40TH ANNIVERSARY face in the depth direction, it can be concluded that their gation. All matrix cracks in these systems, irrespective of advance must be significant and possibly compromises them being PMCs or HMCs, are confined to the surfaces the structural integrity of the laminate of these materials even when the material was subjected to the highest temperature differential. By contrast, HMCS in the thicker multi-layer laminates became much deeper as the applied shock increased in severity 4. Discussion The difference with which each type of ply accom- A number of general observations can be made regarding modates the energy available for crack propagation must the thermal shock behaviour of the laminates analysed be highlighted. The application of more severe shocks in in the previous sections. The main mode of damage af- longitudinal(0%) plies results in the rapid multiplication ter thermal shock treatment in all laminates, irrespective of surface cracks, which also seem to appear at regular of configuration, is the formation of cracks in the ma- intervals along the ply length. This is similar to the sit trix. These cracks were deflected at fibre-matrix interfaces uation under tensile testing(. g. Pryce and Smith [16]) and, thus, did not result in any damage to the fibres, as is and implies that stress transfer may take place during expected for optimally-designed fibre-reinforced CMCs. thermal shock between fibre/matrix and between differ This is in accordance with reports of thermal shock dam- ent plies. This stress transfer mechanism results in the ge in similar materials published in the literature(e. g.[3- energy available for cracking in 0 plies being consumed 7D.However, matrix cracks in different plies advanced to- mainly in multiplying the number of cracks; in contrast, in wards different directions: at right angles to the exposed fi- 90 plies, no transfer takes place and the energy available bre length in longitudinal plies and along the length of the results in the extension(in length and depth) of a small ply surface in transverse plies. This seems to be a general number of cracks that appear at preferential sites. These feature of this class of CMC. More specifically, Kagawa cracks do not increase in number but become deeper and et al.[3], Blissett et al. [4], and Boccaccini et al. [5,6]re- deeper at higher temperature differentials, something that ported PMCs on longitudinal faces of UD Nicalon/Pyrex, would affect the integrity of the material. From this as Nicalon/CAS, and Nicalon/Duran, respectively. In addi- pect, transverse plies behave in a similar way to monolithic tion, Blissett et al. [ 14] presented evidence of PMCs in or particulate-reinforced ceramic materials under thermal the longitudinal plies of (0 2/90%4)s and (0% 90)3s lami- shock loading. In contrast, longitudinal plies show true nates. In contrast, cracks similar to the HMCs described'composite', and thus superior, behaviour under thermal in this paper have been reported for the end faces of UD shock conditions Nicalon/CAS [4] and Nicalon/LAS I [1O], as well as for Finally, this study revealed an important aspect of the the transverse plies of(02 /904)s and0 /90 )3s laminates behaviour of cross-ply Nicalon/CAS under conditions of [14]. This provides evidence for the biaxiality of the sur- thermal shock: cracking always originated in the central face stresses develop as a result of thermal shock, as under plies and, in particular, at or near the centreline of the uniaxial tensile testing only PMCs(as defined here)can face. In addition, crack densities in the central plies were be seen on the surfaces of UD or cross-ply CMCs(e.g. higher at all ATs compared with those of plies located towards the top and bottom edges. Blissett et al. [4] also Fibre failures, such as those reported by Blissett et al. reported that the deep crack due to thermal shock [4], were not detected in this investigation. As these au- always ran along the centreline of the quenched end-face thors associated the occurrence of such damage patterns of a UD Nicalon/CAS. This shows that the highest stress with material degradation after high-temperature expo- at each AT occurs always at the centreline and gradually sure, it can be concluded that their absence in this study reduces as the top and bottom edges are approached. This resulted from the short times the material samples were was true especially for PMCs in longitudinal plies. HMCs held at high temperature, which did not allow the forma-followed random pattern but plies towards the tion of any oxidation products on the material surfaces. centreline always had longer and deeper cracks The lower thickness of the simple cross-ply laminates had a major effect on their thermal shock resistance as these laminates exhibited critical temperature differen tials 100C higher than those of their multi-layer coun- 5. Concluding remarks terparts. This is a well-documented aspect of the thermal Damage due to thermal shock in a range of simple and shock behaviour of ceramic materials. It has been shown multi-layer cross-ply Nicalon/ CAS CMCs was charac that a critical dimension exists above which the ATc be- terised in this paper. comes independent of material dimensions. Samples with Cracking al ways originated in the central plies and con at least one dimension lower than this critical value exhibit sisted of matrix cracks in various directions depending on much higher resistance to thermal shock [1] the orientation of each ply. Application of shocks of in The lower thickness of the simple cross-ply laminates creasing severity resulted in damage being extended to eems also to affect the energy available for crack propa- the remaining plies, although crack density was always40TH ANNIVERSARY face in the depth direction, it can be concluded that their advance must be significant and possibly compromises the structural integrity of the laminate. 4. Discussion A number of general observations can be made regarding the thermal shock behaviour of the laminates analysed in the previous sections. The main mode of damage af￾ter thermal shock treatment in all laminates, irrespective of configuration, is the formation of cracks in the ma￾trix. These cracks were deflected at fibre-matrix interfaces and, thus, did not result in any damage to the fibres, as is expected for optimally-designed fibre-reinforced CMCs. This is in accordance with reports of thermal shock dam￾age in similar materials published in the literature (e.g. [3– 7]). However, matrix cracks in different plies advanced to￾wards different directions: at right angles to the exposed fi- bre length in longitudinal plies and along the length of the ply surface in transverse plies. This seems to be a general feature of this class of CMC. More specifically, Kagawa et al. [3], Blissett et al. [4], and Boccaccini et al. [5, 6] re￾ported PMCs on longitudinal faces of UD Nicalon/Pyrex, Nicalon/CAS, and Nicalon/Duran, respectively. In addi￾tion, Blissett et al. [14] presented evidence of PMCs in the longitudinal plies of (0◦ 2/90◦ 4)s and (0◦/90◦)3s lami￾nates. In contrast, cracks similar to the HMCs described in this paper have been reported for the end faces of UD Nicalon/CAS [4] and Nicalon/LAS II [10], as well as for the transverse plies of (0◦ 2/90◦ 4)s and (0◦/90◦)3s laminates [14]. This provides evidence for the biaxiality of the sur￾face stresses develop as a result of thermal shock, as under uniaxial tensile testing only PMCs (as defined here) can be seen on the surfaces of UD or cross-ply CMCs (e.g. [16]). Fibre failures, such as those reported by Blissett et al. [4], were not detected in this investigation. As these au￾thors associated the occurrence of such damage patterns with material degradation after high-temperature expo￾sure, it can be concluded that their absence in this study resulted from the short times the material samples were held at high temperature, which did not allow the forma￾tion of any oxidation products on the material surfaces. The lower thickness of the simple cross-ply laminates had a major effect on their thermal shock resistance as these laminates exhibited critical temperature differen￾tials ∼100◦C higher than those of their multi-layer coun￾terparts. This is a well-documented aspect of the thermal shock behaviour of ceramic materials. It has been shown that a critical dimension exists above which the
Tc be￾comes independent of material dimensions. Samples with at least one dimension lower than this critical value exhibit much higher resistance to thermal shock [1]. The lower thickness of the simple cross-ply laminates seems also to affect the energy available for crack propa￾gation. All matrix cracks in these systems, irrespective of them being PMCs or HMCs, are confined to the surfaces of these materials even when the material was subjected to the highest temperature differential. By contrast, HMCs in the thicker multi-layer laminates became much deeper as the applied shock increased in severity. The difference with which each type of ply accom￾modates the energy available for crack propagation must be highlighted. The application of more severe shocks in longitudinal (0◦) plies results in the rapid multiplication of surface cracks, which also seem to appear at regular intervals along the ply length. This is similar to the sit￾uation under tensile testing (e.g. Pryce and Smith [16]), and implies that stress transfer may take place during thermal shock between fibre/matrix and between differ￾ent plies. This stress transfer mechanism results in the energy available for cracking in 0◦ plies being consumed mainly in multiplying the number of cracks; in contrast, in 90◦ plies, no transfer takes place and the energy available results in the extension (in length and depth) of a small number of cracks that appear at preferential sites. These cracks do not increase in number but become deeper and deeper at higher temperature differentials, something that would affect the integrity of the material. From this as￾pect, transverse plies behave in a similar way to monolithic or particulate-reinforced ceramic materials under thermal shock loading. In contrast, longitudinal plies show true ‘composite’, and thus superior, behaviour under thermal shock conditions. Finally, this study revealed an important aspect of the behaviour of cross-ply Nicalon/CAS under conditions of thermal shock: cracking always originated in the central plies and, in particular, at or near the centreline of the face. In addition, crack densities in the central plies were higher at all
Ts compared with those of plies located towards the top and bottom edges. Blissett et al. [4] also reported that the major, deep crack due to thermal shock always ran along the centreline of the quenched end-face of a UD Nicalon/CAS. This shows that the highest stress at each
T occurs always at the centreline and gradually reduces as the top and bottom edges are approached. This was true especially for PMCs in longitudinal plies. HMCs followed a more random pattern but plies towards the centreline always had longer and deeper cracks. 5. Concluding remarks Damage due to thermal shock in a range of simple and multi-layer cross-ply Nicalon/CAS CMCs was charac￾terised in this paper. Cracking always originated in the central plies and con￾sisted of matrix cracks in various directions depending on the orientation of each ply. Application of shocks of in￾creasing severity resulted in damage being extended to the remaining plies, although crack density was always 961