V Bianchi et al the matrix, o m, but also the interfacial stress should rication. It can then be stated that, in the case of smooth herefore be taken into account to explain the micro- fibers, the valuc of the thermal residual stresses explains cracking in the composites. Since a, is a tensile stress, a the interfacial decohesion Furthermore the interfacial longitudinal tensile stress lower than the fracture stress,a,is high enough to imply debonding. In con strength of the matrix(<90 MPa for the crystallized trast, in the case of corrugated fibers, the reactivity of matrix,probably lower if the matrix is vitreous)can be the carbon and the topography of the surface seem to sufficient to lead to microcracking. The curves of play a much greater role than the influence of the ther Fig 6(b)and(c)show that the resultant stress would be mal expansion mismatch even greater at higher sintering temperatures When the sintering temperature reaches 1250 C, the In the case of T400H composites, a; is lower than for interfacial normal stress is little changed from its value P25 composites, suggesting that the debonding would at 20oC and should not result in wider decohesion this be less extensive. Nevertheless, a; is a tensile stress but is is confirmed by TEM observations. In addition, another very low if the matrix is kept vitreous [Fig. 7(a)). The phenomenon seems to explain the longer extractions of curves of Fig. 7(a)and(b) show that microcracking the fibers. which are exhibited on surface fractures would appear at higher temperatures if the matrix were and confirmed by push-in rcsults(the interfacial shear maintained in a vitreous state, since the fracture stress falls from 17. 8 MPa to I MPa)and the increase in strength of the vitreous matrix is likely to be lower than the mean distance between matrix microcracks(the that of the crystallized matrix. This result is in agree ment with the observations made for the youngs mod interfacial shear stress). 29 This phenomenon is identified ulus measurements, although the temperatures at which as the reduction of the matrix by the carbon fiber which microcracking appears cannot be determined exactly is confirmed by the amorphous aspect of several resi from the calculation dual contact zones and the fact that the outlines of the decohesion are no longer parallel and are less irre 4 DISCUSSION Concerning the T400H/YMAS composites fabricated It low temperature (composite T1), calculation of the To explain the strength of the fiber matrix interface and normal interfacial stress has shown that the interf the mechanical behavior of the composites, it is neces- should be slightly under tension. However, the TEM sary to take into account not only the thermal residual study has demonstrated the formation of a carbonac- stresses but also physico-chemical reactions between the eous interphase which is well adhered to both the fiber carbon fibers and the matrix. Previously, 2 the interfacial and the matrix. This amorphous carbon interphase shear stress has been measured in push-in tests for dif- about 20 nm in thickness, can originate from the pyr ferent composites and the interfaces have been observed olysis and carbonization of sizing. Hence, this chemical y transmission electronic microscopy (TEM). These reaction plays a priori a dominant role on the very high experiments show that in some cases the carbon of the strength of the interface (t is determined by push-in fiber reacts with the oxygen of the matrix. The aim of tests as close to 108 MPa)and the fracture surfaces. The this paper is to clarify, for the different hot-pressing fibers are broken cleanly in the fracture plane, except for conditions, the thermomechanical or chemical phenom- some non-impregnated bundles of fibers which are na which lead to the nature and strength of the fiber/ pulled-out and explain the non-brittle fracture. The matrix interface magnitude of the interfacial shear stress is confirmed by In P25/YMAS composites, TEM observations have the small distances between matrix microcracks shown that, for composites hot-pressed between 1000 If the hot- perature ind 1 C, slight debonding between the fibers and the facial separation can be observed by tem. the outlines matrix can occur. In this case, i.e. if the aspect of the of the separation indicate that this is mainly due to the er is smooth and the carbon aromatic layers are par- thermal expansion mismatch. This assumption is sup allel to the fiber surface, the outlines of the matrix fol- ported by the increase of the calculated normal stress, w exactly those of the fiber, suggesting that the i, by the decrease of the interfacial shear stress, t. and debonding is essentially due to the thermal expansion at the same time, by the increase of the mean distance atch between the two constituents If the fiber is between microcracks. the weakening of the interface corrugated, therc is no interfacial scparation, owing to then pcrmits a rise of the fr the strong chemical bonds between the fiber and the 1 100 MPa, the debonding and sliding of the fiber in the atrix resulting from the reactivity of disorganized car- matrix blocks allowing a large energy consumption until bon, and also to the topography of the fiber surface the rupture of the fibers. However, some contact zones which permits good mechanical interlocking. Indeed, it are observed formed from the amorphous carbon has been shown that the orientation of surface carbon in the composite TI aromatic layers in P25 fibers is associated with the For T400H/YMAS composites submitted to an aspect of the corrugation, which would be due to a pre- annealing at 1250 C, the interfacial shear stress is not ferential and local oxidation of fibers during their fab- lower, whereas the ultimate strength decreases, the416 V. Bianchi et al the matrix, rr_y, but also the interfacial stress should therefore be taken into account to explain the micro￾cracking in the composites. Since o, is a tensile stress, a longitudinal tensile stress lower than the fracture strength of the matrix (Z 90 MPa for the crystallized matrix,27 probably lower if the matrix is vitreous) can be sufficient to lead to microcracking. The curves of Fig. 6(b) and (c) show that the resultant stress would be even greater at higher sintering temperatures. In the case of T400H composites, gi is lower than for P25 composites, suggesting that the debonding would be less extensive. Nevertheless, oi is a tensile stress but is very low if the matrix is kept vitreous [Fig. 7(a)]. The curves of Fig. 7(a) and (b) show that microcracking would appear at higher temperatures if the matrix were maintained in a vitreous state, since the fracture strength of the vitreous matrix is likely to be lower than that of the crystallized matrix. This result is in agree￾ment with the observations made for the Young’s mod￾ulus measurements, although the temperatures at which microcracking appears cannot be determined exactly from the calculations. 4 DISCUSSION To explain the strength of the fiber/matrix interface and the mechanical behavior of the composites, it is neces￾sary to take into account not only the thermal residual stresses but also physico-chemical reactions between the carbon fibers and the matrix. Previously,* the interfacial shear stress has been measured in push-in tests for dif￾ferent composites and the interfaces have been observed by transmission electronic microscopy (TEM). These experiments show that in some cases the carbon of the fiber reacts with the oxygen of the matrix. The aim of this paper is to clarify, for the different hot-pressing conditions, the thermomechanical or chemical phenom￾ena which lead to the nature and strength of the fiber/ matrix interface. In P25/YMAS composites, TEM observations have shown that, for composites hot-pressed between 1000 and 1150°C slight debonding between the fibers and the matrix can occur. In this case, i.e. if the aspect of the fiber is smooth and the carbon aromatic layers are par￾allel to the fiber surface, the outlines of the matrix fol￾low exactly those of the fiber, suggesting that the debonding is essentially due to the thermal expansion mismatch between the two constituents. If the fiber is corrugated, there is no interfacial separation, owing to the strong chemical bonds between the fiber and the matrix resulting from the reactivity of disorganized car￾bon2s and also to the topography of the fiber surface which permits good mechanical interlocking. Indeed, it has been shown that the orientation of surface carbon aromatic layers in P25 fibers is associated with the aspect of the corrugation, which would be due to a pre￾ferential and local oxidation of fibers during their fab￾rication. It can then be stated that, in the case of smooth fibers, the value of the thermal residual stresses explains the interfacial decohesion. Furthermore, the interfacial stress, ci, is high enough to imply debonding. In con￾trast, in the case of corrugated fibers, the reactivity of the carbon and the topography of the surface seem to play a much greater role than the influence of the ther￾mal expansion mismatch. When the sintering temperature reaches 125O”C, the interfacial normal stress is little changed from its value at 20°C and should not result in wider decohesions. This is confirmed by TEM observations. In addition, another phenomenon seems to explain the longer extractions of the fibers, which are exhibited on surface fractures and confirmed by push-in results (the interfacial shear stress falls from 17.8 MPa to 1 MPa) and the increase in the mean distance between matrix microcracks (the interval between cracks is in inverse proportion with the interfacial shear stress). 29 This phenomenon is identified as the reduction of the matrix by the carbon fiber which is confirmed by the amorphous aspect of several resi￾dual contact zones and the fact that the outlines of the decohesion are no longer parallel and are less irre￾gular. Concerning the T400H/YMAS composites fabricated at low temperature (composite Tl), calculation of the normal interfacial stress has shown that the interface should be slightly under tension. However, the TEM study has demonstrated the formation of a carbonac￾eous interphase which is well adhered to both the fiber and the matrix. This amorphous carbon interphase, about 20nm in thickness, can originate from the pyr￾olysis and carbonization of sizing. Hence, this chemical reaction plays a priori a dominant role on the very high strength of the interface (t is determined by push-in tests as close to 108 MPa) and the fracture surfaces. The fibers are broken cleanly in the fracture plane, except for some non-impregnated bundles of fibers which are pulled-out and explain the non-brittle fracture. The magnitude of the interfacial shear stress is confirmed by the small distances between matrix microcracks. If the hot-pressing temperature is raised, an inter￾facial separation can be observed by TEM. The outlines of the separation indicate that this is mainly due to the thermal expansion mismatch. This assumption is sup￾ported by the increase of the calculated normal stress, D,, by the decrease of the interfacial shear stress, r, and. at the same time, by the increase of the mean distance between microcracks. The weakening of the interface then permits a rise of the fracture strength up to 1100 MPa, the debonding and sliding of the fiber in the matrix blocks allowing a large energy consumption until the rupture of the fibers. However, some contact zones are observed, formed from the amorphous carbon seen in the composite Tl. For T400H/YMAS composites submitted to an annealing at 1250°C the interfacial shear stress is not lower, whereas the ultimate strength decreases, the