《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_Damage characterisation of thermally shocked.pdf_大学文库

40TH ANNIVERSARY J MATER SCI41(006)951-962 Damage characterisation of thermally shocked cross-ply ceramic composite laminates C. KASTRITseas. P A. SMITHJ.A. YEOMANS' School of Engineering(H6, University of Surrey, Guildford, GU2 7XH, Surrey U E-mail: j- yeomans@surrey. ac uk amage due to thermal shock in cross-ply Nicalon/calcium aluminosilicate ceramic matrix composites has been investigated. Heated specimens of two simple [(0/90 )s and (90/0@)s] and two multi-layer [(0%/90)3s and(90/0)3s] materials were quenched into water at room temperature Crack morphologies were assessed by reflected light microscopy and scanning electron microscopy. The use of image assembling software allowed the generation of reflected light microscopy images of all of the thermally-shocked surfaces onto which the crack patterns were then superimposed. This allowed clear identification of damage mechanisms and accurate quantification of damage accumulation with increasing severity of thermal shock. Damage was first detected in the central plies of each composite Composites with 0ocentr plies exhibited slightly higher resistance to thermal shock than their counterparts with 90 central plies. Although damage extended to the outer plies as the severity of the shock increased, crack density was found to vary with position at every shock: it was highest in the central plies and gradually reduced towards the outer plies. Multiple matrix cracking erpendicular to the fibre direction was the damage mode identified in 0 plies while 90 plie contained cracks that ran along the ply length At more severe shocks the morphology of these crack patterns was affected in significantly different ways. In addition, the thinner, simple cross-ply composites exhibited much higher resistance to thermal shock than their multi-layer counterparts. 2006 Springer Science Business Media, Inc 1. Introduction deflecting fibre-matrix interfaces ensures that the fibres As fibre-reinforced ceramic matrix composites(CMCs) remain largely unaffected, thus preserving the integrity of are being considered increasingly for high-performance the material. However, damage due to thermal shock still engines and other applications, it is becoming apparent has an adverse affect on mechanical and thermal proper- that there is a need to understand better their behaviour ties. In addition microstructural changes due to high tem under conditions of"thermal shock. This term describes perature exposure have a detrimental effect on the perfor n event in which a sudden(usually downward)tempera- mance of these materials under thermal shock conditions ture change generates stresses in the material that can lead Recent studies have concentrated mostly on materials to cracking and long-term property degradation [1]. Such with continuous unidirectional (UD) fibres or on various events are common in high-temperature machinery, e.g. types of porous 2-D SiC/SiC prepared by chemical vapour in the case of an emergency shut-down of a gas turbine. infiltration. It was found that multiple matrix cracking Fibre-reinforced CMCs have been shown to exhibit bet- perpendicular to the fibres was the main damage mode ter behaviour under conditions of thermal shock than their on the faces of UD CMCs that contained longitudinal monolithic or particulate-reinforced counterparts [2]. fibres, accompanied by the appearance of ' thumb-nail With optimum selection of fibres and matrices, favourable or 'thermal debond matrix cracks on their end faces [3- residual stress conditions can be established in the matrix, 10]. Large-scale porosity affected the behaviour of 2-D which lead to increased resistance to crack initiation due to SiC/SiC, as the pores acted as crack initiation sites at thermal shock. After cracks appear, the presence of crack- shocks of moderate severity [11-13]. Information on the * Author to whom all correspondence should be addressed. 0022-2461◎2006S Science Business Media, Inc. DOI:10.1007/s10853-006-6594-8

40TH ANNIVERSARY J MATER SCI 4 1 (2 0 0 6 ) 9 5 1 –9 6 2 Damage characterisation of thermally shocked cross-ply ceramic composite laminates C. KASTRITSEAS, P. A. SMITH, J. A. YEOMANS∗ School of Engineering (H6), University of Surrey, Guildford, GU2 7XH, Surrey, UK E-mail: j.yeomans@surrey.ac.uk Damage due to thermal shock in cross-ply Nicalon/calcium aluminosilicate ceramic matrix composites has been investigated. Heated specimens of two simple [(0◦/90◦)s and (90◦/0◦)s] and two multi-layer [(0◦/90◦)3s and (90◦/0◦)3s] materials were quenched into water at room temperature. Crack morphologies were assessed by reflected light microscopy and scanning electron microscopy. The use of image assembling software allowed the generation of reflected light microscopy images of all of the thermally-shocked surfaces onto which the crack patterns were then superimposed. This allowed clear identification of damage mechanisms and accurate quantification of damage accumulation with increasing severity of thermal shock. Damage was first detected in the central plies of each composite. Composites with 0◦ central plies exhibited slightly higher resistance to thermal shock than their counterparts with 90◦ central plies. Although damage extended to the outer plies as the severity of the shock increased, crack density was found to vary with position at every shock: it was highest in the central plies and gradually reduced towards the outer plies. Multiple matrix cracking perpendicular to the fibre direction was the damage mode identified in 0◦ plies, while 90◦ plies contained cracks that ran along the ply length. At more severe shocks the morphology of these crack patterns was affected in significantly different ways. In addition, the thinner, simple cross-ply composites exhibited much higher resistance to thermal shock than their multi-layer counterparts. C 2006 Springer Science + Business Media, Inc. 1. Introduction As fibre-reinforced ceramic matrix composites (CMCs) are being considered increasingly for high-performance engines and other applications, it is becoming apparent that there is a need to understand better their behaviour under conditions of ‘thermal shock’. This term describes an event in which a sudden (usually downward) temperature change generates stresses in the material that can lead to cracking and long-term property degradation [1]. Such events are common in high-temperature machinery, e.g. in the case of an emergency shut-down of a gas turbine. Fibre-reinforced CMCs have been shown to exhibit better behaviour under conditions of thermal shock than their monolithic or particulate-reinforced counterparts [2]. With optimum selection of fibres and matrices, favourable residual stress conditions can be established in the matrix, which lead to increased resistance to crack initiation due to thermal shock. After cracks appear, the presence of crack- ∗Author to whom all correspondence should be addressed. deflecting fibre-matrix interfaces ensures that the fibres remain largely unaffected, thus preserving the integrity of the material. However, damage due to thermal shock still has an adverse affect on mechanical and thermal properties. In addition, microstructural changes due to high temperature exposure have a detrimental effect on the performance of these materials under thermal shock conditions. Recent studies have concentrated mostly on materials with continuous unidirectional (UD) fibres or on various types of porous 2-D SiC/SiC prepared by chemical vapour infiltration. It was found that multiple matrix cracking perpendicular to the fibres was the main damage mode on the faces of UD CMCs that contained longitudinal fibres, accompanied by the appearance of ‘thumb-nail’ or ‘thermal debond’ matrix cracks on their end faces [3– 10]. Large-scale porosity affected the behaviour of 2-D SiC/SiC, as the pores acted as crack initiation sites at shocks of moderate severity [11–13]. Information on the 0022-2461 C 2006 Springer Science + Business Media, Inc. DOI: 10.1007/s10853-006-6594-8 951

40TH ANNIVERSARY thermal shock behaviour of CMCs of other configurations thin glassy layer over the material surfaces, probably a has been limited [14, 15 by-product of oxidation processes, which obscured crack This paper presents comprehensive experimental data observation. To overcome this problem, specimens were on the damage in cross-ply CMCs resulting from a thermal held at the highest temperatures for shorter periods of shock treatment Four different configurations were inves- time. i.e. 7-10 min. tigated for shocks of increasing severity. The results pre- Microscopic examination of the thermally-shocked sented include the determination of the onset of cracking, specimens was carried out mainly using reflected light the identification of the modes of fracture, and damage microscopy. Each surface under investigation was pho- quantification at each shock. The effect of the thickness tographed section by section and the stored images we s of the material on its behaviour under thermal shock is then assembled using suitable image assembling softwar also highlighted to produce an image of the whole surface. The cracking pattern was imposed manually on the resulting image after careful observation of the real surface using microscopy More detailed observation of crack patterns was also per 2. Materials and experimental techniques formed using a scanning electron microscope Two plates of cross-ply CMC comprising Nicalon fibres in calcium aluminosilicate(CAS)matrix were supplied by Rolls-Royce plc. The first was made by stacking together four plies of unidirectional material to create a composite 3. Results with thickness 0.7 mm with a(0%90%)s configuration. 3.1. Simple cross-ply Nicalon/CAs laminates The second plate consisted of twelve plies of unidirec- 3. 1. 1. The(0/90)s laminate tional Nicalon/CAS with a total thickness of 2.2 mm and The description of thermal shock damage on this laminate a(0/90%)3s configuration. In both plates the fibre volume is given with reference to the nomenclature of Fig. 1 fraction was 0.34 As can be seen, the central, thick transverse(90)ply Both plates were cut using a water-cooled diamond saw is designated as Tl (Transverse 1) while the adjacent into specimens with dimensions 6 mm x 6 mm x 0.7 mm longitudinal (0)plies are designated as LI (Longitudinal (0°909))and6mmx6mm×2.2mm(0°/90°)3s).1) Longitudinal faces (6 mm x 0. 7 mm for the(0/90%)s No damage was observed on the surfaces of material and 6 mm x 2.2 mm for the(0/90%)3s)were ground samples after quenching through temperature differen- using silicon carbide paper with grain size 320-4000 grit tials lower than 450C, i.e. for AT10 D)of room- exhibit at least some open porosity crack at 450oC or at temperature (20oC)water. It was then removed from temperature differentials close to this value the water bath and allowed to dry before microscopic The fracture mode identified on the surfaces of mate- examination rial samples shocked through△T≥40° C was matrix The quenching temperature difference, AT, is defined cracking If the direction of matrix cracks relative to the as the difference between the temperature at which the horizontal (i.e. the x-axis) is taken into account, matrix material was held in the furnace and the temperature of cracks can be further divided into those that run paralle the water bath. The critical quenching temperature dif- and those that run perpendicular to the horizontal. These ference, ATe, is the temperature differential that results two types of cracking phenomena are termed 'Horizon- in the onset of cracking. Temperature differentials in the tal Matrix Cracks'(HMCs)and 'Perpendicular Matrix range 100 to 800oC were investigated, with 2 or 3 speci- Cracks(PMCs), respectively. It should be noted that mens used at each AT. All specimens were initially held fibre breaks/failures could be observed even at the highes at high temperature for 15-20 min before quenching. It temperature differential investigated (AT=700-8000C) was found, however, that at the highest ATs investigated HMCs were the first form of damage seen after quench- (AT=700-800oC)this resulted in the formation of a ing through AT=450-500oC(Fig 2). They were located

40TH ANNIVERSARY thermal shock behaviour of CMCs of other configurations has been limited [14, 15]. This paper presents comprehensive experimental data on the damage in cross-ply CMCs resulting from a thermal shock treatment. Four different configurations were investigated for shocks of increasing severity. The results presented include the determination of the onset of cracking, the identification of the modes of fracture, and damage quantification at each shock. The effect of the thickness of the material on its behaviour under thermal shock is also highlighted. 2. Materials and experimental techniques Two plates of cross-ply CMC comprising Nicalon fibres in a calcium aluminosilicate (CAS) matrix were supplied by Rolls-Royce plc. The first was made by stacking together four plies of unidirectional material to create a composite with thickness ∼0.7 mm with a (0◦/90◦)s configuration. The second plate consisted of twelve plies of unidirectional Nicalon/CAS with a total thickness of 2.2 mm and a (0◦/90◦)3s configuration. In both plates the fibre volume fraction was 0.34. Both plates were cut using a water-cooled diamond saw into specimens with dimensions 6 mm × 6 mm × 0.7 mm ((0◦/90◦)s) and 6 mm × 6 mm × 2.2 mm ((0◦/90◦)3s). Longitudinal faces (6 mm × 0.7 mm for the (0◦/90◦)s and 6 mm × 2.2 mm for the (0◦/90◦)3s) were ground using silicon carbide paper with grain size 320–4000 grit and were subsequently polished using diamond paste to a 1 µm finish. By preparing the longitudinal faces adjacent to these initial faces, the effect of thermal shock treatment could be assessed on four different configurations: simple (0◦/90◦)s and (90◦/0◦)s from the samples cut from the first plate and multi-layer (0◦/90◦)3s and (90◦/0◦)3s from the samples obtained from the second. The water-quench test was employed to produce the thermal shock condition. Each specimen, after being heated for a short period of time in an electric muffle furnace at a pre-determined temperature, was dropped into a container with a large quantity (>10 l) of roomtemperature (∼20◦C) water. It was then removed from the water bath and allowed to dry before microscopic examination. The quenching temperature difference, T, is defined as the difference between the temperature at which the material was held in the furnace and the temperature of the water bath. The critical quenching temperature difference, Tc, is the temperature differential that results in the onset of cracking. Temperature differentials in the range 100 to 800◦C were investigated, with 2 or 3 specimens used at each T. All specimens were initially held at high temperature for 15–20 min before quenching. It was found, however, that at the highest Ts investigated (T = 700–800◦C) this resulted in the formation of a thin glassy layer over the material surfaces, probably a by-product of oxidation processes, which obscured crack observation. To overcome this problem, specimens were held at the highest temperatures for shorter periods of time, i.e. 7–10 min. Microscopic examination of the thermally-shocked specimens was carried out mainly using reflected light microscopy. Each surface under investigation was photographed section by section and the stored images were then assembled using suitable image assembling software to produce an image of the whole surface. The cracking pattern was imposed manually on the resulting image after careful observation of the real surface using microscopy. More detailed observation of crack patterns was also performed using a scanning electron microscope. 3. Results 3.1. Simple cross-ply Nicalon/CAS laminates 3.1.1. The (0◦/90◦)s laminate The description of thermal shock damage on this laminate is given with reference to the nomenclature of Fig. 1. As can be seen, the central, thick transverse (90◦) ply is designated as T1 (Transverse 1) while the adjacent longitudinal (0◦) plies are designated as L1 (Longitudinal 1). No damage was observed on the surfaces of material samples after quenching through temperature differentials lower than 450◦C, i.e. for T< 450◦C. Some of the samples quenched through T=450◦C, the majority of the samples quenched through T=480◦C and almost all of the samples tested through T=500◦C showed evidence of cracking in the form of shallow, hair-like cracks. Thus, it was decided that the critical quenching temperature differential for this laminate lies in the range 450–500◦C, i.e. Tc=450–500◦C. The actual value of Tc seems to vary depending on experimental details, such as the angle of impact with the quenching medium and the extent of pre-existing damage on the surfaces of the material. Generally, surfaces that exhibit at least some open porosity crack at 450◦C or at temperature differentials close to this value. The fracture mode identified on the surfaces of material samples shocked through T ≥ 450◦C was matrix cracking. If the direction of matrix cracks relative to the horizontal (i.e. the x-axis) is taken into account, matrix cracks can be further divided into those that run parallel and those that run perpendicular to the horizontal. These two types of cracking phenomena are termed ‘Horizontal Matrix Cracks’ (HMCs) and ‘Perpendicular Matrix Cracks’ (PMCs), respectively. It should be noted that no fibre breaks/failures could be observed even at the highest temperature differential investigated (T = 700–800◦C). HMCs were the first form of damage seen after quenching through T = 450–500◦C (Fig. 2). They were located 952

40TH ANNIVERSARY 10m Figure 3 Photomicrograph of PMC in LI at AT=500C that arrests inside 0 ply could be seen traversing the thickness of ll at at 2.5 800C. while 1-2 longer HMCs ran along the length of O/90),SiC/CAS 2 The application of higher temperature differentials did not result in significant morphological changes in either HMCs or PMCs. both damage mechanisms remained sur- face features of small depth. At all temperature differen- tials PMCs were evenly distributed between the two Oo 0.5 plies termed LI Crack densities for HMCs and Pmcs were determined 450 in terms of crack length per unit area(mm/mm or mm-) 5005506 in order to allow a comparison to be made, as shown in Quenching Temperature Difference(C) at higher temperature differentials and form much longer each AT Relevant trends for each damage mode are also shown nsity Fig 4. The failure of small, individual HMCs to connect Figure 4 Crack densities of PMcs and HMcs and the total crack cracks results in only a moderate increase in crack density with increasing applied shock. By contrast, the density of The main mode of damage due to thermal shock on this PMCs increases at a higher rate. Although PMCs appear laminate was matrix cracking. Horizontal matrix cracks at higher AT, they constitute the larger percentage of (HMCs)developed parallel to the x-axis while perpen- the total damage accumulated at the higher temperature dicular matrix cracks(PMCs)could be seen running at differentials investigated Although an attempt was made to deduce differ- right angles to the x-axis. No damage to the fibres could quenching temperature differentials, the observed scatter PMCs were the first type of damage due to thermal in the measured crack density values did not allow safe shock to appear on the surfaces of this laminate. They conclusions to be reached were exclusively in the central, thick LI ply at AT 500oC. They did not affect the longitudinal fibres on their path and were arrested at fibre-matrix interfaces inside the 0o ply or at the interface between 0 and 90 plies. 3.1.2. The (90/0 s laminate HMCs or thermal debond appeared in the The description of this laminate is similar to that of the Tl plies of this laminate after quenching through (0% /90%)s laminate except that the central, thick ply is in AT=550C. Only a few of these cracks could be ob- the 0 configuration and designated Ll, while the adjacent served and they were deflected at successive fibre-matrix plies are in the 90 configuration and designated as Tl. interfaces. A major, long HMC could not be identified. Damage due to thermal shock was observed using opti- PMCs originating at AT= 500oC were few in num- cal microscopy after quenching through temperature dif- ber and did not penetrate deep inside the matrix ma- ferential higher than 500oC. Thus, the critical quenching terial. In addition, they did not span the full thick temperature differential for this laminate was determined ness of the Ll plies. Most of the PMCs could be seen tobe△Te=500°C. traversing the thickness of the central Ll ply at AT

40TH ANNIVERSARY Figure 3 Photomicrograph of PMC in L1 at T = 500◦C that arrests inside 0◦ ply. could be seen traversing the thickness of L1 at T = 800◦C, while 1–2 longer HMCs ran along the length of T1. The application of higher temperature differentials did not result in significant morphological changes in either HMCs or PMCs. Both damage mechanisms remained surface features of small depth. At all temperature differentials PMCs were evenly distributed between the two 0◦ plies termed L1. Crack densities for HMCs and PMCs were determined in terms of crack length per unit area (mm/mm2 or mm−1) in order to allow a comparison to be made, as shown in Fig. 4. The failure of small, individual HMCs to connect at higher temperature differentials and form much longer cracks results in only a moderate increase in crack density with increasing applied shock. By contrast, the density of PMCs increases at a higher rate. Although PMCs appear at higher T, they constitute the larger percentage of the total damage accumulated at the higher temperature differentials investigated. Although an attempt was made to deduce different trends in crack density increase between successive quenching temperature differentials, the observed scatter in the measured crack density values did not allow safe conclusions to be reached. 3.1.2. The (90◦/0◦)s laminate The description of this laminate is similar to that of the (0◦/90◦)s laminate except that the central, thick ply is in the 0◦ configuration and designated L1, while the adjacent plies are in the 90◦ configuration and designated as T1. Damage due to thermal shock was observed using optical microscopy after quenching through temperature differentials higher than 500◦C. Thus, the critical quenching temperature differential for this laminate was determined to be Tc = 500◦C. Figure 4 Crack densities of PMCs and HMCs and the total crack density at each T. Relevant trends for each damage mode are also shown. The main mode of damage due to thermal shock on this laminate was matrix cracking. Horizontal matrix cracks (HMCs) developed parallel to the x-axis while perpendicular matrix cracks (PMCs) could be seen running at right angles to the x-axis. No damage to the fibres could be detected even at the highest temperature differentials investigated. PMCs were the first type of damage due to thermal shock to appear on the surfaces of this laminate. They were exclusively in the central, thick L1 ply at T = 500◦C. They did not affect the longitudinal fibres on their path and were arrested at fibre-matrix interfaces inside the 0◦ ply or at the interface between 0◦ and 90◦ plies. HMCs or ‘thermal debond’ cracks appeared in the T1 plies of this laminate after quenching through T = 550◦C. Only a few of these cracks could be observed and they were deflected at successive fibre-matrix interfaces. A major, long HMC could not be identified. PMCs originating at T = 500◦C were few in number and did not penetrate deep inside the matrix material. In addition, they did not span the full thickness of the L1 plies. Most of the PMCs could be seen traversing the thickness of the central L1 ply at T = 954

40TH ANNIVERSARY 3.1.3. Summary of observations of simple cross-ply Nicalon/CAS laminates Damage due to thermal shock in simple cross-ply lami- nates was described and quantified in detail in this sec tion The main damage mechanism was found to be matrix cracking Matrix cracks advanced parallel to the horizon tal in transverse plies and at right angles to the horizontal in longitudinal plies. They were deflected at fibre-matrix interfaces at every quenching temperature investigated, so no fibre failures were observed The (90/0%)s laminate exhibited better resistance to thermal shock than the(0/90%s laminate. However, dam age in both laminates originated in the thick, central ply and then, at higher temperature differentials, extended to adjacent plies. Matrix cracks in both laminates were found to remain Figure 5 PMC bridging li thickness at△T=600°C. shallow, surface features irrespective of the severity of thermal shock loading. However, damage in the form of PMCs was more extensive than damage in the form of 550C, while all of them bridged it at ar=600C be. HMCs in both laminates, especially at higher quenching fore being arrested at the interface between 0 and 90 temperature differences plies(Fig. 5). At the highest temperature differentials (AT=700-8000C), some PMCs could be seen propa- gating a short distance inside the adjacent 90 plies A small number of short HMCs were almost evenly 3. 2. Multi-layer cross-ply Nicalon/CAS distributed between the TI plies at all temperature dif- laminates ferential investigated. The increase in AT resulted in a 3.2.1. The (0/90 l3s laminate moderate increase in their length in most cases The description of thermal shock damage on this laminate The depth and opening of both PMCs and is given with reference to the nomenclature of Fig. 7 HMCs were not altered by the application of For this Nicalon/CAS laminate AT =350°C. The form higher temperature differentials. They both remained of thermal shock damage observed was matrix crackin surface features throughout the temperature range which can be further divided into PMCs and HMCs. The investigated. fibres remained unaffected at all she OCKs The number of PMCs increased significantly for higher HMCs were the first form of damage observed after of HMCs shows only quenching through AT=350C. They were located ex- a moderate increase. The large difference in the rate clusively in 90 plies. Depending on the specimen un- of increase between the two types of matrix crack- der observation, these cracks emanated either from flaws, is evident in the graph of Fig. 6. It can be such as pores, and were contained inside the ply, or orig seen that at all ATs, about 2/3 of the total thermal inated from the edges of the ply and ran towards its cen shock damage is due to the formation and extension of tre. They were continuously deflected at successive fibre PMCS matrix interfaces PMCs were detected on the surfaces of thermally 1.8 shocked specimens, exclusively in 0 plies, after quench 1.6(90/0) SiC/CAS ing through AT=400oC. These cracks ran perpendicular to the horizontal (i.e. to the longitudinal fibres of the 0o 攴12 plies), leaving the fibres on their path unaffected, and ar- rested either at a fibre-matrix interface inside the ply or at 08 the interfaces between 00 and 90 plies 20.6 The evolution of both types of damage with increasing 04 FIl applied AT can be seen in the sequence of reflected light microscopy images of Fig. 8. At△T=350400° C only random HMCs could be gen 500 550 erally seen in the thick, central transverse ply (T1). How ever, a much longer crack was also evident in some spec Figure6 Crack densities of PMCs and HMCs and the total crack density imens quenched at this temperature differential. These at each AT Relevant trends for each damage mode are also shown. cracks were limited to the surface of the material

40TH ANNIVERSARY Figure 5 PMC bridging L1 thickness at T = 600◦C. 550◦C, while all of them bridged it at T = 600◦C before being arrested at the interface between 0◦ and 90◦ plies (Fig. 5). At the highest temperature differentials (T = 700–800◦C), some PMCs could be seen propagating a short distance inside the adjacent 90◦ plies. A small number of short HMCs were almost evenly distributed between the T1 plies at all temperature differentials investigated. The increase in T resulted in a moderate increase in their length in most cases. The depth and opening of both PMCs and HMCs were not altered by the application of higher temperature differentials. They both remained surface features throughout the temperature range investigated. The number of PMCs increased significantly for higher Ts while the crack density of HMCs shows only a moderate increase. The large difference in the rate of increase between the two types of matrix cracking is evident in the graph of Fig. 6. It can be seen that at all Ts, about 2/3 of the total thermal shock damage is due to the formation and extension of PMCs. Figure 6 Crack densities of PMCs and HMCs and the total crack density at each T. Relevant trends for each damage mode are also shown. 3.1.3. Summary of observations of simple cross-ply Nicalon/CAS laminates Damage due to thermal shock in simple cross-ply laminates was described and quantified in detail in this section. The main damage mechanism was found to be matrix cracking. Matrix cracks advanced parallel to the horizontal in transverse plies and at right angles to the horizontal in longitudinal plies. They were deflected at fibre-matrix interfaces at every quenching temperature investigated, so no fibre failures were observed. The (90◦/0◦)s laminate exhibited better resistance to thermal shock than the (0◦/90◦)s laminate. However, damage in both laminates originated in the thick, central ply and then, at higher temperature differentials, extended to adjacent plies. Matrix cracks in both laminates were found to remain shallow, surface features irrespective of the severity of thermal shock loading. However, damage in the form of PMCs was more extensive than damage in the form of HMCs in both laminates, especially at higher quenching temperature differences. 3.2. Multi-layer cross-ply Nicalon/CAS laminates 3.2.1. The (0◦/90◦)3s laminate The description of thermal shock damage on this laminate is given with reference to the nomenclature of Fig. 7. For this Nicalon/CAS laminate Tc = 350◦C. The form of thermal shock damage observed was matrix cracking, which can be further divided into PMCs and HMCs. The fibres remained unaffected at all shocks. HMCs were the first form of damage observed after quenching through Tc = 350◦C. They were located exclusively in 90◦ plies. Depending on the specimen under observation, these cracks emanated either from flaws, such as pores, and were contained inside the ply, or originated from the edges of the ply and ran towards its centre. They were continuously deflected at successive fibrematrix interfaces. PMCs were detected on the surfaces of thermallyshocked specimens, exclusively in 0◦ plies, after quenching through T = 400◦C. These cracks ran perpendicular to the horizontal (i.e. to the longitudinal fibres of the 0◦ plies), leaving the fibres on their path unaffected, and arrested either at a fibre-matrix interface inside the ply or at the interfaces between 0◦ and 90◦ plies. The evolution of both types of damage with increasing applied T can be seen in the sequence of reflected light microscopy images of Fig. 8. At T = 350–400◦C only random HMCs could be generally seen in the thick, central transverse ply (T1). However, a much longer crack was also evident in some specimens quenched at this temperature differential. These cracks were limited to the surface of the material. 955

40TH ANNIVERSARY Figure 7 The nomenclature used to describe damage due to thermal shock on a(0 /90%)3s laminate. At AT=400C PMCs appeared in the 0 plies(L1) In general, the application of higher ATs did not affect adjacent to TI while longer, random HMCs could again the morphology of PMCs. In contrast, HMCs located in be seen in Tl Damage, in the form of PMCs and HMCs, transverse plies at or close to the centreline of the face appeared in the 0 plies designated as L2 and the 90 plies became deeper and their opening, as well as their length, designated as T2, respectively, at AT=450C. At this increased significantly at the highest temperature differ temperature differential, a long HMC propagated along entials investigated(Fig 9) the central TI ply while in the t2 plies only individual PMCs were evenly distributed between the longitudinal HMCs appeared plies of the designation (i.e. Ll, L2 or L3). This was Similar patterns were observed at AT=500C. How- not exactly the case for HMCs as these were distributed in ever, almost all PMCs spanned the thickness of the L1, L2 a more random fashion, especially at the higher temper plies while some started to extend into the adjacent 90 ature differentials, between the pairs of transverse plies plies. In addition, some HMCs in the T2 plies connected (i.e. Tl, T2 or T3)depending on the specimen under in- to form longer cracks vestigation. At AT=600C all plies of this system sustained some The increase in PMc density with increasing shock form of thermal shock damage; TI contained a long, deep severity for each set of longitudinal plies (Ll, L2, L3)is HMC, T2 exhibited shorter and shallower HMCs, while shown in the graph of Fig. 10a. It is evident that crack individual, random HMCs could be seen in T3. In addi- density is always higher for the plies located towards the tion, all longitudinal plies (Ll, L2, L3)contained PMCs, centre of the sample surface, i.e. CDLI>CDL2>CDL3 at the number of which decreased on going from the cen- each AT investigated(CD: Crack Density). The rates of treline(C-C) towards the top or bottom edges of the increase of cracking in each set of plies are comparable surface Fig. 10b shows the change in HMC density with in- The application of even higher ATs(700-800oC)lead creasing temperature differential. A significant increase to an increase in the number of PMCs in the longitudinal in cracking can be observed, especially at the higher ther plies, although it again looked as if the plies closer to the mal shocks. The scatter in experimental data is larger at centreline had higher densities of these cracks than those the higher temperature differentials, which reflects the further away. In addition, some PMCs(especially in LI) randomness in the appearance and point of origin of long could be seen to extend into the adjacent transverse plies cracks in Tl and/or T2. TI and T2). HMCs followed a more random pattern. Comparison between PMCs and HMCs(Fig. 10b)re- There was always a long, deep crack that travelled along veals that the rate of increase in density of PMCs is much almost the full length of the ply in either Tl or T2. The higher than that of HMCs and, at high temperature differ- rest of these plies contained shorter and shallower cracks entials, PMCs are the major contribution to the total crack while the cracks located in T3, although continuously density. However, this graph fails to capture the signifi- increasing in number and length, failed to connect into cant differences in morphology between the two types of longer hmcs even at the highest at matrix cracking at△T≥600°C

40TH ANNIVERSARY Figure 7 The nomenclature used to describe damage due to thermal shock on a (0◦/90◦)3s laminate. At T = 400◦C PMCs appeared in the 0◦ plies (L1) adjacent to T1 while longer, random HMCs could again be seen in T1. Damage, in the form of PMCs and HMCs, appeared in the 0◦ plies designated as L2 and the 90◦ plies designated as T2, respectively, at T = 450◦C. At this temperature differential, a long HMC propagated along the central T1 ply while in the T2 plies only individual HMCs appeared. Similar patterns were observed at T = 500◦C. However, almost all PMCs spanned the thickness of the L1, L2 plies while some started to extend into the adjacent 90◦ plies. In addition, some HMCs in the T2 plies connected to form longer cracks. At T = 600◦C all plies of this system sustained some form of thermal shock damage; T1 contained a long, deep HMC, T2 exhibited shorter and shallower HMCs, while individual, random HMCs could be seen in T3. In addition, all longitudinal plies (L1, L2, L3) contained PMCs, the number of which decreased on going from the centreline (C-C’) towards the top or bottom edges of the surface. The application of even higher Ts (=700–800◦C) lead to an increase in the number of PMCs in the longitudinal plies, although it again looked as if the plies closer to the centreline had higher densities of these cracks than those further away. In addition, some PMCs (especially in L1) could be seen to extend into the adjacent transverse plies (T1 and T2). HMCs followed a more random pattern. There was always a long, deep crack that travelled along almost the full length of the ply in either T1 or T2. The rest of these plies contained shorter and shallower cracks while the cracks located in T3, although continuously increasing in number and length, failed to connect into longer HMCs even at the highest T. In general, the application of higher Ts did not affect the morphology of PMCs. In contrast, HMCs located in transverse plies at or close to the centreline of the face became deeper and their opening, as well as their length, increased significantly at the highest temperature differentials investigated (Fig. 9). PMCs were evenly distributed between the longitudinal plies of the same designation (i.e. L1, L2 or L3). This was not exactly the case for HMCs as these were distributed in a more random fashion, especially at the higher temperature differentials, between the pairs of transverse plies (i.e. T1, T2 or T3) depending on the specimen under investigation. The increase in PMC density with increasing shock severity for each set of longitudinal plies (L1, L2, L3) is shown in the graph of Fig. 10a. It is evident that crack density is always higher for the plies located towards the centre of the sample surface, i.e. CDL1>CDL2>CDL3 at each T investigated (CD: Crack Density). The rates of increase of cracking in each set of plies are comparable. Fig. 10b shows the change in HMC density with increasing temperature differential. A significant increase in cracking can be observed, especially at the higher thermal shocks. The scatter in experimental data is larger at the higher temperature differentials, which reflects the randomness in the appearance and point of origin of long cracks in T1 and/or T2. Comparison between PMCs and HMCs (Fig. 10b) reveals that the rate of increase in density of PMCs is much higher than that of HMCs and, at high temperature differentials, PMCs are the major contribution to the total crack density. However, this graph fails to capture the signifi- cant differences in morphology between the two types of matrix cracking at T≥600◦C. 956

40TH ANNIVERSARY The evolution of both types of damage with increasing applied AT can be seen in the sequence of reflected light microscopy images of Fig. 1 At the onset of fracture(△Te=400°C) nly visible in the central plies of this laminate PMCs in LI did not bridge the ply thickness while HMCs in TI were associated mainly with open pores. The application of temperature differentials up to AT 500C did not change the morphology of either type of matrix crack: they remained shallow surface features However PMCs were also visible in L2 and L3 and the presence of HMCs extended to T2.At△T=500°C HMCS of significant length could be seen originating either inside TI and T2 or from the ply edges running towards the centre of the specimen The length and depth of HMCs in Tl became much 9 SEM image showing deep HMC in the central transverse ply at a larger at AT=600oC. At this temperature differential, damage was detected in every ply of the laminate PMCS could be seen bridging the thicknesses of their respective 876 plies and some extended into adjacent transverse plies ( 0/90)s SICICAS The population of these cracks seemed higher the closer the longitudinal ply was located to the centreline(C-C)of the polished surface. The same was true for the transverse 国 plies Application of even higher temperature differentials re- sulted in long, deep HMCs in TI and T2(with depth always being larger in T1), longer HMCs in T3, and mul tiplication of PMCs in all longitudinal plies. However, 300 400 500 600 700 800 900 the depth of PMCs, even in Ll, did not increase In ad- Quenching Temperature Difference(C) dition, only short HMCs were visible in the outer T3 plies PMCs were distributed uniformly between longitu dinal plies with the same designation. By contrast, HMCS (o/90) SiC/CAS accumulated in the pairs of transverse plies in a more regular fashion. The differences in depth between HMCs in Tl, T2, and T3 can be clearly seen in the SEM images of Fig. 12 E15 The accumulation of damage in the 0 plies of this laminate at increasing thermal shocks can be seen in the graph of Fig. 13a. Significant increases in crack density are evident and at each temperature differen- 0 350400450500600700800 tial CDLI >CD 2>CDL3. The rate of increase in crack- ing for LI looks to be higher than the rates of in- Quenching Temperature Difference ('C) crease for L2 and l3. However the scatter of the perimental results does not allow safe conclusions to be reached DLI>CDL2>CD3 at all ATs and(b )Crack densities of PMCs and HMCs The accumulation of damage in the transverse plies of and total crack density at each AT this laminate is shown in Fig. 13b. The density of HMCs increases continuously with the application of higher tem- plies. Although they generally ran horizontally, succes- perature differentials. Note that only the total crack den- sive fibre-matrix interfaces deflected them continuously. that CDT1>CDT2>CDr3 at each temperature differential PMCs were also detected after quenching through the crit- However, whether CDTi or CDr was higher was mostly ical temperature differential, i.e. at ATc= 400C. They random result that depended on the point of origin of appeared only in 0 plies. Their advance was at right an- the respective HMCs. In general, HMCs emanating from gles to the longitudinal fibres, which remained unaffected ply edges ran for longer lengths From the comparison of since Pmcs were deflected at fibre-matrix interfaces crack densities of each type of damage at each AT it can

40TH ANNIVERSARY Figure 9 SEM image showing deep HMC in the central transverse ply at a high temperature differential. Figure 10 (a) Crack density as a function of T for PMCs for each set of longitudinal plies of (0/90)3s Nicalon/CAS laminate. Note that CDL1>CDL2>CDL3 at all Ts and (b) Crack densities of PMCs and HMCs and total crack density at each T. plies. Although they generally ran horizontally, successive fibre-matrix interfaces deflected them continuously. PMCs were also detected after quenching through the critical temperature differential, i.e. at Tc = 400◦C. They appeared only in 0◦ plies. Their advance was at right angles to the longitudinal fibres, which remained unaffected since PMCs were deflected at fibre-matrix interfaces. The evolution of both types of damage with increasing applied T can be seen in the sequence of reflected light microscopy images of Fig. 11. At the onset of fracture (Tc = 400◦C), damage was only visible in the central plies of this laminate. PMCs in L1 did not bridge the ply thickness while HMCs in T1 were associated mainly with open pores. The application of temperature differentials up to T = 500◦C did not change the morphology of either type of matrix crack: they remained shallow surface features. However, PMCs were also visible in L2 and L3 and the presence of HMCs extended to T2. At T = 500◦C, HMCs of significant length could be seen originating either inside T1 and T2 or from the ply edges running towards the centre of the specimen. The length and depth of HMCs in T1 became much larger at T = 600◦C. At this temperature differential, damage was detected in every ply of the laminate. PMCs could be seen bridging the thicknesses of their respective plies and some extended into adjacent transverse plies. The population of these cracks seemed higher the closer the longitudinal ply was located to the centreline (C-C ) of the polished surface. The same was true for the transverse plies. Application of even higher temperature differentials resulted in long, deep HMCs in T1 and T2 (with depth always being larger in T1), longer HMCs in T3, and multiplication of PMCs in all longitudinal plies. However, the depth of PMCs, even in L1, did not increase. In addition, only short HMCs were visible in the outer T3 plies. PMCs were distributed uniformly between longitudinal plies with the same designation. By contrast, HMCs accumulated in the pairs of transverse plies in a more irregular fashion. The differences in depth between HMCs in T1, T2, and T3 can be clearly seen in the SEM images of Fig. 12. The accumulation of damage in the 0◦ plies of this laminate at increasing thermal shocks can be seen in the graph of Fig. 13a. Significant increases in crack density are evident and at each temperature differential CDL1>CDL2>CDL3. The rate of increase in cracking for L1 looks to be higher than the rates of increase for L2 and L3. However, the scatter of the experimental results does not allow safe conclusions to be reached. The accumulation of damage in the transverse plies of this laminate is shown in Fig. 13b. The density of HMCs increases continuously with the application of higher temperature differentials. Note that only the total crack density is plotted at each T. Generally, it can be assumed that CDT1>CDT2>CDT3 at each temperature differential. However, whether CDT1 or CDT2 was higher was mostly a random result that depended on the point of origin of the respective HMCs. In general, HMCs emanating from ply edges ran for longer lengths. From the comparison of crack densities of each type of damage at each T it can 958

40TH ANNIVERSARY (90/0)s SIC/CAS Perpendicular Matrix Cracking 543210 300400500600700800900 Quenching Temperature Difference(c) 4 3.5 (a) 3 2 15 0 40045050060 700 Quenching Temperature Difference (C) Figure 13(a) Crack density as a function of AT for PMCs for each set of longitudinal plies of (90/0)3s Nicalon/CAS laminate. Note that CDLI>CDL?>CDL3 at all ATs and (b) Crack densities of PMCs and HMCs and total crack density at each AT. Relevant trends for each damage mode even higher quenching temperature differences, damage became more extensive, especially the PMCs in longi tudinal plies. However, HMCs in transverse plies were observed, apart from increasing in length, to penetrate deeper and deeper into the matrix. The extent of ther mal shock damage exhibited a gradient across the ma- terial surface: higher crack densities and deeper HMCs were located at or close to the centreline. On moving to- wards the outer plies the extent of the damage reduced significantly In terms of the number of cracks and their measured length as a function of the surface area, damage in the form of PMcs was found to be much more extensive compared with that in the form of HMCs, especially at severe thermal shocks. However, whereas PMCs propa- gated only at the surface of the laminate, HMCs could be seen to extend deeply into the matrix for△T≥600°C Figure /2 SEM images of HMCs in(a)T1, (b)T2, and(c)T3 at AT= was evident from their increased opening. Unfortunately, 800C. The differences in depth can be clearly observed. the depth they penetrated could not be determined with any particular accuracy experimentally. However, judging in the case of the(90%/0%)3s system, in those adjacent to from the openings of the crack surfaces at the surface, the them. With the application of higher differentials, damage extent of their propagation on the surface(from edge to extended to the outer plies until, at intermediate shocks edge for some specimens at severe shocks), and the fact (AT=600C), the surfaces of all plies were fractured. At that in these configurations they cannot meet any ply inter

《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_Damage characterisation of thermally shocked

《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_Damage characterisation of thermally shocked