and carbons. Carbon 1964: 1- 321-8 of unidirectionally reinforced PAN-resin based [38 Mehrotra BN, Bragg RH, Rao AS. Effect of heat treatment opposites. J Mater Sci 1993: 28: 5610-9 temperature(HTT)on density, weight and volume of glass- [16 Dhakate SR, Parashar VK, Raman V, Bahl OP. Effect of like carbon(GC). J Mater Sci 1983: 18: 2671-8 titania(TiO,)interfacial coating on mechanical propertie Savage G. Carbon-carbon composites. Chapman and Hall, J Mater Sci Lett 1993 2000;19(8)699-701. )Lausevic Z, Marinkovic S. Mechanical properties and [17] Ishitani A. Application of X-ray photoelectron spectroscopy chemistry of carbonization of phenol formaldehyde resin. to surface analysis of carbon fiber. Carbon 1981: 19: 269-75 rbon1986;,24:575-80. [18] Nakayama Y, Soeda F, Ishitani A. XPS study of the carbon [41] Manocha LM. Changes in physical and mechanical prop- fiber matrix interface. Carbon 1990- 28: 21-6 erties of carbon fiber-reinforced polyfurfuryl alcohol com- [19] Proctor A, Sherwood PMA, X-Ray photoelectron spectro- posites during their pyrolysis to carbon-carbon composites. scopic studies of carbon fiber surface Il. The effect opposites 1988: 19: 311 electrochemical treatment. Carbon 1983: 21: 53- [42] Dhakate SR, [20] Kozlowski C, Sherwood PMA. X-ray photoelectron spectro- nterface on crystalline parameters, coefficient of thermal scopic studies of carbon fiber surface VIll. A comparison of pansion and electrical conductivity in carbon-carbon type I and ll fibers and their interactions with thin resin film In: Carbon 2001 Conference, Kentucky, KY, n198725:751-60. 01, Session 23, Paper 23.3 [21 Jones C, Samman E. The effect of low power plasmas 43]M LM, Bahl OP. Influence of carbon type and weave carbon fiber surfaces. Carbon 1990- 28: 509-14. pattern on the development of 2D carbon-carbon compos [ 22] Jones C. Effect of electrochemical and plasma carbon fiber surfaces. Surf Interface Anal 1993: 20: 357-67 [44] Mc Enaney B, Mays T. Relation between microstructure and
1202 S.R. Dhakate, O.P. Bahl / Carbon 41 (2003) 1193–1203 [2] Fitzer E. The future of carbon–carbon composites. Carbon [23] Puri BR. In: Walker PL, editor, Chemistry and physics of 1987;5:163–90. carbon, vol. 6, New York: Marcel Dekker, 1971, p. 191. [3] Weisshaus H, Kenig S, Siegmann A. Effect of materials and [24] Bansal RC, Donnet JB, Stoeckli HF. In: Active carbon, New processing on the mechanical properties of c–c composites. York: Marcel Dekker, 1988, p. 119. Carbon 1991;29:1203–20. [25] Boehm HP, Voll M. Basische oberflachenoxide auf kohlen- [4] Dillon F, Thomas KM, Marsh H. The influence of matrix stoff—I. Adsorption von sauren. Carbon 1970;8:227–40. microstructure on mechanical properties of CFRC compos- [26] Contescu A, Vass M, Contescu C, Putyera K, Schwarz JA. ites. Carbon 1993;31:1337–48. Acid buffering capacity of basic carbons revealed by their [5] Fitzer E, Geigle KH, Weiss R. Chemical interactions between continuous pK distribution. Carbon 1998;36:247–58. carbon fiber surface and epoxy resins. Carbon 1980;18:389– [27] Boehm HP. Some aspects of the surface chemistry of carbon 99. blacks and other carbons. Carbon 1994;32:759–69. [6] Bradshaw WG, Vidoz AE. Fiber–matrix interactions in [28] Papirer E, Li S, Donnet JB. Contribution to the study of unidirectional carbon–carbon composites. Ceram Bull basic groups on carbon. Carbon 1987;25:243–8. 1978;57:193–8. [29] Papirer E, Dentzer J, Li S, Donnet JB. Surface group on [7] Fitzer E, Geigle KH, Huettner W. Influence of carbon fiber nitric acid oxidized carbon black sample determined by surface treatment on the mechanical properties of carbon– chemical thermodesorption. Carbon 1991;29:69–72. carbon composites. Carbon 1980;18:265–70. [30] Drago RS, Voel GC, Needham TE. A four-parameter equa- [8] Manocha LM, Yasuda E, Tanabe Y, Kimura S. Effect of tion for predicting enthalpies of adduct formation. J Am carbon fiber surface treatment on mechanical properties of Chem Soc 1971;93:6014–26. c–c composites. Carbon 1988;26:333–7. [31] Drago RS, Parr LB, Chamberlain GS. Solent effects and their [9] Manocha LM, Bahl OP, Singh YK. Mechanical behaviour of relationship to the E and C equation. J Am Chem Soc carbon–carbon composites made with surface treated carbon 1977;99:3203–9. fibers. Carbon 1989;27:381–7. [32] Krekel G, Huttinger KJ, Hoffman WP. The relevance of [10] Bahl OP, Dhami TL. In: Proceedings of the International surface chemistry of carbon fiber in their adhesion to high Conference ‘‘Carbon and Carbonaceous Composite Materi- temperature thermoplastic. Part II. Surface chemistry. J als’’, Malenovice, October 1995. Mater Sci 1994;29:3461–8. [11] Zaldivar RJ, Rellick GS. Some observation on stress [33] Dhakate SR, Bahl OP, Sahare PD. Oxidation behaviour of graphitization in carbon–carbon composites. Carbon PAN based carbon fiber reinforced phenolic resin matrix 1991;29:1155–63. composites. J Mater Sci Lett 2000;19(21):1959–61. [12] Hishiyama Y, Inagaki M, Kimura S, Yamada S. Graphitiza- [34] Jenkins GM, Kawamura K. Polymeric carbons. Cambridge tion of carbon fiber/glassy carbon composites. Carbon University Press, 1978. 1974;12:249–58. [35] Morgan WC. Thermal expansion of graphite crystals. Carbon [13] Manocha LM. The effect of heat treatment temperature on 1972;10:73–80. the properties of polyfurfuryl alcohol based carbon–carbon [36] Choe CR, Lee KH. Effect of processing parameters on the composites. Carbon 1994;32:213–23. mechanical properties of carbonized phenolic resin matrix. [14] Chlopek J, Blzewicz S. Effect on processing variables on the Carbon 1992;30:247–9. properties of carbon–carbon composites. Carbon [37] Kipling JJ, Sherwood JN, Shorter PV, Thompson NR. The 1991;29:127–31. pore structure and surface area of high temperature polymer [15] Takano T, Uruno T, Kinjo T, Tlomak P, Ju CP. Structure and carbons. Carbon 1964;1:321–8. properties of unidirectionally reinforced PAN-resin based [38] Mehrotra BN, Bragg RH, Rao AS. Effect of heat treatment composites. J Mater Sci 1993;28:5610–9. temperature (HTT) on density, weight and volume of glass- [16] Dhakate SR, Parashar VK, Raman V, Bahl OP. Effect of like carbon (GC). J Mater Sci 1983;18:2671–8. titania (TiO ) interfacial coating on mechanical properties of [39] Savage G. Carbon–carbon composites. Chapman and Hall, 2 carbon–carbon composites. J Mater Sci Lett 1993. 2000;19(8):699–701. [40] Lausevic Z, Marinkovic S. Mechanical properties and [17] Ishitani A. Application of X-ray photoelectron spectroscopy chemistry of carbonization of phenol formaldehyde resin. to surface analysis of carbon fiber. Carbon 1981;19:269–75. Carbon 1986;24:575–80. [18] Nakayama Y, Soeda F, Ishitani A. XPS study of the carbon [41] Manocha LM. Changes in physical and mechanical prop- fiber matrix interface. Carbon 1990;28:21–6. erties of carbon fiber-reinforced polyfurfuryl alcohol com- [19] Proctor A, Sherwood PMA, X-Ray photoelectron spectro- posites during their pyrolysis to carbon–carbon composites. scopic studies of carbon fiber surface II. The effect of Composites 1988;19:311–9. electrochemical treatment. Carbon 1983;21:53–9. [42] Dhakate SR, Bahl OP, Mathur RB, Dhami TL. Influence of [20] Kozlowski C, Sherwood PMA. X-ray photoelectron spectro- interface on crystalline parameters, coefficient of thermal scopic studies of carbon fiber surface VIII. A comparison of expansion and electrical conductivity in carbon–carbon type I and II fibers and their interactions with thin resin film. composites. In: Carbon 2001 Conference, Kentucky, KY, Carbon 1987;25:751–60. 14–19 July 2001, Session 23, Paper 23.3. [21] Jones C, Samman E. The effect of low power plasmas on [43] Manocha LM, Bahl OP. Influence of carbon type and weave carbon fiber surfaces. Carbon 1990;28:509–14. pattern on the development of 2D carbon–carbon compos- [22] Jones C. Effect of electrochemical and plasma treatment on ites. Carbon 1988;26:13–22. carbon fiber surfaces. Surf Interface Anal 1993;20:357–67. [44] McEnaney B, Mays T. Relation between microstructure and